Whilst protein coding mutations have been well-catalogued and characterised, changes to cis-regulatory sequences have only recently come to the forefront of gene expression research. Despite this, only a fraction of those interactions in a limited number of organisms, in selected tissues and developmental stage, and under specific conditions, are known. Alterations to cis-regulatory elements can change the course and pattern of gene expression and pave the way for the evolution of species-specific traits[395, 396].
Most components of the transcriptional machinery that regulate gene expression are highly conserved in evolution. As outlined earlier in 1.3, TFs and the sequence of their binding motifs are mostly conserved between human and fruit flies[397-399], and comparisons of the sequence of TFs motifs across different lineages yield a high degree of similarity[394]. It is the cis-regulatory elements, such as
conserved. An examination of the activation profiles in 41 pairs of conserved regulatory elements between human and zebrafish revealed that only a third of these regions displayed any form of conserved activity between the two species[400]. A major study by Villar et al. discovered that turnover of the regulatory activity is pervasive between even more closely related species (Figure 1.8). For example, only 1% of human liver enhancers exhibit consistent activity across the 20 mammals investigated[401]. Taken together, these studies suggest that orthologous regulatory regions display a varied level of activity across species, even though the TFs they bind and their motifs are very much conserved in structure.
It has been suggested that the evolutionary conservation of genes and their regulation is a product of pleiotropic trade-off[399, 402-406]. Pleiotropy is the ability of a single gene/variant to influence several traits simultaneously. If a pleiotropic region is altered by mutations, newly-created variants may confer different effects on the multiple functions they contribute to. These variants may be advantageous in one facet, but gravely deleterious in others[406]. Intuitively, these regions should be under selective pressure to remain more evolutionarily constrained than other, non-pleiotropic regions, and it follows that genes with non-pleiotropic functions are more likely to be found in orthologous regions in other species, and with a comparable expression level[403, 404]. Therefore, TF binding sites that are active tissue-wide and observed at several developmental stages are expected to be conserved between species[399].
Nevertheless, the sequences TFs bind to vary in terms of their information content which correlates strongly with how their occupancy evolves. Genetic drift causes low-affinity, information-poor motifs to evolve rapidly[6, 394, 407-410], but sequence change alone is incapable of fully explaining the evolutionary trajectory of TF binding[407, 408, 411]. Larger, information-rich motifs, such as the CTCF motif, are selectively conserved[269].
Increasingly, investigations into the evolution of mammalian transcriptional regulatory elements are documenting the rapid turnover of enhancers and tissue-specific TF binding sites[412-416]. Findings demonstrate that gene expression across similar tissues in different species is more correlated than different tissues within the same species, suggesting that tissue-specific expression pattern are evolutionarily stable[417, 418]. How this is achieved and maintained in the face of the ever-changing regulatory landscape is a central dilemma in evolutionary genomics.The evolutionary status of regulatory elements ranges from highly conserved to lineage-specific. Taken together, these various findings support the notion that the greater functional influence of a regulatory element, the more conserved across different species, tissues and developmental stages[401, 414, 419, 420]. Conversely, lineage-specific elements appear
to partially compensate for proximally lost events, and are often found in regions with pre-existing regulatory activity[394, 421].
An integrated analysis of transcriptional circuit evolution across >25 animal species examined mRNA expression, transcription factor binding and cis-regulatory motifs. The results revealed that transcriptional regulatory networks evolve at a constant rate across the various lineages, even more so when only chromatin-accessible regions were considered[422]. Another more recent study compared conserved-activity enhancers to species-specific-activity enhancers using liver enhancers in ten diverse mammalian species. Conserved-activity enhancers exhibited greater regulatory potential and activity in humans than their species-specific counterparts. They appeared active across more cellular contexts and the genes they regulated were expressed in more tissues, providing further support to the pleiotropy argument mentioned earlier[423]. Bertholet et al. followed up their 2015 study by analysing promoter and enhancer activity with corresponding gene expression levels in liver samples from 15 species, and reached similar conclusions[424]. They also reported that the evolutionary resilience of transcription is dependent on the number of regulatory elements, with an emphasis on evolutionary conservation. Elements with conserved activity in more species have the most ability to drive stable gene expression. Recently-evolved species-specific enhancers, on the other hand, have a weaker overall regulatory potential[424].
The mechanisms of gene regulation can be influenced by cis- and trans-acting variation with local and pleiotropic effects, respectively. These changes can instigate a much wider effect resulting in changes to gene regulation evolutionary dynamics[425].
A study used the analysis of RNA-seq to measure liver gene expression divergence between two mouse strains: C57BL/6J (Mus musculus domesticus) and CAST/EiJ (Mus musculus castaneus) to establish the extent of allele-specific expression in their F1 hybrid offspring. 535 genes were identified which displayed a parent-of-origin-specific patterns of gene expression, but only few of those genes suffered complete allelic-silencing, indicating that genetic imprinting in somatic mouse tissues accounts for a relatively small number of genes[260]. 32% of non-imprinted genes demonstrated divergent expression between the parental F0 strains., of which only 2% were found out to be exclusively influenced by trans-acting variants. 43% of the set of non-imprinted genes were attributed to variants acting only in cis. The remainder of genes (55%) showed gene expression divergence pattern consistent with a combinatory complex of cis- and trans-acting variation. The genes whose expression divergence is driven by trans-acting variation were additionally observed to have higher sequence constraint than genes whose divergence was caused by variants acting in cis. Gene
opposite directions, suggesting that compensatory regulation due to purifying selection may work to stabilize gene expression levels[260].
Figure 1.8: Enhancer and promoter evolution in 20 mammalian species
Comparative evolutionary genomic analysis in 20 mammals reveals rapid enhancer (orange triangles) and slow promoter (purple triangles) evolution across the evolutionary tree. Enhancers are only rarely constrained, and recently-evolved enhancers are predominant in their regulatory landscape, exhibiting lineage- specific positive selection. Grey arrows on the left indicate divergence time since last common ancestor in millions of years. Figure adapted from Villar et al. [401].
A more recent effort investigated tissue-specific TF (CEBPA, HNF4A, FOXA1) occupancy divergence in the livers of the same two strains and their F1 hybrid to highlight the contributions of cis and trans variation on the dynamics of gene regulation[257]. They also identified that cis-directed mechanisms are
predominant in the birth of new TF binding sites in lineage-specific manner.
Furthermore, they detected apparent coordination in the regulatory networks between TF occupancy, chromatin state, and gene expression in the F1 hybrids[257].