Abstract
Changes in cis-regulatory modules (CRMs) that control developmental gene expression patterns have been implicated in the evolution of animal morphology1-6. However, the genetic mechanisms underlying complex morphological traits remain largely unknown. Here we investigated the molecular mechanisms that induce the pigmentation gene yellow (y) in a complex spot and shade pattern on the abdomen of the quinaria group species Drosophila guttifera. We show that the y expression pattern is controlled by only one CRM, which contains a stripe-inducing CRM at its core. We identified several developmental genes that may collectively interact with the CRM to orchestrate the patterning in the pupal abdomen of D. guttifera. We further show that the core CRM is conserved among D. guttifera and the closely related quinaria group species Drosophila deflecta, which displays a similarly spotted abdominal pigment pattern. Our data suggest that besides direct activation of patterns in distinct spots, abdominal spot patterns in Drosophila species may have evolved through partial repression of an ancestral stripe pattern, leaving isolated spots behind. Abdominal pigment patterns of extant quinaria group species support the partial repression hypothesis and further emphasize the modularity of the D. guttifera pattern.
How complex morphological features develop and evolve is a question of foremost importance in biology. To address this question, we identified genes underlying abdominal pigmentation pattern development in Drosophila guttifera (D. guttifera). The abdomen is decorated with six rows of black spots that run along the anterior-posterior axis, divided by a dark dorsal midline shade. This color pattern shows four sub-patterns: a dorsal, median, and lateral pair of spot rows, plus the dorsal midline shade (Fig. 1a, b). D. guttifera belongs to the quinaria species group, whose members display highly diverse abdominal pigmentation patterns7,8. While D. guttifera shows the most complex pattern of this group, most other quinaria group species lack at least one of the four sub-patterns, illustrating the pattern modularity among species. Interestingly, the stripe patterns of certain species often separate into spots7,8. In this study, we show that the abdominal pigment patterns of quinaria group members may be formed by a combination of localized spot induction and partial stripe repression.
We focused on the regulation of the yellow (y) gene, which is required for the formation of black melanin in insects8-14. Several y gene CRMs have been identified in various Drosophila species, and changes in these CRMs and/or in the deployment of trans-factors that regulate y gene expression have been implicated in the diversification of wing and body pigment patterns12,15-19. In D. guttifera pupae, y gene expression and the location of the Y protein accurately prefigured the complex adult abdominal pigment pattern (Fig. 1c, d). In order to identify putative upstream activators of y, we performed an in situ hybridization screen for genes expressed in ways prefiguring the y gene expression pattern. We found that wingless (wg) expression precisely foreshadowed the six rows of black spots (Fig. 2b). Additionally, decapentaplegic (dpp) expression foreshadowed the dorsal and median pairs of spot rows (Fig. 2c), while abdominal-A (abd-A) expression correlated with the lateral pair of spot rows and the dorsal midline shade (Fig. 2d, e). hedgehog (hh) and zerknullt (zen) were additionally expressed along the dorsal midline of the abdomen (Fig. 2f, g). Thus, the activation of the D. guttifera color pattern appears to be induced in a modular fashion, which is in agreement with our observation that abdominal pigmentation patterns within the quinaria group are variations of the D. guttifera pattern ground plan (Fig. 3). This situation is reminiscent of the wing pattern ground plan in nymphalid butterflies20,21.
We hypothesized that the developmental candidate genes may activate the y gene through four CRMs, each controlling one sub-pattern to assemble the complete melanin pattern. We searched for these CRMs by transforming D. guttifera with DsRed reporter constructs containing non-coding fragments of the 42 kb D. guttifera y gene locus12 (Extended Data Fig. 1). Surprisingly, only one 953 bp fragment from the y intron, the gut y spot CRM, drove expression closely resembling all six spot rows on the developing abdomen (Fig. 4). To isolate possible sub-pattern-inducing CRMs, we subdivided the gut y spot CRM into 8 partially overlapping sub-fragments. Unexpectedly, the 636 bp left sub-fragment displayed horizontal stripe expression along the posterior edges of each abdominal segment, while the 570 bp right fragment was inactive (#1 & #2, Fig. 4.). Further dissection of this CRM revealed a 259 bp sub-fragment, which contained the minimal gut y core stripe CRM with some additional dorsal midline shade activity (#7, Fig. 4). These results suggest that the D. guttifera spots may have evolved from an ancestral stripe pattern that became partially repressed to isolate the spots. Currently, we cannot offer any direct evidence for specific candidate repressor genes. Neither the in situ hybridization experiments nor the bioinformatics analyses, using Jaspar, resulted in putative pigment stripe repressors.
Although we identified 24 Engrailed (En)-binding sites and 19 Homothorax (Hth)-binding sites in the gut y spot CRM (both are known repressors of pigmentation in Drosophila15,22), these sites were not enriched in the right half of the CRM, as we would have expected. However, our transcription factor binding site analysis of the gut y spot CRM sequence revealed putative transcription factor binding sites for most of the developmental genes that we identified as potential activators in our in situ hybridization screen, except for hh. This suggests that localized spot activation by these developmental factors contributes to the formation of the pattern.
Next, we asked whether the abdominal pigment spot pattern of a species closely related to D. guttifera shares a similar developmental basis. We thus performed in situ hybridization experiments in Drosophila deflecta (D. deflecta). This species displays six longitudinal spot rows on its abdomen, but lacks the dorsal midline shade (Extended Data Fig. 2a, b). As in D. guttifera, y mRNA in D. deflecta pupal abdomens was expressed in six rows of spots, except along the dorsal midline (Extended Data Fig. 2c). Similarly, wg foreshadowed all six rows of spots, while dpp expression matched all but the lateral spot rows (Extended Data Fig. 3b, c). In contrast to the D. guttifera results, abd-A, hh, and zen were absent along the dorsal midline, which is in agreement with the lack of pigment in D. deflecta adults (Extended Data Fig. 3d, e, f, g). However, abd-A expression was not detectable where the lateral spot rows will form (Extended Data Fig. 3d), suggesting that these particular spots are controlled differently in D. deflecta. We next cloned the 938 bp orthologous def y spot CRM and transformed it into D. guttifera, using the DsRed reporter assay. The def y spot CRM drove faint dorsal spot row and stripe expression, especially along the dorsal spots (Extended Data Fig. 4). We further subdivided the def y spot CRM into 8 sub-fragments and identified a minimal def y core stripe CRM (288 bp) (#7, Extended Data Fig. 4). This sub-fragment drove a striped pattern, but without the dorsal midline shade activity seen in the D. guttifera minimal gut y core stripe CRM (#7, Fig. 4). We further transformed the gut y spot and def y spot CRMs including all sub-fragments into D. melanogaster to test if D. melanogaster trans-factors can bind to and activate these two quinaria group species’ spot CRMs. As a result, none of the reporter constructs showed any expression (data not shown). This suggests that the hypothetical ancestral stripe pattern of the quinaria group and the pigment stripes found on the D. melanogaster abdominal tergites23 have evolved independently by changes in trans. As the spot CRMs from D. guttifera and D. deflecta are not orthologous to any sequences within the D. melanogaster y locus, changes in cis have also contributed to the diversification of pigment patterns between D. melanogaster and the quinaria species group.
In contrast to the D. guttifera wing spot pattern12, the abdominal pigment pattern develops in the absence of visible physical landmarks. wg, dpp, and hh are homologous to known proto-oncogenes in humans24, while zen and abdA are Hox genes. The abdominal color pattern of D. guttifera appears to be regulated by multiple developmental pathways consisting of activators and repressors acting in parallel, possibly targeting pigmentation genes other than y as well18,19,25,26. Further evidence for the repression of stripes can be seen in Drosophila falleni’s intraspecific pigment variation, another member of the quinaria species group (Extended Data Fig. 5). Our multi-pathway model fits well with the observation that the abdominal pattern variation presented by quinaria group members is largely due to modular derivations from the D. guttifera ground plan (Fig. 3). This scenario is reminiscent of the modularity found in butterfly wing patterns. Because insects use similar genes for color pattern development21,27-30, the quinaria group may serve as a valuable model to understand insect color pattern evolution. Future work should aim to manipulate the genes involved in pigmentation to test if they interact according to the reaction-diffusion model, as predicted by Alan Turing31.
Materials and Methods
Molecular procedures
In situ hybridizations were carried out with species-specific RNA probes, as described in12, but with abdomens cut into halves and cleaned from the internal organs. At least three positive pupae were observed for each result shown. Additional images for verification purposes are provided in Extended Data Figs. 6-16. Immunohistochemistry for the Y protein in abdomens was performed according to15, with abdomens cut in half and cleaned with 1X PBS. D. guttifera CRMs were identified and tested in D. guttifera according to12 and in D. melanogaster as described in23. Transgenic experiments were performed as outlined in32. Pupal stages were identified according to33.
Drosophila stocks
All fly stocks were a kind gift from the Sean B. Carroll Laboratory (University of Wisconsin - Madison) and were cultured at room temperature. We used the D. melanogaster fly strain VK00006 (cytogenic location 19E7), the D. guttifera stock no.15130–1971.10, and the D. deflecta stock no. 15130-2018.00 for gene expression and transgenic analyses.
PCR primer sequences
We used the following primers to amplify the CRM sequences:
(iii) gut y spot CRM: Fwd: 5’-CAGCTGCGGTTGAGTACGAC-3’and Rvs: 5’-GCCAACTCGACGGGAATTC-3’. Restriction sites: KpnI and SacII.
(iv) def y spot CRM: Fwd: 5’-CAGCTGCTGCGGTTCAGTAG-3’ and Rvs: 5’-GCTAGACACACGTTGGTTTGCT-3’. Restriction sites: KpnI and SacII.
(v) gut y spot CRM sub-fragment #1: Fwd: 5’-CAGCTGCGGTTGAGTACGAC-3’ and Rvs: 5’-ACTGAATCTGATTTCGGCTCG-3’. Restriction sites: KpnI and SacII.
(vi) gut y spot CRM sub-fragment #2: Fwd: 5’-AGTTAATCGCCAGTCAATAATGGC-3’ and Rvs: 5’-GAATTCCCGTCGAGTTGGC-3’. Restriction sites: KpnI and SacII.
(vii) gut y spot CRM sub-fragment #3: Fwd: 5’-CAGCTGCGGTTGAGTACGAC-3’ and Rvs: 5’-GCCATTATTGACTGGCGATTAAC-3’. Restriction sites: KpnI and SacII.
(viii) gut y spot CRM sub-fragment #4: Fwd: 5’-AAATGAAGCTCAGTGAGCCGC-3’ and Rvs: 5’-ACTGAATCTGATTTCGGCTCG-3’. Restriction sites: KpnI and SacII.
(ix) gut y spot CRM sub-fragment #5: Fwd: 5’-AGCATCTGAAACTTAAACGCCG-3’ and Rvs: 5’-GAATTCCCGTCGAGTTGGC-3’. Restriction sites: KpnI and SacII.
(x) gut y spot CRM sub-fragment #6: Fwd: 5’-CAGCTGCGGTTGAGTACGAC-3’ and Rvs: 5’-CAGCGATATTAATTTTTTATTCAATGG-3’. Restriction sites: KpnI and SacII.
(xi) gut y spot CRM sub-fragment #7(gut y core stripe CRM): Fwd: 5’-AAATGAAGCTCAGTGAGCCGC-3’ and Rvs: 5’-GCGATTTGTTTGTCAAGTCAAC-3’. Restriction sites: KpnI and SacII.
(xii) gut y spot CRM sub-fragment #8: Fwd: 5’-AAATGAAGCTCAGTGAGCCGC-3’ and Rvs: 5’-GTTGACTTGACAAACAAATCGC-3’. Restriction sites: KpnI and SacII.
(xiii) def y spot CRM sub-fragment #1: Fwd: 5’-CAGCTGCTGCGGTTCAGTAG-3’ and Rvs: 5’-ATTGTCGCAGCTGCCTAACG-3’. Restriction sites: KpnI and SacII.
(xiv) def y spot CRM sub-fragment #2: Fwd: 5’-AACGAAGCTCACTGAGCTGC-3’ and Rvs: 5’-AGCAAACCAACGTGTGTCTAGC-3’. Restriction sites: KpnI and SacII.
(xv) def y spot CRM sub-fragment #3: Fwd: 5’-CAGCTGCTGCGGTTCAGTAG-3’ and Rvs: 5’-GTTAAAAGCAGCCAGTTGGCC-3’. Restriction sites: KpnI and SacII.
(xvi) def y spot CRM sub-fragment #4: Fwd: 5’-CAAAGAATCGAATTCGGAGACAG-3’ and Rvs: 5’-ATTGTCGCAGCTGCCTAACG-3’. Restriction sites: KpnI and SacII. (Clone name: def y 1.1C2)
(xvii) def y spot CRM sub-fragment #5: Fwd: 5’-GAATGAGATTCGTTAGGCAGC-3’ and Rvs: 5’-AGCAAACCAACGTGTGTCTAGC-3’. Restriction sites: KpnI and SacII.
(xviii) def y spot CRM sub-fragment #6: Fwd: 5’-CAGCTGCTGCGGTTCAGTAG-3’ and Rvs: 5’-TTCAACGGATATTCGTTCAATTTC-3’. Restriction sites: KpnI and SacII.
(xix) def y spot CRM sub-fragment #7 (def y core stripe CRM): Fwd: 5’-CAAAGAATCGAATTCGGAGACAG-3’ and Rvs: 5’-GTCAGGCAATGTAAATGTTGTCG-3’. Restriction sites: KpnI and SacII.
(xx) def y spot CRM sub-fragment #8: Fwd: 5’-AACGAAGCTCACTGAGCTGC-3’ and Rvs: 5’-ATTGTCGCAGCTGCCTAACG-3’. Restriction sites: KpnI and SacII.
These forward and reverse primer sequences do not include restriction sites.
Author Contributions
K.K.B.R., M.S. and T.W. conceived and designed the experiments; K.K.B.R, P.M.E.N., T.E.S, E.B., P.P.K, A.McQ., E.M., A.A., A.N. and T.W. performed the experiments; K.K.B.R., S.M. and T.W. analyzed the data; T.W. obtained funding; K.K.B.R and T.W. wrote the paper; K.K.B.R, S.M., P.M.E.N., T.E.S., E.B., P.P.K., A.McQ., E.M., A.A., A.N. and T.W. edited the paper.
Competing Financial Interests
The authors declare no competing financial interests.
Acknowledgements
We thank Ryan Bensen, Abigail Meisel, Bridgette Rebbeck, and Jason Hu for technical assistance. This research was funded by NIH grant #1R15GM107801-01A1 to T.W.