We have demonstrated a striking dissonance between the deep evolutionary conservation of core histone regulatory phenotypes and the profound divergence of their regulatory mechanisms. Core histone genes exhibit similar cell cycle (S phase specific) expression patterns from the yeast S. cerevisiae
to human (Figure ). This regulatory conservation is consistent with the high levels of sequence conservation among core histone proteins. Nevertheless, the regulatory mechanisms that are used to achieve the conserved expression patterns of core histone genes are almost entirely lineage specific. The cis-trans machinery involved in core histone gene regulation has changed substantially between lineages through gain and loss of transcription factor proteins and their cognate binding sites. This suggests that, for families like the core histone genes, phylogenetic footprinting [39
] may have limited utility for identifying functional regulatory elements across all but the most closely related species.
In addition to the divergence of cis sites and trans factors, a distinct level of post-transcriptional regulation of core histones emerged along the metazoan evolutionarily lineage [40
]. Core histone gene 3'-untranslated regions encode a stem loop structure (Figure ) that, when bound by protein, greatly increases mRNA stability. This mechanism is responsible for 70% of the upregulation of core histones in S phase. The sequence that forms the stem loop is conserved across metazoans (Figure ). The emergence of this mechanism may have allowed for some of the turnover of the cis-trans regulatory machinery among metazoan genomes subsequent to their divergence from the yeast evolutionary lineage.
Figure 9 Structure and conservation of the histone 3'-UTR stem loop. (a) Schema of the 3'-UTR stem loop structure present in metazoan mRNAs. (b) Sequence logo representation of the histone 3'-UTR stem loop. The sequences (accession number: RF00032) were obtained (more ...)
There are additional regulatory elements that may help to achieve coordinated regulation of core histone genes in metazoans. For instance, a sequence found in core histone gene encoding regions is important for their expression and may serve as an internal promoter element common to the mammalian lineage [41
]. In addition, the transcription factor NPAT has been implicated as a global regulator of core histone gene expression among metazoans even though it does not seem to bind any DNA sequence directly [44
]. This may provide yet another global lineage specific regulatory mechanism that distinguishes the metazoan mode of core histone gene regulation from that of yeast.
Even though the four core histone gene families (H2A, H2B, H3, and H4) diverged before the species studied here, the regulatory mechanisms are more similar for different family members within species than for the same family members between species (Figures and ). Thus, there is a kind of concerted regulatory evolution operating between members of different core histone gene families. This pattern stands in stark contrast to the pattern of core histone sequence evolution, whereby members of the same family are more similar to one another across species reflecting their more recent common ancestry (Figure ). This suggests that very different modes of evolution exist for histone gene regulation versus protein sequence and structure. The solution space for promoter sequence evolution (the space of functionally viable cis-regulatory binding site sequences) may be far more vast than that of core histone protein sequences. This results in a much more dynamic evolutionary paradigm for promoter sequences and the transcription factors proteins that bind them. Purifying selection may be less efficacious at eliminating variants of cis-regulatory sites because a number of sequence variants may bind transcription factors with similar affinities. In addition, new cis-regulatory sites, which are short and degenerate by nature, may arise relatively quickly through mutation along the promoter. It is possible that these new variants can lead to an exploration of expression space and rapid fixation of adaptive variants by positive selection. Adaptive expression changes of this type may be facilitated by the emergence of intermediate redundant regulatory programs that maintain the ancestral expression pattern and function while simultaneously allowing for selective testing of novel expression patterns [47
]. Such an evolutionary mode, with less pronounced purifying and more prominent adaptive selection, could explain the observation that novel cis-trans combinations are subject to substantial turnover and may be regularly reinvented among evolutionary lineages. In addition, the inherent evolutionary flexibility of regulatory systems may allow for coordinated within-species changes that respond to epistatic pressure from other regulatory pathways in the same lineage that share transcription factors.
It is currently unclear whether the turnover of regulatory mechanisms, in the face of conserved expression patterns, is unique to core histones or also occurs for other gene families. Some studies on the evolution of gene regulation do report evidence of conserved regulatory sequences and expression patterns [47
], whereas others indicate that gene regulatory networks do in fact diverge rapidly [49
]. However, regulatory divergence usually leads to distinct expression patterns [51
]. Interestingly, although yeast core histone transcripts include polyA tails, core histone transcripts are unique among metazoan transcripts in that they lack polyA tails. The absence of polyA tails, which are often bound by poly(A)-binding proteins to promote translation initiation, may necessitate, to some extent, species-specific solutions to core histone gene regulation.
The comparative genomics of core histone gene regulation reveal a novel evolutionary mode, which we dub 'circuitous evolution'. Circuitous evolution of core histone gene regulation is distinct from convergent evolution, because the conservation of the core histone gene regulatory patterns suggests that the same pattern existed in the last common ancestor of all species analyzed here. After divergence from the last common ancestor, the core histone expression patterns remained unchanged but the regulatory mechanisms that give rise to the conserved phenotype diverged dramatically. Thus, with respect to core histone gene regulation, where you are from and where you are are far more important than how you get there.
As an addendum, during revision of the manuscript we became aware of a recently published paper [54
], which confirms that the specific periodic pattern of core histone gene expression is uniquely evolutionarily conserved. The report by Jensen and coworkers also demonstrates how many different regulatory solutions have evolved to control the periodic expression of integrated biological systems that function in the cell cycle.