Common genetic variance at the SMAD9 colorectal cancer risk locus influences tumour mutational phenotype

Abstract

I set out to investigate whether common germline variance associated with colorectal cancer (CRC) risk can influence the somatic mutational landscapes of CRC. Some Mendelian cancer syndromes, such as Lynch syndrome (LS) and MUTYH associated polyposis, associate with defined mutational landscapes and this association can be functionally informative. In this thesis I investigated instead the influence of common risk variants such as those identified by GWAS approaches, the functional impacts of which are currently imperfectly understood. I identified 8 CRC risk loci for investigation that also exhibited eQTL effects in normal colorectal mucosa. Comparisons of tumour mutational loads at each of these showed a significant relationship with the SMAD9 (rs493248) locus and this locus was selected for further investigations. These revealed a relative depletion of highly mutated microsatellite instable (MSI) tumours on the risk allele background. This was verified in a meta-analysis of 5 independent cohorts (OR=0.73, 95%CI 0.63-0.86, p=1x10-4). Separate case-control studies for MSI and microsatellite stable (MSS) CRC risk at the SMAD9 risk locus indicated that the risk allele associated specifically with MSS CRC risk. A corresponding lower risk was demonstrated for MSI CRC (LS excluded) and suggested instead a protective association for this subtype. The latter was not significant, however this comparison lacked statistical power. I further explored two mechanistic hypotheses for these observations. SMAD9 forms part of the Transforming Growth Factor Beta (TGFb) signalling pathway, with tumour suppressive functions in early CRC. I suggested that the eQTL may selectively favour tumours with or without further mutations of TGFb signalling dependent on SMAD9 expression. The correlation with MSS and MSI CRC would come about since these tumours tend to regulate this pathway differently. Comparison of mutations in CRC driver genes demonstrated a correlation of TGFb signalling mutations with the SMAD9 locus but lacked causal support. Subsequent knock-down of SMAD9 in MSI and MSS cell lines did however not indicate a proliferative advantage for MSS CRC. Secondly, I suggested a possible alternate role for SMAD9 risk variance in increasing DNA mismatch repair activity. Comparison of single base substitution signatures showed a depletion of DNA mismatch repair signatures among MSS tumours from risk allele carriers. This appeared to support my second hypothesis although results require replication. In conclusion, I discovered a robust correlation between CRC microsatellite instability and variance at the SMAD9 CRC risk locus and eQTL. This contributes to our functional understanding of this locus and the genetic basis of CRC susceptibility.