GC-content evolution in mammalian genomes: the biased gene conversion hypothesis.

BASE composition is spatially structured in mammaisochores may simply reflect variable mutation processes among genomic regions, consistent with the neutral lian genomes. From sodium chloride centrifugation experiments, Bernardi et al. (1985) defined three model. Such variation in mutational biases has never been demonstrated. Alternatively, isochores might be major classes of genomic fragments with low, median, and high GC content, respectively, and called them the result of natural selection. Bernardi and colleagues have long argued that the isochore structure is an adapisochores. This discrete description now appears artifitation to homeothermy since it has been found in mamcial. Analyses of the complete human genome (Intermals and birds but not amphibians and fish (e.g., Bernational Human Genome Sequencing Consortium nardi 1993). The discovery of isochores in crocodiles 2001) have dismissed the underlying hypothetical picand turtles (Hughes et al. 1999) led to the rejection of ture of sharp boundaries between long homogeneous this hypothesis. Moreover, selection, if any, must be fragments: GC content turns out to vary continuously, unrelated to gene expression: The GC content of genes and somewhat erratically, along chromosomes. These in humans does not correlate positively with their exanalyses, however, have confirmed the existence of a pression level or pattern (Goncalves et al. 2000), and highly significant spatial autocorrelation of GC content, even pseudogenes translocated into a GC-rich region with most of the structure detectable at a relatively large undergo an increase of GC content (Francino and (300-kb) scale. There is a strong correlation between Ochman 1999). Neutralists furthermore argued that, the GC content at third codon positions of codons given the small effective population size of mammalian (GC3) and the GC content of the region in which a species, a selective hypothesis must imply very high selecgene is located (Bernardi 2000). GC3 varies from a tion coefficients (typically 2 orders of magnitude higher typical 40% in low-GC-content regions to 80% and more than in Drosophila) at every position of GC-rich regions, in high-GC-content regions. GC content is correlated including introns and intergenic DNA (Sharp et al. with various genomic features, including repeat element 1995). The existence of such a selective pressure without distribution, methylation pattern (Jabbari and Bernardi apparent correlation with gene expression appeared 1998), and, most remarkably, gene density (Mouchiroud quite speculative. et al. 1991; Duret et al. 1995). GC-rich regions include On the other hand, recent analyses of human polymany genes with short introns while GC-poor regions morphism data sets (Eyre-Walker 1999; Smith and are essentially deserts of genes. This suggests that the Eyre-Walker 2001) unexpectedly contradicted the mudistribution of GC content in mammals could have some tational bias hypothesis. Under neutrality, the substitufunctional relevance, raising the issue of its origin and tion process (accumulation of changes in the long run) evolution. For brevity, in this article, we use the word should reflect the mutation process since each kind of “isochore” as an abbreviation for “the peculiar, strucmutation has an equal fixation probability. At equilibtured distribution of GC content in mammalian gerium, therefore, the number of AT → GC and GC → nomes.” AT mutations arising should be equal. Eyre-Walker, howTwo opposite views about isochore evolution have ever, found significantly more GC → AT than AT → been hotly debated over the last 20 years, as part of GC mutations. G and C alleles seem to have a selective the neutralist/selectionist controversy. One view is that advantage, making the substitution process more GC biased than the mutation process. These reports renewed interest in selective hypotheses for the evolution Corresponding author: N. Galtier, UMR 5000, “Génome, Populations, of isochores (e.g., Bernardi 2000). There are, however, Interactions,” Université Montpellier 2, CC 63, Place E. Bataillon, 34095 Montpellier Cedex, France. E-mail: [email protected] at least two alternative possible explanations for this