Saturday, January 3, 2015

How to identify tumor-associated immunogenic peptides for immunotherapy?

Few weeks back journal Nature has published 5 separate papers in tumor immunology in the same issue. I am going to provide my assessment of the strengths and weaknesses of each paper and its approaches. 

I will start with the study led by Lelia Delamarre and Jennie Lill at Genentech. Genentech is one the oldest biotech company. It has a very good reputation and its scientists are usually trying to publish in high profile journals.

For this study the authors has performed whole exome-sequencing of two tumor models, MC-38 and TRAMP-C1. Initially, they have identified 1290 and 67 non-synonymous mutations in MC-38 and TRAMP-C1, corresponding to 170 and 7 neo-epitopes for MC-38 and TRAMP-C1, respectively (based on NETMHC3.4 algorithm prediction).

In parallel, the authors performed MHC class I peptide elution from these tumors for mass spec analysis. With this method, the authors found only 7 peptides for MC-38 (out of predicted 170 neo-epitopes) and 0 peptides for TRAMP-C1 (out of predicted 7 neo-epitopes).

Direct immunization with these peptide + adjuvant showed that two peptides, Reps1 and Adpgk displayed high immunogenicity (CD8 T cell priming potential), that was predicted by NETMHC3.4 algorithm.

Analysis of tumor-infiltrating lymphocytes (TIL) showed that Adpgk-specific CD8 T cells were the most abundant population (note that Reps1-specific T cells are the most abundant during peptide immunization in absence of tumor and Adpgk-specific T cells are the most abundant during peptide immunization in the presence of tumor).

Moreover, immunized with the 3 peptide-cocktail + adjuvant protected mice from MC-38 challenge (and protection was correlated with the frequency of Adpgk-specific T cells). 

In addition and more clinically relevant, immunization of tumor-bearing mice with 3 peptide-cocktail + adjuvant could augment Adpgk-specific T cells response and inhibit tumor growth.

In summary, the authors suggest that combining whole exome-sequencing with mass spec analysis of MHC class I peptide binding could improve identification of immunogenic peptides specific for tumors.

To be honest, I am a little bit confused with this paper. Even though the authors claim that they have combined whole exome-sequencing with mass spec analysis of MHC class I peptide binding, in the end, they have tested only peptides identified by mass spec. If one can analyse eluted peptides and determine its protein origin, then I do not see what was the function for whole exome-sequencing in the first place in this study? What about trying to test other peptides, at least few of them, predicted by algorithm only? It would have been much more useful, I would imagine. In addition, the authors found 0 immunogenic peptides for TRAMP-C1 with mass spec. How is zero useful here? What about treating the tumor cells with IFN-gamma to increase MHC class I expression to improve peptide recovery. 

David Usharauli


  1. Hi David, I imagine that immunizing with algorithm predicted peptides is a bit of fishing, whereas eluted MHC1 is relatively more definitive because they are there. If resource allows, algorithm peptides would definitely be interesting.

  2. Thanks for commenting. Sure, peptide elution is more definitive. However, for large scale application of personalized tumor immune landscape in clinical setting, reliable algorithms would help a lot (at least for cost reduction and affordability).