Currently, sex differences in behavior are believed to result from sexually dimorphic neural circuits in the central nervous system (CNS). Drosophila melanogaster is a common model organism for studying the relationship between brain structure, behavior, and genes. Recent studies of sex-specific reproductive behaviors in D. melanogaster have addressed the contribution of sexual differences in the CNS to the control of sex-specific behaviors and the development of sexual dimorphism. For example, sexually dimorphic regions of the CNS are involved in the initiation of male courtship behavior, the generation of the courtship song, and the induction of male-specific muscles in D. melanogaster. In this review, I discuss recent findings about the contribution of cell death to the formation of sexually dimorphic neural circuitry and the regulation of sex-specific cell death by two sex determination factors, Fruitless and Doublesex, in Drosophila.
Most animals, including humans, exhibit conspicuous sex differences in a variety of behaviors, particularly reproductive behavior, due to sexually dimorphic neural circuitry (reviewed in Simerly 2002). Currently, there is limited knowledge about the extent of sex differences in the structure of the central nervous system (CNS), the control of sex-specific behavioral patterns by sexually dimorphic neurons, and the development of sexual dimorphism. The fruit fly Drosophila melanogaster is a common model organism for studying innate, sex-specific reproductive behaviors to elucidate the relationship between brain structure, behavior, and genes that control the development and function of the CNS (Hall 1994). Studies about sexual dimorphism in the Drosophila CNS have provided significant insights into these issues.
Early studies of sexual dimorphism in the structure of the CNS in adult Drosophila only revealed a few morphological differences between males and females, such as differences in the projection pattern of taste sensory neurons (Possidente & Murphey 1989; Taylor 1989) and the number of fibers in mushroom bodies (Technau 1984). The lack of significant anatomical sexual dimorphism suggested that sex differences in behavior resulted from differences in fine neural connectivity or physiology, rather than structural differences in neural circuitry (Manoli et al. 2005; Stockinger et al. 2005). More recent investigations demonstrate that the adult Drosophila CNS has sexually dimorphic morphology (Kondoh et al. 2003; Kimura et al. 2005, 2008; Datta et al. 2008; Rideout et al. 2010) and sex-specific programmed cell death is responsible for the formation of sexual dimorphism (Kimura et al. 2005, 2008; Sanders & Arbeitman 2008; Rideout et al. 2010). Here, the role of sex determination genes in the developmental organization of sex-specific neural circuitry and the contribution of sex-specific cell death to the formation of sexually dimorphic neural circuitry for male courtship behavior and male-specific muscles in Drosophila are reviewed.
Sex determination in Drosophila
Sex determination in Drosophila is controlled by the X chromosome-to-autosome ratio (X:A ratio) (reviewed in Burtis 1993; Cline & Meyer 1996) (Fig. 1). In females (XX) with X:A = 1, the Sex-lethal (Sxl) gene and its product, Sxl protein, are expressed. However, in males (XY) with X:A = 0.5, the Sxl gene is suppressed. Sxl is an RNA binding protein that regulates female-specific splicing of transformer (tra) pre-mRNA to generate the Transformer (Tra) protein. As a result, females produce the Tra protein but males do not. Tra protein, which is also an RNA binding protein, acts with Transformer2 (Tra2), a cofactor that is expressed in both sexes, to regulate the sex-specific splicing of downstream genes, such as doublesex (dsx) and fruitless (fru).
In the presence or absence of Tra, female- or male-specific splicing of dsx transcripts results in the female- or male-specific isoform of Dsx protein (DsxF or DsxM), respectively (Burtis & Baker 1989). DsxF and DsxM are zinc finger transcription factors of the DMRT (doublesex and mab-3-related transcription factor) family (Raymond et al. 1998) that have a DNA binding domain called the DM domain (Burtis & Baker 1989). DsxF and DsxM are important determinants of female and male somatic structure and external morphology, respectively. In addition, recent studies have shown that Dsx is involved in the formation of sexual dimorphism of the CNS in Drosophila (Billeter et al. 2006; Rideout et al. 2007, 2010; Shirangi & McKeown 2007).
However, the main determinant of sexual dimorphism of the CNS is fru, which is a large gene (∼140 kb) with four promoters and alternative exons (Ito et al. 1996; Ryner et al. 1996). Alternative splicing of the transcripts produces zinc finger transcriptional factors with the BTB (Broad-Complex, Tramtrack and Bric à brac) domain. Since the transcripts from the most distal promoter (P1) are spliced under control of Tra and Tra2, the female-specific transcripts are not translated into protein (Usui-Aoki et al. 2000; Lee et al. 2000; Baker et al. 2001). As a result, P1-derived proteins are male-specific Fru proteins called FruM.
fruitless and doublesex control courtship behavior in Drosophila males
In Drosophila, the courtship ritual consists of a complex, stereotypic sequence of behaviors (Hall 1994; Yamamoto et al. 1997). The ritual begins when a male recognizes a female by visual and olfactory cues. After orienting his body axis toward her, the male follows her while tapping her with his forelegs, which is believed to convey gustatory pheromone information. He also generates species-specific love songs by vibrating one wing at a time. These songs have an aphrodisiac effect on the female, unless she escapes from the courting male. As a result, she slows down and occasionally stops, allowing the male to lick her genitalia and attempt to mount her. If the female is receptive (e.g. a virgin female), she opens her wings and vaginal plates, which allows the male to copulate with her. During copulation, the male remains on the female for about 20 min before releasing the genital contact. However, if the female is unreceptive (e.g. a recently mated female), she rejects the male by running away, kicking, or extruding her ovipositor.
An important determinant of male courtship behavior is fru. FruM is expressed in small groups of neurons (about 2% of all neurons) scattered throughout the Drosophila CNS (Usui-Aoki et al. 2000; Lee et al. 2000). Mutant males that do not express FruM in these neurons exhibit bisexual or homosexual courtship behavior and much less sexual activity than wild-type males (Ito et al. 1996; Ryner et al. 1996; Goodwin et al. 2000). These results suggest that FruM masculinizes the neurons that establish the neural substrates for wild-type male behavior. However, it was not known which neural circuits fru-expressing neurons construct in the CNS or whether these circuits have sex-specific differences. Recently, several studies have defined the morphology of fru-expressing neurons by visualizing them with a green fluorescent protein (GFP) reporter (Billeter & Goodwin 2004; Kimura et al. 2005; Manoli et al. 2005; Stockinger et al. 2005).
Another important determinant of male courtship behavior is dsx. Mutant males that do not express either DsxM or DsxF, which are intersexual in appearance, exhibit less courtship behavior and sing irregular love songs (Villella & Hall 1996). However, DsxM is not sufficient for male-type courtship behavior. Although chromosomal females that only express DsxM are male in appearance, they do not show any male courtship behavior (Taylor et al. 1994). Recently, Rideout et al. (2010) demonstrated the importance of the dsx-expressing neurons in male courtship behavior by inhibiting their function in males that expressed upstream activation sequence (UAS)-tetanus neurotoxin (TNT) light chain, which targets neural synaptobrevin, in dsxGal4-expressing neurons. This inhibition partly disrupts the early stages of courtship, such as orientation and following, and eliminates the later stages, namely, wing extension, courtship song, and attempted copulation. In addition, male transheterozygotes for dsx and fru show a reduction in male courtship behavior (Shirangi et al. 2006). Moreover, dsx is required to specify a sexually dimorphic population of fru-expressing neurons in the mesothoracic ganglion, and DsxM and FruM are necessary to specify the courtship song (Rideout et al. 2007). Thus, dsx acts in concert with fru to control male courtship behavior.
Sexual dimorphisms in the brain and control of courtship initiation
fru-expressing neurons form sexually dimorphic neural circuitry
The CNS of adult Drosophila has about 100 000 neurons. Immunostaining of the CNS with anti-Fru antibodies shows that there are about 1700 fru-expressing neurons that have been classified into 20 groups (Lee et al. 2000). If FruM is involved in the formation of neural circuits that are important for sex specific behaviors, then fru-expressing neurons would be expected to show structural sexual dimorphism. To test this hypothesis, my laboratory used an enhancer trap strain, fruNP21-gal4, which has a P-element-GAL4 insertion in the second intron of the fru gene (Kimura et al. 2005). In this strain, about 80% of the neurons labeled with the fruNP21-gal4 reporter, mCD8-GFP, express Fru (Kimura et al. 2008). The expression of the reporter indicates the projection and dendritic branching patterns of fru-expressing neurons. Then, by labeling neurons that share the same lineage with the Mosaic Analysis with a Repressible Cell Marker (MARCM) method (Lee & Luo 1999), we were able to resolve and identify distinct groups of fru-expressing neurons (Kimura et al. 2008).
A comprehensive analysis of these neurons in the Drosophila brain reveals two types of sex differences in fruNP21-expressing neurons (Kimura et al. 2005, 2008). One type is the presence of male-specific neurons. For example, a cluster of about 20 Posterior 1 (P1) neurons in the dorsal posterior region of the brain is only present in males (Kimura et al. 2008) (Fig. 2). Each of these neurons extends a primary transversal neurite with extensive ramifications in the bilateral protocerebrum. The other type of sex differences involves the number of cells in and projection patterns of neuron clusters. Specifically, in most clusters, males had significantly more cells than females. For example, a cluster of mAL neurons (medially located just above the antennal lobe) had 30 cells in males but only five cells in females (Kimura et al. 2005) (Fig. 2). These neurons seem to function as a relay between the subesophageal ganglion, the primary gustatory center of the brain, and the superior lateral protocerebrum. mAL neurons have bilateral projections in males and contralateral projections in females. In addition, in the subesophageal ganglion, a forked arborization branching pattern is only found in females. These results suggest the intriguing possibility that the sexually dimorphic mAL interneurons may be involved in the integration of gustatory pheromone inputs (Koganezawa et al. 2009).
If a few sexually dimorphic neurons, such as the P1 neurons, can control sexually dimorphic behavior, then a female fly would be expected to exhibit male-type behavior if those neurons were masculinized. To test this hypothesis, my laboratory used MARCM to generate a few tra1 homozygous clones, which were masculinized since Tra is a female-specific protein, in the brains of female Drosophila flies (Kimura et al. 2008). Then, we determined whether the mosaic females displayed male courtship behavior and which parts of their brains were masculinized. Most mosaic females with a masculinized clone of P1 neurons initiate courtship, follow other females, and spread a wing; however, the subsequent behaviors of singing or attempted copulation are absent. The P1 neurons, which are located near the mushroom body, are in the same general area that Hall (1979) assigned as the focus for the initial steps of male courtship behavior, including tapping, following, and wing extension. The observation that the P1 cluster can initiate male courtship behavior in females indicates that sex-independent neural circuits underlying male-type behaviors are present in the female Drosophila brain. However, in wild-type females, male courtship behavior is not initiated due to the absence of P1 neurons.
Programmed cell death is involved in the formation of sexual dimorphism in the central nervous system
Female-specific cell death may result in the presence of male-specific P1 neurons. To test this hypothesis, my laboratory examined the effect of mutations that prevent cell death (Kimura et al. 2008). Since the primary cell death genes in Drosophila, head involution defective (hid), grim, and reaper (rpr), are aligned in tandem in the genome, they can be removed by a single deletion, Df(3L)H99 (White et al. 1994). In mosaic females with MARCM clones that are homozygous for the H99 deficiency, P1 neurons formed, suggesting that programmed cell death may remove the P1 cluster from the female brain (Fig. 3).
A similar mechanism is involved in the formation of sex differences in mAL neurons (Kimura et al. 2005) (Fig. 3). If female-specific cell death occurs during the formation of mAL neurons, then the inhibition of cell death by the H99 deficiency is expected to increase the number of mAL neurons in females. Indeed, the number of mAL neurons in females increases as they do in males. The mAL neurons that escape cell death in females form the male-type ipsilateral projection (Fig. 3). These observations support the model that the elimination of neurons in male-type projections in females leads to the formation of a sexually dimorphic neural circuit.
FruM may be the sex determination factor that mediates this sex difference since it is expressed in male mAL neurons but not in female ones (Kimura et al. 2005) (Table 1). The absence of FruM in males due to a fru mutation (fru−) reduces the number of mAL neurons to that in females (about five), prevents the ipsilateral projection, and promotes the female-specific forked branching pattern of contralateral projections in the subesophageal ganglion. Conversely, in fruM mutants in which fru is constitutively spliced in the male-specific manner, the number of mAL neurons in females increases to that in males (about 30) and female mAL neurons exhibit male-type projection patterns. These results indicate that FruM inhibits the cell death of mAL neurons in males during the pupal stage, allowing these neurons to develop male-type projections. Thus, the expression of FruM regulates the sex differences in the number and projection pattern of mAL neurons.
Table 1. Phenotypes of wild-type and mutant genes involved in the formation of sexually dimorphic neurons and expression of sex determination factors in Drosophila
For P1 neurons, the formation of sex differences is more complicated (Kimura et al. 2008) (Table 1). Although wild-type males express FruM in P1 neurons, fru− males that do not express FruM still develop these neurons. However, their projection pattern is disrupted. Thus, FruM is required for correct positioning of P1 neurites in the male brain. In fruM females, P1 neurons do not develop. As a result, the expression of FruM does not contribute to the formation of P1 neurons, suggesting that FruM does not control the programmed cell death of these neurons in females.
However, the formation of P1 neurons in tra1 homozygous females (tra− XX) implies that a gene downstream of tra other than fru is involved (Table 1). The most promising candidate is dsx. Immunolabeling with a Dsx antibody shows two clusters of dsx-expressing neurons, posterior cluster 1 (pC1) and pC2, in the posterior brain (Lee et al. 2002). In addition, P1 neurons, which are a subpopulation of pC1 neurons, express both Dsx and FruM (Kimura et al. 2008). In dsx−XX females without either DsxF or DsxM, P1 neurons still form. Moreover, the presence of DsxM does not prevent the formation of P1 neurons if DsxF is absent in females (dsx−/dsxDXX), but the addition of DsxF causes the loss of these neurons in females (dsx+/dsxD). These results demonstrate that DsxF is required for the female-specific death of P1 neurons. Therefore, DsxF contributes to the elimination of P1 neurons from the female brain, which prevents females from exhibiting male-type sexual behavior.
Sexual dimorphism in the thoracic ganglion and control of courtship song
Sex mosaic studies have shown that masculinization of the ventral thoracic ganglion in the adult CNS is necessary to produce a wild-type song (von Schilcher & Hall 1979). This suggests that the neural foci for the courtship song are located in the ventral thoracic ganglia and that sex differences of this region are critical to song production. In the corresponding region of the mesothoracic ganglion, a male-specific subset of Dsx-expressing neurons, called the TN1 cluster, coexpresses FruM and Dsx (Lee et al. 2002; Rideout et al. 2007; Sanders & Arbeitman 2008). Recently, several studies have confirmed this sexual dimorphism by visualizing dsx-expressing neurons with a GFP reporter driven by dsx-gal4 (Rideout et al. 2010; Robinett et al. 2010) (Fig. 2). Furthermore, the projection pattern of TN1 neurons shows that they communicate not only with each other but also with other regions in the brain that are responsible for higher-order processing of sensory cues. As a result, this male-specific neural circuitry may be involved in the production of the male courtship song.
This sex difference in the TN1 cluster is formed during metamorphosis. In the early pupal stage, the number of Dsx-expressing neurons in the cluster is about the same in both males and females; however, during metamorphosis, these neurons disappear in females (Sanders & Arbeitman 2008). A terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay reveals that the TN1 cluster undergoes female-specific cell death (Sanders & Arbeitman 2008). However, the expression of p35, a cell death inhibitor that inhibits caspase activity, in dsx-expressing cells causes a few TN1 neurons to appear in females (Rideout et al. 2010) (Fig. 3). These results support the hypothesis that sex-specific cell death is responsible for the sexual dimorphism in TN1 neurons of the mesothoracic ganglion.
Since FruM and Dsx colocalize in TN1 neurons, it is possible that FruM plays a role in specifying the developmental fate of these neurons. However, in FruM-null males and constitutively active FruM females, the number of TN1 neurons is not significantly different from that in wild-type males and females, respectively (Sanders & Arbeitman 2008; Rideout et al. 2010). Thus, FruM does not contribute to the regulation of female-specific cell death in TN1 neurons (Table 1).
On the other hand, Dsx has a more important role (Table 1). The intersex (ix) gene, which encodes an obligate heterodimer partner of DsxF but not DsxM, is required for female-specific activities (Garrett-Engele et al. 2002). Since adult females with a combination of strong hypomorphic alleles of ix (ix3/Df(2R) en-B) have Dsx-positive TN1 neurons unlike wild-type adult females, female-specific cell death requires DsxF function (Sanders & Arbeitman 2008). However, in mutant females expressing both DsxF and DsxM, the number of TN1 neurons is only intermediate between the number in wild-type males and females (Sanders & Arbeitman 2008; Rideout et al. 2010). As a result, DsxF is not completely sufficient for female-specific cell death and DsxM may antagonize DsxF function by regulating the cell death of TN1 neurons.
Sexual dimorphism in the abdominal ganglion and the formation of a male-specific muscle
During metamorphosis, the development of adult musculature of the Drosophila abdomen also shows sexual dimorphism. For example, a pair of longitudinal muscles called the muscle of Lawrence (MOL) spans the fifth tergite of the abdomen in males but not in females (Lawrence & Johnston 1984) (Fig. 2). The formation of this muscle is associated with developmental cues in the fifth abdominal segment (A5). As a result, homeotic mutations that transform A4 or A6 to A5 induce the development of an ectopic MOL (Lawrence & Johnston 1986). Furthermore, several studies using different experimental approaches, such as mosaic experiments (Lawrence & Johnston 1986), myocyte transplantations (Kimura et al. 1994), and nerve transections (Currie & Bate 1995), demonstrate that the formation and development of the MOL depend on the sex of the innervating motoneuron in A5, rather than the sex of the progenitor myoblasts. The absence or presence of the MOL in fru or dsx mutant males, respectively (Gailey et al. 1991; Taylor 1992), suggests that fru but not dsx is required in males for the development of the MOL (Table 1). Constitutive expression of fru+ in all motoneurons with D42-gal4 or with heat-shock promotor induces the MOL in females and rescues the MOL in fru mutant males (Usui-Aoki et al. 2000). Furthermore, by using a regulatory sequence of the fru P1 promoter (Billeter & Goodwin 2004) with a fru coding mutation that results in the loss of one isoform, Billeter et al. (2006) showed that this isoform alone controlled the differentiation of the MOL and its associated motoneuron. This innervating motoneuron that is essential for inducing the MOL was identified and called the MOL-inducing (Mind) motoneuron (Fig. 2) (Nojima et al. 2010). Moreover, selective expression of fru+ in the Mind motoneuron by using the MARCM method rescues the MOL in fru mutant males (Nojima et al. 2010). These results show that FruM expression in the innervating neurons is required for the induction of the MOL.
The expression of p35 in motoneurons induces the ectopic formation of the MOL in the A4 or A6 segments in males and the A5 segment in females (Nojima et al. 2010) (Fig. 3). These results suggest that the Mind motoneuron and its homologues are eliminated by segment-specific and sex-specific programmed cell death. Indeed, time-lapse imaging shows the degeneration of a fru-expressing motoneuron in the male A6 segment, presumably a Mind homologue, during metamorphosis (Nojima et al. 2010). Since the pattern of ectopic MOL formation induced by artificial expression of p35 is similar to that of FruM, these observations suggest that the FruM expression in males generates a MOL by preventing cell death in the Mind motoneuron and that its suppression in females induces the cell death of Mind homologues. More generally, the suppression of MOL formation in other abdominal segments besides A5 in wild-type males implies that the regulation of cell death is context-dependent. It is possible that homeotic genes, such as abdA and abdB, may act in concert with fru to regulate the cell death of Mind homologues in these other abdominal segments.
Sex-specific programmed cell death is a general mechanism for creating sexual dimorphism in the brain of many organisms. In mammals, sex steroids, such as androgens and estrogens, affect sexual differentiation in the developing brain and programmed cell death creates sex differences in the number of cells in the brain (reviewed in Tsukahara 2009). For example, the sexually dimorphic nucleus of the preoptic area (SDN-POA) of the adult rat brain has more neurons in males than in females. This difference is due to the death of more neurons in the SDN-POA during the early postnatal period in females than in males. The SDN-POA of postnatal males also exhibits a higher expression level of anti-apoptotic Bcl-2 and a lower expression level of pro-apoptotic Bax than those in females. In addition, postnatal treatment of adult female rats with sex steroids increases the volume of their SDN-POA and changes the patterns of expression of Bcl-2 and Bax in this area to male-typical patterns. These results suggest that sex steroids inhibit programmed cell death in the SDN-POA in males by activation of Bcl-2 and the repression of Bax.
In both Drosophila and the nematode Caenorhabditis elegans, a sex determination factor plays a similar role. In C. elegans, a pair of serotonergic motor neurons called HSNs (hermaphrodite-specific neurons) that control egg laying is found in hermaphrodites but not in males (Sulston & Horvitz 1977). The HSNs develop in the embryos of both hermaphrodites and males, but undergo programmed cell death in males shortly after they are formed (Sulston et al. 1983). The C. elegans gene egl-1, which encodes a cell death activator, is required for programmed cell death. In addition, a sex determination factor, TRA-1A represses egl-1 to prevent the elimination of HSNs (Conradt & Horvitz 1999), similar to the activity of FruM in Drosophila.
Specifically, the expression of FruM inhibits cell death in the mAL neurons in the brain and the Mind neuron in the abdominal ganglion in males (Fig. 4). Similarly, DsxF promotes the death of P1 neurons in the brain and TN1 neurons in the thoracic ganglion in females (Fig. 4). These sex determination factors may regulate sex-specific cell death to produce sexual dimorphism in the CNS by binding to and regulating the expression of proapoptotic genes, such as rpr or hid. However, the downstream apoptotic genes targeted by FruM or Dsx are still unknown. In addition, since not all neurons that express DsxF or lack FruM undergo programmed cell death in females, some other mechanism, perhaps involving unidentified cofactors or transcriptional factors, must be involved in this process.
In conclusion, the role of sex-specific cell death in the formation of sexual dimorphism in the CNS is conserved in many organisms, including worms, flies, and mammals. Although the fru gene has only been found in insects, dsx homologues in the DMRT family are critical for sex determination in many animals (Hong et al. 2007). Further studies on the genetic networks involved in sex-specific cell death in Drosophila and other organisms will provide valuable insights into the regulatory mechanisms of sex determination factors and sex steroids that control the formation of sexually dimorphic neural circuitry.
I thank K. Izumi, R. Kimata, C. Sato and A. Urushizaki for technical assistance. This work was supported in part by a Grant-in-Aid for Specially Promoted Research (No. 1802012) from the Ministry of Education, Culture, Sports, Science and Technology and a Grant-in-Aid for Scientific Research (C) (No. 21570072) from the Japan Society for the Promotion of Science to K-I. K.