Factors influencing immunodominance hierarchies in TCD8+ -mediated antiviral responses

A robust deep learning workflow to predict CD8 + T-cell epitopes

BACKGROUND

T-cells play a crucial role in the adaptive immune system by triggering responses against cancer cells and pathogens, while maintaining tolerance against self-antigens, which has sparked interest in the development of various T-cell-focused immunotherapies. However, the identification of antigens recognised by T-cells is low-throughput and laborious. To overcome some of these limitations, computational methods for predicting CD8 + T-cell epitopes have emerged. Despite recent developments, most immunogenicity algorithms struggle to learn features of peptide immunogenicity from small datasets, suffer from HLA bias and are unable to reliably predict pathology-specific CD8 + T-cell epitopes.

METHODS

We developed TRAP (T-cell recognition potential of HLA-I presented peptides), a robust deep learning workflow for predicting CD8 + T-cell epitopes from MHC-I presented pathogenic and self-peptides. TRAP uses transfer learning, deep learning architecture and MHC binding information to make context-specific predictions of CD8 + T-cell epitopes. TRAP also detects low-confidence predictions for peptides that differ significantly from those in the training datasets to abstain from making incorrect predictions. To estimate the immunogenicity of pathogenic peptides with low-confidence predictions, we further developed a novel metric, RSAT (relative similarity to autoantigens and tumour-associated antigens), as a complementary to 'dissimilarity to self' from cancer studies.

RESULTS

TRAP was used to identify epitopes from glioblastoma patients as well as SARS-CoV-2 peptides, and it outperformed other algorithms in both cancer and pathogenic settings. TRAP was especially effective at extracting immunogenicity-associated properties from restricted data of emerging pathogens and translating them onto related species, as well as minimising the loss of likely epitopes in imbalanced datasets. We also demonstrated that the novel metric termed RSAT was able to estimate immunogenic of pathogenic peptides of various lengths and species. TRAP implementation is available at: https://github.com/ChloeHJ/TRAP .

CONCLUSIONS

This study presents a novel computational workflow for accurately predicting CD8 + T-cell epitopes to foster a better understanding of antigen-specific T-cell response and the development of effective clinical therapeutics.

Discordant rearrangement of primary and anamnestic CD8+ T cell responses to influenza A viral epitopes upon exposure to bacterial superantigens: Implications for prophylactic vaccination, heterosubtypic immunity and superinfections

Infection with (SAg)-producing bacteria may precede or follow infection with or vaccination against influenza A viruses (IAVs). However, how SAgs alter the breadth of IAV-specific CD8⁺ T cell (T_CD8) responses is unknown. Moreover, whether recall responses mediating heterosubtypic immunity to IAVs are manipulated by SAgs remains unexplored. We employed wild-type (WT) and mutant bacterial SAgs, SAg-sufficient/deficient Staphylococcus aureus strains, and WT, mouse-adapted and reassortant IAV strains in multiple in vivo settings to address the above questions. Contrary to the popular view that SAgs delete or anergize T cells, systemic administration of staphylococcal enterotoxin B (SEB) or Mycoplasma arthritidis mitogen before intraperitoneal IAV immunization enlarged the clonal size of ‘select’ IAV-specific T_CD8 and reshuffled the hierarchical pattern of primary T_CD8 responses. This was mechanistically linked to the TCR Vβ makeup of the impacted clones rather than their immunodominance status. Importantly, SAg-expanded T_CD8 retained their IFN-γ production and cognate cytolytic capacities. The enhancing effect of SEB on immunodominant T_CD8 was also evident in primary responses to vaccination with heat-inactivated and live attenuated IAV strains administered intramuscularly and intranasally, respectively. Interestingly, in prime-boost immunization settings, the outcome of SEB administration depended strictly upon the time point at which this SAg was introduced. Accordingly, SEB injection before priming raised CD127^highKLRG1^low memory precursor frequencies and augmented the anamnestic responses of SEB-binding T_CD8. By comparison, introducing SEB before boosting diminished recall responses to IAV-derived epitopes drastically and indiscriminately. This was accompanied by lower Ki67 and higher Fas, LAG-3 and PD-1 levels consistent with a pro-apoptotic and/or exhausted phenotype. Therefore, SAgs can have contrasting impacts on anti-IAV immunity depending on the naïve/memory status and the TCR composition of exposed T_CD8. Finally, local administration of SEB or infection with SEB-producing S. aureus enhanced pulmonary T_CD8 responses to IAV. Our findings have clear implications for superinfections and prophylactic vaccination.

Exposure to bacterial superantigens (SAgs) is often a consequence of infection with common Gram-positive bacteria causing septic and toxic shock or food poisoning. How SAgs affect the magnitude, breadth and quality of infection/vaccine-elicited CD8⁺ T cell (T_CD8) responses to respiratory viral pathogens, including influenza A viruses (IAVs), is far from clear. Also importantly, superinfections with IAVs and SAg-producing bacteria are serious clinical occurrences during seasonal and pandemic flu and require urgent attention. We demonstrate that two structurally distinct SAgs, including staphylococcal enterotoxin B (SEB), unexpectedly enhance primary T_CD8 responses to ‘select’ IAV-derived epitopes depending on the TCR makeup of the responding clones. Intriguingly, the timing of exposure to SEB dictates the outcome of prime-boost immunization. Seeing a SAg before priming raises memory precursor frequencies and augments anamnestic T_CD8 responses. Conversely, a SAg encounter before boosting renders T_CD8 prone to death or exhaustion and impedes recall responses, thus likely compromising heterosubtypic immunity to IAVs. Finally, local exposure to SEB increases the pulmonary response of immunodominant IAV-specific T_CD8. These findings shed new light on how bacterial infections and SAgs influence the effectiveness of anti-IAV T_CD8 responses, and have, as such, wide-ranging implications for preventative vaccination and infection control.

MHC-I peptide binding activity assessed by exchange after cleavage of peptide covalently linked to β2-microglobulin

A common approach to measuring binding constants involves combining receptor and ligand and measuring the distribution of bound and free states after equilibration. For class I major histocompatibility (MHC-I) proteins, which bind short peptides for presentation to T cells, this approach is precluded by instability of peptide-free protein. Here we develop a method wherein a weakly-binding peptide covalently attached to the N-terminus of the MHC-I β2m subunit is released from the peptide binding site after proteolytic cleavage of the linker. The resultant protein is able to bind added peptide. A direct binding assay and method for estimation of peptide binding constant (K_d) are described, in which fluorescence polarization is used to follow peptide binding. A competition binding assay and method for estimation of inhibitor binding constant (K_i) using the same principle also are also described. The method uses a cubic equation to relate observed binding to probe concentration, probe K_d, inhibitor concentration, and inhibitor K_i under general reaction conditions without assumptions relating to relative binding affinities or concentrations. We also delineate advantages of this approach compared to the Cheng-Prusoff and Munson-Rodbard approaches for estimation of K_i using competition binding data.

PD-1/PD-L1 co-inhibition shapes anticancer T cell immunodominance: facing the consequences of an immunological ménage à trois

PD-1- and PD-L1-blocking monoclonal antibodies have shown significant promise in clinical settings and rekindled the hope for successful cancer immunotherapy. We recently demonstrated that interfering with PD-1/PD-L1 signaling selectively augments CD8+ T cell (TCD8) responses to subdominant determinants (SDDs) of a model tumor antigen. This was likely due to decreased lysis of SDD-specific TCD8 by neighboring immunodominant clones co-engaging the same antigen-presenting cells (APCs). We therefore proposed that PD-1-based checkpoint inhibitors widen the range of tumor determinants that can be effectively targeted by TCD8. Subsequently and using different tumor models, Chen et al. reported, in Proceedings of the National Academy of Sciences of the United States of America, that PD-L1 protects APCs from the lytic function of immunodominant TCD8 and that PD-L1 blockade narrows, rather than broadens, the overall anticancer T cell response. Here, we briefly compare and contrast the experimental systems employed by the two groups, which may account, at least partially, for the opposing conclusions drawn. We argue that the pathway(s) of tumor antigen presentation, direct presentation versus cross-presentation, and the intensity of PD-1 expression by immunodominant and subdominant TCD8 must be taken into consideration in rational design of anti-PD-1/PD-L1-adjuvanted tumor vaccines and therapies.

Suppression of immunodominant antitumor and antiviral CD8+ T cell responses by indoleamine 2,3-dioxygenase

Indoleamine 2,3-dioxygenase (IDO) is a tryptophan-degrading enzyme known to suppress antitumor CD8⁺ T cells (T_CD8). The role of IDO in regulation of antiviral T_CD8 responses is far less clear. In addition, whether IDO controls both immunodominant and subdominant T_CD8 is not fully understood. This is an important question because the dominance status of tumor- and virus-specific T_CD8 may determine their significance in protective immunity and in vaccine design. We evaluated the magnitude and breadth of cross-primed T_CD8 responses to simian virus 40 (SV40) large T antigen as well as primary and recall T_CD8 responses to influenza A virus (IAV) in the absence or presence of IDO. IDO^−/− mice and wild-type mice treated with 1-methyl-D-tryptophan, a pharmacological inhibitor of IDO, exhibited augmented responses to immunodominant epitopes encoded by T antigen and IAV. IDO-mediated suppression of these responses was independent of CD4⁺CD25⁺FoxP3⁺ regulatory T cells, which remained numerically and functionally intact in IDO^−/− mice. Treatment with L-kynurenine failed to inhibit T_CD8 responses, indicating that tryptophan metabolites are not responsible for the suppressive effect of IDO in our models. Immunodominant T antigen-specific T_CD8 from IDO^−/− mice showed increased Ki-67 expression, suggesting that they may have acquired a more vigorous proliferative capacity in vivo. In conclusion, IDO suppresses immunodominant T_CD8 responses to tumor and viral antigens. Our work also demonstrates that systemic primary and recall T_CD8 responses to IAV are controlled by IDO. Inhibition of IDO thus represents an attractive adjuvant strategy in boosting anticancer and antiviral T_CD8 targeting highly immunogenic antigens.

Cytomegalovirus vectors violate CD8+ T cell epitope recognition paradigms

CD8(+) T cell responses focus on a small fraction of pathogen- or vaccine-encoded peptides, and for some pathogens, these restricted recognition hierarchies limit the effectiveness of antipathogen immunity. We found that simian immunodeficiency virus (SIV) protein-expressing rhesus cytomegalovirus (RhCMV) vectors elicit SIV-specific CD8(+) T cells that recognize unusual, diverse, and highly promiscuous epitopes, including dominant responses to epitopes restricted by class II major histocompatibility complex (MHC) molecules. Induction of canonical SIV epitope-specific CD8(+) T cell responses is suppressed by the RhCMV-encoded Rh189 gene (corresponding to human CMV US11), and the promiscuous MHC class I- and class II-restricted CD8(+) T cell responses occur only in the absence of the Rh157.5, Rh157.4, and Rh157.6 (human CMV UL128, UL130, and UL131) genes. Thus, CMV vectors can be genetically programmed to achieve distinct patterns of CD8(+) T cell epitope recognition.

Subdominant/cryptic CD8 T cell epitopes contribute to resistance against experimental infection with a human protozoan parasite

During adaptive immune response, pathogen-specific CD8⁺ T cells recognize preferentially a small number of epitopes, a phenomenon known as immunodominance. Its biological implications during natural or vaccine-induced immune responses are still unclear. Earlier, we have shown that during experimental infection, the human intracellular pathogen Trypanosoma cruzi restricts the repertoire of CD8⁺ T cells generating strong immunodominance. We hypothesized that this phenomenon could be a mechanism used by the parasite to reduce the breath and magnitude of the immune response, favoring parasitism, and thus that artificially broadening the T cell repertoire could favor the host. Here, we confirmed our previous observation by showing that CD8⁺ T cells of H-2^a infected mice recognized a single epitope of an immunodominant antigen of the trans-sialidase super-family. In sharp contrast, CD8⁺ T cells from mice immunized with recombinant genetic vaccines (plasmid DNA and adenovirus) expressing this same T. cruzi antigen recognized, in addition to the immunodominant epitope, two other subdominant epitopes. This unexpected observation allowed us to test the protective role of the immune response to subdominant epitopes. This was accomplished by genetic vaccination of mice with mutated genes that did not express a functional immunodominant epitope. We found that these mice developed immune responses directed solely to the subdominant/cryptic CD8 T cell epitopes and a significant degree of protective immunity against infection mediated by CD8⁺ T cells. We concluded that artificially broadening the T cell repertoire contributes to host resistance against infection, a finding that has implications for the host-parasite relationship and vaccine development.

Signal peptides and trans-membrane regions are broadly immunogenic and have high CD8+ T cell epitope densities: Implications for vaccine development

Cell mediated immune response has a major role in controlling the elimination of infectious agents. The rational design of sub-unit peptide vaccines against intracellular pathogens or cancer requires the use of antigenic sequence/s that can induce highly potent, long lasting and antigen-specific responses in the majority of the population. A promising peptide selection strategy is the detection of multi-epitope peptide sequences with an ability to bind multiple MHC alleles. While past research sought the best epitopes based on their specific antigenicity, we ask whether specific defined domains have high epitope densities. Signal peptides and trans-membrane domains were found to have exceptionally high epitope densities. The improved MHC binding of these domains relies on their hydrophobic nature and, in signal peptides, also on their specific sequence. The high epitope density of SP was computed using in-silico methods and corroborated by the high percentage of identified SP epitope in the IEDB (immune epitope database). The enhanced immunogenicity of SP was then experimentally confirmed using a panel of nine peptides derived from Mycobacterium tuberculosis (MTb) proteins used in human PBMC proliferation assays and T cell lines functional assays. Our results show the exceptionally high antigen specific response rates and population coverage to SP sequences compared with non-SP peptide antigens derived from the same proteins. The results suggest a novel scheme for the rational design of T cell vaccines using a domain based rather than an epitope based approach.

Complete modification of TCR specificity and repertoire selection does not perturb a CD8+ T cell immunodominance hierarchy

Understanding T cell immunodominance hierarchies is fundamental to the development of cellular-based vaccines and immunotherapy. A combination of influenza virus infection in C57BL/6J mice and reverse genetics is used here to dissect the role of T cell antigen receptor (TCR) repertoire in the immunodominant D(b)NP(366)CD8(+) T cell response. Infection with an engineered virus (NPM6A) containing a single alanine (A) mutation at the critical p6 NP(366-374) residue induced a noncross-reactive CD8(+) T cell response characterized by a novel, narrower TCR repertoire per individual mouse that was nonetheless equivalent in magnitude to that generated after WT virus challenge. Although of lower overall avidity, the levels of both cytotoxic T lymphocyte activity and cytokine production were comparable with those seen for the native response. Importantly, the overdominance profile characteristic of secondary D(b)NP(366)-specific clonal expansions was retained for the NPM6A mutant. The primary determinants of immunodominance in this endogenous, non-TCR-transgenic model of viral immunity are thus independent of TCR repertoire composition and diversity. These findings both highlight the importance of effective antigen dose for T cell vaccination and/or immunotherapy and demonstrate the feasibility of priming the memory T cell compartment with engineered viruses to protect against commonly selected mutants viral (or tumor) escape mutants.