On the relation between promoter divergence and gene expression evolution

Open Access
Publisher Website
Google Scholar
Abstract
Recent studies have characterized significant differences in the cis ‐regulatory sequences of related organisms, but the impact of these differences on gene expression remains largely unexplored. Here, we show that most previously identified differences in transcription factor (TF)‐binding sequences of yeasts and mammals have no detectable effect on gene expression, suggesting that compensatory mechanisms allow promoters to rapidly evolve while maintaining a stabilized expression pattern. To examine the impact of changes in cis ‐regulatory elements in a more controlled setting, we compared the genes induced during mating of three yeast species. This response is governed by a single TF (STE12), and variations in its predicted binding sites can indeed account for about half of the observed expression differences. The remaining unexplained differences are correlated with the increased divergence of the sequences that flank the binding sites and an apparent modulation of chromatin structure. Our analysis emphasizes the flexibility of promoter structure, and highlights the interplay between specific binding sites and general chromatin structure in the control of gene expression. ### Synopsis It is widely accepted that phenotypic differences between closely related species often arise from differences in gene expression, but the principles of evolution of gene expression are largely unknown. Recent studies have used two complementary approaches to characterize the process of gene expression evolution. First, the genome‐wide expression program of related species was measured using microarray technology. Second, instances of loss (or gain) of cis‐ regulatory elements were characterized by analyzing the promoter sequences of orthologous genes in related species. In this work, we wished to understand the connection between these two approaches. Specifically, we asked to what extent the loss of an apparently functional binding site in gene promoter can predict a change in gene expression. As a first step, we used publicly available data. Surprisingly, we found that genes predicted to have lost an apparently functional binding site from their gene promoter still maintained their expression pattern ([Figure 1][1]). This was found for both yeast and mammals, and also when using different expression data sets (e.g. tissue‐specific expression in human versus chimpanzee or in human versus mouse). We reasoned that the apparent lack of influence of binding‐site divergence on gene expression could result from interactions between the multiple regulators that regulate the expression of typical genes. To examine this result in a more controlled setting, we focused on the yeast mating response, which is governed by a single transcription factor (TF) (STE12), and measured the transcription response to pheromone in four closely related yeast species. This analysis identified ∼400 genes that are differentially expressed between the species. Since the promoter sequence to which STE12 binds is well characterized, we were able to characterize also the instances in which this sequence was lost from a gene promoter in only one of the species. Notably, in this case, changes in STE12‐binding sequences could explain ∼50% of the expression differences between the species (Figure 4). This is a much larger fraction compared to the more complicated scenario described before, but is still far from being complete. To try and understand the origin of the remaining fraction of gene expression differences that were not explained by the divergence of STE12‐binding sites, we analyzed promoter regions flanking the binding sites. Interestingly, we found that these flanking sequences are highly diverged among genes with conserved STE12‐binding sites but diverged expression patterns. These flanking sequences could thus influence the chromatin structure and accessibility of STE12‐binding sites. Using a computational model that predicts nucleosome positions from the underlying DNA sequence, we showed that divergence of these flanking sequences could indeed alter the accessibility of several STE12‐binding sites and thus generate the observed expression changes. Taken together, our results suggest a complex interplay between the evolutionary divergence of cis ‐regulatory elements in gene promoters and the divergence of the associated gene expression. First, only rarely does the loss of an apparently functional binding site leads to the loss of gene expression. Promoters appear to be highly flexible and can tolerate such changes, perhaps through interactions with adjacent binding sites to other TFs. Second, expression can change even when the binding sites are conserved. This may reflect other changes in promoter sequence, and is likely to be mediated by changes in chromatin structure. This and other studies thus emphasize the influence of chromatin structure in driving gene expression evolution, but further studies are required to more rigorously describe its role. Mol Syst Biol. 4: 159 [1]: #F1