Rapid evolution of sex pheromone-producing enzyme expression in Drosophila

Abstract

A wide range of organisms use sex pheromones to communicate with each other and to identify appropriate mating partners. While the evolution of chemical communication has been suggested to cause sexual isolation and speciation, the mechanisms that govern evolutionary transitions in sex pheromone production are poorly understood. Here, we decipher the molecular mechanisms underlying the rapid evolution in the expression of a gene involved in sex pheromone production in Drosophilid flies. Long-chain cuticular hydrocarbons (e.g., dienes) are produced female-specifically, notably via the activity of the desaturase DESAT-F, and are potent pheromones for male courtship behavior in Drosophila melanogaster. We show that across the genus Drosophila, the expression of this enzyme is correlated with long-chain diene production and has undergone an extraordinary number of evolutionary transitions, including six independent gene inactivations, three losses of expression without gene loss, and two transitions in sex-specificity. Furthermore, we show that evolutionary transitions from monomorphism to dimorphism (and its reversion) in desatF expression involved the gain (and the inactivation) of a binding-site for the sex-determination transcription factor, DOUBLESEX. In addition, we documented a surprising example of the gain of particular cis-regulatory motifs of the desatF locus via a set of small deletions. Together, our results suggest that frequent changes in the expression of pheromone-producing enzymes underlie evolutionary transitions in chemical communication, and reflect changing regimes of sexual selection, which may have contributed to speciation among Drosophila.

Figure 1. desatF is expressed female-specifically in D. melanogaster oenocytes.

In situ hybridization for desatF performed on 4-d-old adult D. melanogaster flies revealed desatF expression in abdominal oenocytes in females (A, purple stripes) but not in males (B), These cells are the sites of cuticular hydrocarbon production.

Figure 2. desatF expression correlates with evolutionary transitions in diene production from sexual monomorphism to dimorphism.

Left column: phylogenetic relationships of the species used in this study. The D. melanogaster species subgroup is highlighted in red. Transitions in diene production from sexual monomorphism to dimorphism, and from dimorphism to amorphism, are indicated by a green and a black arrowhead, respectively. Middle column: diene production in males and females of these species. Diene production across the Sophophora subgenus shows several transitions: from monomorphism to dimorphism in the most recent common ancestor of the D. melanogaster species subgroup, and from dimorphism to amorphism in D. simulans (and other species not shown in the D. melanogaster species subgroup including D. mauritiana, D. yakuba, D. santomea, D. teissieri, and D. orena; summarized in Figure 7). We refer to dienes as any long-chain hydrocarbon (longer than 20 carbons) with two double bonds. Published reports of hydrocarbon profiles in these species are indicated in the column titled Ref. We have also independently validated the hydrocarbon profiles of D. melanogaster and D. simulans (unpublished data). Our analysis of D. takahashii males and females showed that they both produce a C23 diene (unpublished data). *, Dienes are present on D. sechellia males, but account for less than 2% of the total amount of cuticular hydrocarbons present on the fly . The putative expression of desatF in D. sechellia males is likely to be below the detection capabilities of in situ hybridization analyses. Right column: In situ hybridization of desatF expression performed on 4-d-old adults. In all species studied, we observed expression of desatF in abdominal oenocytes (purple stripes) in accordance with their status of diene production.

Figure 3. cis-regulatory sequence evolution governs the gain of female-specific expression of desatF.

(A) Phylogenetic relationships of the species for which the activity of the desatF upstream regulatory region was assessed. Members of the D. melanogaster species subgroup are indicated in red. (B–M) Confocal images of the dorsal view of the abdomen from 4-d-old D. melanogaster females (B, D, F, H, J, and L) and males (C, E, G, I, K, and M) carrying two copies of the eGFP reporter transgene driven by the desatF CRE of each species indicated on the left. Note that, except for D. sechellia, all transgenes tested recapitulated the endogenous expression of the species they were derived from, indicating that functional differences in cis-regulatory sequences account for the transition from monomorphic to dimorphic expression of desatF.

Figure 4. DSX-F is directly required to activate female-specific expression of desatF in adult oenocytes.

The OK72-Gal4 driver specifically targets oenocyte tissue in females (A) and males (B). X-Gal staining performed on 4-d-old D. melanogaster of the genotype indicated. Blue staining indicates that the activity of the OK72-Gal4 driver is restricted to oenocytes (arrow) and two groups of cells (arrowhead) collinear to the dorsal vessel. These preparations retained fat body tissue, which appear to lack β-Gal activity, indicating that this driver does not target the fat body. (C–H) In situ hybridization for desatF performed on 4-d-old females (C, E, and G) and males (D, F, and H) of the D. melanogaster genotype indicated. Compared to the wild-type-like expression observed in control flies carrying just the GAL4 driver (C and D) and the dsx-RNAi (E and F) transgenes, desatF expression is lost in specimens expressing the dsx-RNAi driven by the OK72-GAL4 construct (G and H). (I) EMSAs were performed on annealed radiolabeled oligonucleotide probes containing the wild-type and mutant (mutated nucleotides in red) DSX-binding sites with increasing amounts of DSX-DBD protein. For probes containing the wild-type binding site, as the amount of DSX-DBD increased (lanes 1–5) a correlative increase in the amount of probe bound was observed. Protein binding was significantly reduced when the DSX-binding site was mutated (lanes 6–10). Arrow and arrowhead points to a single and pair of DSX-DBD monomers bound to the probe respectively. Asterisk identifies free probe. (J–Q) Confocal images of dorsal abdomens from 4-d-old D. melanogaster females (J, L, N, P) and males (K, M, O, Q) carrying two copies of an eGFP-reporter transgene. Reporter constructs are indicated at the top of the columns. The mel-oe (J and K) and ere-oe (N and O) transgenes recapitulate endogenous expression of desatF. However, when the DSX-binding is mutated in these constructs, they fail to drive eGFP-reporter expression in females (L and P, respectively). Reporter expression in males (M and Q, respectively) is not upregulated.

Figure 5. Monomorphic expression of desatF in D. takahashii evolved by functional inactivation of a DSX-binding site.

(A) The DSX-binding site predates the D. melanogaster species subgroup (in red). Left panel: phylogenetic relationships of the species surveyed for the presence of a putative DSX-binding site (adapted from [27]). Others have positioned D. eugracilis and D. ficusphila differently in the phylogeny . Our results are consistent in either case. Black arrowhead: inferred origin of desatF dimorphic expression, based on the phylogenetic distribution of desatF expression (see Figure 2). Green arrowhead: inferred origin of the DSX-binding site. Middle panel: sequences of the putative orthologous DSX-binding site of each species. Purple residues indicate positions that have diverged from the D. melanogaster site. Gray box identifies the critical residue within the core that has diverged in D. takahashii. Right panel: desatF expression summary. For D. trilutea, adult flies were not available and therefore desatF expression could not be assessed. desatF has been deleted in D. eugracilis and therefore assessing its expression was irrelevant. (B) EMSAs comparing the ability of the DSX-DBD protein to bind annealed radiolabeled oligonucleotide probes containing the D. melanogaster DSX-binding site (lanes 1–5) and a mutated version of this site (lanes 6–10) containing a C to A point mutation (in red), as found in D. takahashii. This mutation greatly reduced binding of the DSX-DBD. (C) EMSAs comparing the ability of the DSX-DBD protein to bind annealed radiolabeled oligonucleotide probes containing the D. takahashii putative DSX-binding site and a mutated version of this site containing a C in the core of the putative site instead of an A, which is found in the consensus and in the D. melanogaster site. In contrast to the wild-type D. takahashii site (lanes 1–5), where no significant binding is observed, the DSX-DBD protein binds the mutant site relatively efficiently (lanes 6–10). The arrow and arrowhead point to a single and pair of DSX-DBD monomers bound to the probe. The asterisk marks the position of the free probe. (D–K) eGFP reporter expression in abdomens of 4-d-old D. melanogaster females (D, F, H, and J) and males (E, G, I, and K) and carrying two copies of the transgenes indicated at the top of the columns. Introducing a C to A point mutation in the DSX-binding site of mel-oe1 abolishes eGFP reporter expression in females (compare D and F), while leaving the absence of expression in males unchanged (compare E and G). Introducing an A to C point mutation in the putative DSX-binding site of tak-oe produces sexually dimorphic eGFP expression (J, K), whereas a wild-type tak-oe drives monomorphic expression (H, I).

Figure 6. Cis-regulatory information was gained by deletion during D. melanogaster evolution.

(A) The AATTTG motif is statistically overrepresented in mel-oe2. Schematic representation of the AATTTG motifs (green dot over black bar) in mel-oe2 and its orthologous sequence from D.simulans and D. erecta. (B and C) eGFP reporter expression in abdomens of 4-d-old D. melanogaster females carrying two copies of the transgenes indicated at the top of the images. The introduction of point-mutations in the clustered AATTTG motifs of mel-oe1 (C) abolishes eGFP reporter expression driven in female oenocytes by a wild-type mel-oe1 (B).The absence of reporter activity in males is not altered by these mutations (not shown). (D) Alignment of the desatF upstream region from D. melanogaster, D. simulans, and D. erecta. mel-oe2 and its orthologous sequences from D. simulans and D. erecta are delineated by vertical black bars. The vertical red bar indicates the 3′ end of mel-oe1. mel-oe1 and mel-oe2 begin at the same 5′ position. AATTTG motifs are boxed in green (forward orientation) and blue (reverse orientation). Black stars (*) indicate conservation among the three species. The red plus sign (+) indicates conservation between D. simulans and D. erecta. The beginning of the coding region is in yellow. The D. erecta 190-bp sequence that is necessary (in addition to ere-oe2) to produce a construct capable of full reporter activity in D. melanogaster female oenocytes is represented in grey. Note the very well conserved indels in D. simulans and D. erecta, which disrupt each of the three AATTTG motifs in the cluster, indicating that those hexamer motifs evolved by a series of small deletions.

Figure 7. The desatF locus is rapidly evolving.

(A) Phylogenetic relationships of the 24 species surveyed (adapted from [27]). (B) Schematic of the desatF locus in these species. The blue rectangles indicate the landmarks used in cloning. In D. ananassae, desatF was found in the genome, but not in synteny, which is indicated by the absence of the blue rectangles. The orange boxes indicate the coding region. A striped box indicates a mutation in the coding region leading to a loss of function of the protein (frameshift or nonsense mutation). Black and brown full circles represent regions with repetitive DNA. The six independent gene losses are indicated by red bars. Regulatory losses of expression without gene inactivation are marked by a black bar. Modifications in the sex-specificity of desatF expression are represented by a pink bar. Green “R” refers to regulatory transitions. Altogether, 11 independent evolutionary changes in desatF expression occurred in the approximate 40 millions y during which these species evolved. *, Note that the gene inactivations in D. yakuba and D. lutescens are not counted as such in our tally. In D. yakuba, the regulatory loss of desatF expression appears to have preceded the pseudogenization event. In D. lutescens, the ambiguous phylogenetic relationships in the clade prevents the accurate inference of transitions. Grey full circles indicate independent losses of dimorphism. (C) Status of desatF expression in oenocytes in 4-d-old adults aged. N/R: not relevant (because the gene was not functional based on sequence information).