Control of sexual differentiation and behavior by the doublesex gene in Drosophila melanogaster.

Doublesex proteins, which are part of the structurally and functionally conserved Dmrt gene family, are important for sex determination throughout the animal kingdom. We inserted Gal4 into the doublesex (dsx) locus of Drosophila melanogaster, allowing us to visualize and manipulate cells expressing dsx in various tissues. In the nervous system, we detected differences between the sexes in dsx-positive neuronal numbers, axonal projections and synaptic density. We found that dsx was required for the development of male-specific neurons that coexpressed fruitless (fru), a regulator of male sexual behavior. We propose that dsx and fru act together to form the neuronal framework necessary for male sexual behavior. We found that disrupting dsx neuronal function had profound effects on male sexual behavior. Furthermore, our results suggest that dsx-positive neurons are involved in pre- to post-copulatory female reproductive behaviors.

Significant insights into the genetic and developmental logic supporting sex-specific behaviors have come from the genetically and behaviorally tractable model organism, Drosophila melanogaster. In Drosophila, genes of the sex-determination hierarchy orchestrate the development and differentiation of sex-specific tissues, establishing sex-specific physiology and neural circuitry2. Female-specific expression of Transformer (Tra), with the non-sex-specific Transformer-2, introduces a sex-specific splice in two pivotal downstream transcription factors, doublesex (dsx) and fruitless (fru). In females, dsx transcripts are spliced to give rise to a female-specific isoform: DsxF. In contrast, sex-specific splicing introduces a stop codon into female-specific fru mRNAs, which are then not translated. In males, in the absence of Tra, dsx and fru transcripts undergo default splicing, generating the male-specific isoforms: DsxM and FruM. Together, these key transcriptional regulators establish most aspects of ‘maleness’ and ‘femaleness’.

Dsx proteins are part of the Dmrt (doublesex and mab-3-related transcription factor) family, a structurally and functionally conserved group of zinc-finger transcription factors with important roles in sex determination throughout the animal kingdom3. In Drosophila dsx directs most aspects of somatic sexual differentiation outside the central nervous system (CNS) in both sexes. Many of these differences are implicated in the performance or success of male sexual and reproductive behaviors4-9.

Many recent studies have focused on the control of sexual behavior in males by fru10-12. FruM proteins are expressed from metamorphosis in the male CNS13. In the absence of FruM proteins, males perform little to no courtship towards females, fail to produce the pulse song component of generate aberrant courtship song, never attempt copulation, and exhibit increased inter-male courtship2. Females expressing FruM display many male-specific courtship behaviors; however, they court less than wild-type males, generate aberrant song and never attempt copulation11, 14. Thus FruM expression is not sufficient to specify normal male courtship behavior, suggesting other genes are required for a complete male courtship repertoire.

One obvious candidate is dsx. Males lacking dsx court at diminished levels and generate aberrant fail to generate the sine song component of courtship song15. Dsx is found in the CNS14, 16, 17, and DsxM and FruM proteins are co-expressed in several regions important in establishing male sexual behavior14, 18. Recent studies suggest that dsx and fru co-operate to dictate development of specific neural substrates underlying male sexual behavior14, 18-20, thus both genes, are necessary in the specification of neural systems that generate male patterns of behavior21. However, little is known about the neuroanatomical expression of dsx in the CNS and how this influences male and female sexual behavior.

To explore the anatomy and function of dsx-expressing neurons in both sexes, we targeted the insertion of the yeast transcription factor GAL4 to the dsx locus. Clear differences in neuronal numbers and axonal projections were detected in male and female CNSs. These differences are behaviorally relevant, since disrupting dsx neuronal function leads to highly aberrant behaviors in both sexes, suggesting these neurons instruct sex-specific neural programs. We propose that dsx co-ordinates the development of external sexual morphology and physiology with the development of sex-specific neurons that contribute to the circuitry required for male and female sexual behaviors.

Results

Targeting GAL4 to the doublesex locus

To allow detailed anatomical and functional analyses of dsx-expressing cells we used ends-in homologous recombination to insert the GAL4 coding sequence into the first, non-sex-specific, coding exon of dsx, creating a tandem duplication at the locus (Fig.1a and Supplementary Fig. 1 online). The resulting allele, dsx, produces wild-type dsx transcripts, as well as GAL4-containing transcripts in both sexes (Fig. 1b and Supplementary Fig. 1). When homozygous, or examined in trans to a deficiency of dsx (Df(3R)dsx), external sexual morphology (data not shown) and fertility of dsx flies were unaffected (p>0.05, n>45). We conclude the dsx allele exhibits no overt disruption of gene function.

Figure 1

Expression of GAL4 from the dsx locus recapitulates endogenous dsx expression. (a) Schematic of dsx (dsx+) and GAL4 knock-in allele (dsx). Transcriptional start sites, arrows; black boxes, exons. (b) Male and female dsx locus predicted transcripts. (c,d) dsx 2-day-old male pupa. anti-Dsx (green), anti-ßGal (magenta). (c) Brain and (d) VNC. Ventral views; anterior up. Scale bars = 50 μm. (e-l) dsx transformation of secondary sexual characteristics; dorsal abdominal cuticular pigmentation; external genitalia; T1 leg basitarsal detail. (e) XY and (f) XX wild-type. (g) XX dsx-null intersexual. (h) XX tra-null pseudo-male. (i) XY UAS-dsx;dsx pseudo-female. (j) XX dsx/UAS-dsx pseudo-male. (k) XX UAS-tra. (l) XX dsx/UAS-tra2.

To determine whether dsx reiterates endogenous dsx expression, we used dsxGAL4 to drive expression of a nuclear reporter demonstrating co-localization with all previously described Dsx-expressing neuronal clusters16 (Fig. 1c,d). Thus dsx is a sensitive, specific marker for Dsx-expressing cells.

dsxGAL4 cells dictate sexual differentiation

dsx is required to establish sexually dimorphic morphology and characteristics in male and female flies22, but it is unclear where Dsx is expressed, and whether dsx cells dictate all aspects of external sexual morphology. We determined the temporal and spatial distribution of dsx cells (Supplementary Fig. 2 online). dsx expression was observed in a number of tissues with sexually dimorphic features or physiology, such as the foreleg, cuticle, oenocytes, fat body, and reproductive organs. dsx-expression however is not ubiquitous, suggesting that sex determination need not occur in every cell. Are these dsx cells capable of controlling sexual differentiation?

dsx’s ability to determine secondary sexual characteristics has been established through mutant analyses, or ubiquitous over-expression of Dsx isoforms22. To determine if dsx-expressing cells are capable of directing a sex-specific program of development, we restrictively manipulated the sex of dsx-expressing cells, assaying the consequences of expressing either DsxM or DsxF to external sexual morphology. DsxF over-expression resulted in single-X males with female-like abdominal pigmentation, genitalia, and no sex combs. (Fig. 1i); thus DsxF is sufficient to direct a female-specific program of development even when competing with endogenous DsxM production. DsxM over-expression resulted in masculinized XX females that displayed male-like abdominal pigmentation and sex combs; however the genitalia, while masculinized, were rotated and often malformed (Fig. 1j), possibly due to an inability to overcome the effects of endogenous DsxF.

XX flies homozygous for loss-of-function tra or tra-2 mutations develop as pseudo-males; they are essentially indistinguishable from males in morphology and behavior, though sterile23-25 (Fig. 1h). Using RNA interference we specifically knocked-down endogenous Tra and Tra-2 in dsx cells. Strikingly, XX flies developed as close phenocopies (Fig. 1k,l) of tra and tra-2 loss-of-function variants; they exhibited male-like abdominal pigmentation, genitalia, and sex combs. dsx cells are therefore capable of directing sex-specific morphological development.

dsx-specified sexually dimorphic neural circuitry

While dsx’s specification of external sexual morphology influences sexual behavior5, 26, the anatomical foci central to determination of sex-specific behaviors lie in the CNS2. Therefore, we used dsx to describe dimorphic dsx expression in the developing CNS (Fig. 2; Supplementary Fig. 3, and Supplementary Table 1 online).

Figure 2

Sexually dimorphic expression of dsx-neurons and associated projections and colocalization with FruM in 3-5 day adult CNSs. (a) Male brain, dsx neuronal clusters (arrowheads), male-specific SN neurons (only one cell in plane of focus; arrow). The cell bodies of pC1 ,pC2, and pC3 in dorsal inferomedial, inferolateral, and superomedial protocerebral areas respectively. (b,c) Male brain, SN cells position, boxed. (d-f) Female brain; (d) dsx neuronal clusters (arrowheads). (g) Male VNC, Abg cluster (arrowhead), male-specific TN1 and TN2 neurons (arrow). (h) Female VNC, Abg cluster (arrowhead). (i,j) Male VNC. (k) Female VNC, hindleg contralateral projection (arrow). (l) Female VNC. (a,d,g,h) Neuronal cell bodies expressing UAS-pStingerII (nGFP). anti-GFP (green). (b,e,i,k) Neuronal projections expressing UAS-mCD8::GFP (membrane-bound GFP), anti-mCD8 (green). (c,f,j,l) Synaptic expression; expressing UAS-synaptotagmin (pre-synaptic marker) tagged with HA, anti-HA (green). Neuropil counterstained with anti-nC82 (magenta). Ventral views; anterior top. (m) UAS-RedStinger;dsx male and (n) female brain. Dsx neuronal clusters (arrowheads). anti-RFP(magenta); anti-GFP (green). Horizontal view, ventral top. Scale bars = 50 μm.

In addition to previously described dsx-pC1, -pC2 and -aDN neuronal clusters in the adult posterior brain16, we identified a novel dsx-expressing cluster, designated dsx-pC3 (Fig. 2a,d,m,n​; Supplementary Table 1). The pC1, pC2 and pC3 clusters have higher numbers of neurons in males (Fig. 2a vs. Fig. 2d​; Supplementary Table 1). These neurons lie in the dorsal inferomedial, inferolateral and superomedial protocerebrum (respectively) surrounding the mushroom body calyces, sites implicated in sex-specific behaviors in both sexes2 (Fig. 2m,n; Supplementary Fig. 4 online). In the suboesophageal ganglion (SOG), two male-specific neurons, dsx-SN, are present (Fig. 2a-c).

Despite the striking dimorphism in neuronal numbers in the brain, the topology of dsx-neuronal projections was remarkably similar between male and female brains. The most overt difference is the dramatic increase in density of synapses and projections in males compared to females as determined by visualization of pre-synaptic and membrane markers in dsxGAL4 neurons (Fig. 2b,c,m vs. Fig. 2e,f,n; Supplementary video 1 online).

In both sexes these bilateral dorsal clusters are extensively interconnected (ipsilaterally, with projections extending throughout the medial and lateral protocerebrum; and contralaterally, via an extensive commissural bridge in the superior protocerebrum, to clusters in the opposing hemisegment)20. These connections also extend to the SOG, the point of termination for projections originating directly and indirectly from tarsal gustatory neurons, implicated in mate choice via processing of non-volatile pheromonal cues27.

In the ventral nerve cord (VNC), adult males exhibit expression in the described male-specific dsx-TN1 neuronal clusters and -TN2 cells (Fig. 2g)16. The TN1 neuronal clusters appear to communicate directly with each other, with locally associated TN2 cells, and with regions in the brain responsible for higher order processing of sensory cues via projections through the cervical connective (Fig. 2i,j and 3f; Supplementary video 2 online). It is noteworthy that the ventral Msg, a focus for male-specific unilateral wing extension and courtship song, is innervated by male-specific TN1 and TN2 neurons14, 28, 29. Thus dsx reveals significant dimorphism between males and females in neuronal number, axonal projections and synaptic density in behaviorally relevant regions of the brain.

Figure 3

Sex apecific dsx expression in the foreleg and effect of basitarsal amputations on axonal projections. (a-d) Sexually dimorphic expression in T1 foreleg. (a) Male T1 leg (medial aspect), sex comb (boxed). dsxGAL4 expressing cells in metatarsus = 96 ± 14.4 (n=7) and tarsi 2-5 = 72 ± 10.0 (n=7) (b) Medial and (c) lateral aspect male T1 tarsi and metatarsus, sex comb (boxed). (d) Female T1 tarsi and metatarsus (lateral aspect), area consistent with sex comb (boxed). dsx expressing cells in metatarsus = 77 ± 12.4 (n=8) and tarsi 2-5 = 58 ± 9.7 (n=7) (e) Male T1 tarsi and metatarsus (medial aspect), elav-GAL80 repression in subset of dsx cells, metatarsal sex comb (boxed). (f) Wild-type male and (h) wild-type female prothoracic axonal projections (close-up from Fig. 2k). (g) Atrophied male and (i) atrophied female prothoracic axonal projections, post-amputation. Point of amputation, arrowheads (a). Scale bar = 50 μm.

Extensive innervation in the ventral prothoracic ganglia (Prg) occurs in both sexes. In males these project ipsilaterally, connecting with TN2 cells, as well as contralaterally to form a distinct commissural bridge (Fig. 2i,j and 3f). Females exhibit only ipsilateral projections (Fig. 2k,l and 3h). Previous studies demonstrated gustatory receptors in the foreleg exhibit a similar sexual dimorphism30. The male foreleg exhibited a significant difference in dsx cell numbers, in both metatarsus and tarsi 2-5, compared to females (Fig. 3a vs Fig. 3d). Some of these cells are neuronal, as fewer cells were observed when expression of GAL80, a GAL4 repressor, was targeted to postmitotic neurons using the elav promoter (Fig. 3e). More cells were repressed in males than females, indicating there are more neurons in the male foreleg.

As axons degenerate when severed31, we amputated forelegs of dsx flies expressing membrane-bound GFP. Amputating male forelegs below the sex comb, or an equivalent point in females, showed no overt degeneration in projections (data not shown). With amputations performed above the sex comb, or equivalent point in females, no foreleg projections were observed (Fig. 3g,i). Since only gustatory neurons cross the midline in male VNCs30, at least some of the male dsx-projections are from gustatory neurons involved in non-volatile taste sensation.

The abdominal ganglion (Abg), the anatomical foci for copulatory behaviors in males and females1, is the sole region of the VNC where dsx expression occurs in both sexes, with adult females exhibiting expression in a larger number of cells (Fig. 2g,h; Supplementary Table 1). Neurons from this region send projections through the abdominal nerve trunk to ramify the internal genitalia in both sexes (data not shown). In males four bilateral pairs of axonal fascicles run through the cervical connection, connecting the brain and VNC (Fig. 2i). In females there are five bilateral pairs of axonal fascicles, perhaps reflecting the increased neuronal expression apparent in the Abg (Fig. 2k and 3h).

In summary, we found dsx expression in ~900 neurons (9 groups) in males, and ~700 neurons (5 groups) in females in the adult CNS (Supplementary Table 1). In the development of these neurons, we highlight 48 hr pupae as a critical point of divergence between the sexes. At this stage SN and TN2 cells arise in both sexes (later disappearing in females), TN1 neuronal clusters appear in males (but never females), and until this stage, the number of dsx-Abg neurons is consistently higher in males (Supplementary Fig. 3 and Supplementary Table 1).

Our examination of dsx neurons clearly demonstrates significant differences in cell numbers, projections and synaptic density in regions of the CNS associated with sex-specific behavior. We speculate that assembly of these neural networks, and their sex-specific differences, are responsible for the differences in male and female behavioral programs.

dsx and fru specify sexual dimorphisms in the CNS

DsxM and FruM are co-expressed in a restricted number of regions within the male CNS14, 18, 20. We found dsx-expressing neurons reiterated this co-localization with FruM (Fig. 4). Both DsxM and FruM expression are required for the complete development of a number of specific neuronal populations within the male CNS14, 18, 20. Therefore, to examine the contributions of fru and dsx to the specification of dsx-expressing neurons, we counted the number of dsx neurons in fru and dsx mutant backgrounds (Fig. 5 and Table 1).

Figure 4

Co-localization of FruM neuronal cells and dsx cells (expressing UAS-nGFP) in 3-day-old adult male. anti-nGFP (green), anti-FruM (magenta). (a) Brain; Dorsal view; anterior up. dsx and FruM neuronal cells and clusters designated. (b) VNC. Ventral view; anterior up. dsx and FruM neuronal cells and clusters that colocalize, designated. FruM clusters as previously described20. Scale bars = 50 μm.

Figure 5

dsx expression in CNSs of FruM-null males and females expressing FruM or the anti-apoptotic transgene UAS-p35. (a,b) Schematic of dsx nGFP expression in wild-type adult male (a) and female (b) CNSs. Individual neuronal clusters designated. (c) nGFP and (d) Membrane-bound GFP in FruM-null adult male brain. (e) nGFP and (f) Membrane-bound GFP in FruM-null adult male VNC, reduction in prothoracic contralateral (boxed area) and cervical connection projections (arrowhead). (g) nGFP and (h) membrane-bound GFP in adult female brain expressing FruM. (i) nGFP and (j membrane-bound GFP in adult female VNC expressing FruM. (k) nGFP and (l) membrane-bound GFP in adult female brain expressing both DsxF and DsxM. Supernumerary SN cells (boxed area). (m) nGFP adult female VNC expressing both DsxF and DsxM. Supernumerary male-specific TN1 and TN2 cells (boxed area, arrow). (n) Membrane-bound GFP in adult female brain expressing both DsxF and DsxM. Prothoracic contralateral projections (boxed area). (o) UAS-p35;dsx adult female brain, supernumerary SN cells (boxed area). (p) UAS-p35;dsx adult female VNC, supernumerary TN1 (arrowheads) and TN2 (boxed area) neurons. (q-s) UAS-nLacZ,-mCD8::GFP,-p35;dsx adult female prothoracic ganglion (Prg) (q) Merged image, co-expression of membrane-bound GFP, green (r) and nuclear ßGal, magenta (s). Supernumerary TN1 (arrowheads) and TN2 cells (boxed area) and ectopic contralateral projections. Ventral views; anterior up. (c, e, g, i, k, m) anti-GFP, green. (d, f, h, j, l, n, o, p) anti-mCD8, green and neuropil counter-stained with anti-nC82, magenta. Genotypes as per Methods. Scale bar = 50 μm

Table 1

dsx driven nGFP expression in CNSs of FruM-null males and females expressing FruM, DsxM, or p35

	5-day-old Adult Male		5-day-old Adult Female
	wild-type (CS)	fruF/dsxGAL4, Df(3R)fru4-40	wild-type (CS)	fruM/dsxGAL4, Df(3R)fru4-40	dsxGAL4/dsxDom	UAS-p35; dsxGAL4
FruM	+	−	−	+	−	−
DsxM	+	+	−	−	+	−
DsxF	−	−	+	+	+	+
Neuronal clusters
Brain
-PC1	56.9 ± 5.0	48.2 ± 8.6**	8.7± 2.0	9.6 ± 1.1	45.8 ± 6.4 **	11.1 ± 2.8
-PC2	78.6 ± 3.1	67.4 ± 8.8*	11.2 ± 1.9	10.7 ± 1.5	56.3 ± 4.7 **	16.1 ± 1.8*
-PC3	13.6 ± 1.0	10.7 ± 1.2**	6.4 ± 1.4	6.3 ± 1.9	7.0 ± 1.3	11.7± 2.1**
-SN	1 ± 0	1 ± 0	0 ± 0	0 ± 0	1 ± 0 **	1.0 ± 0 **
-aDN	2 ± 0	2 ± 0	2 ± 0	2 ± 0	2 ± 0	2.2 ± 0.4
Ventral Nerve Cord
-TN1	22.4 ± 1.7	23.0 ± 2.2	0 ± 0	0 ± 0	15.5 ± 2.2**	3.0 ± 0.6**
-TN2	6.9 ± 3.0	6.6 ± 0.7	0 ± 0	0 ± 0	6.0 ± 2.5**	7.0± 2.0**

The presence or absence of FruM, DsxM, and DsxF expression is noted below genotypes. Counts represent one cluster per hemisegment. Mean ± standard deviation. n=10 for all genotypes.

p-values <0.005

p-values <0.0001.

In FruM-null males (expressing DsxM but lacking FruM), dsx-pC1, -pC2 and -pC3 neuronal numbers were significantly reduced compared with wild-type (Fig. 5c,d). A significant reduction in the density of axons was observed in the cervical connection and in the contralateral connections in the Prg (Fig. 5f). No reduction in dsx-SN, -TN1, or -TN2 clusters was observed (Fig. 5e); however we have previously shown a FruM-dependent reduction in dsx cells in the Abg18. FruM is therefore required for the full complement of dsx neurons. To test whether FruM is sufficient to specify dsx neurons, we counted dsx neurons in females expressing FruM, and found the number of dsx neurons was not significantly different from wild-type females (Fig. 5g-j). This demonstrates that FruM expression is not sufficient to specify complete development of male-specific circuitry, as previous studies suggested11.

To determine if dsx is sufficient to specify this neural circuitry, we counted dsx neurons in females expressing both DsxF and DsxM. Neuronal numbers were significantly increased in -pC1, -pC2, and -SN (but not -pC3) in the brain, and all VNC clusters compared to wild-type females with a concomitant increase in projections and formation of contralateral connections (Fig. 5k-n; Table 1). However, the number of dsx-pC1 and -pC2 neurons was still significantly lower than wild-type males (Table 1). Since DsxF precipitates Programmed Cell Death (PCD) in ~20 cells of the fru-P1 neuronal cluster20; and this cluster is a sub-population of dsx-pC1, a reduction in overall cell number in DsxM-expressing females is expected. Similar cell-death processes may occur in the pC2 and pC3 clusters.

Sex-specific PCD is one mechanism for creating sexual dimorphism in the brain12, 20. Thus we investigated its role in sculpting the dsx circuitry by expressing the cell-death inhibitor p35 in dsx cells32; and counting the number of neurons in adult females. dsx-SN and -TN2 cells, present in both sexes in 48hr pupa but specific to males in the adult (Supplementary Table 1), were protected from PCD, as their numbers were not significantly different from wild-type males (Fig. 5o,p; Table 1). Additional neurons were observed in -pC2 and -pC3 clusters as well as cells of the TN1 cluster (3± 0.4, n=14; Fig. 5o-s), which are never observed in females. Intriguingly when TN1 and TN2 neurons are present in females they display contralateral projections ordinarily seen only in wild-type males (Fig. 5q).

Taken together, these results demonstrate that DsxM and DsxF are the primary regulators of dimorphisms in dsx neurons, though FruM function is required to obtain a full complement of male-specific dsxGAL4 neurons, and that sex-specific PCD is one mechanism used to assemble this dimorphic circuitry. Importantly, we show that small changes in neuronal populations can significantly alter the organization and connectivity of the neural network that presumably forms the anatomical basis for sex-specific behaviors.

dsx neurons are required for male sexual behavior

dsx mutant males display aberrant courtship but the neural etiology for this abnormality is unknown2. Having shown that dsx is required in the construction of sex-specific neural substrates; we asked whether dsx-expressing neurons play an active role in courtship behavior. We quantified fertility and courtship behavior in males expressing UAS-tetanus neurotoxin light chain (TNT), which targets neuronal synaptobrevin33, in dsx neurons. After one week in the presence of several virgin females, these males were completely infertile (n=60) (Fig. 6a). We next determined whether the infertility was a consequence of defects in courtship and/or mating. The time taken for UAS-TNTG;dsx males to initiate courtship was significantly increased (Fig. 6b), and once courtship was initiated, the amount of courtship performed was severely reduced and consisted entirely of intermittent following and orientation with no attempted copulation (Fig. 6c). These males did not extend their wings and there was a complete absence of sine- and pulse-courtship song (Fig. 6e).

Figure 6

dsx neurons control male sexual behavior. (a) Male fertility. p-values * <0.05, ** <0.0001 (Fisher Exact Test). (b) Courtship initiation. Mean ± SEM. p-values * <0.05 (Tukey-Kramer HSD statistical test). (c) Courtship index. Mean ± SEM. p-values * <0.05 (Tukey-Kramer HSD test). (d) Percentage males mating in 4-h. p-values ** <0.0001 (Fisher Exact Test). (e) Song-recording, 5-7 day old males. Pulse (P) and sine (S) song components indicated above traces, courtship display below. Each trace represents a fraction of a 10 min recording. Scale bar= 200 ms. Genotypes indicate males. Target females, wild-type. Wild-type males, (15); 18.6 ± 2.2 sine bouts per min (SBPM), 19.9 ± 1.4 pulse trains per min (PTPM), 8.1 ± 0.3 mean pulses per train (MPPT), 31.7 ± 3.0 interpulse interval (IPI, ms). UAS-TNT; dsx males, (10); no recordable data. UAS-TNT; dsx males, (10); 18.4 ± 1.5 (SBPM), 26.6 ± 3.8 (PTPM), 10.0 ± 0.4 (MPPT), 34.0 ± 0.4 (IPI, ms). Mean ± SEM. n’s in parentheses.

During an extended 4hr observation period, ~95% of control flies successfully copulated, but no UAS-TNTG;dsx males copulated (Fig. 6d). Courtship defects in UAS-TNTG;dsx males cannot be explained by more general defects in morphology or sensorimotor function because UAS-TNTG;dsx males do not exhibit gross anatomical abnormalities in their genitalia or reproductive systems, including neuronal innervation (data not shown), and performed at least as well as wild-type and control males in assays for locomotion, flight, olfaction, and taste (Supplementary Fig. 5 online).

Thus inhibition of dsx neuronal function in males disrupts the early steps of courtship (orientation, following) and the complete absence of the later steps (wing extension, courtship song and attempted copulation) suggesting that dsx neurons directly and specifically contribute to male courtship behaviors.

Since dsx is expressed in a variety of tissues outside the nervous system (Supplementary Fig. 2), we tested whether aberrant courtship behavior in UAS-TNTG;dsx males results from disruption of neural function. We used elav-GAL80 to inhibit GAL4-driven expression of TNT specifically in neurons. elav-GAL80 specifically and comprehensively inhibits neural expression of GFP in dsx; UAS-nGFP flies, validating the efficacy of this tool (Supplementary Fig. 3k). Removing TNT expression specifically from neurons reversed the observed courtship defect. Males expressing UAS-TNTG; elav-GAL80,dsx showed greatly improved levels of fertility (Fig. 6a), courtship initiation (Fig. 6b), and a significant recovery in their CI’s (with consequent restoration of courtship modalities such as licking, tapping, and wing extension) when compared with UAS-TNTG;dsx males (Fig. 6c). The percentage of males copulating within a 4-h observation period was also significantly improved (Fig. 6d), as was the ability to produce both sine- and pulse-song bouts (Fig. 6e). These results demonstrate the aberrant courtship phenotypes of UAS-TNTG;dsx males were largely a consequence of disrupting dsx neuronal function. That not all behaviors were restored to control levels likely reflects the inability of GAL80 to fully repress GAL4 function. Collectively, these results demonstrate that dsx neurons are required for male courtship behavior.

dsx neurons are critical for female sexual behavior

The neurobiological basis of female sexual behavior is poorly defined, though it was suggested that dsx’s influence on the sex of the CNS might be critical to female behavior8. We investigated the effects of disrupting dsx neuronal function on female behavior through expression of TNT (Fig. 7). To confirm the neuronal contribution of dsx cells to female sexual behavior, we also used elav-GAL80 to restrict expression of TNT to non-neuronal tissues.

Figure 7

dsx neurons control female sexual behavior. (a) Female fertility. p-values ** <0.0001 (Fisher Exact Test). (b) Egg-laying. Mean ± SEM. p-values * <0.0001 (Dunnett’s Test). (c) Percent copulation over time (10 minute intervals for 1 h). (d) Line crossings during copulation. Mean ± SEM. p-values * <0.05 (Tukey-Kramer HSD test). (e) Copulation duration. Mean ± SEM. p-values * <0.05 (Tukey-Kramer HSD test). (f) Percentage females re-mating with the same male in 4 h. p-values * <0.05 (Tukey-Kramer HSD test). Genotypes indicate females. Target males, wild-type. n’s in parentheses.

UAS-TNTG;dsx females were completely infertile (Fig. 7a), and laid no eggs over five consecutive days post-mating (Fig. 7b). Over time the abdomens of these females became distended, with extruded ovipositors (Supplemental Fig. 6a) and mature eggs atrophying in their oviducts (Supplemental Fig. 6b).

We next observed the courtship behavior of UAS-TNTG; dsx virgin females paired with individual wild-type males. Initially the percentage of females copulating in a 1-h period was investigated (Fig. 7c). Virgin TNTG; dsx females were less receptive than control females within the first 10 minutes (<30% vs. >96%, respectively) but by 40 minutes the percentage of TNTG; dsxGAL4 females that had copulated was near control levels (>93% and 100%, respectively).

We then investigated UAS-TNTG;dsx females’ behavior during, and several hours after, copulation. During copulation, these females actively rebuffed males, exhibited sustained wing flicks, kicking, and failed to remain stationary (Supplementary Video 3 online). Although males often manage to grasp hold of a female, they never manage to spread the female’s wings during copulation. The result of this aberrant copulation with UAS-TNTG;dsx females was a dramatic increase in locomotion of copulating pairs (Fig. 7d; Supplementary Video 3) compared to controls. Indeed we found a significant decrease in the length of copulation with UAS-TNTG;dsx females compared with controls, presumably a consequence of the vigorous rejection displayed by these females (Fig. 7e). Intriguingly, during a 4-h observation period, approximately 30% of UAS-TNTG;dsx females were observed to re-mate with the same male, some as many as four times (Fig. 7f), though they continued to exhibit vigorous rejection behaviors throughout. Control females never re-mated during the observation period.

The observed infertility of UAS-TNTG;dsx females was not due to a lack of sperm transfer during the shorter copulation time, as the reproductive tracts of females who copulated with males with the sperm-enriched mitochondrial marker Don Juan-GFP, were positive for GFP (Supplementary Fig. 6h online)34. We also confirmed transfer of seminal fluids, required for post-mating rejection responses, using a transgene encoding the accessory gland protein Sex-peptide fused to GFP (Supplementary Fig. 6f)35. That no ‘live births’ were observed suggests this infertility is due to an inability to deposit eggs into the uterus, preventing sperm, stored in the seminal receptacle and spermathecae, from fertilizing the mature oocyte. UAS-TNTG;dsx females exhibited no gross anatomical abnormalities in the genitalia or reproductive system, including their neuronal innervation (data not shown); and they performed at least as well as wild-type and control females in assays for locomotion, flight, olfaction, and taste (Supplementary Fig.5).

To interrogate the unusual re-mating phenotype, we compared post-mating responses in mated UAS-TNTG;dsx females against mated control females, and retested for receptivity 24-h later with a second naïve male. During a 1-h observation period 69% of mated UAS-TNTG;dsx females re-mated (n=13), while no mated control females re-mated (n=14). In addition, male courtship behaviors were not suppressed by mated UAS-TNTG;dsx females, as wild-type Drosophila females ordinarily do post-mating2; rather they continued to elicit vigorous courtship (CI= 82.4% ± 4.4) compared with mated control females (CI=28.4% ± 6).

To confirm the infertility and courtship defects observed were a consequence of disrupting dsx neurons, we used elav-GAL80 to inhibit GAL4-driven neuronal expression of TNT. Fertility and egg laying in UAS-TNTG;elav-GAL80,dsx females was not appreciably different from controls (Fig. 7a and 7b). Similarly, locomotion during copulation, copulation duration, and re-mating after copulation, were restored to wild-type levels (Fig. 7d-f). This suggests that the abnormal behaviors exhibited by UAS-TNTG;dsx females were a result of disrupting dsx neurons. These results demonstrate that dsx neurons play a critical role in female courtship and reproductive behaviors. In particular, dsx neurons appear to be important for receptivity to copulation and mating-induced behavioral changes.

Discussion

Using the novel dsx allele we have been able to comprehensively describe dsx expression in neuronal and non-neuronal tissues throughout development. By manipulating the sex of dsx-expressing cells we demonstrated their ability to direct a sex-specific program of morphological development. We also identified profound dimorphisms in neural circuitry in males and females; from differences in cell numbers in homologous clusters; to sex-specific neuronal development; to density, organization and connectivity of axonal projections. Specific inhibition of the function of these neurons in the males and females resulted in disruption of distinct sex-specific behavioral outputs, suggesting that differences in neuroanatomy instruct sex-specific behaviors.

In males significant overlap between fru- and dsx-expressing neurons occurs, allowing functional roles for certain dsx neurons to be inferred from their intersection with fru neurons with defined roles in male sexual behavior14, 20, 36. In males dsx-pC1 colocalizes with the medial fru-P cluster (Fig. 4a), a focus for licking and copulatory behaviors37-39. dsx-pC1 intersects with a subset of fru-P1 neurons, with a demonstrated role in courtship initiation and whose sexual differentiation and axonal morphology is dependent upon both fru and dsx. dsx-pC2 colocalizes with the lateral fru-P cluster20 (specifically P2-P4; Fig. 4a), and dsx-pC3 colocalizes with the fru-pSP2 cluster (Fig. 4a), both regions of the brain associated with initiation, following, tapping and wing extension37-39. Projections to the ventral SOG show intense synapses positionally associated with the fru-mCAL, a region implicated in control of sequential courtship steps13, 40 (Fig. 2c). This coexpression of FruM and DsxM implies cooperation in shaping a shared male neural circuitry, an observation supported by the demonstrated reduction in dsx neurons in pC1, pC2 and pC3 clusters in FruM-null males, analogous to the cooperation demonstrated in the Abg, or Msg14, 18. Early studies identified the ventral Msg as a neural focus for wing extension and song production28. More recently successful song production was correlated with the presence of a male-specific neuronal population in the Msg, and the existence of a localised song pattern generator in the Msg14, 29. Again significant overlap between fru- and dsx-expressing neurons is apparent in the VNC (Fig. 4b). dsx significantly increases neural complexity in this region by specifying the appearance of male-specific neurons and dimorphic projection patterns. Further complexity arises from the dsx-dependent dimorphism in axonal projection paths in the Prg (ispilateral in females versus contralateral in males), which appear to arise as a consequence of inputs from male-specific gustatory receptor neurons in the foreleg30. These receptors have a demonstrated role in non-volatile pheromonal gustation associated with male-specific tapping behaviors in courtship27.

That these dsx neurons direct male-specific behaviors is supported by our finding that male courtship is impaired when their neural function is specifically disrupted. Therefore while fru is necessary in specifying male sexual behaviors, it is not sufficient; rather dsx is also required for complete specification of male courtship behaviors.

Male courtship behaviors, being robust and quantifiable, have long been the focus of behavioral genetics; however, males are not the only participants in copulatory bouts. Females appear superficially passive but exhibit subtle behaviors, consisting mostly of ‘rejection’ behaviors such as wing flicking or kicking2. Little is known about the effects these ‘rejection’ responses have on courting males, and what stimuli trigger these behaviors. From the female’s perspective however she must be able to assess a courting male to make an assured judgment of species type before she will sanction mating. Increased receptivity is indicated when a female slows down and ceases ‘rejection’ behaviors. It is worth noting that some species of Drosophilids actually exhibit an acceptance posture. While no clear acceptance posture has been demonstrated in melanogaster our findings suggest female cooperation does occur to facilitate copulation. A neural focus for receptivity has been identified in the dorsal female brain2; it seems likely that dsx neurons in this region contribute to female mating decisions, just as homologous neurons in males are involved in male decisions20, 41. This is supported by our findings that disrupting synaptic activity of dsx neurons in females impairs distinct female courtship behaviors. These females are seemingly incapable of sampling the male’s display, and so incapable of providing any acceptance response. When copulation occurs, lack of female cooperation is evidenced by continuous movement and rejection behaviors. While sperm and seminal fluids are successfully transferred to UAS-TNTG​;dsx females, they lay no eggs, re-mate, and remain incapable of actively rejecting or suppressing further courtship. This indicates that disrupting dsx neuronal function in females also suppresses post-mating behaviors, and implies that some of these neurons relay information from fru+-sensory neurons in the female reproductive tract in response to Sex-peptide42, 43.

We have demonstrated that dsx cells not only can reprise the functional roles of endogenous dsx in establishing external sexual morphology but also establishes a dimorphic neuroanatomy capable of directing distinct sex-specific behavioral outputs. In addition, dsx expression occurs in sexually physiologically relevant non-neuronal adult tissues such as the fat body and oenocytes, whose correct sexual identity is critical to normal sex-specific behaviors8, 9, 44. Our findings indicate that adult sex-specific behaviors may arise as a consequence of changes to distinct cell groups during development, creating dimorphic neural circuitry in equivalent regions of the male and female brain. Our results demonstrate a fundamental requirement for dsx in the sexual development of both neuronal (sex-specific circuitry) and non-neuronal tissues (sex-specific physiology). Future studies will be aimed at identifying how dsx effects these dimorphic changes, refining our understanding of how individual dsx neurons instruct sex-specific neural programs, and at identifying the relative contributions of both “mind” (fly brain) and body to these behaviors.