A broadened HLA ligandome uncovers new immunotherapy targets for pancreatic cancer + A prime editor mouse to model a broad spectrum of somatic mutations

Abstract

Pancreatic cancer is a lethal malignancy recalcitrant to immune checkpoint blockade and other immunotherapies. A subset of tumors is computationally predicted to harbor potentially immunogenic peptides for MHC class I (MHC-I) presentation, but the nature, expression, and immunogenicity of these peptides has yet to be determined. The only prior study of the pancreatic cancer immunopeptidome focused on profiling MHC-I-associated peptides (MAPs) from canonical proteins in bulk tumor samples; however, non-malignant cell populations comprise most of the pancreatic tumor mass, obscuring the identity of MAPs that derive specifically from cancer cells. In the second chapter of this thesis, I resolve this challenge through extensive profiling of patient-derived organoids with whole-genome sequencing, RNA sequencing, and immunopeptidomics. These data enable a proteogenomics approach that tailors MAP identification to each individual patient sample. Harnessing this platform, my colleagues and I uncovered a diverse cohort of MAPs derived from somatic mutations and transcript isoforms that are functionally unexpressed in most or all healthy tissues. These include MAPs derived from novel, unannotated open reading frames (nuORFs) present within long noncoding RNAs, processed transcripts, and 5’ and 3’ untranslated regions. We found that cytotoxic T cells specific to nuORF-derived MAPs can be readily generated from peripheral blood mononuclear cells of healthy donor individuals. This result highlights the immunogenicity of nuORF-derived MAPs and establishes them as promising targets for immunotherapies in pancreatic cancer.

In Chapter 3, I report the development of a genetically engineered mouse model (GEMM) for performing prime editing in vivo. This system represents a rapid alternative to traditional cancer mouse models, which often take months or years to develop. Through a Creinducible prime editor enzyme encoded in the mouse germline, prime editor GEMMs can mediate rapid and precise engineering of most cancer mutations, including many that are challenging or infeasible to achieve with other CRISPR technology. We demonstrate the utility of this system by mediating secondary Kras mutations and common Trp53 hotspot mutations in model-derived pancreatic organoids. Finally, we model lung and pancreatic cancer in vivo using lentiviral delivery of prime editing guide RNAs or orthotopic transplantation of prime edited organoids. We anticipate that prime editing GEMMs will accelerate preclinical functional studies of cancer-associated alleles that are challenging to model by traditional approaches.