Strip and Cka negatively regulate JNK signalling during Drosophila spermatogenesis

Adult stem cells are present in most metazoan tissues in specialised microenvironments termed ‘niches’, where they contribute to the production of differentiated cell populations and are maintained by self-renewal (Li and Xie, 2005). Stem cell identity is maintained by the contribution of molecular signals from the constituent cells of their niche (Li and Xie, 2005). Defective niche signalling can be a contributing factor to tumourigenic stem cell behaviour, leading to unbridled proliferation and self-renewal (Batlle and Clevers, 2017; Cabarcas et al., 2011; Dalerba et al., 2007; Pardal et al., 2003; Reya et al., 2001; White and Lowry, 2014).

One of the earliest identified and now best-studied stem cell niches is the Drosophila testis, wherein two stem cell populations – the germline stem cells (GSCs) and somatic cyst stem cells (CySCs) – beget the process of spermatogenesis (Fuller, 1993; Greenspan et al., 2015; Hardy et al., 1979). Physically, each GSC is enclosed by a pair of CySCs, and all adhere to a population of non-mitotic ‘hub cells’ – these three cell types make up the niche, and the stem cells receive regulatory signals from the hub cells and from each other (Fig. 1A) (Fuller, 1993; Greenspan et al., 2015; Hardy et al., 1979; La Marca and Somers, 2014). Upon dividing asymmetrically, the germline progeny (gonialblasts) exit the niche and undergo four rounds of transit-amplifying (TA) divisions while differentiating, continually enclosed by terminally differentiated CySC progeny: a pair of somatic cyst cells (CyCs) (Fig. 1A) (Fuller, 1993; Greenspan et al., 2015; Hardy et al., 1979). The resultant 16-cell spermatogonial clusters grow and differentiate into spermatocytes, which undergo meiosis and terminal differentiation to produce sperm; however, sperm production does not begin until the white pre-pupal stage, meaning it is not normally observed in third-instar larval (L3) testes (Fig. 1A) (Fuller, 1993; Hardy et al., 1979). The pre-meiotic stages of spermatogenesis are thought to be identical between the developing larval testis/gonad and the adult testis, particularly in terms of the molecular signals regulating stem cell behaviours – these include the Janus kinase-Signal Transducer and Activator of Transcription (Jak-STAT), Transforming Growth Factor β (TGFβ), Hedgehog (Hh), and Epidermal Growth Factor Receptor (EGFR) signalling pathways (Fig. 1A,B) (Matunis et al., 2012; Zoller and Schulz, 2012).

Previous studies using the Drosophila testis as a model system have almost exclusively used the adult testis, particularly when screening for genetic regulators of spermatogenesis (Castrillon et al., 1993; Hackstein, 1991; Liu et al., 2016; Matunis et al., 1997; Schulz et al., 2004; Terry et al., 2006). Although such approaches have been undeniably successful, they preclude the identification of genes with additional fundamental roles in organism development and survival. To address this bias, we conducted a forward genetic screen of L3 male gonads from stocks homozygous lethal at the late-larval or pupal stage, which had been generated via EMS mutagenesis on each of the major Drosophila chromosomes (Dominado et al., 2016; J.E.L.M., unpublished). Here, we characterise one of the alleles identified in our screen – jam packed (jam) – which we demonstrate is an allele of Strip. Strip is a highly conserved core component of the STRIPAK complex (Goudreault et al., 2009; Ribeiro et al., 2010). Although STRIPAK complexes and their components have been implicated in diverse biological roles (Hwang and Pallas, 2014; Kück et al., 2016; Shi et al., 2016), Strip remains relatively poorly understood. We demonstrate that Strip acts within the somatic lineage of the Drosophila L3 male gonad, cell-autonomously regulating somatic cell differentiation and morphology, and non-cell-autonomously regulating germline lineage proliferation and differentiation. Furthermore, within the somatic lineage, Strip acts in concert with the core STRIPAK component Cka, and together they bind to a JNK kinase [Hemipterous (Hep)] and act as negative regulators of JNK signalling. Strip/Cka therefore plays a novel role in regulating the JNK signalling pathway in Drosophila spermatogenesis.

Our forward-genetic screens of late-larval or pupal homozygous lethal mutants searched for those showing defective spermatogenesis in the L3 male gonad (henceforth referred to as ‘gonads’). In our screen of 3rd chromosome mutants, we isolated an allele we termed ‘jam packed’ (jam), after its phenotype (as visualised by DNA staining using DAPI). In wild-type gonads, DAPI-stained DNA appears much brighter in the apical region of the tissue, where the GSCs, gonialblasts, TA spermatogonia, CySCs and CyCs reside (Fig. 1C). By contrast, supernumerary strongly DAPI-positive cells were observed throughout all jam/jam gonads (Fig. 1D), which were also significantly smaller than wild-type gonads (Fig. S1A).

In order to identify the gene affected by the jam allele, deletion mapping was used to identify a chromosomal region that failed to complement jam. A region of lethality was identified between 63D1 and 63D2 on the 3rd chromosome, which contains nine genes, including Striatin interacting protein (Strip) (Fig. S1B). The jam/jam mutant phenotype of supernumerary DAPI-positive cells was strongly recapitulated in gonads from transheterozygous jam/Df(3L)tc-1 animals (n=6/6) (Fig. 1E) (Sakuma et al., 2014), and transheterozygous jam/Df(3L)HR232 animals (Fig. S1C). Animals transheterozygous for jam and Strip^dogi (n=16/16) (Fig. 1F), an established null allele of Strip (Sakuma et al., 2014), also strongly recapitulated the jam/jam phenotype, while those heterozygous for jam appeared wild type (Fig. S1D). These data suggested jam was an allele of Strip.

DNA sequencing of Strip in jam/jam animals revealed a C/T substitution mutation at base pair 5530 (of the entire gene region), which alters a CAG glutamine codon into a TAG premature stop codon (Fig. 1G). This mutation is present in each predictive isoform of the gene: at base pair 2805 in the transcript of isoforms A, at base pair 2370 in the transcript of isoform B, and at base pair 2542 in the transcript of isoform C. Strip contains two conserved protein domains: a putative transmembrane ‘N1221-like domain’ and a ‘Protein of unknown function DUF3402 domain’, in which the jam lesion occurs. Strip has two mammalian orthologues, STRIP1 and STRIP2, and similar truncation mutations in human STRIP2 have been identified in a variety of cancer cell lines (Madsen et al., 2015). Our sequencing data predicted that each Strip isoform should be ∼15 kDa smaller in jam/jam animals than in the wild type, but western blotting did not detect this change (Fig. S2A), nor did it detect any significant difference in Strip protein levels between the wild type and the mutant (Fig. S2B). The 130 kDa band observed matches the previously reported size for wild-type Strip, which was also observed in embryos homozygous for Strip^dogi and Df(3L)tc-1, which are thought to be null for Strip (Sakuma et al., 2014). The observation of this 130 kDa band might be due to aberrant post-translational modifications or the perdurance of maternal Strip in jam/jam L3s. Regardless, these data collectively suggest that jam is an allele of Strip, the disruption of which leads to supernumerary strongly DAPI-positive cells populating the gonads.

Supernumerary strongly DAPI-positive cells in Strip^jam/jam gonads suggested the presence of ectopic numbers of early stage cells. Immunofluorescence experiments were performed to identify the lineage of the supernumerary cells and characterise any disruptions to their behaviour/morphology.

Stem cell character can be rudimentarily denoted by the expression of Escargot (Esg) (Voog et al., 2014). In wild-type gonads (n=14/14), the Esg-GFP reporter marks the hub cells, GSCs, gonialblasts and CySCs (Fig. 2A, arrow). However, in Strip^jam/jam gonads (n=7/7), Esg-GFP-positive cells were observed throughout the gonad, suggesting the expansion of normally niche-restricted cells (Fig. 2B, arrow). In wild-type gonads (n=17/17), Discs large 1 (Dlg1), a junctional protein, is first detectable at the surfaces of both the germline and somatic cell lineages (Fig. 2C, arrow), before later becoming restricted to the CyCs enclosing the spermatocytes (Fig. 2C, arrowhead) (Papagiannouli and Mechler, 2009). Staining for Dlg1 in Strip^jam/jam gonads (n=13/13) revealed they contained mainly small individualised cells (Fig. 2D, arrow), with differentiated cysts observed only rarely (Fig. 2D, arrowhead). We next examined two germline lineage markers: Vasa, which marks the entire germline lineage (Hay et al., 1988); and Hu li tai shao (Hts), which marks the spectrosome: a germline-specific organelle (Lighthouse et al., 2008; Lin et al., 1994). In wild-type gonads (n=30/30), staining for Vasa revealed germline cells arranged according to growth and differentiation status, while staining for Hts showed normal spectrosomes (small and round) in the GSCs and gonialblasts, (Fig. 2E,E′, arrows). In TA spermatogonia, spectrosomes were larger and branched (termed fusomes) (Fig. 2E,E′, arrowheads). By contrast, in Strip^jam/jam gonads (n=27/27), staining for Vasa revealed tissues populated predominantly by small, individualised germline cells containing small, dot-like spectrosomes, rather than branched fusomes (Fig. 2F,F′, arrows). Staining for the hub cell marker Fasciclin 3 (Fas3) did not reveal any obvious differences in hub morphology between the two genotypes (data not shown). Germline differentiation markers were then examined: Bag of marbles (Bam), which marks the TA spermatogonia (Gönczy et al., 1997; Insco et al., 2009; Kawase et al., 2004; Shivdasani and Ingham, 2003); and Matotopetli (Topi), which marks the spermatocytes (Perezgasga et al., 2004). In wild-type gonads (n=31/31), Bam and Topi have a distinct, sequential expression pattern, and clearly mark the differentiating germline (Fig. 2G,G′, arrowheads and arrows). In Strip^jam/jam gonads (n=21/21), Bam expression was seen in germline cells throughout the gonad, while Topi-expressing cells were rarely observed (Fig. 2H,H′, arrowheads and arrows). Bam expression is repressed by TGFβ signalling, the activity of which is observable by staining for phosphorylated (activated) Mothers against dpp (Mad) (Kawase et al., 2004). Germline pMad expression in wild-type gonads was restricted to the GSCs and gonialblasts (Fig. S3A,A′, arrowheads), while in Strip^jam/jam gonads it was observed throughout the germline cells (Fig. S3B,B′, arrowheads). Altogether, these data demonstrate Strip^jam/jam gonads contain supernumerary, largely undifferentiated, germline lineage cells.

We next examined the status of the somatic lineage, first by staining for the CySC/early CyC marker, Zn-finger homeodomain 1 (Zfh1) (Leatherman and DiNardo, 2008), and the germline cell-somatic cell interface marker Armadillo (Arm) (Joti et al., 2011; Sarkar et al., 2007). In wild-type gonads (n=15/15), Zfh1 expression is detectable within the CySCs/early CyCs, as well as in the pigment cells (Fig. 2I, arrowheads), while Arm expression clearly outlined the enclosure of the germline lineage by their somatic cell partners, by marking somatic cell cytoskeletal extensions (Fig. 2I,I′). In contrast, Strip^jam/jam gonads (n=17/17) showed a loss of germline enclosure, concomitant with a reduction in Arm expression, which was only strongly observed around the tissue border (Fig. 2J,J′, arrows). Additionally, we observed Zfh1-positive cells localised to the surface of the gonads, rather than interspersed between the germline cells, which suggests a severe disruption to enclosure of the germline lineage by the somatic lineage in Strip^jam/jam gonads. This lack of Arm staining, and the disruption to somatic cell cytoskeleton organisation it implies, led to our initial investigation of Strip as the gene affected by the jam mutation: Strip, STRIP1 and STRIP2 had previously been reported as cytoskeleton regulators in vitro (Bai et al., 2011).

To further explore somatic lineage disruptions, we examined expression of the CyC marker, Eyes absent (Eya) (Fabrizio et al., 2003). In wild-type gonads (n=21/21), the somatic lineage transitions from Zfh1 to Eya expression, with a small subpopulation of early CyCs expressing both molecules (Fig. 3A-A″, arrows). Additionally, Eya, unlike Zfh1, does not mark pigment cells (Fig. 3A-A″, arrowheads). Strikingly, in Strip^jam/jam gonads (n=28/28), the majority of somatic cells were both Zfh1 and Eya positive (Fig. 3B-B″, arrows), and were clearly localised near the surface of the gonads. Additionally, the morphology of the somatic cells was disrupted; their nuclei appeared rounded, rather than having the typical elongated form. Despite these disruptions, some somatic cells were exclusively Zfh1 or Eya positive, and found at the anterior or posterior of the gonad, respectively, suggesting that some cells differentiate normally. These results are supported by staining for Traffic jam (Tj), which, like Zfh1, marks CySCs/early CyCs, but does not mark pigment cells (Li et al., 2003). In wild-type gonads (n=27/27), Tj and Eya sequentially marked the differentiating somatic lineage, with a small region of overlap (Fig. 3C, arrow). Again, contrastingly, in Strip^jam/jam gonads (n=10/10) the majority of somatic cells were morphologically disrupted and Tj and Eya positive (Fig. 3D, arrows). The Zfh1-/Eya-positive cells at the posterior of each gonad are the somatic terminal epithelial cells (Fig. 3A-D, double arrowheads), which are not thought to contribute to the early stages of spermatogenesis (Fuller, 1993). Finally, we observed strong expression of pMad in the somatic cells pushed to the surface of Strip^jam/jam gonads (Fig. S3C,C′, arrows), where it is normally restricted to CyCs in the wild type (Fig. S3A,A′, arrows). These data demonstrate that the differentiation and morphology of somatic lineage cells in Strip^jam/jam gonads is disrupted.

Altogether, these results suggest Strip^jam/jam L3 gonads are primarily populated by supernumerary, largely undifferentiated, germline lineage cells. Additionally, somatic lineage differentiation and morphology is disrupted, and germline enclosure by the somatic lineage is disrupted.

The presence of supernumerary germline cells in Strip^jam/jam gonads suggested that some cells might be overproliferating. To investigate, we pulse-labelled dissected gonads with BrdU to detect cells undergoing DNA replication. In wild-type gonads (n=22/22), BrdU incorporation was consistently observed only in those cells in the apical region of the gonad (Fig. 4A), whereas in Strip^jam/jam gonads (n=31/32), we observed BrdU-positive cells throughout the tissue (Fig. 4B). Staining for the M-phase marker phosphorylated Histone H3 (pHis3), gave similar results: in wild-type gonads (n=39/39), individualised pHis3-positive cells (indicative of dividing GSCs or gonialblasts) (Fig. 4C, arrow), as opposed to clusters of pHis3-positive cells (indicative of dividing spermatogonia) (Fig. 4C, arrowhead), were only ever observed within the apical region. By contrast, in Strip^jam/jam gonads (n=50/50), individualised pHis3-positive cells were observed throughout the tissue. These data support the hypothesis of overproliferation occurring in Strip^jam/jam gonads.

To precisely identify the overproliferating cells, we undertook co-staining for pHis3, Zfh1 and Vasa, which, respectively, mark the proliferating cells, the somatic lineage and the germline lineage. By quantifying each pHis3-positive cell and its lineage, statistical analyses revealed that the difference in the average number (±s.e.m.) of pHis3-/Zfh1-positive cells between wild-type (0.26±0.080) and Strip^jam/jam (0.28±0.076) gonads is negligible (Student's t-test, P>0.05) (Fig. 4E). Antithetically, the average number of cells that are pHis3 and Vasa positive in wild-type (0.97±0.17) and Strip^jam/jam (15.12±1.12) gonads is significantly different (Student's t-test, P<0.001) (Fig. 4E), indicating the germline lineage is overproliferating in Strip^jam/jam gonads. We next sought to dissect the composition of the germline lineage. Quantification of the different stages of proliferating germline cells revealed that, in wild-type gonads, the percentages of cells proliferating individually (28.95%) or as clusters of two cells (21.05%), four cells (23.68%) and eight cells (23.68%) are roughly equal, while a lower percentage of cells are proliferating in 16-cell clusters (2.64%) (Fig. 4F). By contrast, in Strip^jam/jam gonads, the percentage of germline cells proliferating individually (84.43%) is conspicuously larger than the percentage of those proliferating as two (12.53%), four (2.38%), eight (0.40%) and 16 (0.26%) cell clusters (Fig. 4F). Taken together, these data demonstrate that the majority of germline cells in Strip^jam/jam gonads are individualised cells undergoing ectopic proliferation, while also failing to undergo TA divisions or differentiate.

Ectopic proliferation could be expected to lead to larger than normal gonads due to increased cell numbers. As Strip^jam/jam gonads were, without exception, smaller than wild-type gonads, we hypothesised that cell death was elevated in Strip^jam/jam gonads. Accordingly, we marked dying cells using an antibody against human cleaved caspase 3, which indicates the activity of the apoptosis protein Death regulator Nedd2-like caspase (Dronc) in Drosophila (Fan and Bergmann, 2010). In wild-type gonads (n=15/15), we saw no Dronc/cleaved Caspase 3 expression (Fig. 4G), which strongly contrasts with the large number of Dronc/cleaved Caspase 3-positive cells observed in Strip^jam/jam gonads (n=14/14) (Fig. 4H). TUNEL assays gave similar data, demonstrating negligible apoptosis levels in wild-type gonads (n=5/5) (Fig. 4I, arrow), but large numbers of apoptotic cells in Strip^jam/jam gonads (n=10/10) (Fig. 4J, arrows). These data suggest that the reduced size of Strip^jam/jam gonads might be due to increased cell death.

In order to identify the lineage(s) in which Strip is necessary, as well as confirm the identity of jam as a Strip allele, we utilised the GAL4/UAS system to induce cell-specific transgene expression. Using the C587-GAL4 driver, which induces expression in the early somatic cells (Demarco et al., 2014), and co-expressing Dicer-2 (Dcr-2), a known positive regulator of the RNA interference (RNAi) process (Kim et al., 2006), we found that the knockdown of Strip via RNAi lines v16211 (n=14/21 at 25°C, n=9/10 at 29°C) (Fig. 5A,A′) and v16212 (n=12/14 at 25°C, n=12/12 at 29°C) (Fig. 5B,B′) strongly recapitulated the Strip^jam/jam gonad phenotypes. Knockdown of Strip in either the hub or germline lineages (Fig. S4A-B′), using the drivers unpaired 1-GAL4 (upd1-GAL4) and nanos-GAL4;UAS-GAL4 (hereafter nos-GAL4), respectively, did not yield any aberrant phenotypes. The somatic lineage specificity of Strip was further confirmed by replicating the Strip^jam/jam-like phenotype using traffic jam-GAL4 (tj-GAL4), another early somatic cell-specific driver, and RNAi line v16212 against Strip (Fig. S4E,E′).

We generated transgenic constructs containing wild-type Strip, under UAS control, tagged with a YFP variant: Venus (Nagai et al., 2002). Two independent insertion lines of this transgene were examined: UAS-Strip-YFP#1 and UAS-Strip-YFP#2. Somatic cell expression of UAS-Strip-YFP#1 (n=22/22) in a wild-type background had no obvious effect on spermatogenesis (Fig. 5C,C′,E,E′). Significantly, somatic cell expression of UAS-Strip-YFP#1 in a Strip^jam/jam background (n=37/37) almost completely rescued the mutant phenotype (Fig. 5D,D′,F,F′), with germline differentiation and somatic cell morphology/differentiation appearing restored. Lethality was not rescued, likely owing to limited expression of the C587-GAL4 driver. Similar results were obtained after somatic cell expression of UAS-Strip-YFP#2 in both wild-type and Strip^jam/jam backgrounds (Fig. S5A-D′). Additionally, Strip-YFP expression in the somatic lineage rescued gonad size, with line #1 the more effective (Fig. S5E). Germline divisions were not rescued completely after Strip-YFP expression – TA spermatogonia and spermatocytes appeared to contain two cells at most, rather than 16 (Fig. 5D,D′, arrows). This may be due to Strip-YFP lacking some functionality, insufficient expression levels for Strip-YFP, dominant-negative behaviour of Strip^jam or some minor germline lineage role for Strip. Note, however, that expression of Strip-YFP#1 in the germline using nos-GAL4 had no discernible effect on the wild-type or Strip^jam/jam phenotypes (Fig. S5F-G′). Collectively, the identification of the molecular lesion in Strip^jam by DNA sequencing, the phenocopying of the Strip^jam/jam gonad phenotypes in transheterozygous and Strip RNAi knockdown larvae and the partial rescue of Strip^jam/jam mutant phenotypes by expression of Strip-YFP all contribute to the confirmation of jam as a Strip allele. Additionally, RNAi and Strip-YFP transgene data indicate that Strip acts in the somatic lineage to autonomously regulate somatic cell differentiation and morphology, and non-cell-autonomously to regulate germline differentiation.

As we were unable to detect Strip in vivo with any available antibodies, we employed clustered regularly interspaced short palindromic repeats/CRISPR associated protein 9 (CRISPR/Cas9) gene-editing technology to tag endogenous Strip (Fig. S6A-E). We generated multiple homozygous viable stocks containing a 3×Myc tag sequence at the 3′ end of the Strip CDS, which is shared across each isoform. Immunofluorescence against the Myc tag epitope showed only background staining in wild-type gonads (n=13/13) (Fig. 5G,G′), but gonads from the CRISPR/Cas9-generated Strip-Myc^16012G stock (n=22/22) revealed cytoplasmic Strip-Myc expression in each gonad lineage (Fig. 5H,H′). Patches of stronger expression were noticeable in the spermatogonia, localised next to their nuclei (Fig. 5H,H′, arrows), possibly indicative of localisation to the endoplasmic reticulum, where orthologues of Strip localise in both yeasts and Caenorhabditis elegans (Frost et al., 2012; Maheshwari et al., 2016). Strip-Myc was also observed in the cytoplasm of the CySCs and CyCs, but, owing to their extremely narrow morphologies, more-specific localisation was not able to be discerned. Notably, these findings were replicated when using Strip-Myc^2-38-4-1 lines concurrently generated by Takahiro Chihara (Fig. S6F-G″) (Sakuma et al., 2016).

Strip is a core member of the highly conserved, but poorly understood, STRIPAK complex – in Drosophila, the STRIPAK complex has an array of dynamic subunits, including phosphatase components [Protein phosphatase 2A at 29B (Pp2A-29B) and Microtubule star (Mts)], kinase components [Germinal centre kinase III (GckIII), Misshapen (Msn)], structural subunits [Cka, Strip, MOB kinase activator 4 (Mob4)] and other subunits with less-well defined roles [e.g. Fibroblast growth factor receptor 1 oncogene partner 2 (Fgop2), Cerebral cavernous malformation 3 (Ccm3), Sarcolemma associated protein (Slmap)] (Fig. 6A) (Ashton-Beaucage et al., 2014; Guruharsha et al., 2011; Hwang and Pallas, 2014; Ribeiro et al., 2010). At the core of the complex is Cka, an orthologue of the human striatin proteins (STR, STR3, STR4). Therefore, we questioned whether knockdown of Cka via RNAi might replicate the phenotypes seen upon Strip knockdown/mutation. Knockdown of Cka in the somatic lineage via RNAi lines v35234 (n=10/10) (Fig. 6B,B′) and v106971 (n=9/22 at 25°C, n=5/7 at 29°C) (Fig. 6C,C′), while co-expressing Dcr-2, successfully recapitulated the Strip knockdown/mutation gonad phenotypes. Knockdown of Cka in the hub or germline lineages had, like Strip knockdown, no effect on the gonad phenotype (Fig. S4C-D′). However, somatic cell-specific knockdown of the STRIPAK complex members Fgop2, GckIII, Mob4, msn and Slmap, via RNAi, had no effect on spermatogenesis (Fig. S7A-F′), even with Dcr-2 co-expression. Owing to this negative result, we confirmed the efficacy of each RNAi via quantitative reverse transcription PCR (RT-qPCR) (Fig. S7G). RNAi lines against Ccm3 (v106841) and mts (v35171) were also tested in vivo and via RT-qPCR, but were found to be non-functional (data not shown). An exception was the case of somatic cell knockdown of Pp2A-29B, which induced a loss of early germline cells (Fig. S8A-B′), a phenotype distinct from that observed for Strip and Cka knockdown.

We next explored whether Strip and Cka interacted physically in L3 tissues via co-immunoprecipitation experiments. In wild-type L3 tissue extracts, we demonstrated that Strip and Cka both immunoprecipitated when baited with antibodies against the other (Fig. 6D,E, lanes 3,4). In Strip^jam/jam L3 tissue extracts, although Strip and Cka were detected following immunoprecipitation with their respective antibodies, the Strip-Cka physical interaction was not observed (Fig. 6D,E, lanes 7,8). These data suggest that Strip and Cka physically interact in the wild-type, but the Strip^jam mutation precludes this interaction. Potentially, the Strip^jam/jam mutant gonad phenotypes derive from this defective interaction. These data suggest that physical interactions between Strip and Cka contribute to proper somatic cell differentiation, morphology and behaviour.

Strip, Cka, and other STRIPAK components have been shown to act as positive regulators of EGFR signalling in the Drosophila wing and eye, and in vitro (Ashton-Beaucage et al., 2014; Friedman and Perrimon, 2006; Horn et al., 2011; Wassarman et al., 1996). Additionally, EGFR signalling is an established regulator of somatic cell enclosure and morphology, and of germline differentiation during spermatogenesis (Fig. S9A) – indeed, disrupting Strip, Cka, or EGFR signalling pathway components yielded strikingly similar gonad phenotypes (Chen et al., 2013; Hudson et al., 2013; Kiger et al., 2000; Sarkar et al., 2007; Schulz et al., 2002; Tran et al., 2000). Therefore, we wanted to explore whether Strip interacted genetically with the EGFR signalling pathway. We first examined EGFR pathway activity, as measured by phosphorylated Rl (also known as ERK/MAPK) abundance, which appeared similar between the wild-type and Strip^jam/jam gonads (Fig. S9B-C′). We next examined whether EGFR signalling suppression phenotypes could be worsened by the presence of Strip^jam (Fig. S9D-E′), and also whether Strip^jam/jam gonad phenotypes could be rescued by EGFR pathway activation (Fig. S9F-I′), but observed no genetic interactions. These data suggested that, rather than interacting with EGFR signalling, a pathway with an established role in spermatogenesis that Strip is known to regulate, Strip has some alternate mode of action.

Cka was originally identified as a positive regulator of c-Jun N-terminal kinase (JNK) signalling, which mediates the physical interactions between Hep, a JNKK, Basket (Bsk), the sole Drosophila JNK, and the AP-1 transcription factor complex, which consists of Jun-related antigen (Jra) and Kayak (Kay) (Fig. 7A) (Chen et al., 2002). The JNK signalling core of Hep and Bsk can be activated by a number of pathways, one of which is the immune deficiency (IMD) signalling pathway (Fig. 7A) (Hoffmann, 2003; Kaneko and Silverman, 2005). Interestingly, it has been demonstrated that Strip, Cka and other STRIPAK components act as negative regulators of JNK signalling when the pathway is activated by IMD signalling after peptidoglycan administration (Ashton-Beaucage et al., 2014; Bond and Foley, 2009). We hypothesised, therefore, that Strip and Cka might be acting as negative regulators of JNK signalling during spermatogenesis.

We first examined staining for phosphorylated JNK (pJNK), which marks JNK signalling activation. Although low in wild-type gonads (n=8/8) (Fig. 7B), pJNK levels were clearly elevated in Strip^jam/jam gonads (n=11/11), particularly in the somatic cells at the gonad surface (Fig. 7C, arrows). Consistent with this were the levels of Matrix metalloproteinase 1 (Mmp1), an established target of JNK signalling (Uhlirova and Bohmann, 2006), which was restricted to the somatic cells and only weakly expressed in somatic cells in wild-type gonads (Fig. S10A, arrow), but strongly expressed in Strip^jam/jam gonads (Fig. S10B, arrow). These data indicate that JNK signalling is elevated specifically in the somatic lineage in Strip^jam/jam animals.

As Cka can act as a scaffold for the physical interactions of Hep, Bsk and the AP-1 complex (Chen et al., 2002), we examined whether Cka and Strip interact physically with Hep in wild-type and Strip^jam/jam L3 tissues via co-immunoprecipitation. We observed that in L3 tissue extracts from both genotypes, Hep immunoprecipitated when baited with antibodies against either Cka or Strip (Fig. S10C,D, lanes 3,4,7,8). This suggests that Hep physically interacts with Cka and Strip in both wild-type and Strip^jam/jam tissues, despite our observation that Strip^jam precludes Cka and Strip themselves from physically interacting (Fig. 6D,E).

We next examined whether reducing JNK signalling via knockdown of Bsk might partially rescue the Strip^jam/jam gonad phenotype. In a wild-type background (n=9/9), expression of RNAi against bsk in the somatic lineage did not modify the L3 gonad phenotype (Fig. 7D,D′). By contrast, somatic cell expression of RNAi against bsk in a Strip^jam/jam background (n=15/15) led to a partial rescue of the mutant phenotype, with proper germline differentiation and somatic cell enclosure clearly visible (Fig. 7E,E′). These results were replicated upon somatic lineage expression of a dominant-negative bsk^DN transgene: in a wild-type background bsk^DN had no discernible effect (Fig. S10E,E′), whereas in a Strip^jam/jam background bsk^DN partially rescued the phenotype (Fig. S10F,F′). Furthermore, somatic lineage expression of bsk^DN rescued the altered expression patterns of Bam and Topi observed in the mutant gonads (Fig. S10G,H), and both methods of bsk disruption significantly increased the size of Strip^jam/jam gonads (Fig. S10I). Furthermore, we examined whether knockdown of bsk might rescue phenotypes induced by Cka knockdown. As before, in a wild-type background, somatic cell co-expression of Dcr-2 and RNAi against Cka (v35234, n=6/7 and v106971, n=10/15) phenocopied the Strip^jam/jam gonad phenotype (Fig. 7F,F′). Knockdown of Cka together with expression of the bsk^DN transgene (v35234, n=10/13 and v106971, n=13/16) led to a striking partial rescue of the germline differentiation and somatic cell enclosure defects of Cka knockdown (Fig. 7G,G′). Collectively, these data indicate that Strip and Cka mutant/knockdown phenotypes are rescued by reducing JNK signalling activity, suggesting that Strip and Cka negatively regulate JNK signalling in the Drosophila gonad.

Previously, Strip and Cka have been shown to negatively regulate JNK signalling initiated via IMD signalling (Ashton-Beaucage et al., 2014; Bond and Foley, 2009). IMD signalling proceeds via the receptor PGRP-LC, which activates Immune deficiency (Imd) and, in turn, activates the kinase complex containing Misshapen (Msn, a JNKKKK), TGFβ activated kinase 1 (Tak1, a JNKKK) and their binding partners: TAK1-associated binding protein 2 (Tab2), TNF-receptor-associated factor 4 (Traf4) and TNF-receptor-associated factor 6 (Traf6) (Fig. 7A). Somatic cell expression of an RNAi line against imd was unable to rescue the Strip^jam/jam gonad phenotype (Fig. S11A-B′), with the RNAi efficacy confirmed via RT-qPCR (Fig. S11C), suggesting Imd activation is not the source of the JNK signalling. However, as alternative activators of Tak1 exist, we examined it directly. In a wild-type background (n=14/14), expression of RNAi against Tak1 in the somatic lineage did not alter the gonad phenotype (Fig. 7H,H′), but in a Strip^jam/jam background (n=22/22), we again observed a partial rescue of the mutant phenotypes (Fig. 7I,I′). Similarly, somatic cell knockdown of Traf6 led to a partial rescue (Fig. S10J-K′).

The Tumour Necrosis Factor (TNF) signalling pathway uses Tak1 to activate JNK signalling, and is initiated by the ligand, Eiger (Egr), binding to either of its receptors: Grindelwald (Grnd) and Wengen (Wgn) (Fig. 7A) (Andersen et al., 2015; Igaki et al., 2002; Igaki and Miura, 2014; Vidal, 2010). To explore a role for TNF signalling, two mutant alleles of egr were used: egr¹ and egr³. In an otherwise wild-type background, egr^1/1 (n=12/12) (Fig. 7J,J′) and egr^3/3 (Fig. S10L,L′) gonads appeared normal, whereas in a Strip^jam/jam background the mutant gonad phenotypes were partially rescued by the presence of egr^1/1 (n=15/15) (Fig. 7K,K′) and egr^3/3 (Fig. S10M,M′). Therefore, we hypothesised that Egr expression might be elevated in Strip^jam/jam gonads relative to the wild type. In wild-type gonads (n=59/59), only low levels of CyC-specific Egr expression was observed (Fig. 7L,L′, arrows). By contrast, in Strip^jam/jam gonads (n=58/58), elevated Egr expression was observed in the somatic cells at the gonad surface (Fig. 7M,M′, arrows), and occasionally in the cytoplasm of the supernumerary germline cells (Fig. S12A,A′, arrows). Egr can be secreted by haemocyte cells identifiable by Nimrod C1 (NimC1) expression (Fogarty et al., 2016), although no NimC1-positive cells were present in either wild-type or Strip^jam/jam gonads (data not shown), suggesting Egr is secreted from another lineage, or acts autocrinically within the somatic lineage. Interestingly, lineage-specific expression of egr in the somatic (Fig. S12B,B′), germline (Fig. S12C,C′) or hub (Fig. S12D,D′) cells did not alter the gonad phenotypes. This suggests overexpression of Egr might not be sufficient to recapitulate the Strip^jam/jam mutant phenotypes, or that in a wild-type background, functional Strip might suppress any elevated Egr activity. Finally, we examined Egr expression in gonads expressing bsk^DN in somatic cells: in an otherwise wild-type background, Egr expression appeared normal (Fig. S12E,E′), whereas in a Strip^jam/jam background, where the mutant phenotype was partially rescued, the ectopic Egr expression seen in non-rescued mutant gonads was absent (Fig. S12F,F′). Collectively, these data indicate that TNF-JNK signalling driven by Egr is upregulated in Strip^jam/jam gonads, and that blocking JNK signalling both rescues the mutant phenotype and reduces expression of Egr.

In this study, we have identified jam, a novel allele of the poorly understood gene Strip, and have characterised the role of Strip in the Drosophila L3 male gonad. We found that Strip acts in the somatic lineage of the gonad, where it cell-autonomously regulates the differentiation and morphology of the somatic cells, and their ability to encapsulate cells of the germline lineage. Non-cell-autonomously, Strip regulates the proliferation and differentiation of the germline lineage. Mechanistically, Strip acts with Cka to physically interact with Hep (JNKK) and negatively regulate TNF-JNK signalling in the somatic cells during spermatogenesis.

Previous studies, largely performed in vitro, have demonstrated JNK signalling is regulated both positively and negatively by STRIPAK complex members (Ashton-Beaucage et al., 2014; Bond and Foley, 2009; Chen et al., 2002). However, Strip and its orthologues also act in a diversity of roles distinct from JNK signalling, both in isolation and with members of the STRIPAK complex. For example, in yeasts, the FAR complex (STRIPAK complex equivalent) has been implicated in cell-cycle arrest in response to pheromones (Kemp and Sprague, 2003), while the Strip orthologue FAR11 regulates Caspase-10-induced cell death (Lisa-Santamaría et al., 2012) and inhibits TORC2 signalling (Pracheil and Liu, 2013; Pracheil et al., 2012). In mammals, STRIP1 and STRIP2 have roles in cytoskeletal organisation (Bai et al., 2011; Rohn et al., 2011), Hippo signalling regulation and cell migration (Madsen et al., 2015), and embryogenesis (Bazzi et al., 2017). Drosophila roles for Strip (variously with and without STRIPAK complex members) are equally diverse, with roles in cytoskeletal organisation (Bai et al., 2011), endocytosis (Sakuma et al., 2014), Ras-MAPK activation (Ashton-Beaucage et al., 2014), microtubule organisation (Sakuma et al., 2015), circadian rhythm (Andreazza et al., 2015), border cell migration (Madsen et al., 2015), Hippo signalling (Ribeiro et al., 2010; Sakuma et al., 2016) and autophagosome transport (Neisch et al., 2017).

This study represents the first in-depth exploration of Strip, Cka and TNF-JNK signalling during Drosophila spermatogenesis, and opens new avenues of investigation for these genes and pathways in other systems. However, roles for Strip and Cka have recently been identified in both the Drosophila adult testis and ovary. In the adult testis, a genome-wide RNAi screen for GSC regulators conducted using the ubiquitous Actin-GAL4 driver revealed that Cka knockdown led to ectopic GSC-like cells in the adult testis, whereas Strip knockdown resulted in GSC loss (Liu et al., 2016); these results both reflect and contrast our own findings. In another study, researchers identified Strip in an RNAi screen for somatic lineage regulators of adult fertility, with Strip knockdown leading to total somatic cell loss (Fairchild et al., 2017). Various explanations exist for these phenotypic discrepancies: different driver choices, different RNAi lines or possibly subtle biological differences exist between the adult testis and larval gonad. Finally, in the adult ovary, Strip and Cka were isolated in an RNAi screen as being required for proper border cell migration (BCM) (Madsen et al., 2015), a process commonly used to model invasive cell behaviours (Montell, 2003; Montell et al., 2012). As there is not thought to be a spermatogenesis process analogous to BCM, it is unlikely that parallels can be drawn between these data and our own, but it is interesting to note that JNK signalling plays a poorly understood role in BCM (Llense and Martín-Blanco, 2008; Mathieu et al., 2007).

The highly conserved JNK signalling pathway has two broad roles: effecting stress response mechanisms or regulating cell shape changes. In either case, the pathway's kinase core of Hep and Bsk is retained, while upstream components can vary greatly (Ríos-Barrera and Riesgo-Escovar, 2013). We demonstrated that Strip and Cka negatively regulate the JNK signalling pathway, something previously demonstrated in the context of IMD signalling (Ashton-Beaucage et al., 2014; Bond and Foley, 2009). However, imd proved dispensable with regard to the activation of JNK signalling during spermatogenesis, while Tak1, Traf6 and the Drosophila TNF egr are all necessary. This strongly suggests that JNK signalling is activated in the L3 male gonad via TNF signalling. However, only a low level of JNK signalling is observed in wild-type gonads – why, then, is the pathway strongly activated in Strip^jam/jam gonads? Previous studies, along with this work, suggest that the ability of Strip/Cka to suppress JNK signalling likely proceeds via physical interactions with Hep and/or Bsk (Ashton-Beaucage et al., 2014; Bond and Foley, 2009; Chen et al., 2002). We propose that JNK signalling during spermatogenesis opposes proper somatic cell morphology and germline enclosure, and that JNK signalling is therefore normally downregulated to the low level we observed by the actions of Strip/Cka (Fig. 8, orange line). Therefore, activation of JNK signalling upon Strip/Cka mutation/knockdown leads to a concomitant disruption of proper somatic cell development.

As we determined Strip and Cka are necessary exclusively within the somatic lineage, why does defective germline differentiation and proliferation occur in Strip^jam/jam gonads? Disruption of EGFR signalling in the testis prevents proper germline enclosure, and results in germline overproliferation phenotypes, similar to our own observations (Chen et al., 2013; Sarkar et al., 2007; Schulz et al., 2002). Furthermore, the somatic lineage is an important source of molecular signals for the germline lineage, and these can regulate behaviours such as differentiation (Leatherman and DiNardo, 2010; Singh et al., 2006; Wang et al., 2008). Thus, the germline phenotypes observed in Strip^jam/jam gonads – ectopic Bam expression, supernumerary early cells, a lack of differentiation and ectopic proliferation – are likely to be secondary phenotypes, deriving from the defective somatic cell behaviours in Strip mutant gonads, and the resultant lack of proper cell-cell contact and enclosure of the germline lineage.

JNK signalling in spermatogenesis was essentially unstudied until recently, when it was found that knockdown of the endocytosis component, Rab5, in the adult testis somatic lineage, led to ectopic spermatogonial TA divisions, a process mediated by JNK and TGFβ signalling (Tang et al., 2017). Interestingly, Strip plays a role in endocytosis during Drosophila neurogenesis, contributing to early endosome organisation, clustering and fusion, processes in which Rab5 is integral (Sakuma et al., 2014). Exploring this link, we found that whilst endocytic components Rab5 (early endosomes) and Rab11 (recycling endosomes) were required in the gonad (Fig. S13A-B′), expression of constitutively active Rab5 did not rescue Strip^jam/jam gonad phenotypes (Fig. S13C-D′), in contrast to its rescue of Strip mutant olfactory projection neuron clones (Sakuma et al., 2014). Additionally, Cka is dispensable for neuronal morphogenesis (Sakuma et al., 2014), but crucial in spermatogenesis, as we have shown. Interestingly, Rab11 knockdown leads to phenotypes highly reminiscent of Strip knockdown (Fig. S14A,A′) (Joti et al., 2011), but Rab5 knockdown leads to ectopically TA spermatogonia (Fig. S14B,B′) (Tang et al., 2017). A more in-depth exploration of potential links between Strip, JNK signalling and endocytosis during spermatogenesis would be informative, particularly given the established links between JNK signalling and endocytosis (Froldi et al., 2010; Igaki et al., 2009; Robinson and Moberg, 2011).

An interesting question to arise from our research is: where does the Egr signal originate? TNF ligands in mammals are believed to be membrane-bound proteins, which can be cleaved, solubilised and then act in an autocrine or paracrine manner (Locksley et al., 2001; Smith et al., 1994). In Drosophila, it is unknown whether Egr behaves similarly or what the upstream regulators of its expression are, and the best-understood situation whereupon Egr expression is induced is after cell polarity loss (Igaki and Miura, 2014; Richardson and Portela, 2018). There, JNK signalling is activated in polarity-deficient cells via Egr secretion from various neighbouring cell types, and the mutant cells are eliminated via apoptosis. As we observed elevated cell death in Strip^jam/jam gonads, a similar situation may be occurring – the disruptions to somatic cell morphology might lead to TNF-JNK-mediated cell death. This is an alternative to our previously suggested model, and places somatic cell morphology upstream of TNF-JNK signalling (Fig. 8, blue line). However, we observed no rescue of the Strip^jam/jam mutant phenotype after blocking effector caspase activity via expression of p35 in either the somatic or germline lineages (data not shown), suggesting JNK signalling plays a different role. JNK signalling has long been established as a cytoskeleton and cell shape regulator (Ríos-Barrera and Riesgo-Escovar, 2013; Rudrapatna et al., 2013), but it is not clear that this activity involves pathway activation via TNF signalling. Another possible mechanism by which Egr expression could be induced in Strip mutant gonads is by TGFβ signalling – during spermatogenesis, TGFβ signalling is induced by JNK activation (Tang et al., 2017). Although it has long been known that dpp, a TGFβ signalling ligand, is a transcriptional target of JNK (Gavin-Smyth et al., 2013; Glise and Noselli, 1997; Hou et al., 1997; Mihaly et al., 2001; Riesgo-Escovar and Hafen, 1997; Takatsu et al., 2000; Zeitlinger et al., 1997), it has only recently been demonstrated that one of the targets of the TGFβ transcription factor, Mad, is egr (Dominguez et al., 2016; Gavin-Smyth et al., 2013), making TGFβ signalling an excellent candidate for explaining the source of the Egr expression observed in Strip^jam/jam gonads in the form of a feedback loop. However, somatic lineage expression of RNAi against Mad [the efficacy of which was confirmed by pMad expression knockdown (data not shown)] did not rescue the Strip^jam/jam phenotype (Fig. S14A-B′), suggesting that the upregulated TGFβ signalling we observed in Strip^jam/jam gonads does not contribute to inducing egr expression. The possibility also remains that activated JNK signalling is itself driving Egr expression (Fig. 8, red line) or perhaps some unexamined pathway is involved (Fig. 8, purple line).

Finally, we considered a role for Jak-STAT signalling, a key regulator of spermatogenesis (Bausek, 2013), and a well-known target of JNK signalling, and which induces Unpaired-family expression in other tissues (Ammeux et al., 2016; Bunker et al., 2015; Jiang et al., 2009). We observed that Jak-STAT signalling was ectopically activated in Strip^jam/jam gonads, as revealed by immunofluorescence for Stat92E (Fig. S14C-D′). Stat92E reporter co-expression with elevated levels of the JNK target Mmp1 was also observed in the somatic lineage (Fig. S14E-H″). However, somatic lineage knockdown of Stat92E did not rescue the Strip^jam/jam phenotype, and only led to early germline cell loss (Fig. S14I-J′), suggesting that Jak-STAT signalling contributes to the established role of the somatic lineage in non-cell-autonomous regulation of germline lineage survival or proliferation (reviewed by Zoller and Schulz, 2012). Thus, although we were not able to dissect the precise upstream effectors of Egr expression, TNF-JNK signalling and how Strip^jam/jam non-cell autonomously affects the germline lineage, our data collectively show the Strip^jam/jam phenotype is highly complex, simultaneously displaying: activation of TNF-JNK signalling, activated TGFβ and Jak-STAT signalling, a loss-of-enclosure phenotype canonically indicative of – but seemingly not linked to – EGFR signalling disruption and increased proliferation alongside increased cell death. It seems likely that the Strip^jam/jam phenotype is a combination of these different disruptions, which also potentially interact with or even effect each other.

Our study has identified novel roles for Strip and Cka in regulating TNF-JNK signalling during spermatogenesis. Furthermore, intriguing parallels exist between the roles for Strip, Cka, and TNF-JNK signalling in spermatogenesis, and their roles in other processes. Given the emerging relevance of the orthologues of Strip in human cancer (Madsen et al., 2015), further research into Strip using the powerful model systems available in Drosophila has the potential to provide important mechanistic insights into pathologically relevant roles of this gene.

The Bloomington Drosophila Stock Center (BDSC) 3rd chromosome deficiency kit, Df(3L)1227 (5410), Df(3L)HR232 (3648), Df(3L)Exel6094 (7573), Df(3L)Exel6095 (7574), Df(3L)Exel6096 (7575), Df(3L)Exel6097 (7576), Df(3L)BSC129 (9294), UAS-rl^SEM (59006), UAS-bsk^DN (9311), UAS-luciferase RNAi (31603), UAS-Rab5^Q88L YFP (9774), UAS-p35 (5072) and w¹¹¹⁸ (3605, used as wild type) were all obtained from the BDSC. The UAS-bsk RNAi (v34138), UAS-Ccm3 RNAi (v106841), UAS-Cka RNAi (v106971), UAS-Cka RNAi (v35234), UAS-Fgop2 RNAi (v47389), UAS-GckIII RNAi (v49558), UAS-GckIII RNAi (v107158), UAS-imd RNAi (v101834), UAS-Mad RNAi (v12635), UAS-Mob4 RNAi (v110742), UAS-msn RNAi (v101517), UAS-mts RNAi (v35171), UAS-Pp2A-29B RNAi (v49671), UAS-Pp2A-29B RNAi (v49672), UAS-Rab5 RNAi (v34096), UAS-Rab11 RNAi (v108382), UAS-Slmap RNAi (v8199), UAS-Stat92E RNAi (v43866), UAS-Strip RNAi (v16211), UAS-Strip RNAi (v16212), UAS-Tak1 RNAi (v101357), UAS-Traf6 RNAi (v16125) and UAS-Dcr-2 (v60009) stocks were all obtained from the Vienna Drosophila Resource Center (VDRC). The Strip^dogi, Df(3L)tc-1 and Strip-Myc^2-38-4-1 stocks were gifts from Takahiro Chihara (Hiroshima University, Japan). The UAS-Egfr^DN, UAS-Egfr^λ, UAS-GAL4;nanos-GAL4, Upd1-GAL4, C587-GAL4 and tj-GAL4 stocks were gifts from Gary Hime (University of Melbourne, Australia). The UAS-egr, egr¹ and egr³ stocks were gifts from Masayuki Miura (University of Tokyo, Japan). The Mad^1-2 and Mad^8-2 stocks were gifts from Donna Denton (University of South Australia, Australia). Other fly stocks used in this study were 10×Stat92E-GFP (Bach et al., 2007) and hsFLP;; Act>>GAL4, UAS-RFP / TM6B. Flies were raised and all crosses undertaken within a controlled environmental incubator at 25°C (unless otherwise indicated) on a standard cornmeal, molasses and yeast medium. In all relevant crosses, GAL4-containing flies were virgin females. UAS construct-containing stocks that effected a modified gonad phenotype when expressed via C587-GAL4 were checked for their gonad phenotype independently of GAL4 expression, and were all found to be wild type in appearance.

Late-larval and early-pupal lethal stocks were generated via administration of ethyl methanesulfonate (EMS) treatment (5 mM) to isogenised FRT82B flies. Mutagenised stocks were maintained using the balancer chromosome, TM6B, Tb¹, Antp^Hu.

L3 male gonads were dissected in 1× phosphate-buffered saline (PBS) (Amresco, E703) and fixed for 15 min in 4% paraformaldehyde (PFA) (ProSciTech, C005; Alfa Aesar, 43368) in PBS with 0.1% Triton X-100 (PBST). During the screening process, gonads were then washed in PBST and mounted in ProLong Gold Antifade Reagent with DAPI (Molecular Probes, P36935), examined using a Nikon Eclipse Ti Inverted Microscope System and imaged using NIS-Elements Imaging Software (Nikon Instruments). During other experiments, gonads were washed in PBST and blocked in PBST containing 5% normal horse serum (NHS) (Sigma-Aldrich, H0146) for 1 h at room temperature, then incubated overnight at 4°C with primary antibodies diluted in 5% NHS in PBST. Tissues were then washed in PBST and incubated in secondary antibody diluted in 5% NHS in PBST at room temperature for 1.5 h, then washed again in PBST and mounted using ProLong Gold Antifade Reagent with DAPI. Samples were imaged using a Zeiss LSM 510 or Zeiss LSM 780 laser scanning confocal microscope, and imaged using Zen 2012 software (Carl Zeiss). Primary antibodies used were: mouse anti-BrdU (1:200, #5292), rabbit anti-Cleaved Caspase-3 (1:5000, #661), anti-Myc tag (1:4000, 2276), mouse anti-phospho Histone H3 (1:1000, 9706), rabbit anti-phospho-p44/42 MAPK (Erk1/2) (1:200, 4370), rabbit anti-phospho Smad 1/5 (1:100, 13820) (all from Cell Signaling Technology); mouse anti-Hts (1:20, 1B1), mouse anti-Discs large 1 (1:50, 4F3), mouse anti-Bam (1:25), mouse anti-Armadillo (1:10, N27A1), mouse anti-Eyes absent (1:100, 10H6), mouse anti-Fas3 (1:25, 7G10), mouse anti-Mmp1 (1:50, 5H7B11, 3B8D12 and 3A6B4 mixed in a 1:1:1 cocktail) (all from the Developmental Studies Hybridoma Bank); goat anti-Vasa (1:100, sc-26877) (Santa Cruz Biotechnology); rabbit anti-ACTIVE JNK pAb (1:500, V793A) (Promega); rabbit anti-Stat92E (1:1000) (a gift from Erika Bach, New York University, USA); rabbit anti-Zn finger homeodomain 1 (1:5000) (a gift from Ruth Lehmann, New York University, USA); rat anti-Matotopetli (1:500) (a gift from Helen White-Cooper, Cardiff University, UK); guinea pig anti-Traffic jam (1:5000) (a gift from Dorothea Godt, University of Toronto, Canada); rabbit anti-Eiger (1:250) (a gift from Masayuki Miura, University of Tokyo, Japan); mouse anti-Nimrod C1 (1:500) (a gift from István Andó, Hungarian Academy of Sciences). Alexa Fluor secondary antibodies (A-11055, A-11058, A-21447, A-11073, A-21202, A-21203, A-31571, A-21206, A-21207, A-31573, A-21208 and A-21209, Molecular Probes) were used at dilutions of 1:500.

Prior to fixation, gonads were incubated in 0.326 mM BrdU (Sigma-Aldrich, B5002) for 1 h at room temperature. After fixation for 15 min in 4% PFA, tissues were washed in PBST, treated with 50 U/ml DNase I (Invitrogen, 18068-015) for 45 min at 37°C, and the reaction halted with 2 μl EDTA (25 mM) (Amresco, 0322). TUNEL assays were performed using a Click-iT Plus TUNEL Assay kit (Molecular Probes, C16019).

To quantify M-phase cells, overlapping z-sections were taken through whole gonads stained for pHis3, Zfh1 and Vasa. Using Zen 2012 software, the numbers of proliferating germline cells (co-expressing pHis3 and Vasa) and somatic cells (co-expressing pHis3 and Zfh1) were quantified. To generate Fig. 3E, synchronously proliferating germline cells within spermatogonia and spermatocytes were counted as single proliferating cells, and the significance was assessed using Student's t-test. To generate Fig. 3F, proliferating somatic cells were ignored, and the number of each type of germline cell undergoing proliferation was judged by appearance and the proximity of pHis3 stains. To quantify gonad sizes, their area was measured using Fiji (Schindelin et al., 2012) at the z-section of their largest circumference. Data were analysed using Student's t-test and visualised using GraphPad Prism 7.

Full-length Strip (isoform A) open reading frame (ORF) sequence was amplified from the expressed sequence tag (EST) AT20596 [Drosophila Genomics Research Center (DGRC), 1035472] via PCR using Pfu DNA polymerase (Promega, #M7741) and the following primers: 5′-CACCATGATGCTCACATCCATAAACAACTC-3′ (forward); and 5′-AGGGCGTCCCAGTCGGTC-3′ (reverse).

The Strip ORF was cloned into a pTVW vector (DGRC, 1091) using a standard Gateway Technology method (Life Technologies, #12535-019). The pTVW-Strip expression clone was purified using a Purelink HiPure Plasmid Filter Midiprep Kit (Life Technologies, K210014) and used to create transgenic fly stocks by BestGene.

L3 carcasses, with the gut removed, were washed and dissected in 1× PBS and homogenised in NETN lysis buffer (cshprotocols.cshlp.org/content/2009/1/pdb.rec11595.full) with added PMSF proteinase inhibitor (Amresco, M145) and a phosphatase inhibitor cocktail (Sigma-Aldrich, P5726). Sample lysate protein concentration was quantified using a DC Protein Assay kit (BioRad, 500-0112). Some lysate was set aside and used as the positive control (input).

For co-immunoprecipitation experiments, primary antibodies were added to sample lysates and incubated overnight at 4°C. Primary antibodies used were: rat anti-Strip (1:100) (a gift from Takahiro Chihara, Hiroshima University, Japan) and rabbit anti-Cka (1:1000) (a gift from Wei Du, The University of Chicago, USA). Negative control samples underwent identical treatments, but without incubation with primary antibodies. Samples were incubated for 4 h with Protein G Sepharose 4 Fast Flow beads (GE Healthcare, 17-0618-01) then washed with TBST.

Samples were boiled for 5 min in 1× Laemmli SDS buffer, and loaded on a 4-15% Mini-PROTEAN TGX Precast Protein Gels (BioRad, 456-1084). Samples loaded were of equal protein quantity (1×), alongside an additional experimental quantity (2×). Gels were run at 200 V and Precision Plus Protein WesternC Standards (Bio-Rad, 161-0376) was used as a ladder. Blotting was performed with a Trans-Blot Turbo Transfer pack (Bio-Rad, 170-4156, 170-4158). Membranes were blocked in 5% skim milk blocking solution and washed with TBST. Primary antibodies used were: mouse anti-β-Tubulin (1:200) (DSHB); rat anti-Strip (1:200) (a gift from Takahiro Chihara); rabbit anti-Cka (1:1500) (a gift from Wei Du); rabbit anti-Hep (1:250, ab1957) (a gift from Julian Ng, University of Cambridge, UK). Secondary antibodies used were: anti-rat HRP (1:1000, P0450), anti-mouse HRP (1:1000, P0447) and anti-rabbit HRP (1:2000, P0448) (all from Dako); and anti-rabbit HRP, light chain specific (1:10000, 211-032-171) (Jackson ImmunoResearch Laboratories). Blots were visualised using the Clarity Western ECL Substrate kit (BioRad, 170-5060) or the Amersham ECL Western Blot Detection Reagents kit (GE Healthcare, RPN220), and imaged using a GelDoc XRS+ system (BioRad).

We aimed to generate a stock containing endogenous Strip tagged at the C terminus with 3×Myc epitope tag sequences via CRISPR/Cas9 gene editing. To generate the guide RNA plasmid, DNA oligos (5′-GTCGACCGACTGGGACGCCCTCT-3′ and 5′-AAACAGAGGGCGTCCCAGTCGGT-3′) were hybridised and subcloned into BbsI-digested pCFD3-dU6:3gRNA plasmid (Addgene, 49410) (a gift from Simon Bullock, University of Cambridge, UK) (Fig. S6A-C). The homology-directed repair (HDR) donor plasmid, consisting of a 3×Myc tag and 500 bp homology arms, was synthesised by GenScript (Fig. S6D,E). Microinjection of flies, stock generation and PCR screening were performed by WellGenetics. Transgenic flies expressing the guide RNA (y², cho², v¹; pCFD3:U6:3-3×Myc-Strip@attP40/CyO) were generated and crossed with Cas9-expressing flies, and ∼200 progeny were injected with the HDR donor plasmid. Following PCR screening (using the primers 5′-TCTTCTGAGATGAGTTTTTGTTCG-3′ and 5′-CTTACCCACTCGGACTTTGC-3′) and DNA sequencing (using the primer 5′-ACCTCAAGCCAAGTGTCGTT-3′), eight stocks possessing Strip::3×Myc (w¹¹¹⁸;; Strip::3×Myc / TM6B) were generated.

RNAi lines were crossed to hsFLP;; Act>>GAL4, UAS-RFP/TM6B and raised at 25°C, and heat-shocked at 37°C for 1 h as first-instar larvae to activate transgene expression. Tissue for RNA extraction (n>10 animals per genotype) was acquired from adults for the cross to the RNAi line against imd, and from pupae for the crosses to RNAi lines against Fgop2, GckIII, Mob4, msn and Slmap, as these latter crosses were lethal at the pupal stage. Tissue from both adults and pupae was obtained for the control crosses to the RNAi line against luciferase. RNA was extracted using an RNeasy Mini Kit (QIAGEN, 74104). DNA digestion was performed using an ezDNase kit (Invitrogen, 11766051). Synthesis of cDNA was performed using a SuperScript III First-Strand Synthesis kit (Invitrogen, 18080-051). RT-qPCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems, 4367659) on a CFX384 Touch Real-Time PCR Detection System (BioRad). Quantification cycle (Cq) values were obtained using BioRad CFX Manager software (v3.1). The data were normalised to expression of the housekeeping gene Gapdh2. The primer sequences used are as follows: Fgop2, forward 5′-TACCCGCACCGAACTGCC-3′ and reverse 5′-AATCTGGCCCACGGATAGGT-3′; GckIII, forward 5′-CGAAGCCGTAGCAAATTCCG-3′ and reverse 5′-ATGTCTGCTCCAGCGTTTCA-3′; Mob4, forward 5′-ACTTCCCTAGCAGGGTGTCC-3′ and reverse 5′-CACAGATACGTCTCGGCCTC-3′; msn, forward 5′-CCCGCTTTTGGTCGATCTCA-3′ and reverse: 5′-TGCACCCTGAGTATGCTTGG-3′; Slmap, forward 5′-GAACTCTGCGATGTGTTGGC-3′ and reverse: 5′-TGGCAGTCAACTGGTACGTT-3′; imd, forward 5′-GGGATCTTGGCATGTCGGAA-3′ and reverse 5′-ACAGCTGGTATATCACCTCTCTA-3′; Gapdh2, forward 5′-GCAAGCAAGCCGATAGATAAACA-3′ and reverse 5′-CGTTGGCGCCCTTATCAATG-3′.

We thank Peter Lock and the LIMS Bioimaging Facility for microscopy equipment and technical support. We thank Keng Chen for his role in generating the EMS mutant collection used in our screen. We thank the Australian Drosophila research community, the BDSC and the VDRC for providing fly stocks, and OzDros for stock importation services. We also thank Linda Parsons, Jan Manent, Peter Burke, Molly Buntine, Sachini Arachchilage, Marta Portela, Kirsten Allan, Monica Ivanyi and Alex Lowdin for suggestions and general experimental assistance. The Drosophila Genomics Resource Center is supported by a National Institutes of Health grant (2P40OD010949). The Developmental Studies Hybridoma Bank was created by the National Institute of Child Health and Human Development of the National Institutes of Health and is based at The University of Iowa. Indicated transgenic fly stocks were obtained from the Vienna Drosophila Resource Center (Dietzl et al., 2007).