Co-expression of sperm membrane proteins CMTM2A and CMTM2B is essential for ADAM3 localization and male fertility in mice

Chemokines are secreted proteins that are well known for their functions in immune cell movement and lymphoid tissue development (Griffith et al., 2014). However, the functions of the transmembrane-type chemokine-like factor (CKLF) and the family of proteins CMTM1-8 (previously named CKLFSF1-8) have not been widely investigated (Han et al., 2001, 2003). Unlike chemokines, CKLF and CMTM1-8 constitute gene clusters in humans and mice (Han et al., 2003; Zlotnik et al., 2006). These genes localize to three regions in human chromosomes 3, 14 and 16 and in mouse chromosomes 8, 9 and 14. Among these gene clusters, two testis-specific genes, CMTM1 and CMTM2, and three ubiquitously expressed genes, CKLF, CMTM3 and CMTM4, map to chromosome 16 in humans (Han et al., 2003; Wang et al., 2004; Shi et al., 2005). The corresponding mouse orthologs, Cklf, Cmtm1, Cmtm2a, Cmtm2b, Cmtm3 and Cmtm4 localize to chromosome 8. Although mouse Cmtm1 is the single homolog of human CMTM1, mouse Cmtm2a and Cmtm2b are both homologs of human CMTM2 which is highly expressed in testicular germ cells (Liu et al., 2007). Moreover, Cmtm2a and Cmtm2b are also testis-enriched genes and it is suggested that they share a similar function (Li et al., 2006). However, their physiological roles remain to be determined in vivo.

Recently, we established a quick and efficient system to analyze male fertility in vivo using CRISPR/Cas9-mediated mutant mice (Mashiko et al., 2013; Miyata et al., 2016). With this system, we found that 54 evolutionarily conserved and testis-enriched genes are not essential individually for male fertility in mice (Miyata et al., 2016). Moreover, we suggested that researchers should determine whether a gene of interest is required for male fertility in vivo before spending significant effort to analyze the molecular function of the gene in vitro.

In this study, we focused on analyzing three testis-specific genes, Cmtm1, Cmtm2a and Cmtm2b, in the CMTM gene cluster on mouse chromosome 8. To examine the physiological roles of these genes, we generated CRISPR/Cas9-mediated mutant mouse lines. Although Cmtm1 mutant mice were fertile, Cmtm2a and Cmtm2b double mutant male mice were sterile. In addition, the disruption of CMTM2A and CMTM2B affected the localization of sperm membrane protein ADAM3 that is essential for male fertility (Fujihara et al., 2018; Okabe, 2018). We conclude that co-expression of sperm membrane proteins CMTM2A and CMTM2B regulates the sperm fertilizing ability and that the two proteins are new members of the ADAM3-associated protein family.

In mice, the CMTM gene cluster localizes to chromosome 8 (Han et al., 2003). The gene cluster consists of six genes Cmtm1, Cmtm2a, Cmtm2b, Cmtm3, Cmtm4 and Cklf. The expression of these six genes in various mouse organs was examined by RT-PCR analysis. Although Cmtm3, Cmtm4 and Cklf were expressed ubiquitously, Cmtm1, Cmtm2a and Cmtm2b were exclusively expressed in mouse testes (Fig. 1A), as similarly reported in humans (Han et al., 2003). Next, the onset of Cmtm1, Cmtm2a and Cmtm2b expression in testes was examined by RT-PCR. All three genes were expressed 2 weeks after birth (Fig. 1B). These data show that mouse Cmtm1, Cmtm2a and Cmtm2b are testis-specific genes within the CMTM gene cluster and are expressed in the testes at the same stage. Mouse Cmtm1 is a homolog of human CMTM1 while mouse Cmtm2a and Cmtm2b are two homologs of human CMTM2 which localizes to human chromosome 16 (Fig. 1C) (Li et al., 2006). Protein sequence similarity between CMTM1 and CMTM2 shows that four putative transmembrane regions are conserved among humans and mice (Fig. S1A-C).

To analyze the physiological role of CMTM1, we generated Cmtm1 mutant mice by CRISPR/Cas9. A 19 bp deletion in the first exon was confirmed by PCR and direct sequencing analysis (Fig. S2A). RT-PCR analysis of the Cmtm1^−19/−19 testis showed a single band around 1.2 kb (Fig. S2B) and a 19 bp deletion from wild-type Cmtm1 mRNA was also confirmed (Fig. S2C). The 19 bp deletion causes a frameshift mutation, leading to a premature termination codon after amino acid 132 (wild-type CMTM1 comprises 392 aa) (Fig. S2D). Cmtm1^−19/−19 mice showed no deleterious effects on testicular histology (Fig. S2E). Spermatozoa produced by Cmtm1^−19/−19 mice were motile and morphologically normal under phase-contrast microscopy (Fig. 1D). To examine male fertility, adult Cmtm1^−19/−19 males were mated with wild-type females for several months. Cmtm1^−19/−19 males were fertile, showing normal mating behavior with successful ejaculation and vaginal plug formation. The mean litter size was 9.8±1.0 in Cmtm1^wt/wt males and 9.3±1.1 in Cmtm1^−19/−19 males (Fig. 1E). Cmtm1 females were fertile as expected, as Cmtm1 is a testis-specific gene (Fig. 1A). These data suggest that Cmtm1 is not essential for male fertility in mice.

We prepared polyclonal antibodies to check the localization of CMTM2A and CMTM2B. Immunostaining of cauda epididymal spermatozoa was performed after the confirmation of antibody specificity (Fig. S3A). CMTM2A and CMTM2B were localized to the membrane of the sperm head when we used antibodies against CMTM2A, CMTM2B, the sperm plasma membrane protein ADAM1B and the sperm acrosomal membrane protein IZUMO1 (Fig. 2A,B and Fig. S3B). CMTM2B was detected in the sperm head using immunoblot analysis (Fig. S3C). These results confirmed that CMTM2A and CMTM2B are sperm plasma membrane proteins. As shown in Fig. 1C and Fig. S1A,C, there is high similarity of genomic localization and amino acid sequence among mouse CMTM2A and CMTM2B, and human CMTM2. There is a possibility that mouse Cmtm2a and Cmtm2b have a compensative role to each other. To examine the function of CMTM2A and CMTM2B, we produced two lines of Cmtm2a and Cmtm2b double mutant mice using CRISPR/Cas9. In one mutant line (line 1), a 2 bp deletion of Cmtm2a and a 444 bp insertion (via a 313 bp deletion and a 757 bp insertion event) of Cmtm2b was confirmed by genotyping PCR, direct sequencing (Fig. 2C, Fig. S3D and Table S1) and RT-PCR analysis (Fig. 2D). In the other mutant line (line 2), a 15 bp deletion of Cmtm2a and an 8 bp deletion of Cmtm2b, was also confirmed by genotyping PCR, direct sequencing (Figs S3E,F and S4A) and RT-PCR analysis (Fig. 2D). In line 1, a frameshift mutation occurs in both Cmtm2a and Cmtm2b (Cmtm2a^−2/−2; Cmtm2b^+444/+444). In line 2, two mutations occur in the two genes, an in-frame mutation of Cmtm2a caused by a five amino acid deletion (residues 13-17) and a frameshift mutation of Cmtm2b (Cmtm2a^−15/−15; Cmtm2b^−8/−8). Immunoblot analysis confirmed that CMTM2A and CMTM2B are disrupted in line 1 testicular germ cells (TGCs) while only CMTM2B is disrupted in line 2 TGCs (Fig. 2E). Furthermore, it was found that the smaller CMTM2A slightly remains in line 2 TGCs. Line 1 and line 2 mice showed no deleterious effects on testicular histology (Fig. S4B). Spermatozoa produced by males in both lines were morphologically normal under phase-contrast microscopy (Fig. 2F). To examine male fertility, we checked the pregnancy rate and the mean litter size of males in the two double mutant lines. The pregnancy rate was 94.4% (29/31 plugs) of female mice mated with wild-type males, 0% (0/25 plugs) of females mated with line 1 males, and 48.4% (15/31 plugs) with line 2 males (Fig. 2G). The mean (±s.d.) litter size was 8.7±3.1 for wild-type males, 0 for line 1 males and 4.0±2.7 for line 2 males (Fig. 2H). Although line 2 males were subfertile, line 1 males were completely sterile despite showing normal mating behavior with successful ejaculation and vaginal plug formation.

To analyze the infertility of Cmtm2a, Cmtm2b double mutant males, we performed in vitro fertilization (IVF) assays using spermatozoa from line 1 and line 2 males mixed with cumulus-intact oocytes. Few spermatozoa collected from the cauda epididymis in line 1 mice could swim-up in sperm incubation medium, in direct contrast with line 2 spermatozoa that could. The major cause of sterility in line 1 mice was due to immotile epididymal spermatozoa. Line 2 spermatozoa could fertilize with eggs as effectively as wild-type spermatozoa [92.4±9.3% (330/357 eggs) and 98.4±2.8% (121/123 eggs) by wild-type and line 2, respectively] (Fig. 3A). To examine the subfertility of line 2 mice, we checked the sperm zona pellucida (ZP)-binding ability of the spermatozoa mixed with cumulus-free oocytes in vitro. Line 2 spermatozoa had a reduced ability to bind to the ZP compared with the wild-type spermatozoa (45.8±11.2 and 3.9±2.2 spermatozoa per egg by wild-type and line 2 spermatozoa, respectively; P<0.001) (Fig. 3B,C). These results indicated that spermatozoa produced by line 1 were immotile, and the spermatozoa of line 2 had an impaired ZP-binding ability.

The impaired ZP-binding phenotype in line 2 mice is shared among mice lacking sperm membrane protein ADAM3 (Fujihara et al., 2018). Thus, we performed an immunoblot analysis of ADAM3-associated proteins in the Cmtm2a, Cmtm2b double mutant mice. While ADAM3 was not detectable in line 1 spermatozoa, there were no differences between wild-type, line 1 and line 2 TGCs (Fig. 3D,E). In contrast, ADAM3 was readily detected in line 2 spermatozoa (Fig. 3E). Moreover, in line 2, CMTM2A remained in TGCs (Fig. 2E) but was not detected in spermatozoa (Fig. 3E). Although both CMTM2A and CMTM2B were absent in line 1 and 2 spermatozoa, ADAM3 was only missing in line 1 spermatozoa.

As shown in Fig. 3, there is a possibility that CMTM2A, CMTM2B and ADAM3 interact in mouse testis and spermatozoa. We checked CMTM2A and CMTM2B (CMTM2A/2B) levels in Adam3 knockout (KO) mice by immunoblot analysis after generating Adam3 KO mice (Fig. S4C). CMTM2B disappeared from spermatozoa in Adam3 KO mice, while CMTM2A remained (Fig. 4A). We reported previously that KO spermatozoa from a TGC-specific GPI-anchored protein encoded by Ly6k possess the same phenotype in which the spermatozoa lose their fertilizing ability, but ADAM3 remains (Fujihara et al., 2014), which is similar to the results in line 2 mice. However, there were no differences in CMTM2A or CMTM2B levels in the Ly6k KO mice (Fig. 4B). These data imply a relationship between CMTM2A/2B and ADAM3 but no relationship with LY6K.

Next, to analyze the interaction of CMTM2A/2B with ADAM3 by immunoprecipitation (IP), we generated Adam3-Flag transgenic (Tg) mice because no anti-ADAM3 antibodies are available for IP analysis. The transgene utilized a truncated form of ADAM3 (736 aa), which removed the intracellular region of ADAM3 (86 aa from the C-terminus), corresponding to a region of low conservation among other species (Table S2). To confirm that the Adam3-Flag transgene was functional, we performed a rescue experiment by generating Adam3-Flag Tg mice on an Adam3 KO background (Fig. 4C). Our results showed that the infertile phenotype of Adam3 KO males was rescued by the transgene (average litter size of 7.0±0.9; mean±s.d.) (Fig. 4D). Expression of the transgene in these mice was further examined by immunoblot analysis. Although ADAM3 and FLAG expression from the transgene was detected in the TGCs of the Adam3 KO-Tg mice, both could not be detected in the spermatozoa (Fig. 4E). Also, CMTM2B could not be detected in the spermatozoa produced by the Adam3 KO-Tg mice (Fig. 4E). To check the interaction of ADAM3 with CMTM2A/2B in TGCs and spermatozoa, we performed IP analysis using FLAG antibody. ADAM3 in TGCs could be pulled down efficiently as shown by the immunoblot analysis using ADAM3 and FLAG antibodies (Fig. S4D), although IP of sperm lysates did not work as well as expected (Fig. S4E). However, CMTM2A/2B, LY6K and PDILT could not be detected in the FLAG immunoprecipitates (Fig. S4D,E). Unfortunately, these IP-immunoblot analyses using Adam3-Flag Tg rescue mice could not confirm the interaction between ADAM3-associated proteins (LY6K and PDILT), as reported previously (Tokuhiro et al., 2012; Fujihara et al., 2014). Under these rescue conditions, CMTM2A/2B could not interact with ADAM3 in the TGCs and spermatozoa. Therefore, we could not clarify whether CMTM2A and CMTM2B directly affect the correct localization of ADAM3 in spermatozoa.

Chemokines are a well-known family of cytokines used to recruit from immune cells (Nagarsheth et al., 2017). The transmembrane-type chemokine like factor (CKLF) and the protein family CMTM (CKLF-like MARVEL transmembrane domain containing) 1-8 constitute gene clusters in humans and mice. In this cluster, there are testis-specific expressed genes; CMTM1 and CMTM2 in humans and Cmtm1, Cmtm2a, and Cmtm2b in mice (Wang et al., 2004; Shi et al., 2005; Li et al., 2006). However, their physiological functions remained to be determined in vivo.

In this paper, we generated mutant mice of three testis-specific CMTM genes (Cmtm1, Cmtm2a and Cmtm2b) using CRISPR/Cas9. When taking human and mouse genomic conservation data into account, we disrupted these genes using two approaches: a single mutation of Cmtm1 and double mutations of Cmtm2a and Cmtm2b. We found that although Cmtm1 mutant mice were fertile, Cmtm2a and Cmtm2b double mutant mice were male infertile due to impaired sperm function. Immunostaining of spermatozoa indicated that CMTM2A and CMTM2B localized to the plasma membrane on the sperm head. Two mouse lines of Cmtm2a and Cmtm2b double mutations, Cmtm2a^−2/−2; Cmtm2b^+444/+444 (line 1) and Cmtm2a^−15/−15; Cmtm2b^−8/−8 (line 2) were produced and it was found that line 1 was completely infertile and line 2 was subfertile. The difference between the two phenotypes is attributed to the genetic mutations induced by CRISPR/Cas9. Line 1 was a complete knockout of Cmtm2a and Cmtm2b while line 2 had an in-frame mutation (15 bp deletion) of Cmtm2a and a frameshift mutation (8 bp deletion) of Cmtm2b. The results indicate that Cmtm2a and Cmtm2b are required for male fertility in mice.

Although the two mouse lines had different mutations, they produced normal-looking spermatozoa, but these epididymal spermatozoa showed different phenotypes: immotility and impaired ZP-binding. The impaired ZP-binding phenotype of line 2 mice is a shared phenotype among mice lacking ADAM3 (Fujihara et al., 2018). The sperm membrane protein ADAM3 is thought to play a pivotal role in sperm-ZP binding and sperm migration through the uterotubal junction (UTJ) (Shamsadin et al., 1999; Yamaguchi et al., 2009). Currently, more than 10 proteins (ACE, ADAM1A, ADAM2, CALR3, CLGN, PDILT, PMIS2, PRSS37, RNASE10, TEX101 and TPST2) have been described to affect the ADAM3 protein and/or its localization in the spermatozoa (Fujihara et al., 2018). Although ADAM3 remains in line 2 spermatozoa, it is not present in line 1 spermatozoa. This finding suggests that testicular CMTM2A may affect the correct localization of ADAM3 onto spermatozoa. Impaired ZP-binding in line 2 may be caused by abnormalities in sperm protein(s) other than ADAM3. Further experiments may be required to examine the cause of immotile spermatozoa and the disappearance of ADAM3 in line 1 mice. Therefore, our Cmtm2a, Cmtm2b double mutant mice may prove useful in elucidating the physiological function of human CMTM2. CMTM2B was absent in Adam3 KO spermatozoa, while CMTM2A remained. Moreover, we recently identified TGC-specific GPI-anchored protein LY6K which is essential for sperm migration through the UTJ and ZP-binding (Fujihara et al., 2014). We showed here that Ly6k KO spermatozoa retained not only ADAM3 but also CMTM2A/2B. These results suggest that LY6K interacts with other unknown factor(s), but not CMTM2A/2B, to regulate sperm fertilizing ability.

To examine the interaction of ADAM3 and CMTM2A/2B, we generated Tg rescue mice using an Adam3-Flag transgene. The transgene used was a truncated form of ADAM3 (736 aa) which removed the protein's intracellular region (86 aa from the C-terminus). The transgene of truncated ADAM3 was functional and could rescue the male infertility of Adam3 KO mice. However, in the Tg mouse spermatozoa, ADAM3-FLAG expression could not be detected by immunoblot analysis. Previously, we reported a similar result in the full-length Adam3 Tg rescue mice (Yamaguchi et al., 2009). The transgenic expression of ADAM3 showed very weak signals in the testis and spermatozoa in Adam3 Tg#47 mice. These results indicated that the amount of ADAM3 protein required for sperm fertilization is much less than the levels of ADAM3 found in wild-type spermatozoa, and that the intracellular region (residues 737-822) was not required for the sperm fertilizing ability. Since ADAM3 and FLAG were expressed in the TGCs of Adam3-Flag Tg mice, IP-immunoblot analysis was performed using FLAG antibody. Even though ADAM3 and FLAG were immunoprecipitated efficiently, they did not interact with either CMTM2A/2B or LY6K and PDILT as control proteins. Although we could not clarify the interaction of CMTM2A/2B and ADAM3 in our experimental conditions, KO mouse experiments indicated that CMTM2A/2B are new members of the identified group of proteins involved in ADAM3 regulation. Unfortunately, our antibodies against CMTM2A/2B and ADAM3 did not work for IP analysis. Further investigations will be required to clarify the interaction of CMTM2A/2B and ADAM3. Our findings support a potential role of human CMTM2 in sperm function and could be used to develop infertility treatments as well as contraceptives.

All animal experiments were approved by the Animal Care and Use Committee of the Research Institute for Microbial Diseases, Osaka University (H25-02-0 and H30-01-0; 21 June 2013 and 4 July 2018, respectively). Mice were maintained under a 12 h:12 h light:dark cycle (lights on from 08:00 to 20:00). Wild-type mice were purchased from CLEA Japan (Tokyo, Japan) and Japan SLC (Shizuoka, Japan).

In this study, we generated genetically modified mouse lines, Cmtm1 (B6D2-Cmtm1<em1Osb>); RBRC10119, Cmtm2a^−2/−2; Cmtm2b^+444/+444 (Line 1) (B6D2-Cmtm2a<em2Osb>Cmtm2b<em2Osb>); RBRC09947, Cmtm2a^−15/−15; 2b^−8/−8 (line 2) (B6D2-Cmtm2a<em1Osb>Cmtm2b< em1Osb>); RBRC09946, Adam3 knockout mice [STOCK-Adam3<tm1(KOMP)Osb>/11B]; RBRC05866, and Adam3-Flag transgenic mice [STOCK-Tg(Clgn-Adam3Flag)1Osb]; RBRC09960. These were deposited to the RIKEN BioResource Research Center (http://mus.brc.riken.jp/en/) and the Center for Animal Resources and Development (CARD), Kumamoto University (http://card.medic.kumamoto-u.ac.jp/card/english/). The Ly6k knockout mouse line (STOCK-Ly6k<tm1Osb>) was described previously (Fujihara et al., 2014).

Mouse cDNA was prepared from multiple adult tissues and from 1- to 5-week-old testes of ICR mice (Fujihara et al., 2010). To confirm the mutations in Cmtm1^−19/−19, Cmtm2a^−2/−2; Cmtm2b^+444/+444 and Cmtm2a^−15/−15; Cmtm2b^−8/−8 mice, RT-PCR was performed using RNA from each mouse testis. The amplification conditions were 1 min at 94°C, followed by 30 cycles of 94°C for 30 s, 65°C for 30 s, and 72°C for 30 s, with a final 7 min extension at 72°C. The primers used are listed in Table S3.

Rabbit anti-mouse CMTM2A and CMTM2B polyclonal antisera were produced by immunization with mouse CMTM2A (residues 6-24: KFPFRPRGGQPREDTTPKR; residues 157-169: QFQHFRGLRLRKW) and CMTM2B (residues 192-210: ERQRELAEKAKRESMDPGW) polypeptides, respectively. The monoclonal antibodies used here were as described previously: TES101 for TEX101, 1D5 for ACE, KS64-10 for SLC2A3, KS64-125 for IZUMO1 and KS107-158 for ADAM1B (Ikawa et al., 2011; Fujihara et al., 2013b). Other antibodies were purchased from Sigma-Aldrich (M2 for FLAG), Santa Cruz Biotechnology (sc-9757 for BASIGIN), and Merck Millipore (7C1 for ADAM3 and 9D for ADAM2). Rabbit antisera against CLGN, CALR3 and PDILT were as described previously (Tokuhiro et al., 2012). Dilutions used were 1:200 to 1:300 for immunostaining and 1:500 to 1:1000 for immunoblot analysis.

Immunostaining was performed as described previously (Fujihara et al., 2012; Muto et al., 2016).

Immunoblot analysis was performed as described previously (Fujihara et al., 2017). Briefly, testicular germ cells were collected from the seminiferous tubules of testes. Sperm samples were collected from the cauda epididymis and vas deferens. Sperm head-tail separation using centrifugation was performed as described previously (Miyata et al., 2015). These samples were homogenized in lysis buffer containing 1% Triton X-100 and 1% protease inhibitor (Nacalai Tesque, Kyoto, Japan) and then were centrifuged (10,000 g for 20 min at 4°C), and the supernatants were collected. Protein lysates were separated by SDS-PAGE under reducing condition and transferred to PVDF membranes (Merck Millipore). After blocking, blots were incubated with primary antibodies overnight at 4°C, and then incubated with secondary antibodies conjugated with horseradish-peroxidase. The detection was performed using an ECL plus western blotting detection kit (GE Healthcare) and Chemi-Lumi One Ultra (Nacalai Tesque).

Cmtm1, Cmtm2a and Cmtm2b mutant mice were produced by microinjection of pX330 plasmid (https://www.addgene.org/42230/) into mouse embryos as described previously (Mashiko et al., 2013; Fujihara and Ikawa, 2014). A search for sgRNA and off-target sequences was performed using CRISPRdirect software (https://crispr.dbcls.jp/) (Naito et al., 2015). After the validation of EGFP expression assay (Fujihara and Ikawa, 2014), the sgRNA sequence used for microinjection were: 5′-AGGTCGTAAAGTTAAGCGTC-3′ for the first exon of Cmtm1, 5′-TTTCCATTTCGGCCCAGAGG-3′ for the first exon of Cmtm2a, and 5′-GATGAAAGAAGAGGGTTCAA-3′ for the first exon of Cmtm2b. The sgRNAs for Cmtm2a and Cmtm2b were co-injected into the pronuclei of fertilized eggs. The 2-cell-stage embryos were transferred into the oviducts of pseudopregnant ICR females the next day. Cmtm1 mutant mice had a 19 bp deletion (5′-GCGTGAGTCCAGACGCTTA-3′) in the first exon. Two lines of Cmtm2a and Cmtm2b double mutant mice were generated; line 1: a 2 bp deletion (5′-AG-3′) in the first exon of Cmtm2a and a total of a 444 bp insertion mutation (313 bp deletion and 757 bp insertion) around the first and second exons of Cmtm2b; and line 2: a 15 bp deletion (5′-GAGGTCAGCCCAGAG-3′) in the first exon of Cmtm2a and an 8 bp deletion (5′-TTCAAAGG-3′) in the first exon of Cmtm2b. The primers used are listed in Table S3. Detailed genotype information of mutant mouse lines is shown in Figs S2-S4.

Sexually mature mutant male mice were caged with 2-month-old B6D2F1 or mutant females for several months, and the number of pups in each cage was counted within a week of birth. Pregnancy rates are presented as the success rate for getting pregnant from natural matings. Average litter sizes are presented as the number of total pups born divided by the number of litters for each genotype.

After breeding studies, males were euthanized by cervical dislocation following anesthesia. Testes were fixed in 4% paraformaldehyde in PBS and were processed for paraffin embedding. Paraffin sections were cut at 5 μm and stained with periodic acid-Schiff (PAS) and then counterstained with Mayer hematoxylin solution (Wako). The cauda epididymal spermatozoa were dispersed in PBS and then the sperm morphology was observed under a phase-contrast microscope (BX50, Olympus).

Sperm ZP-binding assay was performed as described previously (Yamaguchi et al., 2006). Briefly, 30 min after mixing with 2-h-incubated spermatozoa, eggs were fixed with 0.25% glutaraldehyde. The bound spermatozoa were observed with an Olympus IX70 fluorescence microscope after Hoechst 33258 staining.

In vitro fertilization using mouse spermatozoa was performed as described previously (Tokuhiro et al., 2012).

The mouse Adam3 gene targeting vector (DPGS00107_A_C12) was from the International Knockout Mouse Consortium (IKMC) (Skarnes et al., 2011). After linearization with AsiSI digestion, the targeting vector was electroporated into EGR-G01 (129S2×[CAG/Acr-Egfp]C57BL/6NCr) embryonic stem (ES) cells (Fujihara et al., 2013a), and colonies were screened. To disrupt the Adam3 gene, exons 12-16 were replaced with an IRES:lacZ trapping cassette and a floxed promoter-driven neo cassette, and a diphtheria-toxin-A-chain (DTA) expression cassette was used for negative selection. After G418 selection, 1 of 96 drug-resistant clones had a homologous recombination event after PCR analysis. The mutant ES cell clones were injected into 8-cell-stage ICR embryos, and the chimeric blastocysts were transferred into the uterine horns of pseudopregnant ICR females the next day. The obtained chimeric males were mated with B6D2F1 females for germ-line transmission. Offsprings from heterozygous intercrosses were genotyped by PCR. Both a 522 bp band as the wild-type allele and a 358 bp band as the knockout (KO) allele were amplified by PCR. The primers used are listed in Table S3. The phenotype of Adam3 KO mice was reported previously (Shamsadin et al., 1999; Yamaguchi et al., 2009).

A transgenic (Tg) rescue mouse line was produced by injecting the transgene into in vitro fertilized eggs from Adam3 KO mice. The cDNA encoding the mouse Adam3 gene was amplified by PCR using wild-type testis cDNA as a template. The Adam3 cDNA was further modified by the removal of 86 AAs (residues 737-822) from the C-terminus of ADAM3. The XbaI and EcoRV sites included in the primers were used to introduce the modified Adam3 cDNA into the pClgn1.1 vector, a pBluescript II SK (+) vector containing the mouse Clgn promoter and rabbit globin polyA signal (Ikawa et al., 2001). Moreover, the FLAG sequence (DYKDDDK) was inserted into the C-terminus of the modified Adam3 cDNA. The KpnI- and SacI-digested transgene fragment (3.2 kb) was purified by gel electrophoresis. The digested transgene fragment was microinjected into the pronuclei of fertilized eggs, and an Adam3-Flag Tg rescue mouse line [STOCK-Tg(Clgn-Adam3Flag)1Osb] was established. A 614 bp band as the Tg allele was amplified by PCR. The primers used are listed in Table S3.

Immunoprecipitation was performed as described previously (Tokuhiro et al., 2012). Briefly, the protein lysates (supernatants) were incubated with antibody-conjugated Dynabeads Protein G (Life Technologies) for 60 min at 4°C. The protein complexes were then washed three times with TBS buffer containing 40 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.1% Triton X-100 and 10% glycerol before SDS-PAGE and immunoblot analysis.

Statistical analyses were performed using Student's t-test inserted into Microsoft Excel after the data were tested for normality of distribution. Differences were considered significant at P<0.001.

We thank Tomoko Aramaki, Naoko Nagasawa and Biotechnology Research and Development (nonprofit organization) for technical assistance, and Julio M. Castaneda and Ferheen Abbasi for critical reading of the manuscript.