The E2 Ubiquitin-conjugating Enzyme UBE2J1 Is Required for Spermiogenesis in Mice

Background: The physiological roles of many ER quality control components are unknown. Results: Male Ube2j1 -/- mice are sterile with defects in flagella and acrosome function, and cytoplasm removal in sperm cells. Conclusion: Spermatogenesis is a previously unknown process requiring ER quality control. Significance: ER quality control components might serve as diagnostic or therapeutic targets for male infertility in the future. ABSTRACT ER-resident proteins destined for degradation are dislocated into the cytosol by components of the ER quality control machinery for proteasomal degradation. Dislocation substrates are ubiquitylated in the cytosol by E2 ubiquitin conjugating/E3 ligase complexes.

However, the physiological function of Ube2j1 is poorly defined. We find that Ube2j1 -/mice have reduced viability and fail to thrive early after birth. Male Ube2j1 -/mice are sterile due to a defect in late spermatogenesis. Ultrastructural analysis shows that removal of the cytoplasm is incomplete in Ube2j1 -/elongating spermatids, compromising the release of mature elongate spermatids into the lumen of the seminiferous tubule. Our findings identify an essential function for the ubiquitinproteasome-system in spermiogenesis and define a novel, non-redundant physiological function for the dislocation step of ER quality control.
Components of the endoplasmic reticulum (ER) quality control system recognize proteins no longer wanted in the ER and target them for degradation. During this process, misfolded proteins are extracted from the ER (ER dislocation) and delivered to the cytosol for proteasomal degradation (1). Substrate dislocation is driven by concerted ubiquitylation and deubiquitylation reactions (2). Ubiquitylation involves the action of three enzymes, E1 ubiquitinactivating enzymes, E2 ubiquitin-conjugating enzymes, and E3 ubiquitin ligases. E2 ubiquitin-conjugating enzymes known to be involved in ER dislocation include the yeast Ubc6p orthologs UBE2J1 (UBC6e, NCUBE1) and UBE2J2 (UBC6, NCUBE2) (3,4), tail-anchored proteins in the ER membrane with their catalytic domain facing the cytosol. UBE2J1 is part of the SEL1L dislocation complex, which contains several other factors known to be involved in ER dislocation. These include HRD1, Derlin-2, p97, PDI, BiP, Calnexin, AUP1, UBXD8 and OS9 (5).
In the absence of β2-microglobulin, the heavy chain of the human class I MHC molecule HLA-B27 is ubiquitylated, dislocated and degraded by the concerted action of UBE2J1 and the E3 ligase HRD1 (Synoviolin) (5). In contradistinction, UBE2J1 cooperates with the E3 ligase RMA8 and with Derlin-1 in the degradation of CFTRΔ508 (6). Moreover, UBE2J1 is involved in the degradation of other model substrates for ER dislocation, such as the unpaired TCRα chain, and class I MHC heavy chains in the presence of the HCMV-derived immunoevasin US11 (4,6). Despite accumulating data in tissue culture models on the function and mechanisms of ER dislocation, its physiological function in an organismal context is poorly defined, due to the limited number of model substrates and cell lines investigated and due to the lack of appropriate in vivo mammalian models. To obtain further insight into the physiological role of ER dislocation, we generated a floxed allele of Ube2j1 in mice and analyzed Ube2j1 -/mice. Although Ube2j1 has been repeatedly implicated in the dislocation of various misfolded glycoprotein substrates, mice that lack Ube2j1 are viable and show no gross aberrations in morphology or behavior. On the assumption that ER quality control is essential to development and survival, these results suggest remarkable redundancy in its mode of operation. The main defect we observe in Ube2j1 -/animals, surprisingly, is male sterility due to defects in flagella and acrosome function, and defective removal of cytoplasm from elongating spermatids. Our results thus identify a unique, non-redundant role for Ube2j1 in male germ cell development and define a previously unknown physiological function of ER dislocation.
SV40 assays -SV40 was produced and titrated in simian CV-1 cells and crude virus extracts obtained as described (15). For infection experiments, cells were seeded at a confluency of ~60%. On the day of the experiment cells were washed with PBS and infection medium (RPMI, 3% BSA, 10 mM HEPES pH6.8), then covered with SV40 infection inoculum for 4 hours. Subsequently the inoculum was removed and replaced with MEF medium. Cells were detached with trypsin 16 hours later and fixed for FACS staining of the large T antigen (see above). In vitro Fertilization (IVF) -Cumulus oocytes complexes (COCs) were isolated from 4-5 weeks old mice, superovulated C57BL/6NTac (Taconic, Hudson, NY) mice. Sperm from Ube2j1 -/mice and wild type littermates were isolated from caudal epididymides. After 10 minutes at 37°C/5% CO 2 , sperm cells were video documented, counted and normalized before incubation with COCs. Embryos of different developmental stages were counted once a day for three days. The remaining sperm were spread on glass microscopy slides, fixed for 10 minutes in 100% ethanol and morphologically assessed under a phase contrast microscope. Analysis of male reproductive organs -Testes and epididymides were fixed in Bouin's solution or 4% paraformaldehyde overnight and then processed into paraffin wax. Periodic acid-Schiff (PAS) stain performed following the manufacturers instructions (Sigma 395B-1KT). For immunohistology, tissues were rehydrated and antigen retrieval was performed in citrate buffer (Spring Bioscience, Pleasanton, CA). After blocking in 10% normal donkey serum, primary antibodies were applied in PBS for 2 h at 37°C. Sections were then washed and immunolocalization was performed with DyLightconjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA), stained with 50 µg/ml lectin PNA (Molecular Probes) for 5 min and mounted using ProLong Gold with Dapi (Life Technologies). Antibodies used were anti-VASA (AF2030, goat polyclonal, 0.3 µg/ml); R&D biosystems, Minneapolis, MN) and anti-UBE2J1 ((5), rabbit polyclonal serum, 1/200, produced in our laboratory). Electron microscopy -Testes and epididymides were fixed in 2.5% gluteraldehyde, 3% paraformaldehyde with 5% sucrose in 0.1M sodium cacodylate buffer (pH 7.4), and post fixed in 1% osmium tetroxide in veronal-acetate buffer. Epidydimal sperm cells were embedded in agar prior to further processing. Samples were stained in block overnight with 0.5% uranyl acetate in veronal-acetate buffer (pH6.0), then dehydrated and embedded in Embed-812 resin. 50 nm sections were cut on a Leica Ultracut UCT microtome with a Diatome diamond knife, and stained with uranyl acetate and lead citrate. The sections were examined using a FEI Tecnai Spirit at 80KeV.

Conditional ablation of Ube2j1
We inserted LoxP sites into intron 1 and 3 of the Ube2j1 gene to allow conditional ablation of Ube2j1 (Fig. 1A). Correctly targeted ES cell clones were injected into blastocysts and the resulting male chimeras were crossed to FLPe transgenic females to delete the Neo cassette in Exon 3, which is flanked by Frt sites. Whole body knockout mice were obtained by crossing 'floxed' mice to CMV-Cre transgenic animals. Intercrosses of Ube2j1 +/animals yielded progeny of all possible genotypes as assessed by PCR. MEFs were generated from fetuses on gestation day E13.5. RNA-Seq analysis confirmed the deletion of specifically exons 2 and 3 (Fig. 1B). Analysis of lysates from Ube2j1 -/-MEFs showed the complete absence of the UBE2J1 protein as assessed by immunoblot. Levels of UBE2J1 in Ube2j1 +/-MEFs were reduced by half, indicative of a gene dosage effect (Fig. 1C). The additional bands in the wild type and heterozygous samples with apparent molecular weight of ~30 kDa (asterisk in Fig. 1C) might represent a so-far uncharacterized splice variant of Ube2j1. Indeed, exon junction analysis of the RNA-Seq data supported skipping of Exon 7 (data not shown). Ube2j1 -/mice fail to thrive early after birth Litters from heterozygous parents were born at the expected Mendelian ratios (Fig. 1D), and the weight of P0.5 pups was similar for wild type, heterozygous and knockout animals (Fig. 1E, top panel). At weaning (P21), approximately half of the knockout animals had died (Fig. 1D), the survivors were smaller in size and weighed significantly less than wild-type animals (Fig. 1E (Fig. 1C). We identified milk spots one day after birth in both wild type and Ube2j1 -/mice, excluding the inability to feed as possible reason for the failure to thrive. We did not observe enhanced mortality in Ube2j1 -/animals after weaning. Thus, while the lack of Ube2j1 does not impair development or growth of embryos in the gestation phase, Ube2j1 is important for postnatal survival and growth. Specific components of the ER dislocation machinery are upregulated in the absence of UBE2J1 Removal of key components of a quality control pathway is expected to result in accumulation of misfolded proteins and might therefore activate the unfolded protein response (UPR). However, we did not observe constitutive activation of the UPR in Ube2j1 -/-MEFs as judged by the splicing status of the Xbp-1 mRNA ( Fig. 3A and 3B), cleavage status of ATF6 (Fig. 3C), and protein ( Fig. 2A and B) or transcript levels ( Fig. 2D and data not shown) of genes known to be upregulated upon ER stress, unlike what was described in Derlin-2 -/-MEFs (16). However, when we examined protein levels of ER proteins in MEFs ( Fig. 2A) and in lysates prepared from different tissues (Fig 2B and  C), we found that OS9, SEL1L and EDEM1 levels were increased in Ube2j1 -/-MEFs. The observed increase varied with tissue and/or cell-type examined. For example, OS9 was barely detectable in skin from wild type mice, but present at a much higher level in skin from a knockout mouse. While the increase was modest in the large intestine (~2-fold), changes in OS9 levels in MEFs (Fig. 2B), pancreas, thymus and spleen were considerably higher (~5-10-fold, Fig. 2B). Quantitative analysis of RNA-Seq data from MEFs did not show changes similar to those observed in the immunoblots ( Fig. 2D and data not shown), implying that increased transcription of the corresponding genes is not the reason for this observation. Regulation at the translational level also seems unlikely, because levels of other proteins, including components of the dislocation machinery such as HRD1, XTP3B, Derlin-1 or Derlin-2, were unaffected ( Fig. 2A). The RNA-Seq data confirmed the absence specifically of exons 2 and 3 of Ube2j1 in Ube2j1 -/-MEFs (Fig.  1B). Therefore we can conclude with confidence that similar RKPM values for other genes of interest, including those encoding ER quality control components, reflect similar levels of transcripts. Taken together, ablation of Ube2j1 results in the accumulation of specific proteins of the ER quality machinery, probably by their post-translational stabilization. Ube2j1 -/-MEFs are more susceptible to SV40 infection Having shown upregulation of EDEM1, OS9 and SEL1L in Ube2j1 -/-MEFs, we asked whether there are functional consequences to this remodeling of the ER. ER dislocation components are required for transport of SV40 from the ER to the cytosol (15,17). An increase in the concentration of components involved in ER quality control might therefore facilitate escape of SV40 from the ER. We infected knockout and wild type MEFs with SV40 and quantified infection 20 h later by flow cytometry using antibodies for SV40 large T antigen (TAg). Ube2j1 -/-MEFs were indeed more susceptible to SV40 infection than WT MEFs ( Fig  2E). To test whether enhanced susceptibility of Ube2j1 -/-MEFs to infection was not simply due to an increase in the SV40 receptor, the ganglioside GM1, we measured binding of FITC-labeled Cholera Toxin B subunit (FITC-CTxB), which also uses GM1 to bind to cells (18). FITC-CTxB bound equally well to wild-type and knockout MEFs (Fig. 2F). Wild type and Ube2j1 -/-MEFs were equally susceptible to infection with vesicular stomatitis virus (VSV), a virus that does not require ER quality control components for infection (data not shown). SV40 thus infects Ube2j1 -/-MEFs more successfully than wild type MEFs, in agreement with the increased levels of ER quality control components implicated in virus escape from the ER. These components must therefore be properly integrated in the pathway exploited by SV40. Ube2j1-deficiency does not enhance induction of the unfolded protein response (UPR) UBE2J1 is one of three E2 ubiquitin conjugating enzymes involved in ER dislocation, and was identified as part of the HRD1-SEL1L complex. We reasoned that its absence might impair the ability of cells to cope with ER stress and thus render cells more susceptible to the UPR. We therefore exposed MEFs to tunicamycin, an inhibitor of N-linked glycosylation that potently induces the UPR. We neither observed differences in the extent of Xbp-1 splicing (Fig. 3B), nor in the processing of ATF6 (Fig. 3C) in response to tunicamycin treatment. We also did not see differences in response to other inducers of the UPR, such as depletion of ER-resident calcium stores by thapsigargin or by perturbation of redox homeostasis by DTT (data not shown). Notably, steady-state levels of full-length ATF6 were higher in Ube2j1 -/-MEFs (Fig. 3C). ATF6 is itself a substrate that is degraded in a SEL1Ldependent manner (19). Given that mRNA levels of Atf6 are unaltered in Ube2j1 -/-MEFs (Fig. 2D), it is likely that turnover of ATF6 involves the action of UBE2J1, a component of the SEL1Lcontaining dislocon, possibly by directly transferring ubiquitin onto dislocated ATF6. We conclude that Ube2j1-deficiency does not render MEFs more susceptible to ER stress.

B cell development, immunoglobulin maturation and immunoglobulin secretion are normal in Ube2j1 -/-B cells
The prevailing view is that ER quality control is especially important for tissues and cells with a high secretory load, such as liver, pancreas and B lymphocytes.
However, surviving Ube2j1 knockout mice were seemingly healthy once they reached adulthood. Considering the compensatory mechanism(s) observed in MEFs, we reasoned that a phenotype might be observable only under nonhomeostatic conditions or in cells with a high secretory load. We therefore asked if B cell function is affected in the absence of Ube2j1. Ube2j1 -/mice had similar frequencies of immature (B220 int IgM + ), pre-/pro-(B220 + IgM -) and recirculating (B220 hi IgM + ) B cells in the bone marrow as wild type mice (Fig. 4A). We also observed no difference in frequencies of immature (B220 + AA4.1 + ), mature (B220 + AA4.1 -), follicular (B220 + AA4.1 + CD21 lo CD1d lo ), marginal zone B cell precursors (B220 + AA4.1 + CD21 hi CD1d hi CD23 + ) and marginal zone B cells (B220 + AA4.1 + CD21 hi CD1d hi CD23 -) in the spleen (Fig. 4A). The average frequencies of total B1 (B220 lo CD19 + ), B1a (B220 lo CD19 + CD5 + ), B1b (B220 lo CD19 + CD5 -) and B2 (B220 hi CD19 + ) cells in the peritoneal cavity were comparable between Ube2j1 -/and Ube2j1 +/+ control animals (Fig. 4A). We conclude that B cell development is normal in the absence of UBE2J1. B cells have been used extensively to study the impact of chaperone-dependent protein folding as well as other aspects of glycoprotein quality control, precisely because of their exceptionally high rates of synthesis, assembly and secretion of immunoglobulins ((20) and references therein). Therefore, we analyzed bulk glycoprotein synthesis and turnover in B cell plasmablasts. Spleens from CD19-Cre Ube2j1 fl/fl and CD19-Cre Ube2j1 wt/wt were enriched for B cells by exclusion of CD43 + cells. CD43cells were then stimulated with 1µM CpG-DNA for four days to obtain B plasmablasts. Plasmablasts were metabolically labeled with 35 S-cysteine/methionine and the fate of both intracellular and secreted immunglobulins was followed. Maturation and secretion of immunoglobulin or maturation of the class I MHC molecule H2-K b were indistinguishable between Ube2j1 -/and control mice (Fig. 4B). Imposition of additional ER stress by exposure of plasmablasts to tunicamycin for 5 hours prior to pulse-chase analysis did not show any differences in stability or secretion of immunoglobulins either (Fig. 4C). In the aggregate, UBE2J1 is required neither for B cell maturation nor is it essential for synthesis, assembly or secretion of immunoglobulins by B plasmablasts. The lack of detectable alterations in the overall composition of serum proteins, produced mostly by the liver, is also consistent with this conclusion (data not shown).

Male
Ube2j1-/-mice are sterile In our attempt to breed Ube2j1 -/animals we readily obtained litters from Ube2j1 -/females intercrossed with heterozygous males. In contrast, breeding pairs with male Ube2j1 -/animals (n=6) produced no offspring. When co-housing individual male Ube2j1 -/animals overnight with heterozygous females, vaginal plugs were observed, indicating that Ube2j1 -/males were sexually active, but no pregnancies were observed. To understand the underlying defect of this observation we first analyzed epididymal sperm. The total numbers of sperm cells isolated from epididymides of Ube2j1 -/mice were significantly reduced (Fig. 5A). Since the amount of residual sperm might still be sufficient to confer fertility (21,22), we asked if the remaining sperm cells are functional. We found that the motility of spermatozoa isolated from the epididymis was severely impaired in Ube2j1 -/mice (Fig. 5B, Movie S1) as compared with wild type controls (Fig. 5B, Movie S2), pointing to defects in flagella function. Ube2j1 -/animals had a strikingly increased frequency of sperm cells with aberrant morphologies (91% vs. 26% normal, 4% vs. 59% with abnormal head morphology, 5% vs. 15% acephalic sperm in wild type vs. Ube2j1 -/sperm, respectively; Fig. 5C and D). Further, epididymal sperm preparations from Ube2j1 -/mice contained cell debris (Movie S2). Next, we tested if sperm from Ube2j1 -/mice could fertilize eggs in vitro. We incubated sperm from wild type (Movie S3) or Ube2j1 -/-(Movie S4) mice with eggs from wild type mice. Sperm cells from Ube2j1 -/mice rarely bound to isolated eggs, consistent with abnormal sperm function and were unable to fertilize eggs. One day after IVF, we observed a single 2-cell embryo with sperm from Ube2j1 -/animals (n=3 males, ~200 eggs per male), which did not survive to the 4-cell stage. In contrast, ~54% of the eggs fertilized with sperm from wild type males proceeded to the 2-cell stage one day after IVF (n=3 males, ~200 eggs per male). By day 3.5 after IVF, ~36% of the eggs fertilized with control sperm reached the morula stage and ~17% had developed into a blastocyst. In contrast, all eggs that had been incubated with Ube2j1 -/sperm had died or had not proceeded to the 2-cell stage (Fig.  5E). Given its severely reduced mobility (Movie S2) and impaired ability to bind to the zona pellucida (Movie S4), Ube2j1 -/sperm most probably did not efficiently fertilize the oocytes, although we cannot formally exclude a defect in the sperm's ability to support post-fertilization embryo development. Taken together, male Ube2j1 -/mice are sterile. Their sperm cells are morphologically abnormal, display severely impaired mobility and cannot generate multi-cell embryos in an IVF setting.

Male Ube2j1 -/mice have defects in spermatid development
Considering that Ube2j1 -/spermatozoa are defective and given that UBE2J1 is highly expressed in testicular extracts (Fig. 2C), we next investigated the role of Ube2j1 in spermatogenesis. Immunolocalization of UBE2J1 revealed strong expression in the early maturation phase of elongate spermatids between step 12 and 15 (Fig. 6A), a period in which the ER undergoes massive, dynamic changes, as previously described (23). This expression gradually diminished as spermatids reduced their cytoplasmic content, consistent with the regression of the ER at this time (23,24). The specificity of Male Ube2j1 -/mice are sterile the staining with the anti-UBE2J1 antibody is validated by the absence of staining in Ube2j1 -/testis sections (Fig. 6A, right panel). We did not detect UBE2J1 in meiotic or mitotic germ cells (Fig. 6B). The expression of UBE2J1 in elongate spermatids in conjunction with reduced epididymal sperm number, motility and morphology in Ube2j1 -/mice suggested a defect in the final phases of germ cell maturation. To assess spermatogenesis, we performed periodic acid-Schiff (PAS) stains on testis sections. Mouse spermatogenesis is characterized by a defined set of germ cell associations, by which germ cell stem cells (spermatogonia) differentiate and progress through meiosis and spermiogenesis to produce haploid spermatozoa for fertility. At stage VIII of spermatogenesis, elongate spermatids are released into the tubular lumen in a process known as spermiation (25). Haploid round spermatids, previously located immediately basal to elongate spermatids, then start a dramatic morphological transformation into elongate spermatids from late stage VIII onwards. In this way, only one generation of elongate spermatids is observed at any stage of mouse spermatogenesis (26). In Ube2j1 -/mice, we frequently observed abnormal germ cell associations, with retained elongate spermatids from stage IX onwards (Fig. 6C), indicative of defective spermiation (25). These spermatids are subsequently released into the tubular lumen, or phagocytosed by the seminiferous epithelium (asterisks in Fig.6C). We also observed focal regions of mixed hypospermatogenesis in all Ube2j1 -/mice examined (data not shown), indicating that additional spermatogenic defects may contribute to compromised fertility in these males. The average weights of testes from wild type, heterozygous and Ube2j1 -/males were all similar, although we observed substantially smaller testes in a few knockout animals (Fig. 6D). We conclude that spermatid development is defective in Ube2j1 -/mice as manifested by frequent abnormal germ cell associations. Structural abnormalities in Ube2j1 -/sperm To gain further insight into the defects imposed on spermatids by the lack of Ube2j1, we analyzed the ultrastructure of the testis and epididymal sperm by electron microscopy. Fig. 7A shows stage IX spermatids and retained spermatids (asterisks) in a stage IX seminiferous tubule form the testis of a Ube2j1 -/mouse. Both spermatids retained in the testis (Fig. 7A) and epididymal sperm cells from Ube2j1 -/mice (Fig. 7C, D, F and G) showed the presence of cytoplasm around their acrosome and neck, an abnormality not observed in sperm from wild type testes (Fig. 7B) or epididymides (Fid. 7E and H). This excess cytoplasm contained organelles, such as endoplasmic reticulum and vacuoles surrounded by one or more membranes. The massively condensed ER morphology ( Fig.  7D and G) is reminiscent of the 'radial body' described in (23), which is the typical form of appearance of the ER at steps 14-15 of spermatid development. The 'radial body' denotes a condensed, aggregated, glomerulus-like appearance of the ER at the posterior side of the cell (23). We also observed excess residual cytoplasm in the midpiece of Ube2j1 -/epididymal sperm (data not shown), a pathological form of cytoplasmic droplets (27). Cytoplasmic droplet in mice are a cellular structure of epididymal sperm implicated in sperm motility and an indicator for normal spermiogenesis (28). Excess residual cytoplasm around the midpiece in Ube2j1 -/sperm might confer the instability responsible for the elevated number of acephalic sperm in the sperm smear (Fig. 5C), and defects in flagella function (Movie S2). Acrosome formation and the (9+2) organization of microtubules in the sperm tails seem to be intact. However, the subacrosomal and perinuclear space were often expanded, leading to the detachment of the acrosome from the condensed nucleus ( Fig. 7C and F). Together, the EM data suggest that cytoplasm removal cannot be completed in absence of Ube2j1. This failure of cytoplasmic remodeling initiates a cascade of events that ultimately leads to infertility.

DISCUSSION
The removal and degradation of misfolded or unwanted ER proteins is a process pivotal to maintain homeostasis in the ER. Here, we characterized mice deficient in the ER-anchored E2 ubiquitin-conjugating enzyme UBE2J1, which is part of an E2/E3 ubiquitin ligase complex that marks substrates destined for extraction from the ER and degradation by the proteasome. Gestation proceeds normally in Ube2j1 -/mice and adult animals look normal and show no aberrant behaviors. Nonetheless, approximately half of Ube2j1 -/mice die early in life, and surviving adult Ube2j1 -/mice have significantly reduced body weight, suggesting that there is a critical period after birth for which a proper level of UBE2J1 is required. In the absence of Ube2j1, specific proteins involved in ER quality control, such as OS9, EDEM1 and Sel1L, are upregulated resulting in higher susceptibility to infection by SV40. Why the levels of some but not other components implicated in ER quality control show an increase is unclear. The amount of such components, perhaps not surprisingly, may be controlled dynamically (29,30), according to the immediate needs of the cell. In vivo ablation of Ube2j1 -/may thus reveal the existence of a feedback loop in which activation of Ube2j1 attenuates the levels of OS9, EDEM1 and Sel1L. At present we do not know whether ablation of UBE2J1 directly stabilizes OS9, SEL1L or EDEM1 and proteins in the same functional category, or if, owing to the Ube2j1 deficiency, unfolded proteins accumulate first and then trigger upregulation or stabilization of the latter proteins. Regardless of the exact sequence of events, the observed compensation does not seem to involve the UPR. These findings point to a remarkable robustness and redundancy within the ER quality control system to restore and/or maintain ER homeostasis. Notwithstanding the absence of a constitutive activation of the UPR, it is intriguing that ATF6 is stabilized in Ube2j1 -/cells. As part of the HRD1/SEL1L complex, UBE2J1 might required for turnover of ATF6, consistent with the recent finding that ATF6 is stabilized in absence of SEL1L (19). The most striking phenotype observed in Ube2j1 -/mice is a profound defect in spermatid differentiation, known as spermiogenesis, resulting in male sterility. The massive morphological changes in the course of spermiogenesis include synthesis of the acrosome, chromatin condensation and elongation of the nucleus, formation of the flagellum and remodeling of membrane-delimited organelles. In the final step, called spermiation, remaining cytoplasm is eliminated from the spermatid, intercellular cytoplasmic bridges between spermatids are disrupted, and the strong adhesions between spermatids and the Sertoli cell (ectoplasmic specializations), with which they intimately interact, are removed to facilitate their release into the lumen of the seminiferous tubule (25). UBE2J1 protein was abundantly present in elongating spermatids, specifically from step 12 to 15 of sperm differentiation (31), peaking at step 14. Spermatids at these differentiation steps are largely transcriptionally inactive (32, 33), implying that the transcripts are transcribed earlier, and that translation of the Ube2j1 (and presumably other) transcripts is de-repressed in elongate spermatids (34). Thus, UBE2J1 is likely required at a critical time within the elongating spermatid itself for successful completion of the final steps of differentiation and spermiation. Crossing Ube2j1 fl/fl mice with a germ-cell specific (e.g. Stra8-Cre) or a Sertoli cell-specific Cre-deleter strain (e.g. Amh-Cre) may allow us in the future to unequivocally determine the cell-type(s) that is/are responsible for the defect in spermiogenesis. Pups from Ube2j1 +/parents are born at expected Mendelian ratios, suggesting that in heterozygous mice, Ube2j1 -(haploid) sperm are fully functional. This might be explained by the distribution of cellular contents between cells of a clone derived from one Ube2j1 +/spermatogonium via cytoplasmic bridges (35). Ube2j1 transcripts would thereby be shared between adjacent cells of the syncytium, allowing proper differentiation of both Ube2j1 + and Ube2j1spermatids. Why is spermiation not executed correctly in Ube2j1 -/testes? Microscopic examination showed aberrant spermatozoa and cellular debris in epididymal sperm preparations. Ultrastructural examination of epididymal sperm by electron microscopy revealed excess residual cytoplasm in the head and midpiece region of spermatozoa. These observations are in line with the notion that Ube2j1 is critical for cytoplasmic removal during the last steps of spermatid differentiation. Considerable structural changes during spermiogenesis in different organelles, particularly the ER, have been observed by TEM and SEM (23). The structural metamorphosis of the ER starts from a branched network of tubular structures, which are spread evenly throughout the cytoplasm (step 13), over a markedly condensed, glomerular structure (radial body, step 14), to a completely degenerated structure, often aggregated with other organelles (step 15). At step 16, immediately prior to sperm release, the ER is no longer detectable (23). Levels of UBE2J1 are highest precisely when the ER is massively remodeled for eventual elimination (steps 13 to 16). We hypothesize that Ube2j1 is involved in the aforementioned remodeling process, possibly by marking excess proteins for proteasomal degradation. In the absence of Ube2j1, ER luminal proteins that depend on active UBE2J1 for their removal might be trapped in the ER and block further condensation of the ER. This might prevent the cell from completing the necessary removal of cytoplasm, as evidenced by excess residual cytoplasm in the head reagion, accompanied by detached acrosomes, and around the midpiece, probably causing defects in flagella function. Abnormal spermatids are likely sensed by Sertoli cells eventually resulting in the spermatid retention and ultimately, phagocytosis by Sertoli cells. This failure of spermiation likely explains the starkly reduced sperm number in the epididymis. Although essential for the generation of functional sperm cells, how cytoplasmic removal is accomplished in elongating spermatids and which genes are involved in this process is not well understood. In fact, most genes so far associated with defective cytoplasmic removal encode nuclear proteins, such as transition nuclear proteins (Tnp1, Tnp2, (36)) and the protamines (Prm1, Prm2, (37)), none of which are likely to directly participate in cytoplasm removal. An exception might be LKB1 (Liver kinase B1), a kinase that is predominantly found in the cytoplasm and membrane fractions. Their phenotype with respect to male infertility is strikingly similar to Ube2j1 -/mice. However, unlike in Ube2j1 -/mice, the epididymis is completely filled with cell debris in LKB1 -/mice, and LKB1 is expressed in both meiotic and postmeiotic cells. An intriguing case is SPEM1 (spermatid maturation 1), a protein of unknown molecular function, which is expressed exclusively in elongating spermatids between steps 14 and 16, implying a function in cytoplasm removal and/or spermiation (38). Akin to Ube2j1 -/mice, SPEM1 -/males (but not females) are sterile and their sperm are ~85% immotile because of severe deformations in the head/neck region, caused by retention of cytoplasmic remnants. The ultrastructural characteristics are strikingly similar to our observations in Ube2j1 -/sperm (38). SPEM1 interacts with UBQLN1 (ubiquilin 1) at the manchette of elongating spermatids (39). Although it was postulated that UBQLN1 participates in spermiogenesis, in vivo support for this is still pending. UBQLN1 and its homologue UBQLN2 form a complex with p97/VCP and erasin, both of which are central components of the ER dislocation machinery (40). UBE2J1, UBQLN1 and SPEM1 may function cooperatively during the removal of cytoplasm from elongating spermatids. Taken together, we identify Ube2j1 as a key player in the elongation of spermatids, arguably one of the most extreme examples of structural and morphological reorganization of a cell. The role of Ube2j1 in this context is unique and nonredundant, since Ube2j1's paralogue Ube2j2 cannot compensate for the defect in spermiogenesis. Ube2j1 is, to our knowledge, the first reported gene directly linking the little understood process of cytoplasm removal in elongating spermatids with the ubiquitinproteasome pathway, adding a new layer of regulation to spermiogenesis. Furthermore, our data define a previously unknown physiological function for ER dislocation, a cellular process, which until recently has been characterized mostly in established cell lines using biochemical assays. Examination of ER components in mouse models of male fertility might well prove informative to further elucidate the cellular processes that regulate spermiogenesis. Remarkably, components of the ubiquitinproteasome pathway, including Ube2j1, were amongst the most strongly and significantly differentially regulated genes in human in teratozoospermic sperm (41), suggesting that Ube2j1's role in spermiogenesis may be conserved in humans. Considering that the genetic causes underlying spermiogenic failure in men are still largely unknown, Ube2j1 and other components of this pathway warrant closer examination and might be useful as diagnostic markers for male infertility in humans.