Mutation Patterns in the Human Genome: More Variable Than Expected

Abstract

The development, survival, and reproduction of an organism depend on the genetic information that is carried in its genome, yet the transmission of genetic information is not perfectly accurate: new mutations occur at each generation. These mutations are the primary cause of the genetic diversity on which natural selection can operate, and hence are the sine qua non of evolution. A better knowledge of mutation processes is crucial for investigating the causes of genetic diseases or cancer and for understanding evolutionary processes. This knowledge is also important for different practical reasons. First, comparative sequence analysis is widely used to find functional elements within genomes. The basic principle of this approach is that functional elements are affected by natural selection, and hence can be recognized because they evolve either slower or faster than expected given the local mutation rate. Hence, to be able to annotate genomic sequences, it is necessary to have a good knowledge of the underlying pattern of mutation. Moreover, this knowledge is also essential for ensuring the accuracy of the methods that analyze sequence divergence to determine the phylogeny of species or the demographical history of populations. Finally, the study of mutational processes also provides valuable information about genome function in processes such as replication, repair, transcription, and recombination. During the last few years, several important factors affecting mutation rates have been uncovered. However, a paper in this issue of PLoS Biology [1] reveals an unexpected additional layer of complexity in the determinants of mutation rates. A priori, nucleotide mutation rates are expected to depend upon three factors [2]: (i) the intrinsic stability of nucleotides and their sensitivity to mutagenic agents; (ii) the fidelity of DNA replication; and (iii) the efficiency of the DNA repair machinery. The analysis of variations in mutation rate across genomes can shed light on the relative contribution of these different factors and on the genomic features that affect mutation rates. In mammals, current knowledge of mutation processes derives essentially from the analysis of a limited number of germ-line mutations responsible for human genetic diseases [3] and from phylogenetic studies. This latter approach consists of comparing homologous sequences (between species or within populations) to estimate the number and kinds of changes that occurred since their divergence. At neutral sites—i.e., sites where the impact of natural selection is presumed to be null or very limited (pseudogenes, defective transposable elements, noncoding sequences, synonymous codon positions)—substitution rate is expected to be equal to the mutation rate [4]. This approach suffers from several limitations (see below), but thanks to the accumulation of genome sequences and polymorphism data, it has provided indirect estimates of genome-wide mutation patterns. Large-Scale Variations in the Rate and Patterns of Neutral Sequence Evolution Phylogenetic analyses show that in mammals, neutral rates of sequence evolution (measured in number of base changes per site and per year) vary at different scales. First, substitution rates vary between species. Notably, species with short generation time generally evolve faster, presumably because they experience more rounds of germ-cell divisions (and hence more DNA replication errors) during a given unit of time [5]. If most mutations are due to DNA replication errors, then mutation rates are expected to be higher in males than in females, owing to the greater number of cell divisions per generation in the male germ-line. In agreement with that prediction, in apes, substitution rate is two times higher on the Y chromosome than on the X, whereas autosomes show intermediate values (the three classes of chromosomes spend on average—over generations— respectively 100%, 33%, and 50% of their time in the male germ-line) [6,7]. The strength of this male mutation bias in different mammalian species appears to be correlated to the