Edited by: Peter M. Van Endert, Institut national de la santé et de la recherche médicale, France
Reviewed by: James Drake, Albany Medical College, United States; Shouxiong Huang, University of Cincinnati, United States
Specialty section: This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology
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.
Human leukocyte antigen (HLA)-DR15 is a haplotype associated with multiple sclerosis. It contains the two DRB* genes
Human leukocyte antigen (HLA)-DP, -DQ, and -DR are membrane heterodimeric glycoproteins, composed by two chains, α and β, encoded in the class II region of the human major histocompatibility complex (MHC) or HLA. Thus, four different HLA-DP and HLA-DQ molecules are present in heterozygous cells, as each polymorphic α chain can interact with either polymorphic β chain. In the case of HLA-DR, the gene encoding the HLA-DRα subunit is dimorphic (and both forms are functionally equivalent). As such, only two different heterodimers are usually expressed. However, some haplotypes include two functional
It has been known for a long time that different HLA-DR molecules can present common peptides, indicating that HLA-DR molecules are to a certain level promiscuous. In fact, in the HLA class II, antigen processing pathway CLIP, derived from the invariant chain (Ii), must bind essentially all HLA-II molecules for a correct peptide selection. In addition, many other promiscuous binders have been described so far (
Nevertheless, although the promiscuity of HLA-DR molecules is accepted, the degree of overlap between different peptide repertoires has not been extensively addressed. In this regard, we compared the peptide repertoires of four HLA-DR allotypes differentially associated with rheumatoid arthritis and found a low degree of promiscuity in their bound peptidomes (
Human leukocyte antigen-DR15, a haplotype that expresses two functional HLA-DRB genes, is associated with multiple sclerosis (MS). MS is a chronic inflammatory disease of the central nervous system and is considered a T cell-mediated autoimmune disorder with a prevalence of 0.5–1.5 per 1,000 inhabitants in the northern hemisphere [reviewed in Ref. (
The co-expression of both DR2a and DR2b at the RNA level and on the cell surface has been demonstrated in different cell types (
In this study, we have addressed three main questions: first, the fine-mapping of the anchor motifs of DR2a and DR2b by the identification of natural ligands bound to these allotypes; second, the degree of overlap of the peptide repertoires bound to these two HLA-DR molecules (DR2a and DR2b); and third, the contribution of DR2a and DR2b to the peptide repertoire presented on the cell surface. To that end, we used liquid chromatography couple to tandem mass spectrometry (LC–MS/MS) to characterize the peptide repertoires associated with HLA-DRB1*15:01 (555 unique peptides) and HLA-DRB5*01:01 (169 unique peptides) from bare lymphocyte syndrome (BLS) transfected cells. This analysis allowed us to refine the binding motifs of these allotypes, identifying some novel anchor residues. The analysis of these peptide pools indicated that, although these molecules can share some peptides ligands, the overlap of both peptidomes was extremely low. Finally, our estimations indicate that both allotypes contribute similarly to the peptide repertoire presented by HLA-DR15 to CD4+ T cells.
The BLS patient-derived B cell line (BLS) transfected with the genes encoding the molecules DRB1*15:01 (BLS-DR2b) or DRB5*01:01 (BLS-DR2a) were kindly provided by G. Nepom and W. Kwok (University of Washington, Seattle, WA, USA) and were used as the source of the HLA-DR peptide ligands.
For the purification of the peptide pools associated with DRB1*15:01 or DRB5*01:01 the monoclonal antibody B8.11.2 (IgG2b) was used. This antibody recognizes a DR framework structure present on all types of HLA-DR molecules (
Peptides were purified as previously described (
Samples were analyzed in a nano-LC ultra HPLC (Eksigent) coupled online with a 5,600 triple TOF mass spectrometer (AB Sciex) and equipped with a C18 chromXP trapping column (350 µm × 0.5 mm, 3 µm, Eksigent) and a C18 chromXP column (75 µm × 150 mm, 3 µm, Eksigent). Solvent A and B were 0.1% formic acid and 0.1% formic acid in acetonitrile, respectively. Peptides were fractionated at a flow-rate of 300 nl/min at 40°C under gradient elution conditions, as follows: isocratic conditions of 2% B for 1 min, a linear increase to 30% B in 181 min, a linear increase to 40% B in 23 min, a linear increase to 90% B in 15 min, 90% B for 10 min, and back to initial conditions. Total runtime was 250 min. For blank injections, a shorter gradient was employed consisting of isocratic conditions of 5% B for 1 min, a linear increase to 30% B in 109 min, a linear increase to 40% B in 10 min, a linear increase to 90% B in 5 min, 90% B for 5 min, and back to initial conditions. In this latter case, total runtime was 150 min. Each acquisition cycle comprised a survey scan (350–1,250
Raw MS/MS data were converted to mgf files with Peakview 1.2 (AB SCiex) and searched against a concatenated target-decoy database containing the 88,669 Uniprot entries of the Homo sapiens complete proteome set (as of March 2015) and their corresponding reverse sequences. The mgf file corresponding to the DR2b fraction was recalibrated with Protein Pilot (version 4.5, AB Sciex) before the search. MASCOT (Matrix Science, version 2.5) was used as search engine with the following parameters: no enzyme, MS tolerance of 15 ppm, MS/MS tolerance of 0.025 Da and protein N-terminal acetylation, pyroglutamic acid formation from glutamine, and methionine oxidation as variable modifications. Estimation of the false discovery rate (FDR) was carried out by decoy hit counting as previously described (
All the peptides associated with DR2a or DR2b were analyzed with NetMHCIIpan 3.1 Software (
Modeling of complexes between HLA-DR2a and HLA-DR2b and the different peptides was performed using a simulation protocol detailed elsewhere (
The experimental affinity of different peptides to HLA-DR molecules was determined as previously described (
BLS-DR2a and BLS-DR2b cells were used as the source of the peptide–HLA-DR complexes. After purification by immunoaffinity chromatography, the trimeric complexes were denatured under acidic conditions and their associated peptides purified by ultra-filtration. The resulting peptide mixtures were analyzed by LC–MS/MS and peptide identification was carried out using Mascot as search engine. Peptides with a length of 11 amino acids or longer were considered for the analysis. A total of 177 and 560 peptide ligands were identified from the DR2a and DR2b molecules, respectively. Peptides derived from some heterogeneous nuclear ribonucleoproteins were considered as background contaminants on the basis of their particular features (their sequences were notably rich in Gly and Pro) and of our previous observations [similar peptides had been found in previous analysis of other unrelated HLA-II and HLA-I molecules (
The size distribution of the peptide ligands sequenced from the DR2 allotypes followed a normal distribution with an average molecular weight of 1,977.9 and 1,761.0 Da (Figure
Peptide size distribution of the peptides sequenced from the different human leukocyte antigen-DR allotypes.
Most of the HLA-II ligands were peptides derived from proteins located in the endocytic pathway, although some peptides came from cytosolic or nuclear proteins. To test if the DR2a- and DR2b-derived peptidomes were canonical regarding the subcellular location of their source proteins, we analyzed this parameter and found that most of them were located in vesicular compartments (Figure
Major cell location of the parental proteins of the peptides sequenced from the different human leukocyte antigen-DR molecules.
The HLA-DRB1*15:01 and HLA-DRB5*01:01 binding motifs were described years ago (
Refined anchor motifs for DR2a and DR2b molecules.
Since the peptide-binding motifs of DR2a and DR2b showed some common features, we set out to analyze the degree of overlap of the two peptide pools. In our dataset, only 13 peptides were identified from both allotypes (7.7% for DR2a and 2.3% for DR2b) (Table
Shared peptides between DR2a and DR2b.
# | Protein | Peptide |
---|---|---|
1 | Cofilin-1 | ASGVAVSDGVIKVFNDMKVR |
2 | Cofilin-1 | ASGVAVSDGVIKVFNDMKVRK |
3 | Integrin beta | NIQPIFAVTSRMVKTYE |
4 | Invariant chain | KPPKPVSKMRMATPLLMQA |
5 | Invariant chain | KPPKPVSKMRMATPLLMQALP |
6 | Invariant chain | LPKPPKPVSKM |
7 | Invariant chain | LPKPPKPVSKMRMATPLLMQAL |
8 | Invariant chain | LPKPPKPVSKMRMATPLLMQALP |
9 | Invariant chain | LPKPPKPVSKMRMATPLLMQALPM |
10 | Nuclease-sensitive element-binding protein 1 | PPAENSSAPEAEQGGAE |
11 | Protein CutA | PALLPVASRLLLLP |
12 | SWI/SNF complex subunit SMARCC2 | PGTPLPPDPTAPSPGTVTPVPPPQ |
13 | SWI/SNF complex subunit SMARCC2 | PTAPSPGTVTPVPPPQ |
We recently sequenced about 200 HLA-II-bound peptides from the DR2-homozygous B cell lymphoblastoid cell line (B-LCL) MGAR (
Common peptides from MGAR and BLS-DR2a or BLS-DR2b.
PEPTIDE | SEQUENCED FROM | NetMHCIIPan 3.1 | Matrix of this paper | BINDING [IC50 (nM)] | MODELING (ΔG, kJ/mol) | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
DR2a | DR2b | MGAR | CORE DR2a | IC 50 (nM) DR2a | CORE DR2b | IC 50 (nM) DR2b | Core DR2a | Score DR2a | Core DR2b | Score DR2b | DR2a | DR2b | CORE |
DR2a | DR2b | |
X | X | 500.20 | 13.411 | >50 μM | >50 μM | −41 | ||||||||||
X | X | |||||||||||||||
X | X | 564.48 | 15.037 | ND | ND | −58 | ||||||||||
X | X | 148.05 | 13.674 | ND | ND | −50 | ||||||||||
X | X | 1,108.20 | 12.652 | – | 14.5 μM | −52 | ||||||||||
X | X | |||||||||||||||
X | X | 1,365.31 | 13.268 | ND | ND | −35 | ||||||||||
−80 | ||||||||||||||||
X | X | 181.61 | 13.245 | ND | ND | −82 | ||||||||||
X | X | 577.16 | 9.231 | >50 μM | >50 μM | −42 | ||||||||||
X | X | 9.231 | ||||||||||||||
X | X | 10.604 | ||||||||||||||
X | X | 10.604 | ||||||||||||||
X | X | 660.48 | 12.470 | ND | ND | −61 | ||||||||||
X | X | 3,407.84 | 11.077 | ND | ND | −77 | ||||||||||
X | X | 3,028.33 | 17.163 | ND | ND | −58 | ||||||||||
X | X | 2,413.08 | 9.480 | 2.4 μM | >50 μM | L |
−107 | |||||||||
X | X | 10.511 | ||||||||||||||
X | X | 1,690.71 | 16.488 | ND | ND | −47 | ||||||||||
X | X | 62.71 | 13.238 | 0.08 μM | >50 μM | F |
−56 | |||||||||
X | X | |||||||||||||||
X | X | 5,881.78 | 10.940 | 5.8 μM | >50 μM | F |
−82 | |||||||||
X | X | 35.95 | 18.858 | ND | ND | −138 |
To analyze the contribution of the peptides eluted from DR2a and DR2b to the global peptidome presented by the haplotype DR15, we analyzed the peptide pool sequenced in the present work and in a previous report from the B-LCL, MGAR (
This same approach was conducted with the peptide pool isolated from the MGAR cell line. Using that dataset and the here defined score, we estimate that 51.1% of the peptides displayed by DR15 are presented by DR2a and the remaining 48.9% by DR2b (Table S7 in Supplementary Material). Therefore, our data suggest that, in the DR15 haplotype, the contribution of both HLA-DR molecules to the conformation of the HLA-DR15 peptide repertoire is similar, at least in quantitative terms.
To gain insight into the molecular features that govern the interaction of DR15 with its peptide ligands we considered the 8 and 14 peptides associated, respectively, with DR2a and DR2b that we described in a previous report (
We calculated their theoretical binding affinities with NetMHCIIpan 3.1. All the sequences tested showed a high affinity (IC50 < 1,000 nM) except the peptide TPKIQVYSRHP, derived from β2-microglobulin, which probably is not a real ligand, but a contaminant. Of note, all the peptides presented higher affinity for the allotype from which they were sequenced, with the exceptions of the nested set LEEFGRFASFEAQG(A) and the peptide TPKIQVYSRHP, which were identified from the BLS-DR2b cell line but showed a higher theoretical affinity for DR2a (Table
To get an experimental measure of the binding affinities of some natural ligands of DR15, we synthesized the peptides ELEELRAEQQRLKSQDL, QTKEFQVLKSLGKLAMG, and SQAEFEKAAEEVRHL (sequenced from BLS-DR2a and the homozygous B-LCL MGAR) and the peptides LEEFGRFASFEAQG, LPSEKAIFLFVDKTVPQS, and QKKEIHLYQTFVVQ (sequenced from BLS-DR2b and MGAR). These were used in a binding assay to determine their binding affinity to DR2a and DR2b. The three peptides sequenced from DR2a showed a high affinity for DR2a ranging from 0.08 to 5.8 µM, while they failed to bind to DR2b (Figure
Experimental binding affinity of peptides sequenced from MGAR and
Computational models of the complexes of the putative cores obtained from the NetMHCIIpan 3.1 Server and some additional cores of these peptides with both DR2a and DR2b were made. The simulation protocol has been previously described (
Table
Figure
Model structures of the peptides IVIFQSKPE (eluted from DR2b) and FEKAAEEVR (eluted from DR2a) complexed with DR2a or DR2b. Computed binding scores for IVIFQSKPE are −58 kJ/mol to DR2a and −150 kJ/mol to DR2b. Computed binding scores for FEKAAEEVR are −112 kJ/mol to DR2a and −82 kJ/mol to DR2b.
The HLA is polygenic, containing several different HLA genes, which encode classical HLA molecules, which are highly polymorphic and show codominant expression. These features are relevant to the function of the proteins they encode, that is, the presentation on the cell surface of peptides derived from aberrant or non-self proteins to specific T lymphocytes. In the case of HLA-II heterodimers, both chains conform the peptide-binding site. Thus, the combination of two polymorphic α chains with two polymorphic β chains, as occurs in HLA-DP and -DQ proteins, increases the theoretical capacity of peptide presentation. The fact that HLA-DRα is not polymorphic greatly reduces the diversity of the displayed peptide repertoire. However, in some haplotypes, this is compensated with the expression of two different HLA-DRB loci. The selection of these complex haplotypes may indicate that the presentation of different peptides derived from pathogens to CD4+ T lymphocytes constitutes a selective advantage. In addition, the presence of two different functional genes could be the result of a gene duplication that is not a disadvantage (but not necessarily an advantage) and, thus, has not been removed from the genome.
Human leukocyte antigen-DR15 is one of the haplotypes which harbors two functional loci,
Interestingly, P8 showed an increased frequency of positively charged residues (Lys in DR2a and Lys, His and Arg in DR2b). In this regard, we made a similar observation in DRB1*01:01, DRB1*04:01, and DRB1*10:01 (
A relevant aspect that we also evaluated in this study is the promiscuity of peptide binding in HLA-II molecules. It is well-known that some peptides can bind to many HLA-DR molecules. In the case of HLA-DR15, the anchor motifs of DR2a and DR2b are different, both allotypes can accommodate similar or identical residues at major anchor positions. Thus, although with different prevalence, both molecules can accept aliphatic or aromatic residues in P1 and P4, polar residues in P6 and aliphatic and basic residues in P9. Nevertheless, our data show that, although some particular peptides can be presented by both molecules, the global degree of overlap is low. These data agree with our previous finding that, the overlap between the peptide repertoires associated with DRB1*01:01, DRB1*04:01, DRB1*10:01, and DR15 is low, even between the most similar DR1 and DR10 repertoires (
Among the peptides identified in this study, only 13 peptides were common to DR2a and DR2b. Six of them derived from the invariant chain, mirroring the high abundance of Ii-derived ligands in this cell line, a phenomenon particularly evident in DR2a. The seven remaining common peptides derived from five different proteins, two of them located in the plasma membrane, two in the nucleus, and one in the cytosol. The three peptides derived from nuclear or cytosolic proteins presented a low theoretical affinity calculated either with NetMHCIIpan 3.1 (Tables S3 and S4 in Supplementary Material) or with the matrix generated in this work (Tables S5 and S6 in Supplementary Material). Thus, we cannot discard that some of the sequenced peptides are contaminants. Nevertheless, this will not be the case for most of them as: (1) the anchor motif that we observe is similar to that described previously and (2) the subcellular locations of their parental proteins is the expected for the identified ligands. However, considering that background contaminants are more likely to be detected in both cell lines as they are allele independent, the actual overlap could be even lower than reported here.
The low degree of overlap suggests that, at least in the case of DR15, the existence of two DR alleles can increase the diversity of the peptide repertoire presented to T lymphocytes. Moreover, since some peptides can be presented in different registers, as demonstrated previously for the peptide spanning residues 86–105 of the myelin basic protein (
The modeled structures of the peptide ligands complexed with DR2a and DR2b can be rationalized in light of the anchor motifs obtained in this work. Thus, the binding motifs for both alleles show that selection is primarily done at the level of P1, P4, and P9. P1 binds preferentially large aromatic residues in DR2a and smaller aliphatic residues (preferentially Ile) in DR2b. P4 shows the opposite trend, with small aliphatic residues in DR2a and aromatic residues in DR2b. Finally, P9 shows a preference for basic residues in DR2a and aliphatic residues in DR2b. These preferences can be broadly explained by the sequence differences between DR2a and DR2b. Thus, the dimorphism G86V explains the preference of each allotype in P1. As seen in Figure
In this work, we have used an LC–MS/MS-based peptidomic approach to refine the binding model of two HLA-DR allotypes associated with MS. To our knowledge, this is the first time that the peptide repertoires of two HLA-DR molecules belonging to the same haplotype are directly compared. In this context, our data indicate that the overlap in their bound peptidomes is very low. In addition, our data suggest that, quantitatively both molecules contribute to a similar extent to the configuration of the peptide pool displayed by HLA-DR15 on the cell surface. Finally, although the peptides sequenced here do not have special relevance to MS, as they do not come from known autoantigens, we propose that the analysis of peptide ligands eluted from HLA-DR molecules shows to be complementary to bioinformatics tools and can contribute to improve the prediction of new T cell epitopes relevant for autoimmune diseases.
ES contributed with drafting the work and the acquisition, analysis, and interpretation of the data. MM contributed with the acquisition of mass spectrometry data. XD contributed with the computer models. DA-L and EJ contributed with the acquisition and interpretation of binding assays. IA contributed with drafting the work, interpretation of the data, revising, and final approval of the version published.
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 authors thank Annabel Segura for technical help.
The Supplementary Material for this article can be found online at
SDS-PAGE gel of immunoprecipitated HLA-DR2a and -DR2b molecules. Peptide–HLA-DR complexes were eluted in acid medium and passed through a Centricon-10 device. The retained material was loaded on a 12% SDS-PAGE gel. Arrows show α and β HLA-DR chains.
Total ion chromatogram and fragmentation spectra of peptides sequenced from DR2a and DR2b molecules. Pages 1, 2, and 6 show the total ion chromatograms (TICs) of a blank, DR2a and DR2b samples, respectively. Pages 3 and 7 show the extracted ion chromatograms (XICs) of signals 790.9301 (from DR2a) and 746.0306 (from DR2a), respectively. Page 4 shows the fragmentation spectrum of the peptide TPLLMQALPMGALPQ, and page 5 shows the annotation of the corresponding fragments (black fonts correspond to the theoretical masses generated during the peptide fragmentation; red fonts correspond to the masses found in the fragmentation spectrum). Page 8 shows the fragmentation spectrum of the peptide GLQADLSSFKSQELNERNEA, and page 9 shows the annotation of the corresponding fragments (black fonts correspond to the theoretical masses generated during the peptide fragmentation; red fonts correspond to the masses found in the fragmentation spectrum).
Model structures of DR15 ligands complexed with DR2a and DR2b. The interactions of the identified binding cores from peptides sequenced from MGAR (DR15) and from BLS-DR2a or -DR2b and DR2a and DR2b were modeled. Figures show the graphical representation of the models most favorable energetically.