VCP promotes tTAF-target gene expression and spermatocyte differentiation by downregulating mono-ubiquitylated H2A Open Access

Valosin-containing protein (VCP; also known as TER94) is a broadly expressed protein that functions as a hexameric AAA+ ATPase to regulate protein homeostasis by binding ubiquitylated substrates and targeting them for degradation (Ye et al., 2017). Given its essential activities in protein homeostasis, VCP has been studied extensively in relation to aging and degenerative diseases (Darwich et al., 2020; Johnson et al., 2015; Kakizuka, 2008; Laço et al., 2012; Ritson et al., 2010; Scarian et al., 2022). However, VCP also acts in a variety of other biological contexts, including germ-cell development. In oocytes of the nematode Caenorhabditis elegans, CDC-48/VCP is particularly significant during meiotic divisions, when it regulates chromosome condensation and segregation (Mouysset et al., 2008; Sasagawa et al., 2012, 2009, 2007). Although there are indications that VCP may likewise perform regulatory roles during sperm development (Wu et al., 2021), it is unclear at what stage VCP function may be most crucial. Stage-specific transcriptomic data from Drosophila testes suggest that VCP is strongly expressed in meiotic-stage spermatocytes (Shi et al., 2020). Although this hints at a potentially important role for VCP at the spermatocyte stage, the molecular functions of VCP in meiotic spermatocytes are unknown.

Drosophila spermatogenesis presents an ideal model in which to investigate conserved mechanisms regulating male germ-cell development, as the steps involved in making a viable sperm cell are similar in flies and mammals (Griswold, 2016; Hennig, 1992). At the apical tip of the Drosophila testis, germline stem cells divide, producing a daughter cell that undergoes four rounds of mitotic division to yield 16 spermatogonia (Fig. 1A). Subsequently, spermatogonia differentiate into spermatocytes, which enter meiotic prophase (Fig. 1A). Meiotic spermatocytes can be readily identified based on nuclear morphology; at this stage, nuclei are significantly larger than in spermatogonia, homologous chromosomes occupy distinct territories in the nucleus, and actively transcribed regions of the Y chromosome, termed Y-loops, extend into the nuclear interior (Fig. 1A) (Bonaccorsi et al., 1988; Cenci et al., 1994; Fingerhut et al., 2019; Mahadevaraju et al., 2021). Notably, extensive transcriptional rewiring occurs at meiotic prophase during spermatogenesis in flies (Fuller, 1998; Li et al., 2022; Mahadevaraju et al., 2021; White-Cooper, 2010; Witt et al., 2021), as in humans (Jan et al., 2017). These changes to transcription promote meiotic progression and are also needed to support later stages in germ-cell development (Lin et al., 1996; White-Cooper, 2010; White-Cooper et al., 1998). In Drosophila, testis-specific TBP-associated factors (tTAFs) are required to stimulate the spermatocyte gene expression program (Chen et al., 2005, 2011; Fuller, 1998; Hiller et al., 2004; White-Cooper et al., 1998). When tTAFs are non-functional, germ cells arrest as spermatocytes, causing infertility in male flies (Lin et al., 1996). Notably, a common form of human male infertility, non-obstructive azoospermia with spermatocyte arrest, exhibits similar cellular and developmental defects (Meyer et al., 1992; Soderström and Suominen, 1980) as tTAF mutants (Lin et al., 1996). Identifying factors that support robust spermatocyte gene expression, perhaps in cooperation with tTAFs, could reveal fundamental controls relevant to the preservation of male fertility.

Molecularly, tTAFs have been postulated to drive spermatocyte gene expression by antagonizing the activity of a transcriptional repressor, Polycomb (Pc) (Chen et al., 2005, 2011). Pc is a core component of Polycomb repressive complex I (PRC1), which inhibits the expression of target genes, including meiosis-related genes (Endoh et al., 2017; Zdzieblo et al., 2014), by catalyzing H2A mono-ubiquitylation (H2Aub) at repressed gene loci (Blackledge et al., 2020; de Napoles et al., 2004; Endoh et al., 2008; Wang et al., 2004). Interestingly, downregulation of PRC1 activity is a requisite step for meiotic entry and cellular differentiation in several developmental contexts (Tang et al., 2016; Yao et al., 2018; Zdzieblo et al., 2014). Despite these general themes, how downregulation of PRC1 activity is achieved at the molecular level during male germ-cell development remains incompletely understood, especially in relation to upstream signals from tTAFs. In principle, this may involve activation of factors that help to reverse repressive H2Aub epigenetic marks. However, the spatiotemporal regulation of H2Aub in the testis is unclear, as is the identity of enzymes that would facilitate its possible downregulation at meiosis.

In this study, we report that VCP operates downstream of tTAFs to downregulate H2Aub, thereby permitting spermatocyte gene expression and development in the Drosophila testis. Interestingly, we find that VCP is cytosolic in mitotic spermatogonia, but shuttles into the nucleus in meiotic spermatocytes. Nuclear translocation of VCP is dependent on tTAFs, and, like tTAF mutants, VCP-RNAi testes do not contain germ cells that develop beyond the spermatocyte stage. In line with these findings, we demonstrate that VCP is required to activate the expression of a subset of tTAF-regulated genes, and we observe that VCP translocation into the nucleus at meiotic entry coincides with a VCP-dependent downregulation of H2Aub. Remarkably, inhibiting PRC1 ubiquitin ligase activity is sufficient to advance sperm development past meiosis in the absence of VCP, suggesting that H2Aub downregulation is an important function of VCP in spermatocyte differentiation. Overall, our findings identify VCP as an essential regulator of both spermatocyte gene expression and meiotic progression.

We recently generated a collection of fly strains to assess VCP function in multisystem proteinopathy (MSP-1) (Wall et al., 2021). As part of these studies, we used CRISPR to knock-in a GFP tag at the endogenous VCP locus. While working with this strain, we noticed that VCP-GFP signal was exceptionally bright in third instar larval testes (Fig. 1B), which contain stem cells, spermatogonia and spermatocytes, but not later germ-cell stages (Gärtner et al., 2014). Interestingly, previous RNA-sequencing analyses have likewise suggested that VCP is highly expressed in the fly testis (Brown et al., 2014; Leader et al., 2018; Shi et al., 2020). Thus, we sought to analyze VCP localization and regulatory dynamics during Drosophila spermatogenesis in more detail.

We imaged adult VCP-GFP testes such that we could observe endogenous VCP expression patterns in a fully developed male germline. Although the protein was clearly expressed at multiple early stages in male germ-cell development, its localization varied depending on the developmental stage; VCP-GFP was primarily cytosolic in mitotic spermatogonia, but it translocated into the nucleus as cells developed into meiotic spermatocytes (Fig. 1C). Consequently, spermatocytes showed a higher nuclear-to-cytosolic VCP-GFP ratio compared with spermatogonia (Fig. 1D), and cytosolic VCP-GFP fluorescence was higher in spermatogonia than in spermatocytes (Fig. 1E). We confirmed that VCP enters the nucleus as germ cells exit mitosis and enter meiosis by imaging VCP-GFP in testes with germline-specific knockdown of bag of marbles (bam), which is required for mitotic spermatogonia to progress into the meiotic spermatocyte stage (McKearin and Spradling, 1990). Throughout bam-RNAi testes, VCP-GFP signal was restricted to the cytosol and was undetectable in nuclei (Fig. S1), indicating that the nuclear translocation event occurs after the spermatogonia stage. Interestingly, we found that VCP is excluded from the nucleolus (Fig. 1F) and that VCP-GFP signal is enriched in the interior of spermatocyte nuclei, away from Hoechst-labeled DNA (Fig. 1G-I). This subnuclear localization pattern is intriguing, given that open chromatin threads such as Y-loops also extend into the interior of the nucleus at the spermatocyte stage (Fig. 1I) (Bonaccorsi et al., 1988; Cenci et al., 1994; Fingerhut et al., 2019; Mahadevaraju et al., 2021), hinting at a possible role for VCP in transcription and gene expression. Overall, these data indicate that nuclear translocation of VCP is entrained to the spermatogonia to spermatocyte transition (Fig. 1I) and suggest a particularly important function for VCP in spermatocyte nuclei.

Given the redistribution of VCP into the nucleus at meiotic prophase, we hypothesized that VCP may play an important role in post-mitotic stages of sperm development. Because null VCP mutants are lethal (León and McKearin, 1999), we used a germline-specific Gal4 driver, BamGal4, to knock down VCP in the germline. Importantly, this approach successfully knocked down VCP in spermatocytes, but VCP expression in spermatogonia and cyst cells remained unaffected (Fig. S2A); thus, this presented a suitable approach to evaluate the developmental competency of germ cells upon knockdown of VCP in spermatocytes. As an initial assessment of whether mature sperm could even be produced in the absence of VCP, we first analyzed seminal vesicles. Not only were VCP-RNAi seminal vesicles noticeably shrunken, they were also devoid of mature sperm, as indicated by an absence of Hoechst-labeled, needle-shaped sperm nuclei (Fig. 2A). Using differential interference contrast (DIC) imaging, we confirmed that VCP-RNAi testes, which at a larger tissue level were also atypically small, lacked elongated spermatid bundles (Fig. 2B). Instead, the smaller VCP-RNAi testes appeared to show germ cells arrested at an earlier stage of development alongside some degenerating, post-mitotic germ-cell cysts (Fig. 2B, arrowheads). Consistent with a defect in mature sperm production, VCP-RNAi males were completely infertile (Fig. S2B).

To gain information regarding the precise developmental stage at which germ-cell development arrests in the absence of VCP, we stained DNA with Hoechst and performed high-magnification, phase-contrast microscopy on testis squashes. In control squashed preparations, we observed spermatocytes, round spermatids and elongated spermatids (Fig. 2C, left). However, elongated spermatids and round spermatids were absent from VCP-RNAi squashed preparations (Fig. 2C, right). Instead, we observed an accumulation of spermatocytes with well-formed nucleoli and condensed chromosomes that occupied distinct territories in the nucleus, similar to stage-matched control spermatocytes (Fig. 2C,D). Interestingly, VCP-RNAi spermatocyte nuclei appeared to be larger than control spermatocyte nuclei (Fig. 2D). We investigated this observation further by labeling Lamin C and measuring nuclear area in control and VCP-RNAi spermatocytes. Consistent with our initial observation, nuclear area was significantly increased in VCP-RNAi spermatocytes (Fig. 2E,F), suggesting that VCP may extract cargo from spermatocyte nuclei, as it does in Drosophila photoreceptors (Chang et al., 2021). Additionally, we noted that Hoechst staining appeared brighter in VCP-RNAi spermatocytes compared with controls (Fig. 2D), possibly due to hyper-condensation of chromatin. These data indicate that VCP is required for progression beyond the spermatocyte stage, the stage at which VCP enters the nucleus, and suggest that VCP may be important for nuclear and/or chromatin regulation.

The spermatocyte stage lasts approximately 80-90 h (Chandley and Bateman, 1962) and can be classified into sub-stages based on changes to chromatin morphology (Cenci et al., 1994). The chromatin of VCP-RNAi spermatocytes most closely resembled that of S3/S4-stage control spermatocytes (Fig. 2D), which are relatively early in spermatocyte development (Cenci et al., 1994). However, because VCP appeared to possibly regulate chromatin condensation and dynamics (Fig. 2D), we sought to confirm that germ cells arrested at an early-spermatocyte stage in VCP-RNAi testes by using a measure other than chromatin appearance. For this purpose, we analyzed the expression of gene regions of a Y-chromosome gene, kl-3. Notably, kl-3 is transcribed in a stepwise fashion, with 5′ gene regions being transcribed early in spermatocyte development and gene regions distal to the 5′ end being transcribed later in spermatocyte development (Fingerhut et al., 2019). Interestingly, whereas expression of kl-3 exon 2 was not robustly affected by VCP knockdown (Fig. 2G), expression of the more distal kl-3 exons 6 and 14, which are expressed at later spermatocyte stages (Fingerhut et al., 2019), was strongly reduced in VCP-RNAi testes (Fig. 2G). These data suggest that knockdown of VCP causes cells to arrest early in the spermatocyte stage, at a time point shortly after when we observe VCP entering the nucleus (Fig. 1C).

Similar to VCP, tTAFs are required for progression beyond early spermatocyte stages (Lin et al., 1996). Because loss of VCP and tTAF function each cause a common developmental arrest at the spermatocyte stage, we hypothesized that VCP may function in the same pathway as tTAFs. Intriguingly, co-imaging of a GFP-tagged tTAF protein, Spermatocyte arrest (Sa-GFP), with antibody-labeled VCP indicated that VCP enters spermatocyte nuclei shortly after Sa is expressed in spermatocytes (Fig. S3A). Based on this timing, we were curious about whether tTAFs promote the nuclear translocation of VCP in spermatocytes. Indeed, knockdown of sa and another tTAF, meiosis I arrest (mia), impeded nuclear translocation of VCP in spermatocytes (Fig. 3A,B). However, knockdown of VCP did not reciprocally affect Sa expression or localization (Fig. S3B), indicative of a unidirectional pathway. Thus, we conclude that VCP depends upon tTAFs to enter spermatocyte nuclei.

A major function of tTAFs is to activate the spermatocyte gene expression program (Chen et al., 2005, 2011; White-Cooper et al., 1998). Given that VCP enters the nucleus downstream of tTAFs, we hypothesized that VCP regulates the expression of spermatocyte-specific genes as part of this program. To test this hypothesis, we first examined the localization and intensity of phospho-serine 2 RNA polymerase II (Pol II S2p), which is a marker of transcriptional elongation (Mahadevaraju et al., 2021; Phatnani and Greenleaf, 2006). In both control and VCP-RNAi spermatocytes, Pol II S2p was undetectable at the X-chromosome (Fig. 3A), suggesting that the X-chromosome is properly inactivated under each condition (Mahadevaraju et al., 2021; Witt et al., 2021). Unlike the X-chromosome, autosomes showed clear Pol II S2p signal in both control and VCP-RNAi spermatocytes (Fig. 3C). However, Pol II S2p fluorescence intensity was significantly brighter in VCP-RNAi spermatocytes (Fig. 3C,D), consistent with a change in transcription.

Because VCP functions in the same pathway as tTAFs and transcription is downregulated in tTAF mutants (Chen et al., 2005, 2011; Fuller, 1998), we hypothesized that increased Pol II S2p in spermatocytes of VCP-RNAi testes indicates stalled transcription, rather than enhanced transcription. We tested this hypothesis by performing RT-qPCR to measure quantitatively the expression of a subset of tTAF-regulated genes [Mst98C, Mst87F, Mst35Bb (ProtB), janB, gonadal] (White-Cooper et al., 1998) in control and VCP-RNAi testes. As an additional control to verify tTAF dependence, we performed similar analyses on sa-RNAi testes. To rule out possible artifacts caused by developmental arrest at early spermatocyte stages in VCP­-RNAi and sa-RNAi testes, we normalized tTAF-target gene expression to that of Cyclin A, which is uniformly expressed throughout spermatocyte development and is not regulated by tTAFs (Chen et al., 2011; White-Cooper et al., 1998); this allowed us to compare directly differences in gene expression specifically in relation to the gene knockdown of interest. As in sa-RNAi testes (Fig. 3E), we detected a sharp decrease in each of the five tested tTAF-regulated genes in VCP-RNAi testes (Fig. 3E). Thus, we conclude that VCP drives the expression of several tTAF-regulated spermatocyte genes, which could potentially explain the meiotic-arrest phenotype observed upon loss of VCP function.

We next investigated how VCP supports gene expression downstream of tTAFs. Notably, tTAFs have been postulated to antagonize PRC1-mediated repression of meiotic gene expression in spermatocytes (Chen et al., 2005, 2011). Strikingly, we found that H2Aub, a repressive histone mark catalyzed by PRC1, decreased in wild-type testes specifically at the spermatocyte stage; H2Aub was bright in spermatogonia and somatic cyst cells throughout the testis, but H2Aub signal declined dramatically in spermatocytes (Fig. S4A,B). In multiple contexts, VCP binds and extracts ubiquitylated cargo (Ye et al., 2017). As the reduction in H2Aub during spermatogenesis coincided with nuclear entry of VCP, we tested whether H2Aub downregulation in spermatocytes depends upon VCP function. Indeed, high H2Aub signal was retained in spermatocytes and was uniformly distributed throughout the spermatocyte nucleoplasm when VCP was knocked down (Fig. 4A, insets), potentially owing to a failure to extract H2Aub from spermatocyte nuclei. Additionally, H2Aub was significantly more abundant in protein extracts from whole testes (Fig. 4B,C). Knockdown of sa likewise showed an increase in H2Aub in spermatocytes (Fig. S5A) and protein extracts from whole testes (Fig. S5B,C), as would be expected if VCP operated downstream of tTAFs to trigger the downregulation of H2Aub.

To corroborate that H2Aub downregulation during spermatogenesis was dependent on VCP activity, we overexpressed an allele of VCP that lacks ATPase activity (VCP^K2A) (Chang et al., 2021). Because VCP functions as a hexamer (Wang et al., 2003), overexpression of VCP^K2A causes dominant-negative effects by generating hexamers containing the ATPase-dead protein in complex with the wild-type protein, thus impeding VCP function. For this approach, we were unable to drive VCP^K2A overexpression using BamGal4, as this caused lethality prior to adulthood. To overcome this experimental limitation, we utilized an auxin-inducible gene expression system (McClure et al., 2022) to restrict transgene expression to adults. This system utilizes a ubiquitously expressed Gal4 inhibitor, Gal80, fused to an auxin-inducible degron (AID). Upon auxin feeding, auxin binds Gal80 and causes its degradation, thus permitting Gal4 activity and transgene expression. For this experiment, we used the VasaGal4 driver, which is active in all germ cells (Demarco et al., 2014) and should, in principle, provide the strongest transgene expression. Consistent with our VCP-RNAi data, H2Aub signal was higher in spermatocytes of auxin-fed VCP^K2A males relative to controls (Fig. 4D,E). Thus, we conclude that VCP ATPase activity is indeed required for H2Aub downregulation in spermatocytes.

To gain insight into how VCP may downregulate H2Aub, we screened several known VCP co-factors (Fig. S6A) for gene knockdowns that increased H2Aub in spermatocytes. Among these candidates, we found that only knockdown of Ufd1, which encodes a ubiquitin-recognition factor that binds VCP and supports the sequestration and degradation of ubiquitylated proteins (Ramadan et al., 2007; Sasagawa et al., 2009; Shcherbik and Haines, 2007), blocked H2Aub downregulation in spermatocytes (Fig. 5A, Fig. S6A). We conclude that VCP may cooperate with Ufd1 to drive H2Aub downregulation.

Because VCP and Ufd1 commonly act together to drive the extraction and degradation of ubiquitylated substrates, we hypothesized that VCP and Ufd1 may stimulate H2Aub downregulation by promoting the turnover of H2A in spermatocytes. Notably, the nuclear-swelling phenotype observed in VCP-RNAi spermatocytes (Fig. 2E,F) suggests that VCP may indeed control the extraction and degradation of various nuclear proteins, such as H2A. To test this hypothesis, we used BamGal4 to express a transgenic H2A-turnover reporter construct (FRT-H2A-GFP-FRT-H2A-mCherry) (Kahney et al., 2021). This system can be used to measure the turnover of fluorescently tagged, transgenic H2A, but it does not provide information regarding endogenous H2A turnover. To activate this system, we used heat shock to transiently induce expression of a FLPase, which removes the H2A-GFP cassette and permits H2A-mCherry expression. Given time, older H2A will remain marked by GFP, whereas nascent H2A will be marked by mCherry. We first verified the functionality of this system by amplifying the H2A-turnover transgene in extracts from testes with or without the FLPase transgene. In both control and VCP-RNAi testes, we observed more robust amplification of the smaller FRT-H2A-mCherry DNA cassette, as opposed to the full FRT-H2A-GFP-FRT-H2A-mCherry transgene, after FLPase induction (Fig. S6B), confirming that our approach effectively triggered recombination.

We then utilized this system to examine H2A turnover in vivo by microscopy. In control testes 5 days after heat shock, GFP signal was barely detectable above background levels whereas mCherry signal was bright (Fig. 5B, Fig. S6C), consistent with high H2A turnover under normal conditions. We then investigated whether VCP depletion would cause a significant retention in GFP signal, as would be expected if VCP were required for H2A turnover. Indeed, 5 days after heat shock, GFP signal remained bright in many spermatocyte cysts of VCP-RNAi testes (Fig. 5B, Fig. S6C). Quantification of GFP intensity and the GFP/mCherry ratio in spermatocytes within fields of view indicated a significantly higher mean GFP intensity (Fig. 5C) and GFP/mCherry ratio (Fig. 5D) in VCP-RNAi testes compared with controls (Fig. 5D). Despite this general increase in GFP signal relative to mCherry signal in VCP-RNAi testes, we reproducibly observed inter-cyst variability in the degree to which H2A turnover was inhibited (Fig. 5B, Fig. S6C). Whereas some cysts showed bright GFP signal with almost no detectable mCherry signal, other cysts showed dimmer GFP signal and relatively stronger mCherry signal, which nonetheless remained dimmer than in controls (Fig. 5B). Although the reason for this inter-cyst variability is currently unclear, the evident retention of more GFP signal in VCP-RNAi testes compared with controls supports the more general conclusion that VCP activity enables robust H2A turnover in spermatocytes, potentially contributing to VCP-dependent H2Aub downregulation.

Ubiquitylated substrates extracted by VCP are often targeted to the proteasome for degradation (Ye et al., 2017). Therefore, we next examined whether proteasomal degradation was also involved in H2Aub downregulation. Notably, mutation in Rpt2, a gene that codes for a proteasome component, causes a meiotic arrest resembling VCP-RNAi and tTAF mutant testes (Lu et al., 2013). Whereas germline-specific knockdown of Rpt2 in fact blocked downregulation of H2Aub in spermatocytes (Fig. S6D), VCP remained almost completely cytosolic in spermatocytes and failed to effectively translocate into the nucleus (Fig. S6E). Thus, although it remains possible that the proteasome may also act downstream of VCP to degrade extracted H2Aub, proteasome function is apparently important for VCP nuclear translocation, which precludes us from definitively testing whether H2Aub is degraded by the proteasome after extraction in this context.

H2Aub is a well-characterized repressive histone modification (Blackledge et al., 2020; de Napoles et al., 2004; Endoh et al., 2008; Wang et al., 2004). Given that VCP promotes spermatocyte gene expression and H2Aub downregulation, we hypothesized that inhibition of H2Aub in VCP-RNAi testes may at least partially rescue spermatocyte gene expression. To test this hypothesis, we used a BamGal4 driver to simultaneously knock down VCP and sex combs extra (Sce), the PRC1 E3 ubiquitin ligase that catalyzes H2Aub (de Napoles et al., 2004; Wang et al., 2004). As expected, H2Aub levels were dampened in spermatocytes of VCP Sce double-knockdown (DKD) testes, such that the pattern of H2Aub resembled that of control spermatocytes (Fig. 6A). Importantly, H2Aub did not appear to be affected in spermatogonia or cyst cells (Fig. 6A), indicating that Sce knockdown is restricted to spermatocytes. We next tested whether blocking H2Aub in the absence of VCP would restore normal Pol II S2p levels. Indeed, Pol II S2p intensity at autosomes was significantly decreased in DKD testes compared with VCP-RNAi testes (Fig. 6B,C), suggesting that normal transcription may be restored in DKD testes. Notably, each of the five tTAF-target genes we previously found to be downregulated in VCP-RNAi testes (Fig. 3C) increased in expression in DKD testes (Fig. 6D). To broaden our analysis, we analyzed five additional, putative tTAF targets (Jiang et al., 2018; White-Cooper et al., 1998) and confirmed that all but one (HtrA2) were indeed expressed significantly higher in DKD testes compared with VCP-RNAi testes (Fig. 6D). Despite the partial rescue of tTAF-target gene expression in DKD testes, blocking H2Aub in the absence of VCP did not suppress the nuclear swelling phenotype (Fig. 6E,F), suggesting that VCP-dependent regulation of tTAF-target gene expression may not be coupled to its regulation of nuclear size. Collectively, these data indicate that VCP downregulates H2Aub to drive the expression of several tTAF-target genes (Fig. 6G).

Like VCP and sa, multiple de-repressed genes in DKD testes, including Hsc70-2 and Ubi-p63E, are required for meiotic progression (Azuma et al., 2021; Lu et al., 2013). We therefore hypothesized that genetic inhibition of H2Aub may rescue some aspects of germ-cell development in VCP-RNAi testes. To test this hypothesis, we used microscopy to assess how far germ-cell development progressed in DKD testes. Although DKD testes still failed to produce elongated spermatid bundles or mature sperm (Fig. 7A), DKD testes were visibly larger than VCP-RNAi testes (Fig. 7A) and exhibited a significant reduction in degenerating post-mitotic germ-cell cysts, which were prevalent in VCP-RNAi testes (Fig. 7A,B).

We hypothesized that the increased testis size and suppression of cyst degeneration in DKD testes may be indicative of the presence of more-developed germ cells, such as post-meiotic spermatids. Post-meiotic spermatids are characterized by compact chromatin and the presence of a mitochondrial-based structure known as the nebenkern. As an initial assessment of potential spermatid development, we stained testes with Hoechst and a mitochondria-specific dye, MitoTracker. Remarkably, although VCP was non-functional, we observed post-meiotic germ cells with compact chromatin and a circular nebenkern structure adjacent to nuclei in DKD testes (Fig. S7A), indicating that DKD testes were indeed capable of producing post-meiotic spermatids. Importantly, double knockdown of sa and Sce did not similarly permit meiotic progression (Fig. S7B), suggesting that tTAFs likely perform some additional function independent of H2Aub downregulation that is needed to drive meiotic progression.

The presence of a circular nebenkern in DKD spermatids suggested that these cells may arrest at the round spermatid stage, an early sub-stage of spermatid development (Fabian and Brill, 2012). To probe the spermatid stage in more detail, we used phase-contrast microscopy. By this approach, we also detected spermatids with compact chromatin and non-elongated mitochondria in DKD testes (Fig. 7C); however, we noted that the nebenkern did not appear as a phase-dark structure in DKD spermatids (Fig. 7C), as it did in controls (Fig. 7C), hinting that VCP may also regulate mitochondrial morphogenesis in spermatids. Interestingly, although a circular nebenkern is characteristic of round spermatids, we also detected by phase-contrast microscopy a protein dot, which is more characteristic of elongating spermatids (Fabian and Brill, 2012). Thus, it is possible that the spermatids we detected in DKD testes may represent a hybrid and/or transitional state somewhere between the round and elongating spermatid stages.

As a complementary approach to assess the potential rescuing effects of Sce inhibition, we used mosaic analysis with a repressible cell marker (MARCM) to examine the effects of an Sce^KO allele (Gutiérrez et al., 2012) in the germline. In this system (Lee and Luo, 1999), FLP-based recombination produces cells that are homozygous for the allele of interest (i.e. Sce^KO) and do not express Gal80, a Gal4 inhibitor (Duffy, 2002); uninhibited Gal4, in turn, drives expression of a GFP marker that enables the identification of MARCM clones (Lee and Luo, 1999). In contrast, GFP-negative cells in the same tissue, which possess at least one functional copy of the gene of interest (i.e. Sce), express Gal80 and block Gal4 activity. With this set-up, we first ensured that GFP-positive Sce^KO homozygotes could not mono-ubiquitylate H2A by imaging H2Aub in spermatogonia 2 days-post clone induction (dpci). As expected, GFP-positive Sce^KO clones were negative for H2Aub, but neighboring GFP-negative clones were positive for H2Aub (Fig. S7C). Thus, this MARCM strategy provides a means to effectively block H2Aub cell-autonomously in developing germ cells.

Experimentally, the MARCM system can also be applied to drive RNAi-mediated knockdown of genes specifically in GFP-positive MARCM clones, but not in GFP-negative neighboring cells. In GFP-positive cells, Gal80 is absent, which permits Gal4 activity, and therefore expression of an RNAi transgene (i.e. VCP-RNAi). In GFP-negative cells, Gal80 is present, which inhibits Gal4 activity and RNAi transgene expression (Lee and Luo, 1999). We therefore used the MARCM system to generate germ-cell clones expressing VCP-RNAi that were also homozygous for an Sce^KO allele, or, as a control, an Sce^WT allele (Fig. 7D). We tested whether VCP was indeed knocked down in GFP-positive cells by labeling VCP with an antibody in MARCM testes. As expected, VCP was undetectable in GFP-positive germ-cell nuclei, although it remained detectable in neighboring GFP-negative germ cells (Fig. S7D). After confirming that this system functioned properly, we imaged testes at 7-9 dpci, when cells should be in post-meiotic stages if they developed normally (Chandley and Bateman, 1962). As a control, we imaged VCP-RNAi, Sce^WT MARCM testes, which we found never contained post-meiotic GFP-positive cysts (0/16 testes); we observed only GFP-positive spermatocytes in these testes (Fig. 7E), consistent with an essential function of VCP at the spermatocyte stage. In contrast, GFP-positive spermatids were present in 12/18 VCP-RNAi, Sce^KO MARCM testes (Fig. 7E). Similar to DKD testes, GFP-positive spermatids exhibited compact chromatin and a circular nebenkern in VCP-RNAi, Sce^KO MARCM testes (Fig. S7E). Using phase-contrast imaging, we also observed a protein body at spermatid nuclei in VCP-RNAi, Sce^KO MARCM testes (Fig. 7E), but were unable to detect the nebenkern as a phase-dark structure. Together with the DKD analysis, these data indicate that blocking H2Aub in the absence of VCP generates spermatids with features characteristic of both the round and elongating spermatid stages (i.e. circular nebenkern or protein body, respectively). Thus, although it appears that VCP must also execute additional functions to support germline development past the elongating spermatid stage, we conclude that VCP downregulates H2Aub in spermatocytes to promote spermatocyte differentiation.

Although previous studies have demonstrated that VCP influences germ-cell health and development (León and McKearin, 1999; Sasagawa et al., 2012, 2009, 2007; Wu et al., 2021), fundamental questions remain regarding how VCP functions and is regulated in this biological context, particularly in the male germline. In this study, we found that VCP translocates from the cytosol to the nucleus at the mitotic-meiotic transition and downregulates H2Aub to promote changes in gene expression needed for subsequent stages of spermatogenesis (Fig. 8). These findings present a new molecular framework for understanding developmental regulation of VCP and the mechanisms by which it ultimately controls spermatocyte development.

In the Drosophila testis, nuclear entry of VCP at the spermatocyte stage acts as an initiating switch-like event that sets into motion a series of changes needed for developmental progression (Fig. 8). Molecularly, how the regulated nuclear entry of VCP occurs at this developmental stage is unknown, but it appears to be controlled at some level by upstream tTAFs. Although tTAFs can bind and recruit transcriptional machinery important for the expression of spermatocyte and spermatid differentiation genes (Chen et al., 2005, 2011; Hiller et al., 2004), direct binding between tTAFs and VCP has not been documented and seems unlikely given that VCP (Fig. 2) and tTAFs (Fig. S3) (Chen et al., 2005) have distinct localization patterns. Alternatively, other indirect methods of regulation may be at play. Structural studies have indicated that phosphorylation of VCP at a C-terminal tyrosine residue supports its nuclear translocation in other systems (Madeo et al., 1998; Song et al., 2015), raising the possibility that kinase signaling, perhaps downstream of tTAFs, could also feed into this regulation. In the future, it will be important to design and characterize VCP mutants that cannot re-localize to the spermatocyte nucleus to probe further the molecular repercussions of this specific developmental event. Notably, impairments to VCP nucleocytoplasmic shuttling are characteristic of degenerative pathologies, including MSP-1 (Song et al., 2015), and it remains possible that similar factors may directly regulate VCP nuclear entry in somatic and germ cells. Thus, clarifying controls on VCP nuclear translocation during spermatogenesis may not only identify entry points for treating forms of male infertility, including non-obstructive azoospermia, but may also reveal novel targets for treating VCP-related degenerative diseases.

Following nuclear entry, VCP then executes molecular functions important for spermatocyte gene expression. One important function of nuclear VCP appears to be downregulation of H2Aub, an activity that supports spermatocyte differentiation and the expression of at least some tTAF-regulated genes (Fig. 8). It is important to note that although VCP is essential for male germ-cell development, it may control only one branch of signaling downstream of tTAFs; whereas blocking H2Aub in the absence of VCP permitted spermatocyte differentiation, blocking H2Aub in the absence of a tTAF, Sa, did not, consistent with previous studies (Chen et al., 2005, 2011). This suggests that tTAFs support spermatocyte differentiation through multiple signaling branches, which may be independently required. Indeed, tTAFs have been shown to also control other epigenetic modifications, including trimethylation of H3K4 (Chen et al., 2005), hinting at possible branches besides VCP-H2Aub that may act in concert.

Mechanistically, our data suggest that VCP promotes H2Aub downregulation in spermatocytes at least in part via H2A extraction and turnover. Further investigation of whether and/or how endogenous H2A turnover is developmentally coordinated among nuclei within a cyst, and if there are differences in the timing or execution of this event between cysts in the same testis, could reveal potential complexities in regulation pertinent to gene-expression control at the spermatocyte stage. Because PRC1 catalyzes H2Aub at target gene loci (Blackledge et al., 2020; de Napoles et al., 2004; Endoh et al., 2008; Wang et al., 2004), our findings also support the model that PRC1 activity is indeed inhibitory to spermatocyte gene expression and differentiation, a point of contention in the spermatogenesis field (Chen et al., 2005, 2011; El-Sharnouby et al., 2013). Interestingly, the PRC1 components Pc and Sce are still present in spermatocytes (Chen et al., 2005), where H2Aub levels are normally lowered. Thus, it is unclear how spermatocytes prevent mono-ubiquitylation of new H2A in spermatocytes. Because VCP is absent from nucleoli (Fig. 1F) and less abundant at autosomes (Fig. 1G,H), where PRC1 components localize (Chen et al., 2005), it seems unlikely that VCP would directly interact with PRC1 components outside of transient interactions, but it could potentially serve as a barrier to prevent PRC1 components from entering the nuclear interior. Alternatively, VCP may cooperate with an associated deubiquitinase, such as Asx and/or Calypso (ASXL and BAP1 in mammals, respectively) (Scheuermann et al., 2010), or extract polyubiquitylated substrates involved in making this histone modification from target gene loci, similar to how it extracts Aurora B from chromatin in Xenopus embryos (Ramadan et al., 2007). Consistent with the latter possibility, mutation of the major polyubiquitin gene, Ubi-p63E, results in a similar meiotic arrest as VCP-RNAi (Lu et al., 2013). Going forward, it will be important to clarify precisely how VCP downregulates H2Aub at the mitotic-meiotic transition, as this mechanism may also be pertinent to understanding the relief of PRC1 inhibition in other natural, developmental contexts.

Flies were maintained on standard cornmeal/agar food at 25°C, unless otherwise noted. For RNAi experiments, flies were incubated on standard cornmeal/agar food at 29°C for 5-7 days prior to dissection and imaging to boost Gal4 activity unless otherwise noted.

The following fly strains were used in this study: w¹¹¹⁸, VCP-GFP (Wall et al., 2021), UAS-VCP-RNAi (Vienna Drosophila Resource Center, #24354; FBst0455418), BamGal4 (Doug Harrison, University of Kentucky, KY, USA), sa-GFP (Chen et al., 2005), UAS-bam-RNAi [Bloomington Drosophila Stock Center (BDSC), #33631; FBst0033631), UAS-sa-RNAi (BDSC, #36730; FBst0036730), UAS-mia-RNAi (BDSC, #57790; FBst0057790), UAS-Sce-RNAi (BDSC, #67924; FBst0067924), UAS-Ufd1-RNAi (BDSC, #41823; FBst0041823), UAS-CG8042-RNAi (BDSC, #41604; FBst0041604), UAS-Faf2-RNAi (BDSC, #43224; FBst0043224), UAS-casp-RNAi (BDSC, #44027; FBst0044027), UAS-Npl4-RNAi (BDSC, #53004; FBst0053004), UAS-elg1-RNAi (BDSC, #63551; FBst0063551), UAS-Rpt2-RNAi (BDSC, #34795; FBst0034795), Sce^KO (BDSC, #80157; FBst0080157), UAS-GFP^nls (lab stock), VasaGal4; hsFLP.D5, UAS-GFP/cyo; FRT82B, tubGal80/TM6B (Butsch et al., 2022), EyaGal4 (Leatherman and DiNardo, 2008), UASp-FRT-H2A-eGFP-PolyA-FRT-H2A-mCherry (Kahney et al., 2021), tubGal80^AID (McClure et al., 2022) (BDSC, #92470; FBst0092470) and UAS-VCP^K2A (Chang et al., 2021).

Single male flies were placed in a vial with two or three virgin females shortly after eclosion. Males were transferred to a fresh vial with new virgin females 2-3 days later. The presence or absence of progeny was scored after each mating. Males that failed to produce progeny in both matings were scored as infertile. Males that were able to produce progeny in both matings were scored as fertile. Flies were kept at 25°C on standard cornmeal/agar food for all matings.

Third instar larvae were washed several times in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄) to remove food from larvae and prevent potential imaging artifacts. After washing, larvae were placed on a 4% agarose gel pad on a glass microscope slide and subsequently immobilized by firmly pressing a glass coverslip down on the larvae. Imaging was performed on a Leica M165 fluorescence stereomicroscope equipped with a GFP filter and 488 nm laser.

Testis immunohistochemistry was performed using standard procedures. Testes were dissected in PBS and then immediately fixed in 4% paraformaldehyde for 20 min. Testes were washed three times in PBT (0.1% Tween-20 in PBS), then incubated in blocking buffer (3% bovine serum albumin in PBS) for at least 1 h at room temperature. Testes were incubated with the primary antibody diluted in blocking buffer containing 2% Triton X-100 overnight at 4°C. The next day, testes were washed five times with PBT prior to applying the secondary antibody. Testes were incubated with the secondary antibody for at least 3 h at room temperature in the dark. After the secondary antibody solution was removed, testes were washed five times with PBT; 1 µM Hoechst 33342 (Invitrogen, H21492) was added to the first wash to stain DNA. Testes were mounted in Vectashield antifade mounting medium prior to imaging.

Phase-contrast imaging was performed on live testis squashes. Testes were dissected in PBS then stained with 1 µM Hoechst 33342 and 1:5000 MitoTracker Deep Red (Thermo Fisher Scientific, M22426) for 10 min in the dark at room temperature. Testes were then rinsed in PBS and then mounted in PBS.

The following antibodies were used in this study: rabbit anti-GFP (1:1000; Invitrogen, A21311), mouse anti-GFP (1:200; Invitrogen, A11120), rabbit anti-H2Aub (1:100; Cell Signaling Technology, 8240S), rabbit anti-VCP (1:100; Cell Signaling Technology, 2648), rabbit anti-mCherry (1:100; Invitrogen, PA5-34975), mouse anti-fibrillarin (a kind gift from Pat DiMario, Louisiana State University, LA, USA), rat anti-Pol II S2p (1:1000; Millipore, 04-1571, clone 3E10), mouse anti-LamC (1:100; Developmental Studies Hybridoma Bank, LC28.26), goat anti-rabbit 488 (1:500; Invitrogen, A11034), goat anti-rabbit Cy5 (1:500; Invitrogen, A10523), goat anti-mouse 488 (1:500; Invitrogen, A11001), goat anti-mouse Cy5 (1:500; Invitrogen, A10524) and goat anti-rat 568 (1:500; Invitrogen, A11077).

Images were acquired using an inverted Leica SP8 confocal microscope, equipped with a 40× oil immersion objective (NA 1.30), 40× phase contrast objective (NA 0.80) and a white-light laser. Images were processed using Leica LAS X software, and quantifications were performed using Fiji (NIH) on 8-bit images prior to adjusting brightness or contrast.

We used Fiji (NIH) to quantify VCP-GFP intensity in the nucleus and cytosol of germ cells at various stages of spermatogenesis (see below for stage identification information). Briefly, we outlined the nucleus and a small region within the cytosol of five cells per stage per testis (see figure legends for total cell and testis counts) and measured the mean fluorescence intensity. VCP-GFP was imaged using a 488 nm excitation laser and a Hybrid detector set at 500-550 nm. The excitation laser was set at 2.5% for all testes imaged with gain left at its default level, and, thus, GFP intensity can be reliably compared between genotypes. For presentation purposes, the mean intensity of the control group was set to one. We used these same methods to quantify VCP-GFP intensity in the nuclear interior and at autosomes by outlining the nuclear interior (Hoechst negative) and Hoechst-positive chromatin. In this case, the mean intensity of Hoechst-positive chromatin was set to one for presentation purposes.

We quantified Pol II S2p intensity at autosomes using Fiji by outlining autosomes and measuring the mean intensity. For presentation purposes, the mean intensity of control autosomes was set to one.

Germ-cell staging was performed primarily based on chromatin morphology. Spermatocytes were identified based on chromatin features described by Cenci et al. (1994). Spermatids were identified based on chromatin and mitochondrial features described by Fabian and Brill (2012).

Total protein was extracted from 40 pairs of testes per genotype. Testes were dissected in NP-40 lysis buffer containing protease inhibitors (6 mM Na₂HPO₄, 4 mM NaH₂PO₄, 1% NP40, 150 mM NaCl, 2 mM EDTA, 50 mM NaF, 0.1 mM Na₃VO₄, 4 µg/ml leupeptin, one Roche cOmplete^TM protease inhibitor tablet per 10 ml, pH 7.4). After dissections were complete, lysis buffer was removed, and testes were further lysed in 5× SDS sample buffer (1 M Tris-HCl, pH 6.8, 1 mM DTT, 20% SDS, 60% glycerol, Bromophenol Blue) using a sterile pestle. Samples were boiled at 100°C for 10 min, followed by a brief vortex and centrifugation at 20,000 g for 1 min. Proteins were resolved on a 4-12% Bis-Tris gel and transferred to a nitrocellulose membrane. Membranes were blocked in 4% milk for at least 1 h. After blocking, membranes were incubated overnight with primary antibody at 4°C. The next day, membranes were washed three times for at least 10 min in PBT. Membranes were then incubated with secondary antibody for 1 h at room temperature, followed by washing. Membranes were treated with an ECL chemiluminescent reagent (Bio-Rad) and proteins were visualized on a Bio-Rad ChemiDoc imaging system.

The following antibodies were used in this study: rabbit anti-H2Aub (1:2000; Cell Signaling Technology, 8240S), mouse anti-Actin (1:1000; Invitrogen, MA5-11869), goat anti-rabbit IgG HRP (Invitrogen, 31460) and goat anti-mouse IgG HRP (Invitrogen, 31430).

Total RNA was extracted from 50 pairs of testes per genotype by Trizol extraction. Total RNA was treated with DNase (Promega, M610A) to remove genomic DNA, followed by RNA precipitation with isopropanol and washes with 70% ethanol. Five-hundred nanograms of RNA was used for cDNA synthesis using the iScript cDNA synthesis kit (Bio-Rad, 1708891). qPCR was conducted using PowerUp^TM SYBR^TM Green Master Mix (Thermo Fisher Scientific). Data were analyzed using Design and Analysis software (v2.6; Thermo Fisher Scientific). The 2^−ΔΔCt method was used to estimate the relative changes in gene expression. Primers used for each gene were: Actin 5C forward TTGTCTGGGCAAGAGGATCAG, reverse ACCACTCGCACTTGCACTTTC; Cyclin A forward GAAAGTTCAGGTCCTTCCGTGAC, reverse GGGCACCACATTCGATTTCT; janB forward CTCCGTTTCAAAAATGTTACTCAACCG, reverse CAACATCGGCGCCACG; gdl forward GAACTTCCCGAAAACTTGCAGACAC, reverse CGTTCTCCGCCTGCATCC; Mst35Bb forward GTGGAATGGCATAATTTCCATTTCTGC, reverse GTTCACTGGTGGTGACCTTGC; Mst98C forward GCGGTCCTTGTAGTCCATGC, reverse CTCCGGCACAATCTTCTCCG; Mst87F forward TCCGACTTGTCAAACCGATA, reverse GCACGAAGGGTATCCACAAT; Ubi-p63E forward GGCTAAGATCCAAGACAAGGAA, reverse GAGACGAAGGACCAAGTGAAG; Hsc70-2 forward GAACCAGGTGGCCATGAAT, reverse TCAGGTCCTCCTGTATCTTCTT; Cyt-c-d forward CCTCAAGGACCCGAAGAAATAC, reverse TGACTTGAGGAAGGCAATCAA; HtrA2 forward CTATCCGATGGCAGGACTTT, reverse CCGAAAGATTGTTCACCTGTATG; ACXC forward GGAACACAGTTACTTGAGGGAAA, reverse CATCAGACGGACCTGGTAGTA; kl-3 exon 2 forward CCCGAGCATTTAATAACCACAAG, reverse AACGGACATTATCCTTAGCTTCA; kl-3 exon 6 forward GCTGGATCTAAGAGGTCATTGG, reverse GGCTGAATGTAACACCCGTTAT; kl-3 exon 14 forward TATGTCCATTCAACCTAAAGAATCGTC, reverse CCCATTGCAATTAGATGCTGTT.

VasaGal4; hsFLP.D5, UAS-GFP/UAS-VCP-RNAi; FRT82B, Sce^KO/FRT82B, tubGal80 and VasaGal4; hsFLP.D5, UAS-GFP/UAS-VCP-RNAi; FRT82B/FRT82B, tubGal80 males were generated by standard crossing procedures (Butsch et al., 2022). Males were heat shocked at 37°C the same day they eclosed for 1 h to activate FLP expression and drive recombination. Following heat shock, flies were housed at 25°C on standard agar/cornmeal food until dissection. Testes were dissected and imaged at the indicated time points following heat shock (‘clone induction’).

hsFLP.D5/+; BamGal4/UASp-FRT-H2A-eGFP-PolyA-FRT-H2A-mCherry and hsFLP.D5/UAS-VCP-RNAi; BamGal4/UASp-FRT-H2A-eGFP-PolyA-FRT-H2A-mCherry males were generated by standard crossing procedures. Males were heat shocked at 37°C the same day they eclosed for 1 h to activate FLP expression and drive removal of the H2A-eGFP cassette. Following heat shock, flies were housed at 25°C on standard agar/cornmeal food for 5 days until dissection, antibody labeling, and imaging. We quantified H2A-GFP and H2A-mCherry intensity by outlining testes and measuring the mean GFP and mCherry intensities.

We verified the recombination efficiency by performing PCR using a forward primer against UASp (GCTTGTTGAGAGGAAAGGTTGTGTGCG) and a reverse primer against mCherry (GGTGGTCTTGACCTCAGCGTCG). Primers were annealed at 70°C with a 3.5 min extension time for 30 cycles.

VasaGal4; tubGal80^AID flies were generated by standard crossing procedures and subsequently crossed to UAS-VCP^K2A flies to generate VasaGal4; UAS-VCP^K2A/+; tubGal80^AID/+ males. Males were fed 5 mM auxin or standard food (see above) at 29°C for 10 days prior to dissection and imaging. Flies were transferred to fresh food every 2-3 days.

Information on sample size and statistics is provided in figure legends where applicable. Data normality was tested via the D'Agostino-Pearson test in combination with Q-Q plots prior to performing follow-up statistical analyses using GraphPad Prism software. Statistical tests used to determine significance are indicated in figure legends. The Student's unpaired t-test was used when unpaired data for two groups were normally distributed and standard deviation was equal between both groups. Welch's unpaired t-test was used when unpaired data for two groups were normally distributed, but standard deviation was not equal. The Mann–Whitney U-test was used when unpaired data for two groups were not normally distributed. The paired t-test was used when paired data for two groups were normally distributed and standard deviation was equal. The Wilcoxon matched pairs signed rank test was used when paired data for two groups were not normally distributed. The Brown–Forsythe ANOVA with Dunnett's multiple comparisons test was used when there were more than two groups with normally distributed data, but standard deviation was not equal between the groups.

We thank Xin Chen (Johns Hopkins University), Minx Fuller (Stanford University), Doug Harrison (University of Kentucky), Tzu-Sang Kang (National Tsing Hua University, Taiwan) and Pat DiMario (LSU) for generously providing reagents needed to perform this study. We also thank members of the Bohnert and Johnson labs at LSU and members of T.J.B.’s dissertation committee for providing helpful comments regarding the manuscript. Lastly, we thank the LSU Shared Instrumentation Facility and the LSU Genomics Facility for providing equipment needed to complete this study, including a Leica SP8 confocal microscope and a qPCR machine, respectively.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201557.reviewer-comments.pdf.