ABSTRACT
The hermaphrodite (her) gene is necessary for sexual differentiation in Drosophila. Our characterization of her’s zygotic function suggests that one set of female-specific terminal differentiation genes, the yolk protein (yp) genes, is transcriptionally activated by two separate pathways. One is a female-specific pathway, which is positively regulated by the female-specific doublesex protein (DSXF). The other is a non-sex-specific pathway, that is positively regulated by HER. The HER pathway is prevented from functioning in males by the action of the male-specific doublesex protein (DSXM). The HER and DSX pathways also function independently to control downstream target genes in the precursor cells that give rise to the vaginal teeth and dorsal anal plate of females, and the lateral anal plates of males. However, a female-specific pathway that is dependent on both DSXF and HER controls the female- specific differentiation of the foreleg bristles and tergites 5 and 6, and the male-specific differentiation of these tissues does not require the suppression of HER’s function by DSXM.
INTRODUCTION
A hierarchy of regulatory genes controls somatic sex determination and differentiation in Drosophila melanogaster (reviewed, for example, by Burtis, 1993; Burtis and Wolfner, 1992; Cline and Meyer, 1996; McKeown, 1992; Parkhurst and Meneely, 1994). There are two branches in the hierarchy downstream of the transformer (tra) gene. One branch contains the doublesex (dsx) gene (Baker and Ridge, 1980) and the other the fruitless (fru) gene (Ryner et al., 1996). dsx is required for all known aspects of somatic sexual differentiation outside of the CNS (reviewed by Burtis and Wolfner, 1992) as well as some aspects of sexual differentiation in the CNS (Taylor and Truman, 1992; Villella and Hall, 1996). The sex determination function of fru appears to be required for the sexual differentiation of only a small set of cells in the CNS concerned with male sexual behaviors (Gailey et al., 1991; Hall, 1994; Ito et al., 1996; Lawrence and Johnston, 1986; Ryner et al., 1996; Taylor et al., 1994).
At the bottom of the dsx branch of the somatic sex determination hierarchy are the dsx, hermaphrodite (her) (Pultz and Baker, 1995) and intersex (ix) (Chase and Baker, 1995) genes. dsx encodes sex-specific transcription factors (Burtis et al., 1991). The female-specific DSX protein (DSXF) acts together with the her (Pultz and Baker, 1995) and ix (Erdman et al., 1996) gene products to inhibit male differentiation and activate female differentiation in females and, conversely, the male-specific DSX protein (DSXM) acts to inhibit female differentiation and activate male differentiation in males (Jursnich and Burtis, 1993; Taylor and Truman, 1992; Villella and Hall, 1996; reviewed by Burtis, 1993; McKeown and Madigan, 1992). The best-characterized terminal differentiation genes that are sex-specifically regulated by the somatic sex determination hierarchy are the yolk protein (yp) genes yp1, yp2 and yp3. The transcription of yp1 and yp2 in fat body cells is directly activated by DSXF in females and inhibited by DSXM in males, through the binding of the DSX proteins to the fat-body-specific enhancer (FBE) of yp1 and yp2 (reviewed by Bownes, 1994).
Previous genetic results indicated that the zygotic function of her acts together with, or downstream of, dsx, since her and dsx mutations have equivalent phenotypes in females, yet neither the transcription nor the splicing of the dsx pre-mRNAs is affected in her mutants (Pultz and Baker, 1995). Recently, we molecularly characterized the her gene and showed that (1) her encodes a single protein with C2H2-type zinc fingers, (2) her is transcribed throughout all developmental stages in both sexes, and (3) the splicing of the her pre-mRNA is not sex- specific, suggesting that her is not sex-specifically regulated at the levels of transcription or splicing (Li and Baker, 1998). These results suggest that her is expressed non-sex-specifically and is not a downstream target of dsx. The available data thus support a conclusion that her acts together with dsx in sexual differentiation in females.
It is unclear whether the zygotic function of her is also involved in sexual differentiation of males. It was reported that there are extra bristles on the sixth sternite of her mutant males and that males that die as pharate adults often have rotated genitalia (Pultz et al., 1994). These mutant male phenotypes resemble the phenotypes of males that have partially lost the function of dsx (Pultz et al., 1994). Thus it was suggested that her has a sexual differentiation function in males, presumably acting together with DSXM (Pultz and Baker, 1995; Pultz et al., 1994). However, the phenotype of her mutant males could also be interpreted as segmental transformation (Pultz et al., 1994) since her, in addition to its specific roles in sex determination, also has a sex-independent essential function (Pultz et al., 1994).
We report here a characterization of the zygotic sex determination function of the her gene. We show that HER non-sex-specifically activates the transcription of the yp genes in fat body cells, independent of DSXF, and HER’s function in activating yp transcription is inhibited in males by DSXM. We provide genetic evidence that HER and DSXF also function independently to bring about female-specific differentiation in several other sexual dimorphic tissues. In addition, we show that, in some tissues, HER and DSXF function dependently to control female-specific differentiation. Finally, we show that, in one tissue in males, HER and DSXM are both required for normal sexual differentiation.
MATERIALS AND METHODS
Fly stocks
Flies were raised on standard corn meal food. Experiments were done at the temperature indicated. All mutations not referenced in the text and the nomenclatures of standard Drosophila genetics can be found in Lindsley and Zimm (1992). The her alleles used were previously described (Pultz et al., 1994).
Northern analysis
Northern analyses of yp2 and yp3 were performed using 20-30 μg total RNA per lane. yp2 and yp3 [32P]DNA probes were made by random primer labeling. RNAs were electrophoresed, transferred and probed using standard procedures. The rp49 [32P]DNA probe was made using an asymmetrical PCR method (Innis et al., 1990). Exposure of the blots and quantitation of the signals were done using the BioRad PhosphorImager system.
CPRG assay
The lacZ activities were measured according to the previously published protocol (Coschigano and Wensink, 1993) with the following modification. Depending on the β-galactosidase level of each genotype, 15-20 females were homogenized in 1 or 0.5 ml and 15-20 males in 0.5 ml lysis buffer, and 1-200 μl lysate was used to measure the lacZ activity.
Statistical analysis
For simple t-test, we used a pocket calculator. For analysis of variance (ANOVA) of the data in Table 2, we used the computer program StatView (Abacus Concepts, Inc.). To examine if there is interaction between dsx and her in females, we did ANOVA using logarithmically transformed primary data, since if there is no interaction between dsx and her, the expected value in her/+; dsx/+ females is the product of the values in her/her; dsx/+ females and her/+; dsx/dsx females.
RESULTS
her activates the yps in females, but does not repress them in males
Previous results indicated that dsx and her may function together to control somatic sexual differentiation. Since the only characterized target genes of dsx are the yp genes (reviewed by Bownes, 1994), we investigated whether her also regulates the expression of the yp genes and whether her functions similarly to dsx in their regulation.
We initially used northern analysis to examine the effects of her on expression of the yps. Since the complete loss of her function is lethal, the temperature-sensitive allele her1 was used. At 25°C, her1 flies are intersexual and have severely reduced viability while, at 18°C, they are morphologically normal, and have wild-type viability and fertility (Pultz et al., 1994). As shown in Fig. 1A, there is a 10-fold activation of yp2 expression by her+, since her females raised at 25°C showed a 10-fold reduction of yp2 transcript levels as compared to wild-type females and their her1/+ sisters. This is comparable to the activation effect of the dsx+ gene in females (Fig. 1D). The 10-fold activation by her+ could be an underestimate, since her1 may not be null at 25°C with respect to its function in yp expression. The level of yp2 transcript is restored to the wild-type level in her mutant females raised at 25°C when they also carry a transgene (P{her+}) containing a wild-type copy of the her gene (Fig. 1C), demonstrating that it is the her1 mutation, rather than some other mutation on the her1 chromosome, that causes the reduction in yp2 transcript level. To our surprise, yp2 expression was also reduced 10-fold in the her1 homozygous females raised at 18°C (Fig. 1A). However, when grown at 16°C, her1 females have levels of yp2 expression comparable to that seen in wild-type females (Fig. 1B). These results indicate that yp2 expression is more sensitive to the level of her function than is external sexual morphology.
The her and dsx genes control yps expression. (A-D) Northern analysis of yp2 expression. Total RNAs of adult flies were used (20- 30 μg/lane) for the analysis (see Materials and Methods). yp2 indicates the yp2 RNA level and rp49 indicates the rp49 RNA level that serves as the control for the amount of RNA loaded in each lane. In A and D, the signals showing the XY yp2 RNA levels were obtained by longer exposure of the XY blots than the XX blots. Except as indicated, the flies were raised at 25°C and aged as adult at 25°C for 3-4 days. The first number in the parenthesis indicates the temperature (°C) at which the flies were raised and the second number indicates the temperature at which the adult flies were aged. The complete genotypes are as follows. CS, Canton-S wild-type flies; tud: maternal genotype is tud. her1/+: b her1/CyO. her1: b her1/b her1. her1/+; P{her+}/+: b her1/+; P{her+}/+. dsx/+: dsx pp/+. dsx: dsx pp/dsx pp. P{her+} is a transgene containing the her+ genomic DNA fragment B17 (Li and Baker, 1998).
The her and dsx genes control yps expression. (A-D) Northern analysis of yp2 expression. Total RNAs of adult flies were used (20- 30 μg/lane) for the analysis (see Materials and Methods). yp2 indicates the yp2 RNA level and rp49 indicates the rp49 RNA level that serves as the control for the amount of RNA loaded in each lane. In A and D, the signals showing the XY yp2 RNA levels were obtained by longer exposure of the XY blots than the XX blots. Except as indicated, the flies were raised at 25°C and aged as adult at 25°C for 3-4 days. The first number in the parenthesis indicates the temperature (°C) at which the flies were raised and the second number indicates the temperature at which the adult flies were aged. The complete genotypes are as follows. CS, Canton-S wild-type flies; tud: maternal genotype is tud. her1/+: b her1/CyO. her1: b her1/b her1. her1/+; P{her+}/+: b her1/+; P{her+}/+. dsx/+: dsx pp/+. dsx: dsx pp/dsx pp. P{her+} is a transgene containing the her+ genomic DNA fragment B17 (Li and Baker, 1998).
In her males, the yp2 transcript level remains unchanged (Fig. 1A). This is in striking contrast to dsx males where the yp2 level is increased 20-fold compared to that of wild-type males and dsx/+ brothers, consistent with previous findings that DSXM functions to repress the transcription of the yps (Fig. 1D) (Coschigano and Wensink, 1993; Ota et al., 1981). This result reflects a fundamental difference between the her and dsx functions in males, rather than a leakiness of her1 (see below).
Since in wild-type females the yps are expressed both in fat body cells and in ovarian follicle cells (Brennan et al., 1982; Postlethwait et al., 1980), and the ovaries are often underdeveloped in her mutant females, the reduced level of yp2 transcripts in her mutant females could be due to the small size of the ovaries in these females, rather than reduced yp2 expression in the fat body cells. However, this is not the case. The daughters of females homozygous for the tudor mutation have no germlines and therefore have only rudimentary ovaries (Boswell and Mahowald, 1985). In the daughters of tud mothers, the level of yp2 transcript is comparable to that in wild-type females (Fig. 1A), demonstrating that the majority of the yp2 transcripts detected by northern blots are synthesized in female fat body cells.
The loss of yp2 transcripts in her1 females could be due to either an indirect effect, if her has some role in the development of fat body A. tissues per se, or a direct role of her in regulating the yps expression. We believe her’s B. effect on the yps expression is likely to be direct, since (1) the her1 mutation has C. only a 2-fold effect on the expression of a reporter gene pML-58 (Fig. 2) that is specifically expressed in female fat body cells (see below); (2) the her1 mutation has no effect on the expression of a reporter gene FBE-hsp70 promoter-lacZ (Garabedian et al., 1986) that is also specifically expressed in female fat body cells (data not shown); and (3) shifting of the her ts mutant adult females to a non-permissive temperature causes a reduction of the expression of a yp- reporter gene pCR1 (Fig. 2) (see below).
yp-reporter genes and their responses to dsx and her regulation. (A) Simplified view of yp1 and yp2 The numbers indicate the nucleotide positions of the intergenic region of yp1 and yp2. Position +1 and −1225 are the transcription start sites of yp1 and the yp2, respectively. The FBE (from position −196 to −322) is the fat body enhancer element (Garabedian et al., 1986). HRR (from position −322 to −1225) is the her-responsive region. Arrow indicates the direction of transcription. (B) The pCR1 reporter gene structure has been previously reported (Lossky and Wensink, 1995). Adh is the alcohol dehydrogenase gene of Drosophila melanogaster. lacZ is the β-galactosidase gene of Escherichia coli. The αtub 3′UTR indicates the 3′ untranslated region of α-1 tubulin gene of Drosophila melanogaster. (C) The pML-58 reporter gene construct is provided by M. Lossky and P. Wensink. hsp 3′UTR indicates the 3′UTR of the hsp70 gene of Drosophila melanogaster. The degree of response to the regulation by dsx and her is indicated, which is based on the data in Fig. 1 for A, the data in Table 1 for B and the data in Table 3 for C.
yp-reporter genes and their responses to dsx and her regulation. (A) Simplified view of yp1 and yp2 The numbers indicate the nucleotide positions of the intergenic region of yp1 and yp2. Position +1 and −1225 are the transcription start sites of yp1 and the yp2, respectively. The FBE (from position −196 to −322) is the fat body enhancer element (Garabedian et al., 1986). HRR (from position −322 to −1225) is the her-responsive region. Arrow indicates the direction of transcription. (B) The pCR1 reporter gene structure has been previously reported (Lossky and Wensink, 1995). Adh is the alcohol dehydrogenase gene of Drosophila melanogaster. lacZ is the β-galactosidase gene of Escherichia coli. The αtub 3′UTR indicates the 3′ untranslated region of α-1 tubulin gene of Drosophila melanogaster. (C) The pML-58 reporter gene construct is provided by M. Lossky and P. Wensink. hsp 3′UTR indicates the 3′UTR of the hsp70 gene of Drosophila melanogaster. The degree of response to the regulation by dsx and her is indicated, which is based on the data in Fig. 1 for A, the data in Table 1 for B and the data in Table 3 for C.
The expression of the yp1 and yp3 genes is regulated the same way by dsx and her as is the yp2 gene. This was shown by probing the same set of the northern blots (Fig. 1A-D) with probes made from yp1 and y3 genomic DNAs (data not shown). Based on these results, we conclude that HER is required, like DSXF, for the activation of the yps in female fat body cells. But, in contrast to DSXM, HER is not required for the inhibition of the yps expression in males.
her regulates the transcription of the yp genes
The reduction of yp2 transcripts in her mutant females could be due to the involvement of her in the regulation of yp2 transcription or yp2 RNA stability. To distinguish between the two possibilities, we employed the yp reporter gene pCR1 (Fig. 2). In the pCR1 construct, the intergenic regulatory region of the divergently transcribed yp1 and yp2 genes remains intact while the coding sequences of yp1 and yp2 are replaced by the Drosophila melanogaster Adh and the Escherichia coli lacZ genes, respectively (Fig. 2; Lossky and Wensink, 1995). The 3′UTR of the Drosophila melanogaster α-1 tubulin gene is fused to the 3′ end of the lacZ gene (Logan and Wensink, 1990). We used β-galactosidase activity to indicate the level of pCR1 transcripts. The effects of her and dsx on the expression of the lacZ gene of pCR1 were in all cases comparable to their effects on the yps expression as monitored by northern blots, demonstrating that her, like dsx, controls the yps expression at the level of transcription, rather than RNA stability.
Thus DSXF increases pCR1 activity 10-fold in females (Table 1, compare rows 1 and 2, P<0.02, by t-test hereafter except as noted; see Materials and Methods for details of the enzymatic assay), whereas DSXM represses pCR1 activity about 70-fold in males (Table 1, compare rows 3 and 4, P<0.01). When DSXM is present, the transcription of pCR1 is most likely completely turned off, since the level of β-galactosidase activity in dsx/+ males is similar to that seen in wild-type (Canton-S) females that lack pCR1 and only 3-fold higher than the level seen in wild-type males that lack pCR1 (Table 1, compare rows 3, 11 and 12). The observation that dsx has the same effect on the pCR1 activity as it has on the levels of the yps transcripts indicates that dsx regulates the yps expression at the level of transcription, rather than the level of RNA stability, consistent with previous findings (Coschigano and Wensink, 1993).
Similarly, her+ activity is also required in females for the transcriptional activation of the yps, rather than the stability of their transcripts. Thus her+ activates pCR1 10-fold (Table 1, compare rows 7 and 8, P<0.001), comparable to what was seen by northern analysis. Consistent with the northern result, the reduction of pCR1 activity in her1 mutant females is not due to the underdevelopment of ovaries in her1 females, since the lacZ activity from ovaries amounts to only 1% of the total lacZ activity in pCR1; her+; dsx+ females (data not shown). That these effects are due to the her1 mutation is shown by the fact that one copy of the wild-type her gene as a transgene (P{her+}) restores the lacZ activity to the wild-type level in her1 females (Table 1, compare rows 5 and 6, P>0.2). her2 is another ts mutant allele of her, weaker than her1 (Pultz et al.,1994). At permissive temperature (18°C), the pCR1 activity remains the same in her1/her2 mutant females and their wild- type sisters (Table 1, compare rows 15 and 16, P>0.1). However, when the her1/her2 mutant females were shifted to non-permissive temperature (25°C) for a day, the pCR1 activity is reduced 3-fold compared to their wild-type sisters (Table 1, compare rows 13 and 14, P<0.01). This result suggests that the effect of her1 mutation on the pCR1 activity is not the special property of the her1 allele, but rather a property of her mutations per se.
In contrast to dsx, her is not required for the repression of the yps in males. Thus, consistent with the northern result, the pCR1 activity remained the same in her1 homozygous males and their her1/+ male sibs (Table 1, compare rows 9 and 10, P>0.1).
In conclusion, our results demonstrate that her, like dsx, activates the transcription of the yps in females through the intergenic region of yp1 and yp2.
The DSX proteins are the major female-specific and male-specific regulators of the yps
The perceptions that derive from the above experiments concerning the roles of dsx and her in regulating the transcription of the yp genes suggest that both genes function in the activation of the yp genes in females, but that only dsx functions in males, where it acts to repress the yps expression. However, consideration of the quantitative aspects of the data from these experiments indicates that this interpretation is incorrect. In particular, the data with respect to the roles of DSXM in males and DSXF in females indicate that dsx function can account for all of the difference between the sexes in the levels of the yps expression. These findings with regard to dsx clearly contradict the idea that there is a female-specific role for her in the activation of the yp genes. Below we present the data that lead to this contradiction and suggest two alternative views of the role of her in regulating the yps expression that are consistent with these results.
The argument that dsx is the major, if not the only, sex- specific regulator of the yp genes derives from the analysis of the transcriptional regulation of pCR1. The pCR1 lacZ activity in dsx /+ females is about 2000-fold higher than in dsx /+ males (no expression of the yp genes) (Table 1, compare rows 1 and 3, P<0.001). However, the difference is only about 2.6- fold between the dsx homozygous female and male sibs (Table 1, compare rows 2 and 4, P<0.02). Since females homozygous for the X-linked pCR1 transgene were assayed, the 2.6-fold difference in the pCR1 activity between the dsx females and males is largely, if not entirely, due to the 2-fold difference in the gene dosage of pCR1 between females (two copies of the pCR1 transgene) and males (one copy of the pCR1 transgene). This conclusion was confirmed by our data that the lacZ activity is the same in dsx mutant females and males that carry only one copy of the pCR1 transgene (data not shown). These results are consistent with previous data that the activity of the lacZ gene under the control of the hsp70 promoter and the fat body enhancer (FBE) of yp1 and yp2 is the same in dsx/dsx females and males (Coschigano and Wensink, 1993). Thus, these results demonstrate that, in the absence of dsx, the yp genes are expressed at the same levels in both sexes.
In considering these results, it is important to note that two factors contribute to making the levels of the yps expression equivalent in dsx mutant males and females. First, the expression level of the yp genes is elevated in dsx males (compared to wild-type males), due to the absence of repression by DSXM. Second, the expression level of the yp genes in dsx females is reduced, due to the absence of activation by DSXF. Thus in both dsx mutant males and females, there are significant levels of expression of the yp genes, and these levels are equivalent in the two sexes.
There are two ways that we can see to reconcile these observations with regard to dsx with the observation that her appears to control the expression of the yp genes female- specifically. One model is that her does function female- specifically, but that its female-specific function is dependent on DSXF. The second model is that her functions sex- independently to activate the expression of the yp genes, but that its action in males is precluded by DSXM’s repression of the yp’s expression. These two models make different predictions as to the effects expected of her mutants in dsx mutant backgrounds. If the first model is correct, the presence or absence of her should have no effect on the yps when DSXF is absent. If the second model is correct, her should be able to activate the yps in dsx mutant males where DSXM is absent. Experiments that distinguish the two models are described below.
her is a non-sex-specific activator of the yps
To examine the effects of her on the yps expression in the absence of dsx function, we used the pCR1 reporter gene. To obtain sufficient mutant flies in a relatively short time period, we increased the viability of her/her; dsx/dsx flies at 25°C by using a chromosome that contains the her1 allele and a copy of the hsp70 promoter-her cDNA transgene named hsp-her#11 that rescues the lethality and external phenotypes of her mutants, but not their defect in the yps expression (Table 2, compare rows 1 and 3, P<0.0001; data not shown).
To test whether HER activates the yps expression independent of DSXF in females, we examined the responsiveness of pCR1 to HER regulation in the absence of DSXF. When DSXF is absent and HER is present in females (her/+; dsx/dsx females), the pCR1 activity is increased 7-fold as compared to the her/her; dsx/dsx female sibs (Table 2, compare rows 2 and 4, P<0.01). Since HER, even when DSXF is present, can only activate pCR1 14-fold (Table 2, compare rows 1 and 3, P<0.0001), our data suggest that the activation function of HER is not strongly dependent on DSXF.
To test whether her activates the yps expression in males in the absence of the inhibition by DSXM, we examined the responsiveness of pCR1 to her regulation in the absence of DSXM. In males, when DSXM is present, pCR1 is not expressed whether HER is present (her/+; dsx/+ males) or absent (her/her; dsx/+ males) (Table 2, compare rows 5 and 7, P>0.9). However, in males without DSXM, pCR1 is expressed and the pCR1 activity is 5-fold higher when HER is present (her/+; dsx/dsx males) than when HER is absent (her/her; dsx/dsx males) (Table 2, compare rows 6 and 8, P<0.0001). This finding suggests that wild-type her function is normally present in males and capable of activating the transcription of the yp genes, but its activity is normally overridden by the inhibitory function of DSXM.
In conclusion, we have discovered that there are two separate pathways for the activation of the yps. One is the female- specific activation of the yps, which is DSXF-dependent. The other is the non-sex-specific activation of the yps, which is HER-dependent, DSXF-independent and inhibited by DSXM. Our results also suggest that her has the same biological function in both sexes, providing further evidence that the expression of her is independent of the sex determination hierarchy.
DSXF and HER can activate the yps independently in females
We have shown that, in females, HER can act independently of DSXF. We have also asked whether the DSXF-directed female-specific activation of the yps is dependent on the HER- directed non-sex-specific activation pathway, and whether there is an interaction between DSXF and HER in females.
Our data showed that DSXF is still able to activate the yps in her/her females. DSXF increases the pCR1 activity by 11- fold in her/her females (Table 2, compare rows 3 and 4, P<0.0001). DSXF, even when HER is present, can only activate the yps 20-fold (Table 2, compare rows 1 and 2, P<0.0001). Thus, these results suggest that the activation function of DSXF is not strongly dependent on the activation function of HER.
Our data also show that there is a weak interaction between DSXF and HER in females. We used the lacZ activity of pCR1 in the XX; her1/her1; dsx/dsx females as a baseline for comparisons, since this reflects the pCR1 activity in the absence of both dsx and her. When dsx is added (XX; her/her;
dsx/+), the lacZ activity of pCR1 is increased 11-fold (Table 2, compare rows 3 and 4, P<0.0001). If her+ alone is added (XX; her/+; dsx/dsx), the pCR1 activity is increased 7-fold (Table 2, compare rows 2 and 4, P<0.01). When both her+ and dsx+ are added (XX; her/+; dsx/+), the pCR1 activity is increased 150-fold (Table 2, compare rows 1 and 4, P<0.0001). If her and dsx act independently, one would expect a 77-fold increase, rather than a 150-fold increase, in the pCR1 activity when both her+ and dsx+ are added. To examine whether the differences between the observed and the expected values are statistically significant, we did analysis of variance (ANOVA). Our results suggest that there is an interaction between her and dsx in the activation of the yps in females (p <0.02; see Materials and Methods for details). However, this interaction is weak and is not obligatory for the activation functions of either dsx or her.
The DNA sequences mediating her regulation
Our conclusion that her’s activation of the yps appears to be largely independent of dsx is qualified by the fact that we were not able to use null alleles of her in these experiments. To further address the independence of her and dsx, we asked whether the sequences that are necessary and sufficient to mediate dsx’s regulation of the yps expression are also sufficient to mediate her’s regulation of these genes. Previous experiments showed that, in vitro, the DSXF and DSXM proteins bind to the 127 bp FBE element, but not to other sequences of the intergenic region of yp1 and yp2 (Burtis et al., 1991; Coschigano and Wensink, 1993). In vivo, the FBE element is sufficient to direct the DSX-dependent, sex-specific expression of a reporter gene from a heterologous promoter (Coschigano and Wensink, 1993).
To address this topic, we used the yp-reporter gene pML-58 (Fig. 2). In pML-58, the DNA sequences from nucleotide position −322 to +58 of the yp1 gene are fused to the E. coli lacZ gene whose 3′ end is fused to the 3′UTR of the hsp70 gene (position +1 and −1225 are the transcription start sites of yp1 and yp2, respectively) (Fig. 2). Since the DNA sequence from nucleotide position −196 to −322 is the FBE element (Garabedian et al., 1986), the pML-58 construct includes both the FBE element and the yp1 promoter. Thus, the DNA sequences from nucleotide position −322 to −1225 are absent in pML-58, but present in pCR1.
We first examined the regulation of the pML-58 reporter gene by dsx. Our analysis showed that the regulatory sequences in pML-58 are sufficient to mediate regulation by dsx, consistent with previous findings (Coschigano and Wensink, 1993). DSXF increases pML-58 activity 7-fold (Table 3, compare rows 1 and 2, P<0.001). This is only 1.4-fold less than the effect of DSXF on the pCR1 activity in females (Table 1, 10-fold, compare rows 1 and 2, P<0.02), suggesting that pML- 58 contains all of the DSXF-responsive sequences. pML-58 also contains all of the DSXM-responsive sequences since, when DSXM is present, the yp-reporter on pML-58 is completely turned off, as is pCR1 (compare row 3 of Table 1 to row 3 of Table 3).
However, we found that the regulatory sequences of yp1 and yp2 in pML-58 are not sufficient to mediate regulation by her. Thus her+ only increases the pML-58 activity 2-fold in females (Table 3, compare rows 5 and 6, P<0.001) while her+ increases the pCR1 activity 11-fold in females (Table 1, compare rows 7 and 8, P<0.001). In addition, a previously described FBE- hsp70 promoter-lacZ reporter gene that can mediate dsx’s regulation was not regulated by her (data not shown) (Garabedian et al., 1986). Thus, the DNA sequences from nucleotide position −322 to −1225 are necessary for her responsiveness (termed the her responsive region, HRR, hereafter). While pML-58 and pCR1 differ both with respect to the portions of the yp1/2 regulatory region that they contain and in their promoters (pML-58 has the yp1 promoter and pCR1 the yp2 promoter), we believe that the difference in their responsiveness to her are due to the differences in the regulatory regions and not the differences in the promoters. In particular, we showed above by northern analysis that both the yp1 and yp2 genes responded identically to regulation by her. Moreover, the yp1 and yp2 promoters have been shown to be regulated sex-specifically, tissue-specifically and by tra-2 and dsx in the same coordinated manner (Belote et al., 1985; Garabedian et al., 1985; Logan and Wensink, 1990; Tamura et al., 1985).
Since pML-58 responded well to dsx regulation but not to her regulation, a good responsiveness to dsx is not dependent on a good responsiveness to her, consistent with our earlier results that the DSXF-dependent female-specific activation pathway is largely independent of the HER-dependent non- sex-specific activation pathway.
Since the HRR is necessary for about 5-fold responsiveness to her, loss of the HRR should lead to about 5-fold loss of non- sex-specific activation of the yps. This reasoning predicts that (1) in females lacking DSXF, the lacZ activity of pML-58 should be about 5-fold less than that of pCR1, (2) in females with DSXF, the lacZ activity of pML-58 should also be close to 5-fold less than that of pCR1, since the HER-dependent non- sex-specific activation is largely independent of DSXF, and (3) in males lacking DSXM, the lacZ activity of pML-58 should be about 5-fold less than that of pCR1. For all of these three comparisons, the appropriate control is the lacZ activity of pML-58 or pCR1 in wild-type male sibs, since neither reporter gene is expressed in males when DSXM is present. Our data described below are in agreement with these predictions.
The pML-58 activity is 35-fold higher in dsx/dsx females than in their dsx/+ male sibs (Table 3, compare rows 2 and 3, P<0.02), while the pCR1 activity is 180-fold higher in dsx/dsx females than in their dsx/+ male sibs (Table 1, compare rows 2 and 3, P<0.01). Thus, as predicted, in females without DSXF, the lacZ activity of pML-58 is about 5-fold less than that of pCR1.
The pML-58 activity is 240-fold higher in dsx/+ females than in their dsx/+ male sibs (Table 3, compare rows 1 and 3, P<0.001), while the pCR1 activity is 1800-fold higher in dsx/+ females than in their dsx/+ male sibs (Table 3, compare rows 1 and 3, P<0.001). Thus, in females with DSXF, the lacZ activity of pML-58 is 7.5-fold less than that of pCR1. The 7.5-fold reduction is close to the predicted 5-fold reduction. The slightly higher value than predicted could be due to the weak interaction between DSXF and HER that we have detected using pCR1 (see above).
The pML-58 activity is 13-fold higher in dsx/dsx males than in their dsx/+ male sibs (Table 3, compare rows 4 and 3, P<0.01), while pCR1 activity is 70-fold higher in dsx/dsx males than in their dsx/+ male sibs (Table 1, compare rows 4 and 3, P<0.01). Thus, as predicted, in males without DSXM, the lacZ activity of pML-58 is about 5-fold less than that of pCR1.
The reduced non-sex-specific activity of pML-58 could be due to the position effect of the chromosome insertion site of the transgene. However, this is not the case since three independent transformant lines of pML-58 showed similar low levels of the pML-58 activity in wild-type females (data not shown).
In conclusion, our data showed that the FBE is not sufficient to confer her responsiveness and the major her responsive element is located outside of FBE, in HRR (Fig. 2). Thus, HRR is necessary for the HER-dependent non-sex-specific activation of yp1 and yp2.
Independent and dependent functioning of dsx and her in controlling female differentiation
The fact that her and dsx mutant females have similar external phenotypes (Pultz et al., 1994) raises the possibility that dsx and her may regulate other downstream target genes in a similar manner to how they regulate the yp genes. This predicts that the loss of her should masculinize dsx mutant XX flies and vice versa, since HER and DSXF regulate the yp genes independently. In addition, the loss of her should also masculinize dsx mutant XY flies, since DSXM inhibits her’s activation of the yp genes. To examine whether these predictions are true, we compared the phenotypes of five different external cuticular structures, which are sexually dimorphic in wild-type adult flies, among XX and XY sibs of the following four genotypes: (1) her/+; dsx/+, (2) her/her; dsx/+, (3) her/+; dsx/dsx and (4) her/her; dsx/dsx.
The first cuticular structure examined was the number of the vaginal teeth. Our results indicate that, in the precursor cells that give rise to vaginal teeth, her and dsx act independently as in the case of the regulation of the yp genes in fat body. Thus, there are on average 27.7 vaginal teeth on an XX; her/+; dsx/+ fly and 0.0 on an XY; her/+; dsx/+ fly (Table 4, column VT, row 1 and 5). The average number of vaginal teeth on an XX; her/+; dsx/dsx fly and on an XX; her/her; dsx/+ fly is 12.4 and 10.5, respectively, indicating the intersexuality of these flies (Table 4, column VT, compare rows 1 and 2, P<0.0001; compare rows 1 and 3, P<0.0001). However, the number of vaginal teeth on an XX; her/her; dsx/dsx fly is 2.7, significantly less than on an XX; her/+; dsx/dsx fly (12.4) and on an XX; her/her; dsx/+ fly (10.5) (Table 4, column VT, compare rows 4 and 2, P<0.0001; compare rows 4 and 3, P<0.0001). These results show that the loss of her masculinizes dsx mutant XX flies and vice versa, indicating that her+ and dsx+ can act in each other’s absence in these cells.
her’s function in vaginal teeth development is inhibited by DSXM. The average number of vaginal teeth on an XY; her/+; dsx/+ fly is 0.0, while on an XY; her/+; dsx/dsx fly, it is 12.2. However, in an XY; her/her; dsx/dsx fly, the number of vaginal teeth is 3.8, much less than that of XY; her/+; dsx/dsx flies (12.2) (Table 4, column VT, compare rows 6 and 8, P<0.0001). Thus the loss of her masculinizes dsx mutant XY flies, indicating that her is capable of promoting vaginal teeth development in XY flies if the inhibitory function of DSXM is absent.
The second set of cuticular structures examined were the anal plates. The dorsal anal plate of females and the two lateral anal plates of males derive from the same precursor cells (Belote and Baker, 1982). In XX and XY intersex flies, there is a pair of anal plates located dorsolaterally to the anal opening and they are often fused at the dorsoanterior side. This pair of anal plates (referred to as DLAP hereafter) represents the intersexual differentiation of the precursor cells, and they are completely fused to form the dorsal anal plate in wild-type females and are completely separated to form the two lateral anal plates in wild-type males (Table 4, column DLAP, rows 1 and 5; Belote and Baker, 1982). Loss of her masculinizes dsx mutant XX flies and vice versa, since only 5% of DLAP in XX; her/her; dsx/dsx are fused, while 80% and 30% of the DLAP in XX; her/+; dsx/dsx and XX; her/her; dsx/+ flies are fused at the dorsoanterior side, respectively (Table 4, column DLAP, rows 1-4). Thus this result indicates that her and dsx can act independently in the DLAP precursor cells. In XY flies, DSXM inhibits the her’s function in female-specific differentiation of DLAP, since only 5% of DLAP in XY; her/her; dsx/dsx flies are fused while 70% of DLAP in XY; her/+; dsx/dsx flies are fused, and 0% of DLAP in XY; her/+ or her/ her; dsx/+ flies are fused (Table 4, column DLAP, rows 5-8).
The remaining sexual dimorphic cuticular structures examined were (1) the number and morphology of the last (most distal) transverse row of bristles (LTRB) of the basitarsus (LTRB form sex combs in males), (2) the degree of pigmentation of tergite 5 (T5) and (3) the degree of pigmentation of tergite 6 (T6). In XX flies, the intersexual phenotypes of those structures are similar between her mutants and dsx mutants. The loss of her does not masculinize the dsx mutant XX flies in the LTRB precursor cells and vice versa (Table 4, column LTRB, compare rows 2-4). The same is true for T5 and T6 (Table 4, column T5, compare rows 2-4; column T6, compare rows 2-4). These results indicate that, in the LTRB, T5 and T6 precursor cells, DSXF and HER are likely to act together in controlling sexual differentiation. When either HER or DSXF is absent, the control is abolished. If this hypothesis is true, then in XY flies, loss of her should not masculinize dsx mutant flies in the LTRB, T5 and T6 precursor cells, since DSXF is not present in those flies. This is indeed what was observed (Table 4, compare rows 6 and 8 in columns LTRB, T5 and T6).
In summary, our results indicate that, in the precursor cells of vaginal teeth and DLAP, HER controls downstream female- specific differentiation genes non-sex-specifically, and HER’s functioning is independent of DSXF in females and is inhibited by DSXM in males, analogous to HER’s regulation of the yps in fat body cells. However, our results also indicate that this is not the only mechanism by which her and dsx act. In the precursor cells of the LTRB of forelegs, T5 and T6, HER must function together with DSXF.
The function of dsx and her in the development of Sternite 6
In addition to the cuticular structures described above, we also examined the number of the 6th sternite (S6) bristles on dsx mutant, her mutant, and her; dsx mutant XX and XY flies. The results indicate that (1) in XX flies, S6 differentiation follows a default pathway that is independent of dsx (DSXF) and her, and (2) in XY flies, S6 differentiation is dependent on both her and dsx (DSXM).
There are on average 18.6 S6 bristles on an XX; her/+; dsx/+ fly, while there are virtually no S6 bristles on XY; her/+; dsx/+ flies (Table 4, column S6, compare rows 1 and 5). In XX flies, loss of either dsx or her, or both, has no effect on the number of S6 bristles (Table 4, column S6, compare rows 1- 5), indicating that dsx and her are not required for the S6 differentiation in XX flies. However, in XY flies the loss of dsx causes an increase of the number of S6 bristles to the female- specific level (Table 4, column S6, compare rows 1, 5 and 6), indicating that DSXM suppresses bristle formation on S6 of XY flies. Consistent with the previous report (Pultz et al., 1994), HER, like DSXM, is also required for the suppression of the bristle formation on S6 of XY flies. The loss of her in XY flies causes an increase in the average number of S6 bristles from virtually zero to 6.2 (Table 4, column S6, compare rows 5 and 7). This result indicates that a complete suppression of the S6 bristles requires both the DSXM and HER functions (Table 4, column S6, compare rows 5 and 7). We think these results reflect the functioning of HER with DSXM in male-specific sexual differentiation of S6. Note HER functions with DSXF in female differentiation in some tissues and 93% of DSXF sequences are also present in DSXM (Burtis and Baker, 1989). An alternative explanation of her’s role in the suppression of the S6 bristles is that her may be involved in specifying the segmental identity of S6. We think this is unlikely, since there is no other evidence to indicate that her has homeotic functions and the increased pigmentation of T6 in her mutant females (Table 4, column T6, compare rows 1 and 3) is just the opposite of what would be expected if her is required to keep T6 (and S6) from adopting the segmental identity of a more anterior segment.
DISCUSSION
Dependent and independent functioning of her and dsx
Our studies of how the HER and DSX proteins control several different aspects of sexual differentiation have revealed that how these proteins are used varies depending on the particular sexual phenotype being examined. These findings thus provide new insights into the regulation of sexual differentiation.
Based on the sample of sexual phenotypes that we examined, the most prevalent way in which these proteins function is for HER and DSXF to independently promote female differentiation. This is the manner in which they control the expression of the yp genes in the fat body where we have shown that there are two separate pathways for the activation of the yps. One is the female-specific activation of the yps, which is DSXF-dependent. The other is the non-sex-specific activation of the yps, which is HER-dependent, DSXF-independent and inhibited by DSXM. Thus in wild-type females, both HER and DSXF contribute independently to producing the high level of expression of the yps whereas, in wild-type males, the expression of the yps is prevented by the inhibitory function of DSXM, overriding the activation function of HER. We have also shown that her and dsx control downstream target genes in a similar manner in the precursor cells that give rise to the vaginal teeth and dorsal anal plate of females and the lateral anal plates of males (DLAP).
However, in the precursor cells of the LTRB of the foreleg, T5, T6 and S6, her and dsx use different mechanisms to regulate sexual differentiation. It was suggested that DSXF and DSXM control, in opposite ways, a set of genes for the sex-specific differentiation of LTRB, T5 and T6 (Jursnich and Burtis, 1993). Our results suggest that for DSXF to control target genes in these tissues, HER must be present, and vice versa. Thus the loss of either HER or DSXF completely abolishes the regulatory pathway that involves both DSXF and HER. Because of this strong dependent functioning of HER and DSXF, HER is unable to promote female differentiation in males where DSXF is absent. Thus unlike the fat body and the precursor cells of vaginal teeth and DLAP, the male-specific differentiation of LTRB, T5 and T6 does not require the suppression of HER’s function by DSXM. Therefore, the loss of HER has no effect on the phenotypes of either XY; dsx+ or XY; dsx flies. Although HER and DSXF are similarly involved in the female-specific differentiation of LTRB, T5 and T6, DSXM has different roles in the male- specific differentiation of those tissues. While in males, the loss of DSXM leads to intersexual development of LTRB, and to some extent T5 as well (loss of pigmentation at the T5 anterior lateral margins), DSXM is dispensable in the male- specific differentiation of T6 (see below). At the molecular level, DSXM is likely to be responsible for upregulating the genes for pigment production in LTRB and T5, but in T6, DSXM activity is not necessary as previously suggested (Jursnich and Burtis, 1993). Uniquely, DSXM appears to act together with HER for the male-specific differentiation of S6, since DSXM cannot completely suppress the S6 bristles of males when HER is absent.
Three mechanisms for the sexual dimorphism in Drosophila
Our analysis of sexual phenotypes of various tissues in the her and dsx single mutants and the her; dsx double mutants demonstrates that there are three ways by which sexual dimorphism is generated. The first utilizes DSXM in males and does not require DSXF in females. The second utilizes DSXF in females and does not require DSXM in males. The third utilizes both DSXM in males and DSXF in females. HER is involved in the last two modes of regulation, and likely also in at least some cases of the first mode of regulation.
On theoretical grounds, the most parsimonious way to generate differences between homologous tissues in the two sexes during evolution is to have a regulatory gene product present in the tissues of one sex and absent in the other sex, thus affecting the pre-existing non-sex-specific differentiation in one sex, but not in the other. For example, the default pathway for T6 is full pigmentation. The sexual dimorphism of T6 is solely due to the suppression of the T6 pigmentation by DSXF in females, in collaboration with HER, and is irrespective of the presence or absence of DSXM in males. Another example is the formation of S6 bristles. The default pathway is to form 18 bristles on S6. The sexual dimorphism of S6 is caused by the suppression of the bristle formation by DSXM in males, likely in collaboration with HER, and is irrespective of DSXF in females. However, in the presence of selective pressures on both sexes in evolution, one way to increase sexual dimorphism is to have female- and male- specific products of regulatory genes that each have active roles in modifying the effects of pre-existing non-sex-specific regulatory systems in opposite ways, thus generating dramatic sex-specific features. For instance, in the absence of DSXF in females and DSXM in males, the expression levels of the yp genes are equivalent between the two sexes due to the non-sex- specific control by HER. When females have DSXF and males do not have DSXM, there is a 30-fold difference between females and males in the expression levels of the yp genes, and when females do not have DSXF and males have DSXM, there is a 180-fold difference between females and males (see Table 1). However, a maximum difference (2000-fold) is observed only when DSXF is present in females and DSXM is present in males (see Table 1). The sexually dimorphic differentiation of the precursor cells of the vaginal teeth and DLAP is similarly controlled by HER and both DSX proteins. Thus, we may view her as part of a non-sex-specific regulatory system in those tissues, which is subject to the sex-specific modification by DSXF and DSXM.
On the molecular mechanism of the her and dsx functions
Besides providing the above general view of how sexual dimorphism is generated, our studies have also revealed some mechanistic details. We showed that, in females, HER and DSXF activate the yps in a largely independent manner. This result is consistent with our finding that her and dsx act through distinct sequences of the regulatory region of the yps. Since HER is a zinc finger protein, it is likely to regulate the yps directly. Our finding that her activates the yps non-sex- specifically and her function is inhibited by DSXM showed that the her and dsx genes must be expressed independently, and thus answered the question of whether her and dsx are expressed independently at the translational or post- translational levels, since previous studies showed that her and dsx do not regulate each other at the level of transcription or splicing (see Introduction).
That HER and DSXF act largely independently in regulating the yps in the female fat body is in contrast to their strong interaction in controlling differentiation of the LTRB of the foreleg, and the pigmentation of T5 and T6. These differences could be due to different organizations of the regulatory elements of the differentiation genes being controlled in these tissues, or to differences between the arrays of other factors regulating these genes, which HER and DSXF interact with. Alternatively, there may be a very low level of her activity present in the her mutant flies analyzed and that level is sufficient for DSXF to function in the regulation of the yps, but not in the regulation of the genes being controlled in the LTRB of the foreleg or T5 and T6. Nevertheless, the finding that her and dsx act in different ways to control various aspects of sexual differentiation shows that the control of sexual differentiation is more complex than previously thought.
The fat-body-specific expression of the yps is not specified by dsx since, in dsx mutant flies of both sexes, the expression of the yp-reporter genes is still restricted to fat body cells, and this DSXF-independent and non-sex-specific expression is repressed by DSXM (H. L. and B. S. B., unpublished data; Coschigano and Wensink, 1993). Since HER controls the yps non-sex-specifically and independently of DSXF, and HER function is repressed by DSXM, the question arises as to whether HER is required for the fat-body-specific expression of the yps. We think HER is not necessary for the tissue- specific expression of the yps, since the FBE-hsp70 promoter- lacZ and the pML-58 transgenes, which are not regulated by her, are still predominately, if not exclusively, expressed in fat body cells, and the loss of her does not affect fat-body- specific expression of the pCR1 transgene (H. L. and B. S. B., unpublished data). One possibility of how the fat-body specificity of the yps is controlled is by the unknown fat- body-specific factor(s) (referred to as the FBF hereafter) which is required independently of DSXF and HER for yp expression. It has been proposed that the FBF may bind to a site (the bzip1 site) in the FBE that appears to be necessary for the function of DSXF (An et al., 1996; An and Wensink, 1995).
We have shown that the HRR is necessary for her to function. The question arises as to whether the HRR is sufficient for her function. Since the FBF, which likely binds to the FBE, is necessary for HER to activate the yps, as discussed above, the HRR is not likely to be sufficient for yp expression. Consistent with this notion, DNA fragments that contain HRR, but not the FBE, were previously shown to be unable to activate the expression of the yps in fat body (Garabedian et al., 1985, 1986).
Besides dsx and her, ix also acts in the dsx branch of the sex determination pathway. With respect to the regulation of the yp genes, we have observed that ix is also required for the transcriptional activation of the yps in females, but is not required for their repression in males (H. L., Garrett-Engele and B. S. B., unpublished data). Genetic studies indicate that ix interacts with her and with dsx in regulating the yps in females (H. L., Garrett-Engele and B. S. B., unpublished data), thus ix is likely to participate both in the HER-dependent non- sex-specific activation pathway and in the DSXF-dependent female-specific activation pathway. One candidate DNA site in the FBE through which IX may function is the ‘ref’ site, which was shown to function synergistically with the DSX-binding site dsxA (An and Wensink, 1995).
In summary, we have shown that one set of female-specific terminal differentiation genes, the yp genes, is regulated by the dsx-dependent sex-specific pathway and the her- dependent non-sex-specific pathway. The integration of the two regulatory systems occurs at the level of transcriptional control. We have provided further evidence that some other unidentified sex-specific terminal differentiation genes are likely to be regulated similarly by dsx and her. However, this two-pathway mechanism does not apply to all of the sex- specific terminal differentiation genes. Tissue-specific factors and the organization of dsx- and her-binding sites in the regulatory sequences of terminal differentiation genes are likely to play important roles in determining which mechanism is used by dsx and her for the sexual differentiation of a particular tissue. Our results suggest that one way in which sexual differentiation is achieved is by sex- specific modification of the functions of non-sex-specific and tissue-specific regulatory genes.
ACKNOWLEDGEMENTS
We would like to thank Kenneth Burtis, Pieter Wensink and Marie Lossky for generously providing fly strains carrying X-linked yp- reporter transgenes FBE-hsp70 promoter-lacZ, pCR1 and pML-58. We would also like to thank Mike Simon, Paul MacDonald, Mary Ann Pultz and the members of the Baker lab for helpful discussions, and Guennet Bohm for preparation of media and fly food. H. L. is indebted to Qiang Rosa Zhang for long term support. This work is supported by the Jane Coffin Childs Postdoctoral Fellowship (H. L.), by the NIH Developmental and Neonatal Training Grant HD07249-13 and by the NIGMS.