her, a gene required for sexual differentiation in Drosophila, encodes a zinc finger protein with characteristics of ZFY-like proteins and is expressed independently of the sex determination hierarchy

Sex determination and differentiation in Drosophila melanogaster is controlled by a hierarchy of regulatory genes (Fig. 1; reviewed, for example, by Burtis, 1993; Burtis and Wolfner, 1992; Cline and Meyer, 1996; McKeown, 1992; Parkhurst and Meneely, 1994). The key element that determines whether a fly becomes a female or a male is the on/off state of the Sex-lethal (Sxl) gene (reviewed by Cline and Meyer, 1996; Sánchez et al., 1994). Sxl is on in females and off in males. During early stages of embryonic development, Sxl is sex-specifically controlled at the level of transcription by the relative levels of several transcription factors which act through Sxl’s early establishment promoter (P_e) (Fig. 1). Later in development, Sxl activity is maintained in females through autoregulation at the level of splicing. In females, Sxl controls regulatory genes that govern somatic sex determination and dosage compensation (reviewed by Baker et al., 1994; Cline and Meyer, 1996) and is also involved in germline sex determination (reviewed by Mahowald and Wei, 1994). Sxl has no function in males, where no full-length SXL protein is produced.

With respect to its role in female somatic sexual differentiation, the SXL protein directs the splicing of the transformer (tra) pre-mRNA to generate a functional mRNA in females whereas, in males, tra pre-mRNA is spliced in a default pattern that leaves premature stop codons in the mRNA (reviewed by Burtis, 1993; McKeown, 1992). Downstream of tra, the somatic sex determination pathway splits into two branches: one contains the doublesex (dsx) gene (Fig. 1; Baker and Ridge, 1980) and the other the fruitless (fru) gene (Fig. 1; Ryner et al., 1996).

In females, the tra gene product acts together with the transformer-2 (tra-2) gene products, which are sex non-specifically expressed in the soma (Amrein et al., 1988; Goralski et al., 1989; Mattox et al., 1990), to direct the splicing of the dsx pre-mRNA to generate a female-specific mRNA (reviewed by Burtis, 1993; McKeown and Madigan, 1992). In males, where functional tra product is absent, default splicing of dsx pre-mRNA produces the male-specific dsx mRNA. dsx encodes sex-specific transcription factors that are required for all aspects of somatic sexual differentiation outside of the CNS (Burtis et al., 1991). In addition, wild-type dsx function is required for some aspects of sexual differentiation that do, or may, involve the CNS (Taylor and Truman, 1992; Villella and Hall, 1996).

The female-specific DSX protein (DSX^F) acts together with the products of the hermaphrodite (her) (Pultz and Baker, 1995) and the intersex (ix) genes (Chase and Baker, 1995; Erdman et al., 1996) to repress male differentiation and to promote female differentiation in females; conversely, the male-specific DSX protein (DSX^M) acts to repress female differentiation and to promote male differentiation in males (Jursnich and Burtis, 1993; Taylor and Truman, 1992; reviewed by Burtis, 1993; McKeown and Madigan, 1992). The only known direct target genes of dsx are the yolk protein (yp) genes yp1, yp2 and yp3, which are terminal differentiation genes. Transcription of the yp genes in fat body cells is directly activated by DSX^F in females and inhibited by DSX^M in males (reviewed by Bownes, 1994).

The tra and tra-2 products also direct the splicing of fru pre-mRNA into a female-specific mRNA (Ryner et al., 1996). In males, default splicing of fru pre-mRNA produces the male-specific fru mRNAs (Ryner et al., 1996). The male-specific fru products act only in a small part of the CNS where they are necessary for male sexual behavior (Hall, 1994; Ito et al., 1996; Ryner et al., 1996; Taylor et al., 1994) and the development of a male-specific abdominal muscle, the Muscle of Lawrence (MOL) (Gailey et al., 1991; Ito et al., 1996; Lawrence and Johnston, 1986; Ryner et al., 1996). The female-specific fru products have no known functions.

Most genes in the somatic sex determination pathway have been studied extensively both genetically and molecularly. The her gene is one of the few known genes in the pathway that has not been molecularly characterized. Genetic studies (Pultz and Baker, 1995; Pultz et al., 1994) have shown that her functions both maternally and zygotically. In addition, these studies have shown that maternally, as well as zygotically, her has certain functions that are involved in sex determination/differentiation and other functions that are essential for both sexes. Although the exact nature of the sex non-specific vital functions of her are unknown, significant insight has been gained into the nature of her’s sex determination/differentiation functions (Pultz and Baker, 1995; Pultz et al., 1994). The maternal sex determination function of her is required for the activation of the early promoter of Sxl (Pultz and Baker, 1995). It is unknown whether the her maternal sex-specific function regulates the Sxl early promoter directly, or indirectly, through other regulators of Sxl (reviewed by Cline and Meyer, 1996). With respect to the zygotic sex differentiation function of her in females, it was shown that her does not regulate the expression of Sxl, tra or dsx at the level of either transcription or the splicing of their pre-mRNAs (Pultz and Baker, 1995). These results led to the suggestion that the female-specific zygotic function of her acts in parallel with, or downstream of, dsx (Pultz and Baker, 1995). However, it was unknown whether her is sex-specifically regulated in zygotes and whether it is a downstream target of one or more of the somatic sex determination genes.

Whether the zygotic function of her is involved in sexual differentiation of males is unclear. While her mutant males appear slightly intersexual with respect to a very limited subset of sexual dimorphisms (for example, her mutant males have extra bristles on sternite 6), it is not known whether these effects are due to changes in sexual differentiation, or changes in other aspects of differentiation, such as segmental identity (Pultz and Baker, 1995; Pultz et al., 1994).

We report here a molecular genetic characterization of the her gene. her encodes a single protein with four C₂H₂-type zinc fingers. her transcripts are supplied maternally to embryos and are expressed zygotically in both sexes. The transcription and the splicing of her pre-mRNAs are independent of other known sex determination genes. We show that her functions in conjunction with the dsx branch, but not the fru branch, of the sex determination hierarchy. Our results suggest that her is expressed sex non-specifically and acts together with dsx and ix in controlling female sexual differentiation.

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) except her⁸. her⁸ was isolated by the same method as her³ (Pultz et al., 1994)

Standard molecular biology techniques were used according to the protocols described in Sambrook et al. (1989).

DNA from the YAC clone N18-56, which has 160 kb DNA from the 36A3-13 region (Cai et al., 1994), was isolated by pulse-field gel electrophoresis. A probe was prepared from this DNA by random primer labeling and used to screen a λ library of Drosophila genomic DNA. Since a YAC clone of Drosophila DNA may contain some middle repeat sequences, wild-type fly (Canton S) genomic DNA was used to make a probe. This probe was hybridized to a set of duplicate filters of the library. Plaques identified by the YAC probe that did not hybridize to the genomic DNA probe, should contain unique sequences. 15 λ phage clones were isolated and the origin of the DNA inserts were confirmed by in situ hybridization to polytene chromosomes. Based on restriction mapping data, they form four groups (representative clones are shown in Fig. 2B, 6d2-3.4T3.2). Using the DNA inserts of selected λ clones in each group as probes, nine cosmid clones were isolated from a cosmid library of genomic DNA (provided by John Tamkun). Additional cosmid clones and λ phage clones were isolated using probes made from restriction fragments of the ends of cosmid clones to fill the gaps in the walk and to extend the walk proximally. The isolated cosmids form a contig covering 220 kb (Fig. 2A).

Deficiency lines [Df(2L)r10, Df(2L)RI1, Df(2L)RN2, Df(2L)RM5, Df(2L)TE116(R)GW16, Df(2L)TE116(R)GW18, Df(2L)H20] (Ashburner et al., 1990) were crossed to Canton-S flies. In situ hybridization of polytene chromosome preparations of Df(2L)/+ larvae were done according to the protocol described in Ashburner (1989), using probes made from the cosmid and λ phage clones of Fig. 2.

For RFLP mapping of her, two parental chromosomes were used: one carried the cact^F255 mutation which is a P{ry⁺} insertion line and the other the her³ mutation. We generated 43 recombinant chromosomes that carried both the cact^F255 and her³ mutations. Six RFLP markers along the walk were used and their frequencies among the recombinant chromosomes were determined by Southern analysis. A regression line was generated from these data (Fig. 2D). The intersections of the regression line with the 100% and the 0% lines of the axis that represents the percentage of the cact^F255 RFLP markers among the recombinant chromosomes defined the locations of the cact^F255 mutation and the her³ mutation on the walk at map positions 20 kb and 210 kb, respectively (Fig. 2D).

Cosmid clone 7Q6 was digested by BglII and religated. Its insert consists of a 5′ NotI-BglII 1 kb fragment and a 3′ BglII-NotI 17 kb (B17) fragment (Fig. 2E). Since the cosmid vector is a P-element transformation vector, we were able to use this construct to transform w¹¹¹⁸ flies (Rubin and Spradling, 1982; Spradling and Rubin, 1982). One of the three transformant lines (line 54A3) rescued the mutant phenotypes of all her allelic combinations. The 54A3 insert was mobilized by the Δ2-3 chromosome (Robertson et al., 1988) and one new line was obtained. It also rescued her’s mutant phenotypes. The 54A3 line is referred to as P{her⁺} in the text. The cosmid clone 7Q6 was digested by BglII and XbaI and religated to make the 10 kb B17dX transgene (Fig. 2E). None of the five transformant lines carrying the B17dX transgene rescued her’s mutant phenotypes. One of the inserts was mobilized by the Δ2-3 chromosome to generate 45 new lines and none of them rescued her’s mutant phenotypes. A 10 kb HpaI fragment (H10) from 7Q6 was inserted into the pCaSpeR2 transformation vector (Pirrotta, 1988). Ten transformant lines were obtained and all of them rescued her mutant phenotypes (Fig. 2E).

Genomic DNA fragments used for making probes are as follows (Fig. 2E). The 4R fragment is the 4 kb EcoRI-NotI end fragment of the cosmid clone 0.8N1.1. From the λ phage clone 3.4T3.2, the 3.2k fragment is a 3.2 kb EcoRI-EcoRI end fragment, the X1.4 fragment is a 1.4 kb XbaI-EcoRI fragment, the 3.5k and the 1.5k fragments are the 3.5 kb and the 1.5 kb EcoRI-EcoRI fragments, respectively. The 3.4T fragment is the 3.4 kb EcoRI-NotI end fragment of the cosmid clone 7Q6. The 7W fragments is the 7 kb EcoRI-NotI end fragment of the cosmid clone 3.4T1.1.

Probes were made from DNA fragments 4R, 3.2k, 3.5k, 1.5k and 3.4T. A larval imaginal disc cDNA library (provided by A. Cowman) was screened. Four, fifteen, four, twenty two and zero clones were isolated using the 4R, 3.2k, 3.5k, 1.5k and 3.4T probes, respectively. Representative clones from the four groups were used to make probes to hybridize Southern blots of EcoRI restriction digests of all of the isolated cDNA clones. Based on the hybridization data, the 45 cDNA clones form two non-overlapping classes, representing two transcription units. One cDNA clone (4Ra) representing the distal transcription unit was used to make probes to screen separate male and female third instar larval λ libraries (provided by S. Elledge) and a λ ZAP head library (DiAntonio et al., 1993). Two female-specific, seven male-specific and thirteen λ ZAP cDNA clones were isolated.

cDNA inserts were subcloned into Bluescript vectors, or rescued as plasmids from lambda ZAP clones, and sequenced by the dideoxy method using standard procedures. The genomic DNA fragments (4R and 3.2k) that contained her were subcloned into Bluescript vector and sequenced by the same method.

The same CyO balancer was used to balance the b her⁸pr, the b her³pr, the b her^mat and the Df(2L)H20, b pr chromosomes. The parental chromosome from which the her⁸ and her³ mutations were derived is b pr. Genomic DNAs were isolated from adult flies of the following genotypes: b her⁸pr/CyO, b her³pr/CyO, b her^mat/CyO, Df(2L)H20, b pr/CyO and b pr. Southern blots of restriction digests of these genomic DNAs were made and hybridized with probes made from the DNA fragments 5.2V, 4R, 3.2k, X1.4, 3.5k, 1.5k, 3.4T and 7W. Based on the relative intensities and the sizes of the bands on the Southern blots, the breakpoints of the her⁸, her^mat and the Df(2L)H20 mutations were determined (Fig. 2E, data not shown). No genomic DNA aberrations were detected in the her³ mutant chromosome.

A 1.7 kb EcoRI insert of cDNA clone 4Ra6 was put into EcoRI site of pCaSpeR-hs vector. The 4Ra6 clone sequence ends at nucleotide position 2094 (Fig. 3), and thus does not contain any of the polyadenylation signals described in the text. 14 transformant lines were obtained and 12 of them rescued the her mutant phenotypes. Heat shock was given for 30 minutes per day at 37°C for 5 days starting from 24-48 hours after egg-laying.

A developmental northern blot was provided by John Hebert, which contained 20 μg of total RNA per lane. The blot was probed with a ³²P-labeled riboprobe which was made by in vitro transcription of her cDNA clone 4Ra6. The rp49³²P 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 phosphor imaging system Molecular Imager GS363.

1 μg of total RNA was reverse transcribed using random hexamers as primers. One tenth of the sample was used in a 100 μl standard PCR reaction (Innis et al., 1990). PCR conditions were as follows: 94°C, 2 minutes followed by 40 cycles of 94°C, 1 minute; 42°C, 1 minute; 72°C, 1 minute. The sequences of the primers used are:

#6=5′ GCTGAAGGAACATAAGC 3′

#25=5′ GCTGGACAGAAATTGAAGTGCTC 3′

#17=5′ CTGCCCCATAAAGAGCACTTC 3′

#12=5′ TGTCCTTGATTATCTGCAT 3′

Adult flies were dehydrated once in 30%, 50%, 75% and twice in 100% ethanol for at least 12 hours per concentration, followed by 30%, 50%, 70% and 100% (twice) Freon (Electron Microscopy Sciences) treatment for 12 hours at each step, air dried and gold coated using Polaron E5400 High Resolution Sputter Coater. Photos were taken using a Philips 505 SEM.

The procedure has been previously described (Taylor, 1992) except that here rhodamine-conjugated phalloidin (Molecular Probes) was used (10 U/ml in PBS).

her was previously localized to the salivary region 36A6-11 on the left arm of the second chromosome (Pultz et al., 1994). We cloned the region containing her by chromosome walking using cosmid and λ libraries of Drosophila melanogaster genomic DNA (see Materials and Methods for details). Representative cosmid and λ clones from the walk are shown in Fig. 2A,B. They cover a 220 kb region. Selected cosmids from the walk were used to make probes for in situ hybridization to localize the proximal breakpoints of five deficiencies [Df(2L)r10, Df(2L)RI1, Df(2L)RN2, Df(2L)RM5, Df(2L)TE116(R)GW16] that cytologically and genetically had been shown to be close to the distal side of the her gene, and to localize the distal breakpoint of a deficiency [Df(2L)H20] that is near the proximal side of the her gene (Fig. 2C, see Materials and Methods for details) (Ashburner et al., 1990; Pultz et al., 1994). The closest breakpoints that flank the her locus were the proximal breakpoints of Df(2L)TE116(R)GW16 and Df(2L)TE116(R)r10, and the distal breakpoint of Df(2L)H20. These breakpoints turned out to be 150 kb apart. Since no other breakpoints were available to delimit her’s precise location in the walk, RFLP mapping (Davis and Davidson, 1984) was employed to locate her (see Materials and Methods for details). RFLP mapping positioned her approximately 10 kb distal to the breakpoint of Df(2L)H20 (Fig. 2D).

To identify the her gene, genomic DNA rescue experiments were performed. Three overlapping genomic fragments in the region where RFLP analysis had located her were reintroduced into flies by P-element-mediated germline transformation (Rubin and Spradling, 1982; Spradling and Rubin, 1982). Two of the fragments (B17 and H10) rescued all her mutant phenotypes and one fragment (B17dX) did not rescue any of the her mutant phenotypes (Fig. 2E). Using probes made from restriction fragments contained in the genomic DNA fragment B17, two non-overlapping classes of cDNA clones were isolated, representing two transcription units (Fig. 2E, see Materials and Methods for details). The distal transcriptional unit is contained in the genomic DNA fragments B17 and H10 that rescued her mutants and is not contained in the genomic DNA fragment B17dX that failed to rescue her mutants. Sequencing of a cDNA clone from the proximal transcription unit showed that its coding region lies outside of, and proximal to, the rescue fragment H10 (data not shown). Therefore, the distal transcription unit is the her gene.

Genomic rearrangements were observed both in the her⁸ allele (but not in its parental chromosome) and in the her allele (Redfield, 1924) (its parental chromosome is not available). These breakpoints were mapped by quantitative Southern analysis and by detection of rearranged end fragments in genomic Southern blots (Fig. 2E). In her⁸, the entire coding sequences of her and the 5′ region of the transcription unit immediately proximal to her are deleted (Figs 2E, 3, legend). The distal breakpoint of the deficiency Df(2L)H20 was mapped in the same way (Fig. 2E, see Materials and Methods for details).

The three longest independent her cDNA clones were completely sequenced. Nine additional independent her cDNA clones were partially sequenced. A genomic fragment named 4R (a 4 kb EcoRI-NotI end fragment of the cosmid clone 0.8N1) and a genomic fragment named 3.2k (a 3.2 kb EcoRI end fragment of the λ phage clone 3.4T3.2), which overlap with each other at one end, were partially sequenced (Figs 2E, 3). The sequenced portions of the 4R and 3.2k fragments contain all of the sequences of the sequenced her cDNA clones (Fig. 2E). All of the cDNAs have the same open reading frame (ORF), but differ in the lengths of their 3′ untranslated regions (3′UTRs), because of the use of alternative polyadenylation sites (Fig. 3). Of the five cDNAs whose 3′ ends were sequenced, two have a poly(A) sequence beginning at nucleotide position 2229, two at nucleotide position 2360 and one ends at nucleotide position 2885 without a poly(A) sequence (Fig. 3). Since poly(A) addition usually occurs about 10∼50 nt 3′ of the AAUAAA signal (Lamond, 1995), the AAUAAA signals at positions 2215, 2348, 2879 are probably used for poly(A) addition (Fig. 3). Comparison of the genomic and cDNA sequences revealed that the her gene has two small introns; the first one is 66 nt long and the second 60 nt long (Fig. 3).

The long ORF common to all sequenced cDNAs encodes a 487 amino acid protein (Fig. 3). To prove this was the HER protein, one of the completely sequenced cDNAs (4Ra6; see Fig. 3 legend) was cloned into a P element transformation vector (pCaSpeR-hs) containing an hsp70 promoter and an hsp70 3′UTR (Pirrotta, 1988) and introduced into flies by germline transformation. This transgene (hsp-her) rescued all her mutant alleles including a null allele (her⁸, Fig. 2) and thus encodes all her functions including maternal and zygotic, sex-specific and sex non-specific functions (Fig. 4; Table 1; the fertility of her⁸/her⁸ females is not rescued due to a partial deletion of an adjacent gene, Fig. 2). In the absence of heat shock, the basal level of expression of the hsp-her transgene in her mutant mothers is sufficient to rescue the her maternal sex-specific phenotype, since her¹/her^mat; P{hsp-her}/+ females produce equivalent numbers of daughters and sons (data not shown). Since the hsp-her transgene does not contain any of the her endogenous AAUAAA signals (see Materials and Methods), the multiple polyadenylation sites in the her gene are not likely to have essential regulatory functions.

The putative HER protein sequence consists of two domains, the N-terminal domain containing four C₂H₂-type zinc fingers, and the C-terminal domain which has no known structural motifs (Fig. 3). The zinc fingers in the HER protein suggest that it is likely to function as a transcription factor (for a recent review see Klug and Schwabe, 1995). However, all four of the her zinc fingers deviate substantially from the consensus sequences of the C₂H₂-type zinc finger motifs: E-(K/R)-P-(F/Y)-x-C-X_2,4-C-x-(K/R)-x-(F/Y)-x₅-L-x₂-H-x_3,4-H (Gibson et al., 1988; Krizek et al., 1991) (Fig. 5). Most notably, the highly conserved aromatic residue (F or Y) at position 12 is replaced by an S or a T residue and the conserved basic residue (K or R) at position 10 is replaced by an Y residue (Fig. 5).

To estimate frequencies of the HER-type changes at position 10 and 12 among known C₂H₂-type zinc finger motifs, we searched the PIR database (Release 50.0). There were 2314 Cx_2,4Cx₁₂Hx_3,4H motifs in 522 polypeptides. 87% (2022/2314) of them have an F or Y at position 12 and 78% (1799/2314) have a K or R at position 10. Only 3.4% (78/2314) were HER-type motifs having both a Y and a S/T at position 10 and 12, respectively [Cx_2,4CxYx(S/T)x₈Hx_3,4H]. Remarkably, all except 8 of the 78 HER-type zinc fingers are encoded by the X and Y chromosome genes Zinc Finger X (ZFX) and Zinc Finger Y (ZFY) and their homologs from frog, alligator, chicken, mouse and human (their protein products are referred to as ZFY-like proteins hereafter). These fingers correspond to the even-numbered fingers of the ZFY-like proteins (the fingers are numbered according to their relative positions in a protein) (Fig. 5; Dilella et al., 1990). Detailed structural studies have shown that the residue substitutions in ZFY evennumbered fingers retain the ββα secondary structural motif common to most C₂H₂ zinc fingers with known three-dimensional structures (Kochoyan et al., 1991). However, they do alter internal architecture and surface topology relevant to the putative DNA contacting surface (Kochoyan et al., 1991). Another feature shared by the zinc fingers of HER and ZFY-like proteins is a pairwise repeat pattern such that odd-numbered or even-numbered fingers are more similar to each other than are odd-numbered fingers with even-numbered fingers (Fig. 5). No significant sequence similarity was found between HER and ZFY-like proteins outside the zinc fingers.

Besides the zinc fingers of ZFY-like proteins, there are eight other HER-type zinc fingers in the PIR database. Five of them are from the human REST protein (nine zinc fingers in toto) (Chong et al., 1995), two are from the chicken CTCF protein (eleven zinc fingers in toto) (Filippova et al., 1996; Klenova et al., 1993) and one from the sea urchin P3A1 protein (two zinc fingers in toto) (Zeller et al., 1995). All three of the proteins are transcription factors with sequence-specific DNA binding activity.

Besides the zinc fingers, the HER protein sequence does not have significant sequence similarity to proteins in the public databases that we have searched.

To gain insight into how her is regulated, we monitored the levels of her transcripts. Northern analysis shows that her is transcribed throughout development (Fig. 6A,B). The high level of transcripts in 0-2 hour early embryos are mostly maternal, since zygotic transcription is negligible in 0-1.5 hour embryos (Anderson and Lengyel, 1981). The level of transcripts drops substantially during 2-4 hours of embryo development, indicating a lag between loss of maternal transcripts and the synthesis of zygotic transcripts. There is a higher level of her expression during 4-8 hours of embryo development. After about 8 hours of embryo development, her is transcribed at a low level. The patterns of her transcripts are similar in both sexes of adults. The level of her transcripts in adult females is slightly higher (less than two-fold) than in males, most likely due to maternal transcription in ovaries (Fig. 6B). The northern analysis data, together with our result that the hsp-her transgene completely rescues her mutations, strongly suggest that there is no sex-specific regulation of her at the level of transcription.

These northern results also show that, although there are four polyadenylation signals in the her 3′UTR, the frequencies of their usage are different. The polyadenylation signals at nucleotide positions 2731 and 2779 are not used frequently because the sizes of these predicted transcripts would be at least 2.470 kb while the sizes of the majority of the her transcripts on the northern blot fall between 1.9 kb and 2.2 kb (Fig. 6B).

Since much of the regulation in the Drosophila somatic sex determination hierarchy occurs by sex-specific alternative splicing, one possibility was that her might also be controlled at this level. Because the two introns in her pre-mRNA are small (66 nt and 60 nt), we could not ascertain whether her is regulated at the level of splicing based on the northern results. The splicing of the second intron of her is unlikely to be regulated, since the hsp-her construct does not contain the second intron and rescues her mutations in both sexes. However, the hsp-her construct does contain the first intron. We therefore used an RT-PCR assay to examine splicing. Since the alternative splicing of Sxl, tra, dsx and fru pre-mRNAs are readily seen in adults (Horabin and Schedl, 1993; Nagoshi et al., 1988; Ryner et al., 1996), we focused on the analysis of her in adult tissues. Both introns of her pre-mRNAs are spliced in adult males and females, demonstrating that there is no sex-specific splicing of the her pre-mRNA in the majority of adult tissues (Fig. 6C,D). Consistent with this result, partial sequencing of two cDNA clones, one isolated from a male cDNA library and the other from a female cDNA library, showed that the clones have both introns spliced and use the same splice sites (see Materials and Methods; data not shown). These data thus indicate that her is not regulated at the level of splicing by any of the sex determination genes that are expressed sex-specifically (Sxl, tra and dsx). This conclusion was confirmed in the case of tra by the finding that both her introns are spliced in XX; tra /tra and XY; tra /tra flies (Fig. 6C,D). The RT-PCR assay also showed that both her introns are spliced in ix mutant XX and XY flies, demonstrating that ix is not involved in her pre-mRNA splicing (Fig. 6C,D). In addition, we showed that tra-2 does not control her expression at the level of splicing (Fig. 6C,D).

With respect to the regulation of her expression, the available data did not rule out the possibility that her transcription is regulated by the ix gene in sexual dimorphic tissues. If ix had all of its effects on sexual differentiation through transcriptional regulation of her, then the hsp-her transgene should completely rescue the mutant phenotypes of ix flies since the transgene encodes all of her functions and constitutively expresses her mRNAs. To test this possibility, we crossed a chromosome that carries the hsp-her transgene into ix mutant flies and examined whether the hsp-her transgene was able to rescue the adult cuticle phenotypes of the XX; ix mutant flies. We used two independent transformant lines that contain the hsp-her transgene and have levels of her expression that rescue the her mutant phenotypes. The hsp-her transgenes had no effect on the adult cuticular phenotypes of the XX; ix mutant flies (data not shown), demonstrating that ix does not exert any significant part of its sex differentiation functions through transcriptional regulation of her.

The somatic sex differentiation pathway has two branches downstream of tra and tra-2 represented by the dsx gene and the CNS-specific fru gene, respectively (see Introduction). All known her zygotic functions are in conjunction with the dsx branch (see above). However, the question of whether her also functions in the fru branch has not been addressed previously. fru is known to be required for the development of a male-specific abdominal muscle termed the Muscle of Lawrence (MOL) in males; but it is unclear whether it is required for MOL’s absence in females (Gailey et al., 1991; Ryner et al., 1996). We used the presence/absence of the MOL to assess the possible role of her in the fru branch. her¹/her³ is one of the strongest mutant allelic combination that still gives viable adults at 25°C. The MOLs were still present in her¹/her³ mutant males and absent in females whose muscles are indistinguishable from those of wild-type females (data not shown). This result indicates that her is neither required in males for the MOL development, nor required in females for the absence of the MOL development. Moreover, expression of her using the hsp-her transgene has no effects on the development of MOL in females and males (data not shown). Based on these results, we conclude that her does not function in MOL development and therefore is unlikely to be functioning in the fru branch.

With our current knowledge regarding her, how her functions in sex determination can most profitably be addressed by considering two topics: first, whether her is expressed sex-specifically or sex non-specifically, and second, where her functions relative to the other sex determination genes.

With respect to the first of these topics, we note that her cannot be responding directly to the X:A ratio only in females and functioning in parallel with Sxl, since the ectopic expression of tra in males (XY) transform them into phenotypic females, rather than the intersexes that would be expected if her function was absent in males (McKeown et al., 1988). These data also show that her cannot be functioning in parallel with tra, under the control of Sxl, since in those tra-expressing males, Sxl is not expressed and yet they are phenotypic females.

This leaves open the possibilities that her is expressed (1) sex non-specifically, or (2) sex-specifically in the tra-dsx branch of the hierarchy. If it is assumed that the TRA and TRA-2 proteins do not have biochemical functions other than their known activities as RNA-processing factors, then her cannot be a direct target of tra or tra-2, since we have shown that her is not regulated sex-specifically at the RNA-processing level and the splicing of her pre-mRNA is not affected in tra-2 mutants. Moreover, her cannot be indirectly regulated by tra at either the translational or post-translational levels, since HER is capable of activating the expression of the yp genes in males in the absence of the repression exerted by DSX^M (Li and Baker, unpublished data) and, in those males, tra is not expressed. In addition, if it is assumed that the DSX proteins only act as sex-specific transcription factors, then her cannot be a direct target of dsx, since we have shown that her is not regulated sex-specifically at the transcriptional level. Indeed it seems unlikely that her is regulated by dsx at any level, since (1) her activates the expression of the yp genes in the absence of DSX^F (Li and Baker, unpublished data), and (2) dsx regulates the yp genes directly, making it very unlikely that dsx also regulates the yp genes indirectly through her. Finally, her cannot be regulated by ix at the level of transcription or splicing, since (1) an hsp70-her transgene does not ameliorate the phenotype of ix mutants, and (2) ix mutants do not affect the splicing of her pre-mRNA. Taken together, these considerations suggest that the most likely alternative compatible with the data is that her is expressed sex non-specifically.

With respect to the question of where her functions relative to the other sex determination genes, previous work suggested, based on the observations that zygotic her function does not control the transcription or splicing of either Sxl, tra or dsx, that her’s zygotic function was independent of the other genes in the sex determination hierarchy and that her’s zygotic function might be to control sexual differentiation in conjunction with dsx (Pultz and Baker, 1995; Pultz et al., 1994). Our findings are entirely consistent with these suggestions. In particular, we have shown that her is a positive regulator of the yp genes and that HER acts to control the expression of the yp genes through regulatory sequences that are distinct from those through which the DSX proteins regulate the yp genes (Li and Baker, unpublished data).

Genetic studies have shown that her has both maternal and zygotic sex-specific functions, as well as maternal and zygotic sex non-specific essential functions (Pultz and Baker, 1995; Pultz et al., 1994). Our results strongly suggest that the multiple roles of her represent the functioning of a single HER protein. We infer that her’s different functions are due to spatial and/or temporal controls of either HER’s activity, or the activities of factors that interact with HER, or its targets. This is not a unique property of her, since, for example, the sisterless-b (sis-b) gene also encodes a single protein which is necessary for the activation of the Sxl early promoter and for the development of the peripheral nervous system, depending on when SIS-B is expressed and what factors it interacts with (reviewed by Cline and Meyer, 1996; Villares and Cabrera, 1987).

Our studies of her reinforce the view that there are two classes of genes in the somatic sex determination hierarchy. The first class of genes are either sex-specifically expressed, such as Sxl, tra, fru and dsx, or expressed at higher levels in one sex than the other, such as the X-linked zygotic activators of Sxl [sisterless-a (sis-a), sis-b, sisterless-c (sis-c) and runt (run)] (reviewed by Cline and Meyer, 1996). This class of genes plays instructional roles in sex determination and differentiation. The second class of genes are sex non-specifically expressed, such as tra-2, the genes that act to facilitate Sxl auto-regulation [sans fille (snf), fl(2)d and vir] and the maternal or autosome-linked zygotic regulators of Sxl [daughterless (da), extra machrochaetae (emc), groucho (gro) and deadpan (dpn)]. These genes play permissive roles in sex determination and differentiation. Our results indicating that her is expressed sex non-specifically place her in the latter group. Most of the genes in the first class and all of the genes in the second class have functions other than sex determination and differentiation. Thus, only three (Sxl, tra and dsx) of the genes known to be required for sex determination act exclusively in sex determination and/or differentiation, supporting the view that genes participating exclusively in one specific developmental process are rare (Thaker and Kankel, 1992).

Among the genes that are expressed sex non-specifically and play permissive roles in sex determination, there are substantive differences in the nature of their involvement in sex determination. One subset of these genes appears to have rather specific, dedicated roles in sex determination. By this we mean that these genes, examples of which are tra-2, emc and gro, function as specific co-factors for particular steps in the sex determination hierarchy, but do not appear to have broad roles as components of the general cellular machinery concerned with gene expression. For example, the only known functions of tra-2 outside of female somatic sex determination are in the male germline (reviewed by Cline and Meyer, 1996). A second subset of these genes appears to encode general cellular functions. This type is represented by the snf gene, which encodes the U1A/U2B′′ protein, a general splicing factor that is an evolutionarily conserved component of the spliceosome (for an extensive discussion of snf, see Cline and Meyer, 1996). Since her is also expressed sex non-specifically and has a permissive function in sex determination and differentiation, the question arises as to whether HER is a general cellular factor, or has a more restricted function. Two pieces of data point towards her not being a general cellular factor. First, an expected phenotype of a null mutation of many, if not most, genes encoding general cellular factors is cell lethality. Yet mitotic clones that are homozygous for a null her mutation are readily obtained in most, if not all, external tissues (Li and Baker, unpublished data). Second, under conditions where homozygous her females have good viability they exhibit strong sexual transformations (Pultz et al., 1994).

We have shown that the her gene encodes a protein with four C₂H₂-type zinc fingers. The HER protein is likely to be a transcription factor, since the most common role of the C₂H₂-type zinc fingers is to serve as DNA-binding domains within transcription factors (reviewed by Klug and Schwabe, 1995). This is consistent with the prediction from genetic studies that HER might function as a transcriptional regulator, since the two distinct places (upstream of Sxl and either parallel to, or downstream of, dsx) in the somatic sex determination hierarchy where it seemed most likely that her acted involve transcriptional regulation (Pultz and Baker, 1995). The similarity of her zinc fingers to some of the zinc fingers of the REST (Chong et al., 1995), CTCF (Filippova et al., 1996) and the P3A1 (Zeller et al., 1995) proteins, which are known DNA-binding factors, suggests that her is a DNA-binding protein.

It is intriguing that the atypical zinc fingers of HER resemble those of ZFY-like proteins. It is unclear whether the ZFY-like proteins are involved in sex determination in mammals. The Sex determining Region Y gene (Sry) has been found to be responsible for the determination of maleness in humans and mice (reviewed by Goodfellow and Lovell-Badge, 1993; Schafer, 1995). That there might be functional relationship between the mouse Sry and Zfy genes is suggested by the correlation between the narrow spatial and temporal expression patterns in mouse genital ridge of the Sry gene and a Zfy-1::lacZ transgene (Zambrowicz et al., 1994). Among many possibilities, one is that the ZFY-like proteins may provide some permissive functions for the action of the SRY protein in sex determination in mammals. However, it is currently unclear whether the temporal and spatial expression patterns of Sry and Zfy in the embryonic genital ridge are correlated in other vertebrate species. It will be of great interest to know if there exist similar target sequences or interacting factors for the HER and the ZFY-like proteins in Drosophila and vertebrates, respectively.

In summary, our molecular characterization of her shows that her encodes a C₂H₂-type zinc finger protein, and that the expression of her is not sex-specifically regulated. Our findings also indicate that her is expressed independent of the known members of the sex determination pathway, and that her functions with the dsx, but not the fru, branch of the sex determination hierarchy of Drosophila.

We would like to thank Mike Simon, Paul MacDonald, Mary Ann Pultz and the members of the Baker laboratory for helpful discussions, Fran Thomas for help on the SEM, and Guennet Bohm for preparation of media and fly food. We thank Michael Lin for help in isolating some of the her cDNA clones. H. L. would especially like to thank 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 a research grant from the NIGMS.