In the course of our investigations into the physiological functions of mammalian Fbxo7, we generated mice that are either heterozygous or homozygous for an allele of Fbxo7 containing a LacZ insertion between exons 3 and 4 of Fbxo7 (Randle et al., 2015; Patel et al., 2016). This insertion severely disrupts expression of all Fbxo7 isoforms but does not completely abolish it (Randle et al., 2015), and thus the phenotype(s) of the hetero- and homozygous animals can, respectively, be ascribed to moderate or severe under-expression of Fbxo7. In maintaining the colony of LacZ-transgenic animals, we observed that homozygous Fbxo7^LacZ/LacZ males never sired any offspring, while heterozygous males and all genotypes of female were able to produce litters. In heterozygous crosses, there was a mild deficit of homozygous offspring (Figure 1A), suggestive of a small degree of embryonic lethality in this genotype.

Initial characterization showed a significant reduction in mean testis weight for the Fbxo7^LacZ/LacZ compared to heterozygous and WT males (92.4 mg vs 108.2 and 107.7 mg; Figure 1B), indicative of abnormal testis development. Strikingly, there were virtually no mature sperm in the lumen of the epididymis of the Fbxo7^LacZ/LacZ males (Figures 1C–F), demonstrating that these males are sterile due to azoospermia. A very few residual sperm were retrieved from dissected epididymides of Fbxo7^LacZ/LacZ males, with a total yield of fewer than 1,000 cells per cauda epididymis, compared to a normal count of around 10⁸ sperm cells per cauda. The residual sperm were all grossly misshapen, and a high proportion of cells showed abnormal compression of the rear of the sperm head. Heterozygous Fbxo7^LacZ/+ males also showed a slight increase in the frequency of abnormally shaped sperm, with the most severely deformed sperm resembling the homozygous phenotype (Figure 1G). Using a newly developed image analysis program for sperm morphometry (Skinner et al., 2019), we determined that the remaining sperm from Fbxo7^LacZ/LacZ males showed an average 14% reduction in cross-sectional area, with high morphological variability (Supplementary Figures S1, S2).

The relatively small magnitude of the change in testis weight suggested that any germ cell abnormality was likely to only affect later stages of germ cell development. To characterize the stage of germ cell loss in Fbxo7^LacZ/LacZ males, we carried out an initial assessment by FACS to see if there was any gross defect in meiotic progression (Figure 1H). There was no significant alteration in the ratio of cells containing 1C, 2C or 4C DNA content, respectively, representing haploid round spermatids, spermatogonia and primary spermatocytes, indicating no cell loss prior to the onset of spermatid elongation. Elongating and condensing spermatids/spermatozoa often appear as lower than 1C DNA content in FACS staining, but cannot be reliably quantified due to their high variability in staining parameters (Simard et al., 2015).

Histological examination of adult testes using hematoxylin and eosin staining (H&E) showed limited gross changes in testis structure (Figure 2 and Supplementary Figure S2). In both Fbxo7^LacZ/LacZ males and wild type (WT) males, pre-meiotic, meiotic and post-meiotic stages of germ cell development were all present in the testis parenchyma (Figures 2A,B). However, in Fbxo7^LacZ/LacZ testes, very few tubules showed sperm heads adjacent to the lumen (Figures 2B,C), suggesting that germ cells are lost prior to spermiation. Instead, testes from these males contained tubules with no (or virtually no) elongating cells, but where the first layer of spermatids was still round. These are tubules in the first half of the seminiferous cycle but where the late elongating cells have been lost. In these tubules, sperm heads were observed lying deep within the Sertoli cell cytoplasm near to the basement membrane, often in quite dramatic “graveyards” containing multiple cells in advanced stages of karyolysis (Figures 2D–F). Strikingly, however, we did not observe any formation of multinucleate symplasts or any sloughing of degenerating cells into the lumen (note also the lack of sloughed cells in the epididymis in Figure 1F).

Since the normal fate of arrested germ cells is apoptosis followed by either phagocytosis or cell sloughing, we used fluorescent immunohistochemical staining for caspase-2 and LAMP2 (Lysosome-associated membrane protein 2) to trace these processes. Caspase 2 is an apical caspase implicated in stress-mediated germ cell death (Zheng et al., 2006; Lysiak et al., 2007; Johnson et al., 2008), while LAMP2 labels late stage phagolysosomes. In this experiment, we also used fluorescently labelled peanut agglutinin (PNA) to label the acrosomes, allowing for more detailed tubule staging (Figures 2G–K). This showed that the cells in the “graveyards” were most prominent at tubule stage VI, and were positive for both LAMP2 and caspase-2. Lower-level caspase-2 staining was sometimes visible at this stage in spermatids located further from the basement membrane. We hypothesize these latter cells to be in the process of engulfment. The few remaining condensing spermatids near the lumen were still caspase-2 negative. Thus, the spermatids in the “graveyards” are apoptotic cells that have been phagocytosed by the Sertoli cells. We attempted to distinguish cells in later stages of apoptosis by immunohistochemical staining for caspase 3; however, this gave negative results in WT and mutant samples (data not shown). Activation of apical caspases independently of effector caspases is a known alternative cell death pathway in Drosophila germ cells, but has not yet to our knowledge been described in mammalian germ cells (Yacobi-Sharon et al., 2013).

In this experiment, we also noted occasional spermatids at earlier tubule stages (e.g., stage IV, Figure 2L) that appeared to be mis-localized, appearing next to the basement membrane, outside the peripheral ring of spermatogonia. Sperm heads are very rarely seen in this location in wild type testes unless they have been phagocytosed; however, these cells were generally negative for both caspase-2 and LAMP2, indicating that they were not yet apoptotic. Since LAMP2 only labels later stages of phagocytosis, we cannot exclude the possibility that these cells were in early stages of phagocytosis, and that phagocytosis in Fbxo7^LacZ/LacZ testes may precede the induction of apoptosis.

We used periodic acid/Schiff/Hematoxylin (PAS-H) staging to quantify the onset of aberrant localisation of the condensing spermatids in the Fbxo7^LacZ/LacZ testes (Figures 3A–H). Here, the PAS staining labels the acrosome, allowing detection of spermatid location and staging. Only very light hematoxylin counterstaining was used to avoid obscuring the PAS signal. For maximal sensitivity in detecting the earliest stages of disorganization of the seminiferous epithelium, we scored any tubule with even a single spermatid observed at the basement membrane (i.e., appearing to be outside the Sertoli cell tight junctions) as positive.

We observed that mis-localization of late stage spermatids in the Fbxo7^LacZ/LacZ testes initiated as early as spermatid step 13–14 (tubule stage I–II), with the proportion of affected tubules apparently peaking at spermatid step 15 (tubule stage IV). From tubule stage VI onward, the phagocytosed spermatid heads were barely visible by PAS-H, most likely due to digestion of the epitopes detected by the PAS stain, and thus the apparent drop-off after stage IV is a technical artifact (note that dead cells at this stage remained visible via H&E and immunostaining; see Figures 1, 2). This contrasts with the immunostaining data where the mis-localized cells prior to stage VI were LAMP2 and CASP2-negative, but the phagocytosed cells at stages VI-VIII were strongly LAMP2 and CASP2 positive. The two experiments thus probe different aspects of the phenotype: mis-localization followed by apoptosis. Complete data from the PAS-H cell counting are supplied as Supplementary Table S2.

The basis for sterility in ntc-mutant null flies is proposed to be the loss of a stabilizing interaction with DmPI31, leading to reduced proteasome activity (Bader et al., 2011). To address whether this relationship is conserved in spermatogenesis in mice, we tested the expression of Fbxo7 and PI31 in lysates made from testes of mature males (Figure 4A). As expected, the presence of the Fbxo7^LacZ allele caused dose-dependent decreases in Fbxo7 expression, as seen by both Western blot and qRT-PCR (Figures 4A,B). We note that for homozygous Fbxo7^LacZ/LacZ mice, expression of both mRNA and protein for Fbxo7 was reduced by approximately 94%. This is partial effect on Fbxo7 mRNA expression in testes is comparable to the that seen in liver (93%), cerebellum (80%), bone marrow (78%), and spleen (60%) (Randle et al., 2015).

PI31 protein levels were significantly reduced by 39% in adult Fbxo7^LacZ/LacZ testes (Figure 4A), while a 23% reduction in mRNA levels was not statistically significant (Figure 4C). The fact that PI31 protein levels show a more pronounced decline than mRNA levels suggests that Fbxo7 has a role in stabilizing PI31 protein levels in adult testes. We therefore characterized the developmental profile of both proteins in normally developing wild type testes. Both Fbxo7 and PI31 were weakly detected at all ages by Western blot, indicating widespread low-level expression in the testis. Both, however, also showed strong upregulation between postnatal day 15 and day 21, concurrent with the first appearance of haploid spermatids in the testis (Figure 4D). PI31 was further upregulated between day 21 and adult testes, consistent with increased expression in later stage elongating/condensing spermatids. Finally, we examined Fbxo7 and PI31 levels in Fbxo7^LacZ/LacZ testes at days 19 and 24, i.e., at time points before the onset of germ cell mis-localization and loss. At 19 days of age, mutant testes showed a 64–92% reduction, while at 24 days, there was a ∼30% reduction in PI31 protein levels compared to wild type testes (Figure 4E). We conclude that Fbxo7 is required for PI31 stability in earlier stage germ cells.

To test whether the observed reduction in PI31 levels in adult testes led to decreases in proteasome activity, we conducted proteasome activity assays on whole testes from WT, heterozygous and homozygous Fbxo7^LacZ males. However, no reduction in proteasome activity was detected (Figure 4F). Consistent with these data, we observed no changes in the levels of core proteasome subunits α4 or β6 among the different adult WT and mutant testes by Western blot analysis (Figure 4A), indicating stable overall levels of proteasomes. Our attempts to measure PI31 levels and to conduct similar proteasome activity assays on elutriated cell populations were inconclusive due to the poor recovery of later stage spermatids from mutant testes (data not shown). These data indicated there were no major alterations in the overall levels of proteasome activity in testes from Fbxo7 mutant males.

Since Fbxo7 can form part of an E3 ubiquitin ligase, we also assayed for the levels of Skp1, the adaptor protein which recruits F-box proteins into SCF-type E3 ubiquitin ligases. Skp1 levels were unchanged, indicating that other SCF-type E3 ubiquitin ligases would be unaffected in mutant testes. Finally, we also examined levels of cleaved caspase 3 and the pro-apoptotic mediator Bax, both of which are implicated in germ cell death during the first wave of spermatogenesis (Russell et al., 2002; Said et al., 2004). We observed no changes in Bax levels between genotypes, and no expression of cleaved caspase 3, which was consistent with our inability to detect cleaved caspase 3 by IHC staining (data not shown). These data suggest that classical apoptotic pathways were not engaged during germ cell death in Fbxo7 mutant testes.

During spermatogenesis, in addition to the constitutive proteasome, alternative proteasomes are co-expressed. PA200-capped spermatoproteasomes promote the degradation of acetylated histones, which enables their removal from DNA and replacement with protamines, for enhanced nuclear compaction into the spermatid head (Gaucher et al., 2010; Qian et al., 2013). They contain an alternative α4-type proteasome subunit, α4s/PSMA8, a testis-specific subunit, which replaces its 20S counterpart, and enables the recruitment of an alternate lid, PA200. A second proteasome, known as the immunoproteasome, is also expressed and has alternate β-subunits, β1i, β2i and β5i, and a different regulatory 11S lid (Qian et al., 2013). As sperm differentiation requires major cellular remodeling and volume reduction, these alternate proteasomes are thought to play crucial roles in fashioning this specialized cell form (Zhong and Belote, 2007; Kniepert and Groettrup, 2014; Rathke et al., 2014).

Although spermatoproteasome activity cannot be directly assayed independently of total proteasome activity, the spermatoproteasome has a key role in histone degradation during nuclear elongation (Kniepert and Groettrup, 2014). We therefore stained WT and mutant testes for LMP7 (β5i), a component of the immunoproteasome and spermatoproteasome which is not present in normal proteasomes, to determine whether this was altered by Fbxo7 deficiency (Figure 5A), and for histone H3 to determine whether the dynamics of histone removal was perturbed in the knockout males (Figure 5B). This showed no alteration in LMP7 localization in the mutant, with nuclear LMP7 signal being specific to stage IX-XII spermatids in both genotypes. There was also no delay in histone removal in the mutant, with all histone H3 signal being removed from the nucleus by the end of stage XII in both wild type and Fbxo7^LacZ/LacZ testes. Taken together with the overall proteasome assay data shown above, we conclude that both LMP7-containing proteasomes and normal proteasome activity are apparently unaffected by Fbxo7 deficiency in mouse testes.

Finally, while unfortunately, the available Fbxo7 antibodies do not work for immunohistochemical (IHC) staining in mouse testes, we investigated the spatial distribution of PI31 to determine whether this was consistent with a role in nuclear or cytoplasmic remodeling of spermatids. In wild type testes, PI31 was present in the cytoplasm of most cell types, becoming significantly stronger in the cytoplasm of late condensing spermatids from stage V onward and being retained into the residual bodies shed at stage VIII. In addition to the stage-specific cytoplasmic signal, PI31 also showed nuclear staining specifically in wild type elongating spermatids from stages IX through to XII (Figure 6). Thus, there is the potential for Fbxo7 and/or PI31 to regulate proteasomes during spermatid nuclear and/or cytoplasmic remodeling.

Spermiogenesis is a multi-step process that transforms morphologically simple round spermatids into highly specialized mature sperm. It occurs in four successive phases, namely; (a) nuclear elongation and replacement of histones by transition proteins in tubule stages IX to XII, spermatid step 9–12), (b) migration of condensing spermatids into Sertoli cell crypts, protamination of sperm chromatin and cytoplasmic reduction by ∼50% in tubule stages I to VI, spermatid steps 13–15, (c) migration of the mature spermatids to the tubule lumen in tubule stage VII, spermatid early step 16, (d) spermiation, the release of fully formed sperm at tubule stage VIII, spermatid late step 16. In Fbxo7^LacZ/LacZ males, we observe a peculiar and very specific phenotype consisting of mass phagocytosis of condensing spermatids, occurring after nuclear elongation and prior to migration of spermatids back to the tubule lumen. This coincides with the cytoplasmic remodeling of the spermatids during steps 13–15, equivalent to the individualization stage of Drosophila sperm development. The requirement for Fbxo7 at this stage thus appears to be conserved from fruit flies to mammals.

The nature of the sterility phenotype in Fbxo7^LacZ/LacZ males is to our knowledge unprecedented in a mammalian system. Various null mutants with defects in cytoplasmic remodeling have previously been described, including Sept4, Spem1, Capza3, and Ube2j1 mutants. These all show either no phagocytosis or only limited phagocytosis of elongating spermatids during the first half of the cycle, followed by spermiation failure and spermatid retention at the lumen into stage IX and beyond (Kissel et al., 2005; Zheng et al., 2007; Geyer et al., 2009; Koenig et al., 2014). Further mutants have been described that show a step 13 block to spermatid development without defects in cytoplasmic remodeling, including the Bclw and Brd7 null mutants. In these mice, the arrested step 13 spermatids degenerate while still near the tubule lumen, followed by phagocytosis of large symplasts and other cell debris (Russell et al., 2001; Wang et al., 2016). In stark contrast to both of the above types of mutant, the Fbxo7^LacZ/LacZ males showed complete phagocytosis of all developing spermatids with no detectable symplast formation, sloughing of degenerating cells into the lumen, or spermiation failure.

In Fbxo7^LacZ/LacZ testes, mis-localized condensing spermatids are visible by PAS-H staining at the basement membrane from tubule stage ∼I-II onward, and by stage IV almost 100% of tubules have mis-localized cells. At these early stages, however, mis-localized spermatids are negative for caspase-2 and LAMP-2, and retain their acrosomes (i.e., they are stainable by PAS), suggesting that they are not yet apoptotic and/or that phagocytic degradation has not yet initiated. By stage VI, however, the cells have lost their normal orientation, become positive for caspase 2 and LAMP-2, and karyolysis has initiated.

One possible scenario is that the early stages of mis-localization represent an abnormal deepening of the Sertoli cell crypts in mutant testes, and that the condensing spermatids have not yet been phagocytosed at this point. The trafficking of spermatids into and out of Sertoli cell crypts is governed by dynein-coupled motion of a specialized adherens junction complex between germ cell and Sertoli cell, known as the apical ectoplasmic specialisation (AES). If the AES is dissolved prematurely, the Sertoli cell may be unable to eject the spermatids from the crypts, leading ultimately to their death and phagocytosis.

An alternative scenario is that the phagocytosis initiates at stage I-II due to defects in germ cell remodeling, but that the engulfed cells remain alive and non-apoptotic for a short period after being phagocytosed. That is, the developing spermatids are “eaten alive” some time before finally dying. In this case, this would represent death by primary phagocytosis or “phagoptosis,” a relatively newly identified manner of cell elimination (Brown and Neher, 2012). Phagoptosis of elongating spermatids in Fbxo7^LacZ/LacZ testes would explain the absence of symplasts and the lack of sloughing of dead cells into the lumen, since in the case of phagoptosis, the eliminated cells would remain alive until they have already become trapped inside phagocytic vacuoles. Future studies to distinguish between these models using electron microscopy, in conjunction with immuno-labeling to resolve the intracellular machinery, are planned.

In flies, ntc has been shown to be important for caspase activation and sperm individualization through enhancing DmPI31 stability, which acts as a proteasome activator. However, transgenic restoration of PI31 levels in ntc testes restored caspase activity but did not restore individualization complex assembly, suggesting that nutcracker and DmPI31 may both be necessary for sperm production (Bader et al., 2011). We and others have previously shown that the stabilising interaction between Fbxo7 and PI31 is conserved in many different cell types (Kirk et al., 2008; Bader et al., 2011; Shang et al., 2015; Merzetti et al., 2017). Consistent with this, we show here that PI31 protein levels are reduced in adult Fbxo7^LacZ/LacZ testes. Importantly, PI31 protein levels are more strongly reduced than mRNA levels, indicating that this is a stabilizing effect of Fbxo7 on PI31 at the protein level.

The reduction in PI31 levels in Fbxo7^LacZ/LacZ testes is also seen in juvenile testes at days 19 and 24 days post-partum. At postnatal day 19, the most advanced germ cells are at the early round spermatid stage, while at day 24, the most advanced germ cells are early elongating (not condensing) spermatids. Thus, the reduction in PI31 precedes the appearance of condensing spermatids in the testis and is not simply a secondary consequence of the loss of late stage condensing spermatids in mutant testes.

In wild type testes, we show that PI31 and LMP7 are both present in the nucleus of step 10–12 spermatids, indicating that alternate proteasomes play a role in nuclear elongation. PI31 then shifts to the cytoplasm of step 13–16 spermatids at tubule stages I-VIII. The loss of germ cells in the knockout is thus coincident with this shift in PI31 localization from the nucleus to the cytoplasm. This is especially intriguing in the light of recent work showing that PI31 acts as an adaptor to facilitate proteasome transport in axons (Liu et al., 2019). One could imagine a role for PI31 in the re-localization of proteasomes in the remodeling cytoplasm. The spermiogenesis phenotypes could arise due to insufficient levels of PI31 causing deficiencies in localized requirements for proteasomes during remodeling but leaving overall proteasome activity intact.

Although we could not directly measure spermatoproteasome activity, whole-testis proteasome activity showed no changes in Fbxo7^LacZ/LacZ testes. Other proteasome-related knockout males (PA200, PA28γ) have defects in multiple stages of germ cell development, both pre- and post-meiotic. The post-meiotic phenotypes of these knockout models include delayed histone replacement during nuclear remodeling and delayed spermiation, neither of which was seen in Fbxo7^LacZ/LacZ testes (Qian et al., 2013; Huang et al., 2016). Moreover, unlike Fbxo7^LacZ/LacZ males, even the PA200/PA28γ double knockout males were able to produce substantial numbers of morphologically normal sperm in their epididymis (Huang et al., 2016). A knockout of the spermatoproteasome-specific subunit PSMA8 has recently been shown to lead to meiotic abnormalities and early spermatid arrest [Gómez Hernández et al., unpublished data, preprint¹ ], unlike the late stage spermatid loss described in the present study.

Taken together, our data shows that the Fbxo7^LacZ/LacZ male sterility phenotype differs from all other existing mouse knockouts related to proteasome function at both the histological and molecular levels. Hence, the Fbxo7 sterility phenotype appears not to be due to a generalized insufficiency of proteasome activity, although it is plausible that localized proteasome function in the cytoplasm is necessary during spermiogenesis, and our experiments have not addressed this.

Fbxo7 is required for PINK1/Parkin-mediated mitophagy, a process that requires the fragmentation and engulfment of depolarized regions of the mitochondrial network (Burchell et al., 2013), and interestingly, both the nutcracker and parkin null flies show male sterility, with a specific defect during sperm individualization (Greene et al., 2003; Bader et al., 2010). In the parkin null mutant in Drosophila, a specialized mitochondrial aggregate present in insect sperm, known as the Nebenkern, failed to form, and spermatids failed to individualize suggesting that rearrangement of mitochondria is necessary for individualization. However, the mouse parkin null mutant is fertile with no known effects on germ cell development (Itier et al., 2003), and thus the sterility of Fbxo7 mutant males is unlikely to relate directly to its interaction with Parkin.

One possible explanation is that Fbxo7 is more generally involved with specialized cytoplasmic remodeling. In a similar vein to the sperm maturation defects, Fbxo7 has also been shown to be required during the final maturation steps of erythrocytes. We previously reported Fbxo7^LacZ/LacZ mice are anemic due to delayed mitophagy and defects in exiting cell cycle (Randle and Laman, 2016). Importantly, during this maturation step, macrophages in erythroblast islands phagocytose the shed organelles from maturing reticulocytes (Geminard et al., 2002; Zhang et al., 2015; Ovchynnikova et al., 2018), a process requiring the coupling of the autophagy and exocytosis pathways (Mankelow et al., 2015). Could this coupling be coordinated by Fbxo7? If so, then in a testis context, Fbxo7 may enable the fragmentation and isolation of portions of the spermatid cytoplasm to allow phagocytosis by Sertoli cells. In the absence of Fbxo7, failure to correctly package spermatid cytoplasm for elimination could instead lead to wrongful engulfment of complete cells and phagoptotic cell death.

As a third alternative but non-exclusive possibility, we note that the dead cells at tubule stage VI were strongly positive for caspase 2. TRAF2, a target of Fbxo7 ubiquitination (Kuiken et al., 2012), has recently been shown to bind to active caspase 2 dimers and ubiquitinate it to stabilize the activated complex (Robeson et al., 2018). Consequently, Fbxo7 deficiency could lead to over-activity of TRAF2 and subsequently signal the activation of caspase 2 precipitating germ cell death. Identifying the substrates of Fbxo7 underlying the unique phenotypes reported here is an area of future research.

Mice used in this study are Fbxo7^LacZ mice (Fbxo7^{tm 1a(EUCOMM)Hmgu} on a C57BL/6J background) and experiments involving them were performed in accordance with the United Kingdom Animals (Scientific Procedures) Act 1986 and ARRIVE guidelines. Mice were housed in individually ventilated cages with unrestricted access to food and water, and 12-h day-night cycle. Animal licenses were approved by the Home Office and the University of Cambridge’s Animal Welfare and Ethical Review Body Standing Committee. Experiments were performed under the Home Office licenses PPL 80/2474 and 70/9001 (HL).

Testes were fixed in Bouin’s fixative at 4°C overnight and embedded in paraffin. Sections were subjected to standard methods of hematoxylin/eos/in or periodic acid/Schiff/hematoxylin staining for histological examination. For immunohistochemical studies, sections were de-paraffinised in xylene and rehydrated through a graded ethanol series prior to blocking and antibody staining steps. Details of primary and secondary antibody concentrations are given in Supplementary Table S1. Antibody-stained sections were counterstained with DAPI and visualized via epifluorescence (PI31, Histone H3) or confocal fluorescence microscopy (Caspase 2, Lamp-2, LMP7). In some experiments, fluorescently conjugated peanut agglutinin from Arachis hypogaea (PNA-lectin) was included during the secondary antibody incubation step to visualize acrosomal morphology and facilitate staging of seminiferous tubules. Antibody validation and negative controls are shown in Supplementary Figure S3.

Sperm were collected from two Fbxo7^LacZ/LacZ, three Fbxo7^LacZ/+, and two wild type males. The vasa deferentia and caudae epididymides were dissected from each animal, and the contents extracted into 1 mL PBS. Sperm were transferred to a microfuge tube, and tissue clumps were allowed to settle. Then sperm were transferred to a new tube and pelleted at 500 × g for 5 mins. The supernatant was removed, and the sperm fixed dropwise with 3:1 methanol-acetic acid. Sperm were again pelleted at 500 × g for 5 mins, and washed in fixative twice more. Samples were stored at −20°C. Fixed sperm nuclei were stained with DAPI and imaged using a 100× objective on a Nikon Microphot SA epifluorescence microscope equipped with a cooled CCD camera and appropriate filters. Images were captured using SmartCapture 2, exported in 8-bit tiff format and analyzed using the automated morphometric software Nuclear Morphology Analysis v1.13.7² (Skinner et al., 2019). Hierarchical clustering was performed on nuclear shapes to group them into morphological categories, and the proportion of cells from each genotype in each category was calculated. The total numbers of nuclei analyzed for each genotype were 453 for Fbxo7LacZ/LacZ, 1225 for Fbxo7^LacZ/+ and 756 for wild type.

Single-cell suspensions from whole testis tissue were pelleted by centrifugation, and resuspended in 1 mL of ice cold 80% ethanol/PBS while vortexing to disperse clumps. The suspended cells were fixed for at least 1 hr at 4^oC. After fixation, cells were collected by centrifugation and washed once in PBS. The washed cell pellet was resuspended in 1 mL of a solution of 50 μg/mL propidium iodide (PI) staining and 10 μg/mL DNase-free RNase, and incubated for 10 min at 37°C, prior to analysis by flow cytometry (Beckman-Coulter, Inc.).

Whole testis tissue were lysed in RIPA buffer (50 mM Tris–HCl pH 7.6, 150 mM NaCl, 1% NP-40, 0.1% SDS, 0.1% Na deoxycholate, 1x protease inhibitors, 1 mM PMSF, 10 mM sodium fluoride, 1 mM sodium orthovanadate) (all from Sigma-Aldrich), and incubated on ice for 30 min with occasional vortexing. Cell debris was pelleted by centrifugation at 16,000 × g for 10 min at 4°C, and the supernatant retained. Protein concentration was determined via 96-well BCA assay (Pierce). Sample concentrations were standardized by dilution with lysis buffer. For Western blot, samples were mixed with equal volumes of 2x Laemmli buffer, denatured (95°C, 5 min), separated via SDS polyacrylamide gel electrophoresis (SDS-PAGE), and transferred onto polyvinylidenefluoride (PVDF) membrane (Millipore) using a semi-dry transfer system (Biorad). Membranes were blocked for 1 h with 5% non-fat, milk powder/PBS-Tween 20 (0.05%) (PBS-T), and then probed with primary antibody overnight at 4°C in 5% non-fat, milk powder/PBS-T. Membranes were washed in PBS-T and incubated with the appropriate HRP-conjugated secondary antibody in 5% non-fat, milk powder/PBS-T followed by further washes, and detection of HRP bound protein using enhanced chemiluminescence (ECL, GE Healthcare) and exposure onto X-ray film (Konica Minolta). Signal was quantified with background correction and normalized using ImageJ software (NIH, Maryland). Antibodies used in this study are provided in Supplementary Table S1.

Tissue was homogenised in 350 μL RLT buffer with β-mercaptoethanol and RNA isolated using RNeasy Plus kit (Qiagen) as per the manufacturer’s recommendations. One μg of mRNA was converted to cDNA using Quantitect reverse transcriptase (Qiagen), and then diluted 1:10 for subsequent RT-qPCR analysis using SYBR Green JumpStart Taq (Sigma) on a CFX Connect Real-Time PCR machine (Biorad). The following primers for Fbxo7 (5′-CGCAGCCAAAGTGTACAAAG; 3′-AGGTTCAGTACTTGCCGTGTG) and Psmf1(PI31) (5′-CAATCATGCCACCTCTCTGA; 3′-CCGTCCTCATACTAG CAGGC) were used. RT-qPCR reactions were as follows: 95°C for 5 min then 45 cycles of 95°C for 30 s, 60°C for 30 s, 72°C for 30 s, followed by melt curve analysis to confirm a single PCR product was made. Relative gene expression was determined using relative standard curve method, data was normalized to three housekeeping genes (Ppai, Gapdh, Actb) as previously described [see (Birkenfeld et al., 2011; Meziane et al., 2011) for primer sequences], and expressed relative to WT levels.

Tubules were staged using periodic acid/Schiff staining to visualize the stages of acrosomal development (Oakberg, 1956; Russell et al., 1993). Every tubule in a complete testis cross-section was staged for three replicate males per genotype, by an observer blinded to the sample identity. Tubules were scored as positive if there were any mis-located elongating spermatid heads detected beyond the Sertoli cell tight junctions, within the outermost layer of nuclei in the tubule, and negative if there were no elongating spermatid heads within this layer. Tubules were also scored for the presence of “graveyards” defined as 2 or more mis-localized elongated spermatids enveloped by a single Sertoli cell. These definitions were chosen to maintain consistency across the seminiferous cycle, and their sensitivity is discussed in the main text.

Assays for proteasome activity were performed using the Proteasome-Glo^TM Chymotrypsin-Like Cell-Based Assay Kit (Promega) according to the manufacturer’s protocol. Briefly, testes from 10 wk old mice were harvested and lysed in buffer (20 mM Hepes pH 7.6, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 2 mM EDTA, 1 mM DTT) using a Dounce homogenizer in 10X the volume/weight of tissue. Protein concentration was measured, and lysates were diluted so that 100 μg of protein in 100 μL volume/well was loaded into a 96-well plate, and samples were plated in triplicate. Where applicable, samples were pre-incubated with MG-132 for 30 min prior to the addition of reagents, with protease inhibitors (Na₂VO₄, NaF, PSMF), or without inhibitors. Samples were equilibrated to RT for 15 min and then 100 μL of assay reagent was added. After a 10 min incubation, luminescence was measured in triplicate.