doublesex Functions Early and Late in Gustatory Sense Organ Development

Somatic sexual dimorphisms outside of the nervous system in Drosophila melanogaster are largely controlled by the male- and female-specific Doublesex transcription factors (DSXM and DSXF, respectively). The DSX proteins must act at the right times and places in development to regulate the diverse array of genes that sculpt male and female characteristics across a variety of tissues. To explore how cellular and developmental contexts integrate with doublesex (dsx) gene function, we focused on the sexually dimorphic number of gustatory sense organs (GSOs) in the foreleg. We show that DSXM and DSXF promote and repress GSO formation, respectively, and that their relative contribution to this dimorphism varies along the proximodistal axis of the foreleg. Our results suggest that the DSX proteins impact specification of the gustatory sensory organ precursors (SOPs). DSXF then acts later in the foreleg to regulate gustatory receptor neuron axon guidance. These results suggest that the foreleg provides a unique opportunity for examining the context-dependent functions of DSX.


Introduction
Studies in Drosophila melanogaster have revealed that many complex biological processes, such as the specification of somatic structures [1,2], segmental identity [3], and sexual differentiation [4,5], are under the control of master regulatory genes. These genes may act at multiple times in the course of these processes to regulate the expression of distinct sets of target genes. dsx is a unique master regulatory gene in that it functions across a wide variety of tissues to specify nearly all aspects of sexual development and differentiation. To manifest its various functions, dsx is necessarily responsive to three fundamental contexts: 1) the chromosomal sex of the cells in which it is expressed; 2) cell type; and 3) developmental stage of the cells in which it is expressed.
Chromosomal sex impacts dsx function via the sex determination hierarchy, which determines whether dsx transcripts are spliced to encode the transcription factors DSX M in males or DSX F in females [6] (Fig. 1A). DSX M and DSX F contain a shared zinc finger DNA-binding domain but differ in having sex-specific C-termini [7,8,9]. They are thought to bind to, and sex-specifically regulate the transcription of, a common set of target genes. For example, both DSX M and DSX F bind to the enhancer region of the genes encoding Yolk Proteins (YP) 1 and 2, but DSX M downregulates YP production while DSX F up-regulates YP production [8,10]. Thus, the chromosomal sex of a cell determines the sexspecific functions of dsx via post-transcriptional mechanisms.
dsx also functions in various temporal contexts to regulate developmental programs at the appropriate stage. dsx functions in the somatic gonad from embryogenesis through adulthood, while it regulates transcription of the YP genes in the fat body only during adulthood (reviewed in [13,27]). dsx can also function at multiple times within the developmental program of a tissue. In the genital imaginal disc, dsx begins functioning early to direct sexspecific patterns of cell proliferation (reviewed in [6,14]) and then later controls morphogenesis and differentiation of the disc to form sex-appropriate genitalia and analia (reviewed in [6]) [28]. Similarly during foreleg development, dsx functions initially to specify the number of bristles in the male-specific sex comb, then subsequently specifies the morphology of these bristles [29]. Thus temporal context, like cell type, constrains the function of dsx such that it regulates the right genes at the right times in development.
To understand how dsx sculpts sexually dimorphic development, many studies have focused on identifying the genes that dsx directly or indirectly regulates (reviewed in [6,13]) [28,30,31,32,33,34,35,36,37,38,39,40,41]. These studies have offered insight as to how dsx regulates diverse developmental processes. First, dsx regulates a number of patterning and signaling pathway genes in a sex-specific manner during the development of particular tissues, while these same genes are also expressed independent of dsx during the development of other tissues. Second, different genes are regulated by dsx in different cell types. A major focus will then be to determine how dsx function is integrated with spatiotemporal information such that specific target genes are sex-specifically regulated in appropriate cell types and at appropriate times in development.
Here, we address the function of dsx in the development of the gustatory sense organs (GSOs) of the foreleg. GSOs are a class of bristle sensory organs (sensilla) in which multiple gustatory receptor neurons (GRNs) project their dendrites into the shaft of a bristle to detect tastants [42,43]. The foreleg GSOs of Drosophila melanogaster are associated with two sexual dimorphisms. First, males have a greater number of GSOs on their forelegs than do females [43,44], and second, the axons of many of the constituent GRNs project across the midline of the ventral nerve cord (VNC) in males but not in females [45,46]. dsx has a known role in both of these dimorphisms: it regulates the development of at least some of the GSOs in the first tarsal segment (T1) of the foreleg [44], and represses VNC midline crossing by GRN axons [46].
Our study addresses two questions regarding the role of dsx in GSO development. First, we asked if dsx is responsible for the sexually dimorphic number of GSOs across all tarsal segments of the foreleg and when this dimorphism is established. Second, we sought to determine if GSO number and GRN axonal projections arise interdependently as the result of a single upstream developmental decision commanded by dsx or if these dimorphisms each result from independent developmental events that are individually controlled by dsx. We found that dsx does control sex-specific GSO numbers early in GSO development and that this function is temporally separable from the previously described role of dsx in regulating GRN axonal projections in the VNC. Thus, dsx regulates two distinct developmental events that impact the GSO at different times in development.
neur-lacZ (A101) and ase-lacZ were from the Bloomington Drosophila Stock Center at Indiana University (IN, USA).

Generation of Anti-DSX DBD
The DSX DBD antigen was the generous gift of J. Marvin and L. Looger (Janelia Farm Research Campus, Ashburn, VA, USA). A 60-amino acid polypeptide (MSISPRTPPN-CARCRNHGLKITLKGHKRYCKFRYCTCEKCRL-TADRQRVMALQTALRRAQ) corresponding to the DNA binding domain (DBD) of the DSX proteins [50] was encoded by a synthetically generated DNA (DNA2.0) and expressed in E. coli BL21(pLysS) as a direct fusion to an N-terminal 6-histidine affinity tag in a modified pRSET A vector (Invitrogen). The fusion protein was affinity purified over a Ni 2+ -NTA resin (Qiagen) by an imidazole gradient, and conjugated to keyhole limpet hemocyanin. Hybridoma clones from immunized mice were screened by in vitro immunoassay against the DSX DBD polypeptide (RayBiotech, Inc.), and one produced anti-DSX DBD , which stains nuclei of third instar larval tissues in patterns consistent with the expression pattern of dsx GAL4 (Fig. S1 and C. Robinett unpublished data) [11]. Anti-DSX DBD does not stain foreleg discs of dsx nulls (Fig. S2).

Preparation and Examination of Tissues
Tissues were fixed in 4% paraformaldehyde (Electron Microscopy Sciences) in phosphate-buffered saline (PBS). Unless otherwise noted, tissues for immunofluorescence were blocked in 5% normal goat serum in PBS containing 0.1% Triton X-100 (Sigma) (PBST), and Alexa Fluor fluorescently conjugated goat secondary antibodies (Molecular Probes/Invitrogen) were used at 1:500. Tissues were mounted in Vectashield (Vector Laboratories) with or without DAPI. Sample imaging was performed on Zeiss laser scanning confocal microscopes LSM 510 or 710 (Carl Zeiss), and Z-stacks were manipulated using ImageJ (NIH). Images were cropped, rotated, and adjusted for contrast and brightness using Adobe Photoshop.
To examine poxn-GAL4-driven expression of UAS-mCD8::GFP in 48 h APF forelegs, pupae were removed from puparia in PBS, fixed for 30-45 minutes at 22uC, washed three times in PBS, and examined for native GFP fluorescence. For 2-8 h APF, white prepupae were collected at the stage of puparium formation, individually sexed based on the larger size of the male gonads, aged appropriately at 25uC, kept on ice or at 4uC until dissection in PBST, fixed for 25 minutes at 22uC, and washed three times for 15 minutes in PBST. Pupal cuticle secretion around 8 h APF blocks antibody penetration [51,52,53,54], so 8 h APF forelegs were dissected off of fixed pre-pupae by either cutting between T1 Figure 1. dsx regulates the number of foreleg GSOs. (A) The sex determination hierarchy directs the generation of sex-specific DSX and FRU isoforms. The 2:2 ratio of X chromosomes to autosomes in females sets off a female-specific alternative RNA splicing cascade in which TRA directs splicing of dsx and fru transcripts into the female forms. The lack of TRA activity in males results in the production of male forms of these transcripts. (B-D) poxn-GAL4 driving expression of UAS-mCD8::GFP in a (B) male and (C) female foreleg at 48 h APF. Tarsal segments T1-T5 are indicated. Note that there are more clusters of neurons labeled in the male than in the female in T1-T4. (D) Magnified view of two distinct GSOs. The GRNs (arrows) of each GSO project their dendrites into the base of their cognate bristle (arrowheads). (E) Quantitation of foreleg GSOs in T1-T4. 3XP3DsRed was used to distinguish XY flies from XX flies in a dsx-deficient background where chromosomal sex could not otherwise be distinguished. All XY males had a sex chromosome genotype of w/Y. The genotype of the sex chromosomes of dsx-deficient chromosomal females was w/w, 3XP3DsRed, while all other females were w/y w, 3XP3DsRed. Genotype abbreviations: dsx + (UAS-mCD8::GFP; FRT82B dsx 1 , poxn-GAL4/TM6B). dsx 2 (UAS-mCD8::GFP; FRT82B dsx 1 , poxn-GAL4/dsx M+R13 ). dsx D (UAS-mCD8::GFP; FRT82B dsx 1 , poxn-GAL4/dsx D ). dsx + and dsx D are siblings from the same cross. Error bars indicate SEM. Pvalues are for comparisons between the indicated dsx mutant and dsx + of the same chromosomal sex. (*p = .07, **p,.0001, { p = .04, { p = .14, Tukey multiple comparisons of means.). doi:10.1371/journal.pone.0051489.g001 and the femur or plucking the entire foreleg away from the prepupal carcass. T1 was often damaged or had excessive background staining, preventing assessment of GSO numbers. Dissected legs were blocked in PBST plus 5% normal goat serum (PBSTN) for at least 16 hours at 4uC, incubated in primary mouse monoclonal antibody 22C10 [Developmental Studies Hybridoma Bank at the University of Iowa (DSHB), USA] at 1:20 for 40-48 hours at 4uC, washed four times in PBST over the course of 10 hours at 22uC or 24 hours at 4uC, and re-blocked in PBSTN overnight at 4uC. Treatment with the secondary antibody, Alexa Fluor 568 antimouse was the same as per the primary antibody. Legs were mounted in a circle of nail polish painted onto the slide to prevent flattening of the tissue.
GSO lineage cells were identified based on strong fluorescent signals from both native GFP and 22C10. Inspection of the confocal Z-stack slices determined whether a given cell or cell cluster met our criteria to be a GSO: in T3, a cluster of tightly grouped cells or a large single cell or two large paired cells expressing GFP and stained with 22C10 were classified as GSOs; in T2 and T4, criteria were the same as for T3, but the epithelial expression of poxn-GAL4-driven UAS-mCD8::GFP required close inspection of morphology, size, and clustering of the cells, which were discounted if they did not appear similar to the GSOs identified in T3.
Larval tissues were immunostained as described in [11]. The DSX M and AC time-course used rat anti-DSX M 5528 (from B. Oliver [55]) at 1:200 and mouse monoclonal anti-AC (DSHB) at 1:40 followed by Alexa Fluors 568 anti-rat and 488 anti-mouse. In each time-point preparation, male third instar gonads were included as internal controls; successful immunostaining of gonadal somatic cells confirmed reagent competence when DSX M was not detected in foreleg discs (C. Robinett, unpublished data). For each time-point, a foreleg disc pair (identified by the unique physical association of the two discs) from each of three larvae was mounted in a circle of nail polish (described above). DSX and ase-lacZ or neur-lacZ expression were detected with anti-DSX DBD at 1:100 and rabbit anti-b-galactosidase (Cappel/MP Biomedicals, LLC) (1:500), respectively, followed by Alexa Fluors 488 antimouse and 568 anti-rabbit.

DSX M and AC Time-course
Canton S larvae were raised at low-density in standard molasses food bottles for ca. 72 hours at 25uC. The food surface was then overlaid with 20% sucrose in H 2 O at 22uC and gently agitated to ''float'' the larvae into the liquid. 10-30 large larvae having second instar anterior spiracle morphology [57] were transferred to a 2.5-cm diameter food vial supplemented with Brewer's yeast paste; multiple such vials were set up over the course of an hour before being transferred to 25uC. After 2 hours, larvae were collected from these vials by the method above. Males having third instar anterior spiracle morphology were designated 0 h 3I and returned to 25uC on fresh food with Brewer's yeast paste for the duration of the time point, whereupon larvae were collected again and held on ice until dissection. Because of the time taken to handle, sort and stage the larvae, all time points are approximations of 63 hours.

dsx Specifies the Sexually Dimorphic Number of Foreleg GSOs
Male forelegs have more GSOs than do female forelegs across tarsal segments 1 through 4 (T1-T4) [43,58,59]. In T1 (the most proximal tarsal segment), this sexual dimorphism is regulated by dsx [44] (wherein GSOs are referred to as ''bractless bristles''). We revisited the T1 dimorphism and additionally asked if dsx regulates the number of GSOs in T2-T4 by examining the forelegs of males and females that are null for dsx function (dsx 1 /dsx M+R13 ) and comparing them to those with one wild-type copy of dsx (dsx 1 / TM6B, control). GSO cells were marked by expression of Pox neuro-GAL4-14 (hereafter, poxn-GAL4) driving UAS-mCD8::GFP [47] and visualized at 48 hours after puparium formation (h APF) as clusters of GRNs whose dendrites converged toward the surface of the leg (Fig. 1B-D). Consistent with previous quantitations of GSOs based on bristle morphology [43,58,59], our counts of GSOs on control (dsx + ) male and female forelegs showed that males have more GSOs than do females in tarsal segments T1-T4 (Fig. 1E). In contrast, dsx null flies exhibited no significant differences between males and females in the numbers of GSOs in foreleg T1-T4 (Fig. 1E), indicating that sex-specific dsx functions are necessary for this sexual dimorphism. Thus, our results with a transheterozygous allelic combination recapitulated the findings of previous experiments with homozygous dsx 1 mutants in T1 [44], and further, we found that dsx regulates GSO numbers across T2-T4.
To address whether expression of DSX M in XX chromosomal females that lacked DSX F is sufficient to produce male-like numbers of GSOs, we examined dsx D /dsx 1 flies in which only male-specific dsx transcripts are produced regardless of chromosomal sex (see Materials and Methods) [4,60]. The number of GSOs in XX; dsx D /dsx 1 individuals did not significantly differ from that of their XY; dsx D /dsx 1 siblings in T1-T4 (Fig. 1E). Thus, in the absence of DSX F , DSX M is sufficient to generate the male number of foreleg GSOs.
The greater number of GSOs in males relative to females could be due to a GSO-promoting action of DSX M in males, a GSOsuppressing action of DSX F in females, or a combination of these two possibilities. When we compared control males and females to dsx null flies, we found evidence for each of these three possibilities amongst the foreleg tarsal segments (Fig. 1E). In T1, dsx null males and dsx null females had an average number of GSOs (7.360.2 SEM and 7.260.2, respectively) that did not significantly differ from the number of GSOs found in dsx + control females (6.960.1), but were significantly less than the number of GSOs present in dsx + males (10.560.3). Similarly in T3, dsx null males and dsx null females had an average number of GSOs (4.260.1 and 4.360.1, respectively) that did not significantly differ from the number of GSOs in dsx + control females (4.060.0), but were significantly less than the number of GSOs present in dsx + males (6.060.0). Thus in T1 and T3, DSX M induces the male-specific number of GSOs, whereas DSX F appears to have no effect on the number of GSOs. In T4, the number of GSOs in both dsx null males and dsx null females (6.260. and 5.860.1, respectively) were roughly equal to the number present in dsx + control males (6.360.6) and higher than the number in dsx + control females (4.460.1). These data indicate that in T4, DSX F represses the development of two GSOs, while DSX M has no effect. In T2, dsx null males and females had numbers of GSOs (5.360.2 and 5.260.1, respectively) that differed significantly from both control males (7.360.3) and females (4.460.1). Thus in T2, DSX M promotes GSO formation in males, while DSX F represses GSO formation in females.
We conclude that the wild-type function of dsx ensures that each of the T1-T4 tarsal segments elaborates a greater number of GSOs in males than in females, but that this shared outcome arises by differential use of DSX M and DSX F in the tarsal segments. This distinction implies that positional information is integrated with sexual identity (i.e. dsx expression) on a fine-scale along the proximodistal axis of the developing foreleg.
The sexually dimorphic number of GSOs is established by 8 h APF. To address how dsx function specifies the sexspecific number of foreleg GSOs, we sought to determine when this sexual dimorphism was first apparent. We found that GSOs were identifiable by morphology and cell clustering as early as 14 h APF using the poxn-GAL4 marker and that the sexual dimorphism in GSO number was already apparent (C. Robinett, unpublished data). Examination of earlier time points was complicated by the expression of poxn-GAL4 across the leg disc epithelium of T2 and T4 from 0-6 h APF (Figs. 2 and S3), regions where poxn is required for formation of the intertarsal joints [61]. Because cells of the GSOs could not be clearly distinguished from epithelial cells using poxn-GAL4 alone, we incorporated staining with the 22C10 monoclonal antibody, which marks both neurons and other cell types of nascent sensory organs [62] and thus allows us to positively identify the GSOs.
In 8 h APF foreleg discs, GSOs could be recognized as multicell clusters that exhibited strong signals for both GFP and 22C10 (see Materials and Methods) (Figs. 2A,B and S4). There were also single large cells and pairs of large cells exhibiting both GFP and 22C10 signals at strengths similar to what was observed for cells within the GSO cell clusters (Figs. 2A,B, S4 and S5). Given that 22C10 marks nascent sensory organ cells, we consider these large cells to be the single sensory organ precursor cells (SOPs), which each give rise to a single GSO, or to the first pair of daughters of an SOP, respectively. As both of these cell types represent the GSO lineage, they were equated to GSOs in our analysis. We counted GSOs in male and female tarsal segments T2-T4 and found that the numbers were the same as seen in adults (Fig. 2C, compare to Fig. 1E). (T1 was not examined, see Materials and Methods). Therefore, the sexual dimorphism in foreleg GSOs number is established prior to 8 h APF. Because a subset of gustatory SOPs appeared to be undergoing their first cell division around 8 h APF, the developmental mechanism that establishes the sexually dimorphic number of GSOs likely acts prior to the division of gustatory SOPs.
DSX is present in proneural clusters and SOPs. It has been reported that the foreleg gustatory SOPs are specified before or near the time of puparium formation [52]. We therefore examined the distribution of DSX protein in T2-T4 of the foreleg disc before and at 0 h APF using anti-DSX DBD , a monoclonal antibody that recognizes both DSX M and DSX F (see Materials and Methods and Figs. S1 and S2). At this stage of development, foreleg discs of male and female larvae exhibited DSX proteins in a crescent of epithelial cells that occupied the T1 domain as well as in smaller patches of epithelial cells in T2-T4 (Fig. S2A). DSX was not detected across the T5 disc epithelium or in regions of the foreleg disc that were proximal to T1, and there was no staining of the disc epithelium in other leg discs at this stage (Fig. S2D,F,G). This distribution of DSX is in accord with the expression pattern of the dsx promoter [11]. Although the distribution of DSX appeared to be the same in male and female foreleg discs, males showed stronger signal intensity (Fig. S2A,B).
We note two important features in the distribution of DSX. First, the number of epithelial cells marked by DSX M is far greater than the number of GSOs in the adult foreleg, indicating that dsx expression is not restricted to the SOPs that will give rise to the GSOs. Second, the distribution of DSX in broad swaths of epithelial cells is restricted to those leg tissues that produce sexually dimorphic GSO numbers, as foreleg segment T5 and the tarsus of the second leg exhibit sexually monomorphic bristles [43,59,63]. These data thus suggest that dsx may regulate GSO numbers by exerting its function broadly across the epithelium of tarsal segments T2-T4 at a time preceding SOP specification.
In the wing imaginal disc, specification of the thoracic mechanosensory organ SOPs depends on the expression of the proneural genes achaete (ac) and scute (sc), which confer neural potential to patches of epithelial cells called proneural clusters from which the SOP cell is selected (reviewed in [64,65,66]). To ascertain whether dsx might function in proneural clusters of the foreleg disc, we determined the pattern of DSX M with respect to Achaete (AC) in foreleg discs of male late third instar larvae. Nascent third instar larvae (0 h 3I) were isolated, raised at 25uC, and then analyzed at 4-hour increments from 24 h 3I until the time of puparium formation at 48 h 3I, which is equivalent to 0 h APF. At 24 h 3I, AC was detected only in a few cells of T5, while DSX M was absent from the disc (Fig. S7A). AC-positive cells were observed at a greater number in T5 at 28 h 3I (Fig. S7B), as well as in more proximal domains at 32 h 3I (Fig. S7C). However, DSX M was not detected in the foreleg disc until 36 h 3I, when it weakly marked a crescent of epithelial cells in T1 (Fig. 3A). This staining signal intensified at 40 h 3I, and by 44 h 3I DSX M appeared in swaths of epithelial cells in the more distal tarsal segments, T2-T4 (Fig. 3B,C). The pattern of AC also became more complex during 36-44 h 3I as multiple patches of AC-positive cells were distributed across the proximal and distal regions of the foreleg disc ( Fig. 3A-C). Importantly, there was significant overlap between DSX M and AC in patches of cells distal to T1 at 44 h 3I, and this pattern became more pronounced in T2-T4 at 48 h 3I (Fig. 3C,D). Because AC marks cells with proneural potential, we consider these co-expressing patches of cells to be nascent proneural clusters. Further, because leg mechanosensory sense organs are not specified until around 8 h APF [51,52], we assume that these proneural clusters will give rise to gustatory SOPs.
Once an SOP is specified within the proneural cluster, it accumulates high levels of AC and Scute (SC) and concomitantly initiates expression of the neural precursor gene asense (ase), which in the embryonic nervous system and wing imaginal disc promotes the SOP fate and ensures proper development of sensory organs [67,68,69,70]. To determine if DSX M is present early in cells of the GSO lineage following SOP specification, we examined the distribution of DSX M with respect to ase-lacZ, a cytoplasmic reporter of ase expression that is expressed in SOPs and their first daughter cells [68]. In male foreleg discs at 0 h APF, DSX M was present in many of the disc epithelial cells within T1-T4, as well as in the previously mentioned clusters of cells in T5 (Fig. 4, see also Fig. 2). In all tarsal segments, ase-lacZ was expressed in distinct clusters of cells that generally resembled the pattern seen with poxn-GAL4 at later time-points, and some of these cells also contained DSX M (Fig. 4A,B), as indicated by the nuclear-localized DSX M signal surrounded by cytoplasmic b-galactosidase immunoreactivity. In T5, the cells containing DSX M were part of a cluster of cells that expressed ase-lacZ (Fig. 4A). In T4, DSX M was seen in a pair of ase-lacZ cells whose cytoplasm occupied a larger volume than the surrounding epithelial cells (Fig. 4B). That the ase-lacZ cytoplasmic staining corresponded to only two cells was confirmed by nuclear staining (Fig. S6). This pair of tightly associated cells expressing ase-lacZ is assumed to be the immediate daughters of a recently divided SOP. Thus, DSX M is present in the immediate progeny of at least a subset of SOPs in the foreleg tarsal segments.
We also examined the distribution of DSX M with respect to neuralized-lacZ (neur-lacZ), a marker of SOPs and their progeny [71,72], at the later time-point of 6 h APF. DSX M was present in patches of cells on the anterior surface of T1-T4, with broader expression in cells of distal T1 (Fig. 4C,D). Across T1-T4, DSX M colocalized with the expression of neur-lacZ in a few large cells that are likely to be gustatory SOPs (Fig. 4C,D), which is consistent with the presence of dividing poxn-GAL4-expressing cells that were positive for 22C10 at 6 h APF (Fig. S8). Whether the presence of DSX M in a subset of gustatory SOPS results from either continued expression of dsx in these cells or perdurance of the protein from  expression in cells of the proneural clusters, we conclude that dsx is expressed at a time and place that would allow it to regulate the specification of the gustatory SOP fate.

DSX F can Repress Midline Crossing by GRN Axons Independent of its Regulation of Neurogenesis
We previously showed that DSX F in females represses VNC midline crossing by GRN axonal projections [46]. Two possibilities were proposed to account for this phenomenon: 1) DSX F regulates axon guidance in foreleg GRNs to prevent midline crossing in females; or 2) DSX F prevents midline crossing by preventing the birth of male-specific neurons that would send axonal projections across the VNC midline. If DSX F regulates axon guidance independent of its role in regulating GSO formation, then it should be possible to perturb dsx function in such a way as to preserve the wild-type number of male-specific GSOs but alter the axonal morphology of their GRNs. Alternatively, if DSX F acts only to regulate GSO numbers, then perturbing dsx function in males after GSO number has been specified should not affect GRN axonal morphology.
We tested these hypotheses by expressing DSX F in nascent male GRNs during axonal development. Specifically, we drove UAS-DSX F with fru GAL4 [49] which initiates expression in postmitotic GRNs before their nascent axons encounter the VNC midline (D. Mellert, unpublished data). To assess behavior of GRN axons at the VNC midline, fru GAL4 expression was visualized by simultaneously driving UAS-mCD8::GFP while the sensory neuropil was highlighted by counterstaining DN-cadherin [73]. Surprisingly, DN-cadherin staining alone was sufficient to assess midline crossing (see Fig. 5D',E',F'), and corroborated the findings for fru GAL4 -driven expression of UAS-mCD8::GFP. We crossed w; UAS-mCD8::GFP; fru GAL4 females to w; UAS-DSX F /SM6 males and examined three classes of progeny: control males and females that carried the SM6 chromosome, and males that expressed UAS-DSX F under the control of fru GAL4 . As expected, control males ( Fig. 5A) had more fru GAL4 -expressing GRNs in their forelegs than control females (Fig. 5C), and only in control males were GRNs observed to have crossed the VNC midline (Fig. 5D, F). In contrast, when DSX F was expressed in the fru GAL4 -expressing GRNs, midline crossing was eliminated (Fig. 5E), even though the number of GSOs appeared the same (Fig. 5B). This indicates that the roles of DSX F are temporally separable during development of the female foreleg: DSX F first acts early to regulate neurogenesis then acts in the terminally differentiated neuron to regulate axon guidance. Thus, dsx functions in two distinct developmental contexts within the GSO lineage.

Discussion
We report that dsx regulates the sexually dimorphic number of GSOs across all tarsal segments of the foreleg: DSX M promotes and DSX F represses the development of certain GSOs. The effects of this regulation are apparent by 8 h APF, when the GSOs are first identified, and the spatiotemporal pattern of DSX implies that dsx determines the number of gustatory SOPs. dsx exhibits a surprising degree of context sensitivity: the relative importance of DSX M and DSX F varies along the proximodistal axis of the foreleg and, during the course of GSO development, DSX F progresses from regulating cell fate to regulating axon guidance.
Given that dsx controls the formation of the other sexually dimorphic cuticular structures of the fly, as well as the number of GSOs in segment T1 of the foreleg [44], we anticipated that dsx would regulate the sex-specific GSO numbers in segments T2-T4 of the foreleg. However, the manner in which this regulation is achieved across the tarsal segments was surprising. Although each of the T1-T4 foreleg tarsal segments produces more GSOs in males than in females, in two segments this difference is achieved by promoting formation of several GSOs in males (via the action of DSX M ), in one segment it is achieved by repressing the formation of several GSOs in females (via the action of DSX F ), and in another segment, both DSX M and DSX F act to regulate GSO number. This is more complicated than the simpler a priori expectation that the function of dsx would be the same across the T1-T4 foreleg segments.
That DSX M and DSX F can be utilized differentially has been previously established. In the fat body, female-specific expression of Yp1 and Yp2 depends on up-regulation by DSX F in females and down-regulation by DSX M in males [8,10,74]. Thus, in dsx null flies, both sexes express these genes at equivalent levels. Similarly, DSX F activates and DSX M represses expression at the bric-a-brac locus to generate sex-specific pigmentation in the abdominal epithelium [75]. In these two cases, both DSX proteins contribute to regulation of a single trait, similar to the regulation of GSO number in T2. In contrast, desatF is activated by DSX F in oenocytes to produce female-specific pheromones without influence from DSX M [31]. This single isoform-mediated regulation bears similarity to the regulation of foreleg GSOs in T1, T3 and T4. Whereas the previous studies found that DSX M or DSX F were differentially utilized to sculpt sexually dimorphic traits arising from developmentally distinct tissues, we have found that these transcription factors can be differentially utilized across a single developmental field-the epithelium of the foreleg disc. Moreover, the differential roles of DSX M and DSX F in different tarsal segments suggest that each segment may have independently evolved a molecular mechanism for integrating sexual and proximodistal axis information within the foreleg disc to produce more GSOs in the male.
We also sought to determine when the function of dsx impacts neurogenesis to generate the numbers of GSOs. Although the details of foreleg GSO development have not been specifically reported, studies of the mechanosensory macrochaete lineages of the notum provide a basic framework for the multi-step process of sensory organ neurogenesis (reviewed in [64,65,76,77]). The initiating event is patterned expression of the proneural genes ac and sc, which imparts the potential to produce SOPs to specific clusters of epithelial cells across the disc epithelium. Subsequent cell-cell interactions within the cluster typically specify a single SOP (see also [78,79,80]). The nascent SOP must then sustain its fate and undergo a series of stereotyped cell divisions to produce all of the cells of the sensory organ (reviewed in [76,77]). Any of the molecular processes that underlie these stages could be influenced by the functions of dsx.
We were struck by the broad distribution of the DSX proteins across the T1-T4 foreleg disc epithelium before and at 0 h APF, a time when the gustatory SOPs are specified [52]. Because the number of DSX-positive cells far exceeded the number of gustatory SOPs necessary to give rise to the GSOs, we infer that dsx is acting prior to or during SOP formation. This is consistent with the frequent colocalization of DSX with AC in proneural clusters, which suggests that dsx might act within these cells to determine whether the SOP fate is promoted in males or repressed in females. The broad distribution of the DSX proteins could ensure that sexual information is available for integration with positional information across the foreleg disc epithelium to guide sexually dimorphic development.
In contrast to T2-T4, DSX was not apparent in the foreleg epithelium of T5, which produces a sexually monomorphic GSO number. Thus, the presence of DSX in the epithelium correlates with the adult sexual dimorphism in GSO number, consistent with the notion that dsx is expressed at the right time and place to impact SOP selection in the foreleg. However, at 0 h APF we observed DSX in two nascent sensory organs expressing ase-lacZ (Fig. 4A, and C. Robinett, unpublished data) [67,68]. We speculate that these sensory organs correspond to the GSOs containing GRNs that express pickpocket 25 (ppk25), which is enriched in males and required for their normal response to female pheromones [81,82]. Thus, the presence of DSX in the nascent GSOs may forecast sexually dimorphic gene expression in the adult GSO.
In addition to specifying foreleg GSO numbers, we observed a temporally distinct function for dsx in the GRNs. During pupal development, GRN axons project proximally along the leg nerves and into the VNC, and here the behavior of the axon depends on the activities of FRU M or DSX F . In males, FRU M promotes crossing of the VNC midline by the axons, but in females, DSX F represses this behavior [46]. This dual regulation causes GRN axons to project across the VNC midline only in males [45]. We previously proposed two competing hypotheses to explain the action of DSX F : 1) DSX F directly affects axon guidance in differentiating GRNs; or 2) only male-specific GRNs are competent to cross the VNC midline and DSX F indirectly affects midline crossing by repressing formation of the male-specific GSOs. Having shown that post-mitotic expression of DSX F in Fru M -expressing GRNs subsequent to the establishment of GSO number prevents midline crossing, we now reject the second hypothesis. Moreover, the early sexual information that impacts GSO number does not irreversibly determine sex-specific development of the GRNs as they continue to be sensitive to the action of DSX F (and presumably FRU M ). Because dsx and fru are classically thought of as acting in parallel, we were intrigued to find both genes regulating the same phenotype in a common set of GRNs. Determining whether they coregulate a common set of target genes or independently regulate distinct targets will be of great interest.
Because DSX M and DSX F differentially impact GSO numbers in different tarsal segments, and DSX F regulates the later process of axon guidance, the identity of the genes directly regulated by dsx during foreleg development likely changes with spatiotemporal context. Although we currently do not know which genes are directly regulated by dsx in the foreleg epithelium or the GSO lineage, the available data on in vivo DSX binding sites [41] may reveal genes that are known to be involved in peripheral neurogenesis or axon guidance. The challenge will then be to determine if such candidates exhibit sexually dimorphic expression in the different tarsal segments at different developmental time points. In this way, development of the foreleg GSOs presents a unique opportunity for investigating how dsx function is integrated with spatiotemporal context across a changing developmental landscape.  DAPI (blue). Tarsal segments boundaries are indicated with light blue lines in left panels. Cells marked with 22C10 were classified based on both colocalization of poxn-GAL4 and morphology of the cells or cell clusters: GSO lineage cells (magenta arrows); non-GSO cells that lack poxn-GAL4 in T1 and T3 (dark blue arrows); non-GSO cells marked by poxn-GAL4 but lacking GSO morphology in T2 and T4 (light blue arrows). In panel (A), the row of 22C10-positive cells (bracket) in T1 are likely the sex comb SOPs. Scale bars, 50 mm.