Enhancers integrate spatiotemporal information to generate precise patterns of gene expression. How complex is the regulatory logic of a typical developmental enhancer, and how important is its internal organization? This study examined in detail the structure and function of sparkling, a Notch- and EGFR/MAPK-regulated, cone cell-specific enhancer of the Drosophila Pax2 gene, in vivo. In addition to its 12 previously identified protein-binding sites, sparkling is densely populated with previously unmapped regulatory sequences, which interact in complex ways to control gene expression. One segment is essential for activation at a distance, yet dispensable for other activation functions and for cell type patterning. Unexpectedly, rearranging sparkling's regulatory sites converts it into a robust photoreceptor-specific enhancer. These results show that a single combination of regulatory inputs can encode multiple outputs, and suggest that the enhancer's organization determines the correct expression pattern by facilitating certain short-range regulatory interactions at the expense of others (Swanson, 2010).

The goal of this study was to use a well-characterized, signal-regulated developmental enhancer to examine, in fine detail, the regulatory interactions and structural rules governing transcriptional activation in vivo. This study used functional in vivo assays to test the power of the proposed combinatorial code of 'Notch/Su(H) + Lz + MAPK/Ets' to explain the activity and cell type specificity of the spa cone cell enhancer of dPax2. In the course of this work, several surprising properties of spa were discovered that are not accounted for in current models of enhancer function (Swanson, 2010).

The spa enhancer for fine-scale analysis because (1) the known direct regulators and their binding sites are well defined, (2) they could, in theory, constitute the sum total of the patterning information received by the enhancer, and (3) the enhancer, at 362 bp, is relatively small, simplifying mutational analyses. Surprisingly, a large proportion of the previously uncharacterized sequence within spa is vital for normal enhancer activity in vivo, and of that subset, a large proportion directly influences cell type specificity (Swanson, 2010).

In addition to necessary inputs from Lz, Pnt, and Su(H), three segments of spa were identified, regions 4, 5, and 6, that make essential contributions to gene expression in cone cells. In addition, region 2 makes a relatively minor contribution. (Region 1, another essential domain, will be discussed separately.) Fine-scale mutagenesis reveals that within regions 4, 5, and 6, very little DNA is dispensable for cone cell activation. The previously uncharacterized regulatory sites in spa are very likely bound by factors other than Lz/Pnt/Su(H), for the following reasons: no sequences resembling Lz/Pnt/Su(H)-binding sites reside in these regions; mutations in the newly mapped sites have different effects than removing the defined TFBSs or the proteins that bind them; doubling the known TFBSs fails to compensate for the loss of the newly mapped sequences; and, most importantly, mutating the newly mapped regulatory regions does not significantly affect binding of the known activators to nearby binding sites in vitro. It is not known whether the proposed novel regulators are cone cell-specific, eye-specific, or ubiquitous in their expression. It is known that the newly mapped sites are necessary both for normal cone cell expression and ectopic PR expression. Cut, Prospero, and Tramtrack are expressed in cone cells, but are thought to act as transcriptional repressors. The transcription factor Hindsight is required for dPax2 expression and cone cell induction, but acts indirectly, activating Delta in R1/R6 to induce Notch signaling in cone cells (Swanson, 2010).

Unsurprisingly, placing the enhancer closer to the promoter boosts expression of spa(wt), as well as some of the impaired mutants. The spa enhancer is located at +7 kb in its native locus, and nearly all mutational studies place the enhancer immediately upstream of the promoter. If the entire analysis had been performed at −121 bp, the functional significance of several critical regulatory sequences would have been underrated, and region 1 would have been dismissed as nonregulatory DNA. Other well-characterized enhancers, which have been analyzed in a promoter-proximal position only, may therefore contain more critical regulatory sites than is currently realized (Swanson, 2010).

Like many transcriptional activators, all three known direct activators of spa (or their orthologs) recruit p300/CBP histone acetyltransferase coactivator complexes. Doubling the number of binding sites for these transcription factors (to 6 Lz, 8 Ets, and 10 Su(H) sites) does not suffice to drive cone cell expression in the absence of the newly mapped regulatory regions. It may be, then, that factors recruited to the newly mapped regulatory sites within spa employ mechanisms that are distinct from those of the known activators. The remote activity of spa, mediated by region 1, appears to be an example of such a mechanism (Swanson, 2010).

It was possible to convert spa into a R1/R6-specific enhancer in three ways: (1) by moving the defined TFBSs to one side of the enhancer in a tight cluster; (2) by placing Lz and Ets sites next to regions 1, 4, and 6a; and (3) by mutating regions 2, 3, 5, and 6b within spa while maintaining the native spacing of all other sites. From these experiments, it is concluded that spa contains short-range repressor sites that prevent ectopic activation in PRs by Lz + Pnt + regions 4 + 6a. spa contains at least two redundant repressor sites, because both region 5 and regions 2, 3, and 6b must be mutated to attain ectopic R1/R6 expression (Swanson, 2010).

klumpfuss, which encodes a putative transcriptional repressor, is directly activated by Lz in R1/R6/R7, but is also present in cone cells, making it an unlikely repressor of spa. seven-up, another known transcriptional repressor, is expressed in R3/R4/R1/R6 and could therefore act to repress spa in PRs. However, no putative Seven-up-binding sites were identified within spa. Phyllopod, an E3 ubiquitin ligase component, represses dPax2 and the cone cell fate in R1/R6/R7, but the transcription factor mediating this effect is not yet known (Shi, 2009). Perhaps the best candidate for a PR-specific direct repressor of spa is Bar, which encodes the closely related and redundant homeodomain transcription factors BarH1 and BarH2. Bar expression is activated by Lz in R1/R6 and is required for R1/R6 cell fates. Furthermore, misexpression of BarH1 in presumptive cone cells can transform them into PRs. It is unclear whether Bar-family proteins act as repressors, activators, or both. BarH1/2 can bind sequences containing the homeodomain-binding core consensus TAAT, and region 5 of spa contains two TAAT motifs. Future studies will explore the possibility that Bar directly represses spa in PRs (Swanson, 2010).

The combinatorial code of spa, then, requires multiple inputs in addition to Lz, MAPK/Ets, and Notch/Su(H). Indeed, the data suggest that the known regulators can contribute to expression in multiple cell types, depending on context. The newly mapped control elements identified within spa are necessary not only to facilitate transcriptional activation, but also to steer the Lz + Ets + Su(H) code toward cone cell-specific gene expression (Swanson, 2010).

Enhancers are often located many kilobases from the promoters they regulate. Enhancer-promoter interactions over such distances are very likely to require active facilitation. Even so, few studies have focused specifically on transcriptional activation at a distance, and the majority of this work involves locus control regions (LCRs) and/or complex multigenic loci, which are not part of the regulatory environment of most genes and enhancers. Like spa, many developmental enhancers act at a distance in their normal genomic context, yet can autonomously drive a heterologous promoter in the proper expression pattern, without requiring an LCR or other large-scale genomic regulatory apparatus. However, in nearly all assays of enhancer function, the element to be studied is placed immediately upstream of the promoter. In such cases, regulatory sites specifically mediating remote interactions cannot be identified. Because the initial mutational analysis of spa was performed on enhancers placed at a moderate distance from the promoter (−846 bp), it was possible to screen for sequences required only at a distance, by moving crippled enhancers to a promoter-proximal position. Only one segment of spa, region 1, was absolutely essential at a distance but completely dispensable near the promoter. This region, which contains the only block of extended sequence conservation within spa, plays no apparent role in patterning, or in basic activation at close range. Therefore this segment of spa is termed a 'remote control' element (RCE) (Swanson, 2010).

The remote enhancer regulatory activity described in this study differs from previously reported long-range regulatory mechanisms in two important ways. First, the remote function of spa does not require any sequences in or near the dPax2 promoter. This functionally distinguishes spa from enhancers in the Drosophila Hox complexes that require promoter-proximal 'tethering elements' and/or function by overcoming insulators. This distal activation mechanism also likely differs from enhancer-promoter interactions mediated by proteins that bind at both the enhancer and the promoter, as occurs in looping mediated by ER, AR, and Sp1. Second, studies of distant enhancers of the cut and Ultrabithorax genes have revealed a role for the cohesin-associated factor Nipped-B, especially with respect to bypassing insulators, but it has not been demonstrated that Nipped-B, or any other enhancer-binding regulator, is required only when the enhancer is remote (Swanson, 2010).

The spa RCE is the first enhancer subelement demonstrated to be essential for enhancer-promoter interactions at a distance, but unnecessary for proximal enhancer function and cell type specificity. However, the present work contains only a limited examination of this activity, as part of a broader study of enhancer function. These functional studies, testing for potential promoter preferences and distance limitations, and the identities of factors binding to the RCE are being persued(Swanson, 2010).

As discussed above, it is fairly easy to switch spa from cone cell expression to R1/R6 expression (though, curiously, a construct that is active in both cell types has yet to be constructed). The results show that multiple regions of spa mediate a repression activity in R1/R6, but not in cone cells. It is further concluded that these spa-binding repressors act in a short-range manner; that is, they must be located very near to relevant activator-binding sites, because moving Lz and Pnt sites to one side of spa, without removing the repressor sites (KO+synth^CS), abolishes repression. Despite this failure of repression, synergistic interactions among Lz and Ets sites and the newly mapped sites still occur in this reorganized enhancer -- at least in R1/R6 cells. Cone cell-specific expression is lost, however, revealing (along with other experiments) that transcriptional activation in cone cells is highly sensitive to the organization of regulatory sites within spa. Slightly wider spacing of regulatory sites (KO+synth^NS) kills the enhancer altogether, suggesting that synergistic positive interactions within spa, though apparently longer in range than repressive interactions, are severely limited in their range. The structural organization of spa, then, appears to be constrained by a complex network of short-range positive and negative interactions. Activator sites must be spaced closely enough to trigger synergistic activation in cone cells; at the same time, repressor sites must be positioned to disrupt this synergy in noncone cells, preventing ectopic activation (Swanson, 2010).

Recent work has shown that changes to enhancer organization can 'fine-tune' the output of a combinatorial code, subtly changing the sensitivity of the enhancer to a morphogen. Given the importance of the structure of the spa enhancer for its proper function, it is proposed that any combinatorial code model, no matter how complex, is insufficient to describe the regulation of spa, because the same components can be rearranged to produce drastically different patterns (Swanson, 2010).

One might expect that the regulatory and organizational complexity of the spa enhancer, and its extreme sensitivity to mutation, would be reflected in strict evolutionary constraints upon enhancer sequence and structure. Yet, very poor conservation of spa sequence was observed, both in the known TFBSs and in most of the newly mapped essential regulatory elements. The reduced presence of Lz/Ets/Su(H) sites in D. pseudoobscura could potentially be attributed to redundancy of those sites in D. melanogaster, or to compensatory gain of binding sites for alternate factors in the D. pse enhancer. Perhaps more difficult to understand is the apparent loss of critical regulatory sequences in regions 4, 5, and 6a in D. pse; the experiments in D. mel suggest that the absence of those inputs would result in loss of cone cell expression and/or ectopic activation. It remains possible that many of these inputs are in fact conserved, but that conservation is not obvious due to binding site degeneracy and/or rearrangement of elements within the enhancer. Fine-scale comparative studies are ongoing (Swanson, 2010).

spa is by no means the first example of an enhancer that is functionally maintained despite a lack of sequence conservation. The most thoroughly characterized example of this phenomenon is the eve stripe 2 enhancer; its function is conserved despite changes in binding site composition and organization. Note, however, that spa has undergone much more rapid sequence divergence than eve stripe 2, with no apparent change in function. In general, the ability of an enhancer to maintain its function in the face of rapid sequence evolution suggests that enhancer structure must be quite flexible. These observations support the 'billboard' model of enhancer structure, which proposes that as long as individual regulatory units within an enhancer remain intact, the organization of those units within the enhancer is flexible. Yet, the findings concerning the importance of local interactions among densely clustered, precisely positioned transcription factors are more consistent with the tightly structured 'enhanceosome' model. Further structure-function analysis will be necessary to fully understand the players and rules governing this regulatory element (Swanson, 2010).

The transcription factor D-Pax2 is required for the correct differentiation of several cell types in Drosophila sensory systems. While the regulation of its expression in the developing eye has been well studied, little is known about the mechanisms by which the dynamic pattern of D-Pax2 expression in the external sensory organs is achieved. This study demonstrates that early activation of D-Pax2 in the sensory organ lineage and its maintenance in the trichogen and thecogen cells are governed by separate enhancers. Furthermore, the initial activation is controlled in part by proneural proteins whereas the later maintenance expression is regulated by a positive feedback loop (Johnson, 2010).

The development of adult es organs in Drosophila relies on the correct specification of cell types among the SOP progeny cells followed by their differentiation. Specification is controlled by Notch signaling through the divisions, which ultimately results in the expression of unique combinations of differentiation factors in each cell type. The mechanisms by which these differentiation factors are regulated are not well understood. The D-Pax2 gene encodes a critical differentiation factor for proper es development. Its expression pattern is dynamic and complex and can be divided into an early stage, during which it is expressed in the SOP and all its progeny cells, and a late stage, during which it is restricted to the trichogen and thecogen cells. The function of the D-Pax2 transcription factor during early es organ development has not been established, but it is most likely involved as an antagonist of Notch signaling. Loss of function D-Pax2 mutants show occasional cell fate transformations leading to double socket phenotypes and loss of one functional copy of D-Pax2 greatly enhances the dominant double socket phenotype of Hairless mutations. D-Pax2 functions during late es organ development to promote the differentiation of the trichogen and thecogen cells. This has identified two key enhancer regions that control early and late D-Pax2 expression and demonstrates that proneural proteins are in part responsible for driving early D-Pax2 expression, whereas late D-Pax2 expression relies upon a positive feedback loop (Johnson, 2010).

Rescue experiments have demonstrated that the es organ enhancer for D-Pax2 expression os located upstream from the D-Pax2 transcription start site. A region encompassing a small amount of leader sequence and approximately 6.7 kb of upstream sequence driving the D-Pax2 gene was capable of providing a complete rescue of two D-Pax2 bristle mutants, svⁿ and sv^de. These experiments demonstrated a functional requirement for a large regulatory region but did not give a precise boundary for regulatory elements and did not provide information on the spatial and temporal control of D-Pax2 expression. In order to examine the relationship between this regulatory region and the expression of D-Pax2 during es organ development, the upstream region was dissected using GFP reporter constructs. The initial reporter experiments showed that a slightly smaller 5.8-kb region of DNA (Pax2^A-GFP) was able to drive GFP expression in a complete D-Pax2 pattern, including all of the cells of the SOP lineage early and the trichogen and thecogen cells late. This large segment of DNA spans not only the region upstream of D-Pax2 but encroaches upon a neighboring gene, activin-β, which is oriented in the opposite direction. When the initial D-Pax2 es organ enhancer was shortened down to 3.1 kb (Pax2^B-GFP), similar results were obtained, suggesting that all the information required to drive all aspects of D-Pax2 expression in the es organ is located within approximately 3 kb upstream of the transcription start site (Johnson, 2010).

When the 3.1-kb enhancer was further shortened down to a 2.2-kb region (Pax2^C-GFP), the intense late expression of GFP in the differentiating trichogen and thecogen cells was lost. For this reporter construct, early expression of GFP in the SOP and its progeny during the cell divisions remained and weak expression was visible in all four cells at least as late as 36 hr APF. A similar 2.1-kb enhancer region driving D-Pax2 expression in sv mutants has been shown to provide only partial rescue, compared with the greater efficacy of the 6.7-kb region. The failure of this enhancer to effect a complete rescue may therefore result from its inability to maintain a high level of D-Pax2 expression in the trichogen and thecogen cells during their differentiation. In contrast, a 1-kb fragment (Pax2^D-GFP) representing the remaining part of the 3.1-kb enhancer was unable to drive GFP expression in the SOP and its progeny during the divisions. However, by 32 hr APF, GFP was evident in two cells, one large and one small, that showed coincident expression of D-Pax2 protein and so can be identified as the trichogen and thecogen cells. The activities of the 2.2-kb early enhancer and the 1-kb late enhancer are to some extent complementary and these results suggest that the initial expression of D-Pax2 in the lineage is controlled in a separate manner from its later expression during the differentiation of the cells of the es organ. The 2.2-kb early enhancer does generate a clearly visible GFP signal in the four es cells as late as 36 hr APF and the 3.1-kb complete enhancer also shows comparable expression in the tormogen cell and neuron at this time point. Conceivably, the early enhancer region retains some ability to activate D-Pax2 expression at later time points and there may be an unidentified repressor element required to completely extinguish D-Pax2 expression in the tormogen and neuron. Alternatively, the perdurance of GFP protein masks a sharper delineation of the transcriptional regulation (Johnson, 2010).

There are several transcription factors expressed in the SOP that might potentially regulate the early expression of D-Pax2. Of special note are the proneural genes of the ac-sc complex. Of the four members of the complex, three are involved adult es organ development: ac, sc, and ase. Both ac and sc are expressed in PNC and in the SOP and define the SOP fate. Loss of both ac and sc leads to virtually complete loss of SOPs and, therefore, es organs from the surface of the adult fly. The third member, ase, is not expressed in the PNC and whereas ac and sc are found exclusively in the SOP, ase is expressed in other members of the lineage and is regulated directly by ac and sc. However, the functions of all three genes show some redundancy and ectopic expression of each leads to the appearance of supernumerary es organs. Mutants in ase, however, show no obvious defects in notum microchaete development, although es organs in other regions do exhibit phenotypes suggestive of lineage defects. The function of the D-Pax2 early enhancer requires ac and sc, as pupal nota from sc^10-1 flies bearing the early enhancer reporter showed no GFP expression. Furthermore, ectopic expression of sc also leads to ectopic activation of the early enhancer reporter. Given the requirement for the proneural proteins for the establishment and maintenance of the SOP fate, these results are not surprising. The proneural proteins are therefore required for D-Pax2 expression but the question of whether proneural proteins directly regulate D-Pax2 is not addressed by this experiment (Johnson, 2010).

If one or more proneural proteins regulates D-Pax2 directly, one would expect to find proneural binding sites located in the D-Pax2 enhancer. All of the proneural proteins are basic helix-loop-helix transcription factors and recognize a core E box sequence of CAG^G/_CTG. Upon examination of the 2.2-kb early enhancer, four CAGGTG E box sequences were identified. Two appear approximately 1.6 kb upstream of the transcription start site and the other two are located just 3′ of the transcription start site in the 5′ UTR. The presence of these sites is conserved in all the Drosophila strains for which the D-Pax2 ortholog could be identified, although the number and position of the sites varies. To address the function of the Drosophila melanogaster sites, all four E boxes were mutated in both the 3.1-kb full and 2.2-kb early enhancers. Neither mutated construct elicited GFP expression during the early stages of D-Pax2 expression. Therefore, the four proneural E boxes are necessary for the function of the early enhancer. Because the early enhancer does not drive GFP expression in the absence of the proneural proteins Ac and Sc and because it does not function when the proneural binding sites are mutated, it is concluded that one or more of the proneural proteins are involved in the direct regulation of D-Pax2 in the bristle lineage. Mutation of the proneural binding sites in the 3.1-kb full enhancer did not disrupt its ability to drive late expression of GFP. Faint expression of GFP was seen at 20 hr APF and strong expression in the trichogen and tormogen cells was observable by 36 hr APF. This result indicates the involvement of the late enhancer in the maintenance of D-Pax2 expression in these two cells is independent of earlier proneural protein function (Johnson, 2010).

It has not been determined which proneural proteins are directly responsible for D-Pax2 expression. Ac and Sc are expressed only in the SOP and so D-Pax2 expression in the cells of the lineage after SOP division is unlikely to be driven by them. Ase can be found in the lineage but loss of Ase function does not lead to any notum microchaete defects and even mild sv mutants show misshapen shafts. Furthermore, the early enhancer provides weak expression of GFP at late time points, well after all the known proneurals are expressed. Conceivably, the proneural proteins are required to initiate D-Pax2 expression in the SOP and this initial event is required for continued expression which may be controlled by other unidentified factors (Johnson, 2010).

The late enhancer is sufficient to drive gene expression in the differentiating trichogen and thecogen cells well after the fates of the progeny of the SOP have been specified. Interestingly, the late enhancer is dependent upon D-Pax2 protein itself. In a strong sv mutant background, GFP expression driven by the late enhancer alone disappears, indicating that the maintenance of D-Pax2 expression in the trichogen and thecogen cells is governed by a positive feedback loop. The usage of a positive feedback loop to stabilize gene expression in particular cell types is common. Indeed, the activation and maintenance of Pax2 expression along the midbrain–hindbrain boundary in mice has been shown to be controlled by separate enhancers and the maintenance enhancer is regulated directly by Pax2 itself. In this case, there is no evidence that the regulation is direct; there is no full canonical D-Pax2 binding site in the 1-kb enhancer region. Possibly, sites that do not exactly match the full binding site sequence can function in this enhancer. Alternatively, D-Pax2 protein may control other transcription factors that feed back to keep D-Pax2 up-regulated. A positive feedback loop appears to play little if any role in the early expression in the lineage. The 3.1-kb reporter was unaffected by loss of D-Pax2 function at 20 hr APF. Surprisingly, the 3.1-kb reporter also exhibits expression at 32 hr APF in four cells. It is noted that this reporter does show weak expression in the tormogen and neuron normally at this time point. The loss of D-Pax2 prevents the differentiation of the trichogen and thecogen cells and those cells could conceivably be arrested in an “early” state and more responsive to the early enhancer elements. Alternatively, the complete 3.1-kb enhancer may operate in a qualitatively different manner than the separated early and late enhancers do (Johnson, 2010).

The work presented in this study demonstrates separable regulatory regions responsible for the initiation of D-Pax2 expression and its maintenance during the differentiation of the trichogen and thecogen cells. A partial complement of the factors was also identified that control this early and late expression. Factors aside from the proneural proteins are almost certainly involved in the early expression and still need to be uncovered. The maintenance of late expression by means of a positive feedback loop implicates D-Pax2 in its own regulation but the mechanism by which it does so remains unknown (Johnson, 2010).

Enhancers are genomic cis-regulatory sequences that integrate spatiotemporal signals to control gene expression. Enhancer activity depends on the combination of bound transcription factors as well as - in some cases - the arrangement and spacing of binding sites for these factors. This study examined evolutionary changes to the sequence and structure of sparkling, a Notch/EGFR/Runx-regulated enhancer that activates the dPax2 gene in cone cells of the developing Drosophila eye. Despite functional and structural constraints on its sequence, sparkling has undergone major reorganization in its recent evolutionary history. The data suggest that the relative strengths of the various regulatory inputs into sparkling change rapidly over evolutionary time, such that reduced input from some factors is compensated by increased input from different regulators. These gains and losses are at least partly responsible for the changes in enhancer structure that were observe. Furthermore, stereotypical spatial relationships between certain binding sites ('grammar elements') can be identified in all sparkling orthologs - although the sites themselves are often recently derived. It was also found that low binding affinity for the Notch-regulated transcription factor Su(H), a conserved property of sparkling, is required to prevent ectopic responses to Notch in non-cone cells. It is concluded that rapid DNA sequence turnover does not imply either the absence of critical cis-regulatory information or the absence of structural rules. These findings demonstrate that even a severely constrained cis-regulatory sequence can be significantly rewired over a short evolutionary timescale (Swanson, 2011).

Because of spa's rapid structural evolution and binding-site turnover, multispecies sequence alignments do not reveal many conserved features. Only the extreme 5' end of spa is unequivocally alignable across 12 Drosophila genomes. Given spa's complex regulatory circuitry and structure, its unusually rapid sequence divergence between D. mel and D. pse was surprising, especially because both orthologs of spa have identical cell-type specificities (Swanson, 2011).

This study demonstrated that even an enhancer that is subject to structural constraints can be evolutionarily flexible; therefore, an apparent lack of conserved cis-regulatory structure does not imply an absence of organizational rules within an enhancer (Swanson, 2011).

A model for the structural divergence of spa between the melanogaster and obscura groups is proposed, based on sequence analyses and experimental data. Although the remote control element (RCE) and its flanking Lz1-Ets1 pair are relatively stable, many other essential regulatory sites have been relocated. Within regions 4, 5, and 6a, putative novel regulatory motifs, essential for full-strength activation of both spa orthologs, have been identified whose movements are consistent with experimental data on spa's evolutionary restructuring (Swanson, 2011).

Important changes to the Lz/Ets/Su(H) inputs have also occurred: D. pse has fewer Su(H) and Lz sites, relative to the melanogaster group -- which can be compensated by newly acquired, functionally significant 5' Ets and epsilon (AGCCAG) sites. Meanwhile, the melanogaster group has gained a new Lz site and also has a relative abundance of Su(H) sites, which may compensate for relatively few epsilon and Ets sites (Swanson, 2011).

By tracking the reorganization of Su(H), Lz, Ets, and epsilon motifs across multiple species, a speculative phylogeny of the spa enhancer within the genus Drosophila is proposed and the cis-regulatory content of the last common ancestors (LCAs) of several species groups is predicted by reconstructing the gain and loss of sites, and the changing strengths of transregulatory inputs, in specific lineages. The main conclusions to be drawn from this evolutionary view of spa, informed by functional experiments, are: (1) significant enhancer rewiring has occurred since the divergence of the mel and pse lineages; (2) this rewiring involves the loss and gain of individual regulatory motifs, as well as compensatory changes in the overall strength of several trans-regulatory inputs through changes in binding-site number, position, and possibly affinity; (3) despite very rapid site turnover, characteristic configurations of sites ('grammar elements') can be identified; (4) these grammar elements can be relocated within the enhancer, suggesting that a specific arrangement of sites can be more ancient than the individual sites that compose it. These last two points, taken together, may explain how spa can continue to obey structural rules while being significantly reconfigured (Swanson, 2011).

A large proportion of the grammar elements that have been identified involve Lz/Runx and Ets motifs. Unlike the case of linked sites for Dorsal, Twist, and other factors in insect neurogenic enhancers, there is no single, clearly preferred arrangement of Lz and Ets sites within spa: seven distinct types of Lz/Ets grammar element were identified that are at least as ancient as the LCA of the melanogaster group (Swanson, 2011).

Perhaps Runx and Ets factors, which are known to directly interact and to cooperatively activate transcription in flies and vertebrates, can synergize productively in several different spatial configurations. This is consistent with mapped Runx and Ets sites in vertebrate genomes, which are frequently associated with one another in target enhancers, but not with a single rigid arrangement or spacing (Swanson, 2011).

A nonstructural constraint on the sequence of spa was discovered: a requirement for nonconsensus, low-affinity Su(H) sites for proper cone-specific patterning. Because ectopic dPax2 expression in photoreceptor precursors causes faulty cell fate specification and differentiation, resulting in defective eye morphology, it is reasonable to suppose that the expression pattern of spa[Su(H)-HiAff] would have negative fitness consequences for the fly. Taken together with previous work, the data presented in this study suggest that spa requires input from Notch/Su(H) but also requires that input to be attenuated at the cis-regulatory level, in order to generate the proper levels and cell-type specificity of dPax2 expression in a tissue with widespread Notch signaling. Like Notch/Su(H), EGFR/Ets signaling and Lz are also used to specify multiple cell types in the retina, which presents a challenge for combinatorial gene regulation: enhancers must be able to make fine qualitative distinctions in regulatory inputs and often must translate this information into relatively sharp on/off decisions. These pressures could result in a cis-regulatory logic for genes like dPax2 in which many weak inputs are independently tuned (and spatially arranged) to maximize activation in the proper cell type, while minimizing ectopic activation. Previous studies of spa present a picture of an enhancer operating just above a functional threshold, such that the loss of a single regulatory site, or a loss of proper grammar, can result in transcriptional failure in cone cells. One of the main conclusions from this study is that, over a relatively short evolutionary timescale, a cis-regulatory module can find multiple solutions to this complex computational problem (Swanson, 2011).

The presence of weak, nonconsensus binding sites for signal-regulated TFs is a common, but little remarked upon, feature of developmental enhancers. Low-affinity TF binding sites have well-documented functions in shaping a stripe of gene expression across a morphogen gradient and in determining temporal responses to developmental regulators. This study provides direct evidence supporting a role for weak signal response elements in preventing ectopic transcriptional responses to highly pleiotropic signaling pathways such as Notch (Swanson, 2011).

There is one striking question not addressed by this study: why is this enhancer evolving at an unusually high rate, given that its expression pattern is stable? Two plausible explanations are given for which supporting data exist. First, dPax2 is on chromosome 4, the 'dot' chromosome of Drosophila, which has a severely reduced recombination rate, resulting in inefficient selection and relaxed sequence constraint. No other cis-regulatory module on the fourth chromosome has been subjected to an extensive evolutionary analysis, nor are any as well-mapped as sparkling, but enhancers of the fourth-chromosome genes eyeless and toy contain fairly large blocks of sequence conservation, compared to spa. An alternative explanation for the rapid turnover observed within spa involves the presence of nonconsensus, predicted low-affinity sites for Su(H) and, in some cases, Lz and PntP2. For a typical TF, there are many more possible low-affinity binding sites than high-affinity sites: for example, the highest-affinity Su(H) consensus YGTGDGAAM encompasses only 12 variants (TGTGGGAAA, etc.), whereas the lower-affinity consensus of the same length nRTGDGWDn, which accommodates all of the known Su(H) sites within spa, contains 576 possible sequences. Accordingly, it is much more likely that an enhancer will acquire a low-affinity binding site via a single mutational event than a high-affinity site. Thus, an enhancer that does not require high-affinity binding sites for given trans-regulators may rapidly sample a variety of configurations of weak sites and may thereby undergo considerable sequence turnover without losing the input from that regulator. In other words, an enhancer such as spa, which must maintain a weak regulatory linkage with Notch/Su(H), may be less constrained than a high-affinity target with respect to the sequence, number, and position of its Su(H) binding sites. Whatever the reason for the rapid sequence divergence of spa, it provides an opportunity to examine in detail the evolutionary mechanisms by which a complex cis-regulatory module can be significantly reorganized, while still conforming to specific constraints of combinatorial logic and grammar (Swanson, 2011).