The WASp-based actin polymerization machinery is required in somatic support cells for spermatid maturation and release

Spermatogenesis constitutes a remarkable process of cell differentiation, during which small, immature spermatogonia are transformed into elongated and motile spermatids (Fuller, 1993; Roosen-Runge, 1977). A crucial feature of this process is the intimate association and cross-talk between the differentiating germ cells and specialized populations of somatic support cells. Studies in the Drosophila model system have underscored the significance of soma-germline interactions during spermatogenesis and have identified some of the key molecular mechanisms involved. Much of this analysis has focused on the establishment and maintenance of the germline stem cell niche and on the early phases of sperm differentiation (Fabrizio et al., 2003; Kiger et al., 2000; Leatherman and Dinardo, 2010; Schulz et al., 2002; Tulina and Matunis, 2001). Although the close germline-soma association persists throughout Drosophila spermatogenesis, much less is known about the impact of this association on more advanced stages of sperm differentiation.

Spermatogenesis in Drosophila commences at the apical tip of the testis, where the germline stem cells reside, and is initiated by a series of incomplete mitotic divisions and meiosis, producing a cyst with 64 interconnected haploid spermatids. A pair of somatic cyst cells, themselves generated from a separate apical stem cell population, enwraps the germline cyst from the outset, providing it with a thin outer lining (Fuller, 1993) (Fig. 1A). Upon completion of meiosis, the spermatocytes enter a prolonged phase of differentiation, during which they acquire the unique morphological and functional features of individual, elongated and motile spermatids. The differentiating spermatids remain enclosed within cysts throughout this process, as they journey progressively towards the basal end of the testis tube.

The pair of associated somatic cyst cells, which continue to line the differentiating cysts, accommodate the dramatic alterations to cyst size and organization through morphological changes of their own. One of the two cells, termed the tail cyst cell, spreads out so as to cover the elongating sperm tails, whereas the second, so-called head cyst cell, provides an enclosure for the bundled spermatid nuclei. Transfer of sperm from the testis and into the reproductive tract involves a complex series of events, which unfolds once the fully differentiated cysts reach the base of the testis tube (Fig. 1A). The head cyst cell becomes anchored in a terminal epithelium (TE) present at this location, whereupon the ~2 mm long and narrow cyst coils from its apical end to form a compact structure. Shortly thereafter, single, motile spermatids are released through the head cyst cell-TE interface and travel via the short testicular duct into the adjoining seminal vesicle (Miller, 1950).

Although this final phase of spermatogenesis has been the subject of detailed and elegant analysis by electron microscopy (Tokuyasu et al., 1972), identification and characterization of the underlying cellular features and molecular mechanisms are yet to be achieved. Here we describe the involvement of somatic support cells during the final phases of Drosophila sperm maturation, just prior to their release as freely motile cells into the reproductive tract. We present data suggesting that microfilament organization within the head cyst cell, mediated by the WASp-Arp2/3 branched actin polymerization machinery, is essential for cyst anchoring at the base of the testis and subsequent events.

Polymerization of actin into branched microfilament arrays, mediated by the Arp2/3 protein complex, provides a mechanistic basis for a wide variety of basic cellular processes, including cell motility, the formation of cell-cell and cell-matrix adhesions, and endocytosis (Goley and Welch, 2006). Assignment of specific developmental tasks to general cellular machineries such as Arp2/3-based actin polymerization is problematic owing to their ubiquitous nature. One solution is to approach these issues by investigating pathway elements that play tissue-specific or developmental stage-specific roles. Activity of the Arp2/3 complex requires stimulation by nucleation-promoting factors (NPFs), such as WASp and SCAR/WAVE (Kurisu and Takenawa, 2009). The relatively restricted utilization of the Drosophila WASp homolog Wsp as an Arp2/3 NPF makes it a useful tool in this context (Zallen et al., 2002). Furthermore, the NPF activity of WASp family proteins depends upon its proper activation and localization by a variety of factors, which may be utilized only in specialized settings. For example, the restricted expression and function of D-WIP (also known as Vrp1), the Drosophila WASp-interacting protein (WIP) homolog, in developing embryonic muscles was key in revealing the involvement of the WASp-Arp2/3 pathway in myoblast fusion (Kim et al., 2007; Massarwa et al., 2007). Similarly, our discovery that activation of Wsp by the GTP-bound form of the signaling element Cdc42 occurs specifically during late spermatogenesis has now paved the way for the study of Wsp-Arp2/3 function in this particular developmental context.

We show here that Wsp, Cdc42 and the Arp2/3 complex are all required in the somatic cyst cells for the final phase of cyst maturation in the testis: attachment to the basal TE and the subsequent transfer of motile sperm out of the cysts and into the seminal vesicle. A similar requirement is demonstrated for the cell surface protein Sticks and stones (Sns) and for the adaptor element D-WIP, both of which are required, like Wsp, to promote embryonic myoblast fusion. Sns is expressed, along with a testis-specific isoform of D-WIP, in the head cyst cell, where Wsp, D-WIP and Cdc42 are shown to mediate microfilament organization. The concerted activity of Wsp and associated elements therefore appears to take place in the specific cell that associates with the TE and through which sperm ‘escape’ into the seminal vesicle. Finally, we demonstrate an analogous scenario in the mammalian testis, in that murine N-WASp (Wasl – Mouse Genome Informatics) is required in Sertoli cells, the major somatic support cells, for spermatocyte maturation. This suggests an evolutionary conserved function for WASp-based branched actin polymerization during spermatogenesis.

Gene targeting designed to generate the D-WIP^KO allele was performed as described (Tsuda et al., 2010). For constructing the P[donor] plasmid, ~3 kb of both the upstream and downstream regions of the D-WIP locus were amplified and the amplified fragments were subcloned into the polylinker of pP[EndsOut2] (a gift from J. Sekelsky, University of North Carolina). The mini-white gene, a marker for transformation, was inserted between the upstream and downstream target sequences. Targeted lines were backcrossed to the y w^67c23Df(1)w^67c23 stock for six generations. D-WIP^testis constructs were based on PCR amplification of clone AT07076 from an adult fly testis cDNA library (Stapleton et al., 2002). Amplified material was cloned into the appropriate Drosophila Gateway vectors (T. Murphy, Carnegie Institution of Washington) using the Gateway cloning system (Invitrogen). UAS-dsRNA constructs were commonly expressed in flies together with UAS-Dicer2 to enhance RNAi activity (Dietzl et al., 2007). The following lines were used.

Mutant alleles: for Wsp, Wsp¹ and Df(3R)3450 (Ben-Yaacov et al., 2001); for D-WIP, D-WIP^D30 (Massarwa et al., 2007), sltr^S1946 (Kim et al., 2007) and D-WIP^KO (this study).

GAL4 drivers: arm-GAL4 (ubiquitous) (Sanson et al., 1996); eyaA3-GAL4 (cyst cells) (Leatherman and Dinardo, 2008); PpY55A-GAL4 (cyst cells) (Jung et al., 2007); T155-GAL4 (cyst cells) (Hrdlicka et al., 2002); rp298-GAL4 (duf gene) (Menon and Chia, 2001); and Mef2-GAL4 (myogenic mesoderm) (Ranganayakulu et al., 1996).

UAS-Wsp lines: UAS-Wsp, UAS-Wsp^H242D (Tal et al., 2002) and UAS-Wsp-GFP (Massarwa et al., 2007).

UAS-D-WIP lines: UAS-D-WIP (Massarwa et al., 2007), UAS-D-WIP-GFP and UAS-D-WIP^testis-GFP (this study).

Other UAS-based transgenes: UAS-RedStinger (nuclear marker) (Barolo et al., 2004); UAS-Moesin-GFP (microfilament marker) (Chihara et al., 2003; Dutta et al., 2002); and UAS-Dicer2 (Dietzl et al., 2007).

UAS-dsRNA lines: D-WIP (KK102253), Wsp (KK108220), Cdc42 (1253OR-2), sns (GD877), Arp3 (GD35260) and Sop2 (GD42172). GD and KK lines are from the VDRC (Vienna) collection and the Cdc42 line is from NIG-FLY/DGRC (Kyoto).

Other lines: don juan-GFP (spermatid marker) (Santel et al., 1997) and sns-GCN-lacZ (sns reporter) (Zhuang et al., 2009).

Testes were derived from sexually mature (1- to 3-day-old) flies, which were further aged (where appropriate) for 3 days in isolation from females to allow sperm accumulation in the seminal vesicle. Testes (intact or ruptured to expose cysts) were dissected in TB (Arama et al., 2003) and fixed for 15 minutes at room temperature (RT) in 4% paraformaldehyde (PFA). A ~2-hour block at RT in blocking solution [phosphate-buffered saline (PBS) containing 0.1% Triton X-100 (PBT) plus 3% bovine serum albumin (BSA)] was followed by overnight incubation with primary antibodies, diluted in blocking solution, at 4°C in a humid chamber. Primary antibodies used were chicken anti-GFP (Aves; 1:500), guinea pig anti-D-WIP [(Eriksson et al., 2010) 1:500], rabbit anti-DsRed (Clontech; 1:500), rabbit anti-β-galactosidase (Cappel; 1:3000), mouse anti-polyglycylated tubulin [AXO 49 (Bre et al., 1996); 1:100] and mouse anti-Eya (Developmental Studies Hybridoma Bank, University of Iowa; 1:10). Following washes in PBT, testes were incubated for 1 hour with the appropriate fluorophore-conjugated secondary antibodies. Secondary antibodies (Jackson ImmunoResearch) were diluted 1:200 in blocking solution. Where used, phalloidin-TRITC (Sigma, 4 ng/μl) was added together with the secondary antibodies. Preparation and immunostaining of cryosectioned material were performed as described (Kaplan et al., 2010). All stained preparations were mounted in Vectashield/DAPI (Vector Laboratories). Images were acquired using Zeiss LSM 510 and LSM 710 confocal scanning systems.

N-WASp was disrupted in murine Sertoli cells by the following scheme. Mice bearing Dhh-Cre (Jaegle et al., 2003) were crossed to heterozygotes for the null allele N-WASp⁻ (Snapper et al., 2001). Dhh-Cre/+; N-WASp⁻/+ progeny (selected by tail PCR) were then crossed to mice homozygous for the conditional knockout allele N-WASp^flox (Cotta-de-Almeida et al., 2007), and Dhh-Cre/+; N-WASp^−/flox males were identified among the progeny. Incorporation of the Scx-GFP marker (Pryce et al., 2007) was obtained by crossing mice bearing this transgene to Dhh-Cre/+; N-WASp⁻/+, prior to the final cross to the N-WASp^flox homozygotes.

Hematoxylin and Eosin (H&E) histochemistry was performed on 10 μm paraffin sections, following fixation of dissected testes in Bouin's fixative. For immunofluorescence, dissected testes were commonly fixed in 4% PFA overnight at 4°C, washed with PBS and incubated in 30% sucrose overnight at 4°C, following which OCT blocks were prepared and cryosectioned (7 μm). Sections were permeabilized using 0.25% Triton X-100, 1% BSA in PBS (15 minutes at RT), blocked in 1% BSA and 5% goat serum in PBS (1 hour at RT) and incubated with chicken anti-GFP (Aves; 1:500) in a humid chamber overnight at 4°C. An alternative protocol used Bouin-fixed paraffin sections, which were deparaffinized, permeabilized and stained with anti-GFP as above, and then incubated at 37°C for 1 hour with undiluted rat IgM anti-Gcna1 (Enders and May, 1994). Following washes, all sections were incubated with secondary antibodies (1 hour at RT), washed and mounted in Vectashield.

Drosophila Wsp has been shown to contain a conserved and functional GTPase-binding domain that bears high affinity for the active (GTP-bound) form of Cdc42 (Ben-Yaacov et al., 2001). It was therefore surprising that previous analysis had failed to identify specific requirements for Cdc42 in the activation of Wsp within a developmental process. In these studies, Wsp^H242D, a construct encoding a Wsp variant that is incapable of binding GTP-Cdc42 (Tal et al., 2002), fully rescued the lethality of Wsp mutant flies, as well as the prominent neurogenic and myogenic Wsp phenotypes (Massarwa et al., 2007; Tal et al., 2002). Further analysis, however, revealed that Wsp mutant male flies, rescued using the Wsp^H242D construct, are completely sterile.

In order to assess possible spermatogenic defects in the Wsp^H242D-rescued flies, we made use of the spermatid marker Don Juan (Dj)-GFP (Santel et al., 1997). In wild-type testes (Fig. 1B), this marker visualizes the elongated cysts and the seminal vesicle, which is full of individualized, mature sperm. A similar arrangement of elongated cysts and a full seminal vesicle was observed when hemizygous Wsp¹ mutant flies are rescued with an intact UAS-Wsp construct under the control of the ubiquitous armadillo (arm)-GAL4 driver (Fig. 1C). However, when Wsp¹ flies were similarly rescued with the Wsp^H242D construct, the testis contained cysts that were full of elongated spermatids but no sperm was released into the seminal vesicle, which remained empty (Fig. 1D). Strikingly, the cysts of Wsp^H242D-rescued flies did not undergo coiling. Moreover, the extreme basal end of the testis tube, lined by the TE and leading up to the seminal vesicle, was free of cysts altogether (Fig. 1B′-D′).

We monitored several maturation markers in order to determine the differentiation stage of spermatids in Wsp^H242D-rescued flies. One marker is Dj-GFP itself, which is observed exclusively during late post-meiotic stages of spermatogenesis (Santel et al., 1997) and is robustly expressed by Wsp^H242D-rescued flies. We next followed sperm individualization, the process by which the 64 syncytial spermatids within each cyst are enclosed within distinct cell membranes. Dynamic cones of F-actin serve as established hallmarks of the individualization process (Noguchi and Miller, 2003). An examination of the individualization cones in the Wsp^H242D-rescued flies suggested that spermatid individualization in these mutants proceeds normally (see Fig. S1 in the supplementary material). Finally, we stained testes from wild-type and Wsp^H242D-rescued flies with an antibody against polyglycylated tubulin, a post-translational modification that occurs at the onset of spermatid individualization (Arama et al., 2007; Bre et al., 1996), and obtained similar staining patterns in wild-type and mutant flies (see Fig. S1 in the supplementary material). Taken together, these data suggest that the Wsp mutant phenotype represents a late, post-individualization stage block in spermatogenesis, just prior to the onset of the cyst coiling process.

As described above, cyst coiling and release of mature spermatids into the seminal vesicle are associated with a series of cellular events involving both the spermatids and the supporting somatic cyst cells, raising the question of the cell type in which Wsp is required. To address this issue we employed an RNA interference (RNAi) approach to disrupt Wsp activity specifically within the somatic cyst cells, using eyaA3-GAL4 (Leatherman and Dinardo, 2008), a GAL4 driver based on the cyst cell factor eyes absent (eya) (Fabrizio et al., 2003), and a UAS-Wsp^RNAi construct. Knockdown of Wsp in this manner resulted in a male-sterile phenotype, characterized by spermatogenic arrest at the elongated cyst stage and empty seminal vesicles (Fig. 1G), similar to the phenotypes that we obtained following rescue of the Wsp¹ mutant with the Wsp^H242D construct.

Identification of the somatic cyst cells as a site of Wsp activity now opened the door to evaluation of the requirement for Wsp-associated elements in mediating late-stage spermatogenesis and spermatid release. We first sought to complement the results obtained with the Wsp^H242D construct by directly demonstrating a requirement for Cdc42 in these processes. Indeed, cyst cell expression of RNAi directed against Cdc42 led to the same phenotypes obtained with cyst cell Wsp RNAi and Wsp^H242D rescue, i.e. the testis contained mature cysts but the seminal vesicle was empty (Fig. 1H).

Wsp mediates actin filament nucleation through activation of the Arp2/3 complex. Knockdown of Arp2/3 complex activity was achieved using simultaneous eyaA3-GAL4-based expression of RNAi constructs directed against Sop2 and Arp3 (Arp66B – FlyBase), which encode two subunits of the complex (Hudson and Cooley, 2002). Phenotypes similar to those obtained following Wsp and Cdc42 disruption were again observed (Fig. 1I). Taken together, these observations suggest that Wsp activation of the Arp2/3 complex as mediated by GTP-bound Cdc42, which constitutes a fundamental molecular pathway for the generation of branched microfilament arrays (Carlier et al., 1999), is active in the somatic cells that enclose spermatogenic cysts and is essential for the final stages of cyst maturation and spermatid release.

Although WASp activation (in this case by GTP-Cdc42) and Arp2/3 stimulation are core features of the actin microfilament nucleation machinery, additional factors are commonly required to adapt this machinery to specific tissue and cell settings. The study of such factors is warranted, as they can shed light on the mechanistic properties and subcellular localization of the actin nucleation process. Although not capable of WASp activation, members of the conserved WASp-interacting protein (WIP) family have been shown to regulate WASp activity by various means, including stabilization and localization to sites of action (for a review, see Anton et al., 2007). We investigated whether D-WIP, the sole Drosophila WIP homolog, contributes to the final phases of spermatogenesis.

The primary developmental role of D-WIP is to promote embryonic myoblast fusion, which is achieved through its interaction with Wsp (Berger et al., 2008; Kim et al., 2007; Massarwa et al., 2007). This highly specific requirement allowed us to obtain viable flies by expressing a UAS-D-WIP construct (Massarwa et al., 2007) using the muscle-specific Mef2-GAL4 driver (Ranganayakulu et al., 1996) in a D-WIP null mutant background (sltr^S1946/D-WIP^D30). Such male flies, which are mutant for D-WIP in all non-muscle tissues, are completely sterile. These males showed a spermatogenic phenotype that was highly similar to that observed following rescue of the Wsp¹ mutant with the Wsp^H242D construct (Fig. 2A,A′): the testis contained mature cysts that did not undergo appreciable coiling, cysts did not occupy the base of the testis tube, and the seminal vesicle was completely empty of released sperm.

A similar male-sterile phenotype was exhibited by flies homo- or hemizygous for D-WIP^KO, a hypomorphic and viable D-WIP allele (Fig. 2B,B′; see Materials and methods for details of D-WIP^KO generation). Spermatids from both of these D-WIP mutant backgrounds expressed late-stage maturation markers, as observed for Wsp^H242D-rescued males (see Fig. S1 in the supplementary material). Furthermore, the characteristic late-stage arrest of spermatogenesis occurred following RNAi-mediated knockdown of D-WIP in cyst cells (Fig. 1J). Together, these results suggest that somatic regulation of sperm maturation constitutes a second cell-cell communication process, in addition to myoblast fusion, in which D-WIP and Wsp act in concert to mediate tissue differentiation during Drosophila development.

During embryonic myogenesis, D-WIP strongly localizes to sites of myoblast fusion (Berger et al., 2008; Kim et al., 2007; Massarwa et al., 2007), highlighting the cellular location where Wsp-mediated actin polymerization is active in this context. The essential involvement of D-WIP in late spermatogenesis is therefore significant, as it provides a potential tool for determining the location of Wsp pathway activity. We thus sought to localize D-WIP within developing spermatogenic cysts. This approach proved to be informative, as immunofluorescent detection using anti-D-WIP antisera revealed a specific accumulation of the protein within head cyst cells, in close proximity to the sperm nuclear bundle (Fig. 2C-C″). This localization pattern appears to be restricted to particularly mature cysts, which can be distinguished by the tight packing of the spermatid nuclei following individualization (Jayaramaiah Raja and Renkawitz-Pohl, 2005). D-WIP localization suggests, therefore, that the Wsp pathway is activated within the head cyst cell, a somatic support cell that physically associates with both mature spermatids and the TE at the basal end of the testis, near the time of spermatid release.

Since the available antibodies did not provide clear subcellular resolution, we sought to refine our observations of D-WIP localization using GFP-tagged forms. We generated transgenic flies harboring either UAS-D-WIP-GFP, a construct based on the major embryonic form of D-WIP (Kim et al., 2007), and UAS-D-WIP^testis-GFP, representing a novel, testis-specific variant that we had isolated (see Fig. S2 in the supplementary material). For expression of these constructs in the testis, we made use of the moderate cyst cell-specific driver PpY55A-GAL4 (Jung et al., 2007) to avoid possible overexpression artifacts and the cellular debris associated with the strong eyaA3-GAL4 driver. Surprisingly, although the D-WIP-GFP fusion protein displayed a localization pattern that was highly similar to that of endogenous D-WIP when expressed in embryonic myoblasts (not shown), we did not detect any specific localization within late-stage cysts (Fig. 2D-D″). By contrast, D-WIP^testis-GFP displayed a prominent and specific localization within mature head cyst cells (Fig. 2E-E″), similar to the observations made using anti-D-WIP antibodies. Visualization of D-WIP^testis-GFP, which improved the subcellular resolution of the localization pattern, suggested that the protein localizes to the region of the head cyst cell in which the spermatid nuclei are embedded (Fig. 2F-F″). The localization pattern of D-WIP^testis suggests a discrete site within head cyst cells at which D-WIP and Wsp may interact to enable the final phases of spermatogenesis. We note, however, that attempts to rescue the D-WIP male-sterile phenotypes using D-WIP^testis and other isoforms have so far been unsuccessful, possibly owing to complexities related to alternative splicing of D-WIP transcripts or to inaccurate temporal expression patterns of rescuing constructs.

The apparent involvement of both D-WIP and Wsp in sperm maturation raises the possibility that this process employs additional elements of the D-WIP/Wsp machinery that mediates myoblast fusion. Since cell-cell interactions are an essential aspect in both cases, we decided to focus on cell surface proteins as possible common mediating elements. Embryonic myoblast fusion is set in motion following heterotypic interactions between the transmembrane adhesion molecules Dumbfounded (Duf; also known as Kirre) and Sns (Abmayr et al., 2003). To assess whether these elements might be involved in late-stage spermatogenesis, we first used sensitive transcriptional reporters to attempt and detect expression in developing cysts. duf expression was monitored using rp298-GAL4, a GAL4 insertion in the duf locus (Menon and Chia, 2001), to drive UAS-RedStinger (Barolo et al., 2004), a bright nuclear RFP-based reporter. Although a strong signal was observed in the muscle sheath that wraps the testis, no indication of duf expression could be found in other cells of this tissue (Fig. 3A-A″). By contrast, a small population of cells located at the basal end of the testis was found to express sns-GCN-lacZ, a reporter in which an enhancer from the sns gene regulatory region is fused to β-galactosidase (Zhuang et al., 2009). These appeared to be mature head cyst cells, as they also expressed the nuclear cyst cell marker Eya and were closely associated with highly condensed spermatid nuclei (Fig. 3B-B‴).

The sns expression pattern corresponds well to the location and stage at which D-WIP and Wsp mediate sperm maturation. We therefore knocked down sns expression in the cyst cells, using eyaA3-GAL4 and UAS-sns^RNAi, to assess the functional requirement for Sns. Strikingly, knockdown of sns in the cyst cells resulted in the same late-stage mutant phenotypes that we observed for the various components of the Wsp-based machinery (Fig. 3C). The expression and functional assays suggest that Sns contributes to D-WIP/Wsp regulation of spermatogenic cyst maturation, operating together with these elements within somatic head cyst cells.

The phenotypes observed following disruption of the Wsp-based actin polymerization machinery during the final stages of Drosophila spermatogenesis imply a crucial role for microfilament arrays in mediating spermatid release. Consistent with previous reports (Desai et al., 2009) we observed a prominent F-actin structure (Fig. 4A), which, because of its late-stage appearance and positioning at the apical end of coiling cysts, adjacent to the spermatid nuclei, was likely to be involved in the final phases of sperm maturation within the testis. To identify the cell in which this structure is produced, we used the PpY55A-GAL4 cyst cell driver and UAS-Moesin-GFP, which encodes a GFP-tagged microfilament-binding protein (Chihara et al., 2003; Dutta et al., 2002). Moesin-GFP expressed in this fashion fully colocalized with the phalloidin-stained structure (Fig. 4B-B‴), demonstrating that the late-stage microfilament array is formed within the head cyst cell.

Inspection of isolated cysts and sectioned material was employed to examine structural features of the late-stage microfilament array. This approach revealed an apparent temporal progression, during which the array underwent a series of morphological transitions (Fig. 4C-E′). A dense microfilament ‘crown’ that resembled branched/dendritic microfilament networks in organization was observed to form, presumably within the head cyst cell, just beyond the tips of the spermatid nuclei, prior to their full condensation (Fig. 4C,C′). As the spermatid nuclei condensed, the microfilaments appeared to spread over and between them, eventually forming a closely intertwined structure (Fig. 4D,E′). Both lateral views (Fig. 4E,E′) and transverse sections (Fig. 4F,F′) of this structure suggest that the spermatid nuclei come to be encased within actin-lined projections that emanate from the head cyst cell.

Significantly, cyst cell-specific expression of D-WIP^testis-GFP and Wsp-GFP (Massarwa et al., 2007) via the PpY55A-GAL4 driver (Fig. 4G-H‴) demonstrated close spatial overlap of these proteins with the late-stage head cyst cell microfilament arrays.

Prompted by these observations, we sought to assess the involvement of the Wsp-based actin polymerization machinery in the organization of head cyst cell microfilament structures at advanced stages of spermatogenesis (Fig. 5). Visualization of microfilaments and nuclei of late-stage cysts in mutant backgrounds in which D-WIP function was compromised revealed significant structural alterations (Fig. 5B-C″ and see Fig. S3 in the supplementary material). There was a consistent failure to achieve the intertwined and tightly packed organization of head cyst cell F-actin arrays and associated spermatid nuclei. The microfilament arrays appeared sparse and misshapen, and the normally tight and uniformly oriented bundles of nuclei were loosely packed and partially split. Similar phenotypes were also characteristic of late-stage cysts in Wsp^H242D-rescued flies (Fig. 5E-E″) and in flies in which RNAi targeting Cdc42 was specifically expressed in the somatic cells of spermatogenic cysts (Fig. 5F-F″).

The abnormalities in microfilament organization are of particular significance because the genetic and protein localization data reported above imply that the functional defects associated with mutations in Wsp, D-WIP and Cdc42 arise in the head cyst cell. In this context, we find it particularly noteworthy that the robust, transient network of head cyst cell microfilaments (Fig. 4C) was far less frequently observed in Wsp and D-WIP mutant testis preparations as compared with wild type [5% (n=59) versus 25% (n=40)]. Transverse sections through D-WIP (Fig. 5D-D″) and Wsp (not shown) mutant cysts demonstrated, however, that actin-based projections remained properly associated with individual spermatid nuclei at late stages of cyst maturation. Taken together, these observations suggest that the Wsp-based actin polymerization machinery profoundly influences tissue organization during the final stages of cyst maturation via the structuring of specific microfilament arrays within the head cyst cell.

Although the fly and mammalian testes differ substantially in overall structure and organization, spermatogenesis in these systems shares several common underlying principles. Sperm production in both cases is initiated from a stem cell population and proceeds within syncytial units, where the germ cells undergo meiosis, differentiation and individualization in tandem, prior to their release as distinct, mature spermatids. A second key feature is the intimate association of germline cells, throughout differentiation, with specialized somatic cell populations (cyst cells in Drosophila and Sertoli cells in mammals), which provide crucial guidance and support cues (Fuller, 1993; Griswold, 1998). Given the roles that we have identified for the Drosophila Wsp-based pathway in cyst cells, we sought to assess the somatic requirement for this conserved actin polymerization machinery during mammalian (murine) spermatogenesis.

Sertoli cells are the major somatic component of the seminiferous tubules, the site of spermatogenesis within the mammalian testis. Spermatogonial stem cells, which reside at the base of the Sertoli cell epithelium, give rise to syncytia of differentiating spermatids, which transverse the epithelium apically, and release individualized and mature spermatids to the tubule lumen (Fig. 6A-A″). The Sertoli cells communicate and maintain close contact with the spermatogenic lineage throughout this journey. To interfere with the branched actin polymerization pathway in these cells, we chose to disrupt N-WASp, the primary, ubiquitous mammalian WASp isoform (Kurisu and Takenawa, 2009), using a loxP-based conditional knockout (cKO) allele of murine N-WASp (Wasl) (Cotta-de-Almeida et al., 2007) and the Sertoli cell-specific driver desert hedgehog (Dhh)-Cre (Jaegle et al., 2003). We found that male mice in which N-WASp is disrupted in this fashion are viable and reach the age of sexual maturity, but are sterile.

We prepared and examined histological sections of seminiferous tubules to follow the progress of spermatogenesis in these mice as they mature (Fig. 6B,C,E,F,H,I,K,L). Tissue and cell type organization of the testis were properly established in young [post-natal day (P) 6] Dhh-Cre/N-WASp-cKO mice, with spermatogonial stem cells embedded near the basal portion of the circular Sertoli cell epithelium (Fig. 6B,C). At P20, however, the tubules of the mutant mice displayed morphological abnormalities, including an uneven distribution of spermatogonia and spermatocytes and the lack of an internal tubular lumen, which, by contrast, had begun to form in age-matched wild-type mice (Fig. 6E,F). Defective lumen formation and the associated failure to produce mature spermatids were strongly pronounced in older (P35 and P60) Dhh-Cre/N-WASp-cKO mice (Fig. 6H,I,K,L) and were coincident with a marked reduction in differentiating spermatogonia.

Close examination of seminiferous tubules from P35 testes, which normally contain germ cells at all stages of differentiation, suggested that spermatogenesis in Dhh-Cre/N-WASp-cKO testes does not progress beyond the final, round-spermatid stage of meiosis (Fig. 6N,O). In contrast to these germline defects, expression of the Sertoli-specific cytoplasmic marker scleraxis-GFP (Scx-GFP) (Pryce et al., 2007) did not reveal any overt morphological abnormalities or signs of degeneration in the Sertoli cell epithelium of either young or mature Dhh-Cre/N-WASp-cKO mice (Fig. 6D,D′,G,G′,J,J′,M,M′). Arrest of spermatogenesis in Dhh-Cre/N-WASp-cKO mice is therefore likely to be a consequence of specific defects in Sertoli cell communication with the differentiating spermatocytes, rather than a secondary result of abnormal Sertoli cell physiology.

Several novel observations emerge from our study of the involvement of the branched actin nucleation machinery in Drosophila spermatogenesis. A central feature is the requirement for the NPF Wsp at advanced stages, just prior to the release of mature sperm from the testis into the reproductive tract. Our data suggest that Wsp is activated in this particular setting by GTP-bound Cdc42, and, in turn, stimulates the microfilament nucleating capacity of the Arp2/3 complex. Two additional contributing factors are the cell surface protein Sns and the adaptor D-WIP (acting via a testis-specific isoform), which were previously shown to interact with Wsp as mediators of embryonic myoblast fusion. Importantly, both the core actin nucleation machinery (Wsp and Arp2/3) and associated factors (Cdc42, Sns and D-WIP) are required in the somatic cyst cells for execution of the late spermatogenic process leading to spermatid release. The expression and localization patterns of Sns and D-WIP, respectively, further suggest that the coordinated activity of all elements takes place specifically within the head cyst cell, at a late stage of spermatogenesis. Correspondingly, D-WIP, Cdc42 and Wsp contribute to the proper organization of head cyst cell microfilaments, as evidenced by morphological abnormalities following functional disruption of these elements.

We suggest the following tissue-level scenario to account for these findings and link them to the mutant phenotypes (Fig. 7). Attachment of the head cyst cell to the TE present at the base of the testis is a crucial prerequisite for the release of motile sperm from the cyst into the seminal vesicle. Although the mechanism underlying the eventual ‘escape’ of sperm from the testis remains enigmatic, our data suggest that the morphology of the head cyst cell and the proper organization of its internal actin-based cytoskeleton influence the capacity of this cell to associate with the TE. We interpret the failure of mutant cysts to coil as a further indication of this defect in establishing cell-cell contacts: in the absence of a strong head cyst cell-TE association, coiling from the apical end of the cyst is likely to result in ‘pulling away’ at the basal end, accounting for the gap observed between the cysts and the base of the testis tube in all mutants examined.

As we report, prominent microfilament arrays form within the head cyst cell at stages leading up to spermatid release (see also Desai et al., 2009). The presence of these arrays at ‘the right time and place’ and the demonstrated requirement for Wsp machinery elements in their organization strongly suggest that they are key mediators of the crucial cellular associations that govern spermatid release. The identity of the actin structure involved is uncertain, as both a robust, albeit transient, cytoplasmic microfilament network as well as longer-lived actin-lined extensions are generated by the head cyst cell at late stages.

An additional open question is the nature of the molecular scenario underlying Wsp-Arp2/3 function in this particular context. Cdc42 is likely to play an activating role, whereas the basis for the contribution of Sns and D-WIP to this process remains to be elucidated. Models of myoblast fusion (Kim et al., 2007; Massarwa et al., 2007) have suggested that these elements might be part of a complex that determines the cellular site of Wsp-Arp2/3 activity. Although this scenario might apply in the current context as well, our analysis implies some clear distinctions. D-WIP does not appear to be used strictly for Wsp localization (see Fig. S3 in the supplementary material), suggesting a more complex role in mediating Wsp activity. Similarly, Sns would require an alternative binding partner to the myoblast surface molecule Duf, which does not appear to be expressed in spermatogenic cysts and adjacent cells; the Duf homolog Roughest (Bao et al., 2010) could potentially fulfill this role.

Regardless of the molecular details, a key issue is the means by which actin organization influences the relevant cell-cell interactions. One possibility is that head cyst cell microfilaments, possibly via tight association with the compact bundle of nuclei, dictate an overall morphology of the head cyst cell that impacts on its capacity to interact with other cells within the tissue. Alternatively, the well-established involvement of microfilaments in the generation and maintenance of cellular junctions (Hartsock and Nelson, 2008; Mege et al., 2006) might be at play, contributing to the adhesive capacity of the head cyst cell.

As described above, spermatogenesis is arrested in mice following specific disruption in Sertoli cells of N-WASp, which encodes the major mammalian WASp homolog. These findings generalize the significance of branched actin polymerization activity within somatic support cells to the progress of spermatogenesis. Key issues that should now be addressed include determination of the N-WASp-dependent process through which Sertoli cells promote sperm differentiation, the nature of the molecular mechanism involved, and the degree to which parallels may be drawn between the Drosophila and mouse systems.

Microfilament arrays are associated with several unique Sertoli cell structures that mediate different stages of spermatocyte maturation (Vogl et al., 2008). Two of these, the apical ectoplasmic specialization (ES) and the blood-testis barrier (BTB), have recently been recognized as specific localization sites for Arp3, a subunit of the Arp2/3 complex, during key restructuring phases (Lie et al., 2010). The apical ES is a Sertoli cell junctional structure to which spermatids are anchored during the mid and late phases of their maturation, whereas the BTB, which separates the basal and adluminal compartments of the seminiferous epithelium, is crossed by spermatocytes during meiosis (Yan et al., 2007). Arrest of spermatogenesis at the BTB is consistent with the N-WASp mutant phenotypes that we have described. Furthermore, such an arrest would mask a second, later requirement for apical ES construction, which might be more closely related to the role of Wsp during the late stages of Drosophila spermatogenic cyst differentiation. Interestingly, late-stage defects in spermatogenesis leading to partial male sterility have been reported for mice following targeted disruption of CR16 (Wipf3 – Mouse Genome Informatics), a member of the WIP protein family (Suetsugu et al., 2007), expanding the list of conserved elements of the WASp-based machinery that contribute to advanced features of sperm maturation.

In conclusion, our study describes an essential requirement for the WASp-Arp2/3 actin polymerization system during spermatogenesis in both flies and mice. The study highlights a fundamental role of actin nucleation within somatic support cells in orchestrating the intricate relationship between maturing sperm and the surrounding somatic support tissue. Future work should examine the specific cellular functions of WASp-Arp2/3 in each of the systems and explore the possibility of a conserved function, despite the morphological differences.

We thank Susan Abmayr, Elizabeth Chen, Steve DiNardo, George Enders, Therese Eriksson, Dies Meijer, Ruth Palmer, Mar Ruiz-Gomes, Mireille Schäfer and Ronen Schweitzer, as well as the Bloomington, VDRC (Vienna) and DGRC (Kyoto) Drosophila stock centers for providing fly and mouse reagents; R'ada Massarwa for generating the embryonic D-WIP-GFP flies; Moshe Grunspan for images used in Fig. 6; Eli Arama and members of his lab for technical assistance, reagents and advice; the Weizmann Institute Veterinary Resources Department for housing and handling of mice strains; Tamara Berkutzki and Ori Brenner for instruction in preparation and interpretation of histological sections; Julie Brill (Hospital for Sick Children, Toronto) and Sandra Kleinman (Sourasky Medical Center, Tel Aviv) for many helpful discussions; our colleagues Inbal Mor and Eli Arama for critical reading of the manuscript; and all B.-Z.S. lab members for their advice and support. Our research was supported by a grant from the Israel Science Foundation to B.-Z.S. and E.D.S. B.-Z.S. is an incumbent of the Hilda and Cecil Lewis Chair in Molecular Genetics.