Peptide Recognition Research Articles

Multiple environmental antigens may trigger autoimmunity in psoriasis through T-cell receptor polyspecificity.

Psoriasis is a T-cell mediated autoimmune skin disease. HLA-C*06:02 is the main psoriasis-specific risk gene. Using a Vα3S1/Vβ13S1 T-cell receptor (TCR) from a lesional psoriatic CD8+ T-cell clone we had discovered that, as an underlying pathomechanism, HLA-C*06:02 mediates an autoimmune response against melanocytes in psoriasis, and we had identified an epitope from ADAMTS-like protein 5 (ADAMTSL5) as a melanocyte autoantigen. The conditions activating the psoriatic autoimmune response in genetically predisposed individuals throughout life remain incompletely understood. Here, we aimed to identify environmental antigens that might trigger autoimmunity in psoriasis because of TCR polyspecificity. We screened databases with the peptide recognition motif of the Vα3S1/Vβ13S1 TCR for environmental proteins containing peptides activating this TCR. We investigated the immunogenicity of these peptides for psoriasis patients and healthy controls by lymphocyte stimulation experiments and peptide-loaded HLA-C*06:02 tetramers. We identified peptides from wheat, Saccharomyces cerevisiae, microbiota, tobacco, and pathogens that activated both the Vα3S1/Vβ13S1 TCR and CD8+ T cells from psoriasis patients. Using fluorescent HLA-C*06:02 tetramers loaded with ADAMTSL5 or wheat peptides, we find that the same CD8+ T cells may recognize both autoantigen and environmental antigens. A wheat-free diet could alleviate psoriasis in several patients. Our results show that due to TCR polyspecificity, several environmental antigens corresponding to previously suspected psoriasis risk conditions converge in the reactivity of a pathogenic psoriatic TCR and might thus be able to stimulate the psoriatic autoimmune response against melanocytes. Avoiding the corresponding environmental risk factors could contribute to the management of psoriasis.

Use of the energy waveform electrocardiogram to detect subclinical left ventricular dysfunction in patients with type 2 diabetes mellitus

BackgroundRecent guidelines propose N-terminal pro-B-type natriuretic peptide (NT-proBNP) for recognition of asymptomatic left ventricular (LV) dysfunction (Stage B Heart Failure, SBHF) in type 2 diabetes mellitus (T2DM). Wavelet Transform based signal-processing transforms electrocardiogram (ECG) waveforms into an energy distribution waveform (ew)ECG, providing frequency and energy features that machine learning can use as additional inputs to improve the identification of SBHF. Accordingly, we sought whether machine learning model based on ewECG features was superior to NT-proBNP, as well as a conventional screening tool—the Atherosclerosis Risk in Communities (ARIC) HF risk score, in SBHF screening among patients with T2DM.MethodsParticipants in two clinical trials of SBHF (defined as diastolic dysfunction [DD], reduced global longitudinal strain [GLS ≤ 18%] or LV hypertrophy [LVH]) in T2DM underwent 12-lead ECG with additional ewECG feature and echocardiography. Supervised machine learning was adopted to identify the optimal combination of ewECG extracted features for SBHF screening in 178 participants in one trial and tested in 97 participants in the other trial. The accuracy of the ewECG model in SBHF screening was compared with NT-proBNP and ARIC HF.ResultsSBHF was identified in 128 (72%) participants in the training dataset (median 72 years, 41% female) and 64 (66%) in the validation dataset (median 70 years, 43% female). Fifteen ewECG features showed an area under the curve (AUC) of 0.81 (95% CI 0.787–0.794) in identifying SBHF, significantly better than both NT-proBNP (AUC 0.56, 95% CI 0.44–0.68, p < 0.001) and ARIC HF (AUC 0.67, 95%CI 0.56–0.79, p = 0.002). ewECG features were also led to robust models screening for DD (AUC 0.74, 95% CI 0.73–0.74), reduced GLS (AUC 0.76, 95% CI 0.73–0.74) and LVH (AUC 0.90, 95% CI 0.88–0.89).ConclusionsMachine learning based modelling using additional ewECG extracted features are superior to NT-proBNP and ARIC HF in SBHF screening among patients with T2DM, providing an alternative HF screening strategy for asymptomatic patients and potentially act as a guidance tool to determine those who required echocardiogram to confirm diagnosis.Trial registration LEAVE-DM, ACTRN 12619001393145 and Vic-ELF, ACTRN 12617000116325

APE-Gen2.0: Expanding Rapid Class I Peptide-Major Histocompatibility Complex Modeling to Post-Translational Modifications and Noncanonical Peptide Geometries.

The recognition of peptides bound to class I major histocompatibility complex (MHC-I) receptors by T-cell receptors (TCRs) is a determinant of triggering the adaptive immune response. While the exact molecular features that drive the TCR recognition are still unknown, studies have suggested that the geometry of the joint peptide-MHC (pMHC) structure plays an important role. As such, there is a definite need for methods and tools that accurately predict the structure of the peptide bound to the MHC-I receptor. In the past few years, many pMHC structural modeling tools have emerged that provide high-quality modeled structures in the general case. However, there are numerous instances of non-canonical cases in the immunopeptidome that the majority of pMHC modeling tools do not attend to, most notably, peptides that exhibit non-standard amino acids and post-translational modifications (PTMs) or peptides that assume non-canonical geometries in the MHC binding cleft. Such chemical and structural properties have been shown to be present in neoantigens; therefore, accurate structural modeling of these instances can be vital for cancer immunotherapy. To this end, we have developed APE-Gen2.0, a tool that improves upon its predecessor and other pMHC modeling tools, both in terms of modeling accuracy and the available modeling range of non-canonical peptide cases. Some of the improvements include (i) the ability to model peptides that have different types of PTMs such as phosphorylation, nitration, and citrullination; (ii) a new and improved anchor identification routine in order to identify and model peptides that exhibit a non-canonical anchor conformation; and (iii) a web server that provides a platform for easy and accessible pMHC modeling. We further show that structures predicted by APE-Gen2.0 can be used to assess the effects that PTMs have in binding affinity in a more accurate manner than just using solely the sequence of the peptide. APE-Gen2.0 is freely available at https://apegen.kavrakilab.org.

Genome mining of sulfonated lanthipeptides reveals unique cyclic peptide sulfotransferases

Although sulfonation plays crucial roles in various biological processes and is frequently utilized in medicinal chemistry to improve water solubility and chemical diversity of drug leads, it is rare and underexplored in ribosomally synthesized and post-translationally modified peptides (RiPPs). Biosynthesis of RiPPs typically entails modification of hydrophilic residues, which substantially increases their chemical stability and bioactivity, albeit at the expense of reducing water solubility. To explore sulfonated RiPPs that may have improved solubility, we conducted co-occurrence analysis of RiPP class-defining enzymes and sulfotransferase (ST), and discovered two distinctive biosynthetic gene clusters (BGCs) encoding both lanthipeptide synthetase (LanM) and ST. Upon expressing these BGCs, we characterized the structures of novel sulfonated lanthipeptides and determined the catalytic details of LanM and ST. We demonstrate that SslST-catalyzed sulfonation is leader-independent but relies on the presence of A ring formed by LanM. Both LanM and ST are promiscuous towards residues in the A ring, but ST displays strict regioselectivity toward Tyr5. The recognition of cyclic peptide by ST was further discussed. Bioactivity evaluation underscores the significance of the ST-catalyzed sulfonation. This study sets up the starting point to engineering the novel lanthipeptide STs as biocatalysts for hydrophobic lanthipeptides improvement.

Quantitative proteomics analysis identified new interacting proteins of JAL30 in Arabidopsis

Jacalin-related lectins (JALs) are a unique group of plant lectins derived from the jacalin protein family, which play important roles in plant defense responses. JAL30/PBP1 (PYK10 binding protein 1) interacts with inactive PYK10, exerting negative regulatory control over the size of the PYK10 complex, which is formed and activated upon insect or pathogen invasion. However, the precise interplay between JAL30 and other components remains elusive. In this study, we found JAL30 as a nucleocytoplasmic protein, but no obvious phenotype was observed in jal30–1 single mutant. Through immunoprecipitation (IP) enrichment combined with liquid chromatography-tandem mass spectrometry (LC-MS/MS), dozens of new JAL30 interacting proteins were found in addition to several reported ones. Gene Ontology (GO) analysis revealed that these interacting proteins were highly related to the wounding and bacterial stimuli, suggesting their potential involvement in the jasmonate (JA) response. Importantly, the expression of JAL30 was induced by MeJA treatment, further highlighting its relevance in plant defense mechanisms. A novel JAL30 interacting protein, ESM1, was identified and its interaction with JAL30 was confirmed by Co-immunoprecipitation. Moreover, ESM1 was found as an O-GlcNAcylated protein, suggesting that JAL30 may possess glycosylated protein binding ability, particularly in O-GlcNAcylated protein and peptide recognition. Overall, our study provides valuable insights into the interacting protein network and biological function of JAL30, demonstrates the interaction between JAL30 and ESM1, and uncovers the potential significance of JAL30 in plant defense system, potentially through its association with PYK10 complex or JA response. SignificanceThe biological functions of lectin proteins, including defense responses, immunity responses, signal transduction, have been well studied. Lectin proteins were also utilized to enrich glycosylated proteins for their specific carbohydrates binding capability. Jacalin-related lectins (JALs) were found to involve in plant defense mechanism. However, it is not yet clear whether JALs could use for enrichment of glycosylated proteins. In this study, we used label-free quantification method to identify interacting proteins of JAL30. A novel interacting protein, ESM1, as an O-GlcNAcylated protein was found. ESM1 has been reported to take part in defense against insect herbivory. Therefore, our findings provided experimental evidence to confirm that JALs have potential to be developed as the bio-tools to enrich glycosylated proteins. Finally, our data not only illustrated the vital biological role of JALs in plants, but also verified unique function of JAL30 in recognizing O-GlcNAcylated proteins.

In situ sequence recognition of peptides by enzymatic cleavage

In situ sequence recognition of peptides by enzymatic cleavage

EnAMP: A novel deep learning ensemble antibacterial peptide recognition algorithm based on multi-features.

Antimicrobial peptides (AMPs), as the preferred alternatives to antibiotics, have wide application with good prospects. Identifying AMPs through wet lab experiments remains expensive, time-consuming and challenging. Many machine learning methods have been proposed to predict AMPs and achieved good results. In this work, we combine two kinds of word embedding features with the statistical features of peptide sequences to develop an ensemble classifier, named EnAMP, in which, two deep neural networks are trained based on Word2vec and Glove word embedding features of peptide sequences, respectively, meanwhile, we utilize statistical features of peptide sequences to train random forest and support vector machine classifiers. The average of four classifiers is the final prediction result. Compared with other state-of-the-art algorithms on six datasets, EnAMP outperforms most existing models with similar computational costs, even when compared with high computational cost algorithms based on Bidirectional Encoder Representation from Transformers (BERT), the performance of our model is comparable. EnAMP source code and the data are available at https://github.com/ruisue/EnAMP.

Complementary hydrophobic interaction of the redox enzyme maturation protein NarJ with the signal peptide of the respiratory nitrate reductase NarG

In bacteria, NarJ plays an essential role as a redox enzyme maturation protein in the assembly of the nitrate reductase NarGHI by interacting with the N-terminal signal peptide of NarG to facilitate cofactor incorporation into NarG. The purpose of our research was to elucidate the exact mechanism of NarG signal peptide recognition by NarJ. We determined the structures of NarJ alone and in complex with the signal peptide of NarG via X-ray crystallography and verified the NarJ-NarG interaction through mutational, binding, and molecular dynamics simulation studies. NarJ adopts a curved α-helix bundle structure with a U-shaped hydrophobic groove on its concave side. This groove accommodates the signal peptide of NarG via a dual binding mode in which the left and right parts of the NarJ groove each interact with two consecutive hydrophobic residues from the N- and C-terminal regions of the NarG signal peptide, respectively, through shape and chemical complementarity. This binding is accompanied by unwinding of the helical structure of the NarG signal peptide and by stabilization of the NarG-binding loop of NarJ. We conclude that NarJ recognizes the NarG signal peptide through a complementary hydrophobic interaction mechanism that mediates a structural rearrangement.

Peptide recognition by a synthetic receptor at subnanomolar concentrations.

This paper describes the discovery and characterization of a dipeptide sequence, Lys-Phe, that binds to the synthetic receptor cucurbit[8]uril (Q8) in neutral aqueous solution with subnanomolar affinity when located at the N-terminus. The thermodynamic and structural basis for the binding of Q8 to a series of four pentapeptides was characterized by isothermal titration calorimetry, NMR spectroscopy, and X-ray crystallography. Submicromolar binding affinity was observed for the peptides Phe-Lys-Gly-Gly-Tyr (FKGGY, 0.3 μM) and Tyr-Leu-Gly-Gly-Gly (YLGGG, 0.2 μM), whereas the corresponding sequence isomers Lys-Phe-Gly-Gly-Tyr (KFGGY, 0.3 nM) and Leu-Tyr-Gly-Gly-Gly (LYGGG, 1.2 nM) bound to Q8 with 1000-fold and 170-fold increases in affinity, respectively. To our knowledge, these are the highest affinities reported between a synthetic receptor and an unmodified peptide. The high-resolution crystal structures of the Q8·Tyr-Leu-Gly-Gly-Gly and Q8·Leu-Tyr-Gly-Gly-Gly complexes have enabled a detailed analysis of the structural determinants for molecular recognition. The high affinity, sequence-selectivity, minimal size of the target binding site, reversibility in the presence of a competitive guest, compatibility with aqueous media, and low toxicity of Q8 should aid in the development of applications involving low concentrations of target polypeptides.

Non-mutational neoantigens in disease.

The ability of mammals to mount adaptive immune responses culminating with the establishment of immunological memory is predicated on the ability of the mature T cell repertoire to recognize antigenic peptides presented by syngeneic MHC class I and II molecules. Although it is widely believed that mature T cells are highly skewed towards the recognition of antigenic peptides originating from genetically diverse (for example, foreign or mutated) protein-coding regions, preclinical and clinical data rather demonstrate that novel antigenic determinants efficiently recognized by mature T cells can emerge from a variety of non-mutational mechanisms. In this Review, we describe various mechanisms that underlie the formation of bona fide non-mutational neoantigens, such as epitope mimicry, upregulation of cryptic epitopes, usage of non-canonical initiation codons, alternative RNA splicing, and defective ribosomal RNA processing, as well as both enzymatic and non-enzymatic post-translational protein modifications. Moreover, we discuss the implications of the immune recognition of non-mutational neoantigens for human disease.

DeepAMPNet: a novel antimicrobial peptide predictor employing AlphaFold2 predicted structures and a bi-directional long short-term memory protein language model.

Global public health is seriously threatened by the escalating issue of antimicrobial resistance (AMR). Antimicrobial peptides (AMPs), pivotal components of the innate immune system, have emerged as a potent solution to AMR due to their therapeutic potential. Employing computational methodologies for the prompt recognition of these antimicrobial peptides indeed unlocks fresh perspectives, thereby potentially revolutionizing antimicrobial drug development. In this study, we have developed a model named as deepAMPNet. This model, which leverages graph neural networks, excels at the swift identification of AMPs. It employs structures of antimicrobial peptides predicted by AlphaFold2, encodes residue-level features through a bi-directional long short-term memory (Bi-LSTM) protein language model, and constructs adjacency matrices anchored on amino acids' contact maps. In a comparative study with other state-of-the-art AMP predictors on two external independent test datasets, deepAMPNet outperformed in accuracy. Furthermore, in terms of commonly accepted evaluation matrices such as AUC, Mcc, sensitivity, and specificity, deepAMPNet achieved the highest or highly comparable performances against other predictors. deepAMPNet interweaves both structural and sequence information of AMPs, stands as a high-performance identification model that propels the evolution and design in antimicrobial peptide pharmaceuticals. The data and code utilized in this study can be accessed at https://github.com/Iseeu233/deepAMPNet.

Vaccine-elicited and naturally elicited antibodies differ in their recognition of the HIV-1 fusion peptide.

Broadly neutralizing antibodies have been proposed as templates for HIV-1 vaccine design, but it has been unclear how similar vaccine-elicited antibodies are to their naturally elicited templates. To provide insight, here we compare the recognition of naturally elicited and vaccine-elicited antibodies targeting the HIV-1 fusion peptide, which comprises envelope (Env) residues 512-526, with the most common sequence being AVGIGAVFLGFLGAA. Naturally elicited antibodies bound peptides with substitutions to negatively charged amino acids at residue positions 517-520 substantially better than the most common sequence, despite these substitutions rarely appearing in HIV-1; by contrast, vaccine-elicited antibodies were less tolerant of sequence variation, with no substitution of residues 512-516 showing increased binding. Molecular dynamics analysis and cryo-EM structural analysis of the naturally elicited ACS202 antibody in complex with the HIV-1 Env trimer with an alanine 517 to glutamine substitution suggested enhanced binding to result from electrostatic interactions with positively charged antibody residues. Overall, vaccine-elicited antibodies appeared to be more fully optimized to bind the most common fusion peptide sequence, perhaps reflecting the immunization with fusion peptide of the vaccine-elicited antibodies.

Sequence determinants of protein phase separation and recognition by protein phase-separated condensates through molecular dynamics and active learning.

Elucidating how protein sequence determines the properties of disordered proteins and their phase-separated condensates is a great challenge in computational chemistry, biology, and biophysics. Quantitative molecular dynamics simulations and derived free energy values can in principle capture how a sequence encodes the chemical and biological properties of a protein. These calculations are, however, computationally demanding, even after reducing the representation by coarse-graining; exploring the large spaces of potentially relevant sequences remains a formidable task. We employ an "active learning" scheme introduced by Yang et al. (bioRxiv, 2022, https://doi.org/10.1101/2022.08.05.502972) to reduce the number of labelled examples needed from simulations, where a neural network-based model suggests the most useful examples for the next training cycle. Applying this Bayesian optimisation framework, we determine properties of protein sequences with coarse-grained molecular dynamics, which enables the network to establish sequence-property relationships for disordered proteins and their self-interactions and their interactions in phase-separated condensates. We show how iterative training with second virial coefficients derived from the simulations of disordered protein sequences leads to a rapid improvement in predicting peptide self-interactions. We employ this Bayesian approach to efficiently search for new sequences that bind to condensates of the disordered C-terminal domain (CTD) of RNA Polymerase II, by simulating molecular recognition of peptides to phase-separated condensates in coarse-grained molecular dynamics. By searching for protein sequences which prefer to self-interact rather than interact with another protein sequence we are able to shape the morphology of protein condensates and design multiphasic protein condensates.

Caged aminoluciferin probe for bioluminescent immunoproteasome activity analysis.

The immunoproteasome (iCP) can be expressed under inflammatory conditions, such as exposure to interferon-gamma (IFN-γ), that alerts the cell to begin generating iCP preferentially over the standard proteasome (sCP). With the iCP becoming a widely targeted isoform in a variety of diseases, there is a need to understand its activity and expression in cells and in vivo. Activity-based probes for the iCP have been developed but their application has been limited due to their difficult synthesis and cannot be used in tissues or whole animals. Our lab has previously demonstrated we can monitor iCP activity using a 4-mer peptide linked to a fluorophore and a peptoid. This was utilized in the development of the first cell-permeable iCP activity-based probe that did not include a covalent reactive moiety. Here, we demonstrate that this same peptide recognition sequence can be appended to aminoluciferin, caging it, until its interaction with the iCP. This probe should be applicable to monitor iCP activity in animal models where tumor or other tissue has been engineered to produce luciferase. We anticipate it could also be applied to observe iCP activity as tumors are formed in vivo.

Structural engineering of mitochondria-targeted Au-Ag2S photosensitizers for enhanced photodynamic and photothermal therapy.

Much effort has been devoted to designing diverse photosensitizers for efficient photodynamic therapy (PDT) and photothermal therapy (PTT) performance. However, the effect of PS morphology on the PDT and PTT performance needs to be further explored. In this work, a photosensitizer, Au-Ag2S nanoparticles functionalized with indocyanine green, caspase-3 recognition peptides, and mitochondria-targeting peptides (AICM NPs) with different morphologies, including core-shell, eccentric core-shell-I, eccentric core-shell-II, and Janus morphologies, were synthesized to enhance PDT and PTT performance. Among them, AICM Janus NPs with enhanced charge-transfer efficiency and photothermal conversion demonstrate superior PDT and PTT performance compared to those of other morphologies. In addition, AICM NPs exhibit satisfactory surface-enhanced Raman scattering performance for in situ SERS monitoring of caspase-3 during PDT and PTT processes. After PDT and PTT treatment with AICM Janus NPs, the damaged mitochondria released caspase-3. AICM Janus NPs achieved a superior apoptosis rate in tumor cells in vitro. Furthermore, AICM Janus NPs treat the tumors in vivo within only 10 days, which is half the time reported in other work. The AICM NPs demonstrated superior therapeutic safety both in vitro and in vivo. This study investigates the effects of morphology-property-performance of photosensitizers on the PDT and PTT performances, which opens a new pathway toward designing photosensitizers for efficient PDT and PTT.

Evaluation of the humoral and mucosal immune response of a multiepitope vaccine against COVID-19 in pigs.

This study evaluated the immune response to a multiepitope recombinant chimeric protein (CHIVAX) containing B- and T-cell epitopes of the SARS-CoV-2 spike's receptor binding domain (RBD) in a translational porcine model for pre-clinical studies. We generated a multiepitope recombinant protein engineered to include six coding conserved epitopes from the RBD domain of the SARS-CoV-2 S protein. Pigs were divided into groups and immunized with different doses of the protein, with serum samples collected over time to determine antibody responses by indirect ELISA and antibody titration. Peptide recognition was also analyzed by Western blotting. A surrogate neutralization assay with recombinant ACE2 and RBDs was performed. Intranasal doses of the immunogen were also prepared and tested on Vietnamese minipigs. When the immunogen was administered subcutaneously, it induced specific IgG antibodies in pigs, and higher doses correlated with higher antibody levels. Antibodies from immunized pigs recognized individual peptides in the multiepitope vaccine and inhibited RBD-ACE2 binding for five variants of concern (VOC). Comparative antigen delivery methods showed that both, subcutaneous and combined subcutaneous/intranasal approaches, induced specific IgG and IgA antibodies, with the subcutaneous approach having superior neutralizing activity. CHIVAX elicited systemic immunity, evidenced by specific IgG antibodies in the serum, and local mucosal immunity, indicated by IgA antibodies in saliva, nasal, and bronchoalveolar lavage secretions. Importantly, these antibodies demonstrated neutralizing activity against SARS-CoV-2 in vitro. The elicited antibodies recognized individual epitopes on the chimeric protein and demonstrated the capacity to block RBD-ACE2 binding of the ancestral SARS-CoV-2 strain and four VOCs. The findings provide proof of concept for using multiepitope recombinant antigens and a combined immunization protocol to induce a neutralizing immune response against SARS-CoV-2 in the pig translational model for preclinical studies.

PANDORA v2.0: Benchmarking peptide-MHC II models and software improvements.

T-cell specificity to differentiate between self and non-self relies on T-cell receptor (TCR) recognition of peptides presented by the Major Histocompatibility Complex (MHC). Investigations into the three-dimensional (3D) structures of peptide:MHC (pMHC) complexes have provided valuable insights of MHC functions. Given the limited availability of experimental pMHC structures and considerable diversity of peptides and MHC alleles, it calls for the development of efficient and reliable computational approaches for modeling pMHC structures. Here we present an update of PANDORA and the systematic evaluation of its performance in modelling 3D structures of pMHC class II complexes (pMHC-II), which play a key role in the cancer immune response. PANDORA is a modelling software that can build low-energy models in a few minutes by restraining peptide residues inside the MHC-II binding groove. We benchmarked PANDORA on 136 experimentally determined pMHC-II structures covering 44 unique αβ chain pairs. Our pipeline achieves a median backbone Ligand-Root Mean Squared Deviation (L-RMSD) of 0.42 Å on the binding core and 0.88 Å on the whole peptide for the benchmark dataset. We incorporated software improvements to make PANDORA a pan-allele framework and improved the user interface and software quality. Its computational efficiency allows enriching the wealth of pMHC binding affinity and mass spectrometry data with 3D models. These models can be used as a starting point for molecular dynamics simulations or structure-boosted deep learning algorithms to identify MHC-binding peptides. PANDORA is available as a Python package through Conda or as a source installation at https://github.com/X-lab-3D/PANDORA.

Discovery of U2AF1 neoantigens in myeloid neoplasms

BackgroundMyelodysplastic syndromes (MDS) arise from somatic mutations acquired in hematopoietic stem and progenitor cells, causing cytopenias and predisposing to transformation into secondary acute myeloid leukemia (sAML). Recurrent mutations in spliceosome...

TCR-ESM: Employing protein language embeddings to predict TCR-peptide-MHC binding

Cognate target identification for T-cell receptors (TCRs) is a significant barrier in T-cell therapy development, which may be overcome by accurately predicting TCR interaction with peptide-bound major histocompatibility complex (pMHC). In this study, we have employed peptide embeddings learned from a large protein language model- Evolutionary Scale Modeling (ESM), to predict TCR-pMHC binding. The TCR-ESM model presented outperforms existing predictors. The complementarity-determining region 3 (CDR3) of the hypervariable TCR is located at the center of the paratope and plays a crucial role in peptide recognition. TCR-ESM trained on paired TCR data with both CDR3α and CDR3β chain information performs significantly better than those trained on data with only CDR3β, suggesting that both TCR chains contribute to specificity, the relative importance however depends on the specific peptide-MHC targeted. The study illuminates the importance of MHC information in TCR-peptide binding which remained inconclusive so far and was thought dependent on the dataset characteristics. TCR-ESM outperforms existing approaches on external datasets, suggesting generalizability. Overall, the potential of deep learning for predicting TCR-pMHC interactions and improving the understanding of factors driving TCR specificity are highlighted. The prediction model is available at http://tcresm.dhanjal-lab.iiitd.edu.in/ as an online tool.

Unsupervised and supervised AI on molecular dynamics simulations reveals complex characteristics of HLA-A2-peptide immunogenicity.

Immunologic recognition of peptide antigens bound to class I major histocompatibility complex (MHC) molecules is essential to both novel immunotherapeutic development and human health at large. Current methods for predicting antigen peptide immunogenicity rely primarily on simple sequence representations, which allow for some understanding of immunogenic features but provide inadequate consideration of the full scale of molecular mechanisms tied to peptide recognition. We here characterize contributions that unsupervised and supervised artificial intelligence (AI) methods can make toward understanding and predicting MHC(HLA-A2)-peptide complex immunogenicity when applied to large ensembles of molecular dynamics simulations. We first show that an unsupervised AI method allows us to identify subtle features that drive immunogenicity differences between a cancer neoantigen and its wild-type peptide counterpart. Next, we demonstrate that a supervised AI method for class I MHC(HLA-A2)-peptide complex classification significantly outperforms a sequence model on small datasets corrected for trivial sequence correlations. Furthermore, we show that both unsupervised and supervised approaches reveal determinants of immunogenicity based on time-dependent molecular fluctuations and anchor position dynamics outside the MHC binding groove. We discuss implications of these structural and dynamic immunogenicity correlates for the induction of T cell responses and therapeutic T cell receptor design.