Accurate prediction of HLA class II antigen presentation across all loci using tailored data acquisition and refined machine learning.

Accurate prediction of antigen presentation by human leukocyte antigen (HLA) class II molecules is crucial for rational development of immunotherapies and vaccines targeting CD4+ T cell activation. So far, most prediction methods for HLA class II antigen presentation have focused on HLA-DR because of limited availability of immunopeptidomics data for HLA-DQ and HLA-DP while not taking into account alternative peptide binding modes. We present an update to the NetMHCIIpan prediction method, which closes the performance gap between all three HLA class II loci. We accomplish this by first integrating large immunopeptidomics datasets describing the HLA class II specificity space across all loci using a refined machine learning framework that accommodates inverted peptide binders. Next, we apply targeted immunopeptidomics assays to generate data that covers additional HLA-DP specificities. The final method, NetMHCIIpan-4.3, achieves high accuracy and molecular coverage across all HLA class II allotypes.

MHC2AffyPred: A machine-learning approach to estimate affinity of MHC class II peptides based on structural interaction fingerprints.

Understanding how MHC class II (MHC-II) binding peptides with differing lengths exhibit specific interaction at the core and extended sites within the large MHC-II pocket is a very important aspect of immunological research for designing peptides. Certain efforts were made to generate peptide conformations amenable for MHC-II binding and calculate the binding energy of such complex formation but not directed toward developing a relationship between the peptide conformation in MHC-II structures and the binding affinity (BA) (IC50 ). We present here a machine-learning approach to calculate the BA of the peptides within the MHC-II pocket for HLA-DRA1, HLA-DRB1, HLA-DP, and HLA-DQ allotypes. Instead of generating ensembles of peptide conformations conventionally, the biased mode of conformations was created by considering the peptides in the crystal structures of pMHC-II complexes as the templates, followed by site-directed peptide docking. The structural interaction fingerprints generated from such docked pMHC-II structures along with the Moran autocorrelation descriptors were trained using a random forest regressor specific to each MHC-II peptide lengths (9-19). The entire workflow is automated using Linux shell and Perl scripts to promote the utilization of MHC2AffyPred program to any characterized MHC-II allotypes and is made for free access at https://github.com/SiddhiJani/MHC2AffyPred. The MHC2AffyPred attained better performance (correlation coefficient [CC] of .612-.898) than MHCII3D (.03-.594) and NetMHCIIpan-3.2 (.289-.692) programs in the HLA-DRA1, HLA-DRB1 types. Similarly, the MHC2AffyPred program achieved CC between .91 and .98 for HLA-DP and HLA-DQ peptides (13-mer to 17-mer). Further, a case study on MHC-II binding 15-mer peptides of severe acute respiratory syndrome coronavirus-2 showed very close competency in computing the IC50 values compared to the sequence-based NetMHCIIpan v3.2 and v4.0 programs with a correlation of .998 and .570, respectively.

Peptides identified on monocyte-derived dendritic cells: a marker for clinical immunogenicity to FVIII products.

The immunogenicity of protein therapeutics is an important safety and efficacy concern during drug development and regulation. Strategies to identify individuals and subpopulations at risk for an undesirable immune response represent an important unmet need. The major histocompatibility complex (MHC)-associated peptide proteomics (MAPPs) assay directly identifies the presence of peptides derived from a specific protein therapeutic on a donor's MHC class II (MHC-II) proteins. We applied this technique to address several questions related to the use of factor VIII (FVIII) replacement therapy in the treatment of hemophilia A (HA). Although >12 FVIII therapeutics are marketed, most fall into 3 categories: (i) human plasma-derived FVIII (pdFVIII), (ii) full-length (FL)-recombinant FVIII (rFVIII; FL-rFVIII), and (iii) B-domain-deleted rFVIII. Here, we investigated whether there are differences between the FVIII peptides found on the MHC-II proteins of the same individual when incubated with these 3 classes. Based on several observational studies and a prospective, randomized, clinical trial showing that the originally approved rFVIII products may be more immunogenic than the pdFVIII products containing von Willebrand factor (VWF) in molar excess, it has been hypothesized that the pdFVIII molecules yield/present fewer peptides (ie, potential T-cell epitopes). We have experimentally tested this hypothesis and found that dendritic cells from HA patients and healthy donors present fewer FVIII peptides when administered pdFVIII vs FL-rFVIII, despite both containing the same molar VWF excess. Our results support the hypothesis that synthesis of pdFVIII under physiological conditions could result in reduced heterogeneity and/or subtle differences in structure/conformation which, in turn, may result in reduced FVIII proteolytic processing relative to FL-rFVIII.

The Utility of Supertype Clustering in Prediction for Class II MHC-Peptide Binding.

MOTIVATION: Extensive efforts have been devoted to understanding the antigenic peptides binding to MHC class I and II molecules since they play a fundamental role in controlling immune responses and due their involvement in vaccination, transplantation, and autoimmunity. The genes coding for the MHC molecules are highly polymorphic, and it is difficult to build computational models for MHC molecules with few know binders. On the other hand, previous studies demonstrated that some MHC molecules share overlapping peptide binding repertoires and attempted to group them into supertypes. Herein, we present a framework of the utility of supertype clustering to gain more information about the data to improve the prediction accuracy of class II MHC-peptide binding. RESULTS: We developed a new method, called superMHC, for class II MHC-peptide binding prediction, including three MHC isotypes of HLA-DR, HLA-DP, and HLA-DQ, by using supertype clustering in conjunction with RLS regression. The supertypes were identified by using a novel repertoire dissimilarity index to quantify the difference in MHC binding specificities. The superMHC method achieves the state-of-the-art performance and is demonstrated to predict binding affinities to a series of MHC molecules with few binders accurately. These results have implications for understanding receptor-ligand interactions involved in MHC-peptide binding.

Proteochemometrics-Based Prediction of Peptide Binding to HLA-DP Proteins.

Human leukocyte antigens (HLA) class II proteins are involved in the antigen processing in the antigen presenting cells. They form complexes with antigen peptide fragments. The peptide-HLA protein complexes are presented on the cell surface where they are recognized by helper T cells (Th cells). HLA-DP is one of the three HLA class II loci. The HLA-DP proteins are associated with a significant number of autoimmune diseases, as well as with a susceptibility or resistance to a number of infectious agents. In the present study, we apply proteochemometrics-a method for bioactivity modeling of multiple ligands binding to multiple target proteins-to derive and validate a robust model for peptide binding prediction to the 7 most frequent HLA-DP proteins. The model is able to identify 86% of the binders in the top 10% of the best predicted nonamers generated from one protein.

HLA-DP in unrelated hematopoietic cell transplantation revisited: challenges and opportunities.

When considering HLA-matched hematopoietic cell transplantation (HCT), sibling and unrelated donors (UDs) are biologically different because UD-HCT is typically performed across HLA-DP disparities absent in sibling HCT. Mismatched HLA-DP is targeted by direct alloreactive T cell responses with important implications for graft-versus-host disease and graft-versus-leukemia. This concise review details special features of HLA-DP as model antigens for clinically permissive mismatches mediating limited T-cell alloreactivity with minimal toxicity, and describes future avenues for their exploitation in cellular immunotherapy of malignant blood disorders.

Role of HLA-DP in the Presentation of Epitopes from the Truncated Bacterial PE38 Immunotoxin.

Identification of helper T-cell epitopes is important in many fields of medicine. We previously used an experimental approach to identify T-cell epitopes in PE38, a truncated bacterial toxin used in immunotoxins. Here, we evaluated the ability of antibodies to DR, DP, or DQ to block T-cell responses to PE38 epitopes in 36 PBMC samples. We predicted the binding affinities of peptides to DR, DP, and DQ alleles using computational tools and analyzed their ability to predict the T-cell epitopes. We found that HLA-DR is responsible for 65% of the responses, DP 24%, and DQ 4%. One epitope that is presented in 20% of the samples (10/50) is entirely DP restricted and was not predicted to bind to DR or DP reference alleles using binding algorithms. We conclude that DP has an important role in helper T-cell response to PE38.

Proinsulin-specific, HLA-DQ8, and HLA-DQ8-transdimer-restricted CD4+ T cells infiltrate islets in type 1 diabetes.

Type 1 diabetes (T1D) develops when insulin-secreting ß-cells, found in the pancreatic islets of Langerhans, are destroyed by infiltrating T cells. How human T cells recognize ß-cell-derived antigens remains unclear. Genetic studies have shown that HLA and insulin alleles are the most strongly associated with risk of T1D. These long-standing observations implicate CD4(+) T-cell responses against (pro)insulin in the pathogenesis of T1D. To dissect the autoimmune T-cell response against human ß-cells, we isolated and characterized 53 CD4(+) T-cell clones from within the residual pancreatic islets of a deceased organ donor who had T1D. These 53 clones expressed 47 unique clonotypes, 8 of which encoded proinsulin-specific T-cell receptors. On an individual clone basis, 14 of 53 CD4(+) T-cell clones (26%) recognized 6 distinct but overlapping epitopes in the C-peptide of proinsulin. These clones recognized C-peptide epitopes presented by HLA-DQ8 and, notably, HLA-DQ8 transdimers that form in HLA-DQ2/-DQ8 heterozygous individuals. Responses to these epitopes were detected in the peripheral blood mononuclear cells of some people with recent-onset T1D but not in HLA-matched control subjects. Hence, proinsulin-specific, HLA-DQ8, and HLA-DQ8-transdimer-restricted CD4(+) T cells are strongly implicated in the autoimmune pathogenesis of human T1D.

First report on the antibody verification of HLA-DR, HLA-DQ and HLA-DP epitopes recorded in the HLA Epitope Registry.

The International Registry of Antibody-Defined HLA Epitopes (http://www.epregistry.com.br) has been recently established as a tool to understand humoral responses to HLA mismatches. These epitopes can be structurally defined as eplets by three-dimensional molecular modeling and amino acid sequence differences between HLA antigens. A major goal is to identify HLA eplets that have been verified experimentally with informative antibodies. This report addresses class II epitopes encoded by genes in the HLA-D region. Our analysis included reviews of many publications about epitope specificity of class II reactive human and murine monoclonal antibodies and informative alloantibodies from HLA sensitized patients as well as our own antibody testing results. As of July 1, 2014, 24 HLA-DRB1/3/4/5, 15 DQB, 3 DQA and 8 DPB antibody-verified epitopes have been identified and recorded. The Registry is still a work-in-progress and will become a useful resource for HLA professionals interested in histocompatibility testing at the epitope level and investigating antibody responses to HLA mismatches in transplant patients.

A novel family of human leukocyte antigen class II receptors may have its origin in archaic human species.

HLA class II α and ß chains form receptors for antigen presentation to CD4(+) T cells. Numerous pairings of class II α and ß subunits from the wide range of haplotypes and isotypes may form, but most of these combinations, in particular those produced by isotype mixing, yielded mismatched dimers. It is unclear how selection of functional receptors is achieved. At the atomic level, it is not known which interactions of class II residues regulate selection of matched αß heterodimers and the evolutionary origin of matched isotype mixed dimer formation. In this study we investigated assembly of isotype-mixed HLA class II α and ß heterodimers. Assembly and carbohydrate maturation of various HLA-class II isotype-mixed α and ß subunits was dependent on the groove binding section of the invariant chain (Ii). By mutation of polymorphic DPß sequences, we identified two motifs, Lys-69 and GGPM-(84-87), that are engaged in Ii-dependent assembly of DPß with DRα. We identified five members of a family of DPß chains containing Lys-69 and GGPM 84-87, which assemble with DRα. The Lys/GGPM motif is present in the DPß sequence of the Neanderthal genome, and this ancient sequence is related to the human allele DPB1*0401. By site-directed mutagenesis, we inspected Neanderthal amino acid residues that differ from the DPB1*0401 allele and aimed to determine whether matched heterodimers are formed by assembly of DPß mutants with DRα. Because the *0401 allele is rare in the sub-Saharan population but frequent in the European population, it may have arisen in modern humans by admixture with Neanderthals in Europe.

Histidine hydrogen bonding in MHC at pH 5 and pH 7 modeled by molecular docking and molecular dynamics simulations.

Hydrogen bonds play important roles in maintaining the structure of proteins and in the formation of most biomolecular protein-ligand complexes. All amino acids can act as hydrogen bond donors and acceptors. Among amino acids, Histidine is unique, as it can exist in neutral or positively charged forms within the physiological pH range of 5.0 to 7.0. Histidine can thus interact with other aromatic residues as well as forming hydrogen bonds with polar and charged residues. The ability of His to exchange a proton lies at the heart of many important functional biomolecular interactions, including immunological ones. By using molecular docking and molecular dynamics simulation, we examine the influence of His protonation/deprotonation on peptide binding affinity to MHC class II proteins from locus HLA-DP. Peptide-MHC interaction underlies the adaptive cellular immune response, upon which the next generation of commercially-important vaccines will depend. Consistent with experiment, we find that peptides containing protonated His residues bind better to HLA-DP proteins than those with unprotonated His. Enhanced binding at pH 5.0 is due, in part, to additional hydrogen bonds formed between peptide His(+) and DP proteins. In acidic endosomes, protein His(79ß) is predominantly protonated. As a result, the peptide binding cleft narrows in the vicinity of His(79ß), which stabilizes the peptide - HLA-DP protein complex.

Class II HLA epitope matching-A strategy to minimize de novo donor-specific antibody development and improve outcomes.

De novo donor-specific antibody (dnDSA) develops in 15-25% of renal transplant recipients within 5 years of transplantation and is associated with 40% lower graft survival at 10 years. HLA epitope matching is a novel strategy that may minimize dnDSA development. HLAMatchmaker software was used to characterize epitope mismatches at 395 potential HLA-DR/DQ/DP conformational epitopes for 286 donor-recipient pairs. Epitope specificities were assigned using single antigen HLA bead analysis and correlated with known monoclonal alloantibody epitope targets. Locus-specific epitope mismatches were more numerous in patients who developed HLA-DR dnDSA alone (21.4 vs. 13.2, p < 0.02) or HLA-DQ dnDSA alone (27.5 vs. 17.3, p < 0.001). An optimal threshold for epitope mismatches (10 for HLA-DR, 17 for HLA-DQ) was defined that was associated with minimal development of Class II dnDSA. Applying these thresholds, zero and 2.7% of patients developed dnDSA against HLA-DR and HLA-DQ, respectively, after a median of 6.9 years. Epitope specificity analysis revealed that 3 HLA-DR and 3 HLA-DQ epitopes were independent multivariate predictors of Class II dnDSA. HLA-DR and DQ epitope matching outperforms traditional low-resolution antigen-based matching and has the potential to minimize the risk of de novo Class II DSA development, thereby improving long-term graft outcome.

NetMHCIIpan-3.0, a common pan-specific MHC class II prediction method including all three human MHC class II isotypes, HLA-DR, HLA-DP and HLA-DQ.

Major histocompatibility complex class II (MHCII) molecules play an important role in cell-mediated immunity. They present specific peptides derived from endosomal proteins for recognition by T helper cells. The identification of peptides that bind to MHCII molecules is therefore of great importance for understanding the nature of immune responses and identifying T cell epitopes for the design of new vaccines and immunotherapies. Given the large number of MHC variants, and the costly experimental procedures needed to evaluate individual peptide-MHC interactions, computational predictions have become particularly attractive as first-line methods in epitope discovery. However, only a few so-called pan-specific prediction methods capable of predicting binding to any MHC molecule with known protein sequence are currently available, and all of them are limited to HLA-DR. Here, we present the first pan-specific method capable of predicting peptide binding to any HLA class II molecule with a defined protein sequence. The method employs a strategy common for HLA-DR, HLA-DP and HLA-DQ molecules to define the peptide-binding MHC environment in terms of a pseudo sequence. This strategy allows the inclusion of new molecules even from other species. The method was evaluated in several benchmarks and demonstrates a significant improvement over molecule-specific methods as well as the ability to predict peptide binding of previously uncharacterised MHCII molecules. To the best of our knowledge, the NetMHCIIpan-3.0 method is the first pan-specific predictor covering all HLA class II molecules with known sequences including HLA-DR, HLA-DP, and HLA-DQ. The NetMHCpan-3.0 method is available at http://www.cbs.dtu.dk/services/NetMHCIIpan-3.0 .

Peptide binding to HLA-DP proteins at pH 5.0 and pH 7.0: a quantitative molecular docking study.

BACKGROUND: HLA-DPs are class II MHC proteins mediating immune responses to many diseases. Peptides bind MHC class II proteins in the acidic environment within endosomes. Acidic pH markedly elevates association rate constants but dissociation rates are almost unchanged in the pH range 5.0 - 7.0. This pH-driven effect can be explained by the protonation/deprotonation states of Histidine, whose imidazole has a pK(a) of 6.0. At pH 5.0, imidazole ring is protonated, making Histidine positively charged and very hydrophilic, while at pH 7.0 imidazole is unprotonated, making Histidine less hydrophilic. We develop here a method to predict peptide binding to the four most frequent HLA-DP proteins: DP1, DP41, DP42 and DP5, using a molecular docking protocol. Dockings to virtual combinatorial peptide libraries were performed at pH 5.0 and pH 7.0. RESULTS: The X-ray structure of the peptide--HLA-DP2 protein complex was used as a starting template to model by homology the structure of the four DP proteins. The resulting models were used to produce virtual combinatorial peptide libraries constructed using the single amino acid substitution (SAAS) principle. Peptides were docked into the DP binding site using AutoDock at pH 5.0 and pH 7.0. The resulting scores were normalized and used to generate Docking Score-based Quantitative Matrices (DS-QMs). The predictive ability of these QMs was tested using an external test set of 484 known DP binders. They were also compared to existing servers for DP binding prediction. The models derived at pH 5.0 predict better than those derived at pH 7.0 and showed significantly improved predictions for three of the four DP proteins, when compared to the existing servers. They are able to recognize 50% of the known binders in the top 5% of predicted peptides. CONCLUSIONS: The higher predictive ability of DS-QMs derived at pH 5.0 may be rationalised by the additional hydrogen bond formed between the backbone carbonyl oxygen belonging to the peptide position before p1 (p-1) and the protonated ε-nitrogen of His79ß. Additionally, protonated His residues are well accepted at most of the peptide binding core positions which is in a good agreement with the overall negatively charged peptide binding site of most MHC proteins.

Characterizing the binding motifs of 11 common human HLA-DP and HLA-DQ molecules using NNAlign.

Compared with HLA-DR molecules, the specificities of HLA-DP and HLA-DQ molecules have only been studied to a limited extent. The description of the binding motifs has been mostly anecdotal and does not provide a quantitative measure of the importance of each position in the binding core and the relative weight of different amino acids at a given position. The recent publication of larger data sets of peptide-binding to DP and DQ molecules opens the possibility of using data-driven bioinformatics methods to accurately define the binding motifs of these molecules. Using the neural network-based method NNAlign, we characterized the binding specificities of five HLA-DP and six HLA-DQ among the most frequent in the human population. The identified binding motifs showed an overall concurrence with earlier studies but revealed subtle differences. The DP molecules revealed a large overlap in the pattern of amino acid preferences at core positions, with conserved hydrophobic/aromatic anchors at P1 and P6, and an additional hydrophobic anchor at P9 in some variants. These results confirm the existence of a previously hypothesized supertype encompassing the most common DP alleles. Conversely, the binding motifs for DQ molecules appear more divergent, displaying unconventional anchor positions and in some cases rather unspecific amino acid preferences.

Peptide binding prediction for the human class II MHC allele HLA-DP2: a molecular docking approach.

BACKGROUND: MHC class II proteins bind oligopeptide fragments derived from proteolysis of pathogen antigens, presenting them at the cell surface for recognition by CD4+ T cells. Human MHC class II alleles are grouped into three loci: HLA-DP, HLA-DQ and HLA-DR. In contrast to HLA-DR and HLA-DQ, HLA-DP proteins have not been studied extensively, as they have been viewed as less important in immune responses than DRs and DQs. However, it is now known that HLA-DP alleles are associated with many autoimmune diseases. Quite recently, the X-ray structure of the HLA-DP2 molecule (DPA*0103, DPB1*0201) in complex with a self-peptide derived from the HLA-DR α-chain has been determined. In the present study, we applied a validated molecular docking protocol to a library of 247 modelled peptide-DP2 complexes, seeking to assess the contribution made by each of the 20 naturally occurred amino acids at each of the nine binding core peptide positions and the four flanking residues (two on both sides). RESULTS: The free binding energies (FBEs) derived from the docking experiments were normalized on a position-dependent (npp) and on an overall basis (nap), and two docking score-based quantitative matrices (DS-QMs) were derived: QMnpp and QMnap. They reveal the amino acid preferences at each of the 13 positions considered in the study. Apart from the leading role of anchor positions p1 and p6, the binding to HLA-DP2 depends on the preferences at p2. No effect of the flanking residues was found on the peptide binding predictions to DP2, although all four of them show strong preferences for particular amino acids. The predictive ability of the DS-QMs was tested using a set of 457 known binders to HLA-DP2, originating from 24 proteins. The sensitivities of the predictions at five different thresholds (5%, 10%, 15%, 20% and 25%) were calculated and compared to the predictions made by the NetMHCII and IEDB servers. Analysis of the DS-QMs indicated an improvement in performance. Additionally, DS-QMs identified the binding cores of several known DP2 binders. CONCLUSIONS: The molecular docking protocol, as applied to a combinatorial library of peptides, models the peptide-HLA-DP2 protein interaction effectively, generating reliable predictions in a quantitative assessment. The method is structure-based and does not require extensive experimental sequence-based data. Thus, it is universal and can be applied to model any peptide - protein interaction.

Stoichiometry of HLA class II-invariant chain oligomers.

BACKGROUND: The HLA gene complex encodes three class II isotypes, DR, DQ, and DP. HLA class II molecules are peptide receptors that present antigens for recognition by T lymphocytes. In antigen presenting cells, the assembly of matched α and ß subunits to heterodimers is chaperoned by invariant chain (Ii). Ii forms a homotrimer with three binding sites for class II heterodimers. The current model of class II and Ii structure states that three αß heterodimers bind to an Ii trimer. METHODOLOGY/PRINCIPAL FINDINGS: [corrected] We have now analyzed the composition and size of the complexes of class II and Ii using epitope tagged class II subunits and density gradient experiments. We show here that class II-Ii oligomers consist of one class II heterodimer associated with one Ii trimer, such that the DR, DQ and DP isotypes are contained within separate complexes with Ii. CONCLUSION/SIGNIFICANCE: We propose a structural model of the class II-Ii oligomer and speculate that the pentameric class II-Ii complex is bent towards the cell membrane, inhibiting the binding of additional class II heterodimers to Ii.

The HLA-DP2 protein binds the immunodominant epitope from myelin basic protein, MBP85-99, with high affinity.

Myelin basic protein (MBP) is a candidate autoantigen in multiple sclerosis (MS). The immunodominant epitope for T-cell responses is assigned to the amino acid sequence MBP84-102, which binds to human leukocyte antigen (HLA)-DR2a (DRB5*0101) and HLA-DR2b (DRB1*1501) of the HLA-DR2 haplotype carrying the strongest genetic association with MS. In contrast with HLA-DR and -DQ molecules, HLA-DP molecules are poorly characterized with respect to the binding of self-peptides. We show here that HLA-DP2 binds MBP85-99 with high affinity, and that the amino acid residues in position MBP91, MBP92 and MBP93 are influencing the binding, as shown by alanine scans. We further used a series of truncated peptides to identify the core of the binding. Moving the frame along the peptide from residues 87-97 to 89-99 progressively decreased the binding affinity for HLA-DP2, while moving further towards the C-terminal completely abrogated the binding of peptides to HLA-DP2. The data suggest that the docking of the MBP85-99 peptide into the HLA-DP2 groove is dependent on MBP88V and MBP89V and may use either of them as primary anchor for the p1 position. HLA-DP2 might thus present the MBP85-99 peptide in the same register as the HLA-DRB1*1501, where the MBP89V is preferred as the p1 anchor. Notably, full-length MBP was able to compete for peptide binding with an affinity similar to that seen for the high-affinity binding peptides, DRα170-83 and IIP53-65. In summary, the HLA-DP2 molecule binds the immunodominant epitope in MS, MBP85-99, possibly in more than one register.

Towards universal structure-based prediction of class II MHC epitopes for diverse allotypes.

The binding of peptide fragments of antigens to class II MHC proteins is a crucial step in initiating a helper T cell immune response. The discovery of these peptide epitopes is important for understanding the normal immune response and its misregulation in autoimmunity and allergies and also for vaccine design. In spite of their biomedical importance, the high diversity of class II MHC proteins combined with the large number of possible peptide sequences make comprehensive experimental determination of epitopes for all MHC allotypes infeasible. Computational methods can address this need by predicting epitopes for a particular MHC allotype. We present a structure-based method for predicting class II epitopes that combines molecular mechanics docking of a fully flexible peptide into the MHC binding cleft followed by binding affinity prediction using a machine learning classifier trained on interaction energy components calculated from the docking solution. Although the primary advantage of structure-based prediction methods over the commonly employed sequence-based methods is their applicability to essentially any MHC allotype, this has not yet been convincingly demonstrated. In order to test the transferability of the prediction method to different MHC proteins, we trained the scoring method on binding data for DRB1*0101 and used it to make predictions for multiple MHC allotypes with distinct peptide binding specificities including representatives from the other human class II MHC loci, HLA-DP and HLA-DQ, as well as for two murine allotypes. The results showed that the prediction method was able to achieve significant discrimination between epitope and non-epitope peptides for all MHC allotypes examined, based on AUC values in the range 0.632-0.821. We also discuss how accounting for peptide binding in multiple registers to class II MHC largely explains the systematically worse performance of prediction methods for class II MHC compared with those for class I MHC based on quantitative prediction performance estimates for peptide binding to class II MHC in a fixed register.

Peptide binding predictions for HLA DR, DP and DQ molecules.

BACKGROUND: MHC class II binding predictions are widely used to identify epitope candidates in infectious agents, allergens, cancer and autoantigens. The vast majority of prediction algorithms for human MHC class II to date have targeted HLA molecules encoded in the DR locus. This reflects a significant gap in knowledge as HLA DP and DQ molecules are presumably equally important, and have only been studied less because they are more difficult to handle experimentally. RESULTS: In this study, we aimed to narrow this gap by providing a large scale dataset of over 17,000 HLA-peptide binding affinities for a set of 11 HLA DP and DQ alleles. We also expanded our dataset for HLA DR alleles resulting in a total of 40,000 MHC class II binding affinities covering 26 allelic variants. Utilizing this dataset, we generated prediction tools utilizing several machine learning algorithms and evaluated their performance. CONCLUSION: We found that 1) prediction methodologies developed for HLA DR molecules perform equally well for DP or DQ molecules. 2) Prediction performances were significantly increased compared to previous reports due to the larger amounts of training data available. 3) The presence of homologous peptides between training and testing datasets should be avoided to give real-world estimates of prediction performance metrics, but the relative ranking of different predictors is largely unaffected by the presence of homologous peptides, and predictors intended for end-user applications should include all training data for maximum performance. 4) The recently developed NN-align prediction method significantly outperformed all other algorithms, including a naïve consensus based on all prediction methods. A new consensus method dropping the comparably weak ARB prediction method could outperform the NN-align method, but further research into how to best combine MHC class II binding predictions is required.