A systematic survey of embryonic expression patterns has shown that dsx transcripts are not maternally deposited and are specifically expressed in the somatic precursors of the gonads just prior to gonad coalescence 25. However, this global survey did not determine the sex of the embryos or specify the dsx mRNA isoform. We sorted male and female embryos bearing a female-specific Sxl early promoter 24 attached to eGFP and performed RT-PCR experiments using mRNA from sexed 3–10 hour, 10–16 hour and 16–22 hour old embryos (Fig. 1A–C) to determine if dsx^m mRNA is expressed in embryos. dsx^m mRNA was not readily detected in 3–10 hour old embryos of either sex, but was easily detected in 10–16 hour and 16–22 hour old male embryos. Sequencing of the RT-PCR products verified that they derived from predicted dsx^m transcripts. Thus, RT-PCR experiments showed that dsx^mis expressed solely in male embryos.

To further dissect the DSX^M expression pattern, we raised polyclonal antisera against a peptide from the male-specific C-terminus of DSX^M. To determine antibody specificity we performed immuno-labeling experiments focusing on whether: 1) the cell-staining pattern matched the in situ hybridization pattern, 2) the signal was male-specific and 3) the signal was absent from male embryos mutant for dsx.

The anti-VASA antibody was used to detect germ cells, which served as a guide to follow somatic cells of the gonad. In wild-type male embryos, cells intermingled with VASA-positive germ cells were clearly stained with anti-DSX^M during embryonic stage 13 (Fig 1F). Based on position, these were the somatic gonad precursor cells. The DSX^M staining was nuclear based on coincident DAPI staining for DNA (additional file 1). Additionally, the pattern of anti-DSX^M immunofluorescence coincided with the dsx^m transcription pattern detected by in situ hybridization previously 25, and in this study (not shown). Finally, we did not observe anti-DSX^M immunofluorescence in either female embryos (Fig 1D, E), or in embryos expressing a dsx mRNA truncated upstream of the region encoding the epitope used for antibody generation (genotype = Df(3R)dsx¹⁵/In(3R)dsx²³) (Fig 1H, I).

To better determine the identity of DSX^M-expressing cells, we performed co-immunofluorescence staining experiments with anti-DSX^M and antibodies against several somatic gonad precursor markers. We focused on mid- to late embryogenesis when DSX^M is strongly expressed and when gonad formation occurs (Fig 2).

We first ascertained the specific cell type/s that express DSX^M by using anti-VASA to detect germ cells and an antibody against the transcription factor Traffic jam (TJ) to detect the somatic gonad precursors 26. The double labeling with anti-VASA and anti-DSX^M indicated that DSX^M was not expressed in the germline. Anti-TJ and anti-DSX^M staining indicated that DSX^M was expressed in somatic cells intermingled with the germline in stage 13 embryos (Fig. 3A–H), stage 15 embryos (Fig. 3I–P), as well as in later embryonic stages.

The overlap in TJ and DSX^M staining was consistent with DSX^M expression in all somatic gonad precursors in the embryonic testis. However, at the very posterior of the coalescing gonad we detected DSX^M expression in cells that did not show TJ staining (Fig. 3E, H and 3M, P arrows), but were clearly assembled into the embryonic testis. TJ is reported to be expressed in all somatic gonad cell precursors 26, but our data suggest that TJ is not expressed in the posterior somatic precursors. Based on position and the confirming experiments outlined below, these DSX^M-positive and TJ-negative cells were the male-specific somatic gonad cells. This indicates that TJ is not expressed in all somatic gonad cells.

To further investigate DSX^M expression in the somatic gonad we used anti-Eyes absent (EYA). EYA is a transcription factor, which is also expressed in the somatic gonad 27. EYA was found in the entire somatic gonad including the cluster of posterior TJ-negative somatic gonad cells. Staining for EYA and DSX^M revealed co-expression at the cellular level throughout the entire somatic gonad of stage 13 and older embryos (Fig. 4A–D). These data suggest that a small population of somatic gonad cells expresses DSX^M and EYA, but not TJ.

To ascertain the identity of the posterior most cells of the somatic gonad more directly, we applied anti-DSX^M in combination with anti-SOX100B, because the male-specific somatic gonad precursor cells are reported to be the only gonad cells expressing SOX100B 28. Anti-DSX^M and anti-SOX100B co-immunofluorescence revealed DSX^M expression in male-specific somatic gonad precursors (Fig. 4E–L). Absence of TJ in a subset of SOX100B-expressing cells was confirmed by counter-staining with anti-SOX100B (Fig. 4M–P). There were clearly cells expressing SOX100B, but not TJ. A few cells appeared to express both SOX100B and TJ. Taken together, these data indicate that DSX^M is expressed in all somatic gonad cells expressing EYA and either TJ or SOX100B.

Interestingly, SOX100B was also expressed in another population of cells that wrapped around the gonad in stage 17 embryos (Fig. 5A–D). These might have been the precursors of the testis sheath 17, but we did not investigate the fate or function of these cells in this study. We did not observe DSX^M expression in these SOX100B-positive cells surrounding the embryonic testis (Fig. 5A–D).

The testis hub is part of the niche and required to maintain male germline stem cell identity. The somatic hub cells of the male gonad can be identified in stage 17 embryonic testes as an anterior cluster of somatic cells expressing the cell adhesion molecule Fasciclin III (FAS III) 29. In anti-DSX^M and anti-FAS III co-immunofluorescence experiments, we observed strong DSX^M expression in hub cells that were outlined by anti-FAS III labeling (Fig. 5E–H). Taken together, our data indicate that DSX^M is expressed in all the somatic cells of the embryonic testis identified with somatic gonad markers, except the SOX100B-positive cells that surround the embryonic testis at stage 17.

In contrast to many other cell types, the germline stem cells continue to divide in adults in order to produce gametes. If germline sexual identity is irreversibly determined during embryogenesis, there may be no need for post-embryonic somatic DSX^M expression. However, if sex determination or maintenance of sexual identity is an ongoing process, then germline sexual identity might require continued expression of DSX. We readily detected dsx^m transcripts by RT-PCR with RNA from adult testes (not shown). To explore the cellular expression pattern of DSX^M in testes of larvae and adult flies, we performed whole-mount antibody immunofluorescence staining experiments.

Male germline stem cells in close contact with the hub undergo asymmetric divisions to regenerate the stem cell population and produce the cells that develop into sperm 30. Cells remaining at the hub are the stem cells. We observed DSX^M expression in the somatic hub cells of larval (Fig. 6A–D) and adult testes (Fig. 6E–H), where it also co-localized with TJ. These data suggest that male germline stem cells are always in contact with DSX^M-expressing cells.

Germline cells leaving the niche divide four times to produce 16-cell cysts. Germline cysts are enveloped by two somatic cyst cells, which become flattened as the germline cysts enlarge 20-fold in volume in preparation for meiosis. These engorged cells are highly transcriptionally active. Sperm differentiation occurs postmeiotically under translational control 30. The transcription factor TJ is expressed in the cyst progenitor cells at the apex and in early cyst cells enveloping the dividing germ cells and then fades in later cysts 26. EYA expression becomes stronger during this progression, such that the two patterns are partially complementary 31. Co-immunofluorescence with anti-DSX^M, anti-TJ and anti-EYA revealed that there was overlapping expression of the three proteins in these somatic cyst cells (Fig. 7). Anti-DSX^M staining of larval and adult testes revealed expression in early cyst cells as evidenced by position within the gonad and the overlap with TJ. When TJ expression faded, DSX^M expression persisted in late cyst cells and co-localized with EYA (Fig. 7A–D and 7E–H). Interestingly, anti-DSX^M staining was never observed in cyst cells surrounding transcriptionally quiescent post-meiotic stages (not shown). These data are consistent with the idea that the determination or maintenance of germ cell sexual identity requires ongoing DSX^M expression until the end of spermatogenic transcription at meiosis, but is no required once germline transcription ceases.

Testis formation depends on the collaboration of germ cells and distinct somatic cell types. However, somatic gonad formation is independent of germ cells 1732. Therefore, we predicted that functional DSX^M expression should not be dependent on a germline. To investigate whether germ cells were necessary for DSX^M expression in somatic gonadal precursor cells, we examined embryos lacking germ cells due to the maternal grandchildless mutation gs(l)N26 33. Unsurprisingly, anti-DSX^M staining was readily detectable in germlineless embryonic testes (Fig. 8A–D) indicating that the germline is not required to induce or maintain DSX^M expression.

The eya gene was a candidate regulator of dsx expression as EYA was always expressed in DSX^M-positive gonad cells, and preceded DSX^M expression temporally in the embryo (EYA was strongly expressed in somatic gonadal precursor cells of stage 12 embryos, when we did not readily detect DSX^M expression). We therefore determined whether DSX^M expression required EYA function. This experiment was complicated by the fact that EYA is essential for gonadogenesis. In late embryonic stages, eya mutant embryos have germ cells that are scattered throughout posterior regions and only a few somatic gonad precursor cells develop. The few somatic gonad precursors do not coalesce into a gonad, rather they form clumps with associated germ cells 27. In homozygous eya mutant male embryos, we were able to detect DSX^M expression in the sporadically formed somatic gonad precursor cells (Fig. 8E–H). These sporadically formed cells also expressed TJ. Expression of DSX^M and TJ in these rarely formed somatic gonad precursor cells suggests that eya is not obligatory for expression of TJ or DSX^M in the embryonic testis. However, it should be noted that there are very few TJ- and DSX^M-positive cells in eya mutants. Those somatic gonad cells that escape may not be representative.

TJ was expressed in many DSX^M-positive somatic cells and was therefore a potential regulator of dsx. We examined tj^eo2 male embryos to determine if DSX^M expression depends upon TJ. The tj^eo2 allele 26 encodes a truncated protein lacking the two DNA-binding domains, a putative bipartite nuclear localization signal, and the leucine-zipper domain. The allele behaves as a genetic null. In gonads of tj mutant embryos, germline divisions during late embryogenesis are hindered. Additionally, somatic gonad cells do not intermingle with germ cells and remain at the gonad periphery 26. In homozygous tj mutant male embryos, we detected DSX^M expression in all the somatic gonad cells (Fig. 8I–L). Thus, DSX^M expression in somatic gonad cells is not dependent upon the activity of TJ.

The pre-mRNA splicing cascade regulating somatic sexual development has been well studied (Fig 9A). The X-chromosome number is read by the uniform expression of X chromosome transcription factors, which uniformly and transiently activate Sxl expression in XX:AA flies prior to general activation of the zygotic genome 3. This transient expression of the SXL splicing factor results in an autoregulatory loop when Sxl pre-mRNA is produced from a ubiquitous promoter active later in development. In X:AA flies, SXL is not present and dosage compensation complexes form in all X:AA cells to increase X chromosome gene expression 34. Addressing chromosome dose imbalance should be important for all cells. The uniform and early expression of SXL to prevent over-expression of X chromosome genes, via inhibition of male-specific-lethal-2 (MSL2) in females, is therefore quite logical. The absence of SXL also ultimately results in DSX^M production. As many aspects of sexual dimorphism in the soma require DSX function 81435363738, there is no a priori requirement for tissue-specific expression of dsx.

Here we show that DSX^M expression in male embryos is restricted to somatic gonadal cells that form several hours after SXL expression is initiated. Consequently, in addition to the sex determination hierarchy, it is likely that DSX^M expression depends on additional and yet unknown activators expressed in embryonic somatic gonadal precursor cells or repressors in non-gonadal somatic cells. It is also likely that DSX^F is expressed in the corresponding somatic gonad in females, but we have not been able to generate useful antibodies against the predicted female isoform, or the region common to DSX^M and DSX^F. Expression of dsx in subsets of the nervous system 39 raises the possibility that transcriptional deployment of dsx pre-mRNAs in female or male pre-mRNA splicing environments is a common theme.

Two main cell types compose the embryonic Drosophila gonad: germ cells and somatic gonad cells 19. Our results suggest that each somatic gonadal cell of a male stage 13 embryo has a male identity before overt morphological testis development is evident. By stage 17, embryonic gonads are clearly sexually dimorphic (Fig 9B). The hub, a cluster of specialized somatic cells required for germline stem cell maintenance, forms anteriorly in stage 17 testes 29; these hub cells express DSX^M. We have also seen a monolayer of cells around the testis during stage 17, possibly the testis sheath precursors. These late additions to the embryonic testis do not express DSX^M but do express SOX100B.

Although DSX^M is required to promote male development and to repress female development in several somatic tissues, the role of DSX^M activity in embryonic testis morphogenesis is unclear as dsx mutants are strikingly similar to wildtype males 21. However, it is clear that DSX^M regulates the expression of the STAT92E transcription factor in male embryonic germ cells 22. Additionally, DSX^M results in male-specific splicing of Sxl pre-mRNA in adult germ cells 15. Perhaps the major role of somatic DSX^M in the embryo is regulation of the germline gene expression program. The non-autonomous role of DSX^M is not restricted to germline development, as DSX^M also non-autonomously regulates the post-embryonic recruitment of mesodermal cells to the male genital disc 40.

DSX^M-expressing cells are in intimate contact with the germline, which may be important to enable the somatic sex determination pathway to influence germ cell development in the embryonic gonad. Somatic gonadal precursors undergo striking shape changes as the gonad coalesces, producing thin cellular extensions that surround the round germ cells at stage 13 41. DSX^M expression becomes apparent at this stage.

There is also contact between germline and DSX^M-positive somatic cells in post-embryonic stages. We found that DSX^M is expressed in testes of larvae and adult flies in the somatic cyst cells enveloping all premeiotic germ cell cysts as well as in the hub. DSX^M expression was never observed in cyst cells enclosing postmeiotic germ cells, which have completed the transcriptional program required for sperm differentiation 4243. Thus, DSX^M appears to be expressed in all the somatic cells that are closely associated with the germ cells from gonad formation until the transcriptional program of spermatogenesis is complete.

All flies were raised and maintained on standard cornmeal media (Tucson Drosophila stock center, Tucson, AZ) at 25°C. A fly strain bearing a construct containing the early Sex lethal promoter, SxlP_E, 24 upstream of the coding sequence for eGFP on chromosome 3 (Sxl-GFP-3) was used for sexing embryos 44. Only female embryos of this line express eGFP. Dechorionated embryos were sex sorted by distinguishing fluorescent versus non-fluorescent embryos using the Copas Select sorter (UNION Biometrica, Holliston, MA). 60 embryos from each collection were set aside and allowed to mature to score the sexing efficiency. Only collections showing 100% sexing fidelity were used in experiments. The following mutant alleles were used for analyses: gs(l)N26 45; Df(3R)dsx¹⁵ and In(3R)dsx²³ 46; eya^cli-IIEand eya^cli-DI 47 and tj^eo2 26.

Total RNA was prepared using TRIzol reagent (Invitrogen, Carlsbad, CA) from 3–10, 10–16, 16–22 hour old, sex sorted embryos as well as from testes of male adult flies. Preparation of polyA⁺-mRNA from all samples was done using the Oligotex mRNA Mini Kit (Qiagen, Valencia, CA). RT-PCR was performed using the OneStep RT-PCR Kit (Qiagen, Valencia, CA). A pair of primers (RT-sense 5'-CGCGCACCACGTCCACATGGCAGCTG-3'; male-antisense 5'-CTCTGGAGTCGGTGGACAAATCTGTGTG-3') flanking two introns was used to amplify an 801 bp cDNA fragment from the dsx^m transcript. The amplicons were sequenced for verification. For loading control a separate RT-PCR of each mRNA sample was carried out in parallel using a primer pair (β3 sense 5'-ATCATTTCCGAGGAGCACGGC-3'; β3 antisense 5'-GCCCAGCGAGTGCGTCAATTG-3') for the ubiquitously expressed gene β3-tubulin 48, which amplifies a 397 bp fragment from the transcript.

The polyclonal DSX^M antibody was made using standard methods (Covance, Princeton, NJ). A peptide containing the C-terminal amino acids SSNGAYHHGHHL was used for immunization of two rats, both of which gave useful antibodies of good titer and were used here.

For cell staining experiments, pre-stage 17 embryos were fixed, devitellinized, and immunostained as described 49. Because the presence of the cuticle greatly hinders antibody staining, we used a different protocol for stage 17. Fixed stage 17 embryos were rehydrated, washed twice in 1 × PBS + 0.1% Tween 20 (PBT), sonicated in 500 μl PBT (2 pulses of 3 sec. using a Misonix Sonicator 3000 at output setting 1), rewashed twice in PBT and immunostained 49. Gonads of larvae and adult flies were dissected, fixed and immunostained as described 50. All stained samples were mounted on slides in Fluoromount-G (Southern Biotech, Birmingham, AL).

Antibodies were as follows: mouse anti-FASIII and mouse anti-EYA10H6 (Developmental Studies Hybridoma Bank, Iowa City, IA) at 1:50, rabbit anti-SOX100B 28 at 1:1000, guinea pig anti-TJ 26 at 1:3000, rabbit anti-VASA (R. Lehmann) at 1:5000, chicken anti-VASA 51 at 1:1000. For the rat anti-DSX^M at 1:500 in general and the rabbit anti-SOX100B at 1:1000 (in Fig 4N) the TSA (Tyramide Signal Amplification) Cyanine 3 technology (Perkin Elmer, Waltham, MA) in combination with the ABC Kit (Vector Laboratories, Burlingham, CA) was used as the detection system.

The following secondary antibodies were used, all at 1:500: Cy5 goat anti-rabbit, biotin-coupled goat anti-rat, and biotin-coupled goat anti-rabbit (Jackson Immuo-research, West Grove, PA), Alexa 647 goat anti-guinea pig, Alexa 488 goat anti-guinea pig, Alexa 488 goat anti-rabbit, Alexa 647 goat anti-mouse, Alexa 488 goat anti-mouse, Alexa 647 goat anti-chicken (Invitrogen Molecular Probes, Carlsbad, CA). Images were acquired using a Zeiss LSM 510 Meta Confocal Microscope (Zeiss, Thornwood, NY) and processed using Photoshop 7.0 (Adobe, San Jose, CA). All figures were annotated using Illustrator 10.0 (Adobe, San Jose, CA). All images in figures are single optical sections as indicated.

DAPI (4',6-diamidino-2-phenylindole, dihydrochloride) staining was used to visualize the nuclei.