Both microsatellite length and sequence context determine frameshift mutation rates in defective DNA mismatch repair

It is generally accepted that longer microsatellites mutate more frequently in defective DNA mismatch repair (MMR) than shorter microsatellites. Indeed, we have previously observed that the A10 microsatellite of transforming growth factor beta type II receptor (TGFBR2) frameshifts -1 bp at a faster rate than the A8 microsatellite of activin type II receptor (ACVR2), although both genes become frameshift-mutated in >80% of MMR-defective colorectal cancers. To experimentally determine the effect of microsatellite length upon frameshift mutation in gene-specific sequence contexts, we altered the microsatellite length within TGFBR2 exon 3 and ACVR2 exon 10, generating A7, A10 and A13 constructs. These constructs were cloned 1 bp out of frame of EGFP, allowing a -1 bp frameshift to drive EGFP expression, and stably transfected into MMR-deficient cells. Subsequent non-fluorescent cells were sorted, cultured for 7-35 days and harvested for EGFP analysis and DNA sequencing. Longer microsatellites within TGFBR2 and ACVR2 showed significantly higher mutation rates than shorter ones, with TGFBR2 A13, A10 and A7 frameshifts measured at 22.38x10(-4), 2.17x10(-4) and 0.13x10(-4), respectively. Surprisingly, shorter ACVR2 constructs showed three times higher mutation rates at A7 and A10 lengths than identical length TGFBR2 constructs but comparably lower at the A13 length, suggesting influences from both microsatellite length as well as the sequence context. Furthermore, the TGFBR2 A13 construct mutated into 33% A11 sequences (-2 bp) in addition to expected A12 (-1 bp), indicating that this construct undergoes continual subsequent frameshift mutation. These data demonstrate experimentally that both the length of a mononucleotide microsatellite and its sequence context influence mutation rate in defective DNA MMR.