ARRDC5 deficiency impairs spermatogenesis by affecting SUN5 and NDC1 Free

Sperm are highly specialized cells that deliver DNA of the male parent to the offspring; sperm with normal morphology and mobility are crucial for successful fertilization. Mammalian sperm are produced by the complex and precisely orchestrated process of spermatogenesis in the testis, which can be divided into the following three steps: the division of diploid spermatogonia, the formation of haploid round spermatids and morphogenetic transformation events for sperm formation. Morphogenetic transformation events (also called spermiogenesis) include nucleus condensation and elongation, acrosome biogenesis, flagellum formation and cytoplasm removal (Hao et al., 2019; Tanaka and Baba, 2005). Any defects during these processes may lead to azoospermia, to teratozoospermia or to sperm morphology and function defects, all of which can result in male infertility.

The arrestin family of proteins, including visual arrestins and β-arrestins, were originally identified as binding to phosphorylated G protein-coupled receptors and causing their desensitization (Gurevich et al., 1995). A larger family of proteins called α-arrestins (also called arrestin-related trafficking adaptors) were subsequently reported to share the overall arrestin-fold structure. So far, six α-arrestins have been identified, all of them sharing arrestin-like domains, including arrestin domain-containing protein 1 (ARRDC1), ARRDC2, ARRDC3, ARRDC4, ARRDC5 and thioredoxin-interacting protein (TXNIP) (Alvarez, 2008). It has been reported that ARRDC4 is important for sperm maturation by controlling extracellular vesicles release (Foot et al., 2021), but the roles of the other α-arrestins in spermatogenesis and male infertility have remained unclear. According to the mouse ENCODE transcriptome data (Yue et al., 2014), ARRDC5 is different from other α-arrestins by being specifically highly expressed in the testis, which suggests a role in male fertility. Thus, in this study we aimed to explore the expression pattern of ARRDC5 and its physiological functions in spermatogenesis and male fertility. During the preparation of this article, a study reported the role of Arrdc5 in sperm spermatogenesis (Giassetti et al., 2023). However, that study mainly performed a phenotypic analysis by using Arrdc5 knockout mice, and the underlying molecular mechanisms and signaling pathways for how ARRDC5 affects spermatogenesis remain unclear.

Here, we report that ARRDC5 is essential for spermatogenesis. Arrdc5 knockout mice showed male infertility with abnormal sperm morphology, reduced motility and failed capacitation. Through mass spectrometry of testis extracts, we found that ARRDC5 deficiency caused reduced SUN5 and NDC1 protein levels, which are essential for anchoring the sperm head to the tail. Furthermore, we found that ARRDC5 deficiency might affect the SEC22A-mediated transport and localization of SUN5 and NDC1, thus leading to spermatogenesis defects. In addition, we performed intracytoplasmic sperm injection (ICSI) and obtained blastocysts, which can be considered as a rescue therapeutic strategy. Our results reveal the previously unrecognized role of ARRDC5 in SUN5 and NDC1 transport and localization, and reveal a previously unreported signaling pathway involved in sperm head and tail anchoring.

Arrdc5 is specifically highly expressed in the testicular tissue of mice (Fig. 1A), and to investigate the role of ARRDC5, we generated Arrdc5 knockout mice using CRISPR/Cas-mediated genome engineering (Fig. 1B,C). A 4-month period of mating showed that Arrdc5^–/– female mice and Arrdc5^+/+ and Arrdc5^+/– male mice could generate offspring normally, whereas Arrdc5^–/– male mice mating with wild-type female mice could not generate offspring (Fig. 1D). The testis size and the ratio of testis weight to body weight were significantly reduced in Arrdc5^–/– males compared with Arrdc5^+/+ and Arrdc5^+/– males (Fig. 1E,F). Histologically, Hematoxylin Eosin Saffron staining of testis sections showed approximately intact spermatogenesis progression stages in Arrdc5^–/– male mice (Fig. 1G), whereas sperm in the epididymis in Arrdc5^–/– male mice showed a disorganized arrangement and reduced numbers (Fig. 1H).

To explore the effect of ARRDC5 depletion on sperm, we isolated the sperm from the cauda epididymis. The Arrdc5^–/– sperm showed slower speed compared with control sperm. We then analyzed sperm mobility and found defects in ARRDC5 led to significantly reduced sperm motility (P<0.001) and progress rate (P<0.001) (Fig. 2A) together with reduced average path velocity (P=0.006), straight line velocity (P=0.002) and curvilinear velocity (P<0.001) (Fig. 2B). We further evaluated the morphology of the sperm. Arrdc5^–/– sperm under light microscopy showed a significantly (P<0.001) higher bent-head rate (0.93±0.018, mean±s.d.) than wild-type sperm (0.091±0.045) (Fig. 2C,D); this was subsequently confirmed by scanning electron microscopy (SEM) (Fig. 2E). A few other anomalies of morphology were also observed in Arrdc5^–/– sperm, such as abnormal head morphology (0.358±0.02) and double-headed sperm (0.027±0.01) (Fig. 2C,F). To explore the stage at which sperm defects occurred, we examined the testis by transmission electron microscopy (TEM). At stage 12-14 of spermatogenesis, there was full and tight attachment between the head-tail coupling apparatus (HTCA) and the nuclear envelope in wild-type spermatids, whereas spermatids with ARRDC5 deficiency showed an incomplete and loose attachment that led to an offset angle between the head and tail (Fig. 2G). Meanwhile, we assessed manchettes during spermatogenesis, and the manchettes of Arrdc5^–/– spermatids showed inaccurate and disheveled localization compare with wild-type spermatids (Fig. S1). This also partly indicated the role of ARRDC5 in intracellular transport. Correspondingly, mature sperm from the Arrdc5^–/– cauda epididymis were affected and showed a bent head and other abnormalities in the SEM imaging (Fig. 2H). These results illustrate the importance of ARRDC5 for spermatogenesis, sperm morphology and sperm mobility.

We performed in vitro fertilization (IVF) to explore the reason for male mouse infertility and, as we expected, almost no two-pronuclear (2PN) zygotes were formed with Arrdc5^–/– sperm (Fig. 3A,B). Immunofluorescence staining confirmed that most sperm were unable to penetrate into the zona pellucida (ZP) (Fig. 3C). We then measured the presence of protein tyrosine phosphorylation in sperm after incubation in human tubal fluid medium (HTF) to determine the ability for sperm capacitation. Arrdc5^–/– sperm failed to capacitate, as indicated by reduced levels of protein tyrosine phosphorylation (Fig. 3D,E). These results indicated that ARRDC5 deletion impaired the sperm capacitation, which prevented the sperm from crossing the ZP, thus resulting in fertilization failure.

To explore the molecular mechanism behind the role of ARRDC5 in spermatogenesis, we performed whole-protein mass spectrometry of adult wild-type and Arrdc5^–/– testes to search for differentially expressed proteins. A total of 8193 proteins were scored, and the genes that encoded the top 300 downregulated proteins in Arrdc5^–/– mice were selected for Gene Ontology (GO) term enrichment analysis (Fig. 4A,B, Fig. S2A,B). The GO analysis indicated that these proteins were highly enriched in the categories of protein transport and spermatogenesis (Fig. 4B).

Because Arrdc5^–/– sperm showed bent heads and reduced motility, and based on the results of previous studies and reported mouse models, we focused on NDC1 and SUN5. NDC1 transmembrane nucleoporin (NDC1) has been shown to be primarily expressed in the sperm neck between the sperm head and tail (Yeh et al., 2015). NDC1 is a regulator of SEPT12 function, and SEPT12-NDC1 complexes are involved in sperm head and tail formation during mammalian spermiogenesis (Lai et al., 2016). Western blotting showed about a 75% reduction (P=0.0029) of NDC1 protein levels in Arrdc5^–/– testes (0.259±0.086, mean±s.d.) compared with Arrdc5^+/+ testes (1.000±0.178) (Fig. 4C,D), whereas there was no significant difference in mRNA level (Fig. S3A). Sad1 and UNC84 domain-containing 5 (SUN5) has been reported to play an essential role in anchoring the sperm head to the tail, and Sun5-null mice show a globozoospermatozoa-like acephalic spermatozoa (Shang et al., 2017; Zhang et al., 2021b). Western blotting showed about a 50% reduction (P=0.013) of SUN5 protein levels in Arrdc5^–/– testes (0.524±0.188) compared with Arrdc5^+/+ testes (1.000±0.033) (Fig. 4E,F), whereas there was no significant difference in mRNA level (Fig. S3B). Because SUN5 migrates to the coupling apparatus of the sperm during sperm head elongation and differentiation (Shang et al., 2017), we performed immunofluorescence staining in testis smears, and we found that ARRDC5 deficiency led to incorrect localization of SUN5 at stage 13-16 during spermatogenesis (Fig. 4G). Finally, mature sperm from the Arrdc5^–/– cauda epididymis showed a bent head and no detectable SUN5 (Fig. 4H). Previous work has shown that NDC1 is involved in sperm head and tail formation during mammalian spermiogenesis and that SUN5 deletion causes similar defects in sperm head and tail anchoring as Arrdc5 knockout mice sperm (Lai et al., 2016; Shang et al., 2017; Yeh et al., 2015; Zhang et al., 2021b). Taken together, we speculated that ARRDC5 might affect spermatogenesis by influencing NDC1 and SUN5.

Because ARRDC5 deletion caused a sharp decrease in the expression of NDC1 and SUN5 (Fig. 4C,E), we asked how ARRDC5 affects NDC1 and SUN5 during spermatogenesis. Previous studies (Zbieralski and Wawrzycka, 2022) have shown that ARRDC5 shares the overall arrestin-like domains with other α-arrestins (Fig. 5A), and immunofluorescence staining for overexpressed ARRDC5 in HeLa cells indicated that ARRDC5 is a membrane-bound protein. Furthermore, ARRDC3, another member of the α-arrestins, has been reported to be a membrane protein and to play a role in regulating the trafficking of G protein-coupled receptors (Tian et al., 2016; Zbieralski and Wawrzycka, 2022). These findings, combined with the GO analysis presented here suggest that ARRDC5 might have a similar function to ARRDC3 in protein transport. Therefore, we co-expressed ARRDC5 with other ARRDCs in HeLa cells and performed immunofluorescence staining. The results showed that ARRDC5 mainly localized on the cell membrane and co-localized with ARRDC3 and ARRDC4 (Fig. 5B), which supported our hypothesis that ARRCD5 might play a similar function as ARRDC3 and ARRDC4. Based on these results, we asked if ARRDC5 is directly involved in transporting NDC1 and/or SUN5. Thus, we co-expressed ARRDC5 with NDC1 and SUN5 in HeLa cells and performed immunofluorescence staining. However, ARRDC5 did not co-localize with either NDC1 or SUN5 (Fig. S4A,B), thus suggesting that ARRDC5 does not function by directly interacting with NDC1 and SUN5.

To determine how ARRDC5 affects NDC1 and SUN5, we further analyzed the candidate proteins enriched in the protein transport pathway (Fig. 4B). We found that the SEC22A protein was nearly undetectable in Arrdc5^–/– testes according to our mass spectrometry results, but there was no significant difference in the mRNA level (Fig. S2C). SEC22 homolog A, a vesicle trafficking protein (SEC22), belongs to the SEC22 family of vesicle trafficking proteins, one of which (SEC22) has been reported to play diverse roles in vesicle trafficking, membrane fusion and autophagy (Adnan et al., 2019). Because SEC22A might play an intermediate role in protein transport, we hypothesized that ARRDC5 might participate in protein transport by affecting SEC22A. To test this hypothesis, we originally tried to validate the possible role of ARRDC5 in regulating SEC22A-mediated transportation in testis; however, owing to lack of a commercially available SEC22A antibody, we co-expressed SEC22A with ARRDC5, NDC1 and SUN5 in HeLa cells and performed immunofluorescence staining to validate the relationship between ARRDC5 and SEC22A. We found that SEC22A was partly co-localized with ARRDC5 and co-localized with NDC1 and SUN5 (Fig. 5C). Co-immunoprecipitation (co-IP) in HeLa cells also showed that SEC22A directly interacted with ARRDC5, NDC1 and SUN5 (Fig. 5D-F). Because SEC22A is a vesicle trafficking protein and is able to shuttle through cells, it is likely that SEC22A acts as a bridge to establish the link between ARRDC5 and NDC1 and SUN5.

Despite Arrdc5^–/– sperm showing bent heads and reduced motility, the structure of the entire head was relatively intact. We therefore performed ICSI to see whether it could rescue the fertilization failure of Arrdc5^–/– sperm and found that Arrdc5^–/– sperm could accomplish fertilization with wild-type oocytes and that the resulting embryos could ultimately develop into blastocysts (Fig. 6). The result suggested that ICSI might be a treatment strategy for male infertility due to ARRDC5 deficiency.

In this study, we revealed the important role of ARRDC5 in spermatogenesis by affecting the attachment of the HTCA to the nuclear envelope. Deficiency in ARRDC5 led to abnormal sperm morphology and failed capacitation, and thus a failure of fertilization. Our mechanistic studies revealed a possible pathway through which ARRDC5 affects SEC22A in a direct or indirect way, which carries NDC1, SUN5 and other HTCA-related proteins to the correct location and initiates the attachment of the sperm head and tail during spermatogenesis (Fig. 7). ARRDC5 deficiency likely disrupts this protein transport process, and the ectopic localization of NDC1, SUN5 and other HTCA-related proteins without SEC22A transport might be disrupted.

As we prepared this article, Giassetti et al. reported that ARRDC5 was required for normal sperm morphogenesis and male fertility (Giassetti et al., 2023). Both studies observed the abnormal morphology and reduced motility in Arrdc5^–/– sperm with capacitation and fertilization failure. The study of Giassetti et al. demonstrated a conserved role in mammalian by analyzing expression of ARRDC5 was in multispecies testicular cells. Although our study carried out a deeper exploration and found the possible molecular mechanism by which ARRDC5 functions during spermatogenesis, by combining Giassetti et al.’s research with ours, we believe there will be a more complete explanation of the biological functions of ARRDC5 in mammals.

During spermiogenesis the HTCA mediates the integration of the sperm head and tail to ensure subsequent high-speed motility, and HTCA defects lead to the development of abnormal sperm (Wu et al., 2020). Several proteins related to the HTCA have been reported. SPATA6 is required for normal assembly of the sperm connecting piece and for the tight head-tail conjunction, and the absence of SPATA6 in mice impairs the development of the connecting piece (Yuan et al., 2015). PMFBP1 is essential for the attachment of the coupling apparatus to the sperm head, and humans with PMFBP1 mutations and Pmfbp1-deficient mice have both been shown to suffer from acephalic spermatozoa syndrome (Zhu et al., 2018). A recent study showed that centlein (CNTLN) acts as a linker between SUN5 and PMFBP1 to maintain the integrity of the HTCA (Zhang et al., 2021a), and in our Arrd5^–/– mouse testis, SUN5 had reduced protein levels and an incorrect localization; thus, we suspected that the protein would be degraded. However, the phenotype of Arrdc5 knockout sperm during spermatogenesis is not as severe as that of Sun5-null mice, and we suspect that this might be due to the dose of protein levels. Low doses of SUN5 may also provide weak support for HTCA assembly. Although the HTCA can be assembled according to our TEM results, it cannot fully attach to the nuclear envelope, which may be the reason for the sperm head hanging with a loose anchoring, which results in the bent-head phenotype.

The mass spectrometry and GO analysis suggested that many other proteins associated with spermatogenesis and sperm function are also affected by ARRDC5 deficiency. For example, Serpina5 knockout mice are infertile due to abnormal spermatogenesis resulting from destruction of the Sertoli cell barrier (Uhrin et al., 2000), and blocking the SERPINA5 protein reduces the success rate of IVF and reduces the number of sperm that combine with the ZP of oocytes (Cao et al., 2022). SERPINA5 also showed a sharp reduction in expression in the Arrdc5^–/– testis in this study, which might be another possible reason for the fertilization failure with Arrdc5^–/– sperm. SPATA18 has also been reported to be associated with spermatogenesis (Bornstein et al., 2011), and SPATA18 protein expression was also reduced in the Arrdc5^–/– testis. POM121 is a transmembrane nucleoporin that, together with nucleoporin 210 (NUP210) and NDC1, forms a ring of transmembrane nucleoproteins and helps nuclear pore complexes anchor to the nuclear envelope (Strambio-De-Castillia et al., 2010). Both POM121 and NDC1 were reduced in Arrdc5^–/– testis, so it might be the combined loss of the two proteins that results in abnormal formation of sperm head-tail junctions during spermatogenesis. On the other hand, TCP11 has been reported to be important for sperm progressive motility and sperm capacitation through the cyclic AMP/protein kinase A pathway (Castaneda et al., 2020; Fraser et al., 1997). We found that TCP11 was also reduced in the Arrdc5^–/– testis, which we suspected might affect the capacitation of Arrdc5^–/– sperm. All of these proteins can be further studied in the future to gain a more comprehensive understanding of spermatogenesis.

The high conservation of ARRDC5 in mammalian species and our studies on ARRDC5 in mouse models suggest an essential role for ARRDC5 in spermatogenesis in mice. In addition, we show that ARRDC5 might be involved in the anchoring of the sperm head to the tail during spermatogenesis by affecting SEC22A-mediated SUN5 and NDC1 transport and localization. These findings suggest that ARRDC5 might be a new genetic marker for pathogenic variant-targeted sequencing in individuals with similar phenotypes. Finally, we found that ICSI enabled successful rescue of fertility, thus suggesting a therapeutic strategy with which to overcome male infertility caused by ARRDC5 deficiency.

All of our animal experiments were approved by the Ethics Committee of the Medical College of Fudan University and followed all animal testing guidelines and regulations. The mice were euthanized by cervical dislocation according to The AVMA Guidelines for the Euthanasia of Animals: 2020 Edition. The approval number associated with the use of mice is 2021JSIBS-007 of Fudan University.

We fused the coding sequences from mouse tissue cDNA template with FLAG, mclover3 and HA, and inserted them into the pXF4H, pCR3.1 or pCMV6 plasmids to obtain pXF4H-Arrdc5-FLAG, pXF4H-sec22a-FLAG, pXF4H-ndc1-HA, pXF4H-sun5-HA, pCR3.1-Arrdc5-FLAG, pCR3.1-Arrdc3-mclover3, pCR3.1-Arrdc4-mclover3, pCR3.1-Sun5-mclover3, pCR3.1-Ndc1-mclover3 and pCMV6-Sec22a-HA.

Primary antibodies were rabbit anti-SUN5 (17495-1-AP, Proteintech), anti-NDC1 (A17727, Abclonal), anti-vinculin (1390T, CST, Danvers), anti-HA (3724T, CST), anti-FLAG (F7425, Sigma-Aldrich), anti-α-Tubulin (T6199, Sigma-Aldrich), mouse anti-FLAG (GNI4110-FG, GNI) and anti-phosphotyrosine (ab10321, Abcam).

The HeLa cell line was obtained from the Cell Bank of the Shanghai Institute for Biological Sciences, the Chinese Academy of Sciences (Shanghai, China). HeLa cells were cultured in high-glucose Dulbecco's modified Eagle's medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (F2442, Sigma-Aldrich) and 1% penicillin and streptomycin (Gibco) at 37°C with 5% CO₂. Transfection of plasmids was performed with PolyJet (SL100688, SignaGen Laboratories) according to the manufacturer's instructions.

The Arrdc5^–/– mice were generated by CRISPR/Cas-mediated genome engineering in the C57BL/6J mouse model (constructed by Cyagen Biosciences). The Cas9 and gRNA (Table S1) were co-injected into fertilized mouse eggs to generate targeted knockout offspring. F0 founder animals were genotyped by PCR followed by sequence analysis and were bred until homozygous knockout offspring were obtained.

Sperm for motility was referred to Martinez (Martinez, 2022). Sperm from the mouse cauda epididymis were backflushed and incubated in M2 medium for 15 min at 37°C to gain motility. The sperm were subsequently homogenized and transferred to a Leja slide, and their concentration and motility characteristics were assessed using a computer-assisted sperm analysis system (Hamilton Thorne IVOS II).

Sample preparation was performed according to a previously described method (Zhang et al., 2023). For scanning electron microscopy (SEM), washed sperm of adult male mice were spread on a 2×3mm glass slide and air dried. The glass slides were then fixed with 2.5% (vol/vol) glutaraldehyde at 4°C overnight. Images were obtained on an SU801microscope (HITACHI SU8010, Japan) according to standard SEM procedures.

For transmission electron microscopy (TEM), the dissected testes and washed sperm of adult male mice were fixed with 2.5% (vol/vol) glutaraldehyde at 4°C overnight. Tissues were cut into small pieces of ∼1 mm³ and immersed in 1% OsO₄ for 2 h at 4°C. The samples were dehydrated through graded ethanol and acetone, and finally embedded in resin for ultrathin sectioning with an ultramicrotome. The samples were then stained with uranyl acetate and lead citrate, and photographed using a Philips CM-120.

Total RNA from mouse tissues were extracted using an RNeasy Mini Kit (Qiagen). After genomic DNA depletion, 1 μg RNA was used for reverse transcription using a PrimeScript RT reagent Kit with gDNA Eraser (RR047B, TaKaRa). The primer concentration was 50 pmol/ 20 μl. The reverse transcriptase was in the PrimeScript RT Enzyme Mix and the concentration was 200 U/μl. The cDNA sample was diluted 10-fold for qRT-PCR. qRT-PCR was performed using a Quant Studio 5 Real-Time PCR System (Applied Biosystems, Thermo Fisher Scientific), with TaKaRa TB Green Premix Ex Taq (Tli RNaseH Plus). The PCR reactions were performed as follows: hold stage, 50°C for 2 min and 95°C for 2 min; PCR stage, 40 cycles of 95°C for 10 s and 60°C for 30 s; melt curve stage, 95°C for 15 s, 60°C for 1 min and 95°C for 15 s. Design and Analysis Software v2.6.0 (Thermo Fisher Scientific) was used for qPCR analysis. Cq values<38 and Cq shift within 0.5 was valid. Actb and Gapdh were chosen as two housekeeping genes by the Minimum Information for Publication of Quantitative Real-Time PCR Experiments (MIQE) guidelines (Bustin et al., 2009). The relative expression level equaled k×2^−ΔΔCq, where ΔCq=(Cq (candidate gene) – Cq (housekeeping genes)) – ((CqWT(candidate gene)−CqWT (housekeeping genes)) (Pfaffl, 2001). For comparison of related gene expression in wild-type and knockout testes, wild type was chosen as a reference gene and normalized as 1 for relative expression level. Four technical replicates were performed for each sample assay. qPCR specificity has been validated (Fig. S5A,B and Table S2).

At 6-8 weeks of age, C57BL/6J wild-type female mice were mated with 10- to 12-week-old wild-type and knockout male mice for 4 months (two females per male). The vaginal plugs of the female mice were checked to confirm their mating. The number of litters and pups per litter were calculated, and Student's t-test was performed to compare averages in different experimental groups. P<0.05 was considered to be significant.

At 7-8 weeks of age CD-1 (ICR) female mice were superovulated by injecting pregnant mare serum gonadotropin (Ningbo Second Hormone Factory, Zhejiang, China) followed by human chorionic gonadotropin (Ningbo Second Hormone Factory, Zhejiang, China) 46-48 h later. After 12 h, sperm were collected from the cauda epididymis of male mice and capacitated in a drop of HTF at 37°C with 5% CO₂ for 1 h. Cumulus-oocyte complexes from superovulated female mice were then fertilized with capacitated sperm in a drop of HTF at 37°C with 5% CO₂ for 6 h, after which the 2PN zygotes were selected and transferred into KSOM medium (M1430, EasyCheck) at 37°C with 5% CO₂ for blastocyst culture.

At 7-8 weeks of age B6D2F1 (C57BL/6×DBA2) female mice were superovulated as described above. After 13 h, sperm were collected from the cauda epididymis of male mice and capacitated in HTF using the swim up method at 37°C with 5% CO₂ for 1 h. Oocytes were collected from superovulated female mice and then freed from the cumulus by treatment with hyaluronidase (H4272, Sigma-Aldrich). Sperm preparation before injection was performed according to a previously described method (Kimura and Yanagimachi, 1995). The sperm were decapitated by ultrasonication (Skymen JP020S), and then single sperm heads were injected into the oocytes using a micromanipulator with a Piezo apparatus (Eppendorf). Injected oocytes were transferred into KSOM medium at 37°C with 5% CO₂ for subsequent development.

Testes from three pairs of wild-type and knockout mice were homogenized in RIPA lysis buffer (WB0101, Shanghai Wei AO Biological Technology) with 1% protease inhibitor cocktail (B14001, Bimake). The protein lysate was analyzed by mass spectrometry-based label-free quantitative proteomics commissioned by the Key Laboratory of Glycoconjugates Research Ministry of Public Health and Institutes of Biomedical Sciences of Fudan University. The database search was performed using mouse protein sequences obtained from UniProt allowing up to two missed cleavages. Fold change and false discovery rate values were calculated to analyze the expression of different proteins. Statistical analysis was performed using the Limma Bioconductor package to define the probability of variation (P-value) of each protein between two groups. Proteins with P<0.05 were considered significantly different (Tardif et al., 2022; van Ooijen et al., 2018). GO term enrichment analysis was performed using DAVID Bioinformatics Resources (https://david.ncifcrf.gov/), including biological processes (in the process, BP), molecular function (MF) and cell component (CC) enrichment. P-value cutoff=0.05 and the rest defaulted to original settings. The top 10 of BP, CC and MF were chosen for analysis.

The proteins from mouse testes were extracted using RIPA lysis buffer with 1% protease inhibitor cocktail, and the supernatants were collected after centrifugation at 12,000 g for 10 min. The protein concentration was quantified by the bicinchoninic acid method, and equal amounts of protein were separated by 10% SDS-PAGE gels and transferred onto nitrocellulose filter membranes. The membranes were blocked in 5% non-fat milk for 1 h and incubated with primary antibodies overnight at 4°C, and subsequent secondary antibody incubation was performed to detect the primary antibodies. The blots were finally captured using ECL western blotting substrate (180-5001, Tanon). The intensity and quantitation of western blotting results was measured by ImageJ/FIJI (NIH) (Schindelin et al., 2012) from three pairs of independent biological samples. Normal IgG antibodies and originals of western blotting are shown in Fig. S6A,B, Fig. S7 and Fig. S8.

Plasmids with FLAG-tag were co-transfected with the plasmids with HA-tag into HeLa cells. At 36 h after transfection, cell proteins were extracted with NP-40 lysis buffer (50 mM Tris, 150 mM NaCl and 0.5% NP-40 at pH 7.5) containing 1% protease inhibitor cocktail (Bimake). Total protein was incubated with anti-FLAG beads (B23102, Bimake) at 4°C for 3 h. The beads were washed with lysis buffer three times and then boiled with SDS loading buffer for western blotting with anti-FLAG (1:3000 dilution) and anti-HA antibodies (1:3000 dilution).

Round spermatids and elongated spermatids from adult mouse testes were isolated according to a previous study (Bai et al., 2018). Briefly, the testicular tissues were stripped from the albuginea membrane and digested with collagenase IV (1 mg/ml, C4-BIOC, Sigma-Aldrich) for 10 min at 37°C. The dispersed testicular tissues were washed with DMEM and centrifuged at 500 g. The pellet was further digested with 0.25% Trypsin (Gibco) containing DNase I (1 mg/ml, M0303L, NEB) for 10 min at 37°C and centrifuged at 500 g. The pellet was then resuspended in DMEM and obtained a mixture of spermatogenic cells.

Testicular spermatogenic cells were spread on glass slides and fixed in 4% PFA at room temperature for 15 min, rinsed in PBS three times, permeabilized in 0.3% Triton X-100 in PBS for 30 min and then blocked with blocking buffer (3% BSA in PBS) for 1 h at room temperature. The primary antibody was added and incubated at 4°C overnight, followed by incubation with secondary antibody (1:500 dilution) for 1 h at 37°C. Hoechst 33342 (1:500 dilution, 561908, BD Biosciences) was added for 15 min at room temperature to label the DNA. All images were captured on a laser scanning confocal microscope (LSM 880 with Airyscan, Zeiss).

For 2PN zygotes and unfertilized meiotic oocytes, the immunofluorescence staining was performed as described previously (Liu et al., 2022). Briefly, 2PN zygotes and unfertilized meiotic oocytes were fixed with 2% PFA, then permeabilized in 0.5% Triton X-100 and then blocked with blocking buffer (3% BSA, 0.1% Tween 20 and 0.01% Triton X-100 in PBS). Spindles were stained with an anti-β-tubulin antibody (1:500 dilution, F2043, Sigma-Aldrich). Cortical granules were stained with Lens Culinaris Agglutinin (DyLight 649 DL-1048-1) and Hoechst 33342 (1:500 dilution, MedChemExpress) was added to label the DNA. All images were captured on an EVOS M7000 Imaging System (Invitrogen). Normal IgG antibodies of immunofluorescence staining are shown in Fig. S6C,D.

All data are mean±s.d. Statistical analysis was performed using Prism GraphPad 9 software. The statistical significance was analyzed using a two-tailed Student's t-test or a χ² test, and P<0.05 was considered statistically significant. Quantitation of western blotting results was performed using ImageJ software, and all results shown are representative of at least three independent experiments.

We thank Yinghua Jiang for providing help with confocal imaging and the Public Technology Platform of Shanghai Medical College.