Improving the selection of candidate CD8+ T cell epitopes for immunotherapy
Posted 6th January 2020 by Joshua Sewell
Alongside the talks at our conferences, the poster presentations are a huge part of the knowledge sharing that takes place. We’re thrilled to be able to take a closer look at this poster from the Research & Technology Series, presented by Wim van Esch and the team at Sanquin.
Infected and cancerous cells can be recognized and killed by CD8+ (cytotoxic) T cells. They are recognized via the foreign/altered proteins they produce, which are presented on the cell surface as peptides in the context of HLA class I.
These peptides are therefore key candidates for the development of vaccines, immunotherapeutic interventions and monitoring of disease-specific immune responses. The HLA-peptide binding is the most restrictive step in the peptide presentation pathway.
Thus the binding affinity of the peptide for the binding cleft of a given HLA complex is a strong indication of whether the peptide is a potential epitope or not. Narrowing the list of potential (neo)epitopes and reducing the duration time of the identification process are currently major unresolved challenges. We have developed techniques/assays which can speed up (neo)antigen identification and be used for characterization/immunomonitoring of (neo)antigen-specific CD8+ T cell responses in patients using limited cell material.
For prediction of peptide binding to HLA class I alleles prediction methods have been developed, among others the widely used NetMHCpan 4.0. We have shown that by adding an in vitro HLA peptide-binding assay, based on UV-mediated peptide exchange technology*, selection of relevant candidate epitopes can be improved significantly.
In silico-predicted binders and non-binders to e.g. HLA-A*02:01 (n=795) were tested for real binding. Comparison of in vitro peptide binding and in silico binding prediction revealed that 17% of in silico-predicted binders to HLA-A*02:01 do not bind. Furthermore, 17% of in silico-predicted non-binders turned out to be binders. NetMHCpan 4.0 does not discriminate well between strong and weak binders. Ranking for best binders differs significantly. Similar results were found for other HLA alleles.
Interestingly, it was shown that the in vitro HLA peptide-binding assay predicts (neo)antigen immunogenicity. Neo-peptides (n=47) from 2 melanoma patients predicted to bind to HLA-A*02:01 were tested for immunogenicity by Erlend Strønen et al (Science 2016; 352:1337-41). We tested these neo-peptides for binding using our in vitro HLA peptide-binding assay and subsequently ranked. Furthermore, HLA-B*27:05-restricted hepatitis C virus (HCV)-associated predicted binders (n=54) were screened for real binding and subsequently in vitro tested by Jörg Timm et al (Institute of Virology, Essen, Germany) for induction of CD8+ IFNɣ responses in PBMCs from HCV patients.
In both cases ranking based on in vitro binding improved selection of candidate epitopes significantly (i.e. clustering of immunogenic peptides in the upper range of best binders). This has major consequences when candidate epitopes are selected for development based solely on in silico binding predictions: potential good candidates are missed while resources are unnecessarily spent on following up non-binders.
For identification of immunogenic peptides and monitoring treatments, we have developed a multiplex FACS technique for simultaneous enumeration of up to 41 different antigen-specific CD8+ T cell responses in a single cell sample using HLA tetramer combinatorial coding* (HTCC). This allows comprehensive epitope screens on limited patient samples and study of kinetics (quantification of antigen-specific T cells) and broadness of the T cell response detected (T cell responses against single or multiple epitopes).
HTCC combined with phenotypic and/or intracellular staining allows further characterization of the antigen-specific T cell responses. In these assays HTCC is combined with e.g. CD45RA, CD27, CCR7, PD-1, and KLRG1 and/or intracellular staining of e.g. transcription factors (T-bet and Eomes) and granzyme B. Alternatively, staining of peptide pool-induced IFN-gamma, TNF-alpha and CD107a can be coupled with HTCC (antigen specificity) and phenotype. These assays give extra insight into the quantity and quality of cytotoxic T cell responses detected.
1) Ranking based on in vitro HLA peptide-binding improved the selection of candidate CD8+ T cell epitopes significantly (best correlation with immunogenicity) initially identified by in silico analysis;
2) HTCC combined with phenotypic and/or intracellular staining can contribute to in-depth characterization of antigen-specific T cell responses important for immunotherapeutic development and monitoring of immunotherapeutic interventions.
In vitro HLA peptide-binding assay and HLA tetramer combinatorial coding technique improve selection of candidate CD8+ T cell epitopes for immunotherapy
Juk Yee Mok, Giso Brasser, Dionne Geerdes, Astrid Visser and Wim van Esch. Dept. Product Development, Sanquin Reagents B.V., Amsterdam, The Netherlands
*Developed in collaboration with The Netherlands Cancer Institute; patent protected in Europe, US and other countries WO 2010/060439 and WO 2006/080837; licensing or testing services to be offered by Sanquin Reagents B.V.
To view posters, slides, and watch presentations from previous conferences, visit our resources page.
Leave a Reply