Abstract:

Exonuclease domain pathogenic variants of DNA polymerase epsilon (POLE) represent a major clinical challenge. At the constitutional level, they predispose carriers to cancer, particularly colorectal cancer. At the tumoral level, they confer an ultramutated tumor phenotype characterized by an extremely high tumor mutational burden, a molecular signature associated with increased sensitivity to immune checkpoint inhibitors. However, interpretation of many missense variants remains difficult, limiting their use into clinical oncogenetics and precision oncology.

To address these challenges, we combined two complementary approaches. First, a high-throughput functional assay based on saturation mutagenesis of the exonuclease domain of Saccharomyces cerevisiae Pol epsilon allowed systematic evaluation of amino acid substitutions. After validation with known deleterious variants, this strategy generated the first comprehensive functional map of the domain. It revealed that pathogenic variants cluster in critical structural regions: (i) the catalytic pocket, (ii) a loop modulating the positioning of the 3′ end of the nascent strand, and (iii) a helix-loop-helix structure separating the exonuclease and polymerase domains. Outside these regions, most residues tolerated substitutions. Comparison with both germline and somatic clinical data confirmed the relevance of the model, while highlighting the limitations of bioinformatic predictors when faced with direct experimental evidence.

Second, we performed an integrated proteogenomic analysis on a cohort of tumors including POLE-mutated, MSI, and germline POLE variant cases. This approach, combining whole-exome and transcriptome sequencing, proteomic profiling, and bioinformatic prediction of neoantigens, yielded several insights: POLE-mutated tumors showed an excess of missense variants compared with MSI tumors and a higher predicted neoantigen load. Incorporation of HLA typing underscored the importance of the host’s immunogenetic background in shaping neoantigen presentation. Importantly, proteomic analysis confirmed the expression of several neoantigenic peptides derived from somatic variants, validating their immunogenic potential and reinforcing the link between hypermutability and clinical response to immunotherapy. Unlike exome–transcriptome analyses alone, often limited by the absence of detectable expression for certain variants, the addition of proteomics increased confidence in neoantigen prediction and uncovered neoantigens that would otherwise have been missed.

In conclusion, combining a functional atlas of POLE variants with integrated proteogenomic characterization of tumors provides insights into both structure–function relationships and immunological consequences of these variants. This dual approach represents a valuable resource for clinical annotation, refines prediction of immunotherapy response, and paves the way for personalized therapeutic strategies, including the development of vaccines or cell-based therapies targeting specific neoantigens.

Keywords: DNA polymerase epsilon; exonuclease; high throughput functional assay; neoantigens; proteogenomic; functional atlas