The RGS gene loco is essential for male reproductive system differentiation in Drosophila melanogaster

Background The loco gene encodes several different isoforms of a regulator of G-protein signalling. These different isoforms of LOCO are part of a pathway enabling cells to respond to external signals. LOCO is known to be required at various developmental stages including neuroblast division, glial cell formation and oogenesis. Less is known about LOCO and its involvement in male development therefore to gain further insight into the role of LOCO in development we carried out a genetic screen and analysed males with reduced fertility. Results We identified a number of lethal loco mutants and four semi-lethal lines, which generate males with reduced fertility. We have identified a fifth loco transcript and show that it is differentially expressed in developing pupae. We have characterised the expression pattern of all loco transcripts during pupal development in the adult testes, both in wild type and loco mutant strains. In addition we also show that there are various G-protein α subunits expressed in the testis all of which may be potential binding partners of LOCO. Conclusion We propose that the male sterility in the new loco mutants result from a failure of accurate morphogenesis of the adult reproductive system during metamorphosis, we propose that this is due to a loss of expression of loco c3. Thus, we conclude that specific isoforms of loco are required for the differentiation of the male gonad and genital disc.

nal signal the GPCRs are associated with an inactive heterotrimer complex, Gα-GDP/Gβ/Gγ. When a specific ligand binds a GPCR, the intrinsic nucleotide exchange factor (GEF) activity is activated; the resultant Gα-GTP subunit dissociates from Gβ/Gγ, leaving the Gβ/Gγ heterodimer and the Gα-GTP to spread the signal to downstream target molecules. The GTP is slowly hydrolysed by Gα, and the Gα-GDP then returns and binds to the Gβ/Gγ complex rendering the receptor inactive. RGS proteins (Regulator of G-protein signalling) are a family of GTPase activating proteins (GAP) that trigger the intrinsic GTPase activity of the Gα subunits [1,2]. Although a great deal is known about the regulation of G-protein-coupled receptor signalling in a variety of organisms [3,4] less is known in Drosophila and more importantly the involvement of Gprotein-coupled receptor signalling in developmental decisions.
In Drosophila nine genes encoding for RGS proteins have been identified [5], however protein function has only been studied in three of them, axin, gprk2 and loco. Daxin, the Drosophila orthologue of axin [6], is a scaffold protein, that in the absence of Wnt signaling, negatively regulates cytosolic Armadillo by aiding its proteosome-dependent degradation [7][8][9]. The negative regulation of Armidillo by axin is inhibited by the interaction of Axin with the Gαs subunit of Prostaglandin E2-stimulated in colon cancer cells [10]. Gprk2, G-protein-coupled receptor kinase 2, maintains cAMP levels in the ovary and is required for embryonic anterior patterning [11,12]. The Drosophila loco gene encodes a number of LOCO protein isoforms, all of which contain the RGS domain and also a GoLoco motif that acts as a Guanine nucleotide dissociation inhibitor (GDI), ( Figure 1A). Previous studies suggest that LOCO might play an important role during early Drosophila development. For example during asymmetric cell division of the Drosophila neuroblast, LOCO may act as a GTPase activating protein of the Gαi protein via its RGS domain as well as a (GDI) through its GoLoco motif [13]. In the Drosophila embryo loco is essential for the formation, extension, and migration of glial cells, and plays a role in asymmetric cell division of Drosophila neuroblasts [13,14]. Rare adult flies lacking the loco c1 and loco c2 transcripts have locomotive defects [14]. Previously, we showed that loco is expressed in the nurse cells and in specific subsets of the follicle cells of the Drosophila egg chamber, and that it is required for cytoplasmic dumping during oogenesis and for dorsal-ventral axis formation of the egg chamber and embryo [15].
The Drosophila loco gene encodes a number of isoforms of an RGS protein ( Figure 1A). loco c1 and loco c2 were the first transcripts to be identified and were found to be differentially expressed during embryogenesis [14], subsequently we identified a third transcript, loco c3, and showed that it was required for egg and embryo development [15]. The sequence data on loco c3 has been extended to show that more sequence, including a start site, was upstream of the original start site identified for loco c3. The two start sites are in frame, but neither has been shown to be functionally active [13]. To remain in line with the published nomenclature, we will call this extended transcript loco c4.
Until now nothing was known about the involvement of RGS in G protein signalling in the male reproductive system, we therefore set out to determine if LOCO was also essential for male development. We isolated male fertility mutants from a P element mobilisation screen [15] and have shown that the male sterility is due to mutations mapping in the loco gene. Gene expression analysis has not only identified disrupted gene expression in the loco mutant lines but it has also revealed a fourth loco transcript required for correct male development. This alongside the phenotypic analysis of the semi-sterile males suggests a role for loco in the differentiation of the testis from the male gonads and genital discs. Furthermore we analyse loco expression in male gonads and the male adult reproductive tissue in both wild-type and loco mutant lines.
Finally, RGS proteins, such as LOCO, negatively regulate signalling mediated by G-protein coupled receptors, by reducing the time that the Gβ/Gγ subunit is available to signal. However with an additional GoLoco motif, LOCO can also increase the initiation rate of G protein signalling [13]. LOCO may well regulate this signalling pathway in the follicle cells of the Drosophila egg and glial cells of the embryo by binding to the Drosophila Gαi subunit [14,16,17]. In order to analyse the presence of Gα-proteins, which could potentially interact with LOCO in the Drosophila testis, we undertook a candidate PCR approach and identified a further two Gα subunits expressed in the testis.

Screens for male sterility
Previously we carried out a P element mediated mutagenesis screen using a P element located between exons II-1 and I-1 of the loco gene ( Figure 1A) [15]. We established that perfect excision of this element led to fully viable fertile lines, indicating that there were no other mutations in the stock. Most of the 399 lines that we generated were homozygous lethal and many of the viable lines produced very few homozygous adults, indicating a requirement for loco during development. Many of the lines, which generated some adults, also showed reduced fertility in females. Complementation analysis of the 399 lines showed that the mutants fell into two different complementation groups, however, two mutant lines fell into both comple- The loco gene and its effect on male fertility Figure 1 The loco gene and its effect on male fertility. A: The loco gene is comprised of nine exons. The final three exons, exon 2, 3 and 4 are common to all loco transcripts whereas the 5'exons are alternatively spliced to give rise to five different loco transcripts; loco c1 (AF130745), loco c2 (AF130744), loco c3, (AF245455), loco c4 (CG5248 PC) and loco c5 (AI944917). The exon numbers are shown below the genomic sequence and the conserved domains of the LOCO protein and the primers used to identify the transcripts in the testis and pupae are marked above the appropriate exons. Black regions indicate coding sequence, white regions represent untranslated regions. The P element in the original P insertion line, c139, is 322 bp upstream of exon 2. The loco 318 mutant contains two P elements (or one P element and a partial P element) in reverse orientation 322 bp upstream of exon 2. The P elements of loco 318 are flanked by a 9 bp duplication of genomic sequence. B: The sequence surrounding the P element insertion site in c139 was sequenced in loco 358 and loco 387 and aligned to the genomic DNA using Clus-talW [36] and BoxShade [37]. No deletion was observed. The region of genomic DNA that is duplicated in loco 318 is highlighted in red. C: Hemizygous flies, containing the loco mutation and the deficiency chromosome (Df(3R)15CE1 (Df15), Df(3R)17D1 (Df17), or loco Δ13 ) or the wild-type, OrR, chromosome, were crossed with OrR, Df15, Df17 or loco Δ13 virgin females. The ratio between the total number of eggs laid and the number of eggs which hatched is represented as a percentage. Chi squared values were calculated by comparing the heterozygous male semi-steriles to OrR and the hemizygous mutants to the heterozygous deficiency lines. Chi squared values over 6.64 suggest that the reduction in fertility is due to the two lines being closely linked. Asterisks indicate a statistically significant reduction. mentation groups. The complex splicing of loco transcripts makes it likely that both of the complementation groups affect different essential transcripts of the loco gene.
In this study we wanted to determine if any of the semilethal lines which generated some adult progeny were male sterile. We found that 27 of the loco mutant lines generated a few adult males. The fertility of the rare adult males was investigated by crossing the homozygous loco mutant males to OrR virgin-females. Five lines produced none or very rare progeny; loco 318 , loco 358 , loco 455 , loco 370 , and loco 387 . The first three of these are red-eyed lines, which have either resulted from a partial excision or from a mobilised P element. The latter two lines are white eyed and therefore lack at least part of the original P insertion and possibly some flanking genomic DNA. One of these lines (loco 455 ) subsequently ceased producing any adult males and was not further investigated. Complementation analysis showed that the mutations of the semi-sterile males fell into the same complementation group ( Table 1). The rare heteroallelic males produced in these experiments generated very few progeny when crossed to OrR females (this includes line loco 455 ).
None of the loco mutants heterozygous with OrR showed significantly reduced male fertility, indicating that the semi-sterile loco mutant males were fully recessive ( Table  2). To further characterise the loco mutants we analysed male fertility in two other strains lacking loco. The two deficiency strains used were Df(3R)15CE1 (Df15), which lacks the cytogenetic region 93F-;94C-94D, and Df(3R)17D1 (Df17), which lacks 93E-94C, both lack the loco gene which is at position 94B6-94B8. The deficiency lines were crossed to OrR and the resulting heterozygous males were crossed to OrR to assess the effect of the mutation on fertility (Table 2). Only Df15/+ showed a significant reduction in fertility when compared to OrR, this is likely to be due to the loss of genes present in the region 94C-D as Df17/+, when crossed to OrR, did not show any significant reduction in male fertility ( Table 2). Further-more loco Δ113 , a mutant strain, which was previously reported to lack part of the loco gene and 7 kb of downstream sequence [14] also showed no significant reduction in fertility when crossed to OrR (Table 2). This suggests that heterozygous loss of loco does not affect male fertility.
In order to assess whether our loco mutants were in the same or different complementation groups to the deficiency lines [14], the four loco mutant lines were crossed  to each of the above mentioned deficiency lines. All heteroallelic mutants had reduced fertility when compared to the heterozygous mutants ( Figure 1C). Despite all of the heteroallelic loco mutants showing reduced fertility not all were a statistically significant reduction (Table 2). However all of the mutant loco alleles (loco 318 , loco 358 , loco 370 , and loco 387 ) hemizygous with the loco Δ113 allele, showed a statistically significant reduction in male fertility (Table 2 and Figure 1C). This not only suggests that the mutations isolated here fall into the same complementation group but, that the mutations we have isolated reside within the loco gene.

Expression of loco in the testis
The semi-sterile males obtained from the loco mutants isolated in this screen suggested that LOCO may have a role within the testes. To investigate whether loco was expressed in the male gonads a UAS-lacZ reporter strain was crossed to the original P insertion strain (c139), which contained a GAL4 insertion in the loco gene [15]. The resulting male adult gonads were stained for β-galactosidase activity. The testis and seminal vesicles showed very strong lacZ expression (Figure 2A). The loco gene is alternatively spliced to give rise to several isoforms loco c1, loco c2, loco c3, and loco c4 [13][14][15]. Upon database searching we revealed that there was a fifth transcript, isolated from an adult EST testis library (Accession Number Expression of loco in the male gonads  Figure 1A were used to establish that loco c1, loco c2, loco c3 and loco c5 were expressed in OrR testes. The band marked with an asterisk is the true PCR product of the 3' end of loco c2 (3'c2) as determined by sequence analysis. D: Several G-protein transcripts were detected in OrR testis. The asterisk indicates the true PCR product of expressed Gα73B as determined by sequence analysis. R: RNA, G: genomic DNA.
AI944917). The sequence had no conserved domains relating to LOCO and mapped to the genomic region located in the intron between exon II-1 and exon I-1 (Figure 1A). This suggested that there was an additional transcript, not previously described, but expressed in the testis. This new exon has been labelled exon IV-1 and the transcript produced is loco c5 ( Figure 1A).
In order to analyse the expression of the various loco splice variants in the Drosophila testis primers were designed to specifically amplify loco c1, loco c2, loco c3 and loco c5. It should be noted that the primers used to amplify loco c3 could not discriminate between loco c3 and loco c4, furthermore it is possible that these two transcripts form the same transcriptional unit. RT-PCR revealed that loco c1, loco c2, loco c3, and loco c5 were expressed in the OrR male testis ( Figure 2C). Detection of loco c2 using the primer pair FP1 and RP8 ( Figure 1A) produced a specific product at approximately 1 kb and two non-specific bands at lower molecular weights ( Figure 2C, 3'c2). Cloning and sequencing revealed that the 1 kb band was specific to loco c2, the 0.9 kb band aligned to CaBP1 (CG5809) and the 0.4 kb band aligned to myosin binding subunit (CG32156). We also show later in the paper that the novel transcript identified from the testis EST library is not unique to the testis, as it is also expressed in pupae. Thus what we have identified and described is a further novel loco transcript which is expressed during development.

G-protein α subunits are also expressed in the adult
reproductive system LOCO is a regulator of G-protein signalling and has been shown to interact with various Gα subunits [14,17]. Searching the Drosophila genome and various EST databases we found several Gα transcripts and proteins. We wanted to investigate if different Gα subunits were expressed in the testis providing potential binding partners for the isoforms of LOCO that are generated. The three Gα genes we choose to analyse were G-oα47A, Gα73B, and Gα49B.
G-oα47A the Drosophila homologue of the mammalian Goα, is needed for embryonic development [18][19][20] and has more recently been shown to contribute to asymmetric cell division [21]. Furthermore it is expressed in the nurse cells and oocyte and is present in various adult nerve cells [22]. Gα73B encodes a further Gα subunit called Gfα, it is expressed in the embryonic midgut and in the nurse cells after which it is transported to the oocyte [23]. Gα49B, a Gq subunit involved in phospholipase C activation, is involved in the Drosophila visual system [24,25]. Gα49B is known to be expressed in the adult testis [26], and thus acted as a positive control. PCR showed that G-oα47A, Gα73B and Gα49B were expressed in the testis ( Figure 2D). This raises the possibility that in the adult testis there are additional Gα subunits that may interact with LOCO.

Analysis of mutant phenotypes
We have shown that loco and various Gα subunits are expressed in the testis and have identified, genetically, that mutation in loco can lead to significantly reduced fertility in males. We therefore went on and analysed adult gonad morphology in the loco mutants. All of the loco mutants had abnormal reproductive systems and showed a variety of defects ( Figure 3B-E). The phenotype was almost 100% penetrant, there was a range in severity within the same line, and only the very occasional male reproductive system was normal. Often the accessory glands were abnormal and the ejaculatory ducts appeared swollen. Neither testes from virgin wild-type males or aged wild-type males show this peculiar phenotype, suggesting that this phenotype was not attributable to the mating status (data not shown). There were no obvious differences between the mutant lines, which may be attributable to all mutants being in the same complementation group. The testes, even those very abnormal in shape and size, contained mobile sperm ( Figure 3F). This suggested that the differentiation of the sperm does not require the loco isoform affected in this group of mutants, but possibly differentiation and morphogenesis of the derivatives of the genital discs and gonad are altered and often fails in these mutants. Identification of motile sperm in the male mutants may explain why some offspring were produced; it seems likely that the abnormal morphology of the gonads is the causative effect of reduced fertility.

Expression of loco during male development
The phenotypes of the mutant testes and reproductive system suggested that organogenesis and not necessarily spermatogenesis was abnormal. Since differentiation of the testis and reproductive system occurs during metamorphosis we investigated the expression of loco in early and late pupae ( Figure 4A and 4B). We divided the pupae into early pupae and late pupae and performed RT-PCR on the various loco transcripts ( Figure 1A). The loco c1, loco c2, and loco c3 transcripts were expressed during the early stages and late stages of pupal development. The newly identified transcript, loco c5, was found to be differentially expressed and seemed to be detectable only in early staged pupae ( Figure 4A and 4B).
We have now shown that loco is expressed during metamorphosis. In order to ascertain whether loco is expressed during organogenesis of the male gonad, the original P element insertion line, which contains GAL4 in the loco gene [15], was crossed to a UAS-lacZ reporter strain. βgalactosidase activity was detected in the larval male gonads ( Figure 2B, arrow) as well as in the fat body tissue The adult reproductive tissue from the loco mutants surrounding the gonad. The observed β-galactosidase activity in the gonad confirms that expression of the loco gene takes place in the male gonad during development.

Expression of loco in mutant pupae and testes
We have demonstrated that loco is expressed in the developing male and adult gonads, furthermore we have shown that loss of loco expression resulted in reduced fertility and testis with abnormal morphology. In order to ascertain which transcript was affected by the mutation we performed RT-PCR for loco c1, loco c2, loco c3 and loco c5 in the mutant pupae of three of the male semi-sterile lines, loco 318 , loco 358 , and loco 387 ( Figure 4C-E). We detected loco c1, the 5'end of loco c2 and the loco c5 transcripts in the loco 318 mutant. However loco 318 completely lacked expression of the loco c3 transcript ( Figure 4C). Furthermore loco 318 only expressed the 5'end of loco c2. The 3'end of loco c2 was not detected ( Figure 4C, 3'c2) as only the 0.9 kb non-specific band was detected. Pupal expression from the loco gene in the loco 358 mutant produced loco c1, loco c2 and loco c5, however like loco 318 the loco 358 mutant lacked loco c3 ( Figure 4D). Pupal expression of loco c1, loco c2, loco c3 and loco c5 was detected in the loco 387 mutant ( Figure  4E). The testis expression of loco 358 was also found to lack any detectable loco c3 transcript.
The RT-PCR analysis was carried out on multiple different RNA samples and the same result was consistently achieved. Whenever the quality of the RNA was checked on a formaldehyde agarose gel it was found to be intact (data not shown). Furthermore detection of loco c1, loco c2 and loco c5 from the same cDNA samples which lacked loco c3 suggests that the RNA transcribed from the loco gene was intact. Taken together this data suggests that there is a requirement for loco c3 in adult reproductive tissue.

Analysis of the genomic sequence in the mutant lines
To gain further insight into molecular nature of the loco mutations, PCR was performed on genomic DNA isolated from the original c139 strain [15]. PCR utilising a P element primer directed to the 5' end of the P element (pgawb5a inv) in combination with a gene specific primer revealed that the P element was in reverse orientation to the loco gene (the 5' end of the P element was orientated toward the 3'end of the loco gene). Sequencing of the PCR product revealed that the P element had inserted 322 bp upstream of exon I-1 ( Figure 1A).
PCR analysis of the loco 318 mutant revealed that the P element primer directed to the 5' end of the P element (pgawb5a inv) worked in combination with gene specific primers in the forward and reverse orientation. This suggested that there was more than one P element present in loco 318 and that they were in the opposite orientation to one another. Direct sequencing of these products revealed that one P element was present at the original position (322 bp upstream of exon I-1) and the second P element was present at the same position but in the opposite direction ( Figure 1A). The sequencing of the PCR products from loco 318 also revealed a 9 bp duplication of genomic DNA on either side of the P elements ( Figure 1A). It is possible that two P elements 322 bp upstream of exon 2 alters pre-mRNA length such that there is premature dissociation of RNA polymerase II (RNAPol II), hindrance of the folding of the pre-mRNA molecule which prevents the joining of the splice sites, or it may disrupt important splice factor binding sites.
The P element primer used to reveal the position of the P element in the mutant line loco 318 failed to produce PCR products in the mutant line loco 358 . PCR across the original insertion site revealed that the P element had excised and had not deleted any genomic sequence ( Figure 1B). The mutant line loco 358 expresses white, it lacks expression of loco c3 ( Figure 4D) and the mutation genetically maps to the loco gene (Table 2). This suggested that a partial P element, which lacks the P element primer site, was present in the loco gene. PCR with the P element primer and FP8 produced a fragment of 2 kb, suggesting that the P element was 2 kb upstream of exon 2. Genomic PCR was performed across all introns in the loco gene. PCR failed only between exon I-1 and exon 2, whereas a PCR product across this region in wild-type genomic DNA was detected (data not shown). This further suggested that the partial P element in the mutant line loco 358 was located between exon I-1 and exon 2 and thus produced a PCR product too long to be detected by the PCR programme.
Direct sequencing of the insertion site in loco 387 revealed that no sequence had been deleted ( Figure 1B). The loco 387 line was found to express loco c1, loco c2, loco c3, and loco c5 ( Figure 4E). The reduction in male sterility when loco 387 is hemizygous with Df15, Df17 or loco Δ13 was statistically significant (Table 2) and the adult reproductive tissue of loco 387 is morphologically abnormal ( Figure 3E). This suggests that a mutation resides within the loco gene. Large rearrangements can occur during P element mobilisation. The PCR product across the loco insertion site was approximately 500 bp, therefore if a large inversion or rearrangement had occurred in loco 387 it would not be detected by this simple PCR. These data alongside the genetic data strongly suggests that the mutations lie within the loco gene.

Discussion
We isolated a number of homozygous lethal mutant lines of loco. These lines die at a variety of developmental stages, however, among them four lines were able to generate some homozygous adult males, which were semi-sterile.
We suggest that this is most likely to be due to a failure of the correct morphogenesis of the testis and reproductive organ derivatives of the larval gonad. This adds another role to the wide range of developmental decisions that are known to be dependent upon loco. Granderath et al., (1999) [14] showed that loco mutants died as embryos showing abnormalities in the contacts between glial cells [14]. Our previous studies illustrated that there is a requirement for loco in cytoplasmic dumping from the nurse cells to the oocyte and that loco is required for correct patterning of the eggshell and embryo [15]. There is also a large maternal supply of loco in the embryo probably explaining why the embryos die so late in embryogenesis. Finally, it was shown more recently that loco contributes to asymmetric cell division of neuroblasts [13]. These findings suggest that G-protein signalling may be important at wide variety of developmental stages in Drosophila.
The loco gene expresses several splice variants loco c1, loco c2, loco c3, and loco c4 [13][14][15]. Here we describe the expression of a fifth transcript, loco c5. We have analysed the expression of loco c1, loco c2, loco c3, and loco c5 in the wild-type testis and developing pupae and show that there is developmental regulation of loco c5 expression during morphogenesis. In addition we show that several G proteins are expressed in the male gonads and are therefore potential binding partners for the various LOCO isoforms. It is possible that the protein isoforms, expressing different conserved domains, will have different binding specificities and preferences for different G-proteins [27][28][29][30][31]. The G protein Gαi (G-oα65A) binds to loco c2 [14] and it is also co-expressed with loco in a variety of cell types [16,32]. We have shown by PCR that other Gα subunits are expressed (Goα47A, Gα49B, and Gα73B) in the testis and thus there is potential for LOCO to interact with other Gα subunits.
The analysis of the final morphology of the adult reproductive system in all of the flies analysed, strongly suggests that there is a failure in male gonad and genital morphogenesis It is possible that loco c3 expression could be the underlying reason for this phenotype. However the variability in testes morphology between flies may hint that there is some level of redundancy between the loco transcripts. Thus, whilst loco is clearly essential, a lack of or reduction of loco c3 expression does not cause a complete failure of gonad and genital differentiation. The loco mutants we isolated still express several loco transcripts, so further mutants will be needed which disrupt different transcripts or sets of transcripts to discover the role of loco and G-protein signalling in spermatogenesis and to further investigate it in imaginal discs and in the somatic cells of the gonad.

Conclusion
We show that all of the known loco spliceforms are expressed in the pupae and testis. In addition to this we have identified a fifth loco transcript, loco c5. We also show that there are a variety of Gα proteins expressed in the testis that may interact with the various LOCO isoforms. We have generated a set of new alleles of loco that affect the expression of specific loco transcripts. These deletions seem to be highly deleterious to Drosophila, as only a few adults hatch and the majority die as larvae. Mutant pupae and adult gonads of the few males that hatch show a loss of loco c3. We propose that loco c3 is needed for correct morphogenesis of the male gonad and the reproductive system derived from the male genital disc during metamorphosis. The role we have observed for loco in morphogenesis is in some ways similar to its role in glial cells where it has been proposed that G-protein signalling is important for shape changes [14]. Although the reproductive system is derived from the genital disc and the testis from the gonad, both tissues are affected. It, therefore, seems likely that loco is involved in cell-cell interactions during evagination and morphogenesis. During these processes cell and tissue shape changes are crucial.
These results support the well-documented findings that G-protein signalling is crucial throughout development. An extensive investigation is needed to identify the binding specificities of different loco isoforms, the temporal and spatial distribution of different loco transcripts and which Gα subunits co-localise with loco in the gonad and genital discs and in the adult male testis. With this information it will be possible to design genetic and molecular experiments to investigate the developmental mechanisms in which loco participates.

Stocks
Wild-type flies were OrR. Df(3R)17D1 Df(3R)15CE1, loco Δ13 and loco T1 were obtained from Christian Klambt. The original P insertion line was c139; it has an insertion of GAL4 in the loco gene [15]. The mutant lines were generated and described in Pathirana et al 2001 [15]. All Drosophila strains were raised on standard cornmeal-yeastagar medium at 25°C.

β-galactosidase staining of testes and gonads
Testes and male gonads were dissected from the progeny from c139 crossed to the UAS-lacZ reporter line. Staining was carried out as described by Deng et al 1999 [33].

DNA sequencing
The dideoxy chain determination method was used initially in the form of a Sequenase 3.1 kit (US Biochemicals), followed by automated sequencing on Perkin Elmer ABI 373A and 377A machines using dye labeled primers, then dye labeled terminator reactions. Sequenced fragments were assembled using GCG and GENE-JOCKEY software. Sequence analysis was done with GCG GAP, MAP, FASTA, TFASTA and PILEUP software. Conserved domains were predicted using SMART [34] and the NCBI conserved domain database [35].

Reverse transcription (RT)-PCR
Reverse transcription-PCR was carried out as described in Deng et al 1999 [33]. The primers used to identify transcripts were: