Sex determination genes control the development of the Drosophila genital disc, modulating the response to Hedgehog, Wingless and Decapentaplegic signals

In Drosophila melanogaster, regulation of the sex determination genes throughout development occurs by sex-specific splicing of the pre-mRNA of the corresponding downstream gene in the genetic cascade (reviewed by Cline and Meyer, 1996). The first gene in this cascade is Sex-lethal (Sxl). The Sxl protein participates in the female-specific splicing of its own pre-mRNA, resulting in the production of Sxl protein in females but not in males. This protein, in turn, determines the female-specific splicing of the transformer (tra) pre-mRNA, such that functional Tra protein is present only in females. This Tra protein, together with the Tra2 protein from the constitutive gene transformer 2 (tra2), directs the splicing of the doublesex (dsx) pre-mRNA into the female mode, giving rise to the female Dsx protein (Dsx^F) that promotes female sexual development. In males, where no functional Tra protein is available, no Tra/Tra2 complex is formed and the dsx pre-mRNA follows the male mode of splicing, which produces male Dsx protein (Dsx^M). This protein promotes male development (Burtis and Baker, 1989; Hoshijima et al., 1991). The genital disc is of complex organization and bilateral symmetry, and gives rise to the terminalia, which show extreme sexual dimorphism. The terminalia comprise the entire set of internal and external genital and anal structures, with the exception of the gonads (Bryant, 1978). In the embryo, the anlage of the genital disc of both sexes consists of three primordia: the female genital primordium (FGP), located anteriorly, the anal primordium, located posteriorly, and the male genital primordium (MGP), found between these (Nöthiger et al., 1977; Schüpbach et al., 1978). In both sexes, only two primordia develop to form the adult terminalia. The anal primordium grows in both sexes but, depending on the genetic sex, develops into either male or female analia. However, only one of the two genital primordia grows, again depending on the individual genetic sex (Nöthiger et al., 1977; Schüpbach et al., 1978; Epper and Nöthiger, 1982; Wieschaus and Nöthiger, 1982; P. C. Ehrensperger, PhD thesis, University of Zürich, Switzerland, 1983). The genital primordium that does not develop is denoted as the ‘repressed genital primordium’ (RMP and RFP respectively for the male and female genital primordia; Epper and Nöthiger, 1982) (see schemes in Figs 1A,B, 2A,B). The repressed genital primordium of the genital disc contains fewer cells than the corresponding developed genital primordium. It has been estimated by cell count, that the number of cells in the repressed primordium is about 10% of the number of cells in the corresponding developed primordium (Epper and Nöthiger, 1982; P. C. Ehrensperger, PhD thesis, University of Zürich, Switzerland, 1983). As the number of founder cells in both genital primordia is the same, these cells undergo more division in the developing than in the repressed state. Epper and Nöthiger induced clones homozygous for tra in the RMP of tra/+ female genital discs shortly before pupation, and dissected and transplanted the discs into second instar larvae to allow sufficient time for them to grow (Epper and Nöthiger, 1982). The presence of such clones indicated that cells in the RMP continued to divide until the end of the larval period. This suggests that the cells of this primordium show a lower proliferation rate.

The gene dsx controls which of the two genital primordia will develop and which will be repressed. In females, the presence of Dsx^F (female genetic signal) results in the development of the female genital primordium and repression of the male primordium. Conversely in males, the presence of Dsx^M (male genetic signal) results in the development and repression of the male and female genital primordia, respectively. In flies that either lack or have both Dsx^F and Dsx^M products, the male and female genital primordia develop incompletely, giving rise to intersexual flies that are characterized by the simultaneous presence of adult male and female genital structures (reviewed by Steinmann-Zwicky et al., 1990).

The three primordia that form the genital disc are composed of an anterior and a posterior compartment (Freeland and Kuhn, 1996; Casares et al., 1997; Chen and Baker, 1997). Analyses of the roles and the expression pattern of the genes hedgehog (hh), decapentaplegic (dpp), wingless (wg) (Chen and Baker, 1997; Sánchez et al., 1997) and Distal-less (Dll) (Gorfinkiel et al., 1999) in the genital disc have shown that its organization is similar to that of the antenna and leg discs. The dpp and wg expression domains are mostly complementary and abut the expression domain of engrailed (en). Furthermore, the functional relationships between these genes appear to be the same as in the other ventral discs: dpp and wg are repressed by patched (ptc), which in turn is repressed by the Hh protein secreted by cells expressing en (reviewed in Brook et al., 1996).

It has been suggested that Dpp and Wg play similar roles in the development of the genital disc. In contrast with the regulation of these signals in other imaginal discs, it has been proposed that in the genital discs, the expression of both dpp and wg is additionally controlled by the sex determination genes (Sánchez et al., 1997). We have found that the gene dsx controls the development of both male and female genitalia through Hh, Dpp and Wg morphogenetic signals, and that this control is exerted at different levels in each primordium of the male and female genital disc. In addition, it is shown that Dsx products play positive and negative roles in the development of the genital primordia. Furthermore, the time requirement of Dsx^M and Dsx^F products for the development of the male and female genitalia was analyzed. The question addressed is whether these products are continuously required for the development and differentiation of sex specific structures. The results presented show that males and females have different time requirements for Dsx^M and Dsx^F proteins, respectively. It is also shown that this different time requirement is related to the specific induction time of Dpp in the male primordium and by the control of the response to Wg signaling in the female primordium.

The reporter gene dpp-lacZ (Blackman et al., 1991) is expressed as the endogenous RNA. The heat-shock flipase used was FLP122 (Struhl and Basler, 1993). For the description of the dsx¹, dsx¹¹, dsx^Mas and tra2^ts1 mutant alleles see Lindsley and Zimm (1992). (The FlyBase entry of mutations can be found at http://gin.ebi.ac.uk:7081.) For ectopic expression of Dsx^F, the UAS-dsxF (K. C. Burtis, personal communication) was expressed under the control of the C68a-GAL4 line (Brand and Perrimon, 1993) that yields a general expression pattern in the genital disc.

Flies used for the analysis of external terminalia of adults were kept in a mixture of ethanol:glycerol (3:1) for several days. They were then macerated in 10% HOK at 60ºC for 15 minutes, thoroughly washed with H2O, and mounted in Faure´s solution for inspection under a compound microscope.

Clones of mutant cells were generated by FLP-mediated mitotic recombination (Golic, 1991), using heat shock to induce a transient pulse of FLP recombinase at chosen stages of development.

Flies of the following genotypes were subjected to heat shock 0-120 hours after egg laying (AEL) at 37^ºC for 60-90 minutes. For tra2^ts1, y hsFLP122/y; FRT42D cn tra2^ts1/FRT42D Myc; dpp-lacZ/+, larvae were incubated at the restrictive temperature (29ºC) after clone induction for the rest of development. At this temperature, the Tra2 protein is not functional and consequently the cells have the male-specific Dsx^M product. For dsx¹¹ clones in which the Dsx^M product is specifically affected, y hsFLP122/y; FRT82B dsx¹¹/FRT82B FRT82y⁺ flies were analyzed. In all cases, the clones were genetically marked in the genital disc by the absence of Myc, or arm-lacZ staining in the disc. Each recombination event renders twin clones: one clone is homozygous for tra2 or dsx¹¹ and lacks the Myc or the arm-lacZ genes, whereas its partner is wild type for tra2 or dsx¹¹ and homozygous for the markers.

Immunofluorescence staining were performed as described previously (Capdevila et al., 1994). The antibodies used were anti-Omb (Grimm and Pflugfelder, 1996), anti-Wg (Brook and Cohen, 1996), anti-Dac (Mardon et al., 1994), anti-Myc 9E10 (Babco, Berkeley Antibody Company), anti-Sxl (Bopp et al., 1993) and anti-β -gal (Jackson Laboratories). Imaginal discs were examined under a Zeiss Laser Scan microscope.

dpp is expressed in the growing male genital primordium of male genital discs but not in the RMP of female genital discs (Freeland and Kuhn, 1996; Casares et al., 1997; see Figs 1C, 2C). This suggests that the developing or repressed status of the male genital primordium is determined by the regulation of dpp expression. As dsx controls the developmental status of the male genital primordium, and the expression of dpp depends on the Hh signal, we analyzed the relationship between the Hh signal cascade and dsx in the control of RMP development. To this end, a twin clonal analysis for the loss-of-function tra2 mutation was performed in tra2/+ female genital discs. In this way, we analyzed the proliferation and the induction of dpp expression in the clones homozygous for tra2 (male genetic constitution) and that of the twin wild-type clones, both in the repressed male and the growing female primordia (see Materials and Methods). Recall that the effects of tra2 in the genital disc are entirely mediated by its role in the splicing of dsx RNA: the presence or absence of functional Tra2 product gives rise to the production of female Dsx^F or male Dsx^M product, respectively (see Introduction). Clones for tra2 (expressing Dsx^M) induced in the RMP of female genital discs showed overgrowth and were always associated with dpp expression (Fig. 1E), indicating that the lower proliferation shown by the RMP is probably caused by the absence of dpp expression. This activation of dpp was restricted to only certain parts of the clone and never overlapped with Wg expression. As wg is normally expressed in the RMP, the possibility exists that the cells that did not express dpp in the clone were expressing wg, owing to their antagonistic interaction. Double staining of Wg and Dpp in tra2 clones revealed an expansion of the normal domain of wg expression that abuts the dpp-expressing cells (data not shown). In the RMP, the two sister clones showed different size: the tra2 clone (male genetic constitution) was bigger than the wild-type twin clone (female genetic constitution). In contrast, when the clones were induced in the growing female genital primordium, both of them were of a similar size. Moreover, the pattern of dpp expression did not change in the tra2 cells induced in this primordium (Fig. 1E). Previous observations have indicated that optomotor-blind (omb; bi – FlyBase), a target of the Dpp pathway, also responds to Dpp in the genital disc (Gorfinkiel et al., 1999; and see Fig. 1D). Since dpp was de-repressed in tra2 clones induced in the RMP, the activation of omb was monitored in these clones. The activation of dpp in tra2 clones induced the expression of this target gene (Fig. 1F), whose function is required for the development of specific male genital structures (Gorfinkiel et al., 1999; L. G., N. G. and I. G., unpublished). It is concluded that Dsx^F product prevents the induction of Dpp by Hh in the repressed male genital primordium of female genital discs.

In the male genital disc, which has Dsx^M product, the low proliferation rate of the RFP cannot be attributed to a lack of dpp or wg, as both genes are expressed in this primordium (Fig. 2C). Failure to respond to the Dpp signal may also be ruled out because the RFP expresses the Dpp downstream gene, omb (Fig. 2E), indicating that the Dpp pathway is active in this primordium. However, Dll, a target gene for both Wg and Dpp, is not expressed in the RFP but is expressed in the developing female primordium of female genital discs (Fig. 2D). This suggests that the Wg pathway cannot activate some of its targets in the RFP. Thus, the analysis of dsx¹ mutant genital discs, where both male and female genital primordia develop, becomes relevant. These mutant discs show neither Dsx^M nor Dsx^F products. The female genital primordium of these discs now expresses Dll (Fig. 2F). It is concluded that Dsx^M controls the response to the Wg pathway in the RFP of male genital discs.

The gene dachsund (dac) (Mardon et al., 1994) is also a target of the Hh pathway in the leg and antenna (Lecuit and Cohen, 1997). In the present study, it was found that dac is differentially expressed in female and male genital discs. In the female genital discs, which have Dsx^F product, dac expression mostly coincided with that of wg in both the growing female primordium and the RMP (Fig. 3A). In contrast, in male genital discs, which have Dsx^M product, dac was not similarly expressed to wg but its expression partially overlapped that of dpp and no expression was observed in the RFP (Fig. 3B). In pkA⁻ clones, which autonomously activate Wg and Dpp signals in a complementary pattern, dac was ectopically expressed only in mutant pkA⁻ cells at or close to the normal dac expression domains in male and female genital discs (data not shown). In pkA⁻dpp⁻ double clones, which express wg, dac was not ectopically induced in the male primordium of the male genital disc, but was still ectopically induced in both the growing female genital primordium and the RMP of female genital disc (Fig. 3C). Conversely, in pkA⁻wg⁻ double clones, which express dpp, dac was not ectopically induced in the growing female or in the RMP of female genital discs, but was ectopically induced in the growing male primordium of the male genital disc (Fig. 3D). These results indicate that dac respond differently to Wg and Dpp signals in both sexes.

In dsx^Mas/+ intersexual genital discs (Fig. 3E), which have both Dsx^M and Dsx^F products, and in dsx¹ intersexual genital discs (Fig. 3F), which have neither Dsx^M nor Dsx^F products, dac was expressed in Wg and Dpp domains although at lower levels than in normal male and female genital discs. These results suggest that Dsx^M plays opposing, positive and negative roles in dac expression in male and female genital discs, respectively; and that Dsx^F plays opposing, positive and negative roles in dac expression in female and male genital discs, respectively. To test this hypothesis, tra2 clones (which express only Dsx^M) were induced in female genital discs. The expression of dac was repressed in tra2 clones located in Wg territory (Fig. 3G, clones 2, 3 and 4). Therefore, Dsx^F positively regulates dac expression in the Wg domain, and Dsx^M negatively regulates dac expression in this domain, otherwise dac would be expressed in tra2 clones at the low levels found in dsx intersexual genital discs. However, when the tra2 clones were induced in the RMP, in the territory competent to activate dpp, they showed ectopic expression of dac (Fig. 3G, clone 1). Therefore, Dsx^M positively regulates dac expression in the Dpp domain, whereas Dsx^F negatively regulates dac expression in this domain, as in normal female genital discs with Dsx^Fdac is not expressed in Dpp territory. This is further supported by the induction of dac in the Wg domain and repression of dac in the Dpp domain by ectopic expression of Dsx^F in the male genital primordium of male genital discs (Fig. 3H). We conclude that in male genital discs, Dsx^M positively and negatively regulates dac expression in Dpp and Wg domains, respectively; and in female genital discs, Dsx^F positively and negatively regulates dac expression in Wg and Dpp domains, respectively.

Homozygous tra2^ts larvae with two X-chromosomes develop into female or male adults if reared at 18ºC or 29ºC, respectively, because at 18ºC they produce Dsx^F and at 29ºC they produce Dsx^M. A shift in the temperature of the culture is accompanied by a change in the sexual pathway of tra2ts larvae (Belote and Baker, 1982; Epper and Bryant, 1983; Sánchez and Granadino, 1992). Analysis of the growth of genital primordia and their capacity to differentiate adult structures of tra2^ts flies was performed using pulses between the male- and the female-determining temperatures in both directions during development.

Regardless of the stage in development at which the female-determining temperature pulse was given (transitory presence of functional Tra2^ts product; i.e. transitory presence of Dsx^F product and absence of Dsx^M product), the male genital disc developed normal male adult genital structures and not female ones (Table 1A, Fig. 4C; for comparison of external male and female wild-type genital structures, see Fig. 4A and 4B, respectively). This occurred even if the pulse was applied during pupation (data not shown). Pulses of 24 hours at the male-determining temperature (temporal absence of functional Tra2^ts product; i.e. transitory absence of Dsx^F product and presence of Dsx^M product) before the end of first larval stage produced female and not male genital structures (Table 1B). However, later pulses always gave rise to male genital structures (Table 1B, Fig. 4D), except when close to pupation. Further, the capacity of the female genital disc to differentiate adult genital structures was also reduced when the temperature pulse was applied during metamorphosis (data not shown).

When the effect of the male-determining temperature pulses was analyzed in the genital disc, it was found that overgrowth of the RMP was always associated with the activation of dpp in this primordium. However, this activation and the associated overgrowth only occurred when the temperature pulse was given after the end of first larval instar (see Fig. 5A-C). This suggests that there is a time requirement for induction of dpp. The activation of this gene in the RMP and the cell proliferation resumed by this primordium, as well as its capacity to differentiate adult structures was irreversible, as they were maintained when the larvae were returned to the female-determining temperature, which is when functional Tra2^ts product is again available (i.e. presence of Dsx^F product and absence of Dsx^M product; see Table 1B).

This time requirement for induction of dpp was also supported by the fact that dsx¹¹ clones (which lack Dsx^M) induced differentiated normal male adult genital structures in the developing male genital primordium of XY; dsx¹¹/+ male genital discs (which express only Dsx^M) after 24 hours of development (Fig. 4E). However, when the dsx¹¹ clones were induced in the time period between 0 and 24 hours of development, they did not differentiate normally and gave rise to incomplete adult male genital structures (Fig. 4F). This different developmental capacity shown by the dsx¹¹ clones depending on their induction time is explained as follows. When the clones were induced after 24 hours of development, dpp was already activated. Indeed, these clones showed no change in the expression pattern of dpp or their targets (data not shown). Accordingly, these clones displayed normal proliferation and capacity to differentiate male adult genital structures. However, when the clones were induced early in development, dpp was not yet activated, as this gene is not expressed in the male genital primordium of male genital discs early in development (Casares et al., 1997). Therefore,when the male genital disc reaches the state in development when dpp is induced, the cells that form the clones activate this gene as in dsx mutant intersexual flies because the clones have neither Dsx^M nor Dsx^F products. Consequently, these clones do not achieve a normal proliferation rate, and then do not differentiate normal adult male genital structures.

It was shown above that dsx regulates the expression of gene dac. Recall that in male genital discs, Dsx^M positively and negatively regulates dac expression in Dpp and Wg domains, respectively; and in female genital discs, Dsx^F positively and negatively regulates dac expression in Wg and Dpp domains, respectively. The expression of the gene dac was analyzed in genital discs of tra2^ts flies using pulses between the male- and the female-determining temperatures in both directions. It was found that the dac expression pattern switches from a ‘female type’ to a ‘male type’ when male-determining temperature pulses were applied to tra2^ts larvae after first larval instar (Fig. 5D,D´,E,F). Note that dac expression is reduced in the Wg domain of the RMP and is progressively activated in the Dpp domain. It should be remembered that these pulses lead to the transient presence of Dsx^M instead of Dsx^F product. Thus, these results are consistent with the previously proposed suggestion that Dsx^M activates dac in the Dpp domain and represses it in the Wg domain (again the converse is true for Dsx^F). When the pulse was given during first larval instar, dac was not activated in the Dpp domain of RMP (Fig. 5D´), in spite of the fact that there is also a transient presence of Dsx^M instead of Dsx^F. This is explained by the lack of competence of cells to express Dpp, which is acquired after first larval instar (Casares et al., 1997). When the tra2^ts larvae reach such a developmental stage, these cells now produce Dsx^F because they have returned to the female-determining temperature. Dsx^F prevents activation of dpp in the RMP, and consequently no induction of dac expression occurs. In the female genital primordium (Fig. 5D), dac expression is strongly reduced in the Wg domain and absent in the Dpp domain.

Taken together, these results suggest that the development of male and female genital primordia have different time requirements for Dsx^M and Dsx^F products.

Dpp and Wg control development of the genital disc in both sexes and act as morphogens, specifying different cell types of the final adult pattern (Chen and Baker, 1997; Sánchez et al., 1997). The results presented in this work indicate that the sex determination gene dsx controls the development of both male and female genitalia through these morphogenetic signals, but this control is implemented at different levels in the two primordia The developing or repressed status of the male genital primordium seems to be controlled at the level of dpp expression. The RMP of female genital discs, which have Dsx^F, does not express dpp (Freeland and Kuhn, 1996; Casares et al., 1997; Fig. 1C). However, male clones for tra2 (expressing Dsx^M) induced in the RMP of female genital discs showed overgrowth and were always associated with dpp expression. The lack of dpp expression in the RMP may not be attributed to failure of the cells to receive the Hh signal, as Hh induces expression of wg in these cells. Rather, the gene dsx renders cells of the male genital primordium competent to activate dpp. This is consistent with the ectopic dpp induced by Hh in the female genital primordium but not in the RMP of female genital discs. However, ectopic Hh was able to induce dpp expression in the male genital primordium of male genital discs (Sánchez et al., 1997). Thus, the cells of the male genital primordium were able to respond to the Hh signal in male (absence of Dsx^F) but not in female (presence of Dsx^F) genital discs.

The expression of dpp in male clones was always associated with overgrowth. Therefore, the lower proliferation shown by the RMP is likely caused by the absence of dpp expression. This agrees with the lack of dpp expression in the male genital primordium of male genital discs early in development and its expression in second instar larvae, which persists during the rest of larval development (Casares et al., 1997). During the first larval instar, the cells in the genital disc, as in other imaginal discs, are quiescent and resume proliferation in the transition to the second larval stage (Madhavan and Schneiderman, 1977). This proliferation is probably stimulated by the activation of dpp at this stage.

The low proliferation rate of RFP in male genital discs cannot be attributed to the lack of dpp or wg, as both genes are expressed; neither can a failure to respond to the Dpp signal, because the repressed female primordium of male genital discs expresses omb (Gorfinkiel et al., 1999). It is speculated here that this low proliferation is caused by blockage of Wg signaling pathway by Dsx^M. The gene Dll, a Wg and Dpp target in the genital disc (Gorfinkiel et al., 1999), is not expressed in the repressed female genital primordium of male genital discs, where there is Dsx^M product, but is induced in the growing female genital primordium of dsx⁻ mutant genital discs, which lack Dsx^M product.

It is concluded that Dsx^F controls the development of the male genital primordium by preventing the activation of dpp by the Hh signaling pathway and that Dsx^M controls the development of the female genital primordium by blocking the Wg pathway. At which step of these pathways the control is exerted and its molecular mechanism remain to be investigated.

The standard model for the development of the genital disc states that the male and female genital primordia, which are derived from two different cell populations, develop unless Dsx^F and Dsx^M, respectively, repress them (reviewed by Steinmann-Zwicky et al., 1990). This means that these two proteins play a negative rather than a positive role in the developmental control of the genital disc. Our results support the standard model but also raise some uncertainties that may not be explained by the model, such as the positive role of the Dsx products in the development of the genital disc.

The standard model was primarily based on the observation that intersexual dsx mutant flies (which have neither Dsx^M nor Dsx^F) differentiate male and female adult genital structures. However, the inventory of these structures is not complete. This cannot be attributed to mutual interference between both male and female genital primordia as they simultaneously develop, because gynandromorphs have been found with complete male and female adult genitalia (Nöthiger et al., 1977). Rather, it seems that in the dsx intersexual condition, both male and female genital primordia fail to develop completely. Indeed, the number of cells in the mature dsx intersexual genital disc does not add up to the complete male plus female genital disc complement (P. C. Ehrensperger, PhD thesis, University of Zürich, Switzerland, 1983). This suggests that in the absence of both Dsx^M and Dsx^F products, both male and female genital primordia proliferate, although incompletely. This might suggest a positive role for Dsx^M and Dsx^F in the development of male and female genital primordia, respectively. In fact, a positive role for Dsx^M has been observed in the development of the sex comb in males (Jurnish and Burtis, 1993).

The gene dac has been used in this work as a reporter for the control of Dsx^M and Dsx^F products on the response to Wg and Dpp pathways. This gene is regulated by Wg in the female genital disc and by Dpp in the male genital disc, suggesting that this different regulation requires also the function of gene dsx. In normal female genital discs, which have Dsx^F product, dac is expressed in the Wg domain but not in the Dpp domain. In normal male genital discs, which have Dsx^M product, dac is expressed in the Dpp domain but not in the Wg domain. In dsx mutant genital discs (which have neither Dsx^M nor Dsx^F products) dac shows low levels of expression in both Wg and Dpp territories. This agrees with a positive role for Dsx^M and Dsx^F on dac expression. The positive effect of Dsx^F product is supported by the lack of dac activation in tra2 clones (which do not express Dsx^F) induced in the Wg domain of female genital discs. Moreover, the dsx¹¹ clones (which do not express either Dsx^M or Dsx^F) induced in the Wg territory of male genital discs did not show dac expression (data not shown). The positive effect of Dsx^M, which is expressed in the tra2 clones, is supported by the ectopic activation of dac in these clones that is induced in the Dpp domain of the RMP of female genital discs. These results lead us to propose that the expression of dac in female genital discs depends on the concerted action of both Wg and Dsx^F; whereas in male genital discs it depends on the concerted action of Dpp and Dsx^M. Nevertheless, a negative role for these Dsx products is shown when both Dsx^M and Dsx^F are present in the same genital disc, such as in intersexual dsx^Mas/+ genital discs. It was observed in these intersexual discs that dac is expressed in Wg and Dpp territories, although at lower levels than in normal male and female genital discs. Thus, Dsx^M and Dsx^F show a dual role with respect to expression of dac. A similar role has been observed in the regulation of the yolk protein (YP) genes by these Dsx products: Dsx^M represses the YP genes in males, whereas Dsx^F activates the YP genes in females. In dsx⁻ intersexual condition, in which both Dsx products are lacking, the YP genes are expressed at a basal level (Bownes and Nöthiger, 1981; Coschigano and Wensink, 1993).

Other results presented here support the positive role for both Dsx products. dsx¹¹ clones, which have no Dsx^M and are induced in the developing male genital primordium before first larval instar, do not give rise to adult male genital structures. This suggests an early positive effect of Dsx^M on the development of male genital primordium. With respect to the female genital primordium, the continuous requirement of Dsx^F for differentiation of female genital structures observed in the tra2^ts temperature pulses also suggests a positive role for Dsx^F.

We have shown that Dsx^F is continuously required during female larval development for preventing activation of dpp in the repressed male genital primordium; otherwise, this gene becomes activated and, consequently, the cells of this primordium resume normal development. Once activated, dpp maintains its own expression and is no longer repressed by Dsx^F. The following results support this proposition. First, tra2 male clones can be induced in the male repressed primordium of female genital discs throughout larval development. These clones do not have Dsx^F and they express dpp. Consequently, these clones develop and differentiate adult male genital structures in a female terminalia. And second, in the temperature pulse experiment of tra2ts larvae, it was shown that once dpp was induced after the temperature pulse, its expression was maintained even if the genital disc was placed back at the female-determining temperature where functional Tra2 product (equivalent to presence of Dsx^F) was again available. In addition, these temperature pulse experiments indicate that Dsx^F is also required during pupation for female cytodifferentiation. Finally, they also support the idea above that there is a time requirement for the induction of dpp by gene dsx: the proliferation status of the male genital primordium in male genital discs is probably induced by the activation of dpp at the end of first beginning of second larval instar, which persists during the rest of larval development.

In males, Dsx^M seems not to be continuously required during larval development for blocking the Wg signaling pathway in the female genital primordium. Clones that are mutant for dsx¹¹ (they have no Dsx^M) induced in X/Y; dsx¹¹/+ males during or after the first larval instar do not differentiate female genital structures, in agreement with the results reported by Epper and Nöthiger (Epper and Nöthiger, 1982). Thus, the cells in the RFP of male genital discs seem to be irreversibly committed to the repressed state if Dsx^M is present before the end of first larval instar. The temperature pulse experiments show also that Dsx^M does not seem to be needed during pupation for male cytodifferentiation.

dsx controls which of the two genital primordia will develop and which will be repressed. Nevertheless, as it is expressed in each cell, another gene(s) is required to distinguish between the female and the male genitalia. The female genitalia develop from eighth abdominal segment and the male genitalia develop from ninth abdominal segment (Freeland and Kuhn, 1996; Casares et al., 1997). It is also known that Abdominal-B (Abd-B) is responsible for the specification of these posterior segments (reviewed by Duncan, 1996). We have previously proposed that the development of the male and female genitalia requires the concerted action of Abd-B and dsx, and that these two genes control proliferation of each genital primordium through the expression, either directly or indirectly, of dpp and wg (Sánchez et al., 1997). Abd-B produces two different proteins: Abd-Bm and Abd-Br (Casanova et al., 1986; DeLorenzi et al., 1988; Kuziora and McGinnis, 1988; Sánchez-Herrero and Crosby, 1988). Abd-Bm is present only in the female genital primordium, whereas Abd-Br is present only in the male genital primordium (Casares et al., 1997).

We propose here that Dsx^M and Dsx^F combine with Abd-Bm and Abd-Br to make up the signals that determine the dimorphic sexual development of the genital disc (see Fig. 6). In the absence of both Dsx^M and Dsx^F products (dsx intersexes), there is a basal expression of dpp and a basal functional level of the Wg signaling pathway in both male and female genital primordia. In females, the concerted signal made up of DsxF and Abd-Br cause repression of the development of the male genital primordium by preventing the expression of dpp, resulting in the RMP of female genital discs. In males, the concerted signal formed by Dsx^M and Abd-Bm represses the female genital primordium by blocking the Wg signaling pathway, giving rise to the RFP of male genital discs. It is further proposed that DsxM plus Abd-Br increase dpp expression in the male genital primordium of male genital discs, and that Dsx^F plus Abd-Bm enhance Wg signaling pathway function in the female genital primordium of female genital discs. A similar mechanism of modulation of Dpp and Wg responses has been described for the shaping of haltere development by Ultrabithorax (Weatherbee et al., 1998). Therefore, Dsx^M would play a positive and a negative role in male and female genital primordia, respectively, whereas Dsx^F would play a positive and a negative role in female and male genital primordia, respectively. This positive role of both Dsx products serves to explain the expression of dpp and the function of the Wg signaling pathway in growing male and female genital primordia, respectively, in dsx^Mas/+ intersexual flies, where both genital primordia simultaneously have Dsx^M and Dsx^F. Otherwise, dpp would not be expressed in the male genital primordium and the Wg signaling pathway would not be functional in the female genital primordium, as occurs in normal female and male genital discs. If so, this would mean that the two genital primordia of these intersexual genital discs would be kept in the repressed state and would not develop. Contrary to observations, this would result in a lack of male and female adult genital structures in these intersexes.

It has been shown that homothorax and extradenticle genes are involved in the control of the response to Dpp and Wg signals in the proximal part of the leg (GonzalezCrespo et al., 1998; Wu and Cohen, 1999). As these genes are strongly expressed in the repressed male and female primordia of the genital disc (Estrada and Sánchez-Herrero, 2001), it is proposed here that these two genes may form part of the integrated mechanism comprised by Dsx and Abd-B products for the regulation of the morphogenetic signaling response.

During the evolution of the Diptera there has been a tendency towards the fusion of the posterior segments into a single imaginal disc (Crampton, 1942; Matsuda, 1976). In primitive Diptera, such as Tipulidae, males and females still produce an eighth tergite and ninth tergite, respectively. Insects such as Musca and Calliphora, which are considered to represent an intermediate evolutionary step between Tipulidae and Drosophila, have two laterals and one single median genital disc (Dübendorfer 1971; Emmert 1972a; Emmert 1972b). The anlage of the lateral discs corresponds to segment eight and the anlage of the single median disc to the fusion of segments 9 to 11. In females, the lateral discs form the female genitalia, except the parovaria. The median disc develops the parovaria (ninth segment) and the female analia (segments 10-11). In males, the lateral discs produce a reduced eight tergite. The median disc develops the male genitalia (ninth segment) and the male analia (segments 10-11). A further level of fusion occurred in the Drosophila lineage, where segments 8 to 11 form a single genital disc. The model proposed here for the development of the genital disc of Drosophila can be applied to the above primitive dipteran species.

In vertebrates, Dmrt1, the dsx homolog, has been implicated in male gonad development and murine Dmrt1 seems to be required for multiple aspects of testis differentiation (Raymond et al., 2000). This functional similarity could imply a close evolutionary relationship between Dmrt1 and the Drosophila dsx gene. In the same evolutionary context, it has been reported that, in mammals, the signaling molecule Wnt4, one of the mammalian homologs of the Drosophila Wingless gene family, is crucial for female sexual development (Vainio et al., 1999). Although the relationship between sex determination genes and morphogenetic signals has not been found in mammals yet, the findings reported here suggest the possibility that similar signals might be used across species for implementation of sex differentiation.

The authors thank K. Burtis, S. Cohen, G. Mardon, R. Nöthiger, G. Pflugfelder and G. Struhl for stocks and antibodies. We also thank José Felix de Celis for comments on the manuscript. This work was financed by grants PB98-0680 to I. G. and PB98-0466 to L. S., by the D.G.I.C.Y.T. grant 08.9/0001/1998 to I. G., by the Comunidad Autónoma de Madrid and by an institutional grant from the Fundación Areces. N. G. was financially supported by a fellowship from the Spanish Ministerio de Educación y Cultura.