Edited by Wendell Roelofs, Cornell University, Geneva, NY, and approved January 9, 2007 (received for review September 18, 2006)
Drosophila melanogaster produces sexually dimorphic cuticular pheromones that are a key component of the courtship behavior leading to copulation. These molecules are hydrocarbons, with lengths of 23 and 25 carbons in males (mainly with one double bond) and 27 and 29 carbons in females (mainly with two double bonds). Here, we describe an elongase gene, eloF, with female-biased expression. The 771-bp ORF encodes a 257-aa protein that shows the highest sequence identity with mouse SSC1 elongase (33%). The activity of the cDNA expressed in yeast was elongation of saturated and unsaturated fatty acids up to C30. RNAi knockdown in Drosophila led to a dramatic modification of female hydrocarbons, with decreased C29 dienes and increased C25 dienes accompanied by a modification of several courtship parameters: an increase in copulation latency and a decrease in both copulation attempts and copulation. Feminization of the hydrocarbon profile in males by using targeted expression of the transformer gene resulted in high expression levels of eloF, suggesting that the gene is under the control of the sex-determination hierarchy. There is no expression of eloF in Drosophila simulans, which synthesize only C23 and C25 hydrocarbons. These results strongly support the hypothesis that eloF is a crucial enzyme for female pheromone biosynthesis and courtship behavior in D. melanogaster.
Mate recognition represents one of the reproductive isolating mechanisms that are among the first steps of speciation. In Drosophila, cuticular hydrocarbons, some of which act as sex pheromones, may contribute to this process (1, 2). They are long-chain cuticular hydrocarbons, which act at short distances or by contact (3). The information is carried by the chain length and the number and position of double bonds in the molecule. Drosophila simulans and Drosophila melanogaster, although diverged for only ≈2.5 million years (4), show a very different hydrocarbon pattern: the former species is sex monomorphic relative to its hydrocarbons, with the presence of large quantities of Z 7-monoenes, as well as in production of 23 and 25C in both sexes. The latter species is sexually dimorphic, with the presence of Z 7-monoenes with 23 and 25C in males, and of Z,Z 7,11-dienes in females with 27 and 29C (5–7). In both species, the main female hydrocarbon stimulates courtship behavior by conspecific males. These are: 7-tricosene (7-T) in D. simulans and 7,11-heptacosadiene (7,11-HD) in D. melanogaster (3, 5). Other less abundant hydrocarbons, 7,11-nonacosadiene (7,11-ND), 7-pentacosene and 9-pentacosene, have also been shown to excite D. melanogaster males (3, 8).
Hydrocarbon biosynthesis in D. melanogaster occurs by the same elongation–decarboxylation mechanism characterized in Musca domestica (9–12). The first desaturation step involves the Desat1 Δ9 desaturase; this enzyme is present in both sexes of D. melanogaster and D. simulans and transforms palmitic acid to palmitoleic acid (13, 14). Palmitoleic acid is then elongated to long-chain ω7 fatty acids and/or desaturated by the second D. melanogaster female-specific DesatF desaturase to form ω7,11 fatty acids (15). These dienic fatty acids are then elongated by elongase(s) to C28:2 and C30:2, the direct precursors of 7,11-HD (C27:2) and 7,11-nonacosadiene (7,11-ND, C29:2).
Little is known about the elongation system in Drosophila. It is thought that the elongase activities are performed by several elongases with chain-length specificity, because the last step in hydrocarbon biosynthesis, reductive decarboxylation, has little chain-length specificity (12, 16). We found 19 potential elongase genes in the D. melanogaster genome based on homology to yeast and mouse elongases, but no elongase has yet been shown to be involved in hydrocarbon biosynthesis. The only elongase that has been functionally characterized in D. melanogaster is specifically expressed in the male reproductive system and is required for the biosynthesis of cis-vaccenyl acetate; this substance is a nonhydrocarbon male pheromone produced in the ejaculatory bulb, which is involved in courtship and aggregation behavior (17–20).
Because linear hydrocarbons, particularly diene pheromones, are longer in females than in males, we hypothesized that an elongase involved in this process might be absent or expressed at a very low level in males. We found that one elongase gene, CG16905, shows female-biased expression. We thus named this elongase eloF. We confirm that eloF mRNA is present at a much higher level in D. melanogaster females, compared with D. melanogaster males and D. simulans adults of either sex. Heterologous expression in yeast showed that EloF could elongate saturated and unsaturated fatty acids up to 30 carbons. Analysis of a transgenic strain containing an RNAi construct derived from eloF showed a large effect in females, with a marked decrease in 7,11-ND and a large increase in shorter hydrocarbons, especially 7,11-PD (7,11-pentacosadiene, C25:2). This hydrocarbon modification was accompanied by decreased female attractivity affecting wild-type male courtship. When female-specific dienes with 27 and 29 carbons were induced in males by expression of the transformer gene (21), eloF expression was also induced. These results strongly suggest that eloF is involved in diene elongation in D. melanogaster females and is under the control of the sex-determination gene cascade.
Amplification of the ORF of eloF by using cDNA derived from Canton-S wild type yielded a fragment of 771 nt with three modifications compared with the Flybase sequence. The deduced protein contains 257 aa and has the same sequence as the protein predicted in Flybase, except an I residue at position 193 in place of a T in the Flybase sequence. These are likely to reflect eloF-locus polymorphisms. The predicted EloF protein contains a HXXHH motif in its central domain as do the other elongases and five hydrophobic, putative transmembrane domains (Fig. 1A). The Drosophila EloF sequence shares 22–33% identity with mouse elongase sequences and has 30.3% identity with Drosophila Elo68. eloF is most related to mouse Elovl1 (33.1% identity).
Sequence analysis and expression of eloF. (A) Alignment of D. melanogaster EloF (EMBL database accession no. AM292552) with mouse Elovl1 (SSC1), Elovl2 (SSC2), Elovl3 (Cig30), Elovl4, Elovl6 (LCE), and Drosophila Elo68α elongases. Identical amino acids among two to five sequences are highlighted in light gray, among six or seven sequences in dark gray. The dotted lines above the alignment indicate the putative transmembrane regions. The conserved histidine motif is boxed. EloF is 33.1%, 30.3%, 22.5%, 30.3%, 21.6%, and 30.3% identical to ElovlI, Elovl2, Elvl3, Elovl6, and Elo68α, respectively. (B) eloF is almost exclusively expressed in D. melanogaster females. Products of RT-PCR from D. simulans and D. melanogaster RNA (Canton-S and Tai males and females), with eloF- and desat1-specific primers. Amplifications of eloF on genomic DNA (g) and of desat1 on cDNA were used as controls.
As seen in Fig. 1B, eloF showed sex and species specificity. It was expressed almost exclusively in D. melanogaster females. Expression in males was not null, but the amplification of eloF by RT-PCR gave a product that was barely detectable after 30 amplification cycles. No detectable amplification was obtained by using cDNA from either male or female adults of D. simulans, although genomic DNA from the latter species could be amplified under the same conditions. Thus, even though the gene is present in both species, female-biased expression is restricted to D. melanogaster.
To characterize the elongation activity of EloF protein, we cloned the eloF ORF into the yeast expression shuttle vector PYEDP. S. cerevisiae strain W303 was transformed with PYEDP (vector alone) or PYEDP-eloF and grown with myristic acid (14:0) and oleic acid (18:1) under uninduced conditions (2% raffinose) and after induction with 2% galactose.
All of the yeast cultures (transformed with PYEDP alone, with PYEDP-eloF, with or without galactose induction) produced C16 and C18 fatty acids in large and equivalent quantities (Fig. 2A). Without galactose induction, no fatty acids longer than C18 were observed. After galactose induction, very-long-chain saturated and unsaturated fatty acids (from C19 to C30) were obtained: ω7 fatty acids probably derive from the elongation of palmitoleic acid (16:1Δ9) normally produced by the yeast endogenous enzyme OLE1 (22). Oleic acid (18:1Δ9) could also been elongated in very-long-chain ω9 fatty acids (Fig. 2 B and C). C30 compounds were not always detectable because of their very low amounts. The figure shows that saturated and monoenic fatty acids (C16 and C18) are elongated to very long (C26 and C28) fatty acids. The amount of C26:0 produced was ≈3 times more than that of C28:0, the amount of C28:1ω7 and ω9 ≈5 and 1.5 more times than that of C26:1ω7 and ω9. Conversely, very low amounts of C20 to C24 fatty acids were obtained. To determine whether EloF could elongate dienic fatty acids, linoleic acid (18:2 ω6 ω9) was added to the medium. In these conditions, a small quantity of linoleic acid (1%) was transformed to 20:2 ω6 ω9. These data suggest that EloF has broad substrate specificity, because it can elongate saturated and monounsaturated fatty acids up to C30. It is also possible that diunsaturated fatty acids might be generated.
Drosophila EloF elongates fatty acids up to 30 carbons in yeast. Fatty acid profiles of W303 yeast transformed with PYEDP-eloF grown with C14:0 and C18:1Δ9 supplementation. Analysis of fatty acid methyl esters by GC-MS from yeast grown without (A) or with (B and C) induction by the addition of galactose is shown. C is a magnification of B, showing the products from C19 to C28 obtained after induction.
UAS-eloF RNAi flies were crossed to OK72-Gal4 flies to investigate the possible impact of eloF RNAi knockdown in wild-type females (Fig. 3). When the RNAi construct was expressed under control of the OK72-GAL4 driver, there was 98% extinction of eloF expression.
eloF knockdown results in a decrease in C27 + C29 HC and an increase in C23 + C25 HC in females. (A) Mean hydrocarbons (± SEM) of 4-day-old control (OK72/+) and RNAi (OK72/eloFRNAi) female flies. n ≥ 10 for all tests. Means with ∗∗ and ∗∗∗ were significantly different with the Mann–Whitney U test (P = 0.01 and 0.001, respectively). (B) RT-PCR of 4-day-old control (C, OK7/ +) and RNAi (OK72/eloFRNAi) female flies.
Hydrocarbons were analyzed and showed no overall quantitative variation. In contrast, the HC profile was greatly changed in RNAi-expressing females but not in males. In RNAi females, 7,11-dienes were dramatically affected, with a 554% increase (+13.2% of the total HC amount) and a 73% decrease (−12% of the total HC amount) in 7,11-dienes of 25 and 29 carbons, respectively. This hydrocarbon profile suggests a diene-biased deficiency in elongation. There was also a small but significant decrease in C27 monoenic and saturated HC at the expense of C23. These data support a role for eloF in the elongation of fatty acids, especially dienic fatty acids, precursors of female pheromones.
In males, eloF RNAi expression resulted in insignificant effects (data not shown).
These data support a role of eloF in the elongation of fatty acids, precursors of hydrocarbons.
The altered profile of RNAi female hydrocarbons might be expected to alter the behavior of mating couples. We therefore tested the effect of eloF knockdown on courtship behavior, using 4-day-old Canton-S tester males (Fig. 4A). RNAi knockdown of eloF led to a decrease in both the number of copulation attempts (−36%) and the Courtship Index (−58%) and an increase in the copulation latency (+72%). The resulting effect showed a decrease in copulation in couples with RNAi females (43.5% copulation) compared with copulation in couples with control females (83.9% copulation).
eloF knockdown has a marked effect on courtship behavior. (A) Mean courtship parameters (± SEM) of 4-day-old tester Canton-S male flies with OK72/RNAi (in gray) or control (OK72/Cy, in white) females. n ≥ 30 for all tests. Means with ∗∗ and ∗∗∗ were significantly different with the Mann–Whitney U test (P = 0.01 and 0.001, respectively). (B) Mating preferences of wild-type males. In the control test, wild-type males (Canton-S) were given a simultaneous choice between a 4-day-old virgin female with straight wings (+/+) or curly wings (+/Cy). In the RNAi experiment, wild-type males were choosing between an RNAi female with straight wings (OK72/RNAi) or a control female with curly wings (OK72/Cy). In this second experiment none of the males mated with an RNAi female. The numbers of tested males were 48 and 54 in the control and RNAi experiment, respectively.
Because these data showed that eloF inactivation in females resulted in a loss of sexual attractiveness for males, we wondered how this modification in female pheromones would influence male choice in the presence of both control and eloF RNAi females.
When given a simultaneous choice between control females with straight wings (+/+) or curly wings (+/Cy), males showed no preferences, demonstrating no wing effect. On the other hand, males mated exclusively with control Cy females when given the choice between control (OK72/Cy) flies and RNAi (OK72/RNAi) flies (Fig. 4B). This result suggests that the modification of HC profile because of reduction of eloF activity in knockdown females has substantial effects on male D. melanogaster mating preference.
transformer expression was targeted in flies by using the OK72-Gal driver. Under these conditions, a marked feminization of hydrocarbons with a large production of dienes of 27 and 29 carbons occurs in males with feminized oenocytes (11, 14). We obtained the same feminization of male hydrocarbons as previously reported : 27% and 12% 7,11-dienes in C27 and C29, respectively (figure 7 in ref. 11). Whereas no expression of eloF could be detected in control males, high levels of eloF expression were observed in TRANSFORMER feminized males (Fig. 5). In females, which normally produce TRA protein, eloF was expressed in both genotypes. These data indicate that eloF is under the control of the sex-determination hierarchy and that transformer expression in a subset of cells is sufficient to induce high level eloF expression.
Feminization of male oenocytes with the transformer gene results in the expression of eloF. eloF was analyzed by RT-PCR for its expression in four-day-old male flies expressing tra under OK72-Gal4 driver. Amplification of rp49 was used as a positive control.
Here, we described an elongase with a previously uncharacterized activity, able to produce very-long-chain fatty acids. The enzyme is involved in hydrocarbon biosynthesis and especially in female dienic pheromone biosynthesis. The gene is almost exclusively expressed in females. A microarray study found that it is approximately four times more expressed in females in comparison with desatF, a female-specific desaturase, which was found to be 18 times more expressed in females (23).
The elongase described here, EloF, shows limited homology to yeast and higher homology to mouse elongases as well as Drosophila Elo68. EloF appears to be an integral membrane protein containing five transmembrane domains, a characteristic shared by previously identified yeast and mouse elongase enzymes (24–26). EloF contains a histidine-rich motif (HXXHH) which is also present in fatty acid elongases and is believed to act as Fe-chelating ligand used for the electron transfer for O2-dependent redox reactions (27). EloF catalyses the elongation of saturated and unsaturated fatty acids to very-long-chain fatty acids. This is a functional characterization of an elongase involved in the elongation of very long chain fatty acids (up to 30 carbons).
Three fatty acid elongases have been characterized in yeast by genetic and functional studies. Whereas Elo1 is restricted to the elongation of short fatty acids, Elo2 elongates palmitoyl-CoA and stearoyl-CoA up to C22 fatty acids. Elo3 elongates fatty acids of 20 to 26 carbons (24). These latter fatty acids are rapidly incorporated into ceramides and sphin golipids and produced quantities too small to be detectable in fatty acid analysis.
Based on the reversion of yeast mutant phenotypes, two mouse elongases were also involved in the synthesis of very long fatty acids: Elovl3 would be able to elongate fatty acids up to C20 to C24 and Elovl1 up to C26 (26). EloF shows the lowest similarities with Elovl6, which is involved in the synthesis of stearic acid (28). Its highest similarity is with mouse Elovl1, which is involved in the synthesis of saturated and monounsaturated VLCFAs with 26 and 24 carbons, respectively (26, 29, 30).
In invertebrates, several elongases have been characterized, and the polyunsaturated fatty acid biosynthetic pathway in Caenorhabditis elegans has been reconstituted in yeast (31). Contrary to mammals and Drosophila, C. elegans cannot elongate fatty acids beyond 20C. The most similar pathway known to Drosophila pheromones synthesis is found in Musca (32), where an elongation step from C24 to C28 has been described. This elongase is negatively regulated by ecdysteroids (33). This regulation results in the production of 9-heptacosene (C27:1) in males and immature females and of 9-tricosene (C23:1) in mature females, which is the main pheromone in Musca (34). In Drosophila, the elongase described here has a pivotal role in the elongation of shorter compounds (C23, C25), which are more abundant in males to long ones (C27, C29), which are more abundant in females. Contrary to Musca elongase, this elongation step does not seem to be regulated by ecdysteroids. In fact, in ecdysoneless mutants, there is a loss of 7,11-dienes in C27 and C29 and an accumulation of 7-monoenes in C27 and C29 (35), suggesting that the global elongation of C27 and C29 unsaturated HC (dienes + monoenes) is not significantly affected.
In the zebrafish, an elongase was described that could lengthen the range of C18, C20, and C22 polyunsaturated fatty acids and was able to elongate saturated, monounsaturated and three unsaturated fatty acids (36). This observation suggests that the specificity of elongases is quite variable, depending on the enzymes, because one elongase enzyme can perform three elongation steps and use a variety of substrates.
EloF elongates both saturated and unsaturated fatty acids up to C30 in yeast. Saturated and monounsaturated fatty acids are mostly elongated in C26 and C28, respectively. In Drosophila, the absence of significant EloF activity affected all hydrocarbon classes, but there was a particularly severe drop of C29 dienic hydrocarbons in females and an increase in C25 dienic hydrocarbons. Monoenic and saturated hydrocarbons were affected to a lesser degree, with a decrease in C27 at the expense of C23. This suggests that the elongation of C26 dienic fatty acids (precursors of C25 HC) to C30 dienic fatty acids is dramatically decreased and, likewise, that of C24 saturated and monoenic fatty acids to C28 ones. When eloF RNAi was expressed in Tai females, which produce large amounts of 5,9-dienes, similar results were obtained, especially a marked decrease in 5,9-dienes in C29 at the expense of C25 ones (data not shown). This shows that EloF is not only crucial for the production of 7,11-dienes but also of 5,9 ones. The effect of RNAi-mediated disruption of eloF is negligible in males where eloF is expressed at a much lower level. As stated above, males have very few hydrocarbons with 27 and 29 carbons. The EloF gene is expressed in D. melanogaster females, which synthesize long hydrocarbons. Consistent with the HC elongation model, eloF is not expressed in D. simulans, which synthesize shorter (C23 and C25) hydrocarbons. Therefore, there is a good correlation between eloF expression levels in Drosophila and hydrocarbon chain length.
In D. melanogaster, DesatF has no action on unsaturated substrates from C14 to C20 and may use a longer (C22 to C26) fatty acid as substrate (data not shown). Flies expressing desatF RNAi are characterized by a large decrease in C27 and C29 dienes and increase in C27 and C29 monoenes, suggesting that elongation was not significantly affected (14). The action of EloF in vitro (in yeast) and in vivo suggests that it could act on saturated, monounsaturated, and diunsaturated fatty acids. Even if the action of other elongase(s) cannot be excluded, the broad specificity of EloF may be parallel to that of the zebrafish elongase.
We tried to feminize male hydrocarbons by expressing the transformer gene under OK72-Gal4 driver, as already reported (15, 21). The feminization of hydrocarbons under these conditions was accompanied by the expression of desatF, which normally is expressed only in females (13). In this study, we also provide evidence that eloF is highly expressed in transformer males. We also simultaneously expressed desatF and eloF in males, but failed to obtain any hydrocarbon feminization (data not shown). Again, this is an indication that other genes are required for male hydrocarbon feminization. Continued work on characterizing the enzymes in the pheromone production pathway should enable us to reconstitute the pathway in males or perhaps yeast.
We tested the effect of eloF knockdown in females on courtship behavior in males. We saw that the main effect on female hydrocarbons consists in decreased levels of C29 dienes and increased levels of C25 dienes. This shift in diene length resulted in decreased attractiveness, because control males mated half as often with these females, compared with control females. Several parameters of courtship behavior were affected, especially copulation latency and the number of copulation attempts. Female unsaturated hydrocarbons, 7-monoenes and 7,11-dienes with 27 ± 2 carbons have been reported to induce male wing vibration (3, 37, 38) and 9-pentacosene has been reported to induce copulation attempts (8). The 7,11-dienes were also shown to increase the frequency and rapidity of wild-type males to mate (38–40). In a previous study, we showed that the amount of 7,11-dienes markedly affected the number of copulation attempts and copulation latency (15). In this study, we show that the length of dienes is also important for courtship behavior. Moreover, when given a choice between control or RNAi females, males exclusively chose to mate with control flies, confirming the importance of diene length in recognition of females by males and subsequent interactions between such flies.
This study shows the crucial role of an elongase gene, eloF, in female long-chain hydrocarbon biosynthesis and courtship behavior. EloF is able to elongate medium-chain fatty acids up to long chain (C30) in vitro. In Drosophila, eloF expression is correlated with the length of the hydrocarbons formed. eloF is not expressed in D. simulans, which does not synthesize long-chain hydrocarbons. Together with desatF, eloF could have played a role in sexual isolation and in the evolution of species.
Drosophila were kept at 25°C in a 12-hr light and 12-hr dark cycle on standard yeast/cornmeal/agar medium.
D. melanogaster, D. simulans, and Drosophila sechellia strains used in this study have been described (15).
The UAS-tra and OK72-Gal4 strains were obtained from the Drosophila Stock Center (Bloomington, IN) and were used as described (15, 21). OK72-Gal4 drives UAS expression mainly in oenocytes and leads to a feminization of male pheromones in males expressing tra under its control.
Males and females were sexed within 1 hour after emergence and held separately until use at 4 days posteclosion.
Extractions and analyses of hydrocarbons were performed as described (13). Data are shown as mean percentages of hydrocarbons (n ≥ 10 for all tests).
Two types of courtship behavior experiments were performed. In the first, one 4-day-old virgin female (control or mutant) was tested with one 4-day-old wild-type (Canton-S) male, as described (41). Parameters measured were: the time until wing vibration (courtship latency); the time until copulation (copulation latency); the number of attempted copulations in 15 min; the percentage of time spent actively courting by the male (Courtship Index) (42); n ≥ 30 for all tests.
In the second set of experiments, wild-type males (Canton-S) were given a simultaneous choice between a 4-day-old virgin control and a mutant female. n ≥ 40 for all tests.
The sequence of CG16905 was obtained from the Drosophila genome database (http://flybase.bio.indiana.edu). Genomic DNA or RNA extraction and cDNA synthesis were performed as described (15). The CG16905 ORF was amplified by PCR for 30 cycles with the two primers: 5′-ATGTTCGCTCCGATAGATCC-3′ and 5′-TTAATTTTTGTTTTGTTTCGC3′ and cloned into a Bluescript vector, as described (13). Five clones were sequenced and showed the same sequence.
For PCR on cDNA from Canton-S or UAS-tra flies, the control rp49 gene, coding for ribosomal protein 49 (43) was amplified by using the primer pair 5′-CAGAATCTTATGACCATCCGC CCAGCATAC-3′ and 5′-CAGGAATTCAACGTTTACAAATGTGTATTC-3′, providing an amplification fragment of 462 bp.
For PCR on D. melanogaster (Canton-S and Tai strains) and D. simulans, two controls were performed: amplification of CG16905 on genomic DNA and amplification of cDNA with desat1 primers, as described (15). Amplification of cDNA with these primers yielded a fragment of 545 bp.
The eloF ORF was cloned directly into a PYEDP vector at BamHI and KpnI sites (44). This vector includes a galactose-inducible chimeric promoter GAL10-CYC1, a terminator of yeast phosphoglycerate kinase, and an auxotrophic marker, URA3. The resulting galactose-inducible constructs were introduced into Saccharomyces cerevisiae strain W303 (MATα, leu2–3, leu2–1/2, can1–100, ura-3, ade2–1, his3–11, and his3–15)by using the lithium acetate method, and yeast transformants were obtained as described (17). The expression of the transgene was induced by the addition of 2% galactose (and 2% rafinose) and followed by 30 h of incubation. Fatty acids were extracted from yeast in chloroform/methanol, transformed into methyl esters, and analyzed by GC/MS (14). Double positions were analyzed by the generation of the DMDS derivates from the methyl esters, as described (17).
The data are presented as means ± SEM. An analysis of variance (Student's t test or Mann–Whitney U test) was used for the statistical analyses, and P < 0.05 was accepted as statistically significant.
We thank Professors Jean-Marc Jallon, Leonard Rabinow, and Brian Oliver for thoughtful comments on the manuscript. This work was supported by the French Ministry of Research and Education (T.C., L.D., and C.L.) and the Centre National de la Recherche Scientifique (C.W.-T.).
Author contributions: C.W.-T. designed research; T.C., L.D., C.L., and C.W.-T. performed research; T.C., R.U., K.T., and K.S. contributed new reagents/analytic tools; T.C. and C.W.-T. analyzed data; and C.W.-T. wrote the paper.
Conflict of interest statement: The authors declare no conflict of interest.
This article is a PNAS direct submission.
Data deposition: The sequence reported in this paper has been deposited in the EMBL database (accession no. AM29552).
