Sex Determination Signals Control ovo-B Transcription in Drosophila melanogaster Germ Cells
Justen Andrews, Brian Oliver


Nonautonomous inductive signals from the soma and autonomous signals due to a 2X karyotype determine the sex of Drosophila melanogaster germ cells. These two signals have partially overlapping influences on downstream sex determination genes. The upstream OVO-B transcription factor is required for the viability of 2X germ cells, regardless of sexual identity, and for female germline sexual identity. The influence of inductive and autonomous signals on ovo expression has been controversial. We show that ovo-B is strongly expressed in the 2X germ cells in either a male or a female soma. This indicates that a 2X karyotype controls ovo-B expression in the absence of inductive signals from the female soma. However, we also show that female inductive signals positively regulate ovo-B transcription in the 1X germ cells that do not require ovo-B function. Genetic analysis clearly indicates that inductive signals from the soma are not required for ovo-B function in 2X germ cells. Thus, while somatic inductive signals and chromosome karyotype have overlapping regulatory influences, a 2X karyotype is a critical germline autonomous determinant of ovo-B function in the germline.

SEXUAL reproduction in Drosophila requires the coordinated development of both the soma and germline (reviewed by Cline and Meyer 1996). The primary determinant of sexual identity in Drosophila, the X chromosome karyotype, was discovered nearly a century ago. Diploid flies with two X chromosomes (2X) are female, while flies with a single X chromosome (1X) are male. The Y chromosome is required for male fertility, but plays no known role in sex determination per se. Somatic sex determination is relatively well understood, but germline sex determination has been more enigmatic. Both the gametes and the gonads that they develop in are highly sexually dimorphic. Unsurprisingly, the communication between sex-specific gametes and somatic gonads is complex. Unfortunately, this complexity, especially the overlapping influences of cell-autonomous and noncell-autonomous signals, complicates our understanding of how germline sex is encoded in the genome. High fidelity female gametogenesis requires a 2X germline karyotype and an ovarian environment.

Experiments showing that the sex of the soma can override the inherent sexual karyotype of the germline provide strong evidence for the somatic contribution to germline sex determination (Steinmann-Zwickyet al. 1989). Karyotypically female germ cells (2X) develop poorly when transplanted into a male soma, but are able to form sperm at a low frequency, indicating that the male somatic environment can transform germ cells from female (egg) to male (sperm). Likewise, germline phenotypes observed in 2X flies genetically transformed from females into males in the soma indicate that a male soma can transform 2X germ cells from female to male (Seidel 1963; Marsh and Wieschaus 1978; Schüpbach 1982; Cline 1984; Nöthigeret al. 1989; Oliveret al. 1993). However, the sex of the germ cells does not strictly follow inductive somatic instruction. For example, some 2X germ cells in flies transformed from female to male develop recognizable egg chambers in addition to spermatocytes (Brown and King 1961; Seidel 1963; Nöthigeret al. 1989; Oliveret al. 1993). Another example of incomplete sex transformation is the finding that 2X flies transformed into males express both male- and female-specific products in the germline, instead of uniquely male-specific products (Oliveret al. 1993; Horabinet al. 1995). Thus, female germline sexual identity and oogenesis requires an ovarian environment.

Male germ cells (1X) are more resistant to transformation from male to female identity, suggesting that an ovarian environment alone is insufficient for induction of female germline sexual identity. 1X germ cells transplanted into females with no endogenous germline form ovarian tumors instead of eggs (Steinmann-Zwickyet al. 1989; Steinmann-Zwicky 1994). The ovarian tumor cells associated with defective sex determination resemble arrested primary spermatocytes (Dobzhansky and Bridges 1928; Dobzhansky 1931; Schüpbach 1985; Oliver et al. 1988, 1993; Steinmann-Zwickyet al. 1989; Pauliet al. 1993; Steinmann-Zwicky 1994). These data suggest that 1X germ cells attempt to develop according to the 1X male karyotype in an inappropriate ovarian environment. Again, this is not strict. Very limited female germline development and gene expression occur in 1X germ cells in a phenotypically female soma (Waterburyet al. 2000; Janzer and Steinmann-Zwicky 2001). While the precise sexual state of a germ cell is difficult to determine, it is quite clear that proper female germline sex determination requires both 2X germ cells and an ovary.

A network of genes act in germ cells downstream of the autonomous 2X signals, or inductive signals, or both (reviewed by Cline and Meyer 1996). The ovo locus occupies an upstream position in this germline sex-determination hierarchy and acts to regulate ovarian tumor (otu) and 2X germ cell viability. The ovo locus encodes a set of C2H2 zinc-finger proteins that bind to three sites in the otu promoter region (Lüet al. 1998; Lee and Garfinkel 2000; Lü and Oliver 2001) required for otu transcription (Lü and Oliver 2001). Molecular and genetic data suggest that ovo ultimately acts to regulate germline Sex-lethal (Sxl) expression via otu (Oliveret al. 1993; Pauliet al. 1993).

The ovo locus is complex, encoding mRNAs required for germline (ovo) and somatic (shavenbaby) functions during Drosophila development (Garfinkelet al. 1994; Mével-Ninioet al. 1995; Andrewset al. 2000). The locus is transcribed in somatic cells from ill-defined upstream promoters and is transcribed in the germline from either of two closely linked downstream promoters, ovo-A and ovo-B (Figure 1). Alternative promoter use in the germline gives rise to two major classes of OVO proteins with common C-terminal DNA-binding domains and different N-terminal domains (Figure 1B). The ovo-A transcript has an initiating AUG codon in exon 1A that is in frame with the long open reading frame in exon 2. The ovo-B transcript has an initiating AUG codon well into exon 2. The choice between germline promoters and the resulting protein isoform classes, OVO-A and OVO-B, is critical for female germline development. The shorter OVO-B isoform provides necessary and sufficient locus activity for female fertility (Andrewset al. 2000). Expression of the antagonistic OVO-A isoform results in dominant female sterility (Mével-Ninioet al. 1996; Andrews et al. 1998, 2000). This antagonism is evident at the transcriptional level (Figure 1, C and D). OVO-B activates the otu promoter, while OVO-A negatively regulates the otu promoter, as well as both germline ovo promoters (Lüet al. 1998; Andrewset al. 2000; Lü and Oliver 2001).

We know little about how ovo-B and ovo-A promoters are regulated. However, the effect of somatic signals and X chromosome karyotype on the expression of ovo reporter genes (jointly reporting the ovo-B and ovo-A promoters) has been previously reported (Figure 1, C and D; Oliveret al. 1994; Waterburyet al. 2000). One study concluded that the 2X karyotype controls overall ovo expression because 2X flies transformed to males expressed high levels of reporter activity, while 1X flies transformed into females showed feeble expression (Oliveret al. 1994). In contrast, a later study reexamined the same reporters and concluded that the somatic sexual identity controls overall ovo expression because 1X flies transformed into females expressed high levels of reporter activity (Waterburyet al. 2000). These reports are largely contradictory. A shortcoming of both studies is that neither distinguished between ovo-B and ovo-A promoter activity. For these reasons, it is unclear if a 2X germline karyotype (Figure 1C) or a female somatic sexual identity (Figure 1D) is most important for upregulation of the ovo-B promoter.

Figure 1.

Organization and function of the ovo locus. (A) Molecular genetic map of the ovo/shavenbaby locus. The top shows the genomic region surrounding 0, a SalI site in the locus. Just under the scale, the major promoters (Psvb, an ill-defined somatic promoter, and Pgl, the two tightly linked promoters active in germ cells), noncoding and coding exons (open and solid bars, respectively), introns (thin bent lines), and the region encoding the zinc finger DNA-binding domain are shown. Under the transcript map is a summary of the genetic data that separate the somatic and germline ovo functions. Deletion of Psvb by Deficiency (bracket) leaves wild-type ovo function in the germline. Transgenes include the indicated wild-type DNA rescue germline ovo phenotypes, but not shavenbaby phenotypes. (B) An expanded view of the germline promoter region. An AUG in the first exon of ovo-A mRNA appends a repression domain onto the OVO-A transcription factor, resulting in negative activity in the germline. The repression domain is not encoded due to the absence of an AUG in the first exon of ovo-B mRNA. (C) A model of ovo function where ovo responds to the number of X chromosomes in the germline. (D) A model of ovo function where ovo responds to a female somatic environment. (C and D) OVO binding sites are found at ovo and otu promoters. The OVO-A protein directly or indirectly downregulates the ovo and otu promoters, while the OVO-B protein directly upregulates the otu promoter.

Here we determine how the ovo-B promoter responds to 2X and somatic signals. We show that endogenous ovo-B mRNA is readily detectable in the female germline, but is quite difficult to detect in the male germline. We looked at endogenous ovo-B transcripts and ovo-B reporter expression in mutants that transform the sex of the germline or soma to determine which cues are critical for the differential expression of ovo-B. We find that 2X germ cells express high levels of ovo-B regardless of the sexual identity of either the surrounding soma or the germ cells themselves. By analysis of double mutants, we also show that ovo is upstream or independent from somatic sex determination signals, while otu, a direct OVO target gene, is downstream of somatic sex determination signals. These data indicate that ovo-B is functionally controlled by the germ cell sex chromosome karyotype. However, we also show that somatic sex determination signals do promote significant ovo-B expression in 1X germ cells. Thus, while we conclude that ovo-B is regulated primarily by a 2X karyotype and is required only in 2X germ cells, it is also regulated by a female somatic environment.


Flies and histochemistry: We used standard Drosophila techniques throughout. Flies were grown at 25° ± 0.5°. Most alleles and transgenes have been previously described and can be found, with references, at FlyBase ( Relevant ovo and sex determination alleles and FlyBase accessions are as follows: ovoD1rv22 (FBal0013399), ovoD1rv23 (FBal00-13400), ovoD2rvBT2 (a spontaneous revertant of ovoD2 obtained from the Daniel Pauli laboratory), ovoD1r+ (FBql0090210), Df(1)JC70 (FBab0000474), snf1 (FBal0015908), otu1 (FBal0013348), otu17 (FBal0013364), Sxlfs1 (FBal0016682), Sxlfs3 (FBal0034090), Sxl7BO (FBal0016694), tra1 (FBal0017004), tra-2B (FBal0017022), Df(2R)TRIX (FBab0002231), Df(3)dsxM+R15 (FBab0002755), dsxSwe (FBal0003203 or FBal0031142), trahs.PM (FBal0035817 or FBal0035819), ovo::lacZ1.1 (FBal0104821), ovo::lacZDap (FBal-0104823), and ovo::lacZDbp (FBal0104822). Genotypes are listed in the text and figure legends. Gonads were dissected and examined live, under phase contrast and Nomarski optics, or fixed and stained with X-Gal to detect LACZ expression (Andrewset al. 2000). Experimental samples and positive and negative controls were coprocessed in the same tubes.

Reverse transcriptase-PCR: Total RNA was extracted from tissues using TRIZOL (Bethesda Research Laboratories, Gaithersburg, MD). To increase the sensitivity of reverse transcriptase (RT)-PCR, [32P]dCTP was incorporated in the PCR step (Oliveret al. 1993). Touchdown PCR, which increases sensitivity by negatively ramping the annealing temperatures (Donet al. 1991), was also deployed in conjunction with radiolabeling. RT and PCR primers are listed according to 5′ position on the genomic map of Mével-Ninio et al. (1995), the direction (P, plus; M, minus), and length (in nucleotides). A total of 10–20 μg of total RNA was reverse transcribed with 16 ng/μl of ovo-specific primer 1846/M/24, 1 mm dNTPs, 8 mm DTT, 0.8 units/μl RNase inhibitor (Boehringer Mannheim, Indianapolis), 50 mm Tris 8.3, 75 mm KCl, 2 mm MgCl2, and 10 units/μl M-MLV reverse transcriptase (Bethesda Research Laboratories) in a total volume of 25 μl. Other RT primers tested were 1538/M/24 and 1541/M/17. Standard PCR conditions used 2.5 μl of RT reaction, 8 ng/μl of either primer 446/P/20 (ovo-A specific) or primer 858/P/20 (ovo-B specific), 8 ng/μl primer 1480/M/22, 2.5 μCi/33 nm [32P]dCTP, 1X MasterAmp J (Epicentre Technologies, Madison, WI), and 0.02 units/μl Taq polymerase (Bethesda Research Laboratories) in a total volume of 50 μl. The cycle profile was 5 min at 94°, 25 times (1 min each at 94°, 50°, and 72°), and 10 min at 72°. For touchdown PCR we used 2.5 μl of RT reaction, 8 ng/μl of either 446/P/20 (ovo-A specific) or 858/P/20 (ovo-B specific), 8 ng/μl 1480/M/22, 2.5 μCi/33 nm [32P]dCTP, 1× MasterAmp G (Epicentre Technologies), and 0.02 units/μl Taq polymerase (Bethesda Research Laboratories) in a total volume of 50 μl. The cycle profile was 5 min at 94°, 3 times (30 sec each at 94°, 64°, and 72°), 3 times (30 sec each at 94°, 62°, and 72°), 3 times (30 sec each at 94°, 60°, and 72°), 3 times (30 sec each at 94°, 58°, and 72°), 3 times (30 sec each at 94°, 56°, and 72°), 15 times (1 min each at 94°, 55°, and 72°), 5 times (1 min each at 94°, 50°, and 72°), and 10 min at 72°. Other PCR primers used were 455/P/17, 914/P/17, 1127/M/20, 1286/M/24, and 1457/M/17. Amplicons were isolated and sequenced using fluorescent dye terminators (ABI-PRISM, dRhodamine Terminator cycle sequencing, and an ABI-377 sequencer; Perkin-Elmer, Norwalk, CT). Following the initial verification of amplicon sequence, bands were identified by mobility against a known amplicon and by restriction digestion. We amplified internal controls in negative samples, indicating that the absence of product in those samples was not due to failed reactions.


ovo-B mRNA is expressed at high levels in the wild-type female germline and is barely detectable in the male germline: Multiple alleles of ovo result in female sterility, while none result in male sterility (Oliveret al. 1987). The female-specific requirement is for OVO-B (Andrewset al. 2000). However, ovo must be expressed in the male germline too, as otu promoter activity is greatly reduced in males hemizygous for amorphic ovo alleles (Hager and Cline 1997; Lüet al. 1998). The expression of ovo in the male germline is likely to be low, as previous work has failed to detect ovo mRNA in males by Northern blotting (Baeet al. 1994; Garfinkelet al. 1994). We were also unable to detect ovo mRNA in testis by RNAse protection assays (J. Andrews, unpublished results). Expression in the male germline has been detected indirectly, using ovo reporter genes (Oliveret al. 1994; Mével-Ninioet al. 1995). Nothing is known about ovo-A vs. ovo-B expression in the male germline. Low abundance in the male germline and problematic locus structure make this determination difficult. The ovo-A and ovo-B transcripts are nearly the same size, limiting the value of Northern blot analysis using common probes. Further, the short A/T-rich 1A and 1B exons are poor hybridization probes and thus cannot be used to distinguish the isoforms. Therefore, previously deployed methods were probably insufficiently sensitive to detect the limited quantity of ovo mRNA in male germ cells and determination of the isoform expressed is unknown.

Figure 2.

RT-PCR analysis of ovo transcription. (A) Diagram showing the location of the RT-PCR primers (arrows) used to detect the indicated transcripts from the ovo-A and ovo-B promoters. The RT primers are located farther downstream. (B) Radiolabeled touchdown RT-PCR results showing that at least some ovo transcript was present in gonads. Detection of ovo-B in ovaries was facile and robust, but detection of ovo-A in females and detection of either transcript in testis were difficult and sporadic. (C) Radiolabeled standard RT-PCR showing the clear detection of ovo-B transcript from the ovary and in testis only when the ovo copy number is increased to five. The ovo copy number was reduced using ovoD1rv22 and increased with P[w+ ovoD1+] transgenes. N, number scored.

Our interest in this study is the regulation of the ovo-B isoform in the female germline. However, an understanding of ovo expression in 2X female vs. 1X male germ cells is critical for understanding how inductive or autonomous signals regulate ovo. We examined the expression of ovo-A and ovo-B mRNAs in males and females using a sensitive RT-PCR assay, using [32P]dCTP in the reaction and intron spanning oligonucleotide primers (Figure 2A). While we did not experience difficulties amplifying ovo-B mRNA from ovaries by RT-PCR, amplification of ovo-A mRNA from ovaries was not robust. Unsurprisingly, amplification of either ovo-B or ovo-A from testis required extensive exploration of reaction conditions. The best detection of ovo isoforms in gonads was obtained using Touchdown RT-PCR (Figure 2B), an optimization method that uses gradually decreasing annealing temperatures during cycling (Donet al. 1991; materials and methods). We were ultimately able to detect ovo-B and ovo-A amplicons in both testis and ovary preparations. Neither class of ovo transcripts was detected in gonadless carcasses. These data are consistent with the germline specificity of the ovo-A and ovo-B promoters and provide reassuring molecular evidence for ovo expression in the testis, in support of ovo-dependent expression of otu::lacZ in the male germline (Lüet al. 1998).

While the touchdown RT-PCR data suggest that both forms of ovo mRNA are expressed in the gonads, this is not a useful assay for determining which primary sex determination signals are responsible for the high level expression in the female germline. Using standard PCR conditions, we detected ovo-B transcripts in RNA isolated from whole adult females and from ovaries, but not from any adult male tissues or female carcasses (Figure 2C). We failed to detect ovo-A transcripts in any tissue using standard PCR conditions (not shown). Multiple RT-PCR conditions and ovo-B primer pairs gave similar results (see materials and methods). ovo-B mRNA was readily detected using as little as 80 ng of total RNA from females, but not when up to 10 μg of testis or 20 μg of adult male RNA was used. In terms of wild-type gonads, standard RT-PCR provides an essentially plus/minus assay for the nearly female-specific expression of ovo-B.

The ovo locus is X-linked; therefore, wild-type females have two doses of ovo, while males have one. But differential expression of ovo-B is not trivially due to this inherent dose difference between males and females. If we failed to detect ovo-B expression in males because there are fewer copies, then we should be able to detect ovo-B expression in males with two copies of ovo+ (Figure 2C). However, we failed to detect ovo-B expression in testis samples from males with up to three copies of ovo+. We did detect ovo-B expression in samples from males with five copies of ovo+. However, this is not likely to be a reflection of a simple increase in ovo gene number and a linear increase in ovo-B transcript levels, as increased ovo gene dose results in autoregulation of ovo reporters in trans (Lüet al. 1998). Because we failed to detect ovo-B from testis bearing up to three copies of ovo+, any detection of ovo-B in 2X flies transformed from female into male is unlikely to be due to a simple ovo dose artifact.

Reporter genes provide a convenient second assay for the expression of ovo isoforms and are especially useful for determining which cells express the individual forms. Additionally, the reporter genes are autosomal, eliminating the complicating issue of assessing the influence of X chromosome dose on an X-linked gene. The expression patterns of ovo-A-specific and ovo-B-specific reporter genes reflect the higher levels of endogenous ovo transcripts in females vs. males (Figure 3; Andrewset al. 2000). The ovo::lacZΔap reporter shows high ovo-B activity in the female germline and little to no activity in the male germline (Figure 3B). In contrast, the ovo::lacZΔbp reporter shows feeble ovo-A activity in the female germline and little to no activity in the male germline (Figure 3C). Thus, the reporter genes faithfully replicate the expression of the endogenous ovo transcripts, providing another assay for determining which sex determination signals control ovo-B expression.

Figure 3.

The expression of ovo reporter genes in gonads. (A) Reporter of both promoters. (B) ovo-B reporter. (C) ovo-A reporter. Shown are schematics of the reporter constructs (left), genotype with respect to the reporter (left of the images), and X-gal-stained ovarioles (bundles of these functional units exist in the ovary) on the left and apical regions of the testes on the right (the apex contains the stem cells and young primary spermatocytes, the cell types expressing low levels of ovo mRNA).

ovo-B expression in 2X females transformed into males: The ovo-B isoform is required for female germline development, while the ovo-A isoform plays no known zygotic role in the female germline (Andrewset al. 2000). To determine if ovo-B expression in females is positively regulated by inductive female signals, or the number of X chromosomes, or is a simple consequence of female germline differentiation, we assayed for ovo-B expression in mutations that transform sexual identity.

If the somatic sexual identity is a critical determinant of ovo-B expression, then transformation of somatic sexual identity from female to male should result in a marked decrease in germline ovo-B expression. The somatic sex of 2X flies was transformed from female to male by the absence of transformer-2 (tra-2) or by the expression of only the male-specific form of doublesex (dsx) from the dominant allele, dsxswe. Use of the dominant allele is critical, as the absence of all dsx activity results in intersexual flies, not males. In our experiments, microscopic examination showed that ~10% of the sex-transformed 2X flies had testes containing mature primary spermatocytes, spermatids, or sperm. Thus, these 2X male individuals allowed us to test for the dependence of ovo-B expression on both female somatic and germline sexual differentiation. High level expression of ovo-B was noted in the 2X male germ cells, indicating that female sexual identity (somatic or germline) is not an obligate requirement for high level ovo-B expression (Figure 4). As a corollary, these data indicate that a 2X karyotype is important for ovo-B expression.

We do not know how the number of X chromosomes in the germline is assessed (Granadinoet al. 1993; Steinmann-Zwicky 1993; Hager and Cline 1997). In the soma, the effects of multiple X chromosome counting elements contribute to the initiation of female sexual development (Cline and Meyer 1996). As there are multiple X-linked genes in the nascent germline sex-determination hierarchy, we asked if they have additive or synergistic effects on the expression of ovo-B. In addition to the single mutants (Figure 5, C–E), we examined females homozygous for mutations in pairs of X-linked genes (otu1 Sxlfs1 or snf1 otu1) and females heterozygous for multiple X-linked germline sex determination genes [snf+ otu+ Sxl7BO/snf1 otu1 Sxl+, Df(1)JC70 (ovo snf-) otu+ Sxl+/ovo+ snf+ otu1 Sxlfs1, or Df(1)JC70 (ovo snf-) otu+/ovo+ snf1 otu1]. In no case did we observe an overt reduction in ovo-B reporter activity (data not shown). These data indicate that these X-linked loci do not provide overt 2X signals contributing to ovo-B expression.

Figure 4.

The expression of the ovo-B reporter in phenotypic males. (A) Wild-type testis. (B and C) Testes from females transformed into somatic males. Expression is highest in 2X germ cells. The genotypes with respect to the reporter (left) and with respect to sex determination genes (bottom) are shown. The X chromosome status is shown at the top.

Figure 5.

The expression of the ovo-B reporter in phenotypic females. (A) Wild-type female ovariole. (B) Ovary from a 1X fly transformed from a somatic male into a somatic female. (C–E) Ovarian tumors from females mutant for germline sex determination genes. X-gal staining is strongest in 2X gonads, but note that there is detectable expression in some of the tumorous chambers of 1X, trahs.PM females.

Detectable ovo::lacZΔap expression in 1X males transformed into females: Our experiments on 2X females transformed into males indicate that female somatic sexual identity is not required for high level expression of ovo-B. We next determined if ovo-B expression is positively regulated by a female soma in the absence of a 2X karyotype. 1X flies were somatically transformed from male to female using the gain-of-function trahs.PM allele. In the female somatic environment of these flies, the 1X germ cells resemble arrested primary spermatocytes, although some female germline differentiation may also occur (Steinmann-Zwickyet al. 1989; Oliveret al. 1994; Waterburyet al. 2000; Janzer and Steinmann-Zwicky 2001). We compared the expression of ovo-B in 1X females to that in 2X females with wild-type or tumorous ovaries (Figure 5) or to males (Figure 4). The comparison to 2X tumors is especially apt, because 2X and 1X ovarian tumors show similar phenotypes. The 1X germ cells within the ovaries of flies bearing the transgene did not express ovo-B at the levels seen in 2X flies (Figure 5), indicating that a female soma alone does not support high level ovo-B transcription. However, 1X males transformed into somatic females did express germline ovo-B at a level significantly higher than that in 1X males (Figure 4). The low to intermediate expression of ovo-B in 1X females indicates that a female soma has a positive influence on ovo-B expression.

Endogenous levels of ovo-B in sex-transformed flies: We used RT-PCR to directly corroborate the reporter data. To investigate the role of sex chromosome karyotype and somatic sex in regulating ovo-B expression, we performed standard RT-PCRs from sex-transformed flies. Because a female somatic sex had a weak effect on ovo-B reporter expression, we were especially interested in determining if endogenous ovo-B transcripts could be detected in 1X female flies.

We clearly detected ovo-B amplicons in 1X males transformed into females (Figure 6, lane 3), but not in wild-type 1X males (Figure 6, lane 4). These data indicate that a female soma does have a positive influence on the expression of ovo-B mRNA. We also confirmed that endogenous ovo-B transcripts are readily detected in 2X males (Figure 6, lanes 1 and 2), indicating that a 2X karyotype is sufficient for high level ovo-B expression in the absence of a female soma. Detectable expression of ovo-B in 2X flies transformed into males is remarkable, given how few germ cells are present in 2X flies transformed from female to male. Wild-type males with many more germ cells fail to express detectable ovo-B even when there are up to three copies of ovo+ (Figure 1C). Finally, ovo-B expression was readily detected in 2X flies mutant for otu, snf, or Sxl (Figure 6, lanes 6–8). Briefly, these RT-PCR data are fully consistent with the reporter results and strongly indicate that a germline 2X karyotype plays a prominent role in high level ovo-B expression, but that somatic signaling also positively regulates ovo-B expression.

Somatic signal input rests between ovo and otu: We were interested in determining whether the contribution of a female soma to ovo-B expression was important for ovo+ function in the female germline. The characteristic ovo loss-of-function phenotype is female germline death (Oliver et al. 1987, 1994; Staab and Steinmann-Zwicky 1996). Several studies have shown that ovo+ is not required for 1X germ cell viability in males transformed into females by gain-of-function tra alleles (Oliveret al. 1994; Nagoshiet al. 1995; Waterburyet al. 2000). Thus, any effect of the female soma environment on ovo+ function must be assayed in 2X flies. If somatic signals are obligatory for proper ovo+ function, then two related predictions are (1) that transformation of a female soma to a male identity should result in a germline phenotype like that seen in an ovo loss-of-function mutation and (2) that a double mutant for a somatic sex determination gene and ovo should show the same phenotype as flies mutant for only the somatic sex determination gene.

Figure 6.

RT-PCR on sex-transformed flies. Genotypes, karyotypes with respect to the X, and germline and somatic phenotypes of the flies harvested for the reactions are indicated. The location of the ovo-B product and an anonymous amplicon that serves as a useful positive control are also indicated.

Figure 7.

ovo function does not depend on somatic signals. Paired samples were from 2X brothers trans-heterozygous for tra-2B and the tra-2 Df(2)TRIX chromosomes. ovo flies were trans-heterozygous for ovoD1rv22, ovoD1rv23, or ovoD3rvBT2. otu flies were trans-heterozygous for otu1, otu14, or otu17. snf flies were homozygous for snf1. The ovo+, otu+, and snf+ siblings were heterozygous for the mutant alleles and FM7. N, number scored.

Mutations in both ovo and somatic sex transformations result in the loss of germ cells, although this loss is much less extreme in 2X females transformed into males (Nöthigeret al. 1989) than in 2X females without ovo function (Oliver et al. 1987, 1993, 1994). While this argues against a female soma being an obligate regulator of ovo, both phenotypes have been shown to be highly variable (Oliver et al. 1987, 1993, 1994; Nöthigeret al. 1989; Rodeschet al. 1995; Staab and Steinmann-Zwicky 1996). Variability, and the possibility of overlapping regulation, makes it difficult to rule out the proposition that a female soma functionally regulates ovo-B. Given the importance of this point, we extensively examined the phenotype of 2X flies transformed from female to male, in the presence or absence of ovo. For comparative purposes, we examined flies with the same sex-transforming genotype in the presence or absence of the downstream genes otu and snf. There is debate on whether some or all of the somatic sex determination signals pass through the tra and dsx loci, so we restricted our double mutant experiments to tra-2, which has been implicated consistently (Nöthigeret al. 1989; Oliveret al. 1993; Horabinet al. 1995; Waterburyet al. 2000).

Figure 8.

Model of ovo and otu regulation by primary karyotypic and somatic sex determination signals. Shaded somatic signals indicate that while there is molecular evidence for activation of ovo-B via somatic induction, there is no genetic evidence for biological relevance. It is not known if the two somatic signals are the same.

About 10% of control 2X females transformed into males due to the absence of tra-2 had a well-developed germline with advanced primary spermatocytes, spermatids, or sperm. The gonads of the remainder showed mostly debris. In contrast, 2X males that also lacked ovo did not show evidence of germ cells (Figure 7). This highly statistically significant absence of germ cells is due to the 2X karyotype and not due to any requirement for ovo+ in spermatogenesis, as 1X males hemizygous for the same alleles of ovo are fertile. These data unambiguously indicate that ovo+ genetic activity is not fully dependent on somatic sex determination signals transmitted via the tra-2 locus. In contrast to the results with ovo, the double mutant data clearly indicate that otu and snf genetic activities are under the control of somatic signals as has been previously suggested (Nöthigeret al. 1989; Oliveret al. 1993; Horabinet al. 1995; Nagoshiet al. 1995; Waterburyet al. 2000). The 2X flies transformed from somatic females to males and lacking either otu or snf showed the same frequency of germline development (including spermatogenesis) as 2X females mutant for only tra-2. Thus, there is clear somatic signal input downstream from ovo and upstream of otu (Figure 8).


Transcription patterns frequently correlate with tissue requirements. The ovo locus is required for female germline viability, sex determination, and differentiation, but has no known role in the male germline (Oliveret al. 1987). More specifically, ovo is required only in germ cells with a 2X karyotype, as 2X males require ovo+ for germline viability (Oliveret al. 1994; Nagoshiet al. 1995; this study), while 1X females do not (Oliveret al. 1994; Nagoshiet al. 1995; Waterburyet al. 2000). Certainly, the requirement for ovo in 2X germ cells might be expected to be reflected in higher ovo expression in 2X germ cells, but this does not have to be the case. For example, both males and females express some sex determination genes, but achieve sex-specific function by pairing with sex-specific cofactors (reviewed by Cline and Meyer 1996). Initial work suggested that ovo was female-specifically expressed (Baeet al. 1994; Garfinkelet al. 1994). However, later reporter and genetic studies clearly indicated that ovo is also expressed and can function in the male germline (Oliveret al. 1994; Mével-Ninioet al. 1995; Hager and Cline 1997; Lüet al. 1998). Our RT-PCR and reporter data indicate that ovo transcription is highly differential, but not female specific.

The existence of ovo isoforms with opposite activities (Andrewset al. 2000) raised the possibility that production of a particular isoform and not the overall expression of the locus could be sex specific (Andrewset al. 1998). We provide evidence that ovo-A and ovo-B isoforms are expressed in the female and male germline, albeit at markedly lower levels (failure to detect ovo-B expression in males by standard RT-PCR is quite stringent). A simple model where, for example, ovo-B acts in the female germline while ovo-A acts in the male germline has not emerged. For our purposes here, the striking expression differences between males and females as assayed by RT-PCR and reporter gene expression are excellent assays to tease out the regulatory inputs to the ovo-B promoter.

Two studies have examined the regulation of overall ovo transcription by primary sex determination signals (Oliveret al. 1994; Waterburyet al. 2000). Both relied on reporters (Oliveret al. 1994) that were constructed before the ovo-A and ovo-B promoters were mapped (Mével-Ninioet al. 1995). The authors reached opposite conclusions using the same reporters. The first study suggested that high overall ovo transcription is dependent on a 2X karyotype because of high expression in 2X flies transformed into males and low expression in 1X flies transformed into females (Oliveret al. 1994). The second study suggested that high overall ovo expression is controlled by the somatic signals because of high expression in 1X flies transformed into females (Waterburyet al. 2000). Oliver et al. (1994) discounted weaker ovo reporter expression in 1X females as background and Waterbury et al. (2000) did not compare ovo expression in 1X females to 2X male expression. Neither study looked at transcription directly. This study indicates that both a 2X karyotype and a female soma contribute to high ovo-B expression. Epistasis supports the idea that the 2X karyotypic signals are the most important for ovo function.


We thank Virginia Boulais for maintaining Drosophila stocks and The Bloomington Stock Center and Daniel Pauli for providing stocks. We also thank Jurrien Dean, Alan Kimmel, Jining Lü, and Michael Parisi for comments on the manuscript.


  • Communicating editor: R. S. Hawley

  • Received May 26, 2001.
  • Accepted November 9, 2001.


View Abstract