Ocln-KO and B1-V-ATPase-EGFP mice were provided by Prof. Eveline E. Schneeberger and Prof. Sylvie Breton from MGH/Harvard Medical School. The background of these mice has been published previously (23, 24). All animals were housed in a standard specific pathogen-free animal house. Since male Ocln-KO mice are sterile, these mice were produced from heterozygous pairings; otherwise, mice were bred with wild-type (WT) mice (C57BL/6) for maintenance. All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of ShanghaiTech University.

WT and Ocln-KO male mice aged 6, 9, 12, 20, 40 and 60 weeks were mated with two female wild-type mice. At least three pairs of mice of each age were mated for at least two weeks. To determine male mice mating behavior, the female mice (12-week-old) from the mating cage were examined for the plugs and the vaginal plugs were smeared, and the spermatozoa were observed under microscopy.

For in vitro fertilization (IVF) and intracytoplasmic sperm injection (ICSI) assays, sperms were directly obtained from caput or cauda epididymis of WT and Ocln-KO males, the assays were performed as previously reported (25, 26). For computer-assisted sperm analysis (CASA) assays, sperm from the cauda or caput epididymidis were incubated in the EKRB capacitation medium and analyzed using an HTM-TOX IVOS sperm motility analyzer (Hamilton-Thon Research, v14), as we published previously (25, 27).

The testes and epididymides of WT and Ocln-KO male mice were fixed with paraformaldehyde-lysine-periodate (PLP), embedded in O.C.T. compound (Sakura) and sectioned at −18°C or −20°C (Leica Microsystems GmbH), as described previously (28, 29). Cryosections were rehydrated in PBS and stained with Hematoxylin and Eosin Staining kit (Beyotime). Images were acquired using the slide scanner Olympus VS120.

The immunofluorescent labelling procedure was performed as described previously (25). Briefly, Cryosections of testes and epididymides tissue of WT and Ocln-KO male mice were rehydrated in PBS, antigen retrieved with 1%w/v SDS, blocked with 1% w/v BSA and then incubated with primary antibodies, followed by incubated with the secondary antibodies. All slides were mounted in Vectashield with DAPI. The TUNEL assay was performed as the manual of Click-iT Plus TUNEL Assay Kit (Invitrogen). Images were acquired using laser scanning confocal microscopy (LSM880, Carl Zeiss or Nikon A1R) and analyzed with the original software before imported into Adobe Photoshop for typesetting. Details of the antibodies used in this study are listed in Supplementary Table 1 .

Western blot was performed as published previously (27). Cell lysates were subjected to SDS-PAGE and then transferred to PVDF membranes. The membranes were immunoblotted with the indicated antibodies and visualized by SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermo Scientific). Data were acquired using an Amersham Imager 600 or 680 system (GE Healthcare, NJ, USA). Antibodies details are listed in Supplementary Table 1 .

DC2 cell lines were cultured in an incubator at 33°C, 5% CO₂ with Full-IMDM supplemented with 1 nM 5α-dihydrotestosterone and containing 10% (v/v) FBS, penicillin and streptomycin, as described previously (30). Ocln-KO cells were generated in DC2 cells by using the CRISPR/Cas9 system. The Ocln sequence in the second exon 5′-CGGCTGAGAGAGCATCGGC-3′ (Ocln–sgRNA) was targeted with pGL3-U6-sgRNA-PGK-Puro (Addgene). DC2 cells were then co-transfected with Ocln–sgRNA-expressing vector and Cas9 expressed vector pST1374-NLS-flag-linker-Cas9 (Addgene) by Lipofectamine 2000 and selected with puromycin and blasticidin for 24  h. Single cells were isolated by cell sorting. The single clones were validated by immunoblotting analysis and DNA sequencing. The two plasmids were gifts from Xingxu Huang (31, 32).

For pH measurement in vas deferens of adult WT and Ocln-KO male mice, the animals were anesthetized and both sides of the vas deferens were dissected, the blood was removed and kept the vas deferens under a fat pad. The luminal liquid in the sectioned vas deferens was squeezed out and dropped on the pH paper (Hydrion, pH range 5.5-8.0) and recorded it by video. Data values were measured on captured images 10- second after the sample liquid touched the pH paper and the different standards in grey scale were used for calibration using Fiji software.

For lipid extraction, the proximal epididymides from three WT or Ocln-KO mice were pooled into one sample. The distal epididymidum was treated the same way. Three samples per group were used in this study. Lipids were extracted using a modified version of the Bligh and Dyer’s method as described previously (33). Lipidomic analyses were conducted at LipidALL Technologies using an Agilent 1290 II UPLC coupled with Sciex QTRAP 6500 PLUS as reported previously (34). Glycerol lipids including diacylglycerols (DAG) and triacylglycerols (TAG) were quantified using a modified version of reverse phase HPLC/MRM. Heatmaps and volcano plots of TAG and DAG analyses were performed on the free online data analysis website Tutools platform (https://www.cloudtutu.com/).

The protein preparation and TMTs labelling workflow have been described previously (35). Briefly, total proteins were extracted from three pairs of WT and Ocln-KO mice whole epididymis by lysis buffer (50 mM (NH₄)₂CO₃, 8 M urea, 1 mM DTT and protease inhibitor) separately and quantitation using BCA protein assay kit. After alkylated by iodoacetamide and digested by trypsin, the peptide was labeled by TMTs label regent following the manufacturer’s protocol. The resulting peptides were separated and analyzed on an Easy-nLC 1000 system coupled to a Q Exactive HF (Thermo Scientific). The LC-MS/MS procedure and data analysis were performed as reported previously (36).

Total RNA samples of whole epididymis were extracted from three pairs of WT and Ocln-KO mice by Trizol reagent separately according to the manufacturer’s instructions. The RNA quality was checked by Bioanalyzer 2200 (Agilent) and kept at -80°C.The RNA with RIN >6.0 is about to undergo rRNA depletion. cDNA libraries were constructed for each pooled RNA sample by using the NEBNext^® Ultra™ Directional RNA Library Prep Kit for Illumina according to the manufacturer’s instructions. The tagged cDNA libraries were pooled in equal ratio and used for 150 bp paired-end sequencing in a single lane of the Illumina HiSeqXTen. Before read mapping of pair-ends, clean reads were obtained from the raw reads by removing the adaptor sequences, reads with >5% ambiguous bases (noted as N) and low-quality reads containing more than 20 percent of bases with qualities of<20. The clean reads were then aligned to mouse genome (version: GRCm38 NCBI) using hisat2. Differential gene and transcript expression analysis of RNA-seq experiments with HTseq (37) was used to count mRNA, and the RPKM method was used to determine gene expression.

For RNA-Seq data, we applied the DESeq2 algorithm (38) to filter the differentially expressed genes (DEGs); for significance analysis, P values and false discovery rate (FDR) analysis were obtained from the following criteria (39): mRNA under the following criteria: i) Fold change > 2 or< 0.5; ii) FDR< 0.05. For proteomic data, two-tailed student’s t-test was used to verify the significance of the differences between each comparison, proteins with the fold change > 1.2 or< 0.833 and P-value< 0.05 were chosen.

Comparisons were performed using the 2-tailed paired or unpaired Student’s t-test for only two groups, or one-way or 2-way-ANOVA for the comparisons of more than two groups at one- or two-levels, respectively. Significance was defined as a P-value< 0.05. Prism 8.0 (GraphPad) was used for all statistical analyses.

Ocln-KO male mice were reported to be infertile (23). We also monitored the fertility of the Ocln-KO male mice over a long period (from 6-week-old till 69-week-old), which confirmed their sterility after 9-week-old. However, at juvenile ages of about 8 weeks, 1 out 8 Ocln-KO male mice was able to fertilize female mice to give birth to live pups ( Figure 1A ), suggesting the male sterility possibly an acquired phenotype as the mice grow. We then investigated if the male infertility could be due to problems of the epididymis. The Ocln-KO males had smaller overall in size compared to the wild-type (WT) mice, although no difference in proportional size of the epididymis or other organs was found ( Figures 1B, C ). Through detailed morphological analysis, we frequently found the testis of the Ocln-KO was morphological normal ( Supplementary Figures S1A-C ). Whereas the caput (CPT) epididymidis in the head portion of epididymis was enlarged and cauda (CD) epididymidis in the tail portion was degenerated in the Ocln-KO mice as compared to those of the WT ( Figures 1D-F ). Strikingly, in caput to corpus (CPT-CPS) junctional segments of the Ocln-KO, sperm cells were found infiltrated into the interstitial area causing tissue enlargement in this region ( Figure 1G ), a pathology similar to epididymal sperm granulomas (41, 42).

We next examined the epididymal sperm from Ocln-KO mice in comparison with the WT. CASA showed that CPT sperm from the epididymal head were mostly quiescent in both WT and Ocln-KO mice, although CD sperm from the epididymal tail of WT mice gained substantially higher motility and progressiveness as compared to the CPT sperm ( Figure 1H ). However, such a CPT to CD increase in sperm motility was totally absent in Ocln-KO mice. In addition, in WT mice, the sperm count increased from CPT to CD suggesting sperm condensation through epididymal transit. Whereas, in the Ocln-KO mice, the CD sperm count was found at a very low level even lower than their CPT sperm count. Consistently, although the Ocln-KO mice copulated normally with WT females, next-morning analyses of virginal plugs of WT females after copulation revealed very few sperm, compared to control male mice ( Supplementary Figure S1D ). We next took the CD sperm for analyzing their fertilizing capacities in vitro. The IVF assay by incubating CD sperm with WT eggs showed significant reduction in the efficiency of Ocln-KO CD sperm to generate two-cell or blastocyst stage of embryos, as compared to that of WT CD sperm ( Figure 1I ). However, when we injected WT eggs with Ocln-KO CD spermatozoa by ICSI, the eggs were successfully activated and fertilized at a similar rate compared to WT CD spermatozoa ( Figure 1J ). After transplantation in the female oviduct, the ICSI fertilized eggs with WT of Ocln-KO sperm developed normally to birth ( Supplementary Figures S1E, F ), suggesting that the Ocln-KO CD sperm were genetically fine. All these observations suggest that male infertility in Ocln-KO mice is largely due to epididymal-dependent impairment of sperm transportation and thereby motility and fertilization capacity.

To further determine the role of OCLN in the epididymis, we investigated the cellular localization of OCLN on the epididymal cryosections of WT adult mice. The immunofluorescent staining with an anti-OCLN antibody revealed that OCLN protein was enriched in the tight junctions (TJs) of principal cells of the initial segment (IS) and CPT epididymidis, as well as the lateral membranes of WT or Ocln-heterozygous CPT principal cells ( Figure 2A ), but hardly detectable in the distal corpus (dCPS) and CD epididymidis in the tail region of epididymis ( Figure 2B ). No specific OCLN immunofluorescence was detected in Ocln-KO CPT principal cells, confirming the anti-OCLN antibody specificity. Using the same anti-OCLN antibody, only one band was detected in the WT DC2 epididymal cells with a size corresponding to OCLN protein, which was significantly diminished in the Ocln-mutant DC2 cells ( Figure 2C ).

To explore how Ocln-KO leads to the observed epididymal defected, we profiled the transcriptomes and proteomes of the whole epididymis from adult Ocln-KO mice in comparison to those of WT mice to find differentially expressed genes (DEGs). An integrative approach was applied to the analysis of DEGs obtained by combining RNAseq and proteomic datasets. Through the Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis, the results showed that the gene clusters enriched in the acid secretion, and metabolic pathways, including lipid and amino acid metabolic pathways, were the main nodes of downregulation ( Figure 3A , asterisks and triangles). In addition, the significantly downregulated pathways in the Ocln-KO epididymis, including synaptic vesicle cycle and collecting duct acid secretion ( Figure 3A , triangles), also implied the impairment in trafficking of proton pump V-ATPases, which are highly expressed in clear cells and responsible for the luminal acidification of epididymis and maintaining sperm in a quiescent stage (43).

Bioinformatic analysis using Venn diagram of the shortlisted DEGs with significantly downregulated pathways revealed that almost all of them were metabolic enzymes, of which about 40% were involved in lipid catabolic processes ( Figure 3B ). Heatmaps of metabolically clustered genes showed that top enzymes were involved in lipid catabolism ( Figure 3C ). Two of the top shortlisted enzymes Ptgds and Pla2g12a implicated in the metabolism of the fatty acid arachidonic acid. Another top-listed enzyme was CES1, which is also known as a lipase of carboxylic esters to release carboxylic acids and free fatty acids (FFA), such as arachidonic acid (44, 45). Ldhb relating to pyruvate metabolism was also a top shortlisted enzyme, while Psph and Cbs were shortlisted in the amino acid metabolism cluster. Another cluster of top listed genes were the subunits of proton-pump V-ATPase family, such as B1-V-ATPase (Atp6v1b1), a key player in luminal acidification of epididymis, keeping sperm in a quiescent stage (5, 46, 47).

Given the observed downregulation of V-ATPase genes in the Ocln-KO epididymis, we analyzed the cellular localization pattern of clear cells using V-ATPase as the marker. Our results showed that, in the epididymis of Ocln-KO compared to WT mice, not only the V-ATPase-positive clear cells and their cellular debris were found in the epididymal lumen, especially in the caput region. In addition, many keratin-18-positive epithelial cells were shed into the lumen of caput epididymidis ( Figure 4A ). Using TUNEL staining, we observed increased apoptotic index in Ocln-KO epididymis compared to WT controls, including some epithelial cells of the epididymal mucosa and some somatic cellular DNA fragments in the lumen ( Figure 4B ). In the TUNEL positive control wildtype epididymis cryosections treated with DNase I to induce DNA breaks in genome, all the somatic cells showed labelling for TUNEL. Notably, DNase I-induced DNA fragmentations did not occur in sperm cells, suggesting the DNA in sperm cells is insensitive to DNase I and well-protected from somatic enzymatic challenges.

To determine whether the loss of clear cells might be due to dysfunctional differentiation during postnatal development, in addition to increased programmed cell death during epithelial homeostasis, we harvested epididymal tissue from different ages and labelled them for the clear cell marker B1-V-ATPase. Our results showed that the number of clear cells of 2-week-old KO epididymides was only mildly different from that of WT (not shown), but it obviously decreased at puberty and in adulthood ( Figure 4C ). A trend towards increased TUNEL-positive apoptotic epithelial cells was observed in the lumen and epithelium of the Ocln-KO epididymis compared to WT, although statistics showed no difference ( Figure 4D , P<0.06). In the luminal contents, the apoptotic index was decreased from CPT to CD epididymidis, whereas the apoptotic index was increased in the epithelial cells of epithelium ( Supplementary Figure S2 ). The co-localization of TUNEL and B1-V-ATPase was observed only in some but not all B1-V-ATPase cellular debris in the luminal contents of caput epididymal, whereas some B1-V-ATPase-positive cells were not positive for TUNEL ( Supplementary Figure S2 ). Quantification results showed that the numbers of V-ATPase-labelled clear cells in the epididymis were prominently decreased in the pubertal and adult mice but not statistically in early age of Ocln-KO mice ( Figure 4E ).

The acidity of luminal fluid of proximal vas deferens—an organ connecting successively to caudal epididymis with the presence of abundant clear cells—was also significantly decreased, suggesting a disorder of epididymal luminal microenvironment after Ocln deletion. Consistently, the luminal pH of Ocln-KO group was significantly higher than that of WT group ( Figure 4F ).

To determine whether the fatty acid homeostasis is disrupted after Ocln deletion, we determined the contents of stable fatty acid pools triacylglycerols (TAG) and their derivatives diacylglycerols (DAG) in the epididymis ( Figure 5 and Supplementary Figure 3 ). The lipidomic results showed that total of eighty species of TAG components and eighteen DAG components were determined in the mouse epididymis ( Figure 5A ). The total mean TAG in the mouse whole epididymis was ~5.5 µmol per gram of tissue while DAG was ~0.23 µmol per gram of tissue ( Figure 5A ). The volcano plot revealed that only a few specific DAG and TAG components had significantly different levels in the proximal versus distal epididymis ( Figure 5B ). Heatmap analysis showed the heterogeneous levels of specific DAG and TAG components in the proximal and distal epididymis of WT mice ( Figures 5C, D ). Regarding TAG, only two species showed statistically lower levels in the distal epididymis compared to the proximal epididymis, including TAG containing double-bonded unsaturated long-chain fatty acids ( Figure 5E ). As for DAG, most of the statistically distinct components had higher levels in the distal compared to the proximal epididymis of WT mice, including the DAG with double-bonded unsaturated long-chain fatty acids ( Figure 5F ).

In the epididymis of Ocln-KO versus WT mice, the total mean contents of diacylglycerols (DAGs) and triacylglycerols (TAGs) did not differ between the two groups ( Figure 6A ). When all DAG components of proximal and distal epididymis were compared, respectively, the DAG components were overall significantly different in Ocln-KO epididymis from that of WT mice ( Figure 6B ). In the proximal epididymis, three DAG components were significantly increased, including the DAG containing double-bonded fatty acid DAG38:5(16:0/22:5) ( Figures 6C-E ). Four kinds of DAG components were significantly increased in the distal epididymis, including the most accumulated species DAG38:4(18:0/20:4), which contains the double-bonded 20-carbon arachidonic fatty acid chain ( Figures 6F-H ). For the TAG, one component with double-bonded long fatty acid chain, TAG58:7(22:6), was significantly decreased in the KO compared to WT epididymis and between proximal and distal epididymis of KO epididymis ( Figure 6I ). Whereas in the distal epididymis of KO mice, the double-bonded long fatty acid chain containing TAG components, TAG46:1(16:1) and TAG58:8(22:6), were significantly increased ( Figure 6J and Supplementary Figure 4 ). Expression analysis in the transcriptomes and proteomes showed only slight changes in the expression of DAG kinases ( Figure 6K ) and moderate downregulation of a few lipases of DAG or TAG ( Figure 6L ).

To determine the cellular localization of the shortlisted fatty acid catabolic enzymes, immunofluorescent labelling was applied on epididymal cryosections from WT and OCLN-deficient mice, respectively, for commercial antibodies directed against pan-CES1, PTGDS or PLA2g12a ( Figures 7A-C ). The antibody for proton-pump B1-V-ATPase was also used to label clear cells. The immunoreactivity for PTGDS was found in the endoplasmic network of principal cells of WT epididymis. It was found in some luminal contents in the distal epididymis. In addition, it was also enriched in the apical domain of V-ATPase-positive clear cells. In the OCLN-deficient epididymis, PTGDS in the principal cells was obviously decreased throughout the epididymal tubules ( Figure 7A ). Carboxylic esterase CES1 immunofluorescent labelling was enriched in the basal cells ( Figure 7B , yellow arrows), where were marked with anti-keratin-14 antibody ( Figure 7B , white arrows), as we published previously (29). From WT corpus to cauda regions, CES1 was highly abundant in the principal cells of cauda epididymidis ( Figure 7B , arrowheads). In the OCLN-deficient epididymis, CES1 immunofluorescent intensity was increased moderately in the principal cells of caput and corpus epididymidis, but decreased in the basal cells marked with keratin-14 ( Figure 7B , white and yellow arrows). The co-localization of CES1 and WT basal cells ( Figure 7B , yellow arrows), especially in the WT corpus region, was obviously diminished in the Ocln-KO epididymis ( Figure 7B , white arrows). As for PLA2g12a, in the WT epididymis, although its immunofluorescent intensity was obvious in the luminal contents, we consistently observed moderate intensity in the cytoplasm of epithelial cells, especial corpus region ( Figure 7C ). It was enriched levels in the apical domain of clear cells. In the Ocln-KO, PLA2g12a was obviously decreased in the cytoplasm of epithelial cells, but remained in the luminal contents ( Figure 7C ). Heatmap analysis on the transcriptomes and proteomes results showed that several related members were also significantly downregulated or altered in each of these pathways ( Figures 7D-F , red arrows). The redox-sensitive metabolic-associated enzyme catalase was found to be enriched in clear cells of WT epididymis but significantly downregulated in the OCLN-deficient epididymis ( Figure 7G ). Based on all these results, we proposed a model for the OCLN-promoted metabolic pathways, especially lipid catabolism in the epididymal epithelial cells under physiological conditions and in pathological conditions after Ocln ablation, as shown in Figure 8 . This involves impaired luminal acidification due to loss of clear cells ( Figures 8A, B ), as well as downregulated catabolic enzymes in DAG-related lipid metabolism and the subsequent arachidonate and eicosanoid pathways in Ocln-KO epididymis ( Figure 8C ).