The vast majority of cancers result from positive selection of so-called "driver" somatic events, with negative selection playing only a marginal role. This conclusion, established from thousands of tumors, nevertheless relied on the analysis of non-repetitive DNA sequences. However, there exists a subgroup of tumors characterized by microsatellite instability (MSI), a consequence of a deficiency in the DNA mismatch repair (MMR) system. This phenotype, first described in Lynch syndrome and subsequently in sporadic forms of colorectal, gastric, and endometrial cancers, affects thousands of repetitive microsatellite sequences. Because these sequences constitute mutational hotspots regardless of their location and of the function of the genes involved, some of these frequent mutations could be neutral, or even detrimental to the tumor process—a hypothesis that has so far been little explored.
To test it, the authors performed whole-exome sequencing of 47 MSI colorectal cancers, with confirmation in an independent cohort of 53 tumors. They developed a probabilistic model of mutational events within microsatellites, while adapting pre-existing models to the analysis of non-repetitive sequences. The functional impact of coding alterations subject to negative selection was then assessed in colorectal cancer cell lines, using in vitro RNA interference and xenograft models, after which their clinical relevance was examined in a third cohort of 164 patients.
This work demonstrates the simultaneous existence of positive and negative selection of somatic mutations in repetitive sequences, allowing the identification of the true driver genes of the MSI tumor process. Conversely, no negative selection was observed for non-repetitive sequences, in line with the established dogma. Five coding mutational events subject to negative selection (affecting WNK1, HMGXB4, PRRC2C, RFC3, and GART) proved deleterious to tumor cells, affecting apoptosis, proliferation, or migration, with additive effects when several targets were inactivated jointly. Paradoxically, the rare tumors harboring one of these mutations despite counter-selection were associated with a poorer prognosis (hazard ratio of 3; 95% CI: 1.1–7.9; p = 0.03), suggesting that their anticancer effect would be offset by other, as yet unknown, oncogenic processes.
These findings indicate that the genomic instability of MSI colorectal cancers plays a dual role in cellular transformation, generating both favorable driver mutations and deleterious, counter-selected events. The authors highlight the limitations related to the size of the series analyzed by exome sequencing and call for confirmation of these target-gene signatures in larger cohorts.