ABSTRACT: During spermatogenesis, the movement of germ cells across the seminiferous epithelium is associated with extensive junction restructuring. Yet the underlying mechanism (or mechanisms) that regulates these events is largely unknown. If the molecular architecture of the cell-cell actin-based adherens junction (AJ), such as ectoplasmic specialization (ES) and tubulobulbar complex— two testis-specific AJ types, is known, many functional mechanistic studies can be designed. We thus undertook an investigation to study 3 adaptors in the seminiferous epithelium: zyxin, axin, and Wiskott-Aldrich syndrome protein (WASP). All 3 adaptors were shown to be products of Sertoli and germ cells. Zyxin was shown to be a stage-specific protein that was most prominent during stages V—VII and restricted mostly to pachytene spermatocytes, but it could also be detected at the site of basal and apical ectoplasmic specialization (ES). Zyxin, axin, and WASP were shown to be structurally linked to the N-cadherin/β-catenin/α-actinin/actin complex but not to the nectin-3/afadin or the β1- integrin—mediated protein complexes. Interestingly, zyxin, axin, and WASP are also structurally linked to vimentin (an intermediate filament protein) and α-tubulin (the subunit of a microtubule), which suggests that they have a role (or roles) in the regulation of the dynamics of the desmosome-like junction and microtubule. These results illustrate that zyxin, axin, and WASP are adaptors in both AJs and intermediate filament-based desmosome-like junctions. This raises the possibility that classic cadherins are also associated with vimentin-based intermediate filaments via these adaptors in the testis. While virtually no N-cadherin was found to associate with vimentin in the seminiferous tubules, it did associate with vimentin when testis lysates were used. Interestingly, about 5% of the E-cadherin associated with vimentin in isolated seminiferous tubules, and about 50% of the E-cadherin in the testis used vimentin as its attachment site. These data suggest that cadherins in the testis, unlike those in other epithelia, use different attachment sites to anchor the cadherin/catenin complex to the cytoskeleton. The levels of zyxin, axin, and WASP were also assessed during AF-2364-mediated AJ disruption of the testis, which illustrated a time-dependent protein reduction that was similar to the trends observed in nectin-3 and afadin but was the opposite of those observed for N-cadherin and β-catenin, which were induced. Collectively, these results illustrate that while these adaptors are structurally associated with the cadherin/catenin complex in the testis, they are regulated differently.
Zyxin is an adaptor protein found at the site of adherens junctions (AJs) and focal contacts (for a review, see Beckerle, 1997) that recruits AJ- or focal adhesion complex (FAC)-associated structural and signaling molecules to the same site to regulate the cell adhesion function (for a review, see Beckerle, 1997). Zyxin can also be translocated to the nucleus, regulating gene transcription activity (for a review, see Wang and Gilmore, 2003) in a way similar to that of p120ctn and β-catenin. Both are putative AJ-associated peripheral signaling molecules that can interact with the Kaiso and LEF/TCF transcription factor family, activating gene expression in the nucleus (for reviews, see Anastasiadis and Reynolds, 2000; Seidensticker and Behrens, 2000). Furthermore, zyxin is a phosphoprotein per se and can be activated via phosphorylation and/or dephosphorylation. In essence, zyxin is a crucial molecule in the AJ to regulate the cell adhesion function and in the nucleus to signal functions.
Axin is another adaptor and signaling protein that takes part in the Wnt signaling pathway. It recruits β-catenin to glycogen synthase kinase-3β (GSK-3β), which in turn regulates the phosphorylation and degradation of cytosolic β-catenin (for a review, see Seidensticker and Behrens, 2000). Interestingly, the phosphorylated form of GSK-3β was recently shown to be a crucial signaling molecule in the tumor necrosis factor alpha (TNFα)-mediated tight junction regulation in Sertoli cells (Siu et al, 2003a). An increasing number of proteins are shown to be putative axin-binding proteins, such as plakoglobin (Kodama et al, 1999), mitogen-activated protein kinase kinase 1 (MEKK1), and casein kinase I, all of which are implicated in regulating the JNK signaling pathway (Zhang et al, 1999, 2002). Like zyxin, axin is a putative substrate of protein kinases.
Wiskott-Aldrich syndrome protein (WASP) is another adaptor protein that regulates actin dynamics via activation and recruitment of the Arp (actin-related protein) 2/3 complex and profilin-bound monomeric actin (for a review, see Caron, 2002). In its inactive state, WASP is folded, which masks some of its protein-binding sites (for reviews, see Mullins, 2000; Caron, 2002). Upon its activation by cell division cycle 42 (Cdc42) guanosine triphosphatase (GTPase), phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2), and WASP-interacting SH3 protein, WASP unfolds, exposing its binding sites to different interacting proteins (for a review, see Takenawa and Miki, 2001), which include c-Src, actin, WASP-interacting protein, and Arp2/3 (for reviews, see Mullins, 2000; Takenawa and Miki, 2001). As such, the principal role of WASP in AJ functions apparently lies in its ability to recruit different proteins to the multiprotein complexes at the site of AJs, which in turn affects the underlying cytoskeletal network.
Earlier studies have shown that the adhesion of spermatogonia and spermatocytes onto Sertoli cells at or near the basal compartment of the seminiferous epithelium is mediated largely by desmosome-like intermediate filament-based anchoring junctions (Russell, 1977) composed of vimentin, plectin, and other adhesion proteins (for reviews, see Russell and Peterson, 1985; Cheng and Mruk, 2002), whereas the actin-based AJs, such as ectoplasmic specializations (ESs), are restricted to the apical compartment between Sertoli cells and round and elongating/elongate spermatids and between Sertoli cells in the basal compartment (Russell and Clermont, 1976; Russell, 1979). Recent studies have shown that the apical ES is constituted by the α6β1- or α4β1-integrin/laminin γ3, the nectin/afadin, and the cadherin/catenin, whereas the basal ES is constituted by the α6β4-integrin/laminin and the cadherin/catenin complexes (Chapin et al, 2001; Siu and Cheng, 2004; for reviews, see Vogl et al, 2000; Cheng and Mruk, 2002; Lui et al, 2003b). The precise attachment site of the classic cadherin/catenin complex remains a controversial issue. Notably, studies using immunofluorescent microscopy (Johnson and Boekelheide, 2002a,b) and electron microscopy (Mulholland et al, 2001) suggest that the classic cadherin/catenin complex, an actin-based junctional protein complex in other epithelia (for a review, see Gumbiner, 2000), uses the intermediate filament as the attachment site in the testis. It is unknown whether desmocollins and desmogleins (both are desmosome-integral membrane proteins) and desmoplakins, plakophilins, and plakoglobins (the latter 3 are desmosome-peripheral proteins), which are the putative structural proteins of desmosomes in other epithelia (for a review, see Ishii and Green, 2001), are found in the testis (for a review, see Cheng and Mruk, 2002). A recent study using immunoprecipitation with and without a cross-linker has clearly demonstrated that the N-cadherin/β-catenin complex associates exclusively with actin, but not with vimentin, using lysates of seminiferous tubules isolated from rat testes (Lee et al, 2003). This is consistent with an earlier report that colocalized both N-cadherin and β-catenin to the same site at the ES by immunohistochemistry (Wine and Chapin, 1999). In this report, we have also investigated whether the 3 selected adaptor proteins, which are also known to associate with both the actin and intermediate filament cytoskeletal network, can act as linkers between the classic cadherin/catenin complex and the intermediate filament-based and microtubule-based cytoskeletal networks.
Materials and Methods
Animals and Antibodies
Sprague-Dawley rats were purchased from Charles River Laboratories (Kingston, NY). The use of animals was approved by the Rockefeller University Animal Care and Use Committee with protocol numbers 00111 and 03017. Anti-afadin antibody (Cat: 610732, Lot: 1) was obtained from Becton Dickinson Transduction Laboratories (San Diego, Calif). All other antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, Calif). These included antibodies against axin (H-98, Cat: sc-14029, Lot: C012), zyxin (C-19, Cat: sc-6437, Lot: G060), WASP (B-9, Cat: sc-13139, Lot: B222), E-cadherin (H-108, Cat: sc-7870, Lot: K080), N-cadherin (H-63, Cat: sc-7939, Lot: C081), β-catenin (H-102, Cat: sc-7199, Lot: L060), p120ctn (S-19, Cat: sc-1101, Lot: A079), α-actinin (N-19, Cat: sc-7453, Lot: D292), nectin 3 (C-19, Cat: sc-14806, Lot: K261), β1-integrin (M-106, Cat: sc-8978, Lot: E221), vasodilator-stimulated phosphoprotein (VASP) (C-17, Cat: sc-1853, Lot: G262), c-Src (B-12, Cat: sc-8056, Lot: C051), actin (H-196, Cat: sc-7210, Lot: C222), α-tubulin (TU-02, Cat: sc-8035, Lot: G182), and vimentin (V9, Cat: sc-6260, Lot: J2802). For N-cadherin, E-cadherin, α-catenin, β-catenin, axin, β1-integrin, and actin, the antigen used for polyclonal antibody production in rabbits was derived from the corresponding human recombinant protein. The rabbit p120ctn polyclonal antibody and the mouse c-Src monoclonal antibody were raised against peptide fragments derived from mouse p120ctn and human c-Src proteins. Anti-VASP, nectin-3, and α-actinin antibodies were affinity-purified goat polyclonal antibodies raised against peptide fragments derived from the corresponding human recombinant protein. Afadin and WASP were mouse monoclonal antibodies raised against human recombinant proteins. Vimentin and α-tubulin were mouse monoclonal antibodies raised against porcine recombinant proteins. All antibodies used in this study cross-reacted with the corresponding rat proteins as indicated by the manufacturer. The corresponding secondary antibodies were purchased from Santa Cruz Biotechnology.
Primary Testicular Cell Cultures
Sertoli cells were isolated from 20-day-old rat testes and cultured in vitro in Ham F12 medium (F12)/Dulbecco modified Eagle medium (DMEM) as earlier described (Lee et al, 2003). Sertoli cells were plated at a high cell density (0.5 × 106 cells/cm2) on Matrigel (1:8 with DMEM, vol/vol) (Collaborative Research Inc, Bedford, Mass)-coated 12-well dishes (effective area, ∼3.8 cm2 per well) with 3 mL of F12/DMEM per well. Sertoli cells were incubated at 35°C in a humidified atmosphere of 5% CO2/95% air (vol/vol). Cultures were designated D0 (time 0) at the time of cell plating. About 36 hours after plating, cells were hypotonically treated with 20 mM Tris, pH 7.4, for 2.5 minutes to lyse residual germ cells (Galdieri et al, 1981) and washed twice to remove cellular debris. The resulting Sertoli cells had a purity of approximately 95%. Sertoli cells were terminated 24 hours after hypotonic treatment. Germ cells were isolated from 20-, 60-, and 90-day-old rats and terminated within 3 hours (Lee et al, 2003). Cell purity was monitored by microscopic examination (Lee et al, 2003). Germ-, myoid-, Leydig-, and Sertoli cell-specific marker genes, such as c-Kit receptor, fibronectin, 3β-hydroxysteroid dehydrogenase (3β-HSD), and testin, respectively, were used to examine cell contamination by reverse transcriptase-polymerase chain reaction (RT-PCR) (Lee et al, 2003). Seminiferous tubules were isolated from adult rat testes (∼300 gm body weight [bw]) as previously described (Lee et al, 2003).
RT-PCR was performed as described (Lee et al, 2003). Briefly, 2 μg of total RNA was reverse transcribed into complementary DNAs using 1 μg oligo(dT)15 with a Moloney murine leukemia virus RT kit (Promega Corp, Madison, Wis) in a final reaction volume of 25 μL. Thereafter, PCR was performed by combining 2 μL of the RT product and 0.6 μg each of sense and antisense primers of a target gene (coamplified with either rat ribosomal S-16 or rat β-actin using ∼0.01 μg each of the sense and antisense primer) (Table) and the remaining reaction mixture as described (Lee et al, 2003). The cycling parameters for PCR were as follows: denaturation at 94°C for 1 minute, annealing at 46°C–61°C for 2 minutes, and extension at 72°C for 3 minutes, for a total of 20–29 cycles. The linearity of the PCR products was assessed in preliminary experiments as detailed elsewhere (Lee et al, 2003).
Treatment of Rats With AF-2364 to Induce Germ Cell Loss From the Seminiferous Epithelium
Polyacrylamide Gel Electrophoresis and Immunoprecipitation
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions was performed as described (Lee et al, 2003). In brief, cells and tubules were resuspended in immunoprecipitation buffer (10 mM Tris, 0.15 M NaCl, 2 mM PMSF, 2 mM EDTA, 2 mM N-ethylmaleimide, 1% NP-40 [vol/vol], 1 mM sodium orthovanadate [a protein tyrosine phosphatase inhibitor, PTPi], 0.1 μM sodium okadate [a protein Ser/Thr phosphatase inhibitor], and 10% glycerol [vol/vol], pH 7.4, at 22°C) with a cell:buffer and a tissue:buffer ratio of 2 × 106 cells:500 μL and 0.1 g tissue:500 μL, respectively. Samples were sonicated twice for 10 seconds with a 30-second interval in melting ice. Clear lysates were obtained by centrifugation at 12 000 × g for 15 minutes at 4°C. The total protein concentration was determined by the Coomassie blue dye binding assay (Bradford, 1976) using bovine serum albumin as a standard. For immunoprecipitation, 400 μg of protein from the seminiferous tubule lysates was first pretreated with the corresponding normal serum at 1:150 dilution for 4 hours at room temperature in a rotator. Thereafter, 10 μL of Protein A/G-PLUS agarose (Santa Cruz Biotechnology) was added to each sample and incubated at room temperature for 4 hours. Centrifugation was performed to separate the nonspecific complexes that were bound to serum proteins and to the Protein A/G-PLUS agarose from those proteins in the lysates. Then, the resulting lysates in supernatant were incubated with anti-zyxin, axin, or WASP antibodies (1: 150) overnight in a rotator to allow coimmunoprecipitation. Thereafter, 20 μL of Protein A/G-PLUS agarose was added to each sample and incubated for 4 hours to precipitate the immunocomplexes. After washing 4 times with immunoprecipitation buffer (1000 × g, 5 minutes), the resulting immunocomplexes were extracted from the agarose using a SDS-sample buffer [0.125 M Tru, pH 6.8 at 22°C containing 1% SDS (w/v), 1–6% α-mercaptoethanol (v/v), 1 mM EDTA, and 10% glycerol (v/v)], denatured at 100°C for 5–10 minutes, and resolved by SDS-PAGE as described (Lee et al, 2003). A chemiluminescent ECL kit (Amersham Pharmacia Biotech, Piscataway, NJ) was used to detect the immunoreactive target protein band with Kodak BioMax Light Films. Blots were reprobed with a second antibody after the initial antibody was removed by a stripping buffer (62.5 mM Tris-HCl, pH 6.7, at 22°C containing 100 mM 2-mercaptoethanol and 2% SDS [wt/vol]) at 55°C for 30 minutes as described (Lee et al, 2003).
Testes were isolated from normal (control) and AF-2364-treated rats and frozen in liquid nitrogen. Eight-micrometer sections were cut and mounted on poly-l- lysine—coated slides in a cryostat. All sections from one experiment were mounted onto a single slide to ensure that all sections were subjected to the same incubation and color development times in order to minimize interexperimental variations. Each experiment was repeated at least twice using the testes from 2 sets of rats, and representative results are shown. Sections were air dried and fixed in 4% paraformaldehyde (wt/vol in phosphate-buffered saline [PBS], 10 mM sodium phosphate and 0.15 M NaCl, pH 7.4, at 22°C) for 10 minutes at room temperature. Endogenous peroxidase activity was blocked by incubating sections in 1% hydrogen peroxide (in methanol [vol/vol]) for 20 minutes. Nonspecific antibody binding sites were blocked by preincubating the sections with 10% normal goat serum (vol/vol in PBS) (Zymed Laboratories Inc, San Francisco, Calif) for 1 hour. Thereafter, sections were incubated with a goat anti-zyxin antibody (1:150) overnight in a humidified chamber at 35°C. Sections were then incubated with biotinylated rabbit anti-goat immunoglobulin G (IgG; Zymed) for 30 minutes and then with a streptavidin-peroxidase complex (Zymed) for 15 minutes. 3,3′-Diaminobenzidine tetrahydrochloride was used as a substrate to visualize zyxin in the seminiferous epithelium. After counterstaining with hematoxylin and dehydration in 70% ethanol, 90% ethanol, 100% ethanol, and xylene, sections were mounted and examined under a microscope. Immunohistochemistry studies were limited to zyxin, because we could not locate antibodies to WASP or to axin commercially that did not yield multiple bands in addition to the antigen band. The anti-zyxin antibody yielded a prominent immunoreactive band that corresponded to zyxin in immunoblot analysis.
Immunofluorescent microscopy was performed as earlier described (Lee et al, 2003; Siu et al, 2003b). In brief, testes removed from adult rats were fixed in Bouin fixative, embedded in paraffin, and sectioned to 8-μm thickness. Following the removal of the paraffin by the use of xylene, sections were incubated with a mouse anti-N-cadherin antibody (Cat: 33–3900, Lot: 11268187, Zymed) (1:100 dilution) and then with a goat-antimouse IgG-Cy3 (Cat: 81–6515, Lot: 11067429, Zymed). For fluorescent immunocytochemistry, Sertoli cells cultured for approximately 2 days in F12/DMEM at 5 × 104 cells/cm2 were fixed in Bouin fixative, permeabilized with Triton X-100, and stained with DAPI (4′,6-diamidino-2-phenylindole) (Molecular Probes, Eugene, Ore) (specific staining for DNA in the nucleus) and a goat-anti-zyxin antibody, to be followed by a rabbit anti-goat IgG-fluorescein isothiocyanate (FITC) (Zymed) antibodyconjugate complex. Sections or cells were then mounted in Vectashield (Vector Laboratories, Burlingame, Calif), and micrographs were obtained using an Olympus BX40 microscope equipped with Olympus UPlanF1 fluorescent optics (Melville, NY). Controls included sections incubated with either 1) normal mouse serum or 2) a mouse anti-human α1-antitrypsin previously characterized in our laboratory (Silvestrini et al, 1990) at the same dilution as that of the primary antibody.
Seminiferous tubules were removed from the testes, fixed in 2.5% glutaraldehyde (vol/vol in 0.1 M sodium cacodylate buffer, pH 7.4) for 2–4 hours at room temperature, and processed for electron microscopy as earlier described (Lee and Cheng, 2003).
The Student's t test was performed using the GB-STAT Statistical Analysis Software Package (Version 7.0; Dynamic Microsystems Inc, Silver Spring, Md).
Purity of Sertoli and Germ Cells
The purity of Sertoli and germ cells was established by RT-PCR to detect cell-specific RNA products from germ, myoid, Leydig, and Sertoli cells, such as the corresponding c-Kit receptor (Figure 1A), fibronectin (Figure 1B), 3β-HSD (Figure 1C), and testin (Figure 1D), respectively. This study illustrates that the Sertoli and germ cell preparations used in the experiments reported contained negligible contaminations of other cell types.
Relative Levels of Zyxin, Axin, and WASP in Sertoli and Germ Cells
To compare the relative levels of zyxin, axin, and WASP in Sertoli and germ cells, these cells were isolated from 20-, 60-, and 90-day-old rat testes. All 3 adaptors, both mRNAs and proteins, namely zyxin (Figure 2A and D), axin (Figure 2B and D), and WASP (Figure 2C and D), were found in 20-day Sertoli cells. The levels of zyxin (Figure 2A and D) and WASP (Figure 2C and D) were higher in 20-day Sertoli cells than in 20-day germ cells, whereas axin was more predominant in germ cells (Figure 2B and D). Zyxin was virtually undetectable in 20-day-old germ cells, yet it was detected in germ cells from 60- and 90-day-old rats (Figure 2A and D).
Immunohistochemical Localization of Zyxin in the Rat Seminiferous Epithelium and in Sertoli Cells Cultured In Vitro
The localization of zyxin in the seminiferous epithelium of adult rat testes was examined by immunohistochemistry. A negative control was performed using normal goat serum as a substitute for the zyxin antibody. Negligible staining was detected, as demonstrated by the stage V tubule shown in Figure 3A. Yet when sections of testes were incubated with an anti-zyxin antibody, zyxin was localized predominantly in spermatocytes at stages V—VII (Figure 3C and D vs Figure 3B and E) with some staining between elongating/elongate spermatids and Sertoli cells, consistent with its localization at the site of apical and basal ES (Figure 3B through E vs Figure 3A). Figure 3F is a representative fluorescent micrograph showing the immunocytochemical localization of zyxin in Sertoli cells that were cultured in vitro from 2 different experiments. The prominent nucleus of a Sertoli cell was stained with DAPI (Figure 3F). Zyxin was visualized as discrete patches of green fluorescence in the Sertoli cell cytoplasm, which is consistent with its role as an adaptor that recruits other proteins to the AJ site. Figure 3G shows the result of an immunoblotting experiment in which a testis lysate was resolved by SDS-PAGE under reducing conditions; only a prominent band corresponding to the electrophoretic mobility of zyxin, approximately 82 kDa, was detected. Attempts were also made to localize axin and WASP in the seminiferous epithelium; however, these 2 antibodies failed to yield satisfactory results. Also, both of these antibodies were shown to cross-react with other proteins by immunoblotting (data not shown).
Zyxin, Axin, and WASP Structurally Interact With the Cadherin/Catenin/Actin Complex, Intermediate Filaments, and Microtubules in the Seminiferous Epithelium by a Coimmunoprecipitation Technique
Coimmunoprecipitation was performed using zyxin-, axin-, or WASP-specific antibodies, with seminiferous tubule lysate as the starting material. N-Cadherin, an AJ-integral membrane protein, and its binding partners, such as β-catenin, p120ctn, and α-actinin, were shown to associate with axin, zyxin, and WASP (Figure 4). Yet nectin-3, afadin, and β1-integrin were not shown to structurally associate with zyxin, axin, or WASP (Figure 4). Interestingly, c-Src, an AJ-associated signaling molecule, was shown to associate with zyxin, axin, and WASP (Figure 4). Furthermore, actin, vimentin (an intermediate filament component), and α-tubulin (a component of microtubules) interacted structurally with axin, zyxin, and WASP (Figure 4). Unexpectedly, VASP, a zyxin adaptor protein found in other epithelia (for a review, see Reinhard et al, 2001), such as the liver, was not found in the testis (Figure 4). Collectively, these results suggest that zyxin, axin, and WASP structurally associate with the cadherin/catenin/ actin complex but not the nectin/afadin and the integrin/laminin complexes. They possibly recruit AJ-signaling molecules, such as c-Src, to regulate the adhesion function of the cadherin/catenin complex.
Further Structural Analysis of the Association of Zyxin with N-Cadherin, E-Cadherin, Actin, and Vimentin in the Testis
Immunoprecipitation results reported in Figure 4 have shown that the 3 AJ adaptors—namely, zyxin, axin, and WASP—associate with actin and vimentin, which suggests that they serve as a linker between N-cadherin and vimentin. To expand this observation, coimmunoprecipitation was performed using antibodies specific to N-cadherin and E-cadherin, 2 classic cadherins that are found in the testis. It is known that the classic cadherins that are found in the brain are actin based (for a review, see Gumbiner, 2000). Thus, brain lysate served as a control as shown in Figure 5. Virtually all N-cadherin and E-cadherin (∼95%) in the seminiferous epithelium indeed were shown to associate with actin filament (Figure 5A and B), and 5% or less of E-cadherin, but not N-cadherin, was found to associate with vimentin (Figure 5B). However, a substantial amount of both cadherins associated with vimentin in the testis, suggesting that the cadherin/catenin complex outside the blood tumor barrier uses both actin and an intermediate filament as attachment sites. On the contrary, only a negligible amount of vimentin interacted with N-cadherin and E-cadherin in the brain (Figure 5) (note: cadherins in the brain are actin based; see Gumbiner, 2000). This also confirms the results of our earlier immunoprecipitation analysis (Lee et al, 2003). Interestingly, zyxin was also shown to link to N-cadherin and E-cadherin in testes and seminiferous tubules (Figure 5).
To further validate these data that N-cadherin indeed is restricted to the ES in the seminiferous epithelium, which is an actin-based testis-specific AJ type, we used immunofluorescent microscopy and paraffin sections to revisit its localization in the testis. In this connection, it is notable that our earlier studies using frozen sections for immunohistochemistry have consistently localized N-cadherin to the basal ES only, consistent with an earlier report by Wine and Chapin (1999). However, Johnson and Boekelheide (2002a,b) reported that they could also localize N-cadherin to the apical ES, suggesting that N-cadherin in the seminiferous epithelium uses actin as its attachment site, which is consistent with an earlier report (Lee et al, 2003). Using frozen sections, N-cadherin and E-cadherin indeed were restricted to the basal ES. However, when paraffin sections were used, N-cadherin indeed was also found to be localized to the site of apical ES between elongating spermatids and Sertoli cells and was highest in stages V—VI (Figure 6A and B), which is consistent with earlier reported results (Johnson and Boekelheide, 2002a,b). Figure 6C and D are the corresponding controls of the seminiferous epithelium at stages V and VI, respectively, where the primary antibody was replaced with normal rabbit serum and a monoclonal antibody (IgG1 subclass) against human α1-antitrypsin (Silvestrini et al, 1990), illustrating that the staining shown in Figure 6A and B was specific to N-cadherin. We had tried to perform similar experiments using 2 commercially available E-cadherin antibodies without success, possibly because the titer of these antibodies was low. Figure 6E is an immunoblot using testis lysate resolved by SDS-PAGE and stained with the anti-N-cadherin antibody, which was used for the immunofluorescent microscopy study shown in Figure 6A and B. Only a single immunoreactive band corresponding to the electrophoretic mobility of N-cadherin (Mr 127 kDa) was detected, illustrating that the specificity of this antibody and the staining shown in Figure 6A and B indeed represent N-cadherin staining.
Changes in the Protein Levels of the 3 Adaptors and Their Associated Proteins at the AJ Site During AF-2364-Induced AJ Disruption in the Seminiferous Epithelium
Rats were fed AF-2364 (50 mg/kg bw) as described in “Materials and Methods” to induce germ cell loss from the seminiferous epithelium by perturbing the cell adhesion function (Cheng et al, 2001; Grima et al, 2001). While earlier biochemical and molecular studies using markers of the AJs, such as testin, have shown that the anchoring junctions between Sertoli and developing germ cells are the primary targets of AF-2364 (Grima et al, 1997, 2001; Grima and Cheng, 2000; Cheng et al, 2001), its ultrastructural effects in the seminiferous epithelium at the electron microscopy level remain obscure. Figure 7A shows the cross section of an intact seminiferous epithelium from a control rat testis. Four hours after AF-2364 treatment, the intercellular space between germ cells, particularly round spermatids, and Sertoli cells became clearly visible, which was not found in the control rat testis (Figure 7B vs A). By day 5, more intercellular spaces were detected between Sertoli cells and round spermatids, and this pattern was typical throughout the entire seminiferous epithelium in more than 95% of the tubules examined (Figure 7C). This study clearly illustrates that AF-2364 exerts its effects at the cell adhesion sites between Sertoli and germ cells. Yet the basal lamina apparently was not affected by this treatment (Figure 7B and C).
During AF-2364-induced AJ disruption in the testis, a time-dependent reduction in the protein levels of zyxin, axin, and WASP was detected (Figure 8A). This pattern of reduction was similar to those observed for nectin-3 and afadin (Figure 8A). An induction of N-cadherin was detected at 16 hours posttreatment and persisted until day 7 (Figure 8A), which was consistent with the results of a recent report (Lee et al, 2003). The levels of actin and α-tubulin remained relatively stable during AF-2364-induced germ cell loss from the seminiferous epithelium, but a mild increase in vimentin was detected (Figure 8A). Immunohistochemistry was performed using testes from control and AF-2364-treated rats for localizing zyxin. These tests confirmed the results from the immunoblotting analysis (Figure 8B). Virtually no immunoreactive zyxin was detected in the spermatocytes from the seminiferous epithelium 7 days after the AF-2364 treatment. At this time, the elongating/elongate spermatids, round spermatids, and spermatocytes had become detached from the epithelium and were released to the seminiferous tubule lumen.
Zyxin, Axin, and WASP Are Components of the Actin-Based Cadherin/Catenin AJ Complex but Also Associate With the Intermediate Filament-Based Desmosome-like Junctions in the Seminiferous Epithelium
The assembly and disassembly of functional AJs between Sertoli and germ cells in vitro and in vivo were shown to associate with the transient induction of several AJ proteins, such as N-cadherin, β-catenin, p120ctn, β1-integrin, β2-integrin, RhoB, ROCK (Rho-associated protein kinase, a Ser/Thr protein kinase), LIMK (lin-11 isl-1 mec-3kinase), and Rab 8B (see Figure 9) (Chen et al, 2003; Lau and Mruk, 2003; Lee et al, 2003; Lui et al, 2003a; Siu et al, 2003b). This strongly suggests that they have roles in the dynamics of AJs. Furthermore, these proteins likely interact structurally with other peripheral proteins in the cytoplasm, forming a functional multiprotein complex that acts as the platform for signaling transduction to regulate AJ dynamics during spermatogenesis to facilitate the movement of developing germ cells across the seminiferous epithelium. Indeed, studies by coimmuno-precipitation have demonstrated that β-catenin, α-catenin, c-Src, cortactin, integrin-linked kinase, and focal adhesion kinase (FAK) are components of the multiprotein complexes at the apical ES site in the testis (Wine and Chapin, 1999; Chapin et al, 2001; Mulholland et al, 2001; Lee et al, 2003; Siu et al, 2003b). Subsequent in vitro and in vivo studies have shown that AJ dynamics are regulated, at least in part, via the integrin/cadherin/testin/pFAK/PI 3-kinase (phosphatidylinositol 3-kinase)/p130cas (an adaptor protein of Mr 130 kd encoded by the Crkas gene, a Crk-associated protein, with an SH3 domain) (Chen et al, 2003; Lee et al, 2003; Siu et al, 2003b) and the integrin/RhoB/ROCK/LIMK/cofilin (Lui et al, 2003a) signaling pathways. As such, a thorough understanding of the molecular architecture of the AJ complexes, particularly of the adaptors that can recruit peripheral signaling molecules to the site of the ES, will be crucial to the study of the regulation and biology of AJ dynamics. Hereby, we have shown that zyxin, axin, and WASP indeed associate with the cadherin/catenin complex but not with the nectin/afadin complex or the β1- integrin—mediated complex in the seminiferous epithelium. This study used coimmunoprecipitation to show that zyxin structurally associates with α-actinin, which implies that it indirectly links to the cadherin/catenin complex via its binding with α-actinin, which in turn binds to α-catenin. Such linkage is likely because zyxin has been shown to interact directly with α-actinin in other epithelia (Li and Trueb, 2001). Axin and WASP may interact with components of the cadherin/catenin complex directly or via other associated proteins.
Interestingly, zyxin, axin, and WASP also have direct structural association with the underlying actin cytoskeleton, which implies that they have roles in the regulation of actin dynamics. Zyxin is known to bind VASP, which in turn docks profilin and monomeric actin to the junction site in other epithelia (for reviews, see Beckerle, 1998; Reinhard et al, 2001). We have examined whether VASP is an adaptor that tethers actin to the cadherin/catenin complex via zyxin. Surprisingly, no VASP was detected in the testis in a series of coimmunoprecipitation experiments. This result suggests that the testis is using another protein (or proteins) to substitute for the functional role of VASP, as found in other epithelia. It is likely that WASP is being used to replace the tethering function of VASP in the testis because of its structural association with the cadherin/catenin complex and its ability to bind profilin (for a review, see Mullins, 2000). Indeed, zyxin was shown to associate with WASP and vice versa, as reported herein. Furthermore, WASP can regulate actin dynamics via its interaction with and activation of the Arp2/3 complex. For instance, WASP in its native form is unfolded upon its activation by PI(4,5)P2 and Cdc42 GTPase. This exposes the binding sites for Arp2/3, which facilitates the assembly and branching of the actin filament (for reviews, see Mullins, 2000; Caron, 2002). Less is known about the functional role of axin in the actin network; it may act either as a scaffold to bring other peripheral proteins to the site of the actin filament or as a regulator of β-catenin, controlling the stability of the cadherin/catenin/actin complex (for a review, see Seidensticker and Behrens, 2000). Collectively, these data clearly imply that zyxin, axin, and WASP regulate actin cytoskeleton dynamics at the AJ site in the testis.
Other studies have shown that AJ dynamics are regulated by kinases, phosphatases, cytokines, proteases, and protease inhibitors, resulting in either the assembly or the disassembly of the AJ complexes (for reviews, see Cheng and Mruk, 2002; Lilien et al, 2002). In general, the dissolution of the AJ structural protein complexes, such as the cadherin/catenin and integrin/laminin complexes, and the eventual uncoupling from the actin cytoskeleton depend largely on their phosphorylation status (for reviews, see Gumbiner, 2000; Cheng and Mruk, 2002). For instance, when pervanadate-treated leukemia cells with enhanced tyrosine phosphorylation of cadherins and catenins (note: sodium vanadate is a PTPi) were examined, a significant reduction in α-catenin binding to E-cadherin was detected (Ozawa and Kemler, 1998). Furthermore, increased phosphorylation of β-catenin by recombinant c-Src can lead to cell-cell dissociation (Roura et al, 1999). Collectively, the results illustrate that the integrity of the cell adhesion function of the AJ functional units (eg, the cadherin/catenin, the nectin/afadin, and the integrin/laminin complexes) depends at least in part on the phosphorylation status of the constituent proteins. Since zyxin (Crawford and Beckerle, 1991), axin (Ikeda et al, 1998), and WASP (for a review, see Oda and Ochs, 2000) are phosphoproteins per se and are putative substrates of kinases and phosphatases, they are likely subjected to a similar regulatory mechanism. The fact that these 3 adaptors are structurally associated with the cadherin/catenin complex seemingly suggests that their role in regulating the cell adhesion function occurs through changes in their phosphorylation status. In turn, the binding of the adaptors with cadherin and its associating proteins is affected, causing a disruption or an enhancement of the adhesion function of the cadherin/catenin complex. Equally important, we have shown that c-Src, an AJ nonreceptor protein tyrosine kinase, structurally associates with zyxin, axin, and WASP, which further suggests that they are substrates of this protein kinase at the site of AJs.
Zyxin, Axin, and WASP Are Structural Adaptors and Functional Linkers Between the Cadherin/Catenin Complex and the Actin Filaments, Intermediate Filaments, and Microtubules in the Seminiferous Epithelium
Actin microfilaments, intermediate filaments, and microtubules are the 3 cellular cytoskeletons in the testis. Unlike the actin network, less is known about the regulatory functions of zyxin, axin, and WASP in the intermediate filament-based and microtubule-based cytoskeletons. We have shown that zyxin, axin, and WASP associate with vimentin (a structural component of the intermediate filament) and α-tubulin (a subunit of the microtubule). Several recent studies have suggested that axin and WASP are crucial to the regulation of microtubule dynamics. First, axin sequesters GSK-3β away from tau, a microtubule-associated GSK-3β substrate, altering its phosphorylation and microtubule dynamics (Stoothoff et al, 2002). Second, WASP is involved in the regulation of microtubules during podosome formation in primary human macrophages (Linder et al, 2000). On the other hand, these adaptors also regulate the function of different cytoskeletal networks simply by providing a direct linkage of these networks. For instance, the direct interaction between fimbrin, an actin filament-binding protein, and vimentin has been confirmed in macrophages by studies of coimmunoprecipitation and colocalization (Correia et al, 1999). Plectin, an intermediate filament structural protein, was also shown to act as a link that connects the intermediate filament to microtubules and actin cytoskeletons (Svitkina et al, 1996). These observations clearly illustrate that one cytoskeletal network can directly affect the other networks via these interacting linkage proteins, such as zyxin, axin, and WASP. Collectively, these proteins may provide the crucial linkages for the cross-talking that must occur in the seminiferous epithelium among the 3 cytoskeletons—namely, the actin, the intermediate filament, and the microtubule networks.
Does the Classic Cadherin/Catenin Complex in the Testis Use the Actin Filament and/or Intermediate Filament as an Attachment Site?
The actin filament is the underlying cytoskeleton and attachment site for classic cadherins, such as N-cadherin and E-cadherin, in many organs and tissues, including the brain (for a review, see Gumbiner, 2000), as illustrated in this report. In light of the complexity and uniqueness of the junction structures and arrangements in the testis, whether classic cadherins are actin based or vimentin based remains a subject of dispute. To date, several transmembrane proteins have been identified at the ES site, which is a testis-specific actin-based type of AJ. These proteins include α6-integrin, β1-integrin, testin, cadherins, and nectins (for reviews, see Cheng and Mruk, 2002; Takai and Nakanishi, 2003). However, classic cadherins have been shown to associate with the intermediate filament-based desmosome-like junctions in the testis (Mulholland et al, 2001; Johnson and Boekelheide, 2002b). In contrast, 2 other reports have demonstrated that the cadherin/catenin complex is actin based (Wine and Chapin, 1999; Lee et al, 2003). For instance, studies using coimmunoprecipitation have shown that N-cadherin structurally associates with β-catenin, forming a functional cadherin/catenin complex that uses actin filament as the attachment site (Lee et al, 2003). In light of these intriguing findings, we hypothesized that classic cadherin might indeed structurally link to vimentin-based intermediate filament via the vimentin-associating adaptors, such as zyxin, axin, and WASP. Indeed, we have shown that a small amount of cadherin (eg, E-cadherin) is associated with vimentin in the seminiferous epithelium using a Leydig cell-free seminiferous tubule for lysate preparation. Surprisingly, a much higher amount of vimentin can be retrieved with an anti-N-cadherin or E-cadherin antibody using testis lysates. This suggests that cells (probably Leydig cells and others, such as myoid cells, residing outside the seminiferous epithelium) are using vimentin as the attachment sites for the classic cadherin/catenin protein complexes. In conclusion, the data reported support our earlier finding that classic cadherins within the seminiferous tubules are indeed largely actin based, with only 5% or less associated with the vimentin-based intermediate filament, whereas a larger portion (∼50%) of classic cadherins associate with the intermediate filament component in the testis but outside the blood-testis barrier.
Is the Loss of Zyxin, Axin, and WASP From the Seminiferous Epithelium Following AF-2364 a Result of Cytotoxicity?
While the physiological significance of the AF-2364-induced loss of zyxin, axin, and WASP from the seminiferous epithelium at the time germ cells are also depleted from the epithelium is not entirely understood, it does not appear to be a result of cytotoxicity. First, an induction of N-cadherin, in contrast to these adaptors, was detected in the seminiferous epithelium following AF-2364 treatment. This is consistent with several earlier reports, which showed that the loss of cell adhesion function between Sertoli and germ cells that was induced by AF-2364 is also accompanied by an induction of cadherins and catenins (Chen et al, 2003; Lee et al, 2003). Second, the levels of actin and tubulin remained relatively stable throughout the treatment period. Third, and perhaps most significantly, the use of AF-2364 at the dosing reported (50 mg/kg bw) or at dosings up to 20 times higher (ie, Irwin dose range: 100–1000 mg/kg bw, single dose) has safely passed the acute toxicity test in mice and rats conducted by licensed toxicologists (Cheng, unpublished data). Additionally, genotoxicity tests performed by licensed toxicologists according to Food and Drug Administration guidelines have shown that a single dose of AF-2364 (1, 10, 1000, or 2000 mg/kg b.w. by gavage) failed to induce a significant increase in micronucleated polychromatic erythrocytes in either male or female mice. Thus, AF-2364 is negative in the mouse micronucleus assay in vivo. Furthermore, AF-2364 (at 12.5–75 μg/mL) is negative for the induction of numerical chromosome aberrations in CHO cells in vitro in both the nonactivated and activated test systems (Cheng, unpublished data). These results are also consistent with recent serum microchemistry findings that AF-2364 is neither nephrotoxic nor hepatotoxic to rats at doses (up to 50 mg/kg bw) effective enough to induce reversible infertility (Cheng et al, 2001; Grima et al, 2001). It is likely that AF-2364 suppresses these adaptors at the site of Sertoli—germ cell AJs, disrupting the recruitment of AJ-associated proteins to these sites and thereby perturbing the cell adhesion function in the epithelium. However, this possibility must be vigorously investigated in future studies.
We thank M. Y. Mo for his excellent technical assistance in performing nucleotide sequence analyses to verify the authenticity of the PCR products as reported herein.
This work was supported in part by grants from the National Institutes of Health (NICHD, U01 HD45908 [C.Y.C.] and U54 HD29990, Project 3 [C.Y.C.]), the CONRAD Program (CICCR, C1G-01–72 to C.Y.C.), and the Noopolis Foundation.