A mutation usually means “the change in one base” [one nucleotide in the haploid genome; one base-pair (bp) in the diploid genome], but sometimes “mutations” can include any changes –– from a single bp alteration to many bases (two, five, 50, hundreds or thousands) inserted, deleted, and/or inverted). And it has long been known that mutations more readily occur in noncoding regions (introns, intra-genic areas of DNA) than in coding regions (DNA giving rise to the mRNA which in turn gives rise to the protein). The “mutational landscape of the genome” is very complex. Decades ago, traditional models of evolutionary and population genetics had (naïvely) proposed that mutations occur randomly with respect to time, genomic position, and mutation type. Under such models, mutations were expected to occur uniformly throughout the genome, and deviations from this expectation could potentially indicate selective pressures (DNA changes due to environmental pressures, conferring a resultant advantage, i.e. reproduction and survival).
Then CpG dinucleotides (CG on one strand; CG also on the other strand) were found to be mutational hotspots –– having enhanced mutation rates, subsequently shown to be driven by frequent deamination of methylated cytosines –– marked the beginning of the realization that the null model of mutation could not be true. More recently, next-generation sequencing (NGS) of tumors, families, and populations has identified a mutational landscape with biases –– ranging from the level of single-nucleotides to that of the whole genome. At the sequence level, it has now become clear that the probability of a given nucleotide mutating is primarily determined by its flanking upstream and downstream neighbors. Likewise, transcription factors and other DNA-binding proteins exclude repair factors from sequences that span 10-20-30 base pairs.
Chromatin states also influence mutability of sequences of hundreds to thousands of bp in length, DNA replication timing exerts significant pressures on the mutational properties of megabase-scale regions, and the sex and age of parents can influence genome-level patterns of de novo mutation in their gametes (sperm, ova). Interestingly, all of these levels of bias appear to interact with one another to generate a highly complex landscape of genomic mutation. A detailed understanding of this landscape is instrumental to studies of genetic inheritance, evolution, and tumorigenesis. The [attached] report provides insight into yet another layer of mutational complexity –– one that has been hiding right under our noses, yet has gone undetected owing to the difficulty of teasing apart mutational processes from signs of “natural selection”. This new study shows that, independently of purifying selection, mutation rates in gene exons are lower than predicted when analyzing their sequence composition. Rather, authors show that ‘fewer somatic mutations in exons than expected from their sequence content’ are caused by higher mismatch-repair activity in exonic than in intronic regions. These data have important implications in understanding the evolution of eukaryotic genes, and they have practical ramifications for the study of evolution of both tumors and species.
Nat Genet Dec 2o17; 49: 1684–1692 [article] & pp 1673–1674 [editorial]