Evolutionary biologists have long sought to understand the genetic basis of evolutionary change. For theoretical reasons, biologists have predicted that mutations affecting the regulation of gene expression, rather than the physical structure of gene products, will play a central role in phenotypic evolution. Recent empirical work has begun to nail down the molecular details underpinning phenotypic evolution, and in a remarkable number of cases, the evidence strongly supports the theoretical models.
Two disparate schools of theory have predicted the central role for gene regulation in evolutionary change. The critical parameter for both accounts is pleiotropy, the phenomenon whereby a single mutation has multiple phenotypic effects.
One school of thought, growing out of mathematical population genetics and quantitative genetics, holds that mutations fixed by evolution are likely to be those with very limited pleiotropic effects (Fisher, 1930). Because extant organisms must be in some sense fit for survival, a mutation that alters many features at once is likely to change at least some of them for the worse. A mutation that alters only a single feature stands a better chance of improving the organism’s prospects, and therefore spreading through the populations.
Regulatory mutations, particularly those that affect transcriptional regulation, are ideal candidates for mutations with restricted pleiotropy. The temporal and spatial pattern of a gene’s expression, the quantitative profile of its transcription rate, and its responsiveness to a range of inductive cues are influenced by the binding of transcription factors to cis-regulatory DNA elements. Because the array of transcription factors differs among cells, each piece of the total expression pattern may be regulated independently by a unique complement of transcription factors and a discrete segment of regulatory DNA (Arnone and Davidson, 1997). Cis-regulatory mutations that affect expression in one part of the organism may leave expression in other parts unchanged. This modular property of transcriptional regulation makes cis-regulatory DNA conducive to evolutionarily important mutation (Gerhart and Kirschner, 1997; Davidson, 2001).
A second school of thought, growing out of evolutionary developmental biology and molecular evolution, holds that the mutations responsible for the conspicuous phenotypic transformations that account for the diversity of organismal form are likely to be those with quite radical pleiotropic effects. More specifically, a mutation with coordinated pleiotropic effects, radically altering form in a coherent and systemic fashion, is a mutation that could account for a dramatic change in form while retaining functional integration.
While evolutionary change affecting a suite of integrated traits could be due to a large number of separate mutations, a single mutation influencing the whole suite of traits is more likely to preserve functional integration, particularly when the traits are genetically correlated by a shared developmental basis. Even before the molecular mechanisms of development were well understood, this notion found footing in the concept of heterochrony (e.g., Gould, 1977), whereby simple changes in the timing of developmental events resulted in dramatic, but functionally integrated, changes in form. Of course, heterochronic changes in timing are likely to result from changes in the temporal or quantitative regulation of gene expression. The argument for coordinated pleiotropy is famously realized in King and Wilson’s (1975) model for human origins. Because humans and chimpanzees are so similar at the level of structural genes, they argued, the striking phenotypic differences between the species must be due to a small number of mutations with systemic effects.
At the same time, developmental biologists have found that the set of genes regulating development is largely shared among animals (and similarly among plants), a finding confirmed by the proliferation of whole-genome sequences and EST projects. Given a shared “toolkit” of developmentally important proteins, evolutionary developmental biologists have embraced the notion that changes in form are due to changes in the pattern of expression of the shared genes, and the recruitment of whole developmental networks to evolutionarily novel tasks (e.g., Lowe and Wray, 1997). Studies correlating changes in patterns of gene expression to changes in phenotypic form have become a mainstay of evolutionary developmental biology.
The theoretical commitments to limited pleiotropy on the one hand, and to pervasive, integrated pleiotropy on the other hand, have been brought together and reinforced by recent empirical work, which bridges the divide between the developmental biologists, who have focused on comparisons among species, and quantitative geneticists, who have been examining variation within species. Although case studies documenting the role of regulatory mutations in the genetic basis of phenotypic evolution are now numerous (e.g., Sucena et al., 2003; Beldade et al., 2002; Wang and Chamberlin, 2004; Wang et al., 1999; Enattah et al., 2002), an extraordinary recent study of stickleback fish (Gasterosteus aculeatus) exemplifies the role of gene regulation in meeting the competing needs of limited pleiotropy and systemic effects.
Shapiro et al. (2004) genetically mapped the loci underlying the dramatic and very recent evolutionary reduction of the pelvic skeleton in lake populations of sticklebacks. A virtue of this study is that the mapping strategy makes no a priori assumption about the cis-regulatory nature of the underlying locus. Shapiro et al. found that a single major locus accounts for a suite of pelvic traits, consistent with the coordinated pleiotropy model. The locus was identified as Pitxl, a transcription factor, but no differences were seen in the protein sequence. Instead, the fish with and without pelvic skeletons differ in the spatial pattern of Pitxl expression during development. Fish with reduced skeletons show a loss of expression at the site of the presumptive pelvic region. At the same time, expression of the gene in a variety of other developing organs is retained, consistent with the restricted pleiotropy model. Because the complete knockout of Pitxl (in mice) is lethal, the evolutionary importance of the spatially limited mutational effect can hardly be doubted.
Regulatory mutations fit the theoretical bill for evolutionarily important mutations. But phenotypic evolution certainly involves mutations of all sorts – indeed, recent work revealing the extent of positive selection on amino acid substitution in human evolution (Clark et al., 2003) shows that structural mutations are essential contributors to the origin of our own species, despite the predictions of King and Wilson (1975). If regulatory mutations are rare, then their importance to evolution, regardless of their suitability, may be minimal. Three recent findings suggest that regulatory mutations may actually constitute the majority of phenotypically relevant variation in populations. First, comparative genomics has revealed that a surprising fraction of noncoding DNA is conserved among divergent species (e.g., Rat Genome Sequencing Project Consortium, 2004). Because evolutionary conservation implies functional importance, it seems that most of the functional nucleotides (in mammalian genomes, at least) are noncoding and potentially cis-regulatory. These nucleotides, therefore, represent a bigger target for mutation than do coding sequences. Second, surveys of noncoding variants (again, in humans) suggest that a large fraction affect transcription (Rockman and Wray, 2002; Buckland et al., 2004). Because we lack a means to infer the functional import of cis-regulatory DNA directly from sequence, however, our understanding of the extent of regulatory variation remains limited. Third, genome-wide analyses of gene expression variation in a range of species have shown that such phenotypic variation is ubiquitous (e.g. Jin et al., 2001; Cavalieri et al., 2000). Although efforts to map the loci responsible for the variation are highly sensitive to experimental design and statistical analysis issues, one of the best studies to date found that the majority of such loci map to the same genetic location as the variable transcript, implying that the loci may be cis-acting (e.g., Brem et al., 2002). Allele-specific measurement of gene expression, a technique that exclusively discovers the effects of cis-acting variation, also reveals the presence of abundant variation in humans (Lo et al., 2003). Consequently, putting aside all theoretical commitments, we should not be surprised to find regulatory variation underlying phenotypic variation.
Empirical and theoretical results point to changes in gene regulation as a major factor in phenotypic evolution. As molecular quantitative genetics continues to make strides in connecting phenotype and genotype, our understanding of the molecular basis for evolutionary change will grow even richer.
