# BIFIDOBACTERIA AND THEIR ROLE IN THE HUMAN GUT MICROBIOTA, 2nd Edition

EDITED BY : Francesca Turroni, David Berry and Marco Ventura PUBLISHED IN : Frontiers in Microbiology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-354-8 DOI 10.3389/978-2-88963-354-8

#### 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

# BIFIDOBACTERIA AND THEIR ROLE IN THE HUMAN GUT MICROBIOTA, 2nd Edition

Topic Editors:

Francesca Turroni, University of Parma, Italy David Berry, University of Vienna, Austria Marco Ventura, University of Parma, Italy

Atomic Force Microscope image showing bifidobacterial cells with appendices resembling pilus-like structures.

Image by Francesca Turroni

The human intestine is home of an almost inconceivable large number of microorganisms. The human gut microbiota can therefore be pictured as an organ placed within a host organism. The human gut microbiome, which in total may contain >100 times the number of genes present in our genome, endows us with functional features that we did not have to evolve ourselves. It is recognized that intestinal microbiota plays an important role in human health and disease. In fact, gut bacteria other than metabolize dietary components, may play complex roles such as modulation of the immune system and in reduction of gut infections. Variations in the presence and/or abundance of certain components of the intestinal microbiota have repeatedly been observed in patients that suffer from atopic diseases, inflammatory bowel disease, Crohn disease, ulcerative colitis, infectious colitis, colon cancer and diabetes. In this context, bifidobacteria represent one of the most common bacterial members of the human gut microbiota. Bifidobacteria are anaerobic,

Gram-positive, irregular or branched rod-shaped bacteria that are commonly found in the gastro-intestinal tracts (GIT) of humans, especially during the first stages of life and most animal and insects. Bifidobacterial fluctuations seem directly associated with health effects and for these reasons they are being exploited as health-promoting or probiotic bacteria. However, despite the extensive commercial exploitation of bifidobacteria as probiotic bacteria, little is known about their impact or dependency on other members of the human gut microbiota or on their host. Genome analyses have highlighted the existence of gene repertoires encoding products that are responsible for the adaptation of bifidobacteria to the human intestine and intense research efforts at international level are ongoing to understand the molecular details of these interactions. Specifically, the molecular interactions that are presumed to exist between bifidobacteria and the human host, as well as interactions between different residents of intestinal microbiota are the main topic of bifidobacterial research communities.

Publisher's note: In this 2nd edition, the following article has been updated: Cell-Free Spent Media Obtained from Bifidobacterium bifidum and Bifidobacterium crudilactis Grown in Media Supplemented with 3'-Sialyllactose Modulate Virulence Gene Expression in Escherichia coli O157:H7 and Salmonella Typhimurium, by Bondue, P., Crèvecoeur, S., Brose, F., Daube, G., Seghaye, M.-C., Griffiths, M. W., et al. (2016). Front. Microbiol. 7:1460. doi: 10.3389/fmicb.2016.01460

Citation: Turroni, F., Berry, D., Ventura, M., eds. (2019). Bifidobacteria and Their Role in the Human Gut Microbiota, 2nd Edition. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-354-8

# Table of Contents

*06 Editorial:* Bifidobacteria *and Their Role in the Human Gut Microbiota* Francesca Turroni, David Berry and Marco Ventura

#### ECOLOGY


#### PHYSIOLOGY


Pauline Bondue, Sébastien Crèvecoeur, François Brose, Georges Daube, Marie-Christine Seghaye, Mansel W. Griffiths, Gisèle LaPointe and Véronique Delcenserie


#### GENOMICS

*107 Phylogenetic Analysis of the* Bifidobacterium *Genus Using Glycolysis Enzyme Sequences*

Katelyn Brandt and Rodolphe Barrangou

*118 Phylogenomic Analyses and Comparative Studies on Genomes of the*  Bifidobacteriales*: Identification of Molecular Signatures Specific for the Order* Bifidobacteriales *and its Different Subclades*

Grace Zhang, Beile Gao, Mobolaji Adeolu, Bijendra Khadka and Radhey S. Gupta

#### PROBIOGENOMICS AND TECHNOLOGICAL APPLICATIONS

*135 Microencapsulation in Alginate and Chitosan Microgels to Enhance Viability of* Bifidobacterium longum *for Oral Delivery*

Timothy W. Yeung, Elif F. Üçok, Kendra A. Tiani, David J. McClements and David A. Sela

*146 Modulation of the Bifidobacterial Communities of the Dog Microbiota by Zeolite*

Alberto Sabbioni, Chiara Ferrario, Christian Milani, Leonardo Mancabelli, Enzo Riccardi, Francesco Di Ianni, Valentino Beretti, Paola Superchi and Maria C. Ossiprandi

*155 Proteomic Profiling of* Bifidobacterium bifidum *S17 Cultivated Under* In Vitro *Conditions*

Xiao Wei, Simiao Wang, Xiangna Zhao, Xuesong Wang, Huan Li, Weishi Lin, Jing Lu, Daria Zhurina, Boxing Li, Christian U. Riedel, Yansong Sun and Jing Yuan

#### HOST-MICROBE AND MICROBE-MICROBE INTERACTIONS


Claudio Hidalgo-Cantabrana, Francesca Algieri, Alba Rodriguez-Nogales, Teresa Vezza, Pablo Martínez-Camblor, Abelardo Margolles, Patricia Ruas-Madiedo and Julio Gálvez

*203 Effect of* Bifidobacterium *Upon* Clostridium difficile *Growth and Toxicity When Co-cultured in Different Prebiotic Substrates*

L. Valdés-Varela, Ana M. Hernández-Barranco, Patricia Ruas-Madiedo and Miguel Gueimonde

*212 Glycan cross-feeding Activities Between bifidobacteria Under* in vitro *conditions*

Francesca Turroni, Ezgi Özcan, Christian Milani, Leonardo Mancabelli, Alice Viappiani, Douwe van Sinderen, David A. Sela and Marco Ventura


# Editorial: Bifidobacteria and Their Role in the Human Gut Microbiota

#### Francesca Turroni <sup>1</sup> \*, David Berry <sup>2</sup> and Marco Ventura<sup>1</sup>

*<sup>1</sup> Laboratory of Probiogenomics, Department of Life Sciences, University of Parma, Parma, Italy, <sup>2</sup> Division of Microbial Ecology, Department of Microbiology and Ecosystem Science, University of Vienna, Vienna, Austria*

Keywords: bifidobacteria, genomics, gut microbiota, probiotic bacteria, metagenomics

**Editorial on the Research Topic**

#### **Bifidobacteria and Their Role in the Human Gut Microbiota**

Bifidobacteria were originally isolated by Tissier at the beginning of the last century from infant stool samples and until now 57 (sub)species have been included in this bacterial genus (Turroni et al., 2011; Milani et al., 2014). Bifidobacterial biology has captured increasing attention in the last 15 years due to widespread interest in using bifidobacteria as health promoting microorganisms, i.e., known as probiotics, in the food industry. Significant efforts have been expended to dissect the genetics as well as molecular mechanisms underlying the probiotic action(s) of bifidobacteria. This has led to the establishment of a new scientific discipline called probiogenomics, which is providing new insights into the diversity and evolution of bifidobacteria and to the identification of their health-promoting effector molecules (Ventura et al., 2009; Turroni et al., 2014). Furthermore, thanks to recent discoveries about the microbial diversity of the human gut, we have started to achieve detailed insights about the composition of the bifidobacterial communities in this complex ecosystem and to understand the intricate relationship with their host as well as with the other members of the gut microbiota. Altogether, this knowledge will be crucial in order to develop novel bacterial therapeutic strategies based on bifidobacteria.

The 21 articles comprising the Research Topic "Bifidobacteria and their role in the human gut microbiota" illustrate the many key advances that now define our understanding of bifidobacteriahost molecular interactions, as well as the relationship between the various members of the Bifidobacterium genus with other residents of intestinal microbiota.

The current knowledge of the general features of bifidobacteria is reviewed, with particular focus on the metabolic features used to colonize the human gastrointestinal tract (O'Callaghan and van Sinderen) and on the specific molecular mechanisms employed by these microorganismsto interact with host tissue (Ruiz et al.; Wei et al.; Westermann et al.). In addition, the compositional changes of bifidobacterial populations associated with different stages of life are reviewed (Arboleya et al.). Variations in the composition of the human gut microbiota and bifidobacterial communities due to subsistence strategy, i.e., from hunter-gatherer to urban industrial Western lifestyle, has been studied (Soverini et al.). Also, the canine gut microbiota and the contribution of bifidobacterial taxa in this ecosystem have been explored (Sabbioni et al.).

There is growing interest in the mechanisms utilized by bifidobacteria to interact with each other in the gut ecosystem. These include specific metabolic foraging features related to glycans used for cross-feeding (Turroni et al.) as well as metabolic strategies used by bifidobacteria to assimilate nitrogen in their natural ecological niche (Ferrario et al.). The current knowledge regarding compounds that may positively influence human gut microbiota composition, such as short-chain fatty acids (SCFAs), is reviewed (Rivière et al.). Notably, these microbial end-products possibly allow the co-existence of bifidobacterial strains with other butyrate-producing bacteria in the human colon.

#### Edited by:

*Suhelen Egan, University of New South Wales, Australia*

#### Reviewed by:

*Mike Taylor, University of Auckland, New Zealand*

#### \*Correspondence: *Francesca Turroni*

*francesca.turroni@unipr.it*

#### Specialty section:

*This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology*

Received: *11 November 2016* Accepted: *20 December 2016* Published: *06 January 2017*

#### Citation:

*Turroni F, Berry D and Ventura M (2017) Editorial: Bifidobacteria and Their Role in the Human Gut Microbiota. Front. Microbiol. 7:2148. doi: 10.3389/fmicb.2016.02148*

Another important feature of bifidobacterial physiology is their ability to restrain pathogen growth in the intestine. This feature is investigated by two original research articles specifically focused on the competition with Escherichia coli O157:H7 as well as Salmonella enterica serovar Typhimurium (Bondue et al.; Vazquez-Gutierrez et al.), and Clostridium difficile (Valdés-Varela et al.). Furthermore, an investigation of the anti-viral effect of bifidobacteria toward noroviruses has been included (Li et al.).

In addition, the genomics of the order Bifidobacteriales has been explored via a phylogenetic and comparative study on proteins from all publicly available genome sequences belonging to the members of this order (Zhang et al.). Such analyses provide an in-depth overview of their evolutionary relationships and identify molecular markers that are unique to the different members of the order Bifidobacteriales at multiple phylogenetic levels. Moreover, Brandt and Barrangou propose a phylogenetic analysis of the Bifidobacterium genus using glycolysis enzyme sequences as a typing method.

The role of the gut microbiota in metabolism and metabolic disease risk has been described (Connolly et al.). The ecology of bifidobacteria has been also discussed with an opinion article focusing on the contribution of bifidobacteria to the infant gut microbiota in humans (Tannock et al.). The efficacy of the probiotic Bifidobacterium animalis subsp. lactis species, which is widely used in fermented dairy products, in the management of gastrointestinal disorders, has been investigated. In particular, two studies have been included, one investigating the effect of this species on intestinal barrier strength (Martín et al.) and another analysing the mitigating role of exopolysaccharides of B. animalis subsp. lactis in intestinal inflammatory processes such as ulcerative colitis (Hidalgo-Cantabrana et al.). Finally, technological strategies to preserve and protect cell viability of probiotic bifidobacteria that are needed to guarantee the efficacy of probiotic products are amply illustrated in this Research Topic (Yeung et al.).

The integration of various studies to define gut microbiota composition coupled with detailed analyses of the physiology and genomics of bifidobacteria will be crucial in order to improve our understanding of the complex interactions occurring in the human gut. Altogether, these data will undoubtedly help in developing the industrial use of bifidobacteria including for therapeutic use.

# AUTHOR CONTRIBUTIONS

All the authors listed have made substantial, direct and intellectual contribution to this work and approved it for publication.

#### REFERENCES


bacteria: probiogenomics. Nat. Rev. Microbiol. 7, 61–71. doi: 10.1038/nrmicro 2047

**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 Turroni, Berry and Ventura. 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.

# Bifidobacteria and Their Role as Members of the Human Gut Microbiota

Amy O'Callaghan and Douwe van Sinderen\*

Alimentary Pharmabiotic Centre and School of Microbiology, University College Cork, Cork, Ireland

Members of the genus Bifidobacterium are among the first microbes to colonize the human gastrointestinal tract and are believed to exert positive health benefits on their host. Due to their purported health-promoting properties, bifidobacteria have been incorporated into many functional foods as active ingredients. Bifidobacteria naturally occur in a range of ecological niches that are either directly or indirectly connected to the animal gastrointestinal tract, such as the human oral cavity, the insect gut and sewage. To be able to survive in these particular ecological niches, bifidobacteria must possess specific adaptations to be competitive. Determination of genome sequences has revealed genetic attributes that may explain bifidobacterial ecological fitness, such as metabolic abilities, evasion of the host adaptive immune system and colonization of the host through specific appendages. However, genetic modification is crucial toward fully elucidating the mechanisms by which bifidobacteria exert their adaptive abilities and beneficial properties. In this review we provide an up to date summary of the general features of bifidobacteria, whilst paying particular attention to the metabolic abilities of this species. We also describe methods that have allowed successful genetic manipulation of bifidobacteria.

#### Susana Delgado,

Instituto de Productos Lácteos de Asturias-Consejo Superior Investigaciones Científicas, Spain Christian Milani, Instituto de Productos Lácteos de Asturias - CSIC, Spain

> \*Correspondence: Douwe van Sinderen d.vansinderen@ucc.ie

Edited by: Marco Ventura, University of Parma, Italy

Reviewed by:

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 25 April 2016 Accepted: 31 May 2016 Published: 15 June 2016

#### Citation:

O'Callaghan A and van Sinderen D (2016) Bifidobacteria and Their Role as Members of the Human Gut Microbiota. Front. Microbiol. 7:925. doi: 10.3389/fmicb.2016.00925 Keywords: Bifidobacterium, carbohydrate metabolism, genetic modification, probiotics, microbe-host interaction

# INTRODUCTION

The past 20 years has seen a research focus on those members of the gut microbiota that exhibit health-promoting or probiotic effects such as protection of the host against pathogens by competitive exclusion (Bernet et al., 1994; Hooper et al., 1999), modulation of the immune system (O'Hara and Shanahan, 2007), and provision of nutrients through the breakdown of non-digestible dietary carbohydrates (Roberfroid et al., 1995; Leahy et al., 2005). Furthermore, compositional alterations of the gastrointestinal tract (GIT) microbiota have been linked to certain gastrointestinal diseases such as inflammatory bowel disease (Ott et al., 2004) and necrotizing enterocolitis (De La Cochetiere et al., 2004). Particular interest has focused on members of the genus Bifidobacterium, some of which have been included as live components in a variety of so-called functional foods (Ventura et al., 2004). Bifidobacteria were first isolated from the feces of breast-fed infants in 1899 by Tissier and since then bifidobacteria have been isolated from a range of different ecological niches such as the oral cavity, sewage and the insect gut, the GIT of various mammals and more recently from water kefir (Klijn et al., 2005; Ventura et al., 2007; Laureys et al., 2016).

Although, it has been well established that bifidobacteria confer positive health benefits to the human host, there is a clear lack of knowledge concerning the molecular mechanisms that explain these probiotic traits of Bifidobacterium (Cronin et al., 2011). Deciphering whole genome sequences can shed light on the genetic basis of the probiotic action of bifidobacteria, or indeed the associated molecular adaptations that allow this gut commensal to take up residency in its highly competitive ecological niche (Ventura et al., 2014). Although, a significant sequencing effort of bifidobacterial genomes has generated a very extensive set of genomic data, yet this genomic information has hardly been explored at the functional level due to a lack of tools to make bifidobacteria genetically accessible (Serafini et al., 2012).

#### GENERAL FEATURES OF BIFIDOBACTERIAL GENOMES

Since the publication of the first bifidobacterial genome in 2002, there has been a steady increase in the number of publicly available bifidobacterial genome sequences (Lee et al., 2008). The NCBI data base currently (April 2016) holds 254 publicly available bifidobacterial genome sequences, of which sixty one represent complete genome sequences (**Table 1**, source; http://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id =1678NCBI, April 2016). Three or more complete genome sequences are available for certain bifidobacterial species, such as for B. adolescentis, B. animalis, B. breve, B. bifidum, B. longum, and B. angulatum (**Table 1**).

The average size of a bifidobacterial genome is 2.2 Mb, although there is considerable size variation, for example B. indicum LMG11587 harbors a genome with a size of 1.73 Mb, wheras B. scardovii JCM12489 possesses a genome of 3.16 Mb in length. Bifidobacterial genomes typically encode 52–58 tRNA genes per genome, although there are exceptions, e.g., the genome of B. longum subsp. infantis ATCC15697 encompasses 79 tRNA-encoding genes. The number of rRNA operons within bifidobacterial genomes typically ranges from two to five, and it has been suggested that the number of rRNA operons present on a genome is correlated to the adaptation of a particular species to environmental conditions (Klappenbach et al., 2000). The G+C content of complete bifidobacterial genomes ranges from 59.2% (B. adolescentis) to 64.6% (B. scardovii), while the average gene number contained by a bifidobacterial genome is 1825 (**Table 1**). The three species B. indicum, B. coryneforme, and B. animalis possess the lowest number of genes, consistent with their small genome size (Lee et al., 2008; Ventura et al., 2014).

#### IMPACT ON HEALTH AND DISEASE

A diverse microbial community has evolved to adapt and survive in the human GIT and is commonly referred to as the gut microbiota (Guarner and Malagelada, 2003). The large intestine can contain up to 10<sup>12</sup> bacterial cells/g of luminal content making this the most densely populated area of the gastrointestinal tract (Simon and Gorbach, 1984). Members of the gut microbiota interact with their (human) host in a variety of ways, thereby making them innocuous commensals, opportunistic pathogens or health-promoting or probiotic micro organisms (Guarner and Malagelada, 2003). Probiotics are defined as "live microorganisms that, when administered in adequate amounts, confer a health benefit on the host" (FAO/WHO, 2001; Hill et al., 2014), and research into the activities of purported health-promoting bacteria has increased substantially over the last 20 years (Leahy et al., 2005). Probiotic agents have been investigated in many clinical and animal modelbased studies; however, we will summarize just a limited number of studies that specifically relate to bifidobacteria. Bifidobacteria have been commercially exploited as probiotic agents due to their associated health benefits and GRAS (Generally Recognised As Safe) status (Picard et al., 2005).

## Bifidobacteria and Colorectal Cancer

Several studies have investigated the potential of bifidobacteria to prevent and/or treat colorectal cancer. The majority of studies base their findings on murine models, and results suggest that a combination of prebiotics and bifidobacteria may reduce the occurrence of carcinogen-induced cancerous cells in mice (Sekine et al., 1985; Rowland et al., 1998; Rafter et al., 2007; Le Leu et al., 2010). For example, it was shown that B. animalis displays anti-mutagenic activity during growth in MRS broth thereby antagonizing the action of the carcinogen 2-amino-3-methylimidazo [4, 5-f] quinolone (Tavan et al., 2002). It has also been demonstrated under in vivo and in vitro conditions that a B. longum and a B. breve strain provide protection of DNA from induced damage by carcinogens, and inhibit the genotoxic effect of two different carcinogens when tested in a rat model (Pool-Zobel et al., 1996).

#### Bifidobacteria and Diarrhoea

The use of bifidobacteria to treat various gastrointestinal disorders has also been reported. For example, successful treatment of diarrhea following administration of B. longum subsp. infantis CECT 7210 and B. breve K-110 was found to be due to inhibition of rotavirus, the predominant cause of sporadic diarrhea in infants (Bae et al., 2002; Chenoll et al., 2015). Another example involves a double-blind study investigating whether oral treatment with a commercial probiotic formula containing B. bifidum and Streptococcus thermophiles would reduce antibiotic-associated diarrhea in infants. This study found that there was a significant reduction in incidences of diarrhea for those infants fed the probiotic supplemented formula supplemented (Corrêa et al., 2005).

#### Bifidobacteria and Necrotizing Entercolitis

A recent study reported lower incidences of necrotizing enterocolitis in preterm neonates following routine administration of B. breve M-16V (Patole et al., 2016). Administration of B. breve M-16V in association with breastfeeding was shown to be associated with a lower incidence of necrotising enterocolitis in neonates born before 34 weeks gestation, and, although not statistically significant, a lower incidence in this disease was reported for neonates born at a gestation age of less than 28 weeks (Patole et al., 2016).

#### TABLE 1 | Summary of all completely sequenced bifidobacterial genomes.


(Continued)



#### Bifidobacteria and Inflammatory Bowel Disease

Although, the exact mechanism of action is not understood, reduction in the symptoms of inflammatory bowel disease following treatment by probiotic strains has been reported (Venturi et al., 1999). Patients suffering from ulcerative colitis were given a probiotic preparation that includes three Bifidobacterium strains, four Lactobacillus strains and one S. thermophilus strain. Fifteen out of the 20 patients remained in remission throughout the trial, suggesting that administration of this bacterial cocktail is beneficial in maintaining remission from ulcerative colitis (Venturi et al., 1999; Gionchetti et al., 2000).

#### Bifidobacteria and Colon Regularity

A number of studies have reported improvements in colon regularity following ingestion of fermented milk products that contain B. animalis (Marteau et al., 2002; Guyonnet et al., 2007; Meance et al., 2011). Two studies have associated the administration of certain bifidobacterial strains with the alleviation of constipation (Kumemura et al., 1992; Kleessen et al., 1997). However, further investigation is needed in order to identify the precise mechanism(s) of action elicited by bifidobacteria in the prevention and treatment of constipation (Leahy et al., 2005).

#### Bifidobacteria and Competitive Exclusion

Bifidobacteria have also been reported to prevent gastrointestinal infections by competitive exclusion of pathogens based on common binding sites on epithelial cells (Duffy et al., 1994a,b; Perdigon et al., 1995; Picard et al., 2005; Gueimonde et al., 2007). Administration of high levels of bifidobacteria was shown to decrease the viable counts of Clostridium perfringens, a known producer of undesirable toxins (Tanaka et al., 1983).

#### BIFIDOBACTERIA AND FUNCTIONAL FOODS

The inclusion of micro-organisms in the human diet has been on-going for thousands of years (Leahy et al., 2005). Throughout history the most common form of administration of microorganisms was through fermented dairy products and this is still the case today (Leahy et al., 2005). Certain lactic acid bacteria, in particular certain members of the genus Lactobacillus, and members of the Bifidobacterium genus make up the vast majority of the functional ingredients present in currently commercialized probiotic food products (Salminen and Wright, 1998; Ouwehand et al., 2002). Prebiotics have been defined as "selectively fermented ingredients that allow for specific changes, both in the composition and/or activity of the gastrointestinal microflora that confer benefits upon host well-being and health" (Hijova et al., 2009). This definition has been revisited several times since it was first introduced in 1995, although these alternative definitions are in agreement that prebiotics need to be "specific" or "selective" (Gibson and Roberfroid, 1995; Roberfroid et al., 2010; Rastall and Gibson, 2015). In a recent review the definition of prebiotics was revisited and proposed as follows: "a prebiotic is a non-digestible compound that, through its metabolisation by microorganisms in the gut, modulates composition and/or activity of the gut microbiota, thus conferring a beneficial physiological effect on the host" (Bindels et al., 2015).

The newly proposed definition moves away from the requirement of "specific effect" and puts forward the arguments that: (i) our knowledge does not allow for a reliable differentiation between beneficial and detrimental members of the microbiota, (ii) a diverse community is essential for intestinal homeostasis and host physiology, (iii) the metabolic benefits assigned to prebiotics do not require a "selective" fermentation, and (iv) community-wide molecular approaches have revealed that established prebiotics are not as specific as previously assumed (Bindels et al., 2015).

One outcome from the fermentation of prebiotics by the gut microbiota is the production of short chain fatty acids (SCFAs), such as acetate, butyrate and propionate (Broekaert et al., 2011). SCFA production in the GIT results in a lower pH, improved availability of calcium and magnesium, and inhibition of potentially pathogenic bacteria (Teitelbaum and Walker, 2002; Wong et al., 2006). Both bifidobacteria and lactobacilli produce acetate (and lactate), thus contributing to the SCFA-mediated health effects of prebiotics, although these two microorganisms do not produce butyrate and/or propionate (Fukuda et al., 2011; Bindels et al., 2015). The latter SCFAs are produced by members of the Bacteroides phylum and the Clostridium clusters XIVa and IV (Louis et al., 2010; Reichardt et al., 2014; Bindels et al., 2015). Furthermore, a recent study has demonstrated that acetate produced by B. longum NCC2705 acts as an essential co-substrate for butyrate production and growth by Eubacterium rectale ATCC 33656 (Rivière et al., 2015).

Non-digestible oligosaccharides (NDOs), typically obtained from complex carbohydrates or enzymatically produced from disaccharides, represent a group of glycans that include various prebiotics (Grootaert et al., 2007). Examples of this are fructooligosaccharides (FOS) and galacto-oligosaccharides (GOS), which are among the best documented and most commonly used prebiotics on the European and Japanese markets (Grootaert et al., 2007). The prebiotic effects of FOS, GOS, inulin and lactulose have been thoroughly assessed in human trials and many studies suggest that these carbohydrates are selective by increasing bifidobacterial numbers and decreasing the numbers of E. coli and enterococci (Menne et al., 2000; Kolida et al., 2002; Bosscher et al., 2006; Kapiki et al., 2007; Davis et al., 2010; Veereman-Wauters et al., 2011; Walton et al., 2012).

Due to the professed prebiotic effects of arabinoxylan (AX) and its derivatives arabinoxylo-oligosaccharides (AXOS) and xylo-oligosaccharides (XOS), these carbohydrates have in recent times enjoyed increasing scientific interest (Broekaert et al., 2011). The bifidogenic effect of AX has been confirmed in a number of in vitro studies (Van Laere et al., 2000; Crittenden et al., 2002), while the ability of bifidobacteria to metabolize XOS and AXOS in pure culture has also been demonstrated (Jaskari et al., 1998; Van Laere et al., 2000; Crittenden et al., 2002; Palframan et al., 2003; Moura et al., 2007). AXOS consumption amongst members representing eleven different bifidobacterial species suggests that AXOS metabolism is strain dependent and rather complex (Riviere et al., 2014). In this study, five different AXOS utilization clusters were identified based on principal component analysis of the different arabinose substituent and/or xylose backbone consumption patterns. The first and largest cluster (Cluster I) was composed of 15 different strains representing seven different species (B. adolescentis, B. angulatum, B. bifidum, B. breve, B. dentium, B. longum, and B. thermophilum). Strains within this cluster were unable to utilize the substitutions or xylan backbone of AXOS, although some strains were able to utilize the monosaccharides xylose and arabinose. Cluster II was composed of eight B. longum strains that were unable to utilize the xylan backbone, yet were able to utilize the arabinose substitutions on AXOS (both mono- and di-substituted), as well as the arabinose and xylose monosaccharides. Members of the third cluster (Cluster III), encompassing 10 strains representing six different species (B. adolescentis, B. angulatum, B. longum, B. animalis, B. gallicum, and B. pseudolongum), were shown to metabolize the xylan backbone of AXOS, albeit only up to xylotetraose, while eliciting no or limited utilization of the AXOS substitutions. Cluster IV contains two B. longum strains that share the ability to completely utilize AXOS, whereas the only member of Cluster V, B. catenulatum LMG 11043, was shown to display nonpreferential degradation of XOS and a broad degradation pattern of arabinose substitutions (Riviere et al., 2014). A study investigating the in vitro fermentation of wheat-derived AX (AX-W) by human fecal microbiota reported that fermentation of AX-W was associated with the proliferation of bifidobacteria, lactobacilli and eubacteria (Hughes et al., 2007).

Several in vivo studies have also confirmed the bifidogenic effect of AX. An in vivo study in humanized rats demonstrated that long chain AX specifically stimulates the abundance of several different bacterial species in the cecum (relative bifidobacterial abundance in the cecum of the control group was 0.03 ± 0.01%, compared to 2.81 ± 1.46% in the group that were fed long chain AX; Van Den Abbeele et al., 2011). The findings of this latter study were validated by a recent study which detected the presence of two different B. longum species during the fermentation of long chain AX in an in vitro model of the proximal colon (Truchado et al., 2015). Another in vivo study found that when (high-fat) diet-induced obese mice were fed an AX-supplemented diet, a significant increase in caecal bifidobacterial numbers was observed (Neyrinck et al., 2011). Along with this increase in caecal bifidobacteria, AX supplementation restored (some of) the high-fat diet-induced changes to the microbial community.

Synbiotics are mixtures of one or more probiotics combined with one or more prebiotics (Patel and Dupont, 2015). Numerous in vivo studies have been conducted aimed at investigating the efficacy of bifidobacteria-based synbiotics in the treatment of gastrointestinal diseases and conditions. One such study investigated the synbiotic effect of B. animalis subsp. lactis B94 in combination with inulin on acute infectious diarrhea in children. Patients were administered the synbiotic agent once a day for five days and stool was examined for infectious agents such as rotavirus, Salmonella, Shigella, Campylobacter, Cryptosporidium, Adenovirus, Entamoeba histolytica, and Clostridium difficile. A marked decrease in the number of diarrhea stools was reported after 3 days of administration for the synbiotic group as compared to the control group, particularly for patients with rotavirus infection (Islek et al., 2014). A clinical trial investigated the effects of consumption of a synbiotic on the symptoms of Crohn's disease (Steed et al., 2010). The synbiotic, comprised of B. longum, inulin, and oligofructose, was consumed by patients twice daily over a 6 month period, and significant improvements in clinical outcomes were reported including a reduction in some activity indices of Crohn's disease (Steed et al., 2010). As a third example, the beneficial effect of a B. breve strain plus GOS synbiotic was investigated with regards to ulcerative colitis. The bifidobacterial strain was ingested three times a day whereas GOS was consumed once a day for 1 year. The clinical status of the treatment group significantly improved such as a marked improvement in colonoscopy scores and significant decreases in inflammatory markers. Furthermore, although no significant change in bifidobacterial numbers for those consuming the symbiotic was noticed, reduced fecal counts of Bacteriodaceae and reduced fecal pH was noted (Ishikawa et al., 2011).

# BIFIDOBACTERIAL CARBOHYDRATE METABOLISM

The human genome is predicted to encode just eight glycosyl hydrolases (GHs) that are directly linked to carbohydrate digestion. Therefore, many complex dietary carbohydrates remain un-digested and end up in the colon where they may be degraded by members of the microbiota (El Kaoutari et al., 2013).

The human GIT is home to complex microbial community that encompasses approximately 100-fold more genes than the number of genes present in the host genome (Backhed et al., 2005). Colonization of the human GIT, which is believed to occur immediately after birth, is influenced by various factors such as the method of delivery (i.e., vaginal or cesarean), type of feeding (breast-fed or formula-fed), exposure to antibiotics, frequency, and nature of diseases and hygiene conditions (Fanaro et al., 2003). Bifidobacteria dominate the total gut bacterial population in healthy breast-fed infants (Harmsen et al., 2000; Favier et al., 2002; Leahy et al., 2005), although this dominance decreases following weaning (Ventura et al., 2004). During adult life the bifidobacterial population stabilizes to represent 3–6% of the total gut microbial population, whereas in elderly (>65 years) the bifidobacterial numbers usually decline with age (Hopkins et al., 2001; Satokari et al., 2003).

The abundance and make-up of the gut microbiota is (among others) dependent on the diet of its host, and members of the microbiota have evolved effective mechanisms to utilize available nutrients (Vaughan et al., 2005; Ju-Hoon and O'Sullivan, 2010). Digestible and simple sugars such as lactose and sucrose are metabolized in the upper gut by the host and bacteria such as lactobacilli, a prevalent inhabitant of the upper GIT (Ganong, 2005; Vaughan et al., 2005). A diverse set of nondigestible carbohydrates are metabolized in the lower gut, including complex plant-derived polysaccharides (e.g., pectin, gums, hemicellulose, and xylans), host-derived carbohydrates (such as mucin and glycosphingolipids), and extracellular polysaccharides that are produced by members of the gut microbiota (Hooper et al., 2002; Korakli et al., 2002; Pokusaeva et al., 2011a). It is therefore not surprising that on average more than 12% of the annotated open reading frames within bifidobacterial genomes is predicted to encode enzymes involved in carbohydrate metabolism (Milani et al., 2014). In fact a recent study performed on the genome sequences from the type strains of each of the 47 Bifidobacterium (sub)species found that 5.5% of the core bifidobacterial genomic coding sequences (BifCOGs) is associated with carbohydrate metabolism (Milani et al., 2015).

Bifidobacteria present in the infant gut are presumed to metabolize human milk oligosaccharides (HMOs), and the genomes of B. bifidum and B. longum subsp. infantis are indeed tailored toward HMO metabolism (Sela et al., 2008; Duranti et al., 2015). However, other bifidobacterial species such as B. breve and B. longum subsp. longum are also commonly present in the infant gut. Although, they do not encode the same HMO catabolic arsenal found in B. bifidum and B. longum subsp. infantis, they can degrade certain HMOs and may also scavenge on carbohydrates that are released by other (bifido)bacteria (Egan et al., 2014a; Chaplin et al., 2015). After weaning the composition of the bifidobacterial population changes toward species capable of adapting to the metabolism of plant-derived sugars. For example, B. longum subsp. longum and B. adolescentis can utilize such diet-derived carbohydrates, while B. bifidum may shift its HMO-metabolic abilities toward mucin degradation (Schell et al., 2002; Turroni et al., 2010; Sela, 2011; Duranti et al., 2014; Egan et al., 2014a).

Prediction of the number of complete pathways used by bifidobacteria to degrade simple and complex sugars has been performed. The species B. biavatti specifies the largest number of pathways (14 complete pathways), whereas members of the species B. bombi, B. crudilactis, B. longum subsp. infantis, B. minimum, and B. ruminantium specifying just nine complete pathways (Milani et al., 2015). Bifidobacteria lack a number of key enzymes involved in the Emden-Meyerhof Parnas (EMP) pathway, instead, bifidobacteria metabolize hexose sugars through a metabolic pathway named the "bifid shunt" which is centered around the key enzyme, fructose-6-phosphoketolase (EC 4.1.2.2) (**Figure 1**; De Vries and Stouthamer, 1967; De Vuyst et al., 2014). Furthermore, the action of additional enzymes allows for a variety of carbon sources (including pentose sugars) to be channeled through this pathway (Pokusaeva et al., 2011a). Fermentation through the bifid shunt is quite advantageous for bifidobacteria as this pathway allows for the production of more energy from carbohydrates compared to that produced by the EMP fermentative pathway (Salminen and Wright, 1998; Palframan et al., 2003). The bifid shunt theoretically yields 2.5 ATP moles from every 1 mole of glucose fermented, as well as 1.5 mole of acetate and 1 mole of lactate (Palframan et al., 2003). The ratios of acetate to lactate can be influenced, however, by the particular carbohydrate being fermented as well as the growth phase and bifidobacterial species being examined (Palframan et al., 2003). Furthermore, rapid consumption of an energy source was shown to result in the production of large amounts of lactate and low amounts of acetate, ethanol and formate, whereas less lactate is produced along with an increase in production of acetate, formate and ethanol when the energy source is consumed at a slow(er) rate (**Figure 1**; Van Der Meulen et al., 2004, 2006a,b; Falony et al., 2009).

# Carbohydrate Uptake Strategies by Bifidobacteria

Bifidobacteria internalize carbohydrates by ATP-dependent ABC transporters, proton symporters or phosphoenolpyruvatephosphotransferase systems (PEP-PTS) (Turroni et al., 2012). ABC transporters couple ATP hydrolysis to efficient internalization of sugars and appear to represent the primary carbohydrate transport systems for bifidobacteria (Ventura et al., 2007; Davidson et al., 2008; Jojima et al., 2010). PEP-PTS systems allow the concomittant transport and phosphorylation of carbohydrates, while they may also be involved in the regulation of various metabolic pathways (Postma et al., 1993). The PTS component of the system is involved in the internalization and concomitant phosphorylation of carbohydrates, while PEP acts as the (indirect) phosphate donor to the recipient carbohydrate (Ventura et al., 2007). These systems are found in many bacteria

and have also been identified in most, but not all, bifidobacterial genomes (Postma et al., 1993; Pokusaeva et al., 2011a). The action of the PEP-PTS system was demonstrated experimentally in bifidobacteria whereby a PEP-PTS system from B. breve UCC2003 was found to be involved in the internalization of glucose (Degnan and Macfarlane, 1993). Since then a number of PEP-PTS systems have been identified and investigated in other bifidobacterial species (Lorca et al., 2007; Parche et al., 2007; Barrangou et al., 2009; Turroni et al., 2012; O'Connell Motherway et al., 2013).

The B. longum subsp. longum DJO10A and NCC2705 genomes are predicted to represent 10 and 13 different ABC transporters, respectively, responsible for the uptake of carbohydrates, while they each encode just a single glucosespecific PEP-PTS system (Lorca et al., 2007; Parche et al., 2007). Analysis of the B. bifidum PRL2010 genome sequence revealed that this strain encodes two ABC transporters, four PEP-PTS systems and four secondary transporters that are expected to transport mono- and disaccharides (Turroni et al., 2012). Transcriptional analysis revealed that one ABC transporter and two of the PEP-PTS systems are associated with the internalization of degradation products of host-derived glycans, in particular those that are found in mucin. The ABC transporters identified in B. bifidum PRL2010 were found to be linked to the uptake of monosaccharides such as glucose, ribose, fructose and galactose or disaccharides such as turanose (Turroni et al., 2012). Like B. bifidum PRL2010, B. breve UCC2003 is also predicted to encode four PEP-PTS systems and it has been shown experimentally that one system in B. breve UCC2003 is a fructose-inducible fructose/glucose uptake system (O'Connell Motherway et al., 2013). However, B. breve UCC2003 typically employs ABC-type transporters for the uptake of carbohydrates (Pokusaeva et al., 2011a, O'Connell Motherway et al., 2013; Egan et al., 2014b). In contrast, the strain B. animalis subsp. lactis B1-04 does not possess any PEP-PTS system and only encodes two copies of an ATP-binding protein linked to carbohydrate internalization (Barrangou et al., 2009). This low number of carbohydrate uptake systems may be a reflection of genome decay due to the commercial exploitation of B. animalis subsp. lactis, for which purpose it is extensively cultivated in nutritionally rich media (Lee et al., 2008).

#### Bifidobacterial Glycosyl Hydrolases

Carbohydrates can be modified by a range of different enzymes including hexosyl- and phosphotransferases, hydrolases and isomerizes (Pokusaeva et al., 2011a). In the presence of water, glycosyl hydrolases (GHs) hydrolyse the glycosidic bond between two or more sugars or alternatively between a carbohydrate and a non-carbohydrate moiety (Pokusaeva et al., 2011a). GHs are assigned the enzyme commission or EC number EC 3.2.1x whereby the first three numbers indicate that these particular enzymes cleave glycosyl linkages, with the assignment of the final number being based on the target substrate or mechanism of action displayed by the enzyme. Members of one GH family may not only exhibit different substrate specificity, but may also exhibit a different mode of action (Van Den Broek et al., 2008). Classification of GHs can be found at http://www.cazy.org/Glycoside-Hydrolases.html (Lombard et al., 2014). For bifidobacteria GHs are the most prevalent group of carbohydrate-modifying enzymes and since the publication of the first bifidobacterial genome more information on these GHs has become available (Van Den Broek et al., 2008).

None of the carbohydrate active GHs encoded by the human genome appear to be involved in the breakdown of FOS, GOS, XOS, inulin, or arabinoxylan (Guarner and Malagelada, 2003; El Kaoutari et al., 2013). A recent study investigated the distribution of different carbohydrate-active enzymes among 177 bacterial genomes of the human microbiome, including genomes of 12 members of the Actinobacteria phylum, half of which were bifidobacteria (El Kaoutari et al., 2013). Polysaccharide lyases and GHs accounted for 59% of all carbohydrate-active enzymes identified. From these observations it can be said that the microbiota endows metabolic activities that make up for the paucity of GHs encoded by the human genome (El Kaoutari et al., 2013).

According to the current GH classifications, B. scardovii and B. indicum LMG11587 are predicted to encode the highest (126 GHs) and lowest number of GHs (25 GHs), respectively, among the currently sequenced bifidobacterial genomes (**Table 2**). Classification according to the Carbohydrate Active Enzymes (CAZy) system revealed that 3385 genes belonging to the bifidobacterial pan-genome are predicted to represent carbohydrate active enzymes, including members of 57 GH families, 13 GT families and 7 carbohydrate esterases (Milani et al., 2015). Those enzymes belonging family GH13 are the most commonly found in bifidobacterial genomes and are known to be active against a wide range of carbohydrates including the plant-derived carbohydrates starch and related substrates, trehalose, stachyose, raffinose, and melibiose (Pokusaeva et al., 2011a; Milani et al., 2015). The bifidobacterial glycobiome contains a large number of enzymes belonging to the families GH29, GH95, GH20, GH112, GH38, GH125, GH101, and GH129 which are involved in the degradation of hostderived glycans. Members of the B. scardovii, B. longum subsp. infantis, and B. bifidum species in particular encode the most extensive set of enzymes active against host-derived glycans (Milani et al., 2015). Unlike other species, those bifidobacterial species isolated from honey/bumblebees encode a very limited set of GH13 representatives. However, these species specify a larger set of GH43 and GH3 enzymes. These families are active against plant-derived polysaccharides, such are arabino(xylan), (arabino)galactan, and cellodextran (Milani et al., 2015). Furthermore, a substantial number of genes encoding putative arabino(xylan)-degrading enzymes are present in certain bifidobacterial genomes, for example B. longum subsp. longum NCC2705, hinting at the importance of these enzymes toward the colonization of these microorganisms in the GIT (Schell et al., 2002; Van Den Broek et al., 2008).

Enzymes active against such arabinose- and xylose-containing carbohydrates have been characterized from B. longum, B. adolescentis, B. animalis subsp. lactis, and B. breve, and were first described in B. adolescentis (Van Laere et al., 1997; Lagaert et al., 2011). Seven bifidobacterial arabinofuranosidases have currently been characterized belonging to the families GH43 and GH51, five of which are produced by B. adolescentis, while two originate from B. longum (Van Laere et al., 1999; Margolles and De Los Reyes-Gavilan, 2003; Lagaert et al., 2010; Lee et al., 2011; Suzuki et al., 2013). Five intracellular xylanases have been characterized in bifidobacteria to date for both B. breve and B. adolescentis (Lagaert et al., 2007, 2011; Hyun et al., 2012; Amaretti et al., 2013). It is worth noting that all of the 14 currently characterized bifidobacterial arabino(xylan)-degrading enzymes are (predicted to be) intracellular. Due to their sizes, arabino(xylan)s cannot be transported inside the cell and the lack of extracellular enzymes specified by these bifidobacterial strains hints that they may rely on the extracellular hydrolytic activity of other members of the gut microbiota. Furthermore, the degradation of arabino(xylan) by extracellular enzymes has been observed previously for some bifidobacterial strains, and such extracellular enzymes may therefore be of significant interest as they are expected to provide an ecological advantage in the GIT (Riviere et al., 2014).

Investigations have been performed in order to determine those bifidobacterial GHs that are expected to be either secreted into the environment or associated with the cell surface. The majority of GHs located within the bifidobacterial pan-genome are predicted to be intracellular with 10.9% of GHs predicted to be secreted. The members of the family GH13 represent the largest proportion of such (predicted) secreted GHs, followed by members of the GH43 and GH51 families (Milani et al., 2015).

# CARBOHYDRATE CROSS-FEEDING BY BIFIDOBACTERIA

Several recent studies have investigated the impact on the gut microbiome by bifidobacterial cross-feeding of carbohydrates. Various studies have demonstrated that some members of the bifidobacterial community can co-operate in order to degrade large and complex polysaccharides into more simple sugars which are in turn then available to other members of the gut microbiota (De Vuyst and Leroy, 2011). This has been demonstrated for plant derived polysaccharides and also for



Frontiers in Microbiology | www.frontiersin.org



 2 | Continued



**20**

host-derived carbohydrates such as mucin (Milani et al., 2014, 2015; Egan et al., 2014a,b; Turroni et al., 2016).

Bifidobacteria were found to shape the microbiome of the murine gut either through direct action or by cross-feeding activities (Turroni et al., 2016). This study demonstrated that the addition of two or more bifidobacterial strains resulted in the enhanced levels of persistence of such strains within the murine cecum. Furthermore, members of this genus were capable of modulating gene expression toward an increase in glycan metabolism encompassing both host glycans and diet derived carbohydrates. The bifidobacterial strains had a further influence over the production of SCFAs (Turroni et al., 2016).

### CONTROL OF BIFIDOBACTERIAL CARBOHYDRATE METABOLISM

Carbon catabolite repression (CCR) is a regulatory system present in many bacteria in which the expression or activity of proteins involved in the utilization or uptake of available carbon sources is inhibited by the presence of a preferred carbon source (Postma et al., 1993; Saier and Ramseier, 1996). There is evidence that a CCR mechanism is operating in bifidobacteria, although as of yet a CCR-type regulatory system has not been described for any member of this genus. The first report on CCRrelated metabolism in bifidobacteria was in B. animalis subsp. lactis (Trindade et al., 2003). This study reported an induction in sucrose metabolizing activities when this strain was grown on sucrose, raffinose or oligofructose, whereas a repression in the same metabolic activities was reported for growth on glucose (Trindade et al., 2003). An apparent reverse CCR was reported for B. longum NCC2705 in that glucose transport was repressed when lactose was present in the growth medium (Parche et al., 2006). There are two separate reports of CCR-like regulation in B. breve UCC2003 (Ryan et al., 2005; Pokusaeva et al., 2010). The first study reported that transcription of the rbs operon, responsible for ribose metabolism, is induced when B. breve UCC2003 is grown on ribose, whereas transcription of this operon is not induced (or repressed) when the strain is grown on a combination of ribose and glucose (Pokusaeva et al., 2010). The second study on B. breve UCC2003 demonstrated that when grown on sucrose or Actilight (a commercial source of short chain FOS), transcription of the fos operon, which is involved in the metabolism of FOS, is induced (Ryan et al., 2005). However, this operon was not induced (or repressed) when B. breve UCC2003 is grown on a combination of sucrose and glucose, or sucrose and fructose (Ryan et al., 2005).

Transcriptional repressors, such as LacI-type transcriptional regulators, are DNA binding proteins that physically interact with a specific DNA sequence, called the operator, in the vicinity of a regulated promoter, thereby preventing the binding of an RNA polymerase and the initiation of transcription (Ravcheev et al., 2014). Transcriptional repressors may also cause a "road block" for the DNA polymerase thus preventing the progression of transcription (Ravcheev et al., 2014). A substantial number of these repressor proteins have been identified in bifidobacterial genomes, e.g., B. longum NCC2705 is predicted to encode 22 LacI-type transcriptional repressors. The presence of a sugar-binding motif in each of these predicted 22 LacItype proteins indicates that they are predicted to be involved in the regulation of carbohydrate metabolism (Schell et al., 2002). Six different LacI-type regulators have been characterized in B. breve UCC2003 and include LacIfos, GalR, CldR, and RbsR, which regulate transcription of the fos operon, the galactan utilization cluster, the cellodextrin utilization cluster, and the ribose metabolism cluster, respectively (Ryan et al., 2005; Pokusaeva et al., 2010, 2011b; O'Connell Motherway et al., 2011a). More recently, two B. breve UCC2003-encoded LacI-type transcriptional repressors, named MelR1 and MelR2, were shown to control the melezitose utilization cluster (O'Connell et al., 2014).

Other carbohydrate-dependent regulatory systems exist, for example, the RafR regulator is a transcriptional activator of the raffinose metabolism gene cluster in B. breve UCC2003 (O'Connell et al., 2014), while in the same bacterium a GntR-type transcriptional repressor was shown to control transcription of the large nag/nan gene cluster, thereby regulating metabolism of sialic acid (Egan et al., 2015).

# BIOTECHNOLOGY OF BIFIDOBACTERIA

Research into bifidobacterial metabolism has led to the prebiotic concept which in turn confers health benefits to the human host by stimulating the metabolism and activity of bifidobacteria in the GIT (Hijova et al., 2009). However, the actual mechanisms of action that are responsible for such probiotic effects are far from fully understood (Sun et al., 2012). The gold standard approach to investigate the role of a single gene and its products is by site-directed mutagenesis and subsequent phenotypic analysis of the generated mutant(s) (Sun et al., 2012). Unfortunately, bifidobacteria are rather recalcitrant to artificial DNA acquisition methods which means that genetic modification of bifidobacteria has up until recently proven to be impossible. This part of the review will focus on genetic approaches and techniques aimed at improving the genetic accessibility of bifidobacteria.

#### BIFIDOBACTERIAL MUTAGENESIS STRATEGIES

There are a small number of reports on targeted gene inactivation in bifidobacteria, although in recent years a number of techniques have been developed resulting the successful inactivation of genes in bifidobacteria. In this part of the review, we describe the fundamental concept of each strategy and discuss some of benefits and pitfalls associated with each.

#### Single-Crossover Plasmid Insertion

This method involves the use of a non-replicating plasmid to select for homologous recombination events (**Figure 2A**). The first successful mutation created in a bifidobacterial strain was in the gene apuB using this single-crossover plasmid insertion approach whereby a combination of plasmids facilitated the conditional replication of the non-replicative plasmid pORI19

(**Table 3**; O'Connell Motherway et al., 2008). Although, it successfully led to the inactivation of the apuB gene, this method was quite time consuming and laborious (Fukiya et al., 2012).

As already discussed above, by-pass of native R-M systems and the single-crossover plasmid insertion approach using a non-replicating plasmid has been successfully and repeatedly employed for mutagenesis (O'Connell Motherway et al., 2009, 2011b, 2013; Fouhy et al., 2013; Christiaen et al., 2014; Egan et al., 2014a,b, 2015; O'Connell et al., 2014).

#### Double-Crossover and Double-Crossover markerless Gene Deletion

Double-crossover gene disruption was first demonstrated in B. longum NCC2705 (**Table 3**; Fukuda et al., 2011). This method makes use of a non-replicating plasmid and a double-cross over event in order to create a gene deletion (Fukiya et al., 2012). The non-replicative vector harbors two homologous regions of the target gene between which an antibiotic resistance gene is inserted (**Figure 2B**). During the first crossover event, homologous recombination occurs between one of the two homologous regions on the vector and on the chromosome. The second (desired) crossover occurs between the second homologous region on the vector and the chromosome, and results in the replacement of the wild-type allele with the mutated one and an antibiotic marker (**Figure 2B**). In the case of B. longum NCC2705, this approach allows for the successful deletion of (part of) the gene BL0033, a solute binding protein of an ABC transporter that is highly induced when B. longum NCC2705 is grown on fructose (Fukuda et al., 2011).

The double-crossover marker-less strategy has been patented and draws some similarities to the double-crossover gene disruption approach (**Table 3**, **Figure 2C**; Arigoni and Delley, 2013). Like the previous approach, a non-replicative plasmid is used for the first crossover, however, in this case the antibiotic marker is located beside the mutated targeted gene (**Figure 2C**). The second crossover event occurs during longterm sub-culturing of the first crossover integrants. As illustrated in **Figures 1**, **2C** and as described previously for the doublecrossover gene disruption, a markerless gene deletion can be generated following a second crossover. An amended version of this method was used to successfully create a marker-less gene deletion in B. longum 105-A which involved the conditional replication of a vector plasmid (**Table 3**; Hirayama et al., 2012).

The plasmid used for the homologous recombination events is a conditionally replicating plasmid due to a deletion of the gene encoding the replication protein RepA. This plasmid can only replicate in bifidobacteria when the RepA protein is provided in trans. The regions flanking the target gene to be deleted are cloned into the conditional replication vector (Hirayama et al., 2012). Once the plasmid has integrated into the target gene via the first homologous recombination or cross-over

#### TABLE 3 | Summary of bifidobacterial mutagenesis strategies.


event, a second plasmid that encodes RepA is introduced which induces the conditional replication of the integrated plasmid (**Figure 2C**). During the second cross-over and excision, the integrated plasmid is expected to be excised along with the target gene. Second cross-over recombinants that lost the first plasmid can be selected for using appropriate antibiotic markers as the second plasmid is incompatible with the conditional replicating vector (Hirayama et al., 2012).

The double cross-over approaches are more time consuming due to multiple transformations and extensive screening for positive mutants, although the improved double crossover gene deletion system aims to address this (**Table 3**; Fukiya et al., 2012). For single cross-over mutants it is theoretically possible for additional cross-over events to occur between the homologous regions left in the target gene thus resulting in the excision of the integrated plasmid (Fukiya et al., 2012). It is also worth noting that the marker-less method is a superior approach as it does not leave an antibiotic resistance gene, the presence of which may cause polar effects affecting genes down-stream of the mutated gene (**Table 3**; Fukiya et al., 2012).

#### Homologous Recombination Mediated by a Temperature Sensitive Plasmid

The temperature sensitive (Ts) plasmid strategy has been applied successfully to various microorganisms in order to create gene knock-outs (Hamilton et al., 1989; Maguin et al., 1992; Biswas et al., 1993; Takamatsu et al., 2001; Fuchs et al., 2006; Chen et al., 2011). The advantage of this approach over some of the other strategies described above, is that Ts plasmids do not require high transformation efficiencies. Therefore, this would be an ideal and widely applicable approach for (transformation recalcitrant) bifidobacterial strains (**Table 3**; Sakaguchi et al., 2012).

The successful creation of insertion mutants using a Ts plasmid has been achieved in B. breve UCC2003, B. longum 105-A and B. longum NCC2705 (O'Connell Motherway et al., 2008; Sakaguchi et al., 2012). A plasmid harboring repA was first introduced into B. breve UCC2003 followed by the introduction of a non-replicating plasmid that harbored an internal fragment of the target gene, apuB (**Table 3**, **Figure 3**; O'Connell Motherway et al., 2008). Once both plasmids were successfully introduced, growth of B. breve UCC2003 was shifted to 42◦C, thus preventing further replication of the repAcontaining plasmid. As a consequence this also blocks the replication of the pORI19-derivative. Selection on antibiotic results in the integration of the pORI19-derivative into the B. breve UCC2003 genome at the site of the apuB gene (**Figure 3**; O'Connell Motherway et al., 2008). This temperature-sensitive plasmid approach was also used to create insertion mutants in B. longum 105-A and B. longum NCC2705 (Sakaguchi et al., 2012).This method was applied to create an insertion mutant in the strain B. longum 105-A using pyrR as the gene target and was further validated by re-creating a gene deletion in the gene BL0033 in the strain B. longum NCC2705 (**Table 3**; Sakaguchi et al., 2012).

### Transposons for Mutagenesis in Bifidobacteria

A recent study has described the creation of a mutant library in B. breve UCC2003 using a Tn5-based transposon mutagenesis system (Ruiz et al., 2013). This study is the first to demonstrate the application of random-mutagenesis technology in bifidobacteria (**Table 3**). This approach involved the construction of a tetracycline-resistant Tn5 transposon and preparation of a transposome complex that was subsequently introduced into B. breve UCC2003 cells by electroporation (Ruiz et al., 2013). Screening of the mutant bank led to the identification of mutants that display defective growth on selected carbohydrates, such as Bbr\_0010 which encodes a β-galactosidase involved in the metabolism of lactose (Ruiz et al., 2013).

The benefit of this method is that it can be used for high-throughput screening in order to identify genes that are fundamental for a given phenotype (Judson and Mekalanos, 2000). However, similarly to targeted insertion mutant systems, high transformation efficiencies are crucial for this system to be effective (Ruiz et al., 2013).

# CONJUGATION IN BIFIDOBACTERIA

Conjugation can also be used by bacteria to transfer genetic material, as was first described in E. coli and involving cellto-cell contact (Lederberg and Tatum, 1946). Conjugationbased techniques have been widely described for Gram-negative bacteria, though less so for Gram-positive bacteria (Schroder and Lanka, 2005).

Currently, transformation by electroporation is the most popular method used to genetically manipulate bifidobacteria. Shortfalls maybe overcome by the use of a

UCC2003 genome.

conjugation-based approach as it holds a number of advantages over transformation (Dominguez and O'Sullivan, 2013). Firstly, unlike electroporation, the size of the vector does not affect conjugation efficacy (Szostková and Horáková, 1998; Isaacs et al., 2011). Secondly, the presence of R-M systems is not an issue for conjugative methods as during conjugation DNA is transferred as a single strand from donor to recipient, in which form it is insensitive to most R-M systems (Dominguez and O'Sullivan, 2013).

The first, though unsuccessful, attempt at conjugative transfer in bifidobacteria was reported in 1998 (Shkoporov et al., 2008). A recent publication describes a conjugative gene transfer method and the first successful transfer of DNA between E. coli and different Bifidobacterium species (Dominguez and O'Sullivan, 2013). Based on the RP4 conjugative machinery of E. coli WM3064 (pBB109), the E. coli-Bifidobacterium shuttle vector pDOJHR-WD2 was constructed. This plasmid was transformed into E. coli WM3064 harboring pBB109. This latter plasmid encodes the relaxase function to catalyze nicking at the oriT site on pDOJHR-WD2, while RP4, which is integrated into the E. coli WM3064 chromosome, specifies Tra functions (Dominguez and O'Sullivan, 2013). Successful transfer of pDOJHR-WD2 from E. coli to members of four different bifidobacterial species was reported with an efficiency ranging from 1.8 × 10−<sup>4</sup> to 7.5 × 10−<sup>6</sup> transconjugants per recipient (Dominguez and O'Sullivan, 2013).

The recent sequencing of the B. breve JCM 7017 genome led to the discovery of a 190 kb megaplasmid designated as pMP7017 (Bottacini et al., 2015). Analysis of the plasmid sequence led to the identification of genes that are predicted to encode conjugative machinery. Sequence analysis of pMP7017 led to the observation that, although the overall G+C content is akin to that of its host, it appears that this megaplasmid was formed from the cointegration of two separate modules (Bottacini et al., 2015). The functionality of the conjugative mechanism was demonstrated by the successful transfer of pMP7017 from B. breve JCM7017 to B. breve and B. longum representatives (Bottacini et al., 2015).

The recent development of conjugation-based gene transfer systems and the discovery of the first native conjugative bifidobacterial plasmid is quite significant (Dominguez and O'Sullivan, 2013; Bottacini et al., 2015), and may assist in future tool development for the genetic modification of bifidobacteria.

#### MICROBE-HOST INTERACTIONS

Insertion mutagenesis has proven to be a fundamental tool in the development of functional genomics in bifidobacteria. Two studies in particular demonstrate how the use of insertion mutagenesis can reveal how candidates genes are involved in the beneficial effects conferred to the host by bifidobacteria (Fukiya et al., 2012).

In the first example, Fukuda et al. combined comparative genomics, gene expression profiling and gene knock-outs to demonstrate that acetate production by the strain B. longum subsp. longum JCM 1217 is linked to the protection of host epithelial cells from lethal doses of the shiga toxin, Stx2 (Fukuda et al., 2011). Early observations highlighted that germ-free mice pre-colonized with B. longum subsp. longum JCM 1217 and subsequently infected with enterohaemorrhagic E. coli O157:H7 (EHEC O157:H7), survived better than germ-free mice that were infected with EHEC O157:H7. In contrast, the same survival was not observed when, prior to infection with EHEC O157:H7, mice were colonized with B. adolescentis JCM 1275. Comparative genomic analysis performed between protective B. longum subsp. longum and non-protective B. adolescentis implicated an ABCtype transporter system being involved in protection, after which a knock-out mutation was created in the gene encoding the corresponding solute binding protein using a double crossover gene disruption approach (Fukuda et al., 2011). Due to the consequently lower levels of acetate produced, the mutant strain was unable to protect mice from EHEC O157:H7 infection, whereas when the non-protective B. adolescentis strain expressed the ABC transporter cluster this resulted in moderate increases in mouse survival following EHEC O157:H7 infection as well as increased acetate production.

The second example of a functional microbe-host interaction study is the identification of candidate genes directly responsible for the colonization of B. breve UCC2003 in a murine colonization model (O'Connell Motherway et al., 2011b). Genome and in vivo transcriptome analyses of B. breve UCC2003 revealed that a gene cluster, responsible for the production of so-called type IVb or tight adherence (Tad) pili,is required for host colonization (O'Connell Motherway et al., 2011b).

As well as investigating candidate genes directly involved in host-microbe interactions, insertion mutagenesis has also been exploited in functional genomic studies investigating the physiological characteristics of bifidobacteria, such as carbohydrate metabolism (O'Connell Motherway et al., 2011a; O'Connell et al., 2013, 2014; Egan et al., 2014b, 2015).

# CONCLUSION

It is well established that bifidobacteria confer positive health benefits to their host via their metabolic activities. The availability of complete bifidobacterial genomes and corresponding comparative analysis allows for the identification of mechanisms underlying bifidobacterial metabolic activity. Carbohydrate utilization studies and identification of metabolic pathways also provides fundamental information allowing for the identification of novel and effective prebiotic compounds.

Plant-derived and host derived carbohydrates have been shown to stimulate the growth of some bifidobacterial species. To identify and obtain full knowledge of the genes implicated in carbohydrate degradation and utilization, characterization and mutagenesis of candidate genes is required. However, bifidobacteria are notoriously recalcitrant to genetic modification.

It is therefore essential that future studies continue to address the shortage of effective molecular tools. The development of these tools is essential to unravel the underlying molecular mechanisms that explain how bifidobacteria interact with their human host.

### AUTHOR CONTRIBUTIONS

All authors listed, have made substantial, direct and intellectual contribution to the work, and approved it for publication.

#### REFERENCES


#### FUNDING

This publication has emanated from research conducted with the financial support of Science Foundation Ireland (SFI) under Grant Number SFI/12/RC/2273. AO was supported by an enterprise partnership scheme of the Irish Research Council.


co-cultivation with Bifidobacterium bifidum PRL2010 in a mucin-based medium. BMC Microbiol. 14:282. doi: 10.1186/s12866-014-0282-7


Bifidobacterium longum B667. Appl. Environ. Microbiol. 69, 5096–5103. doi: 10.1128/AEM.69.9.5096-5103.2003


Appl. Environ. Microbiol. 72, 1006–1012. doi: 10.1128/AEM.72.2.1006-1 012.2006


**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 O'Callaghan and van Sinderen. 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.

# Gut Bifidobacteria Populations in Human Health and Aging

Silvia Arboleya1,2, Claire Watkins1,2,3, Catherine Stanton1,2 and R. Paul Ross1,2,4 \*

<sup>1</sup> APC Microbiome Institute, University College Cork, Cork, Ireland, <sup>2</sup> Teagasc Food Research Centre, Moorepark, Fermoy, Cork, Ireland, <sup>3</sup> School of Microbiology, University College Cork, Cork, Ireland, <sup>4</sup> School of Science, Engineering and Food Science, University College Cork, Cork, Ireland

The intestinal microbiota has increasingly been shown to have a vital role in various aspects of human health. Indeed, several studies have linked alterations in the gut microbiota with the development of different diseases. Among the vast gut bacterial community, Bifidobacterium is a genus which dominates the intestine of healthy breast-fed infants whereas in adulthood the levels are lower but relatively stable. The presence of different species of bifidobacteria changes with age, from childhood to old age. Bifidobacterium longum, B. breve, and B. bifidum are generally dominant in infants, whereas B. catenulatum, B. adolescentis and, as well as B. longum are more prevalent in adults. Increasingly, evidence is accumulating which shows beneficial effects of supplementation with bifidobacteria for the improvement of human health conditions ranging from protection against infection to different extra- and intra-intestinal positive effects. Moreover, bifidobacteria have been associated with the production of a number of potentially health promoting metabolites including short chain fatty acids, conjugated linoleic acid and bacteriocins. The aim of this mini-review is to describe the bifidobacteria compositional changes associated with different stages in life, highlighting their beneficial role, as well as their presence or absence in many disease states.

Keywords: bifidobacteria, intestinal microbiota, health, aging, probiotics

# INTRODUCTION

The study of the intestinal microbiota, linked with its impact on human health, has become a transforming topic in microbiology research. From the pioneering studies using culture-dependent techniques to the current omics' approach, composition and functionality of the intestinal microbiota has been assessed in a wide variety of studies (Lagier et al., 2015). The complex microbiota and the relationships it plays in the prevention and alleviation of diseases have been a major focus of many works (Turnbaugh et al., 2009; Qin et al., 2010; Karlsson et al., 2013). Members of the genus Bifidobacterium have been identified as almost ubiquitous inhabitants of the human host (Biavatti and Mattarelli, 2006), performing an important role in the gut from birth. Due to the traditional interest in bifidobacteria as probiotics this genus has been studied in depth, and changes in the species composition, diversity or relative abundance have been investigated at different stages of life and in several diseases. The aim of this mini-review is to describe the bifidobacteria compositional changes associated with different stages in life, highlighting their beneficial role, as well as the possible role of their presence in the protection against many disease states.

# Edited by:

Marco Ventura, University of Parma, Italy

#### Reviewed by:

Sinead Christine Leahy, AgResearch, New Zealand Simone Guglielmetti, University of Milan, Italy Silvia Turroni, University of Bologna, Italy

> \*Correspondence: R. Paul Ross p.ross@ucc.ie

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 30 May 2016 Accepted: 20 July 2016 Published: 19 August 2016

#### Citation:

Arboleya S, Watkins C, Stanton C and Ross RP (2016) Gut Bifidobacteria Populations in Human Health and Aging. Front. Microbiol. 7:1204. doi: 10.3389/fmicb.2016.01204

# INTESTINAL MICROBIOTA

fmicb-07-01204 August 17, 2016 Time: 18:44 # 2

The intestinal microbiota plays an important role in human health and has long been associated with such functions as metabolic, protective, and trophic (Guarner and Malagelada, 2003) and more recently functions related to the gut-brain axis or liver-gut axis (Clarke et al., 2014). The current exponential increase in sequencing and the explosion of other "omics" approaches has allowed us to gain a deep knowledge of intestinal microbiota.

The establishment of the intestinal microbiota was initially considered to occur at birth; however, the presence of microorganisms in placenta or amniotic fluid (Collado et al., 2016) suggests a primary fetal colonization. Moreover, the process is determined by different factors which undoubtedly have an effect on microbiota homeostasis (Penders et al., 2006; Faa et al., 2013). Actinobacteria, followed by Proteobacteria and Firmicutes are the main phyla in early childhood, characterized by low diversity and complexity (Koenig et al., 2011; Turroni et al., 2012). The main changes in gut microbiota composition take place in the first stages of life, getting to a relative stability at 1–2 years old (Voreades et al., 2014). The adult-like structure of the gut microbiota is thought to occur after the 3rd year of life (Koenig et al., 2011; Yatsunenko et al., 2012), and reaches a total number of 10<sup>14</sup> microorganisms, comprising of bacteria, eukarya, viruses and archaeal members (Mihajlovski et al., 2010; HMP, 2012). At phylum level, the gut microbiota is made up of 80–90% Firmicutes and Bacteroidetes. At species and strain taxonomic level, the diversity is very high in adult life, and is characterized by high interindividual variability (HMP, 2012). However, the functionality and metabolism of the gut microbiota is highly conserved (Qin et al., 2010). Aging was defined by Imahori (1992) as "the regression of physiological function accompanied by advancement of age." It is a natural process which entails changes in the gastrointestinal tract (GIT), immunosenescence and, in some cases, malnutrition (Woodmansey, 2007; Biagi et al., 2012). In addition, alterations in lifestyle, diet and medication have an unavoidable effect on the elderly microbiota composition and function (Biagi et al., 2012; Claesson et al., 2012). Microbial diversity is reduced in old age; however, Bacteroidetes and Firmicutes still constitute the dominant phyla, with increases in the relative abundance of some other Phyla – most notably Proteobacteria (Biagi et al., 2012; Odamaki et al., 2016).

# BIFIDOBACTERIA AND THEIR RELEVANCE: FROM THE INFANT-TYPE TO THE AGED-TYPE MICROBIOTA

Bifidobacteria are normal inhabitants of the GIT belonging to the Actinobacteria phylum. After the depletion of oxygen by facultative anaerobes, bifidobacterial populations are the most abundant genus present in the healthy infant gut (Favier et al., 2002). During adulthood the levels decrease considerably but remain relatively stable; decreasing again in old age (Odamaki et al., 2016; **Figure 1**). Certain strains of Bifidobacterium are widely used as probiotics – the safety of which is supported by the long historical consumption in fermented milks and the growing knowledge about their physiology and genomes (Arboleya et al., 2016; O'Callaghan and van Sinderen, 2016). The genus Bifidobacterium has been shown to play an important role in the barrier effect, the stimulation of immune system, being associated with a range of beneficial health effects (Picard et al., 2005).

## Early Stages of Life

The initial colonization by bifidobacteria is dependent on a number of extrinsic factors. In terms of vertical transfer, many studies have linked the transmission of bifidobacteria from the mother's vaginal tract, GIT, breast milk, placenta and amniotic fluid to the infant (Makino et al., 2013; Collado et al., 2016). Birth mode, in particular, has a significant impact on this initial colonization, with an increased abundance of bifidobacteria found in infants born vaginally, versus those born by cesarean section (Dominguez-Bello et al., 2010).

Gestational age has been described in terms of its impact on the infant gut, whereby pre-term infants have been characterized in many studies by a dominance of Proteobacteria, with increases in members of Clostridium and Staphylococcus, and much lower levels of Actinobacteria. In contrast, the full-term infant gut has been correlated with much higher levels of Bifidobacterium and Bacteroides, which tend to be dominant in the early weeks of life (Arboleya et al., 2012; Barrett et al., 2013; Arboleya et al., 2015).

Numerous studies investigating the effects of breast feeding versus formula feeding have identified specific bifidobacterial species that correlate with the feeding regime (Roger et al., 2010). Using both culture-dependent and molecular methods, studies have found that Bifidobacterium breve, B. bifidum, B. longum ssp. longum, and B. longum spp. infantis are present in both breast- and formula-fed infants (Mevissen-Verhage et al., 1987; Klaassens et al., 2009). However, B. longum ssp. infantis is more commonly associated with breast fed infants, whereas B. longum ssp. longum has been found more commonly amongst bottle-fed infants (Guaraldi and Salvatori, 2012). B. adolescentis, which is commonly found in adults, was found to be present only in formula-fed babies (Klaassens et al., 2009). Numerous studies have focused on the bifidogenic effect that human breast milk has on the infant gut microbiota (Musilova et al., 2014). Specific glycans found within breast milk are known to be utilized by bifidobacteria; however, it has been shown that breast milk provided by mothers with an inactive allele of the Fucosyltransferase 2 gene (FUT2), an enzyme involved in the transfer of fucose to glycans, delays the establishment of bifidobacteria in the infant (Lewis et al., 2015). Similarly, bifidobacteria provide an important role in the breakdown of human milk oligosaccharides (HMOs; Garrido et al., 2013), creating a clear evolutionary link between the mother, infant and the microbial species present. Indeed, B. longum ssp. infantis has specifically been studied for its ability to digest different HMO structures (Sela et al., 2008). This species is also associated with anti-inflammatory properties, and the ability to decrease intestinal permeability (Underwood et al., 2015). The growth of infant-derived Bifidobacterium, B. longum ssp. infantis, and B. bifidum, in the presence of HMOs has also been shown

to promote the adhesive properties of these strains, as well as the expression of anti-inflammatory cytokines and tight junction proteins in eukaryotic cells (Chichlowski et al., 2012). Interestingly, a study published by Del Chierico et al. (2015) focused on the co-correlation between the metabolome and the OTUs occurrence in infant stool samples. Results found that Bifidobacterium OTUs were positively correlated with the presence of lactate and alanine during this 1st month of life. Another recent work, negatively correlated the lactase gene with Bifidobacterium levels in the gut microbiota of twins, speculating that lactase-persisters babies harbor lower levels of bifidobacteria because of low lactose availability (Goodrich et al., 2016).

In terms of species, B. longum, B. breve, and B. bifidum are most abundant (**Table 1**) in infants (Favier et al., 2002; Gueimonde et al., 2007; Turroni et al., 2012). A recent study described the different Bifidobacterium species profile between monozygotic twins (entirely dominated by B. breve) and their fraternal sibling (exhibiting higher species diversity) at 1 month of life (Murphy et al., 2015; **Table 1**).

#### Adulthood and Old Age

In adulthood, the levels of bifidobacteria are lower (2–14% relative abundance) but remain stable (Odamaki et al., 2016). Gueimonde et al. (2007) identified significantly higher levels of Bifidobacterium in infants than in adults by q-PCR technique. They postulated that B. longum is the most widely abundant species, which is in agreement with other studies (Gavini et al., 2001; Matsuki et al., 2004). However, Matsuki et al. (2004) observed higher levels of B. adolescentis and B. catenulatum in their adult population. Chaplin et al. (2015) reported a decreased frequency of isolation of B. bifidum and B. breve with age and an increased trend in B. adolescentis.

Currently, there is no agreed definition of old-age-specific gut microbiota profile due to the high inter-individual variability, differences in diet and lifestyle, and the unclear definition of the term "elderly." However, some trends are repetitively observed such as the decrease of bifidobacteria in the elderly population, confirmed by several studies using different technologies (Mitsuoka et al., 1974; Mitsuoka, 1992; Gavini et al., 2001; Hopkins et al., 2001; Gueimonde et al., 2010; Biagi et al., 2012; Salazar et al., 2013). Woodmansey et al. (2004) reported that the decline in bifidobacteria population with aging was accompanied by a decrease in species diversity. This decline was associated with the reduction in adhesion to the intestinal mucosa, but it is not clear if it is due to the changes in the microbiota or in the structure of mucus (He et al., 2001).

TABLE 1 | (A) Distribution of the most abundant Bifidobacterium species in the intestinal microbiota at different stages of life analyzed using different techniques.


(Continued)

#### TABLE 1 | Continued

fmicb-07-01204 August 17, 2016 Time: 18:44 # 5


First row indicates the most abundant species found<sup>a</sup> . Second row indicates other abundant species found<sup>b</sup> . (B) Studies focusing on Bifidobacterium species altered in certain diseases.

↑ Increased levels; ↓ Decreased levels.

The bifidobacteria composition in centenary populations was also reported in some studies; however, the results remain somewhat controversial. In a European population, the microbiota composition of centenarians was still similar to that of adults (Biagi et al., 2010), however, higher proportions of bifidobacteria were found in centenarians than in younger elderlies from a region of China (Zhao et al., 2011). Regarding species (**Table 1**), B. longum was the most abundant in Italian centenarians followed by B. adolescentis and B. bifidum (Drago et al., 2012), but B. dentium was dominant in Chinese centenaries (Wang et al., 2015).

Other extrinsic factors indirectly related to the aging process also affect the bifidobacteria composition. The extended use of antibiotics in the older population undoubtedly has a huge impact on the intestinal microbiota composition, decreasing the bifidobacteria population (Woodmansey et al., 2004; O'Sullivan et al., 2013). While antibiotics remain an essential medical tool, therapies targeted toward the reestablishment of microbiota have been explored, in particular the use of probiotics to correct the imbalance in the bifidobacteria population and the alteration in the intestinal microbiota after antibiotic therapy (Rondanelli et al., 2015). In terms of frailty, van Tongeren et al. (2005) did not find differences in bifidobacteria between elderly people divided into a low and high frailty category. The same trend was observed by Bartosch et al. (2004) between elderly living in the local community and elderly in a hospitalized environment.

#### BIFIDOBACTERIA AND DISEASES

Several diseases, both intra- and extra-intestinal, have been associated with alterations in the gut microbiota composition and function (Wu and Lewis, 2013). Although, there is still not a detailed description of "potential alterated-microbiota types," some authors postulate that intestinal microbial alterations could be the prelude to a wide range of disease (Putignani et al., 2014).

#### Bifidobacterial Composition in Diseases

Given the widespread use of bifidobacteria as probiotics, they have been studied extensively and as such the aberrancies in bifidobacteria species composition, diversity or changes in their relative abundance have been reported in several diseases.

Obesity is a worldwide disease affecting children and adults, which is commonly associated with alterations in the microbiota. Some studies have shown lower levels of bifidobacteria, linked to higher prevalence of enterobacteria or Staphylococcus in obese children (Kalliomäki et al., 2008; Gao et al., 2015). Interestingly, women who gain weight during pregnancy have displayed lower levels of Bifidobacterium in contrast to healthy weight pregnant women (8.36 vs. 9.10 log genome equivalents/g feces; Santacruz et al., 2010). These results are correlated with the decreased levels of bifidobacteria in babies whose mother gained significant weight during pregnancy (Collado et al., 2008b), that is over and above the pregnancy itself. In terms of allergic disease, Hevia et al. (2016) observed lower levels of Bifidobacterium in patients with long-term asthma. The same study showed a dominance of B. adolescentis in both short- and long-term asthmatic individuals, in concordance with other previous studies (Stsepetova et al., 2007). This species was only found in infants of allergic mothers, who displayed lower levels of bifidobacteria in their breast milk (Grönlund et al., 2007). One particular Turkish study described a statistical difference between the increased levels of B. longum in healthy children (30.3%) when compared to children with allergic disease (11.1%), suggesting that B. longum may play a beneficial role in the disease and thus may be useful as a probiotic for the prevention of allergic pathologies (Akay et al., 2014). Different studies have focused on the relationship between the intestinal microbiota and the pathogenesis of irritable bowel syndrome (IBS), showing an altered microbiota related to IBS patients and lower levels in the Bifidobacterium genus (Taverniti and Guglielmetti, 2014).

In the elderly gut microbiota, Hopkins and Macfarlane (2002) focused on the impact of Clostridium difficile-associated diarrhea, describing a reduction in numbers of bifidobacteria in elderly people suffering the infection compared to a healthy control group. Decreased numbers of bifidobacteria have also been observed in other illnesses such as cystic fibrosis, hepatitis B and both diabetes Types I and II (Wu et al., 2010; Xu et al., 2012; Duytschaever et al., 2013; Murri et al., 2013).

Overall, there is a repetitive trend between lower bifidobacteria and a variety of common disease states, suggesting a role

of bifidobacteria in health. However, whether bifidobacteria numbers have any causal relationship to any of these conditions remains unknown.

#### Bifidobacteria as Probiotics in Diseases

Numerous health-promoting effects have been ascribed to strains of the Bifidobacterium genus based on their use as probiotics in intervention strategies to address many ill health conditions.

In that regard, the capability of bifidobacteria to stimulate the immune system seems to be species specific as shown in a particular study where the effect of infant derived bifidobacteria on the T-helper 1(TH1)/TH2 balance was examined. Results published by Ménard et al. (2008) demonstrated that B. bifidum, B. dentium, and B. longum were capable of stimulating systemic and intestinal immunity. It is not surprising that bifidobacteria are so widely used as probiotics in the treatment and prevention of infant disease given their dominance in the infant gut. Their application in pathologies such as allergies, celiac disease, obesity, diarrheas, colic, infections or necrotizing enterocolitis has returned very good results (Di Gioia et al., 2014). They have also been extensively used in adults and elderly, in the treatment of gastrointestinal and respiratory diseases (Biagi et al., 2012; Malaguarnera et al., 2012; Tojo et al., 2014). In an attempt to restore the lipoprotein imbalance found in the blood of children with dyslipidemia, Guardamagna et al. (2014) examined the effects of a probiotic which contained three Bifidobacterium strains, selected due to their bile salt hydrolase activity. Their results found a decrease in total cholesterol and low-density lipoprotein cholesterol.

In terms of brain gut disorders, several studies have examined the use of bifidobacteria for their psychobiotic effects in reducing stress, anxiety and depressive like behavior in BALB/c mice. It was concluded that the behavioral effects observed may be partially explained by the differential effects on the immune system, although mechanisms underlying the effects are unknown (Savignac et al., 2014). Furthermore, a particular strain of B. longum was shown to have anxiolytic effects in a model of non-infectious colitis through vagal pathways whereby the fermentation products of this species were capable of modifying the excitation of enteric neurons in the gut (Bercik et al., 2011).

Bifidobacteria have also been studied for their ability to specifically localize at tumor sites (Nakamura et al., 2002). Cronin et al. (2010) showed in that B. breve UCC 2003 was able to target tumors in athymic MF1 nu/nu mice bearing s.c. B16-F10 murine melanoma tumors, as well as C57 mice bearing s.c. Lewis lung carcinoma tumors. Interestingly, an increasing bacterial load of B. breve UCC 2003 was found in the metastatic nodules in the lungs of the B-16 mouse model where oxygen is transferred.

#### REFERENCES


Bifidobacteria have also been shown to have an effect on tumor specific T-cell responses in C57BL/6 mice bearing s.c. B16.SIY melanoma growth (Sivan et al., 2015). In this study bifidobacteria were found to have an effect on dendritic cell function and CD8+ T cell responses, reducing tumor cell growth.

Bifidobacterium species may have positive effects on human health, while it should be stressed that the increase only in fecal bifidobacteria level cannot in itself a health benefit. However, given the strong association of bifidobacteria to health it may provide a contributory biomarker to disease status lined to some illnesses in the future. Moreover the use of bifidobacteria as delivery vehicles for the administration of therapeutic agents to target tumors due to their anaerobic nature, is a very promising and safe form of therapy for this disease state.

## CONCLUSION

Bifidobacteria are one of the most abundant genera present in the healthy infant gut and represent a significant portion of the microbiota throughout a healthy adult life, playing an important role in gut homeostasis and health. During late adulthood and within several diseases, the levels of Bifidobacterium and its species diversity decrease. Nowadays, their association with health and aging is undeniable, however, the jury is still out on whether it is a "cause" or "effect" type relationship. Given the prevalence of bifidobacteria at various stages of a healthy life and the many health promoting attributes associated with their use, it is undoubted that these bacteria play an important role in human health maintenance and protection and also may in the future provide a very important biomarker for certain diseases.

## AUTHOR CONTRIBUTIONS

RR, CS, and SA conceived the manuscript. SA and CW drafted the manuscript. CS and RR reviewed the final version of the manuscript. All the authors approved it for publication.

#### ACKNOWLEDGMENTS

This research was supported by the Science Foundation Ireland (SFI)-funded APC Microbiome Institute; by Department of Agriculture Food and Marine (DAFM)-funded INFANTMET (Ref. No. 10FDairy) and ToddlerFood (Ref. No. 14F821) projects; and by European Union's Seventh Framework Program (EUFP7)-funded MyNewGut project (Ref. No. 613979).

microbiota in preterm neonates. FEMS Microbiol. Ecol. 79, 763–772. doi: 10.1111/j.1574-6941.2011.01261.x


Annu. Rev. Food Sci. Technol. 7, 1–21. doi: 10.1146/annurev-food-041715- 033151


abundance, and community cellular fatty acid profiles. Gut 48, 198–205. doi: 10.1136/gut.48.2.198


spp. in fecal samples from breast-fed and bottle-fed infants with and without iron supplement. J. Clin. Microbiol. 25, 285–289.



**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 Arboleya, Watkins, Stanton and Ross. 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.

# Variations in the Post-weaning Human Gut Metagenome Profile As Result of Bifidobacterium Acquisition in the Western Microbiome

Matteo Soverini<sup>1</sup> , Simone Rampelli<sup>1</sup> \*, Silvia Turroni<sup>1</sup> , Stephanie L. Schnorr<sup>2</sup> , Sara Quercia<sup>1</sup> , Andrea Castagnetti<sup>1</sup> , Elena Biagi<sup>1</sup> , Patrizia Brigidi<sup>1</sup> and Marco Candela<sup>1</sup> \*

<sup>1</sup> Department of Pharmacy and Biotechnology, University of Bologna, Bologna, Italy, <sup>2</sup> Department of Anthropology, University of Oklahoma, Norman, OK, USA

#### Edited by:

Francesca Turroni, University College Cork, Ireland

#### Reviewed by:

Abelardo Margolles, Consejo Superior de Investigaciones Científicas, Spain Jeffrey David Galley, Baylor College of Medicine, USA

#### \*Correspondence:

Simone Rampelli simone.rampelli@unibo.it Marco Candela marco.candela@unibo.it

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 29 April 2016 Accepted: 23 June 2016 Published: 12 July 2016

#### Citation:

Soverini M, Rampelli S, Turroni S, Schnorr SL, Quercia S, Castagnetti A, Biagi E, Brigidi P and Candela M (2016) Variations in the Post-weaning Human Gut Metagenome Profile As Result of Bifidobacterium Acquisition in the Western Microbiome. Front. Microbiol. 7:1058. doi: 10.3389/fmicb.2016.01058 Studies of the gut microbiome variation among human populations revealed the existence of robust compositional and functional layouts matching the three subsistence strategies that describe a trajectory of changes across our recent evolutionary history: hunting and gathering, rural agriculture, and urban post-industrialized agriculture. In particular, beside the overall reduction of ecosystem diversity, the gut microbiome of Western industrial populations is typically characterized by the loss of Treponema and the acquisition of Bifidobacterium as an abundant inhabitant of the post-weaning gut microbial ecosystem. In order to advance the hypothesis about the possible adaptive nature of this exchange, here we explore specific functional attributes that correspond to the mutually exclusive presence of Treponema and Bifidobacterium using publically available gut metagenomic data from Hadza hunter-gatherers and urban industrial Italians. According to our findings, Bifidobacterium provides the enteric ecosystem with a diverse panel of saccharolytic functions, well suited to the array of gluco- and galacto-based saccharides that abound in the Western diet. On the other hand, the metagenomic functions assigned to Treponema are more predictive of a capacity to incorporate complex polysaccharides, such as those found in unrefined plant foods, which are consistently incorporated in the Hadza diet. Finally, unlike Treponema, the Bifidobacterium metagenome functions include genes that permit the establishment of microbe–host immunological cross-talk, suggesting recent co-evolutionary events between the human immune system and Bifidobacterium that are adaptive in the context of agricultural subsistence and sedentary societies.

Keywords: Bifidobacterium, Treponema, gut microbiota, microbiome, co-evolution, Westernization

# INTRODUCTION

The gut microbiota (GM) exerts a vital role in human physiology, being strategic for human nutrition, immune protection, and for the preservation of essential metabolic functions (Kau et al., 2011; Nicholson et al., 2012). This raises questions of whether, and to what degree, the human GM can provide the host with extra physiological and metabolic flexibility for the adaptation to different lifestyles and environments. Since 2010, several studies have been conducted with the

specific aim to explore GM variation across human populations with different subsistence practices, lifestyles and geographical origin. These human GM surveys revealed the existence of robust GM compositional and functional subgroups, so far generally reflective of the variations in subsistence strategy: hunter-gatherer, rural agricultural, and urban industrial Western lifestyle (Yatsunenko et al., 2012; Schnorr et al., 2014; Martínez et al., 2015; Obregon-Tito et al., 2015; Rampelli et al., 2015).

The findings from this new and emerging field of research, which combines human microbiology ecology and anthropology, resulted in two main conclusions with important implications for both human evolutionary history and human health: first, humans co-evolved with symbiont microbial ecosystems, which have co-adapted along the trajectory of subsistence change across human evolutionary history, from hunter-gatherers to rural agricultural to the most recent development of completely industrialized societies (Quercia et al., 2014; Obregon-Tito et al., 2015); second, despite the considerable variation in rural and traditional life-ways, urban industrial populations stand apart as having a distinctly altered GM profile. Indeed, the GM of urban industrial populations seems to universally share certain compositional qualities, such as: (i) an overall compression of microbial diversity as measured by phylogeny and the number of unique taxa (Segata, 2015), (ii) the loss of the so-called microorganisms "old friends", Treponema and Succinivibrio (Blaser and Falkow, 2009; Warinner et al., 2015), and (iii) the acquisition of Bifidobacterium as typical inhabitant of the adult gut (Schnorr et al., 2014; Martínez et al., 2015; Obregon-Tito et al., 2015).

In agreement with the multiple hit hypothesis (Sonnenburg and Sonnenburg, 2014), dietary changes, sanitization and antibiotic usage are all potential triggers that would explain the reduction of ecosystem diversity and the loss of co-adapted microbial communities commonly observed in the GM profile of urban industrial Western populations. Conversely, the factors that favored the correspondent acquisition of Bifidobacterium in the Western microbiome still remain to be defined. Showing a relative abundance that ranges from 3 to 10% of the total ecosystem, bifidobacteria are an abundant bacterial component in the GM of urban industrial, namely Westernized, adults, and also dominates the GM ecosystem of breast-fed infants, where this bacterial family accounts on average for 80% of the total community (Turroni et al., 2012; Bottacini et al., 2014; Milani et al., 2015b). The recent characterization of the bifidobacterial pangenome – 18,181 Bifidobacterium specific Cluster of Orthologous Genes (BifCOGs) from 47 sequenced type strains – revealed the saccharolytic functions of this microorganism, and indicated a strong adaptation to the human gut environment (Milani et al., 2015a). Furthermore, recent work has helped advance a possible hypothesis to explain the adoption of Bifidobacterium into the Western adult GM. Through comparisons of the GM of Hadza hunter-gatherers and urban industrial Italians, Schnorr et al. (2014) highlighted – for the first time to our knowledge – the substantial lack of Bifidobacterium from the GM of some traditional populations. The authors proposed that the lack of bifidobacteria in adult Hadza huntergatherers may be "a consequence of the post-weaning GM composition in the absence of agro-pastoral-derived foods", while the continued consumption of dairy into adulthood is one of the possible vectors by which many Westernized populations maintain a relatively large bifidobacterial presence. Obregon-Tito et al. (2015) and Martínez et al. (2015) reached similar conclusions through comparative research on the GM of huntergatherers from Peru and rural highlanders from Papua New Guinea, respectively, and also found a low proportion of bifidobacteria relative to individuals from the USA. However, these comparative gut metagenome surveys do not specifically explore the impact of Bifidobacterium acquisition on the functional configuration of the GM of Western adults.

Taken together, these findings indicate a scenario of exchange, whereby through Westernization, human populations have lost long-standing commensal microorganisms, in particular Treponema and Succinivibrio, but have potentially compensated through the adult acquisition of bifidobacteria. In order to examine changes to the GM as a result of these community shifts, we investigate how the loss of Treponema and the acquisition of Bifidobacterium influenced the human gut metagenome profile. To this aim, we compare gut metagenome functions assigned to Treponema and Bifidobacterium retrieved from downloadable GM metagenomic data for both Hadza hunter-gatherers and urban Italians. Our findings reveal interesting functional gains in the Western microbiome corresponding to the post-weaning retention of Bifidobacterium as a symbiont microorganism, suggesting an opportunistic yet important role of this taxon in our recent history.

# MATERIALS AND METHODS

# Sample Collection and Shotgun Sequencing

The Illumina shotgun sequences used in this study were previously generated (Rampelli et al., 2015), and are publically available at the National Center for Biotechnology Information – Sequence Read Archive (NCBI SRA; SRP056480, Bioproject ID PRJNA278393. Leipzig Ethik-kommission review board, reference number 164-12-21052012).

# Bifidobacterium and Treponema Species Identification within Italian and Hadza Metagenomes

In order to identify the Bifidobacterium and Treponema species in Italian and Hadza populations, respectively, the 16S rDNA sequences within the assembled metagenomes were taxonomically selected using the assign\_taxonomy.py script of the Qiime pipeline (Caporaso et al., 2010), against the Greengenes database<sup>1</sup> . The assignment at species level was performed by blastn of the Treponema and Bifidobacterium 16S rDNA sequences against the entire NCBI nucleotide database, in particular the top hit results for each sequence were considered.

<sup>1</sup>http://greengenes.secondgenome.com/downloads

# Characterization of the CAZyme Repertoire Assigned to Bifidobacterium and Treponema in the Gut Metagenome

Reads from a total of 38 individual GM metagenomes, 27 Hadza and 11 Italians from Rampelli et al. (2015), were downloaded and included in this study. Reads were assembled into contigs using MetaVelvet (Namiki et al., 2012) with 350 bp as insert length. Predicted open reading frames (ORFs) were determined by FragGeneScan (Rho et al., 2010) on assembled contigs, using the "–w 0" option for the fragmented genomic sequences and the parameter "–t complete". From the translated ORFs we detected the CAZymes-coding sequences using hmmscan tool from the HMMER software package (Eddy, 2011) and the dbCAN database (Yin et al., 2012). The outputs were processed by the script hmmscan-parser.sh<sup>2</sup> , selecting only the sequences that showed a minimum identity of 30% to the query sequences and an alignment length of at least 100 residues. In order to identify CAZymes derived from Bifidobacterium and Treponema, we retrieved the nucleotide sequences of the CAZymes detected with hmmscan from the FragGeneScan output, and we blasted them against the NCBI nucleotide database. Only the sequences that showed as best hit an assignment to Bifidobacterium for the Italian samples or Treponema for the Hadza samples were retained for further analysis. On the basis of the coverage of the contigs, we obtained information concerning the abundance of CAZymes. To compare the data among samples, we obtained a normalized CAZyme abundance by dividing the CAZyme coverages of every correspondent contig for the giga-bases of every correspondent sample. Heatmaps and graphs were generated in R using the packages made4 (Culhane et al., 2005) and stats<sup>3</sup> .

## Read-Mapping Approach for the Detection of Bifidobacterium and Treponema Functions Involved in the Adaptation to the Gut Environment

High quality reads for each sample were aligned to Bifidobacterium- or Treponema-assigned genes encoding bile acid adaptation, host interaction, and polysaccharide catabolism using bowtie2 and setting the alignment parameters to "–sensitive-local" (See Supplementary Table S1 for the list of genes used for this analysis). As reference for the alignment, two different databases containing orthologous genes from the NCBI genomes of the previously detected Treponema or Bifidobacterium species were created. Specifically, the databases contain genes for alpha-amylase, beta-galactosidase, mannanase, cellulase, pectinase, and xylanase. Furthermore, the databases were implemented with the sequences of the bile efflux pump, bile salt hydrolase, exopolysaccharide synthase, fimbrial subunit FimQ, sortase, galactosyl transferase, and undecaprenylphosphate phosphotransferase, since they were recently reported as genes that facilitate commensal-host cross-talk in Bifidobacterium (Gueimonde et al., 2009; Fanning et al., 2012; Turroni et al., 2013; Ferrario et al., 2016). In the event that the Bifidobacterium or Treponema NCBI genomes did not contain the above-mentioned genes, we supplemented the databases with genes belonging to the taxonomically closest annotated microorganism. The reads that aligned with a reference using bowtie2, were extracted and their taxonomy was further verified by blastn against the entire NCBI nucleotide database. Notably, in the case that the best hits of the blastn search were not assigned to Bifidobacterium or Treponema, we did not consider those reads for further analysis. The number of hits for each gene was normalized by the number of base pairs in the input file and in the correspondent reference in order to compare the results.

# RESULTS

## Variation in the Gut Microbiome Carbohydrate-Degrading Repertoire as a Result of the Exclusive Presence of Bifidobacterium or Treponema in the Italian and Hadza Gut Microbial Ecosystems

By studying hunter-gatherers and rural populations, we previously depicted an approximation of the co-evolutional trajectory of the human GM structure in the recent times (Schnorr et al., 2014). In particular, we characterized the taxonomic and functional metabolic potential of the GM in the Hadza hunter-gatherers and urban Italian adults, reporting considerable differences between the two populations that map on to their respective life-ways (Schnorr et al., 2014; Rampelli et al., 2015). Among them, we observed a loss of Treponema and an acquisition of Bifidobacterium in the Italian GM. Considering the importance of Bifidobacterium as a health-promoting commensal of the GM in Western populations, it is important to understand how this beneficial role may have developed through the lens of co-evolution.

We first identified the diversity of the Bifidobacterium and Treponema species in urban Italian and Hadza GM by reconstructing the full 16S rDNA gene from assembled metagenomes. Italian samples contain sequences belonging to Bifidobacterium faecale, Bifidobacterium pseudocatenulatum, Bifidobacterium adolescentis, Bifidobacterium coryneforme, Bifidobacterium bifidum, Bifidobacterium longum, Bifidobacterium angulatum, and Bifidobacterium dentium. On the other hand, we detect the 16S rDNA sequences assigned to Treponema porcinum, Treponema bryantii, Treponema succinifaciens, Treponema parvum, and Treponema berlinense in the GM of the Hadza hunter-gatherers. However, we must acknowledge that these taxonomic assignments are limited by present whole genome for Treponema species, most of which have been characterized by work on human pathogens, rather than commensal members of the GM. Ongoing work with specieslevel variation should be crucial to resolving the disparities in our classification and resolution of these species and their functions in the future (Obregon-Tito et al., 2015). In order

<sup>2</sup>https://github.com/carden24/Bioinformatics\_scripts/blob/master/hmmscanparser.sh

<sup>3</sup>https://stat.ethz.ch/R-manual/R-devel/library/stats/html/00Index.html

to compare the specific saccharolytic functions conferred by Bifidobacterium and Treponema in the Italian and Hadza microbiomes, we identified a total of 5.4 million of ORFs, of which 14,512 mapped to CAZymes for the Italian samples and 74,651 for the Hadza samples (**Figure 1A**). Notably, the Hadza metagenomes contain significantly more CAZymes per subject, in terms of ORFs assigned to CAZymes per million of reads, respect to the Italian metagenomes (mean ± SD, Hadza: 233 ± 86, Italians: 137 ± 78), as reported by Rampelli et al. (2015). We then profiled the saccharolytic repertoire of Bifidobacterium and Treponema in the Italian and Hadza GM as relative abundance at the CAZyme category level based on the coverage of taxonomically assigned contigs (**Figure 2A**). Bifidobacterium showed a higher presence of glycosyl transferase (GT) and carbohydrate esterase (CE), with respect to Treponema (GT relative abundance, rel. ab.: 54% for Bifidobacterium and 43% for Treponema; CE rel. ab.: 6% for Bifidobacterium and 5% for Treponema). On the other hand, Treponema were more enriched in glycoside hydrolase (GH) and carbohydrate binding module (CBM) (GH rel. ab.: 33% for Treponema and 31% for Bifidobacterium; CBM rel. ab.: 15% for Treponema and 8% for Bifidobacterium). At the CAZyme family level, we revealed the four families that constitute the core Bifidobacterium CAZyme repertoire: GH13 (GH family acting on substrates containing α-glycosidic linkages), GH3 (GH family that groups together exo-acting β-D-glucosidases, α-L-arabinofuranosidase, β-Dxylopyranosidase and N-acetyl-β-D-glucosaminidase), GT2 (GT family containing cellulose synthase, mannan synthase, and several monosaccharide-/oligosaccharide-transferases), and GT4 (GT family containing sucrose synthase, glucosyl transferase, and several phosphorylases). The sum of ORFs assigned to these four major families comprises 77% of the total detected CAZyme cohort. The relative abundance of the Bifidobacterium and Treponema CAZyme families detected in the Italian and Hadza samples reveals several differences in the potential saccharolytic functional contributions of these two microorganisms (**Figure 2B**). In particular, Bifidobacterium have a greater abundance of genes involved in the degradation of lactate, which is produced from pyruvate in the fermentation of simple sugar and commonly found in sour milk as well as in other lacto-fermented foods (family GH2). This finding also corresponds to the exclusive contribution of dairy carbohydrates (∼5% of total carbohydrates), in the Italian cohort diet (Supplementary Table S2). Emphasis on monosaccharide catabolism is evidenced by enrichment in gene families containing enzymes that metabolize mannose, xylose and arabinose (GH2, GH31, and GH43), which are highly represented in plant and fruit glycans. As a variety of genes were also found that are involved in the degradation of α- and β-glucans (GH3 and GH31), this illustrates an ability of Bifidobacterium to retrieve energy also from more complex carbohydrates that are commonly present in the cellulosic biomass of plant foods in the Italian diet: salads, fruits, nuts, cereals and their product derivatives. In addition, Bifidobacterium are also enriched in genes involved in the catabolism of sucrose (GH31), which is widely distributed in nature, but robustly manifest in the industrial food products that are consumed daily by most

urban populations. Further evidence of these functions comes from detection of a higher abundance of CBM families for lactose, galactose and β-glucans (CBM4, CBM13, and CBM32) in Bifidobacterium, with respect to Treponema. In contrast, the CAZyme profile of Treponema within the Hadza metagenome is mainly devoted to degradation of glucans, galactans, and fructans (GH16, GH32, and GH53), which are sugar polymers that comprise hemicellulose (galactans) and inulin (fructans). The monosaccharide of galactans, galactose, is also expressed in mucilages and glycoproteins that derive from the human host, as well as a number of vegetable-derived carbohydrates. Both sugar polymers are largely implicit in difficult-to-digest plant polysaccharides that escape small intestine absorption and are instead fermented by the colonic microbiota. The Hadza diet is replete with such unrefined plant foods that contain indigestible polysaccharides such as berries, baobab fruit, and particularly tubers (Schnorr et al., 2015). Treponema are also enriched in two CAZyme α-amylase families (GH57 and GH77), which are unlike the typical α-amylase GH13 family because they have a conserved trans-glycosylating region (MacGregor et al., 2001). Finally, Treponema are better equipped to metabolize peptidoglycans due to a wide range of acetylglucosaminases and peptidoglycan lyases (GH23, GH73, and GH109). These activities were confirmed by the detection of high levels of CBM families for peptidoglycans and α-glucans (CBM50 and CBM48). To specifically investigate the effective differences in polysaccharide-degrading repertoires between Bifidobacterium and Treponema metagenome functions, we aligned the reads of Italian and Hadza samples to a custom database containing the sequences for alpha-amylase, beta-galactosidase, mannanase, cellulase, pectinase, and xylanase. The taxonomic specificity of the aligned reads was verified by blasting the sequences to the nucleotide database of NCBI (**Figure 1B**). Unlike Treponema, which appears to be functionally equipped to derive energy from a broad spectrum of polysaccharides through increased relative representation of pectinase, xylanase, and cellulase, the polysaccharide-degrading functions assigned to Bifidobacterium are enriched in genes coding for β-galactosidase and mannanase (**Figure 3**). Interestingly, both genera are configured to hydrolyze 1–4 α-glycosidic bonds (via α-amylase) to a similar degree. The catabolic configurations of Bifidobacterium and Treponema for polysaccharides are reflective of differences in dietary carbohydrates of Italians and Hadza: pasta, legumes, milk, and dairy products, versus berries, tubers, baobab, and honey, respectively, (Supplementary Table S3).

# Differences in the Gut Metagenome Functions Involved in Bacteria/Host Interaction and Gut Adaptation Corresponding to the Presence of Bifidobacterium or Treponema in the Italian and Hadza Gut Metagenome

We further investigated the presence of the genes involved in host interaction and immune modulation in the ORFs attributed to Bifidobacterium and Treponema. In this scenario,

the sortase-dependent pili are considered key surface molecules in establishing bacterial adherence to the host epithelium and they have been proposed as potential mediator of mucosal immune response (Turroni et al., 2013). Another recognized marker of interaction with the host that affects the immune response is the surface-exopolysaccharide (EPS). Specifically, bacteria producing EPS failed to elicit a strong immune B-cell response compared to EPS-deficient strains (Fanning et al., 2012). Our analyses confirm that the enzymes involved in the production of EPS and pili, namely EPS synthase, undecaprenylphosphate phosphotransferase, galactosyl transferase, sortase, and fimbrial subunit FimQ, are typical of the Bifidobacterium

taxonomy.

ORFs detected in the Italian metagenome, while virtually absent in the Treponema ORFs retrieved from the Hadza GM ecosystem (**Figure 3**). Finally, we investigated the presence of genes involved in bile tolerance as mechanisms of bacterial adaptation to the human host. Bile salts are detergent-like compounds with strong antimicrobial activity (Begley et al., 2005), and intestinal bacteria have had to evolve strategies to tolerate physiological concentrations of bile salts to colonize the intestine (Gueimonde et al., 2009). Interestingly, two representative enzymes, which contribute to bile resistance and adaptation to gut environment, the bile-inducible efflux transporters and the bile salt hydrolase, are present in within the Bifidobacterium gut metagenome functions, but are not detected in Treponema ORFs (**Figure 3**).

#### DISCUSSION

In our work we explore how the apparent dichotomy between two instance-specific mutually exclusive gut inhabitants, Bifidobacterium and Treponema, reflects specific functional roles within the human GM ecosystem. We observe that the complete replacement of these microorganisms within the human gut in modern populations follows a curious pattern of lifestyle-associated differences, which stand at opposite ends of an entire subsistence spectrum, from hunting and gathering to post-industrial urban life. This may hint at a more ancient pattern of transition that occurred along more recent human evolutionary history as early settlements permitted the adoption of a fully agro-pastoral regime in place of mobile foraging. As seen in modern rural farmers, these earlier human groups may have harbored both communities of bifidobacteria and treponemes, suggesting an intermediary phase of co-habitation (De Filippo et al., 2010; Obregon-Tito et al., 2015). These patterns are instructive, and we can hypothesize that the stimulus for bifidobacterial acquisition and proliferation in the human GM followed much earlier changes to diet and lifestyle that occurred during the Neolithic transition. Specifically, these changes include permanent settlements, dense population structures, animal co-habitation, plant domestication, and shifts in the type and amount of carbohydrate-based foods (Childe, 1936; Binford, 1968).

While bifidobacteria are absent from the GM ecosystem of modern hunter-gatherers, and thus possibly acquired as a postweaning symbionts among communities who adopted an agropastoral subsistence, Treponema has been lost from the human GM along the transition from small-scale rural communities to post-industrial urbanized society. This raises key questions such as 'what are the functional gains and losses in the human GM

corresponding to this process?', and 'does an adaptive nature of these changes exist?' To begin answering these questions, we compared metagenome functions assigned to Bifidobacterium and Treponema in the gut metagenome samples from urban Italians and Hadza hunter-gatherers, respectively. In particular, we focused on functional features key in the context of the GM-human host mutualism, such as the metabolism of complex polysaccharides, which is essential to provide the host with shortchain fatty acids (SCFA), and the microbe–host interaction and immunomodulation processes.

As an initial approach, we described the Bifidobacterium and Treponema diversity in the urban Italian and Hadza GM. Eight different bifidobacterial species were detected in the Italian GM ecosystem, including the plant polysaccharides degrader B. adolescentis, and the milk fermenters B. bifidum and B. longum. Other bifidobacterial species detected in the Italian GM were B. faecale, B. pseudocatenulatum, B. coryneforme, B. angulatum, and B. dentium, highlighting considerable bifidobacterial diversity within the Italian GM ecosystem. Given our current limitation in resolving novel Treponema taxa, we were still able to identify five species: T. porcinum, T. bryantii, T. succinifaciens, T. parvum, and T. berlinense. These species were already described in association with the GM of Matses hunter-gatherers from Peru (Obregon-Tito et al., 2015).

According to our findings, Bifidobacterium and Treponema provide the host metagenome with a different configuration of CAZyme categories. Indeed, the Bifidobacterium genes encoding for CAZymes in the Italian gut metagenome are enriched in GT and CE, while the Treponema saccharolytic repertoire in the Hadza gut is principally devoted to GH and CBM functions, suggesting the higher propensity of the latter to act as a primary degrader of complex polysaccharides in the gut. Focusing on the CAZyme family level, we observed that Bifidobacterium provides the host metagenome with an extremely versatile panel of saccharolytic functions. Bifidobacterial CAZyme families enriched in the Italian samples range from the metabolism of simple sugars, such as lactate, mannose, xylose, and arabinose, to more complex carbohydrates from plant sources, such as α- and β-glucans, lactose, or galactose. These findings are in general agreement with the overall structure of the bifidobacterial glycobiome (Milani et al., 2015a). According to Milani et al. (2015a), when comparing the bifidobacterial GH repertoire with the sequenced members of the human GM, Bifidobacterium appears enriched in GH3

and GH43 families for the degradation of plant carbohydrates. These findings are in agreement with our observations that the Bifidobacterium-assigned saccharolytic functions emphasize substrate-specific specializations that are not observed among Treponema-assigned genes. Instead, the saccharolytic functions supplied by Treponema to the Hadza gut metagenome are mainly devoted to the degradation of indigestible polysaccharides, such as galactans, fructans, and glucans, that are present in unrefined and wild plant foods. Other functional peculiarities characteristic of Treponema include a completely unique suite of α-amylases and a greater capacity to bind and metabolize peptidoglycan. When we specifically explored gut metagenome differences in polysaccharide-degrading genes, we confirmed the overall greater metabolic polysaccharolytic potential for Treponema while Bifidobacterium was enriched in beta-galactosidase and mannase.

Interestingly, the specific glycobiomes conferred by Bifidobacterium and Treponema match the respective dietary habits of the two populations. Hadza consume a heavily plantbased diet, particularly in the rainy season, in which ∼70% of kcal are derived from fibrous wild plant foods and the 30% from wild game meat (Marlowe, 2010). Furthermore, Hadza entirely lack dairy in their diet. Correspondingly, the Treponema-assigned glycobiome in the Hadza is devoted to the metabolism of the vast array of refractory glucans, galactans, and fructans that are present in the wild Hadza plant foods. The Mediterranean diet, characteristic for the Italian cohort, is abundant in plant foods (salads, fruits, sauces), pasta, bread and olive oil with low to moderate inclusion of dairy, poultry, fish and read meat (Supplementary Table S3). Curiously, certain glycobiome functions conferred by bifidobacteria to the Italian GM – e.g., GH2, GH31, and GH43, as well as CBM4, CBM13, and CBM32 – are well suited to deal with dairy carbohydrates and some types of plant glycans, all of which are relatively abundant in the Italian diet.

When we looked for functions associated with host–microbe interactions, we found matches only among Bifidobacterium. Conversely, Treponema does not provide the Hadza metagenome with known functionalities for direct host interaction. The Bifidobacterium functional potential encoded within strains retrieved from the Italian samples included enzymes for the biosynthesis of pili and EPS structures. Both cell components are essential in the context of a mutualistic strategy that permits close interaction and initiation of a tolerated immunological dialog between the bacterium and host (Fanning et al., 2012). Finally, as a functional marker of the bacterial adaptation to the gut environment (Begley et al., 2005), we sought functions involved in bile salts tolerance. Interestingly, only bifidobacteria seem to possess bile-inducible efflux transporters

#### REFERENCES

Begley, M., Gahan, C. G., and Hill, C. (2005). The interaction between bacteria and bile. FEMS Microbiol. Rev. 29, 625–651. doi: 10.1016/j.femsre.2004.09.003

and the bile salt hydrolase required for bile salt mitigation and detoxification. Even if biases in the assignment of Treponema functions as a result of the paucity of reference genomes cannot be excluded, our findings could indicate different ecological strategies for Bifidobacterium and Treponema in the human gut.

# CONCLUSION

Our findings suggest possible co-evolutionary implications for the loss of Treponema and the acquisition of Bifidobacterium as a stable component of the post-weaning GM ecosystem from post-industrial urban populations. Capable of heterogeneous saccharolytic metabolism, which ranges from complex plant polysaccharides to simpler sugars such as lactose and sucrose, Bifidobacterium are well suited to handle the degradative demands imposed by a typical Western diet. Conversely, the progressive loss of more challenging microbiota accessible carbohydrates in the Western diet (Sonnenburg et al., 2016), such as hemicellulose and inulin, would help partially explain the extinction of a more specialized fiber degrader such as Treponema from the Western GM ecosystem. Furthermore, unlike Treponema, Bifidobacterium evolved the capacity to establish an intense microbe–host connection, which may help support a continuous and abundant bifidobacterial presence in adults, allowing this commensal to outcompete other opportunistic, but functionally diverse, microbiota. The acquisition of Bifidobacterium as a stable component of the GM ecosystem in small-scale rural agriculturalists, reminiscent of early human farmers, and modern Westernized populations, may therefore engage the functionalities of the host immune system, providing new adaptive solutions in response to changing selective pressures during the restructuring of human diet and society.

# AUTHOR CONTRIBUTIONS

SR and MC conceived the study with the participation of MS in experimental design. MS and SR performed the bioinformatics analysis. MS, SR, SLS, ST, and MC carried out data interpretation and wrote the paper. SQ, AC, EB, and PB participated to the discussion of the results and in revising the final draft of the paper. All authors read and approved the final manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.01058

Binford, L. R. (1968). "Post pleistocene adaptations," in New Perspectives in Archaeology, eds S. Binford and L. R. Binford (Chicago, IL: Adeline), 313–342.

Blaser, M. J., and Falkow, S. (2009). What are the consequences of the disappearing human microbiota? Nat. Rev. Microbiol. 7, 887–894. doi: 10.1038/nrmicro2245


**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 Soverini, Rampelli, Turroni, Schnorr, Quercia, Castagnetti, Biagi, Brigidi and Candela. 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.

# Why Don't All Infants Have Bifidobacteria in Their Stool?

Gerald W. Tannock 1, 2 \*, Pheng Soon Lee3, 4, Khai Hong Wong<sup>3</sup> and Blair Lawley <sup>1</sup>

*<sup>1</sup> Department of Microbiology and Immunology, University of Otago, Dunedin, New Zealand, <sup>2</sup> Riddet Centre for Research Excellence, Massey University, Palmerston North, New Zealand, <sup>3</sup> Mead Johnson Nutrition, Singapore, Singapore, <sup>4</sup> Department of Human Nutrition, University of Otago, Dunedin, New Zealand*

Keywords: bifidobacteria, infants, fecal microbiota, Lachnospiraceae, child development

Members of the genus Bifidobacterium are abundant in the stool of most human infants during the initial exclusively milk-fed period of life, especially at an age of 2–3 months (Harmsen et al., 2000; Favier et al., 2002; Mariat et al., 2009; Coppa et al., 2011; Turroni et al., 2012; Yatsunenko et al., 2012; Tannock et al., 2013; Barrett et al., 2015). Bifidobacteria dominate the stool microbiota regardless of whether the infants are fed human milk or formula based on ruminant milk (cow or goat). However, bifidobacteria have about 20% higher relative abundances in human milk-fed compared to formula-fed babies (Tannock et al., 2013). The greater abundance of bifidobacteria in human-milk-fed infants can, at least in part, be explained by the fact that bifidobacterial species that are enriched in the infant bowel can utilize Human Milk Oligosaccharides (HMO) or their components as growth substrates (Sela et al., 2008; LoCascio et al., 2010; Garrido et al., 2013). It could be anticipated, therefore, that bifidobacteria would be detectable in the stool microbiota of every child nourished at the breast because of the supply of appropriate growth substrates. This expectation is not borne out completely because a proportion of infants have very low abundance or undetectable bifidobacteria as members of the fecal microbiota regardless of breast milk or formula feeding (Young et al., 2004; Gore et al., 2008; Tannock et al., 2013). Antibiotics had not

#### Edited by:

*Francesca Turroni, University College Cork, Ireland*

#### Reviewed by:

*Sabrina Duranti, University of Parma, Italy Francesca Bottacini, University College Cork, Ireland*

\*Correspondence: *Gerald W. Tannock gerald.tannock@otago.ac.nz*

#### Specialty section:

*This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology*

Received: *24 February 2016* Accepted: *17 May 2016* Published: *31 May 2016*

#### Citation:

*Tannock GW, Lee PS, Wong KH and Lawley B (2016) Why Don't All Infants Have Bifidobacteria in Their Stool? Front. Microbiol. 7:834. doi: 10.3389/fmicb.2016.00834* been administered to these infants. How then can the absence of bifidobacteria be explained?

# A GROWTH SUBSTRATE DEFICIT?

The "bifidobacteria-negative" babies have been detected in both human milk and formula-fed infants. Therefore, a bacterial growth substrate effect seems unlikely. While human milk is rich in HMO and ruminant milk lacks these complex molecules (although simpler forms such as sialylated-lactose are present in very small amounts), bifidobacteria are still the most abundant taxon in the feces of infants fed formula un-supplemented with galacto- or fructo-oligosaccharides (Tannock et al., 2013). In this case, lactose and/or glycoproteins and glycolipids are probable growth substrates for bifidobacteria (Turroni et al., 2010; Bottacini et al., 2014; O'Callaghan et al., 2015) in the bowel of exclusively milk fed infants. There is, however, a need to support genomic analysis of bifidobacteria with culture-based investigations of bifidobacterial nutrition based on substrates present in the bowel of exclusively milk-fed babies (other than HMO).

# LACK OF SENSITIVITY OF BIFIDOBACTERIAL DETECTION METHODS?

An obvious reason for bifidobacteria-negative feces is that the detection methods lack sufficient sensitivity. Culture-based methods usually have a lower detection limit of 1 × 10<sup>3</sup> per gram, fluorescent in situ hybridization (FISH) 1 × 10<sup>6</sup> − 10<sup>7</sup> per gram (manual or digital counts respectively) or ∼4 × 10<sup>4</sup> by flow cytometry, and denaturing gradient gel electrophoresis of PCR amplicons ∼1 × 10<sup>5</sup> – 10<sup>6</sup> cells (Welling et al., 1997; Jansen et al., 1999; Zoetendal et al., 2001, 2002) or 1 × 10<sup>4</sup> using internal transcribed spacer targets (Milani et al., 2014). High throughput DNA sequencing methods, such as Illumina, generate tens of thousands of 16S rRNA gene sequences per DNA sample, but there may be several hundred OTU per sample. Thus, taxa present in very low abundance could be missed. However, reference to rarefaction curves (alpha diversity) during sequence analysis will show whether coverage of the microbiota is near complete or not. Therefore, while lack of sufficient sensitivity of detection methods remains a possibility, it probably does not provide the total explanation.

#### BIFIDOBACTERIAL POPULATIONS RISE AND FALL FROM DAY TO DAY?

Most fecal microbiota studies examine a single fecal sample from each participating individual. Comprehensive temporal studies of the fecal microbiota to determine day-to-day variations in composition have not been reported. It is possible that bifidobacteria are present in the feces of all children during early life but that, on some days, the bifidobacterial population falls to undetectable levels. Populations of bifidobacteria in the feces of some adults without diseases are dynamic in terms of strain composition, so there is some support for a concept of temporal instability in the bifidobacterial population of the microbiota (McCartney et al., 1996). **Figure 1A** shows data from feces collected at intervals from infants during the first 12 weeks of life. In the example, fluctuations in the abundances of bifidobacteria were seen, varying from very low abundance to absence, in feces of individual children. Strikingly, bifidobacteria were not detected in any of the fecal samples of one child. Therefore, bifidobacteria-free infants do seem to be a real phenomenon.

# THE WINDOW OF INFECTIVITY (OPPORTUNITY/COLONIZATION) WAS MISSED?

A window of opportunity is a short time period during which an otherwise unattainable opportunity exists. After the window of opportunity closes, the opportunity ceases to exist. Caufield was the first to describe the "window of infectivity" in the acquisition of commensal bacteria. His example was Streptococcus mutans in the oral cavity of children (Caufield et al., 1993; Li and Caufield, 1995). This bacterial species is associated with dental plaque, thus the window of infectivity coincided with the eruption of the first molars. Prior to this, a habitat for S. mutans is not available in the oral cavity of children for this species. The Caufield hypothesis reminds us that many factors have to coincide to favor the establishment of a commensal in a body site. Cesareandelivered babies have lower prevalences of bifidobacteria in their feces in early life (**Figure 1B**). By analogy to Caufield's studies, this probably relates to a lack of favorable opportunities for bifidobacteria to colonize the bowel relative to the vaginal birth process. Notably, we found that 36% of cesarean-derived babies lacked bifidobacteria, whereas 18% of vaginally delivered infants were bifidobacteria-free at 2 months of age (Tannock et al., 2013).

#### OTHER TAXA REPLACE BIFIDOBACTERIA IN SOME BABIES?

If bifidobacteria have not colonized the bowel of certain infants, they are likely to be replaced by other taxa, which may have the requisite metabolic properties to fill the vacant ecological niche. In a study of the fecal microbiotas of Australian babies that were breast milk- or formula-fed, we compared the relative abundances of bacterial taxa in infants that had very low (<10%) or higher (>10%) bifidobacterial content (Tannock et al., 2013). Analysis of the compositions of these microbiotas showed that when Bifidobacteriaceae abundance was low, Lachnospiraceae abundances tended to be greater in babies in all dietary groups (**Figures 1C–E**). There was also a tendency for Erysipelotrichaceae abundances to be greater in formula-fed babies with low bifidobacterial abundances, being much more evident in the case of goat milk-fed infants. These observations suggest that, yes, other taxa might replace bifidobacteria in the fecal microbiota of some children.

#### WHAT ARE THE CONSEQUENCES OF LACKING BIFIDOBACTERIA IN THE BOWEL?

The absence of bifidobacteria in the bowel may be detrimental for infant development. The curious phenomenon whereby mother's milk contains substances not used in the nutrition of the offspring, but which fertilize bifidobacterial growth, is unique to humans. There must be a good reason for this. Enriching bifidobacterial populations in the bowel tends to minimize the abundance of other bacterial species, so a competitive exclusion function could be ascribed to HMO. Additionally, HMO may act as "decoys" in the bowel by binding to pathogens (bacteria and viruses) and their toxins and thus limiting contact with mucosal surfaces (Kunz et al., 2000). The large diversity of HMO structures that is known to occur in human milk suggests a large diversity of decoy functions (Pacheco et al., 2015). Irrespective of where in the World babies live, their gut microbiomes are enriched in genes involved in the de novo biosynthesis of folate (Yatsunenko et al., 2012). In contrast, the microbiome of adults favors synthesis of another B vitamin, cobalamin. Folate synthesis is an attribute of bifidobacteria and folate can be absorbed from the large bowel, so enrichment of bifidobacteria in the infant bowel may provide an important contribution to infant nutrition (Aufreiter et al., 2009; D'Aimmo et al., 2012; Lakoff et al., 2014). Folate functions as a coenzyme or co-substrate in single-carbon transfers in the synthesis of nucleic acids and metabolism of amino acids. One of the most important folate-dependent reactions is the conversion of homocysteine to methionine in the synthesis of S-adenosyl-methionine, an important methyl donor. Another folate-dependent reaction, the methylation of deoxyuridylate to thymidylate in the formation of DNA, is required for proper cell division (Crider et al., 2012).

undetectable by DNA sequencing of 16S rRNA gene amplicons at 6 and 12 weeks in the feces of AF067; 6, 8, 10, and 12 weeks in AF70; 8 and 10 weeks in AF075; and 6 and 12 weeks in AF090. (B) Comparison of bifidobacerial abundances in the feces of infants delivered vaginally or by cesarean. Note that in both groups, bifidobacteria were not detected in some infants. (C) Comparison of abundances of bacterial families in microbiotas of breast milk-fed infants in relation to abundances of bifidobacteria. (D) Comparison of abundances of bacterial families in microbiotas of cow milk formula-fed infants in relation to abundances of bifidobacteria. (E) Comparison of abundances of bacterial families in microbiotas of goat milk formula-fed infants in relation to abundances of bifidobacteria. Note that Lachnospiraceae have increased abundances when bifidobacteria have low relative abundance. Figures after Tannock et al. (2013), reproduced with permission.

Neonatal nutrition could, indeed, be the very important reason for the HMO-bifidobacteria-infant paradigm. The foundation of brain structure and function is set early in life through genetic, biological and psychosocial influences. The rate of neonatal brain growth exceeds that of any other organ or body tissue (Wang, 2012). The infant is born with neurons already formed but the synaptic connections between these cells are mostly established and elaborated after birth causing a large nutritional demand for biosynthesis of gangliosides (Svennerholm et al., 1989). Nutrition of the infant in early life affects brain developmental processes including cognition (Uauy and Peirano, 1999; Uauy et al., 2001). While long-chain fatty acids (such as docosahexaenoic acid) have been the focus of much of the research in this field, tantalizing research evidence now indicates that sialic acid (N-acetyl-neuraminic acid), a 9-carbon carbohydrate, is also an essential nutrient for optimal brain development and cognition (Gibson, 1999; Meldrum et al., 2012; Wang, 2012). Strikingly, cortical tissue from human brain contains up to 4 times more sialic acid than that of other mammals tested (Wang et al., 1998). Moreover, the sialic acid concentration in the brain of breast milk-fed babies is higher than in that of formula-fed infants (Wang et al., 2003). These facts correlate with the unique biochemistry of human milk and the unique bacteriology of the infant bowel. Intriguingly, Ruhaak et al. (2014) have reported the detection of sialylated oligosaccharides (3′ sialyl-lactose, 6′ sialyllactose, 3′ sialyl-lactosamine, 6′ sialyl-lactosamine) that might result from the hydrolysis of HMO, in the blood of human infants. Thus, bifidobacterial biochemistry in the bowel may have extra-intestinal, nutritional influences important in brain development. However, perhaps the taxa that are abundant in the bowel of infants in the absence of bifidobacteria can carry out these same functions? This interesting possibility remains to be investigated.

#### BABIES WITHOUT BIFIDOBACTERIA ARE IMPORTANT SOURCES OF KNOWLEDGE?

Rene Dubos explored in a number of books the interplay between environmental forces and the physical, mental, and spiritual development of humankind. His article published in the journal

#### REFERENCES


Pediatrics entitled "Biological Freudianism: lasting effects of early environmental influences" encapsulated this theme (Dubos et al., 1966). Drawing on the results of experiments conducted with specific-pathogen-free mice, the authors concluded that "From all points of view, the child is truly the father of the man, and for this reason we need to develop an experimental science that might be called biological Freudianism. Socially and individually the response of human beings to the conditions of the present is always conditioned by the biological remembrance of things past."

Biological Freudianism is clearly of relevance to the concept that the first 1000 days, between conception and the child's second birthday, offer a unique window of opportunity to shape healthier and more prosperous futures. Nutrition during this 1000 day window can have a profound impact on a child's ability to grow, and learn. The influences of the microbiota on the development of the child during early life are potentially very important, and much longitudinal research is required to clarify whether there are continuing, medically important impacts of the microbiota, inlcuding the bifidobacteria, that last throughout the lifetime of humans. Comparisons of the cognitive development and general health status of children that had been bifidobacteria-free, and children that were ex-bifidobacteria-free then intentionally exposed to bifidobacteria, in a longitudinal study extending perhaps 10 or 20 years, would tell us whether these bacteria optimize short and/or long term human development and health.

#### AUTHOR CONTRIBUTIONS

GT wrote the article. BL, PL, and KW provided data described in the article.


pathways for host-derived glycan foraging. Proc. Natl. Acad. Sci. U.S.A. 107, 19514–19519. doi: 10.1073/pnas.1011100107


**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 Tannock, Lee, Wong and Lawley. 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.

# Bifidobacteria and Butyrate-Producing Colon Bacteria: Importance and Strategies for Their Stimulation in the Human Gut

#### Audrey Rivière, Marija Selak, David Lantin, Frédéric Leroy and Luc De Vuyst\*

Research Group of Industrial Microbiology and Food Biotechnology, Faculty of Sciences and Bioengineering Sciences, Vrije Universiteit Brussel, Brussels, Belgium

With the increasing amount of evidence linking certain disorders of the human body to a disturbed gut microbiota, there is a growing interest for compounds that positively influence its composition and activity through diet. Besides the consumption of probiotics to stimulate favorable bacterial communities in the human gastrointestinal tract, prebiotics such as inulin-type fructans (ITF) and arabinoxylan-oligosaccharides (AXOS) can be consumed to increase the number of bifidobacteria in the colon. Several functions have been attributed to bifidobacteria, encompassing degradation of non-digestible carbohydrates, protection against pathogens, production of vitamin B, antioxidants, and conjugated linoleic acids, and stimulation of the immune system. During life, the numbers of bifidobacteria decrease from up to 90% of the total colon microbiota in vaginally delivered breast-fed infants to <5% in the colon of adults and they decrease even more in that of elderly as well as in patients with certain disorders such as antibiotic-associated diarrhea, inflammatory bowel disease, irritable bowel syndrome, obesity, allergies, and regressive autism. It has been suggested that the bifidogenic effects of ITF and AXOS are the result of strain-specific yet complementary carbohydrate degradation mechanisms within cooperating bifidobacterial consortia. Except for a bifidogenic effect, ITF and AXOS also have shown to cause a butyrogenic effect in the human colon, i.e., an enhancement of colon butyrate production. Butyrate is an essential metabolite in the human colon, as it is the preferred energy source for the colon epithelial cells, contributes to the maintenance of the gut barrier functions, and has immunomodulatory and anti-inflammatory properties. It has been shown that the butyrogenic effects of ITF and AXOS are the result of cross-feeding interactions between bifidobacteria and butyrate-producing colon bacteria, such as Faecalibacterium prausnitzii (clostridial cluster IV) and Anaerostipes, Eubacterium, and Roseburia species (clostridial cluster XIVa). These kinds of interactions possibly favor the co-existence of bifidobacterial strains with other bifidobacteria and with butyrate-producing colon bacteria in the human colon.

Keywords: bifidobacteria, butyrate-producing colon bacteria, cross-feeding, prebiotics, probiotics, arabinoxylan-oligosaccharides, inulin-type fructans

Edited by:

Marco Ventura, University of Parma, Italy

#### Reviewed by:

Sinead Christine Leahy, AgResearch Limited, New Zealand Christian U. Riedel, University of Ulm, Germany

> \*Correspondence: Luc De Vuyst ldvuyst@vub.ac.be

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 29 April 2016 Accepted: 07 June 2016 Published: 28 June 2016

#### Citation:

Rivière A, Selak M, Lantin D, Leroy F and De Vuyst L (2016) Bifidobacteria and Butyrate-Producing Colon Bacteria: Importance and Strategies for Their Stimulation in the Human Gut. Front. Microbiol. 7:979. doi: 10.3389/fmicb.2016.00979

# INTRODUCTION

Whereas, the human gut microbiota has been studied in the past mainly in the context of infectious diseases, it is known today that this enormous number of microorganisms has an indispensable role in the normal development and functioning of the human body (O'Hara and Shanahan, 2006; Sommer and Bäckhed, 2013). Within the adult gastrointestinal tract, the colon contains the most dense (>10<sup>11</sup> bacteria per mL of luminal content) and metabolically active microbiota (**Figure 1**; Whitman et al., 1998; The Human Microbiome Project Consortium, 2012). The immense number of genes (>100 times the number of genes of the human genome) encoded by this microbiota, expands the host's biochemical and metabolic capabilities substantially (Bäckhed et al., 2005; The Human Microbiome Project Consortium, 2012). Examples of supporting functions of the human gut microbiota are the degradation of otherwise non-digestible food compounds; the transformation of toxic compounds; and the production of essential vitamins, important metabolic end-products, and defending bacteriocins (Sommer and Bäckhed, 2013). Microbial metabolic end-products, which account for one third of the metabolites present in the human blood, play an important role in gut homeostasis and have an impact on host metabolism and health (Wikoff et al., 2009; Hood, 2012; Louis et al., 2014; Sharon et al., 2014; Richards et al., 2016). The short-chain fatty acids (SCFAs) acetate, butyrate, and propionate (typically occurring in a 3:1:1 ratio) are quantitatively (total concentration of 50– 150 mM) and metabolically the most important microbial endproducts of the human colon fermentation process (Louis et al., 2014), as they display several physiological effects (**Table 1**).

Changes in the gut microbiota composition have been associated with disturbed gut barrier functions, increased gut permeability, and increased plasma lipopolysaccharide concentrations (i.e., metabolic endotoxemia), which causes lowgrade inflammation that triggers the development of obesity and metabolic syndrome (Cani et al., 2008). Also other disorders, such as inflammatory bowel disease (IBD, encompassing Crohn's disease and ulcerative colitis), irritable bowel syndrome (IBS), colorectal cancer, and allergies have been linked to changes in the gut microbiota composition (de Vos and de Vos, 2012; Le Chatelier et al., 2013). During the last years, even associations have been made between the gut microbiota composition and behavioral disorders, such as depression, anxiety disorder, regressive autism, and schizophrenia (Collins et al., 2012; Braniste et al., 2014; Dinan et al., 2015). However, whereas increasing numbers of animal studies provide evidence for cause-and-effect relationships between shifts in gut microbiota composition and certain disorders (as in the case of obesity; Ridaura et al., 2013), it has not been proven yet for humans whether changes in the gut microbiota composition can cause disorders or that these changes are a consequence of the disorders themselves (de Vos and de Vos, 2012).

In recent years, a few distinct members of the human gut microbiota have received particular attention because of their dedicated metabolism and central role in gut homeostasis and because their loss adversely affects the remaining microorganisms and/or host's health. Bifidobacterium species are one such bacterial species that fulfill important functions within the human colon (Leahy et al., 2005; Rossi and Amaretti, 2011). Decreased numbers of these species in the colon have been associated with several disorders. Moreover, they have shown to interact with other colon bacteria such as butyrateproducing bacteria by cross-feeding interactions. Furthermore, decreased butyrate concentrations and decreased numbers of butyrate producers in the human colon have been associated with disorders. Therefore, this knowledge has encouraged the development of approaches to stimulate the growth and/or activity of bifidobacteria, i.e., the bifidogenic effect, and butyrate-producing colon bacteria, i.e., the butyrogenic effect, in the human colon. The most prevalent approaches to cause bifidogenic and butyrogenic effects involve the consumption of probiotics and prebiotics.

## BIFIDOBACTERIA AND BUTYRATE-PRODUCING COLON BACTERIA

#### Bifidobacterium Species General Aspects

Bifidobacteria are Gram-positive, anaerobic, saccharolytic bacteria that belong to the phylum Actinobacteria; they mainly occur in the gastrointestinal tract of mammals, birds, and insects, but are present in sewage, human breast milk, fermented milk, cheeses, and water kefir too (Bottacini et al., 2014; Khodayar-Pardo et al., 2014; Laureys and De Vuyst, 2014; Laureys et al., 2016). A typical bifidobacterial genome has an average size ranging from 2.0 to 2.8 Mb and is characterized by a high guanine-plus-cytosine content, with numerous genes involved in the uptake and breakdown of carbohydrates from both diet and host origin (Ventura et al., 2014). Bifidobacteria are among the first bacteria to colonize the human gastrointestinal tract and reach their highest proportion in the colon (up to 90% of the total colon microbiota in vaginally delivered breast-fed infants) during the first 12 months of life (Tannock, 2010; Turroni et al., 2012). This abundance significantly decreases over time to <5% in adult subjects and decreases even more in the elderly (Arumugam et al., 2011; Duncan and Flint, 2013). At the time of writing, the Bifidobacterium genus comprised 51 species (Euzéby, 1997, 2016; Laureys et al., 2016), among which Bifidobacterium longum, Bifidobacterium animalis, Bifidobacterium adolescentis, Bifidobacterium bifidum, Bifidobacterium catenulatum, Bifidobacterium pseudocatenulatum, Bifidobacterium breve, Bifidobacterium pseudolongum, Bifidobacterium gallicum, Bifidobacterium angulatum, and Bifidobacterium faecale are encountered in the human colon (Turroni et al., 2009; Ventura et al., 2011; Choi et al., 2014). In general, B. bifidum and B. longum are the dominant species in infants, whereas B. adolescentis and B. longum dominate the adult gut microbiota (Turroni et al., 2012). Quantitative PCR analyses of fecal samples of 42 Belgian healthy adults have shown that the fecal microbiota of adults contains between zero and four (with an average of two) different bifidobacterial species,

among which B. longum (present in 90% of the adults), B. adolescentis (present in 79% of the adults), and B. catenulatum (present in 38% of the adults) are the most frequently detected species (Ishikawa et al., 2013).

#### Functional Role in the Colon

From the growing body of scientific evidence associating decreased numbers of bifidobacteria with disorders, it emerges that these species have a disproportionally large impact in the human colon in relation to their relatively low numerical abundance in adults. Hence, a decrease in the relative abundances of Bifidobacterium species in the human colon has been associated with antibiotic-associated diarrhea, IBS, IBD, obesity, allergies, and regressive autism (Di Gioia et al., 2014; Grimm et al., 2014). Examples of functions carried out by bifidobacteria include the production and/or liberation of B vitamins, antioxidants, polyphenols, and conjugated linoleic acids; maturation of the immune system during early life and preservation of immune homeostasis during life; preservation of gut barrier functions and protection against pathogens by producing bacteriocins, decreasing luminal pH by the production of acids, and blocking the adhesion of pathogens to the intestinal mucosa (Leahy et al., 2005; Gorissen et al., 2010, 2012; Rossi and Amaretti, 2011; Gagnon et al., 2015). However, these functions are not characteristic for the Bifidobacterium genus or certain species, but are rather strain-specific. Another important function of the bifidobacterial genus that contributes to gut homeostasis and host health is the production of acetate and lactate during carbohydrate fermentation, organic acids that in turn can be converted into butyrate by other colon bacteria through cross-feeding interactions (**Table 1**; De Vuyst and Leroy, 2011; De Vuyst et al., 2014; Rivière et al., 2015).

#### Metabolism

Bifidobacteria display a strictly fermentative metabolism, i.e., they gain energy in the form of ATP by substrate-level phosphorylation during anaerobic carbohydrate breakdown, and play an important role in the human colon with respect to the degradation of carbohydrates that resist digestion and absorption in the upper gastrointestinal tract (Pokusaeva et al., 2011; De Vuyst et al., 2014). Glycoside hydrolases (EC 3.2.1.x) constitute the most important enzyme group that colon bacteria use to degrade poly- and oligosaccharides to fermentable monosaccharides (van den Broek et al., 2008; van den Broek and Voragen, 2008). Compared with the human genome, encoding only 17 glycoside hydrolases for the digestion of food carbohydrates, bifidobacterial genomes possess high numbers of genes encoding these carbohydrases (El Kaoutari et al., 2013). As an example, the genome of B. longum NCC2705 contains 56 genes encoding glycoside hydrolases, one gene encoding a carbohydrate esterase (EC 3.1.1.x), but no genes encoding polysaccharide lyases (EC 4.2.2.x; Schell et al., 2002; Lombard et al., 2014). Bifidobacteria are particularly specialized in efficient uptake of short oligosaccharides into the cell, where

#### TABLE 1 | Overview of the physiological effects of the short-chain fatty acids (SCFAs) acetate, propionate, and butyrate produced by human colon

bacteria (Hamer et al., 2008; Al-Lahham et al., 2010; Havenaar, 2011; Macfarlane and Macfarlane, 2012; Chang et al., 2014; Louis et al., 2014; Tralongo et al., 2014).


they are further degraded to monosaccharides, i.e., they display a preferential oligosaccharide metabolism, providing them a competitive advantage toward other colon bacteria that degrade carbohydrates extracellularly (Van der Meulen et al., 2004, 2006b; Falony et al., 2009a; De Vuyst and Leroy, 2011; De Vuyst et al., 2014). About 5% of the total bifidobacterial gene content is dedicated to carbohydrate internalization, through either ATPbinding cassette transporters, permeases, or proton symporters (Ventura et al., 2009). For example, B. longum NCC2705 contains 15 genes that encode transport systems that could be involved in the transport of oligosaccharides (Schell et al., 2002; Parche et al., 2007). Several laboratory fermentation studies have shown that bifidobacteria can use various non-digestible carbohydrates as energy sources, encompassing plant-derived carbohydrates [such as resistant starch, pectin, inulin, arabinoxylan (AX), cellulose, and their corresponding oligosaccharides] and host-produced carbohydrates (human milk oligosaccharides and mucin), although this ability is strain-dependent too (Klijn et al., 2005; De Vuyst et al., 2014; Rivière et al., 2014; McLaughlin et al., 2015; Selak et al., 2016).

Once internalized into the cytoplasm, hexose monosaccharides (e.g., fructose and glucose) are converted into acetate and lactate by the fructose 6-phosphate phosphoketolase pathway or bifid shunt (De Vuyst et al., 2014). Bifidobacteria will initially cleave, by means of the key enzyme fructose 6-phosphate phosphoketolase, one mole of fructose 6-phosphate into one mole of erythrose 4-phosphate and one mole of acetyl-phosphate (**Figure 2A**). From erythrose 4-phosphate and an additional mole of fructose 6-phosphate, one mole of ribose 5-phosphate and one mole of xylulose 5-phosphate

are formed by the successive action of a transaldolase and a transketolase. Two moles of xylulose 5-phosphate are subsequently converted into two moles of acetyl-phosphate and two moles of glyceraldehyde 3-phosphate by the action of a xylulose 5-phosphate phosphoketolase. These two moles of acetyl-phosphate plus the additional mole of acetyl-phosphate (produced by the fructose 6-phosphate phosphoketolase) are further converted into three moles of acetate by an acetate kinase, which is accompanied by the production of three moles of ATP. The two moles of glyceraldehyde 3-phosphate are oxidized into two moles of pyruvate by enzymes participating in the Embden-Meyerhof-Parnas pathway, which results in an additional production of two moles of ATP. In a last step, pyruvate can be reduced into lactate by means of a lactate dehydrogenase, which is accompanied by NAD<sup>+</sup> recycling. Thus, when fermenting hexose monosaccharides, acetate and lactate are produced in a theoretical molar ratio of 1.5 and three moles of ATP are produced. Pentose monosaccharides (e.g., arabinose and xylose) can also be shuttled into the bifid shunt by their conversion into xylulose 5-phosphate (**Figure 2A**). However, this is not accompanied by the production of an additional mole of acetate (and thus no additional mole of ATP) as in the case of hexose fermentation, leading to a final theoretical molar ratio of acetate to lactate of 1.0 and two moles of ATP (Pokusaeva et al., 2011; De Vuyst et al., 2014). However, these theoretical ratios are rarely found during bifidobacterial carbohydrate fermentation, due to the production of formate, acetate, and ethanol from pyruvate instead of lactate (**Figure 2A**), which depends on the available energy source and its consumption rate (Palframan et al., 2003; Van der Meulen et al., 2004, 2006a,b; Falony et al., 2009b; De Vuyst et al., 2014). The production of formate from pyruvate by a formate acetyltransferase, at the expense of lactate, can be explained by the need for additional ATP production by means of the concomitant production of acetate when bifidobacteria are grown on complex carbohydrates to improve their fitness, despite their lower growth rate compared with simple carbohydrates. Bifidobacteria are also able to produce ethanol from acetyl-CoA with a bifunctional aldehyde-alcohol dehydrogenase, at the expense of acetate, to enable the continuation of pyruvate production by regenerating NAD+. This shift in metabolism away from lactate production has been found for the degradation of complex carbohydrates such as inulin-type fructans (ITF; oligofructose and inulin; Van der Meulen et al., 2004; Falony et al., 2009b) and arabinoxylan-oligosaccharides (AXOS; Rivière et al., 2014, 2015). Bifidobacteria can also regenerate NAD<sup>+</sup> by the production of succinate from oxaloacetate that is in turn formed from phosphoenolpyruvate (**Figure 2A**; Van der Meulen et al., 2006a).

#### Butyrate-Producing Colon Bacterial Species General Aspects

Gene-targeted approaches to investigate the butyrate-producing bacterial communities of the human gut microbiota have led to the consideration that butyrate-producing colon bacteria form a functional group rather than a monophyletic group. Most butyrate producers in the human colon belong to the Firmicutes phylum and in particular clostridial clusters IV and XIVa (Louis and Flint, 2009; Van den Abbeele et al., 2013a; Vital et al., 2014). Clostridial clusters IV and XIVa butyrate producers are Gram-positive, highly oxygen-sensitive, strictly anaerobic, saccharolytic bacteria. The two most dominant bacterial species in the human colon are Faecalibacterium prausnitzii (up to 14% of the total fecal gut microbiota, clostridial cluster IV) and Eubacterium rectale (up to 13% of the total fecal gut microbiota, clostridial cluster XIVa), and are expected to have a significant contribution to butyrate production (De Vuyst et al., 2014; Walker et al., 2014). Other important butyrateproducing bacterial species in the human colon are Roseburia spp. (clostridial cluster XIVa, such as Roseburia faecis, Roseburia inulinivorans, Roseburia intestinalis, and Roseburia hominis), Eubacterium spp. (clostridial cluster XIVa, such as Eubacterium hallii), Anaerostipes spp. (clostridial cluster XIVa, such as Anaerostipes butyraticus, Anaerostipes caccae, and Anaerostipes hadrus), and Butyricicoccus pullicaecorum (clostridial cluster IV). Some of these species (such as E. rectale, F. prausnitzii, and R. intestinalis) preferentially colonize the mucus layer, and consequently increase the butyrate bioavailability for colon epithelial cells, whereas other species such as A. caccae mainly occur in the lumen of the colon (El Aidy et al., 2013; Van den Abbeele et al., 2013a). In contrast to bifidobacteria, clostridial clusters IV and XIVa do not directly colonize the colon in high quantities after birth. In the case of F. prausnitzii, it has been shown that fecal numbers in infants younger than 6 months are undetectable, slightly increase between the age of 6 and 24 months, then suddenly increase to reach a peak during late childhood and adolescence, and finally decrease again during adulthood and especially in the elderly (Miquel et al., 2014).

#### Functional Role in the Colon

Clostridial clusters IV and XIVa have gained a lot of attention during the last years because of their contribution to gut homeostasis, by preserving gut barrier functions and exerting immunomodulatory and anti-inflammatory properties (Velasquez-Manoff, 2015). In addition to the beneficial properties of the butyrate produced (**Table 1**), F. prausnitzii produces antiinflammatory peptides blocking nuclear factor NF-κB activation and cytokine IL-8 production in mice, which provide protection against chemically induced colitis (Qiu et al., 2013; Quévrain et al., 2016). Several studies have shown that the abundance of B. pullicaecorum, E. rectale, F. prausnitzii, and/or R. intestinalis is markedly decreased in IBD patients (Morgan et al., 2012; Eeckhaut et al., 2013; Gevers et al., 2014) and that such patients have lower concentrations of butyrate in their feces than healthy individuals (Marchesi et al., 2007; Nemoto et al., 2012). Less butyrate producers were also found in patients with colorectal cancer (Wu et al., 2013). Therefore, methods are being searched to stimulate butyrate-producing human colon bacterial species by diet (prebiotic approach) or by administering these bacteria orally (probiotic approach). In medical applications, pure butyrate by means of tablets or rectal enemas is used as a therapeutic agent for IBD treatment (Geirnaert et al., 2014).

#### Metabolism

Like bifidobacteria, members of clostridial clusters IV and XIVa carry out a fermentative metabolism and are often able to degrade a wide range of non-digestible carbohydrates in the human colon anaerobically, encompassing resistant starch, ITF, xylo-oligosaccharides (XOS), and AXOS (Falony et al., 2009c; Louis and Flint, 2009; Scott et al., 2014; Rivière et al., 2015; Moens et al., 2016). As an example, the genome of E. rectale ATCC 33656 encodes 52 glycoside hydrolases, encompassing βfructofuranosidases, α-arabinofuranosidases, β-xylosidases, exooligoxylanases, α-amylases, α- and β-glucosidases, α- and βgalactosidases, and cellulases (Lombard et al., 2014). However, inter-genus variations have been found within clostridial clusters IV and XIVa, as not all species and even strains within one species can consume complex carbohydrates to the same extent (Falony et al., 2009c; Scott et al., 2014; Moens et al., 2016). Most of the butyrate-producing colon bacteria use a non-preferential extracellular degradation mechanism for the breakdown of oligo- and polysaccharides, with the release of monosaccharides into the extracellular medium. As illustrated during laboratory batch fermentation experiments, co-cultivation of such butyrate-producing bacteria with bifidobacteria that have a preferential carbohydrate degradation mechanism, can comprise the competitiveness of the butyrate-producing colon bacteria (Falony et al., 2006, 2009c; De Vuyst et al., 2014). For instance, the percentage of oligofructose that was consumed by F. prausnitzii DSM 17677<sup>T</sup> when cocultivated with different bifidobacterial strains decreased with an increasing ITF degradation capacity of the latter (Moens et al., 2016).

Once internalized into the cytoplasm, hexoses and pentoses are degraded to pyruvate by the Embden-Meyerhof-Parnas pathway or pentose phosphate pathway, respectively. Like other fermentative bacteria, clostridial clusters IV and XIVa butyrate producers possess several alternative pathways to form different end-metabolites from pyruvate, depending on the bacterial species, carbohydrate source, hydrogen gas pressure, and necessity of redox balancing. Besides butyrate, they can form lactate, formate, hydrogen gas, and carbon dioxide (**Figure 2B**). Pyruvate can get reduced into lactate by means of a lactate dehydrogenase, which is accompanied by NAD<sup>+</sup> recycling (for instance R. inulinivorans and E. rectale; Falony et al., 2009c; Rivière et al., 2015; Moens et al., 2016). The production of butyrate from pyruvate involves the conversion of pyruvate into acetyl-CoA by a pyruvate:ferredoxin oxidoreductase, with the reduction of ferredoxin and production of carbon dioxide (for instance most clostridial clusters IV and XIVa butyrateproducing colon bacteria; Falony et al., 2009c; Moens et al., 2016) and/or by a pyruvate-formate lyase with the formation of formate (for instance F. prausnitzii and E. rectale; Rivière et al., 2015; Moens et al., 2016). Two moles of acetyl-CoA are then converted via a four-step pathway into butyryl-CoA, in which the last step is carried out by a butyryl-CoA dehydrogenase/electron-transferring flavoprotein complex that catalyzes the NADH + H+-dependent reduction of crotonyl-CoA coupled to the reduction of ferredoxin. The final step from butyryl-CoA to butyrate is either catalyzed by a butyrate kinase (after phosphorylation of butyryl-CoA) or a butyryl-CoA:acetate CoA transferase (Falony et al., 2009c; Louis and Flint, 2009; Mahowald et al., 2009; De Vuyst and Leroy, 2011; Moens et al., 2016). Only a few butyrate-producing colon bacteria, encompassing Clostridium butyricum, Coprococcus eutactus, and Coprococcus comes, are known to use a butyrate kinase to produce butyrate (Louis and Flint, 2009; Vital et al., 2014). The butyryl-CoA:acetate CoA transferase step involves the consumption of external acetate (coming from for instance bifidobacterial carbohydrate breakdown through cross-feeding) as a co-substrate, thereby producing acetyl-CoA and butyrate. The acetyl-CoA produced can be converted via acetyl-phosphate into acetate, with the production of ATP, by acetate kinase, or recycled into the four-step pathway mentioned above (Falony et al., 2009c). The reduced ferredoxin can be reoxidized via a ferredoxin hydrogenase, with the concomitant production of H2, and/or via a membrane-bound ferredoxin oxidoreductase (Rnf) complex, without production of H2, but with the generation of a proton-motive force that allows additional ATP production (for instance R. inulinivorans and F. prausnitzii; Falony et al., 2009c; Moens et al., 2016). The production of butyrate thus not only leads to regeneration of NAD<sup>+</sup> from NADH + H<sup>+</sup> produced in the upper parts of the carbohydrate degradation pathways for ATP production, but also leads to additional ATP production (Falony et al., 2009c; Louis and Flint, 2009; Mahowald et al., 2009; De Vuyst and Leroy, 2011). Some butyrate producers, encompassing A. caccae, A. butyraticus, A. hadrus, and E. hallii, can produce butyrate from lactate instead of carbohydrates (**Figure 2B**; Duncan et al., 2004; Falony et al., 2006, 2009c; Belenguer et al., 2011; De Vuyst and Leroy, 2011; De Vuyst et al., 2014).

#### STIMULATION OF BIFIDOBACTERIA AND BUTYRATE-PRODUCING COLON BACTERIA

Since decreased numbers of Bifidobacterium species and butyrate-producing bacterial species in the human colon have been reported in patients with diverse disorders and because the SCFAs produced by these species have beneficial effects (**Table 1**), these bacteria are potential candidates to be stimulated in the colon to prevent and restore a disturbed gut homeostasis. The most prevalent strategies to stimulate bifidobacteria and butyrate-producing colon bacteria in the human colon involve the consumption of probiotics and prebiotics (Scott et al., 2015).

#### Probiotics

According to the international scientific association for probiotics and prebiotics (ISAPP), probiotics are defined as "live microorganisms that, when administered in adequate amounts, confer a health benefit on the host" (Hill et al., 2014). Selected strains of Bifidobacterium species are commonly used probiotics and are added to food supplements and foods (especially dairy products). The oral consumption of bifidobacteria has been associated with beneficial effects for different digestive problems and disorders, encompassing acceleration of the gut transit time; improvement of lactose intolerance; prevention of antibiotic-associated diarrhea and necrotizing enterocolitis (in pre-term infants that usually harbor reduced numbers of bifidobacteria); and alleviation of IBS and IBD symptoms (Leahy et al., 2005; Di Gioia et al., 2014; Tojo et al., 2014; Saez-Lara et al., 2015). Also, evidence continues to emerge that bifidobacteria influence immune responses and hence may enhance resistance to infections and allergies (Di Gioia et al., 2014; Frei et al., 2015). Further, bifidobacteria display antiinflammatory effects and negatively correlate with metabolic endotoxemia (Everard and Cani, 2013). Moreover, interest is growing to use bifidobacterial strains (such as Bifidobacterium infantis 35624) as psychobiotics, which are "live organisms that, when ingested in adequate amounts, produce a health benefit in patients suffering from psychiatric illness" (Dinan et al., 2013, 2015). The health effects that bifidobacteria exert are of course strain-related; some bifidobacterial strains are effective, whereas others are not. Moreover, the probiotic health benefits are probably not caused by the bifidobacterial strains consumed solely, but are rather the result of interactions with the resident gut microbiota (Cani and Van Hul, 2015; Scott et al., 2015). Indeed, a recent metagenomic and metatranscriptomic study of feces of 12 healthy individuals has shown that the oral administration of the probiotic strain Lactobacillus rhamnosus GG significantly changes the activity of the resident gut microbiota, without influencing the gut microbiota composition itself (Eloe-Fadrosh et al., 2015). Especially genes involved in adhesion, chemotaxis, and/or motility of Bifidobacterium spp., Eubacterium spp., and Roseburia spp. are overexpressed during probiotic consumption, suggesting that the consumption of the probiotic strain promotes interactions between the resident gut microbiota and the host. Nowadays, there is also a growing interest toward the use of other bacterial strains as probiotics, such as Akkermansia muciniphila and butyrate-producing colon bacteria, encompassing B. pullicaecorum, E. rectale, F. prausnitzii, and Roseburia spp. (Marteau, 2013; Geirnaert et al., 2014; Cani and Van Hul, 2015; Scott et al., 2015). For example, the oral administration of B. pullicaecorum 25-3<sup>T</sup> and F. prausnitzii A2-165 in rodents has shown attenuation of chemically induced colitis (Eeckhaut et al., 2013, 2014; Martín et al., 2015). However, whether these strict anaerobic colon bacteria can survive the harsh industrial production steps and deal with the regulatory hurdles (as these bacteria have no history of safe use) will partly determine their application as probiotics in the future human diet (Figueroa-González et al., 2011; Gosálbez and Ramón, 2015; Kumar et al., 2015; Scott et al., 2015).

Since the implementation of EU legislation on health claims in 2009, no health claims for probiotics in foods have been approved by the European Food Safety Authority (EFSA) neither can the term probiotic further be used as a food label in Europe (Glanville et al., 2015). The only approved health claim is the benefit on lactose digestion when consuming live Lactobacillus delbrueckii subsp. bulgaricus and Streptococcus thermophilus strains present in yogurt or fermented milk (EFSA, 2010).

In severe cases of a disturbed gut homeostasis, whereby probiotic treatments do not suffice, the gut microbiota can be restored by transplanting the complete fecal microbiota from a healthy donor into a diseased person. However, the ISAPP recommends that fecal microbiota transplantations (FMTs) should not be considered as probiotics, as they are uncharacterized mixtures of strains (Hill et al., 2014). FMTs have shown to be very effective for curing Clostridium difficile infections, although they have ambiguous outcomes for IBD and IBS (Aroniadis and Brandt, 2014; Pamer, 2014). Furthermore, a step-up FMT strategy has been proposed to treat Crohn's disease and ulcerative colitis, which consists of a FMT, followed by additional FMT steps or standard IBD medications depending on the patient's clinical response to the treatment (Cui et al., 2016). Also, it has been shown that patients with metabolic syndrome display improved insulin sensitivity after being treated with fecal microbiota of healthy individuals (Vrieze et al., 2012). These patients possess increased numbers of butyrate-producing colon bacteria and decreased numbers of Gram-negative bacteria after a FMT. Studies are being performed to see whether FMTs can also cure non-gastrointestinal disorders, such as allergies and behavioral disorders (Xu et al., 2015). However, up to now, few fecal transplants have been performed, as the selection of healthy fecal donors requires a thorough examination to avoid the transfer of pathogens and gut microbiota-associated disorders (Kapel et al., 2014). Therefore, new approaches are being searched to transplant well-defined mixtures of bacteria (de Vos, 2013; Van den Abbeele et al., 2013b). However, an additional challenge in selecting an appropriate healthy donor or bacterial synthetic community is that, despite the large amount of information about the composition and diversity of the human gut microbiota, it is difficult (if not impossible) to define a healthy gut microbiota composition, as each healthy individual harbors a unique gut microbiota (de Vos and de Vos, 2012; Faith et al., 2013; Li et al., 2016).

## Prebiotics

#### General

Another strategy to increase bifidobacteria and butyrateproducing bacteria in the human colon is through the consumption of prebiotics, which are defined according to the ISAPP as "a selectively fermented ingredient that results in specific changes in the composition and/or activity of the gastrointestinal microbiota, thus conferring benefit(s) upon host health" (Gibson et al., 2010). To date, all well-known prebiotics are carbohydrates, although other compounds such as, for instance, polyphenols may display prebiotic properties as well (Bindels et al., 2015). Compared with probiotics, prebiotics are more stable and thus can easily be added to foods, such as yogurts, biscuits, breads, cereals, spreads, ice creams, and drinks (Gibson et al., 2010). The criteria for classifying a compound as a prebiotic have been listed as (i) resistance to gastric acidity, hydrolysis by mammalian digestive enzymes, and gastrointestinal absorption; (ii) fermentation by intestinal microbiota; and (iii) selective stimulation of the growth and/or activity of intestinal bacteria associated with health and well-being (Gibson et al., 2004). In the past, the impact of the consumption of prebiotics on the gut microbiota composition was mainly studied regarding species of Bifidobacterium and Lactobacillus (Verbeke, 2014). However, recent community-wide analyses of the gut microbiota show that prebiotics are not that selective as previously assumed, and that they stimulate other bacteria too (Bindels et al., 2015). It has indeed been shown that butyrate-producing colon bacteria, such as E. rectale, F. prausnitzii, and Roseburia spp., can consume prebiotics such as ITF (Falony et al., 2006, 2009c; Rivière et al., 2015; Moens et al., 2016). Also, the consumption of oligofructose changes the relative abundance of 102 bacterial taxa in mice, of which 16 display a more than 10-fold decrease or increase in abundance (Everard et al., 2011). Therefore, Bindels et al. (2015) proposed to define a prebiotic as "a non-digestible compound that, through its metabolization by microorganisms in the gut, modulates composition and/or activity of the gut microbiota, thus conferring a beneficial physiological effect on the host." Alternatively, the definition of prebiotics has been challenged over time not only according to scientific considerations but also due to its importance for regulators, industry, and consumers (Hutkins et al., 2016). As for probiotics, the term prebiotic cannot be used as a health claim on food products in Europe (Salminen and van Loveren, 2012). Some claims exist for the term fiber (EFSA, 2011a,b, 2015), but not all fibers are prebiotics, whereby the latter are distinguished from the former by the selectivity of their fermentation (Slavin, 2013; Hutkins et al., 2016; Verspreet et al., 2016).

Examples of prebiotic non-digestible carbohydrates that are bifidogenic include poly- and oligosaccharides containing fructose (and a terminal glucose) as in ITF, galactose and glucose (as in galacto-oligosaccharides), glucose (as in isomaltooligosaccharides), galactose and fructose (as in lactulose), xylose (as in XOS), and arabinose and xylose (as in AX and AXOS) (Roberfroid, 2005; Macfarlane et al., 2008; Broekaert et al., 2011; De Vuyst and Leroy, 2011; De Vuyst et al., 2014). Whereas, in the past the target genus for prebiotic stimulation was Bifidobacterium (Gibson et al., 2010), today new prebiotics are searched to stimulate other beneficial bacterial species in the human colon such as butyrate producers. Of special interest are prebiotics that cause both a bifidogenic effect and a butyrogenic effect. ITF, AX, and AXOS are such prebiotics that stimulate both bifidobacteria and the production of butyrate (Falony et al., 2006, 2009b,c; De Vuyst and Leroy, 2011; De Vuyst et al., 2014; Rivière et al., 2015).

#### ITF As an Example of Well-Known Prebiotics

Inulin naturally occurs in fruits and plants such as chicory roots, wheat, onion, banana, garlic, and leek, but is generally extracted from chicory roots on an industrial scale (Roberfroid, 2007). Inulin consists of a linear backbone of β-(2→1)-linked fructose monomers with a degree of polymerization (DP) between 2 and 65 (average DP of 10), which is often linked to a terminal glucose monomer by an α-(1→2)-glycosidic bond (**Figures 3A,B**). Oligofructose is derived from native inulin by partial enzymatic hydrolysis with an inulinase and has a DP that varies between 2 and 8 (average DP of 4). Given the relative simple structures of ITF, only few bacterial enzymes are required for their degradation in the human colon, encompassing enzymes belonging to the β-fructofuranosidase (EC 3.2.1.26) superfamily that cleave terminal fructose residues from the non-reducing ends of the fructose polymers (**Figure 3B**; Scott et al., 2011). Several β-fructofuranosidases have been isolated and characterized in colon bacteria, for instance in Bifidobacterium species (Warchol et al., 2002; Ehrmann et al., 2003; Omori et al., 2010; Jedrzejczak-Krzepkowska et al., 2011) and R. inulinivorans (Scott et al., 2011). Examples of beneficial effects of the consumption of ITF include increased stool frequency, increased colonic absorption of dietary minerals (calcium and magnesium), decreased proteolytic activity, and increased secretion of satiety hormones (Schaafsma and Slavin, 2015).

ITF belong to the most studied prebiotics and their bifidogenic and butyrogenic effects have been well established in various studies (De Vuyst and Leroy, 2011; De Vuyst et al., 2014). For instance, it has been shown that not all bifidobacterial strains benefit in the same way from the presence of ITF in the human colon. A comparative statistical study of 18 bifidobacterial strains, belonging to 10 different species and

coming from different donors and origins, has shown the existence of four different clusters of strains differing in their mechanisms and capabilities to degrade ITF (Falony et al., 2009b). Some strains only consume fructose (cluster A), whereas others consume both fructose and oligofructose, mainly short oligosaccharides (DP up to seven) after import into the cell, i.e., they display a preferential metabolism (cluster B). Certain strains degrade both oligofructose and inulin (short chain length fractions only) extracellularly, accompanied with the release of fructose into the extracellular medium, i.e., they display a nonpreferential metabolism (clusters C and D). A recent study of 190 bifidobacterial strains isolated from different donors and colon regions has shown that these ITF degradation fingerprints are not correlated with the region in the intestine, suggesting that the degradation of ITF is uniform along the human intestine (Selak et al., 2016). Yet, intra-species variability in ITF degradation capacity indicates strain-specific variations. Moreover, within one colon region bifidobacterial strains with different ITF degradation mechanisms occur, which suggests cooperation for the degradation of ITF in the colon, with opportunities for cross-feeding on strain and/or species level. Similar crossfeeding between bifidobacterial strains with complementary degradation mechanisms has also been shown for starch, xylan, and mucin glycoproteins (Egan et al., 2014; Turroni et al., 2015). Also, it has been shown that the consumption of ITF, the bifidogenic effect, and the butyrogenic effect are linked to each other, because of cross-feeding interactions between bifidobacteria and butyrate-producing colon bacteria (**Figure 4**; Belenguer et al., 2006; Falony et al., 2006, 2009c; Moens et al., 2016). As an end-metabolite of the bifid shunt and a co-substrate for the production of butyrate (Section AX and AXOS as an Example of Interesting Prebiotics), acetate plays a key role in cross-feeding interactions between bifidobacteria and butyrate-producing colon bacteria in the human colon. In a first type of cross-feeding, both the bifidobacterial and butyrate-producing strains consume ITF (**Figure 4**). The consumption of ITF by bifidobacteria provides butyrate-producing colon bacteria with exogenous acetate that is used as a co-substrate to produce butyrate by growing on ITF simultaneously [which is, for instance, the case for R. intestinalis DSM 14610 (Falony et al., 2006), R. inulinivorans DSM 16841 (Falony et al., 2009c), and F. prausnitzii DSM 17677<sup>T</sup> (Moens et al., 2016)]. However, such cross-feeding interactions can be either a pure commensal or beneficial relationship between these bacteria or can be dominated by competition, depending on the ITF degradation capacities of the bifidobacterial strains involved (Moens et al., 2016). A second type of cross-feeding takes place between bifidobacteria that consume ITF, and concomitantly produce acetate, and acetate-consuming butyrate-producing colon bacteria that are not able to degrade ITF (**Figure 4**). Instead of ITF, the latter bacteria consume carbohydrate breakdown products (shortchain oligosaccharides) liberated by the bifidobacterial strain (which is, for instance, the case for R. hominis DSM 16839; Belenguer et al., 2006) or lactate (for instance E. hallii DSM 17630; Belenguer et al., 2006, and A. caccae DSM 14662; Falony et al., 2006).

#### AX and AXOS As an Example of Interesting Prebiotics **Physiological effects**

Growing interest is devoted to complex non-digestible carbohydrates that ferment slowly and thereby cause bifidogenic and butyrogenic effects along the entire length of the human colon. AX and AXOS, as a broad class of heteropolysaccharides and -oligosaccharides with complex varying structures (**Figures 3C,D**), belong to these slow-fermenting carbohydrates and hence are able to decrease the production of bacterial toxic metabolites originating from protein and lipid metabolism in the distal colon (Section Bifidobacterium Species; Van Craeyveld et al., 2008; Grootaert et al., 2009; Sanchez et al., 2009; Neyrinck et al., 2011). This is to be explained by a stimulation of saccharolytic activities, an increase in SCFA production, and a lowering of the luminal pH in the distal part of the colon, where carbohydrates are rare and proteolytic bacteria, such as Bacteroides spp., are otherwise favored (Duncan et al., 2009). Examples of additional potential benefits of the consumption of AX and AXOS for human health include improved mineral (calcium and magnesium) absorption; increased stool frequency and improved stool consistency; reduced post-prandial glycemic response; reduced blood cholesterol levels; and increased antioxidant capacity (Grootaert et al., 2007; Broekaert et al., 2011; Damen et al., 2011; Mendis and Simsek, 2013). Moreover, the consumption of AXOS, with the increase of bifidobacterial numbers as a result, may help to restore gut barrier functions and cure metabolic endotoxemia in mice (Neyrinck et al., 2012).

#### **Occurrence, structural properties, and degradation**

AX naturally occur in the endosperm and bran (pericarp, testa, and aleuron layer) of cereal grains such as wheat, rye, rice, barley, oat, and sorghum, but in varying quantities, depending on the cereal species and the location within the cereal kernel (Izydorczyk and Biliaderis, 2006). For instance, the endosperm of wheat kernels contains ca. 2% of AX, whereas the pericarp contains ca. 38% of AX (Benamrouche et al., 2002; Maes and Delcour, 2002). AX consist of a linear backbone of 1500 to 15,000 β-(1→4)-linked xylose monomers, which can randomly be substituted with arabinose monomers on the C-(O)-2 or C-(O)-3 positions (monosubstituted) or on both positions (disubstituted; **Figures 3C,D**; Izydorczyk and Biliaderis, 1995). Distribution patterns of arabinose substituents on the xylose backbone are not regular for wheat AX; highly branched regions are interlinked by sequences of contiguous non-substituted xylose residues (Gruppen et al., 1993). The number of arabinose substituents bound to the xylose backbone is expressed as the arabinose/xylose ratio (A/X) and depends on the cereal species and the location within the kernel. For instance, in the pericarp, testa, aleuron layer, and endosperm of wheat kernels, different A/Xs are found, namely ca. 1.0, 0.1, 0.4, and 0.5, respectively (Izydorczyk and Biliaderis, 1995; Antoine et al., 2003; Barron et al., 2007). The fermentability of AX and AXOS in the human colon is strongly influenced by the complexity of the AXOS molecules and decreases with increasing DP and increasing A/X (Van Craeyveld et al., 2008; Pollet et al., 2012). Additionally, xylose residues can be esterified with glucuronic acid and acetyl groups,

whereas arabinose residues can be esterified with ferulic acid and p-coumaric acid, although in low numbers (**Figure 3D**; Izydorczyk and Biliaderis, 1995). These esterifications are of health and physicochemical importance, since ferulic acid and p-coumaric acid are antioxidants and potent cross-linking sites for attachment to other AX chains (Bunzel et al., 2001; Ou and Sun, 2014). The presence of feruloylated and diferuloylated arabinose substituents reduces the fermentability of AX and AXOS (Hopkins et al., 2003; Snelders et al., 2014). As cereal whole grains only contain low concentrations of AX (varying between 1.8% of AX in sorghum and 12.1% of AX in rye; Izydorczyk and Biliaderis, 2006), and thus the overall intake of AX is low (especially in modern Western-type diets with high intakes of refined cereal products), AX can be extracted from cereal grains and added to food products in higher concentrations (Broekaert et al., 2009). On an industrial scale, AX are usually extracted from wheat bran that is available in large quantities in Europe (Swennen et al., 2006). AXOS, the hydrolysis products of AX, are formed not only in processed cereal-based food products such as bread and beer (Courtin et al., 2009; Broekaert et al., 2011), but can also be produced on an industrial scale by the enzymatic cleavage of AX with β-endoxylanases (Broekaert et al., 2009, 2011). This results in various substituted molecules (i.e., AXOS) and non-substituted molecules (i.e., XOS), differing in DP and A/X.

Given their complex structures, the degradation of AX and AXOS in the human colon requires the cooperative action of debranching and depolymerizing bacterial carbohydrate-active enzymes, encompassing β-endoxylanases (EC 3.2.1.8) that cleave AX into AXOS and XOS; β-xylosidases (EC 3.2.1.37) that cleave terminal xylose residues from the non-reducing ends of the xylose backbones; exo-oligoxylanases (EC 3.2.1.156) that release terminal xylose residues from the reducing ends of the xylose backbones; and α-arabinofuranosidases (EC 3.2.1.55) that remove arabinose substituents from the xylose backbones (**Figure 3D**). Additional enzymes are needed to cleave glucuronic acid [i.e., α-glucuronidase (EC 3.2.1.139)], ferulic acid [i.e., ferulic acid esterase (EC 3.1.1.73)], acetyl groups [i.e., acetyl xylan esterase (EC 3.1.1.72)], and p-coumaric acid [i.e., p-coumaric acid esterase (EC 3.1.1.-)] from AXOS (**Figure 3D**; Dodd and Cann, 2009; Lagaert et al., 2014).

To date, AX and AXOS fall under the definition of dietary fiber (European Commission, 2008; Snelders et al., 2014) but are not considered as prebiotics by the EFSA, although they meet the three criteria of prebiotics (see above; Broekaert et al., 2011). AX and AXOS are neither digested nor absorbed in the upper gastrointestinal tract and reach the human colon intact, where they are fermented by the resident colon bacteria and cause bifidogenic and butyrogenic effects (**Table 2**). However, as is also the case for other prebiotics, the selective stimulation


(Continued)


(Continued)

Frontiers in Microbiology | www.frontiersin.org


criterion can be questioned. Several in vivo and in vitro studies have shown that AX and AXOS stimulate, besides bifidobacteria and butyrate-producing colon bacteria, other saccharolytic colon bacteria too, such as Bacteroides spp. and Lactobacillus spp. (**Table 2**). Moreover, a propionogenic effect is supposed to occur. A few studies have shown that AX and AXOS especially stimulate the production of propionate (**Table 2**; Hopkins et al., 2003; Van den Abbeele et al., 2011; Pollet et al., 2012). For instance, the mucin-consuming propionate-producing A. muciniphila is stimulated in the colon of humanized rats fed with long-chain AX (**Table 2**; Van den Abbeele et al., 2011). Whether this is a direct or indirect effect is not known yet.

#### **Bifidogenic effects of AX and AXOS**

Several in vivo studies (in rodents, pigs, and humans) and in vitro studies [during batch and simulator of human intestinal microbial ecosystem (SHIME <sup>R</sup> ) fermentations with fecal slurries] have shown that AX and AXOS are bifidogenic (**Table 2**). An in vivo study with rats has shown that the bifidogenic effect is only caused by AXOS with low average DPs ≤ 5 and A/Xs ≤ 0.27 (Van Craeyveld et al., 2008), whereas other rodent studies have found a stimulation of bifidobacteria by AX and AXOS with high average DPs up to 284 and A/Xs up to 0.70 (**Table 2**; Damen et al., 2011; Neyrinck et al., 2011; Van den Abbeele et al., 2011). In the latter study, a 60-fold increase of bifidobacteria in the cecum of rats has been found, caused by the consumption of long-chain AX (average DP of 60, A/X of 0.70; Van den Abbeele et al., 2011). Apart from in vitro and animal experiments, human studies have revealed a bifidogenic effect caused by a daily intake of 10 g of AXOS per day (Cloetens et al., 2010), 5.5 g of AXOS per day (Maki et al., 2012), and 2.2 g of AX and AXOS per day (Walton et al., 2012; **Table 2**). However, until recently, many fundamental questions remain unanswered. For instance, how can the low numerical abundant bifidobacteria (<5%) compete with other, more abundant, saccharolytic bacteria in the human colon for AX and AXOS? Do bifidobacteria have a preference for certain AX and AXOS molecules? Are all bifidobacterial strains in the human colon stimulated by AX and AXOS? To answer these questions, a detailed knowledge of the carbohydratehydrolyzing capacity of bifidobacteria was missing. Indeed, in the past, studies of the degradation of AX and AXOS through mono-culture fermentations with bifidobacterial strains were restricted to monitoring of bacterial growth, pH, and SCFA production (Van Laere et al., 2000; Crittenden et al., 2002), or fermentations of purified short-chain AXOS standards were performed (Pastell et al., 2009) without revealing the complete fermentation capacity of bifidobacteria. Recently, the mechanistic variations in AXOS degradation by 36 bifidobacterial strains from different donors and origins have been investigated (Rivière et al., 2014). The results show that not all bifidobacterial strains are stimulated by AXOS to the same extent. AXOS degradation by bifidobacteria is complex and involves the consumption of arabinose substituents, whether or not followed by the consumption of the xylose backbones of AXOS, either up to xylotetraose or longer and either intracellularly or extracellularly. Several bifidobacterial strains use the arabinose

TABLE 2 | Continued

WE-AX,

water-extractable

 AX; WB, AX and AXOS from wheat bran; PSH, AX and AXOS from Psyllium seed husk.

substituents of AXOS solely, whereas others first consume the arabinose substituents and later import the xylose backbones (up to xylotetraose) into the cell. This extracellular arabinose substituent-oriented metabolism of bifidobacteria has been linked to the presence of genes encoding extracellular cellassociated α-arabinofuranosidases (Lagaert et al., 2010, 2014; Rivière et al., 2014). The majority of the bifidobacterial strains cannot use xylose backbones longer than xylotetraose, i.e., they display a preferential metabolism, except for one strain among the 36 tested ones, B. catenulatum LMG 11043<sup>T</sup> , that also uses longer xylose backbones, i.e., they display a non-preferential metabolism (Rivière et al., 2014). This could explain why the bifidogenic effect is strongly influenced by the complexity of the AXOS molecules and decreases with increasing DP (**Table 2**; Van Craeyveld et al., 2008). A multivariate data analysis of the fermentation data of these 36 bifidobacterial strains has revealed five species-independent clusters, representing five different complementary AXOS degradation mechanisms (Rivière et al., 2014). Cluster I strains, albeit not all, consume free arabinose and xylose; cluster II strains have an extracellular arabinose substituent-oriented metabolism; cluster III strains display a preferential metabolism of non-substituted xylose backbones; cluster IV strains combine the degradation mechanisms of clusters II and III; and cluster V strains display a non-preferential AXOS metabolism. The complementary degradation mechanism of bifidobacterial strains and the ability of intracellular and cellassociated degradation of xylose backbones and AXOS, could explain the selective stimulation of bifidobacteria by AXOS in the presence of other saccharolytic colon bacteria in the human colon. Whole-genome sequence annotations have revealed that some bifidobacterial strains contain genes coding for enzymes involved in the debranching of substituents and in the exocleavage of the xylose backbones of AX and AXOS (Schell et al., 2002; van den Broek and Voragen, 2008; van den Broek et al., 2008). Indeed, several AXOS-degrading enzymes have been isolated and characterized in bifidobacterial strains, encompassing β-xylosidases in B. adolescentis ATCC 15703 and B. animalis subsp. lactis BB-12; α-arabinofuranosidases in B. adolescentis ATCC 15703, B. adolescentis DSM 20083, B. longum B667, and B. longum NCC2705; and exo-oligoxylanases in B. adolescentis LMG 10502 (Lagaert et al., 2010, 2011, 2014). However, up to now, no β-endoxylanases have been found in the genome of bifidobacteria. The only gene (i.e., BL1543) that was first annotated as a β-endoxylanase in B. longum NCC2705 (Schell et al., 2002) has shown to be an extracellular membraneassociated α-arabinofuranosidase (Lagaert et al., 2010, 2014; Rivière et al., 2014). For the complete utilization of AX, it is likely that most of the Bifidobacterium species require cooperation with β-endoxylanase-producing bacteria, such as Bacteroides and Roseburia species (Chassard et al., 2007; Dodd et al., 2011). For instance, the genome of R. intestinalis L1-82 contains three genes possibly encoding β-endoxylanases (NCBI Resource Coordinators, 2014).

#### **Butyrogenic effects of AX and AXOS**

Besides a bifidogenic effect, AX and AXOS have shown to cause a butyrogenic effect (**Table 2**). In seven of the 13 in vitro and in vivo studies summarized in **Table 2**, bifidobacteria and butyrate-producing colon bacteria (F. prausnitzii, E. rectale, and Roseburia spp.) are stimulated simultaneously, with a significant increase of butyrate production as a result. As these butyrateproducing colon bacteria are present in high numbers in the colon, a rise in butyrate concentration does not come as a surprise (De Vuyst et al., 2014). In contrast to bifidobacteria, much less is known about the genetic AX- and AXOS-degrading potential of species of butyrate-producing colon bacteria. In silico analysis of the genome sequence of, for instance, E. rectale ATCC 33656 has shown that there are five genes possibly encoding AXOS-degrading enzymes (exo-oligoxylanase, bifunctional β-xylosidase/α-arabinofuranosidase, β-xylosidase, and two α-arabinofuranosidases; Rivière et al., 2015).

In contrast to ITF, the link between the consumption of AXOS, the bifidogenic effect, and the butyrogenic effect has been assessed only recently (Rivière et al., 2015). It has been shown that a third type of cross-feeding can take place in the presence of AXOS (**Figure 4**), for instance in the case of B. longum NCC2705 (an arabinose substituent degrader of AXOS) and E. rectale ATCC 33656 (a complete AXOS degrader). Both strains consume AXOS (as in cross-feeding type 1), but the bifidobacterial strain is additionally stimulated by consuming the monosaccharides released by the extracellular degradation of AXOS by the E. rectale strain, leading to cross-feeding interactions that are mutually beneficial (**Figure 4**). It is likely that these kinds of cross-feeding interactions between bifidobacteria and butyrateproducing colon bacteria, caused by prebiotic consumption, will take place in vivo in the human colon (Boets et al., 2013). However, the presence of other bacterial strains, with their own mechanisms of carbohydrate degradation (preferential vs. non-preferential) and own cross-feeding interactions within and between species and genera (**Figure 4**), complicate the attempts to fully understand the bifidogenic and butyrogenic effects of AX and AXOS in the human colon. Furthermore, the interindividual variations in bacterial composition make it even more intricate to predict the effects of prebiotic consumption in the colon.

### CONCLUSIONS

Human gut microbiota research has grown tremendously over the last years in terms of technology development and implications for human health. For instance, it has been shown that certain key bacteria within the colon, such as bifidobacteria and butyrate-producing colon bacteria, are negatively correlated with disorders such as IBD and colorectal cancer. Of the same importance is the progress that is being made into the modulation of the gut microbiota through the use of probiotics, prebiotics, and FMTs to improve human health. Whereas, in the past, the focus was on straightforward increase of bifidobacterial cell concentrations, shifts in interests are currently emphasizing that the stimulation of butyrate-producing bacteria in the human colon is of importance too. The consumption of prebiotic ITF and AXOS seems to be a promising approach to counteract decreased numbers of bifidobacteria and butyrate-producing colon bacteria. The challenge for the upcoming years will however be to first find out whether these changes in gut microbiota composition are the cause or the consequence of a disorder.

# AUTHOR CONTRIBUTIONS

AR acted as the main author. MS, DL, FL, and LD all contributed substantially to the writing and critical revision of the manuscript and approved its final version.

# REFERENCES


#### ACKNOWLEDGMENTS

The authors acknowledge their financial support of the Research Council of the Vrije Universiteit Brussel (SRP7, IRP2, and IOF342 projects) and the Hercules Foundation (grant UABR09004). MS is the recipient of a PhD fellowship of the Vrije Universiteit Brussel in the framework of a bilateral agreement with the University of Ljubljana. AR was the recipient of a PhD fellowship of the Research Foundation Flanders (FWO-Vlaanderen). DL is the recipient of a PhD fellowship in the framework of an IWT-SBO project (BRANDING IWT130028).

Crit. Rev. Food Sci. Nutr. 51, 178–194. doi: 10.1080/104083909030 44768


characterization of the enzyme expressed in Escherichia coli. Curr. Microbiol. 46, 391–397. doi: 10.1007/s00284-002-3908-1


arabino-xylan-oligosaccharide, in breakfast cereal. Nutrition 28, 1115–1121. doi: 10.1016/j.nut.2012.02.010


rapid uptake and conversion of simple carbohydrates. ISME J. 6, 1415–1426. doi: 10.1038/ismej.2011.212

**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 Rivière, Selak, Lantin, Leroy and De Vuyst. 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.

# Cell-Free Spent Media Obtained from Bifidobacterium bifidum and Bifidobacterium crudilactis Grown in Media Supplemented with 3 ′ -Sialyllactose Modulate Virulence Gene Expression in Escherichia coli O157:H7 and Salmonella Typhimurium

#### Pauline Bondue<sup>1</sup> , Sébastien Crèvecoeur <sup>1</sup> , François Brose<sup>1</sup> , Georges Daube<sup>1</sup> , Marie-Christine Seghaye<sup>2</sup> , Mansel W. Griffiths <sup>3</sup> , Gisèle LaPointe<sup>3</sup> and Véronique Delcenserie<sup>1</sup> \*

<sup>1</sup> Department of Food Science, Fundamental and Applied Research for Animal and Health, Faculty of Veterinary Medicine, University of Liège, Liège, Belgium, <sup>2</sup> Department of Pediatrics, University Hospital Liège-Notre-Dame des Bruyères, Belgium, <sup>3</sup> Canadian Research Institute for Food Safety, University of Guelph, Guelph, Canada

Complex oligosaccharides from human milk (HMO) possess an antimicrobial activity and can promote the growth of bifidobacteria such as Bifidobacterium bifidum and Bifidobacterium longum subsp. infantis. In addition, fermentation of carbohydrates by bifidobacteria can result in the production of metabolites presenting an antivirulence effect on several pathogenic bacteria. Whey is rich in complex bovine milk oligosaccharides (BMO) structurally similar to HMO and B. crudilactis, a species of bovine origin, is able to metabolize some of those complex carbohydrates. This study focused on the ability of B. bifidum and B. crudilactis to grow in a culture medium supplemented in 3′ -sialyllactose (3′SL) as the main source of carbon, a major BMO encountered in cow milk. Next, the effects of cell-free spent media (CFSM) were tested against virulence expression of Escherichia coli O157:H7 and Salmonella enterica serovar Typhimurium. Both strains were able to grow in presence of 3′SL, but B. crudilactis showed the best growth (7.92 ± 0.3 log cfu/ml) compared to B. bifidum (6.84 ± 0.9 log cfu/ml). Then, CFSM were tested for their effects on virulence gene expression by ler and hilA promoter activity of luminescent mutants of E. coli and S. Typhimurium, respectively, and on wild type strains of E. coli O157:H7 and S. Typhimurium using RT-qPCR. All CFSM resulted in significant under expression of the ler and hilA genes for the luminescent mutants and ler (ratios of −15.4 and −8.1 respectively) and qseA (ratios of −2.1 and −3.1) for the wild type strain of E. coli O157:H7. The 3′SL, a major BMO, combined with some

#### Edited by:

Francesca Turroni, University College Cork, Ireland

#### Reviewed by:

Alberto Amaretti, University of Modena and Reggio Emilia, Italy Frédéric Borges, École Nationale Supérieure d'Agronomie et des Industries Alimentaires - Université de Lorraine, France

#### \*Correspondence:

Véronique Delcenserie veronique.delcenserie@ulg.ac.be

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 29 April 2016 Accepted: 01 September 2016 Published: 22 September 2016

#### Citation:

Bondue P, Crèvecoeur S, Brose F, Daube G, Seghaye M-C, Griffiths MW, LaPointe G and Delcenserie V (2016) Cell-Free Spent Media Obtained from Bifidobacterium bifidum and Bifidobacterium crudilactis Grown in Media Supplemented with 3 ′ -Sialyllactose Modulate Virulence Gene Expression in Escherichia coli O157:H7 and Salmonella Typhimurium. Front. Microbiol. 7:1460. doi: 10.3389/fmicb.2016.01460 bifidobacteria strains of bovine or human origin could therefore be an interesting synbiotic to maintain or restore the intestinal health of young children. These effects observed in vitro will be further investigated regarding the overall phenotype of pathogenic agents and the exact nature of the active molecules.

Keywords: Bifidobacterium bifidum, Bifidobacterium crudilactis, bovine milk oligosaccharide, Escherichia coli enterohemorragic O157:H7, Salmonella enterica serovar Typhimurium, virulence expression, 3′ -sialyllactose, whey

### INTRODUCTION

Due to the influence on the microbiota of carbohydrate source present in food, breast-fed children are generally in better health than children fed with formula (Arrieta et al., 2014; Smilowitz et al., 2014; Scott et al., 2015). Human milk oligosaccharides (HMO) are complex oligosaccharides found in human milk. Their concentration can reach 15 g/l and more than 500 structures have been identified (Pacheco et al., 2015). These HMO, produced in mammary glands, cannot be metabolized either by the host or most bacteria, while bifidobacteria species have enzymatic activity able to degrade specific α- and β-bonds (Garrido et al., 2013). These bacteria such as Bifidobacterium bifidum subsp. infantis or Bifidobacterium bifidum are mainly found in the feces of breast-fed children. Indeed, the fecal microbiota of breastfed child contain more than 75% bifidobacteria (Di Gioia et al., 2014). In addition to facilitating the growth of beneficial bacteria such as bifidobacteria, HMO can bind with pathogenic virus or bacteria, limiting adhesion to intestinal epithelium and therefore decreasing pathogens colonization by promoting natural excretion (Smilowitz et al., 2014). Also, other metabolites produced by lactic acid bacteria and bifidobacteria are able to inhibit virulence gene expression of Escherichia coli O157:H7 (Medellin-Pena and Griffiths, 2009; Zeinhom et al., 2012), Salmonella enterica serovar Typhimurium SA 941256 (Medellin-Pena et al., 2007; Bayoumi and Griffiths, 2012; S. Typhimurium) and Campylobacter jejuni (Mundi et al., 2013).

Five monosaccharides can be found in different HMO structures: glucose (Glc), galactose (Gal), N-acetylglucosamine (GlcNAc), fucose (Fuc), and sialic acid, also called Nacetylneuraminic (NeuAc). High quantities of lacto-N-biose type I (LNB: Galβ1-3GlcNAc) and fucosylated HMO are an important characteristic of human milk (Chichlowski et al., 2011; Garrido et al., 2013; Dotz et al., 2014; Smilowitz et al., 2014). B. infantis is able to fully degrade HMO intracellularly, contrary to B. longum, B. breve, and B. bifidum (Sela and Mills, 2010; Underwood et al., 2015). It possesses genes encoding specific transporters and four enzymes necessary for HMO degradation (α-fucosidase, αsialidase, β-galactosidase, and β-N-hexosaminidase) (Sela et al., 2008; Sela, 2011; Smilowitz et al., 2014). HMO degradation by B. bifidum occurs outside the cells. Indeed, B. bifidum possesses a lacto-N-biosidase, which cleaves LNB from HMO. LNB is internalized using a specific transport system and is then degraded using LNB-phosphorylase (Sela, 2011; Smilowitz et al., 2014). Therefore, B. infantis and B. bifidum, despite their different gene clusters, both grow very well in the presence of HMO as sole source of carbon (Asakuma et al., 2011; Barile and Rastall, 2013).

Common oligosaccharides used in infant formula are galactooligosaccharides (GOS) and fructo-oligosaccharides (FOS), including inulin. GOS are composed of Glc and Gal, while FOS are composed of fructose and Glc. Their structures are very simple and linear. They are also bifidogenic, but because of their simple structure, they can also be consumed by other members of the intestinal microbiota such as Bacteroides spp. or Clostridium spp. (Chichlowski et al., 2011; Scholtens, 2014). This is probably why the fecal microbiota of formula-fed children contains only 30% bifidobacteria (Di Gioia et al., 2014) and the species that are present are different from those observed in the feces of breast-fed children. The predominant bifidobacteria in formula-fed children are those encountered in adult feces, such as Bifidobacterium longum subsp. longum and Bifidobacterium adolescentis, which present a less diverse enzymatic arsenal (De Vuyst et al., 2013).

Bovine milk oligosaccharides (BMO) can have similar composition and branching as HMO, so they could share some common properties. In addition, B. infantis or B. bifidum can grow in the presence of these carbohydrates (Sela, 2011; Milani et al., 2014). More than 60 BMO have been identified so far (Pacheco et al., 2015) and whey, a by-product of the dairy industry, is an important low cost source of BMO (Barile et al., 2009; Zivkovic and Barile, 2011). Furthermore, Lactobacillus acidophilus La-5 grown in medium supplemented with dairy ingredients such as whey, presented a protective effect in EHEC-infected mice (Zeinhom et al., 2012). However, BMO concentration in bovine milk is lower than HMO concentration in human milk (Barile et al., 2009; Tao et al., 2009; Kelly et al., 2013). Also, fucosylation occurs at very low frequency compared to sialylation, which is contrary to human milk (Tao et al., 2008). Even if the degree of polymerization in BMO is lower than in HMO, they are also protected by α- and β-bonds which are less accessible to other bacteria (Chichlowski et al., 2011). One of the most important BMO found in cow milk is 3′ sialyllactose (3′ SL: NeuAcα2-3Galβ1-4Glc) and its concentration in colostrum can reach 0.85 mg/ml (Nakamura et al., 2003; Urashima et al., 2013). The majority of bifidobacteria grows only in anaerobic conditions, an environment very difficult to reproduce on an industrial scale. In addition, they have to survive

**Abbreviations:** 3 ′ SL, 3′ -sialyllactose; AE, attaching and effacing; BHI, brain heart infusion; BMO, bovine milk oligosaccharide; CFSM, cell-free spent medium; EHEC, enterohaemorragic Escherichia coli; FOS, fructo-oligosaccharide; Fuc, fucose; Gal, galactose; Glc, glucose; GlcNAc, N-acetylglucosamine; GOS, galacto-oligosaccharide; HMO, human milk oligosaccharide; LacNac, N-acetyllactosamine; LB, Luria Bertani; LNB, lacto-N-biose; LNnT, lacto-Nneotetraose; LNT, lacto-N-tetraose; MRS, De Man, Rogosa and Sharpe medium; NeuAc, N-acetylneuraminic acid or sialic acid; NeuGc, N-glycolylneuraminic acid; OD, optical density; T3SS, type III secretion system.

the acidity of the stomach, bile salts and pancreatic enzymatic activity of the digestive tract in order to reach the colon where they will consume oligosaccharides. Bifidobacterium crudilactis FR/62/B/3, a species isolated from raw cow's milk cheese, is oxygen tolerant (Daube et al., 2006; Delcenserie et al., 2013). The genome encodes enzymes degrading BMO (Delcenserie et al., 2007; Milani et al., 2014). These strains from bovine milk could be an interesting source of probiotics for formula supplementation (Delcenserie et al., 2013).

The type III secretion system (T3SS) has a major role in virulence expression of S. Typhimurium and E. coli O157:H7 by injecting effector proteins in intestinal cells and forming attaching and effacing (AE) lesions on host enterocytes (Bayoumi and Griffiths, 2012). The proteins produced by S. Typhimurium are controlled and activated by hilA and sopD genes and the ssrB2 gene is a major regulator of the T3SS (Guri et al., 2016). Regarding E. coli O157:H7, AE lesions are controlled by a pathogenicity island named locus of enterocyte effacement (LEE) through ler gene, which is regulated by qseA gene also implicated in quorum sensing (Medellin-Pena et al., 2007). The gene luxS, major regulator of quorum sensing and influencing the qseA and ler genes, is also involved in expression of genes encoding flagella and biofilm formation (fliC) or synthesis of shiga-toxin ( B2) (Kaper et al., 2004; Wood et al., 2006).

The aims of this work were therefore to study the growth potential of B. crudilactis FR/62/B/3 compared to B. bifidum BBA1 in culture media supplemented with whey or 3′ SL and to evaluate the effects of CFSM on virulence expression of Escherichia coli O157:H7 and Salmonella Typhimurium. Because of its bovine origin, the hypothesis is that B. crudilactis FR/62/B/3 can metabolize components present in whey, especially BMO, as explained previously. B. bifidum BBA1, a strain isolated from breastfed children feces, was chosen to provide a comparison with a strain of human origin as this strain should be able to use BMO or 3′ SL as a source of carbon, due to their similarity with HMO (Zivkovic and Barile, 2011). Tanimomo et al. (2016) developed a culture medium answering to the specificities of B. crudilactis FR/62/B/3. This formula, more suited to bifidobacteria from bovine origin, was used in this study as an optimized medium in which different sources of carbohydrates have been tested. The effects of CFSM obtained from theses cultures on intestinal pathogens virulence were firstly investigated using rapid-testing on a luminescent reporter mutants. Next, virulence gene expression was more deeply investigated on several virulence genes using RT-qPCR on wild pathogenic strains. A special attention was given to the controls (unfermented media). Indeed, some nutrients have previously been shown to have a repressive effect on virulence gene expression of E. coli O157:H7 (Delcenserie et al., 2012) and therefore, it was important to exclude that potential effect from the observed results with the fermented media.

# MATERIALS AND METHODS

#### Bacterial Strains and Growth Conditions

Bifidobacterium bifidum BBA1 was isolated from feces from a breast-fed child (CHU - Hôpital des Bruyères, Liège, Belgium)

stx and B. crudilactis FR/62/B/3 from Saint-Marcellin, a raw cow milk cheese from Vercors (France). Both strains were stored at −80◦C and grown on De Man, Rogosa, and Sharpe (MRS) medium (Oxoid, Hampshire, UK) supplemented with cysteine-HCl (0.5 g/l) and mupirocin (0.08 g/l) at 37◦C for 48 h in an anaerobic workstation (Led Techno, Heusden-Zolder, Belgium) containing 10% H2, 10% CO2, and 80% N2. Several successive cultures, in the same conditions as described previously, have been realized in MRS broth, prior to use. Pathogenic enterohaemorrhagic E. coli (EHEC) strain O157:H7 ATCC 35150 ( + 2 ) and S. enterica serovar Typhimurium strain ATCC 14028 were stored at −80◦C and grown in Luria Bertani (LB) media (Sigma-Aldrich, Diegem, Belgium). Two reporter mutants, E. coli O157:H7 ATCC 43888 ( <sup>−</sup>, LEE:lux) containing plasmid LEE1 luxCDABE and resistant to ampicillin (Amp<sup>r</sup> ) and kanamycin (Kan<sup>r</sup> ) and S. SA 941 256 containing plasmid pSB377 (hilA::luxCDABE; Amp<sup>r</sup> ) were designed by Medellin-Pena et al. (2007) and Bayoumi and Griffiths (2010), respectively. Both strains were from the Canadian Research Institute for Food Safety Collection and were grown under aerobic conditions at 37◦C in brain heart infusion (BHI) broth (Bio-Rad, Marnesla-coquette, France) supplemented with ampicillin (50 mg/l). A medium optimized for B. crudilactis FR/62/B/3, called MRS2 (Tanimomo et al., 2016) was considered as the reference medium for this study ( ) and was modified by removing or replacing glucose: MRS2 without any glucose (MRS2 G) (control), MRS2 with a mix of glucose and whey (MRS2-Wh) and MRS2 with 3′ SL (MRS2-3′ SL) as the only source of carbohydrate ( ). Whey was collected at the beginning of a curdling process of a Belgian cheese factory (Liège area, Belgium). The quantity of lactose in MRS2-Wh medium was estimated to 25 g/l, based on lactose concentration of sweet whey (50 g/l of lactose; Food and Agriculture Organization/Organisation Mondiale only traces of BMO (Kelly et al., 2013). The 3 SL, added ′ to MRS2-3′ SL, was provided by Carbosynth laboratory (Berkshire, UK). The concentration of 0.85 g/l was chosen to be close to natural concentrations found in colostrum (Nakamura **Table 1 Table 1** stx Typhimurium de la Santé, 1998). However, mature bovine milk contains

#### TABLE 1 | Composition of modified MRS2 media.


et al., 2003). B. bifidum BBA1 and B. crudilactis FR/62/B/3 were grown in three independent experiments under the same anaerobic conditions as previously at 37◦C for 48 h. Five log/ml of bifidobacteria from a fresh 48 h culture of bifidobacteria were inoculated into the fresh media (1% v/v). The concentration of 5 log/ml was confirmed by plating several dilutions of bifidobacteria at day 0 post inoculation. Bacterial growth was determined by viable counts after 48 h incubation. Cell free spent media (CFSM) were obtained after two centrifugation steps at 5000 (Eppendorf Centrifuge 5804, Hamburg, Germany) for 10 min. Supernatants were then sterilized by filtration (Minisart <sup>R</sup> 0.45µm and 0.2µm, Sartorius, Vilvoorde, Belgium). Next, CFSM were freeze-dried (Virtis Benchtop 3.3 EL, SP Scientific, Suffolk, United-Kingdom) and rehydrated with sterile distilled water to obtain a 10x concentration. The same treatment was applied to non-fermented culture media (controls). The pH of rehydrated CFSM was adjusted to 7 using 1 M NaOH. ×

## Measurement of LEE and hilA Promoter Activity

Both E. coli reporter strains were grown overnight in BHI broth supplemented with ampicillin (BHI-Amp). Each overnight culture was diluted 1:100 with fresh BHI-Amp broth supplemented (test samples) or not (control samples) with 10% concentrated CFSM obtained from fermented MRS2-3′ SL. Two hundred microliters of each sample were distributed into triplicate wells of a sterile, opaque 96-well microliter plate (Corning 3610, Fisher Scientific, Ottawa, Ontario, Canada) and incubated at 30◦C. Luminescence was measured every hour for 24 h, with a Victor multilabel counter (Wallac, PerkinElmer Life Sciences Canada, Woodbridge, Ontario, Canada). Luminescence was expressed in counts per second. Optical density (OD) was determined using a Genesys 20 spectrophotometer (Thermo scientific, Erembodegen, Belgium) adjusted to 600 nm.

# Contact between Concentrated CFSM and Wild Type Pathogenic Strains

E. coli O157:H7 ATCC 43890 was grown in LB agar and a single colony was taken from the plate and incubated overnight in LB broth at 37◦C with aeration. The same procedure was applied for S. Typhimurium ATCC 14028 using BHI medium. The cultures were homogenized and 50 µl were diluted in 4.5 ml of LB broth for E. coli and BHI broth for S. Typhimurium. Then, 450 µl of each concentrated CFSM was added to the bacterial suspensions. Triplicate cultures were incubated at 37◦C for 4 h on a shaker at 150 rpm. Bacterial growth was determined by OD measurement at 600 nm. E. coli O157:H7 and S. Typhimurium were grown in LB and BHI broth alone, respectively, as controls.

# Gene Expression Analysis by RT-qPCR

The method was adapted with some modifications from Tellez et al. (2012), Mith et al. (2014) and Guri et al. (2016). After 4 h of incubation (Delcenserie et al., 2012), cells were collected by centrifugation at 5000 × g for 10 min at room temperature (Eppendorf Centrifuge 5804, Hamburg, Germany). The pellet was suspended in 100 µl TE buffer (10 mM Tris and 1 mM EDTA) containing 1% lysozyme (Roche, Mannheim, Germany). Samples were stored at −20◦C overnight. On the next day, RNA was extracted using the RNeasy <sup>R</sup> Mini Kit (Qiagen, Antwerp, Belgium). DNA contamination was eliminated from each sample using the DNase I Recombinant RNase-free Kit (Roche Diagnostics GmbH, Mannheim, Germany). To inactivate the DNase, samples were heated at 75◦C for 10 min. The quantity of RNA was determined by measuring the absorbance at 260 nm using a Nanodrop 2000 Spectrophotometer (Thermo Scientific, USA). The purity and quality of RNA were verified by measuring the ratio of absorbance (260 nm/280 nm) and by using agarose gel electrophoresis (Eurogentec, Seraing, Belgium). The concentration of RNA used for reverse transcription was normalized to 100 ng/µl for E. coli and to 50 ng/µl for S. Typhimurium. Next, the RNA was subjected to reverse transcription polymerase chain reaction (RT-PCR) using a highcapacity cDNA Reverse Transcription Kit (Applied Biosystems, Ghent, Belgium). Briefly, 1 µg of RNA was reverse transcribed with 0.8 µl of desoxyribonucleoside triphosphate (dNTP; 100 mM), 1 µl of Multiscribe <sup>R</sup> Reverse Transcriptase (50 U/µl), 2 µl of 10X RT Random Primers and 2 µl of 10X RT Buffer in an adjusted total volume of 20 µl. For each sample, a no-RT control was included to confirm the absence of DNA contamination. Synthesis of cDNA was performed in a Mastercycler Gradient Thermocycler (Flexigene, Cambridge, UK) under the following conditions: 25◦C for 10 min, 37◦C for 120 min, 85◦C for 5 min and a cooling step at 4◦C. Then, cDNA was stored at −20◦C.

To study the effects of bioactive molecules present in culture supernatant on gene expression of E. coli O157:H7 ATCC 43890, the expression of genes ler (involved in attaching effacing lesions), fliC (involved in mobility), stxB2 (encoding subunit B of Shiga-toxin 2), luxS (major regulator of quorum sensing and producing AI-2), and qseA (involved in quorum sensing and regulator of LEE expression) was determined using qPCR. The same method was used to study effects on S. Typhimurium virulence gene expression of hilA (invasion protein regulator), ssrB2 (Type III secretion system regulator), and sopD (secreted effector protein). Quantitative PCR amplification was conducted using the GoTaq <sup>R</sup> qPCR Master Mix (Promega, Leiden, Netherlands) and using the ABI 7300 Real Time PCR System (Applied Biosystems, Singapore) for E. coli or the Light Cycler 480 (Roche Diagnostics, Mannheim, Germany) for S. Typhimurium. The primers were synthesized by Eurogentec (Liège, Belgium) and have been used in previous studies (**Table 2**). The RT-qPCR was performed in a total volume of 20 µl, containing 10 µl of GoTaq <sup>R</sup> Master Mix, 5.75 µl of molecular grade water, 1 µl of forward primer (10 µM), 1 µl of reverse primer (10 µM), 0.25 µl of carboxy-X-rhodamine (30 µM), and 2 µl of diluted cDNA. The qPCR conditions for E. coli were: initial denaturation at 95◦C for 3 min; denaturation, annealing and elongation repeated 45 times: 95◦C for 15 s, 58◦C for 30 s and 72◦C for 45 s; melting curve program: 60–95◦C with a heating rate of 0.1◦C/s. The qPCR conditions for S. Typhimurium were: denaturation at 95◦C for 10 min; 40 cycles of amplification and quantification: 95◦C for 30 s, 56◦C for 30 s and 72◦C for 30 s; melting curve program: 60–95◦C with a heating rate of 0.1◦C s. The annealing temperature, optimized at 56◦C, was determined experimentally. Each specific amplicon was validated for the presence of a single melting temperature peak and a single band of expected size on a 2% agarose gel after electrophoresis. Cycle threshold (Ct) values were determined using the ABI 7300 System SDS Software for E. coli and the Light Cycler Software 480 version 1.5 for S. Typhimurium. Four housekeeping genes were tested for E. coli: gnd (6-phosphogluconate deshydrogenase), gst (glutathione S-transferase), 16S gene (ribosomal RNA) and recA (recombinase A). Three housekeeping genes were tested for S. Typhimurium: gmk (guanylate kinase), rpoD (sigma factor) and 16S gene (ribosomal RNA gene). Because recA and gmk were the most stable under different treatments, they were selected for normalizing transcript expression levels. The experiments were replicated three independent times. To determine relative changes in gene expression, the formula described by Pfaffl (Pfaffl, 2001) was used: ratio = (Etarget) <sup>1</sup>Ct target(control-sample)/(Ereference) 1Ct reference(control-sample) ,

#### TABLE 2 | Primers of virulence genes used for qPCR (F: forward; R: reverse).


where E is the efficiency of the qPCR, calculated according to the equation: E = 10(−1/slope) .

# Statistical Analysis

The data are means ± standard error of three replicates. A Student's t-test was used to assess the statistical significance of the differences between test and control groups, where p ≤ 0.05 was considered as significant.

# RESULTS

### Growth of Bifidobacterium bifidum and Bifidobacterium crudilactis

B. crudilactis FR/62/B/3 showed increase in viable counts in MRS2, MRS2-Wh and MRS2-3′ SL compared to MRS2 G, but the highest counts were observed on MRS2-Wh (8.9 ± 0.6 log cfu/ml, **Table 3**). The same trend was observed for B. bifidum BBA1 with slightly lower counts (**Table 3**) compared to B. crudilactis FR/62/B/3. The highest counts were also observed for MRS2-Wh (8.1 ± 0.3 log cfu/ml).

## Activity of CFSM from MRS2-3′SL Fermented by Bifidobacteria on Bioluminescent Reporter Gene Expression

Luminescence expression of the plasmids LEE::luxCDABE and hilA:luxCDABE reached its maximum at 4 h for the E. coli mutant and at 13 h for the S. Typhimurium mutant, respectively. In the presence of supernatants from fermented MRS2-3′ SL medium, bioluminescence induction decreased for both mutants (**Figures 1**, **2**) showing a decrease in promoter expression of ler and hilA. These results, statistically significant for both strains, were more pronounced for hilA gene expression of S. Typhimurium (**Figure 2**).

### Effect of CFSM on E. coli O157:H7 Virulence Gene Expression

After incubation of 4 h, the different CFSM had no negative impact on growth. The average OD of E. coli O157:H7 at 600 nm after 4 h exposure was around 1.286 ± 0.119. From every tested housekeeping gene, recA was the most stable and was chosen to normalize the results according to the efficiency of each pair of primers (virulence genes ler, fliC, stxB2, luxS, and qseA) monitored using qPCR (Pfaffl, 2001; Tellez et al., 2012) (**Table 4**). The fermented MRS2 G, MRS2, and MRS2- Wh media did not show significant impact on gene expression

TABLE 3 | Counts of B. bifidum and B. crudilactis after 48 h of incubation in MRS2 G, MRS2, MRS2-Wh, and MRS2-3′SL media.


(RLU/OD600) and where OD is fixed at 0.806 and was taken after 4 h of incubation. LUM: luminescence; OD: optical density. \*\*\*P ≤ 0.005.

(RLU/OD600) and where OD is fixed at 0.909 and was taken after 13 h of incubation. LUM: luminescence; OD: optical density. \*\*\*P ≤ 0.005.

(**Figures 3A–C**) compared to non-fermented control media, meaning that CFSM had no effects. However, significant downregulation of virulence genes of E. coli O157:H7 was observed in the presence of fermented MRS2-3′ SL medium (**Figure 3D**). The medium fermented by B. bifidum BBA1 and B. crudilactis FR/62/B/3 induced a down-regulation of the E. coli ler gene (ratios of −15.4; P ≤ 0.01 and −8.1; P ≤ 0.05, respectively). A down-regulation of the qseA gene was also observed (ratios of −2.1; P ≤ 0.01 and −3.1; P ≤ 0.05, respectively). A nonsignificant trend for up-regulation of the fliC gene (ratios of 25.8 and +20.8, respectively) was noted while a nonsignificant trend for down-regulation of the stxB2 gene (ratios of −4.6 and −4.2, respectively) was observed. In the case of fermentation by B. crudilactis FR/62/B/3, a slight non-significant trend for down-regulation of the luxS gene was observed (ratio of −2.1).

# Effect of CFSM on S. Typhimurium Virulence Gene Expression

After incubation of 4 h, the OD measurements showed no negative impact on growth. The average OD of S. Typhimurium at 600 nm and after 4 h of exposure was 0.862 ± 0.078. From every tested housekeeping gene, gmk was the most stable and was chosen to normalize and adjust the results according to the efficiency of each pair of primers (virulence genes hilA, ssrB2, and sopD) monitored using qPCR (Guri et al., 2016, **Table 5**). The CFSM of MRS2-3′ SL medium fermented by B. bifidum induced a slight down-regulation of the ssrB2 gene (−2.1; P ≤ 0.05). The same trend was observed for the genes hilA and sopD (−2.5 and −1.9, respectively). MRS2-3′ SL CFSM fermented by B. crudilactis FR/62/B/3 did not show any significant effect on virulence gene expression (**Figure 4B**). A significant increase of sopD expression is observed but too light to be biologically meaningful (1.1; P ≤



0.05), and the same but non-significant trend is observed with hilA gene (1.6). Interestingly, a down-regulation of virulence genes was observed with CFSM obtained from MRS2 without glucose. In the case of fermentation by B. crudilactis FR/62/B/3, the genes hilA, ssrB2 and sopD were down regulated (−8.3, −10.9, and −6.2, respectively; P ≤ 0.05). The same but non-significant trend was observed for B. bifidum BBA1 (−8.5, −8.0, and −2.6, respectively, **Figure 4A**).

# DISCUSSION

B. crudilactis FR/62/B/3 presented the best growth potential compared to B. bifidum BBA1, particularly with whey or 3′ SL instead of glucose. This could be explained by the fact that B. crudilactis FR/62/B/3 was originally isolated from raw cow milk and raw milk cheese. This species possesses the genetic machinery suggesting some ability to synthetize specific enzymes for BMO degradation, as highlighted by the presence of genes coding for β-galactosidase and α- or β-glucosidase, genes also present in other bacteria such as B. bifidum, B. longum subsp. infantis, B. mongoliense, B. biavatii, B. kashiwanohense, and B. stellenboschense (Delcenserie et al., 2007; Milani et al., 2014, 2015; Bondue and Delcenserie, 2015). B. bifidum BBA1 grew as well on medium containing 3′ SL as the main carbohydrate source. This species possesses genes encoding some of the enzymes cleaving BMO bonds, so their expression would lead to growth. However, the growth results of bifidobacteria on this media were similar to those observed with MRS2 G (control). This could mean that those genes may not be expressed efficiently under the conditions tested, or another function is necessary. A next step would be to measure the expression of genes involved in carbohydrate metabolism such as genes coding for β-galactosidase and α- or β-glucosidase to validate the expression of those genes in the presence of BMO. Another hypothesis could be attributed to the presence of residual glucose from MRS culture allowing bifidobacteria to grow in medium exempt of carbohydrate. Indeed, the glucose present in initial MRS medium was in excess (Tanimomo et al., 2016), but a negligible part of it (maximum 1 mg) has been transferred with the inoculum. Another source of glucose could be the presence of residual carbohydrate in the yeast extracts or peptone extracts used in MRS media. The positive effects of media supplemented with milk products on growth of probiotics has been demonstrated previously (Champagne et al., 2014). This is confirmed in the present study as the best levels of growth were reached on MRS2-Wh media for both strains studied. In addition to BMO, whey is rich in lactose (Food and Agriculture Organization/Organisation Mondiale de la Santé, 1998), a carbohydrate source that is easily consumed by bifidobacteria (Delcenserie et al., 2007).

As demonstrated previously, the CFSM obtained from some lactic acid bacteria or bifidobacteria can induce a decrease in virulence gene expression of pathogenic bacteria such as C. jejuni (Mundi et al., 2013). Likewise, B. bifidum ATCC 29521, and L. acidophilus La-5 CFSM were able to produce metabolites inhibiting virulence gene expression of E. coli O157:H7 (Medellin-Pena et al., 2007) and S. Typhimurium

dE. coli O157:H7 grown in LB broth supplemented

eE. coli O157:H7 grown in LB broth supplemented

 with fermented concentrated

 with fermented concentrated

 culture media CFSM from B. bifidum for 4 h.

 culture media CFSM from B. crudilactis for 4 h.

(Bayoumi and Griffiths, 2012). In our study, CFSM collected from MRS2-3′ SL medium fermented by B. bifidum BBA1 and B. crudilactis FR/62/B/3 down-regulated most of the virulence genes tested in E. coli O157:H7, except the fliC gene, which tended to be up-regulated. This is not surprising according to the fact that fliC gene is not coded by the LEE operon and therefore not necessarily regulated as other virulence genes involved in T3SS and situated within the LEE operon (Falcao et al., 2004). No significant effect has been observed with CFSM from MRS2 and MRS2-Wh medium. In addition, CFSM obtained from bifidobacteria grown in media enriched in 3′ SL were able to affect virulence gene expression of E. coli O157:H7 without having any impact on its growth, at least during the first 4 h of incubation.

In those media, higher in carbohydrates, more fermentation products such as lactate or acetate are synthetized and could have an inhibiting effect on pathogenic bacteria, as well as acidifying the media. However, all CFSM were neutralized before testing them against E. coli or Salmonella, meaning that the pH did not exert any effect on E. coli O157:H7 growth. Furthermore, under neutral pH, the organic acids were under dissociated form and should not present any bactericidal or bacteriostatic action, contrary to un-dissociated forms (Momose and Hirayama, 2008).

The genes involved in virulence expression such as ler but also fliC genes are regulated by luxS, involved in quorum sensing. However, nutrients can interfere with quorum sensing mechanisms (Henke and Bassler, 2004; Kaper et al., 2004; Nakanishi et al., 2006; Mellies et al., 2007) and induce a decrease in virulence gene expression through a decrease in luxS expression. Delcenserie et al. (2012) previously demonstrated the effects of glucose in down-regulating virulence gene expression of E. coli O157:H7. The present study brought out similar observations with lactose instead of glucose (data not shown). The ler gene was the most affected by the presence of those carbohydrates and the effect was dose-dependent. Media used as controls and containing glucose or lactose (MRS2 and MRS2- Wh) down-regulated this gene but no effect was observed with medium containing mainly 3′ SL as a source of carbohydrate (MRS2-3′ SL).

To be able to metabolize 3′ SL, B. bifidum and B. crudilactis have to secrete sialidases through which NeuAc (sialic acid) can



<sup>a</sup>PCR efficiency: E = [(10(-1/slope) )/2] x 100%.

<sup>b</sup>S. Typhimurium grown in BHI broth for 4 h.

<sup>c</sup>S. Typhimurium grown in BHI broth supplemented with concentrated supernatants from culture media unfermented for 4 h.

<sup>d</sup>S. Typhimurium grown in BHI broth supplemented with concentrated supernatants from culture media fermented B. bifidum for 4 h.

<sup>e</sup>S. Typhimurium grown in BHI broth supplemented with concentrated supernatants from culture media fermented B. crudilactis for 4 h.

be produced. B. bifidum does not use this sialic acid, which is available for other bacteria such as B. breve (Milani et al., 2015). Therefore, if 3′ SL is metabolized by B. bifidum, free sialic acid was probably present in fermented CFSM from 3′ SL medium. Usually, pathogenic bacteria are able to bind this free sialic acid to their cell surface and use it to improve their resistance to the host's innate immune response, or can consume it as a nutrient (Vimr et al., 2004; Severi et al., 2007). NeuAc also exerts a major role in Salmonella enterica subsp. enterica serovar Typhi adhesion to intestinal epithelium (Sakarya et al., 2010). This means that in theory, sialic acid could have an impact on S. Typhimurium and E. coli growth, but no effect on growth as measured by OD has been observed. In addition, NeuAc had probably no impact on virulence gene expression in our study, when supplied in the medium.

The non-significant trend for up-regulation of fliC observed in our study should be clarified. Indeed, this trend seems higher when E. coli O157:H7 was exposed to CFSM from fermented 3 ′ SL. A hypothesis could be that the presence of residual complex carbohydrates affects gene expression of fliC. The presence or absence of some nutrients, or stress, could play a role in virulence expression, including fliC (Mei et al., 2015). Several studies investigated the effects of some stress (oxidative stress, heat shock, long storage) on down-regulation of fliC gene while other virulence genes were upregulated (Carey et al., 2009; Mei et al., 2015; Singh and Jiang, 2015). The experimental protocol of this study submitted E. coli O157:H7 to some stress, which may influence virulence gene expression. Genes involved in general stress (uspA and rpoS), in starvation (phoA and dpS), in cold shock (cspA, cspC, and cspE) and in acid resistance (gadW) have been investigated and the results have shown that the different treatments did not affect expression of stress-related genes (data not shown). This leads us to suppose that virulence factor expression has not been influenced by the experimental conditions.

The results observed using RT-qPCR with CFSM from fermented MRS2-3′ SL on S. Typhimurium virulence gene expression did not confirm the results observed with luminescent reporter strains. However, CFSM obtained after fermentation of MRS2 without any glucose fermented by B. crudilactis FR/62/B/3 down-regulated several virulence genes. Regarding B. bifidum BBA1 CFSM, a trend to down-regulation was observed as well. These down-regulations could be caused by other non-carbohydrate metabolites produced by bifidobacteria and these bioactive molecules could originate from the degradation of proteins. Most known bioactive molecules, such as subpeptin JM4-A and subpeptin JM4-B are antimicrobial peptides synthetized by Bacillus subtilis and active against Salmonella, Staphylococcus aureus, and Bacillus cereus (Sumi et al., 2015). Nisin, a bacteriocin well known, is produced by Lactococcus lactis and has a negative impact on Listeria or Clostridium (Ebbensgaard et al., 2015). The results of our study suggest that the CFSM activity is not due to an antimicrobial effect but due to an antivirulent effect.

In conclusion, this study provides the information that CFSM obtained from MRS2-3′ SL medium fermented by B. bifidum BBA1 and B. crudilactis FR/62/B/3 down-regulated LEE1 expression of the luminescent E. coli reporter strain and hilA expression of luminescent S. Typhimurium reporter strain. These results agree with the decreasing virulence gene expression of ler and qseA for E. coli, but not for S. Typhimurium. The contact between S. Typhimurium and CFSM from fermented MRS without glucose showed down-regulation of genes hilA, ssrB2, and sopD. According to this in vitro study, the antivirulent metabolites issuing from fermentation by bifidobacteria could have a negative impact on T3SS of both pathogens, decreasing expression of genes mainly implicated in this virulence mechanism (ler and qseA genes for E. coli O157:H7; hilA, ssrB2, and sopD genes for S. Typhimurium). The potential upregulation of fliC in E. coli O157:H7 could increase the motility as well as biofilm formation. A phenotypic analysis

## REFERENCES


of the pathogens under the experimental conditions could bring more insights about its virulence pattern. Information is lacking about the nature of the active molecules, but the activity of those CFSM might be due to small peptides or proteins with low molecular weight and resistant to pH modification and heat, or products obtained from carbohydrate metabolism. Size exclusion chromatography could contribute to separating and isolating these bioactive molecules in order to identify them. In the future, the effects of these metabolites will be investigated in a human gastrointestinal model to study the impact on microbiota to mimic in vivo conditions.

# AUTHOR CONTRIBUTIONS

PB did the experiments, interpreted the results and wrote the manuscript. SC and FB participated to the experiments. MS, GD, GL, and MG were involved in the design of the study and provided help for interpretation of the results. VD participated to the design of the study, interpretation of the results and writing of the manuscript.

#### ACKNOWLEDGMENTS

This study was funded by the "Fond Opportunité" from Ulg. The author wishes to acknowledge Nicolas Korda, Emilie Arnould, staff from the Bruyères hospital (Liège, Belgium) and Herve society (Liège, Belgium) for their valuable help on this project.


in infants. Appl. Microbiol. Biotechnol. 98, 563–577. doi: 10.1007/s00253-013- 5405-9


during the prepartum and early lactation. J. Dairy Sci. 86, 1315–1320. doi: 10.3168/jds.S0022-0302(03)73715-1


enterohaemorrhagic Escherichia coli (EHEC) in mice. J. Food Prot. 10, 1796–1805. doi: 10.4315/0362-028X.JFP-11-508

Zivkovic, A. M., and Barile, D. (2011). Bovine milk as a source of functional oligosaccharides for improving human health. Adv. Nutr. 2, 284–289. doi: 10.3945/an.111.000455

**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 Bondue, Crèvecoeur, Brose, Daube, Seghaye, Griffiths, LaPointe and Delcenserie. 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.

# Exploring Amino Acid Auxotrophy in *Bifidobacterium bifidum* PRL2010

*Chiara Ferrario1, Sabrina Duranti1, Christian Milani1, Leonardo Mancabelli1, Gabriele A. Lugli1, Francesca Turroni1, Marta Mangifesta2, Alice Viappiani2, Maria C. Ossiprandi3, Douwe van Sinderen4 and Marco Ventura1\**

*<sup>1</sup> Laboratory of Probiogenomics, Department of Life Sciences, University of Parma, Parma, Italy, <sup>2</sup> GenProbio Ltd., Parma, Italy, <sup>3</sup> Department of Medical-Veterinary Science, University of Parma, Parma, Italy, <sup>4</sup> APC Microbiome Institute and School of Microbiology, University College Cork, National University of Ireland, Cork, Ireland*

The acquisition and assimilation strategies followed by members of the infant gut microbiota to retrieve nitrogen from the gut lumen are still largely unknown. In particular, no information on these metabolic processes is available regarding bifidobacteria, which are among the first microbial colonizers of the human intestine. Here, evaluation of amino acid auxotrophy and prototrophy of *Bifidobacterium bifidum*, with particular emphasis on *B. bifidum* strain PRL2010 (LMG S-28692), revealed a putative auxotrophy for cysteine. In addition, we hypothesized that cysteine plays a role in the oxidative stress response in *B. bifidum*. The use of glutathione as an alternative reduced sulfur compound did not alleviate cysteine auxotrophy of this strain, though it was shown to stimulate expression of the genes involved in cysteine biosynthesis, reminiscent of oxidative stress response. When PRL2010 was grown on a medium containing complex substrates, such as whey proteins or casein hydrolysate, we noticed a distinct growth-promoting effect of these compounds. Transcriptional analysis involving *B. bifidum* PRL2010 cultivated on whey proteins or casein hydrolysate revealed that the biosynthetic pathways for cysteine and methionine are modulated by the presence of casein hydrolysate. Such findings support the notion that certain complex substrates may act as potential prebiotics for bifidobacteria in their ecological niche.

#### Keywords: bifidobacteria, chemically defined medium, genomics, microbiota

#### INTRODUCTION

The gut lumen contains a very complex mixture of compounds from alimentary and endogenous origins together with living microorganisms. The intestinal microbiota is metabolically active and plays a significant role in host physiology and metabolism (Hamer et al., 2012). The ability to metabolize peptides and amino acids is shared by a large number of bacteria ranging from saccharolytic bacteria to obligate amino acid fermenters present in gut microbiota (Davila et al., 2013). Peptides are the preferred substrates over free amino acids for many colonic bacteria, probably due to kinetic advantages of peptide uptake systems. Moreover, nitrogen source, such as amino acids, are fermented to short-chain fatty acids and organic acids, representing energy fuel for the colonic mucosa (Davila et al., 2013).

Milk proteins and peptides such as lactoferrin, lactoperoxidase, and lysozyme are reported to provide a non-immune defense against microbial infections (Schanbacher et al., 1997). In addition, they are known to stimulate growth of several members of the human infant microbiota such

#### *Edited by:*

*David Berry, University of Vienna, Austria*

#### *Reviewed by:*

*Thomas Clavel, Technische Universität München, Germany Clarissa Schwab, ETH Zurich, Switzerland*

> *\*Correspondence: Marco Ventura marco.ventura@unipr.it*

#### *Specialty section:*

*This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology*

*Received: 17 September 2015 Accepted: 12 November 2015 Published: 24 November 2015*

#### *Citation:*

*Ferrario C, Duranti S, Milani C, Mancabelli L, Lugli GA, Turroni F, Mangifesta M, Viappiani A, Ossiprandi MC, van Sinderen D and Ventura M (2015) Exploring Amino Acid Auxotrophy in Bifidobacterium bifidum PRL2010. Front. Microbiol. 6:1331. doi: 10.3389/fmicb.2015.01331*

**87**

*Lactobacillus* and *Bifidobacterium* (Liepke et al., 2002; McCann et al., 2006). In this latter ecological context, the bacterial population is dominated by bifidobacteria, which remain a prominent component of the gut microbiota until weaning (Turroni et al., 2012a; Duranti et al., 2015; Underwood et al., 2015). Member of the genus *Bifidobacterium* are anaerobic microorganisms, typically resident in the gastro intestinal tract of mammals and insects (Lugli et al., 2014), where they are known to interact with their hosts using various genetic strategies (O'Connell Motherway et al., 2011; Fanning et al., 2012; Ventura et al., 2012; Turroni et al., 2014).

Among host-derived nutrients, milk proteins significantly influence the composition of the gut microbiota, supplying these microorganisms with nitrogen and amino acids (Liepke et al., 2002). Enhancement of (bifido)bacterial growth is frequently associated with milk proteins and the peptides that arise from the hydrolysis of these proteins (Nagpal et al., 2011; Lonnerdal, 2013).

Compared to carbon metabolism, for which a large body of scientific data is available (Pokusaeva et al., 2011; Marcobal et al., 2013), only very limited knowledge is available on the acquisition and assimilation processes that are used by members of the infant gut microbiota to retrieve nitrogen from the gut lumen (Liepke et al., 2002). For Gram positive bacteria, nitrogen metabolism has been investigated in *Lactobacillus delbrueckii* subsp. *bulgaricus* (Liu et al., 2012), *Lactobacillus rhamnosus* (Lebeer et al., 2007) and *Bacillus* sp. (Fisher, 1999; Even et al., 2006).

Recently, specific interest has been directed toward sulfurcontaining amino acids and global control of cysteine and methionine metabolism in both Gram positive and negative bacteria, such as *Lactococcus lactis*, *Salmonella* sp., *Vibrio fischeri* and *Clostridium perfringens* (Fernandez et al., 2002; Andre et al., 2010; Alvarez et al., 2015; Singh et al., 2015). Cysteine biosynthesis is the key mechanism by which inorganic sulfur is reduced and incorporated into organic compounds (Kredich, 1992), where it plays an essential role in the formation of the catalytic sites of several enzymes, or protein folding and assembly via the formation of disulfide bonds (Mihara and Esaki, 2002). Sulfur-containing compounds that are used for the synthesis of cysteine and methionine are transported into the bacterial cell through different mechanisms: the first involves sulfate permease related to inorganic phosphate transporters (CysC) and then the reduction of sulfate to sulfide (Mansilla and de Mendoza, 2000) (**Figure 1**). The second involves aliphatic sulfonate ATP-binding cassette (ABC) transporters (SsuBD) (van der Ploeg et al., 1998) and the subsequent conversion into sulfide by an FMNH monooxygenase (**Figure 1**). The following reaction of sulfide with *O*-acetyl-L-serine (OAS) results in cysteine synthesis by the action of an *O*-acetylserine thiol-lyase (Bogicevic et al., 2012). Alternatively, cysteine can be directly transported inside the cell by symporter proteins (TcyBCP) (Burguiere et al., 2004). Methionine biosynthesis is closely linked to cysteine production by the action of serine acetyltransferase, which uses cysteine and an *O*-acetylhomoserine to generate cystathionine, where the latter compound is then converted to homocysteine and methionine (Fernandez et al., 2002) (**Figure 1**).

In this study, in order to understand the role of the bifidobacterial population in the utilization of nitrogen available in the human gut, we evaluated the amino acid metabolism of the infant stool isolate *B. bifidum* PRL2010 (LMG S-28692), a bifidobacterial prototype for analysis of interactions between microbes and the intestinal mucosa (Turroni et al., 2013), by coupling physiological data on a chemically defined medium (CDM) with transcriptional analysis. Specific emphasis was placed on sulfur amino acids/metabolism of PRL2010 since these amino acids are particularly important for the bacterial cells of gut commensals in coping against gut related stresses (e.g., oxidative stress) (Even et al., 2006).

Furthermore, PRL2010 metabolism of complex substrates from milk such as casein hydrolysate and whey proteins was investigated.

# MATERIALS AND METHODS

# Bacterial Strains and DNA Extraction

Bifidobacterial strains used in this study are reported in **Table 1**. Strains were grown anaerobically in de Man, Rogosa, Sharpe (MRS) medium (Scharlau, Spain), which was supplemented with 0.05% L-cysteine-HCl and incubated at 37◦C for 16 h. Anaerobic conditions were achieved by the use of an anaerobic cabinet (Ruskin), in which the atmosphere consisted of 10% CO2, 80% N2, and 10% H2.

# *Bifidobacterium* CDM Development

For amino acid auxotrophy and prototrophy tests, a CDM was employed based on a previously described formulation (Petry et al., 2000; Cronin et al., 2012) for *Lactobacillus* and *Bifidobacterium*, with some modifications. Briefly, to the already reported CDM, 50 mg/l of guanine and 4.0 mg/l of thiamine was added. Several simple sugars were screened including glucose, fructose, galactose, lactose, ribose, xylose, fucose, mannose, and rhamnose. All carbohydrates were added at 2% (w/w). The medium was sterilized by filtration (0.22 μm). When the CDM was prepared without amino acids it is termed basal CDM (bCDM). All components of CDM were purchased from Sigma (USA).

#### Amino Acid and Nitrogen Growth Assay

Cell growth on CDM was monitored by measuring the optical density of cultures at 600 nm (OD 600) using a plate reader (Biotek, Winooski, VT, USA). The plate reader was run in discontinuous mode, with absorbance readings performed after 24 h of incubation and preceded by 30 s of shaking at medium speed. Bacteria were cultivated in the wells of a 96-well microtiter plate, with each well containing a different amino acid, and incubated in an anaerobic cabinet.

For all growth tests, cells were recovered from an overnight MRS broth culture, centrifuged at 3000 rpm for 5 min in anaerobiosis, and washed with bCDM to remove protein and sugar residues. In each of the 96-wells of the microplate, 135 μl of medium was inoculated with 15 μl of washed cells diluted to OD 1.0 with bCDM, obtaining a final

inoculum OD of 0.1. Wells were covered with 30 μl of sterile mineral oil in order to maintain anaerobic conditions. Cultures were grown in biologically independent triplicates and the resulting growth data were expressed as the means from these replicates.

Once it had been established that CDM supports bifidobacterial growth, amino acid assays were performed using CDM in which individual amino acids were omitted or with bCDM supplemented with 0.2 g/l of a particular amino acid. To understand amino acid metabolism and in particular the metabolism of sulfur-containing amino acids derived from complex substrates, bCDM was supplemented with 2–0.5% (w/w) of whey protein or casein hydrolysate (Sigma). To test the influence of reduced sulfur substrate instead cysteine to PRL2010 growth, 5 mM of reduced glutathione (Sigma) was added to bCDM formulation without any other amino acid (bCDM + Glut). To evaluate the utilization of source of sulfur and nitrogen available in the gut environment, 0.5 g/l of taurine was added to bCDM.

## Identification of Genes Involved in Sulfur-containing Amino Acid Metabolism

The identification of genes involved in cysteine and methionine metabolism in PRL2010 and other *B. bifidum* strains was performed by using the BLASTP program (Gish and States, 1993). For the BLAST search, previously identified genes involved in sulfur metabolism of lactic acid bacteria were used (Liu et al., 2012). Twenty bp oligonucleotides for RT-qPCR experiments were manually designed on identified putative genes to obtain

#### TABLE 1 | Bifidobacterial strains used in this study.


*(Continued)*

#### TABLE 1 | Continued


*<sup>a</sup>ATCC, American Type Culture Collection, USA. LMG, Belgian Co-ordinated Collection of Microorganisms-Bacterial Collection, Belgium. DSM, German Collection of Microorganism and Cell Cultures, Germany. JCM, Japan Collection of Microorganisms, Japan.*

amplicons with a size ranging from 150 to 200 bp. Primers were checked with Primer Blast (Ye et al., 2012) and listed in **Table 2**.

#### RNA Isolation, Reverse Transcription and RT-qPCR

Total RNA was isolated from PRL2010 cultures grown in CDM, bCDM supplemented with cysteine, or bCDM supplemented with cysteine and whey protein or casein hydrolysate (2% w/w). PRL2010 cells grown in MRS was used as a control condition. Cultures were grown in biologically independent triplicates. Cells were harvested by centrifugation step at 4000 × *g* for 5 at 4◦C when cells had reached late exponential phase (OD values of 0.8– 1.0, except for bCDM supplemented with glutathione where cells were harvested at OD 0.35). Cell pellets were resuspended in 500 μl of RNAprotect reagent (Qiagen, UK) and mechanically lysed by inclusion of 0.1 mm zirconium–silica beads (Biospec Products, Bartlesville, OK, USA) and by subjecting the sample to three 2 min pulses at maximum speed in a bead beater (Biospec Products, Bartlesville, OK, USA) with intervals of 3 min on ice. RNA was extracted with the RNeasy mini kit (Qiagen) as reported in the manufacturer's instructions. Quality and integrity of the RNA was checked by Tape station 2200 (Agilent Technologies, USA) analysis and only samples displaying a RIN value above seven were used. RNA concentration and purity was then determined with a Picodrop microlitre Spectrophotometer (Picodrop). Reverse transcription to cDNA was performed with the iScript Select cDNA synthesis kit (Biorad) using the following thermal cycle: 5 min at 25◦C, 30 min at 42◦C, 10 min at 45◦C, 10 min at 50◦C and 5 min at 85◦C.

The mRNA expression levels of these genes were analyzed with SYBR green technology in quantitative real-time PCR (qRT-PCR) using SoFast EvaGreen Supermix (Biorad) on a Bio-Rad CFX96 system according to the manufacturer's instructions. Quantitative PCR was carried out according to the following cycle: initial hold at 96◦C for 30 s and then 40 cycles at 96◦C for 2 s and 60◦C for 5 s. Gene expression was normalized relative to a housekeeping genes as previously described (Turroni et al., 2011) and reported in **Table 2**. The amount of template cDNA used for each sample was 12.5 ng.



*aTurroni et al. (2011).*

#### Statistical Analyses

Statistical significance between means was analyzed using the two way ANOVA. Statistically different means were determined using the Bonferroni *post hoc* test at 5% (*P*-value *<* 0.05). Values are expressed as the means ± standard errors from three experiments. Statistical calculations were performed using the software program GraphPad Prism 5 (La Jolla, CA, USA).

#### RESULTS

# Development of a CDM for *B. bifidum* PRL2010

We modified the previously described CDM formulations (Petry et al., 2000; Cronin et al., 2012) based on the nutrient requirements of *B. bifidum* PRL2010. Several growth attempts on CDM minimal modifications (Petry et al., 2000; Cronin et al., 2012), i.e., where various compounds were omitted one after the other, allowed the identification of a number of components that were either essential or non-essential for growth of PRL2010 cells. Notably, folic acid and pyridoxal were eliminated from CDMPRL2010 composition, while guanine and thiamine were supplemented. When testing different sugars it was observed that PRL2010 exhibits the best growth performance with lactose, consistent with previous studies (Turroni et al., 2010, 2012b), and this sugar was therefore used for CDMPRL2010 formulation.

#### Evaluation of Amino Acids Auxotrophy and Prototrophy of PRL2010

When PRL2010 cells were cultivated on CDMPRL2010, they exhibited reduced growth (OD600 value of 1.21 ± 0.3) compared to that observed when grown on a nutrient-rich medium such as MRS (OD600 value of 2.9 ± 0.2). In order to assess PRL2010 amino acid prototrophy/auxotrophy, growth experiments were performed using CDMPRL2010 where an individual amino acid had been omitted at time, and bCDMPRL2010 medium with the inclusion of one amino acid at time. The achieved growth yield was compared to that obtained for complete CDMPRL2010 or bCDMPRL2010 respectively (**Figure 2A**).

In both experiments, only when cysteine was removed or supplied to the medium a significant decrease or increase (ranging from four- to sixfold, *P <* 0.05) of the obtained growth yield was observed, respectively, suggesting that PRL2010 is auxotrophic for cysteine. Furthermore, PRL2010 seems unable to grow on sulfate as its sole sulfur source, such as when this strain is grown in bCDMPRL2010 (a medium that contains MnSO4, MgSO4, and FeSO4).

Another reducing compound, glutathione, was added to bCDMPRL2010 and only very limited growth was detected when PRL2010 cells were cultivated for 24 h (OD600 value of 0.31 ± 0.038). Furthermore, we decided to investigate if taurine, which is a common nitrogen and sulfur sources present in the gut environment (Carbonero et al., 2012) influence the growth yields of PRL2010. However, we did achieved any significant grow (OD600 = 0.10 ± 0.01) of PRL2010 when taurine was used as the unique nitrogen and sulfur sources.

### Assessing Cysteine Auxotrophy of Members of the Genus *Bifidobacterium*

We further investigated the behavior of other strains belonging to the *B. bifidum* species (see **Table 1**), when cultivated under similar growth conditions (**Figure 2B**). These experiments showed that strains LMG11041, 156B, 85B, 324B, and LMG13195 were unable to grow on CDMPRL2010, (OD600 values ≤ 0.3) after 24 h of incubation (**Figure 2B**). The other *B. bifidum* strains investigated (i.e., LMG13200, LMG11582, and LMG11583) reached OD600 values of 0.7–0.9 and exhibited an identical auxotrophic behavior as PRL2010 for cysteine (**Figure 2B**).

Within the genus *Bifidobacterium*, the same auxotrophic behavior for cysteine appears to be widely distributed. In fact, of the currently recognized 48 (sub)species harboring the genus *Bifidobacterium*, only *B. boum* LMG10736, *B. minimum* LMG11592, *B. pullorum* LMG21816, *B. ruminantium* LMG21811, *B. saguini* DSM23967 and *B. scardovii* LMG21589 were shown to be able to grow in CDMPRL2010 without cysteine, though such strains reached OD600 values of just 0.5 ± 0.1 (**Figure 2C**). No bifidobacterial strain was able to grow in bCDMPRL2010.

#### FIGURE 2 | Continued

Growth of *B. bifidum* strains. Growth was measured as the optical density of the medium at 600 nm (OD600). Cultures were grown in triplicates. (A) Reports the growth of *B. bifidum* PRL2010 in CDMPRL2010. In these tests, one amino acid at time was removed (CDM-AA) or supplied (bCDM + AA) to the medium. Amino acids are reported in the horizontal axis as follows: aspartic acid Asp, threonine Thr, leucine Leu, tryptophan Trp, valine Val, histidine His, serine Ser, arginine Arg, isoleucine Iso, methionine Met, lysine Lys, cysteine Cys, proline Pro, glutamine Gln, alanine Ala, glycine Gly, phenylalanine Phe and tyrosine Tyr. (B) Shows an heat map representing the growth performance of all of the type strains of the currently recognized 48 (sub)species belonging to the genus *Bifidobacterium* on CDMPRL2010, CDM-CysPRL2010 and bCDMPRL2010. The different shading represents the optical density reached by the various cultures. (C) Displays the growth of *B. bifidum* strains LMG11041, 85B, 156B, 324B, LMG11583, LMG11582 LMG13200 and LMG13195 in comparison with PRL2010 in CDMPRL2010, CDMPRL2010 without cysteine (CDM – Cys), basal CDMPRL2010 (bCDM), basal CDMPRL2010 with cysteine (bCDM + Cys) and basal CDMPRL2010 with glutathione (bCDM + Glut). (D) Illustrates the growth of *B. bifidum* PRL2010 in CDMPRL2010 supplemented with complex substrates like whey proteins or casein hydrolysate. For both substrates two concentration were tested, 0.5 and 2% (wt/wt). For every concentration was evaluated the presence of amino acid (CDM or bCDM) or other nitrogen sources (N or w/o N).

# Sulfur Amino Acid Metabolism of *B. bifidum* PRL2010

A general prediction based on genomic data about nitrogen metabolism within the genus *Bifidobacterium* was previously reported by (Milani et al., 2014). The presence of genes involved in amino acid biosynthesis appears to be conserved among the seven phylogenetic groups of the genus *Bifidobacterium* (Lugli et al., 2014). However, the genes that are predicted to be involved in sulfur-containing amino acid metabolism were shown to be variably present within bifidobacterial genomes. In this context, an *in silico* analysis of the *B. bifidum* PRL2010 genome (Turroni et al., 2010) for putative genes involved in sulfur-containing amino acid transport did not reveal any positive match.

Aliphatic sulfonates can be used as alternative sulfur sources for the synthesis of cysteine (van der Ploeg et al., 1998). Bioinformatics analyses revealed the occurrence of two genes (BBPR\_0202 and BBPR\_0362) encoding two putative ABC-type permeases, in the chromosome of PRL2010. A low level of homology with genes involved in sulfonate transport (Even et al., 2006) was detected (Supplementary Table S1), possibly explaining why *B. bifidum* PRL2010 cells are unable to grow with sulfate as the only sulfur source (bCDM condition, see **Figure 2A**). Another mechanism to achieve sulfur from the environment is based on the intake of cysteine by symporter proteins. This type of symporter may participate in the uptake of cysteine (Vitreschak et al., 2008). In this context, a putative sodium dicarboxylate symporter gene (BBPR\_0324) was identified in PRL2010 (see Supplementary Table S1). Moreover, two putative genes (BBPR\_0668 and BBPR\_0671) predicted to encode two carriers involved in glutamate transport system (GluA and GluD), exhibited 53 and 26% homology, respectively, with the genes that encode the L-cysteine uptake system of *B. subtilis* (Supplementary Table S1).

As mentioned above, *B. bifidum* PRL2010 cells were shown to be unable to grow in presence of reduced glutathione (and in the absence of cysteine). Such physiological findings are in agreement with *in silico* analyses of PRL2010 chromosome sequences. In fact, the *pepT* and *pepM* genes, which are constituting the pathway for degradation of this compound (Andre et al., 2010) to generate cysteine, are absent in PRL2010 genome. Furthermore, a homolog of the *gshAB* gene, which specifies the glutamate–cysteine ligase/glutathione synthase, is also absent in chromosome of PRL2010 (**Figure 1**).

Genes predicted to be involved in the cysteine biosynthesis I/homocysteine degradation pathway and methionine biosynthesis I pathway were identified in PRL2010 (**Figure 3A**). *In silico* analyses of PRL2010 genome revealed the occurrence of the *cysE* (BBPR\_1340) and *cysK* (BBPR\_ 1344) genes, which encode the predicted serine acetyltransferase that transfers an acetyl group to serine, and the cysteine synthase, respectively (Liu et al., 2012) (**Figure 3A**). In the same genomic region, we also identified the *metB* gene (BBPR\_1343) predicted to encode a cystathionine-γ-synthase, which is catalyzing the conversion of cysteine to cystathionine, as well as the *luxS* gene (BBPR\_1341), encoding an *S*-ribosylhomocysteinase involved in the production of homocysteine, and the *recQ* gene (BBPR\_1342) encoding an ATP-dependent DNA helicase. When the presence of these genes was investigated in the genomes of other *B. bifidum* strains (Duranti et al., 2015) included in this study, a high level of homology (higher than 98% at nucleotide level) was found. Furthermore, in the genome sequences of four *B. bifidum* strains, i.e., LMG13200, LMG13195, LMG11582, and LMG11583 (Duranti et al., 2015), an additional acetyltransferase-encoding gene was identified (**Figure 3A**).

Other genes such as the *cysB* gene (BBPR\_0960), *metC* (BBPR\_1226) and *metA* (BBPR\_1654) that are predicted to be involved in cysteine and methionine metabolism (Fernandez et al., 2002; Liu et al., 2012) are scattered across the PRL2010 genome.

# Growth Evaluation in Complex Substrates

The effects of complex substrates, such as whey proteins or casein hydrolysate, on PRL2010 growth were tested and are reported in **Figure 2D**. Whey proteins and casein hydrolysate were dissolved in CDMPRL2010 and bCDMPRL2010 with or without other nitrogen sources at 0.5 or 2% concentration (wt/wt), respectively. Casein hydrolysate better supports PRL2010 growth in presence of nitrogen, compared to what was observed when this strain was cultivated on whey proteins. In CDM or bCDM without other nitrogen sources, PRL2010 cells seemed to metabolize whey proteins more efficiently as displayed by the higher OD 600 values that were reached (**Figure 2D**).

# Targeted Gene Expression Analyses of PRL2010 with Different Sulfur Substrate

Transcription of genes involved in sulfur metabolism, such as those of cysteine (*cysE, cysK* and *cysB*) and methionine metabolism (*met*A, *metE, metB,* and *metC*), were investigated

statistically significant differences compared to the control. The error bar for each column represent the standard deviation calculated from three replicates.

using a qRT-PCR approach, the results of which are reported in **Figure 3B**.

When PRL2010 cells were cultivated in the complete CDMPRL2010, *cysB, cysE, cysK, metA* and *metB* were overexpressed (*<sup>P</sup> <sup>&</sup>lt;* 0.05) (**Figure 3B**). The occurrence of cysteine in the basal CDMPRL2010 (bCDM + Cys) does not seem to modulate expression of genes involved in sulfur amino acid metabolism. Glutathione (bCDM + Glut) does not allow a significant growth of PRL2010 (OD600 values of 0.31 ± 0.038). However, it enhanced the transcription of the *cysB*, *cysE*, *cysK, metA* and *metB* genes (*P <* 0.05).

Regarding complex substrates, *cys* genes appear to be less induced when PRL2010 cells are cultivated in whey protein compared to basal CDMPRL2010 in the presence of casein hydrolysate (bCDM + Whey and bCDM + Casein respectively in **Figure 3B**), although at significant level (*<sup>P</sup> <sup>&</sup>lt;* 0.05). Moreover, casein hydrolysate significantly increases the transcription level of *metC*, the cystathionine β-liase (*P <* 0.05).

In all conditions tested, no significant transcriptional changes were detected for the *metC* and *metE* genes predicted to encode for cystathionine β-liase and homocysteine methyltransferase, respectively, except for *metC* when PRL2010 was grown in bCDMPRL2010 with casein hydrolysate.

#### DISCUSSION

Following birth, the human intestine is rapidly colonized by a vast array of microorganisms. *Bifidobacterium*, and in particular, *B. bifidum* strains are abundant in breast-fed infants, due to their capacity to grow on mucin and on human milk oligosaccharides (Turroni et al., 2010). The bifidogenic effect of breast milk due to bioactive peptides presence is well known (Liepke et al., 2002). A similar bifidogenic effect was shown for milk-derived k-caseins with loss of activity when the disulfide bonds were oxidized (Poch and Bezkorovainy, 1991).

Here, we investigated for the first time sulfur-containing amino acid metabolism in *B. bifidum* PRL2010 through the development of a CDM called CDMPRL2010 and by the molecular characterization of the putative auxotrophic behavior of this strain for cysteine. Data indicates that bCDMPRL2010 does not support *B. bifidum* PRL2010 growth, unless cysteine addition. The same behavior was extended to three other *B. bifidum* strains, i.e., LMG13200, LMG11582 and LMG11583. Furthermore, cysteine auxotrophy is not a common feature of all the (sub)species harboring the genus *Bifidobacterium*, since representatives of some species such as *B. boum, B. minimum, B. pullorum, B. ruminantium, B. saguini* and *B. scardovii* are able to grow without cysteine, although rather poorly.

*In silico* analyses of PRL2010 genome did not reveal the presence of the genetic arsenal needed to sulfate transport and reduction to sulfide. Growth experiments showed that cysteine is the only amino acid necessary to sustain PRL2010 growth but when the strain is cultivated in basal CDMPRL2010 with cysteine (bCDM + Cys) the transcription of genes involved in cysteine and methionine metabolism was not stimulated by the availability of these amino acid residues. Similar results were reported previously for other bacterial species, such as *Escherichia coli* (Kredich, 1992), *Bacillus subtilis* (Mansilla and de Mendoza, 2000), and *Lactococcus lactis* (Fernandez et al., 2002). Another reduced sulfur compound was used to understand if the role of cysteine in PRL2010 is linked to the reducing effect that it exploits on the redox potential (Even et al., 2006). However, reduced glutathione does not sustain any appreciable strain growth, yet enhanced the transcription of genes predicted to be involved in sulfur amino acid metabolism (*cysB*, *cysE, cysK, metA* and *metB)*. Similar behavior was previously reported for *E. coli* (Kredich, 1992) and *B. subtilis* (Mansilla and de Mendoza, 2000).

Complex substrates from dairy industry such as whey proteins and casein hydrolysate act as a reservoir of amino acid, peptides and free protein. Transcriptional analysis showed that whey proteins and casein hydrolysate increased the transcriptions of genes involved in serine degradation and/or conversion to cysteine and methionine (*cysB*, *cysE, cysK metA* and *metB*).

# CONCLUSION

This study provides new insights into the amino acid utilization ability of the *B. bifidum* species. This work also suggested the existence of a relationship between the sulfur amino acid metabolism and the redox state of the cell. The use of complex nitrogen sources available in the infant gut revealed an enhancement of growth yield and expression of genes involved in sulfur amino acid metabolism in PRL2010. These results could open a new avenue of research for the development of novel functional foods based on milk caseins and whey proteins with high content of cysteine or cysteine precursor's compounds that could act as prebiotics for *Bifidobacterium* enrichment.

#### AUTHOR CONTRIBUTIONS

CF performed the work and wrote the manuscript, SD performed the work, CM, performed bioinformatics analyses, LM performed bioinformatics analyses, GL performed bioinformatics analyses, MM performed the work, AV performed the work, MO contributed data, DvS wrote the manuscript, MV wrote the manuscript.

#### ACKNOWLEDGMENTS

We thank Science Foundation Ireland for financial support to DvS. This study was funded by the Irish Government's National Development Plan (Grant number SFI/12/RC/2273) to DvS.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal*.*frontiersin*.*org/article/10*.*3389/fmicb*.* 2015*.*01331

#### REFERENCES


*Bifidobacterium* by comparative genomics. *Appl. Environ. Microbiol.* 80, 6383– 6394. doi: 10.1128/AEM.02004-14


**Conflict of Interest Statement:** The Guest Associate Editor David Berry declares that despite having hosted a Frontiers Research Topic with the authors Marco Ventura and Francesca Turroni, the review process was handled objectively. 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 Ferrario, Duranti, Milani, Mancabelli, Lugli, Turroni, Mangifesta, Viappiani, Ossiprandi, van Sinderen and Ventura. 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.*

# Hypocholesterolemic and Prebiotic Effects of a Whole-Grain Oat-Based Granola Breakfast Cereal in a Cardio-Metabolic "At Risk" Population

Michael L. Connolly 1, <sup>2</sup> , Xenofon Tzounis 1, 2, Kieran M. Tuohy 1, 3 \* and Julie A. Lovegrove1, 2 \*

*<sup>1</sup> Hugh Sinclair Unit of Human Nutrition, University of Reading, Reading, UK, <sup>2</sup> Institute for Cardiovascular and Metabolic Research, University of Reading, Reading, UK, <sup>3</sup> Department of Food Quality and Nutrition, Research and Innovation Centre - Fondazione Edmund Mach, Trento, Italy*

#### Edited by:

*Francesca Turroni, University College Cork, Ireland*

#### Reviewed by:

*Marco Ventura, University of Parma, Italy Francesca Scazzina, University of Parma, Italy*

#### \*Correspondence:

*Kieran M. Tuohy k.m.tuohy@fmach.it Julie A. Lovegrove j.a.lovegrove@reading.ac.uk*

#### Specialty section:

*This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology*

Received: *06 May 2016* Accepted: *06 October 2016* Published: *07 November 2016*

#### Citation:

*Connolly ML, Tzounis X, Tuohy KM and Lovegrove JA (2016) Hypocholesterolemic and Prebiotic Effects of a Whole-Grain Oat-Based Granola Breakfast Cereal in a Cardio-Metabolic "At Risk" Population. Front. Microbiol. 7:1675. doi: 10.3389/fmicb.2016.01675* Meta-analyses of randomized controlled trials (RTC) have confirmed the hypocholesterolaemic effect of oats and oat based fibers. However, the mechanisms by which oats or oat fractions lower cholesterol is not totally clear. Recognizing the important role of the gut microbiome in metabolism and metabolic disease risk, we examined the impact of whole grain oat Granola (WGO) on the human gut microbiota and cardio-metabolic risk factors using a randomized crossover dietary intervention in at risk individuals (ClinicalTrials.gov Identifier: NCT01925365). We randomized 32 individuals at risk of developing cardio-metabolic disease by virtue of mild hypercholesterolaemia or glucose intolerance, into two groups consuming either 45 g of WGO or non-whole grain (NWG) breakfast cereals daily for two 6-week intervention periods separated by a 4-week wash out period in a randomized, controlled, crossover, double-blinded design. Confirming the cholesterol lowering effect of WGO, we observed a significant time by treatment interaction, for total cholesterol (TC) (*P* = 0.0001) and LDL-cholesterol (LDL-C) (*P* = 0.02) compared to NWG. A significant time by treatment interaction was also observed for the relative abundance of fecal bifidobacteria (*P* = 0.0001), lactobacilli (*P* = 0.001) and total bacterial count (*P* = 0.008), which were all elevated after consumption of WGO. Daily consumption of WGO resulted in a prebiotic effect on the human gut microbiota composition and significant reductions in TC and LDL-C concentrations. Prebiotic modulation of the human gut microbiota may thus constitute a previously unrecognized mechanism contributing to the hypocholesterolaemic effects of whole grain oat Granola.

Keywords: whole-grain oat granola, prebiotic, cholesterol, cardiovascular risk, Bifidobacterium

# INTRODUCTION

Several large epidemiological studies and a number of meta-analyses of nutritional interventions have reported a positive association between increased whole grain intake and reduced risk of developing a range of chronic diseases (Chatenoud et al., 1998; Jacobs et al., 1999; Montonen et al., 2003; Mellen et al., 2008; He et al., 2010; Ye et al., 2012). Consumption of oats or oat based products by individuals with various metabolic disease risk factors (e.g., hypercholesterolemia, obesity, and diabetes) and in different ethnic groups, has been shown to mediate an appreciable normalization of plasma cholesterol levels (Ripsin et al., 1992; Queenan et al., 2007; Wolever et al., 2011; Charlton et al., 2012; Zhang et al., 2012). The cholesterol lowering activity of oats is usually attributed to its ability to reduce intestinal absorption of cholesterol and/or inhibit the enterohepatic circulation of bile acids by increasing carriage of cholesterol and/or bile acids into the colon and facilitating their excretion in feces (Ryan et al., 2007; Gunness and Gidley, 2010; Borneo and León, 2012). Whole grain oats contain a number of potentially bioactive components capable of modulating cholesterol metabolism in mammals, including unsaturated fatty acids, fibers, such as betaglucan, arabinoxylans, arabinogalactans, and resistant starch. Some of these polysaccharides can form viscous gels in aqueous solutions, and/or directly bind cholesterol or bile acids, while all are fermentable by the gut microbiota into short chain fatty acids (SCFA). Oats, and whole grain oat Granola also contains polyphenolic compounds and phytoeostrogens which may also modulate the gut microbiota and impact on host metabolic parameters (Ryan et al., 2007; Borneo and León, 2012). However, it is the gel forming nature of beta-glucans which is most commonly attributed to the cholesterol lowering effect of oats. Tiwari and Cummins (2011) performed a metaanalysis on 126 studies and concluded that there was a significant dose-dependent inverse relationship between beta-glucan from oats and barley and blood total cholesterol, LDL-cholesterol, and triacylglycerol concentrations with 3 g/day being sufficient to lower blood total cholesterol by −0.30 mmol/L (Tiwari and Cummins, 2011). While the ability of oats and beta-glucans to increase excretion of bile acids and cholesterol in feces appears to be well established, the underlying mechanism still remains to be fully elucidated. Although the gel forming nature of betaglucans reducing bile acid/cholesterol absorption is the most commonly proposed mechanism, recent studies with probiotic microorganisms raise the possibility of bile salt hydrolase (BSH) activity as another possible mechanism by which oats and fermentable fibers can lower plasma cholesterol (Gunness and Gidley, 2010; Ooi et al., 2010; Ejtahed et al., 2011; Jones et al., 2012a,b). Studies on non-gel forming prebiotic fibers, which modulate gut bacteria and increase BSH active lactobacilli and bifidobacteria, support the hypothesis that an increased BSH activity due to gut bacterial modulation could reduce plasma cholesterol (Tanaka et al., 1999; Kim et al., 2004). Yet few studies have examined the impact of whole grain oat based food products on human gut microbiota and no study to date has measured the ability of dietary oats to modulate the composition and relative abundance of commensal bacteria in vivo.

The present study aimed to address this knowledge gap by determining the effectiveness of a whole grain oat (WGO) Granola breakfast cereal, compared to a refined breakfast cereal to beneficially modulate gut microbiota and its metabolic output, plasma lipids, gut satiety hormones and inflammation markers in a randomized, controlled, double-blind, crossover dietary intervention study. It tested the hypothesis that whole grain oat based cereal can mediate a prebiotic modulation of the human gut microbiota, particularly increased bifidobacteria abundance and concomitant reduction of plasma LDL cholesterol.

### MATERIALS AND METHODS

#### Subjects

Men and women (age range 23–64 year), recruited from the general population, attended the Hugh Sinclair Unit of Human Nutrition, at the University of Reading in a fasted state for measurement of height, weight, plasma glucose, total cholesterol concentrations, and hematology. To qualify for entry into the study subjects required a BMI of 18–30 kg/m<sup>2</sup> and either glucose intolerance (fasting glucose > 5.5 but < 6.9 mmol/L) or mild to moderate hypercholesterolaemic (total cholesterol > 5.2 but < 7.8 mmol/L). Exclusion criteria for the study were as follows: medical history of heart disease, diabetes mellitus, cancer, pancreatitis or renal disease, use of lipid lowering drugs, systemic corticosteroids or drugs for regulating hemostasis, exposure to any investigational agent < 42 d before the study; presence of gastrointestinal disorder or use of a drug likely to alter gastrointestinal motility or nutrient absorption, a history of substance misuse or alcoholism, a current pregnancy, planned pregnancy, or given birth in the past 12 months, antibiotic treatment 6 weeks previous to study start date, an allergy or intolerance to intervention breakfast cereals components or smoking. The study was conducted according to the guidelines laid down in the Declaration of Helsinki and was given a favorable ethical opinion for conduct by the University of Reading's Research Ethics Committee. All participants gave written informed consent before participation. The study was registered as a clinical trial on ClinicalTrial.gov (NCT01925365).

#### Study Design

The dietary intervention study was a randomized, double blind, controlled, crossover design. Thirty-two volunteers (20 women, 12 men) were recruited onto the study. For a 2-week period prior to dietary intervention, volunteers followed their habitual diet but were required not to consume confirmed prebiotics (such as inulin), probiotics (such as live yogurts or fermented milk drinks), and whole grain products. The subjects were randomly allocated into one of two groups using a random number generator. One group consumed the WGO Granola breakfast cereal (45 g/day) for 6 weeks, and then after a 4-week washout period, consumed the NWG breakfast cereal (45 g/day) for 6 weeks. The other group received the breakfast cereals in the opposite order. During each washout period no breakfast cereal was consumed. All test materials were packaged, labeled and randomized by Jordans Cereals (Biggleswade., UK) prior to the study, neither the investigators nor study subjects were aware of which cereal was allocated.

Study subjects were asked to keep diaries while ingesting the breakfast cereals, to record stool frequency and consistency (constipation, hard, formed, soft, or diarrhea), abdominal pain, intestinal bloating and flatulence (none, mild, moderate, or severe) on a daily basis. Any concomitant medication and adverse events were recorded. Faecal samples, saliva samples, and 20 ml fasting venous blood samples were collected from each volunteer at six time points before and after each treatment and 14 days after each of the treatment periods (i.e., 0, 42, 56, 70, 112, and 126 days).

### Composition of Breakfast Cereals

The nutritional composition of WGO and NWG breakfast cereals were provided by Jordans Cereals (Biggleswade., UK) except for the β-glucan content which was analyzed by Leatherhead Food Research (using the Megazyme kit). The energy content of the WGO and NWG cereals was 417 kcal/100 g and 380 kcal/100 g, respectively. WGO, in the form of Granola breakfast cereal, comprised mainly of whole grain oats with small amounts of almond and dried fruit, and contained (per 100 g) 67.8 g carbohydrates, of which 26.8 g sugars; 8.3 g protein, 12.4 g fat, 6.3 g fiber and 2.9 g β-glucan. NWG, in the form of flaked corn cereal, contained (per 100 g) 84.4 g carbohydrate, of which 8.8 g sugars; 7.4 g protein, 1.1 g fat, 3.0 g fiber and no detectable β-glucan.

# Culture Independent Enumeration of Fecal Bacteria

Fecal bacteria enumeration was the primary outcome measure. Fecal samples were stored in an anaerobic cabinet (10% H2; 10% CO2; 80% N2) for no longer than 2 h prior to processing. Changes in fecal bacteria populations upon consumption of the test and control cereals were monitored using 16S rRNA probes labeled with Cy3 for specific bacterial groups or the nucleic acid stain DAPI for total bacterial counts and Fluorescence in situ Hybridisation (FISH). The bacterial groups selected for enumeration were Bifidobacterium spp., Bacteroides/Prevotella spp., Lactobacillus/ Enterococcus spp., Clostridium coccoides-Eubacterium rectale group, Clostridium histolyticum group, and Atopobium cluster including most Coriobacteriaceae species, using the specific probes Bif164, Bac303, Lab158, Erec482, His150, and Ato291, respectively (Langendijk et al., 1995; Manz et al., 1996; Franks et al., 1998; Harmsen et al., 2000). FISH was performed essentially as described by Rycroft et al. (2001) and Daims et al. (2005). Briefly, fecal samples (375 µl) fixed in 4% paraformaldehyde (pH 7.2) overnight at 4◦C were then centrifuged at 1500 × g for 5 min, washed twice with phosphate buffer saline (PBS 0.1 M, pH 7.0), re-suspended in a mixture of PBS/99% ethanol (1:1 v/v) and stored at −20◦C for up to 3 months.

For the hybridisations, 20 µl of each sample was pipetted onto Teflon- and poly-l-lysine-coated, six-well (10 mm diameter each) slides (Tekdon Inc., Myakka City, FL, USA). Slides were dried at 46◦C for 15 min and then submerged in a series of ethanol solutions (50, 80, and 96%, 3 min each). This process was used for all samples, except those where the Lab158 probe was used. Sample slides probed with Lab158 were subjected to an additional step with 50 µl of lysozyme (1 mg/mL in 100 mM Tris-HCl, pH 8.0) at 37◦C for 15 min prior to being submerged in the ethanol solutions. A probe/hybridization buffer mixture (5 µl of a 50 ng µl −1 stock of probe plus 45 µl of hybridization buffer) was applied to the surface of each well. Hybridisations were performed for 4 h in an ISO20 oven (Grant Boekel). Slides were stored in the dark at 4◦C (for a maximum of 3 days) until cells were counted. Slides were enumerated using a Nikon E400 Eclipse microscope fitted with an EPI-fluorescence attachment, 15 randomized views were counted for each sample.

# SCFA Analysis

High-performance liquid chromatography (HPLC) was performed to determine fecal SCFA concentration. Aliquots (1 ml) of 10% (w/v) fecal suspension were centrifuged at 13000 g for 10 min and the supernatant was stored at −20◦C for up to 3 months. Supernatants were then filtered using 0.2 mm polycarbonate syringe filters (Whatman, Maidstone, Kent, UK) and injected (20µl) into an HPLC system (Merck, Whitehouse Station, NJ, USA) equipped with refractive index detection. For the preparation of the external standard containing the SCFA; acetic, propionic and butyric acid were added to give a final concentration of 25 mM to HCl (6 M) and HPLC gradient water. Dilutions of the external standards were prepared and added to the internal standard (ratio 4:1) to give a final concentration for the internal standard of 20 mM ethyl butyric acid, and a final concentration of external standards as 80, 40, 20, 10, 5.0, 1.0, and 0.5 mM. The column used was an ion-exclusion REZEX-ROA organic acid column (Phenomenex, Inc., Torrance, CA, USA) maintained at 85◦C. H2SO<sup>4</sup> in HPLC-grade H2O (0.0025 mmol/l) was used as the eluent, and the flow rate was maintained at 0.5 ml/min. Quantification of the samples was obtained through calibration curves of acetic, propionic and butyric acids.

#### Blood Samples Collection and Analysis

Blood samples were collected into a 10 ml EDTA tube (BD vacutainer EDTA tube, BD, Cowley, Oxon., UK) for the analysis of fasting total cholesterol (TC), HDL-cholesterol (HDL-C), LDL-cholesterol (LDL-C), triacylglycerol (TAG), CRP, IL-6, TNF-α, PYY, GLP-1 and insulin concentrations; into a 10 ml fluoride/oxalate tube (BD vacutainer fluoride/oxalate tube) for the analysis of fasting glucose concentration. These blood analytes were the secondary outcome measures. Following collection all blood samples were kept on ice until centrifugation. Plasma samples were recovered by centrifugation at 1700 g for 10 min, dispensed into 1.5 ml microcentrifuge tubes and frozen at −20◦C within 1 h from collection. Plasma samples were defrosted and centrifuged for 5 min at 1500 g prior to analysis.

Plasma TAG, glucose, CRP, TC, LDL-C, and HDL-C concentrations were determined on a Monarch Automatic Analyzer ILab 600 using enzymatic kits (Instrumentation Laboratories Ltd, Cheshire, UK). Two quality control samples, (Wako Control Serum I and II. Alpha Laboratories Ltd, Eastleigh, Hants., UK), containing known normal and abnormal concentrations were included at the beginning and end of each batch analysis. Results were accepted if the quality control values were within the range specified by the manufacturers.

Determination of plasma insulin (DAKO Diagnostic Ltd, Cambridgeshire, UK), IL-6 and TNF-α (R&D Systems, Abingdon, UK), PYY and GLP-1 (Yanaihara Institute Inc. Shizuoka, Japan), feacal and saliva sIgA (Immundianostik AG, Bensheim, Germany) calprotectin (NovaTec Immundiagnostica GmbH, Dietzenbach, Germany) concentrations were performed using specific commercial ELISA kits according to manufacturer's instructions. Concentrations of the samples and quality controls were determined automatically by reading from the standard curve using MasterPlex (version EX 2010) computer software.

#### Dietary Analysis

Volunteers were asked on week 5 of each of the intervention arms to complete a 3-day estimated diet diary (2 week and one weekend day). The volunteers received both verbal and written instructions on how to complete the diet diaries. These were analyzed using Dietplan 6.60 (Forestfield Software Ltd), to determine the macro and micronutrient content of the participant's diets during each intervention period.

#### Statistical Analysis

Normality testing was carried out to determine if data was normally distributed and transformed if required. Linear mixed model analysis of variance was performed for testing effects of visit, treatment, and visit by treatment interaction. Tukey's posttest with 95% confidence was used for multiple comparisons after a visit by treatment interaction was established. Minitab 16 for Windows was used for all analysis. P < 0.05 was taken as significant. The sample size of 28 was required to detect 0.5 Log<sup>10</sup> change in bifidobacterial counts with power set at 0.9, and a significance level of 0.05 based on previous prebiotic studies in healthy volunteers conducted with the same microbiological techniques investigating whole grain maize (Carvalho-Wells et al., 2010). To allow for a 10% dropout rate a total of 32 individuals were recruited.

#### RESULTS

## Subjects Characteristics

Of the 32 participants recruited, one individual withdrew before the start of the study due to antibiotic treatment, and another withdrew after visit 1 due to personal reasons. Samples from these individuals were excluded from analysis. Thus, a total of 30 participants were included (women n = 19, men n = 11). The mean age was 42 years with a range between 19 to 60 years. The participant's baseline mean weight was 75.2 kg (SD ± 20.4), BMI 26.4 (± 5.7), were mildly hypercholesterolamic TC 5.4 mmol/L (± 1.0) or glucose intolerant (glucose 5.6± 0.6 mmol/L).

#### Dietary Analysis

Nutrient analysis of the diets during the two intervention periods is shown in **Table 1**. The nutrient intakes in both intervention periods were similar except for total fat as % of total food energy (%E) (P = 0.009) and MUFA (%E) (P = 0.002) which were higher, and starch (%E) (P = 0.007), which was lower while participants consumed the WGO compared with the NWG cereal. These differences broadly reflected compliance to the intervention foods.

#### Faecal Bacteria and SCFA Analysis

The data for the fecal SCFA analysis is shown in **Table 2**. No significant changes were detected either between visits or groups. Fluorescence in situ hybridization was used to enumerate the TABLE 1 | Mean (SEM) of daily macronutrient intake during the WGO and NWG intervention periodsa.


*<sup>a</sup>Determined from estimated 3-day dietary diaries. SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids; TFA, trans fatty acids; NMES, non-milk extrinsic sugars; NSP, non-starch polysaccharide; AOAC, Association of Official Analytical Chemists.*

main gut bacterial groups (**Table 3**). Baseline levels of all bacterial groups were comparable between the two treatments. A time by treatment interaction for bifidobacteria (P = 0.000; **Figure 1A**), lactobacilli (P = 0.001; **Figure 1B**) and total bacterial population (P = 0.008; **Figure 1C**) was observed. After consumption of WGO for a 6-week period the numbers of bifidobacteria (Log<sup>10</sup> bacteria cells/g feces) (P = 0.000), lactobacilli (P = 0.000), and total bacteria population (P = 0.017) significantly increased compared to the respective baseline. A significant decrease in bifidobacteria (P = 0.03) and total bacteria population (P = 0.034) was observed on NWG consumption after the 6-week feeding time. A significant difference at 6 weeks between bifidobacteria (P = 0.000) and total population (P = 0.003) size after WGO and NWG was observed, however they returned to near baseline levels after the 4-week washout. No significant changes in the population size of Bacteroides spp., Atopobium spp., and Clostridium group were observed.

#### Blood Lipid Parameters

Baseline concentrations of all lipid parameters were comparable between the two treatments (**Table 4**). A significant time by treatment interaction for TC (P = 0.000), and LDL-C (P = 0.002) was observed. After consumption of WGO for a 6-week period, TC concentrations significantly decreased (P = 0.0016), in contrast after NWG a significant increase in TC levels was observed (P = 0.0016; **Figure 2A**). Plasma LDL-C was seen to significantly increase (P = 0.0055) after consumption of the NWG cereal (**Figure 2B**). A significant difference at 6 weeks between TC (P = 0.000) and LDL-C (P = 0.009) after WGO and NWG was observed, however they returned to near baseline levels after the 4-week washout. No significant changes in HDLcholesterol, glucose or TAG concentrations were observed.


TABLE 2 | Mean (±SEM) concentrations of fecal short chain fatty acids (SCFA) before during and after WGO and NWG treatment arms.

*(n* = *30), visit 1 represents baseline.*

#### TABLE 3 | Mean (± SEM) fecal bacterial numbers (n = 30) over the trial period.


*Bacterial counts in stool samples as determined by fluorescence in situ hybridisation are shown expressed as mean log*<sup>10</sup> *cells/g feces. WGO, Whole Oat Grain; NWG, Non-Whole Grain; abcdefAll values in one row with a common letter are significantly different from each other (P* < *0.05, Tukey's post-test).*

# Insulin Sensitivity/Resistance and Marker of Inflammation

The homeostasis model for insulin resistance (HOMA IR), as well as QUICKI (model for insulin sensitivity) was calculated for both treatments (**Table 4**). There was no significant effect for either treatment. However, there was a near significant time by treatment interaction for fasting glucose (p = 0.066). After the WGO cereal fasting plasma glucose reduced and was lower at the end of treatment (week 6) than after the NWG. Biomarkers of inflammation including; C-Reactive Protein (CRP), IL-6, TNFalpha from blood plasma, IgA in stool and saliva and calprotectin levels in stool samples are shown in **Table 4**. No significant time by treatment interaction was recorded for any of the inflammation biomarkers after consumption of either cereal treatments.

#### Anthropometric Measures and Satiety Gut Hormones

The data for changes in volunteer's GLP-1 and PYY concentrations and BMI, weight and fat mass are shown in **Table 4**. No significant changes were detected between visits or groups for either of the cereal treatments.

#### Biomarkers of Gut Health

Stool frequency and consistency, qualitatively graded by volunteers as hard, formed, or soft, varied considerably between individuals. No significance differences were observed between treatments. The severity and frequency of reported changes in digestive tolerance varied greatly between volunteers, with neither treatment resulting in adverse symptomology (data not shown).

# DISCUSSION

This novel study was designed to test the hypothesis that a whole grain oat based Granola breakfast cereal can mediate a prebiotic modulation of the human gut microbiota typified by increased relative abundance of bifidobacteria in particular, and concomitant reduction of plasma LDL cholesterol. These data also support a significant body of evidence that ingestion of oats, and oat derived fractions, most notably, β-glucan, at 3 g/day, is associated with a significant reduction in blood TC and LDL-C in hypercholesterolemic groups (EFSA Panel on Dietetic Products, Nutrition and Allergies (NDA), 2010; Tiwari and Cummins, 2011). Although, the bile acid and possibly cholesterol binding abilities of β-glucans have been suggested to be responsible for the hypocholesterolemic effects in vivo, other mechanisms may also be involved, including those linked to the human gut microbiota which have not been addressed to date. In this study the WGO group consumed 1.3 g β-glucan daily, which is under half of the daily recommendation for significant TC and LDL-C reduction. However, despite the low β-glucan dose, WGO consumption resulted in 0.94 mmol/l and 0.4 mmol/l lower concentrations of TC and LDL-c , respectively, compared with the NWG at the end of the 6-week intervention period. These data suggest that other mechanisms of action, other than β-glucan, could have contributed to the observed cholesterol reduction. Similar to β-glucan, other fibers or prebiotics, which do not form viscous gels within the intestinal tract, have been shown to lower blood cholesterol, though these prebiotics including

the fructans, xylooligosaccharides, galactooligosaccharides and resistant starch have not received the same attention from the scientific community as β-glucans (Park et al., 2004; Beylot, 2005; Sheu et al., 2008; Vulevic et al., 2013). In parallel, a number of human feeding studies with well-powered cohorts of hypercholesteroleamic individuals have demonstrated that probiotics selected for bile salt hydrolase (BSH) activity lower plasma cholesterol levels to a similar extent to oats and β-glucan (Jones et al., 2012a,b). Such observations raise the intriguing possibility, that along with the viscous gel forming activities of β-glucan, oats may also possess other biological activities

which could contribute to their cholesterol lowering ability; not least the ability to stimulate bacteria within the gut with BSH activity thereby increasing bile acid deconjugation, facilitating bile acid excretion in feces and triggering de novo hepatic bile acid synthesis from circulating cholesterol. Therefore, the current study was designed to investigate whether WGO mediate a prebiotic modulation of the human gut microbiota with specific increases in bifidobacteria, a group of bacteria with well recognized BSH activity and whether this was associated with reducing plasma cholesterol (Tanaka et al., 1999).

Currently few studies, and none containing oats, have measured the ability of whole grain cereals to modulate bacterial relative abundance in human feces (Costabile et al., 2008; Carvalho-Wells et al., 2010). Andersson et al. (2013) observed gut microbiota modulation in response to a dietary hypocholesterolic intervention with oats in two strains of C57BL/6 mice. We have previously shown that different whole grain cereals, including whole grain oat preparations, have prebiotic potential in vitro, determined by significant stimulation of both bifidobacteria and lactobacilli bacteria (Connolly et al., 2010, 2012). Using the same WGO cereal that induced the highest modulation in bifidobacteria and lactobacilli bacteria in vitro and also had the lowest GI (37), we have shown for the first time in humans that WGO consumption for 6 weeks


TABLE 4 | Mean (±SEM) fasting concentrations of plasma lipids and lipoprotein concentrations, glucose, insulin, markers of insulin resistance, and inflammatory markers at baseline, during and following ingestion of WGO and NWG.

*BMI, fat mass and weight and concentrations of satiety gut hormones GLP-1 and PYY of participants before, after and following washout of both test cereals (n*= *30).* <sup>∞</sup>*Significantly different from baseline (P* < *0.05, Tukey's post-test).*

∇ *Insulin sensitivity calculated using Quantitative Insulin Sensitivity Check Index (QUICKI), and insulin resistance calculated using homeostatic model assessment (HOMA) measured before and after 42 wk of either WGO or NWG treatment. All values are average of volunteer's measurements* ± *SEM, (n*=*30). All measurements were taken in the morning after the subjects had fasted for 12 h.*

can stimulate a prebiotic modulation of the gut microbiota selectively increasing fecal bifidobacteria, lactobacilli, and total bacteria compared to a control treatment (Connolly et al., 2010, 2012). These bacterial numbers returned to baseline levels after 4 weeks washout in the group of metabolically at risk participants. No significant changes in Bacteroides spp., C. histolyticum/perfringens group, or Atopobium spp. was detected in the fecal samples collected, as also found in previous studies with other prebiotics. Furthermore, we observed a significant reduction in TC and LDL-C concentrations after the WGO treatment compared with NWG supporting other findings (EFSA Panel on Dietetic Products, Nutrition and Allergies (NDA), 2010; Tiwari and Cummins, 2011). Although this was associated with a significantly higher monounsaturated fatty acid (MUFA) and lower starch intake in the WGO group, this small absolute change in macronutrient intake is unlikely to have impacted on the circulating lipids, based on the predicted changes reported by Mensink et al. (1% dietary energy exchange of carbohydrate with MUFA was associated with a non-significant reduction in total cholesterol of 0.006 mmol/l and a significant, but relatively small reduction in LDL-C of 0.009 mmol/l) (Mensink et al., 2003). A number of well controlled animal and human feeding studies with strains of bacteria which possess bile salt hydrolase (BSH) activity, like Bifidobacterium and Lactobacillus spp., have reported significant reductions in circulating TC and LDL-C levels to a similar degree as oats, β-glucans and other gel forming fibers like psylleum (Begley et al., 2006; Ooi et al.,

2010; Ejtahed et al., 2011; Jones et al., 2012a,b). In addition in vitro fermentation studies have shown that various oat fractions, including β-glucans can both increase SCFA production and modulate relative abundance of key gut microbiota bacteria (Hughes et al., 2008; Kim and White, 2009; Connolly et al., 2012; Nordlund et al., 2012). The ability of oats and β-glucans to increase SCFA production by the gut microbiota in vitro and in animal studies has been shown, while few studies have examined the ability of oats or β-glucans to increase fermentation in humans (Nilsson et al., 2008; Turunen et al., 2011). It has been postulated that possible mechanisms of action of prebiotics on cholesterol reduction may include physiological actions of SCFA, particularly propionate, to reduce de novo hepatic cholesterol synthesis by down regulation of HMG CoA reductase; increased cholesterol excretion from cholesterol assimilation, deconjugation of bile acids and cholesterol binding to bacterial or plant cell walls (Shapiro and Rodwell, 1971; Levrat et al., 1994).

The mechanisms involved with the observed cholesterol reduction in the present study are not clear. A possible limitation of the study was that plasma bile salts and SCFA were not measured. Although it was observed that fecal SCFA concentrations were similar between treatments, these data may not reflect plasma SCFA concentrations and this mechanism cannot be ruled out. Recent studies have shown that SCFA induce their own active uptake transporters on the gut wall, hindering the use of fecal SCFA concentrations as an indicator of colonic SCFA production from fermentation (Borthakur et al., 2012). Further mechanistic studies are required to examine how prebiotic modulation of the gut microbiota may be linked to cholesterol metabolism, and specifically whether a prebiotic type microbiota modulation increases bile acid deconjugation impacting on bile acid excretion or signaling within the intestine and at other body sites specifically in the liver and brain. In summary we have confirmed that dietary WGO ingestion had an appreciable impact on the composition of the human gut microbiota, and significantly reduces plasma TC and LDL-C, a step toward understanding their gut mediated communication with host energy metabolism.

#### AUTHOR CONTRIBUTIONS

JL, KT, and MC designed the study. MC conducted the study. MC analyzed the data and drafted the paper and

#### REFERENCES


XT performed the dietary analysis. JL, KT, and XT edited the paper and all authors have read and approved the final manuscript. JL and KT had primary responsibility for final content.

#### FUNDING

The study was funded by the Reading Endowment Trust Fund (RETF) and Jordans Cereals (Biggleswade, UK) who kindly supplied the breakfast cereals, however the sponsor had no input into the study hypothesis and design, data analysis and interpretation.

#### ACKNOWLEDGMENTS

The authors would like to thank Jan Luff for her assistance with subject recruitment and the study volunteers for their participation in the study.

pursuant to Article 14 of Regulation (EC) No 1924/2006. EFSA J. 8:1885. doi: 10.2903/j.efsa.2010.1885


**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 Connolly, Tzounis, Tuohy and Lovegrove. 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.

# Phylogenetic Analysis of the Bifidobacterium Genus Using Glycolysis Enzyme Sequences

Katelyn Brandt1,2 and Rodolphe Barrangou1,2 \*

<sup>1</sup> Functional Genomics Graduate Program, North Carolina State University, Raleigh, NC, USA, <sup>2</sup> Department of Food, Bioprocessing and Nutrition Sciences, North Carolina State University, Raleigh, NC, USA

Bifidobacteria are important members of the human gastrointestinal tract that promote the establishment of a healthy microbial consortium in the gut of infants. Recent studies have established that the Bifidobacterium genus is a polymorphic phylogenetic clade, which encompasses a diversity of species and subspecies that encode a broad range of proteins implicated in complex and non-digestible carbohydrate uptake and catabolism, ranging from human breast milk oligosaccharides, to plant fibers. Recent genomic studies have created a need to properly place Bifidobacterium species in a phylogenetic tree. Current approaches, based on core-genome analyses come at the cost of intensive sequencing and demanding analytical processes. Here, we propose a typing method based on sequences of glycolysis genes and the proteins they encode, to provide insights into diversity, typing, and phylogeny in this complex and broad genus. We show that glycolysis genes occur broadly in these genomes, to encode the machinery necessary for the biochemical spine of the cell, and provide a robust phylogenetic marker. Furthermore, glycolytic sequences-based trees are congruent with both the classical 16S rRNA phylogeny, and core genome-based strain clustering. Furthermore, these glycolysis markers can also be used to provide insights into the adaptive evolution of this genus, especially with regards to trends toward a high GC content. This streamlined method may open new avenues for phylogenetic studies on a broad scale, given the widespread occurrence of the glycolysis pathway in bacteria, and the diversity of the sequences they encode.

Keywords: Bifidobacterium, glycolysis, phylogeny, probiotic, evolution

# INTRODUCTION

Bifidobacterium species are an important component of the human gastrointestinal tract (GIT) microbiome, and exert critical functional roles, especially during the establishment of gut microbial composition early in life. Consequently, they are the subject of extensive microbiological and genetics studies, to investigate their probiotic phenotypes, and genotypes, respectively. Actually, many studies are investigating the genetic basis for their health-promoting functionalities, both in industry and academia. This genus is often found in the GIT of animals (Ventura et al., 2014), and is the predominant phylogenetic group early in human life (Turroni et al., 2012a). Indeed, a mounting body of evidence has established vertical transmission between the mother and infants (Milani et al., 2015), notably through the selective nurture of bifidobacteria

#### Edited by:

Francesca Turroni, University College of Cork, Ireland

#### Reviewed by:

Abelardo Margolles, Consejo Superior de Investigaciones Científicas, Spain Gabriele Andrea Lugli, University of Parma, Italy

> \*Correspondence: Rodolphe Barrangou rbarran@ncsu.edu

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 29 February 2016 Accepted: 20 April 2016 Published: 09 May 2016

#### Citation:

Brandt K and Barrangou R (2016) Phylogenetic Analysis of the Bifidobacterium Genus Using Glycolysis Enzyme Sequences. Front. Microbiol. 7:657. doi: 10.3389/fmicb.2016.00657

through diverse non-digestible human-milk oligosaccharides (HMOs) that are a critical component of breast milk (Sela, 2011). These HMOs selectively drive the colonization of the infantile GIT by species that encode prebiotic transporters and hydrolases (Turroni et al., 2012b). Recently, a dichotomy has been established between healthy term babies with a normal gut microbiome, and preterm infants whom have not been colonized by Bifidobacterium species (Arboleya et al., 2015). Several studies have implicated the expansive carbohydrate uptake and catabolism gene repertoire of bifidobacteria as the key driver of adaptation of this genus to the infant diet (Milani et al., 2014). In fact, several species of bifidobacteria have shown unique genome composition adaptation trajectories in their carbohydrate utilization machinery, rendering them competitive in this environment (Pokusaeva et al., 2011; Ventura et al., 2012).

To better understand how these organisms have emerged as potent early-life colonizers, there has been a surge in genome sequencing in recent years. At the time of writing, 47 established species and subspecies have been sequenced (Milani et al., 2016), providing a wealth of genomic information, which serves as a valuable tool for understanding the species and strain diversity within this polymorphic genus, as well as unraveling the key elements that drive health-promoting and colonization phenotypes in humans. However, given the democratization of sequencing technologies in general, and genome and microbiome sequencing in particular, it is imperative that tools and methods be available to analyze this high-throughput data, and specifically allow experimentalists to parse out the complex phylogeny of this broad genus. Indeed, basic questions being addressed regarding the occurrence, diversity and functions of various Bifidobacterium species in the human GIT will require the ability to accurately and consistently assign phylogeny.

Fundamentally, as new sequences become available, it is important to know where to place strains on the phylogenetic tree of Bifidobacterium. Whereas the affordability, accessibility and ability to generate high-throughput data have become somewhat straightforward, a key challenge lies in the analysis of these sequences, regarding assembly, comparative analyses and phylogenetic assignments. Historically, 16S rRNA sequences have been used across the phylogenetics field for classification and sequence tree-based assignments, but there are growing concerns about the adequacy and sustainability of this method (Fox et al., 1992), notably with regards to the availability of proper references (Clarridge, 2004), and the actual levels of conservation of sequences targeted by "universal" primers (Baker et al., 2003). Because of this, new approaches have been suggested, ranging from multi-locus approaches, using housekeeping genes (Eisen, 1995), to core-genome analyses (Medini et al., 2005). For Bifidobacterium, efforts have been focused on creating a phylogeny based on whole and/or conserved genomic sequences, namely the pan-genome and the core-genome, respectively (Lukjancenko et al., 2011; Lugli et al., 2014). While the core-genome is arguably comprehensive, core-genome assembly is time consuming and computationally intense. Alternative methods need to be developed, to allow rapid and convenient phylogenetic screening of new and potentially unknown sequences. Preferably, such a method would provide high resolution, low-throughput, robust, accurate, and affordable information.

Notwithstanding phenotypic diversity between organisms that have specialized metabolic pathway combinations, and the corresponding genomic complement, there are core biochemical pathways and processes that are broadly distributed across the Tree of Life. Noteworthy, glycolysis is a fundamental process for most cells, and may be construed as the biochemical backbone of most, if not all, living organisms. Indeed, this process allows for the genesis of energy through the catabolism of simple carbohydrates. This pathway is, at least partially, present in all genomes (Fothergill-Gilmore and Michels, 1993) and consequently constitutes a promising biochemical, and thus genetic, marker for phylogenetic studies. Because these genes are important, they are typically members of the house-keeping genomic set, and are widely dispersed across the Tree of Life. However, they are likely subject to less selective pressure than other phylogenetic markers (i.e., ribosomal sequences), and thus afford a more diverse set of sequences to encompass a broad range of assorted sequences (Fothergill-Gilmore, 1986). Therefore, we set out to assess the potential of glycolytic genes, and the sequences of the proteins they encode, for bifidobacteria phylogenetic studies. In particular, we determined the occurrence and diversity of these glycolytic enzyme genes in the genomes of bifidobacteria, and compared and contrasted sequence alignment-based trees with one another, and to those derived from alternative sequences, notably the core-genome, and the 16S rRNA-based reference tree. Our results show how the glycolysis protein sequences can be used as suitable markers to create a phylogeny of Bifidobacterium that is as accurate as the core-genome based phylogeny, but much less computationally demanding. We also explore how basic features of the genetic sequences of glycolysis can reveal trends and patterns of evolution among the different Bifidobacterium species and the genus as a whole.

#### MATERIALS AND METHODS

# Genetic Sequences Sampling and Reference Genomes

We used sequences derived from a total of 48 Bifidobacterium genomes from distinct species and subspecies, as listed in **Table 1**. Bifidobacterium stercoris was included in this analysis, as a separate species, but it was recently renamed as a strain of Bifidobacterium adolescentis (Killer et al., 2013). Our results (see below) show that B. stercoris is always a close neighbor of B. adolescentis, consistent with the newest findings. These genomes were mined for the presence of glycolytic enzymes using Geneious version 9.0.5(Kearse et al., 2012). We selectively elected to pursue a scheme based on canonical glycolysis genes, as to generate a broadly applicable method. Nevertheless, the classical glycolysis genes do not universally occur in bacterial genomes. Furthermore, some organisms do carry alternative pathways, such as the bifid shunt in bifidobacterium, which could prove valuable, but are not widely distributed. The nine canonical glycolysis enzymes from bifidobacteria (de Vries and

Stouthamer, 1967) were found in each genome. Four reference species (Bifidobacterium longum subsp. longum, B. adolescentis, Bifidobacterium animalis sub. lactis, and Bifidobacterium breve) were used to make a database of the nine genes. The Annotate from Database feature was used (with 40% nucleotide sequence similarity cut-off) to identify glycolytic orthologs in the other genomes. As all genomes had been previously annotated, we confirmed the original annotation to the database annotation



List of the 48 species and subspecies used in this study. Accession numbers and naming conventions included.

manually to validate this method of mining. In cases where multiple hits were obtained, BLAST (Altschul et al., 1990) analyses were carried out to select the correct homolog. Translated sequences were confirmed using ExPasy (Gasteiger et al., 2003). For the 16S rRNA analysis, the 16S rRNA sequences were extracted manually from each genome. In case of multiple hits, BLAST analyses were carried out to select the right sequences. For increased robustness, the glycolysis enzyme sequences were concatenated in order of occurrence in the glycolysis pathway (Lang et al., 2013).

# Genesis of Sequence Alignment-based Trees

Five different alignments were made for each tree using Geneious version 9.0.5. ClustalW (Larkin et al., 2007) was used, with the BLOSUM scoring matrix, and settings of gap creation at −10 cost, and gap extension at −0.1 cost per element. For the 16S rRNA alignment, ClustalW was set so that the cost matrix was IUB, with a gap opening penalty of 15, and gap extension cost of 6.66. MUSCLE (Edgar, 2004) was used with the setting of eight maximum number of iterations for the amino acid sequences and the 16S rRNA alignments. The Geneious Pairwise Alignment was set so that the alignment type was global alignment with free end gaps and the cost matrix was BLOSUM62 for the amino acid sequences. For the 16S rRNA gene analysis, the alignment type was global alignment with free end gaps and a cost matrix of 65% similarity (5.0/−4.0). MAFFT (Katoh et al., 2002) was used twice, for both the amino acid sequences and the 16S rRNA sequences. For the amino acid sequences the first alignment had an algorithm setting of auto, a scoring matrix of BLOSUM62, a gap open penalty of 1.53, and an offset value of 0.123. The second

FIGURE 2 | Glycolytic proteins concatenated tree. Consensus tree based on alignment of the concatenated amino acid sequences of the glycolysis pathway found in Bifidobacterium. Trees were made using RaxML. Bootstrap values are found on each node. Phylogenetic groups are colored as follows: Bifidobacterium longum is orange, Bifidobacterium adolescentis is green, Bifidobacterium psdeudolongum is purple, Bifidobacterium pollorum is blue-green, Bifidobacterium boum is blue, and Bifidobacterium asteroides is red. Species names follow the naming convention from Table 1.

alignment had an algorithm setting of auto, a scoring matrix of BLOSUM80, a gap open penalty of 1.53, and an offset value of 0.123. For the first 16S rRNA alignment, the algorithm was set to auto, the scoring matrix was set to 100 PAM/k = 2, the gap open penalty was set to 1.53, and the offset value was set to 0.123. The

#### TABLE 2 | Sum of branch lengths for each tree.


Sum of branch lengths for each tree. EC number for each enzyme is also listed.

second alignment for the 16S rRNA was set so that the algorithm was auto, the scoring matrix was 200 PAM/k = 2, the gap open penalty was 1.53, and the offset value was 0.123. trimAl (Capella-Gutiérrez et al., 2009) was used to select a consistent alignment between the five alignments. The parameters were compareset and automated1. Using Geneious, trees were made from the respective consistent alignments. The trees were generated using RaxML version 7.2.8 (Stamatakis, 2006b, 2014). For the protein based trees the parameters were set so that the model was CAT (Lartillot and Philippe, 2004) BLOSUM62, the algorithm was Bootstrap using rapid hill climbing with random seed 1, and the number of bootstrap replicates was 100 (Stamatakis, 2006a). For the 16S rRNA tree, the nucleotide model was GTR CAT, the algorithm was Bootstrap using rapid hill climbing with random seed 1, and the number of bootstrap replicates was 100. A consensus tree was then built using the consensus builder in Geneious, at a 50% support threshold. The consensus tree was used in all further analyses. The sums of branch lengths for each tree were found by adding the branch lengths together in Mega6 (Tamura et al., 2013).

#### Statistical Analyses

All statistical analyses were carried out using R version 3.2.2 (R Core Team, 2015). This software was also used to generate plots, graphs and display quantitative data throughout the manuscript.

#### RESULTS

#### Glycolytic Enzyme Sequence-based Phylogeny

Bifidobacteria contain nine of the 10 traditional enzymes (**Figure 1**) commonly found in the glycolysis pathway (de Vries and Stouthamer, 1967). Phylogenetic analyses were carried out using the amino acid sequences of the proteins encoded by the aforementioned glycolysis genes. A comprehensive tree based on sequence alignment of the concatenated sequences of the glycolytic enzymes found in Bifidobacterium

is shown in **Figure 2**. Six separate phylogenetic groups were identified, as previously established from the core-genome (Milani et al., 2016). These groups are: the B. longum group (orange), the B. adolescentis group (green), the Bifidobacterium pseudolongum group (purple), the Bifidobacterium pollurom group (blue-green), the Bifidobacterium boum group (blue), and the Bifidobacterium asteroides group (red; Bottacini et al., 2014). The number of individuals in each group varied between 3 and 11, with the B. longum group being the most diverse. Bifidobacterium angulatum and Bifidobacterium merycicum were moved to the B. adolescentis group due to a high bootstrap value in the concatenated tree. The concatenated tree has bootstrap values that range from 52 to 100. We observe a total of 34 bootstrap values of 70 and above (Supplementary Figure S1). Trees based on sequence alignments of the individual enzymes of glycolysis can be found in Supplementary Figures S2–S10. Interestingly, all of the individual trees resolved the phylogenetic groups found in the core-genome with only the Gap and Eno trees providing alternative locations for a few branches, notably Bifidobacterium magnum, Bifidobacterium gallicum, and Bifidobacterium thermacidophilum sub. thermacidophilum. **Table 2** shows the sum of branch lengths for each tree. The 16S rRNA tree has the largest sum at 204.99, while the concatenated tree had the smallest sum at 99.56. The consistent clustering into these six phylogenetic trees illustrates how robust and valuable the glycolytic sequences are with regards to phylogenetic information. It also shows that this method is congruent with the core-genome.

#### 16S rRNA-based Reference Phylogeny

A reference phylogeny was generated using the 16S rRNA sequences of each of the 48 species and sub-species included in this study (**Figure 3**). The six phylogenetic groups are identified and colored the same as in the concatenated tree. We elected to assign the B. angulatum and B. merycicum from the B. longum group to the B. adolescentis group, consistent with the concatenated tree. Noteworthy, the tree has bootstrap values that range from 51 to 100, with 17 nodes at values of 70 and above, which is half the amount found in the concatenated tree (Supplementary Figure S1). With regards to size, we point out that the concatenated tree is based on overall sequences ranging between 3,205 amino acids and 3,479 amino acids, which quantitatively compares as approximately twice the amount to the 16S rRNA ∼1,600 nt range, in terms of input-information amounts.

#### Genome-Wide Analyses

The overall genome sizes in this study ranged from 1.73 Mb for Bifidobacterium indicum to 3.26 Mb for Bifidobacterium biavatii, with an average of 2.28 Mb and a median of 2.17 Mb. The GC content ranged from 52.8% for Bifidobacterium tsurumiense to 65.5% for Bifidobacterium choerinum, with an average of 60.4% and a median of 60.2%. This substantiates the perception that bifidobacteria are generally categorized as high-GC content organisms, at the genome-wide level (Ventura et al., 2007). However, a thorough analysis of GC content across the

phylogenetic groups revealed that even among these high-GC organisms there are three distinct subsets of high, medium, and low-GC bifidobacteria (**Figure 4A**). Most of the species fall in the upper medium-GC range, with the low-GC range being the least populated. There are some noteworthy groupings between the phylogenetic groups, specifically the B. pullorum and the B. boum groups, for which the entire groups are packed tightly

FIGURE 4 | GC content by species and glycolytic genes. (A) Shows the total GC content of each species organized by the glycolytic concatenated tree. Spectrum is split into three groups: low GC from 0.52 to 0.567 (gray), Medium GC from 0.567 to 0.613 (pink), and High GC from 0.613 to 0.66 (yellow). Phylogenetic groups are colored as follows: B. longum is orange, B. adolescentis is green, B. psdeudolongum is purple, B. pollorum is blue-green, B. boum is blue, and B. asteroides red. Species names following the naming convention from Table 1. (B) contains notched boxplots of the GC values of each gene and total GC. Boxes are ranked in order of median. Notches that do not overlap are indicative of strong evidence of difference between two medians.

in the high GC region and the medium GC region, respectively. All of the other groups, except the B. longum group, span two of these subsets. For the B. longum group, Bifidobacterium saguini lies just at the border between the low and medium GC subsets. This group has the largest spread, consistent with being the most diverse in the concatenated and 16S rRNA trees.

Next we looked at how the GC content varied across the trees. **Figure 4B** shows boxplots of the GC content of each tree and the total GC content. Except for the 16S rRNA and tpi trees, all other trees had median GC values with strong evidence of being higher than the median total GC content (Chambers, 1983). Looking on an individual basis, over half of the genomes have 16S rRNA and tpi GC values below their total GC, while the other genes are either above or close to their total GC (**Figure 5**). Again, the B. pullorum and B. boum groups are tightly packed in regards to their GC spread amongst their glycolysis genes, 16S rRNA, and total GC. In contrast, the B. longum group has the largest spread, a parallel to its higher diversity in the phylogenetic trees.

# DISCUSSION

Bifidobacterium is a diverse genus of human intestinal beneficial microbes that provide health-promoting functionalities, as illustrated by their broad use as probiotics in foods and dietary supplements (Turroni et al., 2014). Recently, extensive genomic analyses of diverse species, subspecies and phylogenetic groups have provided insights into their adaptation to the human gut, notably with regards to their ability to colonize the intestinal cavity in general, and utilize non-digestible carbohydrates in particular (Milani et al., 2016). Studies investigating the use of human breast milk oligosaccharides illustrate the important contribution of these probiotics in establishing the human gut microbiome at the early stages of life (Sela, 2011). Yet, these studies also reveal that there are many distinct and diverse Bifidobacterium species and phylogenetic groups that colonize the human GIT, perhaps with idiosyncratic genomic attributes, and their corresponding functionalities (Chaplin et al., 2015). These organisms have specifically adapted to their environment to competitively utilize available nutrients (Sánchez et al., 2013). In the human gut, these consist of non-digestible complex oligosaccharides that are not adsorbed, nor broken down in the upper GIT. Whereas, plant-based fibers are important in the adult diet, HMOs are important components of the infant diet. Furthermore, Bifidobacterium have even been successful in helping each other through cross-feeding (Turroni et al., 2015). Thus, we addressed the need to establish practical means to allocate phylogeny with minimalistic information based on sequences that encode glycolysis, the biochemical spine of most cells.

Here, we have shown that a multigene approach using glycolysis sequences can be used to uncover genomic trends and to make an accurate phylogenetic tree, based on a relatively small amount of information. The concatenated glycolysis tree in **Figure 2** is congruent with both the 16S rRNA tree and the established core-genome-based tree (Milani et al., 2016). The only notable exception is the placement of B. merycicum and B. angulatum. However, the relocation was between two

neighboring phylogenetic groups in the concatenated and coregenome based trees. The glycolysis pathway is perhaps as, if not more, robust and accurate than the 16S rRNA tree. Compared to the 16S rRNA, the bootstrap values of the concatenated tree were higher on average. This leads to more confidence in the placement of species and the identification of phylogenetic groups, which in comparison, can appear arbitrarily located on the 16S rRNA. The concatenated tree is able to identify groups as well as the core-genome based tree. In fact, all of the phylogenetic groups from the core-genome were consistently found across the glycolytic pathway based trees. However, the glycolysis-based trees have the advantage of being much less labor intensive than the core-genome approach. This allows for accurate phylogenetic mapping of new strains or species, possibly encompassing unknown species, in less time and with less data than a core-genome. This approach is high resolution, low throughput, affordable, and accurate. Part of the success of this approach is the universality of glycolysis. Glycolysis is the biochemical backbone of the cell, and as such all organisms have at least some part of the glycolysis pathway represented (Fothergill-Gilmore and Michels, 1993). Even though these are slower-evolving genes, the changes that are made are enough to make an accurate phylogeny (Fothergill-Gilmore, 1986), evidenced from the congruence between our trees and the core-genome based tree. Even though the glycolysis enzymes are considered "slow evolvers," our data shows they are evolving at different rates amongst themselves. This can be explained by the fact that the glycolysis pathway is adapted by organisms to best fit their own unique environment and requirements (Bar-Even et al., 2012), as seen here in the Bifidobacterium and their bifid shunt (Sela et al., 2010). Some of the genes have specialized secondary functions, such as enolase acting as a cell surface receptor in Bifidobacterium (Candela et al., 2009). All of this makes the glycolysis pathway an excellent phylogenetic marker candidate. The various rates in evolution and moonlighting abilities also allow for further applications in recognizing adaptive trends.

The functional diversity of bifidobacteria is underpinned by multi-dimensional variety in their genomes, including overall content, organization, sequence diversity, and others. In extreme cases, even a two-fold difference in genome size can be observed. Despite being generally perceived as high GC organisms, they vary enough to have distinct relative classes of high, middle, and low-GC, amongst themselves (**Figure 4A**). Yet, there are non-random patterns and phenomena that drive these differences. The phylogenetic groups are clustered in specific regions of the GC continuum. Some groups are more tightly packed than others. A general trend that is observed across the genus is an evolutionary movement toward a high(er) GC content. The higher end of the spectrum is more densely populated then the lower end of the spectrum, indicative of an upward trend. This is reflected by the increased GC content in the individual glycolysis genes, when compared to the total GC content. Of the glycolysis genes, only one, tpi, does not show strong evidence for being different from the genome-wide (total) GC content. Critically, all of the other genes are above the total GC content. When we combine the overall genomic data with the GC-content groupings and trends discovered using glycolysis as phylogenetic markers, we posit the hypothesis that, over time, the GC content within the genomes of bifidobacteria increases, as to deviate further away from the 50% value, as the organisms adapt, and their genomes evolve accordingly.

Because of the broad occurrence of the glycolysis pathway in the Tree of Life, it is a suitable candidate marker to use in phylogenetic studies, likely beyond its application in bifidobacteria. In addition to being conserved genes that capture genetic diversity, glycolysis genes are consistently amongst the most highly expressed in not only Bifidobacterium (Turroni et al., 2015), but other organisms as well (Barrangou et al., 2006). This reflects both the importance of these sequences genetically (as illustrated by GC content drift), and functionally (as illustrated by their propensity for high levels of constitutive transcription). Because of this, it may be possible to correlate transcriptional data to phylogenetic studies on a broader scale. From here, it could be feasible to assign species and map data to known references using transcriptomic, genomic, or meta-data. Indeed, as the democratization of metagenomic technologies continues, and the need to assign phylogenetic information to partial genomic information increases, we propose that this method be used to provide insights into the phylogeny of un-assigned contigs. Overall, this approach allows for accurate phylogenetic mapping, congruent with a core-genome and more robust than the 16S rRNA phylogenetic approach, as well as inference on genomic adaptation, using either genomic, transcriptomic, or meta-data in a timely fashion and with minimal computation.

# AUTHOR CONTRIBUTIONS

KB and RB designed and carried out experiments, interpreted results, and wrote the manuscript.

# ACKNOWLEDGMENT

We would like to thank the Dr. Todd Klaenhammer lab and the CRISPR lab for providing insights and support during this project.

# FUNDING

This study was supported by startup funds from North Carolina State University. KB is a recipient of a NIEHS training grant.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb.2016. 00657

# REFERENCES


history of an ancient phylum. Microbiol. Mol. Biol. Rev. 71, 495–548. doi: 10.1128/MMBR.00005-07


**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 Brandt and Barrangou. 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.

# Phylogenomic Analyses and Comparative Studies on Genomes of the Bifidobacteriales: Identification of Molecular Signatures Specific for the Order Bifidobacteriales and Its Different Subclades

#### Grace Zhang<sup>1</sup> , Beile Gao<sup>2</sup> , Mobolaji Adeolu<sup>1</sup> , Bijendra Khadka<sup>1</sup> and Radhey S. Gupta<sup>1</sup> \*

*<sup>1</sup> Department of Biochemistry and Biomedical Sciences, McMaster University, Hamilton, ON, Canada, <sup>2</sup> CAS Key Laboratory of Tropical Marine Bio-resources and Ecology, Guangdong Key Laboratory of Marine Materia Medica, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China*

#### Edited by:

*Marco Ventura, University of Parma, Italy*

#### Reviewed by:

*Gabriele Andrea Lugli, University of Parma, Italy Longxian Lv, Zhejiang University, China*

> \*Correspondence: *Radhey S. Gupta gupta@mcmaster.ca*

#### Specialty section:

*This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology*

Received: *29 April 2016* Accepted: *07 June 2016* Published: *27 June 2016*

#### Citation:

*Zhang G, Gao B, Adeolu M, Khadka B and Gupta RS (2016) Phylogenomic Analyses and Comparative Studies on Genomes of the Bifidobacteriales: Identification of Molecular Signatures Specific for the Order Bifidobacteriales and Its Different Subclades. Front. Microbiol. 7:978. doi: 10.3389/fmicb.2016.00978* The order *Bifidobacteriales* comprises a diverse variety of species found in the gastrointestinal tract of humans and other animals, some of which are opportunistic pathogens, whereas a number of others exhibit health-promoting effects. However, currently very few biochemical or molecular characteristics are known which are specific for the order *Bifidobacteriales*, or specific clades within this order, which distinguish them from other bacteria. This study reports the results of detailed comparative genomic and phylogenetic studies on 62 genome-sequenced species/strains from the order *Bifidobacteriales*. In a robust phylogenetic tree for the *Bifidobacteriales* constructed based on 614 core proteins, a number of well-resolved clades were observed including a clade separating the *Scarodvia*-related genera (*Scardovia* clade) from the genera *Bifidobacterium* and *Gardnerella*, as well as a number of previously reported clusters of *Bifidobacterium* spp. In parallel, our comparative analyses of protein sequences from the *Bifidobacteriales* genomes have identified numerous molecular markers that are specific for this group of bacteria. Of these markers, 32 conserved signature indels (CSIs) in widely distributed proteins and 10 signature proteins are distinctive characteristics of all sequenced *Bifidobacteriales* species and provide novel and highly specific means for distinguishing these bacteria. In addition, multiple other molecular signatures are specific for the following clades of *Bifidobacteriales*: (i) 5 CSIs specific for a clade comprising of the *Scardovia*-related genera; (ii) 3 CSIs and 2 CSPs specific for a clade consisting of the *Bifidobacterium* and *Gardnerella* spp.; (iii) multiple other signatures demarcating a number of clusters of the *B. asteroides*-and *B. longum*- related species. The described molecular markers provide novel and reliable means for distinguishing the *Bifidobacteriales* and a number of their clades in molecular terms and for the classification of these bacteria. The *Bifidobacteriales*-specific CSIs, found in important proteins, are predicted to play important roles in modifying the cellular functions of the affected proteins. Hence, biochemical studies on the cellular functions of these CSIs could lead to discovery of novel characteristics of either all *Bifidobacteriales*, or specific groups of bacteria within this order. Some of the functions affected/modified by these genetic changes could also be important for the probiotic/pathogenic activities of the bifidobacteria.

Keywords: molecular signatures for bifidobacteria, phylogeny, taxonomy, conserved signature indels, conserved signature proteins, Bifidobacterium asteroides-clade, Scardovia-clade

### INTRODUCTION

The order Bifidobacteriales contains a large collection of bacterial species, many of which are significant constituents of the gastrointestinal tract of humans, other mammals, birds and honey bees (Biavati et al., 2000; Biavati and Mattarelli, 2006; Turroni et al., 2009, 2011; Biavati, 2012; Milani et al., 2014). In addition to widely recognized health-promoting effects of bifidobacterial species (Leahy et al., 2005; Ventura et al., 2009a; Cronin et al., 2011), some members of the group found in human and animal oral cavities are implicated in the development of dental caries (Huys et al., 2007; Mantzourani et al., 2009; Ventura et al., 2009b). Additionally, Gardnerella vaginalis is indicated to play an important role in the pathogenesis of bacterial vaginosis and urinary tract infections (Smith et al., 1992; Bradshaw et al., 2006; Alves et al., 2014; Kenyon and Osbak, 2014). The order Bifidobacteriales is part of the phylum Actinobacteria (Ventura et al., 2007b; Zhi et al., 2009; Gao and Gupta, 2012) and it harbors a single family, Bifidobacteriaceae, containing >50 recognized species (Biavati, 2012; Lugli et al., 2014; Milani et al., 2014; Parte, 2014) that are grouped into eight genera: Aeriscardovia, Alloscardovia, Bifidobacterium, Gardnerella, Pseudoscardovia, Neoscardovia, Parascardovia, and Scardovia (Jian and Dong, 2002; Simpson et al., 2004; Huys et al., 2007; Biavati and Mattarelli, 2012; García-Aljaro et al., 2012; Storms and Vandamme, 2012; Killer et al., 2013). Of these, the genus Bifidobacterium, encompassing 39 species and 9 subspecies, forms the largest group and accounts for more than 75% of the described taxa within the order Bifidobacteriales (Biavati and Mattarelli, 2012; Milani et al., 2014).

Phylogenetic analyses based on 16S rRNA, as well as sequences for a number of housekeeping genes/proteins, are the main approaches used in the past to distinguish among the Bifidobacteriales species and genera (Miyake et al., 1998; Ventura and Zink, 2003; Ventura et al., 2004, 2006, 2007a; Biavati and Mattarelli, 2006; Yarza et al., 2008; Bottacini et al., 2010; Turroni et al., 2011; Mattarelli et al., 2014). In recent years, complete or draft genome sequences have become available for all currently recognized Bifidobacterium species and subspecies (Ventura et al., 2009b; Milani et al., 2014). Based on these sequences, a panel of multiplex PCR primers has been developed enabling rapid and specific identification of different Bifidobacterium species and subspecies (Ferrario et al., 2015). Based on genome sequences, two recent studies have also examined the evolutionary relationships among Bifidobacterium species employing large datasets of sequences comprising the core proteins of this genus (Lugli et al., 2014; Sun et al., 2015). The robust phylogenetic trees obtained in these studies provide important insights concerning the evolutionary relationships among the Bifidobacterium species and they strongly support the existence of 6-7 distinct clusters within this genus. These clusters are referred to as the B. asteroides, B. pseudolongum, B. longum, B. bifidum, B. adolescentis, B. pullorum, and B. boum groups (Lugli et al., 2014; Sun et al., 2015). Similar clusters are also observed in phylogenetic trees based on the 16S and 23S rRNA genes as well trees based on other gene/protein sequences. Comparative analyses of the Bifidobacterales genomes are also providing useful insights concerning species-specific characteristics that are suggested to play important roles in the adaptation of particular species to either human or insect gut environment (Ventura et al., 2009b; Bottacini et al., 2010, 2012; Turroni et al., 2010).

Due to the health-promoting effects of bifidobacteria, it is of much interest to identify genetic and biochemical characteristics that are specific for the Bifidobacteriales or particular groups/clusters within this order of bacteria. Currently, very few such characteristics are known. One important class of genome sequence-based molecular markers, which have proven very useful for evolutionary, taxonomic and functional studies are conserved signature insertions or deletions (CSIs) that are uniquely present in the genes/proteins homologs from a defined group of organisms (Gao and Gupta, 2005, 2012; Gupta, 2010, 2014). Conserved signature proteins (CSPs), which are genes/proteins that are uniquely found within a monophyletic group of organisms, provide another class of useful molecular makers for evolutionary and functional studies (Gao et al., 2006; Ventura et al., 2007a; Gao and Gupta, 2012; Gupta, 2016a,b). Both these types of markers constitute highly reliable characteristics of specific groups of organisms and they have been extensively utilized for the identification/demarcation of prokaryotic taxa of different ranks in molecular terms (Gao and Gupta, 2012; Gupta et al., 2013a,b, 2016).

In the present work, we report detailed phylogenetic and comparative analyses on protein sequences from the sequenced members of the order Bifidobacteriales in order to identify CSIs and CSPs that are specific for different groups within this order. These studies have led to identification of 32 CSIs in widely distributed proteins and 10 CSPs that are uniquely found in all or most of the genome sequenced Bifidobacteriales species providing novel molecular markers that distinguish this order from all other bacteria. In addition, our work has also identified multiple other CSIs and CSPs that distinguish a number of clades of Bifidobacteriales, including a clade consisting of the Bifidobacterium and Gardnerella species, another clade consisting of the Scardovia-related genera, and

multiple signatures that are specific for different clusters of B. asteroides or B. longum related species. These signatures provide novel means for the identification and demarcation of the members of the described clades in molecular terms and for functional studies that could lead to discovery of novel biochemical and/or other novel properties of these bacteria.

# METHODS

#### Phylogenetic Analysis

A phylogenetic tree for 62 genome-sequenced members from the order Bifidobacteriales was constructed based on concatenated sequences of 614 proteins. The protein families used in this phylogeny were identified using the UCLUST algorithm (Edgar, 2010) to identify proteins families present in at least 80% of the input genomes which shared at least 50% sequence identity and 50% sequence length. Each identified protein family was individually aligned using Clustal Omega (Sievers et al., 2011) and trimmed using Gblocks 0.91b (Castresana, 2000) with relaxed parameters (Talavera and Castresana, 2007). The concatenated dataset of the trimmed sequence alignments contained 197, 777 aligned amino acid residues. A maximumlikelihood tree based on this alignment was constructed using FastTree 2 (Price et al., 2010) employing the Whelan and Goldman model of protein sequence evolution (Whelan et al., 2001) and RAxML 8 (Stamatakis, 2014) using the Le and Gascuel model of protein sequence evolution (Le and Gascuel, 2008). SH-like statistical support values (Guindon et al., 2010) for each branch node in the final phylogenetic tree were calculated using RAxML 8 (Stamatakis, 2014). This process was completed using an internally developed software pipeline.

In parallel, a phylogenetic tree based on the 16S rRNA gene sequences of type strains covering all described species within the order Bifidobacteriales was also constructed. The 16S rRNA sequences were retrieved from Ribosomal Database Project (Cole et al., 2014) and aligned using the SINA aligner (Pruesse et al., 2012) to form a multiple sequence alignment that was 1604 aligned nucleotides long with common gaps removed. A maximum-likelihood phylogenetic tree based on this multiple sequence alignment was created using MEGA 6 employing the General Time-Reversible model of sequence evolution with branch support based on 1000 bootstrap replicates (Tamura et al., 2013).

#### Identification of Conserved Signature Indels

Conserved signature indels (CSIs) were identified by the procedures described in detail recently (Gupta, 2014). Briefly, BLASTp (Altschul et al., 1997) searches were performed on each protein in the genome of Bifidobacterium adolescentis ATCC 15703 (Accession number AP009256.1) against all available sequences in the GenBank non-redundant database. Multiple sequence alignments were then created using ClustalX (Jeanmougin et al., 1998) for proteins that returned high scoring matches from Bifidobacteriales and other prokaryotes. The alignments were then visually inspected for the presence of insertions or deletions that were flanked on both sides by at least 5-6 conserved amino acid residues in the neighboring 30– 40 amino acids. Detailed BLASTp searches were then carried out on short sequence segments containing the indel and the flanking conserved regions (60-100 amino acids long) to determine the specificity of the indels. SIG\_CREATE and SIG\_STYLE (available on Gleans.net) were then used to create Signature files for CSIs that were specific for the Bifidobacteriales order or its subgroups as described in earlier work (Gupta et al., 2013a; Gupta, 2014). Due to space limitations, sequence information for all Bifidobacterium species, particularly for different subspecies of B. longum, B. animalis, B. pseudolongum, and B. thermacidophilum, is not shown in the presented alignment files. However, unless otherwise noted, all of the described CSIs are specific for the indicated groups (i.e., similar CSIs were not present in the protein homologs from other bacteria in the top 500 Blast hits). It should be noted that significant blast hits for a number of CSIs and CSPs described here are also observed for one of the following three Chlamydia trachomatis strains (SwabB1, H1 IMS, and H17 IMS) deposited by the Sanger Institute. We suspect that these anomalous results are due to cross contamination of the sequenced cultures from the above Chlamydia trachomatis strains by a Gardnerella vaginalis strain. We have communicated our concern with the supporting evidence to the Sanger Institute.

# Identification of Conserved Signature Proteins

BLASTp searches were carried out to examine the specificity of some previously described conserved signature proteins (CSPs), which were indicated to be specific for the order Bifidobacteriales (Gao and Gupta, 2012). Additionally, limited work to identify CSPs for the B. asteroides group of species was carried out by conducting BLASTp searches on all proteins from the genomes of Bifidobacterium asteroides (Bottacini et al., 2012) as query sequences. BLASTp searches were performed against all available sequences in the GenBank non-redundant sequence database and the results of these searches were then manually inspected, as described in earlier work (Gao et al., 2006; Gao and Gupta, 2012), for proteins for which all significant hits were from the B. asteroides group of species.

# Homology Modeling of Elongation Factor Tu from Bifidobacterium longum

Homology models of EF-Tu homolog from Bifidobacterium longum were built using the solved crystallographic structure of EF-Tu from Escherichia coli (PDB ID: 3U6K) as the template. Initially, 200 models were generated using MODELER v9.14 (Sali and Blundell, 1993) and ranked/selected using assigned discrete optimized potential (DOPE) scores (Shen and Sali, 2006). The model with the highest DOPE score was then submitted to the ModRefiner program to obtain atomic-level energy minimization and to obtain a model with reliable stereochemistry quality (Xu and Zhang, 2011).

# RESULTS

# Phylogenetic Analysis of the Species from the Order Bifidobacteriales

Phylogenomic analyses of members of the genus Bifidobacterium have been previously reported based on core protein sequences from 45 and 48 described species from this genus (Lugli et al., 2014; Sun et al., 2015). However, these studies did not include the other members of the order Bifidobacteriales such as Gardnerella, Scardavia, Alloscardovia, and Parascardovia, as well as several unnamed Bifidobacterium spp. (viz. strains A11, 7101, AGR2158, MSTE12, 12.1.47BFAA) whose genomes have been sequenced. Additionally, the genome sequence of a recently described species B. aesculapii is also now available (Toh et al., 2015). To comprehensively examine the evolutionary relationships among different members of the order Bifidobacteriales, a phylogenetic tree was constructed for all 62 genome sequenced members of the family which included 54 Bifidobacterium species/strains, 5 species from Scardovia and related genera (viz. Alloscardovia and Parascardovia) and three strains of Gardnerella vaginalis. The tree was constructed based on the concatenated sequences of 614 universally or nearly universally present core proteins for which sequence information could be obtained from the 62 sequenced genomes. A maximum-likelihood tree based on these sequences, which represents the most comprehensive phylogenetic analysis of the order Bifidobacteriales to date, is presented in **Figure 1**.

In the tree shown, members of the order Bifidobacteriales, at the highest level, form two main clusters. One of these clusters referred to as the Scardovia cluster groups together the genera Scardovia, Parascardovia, and Alloscardovia, whereas the second cluster is comprised of members of the genus Bifidobacterium and Gardnerella. Importantly in this tree, as well as in an earlier study in a phylogenetic tree based on concatenated sequences for RpoB, RpoC, and GyrB proteins, different strains of Gardnerella vaginalis were found to branch in between the Bifidobacterium species (Gao and Gupta, 2012), making the genus Bifidobacterium polyphyletic. Earlier phylogenetic studies on members of the genus Bifidobacterium have identified a number of different clusters, which are referred to as the B. asteroides, B. pseudolongum, B. longum, B. bifidum, B. adolescentis, B. pullorum, and B. boum groups (Ventura et al., 2006; Turroni et al., 2011; Lugli et al., 2014; Sun et al., 2015). The existence of these groups/clusters is also confirmed and supported by the tree shown in **Figure 1**. Of these clusters, the species-related to B. asteroides cluster exhibited the deepest branching within the genus Bifidobacterium, as also observed in earlier work (Lugli et al., 2014; Sun et al., 2015). The B. asteroides clade is generally demarcated as being comprised of the B. asteroides, B. indicum, B. coryneforme, and B. actinocoloniiforme species (marked as cluster III in **Figure 1**). However, as discussed later, a number of clusters, marked I, II, and IV, which are either part of the B. asteroides clade or are related to this clade are also distinguished in phylogenetic trees and by the CSIs identified in this work.

We have also created a phylogenetic tree based on 16S rRNA gene sequences of all named Bifidobacteriales species (Supplementary Figure S1). The overall branching pattern in the 16S rRNA tree is similar to that observed in the concatenated protein tree with Scardovia and related genera forming the deepest branches in the tree and the genera Scardovia, Alloscardovia, and Parascardovia were part of one of the deepest branching clusters. The different clusters of the Bifidobacterium spp. that are observed in the concatenated protein tree were also supported by the 16S rRNA tree and G. vaginalis was found to branch in between these clusters. The polyphyletic nature of the genus Bifidobacterium in 16S rRNA gene based phylogenies is also observed in earlier work (Yilmaz et al., 2014).

### Identification of Molecular Markers That Are Specific for the Order Bifidobacteriales

The main focus of this work is the identification of molecular characteristics that are specific for the Bifidobacteriales species and could be used for their identification as well as functional studies. As noted earlier, conserved inserts and deletions (i.e., indels or CSIs) in genes/proteins and conserved signature proteins that are uniquely found in a phylogenetically coherent group of organisms provide very useful molecular markers for such purposes. The indels that provide useful molecular markers are of defined size and they are flanked on both sides by conserved regions to ensure that they are reliable characteristics (Gupta, 1998; Gupta and Griffiths, 2002; Ajawatanawong and Baldauf, 2013). These conserved indels in gene/protein sequences result from highly specific and rare genetic changes, hence when such an indel is uniquely found in a phylogenetically coherent group of species, its simplest explanation is that the genetic change responsible for it occurred once in a common ancestor of the indicated group and then the change was passed on to various descendants (Gupta, 1998, 2014; Rokas and Holland, 2000; Gao and Gupta, 2005). Based upon the presence or absence of a conserved indel in outgroup species, it is also possible to determine whether a given indel represents an insert or a deletion (Gupta, 1998; Gao and Gupta, 2012).

Comparative analyses of protein sequence alignments from bifidobacteia species carried out in this work have led to the identification of 32 CSIs in a broad range of highly conserved proteins, which are specifically found in different Bifidobacteriales taxa (see **Table 1**). One example of a CSI that is specific for all members of the order Bifidobacteriales is shown in **Figure 2**. In this case, a 4 amino acid (aa) insertion is present in a highly conserved region of the protein synthesis elongation factor EF-Tu, which is commonly shared by all sequenced bifidobacteria species, but it is not found in any other bacteria in the top 500 BLAST hits. The protein EF-Tu is a highly conserved protein, which is universally present in all organisms (Harris et al., 2003) and the 4 aa CSI in this protein is a distinctive characteristic of homologs from all sequenced Bifidobacteriales species. Sequence information for 31 other CSIs, which are also specifically shared by members of the order Bifidobacteriales, and which are present in proteins involved in different other functions, is provided in Supplementary Figures S2–S32 and some of their

consistently observed in phylogenetic trees are marked.

#### TABLE 1 | Characteristics of conserved signature indels that are Specific for the order Bifidobacteriales.


*<sup>a</sup>The indel region indicates the region of the protein where the described CSI is present.*

characteristics are summarized in **Table 1**. Barring an isolated exception, all of the CSIs listed in **Table 1** are specifically found in different members of the order Bifidobacteriales and are not present in the protein homologs from other bacteria. Due to their specific presence in bifidobacteria species, the described CSIs provide novel molecular markers for distinguishing and demarcating members of the order Bifidobacteriales from all other bacteria. We have previously described 14 CSPs, whose homologs were specifically found in the 13 different sequenced bifidobacteria species that were available at the time (Gao and Gupta, 2012). Updated BLASTp searches on the sequences of these CSPs confirm that 10 of these CSPs, information for whom is provided in **Table 2**, are still distinctive characteristics of members of the order Bifidobacteriales and they provide additional molecular markers for identification and functional studies on bifidobacteria.

#### Molecular Signatures for Some of the Subclades of Bifidobacteriales

In the phylogenetic tree based on concatenated protein sequences, Bifidobacteriales species form a number of different clusters. At the deepest level, of the two main clusters that are observed, one consists of the genus Scardovia and related genera, whereas the other is comprised of species from the genera Bifidobacterium and Gardnerella. In our analyses, we have also identified a number of CSIs and CSPs which distinguish these two clades of the Bifidobacteriales. **Figure 3** shows one example of a CSI consisting of a 1 aa insertion in the DNA polymerase IV protein that is specifically found in different Bifidobacterium species and Gardnerella, but which is not found in any of the sequenced Scardovia-related genera of the Bifidobacteriales. Two other CSIs in the ribosomal RNA small subunit methyltransferase E protein and GTP-binding


FIGURE 2 | Partial sequence alignment of the protein synthesis elongation factor-Tu showing a 4 aa insertion in a conserved region that is specific for

members of the order Bifidobacteriales. The dashes in this alignment as well as all other alignments show identity with the amino acid on the top line. The *(Continued)*

#### FIGURE 2 | Continued

Genebank Identification numbers of the protein sequences are shown, and the topmost numbers indicate the position of this sequence in the species shown on the top line. Due to space constraints, sequence information for different subspecies is not shown. However, unless otherwise indicated, these CSIs are present in the sequenced subspecies of *B. longum, B. animalis, B. pseudolongum,* and *B. thermacidophilum*. Information for large numbers of other CSIs, which are also specific for the order *Bifidobacteriales* is presented in Table 1 and Supplementary Figures S2–S32.

TABLE 2 | Conserved signature proteins that are uniquely found in the Bifidobacteriales.


*The species that are part of the B. asteroides clusters I, II, and III are indicated in* Figure 1*.*

protein YchF are also specifically shared by members of the genera Bifidobacterium and Gardnerella. Sequence information for these CSIs is presented in Supplementary Figures S33, S34 and some of their characteristics are summarized in **Table 3**. Additionally, we have also confirmed that the homologs of 5 of the 6 previously described CSPs (Gao and Gupta, 2012), information for which is summarized in **Table 2**, are also present in only members of these two genera.

We have also identified a number of CSIs that are commonly and specifically shared by members of the genus Scardovia and related genera for which sequence information is available. One example of a CSI which is specifically found in members of the genera Scardovia, Parascardovia and Alloscardovia, consisting of 1 aa insertion in the triosephosphate isomerase protein, is presented in **Figure 4**. Four other CSIs in four different proteins (viz. FHA domain protein, Glycosyl transferase, PAC2 family protein and Phosphate-ABC- transporter substratebinding protein) are also largely specific for these genera of Bifidobacteriales. Sequence information for these CSIs is provided in Supplementary Figures S36–S39 and their characteristics are summarized in **Table 3**. Interestingly, the CSIs in the Glycosyl transferase and PAC2 family proteins are also commonly shared by G. vaginalis.

A number of distinct clusters within the genus Bifidobacterium are consistently observed in different phylogenetic studies including in the phylogenetic trees constructed in this work (**Figure 1**). A number of CSIs identified in our work serve to distinguish some of the Bifidobacterium clusters. Three of the identified CSIs are specific for the B. longum group and sequence information for one of these CSIs, consisting of a 1 aa insertion in the phosphogluconate dehydrogenase, is shown in **Figure 5**. Sequence information for the other 2 CSIs that are also specific for a subgroup of species from the B. longum clade are presented in Supplementary Figures S40, S41 and their characteristics are summarized in **Table 3**. One additional CSI consisting of a 1 aa insertion in transketolase protein is specifically shared by members of the B. longum, B. bifidum, and B. adolescentis clades. Members of these clusters group together in phylogenetic trees and the shared presence of this CSI supports the view that that the members of these taxa are more closely and specifically related to each other.

The members of the B. asteroides cluster forms the deepest branching group within the genus Bifidobacterium. A number of CSIs identified in this study are specific for group of species, which are either part of the B. asteroides clade or related to this clade. The B. asteroides clade is demarcated as being made up of the species B. asteroides, B. indicum, B. coryneforme, and B. actinocoloniiforme species (marked cluster III in **Figure 1**) (Lugli et al., 2014; Sun et al., 2015). Surprisingly, in our work no CSI was identified that was commonly shared by all of the species from this clade. However, our work identified four CSIs for a cluster (cluster II) comprising of all of other species from the B. asteroides clade, except B. actinocoliniiforme, which shows the deepest branching within this clade. One example of a CSI specific for members of the B. asteroides cluster II consisting of 1 aa insertion in the purine biosynthesis protein purH is shown in **Figure 6A**. Sequence information for three other CSIs that are also specific for the B. asteroides group is presented in Supplementary Figures S43–S45. In our phylogenetic trees as well as in different identified signatures, two Bifidobacterium spp. strains A11 and 7101, isolated from honey bee guts (Anderson et al., 2013), also consistently group with the B. asteroides. Two CSIs identified in our work are specifically shared by B. asteroides and the Bifidobacterium sp. A11 and Bifidobacterium sp.7101 (referred to as B. asteroides cluster I) providing additional evidence of the close relationship of these Bifidobacterium strains to the B. asteroides. Sequence information for these CSIs is presented in Supplementary Figure S46. Lastly, one additional CSI identified in this work, consisting of a 3 aa insertion in a conserved region of the protein 5'-methylthioadenosine nucleosidase, is commonly shared by all the members of the


FIGURE 3 | Partial sequence alignment of DNA polymerase IV showing a 1 aa insertion that is specific for the Bifidobacterium and Gardnerella species, but not found in any other Bifidobacteriales. Information for other CSIs specific for this clade is presented in Table 3 and Supplementary Figures S33–S35.

TABLE 3 | Characteristics of Conserved Signature Indels Distinguishing a number of subgroups within the order Bifidobacteriales.


#*The B. asteroides-related cluster I, II, and IV are demarcated in* Figure 1*.*

B. asteroides as well as by B. crudilactis and B. psychaerophilum. The latter two species form a deeper branching cluster that appears to be specifically related to the B. asteroides clade in the tree based on concatenated protein sequences (marked as B. asteroides cluster IV in **Figure 1**). The shared presence of this CSI by the B. asteroides clade and B. crudilactis and B. psychaerophilum support the inference that these species are specifically related to the B. asteroides clade.

In addition to the described CSIs, BLAST searches on the protein sequences of B. asteroides have also identified 5 CSPs, whose homologs are specifically present in the members of B. asteroides group of species. Information for these CSPs is also presented in **Table 2**. Of these CSPs, three CSPs are specific for the commonly described B. asteroides clade (Cluster III in **Figure 1**), whereas the remaining two are specific for the clusters I and II.

#### DISCUSSION

Members of the order Bifidobacteriales are one of the main groups within bacteria where several members exhibit healthpromoting probiotic effects on humans (Biavati et al., 2000; Biavati and Mattarelli, 2006; Ventura et al., 2007b, 2009a; Cronin et al., 2011; Turroni et al., 2011). Other Bifidobacteriales species are also responsible for implicated in the development of dental caries as well as bacterial vaginosis and urinary tract infections (Bradshaw et al., 2006; Mantzourani et al., 2009; Ventura et al., 2009b; Kenyon and Osbak, 2014). However, very little is known at present concerning the genetic or biochemical characteristics of these bacteria that mediate their beneficial or pathogenic effects. In the present work, we have carried out detailed phylogenetic and comparative analyses of protein sequences from the genomes of Bifidobacteriales species to examine in depth their evolutionary relationships and also to identify molecular markers that are unique to these bacteria at multiple phylogenetic levels. Based on a robust and comprehensive phylogenetic tree for the Bifidobacteriales species based on 614 core proteins from the sequenced genomes, the following inferences regarding the evolutionary relationships among the Bifidobacteriales species could be made. (i) The sequenced Bifidobacteriales species appear to form two main clusters, a deeper clade consisting of the Scardovia-related genera (viz. Scardovia, Parascardovia and Alloscardovia) and another cluster grouping together Bifidobacterium and Gardnerella genera. (ii) Gardnerella vaginalisrather than branching separately is found to consistently branch in between different Bifidobacterium species.


FIGURE 4 | Example of 1 aa conserved signature indel in the protein triosephosphate isomerase that is specific for the Scardovia clade comprising of the genera Scardovia, Parascardovia, Metascardovia, and Alloscardovia. Information for other CSIs specific for this clade is presented in Table 3 and Supplementary Figures S36–S39.


FIGURE 5 | Partial sequence alignment of phosphogluconate dehydrogenase showing a 1 aa insertion that is specific for the B. longum cluster.

(iii) Within Bifidobacterium species, a number of distinct clusters, referred to as the B. asteroides, B. pseudolongum, B. longum, B. bifidum, B. adolescentis, B. pullorum, and B. boum groups, are observed as described in earlier work (Lugli et al., 2014; Sun et al., 2015). Of these clusters, the B. asteroides group forms the deepest branching lineage within the Bifidobacterium (Bottacini et al., 2012; Lugli et al., 2014; Sun et al., 2015).

The present work also identified large number of novel molecular signatures in the forms of CSIs and CSPs, which are specific characteristics of the members of the order Bifidobacteriales at multiple phylogenetic levels. Of these signatures, 32 CSIs and 10 CSPs are specific for the entire order Bifidobacteriales. The identified Bifidobacteriales-specific CSIs are present in assorted widely distributed proteins carrying out wide variety of cellular functions. All of the 10 Bifidobacterialesspecific CSPs are proteins of unknown functions. Given the specificity of these CSIs and CSPs for the Bifidobacteriales, the genetic changes leading to these molecular characteristics have likely occurred in a common ancestor of the Bifidobacteriales (Gao and Gupta, 2005, 2012). Additionally, our analyses have also identified many other molecular signatures (CSIs and CSPs), which independently support the existence of a number of clades of bifidobacteria that are consistently observed in phylogenetic trees. The clades identified by these molecular signatures include, (i) a clade encompassing the genera Scardovia, Parascardovia and Alloscardovia, (ii) signatures that are commonly shared by


FIGURE 6 | Conserved signature indels that are specific for the B. asteroides-related clades of the Bifidobacteriales. (A) Partial sequence alignment of the purine biosynthesis protein purH showing a 1 aa insertion which is specific for the *B. asteroides* cluster II species in the protein tree (Figure 1); (B) Excerpt from sequence alignment of the protein 5′ '-methylthioadenosine nucleosidase showing a 3 aa insertion that is specific for the *B. asteroides-*related cluster IV in the protein tree.

Bifidobacterium and Gardnerella species to the exclusion of other bifidobacteria, and (iii) signatures demarcating specific clusters of B. asteroides- or B. longum- related species.

The order Bifidobacteriales presently contains a single family, Bifidobacteriaceae. Based upon the results of phylogenomic studies and identified molecular signatures, it appears that the members of this order could be divided into two familylevel groups, one comprising of the Scardovia-related genera (viz. Scarodivia, Parascardovia, and Alloscardovia) and the other consisting of the genera Bifidobacterium and Gardnerella. However, genome sequence information for members of several newly described Scardovia-related genera (viz. Aeriscardovia, Neoscardovia, and Pseudoscardovia), is lacking at present (Simpson et al., 2004; García-Aljaro et al., 2012; Killer et al., 2013). In future studies, depending upon whether the species from these genera branch with the Scardovia-clade and their sharing of the molecular signatures specific for this clade, the possibility of dividing the order Bifidobacteriales into two or more families could be considered.

The genus Bifidobacterium, which is comprised of 49 species and subspecies, contains most of the recognized taxa within the order Bifidobacteriales. Although earlier phylogenetic studies have consistently observed 6–7 distinct clusters of Bifidobacterium species (Ventura et al., 2006, 2007b; Turroni et al., 2011; Lugli et al., 2014; Sun et al., 2015), due to lack of any other distinguishing characteristics, no attempt has been made to formally recognize any of these clusters. In our work, we have identified a number of molecular signatures that are either completely or largely specific for the members of two of these clusters (viz. the B. asteroides and B. longum groups). Of these clusters, the distinctness of the B. asteroides group (comprising of the species B. asteroides, B. indicum, B. coryneforme, B. actinocoloniiforme, B. sp. A11, and B. sp. 7101) which forms the deepest branching lineage within the Bifidobacterium, is supported by 2 CSIs and 4 CSPs that are uniquely shared by most of the members of this clade. Further, most of the species which are part of the B. asteroides clade have been isolated from the gastrointestinal tract of honey bees, and unlike other bifidobacteria, they are also capable of carrying out respiratory metabolism (Killer et al., 2010, 2011; Bottacini et al., 2012; Lugli et al., 2014; Sun et al., 2015). All of these characteristics indicate that the members of the B. asteroides clade are a good candidate for recognition as a distinct genus level taxon within the order Bifidobacteriales.

The molecular markers for the order Bifidobacteriales and some of its clades, in addition to their utility for taxonomic and diagnostic studies (Ahmod et al., 2011; Gupta, 2014; Wong et al., 2014), also provide important new tools for genetic and biochemical studies. Earlier work on a number of CSIs in the Hsp60 and Hsp70 proteins has established that both large and small CSIs in conserved proteins are essential for the group of organisms in which they are found (Singh and Gupta, 2009; Gupta, 2016b). Removal of these CSIs, or any significant change in them, was shown to be incompatible with the cellular growth of the CSI-containing organisms. Thus, the identified CSIs are predicted to play essential role in the organisms in which they are found. Structural studies on several studied CSIs show that the sequences corresponding to them are present in the

surface loops of the proteins (Singh and Gupta, 2009; Gupta and Khadka, 2016). Limited structural work on some of the Bifidobacteriales-specific CSIs that we have carried out also shows that these CSIs are located in the surface loops of the proteins. One example of the structural location of a Bifidobacterialesspecific CSIs is illustrated in **Figure 7**. In this case, a homology model of protein synthesis elongation factor Tu from B. longum was created to determine the location of the 4 aa Bifidobacterialesspecific CSI found in this protein. A structural comparison of the EF-Tu from B. longum and E. coli shown in **Figure 7** reveals that the CSI in the B. longum homolog is present in the protein surface loop within the GTPase domain of EF-Tu. The surface loops in proteins play important role in mediating proteinprotein or protein-ligand interactions and it is expected that the identified CSIs are involved in mediating novel interactions that are specific and essential for the CSI-containing organisms (Akiva et al., 2008; Hashimoto and Panchenko, 2010). Similar to the CSIs in the EF-Tu protein, our work has identified numerous other CSIs in different essential proteins, which are specific for the Bifidobacteriales species. Functional studies on proteins harboring these CSIs provide an important means for discovering novel biochemical characteristics that are unique to either all Bifidobacteriales or specific clades of these bacteria, and which could possibly also provide useful insights into the growth-promoting as well as pathogenic effects of some of these bacteria.

# AUTHOR CONTRIBUTIONS

GZ, BG, MA, BK carried out comparative analyses of the bifidobacteriales genomes to identify signatures reported here. ZG and MA constructed phylogenetic trees and BK carried out homology modeling of the protein sequences. BG, MA, and RG were responsible for the writing and editing of the manuscript. All of the work was carried out under the direction of RG.

#### ACKNOWLEDGMENTS

The work from RSG' lab was supported by research grant No. 249924 from the Natural Science and Engineering Research Council of Canada. Work by BG was supported in part by research grants from the National Science Foundation of China (31570011) and Natural Science Foundation of Guangdong Province (2015A030306039). BG is also a scholar

#### REFERENCES


of the "100 Talents Project" of the Chinese Academy of Sciences.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.00978


Eggerthellales ord. nov., containing the family Eggerthellaceae fam. nov. Int. J. Syst. Evol. Microbiol. 63, 3379–3397. doi: 10.1099/ijs.0.048371-0


history of an ancient phylum. Microbiol. Mol. Biol. Rev. 71, 495–548. doi: 10.1128/MMBR.00005-07


**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, Gao, Adeolu, Khadka and Gupta. 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.

# Microencapsulation in Alginate and Chitosan Microgels to Enhance Viability of Bifidobacterium longum for Oral Delivery

Timothy W. Yeung<sup>1</sup> , Elif F. Üçok<sup>1</sup> , Kendra A. Tiani1,2, David J. McClements1,3 and David A. Sela1,3,4 \*

<sup>1</sup> Department of Food Science, University of Massachusetts, Amherst, MA, USA, <sup>2</sup> Commonwealth Honors College, University of Massachusetts, Amherst, MA, USA, <sup>3</sup> Center for Bioactive Delivery, Institute of Applied Life Science, University of Massachusetts, Amherst, MA, USA, <sup>4</sup> Center for Microbiome Research, University of Massachusetts Medical School, Worcester, MA, USA

#### Edited by:

David Berry, University of Vienna, Austria

#### Reviewed by:

Borja Sanchez, Instituto de Productos Lácteos de Asturias – Consejo Superior de Investigaciones Científicas, Spain Chiara Ferrario, University of Parma, Italy

> \*Correspondence: David A. Sela davidsela@umass.edu

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 26 February 2016 Accepted: 27 March 2016 Published: 19 April 2016

#### Citation:

Yeung TW, Üçok EF, Tiani KA, McClements DJ and Sela DA (2016) Microencapsulation in Alginate and Chitosan Microgels to Enhance Viability of Bifidobacterium longum for Oral Delivery. Front. Microbiol. 7:494. doi: 10.3389/fmicb.2016.00494 Probiotic microorganisms are incorporated into a wide variety of foods, supplements, and pharmaceuticals to promote human health and wellness. However, maintaining bacterial cell viability during storage and gastrointestinal transit remains a challenge. Encapsulation of bifidobacteria within food-grade hydrogel particles potentially mitigates their sensitivity to environmental stresses. In this study, Bifidobacterium longum subspecies and strains were encapsulated in core-shell microgels consisting of an alginate core and a microgel shell. Encapsulated obligate anaerobes Bifidobacterium longum subsp. infantis and Bifidobacterium longum subsp. longum exhibited differences in viability in a strain-dependent manner, without a discernable relationship to subspecies lineage. This includes viability under aerobic storage conditions and modeled gastrointestinal tract conditions. Coating alginate microgels with chitosan did not improve viability compared to cells encapsulated in alginate microgels alone, suggesting that modifying the surface charge alone does not enhance delivery. Thus hydrogel beads have great potential for improving the stability and efficacy of bifidobacterial probiotics in various nutritional interventions.

Keywords: microencapsulation, bifidobacteria, probiotics, simulated digestion, oral delivery

# INTRODUCTION

Beneficial bacteria are often incorporated into functional foods and nutritional interventions to be ingested orally as probiotics. This includes humans and livestock animals that receive direct-fed microbials to enhance health outcomes and reduce pathogen load (Braat et al., 2006; Puccio et al., 2007; Neal-McKinney et al., 2012; Watson and Preedy, 2015). Bifidobacterium longum colonizes the human gastrointestinal tract (GIT) and is one of the 48 recognized taxa that are encompassed within the genus Bifidobacterium (Milani et al., 2014; Sun et al., 2015). This obligate anaerobe is one of the earliest colonizers of the infant GIT, and is present in lower concentrations in the adult gut (Schell et al., 2002; Sela et al., 2008). The bifidobacterial taxa longum, infantis, and suis were previously classified as discrete species, but more recently they were reclassified as subspecies of B. longum (Sakata et al., 2002). Their unification as a single species is based primarily on genomic

and phenotypic similarities shared between these groups. While Bifidobacterium animalis subsp. lactis is often used for probiotic applications, B. longum strains are of particular interest due to its likely co-evolution with humans. This is evident in B. longum utilization of human milk oligosaccharides and establishes a protective gut microbiome in infants through adulthood (Sela and Mills, 2010).

Bifidobacterium longum is deployed in several probiotic applications using a variety of delivery formats (Adhikari et al., 2000; Fortin et al., 2011; Amine et al., 2014; Lewis et al., 2015). A relatively large dose of probiotics is recommended to impart health benefits, typically 106–10<sup>7</sup> CFU/g per day (Krasaekoopt et al., 2003; Roy, 2005). However, the direct incorporation of free probiotic cells into food products and supplements results in a significant decrease in cell viability throughout storage and gastrointestinal transit (Sultana et al., 2000; de Vos et al., 2010). Therefore, prolonged storage and the process of ingesting these probiotics may reduce their viability below recommended levels to achieve health benefits. Microencapsulating probiotic cells within hydrogel matrices protects them against extrinsic environmental factors thereby enhancing bacterial survival during processing, storage, and digestion (de Vos et al., 2010; Fareez et al., 2015; Yeung et al., 2016). Encapsulation may also dictate the controlled release of the probiotic at the precise anatomical site of activity within the GIT, thereby enhancing the efficacy of the probiotic through specific targeting after oral delivery (de Barros et al., 2015; Zhang et al., 2015b).

Several biopolymer materials are available to encapsulate microbes in hydrogel matrices, depending on the desired physicochemical properties of the delivery vehicle. The most commonly used food-grade biopolymers are proteins (e.g., whey proteins and caseins) and carbohydrates (e.g., starch and gums; Bagchi et al., 2010; Gaonkar et al., 2014; Etchepare et al., 2015). For many food applications, it is advantageous to encapsulate probiotics within hydrogel beads that trap bacteria within small particles containing cross-linked biopolymer molecules. These microgels must be engineered to encapsulate high concentrations of probiotics and protect them from environmental stresses, such as acidic pH, bile salts, and digestive enzymes (Zhang et al., 2015a). Alginate has been widely used as a biopolymer suitable for food applications as it is relatively inexpensive, easy to gel, biodegradable, and compatible with many food systems (Gombotz and Wee, 2012; Lee and Mooney, 2012). Indeed, recently studies have shown that lactococcal-based probiotics can be encapsulated within alginate microgels to improve their stability (Yeung et al., 2016).

There are appreciable differences between probiotic strain tolerance toward environmental and gastrointestinal stresses. Consequently, it is possible to identify particular strains that are more resistant to these stresses than others, which are therefore more suitable for commercial application (Godward et al., 2000; Krasaekoopt et al., 2004; Capela et al., 2006). As an anaerobe, bifidobacterial species including B. longum differ in their sensitivity to oxygen exposure and other environmental stresses during the preparative phase prior to probiotic deployment (Kawasaki et al., 2006; Ruiz et al., 2012). Therefore, bifidobacterial probiotics may be encapsulated to restrict oxidative damage during preparation and storage and to limit exposure to degradative processes within the GIT.

The aim of this study was to design, fabricate, and characterize a food-grade encapsulation system to protect B. longum cells during simulated storage and gastrointestinal passage. Previously, we demonstrated that encapsulation of probiotics within alginate microgels could improve their viability during storage (Yeung et al., 2016). In the current study, we encapsulated B. longum cells within alginate beads to determine if their viability could be enhanced in storage and gastrointestinal transit. Moreover, the impact of coating these alginate beads with a layer of chitosan was investigated as well. Chitosan coated alginate beads have previously been used to enhance the mucoadhesive properties of probiotic bacteria (Chen et al., 2013).

## MATERIALS AND METHODS

#### Preparation of Bacterial Cultures

Four strains of both Bifidobacterium longum subsp. longum (B. longum) and Bifidobacterium longum subsp. infantis (B. infantis) were studied (**Table 1**). All strains were originally isolated from infant feces. Stock solutions were maintained by storing bacteria at −80◦C in deMann, Regosa, Sharpe (MRS) media with 0.05% L-cysteine in 25% glycerol. Bacteria were propagated in MRS with <sup>L</sup>-cysteine at 37◦C for 24 h, checked for purity, and maintained on MRS agar anaerobically. Anaerobic conditions were maintained in a double chamber anaerobic hood with an airlock (88% N2, 10% CO2, and 2% H2) from Coy Laboratory Products (Grass Lake, Mississippi, USA).

Isolated colonies were routinely propagated in MRS broth (50 mL) for 40 h at 37◦ C. Cells were harvested by centrifugation at 4000 × g for 10 min, washed twice with 0.85% NaCl (physiological saline) solution (25 mL), and suspended in 0.85% NaCl (2 mL). The resulting cell suspensions were used either directly for assessing survival of free cells (i.e., no encapsulation) or subjected to encapsulation as described in section "Microencapsulation of Bifidobacterial Cells." Free cell suspensions (2 mL) were stored in 0.85% NaCl solution (50 mL) at 2–5◦C for up to 5 weeks to model long-term storage conditions.



# General Chemicals Used in Encapsulation and Modeled Digestion

For bacterial culture preparation, MRS broth was obtained from Becton Dickinson and Company (Sparks, MD, USA). Agar, L-cysteine hydrochloric acid, and sodium chloride (NaCl) were purchased from Sigma–Aldrich (St. Louis, MO, USA). Glycerol and sodium citrate dihydrate was purchased from Fisher Scientific (Fair Lawn, NJ, USA).

For encapsulation experiments, sodium alginate (TICA-algin HG 400 powder) was donated by TIC Gums (White Marsh, ML, USA). Calcium chloride hexahydrate, chitosan (medium molecular weight) was obtained from Sigma-Aldrich. Glacial acetic acid was purchased from Fisher Scientific.

For simulated digestion, ammonium nitrate, bile extract porcine, lipase from porcine pancreas type II, pepsin from porcine gastric mucosa, porcine gastric mucin type II, potassium chloride, potassium citrate, potassium phosphate, sodium DLlactate, sodium hydroxide (NaOH), and uric acid sodium salt were also purchased from Sigma-Aldrich. Hydrochloric acid (HCl), phosphate buffer saline (PBS), and urea were purchased from Fisher Scientific.

### Microencapsulation of Bifidobacterial Cells

Bifidobacteria were encapsulated within alginate microgels using an injection–gelation method (Whelehan and Marison, 2011; Seiffert, 2013). Briefly, 1% (w/v) sodium alginate solution was prepared, autoclaved, and then cooled to ambient temperature. The sterile alginate solution (198 mL) was mixed with 2 mL of ∼10<sup>9</sup> CFU/mL probiotic organisms suspended in physiological saline. The polymeric solution was agitated to uniformly distribute cells throughout the mixture. The alginate beads were prepared aseptically using an encapsulator (Büchi B-390 <sup>R</sup> , Büchi Labortechnik AG, Flawil, Switzerland) with a nozzle size of 120 µm, using the manufacturer's standard operating conditions (amplitude 3, frequency 800 Hz, electrode 800 V, pressure 250– 300 mbar). Aliquots of probiotic/alginate solution were injected into 0.1 M calcium chloride solution (350 mL). After 1-h gelation under agitation, the resulting calcium alginate beads were collected by filtration, rinsed with sterile deionized water (200 mL), and re-filtered. Microbeads (∼30 mL) were stored in physiological saline solution (50 mL) at 4◦C for up to 4 weeks to model long-term storage conditions. This process was repeated for all eight strains of bifidobacteria.

Unfilled alginate beads were prepared identically but without the addition of bacterial strains to the alginate solution. 1% alginate solution (200 mL) was extruded into of 0.1 M CaCl<sup>2</sup> (350 mL) solution under continuous agitation. The working parameters (nozzle diameter, frequency, charge, and pressure), filtering steps and storage conditions used were the same as those for the preparation of filled alginate beads.

An aqueous chitosan solution (0.4% w/v) was prepared as described previously by Zhou et al. (1998). Briefly, chitosan (0.4 g) was dissolved in distilled water (90 mL) and glacial acetic acid (0.8 mL). The pH was adjusted to 5.0–5.1 with NaOH, and the total volume was adjusted to 100 mL. The solution was autoclaved and filtered to remove undissolved solids. Subsequently, the alginate beads were submerged in the chitosan solution to provide a secondary coating by electrostatic attraction of the cationic chitosan molecules to the surfaces of the anionic alginate beads. The mixture was agitated for 1 h before filtering and rinsing beads with sterile distilled water. Chitosancoated alginate beads were then stored and analyzed as described in section "Alginate and Chitosan-coated Alginate Microbead Characterization."

#### Alginate and Chitosan-Coated Alginate Microbead Characterization Particle Size Distribution

The particle size distribution was determined by static light scattering (Mastersizer S, Malvern Instruments, Worcestershire, UK). Each sample (1–2 mL) was suspended in distilled water (10 mL) and vortexed to avoid multiple scattering effects and to ensure homogeneity prior to analysis. Volume-weighted (D [4,3]) and surface-weighted (D [3,2]) mean particle diameters were obtained for all samples.

#### Optical Microscopy Characterization

The overall appearance of alginate and chitosan-coated alginate beads was characterized with an optical microscope (C1 Digital Eclipse, Nikon, Tokyo, Japan). Microgel suspensions (1–2 mL) were immersed in physiological saline (10 mL) and vortexed to separate individual beads. Optical images were obtained using a digital camera and further analyzed using the instrument software (EZ CSI version 3.8, Nikon).

#### Scanning Electronic Microscopy (SEM)

The bead microstructure was characterized using a benchtop scanning electron microscope (JCM-6000 NeoScope, JEOL, Tokyo, Japan). To prepare the samples prior to analysis, alginate, and chitosan-coated alginate beads were freeze-dried and sputtercoated with gold (10 nm) before loading onto the microscope. Images of the microgels were documented in representative fields.

#### Electrical Properties

The surface potential (ζ-potential) of alginate and chitosancoated alginate microgels was evaluated by electrophoretic light scattering (Zetasizer Nano ZS, Malvern Instruments, Worcestershire, UK). For each sample, refrigerated microgels (1–2 mL) were suspended in distilled water (10 mL) and vortexed to separate the beads. Samples were then loaded into the measurement cells and analyzed.

#### Modeled Long-term Storage Conditions of Encapsulated Bifidobacteria

Total cell counts of free and encapsulated bifidobacteria were determined by a modified drop plate method as previously described (Herigstad et al., 2001). Briefly, 10 drops (10 µL) of a dilution within a series (100–10<sup>7</sup> ) were deposited on MRS agar plates and counted after incubation under anaerobic conditions at 37◦C.

To determine viable counts of the encapsulated bacteria, beads (1 mL) were re-suspended in 10% sodium citrate dihydrate solution (9 mL) followed by vortexing. The number of released cells was determined by plate count using MRS agar, dilutions of dissolved beads (10−1–10−<sup>7</sup> ) were plated in duplicate and incubated at 37◦C anaerobically for 40 h. For lower viability samples later, beads (2 mL) were re-suspended in 10% sodium citrate dihydrate solution (2 mL) instead, and dilutions (100– 10−<sup>3</sup> ) were plated as before. Samples were taken over a 4-week period on days 0 (initial), 1, 3, 5, 7, 10, 14, 21, and 28. Day 24 was also plated for free cell samples.

#### In Vitro Simulated Digestion of Alginate and Chitosan-Coated Alginate Microbeads

Fluids used in in vitro modeling of digestion were prepared based on the method described by Li et al. (2011). One liter of modeled saliva stock solution was prepared with ammonium nitrate (0.328 g), potassium chloride (0.202 g), potassium citrate (0.308 g), potassium phosphate (0.636 g), sodium chloride (1.594 g), sodium DL-lactate (0.146 g), urea (0.198 g), and uric acid sodium salt (0.021 g) in distilled water. The stock solution was then filter-sterilized. The day before digestion experiments were carried out, the salivary phase was prepared by adding porcine gastric mucin type II (2.4 g) to saliva stock solution (80 mL). The solution was stirred overnight at room temperature to completely dissolve the powder.

One liter of simulated gastric stock solution was prepared by adding sodium chloride (2 g) and 6 M hydrochloric acid (7 mL) to distilled water and filter sterilizing. The simulated intestinal stock solution (500 ml) was prepared by adding calcium chloride hexahydrate (27.38 g) and sodium chloride (109.685 g) to distilled water and autoclaved. Pepsin extracted from porcine gastric mucosa (0.32 g) was then added to gastric stock solution (100 mL).

The day before digestion experiments were carried out, porcine bile extract (0.75 g) was added to PBS solution (14 mL) for the modeled intestinal phase. The solution was stirred overnight at room temperature to completely dissolve the powder. Lipase from porcine pancreas type II (0.24 g) was dissolved in PBS solution (10 mL); the solution (5 mL) was then added with bile salt solution (7 mL) and intestinal stock solution (33 mL).

Free and encapsulated bifidobacteria cells were separately added to simulated saliva fluids (22 mL, pH adjusted to 6.7– 6.8), simulated gastric fluids (45 mL, pH adjusted to 2.5–2.6) or simulated intestinal fluids (45 mL, pH adjusted to 7.0–7.2) and stored in a shaking incubator (MaxQ 6000, Thermo Scientific, Waltham, MA, USA) set at 37◦C with a shaking speed of 110 rpm (Supplementary Figure S1). Dilutions (100–10<sup>5</sup> ) were plated on MRS agar for initial, 5, 10, 15, or 30 min exposure and incubated anaerobically for at least 48 h.

#### Statistical Analysis

The mean of two or three individual determinations was used to calculate particle size, ζ-potential. The mean of 10 replicate drops was used to calculate cell counts. Analysis of variance (ANOVA) followed by Tukey honest significant difference test was use to analyze all data and compare individual means. This was performed using statistical software (GraphPad Prism 6, GraphPad Software, La Jolla, CA, USA).

# RESULTS

# Particle Size Analysis of Alginate and Chitosan-Coated Alginate Microgels

Light scattering was used to determine the mean particle diameter of the different microgel samples (**Table 2**). The mean particle sizes of alginate beads containing similar strains were similar, ranging from 135 to 185 µm (D [3,2]) for encapsulated B. infantis strains and 149–216 µm (D [3,2]) for encapsulated B. longum strains. The chitosan-coated alginate beads were significantly larger compared than the alginate beads, ranging from 191 to 292 µm (D [3,2]). This increase in particle size may have been because of the additional coating formed by the alginate molecules, or because of some aggregation of the microgels. Microgel aggregation may have occurred due to bridging flocculation, which is the ability of the chitosan cation to adsorb to the surfaces of two or more anionic alginate beads. Additional information regarding the structural configuration of the microgels was therefore obtained through microscopy.

#### Optical Microscopy of Alginate and Chitosan-Coated Alginate Microbeads

The structures of samples containing free cells or bacterialloaded microgels were determined using optical microscopy

TABLE 2 | Volume-based (D [4,3]) and surface-based (D [3,2]) mean particle diameters measured by static light scattering alginate and chitosan-coated alginate beads with strains of bifidobacteria.


Values are shown as average particle size ± standard deviation. Values followed by the same letters in the same column are not significantly different (p > 0.05) from each other.

immediately after encapsulation (**Figure 1**). Free cells appeared rod-shaped as expected for bifidobacteria (**Figures 1A,D**). The unfilled alginate and chitosan-coated alginate microgels were similar in morphology, although the individual coated alginate beads did appear larger than the uncoated ones, which is consistent with the particle size analysis (**Figures 1C,F**). Encapsulated bifidobacteria were clearly visualized within the microgels for both alginate and chitosan-coated alginate microgels (**Figures 1E,F**). The bifidobacterial-loaded alginate and chitosan-coated alginate beads had a similar external appearance as the equivalent unloaded beads. The microgels were generally spherical with diameters around 100–300 µm for all samples.

#### Scanning Electron Microscopy

Scanning electron microscopy (SEM) was used to inspect the structure of the alginate and chitosan-coated alginate beads (**Figure 2**). Freeze-dried microgels were uniform in size and shape. However, the surfaces of the microgels observed by SEM appeared wrinkled, whereas they presented as smooth when observed by optical microscopy. This is likely due to sublimation of water originally trapped within the hydrogel matrix, as has been described previously (Yeung et al., 2016). The chitosancoated alginate beads appeared to be more irregular in shape compared to alginate beads. Qualitatively, the alginate beads had smoother wrinkles and microstructures, whereas the chitosancoated beads exhibited sharp jagged edges. This observation suggests that the chitosan layer has been successfully deposited onto the external surfaces of the alginate microgels.

# Zeta Potential Analysis of Microencapsulated Bifidobacteria

Electrophoretic light scattering was used to evaluate the electrical characteristics of the microgels (**Table 3**). The ζ-potentials of all the alginate beads were negative, ranging from −4.2 to −9.4 mV for B. infantis and −2.6 to −4.4 mV for B. longum as predicted with this coating. In contrast, all chitosan-coated alginate bead samples had positive surface potentials ranging from +9.9 to +14.9 mV for B. infantis and +0.8 to +9.0 mV for B. longum. These results indicate that the cationic chitosan molecules formed a secondary shell around the anionic calcium alginate beads.

# Survival of Bifidobacterial Strains during Long-term Storage

#### Non-encapsulated Bifidobacterial Cells

The viability of four B. longum and four B. infantis strains that were not encapsulated was determined during 5 weeks of storage

FIGURE 1 | Optical microscope images of (A) Bifidobacterium longum subsp. infantis UMA 300 (20×), (B) unfilled alginate bead (20×), (C) filled alginate bead with B. longum subsp. longum UMA 306 (20×), (D) B. longum subsp. longum UMA 318 (20×), (E) unfilled chitosan-coated alginate bead (20×), (F) filled chitosan-coated alginate bead with B. longum subsp. infantis UMA 299 (20×). Scale bars represent 100 µm.

FIGURE 2 | Scanning electron micrographs of (A) B. longum subsp. infantis UMA299, (B) unfilled alginate bead, (C) unfilled chitosan-coated alginate bead, (D) B. longum subsp. longum UMA 306, (E) alginate bead containing B. longum subsp. longum UMA 401, (F) chitosan-coated alginate bead containing B. longum subsp. longum UMA 300. Samples were dried before sputter-coating with gold. SEM was set at high-vacuum, 10 kV.

(**Figure 3A**; Supplementary Table S1). As expected, there was a decrease in the viability of the bifidobacteria evaluated, but the rate of the decrease was strain dependent. A sharp decrease in viability was observed for B. infantis UMA318 and B. longum UMA401, diminishing by 9–10 log CFU over the course of a week under aerobic conditions. B. infantis UMA 300 and B. infantis UMA 305 remained viable for slightly longer, with a 10-log reduction observed within 10 days. Whereas, B. infantis UMA 298 and B. infantis 306 exhibited a 9–10 log decrease over 2 weeks of storage. Interestingly, B. infantis UMA 299 and B. longum UMA 402 survived the longest, as viable cell counts diminished by 7–8 logs over 3 weeks before decreasing to undetectable levels.

#### Encapsulated Bifidobacterial Cells

Viability following encapsulation was determined for all eight bifidobacterial strains (**Figure 3B**; Supplementary Table S2). There were distinct differences between the effects of encapsulation depending on strain type. The cell viability of B. infantis UMA 298, B. infantis UMA 305, and B. longum UMA 318 stains rapidly decreased and were undetectable after 3 days. Unexpectedly, B. infantis UMA 298 and B. infantis UMA 305 in alginate were inactivated faster than the corresponding free cells, TABLE 3 | Zeta potential of alginate and chitosan-coated alginate beads with strains of bifidobacteria.


Values are shown as average ζ-potential ± standard deviation. Values followed by the same letters are not significantly different (p > 0.05) from each other.

being undetectable after 24 and 10 days, respectively. Viability of the encapsulated B. longum UMA 318 was identical to that of free cells. B. infantis UMA 300, B. longum UMA 306, B. longum 401,

and B. longum 402 maintained viable populations that decreased by 3–4 log CFU after 10 days of storage before rapidly decreasing to zero. Encapsulated B. infantis UMA 300 survived 3 days longer than corresponding free cells; encapsulated B. longum UMA 401 lasted a week longer than free cells. Encapsulated B. longum UMA 306 survived similarly to free cells, and encapsulated B. longum UMA 402 survived over a week shorter than corresponding free cells. Interestingly, B. infantis UMA 299 viability was enhanced as it experienced a 5 log CFU reduction in 3 weeks compared to an 8 log CFU reduction during this time for the non-encapsulated cells. Thus, encapsulating with alginate extended cell viability of B. infantis UMA 299 and 300 by a few days, and extended viability of B. longum UMA 401 cells by a week. Encapsulating other B. infantis and B. longum strains

did not appear to extend viability over the storage conditions used.

Two strains each of B. infantis (UMA 299 and 300) and B. longum (UMA 401 and 402) were encapsulated in a secondary coating of chitosan applied to the alginate bead core and submitted to testing over time (**Figure 3C**; Supplementary Table S3). B. longum UMA 401 and 402 both decreased 3–4 log within 3 days, and fell to undetectable levels by 2 weeks. Viability of B. infantis UMA 299 and B. infantis UMA 300 decreased only 2 logs in 5 days, before falling to undetectable levels after 2 weeks. Encapsulating B. infantis UMA 299, B. infantis UMA 300, and B. longum UMA 401 cells in chitosan-coated alginate beads did not appear to extend viability compared with uncoated alginate microbeads. B. longum UMA 402 cells in chitosan-coated alginate decreased 2.4 log CFU, whereas B. longum UMA 402 cells in alginate alone decreased 7.2 log CFU, between day 3 and day 14. Hence, encapsulating B. longum UMA 402 in chitosancoated alginate extended detectible viability 4 days more than encapsulating B. longum UMA 402 in alginate alone.

# Survival of Encapsulated Cells during Simulated Digestion

Free and bifidobacterial cells encapsulated in chitosan-coated alginate beads were subjected to simulated digestion in a GIT model as previously described (Li et al., 2011). Free B. infantis UMA 299 and cells encapsulated chitosan-coated alginate were immersed separately in simulated salivary, gastric, and intestinal phases and assessed over time for cell viability (**Table 4**). The strain was selected based on its high viability during storage in free and encapsulated forms (**Table 3**; Supplementary Tables S1–S3). The bacteria appeared to be relatively stable within simulated saliva fluids, as less than a 1 log CFU reduction was experienced in 30 min of exposure regardless of encapsulation. The model salivary juice did not greatly inhibit cell viability in general, as less than one log CFU reduction was experienced in 30 min of exposure regardless of encapsulation. However, microencapsulation provided enhanced protection for UMA299 by shielding the strain from the low pH of the gastric phase. Encapsulated cells decreased by 1.4 logs CFU, whereas untreated cells decreased by 2.7 logs following exposure to pH 2.5 conditions (5 min). This indicates a significant, albeit fleeting protection afforded to the encapsulated cells as viability was abrogated after 10 min of exposure to the gastric phase. Similarly, UMA299 cell viability was not detectible after 5 min of exposure to the intestinal phase. B. longum UMA 402 encapsulated in chitosan-coated alginate was also subjected to simulated digestion in preliminary tests (data not shown). As with B. infantis, cell viability remained stable in the modeled salivary phase, but underwent a 6 log reduction after only a few minutes exposure to gastric phase (pH 2.5).

# DISCUSSION

Initially, chitosan-coating of alginate beads was postulated to enhance the viability of encapsulated probiotics by reducing their exposure to environmental stresses during storage and within the GIT (Kamalian et al., 2014). Accordingly, the influence of encapsulation on a panel of B. longum strains to assess differential viability was systematically studied. The calcium alginate beads formed using an injection–gelation method were roughly spherical in shape, negatively charged, and had dimensions around 130–220 µm. Coating the alginate beads with chitosan caused a small increase in their size and changed their charge from negative to positive. Optical microscopy (**Figures 1C,F**) confirmed that the bifidobacteria were immobilized within the hydrogel beads, which is consistent with previous encapsulation studies (Hansen et al., 2002; Fareez et al., 2015; Yeung et al., 2016).

Interestingly, encapsulation of bifidobacteria in chitosancoated alginate beads led to decreased improvement in their storage or gastrointestinal stability compared with cells in alginate beads. One possible explanation for this observation is that the alginate hydrogel used had relatively large pores, and so small molecules, such as oxygen, acids, bile salts, or digestive enzymes, could easily diffuse into the microgels and inactivate the encapsulated bacteria (McClements, 2015). These results suggest that a simple secondary layer of chitosan alone


Counts based on samples drop-plated on MRS agar and incubated at 37◦C anaerobically. Values are shown as mean cell number ± standard error of replicate counts (n = 10). Means within each column followed by the same lowercase letters are not significantly different (p > 0.05) from each other. Means within each row followed by the same uppercase letters are not significantly different (p > 0.05) from each other.

will not fully protect encapsulated bifidobacteria, and that further optimization is required to engineer more effective delivery systems. Previous studies have shown that alginate has a prebiotic effect on bifidobacteria, which might account for its ability to enhanced viability, potentially through a non-encapsulation mechanism (Wang et al., 2006; Ramnani et al., 2012). In future studies, it may be useful to examine the influence of different biopolymer materials and methods on the ability of microgels to enhance probiotic viability. As an example, the hydrogel pore size may be decreased to limit molecular diffusion, with the addition of anti-oxidants to limit oxidation reactions and prebiotics to stimulate probiotic growth in the colon. Since bifidobacteria ferment oligosaccharides within the gut, a synbiotic approach that integrates prebiotic substrates including plant or milk oligosaccharides may advance bifidobacterialbased delivery (Sela, 2011). Alternatively, judicious selection of strain selection that are resistant to acids, bile salts, or digestive enzymes may enhance the delivery scheme. However, previous studies indicate that most bifidobacteria strains typically exhibit a significant decrease in survival around pH 4 which would necessitate shielding from gastric conditions (Sun and Griffiths, 2000).

Bifidobacteria have been exposed to simulated digestive fluids in previously conducted studies (O'Riordan et al., 2001; Hansen et al., 2002; Kamalian et al., 2014). Although specific strains tested and experimental schemes vary between studies. Hansen et al. (2002) encapsulated several bifidobacterial strains in microgels formed by an emulsion-templating method, and then exposed them to simulated gastric and small intestinal fluids. In this study, B. infantis and B. longum strains showed a 4–6 log CFU/mL decrease between exposure to gastric fluids set at pH 6.0 and pH 2.0 for 2 h, and 3–5 log CFU/mL reduction between exposure to intestinal fluid containing 0 and 1% bile for 24 h. Hansen et al. (2002) also encapsulated B. longum experienced a 5-log CFU/mL reduction after 30 min exposure to gastric juice (pH 2.0). In the study herein, B. infantis UMA 299 encapsulated in chitosan-coated alginate underwent an 8 log reduction in a 10-min exposure to gastric fluid (pH 2.6), and an 8-log reduction in 5 min exposure to intestinal fluid (**Figure 3C**; Supplementary Table S3). This study included 0.75% bile extract, pepsin, and lipases in the gastric and intestinal fluids for the purpose of simulating the harsh conditions of the human GIT. The bifidobacterial general stress response has been studied (Sánchez et al., 2007; Zomer et al., 2009). As with the phylogenetically dissimilar lactic acid bacteria, bifidobacteria employ ATPases to pump protons from the cell when exposed to acidic conditions (Matsumoto et al., 2004; Ventura et al., 2004). In addition, when exposed to bile during gastrointestinal transit, certain bifidobacterial strains deploy bile salt hydrolase to promote cell survival in the small intestine (Ruiz et al., 2014).

In an additional study, an emulsion encapsulation method was performed on B. pseudocatenulatum G4 in chitosan-coated alginate and exposed to gastric conditions (pH 1.5) for 2 h followed by intestinal phase for 5 h (Kamalian et al., 2014). The encapsulated B. pseudocatenulatum experienced a 4-log reduction when encapsulated in alginate and a 2-log reduction in chitosan-coated alginate, relative to the 5-log reduction in the control. However, this was accomplished in the absence of digestive enzymes or bile salts in simulated gastric and intestinal fluids that would present additional hurdles to the bifidobacterial cells. O'Riordan et al. (2001) studied spray-dried Bifidobacterium spp. PL1 in starch and subjected the resultant granules to simulated digestion. After 3 h of exposure to buffer with pH 2.8, they were unable to detect viable cells as well as other sampling points in between 0 and 3 h. This is consistent with the results presented in this study.

In summary, bifidobacterial viability following encapsulation varied between subspecies as well as strains. This suggests that there is a range of genotypic and phenotypic factors contributing to stress responses that promote enhanced viability. Further functional genomic analysis of encapsulated probiotic organisms can aid in matching strains with the particular encapsulation process to optimize cell integrity during storage. Moreover, similar approaches may be used in selecting ideal delivery vehicles to shield bifidobacteria during GIT transit to arrive intact and metabolically poised to exert beneficial activities in the distal colon. Subsequent formulations may optimize delivery vehicles by incorporating antioxidants and cryoprotectants within the encapsulation gel matrix to preserve bifidobacterial cell viability.

#### AUTHOR CONTRIBUTIONS

TY, DM, and DS conceived the experimental plan. TY conducted laboratory experiments, data analysis, and drafted the manuscript. EÜ and KT assisted with experiments, analyses, and contributed to the manuscript. DS supervised execution of the experimental plan, analyzed data, and critically reviewed the final manuscript. All authors read and approved the manuscript prior to submission.

#### ACKNOWLEDGMENTS

The authors thank TIC Gums for donating sodium alginate for the microencapsulation experiments. We thank David Prodanas for assistance with freeze-drying samples prior to SEM and Ezgi Özcan for assistance with statistical analysis. We also thank Jean Alamed and various individuals at the University of Massachusetts for helpful advice. EÜ thanks the Scientific and Technological Research Council of Turkey (TUBITAK) for financial support under Grant BIDEB 2214/A.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.00494

# REFERENCES

fmicb-07-00494 April 15, 2016 Time: 15:39 # 10



**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 Yeung, Üçok, Tiani, McClements and Sela. 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.

# Modulation of the Bifidobacterial Communities of the Dog Microbiota by Zeolite

Alberto Sabbioni<sup>1</sup>† , Chiara Ferrario<sup>2</sup>† , Christian Milani<sup>2</sup> , Leonardo Mancabelli<sup>2</sup> , Enzo Riccardi<sup>1</sup> , Francesco Di Ianni<sup>1</sup> , Valentino Beretti<sup>1</sup> , Paola Superchi<sup>1</sup> and Maria C. Ossiprandi<sup>1</sup> \*

<sup>1</sup> Department of Veterinary Medical Science, University of Parma, Parma, Italy, <sup>2</sup> Laboratory of Probiogenomics, Department of Life Sciences, University of Parma, Parma, Italy

During last decades canine health and well being is becoming an important issue for human owners. In dogs, several factors including diet, pathogenic bacterial and stress conditions can affect the composition of the gut microbiota. In this study, we evaluated the effect of dietary chabazitic zeolitite (CZ) supplementation on the contribution of bifidobacteria to the fecal microbiota in training hunting dogs. Fecal microbiota cataloging based on 16S rRNA microbial profiling analyses highlighted an increase of Lactobacillus and Bifidobacterium in animals treated with CZ, with a simultaneous decrease of pathogens associated with dog gastrointestinal infections, such as Klebsiella and Enterobacter. A detailed profiling of the bifidobacterial population of dogs receiving CZ based on the ITS-based sequencing approach, revealed an enhancement bifidobacterial of species typical of animals such as Bifidobacterium animalis and B. pseudolongum. Moreover, these analyses identified the occurrence of putative new bifidobacterial taxa in both treated and untreated samples.

Keywords: dogs, gut microbiota, chabazitic zeolitite, Bifidobacterium, adsorptive capacity

## INTRODUCTION

Pet population is increasing in western countries, and dogs are the major human companions. Mutual interest has evolved into companion animals being a stable part of human life and therefore, the health and wellbeing of pets have increasingly raised interest during last decades. During history, the dog diet has changed, starting from a carnivorous behavior and a high protein diet (Clauss et al., 2010) to a carbohydrate rich diet and an urban life-style.

Despite the long span history of human-dog co-evolution, the knowledge of canine intestinal microbiota composition is much less complete than for humans. The dog gastro-intestinal tract (GIT) represents a rich ecosystem, composed of a wide range of metabolically active microorganisms (Simpson et al., 2002; Suchodolski et al., 2008; Kerr et al., 2013a). The predominant bacterial phyla in the colon and faeces of dogs are represented by Firmicutes (40–60%), Bacteroidetes (5–10%), Proteobacteria (15–20%), and Fusobacteria (5%) (Kerr et al., 2013b; Deng and Swanson, 2015), representing approximately 99% of the gut microbiota in dogs. However, very little is known about the occurrence of healthy promoting microorganisms such as bifidobacteria in the gut especially using metagenomics based approaches (Gavini et al., 2006; Jia et al., 2010).

#### Edited by:

Francesca Turroni, University College Cork, Ireland

#### Reviewed by:

Claudio Hidalgo Cantabrana, Dairy Research Institute of Asturias, IPLA-CSIC, Spain Francesca Bottacini, University College Cork, Ireland

#### \*Correspondence:

Maria C. Ossiprandi mariacristina.ossiprandi@unipr.it †These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 02 August 2016 Accepted: 07 September 2016 Published: 22 September 2016

#### Citation:

Sabbioni A, Ferrario C, Milani C, Mancabelli L, Riccardi E, Di Ianni F, Beretti V, Superchi P and Ossiprandi MC (2016) Modulation of the Bifidobacterial Communities of the Dog Microbiota by Zeolite. Front. Microbiol. 7:1491. doi: 10.3389/fmicb.2016.01491

Bifidobacteria are Gram positive bacteria that colonize different ecological niches, but represents one of the dominant colonizers of mammals at the very early first stages of life (Milani et al., 2015). The analyses of the gut microbiota of different mammals indicate that some bifidobacterial species, usually detected in the human GIT, were also identified in many other animals (Lamendella et al., 2008). For example, Bifidobacterium bifidum, B. adolescentis, B. catenulatum, and B. dentium are human-type bifidobacteria (Duranti et al., 2015, 2016), but these taxa displayed a cosmopolitan ecological behavior among different mammals (Lamendella et al., 2008).

In hunting dogs, emotional stress to which they are submitted during the training, can alter the habitat of the GIT (Rutgers et al., 1996). Therefore, to keep a suitable function of the GIT through appropriate feeding strategies is interesting, to avoid the intestinal colonization by enteropathogens (e.g., Escherichia coli, Salmonella ssp., Clostridium perfringens, C. difficile) (McKenzie et al., 2010; Kerr et al., 2013a).

To avoid antibiotic therapies, alternative products are under investigation. Zeolitites are aluminosilicates characterized by an open structure, which can accommodate a wide variety of ions. The particle size, crystallite size, and the degree of aggregation of the zeolitic material, as well as the porosity of individual particles, determine the access of ingesta fluids to the zeolitic surface during the passage across the GIT, and strongly affect its ion exchange, adsorption and catalytic properties (Papaioannou et al., 2005). The mechanism of action of zeolite is likely to be multifunctional. Different health and performance promoting properties were highlighted for zeolite in animal diet. These include ammonia binding effect, fecal elimination of p-cresol, retarding effect on digesta transit, enhanced pancreatic ezymes activity, and aflatoxin sequestering effect (Papaioannou et al., 2005). Moreover, recently it was reported the application of zeolite in reducing pathogens counts in broiler chicken (Prasai et al., 2016). Among zeolitites, the chabazitic zeolitite (CZ) has a high cation-exchange capacity and bulk density (Pabalan and Bertetti, 2001). Dietary inclusion of zeolitites has been effective in animals (e.g., pigs, calves) and humans suffering from gastrointestinal disturbances (RodriguezFuentes et al., 1997; Papaioannou et al., 2005). To date, no data exist about the evaluation of the effects of zeolitites on dog intestinal microbiota. The aim of the present study was to assess the effect of dietary CZ supplementation on the fecal microbiota with particular emphasis on bifidobacterial populations in training hunting dogs through culture-dependent methods and 16S rRNA/ITS (internal transcribed spacer) microbial profiling approach.

#### MATERIALS AND METHODS

#### Ethics Statement

This study was carried out in accordance with the recommendations of the ethical committee of the University of Parma. The protocol was approved by the "Comitato di Etica Università degli Studi di Parma", Italy. All animal procedures were performed according to national guidelines (Decreto legislativo 26/2014) on the protection of animals used for scientific purposes.

#### Animals and Experimental Procedure

Twenty adult English Setter dogs, reared in the same kennel, were selected to be homogeneous with reference to age (mean age ± SD: 3.50 ± 1.9 years), body weight (mean weight ± SD: 18.83 ± 2.96 kg) and gender (10 males, 10 not pregnant females). Based on age, weight, and sex animals were equally divided into two groups (10 dogs group−<sup>1</sup> ), individually penned with a rest area inside (2.70 m× 1.40 m) and a paddock outside (4.50 m× 1.40 m). Animals were free of any clinical symptoms indicating gastrointestinal disease and they did not receive medications that are expected to alter the gut microbiota such as antibiotics. Dogs were wormed one month before the start of study. The characteristics of the groups are reported in **Table 1**. During a period of 28 days, both groups received a diet, based on raw poultry meat (25% crude protein, 24% ether extract, 5% ash, 2% crude fiber, and 18.4 MJ kg−<sup>1</sup> ME, on dry matter). The individual ration, administered at about 25 g dry matter kg−<sup>1</sup> of body weight0.75, once a day, was supplemented (group Tr) or not (group NTr) with CZ powder at the dose of 5 g day−<sup>1</sup> . For each dog, zeolitite was weighed and added to the ration at each meal. Free access to water was provided. During the study, all dogs were daily subjected to an aerobic physical activity characterized by gallop for 20 min, according to the trainer's practices. Training was performed in two outdoor next areas, at a mean temperature and relative humidity of 24 ± 3 ◦C and of 67 ± 10%, respectively. Inside each group, five pairs of dogs were identified and each of them assigned alternatively to one or to the other of the training areas.

#### Chabazitic Zeolitite Source

The powder of CZ, was obtained after sterilization at 200◦C for 20 min (Chabasite 70 <sup>R</sup> Verdi S.p.A, Italy). The total zeolitic content was 70 ± 5%, of which 65 ± 3% due to chabazite (Na0.14K1.03Ca1.00Mg0.17) [Al3.46Si8.53O24] × 9.7H2O and 5 ± 3% to phillipsite (Na0.9Ca0.5K0.6) [Si5.2A12.8O16] × 6H2O. No traces of clinoptilolite were found. The composition of zeolitic powder was determined by Rietveld-RIR method (Gualtieri, 2000). The cation-exchange capacity and bulk density in relation to particles size were 2.2 ± 0.1 mEq g−<sup>1</sup> and 0.70−0.90 g (cm<sup>3</sup> ) −1 , respectively (Gualtieri, 2000; Cresswell and Hamilton, 2002). Water retention in relation to particles size was about 30−40% (w/w). The granulometry of the powder was less than 100 µm.


<sup>∗</sup>NTr, untreated group; Tr, treated group.

### Collection of Fecal Samples

fmicb-07-01491 September 19, 2016 Time: 11:54 # 3

Feces consistency was scored using a scale of 1 (hard) to 5 (watery) (Grieshop et al., 2002) at days 0 (Time point 0, T0), 16 (Time point 1, T1) from the beginning of the dietary treatment, and at the end of experimental period (day 29, Time point 2, T2). During the same days, individual fecal samples were collected directly from the rectum, using a sterile glove lubricated with water. The feces were placed in sterile polyethylene bags, immediately transported to the laboratory on ice packs and frozen at −20◦C until analysis.

#### 16S rRNA/ITS Microbial Profiling

Upon arrival at the laboratory, individual fecal samples were aliquoted and combined with other individual samples from the same treatment to form pooled samples. In fact, in animal health it has been shown recently that pooling stool samples allows a rapid assessment of infection intensity and drug efficacy (Mekonnen et al., 2013). Each individual dog sample was equally represented in the respective pooled sample. DNA was extracted from pooled fecal samples using the QIAamp DNA Stool Mini kit following the manufacturer's instructions (Qiagen Ltd., Strasse, Germany).

Partial 16S rRNA gene sequences were amplified from extracted DNA using primer pair Probio\_Uni and /Probio\_Rev, which target the V3 region of the 16S rRNA gene sequence, as previously reported (Milani et al., 2013). Partial ITS sequences were amplified from extracted DNA using the primer pair Probio-bif\_Uni/Probiobif\_Rev as described by Milani et al. (2014b). The PCR conditions used were 5 min at 95◦C and 35 cycles of 30 s at 94◦C, 30 s at 55◦C, and 90 s at 72◦C, followed by 10 min at 72◦C. Amplification was carried out using a Veriti Thermocycler (Applied Bio-systems).

16S rRNA gene and ITS sequencing were performed using a MiSeq (Illumina) according to the protocols previously published (Milani et al., 2013, 2014b).

# 16S rRNA Gene-Based Microbiota Analysis

The achieved individual sequence reads were filtered by the Illumina software to remove low quality and polyclonal sequences. All Illumina quality-approved, trimmed, and filtered data were exported as.fastq files. The.fastq files were processed using a custom script based on the QIIME software suite (Caporaso et al., 2010). Paired-end reads pairs were assembled to reconstruct the complete Probio\_Uni/Probio\_Rev amplicons. Quality control retained sequences with a length between 140 and 400 bp and mean sequence quality score >20 while sequences with homopolymers >7 bp and mismatched primers were omitted. In order to calculate downstream diversity measures (alpha diversity indices, Unifrac analysis), 16S rRNA Operational Taxonomic Units (OTUs) were defined at ≥97 % sequence homology using uclust (Edgar, 2010) and OTUs with less than 10 sequences were filtered. All reads were classified to the lowest possible taxonomic rank using QIIME (Caporaso et al., 2010) and a reference dataset from the SILVA database (Quast et al., 2013). Biodiversity of the samples (alpha-diversity) were calculated with Chao1 index.

#### ITS-Based Microbiota Analysis

For ITS-based microbiota analysis Fastq files obtained by sequencing of the ITS amplicons were analyzed using a custom script, named bif\_ITS\_analysis.sh script<sup>1</sup> . This script requires QIIME (Caporaso et al., 2010) to be installed (or works in a QIIME virtual machine) and accepts.bam or.fastq input files containing sequencing reads. Input data were processed as previously described (Milani et al., 2014b).

## Bacterial Counts

The homogenates fecal specimens were serially diluted with both half-strength Wilkins-Chalgren Anaerobe Broth (WCAB) and Buffered Peptone Water (ThermoScientific-Oxoid, UK). Dilutions in duplicate were plated on MacConkey agar (Merck, Germany) for Enterobacteriaceae, Perfringens agar Base (OPSP) (Oxoid, UK) for C. perfringens, vancomycin and bromocresol green (LAMVAB) agar (Hartemink and Rombouts, 1999) for lactobacilli, and Azide maltose agar (Biolife, Italy) for enterococci counts. MacConkey agar and Azide maltose agar plates were incubated aerobically at 37◦C for 24 and 48 h, respectively. Other media were incubated anaerobically at 37◦C for 48−72 h. The taxonomy of colonies isolated random on selective media were determined at genus and species level by API System (Bio-Merieux, Italy) to verify the reliability of the media utilized.

# In vitro CZ Adsorptive Capacity

The ability of CZ to bind to enteropathogens bacteria was evaluated in pooled feces using two reference strains, i.e., E. coli ATCC 35218 and C. perfringens ATCC 13124. Strains were grown in Mueller-Hinton Broth (Difco, MI, USA) at 37◦C for 24 h, then transferred to 10 ml of broth and grown for another 8 h to reach the final exponential phase.

Adsorptive capacity of CZ was evaluated, measuring spectrophotometrically the OD of the samples (An and Friedman, 1997). Twenty-five grams of pooled feces obtained by NTr groups and collected on days 0, 16, and 29 were placed, in triplicate, into flasks containing 225 ml of Buffered Sodium Chloride-Peptone Solution pH 7.0 (Oxoid, UK). CZ was added in different quantities (0, 0.25, 0.5, 1 g). Lastly, C. perfringens ATCC 13124 or E. coli ATCC 35218 strains were added to medium and incubated at 37◦C. At 0, 2, 4, 6, and 24 h, 150 µl of the suspension were transferred into a microtiter plate in four replicates and the absorbance was immediately evaluated (VICTOR3, 1420 multilabel counter, PerkinElmer, Italy) at 620 nm.

## Statistical Analysis

Data for fecal score and fecal bacteria counts were checked for normality and then analyzed by ANOVA using the GLM procedure in SAS (Version 9.4, SAS Institute Inc., USA). The mixed model included the fixed effects of group (two levels), of sampling time (three levels), the interaction between group and

<sup>1</sup>http://probiogenomics.unipr.it/sw/bif\_ITS\_analysis.zip

sampling time and the random effect of animal. Values of colony forming units (CFU) have been expressed as log<sup>10</sup> g <sup>−</sup><sup>1</sup> of feces.

Statistical significance was reached for P ≤ 0.05 as a P-value >0.05 and ≤0.10 was considered as a trend.

#### Data Deposition

fmicb-07-01491 September 19, 2016 Time: 11:54 # 4

Raw sequences of 16S rRNA gene profiling are accessible through SRA study accession number SRP075756. Raw sequences of ITS profiling are accessible through SRA study accession number SRP080281.

# RESULTS

## 16S rRNA Profiling of CZ Treated Dog

Pooled fecal samples from CZ treated (Tr) and no-treated (NTr) dogs were obtained in order to assess the microbiota composition based on 16S rRNA-sequencing analysis as described previously (Milani et al., 2013). The sequencing produced a total of 589784 reads with an average of 98297 reads per sample (Supplementary Table S1).

Assessment of rarefaction curves, based on the Chao1 biodiversity indexes calculated for 10 subsampling of sequenced read pools, indicated that both curves tend to reach a plateau. Therefore, in all cases the obtained sequencing data was deemed adequate to cover the vast majority of the biodiversity contained within the samples (**Figure 1A**). Moreover, the two curves did not show relevant differences, thus indicating that the analyzed samples have similar biodiversity.

## Gut Microbiota Composition of CZ Treated Dogs

During the study, diarrhea events were not observed in the CZ treated dogs. CZ did not affect the palatability of the feed, which was eaten completely within 30 min after dosing. Fecal scores were not affected by the factors in the statistical model (P > 0.05; **Table 2**). Fecal microbiota differences were observed in relation to group and sampling time (P < 0.05).

Inspection of predicted taxonomic profiles at phylum level for all NTr samples (T0, NTrT1, NTrT2) highlighted that Firmicutes (average 51.15% ± 11.46%) represented the dominant phylum of the cecal community in dogs, outnumbering the Proteobacteria (average 27.06% ± 15.75%), the Fusobacteria (average 8.54% ± 3.46%) and the Bacteroidetes (average 5.49% ± 2.62%) phyla (**Figures 1B,C**).

The comparison of the average relative abundance of NTr and Tr samples at time point T1 revealed a decrease of members of the Enterobacteriaceae family (−66.99 %), such as Escherichia (−67.16%), Klebsiella (−94.75%), and Hafnia (−74.87%), in Tr samples (**Figure 1D**) and an increase of Lactobacillus (205.16%) and Bifidobacterium (75.35%) in CZ treated animals (**Figure 1D**). At time point T2 in CZ treated animals (**Figure 1D**), the decrease in Enterobacteriaceae (−15.34%), includes a reduction of the genera Hafnia (-67.85 %), Klebsiella (−77.18%), and Enterobacter (−84.69%), along with an increase in relative abundance of Lactobacillus (861.64%) and Bifidobacterium (157.73%) (**Figure 1D**).

Notably, data achieved with culture-dependent approaches largely confirmed results obtained with 16S rRNA microbial profiling. In fact, Lactobacillus ssp. and Enterococcus ssp. counts were higher, while Enterobacteriaceae counts were lower in Tr than in NTr group (P < 0.05). Lactobacillus ssp. counts tended to be higher in Tr than in NTr group on day 16 (T1; 7.43 vs. 7.24; P < 0.10) and were higher on day 29 (T2; 8.18 vs. 7.25; P < 0.05). An increase of Enterococcus ssp. concentration (8.10 vs. 7.27) and a decrease of Enterobacteriaceae counts (6.24 vs. 7.14) were found in Tr compared to NTr group on day 29 (T2; P < 0.05). Besides, no change on the fecal C. perfringens counts was reported in relation both to the sampling time and to the treatment (P > 0.05).

#### Bifidobacterial Community Modulation by CZ

Focusing on the contribution of bifidobacteria to the overall dog microbiota, it is worth noticing that at day 0 (T0) and in NTr animals at days 16 and 29 (T1 and T2, respectively), this genus represents 2.32% ± 1.88% of the gut microbiota of hunting dogs. In treated animals (Tr) the presence of the Bifidobacterium genus showed an increase of about 157.73% compared with Tr animals at T2, after 29 days of CZ diet (**Figure 1D**).

In order to precisely catalog the effects on the bifidobacterial population of dogs after CZ treatment, we performed an ITS profiling of bifidobacterial communities in stool samples of Tr and NTr dogs.

Quality filtering of the sequenced ITS amplicons produced an average of 52468 high-quality and full-length reads per sample (Supplementary Table S2) that were taxonomically attributed reaching the minimal taxonomical rank of species.

The composition of bifidobacterial populations of dogs included in the analysis showed the presence of peculiar species, such as B. pseudolongum (average of 60.70% ± 24.61% for T0 and NTr samples) and B. animalis (average of 7.84% ± 7.50% in T0 and NTr dogs) (**Figure 2A**), which have been previously described to be typical of the animal GIT (Milani et al., 2014a) and especially of the dog GIT (Gavini et al., 2006). Notably, other bifidobacterial species previously described to be typical of the human gut such B. catenulatum and B. bifidum were detected at a lower extend (**Figure 2B**).

Furthermore, in untreated animal samples (NTr1 and NTr2), ITS analysis revealed the occurrence of B. longum, B. gallinarum, and B. pseudocatenulatum species, typical human bifidobacterial taxa (Milani et al., 2014a). One possible explanation of the presence of these species in the canine gut microbiota could be a bacterial transmission between animals and trainers as previously reported in literature (Song et al., 2013). However, further investigations will be needed. Notably, a large proportion of the OTUs defined as 'unclassified' in T0 dog samples (**Figures 2A,B**) clusters separately from any current known bifidobacterial taxon, thus putatively representing novel Bifidobacterium taxa. These putative new unclassified bifidobacterial species represents the second most present

FIGURE 1 | Exploration of the taxonomic profile of NTr and Tr groups. (A) Shows the rarefaction curves representing variation of the Chao1 and the Shannon diversity indexes at increasing sequencing depth of NTr and Tr fecal samples. (B) Displays bar plots and heat map of the identified bacterial phyla in the pooled CZ treated or untreated samples. (C) Represents bar plots and heat map of the identified bacterial genera in the pooled Tr or NTr samples. (D) exhibits the variation of taxa at time point T1 (upward) and T2 (below). We reported the bacterial genera with absolute change percentage >0.05% and showing increase >100% or decrease <−50% of relative change percentage in Tr data sets as compared to those obtained from NTr samples. In all panels the term unclassified member is abbreviated to U. m..



<sup>∗</sup>NTr, untreated group; Tr, treated group. ‡SEM, standard error of the difference of means. ‡G, group effect; St, sampling time effect; G × St, interaction. § On a scale of 1 (hard) to 5 (watery).

#### bifidobacterial taxa in the dog microbiota, in both Tr and NTr animals (**Figures 2A,B**).

As reported above, at time point T2 in CZ treated animals, there was an increase in relative abundance of the genus Bifidobacterium (**Figure 1D**). ITS profiling experiments revealed an increase of 803.14 and 54.76% of B. animalis species and B. pseudolongum species, respectively, after the addition of CZ. Moreover, a slight increase was detected also for the here

FIGURE 2 | Exploration of the bifidobacterial population of NTr and Tr groups. (A) Represents the bar plots of the identified bifidobacteria in the pooled CZ treated or untreated samples through the ITS analysis. (B) Shows heat map of the identified bifidobacteria in the pooled Tr-ITS or NTr-ITS samples. (C) Displays the variation of the bifidobacterial population at time point T1 (upward) and T2 (below). We reported the Bifidobacterium species with absolute change percentage >0.05 % and showing increase >100% or decrease <−50% of relative change percentage in Tr-ITS data sets as compared to those obtained from NTr-ITS samples. In all panels the term unclassified member is abbreviated to U. m..

identified putative new bifidobacterial taxa in TrT2 (12.45 %) compared to NTrT2 (**Figure 2C**).

## In Vitro Bacterial Adsorptive Test

fmicb-07-01491 September 19, 2016 Time: 11:54 # 7

CZ showed an adsorptive capacity toward E. coli (**Figure 3A**) and C. perfringens (**Figure 3B**) strains in a dose- and time-dependent trial. Differences among CZ levels were registered for both strains after 2, 4, 6, and 24 h of incubation (P < 0.05). In particular, higher adsorptive capability against E. coli strain, was observed when CZ was added to the medium at a dose of 0.5 and 1 g rather than of 0 and 0.25 g (P < 0.05). When CZ was added at a dose of 1 g, negative values of OD starting from 0 h of incubation was observed for E. coli. During the first six hours of incubation, the adsorptive effect of CZ on C. perfringens strain was higher for levels of 0.5 and 1 g, than of 0 and 0.25 g (P < 0.05; **Figure 3**).

#### DISCUSSION

In hunting dogs, emotional factors, such as those to which they are submitted during the training, can affect the GIT permeability, motility, secretion and mucin production. Thus, ultimately altering the habitat of resident gut bacteria and

promoting changes in the gut microbiota composition (Gagne et al., 2013). Therefore, various feeding strategies have been developed in order to keep a suitable function of the GIT tract. Zeolite and in particular CZ have shown efficacy in animals (such as pigs, calves) and humans suffering from gastrointestinal disturbances (RodriguezFuentes et al., 1997; Papaioannou et al., 2005).

In this study, 20 adult English Setter dogs were trained and fed with a diet supplemented with CZ to evaluate how the microbiota and in particular bifidobacterial population as well as specific gut pathogens, could be modulated.

The results obtained after 29 days of CZ diet, showed that CZ affects the fecal microbial concentration but not the fecal score, which remained in a desirable range (well-formed, soft stools) for healthy dogs (Gagne et al., 2013). Notably, we observed an increase in relative abundance of Lactobacillus ssp. as well as Bifidobacterium ssp. phylotypes, accompanied by a decrease in phylotypes belonging to Enterobacteriaceae family in CZ fecal samples. This could be supported by the adsorptive capacity exploited by CZ toward E. coli and C. perfringens. Furthermore, E. coli and Enterobacter are common causes of extra-intestinal opportunistic infections in in dogs (Ogeer-Gyles et al., 2006), while C. perfringens is strongly related to hemorrhagic gastroenteritis (Schlegel et al., 2012).

Moreover, the major presence of lactobacilli and bifidobacteria could be very interesting since these bacterial taxa are considered to exploit beneficial roles on the health of their hosts (Gibson et al., 2005). In this context, various members of Lactobacillus and Bifidobacterium species are the most exploited probiotic bacteria utilized for pet (Kelley et al., 2010; Strompfova et al., 2014) and some of them have been suggested to improve the health and brain function of dogs (Biagi et al., 2007; Bravo et al., 2011). Increased concentrations of these microorganisms have been associated with decreased fecal concentrations of potentially pathogenic bacteria and decreased levels of carcinogenic and putrefactive compounds in digesta (Grieshop et al., 2002).

This is the first study where the bifidobacterial community of healthy dog was explored through a Next Generation Sequencing approach involving bifidobacterial ITS profiling. The obtained results allowed the identification of a bifidobacterial profile in English setter hunting dogs and revealed the presence of typical animal bifidobacteria such as B. animalis and B. pseudolongum and many putative new taxa. CZ treatment led to an increase of the abundance of B. animalis and B. pseudolongum species, which are characterized by the presence of genes encoding for exopolysaccharides structures that could lead to a special cell protection (Ferrario et al., 2016; Hidalgo-Cantabrana et al., 2016). Increase of the bifidobacterial strains coupled with the adsorptive capacity of CZ could bring to a reduction of species belonging to the Enterobacteriaceae family, such as Klebsiella and Enterobacter, typical dog pathogens (Gibson et al., 2008). Combined CZ treatment with probiotic supplementation, such as bifidobacterial strains, might enhance the reduction of canine pathogens as well as strength the beneficial effects on the animal health.

# CONCLUSION

fmicb-07-01491 September 19, 2016 Time: 11:54 # 8

Dietary CZ supplementation can help to maintain a balanced intestinal microbial ecosystem and to prevent stress-related GIT upsets in healthy dogs, with a decrease of gut pathogens and a remarkable increase of bifidobacteria. This is particularly relevant in training hunting dogs where the mental and physical stress, to which they are subjected during training periods, can affect GI permeability and motility. Further studies are needed to confirm the beneficially effect by CZ also in diseased dogs.

#### AUTHOR CONTRIBUTIONS

AS, PS, VB, and MO designed and performed experiments. MO, LM, AS, PS, and CF wrote the manuscript. LM and CM performed bioinformatic analyses. AS, CF, PS, MO, and VB performed experiments. CM, LM, ER, and FDI commented the manuscript. PS, AS, and MO conceived the study, revised and approved the manuscript. All authors reviewed the manuscript.

#### REFERENCES


### FUNDING

LM is supported by Fondazione Cariparma, Parma, Italy.

### ACKNOWLEDGMENTS

The authors express sincere thanks to Verdi S.p.A, Italy for generously providing the Chabasite 70 <sup>R</sup> product, Della Bassana kennel, Pegognaga (MN), Italy for providing necessary facilities and dogs and Ms Mirella Masini for her operating contribution in the present study.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.01491



**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 FB 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 Sabbioni, Ferrario, Milani, Mancabelli, Riccardi, Di Ianni, Beretti, Superchi and Ossiprandi. 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 *Bifidobacterium bifidum* S17 Cultivated Under *In Vitro* Conditions

*Xiao Wei1, Simiao Wang1, Xiangna Zhao1, Xuesong Wang1, Huan Li1, Weishi Lin1, Jing Lu1, Daria Zhurina2, Boxing Li1, Christian U. Riedel2\*, Yansong Sun1\* and Jing Yuan1\**

*<sup>1</sup> Institute of Disease Control and Prevention, Academy of Military Medical Sciences, Beijing, China, <sup>2</sup> Institute of Microbiology and Biotechnology, University of Ulm, Ulm, Germany*

#### *Edited by:*

*Francesca Turroni, University College Cork, Ireland*

#### *Reviewed by:*

*Michael C. Nelson, University of Connecticut, USA Marco Ventura, University of Parma, Italy Reetta Satokari, University of Helsinki, Finland*

#### *\*Correspondence:*

*Jing Yuan yuanjing6216@163.com; Yansong Sun sunys@qq.com; Christian U. Riedel christian.riedel@uni-ulm.de*

#### *Specialty section:*

*This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology*

*Received: 04 November 2015 Accepted: 18 January 2016 Published: 12 February 2016*

#### *Citation:*

*Wei X, Wang S, Zhao X, Wang X, Li H, Lin W, Lu J, Zhurina D, Li B, Riedel CU, Sun Y and Yuan J (2016) Proteomic Profiling of Bifidobacterium bifidum S17 Cultivated Under In Vitro Conditions. Front. Microbiol. 7:97. doi: 10.3389/fmicb.2016.00097*

Bifidobacteria are frequently used in probiotic food and dairy products. *Bifidobacterium bifidum* S17 is a promising probiotic candidate strain that displays strong adhesion to intestinal epithelial cells and elicits potent anti-inflammatory capacity both *in vitro* and in murine models of colitis. The recently sequenced genome of *B. bifidum* S17 has a size of about 2.2 Mb and encodes 1,782 predicted protein-coding genes. In the present study, a comprehensive proteomic profiling was carried out to identify and characterize proteins expressed by *B. bifidum* S17. A total of 1148 proteins entries were identified by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS), representing 64.4% of the predicted proteome. 719 proteins could be assigned to functional categories according to cluster of orthologous groups of proteins (COGs). The COG distribution of the detected proteins highly correlates with that of the complete predicted proteome suggesting a good coverage and representation of the genomic content of *B. bifidum* S17 by the proteome. COGs that were highly present in the proteome of *B. bifidum* S17 were Translation, Amino Acid Transport and Metabolism, and Carbohydrate Transport and Metabolism. Complete sets of enzymes for both the bifidus shunt and the Embden-Meyerh of pathway were identified. Further bioinformatic analysis yielded 28 proteins with a predicted extracellular localization including 14 proteins with an LPxTG-motif for cell wall anchoring and two proteins (elongation factor Tu and enolase) with a potential moonlighting function in adhesion. Amongst the predicted extracellular proteins were five of six pilin proteins encoded in the *B. bifidum* S17 genome as well as several other proteins with a potential role in interaction with host structures. The presented results are the first compilation of a proteomic reference profile for a *B. bifidum* strain and will facilitate analysis of the molecular mechanisms of physiology, host-interactions and beneficial effects of a potential probiotic strain.

#### Keywords: proteomic profiling, metabolic pathways, pilin proteins, LC-MS/MS, *Bifidobacterium bifidum*

**Abbreviations:** DTT, DL-dithiothreitol; ECM, extracellular matrix; EDTA, ethylene diamine tetraacetie acid; HEPES, 4-(2 hydroxyethyl)-1-piperazineethanesulfonic acid; IECs, intestinal epithelial cells; KEGG, Kyoto Encyclopedia of Genes and Genomes; LC-MS/MS, liquid chromatography coupled to tandem mass spectrometry; LPXTG, Leu-Pro-X-Thr-Gly; MRS, Man-Rogosa-Sharpe; ORF(s), open reading frame(s); PMSF, phenylmethanesulfonyl fluoride; ThDP, thiamine diphosphate; TMDs, transmembrane domains; UHPLC, ultra-high-performance liquid chromatography.

# INTRODUCTION

Bifidobacteria represent an important group of the human intestinal microbiota (Bottacini et al., 2014). Due to their reported ability to reduce cholesterol levels, exclude intestinal pathogens, strengthen the intestinal barrier, alleviate symptoms of constipation, and/or modulate the immune response they are frequently used as active ingredients in food and dairybased products (Leahy et al., 2005; Gareau et al., 2010). In order to reveal the molecular mechanisms responsible for these beneficial effects, several bifidobacterial strains have recently been sequenced (Ventura et al., 2009). Analysis of bifidobacterial genomes has led to the identification of various structures involved in host colonization and probiotic properties. For example, *Bifidobacterium breve* UCC2003 was shown to encode genes for production of exopolysaccharides (Fanning et al., 2012). These exopolysaccharides support persistence in the murine gastrointestinal tract and are required for the protective effect of *B. breve* UCC2003 against infections with the murine pathogen *Citrobacter rodentium*. *B. breve* UCC2003 also possesses type IV tight adherence pili and these pili were shown to support long term colonization of mice (O'Connell Motherway et al., 2011). Similarly, other strains and species of bifidobacteria were shown to contain gene clusters for Tad and/or sortase-dependent pili for some of which interaction with host structures has been shown (Foroni et al., 2011; Turroni et al., 2013, 2014b).

*Bifidobacterium bifidum* strains belong to the infant-type bifidobacteria and show a remarkable adaptation to their ecological niche in the intestinal tract of human neonates. This includes a large number of adhesive structures and a specific ability to utilize host-derived glycans (Turroni et al., 2014a). *B. bifidum* S17 was isolated from feces of a breast-fed infant and displays unusually strong adhesion to IECs (Riedel et al., 2006a; Preising et al., 2010; Gleinser et al., 2012). Additionally, the strain elicits a promising anti-inflammatory capacity both *in vitro* (Riedel et al., 2006b; Preising et al., 2010) and *in vivo* in three different murine models of colitis (Preising et al., 2010; Philippe et al., 2011; Grimm et al., 2015). The genome of *B. bifidum* S17 was sequenced and annotated and contains a predicted 1,782 protein-coding ORFs (Zhurina et al., 2011) including one Tad and three sortase-dependent pili gene clusters as well as several other genes suspected or shown to play a role in adhesion to host structures or host colonization (Gleinser et al., 2012; Westermann et al., 2012).

In recent years, proteomic analysis has become an indispensible tool to analyze the biology of microorganisms, their response to changes in the environmental conditions and their interaction with the host (Otto et al., 2014). One of first the reports on a proteomic analysis of a *Bifidobacterium* sp. strain was a proteomic reference map obtained by 2D electrophoresis and MALDI TOF-TOF mass spectrometry (Yuan et al., 2006). Since then, this technique has been used to study adaptation of various bifidobacteria to bile and oxidative stress (Sánchez et al., 2007; Xiao et al., 2011) and different carbon sources (Liu et al., 2011, 2015), host-induced proteome changes of *B. longum* NCC2705 (Yuan et al., 2008) or its interaction with IECs (Wei et al., 2014). Turroni et al. (2010) performed a proteomic analysis of *B. bifidum* PRL2010 during growth on different sugars and following contact with cultured epithelial cells (Turroni et al., 2010).

However, proteomic approaches that include 2D electrophoresis have limitations in the detection of alkaline and low-abundance proteins (Otto et al., 2014). These limitations can be overcome by 1D gel-based LC-MS/MS (Wickramasekara et al., 2011; Otto et al., 2014). Moreover, LC-MS/MS is more efficient and accurate when analyzing differential global protein expression quantitatively and was used for proteomics of *Burkholderia vietnamiensis* (Wickramasekara et al., 2011). Another example is the analysis of secretion profiles of *B. pseudomallei* MSHR668 (Burtnick et al., 2014). LC-MS/MS has also been employed for comparative proteomics of two *Lactobacillus rhamnosus* strains (Savijoki et al., 2011). For bifidobacteria, LC-MS/MS was used for differential proteomics of two *B. longum* strains (Guillaume et al., 2009) and a proteomic profiling of *B. longum* subsp. *infantis* during growth on lactose, glucose, and galacto-, fructo-, and human milk oligosaccharides (Kim et al., 2013).

In the present study, the proteome of *B. bifidum* S17 grown *in vitro* was analyzed by 1D gel-based ultra-high performance LC-MS/MS representing the first comprehensive proteomic profile for the species *B. bifidum*.

#### MATERIALS AND METHODS

### Bacterial Strains and Growth Conditions

*Bifidobacterium bifidum* S17 was cultured in sealed jars anaerobically at 37◦C in *Lactobacillus* MRS medium (Difco) supplemented with 0.05% L-cysteine. Anaerobic conditions were achieved and maintained using AnaeroGen sachets (Thermo Scientific). Bacteria were harvested for proteomic analysis in midexponential growth phase at an optical density at 600 nm of 0.9 corresponding to approximately 1.5 <sup>×</sup> <sup>10</sup><sup>8</sup> colony forming units/ml.

#### Preparation of Whole Cell Protein Extracts

Bacteria were harvested by centrifugation, washed twice in phosphate-buffered saline (PBS), and pellets (about 0.30 g) were resuspended in 5 mL of lysis buffer (8 M urea, 30 mM HEPES, 1 mM phenylmethylsulfonyl fluoride, 2 mM ethylenediaminetetraacetic acid, 10 mM dithiothreitol) containing one dissolved tablet of complete protease inhibitor (Roche Diagnostics, Mannheim, Germany). Bacteria were then sonicated for 10 min on ice using a Sonifier 750 (Branson Ultrasonics Corp., Danbury, CT, USA) with the following conditions: 2 s of sonication with a 3 s interval set at 25% duty cycle. The cell lysate suspension was centrifuged for 30 min at 20,000 × *g* to collect the supernatant. The proteins were reduced with 10 mM dithiothreitol at 56 ◦C for 1 h, and alkylated with 55 mM iodoacetamide at room temperature for 1 h in the dark. The treated proteins were precipitated in acetone at −20◦C for 3 h. After centrifugation at 20,000 × *g* for 30 min, the protein pellet was resuspended and ultrasonicated in pre-chilled 50% tetraethylammonium bromide buffer with 0.1% sodium dodecyl sulfate (SDS). The proteins were regained after centrifugation at 20,000 × *g* and the protein concentrations were measured by Bradford assay.

# 1D SDS Polyacrylamid Gelelectrophersis (SDS-PAGE) and in-Gel Digestion

The proteome was analyzed by 1D gel-based LC-MS/MS as described previously (Albrethsen et al., 2010). In brief, the proteins were separated via SDS polyacrylamid gelelectrophersis (SDS-PAGE) using precast 4–12 % gradient gels containing Bis-Tris buffer (NuPAGE MES system, Invitrogen). Protein gels were run by using 1x running buffer [259 mM Tris base, 2 M glycine, 1% (w/v) SDS in ddH2O] at a constant voltage of 85 V for 20 min followed by 150 V for 40 min (until the dye runs off the gel). The gels were then stained with Coomassie R-250. A single lane of stained gel was cut into 10 pieces of approximately the same size and transferred into 1.5 mL Eppendorf tubes to perform ingel tryptic digestion. Gel bands were destained and tryptically digested as described previously (Yuan et al., 2006). The digested peptides were extracted with 50% acetonitrile and 2% formic acid solution. After extraction, the peptides were transferred into 500 µL Eppendorf tubes and concentrated using a Speed-Vac Concentrator (Savant) and a volume of 10 µL containing 15 µg of protein was loaded into individual high-performance liquid chromatography (HPLC) autosampler vials and analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS).

#### LC-MS/MS Analysis

Peptide samples were analyzed by ultra-high performance LC-MS/MS on a quadrupole-Orbitrap mass spectrometer (Q-Exactive; Thermo Fisher, Germany) equipped with 15 cm (length) by 75 µm (inside diameter) column packed with 5 µm C18 medium (Thermo Fisher) which was kept at 21◦C throughout the analysis. Mobile phase A was MilliQ water with 0.1% (v/v) formic acid. Mobile phase B was 99.9% (v/v) acetonitrile, 0.1% acetic acid. Gradient was run from 0% B to 30% B over 40 min and then to 80% B for 15 min. The LC was interfaced to a quadrupole-Orbitrap mass spectrometer (Q-Exactive; Thermo Fisher) via nano-electrospray ionization. An electrospray voltage of 1.8 kV was applied. The mass spectrometer was programmed to acquire, by data-dependent acquisition, tandem mass spectra from the top 20 ions in the full scan from 350 to 2,000 m/z. Dynamic exclusion was set to 15 s, singly charged ions were excluded, the isolation width was set to 2 m/z, the full MS resolution was set to 70,000, and the MS/MS resolution was set to 17,500. Normalized collision energy was set to 28, automatic gain control to 1e6, maximum fill MS to 20 ms, maximum fill MS/MS to 60 ms, and the under fill ratio to 0.1%.

### Protein Identification and Bioinformatic Analysis

Peptide identification were performed using Mascot v2.3.01 (Matrix Science Ltd.)1 licensed in-house2 (Albrethsen et al.,

1http://www.matrixscience.com

2010). Peptide mass finger printing searches were performed using Mascot v2.2.06 (Matrix Science Ltd.)1 licensed in-house2 . Monoisotopic peptide masses were used to search the databases, allowing a peptide mass accuracy of 30 ppm and fragment ion tolerance of 0.2 Dalton. Both methionine oxidation and cysteine carboxyamidomethylation were considered in the process. For protein identification by peptide mass fingerprinting, peptide masses were searched against the publically available database for *B. bifidum* S17 (NCBI Reference Sequence: NC\_014616.1). For unambiguous identification of proteins, more than five peptides must be matched and the sequence coverage must be greater than 15%. The complete proteomic dataset is deposited and publically accessible on the iProX database3 under project number IPX00067300.

Calculation of protein molecular weights and isoelectric points were carried out by the protparam software from the Expasy toolbox4 . Functional classification of identified proteins was performed by BLASTPGP (Altschul et al., 1997) searching against the databases of Cluster of Orthologous Groups (COGs5 ) (Tatusov et al., 2000). The cellular localizations of all identified proteins were predicted by PSORTb version 3.0 (Yu et al., 2010). Prediction of signal peptides was carried out with SignalP Version 4.1 (Petersen et al., 2011), TMHMM server 2.0 (Sonnhammer et al., 1998) was used to predict transmembrane helices, and LocateP database (Zhou et al., 2008) was used for prediction of cell wall and lipid anchor motifs.

# RESULTS

The 2.2-Mb genome of *B. bifidum* S17 contains 1,782 predicted protein-coding ORFs. To obtain an overview on the proteins expressed by *B. bifidum* S17 under standard laboratory conditions, whole protein extracts of bacteria grown in MRS medium were subjected to LC-MS/MS and protein identification. A total of 1,148 proteins were successfully matched unambiguously to one of the 1,782 predicted proteins of *B. bifidum* S17, i.e., a coverage of 64.4% of the complete predicted proteome. All detected proteins the peptide data matches the protein sequence as predicted in the genome annotation (data not shown). Moreover, the proteome contained a total of 235 (conserved) hypothetical proteins demonstrating that these proteins are actually expressed by *B. bifidum* S17 at least under *in vitro* conditions, which changes their status from "hypothetical" to proteins with unknown function.

**Supplementary Table S1** summarizes these proteins along with information on predicted pI, molecular mass, function, signal peptides, and subcellular location. 76.7% of the *B. bifidum* S17 proteins detected by LC-MS/MS are acidic with pI between 3 and 7, 12.5% have a pI between 7 and 9, whereas 10.8% of the proteins have a pI greater than 9 (**Figure 1A**). Classification of the 1,148 detected proteins into clusters of

<sup>2</sup>http://www.proteomics.cn

<sup>3</sup>www.iprox.org

<sup>4</sup>www.expasy.ch/tools/protparam.html

<sup>5</sup>ftp://ftp.ncbi.nih.gov/pub/COG/COG2014/static/lists/listCOGs.html

orthologous groups of proteins (COGs) assigned 719 proteins into 22 functional categories (**Table 1**). The remaining 429 unclassified proteins (37,4%) were denoted as COG-absent proteins. For none of the COGs an obvious clustering according to pI and Mw could be observed (**Figure 1A**). Comparison of the proteomes and genomes of *B. bifidum* S17 that the COG distribution of the *B. bifidum* S17 proteome is highly correlated to the distribution of all 1,782 predicted protein-coding ORFs of the genome (**Figure 1B**; coefficient of determination *R*<sup>2</sup> = 0.9769). Also, none of the COGs is skewed markedly off the regression in the proteome/genome comparison. Thus, none of the COGs seems to be over- or underrepresented in the proteome compared to the genome. Collectively, this indicates that the proteome captured by LC-MS/MS is comprehensive and covers all functional categories to a similar extend as encoded by the genomic information of *B. bifidum* S17.

The functional categories with highest representation amongst the detected proteins were Translation (functional category J; 137 proteins, 11.9% of all detected proteins), Amino acid Transport and Metabolism (functional category E; 107 proteins, 9.3%), and Carbohydrate Transport and Metabolism (functional category G; 77 proteins, 6.7%). Collectively, 26% of all detected proteins of the *B. bifidum* S17 proteome are assigned to functional categories J, A, K, and L. Thus, proteins possessing biological functions related to information storage, DNA replication, recombination and repair, RNA processing, transcription, and translation were highly prevalent in the proteome of *B. bifidum* S17. Of note, elongation factors Tu (EF-Tu; BBIF\_1251) was amongst these proteins.

A total of 9.3% of the proteome have a (predicted) role in amino acid transport, biosynthesis, urea cycle, and metabolism of amino groups including 14 aminotransferases (**Table 2**) and 6.7%

TABLE 1 | Relative abundance of functional categories according to clusters of orthologous groups of proteins (COGs) in the *Bifidobacterium bifidum* S17 proteome.


of all identified proteins are related to carbohydrate transport and metabolism. The major and characteristic metabolic pathway of bifidobacteria for fermentation of hexose sugars is the so-called fructose-6-phosphate or "bifid" shunt (Pokusaeva et al., 2011). We successfully detected a complete set of ten enzymes of the bifidus shunt (**Figure 2A**) including the key enzyme of the pathway xylulose-5-phosphate/fructose-6 phosphate phosphoketolase (BBIF\_0798). Moreover, 11 enzymes for a complete Embden-Meyerhof pathway were identified

(**Figure 2B**). Subcellular localization of all 1,148 proteins detected by LC-MS/MS was predicted by PSORTb Version 3.0. Extracellular and cell wall proteins play critical roles in establishing and maintaining interactions between a microbe and its environment. Thus, proteins which, according to the PSORTB prediction, are located in the cell wall or extracellular were further analyzed for transmembrane helices and signal peptides by SignalP v4.1 and TMHMM Server v2.0 and searched or LPxTG cell wall anchor motifs using the LocateP database to confirm their cellular localization (**Table 3**). Interestingly, these analysis suggest a different localization for several of these proteins as predicted by PSORTB. For example, three of the 28 proteins (BBIF\_0312, BBIF\_0337, BBIF\_1026) contained none of these domains/sequence motifs and are thus probably cytoplasmatic. Following, corrections the PSORTB correction of the proteins in **Table 3**, the proteome of *B. bifidum* S17 consists of 743 (64.7%) cytoplasmic, 231 (21.1%) membrane, 14 (1.2%) cell wall, 3 (0.3%) extracellular proteins and 146 (12.7%) proteins with unknown cellular localization


Wei et al. Proteomic Analysis of *B. bifidum* S17

#### FIGURE 2 | Continued

Enzymes of the bifidus shunt (A) and Embden-Meyerhof pathway (B) detected in the proteome of *B. bifidum* S17: BBIF\_1369: glucokinase; BBIF\_0700: alternative glucokinase; BBIF\_0798: xylulose-5-phosphate/ fructose-6-phosphate phosphoketolase; BBIF\_0363: glucose-6-phosphate isomerase; BBIF\_0973: transaldolase; BBIF\_0972: transketolase, BBIF\_1463: probable ribose-5-phosphate isomerase; BBIF\_0944: ribulosephosphate 3-epimerase; BBIF\_0788: acetate kinase; BBIF\_0979: L-lactate dehydrogenase; BBIF\_0175: fructose-bisphosphate aldolase; BBIF\_0981: triosephosphate isomerase; BBIF\_0612: glyceraldehyde 3-phosphate dehydrogenase; BBIF\_0982: phosphoglycerate kinase; BBIF\_1062: phosphoglycerate mutase; BBIF\_1743: phosphoglycerate mutase family protein; BBIF\_0742: enolase; BBIF\_0777: pyruvate kinase.

(**Figure 3**). A number of proteins of the *B. bifidum* S17 proteome with extracellular or cell wall localization have been associated with bacterial adhesion to host structures, colonization and/or immunomodulation. For example, five of the six pilin proteins for sortase-dependent pili including major pilins BBIF\_0301 and BBIF\_1761 and the minor pilins BBIF\_0302; BBIF\_1648, and BBIF\_1761 were detected in the proteome. Further proteins identified in the proteome of *B. bifidum* S17 that may have a role in host colonization are the potential adhesin BBIF\_0636 (BopA), the s subtilisin family peptidase BBIF\_1681, and BBIF\_1317 and BBIF\_1734, two glycoside hydrolases of the fucosidase and sialidase family.

#### TABLE 3 | Proteins of the *B. bifidum* S17 proteome with predicted cell wall or extracellular localization.


<sup>a</sup>*SP, signal peptide predicted using SignalP v4.1.*

<sup>b</sup>*TMH, transmembrane helices, predicted usingTMHMM Server v2.0.*

<sup>c</sup>*Lipid anchor and cell wall (CW) anchor motifs obtained from the LocateP database (http://www.cmbi.ru.nl/locatep-db/cgi-bin/locatepdb.py).* <sup>d</sup>*final prediction for cellular localization cell wall (Cw), extracellular (E), membrane (M), orcytoplasma (Cp).*

# DISCUSSION

extracellular proteins according to Table 3.

*Bifidobacterium bifidum* strains belong to the infant-type bifidobacteria and some of these strains possess interesting properties related to host-health (Turroni et al., 2014a). In an attempt to compile the first proteomic reference profile for the species, *B. bifidum*, we determined the complete proteome of *B. bifidum* S17 grown in MRS, i.e., the standard medium for routine culture of bifidobacteria, by LC-MS/MS. In total 1,148 proteins were detected, i.e., a coverage 64.4% of all predicted proteins of *B. bifidum* S17. This is comparable to the coverage obtained by proteome analysis of two *L. rhamnosus* strains (Savijoki et al., 2011). However, it may still be an underestimation of the complete proteome of *B. bifidum* S17 expressed under the conditions tested. The experimental approach utilized the bacterial pellet of cultures grown in MRS. Thus, most proteins that are secreted into the cell culture supernatant are probably not captured. In line with the proteomes of other bacteria (Yuan et al., 2006; Liu et al., 2012), proteins with a function in DNA replication, recombination and repair, RNA processing, transcription, and translation to were highly prevalent in the proteome of *B. bifidum* S17. One of these proteins is elongation factors Tu (EF-Tu; BBIF\_1251), which was previously shown to posses moonlighting function as an adhesin that mediates binding of *B. longum* NCC2705 and *Lactobacillus* sp. mucus, IECs, and/or ECM components of the host (Granato et al., 2004; Ramiah et al., 2008; Dhanani and Bagchi, 2013; Wei et al., 2014).

Detection of large numbers of enzymes of the bifidus shunt, glycolysis, and amino acid metabolism in the proteome of *B. bifidum* S17 confirms previous findings on the proteome of *B. longum* NCC2705 (Yuan et al., 2006). *B. bifidum* S17 was grown in MRS and *B. longum* NCC2705 in modified Garches medium for proteome analysis. Both media contain high levels of complex components such as yeast extract, peptone and/or beef extract and glucose as additional carbon source (De Man et al., 1960; Krzewinski et al., 1997). This may explain the high numbers of enzymes detected in the two proteomes involved in the uptake, degradation and fermentation of these substrates. Similar to EF-Tu, one of the proteins of the Embden-Meyerhof pathway is a cytoplasmic enzyme that has a moonlighting function. Enolase (BBIF\_0742) was shown to mediate adhesion of *B. bifidum* and other bifidobacteria to IECs and components of the ECM (Candela et al., 2009; Wei et al., 2014). An interesting finding is the detection of 14 aminotransferases in the proteome (**Table 2**). Transamination reactions of amino acid converting pathways have recently attracted attention because they are the first step for the synthesis of important aroma compounds (Christensen et al., 1999) which may affect flavor of probiotic preparations.

A number of proteins of the *B. bifidum* S17 proteome contain domains and signal sequences for extracellular or cell wall location. Some of these proteins have been associated with bacterial adhesion to host structures, colonization and/or immunomodulation. Five of the six pilin proteins for sortasedependent pili were identified in the proteome including major pilins BBIF\_0301 and BBIF\_1761 and the minor pilins BBIF\_0302; BBIF\_1648, and BBIF\_1761. This suggests that at least two of the three sortase-dependent pili encoded on the genome of *B. bifidum* S17 are expressed under laboratory conditions. This is in line with the observation that sortasedependent pili gene clusters are expressed by *B. bifidum* PRL2010 both *in vitro* and in the mouse GIT (Turroni et al., 2013) and pilus-like structures are detectable by atomic force microscopy in the same strain *in vitro* (Foroni et al., 2011). Expression of pili may have an important function in interaction of *B. bifidum* strains with the host. Heterologous expression of one of these gene clusters in *L. lactis* led to increased adhesion to ECM proteins and cultured IECs (Turroni et al., 2013). Moreover, the *L. lactis* strain expressing these pili elicits altered cytokine profiles in the murine gastrointestinal mucosa compared to the control. Another protein with potential role in adhesion to host structures is BBIF\_0636. This protein is an extracellular solute-binding protein of a peptide/nickel transport system, which was also termed bifidobacterial outer protein A (BopA). Previous data obtained by us and others suggests that BopA mediates binding to cultured human IECs (Guglielmetti et al., 2008; Gleinser et al., 2012). However, these findings have been challenged recently (Kainulainen et al., 2013).

Three further proteins of the *B. bifidum* S17 proteome with a potential role in interaction with the host are BBIF\_1317, BBIF\_1681, and BBIF\_1734. BBIF\_1681 is a peptidase of the subtilisin family. The subtilisin family peptidase lactocepin of *L. casei* was shown to degrade pro-inflammatory cytokines contributing to the immunomodulatory effect of this probiotic bacterium (von Schillde et al., 2012). BBIF\_1317 and BBIF\_1734 are a glycoside hydrolases of the fucosidase and sialidase family, respectively. Sialidases are known as important virulence factors of bacterial pathogens that mediate attachment and degradation of host-derived mucus and were also shown to be involved in host colonization by commensal bacteria (Lewis and Lewis, 2012). Genes for mucin degradation pathways including sialidases are conserved amongst *B. bifidum* strains and most *B. bifidum* strains are able to grow on mucin as sole carbon source (Turroni et al., 2010). Similarly, fucosidases are involved in degradation of host-derived glucans such as human milk oligosaccharides and mucus (Turroni et al., 2010). Utilization of host-derived glucans is considered as nutritional adaptation to the intestinal tract of the (human) host (Turroni et al., 2010; Bottacini et al., 2014; Grimm et al., 2014). Thus, detection of proteins for degradation of host derived glycans under *in vitro* is somewhat surprising since these substrates are no present in standard growth medium, i.e., MRS. However, LC-MS/MS is a highly sensitive method that allows detection of even small amounts of a given protein (Otto et al., 2014). Moreover, pilin proteins that are only required *in vivo* were also present in the proteome of *B. bifidum* S17 and expression of pili genes and proteins *in vitro* has been observed previously (Foroni et al., 2011; Westermann et al., 2012). This suggests that regulation of host-interacting proteins may not strictly regulated in bifidobacteria.

In summary a total of 1,148 proteins of the predicted proteome were detected including important metabolic pathways, proteins

#### REFERENCES


known or suspected to be involved in adhesion and colonization as well as a large number of (previously) hypothetical proteins. This represents the first complete proteome analysis of the species *B. bifidum* and confirms previous findings on the proteome level.

#### AUTHOR CONTRIBUTIONS

JY, CR, and XW designed research; XW, DZ, HL, and WL performed research; XW, SW, BL, and XZ contributed new reagents or analytic tools; XW, JL, and CR analyzed data; XW, CR, and YS wrote the paper.

#### FUNDING

This work was supported by a grant from the National Natural Science Foundation of China (31370093) to JY and (81400592) to XW, Mega-projects of Science and Technology Research of China (Grant 2011ZX10004-001), the German Academic Exchange Service/Federal Ministry of Education and Research to CR (Grant D/09/04778), and a grant from the National High Technology Research and Development Program of China (863 Program; grant no. SS2014AA022210).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.00097

#### TABLE S1 | Characteristics of the 1,148 identified proteins.


human intestinal cells and mucins. *Infect. Immun.* 72, 2160–2169. doi: 10.1128/IAI.72.4.2160-2169.2004


*Clostridium sporogenes* and *Enterococcus faecalis*. *Res. Microbiol.* 159, 470–475. doi: 10.1016/j.resmic.2008.06.002


**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 Wei, Wang, Zhao, Wang, Li, Lin, Lu, Zhurina, Li, Riedel, Sun and Yuan. 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.*

# A Critical Evaluation of Bifidobacterial Adhesion to the Host Tissue

#### Christina Westermann<sup>1</sup>† , Marita Gleinser<sup>1</sup>† , Sinéad C. Corr<sup>2</sup> and Christian U. Riedel<sup>1</sup> \*

1 Institute of Microbiology and Biotechnology, University of Ulm, Ulm, Germany, <sup>2</sup> Department of Microbiology, Moyne Institute of Preventative Medicine, School of Genetics and Microbiology, Trinity College Dublin, Dublin, Ireland

Bifidobacteria are common inhabitants of the human gastrointestinal tract that, despite a long history of research, have not shown any pathogenic potential whatsoever. By contrast, some bifidobacteria are associated with a number of health-related benefits for the host. The reported beneficial effects of bifidobacteria include competitive exclusion of pathogens, alleviation of symptoms of irritable bowel syndrome and inflammatory bowel disease, and modulation of intestinal and systemic immune responses. Based on these effects, bifidobacteria are widely used as probiotics by pharmaceutical and dairy industries. In order to exert a beneficial effect bifidobacteria have to, at least transiently, colonize the host in a sufficient population size. Besides other criteria such as resistance to manufacturing processes and intestinal transit, potential probiotic bacteria are tested for adhesion to the host structures including intestinal epithelial cells, mucus, and extracellular matrix components. In the present review article, we summarize the current knowledge on bifidobacterial structures that mediate adhesion to host tissue and compare these to similar structures of pathogenic bacteria. This reveals that most of the adhesive structures and mechanisms involved in adhesion of bifidobacteria to host tissue are similar or even identical to those employed by pathogens to cause disease. It is thus reasonable to assume that these structures and mechanisms are equally important for commensal or probiotic bacteria and play a similar role in the beneficial effects exerted by bifidobacteria.

Keywords: Bifidobacterium, probiotic, adhesion, host interaction

# INTRODUCTION

The mammalian GIT is home to an extremely complex and diverse microbial ecosystem consisting primarily of prokaryotes. This microbial community is collectively referred to as the gut microbiota and exerts a number of profound effects on host health (Marchesi et al., 2016). During the first period of life when newborns are exclusively breast-fed, members of the genus Bifidobacterium are one of the predominant bacterial groups of the microbiota in the lower GIT (Yatsunenko et al., 2012; Bäckhed et al., 2015; Walker et al., 2015). The major source for bifidobacteria is the intestinal microbiota of the mother to which the newborn is exposed during (vaginal) delivery (Grönlund et al., 2011; Matamoros et al., 2013). More recently, breast milk has also been shown to contain viable bifidobacteria (Fernández et al., 2013; Jost et al., 2015). Although their relative proportion

**Abbreviations:** ECM, extracellular matrix; GIT, gastrointestinal tract; IECs, intestinal epithelial cells.

#### Edited by:

Francesca Turroni, University College Cork, Ireland

#### Reviewed by:

Marco Ventura, University of Parma, Italy Simone Rampelli, University of Bologna, Italy

#### \*Correspondence:

Christian U. Riedel christian.riedel@uni-ulm.de †These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 27 May 2016 Accepted: 22 July 2016 Published: 05 August 2016

#### Citation:

Westermann C, Gleinser M, Corr SC and Riedel CU (2016) A Critical Evaluation of Bifidobacterial Adhesion to the Host Tissue. Front. Microbiol. 7:1220. doi: 10.3389/fmicb.2016.01220

decreases over time, bifidobacteria are still a subdominant group amongst intestinal bacteria of adult humans (Arumugam et al., 2011).

With the exception of Bifidobacterium dentium, which has been associated with dental caries (Ventura et al., 2009), bifidobacteria have to date not shown any pathogenic potential. By contrast, a number of health promoting effects have been attributed to the presence of bifidobacteria in the GIT including improvement of symptoms of irritable bowel syndrome, inflammatory bowel disease and infectious diarrhea, modulation of intestinal and systemic immune responses, and resistance against colonization by pathogens (Gareau et al., 2010; Buffie and Pamer, 2013). Of note, a very recent study links bifidobacteria in the gut microbiota to enhanced anti-tumor immune responses and support of checkpoint-inhibition cancer therapy using a monoclonal antibody (Sivan et al., 2015). Based on these findings bifidobacteria are widely used as probiotics, i.e., live microorganisms which when administered in adequate amounts confer a health benefit to the host (Holmes et al., 2012; Foligné et al., 2013).

Besides the health promoting effects, several criteria are applied during selection of a suitable probiotic candidate strain including stability during manufacturing processes, viability during gastrointestinal transit and functionality at the desired target site (Foligné et al., 2013). One of the classical selection criteria for potential probiotic bacteria is adhesion to mucus and/or IECs (Klaenhammer and Kullen, 1999; Tuomola et al., 2001; Papadimitriou et al., 2015).

It may be argued that adhesion is not important for probiotic functionality since probiotic bacteria do not have access to host tissue due to the thick mucus layer covering the (healthy) gut epithelium. However, a number of bifidobacteria were shown to adhere to mucus (He et al., 2001; Izquierdo et al., 2008) and utilize host-derived mucins as a substrate for growth (Tailford et al., 2015). Also, bifidobacteria are discussed as potential treatment options for conditions with an impaired mucus layer (Whelan and Quigley, 2013; Johansson, 2014) facilitating direct access of (bifido)bacteria to the epithelium. Moreover, various bacterial pathogens must overcome the mucosal barriers and gain access to the epithelial layer to cause disease. For example, pathogenic Escherichia coli strains and related organisms use pili, fimbriae, and/or intimin with its translocated intimin receptor for adhesion to epithelial cells (Niemann et al., 2004). Another example is the interaction of InlA of Listeria monocytogenes with E-cadherin on host epithelial cells which is crucially required for infection (Stavru et al., 2011). Only once adhesion of these pathogens to the epithelium has been achieved despite the presence of an intact mucus layer, progression to later stages of infection and disease are possible (Bhavsar et al., 2007).

On the other hand, a number of probiotic traits may be directly linked to adhesion to host structures. One of the proposed health benefits of bifidobacteria is resistance against colonization or infection by pathogens. This may involve a variety of adhesionindependent mechanisms such as competition for nutrients or production of antimicrobial compounds (Buffie and Pamer, 2013; Lawley and Walker, 2013). Nevertheless, adhesion to IECs, mucus and ECM components by commensal and probiotic bacteria may also directly block access of pathogens to these structures (Bernet et al., 1993; Collado et al., 2005; Candela et al., 2008; Serafini et al., 2013) either by competition for attachment sites or steric hinderance. Also, there are numerous reports of immunomodulatory effects of bifidobacteria in vitro and in animal models (Bermudez-Brito et al., 2012). All these effects crucially depend on interaction with (and thus adhesion to) epithelial cells, dendritic cells, monocytes, macrophages and or other immune cells. Finally, even if not directly implicated mechanistically, adhesion might contribute to beneficial effects by allowing initial colonization or prolonging persistence of (probiotic) bifidobacteria in the GIT.

# FACTORS FOR ADHESION OF BIFIDOBACTERIA TO HOST STRUCTURES

A number of factors and structures involved in adhesion to IECs, ECM components, and mucus have been identified in bifidobacteria (**Table 1**). These studies have been performed almost exclusively in in vitro model systems.

Adhesion to mucus is mostly analyzed using microtiter plate assays with immobilized mucus with quantification of adherent bacteria after metabolic labeling using radioisotopes or fluorescent dyes (He et al., 2001; Izquierdo et al., 2008; González-Rodríguez et al., 2012; Kainulainen et al., 2013). Similar assays are performed to analyze adhesion to immobilized ECM proteins (Kainulainen et al., 2013) or detection of ECM proteins bound to bacterial cells or protein extracts by specific antibodies (Candela et al., 2007, 2009, 2010).

The methods and cell lines used to determine adhesion to IECs differ largely between studies and groups. The most widely used cell lines are Caco-2, HT-29, and T84 (Guglielmetti et al., 2009; Preising et al., 2010; Gleinser et al., 2012; González-Rodríguez et al., 2012; Kainulainen et al., 2013; Grimm et al., 2014). In studies that employ more than one cell line, absolute adhesion of different strains may vary between cell lines but relative differences between strains are usually conserved (Riedel et al., 2006; Preising et al., 2010; Gleinser et al., 2012). One observation is that, although there is considerable strain-to-strain variability, strains of the species B. bifidum generally tend to adhere better to IECs than strains of other species (Guglielmetti et al., 2008; Gleinser et al., 2012). Detection of adherent bacteria is performed by metabolic labeling using radionucleotides (Riedel et al., 2006; Kainulainen et al., 2013), enumeration of colony forming units of adherent bacteria (Gleinser et al., 2012; González-Rodríguez et al., 2012), microscopic imaging and calculation of adhesion indices, i.e., the ratio of adherent bacteria and cells (Guglielmetti et al., 2008, 2009, 2010), or expression of fluorescent proteins (Grimm et al., 2014). However, the method of quantification does not seem to impact on adhesion itself as comparable results are obtained using radioactive or fluorescent labeling and plate counting (Riedel et al., 2006; Gleinser et al., 2012; Grimm et al., 2014).

In the following sections, the current knowledge on bifidobacterial adhesion to host structures will be summarized

#### TABLE 1 | Adhesive structures identified in bifidobacteria.


and the involved factors will be compared to adhesion factors of pathogens.

#### Pili

A wide range of Gram-positive and -negative bacteria possess proteinaceous surface appendages termed fimbriae or pili (Proft and Baker, 2009). In general, pili are adhesive structures that are involved in biofilm formation, conjugation, motility, and adhesion to biotic and abiotic surfaces (Maier and Wong, 2015). These hair-like structures extend to some distance (up to 3 µm) from the bacterial cell surface (Proft and Baker, 2009). It is hypothesized that they are able to bridge the repulsive forces between microbial cells and biological substrates, which under physiological conditions are both negatively charged (Proft and Baker, 2009). Pili are well-known for their role as virulence factors of Gram-positive and -negative pathogens and are important for initial attachment to host tissues (Telford et al., 2006; Proft and Baker, 2009).

There is increasing evidence that bifidobacteria also encode and express pilus-like structures on their cell surface. The first report of pili in bifidobacteria was the presence of genes encoding type IVb tight adherence (Tad) pili in B. breve UCC2003 (O'Connell Motherway et al., 2011). Since then, genes for Tad, Type VIa, and/or sortase-dependent pili were found in basically all sequenced genomes of bifidobacteria (**Table 1**). Interestingly, in most cases bifidobacteria posses more than one pilus-coding locus and B. dentium harbours as much as seven gene clusters for sortase-dependent pili (Foroni et al., 2011).

Transcriptional analysis revealed that at least some of the genes are expressed under in vitro conditions and are regulated in response to substrate, presence of other bacteria, growth phase, or stress conditions related to the GIT (Foroni et al., 2011; Westermann et al., 2012; Duranti et al., 2014; Turroni et al., 2014). Moreover, pilin proteins are present in the in vitro proteome of B. bifidum S17 (Wei et al., 2016) and pilus-like structures were observed on several bifidobacteria by electron and atomic force microscopy (Foroni et al., 2011; O'Connell Motherway et al., 2011; Duranti et al., 2014). Collectively, this suggests that bifidobacteria possess functional pili. There is also evidence that bifidobacterial pili have a role in colonization of the host and attachment to epithelial cells. In B. breve UCC2003, expression of the tad locus was up-regulated in the GIT of mice

and is required for efficient colonization in the presence of a competing microbiota (O'Connell Motherway et al., 2011). In a B. adolescentis strain, expression of pilus gene clusterss and presence of pili were enhanced when bacteria were isolated from the murine GIT or grown on starch, cellobiose or maltodextrin, i.e., substrates abundantly present in the GIT (Duranti et al., 2014). Similarly, expression of two of the three sortase-dependent pili clusters of B. bifidum PRL2010 is enhanced in the murine GIT and in the presence of human IECs in vitro (Turroni et al., 2013). Although a direct role for colonization and adhesion by inactivation of the corresponding genes is missing (probably due to the lack of appropriate genetic tools for the species B. bifidum), heterologous expression of the two gene clusters in Lactococcus lactis led to presence of pilus-like structures. Moreover, the recombinant L. lactis strains displayed increased adhesion to cultured IECs (pil2 cluster) and ECM proteins laminin, fibronectin, fibrinogen, and plasminogen (pil2 and pil3 clusters; Turroni et al., 2013). Adhesion to fibronectin seems to be mediated by sugar-binding domains of the pili since enzymatic deglycosylation of fibronectin markedly reduced adhesion of the recombinant L. lactis strains expressing the pil2 and pil3 gene clusters of B. bifidum PRL2010 (Turroni et al., 2013).

## Moonlighting Proteins

A rather obscure group of proteins involved in adhesion of bacteria to host tissues are so-called moonlighting proteins (Huberts and van der Klei, 2010). These proteins are multifunctional and usually have an enzymatic role in bacterial metabolism or other cellular processes but at the same time are involved in totally unrelated biological functions (Huberts and van der Klei, 2010). In more than 90 pathogenic bacteria, proteins with a moonlighting function in virulence have been identified (Henderson, 2014). Interestingly, a large number of moonlighting proteins are cytoplasmic enzymes of the central metabolism that lack secretion signals raising the question if these proteins are actively exported to mediate virulence related functions. The best characterized examples are adhesins of pathogenic bacteria that are involved in primary attachment to host tissue and are important for later stages of infection (Henderson and Martin, 2011).

Enzymes of glycolysis with a moonlighting function in adhesion of pathogens include aldolase (or transaldolase), enolase, and glyceraldehyde-3-phosphate dehydrogenase (Henderson and Martin, 2011). These proteins were detected in proteomes of different bifidobacteria (Yuan et al., 2006, 2008; Ruiz et al., 2009; Gilad et al., 2011; Liu et al., 2011; Wei et al., 2016; Zhu et al., 2016). Transaldolase, a cytoplasmatic key enzyme of the bifidus shunt, was found to be present on the surface of several B. bifidum strains (González-Rodríguez et al., 2012). Using an in vitro binding assay the transaldolase could be identified as a mucin-binding protein and the specificity of this interaction was confirmed by increased mucus binding of recombinant L. lactis strains expressing transaldolase (González-Rodríguez et al., 2012). Enolase of different B. longum, B. bifidum, B. animalis subsp. lactis and B. breve strains was shown to interact with plasminogen (Candela et al., 2007, 2009; Wei et al., 2014). Moreover, the plasminogen binding site in the B. lactis enolase was shown to be homologous to that of Streptococcus pneumoniae and specific amino acid residues crucial for plasminogen binding have been identified (Candela et al., 2009). Another moonlighting protein that serves as an adhesin for bifidobacteria is DnaK, which has a primary function as a chaperone (Henderson and Martin, 2011). For B. animalis subsp. lactis BI07, DnaK was shown to bind plasminogen (Candela et al., 2007, 2010). Further potential moonlighting proteins of B. animalis subsp. lactis BI07 with plasminogen-binding activity are glutamine synthetase, bilesalt hydrolase, and phosphoglycerate mutase (Candela et al., 2007).

For B. longum NCC2705, transaldolase was detected at higher levels incubated in vivo in a rabbit intestinal loop compared to in vitro growth (Yuan et al., 2008) and enolase and transaldolase were more abundant in the proteome following co-cultivation with IECs (Wei et al., 2014). Also, expression of DnaK and enolase is upregulated in several bifidobacteria in response to bile (Savijoki et al., 2005; Candela et al., 2010). This indicates that bifidobacteria might be able to sense the conditions of the intestinal environment and presence of IECs (or receptors on IECs) and respond by enhancing expression of adhesive molecules.

# Other Adhesion Factors

A rather general, and non-specific property of bacteria that has been associated occasionally with adhesion of pathogens to host tissue is surface hydrophobicity (Hirt et al., 2000; Kouidhi et al., 2010). Several studies have tested different strains and species of bifidobacteria for hydrophobicity, autoaggregation and adhesion to IECs (Pérez et al., 1998; Del Re et al., 2000; Pan et al., 2006). Overall, the results suggest that (i) strains with higher surface hydrophobicity show higher autoaggregation and adhesion to IECs and (ii) B. bifidum strains tend to be more hydrophobic than strains of other Bifidobacterium sp. This is in line with other studies showing that B. bifidum strains adhere better to IECs than strains of other species (Preising et al., 2010; Gleinser et al., 2012).

Other non-proteinaceous component of the bacterial envelope that have been associated with adhesion to host tissue of Gram positive pathogens are glycoconjugates including exopolysaccharides, lipoteichoic, and wall teichoic acids (Weidenmaier and Peschel, 2008; Tytgat and Lebeer, 2014; Tan et al., 2015). Despite the presence of genes (potentially) involved in biosynthesis of exopolysaccharides and teichoic acids in most of the sequenced genomes of bifidobacteria (Hidalgo-Cantabrana et al., 2014; Colagiorgi et al., 2015), a contribution to adhesion have not been demonstrated conclusively so far. However, one study links exopolysaccharide production of bifidobacteria with adhesion to mucus by showing that purified exopolysaccharides of two bifidobacteria reduced adhesion of intact bacterial cells of these strains (Ruas-Madiedo et al., 2006).

Besides the abovementioned pili and moonlighting proteins no specific adhesins such as intimin, internalins, lectins, fibronectin-binding proteins as described for a number of pathogens (Niemann et al., 2004; Kline et al., 2009) have been characterized for bifidobacteria. A bioinformatic screen of the genome of B. bifidum S17 yielded a number of proteins with domains such as fibronectin type III domain, concanavalin A-like

lectin, and collagen triple helix repeat domains, suggesting that bifidobacteria might have similar adhesins (Westermann et al., 2012). A definite role of the corresponding proteins in adhesion to host structures has yet to be demonstrated.

One specific protein that has been suspected to play a role in adhesion of bifidobacteria to IECs is BopA, lipoprotein of the cell envelope specifically found in B. bifidum strains (Guglielmetti et al., 2008; Gleinser et al., 2012). However, BopA contains the characteristic domains of a solute-binding protein and is part of an operon that encodes a putative oligopeptide ABC-transporter (Gleinser et al., 2012). Moreover, a recent study has challenged the idea that BopA serves a function in adhesion by showing that blocking BopA using a specific antibody does not affect adhesion of B. bifidum MIMBb75 to IECs (Kainulainen et al., 2013). Thus, BopA might be another example for a moonlighting protein but whether it has a role in adhesion of B. bifidum strains to intestinal tissue in humans needs to be elucidated in further studies.

#### CONCLUSION

A large number of Bifidobacterium sp. strains were shown to adhere to IECs, mucus, and/or ECM proteins. For some bifidobacteria, adhesive structures have been characterized and include pili and different moonlighting proteins. Lactobacilli, another group of potential probiotic, Gram-positive microorganisms use exactly the same structures to adhere to the same target sites on host tissues (Vélez et al., 2007; van Tassell and Miller, 2011). Pathogenic microorganisms employ similar or even identical structures to adhere to host structures. The genus Bacteroides contains highly abundant commensal species as well as opportunistic pathogens that even may cause cancer (Wexler, 2007; Sears et al., 2014). Both commensal and pathogenic strains were shown to adhere to IECs, ECM, or mucus (Brook and Myhal, 1991; Ferreira et al., 2002; Macfarlane et al., 2005; de

#### REFERENCES


O Ferreira et al., 2006; Huang et al., 2011; Ferreira Ede et al., 2013) and pili, specific ECM-binding proteins, EPS etc. (Brook and Myhal, 1991; de O Ferreira et al., 2006; Pumbwe et al., 2006; Ferreira Ede et al., 2013) are involved in the process. Collectively, this illustrates that both pathogenic and commensal, in some cases even beneficial, bacteria employ the same strategies to attach to host structures. There is no doubt that adhesion of pathogens to host tissue is required or helps to promote infection. Bifidobacteria are generally regarded as safe microorganisms, which despite intensive studies of the past decades have not shown any pathogenic potential whatsoever. Instead, there are a number of health-related benefits associated with bifidobacteria. Although definitive proof is missing in most cases, it is reasonable to assume that adhesion to host tissue by beneficial bacteria are also required for or support their health-promoting effects. Moreover, the impressive number of different adhesion factors encoded by individual strains of bifidobacteria suggests that adhesion to host tissue is important for bifidobacteria to colonize and strive in the highly competitive ecosystem of the GIT.

#### AUTHOR CONTRIBUTIONS

All authors listed, have made substantial, direct and intellectual contribution to the work, and approved it for publication.

#### FUNDING

This study was partially supported by the German Academic Exchange Service/Federal Ministry of Education and Research to CUR (Grant D/09/04778). CW was supported by a by Ph.D. fellowship of the "Landesgraduiertenförderung Baden-Württemberg." The funders had no role in the decision to publish or the content of the manuscript.



Bifidobacterium bifidum to Caco-2 cells. Appl. Environ. Microbiol. 74, 4695– 4702. doi: 10.1128/AEM.00124-08




**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 Westermann, Gleinser, Corr and Riedel. 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.

# Anti-viral Effect of Bifidobacterium adolescentis against Noroviruses

Dan Li<sup>1</sup> \*, Adrien Breiman2,3,4,5, Jacques le Pendu2,3,4† and Mieke Uyttendaele<sup>1</sup>†

<sup>1</sup> Laboratory of Food Microbiology and Food Preservation, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium, <sup>2</sup> UMR 892, Institut National de la Santé et de la Recherche Médicale, Nantes, France, <sup>3</sup> UMR 6299, Centre National de la Recherche Scientifique, Nantes, France, <sup>4</sup> Faculty of Medicine, University of Nantes, Nantes, France, <sup>5</sup> Nantes University Hospital, Nantes, France

This study aims to investigate the effect of Bifidobacterium adolescentis against noroviruses (NoVs). Murine norovirus-1 (MNV-1) used as a surrogate was detected by plaque assay and RT-qPCR. Human NoV virus like particles (VLPs) were detected by cell-binding assay. It was shown that the presence of B. adolescentis could inhibit the multiplication of MNV-1 on RAW 264.7 cells within 48 h of co-incubation period at 37◦C. This inhibition did not occur at the viral binding stage, as no difference was observed in MNV-1 genomic copies collected from washed RAW 264.7 cells without and with B. adolescentis after co-incubation for 1 h at room temperature. Meanwhile, the presence of B. adolescentis decreased the binding of human NoV GI.1 VLPs to both Caco-2 cells and HT-29 cells, while no reduction was induced for the binding of human NoV GII.4 VLPs to Caco-2 cells.

Edited by:

David Berry, University of Vienna, Austria

#### Reviewed by:

Douglas Morrison, University of Glasgow, UK Lorena Ruiz, Universidad Complutense de Madrid, Spain

\*Correspondence:

Dan Li dan.li@ugent.be

†These authors are co-senior authors.

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 23 January 2016 Accepted: 23 May 2016 Published: 08 June 2016

#### Citation:

Li D, Breiman A, le Pendu J and Uyttendaele M (2016) Anti-viral Effect of Bifidobacterium adolescentis against Noroviruses. Front. Microbiol. 7:864. doi: 10.3389/fmicb.2016.00864 Keywords: Norovirus, Bifidobacterium adolescentis, probiotics, mechanisms

# INTRODUCTION

Probiotics are defined as "living micro-organisms, which upon ingestion in certain numbers, exert health benefits beyond inherent basic nutrition." Most probiotic microorganisms belong to lactic acid bacteria (LAB), such as Lactobacillus sp., Bifidobacterium sp., and Enterococcus sp (Ljungh and Wadström, 2006). Probiotics may benefit the human and animal host directly, by preventing the infection and combating the causative agent of the intestinal disorder, or indirectly, by balancing the disrupted equilibrium of the enteric flora and augmenting the host's immune responses (Maragkoudakis et al., 2010; Sanders et al., 2013). Clinical evidence has reported that feeding of probiotics can prevent effectively for diarrhea and shedding of rotavirus (Saavedra et al., 1994; Van Niel et al., 2002; Sazawal et al., 2006; Grandy et al., 2010). Accordingly, with the use of animal models, multiple probiotic strains have shown anti-rotavirus effect (Muñoz et al., 2011; Kandasamy et al., 2014; Mao et al., 2016). In addition, in vitro studies have demonstrated that probiotics may have antiviral activity against rotavirus (Maragkoudakis et al., 2010; Muñoz et al., 2011), coxsackievirus (Kim et al., 2014), hepatitis C virus (El-Adawi et al., 2015), as well as noroviruses (NoVs; Aboubakr et al., 2014; Rubio-del-Campo et al., 2014).

Noroviruses, one genera of the Caliciviridae family, were reported as the cause of between 73% to greater than 95% of global epidemic nonbacterial gastroenteritis outbreaks and approximately half of all gastroenteritis outbreaks (Atmar and Estes, 2006). Although NoV infections are generally mild, it may require hospital care and can be associated with mortality in elderly, chronically ill or immune-compromised patients. Taking into consideration their widespreadness, NoVs are causing heavy disease burdens associated with large economic losses (Van Beek et al., 2013). Despite recent

progress, several key challenges remain in assessing the efficacy of vaccines and antiviral drugs for human NoV infection, such as the lack of a robust cell culture system or animal model limits direct study of these viruses, and the extreme genetic heterogeneity among strains (Karst et al., 2014). Therefore, it is of high interest to investigate further if probiotics can be employed for NoV control and treatment.

In this study, due to the non-cultivability of human NoVs, murine norovirus-1 (MNV-1, a very commonly used human NoV surrogate, Kniel, 2014), and human NoV virus like particles (VLPs) were used to study the effect of Bifidobacterium adolescentis against NoVs.

# MATERIALS AND METHODS

#### Bacteria

Bifidobacterium adolescentis (LMG10502, biological origin: adult intestines) was obtained from Belgian Coordinated Collection of Microorganisms (BCCM/LMG). B. adolescentis was cultured in tryptone soya broth (TSB, Oxoid, Thermo) at 37◦C. The anaerobic atmosphere was generated with the use of ANAEROGENTM COMPACT (Oxoid, Thermo).

Before each experiment, B. adolescentis were cultured for 48 h, normalized to an OD<sup>570</sup> of 0.4, and washed twice with phosphate buffered saline (PBS, pH 7.4).

## Cell Lines, Virus, and Virus-Like Particles (VLPs)

Cells of the murine macrophage cell line RAW 264.7 (ATCC TIB-71; kindly provided by Prof. H. W. Virgin, Washington University School of Medicine, St. Louis, MO, USA) were maintained in complete DMEM medium and grown at 37◦C under a 5% CO<sup>2</sup> atmosphere. Complete DMEM consisted of Dulbecco's modified Eagle's medium (DMEM; Lonza, Walkersville, MD, USA) containing 10% low-endotoxin fetal bovine serum (HyClone, Logan, UT, USA), 100 U/ml penicillin, 100 µg/ml streptomycin (Lonza), 10 mM HEPES (Lonza), and 2 mM L-glutamine (Lonza).

RAW 264.7 cells were infected with MNV-1.CW1 and passaged seven times at a multiplicity of infection (MOI) of 0.05 (MNV-1:cells) for 2 days. After two freeze-thaw cycles, low speed centrifugation was used to remove cellular debris from the virus suspension, as described by Wobus et al. (2004). The lysate containing suspended MNV-1 was stored in aliquots at –75◦C.

Cell line Caco-2 (ECACC 86010202) was cultured in Eagle's minimum essential medium with Earle's salts (EMEM; Lonza) supplemented with 10% low-endotoxin fetal bovine serum (HyClone), 100 U/ml penicillin, 100 µg/ml streptomycin (Lonza), and 2 mM <sup>L</sup>-glutamine (Lonza) and grown at 37◦C under a 5% CO<sup>2</sup> atmosphere. HT-29 (ATCC HTB-38) cells were cultured in Dulbecco's modified Eagle's medium (DMEM; GIBCO <sup>R</sup> , Life Technologies.) supplemented with 10% lowendotoxin fetal bovine serum (GIBCO <sup>R</sup> , Life Technologies), 100 U/ml penicillin, 100 µg/ml streptomycin (GIBCO <sup>R</sup> , Life Technologies), and 2 mM L-glutamine (GIBCO <sup>R</sup> , Life Technologies) and grown at 37◦C under a 5% CO<sup>2</sup> atmosphere.

Virus-like particles of human NoV GI.1 (Norwalk virus) and GII.4 (Dijon 1996) were generated using a previously described method (Jiang et al., 1992). Recombinant baculoviruses containing the VP1 protein from NoV GI.1 and GII.4 were generated, and VLPs were produced by infection of Hi5 insect cells. VLPs were released from the infected Hi5 cells by three rounds of freeze-thawing and then clarified by removal of cellular debris (6,000 × g for 30 min) and baculovirus (14,000 × g for 30 min). The VLPs were partially purified through a 30% (wt/vol) sucrose cushion in TNC buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10 mM CaCl2) containing the protease inhibitor leupeptin at 150,000 × g for 2 h. The pelleted VLPs were resuspended in TNC and further purified by isopynic centrifugation in cesium chloride (150,000 × g; 18 h). The resultant VLP bands were collected by puncture, and the solution containing VLPs was dialyzed against PBS prior to quantification by bicinchoninic acid (BCA) protein assay (Thermo Scientific) and stored at –80◦C.

# Cytotoxicity Test-Neutral Red Assay

The cytotoxicity of bacteria on RAW 264.7 cells was measured by a neutral red assay based on the description by Fotakis and Timbrell (2006) with a few modifications. The RAW 264.7 cells were seeded into 96-well plates at a density of 10<sup>5</sup> viable cells per well. On the following day, the RAW 264.7 cells were washed with PBS to remove the antibiotics, the B. adolescentis re-suspended in DMEM (Lonza) were added onto the cells (50 µl per well) and incubated for 1 h at room temperature. Afterward the inoculum was aspirated and fresh complete DMEM without antibiotics was added (100 µl per well). Plates were incubated at 37◦C and 5% CO2.

After 2 days incubation, the medium of the cells was changed to neutral red dye (Sigma–Aldrich, St. Louis, MO, USA, 100 µg/ml) dissolved in DMEM (100 µl per well) and incubated for another 2 h at 37◦C and 5% CO2. Cells were then washed with PBS and the addition of elution medium (EtOH/AcCOOH, 50%/1%, 100 µl per well) followed by gentle shaking for 10 min so that complete dissolution was achieved. The optical density was read at 540 nm (OD540).

# Viral Multiplication Inhibition Test on MNV-1 Detected by Plaque Assay

The RAW 264.7 cells were seeded into six-well plates at a density of 2 × 10<sup>6</sup> viable cells per well. On the following day, the B. adolescentis re-suspended in DMEM (Lonza) was used to make ten-fold dilutions from MNV-1 lysate. The RAW 264.7 cells were washed with PBS to remove the antibiotics, the mixture of bacteria and MNV-1 was added onto the cells (500 µl per well, two wells per sample). Plates were incubated for 1 h at room temperature and manually rocked every 15 min before aspirating the inoculum and overlaying the cells with 1.5% SeaPlaque agarose (Cambrex, Rockland, ME, USA) in MEM (Lonza) supplemented with 10% low-endotoxin fetal bovine serum, 1% HEPES, 1% penicillin/streptomycin, and 2% glutamine (complete MEM) per well. Plates were incubated at 37◦C and 5% CO<sup>2</sup> for 2 days. To visualize plaques, cells were stained with 1.5%

SeaKem agarose in complete MEM containing 1% neutral red (Sigma–Aldrich) per well for 6 h.

Plaque sizes were shown to be associated with the virulence in multiple studies (Takemoto, 1966; Lipton, 1980). In this study, from photos taken from each group with the same format and size, the diameters of plaques were measured and recorded by the use of ImageJ (National Institutes of Health, Bethesda, MD, USA).

# Viral Binding Inhibition Test on MNV-1 Detected by RT-qPCR

The RAW 264.7 cells were seeded into 96-well plates at a density of 10<sup>5</sup> viable cells per well. On the following day, the B. adolescentis was re-suspended in MNV-1 dilutions with DMEM (50 µl per sample, with 5.8 ± 0.1 log MNV-1 genomic copies per sample). The RAW 264.7 cells were washed with PBS to remove the antibiotics, the mixture of bacteria and MNV-1 were added onto the cells (50 µl per well) and incubated for 1 h at room temperature. Afterward the inoculum was aspirated and the cells were washed by PBS for three times.

The RNAs of washed cells were extracted by using the RNeasy Mini kit (Qiagen, Hilden, Germany). For each sample the 100 µl PBS was firstly mixed with 350 µl lysis buffer, and the mixture was pipetted back onto the cells in the 96-well plates in order to lyse the cells. The following procedures were performed according to the RNA Cleanup protocol and the RNA were stored at –75◦C.

The RT-qPCR of MNV-1 was performed by the use of RNA UltraSenseTM One-Step Quantitative RT-PCR System (Life technologies). The primers and probe sequence of MNV-1 were previously shown by Baert et al. (2008). Twenty micro liter of reaction mixture [200 nM each primer, 200 nM probe, 50 nM ROX (Life technologies)] was added to 5 µl of RNA. The RTqPCR assays were performed in an ABI 7300 system (Applied Biosystems). The amplification profile consisted of 50◦C for 15 min, 95◦C for 2 min and 45 cycles of 95◦C for 15 s and 60◦C for 30 s.

An absolute quantification of MNV-1 genomic copies was performed as described by Baert et al. (2008). To obtain representative positive control standards, the plasmid p20.3 containing a full-length cDNA clone of MNV-1.CW1 (Baert et al., 2008) was used for the quantifications. Ten-fold serial dilutions ranging from 10<sup>6</sup> to 10 copies of plasmids per reaction were used to prepare the standard curves.

# Viral Binding Inhibition Test on Human NoV VLPs Detected by Fluorescence Measurement

The HT-29 cells were seeded into 96-well plates at a density of 10<sup>5</sup> viable cells per well. On the following day, the B. adolescentis was re-suspended in NoV GI.1 VLP suspension [5 µg/ml, in PBS-0.1% bovine serum albumin (BSA)]. The mixture of bacteria and VLPs was added onto the cells (50 µl per well) and incubated for 1 h at room temperature. After washing with PBS for three times, the cells were fixed with 4% paraformaldehyde, and stained by anti-VLP rabbit polyclonal antibodies (lp130 for GI.1, diluted in PBS-0.1% BSA, 1-h incubation at 37◦C) followed by Alexa Fluor <sup>R</sup> 488 Goat Anti-Rabbit IgG (H+L) antibody (Life Technologies, diluted in PBS-0.1% BSA, 1-h incubation at 37◦C). Three times washing by PBS was always performed after each step. The fluorescence was read by a fluorimeter (Fluoroskan, Thermo Scientific; Ex/Em = 490/525 nm) in arbitrary unit (a.u.).

The binding test of human NoV GI.1 and GII.4 VLPs (anti-VLP rabbit polyclonal antibodies lp130 for GI.1 and lp132 for GII.4) on Caco-2 cell were similar with the procedures above except that the Caco-2 cells were incubated for 21 days post confluency to be used as differentiated cells.

The binding of NoV GI.1 VLPs to HT-29 cells was also observed with a fluorescence microscope. The HT-29 cells were seeded into eight-well Nunc <sup>R</sup> Lab-Tek <sup>R</sup> Chamber Slide System (Sigma–Aldrich) at a density of 10<sup>5</sup> viable cells per well. On the following day, the binding and staining steps were the same as described above. After the last washing step (from the secondary antibody incubation), the upper structure was removed from the bottom glass slide. The stained cells were mounted on slides with Vectashield <sup>R</sup> (Vector laboratories, Burlingame, CA, USA). The sealed slides were observed under a Zeiss Axiovert 200 fluorescence microscope.

# Data Analysis

Each result was presented as the mean value of three independent replicates with the standard deviation. Statistical analyses were performed by Mann–Whitney U test with SPSS 22 for Windows (SPSS, Inc., Chicago, IL, USA). Significant differences were considered when P was <0.05.

# RESULTS

# Effect of B. adolescentis on MNV-1 Infectivity

First of all, after 48 h incubation, the cell viability determined by neutral red assay showed no significant reduction after incubation of the RAW 264.7 cells with B. adolescentis (OD<sup>540</sup> 1.1 ± 0.1–1.1 ± 0.2, P > 0.05).

MNV-1 lysate was diluted to a concentration that can form countable plaques on the six-well plates, and was seeded onto the cells with or without B. adolescentis. The plaque assay showed that compared with the negative control (MNV-1 on RAW 264.7 cells without bacteria), the MNV-1 plaque forming units (PFU) from cell-culture wells in the presence of B. adolescentis were decreased significantly from 20 ± 3–7 ± 2 PFU/well (**Figure 1A**, P < 0.05).

It was also noticed that the sizes of the plaques from cells incubated with B. adolescentis were smaller than the negative control (examples shown in **Figure 1C**). The average sizes of the plaques measured by ImageJ showed that compared with the negative control (MNV-1 on RAW 264.7 cells without bacteria), the plaque diameters from cell-culture wells in the presence of B. adolescentis were decreased significantly from 9.2 ± 2.7– 3.9 ± 0.6 pixel (**Figure 1B**, P < 0.05). These results indicate that the effect of B. adolescentis on MNV-1 may occur mainly in the viral replication phase instead of showing direct virucidal effect on the MNV-1 or preventing the viruses from binding to the cells.

Therefore lastly, a viral binding inhibition test was performed and indeed no significant difference of MNV-1 genomic copies collected from RAW 264.7 cells between the two groups with (5.77 ± 0.03 log-genomic copies/ml) and without B. adolescentis (5.74 ± 0.01 log-genomic copies/ml) was observed (P > 0.05).

# Effect of B. adolescentis on Human NoV VLPs Cell-Binding Ability

It has been previously reported that both VLPs of human NoV GI.1 and GII.4 could bind to Caco-2 cells (Tamura et al., 2004) and VLPs of human NoV GI.1 could bind to HT-29 cells (Rubiodel-Campo et al., 2014).

The presence of B. adolescentis decreased the binding of human NoV GI.1 VLPs to Caco-2 cells, measured by fluorescence intensity, significantly from 45.2 ± 5.0 (negative control without bacteria) to 33.0 ± 4.6 a.u. (incubation with B. adolescentis; **Figure 2A**, P < 0.05).

The fluorescence intensity of human NoV GI.1 VLPs on HT-29 cells also decreased, although not significantly, from 92.0 ± 25.0 to 52.0 ± 5.0 a.u. by the incubation of B. adolescentis (**Figure 2B**, P = 0.05). A visual example detected by fluorescence microscopy was shown in **Figure 2C**.

As for human NoV GII.4 VLPs, no significant reduction was observed for the binding of the VLPs to Caco-2 cells with the incubation of B. adolescentis (19.0 ± 1.7 to 17.7 ± 0.6 a.u., P > 0.05).

# DISCUSSION

Currently, the research on the effect of probiotics on NoVs are still preliminary. On one hand, due to the non-cultivability of human NoVs, the in vitro studies were conducted with the use of surrogate viruses [e.g., feline calicivirus (FCV), Aboubakr et al., 2014; MNV-1 and Tulane virus (TV), Shearer et al., 2014] and artificially synthesized particles partially mimicking the viral structures (P-particles, Rubio-del-Campo et al., 2014), which may introduce gaps from the authentic NoV reactions. On the other hand, there are an insufficient number of indepth studies on this topic before a general conclusion can be drawn. For instance, it was reported that the fermentation of Donchimi and oysters could effectively reduce the infectivity of FCV and MNV-1 (Lee et al., 2012; Seo et al., 2014). The population of LAB increased largely during the fermentation, however, it was neither clear on the role that the LAB played nor the associated mechanisms (Lee et al., 2012; Seo et al., 2014). For another instance, high infectivity reduction (6 –7 log reduction) was observed for FCV following co-incubation with Lactococcus lactis (both bacterial medium filtrate and cells suspension, Aboubakr et al., 2014). However, the cell-free supernatant of a commercial probiotic mixture of Lactobacillus acidophilus, Lactobacillus rhamnosus, Bifidobacterium bifidum, Lactobacillus salivarius, and Streptococcus thermophiles indicated no reduction for the infectivity of MNV and TV (Shearer et al., 2014). The reasons causing the inconsistence of the studies can be the differences between both of the studied viruses and bacteria strains, as well as the experimental settings such as time/sequence

of addition, comparative ratio of viruses and bacteria, incubation conditions (time, atmosphere and temperature) and matrices, etc.

microscope observation of human NoV GI.1 VLPs bound to HT-29 cells.

In this study, MNV-1 was employed as the surrogate as it was more persistent than FCV in the fermentation of both Donchimi and oysters (Lee et al., 2012; Seo et al., 2014). Human NoV VLPs from GI.1 and GII.4 were used as they could mimic the binding capacity of two representative NoV genotypes. B. adolescentis is a recognized probiotics (Lee et al., 1993; Kim et al., 2008) and has interested dairy manufacturers in producing "therapeutic fermented milk products" (Arunachalam, 1999; Puniya, 2015). Also it was reported recently to exhibit antiviral activity against Coxsackievirus (Kim et al., 2014). The bacterial cells were washed

with PBS before being added to the viruses and cells in order to avoid any direct virucidal effect of the bacterial metabolites in the culture medium, such as organic acids, diacetyl, and bacteriocins. The bacteria and viruses (MNV-1 and human NoV VLPs) were co-incubated with the cells as this strategy was shown to be more effective than pre-treatment of cells, pre-treatment of viruses (Aboubakr et al., 2014; Rubio-del-Campo et al., 2014), or posttreatment of cells attached to viruses (Rubio-del-Campo et al., 2014).

It was demonstrated in this study that the presence of B. adolescentis could inhibit the multiplication of MNV-1 on RAW 264.7 cells. Meanwhile, it was shown that this inhibition effect did not occur on the binding stage of MNV-1 to RAW 264.7 cells. However, in contrast, the presence of B. adolescentis did decrease the binding of human NoV GI.1 VLPs to both Caco-2 cells and HT-29 cells. Although based on different models, these results indicate that the effects of B. adolescentis on different viruses might vary based on different mechanisms.

It was reported previously that NoV P-particles could bind to a series of probiotics with varied binding ability, which was postulated as one of the antiviral mechanisms of the probiotics (Rubio-del-Campo et al., 2014). However, based on a recent study of our group (Li et al., 2015), the B. adolescentis strain used in this study does not express histo-blood group antigen (HBGA) and could not bind to NoV VLPs of either GI.1 or GII.4. This result rules out the possibility of direct competition of B. adolescentis and NoV GI.1VLPs from binding to Caco-2 or HT-29 cells. Instead, as we and others have described a few bacterial lectins binding to HBGAs (Gilboa-Garber et al., 2011; Audfray et al., 2012) and Zhong et al. (2004) have reported that the adhesin of B. adolescentis 1027 could inhibit the adhesion of Escherichia coli to intestinal epithelial cell line Lovo, we assume that certain lectins secreted by B. adolescentis could bind to the HBGAs on the intestinal cell lines and compete with the binding of NoV GI.1VLPs.

As the binding of human NoV GI.1 VLPs to both Caco-2 cells and HT-29 cells was decreased in the presence of B. adolescentis, it is reasonable to assume that the viral multiplication of human NoV GI.1, if an in vitro model could be established, should also be decreased in the presence of B. adolescentis. On the other hand, although no reduction of the binding of human NoV GII.4 VLPs to Caco-2 cells was observed in the presence of B. adolescentis, the possibility cannot be ruled out that B. adolescentis may reduce the viral multiplication of human NoV GII.4 if an in vitro model could be established, or help combat NoVs in vivo. The differences observed between the two strains are not surprising as it is well known that the chemical

#### REFERENCES


structures of the receptor binding interfaces are different between NoV GI.1 and GII.4 (Tan and Jiang, 2011). However, the exact molecular mechanisms casing the influence of B. adolescentis on human NoV VLPs binding to intestinal cell lines still need further investigation.

# CONCLUSION

This study demonstrated the antiviral effect of B. adolescentis against NoVs. Due to the lack of models to study the infection of genuine human NoVs, the results were generated from studying surrogates detected by different methods (infection and binding of MNV-1 on murine macrophage cell line RAW 264.7and binding of human NoV VLPs on human intestinal epithelial cell lines Caco-2 and HT-29). Since the effects of B. adolescentis on MNV-1 and human NoV VLPs as well as their mechanisms were indicated to be different, this study shows the importance of establishing multiplication model for human NoVs in the future. In addition, more probiotic strains will be tested before a general conclusion can be drawn on the antiviral effect of probiotics on NoV infection.

# AUTHOR CONTRIBUTIONS

DL performed most of the experiments and composed the manuscript. AB supplied technical support for the experiments, especially for the fluorescence microscope as well as improvement of the manuscript. JP and MU gave strategic guidance for the work and suggestions to improve the manuscript.

# FUNDING

This study was supported by a postdoctoral grant (DL.) and a financial support for staying abroad (DL.) from the Research Foundation-Flanders (Fonds voor Wetenschappelijk Onderzoek-Vlaanderen) and by a grant from the Region des Pays de la Loire (ARMINA).

# ACKNOWLEDGMENT

The authors also acknowledge the help of imaging core facilities of the University of Nantes (Micropicell).


plaque assay, transfection assay, and real-time reverse transcription-PCR before and after heat treatment. Appl. Environ. Microbiol. 74, 543–546. doi: 10.1128/AEM.01039-07


**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 Li, Breiman, le Pendu and Uyttendaele. 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.

fmicb-07-00864 June 6, 2016 Time: 12:34 # 7

# *Bifidobacterium animalis* ssp. *lactis* CNCM-I2494 Restores Gut Barrier Permeability in Chronically Low-Grade Inflamed Mice

#### *Edited by:*

*Francesca Turroni, University College Cork, Ireland*

#### *Reviewed by:*

*Miguel Gueimonde, Consejo Superior de Investigaciones Científicas, Spain Valentina Taverniti, Universita degli Studi di Milano, Italy*

#### *\*Correspondence:*

*Rebeca Martín rebeca.martinrosique@jouy.inra.fr*

#### *†Present address:*

*Sylvie Miquel, Laboratoire Microorganismes: Genome Environnement (LMGE), UMR CNRS 6023, Clermont Université, Université d'Auvergne, Clermont-Ferrand, France; Johan van Hylckama Vlieg, Chr. Hansen A/S, Hørsholm, Denmark*

#### *Specialty section:*

*This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology*

*Received: 29 February 2016 Accepted: 12 April 2016 Published: 06 May 2016*

#### *Citation:*

*Martín R, Laval L, Chain F, Miquel S, Natividad J, Cherbuy C, Sokol H, Verdu EF, van Hylckama Vlieg J, Bermudez-Humaran LG, Smokvina T and Langella P (2016) Bifidobacterium animalis ssp. lactis CNCM-I2494 Restores Gut Barrier Permeability in Chronically Low-Grade Inflamed Mice. Front. Microbiol. 7:608. doi: 10.3389/fmicb.2016.00608*

*Rebeca Martín1\*, Laure Laval1,2, Florian Chain1, Sylvie Miquel1†, Jane Natividad3, Claire Cherbuy1, Harry Sokol1,4,5, Elena F. Verdu3, Johan van Hylckama Vlieg2†, Luis G. Bermudez-Humaran1, Tamara Smokvina2 and Philippe Langella1*

*<sup>1</sup> Commensal and Probiotics-Host Interactions Laboratory, Micalis Institute, Institut National de la Recherche Agronomique, AgroParisTech, Université Paris-Saclay, Jouy-en-Josas, France, <sup>2</sup> Danone Nutricia Research, Palaiseau, France, <sup>3</sup> Farncombe Family Digestive Health Research Institute, McMaster University, Hamilton, ON, Canada, <sup>4</sup> ERL INSERM U 1057/UMR7203, Faculté de Médecine Saint-Antoine, Université Pierre et Marie Curie, Paris, France, <sup>5</sup> Service de Gastroentérologie, Hôpital Saint-Antoine, Assistance Publique – Hôpitaux de Paris, Paris, France*

Growing evidence supports the efficacy of many probiotic strains in the management of gastrointestinal disorders associated with deregulated intestinal barrier function and/or structure. In particular, bifidobacteria have been studied for their efficacy to both prevent and treat a broad spectrum of animal and/or human gut disorders. The aim of the current work was thus to evaluate effects on intestinal barrier function of *Bifidobacterium animalis* ssp. *lactis* CNCM-I2494, a strain used in fermented dairy products. A chronic dinitrobenzene sulfonic acid (DNBS)-induced low-grade inflammation model causing gut dysfunction in mice was used in order to study markers of inflammation, intestinal permeability, and immune function in the presence of the bacterial strain. In this chronic low-grade inflammation mice model several parameters pointed out the absence of an over active inflammation process. However, gut permeability, lymphocyte populations, and colonic cytokines were found to be altered. *B. animalis* ssp. *lactis* CNCM-I2494 was able to protect barrier functions by restoring intestinal permeability, colonic goblet cell populations, and cytokine levels. Furthermore, tight junction (TJ) proteins levels were also measured by qRT-PCR showing the ability of this strain to specifically normalize the level of several TJ proteins, in particular for claudin-4. Finally, *B. lactis* strain counterbalanced CD4+ lymphocyte alterations in both spleen and mesenteric lymphoid nodes. It restores the Th1/Th2 ratio altered by the DNBS challenge (which locally augments CD4+ Th1 cells) by increasing the Th2 response as measured by the increase in the production of major representative Th2 cytokines (IL-4, IL-5, and IL-10). Altogether, these data suggest that *B. animalis* ssp. *lactis* CNCM-I2494 may efficiently prevent disorders associated with increased barrier permeability.

Keywords: micro-inflammation, apical junction proteins, goblet cells

# INTRODUCTION

The intestinal barrier is an effective defense mechanism that depends on the integrity of the cells and the junctional complexes between them. The gut barrier is a functional unit organized as a multilayer system composed by a physical barrier which prevents bacterial adhesion and regulates paracellular diffusion and a functional layer able to discriminate between pathogens and commensal microorganisms (Lopetuso et al., 2015). The physical barrier is formed by a mucus layer followed by a monolayer of epithelial cells (Denker and Nigam, 1998; Natividad and Verdu, 2013) performing the paracellular transport across the barrier controlled by apical junction proteins (Natividad and Verdu, 2013). The mucus protects the epithelium from harmful microorganisms and antigens being also a lubricant for intestinal motility (Lopetuso et al., 2015). Outer mucus is composed of the highly glycosylated mucin MUC2 protein produced by the goblet cells (Lopetuso et al., 2015). The regulation of its function is mediated by both endogenous and exogenous factors (Agostini et al., 2012; Distrutti et al., 2013) and is a key factor in the development of several diseases involving altered gut permeability and dysfunction such as irritable bowel syndrome (IBS), food allergies, type-1 diabetes, and obesity (Perrier and Corthésy, 2011; Camilleri et al., 2012; Vaarala, 2012). Diverse microorganisms have shown to protect barrier integrity and promote its restoration when damaged. Among them, increasing evidence points out that strains of lactic acid bacteria (Gaudier et al., 2004) and bifidobacteria regulate gut barrier function using different mechanisms (Agostini et al., 2012; Distrutti et al., 2013). For instance, *Lactobacillus rhamnosus* GG (LGG), *B. breve* NCC2950 and a mixture of lactobacilli and bifidobacteria (*L. casei, L. plantarum, L. acidophilus, L. delbrueckii ssp. bulgaricus, B. longum*, *B. breve, and B. infantis*) prevent the increase in intestinal permeability *in vivo* (Ukena et al., 2007; Mennigen et al., 2009; Donato et al., 2010; Natividad et al., 2013).

Bifidobacteria, naturally present in the colonic microbiota, correspond to up to 80% of the cultivable fecal microorganisms in full-term breastfed infants (Picard et al., 2005). They have been traditionally considered as safe microorganisms, due to their Generally Recognized As Safe (GRAS) status and are widely used as health-promoting bacteria in functional foods. Especially, *B. animalis* ssp. *lactis* (*B. lactis*) CNCM I-2494 has a long history of use in fermented dairy products and shows a high gastrointestinal survival (Picard et al., 2005; Rochet et al., 2008). A fermented milk product (FMP)-containing *B. lactis* CNCM I-2494 together with lactic acid bacterial starter cultures has shown positive effects on gut function in several randomized controlled studies (Picard et al., 2005) improving: (i) gastrointestinal wellbeing and digestive symptoms in women reporting minor digestive problems (Guyonnet et al., 2009a), (ii) abdominal girth and gastrointestinal transit (Agrawal et al., 2009), (iii) health related quality of life and symptoms in IBS in adults (Guyonnet et al., 2009b), and (iv) colonic transit time and minor digestive problems in healthy women (Marteau et al., 2002, 2013). The physiological effects of this strain have been also evaluated in animal studies where it has been capable to reduce the aberrant crypts incidence in chemically induced carcinogenesis models in rats (Tavan et al., 2002), improve colitis in mice (Veiga et al., 2010), hydrolyze bile salts in the gastrointestinal tract of pigs (Lepercq et al., 2004), and prevent the increase of intestinal permeability induced by partial restraint stress in rats (Agostini et al., 2012). The molecular mechanisms underlying the positive effects of strain CNCM I-2494 are far from being completely understood although its genome have been sequenced (Chervaux et al., 2011). Recent identification of several restriction and modification systems in this strain and development of specific molecular tools opened the way in studying specific bacterial mechanisms involved in the cross-talk of strain CNCM I-2494 with the host (O'Connell Motherway et al., 2014).

The clear relationship between *B. lactis* CNCM I-2494 and the protection of gut dysfunction in both animal models and clinical trials combined to the industrial importance of this strain has prompted us to deeper analyze its possible effects on an altered permeability and gut dysfunction model. Gut dysfunction was achieved thanks to a first inflammatory insult followed with a second subclinical chemical challenge as previously described (Laval et al., 2015; Martin et al., 2015). The aim of this work was to clarify the direct effect of the strain in the murine intestinal epithelium barrier and function.

# MATERIALS AND METHODS

#### Bacterial Growth Conditions and Animals

*Bifidobacterium animalis* ssp. *lactis* CNCM-I2494 was grown in MRS medium (Difco, USA) supplemented with cysteine (0.5 mg/ml; Sigma–Aldrich) under anaerobic conditions at 37◦C.

Male C57BL/six mice (6–8 weeks old; Janvier, Le Genest Saint Isle, France) were maintained at the animal care facilities of the National Institute of Agricultural Research (IERP, INRA, Jouyen-Josas, France) under specific pathogen-free (SPF) conditions. Mice were housed under standard conditions for a minimum of 1 week before experimentation. All experiments were performed in accordance with European Community rules for animal care and were approved by the relevant local committee (Comethea). Protocol number 02550.01.

#### Experimental Design

Inflammation was induced as previously described (Laval et al., 2015) (**Supplementary Figure S1**). Briefly, mice where challenged, under anesthesia, with a first intra-rectal dose of 100 mg/Kg of dinitrobenzene sulfonic acid (DNBS) solution (ICN, Biomedical Inc.) in 30% ethanol (EtOH). Control mice (without colitis) received only 30% EtOH. Thirteen days after the first DNBS injection, 5 <sup>×</sup> 109 CFU of viable bacteria in 200 <sup>μ</sup><sup>l</sup> of PBS or PBS alone were administered intra-gastrically, daily for 10 days (gavage period). Finally, 21 days after the first challenge, the mice were challenged again with a second administration of 50 mg/kg of DNBS or EtOH. Weight loss was monitored during 3 days following the second DNBS injection to assess possible clinical signs of distress.

To confirm the absence of over inflammation, colonic macroscopic and histological scores as well as colonic myeloperoxidase (MPO) activity (a marker of the degree of infiltration by polymorphonuclear neutrophils) and serum lipocalin-2 levels (an early inflammation marker) were determined as previously described (Shashidharamurthy et al., 2013; Martin et al., 2014; Laval et al., 2015).

#### Histological Features Analysis

Flushed colons were fixed in 4% paraformaldehyde or Carnoy buffer, dehydrated and embedded in paraffin according to a standard protocol. Histological features were analyzed by hematoxylin–eosin–safran (Perrier and Corthésy, 2011) staining. Periodic acid-Schiff (PAS) and Alcian blue (AB) staining were performed as in Wrzosek et al. (2013).

# Intestinal Permeability *In Vivo*

Permeability *in vivo* was assessed using fluorescein isothiocyanate-conjugated dextran (FITC–dextran 3000– 5000 Da, Sigma–Aldrich) tracer as previously described (Tambuwala et al., 2010). Briefly, at the endpoint 0.6 mg/g body weight of FITC–dextran dissolved in PBS was administered to mice by oral gavage. To measure the presence of FITC–dextran in blood, 3.5 h after the gavage blood samples were recovered from the retro-orbital venous plexus and kept in dark at 4◦C until analysis. Mice were housed under standard conditions during this period with un-limited access to water and food. Serum has separated by centrifugation and plasma FITC levels were determined using a fluorescence microplate reader (excitation 485 nm and emission 530 nm; Tecan, Lyon, France).

# Apical Junctional Analysis by Quantitative Real-time PCR (qPCR)

Total RNA was isolated from 20 to 30 mg samples of colon with an RNeasy Mini Kit (Qiagen) as previously described (Laval et al., 2015). qPCR was performed with diluted cDNA (10×) in triplicate and with an iQ5 Real-Time Detection System (Bio-Rad). The reaction mix consisted of Ssofast Evagreen Supermix (Bio-Rad), primers at 0.5 μM (Martin et al., 2015), and 2 μL of diluted cDNA. Values are expressed as relative fold differences normalized to a housekeeping gene, *Gapdh,* by the 2−--*<sup>C</sup>*<sup>T</sup> method. All procedures were performed according to the manufacturers' instructions.

#### Analyses of Lymphoid Populations Present in the Spleen and in the Mesenteric Lymphoid Nodes (MLNs)

Mononuclear cells were isolated from spleens and MLN by gentle extrusion of the tissue through a 50 μm-mesh Nylon cell strainer (BD). Cells were suspended in Dulbecco's Modified Eagle Medium (DMEM) medium supplemented with 10% of fetal calf serum (FCS), 2 mM L-glutamine, 50 U/mg penicillin, and 50 U/mg streptomycin (Lonza, Levallois-Perret, France). Erythrocytes were lysed with red blood-cell lysing buffer (Sigma– Aldrich).

For flow cytometry analysis, aliquots of 106–107 cells per sample were pre-incubated with purified anti-mouse CD16/ CD32 (eBioscience, San Diego, CA, USA) and then labeled with anti-CD4-FITC, anti-CD3e-PE, and anti-CD8-PerCP (all from eBioscience) according to the manufacturer's instructions. The stained cells were analyzed by flow cytometry (Accuri, BDbioscience) with CFlow Sampler software (BD).

For stimulation experiments, 2 <sup>×</sup> 105 cells per well were cultured for 48 h (37◦C, 10% CO2) in DMEM medium in P24 plates pre-coated with anti-CD3/CD28 antibodies (4 μg/mL each; eBioscience) or phorbol 12-myristate 13-acetate (PMA)/ionomycin (cell stimulation cocktail, 1×, ebioscience). Culture supernatant was frozen at −80◦C until processing.

## Cytokine Assays

Blood samples were obtained from the retro-orbital venous plexus before the mice were euthanized and centrifuged, and the sera stored at −80◦C until analysis. One centimeter samples of distal colon were recovered and homogenized in an appropriate volume of PBS (final concentration of 50 mg/ml) in a Tissue Lyser (Qiagen). IL-6, IL-10, IFN-γ, TNF-α, IL-5, IL-2, IL-22, IL-1α, IL-13, IL-17, IL-4, IL-27, and IL-12p70 were assayed in blood and colon samples with a cytometric bead array system (Mouse Th1/Th2/Th17/Th22 13plex Flowcytomix; eBioscience, San Diego, CA, USA). For cytokine quantification in cell culture supernatants the following ELISA tests were performed according to manufacturer's instruction: IL-4, IL-5, IFNγ, IL-17, IL-12p70, and IL-10 (MabTech); TGFβ and IL-22 (ebioscience).

# Statistical Analysis

GraphPad software (GraphPad Sofware, La Jolla, CA, USA) was used for statistical analysis. Results are presented as bar graphs or dot plots with means ± SEM. Comparisons involved the nonparametric Kruskal-Wallis test followed by a Dunn's Multiple Comparison test. A *p* value of less than 0.05 was considered significant.

# RESULTS

# Confirmation of Micro-inflammation in DNBS Challenged Mice

The induction of a low-grade inflammation status following a chronic low-dose DNBS in the mice was confirmed through the follow-up of health, histological and inflammatory parameters (**Supplementary Figure S1**). In particular, weight loss (**Supplementary Figure S2A**), colonic macroscopic and histological scores (**Supplementary Figures S2B,C**) as well as the MPO activity in the colon (**Supplementary Figure S2D**) and the Lipocalin-2 concentration in serum (**Supplementary Figure S2E**) were measured. The absence of differences for all these parameters among the groups, even in presence of *B. lactis* CNCM-I2494, added to the lack of detection of cytokine levels in serum samples (IL-6, IL-10, IFN-γ, TNF-α, IL-5, IL-2, IL-22, IL-1α, IL-13, IL-17, IL-4, IL-27, and IL-12p70, data not shown) discards the presence of an overt and active inflammation in this model.

However, the presence of slightly elevated, although no statistically significant, cytokines IL-13, IL-1α, IL-6, IL-22, IL-2, IL-27, IL-4, IFN-γ, and TNF-α levels in colonic tissues, compared to healthy controls, suggest a local low-grade inflammation (**Figure 1**). Treatment with *B. lactis* CNCM-I2494 reduced these increases in cytokine production (**Figure 1**): Notably, restoration was statistically significant for IL-2, IL-13, and IFNγ (*p* < 0.05).

# *Bifidobacterium lactis* CNCM-I2494 Restores Colonic Permeability by Modulating Apical Junction Protein Levels

The integrity of the gut barrier was assessed by the analysis of the permeability with the paracellular tracer FITC-dextran

*in vivo* at the endpoint. Of note, all the animals were submitted to exactly the same protocol and waiting time to avoid differences due to a minimal possible clearance phenomenon due to renal function. The mice treated with DNBS showed high permeability to the tracer (*<sup>p</sup>* <sup>&</sup>lt; 0.05) (**Figure 2A**) confirming an alteration in the barrier permeability as it has been previously observed (Laval et al., 2015; Martin et al., 2015). The oral administration of *B. lactis* CNCM-I2494 strain resulted in a decrease in permeability (*p* < 0.05). To further analyze the effect on the barrier function the expression of the relevant mRNAs of adherent junction (AJ) and tight junction (TJ) proteins were measured by qRT-PCR (**Figure 2B**). The mRNAs for Claudin-3, 4, E-cadherin, Occludin, and the zona occludens proteins (ZO-1) were all less abundant in DNBS-treated mice than in control mice (*p* < 0.05). CNCM-I2494 tends to partially re-establish the levels of all of them (**Figure 2B**). Notably, this effect was statistically significant for Claudin-4 (*p* < 0.05). Taken together, both the histological analysis and the transcriptional data demonstrate that strain *B. lactis* CNCM I-2494 protects against DNBS-induced chronic barrier dysfunction.

# *Bifidobacterium lactis* CNCM-I2494 Restores Goblet Cell Population Altered by DNBS Chronic Challenge

Histological features, analyzed by hematoxylin–eosin–safran (Perrier and Corthésy, 2011) staining, showed no significant differences in general morphology, crypt depth or total numbers of cells per crypt (data not shown). The numbers of goblet cells stained either by AB (**Figure 3A**), specific for acidic mucopolysaccharides, or PAS (**Figure 3B**), specific for neutral mucopolysaccharides, were significantly lower in DNBS challenged control group (*p* < 0.05). *B. lactis* CNCM-I2494 was able to enhance the percentage of AB or PAS positive cells per crypt (*p* < 0.05) reaching the values of the non-inflamed control group (**Figure 3**).

# *Bifidobacterium. lactis* CNCM I-2494 Modulates CD3**+**/CD4**+** T-Cell Populations in Spleen and MLNs by Increasing T Helper (Th) Profile 2

To study further the mechanism by which *B. lactis* CNCM-I2494 exerts protective function, T-cells from spleen and MLN were isolated and analyzed by flow cytometry. DNBS-treated mice showed lower CD3+/CD4+ T-cell percentages in spleen (**Figure 4A**) than the control group and higher CD3+/CD4<sup>+</sup> cell percentages in MLN (**Figure 5A**; *<sup>p</sup>* <sup>&</sup>lt; 0.05). *B. lactis* CNCM-I-2494 tends to reduce the CD3+/CD4+ decrease in spleen (**Figure 4A**) and significantly control CD3+/CD4<sup>+</sup> increase in MLN (*<sup>p</sup>* <sup>&</sup>lt; 0.05; **Figure 5A**). No variations were observed in CD3+/CD8+ T-cell percentages in spleen or MLN (data not shown).

As variations in CD4+ T-cell populations were found, MLN and spleen cells were cultured after isolation in the presence of two different stimulators during 48 h: CD28+/CD3+ to specifically stimulate lymphocytes and PMA/IO to stimulate all the cells present in the organ disaggregate. Representative cytokines of the major Th profiles (IL-4, IL-5, IFNγ, IL-17, IL-12p70, IL-10, TGFβ, and IL-22) were determined in the culture supernatants (**Figures 4B** and **5B** and data not shown). The IL-17, IL-22, IL-12p70, and TGFβ levels as well as IFNγ in spleen samples were under the ELISA detection limits (2.4, 5.5, 8.6, 10, and 6.5 pg/ml, respectively; data not shown). This fact, in addition to slight increases of Th1 levels (IFNγ) by the DNBS treatment in MLN (**Figure 5B**) confirms the low-grade inflammation status of the mice model. Differences were found in the levels of IL-4, IL-5, and IL-10 in both spleen (**Figure 4B**) and MLN (**Figure 5B**) (*<sup>p</sup>* <sup>&</sup>lt; 0.05). Strain CNCM-I2494 increased Th2 levels as measured by IL-4 and IL-10 augmentation in both spleen (**Figure 4B**) and MLN (**Figure 5B**) and also IL-5 in spleen samples corresponding to an anti-inflammatory patter in this model. This anti-inflammatory patter has been confirmed locally by the INFγ/IL-4 ratio in MLN samples (**Figure 5B**). Nevertheless, CNCM-I2494 was not able to control the small increase in IFNγ caused by the DNBS challenge (**Figure 5**). Finally, DNBS treatment caused also an increase in IL-5 in MLN samples. No significant differences were found between CD3+/CD28+ and PMA/IO stimulations, excepting IL-4 and IL-5 in spleen where a major level of stimulation was achieved with the first one (**Figures 4** and **5**). Taken together these data demonstrate that CNCM-I2494 strain is able to counterbalance the Th1/Th2 ratio altered by the DNBS challenge (which locally augments CD4+ Th1 cells) by increasing the Th2 response as measured by the increase in the production of major representative Th2 cytokines.

# DISCUSION

Epithelial barrier dysfunction is now considered as one of the major contributors to the development of several diseases and syndromes (Perrier and Corthésy, 2011; Camilleri et al., 2012; Vaarala, 2012). In several of them, such as IBS, studies suggest an interplay between luminal factors (e.g, foods and bacteria residing in the intestine), the epithelial barrier, and the mucosal immune system (Barbara et al., 2012). In a healthy state, the epithelial barrier allows a low translocation of luminal antigens by paracellular transport by receptormediated or non-selective endocytosis (Natividad and Verdu, 2013). Therefore, a higher local antigen exposure caused by an increase of intestinal permeability could activate intestinal immune system and inflammation may thus occur (Ohman and Simren, 2007; Natividad and Verdu, 2013). Preclinical studies have shown that selective probiotic strains exhibit the potential to improving mucosal barrier homeostasis (Barbara et al., 2012).

As related above, the administration of fermented milk containing *B. lactis* CNCM I-2494 has been found to prevent *in vivo* the increase of intestinal permeability in rats (Agostini et al., 2012). However, due to possible synergistic interplay of the different strains and/or metabolites contained in this product the specific effect of this *B. lactis* strain on gut

barrier is still unknown. Here, we aimed to clarify the specific effect of *B. lactis* CNCM-I2494 strain on intestinal barrier function.

As previously observed, DNBS-treated mice showed alteration in gut permeability (Laval et al., 2015). *In vivo* values with the paracellular tracer FITC-dextran showed an increase in permeability in DNBS-challenged mice that was restored when mice were treated with *B. lactis* CNCM-I2494 suggesting an effect of the strain on gut barrier function. To better decipher the beneficial effect of *B. lactis* CNCM-I9434 strain on gut permeability, mucus producing cells were analyzed by two different specific staining: AB, which specifically stains acidic mucopolysaccharides and PAS staining, specific for neutral mucopolysaccharides. Both staining protocols reveal the decrease of goblet cell mucus producing cells in mice challenged with DNBS, confirming the functional abnormalities on the tissue despite the lack of macroscopic or microscopic damages. Mice treated with *B. lactis* CNCM-I2494 strain recover the same goblet cell accounts than control mice pointing out a positive effect of the strain in restoring epithelial normal cell composition

and probably the mucus production. In fact, mucus production was previously shown to be affected during inflammation with intestinal dysbiosis (Fyderek et al., 2009). Several bifidobacteria strains, alone or in mixture, induce mucus production and/or are able to adhere to it (He et al., 2001; Gaudier et al., 2005). Even if the goblet cell depletion observed in DNBS challenged mice could explain the differences found in *in vivo* permeability, alterations in apical junction proteins have been also reported previously in this model (Laval et al., 2015). The apical junctions are formed by TJ and AJ proteins. Here, according to our previous results, the expression of TJ proteins measured by RT-qPCR is reduced by the DNBS intra-rectal administration in a protein-specific way (Laval et al., 2015). The treatment with *B. lactis* CNCM-I2494 strain tended to restore F11r, Occludin, E-cadherin and ZO-1 expression, showing this effect especially remarkable for claudin 4. These results are consistent with previous studies in which some lactic bacteria and bifidobacteria prevented changes in occludin, ZO-1, claudin-1, claudin-3, claudin-4, and claudin-5 proteins (Mennigen et al., 2009). Indeed, Agostini et al. (2012) showed that *B. lactis* CNCM-I2494 restored occludin and JAM-A concentrations to control levels after partial restrain stress in rat administration of fermented milk containing *Lactococcus lactis* CNCM-I1631 and two classical yogurt starters.

Changes on mucosal permeability as the ones observed in the DNBS low-dose model can be the cause or the consequence of a low immune activation. To assess the effect of *B. lactis* CNCM-I2494 strain on mucosal immunity and decipher its possible effect on host immune response, colonic cytokine levels, and spleen MLN lymphocyte populations were analyzed. In this specific context, *B. lactis* CNCN-I2494 treatment restored the mild increased IL-13, IL-2, IL-4, and INF-γ colonic values to normal. Several studies pointed out the cytokines as one of the causes of TJ protein modulation. For instance, *in vitro* test have shown a relationship between IL-13 and an increase in paracellular permeability (Prasad et al., 2005) and INF-γ or IL-4 increases have been linked to TJ protein expression alterations (Bruewer et al., 2005; Wisner et al., 2008; Suzuki et al., 2011). Therefore, the effect of *B. Lactis* CNCM-I2494 on cytokine down-regulation could be the factor which triggered permeability restoration.

Mucosal dendritic cells present antigens to the adaptative immune system which directs the polarization of naïve CD4 T cells toward different T-helper cell subsets (Th1 and Th2 among others; Zhu and Paul, 2008). Classically, hapten-mediated colon inflammation protocols trinitrobenzene sulfonic acid (TNBS and DNBS) have been associated with Th1 response (Zuo et al., 2014). Our study confirms that, even in a gut dysfunction model provoked by a low-grade inflammation, DNBS challenge increase lightly Th1 response. Although *B. lactis* CNCM-I2494 was not able to decrease Th1 response, an upper-regulation of Th2 subset has been observed counterbalancing the Th1/Th2 ratio at local level. The increased Th2 cell subset may contribute to the decreased Th1 cell subset due to the mutual antagonizing effects of both Th substets (Donato et al., 2010). Several studies have been performed to assert the role of *Bifidobacterium* strains in modulating T-cell populations, being their results strain and model dependent (Lopez et al., 2011). Our results are consistent with those of Zheng et al. (2014) who showed that one strain of *B. breve* modulates T cell polarization toward Th2 and Treg cellassociated responses *in vitro* and *in vivo i*n a murine model of DSS-induced colitis.

Our results support the hypothesis of Agostini et al. (2012) who pointed out the improvement of the intestinal barrier (epithelial cells and mucus layers) permeability as part of the beneficial effect of the fermented milk commercial product containing CNCM-I2494. In addition, here we firstly point to CNCM-I2494 strain as a possible responsible of this effect. Furthermore, the present study supports that the action mechanism of this protective effect may be mediated by improvement on apical junction proteins and goblet cell population. Finally, the modulation of the host T-cell composition by CNCM-I2494 strain may be the host pathway involved in this phenomenon.

#### AUTHOR CONTRIBUTIONS

RM, TS, JH, EV, CC, LB-H, and PL designed all the experiments. RM, SM, FC, LL, JN, and HS performed the experiments.

#### REFERENCES


RM wrote the manuscript. SM, TS, JH, and PL corrected the manuscript. All authors read and approved the final manuscript.

#### FUNDING

LL and RM receives a salary from Danone Nutricia Research in the framework of a CIFRE contract funded by ANRT and a postdoc contract, respectively. SM receives a salary from FPARIS collaborative project. EV holds a Canada Research Chair. The work was partially funded by CCFC grants to PL and EV.

# ACKNOWLEDGMENTS

We thank the histology facility of UMR 1313 GABI, 78350, Jouyen-Josas, France and the MIMA2 platform for access to the virtual slide scanner (Pannoramic SCAN, 3DHISTECH). Authors would like to thank Sylvia Le Guin and all UEAR personal for their help.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.00608

FIGURE S1 | Low-grade inflammation experimental protocol. Colitis was induced by intra-rectal administration of 100 mg/kg of DNBS in solution in 30% ethanol. Control mice (without colitis) received only 30% EtOH. The effects of DNBS are highest during the first 3 days after its administration (DNBS period). Ten days after the end of the DNBS period bacterial culture or PBS were intra-gastrically administered daily for 10 days (gavage period). Colitis was reactivated 21 days after the first DNBS injection with a second injection of 50 mg/kg of DNBS solution. Three days after reactivation mice were sacrificed. Modified from Martin et al. (2015).

FIGURE S2 | Evaluation of inflammatory status of DNBS-challenged mice. Inflammatory status assessed from (A) the recovery after reactivation (%weight change); (B) macroscopic score; (C) histological score; (D) MPO activity *n* = 24 mice per group, and (E) Lipocalin-2 levels (*n* = 8 mice per group) in the control non-inflamed group (EtOH–PBS), control inflamed group (DNBS–PBS), *B. lactis* CNCM I-2494 strain (DNBS–CNCM-I2494).


murine model of irritable bowel syndrome. *PLoS ONE* 8:e63893. doi: 10.1371/journal.pone.0063893


Zuo, L., Yuan, K. T., Yu, L., Meng, Q. H., Chung, P. C., and Yang, D. H. (2014). *Bifidobacterium infantis* attenuates colitis by regulating T cell subset responses. *World J. Gastroenterol.* 20, 18316–18329. doi: 10.3748/wjg.v20. i48.18316

**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 Martín, Laval, Chain, Miquel, Natividad, Cherbuy, Sokol, Verdu, van Hylckama Vlieg, Bermudez-Humaran, Smokvina and Langella. 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.*

# Effect of a Ropy Exopolysaccharide-Producing Bifidobacterium animalis subsp. lactis Strain Orally Administered on DSS-Induced Colitis Mice Model

Claudio Hidalgo-Cantabrana1, 2 , Francesca Algieri <sup>2</sup> , Alba Rodriguez-Nogales <sup>2</sup> , Teresa Vezza<sup>2</sup> , Pablo Martínez-Camblor 3, 4, Abelardo Margolles <sup>1</sup> , Patricia Ruas-Madiedo<sup>1</sup> \* and Julio Gálvez <sup>2</sup>

<sup>1</sup> Department of Microbiology and Biochemistry of Dairy Products, Instituto de Productos Lácteos de Asturias – Consejo Superior de Investigaciones Científicas, Villaviciosa, Spain, <sup>2</sup> CIBER-EHD, Department of Pharmacology, ibs.GRANADA, Center for Biomedical Research, University of Granada, Granada, Spain, <sup>3</sup> Geisel School of Medicine at Dartmouth, Hanover, NH, USA, <sup>4</sup> Universidad Autónoma de Chile, Santiago, Chile

Exopolysaccharide (EPS)-producing bifidobacteria, particularly Bifidobacterium animalis subsp. lactis strains, are used in the functional food industry as promising probiotics with purported beneficial effects. We used three isogenic strains of B. animalis subsp. lactis, with different EPS producing phenotypes (mucoid-ropy and non-ropy), in order to determine their capability to survive the murine gastrointestinal tract transit, as well as to evaluate their role in improving clinical outcomes in a chemically-induced colitis model. The three strains were able to survive in the intestinal tract of C57BL/6J mice during the course of the intervention study. Furthermore, the disease activity index (DAI) of the animal group treated with the ropy strain was significantly lower than of the DAI of the placebo group at the end of the treatment. However, no significant differences were found among the three strains. The analysis of several immune parameters, such as TNFα and IL-10 quantified in blood plasma and lymphocyte populations enumerated in mesenteric nodes, showed some significant variations among the four experimental animal groups. Remarkably, a higher capability of the ropy strain to increase regulatory T-cells in mesenteric lymphoid nodes was demonstrated, suggesting a higher ability of this strain to regulate inflammatory responses at mucosal level. Our data indicate that strains of B. animalis subsp. lactis producing EPS that confer a mucoid-ropy phenotype could represent promising candidates to perform further studies targeting intestinal inflammatory processes.

Keywords: Bifidobacterium, exopolysaccharide, mucoid, ropy, immune modulation, DSS-colitis, mouse model

#### Edited by:

Marco Ventura, University of Parma, Italy

#### Reviewed by:

Douwe Van Sinderen, University College Cork, Ireland Chiara Ferrario, University of Parma, Italy

\*Correspondence:

Patricia Ruas-Madiedo ruas-madiedo@ipla.csic.es

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 24 April 2016 Accepted: 23 May 2016 Published: 09 June 2016

#### Citation:

Hidalgo-Cantabrana C, Algieri F, Rodriguez-Nogales A, Vezza T, Martínez-Camblor P, Margolles A, Ruas-Madiedo P and Gálvez J (2016) Effect of a Ropy Exopolysaccharide-Producing Bifidobacterium animalis subsp. lactis Strain Orally Administered on DSS-Induced Colitis Mice Model. Front. Microbiol. 7:868. doi: 10.3389/fmicb.2016.00868

# INTRODUCTION

The human gut is one of the most densely populated ecosystems inhabited by a community of microorganisms that actively contributes to the health of the host. Microbiota establishes a symbiotic relationship with the host, but also their members must co-exist in a balanced state. Indeed, disturbances in microbiota composition (dysbiosis) and function, as well as in the microbiota-host cross-talk, are the basis of prevalent gastrointestinal diseases, and also of some extra-intestinal pathology (Tojo et al., 2014). Among others, microbiota shifts have been related to chronic inflammatory disorders, such as inflammatory bowel disease (IBD) (Huttenhower et al., 2014). Up to now, dietary interventions toward restoring the unbalanced microbiota, such as the oral administration of probiotics, have been a realistic approach for human applications (Reid et al., 2011; Collins, 2014). Hill et al. (2014) have recently reviewed the probiotic concept and they have adopted, with a minor grammatical correction, the definition "live microorganisms that, when administered in adequate amounts, confer a health benefit on the host," previously proposed by the FAO and WHO organizations (FAO/WHO. Food and Agricultural Organization of the United Nations and World Health Organization, 2001). Research on probiotics has generated a vast amount of literature, mainly related to the characterisation of the probiotic potential of Lactobacillus and Bifidobacterium, the genera most commonly used for human consumption. However, only a small number of strains have been evaluated for the prevention or treatment of gastrointestinal disorders, such as IBD, diverse diarrheas, irritable bowel syndrome (IBS) or necrotising enterocolitis (NEC) (WGO. World Gastroenterology Organisation, 2011; Sanders et al., 2013). In relation to bifidobacteria, specific strains belonging to the species Bifidobacterium breve, Bifidobacterium longum, Bifidobacterium bifidum, and Bifidobacterium animalis subsp. lactis, have been tested in human intervention studies resulting in positive effects on gastrointestinal dysfunctions (Tojo et al., 2014).

B. animalis subsp. lactis is the species that has been most successfully introduced in food formulations, since it is able to deal with technological challenges during food manufacture (Prasanna et al., 2014). However, analysis of the available genomes shows a very low genetic variability among members of this subspecies (Milani et al., 2013). One of the features shared in all of their genomes is the presence of a cluster involved in the synthesis of exopolysaccharides (EPS) which are carbohydrate polymers surrounding the cell wall (Hidalgo-Cantabrana et al., 2014b). These EPS play a relevant role for the producing strain because they are involved in protection and niche colonization, but EPS also act as intermediaries in the dialog established between bacteria and host (Hidalgo-Cantabrana et al., 2014b). It has been proposed that EPS are able to interact with receptors located in the gut epithelium (Lebeer et al., 2010b) and they could act as effector molecules eliciting different immune responses (Hidalgo-Cantabrana et al., 2012).

In recent years our group has been working on the study of biological properties of EPS synthesized by bifidobacteria. One of the B. animalis subsp. lactis strains from our collection displayed a mucoid-ropy phenotype, denoted by the formation of a long filament from the colony, and it was able to synthesize a high molecular weight (HMW)-EPS. Recently, working with three recombinant strains, we demonstrated the involvement of a mutation (C> 266 >T) in wzz gene in the occurrence of this phenotype and in the synthesis of the HMW-EPS fraction. This gene encodes a hypothetical membrane anchored protein, with a predicted large soluble domain, which is theoretically involved in the chain length determination of the polymer (Hidalgo-Cantabrana et al., 2015). The acquisition of the ropy phenotype favored the in vitro survival and adhesion of the recombinant strain to the gut environment. In turn, this strain also elicited an in vitro and ex vivo anti-inflammatory cytokine profile (Hidalgo-Cantabrana et al., 2015). Thus, in a step forward the current work aims to in vivo evaluate the probiotic potential of the ropy B. animalis subsp. lactis Balat\_1410S89L checking its capability to counteract an inflamed state of colitis chemically induced by dextran sodium sulfate (DSS) administration.

# MATERIALS AND METHODS

### Bifidobacterial Strains and Preparation of Lyophilized Bacteria in Milk

Three isogenic EPS-producing B. animalis subsp. lactis strains from the IPLA-CSIC collection were used in this study to test their probiotic capabilities in an in vivo model. Based on the type strain DSM10140, recombinant strains were obtained from a previous study (Hidalgo-Cantabrana et al., 2015): strain 1Balat\_1410 (DSM10140 lacking the gene Balat\_1410 complemented with pAM1, non ropy), strain Balat\_1410 (strain 1Balat\_1410 complemented with pAM1 + Balat\_1410, non ropy), and Balat\_1410S89L (strain 1Balat\_1410 complemented with pAM1 + Balat\_1410S89L, ropy). Glycerol stocks kept at −80◦C were plated on agar MRSC [MRS Difco (BD, Biosciences, San Diego, CA) containing 0.25% L-cysteine-HCl (Sigma-Chemical Co., St. Louis, MO)] supplemented with erythromycin (2.5 µg/ml) (MRSCE). Plates were incubated at 37◦C under anaerobic conditions (80% N2, 10% CO2, 10% H2) in an MG500 chamber (Don Whitley Scientific, West Yorkshire, UK). A single colony per strain was used to inoculate 50 ml MRSCE broth and after 20 h this culture was used to inoculate 2% (v/v) 1 L fresh medium. After 20 h incubation, cultures were centrifuged, washed once with PBS and resuspended in 100 ml tindalized 11% skimmed milk (Difco). Then, bacterial suspensions in milk were lyophilized for 48 h in the Virtis Freezemobile 12EL equipment (SP Scientific, NY). Each lyophilized bifidobacterial stock was tested for bacterial viability and enumeration by counting in agar-MRSCE and incubated under standard conditions. For oral administration, each lyophilized bifidobacteria was resuspended daily in water (at 5 × 10<sup>9</sup> cfu/ml) and 100 µl was administered per mouse.

# Animals, Experimental Designs and Sample Collection

All studies were carried out following the Directive 2010/63/EU of the European Parliament and the Council on the protection of animals used for scientific purposes (Directive 2010/63/EU, 2010). The experimental protocol was approved by the Ethical Committee of Laboratory Animals of the University of Granada (Spain) (Permit Number CEEA-2010-286). Male C57BL/6J mice (7–9 weeks old, approximately 20 g) were obtained from Janvier Labs (St Berthevin Cedex, France) and kept under conventional conditions with a standard pelleted diet and sterilized water for 1 week before beginning the experiments.

#### Bifidobacteria Survival Study

Animals (32 mice) were randomly assigned to four experimental groups with 8 mice per group: three bifidobacterial groups, each orally receiving 100 µl/day (5 × 10<sup>8</sup> cfu) of the corresponding milk-bifidobacterial suspension in water, and one placebo (control) group which received 100 µl/day of skimmed milk. Placebo and bifidobacteria suspensions were orally administered for 9 days by means of an intra-gastric cannula. Every day, the drinking water and food intake was measured for each group of mice, whereas the body weight was measured individually for each animal. Stool samples were taken, in duplicate, at group level every 2 days, including the first and last day. At the end of the intervention period (day 10), each mouse was anesthetized with isoflurane and then, the animals were sacrificed by cervical dislocation by expert and qualified persons, according to the Federation of European Laboratory Animal Science Associations (FELASA). The colon was excised, its content was collected and its length and weight was measured after washing with PBS. Stool and colon-content samples were immediately processed as described in the next section.

#### Dextran Sodium Sulfate (DSS) Induced Colitis Study

Mice (40 animals) were randomly assigned to five groups, each with 8 animals, as described in **Figure 1**. The reference (non-colitic, non-placebo/bifidobacteria fed) group received daily, through an intra-gastric cannula, 100 µl water during the 15-day experimental period. The other four groups were fed with bifidobacteria, or a placebo, and treated with DSS. During the 15 days of the experimental period these four groups received a placebo (100 µl/day milk, group 1), or different bifidobacterial strains (100 µl/day 5 × 10<sup>8</sup> cfu milk-bifidobacterial suspensions in water, for groups 2, 3, and 4). After 9 days of bifidobacteria/placebo feeding, colitis was induced in the four groups by adding DSS (3%, w/v, 36–50 kDa, MP Biomedicals, CA) in the drinking water (Mähler et al., 1998). This DSS-treatment was kept for a period of 6 days unless the application of humane end-point, as defined below, was needed. At the end of the experimental period (day 15) animals were anesthetized and a blood sample was directly extracted from the heart using heparinized tubes. The blood plasma was obtained by centrifugation (9,300xg, 4◦C, 20 min). The animals were then euthanatized in order to collect colon and colon-content, as described above, as well as mesenteric lymphoid nodes. Samples were immediately processed or stored at −20◦C until their analysis.

Drinking water and food intake was measured at group level each day, whereas body weight, stool consistency and the presence of gross blood in feces were evaluated daily and scored (**Table 1**) for each mouse during the experimental period. The disease activity index (DAI) was calculated as previously reported by Cooper et al. (1993): total score [body weight decrease + stool consistency + rectal bleeding] / 3. The ulcerative colitis (UC) disease is considered for DAI ≥ 1.5 and the humane endpoint was fixed at DAI = 3. Stool samples were taken at group level every 2 days to check viability of bifidobacteria.

#### Microbial Quantification in Fecal Samples Bifidobacterial Quantification by Counting

Stool samples, collected in both studies as previously indicated, were homogenized in PBS (0.1 g/ml) using a Heidolph (Heidolph Instruments GmbH, Schwabach, Germany) stirrer for 1 min. Then, 100 µl was used to obtain serial dilutions in PBS which were plated onto the surface of TOS propionate agar medium (Merck, Darmstadt, Germany), supplemented with lithium-mupirocin as recommended by the manufacturer, to selectively enumerate total bifidobacteria. Additionally, agar lithium-mupirocin-TOS was supplemented with 2.5 µg/ml erythromycin (TOS+Ery) which is the selective marker of the plasmid present in the three B. animalis subsp. lactis strains under study. All plates were incubated at 37◦C in an anaerobic jar using Oxoid-AnaerogenTM (Thermo Fisher Scientific Inc., Waltham, MA) sachets to generate anaerobic conditions for 72 h.

#### Bifidobacterial Identification by 16S rDNA Amplification

Several colonies were picked up from TOS+Ery plates of each of the four experimental groups belonging to the bifidobacteria survival study. The DNA from a total of 67 colonies was isolated using the GenElute Bacterial Genomic DNA kit (Sigma) according to the manufacturer's instructions, but adding a prior step of bacterial lysis with mutanolysin and lysozyme. Primer uses were: pA/pH (Edwards et al., 1989). The PCR mixture (final volume 50 µl) was: 1 µl chromosomal DNA, 0.20 µM of each primer, 200 µM dNTPs (Amersham Bioscience, Upsala, Sweden) and 2.5 U Taq DNA-polymerase (Eppendorf, Hamburg, Germany). After an initial cycle at 95◦C, the PCR amplification consisted of 30 cycles: 95◦C for 1 min, 55◦C for 50 s and 68◦C for 2 min, and ended with a final extension step of 68◦C for 10 min. Reactions were carried out in the thermal cycler UnoCycler (VWR International Eurolab S.L., Barcelona, Spain). Purification and sequencing of the amplicons were performed in Macrogen (Seoul, Korea). The free Chromas 1.45 software (Technelysium Pty Ltd., Australia) was used to process sequences which were compared to those held in the GenBank database (http://www.ncbi.nlm.nih.gov/genbank) using the BLASTn tool.

#### Gene Expression Analyses in Colonic Samples by Reverse Transcriptase qPCR

After colon excision, PBS-washed colonic tissue was sectioned in small fragments which were placed in RNAlater stabilization reagent (Qiagen GmbH, Hilden, Germany) and then stored at −80◦C until RNA extraction. Total RNA was isolated using

TABLE 1 | Disease activity index (DAI) score used to evaluate the DSS-induced colitis. DAI index was calculated as total score (body weight decrease + stool consistency + rectal bleeding) divided by 3.

a ropy EPS). Reference group (non DSS-induced colitis, non-placebo/bifidobacteria fed group) daily receiving 100 µl water.


The humane end-point was established at DAI = 3.

RNeasy Mini Kit (Qiagen) according to the manufacturer's recommendations. Purity and RNA concentration were determined with the NanoDropTM 2000 spectrophotometer. Then, 3 µg RNA was reverse transcribed to obtain cDNA using oligo(dT) primers and reagents from Promega (Promega, Southampton, UK) and the TProfessional basic thermocycler (Biometra GmbH, Goettingen Germany).

The amplification was performed in an EcoTM Real-Time PCR System (Illumina, San Diego, CA USA) using 20 ng of cDNA and specific primers (Table S1). The 2−11Ct method was used to normalize expression results (Livak and Schmittgen, 2001). The values of the house-keeping glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene were first used to normalize the values obtained for each of the genes under study. These values were additionally normalized to those obtained in the reference group, thus this group had a normalized value of 0. Finally the relative expression of each gene was calculated as 2−11Ct, thus the reference group had the relative expression value of 1.

#### Cellular Population in Mesenteric Lymphoid Nodes

The mesenteric lymphoid nodes were carefully mashed with wet slides to decrease friction and the solutions were filtered through a 70 µM cell strainer. Cells were isolated, counted and plated on anti-CD3 (clone17A2, eBioscience, San Diego, CA, USA) and anti-CD28 (clone 37.51, eBioscience)—coated plates for FcγR blocking. Then, the cells were transferred to polystyrene tubes for surface staining with anti-CD4 (PerCP-CyTM5.5, clone RM4-5 BD PharmigenTM, Franklin Lake, NU, USA), anti-CD45 (APC-eFluor <sup>R</sup> 780, clone 30-F11, eBioscience) and Viability Dye (eFluor <sup>R</sup> 660, eBioscience) for 15 min at 4 ◦C in the dark. The cells were then fixed, permeabilized with the Fixation/Permeabilization kit (eBioscience) and intracellular staining was done with anti-Foxp3 (PE, clone FJK-16s, eBioscience) for 30 min at 4◦C in the dark. Data collection was performed using a flow cytometer CANTO II (BD Biosciences San Diego, CA, USA).

#### Cytokine Levels of Blood Samples

The levels of tumor necrosis factor (TNF)-α and interleukin (IL)-10 cytokines in blood plasma were quantified using the mouse TNFα Quantikine and mouse IL-10 Quantikine ELISA kits (R&D Systems, Abingdon, UK) following the manufacturer's instructions.

#### Statistical Analyses

All considered variables were distributed normally (Kolmogorov-Smirnov test was used to check normality); thus, they are described as mean ± standard deviation. Robust Student-Welch or ANOVA tests were used in order to check the equality among groups for all variables, and post-hoc T3 de Dunnet or LSD tests were used to compare each two groups. Standard linear regression models were employed to compute the size of the effect (95% confidence interval) of the different treatments on the DAI; to do this, crude data were adjusted by the initial conditions (DAI score the first day of DSS-induced colitis) and the placebo group was used as reference. Differences below 0.05 were considered statistically significant. All analyses were made with the free statistical software R (www.r-project.org).

# RESULTS AND DISCUSSION

# B. animalis subsp. lactis Strains Are Able to Colonize the Murine Gut

Orally administered probiotics should survive the upper gastrointestinal tract (GIT) transit and reach the colon alive, where they should persist for a certain time in order to be able to exert their beneficial effect. Therefore, an animal experimentation procedure was done to assess the viability of the three B. animalis subsp. lactis strains under study aiming to check whether the presence of the ropy HMW-EPS surrounding Balat\_1410S89L strain could exert a protective role. Additionally, although B. animalis has QPS (qualified presumption of safety) status (EFSA BIOHAZ Panel, 2013), we evaluated the lack of toxicity of our strains, at this oral-dose administered to the animals, by measuring animal behavior and wellbeing parameters. In this regard, no differences (p > 0.05) were found in water and food intake, body weight, stool consistency or colon anatomy between the placebo (skimmed milk-fed) and each of the bifidobacteriafed groups (**Table 2**). These results support the fact that the three bifidobacterial strains used in this study had no side effects at the concentration level administered to this murine model.

The bifidobacterial population in the fecal samples was enumerated using agar-TOS (**Figure 2A**) which is a selective medium for bifidobacteria growing in milk products. After 2 days of administration, the bifidobacterial levels increased in all groups about 1 log unit, including the placebo receiving skimmed milk. However, during prolonged treatment the population of bifidobacteria in the placebo group tended to decline to reach the initial values, whereas counts in the three bifidobacteria-fed groups remained stable. Similarly, counts obtained in TOS+Ery, which is the medium containing the selective marker erythromycin of the plasmid present in the three strains under study, were significantly higher (p < 0.05) in the three bifidobacterial-fed groups with respect to the placebo (**Figure 2B**). Levels of erythromycin-resistant bacteria in the microbiota of placebo-fed mice were around 4 log units and remained unaltered for the duration of the experimentation time. After 2 days of administration, levels of erythromycin-resistant bifidobacteria increased (on average) 0.7 log cfu/g in the mice receiving bifidobacteria; the highest count increase (1.7 log cfu/g) was reached after 4 days, remaining at similar values until the end of the experimentation period. In general, no statistical differences among the three bifidobacteria-fed groups were found and, additionally, the colonies chosen to partially sequence the gene 16S rDNA were identified as members of Bifidobacterium genus; those isolated from Balat\_1410S89L fed animals displayed a ropy phenotype as well (see Table S2). Altogether, these results show that our bifidobacterial strains were able to survive as members of the intestinal microbiota of mice for at least 9 days, following a daily intake of 5 × 10<sup>8</sup> cfu. The increase in total bifidobacteria observed in this study could be related to the ability of EPS-producing B. animalis subsp. lactis to raise the number of other bifidobacterial species in the intestinal microbiota of rodents as previously described (Salazar et al., 2011). This effect could be related to the ability of members from this microbial community to use EPS as fermentable substrates (Salazar et al., 2009). However, the increase of erythromycin resistant members in feces is indicative of a higher survival rate in the gut of the specific strains used in this work. The in vivo performance of the three strains is rather similar, since there was no statistical difference in the number of bifidobacteria present in the fecal samples among the three groups of mice. Thus, it seems that the presence of HMW-EPS on the surface of the ropy Balat\_1410S89L strain does not confer an additional protection of survival in mice gut. Fanning et al. (2012) found that fecal counts (made in Reinforced Clostridium agar with 50 mg/L mupirocin) of mice fed with EPS<sup>+</sup> B. breve UCC2003 were significantly higher than those obtained with two non-EPS producing mutant strains. Thus, these authors conclude that the presence of an EPS on the surface of the wild type B. breve was involved in its higher survival rate in the mice gut. On the contrary, in vivo studies carried out with different EPS-producing Lactobacillus johnsonii strains showed that mutants with deletion of genes from the eps cluster had an increased residence time in the animal's gut (Denou et al., 2008; Horn et al., 2013). It has also been shown that the loss of EPS production in Lactobacillus rhamnosus GG improves the in vitro adhesion of the strain to Caco2 cells (Lebeer et al., 2009); however, the presence of the EPS surrounding the wild type GG strain improves its in vivo performance allowing the strain to persist for longer periods because the EPS protects against the antimicrobial peptide LL-37 (Lebeer et al., 2010a). The intrinsic characteristics of EPS, such as composition and size, seem to promote different effects on bacterial survival under gut challenges. In some cases, polymers could prevent the exposure of molecules from the bacterial envelope involved in adhesion, but EPS can also act as a shield against host immune defense (Hidalgo-Cantabrana et al., 2012).

### The Ropy B. animalis subsp. lactis Balat\_1410S89L Reduces DSS-Induced Damage

As mentioned above, in vitro and ex vivo tests (using human tissues) showed different immune modulation capabilities of the three isogenic bifidobacteria, the ropy strain Balat\_1410S89L having an anti-inflammatory profile (Hidalgo-Cantabrana et al., 2015). To expand on these observations, we wanted to address whether this predicted anti-inflammatory ability is effective in vivo, modifying the immune response in a model of gut inflammation such as mice with DSS-induced colitis. This is a well-established animal model of acute mucosal inflammation that has been used for several decades (Wirtz et al., 2007) and it has been validated as a model for the translation of mice data to humans (Melgar et al., 2008).

In our experimental procedure, four groups of mice were pretreated with an oral administration of bifidobacteria, or placebo, for 9 days before the administration of DSS in drinking water, which was continued for an additional 6 days (**Figure 1**). As was expected, in accordance with results shown in the previous section, the pre-feeding for 9 days in the four groups did not modify the wellbeing and behavior parameters of the mice. This is also consistent with the absence of significant variations in the disease activity index (DAI) score of these groups (values


TABLE 2 | Mean and standard deviation of general parameters measured in four experimental groups of mice to evaluate the absence of toxicity of the bifidobacterial strains administered for 10 days at dose of 5 × 10<sup>8</sup> cfu per mouse and day.

No significant statistical differences (p > 0.05) were found among the four experimental groups after one-way ANOVA study.

<sup>a</sup>Values obtained for the whole mice group for each day of experimental procedure.

<sup>b</sup>body weight at day 10 subtracting the body weight measured at day 0 (27.71 ± 1.11) before the experimental procedure.

<sup>c</sup>colon weight and length determined after 10 days of experimental procedure.

(p < 0.05) and differences among the three bifidobacterial groups are denoted with different letters (p < 0.05).

lower than 0.5, data not shown). The DAI score was kept close to 0 in the non-colitic, non-placebo/ bifidobacteria-fed (reference) group during all experimental procedures. In the four colitic groups, after the second day (day 11) of DSS treatment, the DAI values increased continuously (**Figure 3A**). The maximum score was obtained at the end of the experimental procedure, after 6 days of DSS colitic induction. In general, no statistical differences were detected on each day of treatment among the placebo and the three bifidobacteria-fed groups during DSStreatment; the exception was the group fed with the strain Balat\_1410S89L, which showed a significantly (p < 0.05) lower DAI score than the placebo group at day 15. These results were confirmed when DAI data were adjusted by means of linear regression models taking into consideration the DAI of placebo group each day of DSS-induction (**Figure 3B**). This model allowed us to determine the effect of the treatments on the DAI with respect to the placebo (value = 0). Thus, negative values of the adjusted DAI indicate a reduction in the severity of the disease in comparison with the placebo group. The highest reduction in adjusted DAI obtained for B. animalis subsp. lactis Balat\_1410S89L fed animals at the end of the treatment was mainly associated with a reduction in diarrhea and rectal bleeding. This led to a 44% reduction in the severity of the disease with respect to the placebo (p < 0.007), whereas, in the other two bifidobacteria-fed animals this reduction was not statistically significant, although the reduction of the illness was 23.2% for the strain 1Balat\_1410 and 16.7 % for the strain Balat\_1410. In spite of the remarkable differences among the three strain-fed groups (disease severity in Balat\_1410S89L was reduced by almost two and three-fold compared to 1Balat\_1410 and Balat\_1410, respectively), these differences were not deemed statistically significant. Therefore, we cannot ascertain that the protective effect against DSS-induced colitis exerted by strain B. animalis subsp. lactis Balat\_1410S89L is attributed to the HMW-EPS surrounding the bacteria. Some authors have reported the capability of probiotic strains to reduce the DAI index of animals under DSS-induced colitis. However, a mixture of several bacteria, or combinations of bacteria-bioactive compounds, are often tested, thus the positive effect cannot be assigned to a single bacterium (Dai et al., 2013). Only a few reports have shown the efficacy of single bifidobacterial strains (Hayes et al., 2014; Zheng et al., 2016).

data were compared with placebo during the 5 days of DSS-treatment and statistical differences (p < 0.05) are marked with an asterisk (B). Reference group ( ); DSS-induced colitis groups: placebo ( ), feed B. animalis subsp. lactis 1Balat\_1410 ( ), feed B. animalis subsp. lactis Balat\_1410 ( ), feed B. animalis

#### Different B. animalis subsp. lactis Strains Elicit Specific Host Responses

subsp. lactis Balat\_1410S89L ( ).

In order to understand the mechanism(s) involved in the attenuation of the DSS-colitis by the strain Balat\_1410S89L several immune parameters, targeting adaptive as well as innate responses, were measured in different tissues. In our study, the expression of several genes related to the reinforcement of the intestinal epithelium barrier, such as those coding for chemokines, proteins of tight junctions and mucins, showed no statistical variations (p > 0.05) among the four groups of animals (**Table 3**). Only the non-ropy EPS-producing strain Balat\_1410 was able to slightly induce (p < 0.05) the expression of Mcp1, a chemokine able to attract monocytes, as well as the secretory protein TFF3, which is synthetized by goblet cells and acts to stabilize the mucus layer. Thus, it seems that, based on these results of gene expression, our strains were not able to modify the epithelial barrier in a remarkable way. However, it was reported that probiotics are able to reinforce the intestinal barrier function by multiple mechanisms (Lebeer et al., 2010a; Rao and Samak, 2013). The efficacy of Bifidobacterium infantis BB-02 to reduce intestinal injury in NEC was due to its ability to keep claudin-4 and occludin located in the tight-junction of bifidobacteriafed mice (Bergmann et al., 2013). It was also demonstrated that bioactive factors released by B. infantis (one of the strains present in the VSL#3 mixture) were also able to reduce gut permeability by acting upon tight junction proteins (Ewaschuk et al., 2008). Similarly B. bifidum OLB6378 also favored the location of tight-junction and adherent-junction proteins, also being able to induce the synthesis of mucin-3, TFF3, and IL-6 (Khailova et al., 2009). These works, among others, show that bifidobacteria are able to improve the gut barrier function in a strain dependent manner. Regarding expression of cytokine genes in colonic tissue, statistical differences (p < 0.05) were detected only for pro-inflammatory IL-1β and IFNγ genes and the anti-inflammatory IL-10 (**Table 3**); in general, the two first genes were expressed at a lower rate in the placebo group with respect to the three bifidobacteria-fed groups, whereas IL-10 gene had significantly lower expression rate only for the ropy strain Balat\_1410S89L. Thus, it seems that this strain induced the expression of genes that could lead to a mild pro-inflammatory response, since the ratio TNFα/IL-10 was significantly higher than that obtained for the other three groups (data not shown). A similar pro-Th1 profile was obtained with wild-type EPSproducing B. animalis subsp. lactis strains (one of them also producing a ropy HMW-EPS) co-cultivated in vitro with PBMC and GALT (gut associated lymphoid tissue) cells isolated from naïve rats (Hidalgo-Cantabrana et al., 2014a). Indeed, in the current study the concentration of these cytokines measured in blood plasma (**Figure 4A**) gives TNFα/ IL-10 ratio values of 1.69 ± 0.97 for placebo, 4.32 ± 2.15 for strain 1Balat\_1410, 2.77 ± 1.85 for strain Balat\_1410, and 10.28 ± 9.15 for strain Balat\_1410S89L. The ratio obtained for the animals fed with the ropy strain was significantly (p < 0.05) higher than that obtained for the other three groups of mice. However, it is worth noting that none of the three strains, or the placebo, was able to induce a strong immune systemic response since the values of both cytokines were near (or lower) than value 1, which is the value of the reference animal group. Thus, this suggests that our strains, or the skimmed milk (placebo) used as a protectant for bifidobacteria delivery, did not contribute to increase the inflammatory process induced by DSS.

Immune cells isolated from the lymphoid nodes associated with the gastrointestinal tract were quantified by flow cytrometry and significant (p < 0.05) variations among the four experimental groups were detected (**Figure 4B**). Mice orally receiving the ropy Balat\_1410S89L or the non-ropy Balat\_1410 strains presented a significant reduction in the population of CD4+/CD45<sup>+</sup> cells, which are the markers for helper T-lymphocytes. The number of this population was even slightly lower than that of the reference group (value 1). In addition, the group receiving the strain Balat\_1410S89L was the only one significantly different


#### TABLE 3 | Relative expression of genes related to immunity of intestinal mucosa in colonic tissue.

Values in each DSS-colitic group were referred to the average of those obtained in the control (no-colitic, value = 1) one. P-values from one-way ANOVA tests carried out among four groups; those values within the same row that do not share a common letter are statistically different according to the mean comparison LSD (p < 0.05).

a IL-, interleukin-; TNFα, tumor necrosis factor α; TGF-β, transforming growth factor β; IFNγ , interferon γ ; MUC-, mucin-; TFF-3, trefoil factor 3; ZO-1, zonula occludens 1; MMP-9, matrix metallopeptidase 9; iNOS, inducible nitric oxide synthase; ICAM-1, intercellular adhesion molecule.

<sup>b</sup>The statistical differences among the four bifidobacterial groups are denoted with different letters (p < 0.05).

Bold values underlines statistical differences according to one-way ANOVA test.

FIGURE 4 | Mean values of relative TNFα and IL-10 cytokines measured in blood samples (A) and relative number of CD4+CD25<sup>+</sup> (T helper cells) and FOXP<sup>3</sup> (T regulatory cells) quantified in mesenteric lymphoid nodes (B). Samples were collected after 6-days of DSS-treatment. Values of each DSS-colitic animal group were referred to as the average of those obtained in the reference group (value = 1). Placebo group ( ), feed B. animalis subsp. lactis 1Balat\_1410 ( ), feed B. animalis subsp. lactis Balat\_1410 ( ), feed B. animalis subsp. lactis Balat\_1410S89L ( ). Within each parameter: statistical differences among the four groups are denoted with different letters (p < 0.05).

from the placebo regarding the number of cells expressing Foxp3+, which acts as a regulatory factor for differentiation and function of regulatory T-lymphocytes. Besides, values of Foxp3<sup>+</sup> cells in this group of animals, which also showed a reduced severity in DAI score, were slightly higher than the value of the animal reference group. Thus we suggest that the oral administration of the ropy strain could favor the recruitment of Treg cells to the intestinal mucosa, which could lead to a decrease in the severity of the damage caused by DSS. Jeon et al. (2012) showed in vivo that B. breve Yakult strain was able to ameliorate T cell-dependent colitis in mice through the induction of Treg cells toward producing IL-10. Similar results were obtained with a mixture of five probiotic strains and a mice model mimicking human Crohn's disease (Kwon et al., 2010). These authors also conclude that the efficacy of probiotics was due to an enrichment of CD4<sup>+</sup> Foxp3<sup>+</sup> Treg in the inflamed area. In recent years our group has been working with two models of isogenic, ropy and non-ropy EPS-producing B. animalis subsp. lactis strains: wild-types and the recombinant strains used in the current work. The results regarding the immune response that both types of strains could elicit are consistent when immune cells from the same origin are used. That is, all types of strains are able to induce a mild pro-inflammatory (Th1) response when a murine (rats or mice) model is used, either by in vitro (Hidalgo-Cantabrana et al., 2014a) or in vivo (the current work) approaches. However, the same EPS-producing strains tested in vitro or ex vivo, with immune cells and colonic tissue from human origin, had a reduced capability to elicit an immune response, or displayed an anti-inflammatory cytokine profile (López et al., 2012; Hidalgo-Cantabrana et al., 2015). These results underline the strong influence of the biological model used to test immune capability of strains with probiotic potential. Variations obtained with both models could be related to the differences in composition and function of the microbiota inhabiting the human and mouse intestines (Wos-Oxley et al., 2012).

Finally, due to the absence of statistical differences among the three strains we cannot conclude without doubt that the protective effect of DAI score promoted by strain Balat\_1410S89L was due to the synthesis of the HWM-EPS. However, we can hypothesize that the ropy HMW-EPS (produced in higher amounts by this strain) could act as a physical barrier, like a biofilm, protecting from chemical insults and, consequently, reducing the severity of the mucosa damage. In this regard, we have previously shown that the EPS purified from the wild ropy strain B. animalis subsp. lactis A1dOxR was able to in vitro counteract the cytotoxic effect of bacterial toxins upon humans (Caco2 monolayers) or animals (rabbit erythrocyte) cells (Ruas-Madiedo et al., 2010). Synthesis of EPS by bifidobacteria in intestinal environmental conditions has not been proved to date; however, production of EPS by commensal Bacteroides fragilis in the gut of mice has recently been proved (Geva-Zatorsky et al., 2015). Thus, this fact would support our hypothesis that the HMW-EPS from strain B. animalis subsp. lactis Balat\_1410S89L forms an EPS-biofilm protecting, to a certain degree, the intestinal epithelium from DSS-induced damage.

# CONCLUSION

In this work we have shown that the ropy strain B. animalis subsp. lactis Balat\_1410S89L was able to protect mice from injury caused by DSS. To date, we have not found modifications in the expression of genes related to the reinforcement of the intestinal barrier that could explain the attenuation of chemically-induced damage; the most relevant immune parameter that could be related to this fact is the high capability of strain Balat\_1410S89L to induce Treg cells in mesenteric lymphoid nodes which, in turn, could lead to a localized reduction of the inflammation at mucosal level. In spite of the only difference among the three isogenic recombinant strains tested in this study being the production of a ropy HMW-EPS, we cannot without doubt attribute the capability to reduce the DSS-induced damage to the polymer produced by Balat\_1410S89L .

There is a lack of consistency in the immune modulation capability of the ropy HMW-EPS-producing B. animalis subsp. lactis strains due to the high influence of the biological model used. However, the ability of the strain Balat\_1410S89L to survive in the mice gut and the absence of adverse effects on the animals, together with previous in vitro and ex vivo evidence upon human cells, makes it feasible that the ropy (wild-type) B. animalis subsp. lactis could be a good candidate to check its anti-inflammatory ability in patients suffering from intestinal inflammation.

# AUTHOR CONTRIBUTIONS

JG, AM, and PR contributed with the conception, experimental design and results interpretation of this study. CH carried out all experiments, some of them performed with the collaboration of FA, AR, and TV. PM was in charge of the statistical analyses. PR was in charge of writing the drafted manuscript. All authors performed a critical revision of the manuscript and approved the final version.

# ACKNOWLEDGMENTS

This work was partially financed by FEDER European Union funds through the projects AGL2012-33278 and SAF2011-29648 of the Spanish Ministry of Economy and Competitiveness (MINECO), as well as to the projects AGR-6826 and CTS-164 of the "Junta de Andalucía." CH acknowledges his FPI fellowship to MINECO and FA acknowledges her fellowship to Junta de Andalucía. The CIBEREHD is funded by the "Instituto de Salud Carlos III."

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.00868

# REFERENCES


Bifidobacterium strains abrogate in vitro the cytotoxic effect of bacterial toxins on eukaryotic cells. J. Appl. Microbiol. 109, 2079–2086. doi: 10.1111/j.1365- 2672.2010.04839.x


**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 Hidalgo-Cantabrana, Algieri, Rodriguez-Nogales, Vezza, Martínez-Camblor, Margolles, Ruas-Madiedo and Gálvez. 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.

# Effect of Bifidobacterium upon Clostridium difficile Growth and Toxicity When Co-cultured in Different Prebiotic Substrates

L. Valdés-Varela, Ana M. Hernández-Barranco, Patricia Ruas-Madiedo and Miguel Gueimonde\*

Microbiology and Biochemistry of Dairy Products, Probiotics and Prebiotics, Instituto de Productos Lácteos de Asturias–Consejo Superior de Investigaciones Científicas, Villaviciosa, Spain

The intestinal overgrowth of Clostridium difficile, often after disturbance of the gut microbiota by antibiotic treatment, leads to C. difficile infection (CDI) which manifestation ranges from mild diarrhea to life-threatening conditions. The increasing CDI incidence, not only in compromised subjects but also in traditionally considered low-risk populations, together with the frequent relapses of the disease, has attracted the interest for prevention/therapeutic options. Among these, probiotics, prebiotics, or synbiotics constitute a promising approach. In this study we determined the potential of selected Bifidobacterium strains for the inhibition of C. difficile growth and toxicity in different carbon sources. We conducted co-cultures of the toxigenic strain C. difficile LMG21717 with four Bifidobacterium strains (Bifidobacterium longum IPLA20022, Bifidobacterium breve IPLA20006, Bifidobacterium bifidum IPLA20015, and Bifidobacterium animalis subsp. lactis Bb12) in the presence of various prebiotic substrates (Inulin, Synergy, and Actilight) or glucose, and compared the results with those obtained for the corresponding mono-cultures. C. difficile and bifidobacteria levels were quantified by qPCR; the pH and the production of short chain fatty acids was also determined. Moreover, supernatants of the cultures were collected to evaluate their toxicity using a recently developed model. Results showed that co-culture with B. longum IPLA20022 and B. breve IPLA20006 in the presence of short-chain fructooligosaccharides, but not of Inulin, as carbon source significantly reduced the growth of the pathogen. With the sole exception of B. animalis Bb12, whose growth was enhanced, the presence of C. difficile did not show major effects upon the growth of the bifidobacteria. In accordance with the growth data, B. longum and B. breve were the strains showing higher reduction in the toxicity of the co-culture supernatants.

Keywords: probiotics, prebiotics, inhibition, Clostridium difficile, Bifidobacterium, toxin, HT29, RTCA

# INTRODUCTION

Clostridium difficile is often present in the intestinal microbiota of both infants and adults, where it may be found in about 70 and 17% of the subjects, respectively (Ozaki et al., 2004; Jangi and Lamont, 2010). However, this microorganism is also the main causative agent of antibiotic associated diarrhea in nosocomial environments (Leffler and Lamont, 2015). The epidemiology of

#### Edited by:

David Berry, University of Vienna, Austria

#### Reviewed by:

Amparo Latorre, University of Valencia, Spain Bärbel Stecher, Ludwig Maximilians Universität München, Germany

> \*Correspondence: Miguel Gueimonde mgueimonde@ipla.csic.es

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 16 February 2016 Accepted: 03 May 2016 Published: 18 May 2016

#### Citation:

Valdés-Varela L, Hernández-Barranco AM, Ruas-Madiedo P and Gueimonde M (2016) Effect of Bifidobacterium upon Clostridium difficile Growth and Toxicity When Co-cultured in Different Prebiotic Substrates. Front. Microbiol. 7:738. doi: 10.3389/fmicb.2016.00738 C. difficile infection (CDI) is changing, with an increasing occurrence in populations traditionally considered of low-risk (Carter et al., 2012), likely due to the appearance of hipervirulent strains (Rupnik et al., 2009; Yakob et al., 2015). CDI is treated with antibiotics but a high rate of recurrence is present. In this context, new therapeutic alternatives for treating or preventing CDI are being continuously explored, among them the inhibition of C. difficile growth by the use of probiotics or prebiotics has been tested (Ambalam et al., 2015; Auclair et al., 2015; Forssten et al., 2015).

In general, probiotics and prebiotics have been proposed as biotherapeutic agents to prevent the dysbiosis caused by antibiotics or infections, and to help the microbiota restoration after it (Reid et al., 2011). The development of food products targeting at the inhibition of C. difficile constitutes an interesting approach in the context of the marketing of products bearing health claims. Reducing the intestinal levels of specific pathogens, such as C. difficile, has been considered by the European Food Safety Authority (EFSA) as a beneficial physiological effect [(EFSA Panel on Dietetic Products, Nutrition and Allergies (NDA), 2011)]. Therefore, such an effect would constitute an opportunity for the development of food products bearing a health claim in the area of gastrointestinal health.

To date, different probiotic strains and prebiotic substrates have been reported to increase colonization resistance against C. difficile (Hopkins and Macfarlane, 2003; Kondepudi et al., 2014; Auclair et al., 2015; Forssten et al., 2015). In addition to their microbiota-modulatory properties, probiotics have been found to protect against infections by other mechanisms, such as production of antimicrobial compounds or competition by adhesion sites or nutrients (Servin, 2004). The ability of certain probiotics, mainly bifidobacteria and lactobacilli, to inhibit in vitro the adhesion of C. difficile to intestinal epithelial cells or intestinal mucus is well established (Collado et al., 2005; Banerjee et al., 2009). Similarly, the ability to produce antimicrobials inhibiting the growth of C. difficile in vitro has been repeatedly reported (Lee et al., 2013; Schoster et al., 2013). However, other potential targets of probiotics and prebiotics on CDI, such as their impact on toxin production by the pathogen, and/or toxin activity, have been explored to a lesser extent and have not attracted attention until recently (Kondepudi et al., 2014; Yun et al., 2014; Andersen et al., 2015). Ambalam et al. (2015) recently reported the ability of cell-free supernatants from some Lactobacillus strains, and a probiotic mix, to inhibit the growth of C. difficile strains in variable way depending on the carbon source used. Moreover, the authors observed a reduction of toxin titers in those C. difficile cultures with inhibitory cell-free supernatants added. Moreover, we have demonstrated that incubation of toxigenic C. difficile cell-free culture supernatants with specific bifidobacterial strains reduces the cytotoxic effect upon human epithelial intestinal cells (Valdés et al., 2016). However, the influence of prebiotic substrates upon C. difficile growth and toxicity when co-cultured with bifidobacteria remains largely unknown.

In this context the aim of this study was to evaluate in vitro the potential of four bifidobacterial strains for inhibiting the growth of C. difficile when co-cultured with different prebiotics as carbon source. Moreover, the effect of the strains and prebiotics on the toxicity of the co-culture supernatants upon human intestinal epithelial cells (HT29) was also determined.

# MATERIALS AND METHODS

#### Bacterial Strains and Culture Conditions

The widely used probiotic strain Bifidobacterium animalis subsp. lactis Bb12 and three strains of bifidobacteria from IPLA culture collection, two of them isolated from infant's feces (Bifidobacterium longum IPLA20022 and Bifidobacterium bifidum IPLA20015) (Solís et al., 2010) and the other one from breast-milk (Bifidobacterium breve IPLA20006) (Arboleya et al., 2011), were used. These last three strains were selected based on the good ability to reduce toxicity of C. difficile supernatants (Valdés et al., 2016). With regard to C. difficile we used the strain LMG21717, known to produce TcdA toxin and also, although at lower quantities, TcdB. This strain belongs to ribotype 001, which is one of the most common ones found in Europe (Martin et al., 2016). The Bifidobacterium strains were routinely grown in MRS (Biokar Diagnostics, Beauvois, France) supplemented with 0.25% L-cysteine (Sigma-Chemical Co., St. Louis, MO, USA) in an anaerobic chamber MG500 (Don Whitley Scientific, Yorkshire, UK) and C. difficile was grown in Reinforced Clostridial Medium (RCM, Oxoid, Thermo Fisher Scientific Inc., Waltham, MA) in Hungate tubes as previously described (Valdés et al., 2016). Overnight cultures (18 h) of the bifidobacterial strains and 13 h-old cultures of C. difficile were used to inoculate the batch culture fermentations.

For the batch mono- and co-culture fermentations a defined medium with the following composition was used: proteose peptone (10 g/L) (BD-Difco, New Jersey, EE.UU.), beef extract (10 g/L) (BD-Difco), yeast extract (5 g/L) (BD-Difco), polysorbate 80 (1 mL/L) (Sigma), ammonium citrate (2 g/L) (Sigma), sodium acetate (5 g/mL) (Sigma), magnesium sulfate (0.2 g/L) (Probus, Barcelona, Spain), manganese sulfate (0.056 g/L) (Panreac, Barcelona, Spain), and dipotassium phosphate (2 g/L) (Merck, New Jersey, EE.UU). Pairwise combinations of the C. difficile strain with the different Bifidobacterium strains, as well as the corresponding monocultures, were performed in the medium described above with a 2% (w/v) of different commercial prebiotic substrates added [Synergy 1 (Beneo-Orafti, Barcelona, Spain), Inulin (Sigma) and Actilight (Beghin Meiji and Tereos Syral, Marckolsheim, France)], glucose or without adding any carbon source (used as control). Each media was distributed into Hungate tubes which were inoculated with different Bifidobacterium strains at a final level of about 10<sup>5</sup> CFU/ml in case of B. longum/B. breve and 10<sup>4</sup> CFU/ml in case of B. bifidum/B. animalis, with C. difficile strain at final level of 10<sup>6</sup> CFU/ml or with both of them, in the case of the co-culture. The bifidobacteria were inoculated at a different level depending on the strain with the aim of allowing a balanced growth of both microorganisms (bifidobacteria and clostridia). The appropriate inoculum size was determined in previous experiments (data not shown).

Co-cultures, and the corresponding mono-cultures, in different carbon sources were carried out in triplicate under anaerobic conditions at 37◦C for 24 h. Samples were taken at 0 and 24 h for bacterial growth assessment by quantitative PCR (qPCR), quantification of SCFA by Gas Chromatography (GC), pH measurements (pH meter Basic 20+, Crison Instruments S.A., Barcelona, Spain), and toxigenicity determination. One milliliter of each mono-culture or co-culture was centrifuged (16,000 × g for 10 min), and pellets and supernatants were collected. For toxigenic activity upon HT29 cells, the pH of 0.7 ml cell-free supernatant from each batch culture was adjusted to 7.55 ± 0.05 with 1 and 0.1 N NaOH. All supernatants and pellets were immediately frozen at −80◦C until use.

#### Quantification of Bacterial Growth by qPCR

DNA was extracted from pellets of batch cultures using the GenElute Bacterial Genomic DNA Kit (Sigma) and kept at −80◦C until analyzed. The levels of C. difficile and bifidobacteria in the cultures were determined as DNA copies per ml by qPCR using previously described primers and conditions (Arboleya et al., 2012). Reactions were performed on MicroAmp optical plates sealed with MicroAmp optical caps (Applied Biosystems, Foster City, CA, USA) with a 7500 Fast Real-Time PCR System (Applied Biosystems) using SYBR Green PCR Master Mix (Applied Biosystems). One microlitre of template DNA was used in the 25 mL PCR mixture. Standard curves were made with pure cultures of B. longum NCIMB8809 and C. difficile LMG 21717. In all cultures the levels of the microorganisms were above the corresponding detection limit of the technique (1 × 10<sup>3</sup> and 3 × 10<sup>3</sup> for bifidobacteria and C. difficile, respectively). Samples were analyzed by duplicate in at least two independent PCR runs.

#### Determination of the Production of Short Chain Fatty Acids by GC-MS

Cell-free supernatant (0.1 mL) from each batch culture was mixed with 1 ml methanol, 0.1 ml internal standard solution (2-ethylbutyric 1.05 mg/ml), and 0.1 ml 20% formic acid. This mixture was centrifuged and the supernatant obtained was used for quantification of SCFA by GC in a system composed of a 6890NGC injection module (Agilent Technologies Inc., Palo Alto, Ca, USA) connected to a flame injection detector (FID) and a mass spectrometry (MS) 5973N detector (Agilent) as described previously (Salazar et al., 2011).

## Monitoring the Cytotoxic Effect of the Culture Supernatants upon Intestinal Epithelial Cells

The intestinal cell line HT29 (ECACC 91072201) was purchased from the "European Collection of Cell Cultures" (Salisbury, UK) and stored under liquid N2. McCoy's Medium (MM) supplemented with 10% fetal serum bovine, 3 mM L-glutamine and a mixture of antibiotics (50 µg/ml streptomycin-penicillin, 50 µg/ml gentamicin, and 1.25 µg/ml amphotericin B) was used for HT29 cultivation. All media and reagents were purchased from Sigma-Aldrich. Maintenance of the cell line, between passages 145 and 149, was performed under standard conditions at 37◦C 5% CO<sup>2</sup> atmosphere, in a CO2-Series Shel-Lab incubator (Sheldon Manufacturing Inc., OR, USA). The experimental procedures were carried out with the cell passage 149.

We used an RTCA (real time cell analyser) xCelligence (ACEA Bioscience Inc., San Diego, CA) system, introduced in a Heracell-240 Incubator (Thermo Electron LDD GmbH, Langenselbold, Germany) set at 37◦C with 5% CO<sup>2</sup> atmosphere, to monitor HT29 cells behavior. A method previously described, allowing the assessment of the damage caused by C. difficile supernatants, was used (Valdés et al., 2015). This method is based in the realtime monitoring of the cell index (CI). This CI is an arbitrary unit that measures the impedance, in gold-microelectrodes coating the surface of E-plates, which changes as consequence of the HT29 cells attachment and growth.

In short, 16-well E-plates were seed with 2 × 10<sup>5</sup> HT29 cells (in 100 µl), hold in the RTCA equipment, incubated for 22 h to ensure the formation of a monolayer (confluent state) and the CI was monitored (recording signal every 15 min). After this incubation the medium was removed from the wells and the methodology followed was slightly different depending on the experiment. To determine the effect of the carbon source on the toxicity of C. difficile, 200 µL of MM containing different concentrations (from 0.63 to 40%, v/v) of cell-free neutralizedsupernatants from C. difficile mono-cultures were added to the wells. EC50 values (concentration at which half of the maximum damage was detected) for the cultures, in the different carbon sources tested, were then calculated as previously described (Valdés et al., 2015). To determine the effect of bifidobacteria on the toxigenic capability of C. difficile in the different carbon sources, 200 µL of MM containing a 5% (v/v) of the neutralized supernatant from each mono- and co-culture were added to the wells. Additionally, wells filled with 200 µl of MM (non-cytotoxic control) were included in each experiment. Then, monitoring continued (recording signal every 10 min) up to 20 h under standard incubation conditions. The data analyses were carried out through RTCA software 1.2.1 (ACEA Bioscience). The CI values were normalized as previously described (Valdés et al., 2015) by dividing the CI at every point by the CI at time zero (the time of the supernatant addition, thus making the CI equal to 1 at this time) and then referred to the normalized CI of the control sample (MM) (the normalized-CI of the control sample is then the "0 line" shown in figures).

Toxin A concentration in the supernatant of C. difficile monocultures in different carbon sources was determined by ELISA test (tgcBIOMICS GmbH, Bingen, Germany).

#### Statistical Analysis

To asses differences among carbon sources or between mono- and co-cultures, one-way ANOVAs followed by SNK (Student-Newman-Keuls, p < 0.05) mean comparison test were performed. The statistical package IBM SPSS Statistics for Window Version 22.0 (IBM Corp., Armonk NY) was used to carry out these analyses.

#### RESULTS AND DISCUSSION

## Inhibition of C. difficile Growth When Co-cultured with Bifidobacterium Strains in Different Carbon Sources

There is a great scientific interest on the development of interventions for preventing or treating CDI, including vaccines (Senoh et al., 2015), antimicrobials (Gebhart et al., 2015; Vickers et al., 2015), anti-toxin antibodies (Yang et al., 2015), or genetically engineered bacteria producing them (Andersen et al., 2015), among others. Fecal transplants have demonstrated a high efficacy to treat recurrent CDI (Lee et al., 2016), underlining the importance of the gut microbiota in this disease. Probiotics and prebiotics constitute another interesting option although differences among strains and substrates seem to exist (Allen et al., 2013).

In our study the mono-culture of the Bifidobacterium strains (**Figure 1**) in different substrates (dark colored bars) showed that all the strains grew well in glucose. In agreement with previous reports (Rossi et al., 2005; Kondepudi et al., 2012), the strains showed the ability to grow in short-chain fructooligosaccharides (Synergy and Actilight) (scFOS) but they were not able to grow, or did it poorly, in Inulin (**Figure 1**). This observation was further supported by the production of bacterial metabolites (**Figure 2**, Supplementary File) and the pH (Supplementary Figure 1), which in the case of Inulin remained similar to those of the negative control without carbon source added (WCS). Interestingly, B. longum IPLA20022 showed a significantly higher growth (p < 0.05) in the prebiotics Synergy and Actilight than in glucose (**Figure 1**), whereas no statistically significant differences were observed for B. breve IPLA20006 or B. bifidum IPLA20015 between glucose and these two prebiotics. The mono-cultures of B. animalis Bb12 showed a significantly lower (p < 0.05) growth in all prebiotics than in glucose. This strain exhibited the lowest growth of all bifidobacteria in glucose, Synergy, and Actilight (**Figure 1**), which correlates with the limited drop in pH observed for this strain after 24 h of incubation (Supplementary Figure 1). With regard to the pathogen, C. difficile grew well in Synergy, not differing significantly from glucose, and to a lower extent in Actilight (**Figure 1**). Therefore, in spite of generally claimed high specific fermentation of prebiotic substrates, some intestinal pathogens may also be able to ferment and grow in some of them. This underlines the importance of a careful selection of the most appropriate strains, substrates, and combinations.

When co-cultured with C. difficile in the different carbon sources, the behavior of the bifidobacteria was, in general, similar to that observed in the mono-cultures. We observed increases in bifidobacterial counts in glucose, Synergy, and Actilight and poor grow in Inulin. Regarding C. difficile, it grew better in glucose, followed by Synergy, which is in agreement with the

mono-culture data, but the growth in Actilight was, in general, significantly (p < 0.05) worse in co-culture, the contrary being true for Inulin (**Figure 1**). This growth behavior of C. difficile in the different carbon sources was further confirmed by the metabolites production pattern (Supplementary File), showing in general a lower production of C. difficile metabolites, such as propionate or branched-SCFA, in co-culture with Actilight as carbon source than in the corresponding mono-culture, whilst the contrary was observed for Inulin.

When co- and mono-cultures were compared within the same carbon source, the growth of C. difficile was significantly reduced (p < 0.05) by B. longum IPLA20022, B. breve IPLA20006, or B. animalis Bb12 in glucose. The first two microorganisms also reduced C. difficile growth when co-cultured in Actilight and, in the case of B. longum also when Synergy was used as carbon source (**Figure 1**). On the contrary, no statistically significant differences between mono- and co-cultures were observed for B. bifidum in any carbon source. These results showed a good correlation with the pattern of production of C. difficile metabolites and the drop in pH (Supplementary File). This suggests the production of organic acids, with the concomitant reduction of the pH, as an important mechanism of inhibition (Tejero-Sariñena et al., 2012).

These results point out at B. longum IPLA20022 and B. breve IPLA20006, and the prebiotics Synergy and Actilight, as the most promising alternatives for inhibiting the growth of C. difficile. Moreover, they suggest that the pathogen inhibition is strain and substrate specific, which is in agreement with previous reports (Kondepudi et al., 2012; Tejero-Sariñena et al., 2013; Ambalam et al., 2015). Interestingly, the growth of C. difficile was significantly increased (p < 0.05) by B. breve in the presence of Inulin, indicating a potential risk of such combination and underlining the importance of a careful strain and substrate specific assessment.

Interestingly, effects of the co-culture with C. difficile on the growth of the bifidobacterial strains were also observed. Whilst in glucose the co-culture with the clostridia did not affect the growth of B. longum, it significantly (p < 0.05) reduced that of B. breve but increased that of B. animalis. Moreover, the growth of the latter microorganism was also increased by the presence of C. difficile in the three prebiotics tested, mainly Synergy (**Figure 1**) which was further confirmed by an enhanced production of acetate in the co-culture than in the corresponding monoculture (**Figure 2**) and a higher drop in pH (Supplementary Figure 1).

#### The Carbon Source Determines the Toxicity of C. difficile Supernatants

In addition to bacterial growth, inhibiting the toxicity caused by C. difficile, for example by reducing toxin production or toxic activity, represents another target in CDI (Trejo et al., 2010, 2013). The toxicity of C. difficile culture supernatants has been found to be dependent on the culture media used (Valdés et al., 2015), suggesting a potential role of the carbon source available. Therefore, it is important to know whether the availability of different prebiotics as carbon source may have an impact on the toxicity of C. difficile. To clarify this point we determined the toxicity of neutralized cell-free supernatants, obtained from C. difficile monocultures after 24 h of incubation in the different carbon sources, upon the human epithelial cell line HT29 by using a real-time monitoring system (RTCA). To this end the EC50 values, defined as the concentration of supernatant causing 50% of the maximum cell damage, were calculated (Valdés et al., 2015). Supernatants obtained from the mono-cultures carried out without any carbon source or with Actilight added were significantly (p < 0.05) more toxic than the others (**Figure 3**). They showed EC50 values below 2%, which means that a concentration of monoculture supernatant lower than 2% already produced half of the maximum cell damage. On the contrary, the supernatant of the mono-culture in glucose resulted significantly (p < 0.05) less toxic than all the others (EC50 value over 6%), followed by that on Synergy and the one carried out with Inulin as carbon source (**Figure 3**). The method used (Valdés et al., 2015) allowed us to determine that the C. difficile supernatants' toxicity was higher when no carbon source was added or when the available carbon source supported only a limited growth of

the pathogen, such as in the case of Actilight. On the contrary, the supernatant obtained when the pathogen was grown in glucose, in spite of the good growth of C. difficile, resulted less toxic. The availability of rapidly metabolizable sugars has been reported to inhibit toxin synthesis in C. difficile (Bouillaut et al., 2015). This inhibition is mediated through repression of treR (also known as tdcR), an alternative sigma factor responsible for the positive regulation of toxA and toxB genes (Mani et al., 2002). Our results seem to confirm the higher production of toxins by C. difficile under nutrient limitation or stress conditions in which readily fermentable sugars are not available. Moreover, in C. difficile a co-induction of metabolic pathways, such as that of butyrate production, and toxin production has been reported (Karlsson et al., 2000). In our study the C. difficile monoculture grown in glucose showed, in general, lowest butyrate production than those carried out with Synergy, Actilight or WCS added, which is in good agreement with the lower toxin production in glucose. However, the culture of the strain in Inulin, in spite of a lower production of butyrate than that in glucose, showed higher toxin concentrations, comparable to those found WCS or in the other prebiotics tested. These results indicate that, at least in some circumstances, toxin production by C. difficile is uncoupled from the production of metabolites such as butyrate.

In accordance with the above mentioned toxicity data, the concentration of C. difficile toxin A showed the lowest value in the supernatant from the culture in glucose (**Figure 4**). The supernatants obtained from cultures grown in Synergy, Actilight showed the highest toxin concentrations, whilst those from the growth on C. difficile in Inulin or WCS showed intermediate levels.

#### Co-culture with Bifidobacteria in Different Carbon Sources Reduces C. difficile Toxicity

The ability of certain bifidobacterial strains, such as B. longum IPLA20022, to remove toxins from C. difficile cell-free

supernatants, then diminishing their cytotoxicity, has been recently reported (Valdés et al., 2016). Now we compared the toxicity of the Clostridium-Bifidobacterium co-culture supernatants with that of the pathogen monoculture. In general we observed a significant reduction on the toxicity of the supernatants in co-culture. However, differences depending on the strain and the carbon source used were also observed, confirming the high specificity of these interactions (Trejo et al., 2010). The toxicities obtained for the co-cultures in the different carbon sources were compared by using the normalized cell index (CI) obtained after 12 h of incubation of HT29 cells with a 5% of the culture supernatants. As it was the case for the monocultures, supernatants from co-cultures carried out on the different carbon sources showed differences among them (p < 0.05) (**Table 1**). Similarly to the mono-cultures, supernatants obtained in glucose showed the lowest toxicity whilst those in Inulin, or without any carbon source added, resulted the most toxic. When the supernatants of the co-cultures with the different bifidobacteria were compared with the C. difficile monoculture no statistically significant differences (p > 0.05) were obtained in media WCS added. However, in all the carbon sources tested, either glucose or prebiotics, statistically significant differences (p < 0.05) were observed depending on the bifidobacterial strain used (**Table 1**). Co-culture in Synergy or Actilight of C. difficile with B. longum IPLA20022 or B. breve IPLA20006 significantly (p < 0.05) inhibited the toxicity of the supernatant (i.e., higher normalized CI) when compared with the mono-culture of C. difficile. However, B. bifidum IPLA20015 only was able to reduce (p < 0.05) the toxicity of the pathogen with Actiligh as carbon source whilst B. animalis Bb12 did not produce toxicity inhibition in any prebiotic. The four bifidobacteria tested were able to reduce (p > 0.05) the toxicity of the supernatant when co-cultured in glucose, in comparison to the C. difficile mono-culture, but none of them did it when the carbon source was Inulin. In the latter case, even, an increase in the toxicity was observed when the pathogen was co-incubated TABLE 1 | Normalized cell index (mean ± sd) obtained after 12 h of incubation of HT29 cells with the supernatants (5%) of the C. difficile mono-culture or C. difficile-Bifidobacterium co-cultures grown in different prebiotics, glucose or without any carbon source added (WCS).


\*Different superscripts letters within the same row indicate statistically significant differences (p < 0.05) among carbon sources, whereas different superscript numbers within the same column denote differences (p < 0.05) among cultures.

with B. bifidum (**Table 1**), suggesting a potential risk for such combination.

Our results show that B. longum IPLA20022 and B. breve IPLA20006 reduced the toxicity of the co-cultures with sc-FOS as carbon source. Interestingly these two strains have previously shown the ability to remove C. difficile toxins from solution (Valdés et al., 2016). Although the putative mechanism behind toxin inactivation remains to be elucidated, it has been demonstrated that certain microorganisms produce compounds able to degrade C. difficile toxins or to reduce their toxicity (Castagliuolo et al., 1996; Banerjee et al., 2009; Carasi et al., 2012; Valdés et al., 2016). These mechanisms may be involved in the effect observed by us. However, given that in our case both microorganisms are co-incubated, the direct inhibition of the growth of the pathogen and/or an modulation of the expression of the toxin genes in C. difficile by the presence of bifidobacteria, similarly to that previously reported for Lactobacillus acidophilus (Yun et al., 2014), may also be involved. Previous studies pointed out a role of organic acids, such as lactic acid, in the inhibition of both growth and toxin production by C. difficile (Kolling et al., 2012; Yun et al., 2014). Therefore, the ability of bifidobacteria to produce acids, mainly acetic and lactic acids, and the pH drop caused by them may partially explain our observations. However, the role of other interactions cannot be overruled, especially since behaviors not explained by the acids, such as the increased toxicity of the co-culture C. difficile-B. bifidum in Inulin, were also observed.

#### CONCLUSION

Co-culture with B. longum IPLA20022 or B. breve IPLA20006 in the presence of scFOS, but not of Inulin, reduces significantly the growth of C. difficile. Moreover, co-culture with these two strains

#### REFERENCES

Allen, S. J., Wareham, K., Wang, D., Bradley, C., Hutchings, H., Harris, W., et al. (2013). Lactobacilli and bifidobacteria in the prevention of antibiotic-associated diarrhoea and Clostridium difficile diarrhoea in older inpatients (PLACIDE): in Synergy or Actilight reduced the toxicity of the C. difficile supernatants. Therefore, B. longum IPLA20022 and B. breve IPLA20006, in combination with Synergy or Actilight, are the most promising strains and compounds for the development of probiotic, prebiotic, or synbiotic products targeting at the reduction of CDI. However, future in vitro studies aiming at other clinically relevant C. difficile strains, as well as in vivo evaluation of the efficacy of the products, would be needed before drawing firm conclusions.

### AUTHOR CONTRIBUTIONS

MG and PR contributed with the conception, experimental design, and results interpretation of this study. LV carried out all experiments, AH performed chromatographic analyses. MG was in charge of writing the drafted manuscript. All authors performed a critical revision of the manuscript and approved the final version.

### ACKNOWLEDGMENTS

This work was fund by FEDER European Union funds through the projects AGL2012-33278 and AGL2013-43770R from the Spanish Ministry of Economy and Competitiveness (MINECO), and through the grants EQUIP11 and GRUPIN14-043 from the Program of Science, Technology and Innovation from the "Principado de Asturias." LV acknowledges her JAE-Pre fellowship to CSIC.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.00738

a randomised, double-blind, placebo-controlled, multicentre trial. Lancet 382, 1249–1257. doi: 10.1016/S0140-6736(13)61218-0

Ambalam, P., Kondepudi, K. K., Balusupati, P., Nilsson, I., Wadström, T., and Ljungh, A. (2015). Prebiotic preferences of human lactobacilli strains in co-culture with bifidobacteria and antimicrobial activity against Clostridium difficile. J. Appl. Microbiol. 119, 1672–1682. doi: 10.1111/jam. 12953


for prevention of Clostridium difficile infection in a murine model. Microbiol. Immunol. 58, 552–558. doi: 10.1111/1348-0421.12184


difficile ribotype 027 displacement of endemic strains: an epidemiological model. Sci. Rep. 5:12666. doi: 10.1038/srep12666


**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 Valdés-Varela, Hernández-Barranco, Ruas-Madiedo and Gueimonde. 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.

# Glycan cross-feeding activities between bifidobacteria under *in vitro* conditions

*Francesca Turroni1\*, Ezgi Özcan2, Christian Milani1, Leonardo Mancabelli1, Alice Viappiani3, Douwe van Sinderen4, David A. Sela2,5 and Marco Ventura1*

*<sup>1</sup> Laboratory of Probiogenomics, Department of Life Sciences, University of Parma, Parma, Italy, <sup>2</sup> Department of Food Science, University of Massachusetts Amherst, Amherst, MA, USA, <sup>3</sup> GenProbio s.r.l., Parma, Italy, <sup>4</sup> Alimentary Pharmabiotic Centre and Department of Microbiology, Bioscience Institute, National University of Ireland, Cork, Ireland, <sup>5</sup> Center for Microbiome Research, University of Massachusetts Medical School, Worcester, MA, USA*

#### *Edited by:*

*Joerg Graf, University of Connecticut, USA*

#### *Reviewed by:*

*Michael C. Nelson, University of Connecticut, USA Nicole Marie Koropatkin, University of Michigan Medical School, USA*

#### *\*Correspondence:*

*Francesca Turroni, Laboratory of Probiogenomics, Department of Life Sciences, University of Parma, Parma, Italy francesca.turroni@unipr.it*

#### *Specialty section:*

*This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology*

*Received: 26 June 2015 Accepted: 10 September 2015 Published: 24 September 2015*

#### *Citation:*

*Turroni F, Özcan E, Milani C, Mancabelli L, Viappiani A, van Sinderen D, Sela DA and Ventura M (2015) Glycan cross-feeding activities between bifidobacteria under in vitro conditions. Front. Microbiol. 6:1030. doi: 10.3389/fmicb.2015.01030* Bifidobacteria colonize the gut of various mammals, including humans, where they may metabolize complex, diet-, and host-derived carbohydrates. The glycan-associated metabolic features encoded by bifidobacteria are believed to be strongly influenced by cross-feeding activities due to the co-existence of strains with different glycan-degrading properties. In this study, we observed an enhanced growth yield of *Bifidobacterium bifidum* PRL2010 when co-cultivated with *Bifidobacterium breve* 12L, *Bifidobacterium adolescentis* 22L, or *Bifidobacterium thermophilum* JCM1207. This enhanced growth phenomenon was confirmed by whole genome transcriptome analyses, which revealed co-cultivation-associated transcriptional induction of PRL2010 genes involved in carbohydrate metabolism, such as those encoding for carbohydrate transporters and associated energy production, and genes required for translation, ribosomal structure, and biogenesis, thus supporting the idea that co-cultivation of certain bifidobacterial strains with *B. bifidum* PRL2010 causes enhanced metabolic activity, and consequently increased lactate and/or acetate production. Overall, these data suggest that PRL2010 cells benefit from the presence of other bifidobacterial strains.

Keywords: microbiota, microbe–microbe interactions, RNAseq, transcriptomics

# Introduction

Bifidobacteria are key gut commensals of human beings, reaching a high-relative abundance when their host is an infant (Ventura et al., 2012; Turroni et al., 2014b). These bacteria have a saccharolytic lifestyle and their metabolism is consequently directed toward the utilization of carbohydrates that are naturally occurring in their ecological niches, which not only include dietary glycans, such as resistant-starch and xylan, but also host-derived glycans, i.e., mucin and human milk oligosaccharides (Turroni et al., 2010, 2011b; Duranti et al., 2014, 2015). In order to access these glycans, bifidobacteria have evolved an enzymatic repertoire of extracellular glycosyl hydrolases (GH) that catalyze the breakdown of such polysaccharides, with the production of mono, di/trisaccharides that are then imported into the cell through the action of specific carriers (Turroni et al., 2012). In addition, other species such as *Bifidobacterium longum* ssp. *infantis* possess apparently unique molecular mechanisms to capture intact oligosaccharides to be further processed intracellularly (Sela, 2011). It should, however, be considered that in their ecological niche bifidobacterial populations may interact with different strains/species which may lead to competition for or co-operative sharing of nutrients.

Biotic interactions between bacteria can either positively or negatively influence the fitness of the affected organisms (Pande et al., 2015). Several of these interactions rely on either the active or passive release of chemical molecules into the environment (Phelan et al., 2012; Morris et al., 2013). In this context, interspecies cross-feeding has been observed for *Bifidobacterium bifidum* PRL2010 and *Bifidobacterium breve* UCC2003 cells when cultivated on sialyllactose as the unique carbon source (Egan et al., 2014a,b). Moreover, previous studies have demonstrated metabolic cross-feeding between *Bifidobacterium adolescentis* and lactate-utilizing, butyrate-producing *Firmicutes* bacteria related to *Eubacterium hallii* and *Anaerostipes caccae* (Belenguer et al., 2006). This is significant and relevant to host health as butyrate is widely regarded as a beneficial short-chain fatty acid produced by elements of the microbiota. Furthermore, bacterial cross-feeding opportunities as facilitated by members of the colonic microbiota have been considered to be pivotal for carbohydrate turn-over in this ecological niche (De Vuyst and Leroy, 2011).

In recent years, extensive scientific efforts have been made to decode bifidobacterial genome sequences, which are part of a novel discipline called probiogenomics, directed to understand the genetics sustaining the adaptation of these bacteria to the intestine (Ventura et al., 2009; Milani et al., 2014). In this context, the genome sequence of *B. bifidum* PRL2010, an infant gut isolate, exhibits several genetic adaptations to the human gut and for this reason is employed as a model strain to investigate the biology of infant-associated bifidobacteria (Turroni et al., 2010, 2013; Serafini et al., 2014). The ability of PRL2010 to utilize host-derived glycans such as mucins and human milk oligosaccharides, and the capacity to produce pilus-like structures to facilitate gut colonization and immuno modulation, are clear examples of such genetic adaptations (Turroni et al., 2010, 2013; Serafini et al., 2014).

Here, we investigate possible cross-feeding activities of simple bifidobacterial communities under *in vitro* conditions, targeting specific complex, diet-associated carbohydrates. Such crossfeeding activities were investigated employing transcriptome analysis of bifidobacterial communities by means of RNAseq as well as by assessment of the metabolic profiles of these bifidobacterial consortia. The main findings of the current study are that *B. bifidum* PRL2010 does not utilize starch or xylan, unless co-cultured with a strain that produces extracellular glycoside hydrolases that can degrade these substrates. These findings provide evidence of mutalisitic cross-feeding between certain bifidobacterial strains when co-cultured in media containing starch or xylan.

# Materials and Methods

#### Growth Conditions

*Bifidobacterium bifidum* PRL2010 (Turroni et al., 2010) on its own, or in combination with *B. breve* 12L. (Bottacini et al., 2014), *B. adolescentis* 22L (Duranti et al., 2014), or *Bifidobacterium thermophilum* JCM1207 were cultivated in an anaerobic atmosphere (2.99% H2, 17.01% CO2, and 80% N2) in a chamber (Concept 400; Ruskin) at 37◦C for 24 h in de Man–Rogosa–Sharpe (MRS; Scharlau Chemie, Barcelona, Spain) medium, supplemented with 0.05% (wt/vol) L-cysteine hydrochloride.

#### Co-Cultivation

Viable cells of each of the following strains: *B. bifidum* PRL2010, *B. breve* 12L, *B. adolescentis* 22L, or *B. thermophilum* JCM1207, or these strains in co-cultivation with *B. bifidum* PRL2010 was inoculated in 6 ml of MRS (without any carbohydrate; Scharlau Chemie, Barcelona, Spain) supplemented with 1% of either RS2-resistant starch or xylan (Poly(β-Dxylopyranose[1→4]; Sigma–Aldrich) as the sole carbon source in triplicates. Cell suspensions were mixed and incubated at 37◦C for 24 h under anaerobic conditions. Bacterial cell cultivations were performed in triplicate (biological replicates).

Bacterial strain enumerations at the beginning and at the end of a given growth experiment were determined by quantitative real-time PCR (qRT-PCR).

#### Evaluation of PRL2010 Cell Numbers in Co-Cultivation Trials

Possible enhancement or reduction of PRL2010 growth as a consequence of co-cultivation with other bacteria was monitored by qRT-PCR at the begin as well as at the end of the growth experiments. The amounts of cells for each of the strain used at the begin of the growth evaluation trials determined by qRT-PCR assays are shown in **Supplementary Figure S1**. qRT-PCR experiments were based on strain-specific primers targeting genes present in single copy within the genomes of PRL2010 (BBPR\_0282) 12L (B12L\_0105), 22L (BADO\_1546), and JCM1207 (BTHER\_1915). The copy-number of a gene, and the deduced cell number (since the genes targeted were in single copy per genome) of a given strain used in the cocultivation experiments was evaluated by comparing the cycle threshold (Ct) values obtained with those from a standard curve. Standard curves were calculated from serial dilutions of a culture with a known cell number (as determined by viable count assessment) for each bacterial strain vs. Ct produced for each target gene. In the case of PRL2010 we used the following primer couple: BBPR\_0282-UNI (5 - GCGAACAATGATGGCACCTA-3 ) and BBPR\_282-REV (5 -GTCGAACACCACGACGATGT-3 ). In the case of *B. breve* 12L, we used 12L-UNI (5 -CGAAGTTCCAGTTCACCAT-3 ) and 12L-REV (5 -GTTCTTGGCGTTCCAGATGT-3 ); for *B. adolescentis* 22L we employed the PCR primers 22L-UNI (5 -GACCAAGCCAACCAGTTCAT-3 ) and 22L-REV (5 -TTGGTGGCCTTGTAGTAGCC-3 ); and for *B. thermophilum* JCM1207 the following PCR primers BTHERfw (5 -TTACACGCATCCCAATACGC-3 ) and BTHERrv (5 - CGTGAAGTATGGATGGTCGC-3 ). These primers target genes of the sortase-dependent pilus loci identified in the genomes of these microorganisms (Turroni et al., 2014a).

#### Metabolic Profiling

For quantitative determination of metabolites produced by bifidobacterial fermentation of starch and xylan, cell-free supernatants were analyzed using Agilent 1260 Infinity HPLC system equipped with Wyatt Optilab T-rEX Refractive Index detector. Separation was carried out using a Shodex Sugar SH1011 column (8.0 mm ID × 300 mm) at 60◦C with the detector temperature maintained at 30◦C. The mobile phase was prepared in 20 mM H2SO4 and run at a flow rate of 0.6 mL/min for 30 min. Injection volume was set at 10 μL and speed of draw and eject was set to 100 μL/min. External sugar standards (glucose, maltose, xylose, and maltotriose) and organic acid (acetic acid and lactic acid) standards were purchased from Sigma–Aldrich Co. (USA). Concentrations of individual sugars and organic acids in samples were calculated from calibration curves drawn from external standards for five different concentrations (0.5, 1, 5, 10, and 20 mg/L). Each metabolic profiling experiment was carried out in triplicate (three measurements were performed for each replicate). Glucose and maltose consumption was calculated by subtracting the values at 24 h from time zero. Since the concentration of metabolites produced is dependent upon the cell density at the end of fermentation (i.e., more cells would yield more endproducts), raw metabolite values were normalized by cell density to correct for differences in biomass.

#### RNA Extraction and Purification

Total RNA was isolated using a previously described method (Turroni et al., 2011a). Briefly, cell pellets/tissue materials were resuspended in 1 mL of QUIAZOL (Qiagen, UK) and placed in a tube containing 0.8 g of glass beads (diameter, 106 μm; Sigma). Cells were lysed by shaking the mix on a BioSpec homogenizer at 4◦C for 2 min (maximum setting). The mixture was then centrifuged at 12,000 rpm for 15 min, and the upper RNAcontaining phase was recovered. The RNA sample was further purified by phenol extraction and ethanol precipitation according to an established method (Sambrook and Russel, 2001). RNA quality was checked by analyzing the integrity of rRNA molecules by a Tape Station (Agilent Technologies).

#### RNAseq with Ion Torrent Personal Genome Machine (PGM)

One hundred nanogram of total RNA was used as the starting input for RNA-Seq library preparation. Briefly, 100 ng of total RNA was treated with MICROB*Express*TM Bacterial RNA Enrichment Kit (Ambion) to remove rRNA according to the supplier's instructions. The efficacy of rRNA depletion was checked by a Tape Station (Agilent Technologies), after which rRNA-depleted RNA samples were fragmented using RNaseIII (Life Technologies, USA) followed by size evaluation using a Tape Station (Agilent Technologies). Whole transcriptome libraries were constructed using the Ion Total-RNA Seq Kit v2 (Life Technologies, USA). Barcoded libraries were quantified by qRT-PCR and each library template was amplified on Ion Sphere Particles using Ion One Touch 200 Template Kit v2 (Life Technologies, USA). The samples were loaded on 316 Chips and sequenced by means of a PGM instrument (Life Technologies, USA). Sequencing reads were depleted of adapter sequences and quality filtered (with overall quality, quality window, and length filters) using a custom script implying fastq-mcf1 (settings: –qual-mean=25, -w 5 –q 20 –l 100). The resulting processed sequences were aligned to the reference genomes through BWA (Li and Durbin, 2009) with mismatch and gap penalty increased to six and eight, respectively, in order to avoid cross-mapping during analysis of co-cultivation samples. Reference genomes used for RNASeq bioinformatics analyses are deposited under the following accession numbers: CP006711 (*B. breve* 12L), CP007443 (*B. adolescentis* 22L), CP001840 (*B. bifidum* PRL2010), and GCA\_000741495.1 (*B. thermophilum* JCM1207). Counting of the number of reads that correspond to annotated open reading frames (ORFs) was performed using HTSeq2 and analysis of the count data was performed using the R package DESeq (Anders and Huber, 2010).

#### Statistical Analysis

Statistical significance between means was analyzed using the unpaired Student's *t*-test . Values are expressed as the means ± SE of the mean of a given experiment performed in triplicate.

#### Sequence Accession Numbers

All RNAseq raw data from this study were deposited in the SRA database under the accession number SRP058697 (BioProject accession number PRJNA284883).

# Results and Discussion

#### Evaluation of the Growth Performances on Dietary Polysaccharide

Growth capabilities of *B. bifidum* PRL2010, *B. adolescentis* 22L, *B. breve* 12L, and *B. thermophilum* JCM1207 cultivated on their own (mono-association) on MRS supplemented with starch or xylan as the sole carbon source were evaluated (**Figure 1**; **Supplementary Figure S1**) and compared to those achieved when these strains were co-cultivated in pairs (bi-associations) on identical substrates. In addition, cultivation experiments involving all the strains used in this study were also performed on MRS without carbon source, revealing, as expected, the absence of any sign of growth. These carbohydrates were selected in order to represent carbon sources that may occur in the human gut as they are derived from a plant-derived, glycanbased diet (Chassard and Lacroix, 2013). Notably, of all the possible strain pair combinations, only those bi-associations that included PRL2010 cells displayed significant differences in growth with respect to the respective mono-associations (**Figure 1**; **Supplementary Figure S1**). As expected, PRL2010 cells did not exhibit any significant growth on starch or xylan when grown in a mono-culture (Turroni et al., 2012). However, when this strain was co-cultivated with 22L or 12L cells on a MRS-supplemented with starch, the number of cells increased (about three- and fourfold, respectively, *p <* 0.005) compared to the situation of mono-association, respectively (**Figure 1**).

<sup>1</sup>http://code*.*google*.*com/p/ea-utils

<sup>2</sup>http://www-huber*.*embl*.*de/users/anders/HTSeq/doc/overview*.*html

to all transcribed genes. Asterisks indicate that the presented data display a significant, either ∗*p <* 0.05 or ∗∗*p <* 0.01, deviation from the obtained values of the

In contrast, 12L appears to significant decrease (compared to mono-association; fourfold, *p <* 0.001) during growth on starch when co-cultivated with the presence of PRL2010 cells (**Figure 1**). Conversely, *B. bifidum* PRL2010 did not appear to influence the growth yields (changes less than twofold) of 22L cells on starch, or that of 12L when cultivated on xylan, as displayed by unchanged growth yields of bi-association compared to mono-association of these strains (**Figure 1**). Another interesting increase in deduced PRL2010 cell numbers was observed for the bi-association of PRL2010 with *B. thermophilum* JCM1207 (threefold, *p <* 0.001), when grown on MRS supplemented with xylan (**Figure 1**). In contrast, JCM1207 cells exhibit a modest reduction in growth ability (relative to mono-association), when co-cultivated with PRL2010 (**Figure 1**; **Supplementary Figure S1**). Overall, the concomitant presence of two different bifidobacterial strains was shown in some cases to increase the cell numbers of one partner when cultivated on complex carbohydrates, thus suggesting cross-feeding abilities of bifidobacterial strains.

#### Assessing the Metabolic Profile of Bifidobacteria

In order to investigate if the co-occurrence of two strains influences bifidobacterial metabolism, we evaluated the production of the metabolic endproducts acetate and lactate, along with the depletion of various sugars from the culture supernatant (i.e., glucose and maltose). The results of this metabolic comparison between mono-associations and biassociations (as collected from three independent experiments) are depicted in **Figure 2**. Bifidobacteria produce acetate and lactate as a result of their saccharoclastic fermentative metabolism through the so-called bifid shunt (Sela et al., 2008; Pokusaeva et al., 2011). Interestingly, lactate production from

mono-association.

starch and xylan fermentation showed significant differences between bifidobacterial species. Whereas acetate production from xylan fermentation varied significantly between certain species in mono-associations (PRL2010 vs. 12L, 12L vs. 22L, and 12L vs. JCM1207), and between 12L and all 12L co-cultivations, its production from starch did not show significant variation between species or bi-associations (**Figure 2C**). Since *B. bifidum* PRL2010 did not exhibit significant growth utilizing starch as a sole carbohydrate, acetate, and lactate production was not observed 24 h into the fermentation. Although growth of PRL2010 was enhanced when co-cultivated with 12L, 22L, or JCM1207 in a medium containing xylan as the sole carbon source, it did not appear to influence acetate and lactate production compared to the respective mono-associations (**Figure 2B**). Interestingly, lactate production during cocultivation of 22L and JCM1207 in starch increased compared

to the respective mono-associations (**Figure 2A**). This may be indicative of bacterial proto-cooperation when grown on starch, as the same relationship was not observed with xylan. *B. breve* 12L cells produced significantly more lactate and acetate while fermenting xylan in pure culture as compared to its bi-association with other species (**Figures 2B,D**). This suggests that 12L energy metabolism may be inhibited by the presence of other bifidobacteria. Interestingly, 12L did not produce biomass when grown in axenic culture, suggesting metabolic flux in the absence of cellular growth. Glucose and maltose depletion during starch fermentation varied among species. *B. breve* 12L cells consumed more glucose alone than when co-cultivated with PRL2010, JCM1207, and 22L cells in starch fermentation (**Figures 2E,F**). However, this did not coincide with a significant decrease in lactate production (**Figure 2A**). While fermenting starch individually, 22L and JCM1207 consumed similar amounts of glucose (2.56E-10 and 6.33E-10 mg/cell, respectively) and maltose (2.83E-10 and 8.96E-10 mg/cell, respectively), whereas they consumed onefold more glucose and maltose than their consumption in pure cultures during co-culture (**Figures 2E,F**). This, in turn, resulted in a twofold increase in acetate and lactate production in co-culture compared to the situation in mono-associations (**Figures 2A–C**). Although PRL2010 did not produce significant biomass from xylan or starch fermentation, it appears that glucose and maltose from degradation of starch was depleted from the growth medium regardless. This may be explained by the fact that PRL2010 is utilizing free glucose and/or maltose that may be present in very low amounts in the starch-supplemented growth medium (carbohydrate contaminants), thereby allowing very limited growth (**Figures 2E,F**). In general, organic acid production, and sugar consumption did not exhibit a linear correlation among the tested strains. This is likely due to the hydrolysis of dietary oligosaccharides yielding increased concentrations of monomeric and dimeric sugars derived from xylan or starch before they enter the bifid shunt (Ze et al., 2012).

However, since we are measuring the initial and final metabolites produced, we may not have detected intermediate liberated/produced metabolites from the provided carbon sources. Thus, the consumption kinetics is unclear and we cannot exclude the possibility that metabolite concentrations deviate over time due to glycosyl hydrolase activity.

#### Transcriptomics of the Cross-feeding Features

In order to evaluate the molecular aspects behind the crossfeeding activity as observed for some of the bi-associations involving the PRL2010–22L strain combination, or the PRL2010– 12L strain combination when cultivated on starch, or the PRL2010–JCM1207 combination when grown on xylan, RNAseq experiments of these strain combinations cultivated on either of these substrates were performed. In order to increase the robustness of our RNAseq data, two technical replicates starting from the same library for each RNAseq trial were performed. When compared to the reference condition (mono-association) it was identified that the number of genes whose expression in either PRL2010 or the other strains, was significantly upregulated (greater than or equal to twofold change, *p* < 0.005)

up-regulated genes in *B. bifidum* PRL2010 when was co-cultivated with *B. thermophilum* JCM1207 on xylan-based medium. Colors (black to green) represent the average signal intensity.

ranged from 0 to 121. We used cluster orthologous gene (COG) analysis in order to identify differentially transcribed genes that may contribute to specific biological functions within the gut. As illustrated in **Figure 1**, carbohydrate metabolism, corresponding to COG category [G], is one of the COG functions of PRL2010 most significantly affected by the interaction with another bifidobacterial strain. This is probably due to a response to the presence of specific breakdown capabilities exploited by 22L, 12L, or JCM1207 cells. In this context, we observed an up-regulation of an ABC-type transporter-encoding gene (BBPR\_1824), as well as an major facilitator superfamily (MFS) transporter-encoding gene (BBPR\_0146), thus possibly involved in the uptake of simple sugars when co-cultivated with 22L or 12L cells on MRS containing starch as a unique carbon source (**Figure 3**). This finding can be explained by the fact that the extracellular amylases encoded by strains 22L (Duranti et al., 2014) and 12L (Bottacini et al., 2014) generate simple carbohydrates, which may then be imported by PRL2010 cells through its carbohydrate transporter arsenal (Turroni et al., 2012). Other transcriptionally induced genes of *B. bifidum* PRL2010 cells encompass enolase (BBPR\_0711), glucose-6-phosphate isomerase (BBPR\_354), phosphoglycerate kinase (BBPR\_1038), glyceraldehyde 3-phosphate dehydrogenase (BBPR\_0587), and phosphoglycerate mutase (BBPR\_1487), whose functions are predicted to be related to (carbohydrate-dependent) energy production and conversion through the glycolytic pathway (**Supplementary Figure S1**). These observations indicate that PRL2010 cells have enhanced flux through their central fermentative pathway in the presence of 22L or 12L cells. Notably, we also observed the enhanced transcription of genes encoding the various subunits of the ATPase system of PRL2010 strain, which may occur in response to medium acidification as a result of enhanced metabolic activity. Co-cultivation of PRL2010 cells with strain JCM1207 on a xylan-based medium also increased the transcription levels of genes that are predicted to be involved in the carbohydrate metabolism (**Figure 3**). In particular, 121 genes of PRL2010 exhibited transcriptional up-regulation. The upregulated genes include several ORFs-encoding glycolytic enzymatic repertoire of PRL2010, such as enolase (BBPR\_0711), phosphoglycerate mutase (BBPR\_1487), and pyruvate kinase (BBPR\_0747; **Figure 3**).

Interestingly, strain 12L displayed the transcriptional upregulation of 21 genes when it was co-cultivated with PRL2010 on a starch-based medium. Notably, among the up-regulated genes, we observed B12L\_1818 encoding a putative IIBC component of a phosphotransferase system (PTS), which is involved in carbohydrate metabolism (**Figure 3**). Co-cultivation of strain 12L with PRL2010 caused transcriptional up-regulation of 42 genes of *B. breve* 12L on xylan-based medium, some of which are known to be involved in energy generation through the bif shunt, such as pyruvate kinase (B12L\_0681) and phosphoglucomutase (B12L\_1523; **Figure 3**). This is consistent with the observation that 12L produces significantly more organic acid endproducts in axenic culture than when co-cultured with another bifidobacterial species. In contrast, the transcriptomes of strain 22L when cultivated on starch, and that of JCM1207 strain when grown on xylan-based medium were shown to be unaltered when these strains were co-cultivated with PRL2010 cells compared to transcriptome data obtained when these strains were in mono-association. Such findings confirmed the results achieved by growth experiments (see above).

# Conclusion

In this study, we assessed possible glycan cross-feeding activities between simple bifidobacterial communities when metabolizing complex carbohydrates that, being present in the diet, are expected to be commonly found in the human gut. Our results highlight the existence of a gut commensal relationship between different bifidobacterial species (Supplementary Table S1). We showed the *in vitro* ability of *B. bifidum* PRL2010 to crossfeed on sugars released by the starch- and/or xylan-degrading activities of *B. adolescentis* 22L, *B. breve* 12L, and *B. thermophilum* JCM1207. The generated information advances our knowledge on the metabolic adaptability and versatility of these strains, which no doubt will facilitate colonization of the human gut. Interestingly, when *B. bifidum* PRL2010 was co-cultivated with *B. breve* 12L we observed effects on the transcriptomes of both strains, apparently affecting the glycolytic pathway. The observed transcriptional changes may represent a molecular example of a mutualistic relationship between these two bifidobacterial strains, perhaps being a reflection of their common ecological origin, i.e., the infant gut.

The precise characterization of such complex interactions between gut microbiota members is pivotal in the process of modulation of the composition of the intestinal microbial population, especially during probiotic treatments.

# Acknowledgments

We thank GenProbio s.r.l. for financial support of the Laboratory of Probiogenomics. DS is a member of the APC Microbiome institute funded by Science Foundation Ireland (SFI) through the Irish Government's National Development Plan (Grant number SFI/12/RC/2273). We thank Dr. Albert Byung-Hun UM for his assistance and helpful discussions.

# Supplementary Material

The Supplementary Material for this article can be found online at: http://journal*.*frontiersin*.*org/article/10*.*3389/fmicb*.* 2015*.*01030

FIGURE S1 | Evaluation of the cell density of bifidobacterial strains by qRT-PCR assays. Evaluation of the cell density of mono- and co-cultivation of *Bifidobacterium breve* 12L, *Bifidobacterium adolescentis* 22L, *Bifidobacterium bifidum* PRL2010, and *Bifidobacterium thermophilum* JCM1207 strains on xylanand starch-based medium by qRT-PCR experiments. *y*-axis represents the genome copy number/ml of bacterial culture, *x*-axis showed the name of the strains involved in mono- and bi-associations. Empty bars represent the amount of cells present at T0. Asterisks indicate that the presented data display a significant deviation, i.e., \**p <* 0.05 or \*\**p <* 0.01, with respect to the data obtained for the mono-association.

### References


adaptations for milk utilization within the infant microbiome. *Proc. Natl. Acad. Sci. U.S.A.* 105, 18964–18969. doi: 10.1073/pnas.0809584105


**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 Turroni, Özcan, Milani, Mancabelli, Viappiani, van Sinderen, Sela and Ventura. 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.*

# High Iron-Sequestrating Bifidobacteria Inhibit Enteropathogen Growth and Adhesion to Intestinal Epithelial Cells In vitro

Pamela Vazquez-Gutierrez, Tomas de Wouters, Julia Werder, Christophe Chassard and Christophe Lacroix\*

Laboratory of Food Biotechnology, Institute of Food, Nutrition and Health, ETH Zurich, Zürich, Switzerland

#### Edited by:

Marco Ventura, University of Parma, Italy

#### Reviewed by:

Silvia Arboleya Montes, Teagasc and APC Microbiome Institute, Ireland Lorena Ruiz, Universidad Complutense de Madrid, Spain

> \*Correspondence: Christophe Lacroix christophe.lacroix@hest.ethz.ch

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 06 July 2016 Accepted: 06 September 2016 Published: 22 September 2016

#### Citation:

Vazquez-Gutierrez P, de Wouters T, Werder J, Chassard C and Lacroix C (2016) High Iron-Sequestrating Bifidobacteria Inhibit Enteropathogen Growth and Adhesion to Intestinal Epithelial Cells In vitro. Front. Microbiol. 7:1480. doi: 10.3389/fmicb.2016.01480 The gut microbiota plays an important role in host health, in particular by its barrier effect and competition with exogenous pathogenic bacteria. In the present study, the competition of Bifidobacterium pseudolongum PV8-2 (Bp PV8-2) and Bifidobacterium kashiwanohense PV20-2 (Bk PV20-2), isolated from anemic infant gut microbiota and selected for their high iron sequestration properties, was investigated against Salmonella Typhimurium (S. Typhi) and Escherichia coli O157:H45 (EHEC) by using co-culture tests and assays with intestinal cell lines. Single and co-cultures were carried out anaerobically in chemically semi-defined low iron (1.5 µM Fe) medium (CSDLIM) without and with added ferrous iron (30 µM Fe). Surface properties of the tested strains were measured by bacterial adhesion to solvent xylene, chloroform, ethyl acetate, and to extracellular matrix molecules, mucus II, collagen I, fibrinogen, fibronectin. HT29-MTX mucussecreting intestinal cell cultures were used to study bifidobacteria competition, inhibition and displacement of the enteropathogens. During co-cultures in CSDLIM we observed strain-dependent inhibition of bifidobacterial strains on enteropathogens, independent of pH, organic acid production and supplemented iron. Bp PV8-2 significantly (P < 0.05) inhibited S. Typhi N15 and EHEC after 24 h compared to single culture growth. In contrast Bk PV20-2 showed less inhibition on S. Typhi N15 than Bp PV8-2, and no inhibition on EHEC. Affinity for intestinal cell surface glycoproteins was strain-specific, with high affinity of Bp PV8-2 for mucin and Bk PV20-2 for fibronectin. Bk PV20-2 showed high adhesion potential (15.6 ± 6.0%) to HT29-MTX cell layer compared to Bp PV8-2 (1.4 ± 0.4%). In competition, inhibition and displacement tests, Bp PV8-2 significantly (P < 0.05) reduced S. Typhi N15 and EHEC adhesion, while Bk PV20- 2 was only active on S. Typhi N15 adhesion. To conclude, bifidobacterial strains selected for their high iron binding properties inhibited S. Typhi N15 and EHEC in co-culture experiments and efficiently competed with the enteropathogens on mucusproducing HT29-MTX cell lines. Further studies in complex gut ecosystems should explore host protection effects of Bp PV8-2 and Bk PV20-2 mediated by nutritional immunity mechanism associated with iron-binding.

Keywords: iron sequestration, bifidobacteria, enteropathogens, inhibition, intestinal cell, adhesion

# INTRODUCTION

fmicb-07-01480 September 19, 2016 Time: 11:54 # 2

Bifidobacteria are among the first commensal anaerobic bacteria that reach high levels in the infant gut within the first week of life, representing up to 50–80% of the gut bacteria (Jost et al., 2012; Turroni et al., 2012). The establishment of bifidobacteria in the gut has been associated with a broad range of beneficial effects on host health, such as modulation of intestinal microbiota composition, prevention of infection and immune-modulation (Broekaert and Walker, 2006; Yatsunenko et al., 2012). Inhibition of pathogens in the gut by bifidobacteria might be due to production of inhibitory substances, inhibition of epithelial and mucosal invasion of pathogens, competition for limited nutrients and/or the stimulation of mucosal immunity (Marco et al., 2006; Turroni et al., 2014). Potential inhibition mechanisms include the production of short-chain fatty acids and subsequent local pH decrease (Fukuda et al., 2011), or other antimicrobial compounds such as bacteriocins (Cheikhyoussef et al., 2008; Dobson et al., 2012; Martinez et al., 2013). Bifidobacteria can also compete with pathogens for adhesion to intestinal epithelial sites and nutrients, enhancing resistance to colonization of pathogenic bacteria (Collado et al., 2007; Aires et al., 2010).

The gut microbiota is constantly challenged by different stress factors, including enteropathogens, such as Salmonella and Escherichia coli O157:H45 (EHEC; Wardlaw et al., 2010). Pathogenesis of Salmonella requires its adhesion to host cell surfaces followed by invasion of intestinal epithelial cells, leading to systemic spreading (Sansonetti, 2004; Haraga et al., 2008; Santos et al., 2009). EHEC pathophysiology is attributed to the effects of shiga toxins encoded on the pO157 plasmid, survival to harsh conditions and the formation of attachingand-effacing lesions on epithelial cells (Muller et al., 2009; Melton-Celsa et al., 2012; Thiennimitr et al., 2012). To inhibit pathogen infection in the gut, commensal intestinal microorganisms such as bifidobacteria, should be able to compete for corresponding niches. Bifidobacteria have been reported to occupy attachment sites, therefore preventing pathogen invasion and translocation (Bernet et al., 1994; Goto and Kiyono, 2012). The inhibitory activity and mechanisms of bifidobacteria against enteropathogens have been investigated by microbe-microbe and cell-microbe interaction models (Collado et al., 2007).

Different intestinal epithelial cell lines exhibiting specific characteristics and functions of the gut epithelium are used to study host-pathogen interactions. HT29-MTX cell line is a mucus-secreting clone of the HT-29 intestinal epithelial cell line suitable for mimicking the mucosal surface of the gut epithelium, which acts as the first line of interaction between the microbiota and its host (Lesuffleur et al., 1990; Gagnon et al., 2013). The intestinal mucus layer functions as a physical barrier, separating the epithelium from the bacterial load in the intestinal lumen. Mucus is also an important nutrient source for gut microbes and promotes selective adhesion of gut bacteria to the intestinal mucus layer. Interactions with the intestinal mucus layer is a property of commensal gut bacteria that can enhance the barrier function of the intestinal epithelium by limiting access of pathogens to this specific niche. The adherence to intestinal epithelial cells is therefore an important characteristic for beneficial gut bacteria, enhancing persistence in the gut, pathogen exclusion effects and specific bacterial and hostimmune system interactions (Izquierdo et al., 2008; Bron et al., 2012).

The ability of bacteria to establish in the intestine is heavily dependent on competition for nutrients (Andrews et al., 2003). For example iron is an essential micronutrient for growth, proliferation, and persistence for most gut bacteria, including bifidobacteria and enteropathogens (Turroni et al., 2014). Pathogens such as S. Typhi and EHEC are known to possess efficient iron sequestration mechanisms that contribute to their pathogenicity and competitiveness in the gut (Berkley et al., 2005; Wardlaw et al., 2010; Cassat and Skaar, 2013; Monack and Hultgren, 2013; Winter et al., 2013). These systems have been directly linked to the ability of strains with high iron sequestration properties to establish efficiently in the gut (Weinberg, 2009; Kortman et al., 2012). In a previous study we reported isolation of 56 bifidobacterial strains from stools of breast fed, iron-deficient and anemic Kenyan infants (Vazquez-Gutierrez et al., 2015c). Isolated strains were characterized and compared to public culture collection strains. Bifidobacterium kashiwanohense PV20-2 (Bk PV20-2) and Bifidobacterium pseudolongum PV8-2 (Bp PV8-2) were selected for their high siderophore activity (iron-chelating molecules) and iron internalization. Analysis of the complete genome allowed to identify ferrous and specific ferric iron operons in both strains (Vazquez-Gutierrez et al., 2015a,b,c). Furthermore, a ferrous iron-binding protein and other proteins with adhesive properties were identified in the extracellular fraction of Bk PV20-2 together with. In the extracellular proteome of Bp PV8-2 a ferric ironbinding protein belonging to the ferric iron transport operon was shown. In the present study, the inhibitory activity of Bp PV8-2 and Bk PV20-2 was investigated during co-cultures with S. Typhi N15 and EHEC as a function of iron concentrations (1.5 and 30 µM ). Surface properties were tested by bacterial adhesion to solvent (BATS) and extracellular matrix molecules (ECMs) and the competition for epithelial binding sites was studied in HT29-MTX intestinal cellular model.

# MATERIALS AND METHODS

#### Bacterial Strains and Growth Conditions

Bifidobacterium pseudolongum DSMZ20099 (Bp DSMZ20099) and B. kashiwanohense DSMZ21854 (Bk DSMZ21854) were obtained from the German collection of microorganisms (DSMZ; Leibniz, Germany). B. pseudolongum PV8-2 (Bp PV8-2) and B. kashiwanohense PV20-2 (Bk PV20-2), were obtained from the culture collection of the Laboratory of Food Biotechnology (ETH Zurich, Switzerland). Salmonella enterica ssp. enterica serovar Typhimurium N15 (S. Typhi N15) a clinical isolate obtained from the National Centre for Enteropathogenic Bacteria and Listeria (NENT, University of Zurich, Switzerland) and E. coli O157:H45 (EHEC) were kindly provided by Prof. Roger Stephan. Bifidobacteria were routinely cultured in de Man, Rogosa, and Sharpe (MRS) broth (Biolife, Italy) supplemented with 0.05% of L-cysteine hydrochloride monohydrate (cys; Sigma-Aldrich,

Switzerland). Enteropathogens were cultured in Luria-Bertani (LB) broth (Becton Dickinson, Switzerland) unless otherwise specified. Cells suspensions and serial dilutions were carried out in peptone water at pH 6.5, containing 1.5 g/L peptone water (CDH Bioscience, India) and 0.6 g/L cys (peptone-cys). Bifidobacterial viable cell counts were determined on MRS-cys agar (Becton Dickinson, Switzerland) plates, incubated for 72 h under anaerobiosis in anaerobic jars. S. Typhi and EHEC enumeration was done in Mac-Conkey agar (Oxoid, Switzerland) incubated 24 h at 37◦C. A chemically semidefined low iron medium (CSDLIM) with a low iron concentration of 1.5 µM was used for co-culture interaction assays. The CSDLIM medium was previously used to test siderophore production with the CAS assay (Vazquez-Gutierrez et al., 2015c). Iron supplementation of the CSDLIM medium was achieved by adding 30 µM of ferrous iron (Sigma-Aldrich, Switzerland), corresponding to the iron concentration previously reported to increase Salmonella and EHEC pathogenicity (Cernat and Scott, 2012; Kortman et al., 2012). Iron concentration in CSDLIM was measured by graphic furnace atomic absorption spectrometry (Vazquez-Gutierrez et al., 2015c).

#### Inhibitory Activity of B. pseudolongum PV8-2 and B. kashiwanohense PV20-2 during Co-cultures with Enteropathogens

Growth interactions of bifidobacteria and enteropathogens were investigated in CSDLIM with and without added ferrous iron as follow. The corresponding strains were first cultured twice at 37◦C in MRS-cys for 24 h and LB broth for 12 h, respectively. Bacterial cells were harvested by centrifugation (Biofuge Primo, Heraeus, Switzerland) at 4◦C, 16,000 × g for 10 min. The supernatant was discarded and the pellet was resuspended in peptone-cys water to an OD600 nm of 1.0. Hungate tubes containing 10 mL of CSDLIM with headspace filled with CO<sup>2</sup> were inoculated with log<sup>10</sup> 6.5 ± 0.05 CFU/mL Bp PV8-2, log<sup>10</sup> 6.6 ± 0.13 CFU/mL Bk PV20-2, log<sup>10</sup> 5.5 ± 0.06 CFU S. Typhi N15/mL and log<sup>10</sup> 5.4 ± 0.15 CFU/mL EHEC for both mono- and co-cultures. Hungate tubes were incubated at 37◦C for 24 h and samples were taken at 0, 12, and 24 h from the same tube through septum for absorbance determination at OD600 nm (Biowave, CO8000, Biochrom, Ltd, England), pH and viable cell counts. Short chain fatty acid (SCFA) concentrations were measured by high performance liquid chromatography (HPLC; Thermo Fisher Scientific, Switzerland) as previously described (Cleusix et al., 2008). Briefly, 1 mL culture samples were centrifuged for 12 min at 10,000 × g and 4◦C. Supernatant was filtered with a 0.45 µm nylon membrane (Infochroma AG, Switzerland) directly into HPLC vials. Analysis was performed at a flow rate of 0.4 mL/min with 10 mM sulphuric acid as eluent with an injection volume of 20 µL. Mean metabolite concentrations were expressed in millimolar (mM). Three independent repetitions of mono- and co-cultures in CSDLIM with and without ferrous iron supplementation (30 µM ferrous iron) were carried out.

To test the effects of pH decrease and SCFA on enteropathogens growth inhibition, S. Typhi N15 and EHEC were incubated at 37◦C for 12 h, at pH 4.5 (pH measured at the end of co-cultures) and SCFA concentrations where enteropathogen counts began to decrease during co-cultures. After centrifugation at 4◦C and 16,000 × g for 10 min, cell pellets were suspended in peptone-cys water and adjusted to OD600 nm 1.0. Then hungate tubes containing 10 mL of CSDLIM pH 4.5, 7 mM lactate and 13 mM acetate, were inoculated with log<sup>10</sup> 5.4 ± 0.06 CFU/mL S. Typhi N15 and log<sup>10</sup> 5.3 ± 0.13 CFU/mL EHEC, which were the viable cell counts reached in co-cultures after 12 h incubation. Hungate tubes were incubated for 24 h at 37◦C and 1 mL sample was taken every 4 h to determine pH, absorbance at 600 nm and viable cell counts. The experiment was performed in three independent replicates in CSDLIM with and without ferrous iron supplementation.

# Surface Properties of Bifidobacterial Strains

The BATS assay was used to investigate cell surface properties of bifidobacterial strains according to Xu et al. (2009), with slight modifications. Surface hydrophobicity, electron donor and acceptor properties were determined based on the affinity of bifidobacteria to xylene (apolar solvent), chloroform (polar acidic solvent) and ethyl-acetate (polar basic solvent). Bifidobacteria were cultured in MRS-cys and CDSLIM as described above. Briefly, cell pellets were re-suspended in phosphate-buffered saline (PBS), pH 7.3 to an OD600 nm of 1.0. A volume of 3 mL of cell suspension was mixed with 1 mL of either xylene, chloroform (electron-acceptor), or ethyl-acetate (electron-donor; Sigma-Aldrich, Switzerland). The mixture was vortexed for 1 min and allowed to stand for 5 min to allow separation into two distinct phases. Then 1 mL of the aqueous phase was carefully collected with a pipette and OD600 nm was measured using a UV-Visible spectrophotometer CARY 1Bio (Varian, Switzerland). The decrease in absorbance of the aqueous phase after contact with solvent was used as a measure of the cell surface hydrophobicity or electron-donor/electron-acceptor interaction. BATS was expressed by BATS (%) = (1 – A5 min/A0 min) × 100, where A0 min and A5 min were the absorbance before and after extraction with the solvents, respectively (Xu et al., 2009). Three independent replicates of the experiment were carried out.

# Adhesion to Different Intestinal Cell Surface Molecules

The adhesion affinities of bacteria to the ECMs of intestinal epithelial cells were tested as described previously (Sillanpaa et al., 2008), with slight modifications. Briefly, a solution of type II mucus (Sigma-Aldrich, Switzerland) at 50 µg/mL was prepared in Tris-HCl (0.1 M, pH 8). Collagen I, fibrinogen and fibronectin (all from Sigma-Aldrich, Switzerland) were resuspended at 10 µg/mL in PBS (Gibco, Switzerland), pH 7.5. Bovine serum albumin (BSA; Sigma-Aldrich, Switzerland) was resuspended at 50 µg/mL in Tris-HCl and used as control for unspecific adhesion to protein surfaces. 100 µL of each suspension were applied to wells of a MaxiSorpTM 96-well microtiter plate (Nunc, Switzerland) and kept overnight at 4◦C for adsorption. After removal of the liquid, adsorbed molecules were fixed for 10 min

at 65◦C and subsequently blocked with 100 µL PBS 1% tween 20 per well for 1 h at 37◦C. Before application of bacteria, plates were washed three times with 100 µL PBS 0.005% tween 20 to remove unbound ECM, filled with 100 µL PBS and used within 24 h storage at 4◦C.

Bifidobacteria and enteropathogen strains were first cultured in MRSc or LB media, respectively, then transferred and grown in CDLSIM, as presented above. Briefly, cell pellets (24◦C, 16,000 × g during 10 min) were re-suspended in PBS at pH 5.5 and pH 7.5 to OD600 nm of 1.0. 100 µl of bacterial suspension was applied in triplicates to coated plates and incubated for 1 h at 37◦C to induce bacterial adhesion. Wells were washed three times with 100 µL PBS 0.005% Tween 20 to remove unattached bacteria and dried for 10 min at 65◦C. Adhered bacteria were stained with 100 µL crystal violet (1 mg/mL, Sigma-Aldrich, Switzerland) per well for 45 min at room temperature. Crystal violet retained by fixed bacteria after three washing steps with 100 µL PBS was resolubilized in 100 µL citrate buffer (50 mM; pH 4) under continuous shaking at 37◦C for 1 h. Absorbance of solubilized crystal violet was measured at OD595 nm using a Powerwave XS spectrophotometer (Bio Tek, Switzerland). The experiment was carried out in three independent replicates.

#### In vitro Antagonism of Bifidobacteria against Pathogen Adhesion to HT29-MTX Cells

The mucus-secreting intestinal epithelial cell line HT29-MTX was used to investigate adhesion properties of bifidobacteria and enteropathogens, as described previously with slight modifications (Gagnon et al., 2013). Briefly, HT29-MTX cells were seeded in 24-well tissue culture plates (Bioswisstec, Switzerland) at a concentration of 4 × 10<sup>4</sup> cells/well, and grown in Dulbecco's Modified Eagle Medium (DMEM; Sigma-Aldrich, Switzerland) supplemented with 20% fetal bovine serum (Invitrogen, Switzerland), 1% penicillin/streptomycin (Life Technologies, Switzerland), and 1% non-essential amino acids (Life Technologies, Switzerland), at 37◦C and 10% CO<sup>2</sup> in a humidified incubator (RB150, Revco, Switzerland). Culture medium was changed every other day and experiments were performed 21 days post-seeding on fully differentiated, confluent monolayers with mucus secretion verified using Alcian blue (stains acid mucopolysaccharides) and periodic acid Schiff (stains hexose and sialic acid-containing mucosubstances). After full differentiation the medium was exchanged to antibiotic free medium for 24 h. Tested bacterial cultures were grown in CDLSIM medium and prepared as described above. Bacterial cultures were washed with sterile 0.85% NaCl, and resuspended in DMEM for application to the cell monolayers. Cell monolayers were carefully washed with 500 µL of warm PBS. For all tests, bifidobacteria were added at log<sup>10</sup> 7.7 ± 0.12 CFU/mL and S. Typhi N15 and EHEC were added at approximately log<sup>10</sup> 6.3 ± 0.05 CFU/mL in DMEM to the HT29-MTX monolayer. After 2 h incubation at 37◦C, HT29-MTX monolayers were washed twice with PBS to remove non-attached bacteria and detached using 0.25% trypsin-EDTA solution (Life Technologies, Switzerland). Bacterial cell counts were determined as described above. Adhesion was expressed as the percent ratio of adhered bacteria to number of bacteria added to the HT29-MTX cells monolayer. Experiments were performed in triplicates on three consecutive passages of the HT29-MTX cell line.

To determine the inhibition of pathogen adhesion by bifidobacteria the method of Gagnon et al. (2013) was used with slight modifications. Briefly, bifidobacteria were applied to the cell monolayer for 1 h. Then the well was washed once with PBS to remove non-adhering cells and the tested pathogen was added for a further incubation of 1 h. Enumeration of adhered bacteria was performed after serial dilution on respective media. To examine if adhered pathogenic bacteria could be displaced by the addition of bifidobacteria, enteropathogens were incubated 1 h and, after PBS washing, bifidobacteria were added and incubated 1 h. To investigate the ability of bifidobacteria to competitively exclude enteropathogens, bifidobacteria and pathogenic bacteria were added simultaneously to the HT29- MTX monolayer and incubated for 2 h. All incubations were done at 37◦C and 10% CO2. HT29-MTX monolayers were washed twice with PBS to remove non-attached bacteria, and treated with a 0.25% trypsin-EDTA solution for 15 min, for bifidobacteria and enteropathogens enumeration as stated earlier. Activity of bifidobacteria strains to compete with, displace and inhibit the adhesion of S. Typhi N15 and EHEC to the intestinal epithelial cell line HT29-MTX was expressed by the adhesion ratio. This corresponded to the ratio of the percentage of adhered bifidobacteria or pathogenic bacteria following simultaneous addition divided by the percentage of adhesion of the bacteria added alone to the cell culture (Serafini et al., 2013). All the above tests were carried out in triplicates on three consecutive passages of the HT29-MTX cell line.

# Statistical Analysis

To assess differences between treatments in inhibitory activity in mono- and co-culture experiments, and surface properties of bifidobacteria strains, the means of three independent repetitions were compared using un-paired Student's t-test. ANOVA with post hoc Tukey test was used to assess significant affinity of bifidobacteria and pathogens to ECM when compared to PBS control (P-value < 0.05). Significant differences in competition, inhibition and displacement experiments were tested by comparing the means of three independent repetition of the HT29-MTX cell line using un-paired Student's t-test. Statistical significance was established at P-value < 0.05 and SPSS software 17.0 (SPSS, Inc., Chicago, IL, USA) was used.

# RESULTS

## Inhibitory Activity of B. pseudolongum PV8-2 and B. kashiwanohense PV20-2 against Enteropathogens

The inhibitory activities of Bp PV8-2 and Bk PV20-2 against S. Typhi N15 and EHEC were tested in co-culture experiments and compared to mono-cultures of the same strains (**Figures 1** and **2**). Mono-cultures of Bp PV8-2 reached maximum viable

cell counts of log<sup>10</sup> 8.1 ± 0.1 CFU/mL with and without iron supplementation, and pH of 4.5 ± 0.06 and 4.2 ± 0.01 after 24 h incubation in CDSLIM medium, respectively. In decreasing concentration the main organic acids produced were acetate, lactate and formate (**Table 1**). Iron supplementation significantly (P < 0.05) increased production of all metabolites, by 12, 17, and 29% for acetate, lactate and formate, respectively, when compared to cultures in unsupplemented media. Similarly, Bk PV20-2 in mono-cultures reached viable cells counts of log<sup>10</sup> 7.7 ± 0.2 CFU/mL with and without iron, and pH was 5.3 ± 0.04 and 4.7 ± 0.01, respectively. Organic acid productions were also significantly (P < 0.05) increased with iron, by 27% for acetate and 20% for lactate. No significant effect of iron on growth or metabolite production was observed during monocultures of enteropathogens. EHEC reached cell counts of log<sup>10</sup> 7.9 ± 0.1 CFU/mL after 12 h and remained stable until 24 h of culture. pH after 24 h was 4.88 ± 0.01 and 5.05 ± 0.02, with and without iron supplementation, respectively. Main metabolites were lactate and acetate for S. Typhi N15 and EHEC (**Table 1**).

Viable cell counts of S. Typhi N15 significantly decreased (P < 0.05) by log<sup>10</sup> 2.5 ± 0.1 CFU/mL in co-culture with Bp PV8-2 and by log<sup>10</sup> 1.1 ± 0.1 CFU/mL with Bk PV20-2 after 24 h when compared with mono-cultures of S. Typhi N15 (**Figures 1A,C**). pH after 24 h of co-cultures of S. Typhi N15 with Bp PV8-2 was 4.19 ± 0.02 and 4.21 ± 0.02 and with Bk PV20-2 4.62 ± 0.01 and 4.59 ± 0.01, with or without iron supplementation, respectively. EHEC counts significantly decreased after 12 h when co-cultured with Bp PV8-2 compare to monocultures (**Figure 1B**). No significant differences were observed when co-culturing EHEC with Bk PV20-2. Metabolites produced during co-cultures of S. Typhi N15 and EHEC with Bp PV8-2 were in decreasing order acetate, lactate and formate (**Table 1**), whereas with Bk PV20- 2 only acetate and lactate were identified (**Tables 1** and **2**). No significant differences in metabolites were observed following iron supplementation of the media. No significant effect of iron supplementation was detected on EHEC growth during co-cultures (**Figures 1** and **2**).

Salmonella Typhi and EHEC were tested for the inhibitory conditions observed after 12 h of co-culture with bifidobacterial strains. CDSLIM medium was supplemented with 13 mM of acetate and 7 mM of lactate at pH 4.5 and growth tested for 24 h. S. Typhi N15 and EHEC viable cell counts remained constant (log<sup>10</sup> 5.4 ± 0.12 CFU/mL) during 24 h. In contrast a significant viability decrease of S. Typhi N15 (Bp PV8-2 and Bk PV20-2) and EHEC (Bp PV8-2) was measured during co-cultures with both bifidobacteria between 12 and 24 h for both iron levels (**Figures 1A–C** and **2A–C**).

Typhimurium N15 (S. Typhi N15); (B) Bp PV8-2 and E. coli O157:H45 (EHEC); (C) B. kashiwanohense PV20-2 (Bk PV20-2) and S. Typhi N15; (D) Bk PV20-2 and EHEC. Bp PV8-2, Bp PV8-2 in co-culture, S. Typhi N15, S. Typhi N15 in co-culture, Bk PV20-2, Bk PV20-2 in co-culture, EHEC, EHEC in co-culture. Stars (<sup>∗</sup> ) denote a significant (P < 0.05) difference compared with mono-cultures (mean ± SD, n = 3).

TABLE 1 | pH and concentrations of metabolites (mM) in culture supernatant measured with HPLC after 24 h incubation of mono- and co-cultures of Bifidobacterium pseudolongum PV8-2 in CSDLIM media (mean ± SD, n = 3).


Stars (\*) denote a significant (P < 0.05) difference in the respective metabolite production of cultures performed in iron supplemented compared with unsupplemented medium.

# Surface Properties of Bifidobacteria Strains

Physico-chemical characteristics of bifidobacteria cell surfaces, such as hydrophobicity, electron-donor and electron-acceptor properties, are related to adhesion to intestinal epithelial cells. The affinity of bifidobacteria strains to different solvents, xylene, chloroform and ethyl-acetate, was determined using the BATS assay to quantify surface hydrophobicity. Bp PV8-2 showed similar surface properties as the type strain DSMZ20099 (**Figure 3**). Under iron-limited conditions affinity to hydrophobic xylene was 84.4 ± 3.6% for Bp PV8-2 and 88.3 ± 11.6% for Bp DSMZ20099. Affinity to chloroform, an acidic solvent, was 99.2 ± 1% for Bp PV8-2 and 98.4 ± 3.5% for Bp DSMZ20099. No hydrophobic, electron donor/acceptor properties were observed for Bk PV20-2 (**Figure 3B**), whereas Bk DSMZ21854 showed hydrophobic and electron-donor properties only when grown

TABLE 2 | pH and concentrations of metabolites (mM) in culture supernatant measured with HPLC after 24 h incubation of mono- and co-cultures of Bifidobacterium kashiwanohense PV20-2 in CSDLIM media (mean ± SD, n = 3).


Stars (\*) denote a significant (P < 0.05) difference in the respective metabolite production of cultures performed in iron supplemented compared with unsupplemented medium.

in CDSLIM (**Figure 3D**). Electron-acceptor properties were not observed for any of the strains tested. Bifidobacteria strains showed less affinity to all solvents when grown in MRS-cys compared to CDSLIM.

# Adhesion to Different Intestinal Cell Surface Molecules

The intestinal epithelial cell surface is covered by glycoproteins, such as type II mucus, collagen, fibrinogen and fibronectin, and can serve as attachment sites for microbes. The affinity of bacteria to glycoproteins can therefore influence strain capacity to compete for epithelial binding sites. No significant adhesion to the unspecific protein binding control BSA nor to collagen I was shown for the tested strains when compared with the uncoated control wells (**Figures 4A,B**). Bp PV8-2, showed significant (P < 0.05) adhesion at pH 5.5 to type II mucin and Bk PV20-2 and Bk DSMZ21854 to fibronectin when compared with the uncoated control (**Figure 4A**). S. Typhi N15 showed significant (P < 0.05) adhesion to type II mucin, fibronectin, and fibrinogen. EHEC bound to type II mucin and fibronectin when compared with the uncoated control. At pH 7.5, all strains showed similar affinity to glycoproteins, BSA and the uncoated control (**Figure 4B**), except S. Typhi N15 that showed significantly (P < 0.05) higher adhesion to mucin II, fibrinogen, and fibronectin.

## In vitro Inhibition of Bifidobacteria Strains against S. Typhimurium N15 and EHEC

The ability to adhere to mucus and epithelial cells is an important feature for the barrier effect of bifidobacteria. The adhesion ratios of Bk PV20-2 and Bk DSMZ21854 to mucussecreting HT29-MTX were significantly higher (15.6 ± 6.0% and 12.7 ± 2.4%, respectively) when compared to both Bp PV8-2 and Bp DSMZ20099 (1.4 ± 0.4% and 1.3 ± 0.3%, respectively). Very high adhesion ratios of S. Typhi N15 and EHEC were measured, with 87.8 ± 17.5% and 137.6 ± 51.7%, respectively, likely reflecting growth of the enteropathogens during the test.

The ability of bifidobacteria strains, to compete, displace and inhibit the adhesion of enteropathogens was tested on the HT29-MTX epithelial cell model. Both strains exhibited competitive abilities when added together with S. Typhi N15 in the competition assay, as shown by adhesion ratios significantly higher than 1 (1.88 ± 0.64 for Bp PV8-2 and 1.76 ± 0.51 for Bk PV20-2; **Figures 5A,B**). In contrast, adhesion ratios lower than 1 (P < 0.05) were measured for S. Typhi N15 in competition with Bp PV8-2 (0.67 ± 0.08) and Bk PV20-2 (0.80 ± 0.22), indicating that enteropathogen adhesion was decreased in the presence of both bifidobacteria. In the displacement assay, Bp PV8-2 and Bk PV20-2 strains induced the release of S. Typhi N15 bound to HT29-MTX, indicated by adhesion ratios of 0.43 ± 0.15 and 0.44 ± 0.13, respectively. The inhibition assay showed that adhered bifidobacteria prevented the attachment of S. Typhi N15 and stably occupied a sufficient number of adhesion sites on the surface of HT29-MTX cells. Bp PV8-2 showed the highest degree of inhibition of S. Typhi N15 (0.08 ± 0.04) compared to Bk PV20-2 (0.21 ± 0.12).

Bk PV20-2 was not able to competitively exclude EHEC in the competition assay, as indicated in **Figure 5D**. In the displacement assay with EHEC added first, Bp PV8-2 showed low adhesion ratio of 0.16 ± 0.03 compared with 0.51 ± 0.31 for EHEC which was not significantly different from 1 (**Figure 5C**). This data suggest that Bp PV8-2 could not displace previously adhered EHEC. In the inhibition assay (**Figure 5D**), EHEC adhesion could be significantly (P < 0.05) decreased by the presence of adhered Bp PV8-2. Bk PV20-2 did not reduce adhesion of EHEC when added simultaneously (competition assay) or after the addition of EHEC (displacement assay).

# DISCUSSION

#### Inhibitory Activity of B. pseudolongum PV8-2 and B. kashiwanohense PV20-2 against Enteropathogens

Bifidobacteria play an essential role in the development and homeostasis of the host's immune system in infants where they represent one of the first commensal anaerobic bacteria colonizing the gut (Gupta and Garg, 2009). Efficient competition for iron is a key factor for bacterial growth, persistence and establishment in the intestine (Andrews et al., 2003). In our previous research, B. pseudolongum PV8-2 and B. kashiwanohense PV20-2 isolated from anemic infants in Kenya were therefore selected for this study based on their high iron sequestration capacity (Vazquez-Gutierrez et al., 2015c). Their inhibitory activities against two strains of enteropathogens,

S. Typhi N15 and E. coli O157:H45 which are known to efficiently bind iron, were tested in vitro. Bifidobacteria may exert inhibitory activity against enteropathogens by production of organic acids, competition for essential growth nutrients, production of antibacterial peptides and co-aggregation with pathogens (Turroni et al., 2009; Butel, 2014; Ventura et al., 2014). Organic acids can prevent infections of pathogens by lowering the intestinal pH and hence restricting colonization of pathogenic bacteria that are sensitive to low pH (Bernet et al., 1993; Lievin et al., 2000; Shu et al., 2000; Gopal et al., 2001; Shu and Gill, 2001; Hammami et al., 2013).

Co-cultivation experiments revealed the inhibitory effects of Bp PV8-2 and Bk PV20-2 against S. Typhi N15 and EHEC. Both enteropathogens were not affected by incubation at low pH and organic acid concentrations produced by the bifidobacteria strains during co-cultures. While growth of EHEC was inhibited by Bp PV8-2 and slightly reduced by Bk PV 20-2 (not significantly), S. Typhi N15 was significantly reduced by both bifidobacterial strains during co-cultures compared with single cultures. The reduction of viability of enteropathogens in co-cultures could be the result of different factors combined, such as the fitness of the strain under test conditions, and high iron sequestration mechanisms and the production of inhibitory substances by bifidobacteria (Bailey et al., 2011). Additionally, lactate and acetate, may also function as a permeabilizer of the outer membrane of Gram-negative bacteria and may thus accentuate the effects of other inhibitory substances, such as bacteriocins (Oh et al., 2009).

For example, the extracellular proteome of Bp PV8-2 showed the expression of a lysozyme that might contribute to the inhibitory activity of the strain (Vazquez-Gutierrez, 2014), and the effect might be increased by the production of organic acids (Tejero-Sarinena et al., 2012). Both Bifidobacterium strains tested were less efficient against EHEC, possibly because EHEC has the ability to survive in many adverse conditions when it enters starvation, allowing EHEC to adapt to very harsh conditions with almost no available nutrients, including iron (Chekabab et al., 2013).

#### Adhesion and Competition at the Intestinal Epithelium

The potential of Bp PV8-2 and Bk PV 20-2 to compete for adhesion sites was tested. Occupation of adhesion sites can reduce pathogen adhesion to intestinal epithelium and is mediated by bacterial surface properties like hydrophobicity and cell surface proteins (Botes et al., 2008; Xu et al., 2009). Bp PV8-2 had high affinity for the non-polar solvent xylene, indicating that this strain has hydrophobic cell surface properties. In contrast Bk PV20-2

was hydrophilic and showed no acid-base properties. The growth medium of bifidobacteria had a strong effect on strain surface property, in contrast with iron availability which did not affect the surface characteristics. Affinity to solvents data indicated that Bp PV8-2 has similar physico-chemical cell surface properties to type strain Bp DSMZ20099. In contrast, affinities to solvents of Bk PV20-2 and Bk DSMZ21854 were similar when both strains were grown in MRS-cys but very different when grown in CSDLIM, emphasizing the influence of growth conditions on surface properties of bacterial strains (Xu et al., 2009). Canzi et al. (2005) reported that even very close genetically related bifidobacteria strains can exhibit significantly different adhesion activities to hydrocarbons (xylene and hexadecane), supporting high strain specificity (Del Re et al., 2000).

Adhesion affinity to different binding sites of the intestinal epithelium was subsequently quantified by the adhesion affinity to a representative set of surface glycoproteins. Specific binding affinity to type II mucin was low for all strains, consistent with the findings of Collado et al. (2005) who observed weak adherence of bifidobacterial strains of human origin to human intestinal mucus glycoproteins. Bk PV20-2 and Bk DSMZ21854 showed affinity for fibronectin. Bk strains shared the binding affinity to fibronectin with both enteropathogens (Fujiwara et al., 2001), suggesting possible competition for intestinal binding sites by Bk PV20-2 that could prevent infections (Collado et al., 2005; Sperandio, 2012). Adhesion to extracellular glycoproteins of all strains was increased at pH 5.5 compared with pH 7.5. The acid environment resulting from the colonization of bifidobacteria could further support the competition for the epithelial binding sites, emphasizing the importance of the combined effect of physico-chemical affinity and surface properties (de Wouters et al., 2015; Jans et al., 2016).

Competition between bifidobacteria strains and entero pathogens by competition, displacement and inhibition was

(Bk PV20-2) and E. coli O157:H45 (EHEC); (C) Bk PV20-2 and S. Typhi N15; (D) Bk PV20-2 and EHEC. The adhesion ratio corresponded to the ratio of the percentage of adhered bifidobacteria or pathogenic bacteria following simultaneous addition divided by the percentage of adhesion of the Bifidobacterium strain or pathogenic bacteria added alone to the cell culture. Dotted line (adhesion ratio = 1) indicates no effect of interactions of tested cultures. Columns with a star (\*) indicate significantly different (P < 0.05) values when compared to 1 (mean ± SD, n = 3).

then studied on a differentiated, mucus-secreting HT29-MTX cells monolayer. In agreement with adhesion tests with single surface molecules, Bp strains showed only modest adhesion abilities to mucus-secreting HT29-MTX cells compared with Bk strains, S. Typhi N15 and EHEC. The high adhesion properties of both enteropathogens may reflect growth of the strain during the incubation test with cell layers as previously reported for Salmonella in a similar cell test (Dostal et al., 2014). Several studies suggest correlation between adhesion to intestinal cells and cell surface hydrophobicity measured with the BATS assay (Marin et al., 1997; Del Re et al., 2000), a result which was not confirmed in other studies (Savage, 1992; Ouwehand et al., 1999; Canzi et al., 2005; de Wouters et al., 2015). Even though hydrophobicity did not correlate with adhesion properties, BATS assay showed that cell surface properties of Bp PV8-2 and Bk PV20-2 are different, indicating strain-specificity. Bk PV20-2 strain showed no hydrophobic affinity suggesting that adhesion might be mediated by adhesion-like factors (Turroni et al., 2009; Ventura et al., 2014).

Adhesion properties of beneficial bifidobacteria to the mucosa have been shown to promote gut residence time, pathogen exclusion, protection of epithelial cells and immune modulation. Our data indicated that the degree of competition was dependent on bifidobacteria and enteropathogen strain. While both bifidobacteria strains were able to competitively exclude S. Typhi N15, only Bp PV8-2 was able to decrease the adhesion capacity of EHEC. In the presence of Bk PV20-2 adhesion of EHEC was increased, suggesting a sharing of metabolic activities leading to enhanced adhesion (Collado et al., 2005). Previous studies reported increased enteropathogen adhesion by bifidobacteria. For example under similar experimental conditions, Serafini et al. (2013) investigated antagonistic effects of Bifidobacterium bifidum PRL2010 against various enteropathogens, including S. Typhi and EHEC on HT-29 cells not secreting mucus (Gueimonde et al., 2006; Serafini et al., 2013). Our data suggests a direct competition for binding sites that protect the host against invasion of enteropathogens which might also be influenced by strain fitness related to iron sequestration mechanisms

#### REFERENCES


(Chauviere et al., 1992; Lee and Puong, 2002). Both Bp PV8-2 and Bk PV20-2 resulted in marked reductions in adhesion of S. Typhi N15 and EHEC, indicating that colonization with these potential probiotic candidates selected for high iron sequestration mechanisms might offer at least partial protection from infection with enteropathogenic bacteria (Collado et al., 2007). Further experiments have to be performed in vivo to support these effects.

# CONCLUSION

Ability of commensals such as bifidobacteria to restrain pathogen growth in the intestine is strongly affected by niche and nutrient competition. Our study showed that B. pseudolongum PV8-2 and B. kashiwanohense PV20-2, selected for their high iron sequestration mechanisms, exhibited strain-dependent inhibitory activity against S. Typhi N15 and EHEC. These strains may be potential probiotic candidates especially for inhibiting irondependent enteric pathogens such as enterobacteriaceae in the gut. The biological significance of such competitive probiotics and their potential as preventive or curative probiotics should be further investigated in vitro in presence of complex gut microbiota and in vivo with animal models.

# AUTHOR CONTRIBUTIONS

PV-G, TW, CC, and CL designed the experiment. PV-G, JW performed and analyzed the experiments. TW, CC, and CL supervised the experiments. PV-G, JW, TW, CC, and CL wrote the manuscript. All authors read and approved the final manuscript.

# ACKNOWLEDGMENT

This research was supported by own resources at ETH Zurich, Zurich Switzerland.




**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 Vazquez-Gutierrez, de Wouters, Werder, Chassard and Lacroix. 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.

# Proteinaceous Molecules Mediating Bifidobacterium-Host Interactions

Lorena Ruiz <sup>1</sup> \*, Susana Delgado<sup>2</sup> , Patricia Ruas-Madiedo<sup>2</sup> , Abelardo Margolles <sup>2</sup> and Borja Sánchez <sup>2</sup>

<sup>1</sup> Department of Nutrition, Food Science and Food Technology, Universidad Complutense de Madrid, Spain, <sup>2</sup> Department of Microbiology and Biochemistry of Dairy Products, Instituto de Productos Lácteos de Asturias-Consejo Superior de Investigaciones Científicas, Villaviciosa, Spain

Bifidobacteria are commensal microoganisms found in the gastrointestinal tract. Several strains have been attributed beneficial traits at local and systemic levels, through pathogen exclusion or immune modulation, among other benefits. This has promoted a growing industrial and scientific interest in bifidobacteria as probiotic supplements. However, the molecular mechanisms mediating this cross-talk with the human host remain unknown. High-throughput technologies, from functional genomics to transcriptomics, proteomics, and interactomics coupled to the development of both in vitro and in vivo models to study the dynamics of the intestinal microbiota and their effects on host cells, have eased the identification of key molecules in these interactions. Numerous secreted or surface-associated proteins or peptides have been identified as potential mediators of bifidobacteria-host interactions and molecular cross-talk, directly participating in sensing environmental factors, promoting intestinal colonization, or mediating a dialogue with mucosa-associated immune cells. On the other hand, bifidobacteria induce the production of proteins in the intestine, by epithelial or immune cells, and other gut bacteria, which are key elements in orchestrating interactions among bifidobacteria, gut microbiota, and host cells. This review aims to give a comprehensive overview on proteinaceous molecules described and characterized to date, as mediators of the dynamic interplay between bifidobacteria and the human host, providing a framework to identify knowledge gaps and future research needs.

#### Edited by:

Marco Ventura, University of Parma, Italy

#### Reviewed by:

Ines Martinez, University of Nebraska—Lincoln, USA Christophe Lacroix, Eidgenössische Technische Hochschule Zürich, Switzerland

> \*Correspondence: Lorena Ruiz loruiz@ucm.es

#### Specialty section:

This article was submitted to Microbial Symbioses, a section of the journal Frontiers in Microbiology

Received: 26 April 2016 Accepted: 19 July 2016 Published: 03 August 2016

#### Citation:

Ruiz L, Delgado S, Ruas-Madiedo P, Margolles A and Sánchez B (2016) Proteinaceous Molecules Mediating Bifidobacterium-Host Interactions. Front. Microbiol. 7:1193. doi: 10.3389/fmicb.2016.01193 Keywords: Bifidobacterium, host interaction, proteome, immunomodulation, adhesin

# INTRODUCTION

The human gut is inhabited by a trillion of microorganisms which constitute the gut microbiota. These microorganisms are in close contact with the intestinal mucosa, which represents the largest extension of the human body exposed to external stimuli. A complex molecular interplay is established among microbiota, dietary components and host cells, which regulates immune and metabolic functions in the host (Furusawa et al., 2015). Dysbiosis, defined as changes in the gut microbiota structure associated to healthy individuals, disrupts the microbiome-host cross-talk homeostasis and correlates with metabolic and inflammatory disorders (Evans et al., 2013; Patel et al., 2013; Levy et al., 2015).

Pro- and pre-biotics can improve host health through microbiota modulation and immune system boosting (Picard et al., 2005). Some strains of bifidobacteria, which are among the first colonizers of the human intestine and one of the dominant groups in the breast-fed infant microbiota (Garrido et al., 2012), have been attributed several health benefits, encouraging interest in their use as probiotics. Pathogen inhibition and diarrhea amelioration are their best established outcomes and have been related to the production of organic acids (Fukuda et al., 2011), antibacterial peptides (Moroni et al., 2006), quorum-sensing inhibitors (Cotar et al., 2010), pathogen displacement (Ruas-Madiedo et al., 2006), and virulence attenuation (Tanner et al., 2016). Bifidobacteria also regulate host functions and ferment complex polysaccharides from our diet (Ménard et al., 2005; Heuvelin et al., 2009, 2010; Bermudez-Brito et al., 2013; Furusawa et al., 2015), although there is still limited knowledge on the molecular mechanisms triggering these effects.

Delineating the specific molecules mediating Bifidobacterium cross-talk with the host, will help to understand their beneficial effects and establish microbiome-targeted therapies for human diseases. This review gives an overview on molecules behind the bifidobacterial-host cross-talk, providing a framework to design safe and efficacious probiotic-derived supplements (Licciardi et al., 2010; Shenderov, 2013).

# INTESTINAL MUCOSA ADHESION

Bacterial adhesion to the intestinal surface is mediated by non-specific, hydrophobic or electrostatic interactions, and specific mechanisms involving macromolecular interactions between bacterial and host receptors. Functionally characterized Bifidobacterium adhesins, surface-associated proteins that facilitate bifidobacteria attachment to intestinal cells and/or the extracellular matrixes surrounding them, are reviewed below and summarized in **Figure 1**.

Involvement of pili in bifidobacterial adhesion has been studied in Bifidobacterium bifidum and Bifidobacterium breve (O'Connell Motherway et al., 2011b; Turroni et al., 2013). The B. bifidum PRL2010 genome harbors three pilus clusters. Of these, pil2 and pil3, encode putative sortase-dependent pili that are expressed under both in vitro and in vivo conditions. Heterologous expression of pil3 in Lactococcus lactis significantly increased this bacterium adhesion to the human epithelial cell line Caco-2 (Turroni et al., 2013). The type IVb tight adherence (Tad) pilus-encoding gene cluster from B. breve UCC2003, was found to be essential for the colonization of, and persistence in, the murine gut. Tad inactivation impaired the strain ability to stably colonize the murine intestine, as reflected by reduced shedding level and bifidobacterial numbers in the gut (O'Connell Motherway et al., 2011b). Another surfaceanchored protein potentially involved in intestinal adhesion of bifidobacteria is BopA. This purified lipoprotein competes with B. bifidum MIMBb75 adhesion to Caco-2 cells, and probably facilitates B. bifidum adhesion (Guglielmetti et al., 2008). Indeed, Bifidobacterium strains overexpressing bopA adhere better (Gleinser et al., 2012). However, B. bifidum treatment with anti-BopA antibodies does not reduce the attachment to intestinal cells (Kainulainen et al., 2013), thus the specific adhesion mechanism mediated through BopA must be further elucidated.

Remarkably, bifidobacteria can use some moonlighting proteins, those with multiple functions (Jeffery, 2003), as adhesin-like factors. Surface-exposed glycolytic enzymes, including transaldolase from B. bifidum and enolase from Bifidobacterium animalis, are adhere to mucin and plasminogen, respectively (Candela et al., 2009; González-Rodríguez et al., 2012). Other surface-exposed moonlighting proteins, including the chaperone DnaK from B. animalis and the elongation factor Tu from Bifidobacterium longum, showed high affinity for human plasminogen in vitro and have been proposed as mediators of intestinal attachment (Candela et al., 2010; Wei et al., 2014).

It is worth highlighting that most of the adhesins herein described have been identified on in vitro assays and their relevance for intestinal colonization has not been tested in vivo. Since laboratory models do not accurately mimic all the factors that can affect bacterial attachment to the intestinal mucosa (e.g., due to the absence of a mucus layer and resident microbiota), functional confirmation of the adhesion capacity in vivo is still required (Ouwehand and Salminen, 2003).

# TIGHT-JUNCTIONS

A single layer of epithelial cells provides a selective barrier separating intestinal lumen from subjacent tissues. Tightjunctions (TJ) are multi-protein complexes that control molecule translocation across this barrier (Lee, 2015), and their disruption leads to uncontrolled trafficking of noxious molecules triggering inflammation (Bergmann et al., 2013).

Specific Bifidobacterium strains promote TJ enhancing epithelial barrier integrity (Ohland and Macnaughton, 2010; Mokkala et al., 2016). In animal models, B. bifidum and B. longum strains preserved TJ localization, attenuating intestinal permeability, and decreasing necrotizing enterocolitis incidence (Khailova et al., 2009; Bergmann et al., 2013; Srutkova et al., 2015). Preliminary work suggested that B. bifidum metabolites, like acetate, induced TJ expression in intestinal cells (Hsieh et al., 2015). Soluble factors present in B. longum lysates or secreted by B. infantis or B. breve strains, also mediate epithelial barrier maintenance (Ménard et al., 2005; Ewaschuk et al., 2008; Sultana et al., 2013). Further research to identify the specific molecules mediating this TJ promotion is needed.

# MUCUS, HMO, AND NON-DIGESTIBLE CARBOHYDRATES DEGRADATION

The spatial distribution of bacteria throughout the gastrointestinal tract, is partly controlled by nutrients availability for resident microbiota (Donaldson et al., 2016). Indeed, our wellbeing relates with the nutrient harvesting capability of our gut microbes. These microorganisms, overall, can utilize dietary and host carbohydrates, and glycans produced by other gut bacteria. Indeed, 89 "carbohydrate active enzymes" (CAZyme) have recently been identified in 85% of the microbiomes obtained from 488 individuals (Bhattacharya et al., 2015), suggesting that gut bacteria are highly specialized in using available glycans as their main sustenance. Bifidobacterial genomes are abundant in saccharolytic features whose expression is tightly regulated by available carbohydrates (Khoroshkin et al., 2016), supporting that host glycans were a potent evolutionary force driving their successful gut colonization (Sánchez et al., 2013; Milani et al., 2016).

Numerous studies demonstrated bifidobacteria's capability to use dietary non-digestible oligosaccharides, which is on the basis of the prebiotic concept (Rastall and Gibson, 2015). Glycosyl hydrolases (GH, or glycosidades), many of which are extracellular, have high specificity for the oligosaccharides constituents and cleave the glycosidic bonds (**Table 1**). Special attention has been paid to the GH acting on human milk oligosaccharides (HMO) since these serve as substrates for bifidobacteria, which are the initial gut colonizers in breast-fed infants. HMO are structurally diverse and composed of several monosaccharides (glucose, galactose, N-acetylglucosamine, fucose, or sialic acid). They mainly consist of a lactose core linked to units (n = 0–15) of lacto-N-biose (type I) or to N-acetyllactosamine (type II; Smilowitz et al., 2014). Bifidobacteria secrete GH that cleave specific linkages within the HMO molecules and the best characterized are those synthesized by B. bifidum which, together with B. longum subsp. infantis, are two abundant species in breast-fed neonates (**Table 1**). These species employ different strategies for HMO utilization. Whereas B. bifidum has an array of membrane-associated GH, B. longum subsp. infantis is specialized in the import and intracellular breakdown of HMO (Garrido et al., 2013; Jae-Han et al., 2013). Moreover B. longum strains have similar HMO-utilization patterns, whilst B. bifidum strains are more diverse with some unable to use fucosylated or sialylated HMO (Garrido et al., 2015). Similarly, the B. breve HMO utilization profile is strain dependent and, contrary to B. bifidum, some strains consume fucosylated or sialilated HMOs. B. breve's capability to use these HMOs explains its abundance in breast-fed babies (Ruiz-Moyano et al., 2013).

Some bifidobacteria can also utilize mucins from the mucus layer coating the intestine. Mucin composition and structure resemble that of HMO; consisting of a core of different O-glycans, built on α- and β-linked N-acetyl-galactosamine, galactose, and N-acetyl-glucosamine residues, which can incorporate fucose and sialic acid residues (Tailford et al., 2015). B. longum and B. breve strains' capability to effectively use mucin carbohydrates, has been confirmed in vitro (Ruas-Madiedo et al., 2008). However, GH able to degrade mucins have only been described in B. bifidum (**Table 1**). Indeed, in a comparative genomic study 60% of the GH-encoding genes from B. bifidum were predicted to breakdown mucin-like glycans and most of them were exclusively present in this bifidobacterial species (Turroni et al., 2014). Remarkably, other species could use the monoand oligosaccharides released by B. bifidum GH thus evidencing the existence of cross-feeding mechanisms, as it has been demonstrated in B. breve and B. bifidum co-cultures (Egan et al., 2014).

Bifidobacterium capacity to metabolize specific dietary and host-derived carbohydrates is also dependent on the presence of specific sugar transport systems. These are crucial for their competitive establishment in the gut, thus representing one of the molecular mechanisms by which bifidobacteria interact with the host. Import sugar mechanisms in bifidobacteria are herein described (Bottacini et al., 2014).

First, ATP-binding cassette (ABC) systems are active transporters which couple ATP hydrolysis to translocation uptake across the cell membrane. They are the most frequent sugar transporters in bifidobacteria and have been described for mono- and oligosaccharides in different species (Nishimoto and Kitaoka, 2007; Wada et al., 2008; Wei et al., 2012), although only a few of them have been functionally characterized at protein level (Suzuki et al., 2008; Ejby et al., 2013).

Secondly, some secondary transporters, predicted to consist of single integral membrane-associated proteins, have been characterized at protein level in bifidobacteria. These include permease systems for the uptake of lactose, glucose, and sucrose (Parche et al., 2006). Secondary transporters encoding genes have been identified in different bifidobacterial species, although most of them have not been characterized at protein level (Turroni et al., 2012).

Proton symporters of the glycoside-pentoside-hexuronide (GPH) cation symporter family for melibiose and pentosides were also described in Bifidobacterium (Lee and O'Sullivan, 2010; Turroni et al., 2012), although they remain to be characterized.

Finally, phosphoenolpyruvate-phosphotransferase (PEP-PTS) systems were first characterized in the 90's in B. breve and B. bifidum at protein level (Lee and O'Sullivan, 2010). Later, genome sequence availability revealed their wide spread distribution in bifidobacteria. In particular, B. breve UCC2003 genome contains four PEP-PTS systems, one of which has been characterized as a fructose-specific transporter (Mazé et al., 2007). Also, in silico analysis revealed a putative glucose-specific PEP-PTS uptake system in B. longum (Lorca et al., 2007). However, genome analysis of different B. longum strains showed that glucosespecific PTS transporters are minor in comparison with ABC transporters (Pokusaeva et al., 2011), thus glucose may be transported preferentially by secondary permeases (Parche et al., 2006).

Comparative genomic analysis revealed that sugar PEP-PTS systems are present in all bifidobacterial genomes, except for B. animalis subsp. lactis (Lee and O'Sullivan, 2010) which is hypothesized to have lost most of their carbohydrate transporters due to extended cultivation under industrial conditions.. In fact, the capability to utilize variable carbon sources is considered an adaptation to the gut environment. For instance, the dominant Bifidobacterium species in infant fecal samples (B. longum and B. bifidum) is consistent with their inherent ability to use hostderived oligosaccharides such as mucin and HMO (Bottacini et al., 2014), and their possession of a wide range of host-derived carbohydrate transporters, such as those involved in N-biose import (Suzuki et al., 2008).

# REGULATION T-EFFECTOR CELLS AND Treg

In the absence of disease, the ensemble of molecular interactions taking place in the human gut results in

#### TABLE 1 | Gycosyl hydrolases and sugar transporters characterized and/or described in Bifidobacterium genus.

#### Glycosyl hydrolases


(Continued)

#### TABLE 1 | Continued


<sup>a</sup>OS, oligosaccharide(s).

Human carbohydrates: mucin and HMO.

PTS, phosphotransferase system.

the intestinal homeostasis. Specialized epithelial cells denominated M-cells and antigen presenting cells (APCs) from the gut-associated lymphoid tissue (GALT) continuously sample the intestinal content. Interaction of APCs with the rest of GALT effectors, mainly T and B cells, leads to immunotolerance against commensal microbes and dietary components, whilst the capacity of mounting an acute, quick, and powerful response against enteropathogens is developed.

Differentiation of commensal and pathogenic bacteria is based on the presence of pattern recognition receptors (PRR) on the APC and epithelial cell surfaces. Among them, Tolllike receptors, NOD-like receptors, C-type lectin receptors, and RIG-I-like receptors are in charge of recognizing specific microbial-associated molecular patterns (MAMPs), such as flagellin, teichoic acids, or lipopolysaccharide among others. The type and intensity of the downstream and intracellular signaling cascades deployed after MAMPs-PRR interaction is essential for the APCs interaction with T-cells, which will finally determine the nature of the T-cell response. Roughly, T-cell responses are divided into effector (Th) and regulatory (Treg), its balance being key in the intestinal homeostasis maintenance (Maloy and Powrie, 2011). It is generally accepted that commensal microbiota, by inducing Treg response, modulates the Th1/Th2 balance favoring immune tolerance against the gut microbiota (Ventura et al., 2012). Indeed, the classical MAMP triggering Treg response is the exopolysaccharide A of the commensal bacterium Bacteroides fragilis, molecule also involved in the GALT maturation (Mazmanian et al., 2008).

Bifidobacteria may drive species-specific T-cell responses, as it was revealed by a series of experiments in which the cytokine secretion profiles of monocyte-derived dendritic cells (MoDCs) and full fractions of peripheral blood mononuclear cells (PBMCs) were determined (López et al., 2010). Relative levels of key cytokines (IL-10, IL-17, TNFα among others) suggested a specific immunomodulation mechanism for each species, as reported recently for probiotics (Hill et al., 2014). Challenging immature MoDCs with different strains, followed by co-culture with allogeneic naïve CD4<sup>+</sup> cells and cytokine determination, further confirmed this effect (López et al., 2011).

Remarkably, B. bifidum LMG13195 appeared to induce a Treg response in vitro (**Figure 1**; López et al., 2011). Dendritic cells challenged with membrane vesicles from this strain induced naïve CD4<sup>+</sup> cells polarization into Treg, as deduced from the increases in the expression of foxP3 regulation factor and the CD25 marker (López et al., 2012). Most likely, surface-associated proteins play a role in this process. Several proteins have been identified in the bifidobacterial membrane, among which moonlighting proteins such as fructose-6-phosphate phosphoketolase or enolase, might be behind the immunomodulatory effects of the membrane vesicles (Sánchez et al., 2004). However, the particular proteins involved in this T-cell polarization have not been identified.

Other immunogenic extracellular proteins are pili, proteinaceous structures that self-assemble into filaments on the bacterial surface (Ventura et al., 2012). Specifically, one sortase-dependent pili from B. bifidum PRL2010 induced TNFα production during transient colonization of the murine mucosa, which acted as a macrophage-activating factor during Th1 (Turroni et al., 2013). Another surface-protein able to influence T-cell responses is TgaA from B. bifidum, a peptidoglycan-derived enzyme able to induce DC activation and IL-2 production (Guglielmetti et al., 2014). IL-2 is one of the main cytokines supporting Treg proliferation, which are characterized by the presence of CD25, the T-cell receptor for that interleukin (Zelante et al., 2012). Despite the evidence supporting an immunomodulation role of bifidobacteria, only a few surface-associated proteins have been identified as possible mediators of this effect on a limited number of in vitro experiments. Identifying the molecules behind this effect and confirming their efficacy in clinical trials, might provide keys to ameliorate diseases characterized by exacerbated immune responses.

# INDUCTION OF IgA PRODUCTION

Immunoglobulin A (IgA) is the most abundant antibody in human mucosa and modulates immune responses against commensal bacteria, preventing direct contact with immune cells (Peterson et al., 2007; Brandtzaeg, 2013). Globally, 40% of gut bacteria are IgA-coated although these values are species- and strain-dependent (Talja et al., 2014). In healthy individuals, IgA coating of bifidobacteria is higher than that of other commensals (van der Waaij et al., 2004; De Palma et al., 2010), explaining the immune tolerance to high densities of bifidobacteria. In fact, 44 proteins from B. longum and 24 from B. adolescentis were recognized by IgA (Talja et al., 2014). IgA-coated bifidobacteria also enhanced probiotic attachment to Caco-2 cells and increased production of mucosal defense molecules (Mathias et al., 2010).

Levels of IgA-coated gut commensals are altered in dysbiosis states such as those described in coeliac disease (De Palma et al., 2010), inflammatory bowel disease (van der Waaij et al., 2004), or autoimmunity disorders (Talja et al., 2014). Coeliac children showed reduced levels of bifidobacteria and IgA-coated bacteria (De Palma et al., 2010). Conversely, IgA from children developing islet autoimmunity, bound to more B. adolescentis antigens than those from healthy controls (Talja et al., 2014). Morevoer, bifidobacterial supplements modulate IgA production (Holscher et al., 2012; Kandasamy et al., 2014). A probiotic mixture containing bifidobacteria increased IgA and reduced diarrhea following rotavirus vaccination in a gnotobiotic pig model (Kandasamy et al., 2014). B. animalis Bb12 supplementation to formula-fed infants increased IgA in feces and, in those delivered by C-section, enhanced immune responses as reflected by higher anti-rotavirus and anti-poliovirus IgA production following vaccination (Holscher et al., 2012). The bifidogenic effect of galactooligosaccharides also correlated to increased IgA production (Vulevic et al., 2013; Paineau et al., 2014). Further, research to specifically delineate the bifidobacterial molecules mediating IgA induction and interaction is necessary.

#### OTHER BIFIDOBACTERIUM EFFECTORS OF THE HOST-MICROBE DIALOGUE

A few extracellular proteins, with important physiological roles not discussed in the previous sections deserve further attention. Some Bifidobacterium strains produce surface-exposed Serine Protease Inhibitors of proteinaceous nature (serpins), which participate in a variety of physiological processes. The serpin produced by B. longum NCC2705 inhibits elastase-like proteases, including neutrophil or pancreatic elastases, thus suggesting a role in protecting bifidobacteria against exogenous proteases and potential anti-inflammatory activity (Ivanov et al., 2006). Remarkably, serpins are widely distributed in bifidobacteria and several species harbor serpin-encoding genes in their genomes (Turroni et al., 2010).

# CONCLUSIONS

Some Bifidobacterium proteins have been identified as mediators of the cross-talk bifidobacteria-host, providing bases to understand their beneficial traits and opening new avenues to conceive bifidobacterial-based therapeutic strategies. However,

# REFERENCES


in most cases, the molecular mechanisms triggered remain unknown what has limited their translation into improved functional supplements. Identifying targets for intervention at intestinal level and developing appropriate models to search for bifidobacterial mediators is required to delineate strategies that fine-tune disease-associated alterations in the microbiota-host interplay.

#### AUTHOR CONTRIBUTIONS

LR, SD, PM, AM, and BS contributed to the design and organization of the manuscript, drafted, reviewed, and accepted the final version of the manuscript.

# ACKNOWLEDGMENTS

This research was funded by Grant AGL2013-44039-R from the Spanish "Plan Estatal de I+D+I." LR has received funding from the People Programme (Marie Curie Actions) of the European Union's Seventh Framework Programme FP7/2007- 2013/ under REA grant agreement n◦ 624773. BS was recipient of a Ramón y Cajal postdoctoral contract from the Spanish Ministry of Economy and Competitiveness. SD is hired on a contract supported by BIO2014-55019-JIN project from the Spanish "Plan Estatal de I+D+I."


modulates dendritic cell maturation through its cysteine- and histidinedependent amidohydrolase/peptidase (CHAP) amidase domain. Appl. Environ. Microbiol. 80, 5170–5177. doi: 10.1128/AEM.00761-14


model of necrotizing enterocolitis. Am. J. Physiol. Gastrointest. Liver Physiol. 297, G940–G949. doi: 10.1152/ajpgi.00141.2009


Bifidobacterium longum specific for core 1 disaccharide and lacto-N-biose I. J. Biol. Chem. 283, 13165–13173. doi: 10.1074/jbc.M709777200


sources and a model for their utilization. J. Agric. Food Chem. 61, 4360–4370. doi: 10.1021/jf400792j

**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 Ruiz, Delgado, Ruas-Madiedo, Margolles and Sánchez. 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.