Comparative analysis of mutation frequency of coding and non coding short mononucleotide repeats in mismatch repair deficient colorectal cancers.

Abstract

Coding repeats are usually short mononucleotide tracts (SMT) of 10 or less base pairs in size. A number of such sequences contained in genes suspected to play a role in human carcinogenesis have been found to be mutated in mismatch repair deficient tumors (MSI-H cancers). Because of the high background of instability characterizing these cancers and in the absence of functional criteria, the significance of most frameshift gene alterations is unclear. In the present work, we analysed a series of 22 transcribed but non-coding SMT that are thus unlikely to play a tumorigenic role. Their frequency of size alteration in germline DNA and in DNA from MSI-H and microsatellite stable (MSS) tumors were compared to those previously reported in a series of 25 coding SMT of similar size. Non-coding SMT were either monomorphic or polymorphic in germline DNA whereas coding SMT were all monomorphic. In MSI-H tumors, non-coding SMT showed infrequent alterations (0-44%), as opposed to coding SMT which were altered at extremely variable frequencies (0 to 92%). Seven of the 22 non-coding SMT were monomorphic in MSS tumors but presented size alterations in MSI-H tumors with variable frequencies (3-28%). They were thus selected for further comparative statistical analyses for instability in coding SMT in MSI-H colorectal cancers. Only seven out of 25 of the coding SMT showed a significantly higher mutation frequency in these tumors. In the absence of functional criteria, we propose this as a novel and comprehensive approach for distinguishing real target genes amongst the numerous proposed gene mutations. It should allow identification of those that are genuinely selected for during MSI-H tumoral progression from others that play a less important role, if any, in MSI-H carcinogenesis.