Logo of geneticsGeneticsCurrent IssueInformation for AuthorsEditorial BoardSubscribeSubmit a Manuscript
Genetics. 2008 Dec; 180(4): 1963–1981.
PMCID: PMC2600935

Sexual Back Talk With Evolutionary Implications: Stimulation of the Drosophila Sex-Determination Gene Sex-lethal by Its Target transformer


We describe a surprising new regulatory relationship between two key genes of the Drosophila sex-determination gene hierarchy, Sex-lethal (Sxl) and transformer (tra). A positive autoregulatory feedback loop for Sxl was known to maintain somatic cell female identity by producing SXL-F protein to continually instruct the target gene transformer (tra) to make its feminizing product, TRA-F. We discovered the reciprocal regulatory effect by studying genetically sensitized females: TRA-F from either maternal or zygotic tra expression stimulates Sxl-positive autoregulation. We found female-specific tra mRNA in eggs as predicted by this tra maternal effect, but not predicted by the prevailing view that tra has no germline function. TRA-F stimulation of Sxl seems to be direct at some point, since Sxl harbors highly conserved predicted TRA-F binding sites. Nevertheless, TRA-F stimulation of Sxl autoregulation in the gonadal soma also appears to have a cell-nonautonomous aspect, unprecedented for somatic Sxl regulation. This tra–Sxl retrograde regulatory circuit has evolutionary implications. In some Diptera, tra occupies Sxl's position as the gene that epigenetically maintains female identity through direct positive feedback on pre-mRNA splicing. The tra-mediated Sxl feedback in Drosophila may be a vestige of regulatory redundancy that facilitated the evolutionary transition from tra to Sxl as the master sex switch.

DIPLO-X somatic cells of Drosophila melanogaster maintain their female sexual identity epigenetically through the operation of a direct positive feedback loop on pre-mRNA splicing for transcripts from the master feminizing switch gene Sex-lethal (Sxl) (Bell et al. 1991). The female-specific SXL-F protein thereby generated acts on various regulatory gene targets, including the feminizing switch gene transformer (tra) to elicit female differentiation. In some fly species that do not use an X-chromosome dose to determine sex (Pane et al. 2002; Lagos et al. 2007), and possibly even in the honeybee (Hasselmann et al. 2008), tra occupies the position of Sxl as the master developmental switch gene that epigenetically maintains the female sexual fate commitment by direct positive autoregulation of its own pre-mRNA splicing. In those species, Sxl itself does not appear to be sex-specifically regulated and its function is unknown. Thus an important evolutionary question is how Sxl was recruited to the Drosophila sex-determination pathway. Here we describe functional and structural evidence for an unanticipated additional circuit driving Sxl-positive autoregulation that seems relevant to this evolutionary question.

We discovered this surprising regulatory circuit for Sxl while investigating the developmental basis for a unique female-sterile mutant phenotype encountered when Sxl autoregulation was discovered: Sxl mutant female ovaries seemed to disappear during metamorphosis (Cline 1984). We show here that these ovaries disintegrate due to their ambiguous sexual phenotype, rather than from an upset in the vital process of X chromosome dosage compensation, which Sxl also controls.

The functional relationship between the two promoters of Sxl and the way in which their products are affected by particular Sxl mutations are central to this female-sterile phenotype. For Drosophila, diplo-X somatic cells become female while haplo-X somatic cells become male because the level of regulatory proteins generated from a double dose of X-chromosome signal element genes (XSEs) is sufficient to activate the “sexual pathway establishment” promoter, SxlPe, while the level generated from just a single dose of XSEs is not (reviewed in Cline and Meyer 1996; see also Erickson and Quintero 2007). Although Sxl is required throughout female development to control sexual differentiation and dosage compensation, SxlPe is sensitive to the level of XSE proteins for only a 45-min period that ends soon after fertilization as the embryo transitions from relying predominantly on maternal gene products to relying on zygotic gene products instead. As SxlPe shuts off, the “sexual pathway maintenance” promoter, SxlPm, becomes active. SxlPm remains active thereafter to provide females with the feminizing SXL-F protein that they need throughout development for appropriate sexual differentiation and dosage compensation.

SxlPm also becomes active in males, but they produce no feminizing SXL-F protein because transcripts from SxlPm, unlike those from SxlPe, are productively spliced into mRNAs that encode full-length SXL-F protein only if SXL-F protein is already present. SXL-F is an RNA-binding protein that acts directly on its own pre-mRNA to prevent incorporation of a male-specific exon that would otherwise abort translation early and thereby keep Sxl functionally off. Thus the transient early burst of transcription from SxlPe in diplo-X embryos provides a pulse of SXL-F protein that triggers engagement of a positive feedback loop for the productive splicing of SxlPm transcripts. This feedback loop then maintains the female (“ON”) state for Sxl epigenetically thereafter. Because haplo-X embryos lack this triggering pulse of SXL-F, they do not engage the Sxl feedback loop. Instead, they include the male-specific exon in their Sxl mRNA by default and Sxl remains off. SXL-F continually promotes the productive pre-mRNA splicing of transcripts from the downstream target gene tra, thereby ensuring a continual supply of that gene's actively feminizing RNA-binding protein, TRA-F. TRA-F in turn imposes female-specific pre-mRNA splicing on its gene targets.

The fly's gonad was known to be peculiar with respect to sex determination (reviewed in Oliver 2002) ever since the discovery that tra could be eliminated from diplo-X germ cells without apparent effect (Marsh and Wieschaus 1978). This observation led to the conclusion that tra functioning was limited to somatic cells. In somatic cells, tra seemed to be only a slave to Sxl, continually depending on female-specific SXL-F protein to elicit production of its own feminizing protein product (Nagoshi et al. 1988; Sosnowski et al. 1989). Here we show that female-specific tra expression is not limited to somatic cells and that tra expression in the mother's germline, or in the very young embryo itself, can stimulate Sxl-positive feedback in the female gonadal soma while having no adverse affect on the sexual development of males.

Although Sxl regulation in the Drosophila soma was believed to be strictly cell-autonomous, its regulation in germ cells was known to have a cell-nonautonomous component (Nöthiger et al. 1989). Responding to Sxl in the gonadal soma, somatic tra triggers female-specific splicing of pre-mRNA from Sxl itself in the neighboring diplo-X germ cells (Bopp et al. 1993; Oliver et al. 1993; Steinmann-Zwicky 1994; Waterbury et al. 2000; Janzer and Steinmann-Zwicky 2001; Evans and Cline 2007). As in somatic cells, in germ cells SXL-F is necessary for female development (Schüpbach 1985), but unlike in somatic cells, it is not sufficient (Hager and Cline 1997).

The female-sterile phenotype that is the focus of this study was first seen in a situation where it was inferred that the Sxl-positive feedback loop had become engaged in nearly all somatic cells except those of the gonad (Cline 1984). The disappearance of those Sxl mutant ovaries during metamorphosis was surprising in two respects. First, it was a phenotype that had not been previously reported, much less attributed to mutations affecting sex determination or dosage compensation. While it was known that ovaries partially or completely masculinized by mutations in any of the four somatic sex-determination genes downstream of Sxl (tra, transformer-2, doublesex, and intersex) were abnormal, they remained intact (reviewed by Laugé 1980)—as did ovaries suffering X-chromosome dosage compensation upsets (Kelley et al. 1995). Second, only the gonad appeared to be defective in these adult females, suggesting that there might be something different about the gonadal soma that made it a particularly difficult somatic tissue type in which to trigger full engagement of the Sxl-positive feedback loop.

This unusual female-sterile phenotype is found in situations where the normal XSE-based mechanism for initiating female-specific expression of Sxl has been impaired, but Sxl+ alleles are activated instead by the products of the double-mutant Sxl allele, Sxlf7,M1 (Figure 1A) (Cline 1984, 1986; Granadino et al. 1991). The SxlM1 lesion in Sxlf7,M1 forces female splicing of some SxlPm transcript (Cline 1984; Bernstein and Cline 1994). The mutant Sxl-f7 protein thereby produced can activate an Sxl+ allele in trans, but far less effectively than wild-type protein. Furthermore, since SXL-F7 cannot direct female-specific expression of tra (Cline 1984; Sosnowski et al. 1989), any feminization of somatic tissues that occurs when Sxlf7,M1 is in the presence of a second Sxl allele must result from the productive splicing of Sxl transcripts from that allele. Although Sxlf7,M1 lacks somatic feminizing activity, its germline-feminizing functions are intact, as first shown by the observation that the homozygous Sxlf7,M1 germline clones produce functional eggs (Cline 1983a)—another indication of profound differences between germline and somatic sex determination.

Figure 1.
Sxlf9, a nonsense mutation in the SxlPe-specific exon E1, generates epigenetic mosaicism in SXL-F expression in mutant female embryos. (A) Diagram of the 5′-end of Sxl showing the molecular lesions in Sxlf9 and Sxlf7,M1. (B) Wild-type stage 12 ...

Analysis of this female-sterile phenotype was facilitated by our identification of a purely zygotic mutant genotype in which ovarian disintegration occurred with high penetrance, yet female viability remained high. One key feature of this genotype is an increased dose of sans-fille, a gene that enhances the ability of Sxlf7,M1 to activate Sxl alleles in trans (Cline et al. 1999). The other key feature is the “initiation-defective” mutant allele Sxlf9 whose lesion disrupts the ability of Sxl to respond to X-chromosome dose in the soma without disrupting any other Sxl function (Maine et al. 1985). The Sxlf9 lesion (Figure 1A) is a nonsense mutation in the exon that is unique to mRNAs derived from SxlPe in response to higher X chromosome dose (see materials and methods). Because mRNAs derived from Sxlf9Pm are therefore wild type, if the products of an allele like Sxlf7,M1 acting in trans can induce full engagement of the Sxlf9 feedback loop, the subsequent functioning of Sxlf9 will be indistinguishable from that of Sxl+ in that cell (see Figure 1). But if instead the level of SXL-F protein from Sxlf9 never reaches that self-sustaining threshold—as seems to happen particularly frequently in somatic cells of the mutant gonad—constitutive SXL-F7 from Sxlf7,M1 will nevertheless allow Sxlf9 to maintain a sub-threshold level of wild-type SXL-F in that cell throughout development. Such cells are potentially intersexual, and they are sensitized to even small enhancements of Sxl-positive autoregulation that would raise them above threshold. The subtle enhancement by TRA-F was revealed in this sensitized genotype and led us to highly conserved predicted binding sites for TRA-F in Sxl that had escaped notice for decades.


Drosophila culture and genetics:

Flies were raised at 25° in uncrowded conditions on a standard cornmeal, yeast, sucrose, and molasses medium. Markers, balancers, and transgenes are described at http://flybase.bio.indiana.edu/ except as follows: Df(tud) and tudB45-6 were a gift from R. Lehmann; P{U2af50-traF w+mW.hs}2B allows one to maintain tra homozygous stocks (Evans and Cline 2007). The Binsinscy, y w snx2B let P{w+mC,hs-hid} balancer carries a recessive lethal and the dominant, temperature-sensitive lethal hs-hid transgene described at http://flybase.bio.indiana.edu. The balancer's dominant lethality is tight and remarkably rapid after a 1-hr, 37° treatment of embryos or first instar larvae.

The Sxlf9 molecular lesion:

Genetic fine-structure mapping (Bernstein and Cline 1994) incorrectly located Sxlf9. The strategy was based on intragenic complementation between Sxlf9 and SxlM1,f3, but we subsequently discovered that such females have an unusual propensity for nonhomologous recombination. Consequently, what had seemed to be wild-type intragenic recombinants were later found to be cytologically subtle tandem duplications with both parental alleles in cis. DNA sequencing ultimately showed Sxlf9 to be an A > T substitution in exon E1 at nucleotide 5362 (numbered from the SxlPm start site) (Figure 1A).


Tissues were homogenized in Trizol (Invitrogen), and RNA was isolated according to the manufacturer's protocol. Total RNA (4 ng) was reverse transcribed using random primers. The following primer pairs were used for PCR amplification of cDNAs:

  • tra exon 1, 5′-CCGATGAAAATGGATGCCG-3′;
  • dsxM exon 2, 5′-TGGTAGGTCATCGGGAACATCG-3′;
  • dsxM male-specific exon, 5′-GCCATCGGGGTGTAGTAGTTGTAG-3′;
  • dsxF exon 3, 5′-CGCAGACGCCAACATTGAAG-3′;
  • dsxF female-specific exon, 5′-TCGGGGCAAAGTAGTATTCGTTAC-3′.


Embryos were fixed, stained, and photographed as described by Bernstein et al. (1995). Ovaries were dissected from females 1–3 days after eclosion. Fixation and washes were carried out as described previously (Hager and Cline 1997). Antibodies were used at the following concentrations: mouse anti-SXL (Bernstein et al. 1995) at 1:2000; polyclonal rabbit anti-GFP (Molecular Probes) at 1:1000; mouse anti-Eya 10H6 (Bonini et al. 1993) at 1:25; mouse anti-En 4D9 (Patel et al. 1989) at 1:2; goat anti-mouse-Alexa546 (Molecular Probes) at 1:500 for SXL and 1:400 for all other primaries; and goat anti-rabbit-Alexa488 (Molecular Probes) at 1:400. Images were taken on a Leica AOBS confocal microscope and analyzed with ImageJ software.

Light microscopy:

Live gonads were dissected in Ephrussi and Beadle Ringer's solution (128 mm NaCl, 4.7 mm KCl) and viewed by phase contrast or Nomarski optics.

Generation of labeled FLP-out clones in ovaries:

Female larvae of the genotype described were heat-shocked for 5min in a 37° water bath 3–27 hr after egg laying. Adults were dissected 3 days after eclosion, and live ovaries were scored for GFP expression using a Zeiss Axioskop. An ovariole was scored as containing a somatic clone if any or all follicle cells expressed GFP.

Cloning of Sepsid tra:

Advantage was taken of synteny between tra and l(3)73Ah (Pane et al. 2002; Lagos et al. 2007). We PCR amplified l(3)73Ah using the primers (forward) 5′- CAAGAGTTGCCTGGTGAAGCAC-3′ and (reverse) 5′-GTACTTTTCGATGCCGTTGAGG-3′. Having sequenced Sepsis cynipsea l(3)73Ah, we used that information to design probes (5′-ACACGTTGACCTGCTCGTCGAGTC-3′ and 5′-GAATCCGACTTCCATCGACTCGAC-3′) with which we screened a S. cynipsea fosmid library (generously provided by M. B. Eisen) by the protocol of Han et al. (2000). We then directly sequenced the fosmid containing l(3)73Ah to obtain S. cynipsea tra genomic sequence. Partial Sepsis neocynipsea tra sequence was obtained by sequencing the fragment PCR amplified by the following primers: (forward) 5′-AAATGCCTGTACTCACCCGAGAG-3′) and (reverse) 5′-TGGCATGAGTAACGTCAGCACG-3′.


High-viability Sxl mutant females whose ovaries disappear:

Sxlf7,M1/Sxlf9 mutant females carrying a transgenic copy of sans fille+ (P{snf+}) to boost viability were the focus of this study. Although their viability and the penetrance of their ovarian defects were affected somewhat by genetic background, temperature, and the parent of origin of the extra snf+ allele, their viability was always at least 40% and most females lacked both ovaries.

The molecular basis for the phenotype of these females can be inferred from what is known about Sxl-positive autoregulation, these two mutant Sxl alleles, and the effects of snf+ dose. Above a particular threshold level of SXL-F protein activity, a cell will ramp up to full engagement of the Sxl-positive feedback loop—the normal female state in which all SxlPm transcripts are processed into mRNAs that lack the translation-terminating, male-specific exon. These mRNAs encode full-length SXL-F proteins that direct female somatic development and impose a rate of X-chromosome dosage compensation appropriate for diplo-X somatic cells. Below that triggering threshold, cells will instead damp down to the pre-mRNA splicing state characteristic of males, in which all SxlPm transcripts are processed by default into mRNAs that encode only truncated SXL proteins. These truncated proteins lack all somatic feminizing activity and allow a level of dosage compensation that is appropriate only for haplo-X cells.

The fact that there are generally only two stable Sxl expression states—fully ON and fully OFF—is dramatically illustrated in Figure 1C by Sxlf9 mutant female embryos. Their salt-and-pepper pattern of SXL-F expression reflects genetically identical somatic cells having settled stochastically into one or the other of the two stable Sxl pre-mRNA splicing modes as a consequence of the pulse of SxlPe-derived activity that they generated earlier in response to an X-chromosome dose having been abnormally close to the threshold at which feedback loop engagement is triggered. Sxlf9 is a recessive, female-specific lethal allele impaired in its ability to respond to the primary sex-determination signal in the soma, but otherwise wild type with respect to all Sxl functions (Maine et al. 1985). As diagrammed in Figure 1A, Sxlf9 carries a nonsense point mutation in exon E1 (see materials and methods). Since exon E1 is included only in SxlPe mRNAs (Keyes et al. 1992), this leaky nonsense mutation reduces the initial pulse of SXL-F that normally triggers engagement of the SxlPm transcript splicing feedback loop in diplo-X embryos. On the other hand, any SxlPm transcripts that are spliced in the female mode are fully wild type. Hence, any Sxlf9 diplo-X somatic cell in which the level of SXL-F protein reaches the threshold for engaging the SxlPm splicing feedback loop will ramp up to full female splicing. From that time forward, the Sxl mutant cell will be indistinguishable from a wild-type diplo-X cell with respect to Sxl functions. Conversely, any Sxlf9 cell that fails to reach that threshold will damp down to the male splicing pattern and become indistinguishable from a Sxl diplo-X cell instead. The mutant embryo in Figure 1C is at stage 12, by which time cells have fully ramped up or damped down with respect to immunostaining of protein generated from SxlPm-derived mRNAs. Contrast this mosaic pattern to the uniformly dark immunostaining in Figure 1B of a Sxlf9/Sxl+ sister at the same stage. The pattern of immunostaining of Sxlf9 female embryos initially is uniform but much lighter than wild type (not shown), reflecting the reduced level of SXL-F protein generated from the mutant SxlPe mRNA.

Because SxlPe and the XSEs that activate it seem to not operate in the germline (Keyes et al. 1992; Steinmann-Zwicky 1993), an allele like Sxlf9 that is defective only with respect to SxlPe-mediated functions should be wild type in germ cells. Although a definitive test of this expectation by pole-cell transplantation has not been made, three observations (data not shown) argue that Sxlf9 is functionally indistinguishable from Sxl+ in germ cells: first, Sxlf9 and Sxl+ germline clones induced in young larvae behave in the same way; second, Sxlf9 complements all Sxl mutants that are defective only in germline functions; and third, the rare Sxlf9 escaper females that one finds in 18° cultures are fertile.

As Figure 1A diagrams, the other key allele in this study, Sxlf7,M1, has two significant lesions (Bernstein et al. 1995). M1 is a roo transposon in the sex specifically spliced region of Sxl. It disrupts splicing control by allowing a significant level of female SxlPm transcript splicing even in the absence of SXL-F activity. Through positive feedback, SxlM1 ultimately ramps up to full female activity in most, although not all, haplo-X cells (Cline 1979), killing the chromosomal males by upsetting dosage compensation (Cline 1983b). Nevertheless, Sxlf7,M1/Y males are fully viable and fertile because the f7 missense mutation affects all SXL-F protein isoforms, eliminating their ability to regulate tra and reducing their autoregulatory and dosage compensation activities (Cline 1984; Bernstein and Cline 1994). Homozygous Sxlf7,M1 females are only poorly viable and their soma is completely masculinized, while Sxlf7,M1/Sxl females are inviable. The f7 lesion, like f9, seems to not affect germline functions, since Sxlf7,M1 complements all Sxl mutants that are defective only in germline functions (data not shown) and Sxlf7,M1 germline clones support oogenesis (Cline 1983a).

The utility of Sxlf7,M1 in studies of Sxl-positive autoregulation stems from the fact that it has low but significant constitutive autoregulatory activity but no ability to feminize on its own. Thus any feminization observed in Sxlf7,M1 heteroallelic animals must be due to the expression of the other Sxl allele. In situations where that other allele (e.g., Sxlf9) would not be able to express its feminizing potential by itself, Sxlf7,M1 can elicit in trans at least some of that cryptic feminizing activity. Its effectiveness in this regard is greatly enhanced by increased maternal or zygotic snf+ dose (Cline et al. 1999). As Table 1 shows, one extra copy of snf introduced from the father increased the viability of Sxlf7,M1/Sxlf9 females >50-fold (compare cross 1 to cross 3) and did so without greatly reducing the penetrance of the ovarian defect: 95% of the adult females in cross 3 lacked both ovaries, and the remaining 5% had only one. These ovary defects are illustrated in B and C of Figure 2.

Figure 2.
Imposing a uniform somatic sexual phenotype rescues Sxlf7,M1/Sxlf9 ovaries. Live whole mounts of diplo-X (chromosomally female) adult gonads. (A) Sxlf9/+ ovary illustrating the wild-type phenotype. (B) Asymmetric Sxlf7,M1/Sxlf9 ovary. (C) Typical ...
Rescue of Sxlf7,M1/Sxlf9 mutant ovaries by maternal or zygotic traF

It follows from the information presented above that the abnormal gonadal phenotype of Sxlf7,M1/Sxlf9 females is caused by a failure of Sxlf7,M1 to induce Sxlf9 to fully engage its feedback loop in the gonadal soma. The apparent tissue specificity of this effect in the surviving adults suggested that it is more difficult to induce Sxl autoregulation in the gonadal soma than in any other somatic cell type.

A somatic sexual identity crisis causes Sxlf7,M1/Sxlf9 ovaries to disintegrate during metamorphosis:

To better understand the developmental fate of Sxlf7,M1/Sxlf9 ; +/P{snf+} ovaries during metamorphosis, we tagged their germ cells using a nos:GAL4 germ-cell-specific driver and a UAS-GFP target. The gross morphology of the mutant ovaries (Figure 3D) is remarkably normal prior to metamorphosis (compare to the wild type in Figure 3C). In contrast, the phenotype of homozygous Sxlf7,M1 larval ovaries (Figure 3B) much more closely resembles that of their brothers' testes at the same stage (Figure 3A). The fact that Sxlf7,M1/Sxlf7,M1 larval ovaries are clearly more masculinized than Sxlf7,M1/Sxlf9;+/P{snf+} larval ovaries shows that Sxlf7,M1 must be inducing some female splicing of SxlPm transcripts from the Sxlf9 allele. That induced female splicing must be below the threshold for triggering Sxlf9's self-sustaining splicing feedback loop since such ovaries soon disintegrate.

Figure 3.
Sxlf7,M1/Sxlf9 ovaries become morphologically abnormal only during metamorphosis. Live whole mounts of gonads viewed under Nomarski (A–F and H) or pseudo-darkfield illumination (G), with germ cells tagged with GFP (green overlay). Larval gonads ...

Molecular analysis of Sxlf7,M1/Sxlf9;+/P{snf+} larval ovaries confirmed that Sxl functioning was not as fully female as morphology had implied. We used doublesex (dsx) mRNA as an indicator of Sxl female functioning. dsx is a target of tra (Baker and Ridge 1980; Nagoshi et al. 1988). In the absence of TRA-F protein, dsx generates male-specific dsxM mRNA, while in the presence of TRA-F, dsx splicing follows the alternative, female-specific splicing pattern, generating dsxF mRNA.

Lanes 1 and 2 of Figure 4A show that Sxl+ and Sxlf7,M1 male larvae carrying P{snf+} produce only dsxM mRNA, while Sxl+/Sxlf9;+/P{snf+} control female larvae (lane 4) produce dsxF mRNA, with only a trace of dsxM. In contrast, Sxlf7,M1/Sxlf9;+/P{snf+} female larvae (lane 5) produce a significant amount of dsxM mRNA in addition to dsxF mRNA, clearly signaling molecular intersexuality. To test our hypothesis that the larval ovary was a primary source of this male mRNA, we compared ovaries separated from the fat body tissue in which they were embedded (lane 6) to the fat body tissue (lane 7) from the same Sxl mutant animals. Both tissues produced dsxF mRNA, but only the isolated larval gonads produced a strong dsxM band as well.

Figure 4.
Molecular analysis of dsx regulation in Sxlf7,M1/Sxlf9 females reveals a maternal effect of tra. Sex-specific splicing of dsx transcripts was assessed by RT–PCR of RNA from wandering (late) third instar larvae. Primer pairs used are shown on the ...

Figure 3 shows that the ovarian morphology of these mutant gonads breaks down during the first half of the pupal period. By pupal stage 5 (Bainbridge and Bownes 1981), wild-type ovaries are organized into a series of parallel ovarioles comprised primarily of germaria (Figure 3E). Although the Sxl mutant ovaries still had a full complement of germ cells at this stage, very few were organized into recognizable germaria (Figure 3F). The ruptured appearance of the mutant ovary reflects disorganization, not an artifact of dissection. By the time of eclosion (Figure 3H), disorganization is extreme and most germ cells have been lost (compare to the wild type in Figure 3G). Scored on the basis of morphology alone, 96% of the adult females in this cross appeared to lack gonads; however, a germ-cell-specific marker revealed that 92% of those without recognizable ovaries nevertheless had some undifferentiated germ cells on each oviduct.

To further explore the phenotype of these disorganized adult ovaries, we immunostained them with an antibody against the EYES-ABSENT (EYA) protein, a non-sex-specific marker for gonadal soma, while also marking nuclei with DAPI and germ cells with GFP. As shown in Figure 5A for wild-type ovaries, prior to stage 9 EYA marks the highly organized single-cell layer of somatic cells that surrounds each egg chamber (Boyle and Dinardo 1995; Bai and Montell 2002). Figure 5B shows a Sxl mutant adult female reproductive tract that differentiated a few fairly normal egg chambers on one side, but only highly disorganized gonadal tissue on the other. In that disorganized region (magnified further in Figure 5B′), somatic as well as germ cells were clearly labeled, but both occurred in amorphous multilayered clumps. These clusters show that mutant somatic and germ cells can proliferate significantly during metamorphosis even if they fail to form a recognizable ovary.

Figure 5.
Sxlf7,M1/Sxlf9 adult ovaries are grossly disorganized and intersexual. Gonads from wild-type (A) and Sxl mutant (B and B′) adult females were immunostained for EYA protein (red), which marks gonadal soma, while nuclei were marked by DAPI (blue) ...

To assay whether these grossly abnormal somatic cells had retained some female character, we immunostained for the product of the engrailed (en) gene. EN does not label adult testes, but in the adult ovary it marks terminal filament and cap cells (Forbes et al. 1996), somatic cells normally located at the apical tip of the germarium (Figure 5C). EN immunostaining revealed recognizable terminal filaments that were not otherwise associated with organized ovarioles (Figure 5D). Thus, even in the most grossly disorganized regions of these mutant adult ovaries, at least some somatic cells retain some female character.

Male differentiation was also apparent in these disorganized regions. The testes sheath is an epithelial covering (Figure 5E) whose bright yellow pigmentation serves as a marker of maleness for the underlying gonadal mesoderm (Fung and Gowen 1957). Eighteen percent of the Sxl mutant ovaries (n = 40) contained sporadic patches of cells producing this distinctive pigment (Figure 5F). Hence, by the adult stage, some of the Sxlf7,M1/Sxlf9;+/P{snf+} gonadal soma must have become at least partially masculinized.

Because there was no precedent for intersexuality alone causing such gross disorganization of a diplo-X gonad, we wondered whether the disorganization might instead be caused by an upset in dosage compensation. If it were, uniformly masculinizing or feminizing the Sxl mutant gonads by manipulating tra or dsx should not improve the situation, since neither tra nor dsx affects dosage compensation. In contrast, if confusion over sexual identity were the sole problem, such manipulations would allow Sxl mutant females to differentiate organized gonads.

The results for uniform masculinization are shown in Figure 2 and Table 2. The gonads of Sxlf7,M1/Sxlf9;+/P{snf+} females were invariably rescued when they were masculinized by loss of tra+. All 62 tra gonads (Table 2, second row) were recognizable as pseudotestes (Figure 2F), with a phenotype indistinguishable from that of their Sxl+ tra control sisters (Figure 2E). In contrast, only 3 of 26 tra/+ gonads of Sxl mutant females (Table 2, first row) were recognizable.

Rescue of Sxlf7,M1/Sxlf9; +/P{snf+} mutant gonads by somatic masculinization

Uniform feminization by a TRA-F transgene also rescued, as shown in Figure 2D and Table 1, crosses 2 and 4. Rescue by the u2af50-traF transgene was complete even in cross 2 where the penetrance of the ovarian defect was expected to be highest because no extra copy of snf+ was included to boost autoregulation. The surprising maternal effect of TRA-F is discussed below.

Since dsx controls fewer aspects of sexually dimorphic differentiation than its regulator tra (McRobert and Tompkins 1985; Taylor and Truman 1992; Taylor et al. 1994), we wondered whether masculinization of ovaries by DSX-M protein would rescue less effectively than masculinization by loss of TRA-F. To answer this question, we used dsxD, which can be spliced only in the male pattern (Nagoshi and Baker 1990). Hence, dsxD/Df(dsx) females generate only DSX-M protein. Masculinization by DSX-M did rescue, generating pseudotestes in both Sxl+ control and Sxl mutant females that were indistinguishable from those generated by tra (compare Figure 2, G and H, to Figure 2, E and F). Although not all masculinized females had both gonads (Table 2, cross 2), the difference between Sxl mutant females and their Sxl+/Sxlf9;+/P{snf+} control sisters in this respect was not significant. Thus we could conclude that masculinization by DSX-M rescues as effectively as masculinization by tra.

The experiments with tra and dsx show that the problem with Sxl mutant gonads is one of sexual identity, not dosage compensation. Moreover, since tra and dsx directly affect the sexual identity only of somatic cells, rescue by genetic manipulation of these genes argued that a defect in Sxl autoregulation in somatic cells alone caused disintegration. As a direct test of the strictly somatic basis for this gonadal defect, we asked whether gonadal disintegration would be ameliorated either by genetically eliminating germ cells (Boswell and Mahowald 1985) or by artificially expressing SXL-F specifically in germ cells to stimulate their engagement of the Sxlf9 autoregulatory loop. Although gonads without germ cells, whether sexually transformed or not, are even more rudimentary than sexually transformed gonads with abnormally developing germ cells, they are nevertheless recognizable as organized gonads. We found that Sxlf7,M1/Sxlf9;+/P{snf+} females lacking germ cells invariably lacked one or both ovaries, while their Sxl+ sisters, also without germ cells, always had both (Table 3, cross 1). Moreover, a Sxl cDNA expression construct transcribed specifically in the germline (Hager and Cline 1997) also failed to ameliorate the Sxlf7,M1/Sxlf9 gonadal defect (Table 3, cross 2).

Effect of germ cells on Sxlf7,M1/Sxlf9; +/P{snf+} ovary disintegration

Clonal analysis of unilateral Sxl mutant ovaries shows that Sxlf9 feedback loop engagement in the gonadal soma is likely to be cell-nonautonomous:

When the probability of a Sxlf7,M1/Sxlf9 ; +/P{snf+} adult female having recognizable ovaries is low, most recognizable ovaries are unilateral: an ovary averaging half the normal size will be present on one side of the female, with none on the other side (Table 1). Such asymmetry must reflect the occurrence of some essential event in the developing gonad whose probability is so low that it seldom happens even once in an individual, much less twice. The low-probability event responsible for ovarian differentiation in this case must be activation of the self-maintaining splicing feedback loop for Sxlf9 in precursors of the gonadal soma. On the basis of previous results in situations where feedback loop engagement for a given precursor cell was stochastic (Cline 1984, 1985), we expected that engagement in one cell would not influence the probability of engagement in neighboring cells.

If feedback loop engagement were cell-autonomous in these unilateral Sxl mutant ovaries, the simplest explanation for their gross asymmetry would be that engagement happened early in development in a single gonadal soma precursor cell, which then grew much more rapidly and extensively than it otherwise would have, compensating for its neighbors that had not engaged and therefore could not contribute to the differentiated ovary. If that precursor cell were genetically tagged sufficiently early in development, all the somatic cells in the ovary that developed would be its progeny and be tagged. On the other hand, if Sxlf9 feedback loop engagement in this situation were the consequence of a locally cell-nonautonomous process in which a group of neighboring precursor cells participated, the low-probability event instead might be that group collectively reaching a consensus to engage the Sxlf9 feedback loop. The consensus to engage might or might not be triggered by a single cell initially engaging the feedback loop, but one way or another, cell–cell interactions among neighbors would be involved so that the cells that had engaged would not necessarily be clonally related. Whether this cell-nonautonomous model would require compensatory growth of the cells that had engaged would depend on the timing of the engagement event, the fraction of the precursor cells that were involved, and the final size of the differentiated ovary.

The key distinction between these two alternatives is the growth dynamics predicted for the somatic cells in the unilateral ovaries. We measured those dynamics by genetically tagging precursor cells at random in embryos and then scoring the frequency and size of the marked clones in adults (Figure 6). The simple cell-autonomous engagement model requires that the clones in Sxl mutant female ovaries be much less frequent per differentiated ovary than for the cell-nonautonomous model, since in the former case a smaller fraction of presumptive precursor cells available for tagging would have engaged their Sxl feedback loop so they could contribute to the differentiated ovary. Moreover, since the mutant ovaries were over half the size of the controls, by the simple cell-autonomous model the clones produced by the tagged precursor cell should be much larger in the mutant than in control ovaries, since the progeny of that tagged precursor would have to compensate for their untagged neighbors whose failure to engage had kept them from contributing. In contrast, the cell-nonautonomous model makes no such demands regarding clone frequency and clone size.

Figure 6.
Lack of compensatory growth in Sxlf7,M1/Sxlf9 mutant ovaries indicates cell-nonautonomous stimulation of Sxl-positive feedback loop engagement. Histogram indicating the approximate size and frequency of embryonically induced FLP-out clones in the somatic ...

We tagged embryonic precursors of the gonadal soma by administering heat shocks 3–27 hr after egg laying to females carrying a FLP-out GFP cassette and an hsp-FLPase transgene (Pignoni and Zipursky 1997). Analysis of genetic mosaics has argued that, for wild-type females, a single initial embryonic primordium with 8–10 cells contributes somatic cells to both ovaries (Szabad and Nöthiger 1992). In our experiment, these founder cells would have had some chance to divide before being tagged. The high fraction of unilateral ovaries argues that the stochastic event determining whether a mutant ovary can differentiate normally must occur after the primordia for the two ovaries have separated. The frequency of ovarioles whose follicle cells were GFP positive was compared for 30 3-day-old ovaries from Sxlf7,M1/Sxlf9; +/P{snf+} adult females and 30 from their Sxlf9/+; +/P{snf+} sisters. The genetic background for this experiment proved to be optimal for distinguishing between the two models, since the probability of Sxlf9 feedback loop engagement in this case was particularly low: only 3% of the 1120 mutant females dissected had ovaries, 93% of which were unilateral. On the basis of ovariole number, the mutant ovaries were about half the size of control ovaries (mutant median 9.5, mean 9.9, and range 2–16 vs. 17, 16.8, and range 12–22 for controls).

The simple cell-autonomy model's predictions of lower clone frequency and larger clone size were not met: 28 of the 30 mutant ovaries had clones vs. 22 of 30 for the controls. The high frequency of ovaries with clones in both cases indicates that many mutant and wild-type ovaries had more than one clone, but because we could not distinguish between clones within a single ovary, clone size in Figure 6 is simply the total number of ovarioles with tagged follicle cells. The fact that no ovary had all its somatic cells tagged showed that we were far from saturation with respect to clone induction. Both the total number of tagged ovarioles (140 mutant vs. 136 control) and the distribution of those tagged ovarioles among the two sets of 30 ovaries scored were indistinguishable (Wilcoxon rank-sum test P = 0.81).

A more complicated version of the cell-autonomous model that might fit these data can be imagined, although the number of additional assumptions required for such a fit makes it vulnerable to Occam's razor. Perhaps when metamorphosis begins, any significant salt-and-pepper intermingling of unengaged ovarian somatic precursor cells with their engaged counterparts would cause disintegration. By this hypothesis, the rare event generating unilateral ovaries would not be cell-autonomous feedback loop engagement in a single somatic precursor cell early in development. Indeed, stochastic cell-autonomous engagement could occur at a relatively high frequency perhaps even throughout the larval period. Instead, the rare event would be having a sufficiently large number of contiguous engaged precursor cells at metamorphosis to avoid disintegration. Additional assumptions are necessary to account for the fact that so few mutant gonads reach this threshold, yet those few that do go on to generate adult ovaries that range in size from only 12 to 94% of the median for control ovaries.

Early zygotic and even maternal expression of TRA-F rescues Sxl mutant ovaries by stimulating Sxlf9 autoregulation:

In the experiments described in Table 1 showing that constitutive zygotic expression of TRA-F from our u2af-traF transgene rescues mutant ovaries, we were surprised to see evidence of partial rescue even when the transgene was present only maternally (Table 1, compare control cross 3 with the experimental cross 4 line immediately below). Extraneous genetic background differences were unlikely to be responsible, since we had taken pains to minimize them so that cross 3 would be an appropriate control for cross 4. Since the sons in cross 4 who did not inherit the feminizing transgene had normal morphology and were fertile, a maternal effect of this feminizing transgene would have to leave male development unscathed.

The reality of a traF maternal effect was established beyond question when we observed even stronger maternal rescue with the independently isolated transgene, hsp83-traF (Waterbury et al. 2000). This transgene was explicitly designed to express TRA-F at a high level in germ cells. With hsp83-traF, all the daughters who had not inherited the transgene had two normal ovaries (Table 1, cross 5). Again, all the sons without the transgene were normal.

We assayed for traF mRNA in unfertilized eggs to determine whether rescue of the daughters' ovaries could be a direct effect of tra in the mothers' germ cells. Data in Figure 7A show that eggs generated from wild-type mothers clearly do contain female-specific tra mRNA (lane 3). Thus even though females do not need a functional tra allele in their germ cells to make functional eggs (Marsh and Wieschaus 1978), female germ cells appear to transcribe tra and to splice its transcripts in the female pattern. Lane 4 in Figure 7A shows that mothers carrying hsp83-traF load nearly twice as much traF mRNA into their eggs as mothers without the transgene (line 3), measured relative to an actin internal control. Endogenous and transgenic maternal traF RNA is gone at least by the late third instar stage (Figure 7B, lanes 1–3). Curiously, unlike wild-type female larvae and adults, which have a significant amount of non-sex-specific (nonfunctional) tra mRNA (Figure 7A, lane 2, and Figure 7B, lane 4), eggs contain only the female-specific species. Sxlf7,M1/Sxlf9;+/P{snf+} larvae have significantly more of the non-sex-specific splice form than their +/Sxlf9;+/P{snf+} sisters (compare lanes 4 and 5 of Figure 7B), a result consistent with the indications in Figure 4 of their cryptic intersexuality.

Figure 7.
Molecular analysis of tra transcripts in Sxlf7,M1/Sxlf9 mutants. Sex-specific splicing of tra transcripts was assessed by RT–PCR of RNA using the primer pair shown in the schematic (top), with Actin5c as a loading control. Female-specific tra ...

Why would a traF maternal effect sufficient to rescue the ovaries of mutant daughters not interfere with the sons' sexual development? Perhaps the maternal product that stimulates Sxl autoregulation in this highly sensitized situation acts only early in development and does not persist to later stages where males would be affected. Since antibodies to TRA-F are not available, we could not determine directly whether maternally encoded TRA-F protein fails to perdure. Instead, we used female-specific splicing of dsx transcripts as a proxy for TRA-F. Figure 4A, lane 3, shows that, at least by the late third instar stage, males whose mothers carried a traF transgene were indistinguishable from males whose mothers did not: neither had any female dsx splice product. Hence, any TRA-F protein made from maternal mRNA must be gone well before this point.

We used the GAL4 two-component expression system to determine whether full rescue of mutant ovaries by traF could be achieved without a maternal contribution and to explore the question of when zygotic traF mRNA could rescue. The only UAS-traF GAL4 target transgene available for this purpose (Ferveur et al. 1995) is a pUASt construct that is not expected to be expressed well in germ cells (Rørth 1998). Hence, it was not surprising that we failed to mimic the maternal effect of traF transgenes by driving this GAL4 target with a maternal germline source of GAL4 (nos-GAL4-VP16, data not shown). In the course of attempting to determine whether this UAS-traF target would rescue the tra null phenotype, we were surprised to discover that ubiquitous high-level expression of UAS-traF driven by tub-GAL4 greatly reduces the viability of both sexes—especially males—and even partially masculinizes females! This potential for paradoxical behavior limited the utility of UAS-traF for some but fortunately not all of our purposes.

Zygotic expression of traF in the mesoderm was sufficient to fully rescue the Sxl mutant gonadal phenotype. twist-GAL4 and GAL4-24B are mesoderm-specific drivers (Brand and Perrimon 1993; Andrews et al. 2002). Either transgene driving UAS-traF rescued all the ovaries, while no ovaries differentiated properly among sisters carrying only UAS-traF (Table 1, crosses 6 and 7). We tested the fertility of 40 GAL4-24B females whose ovaries had been rescued and discovered that over a 7-day test period, 42% failed to lay eggs. Ovulation, rather than oviposition, was impaired in the nonlaying females. Moreover, only 61% of the laying females produced viable progeny, suggesting that some females may not have mated. We cannot say whether these behavioral problems reflect something interesting about normal Sxl function, or instead just the potential of UAS-traF for paradoxical behavior.

To determine when in development traF expression could rescue Sxl mutant ovaries, we used a GAL4 driver whose ubiquitous expression could be induced by heat shock to drive UAS-traF any time after the blastoderm stage. To induce zygotic traF mRNA even earlier, we used an NGT source of GAL4 (Tracey et al. 2000). Maternal germline expression of this driver activates zygotic UAS targets as early as nuclear cycle 11 (Ten Bosch 2006). Data in Table 4 show that only pulses of traF generated during the first 6 hr of embryogenesis will rescue the Sxl mutant ovaries. The fact that pulses of TRA-F in very young embryos prevented their Sxl mutant ovaries from exhibiting differentiation defects during metamorphosis is compelling evidence that rescue is a consequence of TRA-F stimulating Sxlf9 feedback loop engagement, a self-maintaining event.

Rescue of Sxlf7,M1/Sxlf9; +/P{snf+} ovaries by a traF pulse as a function of developmental age

If maternally deposited traF rescues mutant ovaries by triggering Sxlf9 autoregulatory feedback loop engagement early in development, we would expect this maternal effect to eliminate the male dsx mRNA that we had seen in immature Sxl mutant ovaries from late third instar larvae (Figure 4A, lane 5). Data in Figure 4B, lane 3 (compare to lane 2), show that indeed it does. Moreover, elimination of dsxM RNA in these female larvae must be due to a much earlier effect of TRA-F, rather than to perdurance of maternally encoded TRA-F protein, since the brothers of these female larvae lacked the dsxF mRNA that perduring TRA-F protein would generate.

If the stimulation of Sxlf9 autoregulation in the gonadal soma by increased TRA-F reflects a normal activity of tra rather than an artifact of TRA-F overexpression, then decreasing TRA-F should have the opposite effect. Data in Table 5 show that it does. In this experiment, 18% of Sxl mutant daughters had differentiated ovaries when both they and their mothers carried two copies of tra+ (cross 1). That number dropped to 9% when mothers were heterozygous for tra+ but daughters were homozygous (P < 0.01) and to 0% when both mothers and daughters were heterozygous (cross 2; P < 0.01). As in all other experiments designed to reveal maternal effects, the mothers used for this comparison were sisters, and the relevant chromosomes had been allowed to freely recombine to homogenize the genetic background.

Enhancement of Sxlf7,M1/Sxlf9; +/P{snf+} ovary disintegration by reduction in maternal and/or zygotic tra+ dose

TRA-F affects Sxl autoregulation even outside the gonad:

An effect of TRA-F on the viability of Sxlf7,M1/Sxlf9 females would indicate effects on Sxlf9 autoregulation outside the gonad. Data in Tables 1 and and55 show that there are such effects but their magnitude is small relative to those in the gonad. Recall that the viability of Sxlf7,M1/Sxlf9 females without an extra copy of snf+ is very low. Raising the level of traF both maternally and zygotically increased their relative viability fivefold (Table 1, crosses 1 and 2). However, even with this additional TRA-F activity, 95% of the females died. Moreover, the rescuing effect of traF disappeared when the viability of Sxlf7,M1/Sxlf9 females was raised by any additional copy of snf+ (compare crosses 3 and 4). Decreasing the level of traF activity from the endogenous alleles led to a statistically significant opposite effect on viability: it dropped from 82% when both mothers and daughters had two tra+ alleles to 57% when both had only one (Table 5).

If TRA-F protein can stimulate Sxl autoregulation in nongonadal tissues even weakly, the stimulation should also be apparent in Sxlf7,M1/Y males that carry Sxl+ and an extra copy of snf+. Like Sxlf7,M1/Sxlf9; +/P{snf+} females, these males are sensitized to effects on Sxl autoregulation, but stimulation for them will be deleterious rather than beneficial. Sxlf7,M1/Y males can tolerate a copy of Sxl+ or an extra copy of snf+individually, but when presented with both, the probability of Sxlf7,M1 engaging the Sxl+ allele's feedback loop becomes significant and viability drops dramatically (Cline et al. 1999). Data in cross 2 of Table 6 illustrate this point. Extra copies of snf+and Sxl+ that had no significant adverse effect on Sxlf7,M1/Y male viability individually (109 and 96%, respectively) together reduced viability to 11%.

TRA-F can kill sensitized males by stimulating Sxl autoregulation

Table 6, cross 1, shows that the hsp83-traF transgene reduced the viability of these sensitized males from 16% without traF to 0.8% with it. There was no maternal effect of traF on these males: their viability was no lower when their mothers carried hsp83-traF (16%) than when their mothers did not (11%). Sxlf7,M1 males not carrying Sxl+ were unaffected by traF, whether or not they carried an extra copy of snf+; however, when Sxl+was present, hsp83-traF had a small but significant effect even without the extra copy of snf+.

Conserved sequences near the Sxl male-specific exon point to direct stimulation of Sxl by TRA-F:

The consensus binding site for TRA-F protein is TC[t/a][t/a]C[a/g]ATCAACA (Hoshijima et al. 1991). Anticipating that the effect of TRA-F on Sxl autoregulation would be indirect, we were stunned to discover a highly conserved tandem pair of TRA-F sites in the sex-specifically regulated region of Sxl (Figure 8). One site is a perfect match to the consensus and is located just 422 bp upstream of the male-specific exon 3 in D. melanogaster. The other site, only 3 bp upstream of the first, is a 1-bp (first position) degenerate sequence. There are no TRA-F consensus sites, or even 1-bp degenerates, anywhere else in Sxl. Although it had been believed that the eight different 13-mers designated as TRA-F sites were functionally equivalent, we found that the one 13-mer in melanogaster Sxl was matched exactly at a comparable position in all but 1 of the 12 published Drosophila species genomes (Drosophila 12 Genomes Consortium 2007) and in Scaptodrosophila lebanonensis, a fly just outside the Drosophila genus (Powell and Moriyama 1997) whose Sxl locus we partially sequenced (EU670259). The one exception is D. ananassae, which has only a single 1-bp (first position) degenerate sequence in the region. All species except lebanonensis and virilis also have a 1-bp (first position) degenerate sequence very near the consensus site. This degenerate sequence is missing in lebanonensis, but in virilis it is replaced by another exact copy of the conserved consensus sequence. A perfect match to this TRA-F consensus site occurs only 34 other times in the D. melanogaster genome. Just as striking, all the various 1-bp degenerate sites shown in Figure 8 together occur only 42 times in melanogaster. The presence of such conserved sites so strategically positioned in the sex-specifically spliced region of Sxl argues that the stimulation of Sxl-positive autoregulation by TRA-F is direct at some point and occurs by the inhibition of splicing to the male exon.

Figure 8.
Highly conserved putative TRA-F protein-binding sites are present exclusively in the sex-specifically regulated region of Sxl. The region between the non-sex-specific Sxl exon 2 and the male-specific, translation terminating exon 3 is shown to scale for ...


It had been believed that the regulatory relationship in Drosophila between the sex-determination switch gene Sxl and one of its immediate downstream targets, tra, was simply one of master to slave. Our effort to understand the developmental basis for an unusual female-sterile phenotype observed in a highly contrived genetic situation revealed that the regulatory relationship between Sxl and tra is not so simple: the slave influences its master and thereby affects itself. Although we could see evidence of this indirect positive feedback of tra through Sxl only in genetically sensitized situations, our discovery that TRA-F binding sites in Sxl have been evolutionarily conserved for >45 million years argues that this unanticipated feedback loop has significance for wild-type Drosophila. Our study generated three additional surprises: (1) intersexuality can cause Drosophila tissues to disintegrate; (2) not all aspects of somatic Sxl regulation may be cell-autonomous; and (3) female-specific tra mRNA is present in unfertilized eggs, with tra exerting a maternal effect on Sxl.

The discovery of a “functionally redundant” positive feedback loop for Sxl that operates through tra—a regulatory circuit that can equally well be thought of as a nonredundant positive feedback loop for tra that operates through Sxl—is of evolutionary interest in light of the fact that, in Tephritid flies, tra occupies the position that Sxl holds in Drosophila: the master sex switch gene that epigenetically maintains the female developmental commitment through direct positive feedback control of its own pre-mRNA splicing (Pane et al. 2002). Very recent work suggests that the honeybee tra ortholog may function like tra in the Tephritids (Hasselmann et al. 2008). In flies for which tra is the master sex switch, tra is expressed in the female germline and that germline expression has been proposed to be an important element in the sex-determination mechanism by which the tra feedback loop is initiated (Lagos et al. 2007). We suggest that both the functionally redundant, tra-mediated circuit for Sxl-positive feedback in D. melanogaster and female germline expression of tra in melanogaster are vestiges of the evolutionary transition between tra and Sxl as the autoregulated master sex switch.

Ambiguities in somatic cell sexual identity, not upsets in dosage compensation, cause Sxlf7,M1/Sxlf9; +/P{snf+} ovaries to disintegrate during the pupal stage:

We were led to the discovery of tra's effects on Sxl in the course of investigating why some Sxl mutant ovaries disintegrate. While precedents exist for genetic imbalances caused by upsets in X chromosome dosage compensation disrupting somatic cell growth and differentiation (Cline 1976; Tanaka et al. 1976; Belote 1983), upsets in the functioning of sexual differentiation switch genes downstream of Sxl, such as tra and dsx that are not involved in X-chromosome dosage compensation, had not been reported to cause such disruptions. Although the gonads of intersexual and transsexual flies generated by mutations in such genes are certainly abnormal, they are easily recognized as gonads. For that reason, we were surprised to discover that the disintegration of Sxlf7,M1/Sxlf9; +/P{snf+} ovaries during metamorphosis is caused by problems with sexual identity, not by dosage compensation problems.

Although Sxlf7,M1/Sxlf9; +/P{snf+} mutant ovaries appear morphologically normal prior to metamorphosis, we found them to be intersexual at the molecular level even at this early, rather quiescent stage. We hypothesize that their disintegration soon after the onset of metamorphosis is a consequence of a sudden increase in the level of dsxF required to promote the rapid female differentiation and counteract any dsxM present, a level that cannot be reached by these mutant cells that have not fully engaged the Sxlf9 positive feedback loop. In this connection it may be relevant that Le Bras and Van Doren (2006) reported that the gonad's requirements for dsxF and dsxM are qualitatively different during embryonic development than at later stages. We can imagine that a sudden shift in sexual identity during a period of rapid differentiation might be more disruptive than an intersexual orientation held consistently throughout development, such as that imposed by a loss of the dsx or ix genes. Another potential factor that might contribute to the ovaries' disintegration is phenotypic sexual mosaicism, reflecting cells that had fully engaged the Sxlf9 feedback loop mixed with cells that had not.

Cell–cell interactions may trigger Sxl feedback loop engagement in the gonadal soma:

All previous evidence had indicated that the regulation of Sxl in the soma is strictly cell-autonomous (discussed in Cline and Meyer 1996). Nevertheless, the lack of compensatory growth of somatic cells in the unilateral ovaries of Sxlf7,M1/Sxlf9; P{snf+}/+ adults, and the wild-type frequency at which those somatic cells could be genetically tagged, showed that the rare stochastic event that allowed these exceptional ovaries to differentiate properly must have involved more than a single somatic stem-cell precursor. We believe that the simplest explanation for these results is that normal ovarian differentiation occurs when full Sxlf9 feedback loop engagement in one somatic gonadal precursor cell stimulates feedback loop engagement in neighboring somatic cells poised just under the threshold for engagement. As mentioned in the results, we cannot exclude a rather complex cell-autonomous alternative model for Sxlf9 feedback loop engagement, but we favor the cell-nonautonomous alternative since female somatic cells in the gonad have already been shown to trigger female expression of Sxl in neighboring germ cells that have been “sensitized” by having two X chromosomes rather than one (Steinmann-Zwicky et al. 1989). Indeed, Evans and Cline (2007) showed that forced expression of TRA-F even in chromosomally male gonadal soma with no SXL-F is sufficient to stimulate SXL-F expression in neighboring diplo-X germ cells to a level adequate for producing fully functional eggs. Hence, our cell-nonautonomous model has these somatic cells simply doing to each other what they are already known to do to germ cells.

Cell-nonautonomy in the sexual differentiation of Drosophila gonads was first reported by Fung and Gowen (1957) and shown most recently by Defalco et al. (2008). However, because none of these examples seemed to involve effects on the sexually determined state of those cells in sensu stricto—the functional state of Sxl—there was no reason a priori to expect autonomous behavior more than nonautonomous behavior. In contrast, cell-nonautonomy for Sxl in the gonadal soma would be more surprising because Sxl is clearly capable of maintaining its activity state within somatic cells without outside input and because Sxl controls somatic dosage compensation. This link to dosage compensation makes it essential that Sxl's expression state in each somatic cell be appropriate for that cell's own X-chromosome dose. For that reason we suspect that the influence of any cell-nonautonomy in this case would be relatively minor—able to increase the fidelity of the process by which the Sxl feedback loop is first engaged in diplo-X somatic cells, but not able to affect haplo-X neighbor cells in a genetic mosaic.

Conserved TRA-F binding sites in Sxl argue that the indirect tra positive feedback circuit is functionally relevant:

The discovery of highly conserved predicted TRA-F binding sites in Sxl indicates that some effect of tra on Sxl must be direct and that this direct effect must be relevant to wild-type flies. While it is formally possible that a cell-nonautonomous effect of TRA-F on Sxl could be mediated by transport of this SR-like splicing factor between cells, such transport would be unprecedented. It seems more likely that the direct effect of TRA-F on Sxl is mechanistically distinct from cell-nonautonomous effects. A direct effect might occur either in zygotic cells that have sequestered maternally synthesized traF product or in cells that have transcribed tra themselves at a point in development when they have enough SXL-F to direct some female tra transcript splicing. Since maternal traF does not disrupt male zygotic development, we can deduce that maternal traF products must not persist. That they need not perdure beyond embryogenesis was indicated by our finding that a pulse of TRA-F generated zygotically can rescue Sxlf7,M1/Sxlf9; +/P{snf+} ovaries well only within 6 hr after egg laying and does not rescue at all after embryogenesis. We propose that the direct effect of TRA-F on Sxl in the gonadal soma influences the initial probability of at least one somatic gonadal precursor cell fully engaging the Sxlf9 feedback loop, but that once one such cell has fully engaged, it stimulates engagement in its somatic neighbors through the same indirect, cell-nonautonomous mechanism by which these cells stimulate Sxl in the germline.

Since we have not yet eliminated tra+ from both the maternal germline and the zygote, the question remains whether such a double loss of tra would affect Sxl autoregulation even in females that had not been genetically sensitized by mutations in Sxl. Marsh and Wieschaus (1978) went part way toward answering that question by transplanting tra female germ cells into tra+ female embryos and mating the resulting chimeric adults to tra males. Since the masculinized tra daughters produced by these chimeric mothers were viable and not malformed, one can infer that Sxl+ activation was normal in most tissues. Nevertheless, since the gonads seem to have not been examined, the question remains whether the aspect of development that seems most sensitive to the effect of tra on Sxl was in fact unaffected. Now that we know that traF transcripts are present in germ cells and can affect Sxl, and know as well that Sxl is required for meiotic recombination (Cook 1993; T. W. Cline, unpublished results; and see Bopp et al. 1999), we need to examine more closely tra germ cells and the eggs that they generate.

The predicted TRA-F binding site sequences in Sxl, as well as their positions relative to the Sxl male exon, are remarkably conserved. The location of those sites argues that TRA-F binding facilitates Sxl-positive autoregulation by directly inhibiting use of the splice acceptor for the male-specific Sxl exon. Although TRA-F has been shown to function only as a splicing activator in D. melanogaster (Hoshijima et al. 1991; Heinrichs et al. 1998), in Tephritid flies it appears to promote the female-specific pattern of tra pre-mRNA splicing by directly inhibiting use of male-specific splice acceptor sites (Ruiz et al. 2007).

It is intriguing that, of the 13 fly species whose Sxl sequence we examined, only D. ananassae did not have at least one perfect match to the 13-bp TRA-F consensus binding sequence near the male-specific exon. D. ananassae is far more closely related to D. melanogaster than many other species in this group; however, as Singh (2000) points out, ananassae is an unusual Drosophila species in many respects, including having a high rate of male genetic recombination. Meiotic recombination is generally a female-specific function in Drosophila, one that requires Sxl (Cook 1993; T. W. Cline, unpublished results; and see Bopp et al. 1999). Perhaps the acquisition by ananassae males of a function normally limited to females is related to the loss of TRA-F binding sites in Sxl.

Redundant positive autoregulation and the evolution of a new master sex switch:

Many years have passed since D. melanogaster was discovered to maintain its female sexual fate decision epigenetically through the operation of a direct positive feedback loop for Sxl pre-mRNA splicing. Surprisingly, in all that time, the Tephritid ortholog of tra is the only other gene found to maintain a cell fate decision by this mechanism. Tephritidae, the “true” fruit flies, are evolutionary cousins of the Drosophilidae. Sxl orthologs have also been identified in the Tephritidae (Saccone et al. 1998) and in insects as distant as honeybees (Dearden et al. 2006) and moths (Niimi et al. 2006); nevertheless, only in the Drosophilidae does Sxl appear to be sex-specifically regulated (Bopp et al. 1996; Dorsett et al. 2003). The view that in the Drosophilidae, Sxl took over tra's ancestral developmental role is supported by the recent discovery and experimental manipulation of feminizer (fem), the honeybee tra ortholog (Hasselmann et al. 2008). The possibility that fem is a positively autoregulating master sex switch is supported by the observation that injection into extremely young female embryos of double-stranded RNA directed against the fem regulator complementary sex determiner (csd) induced those embryos to develop into fertile adult males. Our discovery that TRA-F positively influences Sxl autoregulation in D. melanogaster shows that, even in the Drosophilidae, a positive autoregulatory loop exists for tra, albeit one that is weaker and less direct than that in the Tephritidae. We propose that functional redundancy with respect to positive autoregulation was important in the evolutionary transition between tra and Sxl as the autoregulating master sex switch gene.

For modeling that evolutionary transition, it would be helpful to know with more certainty whether direct positive autoregulation of tra is the ancestral condition. Hoping to answer this question, we sequenced a 5.7-kb region from 1.9 kb upstream to 3.0 kb downstream of the putative tra ORF for S. cynipsea (EU636097) (supplemental Figure S1). The family Sepsidae is closer to the Drosophilidae than to the Tephritidae (Yeates and Wiegmann 1999). Finding TRA-F binding sites in the appropriate region of S. cynipsea tra would indicate direct tra positive autoregulation and hence point to tra autoregulation as being ancestral. The absence of such sites would leave the question open.

Our results leave the question open. Even allowing up to a 3-bp degeneracy in the established 13-bp TRA-F consensus binding site, we found only two overlapping candidate sites in S. cynipsea tra, neither of which seemed likely to be functional: gaATCAATCAACA and TCAA CAATCgcCA (degenerate sites lower case and overlap underlined). TRA-F from D. melanogaster had been shown to not utilize the first sequence (Inoue et al. 1992), and the second sequence lacks the TCAAxx 3′ sequence common to every predicted TRA-F binding site for tra, dsx, and fru orthologs all the way to mosquitos (Pane et al. 2002; Scali et al. 2005; Lagos et al. 2007). Moreover, we sequenced a 722-bp fragment of tra from the sister species S. neocynipsea (EU636098) centered on homology to the S. cynipsea degenerate TRA-F sites and found that neither site was conserved, despite 85% overall sequence identity for the two corresponding regions. The only sequence in S. neocynipsea resembling the TRA-F consensus was a single, different 2-bp degenerate.

Functional redundancy arising by gene duplication has long been considered a basic raw material of evolution (reviewed in True and Carroll 2002). Indeed, duplication of the tra ortholog fem to generate csd in an ancestor of the honeybee appears to have led to the evolution of a new sex-determination signal (Hasselmann et al. 2008). Until very recently with the discovery by Hong et al. (2008) of “shadow enhancers,” much less attention seems to have been given to the possible evolutionary role of functional redundancy in regulatory circuits, such as we have discovered between Sxl and tra.

One can imagine an evolutionary route from tra to Sxl as the autoregulated master sex switch gene on the basis of the sequential development and loss of functional redundancy in positive autoregulation (Figure 9). First, Sxl would become a TRA-F splicing target by acquiring TRA-F binding sites (Figure 9B). The fact that D. melanogaster Sxl has such sites today makes this possibility more plausible than it would otherwise be. Next, tra would acquire SXL-F binding sites that facilitate female tra pre-mRNA splicing, such as those it has today, thereby establishing a functionally redundant circuit for tra-positive autoregulation (Figure 9C). With two positive feedback pathways operating for tra, the indirect circuit running through Sxl could become stronger while the direct circuit weakened (Figure 9D) and eventually disappears (Figure 9E). That disappearance may have been hastened by Sxl's acquisition of its own SXL-F binding sites that generated a weak but direct autoregulatory circuit (Figure 9E). At that point, Sxl rather than tra would be the gene with functionally redundant positive autoregulatory circuits, one direct and one indirect via tra. Again, the circuit that was weaker initially could strengthen as the other weakened, producing the situation that exists today (Figure 9F).

Figure 9.
How regulatory functional redundancy might have facilitated an evolutionary transition from tra to Sxl as the autoregulating master sex-determination switch gene. A hypothetical stepwise scenario is illustrated beginning with the current situation in ...


We thank Brant Peterson of the M. B. Eisen lab for help with the Sepsid tra analysis; Satoru Uzawa of the B. J. Meyer lab for help with confocal microscopy; R. Lehmann for fly stocks; and B. J. Meyer and current members of the Cline lab for helpful discussions and comments on the manuscript. This work was supported by National Institutes of Health grant GM023468 to T.W.C. and a National Science Foundation graduate fellowship to S.G.S.


Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. EU670259, EU636097, and EU636098.


  • Andrews, H. K., Y. Q. Zhang, N. Trotta and K. Broadie, 2002. Drosophila sec10 is required for hormone secretion but not general exocytosis or neurotransmission. Traffic 3 906–921. [PubMed]
  • Bai, J., and D. Montell, 2002. Eyes absent, a key repressor of polar cell fate during Drosophila oogenesis. Development 129 5377–5388. [PubMed]
  • Bainbridge, S. P., and M. Bownes, 1981. Staging the metamorphosis of Drosophila melanogaster. J. Embryol. Exp. Morphol. 66 57–80. [PubMed]
  • Baker, B. S., and K. A. Ridge, 1980. Sex and the single cell. I. On the action of major loci affecting sex determination in Drosophila melanogaster. Genetics 94 383–423. [PMC free article] [PubMed]
  • Bell, L. R., J. I. Horabin, P. Schedl and T. W. Cline, 1991. Positive autoregulation of Sex-lethal by alternative splicing maintains the female determined state in Drosophila. Cell 65 229–239. [PubMed]
  • Belote, J. M., 1983. Male-specific lethal mutations of Drosophila melanogaster. II. Parameters of gene action during male development. Genetics 105 881–896. [PMC free article] [PubMed]
  • Bernstein, M., and T. W. Cline, 1994. Differential effects of Sex-lethal mutations on dosage compensation early in Drosophila development. Genetics 136 1051–1061. [PMC free article] [PubMed]
  • Bernstein, M., R. A. Lersch, L. Subrahmanyan and T. W. Cline, 1995. Transposon insertions causing constitutive Sex-lethal activity in Drosophila melanogaster affect Sxl sex-specific transcript splicing. Genetics 139 631–648. [PMC free article] [PubMed]
  • Beye, M., M. Hasselmann, M. K. Fondrk, R. E. Page and S. W. Omholt, 2003. The gene csd is the primary signal for sexual development in the honeybee and encodes an SR-type protein. Cell 114 419–429. [PubMed]
  • Bonini, N. M., W. M. Leiserson and S. Benzer, 1993. The eyes absent gene: genetic control of cell survival and differentiation in the developing Drosophila eye. Cell 72 379–395. [PubMed]
  • Bopp, D., J. I. Horabin, R. A. Lersch, T. W. Cline and P. Schedl, 1993. Expression of the Sex-lethal gene is controlled at multiple levels during Drosophila oogenesis. Development 118 797–812. [PubMed]
  • Bopp, D., G. Calhoun, J. I. Horabin, M. Samuels and P. Schedl, 1996. Sex-specific control of Sex-lethal is a conserved mechanism for sex determination in the genus Drosophila. Development 122 971–982. [PubMed]
  • Bopp, D., C. Schutt, J. Puro, H. Huang and R. Nöthiger, 1999. Recombination and disjunction in female germ cells of Drosophila depend on the germline activity of the gene Sex-lethal. Development 126 5785–5794. [PubMed]
  • Boswell, R. E., and A. P. Mahowald, 1985. tudor, a gene required for assembly of the germ plasm in Drosophila melanogaster. Cell 43 97–104. [PubMed]
  • Boyle, M., and S. Dinardo, 1995. Specification, migration and assembly of the somatic cells of the Drosophila gonad. Development 121 1815–1825. [PubMed]
  • Brand, A. H., and N. Perrimon, 1993. Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development 118 401–415. [PubMed]
  • Cline, T. W., 1976. A sex-specific, temperature-sensitive maternal effect of the daughterless mutation of Drosophila melanogaster. Genetics 84 723–742. [PMC free article] [PubMed]
  • Cline, T. W., 1979. A male-specific lethal mutation in Drosophila melanogaster that transforms sex. Dev. Biol. 72 266–275. [PubMed]
  • Cline, T. W., 1983. a Functioning of the genes daughterless (da) and Sex-lethal (Sxl) in Drosophila germ cells. Genetics 104 s16–s17.
  • Cline, T. W., 1983. b The interaction between daughterless and Sex-lethal in triploids: a lethal sex-transforming maternal effect linking sex determination and dosage compensation in Drosophila melanogaster. Dev. Biol. 95 260–274. [PubMed]
  • Cline, T. W., 1984. Autoregulatory functioning of a Drosophila gene product that establishes and maintains the sexually determined state. Genetics 107 231–277. [PMC free article] [PubMed]
  • Cline, T. W., 1985. Primary events in the determination of sex in Drosophila melanogaster, pp. 301–327 in Origin and Evolution of Sex, edited by H. Halvorson and A. Monroy. A. R. Liss, New York.
  • Cline, T. W., 1986. A female-specific lethal lesion in an X-linked positive regulator of the Drosophila sex determination gene, Sex-lethal. Genetics 113 641–663. [PMC free article] [PubMed]
  • Cline, T. W., 2005. Reflections on a path to sexual commitment. Genetics 169 1179–1185. [PMC free article] [PubMed]
  • Cline, T. W., and B. J. Meyer, 1996. Vive la différence: males vs. females in flies vs. worms. Annu. Rev. Genet. 30 637–702. [PubMed]
  • Cline, T. W., D. Z. Rudner, D. A. Barbash, M. Bell and R. Vutien, 1999. Functioning of the Drosophila integral U1/U2 protein Snf independent of U1 and U2 small nuclear ribonucleoprotein particles is revealed by snf+ gene dose effects. Proc. Natl. Acad. Sci. USA 96 14451–14458. [PMC free article] [PubMed]
  • Cook, K. R., 1993. Regulation of recombination and oogenesis by the ovarian tumor, Sex-lethal, and ovo genes of Drosophila melanogaster. Ph.D. Thesis, University of Iowa, Iowa City, IA.
  • Dearden, P. K., M. J. Wilson, L. Sablan, P. W. Osborne, M. Havler et al., 2006. Patterns of conservation and change in honey bee developmental genes. Genome Res. 16 1376–1384. [PMC free article] [PubMed]
  • DeFalco, T., N. Camara, S. Le Bras and M. Van Doren, 2008. Nonautonomous sex determination controls sexually dimorphic development of the Drosophila gonad. Dev. Cell 14 275–286. [PMC free article] [PubMed]
  • Dorsett, M., L. A. Megna and T. W. Cline, 2003. Sex-lethal is a female-specific, sex-determination gene in Drosophila virilis, p. a307 in 44th Annual Drosophila Conference Abstracts. Genetics Society of America, Chicago.
  • Drosophila 12 Genomes Consortium, 2007. Evolution of genes and genomes on the Drosophila phylogeny. Nature 450 203–218. [PubMed]
  • Erickson, J. W., and J. J. Quintero, 2007. Indirect effects of ploidy suggest X chromosome dose, not the X:A ratio, signals sex in Drosophila. PLoS Biol. 5 e332. [PMC free article] [PubMed]
  • Evans, D. S., and T. W. Cline, 2007. Drosophila melanogaster male somatic cells feminized solely by TraF can collaborate with female germ cells to make functional eggs. Genetics 175 631–642. [PMC free article] [PubMed]
  • Ferveur, J. F., K. F. Stortkuhl, R. F. Stocker and R. J. Greenspan, 1995. Genetic feminization of brain structures and changed sexual orientation in male Drosophila. Science 267 902–905. [PubMed]
  • Forbes, A. J., A. C. Spradling, P. W. Ingham and H. Lin, 1996. The role of segment polarity genes during early oogenesis in Drosophila. Development 122 3283–3294. [PubMed]
  • Fung, S. T., and J. W. Gowen, 1957. Pigment-inducing potentialities of testes, ovaries and hermaphroditic (hr) gonads. J. Exp. Zool. 135 5–18. [PubMed]
  • Granadino, B., M. Torres, D. Bachiller, E. Torroja, J. L. Barbero et al., 1991. Genetic and molecular analysis of new female-specific lethal mutations at the gene Sxl of Drosophila melanogaster. Genetics 129 371–383. [PMC free article] [PubMed]
  • Hager, J. H., and T. W. Cline, 1997. Induction of female Sex-lethal RNA splicing in male germ cells: implications for Drosophila germline sex determination. Development 124 5033–5048. [PubMed]
  • Han, C. S., R. D. Sutherland, P. B. Jewett, M. L. Campbell, L. J. Meincke et al., 2000. Construction of a BAC contig map of chromosome 16q by two-dimensional overgo hybridization. Genome Res. 10 714–721. [PMC free article] [PubMed]
  • Hasselmann, M., T. Gempe, M. Schiøtt, C. G. Nunes-Silva, M. Otte et al., 2008. Evidence for the evolutionary nascence of a novel sex determination pathway in honeybees. Nature 454 519–522. [PubMed]
  • Heinrichs, V., L. C. Ryner and B. S. Baker, 1998. Regulation of sex-specific selection of fruitless 5′ splice sites by transformer and transformer-2. Mol. Cell. Biol. 18 450–458. [PMC free article] [PubMed]
  • Hong, J-W., D. A. Hendrix and M. S. Levine, 2008. Shadow enhancers as a source of evolutionary novelty. Science 321 1314. [PMC free article] [PubMed]
  • Hoshijima, K., K. Inoue, I. Higuchi, H. Sakamoto and Y. Shimura, 1991. Control of doublesex alternative splicing by transformer and transformer-2 in Drosophila. Science 252 833–836. [PubMed]
  • Inoue, K., K. Hoshijima, I. Higuchi, H. Sakamoto and Y. Shimura, 1992. Binding of the Drosophila Transformer and Transformer-2 proteins to the regulatory elements of doublesex primary transcript for sex-specific RNA processing. Proc. Natl. Acad. Sci. USA 89 8092–8096. [PMC free article] [PubMed]
  • Janzer, B., and M. Steinmann-Zwicky, 2001. Cell-autonomous and somatic signals control sex-specific gene expression in XY germ cells of Drosophila. Mech. Dev. 100 3–13. [PubMed]
  • Kelley, R. L., I. Solovyeva, L. M. Lyman, R. Richman, V. Solovyev et al., 1995. Expression of msl-2 causes assembly of dosage compensation regulators on the X chromosomes and female lethality in Drosophila. Cell 81 867–877. [PubMed]
  • Keyes, L. N., T. W. Cline and P. Schedl, 1992. The primary sex determination signal of Drosophila acts at the level of transcription. Cell 68 933–943. [PubMed]
  • Lagos, D., M. Koukidou, C. Savakis and K. Komitopoulou, 2007. The transformer gene in Bactrocera oleae: the genetic switch that determines its sex fate. Insect Mol. Biol. 16 221–230. [PubMed]
  • Laugé, 1980. Sex determination, pp. 33–106 in The Genetics and Biology of Drosophila, edited by M. Ashburner and T. Wright. Academic Press, New York.
  • Le Bras, S., and M. Van Doren, 2006. Development of the male germline stem cell niche in Drosophila. Dev. Biol. 294 92–103. [PubMed]
  • Maine, E. M., H. K. Salz, P. Schedl and T. W. Cline, 1985. Sex-lethal, a link between sex determination and sexual differentiation in Drosophila melanogaster. Cold Spring Harb. Symp. Quant. Biol. 50 595–604. [PubMed]
  • Marsh, J. L., and E. Wieschaus, 1978. Is sex determination in germ line and soma controlled by separate genetic mechanisms? Nature 272 249–251. [PubMed]
  • McRobert, S. P., and L. Tompkins, 1985. The effect of transformer, doublesex and intersex mutations on the sexual behavior of Drosophila melanogaster. Genetics 111 89–96. [PMC free article] [PubMed]
  • Nagoshi, R. N., and B. S. Baker, 1990. Regulation of sex-specific RNA splicing at the Drosophila doublesex gene: cis-acting mutations in exon sequences alter sex-specific RNA splicing patterns. Genes Dev. 4 89–97. [PubMed]
  • Nagoshi, R. N., M. McKeown, K. C. Burtis, J. M. Belote and B. S. Baker, 1988. The control of alternative splicing at genes regulating sexual differentiation in D. melanogaster. Cell 53 229–236. [PubMed]
  • Niimi, T., K. Sahara, H. Oshima, Y. Yasukochi, K. Ikeo et al., 2006. Molecular cloning and chromosomal localization of the Bombyx Sex-lethal gene. Genome 49 263–268. [PubMed]
  • Nöthiger, R., M. Jonglez, M. Leuthold, P. Meier-Gerschwiler and T. Weber, 1989. Sex determination in the germ line of Drosophila depends on genetic signals and inductive somatic factors. Development 107 505–518. [PubMed]
  • Oliver, B., 2002. Genetic control of germline sexual dimorphism in Drosophila. Int. Rev. Cytol. 219 1–60. [PubMed]
  • Oliver, B., Y. J. Kim and B. S. Baker, 1993. Sex-lethal, master and slave: a hierarchy of germ-line sex determination in Drosophila. Development 119 897–908. [PubMed]
  • Pane, A., M. Salvemini, P. Delli Bovi, C. Polito and G. Saccone, 2002. The transformer gene in Ceratitis capitata provides a genetic basis for selecting and remembering the sexual fate. Development 129 3715–3725. [PubMed]
  • Patel, N. H., E. Martin-Blanco, K. G. Coleman, S. J. Poole, M. C. Ellis et al., 1989. Expression of engrailed proteins in arthropods, annelids, and chordates. Cell 58 955–968. [PubMed]
  • Pignoni, F., and S. L. Zipursky, 1997. Induction of Drosophila eye development by decapentaplegic. Development 124 271–278. [PubMed]
  • Powell, J. R., and E. N. Moriyama, 1997. Evolution of codon usage bias in Drosophila. Proc. Natl. Acad. Sci. USA 94 7784–7790. [PMC free article] [PubMed]
  • Rørth, P., 1998. Gal4 in the Drosophila female germline. Mech. Dev. 78 113–118. [PubMed]
  • Ruiz, M. F., A. Milano, M. Salvemini, J. M. Eirin-Lopez, A. L. Perondini et al., 2007. The gene transformer of anastrepha fruit flies (Diptera, tephritidae) and its evolution in insects. PLoS ONE 2 e1239. [PMC free article] [PubMed]
  • Saccone, G., I. Peluso, D. Artiaco, E. Giordano, D. Bopp et al., 1998. The Ceratitis capitata homologue of the Drosophila sex-determining gene Sex-lethal is structurally conserved but not sex-specifically regulated. Development 125 1495–1500. [PubMed]
  • Scali, C., F. Catteruccia, Q. Li and A. Crisanti, 2005. Identification of sex-specific transcripts of the Anopheles gambiae doublesex gene. J. Exp. Biol. 208 3701–3709. [PMC free article] [PubMed]
  • Schüpbach, T., 1985. Normal female germ cell differentiation requires the female X chromosome to autosome ratio and expression of Sex-lethal in Drosophila melanogaster. Genetics 109 529–548. [PMC free article] [PubMed]
  • Singh, B., 2000. Drosophila ananassae: a species characterized by several unusual genetic features. Curr. Sci. 78 391–398.
  • Sosnowski, B. A., J. M. Belote and M. McKeown, 1989. Sex-specific alternative splicing of RNA from the transformer gene results from sequence-dependent splice site blockage. Cell 58 449–459. [PubMed]
  • Starr, D. J., and T. W. Cline, 2002. A host parasite interaction rescues Drosophila oogenesis defects. Nature 418 76–79. [PubMed]
  • Steinmann-Zwicky, M., 1993. Sex determination in Drosophila: sis-b, a major numerator element of the X:A ratio in the soma, does not contribute to the X:A ratio in the germ line. Development 117 763–767. [PubMed]
  • Steinmann-Zwicky, M., 1994. Sxl in the germline of Drosophila: a target for somatic late induction. Dev. Genet. 15 265–274. [PubMed]
  • Steinmann-Zwicky, M., H. Schmid and R. Nöthiger, 1989. Cell-autonomous and inductive signals can determine the sex of the germ line of Drosophila by regulating the gene Sxl. Cell 57 157–166. [PubMed]
  • Szabad, J., and R. Nöthiger, 1992. Gynandromorphs of Drosophila suggest one common primordium for the somatic cells of the female and male gonads in the region of abdominal segments 4 and 5. Development 115 527–533. [PubMed]
  • Tanaka, A., A. Fukunaga and K. Oishi, 1976. Studies on the sex-specific lethals of Drosophila melanogaster. II. Further studies on a male-specific lethal gene, maleless. Genetics 84 257–266. [PMC free article] [PubMed]
  • Taylor, B. J., and J. W. Truman, 1992. Commitment of abdominal neuroblasts in Drosophila to a male or female fate is dependent on genes of the sex-determining hierarchy. Development 114 625–642. [PubMed]
  • Taylor, B. J., A. Villella, L. C. Ryner, B. S. Baker and J. C. Hall, 1994. Behavioral and neurobiological implications of sex-determining factors in Drosophila. Dev. Genet. 15 275–296. [PubMed]
  • Ten Bosch, J., 2006. The TAGteam DNA motif controls the timing of Drosophila pre-blastoderm transcription. Ph.D. Thesis, University of California, Berkeley, CA. [PubMed]
  • Tracey, W. D., Jr., X. Ning, M. Klingler, S. G. Kramer and J. P. Gergen, 2000. Quantitative analysis of gene function in the Drosophila embryo. Genetics 154 273–284. [PMC free article] [PubMed]
  • True, J. R., and S. B. Carroll, 2002. Gene co-option in physiological and morphological evolution. Annu. Rev. Cell Dev. Biol. 18 53–80. [PubMed]
  • Waterbury, J. A., J. I. Horabin, D. Bopp and P. Schedl, 2000. Sex determination in the Drosophila germline is dictated by the sexual identity of the surrounding soma. Genetics 155 1741–1756. [PMC free article] [PubMed]
  • Yeates, D. K., and B. M. Wiegmann, 1999. Congruence and controversy: toward a higher-level phylogeny of Diptera. Annu. Rev. Entomol. 44 397–428. [PubMed]

Articles from Genetics are provided here courtesy of Genetics Society of America
PubReader format: click here to try


Save items

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...