# PLANTS, STRESS & PROTEINS

EDITED BY: Dipanjana Ghosh, Qingsong Lin, Jian Xu and Hanjo A. Hellmann PUBLISHED IN: Frontiers in Plant Science

#### *Frontiers Copyright Statement*

*© Copyright 2007-2017 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.*

*The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.*

*Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.*

*Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.*

*As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.*

> *All copyright, and all rights therein, are protected by national and international copyright laws.*

*The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use.*

ISSN 1664-8714 ISBN 978-2-88945-267-5 DOI 10.3389/978-2-88945-267-5

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# **PLANTS, STRESS & PROTEINS**

Topic Editors:

**Dipanjana Ghosh,** People's University Bhopal, India and National University of Singapore, Singapore

**Qingsong Lin,** National University of Singapore, Singapore **Jian Xu,** National University of Singapore, Singapore **Hanjo A. Hellmann,** Washington State University, United States

Cover image by "Dipanjana Ghosh"

Biotic and abiotic stress factors deliver a huge impact on plant life. Biotic stress factors such as damage through pathogens or herbivore attack, as well as abiotic stress factors like variation in temperature, rainfall and salinity, have placed the plant kingdom under constant challenges for survival. As a consequence, global agricultural and horticultural productivity has been disturbed to a large extent.

Being sessile in nature, plants cannot escape from the stress, and instead adapt changes within their system to overcome the adverse conditions. These changes include physiological, developmental and biochemical alterations within the plant body which

influences the genome, proteome and metabolome profiles of the plant. Since proteins are the ultimate players of cellular behavior, proteome level alterations during and recovery period of stress provide direct implications of plant responses towards stress factors.

With current advancement of modern high-throughput technologies, much research has been carried out in this field. This e-book highlights the research and review articles that cover proteome level changes during the course or recovery period of various stress factors in plant life. Overall, the chapters in this e-book has provided a wealth of information on how plants deal with stress from a proteomics perspective.

**Citation:** Ghosh, D., Lin, Q., Xu, J., Hellmann, H. A., eds. (2017). Plants, Stress & Proteins. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-267-5

# Table of Contents

#### **Chapter 1: Review and opinions on Plant stress proteomics**

*06 Editorial: How Plants Deal with Stress: Exploration through Proteome Investigation*

Dipanjana Ghosh, Qingsong Lin, Jian Xu and Hanjo A. Hellmann


### **Chapter 2: Signaling events during stress; in the context of proteomics**

*26 Vascular Sap Proteomics: Providing Insight into Long-Distance Signaling during Stress*

Philip Carella, Daniel C. Wilson, Christine J. Kempthorne and Robin K. Cameron

*34 Phloem Proteomics Reveals New Lipid-Binding Proteins with a Putative Role in Lipid-Mediated Signaling*

Allison M. Barbaglia, Banita Tamot, Veronica Greve and Susanne Hoffmann-Benning


Heidi Pertl-Obermeyer, Oliver Trentmann, Kerstin Duscha, H. Ekkehard Neuhaus and Waltraud X. Schulze

*81 Hydrogen Sulfide: A Signal Molecule in Plant Cross-Adaptation* Zhong-Guang Li, Xiong Min and Zhi-Hao Zhou

### **Chapter 3: Abiotic Stress & alterations in plant proteome Section 3.1: Plant proteome alterations during drought stress Section 3.1.1: Drought stress**

*93 Analysis of Drought-Induced Proteomic and Metabolomic Changes in Barley (***Hordeum vulgare L.***) Leaves and Roots Unravels Some Aspects of Biochemical Mechanisms Involved in Drought Tolerance*

Klaudia Chmielewska, Paweł Rodziewicz, Barbara Swarcewicz, Aneta Sawikowska, Paweł Krajewski, Łukasz Marczak, Danuta Ciesiołka, Anetta Kuczyn'ska, Krzysztof Mikołajczak, Piotr Ogrodowicz, Karolina Krystkowiak, Maria Surma, Tadeusz Adamski, Paweł Bednarek and Maciej Stobiecki

#### **Section 3.1.2: Drought stress in combination with other abiotic stress**

*107 Contrasting Changes Caused by Drought and Submergence Stresses in Bermudagrass (***Cynodon dactylon***)*

Tiantian Ye, Haitao Shi, Yanping Wang and Zhulong Chan

*121 The Difference of Physiological and Proteomic Changes in Maize Leaves Adaptation to Drought, Heat, and Combined Both Stresses* Feiyun Zhao, Dayong Zhang, Yulong Zhao, Wei Wang, Hao Yang, Fuju Tai, Chaohai Li and Xiuli Hu

#### **Section 3.2: Light & temperature stress**

*140 Comparative Proteomic Analysis of the Response of Maize (***Zea mays L.***) Leaves to Long Photoperiod Condition*

Liuji Wu, Lei Tian, Shunxi Wang, Jun Zhang, Ping Liu, Zhiqiang Tian, Huimin Zhang, Haiping Liu and Yanhui Chen

*156 Overexpression of Glycolate Oxidase Confers Improved Photosynthesis under High Light and High Temperature in Rice* Li-Li Cui, Yu-sheng Lu, Yong Li, Chengwei Yang and Xin-Xiang Peng

#### **Section 3.3: Salinity stress**

**168** *Comparative Proteomic Analysis of Cultured Suspension Cells of the Halophyte* **Halogeton glomeratus** *by iTRAQ Provides Insights into Response Mechanisms to Salt Stress*

Juncheng Wang, Lirong Yao, Baochun Li, Yaxiong Meng, Xiaole Ma, Yong Lai, Erjing Si, Panrong Ren, Ke Yang, Xunwu Shang and Huajun Wang


#### **Section 3.4: Heavy metal stress**

*223 Analysis of Copper-Binding Proteins in Rice Radicles Exposed to Excess Copper and Hydrogen Peroxide Stress*

Hongxiao Zhang, Yan Xia, Chen Chen, Kai Zhuang, Yufeng Song and Zhenguo Shen

*238 Proteomic Profiling of the Interactions of Cd/Zn in the Roots of Dwarf Polish Wheat (***Triticum polonicum** *L.)*

Yi Wang, Xiaolu Wang, Chao Wang, Ruijiao Wang, Fan Peng, Xue Xiao, Jian Zeng, Xing Fan, Houyang Kang, Lina Sha, Haiqin Zhang and Yonghong Zhou

*249 The Dynamic Changes of the Plasma Membrane Proteins and the Protective Roles of Nitric Oxide in Rice Subjected to Heavy Metal Cadmium Stress*

Liming Yang, Jianhui Ji, Karen R. Harris-Shultz, Hui Wang, Hongliang Wang, Elsayed F. Abd-Allah, Yuming Luo and Xiangyang Hu

#### **Chapter 4: Biotic Stress**

### *267 Alterations in Kernel Proteome after Infection with* **Fusarium culmorum** *in Two Triticale Cultivars with Contrasting Resistance to* **Fusarium** *Head Blight*

Dawid Perlikowski, Halina Wiśniewska, Joanna Kaczmarek, Tomasz Góral, Piotr Ochodzki, Michał Kwiatek, Maciej Majka, Adam Augustyniak and Arkadiusz Kosmala

*277 Changes in the Proteome of Xylem Sap in* **Brassica oleracea** *in Response to*  **Fusarium oxysporum** *Stress*

Zijing Pu, Yoko Ino, Yayoi Kimura, Asumi Tago, Motoki Shimizu, Satoshi Natsume, Yoshitaka Sano, Ryo Fujimoto, Kentaro Kaneko, Daniel J. Shea, Eigo Fukai, Shin-Ichi Fuji, Hisashi Hirano and Keiichi Okazaki

*291 The effector repertoire of* **Fusarium oxysporum** *determines the tomato xylem proteome composition following infection*

Fleur Gawehns, Lisong Ma, Oskar Bruning, Petra M. Houterman, Sjef Boeren, Ben J. C. Cornelissen, Martijn Rep and Frank L. W. Takken

#### **Chapter 5: Miscellaneous**

*308 Proteomic Analyses Provide Novel Insights into Plant Growth and Ginsenoside Biosynthesis in Forest Cultivated* **Panax** *ginseng (F. Ginseng)*

Rui Ma, Liwei Sun, Xuenan Chen, Bing Mei, Guijuan Chang, Manying Wang and Daqing Zhao

# Editorial: How Plants Deal with Stress: Exploration through Proteome Investigation

#### Dipanjana Ghosh1, 2 \*, Qingsong Lin<sup>2</sup> , Jian Xu<sup>2</sup> and Hanjo A. Hellmann<sup>3</sup>

*<sup>1</sup> School of Pharmacy and Research, People's University, Bhopal, India, <sup>2</sup> Department of Biological Sciences, National University of Singapore, Singapore, Singapore, <sup>3</sup> School of Biological Sciences, Washington State University, Pullman, WA, United States*

Keywords: proteomics, abiotic stress, biotic stress, plant stress biology, heavy metal stress

#### **Editorial on the Research Topic**

#### **How Plants Deal with Stress: Exploration through Proteome Investigation**

Biotic and abiotic stress factors serve as consistent threats to a plant's life cycle. Biotic stress factors such as damage through pathogens or herbivore attack as well as abiotic stress factors like variation in temperature, rainfall, and salinity, have placed the members of the plant kingdom under constant challenges for their survival. As a consequence, global agricultural and horticultural productivity has always been below its optimal capacity.

Being sessile in nature, plants cannot escape from the stress, instead they acclimatize to adverse condition by adopting certain systemic changes. These changes include developmental and physiological alterations which influence the genome, proteome, and metabolome of the plant. Since proteins are key regulators of cellular responses, investigations into proteome alterations during stress induction as well as recovery, can provide important information on how plants cope with stress factors. The process has been widely discussed in this Research Topic which comprises nineteen research articles, two reviews, one mini review, and one opinion article.

A significant portion of the research articles included in this Research Topic investigated proteome level alterations in diverse organ parts of the plant body upon exposure to abiotic stress factors such as drought, salinity, and submergence. Comparative proteomic investigations between drought and submergence stress showed that bermudagrass adopt a quiescence strategy in a prolonged submerged state by declining metabolic activities, whereas in a drought environment plants increase their tolerance level through higher levels of photosynthesis and redox potential (Ye et al.). Enhanced levels of molecular chaperones were found to be associated with drought tolerance in multiple studies (Ye et al.; Zhao et al.; Chmielewska et al.) performed on different plant species (bermudagrass, corn, and barley, respectively). Interestingly, when maize plants were challenged by the combination of both drought and heat, up-regulation of a specific set of ethylene-responsive, and ABA-, stress-, and ripening-inducible-like proteins was observed. However, individual exposure to drought or heat stress did not show the above proteome alterations, indicating that these proteins are specifically required to resist the combinatorial effects of the above two stresses (Zhao et al.). Consistently, over-expression of glycolate oxidase (GLO) was found during adaptation of rice plants to both high-light intensity and high temperature but not to high-light intensity alone, with H2O<sup>2</sup> and salicylic acid being the signaling molecules that mediate the adaptive responses (Cui et al.). Studies on salt stress reported the up-regulation of mainly defense-related proteins and proteins related to enhanced energy metabolism (Maršálová et al.; Long et al.). In addition, a quantitative proteomics study explored that vacuolar ATPase AVP1 and sugar transporters of the ERDL (early responsive to dehydration-like) family and TMT2 (tonoplast

#### Edited and reviewed by:

*Norbert Rolland, Centre National de la Recherche Scientifique (CNRS), France*

#### \*Correspondence:

*Dipanjana Ghosh dipanjanaghosh@alumni.nus.edu.sg; dipanjanagh@gmail.com*

#### Specialty section:

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

Received: *09 June 2017* Accepted: *20 June 2017* Published: *30 June 2017*

#### Citation:

*Ghosh D, Lin Q, Xu J and Hellmann HA (2017) Editorial: How Plants Deal with Stress: Exploration through Proteome Investigation. Front. Plant Sci. 8:1176. doi: 10.3389/fpls.2017.01176* monosaccharide transporter 2) were upregulated upon salt stress treatment (Pertl-Obermeyer et al.). Notably, two plant growthpromoting bacteria signal compounds, lipo-chitooligosaccharide (LCO), and thuricin17 (Th17) were found to have a stronger effect on proteins related to the carbon and energy metabolic pathways in salt-treated plants when compared to mock-treated Arabidopsis plants (Subramanian et al.). These compounds thus hold promise for the development of potent agrochemicals that could improve crop productivity under salt stress.

Comparison of growth rate and stress responses between wild type and forest grown varieties of the same plant were investigated through proteomics investigations on Ginseng (Ma et al.) that revealed protein profiles of 25-year-old forest Ginseng is comparable to that of younger wild Ginseng and differentially expressed proteins were involved in energy metabolism, ginsenoside biosynthesis, and stress response.

Heavy metal induced stress is another critical challenge for agricultural productivity. Proteomic investigations on heavy metal induced rice varieties derived that phospholipase D activity is positively correlated with cadmium stress and Cu-binding proteins might help to ameliorate the effects of copper stress. In addition, a wide range of differentially expressed protein datasets have been reported for plants exposed under zinc and nitric oxide (NO) stress (Yang et al.; Wang et al.; Wang et al.; Zhang et al.).

Biotic stress such as fungal infection also contributes toward proteomic changes. Resistance against those fungal pathogens was found to be correlated with the abundance of amylase inhibitor proteins, leucine-rich repeats, and cysteine-containing secreted small proteins found in xylem sap as derived in proteomics investigations on cabbage, tomato, and Triticale plants challenged by several fungal species of the Fusarium genus (Perlikowski et al.; Pu et al.; Gawehns et al.).

Other than the above-mentioned research highlights, Opinion and review articles included in this Research Topic covered various important aspects related to plant stress responses at the protein level. These include vascular sap proteomics for longdistance signaling (Carella et al.), hydrogen sulfide as a signaling molecule in plant cross adaptation (Li et al.), the importance of proteomics in understanding crop stress tolerance (Ahmad et al.), and increasing confidence of proteomics data for identifying stress-responsive proteins in crop plants (Wu et al.; Wu and Wang).

Overall, this Research Topic has provided a wealth of information in the field of plant stress biology from a proteomics perspective. Functional characterization of some of the candidate proteins and pathways may ultimately contribute to the development of novel breeding strategies that will improve agricultural productivity in a changing climate.

### AUTHOR CONTRIBUTIONS

DG: Contribute in compilation of concepts from individual articles of the research topic, manuscript write up and revision; QL and JX: Contributed to the revision of the manuscript writeup; HH: Provided guidance during manuscript write up and contributed to the revision of the manuscript draft.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Ghosh, Lin, Xu and Hellmann. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Increasing Confidence of Proteomics Data Regarding the Identification of Stress-Responsive Proteins in Crop Plants

Xiaolin Wu and Wei Wang\*

State Key Laboratory of Wheat and Maize Crop Science, Collaborative Innovation Center of Henan Grain Crops, College of Life Science, Henan Agricultural University, Zhengzhou, China

Keywords: crop plants, abiotic stress, stress-responsive proteins, proteomics data, 2DE, iTRAQ labeling

## INTRODUCTION

Numerous stresses caused by complex environmental conditions, e.g., drought, heat, cold, salinity, strong light, UV, and heavy metals, negatively affect plant growth and lead to substantial crop losses worldwide. It is estimated that up to 50–70% of declines in crop productivity can be attributed to abiotic stress (Mittler, 2006). Abiotic stress, particularly drought and extreme temperatures, will be more frequent and severe in the near future because of global climate change (Horton et al., 2015). Understanding the abiotic stress response in plants has attracted substantial attention within the plant proteomics community.

#### Edited by:

Silvia Mazzuca, Università della Calabria, Italy

#### Reviewed by:

Christof Rampitsch, Agriculture and Agrifood Canada, Canada Qingsong Lin, National University of Singapore, Singapore

> \*Correspondence: Wei Wang wangwei@henau.edu.cn

#### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 02 March 2016 Accepted: 06 May 2016 Published: 24 May 2016

#### Citation:

Wu X and Wang W (2016) Increasing Confidence of Proteomics Data Regarding the Identification of Stress-Responsive Proteins in Crop Plants. Front. Plant Sci. 7:702. doi: 10.3389/fpls.2016.00702

Quantitative proteomic comparisons are particularly useful in defining proteins that change in abundance, form, location, activity, and these comparisons may indicate involvement in responses to alterations in environmental conditions (Thelen and Peck, 2007). Such analyses can detect proteins involved in the mechanisms underlying plant stress resistance to various abiotic stresses. These proteins can potentially serve as molecular markers in marker-assisted selection by possibly speeding up the identification of relevant targets for stress breeding.

Considering the 2DE and/or iTRAQ analysis methods of proteomics as an example, we briefly analyzed the methodological defects in detecting stress-responsive proteins in plants and propose our opinions for addressing these defects in future plant stress proteomics. The intended audiences of this opinion paper are novice rather than experienced scientists in the plant proteomics research community.

#### METHODOLOGICAL DEFECTS IN PLANT STRESS PROTEOMICS

Comparative proteomics detection of stress-responsive proteins in plants is performed through analyzing protein changes, including protein isoforms and molecular species generated by PTMs, between untreated and stress-treated samples or tolerant and intolerant plants. An increasing number of studies indicate that protein changes are important in plant stress response (e.g., reviews by Kosová et al., 2011; Barkla et al., 2013; Ghosh and Xu, 2014; Wu et al., 2016). Based on briefly reading the abstracts of these studies, it is obvious that 2DE-based approaches and iTRAQ-based approaches currently represent two major types of proteomics techniques in plant stress proteomics.

The 2DE method resolves proteins based on a native charge followed by mass (Rabilloud et al., 2010). The routine 2DE approach allows the detection of lower numbers of protein spots (compared to iTRAQ), and subsequent mass spectrometry-based identification can be applied only to differentially abundant stress-responsive proteins among the analyzed samples. Moreover, 2DE appears to be especially suitable for the detection of changes on the level of protein isoforms (Benešová et al., 2012). A disadvantage of 2DE is that spot matching among a group of 2DE gels can be an arduous task (Thelen and Peck, 2007; Rabilloud et al., 2010). Though DIGE-2DE can make comparisons easier, high-quality 2DE gels with minimal spot streaking and overlap are critical to simplify and maximize the accuracy of spot matching. In good instances, impressive 2DE maps seem like exquisite artistic works compared to the masses of lifeless iTRAQ data.

The iTRAQ analysis method is a second-generation proteomic technique that provides a gel-free shotgun quantitative analysis. It utilizes isobaric reagents to label tryptic peptides and monitor relative changes in protein and PTM abundance (Ross et al., 2004), and it allows for the comparison of up to eight samples. Thus, iTRAQ especially facilitates the analysis of time courses of plant stress responses or biological replicates in a single experiment. However, iTRAQ monitors several thousands of peptides without the ability to pre-select differentially abundant peptides prior to mass spectrometric identification. Compared to 2DE, iTRAQ requires intensive data analysis using appropriate software to detect and quantify the mass tags. To our knowledge, many iTRAQ analyses had been performed by commercial services. Quite often, it works like a "black box." The customers submit samples and get a list of differential proteins with ratios, without knowing the details of experimental processes. During the commercial iTRAQ analysis, experimental design may not be well taken care of, protein isolation may not be properly conducted, and experimental data may not be properly analyzed, which all contribute toward the distortion of iTRAQ data.

Comparative proteomics studies provide a great deal of data and novel insights on plant stress response. However, substantially inconsistent or unreliable results occur in plant stress proteomics research among different research groups. For example, this inconsistency is clear in the comparison of proteomic studies of maize (Zea mays) under salt and drought stresses using 2DE or iTRAQ approaches (**Table 1**).

In the three studies on proteomic changes in maize roots under salt stress, only three identified proteins were common among the dozens of differentially abundant proteins; only four proteins were common in two proteomic studies of maize leaf under drought conditions (**Table 1**). Even in a single experiment, only four stress-responsive proteins were identified by both the 2DE and iTRAQ analyses (Benešová et al., 2012). The partial overlap between the outputs of 2DE and iTRAQ approaches was limited due to their different characteristics (Alvarez et al., 2009; Benešová et al., 2012). Only relatively abundant proteins within a limited pI range (e.g., 3–10) can be detected by standard 2DE, whereas the iTRAQ method allows for the analysis of proteins present in low quantities and ones that tend to be difficult to separate by 2DE. However, a key question is that many abundant proteins detected by 2DE cannot be detected by iTRAQ. This is probably due to the lack of distinction of protein isoforms when the ratios are quantified by iTRAQ. Most of current software for iTRAQ (except for ProteinPilot) cannot discriminate abundance changes of different isoforms; therefore, if a protein abundance change resulted from the increase/decrease of a certain isoform, or the presence/absence of a certain PTM(s), TABLE 1 | Examples of detection of stress-responsive proteins in maize by 2DE and/or iTRAQ proteomic approaches.


the iTRAQ ratio may not show significant change. It is obvious that iTRAQ has also considerable deficits in differential protein detection.

Currently, the identification of stress-responsive proteins in crop plants is poorly overlap among different groups, even using state-of-the-art instrumentation. Except for differences in plant genotype, growth, and stress conditions, the inconsistent or unreliable results regarding identification of stress-responsive proteins mainly originate from erroneous methodology. In particular, three prominent problems affect the accuracy and reliability of proteomic data.

The first problem is inappropriate protein-extraction methods for plant tissues. Compared to model plant Arabidopsis, crop plants are more problematic in protein extraction, because they contains large amounts of secondary compounds such as phenolics, lipids, and organic acids, which severely interfere with protein extraction and proteome analysis (Wu et al., 2014a). Generally, protein-extraction methods need to be optimized and improved dependent on plant species and tissue types. Due to the great variance in sets of secondary metabolites present in various tissues from diverse plant species, no single extraction protocol is effective for every tissue. Adult tissue is usually more problematic than young tissue. For a given tissue, it is recommended that protein extraction starts with simple TCA/acetone precipitation and/or phenol extraction, and is then modified accordingly. Previously, we introduced some cases of protein extraction methods from representative plant tissues for proteomic analysis (Wang et al., 2008). The specific methods used in different labs will cause the disagreement in proteomic results, even for the same tissue and/or treatment.

The second problem is the poor quality of the original proteomics data, especially 2D gels, which is usually due to methodological issues during protein extraction and analysis, e.g., incomplete extraction, interference of non-protein substances, incomplete focusing, and incorrect spot matching. Generally, 2DE is performed manually. Poor-quality 2DE maps will result in erroneous or inconsistent results. Although these common problems intrinsic to 2DE are well known to the proteomics research field, novices in the plant proteomics community should make efforts to improve protein extraction and analysis, dependent on specific experiments, and crop species.

The third problem is insufficient replicates in 2DE or iTRAQ analysis. Many studies using 2DE or iTRAQ analysis have claimed to include three or more independent biological replicates, but these studies did not provide the relevant figures or data. Alvarez et al. (2009) indicated that the quality of iTRAQ results depends on both the number of biological replicates and the number of sample injections. In iTRAQ analysis, despite the application of quality assurance protocols, most errors occur during the pre- and analytical phases. Commercial iTRAQ services quote a price of approximately \$5000–10,000 for a single iTRAQ analysis of four to eight samples. It is conceivable that iTRAQ analysis sometimes lacks sufficient and necessary biological or technical replicates due to expense. Another possibility is the consequence of the pressure exerted by the well-known "publish or perish" dilemma, which often results in the rapid and careless publication of data (Fernández-Marín et al., 2015).

In some instances, differential protein changes in abundance were not as significant as reported. Kim et al. (2015) detected 29 differentially abundant spots in maize leaves under drought stress. However, upon comparing the relative abundance of the differentially expressed proteins, we found that only 10 proteins showed 1.5–2.0-fold changes in abundance, whereas the other seven proteins showed no obvious changes in abundance. Therefore, authors should perform more careful and thorough checking of stress-induced differentially abundant proteins before publication. The novice in plant proteomics also should pay attention to the articles published in the journals by those socall predatory publishers (https://scholarlyoa.com/publishers/).

#### CONCLUDING REMARKS

As discussed above, due to the weakness in the quality of proteomic data and the constraints on biological and technical replicates, it is not surprising that few commonalities and limited biological significance can be drawn from the numerous studies from different groups regarding plant stress proteomics.

To improve the accuracy of detection of stress-responsive proteins in plants, novices in the plant proteomics community must give higher consideration to sample preparation prior to gel-based or gel-free proteomics analysis. The quality of 2DE maps is very straightforward, so many protein extraction protocols have been reported based on a 2DE evaluation, whereas almost none have been reported for protein extraction protocol evaluation by iTRAQ analysis. Recently, we reported in detail a universal protein extraction protocol integrating TCA/acetone precipitation with phenol extraction (Wu et al., 2014b). This protocol made it possible to obtain satisfactory 2DE maps of various crop plant tissues, and it could be suitable for gelfree approaches. In addition, organelle isolation and/or protein fractionation techniques during sample preparation can improve the depth of proteome analysis through reducing proteome complexity.

Another aspect to consider for improving the accuracy of detection of stress-responsive proteins in plants is that the publication of proteomic data should describe biological and technical replicates and provide the necessary proteomic data on the replicates. As proteomics (especially the quantitative approach) is a statistically based method that relies on probability and arbitrary thresholds, there is always the chance of reporting false positives. To obtain proteomic data of confidence, the biological replicates should be at least three times, with three technical replicates in an independent biological experiment.

Moreover, proteomic experiments should be conducted within financial constraints to allow sufficient biological and technical replicates to increase the confidence of the proteomics data. When experiments are designed and performed properly, the technical variation should be comparable between methods, and the results show good agreement and biological significance.

Finally, experimental validation is often required to increase the confidence of the proteomic results, which can be carried out by qRT-PCR or transcriptomic analysis, or can be verified by using specific antibodies through immunochemistry, or directly measuring the changes of enzyme activity.

### AUTHOR CONTRIBUTIONS

XW drafted the paper and WW edited the paper.

### REFERENCES


### ACKNOWLEDGMENTS

Supporting from the plan for scientific innovation talent of Henan province (144200510012) and the program for innovative research team (in science and technology) in University of Henan province (15IRTSTHN015) are acknowledged.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Wu and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Role of Proteomics in Crop Stress Tolerance

Parvaiz Ahmad1,2 \*, Arafat A. H. Abdel Latef3,4, Saiema Rasool<sup>5</sup> , Nudrat A. Akram<sup>6</sup> , Muhammad Ashraf2,7 and Salih Gucel<sup>8</sup>

<sup>1</sup> Department of Botany, Sri Pratap College, Srinagar, India, <sup>2</sup> Department of Botany and Microbiology, King Saud University, Riyadh, Saudi Arabia, <sup>3</sup> Department of Botany, Faculty of Science, South Valley University, Qena, Egypt, <sup>4</sup> Department of Biology, College of Applied Medical Sciences, Taif University, Turubah, Saudi Arabia, <sup>5</sup> Department of Botany, Jamia Hamdard, New Delhi, India, <sup>6</sup> Department of Botany, Government College University, Faisalabad, Pakistan, <sup>7</sup> Pakistan Science Foundation, Islamabad, Pakistan, <sup>8</sup> Centre for Environmental Research, Near East University, Nicosia, Cyprus

Plants often experience various biotic and abiotic stresses during their life cycle. The abiotic stresses include mainly drought, salt, temperature (low/high), flooding and nutritional deficiency/excess which hamper crop growth and yield to a great extent. In view of a projection 50% of the crop loss is attributable to abiotic stresses. However, abiotic stresses cause a myriad of changes in physiological, molecular and biochemical processes operating in plants. It is now widely reported that several proteins respond to these stresses at pre- and post-transcriptional and translational levels. By knowing the role of these stress inducible proteins, it would be easy to comprehensively expound the processes of stress tolerance in plants. The proteomics study offers a new approach to discover proteins and pathways associated with crop physiological and stress responses. Thus, studying the plants at proteomic levels could help understand the pathways involved in stress tolerance. Furthermore, improving the understanding of the identified key metabolic proteins involved in tolerance can be implemented into biotechnological applications, regarding recombinant/transgenic formation. Additionally, the investigation of identified metabolic processes ultimately supports the development of antistress strategies. In this review, we discussed the role of proteomics in crop stress tolerance. We also discussed different abiotic stresses and their effects on plants, particularly with reference to stress-induced expression of proteins, and how proteomics could act as vital biotechnological tools for improving stress tolerance in plants.

#### Keywords: drought, nutrition, plants, proteomics, salts, temperature

### INTRODUCTION

As the population of the world increases exponentially, the agriculture sector worldwide is facing a major challenge of ensuring a sufficient food supply to the masses through enhancing agricultural productivity (FAO, 2012). This challenges further intensified with alterations in weather patterns due to changes in climate that impact crop productivity (Ahmad et al., 2012, 2013; Lake et al., 2012). Adverse environmental conditions alter agro-ecological system that may affect the demand for increased agricultural production (Lake et al., 2012; Ahmad et al., 2013). Of various abiotic stresses, drought, salinity, temperature (freezing/heat), light intensity, and heavy metal contamination are the most prevalent that considerably retard not only plant production, but also the quality of crops (Ahmad et al., 2008, 2010, 2012; Ashraf et al., 2011; Ashraf, 2014; Qadir et al., 2014) (**Figure 1**).

#### Edited by:

Dipanjana Ghosh, National University of Singapore, Singapore

#### Reviewed by:

Georgia Tanou, Aristotle University of Thessaloniki, Greece Kamrun Nahar, Kagawa University, Japan

> \*Correspondence: Parvaiz Ahmad parvaizbot@yahoo.com

#### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 24 March 2016 Accepted: 18 August 2016 Published: 08 September 2016

#### Citation:

Ahmad P, Abdel Latef AAH, Rasool S, Akram NA, Ashraf M and Gucel S (2016) Role of Proteomics in Crop Stress Tolerance. Front. Plant Sci. 7:1336. doi: 10.3389/fpls.2016.01336

Plants being sessile in nature during their course of evolution have developed highly sophisticated and effective strategies to counteract environmental cues (Ahmad et al., 2008, 2010, 2012; Haggag et al., 2015). Also they need to adapt quickly to overcome these stresses during their short lifespan. Stress as it is understood today is a factor that alters normal functioning of a number of mechanisms in an organism. During a stress, these mechanisms are up-regulated at various levels of molecular, morphological, and physio-biochemical responses (**Figure 1**). As the stress comes under control, homeostasis is reestablished. In plants under stress conditions, signaling of kinase cascades, ion channels, accumulation of reactive oxygen species (ROS), and hormones are activated (Ahmad et al., 2010, 2012, 2013; Ashraf et al., 2014; Rejeb et al., 2014; Ziogas et al., 2015; Molassiotis et al., 2016).

Recent technological advances have seen the development of the "omics" technologies which are being applied in plant sciences to identify key proteins or metabolites, are novel covering metabolomics, proteomics and, genomics responsible for plants stress tolerance and also the genes regulating such biomolecules (Ahmad et al., 2013; Shelden et al., 2013; Srivastava et al., 2013; Emon, 2016). Application of these omics facilitates a direct observation of the agents affecting plant development. Proteomics deals with determination, identification of proteins, expression profile, post-translational modifications (PTMs), and protein–protein interactions under stress and non-stress conditions (Hashiguchi et al., 2010; Nam et al., 2012; Mertins et al., 2013; Ghosh and Xu, 2014) (**Figure 1**). A notable change in protein expression always takes place in plants under abiotic stresses, so proteomic approach will be very useful in elucidating the role of protein accumulation under stress conditions and its association with stress tolerance (Witzel et al., 2009; Hossain et al., 2012; Perez-Clemente et al., 2013). Proteomic studies in plants under abiotic stresses are well documented, for example: salt stress (Nam et al., 2012; Zhu et al., 2012), drought stress (Castillejo et al., 2008; Caruso et al., 2009; Mirzaei et al., 2012; Mohammadi et al., 2012; Cramer et al., 2013), waterlogging (Komatsu et al., 2009, 2010, 2012, 2013a,b; Alam et al., 2010a,b), and heat stress (Rollins et al., 2013; Xuan et al., 2013) etc.

Proteomics approach is used to investigate the responses of plants to stresses as well as complexity of biochemical processes (Aghaei and Komatsu, 2013; Ghosh and Xu, 2014; Gong et al., 2015). Plant stress proteomics has the ability of identifying possible candidate genes that can be used for the genetic enhancement of plants against stresses (Cushman and Bohnert, 2000; Rodziewicz et al., 2014; Barkla et al., 2016).

Different signaling pathways are reported to be activated in response to stresses resulting in a complex regulatory network involving transcription factors, ion homeostasis, antioxidants, hormones, kinase cascades, ROS, and osmolyte synthesis (Suzuki et al., 2014; Yin et al., 2015). However, the responses of plant cells to abiotic stresses vary in different organs. Organ-specific proteomics combined with subcellular organelle proteomic studies of developmental mechanisms from leaf to root can provide more detailed information for understanding of cellular mechanisms that regulate stress response and signal transduction in various organelles, and they could be used to enhance crop stress tolerance (Komatsu and Hossain, 2013; Yin et al., 2015). Advances in proteomic technologies have widened our genetic and molecular understanding of plant responses under abiotic stresses. However, the main purpose of this review is to present a critical overview of the recent approaches that promise to enhance the tolerance of the plants with a minimum crop loss under different types of abiotic stresses. A detailed study on the modulations or extent of modulations in proteomics of different plants under salinity, water and/or nutrient deficiency, low/high temperature stress as well as under waterlogging conditions have been explained in the present review. Furthermore, how these studies have contributed to crop improvement under abiotic stresses is summarized below.

#### PROTEOMICS APPROACH

#### Salinity Tolerance

fpls-07-01336 September 6, 2016 Time: 12:3 # 3

Within the next 25 years, salinization of arable lands may result in 30% land loss, and up to 50% by the year 2050 (Yan et al., 2003; Wang et al., 2008; Abdel Latef and Chaoxing, 2014; Ma et al., 2015). Thus, improvement of salt tolerance in crops by genetic engineering is contemplated as one of the powerful tools for overcoming the salinity problem world-over (Ashraf and Akram, 2009; Gupta and Huang, 2014).

Plant responses to salt stress through proteomics approach have been studied in both glycophytes and halophytes. Plant biologists have worked with the model plants under saline stress at the proteomic levels, e.g., Jiang et al. (2007) in Arabidopsis thaliana, Razavizadeh et al. (2009) in Nicotiana tobaccum, Chen et al. (2011) in Populous cathayana, Chattopadhyay et al. (2011) in grasspea, and Xu et al. (2011) in Agrostis stolonifera. Moreover, agricultural plants have also been investigated under saline stress in different studies, e.g., durum wheat (Peng et al., 2009; Jacoby et al., 2010), canola (Bandehagh et al., 2011), sugarbeet (Wakeel et al., 2011), soybean (Sobhanian et al., 2010), peanut (Jain et al., 2006), S. bicolor (Swami et al., 2011; Ngara et al., 2012), maize (Zörb et al., 2009, 2010), tomato (Chen et al., 2009; Manaa et al., 2011), potato (Aghaei et al., 2008), and cucumber (Du et al., 2010) etc.

Apart from glycophytes, halophytes have also been analyzed for the proteomic analysis under salt stress by different workers, e.g., Katz et al. (2007) in Dunaliella salina, Wang et al. (2008) in Physcomitrella patens, Barkla et al. (2009) in Mesembryanthemum crystallinum, Tada and Kashimura (2009) in Bruguiera gymnorhiza, Wang et al. (2009) in Salicornia europaea, Geissler et al. (2010) in Aster tripolium, Pang et al. (2010) in Thellungiella salsuginea, Li et al. (2011) in Suaeda salsa, and Yu et al. (2011) in Puccinellia tenuiflora.

Salt stress can impose its negative effect first on plant roots, because there is an evidence that some salt stress responsive genes and proteins are induced in roots than in shoots (Yan et al., 2003; Hasanuzzaman et al., 2013a). This statement has been authenticated by different workers in soybean (Sobhanian et al., 2010), rice (Liu et al., 2012), wheat (Guo et al., 2012), maize (Zörb et al., 2010), and potato (Aghaei et al., 2008). Proteome of soybean was studied under salt stress by using different tissues (Aghaei et al., 2009; Sobhanian et al., 2010; Ma et al., 2012). They identified 50S ribosome protein which was down-regulated in leaves. This protein is believed to participate in the biosynthesis of soybean protein and causes decrease in plant growth.

Chitteti and Peng (2007) studied changes in the phosphoproteme of roots of rice on exposure to NaCl (150 mM) for a few hours by using Pro-Q Diamond stain. They found 20 proteins upregulated and 18 downregulated. They positively identified 17 of the 20 upregulated proteins and 11 of the 18 downregulated ones. Proteins such as GST, Hsp70, and mannose binding rice lectin up-regulated, while protein kinase, ATP synthase beta-chain, GALP hydrogenase down-regulated. They believed that phosphorylated proteins could be identified using Pro-Q Diamond stain under saline conditions. Of all proteins, 17 overexpressed proteins were responsive to salinity, however, some other proteins identified did not express in any of the proteomic reports on rice on exposure to salinity. All these reports along with some other are listed in **Table 1**.

Following are the major mechanisms that are directly and/or indirectly interlinked with proteins, and proteomics can play a role in regulating them.

#### Photosynthesis

Up-/down-regulation of different proteins predominantly affects photosynthesis by maintaining protein biosynthesis, energy metabolism and detoxification under saline conditions (Rollins et al., 2013; Zhao et al., 2013). Therefore, Parker et al. (2006) attributed the inhibitory effect of NaCl on RuBisCO activase (chaperone protein) which was down-regulated and may be the reason for declined photosynthetic activity in soybean under NaCl stress (Sobhanian et al., 2010). Up-regulation of the 20 kDa chaperonin plays an efficient role in shielding proteins of soybean under salinization (Sobhanian et al., 2010). Furthermore, Zhang et al. (2012) studied the proteomics of 34 different plant species subjected to salinity. They identified 2171 salt responsive proteins and categorized on the basis of cell structure, energy metabolism, CO<sup>2</sup> assimilation/carbohydrate synthesis, stress and defense interaction, transcription, translation, protein transport and folding, cell division/differentiation and fate and many others with unknown functions.

Recently, Wang et al. (2014) reported 53 differentially expressed protein spots on 2DE maps in Kandelia candel subjected to varying salt levels. The results showed the upregulation of proteins involved signal transduction, Na<sup>+</sup> compartmentalization, photosynthesis, protein folding, and respiration. This protein upregulation was reported to be responsible for enhanced salt tolerance in K. candel. Zhu et al. (2012) reported 23 salt responsive proteins in B. gymnorrhiza (a halophyte) under salt stress. Ten proteins were reported to be involved in photosynthesis, antioxidative system, protein folding, and cell organization. They also reported different protein expressions at 200 and 500 mM salt stress. At 200 mM salt, overexpression of enzymatic antioxidants and photosynthesis related proteins led to improved plant growth as well as salt tolerance in B. gymnorrhiza. B. gymnorrhiza is able to sustain severe salt stress due to the upregulation of protein folding and degradation related proteins and cell wall organization related proteins. In salt stressed photosynthetic Chlamydomonas, 3115 proteins were identified, out of which RuBisCO was the most prominent one in the cell (Mastrobuoni et al., 2012).

From the work carried out with different plant species, it is evident that a variety of proteins involved directly/indirectly in the process of photosynthesis are up-/down-regulated under saline conditions. However, it is yet unknown that what types of proteins are specifically involved in all photosynthetic plant species, as protein types and their expression vary from species to species. So, the identification of some specific proteins involved

#### TABLE 1 | Identification and specific roles of different proteins in salt tolerance.


DHAR, dehydroascorbate reductase; HSPs, heat shock proteins; PRPs, pathogen-related proteins; LEA, late embryogenesis-abundant; NHX1, Na+/H+antiporter; APX, ascorbate peroxidase; SOD, superoxide dismutase; ABA, abscisic acid; STH2, Salt tolerance homolog 2.

in salinity tolerance and the extent of their expression in different plants could be helpful in improving crop salt tolerance using modern molecular tools.

#### Late-Embryogenesis Abundant (LEA) Proteins

The LEA proteins are synthesized late during embryogenesis in plant seed development. They have also been reported in vegetative plant tissues under different environmental stresses (Xu et al., 1996; Hand et al., 2011; Amara et al., 2012; Battaglia and Covarrubias, 2013). While investigating changes in LEA proteins in rice plants, Chourey et al. (2003) identified four saltinduced LEA proteins which accumulated under salinity, but were degraded when the stress was over. Aghaei et al. (2009) so showed increased expression of LEA proteins under salinity stress in soybean. Xu et al. (1996) introgressed rice plants by HVA1 (a LEA protein gene) isolated from barley. It was found that modified rice plants showed better growth than wild plants under salt stress. Increase in stress tolerance of transgenic rice plants was associated with the high accumulation of the HVA1 protein. Thus, LEA proteins associated genes could be used significantly as a molecular tool for crop improvement under stress conditions.

#### Oxidative Stress/Antioxidants

While studying the role of different types of proteins in rice cultivars/lines with differential stress tolerance Salekdeh et al. (2002) studied the proteome analysis in the roots of two rice cultivars exposed to NaCl (50 and 100 mM). They identified three proteins: caffeoyl-CoA O-methyltransferase (CCOMT), involved in lignin biosynthesis, auxin and salicylic acid response (ASR1) like protein and APX (ascorbate peroxidase). It is worthy to mention that the tolerant rice cultivar had higher amount of ASR1-like protein and CCOMT than in the salt sensitive cultivar, while both cultivars responded almost uniformly to oxidative stress using increased APX. Thus, constitutively expressed proteins improved salt tolerance of the relatively salt tolerant rice cultivar (Vincent and Zivy, 2007)

Witzel et al. (2009) used 2-D gel electrophoresis to compare the root proteomes of two barley cultivars differing in salinity tolerance under salinity stress. Out of total 39 proteins, 26 proteins were isolated using MS (mass spectrometry). In the tolerant cultivar, proteins improved detoxification of ROS due to high accumulation of glutathione. By contrast, in the sensitive one, Fe absorption proteins increased under saline stress.

Yan et al. (2003) found salinity-induced changes in more than 1100 proteins of the proteome of roots in rice. Twelve different proteins were identified, three of these proteins were identified as enolase, four as salt responsive proteins, and the remaining six were new proteins involved in regulating metabolism, nitrogen and carbon in rice for the removal of ROS and the stability of the cytoskeleton (Roveda-Hoyos and Fonseca-Moreno, 2011).

Guo et al. (2013) showed that 17 unique proteins differentially changed in abundance in response to NaCl in Arabidopsis roots. These identified proteins were believed to be involved in binding catalysts, signal transduction, transport, metabolism of cell wall and energy, and ROS scavenging and defense.

Overall, very few reports as mentioned under this section are available on protein (s) expression involved in oxidative defense system in plants under salinity stress. Furthermore, it is imperative to mention here that this knowledge is limited only to a very few proteins expressed during saline conditions, whereas not a single report is available on what type of proteins up- or down-regulate under saline conditions or what types of proteins are expressed individually or in combination involved in antioxidative system either enzymatic and/or nonenzymatic.

#### Ion Uptake/Homeostasis

fpls-07-01336 September 6, 2016 Time: 12:3 # 5

Understanding the proteins particularly root proteins involved in salt response has shown a significant role of these proteins in salt-tolerance mechanisms. Generally, Na<sup>+</sup> enters the plant roots via apoplastic or symplastic routes, which include many Na<sup>+</sup> transport transmembrane proteins such as Na+/H<sup>+</sup> antiporters and HKT (a high affinity K<sup>+</sup> transporter) (Rus et al., 2001; Tester and Davenport, 2013). Different proteins expressed under stress or non-stress conditions are cultivar specific, while some others depend on the level and duration of salinity stress (Peng et al., 2009; Szopinska et al., 2011). For example, Nohzadeh et al. (2007) performed the proteome investigation of plasma membrane of rice cv. IR651 under saline conditions. The 24 different proteins identified were found associated with protein–protein interaction and signaling (reorins and 14-3-3 protein), and proteins involved in controlling K ion channels (Hashiguchi et al., 2010).

Peng et al. (2009) compared the leaf and root proteome of two wheat cultivars under saline stress conditions. They found that majority of the proteins expressed under stress conditions were cultivar specific, while some others were stress responsive. They suggested that improved salinity tolerance in wheat cv. Shanrong No. 3 was associated with more e-flux of toxic byproducts as well as ionic/osmotic homeostasis.

While working with yeast exposed to 0.4 and 1.0 M NaCl for time intervals of 10, 30, and 90 min, Szopinska et al. (2011) identified 88–109 plasma membrane (PM) proteins. Of which, 12 plasma membrane proteins were expressed at mild salt stress (0.4 M) including some already known and some newly target salt-responsive proteins. However, at both salt levels, 20 PM proteins were down-regulated including ABC and/or amino acid transporters, cell wall biogenesis proteins, Pma1, t-SNAREs, and P-type H+-ATPase. They found that this protein internalization could be due to alteration in ionic homeostasis or plasma membrane morphology.

Overall, although a number of various types of proteins have been found to be up- and down-regulated in different plant species on exposure to saline regimes, more analyses for identification of different stress responsive proteins and their respective genes involved in different key metabolic pathways is necessary because such proteins could be used in improving stress tolerance in different potential crops using different biotechnological tools.

#### Drought Tolerance

Drought is one of the major abiotic constraints that can considerably reduce plant growth and crop yield (Ashraf et al., 2011; Ahmad et al., 2013; Akram and Ashraf, 2013). The reduction in crop yield under drought stress is attributed to water stress-induced osmotic stress, nutritional and hormonal imbalance, activation of oxidative system as well as disturbance in different plant biochemical processes including reduction in carbon uptake through photosynthesis (Hashiguchi et al., 2010; Ashraf et al., 2011; Chugh et al., 2011; Sharma et al., 2012). Shinozaki and Yamaguchi-Shinozaki (2007) reported that there are common genes that are induced during stress in species such as Arabidopsis and Oryza sativa. Comparative analysis of drought stress using microarrays showed that stress-induced genes of Arabidopsis and rice showed a similarity between the two genomes at the molecular level. About 51 genes were identified in Arabidopsis, and 73 were reported with similar function in O. sativa (Roveda-Hoyos and Fonseca-Moreno, 2011). Larrainzar et al. (2007) studied the proteome of roots of Glycine max seedlings under water deficit. They observed a total of 45 proteins under drought stress, only two of them were new proteins. The expression of five proteins was found to be upregulated and that of 21 proteins downregulated. Under the recovery from drought after rewatering soybean plants for 4 days, the concentration of proteins was similar to the control levels. Caruso et al. (2009) studied the proteomics in wheat (T. aestivum) under water stress. They detected 36 protein spots out of which 12 proteins were upregulated and 24 down-regulated. Ke et al. (2009) also studied the proteomics in rice under water deficit conditions using the techniques of 2-DE and mass spectrometry, MALDI-TOF. They detected 18 proteins, and out of these, 12 were up-regulated and 12 down-regulated.

In response to desiccation, Chen et al. (2011) demonstrated the proteomic profiling of seeds of recalcitrant tea. The results showed that 23 proteins up-regulated under desiccation involved in defense against stress as well as redox status under the stress. In another study, wheat plants subjected to drought stress showed 15 bands and out of these 8 protein types were determined to be potential complex forming protein (Zhang et al., 2014). Higher expression levels were found in many proteins of wild genotypes of wheat in response to drought stress. Out of them, 11 protein spots with low peptide matches were identified as candidate unique drought responsive proteins (Budak et al., 2013). Yang et al. (2013) studied protein expression in root of common bean subjected to osmotic stress. They reported 22 proteins differentially regulated by osmotic stress. About 70% of the total expressed proteins were associated with metabolic pathways, such as carbohydrate and amino acid metabolism. Osmotic stress reduced the level of five proteins and increased that of other seven proteins in apoplast (Yang et al., 2013). Sunflower inbred lines (drought tolerant and drought sensitive) were subjected to drought stress and results showed that root metabolism involved proteins that declined in both tolerant and sensitive lines (Ghaffari et al., 2013). The defense related proteins were up-regulated in tolerant lines and down-regulated in sensitive lines (Ghaffari et al., 2013).

Several proteins were found up/down-regulated in different plant species as mentioned in above reports on exposure to drought stress. Some of the important mechanisms perturbed or therein proteins involved for playing their roles in improving drought tolerance in different plant species are listed below such as:

#### Photosynthesis

fpls-07-01336 September 6, 2016 Time: 12:3 # 6

It is well known that the efficiency of photosynthesisdepends on number and activity of proteins (Deeba et al., 2012; Galvan et al., 2013). A study carried-out by de Almeida et al. (2013) on two varieties of sugarcane RB 72910 (drought tolerant) and RB 943365 (drought sensitive) showed change in protein expression. The expression of some proteins in RB 72910 was up-regulated and that in some others down-regulated. However, in cv. RB 943365 all the proteins showed down-regulation. These proteins were associated with functions such as photosynthesis, signal transduction and regulation processes. Valero-Galván et al. (2013) showed a decrease in protein abundance/expression mainly those involved in ATP synthesis and photosynthesis upon water withholding/water deficiency in holm oak (Quercus ilex).

The down-regulation of glycolytic enzymes under osmotic stress may be a strategy for accumulating sugars as an energy source for attaining enhanced growth after recovery of drought stress. Hajheidari et al. (2005) while working with sugar beet leaves identified different proteins involved in redox regulation, photosynthesis (Rubisco), oxidative defense system, and chaperone treated with drought. Recently, Gil-Quintana et al. (2013) have shown that on exposure to drought stress, cell growth of soybean plants reduced due to reduction of many proteins having a role in amino acid metabolism, carbon metabolism, as well as protein synthesis. In an earlier study, Aranjuelo et al. (2011) while carrying out proteomic analysis of alfalfa plants have shown that decreased carboxylation activity due to water shortage was associated with decrease in Rubisco protein content, its activation as well as regeneration. Furthermore, drought-induced reduction in amino acids (glutamic acid and asparagine) showed that N availability was also limited due to decline in nitrogenase activity while increase in that of proteases. In cotton plants, although Deeba et al. (2012) examined that 16 protein spots were up-regulated while 6 down-regulated, however, an additional information on the molecular basis of drought intolerance in cotton plants still needs to be determined. Rollins et al. (2013) determined drought-induced changes in the barley leaf proteome using mass spectrometry and differential GE. They showed that although water stress induced a substantial decrease in plant biomass and yield production of barley, photosynthetic efficiency as well as proteomics remained unchanged due to drought stress.

From the above-mentioned all reports it can be concluded that plant photosynthetic efficiency was adversely affected by drought-induced decrease in activities/levels of different proteins, particularly those of rubisco. Furthermore, many other proteins were also visualized that were markedly affected by drought stress, but their characterization is still underway.

#### Late-Embryogenesis Abundant (LEA) Proteins

Late-embryogenesis abundant proteins are water soluble proteins that are synthesized in high concentration in desiccation-tolerant plant (Alam et al., 2010a,b). The accumulation of dehydrin and ferritin were identified in proteomic investigation of soybean roots under drought stress (Alam et al., 2010a,b; Nouri et al., 2011). Dehydrins are LEA proteins, can effectively improved plant growth under stress by reducing the harmful effect of ROS (Grelet et al., 2005; Nouri et al., 2011; Hossain et al., 2013). Grelet et al. (2005) identified a LEA protein, PsLEAm localized within the matrix space of seed mitochondria of Pisum sativum. PsLEAm shows characteristic of LEA proteins and usually expresses during late seed development. It could not be detected in vegetative tissues, but on exposure to severe water stress, it expresses in leaves. The authors suggested that under drought stress conditions in an in vitro assay, it can play a beneficial role during desiccation. In Medicago truncatula seeds, about 15 polypeptides were reported to be expressed under drought stress, which were significantly associated with drought tolerance of the crop. Of all, 11 polypeptides were identified as LEA proteins including MtEm6, isoform of dehydrins, MtPM25, MP2, PM18, and all isoforms of SBP65. Alteration in the abundance of MtEm6 and MtPM25 in imbibed M. trunculata seeds during the loss of drought tolerance and in developing embryos during the acquisition of drought tolerance confirmed the involvement of these two proteins in drought tolerance (Boudet et al., 2006). Functional diversity among LEA proteins was confirmed in maize by Amara et al. (2012). Protein visualization showed that cells expressing LEA protein Mg3–GFP, were better in controlling cell shrinkage. Another potential mitochondrial LEA protein LEAM was found to be expressed in seeds and this was reversibly folded into α-helices upon water shortage. Generally, this LEA protein protects liposomes by interacting with membranes under water deficit conditions which then protects the inner mitochondrial membrane under desiccation.

Although a variety of LEA proteins have been found to be overexpressed in different plant species under drought stress and have shown their potential role in drought tolerance, the mechanisms of protection from drought are still being researched.

#### Oxidative Stress/Antioxidants

Plants accumulate antioxidants to counteract stress-induced ROS. For example, upregulation of superoxide dismutase (SOD), an ROS scavenger, was reported in soybean (Toorchi et al., 2009) and rice (Ali and Komatsu, 2006) under drought stress. The study of Caruso et al. (2009) identified 21 different proteins including some isoforms and subunits of enzymes under water stress. They reported that 18% of the identified proteins were associated with the routes of glycolysis and gluconeogenesis, 15% proteins were associated with the removal of ROS, 12% in biosynthesis of amino acids, 9% in the Calvin cycle, 6% with defense mechanisms, and the remaining 3% related to posttranscriptional regulation. Benešová et al. (2012) have shown that drought stress induced up-regulation of stress-related protective proteins namely chaperones and dehydrins in two differential tolerant cultivars of maize. The alteration in the concentrations of different detoxification proteins significantly associated with the enzymatic antioxidants, generally lower in the sensitive maize cultivar due to reduced level of proteosynthesis and changes in the translation machinery. In grapevine (Vitis vinifera L.)

Cramer et al. (2013) have reported that proteomic responses to water stress generally involved abundance of proteins for translation, as well as steroid and antioxidants metabolism. Recently, Das et al. (2016) investigated the effect of heat and drought alone or combine in two soybean varieties, Surge and Davison using 2D-DIGE proteomic technique. They found that photosynthesis-related proteins affect RuBisCO regulation, electron transport, Calvin cycle, and carbon fixation under these stresses. In addition, carbonic anhydrase accumulation in the cell helps the cell to become more resistant to cytotoxic concentrations of hydrogen peroxide. While working with sunflower proteomics, Ghaffari et al. (2013) examined that on exposure to drought stress, defense/disease and energy involved proteins reduced significantly in the relatively less tolerant sunflower cultivar, while they increased in the tolerant one. They suggested that better water transport, energy usage, and antioxidant defensive system are essential mechanisms for regulating plant growth under water limited environment. In drought-stressed creeping bentgrass (A. stolonifera L.) plants, fifty-six stress-responsive proteins visualized, of which some proteins those were participated in C and N metabolisms were suppressed due to drought stress. However, glutathione-S-transferase, APX and CAT (antioxidant enzyme) proteins were up-regulated in a relatively drought tolerant cv. Penn-A4, which suggests that proteins have an effective role in drought tolerance by maintaining cell turgor, membrane stability, cell wall expansion and regulation of ROS defensive system under drought stress (Xu and Huang, 2012). It has been identified by Tanou et al. (2012) that local and systemicH2O2 oxidative and NO-nitrosative bursts involved in encoding proteins associated with H2O<sup>2</sup> production such as NOX, Fe SOD, Cu/Zn SOD, and Mn SOD as well as NO biosynthesis (e.g., NOS, NiR, and NR) after 8 days of salinity stress. Recently, Yin et al. (2015) reported that exogenous application of calcium increased the salinity tolerance of soybeanseedlings by promoting protein biosynthesis, inhibiting proteolysis, redistributing storage proteins, regulating protein processing in endoplasmic reticulum, and enriching antioxidant enzymes and activating their activities. Several signaling pathways activated in response to multiple stresses have been revealed in transcriptome, metabolome, and proteome analyses of different plants subjected to different stresses, resulting in a complex regulatory network involving antioxidants, hormones, transcription factors, kinase cascades, ROS, and osmolyte synthesis (Suzuki et al., 2014; Yin et al., 2015). However, differentially adapted species vary in their response to stressinduced oxidative stress. A reasonable number of reports are available in the literature on drought-induced increase or decrease in the levels of a number of enzymatic and nonenzymatic antioxidants. Mostly an increase in ROS served as a signal, which triggered a biochemical response to establish the redox balance of the cell. Identification of a number of PTMs is an important feature of the proteomic approach in response to oxidative stress involves which have been dicussed but have rarely been the focus of studies investigating the response to environmental stress. The implification of proteomics for the functional analysis of plants will benefit from advances in plant phenotyping particularly automated, non-invasive phenotyping of plant collections will assist in characterizing the relevant traits for future crop breeding.

#### Abscisic Acid (ABA) Metabolism

Abscisic acid (ABA) plays a major role in plant response to drought stress as it controls the closing of stomata to minimize water loss. Previously, Zhao et al. (2008) studied a guard cell proteome under drought stress in A. thaliana. They detected 336 new proteins (not detected before in guard cell) with fiftytwo proteins involved in signal transduction. Of these proteins, the myrosinase TGG1 was associated with ABA metabolism and stomatal regulation (Hashiguchi et al., 2010). Also, Nishikawa et al. (2008) stated that the modification of ABA signaling plays a role in drought tolerance of plants. In their work on proteome analysis of Arabidopsis under drought stress, they found that the improvement of fresh weight in Arabidopsis under water stress was through suppression of water vapor loss from stomata. This vapor loss was associated with the high level of sphingosine-1-phosphate. It is worthy to mention that, ABA controlled the sphingosine-1-phosphate level through sphingosine kinase (Hashiguchi et al., 2010)

Recently, Alvarez et al. (2014) have observed significantly more expression of proteins in two differential wheat cultivars, Nesser and Opata. They observed a comparatively higher number of ABA-responsive proteins in the roots of wheat cv. Nesser as compared to those in cv. Opata, which confirmed the role of these ABA responsive proteins in enhancing drought tolerance. ABA has a well known protective role in stomatal closure under water limited environment, but very rare information is available in the literature on proteomic analyses relevant to ABA accumulation. In addition, how and what type of proteins are up- or downregulated and their association with drought-induced increase in ABA still needs to be elucidated.

Overall, a number of reports are available in the literature on identification of different proteins in water stressed plants, but little information is available on their regulation and identification. Thus, there is a need to examine the regulation of all proteins being identified in different plant species exposed to water stress. In addition, the actual function of stress responsive proteins is not fully known which needs to explored with particular reference to plant tolerance mechanisms.

#### Temperature Tolerance High Temperature Stress

High temperature stress (heat stress) results in disturbance in cellular homeostasis and can cause drastic reduction in growth, development and even death of plants (Hasanuzzaman et al., 2013b; Brosché et al., 2014). High temperature induces the synthesis of high (60–110 kDa) and small (15–45 kDa) molecular mass HSPs in plants (Miernyk, 1997; Renaut et al., 2006). Lee et al. (2007) studied the proteomics analysis in rice leaves under heat stress. They identified 48 proteins on exposure to 12 to 24 h of high temperature versus control. Out of all identified proteins, 18 were HSPs including smHSPs, HSP100, HSP70, dnak-type molecular chaperone BiP and Cpn60. Zhu et al. (2006) reported that the induction of HSP70 in transgenic soybean plants by introgressing HsfA1 enhanced tolerance to high temperature

stress. Skylas et al. (2002) detected seven HSPs spots in relatively heat tolerant wheat cv. Fang as compared to heat sensitive wheat cv. Wyuna. High accumulation of HSPs in plants is generally associated with heat tolerance (Hashiguchi et al., 2010; Xu et al., 2011; Hasanuzzaman et al., 2013b). HSPs could be categorized into five different sub-groups depending on their molecular weight (Kosova et al., 2011).

Süle et al. (2004) identified S-adenosylmethioninesynthetase (proteins other than HSPs) as a tolerance marker in heattolerant and susceptible barley cultivars. In other studies, a high accumulation of eIF4F and eIF5A-3 (eukaryotic translation initiation factors) has been reported to induce cellular reorganization leading to PCD on a long-term exposure to high temperature (Zhang et al., 2010; Liu et al., 2013; Rollins et al., 2013).

Rollins et al. (2013) reported 99 proteins expressed differentially in barley under heat stress. These regulated proteins were associated with energy metabolism, photosynthesis, detoxification, and translation. Liu et al. (2013) studied leaf proteome in two cvs. King (heat sensitive) and Vista (heat tolerant) of S. splendens. The results revealed 1213 leaf proteome spots of large size. Of which, 33 proteins were differentially expressed when S. splendens plants were subjected to high temperature stress. These proteins regulate photosynthesis and protein processing under heat stress. Li et al. (2013) reported 81 over-expressed proteins involved in protein synthesis, storage, transport, in/out-flux, signal transduction as well as defensive system against diseases in alfalfa subjected to heat stress. The proteome study of Pinelliaternata leaves subjected to high temperature stress showed 600 protein spots, 7 of which down-regulated and 20 up-regulated (Zhu et al., 2013). Of the 24 proteins identified, maximum of them were sHSPs associated with chlorophyll biosynthesis, RNA processing, photosynthesis as well as protein denaturation/degradation (Zhu et al., 2013). Liao et al. (2014) studied the proteomics at early milky stage of rice grains after exposure to heat stress. The results of 2-DE revealed about 27 up-regulated proteins in rice grains, predominantly from the heat tolerant lines. Out of total 27 proteins, 25 differentially expressed proteins are involved in biosynthesis, energy metabolism, oxidation, heat shock metabolism, and regulation of transcription.

Under high temperature stress, majority of the proteins expressed fall under the heat shock proteins (small and large molecular weight), and their expression varies between heat stress tolerant and sensitive cultivars. Identification and introgression of heat stress tolerant proteins could be used in modern biotechnology tools for the improvement of stress tolerance in economically important crops all-over the world.

#### Low Temperature Stress

Temperature is a main environmental factor which affects growth, productivity and distribution of plants (Shah et al., 2011; Bita and Gerats, 2013; Zia et al., 2014; Zinn et al., 2014). The phenomenon of exposing plants to temperatures from 0 to 15◦C (non-freezing temperatures) is called cold stress or chilling stress (Renaut et al., 2004). Cold stress is associated with reduction of water absorption resulting in cellular desiccation (Sinclair et al., 2013). Also cold stress induces alteration in metabolites leading to an oxidative stres (Renaut et al., 2004; Kosova et al., 2011; Sinclair et al., 2013). The increased freezing tolerance by plants under low temperature is called cold acclimation (Renaut et al., 2004; Timperio et al., 2008; Kosova et al., 2011; Miura and Furumoto, 2013). It includes different changes in protein and gene expression as well as metabolites (Renaut et al., 2006; Miura and Furumoto, 2013).

Kawamura and Uemura (2003) studied plasma membrane proteome of Arabidopsis under low temperature. They detected 38 proteins, 27 of which were soluble, whereas 15 insoluble. Imin et al. (2004) reported 70 proteins including 12 new (47 up- and 11 down-regulated) at pH between 4.0 and 7.0 in rice anthers exposed to cold stress (12◦C) for 48 h. A proteomic work on mitochondria of pea (Pisum sativum) reported by Taylor et al. (2005). Twenty, out of 33 proteins appearing in response to cold stress at 4◦C for 1.5 day. Cheng et al. (2009) in their study on proteomics of soybean seeds subjected to cold stress (4◦C), reported 40 proteins (25 up- and 15 downregulated). These proteins are involved in cell growth/division, storage, defense of cell, energy protein synthesis, transcription, and transport. Toorchi et al. (2009) studied proteomics analysis of soybean seedlings under cold stress for 2 days and observed overexpression of pathogenesis-related protein 1 (PR1), while down-regulation of caffeoyl-CoA 3-O**-**methyltransferase and PR10 proteins. Komatsu et al. (2009) identified a total of 12 N-glycosylated proteins in rice sheaths under cold stress. Of them, a calreticulin protein controlled phosphorylation and glycosylation in leaf sheathes under low temperature stress, indicating that calreticulin may regulate the expression of several other proteins (Hashiguchi et al., 2010).

LB-a is one of the freezing stress responsive low abundant proteins and identified as Hsp70 which decreased in rice plants exposed to low temperature stress (Hashimoto and Komatsu, 2007; Bashir et al., 2010). This decrease in LB-a might be attributed to cold-induced chloroplast degradation (Bashir et al., 2010). The proteomic analysis of plants under low temperature indicated an increase in enzymes involved in ROS scavenging, e.g., Imin et al. (2004) reported high accumulation of different isoforms of APX in tri-nucleate rice pollen under cold stress. Degand et al. (2009) showed that Cu/Zn-SOD abundantly increased in chicory roots under cold stress. While Kosmala et al. (2009) detected an enhancement in enzymes which take part in AsA and GSH metabolism in Festucapratensis plants. Many authors reported the accumulation of chaperonins (chaperonins 60 and 20) and HSPs (HSP90, HSP70) under low temperature stress (Kawamura and Uemura, 2003; Taylor et al., 2005).

Freezing injury is a result of drastic low temperature conditions (Perez-Munuera et al., 2009). Freezing temperature induces desiccation, and imbalances plasma membranes leading to formation of inverted hexagonal phase membrane structure (Sung et al., 2003; Timperio et al., 2008). Timperio et al. (2008) reported that anti-freezing proteins (AFPs) play a significant role in maintaining plant growth against freezing injury. AFPs are similar to the pathogenesis-related protein involved in

eliminating freezing stress and inducing disease resistance (Timperio et al., 2008). Winter cereal such as winter rye and wheat accumulated AFPS in their apoplast to tolerate freezing stress (Marentes et al., 1993).

Xu et al. (2012) reported a significant increase in relative abundance of antioxidant related proteins in low temperature tolerant wheat cv. Shixin 828 compared to those in sensitive cv. Shiluan 02-1. They also reported that carbohydrate metabolism related proteins were more abundant in cv. Shiluan 02-1. Xuan et al. (2013) studied the proteomics of Z. japonica (cold tolerant cv. Meyer) and Z. metrella (cold sensitive cv. Diomond) under cold stress. They showed that 700 proteins were resolved on 2-DE gels, but only 70 protein were considerably over-expressed. They suggested that of all identified proteins, 45 proteins were participated in cellular metabolic processes. Cultivar Meyer showed considerably high concentration/number of accumulated proteins as compared to cv. Diamond and only cv. Meyer showed 15 increased proteins under cold stress. The cold responsive proteins have been associated with the biosynthesis of carbohydrates, proteins, and nucleotides, ROS scavenging, proteolysis, protein folding, and energy storage.

#### CONCLUSIONS AND FUTURE PROSPECTS

Abiotic stresses like salinity, drought, high temperature, freezing stress, water-logging, and mineral toxicity and deficiency severely affect crop productivity and such losses are of major concern for all nations so as to cope with the increasing food demand. Abiotic stresses are known to hinder plant growth and yield by causing a variety of adverse effects including disturbance in regulation of many proteins involved in protein folding, ROS scavenging, proteolysis, metabolic energy supply, biosynthesis of carbohydrates and nucleotides, signal transduction, PCD, RNA processing, redox homeostasis, energy metabolism, secondary metabolites, glycolysis, lipid peroxidation, ethylene biosynthesis and cell wall loosening, etc. However, regulation of different proteins varies among species, therefore complete dissection of all proteins involved in different metabolic processes plant species under a variety of stresses needs to be further carried out.

The vital physiological process, photosynthesis that distinguishes green plants from other organisms entirely depends on the photosynthetic machinery along with the activities/levels of different proteins including rubisco. Furthermore, many other proteins have been also visualized that are markedly affected by stress, but their complete characterization is still underway.

Stress-induced increase or decrease in the levels of a number of enzymatic and non-enzymatic antioxidants in plants is now widely reported, but what type of a particular protein is produced during oxidative stress and what types of proteins are involved to control its accumulation at different levels of stress remains unclear. Furthermore, it is imperative to mention here that this knowledge is limited only to a very few proteins expressed during water stress and saline conditions, whereas not a single report is available on what type of proteins up- or downregulate under different stress conditions or what types of proteins are expressed individually or in combination involved in antioxidative system.

Nutrients (macro/micro) are effectively involved in regulation of plant metabolism. Deficiency of anyone of these triggers a number of pathways, more promising of which are carbohydrate metabolism, protein homeostasis, antioxidative defenses, signal transduction, membrane transductions, etc. Thus, it seems plausible to identify proteins which express under nutrient deficient environments, so they could be considered as potential indicator of mineral deficiency in plants. Furthermore, extensive research has been carried out for the determination of levels, deficiency symptoms, modes of action, and QTLs of different nutrients, but very few reports are available on the introgression of nutrient-related genes or/and QTLs to overcome nutrient deficiency within plant cells/tissues. So, research to explore this knowledge would be more beneficial for improving tolerance to nutrient deficiencies in crop plants.

Under high temperature stress, majority of the proteins expressed fall under the heat shock and LEA proteins. Their expression varies between heat stress tolerant and sensitive cultivars. Thus, identification and introgression of proteins expressed in heat stress tolerant plants could be used in modern biotechnology tools for the improvement of stress tolerance in economically important crops all-over the world.

Proteomic approach has been found to be very important as it helps plant physiologists to understand what is going on in the cell due to an external stimulus. Proteomics has gained attention world-wide due to easy handling of the proteomic analysis tools and accuracy of the results. For example, a number of techniques have been employed for the separation and identification/characterization of different proteins in different plant species including 2 dimensional liquid chromatography (2D-LC), polyacrylamide gel electrophoresis (PAGE), sodium dodecyl sulfate (SDS)- PAGE, pro-Q Diamond stain, 2-D gel electrophoresis, mass spectrometry, Coomassie brilliant blue (CBB)-stained 2-DE, MALDI-TOF, fluorescence, 2-D PAGE, non-gel-based LC-MS, ion-exchange chromatography (IEC), and 2-D difference GE (2D-DIGE). All these techniques have yielded sound results on characterization of proteins.

#### AUTHOR CONTRIBUTIONS

PA, AA, and SR wrote the manuscript. NA, MA, and SG contributed in section 2 of this manuscript. They also reviewed and updated the manuscript.

#### ACKNOWLEDGMENTS

The authors extend their appreciation to the Deanship of Scientific Research, College of Sciences Research Center, King Saud University, Riyadh, Saudi Arabia for supporting the project.

### REFERENCES

fpls-07-01336 September 6, 2016 Time: 12:3 # 10


and their degradation during recovery from salinity stress. J. Plant Physiol. 160, 1165–1174. doi: 10.1078/0176-1617-00909


in the Arabidopsis response to stress. Plant Cell 24, 2200–2212. doi: 10.1105/tpc.111.094946


tips of soybean under flooding stress using proteomics techniques. J. Proteomics 77, 531–560. doi: 10.1016/j.jprot.2012.09.032




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Ahmad, Abdel Latef, Rasool, Akram, Ashraf and Gucel. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Vascular Sap Proteomics: Providing Insight into Long-Distance Signaling during Stress

Philip Carella, Daniel C. Wilson, Christine J. Kempthorne and Robin K. Cameron\*

Department of Biology, McMaster University, Hamilton, ON, Canada

The plant vascular system, composed of the xylem and phloem, is important for the transport of water, mineral nutrients, and photosynthate throughout the plant body. The vasculature is also the primary means by which developmental and stress signals move from one organ to another. Due to practical and technological limitations, proteomics analysis of xylem and phloem sap has been understudied in comparison to accessible sample types such as leaves and roots. However, recent advances in sample collection techniques and mass spectrometry technology are making it possible to comprehensively analyze vascular sap proteomes. In this mini-review, we discuss the emerging field of vascular sap proteomics, with a focus on recent comparative studies to identify vascular proteins that may play roles in long-distance signaling and other processes during stress responses in plants.

#### Edited by:

Hanjo A. Hellmann, Washington State University, USA

#### Reviewed by:

Abu Hena Mostafa Kamal, University of Texas at Arlington, USA Elisabeth Jamet, Laboratoire de Recherche en Sciences Végétales, France

> \*Correspondence: Robin K. Cameron rcamero@mcmaster.ca

#### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 23 March 2016 Accepted: 28 April 2016 Published: 12 May 2016

#### Citation:

Carella P, Wilson DC, Kempthorne CJ and Cameron RK (2016) Vascular Sap Proteomics: Providing Insight into Long-Distance Signaling during Stress. Front. Plant Sci. 7:651. doi: 10.3389/fpls.2016.00651 Keywords: abiotic stress, biotic stress, long-distance signaling, phloem, proteomics, xylem

### INTRODUCTION

Plants are unable to relocate during unfavorable environmental conditions and instead rely on intricate signaling pathways that work at the local and systemic level to withstand stress. At the local level, stress-related signals move cell-to-cell through plasmodesmata (symplastic) or diffuse through the extracellular space (apoplastic). These communication processes are effective for signaling between neighboring cells or adjacent tissues, but are insufficient for systemic communication between organs (Lucas et al., 2013). For long-distance movement between distant tissues, macromolecules often access the plant vasculature; a system of specialized transport conduits connecting all organs of the plant.

#### The Plant Vascular System: Sharing Resources and Information

The xylem vessels and phloem sieve tubes of the plant vascular system connect above- and belowground organs, allowing vascular plants to draw from two distinct resource pools. The xylem provides an avenue for the unidirectional transport of water and mineral nutrients (xylem sap) from roots to aerial tissues that is driven by the transpiration stream. In contrast, the phloem allows for the bidirectional movement of photosynthate and other macromolecules throughout the plant, from areas of synthesis or excess (source) to areas of use (sink) such as storage tissues or zones of active growth. The fluid within the phloem is known as phloem sap, and its movement is thought to be driven by a hydrostatic pressure gradient along sieve tubes according to the pressure-flow hypothesis (reviewed in De Schepper et al., 2013; Lucas et al., 2013). However, this idea is not fully supported as data demonstrating differential pressure between source and sink phloem has not been obtained (discussed in Knoblauch and Oparka, 2012).

In addition to its role in resource allocation, the vasculature serves as an important conduit for the exchange of information between organs. Phloem sap appears to be highly complex, containing a diverse set of molecules such as sugars, lipids, amino acids, peptides, proteins, coding and non-coding RNAs, small molecules, mineral nutrients, and phytohormones (Lough and Lucas, 2006; Turgeon and Wolf, 2009; Lucas et al., 2013). By comparison, xylem sap appears to be less complex, primarily containing mineral nutrients, peptides, proteins, and hormones (Lucas et al., 2013; Turnbull and Lopez-Cobollo, 2013). Many of these molecules have been implicated as signals or signal chaperones in a number of developmental and stressrelated long-distance signaling pathways such as photoperiodinduced flowering, systemic acquired resistance (SAR), induced systemic resistance (ISR), wound responses, RNA silencing, autoregulation during plant-rhizobia symbioses, and systemic nutrient starvation responses. Most of these responses have been reviewed elsewhere (Lough and Lucas, 2006; Champigny and Cameron, 2009; Turgeon and Wolf, 2009; Lucas et al., 2013; Turnbull and Lopez-Cobollo, 2013; Lin et al., 2014), therefore we focus only on those that best illustrate how xylem and phloem facilitate inter-organ communication. The photoperiod-induced flowering response serves as the standard example for long-distance signaling in plants. Many plants sense increasing day length as a cue for the transition from vegetative to reproductive development. Early experiments demonstrated that a signal moves in the phloem from leaves, where day length is perceived, to the shoot apical meristem (SAM) where the transition to flowering is initiated (Zeevaart, 1976). More recent experiments identified the phloem-mobile signal as FT (FLOWERING LOCUS T) protein (Corbesier et al., 2007; Jaeger and Wigge, 2007). Interestingly, FT interacts with phospholipids that are important for FT's function in flowering-time regulation in the SAM (Nakamura et al., 2014). The lipid transfer protein DIR1 (DEFECTIVE IN INDUCED RESISTANCE 1) is an additional lipid-binding protein implicated in phloemmediated long-distance signaling. During the induction of SAR, DIR1 and phloem-mobile immune signals move from locally infected to naïve distant leaves to protect against future pathogen infection (Champigny and Cameron, 2009; Champigny et al., 2013). Several hydrophobic small molecules (azelaic acid, glycerol-3-phosphate, dehydroabietinal, pipecolic acid) have been identified as potential SAR mobile signals in the phloem; however, further experimentation is required to understand how these molecules participate in SAR (discussed in Dempsey and Klessig, 2012). Xylem-mediated long-distance signaling has been implicated in certain abiotic stress responses. For example, nutrient deprivation induces xylem-mobile hormone signals (cytokinins and strigolactones) that travel from roots to shoots to alter plant development (Lucas et al., 2013). Long-distance signaling responses sometimes utilize both xylem and phloem for signal movement. Following infection with symbiotic rhizobacteria, plants produce long-distance signals that travel through the xylem from roots to shoots to inform autotrophic tissues of the impending symbiotic association. Upon the perception of this xylem-borne signal, a shootgenerated signal accesses the phloem and travels back to the roots to regulate the development of symbiotic structures (Staehelin et al., 2011). Xylem-borne CLE (CLAVATA3/EMBRYO SURROUNDING REGION-RELATED) peptide signals from roots are believed to interact with LRR–RLKs (leucine rich repeat–receptor like kinases) in shoots, which in turn induce the accumulation and movement of unidentified shoot-derived signal(s) back to roots (Searle et al., 2003; Staehelin et al., 2011; Wang et al., 2016). Together, these responses illustrate how the plant vasculature acts a conduit for information sharing between distant tissues.

#### Collecting Vascular Sap

Our current understanding of vasculature-mediated longdistance signaling is impacted by the challenge of obtaining pure vascular sap. Standard methods involve collecting the fluid that exudes from the cut ends of petioles or puncture wounds of stems. Phloem sap is easily collected from cucurbits and legumes, which exude large amounts of phloem sap from petiole cut ends and stems (discussed in Turgeon and Wolf, 2009). Notably, the purity of cucurbit exudates has recently come into question, as some members of this family exude primarily from extrafascicular phloem (non-transport), and xylem (Zhang et al., 2010, 2012; Zimmerman et al., 2013). In other plants, phloem sap is collected over the course of several hours by submerging petiole ends in EDTA-containing solutions, which prevents sieve element occlusion by limiting the availability of Ca2<sup>+</sup> (King and Zeevaart, 1974). While EDTA-facilitated exudation enables phloem sap collection from a wide variety of plants, the sample is substantially diluted during the process and prolonged exposure to EDTA may lead to intracellular (nonphloem) contamination caused by tissue softening. A recent improvement to this method was described by Guelette et al. (2012), who demonstrated that an initial 1-h incubation of Arabidopsis petioles in EDTA, followed by a 9-h exudation period in sterile water, was sufficient for metabolomics and proteomics analysis. Another technique for the collection of phloem sap is aphid stylectomy, which uses phloem-feeding insects as tools to collect pure sap directly from phloem cells (discussed in Turgeon and Wolf, 2009). While the collected phloem sap is much less dilute, collection volumes are low and the method is technically challenging.

Xylem sap is typically sampled from the cut stems of larger plants such as Brassica oleracea, Zea mays, and Glycine max via bleeding or root pressure techniques. Bleeding techniques sample xylem sap directly from the cut end of stems or petioles, while root pressure techniques apply pressure (mechanically or through positive pressure using ice) to the rootstock to force liquid through the xylem, which is then collected from the cut end of the stem (Alexou and Peuke, 2013). Despite measures such as the pre-washing and blotting of cut stems, contamination of xylem sap by phloem and/or other cellular contents can be an issue for most xylem sap collection methods (Alexou and Peuke, 2013). Since small proteins and peptides have been implicated in long-distance signaling in the xylem (Neumann, 2007; Lucas et al., 2013), Okamoto et al. (2015) recently optimized a gelfree purification technique to enrich for small proteins/peptides in xylem sap. Combining o-chlorophenol extractions and HPLC

(high performance liquid chromatography) separation, the authors identified small proteins and peptides that were not detected using electrophoresis based-methods (Okamoto et al., 2015). However, a major limitation of these methods is the inability to collect xylem sap from smaller plants, which are often used as molecular-genetic model systems.

#### PROTEOMICS ANALYSIS OF VASCULAR SAP

Proteins play an important role in vasculature-mediated longdistance signaling responses, as demonstrated by the involvement of FT and DIR1 in photoperiod-induced flowering and SAR, respectively. Over the past 10 years, a number of proteomics studies provided information about the protein composition of xylem and phloem sap. Most studies relied on gel-based separation techniques such as 1D or 2D SDS-PAGE followed by standard protein detection approaches. This includes liquid chromatography (LC) coupled to mass spectrometers (MS), which consist of an ionization source (MALDI – matrix-assisted laser desorption/ionization, or ESI – electrospray ionization), and at least one of four types of mass analyzer; FTIC (fourier transform ion cyclotron), ion trap, TOF (time-of-flight), and quadrupole. Combinations of different mass analyzers in tandem MS set-ups improved protein coverage by overcoming particular weaknesses associated with each analyzer. In this review, we refer to these techniques simply as LC–MS/MS; for a comprehensive proteomics review, see Yates et al. (2009). Proteomics techniques have identified proteins in xylem and phloem sap collected from healthy plants growing in normal conditions (**Table 1**). In a recent review, Rodriguez-Celma et al. (2016) analyzed most of these studies and concluded that in general, the vascular fluids of multiple species contain proteins that appear to function in structural maintenance of the vasculature (e.g., cell wall metabolism) as well as constitutive defenses against pathogens (e.g., pathogenesis-related [PR] proteins, chitinases, proteases).

#### COMPARATIVE PROTEOMICS ANALYSIS OF VASCULAR SAP

Comparative proteomics of vascular sap is an excellent approach to identify proteins that may be involved in long-distance signaling responses. Early comparative proteomics studies relied on 1- or 2D difference gel electrophoresis (DIGE) techniques to compare the protein profiles of different samples. Spots present in some samples but not others are excised from the protein gel and analyzed by LC–MS/MS. Unfortunately, this type of analysis performs poorly with more complex samples, since individual proteins cannot be resolved by electrophoresis. Moreover, DIGE techniques have limited coverage since only differentially abundant proteins are analyzed (Gemperline et al., 2016). Modern gel-free comparative techniques overcome both of these issues while also providing quantitative information for every protein that is identified. Gel-free approaches such as ICAT (isotope-coded affinity tag), ICPL (isotope-coded protein labeling), or iTRAQ (isobaric tags for relative and absolute quantitation) rely on the addition of chemical labels to protein samples. Alternatively, label-free comparative proteomics, in which separate MS runs are aligned and compared in silico, is used to identify differentially abundant proteins between samples/treatments. Label-free quantitation is much less labor intensive compared to label-dependent techniques such as iTRAQ, and often allows for superior protein detection and more accurate quantitation (Patel et al., 2009; Trinh et al., 2013). To date, each of these techniques has been successfully employed for comparative proteomics analysis of plant vascular sap during stress responses (**Table 2**). More detailed descriptions of these methods can be found in other reviews (Yates et al., 2009; Lottspeich and Kellermann, 2011).

### Comparative Proteomics Analysis of Xylem Sap

Several comparative proteomics studies have been performed on xylem sap collected from plants experiencing stress (**Table 2**). The first was performed by Rep et al. (2002), who used MALDI-TOF-MS peptide fingerprinting to identify seven PRrelated proteins that accumulate in xylem sap of tomato (Solanum lycopersicum) plants infected with the xylem-infecting fungal pathogen Fusarium oxysporum. This finding was later supported by a 2D-DIGE proteomics study that identified host PR proteins as well as pathogen-derived proteins in tomato xylem sap during F. oxysporum infection (Houterman et al., 2007). Recently, more comprehensive analyses of xylem sap collected during host interactions with F. oxysporum were performed using label-free quantitative proteomics in tomato and Brassica oleracea (Gawehns et al., 2015; Pu et al., 2016). Both studies identified substantially more total proteins (∼150–285) with a relatively high proportion of those proteins showing differential abundance during infection (Gawehns et al., 2015; Pu et al., 2016). Other xylem sap proteomes collected during plant–microbe interactions include Glycine max infected with Fusarium virguliforme (Abeysekara and Bhattacharyya, 2014), G. max during symbiosis with Bradyrhizobium japonicum or during treatment with elicitors from the pathogen Phytophthora sojae (Subramanian et al., 2009), and Brassica napus infected with Verticillium longisporum (Floerl et al., 2008). A common theme among these proteomes is the accumulation of PR proteins such as chitinases and glucanases, which may serve in an antimicrobial capacity to limit the spread of infection via xylem vessels (Sels et al., 2008).

Comparative proteomics analysis of xylem sap during abiotic stress has been studied in Zea mays, B. oleracea, and Gossypium hirsutum. Alvarez et al. (2008) analyzed xylem sap collected from well-watered and drought-stressed Z. mays using 2D-DIGE and LC-MS/MS, identifying 33 proteins that accumulated during drought and 8 that decreased in abundance. Fernandez-Garcia et al. (2011) analyzed xylem sap proteomes of salt-stressed and control B. oleracea plants

#### TABLE 1 | Vascular sap proteomes of healthy plants.


<sup>1</sup>Comparative study where treated and healthy plants displayed identical protein profiles, <sup>2</sup>peptide fragments, <sup>3</sup> identified as differentially abundant in different tissues of healthy plants.

using 2D-DIGE comparative proteomics, identifying 22 proteins that accumulated during salt stress and 18 proteins that decreased in abundance. More recent comparative proteomics analyses of xylem sap have focused on responses to nutrient limitation. Liao et al. (2012) analyzed the Z. mays xylem sap proteome during nitrogen-limiting and -oversupply conditions using 2-DIGE and Zhang et al. (2016) performed labelfree quantitative proteomics on xylem sap collected from cotton (G. hirsutum) seedlings grown under normal and potassium (K)-limited conditions. Interestingly, the differential abundance of PR proteins, proteases, redox-associated proteins, and cell wall metabolism proteins was observed in each of these studies (Alvarez et al., 2008; Fernandez-Garcia et al., 2011; Zhang et al., 2016), which hints at the importance of these proteins in the xylem during abiotic as well as biotic stress.

#### Comparative Proteomics Analysis of Phloem Sap

Comparative proteomics analysis of phloem sap has been performed to investigate long-distance signaling during a number of stress responses (**Table 2**). Using iTRAQ-based proteomics Du et al. (2015) analyzed phloem sap collected from resistant and susceptible rice (Oryza sativa) cultivars that were either unexposed or exposed to phloem-feeding brown plant hopper (BPH) insects. They found that carbohydrate and protein metabolism proteins accumulated in phloem sap of susceptible BPH-infested plants, and that defense-related proteins accumulated in BPH-resistant plants. Responses to virus infection were investigated using phloem sap collected from melon plants (Cucumis melo) that were uninfected or infected with melon necrotic spot virus (MNSV). Using 2D-DIGE and LC-MS/MS, the authors identified a number of cell-death and redox-associated proteins that were differentially abundant during infection with MNSV (Serra-Soriano et al., 2015). Lastly, a label-free quantitative proteomics study was recently undertaken by our group to identify differentially abundant proteins in phloem sap of Arabidopsis thaliana during the induction of SAR with virulent or avirulent strains of Pseudomonas syringae (Carella et al., 2016). Of the 564 proteins identified in Arabidopsis phloem sap, 16 accumulated and 46 decreased in abundance during SAR. Proteins that accumulated in phloem sap during SAR included PR-1, redox-associated proteins, and putative lipid-binding proteins, while proteins with decreased abundance were associated with metabolism. The functional relevance of these proteins was investigated by performing SAR assays



<sup>1</sup>Difference gel electrophoresis, <sup>2</sup>not applicable, <sup>3</sup> label-free quantitative proteomics, <sup>4</sup>cultivar resistant to Fusarium oxysporum, <sup>5</sup>cultivar susceptible to F. oxysporum, <sup>6</sup>host-derived, <sup>7</sup>pathogen-derived, <sup>8</sup>peptide fragments, <sup>9</sup> isotope-coded protein labeling, <sup>10</sup>collected 3 h post wounding, <sup>11</sup>collected 24 h post wounding, <sup>12</sup>isobaric tags for relative and absolute quantitation, <sup>13</sup>salt-intolerant cultivar, <sup>14</sup>salt-tolerant cultivar, <sup>15</sup>spots on protein gel, <sup>16</sup>cultivar susceptible to insects, <sup>17</sup>cultivar resistant to insects.

on corresponding T-DNA knockout mutants, which identified m-type thioredoxins (TRXm1 and TRXm4) and a major latex protein (MLP) as novel phloem-localized proteins that play functional roles in SAR.

Comparative proteomics analyses of phloem sap collected during abiotic stress responses have also been performed. An initial comparative study of phloem sap collected from hybrid poplar (healthy or mechanically wounded) identified 48 total proteins using 2D-DIGE and LC-MS/MS, with two proteins accumulating during wounding stress (Dafoe et al., 2009). A subsequent ICPL-based study of phloem sap collected from wounded and unwounded cucumber (Cucumis sativus) identified substantially more total and differentially abundant proteins (Gaupels et al., 2012). Interestingly, PR-type proteins were identified in the phloem sap of poplar and cucumber (Dafoe et al., 2009; Gaupels et al., 2012) and cucumber phloem also contained cyclophilins, carbon metabolism-related proteins, and other defenserelated proteins (Gaupels et al., 2012). The phloem sap proteomes of plants experiencing other types of abiotic stress have been investigated in cucumber and B. napus. An iTRAQ-based study of phloem sap collected from salt-stressed cucumber identified several salt-responsive phloem proteins in salt-tolerant and -intolerant cultivars (Fan et al., 2015). In addition, comparative 2D-DIGE and LC–MS/MS analysis of phloem sap collected from Fe-deficient and control B. napus plants identified a number of redox-associated proteins with differential abundance during Fe stress (Gutierrez-Carbonell et al., 2015).

#### Common Proteins Present in Vascular Sap Proteomes during Stress

Comparative proteomics analysis of vascular sap has revealed a great deal about how the vascular system responds to stress. Surprisingly, a common theme among many xylem and phloem sap proteomes collected from stressed plants is the accumulation of PR proteins including thaumatin-like proteins, chitinases, glucanases, and MLPs. This may indicate that PR proteins play a role in the protection of the vasculature against pathogen/herbivore attack. Redox-related proteins (thioredoxins, peroxidases, etc.) are similarly associated with stress responses in the vasculature. It has been proposed that these proteins play a protective role in the phloem by preventing damage to proteins caused by oxidative stress (Walz et al., 2002). This may be especially important in phloem sieve elements as these cells lack protein synthesis machinery and cannot quickly replace proteins damaged during stress. In addition, cyclophilins, glycine-rich proteins (GRPs), and putative lipidbinding proteins are commonly identified in both xylem and phloem sap proteomes. The identification of multiple lipidbinding proteins in vascular sap supports the idea of lipid-based long-distance signaling in the vasculature (Benning et al., 2012; Barbaglia et al., 2016). The role of cyclophilins and GRPs in the vasculature is less clear. Extracellular GRPs have been linked to

cell wall formation, while intracellular GRPs and cyclophilins have been implicated in a number of functions including RNAbinding/chaperoning, which may suggest a role in RNA-mediated long-distance signaling in the vasculature (Mangeon et al., 2010; Kumari et al., 2013; Rodriguez-Celma et al., 2016).

#### CONCLUSION AND FUTURE DIRECTIONS

The emergence of sophisticated proteomics techniques has led to a new era in vascular sap proteomics studies. Together, these studies provide fundamental insights into the nature of plant vascular sap under normal and stress conditions. Although significant progress has been made in this field, much remains to be discovered. Further improvements to existing proteomics techniques will make it possible to detect lowabundance proteins essential for vasculature function. Moreover, novel xylem sap collection protocols for genetically tractable model systems are needed to investigate the importance of xylem-mobile proteins at the molecular-genetic level. In the

#### REFERENCES


meantime, researchers can take advantage of the Arabidopsis model system and the recently improved Arabidopsis phloem sap collection method (Guelette et al., 2012; Tetyuk et al., 2013) to expand our understanding of the molecular mechanisms of phloem-mediated long-distance signaling during plant stress.

#### AUTHOR CONTRIBUTIONS

Conceived of the review: PC and RC. Wrote the review: PC, DW, and RC. Analyzed literature and created tables: PC and CK. All authors edited the manuscript.

### FUNDING

This research was funded by a Natural Sciences and Engineering Research Council of Canada (NSERC) grant to RC, an NSERC graduate scholarship to DW, and an Ontario Graduate Scholarship to PC.




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Carella, Wilson, Kempthorne and Cameron. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Phloem Proteomics Reveals New Lipid-Binding Proteins with a Putative Role in Lipid-Mediated Signaling

Allison M. Barbaglia, Banita Tamot, Veronica Greve and Susanne Hoffmann-Benning\*

*Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, MI, USA*

#### Edited by:

*Jian Xu, National University of Singapore, Singapore*

#### Reviewed by:

*Stephen Beungtae Ryu, Korea Research Institute of Bioscience and Biotechnology, South Korea Ruth Welti, Kansas State University, USA*

\*Correspondence:

*Susanne Hoffmann-Benning hoffma16@msu.edu*

#### Specialty section:

*This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science*

Received: *17 February 2016* Accepted: *11 April 2016* Published: *28 April 2016*

#### Citation:

*Barbaglia AM, Tamot B, Greve V and Hoffmann-Benning S (2016) Phloem Proteomics Reveals New Lipid-Binding Proteins with a Putative Role in Lipid-Mediated Signaling. Front. Plant Sci. 7:563. doi: 10.3389/fpls.2016.00563* Global climate changes inversely affect our ability to grow the food required for an increasing world population. To combat future crop loss due to abiotic stress, we need to understand the signals responsible for changes in plant development and the resulting adaptations, especially the signaling molecules traveling long-distance through the plant phloem. Using a proteomics approach, we had identified several putative lipid-binding proteins in the phloem exudates. Simultaneously, we identified several complex lipids as well as jasmonates. These findings prompted us to propose that phloem (phospho-) lipids could act as long-distance developmental signals in response to abiotic stress, and that they are released, sensed, and moved by phloem lipid-binding proteins (Benning et al., 2012). Indeed, the proteins we identified include lipases that could release a signaling lipid into the phloem, putative receptor components, and proteins that could mediate lipid-movement. To test this possible protein-based lipid-signaling pathway, three of the proteins, which could potentially act in a relay, are characterized here: (I) a putative GDSL-motif lipase (II) a PIG-P-like protein, with a possible receptor-like function; (III) and PLAFP (phloem lipid-associated family protein), a predicted lipid-binding protein of unknown function. Here we show that all three proteins bind lipids, in particular phosphatidic acid (PtdOH), which is known to participate in intracellular stress signaling. Genes encoding these proteins are expressed in the vasculature, a prerequisite for phloem transport. Cellular localization studies show that the proteins are not retained in the endoplasmic reticulum but surround the cell in a spotted pattern that has been previously observed with receptors and plasmodesmatal proteins. Abiotic signals that induce the production of PtdOH also regulate the expression of GDSL-lipase and PLAFP, albeit in opposite patterns. Our findings suggest that while all three proteins are indeed lipid-binding and act in the vasculature possibly in a function related to long-distance signaling, the three proteins do not act in the same but rather in distinct pathways. It also points toward PLAFP as a prime candidate to investigate long-distance lipid signaling in the plant drought response.

Keywords: lipid-binding proteins, phospholipids, lipid signaling, abiotic stress, phloem

#### INTRODUCTION

As the world population grows, our need for food and fuel increases. This is aggravated by an encroachment of cities on arable land, competition between food and fuel crops, and the impact of global climate changes on crop yields. Abiotic factors such as drought, heat, and cold commonly affect crop yield. To continuously provide sufficient food and fuel for the increasing world population, we need plants with accelerated growth, higher yields, and increased stress tolerance. Since plants are sessile and cannot escape adverse conditions, it is essential to understand how plants perceive environmental changes and how they transmit the signals that convey developmental changes and the resulting adaptations. This requires intracellular and longdistance signaling. The plant long-distance transport systems are the xylem and the phloem. The two main components for phloem transport are sieve elements and companion cells. To enhance transport of molecules through the sieve elements, they optimize the longitudinal flow in these cells by degrading any obstacles in the form of organelles and ribosomes, leaving only the plasma membrane and a thin cytoplasm which contains ER, phloem-specific plastids, and a few dilated mitochondria (van Bel and Knoblauch, 2000; Turgeon and Wolf, 2009). The residual ER is found near the plasmodesmata which connect the sieve elements with the companion cells. It is thought to participate in controlling and mediating the trafficking of proteins and other molecules from the companion cell, where they are synthesized, into the sieve element for long-distance movement (Lucas et al., 2009, 2013). Transport of photoassimilates as well as signaling molecules is thought to occur from source (photosynthetically active, mature leaves) to sink (immature leaves, roots, fruits, flowers, etc.) in a mechanism driven by the osmotic gradient ("Pressure flow hypothesis"; Münch, 1930; for a review see Froelich et al., 2011; Lucas et al., 2013). Our understanding of the phloem has evolved from simple assimilate movement to a complex trafficking system for environmental- and stress signals as well as developmental regulators (Citovsky and Zambryski, 2000; Ding et al., 2003; Wu et al., 2003; Haywood et al., 2005; Lucas et al., 2013) in the form of small molecules (Chen et al., 2001; Corbesier et al., 2003), proteins (Fisher et al., 1992; Schobert et al., 1995; Kühn et al., 1997; Marentes and Grusak, 1998; Kehr et al., 1999; Xoconostle-Cazares et al., 1999; Haebel and Kehr, 2001; Hoffmann-Benning et al., 2002; Giavalisco et al., 2006; Lin et al., 2009; Guelette et al., 2012; Champigny et al., 2013), nucleic acids (Ruiz-Medrano et al., 1999; Citovsky and Zambryski, 2000; Ding et al., 2003; Yoo et al., 2004; Haywood et al., 2005; Pant et al., 2008; Buhtz et al., 2010; Varkonyi-Gesic et al., 2010; Rodriguez-Medina et al., 2011; Hannapel et al., 2013; Pallas and Gómez, 2013), and lipophilic molecules, including complex lipids such as steroids and phospholipids (Madey et al., 2002; Behmer et al., 2011, 2013; Guelette et al., 2012).

Using proteomics approaches, we along with others have identified several putative lipid-binding proteins in the phloem of several plant species as diverse as Perilla, lupine, Arabidopsis, broccoli, canola, several cucurbits, poplar, and rice (**Table 1**; Hoffmann-Benning et al., 2002; Walz et al., 2004; Giavalisco et al., 2006; Aki et al., 2008; Dafoe et al., 2009; Lin et al., 2009; Cho et al., 2010; Rodriguez-Medina et al., 2011; Guelette et al., 2012;

The questions arise: What is the function of the phloem lipids? How are they solubilized and transported in this aqueous environment? And what is the role of the phloem lipid-binding proteins in this process?

Our findings led us to propose that phloem (phospho-) lipids could act in long-distance developmental signaling in response to abiotic stress: They could facilitate perception by tethering a signaling molecule, receptor, or secondary messenger to the membrane. Alternatively, they could be (part of) a signal themselves. As such they are released, sensed, and moved by phloem lipid-binding proteins (Benning et al., 2012; Hoffmann-Benning, 2015; Barbaglia and Hoffmann-Benning, 2016). Indeed, the proteins we identified include lipases, that could release the signaling lipid into the phloem, putative receptor components, and proteins that could mediate lipid-movement.

The presence of lipids in an aqueous environment is not without precedence in biological systems: Cholesterol is either taken up into cells and incorporated into membranes, or it is moved to the liver for degradation. Its fate depends on the lipoproteins to which it is bound (for a summary see Nelson et al., 2008). Other examples of the lipid movement and signaling are (I) the developmental regulator Wnt in animals, which requires palmitoleic acid for binding to the receptor Frizzled (Frz; Janda et al., 2012); (II) the platelet activation factor is a phospholipid, which controls platelet aggregation and inflammation (Christie, 2014); (III) the regulation of the β-oxidation by fatty acids via the transcription factor PPARα1 (Wahli and Michalik, 2012). Clearly, lipids can act in long-distance signaling using proteinfacilitated mechanisms. The type of protein to which a lipid binds not only determines its transport but also its fate as well as downstream regulatory processes. Despite the fact that these lipid-protein signaling mechanisms are essential for mammalian health and development, their significance in plants is virtually unexplored.

Phloem lipids range from small lipophilic molecules (Jung et al., 2009; Chanda et al., 2011; Chaturvedi et al., 2012; Shah et al., 2014) to lipophilic hormones (Wu et al., 2003; Behmer et al., 2013; Lucas et al., 2013) to (phospho-)glycerolipids (Madey et al., 2002; Guelette et al., 2012; for a summary see Hoffmann-Benning, 2015). Small lipophilic molecules such as oxylipins, dehydroabietinal, a glycerol-3-phosphate-derivative, and azelaic acid (AzA) are studied mostly in the context of biotic stress and systemic acquired resistance (SAR; Howe and Schilmiller, 2002; Chaturvedi and Shah, 2007; Jung et al., 2009; Chanda et al., 2011; Chaturvedi et al., 2012; Shah et al., 2014). The oxylipin jasmonate (JA) is synthesized in response to wounding or herbivory. It moves throughout the plant as its Isoleucine (Ile)- or methyl-ester and elicits a (systemic) defense response (Howe and Schilmiller, 2002; Thorpe et al., 2007; Truman et al., 2007; Mandal et al., 2011; Matsuura et al., 2012; Tamogami et al., 2012). Moreover, a role for the JA-precursor 12-oxophytodienoic acid in response to drought and crosstalk with ABA has been suggested (Savchenko et al., 2014). Behmer et al. (2011, 2013) detected free, acylated, and glycosylated derivatives of cholesterol, sitosterol, campesterol, and stigmasterol in the phloem.

TABLE 1 | Putative lipid-binding proteins that were identified in the phloem exudates of several plant species (Hoffmann-Benning et al., 2002; Walz et al., 2004; Giavalisco et al., 2006; Aki et al., 2008; Dafoe et al., 2009; Lin et al., 2009; Cho et al., 2010; Rodriguez-Medina et al., 2011; Guelette et al., 2012; Anstead et al., 2013; Lattanzio et al., 2013; Tetyuk et al., 2013; Du et al., 2015).


*Expression in companion cells is based on Mustroph et al. (2009), Deeken et al. (2008), Zhao et al. (2005); lipid-binding is based on this paper (GDSL; PIG-P), Chen et al. (2008) (ACBP6), Tzen and Huang (1992) (GRP17), Rescher and Gerke (2004) (Annexins), Nakamura et al. (2014) (FT), Benning et al. (2012) (PLAFP); Lascombe et al. (2008); Shah et al. (2014) (DIR1). Proteins examined in this paper are highlighted in yellow.*

Phospholipids act as intracellular signals regulating development as well as the response to biotic and abiotic stress (Zhu, 2002; Wang et al., 2007; Munnik and Testerink, 2009; Wang and Chapman, 2012; Gillaspy, 2013; Ischebeck et al., 2013; Hung et al., 2014). One of these, phosphatidic acid is generated in the plasma membrane in response to several environmental stresses and ABA via phospholipases D or C and partakes in intracellular signal transduction (Welti et al., 2002; Wang et al., 2007; Munnik and Testerink, 2009; McLoughlin and Testerink, 2013). However, the concept of phospholipids as long-distance signals has not been investigated and provides a novel aspect in lipid signaling.

To test possible protein-based lipid-signaling pathways, three phloem lipid-binding proteins, which could potentially act in a relay, are characterized here:

(I) a putative GDSL-motif lipase that may release lipids into the phloem;


GDSL esterases/lipases are part of a subfamily of hydrolytic/lipolytic enzymes. They contain a distinct Glycine-Aspartic acid-Serine-Leucine (GDSL) motif and have a flexible active site that changes conformation in the presence of different substrates. This flexible active site leads to a broader substrateand regiospecificity. It is situated near the N-terminus, while the active site of other lipases is located near the center of the protein (Akoh et al., 2004). GDSL lipases play a role in seed germination (Ling et al., 2006), plant growth and morphogenesis (Brick et al., 1995), and pathogen response (Lee and Cho, 2003; Hong et al., 2008; Oh et al., 2005). AtGLIP2 plays a role in pathogen defense against Erwinia carotovora through the negative regulation of auxin signaling (Lee et al., 2009).

The PIG-P-like protein is a protein of unknown function with similarity to one subunit of the yeast and human phosphatidylinositol N-acetylglucosaminyltransferase subunit P (PIG-P) of the GPI-N-acetylglucosaminyltransferase. The human enzyme transfers N-acetylglucosamine from UDP-Nacetylglucosamine to phosphatidylinositol and assists in the GPIanchor formation (Watanabe et al., 2000). However, it is much smaller than the putative AtPIG-P, thus their functions are not necessarily related. PIG-P has several homologs in other plants, all containing a DUF4378 at the carboxy-terminus that is shared with the yeast and human PIG-P and could contain the lipidbinding site. The remainder of the plant proteins shows no similarity to any known protein and may have a novel and plant-specific function.

The phloem lipid-associated family protein (PLAFP) is a small putative lipid-binding protein of unknown function. It contains a PLAT/LH2 domain, which is thought to mediate interaction with lipids or membrane-bound proteins (Bateman and Sandford, 1999). Proteins containing the PLAT/LH2 domain are typically stress-induced (Bona et al., 2007; Mhaske et al., 2013). The presence of the PLAT domain has led to the annotation of this protein as a lipase or lipoxygenase, however, PLAFP lacks the catalytic site, suggesting a different function. Hyun et al. (2014) proposed a function in the ER stress response; however, we could not confirm any association with the ER (see Section Localization of Protein and Promoter Activity of GDSL-Lipase, PLAFP, and PIG-P). We have shown that PLAFP specifically binds phosphatidic acid (Benning et al., 2012; see **Figure 1C** or **Figure 1E**), a membrane lipid known to participate in intracellular signaling in response to several stresses (Wang, 2005; Wang et al., 2006; Testerink and Munnik, 2011; Arisz et al., 2013).

To participate in any long-distance function, the proteins need to be expressed in the vasculature of the plant. A previous analysis of the phloem translatome by Mustroph et al. (2009) suggested expression of PLAFP and the putative GDSL-motif lipase in phloem companion cells. Our study here goes beyond the proteomics approach that identified putative lipid-binding proteins in the phloem and provides a functional analysis of three candidates in the context of lipid binding and signaling in response to environmental signals. Our findings show that all three proteins are indeed lipid-binding (**Figure 1**), bind to the same lipid (PtdOH), act in the vasculature (**Table 1**; **Figure 3**), and respond to PtdOH-mediated stresses (**Figure 4**); However, their different response to environmental factors suggests that the three proteins likely do not act in the same pathway.

#### MATERIALS AND METHODS

#### Plant Growth

Arabidopsis seeds were sterilized (20% bleach and 0.5% Triton X-100 for 15 min and washed 6 times with sterile, distilled water) and plated on MS, 1% sucrose, and 0.6% agar. Transgenic lines were selected by growth on plates containing 25 µg/ml kanamycin and confirmed using PCR. For stress experiments plants were germinated on antibiotic-free plates. Next, plates were transferred to 4◦C for 2 days before being placed into a Percival growth chamber; 22/18◦C, 12-h light/12-h dark photoperiod with 60% relative humidity, and a light intensity of 120 µmol photons m−<sup>2</sup> s −1 . After 2 weeks seedlings were either transplanted into soil [equal parts Bacto Soil (Michigan Pear Company, Houston), medium vermiculite, and perlite] and grown to maturity or transferred to hydroponic culture for stress experiments.

### Stress Treatments

Wildtype Col-0 seedlings were carefully transferred to a hydroponic-like system containing water, covered with a clear plastic dome and left to acclimate for 24 h, at room temperature (22◦C.). After this period, 300 mM Mannitol, 150 mM NaCl, 100 µM of ABA, or 30% PEG 6000 were added to the system. Seedlings were harvested at 0, 1, 2, 5, 8, 10, 12, and 24 h post stress (hps). For each time point 3–6 seedlings were pooled. Each time course was performed in triplicate. Asterisks indicate statistical significance as determined by Student's t-test; p < 0.01.

#### Gene Expression Analyses

Total RNA was extracted from 2–3 week old Arabidopsis seedlings or leaves from 5-week- old plants following the instructions provided by the RNEasy Plant Mini Kit (Qiagen). The first strand was synthesized by oligo dT primers using SuperScript First Strand Synthesis III system (Invitrogen). The resultant cDNA was then used for quantitative real-time PCR (qPCR) using SYBR Green (Affymetrix) as the detection probe. Standard conditions (95◦C activation, gene-specific annealing temperature, 72◦C elongation; repeated 40 times) and a melting curve set at 60◦C with a 20 min run time were performed for each run. Primers and annealing temperatures for all the RT-PCR and qPCR are outlined in **Supplementary Table 1**.

#### Protein Expression and Purification

A cDNA clones for GDSL-lipase (At1g29660), U13183; PLAFP (At4g39730), U21720; and PIG-P (At2g39435) were obtained from Arabidopsis Biological Resource Centre, Ohio State University (Columbus, OH, USA). The coding regions of GDSLlipase, PLAFP, and PIG-P excluding the 78 and 69 nucleotide regions encoding the 26 and 23 amino-acid predicted signal peptides for GDSL-lipase and PLAFP, respectively, was PCR amplified using the primers indicated in **Supplementary Table 1**, which introduced NdeI sites at the 5′ end and BamHI at the 3′ end of the GDSL-lipase and PIG-P PCR products and NdeI sites at both ends of the PLAFP PCR product. The PCR products were cloned into pGEMT-Easy vector (Promega), and subcloned into pET15b expression vector (Novagen) using the NdeI site to generate the expression clone, pET15b-GDSL-lipase/PLAFP/PIG-P. E. coli host strain OrigamiB(DE3)pLysS (Novagen) was transformed with pET15b-PLAFP and BL21(DE3)pLysS for pET15b-GDSL-lipase/ PIG-P, respectively. Transformants were selected by Ampicillin (Amp), Kanamycin (Kan), Chloramphenicol (Cm), and Tetracycline (Tet) resistance for PLAFP and Amp and Cm resistance for GDSL-lipase and PIG-P. IPTG up to the final concentration of 0.5 mM was used to induce protein expression. PLAFP protein

was extracted and purified using the HisLinkTM resin (Promega) using the HEPES buffers containing different concentrations of imidazole, following the manufacturer's instructions, and GDSL-lipase and PIG-P proteins were extracted and purified using the Ni-NTA resin (Qiagen) using the phosphate buffers containing different concentrations of imidazole, following the manufacturer's instructions. Purification steps include the clear lysate, flow through, wash fraction, and elution fractions. The purified protein was exchanged into 10 mM KH2PO<sup>4</sup> (Lu and Benning, 2009) using a PD10 column (GE healthcare).

#### Protein–Lipid Overlay Assay

Ten nmol of various phospholipids (Avanti Polar Lipids; di 18:1 Phosphatidylethanolamine: PtdEtn, Phosphatidic acid: PtdOH, Phosphatidylcholine: PtdCho, Phosphatidylserine: PtdSer, Phosphatidylglycerol: PtdG, Phosphatidylinositol: PtdIns) were spotted onto a Hybond-C membrane (GE Healthcare) for PLAFP-lipid binding studies. Pre-spotted membranes for analysis of PIG-P and GDSL-lipase were purchased from Echelon Biosciences Inc. The protein-lipid overlay assay was performed according to Benning et al. (2012) and Awai et al. (2006).

#### Liposome Binding Assay

Liposomes (lipid-bilayer vesicle) were prepared using the above lipids or a mixture of thereof, following the method described in Awai et al. (2006) and Benning et al. (2012). In short, 250 µg liposomes were mixed with 1 µg of purified protein in 50 mM Tris–HCl, pH7.0, 0.1 M NaCl, and centrifuged at 10,000 × g for 10 min at 4◦C after incubation at 30◦C for 30 min. The pellet was washed and then resuspended in SDS-PAGE sample buffer. Western blot analysis was performed using anti-His and HRP-conjugated goat anti-mouse antibodies.

### GUS Reporter Gene Construct, Arabidopsis Transformation, and GUS Assay

The 1 Kb region upstream of the transcription initiation site of PLAFP was PCR amplified using the primers indicated in **Supplementary Table 1**. HindIII and XbaI sites were added at the 5′ and 3′ ends for PLAFP. The PCR product was cloned into pGEMT-Easy vector (Promega) and subcloned into pBI121 (Clontech) vector (from which the 35S promoter was removed by digestion using the restriction enzymes mentioned above) to generate PLAFP1KbPro:GUS which was then transformed into Agrobacterium tumefaciens strain GV3101 and C58C1pGV2260 by electroporation, respectively. Positive transformants were selected by Kanamycin resistance, and further confirmed by colony PCR, purified, and sequenced by the Research Technology Support Facility (RTSF) Genomics Core at Michigan State University, and used to transform Arabidopsis Col-0 by floral dip method (Clough and Bent, 1998). Transgenic lines were selected by Kanamycin resistance and the incorporation of the transgene was confirmed by PCR, using primers indicated in **Supplementary Table 1**.

A GUS assay was performed as described (Martí et al., 2010) using GUS staining solution: 50 mM sodium phosphate buffer, pH 7.0, 0.5 mM potassium ferricyanide, 0.5 mM potassium ferrocyanide, 0.1% triton X-100 and 1mg/ml 5-Bromo-4-chloro-3-indoxyl-beta-D-glucuronide cyclohexylammonium salt (Gold Biotechnology). Seedlings were observed under a Nikon Eclipse Ci light microscope.

#### Fluorescent Reporter Gene Constructs for GDSL-Lipase and PIG-P and PLAFP

The coding sequence of GDSL-lipase, PLAFP, and PIG-P was PCR amplified using the primers indicated in **Supplementary Table 1**, which added the att sites of the Gateway donor/destination vectors at 5′ and 3′ ends. The PCR product was cloned into pGEMT-Easy vector (Promega) and subjected to the Gateway cloning system where the resultant DNA product was subcloned into the donor vector pDNOR 207 followed by the destination vector, pEarleyGate 103 (or pEarleyGate 102—CFP or pEarleyGate 101—YFP) to generate the clones GDSL1KbPro:GFP (CFP), PLAFP1KbPro:YFP, and PIG-P1KbPro:GFP (CFP), which were then transformed into A. tumefaciens strain GV3101 by electroporation. Positive transformants were selected by Kanamycin resistance, further confirmed by colony PCR using the same set of primers mentioned above, sequenced, and used to transiently transform Nicotiana tabacum. Leaf samples were then observed under confocal microscopy (Olympus FV1000SP CLSM; YFP Emission wavelength: 530–555 nm, excitation: 515 nm; RFP emission wavelength: 605–630 nm, excitation: 559 nm; CFP emission wavelength: 475–500 nm, excitation: 458 nm) to detect the subcellular localization of the proteins.

#### RESULTS

#### The Predicted Phloem Lipid-Binding Proteins GDSL-Lipase, PLAFP, and PIG-P-like Protein Bind Lipids

The plant phloem contains several putative lipid-binding proteins (**Table 1**) as well as lipids (Guelette et al., 2012). To participate in any lipid-based signaling pathway, these proteins need to bind specific lipids, including lipids that can be found in phloem exudates. Protein-lipid overlay assays (**Figures 1A–C**) suggest a strong interaction of the putative GDSL-lipase with diacylglycerol, phosphatidylinositol-3,4,5-trisphosphate (PtdInsP3) and a weak interaction with phosphatidic acid (PtdOH); the putative PIG-P shows interaction with phosphatidylserine (PtdSer), phosphatidylinositol-4-phosphate (PtdInsP1), and PtdOH; PLAFP specifically binds PtdOH.

The lipid-binding seen in the protein-lipid overlay was confirmed using independent liposome-binding assays: the purified putative lipid-binding protein was incubated with liposomes consisting of lipids that had been identified in the overlay assay. Proteins that bind to the liposomes of a specific lipid composition can be detected in the (liposome-containing) pellet (**Figures 1D–G**, respectively), while proteins that do not bind to the liposomes are found in the supernatant (illustrated in **Figure 1D**). PtdCho-liposomes were used as negative controls as none of the proteins showed interaction with PtdCho in the protein-lipid overlay. As **Figure 1D** illustrates, GDSL-lipase is not detected in the pellets containing liposomes that contain DAG or PtdCho, or a mixture thereof. However, it is detected in the supernatant. Different compositions of DAG-containing liposomes were used, none of which interacted with the lipase. On the other hand, the GDSL-lipase does associate with PtdOHcontaining liposomes (**Figure 1G**). Since the binding to PtdOH in the protein-lipid overlay assay was weak an increased amount of protein (10 µg) was used for **Figure 1G**, showing that in addition to binding PtdOH, there is a weak interaction with PtdCho. Together this indicates that the lipase binds preferentially to PtdOH.

**Figure 1E** illustrates that PLAFP does not bind to liposomes consisting solely of PtdCho. However, when PtdOH was included in the liposome, PLAFP bound. The amount of protein bound increased with the PtdOH content of the liposomes.

Similarly, PIG-P binds to liposomes containing PtdOH or PtdSer but not to liposomes consisting exclusively of PtdCho (**Figure 1F**).

These liposome binding studies confirmed binding of PtdOH to PLAFP, PIG-P, and GDSL-lipase as well as binding of PtdSer to the PIG-P-like protein. Binding to PtdInsP<sup>3</sup> was not tested since this lipid has so far not been reported in plants and is, thus, not of biological relevance (Munnik and Testerink, 2009). Our results demonstrate that all three proteins are indeed lipidbinding proteins. Most importantly, they all bind PtdOH albeit with different intensities. PtdOH is one of the lipids that was found in the phloem (Benning et al., 2012; Guelette et al., 2012) and that is already known to participate in intracellular signaling (Wang et al., 2007; Xue et al., 2009; Hong et al., 2010; Kim et al., 2013; McLoughlin and Testerink, 2013). Thus, these findings suggest the possibility that all three proteins function in a PtdOHrelated signaling path.

#### Localization of Protein and Promoter Activity of GDSL-Lipase, PLAFP, and PIG-P

During the development of the phloem, many organelles and the nuclei of the sieve elements disintegrate to allow for an unobstructed flow of molecules (Lucas et al., 2013). While some components of the translational apparatus can be found (Lin et al., 2009) they are likely not sufficient for translation and may be remnants of earlier developmental stages. Hence, it is believed, that proteins, RNA, and many other molecules found in the sieve elements are synthesized in the companion cell and move to the sieve elements via plasmodesmata, possibly in an ER-mediated mechanism (Lucas et al., 2013). The GDSL-lipase and PLAFP contain signal peptides, while PIG-P is predicted to be a soluble protein. To understand their localization within the plant cell we generated fusion proteins containing C-terminal fluorescent tags and transiently expressed those in tobacco (**Figure 2**). All three proteins are localized in a dispersed pattern at the periphery of the cell. No co-localization with chloroplasts or nuclei was observed. Similarly, markers for Golgi and plasma membrane also show no overlap (not shown). Overlays with a fluorescent marker for the ER show that there is little co-localization with the ER marker (**Figure 2**). This is particularly obvious for GDSLlipase and PLAFP where cytoplasmic strands containing the ER are clearly visible but show no overlap with the fluorescently tagged protein. PLAFP in particular displays a spotted pattern without ER-colocalization. A similar spotted pattern has been reported for receptors as well as for plasmodesmata-mobile proteins (Kim et al., 2002; Robatzek et al., 2006).

To determine the localization of gene expression we searched several phloem-specific transcriptomes (Zhao et al., 2005; Deeken et al., 2008; Mustroph et al., 2009) for the presence of GDSL-lipase, PLAFP, and PIG-P gene expression. GDSLlipase and PLAFP were found in the companion-cell specific databases suggesting that genes are expressed in the companion cells and could, thus, translocate into the sieve elements via plasmodesmata. Using a GUS reporter gene under the control of the PLAFP promoter we show that PLAFP-promoter activity is indeed associated with the vasculature in roots and expanding leaves, as well as in the hydathodes, which are associated with the vasculature (**Figure 3**: leaf and root). Expression at the branchpoint for lateral roots and in the leaf primordia suggests that PLAFP may be necessary during early vasculature development.

Overall gene expression as determined by RT-PCR showed that all three genes are expressed in all tissues of the plant (stem, root, leaf, and flower) with GDSL-lipase and PLAFP expression at slightly reduced levels in the root (**Supplementary Figure 1**).

#### PLAFP and GDSL-Lipase Gene Expression is Affected by the Same Environmental Factors That Lead to the Production of Their Lipid-Ligand Phosphatidic Acid

We have shown that GDSL-lipase, PLAFP, and the PIG-Plike protein bind PtdOH, which we had detected in phloem exudates (Guelette et al., 2012). PtdOH is a well-known intracellular signal acting in response to various abiotic and biotic stresses such as pathogen response and infection, drought, salinity, wounding, cold, cell death, and oxylipin production (Wang et al., 2007, 2014; Munnik and Testerink, 2009; Xue et al., 2009; Hong et al., 2010; Testerink and Munnik, 2011; Kim et al., 2013; Julkowska et al., 2015). It plays a role in maintaining root system architecture as well as salt tolerance (McLoughlin and Testerink, 2013). In addition, PtdOH is widely known for its role in stomatal closure via the activation of the ABA-signaling pathway under drought/osmotic conditions (Guo et al., 2012; Lu et al., 2013; Yao et al., 2013). To understand if GDSL-lipase, PLAFP, and the PIG-P-like protein are controlled by the same environmental factors as PtdOH, we exposed 3-week-old Arabidopsis seedlings to salt (NaCl) and osmotic stress (Mannitol) as well as to the drought mimic PEG and the stress signal ABA (**Figure 4**; **Supplementary Figure 2**). Concentrations of mannitol, NaCl, PEG, and ABA were based on those used in the literature (Yamaguchi-Shinozaki and Shinozaki, 1994; Nakashima et al., 1997; Zhu, 2002; Fujita et al., 2005; Zhu et al., 2010). Gene expression was monitored for 24 h. PIG-P expression was not affected by any of the stress factors (**Figure 4**). This is not surprising since we had proposed that PIG-P may be part of a receptor and as such, should be constitutively expressed. GDSL-lipase expression was downregulated by osmotic (Mannitol) stress, the signaling molecule ABA, and water stress as mimicked by PEG. This downregulation was significant within 5 h of treatment and was maintained for 24 h (**Figure 4**). PLAFP displayed the opposite response to these stresses: its expression was strongly upregulated by ABA and water stress (PEG) within 2 h, an effect that was maintained for the entire 24-h treatment period (**Figure 4**). A response to mannitol was observed after 12–24 h. This increase in PLAFP parallels the induction of PtdOH synthesis under the same conditions (Wang et al., 2007, 2014; Munnik and Testerink, 2009; Xue et al., 2009; Hong et al., 2010; Testerink and Munnik, 2011; Kim et al., 2013; Lu et al., 2013; McLoughlin and Testerink, 2013; Yao et al., 2013; Julkowska et al., 2015).

#### DISCUSSION

A survey of our proteomics analysis of the phloem exudates of Arabidopsis thaliana as well as publications of other phloem proteomes has shown the presence of lipids and lipid-binding proteins within the translocation stream (**Table 1**; Guelette et al., 2012). This prompted us to propose the possibility of longdistance lipid signaling (Benning et al., 2012). As part of this longdistance path, one protein would release the lipid into the sieve element (GDSL-lipase) where it is bound by a second protein that functions either as transporter or co-signal (PLAFP) and later perceived at a receptor (PIG-P). To participate in the proposed long-distance lipid signaling, these proteins need to fulfill several requirements:



Our results show that all three predicted lipid-binding proteins (GDSL-lipase, PLAFP, and the PIG-P–like protein) bind lipids (**Figure 1**). Most importantly, all three proteins bind PtdOH. This is of particular importance since PtdOH has been found in the same phloem exudates that were used to identify the protein (Guelette et al., 2012). PtdOH is an important intermediate in lipid biosynthesis, a membrane component, and a signaling molecule: As a membrane component it may affect the membrane curvature and, consequently, regulates trafficking and membrane biogenesis (Wang, 2004; Kooijman et al., 2005). Most importantly in the context of this work, PtdOH participates in signaling pathways, often by tethering components of these pathways to the membrane, thus, altering their location and function. PtdOH is rapidly and transiently produced in response to several biotic and abiotic stresses, such as drought, salinity, wounding, cold, pathogen infection, and oxylipin production (Wang et al., 2007, 2014; Munnik and Testerink, 2009; Xue et al., 2009; Hong et al., 2010; Testerink and Munnik, 2011; Kim et al., 2013; Julkowska et al., 2015). The path of its production and the enzymes involved varies depending on the environmental signal (Welti et al., 2002; Uraji et al., 2012; Arisz et al., 2013; McLoughlin and Testerink, 2013; Gonorazky et al., 2014; Julkowska et al., 2015).

We examined if the three PtdOH-binding proteins responded to some of the same environmental factors that induce the production of PtdOH (**Figure 4**), namely osmotic stress, a water-stress mimic (PEG), and the signaling molecule (ABA). PIG-P expression is not influenced by any of those factors. Possible explanations are that PIG-P is part of a receptor and, hence, would likely be constitutively expressed. Alternatively, it could be post-translationally modified or its function could

be unrelated to abiotic stress. The expression of GDSL-lipase is downregulated by ABA, Mannitol, and PEG and upregulated by NaCl. While there have been other GDSL-lipases that exhibit an increase in expression under various abiotic stresses such as ABA, drought, osmotic, salt, and SA stress (Hong et al., 2008), this particular lipase shows the opposite effect. The most interesting finding was that PLAFP expression is upregulated by PEG and ABA within 5 h and by Mannitol within 12 h (**Figure 4**). Drought, osmotic stress, ABA, salt stress and cold activate distinct phospholipases that cleave phospholipids and generate lipid messengers particularly PtdOH, diacylglycerol, and inositol-3-phosphate. They are thought to affect stress tolerance partially through modulating the expression of stressresponsive genes (Zhu, 2002). One example of a PtdOH-based signaling cascade is the response to osmotic stress and drought, which can lead to an increase in ABA. This increase in ABA leads to the activation of phospholipase Dα1, which in turn produces PtdOH. PtdOH prevents abscisic acid insensitive 1 (ABI1), a protein phosphatase 2C, from binding to the ABA receptor by tethering it to the membrane, subsequently leading to a modification in gene expression and an ABA response. In addition, PtdOH has been shown to be involved in the intracellular signaling process by regulating stomatal closure, which leads to a conservation of water when the plant is experiencing water-deficit conditions (Lu et al., 2013), and by regulating the transcription of genes such as GLABRA2 (GL2) through interaction with the MYB transcription factor, WEREWOLF (Yao et al., 2013). The proposed function here is that PtdOH tethers WEREWOLF to the nuclear envelope and facilitates its movement into the nucleus. In long-distance signaling, PtdOH could either act by binding and moving signals from the companion cell into the sieve element, by tethering a receptor to the plasma membrane of the sieve element, by functioning as binding site for a (proteinaceous) signal, or by being part of a mobile signal that would consist of a mobile protein with a hydrophobic pocket for lipid (PtdOH) binding.

In summary, we find that all three lipid-binding proteins and their lipid ligand PtdOH are present in phloem exudates. GDSL-lipase and PLAFP respond to several abiotic stress factors which also regulate PtdOH-production albeit in opposite fashion. This suggests that these proteins may have a long-distance function in response to abiotic stress. The facts that both, PLAFP and its ligand PtdOH are induced by the same environmental factors, that they are both present in the phloem, and that PLAFP is produced in the vasculature (**Figure 3**) allow for the possibility that they act in the same signaling pathway and may

(Student's *t*-test).

be part of a mobile signal. Thus, PLAFP-PtdOH can function as model system to study the possibility and mechanisms of lipid-mediated, long-distance signaling in plants. In addition, they provide a unique opportunity as targets for generating stress tolerant plants.

#### AUTHOR CONTRIBUTIONS

AB generated GDSL- and PIG-P fluorescently tagged proteins and performed localization studies. She also performed the stress response gene-expression studies and analyzed the GUS expression lines. BT generated the promoter-GUS constructs and the protein constructs for overexpression in E. coli. BT, VG, and AB purified the proteins and performed lipid-binding studies. SHB conceived and supervised the experiments. The manuscript was written by SHB with excerpts from AB. BT and VG proofread and approved the manuscript

#### FUNDING

This work was supported by NSF-IOS grant #1144391 to SHB, the USDA-NIFA Hatch project # MICL02233 to SHB, a US Department of Energy graduate assistantship (DE-FG02-91ER20021) and a Cell and Molecular Biology program fellowship to AB, and a MSU-professorial assistantship to VG.

#### REFERENCES


#### ACKNOWLEDGMENTS

We thank Urs Benning for making the PLAFP-YFP constructs and are grateful to Urs Benning and Olena Tetyuk for critically reading the manuscript. We thank Jie Li and Melinda Frame (Center for Advanced Microscopy at MSU) for assistance with the confocal microscopy.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 00563

Supplementary Table 1 | Primers and conditions used for Cloning, RT-PCR, and qPCR.

Supplementary Figure 1 | GDSL (A), PLAFP (B), and PIG-P (C) expression in 5-week old Arabidopsis plants. Values represent mean and standard error of three biological replicates as determined using semiquantitative RT-PCR.

Supplementary Figure 2 | Hydroponic set-up for abiotic stress treatment. Wildtype seedlings were grown on MS plates for 2 weeks and then transferred to the hydroponic-like set up displayed here. After 24 h of acclimation in their new environment, the abiotic stress treatments were added: osmotic stress received 300 mM Mannitol, salt stress received 150 mM sodium chloride (NaCl), and drought stress signal and mimic in the form of 100 µM abscisic acid (ABA) or 30% polyethylene glycol (PEG), respectively. Seedlings were collected after various time points over a 24 h period. Method adapted from communication with Dr. Patricia Ferreira dos Santos, University of Nevada, Reno.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Barbaglia, Tamot, Greve and Hoffmann-Benning. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

## Comprehensive Analysis of the Membrane Phosphoproteome Regulated by Oligogalacturonides in Arabidopsis thaliana

#### Benedetta Mattei\*, Francesco Spinelli, Daniela Pontiggia and Giulia De Lorenzo

*Dipartimento di Biologia e Biotecnologie Charles Darwin,Istituto Pasteur - Fondazione Cenci Bolognetti, Sapienza University of Rome, Rome, Italy*

Early changes in the *Arabidopsis thaliana* membrane phosphoproteome in response to oligogalacturonides (OGs), a class of plant damage-associated molecular patterns (DAMPs), were analyzed by two complementary proteomic approaches. Differentially phosphorylated sites were determined through phosphopeptide enrichment followed by LC-MS/MS using label-free quantification; differentially phosphorylated proteins were identified by 2D-DIGE combined with phospho-specific fluorescent staining (phospho-DIGE). This large-scale phosphoproteome analysis of early OG-signaling enabled us to determine 100 regulated phosphosites using LC-MS/MS and 46 differential spots corresponding to 34 pdhosphoproteins using phospho-DIGE. Functional classification showed that the OG-responsive phosphoproteins include kinases, phosphatases and receptor-like kinases, heat shock proteins (HSPs), reactive oxygen species (ROS) scavenging enzymes, proteins related to cellular trafficking, transport, defense and signaling as well as novel candidates for a role in immunity, for which elicitor-induced phosphorylation changes have not been shown before. A comparison with previously identified elicitor-regulated phosphosites shows only a very limited overlap, uncovering the immune-related regulation of 70 phosphorylation sites and revealing novel potential players in the regulation of elicitor-dependent immunity.

Keywords: oligogalacturonides, Arabidopsis thaliana, elicitors, DAMPs, phosphoproteomics, immunity, LC-MS/MS, 2DE

#### INTRODUCTION

Plants have developed various mechanisms to defend themselves against biotic stresses. Inducible immune defense responses include phytoalexin accumulation, expression of pathogenesis-related proteins, production of ROS, and, in some cases, programmed cell death (Boller and Felix, 2009). Plant innate immunity is driven by the perception of danger signals mediated by recognition proteins (Chisholm et al., 2006). Pathogen-associated molecular patterns (PAMPs) are conserved molecules secreted or present on the surface of most strains of a given microbial taxonomic group that activate the so-called PAMP-triggered immunity (PTI) against a wide range of pathogens (Barrett and Heil, 2012). Plant immunity also relies on the ability to sense danger by means

#### Edited by:

*Qingsong Lin, National University of Singapore, Singapore*

#### Reviewed by:

*Ning Li, Hong Kong University of Science and Technology, China R. Glen Uhrig, ETH Zurich, Switzerland*

> \*Correspondence: *Benedetta Mattei benedetta.mattei@uniroma1.it*

#### Specialty section:

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

Received: *21 April 2016* Accepted: *12 July 2016* Published: *02 August 2016*

#### Citation:

*Mattei B, Spinelli F, Pontiggia D and De Lorenzo G (2016) Comprehensive Analysis of the Membrane Phosphoproteome Regulated by Oligogalacturonides in Arabidopsis thaliana. Front. Plant Sci. 7:1107. doi: 10.3389/fpls.2016.01107* of endogenous molecular patterns that are present only when the tissue is infected or damaged (damage-associated molecular patterns or DAMPs). In these cases, the discrimination between an intact self and an altered self leads to the activation of the immune system (Benedetti et al., 2015). Recognition of both PAMPs and DAMPs is mediated by the so-called pattern recognition receptors (PRRs; Boller and Felix, 2009).

Oligogalacturonides (OGs) are typical plant DAMPs (Ferrari et al., 2013) produced by the action of fungal endopolygalacturonases (PGs) on homogalacturonan (HGA), the main component of pectin (Bellincampi et al., 2014). The interaction between PGs and the plant polygalacturonaseinhibiting proteins (PGIPs) favors the formation of OGs with degree of polymerization from 10 to 15 that activate the plant innate immunity (Mattei et al., 2005; Casasoli et al., 2009; Kalunke et al., 2015). OGs induce accumulation of phytoalexins, glucanase, and chitinase, ROS production [mediated, in Arabidopsis thaliana, by the RESPIRATORY BURST OXIDASE HOMOLOG D (RBOHD)] and callose deposition (Galletti et al., 2008; Ferrari et al., 2013). Exogenous treatment with OGs protects Arabidopsis and grapevine (Vitis vinifera) leaves against infection with the necrotrophic fungus Botrytis cinerea (Aziz et al., 2004; Ferrari et al., 2007), suggesting that, when PGs are secreted by microbes at the site of infection, this elicitor is likely to be released and contributes to activate defenses responses. Because pectin is one of the most accessible targets for many microbial cell wall–degrading enzymes and among the first structures to be altered during an attempted pathogenic attack, the signaling activity of OGs is an indication that plants have evolved mechanisms to monitor HGA degradation for the early detection of tissue injury (De Lorenzo et al., 2011; Nuhse, 2012; Savatin et al., 2014b). Wall-associated kinase (WAK) receptors are potential candidates to monitor pectin integrity (De Lorenzo et al., 2011; Kohorn and Kohorn, 2012). Indeed, WAK1 has been shown to mediate the perception of OGs (Brutus et al., 2010; Gramegna et al., 2016). OGs may also regulate plant growth and development mainly through their antagonism with auxin (Savatin et al., 2011; Ferrari et al., 2013), demonstrating the potential of this molecule to modulate both developmental and defense-dependent signaling.

Perception of elicitors at the plasma membrane (PM) triggers an intracellular signaling cascade that initiates pathogen defense. Early responses induced by OGs and flg22, a peptide PAMP derived from the bacterial flagellin, largely overlap (Denoux et al., 2008). For example, both activate calciumdependent protein kinase (CDPK; Gravino et al., 2015) and mitogen-activated protein (MAP) kinase cascades (Rasmussen et al., 2012). The MAP triple kinases indicated as Arabidopsis NUCLEUS- AND PHRAGMOPLAST-LOCALIZED KINASE1 related protein kinases (ANPs), and the MAP single kinases MPK3 and MPK6 play a role in the response to OGs and PAMPs, including the bacterial elicitors flg22 and elf18 (Galletti et al., 2011; Savatin et al., 2014a). Nonetheless, distinctive features have been described between responses induced by PAMPs and OGs. Microarray analyses show considerable differences in the late responses to these two classes of elicitors (Denoux et al., 2008), and lack of ANPs strongly reduces phosphorylation of MPK3 and MPK6 induced by OGs but enhances that induced by elf18 (Galletti et al., 2011; Savatin et al., 2014a). Moreover, while the Arabidopsis leucine-rich repeat co-receptors BAK1/SERK3 and BKK1/SERK4 are required to achieve full response to elf18 and flg22, only a subset of defense responses induced by OGs is affected by loss of these elements, pointing to a complexity in the OG signaling pathways that is unique among the characterized MAMPs and DAMPs (Gravino et al., 2016). Specific phosphorylation events might regulate the activation of distinct signaling branches leading to different downstream responses.

In previous works, several OG-regulated proteins in the Arabidopsis apoplast and nucleus have been identified by 2-D DIGE (Casasoli et al., 2007, 2008), many of them present in multiple isoforms likely due to post-translational modifications (PTMs). Here we investigated early phosphorylation events regulated by OGs in a large-scale phosphoproteomic study including membrane proteins, to facilitate the detection of low abundance proteins with a signaling role. Analysis of membrane proteins by 2-D gel electrophoresis is limited by solubility constraints, that have a minor impact on shotgun proteomics (Kleffmann et al., 2007). On the other hand, greater proteome coverage can be reached by using gel-based and gel-free methods as complementary strategies (Zhao et al., 2008; Robbins et al., 2013). For a comprehensive picture of the phosphoproteome we therefore used both LC-MS/MS and the combination of 2-D DIGE with ProQ Diamond staining, which is known to selectively stain phosphoproteins (hereon indicated as Phospho-DIGE; Bond et al., 2011; Liu et al., 2015). In the latter, phosphorylated isoforms are particularly well resolved due to the property of the acidic phosphate group(s) to lower the pI of the proteins, thereby facilitating the detection of low-abundance phospho-isoforms.

#### MATERIALS AND METHODS

### Growth Conditions and Plant Treatments

A. thaliana ecotype Columbia-0 (Col-0) was used for this study. Plants were grown on soil (Einheitserde, Germany) in a climatic chamber at 22◦C and 70% relative humidity. Seedlings were grown in a growth chamber at 21◦C. Sterilized seeds (20 per well) were germinated and grown in 1 mL of liquid growth medium [Murashige and Skoog (MS) medium, pH 5.7, 0.5% sucrose] in 12-well plates.

For proteomic analyses, seedlings were grown in 500-ml flasks containing 100 mL of growth medium. Flasks containing seeds (about 100 seeds per flask) were grown in a growth chamber at 21◦C. After 2 weeks, the culture medium was replaced with fresh medium and seedlings were grown for an additional day. About 10 g (fresh weight) of plant material were obtained from each flask. Plants and seedlings were grown under a 16-h light/8-h dark cycle (∼120 µmol m−<sup>2</sup> s −1 ).

OGs with degree of polymerization of 9–16 were prepared as previously described (Pontiggia et al., 2015). The OG stock solution (10 mg/mL) was filter-sterilized before addition to the medium to a final concentration of 50 µg/ml. For proteomic studies, water- (control), and OG-treated seedlings were harvested 10 min after treatment for protein extraction.

### Preparation of Total Protein Extract and Total Microsomal Fraction

To obtain total protein extracts for Phospho-DIGE analysis, 10 g seedlings were homogenized using mortar and pestle in liquid nitrogen in homogenization buffer (1 M NaCl, 1 mM Na3VO4, 1 mM Na2MoO4, 25 mM NaF, protease inhibitor cocktail, Sigma). Total microsomal fractions (TMF) for both Phospho-DIGE and LC-MS/MS analyses were prepared as previous reported (Alexandersson et al., 2004; Tang et al., 2008). Briefly, 10 g seedlings were grinded using a blender with three impulses of 10 s in lysis buffer containing 1 mM Na3VO4, 1 mM Na2MoO4, 25 mM NaF, protease inhibitor cocktail (Sigma). The mixture was filtered through Miracloth and centrifuged at 10,000 × g for 10 min. Supernatant was then centrifuged at 100,000 × g for 1 h and the pellet (TMF) was recovered.

### H+-ATPase Immunoblotting

Protein samples (4 µg each) were boiled for 3 min in Laemmli buffer. Following gel electrophoresis, proteins were blotted onto nitrocellulose membranes using a Trans-Blot Turbo apparatus (BioRad). Subsequently blot was first incubated with the antibody against the plasma membrane H+-ATPase (1:500 dilution in PBS-T, kind gift of Prof M.I. De Michelis, University of Milan) and then with a horseradish peroxidase-coupled secondary antiserum (Santa Cruz, Biotechnology, Santa Cruz, CA; 1:2000 dilution in PBS-T).

#### MPK3/MPK6 Immunoblotting

Immunodetection of phosphorylated MAPKs MPK3 and MPK6 was performed using a polyclonal phospho-p44/42-ERK MAPKspecific antiserum (Cell Signaling Technology, Danvers, MA). Protein samples (30 µg each) were boiled for 3 min in Laemmli buffer. Following gel electrophoresis, proteins were blotted onto nitrocellulose membranes using a Trans-Blot Turbo apparatus (Biorad). Subsequently blots were first incubated with the phospho-p44/42 MAPK-specific antiserum (1:2000 dilution in TBS-T at 4◦C overnight) and then with a horseradish peroxidase-coupled goat anti-rabbit secondary antiserum (Santa Cruz, Biotechnology, Santa Cruz, CA; 1:5000 dilution in PBS-T). Membrane was stripped and re-probed with anti-MAPK3 and anti-MAPK6 (Sigma) and developed as reported above.

### Protein Digestion and Phosphopeptide Enrichment for LC-MS/MS Analysis

TMF from control and OG-treated samples of two independent biological replicates were resuspended in 100 µL of freshly prepared 8 M urea, 10 mM Tris-HCl pH 8.0, 5% sodium deoxycolate. Protein concentration was determined using the BCA Protein Assay Kit (Sigma). For each sample, 150 ug of proteins were subjected to reduction with 10 mM dithioreitol for 45 min at 56◦C followed by alkylation of cysteines with 50 mM iodoacetamide for 25 min at room temperature in darkness. Proteolytic digestion was carried out overnight with proteomics grade Lys-C (Promega, Lys-C:protein ratio 1:50) and for additional 4 h with trypsin (Promega, trypsin:protein ratio 1:50) at room temperature. The digestion mixture was subsequently acidified with 1% (v/v) formic acid and centrifuged to remove insoluble material. Peptides were desalted using home-made microcolumns using R3 beads (Poros) packed in gel loader tips. Phosphopeptide enrichment was performed using a sequential elution from immobilized metal affinity chromatography (SIMAC) as previously described (Thingholm et al., 2008). The procedure includes an initial phosphopeptide enrichment step by IMAC, in which bound peptides are sequentially eluted with an acidic and a basic solution. Next, the IMAC flow-through fraction and the acidic elution fractions are subjected to a second phosphopeptide enrichment step using TiO<sup>2</sup> chromatography. The TiO2-enriched fraction and the basic elution fractions from IMAC were lyophilized and analyzed by LC-MS/MS. For proteomic analysis, also flow-through fractions from TiO<sup>2</sup> chromatography were desalted on home-made R3 stage tips and analyzed by LC-MS/MS.

### Proteomic and Phosphoproteomic Analysis by LC-MS/MS

Peptides were separated on a 15 cm PicoFrit Column (75 µm i.d., New Objective, Woburn, MA) packed with Magic C18AQ (5 µm, 200 Å, Michrom) and analyzed on a LTQ-Orbitrap Discovery (Thermo Fisher Scientific) coupled on-line with a nano-HPLC (Ultimate 3000, Thermo Fisher Scientific). Peptides were eluted using a 0–60% acetonitrile gradient in 0.1% formic acid (180 min, 300 nl/min). MS was acquired at 30,000 FWHM resolution in the FTMS (target value of 5 × 10<sup>5</sup> ions) and MS/MS was carried out in the linear ion trap, with MS/MS on top 5 ions and multistage activation.

### LC-MS/MS Data Analysis

MS data were processed using MaxQuant (Cox et al., 2014) version 1.5.2.8 with a false-discovery rate (FDR) < 0.01 at the level of proteins, peptides and modifications, using default settings. Oxidized methionine, acetylation (protein N-term) and phospho(STY) were selected as variable modifications, and carbamidomethyl cysteine as fixed modification. Proteins were identified using a target-decoy approach with a reversed database, using the Andromeda search engine against the Arabidopsis UniProt database (release-2014, 31565 entries) and LFQ quantification for phosphoproteomics and proteomics was performed by MaxQuant.

Statistical analysis was performed with Perseus (version 1.5.0.31). Hits to the reverse database and contaminants were filtered out, intensities were normalized for unequal protein amounts and log2-transformed. For proteomic analysis quantification was performed on proteins identified with a minimum of 2 unique peptides and at least three valid values among the replicates. Significance was assessed by Student'st-test, using permutation-based FDR to control for multiple hypothesis testing (Data Sheet S2B in Supplementary Material). For the phosphoproteomic study, we first filtered to retain only class I phosphosites (localization probability > 0.75 and score difference > 5). We next considered (1) phosphosites that were exclusive for either the OG-treated samples or the control samples with

two valid values in either the control or the OG-treated group; (2) phosphosites that had at least three valid values (control + OG-treated). These sites are listed in Data Sheet S1B in Supplementary Material. For phosphosites that had three valid values only, first, missing values were imputed with values representing a normal distribution around the detection limit of the mass spectrometer (downshift =1.5; width = 0.3), to allow statistical analysis; then Significance B-values for each replicate (OG-treated vs. control) were calculated using the statistical tool in Perseus (Cox and Mann, 2008). Significance B-values represent outlier probability score weighted for the intensity. Besides the exclusive ones, phosphosites with a significance B p ≤ 0.05 in both replicates were considered as significantly regulated. In addition, a two-sample student's t-test was performed, and phosphosites that had a significance B ≤ 0.05 in one replicate and passed the t-test with p < 0.05 were also considered as significantly regulated.

### 2D DIGE, Pro-Q Diamond Staining, and Image Acquisition

The Clean-up kit (GE Healthcare) was used to eliminate salts and concentrate proteins in both the total protein extracts and the TMF. The protein-containing pellets were finally dissolved in a small volume (50–100 µL) of IEF buffer containing 8 M urea, 2 M thiourea, and 4% (v/v) CHAPS, centrifuged and the supernatants recovered. Protein content was measured using the BIO-RAD protein assay according to the manufacturer's instructions. Total protein extracts and TMF were labeled and analyzed using the same procedure: control and OG-treated protein samples from 3 independent biological replicates (C1, C2, C3; T1, T2, T3) were labeled with Cy3 or Cy5, according to the manufacturer's instructions for minimal labeling (Minimal labeling kit, GE Healthcare) and randomization (ETTAN DIGE System, GE Healthcare). Briefly, each sample (50 µg) was labeled with 200 pmol CyDye DIGE Fluor minimal dyes (GE Healthcare) and incubated on ice in the dark for 30 min. The internal standard (IS) was labeled with Cy2 and consisted of a pooled sample comprising equal amounts of each sample used for each replicate (C1+C2+C3+T1+T2+T3). Equal volumes of a 2X buffer [8 M urea, 2 M thiourea, 2% ASB-14 (or 4% CHAPS; v/v), 20 mg/mL DTT, and 2% IPG buffer/Pharmalytes 4-7] were added to each labeled protein samples. Samples (C+T+IS) were pooled prior to IEF, which was carried out using non-linear IPG strips (pH 4–7, 13 cm, GE Healthcare) rehydrated overnight at room temperature. IEF was performed using the Ettan IPG-phor apparatus (GE Healthcare) as follows: (1) step to 500 V (0.5 kVh); (2) step to 1000 V (0.8 kVh); (3) step to 8000 V (11.3 kVh); (4) step to 8000 V (3.0 kVh), for a total of 25 kVh at 20◦C and a maximum current setting of 50 µA per strip. After IEF, the strips were equilibrated for 15 min in 100 mM Tris pH 6.8, 30% v/v glycerol, 8 M urea, 1% w/v SDS, 0.2 mg/mL bromophenol blue, 5 mg/mL DTT for reduction of disulfide bridges, and alkylated for 15 min in the same equilibration buffer containing 25 mg/mL iodoacetamide. Each strip was then loaded on top of a 12% w/v acrylamide gel for the second dimension. SDS-PAGE was carried out using the Hoefer SE 600 Ruby apparatus (GE Healthcare) at 100 V and 20 mA per gel for 15 min and then 100 V and 40 mA per gel until the bromophenol blue dye front had run off the bottom of the gel. A running buffer of 25 mM Tris pH 8.3, 192 mM glycine, and 0.1% w/v SDS was used.

Each gel was scanned using a Typhoon 9200 imager (GE Healthcare) set at the wavelengths corresponding to each CyDye [532 nm laser and 580 nm band pass (BP) 30 emission filter for Cy3; 633 nm laser and 670 nm BP30 emission filter for Cy5; 488 nm laser and 520 nm BP40 emission filter for Cy2], at high resolution (100 µm). The photo multiplier tube (PMT) voltage was adjusted for each channel between 500 and 700 V to ensure that the image intensity was within a linear range between 40,000 and 60,000 U. For phosphoprotein detection, each gel was then fixed, post-stained with Pro-Q Diamond phosphospecific fluorescent dye (Invitrogen; Bond et al., 2011; Liu et al., 2015) applying the same scanning conditions described above. Pro-Q Diamond and Cy3 have very similar spectra of fluorescence with excitation/emission maxima at 555/580 and 553/569 nm for Pro-Q Diamond and Cy3, respectively. Therefore, phosphoproteins were detected as spots with increased Cy3 fluorescence (Cy3 + Q) in comparison to original Cy3.

#### Phospho-DIGE Statistical Analysis

For data normalization and analysis, gel images acquired before and after Pro-Q Diamond post-staining were compared using the Differential In-gel Analysis (DIA) module of the DeCyder software version 6.5 (GE Healthcare) for co-detection of the three CyDye-labeled forms of each spot and calculation of the ratios between sample and internal standard abundance. Phosphoproteins were detected as spots with increased Cy3 like fluorescence due to ProQ-staining in comparison with original Cy3 emission (Stasyk et al., 2005). Statistical analysis of protein abundance changes between control and treated samples from the three independent biological replicates was performed in the Biological Variation Analysis (BVA) module for quantitative comparisons of protein abundance/phosphorylation across multiple gels. The inter-gel variability was corrected by normalization of the Cy2 internal standard spot map present in each gel. Protein spots whose intensities were significantly different among the control and treated samples were determined by paired one-way ANOVA. Proteins were considered differentially expressed when the FDR-corrected pvalues of the ANOVA analysis were <0.05.

### Protein Spot Identification by MALDI-TOF Mass Spectrometry

For mass spectrometry analysis of proteins identified as differentially phosphorylated by Phospho-DIGE experiments, a preparative gel was run under the same IEF and SDS-PAGE conditions used for the DIGE gels, loading 250 µg each of control and OG treated samples mixed together. Proteins were subsequently visualized using Coomassie Brilliant Blue (CBB) R-250 stain (Sigma) according to the manufacturer's instructions. The preparative gel image was matched to the master gel image (the gel with the highest spot count in the DIGE analytical gel match-set) using DeCyder software. Matching was further improved by land marking and manually confirming potential

spots of interest. By comparing the CBB-stained spot pattern with the corresponding Cy5 protein pattern, spots of interest showing differential fluorescent levels on the 2-D DIGE gels were picked manually from the preparative gel and subjected to trypsin in-gel digestion as previously described (Casasoli et al., 2007).

Protein identification was carried out using a Voyager-DE STR instrument (Applied Biosystems, Framingham, MA). Peptides were desalted using ZipTip C18 microcolumns (Millipore, Bedford, MA, USA) and spotted onto a MALDI target plate using CHCA as the matrix (10 mg/ml α-cyano-4-hydroxy-cinnamic acid in 0.1% TFA, 50% ACN). The mass spectrometer was operated in the positive ion, delayed extraction (200 ns delay time) reflector mode with an accelerating voltage of 20 kV. Each MALDI-TOF spectrum was generated by accumulating data corresponding to 200–500 laser shots. Internal mass calibration was performed by using theoretical masses of the trypsin autodigestion peaks. The mass list was then analyzed using the PeakErazor software (http://www.welcome. to/GPMAW) to eliminate contaminant peaks (keratin, trypsin added for digestion and peaks present in all mass spectra). Proteins were identified by searching the National Center for Biotechnology Information (NCBI) database using the MASCOT (www.matrixscience.com) or Aldente (xpasy.org/tools/aldente) search engines with the following criteria: cleavage by trypsin (one missed cleavage allowed), mass tolerance 30 ppm; carbamidomethyl cysteine as fixed modification, methionine oxidation, and phospho(STY) as variable modifications. Only matches with a Mascot score higher than 60 (p < 0.05; or Aldente score higher than 13.86, p < 0.05), sequence coverage higher than 10% and more than 6 peptide matches were considered significant. A number of photosynthetic proteins were identified as potentially regulated phosphoproteins. Because they corresponded to highly abundant protein spots amenable to mis-quantification, they were excluded from our dataset.

### Bioinformatic Analysis of Phosphoproteins

The PhosPhAt (http://phosphat.mpimp-golm.mpg.de/; Durek et al., 2010) and P3DB (http://www.p3db.org/index.php; Yao et al., 2014) databases were searched to determine if phosphorylation sites had been previously reported for the identified proteins. For the analysis of significantly overrepresented GO terms, differentially phosphorylated proteins were analyzed using AgriGO (http://bioinfo.cau.edu.cn/agriGO/ analysis.php) for Singular Enrichment Analysis (SEA) using A. thaliana TAIR10 as selected species and Arabidopsis genome locus (TAIR10) as selected reference, Fisher as a statistical test method, Yekutieli (FDR under dependency) as z multi-test adjustment method, 0.05 as significance level, 5 as minimum number of mapping entries and complete GO as Gene Ontology type. In order to avoid redundancy at the protein level, in the case of protein groups we considered only the first isoform.

To investigate possible interactions between the OG-regulated phosphoproteins, the STRING (Search Tool for the Analysis of Interacting Genes/Proteins) algorithm was used for the creation of protein interaction networks based on published functional or informatics-predicted interactions (Szklarczyk et al., 2015). The evidence annotation in STRING was filtered out of interactions from text-mining and neighborhood, and only interactions supported by experimental evidence, co-expression and existing database information, with high-confidence score >0.7 were considered (Table S6). The SUBA3 database (http:// suba3.plantenergy.uwa.edu.au/; Heazlewood et al., 2005) was used to assign the subcellular localization of differentially phosphorylated proteins.

### RESULTS

### OG Treatment Induces Early Changes in the Phosphoproteome of Arabidopsis Seedlings

In Arabidopsis, treatment with OGs induces the activation, by phosphorylation, of MPK3 and MPK6 within few minutes, a temporal kinetics similar to that observed upon treatment with PAMPs (Galletti et al., 2011). Two-week-old liquid-grown Arabidopsis ecotype Columbia 0 (Col-0) seedlings were treated with OGs (50 µg/ml) or water as a control. The 10-min time point was chosen for our analyses, because close to that used in previous studies on elicitor-induced phosphorylation (Benschop et al., 2007; Nuhse et al., 2007; Rayapuram et al., 2014). Western blot analysis using α-phospho-p44/42-ERK antibody showed the OG-induced phosphorylation of MPK3 and MPK6 (**Figure 1A**). TMF was obtained to facilitate the identification of less abundant membrane proteins. Enrichment of total membranes was determined by western blot using antibodies specific for the PM H+-ATPase (**Figure 1B**). In addition, total protein extracts were obtained, for Phospho-DIGE analysis only.

#### Phosphoproteomic Analysis of TMF by LC-MS/MS

TMF preparations from seedlings treated with OGs or water as a control were subjected to the SIMAC phosphopeptide enrichment method. The enriched fractions were analyzed by nano-LC-MS/MS, leading to the identification of a total of 2147 unique phosphosites [Data Sheet S1C in Supplementary Material]. We could quantify 1026 phosphosites (Data Sheet S1B in Supplementary Material). A total of 99 phosphosites significantly changed phosphorylation level upon OG treatment; among them, 29 and 36 phosphosites were unique to control and OG-treated seedlings, respectively (**Table 1**).

Proteomic analysis of the TMF did not show substantial changes in overall protein abundance (Data Sheet S2B in Supplementary Material). This is expected, as a significant de novo protein synthesis or protein degradation is unlikely to occur within 10 min of elicitor treatment (Benschop et al., 2007). Identified proteins (Data Sheet S2A in Supplementary Material) were classified according to known gene ontology using AgriGO (http://bioinfo.cau.edu.cn/agriGO/analysis.php). GO term analysis of the cellular component of the total list of proteins demonstrated a clear enrichment for PM-associated proteins, cell wall and intracellular organelles (**Figure 1C**).

MAPKs MPK6 and MPK3 upon treatment of seedlings for 10 min with OGs by immunoblot analysis using an anti-p44/42-ERK antibody (top panel). Thirty micrograms of each protein sample were loaded. Levels of MPK3 and MPK6 total proteins were determined using specific antibodies (bottom panel). The identity of individual MAP kinases, as determined by size, is indicated by arrows. (B) Western blot analysis of total protein extracts, TMF proteins directly solubilized in Laemmli buffer (TMF) or immediately before DIGE analysis (DIGE-TMF), probed with antibody against the plasma membrane H+-ATPase. Four micrograms of each protein sample were loaded. (C) The analysis of enriched GO terms for the cellular component category within the TMF proteins identified (listed in Data Sheet S2A in Supplementary Material) using the AgriGO database. Barplot shows selected GO terms significantly enriched (FDR < 0.05, Yekutieli adjusted). The background/reference represents the proportion of all annotated genes of each GO term within the total genes in the TAIR10 database.

## Identification of OG-Regulated Phosphoproteins by Phospho-DIGE

In a second approach, phospho-DIGE was exploited for detecting changes in protein phosphorylation using both TMFs and total protein extracts. Three independent biological replicates were performed for each treatment and samples were labeled according to the randomization scheme shown in Table S3, to enforce statistical analysis. About 800 fluorescent spots were detected for the total extracts, and about 500 spots for the TMFs. Statistical analysis revealed 30 spots in the total extract and 50 spots in the TMF fraction that changed their phosphorylation status (p < 0.05) after treatment with OGs (representative gel images are shown in Figures S1, S2). The spots were analyzed by MALDI-TOF-MS, leading to the identification of 46 differential phospho-isoforms, corresponding to 34 phosphoproteins (**Table 2**). Representative images showing spots that exhibited an increase or decrease in phosphorylation upon OG treatment, along with representative 3D images to visualize the phosphorylation changes, are shown in **Figure 2**. Proteomic analysis showed that, again, differences in nonphosphorylated protein abundance between OG-treated and control samples were not significant, demonstrating that most changes occurred at the PTM level.

### Computational Analysis of the Differentially Phosphorylated Proteins

To summarize, LC-MS/MS analysis of TMF led to identification of 58 and 42 sites exhibiting an increased or decreased




**55**


Frontiers in Plant Science | www.frontiersin.org August 2016 | Volume 7 | Article 1107 |

**56**




*not* 

 *na, not* 

> >

#### TABLE 2 | List of 2D-DIGE identified phosphoproteins regulated by OGs.



*<sup>a</sup>Full name of the identified protein. Proteins that have never been described before as phosphorylated are underlined.*

*b ID of the identified protein from the TAIR database (The Arabidopsis Information Resource database. www.arabidopsis.org).*

*<sup>c</sup>Subcellular localization, obtained from SUBA (the SUBcellular localization database for Arabidopsis proteins, http://suba3.plantenergy.uwa.edu.au/). PL, plastid; EX, extracellular; CY, cytosol; PM, plasma membrane; ER, endoplasmic reticulum; GO, Golgi apparatus; VO, vacuole; NU, nucleus; PX, peroxisome; MI, mitochondrion; CS, cytoskeleton.* § *indicates localization determined by GFP fusion.*

*<sup>d</sup>Numbers correspond to spots shown in Figure S1 (total protein extracts, indicated by* \**) or in Figure S2 (total microsomal fraction).*

*<sup>e</sup>Value obtained from MASCOT (http://www.matrixscience.com) reported as a measure of the statistical significance of a match (*>*60). In parenthesis are shown the values obtained from ALDENTE (ftp.expasy.ch/tools/aldente) independently (significance*> *13.86).*

*<sup>f</sup>Percentage of protein sequence covered by identified peptides.*

*<sup>g</sup>Numbers of different identified peptides.*

*<sup>h</sup>ANOVA: the Student's t-test p-value represents the probability of obtaining the observed ratio if control and OG-treated spots have the same protein abundance. Significant values (p* < *0.05) are reported.*

*<sup>i</sup>Fold change is calculated as the ratio of the average standardized abundances corresponding to OG-treated and control spots after ProQ Diamond staining. Positive and negative values are indicated for increases and decreases in phosphorylation state, respectively, upon treatment.*

*<sup>j</sup>Fold change of transcript levels after OG treatment at 1 and 3 h with respect to the mock-treated control (Denoux et al., 2008),. Only data of genes for which fold change is* ≥*1.5 and significant in at least one treatment are reported. ns, not significant (P* > *0.01). na, not applicable.*

phosphorylation, respectively (**Table 1**), whereas Phospho-DIGE analysis of both TMF and total extracts revealed a total of 32 and 14 phosphoprotein spots showing an increased or decreased phosphorylation, respectively, corresponding to 34 proteins due to the presence of multispots (**Table 2**). Only Patellin 2 (PATL2) was found to be differentially phosphorylated in both analyses of the TMF. Two spots are shown for PATL2 (**Figure 2** and Figure S2); one exhibits increased phosphorylation in response to OGs, in agreement with our detection of a phosphopeptide exclusively present in the OG-treated sample (**Table 1**). The other PATL2 spot shown in **Figure 2** shows a slight but significant dephosphorylation (−1.15-fold). Many studies reported comparative analyses using both gel-based and gel-free methods, where the match of identified proteins is very low, and emphasized that these two strategies are indeed complementary (Kleffmann et al., 2007; Zhao et al., 2008; Robbins et al., 2013).

Previously described phosphorylation sites were searched using the PhosPhAt (http://phosphat.mpimp-golm.mpg.de/) and P3DB (http://www.p3db.org/) databases. Among all the differential phosphoproteins, sixteen [underlined in **Table 1**

FIGURE 2 | Gel visualization of selected spots from DeCyder analysis. Representative images showing differentially phosphorylated spots upon OG treatment for each protein, gel images of the spots, along with the corresponding 3D images to visualize the phosphorylation changes, are indicated by purple circles. Two spots are shown for PATL2; one (left) exhibits increased phosphorylation in response to OGs, while the other one (right) shows a slight but significant dephosphorylation (−1.15-fold; see Table 2).

(9 proteins) and **Table 2** (7 proteins)] have never been described to be phosphorylated at the identified sites. GO term enrichment analysis with AgriGO was performed to identify significantly over-represented biological process categories in the subset (**Tables 1, 2**) of OG-regulated phosphoproteins (**Figure 3**). The GO terms referring to response to various biotic and abiotic stimuli were significantly enriched in the analyzed subsets of differential phosphoproteins, showing that the molecular processes involved in the response to OGs indeed correlate with a condition of stress. Proteins with increased phosphorylation were specifically enriched in terms referring to Signal transduction and to Cell death, while decreased phosphorylation occurred mainly in proteins associated with Response to biotic stimulus (**Figure 3**).

Connections between all the quantified phosphorylated proteins were searched by using the STRING algorithm (**Figure 4**). Proteins with differentially regulated phosphosites are highlighted with colors corresponding to different functional categories. Many clusters were identified, with several interesting networks standing out. The highly interconnected "transport" cluster includes several primary electrogenic proton pumps found in all eukaryotes: the PM H+-ATPases AHA1 and AHA2, different subunits of the vacuolar H+-ATPase (V-ATPase or VHA; subunits A, C, and E1; Batelli et al., 2007) as well as the subunit beta-1 of the mitochondrial ATP synthase. The V-ATPase subunit C, encoded by the single gene DE-ETIOLATED3 (DET3), is directly connected with the other V-ATPase subunit E1, but also with AHA1 and AHA2.

Multiple interactions are evident among elements involved in signaling (**Figure 4**), such as kinases, receptor kinases, and phosphatases. Many kinases appear to be connected to PP2C-type phosphatase-like protein (At2g20050), an enzyme that acts as negative modulator of protein kinase pathways involved in stress and developmental processes (Kuhn et al., 2006). Moreover, an additional smaller cluster includes several heat-shock proteins (HSP90-1, HSP70-1, HSP93-V, HSC70-1, BiP1) that are connected also to ribosomal proteins (**Figure 4**). Proteins involved in intracellular trafficking, namely syntaxin

FIGURE 3 | Analysis of enriched GO term distributions within the differentially phosphorylated proteins. The analysis was performed on the differentially phosphorylated proteins listed in Table 1 (LC-MS/MS) and Table 2 (Phospho-DIGE) using the GO annotations of biological process in the AgriGO database (Y axis). Barplot shows selected GO terms significantly enriched (FDR < 0.05, Yekutieli adjusted) for proteins with increased or decreased phosphorylation. The background/reference represents the proportion of all annotated genes of each GO term within the total genes in the TAIR10 database.

phosphorylated proteins listed in Tables 1, 2 are highlighted and color coded based on functional categories.

SYP 121 and 122, a calmodulin-binding protein (EDA39), PCaP1 (Plasma membrane-associated cation-binding protein 1), and a phospholipase like protein (PEARLI 4f) form an additional cluster.

### DISCUSSION

Previous proteomic studies have either postulated (Chivasa et al., 2006; Nuhse et al., 2007) or shown (Jones et al., 2006) that PTM-mediated regulation is important for plant immunity. Here we have studied very early protein dynamics induced by OGs, an important class of DAMPs, employing two complementary proteomic approaches, SIMAC phosphopeptide enrichment followed by LC-MS/MS and Phospho-DIGE (a gel-based approach). These analyses led to the identification of 100 regulated phosphosites and 46 differentially phosphorylated protein spots, respectively (**Tables 1, 2**). The functional classification of the OG-regulated phosphoproteins, based on the GO functional categories, identifies typical major PM functions. Three main categories emerged: transporter proteins, signaling proteins (receptors, kinases, phosphatases) and proteins involved in membrane trafficking.

A comparison of the OG-regulated phosphoproteome identified in this study with that regulated by flg22, which includes 127 phosphoproteins identified in three large-scale proteomics works (Nuhse et al., 2007; Benschop et al., 2007; Rayapuram et al., 2014), shows only 18 shared phosphosites, corresponding to 14 phosphoproteins (Table S4). Among these, there are MPK6, which shows OG-induced phosphorylation at the same residues that undergo flg22-induced phosphorylation (i.e., Thr221 and Tyr223), RBOHD and Penetration3 (PEN3), an ATP-binding cassette (ABC) family 36 transporter that undergoes phosphorylation also in response to H2O<sup>2</sup> and methyl jasmonate (Stecker et al., 2014). ABC transporters are integral membrane proteins that transport a wide variety of substrates, such as ABA, auxin, and some plant secondary metabolites across cellular membranes (Kuromori et al., 2010).

Interestingly, twenty-two of the OG-regulated phosphoproteins identified here are reported as targets of MPK3 and MPK6 (Table S5; Lee et al., 2015). Among these, the putative tyrosine phosphatase PTEN2a shows OG-dependent increased phosphorylation. This enzyme has been shown to dephosphorylate also the 3′ phosphate group of PI3P (phosphatidylinositol 3-phosphate) in vitro and to possess strong binding affinity for PA (phosphatidic acid), an important second messenger with roles in stress and hormonal signaling (Pribat et al., 2012). Lipid signaling pathways activated downstream of the MAPKs may therefore be involved in the response to OGs. A phosphotyrosine site that disappears in response to OGs and may be a target of PTEN2a is present in the GSK/Shaggy-like kinase/ASK1 at position 229. Auto-phosphorylation of this residue is required for ASK1 trans-phosphorylation activity (de la Fuente van Bentem et al., 2008) and activation of the cytosolic glucose-6-phosphate dehydrogenase (G6PD), which is essential for maintaining the cellular redox balance (Dal Santo et al., 2012). ASK1 dephosphorylation may thus lead to reduced G6PD activity and elevated ROS levels that contribute to the elicitor-induced oxidative burst.

MPK3 and MPK6 activation promotes also the biosynthesis of indole glucosinolates and the conversion of indole-3 yl-methylglucosinolate (I3G) to 4-methoxyindole-3-ylmethylglucosinolate (4MI3G), through the phosphorylation of ETHYLENE RESPONSE FACTOR 6 (ERF6). 4MI3G is a substrate of the atypical myrosinases PEN2 and PEN3, which shows increased phosphorylation in response to OGs, and is converted to extracellular unstable anti-microbial compounds, possibly isothiocyanates (Xu et al., 2016). OGs may therefore induce production of indole glucosinolates, but this aspect has been scarcely investigated so far.

A few differentially phosphorylated proteins identified here show regulation also at the level of gene expression, with more than a 1.5-fold change in the corresponding transcript levels within 3 h (**Tables 1, 2**; Denoux et al., 2008), pointing to a regulation at two different levels. Most of the identified proteins, however, show mainly a PTM-based regulation, by phosphorylation/dephosphorylation, suggesting that only activation/inactivation or stabilization/degradation, but not de novo protein synthesis, is necessary to exert their physiological role.

### Signal Transduction Proteins

Several identified proteins are likely players in signaling. Among the proteins involved in regulation through phosphorylation/dephosphorylation, we found members of the Receptor-Like Kinases (RLK) family, including members of the Leucine-Rich Repeat RLK (LRR-RLK) subfamily, cyclin-dependent protein kinases, PTEN2a, and the MAP triple kinase MKKK7 (At3g13530, also known as MAP3Kepsilon 1). MKKK7 has been described to interact with FLS2 and to be phosphorylated at Ser452/Ser854 in response to flg22, functioning as a negative regulator of flg22-triggered MPK6 activation (Mithoe et al., 2016). We find, instead, dephosphorylation of MKKK7 on Ser788 after OG perception, suggesting a distinct event in the signaling cascades initiated by the two elicitors. Another immune-related protein, the Cysteine-rich Receptor-like protein Kinase 36 (CRK36), exhibits decreased OG-dependent phosphorylation, along with upregulation at the transcript level. CRK36 interacts with FLS2 and its overexpression leads to enhanced PTI and resistance to virulent bacteria Pseudomonas syringae pv. tomato DC3000, constitutive accumulation of callose and constitutive stomatal closure (Yeh et al., 2015). A role of CRK36 in resistance to insects has also been proposed (Barah et al., 2013). Moreover, CRK36 silencing leads to higher sensitivity to ABA and osmotic stress during the post-germination growth phase, suggesting that the protein negatively controls ABA signaling (Tanaka et al., 2012).

We also identified three phospholipases (PLDγ1 and two putative phospholipases, pEARLI4, and pEARLI4-f) as well as CBL9, a myristoylated PM-localized protein that does not possess enzymatic activity and contains four EF hands (Batistic et al., 2008). CBL9 modulates ABA responses (Kim et al., 2003) and forms complexes with CBL-interacting protein kinases (CIPKs), leading to phosphorylation of AHA2 (Fuglsang et al., 1999) and regulation of V-ATPases (Tang et al., 2012), also found among the OG-regulated phosphoproteins. CBL-CIPK complexes act as a two-module Ca2+−-decoding system where CBL phosphorylation is required for CIPK-mediated phosphorylation of target proteins (Steinhorst and Kudla, 2013) and regulation of channel activity, ion fluxes and ROS formation during environmental adaptation reactions (Kurusu et al., 2015).

It is worth noting that remorin REM1.3, identified as phosphorylated on Thr58 in response to both OGs and flg22 (Kohorn et al., 2016; Benschop et al., 2007), was also identified in our analysis as phosphorylated at the same site, but with a Significance B p-value of 0.1. A tomato remorin was the first protein described to bind and to be phosphorylated in response to OGs (Reymond et al., 1996) and potato REM1.3 has been shown to accumulate in discrete peri-haustorial domains and enhance susceptibility to Phytophthora infestans (Bozkurt et al., 2014). Arabidopsis REM1.3 is differentially recruited to detergent-insoluble membranes/detergent-resistant membranes (DIMs or DRMs), a term that indicates lipids and proteins that have been biochemically associated with sphingolipid (SL)- and sterol-enriched membranes (Tapken and Murphy, 2015).

Two proteins functioning in the transduction of the light signal, PHOT2 and Phytochrome Kinase Substrate 2 (PKS2), exhibit increased phosphorylation in response to OGs. pks2 loss-of-function mutants show altered expression patterns of auxin marker genes (Kami et al., 2014) and it is tempting to speculate that PKS2 is involved in auxin OG/antagonism, the molecular basis of which is still unknown (Savatin et al., 2011). Notably, the auxin-response factor ARF1, a repressor ARF (Chandler, 2016), showed decreased phosphorylation upon OG treatment at two novel phophosites, the adjacent residues Ser399 and Ser400 located in middle region of the protein thought to function as the repression domain (Ulmasov et al., 1999). Ser405, close to this novel dual phosphorylation site, has been described as phosphorylated in seedlings (Wang et al., 2013) but does not appear to be, regulated by OGs. Very little is known about how phosphorylation controls ARF activity. ARF2, also a repressor ARF that is phylogenetically sister to ARF1, loses its DNA-binding capability upon phosphorylation by the brassinosteroid-regulated kinase BIN2 (BRASSINOSTEROID-INSENSITIVE2), providing an example of hormonal crosstalk at the post-translational level (Vert et al., 2008). On the other hand, phosphorylation of the activator ARFs ARF7 and ARF19 increases their DNA binding ability and enhances transcription (Cho et al., 2014). In the light of the OG-auxin antagonism, OGinduced phosphorylation of ARF1 may lead to its activation as a repressor.

#### Trafficking-Related Proteins

OGs induce increased phosphorylation of cytoskeletonrelated phosphoproteins such as myosin-17 (XIK), Actin7 (ACT7), and PCaP1. Myosin XIK is required for normal movements of actin filaments (Park and Nebenfuhr, 2013), turnover of actin and trafficking of vesicles carrying noncellulosic cell wall components (Cai et al., 2014). The protein acts as a key regulator of plant antifungal immunity and contributes to cell wall-mediated penetration resistance, including deposition of callose and accumulation of PENETRATION1 (PEN1)/SYP121 and lignin-like compounds at the penetration site, when highly expressed (Yang et al., 2014). Pharmacological inhibition of myosins affects accumulation of PEN3 more than that of PEN1, suggesting that transport pathways mediating recruitment and export of the two proteins to apoplastic papillae are distinct (Underwood and Somerville, 2013). OG-induced phosphorylation of myosin XIK may regulate the motility of secretory vesicles and PEN recruitment.

ACT7 is one of the three vegetative and partially redundant isoforms of the actin family and plays a role in germination and root growth (Gilliland et al., 2003; Kijima et al., 2015). Expression of ACT7 strongly responds to auxin and wounding and is required for callus formation (Kandasamy et al., 2001); as shown in **Figure 4**, it is also coordinated with the expression of GAPC1 (Glyceraldehyde-3-phosphate dehydrogenase), the product of which also shows increased phosphorylation in response to OGs. How phosphorylation regulates actin function in plants is not known, except for Mimosa pudica actin, which is heavily tyrosine-phosphorylated, and dephosphorylated at different extent depending on the degree of bending of the petiols (Kameyama et al., 2000).

PCaP1 is capable of binding actin and calcium, leading to the destabilization of actin filaments, besides phosphatidylinositol phosphates (Qin et al., 2014). It is localized via N-myristoylation on the PM, from which it dissociates at increased Ca2<sup>+</sup> levels (Li et al., 2011). The protein negatively regulates hypocotyl (Li et al., 2011) and pollen growth (Qin et al., 2014); moreover, it is induced by brassinosteroids and its overexpression partially rescues the morphological phenotype of the det2 mutant, defective in BR biosynthesis (Tang et al., 2008). PCap1 has never been implicated in immunity.

Components of the microtubule (MT) apparatus, namely TUBULIN ALPHA-5 (TUA5) and the 65-kDa MT-associated protein 1 (MAP65-1), a structural element involved in MT bundling and cell division, also show OG-regulated increase in phosphorylation. MTs are disrupted or rearranged in response to elicitors and stable MTs negatively affect defense (Guan et al., 2013). Phosphorylation of MAP65-1 was found exclusively in OG-treated samples, at residues located within the second MTinteracting region. Phosphorylation of this region has been shown to diminish the interaction with MTs, also in tobacco (Guan et al., 2013) and involve CDPKs and MAPKs (Smertenko et al., 2006), which are known to participate in OG signaling cascade (Ferrari et al., 2013; Gravino et al., 2015).

Vesicle trafficking appears to be another important target of the OG regulatory action. SYP122 decreases phosphorylation in response to OGs, while SYP121/PEN1 (Reichardt et al., 2011) as well as PATL2, PATL3, and PATL4 show an opposite behavior. PATL1 is essential for cell plate formation and maturation at late telophase, when dynamic vesicle trafficking and vesicle fusion occur (Peterman et al., 2004); PATL2 is related to the phosphatydylinositol transfer protein Sec14, which plays a role in the interplay between lipid metabolism and membrane trafficking, whereas PATL3 has been shown to interact with the movement protein of Alfalfa Mosaic Virus and interfere with viral movement (Peiro et al., 2014).

A phosphosite of the kinesin motor protein KAC1 (At5g10470) was found only in the presence of OGs. KAC1 is phoshorylated also in response to flg22 (Table S4) and xylanase treatment (Benschop et al., 2007). Kinesins function in vesicle transport along MTs and their phosphorylation is thought to be required for correct folding and dimerization. The C-terminal domain of KAC1 and KAC2, interact with F-actin in vitro and mediate chloroplast movement (Suetsugu et al., 2010). Both proteins belong to the kinesin-14 subfamily (Zhu and Dixit, 2012), greatly expanded in land plants and representing the largest Arabidopsis kinesin subfamily (21 members). Expansion is attributed to the lack of the MT-based motor protein dynein in plants, and some of the kinesin-14 subfamily members may substitute for dynein to perform retrograde transport functions (Muller, 2015).

### ATPases and Other Membrane Transporters

Initial pathogen recognition occurs at the PM and many of the early responses involve membrane transport processes. For example, activity of PM AHAs is dynamically regulated during plant immune responses and multiple pathogens target this family of enzymes. AHAs are responsible for creating and maintaining a negative membrane potential and a transmembrane pH gradient (acidic outside) that control multiple aspects of transport across the PM. Regulatory phosphorylation sites identified in AHAs are located either in the N- or C-terminal domain (Rudashevskaya et al., 2012). Phosphorylation of the penultimate residue, a Thr, within the auto-inhibitory C-terminal domain is critical for interaction with a 14-3-3 regulatory protein (Fuglsang et al., 1999); moreover, phosphorylation within this domain affects enzyme activity incrementally, suggesting a fine modulation (Speth et al., 2010). OGs induce phosphorylation of AHA1 and AHA2 at the conserved residue Ser899, a modification that has been reported to inhibit proton transport (Haruta et al., 2014) and to occur in response to flg22 (Nuhse et al., 2007) This modification may play a role in the extracellular alkalinization induced by elicitors, including OGs. Moreover, OGs induce phosphorylation of AHA1 at the non-conserved Ser544, located in the central part of the protein. Phosphorylation at this site has been recently described (Rudashevskaya et al., 2012) but, to our knowledge, never in response to elicitors.

Several subunits of V-ATPase (VHA-A, VHA-E1, and DET3) also change their phosphorylation status in response to OGs. V-ATPases are multi-subunit enzymes resembling the procariotic ATPases and therefore performing rotational catalysis. V-ATPases are localized on both the tonoplast and the Trans-Golgi Network (TGN) and their impairement affects morphology and function of the Golgi and TGN (Schumacher and Krebs, 2010). In both plants and animals (Robinson, 2014), V-ATPase perform uphill transport of protons and do not only energize secondary active transport, but also play a relevant role in vesicle trafficking and post-translational processing of secretory proteins, by regulating endosomal pH (Huotari and Helenius, 2011). Micro-compartments along the endocytic and secretory pathways have characteristic luminal pH values suited to their functions, and specific luminal pH and ionic concentration may be required not only for appropriate enzyme activities but also for membrane trafficking and sorting and cargo routing. Progressively decreasing pH in the endocytic pathway are thought to provide the cargo information about its location within the pathway (Huotari and Helenius, 2011; Martiniere et al., 2013).

OGs regulates phosphorylation of other plasma membrane transporters such as the acquaporin PIP2-1 and, besides PEN3, two additional ABC transporters, At3g55320 and At3g62700, belonging to subfamily 20 and 14, respectively. Ser280, Thr712 and Ser925 in PIP2-1, At3g55320, and At3g62700, respectively, showed phosphorylation only in the presence of OGs. In PIP2-1, phosphorylation at Ser280 and Ser283, has been associated to the regulation of hydraulic conductivity (Prado et al., 2013). Moreover, the latter phosphosites is necessary for correct targeting to the PM (Prak et al., 2008), showing that specific phosphorylation events control not only activity, but also targeting of transporters. Interestingly, PIP2-1 along with some members of family B and G ABC transporters not identified in this study, are markers of the so-called membrane nanodomains (NDs), ordered membrane SL- and sterol-rich microenvironments with functional assemblies of lipids and proteins that are defined through biophysical or microscopy, and not biochemical, techniques (Tapken and Murphy, 2015), and therefore are distinct from DIMs/DRMs, where PEN3 is instead found.

#### Response to Stress

A small cluster of OG-regulated phosphoproteins is represented by heat-shock proteins, including the cytosolic heat-shock cognate 70-1(HSC70-1), the chloroplast heat shock protein 70-1 (HSP70-1), BiP1 and the cytosolic HSP90. HsP90 is a molecular chaperone responsible for the maturation and stability of a large number of signaling proteins, in particular protein kinases. We found decreased levels of the HSP90 Ser219 phosphosite, in agreement with previous reports (Reiland et al., 2009; Jones et al., 2009). The role of this modification is not known. In yeast, key phosphorylation sites on HSP90 affect the interaction with a selected subset of co-chaperones, and consequently, client proteins (Mollapour et al., 2011); a similar role is possible in plants. Arabidopsis HSP90 interacts with MPK4 and regulates its kinase activity during response to flg22 (Cui et al., 2010). HSP90 isoforms, together with the co-chaperones RAR1 and SGT1, appear to be critical also for the stability of nucleotidebinding leucine-rich repeat receptor (NLR) resistance proteins such as RPM1, MLA, RPS2, RPS4, I2, and N, and the formation of immune receptor complexes (Huang et al., 2014). In addition, they negatively regulate the accumulation of NLRs, likely to avoid autoimmunity (Huang et al., 2014). The decreased level of phosphorylation in response to OGs is likely related to the regulation of HSP90 complex assembly and substrate specificity.

#### Proteins Involved in Redox Homeostasis

Many proteins involved in redox homeostasis and ROS signaling show OG-induced increase in phosphorylation, including an isocitrate dehydrogenase (CICDH; **Table 2**) involved in response to pathogens (Mhamdi et al., 2010), and GAPC1, a glycolytic enzyme that interacts with phospholipase D and transduces hydrogen peroxide signals during stress (Guo et al., 2012). Both proteins appear to be regulated mainly through PTM. In fact, only few proteins involved in redox regulation here identified show both phosphorylation changes and transcript up-regulation. Among these, RBOHD (Kadota et al., 2014), glutathione S-transferases (GSTF9, GST-PM24, GSTF8), the non-photosynthetic NADP-malic enzyme 2 (NADP-ME2) and flavodoxin-like quinone reductase 1 (FQR1) exhibit increased phosphorylation, whereas a monodehydroascorbate reductase (MDAR2) and the protein plant cadmium resistance 8 (CdRes8) show decreased phosphorylation. NADP-ME2 plays a role in basal defense against the hemibiotrophic fungal pathogen Colletotrichum higginsianum, and is involved in the generation of ROS (Voll et al., 2012), whereas FQR1 is a primary auxinresponse gene (Laskowski et al., 2002) that belongs to quinone reductases, classified as phase II detoxification enzymes, that protect organisms from oxidative stress; interestingly, FQR1 acts as a susceptibility factor to B. cinerea (Heyno et al., 2013).

PTM-mediated regulation of oxidative-stress related proteins is in agreement with the notion that the redox state of the cell is dramatically altered in response to OGs.

#### Other Proteins

JAL34 and JAL27 (**Table 2**) are jacalin lectin-like proteins similar to myrosinase-binding protein 1 (Yamada et al., 2011) and to the functionally related PYK10. The latter is a β-glucosidase localized in the endoplasmic reticulum (ER) body, a large compartment specific to the Brassicales. PYK10 forms complexes with JALs and other proteins when tissue is damaged, i.e., upon herbivore or pathogen attack. Engagement in a complex may shield active PYK10 from inhibitors, proteases and other proteins that may reduce its defensive effect (Yamada et al., 2011). JAL34 and JAL27 show also OG-induced increased abundance at a later time point (Casasoli et al., 2008). JAL34 and JAL27 as well as PCaP1, ACT7, PATL2, and MDAR2 have been described in a previous proteomic work as proteins induced by brassinosteroids (Tang et al., 2008); whether these proteins are involved in the BR-mediated inhibition of PTI is still unknown.

#### CONCLUSIONS

In conclusion, we have provided a large-scale study of early phosphoproteome changes in Arabidopsis following OG perception. In a recent phosphoproteomic study of

#### REFERENCES


Arabidopsis seedling total extracts following OG treatment, 51 phosphorylated peptides, representing 49 unique proteins, were identified (Kohorn et al., 2016). A comparison with our study shows only 9 shared phosphosites, corresponding to 9 proteins (**Table 1**), two of which also found in flg22 regulated phosphoproteome. Taking into consideration the other 16 shared flg22-regulated phosphosites, our work performed also on membrane proteins uncovers the immune-related regulation of 73 phosphosites. Protein interaction network analyses point to the main biological processes in which protein phosphorylation events are crucial in OG signaling, and suggest the interplay of several processes, e.g., intracellular trafficking and vesicle dynamics, cytoskeleton rearrangement, signal transduction and phospholipid signaling.

### AUTHOR CONTRIBUTIONS

BM, FS, and DP performed all the experiments and analyzed the data; BM and DP performed the LC-MS/MS analyses; BM and FS performed the 2D-DIGE analyses; BM and GD conceived the project and wrote and revised the article.

#### FUNDING

This work was supported by the European Research Council (ERC Advanced Grant233083), the Italian Ministry of Agriculture Food and Forestry Policies (MIPAAF, Project ALISAL, DM 11008/7303/10) and University of Rome Sapienza (ATENEO, 2014).

### ACKNOWLEDGMENTS

We are grateful to Prof. Felice Cervone (Sapienza University of Rome) for careful review and helpful comments. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaíno et al., 2016) partner repository with the dataset identifier PXD004599.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 01107


proteins by mass spectrometry and peptide chip analysis. J. Proteome Res. 7, 2458–2470. doi: 10.1021/pr8000173


destabilizing cortical microtubules in Arabidopsis. Plant Cell 23, 4411–4427. doi: 10.1105/tpc.111.092684


phosphorylation and dephosphorylation events in Arabidopsis thaliana by quantitative phosphoproteomic analysis. J. Proteome Res. 13, 2137–2151. doi: 10.1021/pr401268v


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Mattei, Spinelli, Pontiggia and De Lorenzo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Quantitation of Vacuolar Sugar Transporter Abundance Changes Using QconCAT Synthtetic Peptides

Heidi Pertl-Obermeyer<sup>1</sup>† , Oliver Trentmann<sup>2</sup> , Kerstin Duscha<sup>2</sup> , H. Ekkehard Neuhaus<sup>2</sup> and Waltraud X. Schulze<sup>1</sup> \*

<sup>1</sup> Department of Plant Systems Biology, University of Hohenheim, Stuttgart, Germany, <sup>2</sup> Plant Physiology, University of Kaiserslautern, Kaiserslautern, Germany

#### Edited by:

Qingsong Lin, National University of Singapore, Singapore

#### Reviewed by:

Giridara Kumar Surabhi, Regional Plant Resource Centre, India R. Glen Uhrig, ETH Zurich, Switzerland

> \*Correspondence: Waltraud X. Schulze wschulze@uni-hohenheim.de

#### †Present address:

Heidi Pertl-Obermeyer, Department Biology I, Plant Biochemistry and Physiology, LMU Munich, Germany

#### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 09 December 2015 Accepted: 17 March 2016 Published: 12 April 2016

#### Citation:

Pertl-Obermeyer H, Trentmann O, Duscha K, Neuhaus HE and Schulze WX (2016) Quantitation of Vacuolar Sugar Transporter Abundance Changes Using QconCAT Synthtetic Peptides. Front. Plant Sci. 7:411. doi: 10.3389/fpls.2016.00411 Measurements of protein abundance changes are important for biological conclusions on protein-related processes such as activity or complex formation. Proteomic analyses in general are almost routine tasks in many laboratories, but a precise and quantitative description of (absolute) protein abundance changes require careful experimental design and precise data quality. Today, a vast choice of metabolic labeling and label-free quantitation protocols are available, but the trade-off between quantitative precision and proteome coverage of quantified proteins including missing value problems remain. Here, we provide an example of a targeted proteomic approach using artificial standard proteins consisting of concatenated peptides of interest (QconCAT) to specifically quantify abiotic stress-induced abundance changes in low abundant vacuolar transporters. An advantage of this approach is the reliable quantitation of alimited set of low-abundant target proteins throughout different conditions. We show that vacuolar ATPase AVP1 and sugar transporters of the ERDL (early responsive to dehydration-like) family and TMT2 (tonoplast monosaccharide transporter 2) showed increased abundance upon salt stress.

Keywords: vacuole, sugar transport in plants, salt stress, drought stress, QconCATs

## INTRODUCTION

Measurements of protein abundance changes allow conclusion about post-transcriptional regulatory processes, for example through correlation with protein activity or complex stoichiometry. Quantitative proteomic analyses are routine tasks in many experiments, and a large selection of methods using metabolic labeling or label-free approaches is available reviewed in Schulze and Usadel (2010) and Arsova et al. (2012a). However, although, for yeast full proteome coverage has been claimed (Walther et al., 2010) and up to 10000 proteins were identified from human tissue in single LC–MS/MS runs (Mann et al., 2013), even with modern, fast, and highly accurate massspectrometers the full proteome measurements are not yet routinely feasible for most labs. Particularly if tissues with highly skewed protein abundance distributions are analyzed, such as plant leaf tissues (Arsova et al., 2012b), uniform protein coverage becomes more challenging. Furthermore, membrane proteomics is still rather pretentious and thus large scale proteomic data sets still are often not complete and particularly lacking information for low abundant membrane proteins such as ion channels and certain transporters.

A common workflow to address this problem is to firstly perform profiling experiments, then, secondly to define a list of relevant target proteins/peptides which will then be studied using dedicated targeted protein analysis methods (Schulze, 2010). Such workflows were successfully used to study protein phosphorylation responses under osmotic stress (Stecker et al., 2014) or to study specific enzyme activities and their regulatory phosphorylation sites (Glinski and Weckwerth, 2005). In yeast, the central carbon and amino acid metabolism was in detail studied by a targeted approach (Costenoble et al., 2011). Most of these targeted approaches require the use of triple quadrupole or quadrupole-Orbitrap mass spectrometers and synthetic versions of the target peptide sequences for tuning and as internal quantitation standards (Gallien et al., 2012; Kelstrup et al., 2012). Setup of such targeted experiments is quite challenging as often redundancy occurs in precursor and product ions resulting in lower quantitation specificity of the selected targets (Sherman et al., 2009). As an alternative strategy, and to avoid expensive synthesis of synthetic peptides, the concatenated peptides (QconCAT) were developed allowing the expression of concatenated peptides as an artificial protein in bacteria (Pratt et al., 2006). These artificial proteins can easily be stableisotope labeled in bacteria and then be digested to obtain tryptic standard peptides for mass spectrometric analysis and quantitation. Heavy labeled peptides are added to the sample of interest in known quantities and ion intensities are measured for the light peptides, originating from the biological sample, and for the heavy peptides, representing the standards. Finally, their ratio of the respective heavy and light signal intensities can be used to calculate the absolute quantity. Such labeled peptides were used in various studies quantifying developmental changes in muscle cells (Rivers et al., 2007) or in studying the glycolytic pathway in yeast (Carroll et al., 2011). In plants, spiked in digested peptides originating from a QconCAT synthetic protein were used to study the stoichiometry in plastidial Clp protease complexes (Olinares et al., 2011), but have not yet been widely used.

The vacuole is an important plant specific organelle with functions in storage of solutes as nutrient reservoirs, but also with important roles in adaptation to stresses, such as cold stress (Wormit et al., 2006; Schulze et al., 2012), salt stress or drought (Rizhsky et al., 2004; Hedrich et al., 2015). Various transporters for solutes into and out of the vacuole were identified and individually characterized in their physiological context (Martinoia et al., 2012). Sugars play important roles as nutrients as well as signal molecules or compatible solutes during stress. Particularly, in the recent years vacuolar sugar transporters were characterized in detail. They are mainly membrane proteins belonging to the major facilitator superfamily (MFS) whose subfamilies of sucrose transporters (SUTs) and monosaccharide transporters (MSTs) are well-studied also under drought stress (Medici et al., 2014). In that context, the family of tonoplast monosaccharide transporters (TMTs) was characterized as glucose transporters involved in the re-distribution of cellular glucose to the vacuole under cold stress conditions (Wormit et al., 2006). More recently, TMT1 and TMT2 were also found to transport sucrose (Schulz et al., 2011). Another family of vacuolar sugar transporters (VGT/ERD) contains importers and exporters of vacuolar sugars (Büttner, 2007). Besides glucose, fructose and sucrose, also sugar alcohols like sorbitol, mannitol, and myo-inositol are transported into vacuoles, suggesting a fine-tuning of the transport activities across the tonoplast dependent on external conditions and nutrient supply.

There is evidence that in Arabidopsis various abiotic stresses, especially cold stress, lead to accumulation of sugars, particularly glucose and fructose in the vacuole (Wormit et al., 2006; Schulze et al., 2012). An increased accumulation of sugars upon drought and heat stress has also been observed (Rizhsky et al., 2004) suggesting a role of vacuolar sugar transporters also under these conditions. Expression of the putative sugar transporter ERD6 (early responsive to dehydration) is induced not only by dehydration but also by cold treatment (Kiyosue et al., 1998), and expression of an ERD6-like transporter (ESL1) is enhanced by drought, salt, and ABA treatment (Yamada et al., 2010). Osmotic stress and salt stress also affects vacuolar transporters, particularly the sucrose transporter SUC4 (Gong et al., 2015) and the v-ATPases (Kirsch et al., 1996).

All of these energy-driven fluxes of metabolites across the tonoplast are channeled by a range of different transporters. Therefore, in addition to sugar transporter, in a typical tonoplast also energizing transporters are found, such as the V-PPase, the V-ATPase, ABC transporters and Ca2<sup>+</sup> pumps. During salt stress the activity of the vacuolar H<sup>+</sup> pumps is increased and accompanied also by induced gene expression (Hasegawa et al., 2000; Maeshima, 2001). In that context, it has been demonstrated that overexpression of the H+-PPase AVP1 increases salinity tolerance in Arabidopsis (Gaxiola et al., 2001). In addition there are several transporters for water and organic solutes, anion channels and cation transporters like potassium channels and zinc or copper transporter (Martinoia et al., 2007). Recently it became obvious that the vacuole as such, or the regulation of tonoplast transporter, plays an important role during adaption to abiotic environmental stress.

Here, we developed and applied a targeted quantitative proteomics workflow using QconCAT synthetic standard peptides to study particularly the abundance changes of sugar transporters in the tonoplast-enriched fractions from Arabidopsis leaves grown under abiotic stress conditions such as salinity or drought.

#### RESULTS

#### Determination of the Linear Range of Quantitation

Unlabeled (L, light) and labeled (H, heavy) QconCATs were mixed in different ratios (L:H ratios of 10:0, 9:1, 7:3, 5:5, 3:7, 1:9, and 0:10) to result in a total of 10 µg of protein before tryptic digest. After tryptic digestion and desalting, 5 µg of each mixture was analyzed by LC–MS/MS. The measured H/L ratio of each mixture of labeled and unlabeled peptides showed a linear relationship with the expected H/L ratio provided by the peptide mixture (**Figure 1**). Only in very low H/L mixing

ratios, the measured ratios were lower than expected suggesting that under these conditions, the signal to noise ratio of the full scan spectra did not fully resolve the very low presence of unlabeled peptide. Different peptides showed slightly different curves reflecting variations in ionization properties.

### Detection of Spiked-in Peptides in Complex Sample Background

To determine the limit of detection within a complex protein background, labeled QconCAT protein was spiked into different amounts of microsomal protein preparations resulting in w/w protein:QconCAT ratios of 5:1, 10:1, and 20:1. After joint tryptic digestion and desalting of these spiked protein mixtures, 5 µg of each mixture was analyzed. As expected, under conditions of the QconCAT protein being mixed with lower amounts of the microsomal protein, higher spectral counts were achieved for each of the QconCAT peptides resulting from the tryptic digestion (**Figure 2**). Based on these results, in further experiments a mixing ratio or protein:QconCAT of 5:1 was used to ensure efficient coverage of the target peptides in the mass spectrometric analysis.

### Reproducibility: Variations in Intensities, Retention Time, and Spectral Counts

Next, we analyzed the reproducibility of QconCAT detection. Thus, a total of 25 µg of QconCAT protein was in-solution digested, desalted and injected five times in 5 µg injections for mass spectrometric analysis. The five injections were done either from different wells on the autosampler plate or repeatedly sampled from the same well. To determine the reproducibility between runs, detected peptides were analyzed according to variation in peak intensities, retention times and the number of recorded spectra. Across all peptides and the five injections, retention time variations from run to run was extremely low. In contrast the ion intensity variation averaged around 0.2 rsd, and the spectral counts showed largest variation (**Figure 3**). These results confirm that our chromatography and subsequent retention time alignment of raw data is very reproducible and that ion intensity (or ion intensity ratio) quantitation is more robust and less prone to variation than spectral counting.

#### Large Alterations in Abundance of Vacuolar Sugar Transporters under Salt and Drought Stress

The QconCAT protein was spiked into vacuolar preparations of plants grown under salt stress or drought stress and respective control conditions. We were particularly interested in following the changes in protein abundance of several vacuolar sugar transporters which play an important role in adjustments of cellular solute concentrations under these stress conditions. The abundance changes in sugar transporter abundance were based on changes of proteotypic signature peptide ratios in stressed plants compared to their control conditions (**Supplementary Table S1**). The peptides were chosen with high probability for good ionization properties (Brownridge et al., 2011) or based on experimentally identified peptides in previous work (Schulze et al., 2012) or from public databases (Joshi et al., 2011). We then used the ratio of light (L, unlabeled) plant-originating peptide version to the heavy (H, labeled) peptide originating from the spiked-in QconCAT for quantitation (**Figure 4**). Using the L/H ratio, with constant amount of the heavy standard peptide (H), the ratio directly is proportional to the abundance ratio of plantderived peptide (L) compared to its standard. Comparing two different conditions (stress treatment and control), the change in L/H ratio depicts the change in protein abundance induced by the stress condition (**Figure 4**) and the ion intensities of the spikedin standard can be used for normalization between different samples. Thus, a change of the L/H ratio to higher (more positive) values is proportional to an increase in abundance of the peptide (and protein) while a ratio change to lower (more negative) values

is proportional to a decrease of peptide and protein abundance (**Figure 4**).

The amount of the spiked in standard protein was 1 µg, but after tryptic digestion within the mixture with the vacuolar protein, each standard peptide was produced to different absolute amounts according to the molecular weight of each peptide, ranging from 10 to 34 ng in the final digested 5:1 mixture of protein:QconCAT (**Supplementary Table S1**).

However, each peptide was produced to the same molar amount: the 1 µg spiked-in QconCAT corresponded to 13 fmol of protein. With knowledge of the molar amount of standard added to the sample, the molar amount of each peptide can be calculated as follows: L/H <sup>∗</sup> (mQconCAT/MWQconCAT). Such calculations revealed AVP1 as the most abundant protein with 45 ± 31 mmol/gprotein and VMA21a and VMA22 as the least abundant proteins with 0.3 ± 0.12 mmol/gprotein. Among the sugar transporters, ERDL4 and TMT2 were most abundant

averaging 24 ± 3.8 mmol/gprotein and 5.8 ± 2.5 mmol/gprotein, respectively.

The L/H ratios in control conditions, salt stress (400 mM NaCl) as well as control and drought conditions could be derived for several vacuolar transporters (**Figure 5**). Thereby, in some cases, such as for high abundant proteins AVP1 (AT1G15690), ERDL4 (AT1G19450) and COPT5 (AT5G20650), the endogenous peptides (L) was in all cases more abundant than the spiked in standards (H) leading to large L/H ratios across conditions. In other cases, such as for low abundant VMA21a (AT2G31710), VMA22 (AT1G20770), ERDL7 (AT2G48020), and ERDL8 (AT3G05150), the endogenous peptides (L) were less abundant than the spiked-in standard (H) leading to low L/H ratios across conditions.

Under salt stress, we observed an increase of protein abundance for most of the monitored proteins. Particularly, for AVP1 (AT1G15690) abundance significantly increased almost threefold when plants were exposed to salt stress of 400 mM NaCl (**Table 1**). Among the sugar transporters, significant and large increases in abundance were observed also for ERDL7 (AT2G48020) and TMT2 (AT4G35300). ERDL4 (AT1G19450) and ENT1 (AT1G70330) were the only sugar transporter for which no significant change in abundance was measured under salt stress. Under drought stress, for most proteins a mild decrease in protein abundances was observed (**Table 1**). These mild changes were significant only for AVP1 (AT1G15690), ERDL7 (AT2G48020), and ENT1 (AT1G70330) which decreased from vacuolar membranes. A significant drought-induced increased abundance was observed for ERDL4 (AT1G19450) and COPT5 (AT5G20650).

The measured abundance changes using the QconCATderived standard peptides were confirmed by an independent label-free quantitation using the same raw files. The very low abundant proteins VMA21a (AT2G31710), VMA22 (AT1 G20770), and ERDL8 (AT3G05150) could not be quantified in the label-free analysis. Overall, in the label-free analysis the abundance changes resulted in smaller amplitudes for some proteins, but with similar trends as in the ratio quantitation (**Table 1**). Only for COPT5 (AT5G20650) under drought stress, an increased abundance was measured based on the QconCAT quantitation while the label-free quantitation resulted in a decreased abundance.

#### DISCUSSION

In this work we explored the use of QconCAT standard proteins to enhance the quantitation of specific target proteins,

such as vacuolar transporters. Thereby, we firstly provide some technical properties on quantitation range, detection limit and reproducibility. Technically, we co-digested vacuolar protein together with the spiked-in standard protein. We specifically decided against alternative to separately digest the standard protein and then add defined amounts of the resulting standard peptides to digested vacuolar protein preparations in order to more precisely spot sample-to-sample variations. 24 peptides (out of 49 peptides within QconCAT) were reproducibly detected in replicates of control and stress treatment samples allowing solid quantitation in both conditions.

In the past years, several methods for absolute protein quantitation strategies were developed, such as emPAI (Ishihama et al., 2005), APEX (Braisted et al., 2008), the TOP3-method (Silva et al., 2006), or iBAQ (Schwannhäusser et al., 2011) which all have in common that they allow protein abundance ranking without the use of spiked-in internal standards. These methods are based on several peptides identified for each protein or on the most optimal three ions as in the TOP3-method. The emPAI values were successfully used to calculate cellular concentrations of metabolic enzymes (Arrivault et al., 2014), and a collection of spiked-in synthetic standard peptides was used to quantify Calvin cycle enzymes in Chlamydomonas (Wienkoop et al., 2010).

The absolute quantitation of protein amounts using spikedin standard peptides is most likely to be less accurate due to lower statistical power of only few peptides, and these peptides may not necessarily be among the most optimal peptides which otherwise are for iBAQ or TOP3 quantitation. It has been shown that quantitation using standard peptides particularly depends on good design of the peptides (Brownridge et al., 2011), and experimental evidence for the chosen peptides from previous analyses is beneficial. Thus, not all peptides are equally suitable to quantify the absolute protein amounts due to low ionization properties or modification sites. The peptides used here were designed to allow clear identification of the target proteins (proteotypic peptides) and in some cases, such as for AVP1 or ENT1 one of the three peptides designed could not be avoided to be cysteine containing due to structural constraints (i.e., avoiding transmembrane spans). In general, the absolute protein amounts calculated here corresponded well (r = 0.89, r <sup>2</sup> = 0.81) with the iBAQ values calculated from a label-free analysis.

The observation that TMT2 protein abundance decrease during drought is surprising, given the high monosaccharide import specificity of latter carrier (Wormit et al., 2006) and given drought conditions provoke accumulation of sugars in leaves of higher plants (Krasensky and Jonak, 2012). However, it might be that during drought monosaccharides have to accumulate in other compartments of the cell. This assumption is based on the observation that sugars act as efficient ROS scavengers (Lineberger, 1980) and it is widely known that under abiotic stress ROS generation takes especially place in chloroplasts, peroxisomes and mitochondria (Cruz de Carvalho, 2008). Furthermore, general protein degradation processes during drought stress could also reduce the abundance of these transporters in the tonoplast.

From none of the ERDL proteins mentioned (**Table 1** and see Results) the transport characteristics have been determined (Hedrich et al., 2015). Thus, it appears unjustified to interpret the changes in protein abundancies of ERDL4 and ERDL7 in response to salt or drought on a mechanistic level. This interpretation becomes even more complex given the known post-translational modification of TMT-type proteins with consequences on transport activity (Wingenter et al., 2011). In other words, the observed changes of ERDL protein abundancies have in the near future to be correlated with exact transport properties (e.g., substrate specificities and modes of transport) and with a detailed search for putative post-translational modifications.

In conclusion, the QconCAT standard peptides are very useful in experiments where a defined set of target proteins is to be analyzed, as for example in this study the focus was specifically on vacuolar (sugar) transporters. We demonstrated that the reproducibility of QconCAT detection is very high and that ion


TABLE 1 | Summary of changes in protein amounts under salt and drought stress comparing QconCAT quantitation with label-free quantitation.

Red and blue indicate down and up regulated protein expression, respectively. Color intensity is proportional to the intensity of change. Significant changes (p-value < 0.05, t-test) are shaded in green.

intensity ratio quantitation is more sensitive and more and robust than spectral counting.

The use of these spiked-in standards clearly does exceed the range of quantitation to include more low-abundant proteins compared to a simple label-free experiment. We showed that salinity stress results in large changes in vacuolar transporter abundances, particularly for vacuolar ATPases AVP1, members of the ERDL-transporter family as well as for the monosaccharide transporter TMT2. Since salinity is known to increase the sugar content in vacuoles (Rizhsky et al., 2004), we suggest that this increase in vacuolar sugar content is achieved by an increase in transporter presence in the tonoplast.

#### MATERIALS AND METHODS

#### Bacterial Strains and Used Vector

Competent Escherichia coli BL21 (DE3) with the genotype B F−dcmompThsdS(r<sup>B</sup> <sup>−</sup>m<sup>B</sup> <sup>−</sup>) galλ(DE3) were purchased from Agilent Technologies. The selected QconCAT peptides (see **Supplementary Table S1**) were concatenated and flanked by a leader N-terminal sequence (MAGK) and a C-terminal sequence (LAAALEHHHHHH) for purification (Polyquant, Bad Abbach, Germany). The full QconCAT1 protein sequence is:

MAGKCLTAVFMLEDEKATILFDFWKHKPDASFVAEAKSLS VFGGLVSPKFSPATFSPVSSEDSNLRQSSMMGSSQVIRHGSLA NQSMILKQTTSMDKGVVALDLGRLYGTHENQSYLARPVPEQ NSSLGLREGWFDLASARDFSTDTDSEIGSPLSPQLRGTFSQLD NLSMANITKHKPDAAFLAEAKLFNLLETSHKDNDDYATDDG AGDDDDSDNDLRLLLMISSIGMTISLVIVAVAFYLKSSGEISPER EPLIKSGNAPIYYPNREATFGELFRITAAQVLEHPWIKSFTQSS PSFTQSSPSFTPTTKTGIPTALWDLMEPYAKVSAVSLSEEEIKM NGGDVTVAGSDDLKMGLTDDFETSLQVLRGATSSDDHALK VTDFGLSAFIEEGKNITDDIPGWGRLTSGVTNTDPALKLAMG PLCDLIGPRADISEEAAEIQDYIETLERSFLALASGDRGGSTMS VLSRQIQMTTTDKMALDPEQQQPISSVSRSSSGVSAPLIPKNT KPPSFSDSTIPVDSDGRQFNTIPGLMEGTAKPDYATCVKGEA SGSVFFIDGSNNQYLRPRLTSDLGASSSGGANNGKIYLHQEGF PGSRRPFIHTGSWYREIGTGNIYACKVSILLGLLKEAGEIQEYL ASLAHLPKAAVIGDTIGDPLKDPLVNLFGSLHEKHSDFEIALQ KLAAALEHHHHHH.

The plasmids encoding the proteins for QconCAT1were cloned into the NdeI/BamHI restriction sites of pET21a (Novagen, Nottingham, UK). A list of tryptic peptides and corresponding proteins they were derived from is shown in **Supplementary Table S1**.

#### Expression, Stable-Isotope Labeling, and Purification of QconCAT1

Subsequent transformation, protein expression, and purification were performed as described in Pratt et al. (2006). In brief, the plasmid carrying the artificial gene for QconCAT1 was introduced into chemically competent E. coli using a standard transformation protocol, and frozen stocks were stored at −80◦C. The QconCAT protein was expressed in E. coli cultures with a full complement of unlabeled amino acids (M9 minimal medium Sambrook et al., 1989) or in the presence of [13C6]arginine (10 mg 100 ml−<sup>1</sup> ) and [13C6]lysine (10 mg 100 ml−<sup>1</sup> ). Expression was induced with IPTG (isopropyl ß-D-thiogalactopyranoside), and after 5 h the cells were harvested by centrifugation at 4,000 rpm for 15 min at 4◦C. Inclusion bodies containing the QconCAT were recovered by breaking the cells using BugBuster Protein Extraction Reagent (Novagen, Nottingham, UK) and after several washing steps centrifuged at 16,000 × g for 15 min at 4◦C. For purification of the labeled and unlabeled QconCATs, the pellets of inclusion bodies were resuspended in IMAC binding buffer (20 mM phosphate buffer, pH 7.4, 500 mM NaCl, 20 mM imidazole, 6 M guanidine hydrochloride) and loaded onto gravity flow Ni2+-NTA sepharose columns (1 ml, IBA GmbH, Goettingen, Germany). After several washing steps the bound QconCAT was eluted with elution buffer (20 mM phosphate buffer, pH 7.4, 500 mM NaCl, 500 mM imidazole, 6 M guanidine hydrochloride) in 5 ml × 1 ml fractions. Purified QconCATwas desalted by dialysis (Slide-A-LyzerTM Dialysis Cassettes, 3,500 MWCO, 3–12 ml, Pierce) against 100 volumes of 50 mM ammonium bicarbonate, 1 mM DTT for 3 h × 2 h at 4◦C. Finally, the protein content was determined using BSA as a standard (Lowry DC, Bio-Rad, Vienna, Austria). To ensure stability, purified QconCAT was aliquoted into 10, 25, 50, and 100 µg samples and dried down in a speed vac to completely dryness and stored at −80◦C. Before use, QconCAT protein was resuspended in 6 M urea, 2 M thiourea, pH 8 in Tris-HCl to result in a final concentration of 1 µg µL −1 .

### In-solution Digest for Mass Spectrometry

For in-solution digest the purified QconCAT protein (5 or 10 µg) was denatured using UTU buffer (6 M urea, 2 M thiourea, pH 8.0). After reduction in 6.5 M DTT and alkylation of cysteine residues by 27 mM iodoacetamide, proteins were digested for 3 h by LysC (Wako, Japan) at room temperature. The solution was then diluted fourfold with 10 mM Tris-HCl, pH 8.0 followed by overnight digestion with trypsin (sequencing grade, Promega) at 37◦C at 350 rpm. Finally, digested peptides were desalted over C18 STAGE- tips and dried down in a speed vacuum concentrator and stored at −80◦C. For mass spectrometric analysis samples were resuspended in resuspension buffer (0.2% v/v TFA, 5% v/v acetonitrile).

Tryptic peptide mixtures were analyzed by LC–MS/MS using nanoflow HPLC (Proxeon Biosystems, Denmark) and a hybrid quadrupole-orbitrap mass spectrometer (Q Exactive Plus, Thermo Scientific) as a mass analyser. Peptides were eluted from a 75 µm × 15 cm C18 analytical column (PepMap <sup>R</sup> RSLC C18, Thermo Scientific) on a gradient using 0.5% acetic acid as aqueous phase and 0.5% acetic acid in 80% acetonitrile as organic phase. The flow rate was set to 250 nL per minute. Peptides were eluted on a linear gradient running from 4 to 64% acetonitrile in 135 min. Spectra were using information-dependent acquisition of fragmentation spectra of multiple charged peptides within the m/z range of 300–1600. Up to 12 data-dependent MS/MS spectra were acquired for each full-scan spectrum acquired at 70,000 fullwidth half-maximum resolution. Fragment spectra were acquired at a resolution of 35,000. Masses of peptides of QconCAT1 were

used as inclusion list to induce preferred fragmentation of the target peptides of interest.

Protein identification and ion intensity quantitation was carried out by MaxQuant version 1.4.1.2 (Cox and Mann, 2008). Spectra were matched against the Arabidopsis proteome (TAIR10, 35386 entries) using Andromeda (Cox et al., 2011) and the following search parameters: multiplicity was set to 2, Arg6 and Lys6 were selected as heavy labels; carbamidomethylation was selected as a fixed modification for cysteines, methionine oxidation was selected as variable modification, two missed cleavages were allowed. Precursor mass tolerance was set to 20 ppm and MS/MS tolerance to 0.5 Da. Peptides were accepted with a length of more than seven amino acids under a protein match false discoveryrate threshold of 0.01, peptide spectral match false discovery rate threshold of 0.01. Retention time alignment was done in a window of 2 min. Quantitation of spiked-in heavy QconCATs and internal unlabeled peptides from the plant extract was performed using the SILAC heavy/light quantitation within MaxQuant.

#### Preparation of Microsomal Protein

Microsomal fractions (MFs) of seedling cultures were prepared by differential centrifugation (Pertl et al., 2001). Frozen Arabidopsis seedlings (approximately 20 g of fresh weight) from Col-0 were smashed into small pieces and resuspended in icecold homogenisation buffer (330 mM sucrose, 100 mM KCl, 1 mM EDTA, 50 mM Tris adjusted with MES to pH 7.5, 5 mM DTT), protease inhibitor cocktail (Sigma-Aldrich), phosphatase inhibitor cocktail 2 (Sigma-Aldrich) and phosphatase inhibitor cocktail 3 (Sigma-Aldrich) were added from stock solutions (50 µL per 10 mL of the homogenization buffer just before use). Tissue was homogenized with a Teflon Potter-Elvehjem on ice. The homogenate was filtered through a 21 µm nylon mesh, and centrifuged at 7,500 × g for 15 min at 4 ◦C. Finally, the supernatant was centrifuged at 48,000 × g

#### REFERENCES


for 80 min at 4◦C. The resulting pellet was the MF and stored at −80◦C.

#### Preparation of Tonoplast Protein

Arabidopsis plants were cultivated for 28 days under short day conditions (10 h light, 14 h dark) at 22◦C. To achieve salt stress the plants were afterward watered using tap water containing 400 mM NaCl at day 29 and at day 31. Vacuoles/tonoplasts were isolated from these plants at day 32. To generate drought stress, watering of the corresponding plants was stopped at day 28 and vacuoles/tonoplast isolation was started when dehydration symptoms were monitored (wilting of plant leaves, about 7–8 days after termination of watering). Vacuoles and tonoplast membranes were isolated following exactly the protocol described (Schulze et al., 2012). Isolated protein (10 µg) was spiked with labeled QconCAT in a protein:QconCAT ratio of 5:1 and was jointly digested by trypsin. After desalting over C18, 5 µg of the resulting peptides were analyzed.

#### AUTHOR CONTRIBUTIONS

HP-O: performed lab experiments, analyzed the data, wrote the paper. KD: performed the salt stress experiments. OT: performed drought stress experiments, sample preparation, contributed to paper writing. HN: contributed significantly to paper writing. WS: conceived the idea and wrote main parts of the paper.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016.00411

TABLE S1 | The peptides on QconCAT1 which was used as a spiked-in standard protein in this study.


of the AVP1 H+-pump. Proc. Natl. Acad. Sci. U.S.A. 98, 11444–11449. doi: 10.1073/pnas.191389398


muscle development using QconCAT. Mol. Cell. Proteomics 6, 1416–1427. doi: 10.1074/mcp.M600456-MCP200


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Pertl-Obermeyer, Trentmann, Duscha, Neuhaus and Schulze. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Hydrogen Sulfide: A Signal Molecule in Plant Cross-Adaptation

Zhong-Guang Li1,2,3 \*, Xiong Min1,2,3 and Zhi-Hao Zhou1,2,3

<sup>1</sup> School of Life Sciences, Yunnan Normal University, Kunming, China, <sup>2</sup> Engineering Research Center of Sustainable Development and Utilization of Biomass Energy, Ministry of Education, Kunming, China, <sup>3</sup> Key Laboratory of Biomass Energy and Environmental Biotechnology, Yunnan Normal University, Kunming, China

For a long time, hydrogen sulfide (H2S) has been considered as merely a toxic by product of cell metabolism, but nowadays is emerging as a novel gaseous signal molecule, which participates in seed germination, plant growth and development, as well as the acquisition of stress tolerance including cross-adaptation in plants. Crossadaptation, widely existing in nature, is the phenomenon in which plants expose to a moderate stress can induce the resistance to other stresses. The mechanism of cross-adaptation is involved in a complex signal network consisting of many second messengers such as Ca2+, abscisic acid, hydrogen peroxide and nitric oxide, as well as their crosstalk. The cross-adaptation signaling is commonly triggered by moderate environmental stress or exogenous application of signal molecules or their donors, which in turn induces cross-adaptation by enhancing antioxidant system activity, accumulating osmolytes, synthesizing heat shock proteins, as well as maintaining ion and nutrient balance. In this review, based on the current knowledge on H2S and cross-adaptation in plant biology, H2S homeostasis in plant cells under normal growth conditions; H2S signaling triggered by abiotic stress; and H2S-induced cross-adaptation to heavy metal, salt, drought, cold, heat, and flooding stress were summarized, and concluded that H2S might be a candidate signal molecule in plant cross-adaptation. In addition, future research direction also has been proposed.

Keywords: cross-adaptation, hydrogen sulfide, signal crosstalk, stress tolerance

## INTRODUCTION

Edited by: Hanjo A. Hellmann,

> Reviewed by: Karl-Josef Dietz,

Sutton Mooney,

\*Correspondence: Zhong-Guang Li zhongguang\_li@163.com

Specialty section: This article was submitted to

Plant Physiology, a section of the journal Frontiers in Plant Science Received: 07 May 2016 Accepted: 13 October 2016 Published: 26 October 2016

Citation:

Washington State University, USA

Washington State University, USA

Bielefeld University, Germany

Cross-adaptation, widely existing in nature, is the phenomenon in which plants expose to a moderate stress can induce the resistance to other stresses (Li and Gong, 2011; Foyer et al., 2016; Hossain et al., 2016). For example, cold pretreatment can improve the heat tolerance of winter rye, salt shock can rapidly induce the cold tolerance in spinach and potato, ultraviolet radiation (UV-B) can enhance the heat tolerance in cucumber and the cold tolerance in Rhododendron, and mechanical stimulation can improve the heat tolerance and the chilling tolerance in tobacco cells (Knight, 2000; Li and Gong, 2011, 2013). Interestingly, Foyer et al. (2016) found that crossadaptation also can be induced between abiotic and biotic stresses. Infection by mycorrhizal fungi can improve the resistance of tomato, sunflower, pea, and rice to drought, chilling, salinity, metal toxicity, and high temperature stress (Grover et al., 2011), while drought stress can reduce aphid fecundity in Arabidopsis (Pineda et al., 2016). Our previous work also showed that heat shock could improve the resistance of maize seedlings to heat, chilling, salt, and drought stress (Gong et al., 2001). Numerous studies found that the acquisition of stress tolerance including cross-adaptation

Front. Plant Sci. 7:1621. doi: 10.3389/fpls.2016.01621

Li Z -G, Min X and Zhou Z -H (2016) Hydrogen Sulfide: A Signal Molecule in Plant Cross-Adaptation.

was involved in a complex signal network consisting of many second messengers such as Ca2+, abscisic acid (ABA), hydrogen peroxide (H2O2) and nitric oxide (NO), as well as their crosstalk (Knight, 2000; Pandey, 2015; Li and Gu, 2016; Li and Jin, 2016; Niu and Liao, 2016; Wang et al., 2016). In tobacco, mechanical stimulation can successively trigger H2O<sup>2</sup> and NO signaling (Li and Gong, 2011, 2013), heat shock can induce Ca2<sup>+</sup> and ABA signaling one after the other (Gong et al., 1998a,b), which in turn induce cross-adaptation to heat and chilling stress, similar results were reported by Gong et al. (2001) in maize seedlings. These results indicate that the acquisition of cross-adaptation is involved in signal crosstalk among Ca2+, H2O2, NO, and ABA in plants. Recently, hydrogen sulfide (H2S) was also found to be a member of this signal network in plants (Calderwood and Kopriva, 2014; Hancock and Whiteman, 2014; Fotopoulos et al., 2015; Guo et al., 2016), indicating that H2S might be a signal molecule in plant cross-adaptation.

For a long time, H2S has been considered as merely a toxic intermediate of cell metabolism due to its strong affinity to Fe2+ containing proteins such as cytochrome oxidase, hemoglobin and myoglobin, which may have been primary cause of the mass extinction of species in the Permian (Li, 2013; Lisjak et al., 2013; Calderwood and Kopriva, 2014; Hancock and Whiteman, 2014; Fotopoulos et al., 2015; Guo et al., 2016; Yamasaki and Cohen, 2016). H2S can inhibit oxygen release from young seedlings of six rice cultivars (Bluebelle, Dawn, Norin 22, Saturn, Yubae, and Zenith) and nutrient uptake such as phosphorus (Li, 2013; Calderwood and Kopriva, 2014; Hancock and Whiteman, 2014). But nowadays, H2S is found to function as gaseous signal molecule at low concentration similar to carbon monoxide (CO) and NO in plants, and it has been shown that plants can actively synthesize endogenous H2S under normal, especially biotic or abiotic stress conditions (Li, 2013; Calderwood and Kopriva, 2014; Hancock and Whiteman, 2014; Yamasaki and Cohen, 2016). The accumulation of endogenous H2S has become a common response of plants to environmental stress, including salt, heavy metal (HM), drought, heat and cold stress, as well as pathogen infection, which may be closely associated with the acquisition of stress tolerance in plants (Li, 2013; Calderwood and Kopriva, 2014; Hancock and Whiteman, 2014). More interestingly, exogenously applied H2S, releasing from its donors such as NaHS and morpholin-4-ium 4-methoxyphenyl(morpholino) phosphinodithioate (GYY4137), shows significant positive effects on seed germination (Li et al., 2012a; Li and He, 2015; Wojtyla et al., 2016), organogenesis and growth (Lin et al., 2012; Fang T. et al., 2014), the regulation of senescence (Zhang et al., 2011), as well as the acquisition of stress tolerance such as salt (Christou et al., 2013), HM (Chen et al., 2013), drought (Christou et al., 2013), heat (Li et al., 2013a,b; Li, 2015c) and cold tolerance (Fu et al., 2013). These results indicate that H2S may be a candidate signal molecule in plant cross-adaptation. In addition, NaHS and GYY4137 are commonly used as H2S donors because they can release H2S when dissolved in water, but NaHS giving a relatively short burst of H2S, while GYY4137 giving a longer more prolonged exposure to H2S (Wang, 2012; Lisjak et al., 2013). However, whether H2S concentration in plant cells or tissues is consistent with that of NaHS and GYY4137 applied as well as actual H2S concentration triggering crossadaptation need to be further investigated. In addition, H2S usually exist in the forms of H2S (approximately 20%) and HS<sup>−</sup> (approximately 80%) in water solution, exact physiological concentration of H2S in plant cells or subcellular organelles is not clear.

Though, there are a lot of excellent reviews which expound potential physiological function of H2S in seed germination, plant growth and development, as well as the acquisition of stress tolerance (Li, 2013; Calderwood and Kopriva, 2014; Hancock and Whiteman, 2014, 2016; Jin and Pei, 2015; Guo et al., 2016; Scuffi et al., 2016; Yamasaki and Cohen, 2016), the role of H2S as a candidate signal molecule in plant cross-adaptation was not summarized in depth. Therefore, in this review, H2S homeostasis in plant cells under normal growth conditions, H2S signaling triggered by adverse environment and H2S-induced cross-adaptation to various abiotic stresses are summarized, which further uncovers that H2S may be a candidate signal molecule in plant cross-adaptation.

## H2S HOMEOSTASIS IN PLANT CELLS

As mentioned above, due to the dual role of H2S, that is, as cytotoxin at high concentration and as cell signal molecule at low concentration, H2S homeostasis in plant cells is very important to exert its physiological functions including crossadaptation induction. In plant cells, there are many metabolic pathways to regulate H2S homeostasis, similar to other signal molecules like H2O2, NO. H2S homeostasis is closely regulated by L-cysteine desulfhydrase (LCD, EC 4.4.1.1), D-cysteine desulfhydrase (DCD, EC 4.4.1.15), sulfite reductase (SiR, EC 1.8.7.1), cyanoalanine synthase (CAS, EC 4.4.1.9), and cysteine synthase (CS, EC 4.2.99.8; Li, 2013, 2015a; **Figure 1**). LCD/DCD catalyzes the degradation of L-/D-cysteine to produce H2S, amine and pyruvate; SiR reduces sulfite to H2S using ferredoxin as electron donor; H2S can be released from cysteine in the present of hydrogen cyanide by CAS; CS, namely O-acetyl- (thiol)-serinelyase (OAS-TL), can incorporate H2S into O-acetyl-L-serine to form cysteine, and its reverse reaction can release H2S (Li, 2013, 2015a; **Figure 1**). Generally, plants synthesize H2S via LCD or DCD, which respond to environment stress and induce the acquisition of stress tolerance. In addition, excess H2S can be released to air (Li, 2013; Calderwood and Kopriva, 2014; Hancock and Whiteman, 2014).

#### H2S SIGNALING TRIGGERED BY ABIOTIC STRESS

Similar to other second messengers such as Ca2+, H2O2, ABA and NO, the rapid production of endogenous H2S in many species of plant can be triggered by numerous stresses (**Table 1**; **Figure 2**), this is a common response of plants to various abiotic stresses, which is closely associated with the acquisition of crossadaptation in plants.

### H2S Signaling Triggered by Heavy Metal Stress

The rapid production of H2S has become a common response of plants to various HM stress, among HMs, Cd is the most severe stress due to its toxicity and stability (Ahmad, 2016). In rice seedlings, 0.5 mM Cd stress resulted in an increment of H2S content from approximately 5 µmol g−<sup>1</sup> fresh weight (FW) to approximately 6 µmol g−<sup>1</sup> FW. The addition of 0.1 mM NaHS caused an even further increase in the level of H2S (approximately 8 µmol g−<sup>1</sup> FW) as compared with Cd treatment alone. Exposure to 0.2 mM hypotaurine (HT, H2S scavenger) with NaHS decreased H2S level compared with NaHS alone, indicating that this elevated level of H2S is correlated with the enhanced Cd tolerance (Mostofa et al., 2015). Zhang et al. (2015) found that the endogenous H2S emission was stimulated by Cd stress in Chinese cabbage. The relative expression of DCD1 and DES1 (cysteine desulfhydrase, OAS-TL homogenous family) genes (responsible for H2S synthesis) was up-regulated after treatment with Cd with a range of concentrations (0, 5, 10, and 20 mM) for 24 h. Expression of DES1 at 5 mM Cd already showed a significant increase, and at 20 mM Cd was 4.7 times of the control. Following a similar pattern, the endogenous H2S concentrations also significantly rose from 0.38 to 0.58 nmol mg−<sup>1</sup> protein min−<sup>1</sup> at 20 mM. Chromium (Cr), existing in the form of Cr3<sup>+</sup> and Cr6+, is regarded as the second most common HM, both forms have become major environmental pollution sources. In foxtail millet seedlings, Fang H. et al. (2014) also reported that the expressions of H2S-emission related genes LCD, DCD2, and DES markedly increased during the first 12 h of Cr6<sup>+</sup> exposure following decline at 24 h, while the expression of DCD1 was consistently increased from 0 to 24 h under 10 mM Cr6<sup>+</sup> stress. Additionally, the H2S production rate is induced by Cr6<sup>+</sup> stress in dose- and time-dependent manner, and this induction was the most significant with 24 h of 10 mM Cr6<sup>+</sup> treatment (from 0.6 to 1.6 nmol mg−<sup>1</sup> protein min−<sup>1</sup> ). These results imply that endogenous H2S synthesis was activated by Cr6<sup>+</sup> stress by activating its emission system in foxtail millet. Inconceivably, in compared with to other plant species, the both species Chinese cabbage and foxtail millet show a remarkable tolerance to HM (Cd and Cr6+). At 20 mM Cd for Chinese cabbage and 10 mM Cr6<sup>+</sup> for foxtail millet, these treatment concentrations are far beyond the physiological level (generally micromolar concentrations) for many plant species, the precise physiological, biochemical, and molecular mechanisms are waiting for being uncovered.

### H2S Signaling Triggered by Salt Stress

Salt stress commonly leads to an osmotic stress response, similar to drought stress, which triggers rapid generation of second messengers like H2S. In alfalfa seedlings, the increasing concentration of NaCl (from 50 to 300 mM) progressively caused the induction of total LCD activity and the increase of

#### TABLE 1 | Different abiotic stresses trigger endogenous H2S production in plants.


The FW and Pr in the table represent fresh weight and protein respectively.

endogenous H2S production (from 30 to 70 nmol g−<sup>1</sup> FW) (Lai et al., 2014). Exposure of strawberry seedlings to salinity (100 mM NaCl) and non-ionic osmotic stress (10% PEG-6000) greatly enhanced H2S concentration (48 and 50 nmol g−<sup>1</sup> FW) in leaves, while 0.1 mM NaHS-pretreated plants subsequently exposed for 7 days to both stress factors were found to accumulate significantly higher amounts of H2S (55 nmol g−<sup>1</sup> FW) in their leaves compared with NaCl-stressed plants (Christou et al., 2013).

### H2S Signaling Triggered by Drought Stress

One of the most severe abiotic stresses being experienced worldwide is drought. In Arabidopsis seedlings, the results of Shen et al. (2013) showed that treating wild type with polyethylene glycol (PEG) 8000, to simulate drought stress, caused an increase in production rate of endogenous H2S (0.8 nmol mg−<sup>1</sup> protein min−<sup>1</sup> ). At early stage of osmotic exposure (PEG 6000 for 2 days), the endogenous H2S in wheat seeds rapidly increased from 1.5 to 3.5 µmol g−<sup>1</sup> dry weight (DW) (Zhang et al., 2010a).

### H2S Signaling Triggered by Low Temperature Stress

Low temperature is a major environmental stress factors that limit plant growth, development and distribution. In grape (Vitis vinifera L.) seedlings, chilling stress at 4◦C induced the expression of L/DCD genes and increased the activities of L/DCD, which in turn enhanced endogenous H2S accumulation (from 7 to 15 µmol g−<sup>1</sup> FW) (Fu et al., 2013). Similarly, Shi et al. (2013) also found that cold stress treatment at 4 ◦C could induce the accumulation of endogenous H2S level (14 nmol g−<sup>1</sup> FW) in bermudagrass [Cynodon dactylon (L). Pers.] seedlings. To uncover the adaptive strategies of alpine plants to the extremely cold conditions prevailing at high altitudes, Ma et al. (2015), using a comparative proteomics, investigated the dynamic patterns of protein expression in Lamiophlomis rotata plants grown at three different altitudes (4350, 4,800, and 5,200 m), and the results showed that the levels and enzyme activities of proteins (OAS-TL, CAS, L/DCD) involved in H2S biosynthesis markedly increased at higher altitudes (4,800 and 5,200 m), and that H2S accumulation increased to 12, 22, and 24 nmol g−<sup>1</sup> FW, respectively, demonstrating that H2S plays a central role during the adaptation of L. rotata to environmental stress at higher altitudes.

### H2S Signaling Triggered by High Temperature Stress

Similar to other stresses, high temperature also can induce endogenous H2S generation in many species of plant. In 3-week-old seedlings of tobacco, Chen et al. (2016) found that treatment with high temperature at 35◦C increased the activity of LCD, which in turn induced the production of endogenous H2S (8 nmol g−<sup>1</sup> FW) in tobacco seedlings, and that H2S production remained elevated level after 3 days of high temperature exposure. More interestingly, H2S production by high temperature can induce the accumulation of jasmonic acid, followed by promoting nicotine synthesis. These data suggest that H2S and nicotine biosynthesis is linked in tobacco plants subjected to high temperature stress. Additionally, heat stress caused a marked modulation in H2S content in strawberry seedlings, as indicated in a significant increase after 1, 4, and 8 h of exposure to 42◦C compared with control plants. A significant increase in H2S content was also observed in 0.1 mM NaHS-pretreated plants after 1 h exposure to heat stress, gradually lowering to control levels thereafter (Christou et al., 2014).

### H2S Signaling Triggered by UV-B Radiation

Recently, Li et al. (2016) found that UV-B radiation could induce H2S production in leaves of barley seedlings, reaching a peak of approximately 230 nmol g−<sup>1</sup> FW after 12 h of exposure, which in turn promoted the accumulation of UV-absorbing compounds flavonoids and anthocyanins. H2S began to decline with time, but it is overall significantly higher than that of the control (approximately 125 nmol g−<sup>1</sup> FW) at 48 h of exposure. A similar trend was observed for LCD activity, which was corroborated by the application of DL-propargylglycine (PAG, an inhibitor of LCD) that resulted in complete inhibition of the H2S production and the accumulation of UV-absorbing compounds induced by UV-B radiation (Li et al., 2016).

### H2S Signaling Triggered by Hypoxia and Fungal Infection

Flooding often leads to hypoxia in plant roots, which significantly limits agriculture production. In pea (Pisum sativum L.) seedlings, Cheng et al. (2013) found that hypoxia could activate H2S biosynthesis system (LCD, DCD, OAS-TL, and CS), which in turn induced the accumulation of endogenous H2S from approximately 0.9 (control) to 5.1 µmol g−<sup>1</sup> FW (hypoxia for 24 h), indicating that H2S might be a hypoxia signaling that triggers the tolerance of the pea seedlings to hypoxic stress, this hypothesis was further supported by exogenously applied NaHS.

Pathogen infection is a common biotic stress in plants. In oilseed rape (Brassica napus L.) seedlings, fungal infection with Sclerotinia sclerotiorum led to an even stronger increase in H2S, reaching a maximum of 3.25 nmol g−<sup>1</sup> DW min−<sup>1</sup> 2 days after infection, suggesting that the release of H2S seems to be part of the response to fungal infection (Bloem et al., 2012).

### H2S Signaling Triggered by Exogenously Applied NaHS or Up-regulating the Expression of L/DCD

In addition to above-described abiotic and biotic stressors, H2S signaling in plant cells also can be triggered by exogenously applying NaHS (H2S donor) or up-regulating the expression of genes involved in H2S biosynthesis like L/DCD under normal growth conditions. In strawberry seedlings, treatment of root with 0.1 mM NaHS resulted in significantly elevated H2S concentration (35 nmol g−<sup>1</sup> FW) in leaves compared with control plants (25 nmol g−<sup>1</sup> FW) (Christou et al., 2013). In wheat seeds, the endogenous H2S level [4.5 µmol g−<sup>1</sup> dry weight (DW)] in NaHS-treated seed was slightly higher than that of control (1.7 mol g−<sup>1</sup> DW) (Zhang et al., 2010a). These results indicated that H2S is easy to enter into plant cells and follow on being transported to other tissues or organs due to its highly lipophilic property, which in turn exert its physiological role in plants.

Additionally, Jin et al. (2011) found that the Arabidopsis seedlings expressing L/DCD showed higher endogenous H2S content under both normal (6 nmol mg−<sup>1</sup> protein min−<sup>1</sup> ) and drought stress conditions (14 nmol mg−<sup>1</sup> protein min−<sup>1</sup> ) compared with the control (3 nmol mg−<sup>1</sup> protein min−<sup>1</sup> ), and the expression pattern of L/DCD was similar to the drought associated genes dehydration-responsive element-binding proteins (DREB2A, DREB2B, CBF4, and RD29A) induced by dehydration, while exogenous application of H2S (80 µM) was also found to stimulate further the expression of drought associated genes. In addition, drought stress significantly induced endogenous H2S production in both transgenetic plant and wild type, a process that was reversed by re-watering (Jin et al., 2011). Interestingly, Arabidopsis seedlings overexpressing LCD or pre-treated with NaHS exhibited higher endogenous H2S level (from 2 to 10 nmol g−<sup>1</sup> FW), followed by improving abiotic stress (drought, salt, and chilling) tolerance and biotic stress (bacteria) resistance, while LCD knockdown plants or HT (H2S scavenger) pre-treated plants displayed lower endogenous H2S level and decreased stress resistance (Shi et al., 2015).

In conclusion, above-mentioned researches in this section display that: (1) under normal growth conditions, the content of endogenous H2S or production rate in various plant species are different, ranging from 2 nmol g−<sup>1</sup> FW to 7 µmol g−<sup>1</sup> FW or 0.38 to 6 nmol mg−<sup>1</sup> protein min−<sup>1</sup> . These differences may be relative to measurement methods, plant species and development stage, and experiment system. (2) Under abiotic stress conditions, the level of endogenous H2S in various plant species is averagely increased by 2∼2.5-fold, indicating that different environment stresses can trigger the H2S signaling, which may be a trigger that induces the acquisition of cross-adaptation in plants.

## H2S-INDUCED CROSS-ADAPTATION

As described above, not only there are a broad range of environmental stressors can trigger H2S signaling in plants, but pretreating plants with exogenously applied H2S can provide additional resistance to subsequent stress exposure. The next section explores the role of H2S as an important signaling molecule for cross-adaptation to HM, salt, drought, cold, heat and flooding stress by enhancing antioxidant system activity, accumulating osmolyte, synthesizing heat shock proteins (HSPs), as well as maintaining ion and nutrient balance (**Table 2**; **Figure 3**), which may be common mechanism of crossadaptation induced by H2S.

### H2S-Induced Metal and Metalloid Tolerance

Heavy metals refer to a group of metal elements with a density greater than 6 g/cm<sup>3</sup> , including Cr, Cu, Zn, and so forth (Gupta et al., 2013; Ahmad, 2016). Due to their toxicity and stablility, HM has become the major abiotic stress in plants, and even threatens human health by way of the food chain. HM stress commonly results in oxidative stress, that is, the excessive accumulation of ROS, which leads to lipid peroxidation, protein oxidation, enzyme inactivation, and DNA damage (Yadav, 2010; Gupta et al., 2013; Ahmad, 2016). However, higher plants have evolved a sophisticated antioxidant defense system to scavenge excessive ROS and maintain its homeostasis in plants (Foyer and Noctor, 2009, 2011).

Arsenic (As) is a highly toxic metalloid, it is major pollutant in the soil. In pea seedlings, As treatment increased the accumulation of ROS, which in turn damage to lipids, proteins and biomembranes. Meanwhile, higher cysteine level was observed in As-stress seedlings in comparison to all other treatments (As-free; NaHS; As + NaHS), while these effects were alleviated by the addition of NaHS (Singh et al., 2015). Further experiments showed that As treatment inhibited the activity of the enzymes involved in the ascorbic acid (AsA)– glutathione (GSH) cycle, whereas their activities were enhanced by application of NaHS (Singh et al., 2015). In addition, the redox status of AsA and GSH was disturbed, as indicated by a steep decline in their reduced/oxidized ratios. However, exogenously applied NaHS restored the redox status of the AsA and GSH pools under As stress (Singh et al., 2015). Furthermore, NaHS treatment ameliorated As toxicity, which was coincided with the increased accumulation of H2S. The results demonstrated that H2S might counterbalance ROS-mediated damage to macromolecules by reducing the accumulation of As and triggering up-regulation of the AsA–GSH cycle, further suggesting that H2S plays a crucial role in plant priming, and in particular for pea seedlings in mitigating As stress.

Under Cr stress, exogenous application of NaHS could improve the germination rate of wheat seeds in a dosedependent manner and the activities of amylase, esterase as well as antioxidant enzymes superoxide dismutase (SOD), catalase (CAT), ascorbate peroxidase (APX) and glutathione peroxidase

#### TABLE 2 | NaHS (H2S donor)-induced cross-adaptation in plants.


(GPX), whereas reduced the activity of lipoxygenase and overproduction of malondialdehyde (MDA) as well as H2O<sup>2</sup> induced by Cr, and sustained higher endogenous H2S level (Zhang et al., 2010b). Additionally, NaHS pretreatment increased the activities of SOD and CAT, but decreased that of lipoxygenase in wheat under Cu stress (Zhang et al., 2008), these results were consisted with the response of wheat to Cr stress (Zhang et al., 2010b).

Also, NaHS could alleviate the inhibitory effect of Cu stress in wheat in a dose-dependent manner, and H2S or HS<sup>−</sup> derived from NaHS rather than other sulfur-containing components (S2−, SO<sup>4</sup> <sup>2</sup>−, SO32−, HSO<sup>4</sup> <sup>−</sup>, and HSO<sup>3</sup> <sup>−</sup>) attribute to the potential role in promoting seed germination under Cu stress (Zhang et al., 2008). Further experiments showed that NaHS could increase amylase and esterase activities, reduced the disturbance of plasma membrane integrity induced by Cu in the radicle tips, and sustain lower MDA and H2O<sup>2</sup> levels in germinating seeds (Zhang et al., 2008), similar to the reports by (Zhang et al., 2010b).

Aluminium (Al), a non-essential element for plants, adversely affects plant growth, development and survival, especially in acid soil. In barley (Hordeum vulgare L.) seedlings, Al stress inhibited the elongation of roots, while pretreatment with NaHS partially rescued the inhibition of root elongation induced by Al, and this rescue was closely correlated with the decrease of Al accumulation in seedlings (Chen et al., 2013). Additionally, application of NaHS significantly alleviated citrate secretion and oxidative stress (as indicated in lipid peroxidation as well as ROS burst) induced by Al by activating the antioxidant system (Chen et al., 2013). Similar results were reported by Zhang et al. (2010c) in wheat (Triticum aestivum L.).

Though zinc (Zn) is an essential element for plants, its toxic effects can be observed when being excessive accumulation in plants. In Solanum nigrum L. seedlings, H2S ameliorated the inhibition of growth by excess Zn, especially in roots, and an increase in free cytosolic Zn2<sup>+</sup> content in roots, which was correlated well with the down-regulation of Zn uptake and homeostasis related genes expression like zincregulated transporter (ZRT), iron-regulated transporter (IRT) like protein (ZIP) and natural resistance associated macrophage protein (NRAMP) (Liu et al., 2016). In addition, H2S further enhanced the expression of the metallothioneins to chelate excessive Zn and alleviated Zn-oxidative stress by regulating the genes expression of antioxidant enzymes (Liu et al., 2016).

## H2S-Induced Salt Tolerance

Salts stress is negative effects of excessive salt on seed germination, plant growth and development, and even survival, which is a major abiotic stress in agriculture production worldwide. Salt stress commonly leads to direct and indirect injury, namely ion toxicity, osmotic stress, nutrient imbalance, and oxidative stress (Ahmad et al., 2013a,b). To combat with salt injury, plants have evolved many protective strategies, including osmotic adjustment by synthesizing osmolytes such as proline (Pro), glycine betaine (GB), trehalose (Tre), and total soluble sugar (TSS); ion and nutrient balance by regulating transporter; and enhancement of antioxidant capacity by activating the activity of antioxidant enzymes SOD, CAT, APX, GPX and glutathione reductase (GR), as well as by synthesizing antioxidants like AsA and GSH (Ahmad et al., 2013a,b). In saltsensitive wheat cultivar LM15, the results of Bao et al. (2011) showed that wheat seed priming with different concentrations of NaHS (0.01, 0.05, 0.09, 0.13 mM) for 12 h could significantly alleviate the inhibition of seed germination and seedling growth induced by 100 mM NaCl in a concentration-dependent manner, as indicated in germination rate, germination index, vigor index and growth of seedlings of wheat. In alfalfa (Medicago sativa), NaHS pretreatment differentially activate total and isoenzymatic activities as well as corresponding transcripts of antioxidant enzymes (SOD, CAT, POD, and APX) under 100 mM NaCl stress, thus resulting in the alleviation of oxidative damage induced by NaCl (Wang et al., 2012). In addition, NaCl stress inhibited seed germination and seedling growth, but pretreatment with NaHS could significantly attenuate this inhibitive effect and increase the ratio of potassium (K) to sodium (Na) in the root parts (Wang et al., 2012). Also, under 100 mM NaCl stress, Arabidopsis roots displayed a great increase in electrolyte leakage and Na+/K<sup>+</sup> ratio, indicating that Arabidopsis was sensitive to salt stress, while treatment with NaHS enhanced the salt tolerance by maintaining a higher K+/Na<sup>+</sup> ratio (Li J. et al., 2014). In addition, the level of gene expression and the phosphorylation of plasma membrane H+-ATPase and Na+/H<sup>+</sup> antiporter protein was promoted by H2S, while the effect of H2S on the plasma membrane Na+/H<sup>+</sup> antiporter system was removed by diphenylene iodonium (DPI, a PM NADPH oxidase inhibitor) or dimethylthiourea (DMTU, an ROS scavenger) (Li J. et al., 2014), suggesting that H2S can maintain ion homeostasis in salt-stress Arabidopsis root in the H2O2 dependent manner.

### H2S-Induced Drought Tolerance

Similar to other stressors, drought stress, namely water deficiency, usually leads to osmotic stress and oxidative stress, which adversely affects plant growth, development and production (Iqbal et al., 2016). Plants can maintain water balance and ROS homeostasis by osmotic adjustment and antioxidant system (Foyer and Noctor, 2009, 2011; Iqbal et al., 2016). Zhang et al. (2010d) found that the germination rate reduced gradually with the increasing concentrations of PEG-6000, which mimicked osmotic stress, while NaHS treatment could promote wheat seed germination under osmotic stress in a dose-dependent manner, Na<sup>+</sup> and other sulfur-containing components (S2−, SO<sup>4</sup> <sup>2</sup>−, SO32−, HSO<sup>4</sup> <sup>−</sup>, and HSO<sup>3</sup> <sup>−</sup>) were not able to replace NaHS, confirming H2S or HS<sup>−</sup> derived from NaHS contribute to the protective roles (Zhang et al., 2010d). Further experiments showed that NaHS treatment significantly increased CAT and APX activities, reduced that of lipoxygenase as well as the accumulation of MDA and H2O<sup>2</sup> in seeds (Zhang et al., 2010d). Additionally, exogenously applied NaHS increased the activities of APX, GR, dehydroascorbate reductase (DHAR) and gamma-glutamylcysteine synthetase in wheat seedlings, as well as the contents of AsA, GSH, total ascorbate and total glutathione under water stress compared to the control without NaHS treatment, which in turn decreased the MDA content and electrolyte leakage induced by water deficiency in wheat

seedlings (Shan et al., 2011). In Arabidopsis seedlings, under drought stress, the expression pattern of L/DCD was similar to the drought associated genes, whose express was stimulated further by H2S (Jin et al., 2011). Also, seedlings treated with NaHS exhibited a higher survival rate and a significant reduction in the size of the stomatal aperture compared to the control (Jin et al., 2011). In addition to these, García-Mata and Lamattina (2010) also found that, in Vicia faba (L.) var. major and Impatiens walleriana Hook. f., H2S treatment could increase relative water content (RWC) and protect plants against drought stress.

## H2S-Induced Cold Tolerance

Low temperature stress includes chilling stress (>0 ◦C) and freezing stress (<0 ◦C). Low temperature usually leads to osmotic stress and oxidative stress, plants can reduce the low temperature injury by osmotic adjustment and activating antioxidant system (Foyer and Noctor, 2009, 2011; Iqbal et al., 2016). Shi et al. (2013) found that exogenous application of NaHS conferred multiple stress tolerance including freezing tolerances in bermudagrass, in reflected in decreased electrolyte leakage and increased survival rate under freezing conditions. Additionally, NaHS treatment mitigated the ROS burst and cell damage induced by freezing stress via modulating the activities of antioxidant enzymes CAT, GPX and GR, as well as non-enzymatic GSH pool and redox state (Shi et al., 2013). In grape (Vitis vinifera L) seedlings, Fu et al. (2013) reported that treatment with NaHS showed the high activity of SOD and gene expression of VvICE1 and VvCBF3, lowed superoxide radical and MDA levels as well as cell membrane permeability under chilling stress at 4◦C, while HT treatment displayed contrary effect under the chilling stress. Also, Arabidopsis seedlings overexpressing LCD or pretreating with NaHS exhibited higher endogenous H2S level and stronger chilling stress tolerance, while LCD knockdown or HT pretreated plants displayed lower endogenous H2S level and weaker stress resistance. Moreover, H2S could up-regulate the expression of genes involved in multiple abiotic and biotic stress and inhibited ROS accumulation (Shi et al., 2015). Ma et al. (2015) found that the levels and enzyme activities of proteins involved in H2S biosynthesis (L/DCD, CAS, OAS-TL) markedly increased at higher altitudes at 4800 and 5200 m, which in turn maintained higher H2S level. Exogenous H2S application reduced ROS and RNS (reactive nitrogen species) damage by increasing antioxidant enzyme and GSNOR (S-nitrosoglutathione reductase) activities, activated the downstream defense response, resulting in protein degradation as well as Pro and SS accumulation. However, such defense responses could be reversed by HT and PAG, respectively. These results illustrated that H2S plays a central role in L. rotata uses multiple strategies to adapt to the alpine stress environment. Also, H2S fumigation maintained higher values of lightness and peel firmness of banana fruit and reduced the accumulation of MDA under chilling stress (Luo et al., 2015). In addition, H2S could increase the activities of GPX, SOD, CAT, APX, GR and the phenylalanine ammonia lyase and total phenolics content, which in turn improved antioxidant capacity of banana fruits, reducing H2O<sup>2</sup> and superoxide anion accumulation (Luo et al., 2015). Further experiments also found that H2S fumigation elevated Pro content by activating P5CS activity and decreasing that of ProDH, which might be related to chilling injury tolerance improvement (Luo et al., 2015), similar to the report by Li and Gong (2013). These data indicate that H2S alleviated the chilling injury may be achieved through the enhancement of antioxidant system and Pro accumulation in banana fruit.

## H2S-Induced Heat Tolerance

Along with global warming, high temperature has already become a noticeable abiotic stress worldwide, and the mechanisms of high temperature injury and heat tolerance have attracted much attention (Wahid et al., 2007; Asthir, 2015; Hemmati et al., 2015). Christou et al. (2011) found that pre-treatment of roots with NaHS effectively alleviated the decrease in leaf chlorophyll fluorescence, stomatal conductance and relative leaf water content in strawberry (Fragaria x ananassa cv. Camarosa) under heat stress at 42◦C, as well as an increase in ion leakage and MDA accumulation in comparison with plants directly subjected to heat stress. In addition, NaHS pretreatment preserved AsA/GSH homeostasis, as evidenced by lower AsA and GSH pool redox disturbances and enhanced transcription of AsA and GSH biosynthetic enzymes, 8 h after heat stress exposure. Furthermore, NaHS root pretreatment increased the gene expression of antioxidant enzymes (cAPX, CAT, MnSOD, GR), heat shock proteins (HSP70, HSP80, HSP90), and aquaporins (PIP) (Christou et al., 2014). These results suggest that H2S root pretreatment activates a coordinated network of heat shock defense-related pathways at a transcriptional level and systemically protects strawberry plants from heat stress-induced damage. Our previous study also showed that 0.7 mM NaHS treatment increased the activities of CAT, GPX, SOD and GR, and the contents of GSH and AsA, as well as the ratio of reduced antioxidants to total antioxidants [AsA/(AsA+DHA) and GSH/(GSH +GSSG)] in maize seedlings under normal culture conditions compared with the control (Li Z.G. et al., 2014). Under heat stress, antioxidant enzymes activities, antioxidants contents and the ratio of the reduced antioxidants to total antioxidants in control and treated seedlings all decreased, but NaHS-treated seedlings maintained higher antioxidant enzymes activities and antioxidants levels as well as reduced antioxidants/total antioxidants ratio (Li Z.G. et al., 2014), similar results also were found in tobacco cells (Li et al., 2015). In addition, NaHS pretreatment significantly increased the survival percentage of tobacco cells under heat stress and regrowth ability after heat stress, alleviated a decrease in vitality of cells and an increase in electrolyte leakage and MDA accumulation (Li et al., 2012b). Meanwhile, the heat tolerance induced by NaHS was markedly enhanced by exogenous application of Ca2<sup>+</sup> and its ionophore A23187, respectively, while was weakened by addition of Ca2<sup>+</sup> chelator ethylene glycol-bis(b-aminoethylether)- N,N,N<sup>0</sup> ,N0 -tetraacetic acid, plasma membrane channel blocker La3+, as well as calmodulin antagonists chlorpromazine and trifluoperazine, respectively (Li et al., 2012b). Similarly, in maize, pretreatment with NaHS markedly improved the germination percentage of seeds and the survival percentage of seedlings under heat stress, alleviated an increase in electrolyte leakage

of roots and a decrease in tissue vitality and accumulation of MDA in coleoptiles of maize seedlings (Li et al., 2013a). Furthermore, NaHS pretreatment could improve the activity of 1<sup>1</sup> -pyrroline-5-carboxylate synthetase (P5CS) and lowered that of Pro dehydrogenase (ProDH), which in turn induced the accumulation of endogenous Pro in maize seedlings (Li et al., 2013a). Also, exogenously applied Pro could increase endogenous Pro content, followed by increase in the survival percentage of maize seedlings under heat stress (Li et al., 2013a). These results suggest that NaHS pretreatment can improve the heat tolerance in plants and the acquisition of heat tolerance induced by NaHS may require the synergistic effect of antioxidant system, calcium messenger system, HSPs and Pro.

### H2S-Induced Flooding Tolerance and Pathogen Resistance

Flooding stress usually causes hypoxia, and even anoxia in plant roots, plants can improve hypoxia tolerance by reducing oxidative damage (van Dongen and Licausi, 2014). Cheng et al. (2013) found that hypoxia could induce root tip death of pea seedlings, while pretreatment with exogenous H2S dramatically alleviated cell death by protecting root tip cell membranes from ROS damage induced by hypoxia and by inhibiting ethylene production. Conversely, root tip death induced by hypoxia was strongly enhanced by inhibiting the key enzymes responsible for endogenous H2S biosynthesis (adding hydroxylamine to inhibit LCD activity). These results demonstrated that H2S can enhance the tolerance of the plant to hypoxic stress by alleviating hypoxiainduced root tip death in pea seedlings.

More interestingly, H2S also could transcriptionally regulate MIR393-mediate auxin signaling, including MIR393a/b and their target genes (TIR1, AFB1, AFB2, and AFB3), and this regulation was related with H2S-induced antibacterial resistance (Shi et al., 2015).

All of the above studies in this section show exogenous application of NaHS (a H2S donor) can induce cross-adaptation to HM, salt, osmosis, drought, cold, heat and hypoxia stresses in different plant species, and the optimal NaHS concentration range from 0.05 to 1.5 mM (**Table 2**), while higher NaHS concentration (>1.5 mM) exhibits negative effect on plant growth, development, survival, and even the acquisition of stress tolerance. Therefore, the optimal concentration of NaHS should be carefully selected according to plant species and experimental system.

### CONCLUSION AND FUTURE PROSPECTIVE

In general, after undergoing a moderate stress, plants not only can improve the resistance to this stress, but also can increase the tolerance to subsequent other stresses, which known as crossadaptation. Many studies found that signaling triggered by a moderate stress, such as Ca2+, ABA, H2O2, and NO signaling, is a common response of plants to abiotic and biotic stress, which in turn induces the acquisition of cross-adaptation. In addition, exogenously applied these signal molecules also can trigger corresponding signaling, followed by improving stress tolerance of plants, thus Ca2+, ABA, H2O2, and NO are considered to be candidate signal molecules in cross-adaptation in plants (Knight, 2000; Gong et al., 2001; Li and Gong, 2011; Li et al., 2012b; Fang H. et al., 2014; Qiao et al., 2015; Chen et al., 2016). More recently, many research groups found that a number of abiotic stresses also can trigger H2S signaling, while exogenously applied H2S can induce cross-adaptation to multiple stresses, indicating that H2S represents a potential candidate signal molecule in crossadaptation in plants (Li, 2013; Lisjak et al., 2013; Calderwood and Kopriva, 2014; Hancock and Whiteman, 2014; Fotopoulos et al., 2015; Guo et al., 2016). However, H2S acts as a signal molecule in plants cross-adaptation, the following questions need to be further answered: (1) Receptor or target of H2S. Due to H2S is easy to penetrate the cell membrane, maybe there is no H2S receptor in plant cells, but Li et al. (2011) and Aroca et al. (2015) found that H2S could modify the activity of some proteins with sulfhydryl (-SH) by sulfhydrylation (-SSH), whether these proteins are the receptors of H2S needs to be further research. (2) Physiological concentration of H2S. Many assay methods for H2S including colorimetric, fluorescence-based, gas chromatographic and electrochemical methods give highly contrasting results (**Table 2**; Peng et al., 2012; Li, 2015b), so accurate physiological concentration of H2S in plant cells or organelles is waiting for uncovering. It will be important to design the stress treatments closer to physiologically relevant stress intensities, thus low micromolar rather than millimolar HM concentration should be investigated in order to strengthen the conclusions. (3) Crosstalk between H2S and other signal molecules in crossadaptation. The acquisition of abiotic tolerance is involved in a signal network consisting of many signal molecules including H2S, interaction among signal molecules needs to be updated and perfected. (4) Physiological, biochemical and molecular mechanisms of H2S-induced cross-adaptation. The study on H2S-induced abiotic tolerance including cross-adaptation has just started, many physiological, biochemical and molecular mechanisms require being expounded using transcriptome, proteome and metabolome approaches.

### AUTHOR CONTRIBUTIONS

Z-GL wrote and revised the paper, XM and Z-HZ provided the idea.

#### ACKNOWLEDGMENTS

This research is supported by National Natural Science Foundation of China (31360057) and Doctor Startup Foundation of Yunnan Normal University China (01200205020503099). We appreciate the reviewers and editors for their exceptionally helpful comments about the manuscript.

### REFERENCES

fpls-07-01621 October 24, 2016 Time: 16:15 # 10



Zhang, L., Pei, Y., Wang, H., Jin, Z., Liu, Z., Qiao, Z., et al. (2015). Hydrogen sulfide alleviates cadmium-induced cell death through restraining ROS accumulation in roots of Brassica rapa L. ssp. pekinensis. Oxid. Med. Cell Longev. 2015, 1–11. doi: 10.1155/2015/714756

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer SM and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2016 Li, Min and Zhou. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

fpls-07-01621 October 24, 2016 Time: 16:15 # 12

# Analysis of Drought-Induced Proteomic and Metabolomic Changes in Barley (Hordeum vulgare L.) Leaves and Roots Unravels Some Aspects of Biochemical Mechanisms Involved in Drought Tolerance

Klaudia Chmielewska<sup>1</sup>† , Paweł Rodziewicz<sup>1</sup>† , Barbara Swarcewicz<sup>1</sup>† , Aneta Sawikowska<sup>2</sup>† , Paweł Krajewski<sup>2</sup> , Łukasz Marczak<sup>1</sup> , Danuta Ciesiołka<sup>1</sup> , Anetta Kuczynska ´ 2 , Krzysztof Mikołajczak<sup>2</sup> , Piotr Ogrodowicz<sup>2</sup> , Karolina Krystkowiak<sup>2</sup> , Maria Surma<sup>2</sup> , Tadeusz Adamski<sup>2</sup> , Paweł Bednarek<sup>1</sup> \* and Maciej Stobiecki<sup>1</sup> \*

1 Institute of Bioorganic Chemistry – Polish Academy of Sciences, Poznan, Poland, ´ 2 Institute of Plant Genetics – Polish Academy of Sciences, Poznan, Poland ´

In this study, proteomic and metabolomic changes in leaves and roots of two barley (Hordeum vulgare L.) genotypes, with contrasting drought tolerance, subjected to water deficit were investigated. Our two-dimensional electrophoresis (2D-PAGE) combined with matrix-assisted laser desorption time of flight mass spectrometry (MALDI-TOF and MALDI-TOF/TOF) analyses revealed 121 drought-responsive proteins in leaves and 182 in roots of both genotypes. Many of the identified drought-responsive proteins were associated with processes that are typically severely affected during water deficit, including photosynthesis and carbon metabolism. However, the highest number of identified leaf and root proteins represented general defense mechanisms. In addition, changes in the accumulation of proteins that represent processes formerly unassociated with drought response, e.g., phenylpropanoid metabolism, were also identified. Our tandem gas chromatography – time of flight mass spectrometry (GC/MS TOF) analyses revealed approximately 100 drought-affected low molecular weight compounds representing various metabolite types with amino acids being the most affected metabolite class. We compared the results from proteomic and metabolomic analyses to search for existing relationship between these two levels of molecular organization. We also uncovered organ specificity of the observed changes and revealed differences in the response to water deficit of drought susceptible and tolerant barley lines. Particularly, our results indicated that several of identified proteins and metabolites whose accumulation levels were increased with drought in the analyzed susceptible barley variety revealed elevated constitutive accumulation levels in the drought-resistant line. This may suggest that constitutive biochemical predisposition represents a better drought tolerance mechanism than inducible responses.

Keywords: abiotic stress, barley, drought stress, mass spectrometry, metabolome, proteome

#### Edited by:

Qingsong Lin, National University of Singapore, Singapore

#### Reviewed by:

Ján A. Miernyk, University of Missouri, USA Uener Kolukisaoglu, University of Tübingen, Germany

#### \*Correspondence:

Maciej Stobiecki mackis@ibch.poznan.pl Paweł Bednarek bednarek@ibch.poznan.pl

†These authors have contributed equally to this work and should be considered co-first authors.

#### Specialty section:

This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science

> Received: 06 May 2016 Accepted: 12 July 2016 Published: 26 July 2016

#### Citation:

Chmielewska K, Rodziewicz P, Swarcewicz B, Sawikowska A, Krajewski P, Marczak Ł, Ciesiołka D, Kuczynska A, Mikołajczak K, ´ Ogrodowicz P, Krystkowiak K, Surma M, Adamski T, Bednarek P and Stobiecki M (2016) Analysis of Drought-Induced Proteomic and Metabolomic Changes in Barley (Hordeum vulgare L.) Leaves and Roots Unravels Some Aspects of Biochemical Mechanisms Involved in Drought Tolerance. Front. Plant Sci. 7:1108. doi: 10.3389/fpls.2016.01108

### INTRODUCTION

fpls-07-01108 July 26, 2016 Time: 12:2 # 2

Water scarcity (physical and economic) is a problem faced by modern agriculture and has the greatest impact on reduced field production. This issue will remain a serious threat to world food security unless means to circumvent the impacts of water deficit is developed. Plants, as sessile organisms, have evolved a variety of mechanisms to confront challenges from adverse abiotic and biotic environmental factors. For example, during drought stress, to limit water loss through transpiration plants close their stomata, but as a side effect it leads to diminished internal CO<sup>2</sup> concentration which eventually results in lower photosynthetic rate (Cornic, 2000). The lower CO<sup>2</sup> fixation depletes glycerate-3-phosphate, which decreases NADPH use in the Calvin cycle. The result is lower level of NADP+, which is the primary electron acceptor in photosystem I (PSI); O<sup>2</sup> reduction is accelerated which leads to excessive generation of reactive oxygen species (ROS; Asada, 2006; Murata et al., 2007; Chen and Murata, 2008). ROS, unlike atmospheric oxygen, can oxidize various biomolecules and internal cell structures, thereby causing cellular damage (Dat et al., 2000). During drought, plants may increase activities of ROS-scavenging pathways, including water-water and ascorbate–glutathione cycles and enzymes like superoxide dismutase and catalase (Mittler, 2002; Asada, 2006). Drought may also trigger molecular chaperone accumulation (Wang et al., 2004) and the overproduction of compatible organic solutes (Serraj and Sinclair, 2002). Glycine betaine (GB) and proline are two main plant osmolytes that accumulate in response to various abiotic stresses (Ashraf and Foolad, 2007).

Proteomics and metabolomics became powerful tools for analyzing plant reactions to various environmental stimuli. Especially comparative studies of genetically diverse germplasms subjected to adverse conditions like drought provide valuable insights into plant responses to pre-determined stress and give information of the biochemical pathways that participate in acclimation to environmental constrains. Proper evaluation of proteomic and metabolomic data can contribute to a process of biomarker discovery. In the case the potential candidates are successfully correlated with corresponding quantitative trait loci (QTLs), they can be further integrated into markerassisted breeding strategy to enhance selection of plants with desired traits (Tuberosa and Salvi, 2006). Many studies were performed to determine the effects of drought and other abiotic stresses on crop plants at the proteomic and metabolomic level (Kopka et al., 2004; Riccardi et al., 2004; Cramer et al., 2007; Zuther et al., 2007; Caruso et al., 2009; Skirycz et al., 2010; Ford et al., 2011; Bowne et al., 2012; Witt et al., 2012). In these studies abundantly identified proteins and metabolites were usually involved in defense mechanisms, including detoxification enzymes, redox status regulation, signaling pathways, protein folding and degradation, photosynthesis, and primary metabolism.

Barley is the fourth most important cereal in total worldwide production after wheat, maize, and rice<sup>1</sup> . It is used for animal feed, malt production, and human food. The species is a good

<sup>1</sup>http://www.faostat.fao.org

experimental model for studying cereal plant biology, primarily due to its short life cycle (∼90 days), autogamous nature and relatively small diploid genome (5.3 Gbp), particularly when compared to hexaploid wheat (18 Gbp). Recently, the physical, genetic, and functional sequence of the barley genome was assembled (The International Barley Genome and Sequencing Consortium, 2012). Drought or salt susceptible and tolerant barley varieties were also recently examined under abiotic stresses and subjected to proteomic or metabolomic analyses (Ashoub et al., 2013). However there is a lack of publications that cover simultaneously the area of proteomic and metabolomic research performed in one analysis.

The aim of the present study was to evaluate with mass spectrometry techniques proteome and metabolome changes in leaves and roots of two barley (Hordeum vulgare L.) genotypes of different origin subjected to drought. The phenotypic traits measured at the harvest point indicated significant differences in the tolerance to water deficit between those two lines. This distinctive effect of drought on growth of both barley varieties allowed us to correlate some of the differences observed in their proteomes and metabolomes with enhanced drought resistance.

#### MATERIALS AND METHODS

#### Plant Growth and Stress Treatment

Barley (Hordeum vulgare L.) plants were cultivated under partially controlled greenhouse conditions. Two spring varieties — Maresi and Cam/B1//CI08887/CI05761 (referred further as Cam/B1/CI) — were used for the experiments. Maresi is a German semi-dwarf variety and Cam/B1/CI is a Syrian breeding line; the two varieties were chosen based on the results on their tolerance to reduced water supply reported (Górny, 2001). Seeds of the Syrian genotype were kindly supplied by the International Center for Agricultural Research in the Dry Areas (ICARDA; Syria), and seeds of Maresi originated from the Plant Genetic Resources (Czech Republic).

The plants were grown in pots containing 8 kg of soil (loamy sand mixed with sand at a weight ratio of 7:2). Water properties of soil were characterized by water retention curve. Soil moisture at 2.2 pF (15.8 kPa) for control and 3.2 pF (158.5 kPa) for drought was established. The amount of added fertilizer was established on the basis of soil-tests. Each variety was grown in both control and drought conditions (16 pots in total). In each pot 10 plants were cultivated and when harvested treated as one pooled biological repetition. Half of the plant material was used for metabolomic and proteomic studies and the other half for phenotypic traits analysis. This experiment was repeated twice.

The drought began at the three-leaf stage (phase 13 in the BBCH scale), which was achieved 16 days after sowing and lasted 10 days. The soil moisture was controlled daily using an FOM/mts TDR soil moisture meter according to the reflectometry method (EasyTest, Institute of Agrophysics PAS, Poland) and adjusted by adding the appropriate quantity of water.

The plant material (leaves and roots) for proteomic and metabolomic analysis was collected directly after the 10-days

drought period. The harvested samples were immediately frozen in liquid nitrogen. The tissue samples were ground in liquid nitrogen in precooled adaptors for 45 s at 30 Hz frequency using a ball mill MM400 (Retsch, Germany). Pulverized tissue was stored at −80◦C until further analyses.

Plants used for the phenotypic traits analysis were grown after the drought period maturity under control conditions. Traits related to plant architecture, growth and productivity were measured on the basis of four replicates.

#### Protein Extraction

The total protein from leaves and roots was extracted using the procedure described by Hurkman and Tanaka (1986) with certain modifications. In brief, homogenized tissue was weighed (300 and 600 mg of leaf and root samples, respectively) and suspended in 750 µl of cold lysis buffer [500 mM Tris-HCl pH 8.0; 700 mM saccharose; 100 mM KCl; 50 mM EDTA; 2 mM phenylmethylsulfonyl fluoride; and 2% (v/v) β-mercaptoethanol] and incubated for 10 min on ice. Subsequently, an equal volume of water-saturated phenol (750 µl) was added to each sample. The solutions were incubated on a shaker for 10 min at room temperature. The aqueous and organic phases were separated by centrifugation for 10 min at 11, 000 g at 4◦C. The phenolic phase was recovered and re-extracted with an equal volume of extraction buffer. 5 volume of 0.1 M ammonium acetate in methanol was added to phenolic phase and the proteins were precipitated overnight in −20◦C. The precipitated proteins were collected by centrifugation for 5 min at 18, 000 g and 4◦C and then washed three times with 0.1 M ammonium acetate in methanol. One percent (w/v) polyvinylpyrrolidone (PVP) was added to the first wash to facilitate phenolic contamination removal. Finally, the precipitated proteins were washed with 80% acetone. The samples were centrifuged for 2 min at 18, 000 g and 4 ◦C after each wash step and air dried.

#### Metabolite Extraction

50 mg of frozen barley samples (leaves or roots) was suspended in 1.4 ml of 80% methanol. The ribitol was used as an internal standard (25 and 10 µl of 1.0 mg/ml solution for leaves and roots, respectively). The mixtures were shaken vigorously for 10 min at room temperature in a thermomixer (TS-100, Biosan, Latvia) at 950 rpm. The suspensions were centrifuged at 11, 000 g in RT. For leaf and root samples respectively, 200 and 500 µl aliquots of the supernatant was evaporated in RT using vacuum concentrator (Eppendorf, Germany). The dried extract samples were redissolved in 50 µl of 20 mg/ml methoxyamine hydrochloride solution in pyridine, and the derivatization reaction was performed for 1.5 h followed by a 30 min reaction with 80 µl of N-methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA). Both reactions were performed at 37◦C. A mixture of C10–C36 alkanes in hexane was used as the retention index marker.

#### 2D Gel Electrophoresis of the Proteins

Isoelectric focusing (IEF, first dimension) was performed by dissolving the total protein extracts in the rehydration buffer [7 M urea, 2 M thiourea, and 2% (w/v) CHAPS]. 2D Quant Kit (GE Healthcare, USA) was used to quantify the protein concentration. 300 µg of the protein sample was loaded onto an 11 cm Immobiline DryStrip with a linear gradient of pH 4–7 (GE Healthcare, USA). Isoelectric focusing (IEF) was performed using an ETTAN IPGphor 3 System (GE Healthcare, USA) according to the manual. The IPG strips were equilibrated twice for 15 min in 7.5 ml of equilibration buffer. The first equilibration solution contained 1.5 M Tris-HCl, pH 8.8, 6 M urea, 30% (v/v) glycerol, 2% (w/v) SDS, 0.002% (w/v) bromophenol blue, and 1% (w/v) dithiothreitol. The second equilibration solution was modified by replacing the dithiothreitol with 2.5% (w/v) iodoacetamide. In the second dimension proteins were separated in 12% SDS-polyacrylamide gels on Ettan DALTtwelve System (GE Healthcare, USA). SDS-PAGE was performed using the following conditions: the temperature was maintained at 19◦C, and the wattage was 1.25 W per gel for 30 min and 7.5 W per gel for 2 h. Two technical replicates were conducted for each biological replicate; therefore, eight gels for the leaf samples and eight gels for the root samples were analyzed for each cultivar, giving in total 64 gels.

### Gel Image and Statistical Analysis

The gels were stained with Colloidal Coomassie Brilliant Blue (Neuhoff et al., 1985). ImageScanner III (GE Healthcare, USA) was used to scan the gels. Image Master 2D Platinum 7.0 software (GE Healthcare, USA) was used to quantitatively analyze the spots. The protein spots across the gels were matched automatically, however, manual edition was necessary to improve the analysis. Protein spots relative volume (% vol) was quantified. This parameter is relatively independent of variation due to protein loading and staining. Statistical significance of the relative change of accumulation of protein spots was determined using Student's t- test.

#### In-Gel Digestion and Protein Identification by Mass Spectrometry

Protein spots were excised from the gels and digested with trypsin following the protocol described by Shevchenko et al. (1996) with certain modifications. Each excised gel piece was rinsed for 15 min with 100 µl 50 mM ammonium bicarbonate/acetonitrile and for 15 min with 100 µl 50 mM ammonium bicarbonate. Subsequently, the gel pieces were dehydrated with 100 µl acetonitrile and dried under vacuum centrifugation. A trypsin solution (Promega, USA) was added to the dry gel pieces, and the samples were incubated overnight at 37◦C. Next, 1 µl of acetonitrile was added to the gel pieces, and the samples were subjected to sonication for 5 min and centrifugation (10, 000 g, 1 min). Subsequently, the isolated peptides were analyzed using MALDI-TOF or MALDI-TOF/TOF mass spectrometers model Autoflex and UltrafleXtreme (Bruker Daltonics, Germany) respectively. The matrix used was α-cyano-4-hydroxycinnamic acid dissolved in 50% acetonitrile with addition of 0.1% of trifluoroacetic acid. Mass spectra were registered in reflectron positive ion mode. The peptide mass fingerprinting (PMF) data were submitted to databases using the MASCOT in-house server. For cases in which the PMF analysis did not yield suitable results,

an MS/MS analysis was performed on five peaks selected from the PMF spectrum. Database queries were restricted to Green Plants (Viridiplantae) and monoisotopic peptide masses were search. The criteria applied to accept the results were based on the molecular weight search score (MOWSE), percent sequence coverage, and matched peptide numbers.

### Metabolite Profiling Using Gas Chromatography/Mass Spectrometry

Qualitative and quantitative analyses were performed using a 6890 N gas chromatograph (Agilent, USA) and a GCT Premier mass spectrometer (Waters, USA) with Waters MassLynx software version 4.1. Gas chromatography was performed using a DB-5MS capillary column (30 m × 0.25 mm with a 0.25 µm film thickness; J&W Scientific, USA). The injection temperature was set to 230◦C, the MS transfer line was at 230◦C, and the ion source was adjusted to 250◦C. Pure helium was used as the carrier gas at a constant flow of 1 ml/min. The oven temperature was maintained at 70◦C for 2 min, then ramped at 10◦C/min to 300◦C, and finally maintained at 300◦C for 10 min. Mass spectra were recorded in the m/z range 50–650 with electron ionization (70 eV) in the positive ion mode. Each biological replicate was analyzed using four technical repetitions. The obtained mass spectra were analyzed by the TargetSearch software package (Cuadros-Inostroza et al., 2009), using the Golm Metabolome Database (GMD) as the reference library (Kopka et al., 2005).

Two peaks that originated from the oligosaccharides raffinose and 1-kestose were not fully resolved on the TIC due to the co-elution on the GC column and high similarity of mass spectra, which was also found during previous investigations (Bowne et al., 2012). This lack of resolution disabled automated assignment and alignment with TargetSearch. For this reason, manual integration of raffinose and 1-kestose peaks was performed.

#### Statistical Data Analysis

The metabolomic and phenotypic data were submitted to an analysis of variance in Genstat 16 (VSN International, Ltd<sup>2</sup> ) to identify significant mean differences among the varieties, between control and drought conditions, and to identify significant interaction of varieties with different conditions.

### RESULTS

#### Drought Influence on the Phenotypic Traits of Analyzed Barley Genotypes

Studied genotypes clearly differed in response to the applied 10 days drought period in terms of maintaining turgor pressure (**Figure 1**). At this time point relative water content (RWC) was comparable in control plants of both lines tested barley lines (Maresi: 97 ± 2%; Cam/B1/CI: 94 ± 1%). However, the RWC measured in plants exposed to drought conditions indicated that Cam/B1/CI is less prone to water loss (Maresi: 75 ± 2%; Cam/B1/CI: 87 ± 3%). The results of phenotypic measurements performed for mature plants revealed several differences in drought impact on the performance of both tested barley lines. Significant effects (P < 0.05) for the drought effect were observed for eight traits in Maresi and for four traits in Cam/B1/CI, with one trait in common (**Table 1**). The relative reduction of yield in drought was similar for both varieties (about 30%), but the absolute loss for Maresi (P < 0.05) was about two times bigger

<sup>2</sup>www.vsni.co.uk


#### TABLE 1 | Drought induced changes in phenotypic traits observed in mature barley plants.

Asterisks indicate a statistically significant difference between varieties (P < 0.05). Arrows indicate the direction of statistically significant drought-induced changes of the trait (P < 0.05, n = 4).

than the loss for Cam/B1/CI. The two varieties attained the yield reduction in a different way. For Maresi, the lower yielding was due to a lower grain weight and number of grains on main spike, length of spikes and number of productive tillers. Cam/B1/CI reacted to drought by significantly reducing its 1000-grain weight and number of spikelets in lateral spikes. The protein content in seeds was significantly increased in drought only in Maresi. Overall, our results suggest that Cam/B1/CI is better adapted to drought conditions than Maresi.

#### Drought-Induced Changes in the Barley Leaf and Root Proteomes and Metabolomes

2D gel electrophoretic separation enabled reproducible detection of over 1000 protein spots (**Figure 2**; Supplementary Images S1–S4). The leaf proteome reaction to drought clearly differentiated tested genotypes. Prevalent number of all 81 drought-affected Maresi leaf proteins decreased their accumulation when compared to control (**Figure 3A**). In Cam/B1/CI leaves, only 40 proteins were influenced by drought conditions, but opposite to Maresi bigger fraction of them enhanced their accumulation (**Figure 3A**). In contrast to the leaf tissue, greater number (104) of root drought-responsive proteins was observed in Cam/B1/CI compared to 78 identified in Maresi. Also, a clearly higher number of proteins responded to stress by reducing their accumulation in Cam/B1/CI roots as compared to Maresi. We were able to identify ∼70% of leaf and ∼60% of root drought-responsive proteins by mass spectrometry and grouped them based on their function in biological processes (**Figure 3A**). The highest number of identified leaf and root proteins represented defense mechanisms and carbon metabolism respectively. In Cam/B1/CI leaves, opposite to Maresi, we did not find any proteins belonging to carbon metabolism category with reduced concentrations (**Figure 3A**). Accumulation levels of all proteins linked with gene expression identified in Cam/B1/CI were reduced in drought conditions, while in Maresi some representatives of this functional category increased their amounts in leaves and roots. Drought affected nitrogen metabolism in leaves, but not in roots of both tested barley lines. Finally, all identified enzymes linked with secondary metabolites reduced their accumulation in plants exposed to drought.

GC/MS analyses of leaf and root extracts enabled detection of 86 and 85 drought-affected metabolites respectively (**Figure 3B**, Supplementary Table S1). Primary metabolites constituted the main group among these compounds. Due to drought treatment significant (P < 0.01) changes in accumulation were observed for 23 leaf metabolites and 13 identified in roots in at least one variety. Mean drought effects were more frequent than interaction effects both in leaves (23/8) and roots (13/2). The metabolomic reaction to drought varied between the two studied genotypes with higher number of changes in metabolite accumulation observed in Maresi. There was also a clear difference between the total numbers of up-regulated, but not down-regulated, metabolites between tested genotypes (**Figure 3B**). 18 leaf and 21 root metabolites that were found to be up-regulated in Maresi did not change their accumulation levels in the same organs of Cam/B1/CI. Identified metabolites were grouped into respective metabolite classes. Among those the highest number of drought-responsive compounds represented amino acids (**Figure 3B**). Particularly response of Maresi roots to drought was linked with increased accumulation of 13 compounds representing this group. Interestingly, we did not find any amino acids whose accumulation was significantly reduced in roots of both tested barley lines.

#### Defense Related Proteins

Our results showed that accumulation of heat shock proteins (HSPs) was strongly affected by drought. Proteins belonging to

family of small HSP (sHSP) increased their accumulations in roots of both tested barley genotypes (**Figure 4**; Supplementary Tables S2 and S3). Many of the identified HSP70 isoforms diminished their abundance in both varieties; however, constitutive accumulation levels of these proteins were higher in Cam/B1/CI than in Maresi (**Figures 4** and **5**; Supplementary Tables S2–S5). Members of HSP100 family, caseino-lytic protease (ClpP) and its associated chaperone isoforms (ClpC), decreased their accumulation in Maresi leaves during the response to water deficit. Opposite, in Cam/B1/CI leaves ClpP accumulated to higher levels with drought (**Figure 5**). In addition, an increase in the abundance of cold-regulated protein (COR), which is typically associated with tolerance to low temperatures (Hajela et al., 1990) was observed in roots of both varieties (**Figure 4**).

#### Photosynthesis and Glycolysis

Our proteomic analysis revealed drought induced changes in concentration of proteins constituting the photosystem I (PSI) and PSII reaction centers and energy transfer. Two isoforms of the oxygen-evolving enhancer protein decreased in both cultivars, while accumulation of yet another isoform increased in Maresi (**Figure 5**). The levels of ferredoxin-NADP+ oxidoreductase were diminished in Maresi, whereas cytochrome b6-f exhibited increased accumulation upon drought treatment (**Figure 5**).

We observed drought-induced decrease in the levels of ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) large subunit in Maresi, whereas in Cam/B1/CI such change was not detected. In both tested genotypes RuBisCO activase B exhibited strong drought-induced accumulation, whereas the level of isoform A was decreased in Maresi. At the same time accumulation of RuBisCO large subunit binding protein declined in both cultivars, especially in Maresi (**Figure 5**).

An isoform of fructose-bisphosphate aldolase that could be involved in Calvin cycle and glycolysis decreased its abundance in drought-stressed Maresi leaves. At the same time the level of another isoform of this enzyme increased in both varieties (**Figure 5**). Accumulation of enolase, which is a penultimate enzyme in glycolysis was reduced in Maresi, but increased in Cam/B1/CI leaves. Similarly, the abundance of chloroplastic glyceraldehyde 3-phosphate dehydrogenase declined in Maresi, while another isoform increased its concentration in Cam/B1/CI leaves (**Figure 5**). Decrease in fructokinase accumulation level was observed in root tissue of both tested varieties; however, in Cam/B1/CI one isoform of this enzyme accumulated to higher extent (**Figure 4**). Observed changes suggest that glycolysis is compromised in Maresi, but maintained or even elevated in Cam/B1/CI during the response to drought.

Mitochondrial pyruvate dehydrogenase E1 subunit alpha (PDHA1) and NADP dependent malic enzyme (NADP-ME), which are involved in redox reactions in glycolysis and Krebs cycle (Tovar-Méndez et al., 2003; Wheeler, 2005), increased their abundances in roots of Maresi and both cultivars, respectively. Observed changes in PDHA1 and NADP-ME accumulation may suggest enhanced activity of Krebs cycle under drought conditions. In accordance with this suggestion, we observed elevated levels of α-ketoglutaric acid in drought stressed leaves and roots of both tested barley lines (**Figure 6**). Contrary, accumulation of succinic acid decreased with drought. Fumaric and citric acids differentiated roots of tested barley lines exhibiting clear up-regulation in Maresi. From other carboxylic acids the most striking drought-induced changes were observed for maleic acid, which increased its accumulation in roots of both tested lines, but only in Maresi leaves. Overall, Maresi revealed higher increases in carboxylic acid accumulation than Cam/B1/CI (**Figure 6**).

### Reactive Oxygen Species

plants.

Proteins implicated in ROS generation: hydroxy acid oxidase (HAO, also known as glycolate oxidase) and oxalate oxidase (OXO) exhibited increased accumulation in roots of Maresi and Cam/B1/CI, respectively (**Figure 4**). On the other hand, the putative drought-induced increase in ROS generation was compensated with changes in accumulation of several proteins that may aid plants in circumventing oxidative damage, including enzymes associated with the ascorbate–glutathione cycle. The increased level of ascorbate peroxidase (APX) was observed in leaves of both tested barley lines, but glutathione reductase, which catalyzes the last step in the cycle was upregulated only in Cam/B1/CI leaves (**Figure 5**). In addition, the accumulation levels of glutathione S-transferase (GST), which is involved in glutathione mediated ROS scavenging (Noctor et al., 2012), and 2-cystein peroxiredoxin (2-CP) increased in both lines in response to water deficit (**Figure 5**). Overall, the majority of identified leaf proteins involved in ROS scavenging accumulated to higher extent in both genotypes under water deficiency conditions. In root tissue, the abundance of glyoxalase I/lactoylglutathione lyase was increased in both lines (**Figure 4**). Interestingly, opposite to leaves, other proteins involved in ROS


scavenging did not change or even reduced their accumulation levels in roots of both tested lines subjected to drought. This concerns GSTs, APXs, and superoxide dismutases.

A few non-enzymatic components of ROS scavenging system were also identified. Ascorbic acid, which is a principal ROS scavenger in plant cells (Asada, 1992), was detected only in leaves, and its concentration significantly increased after drought treatment in Maresi (**Figure 6**). Similarly, accumulation of myoinositol, which belongs to another group of ROS scavengers

FIGURE 6 | Drought-induced changes in the accumulation levels of metabolites identified in both tested barley lines. Quantitative results and statistical significance are shown in Supplementary Table S1.

(Smirnoff and Cumbes, 1989), was found to be up-regulated exclusively in Maresi leaves. At the same time accumulation of fructose-6-phosphate, which also possess antioxidant properties (Spasojevic et al., 2009 ´ ), decreased in roots of this barley line (**Figure 6**). Also some of the identified sugars were proposed to contribute to ROS scavenging during plant response to drought (Foyer and Shigeoka, 2011). This concerns for instance galactinol (Nishizawa et al., 2008), which revealed the most striking increase in concentration among all detected sugars in roots of both tested genotypes, but only in Maresi leaves (**Figure 6**).

#### Osmolytes

Sugars may also play a role in osmotic adjustment. Our analysis revealed that accumulation of several carbohydrates was affected in tissues of both genotypes subjected to drought. Sucrose was the most abundant metabolite identified in studied samples. Despite its high constitutive concentrations, accumulation of this metabolite significantly increased in roots of drought-stressed Maresi plants. Also fructose, ribose and 1-kestose increased their concentrations stronger in Maresi than in Cam/B1/CI roots (**Figure 6**).

Glycine betaine (GB) and proline are two main osmolytes that accumulate in response to various abiotic stresses (Ashraf and Foolad, 2007). The accumulation of betaine aldehyde dehydrogenase (BADH), crucial enzyme involved in GB synthesis, was elevated in roots of both cultivars, but only in Cam/B1/CI leaves (**Figure 5**; Supplementary Tables S2 and S3). However, we did not detect GB during our GC/MS analysis.

The key enzyme involved in proline synthesis, delta-1-pyrroline-5-carboxylate synthetase (P5CS) did not exhibit significant changes in its accumulation pattern in leaves of Cam/B1/CI; however, in Maresi the level of this enzyme was clearly reduced (**Figure 5**). The abundance of an enzyme involved in catabolism of proline, pyrroline-5-carboxylate dehydrogenase, was significantly increased in Cam/B1/CI roots (Supplementary Table S3). Opposite to the changes in P5CS accumulation, metabolomic analyses showed increased proline levels in leaves of both tested cultivars subjected to drought. In Maresi we observed more than 20-fold increase in accumulation of this amino acid, but only 1.8-fold change was registered for Cam/B1/CI leaves (**Figure 6**). Similar pattern was observed in roots; proline levels increased more than 70-fold in Maresi and 15-fold in Cam/B1/CI (Supplementary Table S1). However, this higher fold of change in the concentration of proline in German cultivar was mainly due to clearly lower constitutive content of this amino acid in comparison to Cam/B1/CI plants grown in optimal conditions.

#### Amino Acids and Nitrogen Metabolism

Apart of proline, several other amino acids were clearly induced in Maresi, but not in Cam/B1/CI roots. Opposite to roots, in leaves the response of both lines was more similar with predominant reduction in the amino acid cocnentrations. Changes in accumulation of enzymes involved in amino acid biosynthesis also discriminated tested genotypes. In Maresi leaves their accumulation was reduced, whereas in Cam/B1/CI it remained mainly unchanged. Additionally, two enzymes involved in nitrogen flow that use glutamate as a substrate, glutamate dehydrogenase and glutamine synthetase, exhibited increased abundances exclusively in Cam/B1/CI leaves (**Figure 5**). Nitrogen-containing compounds, putrescine and hydroxylamine, significantly decreased their levels in leaves of both genotypes subjected to drought (**Figure 6**). In contrast, the level of allantoin, which is a nitrogen-transporting compound (Watanabe et al., 2014), significantly increased, but only in Maresi leaves. Interestingly, this particular metabolite has been reported as an important player in the abscisic acid-mediated abiotic stress tolerance (Watanabe et al., 2014).

#### Secondary Metabolism

We observed a clear decrease in the levels of the phenylalanine ammonia-lyase (PAL), which is an important regulatory point between primary and secondary metabolism (Vogt, 2010), in Cam/B1/CI roots and leaves (**Figures 4** and **5**). However, accumulation of PAL substrate phenylalanine in the Syrian genotype did not reveal statistically significant changes in response to drought (**Figure 6**). During our GC/MS analysis we identified only a limited number of phenylpropanoid-type products including p-coumaric acid in roots as well as ferulic and chlorogenic acid in leaves. However, only the chlorogenic acid content decreased in both analyzed genotypes (**Figure 6**).

#### DISCUSSION

### Differential Adaptation to Drought of Both Tested Barley Lines

Our phenotypic analysis revealed a significant drought-related reduction of plant yield for Maresi, but not for Cam/B1/CI (**Table 1**). Other phenotypic traits and RWC were also more affected by drought in Maresi. The worse adaptation of this cultivar to water deficiency could be additionally inferred from the observed negative effects of drought on proteins involved in carbon assimilation and catabolism. For instance, we found a significant reduction in the abundance of RuBisCO large subunit and RuBisCO large subunit binding protein in Maresi, but not Cam/B1/CI leaves (**Figure 5**). Similarly, the recent proteomic studies revealed decrease in the accumulation levels of RuBisCO large subunit, RuBisCO activases and RuBisCO large subunit binding protein in drought-sensitive barley genotypes (Ashoub et al., 2013; Kausar et al., 2013). In addition to RuBisCO, also enzymes involved in nitrogen assimilation and amino acid biosynthesis were more negatively affected by drought in Maresi than in Cam/B1/CI. The level of leaf nitrogen plays an important role in photosynthetic acclimation of the plant to environmental variables (Ainsworth et al., 2003). It was shown in several studies that leaf nitrogen decreases gradually as the drought progresses and this can be related to photosynthetic apparatus damage (Sinclair et al., 2000; Xu and Zhou, 2006).

The enhanced susceptibility of Maresi plants to water deficit seems to correspond with proteomic and metabolomic responses in leaves of this genotype, where higher number of proteins and metabolites was influenced by drought as compared with Cam/B1/CI. However, in roots, the situation

was opposite – more proteins and metabolites changed their accumulation pattern in Cam/B1/CI. It is possible that the rapid and pronounced induction of defensive mechanisms in roots contribute to the enhanced tolerance of this line to water deficit. Conversely, some of the changes in the protein and metabolite accumulation observed in Maresi leaves may be linked with negative consequences of drought rather than with droughtprotective mechanisms.

### Molecular Chaperons May Help in Drought Tolerance

Many of the proteins identified during this study are known as general abiotic stress indicators. These include several of HSPs, which constitute a protein family ubiquitous among prokaryotes and eukaryotes. For instance, we found constitutive accumulation levels of HSP70s higher in Cam/B1/CI than in Maresi (Supplementary Tables S2–S5). Similarly, Wendelboe-Nelson and Morris (2012) reported higher constitutive HSP70 levels in the drought-tolerant compared with the susceptible barley genotype. In addition, accumulation of one of the HSP70 leaf isoforms increased with drought in Cam/B1/CI, but decreased in Maresi (**Figure 5**). Kausar et al. (2013) also reported increase of HSP70 accumulation in drought-tolerant barley genotype, whereas in the drought-sensitive genotype the decline was observed. This indicates that in barley constitutive levels of HSP70 as well as changes in the accumulation of this chaperone could correlate with drought resistance. Similarly, to the mentioned above HSP70 isoform, we observed drought induced decrease in ClpP accumulation in Maresi leaves, while at the same time levels of this protein increased in Cam/B1/CI (**Figure 5**). Rosano et al. (2011) revealed a molecular chaperone function of Clp/HSP100 from Arabidopsis thaliana and showed its contribution to protein import into the chloroplast. In this species, the regulatory domain of chloroplast targeted Clp complex, encoded by the nuclear gene Early Responsive to Dehydration 1, is induced by water stress in ABA-independent manner (Nakashima et al., 1997). Up-regulation of Clp proteins observed in wheat seedlings subjected to water deficiency was proposed to be linked with drought-induced changes in photosynthesis and RuBisCO content (Demirevska et al., 2008).

It is known that the refolding of non-native proteins by ATPdependent chaperones, including HSP70s is facilitated by the activity of sHSPs (Lee and Vierling, 2000). We found that the sHSP 16.9 isoforms increased their accumulation in roots of both tested lines during the response to drought (**Figure 4**). Similarly, Guo et al. (2009) reported that a gene encoding the sHSP family member was induced under drought stress in tolerant and sensitive barley genotypes. Overall, these findings suggest that changes in the accumulation levels of HSP70 and Clp/HSP100 may contribute to the enhanced drought tolerance of the Cam/B1/CI line. Opposite to these two HSP subclasses, members of the sHSP subfamily respond to water deficit, but are rather not among the traits that determine the difference in drought resistance observed between the two tested barley lines.

### ROS Production and Scavenging Contributes to the Difference in Drought Resistance between Cam/B1/CI and Maresi

Our proteomic analysis revealed drought induced changes in HAO and OXO accumulation (**Figure 4**). These two enzymes are involved in H2O<sup>2</sup> production in peroxisomes and apoplast respectively (Mittler et al., 2004). Similar to our results, Kausar et al. (2013) reported increased accumulation of OXO under drought stress only in a drought-tolerant barley genotype indicating that enhanced production of H2O<sup>2</sup> is not necessarily disadvantageous for plant fitness during the response to drought. Plants possess a number of mechanisms facilitating efficient neutralization of free radicals, including water/water cycle in chloroplasts, ascorbate–glutathione cycle and reactions related to the activity of peroxidases and catalases (Mittler, 2002). The observed increase in APX accumulation in leaves of Maresi and even stronger in Cam/B1/CI combined with elevated levels of glutathione reductase in Cam/B1/CI indicates increased activity of ascorbate–glutathione cycle pathway during drought, particularly in the more resistant Syrian genotype. Similarly to our findings, Wendelboe-Nelson and Morris (2012) also reported more pronounced increase in APX accumulation in the drought-tolerant barley line. Opposite to the changes in APX levels, ascorbic acid accumulation increased in Maresi, but not in the Cam/B1/CI (**Figure 6**). However, leaves of the Syrian line accumulated constitutively higher amounts of ascorbic acid as compared with Maresi (Supplementary Image S5), suggesting that elevated levels of this metabolite in naïve plants could be linked with adaptation to dry climate. Other enzymes linked with the ascorbate–glutathione cycle, which increased their accumulation during drought included a GST and a glyoxalase (**Figure 4**). However, from those two only GST was up-regulated in Cam/B1/CI, but not Maresi (**Figure 4**). The GST activity can be mediated by H2O<sup>2</sup> and overexpression of a GST from Pyrus pyrifolia in tobacco increased substantially available amount of enzymatic and non-enzymatic components of ascorbate–glutathione cycle suggesting positive contribution of GSTs to abiotic stress tolerance (Liu et al., 2013). Similarly to our results other studies on barley indicated that GST accumulation during the response to drought only in drought tolerant, but not in sensitive genotypes (Guo et al., 2009; Kausar et al., 2013).

Apart from glutathione and ascorbic acid also other small molecules may contribute to ROS detoxification. For instance, Spasojevic et al. (2009) ´ showed that fructose-phosphates have ROS scavenging properties that are more pronounced from those of fructose or glucose. This suggested that fructokinases may play a role ROS scavenging under drought stress (Fulda et al., 2011). In this context it seems of importance that that the ratio of fructosephosphate to fructose was clearly decreased in drought stressed Maresi, but not Cam/B1/CI roots. These changes correlated with the observation that one out of two identified fructokinases was reduced its accumulation in roots of Maresi, but not of Cam/B1/CI (**Figure 4**).

Overall, our data indicated that water deficit affects several key players of redox balance in barley. Among those we found APX and glutathione reductase as the components whose response to drought differentiated between Maresi and Cam/B1/CI and therefore could be considered to contribute to the enhanced drought tolerance of the Syrian genotype. In addition higher constitutive ascorbic acid accumulation and the maintained during drought fructose-phosphate to fructose ratio could be of significance for the response to water deficit.

### Constitutive Accumulation of Osmoprotective Metabolites Enhances Drought Resistance

Several proteins and metabolites identified in this study are related to osmotic processes, which are severely affected by drought. For instance, elevated proline levels are supposed to enhance cell tolerance toward different kinds of osmotic stresses (Zhu et al., 1998). We observed substantial differences in accumulation of enzymes involved in proline biosynthesis and catabolism (5PCS and P5CDH), as well as clear differences in abundance of this amino acid between Maresi and Cam/B1/CI. Proline accumulation increased significantly during the response to drought in Maresi leaves and roots (**Figure 6**). Interestingly, the synthesis of amino acids with osmoprotective properties (including proline) could be linked with the reallocation of nitrogen derived from RuBisCO degradation (Aranjuelo et al., 2011). This correlated with the elevated decrease in the accumulation of RuBisCO-related proteins that we observed in Maresi, as compared with Cam/B1/CI (**Figure 6**). Opposite to the changes in proline accumulation, at the same time 5PCS levels decreased in leaves of this variety (**Figure 5**). As high proline amounts may trigger cell damage (Hellmann et al., 2000), observed in Maresi drop of P5CS accumulation may indicate a feedback regulation initiated by the huge drought-induced increase in proline concentration. Similar to glutathione, constitutive accumulation level of proline in Cam/B1/CI leaves was already relatively high and increased only moderately upon drought, which correlated with lack of changes in P5CS accumulation (**Figures 5** and **6**; Supplementary Image S5). Interestingly, wheat seedlings with altered tolerance to water deficit did not differ regarding proline accumulation. However, the resistant variety was able to accumulate and metabolize this compound more rapidly (Nayyar and Walia, 2003). Thus, it seems that the key aspect of drought tolerance is not only proline accumulation level, but also appropriate regulation of the biosynthesis and metabolism of this amino acid. In addition to proline, we also observed significant differences in the accumulation levels of potentially osmoprotective disaccharides and trisaccharides in the control and stressed plants. Cam/B1/CI was the genotype with a substantial difference in 1-kestose concentration between control and drought conditions (**Figure 6**). In addition, several monosaccharides accumulated constitutively to higher levels in Cam/B1/CI as compared with Maresi (Supplementary Image S5). It seems possible that the elevated sugar levels contribute to the better osmotic protection in the Syrian line to tolerate osmotic stress.

### A Link between Phenylpropanaoid Pathway and Drought Tolerance

We found PAL among the enzymes whose accumulation was consistently down-regulated in Cam/B1/CI but not in Maresi (**Figures 4** and **5**). Initiated with this enzyme phenylpropanoid pathway is crucial for the biosynthesis of different products which are critical in plant development and plant adaptation to the environment (Dixon and Paiva, 1995). These metabolites include lignin, phenolic acids, and flavonoids. Interestingly, accumulation of caffeoyl-CoA O-methyltransferase, which is involved in the phenolic acid/lignin branch of phenylpropanoid metabolism (Boerjan et al., 2003), also decreased in drought stressed Cam/B1/CI roots (**Figure 4**). A proteomic study on maize suggested that lignin biosynthesis suppression might play a role in β-aminobutyric acid-induced drought resistance (Macarisin et al., 2009). In accordance with this functional correlation, A. thaliana pal1 pal2 double-knockout mutant, which produces only 30% of the wild-type lignin content, is significantly more resistant to drought than the wild-type plants (Rohde et al., 2004; Huang et al., 2010). Cumulatively, these results support a potential link between reduced lignin biosynthesis/deposition and drought tolerance. However, molecular mechanisms of this link remain obscure.

### CONCLUSION

Our results indicate that activity of molecular chaperons as well as production of osmoprotectants and antioxidants may collectively mediate enhanced resistance to water deficit observed in Cam/B1/CI line. Interestingly, it seems that the elevated constitutive accumulation of several molecular components of these defense mechanisms, including HSP70, proline, carbohydrates and ascorbic acid, is indispensable for the proper response to water deficit. Additional droughttriggered increase in accumulation of certain proteins and metabolites including HSP70, Clp, APX, GST, and carbohydrates is also of significance. Moreover, down-regulation of secondary metabolism, particularly lignin biosynthesis, may also contribute to the enhanced drought tolerance observed in Cam/B1/CI. However, it should be noted that we observed several clear changes in protein and metabolite accumulation that were similar in both tested lines. At the currant stage it can be not excluded, that although these changes were not discriminating the response of Cam/B1/CI and Maresi, they may still significantly contribute to drought resistance.

#### AUTHOR CONTRIBUTIONS

The experimental set up was designed by KC, PR, BS, AS, MSu, TA, and MSt. KC, PR, BS, ŁM, DC, AK, KM, PO, and KK performed the experiments and acquired the data. KC, PR, BS, AS, and PK analyzed the data. KC, PR, BS, AS, PK,

PB, and MSt interpreted the results and wrote the manuscript with input from all other authors. MSt supervised the project.

#### FUNDING

This work was supported by the European Regional Development Fund through the Innovative Economy Program for Poland 2007–2013 (project WND-POIG.01.03.01-00-101/08

#### REFERENCES


POLAPGEN-BD), and additionally by the Polish Ministry of Science and Higher Education, under the KNOW program.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016.01108



fructose to scavenge the hydroxyl radical. Carbohydr. Res. 344, 80–84. doi: 10.1016/j.carres.2008.09.025


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Chmielewska, Rodziewicz, Swarcewicz, Sawikowska, Krajewski, Marczak, Ciesiołka, Kuczynska, Mikołajczak, Ogrodowicz, Krystkowiak, Surma, ´ Adamski, Bednarek and Stobiecki. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Contrasting Changes Caused by Drought and Submergence Stresses in Bermudagrass (Cynodon dactylon)

Tiantian Ye1, 2, Haitao Shi <sup>1</sup> , Yanping Wang<sup>1</sup> and Zhulong Chan<sup>1</sup> \*

*<sup>1</sup> Key Laboratory of Plant Germplasm Enhancement and Specialty Agriculture, Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan, China, <sup>2</sup> University of Chinese Academy of Sciences, Beijing, China*

In this study, we investigated the mechanisms by which bermudagrass withstands the drought and submergence stresses through physiological, proteomic and metabolomic approaches. The results showed that significant physiological changes were observed after drought treatment, while only slight changes after submergence treatment, including compatible solute contents, ROS levels and antioxidant enzyme activities. Proteomics results showed that 81 proteins regulated by drought or submergence treatment were identified by MALDI-TOF-MS. Among them, 76 proteins were modulated by drought stress with 46 increased abundance and 30 decreased abundance. Forty-five showed abundance changes after submergence treatment with 10 increased and 35 decreased. Pathway enrichment analysis revealed that pathways of amino acid metabolism and mitochondrial electron transport/ATP synthesis were only enriched by drought treatment, while other pathways including photosynthesis, biodegradation of xenobiotics, oxidative pentose phosphate, glycolysis and redox were commonly over-represented after both drought and submergence treatments. Metabolomic analysis indicated that most of the metabolites were up-regulated by drought stress, while 34 of 40 metabolites contents exhibited down-regulation or no significant changes when exposed to submergence stress, including sugars and sugar alcohols. These data indicated that drought stress extensively promoted photosynthesis and redox metabolisms while submergence stress caused declined metabolisms and dormancy in *Cynodon dactylon*. Taken together, the quiescence strategy with retarded growth might allow bermudagrass to be adaptive to long-term submerged environment, while activation of photosynthesis and redox, and accumulation of compatible solutes and molecular chaperones increased bermudagrass tolerance to drought stress.

Keywords: bermudagrass, drought stress tolerance, sumbergence stress, Proteomic analysis, reactive oxygen species, carbohydrate metabolism

#### Edited by:

*Jian Xu, National University of Singapore, Singapore*

#### Reviewed by:

*Alberto A. Iglesias, Universidad Nacional del Litoral, Argentina Ping Lou, Dartmouth College, USA*

> \*Correspondence: *Zhulong Chan zhulongch@wbgcas.cn*

#### Specialty section:

*This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science*

Received: *16 September 2015* Accepted: *19 October 2015* Published: *10 November 2015*

#### Citation:

*Ye T, Shi H, Wang Y and Chan Z (2015) Contrasting Changes Caused by Drought and Submergence Stresses in Bermudagrass (Cynodon dactylon). Front. Plant Sci. 6:951. doi: 10.3389/fpls.2015.00951*

**Abbreviations:** 2-DE, two-dimensional gel electrophoresis;ABA, abscisic acid; CAT, catalase; DHAR, dehydroascorbate reductase; DW, dry weight; EL, Electrolyte Leakage; FW, fresh weight; GA, gibberellic acid; GR, glutathione reductase; GST, glutathione S-transferase; IEF, isoelectric focus; LWC, leaf water content; MDA, malondialdehyde; PC, plastocyanin; POD, Peroxidase; PQ, plastoquinone; PRX, peroxiredoxin; PSI, photosystem I; PSII, photosystem II; ROS, reactive oxygen species; RuBisCO, Ribulose-1,5-bisphosphate carboxylase/oxygenase; SDS-PAGE, SDS polyacrylamide gel electrophoresis.

### INTRODUCTION

Drought and flooding are major abiotic factors limiting plant growth and development which happened from time to time worldwide. Under drought stress condition, limited water supply greatly decreases leaf water content and causes tissue dehydration which is characterized by extensive changes at physiological, biochemical, molecular, and cellular levels (Ashraf, 2010; Fleury et al., 2010). Drought tolerance is a very complex trait depending on severity of the drought, plant developmental stage as well as the stress duration (Zhu, 2002). Drought stress induces the accumulation of the plant hormone abscisic acid (ABA), which leads to stomatal closure for maintaining water status in plant cells under water-deficit conditions (Ren et al., 2010; Zhao et al., 2013).

Flooding is another form of water stress that results from excess water, which affects about 10% of the global land area. Flooding, including waterlogging and submergence, can negatively affect plant growth and crop production (Setter and Waters, 2003). Waterlogging is defined as the saturation of the soil with water around the roots, while submergence describes the condition in which the whole plant is completely covered by water (Liu and Jiang, 2015). Under submergence environment, gases such as O2, CO2, and ethylene diffuse very slowly in water and the cellular O<sup>2</sup> level decreases and inhibits aerobic respiration (Gibbs and Greenway, 2003; Fukao and Bailey-Serres, 2004). Despite knowledge of adaptive mechanisms to drought, understanding of the mechanisms behind plant response to submergence is very limited. Plants develop different strategies in response to submergence. Recent studies showed that many genes were involved in submergence responses (Gonzali et al., 2005; Xu et al., 2006; Hattori et al., 2009). In rice, flood-tolerant cultivars invoke a quiescence strategy that is controlled by transcription factors SUB1. SUB1A is induced by ethylene under submergence condition and negatively regulates expression of SUB1C, leading to repressed carbohydrate metabolism and retarded cell elongation. Flood-susceptible rice cultivars avoid submergence via activation of SUB1C expression which is promoted by gibberellic acid (GA) and is associated with rapid degradation of carbohydrate reserves and enhanced elongation of leaves and internodes (Bailey-Serres and Voesenek, 2008; Xu et al., 2006).

Grass plants were exposed to either drought or flooding conditions frequently. Several groups reported growth changes of perennial grass under waterlogging condition. The results showed that waterlogging reduced shoot and root dry weight in cool-season grass species including creeping bentgrass (Agrostis stolonifera) (Huang et al., 1998; Jiang and Wang, 2006) and Kentucky bluegrass (Poa pratensis) (Wang and Jiang, 2007). However, waterlogging stimulated plant growth in the tolerant warm-season grass species such as knotgrass (Paspalum paspaloides) and spiny mudgrass (Pseudoraphis spinescens), while inhibited the growth in the intolerant seashore paspalum (Paspalum vaginatum) and centipedegrass (Eremochloa ophiuroides) (Zong et al., 2015). Comparative physiological analysis showed that submergence caused greater damage in perennial ryegrass (Lolium perenne) than waterlogging, increased greater reductions in leaf chlorophyll and total carotenoid concentrations (Liu and Jiang, 2015). The responses of diverse perennial ryegrass accessions to submergence and their recovery following de-submergence were also reported by the same group. The results indicated that large phenotypic variations in leaf color, plant height, and growth rate were observed under submergence condition (Yu et al., 2012).

As one of the most important warm-season turfgrasses, bermudagrass (Cynodon dactylon) exhibited high tolerance to several abiotic stresses including drought and submergence. Recently, we identified bermudagrass varieties that were differing in drought tolerance. Comparative physiological analysis showed that changes of water status, osmolyte accumulation and antioxidant defense system might be contributed to the natural variation of drought tolerance between bermudagrass varieties (Lu et al., 2009; Shi et al., 2012). Net CO<sup>2</sup> assimilation and stomatal conductance to water vapor were inhibited by drought stress (Carmo-Silva et al., 2008a). However, activity of the enzymes involved in the assimilation of CO<sup>2</sup> did not show significant change by drought treatment in three C4 grasses of different subtypes (Carmo-Silva et al., 2008b). Proteomic profiling identified 39 and 54 proteins that were regulated by drought stress in different bermudagrass cultivars, respectively (Zhao et al., 2011; Shi et al., 2014). Exogenous application of small molecules increased drought stress tolerance of C. dactylon. Totally 36 and 76 proteins were induced by polyamine and melatonin, respectively, in C. dactylon based on proteomics approach (Shi et al., 2013, 2015b). Additionally, the macroarray and RNA sequencing analyses identified stressresponsive candidate genes from C. dactylon (Kim et al., 2009; Shi et al., 2015a). Overexpression of a C. dactylon stress-responsive nuclear factor Y gene (Cdt-NF-YC1) in rice resulted in increased tolerance to drought and salt as well as increased sensitivity to ABA (Chen et al., 2015).

As indicated above, responses of C. dactylon to drought condition have been well characterized by several groups. However, limited information is available for the responses of C. dactylon to submergence condition. Field survey data in the water level fluctuation zone of the Three Gorges Reservoir in China demonstrated that most original vegetation disappeared due to winter flooding for up to 6 months, while perennials including C. dactylon could tolerant deep and long-term flooding condition (Ye et al., 2013; Wang et al., 2014). Physiological analysis showed that submergence increased antioxidant enzyme activities, but decreased total soluble carbohydrate and starch contents (Tan et al., 2010). However, the detailed proteomic and metabolomic changes in C. dactylon in response to sumbergence are largely unknown. Moreover, studies to directly compare contrasting responses after drought and submergence in C. dactylon were lacking and the underlying mechanisms remained elusive. Here comparative proteomics and metabolomics approaches were applied to investigate the mechanisms by which bermudagrass withstands the drought and submergence stresses. The results showed that drought stress extensively promoted photosynthesis and redox metabolisms while submergence stress caused declined metabolisms and dormancy in C. dactylon. Therefore, growth of C. dactylon was severely inhibited by drought, but completely by submergence, indicating different strategies resulted in contrasting growth adaption in C. dactylon in response to drought and submergence stresses.

### MATERIALS AND METHODS

#### Plant Materials and Growth Conditions

The bermudagrass seeds Yukon were kindly provided by American Seed Research of Oregon Company. After 3 days of stratification at 4◦C in the dark, the seeds were sown in the flowerpot filled with soil in the greenhouse and were grown under long-day lighting conditions (16 h light/8 h dark), with about 65% relative humidity at 25 ± 2 ◦C and light irradiance of about 150µmol quanta m−<sup>2</sup> s <sup>−</sup><sup>1</sup> per day. The plants were irrigated with nutrient solution twice every week.

#### Experimental Design of Stress Treatments

To compare the differences of bermudagrass responses to drought and submergence, 21-day-old seedlings were subjected to control condition and stress conditions. For drought treatment, water was withheld for 21 d. For submergence treatment, plants were fully submerged in larger plastic containers (60×40 × 27 cm) for 21 d. The survival rate of stressed bermudagrass was recorded at 7 d after re-watering (for drought treatment) or de-submergence (for submergence treatment). The leaf samples were collected at 0, 7, 14, 21 days after control and stress treatments for physiological indexes analyses. The leaf samples at 14 days subjected to control and stress conditions were harvested for proteomic and metabolomic assays based on measured electrolyte leakage data (**Figure S1**). For each independent experiment, every plant sample was extracted from at least 30 bermudagrass plants. All the experiments in this study were repeated three times.

#### Determination of Leaf Water Content (LWC) and Electrolyte Leakage (EL)

For the relative LWC analysis, the leaf samples were harvested from at least 30 independent lines of different treatments at different time points (0, 7, 14, and 21 days). The fresh weight (FW) was weighed immediately after collection, and the dry weight (DW) was quantified after incubation for 16 h at 80◦C, and the LWC (%) was measured as (FW-DW)/FW × 100 (Shi et al., 2012, 2014).

EL was determined from detached leaves, which were collected from at least 30 plants each treatment (about 0.2 g), The detached leaves were placed in 50 ml tubes containing 15 mL deionized water. After gently shake at room temperature for 6 h at 150 rpm, the initial conductivity was determined. The fully releasing conductivity was measured after boiling at 121◦C for 20 min using previous samples. The conductivity was measured using a conductivity meter (Leici-DDS-307A, Shanghai, China). The percentage of electrolyte leakage was determined as the ratio of the initial conductivity to fully releasing conductivity as described previously (Shi et al., 2012, 2014).

#### Quantification of Sucrose and Soluble Total Sugars

The sucrose and soluble total sugars were measured using the method as previously described by Shi et al. (2012). The sucrose content and soluble total sugar content of samples were measured at 480 nm of absorbance and calculated by using the standard curve with known concentration of sucrose and glucose.

### Measurement of Malondialdehyde (MDA) and Proline Contents

The MDA content in control and stressed plant samples was extracted using thiobarbituric acid (TBA) regent and boiled at 100◦C for 20 min as previously described by Yang et al. (2010). After cooling to room temperature and centrifugation at 15,000 g for 10 min, the supernatant was quantified at 450, 532, and 600 nm of absorbance with a spectrometer. The MDA concentration can be estimated through the following formula (µmol l−<sup>1</sup> ) = 6.45(A<sup>532</sup> - A600) – 0.56A450.

Proline content was measured by a spectrometric method using known concentration of L-proline to form standard curve. Briefly, 0.25 g leaf samples were grinded to power and then extracted in 3% (w/v) sulfosalicylic acid for 10 min at 100◦C, then 2 ml ninhydrin reagent and 2 ml glacial acetic acid were added to the 2 ml extraction solution. The mixed solution was boiled at 100◦C for 40 min. After cooling to room temperature, the proline level of sample was measured absorbance at 520 nm and calculated according to the standard curve as described previously (Shi et al., 2012).

### Determination of ROS Accumulation and Antioxidant Enzyme Activities

The protein concentration was quantified using the Bradford method (Bradford, 1976). For H2O<sup>2</sup> content analysis, supernatant of the plant extracts and 0.1% (w/v) titanium sulfate regent [in 20% (v/v) H2SO4] were mixed at 1/1 (v/v) to precipitate the peroxide—titanium complex. The absorbance of solution was quantified at 410 nm. For the O• 2 - content assay, a plant O• 2 - ELISA Kit (Dingguo, Beijing, China) was used. The absorbance was quantified at 405 nm.

The catalase (CAT, EC chsdateIsROCDateFalseIsLunarDateFa lseDay30Month12Year18991.11.1.6), glutathione reductase (GR, EC 1.6.4.2) and peroxidase (POD, EC 1.11.1.7) activities were determined using CAT Assay Kit (Beyotime, Shanghai, China), GR Assay Kit (Beyotime, Shanghai, China) and Plant POD Assay Kit (Nanjing Jiancheng Bioengineering Institute, Nanjing, China), respectively, as described previously (Shi et al., 2012).

### Protein Extraction and 2-DE

Total protein was extracted according to the previously described method with slight modifications (Chan et al., 2007). Briefly, 1 g frozen powder from plant leave were homogenized extensively with 5 ml of pre-cooled homogenization buffer [20 mM Tris-HCl (pH 7.5), 1.05 M sucrose, 10 mM EGTA, 1 mM DTT, 1 mM PMSF and 1% (v/v) Triton X-100] on ice, and centrifuged at 10,000 g for 30 min at 4◦C. The supernatant was then mixed with equal volume of Tris-HCl (pH 7.8) buffered phenol. After centrifugation at 10,000 g for 30 min at 4◦C, the above phenol phase was mixed with five volumes of ice-cold saturated ammonium acetate in methanol overnight at −20◦C. The total proteins were collected through centrifugation was stored at −80◦C or dissolved in the lysis buffer [7 M urea, 2 M mithiourea, 4% (w/v) of 3-[(3-cholamidopropyl)-dimethylammo-nio]-1 propane sulfonate (CHAPS), 65 mM DTT and 0.2% (w/v) of carrier ampholyte (pH3.5–10)]. After dissolving extensively and centrifugation, the protein supernatant was quantified through the Bradford's method (Bradford, 1976).

The 2-DE was performed as described by Shi et al. (2013) with minor modification. Briefly, 1 mg of total proteins was applied onto an immobilized pH gradient (IPG) strip (17 cm, pH 4–7, Bio-Rad, USA) and rehydrated extensively at room temperature overnight. The next day, the rehydrated strips were transferred to isoelectric focus (IEF) in the Protein IEF system (Bio-Rad, USA). The conditions of IEF and SDS-PAGE were the same as described by Shi et al. (2012).

#### Gel Image Analysis and Protein Spot Identification by MALDI-TOF-MS

The 2-D gels were stained in Coomassie brilliant blue R250 staining buffer for 4 h and distained overnight. After scanning with an EPSON PERFECTION V700 PHOTO scanner (Epson), the protein spot images of 2-D gel were analyzed using PDQuest

FIGURE 2 | Osmolytes accumulation of bermudagrass after drought and submergence treatments. Changes of proline content (A), soluble sugars (B), and sucrose content (C) of bermudagrass during control and stressed conditions at indicated days. The results shown are means ± SE (*n* = 4), and the results followed by different letters are significantly different from each other at *P* < 0.05 according to Duncan's method.

2-DE Analysis Software (BIO-RAD, USA). Protein spots with more than 2-fold abundance change were used for trypsin digestion and MALDI-TOF-MS analysis with AXIMA-CFR plus (Shimadzu Biotech, Kyoto, Japan) as reported by Shi et al. (2013). MASCOT software (Mascot Wizard chsdateIsROCDateFalseIs LunarDateFalseDay30Month12Year18991.2.0, Matrix Science Ltd., http://www.matrixscience.com) was used to analyze the MS data. Since bermudagrass is an un-sequenced species, the homologous proteins were blasted against sequenced plant species. In the searching process against NCBInr and Swiss-Port protein sequence databases, peptide masses were assumed to be monoisotopic, and 100 ppm was used as mass accuracy, and one missing cleavage site was the maximum, and modifications were also considered. The minimum score of 43 and the minimum sequence coverage of 6% in MOWSE analysis were used to keep the confidence of the identification results.

#### Quantification of Metabolites

The metabolites extraction and derivatization were performed as described by Lisec et al. (2006) and Sanchez-Villarreal et al.

FIGURE 3 | Proteins changed by drought and submergence. (A) A sketch map to show proteome patterns of bermudagrass in responses to drought and submergence. The protein spots induced at least two folds by drought and submergence were marked with arrows. Proteins were separated in the first dimension on the IPG strip (pH 4–7), and in the second dimension on 12.5% SDS-PAGE. (B) Total number of proteins changed by drought and submergence. (C) Venn diagram showing the number and of proteins that overlapped among three types of drought and submergence. (D) Hierarchical cluster analysis of proteins modulated by drought and submergence treatments. Resulting tree figure was displayed using the software package and Java Treeview. The detailed protein information was listed in Table S1.

(2013). The metabolites were then determined using GC-TOF-MS (Agilent 7890A/5975C, CA, USA) according to the procedure of Lisec et al. (2006). For GC-TOF-MS, 1 mL of derivatizated extract was injected into a DB-5MS capillary (30 m × 0.25 mm × 0.25 mm, Agilent J&W GC Column, USA). The metabolites were identified based on retention time index specific masses, via comparing with reference spectra in mass spectral libraries (NIST 2005, Wiley 7.0). After metabolite identification, quantification of metabolites was performed based on the pre-added ribitol in the process of metabolite extraction that was used as an internal standard.

#### Cluster Analyses

Hierarchical cluster analysis was performed using CLUSTER program (http://bonsai.hgc.jp/~mdehoon/software/cluster/) (de Hoon et al., 2004). The resulting tree figures were displayed using the software package and Java Treeview (http://jtreeview.sourceforge.net/). The pathway graph of carbon metabolism was obtained from KEGG (http:// www.genome.jp/kegg/pathway.html). The proteins with

TABLE 1 | Pathway enrichment analysis of proteins modulated by drought and submergence treatments in bermudagrass.


*<sup>a</sup>NF, normalized frequency of each functional category in genome.*

*Black background means NF* ≥ *10 and P* ≤ *0.05 and gray background means NF* ≥ *2 while P* ≥ *0.05.*

different abundance changes were classified using the Classification SuperViewer Tool (http://bar.utoronto.ca/ ntools/cgi-bin/ntools\_classification\_superviewer.cgi) (Provart and Zhu, 2003) and functional categories of every protein were assigned using MapMan (http://mapman.mpimp-golm. mpg.de/general/ora/ora.html) (Thimm et al., 2004). Normalized frequency (NF) of each functional category was assayed as sample frequency of each category in this experiment/background frequency of each category in genome.

### Statistical Analysis

All the experiments in this study were conducted three times, and the data shown are the means ± SEs, while the mean is the average of three replicates. For each independent experiment, every plant sample was extracted from at least 30 bermudagrass plants. Different letters above the columns in every figure indicate significant differences at P < 0.05 (according to Duncan's method).

### RESULTS

#### Drought Severely while Submergence Completely Inhibited Growth of Bermudagrass

Bermudagrass seedlings under control condition grew well with the shoot length from 2.1 cm at 0 d to 13.5 cm at 21 d after treatment. Drought severely and submergence completely inhibited seedling growth. The shoot length only reached 6.8 cm at 21 d under drought condition, while remained 2.6 cm after submergence treatment, which was only 19% of control seedlings (**Figure 1A**). Relative leaf water content decreased significantly after 14 and 21 d of drought treatment, but no differences were observed under submergence condition (**Figure 1B**). Both drought and submergence treatments significantly increased electrolyte leakage, resulting in increased cell membrane damages. At 21 after treatments, less than 16% seedlings survived under drought condition, while 38.6% seedlings survived under submergence condition. These results indicated that both drought and submergence treatments caused severe cell membrane damages and greatly inhibited bermudagrass growth.

### Contrasting Effect of Drought and Submergence on Compatible Solute Accumulation

Compatible solutes including soluble sugar and proline protect macromolecule structure and at the same time increase the osmotic pressure of the cytoplasm and thereby counteract water loss from cells. Compatible solutes also play key roles during plant redox metabolism (Couee et al., 2006). Under drought condition, proline content increased significantly when compared to the control, but no significant differences were observed in seedlings after submerged (**Figure 2A**). Interestingly, drought stress treatment significantly increased soluble sugar and sucrose contents in bermudagrass, while submergence caused declined accumulation of soluble sugar and sucrose (**Figures 2B,C**). These results showed that bermudagrass might

develop contrasting strategies to accumulate compatible solute under drought and submergence conditions.

#### Protein Level Changes after Drought and Submergence Treatments

To identify proteins simultaneously involved in drought and submergence stress responses in bermudagrass, proteomic analyses based on 2-DE were performed using 14 d stressed samples which showed about 50% EL (**Figure 1C**). Through proteomics approach, totally 81 proteins regulated by drought or submergence treatment were identified by MALDI-TOF-MS (**Figure 3A**). Among them, 76 proteins were regulated by drought stress with 46 increased abundance and 30 decreased abundance. Forty-five showed abundance changes after submergence treatment with 10 increased and 35 decreased (**Figure 3B**). The MS results were matched against NCBInr and Swiss-Port protein sequence databases using MASCOT software, and the best matched protein with high confidence score was selected as the final result of each protein spot (**Table S1**). Although, Viridiplantae (Green Plants) was chosen as taxonomy during Mascot database search, most putatively identified proteins were matched to those in Poaceae like Oryza sativa, Triticum urartu, Zea mays, and Setaria italica, which are very close to bermudagrass based on gene sequence alignment analysis.

Overlapping and cluster analyses showed that 9 and 15 proteins were commonly up- and down-regulated by both treatments, respectively (**Figures 3C,D**). Abundance of 52 and 21 proteins was specifically modulated by drought and submergence stress treatments, respectively (**Figures 3C,D**). Moreover, we previously identified 27 proteins which showed increased abundances in Yukon after drought treatment. Among them, at least 8 proteins were also significantly up-regulated by drought in this study, including Chitinase, SOD, and heat shock proteins (**Table S3**).

#### Photosynthesis and Redox Related Pathways were Enriched after Drought and Submergence Treatment

Pathway enrichment analysis was then performed. Because of limited reference genome information for bermudagrass, the homologous proteins were blasted against sequenced plant species and functional categories were also assigned using MapMan. The information of homologous protein and functional category of each protein was shown in **Table 1** and **Table S1**. The MapMan pathway enrichment analysis revealed that pathways of amino acid metabolism and mitochondrial electron transport/ATP synthesis were only enriched by drought treatment (**Table 1**), however, several other pathways including photosynthesis, biodegradation of


TABLE

2


Ye et al. Responses of bermudagrass to drought and submergence stresses

TABLE

2


Continued

xenobiotics, oxidative pentose phosphate, glycolysis, and redox were commonly over-represented after both drought and submergence treatments. Further, analysis showed that 14 proteins changed by drought and submergence were involved in carbon fixation in photosynthetic organisms (**Figure 4**). These results indicated that drought and submergence stresses commonly affected photosynthesis and redox related pathways in bermudagrass.

### Redox and ROS Metabolism Related Proteins were Extensively Changed after Drought and Submergence Treatments

Since pathways related to redox were largely enriched after drought and submergence treatments, we then checked detailed fold changes of proteins involved in redox and ROS pathways. The results showed that 31 proteins playing key roles during photosynthesis, including RuBisCO activase, Cytochrome b6-f complex, and oxygen-evolving enhancer (**Table 2**), were mainly induced by drought, but inhibited by submergence treatment. Several redox metabolism related proteins, like peroxidase, and superoxide dismutase showed increased intensities by drought, but decreased intensities by submergence. Dehydrogenase was commonly inhibited by both drought and submergence (**Table 2**). Chaperonin and heat shock proteins were induced by drought and inhibited by submergence (**Table 2**). These results showed that redox and ROS related proteins were extensively changed under drought and submergence conditions.

### Modulation of ROS Metabolism in Bermudagrass after Drought and Submergence Treatments

To further investigate ROS homeostasis caused by drought and submergence stresses, the detailed content changes of reactive oxygen species were determined. After drought treatment, both H2O<sup>2</sup> and O<sup>−</sup> 2 contents increased after 14 d stress treatment. However, under submergence condition, H2O<sup>2</sup> content decreased and O<sup>−</sup> 2 content showed no significant changes (**Figures 5A,B**). MDA is one of the most frequently used indicators of lipid peroxidation, and MDA content reflects the degree of membrane lipid peroxidation. Drought treatment significantly increased MDA content while submergence slightly increased MDA content (**Figure 5C**). Antioxidant enzymes activities, including CAT, GR, and POD, were then analyzed to reveal changes of enzymatic defense systems. Both drought and submergence treatments increased CAT, GR, and POD activities (**Figures 5D–F**). These results indicated that drought and submergence treatments modulated antioxidant enzyme activities and caused contrasting ROS content changes in bermudagrass.

### Modulation of Metabolites in Bermudagrass after Drought and Submergence Treatments

Since several proteins involved in carbon fixation were changed after stress treatments (**Figure 4**), primary metabolite contents were then determined through chromatography time-of-flight mass spectrometry (GC-TOF-MS). In total, 40 metabolites were measured, including 15 amino acids, 14 sugars, 5 organic acid, 2 sugar alcohols, 2 fatty acid and 2 others (**Figures 6A**, **4B**; **Table S2**). After drought and submergence treatments, contents of most amino acid increased, including theronine, serine, and proline. However, contents of most sugars, organic acid, sugar alcohols, and fatty acid increased by drought, but decreased by submergence. Among 40 metabolites, 22 metabolites involved in carbon and amino acid metabolic pathways (**Figure 6B**) were commonly modulated by drought and submergence stresses,

further confirming the carbon and amino acid metabolisms were extensively changed in response to abiotic stresses.

#### DISCUSSION

Plants periodically exposed to drought and submergence stresses in field condition which greatly inhibited plant growth, development and production. Abiotic stresses trigger complex signaling transduction pathways which may lead to an imbalance between antioxidant defenses and the amount of ROS, resulting in oxidative stress (Pastori and Foyer, 2002; Xiong et al., 2002). ROS are harmful by-products of normal cellular metabolism in aerobic organisms (Apel and Hirt, 2004; Miller et al., 2010) and can directly attack membrane lipids, resulting in lipid peroxidation and oxidation of proteins and nucleic acids (Kranner et al., 2010; Alhdad et al., 2013). In addition to the toxicity of ROS, ROS are necessary for inter- and intracellular signaling and considered to be signaling molecules that regulate plant growth and development, adaptation to abiotic and biotic stress factors (Apel and Hirt, 2004; Mittler et al., 2004). To scavenge ROS, plants have evolved an efficient enzymatic and nonenzymatic antioxidative system to protect themselves against oxidative damage and fine modulation of low levels of ROS for signal transduction. Enzymatic antioxidants in plant include SOD, CAT, POD, GR, DHAR, GST, and PRX (Miller et al., 2010; Meyer et al., 2012; Noctor et al., 2014). Non-enzymatic antioxidants including glutathione (GSH), ascorbic acid (AsA), carotenoids, tocopherols, and flavonoids are also crucial for ROS homeostasis in plant (Gill and Tuteja, 2010). In this study, enzyme activities of POD, CAT, and GR increased after drought and submergence treatments (**Figure 5**), while protein abundances of SOD, POD, and PRX were enhanced by drought but inhibited by submergence (**Table 2**). However, H2O2, O2• <sup>−</sup> and MDA contents increased only after drought treatment (**Figure 5**), and no significant changes were found for submerged bermudagrass (**Figure 5**). These results showed that bermudagrass under drought condition suffered from oxidative stress while submerged plants did not.

Besides traditional enzymatic and non-enzymatic antioxidants, increasing evidences indicated that soluble sugars have a dual role with respect to ROS (Couee et al., 2006; Keunen et al., 2013). Soluble sugars were directly linked with the production rates of ROS by regulation ROS producing metabolic pathways, such as mitochondrial respiration or photosynthesis. Conversely, they also feed NADPH-producing metabolism such as the oxidative pentose-phosphate pathway to involved in antioxidative processes (Couee et al., 2006). Drought stress caused significant increases of soluble sugars and sucrose (**Figure 2**). Proteomic analysis also revealed that 14 proteins involved in photosynthesis and carbon fixation were highly induced under drought condition (**Figure 4**; **Table 2**). These data was confirmed by metabolomic results which showed that sugars, organic acid, sugar alcohols, and fatty acid increased after drought treatment (**Figure 6**). However, only slight changes were observed after submergence treatment. Proline, acting as osmoprotectors, protects protein structures from stress caused damages. Proline also functions as a ROS scavenger, especially for hydroxyl radical (Smirnoff and Cumbes, 1989). Higher proline content in plants has been shown to be associated with increased tolerance to oxidative stress (Arbona et al., 2008). In this study, drought stress increased proline content in bermudagrass while submergence had no significant effect on proline accumulation (**Figures 2**, **6**). Taken together, the decreases or insignificant changes of 85% metabolites in submerged bermudagrass may be probably related to its physiological dormancy encountered deep submergence stress (Gibbs and Greenway, 2003; Bailey-Serres and Voesenek, 2008).

Photosynthesis has a high capacity for production of ROS. The primary event of photosynthesis is light-driven electron transfer–a redox reaction. During photosynthesis, electrons produced from water are transferred from the reaction center of photosystem II (PSII) to the cytochrome b6f (Cyt b6f) complex by the mobile electron carrier plastoquinone (PQ). Electrons from the cytochrome b6f complex are then transferred to photosystem I (PSI) by plastocyanin (PC). Under adverse environmental condition, electrons of PSI can also be transferred to oxygen, which results in the generation of ROS (Pfannschmidt et al., 2001; Pfannschmidt, 2003). Three proteins identified as Cyt b6f were inhibited by both drought and submergence in bermudagrass (**Table 2**), indicating that transfer of electrons from PS II to PS I became impaired. In addition, ATP synthase and ATPase showed more than 8-fold increases only by drought treatment in bermudagrass (**Table 2**). These results indicated that both drought and submergence affected photosynthesis, however, drought promoted while submergence declined ATP biosynthesis. Moreover, 7 RuBisCO related proteins showed

2.3–25.6 folds intensity change in bermudagrass after drought treatment (**Table 2**). RuBisCO is involved in the first key step of carbon fixation during calvin cycle. These data verified that photosynthesis was promoted by drought, but inhibited after submergence.

bermudagrass growth.

Several amino acids such as leucine, isoleucine, methionine were significantly increased after submergence treatment, but decreased after drought stress (**Figure 6**). For example, the content of methionine using for ethylene synthesis was significantly induced by submergence stress, but decreased by drought stress. The ethylene accumulation is very important for plants to cope with submergence stress (Hattori et al., 2009; Niroula et al., 2012). In addition, some carbohydrates such as glucose, sucrose, sorbose, melibiose, and fructose were significantly down-modulated by submergence. Therefore, the bermudagrass could develop specific mechanism such as restriction of carbohydrate consumption and ethylene accumulation to cope with submergence stress during physiological dormancy period. Therefore, under submergence condition, bermudagrass may invoke a quiescence strategy with repressed carbohydrate metabolism and retarded cell elongation. This hypothesis was confirmed by completely inhibited growth after submerged (**Figure 1**).

It has been reported that waterlogging reduced biomass in cool-season creeping bentgrass (Huang et al., 1998; Jiang and Wang, 2006) and Kentucky bluegrass (Wang and Jiang, 2007), as well as in warm-season seashore paspalum and centipedegrass (Zong et al., 2015). However, waterlogging stimulated plant growth in other warm-season grass species such as knotgrass and spiny mudgrass (Zong et al., 2015). According to the field survey results in the Three Gorges Reservoir in China, bermudagrass can tolerate deep and prolonged submergence stress for half a year (Tan et al., 2010). Through, physiological analysis, we observed many parameters showed significant changes after drought treatment, while only slight changes after submergence treatment, including osmolytes accumulation and ROS level and antioxidant enzyme activities (**Figures 2**, **5**). Proteomics results showed that abundance of only 10 proteins increased by submergence, while 46 proteins by drought (**Figure 3**). Metabolomic analysis indicated that most of the metabolites were up-regulated by drought stress, while 34 of 40 metabolites contents exhibited down-regulation or no significant changes when exposed to submergence stress (**Figure 6**). These data were consistent with results observed by Tan et al. (2010) that submergence decreased total soluble carbohydrate and starch contents in bermudagrass. As reported previously (Shi et al., 2014), 27 proteins were induced by drought in Yukon leaf and 8 of them were identified to be drought stress inducible in this study, including, chitinase, SOD, and heat shock proteins (**Table S3**). All these data indicated that ROS and stress related proteins played important role during bermudagrass stress response.

#### REFERENCES


In conclusion, bermudagrass might slow down metabolisms such as carbonhydrate degradation and energy supply under submergence stress, resulting in completely inhibited growth (**Figures 1**, **7**). The quiescence strategy with retarded growth might allow bermudagrass to be adaptive to long-term submerged environment. However, bermudagrass developed drought stress tolerance through activation of photosynthesis and redox, leading to accumulation of compatible solutes and molecular chaperones (**Figure 7**).

#### ACKNOWLEDGMENTS

This research was supported by "the Hundred Talents Program," the Knowledge Innovative Key Program of Chinese Academy of Sciences (Grant No. Y154761O01076 and No.Y329631O0263) to ZC.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00951

Figure S1 | The detailed design of the experiment.

Table S1 | Details of different expression.

Table S2 | Quantitive value mg per g F.W.

Table S3 | Comparison of the proteomic results with the previous research (Shi et al., 2014).


reduction of oxidative stress in bermudagrass (Cynodon dactylon (L). Pers.). J. Pineal Res. 59, 120–131. DOI: 10.1111/jpi.12246


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2015 Ye, Shi, Wang and Chan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Difference of Physiological and Proteomic Changes in Maize Leaves Adaptation to Drought, Heat, and Combined Both Stresses

Feiyun Zhao1 †, Dayong Zhang2 †, Yulong Zhao1 †, Wei Wang<sup>1</sup> , Hao Yang<sup>1</sup> , Fuju Tai <sup>1</sup> , Chaohai Li <sup>1</sup> and Xiuli Hu<sup>1</sup> \*

*<sup>1</sup> State Key Laboratory of Wheat and Maize Crop Science, Collaborative Innovation Center of Henan Grain Crops, College of Life Science, Henan Agricultural University, Zhengzhou, China, <sup>2</sup> Provincial Key Laboratory of Agrobiology, Institute of Biotechnology, Jiangsu Academy of Agricultural Sciences, Nanjing, China*

### Edited by:

*Hanjo A. Hellmann, Washington State University, USA*

#### Reviewed by:

*Georgia Tanou, Aristotle University of Thessaloniki, Greece Mohammad-Zaman Nouri, Rice Research Institute of Iran in Mazandaran, Iran*

\*Correspondence:

*Xiuli Hu xiulihu@126.com † These authors have contributed equally to this work.*

#### Specialty section:

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

Received: *22 April 2016* Accepted: *15 September 2016* Published: *26 October 2016*

#### Citation:

*Zhao F, Zhang D, Zhao Y, Wang W, Yang H, Tai F, Li C and Hu X (2016) The Difference of Physiological and Proteomic Changes in Maize Leaves Adaptation to Drought, Heat, and Combined Both Stresses. Front. Plant Sci. 7:1471. doi: 10.3389/fpls.2016.01471*

At the eight-leaf stage, maize is highly sensitive to stresses such as drought, heat, and their combination, which greatly affect its yield. At present, few studies have analyzed maize response to combined drought and heat stress at the eight-leaf stage. In this study, we measured certain physical parameters of maize at the eight-leaf stage when it was exposed to drought, heat, and their combination. The results showed an increase in the content of H2O<sup>2</sup> and malondialdehyde (MDA), and in the enzyme activities of superoxide dismutase (SOD), ascorbate peroxidase (APX), and glutathione reductase (GR), but a decrease in the quantum efficiency of photosystem II (8PSII). The most obvious increase or decrease in physical parameters was found under the combined stress condition. Moreover, to identify proteins differentially regulated by the three stress conditions at the eight-leaf stage, total proteins from the maize leaves were identified and quantified using multiplex iTRAQ-based quantitative proteomic and LC-MS/MS methods. In summary, the expression levels of 135, 65, and 201 proteins were significantly changed under the heat, drought and combined stress conditions, respectively. Of the 135, 65, and 201 differentially expressed proteins, 61, 28, and 16 responded exclusively to drought stress, heat stress, and combined stress, respectively. Bioinformatics analysis implied that chaperone proteins and proteases play important roles in the adaptive response of maize to heat stress and combined stress, and that the leaf senescence promoted by ethylene-responsive protein and ripening-related protein may play active roles in maize tolerance to combined drought and heat stress. The signaling pathways related to differentially expressed proteins were obviously different under all three stress conditions. Thus, the functional characterization of these differentially expressed proteins will be helpful for discovering new targets to enhance maize tolerance to stress.

Keywords: proteomics, iTRAQ labeling, combined drought and heat stress, maize, physiological characterization

### INTRODUCTION

Under field conditions, crops are often subjected to a combination of several stresses, which have an adverse effect or may even prove lethal. Recently, researchers have begun to pay more attention to the potential molecular mechanisms involved in crop endurance to combined stress (Rampino et al., 2012; Liu et al., 2015; Obata et al., 2015). The evidence shows that crops exhibit unique physiological and molecular responses to combined stress, which cannot be directly inferred from plant responses to single stresses. Moreover, the simultaneous occurrence of several stresses brings about a complexity of plant responses that are highly controlled by different or opposing signaling pathways (Rollins et al., 2013; Johnson et al., 2014; Suzuki et al., 2014).

Heat, drought, and their combination are the main stress factors for field crops and are responsible for most production losses (Lobell et al., 2011a; Suzuki et al., 2014). Moreover, global climate change is gradually increasing the occurrence and distribution of these stressors, causing further reductions in crop yield (Rasul et al., 2011). Thus, to meet food demand, it is necessary to develop crops with elevated endurance to drought, heat stress, and their combination. Some studies have looked specifically at the effects of drought, heat stress, and their combination on barley (Rollins et al., 2013; Ashoub et al., 2015), wheat (Rampino et al., 2012; Liu et al., 2015), Sorghum bicolor (Johnson et al., 2014), and maize (Hu et al., 2010, 2015). However, the functions of many proteins involved in crop responses to combined drought and heat stress remain unclear.

In recent years, global quantitative analysis to determine protein expression levels has been performed using iTRAQbased (isobaric tags for relative and absolute quantitation) methods and quantitative proteomic and LC-MS/MS (liquid chromatography/tandem mass) assays (Alvarez et al., 2014; Han et al., 2014), which facilitate the simultaneous analysis of the differential expression of proteins under control and stress conditions. Large-scale proteomic analyses have been conducted regarding crop responses to stress (Alvarez et al., 2014; Xie et al., 2016). For example, in the response of wheat to drought stress, a large number of proteins inherently exhibited different levels of expression between two varieties with different tolerances to drought stress (Alvarez et al., 2014). The genetic basis of proteome variation in crop responses to stress may represent mechanisms of stress adaptation that can be exploited in future crop-breeding efforts; this is a feasible strategy for developing drought- and heat-tolerant crop cultivars to help increase crop production under future challenging environments.

Maize (Zea mays L.) not only constitutes a major cereal crop, and food for both humans and animals, but has also become a critical resource for industrial use and for bio-energy production throughout the world. Maize is highly productive under suitable growth conditions. However, in many regions of the world, maize is mainly grown in semi-arid environments characterized by water scarcity, high temperature, and a combination of these conditions in the field. Maize originated from the tropics but is still sensitive to drought and heat, particularly after reaching the eight-leaf stage (Chen et al., 2010). In the maize-growing areas of China, ∼60% of crops are often subjected to drought and heat, which may result in an ∼30% yield loss per year. Along with global climate change, it is predicted that these stresses will become major challenges to maize yields and will lead to a loss of 15∼20% of world maize production each year (Lobell et al., 2011b; Chen et al., 2012). Thus, in terms of maize breeding programs, the need to improve maize tolerance to drought, heat, and their combination has become a top priority (Chen et al., 2012).

However, at present, few studies have analyzed maize response to combined drought and heat stress at the eight-leaf stage. In this study, to discover more about such responses, we analyzed the changes in certain physical parameters and iTRAQbased proteomes in maize exposed to heat, drought, and these conditions in combination. Furthermore, we conducted bioinformatics analyses to confirm the functions of the differentially expressed proteins in the adaptive response of maize to combined drought and heat stress. Such work should help advance our understanding of the molecular mechanisms involved in the response of maize plants to combined drought and heat stress.

#### MATERIALS AND METHODS

#### Plant Material and Stress Treatments

According to the method we have described previously (Hu et al., 2010), maize seeds (Zhengdan 958) were used in the experiments. Zhengdan 958 is a high-yield maize hybrid that is grown in China. The seeds were surface-sterilized for 10 min in 2% hypochlorite, washed in distilled water and germinated on moistened filter paper. The maize plants were grown in Hoagland's nutrient solution in a light chamber under 400 µmol m−<sup>2</sup> s <sup>−</sup><sup>1</sup> of photosynthetically active radiation, a 14-/10 h day/night cycle, a day/night temperature of 28/22◦C, and a relative humidity of 75%. When the eighth leaf was fully expanded, the plants were subjected to drought, heat, and combined stress treatments.

According to our previously described procedure (Hu et al., 2010), drought stress was imposed by placing the plants in polyethylene glycol (PEG) solution (−0.7 MPa, moderate drought) for 8 h at 28◦C and 40% relative humidity. Heat stress was applied by raising the temperature from 28 to 42◦C at a rate of 2◦C/h and then maintained at 42◦C for 1 h, for a total of 8 h. Therefore, each stress treatment lasted 8 h. The combined stress consisted of simultaneous treatment with PEG and heat stress. The control seedlings were maintained at 28◦C and 75% relative humidity. Next, the expanding leaves (the eighth from the bottom) of the treated and untreated seedlings were sampled, immediately frozen in liquid nitrogen, and stored at −80◦C until analysis. Three biological replicates were performed for each treatment.

#### Quantum Efficiency of Photosystem II

The quantum efficiency of photosystem II (8PSII) was measured using an OS-30p Chlorophyll Fluorometer (Opti-Sciences, Tyngsboro, MA, USA) on the eighth fully expanded leaf.

#### Malondialdehyde (MDA) Content

Malondialdehyde (MDA) content was measured according to the method described by Hodges et al. (1999): 50 mg fresh weight (FW) of leaves were homogenized in 1 ml of 80% (v/v) ethanol using a mortar and pestle. After centrifugation, the supernatant reacted with thiobarbituric acid to produce the pinkish-red chromogen, thiobarbituric acid-malondialdehyde (TBA-MDA). Absorbance was measured at 440, 532, and 600 nm by UV-vis (ultraviolet–visible) spectrophotometry. The MDA content was calculated as nmol/g FW tissue.

#### Enzyme Assays

According to the method we described previously (Hu et al., 2010), frozen leaf samples were homogenized (1:20 g/ml) in an extraction buffer consisting of 50 mM potassium phosphate, pH 7.0, 1 mM EDTA, and 1% polyvinylpyrrolidone, plus 1 mM ascorbate in the case of the APX assay. The homogenate was centrifuged at 15,000 × g for 20 min at 4◦C and the supernatant was immediately used for antioxidant enzyme assays.

The activities of antioxidant enzymes were also determined by the method described previously (Hu et al., 2010). Superoxide dismutase (SOD: EC 1.15.1.1) activity was assayed by monitoring the inhibition of photochemical reduction of nitroblue tetrazolium at 560 nm. One unit of SOD activity was defined as the amount of enzyme required to cause 50% inhibition of the nitro-blue tetrazolium reduction. APX (EC 1.11.1.11) activity was measured by monitoring the absorbance decrease at 290 nm as the ascorbate was oxidized. Glutathione reductase (GR: EC 1.6.4.2) activity was measured by following the change in oxidation at 340 nm in the glutathione-dependent oxidation of NADPH.

### Cytochemical Detection of Hydrogen Peroxide

Hydrogen peroxide (H2O2) was visualized at the subcellular level using cerium(III) chloride (CeCl3) for localization (Bestwick et al., 1997; Hu et al., 2005). Electron-dense CeCl<sup>3</sup> deposits are formed in the presence of H2O<sup>2</sup> and are visible by transmission electron microscopy. Tissue pieces (1∼2 mm<sup>2</sup> ) were excised from the treated and untreated leaves and incubated in freshly prepared 5 mM CeCl<sup>3</sup> in 50 mM 3-(N-morpholino) propanesulfonic acid (MOPS) at pH 7.2 for 1 h. The leaf sections were then fixed in 1.25% (v/v) glutaraldehyde and 1.25% (v/v) paraformaldehyde in 50 mM sodium cacodylate buffer, pH 7.2, for 1 h. After fixation, tissues were washed twice for 10 min in the same buffer and post-fixed for 45 min in 1% (v/v) osmium tetroxide, and then dehydrated in a graded ethanol series (30∼100%; v/v) and embedded in Eponaraldite (Agar Aids, Bishop's Stortford, UK). After 12 h in pure resin, followed by a change of fresh resin for 4 h, the samples were polymerized at 60◦C for 48 h. Blocks were sectioned (70∼90 nm) on a Reichert-Ultracut E microtome, and mounted on uncoated copper grids (300 mesh). Sections were examined using a transmission electron microscope at an accelerating voltage of 75 kV.

### Protein Extraction

As reported in our earlier study (Hu et al., 2010), total proteins from the eighth leaf of the maize plants were extracted according to the method reported by Wang et al. (2013) and Zhang et al. (2014). Briefly, ∼0.5 g fresh leaves from each biological replicate were ground into a fine power in liquid nitrogen using a mortar and pestle and further ground in 4 ml of SDS buffer (30% sucrose, 2% SDS, 100 mM Tris-HCl, pH 8.0, 50 mM EDTA-Na2, 20 mM DTT) and 4 ml phenol (Tris-buffered, pH 8.0), then 1 mM phenylmethanesulfonyl fluoride (PMSF) and PhosSTOP phosphatase inhibitor cocktail (one tablet/10 ml; Roche, Basel, Switzerland) was added to inhibit protease and phosphatase activity. The mixture was thoroughly vortexed for 30 s and the phenol phase was separated by centrifugation at 14,000 × g and 4◦C for 15 min. The upper phenol phase was pipetted into new 10 ml tubes, and four-fold volumes of cold methanol plus 100 mM ammonium acetate were added. After centrifugation at 14,000 × g and 4◦C for 15 min, the supernatant was carefully discarded and the precipitated proteins were washed twice with cold acetone. Finally, the protein mixtures were harvested by centrifugation. Protein concentrations were measured using a 2-D Quant Kit (Amersham Biosciences, Piscataway, NJ, USA), with bovine serum albumin (BSA; 2 mg/ml) as the standard. To enhance the quantitative accuracy, extracted proteins from every biological replicate were adjusted to the same concentration for the subsequent analysis.

#### Protein Digestion and ITRAQ Labeling

Protein digestion was performed according to the FASP (filteraided sample prep) procedure described by Wi´sniewski et al. (2009) and Lv et al. (2014), and the resulting peptide mixture was labeled using 4-plex iTRAQ reagent according to the manufacturer's instructions (Applied Biosystems, Foster City, CA, USA). Briefly, 200 µg of protein from each sample was mixed with 30 µl of STD buffer (4% SDS, 100 mM DTT, 150 mM Tris-HCl pH 8.0). The detergent, DTT, and other lowmolecular-weight components were removed using UA buffer (8 M urea, 150 mM Tris-HCl pH 8.0) with repeated ultrafiltration (Microcon units, 30 kDa). Next, 100 µl of 0.05 M iodoacetamide in UA buffer was added to block reduced cysteine residues, and the samples were incubated for 20 min in darkness. The filters were washed three times with 100 µl of UA buffer, then twice with 100 µl of DS buffer (50 mM triethylammonium bicarbonate at pH 8.5). Finally, the protein suspensions were digested with 2 µg of trypsin (Promega, USA) in 40 µl of DS buffer overnight at 37◦C, and the digested peptides were collected as a filtrate. The peptide content was estimated via UV absorption at 280 nm using an extinction coefficient of 1.1 per 0.1% (g/l) solution, which was calculated based on the proportion of tryptophan and tyrosine residues in vertebrate proteins.

For labeling, each iTRAQ reagent was dissolved in 70 µl of ethanol and added to the respective peptide mixture. The samples were referred to as control (under no stress), drought, heat, and combined drought and heat stress and were labeled with reagent and vacuum dried.

### Peptide Fractionation with Strong Cation Exchange Chromatography

iTRAQ-labeled peptides were fractionated by strong cation exchange (SCX) chromatography using the AKTA Purifier system (GE Healthcare, USA). The dried peptide mixture was reconstituted and acidified with 2 ml buffer A (10 mM KH2PO<sup>4</sup> in 25% of Acetonitrile, pH 2.7) and loaded onto a PolySULFOETHYL 4.6 × 100 mm column (5 µm, 200 Å, PolyLC Inc, MD, USA). The peptides were eluted at a flow rate of 1 ml/min with a gradient of 0–10% buffer B (500 mM KCl, 10 mM KH2PO<sup>4</sup> in 25% of acetonitrile, pH 2.7) for 2 min, 10– 20% buffer B for 25 min, 20–45% buffer B for 5 min, and 50–100% buffer B for 5 min. The elution was monitored by absorbance at 214 nm, and fractions were collected every 1 min. The collected fractions (about 30 fractions) were finally combined into 10 pools and desalted on C18 cartridges [EmporeTM SPE cartridges C18 (standard density), bed I.D. 7 mm, volume 3 ml, Sigma]. Each pool was concentrated by vacuum centrifugation and reconstituted in 40 µl of 0.1% (v/v) trifluoroacetic acid. All samples were stored at −80◦C until LC-MS/MS analysis.

### Liquid Chromatography Electrospray Ionization and Tandem MS (MS/MS) Analysis by Q-Exactive

Analyses were performed using a Q-Exactive mass spectrometer that was coupled to an Easy-nLC system (Thermo Fisher Scientific, Odense, Denmark). Ten microliters of each fraction was injected for nanoLC-MS/MS analysis. The peptide mixture (5 µg) was loaded onto a C18 reversed-phase column (Thermo Scientific Easy Column, 10 cm long, 75 µm inner diameter, 3 µm resin) in buffer A (0.1% formic acid) and separated with a linear gradient of buffer B (80% acetonitrile and 0.1% formic acid) at a flow rate of 250 nl/min controlled by IntelliFlow technology over 140 min. MS data was acquired using a data-dependent "top10" method, which dynamically chooses the most abundant precursor ions from the survey scan (300–1800 m/z) for HCD (higher collision dissociation) fragmentation. Determination of the target value is based on predictive automatic gain control (pAGC). Dynamic exclusion duration was 60 s. Survey scans were acquired at a resolution of 70,000 at m/z 200, and resolution for HCD spectra was set to 17,500 at m/z 200. Normalized collision energy was 30 eV and the underfill ratio, which specifies the minimum percentage of the target value likely to be reached at maximum fill time, was defined as 0.1%. The instrument was run with peptide recognition mode enabled.

### Sequence Database Searching and Data Analysis

MS/MS spectra were searched using Mascot 2.2 (Matrix Science) embedded in Proteome Discoverer 1.4 against the uniprot\_Zea\_mays\_87227\_20150504.fasta (87227 sequences, downloaded on May 4, 2015) and the decoy database. For protein identification, the following options were used: peptide mass tolerance, 20 ppm; MS/MS tolerance, 0.1 Da; enzyme, trypsin; missed cleavage, 2; fixed modification Carbamidomethyl

The protein and peptide probabilities were set at 50 and 60%, respectively. Only proteins with at least two unique peptides with a Mascot score of at least 25 and detected in at least two replicates were further analyzed.

For each replicate of proteomics, iTRAQ ratios between drought/heat/combined stress and controls for each run were converted to z-scores to normalize the data.

### Bioinformatics

The molecular functions of the identified proteins were classified according to their gene ontology annotations and their biological functions. The subcellular localization of the proteins identified in this study were predicted using the publicly available program WolfPsort (http://wolfpsort.org). Protein–protein interaction networks were predicted using the publicly available program STRING (http://string-db.org/). STRING is a database of known and predicted protein–protein interactions. The interactions include direct (physical) and indirect (functional) associations, and they are derived from four sources: the genomic context, high-throughput experiments, co-expression, and previous knowledge. STRING quantitatively integrates the interaction data from these sources for a large number of organisms, and where applicable, transfers information between these organisms.

According to the known or predicted cellular localization and molecular function of the proteins, as determined by Blast2Go (http://www.blast2go.com), specific groups of proteins were selected and analyzed on the basis of, for example, stimulus responses, chloroplasts proteins and enzymes.

### Statistical Analysis

The mean of three replicates was used to ascertain the protein assays. Means were compared using one-way analysis of variance and Duncan's multiple range test at a 1% level of significance.

### RESULTS

### Comparison of Physical Parameters Affected by the Three Stress Conditions

To investigate the level of H2O<sup>2</sup> accumulation in the leaves of maize plants exposed to the drought, heat and combined stress conditions, we used a cytochemical technique whereby CeCl<sup>3</sup> reacts with H2O<sup>2</sup> to form electron-dense deposits of cerium perhydroxide (CeH8O4; Bestwick et al., 1997). Under normal conditions (control), no CeH8O<sup>4</sup> deposit¡<sup>a</sup> as an indication of H2O<sup>2</sup> accumulation¡awas observed in the mesophyll cells and chloroplasts (**Figures 1A,E**). Under the drought, heat, and combined stress conditions, H2O<sup>2</sup> accumulation was visible in the walls of mesophyll cells (**Figures 1B–D**) and in chloroplasts (**Figures 1F–H**). Both in the walls of the mesophyll cells (**Figures 1B–D**) and in the chloroplasts (**Figures 1F–H**), the highest level of H2O<sup>2</sup> accumulation was found under the combined stresses, and the second-highest level was observed under heat stress.

MDA is generated by lipid peroxidation, so a change in MDA content reflects the extent of membrane damage. In

FIGURE 1 | Cytochemical localization of H2O2 accumulation in mesophyll cells of maize variety Zhengdan 958. Arrows indicate CeCl3 precipitates. (A,E), control; (B,F), drought; (C,G), heat; (D,H), combined drought and heat stress; C, chloroplast; CW, cell wall; IS, intercellular space; M, mitochondrion; N, nucleus; V, vacuole. Bar = 1 µm. All experiments were repeated at least three times with similar results.

TABLE 1 | Comparisons of physiological indexes in maize leaves under CK, D, H, and DH conditions.


*CK, control; D, drought stress; H, heat stress; DH, combined drought and heat stress. Each value represents the average of three biological replicas. For Duncan's Results, different characters are considered to be significant among different treatments.*

the present study, MDA content was prominently elevated by drought stress, heat and combined stress compared with the control (**Table 1**). The most obvious elevation was that under combined stress, followed by that observed under heat stress. 8PSII is a chlorophyll fluorescence parameter that is classically used to monitor changes in photosynthetic performance. 8PSII was significantly decreased by these three stresses. The most obvious decrease was found under the combined stress, followed by that under the heat stress.

SOD catalyzes the dismutation of O<sup>−</sup> 2 to O<sup>2</sup> and H2O2. APX and GR are the two key enzymes of the Halliwell–Asada pathway for the removal of H2O2. Compared with the control, the drought, heat, and combined stress conditions enhanced the activities of SOD, APX, and GR. The most obvious elevation was under the combined stress, followed by that under the heat stress (**Table 1**). Taken together, these results indicate that the combined drought and heat stress had the most significant effect on these parameters, followed by the heat stress.

#### Identification of Differentially Expressed Proteins under the Three Stress Conditions

After the maize plants at the eight-leaf stage were subjected to the drought, heat and combined stress conditions, newly expanded leaves were used to extract the total proteins, and then multiplex iTRAQ-based quantitative proteomic and LC-MS/MS assays were performed on the total proteins, resulting in the identification of 5238 proteins in these treatments at a false discovery rate (FDR) of 1%. In detail, based on a significant linear regression (p < 0.01) and a threshold of ≥ 1.5-fold or ≤ 0.66-fold change ratio of stress-induced protein expression levels compared with control: under the heat stress, the expression level of 135 proteins showed significant changes, of which 67 were upregulated and 68 were down-regulated; under the drought stress, the expression level of 68 proteins showed significant changes, of which 46 were up-regulated and 22 were down-regulated; and under the combined stress, the expression level of 201 proteins showed significant changes, of which 113 were up-regulated and 88 were down-regulated (**Figure 2**). Among 246 proteins that showed prominent changes, 18 were commonly found under all three stress conditions (**Table 1**), 104 proteins were common to the heat stress and combined stress conditions (Table S1), 21 were common to the drought stress and combined stress conditions (Table S2), and one was common to the drought stress and heat stress (Table S3), while 15 proteins were identified under the heat stress alone (Table S4), 28 proteins were identified under the

drought stress alone (Table S5), and 59 proteins were exclusively identified under the combined stress (Table S6).

#### Proteins Related to Stimulus Response under the Three Stress Conditions

In this study, the expression level of 19, 39, and 59 proteins related to stimulus response showed significant changes under the drought, heat, and combined stress conditions, respectively (**Table 3**). Under all three stress conditions, ribonucleoprotein A and fatty acid desaturase were down-regulated in common (**Table 1**). Under the drought stress and combined stress conditions (Table S2), RAB17 protein, MTN3, uncharacterized protein (B4G1H1), glutathione S-transferase GST6, dehydrin, ABA-responsive protein and aquaporin PIP2-6 were downregulated in common, and with the exception of MTN3, the other six proteins were significantly up-regulated. Under the heat stress and combined stress conditions (**Table 3**), 36 proteins were up-regulated in common, of which 20 were shock proteins (HSPs), including 14 small HSPs (sHSPs). All of these HSPs were obviously up-regulated by the heat stress and combined stress, but were only slightly affected by the drought stress (Table S1). In addition, 9 and 13 proteins related to stimulus response were found to be differentially expressed only under the drought stress and combined stress, respectively (**Table 3**). Of particular note was the finding that the expression levels of abscisic acid (ABA) stress ripening protein 2, ethylene-responsive protein, and ABA-, stress- and ripening-inducible-like protein were significantly upregulated under the combined stress (Table S6). All of these proteins related to stimulus response under heat stress were found to be differentially expressed under the combined stress (**Table 3**).

### Chloroplast Proteins Showing Significant Changes

The chloroplast proteome of photosynthetic plants includes ∼3000 different proteins, of which components of the photosynthetic apparatus are very abundant. In this study, 13, 21, and 32 chloroplast proteins were identified under the drought, heat and combined stress conditions, respectively. Moreover, most of the chloroplast proteins found were identified as uncharacterized proteins with molecular functions relating to nucleotide binding or catalytic activity. Under the three stress conditions (**Table 2**), ribonucleoprotein A, putative uncharacterized protein (B6UCG5) and uncharacterized protein (K7U7W9) were down-regulated in common. Under the heat stress and combined stress conditions (Table S1), except for iron-sulfur assembly protein IscA, the other chloroplast proteins were down-regulated in common. Eight, one and 11 chloroplast proteins were specific to the drought, heat, and combined stress conditions, respectively (**Table 3**). These results showed that heat and combined stress conditions may exert a more obvious effect on maize chloroplast function than drought stress.

#### Responses of Kinases and Phosphatases to the Three Stress Conditions

It is also notable how various enzymes (including in particular kinases and phosphatases) responded to the stresses. Under the drought stress, 12 enzymes, including one kinase and one phosphatase, were identified as differentially expressed. Under the heat stress, 27 enzymes, including two kinases and one phosphatase, were identified. Under the combined stress, 38 enzymes, including four kinases and one phosphatase, were identified (**Table 3**). In addition, alpha-galactosidase, fatty acid desaturase, and asparagine synthetase (B5U8J8) were commonly found under all three stress conditions. In particular, the expression level of asparagine synthetase (B5U8J8) had a 5.28 and 10.72-fold increase under the drought stress and combined stress compared with the control (**Table 2**), respectively, while asparagine synthetase (B6ETR5) was significantly increased only by the combined stress (Table S6). Three isoforms of stachyose synthase (B6SYY2, B6SRV6, and B6UBW7) were identified under the drought stress and combined stress, of which B6SYY2 and B6SRV6 were significantly increased by the drought stress, while B6UBW7 was increased by the combined stress; 21 enzymes were commonly found under the heat stress and combined stress (**Table 3**). The remaining eight, three and 13 enzymes were found only under the drought stress, heat stress, and combined stress, respectively.

To identify the interactions of enzymes and HSPs with other proteins under all three stress conditions, the protein interactions among differentially expressed proteins were analyzed using STRING software (**Figures 3**–**5**). Under the drought stress and the combined stress, asparagine synthetase (4332506) was found to interact with decarboxylase (4329593). In fact, for all three stress conditions, extensive interactions were


*CK, control; D, drought stress; H, heat stress; DH, combined drought and heat stress. Each ratio was the average of three replicates. a Each value represents the average of three biological replicas. For Duncan's Results, different characters are considered to be significant between different treatments.*

found among all of the chloroplast proteins (**Figures 3**–**5**). Under the heat stress (**Figure 4**) and the combined stress (**Figure 5**), extensive interactions were found amongst HSPs, or between HSPs and other proteins. For example, some HSPs (dnaK-family proteins 4332080, 4332420, 4327388; chaperone protein clpB1-4328515; HSP101-4339343) exhibited interactions with phosphosulfolactate synthase-related protein (4341866), while other HSPs (4334919, 4342077, 4330134) exhibited interactions with CS domain-containing protein. Interactions between HSPs were also observed under the combined stress (**Figure 5**). An HSP70-family protein (4330492) exhibited extensive interactions with dnaK-family proteins (4332080, 4332420, 4327388, LOC\_Os09g31486.1), heat shock protein ST1 (4330134), HSP (4334919), HSP101 (4339343), HSP20/alpha crystalline family proteins (4332357, 4325697, 4332363), and HSP18 (rice protein query sequences corresponding to maize protein query sequences; see Table S7 for the drought stress, Table S8 for the heat stress, and Table S9 for the combined stress). These results indicate that HSPs as chaperones probably play a role in protecting protein functions under heat stress and combined stress conditions.

#### Changes in Receptor Proteins

Receptors can make cells detect changes in the internal or external environment. In this study, the expression levels of three receptor proteins were significantly regulated under the drought stress, heat stress, and combined stress conditions. The expression level of brassinosteroid LRR receptor kinase (B6SV61) was reduced by the heat stress and combined stress (Table S1). The expression level of mitochondrial import receptor subunit TOM22 (B6U2X6) was increased by all three stress conditions, but under the combined stress alone there was an increase of up to 1.5-fold (Table S6). The expression level of gibberellin receptor GID1L2 (B6TC25) was decreased by the heat stress and combined stress, but under the combined stress alone there was a decrease of up to 1.5-fold (Table S6).

### The Signaling Pathways Related to Differentially Expressed Proteins under the Three Stress Conditions

All identified proteins were classified by gene ontology (GO) annotation software and then classified as three functional groups: molecular function, biological process, and cellular component. The results of the GO analyses for the drought, heat and combined stress conditions are shown in **Figures 6**– **8**, respectively. Most of the annotated molecular functions were found to relate to binding and catalytic activity, while most of the annotated biological processes were found to relate to cellular and metabolic processes.

On the basis of biological process analysis using the BLAST2GO program, among differentially expressed proteins: for the drought stress, 19 proteins were classified as "response to stimulus," two were involved in transport, and 11 were classified as binding proteins involved in DNA binding, protein binding and nucleotide binding (**Figures 6A,B**; **Table 2**); for the heat stress, 39 proteins were categorized as "response to stimulus," one was involved in transport, and 56 were classified as binding proteins involved DNA binding, protein binding, and nucleotide binding (**Figures 7A,B**; **Table 2**); for the combined stress, 59 proteins were categorized as "response to stimulus," three were

TABLE 3 | Stimulus response proteins, chloroplast proteins and enzymes with significant expression changes under D, H, and DT respectively.


mRNA/K7VC62

2/B6SU65

*(Continued)*


*(Continued)*

Frontiers in Plant Science | www.frontiersin.org October 2016 | Volume 7 | Article 1471 |


involved in transport, and 84 were classified as binding proteins involved in DNA binding, protein binding, and nucleotide binding (**Figures 8A,B**; **Table 2**).

In the light of KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis: under the drought stress, the differentially expressed proteins were found to be mainly involved in the galactose metabolism, photosynthesis, and carbon metabolism pathways (**Figure 6D**); under the heat stress, the differentially expressed proteins were found to be mainly involved in protein processing in the endoplasmic reticulum (ER), in antigen processing and presentation, and in estrogen signaling pathways (**Figure 7D**); under the combined stress, the signaling pathways were found to be similar to those found under the heat stress alone (**Figure 8D**). These results indicate that the signaling pathways mediated by the heat stress and combined stress were obviously different to those mediated by the drought stress. In particular, of the differentially expressed proteins that related to protein processing in endoplasmic reticulum, 20 were observed under the heat stress and 22 under the combined stress, indicating that the signaling pathways related to protein processing play an important role in maize response to heat stress and combined drought and heat stress conditions.

#### DISCUSSION

The final physiological response is dictated by the growth stage and plant tissue type, along with the severity and duration of the stress exerted on the plants. After developing to the eightleaf stage, maize is sensitive to heat stress, especially to combined drought, and heat stress. In this study, we measured the changes in physical parameters and comprehensively analyzed the differentially expressed proteins in maize leaves in response to drought, heat, and their combination using iTRAQ-based quantitative proteomic and LC-MS/MS methods. The combined stress caused very significant changes in the level of protein expression in the maize leaf, and some changes exclusively resulted from the combined drought and heat stress.

#### Physiological Parameters Affected by Stress

The generation of reactive oxygen species (ROS) often leads to the destruction of cellular structures, which ultimately causes cell death. MDA is widely used as a marker of oxidative lipid injury. In this study, the accumulation of H2O<sup>2</sup> and MDA, and the activities of SOD, APX, and GR, were enhanced by these three stress treatments, especially by the heat stress and combined stress. Our results indicated that all three stress conditions induced and aggravated membrane injury; in addition, they showed that the maize plants triggered an anti-oxidative defense mechanism to alleviate the ROS, which enhanced their tolerance to stress.

#### Chloroplast Proteins Affected by Combined Stress

Abiotic stresses bring about serious damage to plant photosynthetic systems. In photosynthetic systems, photosystem

*D, drought stress; H, heat stress; DH, combined drought and heat stress. "*+*" or "*−*" indicates common proteins or non-common*

 *proteins.*

II (8PSII) is one of the most sensitive components under drought and heat stresses. In soybean, 25 differentially expressed proteins of photosynthesis were involved in RuBisCO regulation, electron transport, the Calvin cycle, and carbon fixation under drought and heat stress conditions (Das et al., 2016). In our study, 13, 21, and 32 proteins were related to chloroplast function under the drought, heat, and combined stress conditions, respectively. Remarkably, under the combined stress, the 32 chloroplast

proteins were mainly found to be involved in chlorophyll biosynthesis, electron transport, carbon fixation, transcription regulation, lipid metabolism, and chaperone function.

Four uncharacterized proteins (B4FHM6, B4FZN7, K7USR3, and K7UWZ6) related to chlorophyll syntheses were down-regulated , while PSI reaction center subunit V/N, HSP101, and FtsH protease were up-regulated by the heat stress and combined stress. FtsH protease is an ATP-dependent metalloprotease. In soybean, FtsH protease was up-regulated under the heat stress (Das et al., 2016). In Arabidopsis, FtsH protease was directly involved in turnover of the 8PSII reaction center D1 protein (Kato et al., 2009). Taken together, our results also suggest that chloroplastic FtsHs may protect chloroplast photosynthesis under heat stress and combined stress. In rice, it was reported that 8PSI was more susceptible to heat stress than 8PSII (Essemine et al., 2016), which explained why the two 8PSI reaction center subunits V and N showed significant changes under heat stress and combined stress in this study.

#### Chaperone and Senescence-Related Proteins in Response to Stress Stimuli

Heat stress has a negative effect on protein stability and enzyme functions in the cell. Responding to this stimulus, plants synthesized HSPs and chaperone-like proteins in order to restore the correct configuration of proteins and impede aggregation (Wang et al., 2004). sHSPs play important and comprehensive

roles in the ability of plants to combat heat stress (Eisenhardt, 2013; Mu et al., 2013).

In this study, protein profiles were found to be more similar under heat stress and combined stress, and were significantly different under drought stress. Among the proteins found to be differentially regulated in common under heat stress and combined stress, HSPs, including 14 sHSPs, were the mostrepresented. Similar results were found in terms of the response of wild barley and soybean to heat stress and combined drought and heat stress (Ashoub et al., 2015; Das et al., 2016), but more HSPs were identified in the present study, which supports our finding that maize is highly sensitive to heat at the eight-leaf stage. In this study, HSPs interacted strongly with other proteins or with HSPs under heat stress and combined stress. These results are the first to demonstrate the similarity of HSP expression in the response of maize to heat stress and combined drought and heat stress, and reaffirm that HSPs are important in terms of plant responses to heat stress and combined drought and heat stress.

The abundance of some proteins was exclusively changed under the combined stress. In particular, the combined stress increased the expression of ftsH6-Z. mays FtsH protease, which had a predicted interaction with photosystem I reaction center subunit, ATP synthase protein I. It has been reported that ATP-dependent zinc metalloprotease FTSH 1 is involved in the turnover of oxidatively damaged D1 proteins of 8PSII (Adam et al., 2001) and contributes to the heat tolerance of grapevine (Rocheta et al., 2014). These results indicate that the distinct forms of metalloprotease FTSH may play an active role in protecting chloroplast from heat stress and combined drought and heat stress. Moreover, the expression of ABA stress ripeningrelated protein 2, ethylene-responsive protein, and ABA-, stressand ripening-inducible-like protein was found to be significantly increased under all three stress conditions. These three proteins are all associated with leaf senescence and fruit ripening. In plants, leaf senescence promotes the transfer of nutrients to developing and storage tissues. It has been reported that the

senescence and abscission of older leaves, and the subsequent transfer of nutrients, increases plant survival under drought and heat stress (Munné-Bosch and Alegre, 2004; Lim et al., 2007). In addition, studies on transgenic tobacco have indicated that delayed leaf senescence increases tobacco endurance to drought stress (Rivero et al., 2007). Furthermore, studies have shown that ABA affects mango fruit ripening by regulating ethylene changes (Zaharah et al., 2013), and promotes leaf senescence by enhancing ethylene production in submerged aquatic plants (Jana and Choudhuri, 1982). Nevertheless, with regard to the response of Arabidopsis to drought stress, the study by Zhao et al. (2016) found that ABA promotes leaf senescence in an ethyleneindependent manner. However, further research is needed to further prove whether ABA and ethylene can enhance maize tolerance to drought, heat and combined stress conditions by promoting leaf senescence.

#### Protein Processing in the ER

The endoplasmic reticulum (ER) is an important organelle responsible for proteostasis. The accumulation of misfolded proteins in the ER disturbs ER homeostasis and thus brings about ER stress. Misfolded proteins may bind to chaperone BiP and be degraded through the proteasome (Perri et al., 2016). The protein disulfide isomerase (PDI) is an abundant oxidoreductase in eukaryotic ER and catalyzes the folding of proteins (Gruber et al., 2006). HSPs may not only prevent the inappropriate interaction of proteins and promote correct folding, but may also play a significant role in the degradation pathways (Bozaykut et al., 2014).

In this study, the protein processing that occurred in ER was the most prominent pathway under the heat stress and combined stress stimulus. In particular, we observed one down-regulated PDI and 17 up-regulated HSPs (including 12 sHSPs) under the heat stress and combined stress. The results suggest that the depression of PDI expression may cause the accumulation of misfolded proteins in ER. Thus, the expression of HSPs was significantly elevated in order to eliminate misfolded proteins. It is important to uncover the role of HSPs in protein turnover under heat stress and combined stress.

differentially expressed proteins were identified in this study and classified according to their known or predicted cellular localization, using the Blast2Go

Phosphatases and Kinases

(http://www.blast2go.com) program.

The interplay between phosphatases and kinases strictly controls many biological processes in plants (Johnson, 2009; Pjechová et al., 2014). Brassinosteroids (BRs) regulate various aspects of plant development (Yang et al., 2011). It is well-known that BR and ABA exert an antagonistic effect on plant development. BR signaling mutant bak1 (BRI1-associated receptor kinase 1) has been found to lose more water than wild-type and to be insensitive to ABA in stomatal closure, suggesting that BAK1 is involved in stomatal closure induced by ABA (Shang et al., 2016). Our own results showed that heat stress and combined stress down-regulated the expression of brassinosteroid LRR receptor kinase. So, we hypothesized that brassinosteroid LRR receptor kinase may have a similar function to BAK1 under heat stress and combined stress. Namely, in relation to the control plants, the decrease in the level of BAK1 expression may have caused the maize leaf to lose more water, which would be helpful in adapting to the heat stress and combined stress conditions.

Plant glycogen synthase kinase-3 (GSK-3) belongs to a multigene family that regulates various different physiological responses (Saidi et al., 2012; Youn and Kim, 2015). It has been reported that alfalfa MsK4 and Arabidopsis AtGSK1/ASK1 and ASKα promote enhanced salt tolerance (Piao et al., 2001; Kempa et al., 2007; Santo et al., 2012); However, rice GSK1 has been shown to reduce salt tolerance (Koh et al., 2007). The study by Youn and Kim (2015) revealed that AtSK21/BIN2 and AtSK12 was critical in BR signaling. Our own results indicated that these three stress conditions downregulated the expression of putative glycogen synthase kinase family protein. However, in terms of the response of maize response to drought, heat and combined stress conditions, the role of GSKs in BR and ABA signaling needs further investigation.

The expression of protein kinase Kelch repeat/Kelch was elevated by the heat stress and combined drought and heat stress. In rice, Kelch domain containing 10 has been shown

to be involved in oxidative stress-induced cell death (Sekine et al., 2012), and OsFBK12 (an F-box protein containing a Kelch repeat motif) regulated pleiotropic phenotypes and leaf senescence (Chen et al., 2013). Purple acid phosphatase (PAP) family members were involved in extensive aspects of plant development, mineral homeostasis and stress responses (González-Muñoz et al., 2015). In Arabidopsis thaliana, the elimination of AtPAP26 disturbed phosphorus remobilization and delayed leaf senescence (Robinson et al., 2012). In the current study, heat stress and combined stress down-regulated PAP expression. However, the role of protein kinase Kelch repeat and PAPs in maize endurance to heat stress and combined stress remains to be elucidated.

### Proteins Involved in K+, Sugar, and Water Transport

Transporter proteins play important roles in maintaining turgor pressure and regulating water potential, which is vital for plant growth and survival in the stress response. For example, plasma membrane intrinsic proteins (PIPs) are primary channels that mediate the transfer of water and other small molecules across vacuolar and plasma membranes, and are associated with plant tolerance to stress. In this study, drought stress and combined stress up-regulated the expression of aquaporin PIP2-6. Other results have shown that the over-expression of MzPIP2:1 in Arabidopsis enhances plant tolerance to drought (Wang et al., 2015). However, the over-expression of AtPIP1:2 in tobacco has been found to reduce plant tolerance to drought (Aharon et al., 2003).

K <sup>+</sup> is involved in many cellular processes, including enzyme activation, protein synthesis, and osmotic regulation (Anschütz et al., 2014; Demidchik et al., 2014). In Arabidopsis responses to drought, it is essential to regulate the homeostasis of intracellular K <sup>+</sup>. KZM2 has a voltage-gated K<sup>+</sup> channel activity (Shabala and Pottosin, 2014). The study by Büchsenschütz et al. (2005) found that KZM2 in maize epidermis was responsible for stomatal opening. In the present study, both heat stress and combined stress up-regulated the level of KZM2 expression. Taken together, these results indicate that KZM2 enhances plant tolerance to heat stress and combined drought and heat stress by regulating stomatal opening, which helps to release heat by transpiration.

Under abiotic stress conditions, carbohydrates accumulate in plant cells. In Arabidopsis leaves, the carrier protein SWEET17 is a major factor controlling fructose metabolism. The decrease of SWEET17 expression by stress causes fructose to accumulate in leaves (Chardon et al., 2013). SWEET16 is a vacuole-located carrier involved in glucose, fructose, and sucrose transportation. The over-expression of AtSWEET16 has been shown to modify Arabidopsis tolerance to stress (Klemens et al., 2013). In this study, sugar carrier protein C had a significant decrease under these three stress conditions. Taken together, these results indicate that the reduced expression of a sugar transporter may facilitate the accumulation of sugar in leaves in order to increase stress endurance.

#### CONCLUSIONS

Owing in part to climate change, food resources are being challenged by drought, heat, and the combination of these factors. Plants apply different mechanisms to adapt to combined stresses than to adapt to a single stress. Among the drought, heat, and combined drought and heat stress conditions, we found more similar proteins between the heat stress and combined stress conditions. HSPs, especially sHSPs, showed abundant expression under the heat stress and combined stress, and were found to play a role in extensive signaling pathways, suggesting that HSPs play a crucial role in maize tolerance to heat stress and combined stress. Even though similar signaling pathways were found in response to the heat stress and combined stress, relative to the drought stress and heat stress, the combined stress led

#### REFERENCES


to the greater expression of chloroplast proteins, enzymes and stimuli response proteins, which led to the development of more extensive signaling pathways and protein interaction networks. Our results also implied that ethylene-responsive protein and ripening-related proteins, which promote leaf senescence, may also have a potential role in maize endurance to combined drought and heat stress. Therefore, our results could be used to further our understanding of the mechanisms of crop response to combined stresses.

#### AUTHOR CONTRIBUTIONS

XH conceived and designed the research. FZ, YZ, and FT performed the experiments. DZ, HY, and XH analyzed the data. WW and CL contributed reagents/materials/analysis tools. XH and DZ wrote the paper. All authors read and approved the manuscript.

#### ACKNOWLEDGMENTS

This work was supported by the Program for Scientific Innovation Talent for Henan Province (grant no. 154100510005), the Program for Science and Technology Innovation Talents in Universities of Henan Province (grant no. 13HASTIT001), and the National Natural Science Foundation of China (grant no. 31171470).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 01471/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Zhao, Zhang, Zhao, Wang, Yang, Tai, Li and Hu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Comparative Proteomic Analysis of the Response of Maize (Zea mays L.) Leaves to Long Photoperiod Condition

Liuji Wu1, 2 †, Lei Tian1, 2 † , Shunxi Wang1, 2, Jun Zhang<sup>3</sup> , Ping Liu1, 2, Zhiqiang Tian1, 2 , Huimin Zhang1, 2, Haiping Liu<sup>4</sup> and Yanhui Chen1, 2 \*

<sup>1</sup> Henan Agricultural University and Synergetic Innovation Center of Henan Grain Crops, Zhengzhou, China, <sup>2</sup> Key Laboratory of Physiological Ecology and Genetic Improvement of Food Crops in Henan Province, Zhengzhou, China, <sup>3</sup> Food Crops Research Institute, Henan Academy of Agricultural Science, Zhengzhou, China, <sup>4</sup> Department of Biological Science, Michigan Technological University, Michigan, MI, USA

#### Edited by:

Hanjo A. Hellmann, Washington State University, USA

#### Reviewed by:

Francisco Javier Medina, Consejo Superior de Investigaciones Científicas, Spain Letizia Bernardo, Università Cattolica del Sacro Cuore, Italy

> \*Correspondence: Yanhui Chen chy9890@163.com

† These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 22 November 2015 Accepted: 17 May 2016 Published: 02 June 2016

#### Citation:

Wu L, Tian L, Wang S, Zhang J, Liu P, Tian Z, Zhang H, Liu H and Chen Y (2016) Comparative Proteomic Analysis of the Response of Maize (Zea mays L.) Leaves to Long Photoperiod Condition. Front. Plant Sci. 7:752. doi: 10.3389/fpls.2016.00752 Maize (Zea mays L.), an important industrial material and food source, shows an astonishing environmental adaptation. A remarkable feature of its post-domestication adaptation from tropical to temperate environments is adaptation to a long photoperiod (LP). Many photoperiod-related genes have been identified in previous transcriptomics analysis, but proteomics shows less evidence for this mechanism of photoperiod response. In this study, we sampled newly expanded leaves of maize at the three- and six-leaf stages from an LP-sensitive introgression line H496, the donor CML288, LP-insensitive inbred line, and recurrent parent Huangzao4 (HZ4) grown under long days (15 h light and 9 h dark). To characterize the proteomic changes in response to LP, the iTRAQ-labeling method was used to determine the proteome profiles of plants exposed to LP. A total of 943 proteins differentially expressed at the three- and six-leaf stages in HZ4 and H496 were identified. Functional analysis was performed by which the proteins were classified into stress defense, signal transduction, carbohydrate metabolism, protein metabolism, energy production, and transport functional groups using the WEGO online tool. The enriched gene ontology categories among the identified proteins were identified statistically with the Cytoscape plugin ClueGO + Cluepedia. Twenty Gene Ontology terms showed the highest significance, including those associated with protein processing in the endoplasmic reticulum, splicesome, ribosome, glyoxylate, dicarboxylate metabolism, L-malate dehydrogenase activity, and RNA transport. In addition, for subcellular location, all proteins showed significant enrichment of the mitochondrial outer membrane. The sugars producted by photosynthesis in plants are also a pivotal metabolic output in the circadian regulation. The results permit the prediction of several crucial proteins to photoperiod response and provide a foundation for further study of the influence of LP treatments on the circadian response in short-day plants.

Keywords: proteomic analysis, maize leaves, introgression line, iTRAQ, long photoperiod, cytoscape, circadian

### INTRODUCTION

Maize (Zea mays L.) is a key food source and industrial material that has rapidly spread in cultivation since originating in Southern Mexico 6000–10,000 years ago from Balsas teosinte (Zea maysssp. parviglumis; Matsuoka et al., 2002). Balsas teosinte required short-day (SD) conditions for flowering (Emerson, 1924). One remarkable determinant enabling the spread of maize across latitudes was the post-domestication adaptation to changing in daylight hours (Piperno et al., 2009; van Heerwaarden et al., 2011). Under the longer days experienced at higher latitudes, tropical maize cultivars do not flower or show delayed flowering (Betran et al., 2003). Plants integrate signals from endogenous regulatory pathways or the environment to modulate the timing of flowering (Colasanti and Coneva, 2009). In the model plant Arabidopsis thaliana, numerous components associated with the plant circadian clock and photoperiod have been studied to understand the regulation and molecular mechanism of flowering in higher plant (Matsubara et al., 2008; Kumimoto et al., 2010; Lazakis et al., 2011; Knuesting et al., 2015). However, only a small number of genes such as ZCN1, ZCN8, and conzl identified by the Arabidopsis orthologues AtTFL1, AtFT, and AtCO, respectively (Danilevskaya et al., 2008; Matsubara et al., 2008; Miller et al., 2008; Lazakis et al., 2011), have been shown to be involved in the regulation of flowering time and the vegetative to reproductive transition in maize. Recently, numerous quantitative trait loci (QTLs) were affecting flowering date and response to photoperiod were detected, each with a small effect (Buckler et al., 2009). The photoperiod response may be influenced by only a small number of these floweringtime QTLs, including ZmCCT which encodes a CCT domaincontaining protein (Ducrocq et al., 2009; Coles et al., 2010; Hung et al., 2012). Previously, our research group identified CACTAlike transposable elements in ZmCCT, which were shown to attenuate the photoperiod sensitivity and to accelerate the postdomestication spread of maize (Yang L. T. et al., 2013).

The circadian system influences expression of a substantial fraction of the genes in a variety of species because of the diversity of clock outputs. Approximately 10 and 30% of genes are estimated to be regulated by the circadian system in mammals and plant, respectively (Panda et al., 2002; Michael and McClung, 2003; Covington et al., 2008). Circadian rhythms are entrained by environmental signals, such as temperature and light, and by endogenous sugar production by photosynthesis to enable a plant to adapt the local environment (Harmer et al., 2000; Haydon et al., 2013).

Recent studies on photoperiod response have highlighted the emergence of proteomic analysis as a promising tool. To our knowledge, our group is responsible for the only previous proteomic analysis of photoperiod responses in maize, using classical 2-D electrophoresis (2-DE) combined with mass spectrometry (MS; Wang et al., 2015). In that study, however, only a few proteins responsive to long photoperiod (LP) were identified (Wang et al., 2015). And in our study, we also used the bioinformatics tools WEGO (GO annotation) and Cytoscape (v3.0.2) plugin ClueGO + Cluepedia v2.1 (GO-KEGG network) for functional classification and enrichment analysis, and argue that photoperiod response to LP will show a close relationship with protein synthesis, metabolism process, post-transcriptional regulation and mitochondrial outer membrane. None of this is included in the Wang et al. (2015) article. For each functional category, we identified more proteins compared with the Wang et al. (2015) study. Especially for these "circadian" related proteins. Therefore, the current study lays a foundation for future elucidation of the protein network regulatory mechanism underlying the photoperiod response.

Stevia rebaudiana plants grown under long-photoperiod (LP) conditions show increased leaf size, internode length and dry weight, but reduced intervals between successive leaf pairs, compared with plants grown under SD (Metivier, 1979). However, few studies on proteomic fluctuations in response to LP in the maize leaf have been undertaken. To clarify the mechanism involved in alterations of the proteome, in the present study we collected newly expanded third and sixth leaves from the photoperiod-insensitive maize inbred line Huangzao4 (HZ4) and the photoperiod -sensitive inbred line H496 obtained through crossing the recurrent parent of HZ4 with CML288 (non-recurrent parent). A total of 5259 proteins and 14 proteins directly related to the photoperiod were identified by isobaric tags for relative and absolute quantitation (iTRAQ) labeling in response to the LP condition.

#### MATERIALS AND METHODS

#### Plant Materials

The near-isogenic lines H496, which is highly photoperiodsensitive, was derived from a cross between HZ4 (the recurrent parent) and a tropical maize inbred line, CML288, The latter was acquired from the National Maize and Wheat Improvement Center in Mexico, whereas HZ4 is a representative of the Chinese Tangsipingtou heterotic group. Four plants were grown in each 15 cm pots under LP conditions (15/9 h, light/dark; Ku et al., 2011). Newly developed third and sixth leaves were collected for proteomic analysis. All leaf samples were immediately frozen in liquid nitrogen stored at −80◦C until use.

#### Sucrose and Glucose Measurement

Fresh leaf material of HZ4 and H496 sampled at three- and six-leaf stages (15/9 h, light/dark) were separately ground to fine powder with a mortar and pestle in liquid nitrogen. The sucrose and glucose content was determined by enzyme-coupled reactions using the Sucrose/D-Glucose/D-Fructose assay kit (R-Biopharm, Darmstadt, Germany) as described by Thalor et al. (2012). A sample (200 mg) of the powder was immediately boiled with 600µl distilled water for 15 min in a water bath. After centrifugation (16,000 × g, 15 min at 4◦C), 100µl of the supernatant was used for absorbance determination in the sucrose assay by the spectrophotometer (Hitachi U-2900, Hitachi, Tokyo, Japan).

#### Fe Content Measurement

Leaves at the three- and six-leaf stages of the two maize inbreds were collected in three biological replications for Fe concentration analyses. One hundred microgram of the leaves were dried for 2–3 days at 70◦C, then digested with 1 ml of 13 M HNO3 and 1 ml of 8.8 M H2O2 (Wako, Japan) at 220◦C for 20 min using MARS Xpress oven (CEM, USA) as described by Masuda et al. (2008); After digestion, the samples were diluted to 5 ml and analyzed using a SPS1200VR ICPAES (Seiko, Japan).

#### Protein Digestion and iTRAQ Labeling

Digestion of protein was carried out in accordance with the filteraided sample preparation (FASP) protocol used by Wisniewski et al. (2009). Briefly, the method used was as follows. For each sample, 200µg proteins were suspended in 30µl STD buffer (4% SDS, 150 mM Tris-HCl, 100 mM DTT, pH 8.0), incubated in boiling water for 5 min and then cooled to room temperature. The DTT (detergent) and other low-molecularweight components were diluted with 200µl UA buffer (150 mM Tris-HCl, 8 M urea, pH 8.0) and transferred by repeated ultrafiltration (Microcon units, 30 kD). Next, 100µl of 0.05 mol·L-1 iodoacetamide (IAA) was added to the UA buffer to block the reduced cysteine residues. The mixture was incubated in darkness for 20 min. The filters were washed three times with 100µl UA buffer and then twice with 100µl DS buffer (50 mM triethylammonium bicarbonate, pH 8.5). Finally, 2µg trypsin (Promega, Madison, USA) was used to digest the protein suspensions in 40µl DS buffer at 37◦C overnight. The digested peptides were collected as a filtrate. The concentration of peptides was measured by UV light spectral density at 280 nm using an extinction coefficient of 1.1 of 0.1% (g/l) solution, which was calculated based on the frequency of tyrosine and tryptophan in vertebrate proteins.

For labeling, the digested products of the peptide mixture were labeled with 8plex iTRAQ <sup>R</sup> Reagents following the manufacturer's instructions (Applied Biosystems). Briefly, 70µl of ethanol was used to dissolve each iTRAQ reagent and then the solution was combined with respective peptide mixture. The samples were labeled (496-6Y)-113, (496-3Y)-114, (HZ4-6Y)- 115, and (HZ4-3Y)-116 and were multiplexed and vacuum-dried.

### Peptide Fractionation with Strong Cation Exchange (SCX) Chromatography

An AKTA purifier system (GE Healthcare) was used to fractionate iTRAQ-labeled peptides. Reconstitution and acidification of the dried peptide mixture were performed using 2 ml of buffer A [10 mM KH2PO4 in 25% (v/v) acetonitrile, pH 2.7]. The products were loaded onto a polysulfethyl 4.6×100 mm column (5µm, 200 Å; PolyLC Inc., Columbia, MD, USA). A gradient of 0–10% buffer B [10 mM KH2PO4 in 25% (v/v) acetonitrile, 500 mM KCl, pH 2.7] was used for elution of the peptides at a flow rate of 1 ml min-1 for 2 min, 10–20% buffer B for 25 min, 20–45% buffer B for 5 min, and 50–100% buffer B for 5 min. The elution was monitored by absorbance at 214 nm, and fractions were collected at 1-min intervals. All collected fractions (∼30) were finally grouped into 10 pools and desalted on C18 cartridges [EmporeTM SPE cartridges C18 (standard density), bed i.d. 7 mm, volume 3 ml, Sigma]. After concentration by vacuum centrifugation, each fraction was reconstituted in 40µl of 0.1% (v/v) trifluoroacetic acid. Before liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis, all samples should be stored at −80◦C.

#### LC-ESI MS/MS Analysis

Q ExactiveTM mass spectrometer coupled with an EasynLC chromatography system (Proxeon Biosystems, now Thermo Fisher Scientific) were used to perform the following experiments. For nano LC-MS/MS analysis, totally 10µl of each fraction was used. The peptide mixture (5µg) was loaded into the C18 reversed-phase column (15 cm length, 75µm id) packed in-house with RP-C18 5µm resin in buffer A (0.1% formic acid) and separated by buffer B with a linear gradient (0.1% formic acid and 80% acetonitrile) at the flow rate of 250 nl/min controlled by an Intelli Flow Technology controller over 140 min. Data-dependent top 10 method, which dynamically chose the most abundant precursor ions from the survey scan (300–1800 m/z) for HCD fragmentation, was used to acquire the MS data. Predictive automatic gain control (pAGC) was applied to determinate the target value. Dynamic exclusion duration was 60 s. Resolution for survey scans was set to 70,000 at m/z 200, while for HCD spectra, the resolution 17,500 at m/z 200. 30 eV was applied for normalized collision energy and 0.1% was defined for the underfill ratio which specifies the minimum percentage of the target value likely to be reached at maximum fill time. Peptide recognition mode was enabled during the running of the instrument.

### Sequence Database Search and Data Analysis

MASCOT engine (Matrix Science, London, UK; version 2.2) was embedded into the Proteome Discoverer 1.3 (Thermo Electron, San Jose, CA, USA) for searching MS/MS spectra against the decoy database and UniProt Plant database (134,648 sequences, downloaded on May 5, 2013). The following parameters were used for identifying proteins. ±20 ppm is set for peptide mass tolerance, 0.1 Da for MS/MS tolerance, 2 for missed cleavage, enzyme is trypsin, fixed modification: carbamidomethyl (C), iTRAQ4/4plex(K), iTRAQ4/4plex(Nterm), Variable modification: oxidation (M), iTRAQ4plex (Y), 20 ppm for integration window tolerance, 0 for minimum quan value threshold, 2 for fold change threshold for up/down regulation, 100 for maximum allowed fold change, and FDR is no more than 0.05 (Sandberg et al., 2012). For iTRAQ studies, we used confidence scores >1.2-fold, FDR ≥0.05, as the qualification criterion, which corresponded to a peptide confidence level of 95% (Yang Q. et al., 2013).

#### Bioinformatics

As described by Ye et al. (2006), Gene Ontology (http://www. geneontology.org/) and WEGO (http://wego.genomics.org.cn/) online tools were used for functional analysis of the proteins. The statistically enriched gene ontology (GO) categories for the identified protein interactome were determined by Cytoscape (v3.0.2) plugin ClueGO + Cluepedia v2.1.3 (Bindea et al., 2009, 2013). The analysis was carried out using the proteins identified in the three-leaf and six-leaf stage in these two inbred lines. Biological processes, subcellular locations, molecular function and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathways (Kanehisa and Goto, 2000), which were inferred electronic annotation and experimental data, were all in the identified GO categories. A minimum level of 5 and a maximum level of 11 were set as the GO level interval with a minimum of two genes per category. And a right-sided hypergeometric test for enrichment analysis was elicited applying against the ClueGO Z. Mays reference genome.

#### Post-hoc Test

To verify the main variable contributing to the differences, a twoway analysis of variance with post-hoc test was performed using SAS software (SAS Institute Inc., Cary, NC, USA).

#### RESULTS AND DISCUSSION

### Phenotypes and Growth Parameters between HZ4 and H496 in Maize

To examine the protein changes response to the photoperiod in maize leaves, plant phenotypes were periodically observed in the two inbred lines HZ4 and H496 under LP at the three- and sixleaf stages, and the individual samples were collected. Under the LP condition, plants of the H496 line were considerably taller than those of the HZ4 line. HZ4 plants showed less photoperiod sensitivity than H496 plants, in which flowering was delayed by 1 week (**Table 1**). The two lines showed similar leaf and shoot apex phenotypes at each of the three- and six-leaf stages (**Figure 1**).

Previously, the number of leaves and morphologies of the shoot apical meristem were used to indicate the inductive phase changes of photoperiod sensitivity in maize. This showed that the juvenile vegetative stage was completed between the fourand five-leaf stages in HZ4 and CML288 under LP condition (Wu et al., 2008). In the present study, we observed that the shoot apical meristem was elongated in the six-leaf stage in both HZ4 and H496 inbred lines (**Figures 1E–H**), which proved that plants were in different developmental phases at the three- and six-leaf stages. Thus, it seems that examination of photoperiodsensitive phenotypic traits is significant during improvement of maize germplasm.

#### iTRAQ Analysis of Protein Expression in Response to LP Condition

In this study, 23,767 unique peptides were analyzed. A total of 5259 proteins were identified by MS/MS (**Table S 1**). The peptides of the identified proteins are listed in **Tables S 1**, **S 2**. According to the criteria for recognition of differentially expressed proteins (fold change ratio >1.2 and p < 0.05), 943 proteins differentially expressed between H496 and HZ4 were

TABLE 1 | Phenotypes of the maize inbred line HZ4 and near-isogenic line H496 grown under long-photoperiod conditions.


identified, of which 185 proteins were differentially expressed at both developmental stages (**Table S 3**), 398 proteins showed differential expression at the three-leaf stage (**Table S 4**), and 360 proteins were only differentially expressed at the six-leaf stage (**Figure 2A**, **Table S 5**). The results showed that the difference between the inbred lines lead to the changes of the proteins. Of the differentially expressed proteins, 278 upregulated and 305 downregulated proteins were identified at the three-leaf

FIGURE 1 | Morphology of the shoot apex and leaves at the three- and six-leaf stages of maize inbred line HZ4 and H496 grown under long-photoperiod conditions. (A–D) Shoot apex of HZ4 at the three-leaf stage (A) and six-leaf stage (B). Shoot apex of H496 at the three-leaf stage (C) and six-leaf stage (D). (E–H) Leaves of HZ4 at the three-leaf stage (E), and six-leaf stage (F); shoot apex of H496 at three-leaf stage (G), and six-leaf stage (H).

stage (**Figure 2B**, **Tables S 3**, **S4**), and 223 upregulated and 322 downregulated proteins were detected at the six-leaf stage (**Figure 2B**, **Tables S 3**, **S5**).

Analysis of variance confirmed that there were significant differences in protein expression between the inbred lines, but there was no significant difference between the two developmental stages (**Table S 6**). Previously, our group used the H496-10 line which was produced after one less generation of back-crossing with HZ4 than H496 via gel-based proteomic approach to provide novel insights into the influences of longphotoperiod treatments on short-day plants (Wang et al., 2015), but there are significant differences and considerable novelty in this study, we choosed three- and six-leaf stage as two distinct phases in order to verify what proteins change in expression before and after the onset of the induction phase according to Wu et al. (2008). And 943 proteins differentially expressed were identified, while only 22 differentially expressed proteins between HZ4 and H496-10 (Wang et al., 2015). Clearly, this current report by iTRAQ method identified many additional proteins and presents further evidence with which to understand the photoperiod response in maize.

### Functional Characterization of Protein Interaction Network

We analyzed the GO annotation of the 943 proteins that were expressed differentially at the three- and six-leaf stages of H496 compared with HZ4 (**Tables S 3**–**S 5**) to gain insights into the functions of the proteins and the mechanism involved in the photoperiod pathway. The WEGO tool was used to plot the distribution of GO annotations (**Figure 3**). The differentially expressed proteins were grouped into three hierarchically structured GO terms, namely biological process, cellular component, and molecular function. The differentially identified proteins were subcategorized into 16 main hierarchically structured GO classifications including 4 biological processes, 10 cellular components, and 2 molecular functions (**Figure 3**). Specifically, "metabolic process" and "cellular process" were highly represented in "biological process"; "cell part", "cell" and "intracellular" were incorporated in "cellular component"; and "binding," "catalytic activity" were represented in "molecular function" (**Figure 3**). This analysis indicated that the identified proteins involved in these GO categories may play the most important roles in regulation of the photoperiod response to LP.

Based on molecular functions, biological processes and KEGG pathways (Kanehisa and Goto, 2000), we also generated a GO annotation and KEGG network (KEGG-GO; Reference Genome Group of the Gene Ontology 2009) using the Cytoscape plugin Cluego + Cluepedia (Bindea et al., 2009, 2013). Twenty terms were connected by 38 edges with the kappa scores, and showed considerable enrichment (p < 0.05) in the identified protein interactome (**Figure 4**). The most significant terms comprised those associated with protein processing in endoplasmic reticulum, splicesome, ribosome, glyoxylate, dicarboxylate metabolism, L-malate dehydrogenase activity, and RNA transport (**Figure 5A**). With regard to subcellular location, all proteins showed significant enrichment in the mitochondrial outer membrane (**Figure 5B**).

#### Protein Synthesis

Protein turnover, which represents the balance between protein synthesis and degradation, is one of the many forms of regulation that is employed to achieve a unified cellular response (Reinbothe et al., 2010). In the present study, the most significant function enrichment pathway terms were ribosome, protein processing in endoplasmic reticulum, and RNA transport, which are involved in protein synthesis (**Figure 5A**). Missra et al. (2015) calculated the rate of protein synthesis by multiplying transcript abundance by translation state in Arabidopsis to show that high translation rates of TOC1 and LUX mRNAs at night may allow many related proteins to continue to repress transcription of the morning genes CCA1 and LHY, and of day genes such as GI and PRR9, which argued that it is plausible that differences in the waveform of protein synthesis rates may help to fine-tune circadian gene function (Missra et al., 2015). Recently, it has been reported that DNA replication during the cell cycle causes protein synthesis rates to show sharp, periodic jumps that can entrain the circadian clock in the cyanobacterium Synechococcus elongates (Paijmans et al., 2016). The present results provide additional evidence that protein synthesis has an important role in circadian regulations.

#### Metabolism Process

As **Figure 3** shown, Metabolism process constituted a high percentage (>70%) of the GO terms and Glyoxylate and dicarboxylate metabolism also elucidated significant enrichment in the KEGG-GO network (**Figures 5A,B**). These results indicated that amino acid metabolism may show distinct differences between HZ4 and H496 in response to LP. Connections between circadian clocks and carbon metabolism

has been reported previously by Müller et al. (2014), and recently quantitative circadian phosphoproteomic analysis of Arabidopsis has revealed extensive clock control of key components in physiological, metabolic and signaling pathways and these findings showed new interaction networks that confer previously uncharacterized rhythms onto metabolism and physiology (Choudhary et al., 2015).

#### Post-Transcriptional Regulation

In 2011, Staiger D. and Köste T. have reviewed that posttranscriptional control in the circadian system of modern organisms, Drosophila, mammals, Neurospora, Chlamydomonas and Arabidopsis (Staiger and Köster, 2011). Notably, the next year alternative splicing (AS), as one type of post-transcriptional regulation, was reported as a way of linking the circadian clock to temperature response in Arabidopsis by AS of circadian gene CCA1(Park et al., 2012). Recently, Papasaikas and collaborators observed that GEMIN2, the only component of the SMN complex that is conserved from yeast to humans, controls the pace of the circadian clock under standard growth conditions in Arabidopsis by controlling the AS of TOC1 and other core clock genes (Papasaikas et al., 2015).

#### Mitochondrial Outer Membrane

In a study of the mitochondrial outer membrane in Arabidopsis, Duncan et al. (2011) developed a statistically rigorous quantitative proteomic workflow to confidently determine components of the outer mitochondrial membrane proteome of Arabidopsis. The proteins identified range from plantspecific proteins with unknown functions to proteins that have putative functions in mitochondrial signaling, morphology, and defense responses. In the present study, the mitochondrial outer membrane, as the only significantly enriched subcellular location, may be an important and novel location for proteins associated with photoperiod response (**Figure 5B**).

### Protein Species Expressed Specifically Related to Photosynthesis in Three-Leaf Stage or Six-Leaf Stage in Response to LP between HZ4 and H496

Under LP condition, some differentially expressed proteins identified in H496 compared with HZ4 exhibited similar proteome patterns at both the three- and six-leaf developmental stages (**Figure 2A**)., Such proteins included the Gibberellin receptor GID1L2 (B6TKC8), pollen-specific protein (B6TIS3), cryptochrome 2 (C9DQ39), and cyptochrome P450 super family protein (B4FQH2) (**Table S 3**). The majority of proteins identified in the two inbred lines differed between the threeleaf and six-leaf stages (**Figure 2A**, **Tables S 4**, **S 5**). In addition, photoperiod altered the expression of all differentially expressed proteins involved in carbon and energy metabolism, including

#### FIGURE 5 | Continued

Cytoscape plug-in ClueGo + Cluepedia to identify statistically enriched GO categories compared with the ClueGO maize reference genome. (A) GO categories searched include biological processes, molecular function, KEGG pathways, and (B) cell component. Nodes represent a specific GO term and are grouped based on the similarity of their associated proteins. Each node represents a single GO term and is color-coded based on enrichment significance (pV = p-value). Node size indicates the number of proteins mapped to each term. Edge thickness represents the calculated kappa score based on the number of proteins shared between terms. Functional groups are labeled by the most significant term in the group. Arrow indicates positive regulation.

ribosome, L-malate dehydrogenase activity, glyoxylate and dicarboxylate metabolism (**Figure 5A**). Some specially expressed proteins play a specific role in photosynthesis at both the threeand six-leaf stages (**Figure 2A**, **Tables 2, 3**). The three main classes were carbohydrates, amino acids, and lipids (**Tables 2, 3**), the protein amounts of these three metabolites (34/56, 60.7%) were higher in H496 compared with HZ4 at the three-leaf stage; but at the six-leaf stage only half of these proteins showed elevated expression. This result indicated that photosynthesis at the two developmental stages of HZ4 and H496 may show different regulatory mechanism responses to the LP condition.

#### Carbohydrate

The proteins involved in carbon assimilation showed significant changes in abundance under LP. A previous report indicated that enzymes functioning during the reduction phased of the Calvin-Benson cycle accumulated to higher levels shoot tips under LP, whereas the level of enzymes involved in carboxylation and regeneration phases was increased in shoot tips under SD condition (Victor et al., 2010). In the present study, triose phosphate isomerase (B6AMV7), an important enzyme in the Calvin-Benson cycle, was less abundant in LP leaves (**Table 3**). This result contradicts previous observations and may be owing to differences between tissues, or the mechanisms of the response to LP in leaf may be more complex and involve additional regulators than compared with that in the shoot tips.

Potentially increased availability of carbohydrate under LP may be the reason for elevated accumulation of enzymes responsible for glycolysis, such as malate dehydrogenase (B4FZU8, B4FG53, B4FRJ1, F6MFD6; **Table 3**), which are involved in the pathway following glycolysis and were also more abundant in leaves. In Arabidopsis, the activities of enzymes involved in the glycolysis pathway were decreased in response to a shortened photoperiod, whereas activity of enzymes participating in photosynthesis and starch synthesis remained high.

As a diurnally regulated carbohydrate, sucrose content increases during light conditions and decreases during dark conditions, consistent with previously reports for other plants, such as potato (Urbanczyk-Wochniak et al., 2005). Glucose-6-phosphate, which is responsible for sucrose biosynthesis as well as degradation, exhibited a similar pattern to that of sucrose (Urbanczyk-Wochniak et al., 2005). Hoffman et al. (2010) reported that diurnal fluctuations were regulated by several Krebs-cycle intermediates in pool sizes. In our study, we measured the sucrose and glucose contents in the leaf of HZ4 and H496 at the three- and six- leaf stages. The sucrose and glucose contents in H496 were slightly lower than those of HZ4 at the three-leaf stage, but higher at the six-leaf stage (**Figures S 1A, B**). This finding indicated that sucrose and glucose showed homeostatic changes in response to LP in the development of the two lines. Malate dehydrogenase showed an activated pattern at the three-leaf stage in H496, with an increased level compared to HZ4. Significant differences also observed for several proteins involved the metabolites between the two species at both stages (**Tables 2, 3**). These results provide new evidence to further verify carbohydrate will mediated the circadian response.

#### Chloroplast Proteins

Adequate light harvesting for photosynthesis is closely related with the abundance of chloroplast proteins, such as chlorophyll a/b binding protein, which is responsible for energy transfer in the reactive center in photosystem II (Kovács et al., 2006). This protein is responsible for balancing the distribution of excitation energy between photosystems: I and II (Kovács et al., 2006). Interestingly, one of these genes encoding chloroplast a/b binding protein was homologous to known genes that are responsible for the floral transition or morphology and the circadian rhythm photoperiod response in maize, rice and Arabidopsis (Coles et al., 2010). In the present study, expression of two chlorophyll a/b binding proteins, K7TWD9 and B4FV94, was increased in H496. Thus, these two proteins are predicted to be involved in the circadian rhythm response, but confirmation requires further investigation.

#### Ribosomal Proteins

We detected 27 ribosomal proteins, of which 16 proteins were upregulated and 11 proteins were downregulated (**Tables 2, 3**). Previously, the ribosomal protein gene L34 (rpL34), which encodes a cytoplasmic ribosomal protein with high homology to the rat 60S r-protein, was isolated from a genomic library of tobacco (Nicotiana tabacum cv. Xanthi-nc), and histochemical GUS staining showed that rpL34 promoter activity was high in actively growing tissues, including various meristems, floral organs and developing fruits (Dai et al., 1996). In the present study, the ribosomal protein L34 was downregulated in HZ4 only at the six-leaf stage, which is the stage at which shoot apex morphology changes (Wu et al., 2008). Thus, the early flowering habit of HZ4 may be caused by activation of the ribosomal protein. Conversely, translation, especially the production of ribosomal proteins, is positively correlated with the abundance of phosphorylated S6 protein (Williams et al., 2003; Turck et al., 2004). The phosphorylation of L29-1, a 60S ribosomal protein, is enhanced under moderate "day time," and the possibility of diurnal regulation of translation in plants is indicated by differential phosphorylation of at least three ribosomal proteins: the 40S ribosomal proteins S6-1 and S6-2, and the 60S ribosomal protein L29-1 (Turkina et al., 2011).


#### TABLE 2 | Functional classification of identified proteins significantly differentially expressed at the three-leaf stage of maize HZ4 and H496 plants exposed to long-photoperiod conditions.


Annotations were obtained from the UniProtKB/Swiss-Prot databases (http://www.expasy.org/).

In the current study, a higher number of ribosomal proteins were upregulated in H496, and the 60S ribosomal protein L29 showed a higher expression level in H496, which is consistent with the above-mentioned report (Turkina et al., 2011). However, information on the exact mechanisms to explain increased protein production in the light phase of the photoperiod is extremely limited. The present findings shed some light on this conundrum by indicating that a portion of the enhanced protein synthesis may result from diurnal regulation of translation by complex combinatorial phosphorylation of ribosomal proteins. Overall, whether the ribosomal proteins are upregulated or downregulated, it would be an important cue in the regulation of flowering and photoperiod response, but the molecular function and regulatory mechanism for each ribosomal protein are poorly known and require further investigation in the future.

#### Expression Pattern of Iron Metabolism-Related Proteins under LP Condition

Anti-oxidative molecules, such as ferritin, are essential to detoxify reactive oxygen species or buffer irons to prevent oxidative stress (Ravet et al., 2009). Iron is a critical component for the function of many photosynthetic proteins, and iron deficiency causes an extended free-running period of rhythm changes and increases the production of reactive oxygen species (Salome et al., 2013). Thus, a higher level of ferritin 1 may correspond to enhanced detoxifying process of reactive oxygen species and distinct reactions to LP treatments. In the present research, the increase in accumulation of ferritin 1 (K7U2L3) of inbred line H496 leaves at six-leaf stage was higher than that in HZ4 under LP (**Figure S1 C**), suggesting that the strength of photosynthesis and the production of photosynthetic protein were higher in H496 compared with HZ4.

### Expression Pattern of Circadian-Associated Proteins under LP Condition

Sorts of categories of circadian-associated proteins have been identified, and the expression patterns of 14 circadian associated proteins were examined (**Figure 4**). We found that eight proteins were upregulated in HZ4 in three-leaf stage, and five in sixleaf stage. But compared with that in H496, many proteins were elevated in HZ4 of three-leaf (11/14, 78.57%) and sixleaf stage (9/14, 64.28%), except that Q9ZR52 (CK2 alpha) were downregulated at the three-leaf stage, as well as C0P8K7 (AtHXK1), B6U4K6 (AtHXK1) and B4F864 (NF-YB12; NF-YB13) in the six-leaf stage. Interestingly, the amount of B4FBL9 (ELF5) decreased at both stages while B4FVS0 (ATFYPP3) and Q1A5Y4 (PHYB; PHYD) have no change separately in three-leaf and six-leaf stage.

Post-translational regulation of CONSTANS (CO) protein is another key element of the photoperiodic induction of FLOWERING LOCUS T (FT) transcription (Möglich et al., 2010). Phytochrome is an important regulator coordinating downstream signaling components, and many studies have focused on elucidating novel components involved in light signal transduction (Paul and Khurana, 2008). The combination of CO with NF-Y transcription factors activates FT during floral initiation, which is dependent on photoperiod (Kumimoto et al., 2010). An ELF5 (B4FBL9) mutation, elf5, partially suppresses the photoperiod pathway and causes early flowering under SD, suggesting that ELF5 controls flowering independent of FLOWERING LOCUS C (FLC), a floral repressor upon which many of the flowering pathways converge (Noh et al., 2004). Moreover, ELF4 regulates the access of GIGANTEA (GI) to chromatin by sequestering GI from the nucleoplasm into subnuclear bodies preferentially during the night, thus restricting its ability to bind to the CO promoter (Kim et al., 2013). Recently, the FPF1 (B6TP05) class genes have been explored, which may act as a regulator of flowering and the formation of wood in poplar (Hoenicka et al., 2012). To attain synchrony with day and night, the clock is entrained via the red/far-red-absorbing PHYTOCHROMES (PHYA-PHYE), the blue light-absorbing CRYPTOCHROMES (CRY1 and CRY2), and the LOV (LIGHT, OXYGEN, VOLTAGE) domain proteins ZEITLUPE (ZTL), FLAVIN BINDING, KELCH REPEAT, F-BOX1 (FKF1), and LOV KELCH PROTEIN 2 (LKP2) (Devlin, 2002; Fankhauser and Staiger, 2002) (**Figure 6**).


TABLE 3 | Functional classification of identified proteins significantly differentially expressed at the six-leaf stage in maize HZ4 and H496 plants exposed to long-photoperiod conditions.

Annotations were obtained from the UniProtKB/Swiss-Prot databases (http://www.expasy.org/).

Some additional proteins related to plant circadian rhythms have been identified. overexpression of Arabidopsis hexokinases (AtHXK1, C0P8K7, and B6U4K6) in tomato plants may reduce photosynthesis, inhibit growth and accelerate senescence (Dai et al., 1999), These results indicate that the activity of endogenous hexokinase is not a factor limiting growth rate, but functions to regulate photosynthesis in photosynthetic tissues. Overexpression of AtHXK1 in tomato plants also reduced the chlorophyll content. From this result, we assume that HXK, as a sugar phosphorylation enzyme, is a negative regulator of photosynthesis. The study by Miao et al. (2013) reinforces and extends the argument that the promoted biosynthesis of aliphatic glucosinolate by glucose is involved in HXK1 and/or RGS1-mediated signaling through the transcription factors MYB29, MYB28 and ABI5. In a previous study we demonstrated that in transgenic Arabidopsis plants, ZmHd6, encoding a protein similar to the Arabidopsis of CASEIN KINASE2 alpha subunit (CK2 alpha, Q9ZR52), affected the

to activate FT expression. NF-Y complex enhances the binding of CO protein to the FT promoter. CIB1 is activated by blue light absorbed by CRY2 and stabilized by blue light absorbed by ZTL. CIB1 directly activates the expression of FT in the afternoon. These proteins prevent flowering under unfavorable conditions, such as short days (SD). Under SD, the expression peaks of FKF1 and GI do not coincide. In the absence of the FKF–GI complex, CO expression is continuously suppressed by CDF proteins during the day.

flowering time through the photoperiodic pathway in maize (Ku et al., 2011). PSEUDO-RESPONSE REGULATOR 7 (PRR7), which is considered a "morning-expressed" gene, was isolated recently (Haydon et al., 2013). By inhibiting photosynthesis, the authors described that endogenous fluctuations in sugar levels supplied feedbacks at metabolic level to circadian oscillator via PRR7. In addition, ppr7 mutants are insensitive to the oscillations of sucrose levels during circadian rhythms. Consequently, in Arabidopsis, robust circadian rhythms are stringently maintained by photosynthesis, demonstrating that the circadian clock is regulated by metabolism to a large extent (Haydon et al., 2013).

#### CONCLUSIONS

In this study, 5259 proteins were detected in maize leaves in the inbred lines HZ4 and H496. On the basis of MS/MS identification, 943 proteins were expressed differentially between HZ4 and H496, and those proteins were commonly shared by the newly expanded leaves from three- and six-leaf stages. Fourteen circadian associated proteins were also examined. The protein expression patterns of the inbred lines differed significantly even though the two lines share a similar genetic background. The proteomic changes in the maize leaf induced by LP treatment were highly function-specific, such as endoplasmic reticulum, splicesome, ribosome, glyoxylate, dicarboxylate metabolism, L-malate dehydrogenase activity, and RNA transport. The protein species differentially expressed between HZ4 and H496 were associated with photosynthesis including carbohydrate, chloroplast and ribosomal proteins at the three- or six-leaf stages in response to LP. To adapt to the outside environment, the phase of rhythms are adjusted in response to environmental signals, such light and external sugar supplement. The regulation patterns of light and circadian-associated protein under LP condition are discussed. The iron metabolism-related proteins and circadian-associated protein, such as K7U2L3, C0P8K7, and Q9ZR52, may mediate the photoperiodic pathway. The results offer novel insights into the influence of LP and provide additional information on the mechanism of circadian response in short-day plants at the proteomic level.

#### AUTHOR CONTRIBUTIONS

YC, LW, and LT conceived and designed the experiments. LW, LT, JZ, SW, PL, HZ, and HL Performed the experiments. LW and JZ Analyzed the data. YC and SW contributed reagents, materials and analysis tools. LW, YC, and LT wrote the manuscript.

#### ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (no. 31471503), the National Basic Research Program of China (973 Program, no. 2011CB111500), and the China Postdoctoral Science Foundation (no. 20100470993).

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 00752

Figure S 1 | Accumulation content of sucrose (A), glucose(B) and Fe(C) between HZ4 and H496 in three- and six-leaf stages. Two-Way ANOVA followed by Student-Neuman-Keuls post-hoc test (∗∗p < 0.01,∗p < 0.05).

Table S 1 | All proteins were identified by MS/MS in 3-leaf stage and 6-leaf stage of HZ4 and H496 under LP condition.

Table S 2 | All peptides of identified proteins in three- and six-leaf stages of HZ4 and H496 under LP condition were identified by MS/MS.

Table S 3 | Differentially expressed protein in H496 compared to HZ4 in both two stages.

Table S 4 | Different expressed specially proteins in H496 compared to HZ4 in 3-leaf stage.

Table S 5 | Different expressed specially protein in H496 compared to HZ4 in 6-leaf stage.

Table S 6 | Two-way analysis of variance of proteins differentially expressed at the three- and six-leaf stages in HZ4 and H496 grown underLP conditions. IL, inbred lines; DS, developmental stages.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Wu, Tian, Wang, Zhang, Liu, Tian, Zhang, Liu and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Overexpression of Glycolate Oxidase Confers Improved Photosynthesis under High Light and High Temperature in Rice

#### Li-Li Cui<sup>1</sup>† , Yu-sheng Lu<sup>1</sup>† , Yong Li<sup>1</sup>† , Chengwei Yang<sup>2</sup> and Xin-Xiang Peng<sup>1</sup> \*

<sup>1</sup> State Key Laboratory for Conservation and Utilization of Subtropical Agro-Bioresources, College of Life Sciences, South China Agricultural University, Guangzhou, China, <sup>2</sup> College of Life Sciences, South China Normal University, Guangzhou, China

While glycolate oxidase (GLO) is well known as a key enzyme for the photorespiratory metabolism in plants, its physiological function and mechanism remains to be further clarified. Our previous studies have shown that suppression of GLO in rice leads to stunted growth and inhibited photosynthesis (Pn) which is positively and linearly correlated with decreased GLO activities. It is, therefore, of interest to further understand whether Pn can be improved when GLO is up-regulated? In this study, four independent overexpression rice lines, with gradient increases in GLO activity, were generated and functionally analyzed. Phenotypic observations showed that the growth could be improved when GLO activities were increased by 60 or 100%, whereas reduced growth was noticed when the activity was further increased by 150 or 210%. As compared with WT plants, all the overexpression plants exhibited significantly improved Pn under conditions of high light and high temperature, but not under normal conditions. In addition, the overexpression plants were more resistant to the MV-induced photooxidative stress. It was further demonstrated that the antioxidant enzymes, and the antioxidant metabolite glutathione was not significantly altered in the overexpression plants. In contrast, H2O<sup>2</sup> and salicylic acid (SA) were correspondingly induced upon the GLO overexpression. Taken together, the results suggest that GLO may play an important role for plants to cope with high light and high temperature, and that H2O<sup>2</sup> and SA may serve as signaling molecules to trigger stress defense responses but antioxidant reactions appear not to be involved in the defense.

Keywords: glycolate oxidase, photosynthesis, hydrogen peroxide (H2O2), salicylic acid (SA), rice (Oryza sativa L.)

## INTRODUCTION

Photorespiration (PR) is the second highest metabolite flux only next to photosynthesis (Pn) in C3 plants, with flux rates amounting to 25–30% of Pn (Sharkey, 1988; Peterhansel and Maurino, 2011). The rate can be even higher under stress conditions such as high temperature, high light and CO<sup>2</sup> or water deficit (Foyer et al., 2009; Peterhansel and Maurino, 2011). PR is also considered as a major source for hydrogen peroxide (H2O2) in plants, likely accounting for more than 70% of total cellular H2O<sup>2</sup> in photosynthetic leaves of C3 plants (Noctor et al., 2002; Foyer et al., 2009; Peterhansel and Maurino, 2011). Cellular H2O<sup>2</sup> is an important reactive oxygen species (ROS),

#### Edited by:

Hanjo A. Hellmann, Washington State University, USA

#### Reviewed by:

Wei Huang, Kunming Institute of Botany, China Congming Lu, Institute of Botany, Chinese Academy of Sciences, China

> \*Correspondence: Xin-Xiang Peng xpeng@scau.edu.cn †These authors have contributed

### equally to this work. Specialty section:

This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science

Received: 28 April 2016 Accepted: 20 July 2016 Published: 04 August 2016

#### Citation:

Cui L-L, Lu Y -s, Li Y, Yang C and Peng X -X (2016) Overexpression of Glycolate Oxidase Confers Improved Photosynthesis under High Light and High Temperature in Rice. Front. Plant Sci. 7:1165. doi: 10.3389/fpls.2016.01165

which function as a signaling molecule to regulate various physiological and defense processes. While different sources for H2O<sup>2</sup> have been reported in plants, the peroxisomal H2O<sup>2</sup> has recently received increasing attention and shown to play important roles in the programmed cell death (PCD) and biotic defense responses (Chaouch and Noctor, 2010; Sewelam et al., 2014). The peroxisomal H2O<sup>2</sup> is mainly contributed by the glycolate oxidation reaction catalyzed by glycolate oxidase (GLO) (Noctor et al., 2002; Foyer et al., 2009). As a result of this, physiological functions of GLO are often considered to link with H2O<sup>2</sup> signaling (Chamnongpol et al., 1998; Peterhansel and Maurino, 2011). GLO could be induced in response to various environmental stresses, as noticed in Vigna, pea and tobacco (Mukherjee and Choudhuri, 1983; Mulligan et al., 1983; Mittler and Zilinskas, 1994; Rizhsky et al., 2002). GLO was also implicated in plant resistance to pathogens (Mukherjee and Choudhuri, 1983; Taler et al., 2004; Rojas et al., 2012; Gilbert and Wolpert, 2013). Taler et al. (2004) identified "enzymatic resistance" genes in melon and suggested that the enhanced expression of the peroxisomal serine/glyoxylate aminotransferase (SGAT) correlated with higher GLO activity which was proposed to play a role in the resistance to Psilocybe cubensis by greater production of H2O<sup>2</sup> (Taler et al., 2004). It is more recently demonstrated that GLO is an alternative source for the production of H2O<sup>2</sup> during both gene-for-gene and non-host resistance in Nicotiana benthamiana and Arabidopsis (Rojas et al., 2012; Gilbert and Wolpert, 2013).

In addition, GLO has been frequently implicated to markedly affect Pn, mainly through studies using mutants or RNAi transgenic plants (Yamaguchi and Nishimura, 2000; Xu et al., 2009; Zelitch et al., 2009; Lu et al., 2014). Interestingly, all these studies consistently found that GLO-deficient C3 plants, or even C4 maize, displayed typical PR phenotypes, i.e., the plants are lethal or stunted in air while normal under high CO2. This phenotype is similar to what was observed in mutants with defects of the other photorespiratory enzymes (Somerville, 2001; Boldt et al., 2005; Timm and Bauwe, 2013). The PR phenotype in the C4 maize glo mutant may implicate that either the photorespiratory pathway is equally important in C4 plants as in C3 plants (Zelitch et al., 2009), or that GLO plays a second essential, yet unidentified, role in plants, as once proposed by Somerville and Ogren (1982). More intriguingly, our previous work has shown that suppression of GLO led to inhibited Pn, which was positively and linearly correlated with the decreased GLO activities (Xu et al., 2009). A few studies have reported that increased levels of photorespiratory enzymes in plants improved Pn or even growth parameters (Timm et al., 2012, 2015, 2016). So it is of curiosity to further know if Pn can be improved when GLO is up-regulated? In this study, various GLO overexpression rice lines, with gradient increases in activity, were generated in order to address the above question. Further functional analyses on these plants indicate that GLO may play an important role for plants to cope with high light and high temperature, and that H2O<sup>2</sup> and salicylic acid (SA) may serve as signaling molecules to trigger stress defense responses but antioxidant reactions appear not to be involved in the defense.

## MATERIALS AND METHODS

### Growth Conditions and Treatments

Pre-germinated rice seeds and transgenic plants were normally grown in Kimura B complete nutrient solution (Yoshida et al., 1976) under natural conditions [average temperature of 30– 35/23–26◦C (day/night), photosynthetically active radiation 600– 1500 µmol m−<sup>2</sup> s −1 and photoperiod of 12 h day/12 h night]. The solution was adjusted to pH of 4.8–5.0 and was renewed once in a week. Various treatments are specified in the corresponding figure legends.

### Construction of the GLO-Overexpression Transgenic Rice Lines

Rice (Oryza sativa L. cv. Zhonghua 11) was used for constructing transgenic lines in this study. The complete cDNA sequences of OsGLO1 (Os03g0786100) or OsGLO4 (Os07g0152900) were amplified by RT-PCR, then the sequence was inserted into an overexpression vector named pYLox.5. PCR with specific primers and cutting with restriction enzymes proved that the target fragment had been correctly ligated. DNA sequencing finally confirmed the correct orientation and 100% cDNA identity to that reported in the GeneBank. The constructed vectors were then transformed into rice callus via Agrobacterium-mediated infection (strain EHA105). T<sup>0</sup> lines were analyzed by Southern blot, and T1 seeds with a single T-DNA insertion were grown to produce T2 seeds. Homozygous lines were finally obtained with hygromycin-resistance screen.

### Transcript Analysis, Enzyme Activity and Metabolite Assays

#### Semi-quantitative and Real-Time PCR

Total RNA was isolated using TRIZOL reagent. The isolated total RNA was then further treated with DNase I and used as a template for first-strand cDNA synthesis using ReverTra Ace (Toyobo, Osaka, Japan) with random hexamers according to the manufacturer's instructions. For semi-quantitative RT-PCR analysis, the optimal number of PCR cycles was first tested gene by gene. The PCR products were separated on 1% (w/v) agarose gels and visualized by Goldview staining. For real-time quantitative RT-PCR, the PCR reaction consisted of 10 µL of 2 × SYBR Green PCR Master Mix (Toyobo), 200 nM primers, and 2 µL of 1:40-diluted template cDNA in a total volume of 20 µL. No template controls were set for each primer pair. The analysis was conducted by a DNA Engine Option 2 Real-Time PCR Detection system and Opticon Monitor software (Bio-Rad, USA).

#### Enzyme Activity Assays

Glycolate oxidase activity was assayed according to Hall et al. (1985) with some modifications (Xu et al., 2009). Superoxide dismutase (SOD) activity was assayed by monitoring the inhibition of the photochemical reduction of nitroblue tetrazolium (NBT) according to the method of Beauchamp and Fridovich (1971), Catalase (CAT) activity was determined by following the consumption of H2O<sup>2</sup> (extinction coefficient

43.6 M−<sup>1</sup> cm−<sup>1</sup> ) at 240 nm for 1 min (Aebi, 1984). The crude extract for guaiacol peroxidase (POD) measurements was isolated according to Polle et al. (1994). Ascorbate peroxidase (APX) activity was determined in the soluble fraction and in the chloroplast membrane fraction in 2 mL reaction mixture containing 50 mM potassium phosphate (pH 7.0), 0.5 mM ascorbate (extinction coefficient 2.8 mM−<sup>1</sup> cm−<sup>1</sup> ), 0.1 mM H2O2, and leaf extract causing a linear decrease in absorbance at 290 nm for 1 min (Nakano and Asada, 1981). Protein concentration was determined according to Bradford (1976).

#### MV Treatment

The youngest fully expanded leaves were detached and treated with 6 µM Methyl viologen (MV, N, N 0 -dimethyl-4, 4<sup>0</sup> bipyridinium dichloride) at 30◦C under continuous illumination (100 µmol m−<sup>2</sup> s −1 ) for 0, 3, 6, 9, 12 h to induce photooxidative stress (Kim and Lee, 2002).

#### Metabolite Assays

Glutathione (GSH) and glutathione disulfide (GSSG) were determined according to Rahman et al. (2006). SA was measured according to Meuwly and Métraux (1993). SA was quantified fluorimetrically (G1321B scanning fluorescence detector, Agilent, USA), with excitation at 305 nm and emission at 407 nm. Hydrogen peroxide (H2O2) production was detected by staining with a freshly prepared 3, 3<sup>0</sup> -diaminobenzidine (DAB) solution (1 mg/ml, pH 3.8) for 2 h in light at 30◦C. The experiment was terminated by boiling the leaves in ethanol for 30 min (Thordal Christensen et al., 1997).

#### Gas Exchange Measurements and Chlorophyll Fluorescence analysis

Gas exchange characteristics including net photosynthetic rate (Pn), stomatal conductance (Gs) and internal CO<sup>2</sup> concentration (Ci) were analyzed in situ using a portable Pn system (LI-6400, LI-COR). The plants were grown in normal natural condition or in an environment-controlled growth chamber, and the youngest fully expanded leaves were used to determine the photosynthetic parameters. Measurements were performed in the morning (10:00–12:00), unless specified elsewhere. The other conditions were set as follows: leaf temperature 30◦C, humidity 60%, CO<sup>2</sup> concentration 400 µmol mol−<sup>1</sup> , photosynthetic photon flux density (PFD) 1000 µmol m−<sup>2</sup> s −1 . For determining the curves of Pn versus PFD, light intensity was controlled by a LI-COR LED irradiation source.

The chlorophyll fluorescence was measured with a PAM 2100 portable chlorophyll fluorometer. Leaves were dark adapted for at least 20 min prior to the measurement. Two measurements were taken from each seedling to determine Fo and Fm, and the maximal photochemical efficency of PSII (Fv/Fm) was calculated according to Krause and Weis (1991).

#### Statistical Analysis

The data were subjected to statistical analysis using Duncan's multiple range test at the 5% (P < 0.05) confidence levels. Data Processing System (DPS) software (Tang and Zhang, 2013) were used for data statistics analysis.

#### RESULTS

#### Generation of GLO Overexpression Rice Lines

Differential GLO overexpression rice lines were generated by upregulating either GLO1 or GLO4. Four independent homozygous lines (two each for either GLO1 or GLO4) were selected for this study. As shown in **Figure 1**, when GLO was upregulated at the mRNA level (**Figures 1A,B**), its catalytic activity was differentially increased, ranging from +60% to +210% (**Figure 1C**). Since we have previously demonstrated that GLO1 and GLO4 were responsible for controlling GLO

FIGURE 1 | Expressional verification of the glycolate oxidase (GLO) overexpression lines. The plants were grown in Kimura B nutrient solution under normal natural conditions [temperature of 30–35/23–26◦C (day/night), photosynthetically active radiation of 600–1500 µmol m−<sup>2</sup> s <sup>−</sup><sup>1</sup> and photoperiod of 12 h day/12 h night]. The fully expanded leaf was detached at four-leaf stage for assay of transcripts (A,B) and activity (C). The OsActin gene was used as an internal control. The data are means ± SD of three biological replicates, and representative of three independent experiments. Different letters on the top of columns indicate significant difference at p < 0.05 according to Duncan's multiple range test.

catalytic activities and that specific silencing of either GLO1 or GLO4 exhibited same phenotypes, indicating both play same physiological roles in rice (Zhang et al., 2012). Thus, for further functional analyses, we shall be able to use these four independent transgenic rice lines with gradient activity increases by overexpressing either GLO1 or GLO4.

#### Phenotypes of GLO Overexpression Lines

Under normal natural conditions, the lines with 60 and 100% increases in GLO activities had significantly higher growth than WT (**Figure 2**). But, as the activity was further increased by 150 or 210%, the growth was inhibited (**Figure 2**).

#### Photosynthesis of GLO Overexpression Lines under Normal and Stressful Conditions

Under normal natural conditions, as has been previously reported, the photosynthetic rate (Pn) was heavily inhibited if GLO was suppressed in either high photorespiratory C3 or low photorespiratory C4 plants (Xu et al., 2009; Zelitch et al., 2009; Lu et al., 2014). More intriguingly, a positive and linear correlation was noticed between Pn and GLO activities when the enzyme was differentially down-regulated by an inducible antisence approach (Xu et al., 2009). Thus, we are curious to know whether Pn can be improved when GLO is upregulated. Here, we generated differential GLO overexpression lines to study their photosynthetic performance. Unexpectedly, all the GLO overexpression lines showed no preference in photosynthetic capacities under normal natural conditions as compared with WT (**Figures 3A–D**). However, measurements of the Pn response to light intensity pointed to a tendency that the overexpression plants may have photosynthetic preference under high light conditions because, as light intensity was increased to high levels (over 1200 µmol m−<sup>2</sup> s −1 ), Pn in WT became leveled off (light saturation point) while it was still gradually increased in the overexpression plants (**Figure 4**).

Under high light and high temperature conditions. We further tested whether differences may occur under stressful conditions. The plants were first grown in a greenhouse under normal natural conditions, then transferred to a growth chamber with temperature of 40◦C (day)/30◦C (night) and light intensity of 900 µmol m−<sup>2</sup> s −1 . Three days after the treatment, photosynthetic parameters were determined. As shown (**Figures 3E–H**) Pn, transpiration rate and stomatal conductance were all significantly improved in the overexpression lines as compared with WT plants, except that the internal CO<sup>2</sup> concentration stayed unaltered for all the plants.

A comparative study was further conducted to verify the above results. The plants were first grown under light of 400 µmol m−<sup>2</sup> s −1 and temperature of 30◦C (day)/25◦C (night) in a growth chamber, and then treated with two conditions: (i) light of 900 µmol m−<sup>2</sup> s −1 and temperature of 30◦C (day)/25◦C (night); (ii) 900 µmol m−<sup>2</sup> s −1 and 40◦C (day)/30◦C (night). The results found that the overexpression plants had significantly higher Pn than WT plants only under the high temperature plus high light conditions, but not different under only this high light (900 µmol m−<sup>2</sup> s −1 ) (**Figure 5A**). For further reinforcement, another experiment was carried out under a natural condition. The plants were grown in a greenhouse during summer season. The temperature was artificially increased by attenuating the air circulation, where temperature and light could be quickly increased to high levels during the noon (**Figure 5B** inset). At 3 days after such condition, Pn was determined. As the temperature and light intensity were increased during the day, Pn of WT plants was decreased while it remained stable for the overexpression plants (**Figure 5A**), further demonstrating photosynthetic preference for the overexpression plants under high light and high temperature conditions.

conditions. The plants were grown under normal natural conditions as described in Figure 1. Until 5-leaf stage, net photosynthesis rates (A), internal CO<sup>2</sup> concentration (B), transpiration rates (C), and stomatal conductance (D) were measured using the youngest fully expanded leaves at between 10:00 and 12:00 of the day. The plants were transferred to a growth chamber with temperature of 40◦C (day)/30◦C (night) and light of 900 µmol m−<sup>2</sup> s <sup>−</sup><sup>1</sup> and 65% humidity. At 3 days after the treatment, the photosynthetic parameters were determined in (E–H). The data are means ± SD of four measurements on different plants and representative of three independent experiments. The other legends are the same as those in Figure 1.

### Resistance of GLO Overexpression Lines to MV-Induced Oxidative Stress

legends are the same as those in Figure 3.

MV is known to be able to induce oxidative stress in plants, particularly under photosynthetic conditions (Kim and Lee, 2002). In addition, MV is also reported to inhibit cyclic electron flow that is essential for photoprotection (Fan et al., 2007, 2008). As shown in **Figure 6**, when the detached rice leaves were treated with MV, all the GLO overexpression lines showed more resistance than WT plants to the MV-induced photooxidative stress.

#### H2O<sup>2</sup> and SA Accumulation in Response to GLO Overexpression

As described previously, GLO is always linked to the photorespiratory H2O<sup>2</sup> production in plants. Here we estimated the H2O<sup>2</sup> content in rice leaves by DAB staining. As shown in **Figure 7A**, H2O<sup>2</sup> was increased in all the GLO overexpression lines under both normal and stressful conditions. It has been documented that H2O<sup>2</sup> and SA may function together in a self-amplifying feedback loop, in which H2O<sup>2</sup> induces SA accumulation and SA in turn enhances H2O<sup>2</sup> accumulation (Chaouch and Noctor, 2010; Miura et al., 2013; Xia et al., 2015). So we further determined the response of SA to the GLO overexpression. The results showed that the contents of both free and total SA were increased in all the overexpression lines compared with WT, similar to the H2O<sup>2</sup> accumulation.

#### Responses of Antioxidant Reactions to GLO Overexpressions

As noticed above, H2O<sup>2</sup> was increased in all the GLO overexpression lines under both normal and the stressful conditions, so it is interesting to know if the antioxidant defense

FIGURE 5 | A further test for Pn of the GLO overexpression lines under high light and high temperature conditions. (A) The plants were first grown under light of 400 µmol m−<sup>2</sup> s <sup>−</sup><sup>1</sup> and temperature of 30◦C (day)/25◦C (night) in a growth chamber to five-leaf stage, then treated with two conditions: (i) light of 900 µmol m−<sup>2</sup> s <sup>−</sup><sup>1</sup> and temperature of 30◦C (day)/25◦C (night); (ii) 900 µmol m−<sup>2</sup> s <sup>−</sup><sup>1</sup> and 40◦C (day)/30◦C (night). At 3 days after the treatment, Pn was determined. (B) The plants were first grown to five-leaf stage in a greenhouse during summer season. The temperature was artificially increased by attenuating the air circulation, where temperature and light can be quickly increased to high levels during the noon (see the inset). At 3 days after such condition, Pn was determined. "µE" represents "µmol m−<sup>2</sup> s −1 ." The data are means ± SD of four measurements on different plants and representative of three independent experiments. The other legends are the same as those in Figure 1.

(0, 3, 6, 9, and 12 h). The data are means ± SD of four biological replicates and representative of three independent experiments.

reactions are activated by the increased H2O2. Unexpectedly, the antioxidant enzymes such as SOD, CAT, POD, and APX, were little altered in the overexpression lines compared with WT plants and the antioxidant metabolite glutathione was also not affected by overexpressing GLO in rice (**Figure 8**).

#### DISCUSSION

Previous results have demonstrated that photosynthetic inhibition occurred either in C3 or C4 plants if GLO was suppressed (Xu et al., 2009; Zelitch et al., 2009; Lu et al., 2014). More intriguingly, a positive and linear correlation was noticed between GLO activities and Pn in rice (Xu et al., 2009). Thus we are curious about what will occur and whether Pn can be improved when GLO is upregulated (Timm et al., 2016). In order to address this question, we further generated differential GLO everexpression rice lines and then conducted detailed functional analyses on these plants, including phenotypic, physiological and biochemical analyses. Phenotype observations showed that, under normal conditions, the lines with 60 or 100% increase in GLO activity showed improved growth whereas the lines with further increases (+150% or +210%) conferred reduced growth (**Figure 2**). But, under normal conditions, photosynthetic parameters were not improved in all these overexpression lines (**Figure 3**). It appears that the improved growth for the first two lines is not correlated with the Pn, but possibility still exists that Pn may have been transiently improved sometimes during the whole growth stage under normal conditions, which failed to be detected by our limited time-point measurements.

Measurements on the Pn response to light intensities pointed to a tendency that the GLO overexpression plants have photosynthetic preference under high light conditions (**Figure 4**). This led us to further test the photosynthetic performance under stressful conditions. Resultantly, under conditions of high light plus high temperature, photosynthetic capacities were significantly improved in the overexpression plants (**Figures 3** and **5**). Moreover, the overexpression plants were more resistant to the MV-induced photo-oxidative stress than WT plants (**Figure 6**). These results collectively suggest that GLO may play a critical role for Pn to cope with high light plus high temperature or the induced oxidative stress. Pn is known as the most sensitive physiological process to stresses, and any alterations in photosynthetic attributes under stresses are good indicators of the plant stress tolerance, and thus, in any species the ability to sustain leaf gas exchange under stress has direct relationship with the stress tolerance (Wahid et al., 2007). In other words, it may be extended that GLO may play important roles for rice plants to cope with high light and high temperature, or the induced oxidative stress. Such a role is of far-reaching practical significance as rising atmospheric CO<sup>2</sup> is driving temperature increases (i.e., global warming) in many already stressful environments, such as strong light and drought, particularly for rice, as a staple food crop (Singh et al., 2014).

Photorespiration is generally stimulated as light intensity is increased (Brown and Morgan, 1980; Gerbaud and André, 1980; Vines et al., 1982; Haupt-Herting et al., 2001), which is even more dependent on light intensity when coupled with other stresses, such as high temperature, water stress or CO<sup>2</sup> deficit (Kangasjärvi et al., 2012). High temperature can stimulate photorespiratory flux even if light intensity is constant, because (i) the solubility of CO<sup>2</sup> in water decreases with temperature more than the solubility of O2, resulting in a lower CO2:O<sup>2</sup> ratio at the active site of Rubisco (ribulose-1, 5-bisphosphate carboxylase/oxygenase); and (ii) the enzymatic properties of Rubisco shift with temperature, stimulating RuBP oxygenation to a greater degree than RuBP carboxylation (Foyer et al., 2009). In addition, high temperature and high light can result in stomatal closure, which reduces the C:O ratio around Rubisco, thereby promoting PR as an indirect result (Kangasjärvi et al.,

s −1 light intensity, and 65% humidity] for 3 days, then the temperature and light intensity were increased to 40◦C (day/12 h)/30◦C (night/12 h) and 900 µmol m−<sup>2</sup> s −1 , respectively, for 24 h. H2O<sup>2</sup> and SA were measured before and after the treatments. (A) H2O2-3<sup>0</sup> -diaminobenzidine (DAB) staining and the result is representative of three independent experiments; (B,C) SA was determined by HPLC chromatography. The other legends are the same as those in Figure 1.

2012). Thus, high light plus high temperature may be able to markedly stimulate photorespiratory metabolism, leading to the overproduction of glycolate. If such glycolate is not removed timely and accumulated within chloroplasts, it may be converted into glyoxylate by a possibly existing photosystem I-dependent oxidation system (Murai and Katoh, 1975; Goyal and Tolbert, 1996; Goyal, 2002). The accumulated glyoxylate in chloroplasts has been known to inhibit Pn (Chastain and Ogren, 1989; Campbell and Ogren, 1990; Lu et al., 2014). Under such circumstances, therefore, a higher level of peroxisomal GLO is able to facilitate a timely scavenging of the overproduced glycolate so as to avoid its toxicity to chloroplasts.

In contrast with the above notion, Nölke et al. (2014) recently reported that promoting glycolate oxidation within chloroplasts even improved Pn and yield in potato (Nölke et al., 2014). In addition, the above notion may not explain the result that the overexpression lines show more resistance to the MV-induced photooxidative stress (**Figure 6**). It has been demonstrated that GLO plays important roles in both biotic and abiotic responses or resistance (Mukherjee and Choudhuri, 1983; Bohman et al., 2002; Taler et al., 2004; Rojas et al., 2012). Considering mechanisms, the researchers always link it to the GLO-catalyzed H2O<sup>2</sup> production, as is known to play a signaling role in various physiological processes (Foyer et al., 2009). It is extensively documented that H2O<sup>2</sup> originates mainly in apoplasts associated with the plasmalemma, but evidences are accumulating to show that other intracellular sources of H2O2, notably chloroplasts, peroxisomes and mitochondria, could be also involved. Peroxisomes and chloroplasts may accumulate 30– 100 times higher H2O<sup>2</sup> as compared to mitochondria (Hossain et al., 2015). The peroxisomal H2O<sup>2</sup> is well known to be ultimately contributed by the GLO-catalyzed glycolate oxidation (Noctor et al., 2002; Foyer and Noctor, 2003; Kangasjärvi et al., 2012). Accumulation of the peroxisomal H2O<sup>2</sup> stimulated the isochorismate-dependent SA synthesis and then triggered SArelated pathogenesis responses and defense gene expressions in plants (Chamnongpol et al., 1998; Chaouch et al., 2010; Kangasjärvi et al., 2012). The peroxisomal H2O<sup>2</sup> can also induce oxidative stress that would activate programme cell death (PCD) under long day and high light if not controlled by CAT activity (Chaouch et al., 2010; Mhamdi et al., 2010; Suzuki et al., 2011). Sewelam et al. (2014) most recently revealed that the peroxisomal H2O<sup>2</sup> induced transcripts for stress tolerance, and Rojas et al. (2012) presented more strong evidence indicating that the GLO-catalyzed H2O<sup>2</sup> production contributed to both genefor-gene and non-host resistance in Nicotiana benthamiana and Arabidopsis.

In this study, we observed that both H2O<sup>2</sup> and SA were correspondingly induced but the antioxidant reactions were not responsive upon the GLO overexpression (**Figures 7** and **8**), although the result that the overexpression lines are more resistant to the MV-induced photooxidative stress (**Figure 6**) points toward possibilities that the antioxidant systems have been activated in these plants. While many publications have demonstrated that exogenous or stress-induced H2O<sup>2</sup> is able to activate the antioxidant defense system, including both non-enzymaitc and enzymatic, the correlation between the

endogenous H2O<sup>2</sup> and antioxidant systems is not well established so far (Neill et al., 2002; Winfield et al., 2010; Del, 2015). During the past years, by using CAT-deficient mutants and/or GLOupregulated transgenic plants as an endogenous H2O<sup>2</sup> burst producer, it was only observed that some antioxidant enzyme genes, such as APX and GPX, were induced at both transcript and protein levels as the endogenous H2O<sup>2</sup> is enhanced, but few data at the activity level confirmed these responses (Neill et al., 2002; Mhamdi et al., 2010; Sewelam et al., 2014; Del, 2015; Xia et al., 2015). It seems that exogenous or stress-induced H2O<sup>2</sup> could be different from endogenous H2O<sup>2</sup> in triggering metabolic or physiological responses, likely the former being mostly a stressor while the latter acting mostly as a signal. In contrast, the results that H2O<sup>2</sup> and SA were correspondingly induced in the overexpression plants (**Figure 7**) are in well agreement with previous results (Chaouch et al., 2010; Mhamdi et al., 2010). These two substances have been well known as key signaling molecules to be able to trigger various defense responses (Herrera-Vásquez et al., 2015; Xia et al., 2015), both of which may function together in a self-amplifying feedback loop, in

#### REFERENCES


which H2O<sup>2</sup> induces SA accumulation and SA in turn enhances H2O<sup>2</sup> accumulation (Khokon et al., 2011; Miura et al., 2013; Xia et al., 2015). Therefore, it can be inferred that both H2O<sup>2</sup> and SA are involved in triggering some stress defense responses, but not including antioxidant reactions, for the GLO overexpression plants to cope with high light and high temperature.

#### AUTHOR CONTRIBUTIONS

X-XP conceived the idea and designed the experiments. L-LC, Y-sL, and YL performed the experiments. X-XP wrote the manuscript. CY and L-LC edited the manuscript. All the authors approved the final manuscript.

#### ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (U1201212; 31470343).


during irradiation. Plant Cell Physiol. 41, 1397–1406. doi: 10.1093/pcp/ pcd074


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Cui, Lu, Li, Yang and Peng. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Comparative Proteomic Analysis of Cultured Suspension Cells of the Halophyte *Halogeton glomeratus* by iTRAQ Provides Insights into Response Mechanisms to Salt Stress

Juncheng Wang1, 2 †, Lirong Yao1, 2 †, Baochun Li 1, 3, Yaxiong Meng1, 2, Xiaole Ma1, 2 , Yong Lai <sup>4</sup> , Erjing Si 1, 2, Panrong Ren1, 2, Ke Yang1, 2, Xunwu Shang<sup>2</sup> and Huajun Wang1, 2 \*

#### *Edited by:*

Dipanjana Ghosh, National University of Singapore, Singapore

#### *Reviewed by:*

Torsten Kleffmann, University of Otago, New Zealand Alberto A. Iglesias, Instituto de Agrobiotecnología del Litoral, Argentina Asish Kumar Parida, Central Salt & Marine Chemicals Research Institute, India

*\*Correspondence:*

Huajun Wang whuajun@yahoo.com

† These authors have contributed equally to this work.

#### *Specialty section:*

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

*Received:* 09 December 2015 *Accepted:* 21 January 2016 *Published:* 09 February 2016

#### *Citation:*

Wang J, Yao L, Li B, Meng Y, Ma X, Lai Y, Si E, Ren P, Yang K, Shang X and Wang H (2016) Comparative Proteomic Analysis of Cultured Suspension Cells of the Halophyte Halogeton glomeratus by iTRAQ Provides Insights into Response Mechanisms to Salt Stress. Front. Plant Sci. 7:110. doi: 10.3389/fpls.2016.00110 <sup>1</sup> Gansu Provincial Key Lab of Aridland Crop Science/Gansu Key Lab of Crop Improvement and Germplasm Enhancement, Lanzhou, China, <sup>2</sup> Department of Crop Genetics and Breeding, College of Agronomy, Gansu Agricultural University, Lanzhou, China, <sup>3</sup> Department of Botany, College of Life Science and Technology, Gansu Agricultural University, Lanzhou, China, <sup>4</sup> Department of Agriculture and Forestry, College of Agriculture and Animal Husbandry, Qinghai University, Xining, China

Soil salinity severely threatens land use capability and crop yields worldwide. An analysis of the molecular mechanisms of salt tolerance in halophytes will contribute to the development of salt-tolerant crops. In this study, a combination of physiological characteristics and iTRAQ-based proteomic approaches was conducted to investigate the molecular mechanisms underlying the salt response of suspension cell cultures of halophytic Halogeton glomeratus. These cells showed halophytic growth responses comparable to those of the whole plant. In total, 97 up-regulated proteins and 192 down-regulated proteins were identified as common to both 200 and 400 mM NaCl concentration treatments. Such salinity responsive proteins were mainly involved in energy, carbohydrate metabolism, stress defense, protein metabolism, signal transduction, cell growth, and cytoskeleton metabolism. Effective regulatory protein expression related to energy, stress defense, and carbohydrate metabolism play important roles in the salt-tolerance of H. glomeratus suspension cell cultures. However, known proteins regulating Na<sup>+</sup> efflux from the cytoplasm and its compartmentalization into the vacuole did not change significantly under salinity stress suggesting our existing knowledge concerning Na<sup>+</sup> extrusion and compartmentalization in halophytes needs to be evaluated further. Such data are discussed in the context of our current understandings of the mechanisms involved in the salinity response of the halophyte, H. glomeratus.

Keywords: halophyte, *H. glomeratus*, iTRAQ, salt tolerance, cells, response mechanisms

### INTRODUCTION

Soil salinity severely limits robust plant growth and development, as well as the attainment of an adequate crop yield. It is estimated that approximately one-third of arable land throughout the world is affected by natural and secondary salinity to varying degrees (Munns, 2002, 2005; Munns and Tester, 2008; Li et al., 2012; Wang et al., 2013). Thus, improving the tolerance of crops to salinity in order to increase food production has become an urgent goal for plant breeders. Soil salinity mainly induces osmotic stress and ion toxicity, important processes that are considered most harmful to plants (Flowers and Colmer, 2008; Munns and Tester, 2008). For many plants, salt stress can cause a slowing of growth, wilting or even death. To survive under salinity stress, plants have evolved sophisticated response and adaptive mechanisms at biochemical, physiological, cellular, and molecular levels (Zhang et al., 2011). Many salt-tolerantrelated candidate genes and gene products involved in salt uptake and transport (Munns et al., 2012; Guan et al., 2014), the elimination of reactive oxygen species (ROS; Suzuki et al., 2012; Peng et al., 2014), the accumulation of organic compounds (Ashraf and Foolad, 2007), the regulation of hormones (Jiang et al., 2013; Osakabe et al., 2014), and other processes have been conclusively identified. Unfortunately, however, salt-tolerant crops using some of these genes have not been developed due to the complexity of the mechanisms of salt tolerance (Witzel et al., 2010; Shabala and Munns, 2012). Understanding the molecular mechanisms of a salt response and defense in plants will help in the development of crops with salt tolerance (Deinlein et al., 2014; Munns and Gilliham, 2015). However, we have an extremely limited understanding of the molecular ion transport and regulatory mechanisms activated in plants under salt stress.

Halophytes are plants capable of surviving to reproduce in highly saline soils. Such a robust tolerance of salt is mainly due to the more effective control of the uptake and transport of Na<sup>+</sup> and Cl<sup>−</sup> compared with glycophytes (Flowers and Colmer, 2008). These special adaptive strategies make halophytes an excellent model to study molecular mechanisms of salt tolerance in plants. Recent advances in functional genomic technologies, including transcriptomics (Dang et al., 2013; Garg et al., 2013), proteomics (Wang et al., 2013, 2015b; Cheng et al., 2015), and metabolomics (Ruan and Teixeira da Silva, 2011; Obata and Fernie, 2012), have increased our knowledge of the complex regulatory networks associated with stress adaptation and tolerance of halophytes. Although changes in gene expression at the transcriptional level are not always reflected at the protein level (Lan et al., 2011), such biological processes are ultimately controlled by proteins. Therefore, clarification of changes in protein profiles using a proteomics approach will reveal a more realistic picture of the metabolic adjustments occurring during salinity stress. In the past two decades, a series of novel techniques have been developed and widely applied in the field of proteomics research. Isobaric tags for relative and absolute quantification (iTRAQ) is a powerful, gel-free quantification method that allows for the relative and absolute quantification of peptides. This technique is based on stable isotope labeling of peptides and measures peak intensities of reporter ions produced by precursor ion fragmentation in tandem mass spectrometry (Ross et al., 2004). In view of obvious advantages, such as high accuracy in identification and quantification of proteins, a sensitivity for low-abundance proteins, and the relatively high through-put multiplexing of up to eight samples (Bindschedler and Cramer, 2011), iTRAQ-based proteomics has been used to study several adaptive strategies of plants in response to abiotic challenges using leaves (Schneider et al., 2009; Liu et al., 2014), roots (Lan et al., 2011; Owiti et al., 2011; Wang et al., 2014), and suspension cell cultures (Rao et al., 2010; Böhmer and Schroeder, 2011). In an iTRAQ-based proteomics analysis on the mechanisms of a salinity response in halophytes, proteins involved in protein synthesis, photosynthesis, metabolism, and energy changed significantly under salt stress in Thelungiella halophila compared to Arabidopsis thaliana (Pang et al., 2010). Cheng et al. (2015) analyzed the dynamic protein expression patterns of leaves from Tangut Nitraria under salinity stress and showed that proteins related to redox homeostasis, photosynthesis, and energy metabolism made up the primary response networks to salinity (Cheng et al., 2015). However, this proteomics technique has not been used to explore the response mechanisms of the halophyte, Halogeton glomeratus, to salt stress.

The culture of plant cells in suspension offers a simplified model system for the study of cellular and molecular processes. Its predominant advantage is that a relatively homogenous, single cell population allows a rapid and uniform response to external stimuli, thereby avoiding the complications of multicellular types at the whole plant level (Mustafa et al., 2011). Based on the above advantages, suspension cell cultures have been widely used in investigating the physiological and molecular mechanisms involved in plant responses to salt stress. In their study of halophytes, Rosario Vera-Estrella et al. (1999) established a salt-tolerant, stable suspension culture of Mesembryanthemum crystallinum cells that showed a halophytic growth response comparable to that of the whole plant (Vera-Estrella et al., 1999). Their analysis of metabolic pathways suggested that NaCl stress induced programmed cell death in T. halophila suspension cell cultures that is similar to apoptosis in mammalian cells (Wang et al., 2010). Comparing ion content and distribution under NaCl stress in suspension cell cultures of the mangrove halophyte, Sonneratia alba, suggested that effective transport of Na<sup>+</sup> and Cl<sup>−</sup> into their vacuoles improved the salt tolerance of these cells (Hayatsu et al., 2014). Furthermore, during investigations of the salinity-induced proteomics of Nitraria sphaerocarpa suspension cells, it was found that proteins involved in signal transduction, cell rescue/defense, the cytoskeleton, the cell cycle, and in protein folding and assembly changed significantly after salt stress (Chen et al., 2012).

The halophyte, H. glomeratus, belongs to the Chenopodiaceae family, which has a strong tolerance to salinity; a physiological analysis of seedlings suggested that osmotic adjustment is one of the primary mechanisms of salt tolerance (Wang et al., 2015b). However, for osmotic adjustments in biological processes under salt stress, the primary transporters associated with Na<sup>+</sup> accumulation in cells and its compartmentation in the vacuole remain are unclear (Wang et al., 2015a). The culture of salt-tolerant suspension cells is a useful tool for clarifying biological processes and primary transporters associated with Na<sup>+</sup> accumulation and compartmentation under salt stress in halophytes (Vera-Estrella et al., 1999).

In the current study, we present a comprehensive proteomic analysis of suspension cell cultures of halophytic H. glomeratus treated with different NaCl concentrations using an iTRAQbased approach. The objective of our work is to explore protein expression changes in response to NaCl and to Wang et al. Proteomics of H. glomeratus in Salinty

highlight any potential response mechanisms of salt stress at a cellular level. This work will increase our understanding of halophyte cellular responses to salinity in H. glomeratus and will be of great interest in the rapidly developing field of salt-tolerant mechanisms. To our knowledge, this is the first report of the suspension cell culture of H. glomeratus in response to salinity stress using a comparative proteomics approach.

#### MATERIALS AND METHODS

#### Suspension Cell Cultures and Salt Stress Treatments

Seedlings of H. glomeratus were used for generating calluses. Seeds of H. glomeratus were sterilized in 25% sodium hypochlorite and germinated on Murashige and Skoog (MS) solid medium under continuous light (300 µmol m−<sup>2</sup> s −1 ) at 25◦C and 70% relative humidity. For the 12 days culture, sterile apical meristems were cut off from seedlings and cultured on MS solid medium containing 2 mg/L 2,4-D (2,4-dichlorophenoxyacetic acid) at 25◦C with continuous white light (100 µmol m−<sup>2</sup> s −1 ). The medium was refreshed every 2 weeks. After 28 days, each 2.0 g of callus was transferred to 50 mL MS liquid medium containing 2 mg/L 2,4-D and grown in continuous white light (100 µmol m−<sup>2</sup> s −1 ) with shaking at 120 rpm at 25◦C. Cell suspensions were subcultured every 7 days for 5 weeks to obtain mainly synchronized cells. Salt stress treatments were performed, 7 days after changing the medium, by supplementing with 200 (moderate salinity stress) or 400 (severe salinity stress) mM NaCl. NaCl was not added to control medium. Cells were collected by filtration after the 5 days stress period (Whatman 113; Whatman International Ltd., Brentford, UK), frozen in liquid N2, and stored at −80◦C for proteome analysis. Each treatment was made up of three biological replicates.

#### Measurement of Ion Concentrations

Na<sup>+</sup> and K<sup>+</sup> contents was determined as described previously (Munns et al., 2010) using atomic absorption spectrometry. In brief, at the end of treatments, cells were rapidly rinsed three times with ultrapure water and freeze-dried at −50◦C. Concentrations of Na<sup>+</sup> or K<sup>+</sup> were determined after digestion with 0.5% 0.5% nitric acid and the recovery of dry material as a fine powder.

### Cell Viability Assay

NaCl stress tolerance of cells was determined by assessing the relative cell viability of suspension cell cultures. Relative cell viability was measured using a 2,3,5-triphenyltetrazolium chloride (TTC) reduction method after a freeze-thaw cycle (Li et al., 2012). About 0.2 g of fresh cells per sample were used and three independent experiments were performed.

#### Protein Extraction and iTRAQ Labeling

For protein extraction, cells were suspended in lysis buffer (7 M urea, 2 M thiourea, 4<sup>c</sup> /oCHAPS, 40 mM Tris-HCl pH 8.5, 1 mM phenylmethylsulfonyl fluoride [PMSF], 2 mM ethylenediaminetetraacetic acid [EDTA]) and the mixture sonicated on ice. Samples were reduced with 10 mM dithiothreitol (DTT; final concentration) at 56◦C for 1 h and then alkylated using 55 mM iodoacetamide (IAM; final concentration) for 45 min in darkness at room temperature. The reduced and alkylated protein mixtures were precipitated by adding a 4 × volume of chilled acetone at −20◦C for 14 h. After centrifugation at 25,000 × g for 10 min at 4◦C, the pellet was dissolved in 0.5 M tetraethylammonium borohydride (TEAB; Applied Biosystems, Milan, Italy) and sonicated on ice for 15 min. Proteins in the supernatant were collected after centrifuging at 25,000 × g at 4 ◦C for 20 min, and protein concentrations determined by the Bradford method using bovine serum albumin as a standard (Bradford, 1976).

Protein (100 µg) was digested with 5 µg Trypsin Gold (Promega, Madison, WI, USA) at 37◦C for 16 h. Peptides were reconstituted in 0.5 M TEAB and processed according to the manufacturer's protocol for 8-plex iTRAQ reagent (Applied Biosystems, Foster City, CA, USA). Each treatment was made up of two (Control) or three (Treated with 200 or 400 mM NaCl) biological replicates. Briefly, peptides were labeled with iTRAQ reagents 113 and 115 for control samples, 114, 116, and 118 for 200 mM NaCl-treated samples, and 117, 119, and 121 for 400 mM NaCl-treated samples, then pooled, and dried by vacuum centrifugation.

#### Separation of Peptides and LC-MS/MS Analysis

Offline strong cation exchange (SCX) was performed as previously reported by Patterson et al. (2007) using a LC-20AB HPLC Pump system (Shimadzu, Kyoto, Japan) to fractionate complex peptides (Patterson et al., 2007). In total, SCX peptide fractions were pooled into 20 fractions, desalted using a Strata X C18 column (Phenomenex, Torrance, CA, USA) and vacuumdried for liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. Each dried fraction was resuspended in buffer (5% acetonitrile [ACN], 0.1% formic acid [FA]), centrifuged at 20,000 × g for 10 min; the final concentration of peptide was 0.5 µg/µL. A 5 µL peptide was loaded onto a LC-20AD nano high-pressure liquid chromatography analyzer (HPLC; Shimadzu, Kyoto, Japan) by an autosampler into a 2 cm C18 trap column (inner diameter 200 µm). Peptides were then eluted into a 10 cm analytical C18 column (inner diameter 75 µm) packed in-house.

Liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) analysis of fractionated samples was performed using a TripleTOF 5600 System (AB SCIEX, Concord, ON, Canada), as described previously, at Beijing Genomics Institute (BGI, Shenzhen, China; Yang et al., 2013). An ion spray voltage of 2500 V was used for data acquisition. For information dependent acquisition (IDA), survey scans were acquired in 250 ms and as many as 30 product ion scans were collected if a threshold of 120 counts per second and with a charge state of 2+ to 5+ was exceeded. A sweeping collision energy of 35 eV was applied to all precursor ions for collision-induced dissociations.

### Database Search and Quantification

Protein identification and quantification were analyzed against a H. glomeratus protein database containing 30,124 non-redundant sequences (Source: NCBI [BioProject ID: PRJNA 254029]) using a Mascot search engine (Matrix Science, London, UK; version 2.3.02). For protein identification, the search parameters used were as follows: fragment mass tolerance of 0.1 Da, peptide mass tolerance of 0.05 Da, trypsin was chosen as the enzyme, with a maximum of one missed cleavage, Gln->pyro-Glu (Nterm Q), Oxidation (M), Deamidated (NQ) as the potential variable modifications, and Carbamidomethyl (C), iTRAQ8plex (N-term), iTRAQ8plex (K) as fixed modifications. To reduce the probability of false peptide identification, only peptides at a 95% confidence interval (P < 0.05) with a false discovery rate (FDR) estimation ≤1.04% were counted as being successfully identified. Each positive protein identification contained at least one unique peptide. The MS-based proteomic data are available via ProteomeXchange with identifier PXD003358.

For protein quantitation, iTRAQ 8-plex was chosen for quantification during the search, and the minimum peptide was set to two. Proteins containing at least two unique spectra were selected for quantification analysis. The control sample was used as reference based on the weighted average of the intensity of reported ions in identified peptides. Quantitative protein ratios were weighted and normalized by the median ratio in Mascot. Ratios with P values < 0.05, and a >1.5 fold or <0.67-fold cut-off were considered as up- or downregulated proteins, respectively. Differentially expressed proteins were enriched to Gene Ontology (GO) Terms by the agriGO web service (http://bioinfo.cau.edu.cn/agriGO/index.php) on the basis of their biological functions. Protein classification was performed essentially on the basis of metabolic and functional features as described by Bevan et al. (1998).

### RESULTS

### Effect of NaCl Treatment on Morphological and Physiological Changes

H. glomeratus is an extreme halophyte that compartmentalizes toxic ions into vacuoles as its primary salt-tolerance mechanism. In our studies, a large central vacuole was evident within suspension cells cultured with different salt treatments (**Figures 1A–C**). When H. glomeratus suspension cells were treated with 200 or 400 mM of NaCl for 5 d the size of the large central vacuole clearly confined the cytoplasm to a thin peripheral layer (**Figures 1B,C**). To evaluate the effect of ion toxicity on cell viability, we examined suspension cells cultured under different NaCl concentrations. As shown in **Figure 1D**, salt stress resulted in a significant inhibition of cell viability, showing a decrease to 86.39 and 73.74% of the control for 200 and 400 NaCl mM conditions, respectively (P < 0.05). When compared with control cells, the Na<sup>+</sup> and K<sup>+</sup> contents of cells significantly increased and decreased, respectively (**Figure 1E**). The Na<sup>+</sup> concentration of cells treated with 200 or 400 mM NaCl were 20.18- or 32.81-fold higher than that of control cells. As a result, the K+/Na<sup>+</sup> ratio in cells markedly decreased under salinity stress (**Figure 1E**). Thus, the analysis of these physiological features indicated that suspension cell cultures of H. glomeratus displayed a robust ability to tolerate NaCl stress similar to that observed at the whole plant level. The transport of Na<sup>+</sup> across the plasma membrane and its segregation in the vacuole still play key roles in the regulation of ion homeostasis under salt stress in suspension cell cultures of H. glomeratus.

### Changes in Suspension Cell Protein Profiles Under Salt Stress

To determine any proteomic changes in suspension cell cultures of H. glomeratus in response to NaCl stress at an adaptive stage, protein profiles for cells exposed to 0, 200, and 400 mM NaCl stress for 5 days were explored using an 8-plex iTRAQ method. At least two biological repeats were prepared for each condition. After analysis using mascot software by searching the total 444,052 spectra, a total of 23,893 peptides, and 5,649 proteins were identified. The detailed information of proteins and peptides is listed in Supplementary File 1. Proteins with at least two unique peptides (3,796, 67.2%) were quantified according to the criteria of a 95% confidence level (P < 0.05) and a 1.5-fold change upon up- or downregulation. Compared to untreated control cells, 489 and 473 proteins changed significantly during 200 or 400 mM NaCl stress treatments, respectively (Supplementary Files 2, 3). The significantly enriched biological processes of these differentially expressed proteins were annotated by their GO annotation and enrichment analysis with AgriGO and are summarized in **Figure 2** and Supplementary Files 4, 5. In 200 mM NaCl stressed cells, the prevalent categories included "metabolic process," "cellular process," "single-organism process," "response to stimulus," and "cellular component organization or biogenesis." However, in 400 mM NaCl stressed cells, the number of proteins involved in "cellular process," "metabolic process," "cellular component organization or biogenesis," and "multi-organism process" slightly increased, and the number of proteins involved in "regulation of biological process," "biological regulation," and "multicellular organismal process" decreased slightly when compared to protein categories associated with the 200 mM NaCl treatment (**Figures 2A,B**). This suggests that similar strategies might be activated in response to salt stress in suspension cells treated with Na<sup>+</sup> concentrations ranging from 200 to 400 mM for 5 days.

### Potential Proteins Regulated by NaCl Stress

The present study was concerned with identifying proteins involved in tolerance to salt stress and any associated tolerance mechanisms in suspension cell cultures. Of the differentially expressed proteins, 177 were up-regulated and 312 were downregulated in the presence of 200 mM NaCl. Similarly, in 400 mM NaCl-treated cells, 160 proteins were up-regulated and 313 proteins were down-regulated. Notably, more down- than up-regulated proteins were identified when cells were exposed to different concentrations of NaCl stress. The numbers of differentially expressed proteins and how they overlapped under different NaCl treatments are illustrated by Venn diagram analysis as shown in **Figure 3**. Of 337 up-regulated proteins, 87

proteins were observed in cells under salinity stress at both NaCl concentrations (**Figure 3A** and Table S1). The functions of these proteins were classified into 10 categories, including energy (15, 17.24%), carbohydrate metabolism (6, 6.90%), stress defense (25, 28.74%), protein metabolism (2, 2.30%), signal transduction (3, 3.45%), cell growth/division (4, 4.60%), metabolism (10, 11.49%), secondary metabolism (16, 18.39%), unclassified (1, 1.15%), and unknown (5, 5.75%; **Figure 3C**). Of the 431 down-regulated proteins, 194 proteins were down-regulated when cells were exposed to both NaCl treatments (**Figure 3B** and Table S2). The functions of these down-regulated proteins were grouped into 13 categories involving energy (7, 3.16%), carbohydrate metabolism (11, 5.67%), stress defense (16, 8.28%), protein metabolism (41, 21.13%), transportation (3, 1.55%), signal transduction (10, 5.15%), cell growth/division (22, 11.34%), cytoskeleton metabolism (22, 11.34%), metabolism (11, 5.67%), secondary metabolism (5, 2.58%), transcription (16, 8.25%), unclassified (6, 3.09%), and unknown (24, 12.37%; **Figure 3C**).

In addition, a comparison between the differently expressed proteins and their categories was carried out (**Figure 3C**). The proportion of up-regulated proteins related to energy, carbohydrate metabolism, stress defense, metabolism, and secondary metabolism was higher than that of down-regulated proteins. However, the modulation of proteins involved in transportation and transcription was not observed at a 200 mM NaCl stress level (**Figure 3C**). This indicated that only high-salt stress condition significantly affected the biological processes of material transport and DNA transcription in suspension cells of H. glomeratus. In addition, many hypothetical proteins were detected among these differentially expressed proteins. Further studies are required on such proteins with unknown functions.

### Proteins Related to Toxic Ion Transport

To investigate a possible ion-regulating model for proteins involved in transporting toxic ions (mainly Na+), we visualized the expression patterns of Na<sup>+</sup> and K<sup>+</sup> transport-related proteins in response to salt stress. As shown in **Figure 4**, a total of 21 proteins were identified related to Na<sup>+</sup> and K<sup>+</sup> transport, of which the category "tonoplast H<sup>+</sup> pumps" represented the largest group (9, 42.85%), followed by "ion transporter" proteins (6, 28.57%), "plasma membrane H<sup>+</sup> pumps" (3, 14.29%), and "channel" proteins (3, 14.29%). Contrary to our expectations, after using a cutoff value of a 1.5-fold-change threshold to stringently assess a protein as being responsive to salinity stress, none of these proteins showed markedly different expressions in all NaCl stress conditions compared to untreated cells. This indicated that ion transport in suspension cell cultures had achieved a state of balance in the extracellular space, cytoplasm, and vacuole after vacuole after being exposed to salinity for 5 d. Furthermore, ion transporter proteins, such as cation-chloride cotransporter 1 isoform 2, putative potassium transporter KUP3, and Na+/H<sup>+</sup> antiporter, were only detected by one unique peptide, but its low expression abundance could not satisfy the quantitative study.

#### DISCUSSION

### Na<sup>+</sup> Accumulation and Compartmentation in Suspension Cell Cultures of *H. glomeratus*

We successfully established a suspension cell culture from seedlings of H. glomeratus. We then exposed suspension cell

#### FIGURE 3 | Graphical representation and functional cataloging of differentially expressed proteins. Venn diagrams showing common, significantly up-regulated (A) and down-regulated proteins (B) in cells exposed to salt stress. Functional classification of common up-regulated (left) and down-regulated (right) proteins based on Bevan et al. (1998) (C). The histogram shows the distribution of the differentially expressed proteins into their functional classes represented by percentages.

cultures to different NaCl stress conditions and analyzed any physiological changes at the cellular level. Suspension cells of H. glomeratus seedlings were capable of growing at high salinity levels, with their relative cell viability reaching up to 78% in 400 mM NaCl-containing medium compared to untreated, control cells. Moreover, an accumulation of Na<sup>+</sup> was observed in salttreated suspension cells, with the highest intracellular content of Na<sup>+</sup> found when cells were grown in the presence of 400 mM NaCl for 5 d. These characteristics were similar to those reported previously for whole plants of H. glomeratus (Wang et al., 2015b) and related halophytes such as M. crystallinum (Vera-Estrella et al., 1999), and indicate that these cells are salt-tolerant and show growth characteristics in response to salinity comparable to those at the whole plant level. Although not all responses of cell suspensions to salinity were similar to those of whole plants, the large central vacuole that occupied most of the total cell volume, and high concentrations of Na<sup>+</sup> accumulated within cells of salt-treated H. glomeratus are characteristic of whole plant responses to salinity stress (**Figure 1**). We hypothesize that a high concentration of Na<sup>+</sup> was probably sequestered within the vacuole to avoid ion toxicity of the cytoplasm and to increase cellular osmolarity under salinity stress. The compartmentalization of Na<sup>+</sup> into a vacuole is one of the most important strategies employed by suspension cell cultures of halophytes in response to salt stress (Vera-Estrella et al., 1999; Mimura et al., 2003).

### Protein Identification and Regulation under NaCl Stress

The objective of our work was to identify response mechanisms of suspension cells to salt stress from the perspective of comparative proteomics. Liu et al. (2013) integrated information from proteomic and metabolomic studies of suspension cell cultures of rice under salinity stress. They demonstrated that salt-responsive networks of suspension cells were extremely complex and identical to those observed at the plant level (Liu et al., 2013). In our experiments, suspension cells of the halophyte, H. glomeratus, were treated with NaCl concentrations of 0, 200, and 400 mM for 5 days and then subjected to proteomics analysis using the gel-free iTRAQ system. Among 5,649 identified proteins, a total of 87 up- and 194 downregulated proteins responded to both NaCl stress treatments (commonly responded). The functions of these salt responsive (mainly up-regulated) proteins and their main pathways are discussed in the following section.

#### Proteins Involved in Energy and Carbohydrate Metabolism

A plant energy deficit is among the primary symptoms induced in response to salinity. The observed reduction in photosynthesis and/or respiration is closely associated with growth arrest and cell death (Baena-González and Sheen, 2008). We found that in suspension cells of H. glomeratus, 15 upregulated proteins were noted after NaCl treatment, including ribulose bisphosphate carboxylase/oxygenases (Rubisco) activase (CL3673.Contig2 and CL3673.Contig4), ribulose bisphosphate carboxylase small chain 1 (Unigene1053), ribulose-phosphate 3 epimerase (Unigene20305) and phosphoribulokinase (Unigene 25793), which are all Calvin cycle-specific enzymes. An increase in abundance of these proteins may contribute to an enhancement of Calvin cycle activity, leading to improvements of photosynthetic CO<sup>2</sup> assimilation in response to salinity. Consistent with our previous proteome results at the wholeplant level, the relative abundance of Rubisco was a rapid increase in salt stress (Wang et al., 2015b). Two proteins involved in light reaction processes, namely, a 23 kDa precursor protein of the oxygen-evolving complex (Unigene6788), and a photosystem II reaction center PSB28 protein, have been shown to up-regulate light reaction-related proteins under NaCl stress in order to provide an adequate proton gradient and energy for other metabolic processes to operate normally (Pang et al., 2010). Furthermore, we also identified a serine hydroxymethyltransferase (Unigene11766) that belongs to a photorespiratory pathway.

The expression of electron transfer-related proteins, such as ferredoxin-1 (CL1018.Contig1, Unigene18531), a precursor of plastocyanin (Unigene13000), and glycerate dehydrogenase, (Unigene588) was up-regulated; these can modulate electron transfer effectively and improve ATPase synthesis and NADPH formation (Zhang et al., 2011). In addition, we identified an ATP-dependent zinc metalloprotease FTSH 2 (Unigene1270), a mitochondrial substrate carrier (Unigene14606), and an ATP synthase delta subunit precursor (Unigene523), which are classified as ATPase activity proteins. These results suggest that at high NaCl concentrations, suspension cells of H. glomeratus are able to deal with salt stress by sustaining appropriate level of the energy metabolism.

It is well established that carbohydrate metabolism largely changes in response to an energy shortage caused by salt stress in plants (Zhang et al., 2011). In our study, we, firstly, found a greater number of down- as opposed to up-regulated proteins involved in carbohydrate metabolism with salt treatment. Secondly, with regard to the downregulated proteins, the abundance of most was lower for the 400 mM than the 200 mM NaCl treatment, indicating that normal pathways of carbohydrate metabolism were inhibited by a high concentration of NaCl. In suspension cells of H. glomeratus, four proteins belonging to the aldoketo reductase (AKRS) superfamily, aldo-keto reductase yakc (CL1157.Contig1), aldo-keto reductase 2 (CL1157.Contig3, Unigene18105), and aldo/keto reductase (Unigene31104), were up-regulated under NaCl stress. AKRS, when induced by various stress treatments, confers tolerance to abiotic stresses (Éva et al., 2014; Kanayama et al., 2014) while overproduction of the OsAKR1 enzyme increases oxidative and heat stress tolerance through malondialdehyde and methylglyoxal detoxification in rice (Turóczy et al., 2011). The chloroplast glyceraldehyde-3 phosphate dehydrogenase (Unigene26663 and Unigene3804) was also increased in Na+-stressed suspension cells; this enzyme is not only involved in the glycolytic pathway, but also mediates plant metabolism and development under salt stress (Chang et al., 2015).

### Stress Defense and Signal Transduction-Related Proteins

By regulating the expression of specific, stress-related genes and metabolites to counter ion imbalance, hyperosmotic stress, ROS production and oxidative damage (Huang et al., 2012), salt-induced defense responses, and signal transduction are crucial for salt tolerance. In our study, one of the most remarkable changes observed was the greater number of upregulated proteins identified in stress defense after salt treatment. Of these, five proteins were determined to be pathogenesisrelated proteins (PR proteins; CL3714.Contig2, CL4645.Contig1, Unigene1148, Unigene13991, and Unigene15500). Furthermore, the activation of six proteins: defense-related hydrolase chitinase 3 (CL3889.Contig1, Unigene26473), beta-1, 3-glucanase 31 (CL4645.Contig1), acidic endochitinase SP2 (CL39.Contig1), endochitinase PR4 (Unigene14744), and glycosyl hydrolase family 18 protein (CL3889.Contig2), was also detected after salt stress. PR proteins are usually induced by various abiotic and biotic stresses, which play a crucial role in plant defense (Liu and Ekramoddoullah, 2006). For instance, overexpression of PR protein genes enhanced tolerance to heavy metal and pathogen stresses (Sarowar et al., 2005) and to drought stress environments in tobacco (Jain et al., 2012).

Up-regulated glutathione transferase (Unigene22454, Unigene 3942) and glutathione peroxidase (GSH-Px) are detoxifying enzymes, belonging to most families of the cell's antioxidant defense system. NaCl induces a quinone oxidoreductase-like protein (CL769.Contig1) that detoxifies quinones and their derivatives in leaves of salt-treated tomato plants (Zhou et al., 2009). Thaumatin-like protein (CL3555.Contig1) and osmotinlike protein (Unigene4336) are associated with osmotic adaption under salt stress (Ramos et al., 2015). A methionine sulfoxide reductase A4 (Unigene12027) repairs oxidized methionine and protects organisms from oxidative damage caused by ROS. In addition to these proteins, five stress response proteins, including putative prolyl aminopeptidase (Unigene15105), serine carboxypeptidase (Unigene25730), hypersensitive-induced response protein (Unigene2976) were identified as being upregulated after salt stress.

Understanding salt-responsive signaling pathways is currently a hot topic in salt stress research. Here, three proteins, phosphatase 2C (CL2317.Contig1), proline-rich receptor-like protein kinase (CL3178.Contig2), and phospholipase D (CL5365. Contig1), were identified as having increased after NaCl stress. Previous reports have shown that a phosphatase 2C family protein phosphatase mediates ABA (abscisic acid)-regulated stomatal movement, water loss during leaf senescence, and abiotic stress tolerance in Arabidopsis (Zhang and Gan, 2012; Singh et al., 2015). The proline-rich receptor-like protein kinase belongs to the receptor-like protein kinase family, which may be part of a response to external challenges presented by an ever-changing environment (Morris and Walker, 2003). In short, the changes observed in this study suggest that proteins related to stress defense and signal transduction may be of particular importance for salt tolerance in cells.

### Proteins Related to Other Metabolic Systems

The balance between synthesis and degradation of proteins plays an important role in a plant's survival during abiotic adaptation (Hinkson and Elias, 2011). Generally, protein synthesis is inhibited by salt stress altering protein metabolic pathways (Zhang et al., 2011). From our iTRAQ data, we observed 44 proteins involved in protein translation, processing, and degradation were differently co-expressed after NaCl treatment. Of these differentially expressed proteins, only two

were identified as up-regulated proteins, including aspartate aminotransferase (CL4696.Contig1), and elongation factor P (Unigene25772).

The cell's response to osmotic stress includes changing cell size and morphology to maintain turgor and this plays an important role in its survival under salt stress (Chen et al., 2012). We found that levels of the majority of cell growth and cytoskeleton-related proteins decreased in the presence of salinity (Tables S1,S2). The four up-regulated, growth-related proteins in response to NaCl stress were identified as auxin-induced atb2 (CL1157.Contig5), auxin-induced, protein PCNT115-like isoform 1 (CL6230.Contig2, Unigene17232), and indole-3-acetic acid-amido synthetase GH3.1-like (CL3188.Contig1). Of these, auxin-induced protein PCNT115-like isoform 1 was detected in a proteomics analysis in response to stimulation (Margaria et al., 2013) and development (Zhang et al., 2015). The indole-3-acetic acid-amido synthetase GH3.1 conjugates indole-3-acetic acid (IAA) to amino acids in an effort to maintain appropriate IAA concentrations in plants in order to regulate growth and development (Böttcher et al., 2010). Additionally, high intracellular Na<sup>+</sup> levels evidently induced changes in the abundance of proteins involved in metabolism and secondary metabolism.

### Proteins Related to Ion Homeostasis under Salinity Stress

For ion homeostasis in the cell under salinity stress, ions that are extruded from the cytoplasm across the plasma membrane and compartmentalized in vacuoles cross the tonoplast using transporters or channels. This not only effectively alleviates the toxic effect of Na<sup>+</sup> and Cl−, but also markedly increases cellular osmolarity to counter osmotic stress (Flowers and Colmer, 2008). Generally speaking, the plasma membrane Na+/H<sup>+</sup> exchanger (SOS 1) and tonoplast Na+/H<sup>+</sup> exchanger (NHX) are considered the main transporters mediating the efflux and compartmentalization of Na<sup>+</sup> (Garcia de la Garma et al., 2015). Furthermore, increasing the activity of the H+-pump and V-ATPase in response to salinity correlates with an improvement in Na+/H<sup>+</sup> antiport activity (Adolf et al., 2013). We have identified and summarized tonoplast H<sup>+</sup> pumps (**Figure 4A**), plasma membrane H<sup>+</sup> pumps (**Figure 4B**), transporters (**Figure 4C**), and channels (**Figure 4D**) related to transmembrane transport of ions in this study (**Figure 4**). Contrary to our expectations, none of these showed significantly changed abundances after salt stress. Of 21 proteins in the NaCl-stressed proteome, the tonoplast H<sup>+</sup> pumps subfamily was the largest group, which probably provides energy required for vacuolar Na<sup>+</sup> deposition; a relatively low abundance level of Na+/H<sup>+</sup> antiporter was detected and this was partly consistent with our previous studies of the transcriptome (Wang et al., 2015a) and proteome (Wang et al., 2015b) of the salt stress response in H. glomeratus. Similarly, in Arabidopsis, tonoplast-localized NHX proteins mediate active potassium uptake into vacuoles but do not enhance the ability to compartmentalize Na<sup>+</sup> in vacuoles (Barragán et al., 2012). Furthermore, for Na<sup>+</sup> extrusion, the abundances of plasma protein H+-ATPase and SOS1 were slightly decreased in salt stress. Obviously, the assumption of the plasma membrane Na+/H<sup>+</sup> exchanger (SOS 1) and tonoplast Na+/H<sup>+</sup> exchanger (NHX) mediating the efflux of Na<sup>+</sup> from the cytoplasm and its compartmentalization into the vacuole is not in accordance with proteomic results (**Figure 5** and Table S3). Meanwhile, recently in tobacco salt-acclimated cells, the subcellular analysis showed that Na<sup>+</sup> compartmentalization in the cell vacuoles could be mediated through vesicle trafficking, so avoiding Na<sup>+</sup> toxicity in the cytoplasm (Garcia de la Garma et al., 2015). Our quantitative proteomics data cannot fully explain the current view of Na<sup>+</sup> extrusion and compartmentalization in H. glomeratus. Further work is needed to better understand the mechanisms of Na<sup>+</sup> transport, including the Na<sup>+</sup> efflux system of the plasma membrane and the Na<sup>+</sup> uptake system of the tonoplast, using both functional assays and proteomics of subcellular compartments such as the tonoplast and plasma membrane proteomics analysis of H. glomeratus under salt stress.

#### CONCLUSION

In conclusion, to our knowledge, this is the first report that identifies a large number of proteins associated with a salt response after treatment, with different concentrations of NaCl, of suspension cell cultures of the halophyte, H. glomeratus. Notably, suspension cell cultures of H. glomeratus showed a similar halophytic growth response to that of whole plants. We, here, have provided a comprehensive perspective on the salinity response of suspension cell cultures of H. glomeratus using a comparative proteomic approach. A total of 87 proteins were similarly up-regulated when exposed to two different NaCl concentrations; these proteins were mainly involved in energy, carbohydrate metabolism, stress defense, protein metabolism, signal transduction, cell growth, and cytoskeleton metabolism. In addition, known proteins that regulate the efflux of Na<sup>+</sup> from the cytoplasm and its compartmentalization into the vacuole

#### REFERENCES


were not significantly changed under salinity stress. Whether suspension cells of H. glomeratus have specific transporters that are responsible for Na<sup>+</sup> compartmentalization requires further study. Taken together, our results have contributed to crucial insights into the mechanisms involved in a salinity response by the halophytic H. glomeratus.

#### AUTHOR CONTRIBUTIONS

JW and LY prepared suspension cell cultures of H. glomeratus samples for iTRAQ-based proteomic analysis. YL, BL, YM, and XM performed the general statistical analysis on the proteomics data. ES, KY, and PR participated in interpreting physiological characteristics results. JW, LY, and YL also involved in proteomics analysis and wrote the manuscript. HW and XS designed the experiment and provided guidance on the whole study. All authors have read and approved the final manuscript.

#### ACKNOWLEDGMENTS

This work was supported by National Basic Research Program of China (973 program, 2014CB160313), National Natural Science Foundation of China (No.31171558, 31460347), China Agriculture Research System (CARS-05), and Fostering Foundation for the Excellent Ph.D. Dissertation of Gansu Agricultural University (YBPY2014001).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 00110


aluminum stress tolerance mechanisms in plants. J. Proteomic 98, 189–205. doi: 10.1016/j.jprot.2013.12.023


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Wang, Yao, Li, Meng, Ma, Lai, Si, Ren, Yang, Shang and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Comparative Proteomic Analysis Reveals Differential Root Proteins in *Medicago sativa* and *Medicago truncatula* in Response to Salt Stress

Ruicai Long<sup>1</sup> , Mingna Li <sup>2</sup> , Tiejun Zhang<sup>1</sup> , Junmei Kang<sup>1</sup> , Yan Sun<sup>2</sup> , Lili Cong<sup>1</sup> , Yanli Gao<sup>1</sup> , Fengqi Liu<sup>3</sup> and Qingchuan Yang<sup>1</sup> \*

*1 Institute of Animal Sciences, Chinese Academy of Agricultural Sciences, Beijing, China, <sup>2</sup> Department of Grass and Forage Science, College of Animal Science and Technology, China Agricultural University, Beijing, China, <sup>3</sup> Institute of Pratacultural Science, Heilongjiang Academy of Agricultural Sciences, Haerbin, China*

#### *Edited by:*

*Qingsong Lin, National University of Singapore, Singapore*

#### *Reviewed by:*

*Pannaga Krishnamurthy, National University of Singapore, Singapore Chien-Chen Lai, National Chung Hsing University, Taiwan*

*\*Correspondence:*

*Qingchuan Yang qchyang66@163.com; xms\_grass168@163.com*

#### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

*Received: 28 October 2015 Accepted: 18 March 2016 Published: 31 March 2016*

#### *Citation:*

*Long R, Li M, Zhang T, Kang J, Sun Y, Cong L, Gao Y, Liu F and Yang Q (2016) Comparative Proteomic Analysis Reveals Differential Root Proteins in Medicago sativa and Medicago truncatula in Response to Salt Stress. Front. Plant Sci. 7:424. doi: 10.3389/fpls.2016.00424* Salt stress is an important abiotic stress that causes decreased crop yields. Root growth and plant activities are affected by salt stress through the actions of specific genes that help roots adapt to adverse environmental conditions. For a more comprehensive understanding of proteins affected by salinity, we used two-dimensional gel electrophoresis and mass spectrometry to characterize the proteome-level changes associated with salt stress response in *Medicago sativa* cv. Zhongmu-1 and *Medicago truncatula* cv. Jemalong A17 roots. Our physiological and phenotypic observations indicated that Zhongmu-1 was more salt tolerant than Jemalong A17. We identified 93 and 30 proteins whose abundance was significantly affected by salt stress in Zhongmu-1 and Jemalong A17 roots, respectively. The tandem mass spectrometry analysis of the differentially accumulated proteins resulted in the identification of 60 and 26 proteins in Zhongmu-1 and Jemalong A17 roots, respectively. Function analyses indicated molecule binding and catalytic activity were the two primary functional categories. These proteins have known functions in various molecular processes, including defense against oxidative stress, metabolism, photosynthesis, protein synthesis and processing, and signal transduction. The transcript levels of four identified proteins were determined by quantitative reverse transcription polymerase chain reaction. Our results indicate that some of the identified proteins may play key roles in salt stress tolerance.

Keywords: *Medicago*, salt stress, root, protein, 2-DE, gene expression, function

#### INTRODUCTION

Plant growth and productivity are adversely affected by various natural abiotic and biotic factors, which cause considerable crop losses worldwide. These factors prevent plants from reaching their full genetic potential and limit crop productivity (Cramer et al., 2011). Salt stress is an important abiotic factor in many parts of the world, especially on irrigated lands (Munns and Tester, 2008). Soil salinity is a major abiotic stress influencing crop production, and researchers have investigated plant salt tolerance mechanisms with the aim of improving crop plants (Duzan et al., 2004). The metabolic imbalances caused by ion toxicity, osmotic stress, and nutritional deficiency due to salinity may also lead to oxidative stress (Zhu, 2002). These negative effects trigger changes to root morphology and suppression of plant growth, and can ultimately result in plant death.

Salinity regulates the expression of many plant genes at the transcriptional and post-translational levels. The molecular mechanism of plant salt tolerance is very complex (Zhu, 2001, 2002; Munns and Tester, 2008). To investigate this mechanism, several studies have been conducted in many plant models. Published analyses have helped characterize the expression profiles of many genes and proteins involved in salt stress responses in Arabidopsis thaliana, rice, wheat, soybean, tobacco, barrel medic (Medicago truncatula), and other plant species (Merchan et al., 2007; Cheng et al., 2009; Kumari et al., 2009; Razavizadeh et al., 2009; Zhang et al., 2009; Sobhanian et al., 2010; Capriotti et al., 2014; Ghaffari et al., 2014). The root is the primary tissue involved in salinity perception and is one of the first to be injured following exposure to several types of stresses. The sensitivity of the root to stress often limits the productivity of the entire plant (Steppuhn et al., 2010). Therefore, a comprehensive understanding of root molecular responses to salt stress is necessary for researchers to be able to increase crop tolerance to salt stress.

Plants differ considerably in their tolerance to salinity, as reflected by their different growth responses. For instance, several legumes, including Medicago sativa (alfalfa) and M. truncatula, have cultivars that have adapted to saline soils. This adaptive process is associated with a number of biochemical and physiological changes. The majority of these modifications are regulated by salt through alterations in gene expression (Zhu, 2002; Munns and Tester, 2008). Proteomics-based technologies have become powerful tools in the study of protein expression (Faurobert et al., 2007). For example, the combination of twodimensional gel electrophoresis (2-DE) and mass spectrometry has been one of the most widely used techniques to study plant proteomes. Investigations into plant proteome changes during exposure to salt stress have been conducted for many plants, such as A. thaliana (Jiang et al., 2007), tomato (Manaa et al., 2011), soybean (Sobhanian et al., 2010), rice (Zhang et al., 2009; Ghaffari et al., 2014), tobacco (Razavizadeh et al., 2009), durum wheat (Capriotti et al., 2014), and barley (Witzel et al., 2014).

In this study, we explored new potential regulatory proteins of salt stress tolerance in M. sativa and M. truncatula roots. Many M. truncatula cultivars are more salt sensitive than M. sativa. Some M. sativa cultivars are highly tolerant to salinity stress (Munns and Tester, 2008), particularly M. sativa L. cv. Zhongmu-1, which we developed. High salt concentrations surrounding plant roots can induce rapid changes to cell growth and associated metabolic activities. The accumulation of salts inside plants can be toxic (Munns and Tester, 2008). There have been many genetic studies of Medicago species, but most of them focused on the model legume, M. truncatula. The objective of our study was to identify novel proteins regulated by salt stress in M. sativa and M. truncatula roots. We also aimed to determine differences in protein expression patterns between these two leguminous plants. We prepared total protein extracts from M. sativa and M. truncatula seedling roots treated with 300 mM NaCl and compared them to those of control roots using 2- DE. We identified novel salt stress-responsive root proteins and differentially expressed proteins.

### MATERIALS AND METHODS

#### Plant Materials

We used M. sativa cv. Zhongmu-1 and M. truncatula cv. Jemalong A17 in this study. Zhongmu-1 is a salt-tolerant cultivar of alfalfa (M. sativa, tetraploid, 2n = 4× = 32), which is widely cultivated in China. Jemalong A17 is a cultivar of M. truncatula (diploid, 2n = 16), which is salt-sensitive. The genome sequence of Jemalong A17 is already known. Seeds of both cultivars were surface-sterilized in 75% ethanol for 10 min followed by three washes with sterile water. Seeds were germinated on moistened Whatman filter paper placed in Petri dishes (10 cm diameter). After a week, the seedlings were transferred to hydroponic cultures containing full-strength Hoagland's solution in a growth chamber with a 16 h/8 h light/dark photoperiod at 25◦C and 65% relative humidity. The Hoagland's solution was renewed every 3 days.

### Physiological Analysis

One-month-old Zhongmu-1 and Jemalong A17 seedlings were treated with Hoagland's solution supplemented with 300 mM NaCl for 0, 2, 8, 24, and 48 h. They were then analyzed for relative water content (RWC), electrolyte leakage, and proline content. The RWC was used to evaluate plant water status. Leaf RWC was calculated as RWC = (FW − DW) / (WS − DW), where FW refers to fresh weight, DW refers to dry weight, and WS refers to saturated water weight. Membrane damage was assessed by measuring electrolyte leakage. For each measurement, 10 g seedlings were added to 30 ml double deionized water in 50-ml tubes. Air was removed from the tubes using a vacuum pump until all seedlings were submerged in the water. Seedlings were maintained in the water for 4 h at 25◦C.

We measured the conductivity of the bathing solution with a conductivity meter (Mettler Toledo) (as L1). The tubes were then incubated at 100◦C for 15 min. The conductivity of the incubated solution was measured again after cooling to room temperature (as L2). For each sample, the relative conductivity (%) was calculated as L1/L<sup>2</sup> × 100. Samples treated with 300 mM NaCl were harvested, frozen at −80◦C and ground to a fine powder in liquid nitrogen using a mortar and pestle. We measured the free proline content using a colorimetric assay as described (Bates et al., 1973). Proline concentration was determined using a calibration curve and expressed as mg proline g−<sup>1</sup> FW.

### Sample Preparation and 2-DE

To identify M. sativa and M. truncatula proteins potentially involved in regulating salt tolerance, three independent replicates of 1-month-old seedling root samples were collected from Zhongmu-1 and Jemalong A17 plants treated with 300 mM NaCl for 8 h (Long et al., 2015). Untreated roots were used as controls. Root samples were ground to a fine powder in liquid nitrogen. Total protein extracts were prepared from the ground root samples using an optimized TRIzol method (Xiong et al., 2011). The final protein pellets were washed three times in 1 ml ethanol and resuspended in 1 ml lysis buffer (8 M urea, 4% w/v CHAPS, and 2% w/v DTT). Protein samples were sonicated for 10 min (4◦C) and incubated at room temperature for 2 h. The protein solutions were centrifuged (40,000 × g, 40 min, 4◦C) and the supernatants were collected. The protein concentrations of the supernatants were determined using the 2-D Quant kit according to the manufacturer's protocol (GE Healthcare). We diluted protein solutions with rehydration buffer [8 M urea, 2% w/v CHAPS, 1% w/v DTT, 0.5% v/v immobilized pH gradient (IPG) buffer pH 4–7, and 0.002% w/v bromophenol blue]. We then loaded 120 mg protein (in 450 µl) onto pH 4–7 IPG strips (24 cm). Isoelectric focusing (IEF) was completed using the Ettan IPGphorII system (GE Healthcare). The IEF and second dimension sodium dodecyl sulfate polyacrylamide gel electrophoresis were performed as described (Xiong et al., 2011). The IEF running conditions were as follows: 30 V for 12 h, 150 V for 250 Vh, 200 V for 300 Vh, 500 V for 250 Vh, 1000 V for 1000 Vh, 8000 V for 3 h, and 8000 V for a total of 30,000 Vh. Gel electrophoresis was performed using 12% polyacrylamide gels and the Ettan DALTsix electrophoresis gel system (GE Healthcare). The proteins were visualized using colloidal Coomassie brilliant blue G-250.

#### Protein Visualization and Image Analysis

The stained gels were scanned using a UMAX Power Look 2100XL scanner (UMAX) at a resolution of 600 dots per inch. Gel images were analyzed using ImageMasterTM 2D Platinum Version 5.0 (GE Healthcare Bio-Science). We estimated the isoelectric point (pI) of the proteins based on the relative migration of the protein spots on the IPG strips. All spot volumes were normalized as a percentage of the total volume of all spots present in the gel. We used ImageMasterTM 2D Platinum Version 5.0 to perform ANOVA. Comparisons of the mean differences were completed using Duncan's multiple range test at P < 0.05. The protein spots were determined to be significantly up- or down-regulated when the abundance fold change was more than 1.5 at P < 0.05.

### Protein Identification and Analysis of Function

Significantly up- or down-regulated protein spots were excised from gels and destained for 2 h at room temperature using a freshly prepared wash solution consisting of 100% acetonitrile, 50 mM NH4CHO<sup>3</sup> (50:50 v/v). Proteins were digested using a trypsin solution according to an established method (Xiong et al., 2011). Peptide mixtures were analyzed using the 4800 Plus MALDI TOF/TOFTM Analyzer (ABI), which is a matrix-assisted laser desorption ionization time of flight (MALDI-TOF/TOF) mass spectrometer. Mass spectrometry was completed using an established method (Li et al., 2011). Proteins were identified using peak lists for searches against the NCBInr database with the Mascot search engine (http://www.matrixscience.com/). The search criteria consisted of the following: Enzyme, Trypsin; Variable modifications, Oxidation (M); Peptide tolerance, 200 ppm; MS/MS tolerance, 0.8 Da; Instrument, MALDI-TOF/TOF; and Carbamidomethyl (C) as a fixed modification for all alkylated samples. Blast2GO software was used for gene ontology and Kyoto Encyclopedia of Genes and Genomes pathway analyses of identified proteins (Conesa et al., 2005).

### Transcript Analysis Using Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR)

We used 1-month-old Zhongmu-1 and Jemalong A17 seedlings for transcript analyses. Total RNA was isolated from roots treated with NaCl (0, 2, 8, and 24 h) using TRIzol (Invitrogen, USA) according to the manufacturer's instructions. The RNA was then reverse transcribed and the synthesized cDNA was used as the template for qRT-PCR. The real-time fluorescent quantitative PCR was completed using the ABI 7500 system (Applied Biosystems). The β-actin gene served as a housekeeping gene to normalize target gene quantities. The real-time PCR primers used for the amplification of β-actin and the genes of ten identified proteins are listed in **Supplementary Table 1**. The PCR program consisted of a maximum of 40 cycles of 95◦C for 15 s and 60◦C for 30 s, followed by melting curve analysis. Transcript abundance for each gene was normalized to that of β-actin. The relative expression levels were calculated as follows: ratio = 2 <sup>−</sup>11Ct = 2 <sup>−</sup>[Ct,t−Ct,r], where Ct refers to cycle threshold, Ct,t refers to Ct of the target gene and Ct,r refers to Ct of the β-actin control gene.

## STATISTICAL ANALYSIS

All experiments were repeated with three independent biological replicates. All data obtained were subjected to a one-way ANOVA. The mean differences were compared using Duncan's multiple range t-test. Comparisons with P < 0.05 were considered significantly different. The values provided in the figures and tables are the means ± standard errors.

### RESULTS

### Physiological Parameters Related to Salt Tolerance

The responses of Zhongmu-1 and Jemalong A17 to salt stress were compared in terms of leaf RWC, electrolyte leakage, and proline content. We observed that the leaf RWC of Zhongmu-1 and Jemalong A17 decreased by about 10% and over 20%, respectively, after exposure to salt stress for 48 h (**Figure 1A**). Relative electrolyte conductivity and proline content increased dramatically in salt stressed plants (**Figure 1B**). Following salt treatment, the relative electrolyte conductivity of Zhongmu-1 was lower than that of Jemalong A17. Conversely, proline accumulation in Zhongmu-1 was higher than that of Jemalong A17 (**Figure 1C**). Acording to the phenotypic observation the wilting degree of Jemalong A17 seedlings was much more obvious than that of Zhongmu-1 seedlings after treating with 300 mM NaCl for 8 h (**Figures 1D–G**). The physiological and phenotypic observations confirmed that M. sativa cv. Zhongmu-1 is more salt-tolerant than M. truncatula cv. Jemalong A17.

#### Protein Responses to Salt Stress

The root is the first plant organ to be affected by salt stress. Representative 2-DE gel images of the Zhongmu-1 and Jemalong A17 root proteomes following salt treatment

are presented in **Figure 2**. There was a broad distribution of the proteins in terms of pI (4.0–7.0) and mass (10–70 kDa). Of the approximately 800 detected Zhongmu-1 protein spots, 93 exhibited significant changes to spot abundance (P < 0.05) (**Figure 2A**, **Supplementary Figure 1**). Fifty-three of these proteins were up-regulated by salt stress, with the remaining 40 being down-regulated (**Supplementary Table 2**, **Figure 3**). Of the approximately 900 detected Jemalong A17 protein spots, 30 protein spots exhibited significant changes to spot abundance (P < 0.05) (**Figure 2B**, **Supplementary Figure 1**). Twenty-two of these proteins were up-regulated by salt stress, with the remaining eight being down-regulated (**Supplementary Table 2**, **Figure 3**).

### Protein Identification

After mass spectrometric analysis, a total of 60 and 26 protein spots were identified in Zhongmu-1 and Jemalong A17, respectively (**Table 1**, S and T correspond to protein spots of Zhongmu-1 and Jemalong A17). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaino et al., 2014) partner repository with the dataset identifier PXD003761. These proteins were classified into five groups according to their molecular function as follows: Anti-oxidation, photosynthesis, metabolism, signal transduction, and protein synthesis and processing. Only the following proteins were identified in Zhongmu-1 and Jemalong A17: chaperonin CPN60-like protein (S44 and T2), fructose-bisphosphate aldolase (S3 and T1), and heat shock protein (S28 and T26). The abundance of chaperonin CPN60 like protein and fructose-bisphosphate aldolase increased in Zhongmu-1 and Jemalong A17 following salt treatment. However, the response of the heat shock protein corresponding to S28 and T26 differed between Zhongmu-1 and Jemalong A17. We observed an increase in S28 abundance and a decrease in T26 following salt treatment. Some proteins were identified in more than one spot on the same gel (**Table 1**). For example, S3 and S4 were identified as fructose-bisphosphate aldolase, S17 and S53 were identified as phosphopyruvate hydratase, S54 and S55 were identified as aconitate hydratase, and S61, S62, S63, and S64 were identified as a translation initiation factor (eIF-5A). Further examination of electrophoresis patterns indicated that the inferred mass and pI values of these spots differed, perhaps because of post-translational modifications or degradations.

## Functional Analysis of Identified Proteins

The identified protein sequences were blasted by BLASTP in the NCBI database. These identified proteins in Jemalong A17 were classified into 11 functional groups based on GO prediction (**Figure 4A**), including binding, catalytic activity, nucleotide binding, hydrolase activity, small molecule binding, protein binding, transferase activity, RNA binding, lipid binding, transporter activity, nucleic acid binding and not determined. These identified proteins in Zhongmu-1 were classified into 15 functional groups based on GO prediction (**Figure 4B**), including binding, catalytic activity, nucleotide binding, small molecule binding, transferase activity, hydrolase activity, protein binding, RNA binding, nucleic acid binding, translation factor activity, transporter activity, kinase activity, transferase activity, DNA binding, enzyme regulator activity and not determined. Molecule binding group and catalytic activity group were the two mainly functional groups.

respectively, showing at least a 1.5-fold change following 300 mM NaCl treatment (*P* < 0.05). M, protein marker.

#### Transcript Analysis of Selected Proteins

Ten differentially accumulated root proteins identified from Zhongmu-1 and Jemalong A17 were chosed to perform transcript expression analyses. The expression levels of fructosebisphosphate aldolase (S3/T1), heat shock protein (S28/T26), TCP-1/cpn60 chaperonin family protein (S44/T2), and cinnamyl alcohol dehydrogenase-like protein (S76) based on qRT-PCR analyses are provided in **Figure 4**. The expression analyses results of all the 10 genes are provided in **Supplementary Table 3**. Compared with the expression level at 0 h, transcript abundance of fructose-bisphosphate aldolase, heat shock protein, and TCP-1/cpn60 chaperonin family protein increased considerably after salt treatment in Zhongmu-1. The transcript abundance of cinnamyl alcohol dehydrogenase-like protein decreased significantly (P < 0.05) after salt treatment in Zhongmu-1, whereas that of Jemalong A-17 did not significantly change, though there was a decreasing trend (**Figure 5**). After salt treatment, the heat shock protein transcript abundance increased

more than 6-fold in Zhongmu-1, whereas that of Jemalong A-17 showed no significant changes, but did exhibit an increasing trend (**Figure 5**). These results along with those from the 2-DE analyses suggest that the transcript and protein level changes of most analyzed proteins were similar.

### DISCUSSION

The leaf RWC is one of the factors used to determine the extent of wilting after certain abiotic stresses. We observed that after salt treatment, the RWC of Zhongmu-1 was significantly higher than that of Jemalong A17, indicating Jemalong A17 was more wilted. Proline is an organic solute that helps regulate cellular osmolarity during plant responses to osmotic stress. Proline accumulation has been used as a drought-tolerance selection criterion related to membrane integrity in different plant species (Misra and Gupta, 2005). After salt treatment, Zhongmu-1 accumulated more proline than Jemalong A17. An increased rate of electrolyte leakage has been used as an indicator of cell membrane physical damage during exposure to abiotic stresses (Thiaw and Hall, 2004). Electrolyte leakage in Zhongmu-1 was lower than that of Jemalong A17, indicating that the cell membranes of Jemalong A17 were more damaged than those of Zhongmu-1. The physiological characteristics of Zhongmu-1 and Jemalong A17 were consistent with their salt tolerance levels. Based on their phenotypes, Zhongmu-1 was much more tolerant than Jemalong A17. The number of Zhongmu-1 protein spots exhibiting significant changes in abundance in response to salt stress was 3-fold higher than that of Jemalong A17. This result may be related to the fact that Zhongmu-1 is considerably more salt tolerant than Jemalong A17.

Some of the identified salt stress-regulated proteins were also reported in other plant species. For example, heat shock proteins, fructose-bisphosphate aldolase, peroxidase, DNA/RNA binding protein, and caffeoyl-CoA O-methyltransferase were detected in tomato roots after exposure to salt stress (Jiang et al., 2007; Manaa et al., 2011; Witzel et al., 2014). To evaluate the correlation between mRNA and the corresponding protein levels, the expression of 10 proteins with significant salt-induced changes in protein spot abundance was quantified by qRT-PCR.


#### TABLE 1 | Identities of salt-responsive proteins in *M. sativa* cv. Zhongmu-1 and *M. truncatula* cv. Jemalong A17 based on mass spectrometry and a Mascot search.



*† Species of the matched protein based on a Mascot search.*

*†† Mascot search score.*

*††† Spot volume fold change corresponding to spot volume after 8 h salt treatment/spot volume before salt treatment (0 h).*

significant difference at *p* < 0.05 (Student's *t*-test).

The qRT-PCR and 2-DE results suggest that the mRNA and protein level changes exhibit similar trends. These results also support the concept that post-transcriptional regulation plays an important role in stress-responsive gene expression, and indicates the importance of a combined transcriptomic and proteomic analyses (Mooney et al., 2006; Jiang et al., 2007).

We observed only a few salt stress-regulated proteins that were common between Zhongmu-1 and Jemalong A17 and there were differences in the expression patterns of these proteins. For intance, heat shock protein (70 kDa) was up-regulated in Zhongmu-1 (S28), but down-regulated in Jemalong A17 (T26). Heat shock proteins play important roles in a variety of cellular processes. They maintain proteins in their functional state and are also involved in protein translocations to subcellular compartments (Goswami et al., 2010). Heat shock proteins were previously reported to be up-regulated in tomato following exposure to cold stress (Page et al., 2010) and salt stress (Manaa et al., 2011).

Some proteins were identified in more than one spot in the same gel, such as fructose-bisphosphate aldolase (S3, S4, and S85) and phosphopyruvate hydratase (S17 and S53). Glycosylation, phosphorylation, and other post-translational modifications, which can alter the molecular weight and/or charge of proteins, may be responsible for these results. It is also possible that proteins were identified from multiple spots because of translation of alternatively spliced mRNAs (Yoshimura et al., 1999; Ndimba et al., 2005; Jiang et al., 2007). Proteomic studies have also shown that some proteins may be degraded during exposure to abiotic stress. For example, 19 different ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) large subunit fragments were detected in salt-treated rice roots (Yan et al., 2006). Similar phenomena have been reported in tomato root proteomes affected by salt stress (Manaa et al., 2011). The multiple fragments may also be the result of protein degradation by reactive oxygen species (ROS) during stress responses (Kingston-Smith and Foyer, 2000). Excess ROS can seriously disrupt normal plant metabolism through oxidative damage to lipids, proteins, and nucleic acids (Apel and Hirt, 2004; Askari et al., 2006; Bhushan et al., 2007). Anti-oxidative enzyme peroxiredoxin (T19) was detected in our study, which has been observed in responses to various abiotic stresses, including cold (Sarhadi et al., 2010), salinity (Ghaffari et al., 2014), and drought (Ali and Komatsu, 2006).

The proteins identified in this study are involved in various molecular processes. According to our results, some proteins associated with photosynthesis and metabolism were differentially expressed in Zhongmu-1 and Jemalong A17 following salt treatment. Ribulose-1,5-bisphosphate carboxylase/oxygenase (S1 and S2) is the most prevalent plant enzyme. It forms approximately 30-50% of the total soluble protein content in chloroplasts. The small subunit of Rubisco may be degraded because of oxidative stress (Sobhanian et al., 2010). Subsequently, the production of the large subunit may be inhibited. In our study, the abundance of the small (S1) and large (S2) subunits of Rubisco increased significantly in Zhongmu-1 8 h after salt treatment. This increase in abundance indicates that M. sativa can survive and photosynthesize even during moderate levels of salt stress. The increased activity of Rubisco subunits in tobacco and rice under salt stress has also been demonstrated (Kim et al., 2005; Razavizadeh et al., 2009). Therefore, it is possible that the accumulation of Rubisco in M. sativa reflects the increase in photorespiration during exposure to salt stress. Our results showed that the abundances of some proteins associated with energy production or transport, such as cytosolic malate dehydrogenase (S90) and glyceraldehyde-3-phosphate dehydrogenase (T2), were affected by salt in Zhongmu-1 and Jemalong A-17. Cytosolic malate dehydrogenase was reported to

#### REFERENCES


be responsive to salinity stress in A. thaliana roots (Jiang et al., 2007). However, the function of some identified proteins (such as S7, S12, and S30) are still unknown. Further research is needed to determine the functions of these proteins.

#### CONCLUSION

Our physiological and phenotypic observations confirmed that M. sativa cv. Zhongmu-1 is considerably more salt tolerant than M. truncatula cv. Jemalong A17. We used 2-DE to explore the changes in the root proteomes of these leguminous plants as a result of exposure to salt stress. Differentially accumulated proteins identified in Zhongmu-1 and Jemalong A17 were determined to be involved in various molecular processes, most of which belonged to molecule binding and catalytic activity. Some of the identified proteins were validated or predicted to play critical roles in salt stress regulation. The identification of salt-responsive proteins provides new insights into salt stress responses and the basis for further studies to improve the salt tolerance of alfalfa and other plants.

### AUTHOR CONTRIBUTIONS

RL and QY Conceived and designed the experiments, RL, ML, and TZ performed the experiments, JK, YS, LC, YG, FL contributed to data analysis, RL, ML, and QY wrote the manuscript.

#### ACKNOWLEDGMENTS

This work was supported by the China Forage and Grass Research System (CARS-35-04), National Key Basic Research Program of China (2015CB943500), and Basic Scientific Research Fund of IAS-CAAS (2014ywf-zd-2).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 00424

Supplementary Figure 1 | The 2-DE images of 3 biological replicates for control and the treated samples.

Supplementary Table 1 | Primers used for qRT-PCR to analyze transcript abundance of four proteins identified during 2-DE and mass spectrometry.

Supplementary Table 2 | Protein spot volume fold changes of all proteins identified in Zhongmu-1 and Jemalong A17.

Supplementary Table 3 | Transcript expression analyses results of 10 differentially accumulated root proteins in Zhongmu-1 and Jemalong A17.


expressed proteins in chickpea extracellular matrix during dehydration stress. Mol. Cell. Proteomics 6, 1868–1884. doi: 10.1074/mcp.M700015-MCP200


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer, PK, and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2016 Long, Li, Zhang, Kang, Sun, Cong, Gao, Liu and Yang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Proteomic Response of Hordeum vulgare cv. Tadmor and Hordeum marinum to Salinity Stress: Similarities and Differences between a Glycophyte and a Halophyte

Lucie Maršálová<sup>1</sup> , Pavel Vítámvás<sup>2</sup> , Radovan Hynek<sup>1</sup> , Ilja T. Prášil<sup>2</sup> and Klára Kosová<sup>2</sup> \*

<sup>1</sup> Department of Biochemistry and Microbiology, Faculty of Food and Biochemical Technology, University of Chemistry and Technology, Prague, Czech Republic, <sup>2</sup> Laboratory of Plant Stress Biology and Biotechnology, Division of Crop Genetics and Breeding, Crop Research Institute, Prague, Czech Republic

#### Edited by:

Dipanjana Ghosh, National University of Singapore, Singapore

#### Reviewed by:

Abu Hena Mostafa Kamal, University of Texas at Arlington, USA Arkadiusz Kosmala, Institute of Plant Genetics of the Polish Academy of Sciences, Poland

> \*Correspondence: Klára Kosová kosova@vurv.cz

#### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 26 April 2016 Accepted: 19 July 2016 Published: 03 August 2016

#### Citation:

Maršálová L, Vítámvás P, Hynek R, Prášil IT and Kosová K (2016) Proteomic Response of Hordeum vulgare cv. Tadmor and Hordeum marinum to Salinity Stress: Similarities and Differences between a Glycophyte and a Halophyte. Front. Plant Sci. 7:1154. doi: 10.3389/fpls.2016.01154 Response to a high salinity treatment of 300 mM NaCl was studied in a cultivated barley Hordeum vulgare Syrian cultivar Tadmor and in a halophytic wild barley H. marinum. Differential salinity tolerance of H. marinum and H. vulgare is underlied by qualitative and quantitative differences in proteins involved in a variety of biological processes. The major aim was to identify proteins underlying differential salinity tolerance between the two barley species. Analyses of plant water content, osmotic potential and accumulation

of proline and dehydrin proteins under high salinity revealed a relatively higher water saturation deficit in H. marinum than in H. vulgare while H. vulgare had lower osmotic potential corresponding with high levels of proline and dehydrins. Analysis of proteins soluble upon boiling isolated from control and salt-treated crown tissues revealed similarities as well as differences between H. marinum and H. vulgare. The similar salinity responses of both barley species lie in enhanced levels of stress-protective proteins such as defense-related proteins from late-embryogenesis abundant family, several chaperones from heat shock protein family, and others such as GrpE. However, there have also been found significant differences between H. marinum and H. vulgare salinity response indicating an active stress acclimation in H. marinum while stress damage in H. vulgare. An active acclimation to high salinity in H. marinum is underlined by enhanced levels of several stress-responsive transcription factors from basic leucine zipper and nascent polypeptide-associated complex families. In salt-treated H. marinum, enhanced levels of proteins involved in energy metabolism such as glycolysis, ATP metabolism, and photosynthesis-related proteins indicate an active acclimation to enhanced energy requirements during an establishment of novel plant homeostasis. In contrast, changes at proteome level in salt-treated H. vulgare indicate plant tissue damage as revealed

by enhanced levels of proteins involved in proteasome-dependent protein degradation and proteins related to apoptosis. The results of proteomic analysis clearly indicate differential responses to high salinity and provide more profound insight into biological mechanisms underlying salinity response between two barley species with contrasting salinity tolerance.

Keywords: glycophyte, halophyte, salinity, proteome, stress acclimation, stress damage, Hordeum marinum, Hordeum vulgare

#### INTRODUCTION

Salinity represents an important threat to agricultural production worldwide, especially in hot arid and semi-arid areas. It is estimated that salinity affects ca 7% of agricultural land globally while in irrigated agriculture, it is nearly one third of irrigated aricultural land and the salt-affected area is steadily increasing. Plants exposed to increased salt levels generally reveal a twophase salinity response: first, a non-specific osmotic effect, i.e., cellular dehydration caused by a decreased water potential of salt solution; and second, a salt-specific ionic effect, i.e., an enhanced salt ion influx into cell cytoplasm leading to significant alterations in cell ion homeostasis (Zhu, 2001; Munns, 2002). Osmotic effect is common to all dehydration stresses and leads to an accumulation of osmolytes in cell cytoplasm to counteract the adverse effects on plant water uptake. In contrast, ionic effect is specific for salinity stresses and lies in enhanced concentrations of salt ions, namely Na+, in soil water solution which results in enhanced salt ion levels in cell cytoplasm resulting in an activation of salt ion exclusion and compartmentation mechanisms (Munns, 2002; Munns and Tester, 2008).

Barley (H. vulgare) is a relatively dehydration- and salttolerant cereal, the most tolerant of cultivated Triticeae, which can grow up to 250 mM NaCl (Colmer et al., 2006). However, several wild Triticeae species are halophytes, i.e., they can grow in areas with salt concentrations approaching those in sea water, i.e., around 500 mM NaCl, and can complete their life cycle at salinities above 200 mM NaCl (Flowers, 2004; Colmer et al., 2006). In Triticeae, crown tissues are crucial for the whole plant survival. Sea barleygrass (H. marinum) is a wild halophytic Triticeae species growing in salt marshes and other coastal areas in Mediterranean region and the Middle East. Previous experiments aimed at an investigation of H. marinum salt response mechanisms revealed that H. marinum maintained low levels of salt ions in the shoot indicating efficient ion exclusion mechanisms at the level of xylem transport (Garthwaite et al., 2005; Islam et al., 2007). In contrast, cultivated barley H. vulgare reveals relatively high Na<sup>+</sup> levels in shoots when exposed to salinity resulting in enhanced Na<sup>+</sup> vacuolar compartmentation. These processes seem to be compensated by cytosolic osmolyte accumulation leading to decreased shoot osmotic potential levels (Garthwaite et al., 2005).

Proteins play a crucial role in plant stress response since they are directly involved in the processes aimed at an enhancement of stress tolerance. Similarly to other stresses, salinity induces complex adaptations including an enhanced abundance of several stress-related proteins (dehydration-induced proteins, ion transporters, and ROS scavenging enzymes) as well as changes in cell signaling, gene expression, cellular metabolism, and regulatory processes. Despite an increasing number of proteomic studies aimed at crop stress response (reviewed in Kosová et al., 2011, 2013a,b, 2014a; Komatsu et al., 2014), there have been published only a few studies aimed at proteome response to salinity in halophytes (reviewed in Kumari et al., 2015) as well as a comparison of proteome response between related plant species with contrasting salt tolerance (a glycophyte vs a halophyte; reviewed in Kosová et al., 2013d). The few studies aimed at comparison of salt response between a glycophyte and a halophyte at transcript and protein levels include a comparison of Arabidopsis thaliana and Thellungiella halophila (Taji et al., 2004; Gong et al., 2005; Pang et al., 2010), common wheat Triticum aestivum and T. aestivum-Thinopyrum ponticum amphiploid (Wang et al., 2008; Peng et al., 2009), and rice and its wild relative Porteresia coarctata (Sengupta and Majumder, 2009). The results of the few studies indicate that salt-tolerant species revealed higher constitutive levels of stress-related proteins (ROS scavenging enzymes, salt ion transporters SOS1, V-ATPase), and, under stress, thus they revealed relatively higher levels of anabolism-related proteins (RubisCO activase and other proteins involved in photosynthesis such as OEE proteins) than salt-sensitive species. In contrast, salt-sensitive species reveal relatively higher levels of catabolism-related proteins such as

**Abbreviations:** ALDO, fructose-bisphosphate aldolase; APX, ascorbate peroxidase; bZIP, basic leucine zipper (protein motif); CBS, cystathione β synthase; cdc48, cell division cycle 48 (protein); DHN, dehydrin; eEF, eukaryotic elongation translation factor; eIF, eukaryotic initiation translation factor; ENO, enolase; ENR, enoyl(acyl-carrier-protein) reductase (NADH); ESI, electrospray ionization; Fv/Fm, variable-to-maximum chlorophyll fluorescence; GAP, GTPase activating factor; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GEF, GDP/GTP exchange factor; GRP, glycine-rich protein; HMC, Hordeum marinum, control plants; HMN, H. marinum, salt-treated plants; HSP, heat-shock protein; HVC, Hordeum vulgare, control plants; HVN, H. vulgare, salt-treated plants; LC-MS, liquid chromatography mass spectrometry; LEA, late embryogenesisabundant (protein); NAC, nascent polypeptide associated complex; nsLTP, non-specific lipid transfer protein; OEC, oxygen evolving complex; OEE, protein component of oxygen evolving complex; PCD, programmed cell death; PDI, protein disulfide isomerase; PGK, 2,3-bisphosphoglycerate kinase; PGM, 2,3 bisphosphoglycerate-independent phosphoglycerate mutase; POX, peroxidase; PP2C, protein phosphatase 2C; PPI, peptidyl-prolyl cis-trans isomerase; PR, pathogenesis-related (protein); Prx, peroxiredoxin; PTM, posttranslational modification; PWC, plant water content; RC PSII, reaction centre of photosystem II; ROS, reactive oxygen species; SAM, S-adenosylmethionine; SOD, superoxide dismutase; STI, stress-induced (protein); TCA, tricarboxylic acid (cycle); TCTP, translationally-controlled tumour protein; TF, transcription factor; TPI, triose phosphate isomerase; TPR, tetratricopeptide repeat (protein motif); USP, universal stress protein; WD, tissue dry weight; WF, tissue fresh weight; WT, tissue turgescent weight; WSD, water saturation deficit; ZFHD, zinc-finger homeodomain (protein motif); ψ<sup>s</sup> , osmotic potential

glycolytic and respiratory enzymes. However, most important aspects of the differential salinity response between glycophytes and halophytes still remain unexplored.

The manuscript represents the first study aimed at a comparison of proteome response to high salinity (300 mM NaCl) in cultivated barley H. vulgare and its halophytic relative H. marinum. High salinity (300 mM NaCl) was applied in order to induce stress response in both H. vulgare and H. marinum. First, both salt acclimation treatments as well as one-step direct transfers from control to high salinity conditions were applied in order to investigate plant stress response by using simple physiological assessments of tissue dehydration (WSD; PWC; leaf sap osmotic potential ψs), osmolyte accumulation (proline, dehydrins) and impairment of photosynthetic apparatus (maximum quantum yield efficiency Fv/Fm). Next, an one-step transfer from control to high salinity (0–300 mM NaCl) was chosen for proteomic analysis since only this kind of salinity treatment induced a significant stress response in salt-tolerant H. marinum. An analysis of boiling-stable proteins enriched fractions led to an identification of several stress-responsive proteins underlying differential salinity tolerance in salt-sensitive H. vulgare and salt-tolerant H. marinum, respectively, which would help us to understand differential mechanisms employed in salinity response by related species of glycophytes and halophytes.

#### MATERIALS AND METHODS

#### Plant Material and Growth Conditions

Hordeum vulgare cv. Tadmor is a black-seeded barley cultivar derived from a Syrian landrace Arabi Aswad by ICARDA, Aleppo, Syria. H. marinum ssp. gussoneanum is a wild halophytic barley species adapted to salt marshes and other habitats with enhanced soil salinity. Seeds of H. vulgare cv. Tadmor were obtained from Dr. L. Holková, Mendel University, Brno, Czech Republic. Seeds of H. marinum ssp. gussoneanum, accession H818 from Iran, were obtained from Nordic Gene Bank, Alnarp, Sweden. Seeds of both barley species were surface sterilized with 1% (m/v) sodium hypochlorite for 5 min followed by rinsing with distilled water. The seeds were let germinate on moist filter paper in the dark for 5 days under temperature of 20◦C and then plants were grown hydroponically in continuously aerated pots in a growth chamber (Tyler T-16/4, Budapest, Hungary) under an irradiance of 350 µmol m−<sup>2</sup> s −1 , a 12 h photoperiod, and temperature of 20◦C. The pots were filled with a commercially available solution Hydropon (Lovochemie, Lovosice, Czech Republic) corresponding to the Hoagland 3 nutrient solution including microelements, the final dilution was 1:200 (v/v); pH of the solution was adjusted to 6.5 by addition of KOH as described previously (Prášil et al., 2005). The hydroponical solution was changed every 3rd day in order to prevent nutrient depletion. All plants were first exposed to the control treatment of 0.2 mM NaCl (osmotic potential of the control solution ψ = −0.095 MPa) for 7 days similarly to previous publications (Islam et al., 2007; Kosová et al., 2015) due to a halophytic nature of H. marinum. NaCl was added to the hydroponical solution every day in a gradually increasing manner by 50 mM NaCl up to the final concentrations of 100 mM NaCl (100; a moderate salt stress ψ = −0.608 MPa) and 300 mM NaCl (300; a high salt stress ψ = −1.64 MPa), respectively. The alternative way represented direct transfers from 0.2 to 300 mM NaCl (0.2–300) and from 100 to 300 mM NaCl (100–300). All variants were sampled after 7 days of reaching 300 mM NaCl and the transfers, i.e., after 20 days of hydroponic cultivation. A scheme of the whole experiment is given in Supplementary Figure S1.

### Determination of Physiological Parameters and Dehydrin Relative Accumulation

Determination of WSD, osmotic potential (ψs), proline accumulation, chlorophyll fluorescence parameter maximum quantum yield of photosystem II photochemistry (Fv/Fm) and dehydrin relative accumulation was performed as described previously in Kosová et al. (2015). Youngest fully developed leaves were used for the analyses. Briefly, WSD was determined according to Slavík (1963) as WSD (%) = (WT–WF)/(WT– WD) ∗ 100 where W<sup>D</sup> was dry weight, W<sup>F</sup> was fresh weight and W<sup>T</sup> was turgescent weight of the leaf sample. PWC was determined as ml (g) of water per 1 g of dry mass and it was calculated as follows: PWC = (WF–WD)/(WD).

Leaf osmotic potential (ψs) was determined as osmolarity using VAPRO Dew Point Osmometer (WESCOR Inc., MtLogan, UT, USA). Osmotic potential values were calculated using van't Hoff equation ψ<sup>s</sup> = −n<sup>i</sup> ∗ ci ∗R <sup>∗</sup>T where n<sup>i</sup> is the molar amount of the dissociated ions, c<sup>i</sup> is a concentration of a given solute, R is an universal gas constant and T is a thermodynamic temperature (K).

Proline accumulation was determined spectrophotometrically at 520 nm as proline-ninhydrin complex extracted from leaf tissue using a mixture of 2.08 M orthophosphoric acid and 13.9 M acetic acid, a 60 min heating step (90◦C) and elution into toluene overnight according to Jiménez-Bremont et al. (2006) using <sup>L</sup>-proline as a standard (Sigma–Aldrich).

Chlorophyll fluorescence parameter maximum quantum yield of photosystem II photochemistry (Fv/Fm) was determined on dark-adapted plants using a portable fluorometer FluorPen FP100 (PhotonSystems Instruments, Drasov, Czech Republic).

Dehydrin protein relative accumulation was determined using boiling-enriched soluble protein fractions extracted from leaf (L) and crown (C) tissues using 0.1 M Tris-HCl extraction buffer, pH 8.5 with added protease inhibitor (Complete EDTA-free Protease Inhibitor Cocktail Tablets, Roche, Basel, Switzerland). A boiling step was 10 min followed by acetone precipitation. Dry protein pellets were dissolved in Laemmli sample buffer (Bio-Rad, Hercules, CA, USA), concentration of the dissolved proteins was determined using RC DC Protein kit (Bio-Rad) and the samples were loaded on 1D SDS-PAGE gels (12.5% resolving gel) using equal protein loading with respect to protein concentration in plant tissue (fresh weight). For estimation of apparent (electrophoretic) MW of dehydrin proteins, protein marker All Blue Precision Plus Protein Marker Standard (Bio-Rad)

was loaded onto the gels. Proteins resolved on 1D SDS-PAGE gel were transferred onto nitrocellulose membrane and were incubated with anti-dehydrin primary antibody raised against dehydrin K-segment (Enzo Life Sciences, Farmingdale, NY, USA) resolved in Tween-20 Tris-buffered saline (TTBS) and then in anti-rabbit alkaline phosphatase and the resulting complexes were visualized using nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl-phosphate (NBT/BCIP) colorimetric detection (Bio-Rad Manual). Immunoblot membranes with visualized dehydrin bands were scanned using a densitometer GS-800 (Bio-Rad) and densitometric analysis of detected dehydrin bands was carried out using Quantity One software, version 4.6.2 (Bio-Rad). Dehydrin band density was calculated relative to a mixed sample containing equal amounts of all samples used in the analysis and loaded on each gel as an internal standard.

Statistical analysis of the data on WSD, ψ<sup>s</sup> , Fv/Fm, proline and dehydrin relative accumulation was performed using STATISTICA, version 12 (StatSoft Inc., Tulsa, OK, USA). Significant differences between the individual experimental variants were evaluated using ANOVA analysis, multiple comparisons, Duncan's multiple range test (DMRT) at p < 0.05.

#### SDS-PAGE and In-Gel Digest

Pellets of boiling-soluble proteins-enriched samples for dehydrin analysis extracted from H. vulgare and H. marinum crowns were dissolved in 50 µl 8 M urea (Sigma–Aldrich, Prague, Czech Republic). Three biological replicates (n = 3) were used. The samples were 10 min sonicated (Sonorex, Bandelin, Berlin, Germany) and 1 h incubated at room temperature in thermo shaker (Eppendorf, Rí ˇ cany, Czech Republic) at 1400 rpm. ˇ The sample concentration was determined by BCA Protein Assay (Fisher Scientific, Pardubice, Czech Republic). For an electroforetic separation, the samples were incubated 10 min at 30◦C in the thermo shaker and 5 µg of total proteins were loaded on 12% separating gel. The electrophoretic separation (Mini-PROTEAN Tetra Cell, Bio-Rad, Prague, Czech Republic) was carried out for 15 min at 160 V. The gel was stained by Imperial Protein Stain (Fisher Scientific, Pardubice, Czech Republic) according to standard protocol. An in-gel digest was performed according to standard protocol. The protein digestion in bicarbonate buffer by trypsin (Sequencing grade modified trypsin, East Port, Prague, Czech Republic) was carried out for 3 h at 37◦C in the thermo shaker. The peptides were extracted by two-stage extraction with mixtures of 35% acetonitrile (AppliChem, Darmstadt, Germany) and 0.1% TFA (Sigma–Aldrich, Prague, Czech Republic) and 70% acetonitrile and 0.1% TFA. The peptides were lyophilized and dissolved in 0.1% TFA prior to a purification by ZipTip C18 (Merck, Prague, Czech Republic). After the purification, the peptides were dried at room temperature.

#### NanoLC-ESI-Q-TOF MS

Mass spectrometric analysis was performed using liquid chromatograph UHPL Dionex Ultimate3000 RSLCnano (Dionex, Dreieich, Germany) with mass spectrometer ESI-Q-TOF Maxis Impact (Bruker Daltonics, Bremen, Germany). The lyophilized samples were dissolved in 10 µl of loading buffer (mixture of water:acetonitrile:formic acid in the ratio of 97:3:0,1). Three microliter of the samples were applied to the trap colon Acclaim PepMap 100 C18 (100 µm × 2 cm, reverse phase particle size 5 µm; Dionex, Dreieich, Germany) with a flow rate of 5 µl/min for 5 min. Then, the samples were separated by reverse phase chromatography carried out with a flow rate of 0.3 µl/min through the commercially produced column Acclaim PepMap RSLC C18 (75 µm × 150 mm, reverese phase particle size 2 µm; Dionex, Dreieich, Germany). The separation took place with the following gradient: 0 min 3% B, 5 min 3%B, 85 min 50% B, 86 min 90% B, 95 min 90% B, 96 min 3% B, 110 min 3% B; composition of mobile phase A = 0.1% formic acid in water and mobile phase B = 0.1% formic acid in acetonitrile. The peptides were eluted directly into the ESI source (Captive spray; Bruker Daltonics, Bremen, Germany). The measurement took place in DDA mode with the selection of precursor in the range of 400–2000 Da. From each MS spectra, up to ten precursors could be fragmented.

#### Data Analysis

The peaklists were extracted by Data Analysis 4.1 (Bruker Daltonics, Bremen, Germeny). The proteins were identified by ProteinScape 3 (Bruker Daltonics, Bremen, Germany) with software Mascot 2.4.01 (Matrix Science, London, UK) and the database composed of protein sequences H. vulgare contained in the database Uniprot (accessed 30. 01. 2014; 52,296 sequences; 17,742,792 residues belonging to H. vulgare) with following parameters: carbamidomethyl (C) as fixed modification, oxidation (M) as variable modification, accuracy 10 ppm in MS mode, MS/MS peptide mass assignment accuracy 0.05 Da and FDR 1%. The proteins which were identified as predicted or uncharacterized were compared to NCBI plant database using BLAST software. Protein domains were determined using PFAM database<sup>1</sup> . Protein functional classification was performed using Gene Ontology database<sup>2</sup> using the criterion of biological process. The proteins from different samples were compared by Venn diagrams. The proteins which were identified in control and salt-treated samples were quantified by TOP3 methods. This method is based on comparing the average of three most intense peptides of unique protein. This average is directly proportinal to the protein amount in the sample (Ahrné et al., 2013). The ratio of average values and variance of the ratio were calculated using the Taylor series. From the variance was calculated coefficient of variation, which should not exceed 20% (which is a permissible deviation for quantification of proteins based on marking of peptides or proteins). The relevance of quantification was analyzed by Student's two selection unpaired t-test at a level of significance of 0.05 and 0.01 using STATISTICA, version 12.

Cluster analysis on selected 108 proteins present in at least two of the experimental variants was performed using Permut Matrix software, version 1.9.3 (Caraux and Pinloche, 2005). For data analyses, Euclidean distances and Ward's minimum criteria were used.

<sup>1</sup>http://pfam.xfam.org

<sup>2</sup>http://www.geneontology.org/

#### RESULTS

### Physiological Parameters (WSD, PWC, ψs, Fv/Fm, Proline) and Dehydrin Relative Accumulation

All salt treatments applied on plants (100; 300; 0.2–300; 100– 300) led to an increase in leaf WSD (%), a decrease in osmotic potential of leaf sap (ψs), and an increase in proline levels in both barley species (**Figures 1A–C**). A comparison of both barley species indicated an enhanced increase in WSD while a lower decrease in ψ<sup>s</sup> in H. marinum with respect to H. vulgare. Higher WSD levels in H. marinum than in H. vulgare correspond to lower PWC expressed as ml per 1 g of dry weight in H. marinum than in H. vulgare which was found not only in plants exposed to high salinity (300 mM NaCl), but also in control plants (Supplementary Figure S2). Only the high salinity treatments (300; 0.2–300; 100–300) led to decreased levels in chlorophyll fluorescence parameter maximum quantum yield of photosystem II efficiency Fv/F<sup>m</sup> in H. vulgare while no significant decrease in Fv/F<sup>m</sup> was observed in H. marinum (**Figure 1D**).

All salinity treatments led to an accumulation of dehydrin proteins in both barley genotypes (**Figure 2**). Both quantitative and qualitative differences were found between both barley species. The level of dehydrin protein relative accumulation in H. marinum was significantly lower than in Tadmor under the same treatment. Both barley species accumulated highmolecular and low-molecular dehydrin proteins in response to salinity; however, the number and position of detected dehydrin bands were different between the species. H. vulgare cv. Tadmor accumulated one high-molecular dehydrin band of 82 kD corresponding to DHN5 and four low-molecular dehydrin bands of 26, 21, 19, and 18 kD in accordance with our previous results (Kosová et al., 2015). In contrast, H. marinum reveals two high-molecular dehydrin bands of apparent MW 99 and 74 kD and five low-molecular dehydrin bands of apparent MW 23, 21, 20, 19, and 18 kD (**Figure 2A**). However, in H. marinum, the lowmolecular-weight dehydrins were detected only in the samples transferred directly from control to high salinity (0.2–300 mM NaCl) while they were absent in the samples gradually acclimated to 300 mM NaCl. In both species, a significantly higher dehydrin accumulation was found in crown tissues with respect to leaf tissues (**Figures 2A,B**).

#### Proteomic Analysis

Crown tissues of two barley species H. vulgare (HV-) and H. marinum (HM-), both control (0.2 mM NaCl; HVC, HMC) and salt-treated samples (300 mM NaCl; HVN, HMN) after the one-step transfer to high salinity (0.2–300 mM NaCl),

were used for proteomic analysis. Three biological replicates (n = 3) of each sample were used for the analysis. Proteins which remain soluble upon a short boiling step of 10 min. were separated by 1D SDS-PAGE gels which were divided into four fractions according to protein electrophoretic molecular weight for protein identification (Supplementary Figure S3). In total, 284 differential proteins were identified by LC-MS (ESI-Q-TOF) in the four experimental variants (**Figure 3**; Supplementary Tables S1 and S2). In H. vulgare control plants (0.2 mM NaCl; HVC), 107 proteins were identified, in H. vulgare salt-treated plants (300 mM NaCl; HVN), 105 proteins were identified, in H. marinum control plants (0.2 mM NaCl; HMC), 141 proteins were identified, and in H. marinum salt-treated plants (300 mM NaCl; HMN), 200 proteins were identified (**Figure 3A**). A comparison of the proteins identified in both control and salttreated plants has revealed that H. vulgare exhibited a higher number of proteins decreased under high salinity with respect to control than proteins increased in response to high salinity (4↑12↓); in contrast, H. marinum revealed a higher number of proteins increased in response to salinity with respect to the number of decreased proteins (13↑6↓; **Figure 3B**). Similarly, a relatively high number of proteins (25) revealed a decreased relative abundance in salt-treated H. vulgare than in salttreated H. marinum (**Figure 3C**). A list of identified proteins belonging to the individual groups in Venn diagrams is given in Supplementary Table S3.

Out of 284 proteins identified in all four variants, 136 proteins were identified in one variant only while 148 proteins were identified in at least two experimental variants. Cluster analysis was carried out for 108 proteins present in at least two experimental variants except for the proteins found in HVC ∩ HMC (21), HVN ∩ HMN (15), HVC ∩ HMN (4), and HVN ∩ HMC (0) only. Cluster analysis has distinguished 10 clusters regarding protein dynamics between the variants (**Figure 4**). Cluster 1 (14 proteins) includes proteins identified only in H. vulgare while cluster 2 (51 proteins) includes proteins identified only in H. marinum. Cluster 3 (five proteins) and cluster 4 (seven proteins) include proteins with an increased

relative abundance in H. marinum with respect to H. vulgare. Cluster 5 (eightproteins) includes proteins with relatively enhanced abundance in HVC and HMN with respect to HVN and HMC variants. Cluster 6 (eight proteins) includes proteins with relatively increasing pattern from HVC via HVN and HMC to HMN, i.e., proteins with the highest abundance in HMN variant while the lowest abundance in HVC variant. Cluster 7 (2 proteins) and cluster 9 (10 proteins) include proteins with relatively enhanced abundance in salt-treated variants HVN, HMN) with respect to control variants (HVC, HMC) while cluster 8 (two proteins) includes proteins with relatively enhanced abundance in salt-treated variants (HVN, HMN) than in control samples (HVC, HMC). Cluster 10 (one protein) reveals relatively higher abundance in HVC and HMN than in HVN and HMC. The list of identified proteins belonging to the individual clusters is given in Supplementary Table S4.

The 284 identified proteins were classified into 24 functional groups based on Gene Ontology classification of biological processes: amino acid metabolism (5 proteins), apoptotic process (7 proteins), ATP metabolism (6 proteins), carbohydrate metabolism (18 proteins), cell adhesion (8 proteins), chlorophyll metabolism (1 protein), cytoskeleton (6 proteins), defense response (12 proteins), gene expression and replication (23 proteins), lipid metabolism (2 proteins), nucleic acid assembly (20 proteins), nucleic acid metabolism (3 proteins), one-carbon metabolism (2 proteins), photosynthesis (3 proteins), protein biosynthesis (28 proteins), protein degradation (23 proteins), protein folding (32 proteins), redox metabolism (24 proteins), regulatory proteins (7 proteins), secondary metabolism (2 proteins), proteins involved in signaling (15 proteins), proteins of tricarboxylic acid cycle (5 proteins), proteins involved in cellular transport (15 proteins), and proteins of unknown function (17 proteins). Several proteins were identified as predicted proteins or uncharacterised proteins and their possible identity was determined using BLAST search against Uniprot database (downloaded January 30, 2014). An overview of all 284 proteins identified in the experiment ordered according to their biological functions is given in **Figure 5**; Supplementary Tables S1 and S5.

## DISCUSSION

#### Plant Physiological Response and Dehydrin Relative Accumulation

All salt treatments applied on plants (100; 300; 0.2–300; 100– 300) indicated a significant dehydration effect of salinity on both genotypes. A comparison of the two barley species has shown higher leaf tissue dehydration in H. marinum with respect to H. vulgare while a lower decrease in ψ<sup>s</sup> associated with an enhanced accumulation of both low-molecular osmolytes and osmotically active proteins (dehydrins) in H. vulgare which may compensate enhanced Na<sup>+</sup> accumulation in H. vulgare shoot vacuoles as shown in a previous study (Garthwaite et al., 2005). Higher WSD levels in H. marinum than in H. vulgare correspond to lower PWC in H. marinum than in H. vulgare which was found not only in plants exposed to high salinity (300 mM NaCl), but also in control plants (Supplementary Figure S1). These data are in accordance with the previously found results (Garthwaite et al., 2005) and indicate constitutive differences in shoot water relations between H. marinum and H. vulgare which may be associated with differential levels of Na<sup>+</sup> ions in shoot tissues.

Regarding dehydrins, both quantitative and qualitative differences in dehydrin accumulation under salinity treatments indicate differential roles of dehydrins in salt-treated H. marinum and H. vulgare, respectively. Higher dehydrin levels found in crowns than in leaves correspond to the crucial role of crown tissues for the whole plant survival. Similarly, an enhanced

dehydrin accumulation was found in cold-treated crown tissues of winter wheat (Houde et al., 1995). One-step transfers from control and low-salinity conditions to high-salinity conditions resulted in higher accumulation of dehydrin proteins with respect to acclimation to gradually increasing salt levels, especially in salt-tolerant H. marinum. This corresponds to the results of our previous study (Kosová et al., 2015). Crowns are also closer to root salt ion uptake than leaves so it can be expected that they contain higher Na<sup>+</sup> levels. Thus, increased dehydrin accumulation in crown cell cytoplasm may help to compensate increased Na<sup>+</sup> levels in crown vacuoles.

Only the high salt treatments (300; 0.2–300; 100–300) led to a significant decrease in Fv/F<sup>m</sup> levels indicating an adverse stress impact of salinity on the efficiency of photosynthetic processes (**Figure 1D**). Under optimum conditions, Fv/F<sup>m</sup> values range around 0.83 (Lichtenthaler and Rinderle, 1988); however, severe stress leads to a decrease in Fv/F<sup>m</sup> values in sensitive plants indicating a reduced energy trapping efficiency by damaged photosystem II reaction centre (RC PSII), partially due to a damaged OEC (Rizza et al., 2011; Athar et al., 2015). Decreased Fv/F<sup>m</sup> values in H. vulgare cv. Tadmor exposed to high salt stress (300 mM NaCl) were also found in our previous study (Kosová et al., 2015). Significantly decreased Fv/F<sup>m</sup> levels in leaf tissues of H. vulgare cv. Tadmor exposed to high salinity levels (300; 0.2–300 and 100–300 mM NaCl treatments) indicate impairments in primary photosynthetic processes of salt-affected plants. Similar results were found in other salt-sensitive plants such as rice (Kim et al., 2005). In contrast, Fv/F<sup>m</sup> values in leaf tissues of H. marinum remained unchanged under high salinity with respect to control plants indicating no significant adverse effect of salinity on the efficiency of primary photosynthetic processes. The observed differences in Fv/F<sup>m</sup> between H. vulgare and H. marinum exposed to high salinity treatments indicate differential response of photosynthetic apparatus to salinity in these two species and they are in accordance with differential levels of OEE proteins and several chloroplast-located chaperones found by proteomic analysis (see below).

The differences between the two barley genotypes observed in tissue dehydration (WSD, PWC, and ψs) and osmolyte accumulation (proline, dehydrins) are in accordance with the previously published data (Garthwaite et al., 2005; Islam et al., 2007). These studies have reported differential levels of salt ions and differential mechanisms of salt tolerance in H. marinum and H. vulgare, respectively. Whereas H. marinum reveals low Na<sup>+</sup> levels in the shoots under salinity due to efficient Na<sup>+</sup> exclusion at xylem level, H. vulgare seems to employ Na<sup>+</sup> vacuolar compartmentation which is compensated by enhanced accumulation of osmotically active compounds such as proline and dehydrin proteins in cell cytoplasm. Therefore, higher levels of proline and sum of dehydrin proteins as well as lower levels of osmotic potential found in H. vulgare cv. Tadmor with respect to H. marinum are in accordance with the observed differences in salinity response between the two barley species.

#### Proteomic Analysis

The results of physiological analyses including characteristics related to tissue dehydration (WSD, PWC, and ψs) and osmolyte accumulation (proline, dehydrins) as well as efficiency of photosynthetic apparatus (Fv/Fm) led us to selection of one-step direct transfer to high salinity (0–300 mM NaCl) for further proteomic study since this treatment induced a significant stress response in both H. vulgare and H. marinum. For example, only the one-step direct transfers to high salinity induced an accumulation of low-molecular weight dehydrins in H. marinum (**Figure 2A**). Crown tissues were used for protein isolation since crowns are crucial for whole plant survival in cereals. We decided to use the boiling step commonly applied for dehydrin isolation

for sample enrichment with several stress-responsive proteins and chaperones.

An analysis of 1D SDS-PAGE protein fractions has led to an identification of a total of 284 proteins in both barley genotypes under control and high salinity treatments (**Figure 3A**). A comparison of both barley species in Venn diagrams has revealed more proteins enhanced in salt-tolerant H. marinum with respect to H. vulgare under both control (15↑) and high salinity (32↑) conditions (**Figure 3C**). These results are in accordance with those reported by Pang et al. (2010) who found constitutively enhanced levels of several stress-responsive proteins in salt-tolerant T. halophila with respect to salt-sensitive A. thaliana.

A comparison of control variants versus high salinity variants based on protein functional categories (**Figure 5**) revealed that salt-treated samples with respect to control ones exhibited a higher percentage of identified proteins involved in defense response (HVN 6.7%, HMN 4.5%, HVC 0.93, and HMC 0.7% of total identified proteins in the given experimental variant) and in redox metabolism (HVN 7.6%, HMN 8.5%, HVC 5.6%, and HMC 5.7% of total identified proteins in the given experimental variant). In contrast, control samples with respect to salt-treated ones revealed higher percentage of identified proteins involved in carbohydrate metabolism (HVC 9.3%, HMC 9.2%, HVN 3.8%, and HMN 6% of total identified proteins in the given experimental variant) and protein biosynthesis (HVC 8.4%, HVN 6.7%, HMC 12%, and HMN 8.5% of total identified proteins in the given experimental variant). Salt-treated H. vulgare (HVN) revealed relatively higher percentage of proteins involved in apoptotic processes (5.7% of total identified proteins) with respect to other variants (1.5–3.7% of total identified proteins in the individual variants; **Figure 5**; Supplementary Table S5).

The presence of several boiling-unstable proteins in a boilingsoluble protein fraction can be explained by presence of several chaperones and other proteins with protective functions which can protect other proteins from a lack of their conformation and precipitation. The identification of several ubiquitious proteins such as enzymes of tricarboxylic acid cycle and several structural proteins such as transmembrane channels (e.g., components of mitochondrial outer and inner envelope complexes) only in some

experimental variants can be explained by differential relative abundance of these proteins with respect to other proteins in the whole protein sample which can be under detection threshold in some samples as well as by strict identification criteria (a minimum of two matched peptides) which could not be fulfilled in all samples.

### Possible Roles of Identified Proteins in H. marinum and H. vulgare Salinity Response with Respect to Their Functional Classification

#### Signaling and Signal Transduction

Proteins revealing significant quantitative or qualitative differences between experimental variabnts included PP2C, small calcium-binding proteins and 14-3-3 proteins (**Figure 5**; Supplementary Table S1).

Protein phosphatase 2C is known to act antagonistically to MAPKKK kinase cascade and SnRK involved in ABAmediated signaling and signal transduction from plasmalemma to nucleus. It is known that ABA receptor PYR1 activated by ABA inhibits PP2C (Park et al., 2009). A decrease in PP2C 10 (M0Z9V5) levels under salinity with respect to control plants may indicate alterations in regulation of ABA-mediated signaling pathways under salinity stress with respect to control conditions. Differential phosphorylation of PP2C under drought with respect to control conditions was found in wheat (Zhang et al., 2014).

Three small calcium-binding proteins, calcium-binding protein CML7 (F2E2L5), calmodulin (F2CQ91) and calreticulin (F2CWX0) and one protein with an IQ calmodulin-binding motif (M0WVJ8) were identified in the experiment. An increase in calmodulin levels was found in HVN with respect to HVC, while decreased levels of calmodulin and calreticulin were found in HVN with respect to HMN. Small calcium-binding proteins are known to play an important role in stress-induced signaling as second messengers transducing the original signal via interactions with several kinases and phosphatases. Increased levels of calmodulin and calreticulin were found under salinity in Arabidopsis roots (Jiang et al., 2007), in rice roots (Cheng et al., 2009), in grapevine shoots (Vincent et al., 2007), and others.

Three 14-3-3 proteins (14-3-3A F2CRF1; 14-3-3B M0XMV1; 14-3-3E A1X810) were identified in three experimental variants except HVN. 14-3-3 proteins contain a TPR motif and they are known to be involved in modulation of cell signaling, interactions with other proteins and plant-pathogen interactions. They can also regulate activity of vacuolar H+-ATPases thus affecting Na<sup>+</sup> vacuolar sequestration (Finnie et al., 1999). Some 14-3-3 proteins can be found even in extracellular space where they interact with proteins produced by plant pathogens (Denison et al., 2011). A decrease in 14-3-3 like protein B level was found in HMN with respect to HMC which is consistent with the results observed in salt-tolerant Puccinellia tenuiflora (Yu et al., 2011); in contrast, an increase in 14-3-3 protein was found in salt-treated Kandelia candel (Wang et al., 2014).

#### Gene Expression and Nucleic Acid Assembly

Nascent associated-polypeptide complex (NAC α) proteins are TFs involved in both ABA-dependent and ABA-independent pathways which bind to NACR cis-regulatory elements and regulate alone or in cooperation with zinc-finger homeodomain TFs (ZFHD TFs) expression of several dehydration-responsive genes such as RD22 in A. thaliana (Yamaguchi-Shinozaki and Shinozaki, 2006; Puranik et al., 2012). In our study, three NAC proteins (F2D1I7, M0VX35, and M0YAP1) were identified in both salt-treated barleys with HMN revealing significantly higher abundance of NAC proteins than HVN. Enhanced levels of NAC α were also reported in salt-treated rice roots (Yan et al., 2005) indicating the importance of NAC α in regulation of salt-responsive protein pathways. An identification of a bZIP (M0Z1P5) in HMN may indicate an enhanced expression of several ABA-inducible stress-responsive genes including several LEA proteins (Choi et al., 1999).

Three proteins (M0ZC67, M0ZC68, and M0XUE8) identified as nucleosome assembly protein 1-A as well as protein SET (M0VVZ7) are involved in interaction with nucleosomal histones thus affecting the dynamics of euchromatinheterochromatin transitions and DNA accessibility for transcription. Nucleosome assembly proteins also participate on transcriptome reprogramming in plants exposed to altered environmental conditions as reported for cold-treated winter barley (Janská et al., 2011). Decreased levels of SET protein were found in H. vulgare with respect to H. marinum under both control and salinity conditions. DNA repair protein RAD23 (M0YFP6, M0Z6M4, and F2DUJ6) is involved in nucleotide excision and repair (GO:0006289). However, RAD23 is also known to inhibit a polyubiquitin chain formation and 26S proteasome-dependent degradation of proteolytic substrates in yeast thus linking DNA repair to ubiquitin/proteasome pathway (Schauber et al., 1998).

#### Protein Metabolism

#### **Protein biosynthesis**

Several proteins involved in protein biosynthesis were identified including both cytosolic and organellar ribosomal proteins and translation initiation and elongation factors. Some of them were detected only in control variants (HVC, HMC; 40S ribosomal protein SA M0WYK5; 60S acidic ribosomal protein P0 F2DBD4; 60S ribosomal protein L22-2 F2EAX5; organellar elongation factor Tu M0ZD98) while others were detected only in H. marinum (HMC, HMN) variants (elongation factor 1 β M0YVB7; elongation factor 1 γ 2 M0WF40; eukaryotic translation initiation factor 5A2 M0YU33; 40S ribosomal protein S12 M0VTY9; 40S ribosomal protein S19 F2E598; 30S ribosomal protein 2 M0XX22). Several ribosomal proteins reveal specific functions in translation regulation; for example, 60S acidic ribosomal proteins (60S acidic ribosomal protein P2B M0V315; 60S acidic ribosomal protein P0 F2DBD4; 60S acidic ribosomal protein P2 M0YSB6; 60S acidic ribosomal protein P3 M0XHQ5) are known to be regulated by phosphorylation and they are involved in interaction with elongation factor EF1. A decrease in 60S acidic ribosomal protein P2 (M0YSB6) was found in HVN with respect to HMN. The 30S ribosomal

protein S1 is important for translation initiation due to mRNA recognition and binding to mRNA upstream of Shine–Dalgarno sequence (Boni et al., 1991). An enhanced relative abundance of 30S ribosomal protein S1 (M0YSG8) found in H. marinum with respect to H. vulgare under control conditions indicates possible alterations in organellar proteosynthesis between the two barley species. Identification of several ribosomal proteins and translation initiation and elongation factors indicates active protein biosynthetic processes which are crucial for both control and stress-treated plants to underlie an active stress acclimation. 60S ribosomal protein L12 and 50S ribosomal protein L12- 2, respectively, are known to bind to rRNA during ribosome assembly and are involved in maintenance of ribosomal structure and function. In our experiment, these proteins were detected in salt-treated plants of both species (HVN, HMN) and in H. marinum, respectively. Alterations in the levels of 60S acidic ribosomal protein P1, 30S ribosomal protein S1 and ribosomal protein L12 homolog were found in salt-treated barley leaves (Fatehi et al., 2012). An enhanced abundance of ribosomal protein L12 was reported in salt-treated rice leaves (Kim et al., 2005).

Some translation initiation and elongation factors reveal multiple functions besides regulation of protein synthesis. Eukaryotic translation initiation factor 5A (eIF-5A) is known to be activated by PTM of hypusination and it is involved in regulation of cell cycle. It is also known that eIF-5A isoforms differing in hypusination level play differential roles in cell cycle regulation and cell fate determination with eIF-5A1 isoform involved in induction of cell apoptosis while eIF-5A2 isoform involved in induction of cell division (Thompson et al., 2004). In our experiment, eIF-5A1 isoform was found in H. vulgare while eIF-5A2 isoform was found in H. marinum which is consistent with the presence of several proteins involved in apoptosis induction in salt-treated H. vulgare while their absence in H. marinum.

#### **Protein degradation**

Proteins involved in protein ubiquitination resulting in protein targeting for proteasome degradation (ubiquitin-like protein M0V112; E3 UMF1-protein ligase 1 homolog M0W4G1; deubiquitination-protection protein dph1 F2CQR1; small ubiquitin-related modifier 1 M0UGE2) are increased in salttreated variants (HVN, HMN) with respect to control ones (HVC, HMC) indicating enhanced protein damage under salinity. In addition, 26S protease regulatory subunit 6A-like protein A (M0ZB49) reveals relatively enhanced level in HVN with respect to HMN indicating relatively enhanced proteasome-dependent protein degradation in HVN with respect to HMN. 26S protease regulatory subunit 6A-like protein A reveals ATPase activity and contains AAA+-ATPase domain. AAA+-ATPases are present in all living organisms and reveal a wide array of functions including facilitation of protein folding and defolding, assembly and disassembly of protein complexes, and proteolysis. Enhanced levels of proteins involved in proteasome-dependent protein degradation were reported in salt-treated Nitraria sphaerocarpa (Chen et al., 2012).

Four proteins with cysteine proteinase activity (papain-like cysteine proteinase B4ESE6, M0ZD30; thiol protease aleurain M0VNM9; triticain β 2 M0YY63) and three cystatins (M0V101, F2CUF5, and F2DNM2), cysteine proteinase inhibitors, were identified in the samples, predominantly in HMN. In HMN, relatively enhanced levels of papain-like cysteine protease and cystatin Hv-CPI14 with respect to control indicate an activation of stress-responsive protein protective mechanisms under salinity. Cystatins are known to respond to pathogen attack via the inhibition of pathogen cysteine proteinases (Benchabane et al., 2010). They could be induced also by various abiotic stresses including cold (Kosová et al., 2013b) and salinity, as seen in the present study.

#### Energy Metabolism

#### **Photosynthesis**

Three OEE proteins involved in formation of OEC, a crucial component of photosystem II (PSII) where photolysis of water occurs, were identified (OEE1 F2CRK1; OEE2 M0WTH3; OEE3- 1 M0YTD9). Two of them (OEE1, OEE2) were present in all four experimental variants and revealed a significant increase in salt-treated H. marinum with respect to control plants as well as enhanced levels in H. marinum with respect to H. vulgare, especially in salt-treated variants (HVN, HMN). Relatively enhanced abundance of OEE1 and OEE2 proteins in H. marinum with respect to H. vulgare, especially under high salt treatment, indicates an active stress acclimation of H. marinum when compared to H. vulgare. Enhanced levels of some PSII components such as D2 protein, a crucial component of RC PSII, and OEE1 and OEE2 proteins, under salinity were found in tolerant halophytic species such as Suaeda aegyptiaca (Askari et al., 2006) and Aeluropus lagopoides (Sobhanian et al., 2010), respectively; in contrast, a decrease in PSII protein components was found in salt-sensitive species such as durum wheat (Caruso et al., 2008).

#### **ATP metabolism**

Six proteins identified include proteins involved in ATP biosynthesis such as mitochondrial ATP synthase precursor as well as proteins involved in cleavage or transfer of phosphate groups resulting in interconversion of nucleotide phosphates (adenosine kinase 2 M0XLC4; adenylate kinase A F2DTH4; nucleoside diphosphate kinase F2CXV3; soluble inorganic pyrophosphatase 1 F2EJL5, F2CVI3). ATP metabolism plays a crucial role in plant stress acclimation since an active stress acclimation process requires a large portion of energy. Soluble inorganic pyrophosphatase is involved in cleavage of macroergic phosphate bonds in pyrophosphate to yield single phosphates. Tonoplast-bound inorganic pyrophosphatase is involved in Na+/H<sup>+</sup> antiport across tonoplast and was found enhanced upon salt stress in both glycophytes and halophytes (Jiang et al., 2007; Wang et al., 2008; Du et al., 2010; Pang et al., 2010; Xu et al., 2010). Two proteins (F2EJL5, F2CVI3) identified as soluble inorganic pyrophosphatase 1 reveal a decreased abundance in HVN with respect to HMN indicating differential energy requirements. The presence of mitochondrial ATP synthase precursor (M0XMC3) in salt-treated variants (HVN, HMN) corresponds with enhanced need for energy during an active stress acclimation as well as with the results of other proteomic studies on halophytes A. lagopoides (Sobhanian et al., 2010) and Tangut nitraria (Cheng et al., 2015).

#### **Tricarboxylic acid (TCA) cycle**

fpls-07-01154 August 1, 2016 Time: 13:26 # 12

Five proteins belonging to TCA cycle (aconitate hydratase M0Y4D4; isocitrate dehydrogenase M0YIA7; succinyl-CoA ligase subunit β F2D1J8; malate dehydrogenase M0Z0D3, F2CQR0) were detected only in control plants (HVC, HMC); all of them were identified in H. marinum while two of them in H. vulgare (Supplementary Tables S1 and S5). Detection of TCA cycle enzymes only in control plants may indicate an enhanced abundance of these enzymes in non-stressed plant materials with respect to salt-treated ones. Moreover, higher levels of several protein chaperones in H. marinum with respect to H. vulgare may contributed to the preservation of relatively larger amounts of TCA cycle enzymes in HMC with respect to HVC during sample preparation. Similarly, a relatively enhanced level of malate dehydrogenase was found in high frost-tolerant winter wheat Mironovskaya 808 with respect to less frost-tolerant winter wheat Bezostaya 1 (Vítámvás et al., 2012).

#### Carbohydrate Metabolism

#### **Catabolic enzymes**

In the samples, six out of twelve glycolytic enzymes were identified including fructose bisphosphate aldolase (ALDO; F2CXT7), triosephosphate isomerase (TPI; F2EHF8, M0WDD4), glyceraldehyde-3-phosphate dehydrogenase 2 (GAPDH; F2D6I8), phosphoglycerate kinase (PGK; M0Y9H9), PGM (M0WLI6), and ENO (M0WLI6). A significant decrease in relative abundance of four glycolytic proteins detected (ALDO, ENO, PGM, and TPI) was found in salt-treated samples with respect to control ones, especially in H. vulgare, indicating a downregulation of crucial metabolic pathways under severe stress. In halophytes such as Bruguiera gymnorhiza and Halogeton glomeratus, enhanced levels of several glycolytic enzymes such as ALDO and GAPDH were found (Tada and Kashimura, 2009; Wang et al., 2015). An induction of several glycolytic enzymes (GAPDH, TPI) was found in salt-treated durum wheat (Caruso et al., 2008); however, the plants were exposed to 100 mM NaCl indicating a mild stress treatment only. A decrease in glycolytic enzymes levels in HVN indicates a severe stress treatment; as found in our previous study on spring barley Amulet exposed to two intensities of drought, a mild stress usually leads to enhanced levels of glycolytic enzymes indicating an active plant stress acclimation while a more severe stress can result in a decrease in glycolytic enzymes indicating a disruption of cellular metabolism (Vítámvás et al., 2015). It can be concluded that high salt treatment induced an active acclimation in H. marinum while damage in H. vulgare.

#### **Anabolic enzymes**

Phosphoglycerate kinase involved in gluconeogenesis and transaldolase involved in pentose phosphate pathway were found in control variants of both barley species which may indicate relatively enhanced need for hexoses and pentoses in control plants with respect to salt-treated ones. Sucrose synthase catalyzes a reversible cleavage of sucrose to yield fructose and UDP-glucose which can be then incorporated into polysaccharide chains. The identification of sucrose synthase (M0UKI5) in control variants (HVC, HMC) only may indicate enhanced glycosidic bond formation in control variants which corresponds to an enhanced need for energy, i.e., rather a cleavage than a formation of glycosidic bonds, in salt-treated variants. Several enzymes involved in modification of cell wall polysaccharides such as UDP-glucuronate decarboxylase (F2D5Z0), UTP-glucose-1 phosphate uridylyltransferase (M0USZ9) were found only in control plants while endo-1,3:1,4-β-D-glucanase (F2DSV5) was found only in HMN variant. Protein identified as invertase inhibitor (M0W0Q7) in HMN variant is known to bind to invertases and pectinesterases inhibiting their activity thus affecting cell wall elongation and fruit ripening. These results indicate profound alterations in composition of cell wall polysaccharides in salt-treated plants with respect to controls, especially in HMN, which corresponds to the results of Mostek et al. (2015).

#### Other Metabolic Pathways **Amino Acid Metabolism**

Five proteins involved in amino acid metabolism were identified, four of them were identified only in H. marinum (both control and salt-treated samples). Tryptophan aminotransferase 1 (F2E3H9), aspartate aminotransferase (M0USC9) are involved in transfer of amino groups resulting in interconversion between aminoacids and oxoacids. Cysteine synthase (M0VBS3) is involved in biosynthesis of cysteine while serine hydroxymethyltransferase (F2DET3) is involved in interconversion between glycine and serine. Amino acids seem to play a protective role under salinity due to their osmoprotectant and ROS scavenging effects (Sobhanian et al., 2010; Flowers and Colmer, 2015).

#### **Chlorophyll metabolism**

One protein was identified as magnesium chelatase which is an enzyme catalyzing ATP-dependent incorporation of Mg2<sup>+</sup> into protoporphyrin IX during chlorophyll biosynthesis. This protein was found only in HVC. Magnesium chelatase was found decreased in salt-treated halophyte Tangut nitraria (Cheng et al., 2015).

#### **Lipid metabolism**

Two enzymes, ENR (M0YUL2) and GDSL esterase/lipase (M0VZB7), were identified only in HMC. ENR is a component of fatty acid synthetase type II in plastids and it is involved in a reduction step during fatty acid chain biosynthesis in thylakoid membranes which is crucial for thylakoid membrane integrity upon salt stress. The presence of ENR in HMC may indicate constitutive protection of thylakoid membranes in H. marinum. Changes in ENR levels were also found in salt-treated rice panicles (Dooki et al., 2006).

#### **One-carbon metabolism**

Two enzymes associated with one carbon metabolism were identified. S-adenosylhomocysteine hydrolase (M0ZB73) cleaves S-adenosylhomocysteine, a product after the methyl transfer from SAM, into adenosine and homocysteine. Cyanate hydratase,

also known as cyanase, is involved in detoxification of cyanate since it catalyzes a conversion of cyanate to carbamate which is then spontaneously decomposed into carbon dioxide and ammonia. Cyanate in plants arises as a by-product during ethylene biosynthesis. Cyanate hydratase (M0UZA5) was identified only in HMN. Similarly, cyanate hydratase was identified in salt-treated halophyte S. aegyptiaca by Askari et al. (2006) and the authors speculated about its role not only in detoxifying cyanate, but also in supplying salt-treated plants with alternative nitrogen and carbon sources. Thus, the presence of cyanate hydratase in HMN may underlie its superior salt tolerance under high salt stress.

#### Stress and Defense-Related Proteins

#### **Protein folding**

Several proteins involved in protein folding revealed enhanced levels in salt-treated plants with respect to control ones as well as in salt-treated H. marinum with respect to H. vulgare (M0X093 luminal binding protein 4; M0WRW4, F2DA67, M0WCQ8 PPI; M0UZ78, F2DU3 20 kDa chaperonin; M0X640 RubisCO large subunit-binding protein subunit β; M0Y631, M0V4Z9 chaperonin CPN60-2; M0UEM7 HSP STI; M0WVW3, M0VTF2 stromal 70 kDa heat shock-related protein), out of which, a substantial amount represent chloroplast-located proteins involved in protection of photosynthetic apparatus and RubisCO carbon assimilation activity (luminal binding protein 4; RubisCO large subunit-binding protein subunit α and β, stromal 70 kDa heat shock-related protein, chaperonin CPN 60-2) indicating enhanced protection of photosynthetic apparatus in halophytic H. marinum with respect to glycophytic H. vulgare. An increase in chloroplast RubisCO large subunitbinding protein subunit β was found in salt-sensitive durum wheat (Caruso et al., 2008) and barley (Rasoulnia et al., 2011; Fatehi et al., 2012) as well as in salt-tolerant P. tenuiflora (Yu et al., 2011). Several low-molecular chaperones (20 kDa chaperonin, 23.5 kDa HSP) were identified in our experiment. An increased relative abundance of several small HSPs including cytoplasmic as well as chloroplast and mitochondrial proteins was found in salt-treated tomato hypocotyls (Chen et al., 2009) and Aster tripolium leaves (Geissler et al., 2010). Increased relative abundance of STI1 protein, a stress-responsive phosphoprotein with two heat shock chaperonin-binding motifs and three TPR, was found in salt-treated rice panicles (Dooki et al., 2006) which points toward a regulatory network affected by salt stress since TPR-containing proteins were reported as being involved in myriads of processes including HSP90 signaling, gibberellin signaling and protein mitochondrial transport. GrpE protein homolog (M0YSD3, M0ZDK9) reveals an increase in salt-treated plants with respect to control plants in both H. marinum and H. vulgare. GrpE proteins are chaperones involved in ATP-dependent protein folding, interactions with client protein transit peptides and the client protein translocations to subcellular organelles such as mitochondrial matrix. Endoplasmin (M0VKI5, M0Z0U7) belongs to Hsp90 protein family and it is involved in ATPdependent protein folding and translocation of client proteins into chloroplast stroma.

Protein disulfide isomerase (F2D284, M0UZB9) catalyzes a reversible oxidation of sulfhydryl groups in cysteine residues to form disulfide bridges thus affecting cellular redox homeostasis and protein conformation. PPI (M0WRW4, F2DA67 and M0WCQ8) catalyzes a reversible isomerisation between cis- and trans-conformation of peptide bonds next to proline residue. Enhanced PDI was found in drought-treated barley (Ashoub et al., 2013). A relatively decreased level of PDI (F2D284) was found in HVN with respect to HMN which corresponds with the findings of Mostek et al. (2015) who reported relatively higher PDI level in salt-tolerant barley cultivar with respect to the salt-sensitive one.

#### **Defense response**

Six LEA proteins (late embryogenesis abundant protein M0ZDL8; ABA inducible protein from LEA 4 family M0Z6A4; late embryogenesis abundant protein F2CRD9; late embryogenesis abundant protein D34 F2DNE8; late embryogenesis abundant protein 1 F2ECH4; dehydrin 8 M0UW32) and three PR proteins (pathogenesis-related protein M0Z8T3; germin F M0VST4; salt tolerant protein M0YJN2) were found only in salt-treated variants (HVN, HMN) while they were absent in control variants (HVC, HMC). An enhanced accumulation of dehydrin proteins in salt-treated H. vulgare was found in our previous study (Kosová et al., 2015) as well as in durum wheat (Brini et al., 2007). It is known that H. vulgare genome encodes 13 dehydrin genes out of which Dhn1 to Dhn11 are expressed in vegetative tissues (Choi et al., 1999) while Dhn12 reveals an embryo-specific expression (Choi and Close, 2000) and Dhn13 reveals an anther-specific expression (Rodriguez et al., 2005). Barley dehydrins belong to four structural types – K<sup>n</sup> type (DHN5), SK<sup>n</sup> type (DHN8), KnS type (DHN13) and YxSK<sup>n</sup> type (DHN1,2,3,4,6,7,9,10,11,12; Tommasini et al., 2008). Two major groups of dehydrin proteins, high-molecular dehydrins and low-molecular dehydrins, were detected in both salt-treated barley species on the immunoblots (**Figure 2A**). Of high-MW dehydrins, only one band corresponding to DHN5 protein was detected in H. vulgare while two high-molecular bands were detected in H. marinum. Of low-molecular dehydrins, four distinct bands were detected in H. vulgare while five were detected in H. marinum. Proteomic analysis revealed the presence of acidic SK<sup>3</sup> dehydrin 8 in both barley species. It is known that both Dhn5 and Dhn8 genes are significantly induced by cold; however, they can also be induced by dehydration stresses and ABA (Choi et al., 1999; Tommasini et al., 2008; Kosová et al., 2013c, 2014b). The majority of H. vulgare dehydrins are of YxSK<sup>n</sup> type, relatively low-MW (14–30 kDa) and are induced by dehydration and ABA (Choi et al., 1999; Tommasini et al., 2008). A similar pattern of dehydrin expression seems to be found in halophytic H. marinum. Immunoblot results obtained on dehydrin proteins represent LC-MS/MS data validation since dehydrins as LEA proteins were detected only in salt-treated samples (HVN, HMN).

Germin F (M0VST4) was found in both salt-treated barley species (HVN, HMN) while it was absent in control variants (HVC, HMC). An increased abundance of germin-like proteins was reported in salt-treated leaves of barley (Fatehi et al., 2012),

in barley response to powdery mildew (Zhou et al., 1998), in barley root exposed to salinity (Hurkman et al., 1994), aluminum (Tamás et al., 2004), as well as cadmium (Valentovicová et al., 2009), in cold-treated winter wheat crowns (Kosová et al., 2013b) and in salt-treated Arabidopsis roots (Jiang et al., 2007).

Universal stress protein (F2DHC4) is known to mitigate various abiotic stress impacts (Udawat et al., 2016); however, a precise role of plant USP proteins in stress protection remains largely unknown. Constitutive presence of USP protein in H. marinum while its absence in H. vulgare may contribute to superior salt tolerance of H. marinum.

#### **Redox metabolism**

Twenty-three proteins identified include enzymes involved in cleavage of superoxide anion radical such as cytosolic Cu/Zn superoxide dismutase (Cu/Zn-SOD M0VCA1, K4KCI3, M0YTZ0, M0VYA3), enzymes involved in cleavage of hydrogen peroxide (peroxidases POX M0X820, M0X558, F2CTB8, M0XHU1, M0YYV7; and peroxiredoxins Prx M0VDA7, M0X4Z3), enzymes involved in cleavage of lipid peroxides (lipoxygenases M0VUI3, M0Y2M0) as well as enzymes of ascorbate-glutathione cycle APX (M0UT11, M0X2A1, and M0YH83; monodehydroascorbate reductase MDAR F2D5M0). The majority of the proteins (17 out of 23) involved in redox metabolism were identified in salt-treated H. marinum (HMN) indicating an active stress acclimation. Fine regulation of ROS levels in plant cells represents an efficient tool in coordination of plant stress response (Suzuki et al., 2012).

Cystathione β-synthase domain-containing protein CBSX3 (M0Z0Z8) was found in all variants except for HVC. CBS domain proteins are involved in regulation of thioredoxin activity thus affecting cellular redox homeostasis (Bertoni, 2011). Previous study has reported improved salinity, redox and heavy metal tolerance in transgenic tobacco overexpressing rice CBS domain protein (Singh et al., 2012).

Lactoylglutathione lyase (F2CQP8) reveals glyoxalase activity using glutathione as a cofactor. It is known to be involved in glutathione-dependent detoxification of methylglyoxal which is a toxic byproduct of carbohydrate and amino acid metabolism. Lactoylglutathione lyase was detected only in HMC which corresponds to a constitutively enhanced level of this enzyme in roots of salt-tolerant barley cultivar Morex with respect to saltsensitive barley cultivar Steptoe (Witzel et al., 2009) and indicates constitutively enhanced salt tolerance of H. marinum with respect to H. vulgare.

#### Regulatory Proteins

Cdc48 is cyclin involved in regulation of cell cycle and cell division. Cdc48 (M0Z3K0) was detected in HVC only which is consistent with a decline in Cdc48 observed in salt-treated halophyte N. sphaerocarpa (Chen et al., 2012).

Several glycine-rich RNA binding proteins (GRP; M0XEV2, F2DKA4, and M0WJV7) and low-temperature-responsive RNA binding protein (M0V3Q4) were identified, mostly in HMN. Small GRPs are known to be involved in the regulation of transcription and RNA processing. Several GRPs were reported to function as repressors of the major flowering repressor FLC and the members of an autonomous flowering pathway in A. thaliana (Quesada et al., 2005). Increased GRPs were found to be associated with vernalization fulfillment and a transition to flowering in winter wheat (Rinalducci et al., 2011) as well as in A. thaliana (Streitner et al., 2008). Increased abundances of several GRPs were reported in stress-treated plants such as in transgenic A. thaliana, poplar, and Nicotiana tabacum under salt and flooding stresses, respectively (Kwak et al., 2005; Durand et al., 2010; Wang et al., 2012).

Ricin B lectin (F2DIC8, M0VWY2) belongs to lectins which are glycoproteins involved in saccharide signaling as well as in regulation of cell development. Ricin B lectin was found to be increased in cold-treated winter barley and winter wheat crowns, respectively (Hlavácková et al., 2013 ˇ ; Kosová et al., 2013b).

#### Apoptosis-Related Proteins

Proteins involved in regulation of PCD include seven proteins identified, three of them found specifically in HVN variant. The proteins specifically found in HVN include apoptotic chromatin condensation inducer in nucleus (F2E093) and FAS associated factor 2B (M0YTK2) which is involved in the interaction of FAS antigen with FAS ligand leading to initiation of apoptotic processes. The FAS associated factor binds to the FAS antigen thus enhancing its interaction with FAS ligand and inducing signaling processes leading to apoptosis. The presence of these proteins specifically in HVN indicates induction of processes leading to PCD in H. vulgare crown tissues by high salt stress. Plasminogen activator inhibitor 1 RNA-binding protein (M0WMM0, M0YZ72, and M0YVX6) is, according to GO annotation, involved in apoptotic processes (GO: 0042981) and regulation of mRNA stability (GO: 0043488). Guaninenucleotide binding protein (G protein) subunit β (F2DSU6) is a phosphoprotein with WD-repeat motif which is known to be induced by exogenous ABA and is involved in a variety of cellular processes including signal transduction, transcription regulation, cell cycle regulation and apoptosis. Guanine-nucleotide binding protein subunit β was reported to be increased in salinityand drought-treated rice (Dooki et al., 2006; Ke et al., 2009), respectively.

Translationally controlled tumor protein homolog (TCTP; F2DWT1) was identified in all experimental variants revealing a significantly increased level in HMN with respect to HVN. It is known that TCTP are calcium-binding proteins which are involved in an inhibition of p53 tumor suppressor-dependent apoptosis by binding to p53 thus down-regulating p53 activity. The obtained results thus indicate better suppression of apoptotic processes in HMN when compared to HVN. An increased abundance of TCTP was reported in barley plants exposed to salinity (Mostek et al., 2015) and drought (Ghabooli et al., 2013).

#### Structural Proteins

#### **Cell adhesion**

Out of eight proteins categorized to cell adhesion, four proteins (M0X5B5, M0WI23, M0XHE6, and M0YDA0) were identified as ankyrins (ankyrin-1 and ankyrin repeat domaincontaining protein 2) which are known to be involved in anchoring integral membrane proteins in plasma membrane

and assisting their attachment to spectrin-associated membrane skeleton. Other three proteins were identified as fasciclinlike proteins (fasciclin-like protein FLA5 M0Z9S4; fasciclin-like protein FLA10 M0W0B1; fasciclin-like protein FLA15 M0Z3D1). Fasciclins are secreted or membrane-anchored glycoproteins involved in cell wall architecture and biomechanics affecting plant stem strength (MacMillan et al., 2010). An increase in fasciclin-like protein FLA15 was found in HMN with respect to HMC while a relatively decreased level of ankyrin-repeat domain containing protein 2 was found in HVC with respect to HMC indicating both genotype-specific and salinity-induced alterations in cell biomechanical properties.

#### **Cellular transport**

Increased levels of Na<sup>+</sup> lead to enhanced Na<sup>+</sup> sequestration into vacuole. Vacuolar proton ATPase subunit E involved in Na+/H<sup>+</sup> antiport was found in HMN. V-ATPases were reported to be increased upon salinity (Wang et al., 2008; Pang et al., 2010) and their activity is supposed to be regulated by SOS2 component of SOS1/SOS2/SOS3 signaling pathway (Batelli et al., 2007) and by glycolytic enzymes aldolase and ENO (Barkla et al., 2009).

Processes associated with mRNA and protein transport across nuclear pores are regulated by small GTPases from Ran family (Ran GTPases) and related proteins involved in Ran GTPase activation. Two Ran GTPase activating proteins (Ran GTPase-activating protein 1-like M0X1Z2; Ran-specific GTPase-activating protein 2 M0YLZ9) and one Ran-binding 1-c like protein (M0ZCE0) were identified in salt-treated variants (HVN, HMN) which is consistent with the finding that GTP-binding nuclear protein Ran1 was found increased in salt-treated shoots of A. lagopoides (Sobhanian et al., 2010).

Non-specific lipid transfer proteins are small proteins with a hydrophobic lipid binding site which are involved in lipid transfer from donor membranes such as endoplasmic reticulum to acceptor membranes such as chloroplasts, mitochondria,

peroxisomes, and glyoxysomes. LTPs are also found in cell walls where they are involved in cuticle formation. LTPs were reported to be induced by several pathogens and environmental stresses including drought and salinity in wheat (Jang et al., 2004). Two nsLTPs (Q5UNP2, F2CY84), one lipid binding protein precursor (M0Z7W8) and one protein with sterolbinding domain (M0V4I6) were identified in salt-treated variants (HVN, HMN).

#### **Cytoskeleton**

Components of both microfilamental (actin cytoskeletonregulatory complex protein PAN1 M0YG53; actin depolymerizing factor 4 F2DY31) and microtubular (tubulin alpha M0YMF1; β tubulin 6 A5CFY9; tubulin folding cofactor B F2CRA1) cytoskeleton were found in the samples; however, except for actin depolymerizing factor 4 found in salt-treated plants of both barley species (HVN, HMN), other cytoskeletonassociated proteins were reliably detected in one variant only. A decline in β tubulin and γ tubulin-interacting protein was found in salt-treated N. sphaerocarpa (Chen et al., 2012).

### CONCLUSION

The present study has revealed a differential response to high salinity of 300 mM NaCl between cultivated barley H. vulgare cv. Tadmor and a halophytic wild barley H. marinum (**Figure 6**). H. marinum revealed constitutively enhanced tissue dehydration with respect to H. vulgare. In contrast, H. vulgare revealed lower osmotic potential and higher levels of osmolytes such as proline and dehydrins with respect to H. marinum which is consistent with the results of previous studies and corresponds with enhanced Na<sup>+</sup> vacuolar accumulation in salt-treated H. vulgare shoot tissues as reported previously (Garthwaite et al., 2005; Islam et al., 2007).

Proteomic analysis revealed significant differences between proteomes of control and salt-treated plants as well as between both barley species. Proteins identified specifically in control plants (HVC, HMC) include proteins associated with active biosynthetic processes including cytoplasmic and organellar protein biosynthesis (40S ribosomal protein SA, 60S acidic ribosomal protein P0, 60S ribosomal protein L22-2; translation elongation factor Tu); proteins involved in carbohydrate biosynthesis (sucrose synthase), fatty acid biosynthesis [enoyl-(acyl-carrier-protein) reductase (NADH)], chlorophyll biosynthesis (magnesium chelatase 40 kDa subunit) as well as proteins involved in amino acid biosynthesis (cysteine synthase, serine hydroxymethyltransferase) and amino acid-oxoacid interconversions (aspartate aminotransferase). In contrast, proteins identified specifically in salt-treated plants (HVN, HMN) include proteins involved in ATP biosynthesis and metabolism (mitochondrial ATP synthase precursor, soluble inorganic pyrophosphatase 1) indicating enhanced need for immediately available energy in the form of ATP during the process of an active stress acclimation. Moreover, several proteins involved in stress and defense responses and redox metabolism (germin F, LEA protein D34, dehydrin 8, peroxidase 2) were found only in salt-treated samples.

Hordeum vulgare subjected to salinity reveals severe damage under high-salinity stress (300 mM NaCl) indicated by a presence of several proteins involved in apoptotic processes (apoptotic chromatin condensation inducer in the nucleus; FAS-associated factor 2-B) and proteins involved in protein ubiquitination resulting in protein targeting to proteasomal degradation (E3 UMF1-protein ligase 1 homolog; deubiquitination-protection protein dph1). A relative decrease in several proteins associated with energy metabolism such as glycolytic proteins ALDO, ENO, PGM, TPI indicates serious damage of HVN metabolism under high salinity. Changes at proteome level thus indicate processes leading to tissue damage and PCD. Our previous study indicates that H. vulgare cv. Tadmor plants subjected to one-step transfer to high salinity still not fully recovered after 1 week of recovery treatment when compared to Tadmor plants subjected to a gradual salt acclimation (Kosová et al., 2015). In the present study, the data obtained by proteomic analysis indicate that an one-step transfer to 300 mM NaCl will finally result in irreversible crown tissue damage in H. vulgare cv. Tadmor.

In contrast, H. marinum transferred to 300 mM NaCl reveals an active acclimation to high salinity indicated by enhanced levels of several proteins involved in energy metabolism (ATP metabolism), protection of intracellular structures (proteins involved in protein folding, redox reactions), and even proteins involved in sensitive processes of energy metabolism such as photosynthesis (OEE proteins in PSII). These proteins indicate an enhanced protection of stress-sensitive processes such as photosynthesis in halophytic H. marinum with respect to glycophytic H. vulgare. Enhanced levels of proteins involved in cleavage of macroergic phosphate bonds (adenylate kinase A, soluble inorganic pyrophosphatase) indicate enhanced need for immediately available energy during salt stress acclimation. The presence of several isoforms of ROS scavenging enzymes as well as detoxification enzymes such as CBS domain-containing protein CBSX3, USP, cyanate hydratase, and lactoylglutathione lyase indicates fine tuning and efficient decomposition of toxic byproducts of cellular metabolism. Moreover, enhanced levels of dehydration stress-responsive TFs such as NAC and bZIP TFs underlie enhanced levels of several stress-responsive proteins indicating an active stress acclimation. Differential states of H. vulgare and H. marinum subjected to high salt levels are also indicated by the presence of different isoforms of eIF-5A with eIF-5A1 isoform involved in apoptosis induction in H. vulgare while eIF-5A2 isoform involved in cell division induction in H. marinum.

### AUTHOR CONTRIBUTIONS

LM performed protein identification using nanoLC-ESI-Q-TOF and analysed the data. RH analyzed protein MS/MS data. KK designed the whole experiment, determined plant physiological characteristics, prepared literature review and wrote the manuscript. PV and IP participated on the manuscript and figure captions preparation.

#### ACKNOWLEDGMENTS

fpls-07-01154 August 1, 2016 Time: 13:26 # 17

The work was supported by Institutional project RO0415 for Crop Research Institute from Czech Ministry of Agriculture (MZe CZ) and by projects LD11069, LD14087 and LD15167 supported by Czech Ministry of Education, Youth and Sports (MEYS CZ) as a part of international COST actions FA0901 "Putting Halophytes to Work from Genes to Ecosystems," FA1208 "Pathogen-informed Strategies for Sustainable Crop Resistance" and FA1306 "The

#### REFERENCES


Quest for Tolerant Varieties - Phenotyping at Plant and Cellular Level," respectively. This work was supported by a grant from Specific University Research (MEYS CZ No 20/2016).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016.01154

diverse resposne pathways critical for high salinity tolerance. Front. Plant Sci. 6:30. doi: 10.3389/fpls.2015.00030




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Maršálová, Vítámvás, Hynek, Prášil and Kosová. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Proteomic Studies on the Effects of Lipo-Chitooligosaccharide and Thuricin 17 under Unstressed and Salt Stressed Conditions in Arabidopsis thaliana

Sowmyalakshmi Subramanian, Alfred Souleimanov and Donald L. Smith\*

Department of Plant Science, McGill University, Montréal, QC, Canada

Plants, being sessile organisms, are exposed to widely varying environmental conditions throughout their life cycle. Compatible plant-microbe interactions favor plant growth and development, and help plants deal with these environmental challenges. Microorganisms produce a diverse range of elicitor molecules to establish symbiotic relationships with the plants they associate with, in a given ecological niche. Lipochitooligosaccharide (LCO) and Thuricin 17 (Th17) are two such compounds shown to positively influence plant growth of both legumes and non-legumes. Arabidopsis thaliana responded positively to treatment with the bacterial signal compounds LCO and Th17 in the presence of salt stress (up to 250 mM NaCl). Shotgun proteomics of unstressed and 250 mM NaCl stressed A. thaliana rosettes (7 days post stress) in combination with the LCO and Th17 revealed many known, putative, hypothetical, and unknown proteins. Overall, carbon and energy metabolic pathways were affected under both unstressed and salt stressed conditions when treated with these signals. PEP carboxylase, Rubisco-oxygenase large subunit, pyruvate kinase, and proteins of photosystems I and II were some of the noteworthy proteins enhanced by the signals, along with other stress related proteins. These findings suggest that the proteome of A. thaliana rosettes is altered by the bacterial signals tested, and more so under salt stress, thereby imparting a positive effect on plant growth under high salt stress. The roles of the identified proteins are discussed here in relation to salt stress adaptation, which, when translated to field grown crops can be a crucial component and of significant importance in agriculture and global food production. The mass spectrometry proteomics data have been deposited to the ProteomeXchange with identifier PXD004742.

Keywords: Arabidopsis thaliana, Lipo-chitooligosaccharide, Thuricin17, NaCl salt stress, shotgun proteomics

## INTRODUCTION

Microbes are a key component of all ecosystems on earth, playing major roles in the biogeochemical cycles (Falkowski et al., 2008). Compounds secreted by the bacterial population of a rhizosphere are very species and environment dependent. Two bacterial signal compounds, lipochitooligosaccharide (LCO) from Bradyrhizobium japonicum strain 532C and Thuricin 17 (Th17),

#### Edited by:

Qingsong Lin, National University of Singapore, Singapore

#### Reviewed by:

Renu Deswal, University of Delhi, India Chiew Foan Chin, University of Nottingham Malaysia Campus, Malaysia

> \*Correspondence: Donald L. Smith donald.smith@mcgill.ca

#### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 06 June 2016 Accepted: 16 August 2016 Published: 30 August 2016

#### Citation:

Subramanian S, Souleimanov A and Smith DL (2016) Proteomic Studies on the Effects of Lipo-Chitooligosaccharide and Thuricin 17 under Unstressed and Salt Stressed Conditions in Arabidopsis thaliana. Front. Plant Sci. 7:1314. doi: 10.3389/fpls.2016.01314

a bacteriocin from Bacillus thuringiensis strain NEB17, were isolated from bacteria that reside in the soybean rhizosphere. These signal compounds were successfully isolated and characterized, with regard to plant growth promotion, in our laboratory in 2000 and 2006, respectively (Prithiviraj et al., 2000; Gray et al., 2006a,b). These two compounds are under evaluation for their capacity to promote plant growth and development in both legumes and non-legumes under laboratory and field conditions, and are being developed as low-input components of crop production systems for deployment under Canadian climatic conditions. While LCO technology is already in the market for commercial application, with products such as Optimize, marketed by Novozymes (now a part of BASF), Th17 is under evaluation for potential commercialization.

lipo-chitooligosaccharides, also referred to as Nod factors, have been reported to positively and directly affect plant growth and development in legumes and non-legumes; as signal compounds they were first reported by Dénarié and Cullimore (1993). Nod factors have since been reported to affect plant growth in diverse plant species such as tobacco (Schmidt et al., 1993), Norway spruce and Picea abies (Dyachok et al., 2000, 2002; Oláh et al., 2005), soybean (Zhang and Smith, 2001; Lindsay, 2007; Wang et al., 2012; Prudent et al., 2014), canola (Schwinghamer et al., 2014), corn (Souleimanov et al., 2002a,b; Khan, 2003; Tanaka et al., 2015) and tomato (Chen et al., 2007). The LCO induced enod genes in non-legumes code for defense related responses, such as chitinase and PR proteins (Schultze and Kondorosi, 1996, 1998), peroxidase (Cook et al., 1995; Lindsay, 2007; Wang et al., 2012) and enzymes of phenylpropanoid pathway, such as L-phenylalanine ammonia-lyase (PAL; Inui et al., 1997).

Bacillus thuringiensis NEB17 was isolated from soybean root nodules as a putative endophytic bacterium in 1998, in our laboratory; when co-inoculated with B. japonicum under nitrogen free conditions, it promoted soybean growth, nodulation and grain yield (Bai et al., 2002, 2003). Subsequently, the causative agent of plant growth promotion, a bacteriocin, was isolated from B. thuringiensis NEB17, and is now referred to as Thuricin 17 (Gray et al., 2006b). Thuricin 17 (Th17), applied either as leaf spray or as root drench, has positive effects on soybean and corn growth. This report, from our laboratory, was the first to indicate plant growth stimulation by a bacteriocin (Lee et al., 2009). Th17 is now being tested under field conditions and DuPont Canada Crop Protection and Pioneer Canada have confirmed the stimulation of plant growth by Th17 (unpublished data).

Plants are sessile multi-cellular organisms that cope with various environmental stressors that play a major role in the growth and development of plants. Under field conditions, they face a range of challenges, the most common being soil salinity, cold temperature and drought. During the late 1990s and the early 2000s, intense gene expression and mutant studies were conducted to identify the probable signal transduction pathways; and to understand the differences and commonalities between salt, drought and cold temperature stresses. Some of the key findings are summarized herein. These three abiotic stressors are physically different and yet elicit both specific and common gene responses. With nearly every aspect of plant physiology and metabolism being affected, a very complex network of signaling pathways exist, and help plants respond to these conditions (Zhu, 2001a,b). Salt stress creates both osmotic and ionic stress in plants. The ionic stress is very distinct, associated with high sodium (Na+) and potassium (K+) deficiency, and occurs a few days after the salt stress is perceived (Munns, 2002; Xiong et al., 2002). However, the osmotic stress component is common to all three mentioned abiotic stressors, thereby converging into the induction of common sets of genes (Shinozaki and Yamaguchi-Shinozaki, 1997; Zhu, 2001a,b). Excess salt in plants results in irregularities in ion homeostasis that are controlled by the cell via various ion transporters (SOS1, 2 and 3) that restrict Na<sup>+</sup> entry into the cytoplasm and regulate its accumulation in the vacuoles, and simultaneously selectively import K<sup>+</sup> ions (Hasegawa et al., 2000; Zhu, 2000). SOS1 is now known to encode for the plasma membrane localized Na+/H<sup>+</sup> antiporter which removes Na<sup>+</sup> from the cell to the outside and SOS2 encodes for a serine/threonine protein kinase. SOS3 encodes for a myristoylated calcium-binding protein and senses salt specific cytosolic Ca2<sup>+</sup> concentration. It interacts with SOS2 using calcium as the second messenger and targets vegetative storage protein 2 (VSP2) to impart salt tolerance (Gong et al., 2001), simultaneously controlling the Na+/H<sup>+</sup> antiporter system (Qiu et al., 2002). About 5% of Arabidopsis thaliana genes are involved in ion regulation (Lahner et al., 2003). Differences in calcium concentration trigger protein phosphorylation cascades that provoke mitogen-activated protein-kinases, which in turn, regulate the stress response (Chinnusamy et al., 2004).

In our previous study, regarding phytohormone quantification, we observed that LCO treated A. thaliana rosettes had increased levels of ABA and free SA, while the Th17 treated rosettes showed increased levels of IAA and SA (Subramanian, 2014). Since ABA regulation is observed in abiotic stress tolerance and IAA regulates protein degradation using the ubiquitin proteasome pathway, which decreases the toxic effects of ROS, we wanted to assess the role of LCO and Th17 in regulation of the proteome for plant growth promotion both under optimal and salt stressed conditions.

#### MATERIALS AND METHODS

#### Plant Material and Treatments

Seeds of A. thaliana Col-0 were procured from Lehle Seeds (Round Rock, TX, USA), the seeds were planted in peat pellets and the resulting plants grown in a walk-in growth chamber (Conviron Model No. PGR15, Controlled Environments Ltd, Winnipeg, MB, Canada), set at 22 ± 2 ◦C, with a photoperiod of 16/8 h day/night cycle and 60–70% relative humidity and photosynthetic irradiance of 100–120 µmol quanta m−<sup>1</sup> s −1 .

#### Extraction and Purification of Lipo-Chitooligosaccharides (LCOs) and Thuricin 17 (Th17)

The extraction and purification of LCOs was carried out and chromatography conducted for 45 min using a linear

gradient of acetonitrile from 18 to 60%, as described by Souleimanov et al. (2002a). Identification of Nod factors was conducted by comparing the retention time of standard Nod factors from strain 532C (identified by mass spectrometry).

Bacillus thuringiensis NEB17 was cultured in King's B medium (King et al., 1954) as previously described (Gray et al., 2006a). Th17 isolation and purification was carried out using a HPLC following the procedures of Gray et al. (2006b). The collected material was denoted as partially purified Th17, stored at 4◦C and diluted to required concentrations for all the experiments.

In all the germination experiments LCO concentrations of 10−<sup>6</sup> and 10−<sup>8</sup> M (referred to as LCOA and LCOB, respectively, in Figures and Tables), and Th17 concentrations 10−<sup>9</sup> and 10−<sup>11</sup> M (referred to as THA and THB, respectively) were used, the concentrations of which were found to be the best in plant growth response studies (Prithiviraj et al., 2000; Souleimanov et al., 2002b; Lee et al., 2009).

### Petri Plate Assay for Screening for Salt Stress

Seeds of A. thaliana were surface sterilized in 90% alcohol for 1 min and rinsed several times with sterile water. These seeds (25 per plate) were placed on agar plates comprised of control, 10−<sup>6</sup> and 10−<sup>8</sup> M LCO and 10−<sup>9</sup> and 10−<sup>11</sup> M Th17 treatments, to score for germination. To assess salt tolerance, the seeds (25 per plate) were placed on agar plates comprising 0, 100, 150, 200, and 250 mM NaCl in combination with 10−<sup>6</sup> and 10−<sup>8</sup> M LCO and 10−<sup>9</sup> and 10−<sup>11</sup> M Th17. Control plates were comprised of only <sup>1</sup>/<sup>2</sup> MS medium with agar and the salt controls were 100, 150, 200, and 250 mM NaCl. After 48 h of stratification, the seeds were allowed to germinate and the seedlings were allowed to grow for 20 days in a growth chamber at 22 ± 2 ◦C, with a photoperiod of 16/8 h day/night cycle and 60–70% relative humidity and photosynthetic irradiance of 100–120 µmol quanta m−<sup>1</sup> s −1 , after which the samples from the plates were assessed for visual differences in growth. Since plants in Petri plate conditions are good for screening and not for long term growth, plants were grown in trays to assess salt stress tolerance and recovery and for label free proteomic studies.

#### Tray Assay for Assessing Salt Stress Recovery of A. thaliana

Jiffy-peat pellets (Jiffy products, Plant Products Ltd., Brampton, ON, Canada) were soaked in water to saturation and seeds of A. thaliana sown on them. The trays were covered and the seeds allowed to germinate. Two and half-week-old plants were subjected to 10−<sup>6</sup> and 10−<sup>8</sup> M LCO and 10−<sup>9</sup> and 10−<sup>11</sup> M Th17 treatments, followed up by fulminant salt stress at 200, 250, and 300 mM NaCl, 48 h post bacterial signal treatments. The plants were watered regularly and allowed to grow for 15 days, after which the plants were assessed for visual symptoms of salt stress and loss of turgor.

### Leaf Proteomics using Shotgun Approach

For the proteome analysis, the rosettes sampled at 24 h post bacteria signal treatments (from control, 10−<sup>6</sup> M LCO and 10−<sup>9</sup> M Th17) comprised the unstressed group. The remaining plants were fulminant salt stressed at 250 mM NaCl, 48 h post bacterial signal treatments. Plants from 7 days of salt stress at 250 mM NaCl in combination with 10−<sup>6</sup> M LCO and 10−<sup>9</sup> M Th17 treatments were sampled as the salt stressed group. 250 mM NaCl served as the salt control. The samples were flash frozen in liquid nitrogen and stored in −80◦C until protein extraction. Total proteins from the samples were extracted using a protein extraction kit (Sigma–Aldrich, PE-2305, St. Louis, MO, USA).

#### Protein Extraction

In brief, the sampled (pool of three plants per replicate) rosettes were ground to a fine powder in liquid nitrogen. Approximately 100 mg of the fine powder was placed in sterile eppendorf tubes and 1 mL of ice cold methanol (Cat no. 15468-7, Sigma–Aldrich Co., St. Louis, MO, USA) was added, vortexed, incubated in −20◦C for 20 min. and centrifuged (Micro12, Fisher Scientific, Denver Instrument Co., USA) at 13,000 rpm for 7 min at 4◦C. The supernatant was discarded and the procedure was repeated twice more, followed by similar incubation in acetone (Cat. no. 179124, Sigma–Aldrich, Co., St. Louis, MO, USA), both steps in order to remove phenolics and secondary metabolites that might otherwise interfere with LC-MS/MS analysis. The RW2 solution was added to the samples after removing acetone, vortexed for 30 s and incubated at room temperature (22◦C) for 15 min. The samples were then centrifuged at 13,000 rpm for 10 min and the supernatant carefully collected in a fresh sterile tubes. The supernatant constituted total proteins from that sample. The proteins were then diluted and quantified using the Lowry method, and samples of 10 µg in 20 µL of 1 M urea were taken to the Institut de recherches cliniques de Montréal (IRCM) for label free proteomic analysis using LC-MS/MS.

#### Protein Profiling

The total protein extracts were then digested with trypsin and subjected to LC-MS/MS using LTQ-Velos Orbitrap (Thermo Fisher, Waltham, MA, USA). Tandem mass spectra were extracted, charge state deconvoluted and deisotoped, and all MS/MS samples were analyzed using Mascot software (Matrix Science, London, UK; version 2.3.02). Mascot was set up to search the A. thaliana database (txid\_3702, 80416 entries) assuming the digestion enzyme trypsin. Mascot was searched with a fragment ion mass tolerance of 0.60 Da and a parent ion tolerance of 15 ppm. Carbamidomethyl of cysteine was specified in Mascot as a fixed modification. Oxidation of methionine was specified in Mascot as a variable modification.

#### Criteria for Protein Identification

Scaffold (version Scaffold 4.0, Proteome Software Inc., Portland, OR), was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95.0% probability, as specified by the Peptide Prophet algorithm (Keller et al., 2002). Protein

identifications were accepted if they could be established at greater than 99.0% probability and contained at least two identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm (Nesvizhskii et al., 2003). Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony.

#### Data Analysis

Experiments were structured following a completely randomized design. The SAS Statistical Package 9.3 (SAS Institute Inc., Cary, NC, USA) was used, and within this the Proc Mixed procedure and Tukey's multiple means comparison when there was significance at the 95% confidence level. Data transformation was applied when necessary to meet the criteria for analysis of variance for seed germination.

Scaffold 4.0 was used for analyzing the proteomics data for fold change and Fisher exact test of the identified proteins after subjecting the quantitative value of the spectra to the embedded normalization. The FASTA file generated was analyzed using Blast2GO-Pro V.2.6.6 (Conesa et al., 2005; Conesa and Götz, 2008; Götz et al., 2008, 2011), for the functional annotation and analysis of the protein sequences. Apart from these, Enzyme code (EC), KEGG maps and InterPro motifs were queried directly using the InterProScan web service. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium<sup>1</sup> via the PRIDE partner repository (Vizcaino et al., 2013) with the dataset identifier PXD004742.

#### RESULTS

### A. thaliana Seed Germination in Petri Plates and Trays under Unstressed Conditions and in the Presence of Salt Stress Screening, When Treated with LCO and Th17

LCO and Th17 (LCO – 10−<sup>6</sup> and 10−<sup>8</sup> M; Th17 – 10−<sup>9</sup> and 10−<sup>11</sup> M) treatments generally had no effect on A. thaliana seed germination except at 30 h, when conditions were carefully maintained as optimal (**Table 1**). Hence an evaluation of the effects of these signal compounds in the presence of salt stress was conducted; a NaCl dosage response screening was performed. This work suggested that, in the presence of signal compounds, the plants could withstand up to 200 mM NaCl in Petri plates while the 250 mM NaCl stress completely inhibited root growth (**Figure 1**). However, the Petri plate assay is only good for early growth determinations as the seedlings are in an enclosed environment and this can induce other stresses. Hence, plants were grown in trays and two and half-week-old plants were screened with 200, 250, and 300 mM NaCl. After 48 h of treatment with LCO and Th17, the plants were treated with 200, 250, and 300 mM NaCl, allowed to recover from the shock and assessed for visual signs of stress 15 days after the NaCl stress was imposed. The plants could tolerate 250 mM NaCl while at 300 mM, the visible signs of stress were obvious such as retarded plant growth and loss of turgor (**Figure 2**). This also reflected on the plants fresh and dry weight (**Figure 3**). Hence, there were clear beneficial effects of LCO and Th17 for 250 mM NaCl treated plants and a 7 days post exposure condition was selected for rosette proteomics since the effects of salt stress are known to manifest from day 7 of exposure to NaCl salt.

#### Protein Profiling

To understand the effect of LCO and Th17 on unstressed and salt stressed A. thailana rosettes, total proteins were extracted from the samples and subjected to LC-MS based proteome profiling. Based on the quantitative value of the identified spectra, and at 99% protein probability, with two minimum peptides and 95% peptide probability, a number of proteins were identified in the unstressed and salt stressed treatments (**Table 2**). The treatment contrasts were then analyzed for fold-change after normalization, and Fisher's Exact test was used to narrow down the up- and down-regulated proteins, to predict their probable functions at 24 h after signal compound treatment and 7 days after NaCl stress imposition. It is likely that we missed some of the relevant proteins due to very strict criteria for difference detection during data analysis; this level of stringency was utilized for ease of subsequent functional interpretation. According to the fold-change patterns and Fisher's Exact test of the contrasts, the proteins were categorized as known proteins, putative proteins, hypothetical and unknown proteins (**Tables 3A,B**; Supplementary Datasheets 1 and 2).

Based on the known and predicted proteins, prominent proteins functioning under unstressed control conditions included O-methyltransferase, pyrophosphatase, COR15A and B, legume lectin family, actin 7, membrane associated progesterone binding protein, legume lectin family, a chloroplast drought induced stress protein, phosphoglycerate kinase, mitochondrial HSP70, thiamin C, profilin, TIC 40 (TRANSLOCON AT THE INNER ENVELOPE MEMBRANE OF CHLOROPLASTS) universal stress protein and a putative jasmonate inducible protein. Some of the notable proteins up-regulated in LCO treated plants were members of 40 and 60 S ribosomal protein family, phosphoenolpyruvate carboxylase (PEPC) 1 and 2, proteins of the photosystem I subunit and photosystem II 47kD, D1 and D2 proteins, acetyl CoA carboxylase, calcium sensing receptor (CaS), lipoxygenase, RUBISCO large subunit, members of oxidoreductases, fibrillin family, TIC 40 and 110, LEA protein family, peroxisomal glycolate oxidase, cadmium sensitive, cell division cycle related protein and vestitone reductase. Th17 treated rosettes had all of the above proteins up-regulated in LCO treated rosettes in addition to COR13, hydroxyl-proline rich, a major latex protein, catalase, light harvesting complex proteins LHB1B1, LHCA2, LHCB5, LHCB6, nodulin related, Zn binding oxidoreductases and progesterone binding protein (Supplementary Datasheets 1 and 2 for fold change and Fisher's exact test results for A. thaliana signals group contrasts).

The number of significant proteins identified in the salt stressed signals group did not alter much in number, but

<sup>1</sup>http://proteomecentral.proteomexchange.org


TABLE 1 | Least square means of Arabidopsis thaliana percentage germination – seeds treated with lipo-chitooligosaccharide and Thuricin 17 under optimal conditions.

Means associated with the same letter are not significantly different at p ≤ 0.05. Bold values indicate the only times when statistical significance occurred under optimal germination conditions.

did so in the types of proteins that seemed to be regulated in response to salt stress. Some of the prominent proteins of the 250 mM salt stress control were 40, 50, and 60 S ribosomal proteins, a methyl jasmonate esterase, CaS, cadmium sensitive protein 1, pyruvate decarboxylase, phosphoglycerate kinase, seed maturation protein, stromal ascorbate peroxidase, TIC 40, cinnamyl-alcohol dehydrogenase, pyruvate kinase, peroxidase, Fe-superoxide dismutase, chitinase, plastocyanin, and profiling 1. The LCO treated and salt stressed group, however, up-regulated a very different set of proteins comprised of COR15A, cytochrome B5 isoform E, glucose-phosphate-6 isomerase, LHCB 4.2 and protein D1 of photosystem II, nodulin related protein, photosystem 1 P700 chlorophyll apoprotein A1, plastid-lipid associated protein, NADH-cytochrome B5 reductase, NADPH oxidoreductases, allene oxide synthase and cyclase. Th17 treated and 250 mM NaCl stressed rosettes upregulated proteins, some of which were common to both the salt control and LCO with salt groups. Apart from these, also affected were ATP citrate synthase, alcohol and aldehyde dehydrogenases, seed maturation protein, cinnamyl-alcohol dehydrogenase 4, cadmium sensitive, glutathione synthase transferase, PIPIB (a plasma membrane water channel protein), importin subunit, APE2 (Acclimation of leaf photosynthesis), myo-inositol, isocitrate dehydrogenase and vestitone reductase. (Supplementary Datasheets 1 and 2 for Fold change and Fisher's exact test results for A. thaliana signals with stress group contrasts).

Based on Blast2GO Pro results, the enzyme code distribution for both the unstressed and salt stressed rosettes were studied. A sharp increase in some of the main enzyme classes was observed within the salt stress group, as compared to those within the unstressed group (**Table 4**). The sharp increase in oxidoreductases in LCO with stress and hydrolases in Th17 with stress could be explained through possible roles in salt stress alleviation.

The GO function distribution characteristics of the unstressed and salt stressed groups also indicated that the proteins identified were mostly associated with ATP, GTP, protein and nucleotide binding, metal ion binding and specific to zinc, copper, cadmium, cobalt, magnesium, and calcium, response to salt and cold, glycolysis, pentose-phosphate shunt, gluconeogenesis, thylakoid associated, photorespiration, oxidation-reduction processes and photosystem II assembly (**Figures 4–6**).

Molecular function, biological processes and cellular components were all affected in both unstressed and salt-stressed conditions. Translation, translation elongation factor activity,

FIGURE 3 | Screening assay in trays for A. thaliana – fresh weight and dry weight of A. thaliana rosettes in response to different levels of salt stress in the presence of LCO and Th17, 15 days after imposition of salt stress. (Optimal: Control – Water; LCOA – 10−<sup>6</sup> M, LCOB – 10−<sup>8</sup> M, THA – 10−<sup>9</sup> M, THB – 10−<sup>11</sup> M; 150, 200, 250, and 300 mM NaCl in combination with LCOA – 10−<sup>6</sup> M, LCOB – 10−<sup>8</sup> M, THA – 10−<sup>9</sup> M, THB – 10−<sup>11</sup> M treatments).



TABLE 3A | Grouping of proteins in A. thaliana rosettes, that was significant in contrasts based on fold change.


TABLE 3B | Grouping of proteins in A. thaliana rosettes, that were significant in contrasts based on Fisher's Exact test.


TABLE 4 | Enzyme code distribution in un-stressed and salt stressed groups.


The % increase or decrease in the main enzyme classes within the groups (Optimal and Salt stress) is mentioned in the brackets.

thylakoid membrane organization proteins, starch biosynthetic process, proteins of the stromule and the extracellular region were all arrested in the salt stressed group. However, magnesium ion binding proteins, proteins related to misfolded protein responses, proteasome core complex assembly and hyperosmotic stress response, cytosolic ribosome and plant-type cell wall were all prominent in the salt stressed group. Apart from these, proteins in the cytosol, plasma membrane, chloroplast and its envelope, apoplast, plasmodesmata, nucleus and the vacuole were all up-regulated. Very little change was observed in GTP binding and GTPase activity, copper ion binding, rRNA processing, photosystem II assembly, plastoglobule, vacuole membrane, chloroplast thylakoid and stroma. The protein report for all the treatments is included as Supplementary datasheet 3 and peptide report for all the treatments is included as Supplementary datasheet 4.

#### DISCUSSION

Nod Bj V (C18:1;MeFuc), a major LCO molecule produced by B. japonicum 532C, isolated and identified in our laboratory, has been reported to have a positive and direct effect on both legume and non-legume seed germination, plant growth and development (Prithiviraj et al., 2003). Other than soybean and common bean, LCO can also enhance seed germination and seedling establishment in maize, rice, canola, apple, and grapes, and is accompanied by increased photosynthetic rates (Zhang and Smith, 2001). Investigations into these effects, at the molecular level, led us to transcriptomic studies, and microarray studies on soybean leaves sprayed with LCOs under optimal and sub-optimal growth conditions. The optimal condition microarray revealed 639 differentially expressed genes out of which 13 were related to abiotic stress, 14 related to biotic stress,

treatments in A. thaliana rosettes.

3 to salicylic acid and 7 to cytochrome P450s at 48 h post treatment (Lindsay, 2007). The sub-optimal stress microarray revealed the differential expression of over 600 genes. Many of these were defense and stress response related, or transcription factors suggesting the effects of LCO on the transcriptome of the leaves at 48 h post treatment (Wang et al., 2012). These results suggest a need to further explore the mechanisms by which microbe-to-plant signals might help plants accommodate abiotic and biotic stress conditions. Th17 however, has not been studied as well as LCO, as it is more recently isolated. We have some information regarding its effects on soybean and corn plant growth. The leaves of 2-week-old soybean leaves sprayed with Th17 showed increased activities of lignification-related and antioxidative enzymes and their isoforms. Both leaf spray and root drench of soybean and corn with Th17 stimulated plant growth (Jung et al., 2008; Lee et al., 2009).

Hence, in this study we subjected A. thaliana plants to salt stress to evaluate the efficacy of both LCO and Th17 under unstressed and salt stressed conditions for proteome profiling. A. thaliana is a glycophyte and sensitive to salt. The roots of A. thaliana seedlings were severely affected at 200 mM NaCl, in a Petri plate assay used to study salt stress (0, 50, 100, 150, 200, and 250 mM NaCl; Jiang et al., 2007). Our study shows that, in the presence of the bacterial signal compounds A. thaliana showed retarded root growth only at 250 mM NaCl. These compounds alleviated salt stress up to 250 mM NaCl when A. thaliana grown in trays were exposed to NaCl stress. At 300 mM NaCl stress, obvious stress related symptoms such as retarded plant growth and loss of turgor in the leaves were observed.

Proteins play central roles in essentially all metabolic processes. The advances in instrumentation and bioinformatic analysis have increased our understanding of proteins and their

treatments in A. thaliana rosettes.

effects, which can provide key evidence regarding shifts in plant physiology. Despite these advances, proteome profiling in systems biology are still a major challenge; but the amount of information they can add to the understanding of a biological system is impressive. In this study we used the label free proteomics approach to understand A. thaliana proteomic responses in the presence of microbial signal compounds and under unstressed and 250 mM NaCl stressed conditions and discuss some of the key proteins identified.

Translation, translation elongation factor activity, thylakoid membrane organization proteins, starch biosynthetic process, proteins of the stromule and the extracellular region were all arrested in the salt stressed group, suggesting that the plants were arresting these processes in order to compensate for energy dependent activities associated with Na<sup>+</sup> ion flushing from the cytosol. However, magnesium ion binding proteins, proteins related to cadmium response, misfolded protein responses, proteasome core complex assembly and hyperosmotic stress response, cytosolic ribosome and plant-type cell wall were all prominent in the salt stressed group. Ligands for metals such as cadmium (Cd), copper (Cu), nickel (Ni), and zinc (Zn) are seen in all plant tissues and in abundance in the xylem sap where they form complexes with histidine and citrates in the xylem sap moving from roots to leaves. The Cd binding complexes are found in both the cytosol and, predominantly, in the vacuole of the cell (Rauser, 1999). Apart from these, proteins in the cytosol, plasma membrane, chloroplast and its envelope, apoplast, plasmodesmata, nucleus and the vacuole were all upregulated, while very little change was observed in GTP binding and GTPase activity, copper ion binding, rRNA processing, photosystem II assembly, plastoglobule, vacuole membrane, chloroplast thylakoid and stroma.

The significant findings in our study were the up-regulation of the chloroplast proteins and the proteins from photosystems I and II in the LCO and Th17 treatments, since these are generally strongly and negatively affected by salt stress. During abiotic stresses, photosynthetic capacity is reduced due to damage to photosynthetic pigments of the photosystems I and II, resulting in reduced light absorption capacity (Zhang et al., 2011; Ashraf and Harris, 2013). The stromal proteome of Arabidopsis chloroplasts represented 10% of the 241 proteins identified to be involved in chloroplast protein synthesis and biogenesis, with 75% being associated with the oxidative pentose phosphate pathway, glycolyis and Calvin cycle, 5–7% with nitrogen metabolism and the rest with other biosynthetic pathways such as fatty acid metabolism, amino acid metabolism, nucleotides, vitamins B1 and 2, tetrapyrroles, lipoxygenase 2 and a carbonic anhydrase (Peltier et al., 2006). The plastoglobule proteome of the chloroplast includes a M48 metallopeptidase, Absence of bc1 complex (ABC1) kinases and fibrillins, together constituting about 70 % of the plastoglobule protein biomass. The fibrillins present in other parts of the chloroplast are partitioned, probably based on their isoelectric point and hydrophobicity, to specific functions such as chlorophyll degradation and senescence, plastid proteolysis, isoprenoid biosynthesis, redox and phosphoregulation of the electron flow, although most of the functions of the associated proteins are still not clear (Lundquist et al., 2012).

Plants are exposed to various levels of light in nature and one of the ways they compensate for this is by regulating their thylakoid membrane proteins. Light harvesting protein complex protein phosphorylation is catalyzed by light-dependent protein kinase mediated by plastoquinone and is driven by the electron transport system based on light dosage (Ranjeva and Boudet, 1987). Photosystem II light interception is mediated by pigment proteins that belong to a large class of antenna pigments. Light harvesting complex II (LHC II) is the most

abundant of the photosystem II proteins; the apoprotein and pigment-protein holocomplex is structurally very heterogenous. The LHC II apoproteins Lhcb1, Lhcb2, and Lhcb3 are coded by Lhcb1, Lhcb2, and Lhcb3 genes. Once synthesized in the cytoplasm as precursors, these are transported into the chloroplast by post-translational modifications (Jackowski et al., 2001). A short term post-translational redistribution of LHC and a long term chloroplast DNA transcription, balance the regulation of photosystems I and II, and is dictated by the redox state of plastoquinone which is in turn controlled by chloroplast sensor kinase (CSK; Allen et al., 2011).

Exposure to light stress causes oxidative and nitrosative stress and the proteins of photosystem I and II are affected differentially. Amino acid oxidation products are determined mostly in the photosystem II reaction center, and this often leads to tyrosine and tryptophan oxidation or nitration (Galetskiy et al., 2011). About 80–200 proteins present in the thylakoid lumen are closely associated with the light harvesting complexes and the other proteins regulating photosynthesis. Following an 8 h light exposure, PsbP and PsbQ subunits of photosystem II were seen to increase along with a major plastocyanin and various proteins of unknown function. These proteins also seem to be similarly expressed at the transcription level (Granlund et al., 2009). The excess light energy perceived by plants is channeled into the chloroplasts and dispersed by the mitochondrial respiratory chain. The type II NAD(P)H dehydrogenases in the inner membrane of the mitochondria, cyanide-resistant alternative oxidase and phosphorylating pathway complexes I, III, and IV regulate this energy processing. Along with the glycine decarboxylase complex (GDC), these pathways regulate the energy balance between chloroplast and mitochondria under stressful conditions, wherein these pathways are up-regulated to maximize photosynthetic efficiency (Noguchi and Yoshida, 2008).

Carbonylation of proteins in organisms is an irreversible and oxidative process that increases with age and leads to disfunction of modified proteins in the system. In Arabidopsis, protein carbonylation is found to increase as the plant grows, but is seen to decrease drastically during the onset of bolting and flowering. Hsp70, ATP synthases, RUBISCO large subunit and proteins of the light harvesting complex and of energy transfer are all targets of this mechanism (Johansson et al., 2004). Despite the high salt stress levels imposed on the plants in this experiment, the photosystem proteins were still up-regulated in the LCO and Th17 treatments, suggesting that these two signals are preventing the damage of these photosystem proteins in a way we still do not understand.

The other up-regulated proteins in this experiment include those of PEPC and phosphoenolpyruvate carboxylase kinase (PPCK) in the LCO and Th17 treatments, which are two major cytosolic enzymes central to plant metabolism. During phosphate deprivation, A. thaliana responds by phosphorylating PEPC subunits to modulate the metabolic adaptations of lower phosphate availability (Gregory et al., 2009). The plant type – PEPC genes encode for 110-kDa polypeptides. These peptides are homotetrameric and contained several conserved sites for serinephosphorylation and lysine-mono-ubiquitination (O'Leary et al., 2011; and references therein).

The up-regulation of the proteasome pathway indicates that the stress generated toxic proteins might be degraded by this system. In previous studies, the 20S proteasome pathway was seen to be up-regulated in both RNA and protein levels of cadmium stressed Arabidopsis leaves, suggesting that this proteasome pathway might help with degrading stress generated oxidized proteins (Polge et al., 2009). Also, the 26 S proteasome of A. thaliana contains 26 unique proteins with at least 13 of them containing tryptophan residues as identified using nano-flow liquid chromatography (Russell et al., 2013). Post-transcriptional gene regulation is, in part, controlled by RNA-binding proteins (RBP). The A. thaliana genome encodes for more than 200 RBPs and they contribute to diverse developmental processes, chromatin modification and environmental adaptation (Lorkovic, 2009).

LEA proteins, initially discovered and researched in seeds, are now reported to be present in other vegetative tissues and have a wide range of sequence diversity and intercellular localization, with expression patterns depending on environmental conditions. The majority of the predicted LEA proteins are highly hydrophilic and found mostly in unfolded conditions, being involved largely with cellular dehydration tolerance. In Arabidopsis, nine distinct groups of LEA proteins, encoded by 51 different LEA protein genes, have been reported; most harbor abscisic acid response (ABRE) and/or low temperature response (LTRE) elements in their promoters (Hundertmark and Hincha, 2008). Up-regulation of LEA proteins in LCO and Th17 under salt stress correlates with these findings in this abiotic stress tolerance mechanism.

Up-regulation of A. thaliana membrane-associated progesterone binding protein was observed in Th17 treatment. Progesterone 1 was detected in apple seeds as early as 1968, but due to technical challenges in instrumentation and reliability of assays, the role of progesterone in plants was conclusively established by Saden-Krehula et al. (1991; Janeczko and Skoczowski, 2005). It has now been detected in a variety of dicots and monocots such as adzuki bean, mung bean, pea, tomato, potato, apple, onion, rice, and Arabidopsis, with the shoots having relatively more abundant progesterone 1 than inflorescences, seeds, roots, and tubers. Progesterone 1 was also seen to promote plant growth at very low concentrations (range of 0.01–1 µM) suggesting that this could be playing the role of a hormone in plants, regulating growth and development (Iino et al., 2007; Nakano and Yokota, 2007). It is possible that this membrane associated protein functions in a way better than LCO in promoting plant growth.

Nodulin genes once thought to be specialized genes present only in legumes have been observed in some non-legumes. The role of nodulin genes might be diverse and related to general organogenesis, rather than restricted to nodulation. Nodulin genes are found in high transcript levels in floral tissues (Szczyglowski and Amyot, 2003). Other enod40 genes have been cloned from non-legumes include tomato (Vleghels et al., 2003),

maize (Compaan et al., 2003), rye grass (Lolium) and barley (Hordeum) (Knud, 2003). Enod40 levels are elevated during arbuscular mycorrhizal root colonization of tobacco (Nicotiana bentana) and alfalfa (Medicago truncatula) (Sinvany et al., 2002). Azorhizobium caulinodans ORS571 colonized the roots of A. thaliana through lateral root cracks and the colonization was improved upon addition of flavonoids naringenin and daidzein. Both colonization and flavonoid stimulation were Nod gene independent (Gough et al., 1997). Early nodulin like protein has been observed to accumulate during the early stages of sieve cell differentiation (Khan et al., 2007). Investigation of transgenic lines of Arabidopsis for early nodulin gene enod40 function showed reduction in cell size of selected tissues in the plant, such as the leaf mesophyll and the epidermal internode cells (Guzzo et al., 2005). Nodulin protein analogs in watermelon control fruit development and ripening (Wechter et al., 2008). Roles for nodulin genes have been reported in tomato fruit development and ripening (Lemaire-Chamley et al., 2005). Up-regulation of nodulin related proteins in both LCO and Th17 with NaCl stress suggests that this nodulin related protein is mostly playing a role in plant development during stress tolerance.

#### CONCLUSION

In this study, we compared the effects of LCO and Th17 under unstressed and salt-stressed conditions; this is the first study conducted to determine the effects of these signals, in combination with stressful levels of salt, on A. thaliana proteome. A. thaliana is a glycophyte and is sensitive to salt stress. LCO is commercially available (products such as Optimize with

#### REFERENCES


LCO promoter technology) and is known to accelerate plant growth in the field. The comparison between LCO and Th17 and the effects on the proteome of the rosettes under stressed and unstressed conditions is another step to understanding the effects of these compounds, at the proteome level, during plant growth. This study also increases our understanding of plant– microbe interactions, mainly in the use of such growth promoting technologies, boosting the potential for decreased use of synthetic chemical inputs on cultivated land, and perhaps enhanced crop productivity on salinized soils around the world.

#### AUTHOR CONTRIBUTIONS

SS designed, performed experiments and data analysis; AS and DS contributed to reagents/materials/analysis tools; SS and DS wrote the manuscript.

#### ACKNOWLEDGMENTS

The authors are grateful to the Natural Sciences and Engineering Research Council of Canada (NSERC), Green Crop Network for the funding. LC-MS/MS service provided by Institut de recherches cliniques de Montréal (IRCM) for label free proteomic analysis is highly appreciated.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016.01314




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Subramanian, Souleimanov and Smith. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Analysis of Copper-Binding Proteins in Rice Radicles Exposed to Excess Copper and Hydrogen Peroxide Stress

Hongxiao Zhang<sup>1</sup> , Yan Xia<sup>2</sup> , Chen Chen<sup>2</sup> , Kai Zhuang<sup>2</sup> , Yufeng Song<sup>2</sup> and Zhenguo Shen<sup>2</sup> \*

<sup>1</sup> College of Agriculture, Henan University of Science and Technology, Luoyang, China, <sup>2</sup> College of Life Sciences, Nanjing Agricultural University, Nanjing, China

Copper (Cu) is an essential micronutrient for plants, but excess Cu can inactivate and disturb the protein function due to unavoidable binding to proteins at the cellular level. As a redox-active metal, Cu toxicity is mediated by the formation of reactive oxygen species (ROS). Cu-binding structural motifs may alleviate Cu-induced damage by decreasing free Cu2<sup>+</sup> activity in cytoplasm or scavenging ROS. The identification of Cu-binding proteins involved in the response of plants to Cu or ROS toxicity may increase our understanding the mechanisms of metal toxicity and tolerance in plants. This study investigated change of Cu-binding proteins in radicles of germinating rice seeds under excess Cu and oxidative stress using immobilized Cu2<sup>+</sup> affinity chromatography, two-dimensional electrophoresis, and mass spectra analysis. Quantitative image analysis revealed that 26 protein spots showed more than a 1.5-fold difference in abundances under Cu or H2O<sup>2</sup> treatment compared to the control. The identified Cu-binding proteins were involved in anti-oxidative defense, stress response and detoxification, protein synthesis, protein modification, and metabolism regulation. The present results revealed that 17 out of 24 identified Cu-binding proteins have a similar response to low concentration Cu (20µM Cu) and H2O<sup>2</sup> stress, and 5 out of 24 were increased under low and high concentration Cu (100µM Cu) but unaffected under H2O<sup>2</sup> stress, which hint Cu ions can regulate Cu-binding proteins accumulation by H2O<sup>2</sup> or no H2O<sup>2</sup> pathway to cope with excess Cu in cell. The change pattern of these Cu-binding proteins and their function analysis warrant to further study the roles of Cu ions in these Cu-binding proteins of plant cells.

Keywords: Cu stress, Cu-binding protein, H2O2 stress, immobilized metal affinity chromatography, germinating rice seed

### INTRODUCTION

Copper (Cu), an essential micronutrient required for growth and development in all plants, is a structural and catalytic cofactor of several proteins and enzymes involved in electron transfer and redox reactions. More than 100 proteins comprising two groups are estimated to have the ability to complex with Cu in Arabidopsis: Cu-binding proteins/chaperones and enzymes (Häensch and Mendel, 2009). However, excess Cu is toxic to most plants, causing a wide range of deleterious effects such as the inhibition of photosynthesis and pigment synthesis, damage to

#### Edited by:

Dipanjana Ghosh, National University of Singapore, Singapore

#### Reviewed by:

Arkadiusz Kosmala, Polish Academy of Sciences, Poland Yanwei Cheng, Luoyang Normal University, China Jisen Shi, Nanjing Forestry University, China

> \*Correspondence: Zhenguo Shen zgshen@njau.edu.cn

#### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 08 May 2016 Accepted: 02 August 2016 Published: 17 August 2016

#### Citation:

Zhang H, Xia Y, Chen C, Zhuang K, Song Y and Shen Z (2016) Analysis of Copper-Binding Proteins in Rice Radicles Exposed to Excess Copper and Hydrogen Peroxide Stress. Front. Plant Sci. 7:1216. doi: 10.3389/fpls.2016.01216 plasma membranes, and other metabolic disturbances. At the cellular level, excess Cu can inactivate and disturb the protein structure via unavoidable protein binding (Yruela, 2009). To control metal homeostasis and redox status, plants have several mechanisms of metal tolerance, including exclusion, compartmentalization, and binding to organic ligands such as organic acids, amino acids, peptides, and proteins (Hall, 2002; Yruela, 2009). Recently, the molecular and physiological basis for plant interactions with metals has attracted considerable interest. The identification of metal-binding proteins involved in the responses of plants to metal toxicity may improve our understanding regarding the mechanisms of metal toxicity and tolerance in plants.

Moreover, as a redox-active metal, Cu<sup>+</sup> can catalyze the formation of reactive oxygen species (ROS) such as the superoxide anion (O•− 2 ), hydrogen peroxide (H2O2), and hydroxyl radical (HO·) via Fenton-type reactions (Schützendübel and Polle, 2002). ROS can oxidize proteins, unsaturated fatty acids, and nucleic acids, resulting in cellular damage and cell death. To scavenge ROS and alleviate their deleterious effects, plants have evolved various protective mechanisms that use superoxide dismutase (SOD), catalase (CAT), peroxidase (POD), ascorbate peroxidase (APX), and glutathione reductase (GR). Some antioxidant enzymes such as SOD have high affinities for binding to Cu, zinc (Groppa et al., 2008), manganese (Weeks et al., 2006), or iron (Fe). However, ROS can serve as signaling molecules for the induction of plant responses to environmental stresses such as metals (Babu et al., 2003; Maksymiec, 2007; Tamás et al., 2010). Cho and Seo (2005) reported that a reduced H2O<sup>2</sup> accumulation increases cadmium (Cd)-tolerance in Arabidopsis seedlings. Exogenous H2O<sup>2</sup> supplied to rice seedlings increased glutathione (GSH) levels and protected against subsequent Cd stress (Chao et al., 2009). H2O<sup>2</sup> may be involved in the regulation of Cd- and heat shock-increased APX and GR activities in rice leaves (Chou et al., 2012). The improved Cd tolerance in rice seedlings by H2O<sup>2</sup> may be due to stimulation of the antioxidant system and Cd sequestration (Hu et al., 2009). Although numerous physiological and biochemical analyses have examined the responses of plants to metal toxicity and the role of H2O<sup>2</sup> as signaling molecules regulating metal-responsive protein accumulation in plants, the process remains unclear.

Immobilized metal affinity chromatography (IMAC) combined with mass spectrometry (MS) has been used to investigate the metal-binding proteome (She et al., 2003; Smith et al., 2004; Kung et al., 2006; Tan et al., 2010; Sun et al., 2011). This technique can separate proteins from biological samples based on specific interactions between proteins in solution and metal ions immobilized on a solid support (Porath et al., 1975; Sun et al., 2005). Metal ions are typically complexed with chelating ligands such as iminodiacetic acid (IDA). The proteins are separated according to their affinity for the chelated metal ions. In the bacterium Streptococcus pneumoniae, 232 and 166 putative metal-binding protein species were respectively isolated using a Cu- and Zn-IMAC column (Sun et al., 2011). Metals often bind proteins at specific coordination sites involving Cys, His, and Met residues (Harding, 2004). Smith et al. (2004) used a Cu-IMAC approach to enrich Cu-binding proteins in hepatocellular cells and reported nine putative metal-binding domains, namely, H–(X)n–H (n = 0–5) and C–(X)m–C (m = 2–4). Kung et al. (2006) identified 35 putative Cu-binding proteins in Arabidopsis roots, and found that 29 protein species possessed one or more of the H–(X)n–H (n = 0–5) and C– (X)m–C (m = 2–4) metal-binding motifs proposed by Smith et al. (2004). Kung et al. (2006) further identified the top six candidate motifs (H–(X)5–H, H–(X)7–H, H–(X)12–H, H–(X)6– M, M–(X)7–H, and H–(X)3–C), which accounted for 31 of the 35 proteins (89%). Tan et al. (2010) identified 35 weak and 48 strong Cu2+–IMAC-interactions in Arabidopsis mitochondria. Based on their data, 72% of the identified Cu-binding proteins contained one or more of the top six Cu-binding motifs (H–(X)5–H, C–(X)7–H, H–X–C, H–(X)2–M, M–(X)3–H, or M–(X)7-H). However, limited information is available on the metal-binding proteome in plants and other organisms under excess metal stress conditions.

Rice (Oryza sativa L.), an important food crop worldwide, is often used as a model for monocotyledons because of its well-established database. Several proteomic studies have been conducted on seed germination, growth regulation, and stress responses in rice (Ahsan et al., 2007; Aina et al., 2007; Yang et al., 2007; Zang and Komatsu, 2007; Zhang et al., 2009; Lee et al., 2010; Wu et al., 2011; Song et al., 2013). In a previous study, we developed a novel IMAC method, in which the IDA-Sepharose column was applied prior to a Cu-IMAC column to remove metal ions from protein samples for separating and isolating Cubinding proteins from Cu-treated rice roots (Song et al., 2014). By comparing the difference of Cu-binding proteins in the roots of Cu-tolerant and Cu-sensitive rice varieties exposed to excess Cu (Chen et al., 2015), we had found some Cu-binding proteins involved in Cu tolerance in rice, but we did not know by which pathway these proteins were accumulated. We hypothesized that ROS signal molecules, especially those induced by Cu, might be involved in the Cu-binding proteins accumulation. In this study, we further identified soluble proteins isolated from the Cu-IMAC column that are regulated by Cu or H2O2. The aim of this report was to characterize the mechanisms involved in excess Cu stress responses and the role of H2O<sup>2</sup> as a signaling molecule or redox substrate in the expression of soluble Cu-binding proteins in plants.

#### MATERIALS AND METHODS

#### Plant Growth and Treatment

Rice seeds (O. sativa L. cv. Wuyunjing No. 7, obtained from company of Nanjing Shenzhou Seed) were surface-sterilized with 5% (v/v) sodium hypochlorite (NaClO) for 15 min and thoroughly washed in distilled water. Each treatment was performed in triplicate. For one replicate, 100 seeds were randomly placed on moist filter paper in 200 mm Petri dish. The seeds were germinated in the dark at 25◦C with renewal of distilled water every day. After 4 days, these germinating rice seeds were transferred to the mesh over 2.5 L vessel containing different concentrations of Cu sulfate pentahydrate (CuSO4·5H2O) solution (0–200µM) for 0–48 h, 1 mM ascorbic acid (Asc) for 12 h or 10 mM H2O<sup>2</sup> solution for 6 h. A certain number of radicles from each replicate were obtained for the below experiments.

## Histochemical Detection of H2O<sup>2</sup>

H2O<sup>2</sup> formation in situ in rice radicles was visually detected based on the infiltration of 3,3′ -diaminobenzidine (DAB) as described by Romero-Puertas et al. (2004) with minor modifications. Briefly, six radicles from each replicate (each Petri dish) were immersed in a 1 mg/mL solution of DAB (pH 3.8) and incubated at room temperature for 20 min in the absence of light. After staining, images were captured with a Coolpix 4500 digital camera (Nikon, Tokyo, Japan).

## H2O<sup>2</sup> Determination in Extracts

The concentration of H2O<sup>2</sup> in rice radicles from Cu-treated plants was measured by monitoring the A415 of the titaniumperoxide complex following the method described by Jiang and Zhang (2001). Absorbance values were calibrated to a standard curve established with 0.1–1.0µM H2O2.

### Protein Extraction

Rice seeds germinated for 4 days were treated with 10 mM H2O<sup>2</sup> for 6 h or with 20 and 100µM Cu for 12 h, referred as low and high concentration Cu treatment, respectively. Seeds germinated in deionized water without Cu and H2O<sup>2</sup> were used as controls. Radicles were harvested and ground with a mortar and pestle in liquid nitrogen to obtain a fine powder, and then suspended in four volumes of protein binding buffer (20 mM sodium phosphate, pH 5.8, 500 mM NaCl, 0.1% (w/v) Triton X-100) containing 1 mM phenylmethyl sulfonyl fluoride (PMSF), incubated for 30 min at 4◦C, and centrifuged for 30 min at 15,000 g at 4◦C. The proteins in the supernatant were used for protein analysis, and the protein concentration was determined according to the Bradford method using bovine serum albumin (0, 0.2, 0.4, 0.6, 0.8, 1.0 mg/mL) as the standard (Bradford, 1976).

#### Separation and Isolation of Cu-Binding Proteins Based on Cu-IMAC

Experimental design for proteomic analysis of Cu-binding proteins in rice radicles was shown in Supplementary Figure S1. The used method for separating and isolating Cu-binding proteins was based on the Cu-IMAC method of Song et al. (2014). Protein samples were pre-chromatographed on a column with IDA-Sepharose for removing metal ions from proteins samples before flowing over a Cu-IDA-Sepharose column (referred to as Cu-IMAC) for separating Cu-binding proteins. IDA-Sepharose and Cu-IMAC were connected in tandem with a tube (inner diameter of 0.5 mm). For the pre-chromatography column, IDA-Sepharose with a 2 mL bed volume for each column was poured into a 10-mL glass column with an inner diameter of 10 mm and washed with 10 mL of water at a rate of 0.5 mL/min. For the Cu-IMAC column, IDA-Sepharose with a 2 mL bed volume for each column was poured and washed with 10 mL of water at a rate of 0.5 mL/min, after which the bed volume of 0.2 M CuSO<sup>4</sup> was applied to the column, followed by washing with 15 mL distilled water to remove excess Cu ions at a rate of 0.5 mL/min. Columns were equilibrated with 10 bed volumes of binding buffer at a linear flow rate of 0.5 mL/min after they were connected in tandem.

A total of 20 mg of protein sample solution was loaded onto the pre-chromatography column at a linear flow rate of 0.25 mL/min, after which the column was washed with binding buffer until the ultraviolet absorbance of the effluent from the Cu-IMAC column reached baseline. Binding buffer containing 10 mM imidazole was added to the per-chromatography column at a linear flow rate of 0.5 mL/min until the UV-absorbance of the effluent from the Cu-IMAC column reached baseline, suggesting that non-specific proteins were removed. The Cu-binding proteins were eluted with elution buffer (10 mM sodium acetate, 500 mM NaCl, pH 5.5) containing 40 mM imidazole at a linear flow rate of 0.5 mL/min until the UV-absorbance of the effluent from the Cu-IMAC column reached baseline.

The Cu-binding proteins eluted from Cu-IMAC were precipitated with four volumes of cooled 10% (w/v) acetone (containing 0.07% (w/v) dithiothreitol, DTT) overnight at −20◦C for 1 h, followed by centrifugation for 15 min at 10,000 g at 4◦C. Afterward, the pellets were dissolved in lysis buffer (7 M urea, 2 M thiourea, 4% (w/v) CHAPS, 0.2% (w/v) Bio-lytes pH 3–10, 65 mM DTT). Protein concentrations were assayed using a Bio-Rad RC DC Protein Assay Kit 1.

### Two-Dimensional Electrophoresis (2-DE), Gel Scanning, and Image Analysis

For each replicate, 100µg of total protein extract was loaded onto IPG dry strips (17 cm, pH 4–7 linear gradient; Bio-Rad, Hercules, CA, USA) during the rehydration step (13 h), followed by focusing for a total of 60,000 V·h using a Protean IEF Cell (Bio-Rad). Following isoelectric focusing (IEF), the gel strips were equilibrated for 15 min in 5 mL equilibration buffer containing 0.375 M Tris–HCl (pH 8.8), 6 M urea, 20% (v/v) glycerol, 2% (w/v) sodium dodecyl sulfate (SDS), and 2% (w/v) DTT. The strips were then equilibrated in the same buffer as described above, but including 2.5% w/v iodacetamide instead of DTT. SDS-polyacrylamide gel electrophoresis (PAGE) in the second dimension was performed using 12% SDS-polyacrylamide gels sealed with 0.5% agarose. Electrophoresis was performed at 50 V for the first 30 min, followed by 150 V for 8 h using a Protean Plus Dodeca Cell apparatus (Bio-Rad). Protein spots were visualized using MS-compatible silver staining (Yan et al., 2000). To prevent the gels being overexposed, the developing course was divided into two steps: firstly, the developing solution was drained off after becoming yellow; secondly, the developing course was terminated when the small protein spots begin to become clear.

The gels were scanned using the image scanner UMAX Powerlook III (UMAX Technologies, Dallas, TX, USA) at 300 dpi resolution; image and data analyses of the gels were performed using PDQuest software (version 8.0; Bio-Rad) and a multivariate statistical package (DeCyder EDA, Unscrambler, Samespots), which can automatically deal with missing values during analysis (Valledor and Jorrín, 2011). The abundance of spot mean a summation of the pixel intensities localized within the defined spot area, which obtained by PDQuest (Bio-Rad) image analysis software. Spot quantity was normalized in the "total quantity of

valid spot" mode for possible staining differences between gels. Duplicate 2-DE gels were run for each treatment from three independent tissue extractions, only spots with significant and reproducible changes were considered to represent differentially accumulated proteins. The results for the control and Cuor H2O2-treated samples were analyzed for differences using Student's t-test with a significance level of 5%. Protein spots were selected for MS analysis when a difference of 1.5-fold or greater was observed in the level of accumulation between the treatment and control.

### In-Gel Digestion of Protein, MS Analysis, and Functional Classification

Protein spots were excised and destained (Gharahdaghi et al., 1999). The samples were incubated in 50 mM ammonium bicarbonate for 5 min, dehydrated with acetonitrile (ACN), and dried. The peptides were extracted with 60% ACN and 0.1% trifluoroacetic acid after the proteins were digested with trypsin, and were then extracted and desalted with ZipTip C18 columns (Millipore, Bedford, MA, USA). The peptide solution was saturated with α-cyano-4-hydroxycinnamic acid and then air-dried on an MS sample plate.

Peptide mass spectra were obtained using a 4700 Proteomics Analyzer MALDI-TOF/TOFTM mass spectrometer (Applied Biosystems, Framingham, MA, USA) in positive ion reflector mode. The subsequent MS/MS analysis was performed in a datadependent manner, and the 10 most abundant ions fulfilling certain preset criteria were subjected to high-energy collisional dissociation (CID) analysis. The collision energy was set to 1 keV, and nitrogen was used as the collision gas.

All protein spectra were submitted for database searching using the online MASCOT program (http://www.matrixscience. com) against NCBInr databases (http://www.ncbi.nlm.nih.gov/ protein). The taxonomic category selected was Oryza sativa. The searching parameters were as follows: 0.15 Da mass tolerance for peptides and 0.25 Da mass tolerance for TOF–TOF fragments, one allowed trypsin miscleavage, Cys carbamidomethylation as a fixed modification, and Met oxidation as a variable modification. Only significant hits, as defined by the MASCOT probability analysis (P < 0.05), were accepted.

Kyoto Encyclopedia of Genes and Genomes (KEGG) (http://www.genome.jp/kegg/ or http://www.kegg.jp/) was used to predict molecular function, biological processes, and significant pathways involved in response to stress.

#### Total RNA Isolation, cDNA Synthesis, and Quantitative RT-PCR

Total RNA was extracted using the RNA simple Total RNA Kit (LifeFeng, Shanghai, China) according to the manufacturer's instructions and then converted to cDNA after DNase I treatment using a PrimeScriptTM RT Master Mix (TaKaRa). Real-time quantitative RT-PCR was performed on a MyiQ Real-Time PCR Detection System (Bio-Rad Hercules, CA, USA) using SYBR Premix Ex Taq (TaKaRa). The primers for protein mRNA were listed in Supplementary Table S1. The PCR protocol included an initial 7-min incubation at 95◦C for complete denaturation followed by 40 cycles at 94◦C for 30 s, 60◦C for 30 s, and 72◦C for 30 s. The specificity of the PCR amplification was examined based on a heat dissociation curve (65–95◦C) following the final cycle. Normalized relative expression was calculated using the 2−11Ct (cycle threshold) method.

#### Statistical Analysis

Data were analyzed using SPSS ver. 16.0 (Statistical Package for Social Science for Windows, SPSS, Inc., Chicago, IL, USA). All values reported in this paper are means ± SE (n = 3) of three separate experiments. Means denoted by the same letter did not significantly differ at P < 0.05 according to Duncan's multiple range test.

### RESULTS

### Induction of Cu on H2O<sup>2</sup> Production in Radicles

The accumulation of H2O<sup>2</sup> in the radicles of germinating rice seeds was examined using histochemical DAB staining. Our results showed that exposure to excess Cu for 12 h caused an evident accumulation of H2O<sup>2</sup> in the radicles (**Figure 1A**). The production of Cu-induced H2O<sup>2</sup> could be decreased by infiltration with the H2O<sup>2</sup> scavenger, Asc. In the presence of 100µM Cu, H2O<sup>2</sup> accumulation in radicles gradually increased during the first 12 h of exposure and then decreased slightly but remained higher than the control (**Figure 1B**). The concentrations of H2O<sup>2</sup> assayed by spectrophotometry were consistent with the results of histochemical detection by DAB staining (**Figure 1C**).

### Identification of Cu-Binding Proteins Modulated by Cu and H2O<sup>2</sup>

UV detections to Cu-binding proteins of rice radicles via Cu-IMAC were shown in Supplementary Figure S2. The Cubinding protein yields from the control, H2O2-treated, 20µM Cu, and 100µM Cu-treated rice radicles, estimated with the percents of peak area (of total peak area), were not significantly different. Protein maps produced from 2-DE gels showed a high reproducibility among the three independent extractions (**Figure 2**, Supplementary Figure S3). When analyzed using PDQuest, 780 ± 15, 772 ± 25, 793 ± 13, and 695 ± 24 proteic spots were identified in the range of pH 4–7 and relative molecular masses of 10–120 kDa with the control, H2O2-treated, 20µM Cu and 100µM Cu-treated rice radicles, respectively. The significantly differential spot patterns between 100µM Cu treatment and the other treatments might be explained by differing degrees of protein loss resulting from the Cu-binding proteins eluted from the Cu-IMAC column. Among all of the spots, 656 spots were present with all four treatment, other protein spots were treatment-specific. 26 protein spots, exhibited more than 1.5-fold differences in the abundances under at least one treatment (20, 100µM Cu, or 10 mM H2O2) compared to the control. In order to observe more clearly, these protein spots were artificially divided into four regions (A, B, C and D) in gels, and the four regions were enlarged and shown in **Figure 3**.

Compared with the control, the H2O<sup>2</sup> treatment increased the abundances of 11 spots and decreased that of nine spots among the 26 protein spots. Among the 11 H2O2-increased protein spots, six protein spots were simultaneously increased under both Cu treatments (20 and 100µM Cu), and five protein spots were increased under 20µM Cu. Among nine protein spots decreased under H2O2, six protein spots were similarly decreased under both Cu treatment, one spot decreased under 20µM Cu treatment. In addition, the abundances of six protein spots only increased under Cu treatment and unaffected under H2O<sup>2</sup> treatment (Supplementary Figure S4).

The Cu-binding proteins in 26 spots were analyzed using MALDI-TOF/TOF MS, and all spectra of proteins were submitted to a NCBInr protein database search using the online MASCOT program. Supplementary Table S2 shows the identity of Cu-binding proteins in 26 spot after a database search. The molecular masses (Mr) and isoelectric point (pI) values of each identified protein are listed in **Table 1**. These identified Cubinding proteins were found to be involved in different cellular

FIGURE 2 | Representative 2-DE maps of copper-binding proteins obtained from radicles of germinating rice seeds treated with Cu and H2O2 via Cu-IMAC plus IDA-Sepharose pre-chromatography. Germinating rice seeds were treated with control (deionized water without Cu and H2O2), 10 mM H2O2 for 6 h, 20 and 100µM Cu for 12 h. A 20 mg proteins extracts from radicles of germinating rice seeds was loaded onto the column with IDA-sepharose to removal metal ions in protein samples before onto Cu-IMAC. These Cu-binding proteins eluted from a Cu-IMAC column were subjected to 2-DE separation. One-hundred microgram of total protein were loaded onto IPG dry strips (17 cm, pH 4–7 linear gradient), the second dimension was carried out using 12% SDS-PAGE. The protein spots were visualized by mass spectrometry compatible silver staining.

FIGURE 3 | Enlargements of the framed areas (A), (B), (C), and (D) shown in Figure 2 and the relative abundance of differentially accumulated proteins. Arrows indicate the differentially expressed proteins in response to Cu and H2O2 stress. The vertical axis (abundance) mean a summation of the pixel intensities localized within the defined spot area, which obtained by PDQuest (Bio-Rad) image analysis software.

responses and metabolic processes, including antioxidative defense (6 proteins), stress response and detoxification (4 proteins), protein synthesis (5 proteins), protein modification (1 protein in 2 spots), protein metabolism (2 protein in 3 spots), carbohydrate metabolism (3 proteins), nucleotide metabolism (1 protein), and secondary metabolite metabolism (2 proteins).

### Analyses of Metal-Binding Motifs

In this study, among 24 proteins in 26 spots identified, 18 proteins contained one or more of nine metal-binding motifs reported by Smith et al. (2004), and 20 protein species contained one or more of the top six motifs (H–(X)5–H, H–(X)7–H, H– (X)12–H, H–(X)6–M, M–(X)7–H, and H–(X)3–C) reported by Kung et al. (2006) in Arabidopsis roots (**Table 2**). Fifteen proteins contained motifs reported by Smith et al. (2004) and the top six motifs by Kung et al. (2006). However, one protein (spots 18 and 19; putative legumin) contained neither the motifs reported by Smith et al. (2004) nor the top six motifs reported by Kung et al. (2006).

### Transcriptional Analysis of Genes for Some Cu-Binding Proteins

In order to assess the correlation between mRNA expression and protein accumulation, Real-time quantitative RT-PCR was applied to four mRNAs of identified Cu-binding proteins, copper/zinc superoxide dismutase (CuZn-SOD, spot 16), L-ascorbate peroxidase (APX, spot 8), peroxiredoxin (Prx, spot 15), and glutathione S-transferase 2 (GST2, spot 6), involved in antioxidative defense (**Figure 4**). By compared with the corresponding spots of **Figure 3**, the expression analyses of four genes were consistent with the proteins accumulation except for CuZn-SOD mRNA change under high Cu treatment, indicate that the accumulation of these proteins have been largely regulated at the transcriptional level.

## DISCUSSION

### Cu-Induced Accumulation of H2O<sup>2</sup> and Cu-Binding Proteins

Numerous studies have shown that excess Cu can induce the formation of ROS (including H2O2) and cause oxidative stress. In this study, the formation of H2O<sup>2</sup> was observed with increasing Cu concentrations and with Cu treatment time in the Cu-treated radicles of rice (**Figure 1**). Accumulation of H2O<sup>2</sup> has also been observed in other Cu-exposed plant species using histochemical staining (Tewari et al., 2006; Sgherri et al., 2007; Zhang et al., 2008, 2010). Because H2O<sup>2</sup> is relatively stable and diffusible through membranes, it is known to modulate gene expression and participate in various physiological processes (Neill et al., 2002; Ahmad et al., 2008). So far, no proteins simultaneously response to H2O<sup>2</sup> and Cu stress were reported by searching web of science.

In the present study, 24 Cu-IMAC-binding proteins in 26 spots were identified that were differentially accumulated at least by one treatment (20, 100µM Cu, or 10 mM H2O2). The same protein (e.g., protein disulfide isomerase and putative legumin) in varied spots is possible since the spot change its position in the gel due to changes in pI or Mr as a

#### TABLE 1 | Differentially accumulated Cu-binding proteins of rice radicles identified by MS/MS.


<sup>a</sup>Spot abundance is accumulated as the ratio of intensities of proteins between stress and control. Fold changes indicate a statistically significant difference (P < 0.05) between treated samples and control samples by Duncan's test; ↑, up-regulated; ↓, down-regulated (alone↓, disappearance of spot); –, no change. H2O2, 20 Cu, and 100 Cu represent 100µM H2O2, 20µM Cu and 100µM Cu treatment concentrations, respectively.

<sup>b</sup>PM, number of peptides matched.

<sup>c</sup>SC, sequence coverage by MS/MS.

consequence of post-translational modifications. Among these identified proteins, elongation factor EF-2 (EF-2), GST, Prx, APX, quinone-oxidoreductase QR2 (QR2), protein disulfide isomerase (PDI), glyceraldehyde 3-phosphate dehydrogenase (GAPDH), triosephosphate isomerase (TPI), NADPH-dependent mannose 6-phosphate reductase (M6PR), and cytidine/deoxycytidine deaminase-like (CDC) have been identified as Cu-IMAC-binding proteins in Arabidopsis (Kung et al., 2006; Tan et al., 2010), soybean (Wang et al., 2014), microalgae (Smith et al., 2014), and rice (Song et al., 2014). Moreover, the analogs of eukaryotic

#### TABLE 2 | Potential Cu-binding motifs of identified proteins.


<sup>a</sup>Motifs that were reported by Smith et al. (2004).

<sup>b</sup>Motifs that were reported by Kung et al. (2006). "–"indicates not present; H, C, and M indicates respectively the three amino acids His, Cys and Met, X indicates any one of 20 amino acids.

translation initiation factor 5A (eIF5A), transcription factor (TF), and SOD, such as translation initiation factor Tu, G, 3A, and 4A, iron-dependent transcriptional regulator, and Fe-SOD were identified as Cu-IMAC-binding proteins in hepatoma cells (Smith et al., 2004), in Arabidopsis (Kung et al., 2006) and S. pneumoniae (Sun et al., 2011). SODs are metalloenzymes found in three different molecular forms containing Cu and Zn (CuZn-SOD), Mn (Mn-SOD), or Fe (Fe-SOD) as prosthetic metals. However, to the best of our knowledge, the other eight proteins, including pathogenesis-related protein (PR) known as PR-10a, and Bet v I family protein (PR-b), cytochrome P450-like protein (CYP-L), salt stress-induced protein (SSI), legumin, cathepsin Blike cysteine protease (CBCP), arginine decarboxylase (ADC), and chalcone-flavonone isomerase (CHI), have not been reported as Cu-IMAC-binding proteins in plants. The present data showed that the abundance of 11 spots increased under exogenous H2O<sup>2</sup> treatment out of 17 spots increased under 20µM Cu, and that all 20µM Cu-decreased spots decreased under H2O<sup>2</sup> (**Table 1,** Supplementary Figure S3), which may be ascribed to the same amount H2O<sup>2</sup> produced by low concentration Cu as H2O<sup>2</sup> treatment. However, H2O<sup>2</sup> or Cu at high levels can cause oxidative stress and cell damage, which could be the reason that the abundance of five protein spots are decreased under 100µM Cu stress but increase under H2O<sup>2</sup> and 20µM Cu stress.

#### Cu-Binding Proteins Simultaneously Accumulated under Low Cu and H2O<sup>2</sup> Treatments

Four identified Cu-binding proteins such as CuZn-SOD (spot 16), QR2 (spot 13), Prx (spot 15), and APX (spot 8), displayed similar behavior under 20 µM Cu and H2O<sup>2</sup> treatments, which may play important roles in plant antioxidant defense responses. SODs are key players in the antioxidant defense system through the dismutation of O•− 2 to H2O2. SODs as metal chelators may also regulate the intracellular Cu level. In plants, quinones are redox-active compounds that oxidize the thiol groups of proteins and GSH. QR2 catalyzes two electron reductions of quinones to hydroquinones (Malakshah et al., 2007; Vannini et al., 2012). Prx, which consists of many different thiol-disulfide exchange proteins, such as thioredoxins and glutaredoxins, is an H2O2-scavenging enzyme that reduces H2O<sup>2</sup> to H2O, and Prx possesses a highly reactive Cys that is oxidized to form a disulfide bond coupled with the reduction in H2O<sup>2</sup> (Dietz et al., 2006). Thus, increase of these Cu-binding proteins may alleviate Cu-induced damage by decreasing free Cu2<sup>+</sup> activity in the cytoplasm and/or scavenging ROS.

Notably, the abundance changes of specific Cu-binding proteins responded differentially to excess Cu: the abundances

of Prxs and QR2 increased, while that of APX decreased. The abundance of CuZn-SOD increased under 20µM Cu but decreased under higher levels of Cu (100µM). These different changes of antioxidative enzymes following exposure to excess Cu may be due to their varied functions. Excess Cu increased SOD expression (Sunkar et al., 2006; Cohu et al., 2009; Zhang et al., 2010) and activities (Tewari et al., 2006; Zhang et al., 2008, 2010). However, Cu ions can be dangerous to cellular compartments as free ions. Thus, high Cu treatment (100µM Cu) decreased the abundance of CuZn-SOD in this study. APX and CAT are two major scavengers of H2O2. APX is present throughout the cell and has a higher substrate affinity in the presence of Asc as a reductant. Cu and H2O<sup>2</sup> have been reported to increase APX expression and activity (Lee et al., 2007). In this study, the abundance of APX decreased under excess Cu and exogenous H2O<sup>2</sup> treatments. Decrease of APX could lead to the accumulation of H2O<sup>2</sup> and enhance oxidative stress. Similar decrease of APX was observed in rice leaves (Wan and Liu, 2008) and tobacco cells (Vannini et al., 2012) in response to H2O<sup>2</sup> stress at high doses (50 mM) or over extended times (6 h). H2O<sup>2</sup> was suggested to directly inhibit APX activity by causing protein oxidation at concentrations over a threshold value (de Pinto et al., 2006). The abundance changes of GST (spots 6 and 14) responded differentially to excess Cu and exogenous H2O<sup>2</sup> treatment. A major function of GSTs is to detoxify a variety of hydrophobic and electrophilic compounds by catalyzing their conjugation with GSH (Jwa et al., 2006). Consistent with our results, an increased GSTs was detected in Cu-treated (Song et al., 2013, 2014) and H2O2-treated rice (Wan and Liu, 2008). In contrast, a decrease in GST levels in rice exposed to Cu (Ahsan et al., 2007), H2O<sup>2</sup> (Vannini et al., 2012), and selenium (Se) (Wang et al., 2012) has been observed.

The gene products of four identified antioxidant proteins (CuZn-SOD, APX, Prx and GST2) showed similar changes obtained from proteomics experiments except for CuZn-SOD change under high Cu treatments (**Figure 4**). Previous studies showed that Cu availability is the major factor that determines whether Fe-SOD or CuZn-SOD are expressed (Cohu et al., 2009), the CuZn-SOD accumulation is mediated by a microRNA, miR398, which targets CuZn-SOD mRNA for degradation under some condition (Sunkar et al., 2006), and CuZn-SOD proteins accumulated only when Cu ions were available for final assembly and stability. Thus, it is possible that the abundance decrease of CuZn-SOD protein spot under high Cu treatment was not consistent with the results of gene expression analyses.

Three proteins including PR-b, CYP-L, and SSI out of four Cu-binding proteins involved in the stress response and detoxification displayed similar behavior under low concentration Cu and H2O<sup>2</sup> treatment. In plants, CYP proteins are involved in the synthesis of fatty acids, lignin, hormones, and flavonoids, as well as xenobiotic metabolism in higher plants (Schuler and Werck-Reichhart, 2003). In this study, the abundance of one CYP-L (spot 17) increased slightly under 20µM Cu and H2O<sup>2</sup> but significantly decreased under 100µM Cu treatment. Li et al. (2009) observed an increase of the CYP-like protein in soybean at 2 h post inoculation. In contrast, decrease of CYP proteins was observed in Cu-treated rice germinating embryos (Zhang et al., 2009) and a Cd-treated Phytolacca americana leaf (Zhao et al., 2011). Wan and Liu (2008) observed that one putative salt-induced protein increased under H2O<sup>2</sup> in rice leaves. In this study, Cu and H2O<sup>2</sup> treatments significantly increased the abundance of SSI (spot 23), but its function in Cu-stressed plants remains unknown. PR proteins play a role in a wide range of cell functions, including cell wall rigidification, signal transduction, and antimicrobial activity (Markovic-Housley et al., 2003). Elevated levels of ROS have been reported to induce PR proteins in rice (Jwa et al., 2006). The increase of PR proteins has also been observed in Cu-treated Phaseolus vulgaris (Cuypers et al., 2005), Elsholtzia splendens (Li et al., 2009), and rice (Zhang et al., 2009). In this study, treatment with H2O<sup>2</sup> and Cu decreased or did not affect the abundance of PR-10a (spot 1) and PR-b (spot 2), although 100µM Cu increased that of PR-10a. The opposite change patterns of PR proteins by excess Cu suggest that they have different roles.

All Cu-binding proteins involved in protein synthesis displayed similar behavior under low concentration Cu and H2O<sup>2</sup> treatments (**Table 1**). The abundances of susceptibility homeodomain transcription factor (SHTF, spot 3), EF-2, and (spot 21) increased under excess Cu and H2O2, excluding the high Cu treatment that did not affect that of SHTF and decreased that of legumin. In contrast, the abundances of eIF5A (spot 11) and eIF5A-2 (spots 9 and 10) decreased under both Cu and H2O2. eIF5A was also thought to play a role in translation elongation (Saini et al., 2009) and other aspects of RNA metabolism such as RNA export (Liu et al., 2008). The expression of eIF5A in plants usually increases in response to abiotic stress (Li et al., 2009; Xu et al., 2011; Meng et al., 2014; Parkash et al., 2014). In agreement with our result, a significant decrease of eIF5A was observed in rice after a longterm salt stress (Parker et al., 2006), which may be associated with premature senescence. EFs (EF1A, EF1B, and EF-2) are fundamental regulatory proteins of the translational elongation step in higher plants, as well as other eukaryotic organisms. EF-2 catalyzes GTP-dependent translocation of peptidyl-tRNA from the A site to the P site of the ribosome during peptide chain elongation (Browne and Proud, 2002). In this study, the abundances of EF-2 increased by 30.0-, 11.4-, and 6.1-fold in the presence of H2O2, low and high concentration Cu, respectively. Similar increase of EF-2 was observed in Schizosaccharomyces pombe in response to H2O<sup>2</sup> stress (Weeks et al., 2006). In contrast, decrease of the EF-2 protein was observed in Cu-treated E. splendens roots (Li et al., 2009), Cd-treated P. americana (Zhao et al., 2011), and B-deficient Brassica napus (Wang et al., 2010).

Legumin is a major storage protein in plant seeds, including α and ß basic polypeptides of 40 and 20 kDa, respectively, bound by a disulfide bridge (Sabir et al., 1973). This protein contained neither the motifs reported by Smith et al. (2004) nor the top six motifs reported by Kung et al. (2006), but contained 8 of 117 potential metal-binding motifs (C-(X)n-C, C-(X)n-H, C-(X)n-M, H-(X)n-C, H-(X)n-H, H-(X)n-M, M-(X)n-C, M-(X)n-H, and M-(X)n-M, where n = 0–12) reported by Kung et al. (2006). Cu caused a reduction in the germination rate of bean, which increased the level of storage proteins compared to the control (Karmous et al., 2011). In this study, the abundances of legumin (spots 18 and 19) increased under excess Cu and H2O2. It is unknown whether increase of legumin protein abundance may alleviate Cu-induced damage by decreasing free Cu2<sup>+</sup> activity in the cytoplasm or be a consequence of Cu toxicity and oxidative stress.

Two enzymes (GAPDH and M6PR) involved in carbohydrate metabolism and two enzymes (ADC and CHI) involved in secondary metabolism displayed similar behavior under Cu and H2O<sup>2</sup> treatments. Treatments with both Cu and H2O<sup>2</sup> decreased the abundance of GAPDH (spot 12) and increased M6PR (spot 20). Arabidopsis Cytosolic GAPDH may be a potential target of H2O2-dependent oxidation in plant protein extractions (Hancock et al., 2005). TPI and GAPDH are important enzymes in the glycolytic pathway. Here, the decrease of GAPDH and increase of TPI may favor the accumulation of glyceraldehyde 3-phosphate under stress conditions. M6PR, a key enzyme in mannitol biosynthesis, catalyzes the conversion of mannose 6 phosphate into mannitol 1-phosphate. Overexpression of M6PR genes from celery and Arabidopsis result in increased tolerance to salt stress (Sickler et al., 2007; Chan et al., 2011). Treatments with Cu and H2O<sup>2</sup> increased abundance of ADC (spot 4) and decreased that of CHI (spot 5). Cu-induced increases in ADC activity were also observed in previous reports (Groppa et al., 2008; Xu et al., 2011). The roles of ADC and CHI as Cu-binding proteins in the tolerance to Cu and oxidative stresses are still unknown.

### Cu-Binding Proteins Accumulated under Cu not H2O<sup>2</sup> Treatment

Whereas five identified Cu-binding proteins including GST (spot 6), PDI (spots 25 and 26), CBCP (spot 24), TPI (spot 7), and CDC (spot 22) were increased only under low and high Cu but unaffected under H2O<sup>2</sup> stress, which hint Cu ions can regulate the Cu-binding proteins accumulation by no H2O<sup>2</sup> pathway. PDI has been identified as a Cu-binding protein in previous reports (Smith et al., 2004; Song et al., 2014). PDI is a thioredoxin superfamily oxidoreductase from the endoplasmic reticulum, and catalyzes a wide range of thioldisulfide exchange reactions, including oxidation, reduction, and

isomerization, and also displays chaperone, calcium-binding, and Cu-binding activity (Hatahet and Ruddock, 2009; Laurindo et al., 2012). Overexpression of PDI gene from Methanothermobacter thermoautotrophicum enhances mercury tolerance in transgenic rice (Chen et al., 2012). Proteomic analyses showed that PDI accumulation increased in rice roots in the presence of Cu (Song et al., 2014), in rice leaves in the presence of H2O<sup>2</sup> (Wan and Liu, 2008), and in soybean leaves during salt stress (Ma et al., 2012). Here, the abundance of PDIs was markedly higher in Cu-treated rice, which can enhance Cu tolerance in germinating rice seed by binding Cu and thiol-disulfide exchange reactions. In contrast, PDIs accumulation were down-regulated by Cu stress in roots of the tolerant plant E. splendens (Li et al., 2009), by H2O<sup>2</sup> in rice root apoplasts (Zhou et al., 2011), and by flooding stress in soybean roots (Khatoon et al., 2012).

### CONCLUSIONS

The present results revealed that 17 out of 24 identified Cu-binding proteins have a similar response to 20µM Cu and H2O<sup>2</sup> stress in rice radicles. These Cu-binding proteins involved in antioxidative defense, stress response, and detoxification, protein synthesis and metabolism, and can play important roles on reconstructing homeostasis of cell under stress condition by H2O<sup>2</sup> signal pathway. The accumulation of five identified Cubinding proteins were up-regulated by 20 and 100µM Cu but unaffected by H2O2, which hint Cu ions can regulate Cu-binding proteins accumulation by no H2O<sup>2</sup> pathway to cope with excess Cu in cell. A putative model of Cu-binding proteins in rice radicles to Cu and H2O<sup>2</sup> stress responses was shown in **Figure 5**. Further studies are required to clarify the roles of Cu ions in these putative Cu-binding proteins in plant cells to determine if they are passive molecular targets of metal ions or active participants in metal tolerance.

#### AUTHOR CONTRIBUTIONS

ZS and HZ designed research. HZ, YX, YS, and KZ conducted sampling, biochemical and data analysis. YS, CC, and KZ contributed with proteomic analysis. HZ, CC, and ZS wrote the manuscript. All authors read, reviewed and approved the manuscript.

#### ACKNOWLEDGMENTS

This research project was supported by Specialized Research Fund for the Doctoral Program of Higher Education

#### REFERENCES


(20120097130004), the National Natural Science Foundation of China (31471938), Key Scientific Research Projects of Education Department of Henan Province (16A180004), Innovation Team Foundation of Henan University of Science and Technology (2015TTD002), and the Development Program to National Natural Science Foundation of China of Henan University of Science and Technology (13000786).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 01216

chloroplast and cytosolic copper/zinc-superoxide dismutases: regulation and unexpected phenotypes in an Arabidopsis mutant. Mol. Plant 2, 1336–1350. doi: 10.1093/mp/ssp084


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Zhang, Xia, Chen, Zhuang, Song and Shen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Proteomic Profiling of the Interactions of Cd/Zn in the Roots of Dwarf Polish Wheat (Triticum polonicum L.)

Yi Wang1 †, Xiaolu Wang<sup>1</sup> † , Chao Wang1 †, Ruijiao Wang<sup>1</sup> , Fan Peng<sup>1</sup> , Xue Xiao<sup>1</sup> , Jian Zeng<sup>2</sup> , Xing Fan<sup>1</sup> , Houyang Kang<sup>1</sup> , Lina Sha<sup>1</sup> , Haiqin Zhang<sup>1</sup> and Yonghong Zhou<sup>1</sup> \*

*<sup>1</sup> Triticeae Research Institute, Sichuan Agricultural University, Sichuan, China, <sup>2</sup> College of Resources, Sichuan Agricultural University, Sichuan, China*

#### Edited by:

*Hanjo A. Hellmann, Washington State University, USA*

#### Reviewed by:

*Ramesh Katam, Florida A&M University, USA Chiew Foan Chin, University of Nottingham Malaysia Campus, Malaysia*

> \*Correspondence: *Yonghong Zhou Zhouyh@sicau.edu.cn*

*† These authors have contributed equally to this work.*

#### Specialty section:

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

Received: *16 May 2016* Accepted: *30 August 2016* Published: *14 September 2016*

#### Citation:

*Wang Y, Wang X, Wang C, Wang R, Peng F, Xiao X, Zeng J, Fan X, Kang H, Sha L, Zhang H and Zhou Y (2016) Proteomic Profiling of the Interactions of Cd/Zn in the Roots of Dwarf Polish Wheat (Triticum polonicum L.). Front. Plant Sci. 7:1378. doi: 10.3389/fpls.2016.01378* Cd and Zn have been shown to interact antagonistically or synergistically in various plants. In the present study of dwarf polish wheat (DPW)roots, Cd uptake was inhibited by Zn, and Zn uptake was inhibited by Cd, suggesting that Cd and Zn interact antagonistically in this plant. A study of proteomic changes showed that Cd, Zn, and Cd+Zn stresses altered the expression of 206, 303, and 190 proteins respectively. Among these, 53 proteins were altered significantly in response to all these stresses (Cd, Zn, and Cd+Zn), whereas 58, 131, and 47 proteins were altered in response to individual stresses (Cd, Zn, and Cd+Zn, respectively). Sixty-one differentially expressed proteins (DEPs) were induced in response to both Cd and Zn stresses; 33 proteins were induced in response to both Cd and Cd+Zn stresses; and 57 proteins were induced in response to both Zn and Cd+Zn stresses. These results indicate that Cd and Zn induce differential molecular responses, which result in differing interactions of Cd/Zn. A number of proteins that mainly participate in oxidation-reduction and GSH, SAM, and sucrose metabolisms were induced in response to Cd stress, but not Cd+Zn stress. This result indicates that these proteins participate in Zn inhibition of Cd uptake and ultimately cause Zn detoxification of Cd. Meanwhile, a number of proteins that mainly participate in sucrose and organic acid metabolisms and oxidation-reduction were induced in response to Zn stress but not Cd+Zn stress. This result indicates that these proteins participate in Cd inhibition of Zn uptake and ultimately cause the Cd detoxification of Zn. Other proteins induced in response to Cd, Zn, or Cd+Zn stress, participate in ribosome biogenesis, DNA metabolism, and protein folding/modification and may also participate in the differential defense mechanisms.

Keywords: dwarf polish wheat, iTRAQ, cadmium, zinc, interaction, proteomic

### INTRODUCTION

Environmental toxicity from non-essential heavy metals such as cadmium (Cd), which is released from human activities and other environmental causes, is rapidly increasing (Ahsan et al., 2009). In humans, Cd causes diseases such as osteoporosis and emphysema by damaging the lungs, kidneys, and bones (Kazantzis, 2004; Straif et al., 2009). In plants, Cd damages the photosynthetic apparatus, interrupts respiratory and nitrogen metabolism, and unbalances water and nutrient uptake (Herbette et al., 2006; Balen et al., 2011), ultimately reducing biomass, causing leaf chlorosis, inhibiting root growth, and even leading to plant death (Lin et al., 2007; Yadav, 2010; Lin and Arats, 2012). Additionally, plants can accumulate high Cd contents in their edible parts, which poses a potentially major hazard to human health (Satarug et al., 2003).

Zinc (Zn), an essential metal and a cofactor of numerous plant proteins and enzymes, plays several crucial roles in protein binding, enzyme activity, transcriptional and translational regulation, and signal transduction (Broadley et al., 2007; Lin and Arats, 2012). However, excess Zn can also cause toxicity, as it can damage DNA replication and disrupt enzyme activities and protein folding and function, ultimately inducing chlorosis and inhibiting plant growth and development (Broadley et al., 2007; Lin and Arats, 2012; Schneider et al., 2013).

Cd and Zn coexist naturally in the soil. Due to their physical and chemical similarities (Chesworth, 1991), their uptake and transport in plants use similar pathways (Grant et al., 1998). Many metal transporters that transport both Cd and Zn have been identified, including AtNRAMP3 and AtNRAMP4 (Thomine et al., 2000; Lanquar et al., 2010). In response to Cd and Zn stresses, plants have developed strategies to prevent Cd-induced damage and maintain Zn homeostasis. Therefore, researchers have investigated the various synergistic and/or antagonistic interactions of Cd/Zn and found these interactions to depend on species, external bioavailable metal concentration, tissue type, and developmental stage. For example, some durum and bread wheat show antagonistic interactions of Cd/Zn in which Cd uptake is inhibited by Zn and Zn uptake is inhibited by Cd in roots, stems, and leaves (Hart et al., 2002, 2005; Sun et al., 2005). Conversely, some wheat under field conditions has shown synergistic interactions in which Cd and Zn uptake are promoted by each other (Nan et al., 2002).

However, previous studies on Cd/Zn interactions mainly focused on their transport and biochemical responses (Hart et al., 2002, 2005; Nan et al., 2002; Hassan et al., 2005; Sun et al., 2005). Although proteomic changes in response to Cd or Zn have been successfully investigated using a proteomics approach (Kieffer et al., 2008, 2009; Ahsan et al., 2009; Fukao et al., 2011; Lin and Arats, 2012; Schneider et al., 2013), the molecular mechanisms of Cd/Zn interactions are unknown, which limits our understanding of the interactions of Cd/Zn. Polish wheat (2n = 4x = 28, AABB, Triticum polonicum L.), which has low genetic similarity with T. aestivum (Wang et al., 2013; Michalcová et al., 2014), accumulates high concentrations of Zn, Fe, and Cu and therefore has attracted the interest of producers and breeders (Wiwart et al., 2013). Meanwhile, dwarf polish wheat (DPW), collected from Tulufan, Xingjiang, China, shows high tolerance to Cd and Zn because its growth is not affected by the accumulation of high concentrations of these metals in seedlings (Wang X. et al., in press). However, molecular responses to Cd and Zn remain unknown. Since DPW accumulates high concentrations of Cd and Zn in seedlings, it is a useful system for studying Cd/Zn interactions.

The purposes of this study are to understand molecular responses to Cd and Zn stresses, to investigate Cd/Zn interactions in DPW seedlings and to understand the molecular mechanisms of Cd/Zn interactions in DPW roots using isobaric tags for relative and absolute quantification (iTRAQ). iTRAQ is a highthroughput proteomic technology (Karp et al., 2010) that has been successfully used to reveal plant responses to heavy metals (Ahsan et al., 2009; Fukao et al., 2011).

### MATERIALS AND METHODS

#### Plant Material and Growth Conditions

DPW seeds were sterilized with 1% NaOCl and germinated in the dark for 5 days. The seedlings were cultured in full Hoagland nutrient solution in a growth chamber at 25◦C with a 16 hlight/8 h-dark cycle. At the two leaf stage, the seedlings were treated with null (CK), 40µM CdSO<sup>4</sup> (Cd), 800µM ZnCl<sup>2</sup> (Zn), or 40µM CdSO<sup>4</sup> + 800µM ZnCl<sup>2</sup> (Cd+Zn). Two days after treatments, the roots (three biological replications, each replication including 15 plants) were washed with 0.1µM EDTA and ddH2O, snap frozen in liquid nitrogen and stored at <sup>−</sup>80◦<sup>C</sup> for iTRAQ analysis. Other roots and leaves were dried for 2 days at 80◦C for measuring metal concentrations.

### Measurement of Cd and Zn Concentrations

Cd and Zn concentrations were measured as described by Wang et al. (2014). Briefly, the dried roots and leaves were ground to particle powders. Then, 0.2 g of powder was digested using concentrated sulfuric acid and hydrogen peroxide at 320◦C and then diluted to 50 ml. Metal concentrations were then determined using an atomic absorption spectrometer, Analyst 400 (PerkinElmer, CT, USA). Standard solutions of Cd and Zn were purchased from Fisher Scientific Ltd. (China). All data and figures were analyzed (t-test was conducted for the statistical analysis) and drawn using Sigmaplot 12.0.

### Total Protein Extraction

Roots (two randomly selected biological replications) with 0.1 mg of polyvinylpyrrolidone (PVPP) were ground into powders using liquid nitrogen and then homogenized in Tris-phenol (pH 8.0) and protein extraction buffer (0.7 M sucrose, 0.1 M KCl, 50 mM EDTA, 0.5 M Tris, pH 7.5, 2% β-mercaptoethanol, and 1 mM PMSF). After centrifuging for 20 min at 6000 rpm, the supernatants were collected and re-purified using protein extraction buffer. Proteins were precipitated using ammonium acetate methanol and then washed with methanol and acetone. Finally, protein samples were diluted using RIPA reagent, and protein concentrations were measured using a BCA Assay Kit (Biotech).

#### iTRAQ Labeling and LC-MS Analysis

iTRAQ labeling was performed according to Wu et al. (2013) with modifications. Briefly, 200µg of protein from each sample (two biological replications) was reduced, alkylated and then subjected to tryptic hydrolysis. iTRAQ labeling was performed using an iTRAQ <sup>R</sup> reagents-8plex Kit (Applied Biosystems). Peptides of CK, Cd, Zn, and Cd+Zn samples were labeled singly with the iTRAQ reporters 113, 114, 115, and 116, respectively. LC-MS (TripleTOF5600, Applied Biosystems) analysis was performed as described by Wu et al. (2013).

#### Protein Identification and Quantification

Protein identification and relative quantification were also performed according to Wu et al. (2013). Protein Pilot software v. 4.0 (Applied Biosystems) was used to convert the raw data (.wiff) into peak lists (.mgf). Each MS/MS spectrum was searched against the protein database Uniprot-147389. The search parameters were as follows: Paragon method: iTRAQ-8plex, Cys alkylation: MMTS, Digestion: Trypsin, Instrument: TripleTOF 5600, ID focus: Biological modifications and Amino acids substitutions, Detected Protein Threshold [Unused ProtScore (Confidence)]: ≥ 1.3, Competitor Error Margin (ProtScore): 2.0, and No. Distinct Peptides (Confidence): ≥ 95%. The tolerances were specified as ± 0.05 Da for peptides and ± 0.05 Da for MS/MS fragments. The relative abundance (fold-change ratios of differential abundance between labeled samples), P-value, error factor, lower confidence interval and upper confidence interval were calculated using the ProteinPilot software. Proteins containing at least two distinct peptides and fold change ratios ≥ 1.5 or ≤ 0.67 were considered as more abundant or less abundant proteins, respectively.

#### RESULTS

#### Metal Concentrations

No Cd was detected in CK (control) samples (**Figure 1A**). Two days after treatment, the Cd concentration in roots treated with Cd (752.55 ± 6.51 mg/Kg) was significantly higher (P < 0.01) than that in roots treated with Cd+Zn (76.75 ± 3.312 mg/Kg; **Figure 1A**). Meanwhile, the Cd concentration in leaves under Cd stress (40.87 ± 3.69 mg/Kg) was also significantly higher (P < 0.01) than that in leaves under Cd+Zn stress (9.20 ± 1.24 mg/Kg; **Figure 1A**). These results indicate that Zn inhibits Cd uptake in roots as well as its transport from roots to shoots.

Zn concentrations in the roots were always higher than those in the leaves (**Figure 1B**). Zn concentrations in leaves were similar between Zn (139.26 ± 32.12 mg/Kg) and Cd+Zn (147.00 ± 20.15 mg/Kg) stresses (**Figure 1B**). In roots treated with Zn (675.36 ± 41.67 mg/Kg), the Zn concentration was significantly higher (P < 0.01) than that in roots treated with Cd+Zn (557.63 ± 26.30 mg/Kg; **Figure 1B**). These results suggest that Cd inhibits Zn uptake in roots but does not affect its transport from root to shoot.

#### A Total of 432 Proteins Were Altered by Cd, Zn, or Cd+Zn Stresses

A total of 960 proteins with one or more distinct peptides and an Unused ProtScore ≥ 1.3 (with a peptide confidence ≥ 95%) (**Data Sheet 1**) were identified from the protein database Uniprot-147389. Compared with null, the expression levels of 206 (**Data Sheet 2**), 303 (**Data Sheet 3**), and 190 (**Data Sheet 4**) proteins were altered by Cd, Zn, and Cd+Zn stresses, respectively. Further analysis indicated that these proteins could be grouped into seven sub-groups (**Table 1**, **Figure 2**).

#### 53 Proteins Were Altered by All Three Stresses (Cd, Zn, and Cd+Zn)

The relative abundances of 53 proteins were altered significantly by all three stresses (Cd, Zn, and Cd+Zn; **Figure 2**, **Data Sheet 5**). Among these, 13 noteworthy proteins participated in either sucrose metabolism (5 proteins), glutathione (GSH) metabolism (5 proteins), or the oxidation-reduction process (3 proteins; **Table 1**). However, the relative abundances of other proteins were differentially altered by Cd, Zn, and combined Cd+Zn stresses (**Table 1**). For example, the relative abundance of glucose-6-phosphate isomerase (protein 554) was increased by Cd stress but was decreased by both Zn and Cd+Zn stresses. Contrary results were observed for lactoyglutathione lyase (protein 424), as its relative abundance was decreased by Cd stress but was increased by both Zn and Cd+Zn stresses. Further, the relative abundance of cytochrome c oxidase subunit 6B (protein 616) was increased by both Cd and Cd+Zn stresses but was decreased by Zn stress (**Table 1**). Thus, our analysis revealed differential molecular responses to Cd, Zn, and Cd+Zn stresses.

#### TABLE 1 | Some noteworthy proteins identified in differentially interactive groups.




*a represents protein identified number, <sup>b</sup> represents score, and <sup>c</sup> represents number of identified peptides.*

#### 58 Proteins Were Induced only in Response to Cd Stress

We identified 58 proteins whose relative abundances were induced only in response to Cd stress (**Figure 2**, **Data Sheet 6**). Of these, the relative abundances of 23 proteins were increased, and those of 35 proteins were decreased (**Figure 2**, **Data Sheet 6**). These proteins were not induced by either Zn or combined Cd+Zn stress, which suggests that they might participate in Zn inhibition of Cd uptake and transport. Among the 58 proteins we identified were 18 noteworthy proteins that participated in the oxidation-reduction process (4 down and 4 up), GSH metabolism (2 up and 1 down), sucrose metabolism (2 up), calcium (Ca) metabolism (3 up), or S-adenosyl-L-methionine (SAM) metabolism (1 down and 1 up; **Table 1**).

#### 131 Proteins Were Induced only in Response to Zn Stress

The relative abundances of 131 proteins were induced only in response to Zn stress (46 up and 85 down; **Figure 2**, **Data Sheet 7**). That these proteins were not induced in response to either Cd or combined Cd+Zn stress suggests that they might participate in Cd inhibition of Zn uptake. Among the 131 proteins we identified, we classified 26 noteworthy DEPs into four functional groups (**Table 1**): Sucrose metabolism (6 down and 2 up), organic acid metabolism (4 down and 2 up), the oxidationreduction process (4 up and 7 down), and cation transport (1 down).

#### 61 Proteins Were Induced in Response to both Cd and Zn Stresses

We observed 61 DEPs whose relative abundances were altered in response to both Cd and Zn stresses (**Figure 2**, **Data Sheet 8**). However, under Cd stress, the relative abundances of 24 proteins were increased and 37 were decreased, whereas under Zn stress, 29 were increased and 32 were decreased. We also observed 12 proteins whose relative abundances were altered inversely in response to Cd and Zn stress (**Data Sheet 8**, marked by yellow). These proteins were not induced in response to combined Cd+Zn stress, indicating that they might be involved in the mutual inhibition of Cd/Zn. Among the 61 proteins we identified were 5 noteworthy proteins that we divided into three functional pathways (Table 1): The oxidation-reduction process (2), sucrose metabolism (1) and SAM metabolism (2).

#### 33 Proteins Were Induced in Response to both Cd and Cd+Zn Stresses

We identified 33 proteins whose relative abundances were induced in response to both Cd and combined Cd+Zn stresses (**Figure 2**, **Data Sheet 9**). Under Cd stress, 13 proteins were more abundant, and 20 were less abundant, whereas under Cd+Zn stress, 18 were more abundant, and 15 were less abundant. These results indicate that Cd and Cd+Zn stresses induce differential molecular responses. We also identified 11 proteins whose relative abundances were altered inversely in response to Cd and Cd+Zn, including nicotianamine synthase 2 (NAS; **Data Sheet 9**, marked by yellow). That these proteins were not induced in response to Zn stress suggests that they might participate in Cd detoxification of Zn. Among the 33 proteins we identified, 2 proteins were key enzymes in SAM metabolism, 1 was involved in Ca metabolism, and 1 participated in sucrose metabolism (**Table 1**).

#### 57 Proteins Were Induced in Response to both Zn and Cd+Zn Stresses

We identified 57 proteins whose relative abundances were induced in response to both Zn and combined Cd+Zn stresses (**Figure 2**, **Data Sheet 10**). Under Zn stress, 28 proteins were more abundant, and 29 were less abundant, whereas under Cd+Zn stress, 39 were more abundant, and only 18 were less abundant (**Data Sheet 10**). These results indicate that Zn and Cd+Zn stresses induce differential molecular responses. We also identified 28 proteins whose relative abundances were altered inversely in response to Zn and Cd+Zn, including glutaredoxin-C8 and glucan 1,3-beta-glucosidase, (**Data Sheet 10**, marked by yellow). These proteins were not induced in response to Cd stress, which suggests that they might participate in Zn detoxification of Cd. Among the 57 proteins we identified, 4 proteins functioned in GSH metabolism, 3 were peroxidases involved in the oxidationreduction process, and 1 was involved in sucrose metabolism (**Table 1**).

#### 47 Proteins Were Induced only in Response to Cd+Zn Stress

We identified 47 proteins whose relative abundances were induced only in response to Cd+Zn stress (**Figure 2**, **Data Sheet 11**). These proteins were not induced in response to either Cd or Zn stress alone, which suggests that the molecular response induced by Cd+Zn stress differs from that induced by Cd and Zn stresses individually. Of the 47 proteins we identified, 24 proteins were more abundant, and 23 proteins were less abundant (**Data Sheet 11**). Among these, we identified 11 noteworthy proteins that functioned in Ca metabolism, the oxidation-reduction process, organic acid metabolism, and sucrose metabolism (**Table 1**).

### DISCUSSION

Interactions between Cd and Zn have previously been shown to be antagonistic and/or synergistic in various plants (Hart et al., 2002, 2005; Sun et al., 2005; Tkalec et al., 2014). In the present study, Cd uptake was inhibited by Zn and Zn uptake was inhibited by Cd in DPW roots (**Figure 1**). Cd transport from root to shoot was inhibited by Zn (**Figure 1A**) but was promoted by Zn after 5 days after treatment (unpublished data). Meanwhile, Zn transport from root to shoot was not affected by Cd (**Figure 1B**). These results indicate that Cd and Zinc interact antagonistically in DPW seedlings, as previously reported in bread and durum wheat (Hart et al., 2002, 2005; Sun et al., 2005) and unlike the synergistic interactions that have been reported in other wheat under field conditions (Nan et al., 2002).

Proteomic changes in the roots implicated several proteins in the antagonistic interactions of Cd/Zn (**Data Sheets 1**–**11**). Two days after treatment, the relative abundances of 206 (**Data Sheet 2**), 303 (**Data Sheet 3**), and 190 (**Data Sheet 4**) proteins were induced in response to Cd, Zn, and Cd+Zn stresses, respectively (**Figure 2**). Among these, 53 proteins were induced in response to all three treatments, and 58, 131, and 47 proteins were induced in response to only Cd, Zn, or Cd+Zn stresses, respectively (**Figure 2**). We grouped these proteins into different interactions of Cd/Zn (**Figure 2**). Our results indicate that although Cd and Zn have similar physical and chemical properties (Chesworth, 1991) and pathways for uptake (Grant et al., 1998), they induce differential molecular responses (Lin and Arats, 2012), which result in the antagonistic interactions of Cd/Zn in DPW roots (**Figure 1**) and the high tolerance of DPW to Cd and Zn toxicity (Wang X. et al., in press). Some proteins identified in this study that are involved in noteworthy processes are discussed below.

To overcome oxidative toxicity caused by heavy metal stresses (Ranieri et al., 2005; Lin et al., 2007; Kieffer et al., 2008; Di Baccio et al., 2011; Zeng et al., 2011), plants utilize an effective antioxidant system that protects their cells against oxidative damage (Kieffer et al., 2008; Di Baccio et al., 2011) by inducing the expression of oxidation-reduction-related proteins (Lin et al., 2007; Kieffer et al., 2008, 2009; Di Baccio et al., 2011; Zeng et al., 2011; Schneider et al., 2013). In this study, 31 oxidationreduction-related proteins were observed (**Table 1**). Of these, 8, 11, and 4 proteins were altered in response to Cd, Zn, and Cd+Zn stresses, respectively (**Table 1**). These results suggest that Cd, Zn, and Cd+Zn stresses cause differential oxidative threats which are detoxified through the induction of different oxidation-reduction-related proteins. Conversely, 8 Cd-induced proteins, 11 Zn-induced proteins, and 2 proteins induced by both Cd and Zn stresses were not induced in response to combined Cd+Zn stress, which suggests that the oxidative threats caused by Cd and Zn stresses are not the same as those caused by Cd+Zn stress. These results indicate that Cd and Zn detoxify each other in combined Cd+Zn stress, resulting in their uptakes being inhibited by each other. As described in previous studies (Kieffer et al., 2008, 2009; Schneider et al., 2013), Cd and Zn induced a greater abundance of some oxidative stress-related proteins but also induced a lower abundance of other oxidative stressrelated proteins (**Table 1**). Among these, 3 proteins were induced by all 3 stresses (Cd, Zn, and Cd+Zn) (**Table 1**), including ascorbate peroxidase (protein 285), L-ascorbate peroxidase 2 (protein 573), and peroxisome type ascorbate peroxidase (protein 511), which are key peroxide detoxification enzymes (Raven et al., 2004). These results suggest that ascorbate mediates Cd- and Zn-induced oxidative stress in plants (Kieffer et al., 2008).

In response to Cd and Zn stresses, plants form heavy metalglutathione (GSH) or metal-phytochelation (PC) compounds for metal detoxification (Seth et al., 2012; Jozefczak et al., 2015). GSH metabolism-related proteins, such as glutathione S-transferase (GST) and glutaredoxin (Grx), are differentially induced by Cd or Zn stress (Ahsan et al., 2009; Alvarez et al., 2009; Kieffer et al., 2009; Smiri et al., 2011; Zeng et al., 2011; Schneider et al., 2013). Meanwhile, GSTs translocate compounds of GSHcytotoxic substrates into vacuoles for detoxification (Kumar et al., 2013). In this study, all three stresses (Cd, Zn, and Cd+Zn) induced GST (protein 405), Grx (protein 141), lactoyglutathione lyase (proteins 424), and 2 sulfate metabolism-related proteins [sulfurtransferase (protein 540) and ATP sulfurylase (protein 112)] (**Table 1**), suggesting that sulfate availability for the synthesis of metal chelations such as GSH (Speiser et al., 1992) determines Cd and Zn tolerance (Nocito et al., 2006; Alvarez et al., 2009). Additionally, 3 GSH metabolism-related proteins, including glutamine synthetase cytosolic isozyme 1-2 (protein 210), GST (protein 696) and lactoyglutathione lyase (proteins 135), were induced only in response to Cd stress (**Table 1**), suggesting that Cd is detoxified through sequestration of GSH-Cd compounds into vacuoles and subsequent reduction of oxidative stress (Seth et al., 2012; Jozefczak et al., 2015). However, these proteins were not induced in response to combined Cd+Zn stress, which partly illustrates Zn detoxification of Cd. Interestingly, 2 GSTs (proteins 79 and 731) and 2 Grxs (proteins 384 and 472) were induced in response to both Zn and Cd+Zn stresses (**Table 1**), similar to the results obtained for some GSTs induced by Zn stress in Noccaea caerulescens (Schneider et al., 2013). These results suggest that these proteins participate in the detoxification of Zn stress-induced reactive oxygen species (Dixon et al., 2009; Schneider et al., 2013).

As a precursor of GSH, SAM plays important roles in protecting against Cd stress-induced reactive oxygen species (ROS) (Noriega et al., 2007). In the present study, protein levels of serine hydroxymethyltransferase (SHMT) and spermidine synthase 1, key enzymes in SAM metabolism, were altered in response to both Cd and Zn stresses, suggesting that SAM plays important roles in protecting against these stresses. Sadenosylmethionine synthase (SAMS) synthesizes SAM, which is a precursor of nicotianamine (NA) (Schneider et al., 2013). The protein level of nicotianamine synthase 2 (NAS), which synthesizes nicotianamine (NA) from SAM, increased in response to Cd stress. NA is an essential compound for cell-to-cell transport of Zn, Fe, and Cu (Takahashi et al., 2003; Klatte et al., 2009). However, the abundances of both SAMS and NAS decreased in response to combined Cd+Zn stress, whereas a previous report in N. caerulescens showed increased SAMS and NAS levels in response to Zn stress (Schneider et al., 2013). Our results partially illustrate Cd inhibition of Zn uptake. Cd stress also causes the lignification of roots (Finger-Teixeira et al., 2010). SAM provides the methyl donor to caffeic acid 3-Omethyltransferase (COMT) in lignin biosynthesis (Wang Y. et al., 2016). COMT levels were increased only in response to Cd stress (**Table 1**), suggesting that Cd also causes root lignification.

Some organic acids such as oxalate, malate, citrate, and fumarate are induced by Cd and Zn stress (Ueno et al., 2005; López-Millán et al., 2009; Zhu et al., 2011; Schneider et al., 2013) and form metal-organic acid complexes to act as metal chelators to promote detoxification in planta (Verbruggen et al., 2009). Further, Cd and Zn also induce key enzymes that participate in organic acid metabolism (López-Millán et al., 2009; Schneider et al., 2013). In this study, Zn stress induced several of these enzymes (**Table 1**), including malate dehydrogenase (protein 58), isocitrate dehydrogenase (protein 225), aconitate hygratase (protein 39), citrate synthase 4 (protein 82), and oxalate oxidase GF-2.8 (protein 376). These results suggest that detoxification of Zn could be achieved through the formation of Zn-organic acid complexes and subsequently, the complexes are deposited into vacuoles (Schneider et al., 2013). Conversely, organic acid secretion is associated with Cd and Zn exclusion (Zhu et al., 2011). Combined Cd+Zn stress resulted in decreased abundances of furmarate hydratase 2 (protein 116), malate dehydrogenase 1 (protein 537) and succinate dehydrogenase (815), which are key enzymes in furmarate, malate, and citrate metabolisms, respectively (**Table 1**). Thus, our results partially illustrate the mutually inhibited uptake of Cd/Zn in the roots (**Figure 1**).

Cellulose and pectic polysaccharides are major components of the plant cell wall (Cosgrove, 2005), which can be modified by some heavy metals. For example, Cd enhances the contents of glucose and polysaccharides in cell walls (Li et al., 2015). Further, exogenous glucose alleviates Cd toxicity by increasing Cd fixation in root cell walls (Shi et al., 2015). In this study, several sucrose metabolism-related proteins were induced by Cd, Zn, or Cd+Zn stress (**Table 1**), suggesting that glucose and/or polysaccharides participate in Cd and Zn fixation, exclusion or sequestration in root cell walls (Li et al., 2015; Shi et al., 2015). However, 8 sucrose metabolism-related proteins were induced in response to Zn stress but not combined Cd+Zn stress (**Table 1**), which suggests that Cd detoxifies excess Zn by inhibiting its uptake, resulting in Cd-induced inhibition of Zn modification of sucrose metabolism.

Zn stress also affects the expression of P-type ATPases and several other metal transporters (Schneider et al., 2013). P-type ATPases, such as AtHMA4 from Arabidopsis, GmHMA3 from soybean and AhHMA3 from A. halleri, have Zn uptake activity (Becher et al., 2004; Hussain et al., 2004; Wang et al., 2012). Further, AtHMA2, 3, and 4 have been shown to transport Zn from root to shoot (Williams and Mills, 2005). In this study, the abundance of a P-type proton pump ATPase (protein 679) decreased in response to Zn stress but not combined Cd+Zn stress (**Table 1**), a result which contradicts previous work in N. caerulescens showing increased abundance of two P-type ATPases in response to Zn stress (Schneider et al., 2013). However, we found that other Zn-induced metal transporters were not observed 2 days after treatment, but their transcripts were regulated by Cd, Zn, and Cd+Zn 5 days after treatment (unpublished data).

Finally, as reported by previous studies (Di Baccio et al., 2011; Zeng et al., 2011), proteins that were similarly or differentially induced in response to Cd, Zn, and/or Cd+Zn stresses also participated in other processes, including ribosome biogenesis, DNA metabolism, protein folding/modification (all SFiles), suggesting that these proteins might contribute to differential defense mechanisms against these stresses (Zeng et al., 2011).

#### CONCLUSION

Taken together, our results indicate that Cd and Zn interact antagonistically in DPW seedlings. Although 206, 303, and 190 proteins were induced in response to Cd, Zn, and Cd+Zn stresses, respectively, only 53 proteins were induced in response to all three stresses. 58, 131, and 47 proteins were induced only in response to Cd, Zn, and Cd+Zn stresses, respectively (**Figure 2**). These proteins could be divided into groups that resulted in different Cd/Zn interactions. Our results suggest that Zn and Cd stresses cause differential molecular responses in DPW. Under these stresses, oxidative stress-related proteins, metal chelators, metabolism-related proteins, sucrose metabolismrelated proteins, and metal transporters are differentially induced to participate in metal detoxification, which ultimately causes antagonistic interactions and enhanced tolerance of Cd and Zn.

#### AUTHOR CONTRIBUTIONS

YW, XW, XX, and YZ conceived and designed research, and wrote the manuscript. YW, XW, XX, CW, FP, and

#### REFERENCES


RW conducted experiments. YW, XW, JZ, HK, XF, LS, and HZ analyzed data. All authors read and approved the manuscript.

#### ACKNOWLEDGMENTS

The authors thank the National Natural Science Foundation of China (No. 31301349, 31270243, and 31470305), Bureau of Science and Technology and Bureau of Education of Sichuan Province, China. We would like to thank Lu Gao (Guangzhou Fitgene Biotechnology Co., Ltd) for useful advice and discussion.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 01378

Data Sheet 1 | Information about proteins identified in this study.

Data Sheet 2 | Proteins induced by Cd stress.

Data Sheet 3 | Proteins induced by Zn stress.

Data Sheet 4 | Proteins induced by Cd+Zn stress.

Data Sheet 5 | Proteins induced by all three treatments.

Data Sheet 6 | Proteins induced only by Cd stress.

Data Sheet 7 | Proteins induced only by Zn stress.

Data Sheet 8 | Proteins induced by both Cd and Zn stresses.

Data Sheet 9 | Proteins induced by both Cd and Cd+Zn stresses.

Data Sheet 10 | Proteins induced by both Zn and Cd+Zn stresses.

Data Sheet 11 | Proteins induced only by Cd+Zn stress.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Wang, Wang, Wang, Wang, Peng, Xiao, Zeng, Fan, Kang, Sha, Zhang and Zhou. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Dynamic Changes of the Plasma Membrane Proteins and the Protective Roles of Nitric Oxide in Rice Subjected to Heavy Metal Cadmium Stress

Liming Yang1, 2, 3 †, Jianhui Ji <sup>1</sup> † , Karen R. Harris-Shultz <sup>4</sup> , Hui Wang<sup>2</sup> , Hongliang Wang<sup>4</sup> , Elsayed F. Abd-Allah<sup>5</sup> , Yuming Luo<sup>1</sup> \* and Xiangyang Hu<sup>6</sup> \*

#### Edited by:

Dipanjana Ghosh, National University of Singapore, Singapore

#### Reviewed by:

Christian Lindermayr, Helmholtz Zentrum München - German Research Center for Environmental Health, Germany Jisen Shi, Nanjing Forestry University, China

#### \*Correspondence:

Yuming Luo yumingluo@163.com; Xiangyang Hu huxiangyang@mail.kib.ac.cn † These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 05 November 2015 Accepted: 04 February 2016 Published: 26 February 2016

#### Citation:

Yang L, Ji J, Harris-Shultz KR, Wang H, Wang H, Abd-Allah EF, Luo Y and Hu X (2016) The Dynamic Changes of the Plasma Membrane Proteins and the Protective Roles of Nitric Oxide in Rice Subjected to Heavy Metal Cadmium Stress. Front. Plant Sci. 7:190. doi: 10.3389/fpls.2016.00190 <sup>1</sup> Jiangsu Key Laboratory for Eco-Agriculture Biotechnology around Hongze Lake, Jiangsu Collaborative Innovation Center of Regional Modern Agriculture and Environment Protection, Huaiyin Normal University, Huaian, China, <sup>2</sup> Department of Plant Pathology, University of Georgia, Tifton, GA, USA, <sup>3</sup> Crop Protection and Management Research Unit, United States Department of Agriculture, Agricultural Research Service, Tifton, GA, USA, <sup>4</sup> Crop Genetics and Breeding Research Unit, United States Department of Agriculture, Agricultural Research Service, Tifton, GA, USA, <sup>5</sup> Department of Plant Production, Faculty of Food and Agricultural Sciences, King Saud University, Riyadh, Saudi Arabia, <sup>6</sup> Shanghai Key Laboratory of Bio-Energy Crops, School of Life Sciences, Shanghai University, Shanghai, China

The heavy metal cadmium is a common environmental contaminant in soils and has adverse effects on crop growth and development. The signaling processes in plants that initiate cellular responses to environmental stress have been shown to be located in the plasma membrane (PM). A better understanding of the PM proteome in response to environmental stress might provide new insights for improving stress-tolerant crops. Nitric oxide (NO) is reported to be involved in the plant response to cadmium (Cd) stress. To further investigate how NO modulates protein changes in the plasma membrane during Cd stress, a quantitative proteomics approach based on isobaric tags for relative and absolute quantification (iTRAQ) was used to identify differentially regulated proteins from the rice plasma membrane after Cd or Cd and NO treatment. Sixty-six differentially expressed proteins were identified, of which, many function as transporters, ATPases, kinases, metabolic enzymes, phosphatases, and phospholipases. Among these, the abundance of phospholipase D (PLD) was altered substantially after the treatment of Cd or Cd and NO. Transient expression of the PLD fused with green fluorescent peptide (GFP) in rice protoplasts showed that the Cd and NO treatment promoted the accumulation of PLD in the plasma membrane. Addition of NO also enhanced Cd-induced PLD activity and the accumulation of phosphatidic acid (PA) produced through PLD activity. Meanwhile, NO elevated the activities of antioxidant enzymes and caused the accumulation of glutathione, both which function to reduce Cd-induced H2O<sup>2</sup> accumulation. Taken together, we suggest that NO signaling is associated with the accumulation of antioxidant enzymes, glutathione and PA which increases cadmium tolerance in rice via the antioxidant defense system.

Keywords: rice, cadmium pollution, nitric oxide, reactive oxygen species, lipid hydrolysis, quantitative proteomics

## INTRODUCTION

The heavy metal cadmium is a common environmental contaminant with a long biological half-life in the soil and has adverse effects on crop growth and development (Das et al., 1997; Tuan Anh and Popova, 2013; Choppala et al., 2014). Cadmium pollution is caused by the application of phosphate fertilizer and certain industrial processes and poses a critical threat to human health due to its over-accumulation in the soil and in crop food (Prasad, 1995; Das et al., 1997; Zhang et al., 2015).

Cadmium is absorbed by plant roots and is quickly transported to the leaves via the xylem. Most plants are very sensitive to trace amounts of cadmium, responding with retardation in growth and development, due to decreased photosynthesis in the leaves. This impairs photosynthetic supply and accelerates apoptosis and necrosis of the leaves. In plants, cadmium can be detoxified by phytochelatins, a class of glutathione-derived peptides containing repeating units of Glu and Cys that function by binding metal ions and transporting them to the vacuole (Howden et al., 1995; Di Toppi and Gabbrielli, 1999). Plant cells also can use antioxidant enzymes to degrade over-produced reactive oxygen species (ROS) to diminish damage from cadmium exposure (Cobbett et al., 1998; Singh and Tewari, 2003; Balestrasse et al., 2006). In addition, nitric oxide (NO), an important signaling molecule, participates in the cadmium stress response in plants and protects against cadmium-induced damage in Helianthus (Laspina et al., 2005) and aluminum toxicity in rice (Yang et al., 2013). Furthermore, NO is required for cadmium-induced programmed cell death in Arabidopsis (Balestrasse et al., 2006; De Michele et al., 2009).

The plasma membrane of a plant cell is a highly organized system that mediates the exchange of information and materials between the cell interior and the extracellular environment. The enzymes or molecules located in the plasma membrane play important roles in transferring stress signals in plants. Phospholipases, including phospholipase D (PLD), phospholipase C (PLC), and phospholipase A (PLA), play an important role in lipid hydrolysis of the plasma membrane and in the mediation of the stress response in plants (Wang et al., 2002; Wang, 2006). For example, heavy metal stress from copper ions can trigger PLD activity in wheat roots (Wang et al., 2002; Navari-Izzo et al., 2006). In Arabidopsis, phosphatidic acid (PA) produced by PLD interacts with NADPH oxidase in the plasma membrane to generate H2O<sup>2</sup> in response to abscisic acid (ABA; Wang, 2006; Zhang et al., 2009). However, the cross-talk between NO, PLD activity, and the antioxidant system during the stress response to heavy metals remains uncharacterized.

The plasma membrane is the first site of extracellular biotic or abiotic sensing, so an understanding of proteome dynamics may facilitate the development of new strategies for stress resistance in crops (Sussman, 1994; Santoni et al., 1998; Alexandersson et al., 2004). As the primary environmental barrier, the plasma membrane of a plant cell controls many biological processes such as ion transport, endocytosis, cell differentiation and proliferation, and signal transduction. However, functional protein localization is complicated because some membrane proteins are tightly associated with the dual lipid core, whereas others are loosely and reversibly associated (Sussman and Harper, 1989; Sussman, 1994). Technological advances in the extraction of purified plant plasma membrane proteins have made it possible to profile the proteome of the entire plant membrane (Santoni et al., 2000; Alexandersson et al., 2004). Proteome profiling may elucidate which systemic protective responses are initiated in plant plasma membranes exposed to environmental stress.

To investigate the accumulation patterns of proteins localized in the plasma membranes of rice seedlings exposed to cadmium, we conducted an iTRAQ-based quantitative proteomic analysis for proteins in the plasma membrane. We found that cadmium treatment induced a rapid increase in a set of transporter proteins and ATPases, as well as a PLD protein. Subsequent experiments showed that cadmium induced rapid generation of NO and PA (produced assumedly through PLD activity). The exogenous addition of either NO or PA with Cd mitigated cadmium toxicity in rice seedlings and augmented synthesis of glutathione that functions to reduce oxidative damage. Based on these results, we propose a model that depicts how NO and PLD improves Cd tolerance in rice seedlings.

### MATERIALS AND METHODS

#### Growth of Rice Seedlings and Cadmium Treatment

Seeds of the rice variety (Oryza sativa ssp. Japonica cv Zhonghua11) were surface-sterilized, washed, and germinated on wet filter paper. Seeds were then grown in 1/4 Hoagland's nutrient solution (pH 5.5) at 23◦C and 16 h light/8 h dark conditions. When the third leaves of the seedlings emerged, CdCl<sup>2</sup> was added to the culture solution to a final concentration of 10µM. To examine the impact of NO, rice seedlings were pretreated with a 30µM solution of S-nitroso-N-acetylpenicillamine (SNAP), a spontaneous NO donor, for 2 h prior to cadmium exposure. The SNAP solution was refreshed every 3 days, and the pH of solution was maintained at 5.5. Aliquots of the seedlings were harvested for subsequent assays at the indicated times. For the inhibitor treatment, different inhibitors including 2-(4-carboxyphenyl)-4,4,5,5- tetra methylimidazoline-1-oxyl-3 oxide (cPTIO), and 1-butanol (1-Bu) were added into the rice culture solution for 2 h prior to cadmium treatment, respectively.

### Extraction and Purification of Plasma Membranes from Rice Seedlings

The rice plasma membrane was enriched by aqueous two phase partitioning as previously described (Nohzadeh Malakshah et al., 2007). In brief, about 10 g of rice seedlings were ground to a powder in liquid nitrogen and homogenized in 50 ml of ice-cold 50 mM 3-(N-morpholino) propanesulfonic acid (MOPS)/KOH buffer (pH 7.5) containing 330 mM sucrose, 5 mM EDTA, 5 mM DTT, 5 mM ascorbate, 0.5 mM phenyl-methyl-sulfonyl fluoride (PMSF), 0.2% BSA, 0.2% casein, and 0.6% polyvinylpyrrolidone (PVP) at 4◦C. The homogenate was centrifuged at 2000 × g for

10 min at 4◦C, and the supernatant was filtered through a 260 µm filter. The filtrate was centrifuged again at 12,000 × g for 10 min at 4◦C, and the resulting supernatant was centrifuged at 50,000 × g for 60 min at 4◦C to precipitate the microsomal pellets. To enrich for plasma membranes, the microsomal pellets were re-suspended in 10 ml of resuspension buffer (330 mM sucrose, 5 mM potassium phosphate [pH 7.8], 2 mM potassium chloride, 1 mM DTT, and 0.1 mM EDTA) and were mixed with a phase mixture containing 6.3% (w/w) Dextran T500 (Sigma-Aldrich, St. Louis, MO, US), 6.3% (w/w) PEG3350 (Sigma-Aldrich), 330 mM sucrose, 5 mM phosphate buffer [pH 7.8], 1 mM KCl, 0.5 mM EDTA, and 1 mM DTT to yield a 36-g phase system. After mixing, phase separation was conducted at 4◦C. The upper (PEG) phase containing a second nascent partitioning was further purified into two sub-phases. The upper phase of this partition was diluted with wash buffer (250 mM sucrose, 10 mM MOPS) and then centrifuged at 100,000 × g for 60 min. The resulting plasma membrane pellets were dissolved into 1 ml of sample buffer containing 330 mM sucrose and 50 mM 2-(N-morpholino) ethanesulfonic acid (MES)/KOH (pH 6.0). The purity of the isolated plasma membranes was evaluated by measuring the activities of several marker enzymes for different membrane fractions.

### Protein Digestion, iTRAQ Labeling, and Protein Quantification

The proteins of the plasma membrane were prepared for iTRAQ labeling as previously described (Kong et al., 2014), and then was dissolved in 1% SDS, 100 mM triethylammonium bicarbonate, pH 8.5, followed by reduction, alkylation, trypsin digestion, and labeling using 8-plex iTRAQ reagent kits, according to the manufacturer's instructions (AB Sciex, Framingham, MA). Labeled samples were lyophilized, and the peptide mixture was dissolved in a strong cation exchange (SCX) solvent A (25% v/v acetonitrile, 10 mM ammonium formate, pH 2.8). The resulting peptides were fractionated on an Agilent HPLC system 1100 with a polysulfethyl A column (2.1 × 100 mm, 5µm, 300 A, PolyLC, Columbia, MD). Peptides were eluted at a flow rate of 200µL/min with a linear gradient of 0-20% of solvent B consisting of 25% v/v acetonitrile, 500 mM ammonium formate for over 50 min, followed by ramping up to 100% solvent B for 5 min and holding for 10 min. The absorbance at 214 nm was monitored, and a total of 12 fractions were collected. Each SCX fraction was lyophilized and dissolved in solvent A (3% acetonitrile v/v, 0.1% formic acid v/v) and then was analyzed with a Q-Exactive Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo Finnigan Scientific, San Jose, CA). Samples were separated on a Hypersile Gold C18 column (100 mm × 2.1 mm, 1.9µm; Thermo Fisher Scientific, Pittsburgh, PA). Peptides were eluted with a linear gradient of acetonitrile/0.1% formic acid from 3 to 50% for 90 min at a flow rate of 250 nL/min. Peptides were then sprayed into the orifice of the Q-Exactive MS/MS system with a spray voltage of 2.2 kV. Full-scan mass spectra were performed over 200–1800 m/z at high resolution at 60,000. At least the four most intense precursor ions were selected for collision-induced fragmentation in the linear ion trap with 50–2000 m/z and 30–2000 ms at a resolution of 7500. Dynamic exclusion was employed within 40 s to prevent repetitive selection of the peptides.

The raw LC-MS/MS files were analyzed using the software, Proteome Discoverer 1.3 (Thermo Fisher Scientific, Pittsburgh, PA), which was connected to the Mascot Search Engine server, version 2.3 (Matrix Science, Boston, MA). The spectra were searched against the NCBInr protein database (Taxonomy: O. sativa, which contains 132,343 sequences). Search parameters included iTRAQ 8-plex quantification, carbamidomethylation of cysteine was set as a fixed modification, and oxidation of methionine was set as a variable modification. Trypsin was specified as the proteolytic enzyme, and one missed cleavage was allowed. Peptide mass tolerance was set at 10 ppm; fragment mass tolerance was set at 0.1 Da. An automatic decoy database search was performed as part of the search. False discovery rates (FDRs) for peptide identification of all searches were <1.0%. The data were pre-filtered to exclude MS/MS spectra containing fewer than three peaks. Mascot results were filtered with the Mascot Percolator package to improve the accuracy and sensitivity of peptide identification. For differential analyses, all proteins identified and quantified with at least four independent peptides with a high degree of confidence (FDR 1%) were selected. The quantification was performed by normalizing the results of all of the measured iTRAQ reported ratios value using the software Proteome Discoverer 1.3 software. Only the significant ratios from the replicates were used to calculate the average ratio for the protein. It should be noted that each p value was generated based on quantitative information derived from at least three independent peptides in each replicate. Cut-offs of 1.2- or 0.6 fold were set to indicate up-regulation or down-regulation of proteins, and a p < 0.05 was used to indicate significance.

### Analyses of Electrolyte Leakage and Chlorophyll Fluorescence Analyses

One gram of treated rice seedling roots were excised and transferred to tubes containing 10 ml of deionized water. The conductivity of the solution was measured after shaking overnight at room temperature. After measurement, samples were autoclaved, and the conductivity of the solution was measured again. The percentage of electrolyte leakage was then calculated, as previously described (Xia et al., 2009). The chlorophyll fluorescence was analyzed with a PAM (pulseamplitude modulation) Chlorophyll Fluorometer (Heinz-Walz-GmbH, Effeltrich, Germany) as previously described (Yang et al., 2013; Ma et al., 2015).

### Analyses of H2O2, NO, Glutathione, and Phytochelatin

H2O<sup>2</sup> levels were measured as described previously (Yang et al., 2013). NO levels were detected using the NO-specific fluorescence probe, 4-amino-5-methylamino-2 ′ ,7′ -difluorofluorescein diacetate (DAF-FMDA), under epifluorescence microscopy (PCM 2000, excitation 488 nm, emission 515–560 nm, Nikon, Tokyo, Japan) or by using a hemoglobin assay (Hu et al., 2005). The levels of glutathione and

phytochelatin were analyzed using a reported method (Gupta and Goldsbrough, 1991).

### Analyses of Lipid Molecular Species and PLD Activity

After different treatments, about 1 g of rice seedlings was immediately immersed in 3 mL of isopropanol with 0.01% butylated hydroxytoluene to terminate lipolytic activities. Lipid extraction, ESI-MS/MS analysis, and quantification were done as described previously (Zhang et al., 2009). Five replicates of each treatment were carried out and analyzed. The PLD activity was measured as described (Devaiah et al., 2007).

### Measurement of Antioxidant Enzyme Activity

Leaves (1–2 g) were homogenized in 50 mM sodium phosphate buffer (pH 7.0) containing 1.0 mM ethylenediaminetetraacetic acid (EDTA), 0.5% (v/v) Triton X-100 and 1% (w/v) polyvinylpyrrolidone (PVPP; 100 mg tissue/mL buffer). For the analysis of APX, the extraction buffer also contained 5 mM AsA. The homogenates were centrifuged at 15,000 × g for 20 min at 4◦C, and the resulting supernatant was immediately used for the antioxidant enzyme assays. The total activities of ascorbate peroxidase (APX, EC1.11.1.11), superoxide dismutase (SOD, EC1.15.1.1), and glutathione reductase (GR, EC1.6.4.2) in the supernatants were determined as described previously (Zhang et al., 2010; Yang et al., 2015).

### Plasmid Construction and Transient Transformation

A reporter protein construct, named GFP-PLDa, comprised of PLD fused with green fluorescent protein (GFP) was generated by amplifying the full-length PLD alpha (PLDa) cDNA (GenBank: NM\_001064550) using the primers of 5′GGCCGAATTCatggcggagcagcagctgatgc3′ and 5 ′GGCCAAGCTTctacgaggt gatatcgggggtca3′ . The resulting PCR fragments were digested with EcoRI and HindIII. The digested fragments were inserted into the binary vector pEGAD to fuse with EGFP at C-terminus. The final construct was confirmed by sequencing. The protoplast preparation and transient transformation were manipulated as described previously (Sheen, 2001; Yoo et al., 2007).

### SDS-PAGE and Western Blotting

Samples (1 g) were ground to a powder in liquid nitrogen, and were homogenized with 10 mL of extraction buffer containing 125 mM Tris-Cl [pH 7.5], 10% SDS, and 10% mercaptoethanol at 4 ◦C. The mixture was centrifuged at 12,000 × g at 4◦C for 10 min, and proteins of the resulting supernatant were separated by SDS-PAGE. Following electrophoresis, separated proteins were transferred to a Hybond-C nitrocellulose membrane (GE Life Sciences, Buckinghamshire, UK) using the Multiphor II semi-dry blotting apparatus, according to the manufacturer's instructions (GE Life Sciences). The transferred proteins were visualized using an ECL detection kit (Roche, Mannheim, Germany), and the antibodies against plant SOD (1:2000), GR (1:1000), and APX (1:2000) were obtained from Agrisera (Agrisera, Vannas, Sweden). To detect PLD proteins in the plasma membrane, the rabbit polyclonal anti-PLDa antibody was prepared by immunizing rabbits with the 14-amino acid N-terminal sequence of the rice PLDa protein (MAHLLMHGTLDATI; GenBank: BAD35530) conjugated to the keyhole limpet hemocyanin.

## RESULTS

### NO Reduces Cadmium Toxicity to Leaf Photosynthesis and Increases Root Growth

To investigate the effect of cadmium on the growth and development of rice, we first assessed the dose effect of cadmium on 2-week old rice seedlings. As shown in **Supplemental Figure 1**, increasing the concentration of cadmium decreased shoot and root growth. Seedlings treated with 10µM of cadmium showed a 50% inhibitory effect on rice growth. Thus the concentration used for cadmium treatment was 10µM for the subsequent experiments to allow for direct comparisons. As shown in **Figure 1A**, cadmium treatment for 1 week significantly suppressed rice seedling growth, and caused yellowing of leaves as compared to the control plants. The chlorophyll content in the leaves of rice is an important index to evaluate the photosynthesis capability and plant tolerance to environmental stress. Here we found that Cd treatment for 1 week reduced the chlorophyll content by 59.8% as compared to control plants (**Figure 1B**). Cd treatment also suppressed the root and stem length by 64.1 and 60.0%, respectively, compared the control seedlings without cadmium treatment (**Figure 1C**). The ratio of Fv/Fm and ion leakage is a measurement of the photosynthetic capability and the degree of membrane damage, respectively. Cd treatment decreased Fv/Fm from 0.64 to 0.34 and increased the ion leakage by 13.9-fold (**Figure 1D**). Cd treatment suppressed the biomass of rice by 48.3% compared with the control lines without Cd treatment (**Figure 2A**). Phytochelatins were increased with cadmium treatment and peaked after 48 h of Cd treatment (**Figure 2B**). This molecule plays an important role in heavy metal detoxification in plants. These data indicate that Cd treatment for 1 week damaged the rice seedling viability.

NO signaling is reported to reduce the toxic effects of heavy metals on plants. Therefore we investigated the role of NO in the response of rice seedlings to Cd stress. The NO artificial donor SNAP treatment significantly improved seedling tolerance to cadmium by facilitating growth (**Figures 1A,C**), reducing ion leakage (**Figure 1D**), increasing leaf chlorophyll synthesis capability (including the Fv/Fm ratio and chlorophyll content; **Figures 1B,D**), and increasing biomass and phytochelatin content (**Figures 2A,B**). However, these SNAP effects were abolished when the NO scavenger, 2-(4-carboxyphenyl)-4, 4, 5, 5-tetramethylimidazoline-1-oxyl-3-oxide (cPTIO), was applied (**Figures 1**, **2**). These data suggest that NO plays an essential role in the rice seedling response to Cd treatment.

In addition, PA treatment also significantly improved seedling tolerance to cadmium (**Figures 1A,C**), increasing total chlorophyll content (**Figures 1A,B**), and increasing biomass and phytochelatin content (**Figures 2A,B**). Moreover, the addition of

Cd and 1-Bu, an inhibitor of PLD, decreased the total chlorophyll content, stem length, and biomass as compared to control plants (**Figures 1A–C**, **2A**).

To further understand the role of NO during the rice response to Cd treatment, we observed the in-situ NO accumulation in rice roots subjected to Cd treatment. Using a NO-specific fluorescence probe DAF-FMDA staining, we found that cadmium exposure rapidly increased NO production (**Figures 3A,B**). NO fluorescence was enhanced with the addition of SNAP and abolished in the presence of cPTIO. By directly measuring the NO content in the roots, we found that Cd treatment increased the NO content as compared to control plants, but this increase could be suppressed by the NO scavenger cPTIO treatment (**Figure 3B**). Consistent with the NO staining fluorescence results, the addition of Cd and SNAP could further increase NO content. These data suggest that rice seedlings treated with Cd induced NO biosynthesis.

### When Plants are Treated with Cadmium or Cadmium and NO, Dynamic Changes Occur in Plasma Membrane Proteins

The plant plasma membrane plays an important role during environmental stress, and most of the previous proteomics studies used total plant proteins to investigate the underlying mechanisms of the plant response to Cd treatment. In this study, we focused on the role of the plasma membrane proteins by isolating plasma membrane proteins and comparing the treatments using iTRAQ (isobaric tags for relative and absolute quantitation) proteomics. To understand the functions of the plasma membrane proteins of rice in stress response, an iTRAQ approach was used to measure the change in plasma membrane proteins isolated from 2-week old seedlings after 12 h or 1 day of 10µM Cd treatment, or 12 h or 1 day of 10µM Cd treatment with 30µM SNAP. The seedlings without Cd or SNAP treatment were used as the control. We isolated and purified the plasma membrane proteins from rice seedlings after Cd treatment using the two-phase partitioning method. The purity of the plasma membrane fraction was evaluated using marker enzymes associated with various subcellular membranes (Natera et al., 2008). Orthovanadate-sensitive ATPase, nitratesensitive ATPase, and azide-sensitive ATPase were selected as plasma membrane, vacuolar, and mitochondrial membrane markers, respectively. The ATPase activity of the prepared plasma membrane was primarily sensitive to sodium orthovanadate (**Supplemental Table 1**). Furthermore, we assessed the purity of the isolated plasma membrane proteins using antibodies against marker proteins corresponding to various subcellular membranes. H+-ATPase is a marker protein for the plasma membrane protein fraction, V-ATPase is a tonoplast membrane

means ± SEs of at least three independent experiments (n = 10/experiment). Different symbols above the bars indicate significant differences (Tukey's test, p < 0.05).

protein, and molecular chaperone binding protein (BiP) is a protein of the endoplasmic reticulum (ER). Compared with the total protein fraction, the stronger immunoblot signal was detected in the PM fraction using the anti-H+-ATPase antibody, suggesting that H+-ATPase was strongly enriched in the PM fraction (**Supplemental Figure 2**). Furthermore, the anti-H+-ATPase antibody immunoblotting signal was not detected in other vesicles such as tonoplast, mitochondria, or ER (**Supplemental Figure 2**). These data suggest that our method is sufficient to obtain high-quality plasma membranes to perform a plasma membrane proteomics study.

We determined the ratios of protein abundance for the following four groups: (1) Cd treatment for 12 h (12 h Cd)/the control line without treatment (CK), (2) Cd treatment for 1 day (1 d Cd)/the control line without treatment (CK), (3) Cd treatment for 12 h with SNAP (12 h Cd+SNAP)/the control line without treatment (CK), (4) Cd treatment for 1 day with SNAP (1d Cd+SNAP)/the control line without treatment (CK), to identify proteins affected by Cd or Cd and NO treatments. With a threshold of fold-change cutoff of 1.2-fold for increased accumulation and <0.6-fold for decreased accumulation, a total of 66 proteins showed differential accumulation (p < 0.05) when subjected to Cd or Cd and SNAP treatment as compared to the control line without treatment (**Figure 4**, **Table 1**, **Supplemental Table 2**). Among these differentially regulated proteins, 27 proteins were upregulated and 17 proteins were downregulated after 12 h of Cd treatment, 30 proteins were upregulated and 23 proteins were downregulated after 1 day of Cd treatment, 29 proteins were upregulated and 13 proteins were downregulated after 12 h of Cd and SNAP, and 31 proteins were upregulated and 24 proteins were downregulated after 1 day of Cd treatment and SNAP treatment. SNAP treatment did not increase the number of proteins showing up-regulation or down-regulation (**Figure 4A**), but rather only affected the expression intensities for these proteins. These identified proteins were divided into nine groups based on their biological functions (**Table 1**). The majority of these proteins were plasma membrane transporters (16), including nitrate transporters, phosphate transporters, oligopeptide transporters, a sucrose transporter, iron transporter, and a monosaccharide transporter etc. followed by ATPases (9) and kinases (9). Additionally, phosphatases and phospholipases (5), metabolism enzymes (6), antiporters (3), structural proteins (6), aquaporins (4), and signal and hormone related proteins (8) showed differential regulation. A hierarchical cluster analysis was conducted to categorize the proteins that showed differential expression profiles during Cd or Cd and SNAP treatment (**Figure 4B**). We found the expression profile for the identified proteins could be clustered into two groups. The proteins belonging to one group showed up-regulation after Cd treatment and the Cd+SNAP treatment could further enhance their upregulation. The proteins belong to the other group showed down-regulation and Cd+SNAP treatment further suppressed their accumulation. We also found most proteins that were aquaporins or involved in signal transduction and hormone response, were also differentially regulated by Cd or Cd+SNAP treatment (**Table 1**). Among these identified proteins, we observed one PLD protein that was up-regulated after Cd and Cd+SNAP treatments (**Table 1**-protein is highlighted in bold). Because it has not been reported before that PLD proteins could be localized to the plasma membrane after Cd or Cd and NO treatments, or that PLD protein accumulation could be regulated by Cd or Cd and NO treatments, these data hint that PLD has a function in the rice seedlings response to Cd stress. Thus, the role of PLD after Cd and Cd and NO treatments was further investigated.

#### Exogenous NO Treatment Promotes the Translocation of PLD from the Nucleus to the Plasma Membrane

To understand the role of PLD during the rice response to Cd stress, we analyzed PLD localization in rice protoplasts transiently transformed with the vector expressing the pUBI:GFP-PLDa fusion gene. Prior to cadmium treatment, we observed strong GFP fluorescence in the nuclei before Cd treatment, indicating that GFP-PLDa was sequestered to the nuclei of transformed protoplasts before Cd treatment (**Figure 5A**). After 12 h of Cd treatment, a strong GFP fluorescence were also observed in the nuclei and around

the plasma membrane, suggesting Cd treatment affected the expression of GFP-PLDa after Cd treatment (**Figure 5A**). We then found that addition of Cd and SNAP treatment further increased GFP fluorescence around the plasma membrane (**Figure 5A**). This effect was diminished by the NO scavenger cPTIO pretreatment. These data suggest that NO promotes the localization of PLDa to the plasma membrane, which is consistent with the proteomic data that Cd and Cd and SNAP treatment increased the accumulation of PLD in the plasma membrane (**Table 1**). Additionally, the anti-PLDa antibody was used to analyze PLDa protein accumulation in the total extracted protein or in the plasma membrane fraction from the seedlings after Cd and Cd+SNAP treatment. We found that Cd or Cd+SNAP treatments increased the level of PLD protein in the plasma membrane (**Figure 5B**) and partially increased the total PLD protein level in the total protein (**Figure 5C**). This increase in PLDa could be decreased in the PM and total protein fractions by pretreatment with cPTIO (**Figures 5B,C**).

We further measured PLD activity after Cd or Cd and SNAP treatment. Cd treatment induced PLD activity and PLD activity was highest between 12 and 24 h after Cd treatment (**Figure 5D**). Cd+SNAP treatment could further enhance the PLD activity, and the Cd+SNAP+cPTIO treatment (cPTIO is a NO metabolism scavenger) could suppress Cd or Cd+SNAP-induced increase in PLD activity (**Figure 5E**). These data correlated to our previous

#### Accession number<sup>a</sup> Protein names Cd (12 h): Control<sup>b</sup> Cd (1 day): Control<sup>c</sup> Cd+SNAP (12 h): Control<sup>d</sup> Cd+SNAP (1 day): Control<sup>e</sup> Rep 1 Rep 2 Rep 3 p-value p-value p-value Average Rep 1 Rep 2 Rep 3 p-value p-value p-value Average Rep 1 Rep 2 Rep 3 p-value p-value p-value Average Rep 1 Rep 2 Rep 3 p-value p-value p-value Average ATPases gi|218179 H-ATPase 1.664 1.586 1.624 0.037 0.023 0.034 1.625 1.769 1.770 1.742 0.013 0.005 0.006 1.760 1.926 1.800 1.871 0.038 0.035 0.037 1.866 1.932 1.961 1.926 0.004 0.005 0.009 1.940 gi|20302439 Plasma membrane H+-ATPase 1.292 1.269 1.397 0.017 0.009 0.008 1.319 1.384 1.541 1.440 0.021 0.032 0.023 1.455 1.493 1.477 1.420 0.033 0.003 0.021 1.463 1.689 1.699 1.667 0.046 0.005 0.008 1.685 gi|31432331 AAA family ATPase 0.811 0.843 0.838 0.041 0.012 0.032 0.531 0.971 0.887 0.859 0.003 0.036 0.008 0.706 0.986 0.940 0.994 0.034 0.009 0.012 1.273 1.510 1.457 1.684 0.003 0.004 0.008 1.550 gi|50252047 Putative calcium-transporting ATPase 0.850 0.785 0.794 0.006 0.009 0.008 0.510 0.956 0.805 0.883 0.016 0.009 0.021 0.881 0.850 0.983 0.955 0.029 0.028 0.001 0.929 1.611 1.633 1.603 0.042 0.035 0.021 1.616 gi|20302437 Plasma membrane H-ATPase 0.945 0.854 0.842 0.034 0.021 0.024 0.880 1.194 1.282 1.268 0.004 0.002 0.031 1.248 1.265 1.268 1.319 0.021 0.021 0.018 1.384 1.552 1.720 1.682 0.021 0.035 0.021 1.651 gi|544586339 Type IIB Ca2+ATPase 1.589 1.483 1.616 0.035 0.031 0.025 1.563 1.661 1.637 1.641 0.021 0.015 0.018 1.646 1.561 1.424 1.487 0.021 0.005 0.017 1.791 1.651 1.652 1.519 0.012 0.035 0.007 1.607 gi|77552962 Calcium-transporting ATPase 4 1.451 1.421 1.450 0.011 0.016 0.015 1.441 1.702 1.870 1.689 0.010 0.010 0.016 1.754 1.515 1.527 1.575 0.034 0.007 0.007 1.539 1.750 1.715 1.709 0.030 0.058 0.024 1.725 gi|31432100 Calcium-transporting ATPase 13 1.318 1.337 1.371 0.016 0.011 1.342 1.530 1.555 1.592 0.042 0.021 1.559 1.685 1.624 1.677 0.029 0.024 1.662 1.887 1.866 1.855 0.016 0.005 1.869 gi|122248711 ATP-citrate synthase alpha chain protein 2 0.554 0.552 0.637 0.030 0.022 0.025 0.581 0.445 0.453 0.442 0.021 0.027 0.021 0.447 0.554 0.552 0.005 0.020 0.023 0.369 0.554 0.552 0.006 0.025 0.011 0.369 TRANSPORTERS gi|23600439 Putative phosphate transporter OsPT1 0.777 0.592 0.624 0.017 0.020 0.021 0.664 0.573 0.498 0.480 0.043 0.035 0.017 0.517 0.854 0.687 0.583 0.027 0.028 0.025 0.708 0.829 0.736 0.835 0.009 0.005 0.007 0.800 gi|52550769 Phosphate transporter 7 0.762 0.778 0.744 0.011 0.020 0.007 0.861 0.798 0.759 0.771 0.048 0.021 0.035 0.576 0.678 0.646 0.648 0.036 0.036 0.032 0.657 0.474 0.509 0.482 0.010 0.021 0.009 0.488 gi|52550767 Phosphate transporter 5 0.514 0.492 0.454 0.003 0.028 0.025 0.487 0.434 0.442 0.345 0.017 0.021 0.021 0.407 0.505 0.511 0.499 0.006 0.007 0.008 0.505 0.480 0.489 0.545 0.038 0.035 0.032 0.505 gi|20279475 Putative ABC transporter 1.093 1.144 0.993 0.039 0.036 0.032 1.077 1.569 1.696 1.705 0.019 0.021 0.022 1.657 1.467 1.424 1.412 0.026 0.028 0.025 1.434 1.765 1.756 1.771 0.022 0.035 0.035 1.764 gi|37548736 Sucrose transporter SUT2 1.356 1.471 1.467 0.044 0.022 0.023 1.431 1.519 1.450 1.589 0.035 0.025 0.032 1.519 1.517 1.563 1.465 0.013 0.023 0.032 1.515 1.566 1.647 1.669 0.029 0.007 0.012 1.627

#### TABLE 1 | List of proteins differentially regulated by Cd and Cd+SNAP in the plasma membrane as determined by using iTRAQ analysis.





Protein Phospholipase D was bold-marked for further analysis in this study.

The differentially regulated proteins induced by cadmium and cadmium and S-nitroso-N-acetylpenicillamine (SNAP) in enriched plasma membrane from rice analyzed after using isobaric tags for relative and absolute quantification (iTRAQ).

<sup>a</sup>NCBI gene number ID.

b three repeated iTRAQ experiments between Cd treatment for 12 h and control without any treatment.

c three repeated iTRAQ experiments between Cd treatment for 1 day and control without any treatment.

d three repeated iTRAQ experiments between Cd treatment plus SNAP for 12 h and control without any treatment.

e three repeated iTRAQ experiments between Cd treatment plus SNAP for 1 day and control without any treatment.

results and suggest that Cd and Cd+SNAP treatment could increase PLD protein levels and increase PLD enzyme activity.

#### The Cross Talk of NO and PLD-Mediated PA Accumulation in Rice Seedlings under Cadmium Stress

It has been previously shown that PLDa was activated in the ABA response and that phosphatidylcholine (PC) was hydrolyzed to PA in leaf protoplasts labeled with fluorescent PC (Zhang et al., 2004). Our above results showed that Cd and Cd and SNAP treatments increased PLD activity (**Figures 5D,E**). To further characterize the PA change in response to Cd or Cd and SNAP, we used electrospray ionization–tandem mass spectrometry (ESI-MS/MS) to analyze PA species in rice leaves. Cd induced a gradual increase in total PA, with a maximum increase occurring at 12 h after Cd treatment (**Figure 6A**). The addition of Cd and SNAP could further enhance total PA content but the effect could be suppressed by Cd+SNAP+cPTIO treatment (**Figure 6B**). Further analysis of PA molecular species revealed the distinguishable changes after Cd or Cd and SNAP treatment. The major molecular species of PA responsive to Cd treatment in rice seedlings were 34:2 (16:0-18:2), 34:3 (16:0-18:3), 36:4(18:2-18:2), 36:5 (18:2-18:3), and 36:6 (18:3-18:3; **Figure 6C**). The level of

after 1 day of treatment (E). Values reflect means ± SEs of three independent experiments (n = 10/experiment). Different symbols above the bars indicate significant differences (Tukey's test, p < 0.05).

these PAs gradually increased after Cd treatment and reached the highest level after 12 h of treatment and decreased at 24 h after treatment (**Figure 6C**). Similarly, Cd and SNAP treatment could enhance, while Cd+SNAP+cPTIO treatment could suppress the Cd-induced increase of these PAs (**Figure 6D**). These results show that Cd and Cd and NO could induce 34:2, 34:3, 36:4, 36:5, and 36:6 PA accumulation through increased PLD activity.

### NO Reduces Cadmium-Induced H2O<sup>2</sup> Generation and Increases Glutathione Accumulation

Heavy metal treatments can induce the production of large amounts of ROS, which damages cell viability (Shahid et al., 2014). NO treatment can reduce the damaging effects of ROS (Beligni and Lamattina, 2002). We then examined whether NO treatment could inhibit cadmium toxicity in rice seedlings by increasing antioxidant levels and reducing ROS damage. We measured H2O<sup>2</sup> and glutathione levels in rice roots following cadmium or Cd and NO treatments. Cadmium exposure sharply increased H2O<sup>2</sup> production (**Figure 7A**). The addition of Cd+SNAP reduced cadmium-induced H2O<sup>2</sup> accumulation and sharply increased glutathione accumulation. This effect was abolished with Cd+SNAP+cPTIO pretreatment (**Figure 7A**). 1- Butanol (1-Bu) is the special inhibitor of PLD and suppressed PA production. We found that the addition of 1-Bu treatment increased the Cd-induced H2O<sup>2</sup> accumulation and decreased glutathione accumulation. PA produced by PLD plays an essential role in many plant physiological process, such as stomatal closure, root growth, plant tolerance to salinity and water deficits, and nitrogen deficiency stress (Zhang et al., 2009). Here, we found that Cd and PA (mainly 16:0-18:2 PA) treatment also reduced Cd-induced H2O<sup>2</sup> accumulation and enhanced glutathione accumulation. This suggests that NO and PA prevent Cd induced H2O<sup>2</sup> accumulation and increase glutathione accumulation. The glutathione-ascorbate pathway is essential for scavenging ROS in plant cells (Foyer and Noctor, 2011). APX and GR are enzymes of the glutathioneascorbate pathway. SOD also plays a role in reducing ROS overaccumulation. The addition of cadmium increased GR, SOD, and APX activities as compared to untreated plants (**Figure 7B**). We found that Cd and PA (16:0-18:2) and Cd and SNAP treatments further increased cadmium-induced activities of GR, SOD, and APX enzymes except for Cd and PA treatment for GR activity where the effect was not significant (**Figure 7B**). Additionally the Cd+SNAP+cPTIO and Cd and 1-Bu treatments reduced the cadmium-induced antioxidant enzyme activities (**Figure 7B**).

Furthermore the enzyme levels were examined using antibodies to the APX, SOD, and GR enzymes (actin was used as a control; **Figure 7C**). As compared to the control plants, levels of APX, SOD, and GR increased with Cd, Cd+PA, and Cd+SNAP treatments. This increase in protein levels was abolished by the treatments Cd+SNAP+cPTIO and Cd and 1-Bu. Thus NO and PLD are required for the Cd-induced increase in these antioxidant enzymes.

We found that Cd treatment damaged rice seedling viability, including yellowish leaves, lower Fv/Fm and chlorophyll content, shorter shoot and root lengths, and higher ion leakage. However, the addition of NO or PA to cadmium treated rice seedlings could increase total chlorophyll content in leaves and reduce the negative effects on growth (**Figures 1B,C**). Suppressing PLD activity with 1-Bu, or NO (using SNAP) with cPTIO could increase Cd induced damage on cell viability (**Figure 1A**), including increasing ion leakage and further reducing Fv/Fm (**Figure 1D**), decreasing leaf chlorophyll content (**Figure 1B**), and decreasing the stem and root length (**Figure 1C**). These data suggest that both NO and PA play a role in relieving Cd damage to rice seedlings.

### DISCUSSION

Cadmium is readily deposited into human and animal bodies after absorption and can damage the nervous system (López et al., 2006). In plants, cadmium reduces the absorption of nitrates and iron, damages photosynthetic capabilities, inhibits stomatal opening, and induces oxidative stress (Das et al., 1997; Lombi et al., 2002; Lin et al., 2007). Previous studies have primarily focused on physiological processes in the cytosol during cadmium stress in plants. The role of the plasma membrane in this process has received less attention, despite it being the initial site of cadmium sensing. We isolated and purified plasma membranes from 2-week-old rice seedlings and applied the quantitative iTRAQ method to investigate protein accumulation patterns following cadmium stress. Classification of the differentially accumulated proteins revealed these proteins were mainly membrane transporters, ATPases, and kinases and involved in catalytic function, suggesting the proteins involved in transport of ions and other molecules across membranes play an essential role during the response of rice seedlings to Cd stress.

Metal ions such as Cd disrupt the plasma membrane by binding to the proteins and lipids of the plasma membrane and can replace calcium ions in the membrane. This disruption to the plasma membrane increases non-specific membrane permeability and decreases specific transporting activities (Janicka-Russak et al., 2012). In this study, phosphate, iron, oligopeptide transporters in the plasma membrane were down-regulated whereas sucrose, monosaccharide, and nitrate transporters were up-regulated (**Table 1**). Ion homeostasis is regulated by the ATP dependent proton pump of the plasma membrane and Cd can reduce the enzyme activity of the H+ATPase (Janicka-Russak et al., 2012). H+ATPase, when in an active state, is bound to a 14-3-3 protein. In this study four H+ATPase proteins and a 14-3-3 protein in the plasma membrane were differentially regulated by Cd.

Exposure to Cd is known to increase the levels of jasmonate, abscisic acid, ethylene, auxin, and salicylic acid (Dalcorso et al., 2008; Chmielowska-B ˛ak et al., 2014). In this study an auxin transport protein REH1, two auxin efflux carrier proteins, and one gibberellin response modulator were all down-regulated after Cd treatment. This suggests that in rice seedlings that auxin or gibberellin may play a role in the Cd stress response in the plasma membrane.

In this study, six aquaporins in the plasma membrane were repressed by Cd stress. Water transportation through aquaporins might be involved in the tolerance and accumulation of Cd in pea (Pisum sativum L.; Belimov et al., 2015). The role of aquaporins in response to Cd may need further examination.

In response to Cd, ROS production can be triggered by calmodulin, protein kinases, phospholipase C, and phospholipase D (Chmielowska-B ˛ak et al., 2014). These phopholipases, through PA, activate secondary messengers such as lipid and protein kinases. Mitogen-activated protein kinases (MAPK) can affect transcription factors and thereby altering gene expression. In this study calmodulin, phospholipase C, phospholipase D, protein kinases including MAPK, and calcium dependent were differentially regulated in the plasma membrane. Based on the plasma membrane proteomic data, we propose that the rice cell initiated multiple strategies at the plasma membrane in response to Cd stress.

Previous studies showed that NO enhances the tolerance to cadmium stress in Arabidopsis and rice (Zhang et al., 2012; Yuan and Huang, 2016). PLD-dependent PA accumulation is also reported to alter plant response to salt, heat, and aluminum stress (Mishkind et al., 2009; Yu et al., 2010; Zhao et al., 2011). PLD activity is functional downstream of NO to control ABAinduced stomatal closure in Arabidopsis (Distéfano et al., 2012). Our proteomic data showed a PLD protein was upregulated substantially following cadmium treatment, and additional NO donor SNAP treatment could enhance PLD activity (**Figure 5E**). Western blotting confirmed that inhibiting NO generation reduced cadmium-induced accumulations of PLDa (**Figure 5B**), supporting the possibility of cross-talk between PLD and NO during cadmium exposure. Our physiological experiments have demonstrated that cadmium induces rapid generation of NO and the PLD product PA (**Figures 3**, **6**). When NO accumulation was reduced by the NO scavenger cPTIO, cadmium-induced PA also was reduced (**Figure 6**). Applying rice with NO donor SNAP or PA reduced Cd toxicity to rice (**Figures 1**, **2**), suggesting the critical roles of NO and PLD-mediated PA for rice tolerance to Cd stress. Interestingly, when PA was added to Cd treated seedlings, nitric oxide levels increased (**Figure 3B**) suggesting that the level of PA (via PLD) may affect the generation of NO. We also found that NO treatment promoted the translocation of GFP-PLDa from the nucleus to the plasma membrane. Such an effect could be abolished by NO scavenger cPTIO treatment. Furthermore, it is possible that NO promotes the localization of PLD to the plasma membrane, which is efficient strategy for the PLD product PA to easily bind to other important proteins. It is reported that PA may bind to NADPH oxidase to strictly controlling the ROS in Arabidopsis (Zhang et al., 2009). Thus PLD also plays a role in controlling ROS production in rice subjected to Cd stress.

ROS can induce cell death in Arabidopsis, of which, such effect could be reduced by adding PA (Zhang et al., 2003). Exogenous NO depletes Cd-induced toxicity by eliminating oxidative damage (Liu et al., 2015). Cd stress also caused the over accumulation of ROS in rice (Wang et al., 2015). Consistent with these reports, our data found that Cd treatment induced significant accumulation of H2O2, which damaged cell viability as shown by a reduction in Fv/Fm, biomass and phytochelatins. NO was rapidly induced by Cd treatment. NO also enhanced Cd-induced PLD activity and PA accumulation. Increases of NO or PA to cadmium-treated rice seedlings could increase antioxidant enzymes activities associated with glutathioneascorbate pathway, including APX and GR activity, as well as the SOD activity, and reduced Cd-induced accumulation of ROS. Suppressing NO signal by a NO scavenger or PA generation by suppressing PLD activity increased the damage by Cd to rice seedlings. It also compromised the activities of APX, GR and SOD and their proteins accumulations. We also found increasing levels of NO or PA increased the content of glutathione (**Figure 7**). Glutathione was reported to contribute to control redox homeostasis under toxic metal and metalloid stress (Hernández et al., 2015). It is also the precursor of phytochelatins, which is induced by Cd stress in our study (**Figure 2**). These data coincide with previous studies and demonstrated that both NO and PA could reduce Cd stress by enhancing antioxidant protein accumulation and enzyme activities.

Based on the proteomics and physiological results, we propose a model to explain the central roles of NO and PLD during the rice response to cadmium stress (**Figure 8**). Upon cadmium exposure, NO generation was rapidly induced to activate PLD activity and increase its accumulation, and accelerated its migration from the cell nucleus to the plasma membrane, which possibly lead to increased PA synthesis at the plasma membrane. Then PA acts as a signal to prevent the production of ROS and further induce the accumulation of antioxidant proteins and increase their activities. The accumulation of antioxidants and thus the reduction in ROS effectively protect against

cadmium-induced toxicity. Altogether, our data support the multiple roles of the plasma membrane proteins during the rice response to Cd stress, and demonstrates the essential functions of NO and PLD-mediated signaling during this process. These findings should be useful to enhance rice tolerance to Cd stress through genetic modification of NO and PLD signaling.

### CONCLUSION

In this study, a quantitative proteomic approach was applied to obtain a comprehensive proteomic description of the plasma membrane from rice seedlings in response to cadmium stress or combining cadmium and NO treatment. Among the 66 differentially regulated proteins identified in the plasma membrane of rice seedlings, the majority of proteins were enzymes involved in metabolism, transporters, ATPases, kinases, phosphatases, and phospholipases. Damage symptoms induced by Cd stress, including morphological and biomass changes can be alleviated by NO or PA application. Addition of NO or PA resulted in the accumulation of glutathione and increased the activities of ROS scavenging enzymes, to alleviate cadmiuminduced damage to rice plants. Addition of NO to Cd treated plants increased the synthesis of the Cd-responsive PA molecular species. Similarly, addition of PA to Cd treated plants increases nitrate oxide production. Taken together NO and PA serve a protective role in rice seedlings treated with Cd.

### AUTHOR CONTRIBUTIONS

LY and JJ performed the experiments, data analysis, and drafted the manuscript. KH, HW, HLW and EA assisted with data analysis, manuscript preparation, and revision. XH and YL served as the principal investigator, conceived the project, and finalized the manuscript.

### ACKNOWLEDGMENTS

This work was supported by National Science Foundation grant of China (30900871, 30971452, 31170256, 31400169), Natural Science Foundation of Jiangsu Province (BK2011409, BK20140454), the Opening Foundation of the Jiansu Key Laboratory for Eco-Agricultural Biotechnology around Hongze Lake (No.HZHL1002), Jiangsu Collaborative Innovation Center of Regional Modern Agriculture &Environment Protection (HSXT305), Jiangsu Government Scholarship for Overseas Studies and Qinglan Project of Jiangsu Province. The authors would like to extend their sincere appreciation to the Deanship of Scientific Research at King Saud University for funding this

### REFERENCES


research (Research Group NO. RG-1435-014). Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the USDA. The USDA is an equal opportunity provider and employer.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 00190

Supplemental Figure 1 | Effect of cadmium concentration on shoot and root growth of rice seedlings.

Supplemental Figure 2 | Western blot measuring the enriched plasma membrane purity using different membrane marker antibodies.

Supplemental Table 1 | The purity of the plasma membrane fraction was evaluated using marker enzymes associated with various subcellular membranes.

Supplemental Table 2 | The detailed peptide information on proteins that are differentially regulated in the plasma membrane due to cadmium or cadmium and SNAP treatment.


different ecotypes of the hyperaccumulator Thlaspi caerulescens. Plant Physiol. 128, 1359–1367. doi: 10.1104/pp.010731


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Yang, Ji, Harris-Shultz, Wang, Wang, Abd-Allah, Luo and Hu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Alterations in Kernel Proteome after Infection with *Fusarium culmorum* in Two Triticale Cultivars with Contrasting Resistance to *Fusarium* Head Blight

#### *Edited by:*

Dipanjana Ghosh, National University of Singapore, Singapore

#### *Reviewed by:*

Sebastien Carpentier, Katholieke Universiteit Leuven, Belgium Sun Tae Kim, Pusan National University, South Korea

#### *\*Correspondence:*

Arkadiusz Kosmala akos@igr.poznan.pl

† These authors have contributed equally to this work.

#### *Specialty section:*

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

*Received:* 26 April 2016 *Accepted:* 02 August 2016 *Published:* 17 August 2016

#### *Citation:*

Perlikowski D, Wisniewska H, ´ Kaczmarek J, Góral T, Ochodzki P, Kwiatek M, Majka M, Augustyniak A and Kosmala A (2016) Alterations in Kernel Proteome after Infection with Fusarium culmorum in Two Triticale Cultivars with Contrasting Resistance to Fusarium Head Blight. Front. Plant Sci. 7:1217. doi: 10.3389/fpls.2016.01217 Dawid Perlikowski 1 †, Halina Wisniewska ´ 1 †, Joanna Kaczmarek <sup>1</sup> , Tomasz Góral <sup>2</sup> , Piotr Ochodzki <sup>2</sup> , Michał Kwiatek <sup>1</sup> , Maciej Majka<sup>1</sup> , Adam Augustyniak <sup>1</sup> and Arkadiusz Kosmala<sup>1</sup> \*

1 Institute of Plant Genetics, Polish Academy of Sciences, Poznan, Poland, <sup>2</sup> Plant Breeding and Acclimatization Institute—National Research Institute, Blonie, Poland

Highlight: The level of pathogen alpha-amylase and plant beta-amylase activities could be components of plant-pathogen interaction associated with the resistance of triticale to Fusarium head blight.

Triticale was used here as a model to recognize new components of molecular mechanism of resistance to Fusarium head blight (FHB) in cereals. Fusarium-damaged kernels (FDK) of two lines distinct in levels of resistance to FHB were applied into a proteome profiling using two-dimensional gel electrophoresis (2-DE) to create protein maps and mass spectrometry (MS) to identify the proteins differentially accumulated between the analyzed lines. This proteomic research was supported by a measurement of alpha- and beta-amylase activities, mycotoxin content, and fungal biomass in the analyzed kernels. The 2-DE analysis indicated a total of 23 spots with clear differences in a protein content between the more resistant and more susceptible triticale lines after infection with Fusarium culmorum. A majority of the proteins were involved in a cell carbohydrate metabolism, stressing the importance of this protein group in a plant response to Fusarium infection. The increased accumulation levels of different isoforms of plant beta-amylase were observed for a more susceptible triticale line after inoculation but these were not supported by a total level of beta-amylase activity, showing the highest value in the control conditions. The more resistant line was characterized by a higher abundance of alpha-amylase inhibitor CM2 subunit and simultaneously a lower activity of alpha-amylase after inoculation. We suggest that the level of pathogen alpha-amylase and plant beta-amylase activities could be components of plant-pathogen interaction associated with the resistance of triticale to FHB.

Keywords: amylase, cereals, FHB, *Fusarium*, inhibitors, mycotoxins, proteome

### INTRODUCTION

Fusarium head blight (FHB) is a serious plant disease resulting in a significant reduction of kernel quality and yield in small grain cereals. This head infection is caused by several widespread necrotrophic mycotoxigenic fungi of Fusarium genus: F. culmorum (W.G. Smith.), F. avenaceum (Corda ex Fries) Sacc., and F. graminearum (Schwabe; Bottalico and Perrone, 2002). The disease symptoms are mainly a result of contamination of Fusarium-damaged kernels (FDK) with toxic fungal secondary metabolites (mycotoxins), including e.g., zearalenone (ZEA) and trichothecene B toxins such as deoxynivalenol (DON), nivalenol (NIV), and DON derivatives— 3-acetyldexynivalenol (3-AcDON) and 15-acetyldeoxynivalenol (15AcDON; Bottalico and Perrone, 2002; Chakraborty et al., 2006; Buerstmayr et al., 2009; Marin et al., 2013). The Fusarium species are harmful mostly to bread wheat (Triticum aestivum L.), durum wheat (T. durum Desf.), maize (Zea mays L.), triticale (×Triticosecale Wittm.), oat (Avena sativa L.), and rice (Oryza sativa L.; Miedaner et al., 2001; Langevin et al., 2004). The selection of cereal genotypes with improved resistance to FHB is a relatively difficult process since the resistance is a quantitative trait governed by genetic factors located both in the host plant and pathogen, and also by environmental conditions, particularly temperature and rainfall, from flowering to the soft-doughstage of kernel development (Mesterhazy, 1995; Miedaner, 1997; Chełkowski et al., 2000; Snijders, 2004; Cowger et al., 2009). The Quantitative Trait Loci (QTLs) controlling resistance to FHB have been identified in wheat on most chromosomes (Buerstmayr et al., 2009). The QTL with the largest effect was located on 3B chromosome (Qfhs.ndsu-3BS) in the Chinese wheat cultivar Sumai 3 and it was shown to be associated with the FHB resistance gene Fhb1 (Cuthbert et al., 2006). The other QTLs were also mapped and named e.g., Fhb2, Fhb4, Fhb5 (Cuthbert et al., 2007; Xue et al., 2010, 2011), however, these chromosomal regions confer only partial resistance (Bai and Shaner, 2004). To date, the resistance to FHB was classified into five types including resistance against initial infection (I type), resistance to Fusarium spread within the spike (II type), resistance to kernel infection (III type), tolerance to FHB and toxins (IV type), and resistance to toxin accumulation (chemical modification or synthesis inhibition; V type; Mesterhazy, 1995; Boutigny et al., 2008; Foroud and Eudes, 2009).

Triticale has been obtained by crossing of hexaploid or tetraploid wheat as a female parent with diploid rye (Secale cereale L.) as a male parent (Cichy et al., 2002; Oettler, 2005). This intergeneric, man-made hybrid combines the complementary traits of both parental species, high yielding capacity of wheat and stress tolerance of rye, however, there are only limited reports concerning resistance to FHB in triticale (Miedaner et al., 2001, 2004; Góral et al., 2002; Góral and Ochodzki, 2007). Although, this species is thought to be less susceptible to FHB, compared to wheat and more susceptible, compared to rye (Arseniuk et al., 1993; Langevin et al., 2004), the other studies revealed that susceptibility to FHB in triticale may be equal to that observed in wheat or even exceeds it (Miedaner et al., 2001; Langevin et al., 2004; Comeau et al., 2008; Veitch et al., 2008). Recently, the report on QTLs associated with FHB resistance in triticale has been published (Kalih et al., 2015) but the molecular nature of this trait still remains unrecognized in detail. Taking into account a genomic constitution of triticale and its origin, this hybrid could be a good model for the other cereals to recognize new crucial components of resistance to FHB not revealed to date in the parental species.

Proteome profiling to investigate mechanisms of resistance to FHB in cereals has been shown before e.g., in wheat (e.g., Zhou et al., 2005, 2006; Eggert et al., 2011) and barley (Hordeum vulgare L.; Yang et al., 2010a,b; Eggert and Pawelzik, 2011) infected with F. culmorum or/and F. graminearum (for review: Yang et al., 2013). Also our earlier work on winter wheat infected with F. culmorum could be treated as a good example of such research (Perlikowski et al., 2014). The aspect of proteomic approach to recognize in cereals markers associated with their resistance to the selected biotic stresses has been recently reviewed by Kosová et al. (2014).

Here, we demonstrate the first proteomic research for triticale, including: (1) the analysis of protein abundance in the FDK of two lines, more susceptible and more resistant to FHB using two-dimensional gel electrophoresis (2-DE) and (2) mass spectrometry (MS) to identify differentially accumulated proteins. The proteome screening was followed by the alphaand beta-amylase activity assays to reveal a potential involvement of these enzymes into the resistance of triticale to FHB. This proteomic research was supported by the evaluation of fungal biomass as well as mycotoxin content in the analyzed kernels.

### MATERIALS AND METHODS

### Field Experiments

The scientific approach for field experiments was similar to that described previously for wheat by Perlikowski et al. (2014). The plant materials for the research reported here were two lines of hexaploid triticale (×Triticosecale Wittm.)—DS 9, a line with a relatively high level of resistance to FHB (RL) and DANKO 1, a line with a relatively high level of susceptibility (SL), both developed by Danko Plant Breeding Ltd., Co. (Poland). The resistance levels of the analyzed triticale lines were estimated in 2014, in two locations under the field conditions: Cerekwica (western Poland; GPS coordinates: N 52.521012, E 16.692005) characterized by poor, sandy-clay soil and Radzikow (central Poland; GPS coordinates: N 52.211754, E 20.631954) with rich sandy-clay soil. The rainfalls and mean temperature during the experiments performed in Cerekwica and Radzikow, are presented in Table S1. The experiments in both locations were carried out according to the same design. The experimental field in each location consisted of four plots for each tested line. The seeds were sown in plots of 1 m<sup>2</sup> size with the sowing rate 300 seeds (September, 2013). The fungal material for inoculation was a mixture of three isolates of F. culmorum (W.G.Sacc.): KF 846 (DON chemotype) and KF 350 (NIV chemotype) derived from the collection of Institute of Plant Genetics, Polish Academy of Sciences (Poznan, Poland), and ZFR 112, producing zearalenone (ZEA), derived from the collection of Plant Breeding and Acclimatization Institute—National Research Institute (Radzikow, Poland; Snijders and Perkowski, 1990; Wi´sniewska and Kowalczyk, 2005). In the case of each analyzed line, one plot, used as a control, was not treated with Fusarium isolates, and the flowering triticale heads at three other field plots were sprayed with the spore suspension at a rate of ∼100 ml m−<sup>2</sup> (May, 2014). The developmental stage of the heads was 65, in commonly used BBCH scale, which means: full flowering and 50% of anthers mature. The concentration of conidia was adjusted to 5 × 10<sup>4</sup> ml−<sup>1</sup> . A micro-irrigation was applied during 2 days after inoculation, and after next 15 days, a progress of the disease was evaluated visually. The percentage of heads infected per plot and percentage of head infection were determined. The FHB index (FHBi), associated with the resistance type I and II (Mesterhazy, 1995), was calculated, separately for each line and location, according to the formula:

#### FHBi (%) = (% of head infection × % of heads infected per plot)/100

In August 2014, 20 randomly selected heads from each experimental plot for the RL and the SL, were threshed manually. Kernels were visually scored and divided into two categories: Healthy-looking kernels (HLK) and FDK. Kernel weight [g] and number were estimated. Percentage of FDK (% FDK) was calculated as a proportion of infected kernels per sample. Mean values and standard deviations of this parameter calculated on the base of three inoculated plots were shown in the paper, separately for each location and line. The % FDK is associated with the type III FHB resistance (Mesterhazy, 1995). Total kernel number and weight per head from 20 randomly selected heads from one control plot, separately for each line and location, were also estimated.

Analysis of variance in FHBi, FDK (weight and number), a total kernel number per head and a total kernel weight per head was performed using the ANOVA procedure of XLSTAT (Microsoft <sup>R</sup> Excel 2010/XLSTAT©-Ecology Version 2016.02.28540, Addinsoft, Inc., Brooklyn, NY, USA). Multiple comparisons of means of lines in locations were performed using Fisher (LSD) test.

#### Mycotoxin Analysis

The accumulation level of trichothecene B in the kernels of the RL and the SL triticale lines was evaluated. This analysis involves deoxynivalenol, nivalenol, 3-acetyldexynivalenol, and 15-acetyldeoxynivalenol. Additionally, the level of zearalenone was also estimated.

#### Trichthecenes B

The amount of 5 g of the ground sample was placed in a conical 50 ml Falcon centrifuge tube and then 25 ml of the solvent (acetonitrile-water 84:16 v/v) was added. The sample was extracted for 2 h on a shaker and then centrifuged (1620 g, 5 min). A clear extract volume of 6 ml was purified on Trich 227 SPE column (Multisep <sup>R</sup> 227 Trich+, Romer Labs, <sup>R</sup> Inc. Union, Mo, USA) and 4 ml of purified extract was transferred to vial, 100µl of internal standard (chloralose 10µg/ml) was added and then evaporated to dryness in heating block under stream of nitrogen in 40◦C. To the dried extract in 4 ml vial 75µl of Sylon BTZ [BSA (N,O-Bis(trimethylsilyl)acetamide]:TMCS (Chlorotrimethylsilane):TMSI [1-(Trimethylsilyl) imidazole), 3:2:3] (Supelco, Bellefonte, USA) was added and heated for 30 min in 60◦C in oven. After cooling sample was dissolved in 1 ml of isooctane, the excess of silylating agent was removed by washing twice with 1 ml of distilled water and organic layer was transferred to 2 ml autosampler vial. Chromatographic analysis was conducted on SRI 8610C gas chromatograph (SRI Instruments, Torrance, USA) equipped with electron-capture detector (ECD) and capillary column BGB-5MS, 30 m, 0.25 mm, 0.2µm film thickness. Hydrogen was the carrier gas at constant pressure and nitrogen was the make-up gas, constant flow 60 ml/min was applied. Injector temperature was established at 250◦C and detector temperature at 320◦C. A sample volume of 1µl was injected in splitless mode. The initial temperature of column was 170◦C, held for 2 min, then increased by 5◦C/min to 245◦C, held for 2 min, increased again by 25◦C/min to 300◦C, and held finally for 7 min. The content of each toxin was expressed as toxin weight [mg] per kernel weight [kg].

#### Zearalenone

A 5 g of the ground sample was placed in a conical 50 ml Falcon centrifuge tube and then 25 ml of the solvent (methanol-water 70:30 v/v) was added. The sample was extracted for 1 h on a shaker and then centrifuged (1620 g, 5 min). The obtain extract was analyzed with ELISA method according to the procedure described by RomerLabs, Agraquant (http://shop.romerlabs.com/en/AgraQuant-ELISA/ AgraQuant-Mycotoxins). The content of zearalenone was expressed as toxin weight [mg] per kernel weight [kg].

### Evaluation of Pathogen Biomass in the Kernels

To produce mycelium KF 846 isolate of F. culmorum was maintained during 5 days on Potato Dextrose Agar plates (PDA, Sigma, UK) with streptomycin (100µg/ml) at room temperature. Fungal tissue was lyophilized, frozen in liquid nitrogen, and ground to a fine powder. Further, it was mixed with F. culmorumfree ground triticale kernels in 10-fold dilution series from 10 to 0.001 mg/g (Horevaj et al., 2011). DNA was extracted from each serial dilution and 0.5 g triticale kernels of the RL and SL lines using the CTAB (cetyltrimethyl ammonium bromide). The samples were suspended in 650µl CTAB and incubated at <sup>65</sup>◦C for 20 min. The volume of 500µL CHCl<sup>3</sup> was added and centrifuged at 12 879 g for 15 min. DNA was precipitated with 65µl 3M sodium acetate, pH 5.4, and two volumes of ice cold 99.8% ethanol. The tubes were stored at −20◦C overnight and centrifuged at 17 530 g for 5 min. The pellets were washed with 70% ethanol, centrifuged at 17,530 g for 5 min and fully dissolved in 100µl of TE buffer.

The real-time PCR was performed in 10µl containing 7.5µl AmpliQ Real-Time PCR Opti Probe Kit (Novazym, Poznan, ´ Poland), 100 nM of FAM-labeled probe and 300 nM of forward and reverse F. culmorum primers (Waalwijk et al., 2004). Thermal cycling parameters for a quantitative fungal DNA detection were: 95◦C for 2 min followed by 40 cycles of 95◦C for 15 s and 60◦C for 30 s. Nuclease-free water was used as the no-template control. A standard curve was generated by plotting the C<sup>t</sup> value for each sample of standard series of the amount of fungal biomass (10–0.001 mg/g). All the samples were tested in triplicate.

Analysis of variance in Fusarium biomass was performed using the ANOVA procedure of XLSTAT (Microsoft <sup>R</sup> Excel 2010/XLSTAT©-Ecology Version 2016.02.28540, Addinsoft, Inc., Brooklyn, NY, USA). Multiple comparisons of means of lines in locations were performed using Fisher (LSD) test.

#### Proteome Profiling and Identification of Differentially Accumulated Proteins

The plant materials derived from one location (Cerekwica) were used for further molecular research. The FDK derived from 20 heads were pooled, separately for each inoculated plot, giving three separate pooled samples (bulk flour) for each analyzed line, the RL and the SL. The kernels derived from 20 heads of the control plot were also pooled for each analyzed line. The pooled samples (bulk flour) were used for proteomic research—each one in two technical replicates. A diagram outlining the workflow of sample preparation for proteome analysis is shown in the **Figure 1**.

The proteomic protocol used, including 2-DE and MS to identify differentially accumulated kernel proteins between the RL and SL of triticale, was the same as that described in detail by Perlikowski et al. (2014). Proteins were extracted as described by Hurkman and Tanaka (1986) and their concentration in samples estimated using 2-D Quant Kit (GE Healthcare, Buckinghamshire, UK). In isoelectrofocusing (IEF), strip gels with linear pH range 4–7 (24 cm) were used to focus 500µg of proteins extracted from 25 mg of triticale flour. This pH range was selected on the basis of our earlier work on wheat and our preliminary proteome screening in triticale. It was shown to be a good compromise between gel quality, its resolution, and spot numbers. In the second dimension (sodium dodecyl sulfatepolyacrylamide gel electrophoresis) the proteins were separated using 13% polyacrylamide gels (1.5 × 255 × 196 mm). The gels were stained with colloidal coomassie brilliant blue G-250 as described by Neuhoff et al. (1988), scanned by Image scanner III (GE Healthcare, Buckinghamshire, UK) and subjected to Labscan 6.0 program (GE Healthcare, Buckinghamshire, UK) processing. The image analysis was performed with Image Master 2-D Platinum software (GE Healthcare, Buckinghamshire, UK). The abundance of each protein spot was normalized as a relative volume (% Vol) and calculated as a ratio of the volume of particular spot to the total volume of all the spots present on the gel. The spot had to be detected in all the replicates to be consider as "present." The significance of the differences was assessed using Kolmogorov–Smirnov Test (three biological replicates, each one with two technical replicates used as means). The protein spots which showed at least two fold differences in protein content between two analyzed lines were analyzed by liquid chromatography coupled to the Orbitrap Velos type mass spectrometer (Thermo Fisher, Waltham, MA,


USA), working in the regime of data dependent MS to MS/MS switch.

The MS analysis was performed in the Laboratory of Mass Spectrometry, Institute of Biochemistry and Biophysics, Polish Academy of Sciences (Warsaw, Poland) as shown earlier by Kosmala et al. (2012) and Perlikowski et al. (2014). The data was analyzed with Mascot Distiller software (version 2.3, MatrixScience London, UK) with standard settings for the Orbitrap low resolution measurements (available at http:// www.matrixscience.com/distiller.html) to extract MS/MS peak-lists from the raw files. The obtained fragmentation spectra were matched to the National Center for Biotechnology Information (NCBI) non-redundant database (57412064 sequences; 20591031683 residues), with a Viridiplantae filter (2874321 sequences) using the Mascot search engine (Mascot Daemon v. 2.3.0, Mascot Server v. 2.4.0, MatrixScience, London, UK). The search parameters were the same as those described in details by Perlikowski et al. (2014). The MS proteomics data has been deposited to the ProteomeXchange Consortium via the PRIDE (https://www.ebi.ac.uk/pride/ archive/) partner repository with the dataset identifier PXD004464.

#### Alpha- and Beta-Amylase Activity Assays

Alpha-amylase activity in triticale kernels was evaluated using the Ceralpha α-Amylase Assay Kit (Megazyme International Ireland Inc., Bray, Ireland) as described in our earlier work on wheat (Perlikowski et al., 2014).

Beta-amylase activity was tested using the "Betamyl-3 <sup>R</sup> method" Assay Kit (Megazyme International Ireland Ltd., Bray, Ireland). One unit of activity was defined as the amount of enzyme, in the presence of excess thermostable β-glucosidase, required to release one micromole of p-nitrophenol from p-nitrophenyl-β-D-maltotrioside in 1 min under the defined assay conditions.

Three biological and two technical replicates were used (**Figure 1**). Each technical replicate contain a flour in an amount of 0.5 g. The enzyme activity was shown in Ceralpha Units (CU) per gram of flour and the significance of differences between the RL and SL was assessed using ANOVA (p ≤ 0.05).

### RESULTS AND DISCUSSION

#### Field Experiments and Mycotoxin Analysis

The two analyzed triticale lines were revealed to have significantly different levels of resistance to FHB. This phenomenon was manifested by the values of FHBi in Cerekwica and % FDK in two locations (**Table 1**). These triticale lines were also significantly different with respect to a mycotoxin content (**Table 2**). In the control conditions the resistant line showed a higher yield level, compared to the susceptible line (**Table 1**). The differences revealed for both locations could be a result of different soil quality and weather conditions (Table S1).

#### Pathogen Biomass in the Kernels

The differences in the levels of resistance to FHB between the triticale lines were also manifested by the values of fungal biomass in the analyzed kernels. Consistently with the visual assessments of FHB, the line with a higher disease index (SL) revealed simultaneously a higher amount of fungal biomass assessed by qPCR. No F. culmorum tissue was detected in the kernels collected in the control conditions (**Table 1**).

#### Proteome Profiles and Identities of Differentially Accumulated Proteins

The comparative analyses indicated a total of 23 spots that showed significant differences in a protein abundance between the more resistant and more susceptible triticale lines after infection (**Figures 2**, **3** and Figures S1, S2), including 16 spots with a significantly higher protein abundance in the more susceptible line (spots no. 1–16) and seven spots with a significantly higher abundance in the more resistant line (spot no. 17–23). All the selected protein spots were subjected to MS identification (**Table 3**) and in all the cases these selections were identified as homologs of proteins from related plant species (**Table 3**). A majority of the identified proteins (according to UniProt categories; www.uniprot.org) were involved in a cell carbohydrate metabolism with 10 proteins highly accumulated in the SL (spots no. 1, 2, 4, 5, 6, 7, 8, 9, 10, and 15) and two in the RL (spots no. 20 and 22), stressing the importance of this protein group in plant response to Fusarium inoculation. The relevance of the regulation of carbohydrate metabolism for

TABLE 1 | The components of the resistance to *Fusarium* head blight in the more resistant (RL) and more susceptible (SL) triticale lines and their yields under control conditions.


FHBi, Fusarium head blight index; FDK, Fusarium-damaged kernels; RL, more resistant line; SL, more susceptible line; mean values and standard deviations of each parameter calculated after inoculation (three plots) and data from one plot calculated for the control conditions, are shown. Values marked with the same letter are not significantly different according to Fisher (LSD) test at p = 0.05.

TABLE 2 | The toxin content in the kernels of the more resistant (RL) and more susceptible (SL) triticale lines.


RL, more resistant line; SL, more susceptible line.

plant-pathogen integrations was postulated earlier (for review: Berger et al., 2007). The other identified proteins were involved in an amino acid metabolism (spots no. 3, 11, and 12), a protein biosynthesis and folding (spots no. 13 and 21), a nutrient storage (spot no. 18), a cell redox homeostasis (spot no. 19), a detoxification system (spot no. 17), a mitochondrial electron transport (spot no. 14), and a RNA processing (spot no. 16). The earlier studies performed on other cereal species, revealed also that fungal infection is followed by alterations in plant proteome. In hexaploid wheat infected with F. graminearum, 15 proteins were induced or up-regulated. Among them, the antioxidants, such as superoxidase dismutase and glutathione S-transferase, as well as pathogenesis-related proteins, such as beta-1, 3 glucanase, were detected (Zhou et al., 2005). In T. dicoccum infected with the same fungal species, 10 proteins changed abundance, including globulin-2 (3) and beta-amylase, identified also in our present study (Eggert et al., 2011). The potential involvement of amylases into plant-fungal interactions and mechanisms of resistance to FHB was suggested earlier e.g., for H. vulgare (Yang et al., 2010b) and Triticum species (Packa et al., 2013; Perlikowski et al., 2014). The relationship between amylase activities and resistance to FHB exists also, to a high probability, in triticale, as indicated by our first study regarding this matter.

### Alpha- and Beta-Amylase Activity Could be a Component of the Susceptibility to FHB in Triticale

Natural kernel sprouting requires starch decomposition by plant alpha- and beta-amylases (Lunn et al., 2001). However, Fusarium pathogens can also use plant or their own hydrolytic enzymes to colonize kernels (Wang et al., 2005). The triticale beta-amylase, which significantly accumulated in the SL samples, was identified in six spots (spots no. 1–2, 4–6, and 9; Figure S2). The isoform present in spots no. 1, 2, and 5 possesses 94% of amino acid sequence identity with the isoform present in spots no. 4, 6, and 9 (Figure S3). The particular proteins identified as amylase isoforms might vary in post-translation modifications, resulting in different isoelectric points and molecular weights affecting spot positions in the 2-D gels (**Figures 2**, **3** and **Table 3**). Here, the increased accumulation levels of different isoforms of betaamylase were observed for the SL triticale line after inoculation (spots no. 1–2, 4–6, and 9) and also in the control conditions (spots no. 6 and 9). However, these elevated accumulation levels were not followed by a total level of activity of that enzyme

FIGURE 2 | One representative 2-DE protein map of triticale kernel after *Fusarium culmorum* infection (*Fusarium*-damaged kernels) for the line more susceptible (SL) to *Fusarium* head blight. The spots with differentially accumulated proteins (1–16) identified in the SL, are circled with a solid line.

observed after inoculation. The activity was shown to be a slightly higher in the RL and, furthermore, comparable to the level observed in the control conditions in that line (**Figure 4A**). Thus, the level of beta-amylase activity did not change after infection with Fusarium in the resistant line. On the other hand, the susceptible line in the control conditions revealed a higher activity level of beta-amylase, compared to the RL (**Figure 4A**), indicating simultaneously a higher level of starch degradation in this line before inoculation. After inoculation the enzyme activity



<sup>a</sup>Spot numbering was the same as in *Figures 2, 3*.

<sup>b</sup>Database accession (according to NCBInr) of a homologous protein.

<sup>c</sup>Homologous protein and organism from which it originates.

<sup>d</sup>Mascot MudPIT (Multidimensional Protein Identification Technology) score.

<sup>e</sup>Amino acid sequence coverage for the identified proteins; amino acid sequences for the proteins were shown in Figure S3.

<sup>f</sup> Theoretical molecular weight and isoelectric point revealed by Mascot software.

<sup>g</sup>Experimental molecular weight and isoelectric point calculated based on 2-D protein maps.

was significantly reduced. This phenomenon could explain to a certain degree a higher level of susceptibility of DANKO 1 line to F. culmorum before infection. Interestingly, as mentioned earlier, two isoforms of beta-amylase (spots no. 6 and 9) showed also higher accumulation levels in the SL in the control conditions. A discrepancy between accumulation and activity levels of betaamylase after inoculation has not been explained here in detail. However, it is highly probable that the accumulated isoforms can play also a role in the other metabolic pathways.

It was shown earlier that in field conditions the inoculation with F. culmorum led to an increased alpha-amylase activity in kernels of T. monococum, T. dicoccum, and T. aestivum (Packa et al., 2013; Perlikowski et al., 2014). Our earlier work on winter wheat showed that the alpha-amylase activity level was lower in the wheat RL, both in the control conditions and after inoculation, and we suggested that this could be due to the presence of monomeric alpha-amylase and dimeric alphaamylase inhibitors, accumulated to a higher degree in the wheat line with a higher resistance to FHB (Perlikowski et al., 2014). Here, a higher accumulation level of alpha-amylase inhibitor CM2 subunit (spot no. 22) both in the control conditions and after inoculation was also detected in the RL triticale line (Figure S2). These results were supported by a lower level of alpha-amylase activity after Fusarium inoculation in the RL (**Figure 4B**). Tetrameric, CM amylase inhibitors are generally composed of one CM1 or CM2 subunit, plus one CM16 or CM17 subunit and plus two CM3 subunits. The inhibitory activity of the protein is dependent on the combination of subunits, however, it was proved to be active against different pathogen alpha-amylases, but not against cereal enzymes (Altenbach et al.,

2011). Plant and pathogen alpha-amylase activities were not precisely distinguished here, however, it is highly probable that this amylase activity had its source in both organisms. In the control conditions, when the kernels lacked pathogen biomass (**Table 1**), this activity could have been fully an attribute of triticale. On the other hand, after inoculation, it could have been mostly an attribute of Fusarium, especially in the SL. The resistant line revealed after inoculation the level of alphaamylase activity comparable to that observed in the control conditions (**Figure 4B**). Thus, we suggest here that the increased accumulation and activity levels of that enzyme after inoculation in the SL could be a result of its production by Fusarium and this phenomenon simultaneously could improve pathogen propagation. This might be possible to a certain degree because of less effective alpha-amylase inhibitory system present in the SL.

#### CONCLUSIONS

Although, the infection of triticale with F. culmorum resulted in abundance alterations of different proteins, the group associated

#### REFERENCES


with carbohydrate metabolism was revealed to be the most numerous. The majority of identified proteins in that group were the components of cell amylase machinery, including plant alpha-amylase inhibitor and isoforms of plant betaamylases. The plant alpha-amylase inhibitors were proved earlier to be the important components of the active resistance of plants to necrotrophic pathogens (Svensson et al., 2004). Thus, the inhibition of pathogen alpha-amylase activity, observed in our study, could also prevent infection progress in the analyzed here more resistant triticale line, however, next experiments are required using more cultivars, locations and different environmental conditions. The similar phenomenon was observed earlier in winter wheat (Perlikowski et al., 2014). Moreover, the activity level of plant beta-amylase before Fusarium inoculation in triticale could be responsible, at least partially, for a different susceptibility of the analyzed lines to a pathogen infection. Further research is required to go deeper into the mechanisms of cereals' resistance to FHB, including the involvement of particular amylases into that process. The next experiments should involve work on the other cereals, including rye and other proteomic methods based on fluorescent dyes and gel free protein quantifications.

#### AUTHOR CONTRIBUTIONS

AK and HW conceived and designed the experiments. DP, HW, JK, TG, PO, AA, and AK performed the experiments. DP, HW, JK, TG, MK, MM, and AK analyzed the data. HW, TG, JK, PO, and AK contributed reagents/materials/analysis tools. AK wrote the first version of the manuscript. All the authors read and approved the manuscript.

#### ACKNOWLEDGMENTS

The study was supported by the Polish Ministry of Agriculture and Rural Development (no. 14, 2015). The equipment used for MS was sponsored in part by the Centre for Preclinical Research and Technology (CePT), a project co-sponsored by European Regional Development Fund and Innovative Economy, The National Cohesion Strategy of Poland.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 01217


development in six cereal species. Eur. J. Plant Pathol. 110, 735–746. doi: 10.1023/B:EJPP.0000041568.31778.ad


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Perlikowski, Wi´sniewska, Kaczmarek, Góral, Ochodzki, Kwiatek, Majka, Augustyniak and Kosmala. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Changes in the Proteome of Xylem Sap in *Brassica oleracea* in Response to *Fusarium oxysporum* Stress

Zijing Pu<sup>1</sup> , Yoko Ino<sup>2</sup> , Yayoi Kimura<sup>2</sup> , Asumi Tago<sup>1</sup> , Motoki Shimizu1, 3, Satoshi Natsume<sup>3</sup> , Yoshitaka Sano<sup>1</sup> , Ryo Fujimoto<sup>4</sup> , Kentaro Kaneko<sup>1</sup> , Daniel J. Shea<sup>1</sup> , Eigo Fukai <sup>1</sup> , Shin-Ichi Fuji <sup>5</sup> , Hisashi Hirano<sup>2</sup> and Keiichi Okazaki <sup>1</sup> \*

<sup>1</sup> Graduate School of Science and Technology, Niigata University, Niigata, Japan, <sup>2</sup> Advanced Medical Research Center, Yokohama City University, Kanazawa, Japan, <sup>3</sup> Iwate Biotechnology Research Center, Kitakami, Japan, <sup>4</sup> Graduate School of Agricultural Science, Kobe University, Kobe, Japan, <sup>5</sup> Faculty of Bioresource Sciences, Akita Prefectural University, Akita, Japan

Fusarium oxysporum f.sp. conlutinans (Foc) is a serious root-invading and xylem-colonizing fungus that causes yellowing in Brassica oleracea. To comprehensively understand the interaction between F. oxysporum and B. oleracea, composition of the xylem sap proteome of the non-infected and Foc-infected plants was investigated in both resistant and susceptible cultivars using liquid chromatography-tandem mass spectrometry (LC-MS/MS) after in-solution digestion of xylem sap proteins. Whole genome sequencing of Foc was carried out and generated a predicted Foc protein database. The predicted Foc protein database was then combined with the public B. oleracea and B. rapa protein databases downloaded from Uniprot and used for protein identification. About 200 plant proteins were identified in the xylem sap of susceptible and resistant plants. Comparison between the non-infected and Foc-infected samples revealed that Foc infection causes changes to the protein composition in B. oleracea xylem sap where repressed proteins accounted for a greater proportion than those of induced in both the susceptible and resistant reactions. The analysis on the proteins with concentration change >= 2-fold indicated a large portion of up- and down-regulated proteins were those acting on carbohydrates. Proteins with leucine-rich repeats and legume lectin domains were mainly induced in both resistant and susceptible system, so was the case of thaumatins. Twenty-five Foc proteins were identified in the infected xylem sap and 10 of them were cysteine-containing secreted small proteins that are good candidates for virulence and/or avirulence effectors. The findings of differential response of protein contents in the xylem sap between the non-infected and Foc-infected samples as well as the Foc candidate effectors secreted in xylem provide valuable insights into B. oleracea-Foc interactions.

Keywords: *Brassica oleracea*, *F. oxysporum* f. sp. *conglutinans*, xylem sap, proteomics

#### INTRODUCTION

The xylem vessel is formed via the programmed death of xylem tracheary elements, followed by connection of the elements to long tubes. The important function of xylem is transporting water and minerals from the root to the aerial tissues of the plant (De Boer and Volkov, 2003). Furthermore, the observed macromolecules in the xylem sap, such as carbohydrates and proteins,

#### *Edited by:*

Hanjo A. Hellmann, Washington State University, USA

#### *Reviewed by:*

Letizia Bernardo, Università Cattolica del Sacro Cuore, Italy Reinhard Turetschek, University of Vienna, Austria

> *\*Correspondence:* Keiichi Okazaki okazaki@agr.niigata-u.ac.jp

#### *Specialty section:*

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

*Received:* 17 September 2015 *Accepted:* 10 January 2016 *Published:* 01 February 2016

#### *Citation:*

Pu Z, Ino Y, Kimura Y, Tago A, Shimizu M, Natsume S, Sano Y, Fujimoto R, Kaneko K, Shea DJ, Fukai E, Fuji S-I, Hirano H and Okazaki K (2016) Changes in the Proteome of Xylem Sap in Brassica oleracea in Response to Fusarium oxysporum Stress. Front. Plant Sci. 7:31. doi: 10.3389/fpls.2016.00031 are thought to have function in the response against biotic and abiotic stresses (Biles and Abeles, 1991; Satoh et al., 1992; Masuda et al., 1999; Sakuta and Satoh, 2000).

Xylem sap proteins come from two pathways; when xylem elements are dead, the protoplast is autolyzed and some of the proteins are released into the xylem. Since the differentiated xylem cells are not able to produce proteins by themselves, after completion of the xylem vessel, most xylem sap proteins are secreted from the stele cell in the root (Ligat et al., 2011). Earlier studies reported relatively abundant xylem sap proteins such as chitinase, peroxidase, and β-1,3-glucanases in cucumber (Masuda et al., 1999; Sakuta and Satoh, 2000) and tomato (Rep et al., 2002), using one-dimensional gel electrophoresis (1-DE). Subsequently, hundreds of proteins were detected in xylem sap in different species using 2-DE and high-throughput proteomics techniques (Alvarez et al., 2006; Djordjevic et al., 2007; Aki et al., 2008; Floerl et al., 2008; Dafoe and Constabel, 2009; Fernandez-Garcia et al., 2011; Ligat et al., 2011; Zhang et al., 2015). Ligat et al. (2011), in their study using liquid chromatographytandem mass spectrometry (LC-MS/MS) and Brassica EST and cDNA sequences, classified 189 B. oleracea xylem sap proteins into eight of the nine functional classes previously defined for Arabidopsis thaliana cell wall proteins; most of them belong to those acting on carbohydrates (e.g., β-1,3-glucanases and chitinase), oxido-reducatses and proteases (29.2, 23.8, and 17.1%, respectively), which are also abundant/dominating protein classes in other plant xylem saps. Other small protein classes reported in Ligat et al. (2011), such as proteins related to lipid metabolism (4.9%), proteins with domains interacting with carbohydrates or proteins (e.g., lectin and protease inhibitor) (4.9%), miscellaneous proteins (8.5%) and proteins involved in signaling (5.5%), have also been described in Glycine max, B. napus, and Oryza sativa (Kehr et al., 2005; Djordjevic et al., 2007; Aki et al., 2008).

Many of xylem sap proteins contribute to a plant defense reaction, and the common presence of defense related proteins in plant xylem indicates the important role of xylem sap proteins in response to biotic and abiotic stresses in plants. Several xylem sap proteomics studies have suggested infection of pathogens induces some of the plant pathogenesis-related (PR) proteins. In oilseed rape, infection of Verticillium longisporum induced the PR-4 (chitinase) and PR-2 (β-1,3-glucanase) proteins in xylem sap (Floerl et al., 2008). β-1,3-glucanases and some other PRs were also induced in soybean by F. virguliforme infection (Abeysekara and Bhattacharyya, 2014). F. oxysporum f. sp. lycopersici (Fol) colonization induces PR-5 protein in both resistant and susceptible tomato, and the accumulation of PR-1, PR-2 as well as PR-3 concomitantly appeared with disease symptoms in susceptible tomato plants (Rep et al., 2002). In contrast, XSP10 which has similarity with lipid-transfer proteins (PR-14), declined in xylem sap with the infection of Fol, indicating that modification of secretion of this protein may be induced by the pathogen (Rep et al., 2003; Krasikov et al., 2011).

In addition to those PR proteins regulated in infected plants, xylem sap analyses also detected fungal proteins secreted into plant xylem during infection, including so-called effectors. Certain fungal effectors are specifically recognized by plant resistant gene (R gene) products and activate effector-triggered immunity (ETI) in the plant. The effectors recognized by R gene products are called avirulence (Avr) effectors and this genetic interaction between R and Avr genes is described as "gene-for-gene" theory (Flor, 1942; Jones and Dangl, 2006). A representative group of such effectors from vascular-invasion fungus are the Six (secreted in xylem) effectors which are commonly small and cysteine-rich secreted proteins in F. oxysporum species. Houterman et al. (2007) carried out a mixed xylem sap proteome analysis of Fol-infected tomato by 2-DE combined with mass spectrometry. Due to the lack of genomic information of Fol at that time, only seven Fol proteins including four Six (Six1-Six4) proteins were identified. Surprisingly, three of the four Six proteins (Six1, Six3, and Six4) have been functionally confirmed as Avr effectors (Avr3, Avr 2, and Avr1, respectively) in further studies (Rep et al., 2004, 2005; Houterman et al., 2008, 2009). After the whole genome sequence of Fol, more Six genes have been disclosed (Ma et al., 2010). In total, 14 Six genes have been identified so far in Fol (Schmidt et al., 2013). Some Six genes are also present in other formae speciales, for example, Six1, Six4, Six8, and Six9 have homologs in f. sp. conglutinans, Six6 also present in f. sp. melonis and f. sp. radiciscucumerinum, Six7 was also detected in f. sp. lili (Lievens et al., 2009; Thatcher et al., 2012).

Fusarium-wilt in B. oleracea, caused by F. oxysporum f. sp. conglutinans (Foc), is a destructive disease that results in severe losses in both yield and quality during B. oleracea production. Four homologs (Six1, Six4, Six8, and Six9) of the 14 so far reported Six genes in Fol have been found in Foc (Thatcher et al., 2012). Only Foc-Six4 was confirmed as a virulence factor and the function of the remaining three Foc-Six homologs remain unknown (Thatcher et al., 2012; Kashiwa et al., 2013). On the other hand, major resistance locus against Foc has been mapped on the C7 chromosome of B. oleracea, and the resistant gene (named Foc-Bo1) was successfully cloned by map-based cloning (Pu et al., 2012; Shimizu et al., 2015). To date, however, there is no comprehensive study regarding the interaction between Foc and B. oleracea. The Avr effector in Foc is still unknown. Given the effectiveness of analyzing the xylem sap proteome in the tomato-Fol study (Rep et al., 2002, 2004), we therefore carried out a xylem sap proteomics analysis among resistant and susceptible cabbage, that were non-infected or infected with Foc, using in-solution digestion method before LC-MS/MS analysis. Whole genome sequencing for Foc was also performed with the expectation of improved accuracy in protein identification. The main purposes of this study are (1) to obtain a more comprehensive overview of the response in B. oleracea against Foc infection and (2) to investigate candidate effectors in Foc that contribute to virulence and/or avirulence toward B. oleracea.

### MATERIALS AND METHODS

#### Plant and Pathogen Materials

Commercial cabbage F1 cultivars, YCR-Rinen (Nippon Norin Seed Co., Japan) and Delicious (Watanabe Seed Co., Japan) resistant and susceptible to Fusarium-wilt, respectively, were used in this study. YCR-Rinen contains the fusarium-wilt resistant gene (Foc-Bo1), while Delicious does not (Shimizu et al., 2015). The Cong: 1-1 strain of F. oxysporum f. sp. conglutinans (Foc), obtained from cabbage, was provided by Dr. Kadota (National Agricultural Research Center for Tohoku Region, Japan), and was used to prepare inocula. Inoculation was carried out as described in our previous paper (Pu et al., 2012).

### Visualizing of the Infection Process

To determine the optimal time for xylem sap collection, the infection site of Foc in B. oleracea was visualized by 5-bromo-4-chloro-3-indoxyl-α-L-arabinofuranoside (X-Ara) staining as reported by Diener (2012). Infected roots collected at 1 day post-infection (dpi), 3, 7, and 12 dpi were washed well with aqueous solution containing 0.1% Triton X-100 and 20 mM EDTA, pH 8.0. Peat attached to roots was removed carefully by sharp tweezers. The cleaned roots were then incubated with X-Ara in 40-fold volume staining solution (0.02% X-Ara, 10 mM EDTA, 1 mM K3Fe(CN)6, 0.1% Triton X-100 and 0.1 M sodium phosphate, pH 7.2) at 28◦C overnight and the resulting blue color was observed in the Foc-invaded roots.

### Xylem Sap Collection

Four-week old plants were transplanted into Foc-infected or non-infected soil. Xylem sap collection was carried out 12 dpi when the susceptible plants show disease symptoms. Xylem sap collection was carried out according to the method reported by Buhtz et al. (2004). Briefly, xylem sap samples were obtained after cutting stems approximately 3 cm above soil level. To avoid the contamination from the phloem, cut-surface was thoroughly washed by distilled water and the first droplets appearing on the cut surface were removed with blotting paper. The following droplets resulting from "root pressure" were collected with a hand-held pipette. It was difficult to quantify the amount of proteins in the sap using the Bradford method, possibly due to low concentration of xylem sap proteins, which is consistent with the result of Ligat et al. (2011). Since protein concentration in the xylem sap is thought to be quite low, about 15 individual plants were used for collection of each sample, and the collected sap was pooled, frozen in liquid nitrogen, and stored at −80◦C for further analysis.

### Protein Precipitation

1 ml pooled xylem sap was concentrated by vacuum freezing centrifugal drying until approximately 200µl. Then 1 mM HCl was added to the sample to adjust the pH around 4–5. Acetone precipitation was then carried out by adding 1 ml Acetone, followed by store at −80◦C for 1 h. The precipitated proteins were collected by centrifugation for 30 min at 8000 g at 4◦C. The pellet was air-dried and dissolved in 8 M urea and 2 M thiourea buffer. Protein concentration was determined by Bradford method. One-dimensional SDS-PAGE electrophoresis was carried out to get a primary image of protein patterns in each sample. Gelseparated proteins were visualized by silver staining (Wako Pure Chemical Industries, Osaka, Japan).

#### LC-MS/MS Analyses

Proper volume of the dissolved protein (about 1µg) sample was reduced with 10 mM dithiothreitol (DTT) at 60◦C for 30 min and carbamidomethylated with 25 mM iodoacetamide at room temperature for 15 min. Samples were diluted 2-fold with 50 mM ammonium bicarbonate followed by digestion with lysyl endopeptidase (Lys-C; Wako, Osaka, Japan) at 37◦C for 3 h. Samples were then further diluted 2-fold with 50 mM ammonium bicarbonate, and subsequently digested with trypsin (Promega, Madison, MA, USA) at 37◦C for 16 h. The resulting peptides were desalted using C18 stage tips according to the method published by Rappsilber et al. (2003). Label-free protein relative quantitation of each experimental plot with shotgun LC-MS/MS analysis was performed using equal amounts of purified tyrosinephosphorylated peptides. LC-MS/MS analysis was performed on a LTQ Orbitrap Velos hybrid mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) using the Xcalibur version 2.0.7. UltiMate <sup>R</sup> 3000 LC system (Dionex, LC Packings, Sunnyvale, CA, USA) was used to provide the gradient for online reversedphase nano-LC at a flow rate of 300 nL/min. A C18 PepMap™ column (LC Packings) and a nanoscale C18 PepMap™ capillary column (75µm id × 15 cm) (LC Packings) were used as analytical columns. The mobile phases were A (2% acetonitrile, 98% water, 0.1% formic acid) and B (95% acetonitrile, 5% water, 0.1% formic acid). Peptides were separated using a 145 min gradient program consisting a gradient of 2–33% B over 120 min. The full-scan mass spectra were measured from m/z 350 to 1200 in the positive-ion electrospray ionization mode on an LTQ Orbitrap Velos mass spectrometer (Thermo Fisher Scientific) operated in the data-dependent mode. The general mass-spectrometric conditions were as follows: spray voltage, 1.8 kV; capillary temperature, 250◦C; normalized collision energy, 35.0%; isolation width, 2 m/z; activation time, 10 ms; activation Q, 0.25; dynamic exclusion, 180 s; resolution, 60,000; datadependent mode, TOP15 strategy.

In-gel digestion followed by MS analysis identified the protein for each selected band. Proteins of interest were excised from silver stained gel and de-staining. Gel pieces were soaked in trypsin solution and incubation at 37◦C for 16 h to digest after dehydration with acetonitrile. The resulting peptides were desalted using C18 stage tips. LC-MS/MS analysis was performed on a QSTAR XL mass spectrometer (Applied Biosystems, Foster City, CA, USA). To identify the sequence of peptides, peak lists were created using Analyst QS software. The obtained MS and MS/MS data were used for database searches using MASCOT Version 2.4.1 (Matrix Science, London, UK). The search parameters were as follows: trypsin digestion with two missed cleavages permitted, variable modifications (oxidation of methionine, carbamidomethylation of cysteine and propionamidation of cysteine), peptide mass tolerance for MS data ± 0.5 Da, and fragment mass tolerance ± 0.5 Da.

### Database Construction and Protein Identification

Whole genome sequencing of Cong: 1-1 strain was carried out using a hybrid sequencing strategy, utilizing Hiseq1000 (Illumina) and GS Junior (Roche) next generation sequencers. The CLC Genomics Workbench program was used for de novo assembly of 6425 contigs with an average size of 8458 bp. The protein sequences were predicted by Augustus (http://bioinf.unigreifswald.de/augustus/; Stanke and Morgenstern, 2005), and generated a Cong:1-1 protein sequences database with 17,009 predicted proteins (manuscript in preparation). The protein sequences of B. rapa subsp. pekinensis (41,212 proteins) and B. oleracea (1429 proteins) were downloaded from Uniprot (http://www.uniprot.org/). The Foc and plant protein sequences were combined together to generate a database. Two predicted proteins of the candidate Foc resistant genes, FocBo1 in B. oleracea (GeneBank access to AB981182.1. Shimizu et al., 2015) and FocBr1 in B. rapa (Bra012688. Shimizu et al., 2014) were also added into the database. Totally, the generated database contains 59,647 protein sequences for database searching. Additionally, we screened for contaminants by using a combined database of the contaminants database (http://www.matrixscience.com/help/ seq\_db\_setup\_contaminants.html) and our constructed protein database. As a result, several human proteins like trypsin and keratins were detected and then eliminated from the data. Furthermore, the same plant proteins and Foc-proteins were detected in the database search with/without the contaminants database.

To identify the sequence of peptides, peak lists were created using Proteome Discoverer (version 1.3). The obtained MS and MS/MS data were used for database searches using MASCOT Version 2.4.1 (Matrix Science, London, UK) with the following parameters: enzyme, trypsin; peptide mass tolerance, ±5 ppm; fragment mass tolerance, ±0.5 Da; max missed cleavages, 2; variable modifications, carbamidomethyl (C) and oxidation (M). In addition to the criterion of an FDR of 1% used as the acceptance criteria for all protein identifications, only proteins which were positively identified throughout the LC-MS/MS analysis (repeated three times) with score>=95 (at least two times among the repeats), were reported in this study. The proteins satisfying those criteria were also applied to protein quantitation in the Progenesis LC-MS deta analysis program (version 4.1, Non-linear Dynamics, Newcastle, UK). The protein content levels within each cultivar, i.e., between Del-Con and Del- Inf as well as between Ri-Con and Ri-Inf were compared in the Progenesis LC-MS analysis where the MS/MS data (three times repeat per experimental pot) were used for quantitation and the obtained normalized abundances of each peptide were subjected to statistical analysis using one-way analysis of variance (ANOVA). In this study, the detected feature was assumed to be statistically significant when the p-value for a given peptide was <0.05 and minimum fold change was >= 2.

#### Bioinformatics

The identified plant proteins were annotated based on Uniprot website information and the remaining uncharacterized proteins were further annotated with the InterPro domain annotation by searching the corresponding gene name in Brassica Database (http://brassicadb.org/brad/index.php). The corresponding A. thaliana genes were obtained from the known orthologs listed in the Brassica Database, or determined by a BLAST search against the Uniprot database. SignalP (http://www. cbs.dtu.dk/services/SignalP/) and TargetP http://www.cbs.dtu. dk/services/TargetP/) were used for sub-cellular localization prediction. The identified Foc proteins were annotated by NCBI protein BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) in Non-redundant protein sequences (nr) database using blastp algorithm.

Our MS/MS data was deposited to the PRIDE Archive, PRIDE Accession PXD003378 (http://www.ebi.ac.uk/pride/archive/). This Whole Genome Shotgun project has been deposited at DDBJ/EMBL/GenBank under the accession LPZQ00000000. The version described in this paper is version LPZQ01000000.

#### RESULTS

### Determining the Time-Point for Harvesting the Samples

Preliminary experiments indicated that YCR-Rinen was resistant, while Delicious showed susceptibility to Foc. 12-day-old plants were used for inoculation test and the symptoms were observed 6- to 7 dpi in our inoculation system (data not shown). The differences in the infected roots of YCR-Rinen and Delicious were visualized by X-Ara staining (**Figure 1**). In the susceptible cultivar (Delicious), subtle staining was observed at 1 dpi at the outer layers of the root apex and the lateral root primordial (LRP) (**Figure 1A**), supporting the previous study that the pathogen penetrates the surface of root apex and LRP, and traverses tissues outside the vascular cylinder in the primary phase of initial Foc penetration to A. thaliana (Diener, 2012). More infection points were observed at 3 dpi in the susceptible cultivar. The staining went through the root apex region, and developed into more basal tissue, especially in the vascular cylinder (**Figure 1B**). A thorough infiltration of the central vascular cylinder was observed at 7 dpi in the susceptible cultivar, and more extensive inroads into the road lines at 12 dpi was indicated by pervading of the precipitated blue (**Figures 1C,D**). The growth of Foc in host plant was associated with the development of yellowing symptoms in the aerial part: yellowing appeared from 6 to 7 dpi, developed as time went on, and finally the susceptible plant was dead.

The resistant cultivar showed no yellowing, but some plants showed stunted and retarded growth after infection. In the resistant cultivar, no infection was observed at 1 dpi (**Figure 1E**). Faint blue spots were detected from 3 dpi and the staining mostly confined in the root apex and LRP throughout the infection period tested in our study, indicating the restricted development of Foc in the resistant cultivar (**Figures 1F–H**). Early stage of infection showed limited growth of Foc, leading to less fungal proteins transported into the xylem of plants, while late stages of infection hinder xylem sap collection in susceptible plants, due to the advanced progression of symptoms. We therefore decided upon 12 days after inoculation as the appropriate point in time for the collection of xylem sap.

#### SDS-PAGE and Xylem Protein Identification via In-Gel Digestion

The collected xylem was first separated on a one-dimensional SDS gel (**Figure 2A**). As expected, similar band patterns were observed in the non-infected samples in the different

cultivars, YCR-Rinen and Delicious. On the other hand, the SDS gel analysis clearly showed that Foc infection induced different types of bands in the resistant and susceptible cultivars. We selected several major bands from the infected sample and proceeded with in-gel digestion so that only plant proteins such as arabinofuranosidase, glycoside hydrolase, and peroxidase etc. were detected with low score values (**Figure 2B**). Although the identification via the in-gel digestion analysis should be preliminary, this result indicates the overwhelming concentration of plant proteins in the xylem. The proteins identified by the in-gel digestion followed by LC/MS/MS were also identified by the following in-solution digestion analysis.

#### Plant Proteins Identified in the Non-Infected and Infected *B. oleracea* Xylem Sap

In solution digestion was carried out to make a detailed investigation of protein content in each xylem sap sample, collected from the non-infected and infected plants of Delicious and YCR-Rinen (Del-Con, Del-Inf, Ri-Con, and Ri-Inf, respectively). For the xylem sap proteins of the non-infected plants, the shotgun LC-MS/MS proteomics analysis identified 270 and 254 plant proteins from Del-Con and Ri-Con,

indicate non-infected or Foc-infected plants of susceptible Delicious and resistant YCR-Rinen, respectively. (A) Several bands that show induced or repressed were selected for further analysis. (B) In-gel digestion followed by MS analysis identified the protein for each selected band.

respectively (**Figure 3**, Table S1). VENNY 2.0 (http://bioinfogp. cnb.csic.es/tools/venny/index.html) was then used to investigate the overlap of these proteins between the two samples. Among the proteins detected in non-infected samples, 225 proteins were common between Del-Con (83% of 270) and Ri-Con (89% of 254). The difference of the proteins identified in the two non-infected B. oleracea xylem sap samples could be due to differences in the cultivars.

The molecular mass (MM) of B. oleracea xylem sap plant proteins varied with a range of 7.6 kDa (M4CY57, Uncharacterized protein) to 160.3 kDa (M4CW30, Tyrosine-protein kinase). However, the majority of plant proteins (86∼88%) had a MM between 10 and 70 kDa in size (Figure S1), which coincided with band patterns observed in one-dimensional SDS-PAGE (**Figure 2**). These results indicate B. oleracea xylem sap proteins mainly consist of relatively small sized proteins.

Sub-cellular localization of proteins identified in healthy samples was predicted by bioinformatics software and the proteins were classified into (1) predicted intracellular proteins (about 20% of total) that were devoid of signal peptide and (2) secreted proteins (about 80% of total). The secreted proteins were further sorted into eight functional classes based on the previous study of A. thaliana cell wall proteins (Jamet et al., 2006) and B. oleracea xylem sap proteome (Ligat et al., 2011); proteins

acting on carbohydrates (31∼30%, differ between Delicious and Rinen), oxido-reductase (27∼28%, as above), proteases (8∼7%), proteins involved in lipid metabolism (8∼6%, as above), proteins involved in signaling (4%), proteins with domains interacting with carbohydrates or proteins (7∼9%, as above), proteins with diverse functions (11%), and proteins with yet unknown function (4%; **Figure 4**). The main group in proteins acting on carbohydrates was glycoside hydrolase; oxido-reductases mainly consisting of FAD-binding, peroxidase, and plastocyaninlike family; almost all the proteins in proteases group were peptidases; LTPs were the main content of lipid metabolism; all the proteins in signaling group were FAS1 (fasciclin-like) domain proteins, which may function in cell communication and adhesion (Johnson et al., 2003).

For the xylem sap proteins of the infected plants (Del-Inf and Ri-Inf), 255 plant proteins were identified from Delicious (Del-Inf), and 209 were identified from YCR-Rinen (Ri-Inf), (**Figure 3**, Data Sheet 1). Foc proteins were only identified from the xylem sap collected from susceptible Delicious that was infected by Foc (later mentioned). The relatively small number of proteins detected in Foc-infected YCR-Rinen (resistant) may indicate a resistant reaction, because tyloses are expected to induce in resistant plants to occlude the xylem, and thus limit the pathogen's growth as well as the protein transport in plant xylem (reviewed in Yadeta and Thomma, 2013). The functional classification of the secreted proteins detected in the Foc-infected B. oleracea is shown in **Figure 5A**. This result enables us to analyze changes that have occurred between healthy and infected samples, as follows.

#### Identification of Differential and Unique Proteins in *B. oleracea* during Foc Infection

Profile comparisons between Foc-infected and non-infected within each cultivar, i.e., Del-Con vs. Del- Inf as well as Ri-Con vs. Ri-Inf, were carried out (**Figure 5**, Data Sheet 2). The results are as follows. First, the total number of proteins

repressed after infection was larger than induced in both the resistant and susceptible groups. In YCR-Rinen (resistant), 112 secreted proteins (among 155 proteins, including 43 intercellular proteins) were either unique or show a fold change >= 2, including 78 repressed and 34 induced. On the other hand, in Delicious (susceptible), the 164 secreted proteins (among 204 proteins, including 40 intercellular proteins) had a fold change >= 2 in Delicious (susceptible), including 103 repressed and 61 induced (Data Sheet 2). The greater number of repressed proteins may indicate a suppressed transporting/metabolism in xylem in both Foc-resistant and -susceptible system. Indeed, no signaling related protein (FAS1 domain proteins) was induced in either of the reactions (**Figure 5**). The suppressed metabolism in xylem is considered to coincide with the symptom of stunned and retard growth triggered by Foc. Second, similar percentage of proteins acting on carbohydrates was induced or suppressed in both the resistant and susceptible reactions, which demonstrate a role of these proteins in Foc resistance (**Figure 5**). Third, lipid metabolism related proteins, including the so-called PR-14 (lipidtransfer proteins, LTP) that contribute to plant defense response, also showed up- and down-regulation in similar percentages in both the resistant and susceptible systems. Additionally, in the groups of proteins with interacting domains and miscellaneous proteins having diverse functions, a higher number of proteins were induced by Foc-infection (**Figure 5**, Data Sheet 2). This is mainly due to the induction of lectins, leucine-rich repeat (LRR) and thaumatins by Foc-infection. Furthermore, although both up- and down-regulated oxido-reductases and proteases were detected in the susceptible reaction, only a few oxidoreductases were induced in resistant system (**Figure 5**). Such a differential response on oxido-reductases indicates that the induced oxido-reductases were considered to be related to symptom development in susceptible plant.

### Foc Proteins Identified from the Xylem Sap of the Infected Delicious

Whole genome sequencing for Foc was carried out and generated a Cong: 1-1 protein sequences database with 17,009 predicted proteins (manuscript in preparation) using for Foc protein identification in this study. Foc proteins were only identified from the xylem sap collected from susceptible Delicious that was infected by Foc (Del-Inf). In total, 25 predicted Foc proteins were detected in Del-Inf (**Table 1**). The MM of Foc proteins identified varied from 12.4 kDa (P11242) to 104.8 kDa (P01134). Twentyfour proteins (96% of total number proteins detected) were distributed among 10∼70 kDa in size (Figure S1, Foc protein), and P01134 is the only Foc protein that is larger than 70 kDa identified in our study.

Since small cysteine-rich proteins secreted by phytopathogenic fungi into their host are implicated as effectors involved in disease development and R-gene-mediated resistance (Reviewed in Rep, 2005; Stergiopoulos and De Wit, 2009), we predicted the sub-cellular localization and analyzed total protein length as well as the cysteine content calculation (cysteine number divided by total protein length) of identified proteins (**Table 1**). Of the 25 proteins, 22 were predicted as secreted protein by SingalP or TargetP, while the remaining three (P00041, P05132, P13299) were predicted as intracellular proteins. Eleven of the twenty-three secreted proteins were regarded as small proteins with a length less than 300 amino acids and one of them (P16246) contains no cysteine. That is, we detected 10 small secreted cysteine-containing Foc proteins during the interaction between Foc and B. oleracea.

To annotate the identified Foc proteins, we performed similarity search of the predicted protein sequences using the NCBI protein BLAST tool. Based on the alignment list generated by the NCBI blast tool, we selected the top match from in the BLAST results for annotation (Table S4). Since most of them were hypothetical proteins, a functional/ putative functional protein item in each BLAST result with max score >150 and query cover >60% was also selected as reference (Table S4) and list in the **Table 1** as description. Among the 25 identified Foc proteins, five of them (P11242, P13298, P13299, P13373, and P16923) have no reliable functional/ putative functional protein item in the BLAST result (**Table 1**, Table S4). Most of the annotated proteins were enzymatic active protein, such as chitinase (P11164), putative oxidoreductase (P11347), and xylanase (P10456 and P16246).

P14728 and P15981 show 100% identity with the reported Foc-Six4 and Foc-Six1 gene, respectively. Foc-Six4 has been demonstrated as a virulence factor toward A. thaliana and cabbage (Thatcher et al., 2012; Kashiwa et al., 2013). However, the expression of Foc-Six1 during infection and its effector function has yet to be confirmed. Our study indicated a secretion of Foc-Six1 into susceptible plant xylem during Foc-infection. In addition, P11311 containing two Lysin motifs (LysMs), was determined as a LysM-contains protein (Figure S2); P03403 contains a cerato-platanin (CP) domain and its predicted amino acid sequence shares 91.9% similarity with some reported CP proteins including MgSM1 (a CP protein from Magnaporthe grisea), thus determined as a CP protein (**Figure 6**).

### DISCUSSION

### Determining the Time-Point for Harvesting the Samples

In this study, we demonstrated that the mycelia penetrate the surface of root apex and LRP, traverse cortical tissues and reach the vascular cylinder at 7 dpi in the susceptible cultivar. After that the mycelia quickly proliferated in the root in the susceptible plants. In contrast, in the resistant plant, the mycelia were restricted at the point of infection, such as the root apex. This observation is in agreement with the result reported by Li et al. (2015), who performed microscopical analysis using the GFPexpressing strain of Foc to the 2–3 leaf stage of cabbage. Li et al. (2015) reported that the Foc-colonization in the root of the susceptible plants reached a maximum at the 11 dpi. Therefore, the time point at 12 dpi is thought to be suitable for the collection of xylem sap.

#### Plant Proteins in the Xylem Sap of the Non-Infected *B. oleracea*

Previous studies examined the xylem sap through a precise geldependent separation and digestion (Rep et al., 2002; Buhtz et al., 2004; Houterman et al., 2007; Floerl et al., 2008). However, the number of proteins identified was limited because not all of the proteins could be resolved by gel electrophoresis and further analysis for the detected proteins patterns (one-dimensional) or spots (two-dimensional) would be time and labor consuming. For example, Kehr et al. (2005) carried out proteome analysis for B. napus xylem sap based on 2-dimensional analysis and identified 69 proteins. Thereafter, application of high-throughput LC-MS/MS improved resolution for proteins and thus increased the number of proteins identified. A representation of xylem sap proteome analysis is Ligat et al. (2011) who reported 189 proteins in B. oleracea xylem. Our study adopted in-solution digestion, which yielded about 200 proteins in B. oleracea xylem. Despite the different techniques and databases for protein identification, close similarity between A. thaliana and Brassica genes allowed the comparison of our study with previous studies (Table S1). For example, 32 and 150 unique A. thaliana homologous genes were reported in the study of Kehr et al. (2005) and Ligat et al. (2011), respectively. About 84% (27 of 32) unique A. thaliana genes homologous with the B. napus protein identified by Kehr et al. (2005) were present in our study, demonstrating a relationship between B. oleracea and B. napus that B. oleracea (C genome) is a progenitor species of B. napus (AC genome) (Table S1). TABLE 1 | Foc proteins identified in the xylem of Foc-infected *B. oleracea* cv. Delicious.


<sup>a</sup>Serial number/contig number of Foc Cong: 1-1 genome database used in this study. The numbers given in bold indicate a small (less than 300 amino acids) cysteine-contained secreted protein.

<sup>b</sup>Genes were annotated by NCBI blastp. Descriptions of the functional/putative functional protein item were used. Descriptions given in bold were discussed in Discussion. A detail result of protein BLAST consults Table S4.

<sup>c</sup>SP indicates a predicted signal peptide denoting secretion. The signal peptide was predicted using SignalP (http://www.cbs.dtu.dk/services/SignalP/) and TargetP (http://www.cbs.dtu.dk/services/TargetP/). "Y" and "N" indicate whether or not the sequence contains a signal peptide in secretory pathway. When both predictions are consistent, only the SignalP result is shown. When it is not the case, both predictions are shown. "M" indicates that TargetP predicted a mitochondrial targeting peptide.

<sup>d</sup>Length given in amino acid number is count from the full length of predicted peptide sequence generated from Foc genome information.

<sup>e</sup>Number before the slash indicates the score and after the slash indicated the coverage (given in percentage) evaluated by MASCOT.

Compared with the previous report by Ligat et al. (2011), about 75% of the A. thaliana homologous genes (112 of 150 unique ID) identified in Ligat et al. (2011) are present in our study (Table S1). There appears to be no remarkable differences in protein families identified between the studies, however there are variations in the constitution and protein numbers in each protein family observed between Ligat et al. (2011) and our study. The difference may result from cultivars, sampling time points or the different technologies used in the bioinformatic analysis.

It has been suggested that xylem sap protein composition is conserved among species, since some of the most abundant proteins are commonly present within plant xylem of various cultivars (Buhtz et al., 2004; Dafoe and Constabel, 2009). Many of these proteins, such as PRs, oxido-reductases and proteases, could be induced by the stress of the unavoidable decapitation step during xylem sap collection (Kehr et al., 2005; Alvarez et al., 2006; Djordjevic et al., 2007; Aki et al., 2008; Floerl et al., 2008; Dafoe and Constabel, 2009; Ligat et al., 2011; Zhang et al., 2015). In the present study, these proteins were also detected in B. oleracea xylem sap in both the non-infected and Foc-infected plants (Data Sheet 1).

#### Comparison of the Xylem Sap Proteins in the Non-Infected with Foc-Infected Plants

Our study suggests a regulation of xylem sap proteins driven by Foc-infection (**Figure 5**). It is well known that proteincarbohydrate interactions play an important role in the recognition of pathogens. Carbohydrate structures, which are either present at the surface of the invading pathogen cell or released from degraded plant cell wall damaged by pathogen entry, are the main part of the pathogen/damage-associated molecular patterns (P/ DAMPs) perceived in the plant and triggered by the innate immunity response of the plant (reviewed in Lannoo and Damme, 2014). In our study, the large portion of up- and down-regulated proteins acting on carbohydrates in both resistance and susceptible system also suggests complex and intense protein-carbohydrate interactions in B. oleracea

xylem sap driven by Foc-infection (**Figure 5**, Data Sheet 2). Among these carbohydrate interacting proteins, β-1,3-glucanases and chitinase have been reported to contribute to biotic stress response and are ubiquitous in plant xylem (Punja and Zhang, 1993; Buhtz et al., 2004; Ligat et al., 2011; Ahmed et al., 2012). Rep et al. (2002) reported an accumulation of β-1,3 glucanases and chitinase in tomato xylem sap after Fol-infection. They were also up-regulated in B. napus upon V. longisporum infection (Floerl et al., 2008). In our study, Foc infection in both resistant and susceptible plants activated chitinases and β-1,3-glucanase. Although they may be involved in PTI (PAMP-triggered immunity) in B. oleracea, such a defense mechanism was not strong enough in the susceptible plants, such that Foc successfully persisted in parasitic growth. However, in the resistant plant, certain effector(s) triggered an R-gene (FocBo1, later mentioned) mediated resistance reaction (ETI) that restricts Foc growth.

Our previous study identified the Foc-Bo1, a candidate gene, conferring Fusarium resistance to B. oleracea (Shimizu et al., 2015). Its homologous gene in B. rapa, named Foc-Br1, was also identified which was shown to be a candidate of Foc resistant gene (Shimizu et al., 2014). Since both of them are NBS-LRR type resistant genes, they are considered as the main R-gene involved in mediated resistance against Foc. We did not detect the Foc-Bo1 protein in the xylem sap of YCR-Rinen, a cultivar that contains the Foc-Bo1 gene (Shimizu et al., 2015). In the tomato-Fol system, the nuclear migration of Avr2 from Fol is required to activate the tomato resistance protein I-2, which triggers cell death as a consequence of recognition of Avr2 (Ma et al., 2013). It is therefore plausible that the R-gene mediated resistance against Foc is also an in-cell reaction. That is to say, FocBo1 perceives Foc in the cell and immediately triggers HR to limit the further growth of Foc. The effective programmed cell death limits the transport of Foc-Bo1 protein in xylem sap of B. oleracea.

Different from the group of proteins acting on carbohydrates which showed a similar number of up- and down- regulated proteins due to Foc-infection, a larger portion of proteins with interacting domains and miscellaneous proteins having diverse functions were induced after Foc-infection. In the interacting domains group, the induced proteins were mainly leucine-rich repeat (LRR) protein and legume lectin (Table S2). The LRR domain plays an important role in direct/indirect recognition of pathogenic effector proteins, and lectin domains are implicated in the recognition of carbohydrate structures that are perceived as "danger" molecules (Reviewed in Lannoo and Damme, 2014). In the miscellaneous protein group, the induction mainly comes from Thaumatin. Thaumatin (PR-5) is a kind of sweet protein that was first reported in Thaumatococcus danielli. Since then, studies have identified Thaumatin-like proteins (TLPs) from many different plants and classified these TLPs into PR-5 proteins, based on their ability to respond to biotic and abiotic stress (Van Loon et al., 2006). The anti-fungal activity of TLPs against plant pathogenic F. oxysporum has been demonstrated in A. thaliana (Hu and Reddy, 1997), French bean (Ye et al., 1999), tomato (Rep et al., 2002), and cotton (Munis et al., 2010). Our study showed that thaumatin proteins were induced at a similar level in both resistance and susceptible systems, with the exception of M4EA13 and M4CVH7 that was specifically accumulated in YCR-Rinen and Delicious, respectively (Table S3). The fact that Foc colonization proceeds in Delicious in spite of the accumulation of PR-5 proteins implies that Foc could avoid or resist potential anti-fungal activity of these TLPs. Further study is required to determine whether the M4EA13 produced in YCR-Rinen contributes specifically against Foc infection, or the accumulation of thaumatins in Delicious occurs too late (than YCR-Rinen) to prevent Foc infection.

Interestingly, oxido-reductases are repressed in both cultivars but induced only in the susceptible system (**Figure 5**). Oxidative bursting is one of the earliest responses in the root against F. oxysporum infection (Plancot et al., 2013). Previous studies suggest reactive oxygen species (ROS) act as signaling molecules thus contributing to plant defenses (Torres, 2010). However, it has been recently suggested that ROS may also promote disease development of some pathogens. Lyons et al. (2015) reported that PRX33, which is required for ROS formation and MAMP-triggered ROS production, promotes susceptibility to F. oxysporum. Therefore, the induction of oxido-reductases in this study may relate to the development of Fusarium-wilt.

## Fungal Proteins Detected in *B. oleracea* Xylem

Identification of Fusarium proteins in plant xylem has been thoroughly carried out in tomato-Fol system (Rep et al., 2002; Houterman et al., 2007). However, due to the lack of genome information of Fol at that time, identification of Fol original proteins was dependent on the information of homologous peptides of other species. This hindered previous identification work and limited the number of proteins that could be identified. To facilitate the accuracy of protein identification, whole genome sequencing of Foc was carried out and a Foc protein database with 17,009 predicated proteins was generated in this study. Finally, we detected 25 Foc-proteins in infected B. oleracea xylem, which strongly indicated these proteins play some roles during infection. Thus, functions of these proteins are also of great interest. Among the detected proteins, 22 were predicted to have a signal peptide that connotes active secretion. Given small proteins secreted by plant-pathogenic fungi in the hosts have been implicated in disease symptom development as well as in R-gene mediated disease resistance (Reviewed in Rep, 2005), proteins containing less than 300 amino acids are of great interest. Since the cloning of the first Six gene (Six1/Avr3) from Fol, 14 Fol-Six effector candidate proteins have been identified (Rep et al., 2004; Schmidt et al., 2013). Six1, Six4, Six8, and Six9 homologs have been detected in Foc but only Six4 has been cloned and confirmed as a virulence factor to A. thaliana and cabbage (Thatcher et al., 2012; Kashiwa et al., 2013). Our study detected the reported Foc-Six4 protein as well as the not yet characterized Foc-Six1 protein in infected xylem sap. Foc-Six1 shares 80% amino acid similarity with Fol-Six1 which is required for full virulence and recognized by the I-3 resistance gene in tomato (Rep et al., 2004, 2005; Thatcher et al., 2012). However, the expression and function of Foc-Six1 during infection has not been reported. The detection of Foc-Six1 proteins in infected cabbage strongly indicates the expression of Foc-Six1 during infection. Further studies should focus on the function of Foc- Six1.

Protein annotation indicates P03403 is a Snodprot protein that has similarity with the reported cerato-platanin (CP) family proteins, such as Snodprot1 of Neurospora crassa (Q9C2Q5), Snodprot-FG (Q5PSV6) of Gibberella zeae, MgSM1 of Magnaporthe grisea, and BcSpl1 in Botrytis cinerea (Jeong et al., 2007; Yang et al., 2009; Frias et al., 2011; **Figure 6**). To date, CP proteins have only been reported in fungi, are widespread within fungi, conserved in structure, and abundantly secreted. Additionally, CP proteins have been reported to act as pathogenassociated molecular patterns (PAMPs). Ectopic expression of MgSM1 gene in A. thaliana activates plant defense response, resulting in broad-spectrum resistance against different fungal and bacterial pathogens (Yang et al., 2009). A. thaliana lacking the BAK1gene of the PAMP signaling pathway prevented the induction of necrosis in this mutant by BcSpl1 (Frias et al., 2011). Furthermore, some of the CP proteins have been shown to act as virulence factors, i.e., knockout of BcSpl1 in B. cinerea and MSP1 in M. grisea showed reduced virulence to their host plant, respectively (Jeong et al., 2007; Frias et al., 2011). Recent studies indicate some MpCPs (CP proteins from Moniliophthora perniciosa) are able to bind chitin with high affinity, and consequently, this strong affinity for chitin could sequester the excitation of the plant immune system elicited by fungal chitin fragments (De O Barsottini et al., 2013; Baccelli et al., 2014). To our knowledge, the function of CP proteins has not been reported in F. oxysporum species. The detection of P03403 protein strongly indicates it plays a role in Fusarium-host interaction. Gene knockout experiments are ongoing to confirm this hypothesis.

In addition to CP proteins, LysM (Lysin motif)-containing effectors are also widespread in fungi and are proposed to contribute to virulence by sequestration of chitin oligosaccharides released from fungus during infection, thereby blocking the activation of host chitin receptors (Reviewed in De Jonge and Thomma, 2009). For example, the secreted LysM Protein1 (Slp1, required for full virulence toward rice) from M. oryzae is accumulated at the interface between the fungal cell wall and the rice plasma membrane. It competes with the chitin elicitor binding protein (CEBiP) for the binding of chitin oligosaccharides and it is therefore proposed to suppress chitininduced plant immune response, facilitating virulence (Mentlak et al., 2012). Ecp6 from Cladosporium fulvum is also a famous LysM-containing effector, which suppress chitin-triggered immunity through intra-chain LysM dimerization and/or through binding to chitin oligomers thereby physically blocking host immune receptor dimerization (De Jonge et al., 2010; Sánchez-Vallet et al., 2013). In our study, the detected xylem secreted Foc protein P11311, contains two LysMs. Thus, this result generates a question for the function of LysM-containing protein during Foc infection (Figure S2). Further studies should reveal whether, and how, perturbation of chitin-triggered immunity by LysM occurs in the interaction of Foc with B. oleracea.

The only cysteine-free small secreted protein is P16246 which has 94.2% similarity with FGSG\_03624, an endo-1,4-β-xylanase of Fusarium graminearum (Sella et al., 2013). In addition to P16246, another endo-1,4-β-xylanase, P10456, was also detected in this study. P10456 is 328 amino acids in length and has 99.2% identity with xyl2 of Fol (GenBank accession No. AF052583, Ruiz-Roldan et al., 1999). Endo-1,4-β-xylanase are produced by many plant pathogenic fungi and are likely to be involved in the degradation of cell walls during host colonization thereby facilitating infection/pathogenicity (Walton, 1994; Kikot et al., 2009). However, xyl2 is only expressed in the final stages of tomato wilt, and thus may associate with saprophytic growth (Ruiz-Roldan et al., 1999). It was suggested recently that xylanase contributes to virulence not by enzymatic activity but also with its necrotizing activity (Enkerli et al., 1999; Noda et al., 2010). Both of the FGSG\_03624 (Sella et al., 2013) and the P16246 (our study) share similarity with the amino acids in regards to necrosis elicitation. However, whether P16246 has a similar ability to FGSG\_03624 to induce necrosis in infected tissue remains elusive.

In conclusion, to our knowledge, the present study is the first report regarding to identification of protein changes driven by Foc-infection in B. oleracea xylem sap in both the resistant and susceptible systems. The large number of up- and downregulated proteins acting on carbohydrate as well as the induced LRR and legume lectin domain proteins suggests a complex recognition of Foc in B. oleracea. In addition, the induced oxidoreductases in the susceptible reaction may indicate a contribution of ROS to disease development. Importantly, our study also reported 25 predicted Foc proteins in susceptible plant xylem sap infected by Foc. Eleven of them are small (less than 300 amino acid in length) secreted Foc proteins in the infected B. oleracea with 10 of them containing cysteine. These Focproteins are of great interest because they are good candidates for virulence and/or avirulence factors. Thus, the present study provides important resources for study on Foc effector proteins as well as the mechanisms of the interaction between Foc and B. oleracea.

### AUTHOR CONTRIBUTIONS

ZP contributed to the study as the first author. YI, KK, and YK partly conducted LC-MS/MS analysis. SF conducted Illumina sequence. AT, MS, DS, and SN conducted bioinformatics analysis. YS took charge of fungal culture. RF conducted experimental design and statistical analysis. HH contributed to proteome experimental design. KO partly contributed to writing manuscript, provided funding, and managed the whole project.

#### FUNDING

The support provided by China Scholarship Council (CSC) during a visit of ZP to Niigata University is deeply acknowledged. This work was supported by the Programme for Promotion of Basic and Applied Researches for Innovations in Bio-oriented Industry to KO.

### ACKNOWLEDGMENTS

The authors sincerely thank Dr. Y. Kadota at the National Agriculture and Food Research Organization/NARO Tohoku Agricultural Research Center, Japan, for kindly providing Cong: 1-1 strain of F. oxysporum f. sp. conglutinans. The authors also feel grateful to Dr. R. Terauchi in Iwate Biotechnology Institute, Kitakami, Japan, for his valuable suggestions to this study.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 00031

## REFERENCES


is a necrotizing factor but is not essential for virulence. Plant Physiol. Biochem. 64, 1–10. doi: 10.1016/j.plaphy.2012.12.008


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Pu, Ino, Kimura, Tago, Shimizu, Natsume, Sano, Fujimoto, Kaneko, Shea, Fukai, Fuji, Hirano and Okazaki. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The effector repertoire of Fusarium oxysporum determines the tomato xylem proteome composition following infection

Fleur Gawehns 1 †, Lisong Ma1 †, Oskar Bruning<sup>2</sup> , Petra M. Houterman<sup>1</sup> , Sjef Boeren<sup>3</sup> , Ben J. C. Cornelissen<sup>1</sup> , Martijn Rep<sup>1</sup> and Frank L. W. Takken<sup>1</sup> \*

*<sup>1</sup> Molecular Plant Pathology, Faculty of Science, Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, Netherlands, <sup>2</sup> RNA Biology and Applied Bioinformatics Research Group and MAD: Dutch Genomics Service and Support Provider, Faculty of Science, Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, Netherlands, <sup>3</sup> Laboratory of Biochemistry, Wageningen University, Wageningen, Netherlands*

#### Edited by:

*Hanjo A. Hellmann, Washington State University, USA*

#### Reviewed by:

*Curtis G. Wilkerson, Michigan State University, USA Torsten Kleffmann, University of Otago, New Zealand*

> \*Correspondence: *Frank L. W. Takken f.l.w.takken@uva.nl*

#### †Present Address:

*Fleur Gawehns, Naktuinbouw, Netherlands Inspection Service for Horticulture, Roelofarendsveen, Netherlands; Lisong Ma, Saskatoon Research Centre, Agriculture and Agri-Food Canada, Saskatoon, Canada*

#### Specialty section:

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

Received: *04 September 2015* Accepted: *22 October 2015* Published: *04 November 2015*

#### Citation:

*Gawehns F, Ma L, Bruning O, Houterman PM, Boeren S, Cornelissen BJC, Rep M and Takken FLW (2015) The effector repertoire of Fusarium oxysporum determines the tomato xylem proteome composition following infection. Front. Plant Sci. 6:967. doi: 10.3389/fpls.2015.00967* Plant pathogens secrete small proteins, of which some are effectors that promote infection. During colonization of the tomato xylem vessels the fungus *Fusarium oxysporum* f.sp. *lycopersici* (Fol) secretes small proteins that are referred to as SIX (Secreted In Xylem) proteins. Of these, Six1 (Avr3), Six3 (Avr2), Six5, and Six6 are required for full virulence, denoting them as effectors. To investigate their activities in the plant, the xylem sap proteome of plants inoculated with Fol wild-type or either *AVR2, AVR3, SIX2, SIX5,* or *SIX6* knockout strains was analyzed with nano-Liquid Chromatography-Mass Spectrometry (nLC-MSMS). Compared to mock-inoculated sap 12 additional plant proteins appeared while 45 proteins were no longer detectable in the xylem sap of Fol-infected plants. Of the 285 proteins found in both uninfected and infected plants the abundance of 258 proteins changed significantly following infection. The xylem sap proteome of plants infected with four Fol effector knockout strains differed significantly from plants infected with wild-type Fol, while that of the *SIX2-*knockout inoculated plants remained unchanged. Besides an altered abundance of a core set of 24 differentially accumulated proteins (DAPs), each of the four effector knockout strains affected specifically the abundance of a subset of DAPs. Hence, Fol effectors have both unique and shared effects on the composition of the tomato xylem sap proteome.

#### Keywords: label free proteomics, effectors, pathogenicity, virulence, xylem sap

### INTRODUCTION

Pathogens such as bacteria, fungi, oomycetes, protozoa and nematodes continuously challenge the plant immune system. Although most attacks are unsuccessful, sometimes plants are infected and disease develops. Diseases in crops do not only result in yield losses, but can also affect food-safety and quality (Agrios, 2005). To facilitate infection and host colonization, pathogenic microorganisms deploy small, secreted proteins, often called effectors (Win et al., 2012). Some effectors suppress or evade plant basal immunity thereby conferring Effector Triggered Susceptibility (ETS), while others manipulate host factors required for a sustained compatibility (Jones and Dangl, 2006; Win et al., 2012). To unravel how pathogens manipulate the host one needs to study effector action.

Among the most devastating plant pathogens are fungi and they pose a widespread threat to our crops (Fisher et al., 2012). Of the soil-borne fungus Fusarium oxysporum, pathogenic and host-specific forms have evolved that cause wilt disease or foot, root or bulb rot on a variety of economical important crops like cotton, banana, melon and tomato (Tjamos and Beckman, 1989; Michielse and Rep, 2009). The tomato-infecting form, F. oxysporum f.sp. lycopersici (Fol) is the causal agent of Fusarium wilt disease. Fol invades the roots and subsequently colonizes the xylem vessels, thereby compromising water transport resulting in wilting of the plant (Michielse and Rep, 2009). So, during early stages of infection the interface and communication between pathogen and host is largely confined to the xylem sap (Michielse and Rep, 2009). Since the xylem sap can easily be collected this pathosystem is perfectly suited to identify the proteins acting at the plant-pathogen interface (Rep et al., 2002; Houterman et al., 2007). In the sap of infected tomato plants 14 small, Folderived proteins have been identified so far that are called SIX (Secreted In Xylem) proteins (Houterman et al., 2007; Lievens et al., 2009; Ma et al., 2010; Schmidt et al., 2013). Gene knockout studies revealed that Six1 (Avr3), Six3 (Avr2), Six5, and Six6 are required for full pathogenicity (Rep et al., 2004; Rep, 2005; Houterman et al., 2008, 2009; Gawehns et al., 2014) designating them as effectors sensu strictu (i.e., having a role in disease development).

A commonly observed response to pathogen attack is the production of so-called pathogenesis-related (PR) proteins by the plant. Since many of them possess antimicrobial activity, it is generally assumed that they play a role in defense (van Loon et al., 2006). In response to Fol infection the abundance of specific tomato proteins, including PR proteins, changes in the xylem sap (Rep et al., 2002; Houterman et al., 2007). Besides the appearance of these specific proteins, the abundance of others decreases, sometimes below the detection level resulting in an apparent disappearance. One protein whose abundance strongly decreases upon infection is XSP10, a 10 kDa protein with lipidbinding properties (Rep et al., 2003; Krasikov et al., 2011), which is required for full susceptibility of tomato to Fusarium wilt (Krasikov et al., 2011).

Since all effector knockout strains tested so far showed decreased pathogenicity, Fol effectors are predicted to have unique and non-redundant functions. Hence, each effector is expected to affect only a specific subset of plant responses. To test this hypothesis and to gain insight into these functions, we here set out to determine the changes in the xylem sap proteome following Fol infection using a label-free quantitative, large-scale proteomics approach. Thereto, xylem sap of tomato inoculated with either wild-type Fol or the available AVR3, AVR2, SIX5, or SIX6 knockout strains and a newly created SIX2 knockout was isolated, and analyzed using qLC-MS. To reveal changes in abundance of xylem sap proteins a bioinformatics pipeline was developed. Next, the xylem sap proteomes were compared with each other and with the proteome of either mock (water) or wild-type Fol inoculated plants. Surprisingly, not only a specific effect of Fol effectors on the xylem sap composition was revealed, but also a common one implying that effector proteins not only exert non-redundant activities but also have shared functions.

### MATERIALS AND METHODS

#### Plant Growth Conditions

Tomato (Solanum lycopersicum) variety C32 was used for the Fol-disease assays and is susceptible to Fol races 2 (Fol007) (Kroon and Elergsma, 1993). Seedlings were transferred to pots 10 to 14-days-after-sowing and subsequently grown in a climatized greenhouse at 25◦C, 65% relative humidity and a 16 h photoperiod.

### Generation of the SIX2 Knockout-constructs

To create the SIX2 knockout construct, the DNA sequence from the position 1870–64 bp upstream of the SIX2 open reading frame was PCR-amplified using primers with HindIII and XbaI linkers (5′ -AAAAAGCTTGGACCGTACATAATGC TGCA-3′ and 5′ -AAATCTAGAGCGGATAGAGATGAGATGA-3 ′ ) and inserted into the binary vector pRWh2 next to the hygromycin resistance cassette (Houterman et al., 2008). The sequence from 147 to 870 bp downstream of the SIX2 stop codon was amplified using primers containing a KpnI linker (5′ - AAAGGTACCAAATCTATCCTCCAGGTT-3′ and 5 ′ - AAAGGTACCATCATGCACGTTAATGAAAGTA-3′ ), and inserted on the other side of the hygromycin cassette.

### Transformation of Fol, Targeted Knockout of SIX2 and Pathogenicity Test

The transformation was performed using Agrobacterium tumefaciens mediated transformation (Takken et al., 2004) as described before (Gawehns et al., 2014). Briefly, spores of Fol007 (2 × 10<sup>6</sup> spores/ml) and A. tumefaciens carrying the SIX2 knockout-construct were co-cultivated on ME25 filters placed on IM plates (10 mM glucose, 10 mM K2HPO4, 10 mM KH2PO4, 2.5 mM NaCl, 4 mM (NH4)2SO4, 0.7 mM CaCl, 2 mM MgSO4, 9µM FeSO4, 0.5% (w/v) glycerol, 5mM glucose, 1.5% Bacto-agar, 40 mM MES pH 5.3). Putative transformants were obtained after transfer of the filters on CDA (Czapek Dox Agar) containing cefotaxime. The hygromycin resistant monospores were PCR tested for successful transformation and deletion of SIX2 using primer pairs 5′ - TGGGCGGAATATATGACCAT-3′ / 5′ -GCATGTTTCTTCCTT GAACTCTC-3′ and 5′ -TAGAGATCATGCTATATCTC-3′ /5′ - CGACACTCGCTTATCATGCA-3′ .

To analyze pathogenicity of the SIX2 knockout, 10-day-old seedlings were uprooted from the soil and for 5 min inoculated with a 5-day-old spore suspension (10<sup>7</sup> spores/ml) of the SIX2 knockout, the Fol007 wild-type or mock-treated (no spores) and subsequently potted (Wellmann, 1939). Disease symptoms of 15 plants/treatment were scored by means of plant weight and disease index (Gawehns et al., 2014) after 3 weeks. Significant differences between the treatments were tested using ANOVA (Fisher PLSD significant at 95%) and presented by the clustering they show in a dot plot.

### Plant Inoculations and Xylem Sap Collection

Spore suspensions (0.5 × 10<sup>7</sup> spores/ml) were prepared from 5-day-old cultures of Fol007, 1AVR2, 1AVR3, 1SIX2, 1SIX5, and 1SIX6. The soil, and part of the main root system, of 4-week-old C32 tomato plants was removed. Twenty-five plants per replicate were inoculated with Fol007, the knockout strains, or were mock-treated (water without spores) for 5 min and planted (Wellmann, 1939). Upon appearance of the disease symptoms (formation of air-roots, yellowing and wilting of the lower leaves), at approximately 14 dpi the xylem sap was collected as described (Rep et al., 2002; Krasikov et al., 2011). Briefly, plants were watered and the temperature was set at 22◦C. The stems were cut just below the first real leaves and plants were placed horizontally to "bleed" for 6 h into a 12 ml polystyrene tube that was placed on ice. The collected xylem sap was stored at −20◦C until further processing. Inoculation and xylem sap harvesting was independently repeated four times/experiment and the experiments were carried out in four subsequent weeks. Experiment 1 was done in the autumn of 2011, Experiment 2 early spring 2012.

#### Sample Preparation and Mass Spectrometry and Label-free Quantitative Proteomics

Analysis of the samples was performed as described before (Schmidt et al., 2013). In summary, after removing the spores from the xylem sap by centrifugation, the sap of 25 plants was concentrated and trichloroacetic acid/aceton precipitated xylem sap samples (45µg of protein) were boiled in SDS loading buffer (2% SDS, 10% glycerol, 50 mM Tris pH 6.8, 100 mM DTT, 0.05% bromphenol blue) and loaded on a SDSpolyacrylamide gel and shortly (1 cm) separated on SDS-PAGE (Mini-PROTEAN gel electrophoresis, Bio-Rad). Proteins were stained with Commassie PageBlue (ThermoFisher) revealing approximate equal amounts of proteins for each treatment. From each sample one slice containing all proteins was cut from the gel. Following in-gel digestion (Rep et al., 2002) the samples were dissolved into 50µl 1 ml/l formic acid in water and the obtained peptides were analyzed by nanoLC-MS/MS. The samples were analyzed by injecting 18µl sample over a 0.10 <sup>∗</sup> 32 mm Magic C18AQ 200A 5µm beads (Michrom Bioresources Inc., USA) pre-concentration column (prepared in-house) at a constant pressure of 270 bar (normally resulting in a flow of ca. 7µl/min). Peptides were eluted from the pre-concentration column onto a 0.10 <sup>∗</sup> 250 mm Magic C18AQ 200A 3µm beads analytical column (prepared in-house) with an acetonitril gradient at a flow of 0.5µl/min with a Proxeon EASY nanoLC. The gradient consisted of an increase from 8 to 33% acetonitril in water with 5 ml/l acetic acid in 50 min followed by a fast increase in the percentage acetonitril to 80% (with 20% water and 5 ml/l acetic acid in both the acetonitril and the water) in 3 min as a column cleaning step.

A P777 Upchurch microcross was positioned between the preconcentration and analytical column. An electrospray potential of 3.5 kV was applied directly to the eluent via a stainless steel needle fitted into the waste line of the microcross. Full scan positive mode FTMS spectra were measured between m/z 380 and 1400 on a LTQ-Orbitrap XL (Thermo electron, San Jose, CA, USA) in the Orbitrap at high resolution (60,000). CID fragmented (Isolation width 2 m/z, 30% normalized collision energy, Activation Q 0.25 and activation time 15 ms) MSMS scans of the four most abundant 2 and 3+ charged peaks in the FTMS scan were recorded in data dependent mode in the linear trap (MSMS threshold = 5.000, 45 s exclusion duration for the selected m/z ± 25 ppm).

The MaxQuant software (Cox and Mann, 2008; Hubner et al., 2010) and MaxQuant 1.1.36 settings (Peng et al., 2012) were used to analyze the raw data generated by an LTQ-OrbitrapXL for protein identification and label-free quantification. The principles and algorithms underlying the label free quantification (LFQ) method is described elsewhere (Cox et al., 2014). Default settings for the Andromeda search engine were used (Cox et al., 2011) including 1% FDR cutoff values for peptides and proteins, except that extra variable modifications were set for de-amidation of N and Q. Identification of the tomato proteins was based on the SGN tomato protein database ITAG2 version 3 (34,727 entries) (ftp://ftp.solgenomics.net/../../ proteins/protein\_predictions\_from\_unigenes/single\_species\_

assemblies/Solanum\_lycopersicum/), while for fungal proteins the database from the Fusarium Comparative Genome website (http://www.broadinstitute.org/annotation/genome/fusarium\_

group/MultiHome.html) was used. The latter database (total 17,652 entries) was supplemented by adding the sequences of Six proteins that are not annotated in the public database. A "contaminant" database (59 entries) was used to identify proteins such as trypsin and human keratins (Peng et al., 2012). The "label-free quantification" as well as the "match between runs" (set to 2 min) options were enabled. De-amidated peptides were allowed to be used for protein quantification and all other quantification settings were kept default. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (Vizcaíno et al., 2013) via the PRIDE partner repository with the dataset identifier PXD003010.

Filtering and further bioinformatics analysis of the MaxQuant/Andromeda workflow output and the analysis of the abundances of the identified proteins were performed with the Perseus 1.3.0.4 module (available at the MaxQuant suite). Accepted proteins had at least 2 identified peptides of which at least one should be unique and at least one should be unmodified. Reversed hits were deleted from the MaxQuant result table. LFQ values were log10 transformed.

#### Data Processing

For proteins, which were not detected in one sample but were present in one of the other samples of the same experiment, fixed log10 LFQ values (6.0 for Experiment 1 and 5.5 for Experiment 2) were imputed. Proteins that were quantified by at least 2 peptides in at least 3 out of 4 biological replicates were annotated as "Present" (P), those quantified in 1 or 2 biological replicates were referred to as "Marginal" (M) while the others were called "Absent" (A). To allow for making statements about significance and fold change all proteins (P+M) were kept in the datasets of each experiment for follow-up analyses. Data handling was performed mostly in the R-2.15.1 program (http://www.r-project. org/) together with the bioconductor (http://www.bioconductor. org/) packages limma and maanova. As technical differences between Experiments 1 and 2 were present, these datasets were handled independently from one another on the level of data processing and low-level analysis.

Subsets of plant specific proteins were generated by removing the fungal and contaminating non-plant proteins. Next, the log10 LFQ values were median normalized over the samples per experiment to obtain more comparable data distributions. PCA plots of these normalized data were then generated. For each dataset the following mixed linear model was fitted per protein on all samples (Wolfinger et al., 2001; Cui and Churchill, 2003):

$$\mathcal{Y} = \mu \, \, \, \, T \, \, \, \, W \, \, \, \, \varepsilon$$

Where µ captures the average protein abundance, T captures the variation for the different treatments, W is a random factor that captures the variation for the different weeks in which the experiments were performed and ε is the residual error for each individual sample.

For hypothesis testing a permutation-based F1 test (2000 permutations) was applied to the values for treatment from the model fit with all pairwise comparisons of each samples to one another (Cui and Churchill, 2003). False discovery rate (FDR) correction was performed using q-values (Storey and Tibshirani, 2003) over all tests from the different comparisons combined. This is an approach that is specifically designed to correct for false discovery in multiple-hypothesis testing with features that are represented in a genome. FDR adjusted p-values <0.1 were considered as statistically significant (**Supplemental Table 1**). This workflow was adapted from Ting et al. (2009).

#### Gene Ontology (GO) Analysis

To batch-classify the xylem sap proteins into functional plant categories the software Mercator (http://mapman.gabipd.org) was used. Mercator uses MapMan categories, which categorize proteins in metabolic pathways and enzyme functions. For the analysis three sequence classifications were performed: Blast searches against Arabidopsis TAIR10, plant proteins from swissprot and UniRef90, RPS-Blast searches against cdd and KOG and an InterPro scan. Default settings were used to analyze the xylem sap proteins. The obtained MapMan bin-codes were sorted manually into 10 gene ontology classes as shown in **Supplemental Table 2** and every protein was assigned a single bin-code.

#### RESULTS

#### Characterization of the Tomato Xylem Sap Proteome upon Fol Infection

To determine the xylem sap proteome composition and the quantitative changes therein due to specific effector proteins, we collected xylem sap from tomato plants inoculated with F. oxysporum isolate Fol007 and different effector knockout strains in the Fol007 background. Besides the previously described AVR2, AVR3, SIX5, and SIX6 effector knockout strains (Rep et al., 2005; Houterman et al., 2009; Schmidt et al., 2013; Gawehns et al., 2014) also an SIX2 knockout strain was included. Three SIX2 knockout strains were made, of which only one strain (1SIX2#1) showed reduced virulence in a seedlings bioassay, resulting in increased plant weight and a reduction of the disease index (i.e., vessel browning, stunting and wilting, Gawehns et al., 2014); the other two knockout strains and a reference strain with the SIX2 knockout construct ectopically inserted into the genome behaved like wild-type (**Figure 1**). Transformant 1SIX2#1 was used for further analysis in this study.

Two experiments were performed in a climatized greenhouse, one in autumn (Experiment 1) and one in spring (Experiment 2). Besides mock and Fol007 inoculations the 1AVR2 and 1SIX5 strains were included in Experiment 1, while 1AVR3, 1SIX2, and 1SIX6 were used in Experiment 2 (see below). Roots of 4-week-old plants were inoculated and xylem sap was collected approximately 14-days-post-inoculation (dpi) after early disease symptoms developed (formation of air-roots, yellowing and wilting of the lower leaves) (Rep et al., 2002). The procedure resulting in the identification and quantification of the xylem sap proteome is schematically depicted in **Figure S1**. Briefly, the xylem sap of the 25 plants/treatment was pooled, resulting in a typical yield of 15–70 ml. The collected xylem sap proteins were concentrated, separated (SDS-polyacrylamide gel) and subjected to in-gel tryptic digestion. Identification of the released peptides was done using nLC-MS/MS mass spectrometry followed by MaxQuant analysis to identify and quantify the progenitor proteins (**Supplemental Tables 3**, **4**) For reliable identification only proteins were included that matched with at least two peptides of which at least one was unique.

In some cases, proteins were not found consistently among all replicates of a treatment. To allow inclusion of these proteins in subsequent analyses the following definitions were used:

FIGURE 1 | One SIX2 knockout is reduced in virulence. Susceptible tomato seedlings were inoculated with wild-type Fol (Fol007) or strains in which *SIX2* was deleted (1*SIX2*#1-3). As controls mock-inoculated plants were used (mock) or one transformant (ect#1), in which the *SIX2* deletion construct was integrated ectopically. Average plant weight of 15 plants was plotted against the average disease index of the same plants. In 1*SIX2*#1 pathogenicity was impaired as shown by the increased plant weight as compared to infection with Fol007 and the ectopic transformant. Error bars represent the standard error. Clustering is based on an ANOVA (Fisher PLSD significant at 95%) using either the disease index (solid line) or plant weight (dashed line).

"Absent" (A) for proteins that were not identified in any of the four replicates, "Marginal" (M) for proteins found in only one or two replicates or "Present" (P) when found in three or all four replicates. A protein was considered "found" when either labeled "Marginal" or "Present." To allow for statistical testing of significant differences between the xylem sap proteins of mock and Fol-inoculated plants the LFQ intensity data, with imputed values for "Absent" proteins, were normalized using log10 transformation and a median scaling of the protein abundance data distributions over the replicates of all treatments for each experiment separately (**Figure S2**). **Figure 2** shows a principal component analysis (PCA) plot, generated from these normalized data, representing all treatments and replicates per experiment. The distribution of all replicates of each treatment was found to be comparable along the PC1-axis. This even distribution indicates that each replicate was influenced by similar effects allowing the replicates of each treatment to be analyzed together. In Experiments 1 and 2, respectively, a total of 343 and 292 plant proteins (**Supplemental Table 1**), 43 and 19 Fol proteins and 7 contaminating proteins (e.g., trypsin, keratin etc.) were identified and quantified. Analysis of the Fol-encoded proteins has been reported elsewhere (Schmidt et al., 2013); here we solely focus on the plant proteome.

To assess the purity of the identified xylem sap the proteome was analyzed for the presence of typical proteins from xylem and absence of phloem-derived and intracellular proteins. Typical xylem sap proteins that are conserved among different plant species, like peroxidases, glycine-rich proteins, serine and aspartyl proteases, chitinases and lipid transfer proteinlike polypeptides (Buhtz et al., 2004) were present in both experimental sets. In addition 16 intracellular proteins were identified, including enzymes like Ribose-5-phosphate isomerase, Triose-phosphate-isomerase, Fructose-bisphosphatealdolase and Fructose-1,6-bisphosphatase catalyzing reactions in gluconeogenesis, glycolysis and the Calvin cycle. Also a Polyphenoloxidase was identified, which is a characteristic protein for phloem sap (Dafoe and Constabel, 2009). Other typical phloem proteins, like PP1 and WIN4 (Dafoe and Constabel, 2009; Subramanian et al., 2009), or the intracellular Ribulose-1,5-bisphosphate carboxylase however, were absent. Notably, of the 16 intracellular proteins only one, a ferredoxin-related protein (Solyc07g063740.2.1), was consistently found in all 4 replicates in Experiments 1 and 2. Only 3 of the 15 others were found in both experiments and their presence where mostly scored as marginal/absent in Experiment 2. Together, this suggests some, but minimal contamination of phloem sap in the xylem sap preparation.

### The Tomato Xylem Sap Proteome Composition Changes Distinctively after Fol Infection

To assess the effect Fol has on the xylem sap proteome composition, the identities of the plant proteins present in the Fol- and mock-treated plants of Experiment 1 were compared. The data obtained in Experiment 1 were preferred over the data set of Experiment 2 because of the larger number of proteins identified (342 vs. 292). In Experiment 1 297 and 330 tomato proteins were identified in Fol- and mock-treated plants, respectively. The possible functions of the xylem sap proteins were determined by Gene Ontology (GO) annotation using the plant optimized and homology-based annotation software MapMan (Klie and Nikoloski, 2012; Lohse et al., 2014). The obtained bin-codes were classified into 10 categories: "cell wall," "metabolism," "stress responses," "redox," "peroxidases," "DNA/RNA," "protein modification," "signaling," "others" including mostly photosynthesis and development related proteins and "not assigned" (**Supplemental Table 2**). The largest category is formed by proteins belonging to "stress responses" with 21 and 20% of the total proteins found in the Fol and mock proteome respectively (**Figure 3A**). This is followed by the categories "protein modification and degradation," "cell wall," "peroxidases," and "metabolism." The categories containing the lowest number of proteins are "others," "redox," and "signaling" The latter group is most distinct in relative size between Fol and mock. In general, however, the size and composition of the categories are similar for both treatments.

Only 12 proteins were identified exclusively in Fol-treated plants, while 45 were solely found in the mock. In total, 285 plant proteins were identified in both Fol and mock (**Figure 3B**). A data pipeline was developed to identify proteins whose abundance changes significantly following infection. Thereto, a

mixed linear model was fitted per protein on all samples to correct the data for the non-treatment related noise caused by sampling the replicates in four different weeks, and to estimate the effect of the different treatments. Next, the contrasts between xylem sap proteins in mock and Fol-inoculated plants were tested for significant differences using 2000 permutations to relax restrictions of normality. Subsequently, a correction for false discovery rate was done (p < 0.1). Proteins exhibiting a significantly altered abundance after Fol inoculation compared to the mock treatment were called DAPs, for Differentially Accumulated Proteins (DAPs). This analysis revealed that of the 285 plant proteins common to both datasets, the abundance of 27 proteins was unaltered (**Figure 4**, top left panel, blue circles and **Supplemental Table 5**), whereas that of the other 258 proteins varied significantly (p < 0.1) between Fol and mock treated plants (**Figure 4**, top left panel, red circles). The abundance of 156 proteins was decreased (**Figure 4**, red circles left of the "0" axis) and that of 102 proteins increased upon Fol inoculation (**Figure 4**, red circles right of the "0" axis and **Supplemental Table 1**). In total, 315 DAPs were identified; 258 are present in both Fol and mock samples, 12 are unique for Fol treatment and 45 are unique for the mock. **Figure 3C** shows the distribution of the 315 DAPs over the GO categories. We separated them in two groups: the "up" group consisting of emerging proteins and those whose abundance increases upon Fol treatment, and the "down" group contains all DAPs with reduced amounts and proteins undetected after Fol treatment. Most proteins in the "down" group belong to the categories "cell wall" or "stress responses." These categories each account for respectively 15 and 16% of the total dataset. The category "protein

FIGURE 4 | The number of DAPs differs substantially between the different contrasts that were tested. Volcano plots showing the relation between -log10 False Discovery Rate-corrected *P*-values and log10 protein abundance ratios, based on the group average per treatment from the linear model fit. Blue circles indicate proteins that are not differential while red circles were annotated as DAPs. The black encircled DAPs, with names included, represent the top-10 combinations of lowest P-value and strongest up- and down-regulated proteins of Fol compared to Mock (top left), dAVR2 compared to Fol (top right), dSix5 compared to Fol (center left), dSix6 compared to Fol (center right), dAvr3 compared to Fol (bottom left) and dSix2 compared to Fol (bottom right). Names of the proteins can be found in Table 2.

modification and degradation" covers 14%. Of the "up" DAPs the category "cell wall" is depleted and represents only 5% of the proteins. The same pattern is observed regardless whether the 45 mock-unique proteins were included in the analysis or not (data not shown). The "up" DAPs were especially enriched for the category "stress responses," harboring 27% of the total number of identified proteins.

In summary, the abundance of 92% of the tomato proteins present in the xylem sap changes after infection while that of only 8% remained unaltered. Of the altered proteins 6% was identified solely after Fol inoculation. Hence, upon infection the overall xylem sap proteome composition—in terms of GO categories—remains relatively constant, but the abundance of specific proteins changes distinctively. Generally, the abundance of proteins related to stress-responses increased while those having cell wall-related activities became less abundant.

#### Data from Different Experiments can be Combined

Also in conditioned greenhouses the symptoms of Fusarium wilt disease are subject to seasonal influences. **Figure 5A** shows plants treated with either water or with one of the three Fol strains (Fol007, 1SIX2, and 1SIX6) in two different seasons (summer 2010 and early spring 2012). Inoculation in spring typically results in shorter plants, later onset of wilting symptoms, and a stronger epinastic response as compared to bioassays performed in summer. Nevertheless, the same relative effect of the inoculations on disease symptoms were observed: plants inoculated with Fol007 or 1SIX2 showed the severest disease symptoms while inoculation with 1SIX6 or mock treatment caused mild or no symptoms.

To assess the overall quality and comparability of the data sets obtained in Experiments 1 and 2, PCA analysis of the normalized LFQ values of all plant proteins was performed for the mock- and Fol-treated plants. The PCA plots for both data set 1 (**Figure 5B**) and data set 2 (**Figure 5C**) show a clear separation of the mock treatment from the Fol treatment on the PC1 axis. Hence, the biological effects were stronger than the technical effects and therefore we considered the quality of the data to be sufficient for further analysis. In addition, a PCA based on all proteins found in the four datasets (**Figure S3**) shows that, although the technical and biological variation explains most of the variance in the data (PC1), the Fol007 and mock clearly separate on the PC2 axis for both experiments. To determine the similarity between both sets on the biological level, a GO analysis was performed using the xylem sap proteome of either mock- or Fol-inoculated plants. The results of this analysis are plotted in a bar chart (**Figure 5D**). The largest group of proteins in all treatments belonged to the category "stress response," followed by the categories "protein modification and degradation," "cell wall," "peroxidases," and "metabolism." The categories "signaling," "redox," "DNA and RNA related," "others," and "not assigned" were least represented. In the data set of Experiment 1 relatively more metabolism-related proteins were identified, whereas in Experiment 2 relatively more proteins with a function in protein modification and degradation were detected. The abundance of proteins assigned to "others" and "DNA and RNA related" was relatively high in Experiment 1 compared to Experiment 2.

A correlation analysis was performed with the data from Experiments 1 and 2 (**Figure S4**) to analyze the comparability of the data. In the scatter plot the data distribute on a linear line through 0 by trend. The Pearson correlation coefficient of this analysis is 0.6, which indicates a positive correlation between the data sets. In conclusion, although differences between the severity of disease symptoms were observed between Experiments 1 and 2 both data sets are similar regarding overall biological differences and data-quality, allowing direct comparison of the DAPs identified in both sets for further analysis.

### Avr2, Avr3, Six5, and Six6 have Specific and Common Effects on the Xylem Sap Proteome Composition

Next, we examined whether reduced pathogenicity can be correlated to a specific change in the xylem sap proteome composition. Comparing the DAPs identified in Experiments 1 and 2 revealed changes in the xylem sap proteome after inoculation with the five Fol effector knockout strains. To analyze the differences in the xylem sap proteome in relation to the absence of specific effectors, the DAPs for each knockout were determined using a p < 0.1. The contrast between the AVR2 knockout and Fol wild-type revealed 209 DAPs; 37 DAPs were found for 1AVR3, 0 for 1SIX2, 206 for 1SIX5 and 115 for 1SIX6 (**Figure 4**). The absence of DAPs between 1SIX2 and Fol wild-type is noticeable and corroborates the chosen settings for FDR and p-values as being sufficiently stringent. Because the xylem proteome composition after inoculation with the 1SIX2 knockout was identical to that of the wild-type Fol (**Figure 4**, bottom right panel) these contrasts are not depicted separately in the subsequent analyses. Note that also the disease symptoms of the 1SIX2 Fol inoculation of the 4-week-old plants were indistinguishable from that of the wild-type (**Figure 5A**).

The overlap between the DAPs from different knockouts is depicted in **Figure 6A**. In line with their combined function in the plant (Ma et al., 2015), the largest number of DAPs (110) is shared between the AVR2 and SIX5 knockouts. Twentyfour DAPs are shared by the AVR3, AVR2, SIX5, and SIX6 knockout strains (**Figure 6A**). Ten of those DAPs belong to the category "stress response" and are mostly PR proteins (**Table 2** and **Figure 6B**). Also four proteases from the category "protein modification and degradation" are present in this set. In all xylem sap samples from plants treated with one of the above four effector-knockout strains the abundance of those proteases was decreased compared to the Fol wild-type. Also from each of the categories "signaling," "redox," "peroxidases," and "metabolism," two proteins belonged to the common DAPs. These 24 DAPs are likely virulence-associated DAPs as their abundance is affected by all Fol strains that show a reduction in virulence and nearly half of them have a function in the stress response (**Table 2**).

Next, DAPs were identified whose abundance uniquely alters in xylem sap isolated from plants infected with a specific effector knockout strain. The gene ontology analysis of the specific DAPsets showed a profile distinct from that of the common DAPs

(**Figure 6B**). The largest change was observed in the percentage of proteins belonging to "stress responses" covering 42% for the common DAPs and, e.g., only 10% for the AVR2-specific DAPs. The profiles of the specific DAPs differed significantly from each other, as each profile was unique for a specific knockout.

sap proteins per GO category for the Fol and mock treatments of data set 1 and data set 2.

In more detail, six DAPs were found in the xylem sap from plants inoculated with the Fol AVR3 knockout when compared to the Fol wild-type (**Figure 6A**). These DAPs are a Kunitz-type proteinase inhibitor (Solyc03g098710.1.1), which is increased in abundance in the xylem sap, a polygalacturonase-like protein (Solyc09g075460.1.1), two serine carboxypeptidases (Solyc06g083040.1.1 and Solyc02g088820.1.1), one superoxide dismutase (SOD) (Solyc01g067740.1.1) and one PR-2 (Solyc04g080260.1.1 a Glucan endo-1 3-beta-glucosidase) protein, of which abundances were all decreased in the absence of Avr3. Three of these DAPs are unique to the AVR3 knockout (Solyc09g075460.1.1, Solyc06g083040.1.1, and Solyc03g098710.1.1) while the other three (Solyc01g067740.1.1, Solyc04g080260.1.1, and Solyc02g088820.1.1) exerted an Avr3 specific change in their abundance. For instance PR-2 abundance (Solyc04g080260.1.1) is decreased in 1AVR3 while it is increased in the 1SIX5 strain as compared to xylem sap of Fol wild-type infected plants. The two other DAPs are reduced in abundance in the Avr3 knockout while their abundance is increased in both the Avr2 and Six5 knockouts, bringing the total number of Avr3-specific DAPs to six.

In the xylem sap of tomato plants that were inoculated with the AVR2 knockout strain 21 unique DAPs were found (**Figure 6A**). Also one DAP, a Endo-1 4-beta-xylanase (Solyc11g040330.1.1), is found that is shared with the SIX6 knockout but it abundance is increased rather than decreased in the 1AVR2 strain. Those proteins represent all GO categories,

except "signaling," "others," and "DNA/RNA related." Notably, only 10% of these DAPs belong to "stress responses," 24% to "cell wall" and 19% to "others" (**Figure 6B**). **Table 1** shows all 22 1AVR2-specific DAPs with either increased or decreased abundance. Next to common DAPs like PR proteins, peroxidases, proteinases and different hydrolases, a Translationally Controlled Tumor Protein (TCTP) homolog and a Group II intron splicing factor CRS1-like protein was found. The first one showed a decrease in abundance the latter an increase.

In the xylem sap proteome of SIX5 knockout strain-treated plants 18 unique DAPs were identified (**Figure 6A**). When not only the absolute change in protein abundance was taken into account, as compared to Fol wild-type infected plants, but also the direction, in- or decrease, 6 additional 1SIX5-specific DAPs were found: an unknown Protein (Solyc10g074820.1.1), a Peroxidase (Solyc09g007520.1.1), a Kunitz trypsin inhibitor (Solyc06g072230.1.1), a Glucan endo-1 3-beta-glucosidase (PR-2), (Solyc04g080260.1.1), a Receptor-like kinase (Solyc01g108840.1.1) and a Serine carboxypeptidase K10B2.2 (Solyc12g099160.1.1) (**Table 2**). The category "metabolism" is overrepresented as compared with the 1AVR2-specific DAPs as 39% of all 1SIX5-specific DAPs belong to this category (**Figure 6B**). No proteins from the category "cell wall" were found. Generally, more DAPs were increased (17) than decreased (7) in this xylem sap proteome. Remarkably, the enzymes phosphoglucomutase (Solyc04g045340.1.1) and Fructose-1-6-bisphosphatase (Solyc05g052600.1.1) were identified in the latter group of DAPs, which have catalytic functions in glycogenesis.

The 29 specific DAPs, and the 15 DAPs (Solyc12g099160.1.1, Solyc11g040330.1.1, Solyc10g074820.1.1, Solyc09g007520.1.1, Solyc06g072230.1.1, Solyc01g108840.1.1, Solyc09g007010.1.1, Solyc08g066810.1.1, Solyc06g072220.1.1, Solyc05g054710.1.1, Solyc05g052280.1.1, Solyc04g072000.1.1, Solyc02g077040.1.1, Solyc02g024050.1.1, and Solyc01g105070.1.1) whose abundance was either specifically increased or decreased in the xylem sap of Fol SIX6 knockout compared to other effector knockout strains inoculated tomato plants (**Table 2**, **Figure 6A**) distribute over all GO categories except "others." Relatively more DAPs belonging to the categories "protein modification and degradation" and "redox" were identified than in Fol007 vs. mock (**Figure 6B**) and relatively fewer proteins from "stress responses," "peroxidases," and "cell wall." All peroxidases were decreased in abundance while the abundance of protein-degrading enzymes was mostly increased (**Table 2**).

In summary, knockout of a single effector gene in Fol specifically affects the composition of the xylem sap proteome of infected tomato plants. All effector knockout strains that were compromised in virulence affected the abundance of a core set of 24 xylem sap proteins. In addition, each effector knockout strain had a unique effect as shown by their specific GO fingerprint profile affecting distinctive DAPs. The SIX2 knockout did not affect the xylem sap proteome in comparison to the wild-type. This is in line with the observation that disease development of tomato plants infected with the SIX2 knockout strain is identical to that of plants infected with the wild-type strain.

#### DISCUSSION

#### Changes in the Xylem Sap Proteome upon Fol Infection

The tomato xylem sap proteome changes dramatically upon infection with Fol. Using a quantitative MS approach, 388 plant and 43 fungal proteins were identified when the data from both experiments were merged. When focusing on Experiment 1, 12 new tomato proteins were found while 45 proteins were no longer detected in the xylem sap proteome as compared to mock-infected plants. Gene ontology analysis revealed that mostly proteins from the categories "stress responses" (mostly PR proteins), "signaling" (many LRR class receptor like kinases), "protein modifications" (many peptidases) and "not assigned" were no longer detectable. This suggests that either expression of the encoding genes is reduced or that protein turnover or location is altered resulting in reduced abundances.

Of all proteins identified in both experiments 97% showed changes in abundance following infection. Xylem sap proteins are produced in the cells adjacent to the root xylem and subsequently secreted into the xylem sap (Kehr et al., 2005). Upon Fol


#### TABLE 1 | List of proteins, whose abundance changes specifically upon infection with a SIX knockout as compared to wild-type Fol.


*u, increased abundance; d, decreased abundance, Log10 Label Free Quantification (LFQ) values compared to Fol Wildtype (Fol007) and adjusted P-values (adjPval) are depicted. sr, stress responses; s, signaling; r, redox; po, peroxidases; p, protein modification/degradation; o, others; na, not assigned; m, metabolism; d, DNA/RNA related; cw, cell wall.*


*u, increased abundance; d, decreased abundance compared to xylem sap infected with Fol-wildtype; sr, stress responses; s, signaling; r, redox; po, peroxidases; p, protein modification/degradation; o, others; na, not assigned; m, metabolism; d, DNA/RNA related; cw, cell wall.*

infection, expression of xylem sap protein-encoding genes is likely to differ between colonized and uninfected tissues. Since the xylem sap proteome is produced by a combination of infected and non-infected root and xylem tissue, it is surprising that a relatively large number of proteins (45) disappeared from the proteome upon infection. This suggests a systemic signal originating from infected tissues that affects gene expression in healthy tissues. The nature of this signal is unknown, but could be one of the known root generated systemic signals (Fu and Dong, 2013).

In previous studies employing 1-D and 2-D gel electrophoresis, thirteen proteins, including PR-1, two PR-2 isoforms, PR-3, PR-5, and peroxidases, were found to accumulate in tomato xylem sap following inoculation with Fol (Rep et al., 2002; Houterman et al., 2007) whereas abundance of the Lipid Transfer Like protein (LTP) XSP10 was decreased (Rep et al., 2002, 2003). Silencing XSP10 showed that the encoded protein is required for disease symptom development (Krasikov et al., 2011). Since expression of the XSP10 gene is constitutive during infection it was suggested that either synthesis or secretion of XSP10 is suppressed or that XSP10 is turned over during infection (Rep et al., 2003). In this study a clear reduction in the abundance of XSP10 was observed in Experiment 1, but not in Experiment 2. Since Experiments 1 and 2 were similar regarding both biological and data-quality, this difference is noteworthy. Sampling of both experiments was performed at similar time points after infection. However, in the second experiment plants clearly showed less disease symptoms. This lower disease severity corresponds to the lower number of fungal proteins identified in the latter experiment (19 vs. 43), suggesting a less successful Fol infection (**Figure 5A**). Plant stage, fitness and the nutritional status as well as environmental conditions are known to determine the speed and severity of disease development (Yadeta and Thomma, 2013). Likely, those factors are affected by the season, explaining the quantitative difference in disease development and severity even though the experiments were conducted in climatized greenhouse compartments. A reduced severity of fungal infection may have caused the less effective reduction of XSP10 abundance in Experiment 2. However, because LFQ values between Experiments 1 and 2 were not directly comparable, we could not estimate whether the absolute abundance of XSP10 was indeed higher in Experiment 2 than in Experiment 1.

#### One Fourth of All Xylem Sap Proteins does not Carry a Classical Signal Peptide

As xylem vessels consist of dead cells, xylem sap proteins originate from neighboring living cells. The vast majority of extracellular proteins are secreted through the endoplasmic reticulum—Golgi pathway (Kehr et al., 2005). Secretion of such proteins is mediated by an amino-terminal signal peptide and the majority (283) of the identified xylem sap proteins are indeed predicted to contain such a signal peptide. Of the 105 proteins that did not have a predicted signal peptide, 77 were suggested by SecretomeP to use the non-classical secretory pathway (N score > 0.5) ("unconventional secreted proteins"). The 28 remaining xylem sap proteins contained neither a signal peptide nor were predicted to be unconventionally secreted. Two of these, a beta-1,3-glucanase (PR-2) (Solyc02g086700.2.1) and a 5-methyltetrahydropteroyltriglutamate-homocysteine methyltransferase (Solyc10g081510.1.1), were highly abundant in both data sets having LFQ values larger than nine. Similarly, in the xylem sap of Brassica oleracea a putative intracellular methionine synthase was identified in relatively high amounts next to other less abundant intracellular proteins (Ligat et al., 2011). The overall low number of putative intracellular proteins retrieved in the xylem sap in our experiments makes it unlikely that these are contaminants originating from tissues other than xylem vessels. Possibly, these 28 proteins originated from developing xylem precursor cells that are released after cell death to form tracheary elements (Dafoe and Constabel, 2009; Yadeta and Thomma, 2013). Alternatively, annotation of some of these proteins could be incorrect or incomplete, resulting in a mis-annotation of the N-terminus (i.e., missing the signal peptide).

It has been proposed that non-classical protein secretion is preferably activated upon biotic and abiotic stresses (Agrawal et al., 2010). If so, one expects enrichment of proteins lacking a signal peptide in the proteome of Fol-infected plants as compared to the mock. To see if this was the case we looked at the abundance of the 77 xylem sap proteins without a predicted signal peptide before and after infection. Of the 12 plant proteins unique for the "infected" proteome only five are predicted to have a signal peptide—the other eleven do not. In the control proteome the majority (35) of the 45 unique proteins were predicted to contain a signal peptide. This appears to support the hypothesis of increased non-classical secretion upon infection. Within the population of DAPs present in both the "mock-" and "Fol-treated" proteome, no clear over-representation could be detected of proteins lacking a signal peptide, regardless of whether their abundance increased or decreased.

Regarding the total protein dataset from both experiments, the majority of proteins related to "stress responses" had a signal peptide (**Figure 7**). The same was found for "peroxidases," "signaling related proteins" and proteins associated to protein modification and degradation. Relatively speaking, proteins with a function in the "redox system," in "metabolism" or in "cell wall" were more likely to be secreted via an alternative route. Apparently, non-classical-secretion preferably occurs for specific functional categories of xylem sap proteins, but the relationship between infection and their route of secretion needs to be clarified further.

#### The Absence of Single Effectors Affects Abundance of Specific Xylem Sap Proteins

Four out of five SIX gene knockouts had both shared and specific effects on the xylem sap proteome composition. These four SIX genes encode virulence factors (Avr3, Avr2, Six5, Six6) (Rep et al., 2005; Houterman et al., 2009; Gawehns et al., 2014). Avr2 and Six5 interact in a Yeast-Two-Hybrid and in planta and both are required to induce full resistance in I-2 carrying plants (Ma et al., 2015). Here we observed a significant higher overlap between the DAPs of the SIX5 and AVR2 knockout than between other knockouts. This finding is in support of a functional overlap of these effectors, although they might also have specific activities.

The SIX2 knockout was slightly compromised in virulence on seedlings, but its disease symptoms in 4-week-old plants were indistinguishable from the wild-type. Correspondingly, its xylem sap proteome was also identical as wild-type, which shows the robustness and reproducibility of the quantification method.

Characterization of effector action by analyzing the xylem sap proteome is a novel strategy to study microbial virulence. The most similar plant study that we found was one where a Pseudomonas syringae hrpA- strain, which is unable to secrete any type III secreted effectors (TTEs), was added to an A. thaliana cell suspension (Kaffarnik et al., 2009). In that study, among others, accumulation of a SOD was suppressed while an enolase was induced by the TTEs. Notably, abundance of a SOD was also significantly decreased in tomato xylem sap upon inoculation with the AVR3 knockout strain. Hence, the presence of Avr3 apparently enhances SOD abundance. As a scavenger of reactive oxygen species (ROS) SOD functions in abiotic stress tolerance and its reduced abundance might be due to the reduced pathogenicity of the AVR3 knockout strain.

Upon inoculation with the AVR2 knockout strain the most notable increase in abundance was observed among cell wall degrading enzymes, specifically galactosidases, which degrade cell walls through galactomannan mobilization, glucosidases, which can hydrolyze xyloglucan but also salicylic acid or abscisic acid, glucosides and endo 1,4 xylanases that convert xylan into xylanose. Also a decrease was found for a TCTP, whose expression levels normally increase in response to abiotic stresses and upon pathogen infection (Berkowitz et al., 2008). In conclusion, Avr2 appears to suppress abundance of cell-wall degrading proteins while it triggers abundance of a TCTP.

The abundance of an enolase was decreased specifically upon inoculation with the SIX5 knockout strain compared to the wildtype. Hence, Six5 may trigger enolase abundance, possibly in a manner comparable to the TTSs of P. syringae. Upon infection with both the AVR2 and SIX5 knockout strains, abundance of a thaumatin-like protein (PR-5) was higher than with WT infection. Furthermore, abundance of two other PR-5 isoforms was higher upon inoculation with both knockouts than upon infection with the wild-type fungus. Hence, the presence of Avr2 and Six5 together appear to have an inhibitory effect on the abundance of PR-5, a protein that has been proposed to display antifungal activities against i.a. F. oxysporum (Hu and Reddy, 1997).

Deletion of SIX6 has the strongest effect on specific DAPs, altering abundance of 44 proteins; 22 proteins increased in abundance and 22 decreased. The largest functional groups of proteins affected are proteases and cell wall degrading enzymes, among which also proteins involved in pectin degradation. Notably, abundance of four peroxidases was reduced. Generally, peroxidases catalyze the oxidation of different substrates using H2O<sup>2</sup> under the production of ROS. This process is active in different physiological processes like phenol oxidation and lignification, which play a role in the defense against pathogens and in cell wall formation (Passardi et al., 2005).

The altered virulence of an effector knockout strain corresponds to a unique "fingerprint" of the xylem sap proteome. We consider two scenarios to explain this observation: (I) the effector indirectly mediates abundance of a set of host proteins by affecting specific signaling pathways that control their expression or (II) the absence of an effector directly changes the xylem sap protein content, for instance by interacting with specific proteins, affecting their turnover or their mobility in the sap. The common effect on the xylem sap proteome composition induced by all effector knockouts except SIX2 complies with the first scenario and is related to the degree of virulence of the strains. The affected group set was dominated by proteins from the group "stress responses," including pathogenesis related (PR) proteins like PR-1, β-1,3-glucanases (PR-2), several chitinases (PR-3) and PR-5 (**Table 2**). This suggests a shared signaling pathway, which is manipulated only upon infection with a fully virulent pathogen, as the abundance of the proteins controlled by this pathway was altered upon infection of a knockout strain. The effector-specific proteome changes could fit either scenario. Future studies, aimed at revealing the identity of the signaling

components targeted by the fungal effector proteins, are required to resolve this issue.

In summary, our study implies that the defense against a vascular pathogen like Fol is mostly based on changes in cell wall remodeling proteins and secretion of proteins with proposed antifungal activity. Individual Six proteins might play exclusive roles to overcome some of those physiological adaptations to allow the fungus to infect and spread.

### AUTHOR CONTRIBUTIONS

FG, LM, and PH collected the xylemsap and prepared the samples for MS analysis. SB performed the LC MS/MS measurements and FG, OB, and SB analyzed the data. BC provided intellectual input and critically evaluated the manuscript. FG, MR, and FT designed the experiments and FG and FT wrote the manuscript. All authors provided intellectual input and approved the manuscript and are accountable for accuracy and integrity of this study.

### ACKNOWLEDGMENTS

FG, SB, and FT are funded by the Center for BioSystems Genomics (CBSG) program (an NGI initiative), LM, OB, PH, FT and BC by the University of Amsterdam and MR by a NWO VICI (project number 865.10.002). All proteomics LC MS/MS measurements were done at Biqualis Wageningen and funded by the CBSG. We are grateful to Ludek Tikovsky, Harold Lemereis and Thijs Hendrix for taking care of the plants, to Alexander Förderer and Fang Xu who assisted with xylem sap collection and Myriam Clavijo for generating the SIX2 knockout.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00967

Figure S1 | Pipeline for the identification and functional annotation of Differential Accumulated xylem sap Proteins (DAPs). Upon inoculation of tomato with different Fol strains the xylem sap was harvested after appearance of the first disease symptoms. The concentrated proteins were shortly run in an SDS-polyacrylamide gel electrophoresis (PAGE), protein-containing bands were cut out of the gel and used for in gel tryptic digestion. The obtained peptides were identified and quantified using nanoLC-MS/MS coupled to a bioinformatics pipeline (MaxQuant). The data were filtered on reliability, normalized and a mixed linear model was fitted. Differences of the xylem sap proteome between different treatments were tested for permutated significance resulting in the identification of DAPs, which were subsequently functionally annotated.

Figure S2 | The protein abundance distributions between replicates and treatments become more comparable after median normalization. Boxplots showing the protein abundance distributions based on the LFQ values for Experiment 1 (top left) and experiment 2 (top right) and after median normalization on the LFQ values for Experiment 1 (bottom left) and Experiment 2 (bottom right).

Figure S3 | Differences between Fol- and mock-treated plants are similar for both experiments. PCA plot based on the combined median normalized log10 data of the Fol- or mock-treated plants from both data sets. PC1 is plotted on the x-axis, PC2 is plotted on the y-axis.

Figure S4 | The data sets of Experiments 1 and 2 correlate. Scatter plot generated from the FC (log10 Fold Change) values obtained in Experiments 1 and 2. The data points align on a line through 0 by trend.

Supplemental Table 1 | Overview of protein annotation information (columns A-G), SignalP analysis (column H), group averaged normalized log10 LFQ values from the ANOVA fit (columns I-Q), "Present," "Marginal" and "Absent" calls (columns R-Z), normalized log10 LFQ ratio data (columns AA-AG), FDR adjusted p-values (columns AH-AN), DAPs

increase or decrease (columns AO-AU), observed F-values from the F1 test (columns AV-BB), processed Max Quant log10 LFQ data (columns BC-CL).

#### Supplemental Table 2 | List of gene ontology terms categories grouped by the corresponding bin-codes.

#### Supplemental Table 3 | Identified proteins and log10 LFQ values for

dataset 1. Table lists, identity (id), protein identity (Protein IDs and Majority Protein IDs), Peptide Counts (unique), Protein Descriptions, Proteins, Peptides, Unique Peptides and peptides identified in the different replica's. The samples are labeled as follows: "non-infected" for mock, "d65" for the Six6 knockout, "07" for Fol 007 wild-type, "dAvr2" for the Avr2 knockout and "d65" for the Six5 knockout. Sequence Coverage [%] and Mol. Weight [kDa] are given for all samples. Log10 LFQ values and MS/MS counts for the samples are provided including alternative identification methods: Only identified by site or Reverse. Contaminants are marked with a +.

#### REFERENCES


#### Supplemental Table 4 | Identified proteins and log10 LFQ values for

dataset 2. Table lists, identity (id), protein identity (Protein IDs and Majority Protein IDs), Peptide Counts (unique), Protein Descriptions, Proteins, Peptides, Unique Peptides and peptides identified in the different replica's. The samples are labeled as follows: "d007" for Fol 007 wild-type, "Six1" for the Avr3 knockout, "Six2" for the Six2 knockout and "Six6" for the Six6 knockout. Sequence Coverage [%] and Mol. Weight [kDa] are given for all samples. LFQ values and MS/MS counts for the samples are provided including alternative identification methods: Only identified by site or Reverse. Contaminants are marked with a +.

Supplemental Table 5 | Differentially identified proteins in the xylem sap of Fol Wild-type (Fol007) and mock-inoculated tomato plants. Protein identity (ID), functional description and Log10 LFQ values of Fol007, Mock and the contrasts are depicted together with adjusted *P*-values (adjPval).


per os infectivity factor complex. J. Virol. 86, 4981–4988. doi: 10.1128/JVI. 06801-11


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2015 Gawehns, Ma, Bruning, Houterman, Boeren, Cornelissen, Rep and Takken. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Proteomic Analyses Provide Novel Insights into Plant Growth and Ginsenoside Biosynthesis in Forest Cultivated Panax ginseng (F. Ginseng)

Rui Ma1, 2 †, Liwei Sun<sup>1</sup> \* † , Xuenan Chen2, <sup>3</sup> , Bing Mei <sup>2</sup> , Guijuan Chang<sup>2</sup> , Manying Wang<sup>1</sup> and Daqing Zhao<sup>2</sup> \*

<sup>1</sup> Jilin Technology Innovation Center for Chinese Medicine Biotechnology, College of Chemistry and Biology, Beihua University, Jilin, China, <sup>2</sup> Ginseng Research Center, Changchun University of Chinese Medicine, Changchun, China, <sup>3</sup> The first affiliated hospital to Changchun University of Chinese Medicine, Changchun, China

#### Edited by:

Dipanjana Ghosh, National University of Singapore, Singapore

#### Reviewed by:

Ramesh Katam, Florida Agricultural & Mechanical University, USA Carla Pinheiro, Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa, Portugal

#### \*Correspondence:

Liwei Sun sunnylilwei@163.com; Daqing Zhao zhaodaqing1963@163.com † These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 16 September 2016 Accepted: 05 January 2016 Published: 26 January 2016

#### Citation:

Ma R, Sun L, Chen X, Mei B, Chang G, Wang M and Zhao D (2016) Proteomic Analyses Provide Novel Insights into Plant Growth and Ginsenoside Biosynthesis in Forest Cultivated Panax ginseng (F. Ginseng). Front. Plant Sci. 7:1. doi: 10.3389/fpls.2016.00001 F. Ginseng (Panax ginseng) is planted in the forest to enhance the natural ginseng resources, which have an immense medicinal and economic value. The morphology of the cultivated plants becomes similar to that of wild growing ginseng (W. Ginseng) over the years. So far, there have been no studies highlighting the physiological or functional changes in F. Ginseng and its wild counterparts. In the present study, we used proteomic technologies (2DE and iTRAQ) coupled to mass spectrometry to compare W. Ginseng and F. Ginseng at various growth stages. Hierarchical cluster analysis based on protein abundance revealed that the protein expression profile of 25-year-old F. Ginseng was more like W. Ginseng than less 20-year-old F. Ginseng. We identified 192 differentially expressed protein spots in F. Ginseng. These protein spots increased with increase in growth years of F. Ginseng and were associated with proteins involved in energy metabolism, ginsenosides biosynthesis, and stress response. The mRNA, physiological, and metabolic analysis showed that the external morphology, protein expression profile, and ginsenoside synthesis ability of the F. Ginseng increased just like that of W. Ginseng with the increase in age. Our study represents the first characterization of the proteome of F. Ginseng during development and provides new insights into the metabolism and accumulation of ginsenosides.

Keywords: Panax ginseng, energy metabolism, ginsenosides biosynthesis, growth, proteomic analysis

### INTRODUCTION

Ginseng (Panax ginseng C. A. Meyer) is a perennial plant that has long been used in Chinese herbal medicine. The main part of ginseng that has therapeutic effects is the root. Ginseng root has clinical and pharmacological effects such as anti-aging activity, anticancer activity, protection against circulatory shock effects; it also promotes immune function in human beings (Wang et al., 2005; Li et al., 2012; Kim and Cho, 2013; Bae et al., 2014). However, wild-type ginseng (W. Ginseng) is very scarce and unsustainable. Ginseng growing in forests (F. Ginseng) is a type of ginseng cultivar, which is planted to alleviate the resource scarcity and ensure the external morphology.

F. Ginseng requires 15–20 years or even longer to attain the medicinal properties with efficacy similar to that of W. Ginseng (Li, 2008). F. Ginseng undergoes dramatic changes in size, weight, composition, and accumulating ginsenosides in the root over time (Deng et al., 2013). Folklore suggests that a change in the morphology of F. Ginseng is a good indicator of its growth period, and helps to assess its effectiveness and differences as compared with W. Ginseng. Moreover, recent studies the content of ginsenosides-the main active substancesincreases every improve its medicinal value (Soldati and Tanaka, 1984; Chan et al., 2000; Lee et al., 2001; Lin et al., 2010). Thus, the medicinal value of F. Ginseng is positively correlated to its age. The longer the growth time, the more similar F. Ginseng is to W. Ginseng. Therefore, comparative studies on the growth and development of F. Ginseng with W. Ginseng will provide deeper insights into the molecular mechanism of ginsenoside formation.

In fact, the plant growth is the summation of biochemical and physiological changes. These changes include synthesis of sugars, alterations in secondary metabolite biosynthesis, response to stress, and accumulation of antioxidant compounds (Giovannoni, 2004). Differential proteomic approaches enable the identification of protein species with changes in abundance levels during the process of growth. This allows the identification of proteins that are specifically relevant to the control of the metabolic pathways responsible for plant growth and development (Andrade et al., 2012). Major protein variations that occur during plant development, ripening and response to stress are well-studied in many commercially important plants including the model plant Arabidopsis thaliana (Lee et al., 2012), rice (Nozu et al., 2006), tomato (Rocco et al., 2006; Faurobert et al., 2007), and orange (Bianco et al., 2009). Previously, studies on ginseng focused on differences between cultivars (Lum et al., 2002; Nagappan et al., 2012). The protein changes that occur in ginsengs during growth and ginsenosides biosynthesis are not well-studied.

Traditionally, two-dimensional polyacrylamide gel electrophoresis (2DE) has been the gold standard for proteomic analysis. However, this platform is limited by protein identification and quantification capabilities. For instance, the low-abundance proteins, such as membrane proteins and hydrophobic proteins are difficult to detect on 2D gel-electrophoresis (Zieske, 2006). To overcome the disadvantage of this technique, non-gel-based quantitative proteomic methods have been developed in recent years. Isobaric mass tagging (e.g., iTRAQ) is a precise and sensitive multiplexed peptide/protein quantification technique in mass spectrometry (Ghosh et al., 2013), which has been extensively used for revealing the differentially expressed proteins under any given conditions including plant growth and development (Fukao et al., 2011), and biotic and abiotic stresses (Wang et al., 2014; Li et al., 2015).

In the present study, we used a proteomic approach involving 2DE and iTRAQ to investigate the differentially expressed proteins during F. Ginseng root growth, analyze the changes on metabolic process related to growth, and investigate the relationship between F. Ginseng and W. Ginseng. Our findings will help in understanding the molecular mechanism of ginsenoside biosynthesis. Moreover, our results can also be extrapolated to studying the medicinal use of F. Ginseng.

### MATERIALS AND METHODS

#### Plant Materials

Up to 80% of the world's total supply of F. Ginseng roots comes from the major ginseng farming region in the Jilin Province, China. Therefore, F. Ginseng roots were collected from August in 2014 from the FuSong County in Jilin Province. This area is located at 127 degrees, 46 min east longitude and 42◦ , 48 min north latitude and is situated in a humid mountainous climate at an altitude of 520 m. F. Ginseng samples were collected during four kinds of growth years: 10, 15, 20, and 25 years. Fifteen samples/years were taken and frozen in liquid nitrogen. W. Ginseng roots were provided by Wujie wild ginseng planting base located in the FuSong county. Ten W. Ginseng roots samples were 40 years old, collected at 127◦ 30–50 min east longitude and 35–58 min north latitude. The dissected samples were immediately frozen in liquid nitrogen.

### Estimation of Growth parameters and phenotypic Plasticity Indexes

As described previously (Lum et al., 2002), we divided ginseng into its main root (MR), lateral root (LR) and rhizome head (RH) (**Figure 1**). Different parts were weighed on an electronic scale (0.01 g), and the root mass ratio (RMR) was calculated (g·g −1 ) (Gregory et al., 1995). We placed the plant parts on a glass board covered with graph paper to measure their length (0.1 mm) and calculated the specific root length (SRL) (cm·g −1 ) (Ostonen et al., 2007). The total length and biomass were determined as the sum of every part. Average values were calculated from 25 samples per developmental stage of F. Ginseng. Relative growth rate (RGR) was measured as the increase in mass per biomass per year and was calculated using the following equation: RGR = (ln W2 − ln W1) / (t2 − t1); where ln = natural logarithm, t1 = time one (in years), t2 = time two (in years), W1 = weight of plant at time one (in grams), W2 = weight of plant at time two (in grams). The phenotypic plasticity index [PPI, (F. Ginseng mean − W. Ginseng mean)/W.

Ginseng mean] was calculated for each trait (Caplan and Yeakley, 2013), which was used to evaluate the morphological difference between F. Ginsengs (in different growth time) and W. Ginseng.

#### Protein Extraction

The ginseng root proteins were extracted using a phenol procedure (Wang et al., 2009). Ground tissue was precipitated with cold acetone with 0.07% b-mercaptoethanol (at least three times) and resuspended in lysis buffer [7 M urea, 2 M thiourea, 2% (w/v) CHPAS, 1% (w/v) plant protease inhibitor]. Then, an equal volume of tris-saturated phenol was added and centrifuged at 10600 g/min at 4◦C for 15 min, and the water phase was discarded. The phenol phase was washed with methanol containing 0.1 M ammonium acetate and acetone two and three times, respectively. After the complete evaporation of acetate, the proteins were dissolved in the appropriate volume of rehydration solution [5 M urea, 2 M thiourea, 2% (w/v) CHAPS, 2% (w/v) N-decyl-N,Ndimethyl-3-ammonio-1-propane-sulfonate (SB3-10)] (Chinnasamy and Rampitsch, 2006). The protein concentrations were measured using Bradford's method.

### 2DE and Image Analysis

The protein samples were first separated by isoelectric focusing using linear precast IPG strips (24 cm, 3–10 linear pH gradients, GE Healthcare, UK). IPG strips with 1.2 mg of protein were rehydrated for 12 h. and focused at 72,000 Vhs, as described previously (Sun et al., 2011). Focusing was performed under the following conditions: a rapid gradient increase from 30 to 200 V for 1 h, 200 to 500 V for 1 h and then a linear increase from 500 to 1000 V for 2 h, 8000 V for 4 h, and, at last, a rapid gradient of 8000 V until 60,000 V.h. After IEF, first-dimension strips were equilibrated immediately. Second-dimension SDS– PAGE was performed using 12.5% polyacrylamide gels at 2 W per gel for 30 min and 15 W per gel for 5–6 h in six EttanDalt systems (GE Healthcare, UK). Finally, the gels were stained in the staining solution (0.05% CBBR in 25% isopropanol and 10% acetic acid) for 6 h or overnight and destained by 10% (v/v) ethanol and 10% (v/v) acetic acid until the background was clear (Borejdo and Flynn, 1984; Westermeier, 2006; Yang et al., 2016).The stained gels were scanned by a 600 dpi Image Scanner (GE Healthcare, UK). All spots were matched by gel-to-gel comparison using Image Master 2D Platinum Software Version 6.0 (GE Healthcare, UK). The spots with statistically significant (Student's t-test with a P < 0.05) and reproducible changes (quantitative changes > 1.5-fold in abundance) in abundance were considered to be differentially expressed protein spots.

Significant differences were analyzed using a two-way hierarchical clustering methodology using the PermutMatrix software (Meunier et al., 2007). For this purpose, the data produced by the analysis of 2DE gels were converted into a binary matrix, and the missing values were replaced with zeros (Negri et al., 2008). The row-by-row normalization of data was performed using the classical zero-mean and unit-standard deviation technique. Pearson's distance and Ward's algorithm were used for the analysis.

### Two-Dimensional Gel Excision, Tryptic Digestion, and Desalting

Protein extracts were separated on preparative gels and proteins of interest were recovered from the gels for identification. Proteins from the different years of F. Ginseng samples were resolved on separate preparative polyacrylamide gels and were visualized by staining with a Coomassie blue staining method compatible with subsequent mass-spectrometric analysis. All of the differentially expressed spots were selected and excised manually from the four preparative gels. Protein spots of interest were cut from the preparative gels de-stained for 20 min in 25 mM NH4HCO3/50% acetonitrile and washed with Milli-Q water until the gels were de-stained. The spots were incubated in 0.2 M NH4HCO<sup>3</sup> for 20 min and then lyophilized. Each spot was digested overnight in 12.5 ng/µl trypsin in 25 mM NH4HCO3. The peptides were extracted three times with 60% acetonitrile (ACN)/0.1% trifluoroacetic acid (TFA). The extracts were pooled and dried completely using a vacuum centrifuge.

### Protein Identification by MALDI-TOF/TOF MS

MS and MS/MS data for protein identification were obtained by using a MALDI-TOF-TOF instrument (4800 proteomics analyzer; Applied Biosystems). Instrument parameters were set using the 4000 Series Explorer software (Applied Biosystems). The MS spectra were recorded in reflector mode in a mass range from 800 to 4000 with a focus mass of 2000. MS was used a CalMix5 standard to calibrate the instrument (ABI 4700 Calibration Mixture). For one main MS spectrum, 25 subspectra with 125 shots per sub-spectrum were accumulated using a random search pattern. For MS calibration, autolysis peaks of trypsin ([M+H]+842.5100 and 2211.1046) were used as internal calibrates, and up to 10 of the most intense ion signals were selected as precursors for MS/MS acquisition, excluding the trypsin autolysis peaks and the matrix ion signals. In MS/MS positive ion mode, for one main MS spectrum, 50 sub-spectra with 50 shots per sub-spectrum were accumulated using a random search pattern. The collision energy was 2 kV, collision gas was air, and default calibration was set by using the Glu1-Fibrino-peptide B ([M+H]<sup>+</sup> 1570.6696) spotted onto Cal 7 positions of the MALDI target. Combined peptide mass fingerprinting PMF and MS/MS queries were performed by using the MASCOT search engine 2.2 (Matrix Science, Ltd.), embedded into GPS-Explorer Software 3.6 (Applied Biosystems) on the NCBI viridiplantae database with the following parameter settings: 100 ppm mass accuracy, trypsin cleavage one missed cleavage allowed, carbamidomethylation set as fixed modification, oxidation of methionine was allowed as variable modification, MS/MS fragment tolerance was set to 0.4 Da. A GPS Explorer protein confidence index ≥95% were used for further manual validation.

#### iTRAQ Analysis

Protein (100 mg) was reduced by adding dithiothreitol to a final concentration of 10 mM and incubated for 1 h at room temperature. Subsequently, iodoacetamide was added to a final

concentration of 40 mM, and the mixture was incubated for 1 h at room temperature in the dark. Then, dithiothreitol (10 mM) was added to the mixture to consume any free iodoacetamide by incubating the mixture for 1 h at room temperature in the dark. Proteins were then diluted in 50 mM triethylammonium bicarbonate and 1 mM CaCl<sup>2</sup> to reduce the urea concentration to less than 0.6 M and digested with 40 mg of modified trypsin at 37◦C overnight. The resulting peptide solution was acidified with 10% trifluoroacetic acid and desalted on a C18 solid-phase extraction cartridge.

Desalted peptides were then labeled with iTRAQ reagents (Applied Biosystems) according to the manufacturer's instructions. Ten-year-old ginsengs were labeled with reagent 114; 15-year-old ginsengs were labeled with reagent 115; 20 year-old ginsengs were labeled with reagent 116, and 25-year-old ginsengs were labeled with reagent 117. The reaction was allowed to proceed for 1 h at room temperature.

Subsequently, Nano-HPLC-MS/MS analysis was performed on a nanoAcquity system (Waters) connected to an LTQ-Orbitrap XL hybrid mass spectrometer (Thermo Electron) equipped with a PicoView nanospray interface (New Objective). Peptide mixtures were loaded onto a 75-mm i.d., 25-cm length C18 BEH column (Waters) packed with 1.7-mm particles with a pore size of 130 A◦ . They were separated using a segmented gradient in 90 min from 5 to 40% solvent B (acetonitrile with 0.1% formic acid) at a flow rate of 300 nL /min and a column temperature of 35◦C. Solvent A was 0.1% formic acid in water. The LTQ-Orbitrap XL hybrid mass spectrometer was operated in positive ionization mode. The MS survey scan for all experiments was performed in the Fourier transform cell recording a window between 350 and 1600 mass to charge ratio (m/z). The resolution was set to 60,000 at m/z 400, and the automatic gain control was set to 500,000 ions. The m/z values triggering MS/MS were put on an exclusion list for 90 s. The minimum MS signal for triggering MS/MS was set to 5000. In all cases, one microscan was recorded. For high-energy collision dissociation, the applied acquisition method consisted of a survey scan to detect the peptide ions followed by a maximum of three MS/MS experiments of the three most intense signals exceeding a minimum signal of 5000 in survey scans. For MS/MS, we used a resolution of 7500, an isolation window of 2 m/z, and a target value of 100,000 ions, with maximum accumulation times of 400 ms. Fragmentation was performed with a normalized collision energy of 50% and an activation time of 30 ms. We performed three technical replications for each experiment.

#### Database Search and Quantification

The 2.3.02 version of the Mascot software (Matrix Science) was used to identify and quantify proteins simultaneously. In this version, unique peptides used for protein quantification can be chosen, which is more precise to quantify proteins. Searches were made against the green plants protein database (TAIR9\_pep\_20090619, 33,410 sequences; ftp://ftp.arabidopsis. org/home/tair/Sequences/blast\_datasets/TAIR9\_blastsets/) concatenated with a decoy database containing the randomized sequences of the original database. For each technical repeat, spectra from the 20 fractions were combined into one MGF (Mascot generic format) file after loading the raw data, and the MGF file were searched. For biological repeats, spectra from the three technical repeats were combined into one file and searched. The search parameters were as follows: trypsin/P was chosen as the enzyme with two missed cleavages allowed; fixed modifications of carbamidomethylation at Cys, variable modifications of oxidation at Met and iTRAQ 4plex at Tyr; peptide tolerance was set at 10 ppm, and MS/MS tolerance was set at 0.6 D. Peptide charge was set Mr, and monoisotopic mass was chosen. iTRAQ 4plex was chosen for quantification during the search simultaneously. The search results were passed through additional filters before exporting the data. For protein identification, the filters were set as follows: Significance threshold P, 0.05 (with 95% confidence) and ion score or expected cutoff less than 0.05 (with 95% confidence). For protein quantitation, the filters were set as follows: "Weighted" was chosen for protein ratio type (http:// mascot-pc/mascot/help/quant\_config\_help.html); minimum precursor charge was set to 1, and minimum peptides were set to 2; unique peptides were used to quantify proteins. Summed intensities were set as normalization, and outliers were removed automatically. The peptide threshold was set as above for homology.

#### Enzymatic Activity Analysis

Amylase (AMY, EC 3.1.1.2) activity was detected by the 3,5-dinitrosalicylic acid colorimetric method (Hao et al., 2007). Malate dehydrogenase (MDH, EC 1.1.1.37) activity was examined as described by Husted and Schjoerring (1995), with some modifications. Ten microliter samples were added to a 3 ml reaction mixture containing 0.17 mM oxalacetic acid and 0.094 mM β-NADH disodium salt in 0.1 M Tris buffer, pH 7.5. The reaction was measured by the decrease in absorbance at 340 nm for 180s in a spectrophotometer (Hitachi U-2001 Japan), the same reaction system only with sample buffer added in was used as a blank. Superoxide dismutase (SOD, EC 1.14.1.1) activity was measured according to the method of Zhang and Kirkham (1996), and Xu and Huang (2004). One unit of SOD activity is defined as the amount of SOD required to cause 50% inhibition of nitroblue tetrazolium (NBT) reduction at 560 nm min-1. Catalase (CAT, EC 1.11.1.6) and peroxidase (POD, EC.1.11.1.7) activity were determined based on the method of Chance and Maehly (1955) as described in detail for creeping bentgrass in Xu and Huang (2004). Enzyme activities were based on the absorbance change of the reaction solution per minute at a given wavelength for each enzyme: CAT at 240 nm and POD at 470 nm.

The activities of farnesyl diphosphate synthase (FDPS, EC. 2.5.1.10), cycloartenol synthase (CAS, EC. 5.4.99.8), squalene epoxidases (SE, EC:1.14.13.132), and squalene synthase (SS, EC. 2.5.1.21) involved in ginsenosides biosynthesis, were quantified by an indirect competitive enzyme-linked immunosorbent assay (ELISA). The optical density (OD) values of each sample were read by a BioTek ELx800 microplate reader at 450 nm. The primary concentration of each test sample was calculated from the linear regression equation based on the OD values of the standards.

#### Metabolite Content Analyses

Starch was measured via an enzyme hydrolysis method. Starch was hydrolyzed into dual sugars by amylase, hydrolyzed into monosaccharides by hydrochloric acid, and finally determined by reducing sugar, which is converted to starch (Rose et al., 1991).

The contents of pyruvate in the sample were determined according to the methods of Lin et al. (1995). Protein was removed from the samples by TCA precipitation, and in the resulting sample, pyruvate reacted with 2,4 nitrophenylhydrazine. The product turned red in the presence of an alkali solution, and the intensity of the color change was measured by a spectrophotometer. A standard curve for calibration was obtained using sodium pyruvate as a reagent with a gradient of concentrations of pyruvic acid. Absorbance values were obtained to generate a standard curve to calculate the pyruvate concentration.

For glutathione (GSH), roots were ground in liquid nitrogen and homogenized in 1 mL 5% (w/v) m-phosphoric acid containing 1 mM diethylene triamine pentaacetic acid (DTPA) and 6.7% (w/v) sulfosalicylic acid. Root extracts were centrifuged at 12,000 × g for 15 min at 4◦C. GSH contents were determined according to the methods of Kortt and Liu (1973) and Ellman (1959) with some modifications.

The ascorbic acid (AsA) content was determined according to Egea et al. (2007) with slight modifications. Ginseng roots were ground in an ice bath with 10 mL 5% metaphosphoric acid stored at 4◦C, and then the final mix was homogenized by vortex. The final solution was maintained on the ice bath, in darkness, for 30 min and then centrifuged at 20,000 × g for 25 min at 4◦C. Ascorbate was spectrophotometrically detected by measuring absorbance at 254 nm with a UV detector. For quantification of the compound, a calibration curve in the range of 10–100 mg kg−<sup>1</sup> was prepared from standard ascorbic acid. Results were expressed as mg 100 g−<sup>1</sup> FW.

Root extracts were centrifuged at 12,000 × g for 15 min at 4◦C. The extraction and determination of ginsenosides was performed following the method of Yu et al. (2002).

#### Semi-Quantitative RT-PCR Analysis

Total RNA was extracted from various P. ginseng samples collected over various growth years using RNeasy mini kit (Takara Bio, China). RT-PCR was conducted using 200 ng of total RNA as a template for reverse transcription using oligo(dT)15 primer (0.2 mM) and AMV Reverse Transcriptase (10 U/µl) (Takara Bio, China) according to the manufacturer's instructions. RT-PCR was performed using a 1-µl aliquot of the first strand cDNA in a final volume of 20µl reaction volume. Five pmol of specific primers for pathogenesis-related protein gene (PR5) and glutaredoxin gene (Grx) were used for performing PCR. The actin gene (ACT) primers were used as internal control (**Table S1**). The thermal cycler conditions recommended by the manufacturer were used as follow: an initial denaturation for 10 min at 94◦C, 35 amplification cycles (30 s at 94◦C, 30 s at 58◦C, and 30 s at 72◦C), followed by a final elongation for 10 min at 72◦C. Ten microliters of the reaction mixture were analyzed on a 1% (W/V) agarose gel in 1 × TAE buffer and then photographed for expression analysis.

Images of the RT-PCR ethidium bromide-stained agarose gels were acquired with a Cohu High Performance CCD camera (Cohu Inc. San Diego, CA). The quantification of bands was performed by Phoretix 1 D (Phoretix International Ltd., Newcastle upon Tyne, UK). Band intensity was expressed as relative absorbance units. The ratio between the sample RNA to be determined and actin was calculated to normalize for initial variations in sample concentration and as a control for reaction efficiency. Means and standard deviations of all experiments performed were calculated after normalization to actin.

#### Statistical Analysis

Values in figures and tables are expressed as the mean ± SD. Statistical analysis was carried out with three biological replicates for proteomic and biochemical analyses. The results of the spot intensities and physiological data were statistically analyzed by a One-way ANOVA and the Duncan's new multiple range test (DMRT) to determine the significance of differences between group means. P ≤ 0.05 was considered statistically significant (SPSS for Windows, version 12.0).

#### RESULTS

#### Morphological Changes in F. Ginseng growth

In order to analyze the changes in physiological shape along with the growth years, we tested the growth parameters of F. Ginseng root (**Table 1**). The morphological characteristics of F. Ginseng showed that the total length (TL) increased with years of growth while specific root length (SRL), specific main root length (SMRL), specific lateral root length (SLRL) and specific rhizome head length (SHRL) demonstrated an opposite trend. The SRL of young F. Ginseng (**Figures 2A,B**) was greater than that of older F. Ginseng (**Figures 2C,D**), which decreased significantly due to yearly increases in weight and diameter to guarantee root absorption during the period of vigorous growth (Maurice et al., 2010), similar to W. Ginseng (**Figure 2E**).

Biomass is based on energy and the accumulation of nutrients (Guo et al., 2002). The increase in total biomass (TB) of F. Ginseng was mainly allocated to the main root and rhizome head, and the lateral root mass ratio (LRMR) decreased year by year (**Table 1**). This was mainly due to the degeneration of most of the tiny lateral roots and the survival of several strong roots (**Figure 2**), which enabled the plant to adapt to the environment by reducing the risk of root damage.

The RGR can capture dynamic changes in the physiology and morphology of F. Ginseng and is used to quantify the speed of plant growth (Useche and Shipley, 2010). The results showed that the RGR decreases over time as the biomass of F. Ginseng increases and then becomes stable in mature ginseng, consistent with the other growth parameters (**Table 1**).

The year-to-year variation of SLRL and SHRL in F. Ginseng growing is closer to W. Ginseng. The phenotypic plasticity index (PPI) between 25-year-old F. Ginseng and W. Ginseng was minimum, indicating that F. Ginseng mainly adapts to changes in growth by regulating morphology and allocating biomass TABLE 1 | Growth parameters and phenotypic plasticity indexes of F. Ginsengs during different growth years and W. Ginseng.


(**Table 1**). Our results are in agreement with those of Huang et al. (2012).

#### Differences between Relative Abundance of F. Ginseng and W. Ginseng

Proteins, the expression of which could explain the life characteristics of organisms with a specified status, ultimately control biological processes. There is a negative correlation between the abundance of protein and the rate of evolution (Wolf et al., 2010). The proteins with high and intermediate abundance with central importance in cells may reflect the genetic relationships between different species to some extent (Beck et al., 2011; Zhong et al., 2012). We compared the proteomic profiles of F. Ginseng and W. Ginseng to explore similar characteristics between the two ginseng types (**Figure S1**). The experimental workflow is shown in **Figure S2**.

We analyzed the proteins present at different levels between F. Ginseng and W. Ginseng by hierarchical clustering analysis using PermutMatrix software. We compared the protein levels in the five samples including 10-, 15-, 20-, and 25-year-old of F. Ginseng and W. Ginseng (**Figure 3**). The clustering of the differentially accumulated proteins revealed two major clusters as seen from the dates of the columns. The 10- and 15-yearold F. Ginsengs clustered into one group preferentially, with the next level clustering with the 20-year-old ginseng. These findings indicate that the 10-, 15-, and 20-year-old ginseng plants are more closely related on the protein level than the other plants tested here. Notably, 25-year-old ginseng and wild ginseng clustered into one group. This finding shows that the morphological and physiological parameters of F. Ginseng and W. Ginseng are reshaped over the years of growth, and these changes are regulated by protein abundance. On the protein level, this result validates the traditional view that the morphology of older F. Ginseng becomes closer to W. Ginseng.

#### 2DE analyses of Differential Expression Protein

The protein expression patterns of F. Ginseng in different growth years were analyzed by 2DE. We identified 47 proteins as continuously changing based on their accumulation patterns (**Figure S1D**). Among these proteins, 31 were up-regulated, and 16 were down-regulated. We divided the mass spectral data

into six groups with clustering analysis (**Figure 4**, **Table S2**). We further grouped the up-regulated proteins into three clusters. Proteins expressed at continuously increasing levels with years of growth of F. Ginseng were grouped in cluster A. Cluster B proteins were reduced in abundance during the 15th and 20th year of growth. Proteins that first increased in abundance and then decreased were placed in cluster C. A total of 16 proteins decreased in abundance during the 25th year of F. Ginseng growth compared with other growth periods. We defined these as growth-related proteins because they either re-accumulated in 20-year-old F. Ginseng (after first decreasing in abundance in 15-year-old plants, cluster D) or increased in abundance in 15-year-old plants (clusters E and F).

#### iTRAQ Analyses of Differentially Expressed Protein

We used 2DE to analyze the proteins with high and intermediate abundance. However, the low-abundance proteins were difficult to detect; therefore, we used iTRAQ for the quantitative analysis of the differentially expressed proteins of F. Ginseng in different growth years to corroborate and supplement

the 2DE analysis data. We used three biological replicates for each condition to identify 145 differentially expressed proteins. This number excluded the unknown and predicted proteins. Our results revealed a spectrum of different temporal expression patterns (**Figure 5**, **Table S3**), ranging from proteins up-regulated primarily during the latter stages of the time course (clusters A–D) to proteins down-regulated primarily during the latter phases of the time course (clusters E–H). Proteins in the growth process category were statistically over-represented in clusters A–D and under-represented in clusters E–H. The proteins increased in abundance in 15 years old in clusters B and G, and in 20 years old in clusters A, F, and H in the growth process category. The abundance of proteins in clusters D and E exhibited no significant changes and returned to control levels in 25-year-old plants. Clusters A, B, and G also included most of the highly induced proteins (>2-fold increase compared with the 15-year-old plants). These changes in protein abundance demonstrate that the growth and development of F. Ginseng could be controlled by changes in the expression of these differentially expressed proteins (Ma et al., 2013).

#### Functional Analysis of Identified Proteins

We grouped the proteins identified by 2DE and iTRAQ according to their biological functions (**Figure 6**), which were determined using the iProClass Gene Ontology (GO) analysis tool in the Protein Information Resource (PIR) database (http://pir.georgetown.edu/). The GO classification will help to identify the metabolic events associated with the growth of F. Ginseng.

The main biological functions were: energy metabolism (29%), disease/defense (21%), transcription related(14%), cell growth/division (9%), protein synthesis and degradation (9%), and secondary metabolism (8%). In addition, some proteins were categorized in unknown biological processes.

#### Analysis of Differentially Expressed Enzymes Activity

The level of activity is positively correlated with the enzyme protein abundance (Yang et al., 2013). To validate the proteomics data, two enzymes involved in glycometabolism, four enzymes involved in ginsenosides biosynthesis, and three enzymes involved in ROS scavenging between F. Ginseng and W. Ginseng were selected for activity analysis (**Figure 7**). The activities of AMY (**Figure 7A**) were lower, whereas the activities of MDH (**Figure 7B**) were higher in older F. Ginseng. The activities of FDPS (**Figure 7C**), CAS (**Figure 7D**), SE (**Figure 7E**), and SS (**Figure 7F**) increased with growth years of F. Ginseng. The activities of SOD (**Figure 7G**) and CAT (**Figure 7H**) changed

dynamically in a parabolic pattern, and POD (**Figure 7I**) showed a downward trend over the course of F. Ginseng growth. These results concur with the protein profiles of the 2DE and iTRAQ analysis. Moreover, the changes in enzyme activity of F. Ginseng (with growth time) should be close to that of W. Ginseng.

FIGURE 6 | Functional analyses of differentially expressed proteins in the biological process. Percentage distributions of the GO terms were calculated by iProClass GO tool in PIR database.

### Changes in Metabolite(s) Content

Metabolite synthesis is influenced by the activities of the enzymes that regulate the metabolic pathways. In this study, we found that the starch (**Figure 7J**) content of F. Ginseng increased with increase in growth years, which showed a negative correlation with the activity of amylase. Pyruvate (**Figure 7K**) is an intermediate metabolite of glucose metabolism and its content increased with growth years indicating a gradually active glucose metabolism. The content of ginsenoside (**Figure 7L**), as well as the activities of the enzymes in ginsenoside synthesis also increased with growth years. Moreover, antioxidants such as AsA (**Figure 7M**) and GSH (**Figure 7N**) accumulated to enhance the antioxidant capacity of F. Ginseng. The above results indicate that F. Ginseng becomes increasingly similar to wild ginseng as growth years increase.

#### The Expression Analysis of Grx and PR5 at the mRNA Level

In order to assess the correlation of expression levels between mRNA and protein, we did a semi-quantitative PCR for the defense genes, Grx and PR5 (**Figure 8**). The expression of Grx gene was down-regulated in response to stress year after year. Conversely, the expression of PR5 was up-regulated year after year. Therefore, the expression of these proteins, in response to stress, is regulated at the transcriptional level in growing F. Ginseng.

#### DISCUSSION

#### Active Energy Metabolism for Growth and Development

The growth and development of F. Ginseng require active energy metabolism to produce the necessary material and energy. Our data shows that the largest category of proteins in F. Ginseng is associated with energy metabolism. This category primarily includes proteins involved in starch metabolism, glycolysis, the pentose phosphate pathway, and the tricarboxylic acid cycle.

Fructokinases—that induce starch synthesis—were overexpressed over the time course of F. Ginseng growth (spot 15, 16) (**Figure 4** clusters C). Amylase (spot 58) (**Figure 4** clusters B), which reduce starch degradation, were down-regulated over time. Starch content (**Figure 7J**) in F. Ginseng was positively correlated with the fructokinase expression and negatively correlated with the amylase activity (**Figure 7A**); thus, corroborating the study of Yang et al. (2001). F. Ginseng consumes many nutrients in the early growth period to provide enough material and energy for morphogenesis and development. Starch accumulates as F. Ginseng matures, and the same trend is observed in tomato, maize endosperm, and Arabidopsis seeds (Schaffer and Petreikov, 1997; Angeles-Núñez and Tiessen, 2011; Spielbauer et al., 2013). The starch accumulation of W. Ginseng less than F. Ginseng probably could due to a lack of nutritional growth environment.

Glycolysis is active as the downstream pathway of starch metabolism. Enolase (spot 25, 27, 31) (**Figure 4** clusters A and C), Glucose-6-phosphate dehydrogenase (6PGDH, spot 30) (**Figure 4** clusters A), and glyceric acid phosphate mutase, which is involved in glycolysis, was up-regulated in mature F. Ginseng, providing energy for the growth and development of F. Ginseng. Pyruvic acid, as the end product of glycolysis, increased with an increase in the growth of F. Ginseng (**Figure 7K**). Thereafter, the intermediate products of active glycolysis promote other energy metabolic pathways (Gómez et al., 2012). MDH (spot 42) (**Figure 4** clusters C. **Figure 7B**), aconitic acid hydratase and isocitrate dehydrogenase (spots 40 and 41) (**Figure 4** clusters A) related to Krebs cycle and pentose phosphate pathway accumulated from year to year. Energy metabolism of the older F. Ginseng was active to provide a large number of ATPs for the growth and metabolism, and this trend approached that of W. Ginseng year after year.

In addition to the energy metabolism pathways discussed above, we also observed an increase in the expression of related proteins in F. Ginseng. This suggests that the energy metabolism increases as F. Ginseng matures, providing energy to support its morphological changes. This result was consistent with garden ginseng growth, wherein active energy metabolism promotes root enlargement and weight gain (Ma et al., 2013). Moreover, large amounts of precursor substances are synthesized to promote effective substance biosynthesis to improve the quality of F. Ginseng medicinal components.

#### Ginsenosides Biosynthesis Promotes the Medicinal Properties of F. Ginseng

Ginsenoside—active ingredients of ginseng—are a class of natural product steroid glycosides and triterpene saponins. Ginsenosides have a long history of use in traditional medicine such as anti-tumor and neurotrophic activity, and enhancement of immunity (Rudakewich et al., 2001; Zhang et al., 2008; Song et al., 2010). Studies have shown that the ginsenoside content continues to increase along with the age of ginseng (Soldati and Tanaka, 1984). An increase in the content of ginsenosides is directly related to an increase in their pharmacological effect (Lin et al., 2010). The synthesis and effects of ginsenosides present positive correlation in ginseng growth.

We identified enzymes involved in the biosynthesis of ginsenosides in our study. These enzymes (proteins) were differentially expressed in F. Ginseng in different growth years. FDPS (**Figure 5** clusters H, **Figure 7C**), CAS (**Figure 5** clusters F, **Figure 7D**), SE (**Figure 5** clusters E, **Figure 7E**), and SS (**Figure 5** clusters F, **Figure 7F**) were up-regulated to increase the synthesis ability of the F. Ginseng ginsenoside. Mature F. Ginseng plants had higher ginsenoside content (**Figure 7L**) as compared with their younger counterparts. Our results suggest F. Ginseng that ginsenoside synthesis in F. Ginseng increases every year to enhance its medicinal properties and become more like W. Ginseng.

### Scavenging ROS to Maintain Redox Balance

Plants are vulnerable to growth environmental stresses, which result in excessive accumulation of ROS, causing cell damage. ROS scavenging enzymes such as peroxiredoxin (spot 18) (**Figure 4** clusters D), SOD (**Figure 7G**), CAT (**Figure 7H**), and Grx gene (**Figures 8A,B**; Ding et al., 2010) were up-regulated in the early growth stage and down-regulated in the mature stage. Thus, F. Ginseng activated the antioxidant defense system to cope with the oxidative stress by maintaining a relatively stable level of ROS (Arbona and Gómez-Cadenas, 2012).

A key enzyme 6PGDH of the pentose phosphate pathway, is expressed at high levels in response to drought, low temperature and high salt stress (Airaki et al., 2012; Signorelli et al., 2013; Wang et al., 2013). Our results show up-regulated 6PGDH possibly to activate the pentose phosphate pathway and produce large amounts of reducing substances (NADPH), which remove active oxygen via the AsA-GSH cycle. We tested the AsA (**Figure 7M**) and GSH levels (**Figure 7N**), which increased with F. Ginseng growth. Our results showed that older F. Ginseng exhibitsW. Ginseng enhanced stress tolerance (similar to W. Ginseng) to maintain redox homeostasis within the cells.

### Enhanced Expression of Resistance Genes and Proteins to Ensure Health

Plant disease resistance is accomplished by temporal and spatial changes in gene expression that cause changes in the physiological and biochemical reactions to aid in resistance to pathogens. The NBS-LRR region is an important component of the disease resistance genes in the plant immune system (Nimchuk et al., 2003; Belkhadir et al., 2004), exists in DNA fragments of certain conservative domains in plant genomes, such as mosaic virus resistance genes (N gene) in Soybean and potato (Bakker et al., 2011; Zhang et al., 2012). We identified NBS-LRR (**Figure 5** clusters A) and CC-NBS-LRR (**Figure 5** clusters B) homologs in F. Ginseng, which were up-regulated over the growth period, and could improve the disease resistance and maintain the health of F. Ginseng. Meanwhile, the NBS (nucleotide binding site) participates in hydrolyzing ATP and releasing signals (Ade et al., 2007) such as ATPase (**Figure 4** clusters E) and RHO protein GDP dissociation inhibitor (Rho GDI) (**Figure 5** clusters E), which are involved in signal transduction of stress response (Wong et al., 2004).

Furthermore, the expression of resistance genes can result in the production and accumulation of the pathogenesis-related protein (PR), which can be induced by components of the plant's self-defense mechanism. We identified PR10 (**Figure 5** clusters D and F) by iTRAQ and PR5 gene (**Figures 8A,C**) was assayed by semi-quantitative RT-PCR. They were up-regulated in response to stress year after year. We observed the expression of several of these proteins, including major latex-like protein (**Figure 4** clusters F and **Figure 5** clusters F), ribonuclease (**Figure 5** clusters B), and POD. The up-regulation of these pathogenesisrelated proteins could increase the adaptive ability of roots to adverse growth environments, which may be part of the defense mechanism of F. Ginseng (Sun et al., 2010).

#### Transcriptional Regulation for Growth

DNA methylation, histone modification, and chromatin reorganization are some of the mechanisms that regulate transcription (Turner, 2009). We observed upregulated expression of cytosine-specific methyltransferase (**Figure 5** clusters B) and histone-lysine N-methyltransferase (**Figure 5** clusters A), which participate in DNA methylation and histone methylation. Both of these enzymes have a synergistic effect (Fuks et al., 2003) to regulate the response of F. Ginseng to various environmental stresses (Sharma et al., 2009; Bilichak et al., 2012; Chervona and Costa, 2012). Moreover, transposons and retrotransposons have DNA methylation sites, which are methylated and silenced to protect the host genome (Cam et al., 2008). The upregulated retrotransposon (**Figure 5** clusters F and G) and En/Spm-like transposon protein (**Figure 4** clusters A and **Figure 5** clusters F) indicate their active involvement in inhibiting improper transcription, insertion mutations, and recombination in the F. Ginseng genome.

SWI/SNF (SWItch/Sucrose NonFermentable) is a chromatin remodeling complex that helps in DNA packaging. In our study, both chromatin remodeling complex protein (**Figure 5** clusters A; Reyes, 2014) and chromatin assembly factor (**Figure 5** clusters E; Margueron and Reinberg, 2010) involved in SWI/SNF were differentially expressed over the course of F. Ginseng growth. These proteins maintain the right chromatin structure to promote the appropriate combination of transcription factors and basic transcription elements that help in the expression of resistance genes. We also identified some transcription factors that are overexpressed in the growth process. Eukaryotic translation initiation factor 5A (spots 21, 36; **Figure 4** clusters C) involved in resistance to oxidative and osmotic stress (Xu et al., 2011). WRKYs (**Figure 5** clusters B, C, E and F) participate in the regulation of related gene expression in response to changes in the growth environment (Mondini et al., 2012). These changes on transcription regulation in different growth stages of F. Ginseng provide a mechanism to adapt to the rapidly changing growth environment as well as to W. Ginseng.

## CONCLUSION

To our knowledge, this work is the first large-scale proteomic investigation F. Ginsengof F. Ginseng growth. As F. Ginsengs grow, their morphological characteristics and metabolism become comparable to that of W. Ginseng. The changes in protein abundance revealed that 25-year-old F. Ginseng is the most similar to W. Ginseng. Proteins identified from 192 spots and discussed in relation to F. Ginseng highlighted the metabolic changes occurring during the growth, with particular regard to proteins associated with energy metabolism, ginsenosides biosynthesis, and antioxidation (**Figure 9**). Therefore, 25-yearold F. Ginseng plants have better external morphological and medicinal effect, which is similar to wild ginseng, than their younger counterparts. This F. Ginseng study improves our understanding of the medicinal properties of F. Ginseng and provides a future direction for engineering new varieties with higher medicinal value.

### AUTHOR CONTRIBUTIONS

LS, DZ: Substantial contributions to the conception or design of the work. RM, XC, BM, GC, MW: the acquisition, analysis and interpretation of data for the work. RM, LS, XC: Drafting the work or revising it critically for important intellectual content. RM, LS: Final approval of the version to be published. All authors agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

### ACKNOWLEDGMENTS

This research was financially supported by two foundations of national natural science (No. 81373932 and No. 81274038), two national key technology R&D programs (No. 2011BAI03B01 and No. 2012BAI29B05), a specialized research fund for the doctoral program of higher education (No. 20122227110005) and Jilin provincial scientific and technologic development project (No. 20150520138JH).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 00001

Figure S1 | Comparison of F. Ginseng proteome pattern in different growth years, which were 10 years old (A), 15 years old (B), 20 years old (C), and 25 years old (D) respectively, and W. Ginseng (E). The arrows indicate the spot numbers corresponding to the differentially accumulated spots among F. Ginsengs of different ages. Statistics on the indicated spots can complete identification data in Supporting Information Table S2.

Figure S2 | Experimental workflow of proteomics analysis.

#### Table S1 | Oligonucleotide sequences of primers used for semiquantitative RT-PCR.

Table S2 | Identification of differentially expressed proteins from F. Ginseng in different growth years by MALDI-TOF/TOF-MS/MS. <sup>a</sup>Clusters, The clusters of abundance of differentially expressed proteins in Figure 4; <sup>b</sup>Spot no, spot numbers correspond with 2-DE gel as shown in Figure S1D; <sup>c</sup>Accession number in NCBI database. <sup>d</sup>Pep. count, Number of matched peptides.

#### REFERENCES


Table S3 | Identification of different expressed proteins from F. Ginseng in different growth years by iTRAQ. Expression differences were defined as

protein abundance ratios greater than 1.4 or less than 0.6. <sup>a</sup>Clusters, The clusters of abundance of differentially expressed proteins in Figure 5; <sup>b</sup>Accession number in NCBI database.c15/10, the ratios of protein abundances between 15 and 10 years of F. Ginseng; <sup>d</sup>20/10, the ratios of protein abundances between 20 and 10 years of F. Ginseng; <sup>e</sup>25/10, the ratios of protein abundances between 25 and 10 years of F. Ginseng.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Ma, Sun, Chen, Mei, Chang, Wang and Zhao. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.