Impact of proteasomal processing on immunopeptidome repertoires and therapeutic immune recognition.

Abstract

Acute myeloid leukemia (AML) is the most common acute leukemia in adults. AML often responds to initial chemotherapy, but a majority of patients will relapse with resistant disease leaving a critical need for more effective therapies. Hematopoietic stem cell transplantation can prevent relapse long term through donor T cell-mediated elimination of leukemic cells. However, this anti-leukemic effect is unpredictable, and transplant can have detrimental side effects, including graft versus host disease (GVHD). New therapies focus the anti-leukemic effect and bypass potential GVHD by using engineered T cells with a defined T cell receptor (TCRs) that recognize proteasome-generated peptide fragments from tumor-associated proteins. Such ‘TCR-T’ cells rely on the targeted peptide to be processed and then presented by the restricting human leukocyte antigen (HLA). Cells can express standard- and immuno-proteasome isoforms, which can generate ‘distinct’ (isoform-dependent) or ‘mutual’ (isoform-independent) peptides. We investigated a clinical scenario in which the targeted AML modulated proteasome composition to eliminate processing of the targeted ‘distinct’ peptide, leading to loss of recognition by the adoptively transferred TCR-T cells and AML progression. Instead, an alternative TCR recognizing a ‘mutual’ peptide could respond to the progressive AML. Our data point to a mechanism whereby AML can evade immune recognition through modulation of proteasome isoform expression. This further implies that TCR-T cell therapy could be enhanced if tailored to target ‘mutual’ peptides. To determine the relative proportions of HLA-restricted ‘distinct’ and ‘mutual’ peptides and to identify promising TCR-T cell targetable peptides less likely to evade recognition, we utilized an emerging peptide discovery platform, Artemis, that employs soluble HLAs to bypasses immunoprecipitation of endogenous HLAs. Artemis-identified peptides revealed that the proportions of ‘mutual’ and ‘distinct’ peptides were specific to each HLA allele, with HLA-A*02:01 presenting the fewest ‘mutual’ peptides and HLA-A*11:01 presenting the most. We recovered 37 peptides from three AML-associated proteins of interest, 19 (51%) were ‘mutual’ peptides and some peptides (11%) were previously identified. Our findings demonstrate that defining isoform-specific proteasomal processing is critical to optimal peptide selection for TCR-T cell therapy and Artemis is an effective platform to identify ‘mutual’ peptides.