A. aegypti midline gene expression resembles that of other arthropods

We previously reported that A. aegypti sim, unlike its orthologs in other studied arthropods, shifted expression from the ventral midline to lateral regions of the embryonic CNS sometime during the midline primordium stage of midline development; this pattern was similarly adopted by other genes of the midline GRN functioning downstream of sim [18]. We sought to establish a more thorough timeline for this shift in expression, as well as to determine more definitively whether sim and other midline genes were fully absent or merely severely reduced in expression in the late midline, using the sensitive hybridization chain reaction (HCR) method for in situ hybridization [42]. Surprisingly, HCR revealed that sim expression remains strong in the A. aegypti ventral midline, with no shift to lateral CNS regions, through to late embryonic stages (Fig 1A-1F). Somatic muscle expression, similar to what has been observed in both Drosophila and in Apis mellifera (honeybee), was also observed (S1 Fig; [18,24]). HCR probes for short gastrulation (sog; Vectorbase AAEL024210) and shotgun (shg; Vectorbase AAEL012421), two other genes previously reported to shift from midline to lateral in A. aegypti [18], similarly reveal a retained midline expression pattern (Fig 1K-1N).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Expression of sim, sog, and shg in A. aegypti embryos.

(A-F) Timecourse of sim expression in A. aegypti embryos from 14 hr. through 52 hr. using HCR. Note that expression is present in the midline at all stages. Lateral-appearing expression seen in panel E (arrows) is non-CNS expression in the somatic mesoderm (confirmed via mef2 expression; see S1 Fig). (G-H) GFP expression from a sim::GFP fusion gene inserted in the native sim locus, visualized using anti-GFP antibody staining. Timepoints correspond to those in panels C-F. Arrows in G, H and I mark expression in the somatic musculature, not in the plane of focus in C and D (but see S1C and S1D Fig). (K, L) sog expression in 35 and 45 hr. embryos as visualized by HCR. Limited expression is visible at the periphery of the CNS at 35 hr. (arrows); however, this is distinct from the more significant expression reported by Suryamohan et al. [18]. Expression at 45 hr. is in cells whose morphology and position resemble the Drosophila dorsal median (DM) cells, which are not known to express sog and thus may represent a late embryonic expression difference between these two species. (M, N) Expression of shg as revealed by HCR in 35 and 45 hr. embryos. Note the prominent expression in the midline. In all panels, embryos are shown in their ventral aspect with anterior to the left.

https://doi.org/10.1371/journal.pgen.1010891.g001

To better understand the discrepancy between the current and previous results, we repeated the in situ hybridizations for sim and sog using the standard digoxygenin-labeled riboprobe/alkaline-phosphate detection protocol [43] on 45-hour A. aegypti embryos, with an extensive set of control experiments. We determined that the A. aegypti nervous system is largely refractory to this protocol during this mid-embryonic stage, yielding little true signal but a significant number of positive-appearing cells scattered throughout the tissue (S2 Fig). We detected little difference between sense and anti-sense probes (cf. panels E&H and F&I in S2 Fig), and obtained similar results with probes for the Saccharomyces cerevisiae Gal4 gene, which has no mosquito homolog (S2D and S2G Fig). The sog probe did have some limited activity in what appears to be a legitimate site of A. aegypti sog expression (analogous to the Drosophila dorsal median cells, S1J and S1K Fig, arrows), but also ectopic-seeming expression elsewhere (S1K Fig, arrowhead). In gut tissues, by contrast, we observed good overlap between the HCR and riboprobe methods (S2L and S2M Fig arrows), suggesting that the artifacts may be present primarily in nervous tissue (note the ectopic expression in the brain; S2L and S2M Fig, arrowheads). Taken in sum, these results argue for greater reliability of the HCR data, and suggest that our previous reported results are incorrect.

To establish the sim expression pattern definitively, we used CRISPR/Cas9 genome editing to create a sim::GFP fusion gene in the sim locus (see Methods). As this fusion gene is under the control of the native sim genomic regulatory sequences, it should produce a faithful readout of sim expression. Sim::GFP was expressed in the midline and ventral somatic musculature identical to what we observed for sim HCR (Fig 1G-1J, cf. Figs 1C-1F, S1C and S1D). Based on the HCR analysis and sim::GFP fusion data, therefore, we conclude that A. aegypti midline gene expression resembles that of D. melanogaster and other studied arthropods throughout embryogenesis.

Identification of A. aegypti sim enhancers

Although the pattern of sim gene expression is similar between D. melanogaster and A. aegypti, we were curious to see what regulatory changes have occurred between the two species. The Drosophila sim locus is an order of magnitude smaller than its A. aegypti counterpart (S3 Fig) and, as is typical with insect species as many years divergent as D. melanogaster and A. aegypti, there is no readily detectable conservation of the non-coding sequences [5,44]. The Drosophila locus has been extensively assayed for regulatory sequences. At least four distinct enhancers are known to drive sim expression during the midline primordium stage, and one at the mesectoderm stage (S3B Fig) [31,35,38]. CRMs responsible for late midline expression remain unidentified. No A. aegypti sim enhancers are known.

We chose eight initial sequences from the A. aegypti sim locus to assay for regulatory activity (S3A Fig). Four were chosen based on sequence conservation with the related mosquito Aedes albopictus (estimated divergence 48 MYA; [41]), two were chosen as lying just upstream of the two annotated sim promoters, respectively, and two were selected essentially randomly from within the large first intron of the “B” transcript of the gene. As an initial screening assay, each selected sequence was tested for its ability to drive reporter gene expression in transgenic Drosophila embryos. Of the eight sequences, five had no observable expression (S3A Fig), one drove expression in a pattern largely distinct from that of sim (S3A and S4 Figs), and two drove expression in the ventral midline (see below).

A mid-stage autoregulatory sim enhancer

The intP2 sequence is located in the A. aegypti sim first intron between the two annotated promoters and is roughly 40% conserved with A. albopictus (Fig 2A). In Drosophila embryos, it drove reporter gene expression in a pattern identical to that of sim beginning at early stage 9 and persisting throughout embryogenesis (Fig 2B and 2D). Closer examination of the sequence revealed that there are two primary blocks of sequence conservation (each about 75% identity) (Fig 2A). We therefore tested each of these smaller conserved fragments individually. Whereas sequence intP2A failed to drive embryonic reporter gene expression (Fig 2E), the 826 bp intP2B fragment has activity identical to that of the larger intP2 sequence (Fig 2F) and recapitulates endogenous sim expression (Fig 2G). Consistent with a role as a sim enhancer, intP2B fully overlaps a region of open chromatin as assayed by FAIRE-seq (Fig 2A; [45,46]).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Characterization of the intP2 enhancer.

(A) Top: Schematic of the A. aegypti sim locus with the intP2 sequence highlighted in cyan. Bottom: Close-up of the intP2 region showing tested subfragments, conservation with A. albopictus, and FAIRE data. (B-O) Experiments in Drosophila melanogaster embryos are indicated by a fly icon, while those in A. aegypti embryos have a mosquito icon. Fly embryos are all stage 11, except for panels C (st. 14) and D (st. 15). All embryos are oriented ventrally with anterior to the left. (B) Reporter gene expression driven by intP2 in transgenic Drosophila at stage 11, (C) at stage 14, and (D) at stage 15. (E) The intP2A fragment fails to drive reporter gene expression. (F) Reporter gene expression driven by intP2B in transgenic Drosophila. (G) Double-labeling confirms that intP2B activity (magenta) is in all midline cells, as marked by sim-Gal4::UAS-tauGFP (green). (H, I) Reporter gene activity driven by intP2B in 24 hr. (germ-band extended) and 36 hr. (germ-band retracted) A. aegypti embryos, respectively. (J) Mutation of CME sites in intP2B abrogates enhancer activity in transgenic Drosophila. (K) No intP2B activity is observed in a sim null background. (L, M) As in Drosophila, mutation of CME sites in intP2B abrogates enhancer activity in transgenic A. aegypti. (N) The Drosophila sim_st10 enhancer drives reporter gene expression in the midline and in the ventral musculature (arrows) in A. aegypti transgenic embryos (cf. Fig 1G). A 24 hr embryo is shown. (O) Similarly, the sim_st10 enhancer drives expression in the midline and ventral musculature (arrows) in Drosophila embryos (cf. S1A Fig).

https://doi.org/10.1371/journal.pgen.1010891.g002

To ensure that the activity we observed was not due to the sequence being assayed in a Drosophila background, we generated transgenic A. aegypti with intP2 and intP2B driving a GFP reporter gene. Reporter gene expression in the transgenic mosquitoes was fully consistent both with the Drosophila transgenic results and with endogenous A. aegypti sim expression as revealed by HCR and sim::GFP (Figs 2H, 2I and S5; cf. Figs 2B and 1A-1J). Reporter expression was fully present by 16 hours of development (early germ-band extended) and persisted past germ band retraction to at least 45 hours.

In Drosophila, sim expression at the midline primordium stages is autoregulated via binding of Sim:Tango (Tgo) heterodimers to RCGTG “CNS midline enhancer” (CME) sites [26,31,47]. Inspection of the intP2B sequence revealed four of these potential autoregulatory sites (S6 Fig), mutagenesis of which abrogated enhancer activity in both transgenic D. melanogaster and A. aegypti (Figs 2J, 2L 2M and S7). Furthermore, intP2B was unable to drive reporter gene expression in D. melanogaster when recombined into a sim null background (Fig 2K). Collectively, these results demonstrate that A. aegypti midline sim expression at the midline primordium stage is regulated through an autoregulatory enhancer, similar to what has been characterized in D. melanogaster.

The intP2B enhancer is remarkably similar to the Drosophila sim_st10 enhancer [38], one of at least four Drosophila CRMs with reported midline primordium regulatory activity. Despite sharing no appreciable sequence conservation, the two enhancers are alike both in overall position within the sim first intron and in their complement of 3–4 Sim:Tgo “CME” autoregulatory sites (S6 Fig). Other putative binding sites are also in common, including those for Ventral veins lacking (Vvl) and Twist (Twi), although roles for these have not been tested empirically. We tested the Drosophila sim_st10 enhancer in transgenic A. aegypti to determine if the fly enhancer was functionally equivalent to the mosquito enhancer. Surprisingly, we observed not only strong midline activity, but also activity in the Sim-positive muscle precursors (Fig 2N). No such muscle activity was observed in either species with the intP2 or intP2B enhancers. As this activity was not previously reported in Drosophila for sim_st10 [38], we generated transgenic flies using the sim_st10 sequence and assayed reporter gene expression at stages 11 and later. Consistent with what we observed in transgenic A. aegypti, clear muscle expression driven by sim_st10 was observed in the Drosophila embryos (Fig 2O). Therefore, despite substantial similarity in both position and binding site composition, intP2B and sim_st10 do not have identical activity.

A novel second midline enhancer

The 5P3 sequence is located in a conserved region in the sim 5’ intergenic region roughly 19.8 kb upstream of the sim-RB promoter (Fig 3A). When assayed in Drosophila embryos, it drove expression in a subset of midline cells as well as in a variety of non-midline interneurons in more lateral regions of the CNS (Fig 3B). By testing a series of smaller overlapping sequence fragments (Fig 3A), we were able to separate the midline activity from the lateral activity (Fig 3C-3I). The lateral activity, which is ectopic with respect to native sim expression in both fly and mosquito, does not become apparent until germ band retraction at stage 13 and maps to at least two independent sequences, 5P3G and 5P3B (Fig 2C and 2G). 5P3G lies within the region of conservation with A. albopictus. The midline activity is contained within the 173 bp 5P3F sequence, begins at late stage 9/stage 10, and persists through late embryogenesis (Fig 3H and 3I). Testing in transgenic A. aegypti confirmed that 5P3 behaves in its native trans environment identical to how it behaves in Drosophila; expression is consistently seen in the midline by 24 hours of embryogenesis (mid germ-band extended stage) and persists through at least 45 hours (Figs 3J, 3K and S5). Strikingly, we see lateral ectopic expression in the later stages, following germ band retraction (approximately 32 hr.), similar to what was observed in the Drosophila assay (Figs 3K, arrows and S5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Characterization of 5P3 and its subfragments.

(A) Top: Schematic of the A. aegypti sim locus with the 5P3 sequence highlighted in green. Bottom: Close-up of the 5P3 region showing tested subfragments and conservation with A. albopictus. Fragments with midline activity are colored cyan, those with lateral activity yellow, and those with both, green. Fragments with no enhancer activity are shown in gray. (B-K) Experiments in Drosophila melanogaster embryos are indicated by a fly icon, while those in A. aegypti embryos have a mosquito icon. Fly embryos are all stage 14–15 except for H, which is stage 11. All embryos are oriented ventrally with anterior to the left. (B) The 5P3 sequence drives reporter gene expression in midline (arrowheads) and lateral (arrows) CNS cells. (C) Activity of 5P3B is restricted to lateral cells (arrows). (D) 5P3A drives both midline (arrowheads) and lateral (arrows) reporter gene expression, whereas (E) 5P3D is only active in midline cells (arrowheads). Non-CNS expression in the lateral bodywall (thin arrows) is a product of the expression vector, not the enhancer fragment. (F) 5P3E drives both midline (arrowheads) and lateral (arrows) reporter gene expression, but (G) 5P3G is confined to lateral regions only. Somewhat more medial expression (thin arrows) is not in the midline and is present, but out of the plane of focus, in the other fragments having lateral activity as well. (H, I) 5P3F has midline expression only (arrowheads), pictured here at stages 11 and 15 respectively. Thin arrows in (I) indicate vector-induced expression. (J, K) The 5P3 fragment drives midline expression (arrowheads) in midline primordium-stage A. aegypti embryos, as well as ectopic expression following germ-band retraction (arrows). (L) Expression driven by 5P3F is still observed in a sim null background (arrowheads), showing that 5P3F is not subject to autoregulation.

https://doi.org/10.1371/journal.pgen.1010891.g003

Labeling using a variety of midline-expressed markers enabled us to identify the cells where the 5P3F reporter gene is active in Drosophila embryos (Fig 4). The stage 10 midline consists of 16 cells comprising three equivalence groups, the MP1, MP3, and MP4 groups (Fig 4A; [48]). Subsequent signaling via the Notch pathway leads to specification of the anterior and posterior midline glia (AMG, PMG) and six neural precursors, MP1, MP3, MP4, MP5, MP6, and the median neuroblast (MNB). At late stage 11, these cells undergo Notch-mediated asymmetric cell divisions to form the MP1 neurons, the H-cell and H-cell sib neurons, and the Ventral Unpaired Median motorneurons and interneurons (mVUMs and iVUMs, respectively). The latter arise from MP4, MP5, and MP6, each of which divides to yield one mVUM and one iVUM. Specific combinations of gene expression allow each of these cells types to be uniquely identified [49].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Cell-type activity of the 5P3F enhancer in transgenic Drosophila embryos.

(A) Cartoon of the midline cell lineages at the midline primordium stages (adapted from [68]). Cells of the MP4 equivalence group are shown in shades of blue. Cells with heavy outlining express Engrailed. (B) Co-labeling of the 5P3F reporter (magenta) with Nubbin (Nub; green) at stage 13 shows that 5P3F is not active in H-cell or H-cell-sib. (C) Co-labeling with Slit (Sli; green) indicates 5P3F activity in posterior midline glia, shown in dissected embryos at stage 16. (D) 5P3F activity is both in and adjacent to cells expressing Mab 22C10 (Futsch; green), which marks the mVUMs, shown here at stage 16. (E-G) Sagittal sections showing co-labeling of the 5P3F reporter (magenta) with Engrailed (En; green) at stage 13. 5P3F reporter gene expression is observed in the mVUMs (white arrowheads), iVUMs (yellow arrowheads), MNB and progeny (yellow arrow), and posterior midline glia (PMG; thick white arrow). Not all of the PMG are expressing the reporter (thin white arrows). (H) Cartoon of the stage 13 midline (adapted from [69]). (I) Co-labeling of the 5P3F reporter (magenta) with Engrailed (En; green) at stage 11 (ventral view). All reporter-gene expressing cells are positive for En with the exception of a single anterior cell (arrow), likely one of the MNB progeny. Embryos in B, C, D, and I show a ventral section while E-G are sagittal sections; all embryos have anterior to the left.

https://doi.org/10.1371/journal.pgen.1010891.g004

At stages 15–16, coexpression with Slit and Mab 22C10 (Futsch) revealed 5P3F reporter gene activity in the posterior midline glia (PMG) and mVUMs, respectively (Fig 4 C and 4D), while stage 13 expression just posterior to Nub expression rules out activity in the H and H-sib cells (Fig 4B). At stages 13–14 we see coexpression with En, which marks the iVUMs, MNB and its progeny, and PMG, although reporter expression in PMG is inconsistent (Fig 4E-4H). At late stage 11, reporter expression is similarly coincident with En (Fig 4I); the exception is the anteriormost cell (arrow), which we interpret to be one of the MNB progeny which loses En expression at about this stage [49]. Taken together, these data suggest that the 5P3F enhancer is specific to the early “MP4 equivalence group” (Fig 4A, top), the earliest clear differentiation of the midline cells into specific fates [48].

Surprisingly, the 5P3F sequence does not contain any sim-binding CME sites, suggesting that unlike all known Drosophila midline primordium-stage sim enhancers, it is not autoregulated. Consistent with this, reporter gene expression is maintained even in a sim null background (Fig 3L). The A. aegypti 5P3F enhancer is therefore novel in several distinct ways as compared to both the intP2 enhancer and to any of the known Drosophila sim enhancers: its activity begins later, at the stage 9/10 boundary; it is expressed in only a subset of the midline cells; it lacks autoregulatory capability; and it is located substantially upstream of the transcription start site.

Functional assessment of conserved regions from Aedes albopictus

The intP2 and 5P3 sequences were chosen as candidate enhancers in part due to sequence conservation with the related Aedine mosquito A. albopictus. We were therefore curious to determine whether these partly conserved A. albopictus sequences were also able to function as midline enhancers. For intP2, we selected a 648 bp A. albopictus sequence similar to the A. aegypti intP2B fragment (Fig 2A). The two sequences have 68% identity over a central 600 bp of alignment, including conservation of all four Sim:Tgo binding sites. For the 5P3 region, we chose the primary conserved region (~70% identical) that aligns from the middle of the 5P3D fragment through the 5P3H fragment (Fig 3A). Both of these A. albopictus fragments drove reporter gene expression in transgenic D. melanogaster in a pattern similar to their A. aegypti counterparts (S8 Fig).

Conserved and novel enhancers

Although the intP2B enhancer is remarkably similar to the Drosophila sim_st10 enhancer, it lacks the muscle activity regulated by the latter. Several possibilities may account for the differences in activity: the sim_st10 sequence may be a composite of two enhancers, a midline enhancer and a muscle enhancer, and intP2B is the counterpart of just the midline one; the proper counterpart to int2B in Drosophila may be one of the other three or more identified sim midline enhancers (none of which have described muscle activity), despite their being less similar to int2B in terms of position and binding site composition; or the sequence may have evolved different capabilities in the two species. Future finer-scale mapping of both the A. aegypti and Drosophila loci, with identification of additional enhancers and further dissection of known ones, will be necessary to distinguish between these scenarios.

The 5P3F enhancer, by contrast, is completely novel compared to known Drosophila sim enhancers with respect to its spatial activity, temporal activity, and lack of autoregulatory capability. The latter is particularly striking, as all studied Drosophila midline enhancers contain CME sites [25–31]. Full activity of all tested midline enhancers requires functional CMEs, although in a minority of cases reduced enhancer activity can still be observed in their absence [25]. Hong et al. [31] have noted a correlation between number of CME sites and onset of enhancer activity, with a high number of binding sites (four) needed for early midline primordium activity. Although 5P3F has a later onset than intP2B and the known Drosophila sim enhancers, its activity by stage 10 still places it in the class requiring at least two CME sites by their analysis. 5P3F function, therefore, is likely to be governed by novel regulatory mechanisms.

5P3F is also unique in that it appears to initiate activity only in the MP4 equivalence group. All currently known Drosophila sim enhancers that initiate at the midline primordium stage do so in the complete set of midline cells. 5P3F activity is most like that of the mfas_780 enhancer [25], which is active in the MP4-6 cells as well as in anterior midline glia. 5P3F, however, only drives expression in the posterior, not anterior, midline glia. When tested in Drosophila embryos, 5P3F reporter gene expression is similar to that of En at stages 10–11; later deviations may simply result from perdurance of the lacZ reporter in cells where En has already ceased expression. Further studies will be necessary to determine the off times of both the intP2B and 5P3F enhancers. Interestingly, despite the extensive characterization of the Drosophila sim locus, enhancers active in late midline stages have not been isolated. Thus, sequences regulating sim expression following germ band retraction, if distinct from those active at the midline primordium stage, remain unknown.

Importantly, it is also not currently known if the cellular makeup of the A. aegypti midline is one-for-one identical to that of Drosophila or not, and if gene expression patterns are completely conserved. For example, while the overall morphology, number, and arrangement of neuroblasts in the beetle Tribolium castaneum is conserved with Drosophila, there is substantial variation in gene expression patterns in the neuronal lineages [50]. Thus, extensive additional analysis of the A. aegypti CNS remains necessary in order to put our results fully in context. The reporter lines we have generated will be of considerable aid in this regard.

While the regulatory mechanisms governing 5P3F are novel compared to known Drosophila sim enhancers, we note that 5P3 reporter gene activity is identical in all respects when tested in both A. aegypti and D. melanogaster transgenic embryos. This includes its later onset of expression at the mid, not early, midline primordium stage and presence of ectopic activity in lateral CNS cells following germ band retraction. Therefore, the same regulatory mechanisms are available for use in both species. It will be interesting to determine whether a 5P3F-like enhancer exists in the Drosophila sim locus, or if the A. aegypti sequence is a novel regulatory element that is tapping into an available, but not accessed, regulatory capability. Kalay et al. [15] have described “cryptic enhancers,” sequences with regulatory capability that are repressed by surrounding sequences in some species, but which constitute latent enhancers that could potentially be activated in other species to evolve a new expression pattern. Further characterization of the Drosophila sim locus, including finer-scale dissection of known enhancer regions, will be necessary to establish whether this is the case, whether there is an active but as-yet unidentified MP4-equivalence group sim enhancer, or whether the exogenous 5P3F sequence is independently responding to signals in the Drosophila trans environment without having a Drosophila genomic counterpart.

We used here Drosophila transgenic reporter lines as an expedient method to test our putative enhancers for activity, and only then tested the best candidates in their native A. aegypti host. This is a common strategy that has served well in the past. However, it will still be necessary at some point to confirm whether the sequences that failed in the Drosophila assay also fail in their native A. aegypti trans environments. This is especially true for the intP3 sequence, which was able to drive expression in a non-sim like pattern in Drosophila.

Challenges of non-model organisms

Special challenges come into play when studying regulatory evolution in non-traditional laboratory species [51] and at large evolutionary distances. Fortunately, A. aegypti has proven a tractable experimental system, with robust transgenic capability and straightforward husbandry. Nevertheless, our experiences underscore the care that must be taken when working in relatively unexplored territory, as “standard” techniques such as digoxygenin/riboprobe based in situ hybridization may prove unreliable, and orthogonal methods and corroborating data are not always available. In this case, use of the sensitive HCR method plus an extensive set of control experiments, along with generation of an engineered GFP fusion gene, enabled us to determine that previous characterizations of the A. aegypti sim expression pattern were incorrect, and allowed for a convincing updated description. However, other challenges remain. As mentioned above, detailed morphological and molecular-level investigations of the A. aegypti CNS are still required to properly establish homology with Drosophila between the midline and other nervous system cell types. Moreover, the high degree of non-coding sequence divergence between the two species makes identifying regulatory sequences a particular challenge. While we demonstrate some success here by using conservation between the two Aedine species A. aegypti and A. albopictus, sequence alignment in general provides a limited means for enhancer discovery [52]. The recent availability of additional Aedine genomes [53] may be of future help, as will success we have had using computational methods such as SCRMshaw to identify functionally-related regulatory sequences across the holometabolous insects [5,18,54,55].

Regulatory evolution of a locus

Enhancer activity within a locus can be maintained over extraordinary evolutionary distances. For example, sequences from the sponge Amphimedon queenslandica Islet locus are able to drive reporter gene expression in transgenic zebrafish in patterns reminiscent of (although not identical to) the zebrafish Islet2a gene [7], and in similar tissues in transgenic mice, despite over 750 MY of divergence. Similarly, a deeply conserved region of the SoxB2 locus maintains related although not identical function in species from vertebrates to cnidarians (~700 MY). Strikingly, both human [6–8] and sea urchin versions of this enhancer are also able to drive reporter gene expression in a Sox-related pattern in Drosophila, despite no obvious sequence homology [6]. While these studies make clear the remarkable ability of enhancers to retain functional conservation over hundreds of millions of years, the vast distances and limited characterization of the respective loci make it difficult to infer how the complete regulatory locus may have evolved.

The overall regulatory evolution of a single locus has previously been evaluated extensively in the case of the Drosophila yellow (y; FlyBase FBgn0004034) gene, which encodes a protein in the pigmentation pathway. Kalay et al. [12] found that enhancers with similar tissue-specific activity in three Drosophila species were often found in different locations within their respective loci. However, the authors note that y expression is a highly evolving trait, and speculate that enhancers controlling conserved aspects of expression may be more constrained in genomic location. The analysis is complicated by the subsequent identification of numerous redundant and semi-redundant enhancers in the y locus [15]. Determining which enhancers have changed location versus which have maintained position is thus a non-trivial exercise.

Frankel et al. [56] undertook a thorough analysis of the shavenbaby (svb; FlyBase: ovo, FBgn0003028) locus in D. melanogaster and its 30–40 MYA diverged relative D. virilis. In the case of svb, which encodes a transcription factor that serves as a “master regulator” of trichome development, the position and overall function of enhancers appears to be maintained. Although like pigmentation trichome patterning is a rapidly-evolving trait, svb’s more pleiotropic role as a master transcriptional regulator may induce stronger constraints on the overall regulatory architecture of its locus (consistent with the speculations of Kalay et al. [12] discussed above).

Enhancer positioning has also been explored in several cases of much larger evolutionary divergence than the 25–50 MYA range examined for y and svb regulation. Cande et al. [4] looked at the locations of enhancers for several developmental genes in species as diverse as D. melanogaster, T. castaneum, Anopheles gambiae (mosquito), and Apis mellifera (honeybee). Many of these enhancers appear to be maintained in a similar genomic position even over these large distances (up to 345 MYA), although the locations of others seem not to have been conserved. Similar results were observed by Kazemian et al. [5]. These analyses are complicated by the fact the overall regulatory loci are not defined in detail, making it difficult to assess the full regulatory landscape in the way that was done for y and svb, albeit among more closely-related species.

The Drosophila sim locus, by contrast, has been characterized in detail, with almost all non-coding sequences in the ~30 kb locus tested for enhancer activity. sim is a core developmental gene with embryonic midline activity conserved over more than 350 MY of arthropod evolution. Our initial analysis of A. aegypti sim enhancers suggests both conserved position and mechanism of action for some enhancers, but also potentially novel regulatory mechanisms acting through non-homologous enhancers at new locations. Further studies of this locus will be necessary to determine whether or not there are additional novel regulatory activities, and to what degree the various midline as well as non-midline enhancers are retained, lost, or altered. It has been well-established that developmental genes often have highly conserved expression patterns mediated by enhancers with common sequence characteristics, despite these enhancers having frequently diverged past the point of recognition via simple sequence comparisons. Unresolved, however, is the question of whether or to what extent such functionally-equivalent enhancers are conserved by direct descent or result from convergent processes [5]. Understanding how the genomic organization of these enhancers changes over evolutionary time aids in clarifying this question. Although an analysis of only two species, such as that presented here, cannot answer the descent vs. convergence question, the observation that at least some enhancers appear to be maintained in analogous locations suggests one of two possibilities: the enhancer has been conserved through direct descent, or there is a functional constraint on the positioning of a regulatory element with the necessary function, leading to convergence of not just sequence composition but also of positioning for a newly-evolved enhancer. Although the latter seems to us the less probable alternative, it raises intriguing questions about locus-level regulation of gene expression. While the “grammar” of individual regulatory elements has often been discussed [57], potential larger-scale grammars have received much less attention. Whether the arrangement, spacing, or other aspects of the relative configuration of the various enhancers in a locus has regulatory consequences remains an open and largely unexplored frontier. Continued analysis of the well-explored and tractable sim locus should help to shed light on this and other important questions relating to how the regulatory landscape of a major and highly-conserved developmental regulator evolves.

Reporter constructs and transgenic Drosophila

Sequences were amplified by PCR on genomic DNA isolated from A. aegypti Liverpool (LVP) wildtype mosquitoes. Eighteen putative sim enhancer sequences were cloned, using the Gibson Assembly method [58], into plasmid pLacZattB, a φC31-enabled Drosophila transformation vector containing the LacZ gene under the control of a minimal hsp70 promoter [59]. Primer sequences are provided in Table A in S1 Text. Transgenic flies were produced by Rainbow Transgenic Flies (Camarillo, CA) by injection into line attP2 8622.

Reporter constructs and transgenic A. aegypti

Sequences were amplified by PCR on genomic DNA isolated from A. aegypti LVP wildtype mosquitoes, with the exception of sim_st10, which was synthesized as a gBlock (IDT, Coralville IA) based on D. melanogaster dm6 sequence 3R:13060273..13060969. The intP2, 5P3, and sim_st10 sequences were cloned into pENTR-D-Topo (Thermo Fisher) and then by Gateway recombination [60] into pgPhiGUE, a piggyBac transformation vector containing enhanced green fluorescent protein (EGFP) under the control of the Drosophila synthetic core promoter (gift of Y. Tomoyasu, Miami University). Primer sequences are provided in Table A in S1 Text. Transgenic mosquitoes for intP2 and 5P3 were produced by the Insect Transformation Facility at the Institute for Bioscience & Biotechnology Research (Rockville, MD) by injecting preblastodermal A. aegypti LVP embryos; the other transgenic mosquito lines were generated in the Halfon lab. Genomic insertion sites for the transgenic lines were identified using the Splinkerette method [61] and are provided in Table B in S1 Text. Control lines consisting of the pgPhiGUE vector with a non-regulatory dummy sequence inserted in lieu of an enhancer confirmed that there is no vector-dependent embryonic reporter gene activity (S9 Fig).

Construction of the sim::eGFP fusion line of A. aegypti

An active sgRNA within Exon VII of the sim locus was identified by microinjection of A. aegypti LVP embryos co-injected with 80 fmol/μL each of the pHsp70-Cas9 vector (Addgene 45945) and an sgRNA expressing plasmid containing the U6:3 promoter for AAEL017774 (U6 spliceosomal RNA) followed by the sgRNA scaffold from [62] programmed for the protospacer target site. The active sgRNA site was identified 39 bp upstream of the sim stop codon at position Chr1:73549491–73549513. A homology arm donor was then synthesized as a gBlock (IDT, Coralville IA) with the donor cargo flanked by 400 nucleotides up- and downstream of the CRISPR/Cas9 cleavage site (Table C in S1 Text). The donor consisted of a synonymous SNP recoding of the sim 39 C-terminal nucleotides, a 39 nucleotide fusion linker terminating with an EcoRI cut site followed by five additional nucleotides, and an XhoI restriction site. The U6:3 promoter for AAEL017774 followed by the sgRNA scaffold from [62] programmed for the active cut site in sim Exon VII was further included in the gBlock, downstream of the right homology arm (see S10 Fig). The gBlock was cloned into pBluescript using ApaI and NheI/XbaI. The eGFP fusion, terminator, and 3xP3-dsRed marker were then amplified from the pgPhiGUE vector using primers eGFP-fus-EcoRI-F and 3xP3-fus-XhoI-R (Table C in S1 Text), and cloned into the intermediate plasmid using EcoRI and XhoI. Lines were generated in the Halfon lab where preblastoderm embryos of A. aegypti LVP were microinjected with 80 fmol/μL each of the donor and the pHsp70-Cas9 helper. Surviving G0 animals were backcrossed to wildtype LVP and the G1 progeny were screened for the presence of the 3xP3-dsRed eye marker. Correct insertion of the donor construct was confirmed by Nanopore sequencing of a PCR product using primers Aeg_sim_E1_F1 and 3xP3-Forward, which spans the 3xP3 marker and runs into the sim gene beyond the region included for the homologous recombination (S8 Fig).

Mosquito husbandry

This project used the A. aegypti LVP_ib12 strain, obtained from the Malaria Research and Reference Reagent Center (MR4; BEI Resources). A. aegypti lines were reared in an insectary at constant conditions of 27°C, 80% relative humidity, and a 12:12 hr light:dark cycle. Larvae were reared in pans and fed on ground Tetramin fish flakes (Spectrum Brands Pets, Blacksburg, VA). Adult mosquito diet was raisins, with adult females fed three times every five days with defibrinated sheep blood (Hardy Diagnostics).

Immunohistochemistry

For all analyses, a minimum of ten embryos were analyzed in detail and for transgenic mosquitoes, at least four independent insertion lines were examined. Drosophila embryo preparation and immunohistochemistry was performed using standard methods. A. aegypti embryo preparation and staining was performed essentially as described by Clemons et al. [63,64] with modifications as detailed in Reid et al. [65]. Primary antibodies used were mouse anti-β-galactosidase (1:500; Ab-cam), rabbit anti-GFP (1:500; Ab-cam, ab290), rabbit anti-β-galactosidase (1:500; Ab-cam), mouse anti-Slit C555.6D (1:10; Developmental Studies Hybridoma Bank), mouse anti-Nubbin 2D4 (1:10, DSHB), mouse anti-En 4D9 (DSHB), and Mab 22C10 (1:20, DSHB). The ABC kit (Vector Laboratories) was used for immunohistochemical staining. For fluorescent staining, the secondary antibodies used were anti-mouse Alexa Fluor 488 (Molecular Probes, A11029; 1:250) and anti-rabbit Alexa Fluor 647 (Molecular Probes, A21245; 1:500). Fluorescent staining was visualized by confocal microscopy using a Leica SP8 confocal microscope.

In situ hybridization

A. aegypti embryos were prepared as described by Reid et al. [65]. In situ hybridization using digoxygenin-labeled riboprobes was performed as in Haugen et al. [43]. For HCR, reagents were obtained from Molecular Instruments (Los Angeles, CA). Embryos were stained as described by Bruce et al. [66] with the following modifications [67]: Embryos were incubated at 37°C overnight with HCR probes at a concentration of 16 μL/mL, washed 5x in probe wash buffer, then incubated overnight in the dark in 100μL of a 6% solution (each) of the labeled HCR hairpins. Embryos were subsequently washed three times in 5xSCCT, one time with 1xPBS, then resuspended in 80% glycerol containing 4% n-propyl gallate and visualized by confocal microscopy using a Leica SP8 confocal microscope.

Motif analysis and mutations

The intP2B enhancer sequence was scanned for exact matches to the Sim binding site consensus sequences ACGTG [47] and DDRCGTG [31] using MacVector (Apex, NC). Four sites were identified, and each motif was mutated to the BamHI restriction site GGATCC as in Wharton et al. [26]. The mutated enhancer sequence was synthesized as a gBlocks Gene Fragment (IDT, Coralville IA) and inserted into pattBnucGFPh using the NEBuilder HiFi DNA Assembly kit (New England Biolabs, Beverley, MA).

Test for autoregulation

The intP2B and 5P3F reporter constructs were recombined onto a sim null mutant chromosome (sim², Bloomington Stock Center stock #2055) and balanced over a ftz-lacZ containing TM3 balancer.

Image credits

The following graphical elements were obtained from The Noun Project (thenounproject.com) under a Creative Commons CC-BY 3.0 licence: fly, Georgiana Ionescu; mosquito, Cristiano Zoucas.