Rational design of peptide-based tumor vaccines

Attentive Variational Information Bottleneck for TCR-peptide interaction prediction

MOTIVATION

We present a multi-sequence generalization of Variational Information Bottleneck and call the resulting model Attentive Variational Information Bottleneck (AVIB). Our AVIB model leverages multi-head self-attention to implicitly approximate a posterior distribution over latent encodings conditioned on multiple input sequences. We apply AVIB to a fundamental immuno-oncology problem: predicting the interactions between T-cell receptors (TCRs) and peptides.

RESULTS

Experimental results on various datasets show that AVIB significantly outperforms state-of-the-art methods for TCR-peptide interaction prediction. Additionally, we show that the latent posterior distribution learned by AVIB is particularly effective for the unsupervised detection of out-of-distribution amino acid sequences.

AVAILABILITY AND IMPLEMENTATION

The code and the data used for this study are publicly available at: https://github.com/nec-research/vibtcr.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Protective immune response against P32 oncogenic peptide-pulsed PBMCs in mouse models of breast cancer

High expression of p32 in certain tumors makes it a potential target for immunotherapy. In the present study, the first goal was to design multi-epitope peptides from the P32 protein and the second goal was to compare the prophylactic effects of DCs- and PBMCs- based vaccines by pulsing them with designed peptides. For these purposes, 160 BALB/c mice were vaccinated in 5 different subgroups of each 4 peptides using PBS (F1-4a), F peptides alone (F1-4b), F peptides with CpG-ODN (F1-4c), F peptides with CpGODN and DCs (F1-4d), and F peptides with CpG-ODN and PBMCs (F1-4e). We found a significantly higher interferon-γ (IFN-γ) and granzyme B levels in T cells of F4d and F4e subgroups compared to control (p ≤ 0.05). The result of challenging spleen PBMCs of vaccinated mice with 4T1 cells showed significant up- and down- regulation of Fas ligand (FasL) and forkhead box P3 (Foxp3) gene expression between F4d and F4e subgroups with control, respectively. In addition, a significant change was seen in Caspase3 gene expression of F4d subgroup compared to control (p ≤ 0.05). Supernatant levels of IFN-γ and perforin were significantly increased in F4d and F4e subgroups compared to control. Consequently, significantly lower tumor sizes and prolonged survival time were detected in F4d and F4e subgroups compared to control after challenging mice with 4T1 cells. Accordingly, these results demonstrated that PBMCs pulsed F4 peptide-based vaccine could induce a protective immune response while it is a simple and less expensive vaccine.

Analogue peptides for the immunotherapy of human acute myeloid leukemia

The use of peptide vaccines, enhanced by adjuvants, has shown some efficacy in clinical trials. However, responses are often short-lived and rarely induce notable memory responses. The reason is that self-antigens have already been presented to the immune system as the tumor develops, leading to tolerance or some degree of host tumor cell destruction. To try to break tolerance against self-antigens, one of the methods employed has been to modify peptides at the anchor residues to enhance their ability to bind major histocompatibility complex molecules, extending their exposure to the T-cell receptor. These modified or analogue peptides have been investigated as stimulators of the immune system in patients with different cancers with variable but sometimes notable success. In this review we describe the background and recent developments in the use of analogue peptides for the immunotherapy of acute myeloid leukemia describing knowledge useful for the application of analogue peptide treatments for other malignancies.

Cytolytic activity of the human papillomavirus type 16 E711-20 epitope-specific cytotoxic T lymphocyte is enhanced by heat shock protein 110 in HLA-A*0201 transgenic mice

Heat shock proteins (HSPs) have been successfully applied to a broad range of vaccines as biological adjuvants to enhance the immune response. The recently defined HSP110, in particular, exhibits strong protein binding affinity and is capable of enhancing the immunogenicity of protein antigens remarkably more than other HSP family members. In our previous study, we verified that murine HSP110 (mHSP110) significantly enhanced the immune response of a C57BL/6 mouse model to the H-2(d)-restricted human papillomavirus (HPV) E749-57 epitope (short peptide spanning the 49th to 57th amino acid residues in the E7 protein). To determine whether HSP110 similarly enhances the immunogenicity of human epitope peptides, we used the HLA-A2 transgenic mouse model to investigate the efficacy of the mHSP110 chaperone molecule as an immunoadjuvant of the human HLA-A2-restricted HPV16 E711-20 epitope vaccine. Results showed that mHSP110 efficiently formed a noncovalently bound complex with the E711-20 epitope. The mHSP110-E711-20 complex induced epitope-specific splenocyte proliferation and E711-20-specific gamma interferon (IFN-γ) secretion. Importantly, cytotoxic T lymphocytes primed by the mHSP110-E711-20 complex exerted strong cytolytic effects on target T2 cells pulsed with the E711-20 peptide or TC-1 cells transfected with the HLA-A2 gene. In addition, the mHSP110-E711-20 complex elicited stronger ex vivo and in vivo antitumor responses than either emulsified complete Freund's adjuvant or HSP70-chaperoned E711-20 peptide. These collective data suggest that HSP110 is a promising immunomodulator candidate for peptide-based human cancer vaccines, such as for the HLA-A2-restricted E711-20 epitope.

Structure-based prediction of MHC-peptide association: algorithm comparison and application to cancer vaccine design

Peptide vaccination for cancer immunotherapy requires identification of peptide epitopes derived from antigenic proteins associated with the tumor. Such peptides can bind to MHC proteins (MHC molecules) on the tumor-cell surface, with the potential to initiate a host immune response against the tumor. Computer prediction of peptide epitopes can be based on known motifs for peptide sequences that bind to a certain MHC molecule, on algorithms using experimental data as a training set, or on structure-based approaches. We have developed an algorithm, which we refer to as PePSSI, for flexible structural prediction of peptide binding to MHC molecules. Here, we have applied this algorithm to identify peptide epitopes (of nine amino acids, the common length) from the sequence of the cancer-testis antigen KU-CT-1, based on the potential of these peptides to bind to the human MHC molecule HLA-A2. We compared the PePSSI predictions with those of other algorithms and found that several peptides predicted to be strong HLA-A2 binders by PePSSI were similarly predicted by another structure-based algorithm, PREDEP. The results show how structure-based prediction can identify potential peptide epitopes without known binding motifs and suggest that side chain orientation in binding peptides may be obtained using PePSSI.

Secondary anchor substitutions in an HLA-A*0201-restricted T-cell epitope derived from Her-2/neu

We investigated analogues of GP2 (IISAVVGIL), an HLA-A*0201-restricted T-cell epitope derived from residues 654-662 in the tumor-associated antigen (TAA) Her-2/neu. One limiting factor of GP2 is its poor affinity for HLA-A*0201. Conformational analysis revealed the P5-P7 region in GP2 appears to be linked to the stability of P9 side chain interaction with the MHC molecule. To identify variants of GP2 with enhanced presentation to HLA-A*0201, we tested V6S, V6T, V6Q, G7P, G7F, T6F7, and Q6F7 for their capacity to stabilize cell surface HLA-A*0201 molecules. Of the mono-substituted variants, V6Q and G7F exhibited superior stabilization as compared to GP2. Molecular dynamics simulations suggest the improved binding can be attributed to concerted motions in the central and C-terminal regions of the peptide. These data support the notion that amino acids in HLA-A*0201 epitopes may be inter-dependent. Priming HLA-A*0201 transgenic mice with G7F-loaded syngeneic dendritic cells stimulated mouse T cells to produce a higher level of INFgamma than mice immunized with GP2.

Successful identification of novel agents to control infectious diseases from screening mixture-based peptide combinatorial libraries in complex cell-based bioassays

Mixture-based peptide synthetic combinatorial libraries (SCLs) represent a valuable source for the development of novel agents to control infectious diseases. Indeed, a number of studies have now proven the ability of identifying active peptides from libraries composed of thousands to millions of peptides in cell-based biosystems of varying complexity. Furthermore, progressing knowledge on the importance of endogenous peptides in various immune responses lead to a regain in importance for peptides as potential therapeutic agents. This article is aimed at providing recent studies in our laboratory for the development of antimicrobial or antiviral peptides derived from mixture-based SCLs using cell-based assays, as well as a short review of the importance of such peptides in the control of infectious diseases. Furthermore, the use of positional scanning (PS) SCL-based biometrical analyses for the identification of native optimal epitopes specific to HIV-1 proteins is also presented.