Mismatch repair deficiency (MMRd) occurs in ~15% of colorectal cancers (CRC). Its causes can be sporadic or familial (Lynch syndrome). MMRd results in a high burden of single nucleotide variants and indels and microsatellite instability (MSI). MMRd CRC demonstrate profound clonal complexity and evolvability under immune selection. High mutational load translates into elevated neoantigen numbers which is thought to be the basis of their excellent response to immune checkpoint inhibitors (ICIs). However, for reasons yet unclear, ~50% of MMRd CRC do not respond to ICIs. Thus, improved biomarkers for patient stratification are required. The aim of my project is to understand genomic differences between MMRd genotypes that likely contribute to the variable clinical outcomes.

CRISPR-Cas9 was used to knock out four genes implicated in MMR (MLH1, MSH6, MSH3, MBD4) in different combinations in a human colonic epithelium cell line (HCEC). The gene knockouts were validated through DNA sequencing (Sanger and Illumina MiSeq) and Western blotting. The cell lines were subjected to two rounds of subcloning and 3-4 weeks of mutation accumulation, followed by WGS at 30x. The cell doubling time was investigated using live-cell fluorescent imaging and flow cytometry. The WGS dataset was processed using an established bioinformatic pipeline for variant calling and mutational signature discovery. CUT&TAG to profile the distribution of H3K36me3 was conducted on select cell lines and analysed with a published protocol (Zheng Y et al (2020). Protocol.io).

29 colonic epithelial cell lines were produced that model eight MMRd genotypes observed in clinical setting. Cell cycle profiles and cell doubling times were demonstrated to be similar, irrespective of the genotype. Evidence from Western blot experiments suggested that knockout of one MMR gene in cells does not significantly affect expression levels of other MMR genes. WGS data analysis (in progress) yielded reliable variant counts proving that the cell lines exhibit hypermutation, reproduced characteristic MMRd mutational signatures and provided novel insight into the mutational profile of MSH3. Preliminary CUT&TAG data processing revealed similar H3K36me3 distributions in WT and MSH6 -/- lines and identified peaks of marker abundance that will be used to assess mutations in transcribed genomic regions.

A suite of valuable cell models was established to study genomic impact of MMRd. Further analysis of the WGS dataset obtained through carefully planned rounds of cell line subcloning, as well as the complementary CUT&TAG data, will provide detailed insights into the burden, rate and types of mutation caused by various forms of MMRd. Overall, this work adds to the view of MMRd as a multifaceted genetic phenomenon and the catalogue of mutational signatures which carries potential as future biomarkers.