Genetic Control of Courtship Behavior in the Housefly: Evidence for a Conserved Bifurcation of the Sex-Determining Pathway

In Drosophila melanogaster, genes of the sex-determination hierarchy orchestrate the development and differentiation of sex-specific tissues, establishing sex-specific physiology and neural circuitry. One of these sex-determination genes, fruitless (fru), plays a key role in the formation of neural circuits underlying Drosophila male courtship behavior. Conservation of fru gene structure and sex-specific expression has been found in several insect orders, though it is still to be determined whether a male courtship role for the gene is employed in these species due to the lack of mutants and homologous experimental evidence. We have isolated the fru ortholog (Md-fru) from the common housefly, Musca domestica, and show the gene’s conserved genomic structure. We demonstrate that male-specific Md-fru transcripts arise by conserved mechanisms of sex-specific splicing. Here we show that Md-fru, is similarly involved in controlling male courtship behavior. A male courtship behavioral function for Md-fru was revealed by the behavioral and neuroanatomical analyses of a hypomorphic allele, Md-traman, which specifically disrupted the expression of Md-fru in males, leading to severely impaired male courtship behavior. In line with a role in nervous system development, we found that expression of Md-fru was confined to neural tissues in the brain, most prominently in optic neuropil and in peripheral sensory organs. We propose that, like in Drosophila, overt sexual differentiation of the housefly depends on a sex-determining pathway that bifurcates downstream of the Md-tra gene to coordinate dimorphic development of non-neuronal tissues mediated by Md-dsx with that of neuronal tissues largely mediated by Md-fru.


Introduction
The genetic pathway controlling sex-specific behaviors in the invertebrate model system Drosophila melanogaster has been elucidated in great detail [1,2]. These studies revealed that the genetic control of male mating behavior in Drosophila melanogaster is largely governed by the transcription factor fruitless (fru) [3,4]. This gene acts as a pivotal switch for the correct developmental implementation of male-specific behaviors. A specific class of fru alleles have been identified which impairs courtship performance of males, but leaves females unaffected [5,6,7]. Moreover, it was demonstrated that expression of the male-specific fru functions in genotypically female flies elicits some aspects of male courtship behavior [8,9,10,11]. However, the conclusion that fru is not only necessary but also sufficient for implementing male mating behavior in Drosophila has been challenged in recent studies [12,13,14,15,16]. These studies suggest that the sex-determining double-switch gene doublesex (dsx) also contributes to the regulation of male mating behavior.
The fru gene in Drosophila melanogaster is very large in size spanning approximately 150 kb of genomic sequences. Transcrip-tion initiates from four independent promoters (P1-P4) and the resulting transcripts undergo a complex pattern of alternative splicing events at both the 59 and the 39 ends, generating 18 different transcript variants which code for a family of transcription factors containing a common BTB domain and one of four alternative zinc-finger DNA binding domains [17]. The distal most promoter P1 gives rise to transcripts, which are implicated in regulating sex-specific behaviors [3,4]. Transcripts derived from the other three promoters are non sex-specific and have vital developmental functions in both males and females [18]. Sexspecific expression is achieved by alternative splicing at the 59 end of P1 transcripts mediated by the splice regulator transformer (tra), a key component of the sex-determining pathway. In females, TRA protein forms a splice enhancer complex with TRA2 and RBP1 (RNA Binding Protein 1) binding to nascent fru RNAs where they act to alter splice site choice, resulting in the inclusion of a pre-mature stop codon signal and thereby generating femalespecific transcripts (fru F ) with a severely truncated ORF [3,19]. In males, where the tra gene is functionally OFF, the default splicesite is used and the resulting male-specific transcripts (fru M ) encode functional full-length BTB-Zn-finger type transcription factors.
These sex-specific transcripts incorporate alternate 39 exons A, B, or C to encode the Fru M isoforms: Fru MA , Fru MB and Fru MC . Fru M isoforms are translated only in males and are expressed in all regions of the nervous system implicated in male courtship behavior [12,20,21,22,23]. These FRU M positive neurons located in widely distributed clusters throughout the male brain and nerve chord are key parts of the neuronal circuitry required for behavioral responses to sex pheromones and the display of proper male courtship behavior.
Given the critical role of fru in controlling sex-specific behaviors the question arises whether this function is conserved in other insects. The existence of sex-specifically expressed fru homologs in various dipteran and in hymenopteran species (Anopheles gambiae [24], Aedes aegypti [25], Ceratitis capitata [26] and Nasonia vitripennis [27]) suggests that fru has a conserved male-specific function in holometabolous insects. However, functional studies have yet to establish a definitive role in male courtship behavior in these species.
In our study we used the common housefly, Musca domestica, as an experimental system to investigate the genetic control underlying male courtship behavior. In previous studies we showed that all aspects of sexually dimorphic development including behavior are governed by a key switch gene, previously referred to as F and more recently identified as an ortholog of the transformer gene, Md-tra [28,29]. A male-determining M-factor, which is located on the Y chromosome in standard strains but can also be found on autosomes in natural populations, serves as the primary instructive signal [29]. When M is absent, zygotic Md-tra is spliced in a mode that generates transcripts with an intact ORF. These Md-tra messages code for a RS-type of splicing regulator, which through an autocatalytic function upholds the productive splicing of its own pre-mRNA. Maternally deployed Md-tra activity is required to engage this positive feedback loop in the early zygote [28]. Presence of M in the zygote invokes a change in splicing of Md-tra, which results in the production of messages with a truncated ORF. Likewise, a previously identified target of Mdtra, the Musca doublesex homologue, Md-dsx, is regulated by sexspecific processing of its pre-mRNA [30]. Md-TRA together with its co-factor Md-TRA2 directs the female-specific splice mode of Md-dsx generating a female-specific variant which instructs female differentiation [31]. In M containing individuals, where Md-tra remains OFF, the male-specific splice mode of Md-dsx is established by default and a male-specific isoform is generated thereby ensuring male development.
Here we report the identification of the housefly ortholog of fruitless, Md-fru, and show that this gene is a downstream target of Md-tra. In an analogous way to Drosophila, Md-fru expresses functional splice variants only in males where Md-tra is OFF. Residual levels of Md-tra activity in males homozygous for the hypomorphic Md-tra man allele, prevents expression of functional Md-fru products leading to sterility. Our findings suggest that the correct implementation of male mating behavior is governed by a specific branch of the sex-determining pathway mediated by fru in both Musca and Drosophila.

Results
Md-fru is Sex-specifically Regulated and a Target of Mdtra In Drosophila, the pathway downstream of tra bifurcates and employs the gene fruitless (fru) to control various aspects of sexually dimorphic CNS development [18,32]. To establish if Md-tra also regulates housefly courtship through a fru-like downstream effector we set out to isolate the fru ortholog in Musca. PCR amplification with degenerate primers flanking the conserved BTB domain of fru yielded a genomic fragment of 190 bp in length. Subsequent extension by 39 RACE of cDNAs prepared from adult houseflies revealed the existence of four alternative mRNA transcripts (Fig. 1). All 4 variants were present in female and male cDNA preparations and each encoded a different zinc finger type motif at its carboxyterminal end. The same complex pattern of 39 end processing has been reported in fru orthologs of Drosophila melanogaster, Anopheles gambiae, Aedes aegypti, Nasonia vitripennis, and Tribolium castaneum, [24,25,27]; Nicole Meier and Daniel Bopp, unpublished results). Sequence alignments of the differentially processed exons showed a high degree of similarity to the corresponding Drosophila exons (Fig. S1), so the genomic order of the differentially spliced 39 exons is additionally conserved.
In Drosophila, the genes acting immediately before fru in the sex determination hierarchy direct sex-specific splicing of transcripts derived from the most distally located P1 fru promoter [3,19]. To examine the corresponding upstream sequences of the putative housefly fru ortholog, 59 RACE was performed on cDNA templates prepared from RNA extracted from adult male and female heads. In male preparations, we detected transcripts that contained two additional exons (P1 and S) upstream of the BTB coding exons (C1-C3) extending the ORF by an additional 72 amino acids at the amino terminal end (Figs. 1, S1). In female transcripts we observed that exon S was significantly larger in size due to the utilization of a more downstream located splice donor site (S f ). In addition, we identified a female-specific exon (f) downstream of exon S f (Fig. 1). These additional exonic sequences, which are absent in male transcripts, contain several in-frame stop codons, which prematurely truncate the long ORF as defined by the male transcripts (dots in Fig. 1A). Four repeat elements were identified in these female-specific sequences that perfectly match the cis-regulatory elements (TRA/TRA2 binding sites) known to play an important role in sex-specific splicing of Drosophila fru (Fig. 2). The male-specific transcripts encode full-length products, which show a high degree of protein sequence similarity to those of P1-derived male transcripts in Drosophila and are therefore referred to as Md-fru M (GenBank Accession Number KC750909, KC750911 and KC750913). The female-specific transcripts, on the other hand, have a severely truncated ORF and thus are likely to represent non-functional splice variants of the same locus. Because of their structural correspondence to the non-functional products of P1-derived female transcripts in Drosophila they are collectively named Md-fru F (GenBank Accession Number KC750910, KC750912 and KC750914). Taken together we conclude that the Musca genome harbors an ortholog of fru, Md-fru, which expresses a set of male-specific products and hence is likely to be a target of the sex determining pathway.
The presence of conserved TRA/TRA2 binding sites in Md-fru, suggested that it is sex-specifically spliced and directly targeted like Md-dsx, by Md-tra. To explore this possibility, we tested expression of Md-fru in different genetic backgrounds in which Md-tra activity was experimentally altered (Fig. 2). Previous studies demonstrated that silencing by early dsRNA injections is an effective means to deplete Mdtra activity in XX embryos, resulting in a complete shift of Md-dsx splicing from female to male mode [28]. Likewise, we observed that the splicing mode of Md-fru changed from female to male in Md-tra silenced XX individuals ( Fig. 2A). The same result was obtained when Md-tra2, an essential co-factor of Md-tra, was depleted by RNAi (data not shown). We conclude that female splicing of Md-fru depends on the presence of Md-tra and Md-tra2 activity supporting our notion that Md-fru is a downstream target of Md-tra in the sex determination pathway. To extend this type of epistatic analysis we investigated whether ectopic expression of the female activity of Md-tra is sufficient to direct female-specific splicing of Md-fru in genotypic males. To this end, we used a transgenic line in which Md-tra F is constitutively expressed in flies carrying the male determining factor M III on the third chromosome [28]. These individuals develop into fully fertile females demonstrating that forced Md-tra F expression overrides repression by M III and is sufficient to direct all aspects of female development. RNA samples analyzed from the heads of these sex-reverted M III / +; p[Md-tra F ] individuals showed the presence of only femalespecific splice variants of Md-fru ( Fig. 2A). Hence, Md-tra activity is not only required but also sufficient to impose female splicing mode of Md-fru. The presence of well-conserved TRA/TRA2 binding sites around the critically involved splice sites suggested that Md-fru is a direct target of Md-tra (Fig. 2B,C).

Md-tra man Males Express Female Md-fru Transcripts and are Impaired in Courtship
We previously described the masculinizer (man) allele of Md-tra to be a hypomorphic mutation with residual Md-tra activity [28,33]. Flies homozygous for Md-tra man develop into phenotypically normal looking males (from here on referred to as Md-tra man1 males). However, they are reduced in fertility, as only a fraction (about 10%) produces progeny in single mating tests. Sterility is 100% in homozygous Md-tra man males when their mother carried the dominant gain-of-function allele Md-tra D in trans [28,34,35]. Males with this maternal background were referred to as Mdtra man2 males. The genital apparatus of Md-tra man1 and Md-tra man2 males looked indistinguishable in morphology from those of wildtype males, and normal amounts of motile sperm were present in squashed preparations of mutant testes. To test for gonad and sperm functionality, the larval anlagen of Md-tra man2 testes were transplanted into wildtype male hosts (Fig. S2). As these mosaic animals were capable of siring a large number of donor-derived offspring, we concluded that sperm maturation was normal in Mdtra man2 mutant testes.
Mutations affecting neuronal development can disturb behavioral outputs such as male courtship and, thus, can lead to sterility. Therefore, we decided to test courtship performance of Md-tra man1 and Md-tra man2 males with wild-type females. Meffert [36] described courtship in the housefly as a very brief and stereotypical ritual (see Fig. 3A). Behavioral analysis of Md-tra man mutant males showed that their overall levels of courtship were substantially lower than those of wild-type XY males (30% of Md-tra man1 and 5% of Md-tra man2 males compared to 80% in wild-type males) ( Fig 3B). In single courtship assays, only half of the Md-tra man1 males attempted copulation, while no copulation attempts were observed when Md-tra man2 males were tested. In contrast, 80% of courting wild-type males attempted copulation and almost 50% successfully copulated during the 20 minutes observation period (Fig. 3B). Of the mutant males which did engage in courting, courtship latency was comparable to that of wild-type males, but copulation duration was on average three times longer than wildtype males (Table S1). To assess whether specific parts of the mating ritual were affected, we performed a second set of courtship assays using the courting scheme described by Meffert and Hagenbuch [37]. Md-tra man1 and Md-tra man2 males which engaged in courting performed two of the last three steps with substantially reduced frequencies, in contrast to control wild-type males which performed each step of the sequence at comparable rates ( Fig. 3C and Table S2). In particular the latter steps involving wing vibrations and backward leaning of the males on top of the females are severely affected in Md-tra man males. Premature termination of the courting ritual was more often observed in Md-tra man2 males than in Md-tra man1 males of which 19% performed the last steps and even attempted copulation. These data suggest that the infertility observed in homozygous Md-tra man males was due to failure to properly complete the sequence of courting steps prior to copulation.
We hypothesized that residual feminizing activity of the Mdtra man allele caused male courtship failure. We tested this hypothesis by injecting dsRNA of Md-tra into syncytial Md-tra man2 embryos. Silencing of Md-tra at this early stage leads to a complete and lasting loss of Md-tra activity [28]. Of 52 injected Md-tra man2 males, 41 were fertile in single mating tests and, due to the absence of an M factor, produced only female offspring when outcrossed with wild-type females. The percentage of successful matings (about 80%) was comparable to those of wild-type control males in single crosses. In contrast, all 20 Md-tra man2 males injected with Ringer's buffer remained infertile. Hence, the sterility of Md-tra man2 homozygous males can be rescued by targeted silencing of Md-tra. In a previous study we showed that silencing of Md-tra2, a required co-regulator for female expression of Md-tra, could also completely restore fertility of Md-tra man2 mutant males [28,31]. These results suggested that the residual feminizing activity of Md-tra man is sufficient to cause male courtship failure.
This result prompted us to test whether this phenotype may be caused by misregulation of the newly identified Md-tra target Mdfru. The sex-specific splicing of Md-fru was investigated in Mdtra man2 males. RT-PCR experiments on RNA extracted from Mdtra man2 male heads detected only non-functional Md-fru F transcripts ( Fig. 4A), suggesting residual Md-tra F activity in Md-tra man males is sufficient to impose female-specific splicing of Md-fru. In contrast, the previously described target of Md-tra, Md-dsx, seems to be unaffected as only male products were detected in Md-tra man2 males (Fig. 4B). It seems likely that, in Md-tra man2 males, lack of functional Md-fru M products is the primary cause of impaired courtship performance. If normal male courtship indeed depends on presence of Md-fru M we expected that male splicing of Md-fru must be restored in rescued Md-tra man2 males. Consistent with this interpretation, we detected substantial levels of Md-fru M transcripts in rescued males (Fig. 4A) indicating a shift in the splicing mode from female back to male. We conclude that normal male behavior correlates with presence of functional Md-fru M transcripts suggesting that this gene plays a crucial role in the correct implementation of male behavior in the housefly.

Md-fru is Expressed in the CNS and Peripheral Sensory Neurons
In Drosophila sex-specific fru expression is confined to neuronal tissues [12,20,21,22,23]. To investigate the spatial expression pattern of Md-fru in houseflies we performed in situ hybridization assays. Whole mount preparations of third instar larval tissues were initially investigated. We detected ubiquitous expression of Md-fru P1 transcripts in both larval brain hemispheres and in the ventral ganglion. Staining was neither observed in imaginal discs nor in other non-neural tissues (data not shown). If Md-fru were indeed part of a developmental program involved in establishing the neuronal circuitry involved in courtship behavior, we would  . Impaired courtship performance of Md-tra man males. (A) Sketch of the housefly courtship ritual as described by [36]: (1) The male fly (blue) approaches the female (pink) and performs an abortive form of walking and stalking towards the female. The male fly will interdigitate his legs with those of the female while making tapping movements. The female may respond with similar fencing. (2) Next the male will mount the female and flutter his wings. (3) While he is lunging forward to the head of the female he will continue to flutter. He will then lift the female's forelegs. At this point, he holds his wings in a position that is roughly 180u from their resting orientations. (4) When mounted, the female thrusts her wings 90u to her body and perpendicular to the substrate. (5) Finally, the male will lean backward and attempt to copulate. The whole ritual lasts around 2 seconds and will be repeated by the male when the first attempt fails. (B) Courtship activity of standard wild-type XY males, Md-tra man1 males, and Md-tra man2 males in single mating assays with wild-type females; percentage of males which courted at least once during a 20 minutes observation period (total number of males per genotype observed n = 20). A marked reduction of courting activity is observed for Md-tra man1 males and almost no activity is seen for Md-tra man2 males. (C) Percentage of males performing individual steps of the courtship ritual (total number of courting males per genotype observed n = 19). While the percentage of wildtype XY males initiating courtship remains equally high throughout the various steps of the ritual, the percentage of Md-tra man males engaged in courtship declines precipitously during the last steps, in particular step 3 (fluttering) and step 5 (copulation attempt). doi:10.1371/journal.pone.0062476.g003 Md-tra man2 males. Primer sequences are published in [28]. In contrast to Md-fru, Md-dsx is almost exclusively spliced in the male mode in Md-tra man2 males. doi:10.1371/journal.pone.0062476.g004 expect a locally restricted expression pattern in the adult CNS similar to that reported in Drosophila. RNA in situ hybridizations were performed on cryosections of adult female and male housefly heads with an antisense probe against 59 sequences (P1-derived) of Md-fru (Fig. 5). Strong ubiquitous staining was detected in areas which process visual information, namely in and around the four optic neuropils lamina, medulla, lobula and lobula plate (Fig. 5A-C). Prominent fru expression has also been detected in the visual systems of D. suzukii and D. virilis, in particularly in the lamina, but is completely absent in D. melanogaster [38]. Overall, P1-specific transcripts of Md-fru appeared to be more broadly expressed in the mid brain than the corresponding P1 transcripts in D. melanogaster (Ryner et al., 1996). The Md-fru probe also labeled neurons located in the base of taste sensilla of the labellum (Fig. 5D-F). We propose these neurons belong to the gustatory system, as corresponding gustatory receptor neurons in Drosophila are known to express FRU M [22].
Within the sexually dimorphic neural circuitry identified in the CNS of Drosophila melanogaster is a cluster of 10 serotonergic neurons located in the posterior dorsal part of the abdominal ganglion which are present in males but not in females. These SAbg neurons send their projections to the main trunk that innervates the male internal reproductive organs such as the testicular ducts, seminal vesicles, accessory glands, and ejaculatory duct [39]. In fru mutant males these SAbg clusters are substantially diminished or completely absent suggesting that FRU M is responsible for the formation and/or neurotransmitter specification of these neurons [12]. This well-documented example of fru function in the male CNS led us to investigate whether a similar neuroanatomical dimorphism exists in the housefly. We performed anti-5HT (serotonin) antibody staining on whole mount preparations of adult male and female CNS ventral ganglions (Fig. 6). As in Drosophila, we identified a small cluster of serotonergic neurons which are present at the posterior end of the abdominal ganglion in male flies, but absent in the corresponding region of the female ganglion ( Fig. 6B-C). Also, projections of these serotonergic neurons are sent to the main trunk and innervate male internal reproductive organs ( Figure 6E,F). Significantly, this cluster of serotonergic neurons and its projections to the male internal reproductive organs were absent in abdominal ganglions dissected from behaviourally sterile Md-tra man2 mutant males (Fig. 6D,G). There was a complete lack of serotonin immunoreactivity in neurons innervating the male reproductive organs in 75% of Mdtra man2 mutants (n = 12), compared with 100% immunoreactivity in wild-type controls and in Md-tra man1 (n = 10 each). The loss of serotonin staining of this cluster correlates with the lack of Md-fru M expression suggesting that, just as Drosophila, Md-FRU M plays a key role in the specification of these male serotonergic cells [13].

Male Mating Behavior in Houseflies is Controlled by Mdtra
We previously reported that the gene Md-tra acts as the key switch in the sex determining hierarchy of the house fly governing every aspect of somatic dimorphic development [28]. In this study we show that courtship of male houseflies is an innate behavior that is also under the control of Md-tra. Our analysis of Mdtra man males revealed that infertility is primarily a result of anomalous behavior rather than of aberrant differentiation of reproductive tissues. We propose that the primary cause for anomalous courtship behavior in mutant males is the presence of female Md-tra activity. When Md-tra is knock downed by RNAi, fertility of these males is restored suggesting the absence of Md-tra activity is a pre-requisite for proper implementation of male courtship behavior. In females, on the other hand, Md-tra is activated in order to prevent the establishment of these behaviors. We propose that Md-tra acts as an upstream switch, coordinating dimorphic differentiation of both morphological and behavioral traits. The central role of tra in coordinating these two facets of dimorphic development has been extensively investigated in Drosophila. For instance, genotypically female XX animals which are mutant for tra develop into morphologically normal males which behave and court like males [40]. Reciprocally, feminization of XY individuals by activating tra results in the development of pseudofemales, which do not display male behaviors. Morphological and behavioral dimorphic development, however, differ with regard to the spatial and temporal requirements of tra. While tra's functions in somatic differentiation are needed as early as during embryonic development (e.g., gonad differentiation), the critical developmental window for the control of sex-specific behaviors was mapped to much later stages of development around larvalpupal transition [41]. When tra is activated at these stages in genotypically male individuals, they will develop into normal looking males however courtship behavior is severely impaired. It suffices to target expression of tra to specific areas of the brain to accomplish a complete loss of courtship behavior in otherwise normally developed males [42]. This phenotype is reminiscent of the situation in Md-tra man males suggesting that residual Md-tra activity in these males is primarily affecting late events in dimorphic development in particular those during the critical stages when male courtship behavior is established. As feminization is only observed at the behavior level this suggests that the female activity provided by the Md-tra man alleles is either below a threshold needed to differentiate female structures or is confined to neural tissues where it specifically affects CNS related aspects of dimorphic development.

Md-fru is a Target of the Sex-determining Pathway Downstream of Md-tra
In this study we have identified the house fly homologue of the key Drosophila sex-determination gene fru, Md-fru. We have shown that, in an analogous manner to Drosophila, Md-fru RNAs are alternatively spliced leading to male-specific expression of Md-FRU M proteins from the P1 promoter. The presence of four putative TRA/TRA2 bindings sites located close to the regulated splice sites of the female-specific exons of Md-fru suggests that this regulation is due to its targeting by the female-specific splicing factor Md-tra. This is similar to the regulation of the previously proposed targets of Md-tra, Md-dsx and Md-tra itself [28,30]. We show that changes in Md-tra activity lead to corresponding changes in Md-fru expression. For instance, ablating Md-tra function by RNAi in karyotypically female individuals shifts the mode of Mdfru splicing from female to male. Reciprocally, when Md-tra functions are ectopically expressed in karyotypic males, Md-fru expression switches to the female nonfunctional mode. We additionally show that Md-tra2, a functional splicing co-factor with tra in Drosophila, is involved in Md-fru splicing regulation, by silencing Md-tra2 in Md-tra man2 males and showing a shift of Md-fru splicing from the female to the male mode (data not shown) [31].
Taken together we propose that, like Md-dsx and Md-tra, Md-fru is another downstream and possibly direct target of Md-tra in the house fly sex determination pathway. In D. melanogaster, malespecific FRU M products are required for the proper display of male courtship behavior. When tra is ectopically activated in the developing male CNS, it prevents the production of FRU M and, as a consequence, these males exhibit impaired and anomalous courtship [3,43,44]. Along the same lines, we propose that Md-fru is a target of Md-tra and that misregulation of Md-fru by residual Md-tra activity is the primary cause for courtship failure of Mdtra man males. Interestingly, the sex-specific splicing of Md-fru, and not Md-dsx, seems to be affected by the hypomorphic activity produced by the Md-tra man alleles. In Md-tra man mutant males the splicing of Md-fru shifts from the male to female mode, resulting in the loss of ability to produce male-specific FRU proteins.
The disparity in splicing behavior of Md-dsx and Md-fru in the hypomorphic Md-tra man background can be explained by either different thresholds requirements or by differences in temporal and spatial requirements for activating the female splicing mode. For instance, Md-fru pre-mRNAs may have a higher affinity for Md-TRA binding than Md-dsx pre-mRNAs or may require lower levels of bound complexes to activate the female-specific splice sites. On the other hand, the differential effects on Md-dsx and Md-fru may result from discontinuous and/or tissue-specific Md-tra man expression during development. In either case, its prominent impact on behavioral dimorphic development suggests that Md-tra man must be expressed in the nervous system around the time when the CNS shapes the neuronal circuits required for proper male behavior [41]. Consistent with this interpretation, we detected female fru transcripts (Md-fru F ) in the heads of male Md-tra man adults. We propose that Md-fru is a direct target of Md-tra thereby defining an Md-dsx independent branch in the sex-determining pathway, which specifies the fates or activities of neurons that carry out higher order control functions to elicit and coordinate the activities comprising male reproductive behavior.
Is Md-fru Functionally Conserved?
We propose that male-specific Md-fru M expression is required for the correct display of male courtship in the housefly. This conclusion is based on the strict correlation between proper courtship performance and male expression of Md-fru. In our courtship assays we found that only males that expressed the Mdfru M variants performed as well as normal males, while those that expressed non-functional Md-fru F variants either performed poorly or did not engage in courtship at all. The most compelling argument comes from the observation that silencing of Md-tra in Md-tra man2 males not only shifts Md-fru expression from the female to male mode but also completely restores fertility. A more direct test to confirm the role of Md-fru in courtship behavior would require a targeted disruption of Md-fru M function (cf. [8]). We expect that silencing of Md-fru M causes behavioral phenotypes similar to those observed in Md-tra man males. Unfortunately, several attempts to specifically disrupt Md-fru M functions have thus far failed. Injections of dsRNA against P1 derived sequences of Md-fru into early syncytial blastoderm embryos failed to downregulate Md-fru expression in larval and pupal stages (data not shown). To overcome this problem we introduced a construct into the housefly genome which contains inverted repeats against P1 derived Md-fru transcripts driven by a constitutively expressed promoter. Unfortunately, none of the tested transgenic lines effectively suppressed Md-fru (data not shown). On the other hand, downregulation of the respective fru homolog in the hemimetabolous species, Blatetella germanica and Schistocerca gregaria, both of which are responsive to systemic effects of dsRNA silencing, have been shown to disrupt proper male behavior suggesting that fru plays an important and widely conserved role in male behavior [45,46,47]. However, it is not known whether the fru homologs in these species are direct targets of the sex-determining cascade and whether of not these homologs are specifically expressed in the CNS.
The functions of fru in male courtship have been studied in great detail at the cellular level in Drosophila. FRU M is expressed in about 1700 neurons which appear to be involved in various aspects of male courtship including olfactory and gustatory neurons required for behavioral responses to sex pheromones. Likewise, we observe that expression of Md-fru is confined to specific areas of the CNS and to neurons of the peripheral nervous system. Though expression appears much broader than in Drosophila and apparent in the optic neuropils, its neural-specific expression suggests that Md-fru is also involved in building and specifying neuronal circuits required for male behavior. High levels of expression in the neuropils which process visual cues hint at a prevalent role of the visual system in male behavior of the housefly. Previous studies showed that visually guided behavior patterns in houseflies differ between the sexes. Male houseflies chase females in the air using visual cues, but are not chased by females [48], coinciding with this sex-specific behavior are the presence of male-specific neurons in the visual system [49,50]. It is thus possible that expression of Mdfru in the optic neuropils is involved in establishing the sexual difference of neural architecture involved in motion detection.
One striking dimorphic neural trait common to houseflies and fruit flies is the presence of a male-specific cluster of serotonergic neurons in the dorsal posterior part of the abdominal ganglion. In both types of dipterans these neurons send projections to the male internal reproductive organs where they innervate contractile muscles. It has been proposed that these muscles are required for transferring sperm and seminal fluids during copulation [51,52]. In semi-fertile fru mutant males this cluster is substantially reduced in size and projection patterns are aberrant [51]. A similar phenotype was found in Md-tra man2 housefly males where the cluster is substantially diminished in size and projections to the internal reproductive organs are lost. In Drosophila the formation and/or specification of the male-specific serotonergic cluster depends on presence of FRU M and DSX M activity [12,21]. It is thus conceivable that lack of Md-FRU M is the cause for their absence in Md-tra man2 males, suggesting a conserved mechanism in the specification of these male-specific neurons.
Our study in the housefly suggests that merging the control of the two terminal pathway effectors, fru and dsx, at the level of tra is a widely used strategy in holometabolous insects (Fig. 7). We propose that this linked control ensures that sex-determining instructions are coordinately relayed and implemented in neural and non-neural tissues.

Rearing of the Flies
Rearing of Musca domestica larvae and flies has been described previously [33]. Since low density of larvae on standard medium can cause substantial decrease in survival rates, we reared transplanted host larvae and larvae from injected embryos on porcine manure. To dispose of mites and other parasites and to avoid contamination with eggs or larvae from wild-type populations, manure was stored at 270uC for at least two weeks prior to use.

Transplantation of Larval Gonads
Crossing schemes and used genotypes are shown in Fig. S2. Larval anlagen of Mdtra man2 , bwb + testes were dissected and transplanted into the posterior region of bwb male host larvae in Ringer's solution. These larvae were then reared on porcine manure and surviving adults were outcrossed with bwb females of the host strain. In the progeny we screened for presence of donor derived bwb + individuals.

Sperm Analysis
Testes of 5-days old adult males were dissected in Ringer's solution and transferred to a coverslip where they were gently squeezed to release sperm which was then dispersed with a needle. Morphology and mobility of sperm were examined by phasecontrast microscopy.

Courtship Assays
Male and female flies collected for courtship assays were separated shortly after hatching and kept at 25uC for 6 days in plastic beakers with sugar water and milk powder. The air humidity was kept between 65% and 75% and flies were exposed to a 12 hours light/dark cycle. The courtship assays were performed at the same time of the day. The courtship arena consisted of a see-through plastic cylinder with a height of 40 mm and a length of 160 mm. For each couple courtship activities were examined during a period of at least 20 minutes.

Md-tra dsRNA Injections
The same primers were used for synthesis of Md-tra dsRNA as described in [28]: Md-tra forward primer located in the 59UTR of exon 1 linked to T7 promoter sequences (small letters): (59 gtaatacgactcactatagggTGGTGTAATATGGCTCTATCG 39) and Md-tra reverse primer located in exon 5 linked to T7 promoter sequences (small letters): (59 gtaatacgactcacta-tagggGCTGCCATACAAACGTGTC 39).
Total RNA from adult females was reverse transcribed and the resulting cDNA was amplified with this primer pair to generate a female-specific template of 620 bp in length. In vitro transcription of this template with T7 RNA polymerase yielded dsRNA that was ethanol precipitated and resuspended in injection buffer to a final concentration of 1 mg/ml. Embryos were collected within 1 h after egg laying and dechorionated. Then 1-2 nl of the dsRNA solution was injected into the posterior pole of syncytial blastoderm embryos [28]. As a negative control, embryos of the same stage were injected with a similar volume of Ringer's solution. Injected embryos were allowed to develop at room temperature and placed in porcine manure for further development. Hatching flies were mated to virgins of the multimarked strain.

Isolation of Md-fru
A small Md-fru fragment spanning the conserved BTB domain was previously recovered by David Dolezel (University of Ceske Budejovice, Czech Republic) using degenerated primers and made available to us for further investigation. Based on sequence alignments between fru sequences of Drosophila and Anopheles gambiae [24] we generated degenerated primers for the different zinc finger-domains: domain These reverse primers were used in combination with the forward primer MdFRU-3E (59CAGTTAAGGCTCAT-CAAGCC39) located in the BTB domain to amplify cDNA variants containing zincfinger domains B and C. Forward primer MdFRU-15s (59CATGGGCCTCAATCAGTCG39) was used to amplify the variant which contains zinc finger domain A. The zinc finger domain D was identified by bioinformatics in the genomic region between exon C4 and exon C5.

and 39 RACE of Md-fru cDNA
Female and male heads were collected from freshly hatched adults and total RNA was prepared according to the protocol of the manufacturer TRI REAGENT (SIGMA-ALDRICH, St Louis, USA) or to the protocol of the ''NucleoSpin RNAII'' kit (Macherey-Nagel, Düren, Germany). For long term storage, total RNA was dissolved in 70% ethanol and stored at 280uC. 39

RT-PCR Analysis of Sex-specific Md-fru Splice Variants
The following primers were used to monitor amplify the sexspecific splice variants of Md-fru transcripts. Forward primers: Mdfru-18 (59 TCCATCATGAGCGAACTCTGT 39) is located in the female-specific sequences of the P1 transcripts, Md-fru-27 (59 GTGAAGTGAACTTAGTACATGG 39) is located in the common sequences of the P1 transcripts; reverse primer: Mdfru-1 (59 TCTGAGTATCTGACATCCTTC 39), located in the C1 exon. As an internal control the cytochromatic gene CYP6D3 was amplified with the following primers: CYP6D3-1 (59 Figure 7. Bifurcation of the sex-determination pathway: a general principle in holometabolous insects? The phylogenetic relationship and the sex-determining pathways of several dipteran species and the hymenopteran species Nasonia are depicted in this scheme. Despite considerable differences regarding the nature of the instructive sex-determining signal (X-counting in Drosophila, male [M] dominant determiners in Musca and Ceratitis, haplo-maternal [mat] versus diplo-maternal-paternal [pat+mat] in Nasonia) many insects use tra as the binary ON/OFF switch downstream of the signal [54], though the molecular basis of this switch in A. gambiae remains still elusive [24,55,56]. The dsx gene, expressing sexspecific variants DSX F and DSX M , appears to be a conserved target of the binary switch and its use as a bifunctional executor of the instructive signal was validated by gene function studies in the depicted insect species [26,30,56,57,58] Also, homologs of fru were identified in these species that express sex-specific variants FRU F and FRU M [24,26,27,59]. For Musca we here present evidence that fru is a downstream target of the tra branch which is involved in controlling proper display of courtship. Though no functional fru data from other holometabolous insects are available as yet, we propose that fru independently from dsx represents another conserved effector of the instructive signal. Hence, bifurcation of the pathway downstream of tra may have already existed before the divergence of holometabolous insect orders. doi:10.1371/journal.pone.0062476.g007 GTTCGGTAATATTTGGCTTGG 39) and CYP6D3-2 (59 CCCGTATTCCGTAGTTGAATT 39). Primer pair Md-fru-18 and Md-fru-1 yields a fragment of 289 bp in females, and primer pair Md-fru-27 and Md-fru-1 generates a 602 bp long PCR product in males.
In situ Hybridisation of Md-fruP1 Transcripts on Adult Head Sections Probes were derived from templates generated by PCR with primer Md-fru-27 (forward; 59 GTGAAGTGAACTTAGTA-CATGG 39) and Md-fru-29 (reverse; 59 CTGTCAAGGGTTCTCGGCTT 39): A T7 promoter sequence (GTAATACGACTCACTATAGGG) was added to the Md-fru-27 primer for preparing sense probes (control) and to the Md-fru-29 primer for preparing antisense probes. Both amplicons were transcribed with T7 polymerase to generate labelled single stranded RNA probes (Roche, Madison, USA). Heads were dissected from adult flies, mounted in O.C.T medium and frozen at 220uC prior cryosectioning with a Leica rotary microtome. In situ hybridizations were carried out on adult head sections according to the protocol of [53].

Serotonin Staining of Abdominal Ganglions
Abdominal ganglions were dissected in PBS solution and fixed in 4% paraformaldehyde in PBS for 1 hour on ice while gently shaking. Then the tissues were washed three times in PBS for 15 minutes each. Following a brief 5 minutes incubation in 1 N HCl, the tissues were washed in PBS for 15 minutes and three times in 0.4% PBT for 10 minutes each. Afterwards the tissue samples were incubated for 1 hour in PTN (0.4% PBT, 5% (v/v) normal goat serum), followed by the incubation with the primary antibody for 24-36 hours at 4uC. The rabbit anti-serotonin (5-HT) antibody was purchased from SIGMA-ALDRICH (cat # S5545). Dilution of the antibody was 1/400 in PTN. After the incubation of the primary antibody the ganglions were washed four times in PBT for 30 minutes each, followed by the incubation of the secondary antibody at 4uC overnight or at room temperature for 3 to 4 hours. Dilution of the secondary antibody (anti-rabbit FITC, SIGMA-ALDRICH, anti-rabbit AlexaFluor 488, Molecular Probes Eugene, USA) was 1/600 in PTN. To remove unbound secondary antibody the samples were washed four times in PBT for 1 hour each. Thereafter, the samples were washed several times in PBS for several hours each (1-2 days in total). Finally, the samples were mounted in VectaShield medium (Vector Labs, Burlingame, USA) on glass slides and covered with cover slips. The mounted samples were examined and analysed by confocal laser scanning microscopy (Zeiss LSM 510). The genotypes of all animals were coded before dissections to avoid bias in either preparation or scoring of the samples.  Figure S2 Larval testes transplantations. Male Md-tra man2 progeny (donor) was generated by crossing bwb + ; Md-tra man males with bwb + ; Md-tra man /Md-tra D females. Gonads were dissected from these male larvae and transplanted into male bwb, M III ; Md-tra + larvae (host). These larvae were reared to adulthood and outcrossed with females of the host strain. Three different phenotypes of progeny were expected, bwb males and females (light brown) from host derived sperm and bwb + females (dark brown) from donor derived sperm. We recovered 7 successfully transplanted males which sired in total 728 individuals with a host genotype (bwb) and 154 individuals with a donor genotype (bwb + ). (TIF)