Cell Autonomous and Nonautonomous Function of CUL4B in Mouse Spermatogenesis*

CUL4B ubiquitin ligase belongs to the cullin-RING ubiquitin ligase family. Although sharing many sequence and structural similarities, CUL4B plays distinct roles in spermatogenesis from its homologous protein CUL4A. We previously reported that genetic ablation of Cul4a in mice led to male infertility because of aberrant meiotic progression. In the present study, we generated Cul4b germ cell-specific conditional knock-out (Cul4bVasa),as well as Cul4b global knock-out (Cul4bSox2) mouse, to investigate its roles in spermatogenesis. Germ cell-specific deletion of Cul4b led to male infertility, despite normal testicular morphology and comparable numbers of spermatozoa. Notably, significantly impaired sperm mobility caused by reduced mitochondrial activity and glycolysis level were observed in the majority of the mutant spermatozoa, manifested by low, if any, sperm ATP production. Furthermore, Cul4bVasa spermatozoa exhibited defective arrangement of axonemal microtubules and flagella outer dense fibers. Our mass spectrometry analysis identified INSL6 as a novel CUL4B substrate in male germ cells, evidenced by its direct polyubiquination and degradation by CUL4B E3 ligase. Nevertheless, Cul4b global knock-out males lost their germ cells in an age-dependent manner, implying failure of maintaining the spermatogonial stem cell niche in somatic cells. Taken together, our results show that CUL4B is indispensable to spermatogenesis, and it functions cell autonomously in male germ cells to ensure spermatozoa motility, whereas it functions non-cell-autonomously in somatic cells to maintain spermatogonial stemness. Thus, CUL4B links two distinct spermatogenetic processes to a single E3 ligase, highlighting the significance of ubiquitin modification during spermatogenesis.

Protein degradation via the ubiquitin-proteasome system plays a critical role during mammalian spermatogenesis. Timely removal of outlived proteins is crucial to ensure progression through different phases of spermatogenesis including mitotic, meiotic divisions, and postmeiotic morphogenesis. The ubiquitin-proteasome system selects its targets via a group of E3 ubiquitin ligases, which upon interaction with E2 ubiquitin-conjugating enzymes tag substrates for proteasomal degradation by the 26S proteasome. The largest mammalian E3 ligase family is the CRL (cullin-RING finger ubiquitin ligase) family, which includes eight members. Each cullin family member serves as a molecular scaffold and forms an E3 ligase complex with substrate-recruiting receptors and a linker protein (1). The Cul4b gene, located on the X chromosome, shares extensive sequence homology and functional redundancy with the other CRL4 family member, Cul4a. Both proteins employ DDB1 (DNA damage-binding protein 1) as the linker protein, which in turn recruits common or distinct DDB1-CUL4 associated factors for substrate binding. The DDB1-CUL4-mediated protein modification/degradation is involved in critical cellular processes including DNA replication, DNA repair, cell cycle control, and histone modifications (2).
Male infertility accounts for ϳ40 -50% of infertility cases in humans (3,4). Many factors contribute to male infertility; however, the vast majority of male infertility cases are associated with oligozoospermia (decreased sperm number), asthenozoospermia (reduced sperm motility), and/or teratozoospermia (abnormal sperm morphology). In the mammalian testis, a single pluripotent spermatogonial stem cell (SSC) 2 undergoes several rounds of mitotic divisions followed by meiotic divisions and complex postmeiotic morphogenesis, eventually giving rise to a cohort of mature spermatozoa. We have previously reported that the CRL4 proteins exhibited complementary expression patterns in adult mouse testis, where CUL4A was predominantly detected in primary spermatocytes, whereas CUL4B was highly expressed in Sertoli cells, spermatogonia, and spermatids (5). The absence of CUL4B in spermatocytes is likely due to meiotic sex chromosome inactivation, a transient X chromosome inactivation caused by sex chromosome condensation and subsequent gene silencing (6). However, the absence of Cul4a gene expression in Cul4b-expressing cells suggests that these two family members play distinct and nonoverlapping/nonredundant roles in specific testicular cell populations. We reported previously that loss of Cul4a in null mutant mice led to male infertility because of meiotic defects (5). Here we investigate the role of Cul4b in murine spermatogenesis. Our previous studies showed that Cul4b mutant mouse embryos died in midgestation because of a G 2 /M arrest in extraembryonic tissues, whereas epiblast-specific null mutants survived to adulthood with no overt developmental defects (7). To investigate the role of CUL4B protein in mammalian spermatogenesis, we generated and characterized germ cell-specific as well as global Cul4b mutant mice. We demonstrated that CUL4B is required for proper progression of spermiogenesis and spermatozoa motility and uncovered an unexpected noncell-autonomous role for CUL4B in maintaining the SSC niche.

Experimental Procedures
Mice-Generation of conditional Cul4b mice was described previously (7). The Vasa(Ddx4)-Cre (stock no. 006954), Amh-Cre (stock no. 007915), and Sox2-Cre (stock no. 008454) transgenic mice were purchased from the Jackson Laboratory (Bar Harbor, ME). All mouse colonies were maintained in a barrier facility at Washington University in St. Louis, and all animal experiments were performed in accordance with the institution's regulations under an approved protocol.
Sperm Motility Assay-Cauda epididymides were dissected, pierced, and incubated in EmbryoMax human tubal fluid (EMD Millipore) for 10 min at 37°C in 5% CO 2 to disperse spermatozoa. Sperm motility was determined using a computerassisted sperm analysis system (Hamilton-Thorne Research, Beverly, MA).
Flow Cytometric Analysis-Single testicular cell suspensions and epididymal spermatozoa were prepared as previously described (5). Live spermatozoa were stained with JC1 dye (Life Technologies, Inc.) according to the manufacturer's instructions, followed by flow cytometry on the BD FACScan TM system (BD Biosciences, San Jose, CA). To prepare splenocyte suspensions, spleen removed from one mouse was meshed through a 70-m cell strainer (BD Biosciences) with a syringe plunger and washed with cold PBS (calcium-and magnesiumfree). After a brief spin, splenocyte pellets were resuspended and incubated in 5 ml of red blood cell lysis buffer (Sigma-Aldrich). The remaining cells were washed once in PBS, centrifuged, and processed for marker staining and FACS analysis.
Isolated splenocytes and testicular cells were incubated in blocking solution (PBS with 1% bovine serum albumin and 1% mouse Fc receptor block (BioLegend, San Diego, CA)) for 15 min, followed by incubation with antibodies diluted in blocking solution for 30 min on ice, protected from light. Dilutions for antibodies and fluorescent dyes used were: 1:100 for antimouse GR1-FITC, anti-mouse F4/80-PE, anti-mouse CD11b-APC, and anti-mouse CD45-Pacific Blue, and 1:500 for 7AAD (all antibodies were from AnaSpec Inc., Fremont, CA). Following antibody incubation, the cells were fixed in 4% paraformaldehyde with actinomycin (ACROS Organics, Geel, Belgium) overnight at 4°C and analyzed on a BD LSRFortessa TM cytometer the following day. The data analyses were completed by using FlowJo V10 software (TreeStar Inc., Ashland, OR).
Determination of Sperm ATP Level-Sperm extracts were prepared as previously described (8). The ATP levels were quantified by using the Molecular Probes ATP determination kit (Life Technologies, Inc.) following the manufacturer's instructions. Bioluminescence was measured on a GloMax microplate reader (Promega, Madison, WI), and a standard curve for a series of ATP standard solutions with concentrations ranging from 10 pM to 2.5 M was generated. The ATP concentration of each sample was determined according to the standard curve, and sperm mass (total protein content) was determined by standard Bradford assay. Normalized ATP levels were calculated as ATP (micromoles)/sperm mass (gram). n ϭ 4 for each genotype in two replicates.
Determination of Sperm Phosphoenolpyruvate (PEP) Level-A commercial fluorometric assay kit was used to determine sperm intracellular PEP level (Cayman Chemical Company, Ann Arbor, MI). Epididymal spermatozoa were collected, lysed, and deproteinated in the metaphosphoric acid buffer provided in the kit. Protein pellets were reconstituted, and concentration was determined for normalization purposes. Reaction was set up according to the manufacturer's instructions, and fluorescence was measured on the Synergy TM H4 hybrid microplate reader (Ex ϭ 535 nm, Em ϭ 590 nm). A PEP standard curve was plotted using a series of PEP solutions with concentrations ranging from 0 to 100 M, and the PEP levels of individual samples were determined accordingly. Normalized PEP level was calculated as PEP concentration (M)/sperm mass (mg). Four biological replicates were used for each genotype assayed in duplicate.
Affinity Purification and Identification of Ubiquitinated Testicular Proteins-Seminiferous tubules from control or Cul4b Vasa testes were minced in PBS, and the suspension was pipetted repetitively for 1 min. Large fragments were allowed to settle for 15 min, and the remaining cell suspension was centrifuged at 600 ϫ g for 5 min. Cell pellets were resuspended in 1 ml of PBS. To enrich for germ cells, cell suspension was carefully layered on top of a Percoll gradient in a 15-ml conical tube, composed of 1 ml each of 45, 30, 22, and 15% Percoll solutions. After centrifugation at 600 ϫ g for 25 min, cell population in the lower 22% Percoll fraction was recovered by slow pipetting. This fraction contained largely round and elongated spermatids, with some remaining spermatocytes. Cell pellets were washed three times in PBS and resuspended in 0.2-0.5 ml immunoprecipitation assay buffer (1% Triton X-100, 150 mM NaCl, 20 mM Tris-Cl, pH 7.4, 1 mM EGTA, 1ϫ protease inhibitor mixture; Roche Applied Science). To affinity-purify ubiquitinated proteins, 10 -15 l of p62-derived ubiquitin-associated (UBA) agarose beads (agarose immobilized recombinant ubiquitin-binding domain of p62/SQSTM1 protein; ENZO, catalog no. UW9010) were added to cell lysates and incubated overnight at 4°C on a shaking platform. After removing supernatant containing unbound proteins, affinity-purified complexes were pelleted, washed in PBS three times, and submitted to the Proteomics and Mass Spectrometry Program core facility at Washington University School of Medicine for mass spectrometry analysis (detailed method available upon request). Scaffold (Proteome Software, Portland, OR) was used to validate protein identifications using PeptideProphet and to derive corresponding protein ID probability scores (9). The mass spectrometry data processing pipeline and detailed method for peptide quantification were described previously (10).

Germ Cell-but Not Sertoli Cell-specific Cul4b-null Males Are
Infertile-We previously demonstrated dynamic and complimentary expression patterns for the two Cul4 genes in the mouse testis and reported that loss of Cul4a led to male infertility caused by defects in meiotic progression (5). Because Cul4b is highly expressed in both somatic and germ cells in the testis, we first used the well established Vasa-Cre line (13) to generate germ cell-specific Cul4b mutants. These germ cellspecific mutants are hereafter referred to as Cul4b Vasa , as compared with their wild type congeners, termed Cul4b f/Y . As shown in Fig. 1C and our previous study (5), CUL4B protein is predominantly detected in Sertoli cells, spermatogonia, and spermatids in adult wild type mouse testes. Close examination of wild type seminiferous tubule sections revealed that as spermiogenesis progresses, CUL4B protein gradually localizes to the cytoplasm adjacent to the postacrosomal segment of elongating spermatids, coinciding with the location of the caudala manchette, a transient microtubule-based spermatid structure (Fig. 1B, compare with Fig. 1A, arrows). In the Cul4b Vasa testis, however, CUL4B protein was no longer detectable in germ cells, even though its expression was unaffected in Sertoli cells (Fig. 1D). To further confirm this finding, double immunofluorescence staining was performed using antibodies against CUL4B and an SSC marker, PLZF (14). High magnification epifluorescence microscopy showed complete removal of CUL4B from PLZF-positive spermatogonia, as well as round spermatids residing in the same seminiferous tubules (Fig. 1, E-J), demonstrating successful deletion of Cul4b in male germ cells.
To address whether in the absence of CUL4B, spermatocytes turn on Cul4a expression to compensate for the CRL4 function, we performed double immunofluoroscence staining of CUL4A and PLZF proteins in the Cul4b Vasa testis. No CUL4A staining was detected in PLZF-positive spermatogonia or spermatids of the Cul4b Vasa testis ( Fig. 1, K-N), suggesting that loss of CUL4B protein does not lead to compensatory Cul4a gene expression in these germ cell populations.
Cul4b Vasa males exhibit no defects during development and have a normal life span. However, mutant males are completely infertile. Fertility records showed regular and successful copulatory activities of mutant males, evidenced by the presence of vaginal plugs in females housed in the same cages. However, none of the females mated with Cul4b Vasa males produced any pups, in contrast to those mated with control males, averaging 11.2 pups per litter (Table 1).
To address the function of CUL4B in Sertoli cells, we generated Sertoli cell-specific Cul4b conditional knock-out mice using the Amh-Cre mouse (15), hereafter referred to as Cul4b Amh . Interestingly, despite its high expression in Sertoli cells (Figs. 1E and 2A), abolishing Cul4b in these somatic cells did not cause any reproductive phenotype. Immunofluorescent staining on Cul4b Amh testis sections showed complete absence of CUL4B in their Sertoli cells (Fig. 2B); however, these males exhibited normal fertility and seminiferous epithelium histol-ogy (Fig. 2F). These results suggest that cell autonomous CUL4B function is critical to male germ cell lineage but is dispensable in Sertoli cells.
Spermatozoa in Cul4b Vasa Testis Are Immotile-Anatomical examination of the Cul4b Vasa mice revealed normal internal male reproductive organs, including the testes, epididymides, vas deferens, and seminal vesicles. No significant differences in size, weight, or morphology of the testes were noted between the two genotypes. As shown in Fig. 1 (O-R), seminiferous tubule histology of the two genotypes is indistinguishable, and germ cells of all phases are present in the Cul4b Vasa testis.
To investigate the cause of infertility in Cul4b Vasa males, we harvested epididymal spermatozoa from these mice and used the computer-assisted sperm analysis to evaluate the quality of the mutant spermatozoa. As shown in Fig. 3A, Cul4b Vasa produced equivalent numbers of spermatozoa as compared with their wild type counterparts (54.5 Ϯ 5.1 M/ml, n ϭ 4, versus 54.8 Ϯ 5.6 M/ml, n ϭ 5 in controls; p ϭ 0.925). In contrast, sperm motility in the mutants was dramatically compromised: only 6.0 Ϯ 1.4% mutant spermatozoa were progressive motile (n ϭ 4), as compared with 87.2 Ϯ 9.3% in control mice (n ϭ 5, p Ͻ 0.0001). We next examined the morphology of epididymal  (47). C, CUL4B protein was detected in Sertoli cells (cells lining the seminiferous tubules with irregularly shaped nuclei and tripartite nucleoli), spermatogonia (Spg, cells lining the seminiferous tubules with round nuclei and often one visible nucleolus), round spermatids (rSpd, haploid cells located closer to the tubular lumen with smaller round nuclei and a large densely stained nucleolus), and elongated spermatids (eSpd, haploid cells located toward the tubular lumen with elongated nuclei) in a 4-month-old control Cul4b f/Y testis. D, Cul4b deletion through Vasa-Cre (Cul4b Vasa ) showed complete absence of the protein in all germ cell lineages, without affecting its expression in Sertoli cells. E-J, double IF of CUL4B and spermatogonial marker PLZF. CUL4B protein was readily detected in the PLZFpositive spermatogonia in the control testis (E-G) but not in the mutant testis (H-J). K-N, double IF of CUL4A and PLZF in Cul4b Vasa testis showing no ectopic CUL4A expression in the absence of Cul4b. O-R, H&E staining showed normal histology of the Cul4b Vasa testis, containing seminiferous tubules at all stages. K and L showed stage XII tubules of the two genotypes at higher magnification.

CUL4B Regulates Mammalian Spermatogenesis
spermatozoa using epifluorescence microscopy after staining with sperm structure-specific dyes. In typical wild type spermatozoa, the sickle-shaped heads contained highly condensed DNA, which stained positive for DAPI (Fig. 3B, blue), and were covered by a crescent-shaped acrosome stained by peanut agglutinin-lectin (green). The midpiece of the spermatozoa is covered by a helical mitochondrial sheath, and active mitochondria were stained bright red after live spermatozoa were incubated with a cell-permeable fluorescent dye, MitoTracker Red CMXRos (Fig. 3B). Most Cul4b Vasa sperms exhibited normal morphology when examined under light microscopy; however, more than half of the mutant spermatozoa stained very weakly or negatively for MitoTracker CMXRos (Fig. 3, D, G,  and H), suggesting a defect in mitochondrial membrane potential. Intriguingly, a small percentage (6.6 Ϯ 3.4%) of Cul4b Vasa spermatozoa exhibited a double-headed phenotype (Fig. 3, E  and G, asterisk). The individual heads of these spermatozoa were morphologically normal, with properly condensed nuclei and visible acrosomes. Nevertheless, they were often joined at the connecting piece and shared a common flagellum.
To further characterize the impaired mitochondrial activity observed in the Cul4b Vasa spermatozoa, we stained epididymal spermatozoa with a vital carbocyanine dye, JC1, which fluoresces at different wavelengths when binds to mitochondria with differential membrane potentials, and analyzed JC1 staining profiles by FACS (16). Fig. 3I showed the FACS results of JC1-stained spermatozoa from representative control and Cul4b Vasa mice. On average, 90.2 Ϯ 6.2% of the control spermatozoa (n ϭ 5) exhibited high mitochondrial membrane potential, evidenced by the orange JC1 staining (Fig. 3, I and J). However, this population in Cul4b Vasa mice was significantly reduced (62.3 Ϯ 10.8%, n ϭ 3, p ϭ 0.045), accompanied by an increase in the population of spermatozoa with low mitochondrial membrane potential stained green with JC1 (Cul4b Vasa , 16.8 Ϯ 3.3%, n ϭ 3; control, 6.2 Ϯ 1.1%, n ϭ 5; p ϭ 0.024). These results were consistent with our findings with MitoTracker staining and statistically demonstrated that lack of Cul4b in mouse germ cells led to defective mitochondrial activity in some spermatozoa.
ATP is the energy source that is hydrolyzed by the dynein ATPase to generate the propelling and coordinated force along the axoneme that enables proper sperm movement (17,18). The immotile phenotype of Cul4b Vasa prompted us to examine ATP production in the mutant spermatozoa by analyzing intracellular ATP level using a previously described method (8). Not surprisingly, we observed low ATP production in Cul4b Vasa spermatozoa (2.4 Ϯ 0.5 ϫ 10 3 moles/g versus 6.7 Ϯ 1.3 ϫ 10 4 moles/g in controls, n ϭ 4, p ϭ 0.002) determined by a bioluminescence assay (Fig. 3K). There is continuous debate on the source of sperm ATP, and two metabolic pathways are believed to be involved in this process. Mitochondria, localized in the sperm tail midpiece, have long been considered to be a major source of ATP production because of their high efficiency of energy production via oxidative respiration (19,20). Recent evidence, however, suggests that locally produced ATP via less efficient, but high throughput glycolysis in the principle piece of the sperm tail is the major energy source for sperm motility in the mouse (21)(22)(23). In our Cul4b Vasa mutants, the reduction in mitochondrial membrane potential in spermatozoa may contribute to lowered ATP production. However, loss of ATP production in mutant spermatozoa may also reflect a defective glycolytic pathway. To test this hypothesis, we measured the spermatozoa intracellular PEP level using a fluoremetric assay kit. PEP is the intermediate produced during glycolysis that is metabolized to the final product, pyruvate, by pyruvate kinase. We chose PEP instead of pyruvate as the indicator of glycolysis because pyruvate could also come from other metabolic pathways including amino acid metabolism (serine, alanine, and glycine), as well as conversion from lactate by lactate dehydrogenase. As shown in Fig. 3L, a significant decrease in PEP levels was detected in the mutant spermatozoa (Cul4b f/Y , 374.8 Ϯ 68.0, n ϭ 3; Cul4b Vasa , 116.4 Ϯ 18.8, n ϭ 3, p ϭ 0.017), indicating perturbation of glycolysis in the mutant spermatozoa. Taken together, these results suggest that a lack of ATP production resulting from impaired mitochondrial activity and glycolysis was a major cause of low motility/asthenozoospermic phenotype in Cul4b Vasa spermatozoa.
We next examined the ultrastructure of mutant spermatozoa by TEM because a loss of sperm motility is often associated with defective axoneme structure. The TEM analysis revealed that Cul4b Vasa mutant sperm axonemes frequently displayed missing, supernumerary, or displaced microtubule doublets (Fig. 3, Q-T). These defects were observed mostly in peripheral microtubule doublets, as opposed to the central microtubule doublet, and it was not clear whether the displaced microtubule doublets were still associated with dynein arms or radial spokes.
The mitochondrial sheaths of the mutant spermatozoa appeared morphologically normal (Fig. 3P). These data collectively demonstrated that deleting Cul4b from mouse spermatogonia, and subsequent germ cell populations did not affect the mitotic and meiotic phase of spermatogenesis but compromised spermiogenesis and biogenesis of the sperm flagellum in particular. Lack of Cul4b in these cells resulted in abnormal axonemal microtubule structure and reduced mitochondrial activity in the midpiece, which rendered Cul4b Vasa spermatozoa immotile and subsequently male infertile.
INSL6 Is a Direct CUL4B Substrate in Mouse Testes-Given the function of CUL4B as an E3 ubiquitin ligase, it is reasonable to hypothesize that in the Cul4b-null germ cells, certain CUL4B substrates may not be efficiently ubiquitinated, which led to defective spermiogenesis. In an attempt to identify potential CUL4B ligase substrates, we performed a mass spectrometry analysis on ubiquitinated proteins purified from enriched spermatids of control and Cul4b Vasa mice. Ubiquitinated proteins from pooled spermatid lysates were affinity purified through binding to p62-UBA immobilized agarose beads, and CUL4B target candidates were expected to be under-represented in Cul4b Vasa samples. The identity of potential targets were predicted and visualized by Scaffold 4 following mass spectrometry (9). In total, 2504 proteins were identified with a minimum protein threshold of 95.0% and a prophet false discovery rate of 0.6% or lower (supplemental Table S1). Among these, an insulin family member, INSL6, showed high spectrum count in the control testis but was absent in the p62-UBA-enriched testis lysates from Cul4b Vasa , suggesting INSL6 as a potential CUL4B substrate in the testis. INSL6 was previously reported to be highly expressed in germ cells, particularly in spermatids of both rodent and human testes (24) (HPA021364 SIGMA), although its function in spermatogenesis remains unknown. To test the ability of CUL4B to directly mediate polyubiquitination of INSL6, a plasmid encoding MYC-labeled full-length INSL6 was co-transfected into HEK293 cells along with His-tagged ubiquitin-expressing plasmid, with or without HA-CUL4B-expressing plasmid. As shown in Fig. 4A, the presence of CUL4B evidently increased polyubiquitinated INSL6, demonstrating that INSL6 is a direct substrate for CUL4B. Next HEK293 cells were co-transfected with INSL6-coding plasmid (INSL6-V5) along with various amount of HA-CUL4B plasmids as indicated in Fig. 4B, and a clear CUL4B dose-dependent decrease in INSL6 protein level was detected, indicating that INSL6 is directly degraded through the CUL4B-mediated ubiquitin-pro- teasome system mechanism. To further investigate the role of CRL complex in INSL6 polyubiquitination, we tested interactions between INSL6 and two key components of CRL4, the N-terminal binding DDB1 and the C-terminal interacting RING finger protein RBX1 (Rbx1/ROC1/Hrt1) (25). As shown in Fig. 4C, attenuating DDB1 and RBX1 functions, respectively, by RNA silencing in HEK293 cells resulted in reduced levels of INSL6 polyubiquitination. Moreover, DDB1 co-immunoprecipitated with INSL6 after co-transfection of MYC-INSL6 and HA-DDB1 plasmids into HEK293 cells, which indicates direct interactions between the two proteins (Fig. 4D). Together these data demonstrate that CUL4B directly polyubiquitinates INSL6 via the CRL complex in vitro, which leads to its degradation through ubiquitin-proteasome system.
Global Cul4b Deletion Leads to Germ Cell Depletion in Aged Mice-We have previously reported that Cul4b knock-out embryos died soon after implantation because of increased apoptosis in extraembryonic tissues, whereas epiblast-specific deletion of Cul4b via Sox2-Cre gave rise to viable Cul4b-null mice (Cul4b Sox2 ) (7). Because Cul4b is expressed in many somatic cell types, we characterized the fertility phenotype of Cul4b Sox2 males and compared with that of Cul4b Vasa to determine whether Cul4b also functions in somatic cells to support spermatogenesis. Cul4b Sox2 males were completely infertile as expected; however, they exhibited a different and more severe age-dependent testicular/sperm phenotype when compared with their Cul4b Vasa counterparts. Younger adult Cul4b Sox2 (4-month-old) testes exhibited no distinct defects (Fig. 5B), and their sperm phenotype resembled that in the Cul4b Vasa mice (data not shown), suggesting relatively normal mitotic and meiotic progression despite impaired spermiogenesis. However, as Cul4b Sox2 mutants aged (8-month-old), their testes shrank, and many seminiferous tubules started to lose their germ cell populations (Fig. 5D). Such a germ cell depletion phenotype varied between individuals (Fig. 5D, inset), but was not detected in older Cul4b Vasa mice (Fig. 5C, inset). As a result, oligoastheno- zoospermia was often observed in older Cul4b Sox2 mutants ( Fig.  5F; 54.8 Ϯ 5.6 M/ml, n ϭ 5 in controls; 16.0 Ϯ 13.6 M/ml, n ϭ 4 in Cul4b Sox2 ; p ϭ 0.007). Accompanied by low sperm counts, abnormal morphological phenotypes including deformed sperm heads, double-or triple-headed spermatozoa, and sperm heads detached from the flagella were common in Cul4b Sox2 mutants (Fig. 5F, insets). This germ cell depletion phenotype may have resulted from a failure to maintain the SSC niche.
Sertoli cells play a crucial role in SSC niche maintenance (26). We tested the hypothesis whether Cul4b expression in Sertoli cells is required to maintain the SSC niche. As mentioned earlier, knocking out Cul4b in Sertoli cells alone did not cause an infertility phenotype (Fig. 2). We followed these mice as they aged (up to 10 months) but did not observe the germ cell depletion phenotype described above. These data argue against a role for Cul4b in Sertoli cells to maintain the SSC niche and suggest that other somatic populations are responsible for this phenotype. One such somatic cell population could be macrophages. It was estimated that macrophages represent as much as one quarter of the interstitial cell population in rodent testes (27,28), and they play essential roles in regulating the SSC niche (29). In humans, increased number of testicular macrophages was often reported in patients with male infertility (30). Because the germ cell depletion phenotype in aging Cul4b Sox2 mutants was often accompanied by splenomegaly (data not shown), we hypothesized that loss of Cul4b in the spleen may have caused a systemic perturbation of macrophages, which may in turn adversely affected spermatogonial maintenance and self-renewal. To address this possibility, surface marker staining followed by FACS analysis was performed on dissociated splenocytes and testicular cells of control and Cul4b Sox2 animals at 9 month of age. We employed four-color FACS to distinguish different immunocyte populations: CD45 as a marker for nucleated hematopoietic cells (31), CD11b as a marker for macrophage/microglia (32), F4/80 as a mature macrophage marker (33), and GR1 as a neutrophil marker in peripheral organs (34). The splenocytes isolated from control animals contained 11.05 Ϯ 1.96% CD45 ϩ CD11b high macrophages/granulocytes of all CD45 ϩ lymphocytes (Fig. 5G), among which 5.81 Ϯ 1.61% were F4/80 high GR1 low macrophages (n ϭ 5; Fig. 4I). No statistically significant changes were observed in the Cul4b Sox2 splenocytes (CD45 ϩ CD11b high : 12.02 Ϯ 1.65%, n ϭ 6, p ϭ 0.41; F4/80 high GR1 low : 5.50 Ϯ 1.66%, n ϭ 6, p ϭ 0.76; Fig. 5, H and J). The isolated testicular cells from control animals comprised of 21.12 Ϯ 4.54% CD45 ϩ CD11b high macrophages/granulocytes of all lymphocytes, and the vast majority of them were macrophages (F4/80 high GR1 low : 91.84 Ϯ 2.32%, n ϭ 5; Fig. 5, K and M). CD45 ϩ CD11b high population remained largely unchanged in Cul4b Sox2 testes (17.20 Ϯ 5.30%, n ϭ 6, p ϭ 0.22; Fig. 5L); no statistically significant difference in macrophages was noted in these mutants (88.81 Ϯ 3.36%, n ϭ 6, p ϭ 0.11; Fig. 5N). Comparison between the lymphocyte compositions in the spleen and testis of old (10 -12 months old) and young (4 months old) adult Cul4b Sox2 mice revealed no significant difference. Altogether, these data demonstrated that lack of Cul4b in both germ cells and somatic cells led to male infertility caused by spermiogenesis defects similar to that found in the Cul4b Vaas mice and age-dependent germ cell depletion. The cause of the latter phenotype is yet to be determined but is unlikely to be associated with abnormal macrophage infiltration to the mutant testis.

Discussion
In the present study, we assessed the function of the E3 ubiquitin ligase CUL4B in spermatogenesis using genetically modified mouse models. We discovered a male infertility phenotype in both germ cell-specific conditional Cul4b knock-outs, as well as global Cul4b-null mutants.
Abolishing Cul4b in male germ cells did not cause any apparent morphological changes to the testis, and the progression of spermatogenesis appeared to be normal. The seminiferous tubules of Cul4b Vasa conditional knock-outs contained male germ cell populations of all phases, and a normal number of spermatozoa were recovered from these animals. The cause of male infertility in the mutants is low sperm motility (93% reduction compared with controls). Several factors could contribute to this decrease in motility. We detected a varied level of decreased mitochondrial activity and loss of mitochondrial membrane potential, evidenced by MitoTracker Red CMXRos and JC1 staining (Fig. 3). The mitochondrial sheath, tightly wrapped around the axoneme of the sperm tail midpiece, may provide energy in the form of ATP required for flagella movement. In humans, numerous cases of idiopathic and varicocelerelated asthenozoospermia have been reported to be associated with defective mitochondrial respiratory functionalities, with or without phenotypic abnormalities of the mitochondrial sheath (35)(36)(37). Furthermore, a positive correlation was found between high mitochondrial membrane potential and increased sperm motility in a human male infertility study (37). Oxidative phosphorylation relies on the H ϩ concentration gradient generated and maintained by the electron transport chain, which requires the impermeability of inner mitochondrial membrane. A decrease in mitochondrial membrane potential therefore indicates an inner membrane breach that couldpotentiallyleadtoimpairedATPproduction.Indeed,measurement of sperm ATP level revealed an almost complete loss of ATP in the Cul4b Vasa spermatozoa. However, the partial reduction in mitochondrial activity cannot fully account for the near total loss of ATP production in mutant spermatozoa. Therefore, the alternative pathway for ATP production, glycolysis, could also be impaired. We measured spermatozoa intracellular levels of PEP, the second to last intermediate of glycolysis, and revealed a significant decrease compared with the controls (Fig. 3). Therefore, loss of CUL4B simultaneously caused reduction of mitochondrial activity in the midpiece and compromised glycolysis, which resulted in lack of energy supply to mutant spermatozoa, ultimately leading to the immotile phenotype. In addition, we observed structural defects in the mutant spermatozoa. Specifically, the axonemes of many mutant spermatozoa showed microtubule anomalies, including supernumerary, missing, or misplaced outer microtubule doublets (Fig. 3). During murine spermiogenesis, the distal centriole nucleates microtubule polymerization to form the axoneme and the proximal centriole organizes the anlage of the capitulum, followed by disintegration of the centrosome in spermatids. Abnormal microtubule arrangements found in Cul4b Vasa sperm flagella suggest potential structural defects in the distal centriole or failure, at least in part, of microtubule nucleation.
To uncover the molecular mechanisms leading to these defects, we performed proteomic analysis comparing ubiquitinated proteins isolated from control and Cul4b Vasa spermatids by affinity purification using p62-UBA domain-immobilized beads. Among the proteins identified as potential CUL4B targets, INSL6 showed relatively high spectrum count in control samples but was completely absent in Cul4b Vasa samples. INSL6 is an insulin-like peptide that shows significant sequence similarity to the relaxin B chain (38). Intriguingly, high INSL6 expression has been reported in both mouse and human male germ cells, specifically in spermatids (24) (HPA021364 SIGMA), and INSL6 mutation in mouse led to compromised fertility in males only (39). Rearrangement of INSL6-containing chromosome 9q24.1 region in humans has been reported to be associated with testicular development failure (40), even though its function in spermatogenesis remains ambiguous. In this study, we performed in vitro experiments and provided solid evidence that CUL4B mediates both INSL6 polyubiquitination, as well as its degradation, implicating INSL6 as a direct substrate of the CUL4B E3 ubiquitin ligase (Fig. 4). Direct interaction between INSL6 and DDB1 was also demonstrated. In addition, knocking down either DDB1 or RBX1 resulted in a reduction in INSL6 polyubiquitination (Fig. 4). It is unfortunate that there is currently no commercially available antibody against mouse INSL6, which prevented us from analyzing INSL6 expression and accumulation in mouse germ cells. Further investigations aiming at revealing the function of INSL6 is underway, which will shed light on understanding the underlying mechanisms of its involvement in spermatogenesis.
Interestingly, Cul4b-null mutation by Sox2-Cre resulted in a different male fertility phenotype, in that the aged mutant mice exhibited a germ cell depletion phenotype. This led us to hypothesize that CUL4B is required to maintain the SSC niche. It is believed that the SSCs are not randomly distributed along the basement membrane of seminiferous tubules but rather preferably located to "SSC niches" that are adjacent to the interstitial tissues rich in blood vessels (41,42). Several types of somatic cells are responsible for maintaining the SSC niche. Inside the seminiferous tubules, Sertoli cells are thought to provide growth factors regulating SSC self-renewal/proliferation, including FGF2, GDNF (glial cell line-derived neurotrophic factor), and SCF (stem cell factor) (43)(44)(45). Because Cul4b was found to be highly expressed in Sertoli cells and loss of Cul4b was not compensated by Cul4a, we first speculated that it played a cell autonomous role in the Sertoli cells to maintain the SSC niche. However, we found no germ cell depletion phenotype in Sertoli-specific Cul4b mutants (Cul4b Amh ; Fig. 2). This result suggests that Cul4b is dispensable in Sertoli cells. The interstitial Leydig cells are thought to regulate Sertoli cell function through growth factor production via testosterone signaling. There is a possibility that loss of CUL4B in Leydig cells altered this signaling, causing disturbance in Sertoli cell functions. However, measurements of circulating testosterone in Cul4b Sox males revealed no significant difference compared with controls (data not shown). It was also proposed that interstitial macrophages play essential roles in maintaining SSC niche, and infiltration of macrophages in the testis was reported in patients with male infertility (30). High expression of Cul4b was reported in macrophages, and loss of CUL4B protein by shRNA knockdown significantly reduced TNF␣ protein in the activated macrophages (46). This led us to test whether Cul4b loss caused changes in testicular macrophage populations, which may in turn cause SSC loss. We tested this hypothesis by analyzing immunocyte composition of the spleen and testis tissues of control and Cul4b Sox2 animals. We used F4/80 as a macrophage marker for FACS analysis, and no noticeable differences were discovered between the two genotypes or mutants of different ages (Fig. 5). Thus, loss of SSCs in Cul4b Sox2 mutant testis is not the result of a change in testicular macrophage number. However, our study does not exclude the possibility that loss of Cul4b may change the properties of testicular macrophages, thereby affecting their ability to maintain the SSC niche. One final cell population that could contribute to the germ cell loss phenotype is the peritubular myoid cell. However, we did not detect either CUL4B or CUL4A expression in this cell type; hence it is unlikely that CUL4B is required in these cells to maintain the stem cell niches. It is noteworthy that our global mutants manifest a male germ cell depletion phenotype with a rather late adult onset, in contrast to most other germ cell depletion mouse models exhibiting early onset. This makes our animals an excellent model for studying long term maintenance of the SSC niche. In conclusion, CUL4B is required in the male germ cells for spermiogenesis and in somatic cells to maintain the SSC population of the testis.
Author Contributions-L. M. and P. Z. designed the study, and Y. Y., L. L., C. Y., C. L., G. M. V., C. W., and P. S. contributed to data acquisition and analysis. Y. Y., L. M., and P. Z. wrote the paper.