Over the last 2 decades, the concept of alteration in cis
-regulatory elements, rather than coding sequences, has gained increasing support as a key driver for the evolution of morphological differences (reviewed in references 10
). The majority of studies so far have focused on the analysis of cis
-regulatory elements of invertebrate genes, alterations of which cause striking morphological changes in closely related species (3
), with only a few examples in vertebrates (3
). Compared to the evolution of altered gene expression patterns, much less is known about the mechanisms responsible for highly conserved patterns of expression. The latter might reflect highly conserved cis
-regulatory mechanisms; some invertebrate data support this idea (3
), but little is known about the mechanisms operating in vertebrates. In this study, we have used comparative genomic sequence analysis to identify the chicken SCL cis
-regulatory elements. Where long-rage DNA alignments failed to identify conserved regions, we used our previous characterization of transcription factor binding sites essential for the activity of murine SCL
regulatory elements to design targeted searches. To analyze the function of the chicken cis
-regulatory elements and avoid possible functional redundancy, each element was analyzed independently. The function of each element was analyzed in vitro
, using luciferase assays, in different hematopoietic cell lines and in vivo
using transgenic mice and frogs. Both approaches showed that despite the conservation of gene expression patterns between mice and chickens, each independent element shows considerable functional differences compared with its mouse orthologue.
Several aspects of our data are of general significance. First, they demonstrate that a highly conserved expression pattern may hide unexpected cis
-regulatory plasticity (Fig. ). The functions of individual SCL
elements are different in mammals and lower vertebrates. In some cases, the altered function reflects binding site turnover between distinct elements—the mammalian promoter has lost ETS sites, but these have been acquired by the mammalian +19 enhancer. In acquiring its ETS sites, the mammalian +19 enhancer has lost an E box present in the equivalent ancestral enhancer and has switched from an erythroid element to a hematopoietic stem cell/endothelial enhancer. The ancestral SCL
locus contained two enhancers with erythroid GATA/E-box motifs (orthologous to the murine +19 and +40 enhancers), and this redundancy may have permitted the +19 equivalent to lose its E box without deleterious consequences. In the mouse, the +40 is the major element driving activity to the circulating erythroid cells, whereas in the chicken, the element corresponding to the +19 fulfils this role, raising the possibility that the chicken +40 orthologue functions primarily in the midbrain. Despite the multiple differences in the cis
regulation of the mouse and chicken, the net effect is a constancy of regulatory inputs and spatiotemporal transcriptional output. However, this constancy conceals significant cis
-regulatory alterations throughout the locus, with orthologous elements from different species exhibiting structural and functional differences. These observations also demonstrate that fixation of mammal-specific regulatory elements (2
) is not necessarily accompanied by the acquisition of new transcriptional inputs or changes in the overall expression pattern of a gene but may reflect “rewiring” within a locus. It is unclear how new transcription factor binding sites “appear” within an element, although it has recently been hypothesized that in addition to conventional evolutionary processes, specific binding sites may also be transferred between elements through recombination-based mechanisms (9
Secondly, the data emphasize that when using comparative genomic sequence analysis to identify regulatory elements, an absence of evidence is not evidence of absence. Sequence alignments were unable to detect enhancers that could be identified given detailed knowledge of functionally important motifs. The +19 enhancer provides a particularly dramatic example, where a few conserved nucleotides flanking a consensus motif provided the clue to locating the chicken equivalent. Conversely, sequence conservation between orthologous regulatory elements may not equate to conserved function. The SCL promoter region exhibits significant sequence conservation throughout vertebrate evolution, and yet the mammalian promoter lacks hematopoietic/endothelial activity displayed by the SCL promoter from lower vertebrates.
Thirdly, our results have revealed an unexpected pattern of sequence conservation: the chicken sequence for each SCL
regulatory element is more closely related to its frog orthologue than to its mammalian equivalent. This unexpected pattern of sequence conservation is not a general feature of the SCL
locus, since alignments of the SCL
coding region and the conserved SCL
poly(A) region all show the expected pattern of sequence conservation (see Table S1 and Fig. S7 at http://hscl.cimr.cam.ac.uk/Gottgens_et_al_MCB_Supplemental_Information.pdf
). These data indicate that after the divergence of birds and mammals, multiple SCL cis
-regulatory elements accrued mutations at an accelerated rate in mammalian ancestors. The same “reversed” pattern of frog/chicken/mammal sequence conservation was identified for a GATA-2 hematopoietic/endothelial enhancer but not for Runx-1
(see Table S1 and Fig. S8 at the above URL), suggesting that cis
-regulatory remodelling coupled with accelerated evolution of regulatory elements may be a more widespread phenomenon. We postulate that this accelerated acquisition of mutations in multiple cis
-regulatory elements may have been triggered by the creation of a redundant element. The platypus and opossum sequences demonstrate that the switch of the +19 equivalent from a GATA/E-box to an ETS/ETS/GATA motif was an early event. This switch would have resulted in redundancy with the ancestral orthologues of both the promoter and the −4 enhancer, thus freeing other elements to accumulate alterations more rapidly without deleterious consequences. The ancestral SCL
promoter is active in brain as well as in blood/endothelium, and so there may also be adaptive value in disentangling the individual components of such a hybrid bifunctional element, thus allowing unconstrained evolution of the individual components. We have shown that analogous departures from the expected pattern of regulatory element sequence conservation occur at other loci (e.g., GATA-2) but, depending on the timing of the initial trigger, will not always be revealed by frog/chicken/mammal comparisons.
In summary, our article contains a detailed characterization of the evolution of multiple elements across a key vertebrate regulatory locus, using stringent transgenic functional assays for individual cis elements. This analysis revealed remarkable remodeling of cis-regulatory elements across the locus despite constancy of transcriptional inputs and output (i.e., expression pattern). Moreover, an unexpected pattern of sequence conservation in regulatory elements (but not in coding regions) suggests an accelerated accumulation of mutations between 310 million and 200 million years ago, which our results suggest is probably the consequence of the formation of a new but redundant element. Given that redundancy of regulatory elements has been observed for many other vertebrate gene loci, cis-regulatory remodeling is likely to be a general phenomenon, suggesting that the comprehensive functional data presented here are of broad relevance for our understanding of vertebrate gene regulation and evolution.