Pigmentation pattern and developmental constraints: flight muscle attachment sites delimit the thoracic trident of Drosophila melanogaster

Abstract

In their seminal paper published in 1979, Gould and Lewontin argued that some traits arise as by-products of the development of other structures and not for direct utility in themselves. We show here that this applies to the trident, a pigmentation pattern observed on the thorax of Drosophila melanogaster. Using reporter constructs, we show that the expression domain of several genes encoding pigmentation enzymes follows the trident shape. This domain is complementary to the expression pattern of stripe (sr), which encodes an essential transcription factor specifying flight muscle attachment sites. We demonstrate that sr limits the expression of these pigmentation enzyme genes to the trident by repressing them in its own expression domain, i.e. at the flight muscle attachment sites. We give evidence that repression of not only yellow but also other pigmentation genes, notably tan, is involved in the trident shape. The flight muscle attachment sites and sr expression patterns are remarkably conserved in dipterans reflecting the essential role of sr. Our data suggest that the trident is a by-product of flight muscle attachment site patterning that arose when sr was co-opted for the regulation of pigmentation enzyme coding genes.

Introduction

Body pigmentation is at the interface between the organisms and their environment and fulfils many ecologically relevant functions. Indeed, the adaptive role of pigmentation seems often so obvious – i.e. in crypsis, mimicry, aposematism, mate recognition, UV protection or thermoregulation - that it is generally assumed that the observed patterns have been selected. Direct selection on pigmentation has been demonstrated in many cases, such as the famous industrial melanism of the peppered moth, Biston betularia, whose molecular basis has recently been identified^1,2. Pigmentation patterns can also be the object of a trade-off between opposite selection forces. For example, in guppy Poecilia reticulata males, pigmentation patterns result from a balance between selection for crypsis, an anti-predator strategy, and selection for conspicuousness to attract females³. Further evidence for selection is the convergent loss of pigmentation when selection is relaxed in organisms living in the absence of light, such as cave animals^4,5. However, since Gould and Lewontin’s seminal paper in 1979⁶, it is widely acknowledged that adaptationist explanations should sometime be used with caution and developmental constraints also taken into account. Gould and Lewontin used the example of the spandrels decorated with splendid mosaics between the arches supporting the dome of the basilica of Saint-Mark in Venice. Although the mosaics fit remarkably well on the spandrels, the spandrels were not designed for them but result from architectural constraints imposed by the structure supporting the dome. Thus, Gould and Lewontin argued that some biological traits arose as by-products of developmental constraints on a crucial trait and were not selected for their direct utility^6,7. Hence, developmental constraints favour particular patterns or morphologies whereas they forbid others. A few studies on pigmentation have addressed this question by exploring morphospaces with artificial selection experiments^8,9. The size of the eyespots on butterfly Bicyclus anynana wings responds to selection but their colour is more constrained as only coordinated changes of pigments are possible⁸. Consequently, only some area of the morphospace can be occupied. Furthermore, the eyespot pattern through the genus Bicyclus follows a similar path of diversification, suggesting that evolution is partly constrained and that selection (or drift) can operate only in particular directions⁸.

Drosophila pigmentation is an interesting model to study the impact of developmental constraints¹⁰. This highly evolvable trait has been the focus of many studies analysing the genetic bases of morphological variation within and between species^11,12. Body or wing pigmentation relies on the coordinated action of trans-regulatory factors on genes encoding pigmentation enzymes^13–20. Most of these factors, for example Engrailed or Abdominal-B, are spatially restricted components of a deeply conserved regulatory landscape involved in development of many essential traits. Hence, pigmentation patterns can be interpreted as targets of indirect selection due to their association with another trait²¹.

The trident is a melanic pattern observed on the dorsal thorax of Drosophila melanogaster²². In natural populations, the intensity of the trident shows clinal genetic variation, with darker tridents observed at higher latitudes in Europe, India and Australia and higher altitudes in India and Africa^23–26. The variable intensity of the trident is therefore thought to be an adaptation to temperature and/or UV^24,25. Differences in the intensity of the trident were shown to be linked to genetic variation in the pigmentation gene ebony (e), which is less expressed in flies with a darker trident^26,27. Besides, the trident is more clearly visible in an e mutant background²⁸. Furthermore, the intensity of the trident is sensitive to developmental temperature²⁴. This trait is therefore an example of phenotypic plasticity, “the ability of a given genotype to produce different phenotypes in response to distinct environmental conditions”²⁹. Interestingly, in Drosophila busckii, a pattern similar to the trident is very clearly delineated on the thorax, being dark brown on a yellow background^22,30. Thus, the intensity of the trident is variable, plastic and highly evolvable. By contrast, the developmental bases of its shape have attracted little attention. Regulatory genes, whose mutation modifies it, are the most promising candidates to address the developmental constraints exerted. Mutation in the stripe gene (sr¹ allele) affects the shape of the trident³¹. This gene encodes an Egr-like zinc-finger transcription factor specifying epidermal tendon cells, to which muscles attach^32–35. Therefore, sr expression labels the flight muscle attachment sites on the pupal thorax^32,35. The expression pattern of sr on the pupal thorax³⁶ seems complementary to the trident, which suggests that sr might delimit the trident by repressing melanin production on the thorax. Using reporter constructs made with the regulatory sequences of pigmentation enzyme coding genes, we investigate here the role of sr in the establishment of the trident’s pattern and its possible relationship with the positioning of the flight muscles. Using mutants, we show that sr represses the expression of several pigmentation enzyme-coding genes in the thorax epidermis, thus shaping the trident. The flight muscle attachment sites and sr expression pattern are remarkably conserved in dipterans, reflecting the essential role of sr^36,37. Our data suggest that the shape of the trident is a by-product of flight muscle attachment site patterning that arose when sr was co-opted for the regulation of pigmentation enzyme-coding genes.

Results and Discussion

stripe represses melanin production on the thorax

The pattern of the trident varies among Drosophila melanogaster lines, being absent in some of them (w¹¹¹⁸) whereas clearly visible in others (41Jd) (Fig. 1a,b). In ebony mutant (e¹), the trident is very visible (Fig. 1c) as previously reported²⁸. The trident is not limited to Drosophila melanogaster as it is reminiscent of the pigmentation pattern observed in another Drosophila species, Drosophila busckii (Fig. 1d).

Figure 1.

The trident pattern in Drosophila. (a) Absence of trident in w¹¹¹⁸. (b) Trident clearly visible in 41Jd. (c) Dark trident of an ebony e¹ mutant. (d) Distinctive trident in Drosophila busckii. (e) Trident of an ebony e^s mutant. (f) Dark longitunal stripe in an e^s sr¹ double mutant. (g) Trident still visible in an y¹, e^s mutant background. (g) Longitudinal dark band in a y¹; e^s sr¹ mutant.

Apparent complementarity between sr expression pattern and the trident in the pupal thorax³⁶ suggests that sr might be involved in trident patterning. To address this question, we took advantage of the sr¹ allele of sr. This is a very old allele, as it is reported to have been isolated by Calvin Bridges (1889–1938)³¹. In sr¹ flies, the trident is replaced by a broad longitudinal stripe, hence the name of the gene³¹. It is only thanks to the much more recent description of the sr expression pattern and to the characterization of the sr¹ allele that this phenotype can be interpreted. It was shown that sr¹ is a regulatory mutant, which loses the most dorsal domains of sr expression on the thorax³⁶. To render the dark longitudinal stripe more visible, we combined sr¹ with the e^s hypomorphic allele³¹. The dark dorsal longitudinal stripe induced by sr¹ was perfectly visible in the e^s background (compare Fig. 1e,f). These data suggest that the dark longitudinal stripe is a modified trident, in which spaces between teeth are filled with melanin.

These spaces correspond to the dorsal domains of sr expression that are missing in sr¹, which suggests that sr shapes the trident by repressing melanin production. This repression could occur through different mechanisms implicating the expression, the stability or/and the activity of one or several pigmentation enzymes. Production of cuticle pigments involves many enzyme-coding genes arranged into a pathway^38,39. Interestingly, among them, ple (encoding the Tyrosine Hydroxylase), Ddc (encoding the Dopa decarboxylase), yellow (y), and e were shown previously to be expressed in the trident^28,40,41. Over-expression of y in the dorso-medial domain in an e mutant background is sufficient to generate a homogenous black pigmentation²⁸. Thus, restriction of y expression to the trident is sufficient to explain its delimitation. However, in a y¹, e^s double mutant, the trident is still visible (Fig. 1g) as previously reported²⁸ and in a y¹; sr¹, e^s triple mutant, the longitudinal pattern typical of sr¹ is clearly visible (Fig. 1h). This implies that other pigmentation enzymes than y and e are involved in the patterning of the trident downstream of sr.

Expression of stripe and tan in the thorax are complementary

Most pigmentation enzyme genes were previously shown to be expressed in the trident^28,40,41. However, expression of tan (t), which encodes an enzyme involved in melanin synthesis, was never analysed in the thorax. Using a transgene, in which the expression of nuclear enhanced green fluorescent protein (nEGFP) was driven by the t abdominal enhancer t_MSE^42,43, we observed nEGFP expression in the trident (Fig. 2a), showing than t_MSE was also activated in this motif. To reveal sr expression, we used the enhancer trap line sr^md710 (sr-Gal4)³⁶ and the UAS-mCherry-NLS transgene. mCherry expression was visible in the notum of pupae, where it precisely labelled the flight muscle attachment sites³⁶ (Fig. 2b). By combining (sr^md710, UAS-mCherry-NLS) with the t_MSE-nEGFP transgene, we observed a remarkable complementarity between the patterns of mCherry and nEGFP (Fig. 2c). All flight muscle attachment sites expressing mCherry corresponded to regions where nEGFP was absent, and a high nEGFP level was observed outside of the flight muscle attachment sites.

Figure 2.

Complementary expression of stripe and tan in the thorax. Thorax of a UAS-mCherry-NLS/t_MSE-nEGFP; sr-Gal4/+ freshly eclosed fly. (a) nEGFP showing the activity of t regulatory sequences; (b) mCherry showing sr expression; (c) Merge.

stripe delimits the shape of the trident by repressing multiple pigmentation genes

Complementarity between sr and t expression domains on the thorax suggests that sr might repress the expression of t. This could also be the case for all pigmentation enzyme genes. Then, to draw up a precise and complete analysis of the expression domains of pigmentation enzymes in the thorax, as compared to the expression domain of sr, we used Ddc-Gal4⁴⁴ and ple-Gal4⁴⁵ associated with the UAS-mCherry-NLS transgene as well as transgenes expressing nEGFP under the control of t, e or y regulatory sequences^13,41,42. In the control background, all reporters were expressed in the thorax with patterns that resembled the trident (Fig. 3a,c,e,g,i) suggesting that the expression of pigmentation enzyme genes was constrained to this motif. By contrast, in the sr¹ background, expression of the reporters was extended dorsally and the trident motif disappeared (Fig. 3b,d,f,h,j as compared to a, c, e, g, i, respectively). These data indicate that sr represses Ddc, ple, t, e and y in the thorax. Hence, the shape of the trident likely reflects the spatial regulation of pigmentation enzyme coding genes by Stripe.

Figure 3.

Effect of the sr¹ mutation on the expression of pigmentation enzyme coding genes. (a) Ddc expression in thorax of pharates visualized using Ddc-Gal4 and UAS-mCherry-NLS transgenes (UAS-mCherry-NLS/+; Ddc-Gal4). (b) Ddc expression in a sr¹ background (UAS-mCherry-NLS/+; Ddc-Gal4, sr¹/sr¹). (c) ple expression visualized using ple-Gal4 and UAS-mCherry-NLS transgenes (UAS-mCherry-NLS/+; ple-Gal4/+). (d) ple expression in a sr¹ background (UAS-mCherry-NLS/+; ple-Gal4, sr¹/sr¹). (e) t expression followed using the t_MSE-nEGFP transgene. (f) t expression in a sr¹ background (t-MSE-nEGFP; sr¹). (g) e expression visualized using the e-nEGFP transgene. (h) t expression in a sr¹ background (e-nEGFP, sr¹/sr¹). (i) y expression followed using the y-wb-nEGFP transgene. (j) y expression in a sr¹ background (y-wb-nEGFP; sr¹).

Conclusion

We show here that sr regulates several pigmentation enzyme-coding genes on the thorax, although we do not know whether this regulation is direct or not. The expression of sr is conserved in Calliphora vicina, a species that diverged from Drosophila lineage about 100 million years ago, suggesting that this gene is a member of a deeply conserved regulatory landscape³⁷. sr plays an essential role in the establishment of flight muscle attachment sites, and conservation of its expression is mirrored by a remarkable conservation of the flight muscle apparatus in dipterans⁴⁶. In contrast, a clear thoracic trident complementary to the flight muscle attachment sites is observed in only a few species of flies, notably in Drosophila melanogaster²². Our results suggest that the co-option of sr for the regulation of pigmentation enzyme coding genes has led to the generation of the trident, a pigmentation pattern complementary to the flight muscle attachment sites. Therefore, the shape of the trident primarily results from a developmental constraint imposed by the flight muscle pattern. It is typically a “spandrel” in the sense of Gould and Lewontin (1979), a by-product of the development of another structure. The same applies to the position of large bristles (macrochaetes) on the thorax. Indeed, macrochaetes are excluded from the flight muscle attachment sites, and it was shown that the development of macrochaetes and tendon cells on the thorax are mutually exclusive³⁶.

The fact that the trident was originally an indirect target of selection does not exclude that it has later become a direct target of selection. Indeed, clinal variation in the intensity of the trident strongly suggests an adaptive role in thermoregulation^23,24,26. However, natural selection has targeted variation in the intensity of the trident, rather than in its shape that is highly constrained by fixed muscle attachment sites. Furthermore, it is possible that the expression of pigmentation enzymes in the trident confers new properties to the thorax cuticle that are important for flight, such as flexibility or mechanical endurance.

Methods

Fly stocks

w¹¹¹⁸ is an inbred line used as control. The line 41Jb was established by Jean-Michel Gibert from a female caught in Marsais (France). The Drosophila busckii line was established from a female caught in Niort (France) and kindly provided by Dr Laure Teysset. The following stocks were obtained from the Bloomington Drosophila stock centre: e¹ (BL-1658), e^s (BL-498), Ddc-Gal4 (BL-7009), ple-Gal4 aka TH-Gal4 (BL-8848), sr-Gal4 aka sr^md710 (BL-26663) and UAS-mCherry-NLS (BL-38425). The lines y-wing-body-nEGFP, e-nEGFP (containing the regulatory regions ABC + intron) and t_MSE-nEGFP were kindly provided by Dr. Sean Carroll’s laboratory. Flies were grown on standard medium at 25 °C.

Image acquisitions

Thoracic cuticles of flies immerged in 75% ethanol were imaged with a binocular equipped with Leica DC480 digital camera, using the Leica IM50 Image Manager software. Stacks of 4–10 images were generated for each thorax. They were merged using Photoshop. Identical settings were used for all acquisitions.

Fluorescent images were acquired with a Macro-Apotome (Zeiss) with a 63 × objective on freshly decapitated flies immerged in PBS on an agarose substrate. Stacks were composed of around 75–115 pictures. Maximum intensity projections were created. Brightness and contrast were slightly adjusted in Photoshop.

Data availability

Data and materials used in this work are available on request.

Author Contributions

J.M.G. designed the study, performed the experiments and analysed the data. J.M.G., E.M.V. and F.P. wrote the paper.

Competing Interests

The authors declare no competing interests.