Investigating the ability of deep learning-based structure prediction to extrapolate and/or enrich the set of antibody CDR canonical forms.

Deep learning models have been shown to accurately predict protein structure from sequence, allowing researchers to explore protein space from the structural viewpoint. In this paper we explore whether "novel" features, such as distinct loop conformations can arise from these predictions despite not being present in the training data. Here we have used ABodyBuilder2, a deep learning antibody structure predictor, to predict the structures of ~1.5M paired antibody sequences. We examined the predicted structures of the canonical CDR loops and found that most of these predictions fall into the already described CDR canonical form structural space. We also found a small number of "new" canonical clusters composed of heterogeneous sequences united by a common sequence motif and loop conformation. Analysis of these novel clusters showed their origins to be either shapes seen in the training data at very low frequency or shapes seen at high frequency but at a shorter sequence length. To evaluate explicitly the ability of ABodyBuilder2 to extrapolate, we retrained several models whilst withholding all antibody structures of a specific CDR loop length or canonical form. These "starved" models showed evidence of generalisation across CDRs of different lengths, but they did not extrapolate to loop conformations which were highly distinct from those present in the training data. However, the models were able to accurately predict a canonical form even if only a very small number of examples of that shape were in the training data. Our results suggest that deep learning protein structure prediction methods are unable to make completely out-of-domain predictions for CDR loops. However, in our analysis we also found that even minimal amounts of data of a structural shape allow the method to recover its original predictive abilities. We have made the ~1.5 M predicted structures used in this study available to download at https://doi.org/10.5281/zenodo.10280181.

Structural definition of HLA class II-presented SARS-CoV-2 epitopes reveals a mechanism to escape pre-existing CD4+ T cell immunity.

CD4+ T cells recognize a broad range of peptide epitopes of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which contribute to immune memory and limit COVID-19 disease. We demonstrate that the immunogenicity of SARS-CoV-2 peptides, in the context of the model allotype HLA-DR1, does not correlate with their binding affinity to the HLA heterodimer. Analyzing six epitopes, some with very low binding affinity, we solve X-ray crystallographic structures of each bound to HLA-DR1. Further structural definitions reveal the precise molecular impact of viral variant mutations on epitope presentation. Omicron escaped ancestral SARS-CoV-2 immunity to two epitopes through two distinct mechanisms: (1) mutations to TCR-facing epitope positions and (2) a mechanism whereby a single amino acid substitution caused a register shift within the HLA binding groove, completely altering the peptide-HLA structure. This HLA-II-specific paradigm of immune escape highlights how CD4+ T cell memory is finely poised at the level of peptide-HLA-II presentation.

Molecular characterization of HLA class II binding to the LAG-3 T cell co-inhibitory receptor.

Immune checkpoint inhibitors (antibodies that block the T cell co-inhibitory receptors PD-1/PD-L1 or CTLA-4) have revolutionized the treatment of some forms of cancer. Importantly, combination approaches using drugs that target both pathways have been shown to boost the efficacy of such treatments. Subsequently, several other T cell inhibitory receptors have been identified for the development of novel immune checkpoint inhibitors. Included in this list is the co-inhibitory receptor lymphocyte activation gene-3 (LAG-3), which is upregulated on T cells extracted from tumor sites that have suppressive or exhausted phenotypes. However, the molecular rules that govern the function of LAG-3 are still not understood. Using surface plasmon resonance combined with a novel bead-based assay (AlphaScreenTM ), we demonstrate that LAG-3 can directly and specifically interact with intact human leukocyte antigen class II (HLA-II) heterodimers. Unlike the homologue CD4, which has an immeasurably weak affinity using these biophysical approaches, LAG-3 binds with low micromolar affinity. We further validated the interaction at the cell surface by staining LAG-3+ cells with pHLA-II-multimers. These data provide new insights into the mechanism by which LAG-3 initiates T cell inhibition.

Immune Remodeling of the Extracellular Matrix Drives Loss of Cancer Stem Cells and Tumor Rejection.

The nature of the tumor microenvironment (TME) influences the ability of tumor-specific T cells to control tumor growth. In this study, we performed an unbiased comparison of the TME of regulatory T-cell (Treg)-replete and Treg-depleted carcinogen-induced tumors, including Treg-depleted responding (regressing) and non-responding (growing) tumors. This analysis revealed an inverse relationship between extracellular matrix (ECM) and T-cell infiltrates where responding tumors were T-cell rich and ECM poor, whereas the converse was observed in non-responder tumors. For this reason, we hypothesized that the ECM acted as a barrier to successful T-cell infiltration and tumor rejection. However, further experiments revealed that this was not the case but instead showed that an effective T-cell response dramatically altered the density of ECM in the TME. Along with loss of ECM and high numbers of infiltrating T cells, responder tumors were distinguished by the development of lymphatic and blood vessel networks with specialized immune function. ECM-rich tumors exhibited a stem cell-like gene expression profile and superior tumor-initiating capacity, whereas such features were absent in responder tumors. Overall, these findings define an extended role for an effective immune response, not just in direct killing of tumor cells but in widescale remodeling of the TME to favor loss of ECM, elimination of cancer stem cells, and propagation of adaptive immunity.

CD4+ T Cells Recognize Conserved Influenza A Epitopes through Shared Patterns of V-Gene Usage and Complementary Biochemical Features.

T cell recognition of peptides presented by human leukocyte antigens (HLAs) is mediated by the highly variable T cell receptor (TCR). Despite this built-in TCR variability, individuals can mount immune responses against viral epitopes by using identical or highly related TCRs expressed on CD8+ T cells. Characterization of these TCRs has extended our understanding of the molecular mechanisms that govern the recognition of peptide-HLA. However, few examples exist for CD4+ T cells. Here, we investigate CD4+ T cell responses to the internal proteins of the influenza A virus that correlate with protective immunity. We identify five internal epitopes that are commonly recognized by CD4+ T cells in five HLA-DR1+ subjects and show conservation across viral strains and zoonotic reservoirs. TCR repertoire analysis demonstrates several shared gene usage biases underpinned by complementary biochemical features evident in a structural comparison. These epitopes are attractive targets for vaccination and other T cell therapies.

VDJdb in 2019: database extension, new analysis infrastructure and a T-cell receptor motif compendium.

Here, we report an update of the VDJdb database with a substantial increase in the number of T-cell receptor (TCR) sequences and their cognate antigens. The update further provides a new database infrastructure featuring two additional analysis modes that facilitate database querying and real-world data analysis. The increased yield of TCR specificity identification methods and the overall increase in the number of studies in the field has allowed us to expand the database more than 5-fold. Furthermore, several new analysis methods are included. For example, batch annotation of TCR repertoire sequencing samples allows for annotating large datasets on-line. Using recently developed bioinformatic methods for TCR motif mining, we have built a reduced set of high-quality TCR motifs that can be used for both training TCR specificity predictors and matching against TCRs of interest. These additions enhance the versatility of the VDJdb in the task of exploring T-cell antigen specificities. The database is available at https://vdjdb.cdr3.net.

Human leukocyte antigen (HLA) class II peptide flanking residues tune the immunogenicity of a human tumor-derived epitope.

CD4+ T-cells recognize peptide antigens, in the context of human leukocyte antigen (HLA) class II molecules (HLA-II), which through peptide-flanking residues (PFRs) can extend beyond the limits of the HLA binding. The role of the PFRs during antigen recognition is not fully understood; however, recent studies have indicated that these regions can influence T-cell receptor (TCR) affinity and pHLA-II stability. Here, using various biochemical approaches including peptide sensitivity ELISA and ELISpot assays, peptide-binding assays and HLA-II tetramer staining, we focused on CD4+ T-cell responses against a tumor antigen, 5T4 oncofetal trophoblast glycoprotein (5T4), which have been associated with improved control of colorectal cancer. Despite their weak TCR-binding affinity, we found that anti-5T4 CD4+ T-cells are polyfunctional and that their PFRs are essential for TCR recognition of the core bound nonamer. The high-resolution (1.95 Å) crystal structure of HLA-DR1 presenting the immunodominant 20-mer peptide 5T4111-130, combined with molecular dynamic simulations, revealed how PFRs explore the HLA-proximal space to contribute to antigen reactivity. These findings advance our understanding of what constitutes an HLA-II epitope and indicate that PFRs can tune weak affinity TCR-pHLA-II interactions.

Diversity within the adenovirus fiber knob hypervariable loops influences primary receptor interactions.

Adenovirus based vectors are of increasing importance for wide ranging therapeutic applications. As vaccines, vectors derived from human adenovirus species D serotypes 26 and 48 (HAdV-D26/48) are demonstrating promising efficacy as protective platforms against infectious diseases. Significant clinical progress has been made, yet definitive studies underpinning mechanisms of entry, infection, and receptor usage are currently lacking. Here, we perform structural and biological analysis of the receptor binding fiber-knob protein of HAdV-D26/48, reporting crystal structures, and modelling putative interactions with two previously suggested attachment receptors, CD46 and Coxsackie and Adenovirus Receptor (CAR). We provide evidence of a low affinity interaction with CAR, with modelling suggesting affinity is attenuated through extended, semi-flexible loop structures, providing steric hindrance. Conversely, in silico and in vitro experiments are unable to provide evidence of interaction between HAdV-D26/48 fiber-knob with CD46, or with Desmoglein 2. Our findings provide insight into the cell-virus interactions of HAdV-D26/48, with important implications for the design and engineering of optimised Ad-based therapeutics.

The nature of the human T cell response to the cancer antigen 5T4 is determined by the balance of regulatory and inflammatory T cells of the same antigen-specificity: implications for vaccine design.

The oncofoetal antigen 5T4 is a promising T cell target in the context of colorectal cancer, as demonstrated by a recent clinical study where 5T4-specific T cell responses, induced by vaccination or cyclophosphamide, were associated with a significantly prolonged survival of patients with metastatic disease. Whilst Th1-type (IFN-γ+) responses specific to 5T4, and other oncofoetal antigens, are often readily detectable in early stage CRC patients and healthy donors, their activity is suppressed as the cancer progresses by CD4+CD25hiFoxp3+ regulatory T cells (Treg) which contribute to the immunosuppressive environment conducive to tumour growth. This study mapped the fine specificity of Th1 and Treg cell responses to the 5T4 protein. Surprisingly, both immunogenic peptides and those recognised by Tregs clustered in the same HLA-DR transcending epitope-rich hotspots within the 5T4 protein. Similarly, regions of low Th1-cell immunogenicity also did not contain peptides capable of stimulating Tregs, further supporting the notion that Treg and Th1 cells recognise the same peptides. Understanding the rules which govern the balance of Th1 and Treg cells responding to a given peptide specificity is, therefore, of fundamental importance to designing strategies for manipulating the balance in favour of Th1 cells, and thus the most effective anti-cancer T cell responses.

Using X-ray Crystallography, Biophysics, and Functional Assays to Determine the Mechanisms Governing T-cell Receptor Recognition of Cancer Antigens.

Human CD8+ cytotoxic T lymphocytes (CTLs) are known to play an important role in tumor control. In order to carry out this function, the cell surface-expressed T-cell receptor (TCR) must functionally recognize human leukocyte antigen (HLA)-restricted tumor-derived peptides (pHLA). However, we and others have shown that most TCRs bind sub-optimally to tumor antigens. Uncovering the molecular mechanisms that define this poor recognition could aid in the development of new targeted therapies that circumnavigate these shortcomings. Indeed, present therapies that lack this molecular understanding have not been universally effective. Here, we describe methods that we commonly employ in the laboratory to determine how the nature of the interaction between TCRs and pHLA governs T-cell functionality. These methods include the generation of soluble TCRs and pHLA and the use of these reagents for X-ray crystallography, biophysical analysis, and antigen-specific T-cell staining with pHLA multimers. Using these approaches and guided by structural analysis, it is possible to modify the interaction between TCRs and pHLA and to then test how these modifications impact T-cell antigen recognition. These findings have already helped to clarify the mechanism of T-cell recognition of a number of cancer antigens and could direct the development of altered peptides and modified TCRs for new cancer therapies.

T-cell libraries allow simple parallel generation of multiple peptide-specific human T-cell clones.

Isolation of peptide-specific T-cell clones is highly desirable for determining the role of T-cells in human disease, as well as for the development of therapies and diagnostics. However, generation of monoclonal T-cells with the required specificity is challenging and time-consuming. Here we describe a library-based strategy for the simple parallel detection and isolation of multiple peptide-specific human T-cell clones from CD8(+) or CD4(+) polyclonal T-cell populations. T-cells were first amplified by CD3/CD28 microbeads in a 96U-well library format, prior to screening for desired peptide recognition. T-cells from peptide-reactive wells were then subjected to cytokine-mediated enrichment followed by single-cell cloning, with the entire process from sample to validated clone taking as little as 6 weeks. Overall, T-cell libraries represent an efficient and relatively rapid tool for the generation of peptide-specific T-cell clones, with applications shown here in infectious disease (Epstein-Barr virus, influenza A, and Ebola virus), autoimmunity (type 1 diabetes) and cancer.