CCDC38 is required for sperm flagellum biogenesis and male fertility in mice

The sperm flagellum is essential for sperm motility (Freitas et al., 2017; Pereira et al., 2017), which is a fundamental requirement for male fertility. The flagellum contains four parts: connecting piece, midpiece, principal piece and end piece. The core of the sperm flagellum is the central axoneme, which consists of a central microtubule pair connected to nine peripheral outer microtubule doublets to form a ‘9+2’ structure (Sironen et al., 2020). The axoneme possesses radial spokes that connect the central and peripheral microtubules and are related to the mechanical movement of the flagellum (Inaba, 2011). Besides the axoneme, the sperm flagellum contains unique structures, the outer dense fibers (ODFs) and the fibrous sheath (FS), that are not present in cilia or unicellular flagella (Fawcett, 1975). The ODFs are the main cytoskeletal elements of sperm flagella, which are required for sperm motility (Inaba, 2011). ODFs contain nine fibers in the midpiece, each of which is associated with a microtubule doublet. In humans, ODFs 3 and 8 in the principal piece are replaced by two longitudinal columns of FS, with diminished fibers 3 and 8 ending at the annulus (Azizi and Ghafouri-Fard, 2017; Kim et al., 1999). There are at least 14 proteins that make up the ODFs, including ODF1 and ODF2 (Lehti and Sironen, 2017). Any defects in the axoneme structure can cause abnormalities in the sperm flagellum by changing the morphology and causing severe sperm motility disorders (Sha et al., 2014). Thus, axoneme structures are very important to sperm morphology and the function of the flagellum.

The ‘multiple morphological abnormalities of the flagella’ (MMAF) phenotype is a severe teratozoospermia (Coutton et al., 2015) that is characterized by various spermatozoa phenotypes with absent, short, coiled or irregular flagella. Many flagellar axoneme defects exist in individuals with MMAF, including disrupted microtubule doublets, ODFs, FSs, outer or inner dynein arms and others (Jiao et al., 2021). Over the past several years, many mutations have been found to be associated with MMAF individuals, and many mouse models display MMAF-like phenotypes, such as those containing loss-of-function mutants for Dnah2 (Li et al., 2019), Dnah8 (Liu et al., 2020), Cfap44, Cfap65 (Tang et al., 2017; Li et al., 2020), Qrich2 (Shen et al., 2019), Cep135 (Sha et al., 2017) and Ttc21a (Liu et al., 2019), all of which have been reported as MMAF-related genes. Despite rapid progress having been made in understanding the mechanisms underlying MMAF, the pathogenesis associated with human idiopathic MMAF remains unknown.

Coiled-coil domain-containing (CCDC) proteins are involved in a variety of physiological and pathological processes. An increasing number of CCDC proteins have been suggested to be involved in ciliogenesis (Priyanka and Yenugu, 2021), yet only some of those genes are involved in spermatogenesis, such as Ccdc9, Ccdc11, Ccdc33, Ccdc42, Ccdc63 and Ccdc172. These genes are associated with sperm flagellum biogenesis and manchette formation, and their defects lead to male infertility (Sha et al., 2019; Wu et al., 2021; Tapia Contreras and Hoyer-Fender, 2019; Young et al., 2015; Yamaguchi et al., 2014). CCDC42 is a highly expressed protein in mouse testis, and localizes to the manchette, head-to-tail coupling apparatus (HTCA) and sperm tail during spermatogenesis. CCDC42 is necessary for HTCA assembly and sperm flagellum biogenesis (Tapia Contreras and Hoyer-Fender, 2019); however, the functional role of CCDC42 in spermatogenesis is still not understood.

Here, we find that CCDC38 interacts with CCDC42, is highly expressed in the testis and associates with the manchette in the elongating spermatid. Importantly, Ccdc38 knockout in mice resulted in an abnormally elongated manchette and an MMAF-like phenotype. Furthermore, we find that CCDC38 interacts with the proteins intraflagellar transport 88 (IFT88) and ODF2 to facilitate ODF2 transport in the flagellum. Our results suggest that the CCDC42-CCDC38 interaction mediates ODF2 transport during flagellum biogenesis, and both proteins are essential for flagellum biogenesis and fertility in male mice, suggesting mutations of these two genes might be associated with male infertility in humans.

Many CCDC proteins participate in flagellum biogenesis during spermiogenesis (Priyanka and Yenugu, 2021). CCDC42 localizes to the centrosome, HTCA, manchette and sperm tail in male germ cells, and it is involved in the biogenesis of motile cilia and flagella in mice (Perles et al., 2012; Tapia Contreras and Hoyer-Fender, 2019; Pasek et al., 2016; Silva et al., 2016). To understand the underlying mechanism of CCDC42 in flagellum biogenesis during spermiogenesis, we used the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database to search for CCDC42-binding candidates (Fig. 1A), and chose to analyze the role of CCDC38, which is exclusively expressed in testes (Lin et al., 2016). Epitope-tagged CCDC42 and CCDC38 were expressed in human embryonic kidney 293 expressing SV40 large T antigen (HEK293T) cells and followed by immunoprecipitation experiments using the anti-MYC antibody, which demonstrated that CCDC38 co-immunoprecipitated with CCDC42 (Fig. 1B). An overlapping immunostaining pattern was clearly seen in HeLa cells transiently expressing GFP-CCDC38 and MYC-CCDC42 and GFP-CCDC38 also colocalized with γ-tubulin as reported (Firat-Karalar et al., 2014) (Fig. 1C; Fig. S1). These results suggest that CCDC42 can indeed interact with CCDC38.

Next, we examined the localization of endogenous CCDC38 during spermatogenesis. CCDC38 was detected as two adjacent spots near the nuclei of spermatocytes or round spermatids, and it localized to the skirt-like structure encircling the spermatid head from steps 9 to 14 (Lehti and Sironen, 2016) and the testicular sperm tail (Fig. 1D). We, therefore, speculate that CCDC38 might participate in flagellum biogenesis during spermiogenesis.

Reverse transcription PCR (RT-PCR) analysis detected Ccdc38 expression in the testis and revealed that Ccdc38 was first expressed at postnatal day (P) 14, peaking at P35 (Fig. 2A,B). To determine the physiological role of CCDC38, we generated Ccdc38-deficient mice using the CRISPR/Cas9 system to delete exons 5-11 of the Ccdc38 gene (Fig. 2C). The Ccdc38-knockout mice were genotyped by genomic DNA sequencing and further confirmed by PCR, as seen by the 591 bp band for the wild-type (WT) allele, 750 bp band for homozygotes, and two bands of 750 bp and 591 bp for heterozygotes (Fig. 2D). Subsequent western blotting analysis validated complete ablation of CCDC38 in the total protein extracts from Ccdc38^−/− testes (Fig. 2E). We next examined the fertility of Ccdc38^+/+, Ccdc38^+/− and Ccdc38^−/− mice. Male Ccdc38^−/− mice exhibited normal mounting behaviors and produced coital plugs but failed to produce any offspring after mating with WT adult female mice (Fig. 2F). In contrast, female Ccdc38^−/− mice could generate offspring after mating with WT adult males, similar to the Ccdc38^+/+ female mice (Fig. 2G). However, the knockout of Ccdc38 did not affect either testis size (Fig. 2H) or the ratio of testis weight to body weight (Fig. 2I-K). Taken together, Ccdc38 knockout leads to male infertility.

To explore the cause of male infertility further, we examined the cauda epididymis of Ccdc38^+/+, Ccdc38^+/− and Ccdc38^−/− mice by Hematoxylin and Eosin (H&E) staining, and found fewer spermatozoa in the epididymal lumen of Ccdc38^−/− mice compared with Ccdc38^+/+ and Ccdc38^+/− mice (Fig. 3A). We next released spermatozoa from the epididymis and found that the number of sperm from Ccdc38^−/− mice was significantly less than that of Ccdc38^+/+ and Ccdc38^+/− mice (Fig. 3B). The percentage of spermatozoa that were motile decreased sharply for Ccdc38^−/− mice compared with Ccdc38^+/+ and Ccdc38^+/− mice (Fig. 3C), but sperm number and motility showed no obvious differences between Ccdc38^+/+ and Ccdc38^+/− mice (Fig. 3B,C). Thus, we focused our attention on only Ccdc38^+/+ and Ccdc38^−/− mice in subsequent studies. We also noticed that Ccdc38^−/− spermatozoa showed morphological aberrations, including abnormal nuclei and an MMAF-like phenotype of either a short tail (type 1), disordered filaments (type 2), impaired spermatozoon head (type 3), curly tail (type 4) and tailless (type 5) (Fig. 3D). The percentage of spermatozoa with abnormal heads and flagella is shown in Fig. 3E. Scanning electron microscopy (SEM) further revealed the morphological abnormalities of Ccdc38^−/− spermatozoa (Fig. 3F). Therefore, the knockout of Ccdc38 results in an MMAF-like phenotype in mice.

To investigate further why Ccdc38 knockout led to an MMAF-like phenotype, we first used Periodic acid-Schiff (PAS) staining to determine in which stages the defects occurred. During spermatogenesis, the cycle of the seminiferous epithelium can be subdivided into 12 stages, which can be distinguished from one another by steps in spermatid development. Meanwhile, the spermatid development can be subdivided into 16 steps, according to acrosome and sperm nuclear morphology (Griswold et al., 2016). In testis sections from Ccdc38^+/+ mice, round spermatids differentiated into elongating spermatids from stage IX (Griswold et al., 2016), whereas abnormal elongated spermatids occurred at stage X in the testes of Ccdc38^−/− mice (Fig. 4A). To delineate the detailed defects of Ccdc38^−/− spermatids, we analyzed steps 1-16 for spermiogenesis in both Ccdc38^+/+ and Ccdc38^−/− mice, and found that at steps 1-8, the morphology of acrosomes and nuclei of Ccdc38^−/− spermatids was similar to that of Ccdc38^+/+ spermatids. In Ccdc38^+/+ mice, spermatid head elongation and maturation began at step 9, whereas in Ccdc38^−/− mice, spermatid head elongation and maturation showed abnormalities at step 9, eventually leading to abnormal spermatozoa at step 16 (Fig. 4B). These results support the hypothesis that CCDC38 plays an essential role during spermiogenesis.

To study the causes of abnormal sperm morphology after Ccdc38 depletion, H&E staining was used to determine the morphology of seminiferous tubules in Ccdc38^+/+ and Ccdc38^−/− mice. Compared with Ccdc38^+/+ testes, obvious shortened tails and tailless spermatids could be detected in Ccdc38^−/− testes (Fig. 5A). Immunofluorescence staining for the flagellum marker acetylated tubulin (Ac-tubulin) further confirmed the flagellum biogenesis defects in Ccdc38^−/− testes (Fig. 5B). We conducted immunofluorescence analysis of lectin peanut agglutinin (PNA) which marks the acrosome, and Ac-tubulin to determine which stages were affected by Ccdc38 knockout and found that the flagella of Ccdc38^−/− spermatids were short and curly from stages IV-V compared with those of Ccdc38^+/+ spermatids (Fig. 5C). Using transmission electron microscopy (TEM), we observed that the ODFs, FS, mitochondrial sheath and axoneme were also abnormally organized in the Ccdc38^−/− elongating spermatids (Fig. 5D).

When spermatids were elongated, the sperm head was found to be abnormal in the Ccdc38^−/− spermatids, indicating that the manchette might be abnormally formed (stage X-XI, Fig. 5C). The manchette is important for sperm head shaping (Wei and Yang, 2018), so we turned our attention to evaluating manchette structure. We found the manchettes of Ccdc38^−/− spermatids were roughly normal at steps 8-10 but were abnormally longer at steps 11-12 compared with control spermatids (Fig. 6A). We also used TEM to analyze the manchette in further detail and found Ccdc38^−/− spermatids, but not Ccdc38^+/+ spermatids, became abnormally elongated at step 11 (Fig. 6B). In support of these results, we found that CCDC38 colocalized with α-tubulin to the manchette in control mice (Fig. 6C). All these results suggest that CCDC38 is directly involved in flagellum biogenesis.

It has been reported that CCDC42, IFT88, CFAP53 and KIF3A are involved in anterograde transport during flagellum biogenesis (Wu et al., 2021). To test whether CCDC38 also participates in anterograde transport by interacting with CFAP53 and intraflagellar transport (IFT) complexes, such as IFT88 and IFT20, we co-transfected pCSII-MYC-IFT88, pCSII-MYC-CFAP53 or pRK-FLAG-IFT20 with pEGFP-C1-CCDC38 in HEK293T cells. We next immunoprecipitated CCDC38 using the anti-GFP antibody and found that IFT88 and CFAP53, but not IFT20, immunoprecipitated with CCDC38 (Fig. 7A-C). We also analyzed protein expression levels in Ccdc38^+/+ and Ccdc38^−/− mice testes and found that IFT88, CFAP53 and IFT20 expression levels were all significantly decreased in Ccdc38^−/− mice testes compared with Ccdc38^+/+ mice testes (Fig. 7D,E). Because IFT88 can participate in anterograde transport, we next evaluated the distribution of IFT88 in spermatids at different steps and found that IFT88 was present in the manchette and elongating sperm tails in Ccdc38^+/+ mice, whereas in the Ccdc38^−/− spermatids, IFT88 also localized to the manchette and elongating sperm tails in spermatids (Fig. 7F). Therefore, CCDC38 might regulate sperm flagellum biogenesis by interacting with IFT88.

It has been reported that ODF1 and ODF2 can interact with CCDC42 and have been found to be involved in the formation of the male germ cell cytoskeleton (Tapia Contreras and Hoyer-Fender, 2019). To study the relationship between CCDC38 and ODF2, reciprocal co-immunoprecipitation assays were carried out. We transfected the pCDNA-HA-ODF2 and pEGFP-C1-CCDC38 plasmids into HEK293T cells and found that CCDC38 and ODF2 were able to immunoprecipitate with each other (Fig. 8A). These findings suggest that CCDC38 interacts with ODF2.

As the main scaffolding protein in ODFs, ODF2 is essential for sperm flagellum integrity and beating (Donkor et al., 2004; Ito et al., 2019; Fawcett, 1975). To determine the effects of Ccdc38 knockout on ODF1 and ODF2 protein levels, we examined testicular extracts and found that ODF2, but not ODF1, levels were significantly decreased in Ccdc38^−/− testicular extracts (Fig. 8B,C) compared with those of Ccdc38^+/+. Next, we used immunofluorescence to analyze the localization of ODF2 in spermatids and epididymal spermatozoa. We found that ODF2 localized on the manchette, along with the sperm tail, in elongated spermatids of Ccdc38^+/+ mice, whereas it localized on the manchette without tail staining in most of the elongated spermatids of Ccdc38^−/− mice (Fig. 8D). Of note, ODF2 colocalized with α-tubulin in the midpiece and principal piece of Ccdc38^+/+ sperm tails, whereas it displayed discontinuous, punctiform signals in only short or curly tails in Ccdc38-knockout spermatozoa (Fig. 8E), suggesting that ODF defects in Ccdc38-knockout spermatozoa might arise from ODF2 transport defects during spermiogenesis.

Ccdc38 is exclusively expressed in testes (Lin et al., 2016), but its role during spermiogenesis has not yet been investigated. To study its role during spermiogenesis, we generated a Ccdc38^−/− mouse model and found Ccdc38^−/− male mice to be sterile (Fig. 2F) owing to significantly reduced spermatozoa number and motility (Fig. 3B,C). As for which kinds of cells were affected by CCDC38 function, a discrepancy exists between our results and others. Lin et al. (2016) previously reported that CCDC38 is mainly localized to spermatogonia and spermatocytes, but we found that CCDC38 mainly participates in spermatid elongation and is localized on the manchette and the sperm tail (Fig. 1D, Fig. 6C). The discrepancy may have arisen from different CCDC38 antibodies being used. Our CCDC38 antibody was validated by both western blotting (Fig. 2E) and immunofluorescence (Fig. 6C), supporting the robustness of our antibody labeling and the findings of our protein localization in this study. Further support for the manchette and sperm tail localization of CCDC38 is shown by the knockout of Ccdc38, which results in MMAF but does not affect ciliogenesis in the lung and trachea of Ccdc38^−/− mice (Fig. S2). These results suggest that CCDC38 could be directly involved in flagellum biogenesis but not in other processes.

The manchette is a transient structure in developing germ cells, which is required for sperm nuclear condensation and flagellum biogenesis (Wei and Yang, 2018) by providing the structural basis for intra-manchette transport (IMT). IMT transfers structural and functional proteins to the basal body and is essential for nucleocytoplasmic transport (Kierszenbaum, 2002; Kierszenbaum et al., 2002). As an IMT component, CCDC42 localizes to the manchette, connecting piece, and sperm tail during spermiogenesis, and can interact with ODF1 and ODF2 to regulate germ cell cytoskeleton formation (Tapia Contreras and Hoyer-Fender, 2019; Pasek et al., 2016). In this study, we found that CCDC38 interacts with CCDC42, colocalizes with CCDC42 on centrosomes in HeLa cells (Fig. 1A-C) and interacts with ODF2 and ODF1 (Fig. 8A; Fig. S3). Both ODF1 and ODF2 are components of ODFs and are important for male fertility. The knockout of Odf1 in mice impairs sperm head-to-tail coupling (Yang et al., 2014, 2012), whereas the knockout of Odf2 leads to pre-implantation lethality. Even the absence of a single copy of this gene results in separation of the sperm neck and midpiece (Qian et al., 2016; Tarnasky et al., 2010). Furthermore, recent human infertility studies showed that mutations in ODF2 caused MMAF (Zhu et al., 2022). Consistent with the clinical studies, we found that in Ccdc38-knockout mice, ODF2 but not ODF1 protein levels decreased in the testis (Fig. 8B,C) and ODF2 distribution was altered in flagella (Fig. 8D,E). Thus, CCDC38 either works as a partner of ODF2 to support its stability or participates in IMT to mediate ODF2 transport during flagellum biogenesis. As CCDC38 also interacted with CCDC42, we believe CCDC38 plays a role in ODF2 transport, and this is further supported by its interaction with IFT88.

In addition to IMT, IFT is also required for flagellum biogenesis, and is responsible for protein transport during the development of the flagellum. During IFT, various cargos are transported from the basal body to the tip of the flagellum and then back to the sperm head along the axoneme (Scholey, 2003; Taschner and Lorentzen, 2016; Ishikawa and Marshall, 2017). IFT88 is a component of the IFT-B complex and is present only in the heads and tails of elongating spermatids, and not in mature sperm (San Agustin et al., 2015). IFT88 has been found to interact with KIF3B to regulate anterograde transport along the axoneme (Rosenbaum and Witman, 2002; Funabashi et al., 2018). As an IFT88-interacting protein (Fig. 7A), CCDC38 may also participate in anterograde transport along the flagellum. Previously, we found that CFAP53 could interact with both CCDC42 and IFT88 to regulate the anterograde transport along the flagellum (Wu et al., 2021). In this study, we found that CCDC38 also interacts with CFAP53 (Fig. 7B). Thus, through its interactions with CFAP53, CCDC42 and IFT88, CCDC38 may regulate cargo transport by IMT and IFT during flagellum biogenesis. Further investigations are needed to reveal the detailed roles of these proteins and their relationships during flagellum biogenesis.

In summary, we have identified a new CCDC42-interacting protein, CCDC38, that is essential for spermiogenesis and flagellum biogenesis, as the knockout of Ccdc38 results in an MMAF-like phenotype in mice. Given that these genes are evolutionarily conserved in humans, we believe that mutations of these genes may exist in individuals with MMAF and further studies should be undertaken to uncover how these mutations affect disease presentation.

Mouse Ccdc42, Cfap53 and Ift88 were obtained from mouse testis cDNA and were cloned into the pCSII-Myc vector using the Clone Express Ultra One Step Cloning Kit (C115, Vazyme). Mouse Ccdc38 was obtained from mouse testis cDNA and it was cloned into the pEGFP-C1 vector. Mouse Ift20 was obtained from mouse testis cDNA and it was cloned into the pRK vector. Mouse Odf2 was obtained from mouse testis cDNA and it was cloned into the pCDNA-HA-3.0 vector. pcDNA-HA-3.1 vector was obtained from Addgene (128034). pCSII-Myc and pEGFP-C1 vectors were obtained from BioVector NTCC. pRK-flag vector was modified from pRK7-N-Flag (BioVector NTCC).

The mouse Ccdc38 gene is 1692 bp and contains 16 exons. Ccdc38-knockout mice were generated using the CRISPR/Cas9 system from Cyagen Biosciences. The genotyping primers used for knockout were: F1, 5′-GTAGCTGTTTCTAAGCGATCATCA-3′; R1, 5′-ACTAGGTACCTCAAGCTGGTTTAGA-3′. The genotyping primers for WT mice were: F1, 5′-GTAGCTGTTTCTAAGCGATCATCA-3′; R2, 5′-GTCATGGGACAGATGTGGAACTA-3′.

All animal experiments were performed according to approved institutional animal care and use committee (IACUC) protocols (#08-133) of the Institute of Zoology, Chinese Academy of Sciences, Beijing, China. All surgery was performed under sodium pentobarbital anesthesia, and every effort was made to minimize suffering.

The following primary antibodies were used for immunofluorescence (IF) and western blotting (WB): mouse anti-GFP (M20004, Abmart; 1:1000 for WB), rabbit anti-MYC (BE2011, EASYBIO; 1:1000 for WB), anti-ODF2 (12058-1-AP, Proteintech; 1:1000 for WB, 1:200 for IF), mouse anti-α-tubulin antibody (AC012, ABclonal, 1:100 for IF), mouse anti-α/β-tubulin antibody (ab44928, Abcam; 1:100 for IF), mouse anti-GAPDH antibody (AC002, ABclonal; 1:10,000 for WB), mouse anti-ODF1 antibody (sc-390152, Santa Cruz Biotechnology; 1:500 for WB), mouse anti-Ac-tubulin antibody (T7451, Sigma-Aldrich; 1:200 for IF), mouse anti-CCDC38 (generated by Dia-an Biotechnology, Wuhan, China; 1:20 for IF) and rabbit anti-CCDC38 (generated by Dia-an Biotechnology, Wuhan, China; 1:500 for WB). The Alexa Fluor 488 conjugate of lectin PNA (1:400, L21409, Thermo Fisher Scientific) was used for immunofluorescence. The following secondary antibodies were used: goat anti-rabbit FITC (1:200, ZF-0311, http://www.zsbio.com/), goat anti-TRITC (1:200, ZF-0316, Zhong Shan Jin Qiao), goat anti-mouse FITC (1:200, ZF-0312, Zhong Shan Jin Qiao) and goat anti-rabbit TRITC (1:200, ZF0313, Zhong Shan Jin Qiao).

As previously reported (Liu et al., 2016), the tunica albuginea of the testis was peeled and added to RIPA buffer supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF) and protease inhibitor cocktail (PIC; Roche Diagnostics, 04693132001). The solution was sonicated transiently and then placed on ice for 30 min. The samples were centrifuged at 13,523 g for 15 min at 4°C. Next, the supernatants were collected in new tubes. Protein lysates were electrophoresed and electrotransferred to a nitrocellulose membrane. The membrane was then incubated with a primary antibody followed by a secondary antibody. Finally, the membrane was scanned using an Odyssey infrared imager (LI-COR Biosciences; RRID: SCR_014579).

Transfected cells were lysed in a lysis buffer (50 mM HEPES, pH 7.4, 250 mM NaCl, 0.1% NP-40 containing PIC and PMSF) on ice for 30 min and centrifuged at 13,523 g at 4°C for 15 min. Cell lysates were first incubated with a primary antibody (mouse anti-GFP antibody, M20004L; Abmart) overnight at 4°C and then incubated with protein A-Sepharose (GE, 17-1279-03) for 2 h at 4°C. The immunoprecipitates were washed three times with lysis buffer, the SDS loading buffer (2% SDS with 1 M DTT) added, heated for 10 min at 95°C and subjected to western blotting analysis.

The cauda epididymis was isolated from 8-week-old mice. Sperm was released from the cauda epididymis in human tubal fluid (HTF) medium (MR-070-D, Sigma-Aldrich) and incubated at 37°C for 15 min. The medium was then diluted to 1:100 and sperm were counted using a hemocytometer.

As previously reported (Wang et al., 2018), the testes were dissected after euthanasia, and fixed with Bouin's fixative for 24 h at 4°C. Next, the testes were dehydrated with graded ethanol, embedded in paraffin and 5 μm sections were cut and placed on glass slides. Sections were stained with H&E and PAS for histological analysis after deparaffinization.

Testes were processed using methods previously reported, with some modifications (Liu et al., 2016). Briefly, the testis from WT and Ccdc38-knockout mice testes and epididymides were dissected and fixed in 2.5% (v/v) glutaraldehyde in 0.1 M cacodylate buffer (6131-99-3, Sigma-Aldrich) at 4°C overnight. After washing in 0.1 M cacodylate buffer, samples were cut into small pieces, then immersed in 1% OsO₄ for 1 h at 4°C. Samples were dehydrated through a graded acetone series (50%, 60%, 70%, 80%, 90%, 95%, 100%) and embedded in resin [DDSA, NMA, enhancer, 812 (Sigma-Aldrich)] for staining. Ultrathin sections of 60 nm were cut and stained with uranyl acetate and lead citrate. Images were acquired and analyzed using a JEM-1400 transmission electron microscope (JEM-1400, JEOL).

Sperm was released from the epididymis in HTF medium at 37°C for 15 min, centrifuged for 5 min at 500 g, washed twice with a phosphate buffer (NaH₂PO₄ and Na₂HPO₄·12H₂O) and fixed in 2.5% glutaraldehyde solution overnight. Samples were finally dehydrated in graded ethanol, subjected to drying and coated with gold. Images were acquired and analyzed using a SU8010 scanning electron microscope (SU8010, JEOL).

The tunica albuginea of the testis was peeled and incubated with collagenase IV (17104019, Gibco) and hyaluronidase (H1115000, Sigma-Aldrich) in PBS for 15 min at 37°C, then washed twice with PBS. Next, the sample was fixed with 4% paraformaldehyde for 5 min, and then coated on a glass slide to dry. The slides were washed with PBS three times, treated with 0.5% Triton X-100 for 5 min, and finally blocked with 5% bovine serum albumin for 30 min. The samples were incubated with primary antibodies at 4°C overnight, followed by incubation with secondary antibodies for 1 h at room temperature and DAPI staining. Images were taken using LSM880 (Zeiss) and Sp8 (Leica) microscopes.

Total RNA was extracted from mice tissues using TRIzol (Invitrogen; Thermo Fisher Scientific). The cDNA was synthesized using PrimeScript RT Regent Kit (Takara, RR037A). The primer sequences were: Gapdh F,5'-AGTGGCAAAGTGGAGATT-3′; Gapdh R, 5′-GTGGAGTCATACTGGAACA-3′; Ccdc38 F, 5′-CTTGTCCTGTTAGTCCTGTATAG-3′; Ccdc38 R, 5′-CGTAGAGATGAAGTGTGATGAT-3′.

All data are presented as the mean±s.d. Statistical significance of the differences between the mean values for the various genotypes was measured by Student's t-tests with paired, two-tailed distribution. The data were considered significant for *P<0.05, **P<0.01 or ***P<0.001.

We thank the State Key Laboratory of Membrane Biology, Institute of Zoology, Chinese Academy of Sciences for our electron microscopy work. We are grateful to Pengyan Xia for his help in preparing the electron microscopy samples and we thank Jingnan Liang (Institute of Microbiology, Chinese Academy of Sciences) for TEM technical support.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.200516.