The Homeodomain Protein Defective Proventriculus Is Essential for Male Accessory Gland Development to Enhance Fecundity in Drosophila

The Drosophila male accessory gland has functions similar to those of the mammalian prostate gland and the seminal vesicle, and secretes accessory gland proteins into the seminal fluid. Each of the two lobes of the accessory gland is composed of two types of binucleate cell: about 1,000 main cells and 40 secondary cells. A well-known accessory gland protein, sex peptide, is secreted from the main cells and induces female postmating response to increase progeny production, whereas little is known about physiological significance of the secondary cells. The homeodomain transcriptional repressor Defective proventriculus (Dve) is strongly expressed in adult secondary cells, and its mutation resulted in loss of secondary cells, mononucleation of main cells, and reduced size of the accessory gland. dve mutant males had low fecundity despite the presence of sex peptide, and failed to induce the female postmating responses of increased egg laying and reduced sexual receptivity. RNAi-mediated dve knockdown males also had low fecundity with normally binucleate main cells. We provide the first evidence that secondary cells are crucial for male fecundity, and also that Dve activity is required for survival of the secondary cells. These findings provide new insights into a mechanism of fertility/fecundity.


Introduction
In many higher insects, the reproductive behavior of females drastically changes after mating [1,2]. Stimulation of egg laying and suppression of remating are induced by factors present in the male seminal fluid. The Drosophila male accessory gland secretes accessory gland proteins (Acps) into the seminal fluid, which are essential for male fertility/fecundity [3,4]. Each of the two lobes of the accessory gland is composed of two types of binucleate cell: about 1,000 main cells and 40 secondary cells [5]. Adult main cells are flat hexagonal cells and secondary cells are large spherical cells interspersed among the main cells at the distal tip of each accessory gland lobe. A well-known Acp, sex peptide (SP, also known as Acp70A), is secreted from the main cells and induces long-term postmating response, such as increased egg laying and reduced sexual receptivity, to increase progeny production [6,7,8,9,10]. These postmating responses are critically regulated through SP binding to the G-protein-coupled SP receptor in the female reproductive tract [11,12,13]. In addition, SP-sperm interaction is also required for long-term postmating response through localization of SP to sperm storage organs, and the Cterminal part of SP is gradually released from sperm tails [14,15]. In contrast to the increasing knowledge of Acps secreted from the main cells, little is known about physiological significance of the secondary cells.
Cell-fate determination of accessory gland primordia depends on fibroblast growth factor (FGF) signaling, and the mesodermal cells expressing an FGF receptor, Breathless (Btl), are recruited into a part of the male genital disc during late larval development [16]. The btl-expressing cells become epithelial, and give rise to accessory glands (paragonia) and seminal vesicles (vas deferens). Subsequent cell proliferation and functional differentiation of the accessory gland are regulated by the homeodomain transcription factor Paired (Prd) [17]. Mutant males for prd are sterile as they have severely reduced or no accessory glands [18,19,20], indicating that seminal fluid components from the accessory gland are essential for male fertility.
The homeodomain transcriptional repressor Defective proventriculus (Dve) is involved in various functions including wing morphogenesis, leg joint formation, head vertex specification, ommatidial cell-type specification, and functional specification of the midgut [21,22,23,24,25,26,27]. Here, we provide evidence that Dve is required for male accessory gland development, binucleation of main cells and survival of secondary cells, and also that secondary cells are essential for male fecundity.

Spatio-temporal pattern of Dve expression during accessory gland development
We have found that Dve is expressed in the male accessory gland at least from 24 hr after puparium formation (APF) but not in the male primordia of the genital disc in the late third-larval instar ( Figure 1A, 1B). Dve is expressed strongly in secondary cells and weakly in main cells at 72 hr APF ( Figure 1C), but Dve expression in main cells is undetectable in the adult stage ( Figure 1D). To generate a GAL4 driver line that can induce gene expression during early stages of accessory gland development, we established a dve-GAL4 line, in which GAL4 expression is under the control of the 13-kb regulatory element upstream of the first exon of dve. Expression pattern of the green fluorescent protein (GFP) driven by the dve-GAL4[35A] (dG35A.GFP) is nearly identical to that of endogenous Dve protein, although low-level expression in main cells could be detected in the adult stage ( Figure 1E-H).
The homeobox gene paired (prd) is required for accessory gland development and its mutation leads to greatly reduced size of the accessory gland [17]. We compared Dve expression with that of Prd, and found that the onset of their expression was nearly the same ( Figure 1I-K). However, Prd expression in main cells is maintained in the adult stage ( Figure 1L). In the midgut, Dve is coexpressed with the copper cell determinant Labial in embryonic precursor cells, and the subsequent dve repression in the larval copper cells is regulated by Labial and Dve itself [24]. We hypothesized that the temporally regulated dve repression in accessory gland main cells is also regulated in a similar manner. Therefore, we examined the effect of prd mutation on Dve expression. Interestingly, Dve was derepressed in prd mutant main cells, which also had lost their binucleate character ( Figure 1M). In contrast, prd mutant secondary cells were normally binucleate and Dve expression was maintained in these cells ( Figure 1N). Thus, the Prd activity is required for temporally regulated dve repression in the main cells and it is dispensable for Dve expression in the secondary cells.

Dve-A is required for secondary cell development
Two transcripts, type A (4.9 kb) and type B (3.5 kb), have been identified in the dve locus. The type-A (dve-A) null allele dve E181 has about 4-kb deletion including the first exon for the dve-A transcript, while the dve E144 allele has a smaller deletion [22]. Homozygosity for dve E181 is semi-lethal, and escaper adult males have smaller accessory glands (Figure 2A, 2B). Because the body size of dve E181 homozygous males was also small ( Figure 2D), we measured the size of ejaculatory bulb as an internal control. The size ratio of cross-section area (accessory gland/ejaculatory bulb, AG/EB) was significantly reduced in males of dve E181 homozygotes and heteroallelic combination (dve E181 /dve E144 ) ( Figure 2E). Similar phenotype was also observed in RNAi-mediated dve knockdown (dve KD, dG30A.dve-IR dve 1 ) ( Figure 2E and Figure S1). To perform efficient knockdown, we recombined UAS-dve-IR with a loss of function allele, dve 1 [24,25], to generate a recombinant chromosome, UAS-dve-IR dve 1 . Notably, these dve-A mutant accessory glands completely lacked cells with the appearance of mature secondary cells and some of their main cells were mononucleate ( Figure 2B9, 2B0, 2F). Secondary cells can be distinguished by their spherical cell shape and the presence of large vacuoles. The vacuolar components are non-specifically stained with anti-Prd antibody (red in Figure 2A9, 2A0). Interestingly, they were rarely detected in very small atrophic cells of dve-A mutants (arrow in Figure 2B0), suggesting that cell fate for the secondary cell could be induced in dve-A mutants.
Because prd mutant main cells are mononucleate ( Figure 1M), we checked whether these dve-A mutant phenotypes are due to the absence of Prd activity. Prd expression was unaffected in dve E181 mutants ( Figure 2B0), indicating that the Dve activity is crucial for growth of the accessory gland, binucleation of main cells, and secondary cell differentiation even in the presence of Prd proteins. To further confirm the Dve functions in accessory gland development, we performed mosaic analysis using the loss of function allele dve 1 . Mutant clones for dve 1 showed the same phenotypes as the accessory gland did in a dve-A mutant male. Cells with secondary morphology were only detected in GFPpositive wild-type cells but not within dve 1 mutant clones ( Figure 2C, 2C9).
Seminal fluid from dve mutant males is defective despite the presence of sex peptide As prd mutant males who have very small accessory glands are sterile due to a defect in ejaculation [18,19,20], we checked the fertility/fecundity of dve E181 mutant males. The dve-A mutant sperms had normal motility and they were correctly localized to the female sperm storage organs, seminal receptacle (src) and spermathecae (spt) ( Figure 3A-G). However, the average number of progeny from a wild-type female mated with a dve-A mutant male was significantly lower than those mated with control males ( Figure 4A). Therefore, it is assumed that the low fecundity of dve-A mutant males reflects the defects of seminal fluid proteins that are secreted from accessory gland cells. We examined two potential reasons for the low fecundity: (1) failure of egg-laying stimulation, and (2) reduced rate of fertilization as reported in wasted mutant sperm [28]. Females mated with dve-A mutant males laid substantially fewer eggs than those mated with control males, although the hatching rate was indistinguishable between females mated with dve-A mutant or control males ( Figure 4B and data not shown). These results strongly suggest that the low fecundity of dve-A mutant males is due to the failure of egg-laying stimulation to mated females.
The sex peptide (SP), also known as Acp70A, is a major accessory gland protein that induces changes in female postmating response such as egg laying and receptivity [7,10]. Females mated with SP null mutant males lay fewer eggs, and they show high receptivity to the second males (i.e., remating activity) at least 12 hr after mating. The mating plug protein PEBII, which is secreted from the ejaculatory bulb, contributes to reduced receptivity at earlier period (4 hr after mating) [29]. Thus, we examined remating activity of a female firstly mated with a dve-A mutant male and secondly with a wild-type male. The remating activity at 4 hr after first mating was not evident ( Figure 4C). Interestingly, females at 48 hr after first mating with dve-A mutant males showed high remating activity ( Figure 4D). These results raised the possibility that SP expression or its transfer is impaired in dve-A mutant males. Thus, the SP expression in main cells was monitored using a GFP fusion protein (SP::GFP), whose expression is under the control of the endogenous SP promoter [30]. Unexpectedly, SP::GFP was normally expressed and secreted into the lumen of the dve-A mutant accessory glands, and it was normally transferred into the female reproductive tract after mating ( Figure 3H-J).

Secondary cells are essential for male fecundity
To examine whether the low fecundity of dve mutant males is due to the mononucleate state of SP-expressing main cells, we checked the postmating response of females after they had mated with dve KD males. In dve KD males, binucleation of main cells was relatively normal in contrast to mononucleate main cells seen in dve-A mutants ( Figure 2F and Figure 5A). This does not mean that dve KD main cells are functionally normal, however, secondary cell differentiation was severely inhibited and small atrophic cells were frequently observed ( Figure 5A9). Females mated with the dve KD or dve-A mutant males showed the same phenotypes in egg laying and remating activity ( Figure 5B-D), suggesting the importance of mature secondary cells for male fecundity.
Although little is known about physiological significance of secondary cells, Abd-B mutant males also lack mature secondary cells and show low fecundity similar to that of dve-A mutants (personal communication from M. F. Wolfner and F. Karch). These results further support the importance of secondary cells and prompted us to examine the relationship between Dve and Abd-B during secondary cell development. Expression patterns of Dve and Abd-B are nearly complementary until 24 hr APF and change to the same pattern from 48 hr APF to the adult stage ( Figure S2A-H). Abd-B expression after the late pupal stage appears to be dependent on the Dve activity, because it was greatly reduced in dve mutant clones and also in adjacent wild-type cells ( Figure S2E). Thus, it seems likely that the low fecundity in dve mutant males reflects the absence of Abd-B expression with loss of mature secondary cells. Taken together, these results strongly suggest that unknown factors secreted from secondary cells are essential for male fecundity.

The common region of Dve isoforms is required for survival of secondary cells
In dve-A mutant or dve KD accessory glands, we could detect very small atrophic cells ( Figure 2B0 and Figure 5A9). These observations suggest a mechanism that Dve is required for cell survival rather than cell-fate determination of secondary cells. Consistent with this notion, we could detect a few secondary cell precursors in dve-A mutants during pupal development. These mutant cells had a characteristic feature of secondary cells in their nuclear arrangement and also showed reduced Prd expression ( Figure 6A, 6B). Because Prd expression in main cells was unaffected in dve mutants ( Figure 2B9, 2C and Figure S2E), these cells appear to be precursors of atrophic secondary cells and they were detected only in the distal region ( Figure 6B). It seems likely that most precursor cells rapidly die in dve-A mutants, because it was quite difficult to detect these precursor cells even in earlier stages. Loss of secondary cells in dve-A mutants was rescued by dve-A expression ( Figure 6C). Interestingly, expression of dve-B also rescued loss of secondary cells in the dve-A mutant background ( Figure 6D). These results imply that the common region of two isoforms including a compass domain (CMP) and two homeodomains (HD-N and HD-C) is responsible for survival of secondary cells.

Discussion
Mutant males for dve-A showed low fecundity together with loss of secondary cells and reduced size of accessory glands. It has been reported that greatly reduced size of accessory glands results in sterility, and also suggested that there is a minimum size to maintain fertility [31]. If a male is selected for larger size of accessory glands with 16 generations, the selected males have good fecundity. However, they have only about 1.4-fold larger size compared to the control males, suggesting that there is also a maximal size of accessory gland not to waste energy [31]. Thus, the size of accessory glands should be controlled in an appropriate range and binucleation seems to be the best strategy to provide highly plastic change of the size [32]. Although reduced size of dve-A mutant accessory glands may have some effects on fecundity, dve KD males had similar size of accessory glands to the dve-A heterozygous controls and dve KD males showed low fecundity with loss of secondary cells ( Figure 2E and Figure 5). Thus, it is most likely that the low fecundity in dve-A mutant males is due to the absence of mature secondary cells (Figure 7). This is consistent with independent findings that Abd-B is required for maturation of secondary cells and for maintaining female postmating response (personal communication from M. F. Wolfner and F. Karch). Although we cannot exclude the possibility that some defects in dve mutant main cells affect the fecundity, SP was normally expressed and transferred into the female reproductive tract (Figure 3H-J). Thus, the following mechanisms should be considered for SP activation to induce long-term postmating response; (1) secondary cell products cooperate in parallel with SP-mediated signaling; (2) secondary cell products enhance SP binding to its receptor; (3) secondary cell products stabilize SP binding to sperm; and (4) secondary cell products are involved in modification and/or stabilization of SP secreted from main cells. Interestingly, egg laying of females mated with dve mutant or dve KD males was gradually reduced over time ( Figure 4B, 5B), suggesting that the last two interpretations, stabilization of SP by secondary cell products, are more plausible. Identification of unknown factors secreted from secondary cells will provide new insights into a mechanism that is crucial for activation of seminal fluid functions.
It seems likely that Dve functions are crucial to inhibit cell death of secondary cells, and an intriguing possibility is that inactivation of the Dve activity is closely linked to the regulated cell death to adjust the number of secondary cells. Dve and the special AT-rich sequence binding proteins (SATBs) belong to the cut superclass of homeobox genes and have an evolutionarily conserved CMP [33]. It is reported that SATB1 is cleaved and inactivated by Caspase 6 in response to the apoptotic signaling pathway [34]. Expression of the BCL2 gene, which is a key regulator to inhibit apoptosis, is finely tuned by a variety of stimuli and activated through SATB1mediated chromatin looping. Thus, SATB1 is required for cell survival through inhibition of programmed cell death [35,36]. The functional similarity between Dve and SATB1 for inhibition of cell death raises a possibility that an evolutionarily conserved CMP plays important roles to inhibit cell death. The CMP of SATB1 is characterized as a PDZ-like domain (amino acids 90 to 204) involved in protein-protein interaction, and the Caspase 6dependent cleavage of SATB1 at amino acid position 254 disrupts the dimerization of SATB1. Further characterization of CMPinteracting proteins will clarify the underlying mechanism and provide new insights into a regulatory mechanism of fecundity/ fertility.

Fly stocks
Flies were reared on standard yeast-glucose-cornmeal-agar medium at 25uC. Oregon-R flies were used as wild-type controls. dve E181 and dve E144 are dve-A-specific mutant alleles that remove the first exon [22]. dve 1 is a severe loss of function allele that has no dve-A and a very weak dve-B activity in the larval midgut [22,24]. dve L186 is a null allele that completely removes the gene [37]. To knockdown the dve activity in a spatio-temporal manner, we used UAS-dve-IR (v109538, Vienna Drosophila RNAi Center), which can induce RNA interference by expressing the inverted repeat (IR) sequence of exon 3. To perform efficient knockdown, a recombinant line, UAS-dve-IR dve 1 , was used for experiments. SP null mutant males were produced by crossing SP 0 /TM3 Sb males with D130/TM3 Sb females. The resulting SP 0 /D130 (SP null) males produce no SP [10]. The following GFP marker strains were used: protamineB::GFP for sperm nuclei (Drosophila Genetic Resource Center, DGRC) [38] and SP::GFP for expression of sex peptide [30].

Establishment of dve-GAL4 lines
The lacZ sequence of pCaSpeR-AUG-b-gal vector [39] was replaced with the GAL4 sequence from pGaTB [40]. This pCaSpeR-GAL4 vector has unique restriction enzyme sites (KpnI, EcoRV, and BamHI) upstream of the GAL4 open reading frame. The 5.8-kb fragment, which includes the upstream promoter region and the first exon (to +557) of the dve gene, was inserted into the KpnI-BamHI sites of pCaSpeR-GAL4 to generate pCaSpeR-5.8EDE-GAL4. The 5.4-kb fragment, which includes further upstream sequences, was inserted into the KpnI-EcoRI sites of pCaSpeR-5.8EDE-GAL4 to generate pCaSpeR-5.4-5.8E-GAL4. The 1.6-kb EcoRI fragment was inserted into the junctional EcoRI site between 5.4-kb and 5.8-kb fragments of pCaSpeR-5.4-5.8E-GAL4 to generate the dve-13kb-GAL4. This construct expresses

Behavior assays
All flies were cultured at 25uC under 12 hr light/dark cycles. Males and females were collected at eclosion and aged separately for 4-6 days in groups of 10-30 flies per vial. At circadian times 0 to 3 (CT0, subjective dawn), a pair of male and female flies was placed on a food medium with an observation chamber (30-mm diameter64-mm depth) for 1 hr. A mated female was mildly aspirated into a vial and used for the following assays.
Progeny assay. Male fertility/fecundity was measured as the number of viable progeny per mated wild-type female during 4 days after mating. A mated female was transferred twice to a fresh vial and allowed to lay eggs for 2 days. The number of adult progeny from each vial was counted.
Egg-laying/hatching assay. A mated female was transferred to a fresh vial every 24 hr for 7 days, and eggs laid per day were counted. Hatching rate was determined by counting the number of unhatched eggs during next 24 hr.
Remating assay. At 4 hr and 48 hr after first mating, mated females were individually paired with one new virgin male (4-6 days old) for 1 hr. Several independent sets of remating assay were performed and the cumulative percentage of remated females was calculated.

Statistical analyses
The significance of differences between the control and test progenies was analyzed with t-tests or one-way ANOVA using Kaleidagraph software version 3.6 (HULINKS). The levels of significance are indicated by asterisks: *P,0.01, **P,0.001.