Enhancer modularity and the evolution of new traits

ABSTRACT

Animals have modular cis-regulatory regions in their genomes, and expression of a single gene is often regulated by multiple enhancers residing in such a region. In the laboratory, and also in natural populations, loss of an enhancer can result in a loss of gene expression. Although only a few examples have been well characterized to date, some studies have suggested that an evolutionary gain of a new enhancer function can establish a new gene expression domain. Our recent study showed that Drosophila guttifera has more enhancers and additional expression domains of the wingless gene during the pupal stage, compared to D. melanogaster, and that these new features appear to have evolved in the ancestral lineage leading to D. guttifera.¹ Gain of a new expression domain of a developmental regulatory gene (toolkit gene), such as wingless, can cause co-option of the expression of its downstream genes to the new domain, resulting in duplication of a preexisting structure at this new body position. Recently, with the advancement of evo-devo studies, we have learned that the developmental regulatory systems are strikingly similar across various animal taxa, in spite of the great diversity of the animals' morphology. Even behind “new” traits, co-options of essential developmental genes from known systems are very common. We previously provided concrete evidence of gains of enhancer activities of a developmental regulatory gene underlying gains of new traits.¹ Broad occurrence of this scenario is testable and should be validated in the future.

Cis-regulatory regions have modular structures

A cis-regulatory region consists of non-coding DNA that resides in the vicinity of a gene and regulates the gene's expression level. Within a cis-regulatory region, an enhancer, which typically ranges from a hundred base pairs to a few kilobases of DNA, positively regulates the gene's expression level.^2,3 An enhancer usually contains many binding sites for transcription factors, enabling integration of regulatory information and controlling effective output as the gene's expression level. A single gene can have multiple enhancers, and they often work independently.^4-9 For example, the wingless (wg) gene of Drosophila melanogaster (Wnt1 homolog) is a pleiotropic gene required for many aspects of development, including segment polarity function during embryogenesis, and hence a null mutation causes a lethal phenotype. The wg gene was originally found as a mutant allele, wg¹.¹⁰ The genetic entity of wg¹ is a small deletion of a cis-regulatory region of the wingless gene, and the wg¹ fly lacks only wing-related developmental function, resulting in viable adults that lack wings.^11,12 This example illustrates that the deletion of a single enhancer does not necessarily affect the functions of other enhancers; in other words, cis-regulatory regions have modular structure and function.

Enhancer evolution and loss of traits

If one thinks about organismal development and evolution, the discrete activities of enhancers as independent modules fit well with our current understanding of evolutionary mechanisms. Increasing evidence supports the idea that cis-regulatory changes, especially gain or loss of enhancers, contribute to origins of novel traits during evolution.^8,13-16 In the above example of wg, the enhancer function was lost in the laboratory. Some important studies have shown that similar phenomena have occurred in nature over a long time scale. For example, freshwater stickleback fish populations lack a girdle and associated spines, which was explained as an adaptation to an environment with no gaping predatory fish, with low calcium availability, and/or with the presence of grasping predators such as dragonfly naiads.¹⁷ The genetic basis of this loss of girdle and spines was linked to the loss of an enhancer regulating expression of the gene encoding the transcription factor Pitx1.^18,19 In a fruit fly species found in the Seychelles, Drosophila sechellia, denticle belts on the surface of larvae were lost in comparison to those on closely related species. The genetic basis of this was associated with the loss of enhancer functions for a transcription factor gene, shaven-baby, due to a combination of nucleotide substitutions.^20-22 Thus, in both laboratory and natural populations, complete or partial loss of an enhancer(s) or accumulation of nucleotide substitutions in an enhancer(s) can result in the loss of certain body structures, and this can provide material for adaptive evolution.

Enhancer evolution and gains of traits

Then the question arises: are there cases in which enhancer evolution causes gain of a trait? In some Drosophila species, yellow gene expression leading to pigmentation in wings is caused by gains of new enhancer functions.^23,24 Comparison of chicken and mouse development revealed that a cis-regulatory change in Hoxc8 during evolution altered expression domains in their paraxial mesoderm and ultimately affected regional identity of vertebrae.²⁵ A gene-expression difference of bric-à-brac, which is a negative regulator of Drosophila abdominal pigmentation, was also explained by evolutionary changes in cis-regulatory elements of the gene.^26,27

In the evolutionary processes leading to more complex and diversified animal body plans, not only loss of structures, but also gain of structures, must have played a fundamental role. Specifically, it is likely that gain of new expression domains of developmental regulatory genes (toolkit genes) were key events during long-term macroevolution. However, specific examples whereby a new enhancer caused gain of a new expression domain of a developmental regulatory gene, and ultimately a new trait, have not been described until recently.¹

Discovery of the evolution of wg expression domain by addition of enhancer activity

In a fruit fly species with polka-dotted wings, Drosophila guttifera, the wg expression pattern was found to have uniquely evolved, and wg was found to have an activity that induces pigmentation in the wings (Fig. 1A, B and C).²⁸ Wg is expressed on the margin and crossveins of pupal wings in Drosophila melanogaster. However, Drosophila guttifera expresses wg also on the tips of the longitudinal veins of the wings and in the developing campaniform sensillum precursors. This expression pattern was found only in D. guttifera, but not in closely related species, such as D. deflecta, D. nigromaculata, D. palustris or D. quinaria.¹ Therefore, this trait must have been acquired at some point in the lineage leading to D. guttifera, after bifurcation of its common ancestor from these other, closely related species.^1,29

Figure 1.

(A) An adult wing of Drosophila melanogaster. (B) An adult wing of Drosophila guttifera. (C) Modular additions of enhancer functions underlie the evolution of wingless expression domains.¹ Top (left to right): Schematics of D. melanogaster pupal thorax, pupal wing, and wingless locus. Bottom (left to right): Schematics of D. guttifera pupal thorax, pupal wing and wingless locus. The color code indicates the correspondence between expression domains in the pupal tissue and the enhancers driving gene expressions in these domains (green: wing margin, blue: crossveins, magenta: longitudinal vein tips, orange: campaniform sensilla, red: thoracic stripes). (D) Most gecko species have pilose pads on the ventral side of their digits (black arrows). Among geckos, species of the genus Lygodactylus have similar structures on the ventral tip of their tails (red arrow). This is thought to be an example of heterotopic evolution.^40,41 (E) Most Old World leaf warbler species of the genus Phylloscopus have a pale eyestripe (black arrow), and some species also have a crown-stripe, wing-bars, and a rump spot (red arrows). All pale parts share the same mode of loss of pigmentation, that is, a lack of melanin in the feather tip, and this is thought to be another example of heterotopic evolution.⁴²

My collaborators and I¹ investigated the genomic regions around the wg gene in D. guttifera and D. melanogaster using a transgenic reporter assay and found guttifera-specific enhancer activities (Fig. 1C). One of these enhancers drives expression at longitudinal vein tips (gutCV-T [guttifera crossvein-tip] enhancer), resides just downstream (3′) of the wg gene and overlaps with a crossvein enhancer. The evolutionary origin of the longitudinal vein tip enhancer activity is more recent than that of the crossvein enhancer activity, and thus it should have evolved on top of the preexisting crossvein enhancer activity. The wg regulatory region is large, and it seems counterintuitive that the new enhancer did not evolve in a “naive” region, but in partially overlapping fashion with the preexisting enhancer. One possibility why co-option of preexisting transcription factor binding sites was realized is efficiency: using pre-existing regulatory DNA elements, instead of generating a completely novel enhancer from a “naive” DNA segment, requires a relative small number of mutations.

I note that 2 additional enhancers were found far downstream (approx. 70 kb) of wg, beyond Wnt6, within an intron of Wnt10 (Fig. 1C). One of them, the gutCS (guttifera campaniform sensillum) enhancer, drives expression in developing campaniform sensilla on pupal wings, while the gutTS (guttifera thoracic stripe) enhancer drives expression in stripes on the dorsal side of the pupal thorax. I propose that an ancestor of D. guttifera had an enhancer in this region driving wg expression somewhere during development, and that the new enhancers might have emerged by co-option of this preexisting enhancer, albeit direct evidence for this idea is currently lacking. Of course, there is the possibility that they are wholly new enhancers. In any case, the modular addition of these 3 enhancer activities enabled D. guttifera to acquire new gene expression domains without disturbing the preexisting gene functions, and this contributed to the gain of novel traits in D. guttifera.

Are there truly “new” enhancers?

How could a new enhancer activity emerge? There are several possible scenarios for the origin of enhancer activities.^30,31 The most likely scenario is co-option of a preexisting enhancer. Recruitment or expansion of an enhancer function by addition of a new transcription factor binding site can result in the generation of a new enhancer activity. Especially if the developmental timing and/or the domain of a preexisting enhancer function is close to that of a new enhancer, trans-regulatory factors are already present in the cis-regulatory region of the gene, and therefore, a relatively small number of mutations is sufficient to generate the new enhancer function. In addition to the example of the gutCV-T enhancer discussed above, a spot enhancer, left^{bia large} (spot), of the yellow gene of Drosophila biarmipes also emerged through modification of a preexisting wing blade enhancer.²³ Likewise, an optic lobe enhancer in the Nep-1 gene in Drosophila yakuba co-opted an ancestral enhancer.²⁹ Questions related to the nature of enhancer co-option can be tested by a comparison of enhancers over multiple lineages of animals and by functional experimental assays, whereby enhancers are modified and tested in vivo using transgene-reporter constructs.

In cases of neofunctionalization, through which one copy of a duplicated gene gains a new function, functional diversification of an enhancer can facilitate the evolution of novel expression domains.^32,33 In such a case, the new enhancer might co-opt an old enhancer function. In other cases, transposon insertions have been found to generate a new enhancer.^34,35 Alteration of enhancer-promoter specificity is another way to obtain a new enhancer function for a particular gene. For example, a local inversion of a genomic sequence can switch the activation target from one gene to another.³⁶ And finally, de novo generation is the theoretically simplest, but a rarely observed, path to acquire a new enhancer. One such example has been reported in the zebrafish (Danio rerio). Eichenlaub and Ettwiller³⁷ focused on whole genome duplication and subsequent generation of pseudogenes in the zebrafish lineage. They reported multiple examples of emergence of new enhancers within pseudogenes, and concluded that these enhancers were generated de novo. By investigating the putative activity of the specific DNA region in the pseudogenes and the paralog gene, and the corresponding region in the single ortholog gene of the related species, they could show that the original coding sequences did not have an enhancer activity.

Heterotopy caused by enhancer addition

Biological traits are thought to be produced modularly.³⁸ Not only genomic structures such as cis-regulatory regions, but also body structures, are thought to undergo modular evolution. A phenomenon in which a similar organ appears ectopically is termed “heterotopy.”^39,40 One typical case of heterotopy is the evolution of the epidermal pad of geckos.⁴¹ Many species of geckos have epidermal pads on the ventral side of their digits that enable them to stick to vertical surfaces such as a tree trunk. In some species, epidermal pads are formed on the ventral tip of the tail, too (Fig. 1 D). This tail pad might have evolved de novo, without any relationship to finger pads. But seeing the anatomical similarity, it seems plausible that the tail pad evolved by co-option of the developmental system of the finger pads. Another example of heterotopy is the feather pattern of certain Old World warblers.⁴² In the genus Phylloscopus, many species have pale eyestripes. Among them, some species also have pale wing-bars (on their median and greater coverts), a pale spot on the rump, and a pale crown-stripe (Fig. 1E). The source of similarity between the old structure and the new structure in heterotopy is thought to be the similarity of the expression pattern of genes. In other words, the same gene sets expressed in different anatomical positions act to form similar body structures. Modular addition of new enhancers to a developmental regulatory gene might contribute to heterotopy, and this idea should be testable for the above-noted and other organisms in the future. The author and colleagues demonstrated that at least in one case, a gain of new expression domains of wg is a likely the cause of heterotopic gain of pigmentation in D. guttifera (Fig. 1A-C).^1,28

Modular addition of enhancers enables evolution of new traits

As we have obtained the genome sequences of many different animals, we have come to realize that developmental regulatory genes are well conserved among animals. On the other hand, animal morphologies are extremely diversified. One prominent theory to explain this seeming paradox is that the source of variation is not a consequence of the presence or absence of genes, but rather a consequence of different uses of shared genes, an idea that is gaining increasing acceptance.^8,16 Known cis-regulatory regions of developmental regulatory genes contain many modular enhancers.^5-7 Having started from the primitive ancestral condition with a single enhancer per gene, most genes in higher organisms must have evolved to possess many enhancers during the process of morphological elaboration. Modular addition of an enhancer is one way to add a new expression domain with minimal deleterious effects on preexisting functions, and such gains of enhancers should play a crucial role in the evolution of new traits.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

I thank Sean B. Carroll, Matt W. Giorgianni and Naoyuki Fuse for their comments on the manuscript, Kiyokazu Agata, Tsutomu Hikida, Akira Mori, Teppei Jono and Yuichi Fukutomi for their scientific advice, and Elizabeth Nakajima for constructive comments and English editing. I also thank 2 anonymous reviewers for helpful suggestions on the manuscript.

Funding

This work was supported by JSPS KAKENHI Grant Number 15K18586 and the Uehara Memorial Foundation.