Ezrin is an Actin Binding Protein That Regulates Sertoli Cell and Spermatid Adhesion During Spermatogenesis

Abstract

During spermatogenesis, the transport of spermatids and the release of sperms at spermiation and the remodeling of the blood-testis barrier (BTB) in the seminiferous epithelium of rat testes require rapid reorganization of the actin-based cytoskeleton. However, the mechanism(s) and the regulatory molecule(s) remain unexplored. Herein we report findings that unfold the functional significance of ezrin in the organization of the testis-specific adherens junction at the spermatid-Sertoli cell interface called apical ectoplasmic specialization (ES) in the adluminal compartment and the Sertoli cell-cell interface known as basal ES at the BTB. Ezrin is expressed at the basal ES/BTB in all stages, except from late VIII to IX, of the epithelial cycle. Its knockdown by RNA interference (RNAi) in vitro perturbs the Sertoli cell tight junction-permeability barrier via a disruption of the actin microfilaments in Sertoli cells, which in turn impeded basal ES protein (eg, N-cadherin) distribution, perturbing the BTB function. These findings were confirmed by a knockdown study in vivo. However, the expression of ezrin at the apical ES is restricted to stage VIII of the cycle and limited only between step 19 spermatids and Sertoli cells. A knockdown of ezrin in vivo by RNAi was found to impede spermatid transport, causing defects in spermiation in which spermatids were embedded deep inside the epithelium, and associated with a loss of spermatid polarity. Also, ezrin was associated with residual bodies and phagosomes, and its knockdown by RNAi in the testis also impeded the transport of residual bodies/phagosomes from the apical to the basal compartment. In summary, ezrin is involved in regulating actin microfilament organization at the ES in rat testes.

In the mammalian testis, junction remodeling takes place at the spermatid-Sertoli cell interface known as apical ectoplasmic specialization (ES) to facilitate the transport of spermatids across the epithelium during the epithelial cycle (1, 2). Furthermore, junction restructuring also takes place at the Sertoli cell-cell interface called basal ES at the blood-testis barrier (BTB) to facilitate the transport of preleptotene spermatocytes across the barrier (3, 4). Also, adhesion protein complexes at the apical ES and basal ES that use F-actin for attachment undergo rapid deadhesion and readhesion (5–7). Although morphological details of germ cell transport involving actin-based cytoskeleton during spermatogenesis in rodents are known, molecular mechanism(s) that regulates cytoskeletal reorganization remains elusive. Because apical and basal ES are constituted by bundles of actin filaments that lie between cisternae of the endoplasmic reticulum and the apposing plasma membranes (5, 8), these actin filament bundles must be rapidly reorganized via debundling and rebundling and vice versa during germ cell transport (3). However, the protein(s) that supply regulated linkage between integral membrane proteins plus peripheral proteins (eg, adaptors, nonreceptor protein kinases, and phosphatases) and the actin cytoskeleton at the ES remains unknown. A better understanding of the proteins that organize the ES is important because this information can unravel the mechanism(s) that regulates changes in cell adhesion and deadhesion during germ cell transport.

Ezrin, radixin, and moesin family proteins that tether actin microfilaments to integral membrane proteins as well as peripheral proteins (eg, adaptors) in mammalian cells to organize apical membrane domain including tight junction (TJ) and adherens junction (AJ), which thus create a scaffold for signaling molecules to regulate cell migration, proliferation, adhesion, and polarity (9–12). However, there was a misconception that these three proteins functionally overlap. In fact, ezrin, radixin, and moesin proteins rarely coexist in the same mammalian cell, and they are functionally distinct. For instance, ezrin is expressed mostly in polarized epithelial and mesothelial cells (13, 14), radixin in hepatocytes (15, 16), and moesin primarily in endothelial and lymphoid cells (13, 17), In ezrin^−/− mice, neonates are normal at birth, but they die by 21 days postpartum due to defects in the epithelium of the small intestine in which ezrin is highly expressed (14), illustrating radixin and moesin fail to supersede the lost function of ezrin. Radixin^−/− mice are viable but develop liver injury by 8 weeks of age due to defects in the localization of Mrp2 (multidrug resistance related protein 2, an efflux drug transporter) at the bile canalicular membranes to secrete conjugated bilirubin (18). Moesin-deficient mice are viable without gross abnormalities, and both male and female moesin^−/− mice are fertile (19). A study by immunohistochemistry has detected ezrin, radixin, and moesin in the mouse testis and the association of ezrin with residual bodies, phagosomes, and apical ES structures (20). However, the function of these proteins, such as ezrin, remains unknown. In humans, ezrin is known to associate with spermatozoa, involved in sperm capacitation (21). Ezrin is also structurally involved in assembling intercellular bridges, also called tunneling nanotubes (TNT) (22–24) for cell-cell communication in cancer cells (25). Herein we report findings on ezrin and its involvement in ES organization.

Materials and Methods

Animals and antibodies

Sprague Dawley rats were from Charles River Laboratories. The use of animals was approved by the Rockefeller University Institutional Animal Care and Use Committee with protocol number 12-506.

Isolation and culture of testicular cells

Sertoli cells were isolated from 20-day-old rat testes and cultured in serum-free F12/DMEM medium (26). Cells were plated on Matrigel (BD Biosciences)-coated coverslips, 12-well culture dishes, or Millicell HA (mixed cellulose esters) cell culture inserts (diameter 12 mm; pore size 0.45 μm; effective surface area 0.6 cm²) (EMD Millipore) at 0.05 (or 0.005 to monitor the presence of intercellular bridges or tunneling nanotubes), 0.5, and 1.2 × 10⁶ cells/cm², respectively, for the corresponding experiments as follows: (1) dual-labeled immunofluorescence analysis, 2) lysate preparation for immunoblotting or RNA extractions for RT-PCR, and 3) assessing the Sertoli cell TJ-permeability barrier function by quantifying transepithelial electrical resistance (TER) across the cell epithelium (27, 28). Germ cells were isolated from adult rat testes as described (29) and used for RNA extraction or lysate preparation within 12 hours.

Ezrin silencing in primary Sertoli cells cultured in vitro by RNA interference (RNAi)

Sertoli cells were cultured in vitro alone for 3 days to allow the establishment of a functional TJ-permeability barrier, containing ultrastructures of TJ, basal ES, gap junction, and desmosome (30, 31), mimicking the BTB in vivo (3). Thereafter cells were transfected with ezrin-specific small interfering RNA (siRNA) duplexes vs nontargeting control siRNA duplexes at 100 nM for 24 hours using RiboJuice (Novagen, EMD Biosciences) as a transfection medium. siRNA duplexes that specifically targeted ezrin were as follows: sense, 5′-GGACUUAACAUUUAUGAGAtt-3′, antisense, 5-UCUCAUAAAUGUUAAGUCCaa-3 (s132626); and sense, 5′-GGACAGUGCUAUGCUCGAAtt-3′, antisense, 5′-UUCGAGCAUAGCACUGUCCtt-3′ (s132627) (Ambion). Nontargeting siRNA duplex (Silencer Select Negative Control 1 siRNA; Ambion) served as the negative control. These duplexes were selected since other siRNA duplexes showed low efficacy or no effects based on pilot experiments. For studies that assessed the effects of RNAi on the Sertoli cell TJ-barrier function, 150 nM siRNA duplexes were used. After transfection for 24 hours, cells were washed and cultured with fresh F12/DMEM, and cells were collected for RNA extraction for RT-PCR analysis. For the immunoblot and immunofluorescence analysis, cells were cultured for an additional 24 hours before termination. In some RNAi experiments, the cells were cotransfected with 1 nM siGLO red transfection indicator (Dharmacon) to track successful transfection (27).

Assessment of TJ-permeability barrier in vitro

Sertoli cell TJ-permeability barrier was assessed by quantifying TER across the cell epithelium as described (26). Each time point had triplicate inserts, and each experiment was repeated three times using different cell preparations.

Ezrin silencing in adult rat testis in vivo

To knock down ezrin in vivo, rats (∼350–375 g body weight) were transfected with siRNA duplexes via intratesticular injection using a 28-gauge needle as described (27). On day 0, a testis of the same rat received nontargeting control siRNA duplexes vs the other testis received ezrin-specific siRNA duplexes shown to perturb the Sertoli cell TJ-barrier in vitro. siRNA duplexes at the desired concentration (100 nM) were constituted in a transfection mix of 200 μL containing 7.5 μL Ribojuice siRNA transfection reagent (EMD Millipore) and 192.5 μL Opti-MEM reduced serum medium (Invitrogen). Thus, each testis (∼1.6 g, with a volume of ∼1.6 mL) received 200 μL of this solution for transfection. In short, the 28-gauge needle (13 mm long) attached to a 1-mL syringe was loaded from the apical to near the basal end of the testis vertically, and as the needle was withdrawn apically, transfection mixture was gradually and gently released from the syringe so that the entire testis was filled with the 200 μL mixture that spread throughout the testis (32) to avoid an acute rise in intratesticular hydrostatic pressure. Transfection was performed on day 0, to be followed by a second transfection 48 hours thereafter (ie, on day 2). Rats were euthanized on day 4 (n = 5 rats) and day 6 (n = 5 rats). Because the pilot experiments had shown that the phenotypes were similar when rats were terminated on day 4 vs day 6, data were thus pooled for analysis. For immunoblotting, testes were snap frozen in liquid nitrogen and stored at −80°C until used. Some testes were fixed in Bouin's fixative to obtain paraffin sections for histological analysis using hematoxylin and eosin staining.

Immunoblotting, coimmunoprecipitation (Co-IP), and RT-PCR

Lysates were obtained from Sertoli and germ cells, testes, and seminiferous tubules. Tubules isolated from adult rat testes (33) were used within 2 hours after their isolation, which were devoid of Leydig cell contamination (33). Antibodies used for immunoblotting or Co-IP are listed in Supplemental Table 1. Co-IP was performed using lysates (∼600 μg protein) from testes or tubules (34, 35). Chemiluminescence was performed and the immunoblots were analyzed as described (36). RNA extraction and PCR using specific primer pairs (Supplemental Table 2) were performed as described (27, 37). The PCR products were verified by direct DNA sequencing performed at Genewiz.

Dual-labeled immunofluorescence analysis and immunohistochemistry (IHC)

Dual-labeled immunofluorescence analysis was performed using frozen cross-sections of testes and specific primary antibodies and corresponding secondary antibodies (Supplemental Table 1) including negative controls as described (38). F-actin was stained by either fluorescein isothiocyanate (FITC)- or rhodamine-conjugated phalloidin as described (28, 38). Cell nuclei were visualized by 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen). Fluorescence images were obtained in an Olympus BX61 fluorescence microscope (38). Image files were analyzed using Photoshop in Adobe Creative Suite (version 3.0) such as for image overlap to assess protein colocalization (27, 39). IHC was performed in which Bouin-fixed, paraffin-embedded sections were deparaffinized, rehydrated, and then subjected to antigen retrieval in 10 mM citrate buffer (pH 6.0) for 10 minutes in a microwave. Sections were blocked with 10% normal goat serum, followed by overnight incubation at 4°C with antiezrin antibody (Supplemental Table 1). Thereafter sections were incubated with biotinylated goat antirabbit IgG and then with streptavidin-horseradish peroxidase, and color development was performed using 3-amino-9-ethylcarbazole.

Assessment of defects in spermatogenesis after silencing of ezrin in the testis

Histological analysis was used to assess defects in spermatogenesis. Changes were scored by examining approximately 80 randomly selected cross-sections of stage VIII and stage IX tubules from a rat testis, and a total of four rats were examined. Thus, a total of approximately 320 stage VIII and IX tubules were scored. Data were compared with control testes transfected with nontargeting duplexes, and changes were expressed as a percentage of total scored tubules in treatment vs control groups. First, defects in spermatid transport that led to defects in spermiation were defined by the presence of more than five step 19 spermatids that were retained within the seminiferous epithelium in the cross-section of a stage VIII or stage IX tubule. Second, defects in spermatid polarity that might contribute to spermatid transport/adhesion as reported (35) was defined by the presence of more than five step 19 spermatids in which their heads deviated by at least 90° from the intended orientation of pointing toward the basement membrane. Third, defects in the transport of phagosomes were determined. In normal testes, at stage VIII, residual bodies/phagosomes are found in the adluminal compartment near the tubule lumen, which are transported to the basal compartment at stage IX for degradation. A tubule with defects in phagosome transport is defined by the presence of more than three phagosomes found in the adluminal compartment at stage IX.

BTB integrity assay to assess Sertoli cell barrier integrity in vivo

A functional assay was performed to assess whether the knockdown of ezrin in the testis would impede BTB integrity as described (40). This assay monitored the ability of an intact Sertoli cell barrier to block the passage of a fluorescence tag, FITC-conjugated inulin (Mr 4.6 kDa) (Sigma-Aldrich), into the adluminal compartment in control rats transfected with nontargeting duplexes (n = 3 rats) vs ezrin siRNA duplexes (n = 3 rats). In brief, transfection was performed on day 0, to be followed by a second transfection 48 hours thereafter (ie, on day 2). On day 4, the integrity of the BTB was assessed (40). Approximately 80 tubules were randomly scored for BTB disruption in cross-sections of a rat testis, and a total of approximately 240 tubules from three testes were examined and scored in each group.

Image analysis

For in vitro RNAi in Sertoli cells, at least 200 cells were randomly selected and examined in control vs experimental group with three experiments. Fluorescence intensity of a target protein in Sertoli cells or in the seminiferous epithelium of testes was quantified (27, 39) using ImageJ 1.45 software (US National Institutes of Health, Bethesda, MD; http://rsbweb.nih.gov/ij). At least 50 randomly selected stage VIII vs stage IX tubules from cross-sections of a rat testis were examined with four rats. Analysis was focused on these two stages because defects in spermatid transport that impeded spermiation, loss of spermatid polarity, and defects in phagosome transport were readily detected.

Statistical analysis

For studies using Sertoli cell cultures, triplicate coverslips, dishes, or bicameral units were used. Each data point (or bar graph) is a mean ± SD of three to five experiments (or n = 5 rats). For each experiment, data in treatment groups were normalized against the corresponding control, which was arbitrarily set at 1. Statistical analysis was performed using the GB-STAT software package (version 7.0; Dynamic Microsystems). Statistical analysis was performed by two-way ANOVA followed by Dunnett's test. In selected experiments, a Student's t test was used for paired comparisons.

Results

Stage-specific expression of ezrin at the ES in the rat testis

Ezrin, an 85-kDa actin-binding protein, was expressed by both Sertoli and germ cells in the rat testis when examined by either RT-PCR (Figure 1A) using a primer pair specific to ezrin (Supplemental Table 2) or immunoblotting (Figure 1B) using a specific antiezrin antibody (Supplemental Table 1). When Sertoli cells were cultured at 5 × 10⁴ cells/cm² for 4 days, ezrin was shown to partially colocalize with actin microfilaments in cell cytosol (Figure 1C). When Sertoli cell density was reduced by approximately 10-fold to 5 × 10³ cells/cm², ezrin was found to colocalize with actin microfilaments, constituting the intercellular bridges (or TNTs) (Figure 1D), analogous to its involvement in organizing TNT in human cells (25). The specificity of this antiezrin antibody was illustrated by immunoblotting using the lysate of either Sertoli or germ cells (Figure 1E and Supplemental Table 1).

Figure 1.

A–G, Expression of ezrin by Sertoli cells and germ cells, and its stage-specific localization in the seminiferous epithelium of adult rat testes. A, Relative expression of ezrin in adult rat testis (T), Sertoli cells (SC), and germ cells (GC) vs kidney (K; served as a positive control) was analyzed by RT-PCR. S-16 served as a loading and PCR control. M, DNA size markers in base pairs. B, Lysates of testes (T) from adult rats, Sertoli cells (SC), and germ cells (GC) were used (∼30 μg protein per sample) for immunoblotting to assess the steady-state protein level of ezrin. β-Actin served as a protein loading control. Immunoblot data were summarized in this histogram, with each bar representing a mean ± SD of three samples, normalized against actin. The relative protein level of ezrin in the testis was arbitrarily set at 1. C, Dual-labeled immunofluorescence analysis to assess colocalization of ezrin (red fluorescence) with F-actin (green fluorescence) in Sertoli cells. Sertoli cell nuclei were visualized by DAPI. Scale bar, 15 μm (applies to all micrographs). D, Colocalization of ezrin (red) with F-actin (green) at the intercellular bridge (yellow arrows) between two Sertoli cells. Sertoli cell nuclei were visualized by DAPI (blue). Scale bar, 40 μm (applies to all micrographs). E, Specificity of the antiezrin antibody was assessed by immunoblotting using lysate of Sertoli (SC) and germ cells (GC). β-Actin served as a protein loading control. F, Localization of ezrin (red) in the small intestine (SI) at duodenum using frozen sections to confirm ezrin expression in the epithelial cells to serve as a positive control. Primary antibody replaced by normal mouse IgG served as a negative control. Scale bar, 120 μm on the left panel; 300 μm on the right panel. G, IHC localization of ezrin (reddish brown precipitates) in the testis, illustrating the localization of ezrin at the apical ES. Ezrin was mostly localized at the convex (dorsal) side and the tip of spermatid heads but was limited to stage VIII tubules (see enlarged boxed image). In a stage IX tubule, ezrin was detected in residual bodies (green arrows) and also phagosomes (red arrow). Leydig cells were not ezrin positive, but ezrin occasionally stained a few cells in the interstitium, apparently macrophages, as shown in this stage IX tubule; blue arrow illustrates a peritubular myoid cell stained positive for ezrin. Scale bar, 50 μm in all micrographs; scale bar 25 μm in green boxed inset. In negative control, primary antibody was replaced with normal rabbit IgG; scale bar 150 μm. H, Localization of ezrin (red) in the seminiferous epithelium of adult rat testes. Expression of ezrin at the apical ES was mostly limited to stage VIII tubules. However, at stage IX, ezrin also stained prominently with residual body. At the basal ES/BTB, ezrin was detected in all stages of the cycle except at stage IX when its expression was considerably diminished. Insets represent enlarged images of ezrin in corresponding boxed areas, illustrating ezrin was detected mostly at the convex side of spermatid heads. Scale bar, 50 μm (applies to other micrographs); scale bar, 25 μm in inset. Findings shown herein are representative data of three to five experiments.

To further confirm the specificity of the antibody, we examined the localization of ezrin in the small intestine by immunofluorescence microscopy (IF). Ezrin (red fluorescence) was found to localize exclusively to the epithelial cells of the apical microvilli (Figure 1F), consistent with findings of an earlier report (13). Also, the substitution of the primary antiezrin antibody (Supplemental Table 1) with normal mouse IgG did not yield any staining, illustrating the antibody specificity (Figure 1F). Thus, this antibody was used to examine the spatiotemporal expression of ezrin by IHC using paraffin-embedded (Figure 1G) and IF using frozen (Figure 1H) sections of adult rat testes. By IHC (Figure 1G), the expression of ezrin was stage specific, prominently expressed at the apical ES at stage VIII, surrounding the spermatid head but most notably on the dorsal (convex) side.

Ezrin was also highly expressed and associated with the residual body undergoing phagocytosis (green arrows) that formed phagosome (red arrow) at stage IX, consistent with findings of an earlier report (20) (Figure 1G). Peritubular myoid cells were also ezrin positive (Figure 1G). Leydig cells in the interstitial space were negative for ezrin; however, a few cells, apparently resident macrophages, were found to be positive for ezrin in the interstitium, and the ezrin staining by IHC appeared to be specific because negative control yielded no staining (Figure 1G). As noted in Figure 1H, although ezrin (red fluorescence) was detected near the base of the seminiferous epithelium, consistent with its localization at the BTB, at all stages of the epithelial cycle except stage IX; its expression at the apical ES was highly stage specific (Figure 1H). At the apical ES, ezrin was weakly detected in stage VII, but it was robustly expressed at stage VIII and almost exclusively on the dorsal (convex) side of the spermatid head (Figure 1H), consistent with the findings of IHC shown in Figure 1G. At stage IX, ezrin was also found to associate with residual bodies undergoing phagocytosis by the Sertoli cell (green arrows), which is an F-actin-dependent event (41, 42). These findings illustrate ezrin is a component of the basal and apical ES, and phagocytosed residual body/phagosome, likely involved in spermatid and phagosome transport.

Ezrin is an ES component and structurally interacts with F-actin and other actin binding/regulatory proteins

By Co-IP, ezrin was found to structurally interact with TJ-protein junctional adhesion molecule-A (JAM-A), and basal ES-protein N-cadherin at the BTB (Figure 2A). Ezrin also interacted with apical ES protein laminin-γ3 chain but not nectin-3 (Figure 2A). Ezrin also interacted with the following: 1) apical ES regulatory protein p-FAK-Tyr³⁹⁷ (39); 2) ES regulatory protein c-Src (43); 3) actin-related protein (Arp)-3 [which together with Arp2 forms the Arp2/3 complex known to induce barbed end actin polymerization, effectively converting actin filaments from a bundled to an unbundled/branched network (3)]; and 4) actin (Figure 2A). Data shown in Figure 2B also demonstrated partial colocalization of ezrin with F-actin. To further confirm that ezrin is an integrated component of the apical and basal ES, dual-labeled immunofluorescence analysis was performed using specific markers of apical ES: p-FAK-Tyr³⁹⁷ and β1-integrin (Figure 2B) as well as TJ proteins: JAM-A and occludin and basal ES protein N-cadherin at the BTB (Figure 2C). Ezrin colocalized with both apical ES proteins p-FAK-Tyr³⁹⁷ and β1-integrin in the adluminal compartment (Figure 2B), and TJ/basal ES proteins JAM-A, occludin, and N-cadherin at the BTB (Figure 2C). Collectively these findings illustrate that ezrin is an integrated component of the apical ES in the adluminal compartment and the basal ES/TJ at the BTB.

Figure 2.

A–C, A study to assess whether ezrin is an integrated component of the apical and basal ES in adult rat testes. A, Co-IP using lysates of seminiferous tubules was performed using specific markers of basal ES/BTB, apical ES, actin regulatory, and cytoskeletal proteins to identify specific protein-protein interaction with ezrin. IgG, both heavy (H) and light (L) chains, served as the protein loading control. Seminiferous tubule lysate (ST; 30 μg protein) served as positive control. +, positive protein-protein interaction with ezrin; −, negative protein-protein interaction with ezrin. B and C, To further confirm data from the Co-IP experiment, dual-labeled immunofluorescence analysis was performed to assess colocalization of ezrin (red) with either apical ES (green) proteins (B) or basal ES/BTB (green) proteins (C). Scale bar, 30 μm, in panels B and C, which applies to other micrographs. Data are representative findings from three experiments.

Knockdown of ezrin perturbs Sertoli cell TJ-permeability barrier, mediated by changes in the organization of actin microfilaments and protein localization

When ezrin was knocked down by RNAi with an efficacy of approximately 90% (Figure 3, A and B), the expression of several proteins at the Sertoli cell BTB was not affected except for N-cadherin and palladin, which displayed a down-regulation (Figure 3A and Supplemental Figure 1). When the fluorescence intensity of ezrin in Sertoli cells was monitored after the ezrin knockdown, ezrin was found to be silenced by at least 80% (Figure 3C), consistent with data shown in Figure 3, A and B. Interestingly, a knockdown of ezrin also perturbed the Sertoli cell TJ-permeability barrier (Figure 3D). Knockdown of ezrin appeared to be specific for ezrin because the expression of two interferon-stimulated genes, OAS1 and STAT1, were found not to be affected (Figure 3E) because these two genes were up-regulated nonspecifically when mammalian cells transfected by RNAi vectors for gene silencing were having off-target effects (44–46).

Figure 3.

A–F, A study to assess effects of ezrin knockdown on the Sertoli cell TJ-permeability barrier, F-actin organization, and protein localization at the cell-cell interface in vitro. A and B, Sertoli cells cultured for 2 days (day) with a functional TJ-barrier were transfected with ezrin-specific siRNA duplexes (Ezrin RNAi) to knock down its expression by almost 90% vs nontargeting control siRNA duplexes (Ctrl RNAi). Effects of ezrin knockdown on the expression of target proteins were assessed by immunoblotting, with actin, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as protein loading controls (A). Knockdown of ezrin affected the expression of N-cadherin and palladin but not other BTB protein markers (A; see also Supplemental Figure 1). Each bar in panel B and Supplemental Figure 1 is a mean ± SD of three experiments. The level of a marker protein including ezrin in control cells was arbitrarily set at 1, against which statistical comparison was performed. **, P < .01. C, Localization of ezrin (green) in ezrin knockdown Sertoli cells vs controls (top panel). Red fluorescence represents siGLO red transfection indicator (Dharmacon), illustrating successful transfection. Bar, 20 μm, which applies to other micrograph. Image analysis (lower panel) to confirm the intensity of fluorescence signals of ezrin (green) in the Sertoli cells using ImageJ. *, P < .05. D, Changes in the Sertoli cell TJ-permeability barrier after ezrin knockdown vs control were assessed by quantifying TER across the cell epithelium at specified time points. Each data point is a mean ± SD of four bicameral units from a representative experiment with n = 3 experiments that yielded similar results. *, P < .05. E, Expression of two ISGs: OAS1 and STAT1 in addition to ezrin were used to assess nonspecific off-target effects. Neither OAS1 nor STAT1 was up-regulated after ezrin knockdown. M, DNA size markers in base pairs. F, Changes in the expression, organization, and/or localization of ezrin, F-actin and marker proteins in Sertoli cells or at the cell-cell interface after ezrin knockdown vs controls were assessed by IF. Ezrin (green) signal was considerably diminished after its knockdown. Ezrin knockdown altered F-actin organization in which actin microfilaments in Sertoli cells were truncated and disorganized (annotated by asterisks) vs controls. These changes thus impeded adhesion protein localization that used F-actin for attachment, thereby perturbing the Sertoli cell barrier, supporting findings shown in panel D. Furthermore, the localization of palladin, N-cadherin, and ZO-1 (but mildly for ZO-1) was disrupted, and the expression was also diminished. Sertoli cell nuclei were visualized by DAPI (blue). Sertoli cells were cotransfected with 1 nM siGLO red transfection indicator (Dharmacon) to track successful transfection. Scale bar, 50 μm, which applies to other micrographs. G, In control cells, a TNT between two Sertoli cells constituted by ezrin and F-actin was clearly visible (annotated by yellow arrows). However, in ezrin-silenced cells, TNT failed to establish as annotated by red arrows. Scale bar, 50 μm.

We next examined the likely mechanism that caused the disruption of the Sertoli cell TJ-barrier function. As shown in Figure 3F, the organization of F-actin after the ezrin knockdown was disrupted because truncation of actin microfilaments was detected in Sertoli cells, and actin filaments were no longer organized as bundles (Figure 3F). This perhaps was due to the down-regulation of palladin (Figure 3F), an actin bundling protein in the testis (47). The localization of zonula occludens-1 (ZO-1) at the Sertoli cell-cell interface appeared to be mildly disturbed; N-cadherin, a basal ES protein at the Sertoli cell BTB, was no longer distributed tightly to the cell-cell interface (Figure 3F). Furthermore, a knockdown of ezrin also impeded the establishment of TNT between Sertoli cells in which F-actin, in the absence of ezrin, failed to assemble an intact TNT (Figure 3G).

Knockdown of ezrin in the testis in vivo perturbs apical ES function

Changes in the phenotype after ezrin knockdown were next examined. Efficacy of ezrin knockdown in the testis was first assessed by RT-PCR (Figure 4A). When ezrin was knocked down in the testis in vivo by approximately 60% based on immunoblot analysis, no significant changes in levels of ES proteins were detected (Figure 4B and Supplemental Figure 2). We focused our analysis in stage VIII tubules because ezrin expression at the apical ES was most prominent at this stage (see Figure 1G). We also elected to assess defects in spermatid transport and spermiation as well as spermatid polarity, which are the primary functions of the apical ES (2, 6). We included stage IX tubules in our analysis because this is the stage that follows stage VIII, and if there were defects in spermatid transport and spermiation, step 19 spermatids were expected to be seen in this stage. Also, pilot experiments illustrated no visible changes were detected in all other stages at the apical ES. Figure 4C illustrates the efficacy of the ezrin knockdown at the apical ES in vivo in which the expression of ezrin at the convex side of the spermatid head was considerably diminished.

Figure 4.

A–E, Effects of ezrin knockdown on spermatid transport, spermatid polarity, and phagosome transport in the adult rat testes in vivo at stage VIII-IX of the epithelial cycle. A and B, Efficacy of the ezrin knockdown in the testis in vivo was estimated by RT-PCR (A) and immunoblotting (B). M, DNA size markers in base pairs. Data in panel B are representative findings of three experiments (see Supplemental Figure 2). C, After the knockdown of ezrin (Ezrin RNAi) vs controls (Ctrl RNAi) in vivo, the signal of ezrin (red) in the seminiferous epithelium as illustrated in this stage VIII tubule (top panel) was considerably diminished. Ezrin signal at the apical ES was mislocalized (see insets, which are enlarged images of the boxed areas). Ezrin signal at the basal ES/BTB (yellow arrowheads) was also considerably diminished (red arrowheads). Polarity of spermatids was also lost (see yellow arrow in inset). Findings are representative findings of n = 4 rats. A histogram (bottom panel) that summarizes changes in ezrin fluorescence signal in the testis in randomly selected stage VIII tubules after ezrin knockdown vs control with each bar a mean ± SD of four rats. *, P < .05. Bar, 50 μm, which applies to the other panel; bar = 25 μm in inset. D, In this representative stage VIII tubule, a loss of spermatid polarity was noted in multiple step 19 spermatids (annotated by yellow arrows) in ezrin-silenced rats. Boxed area in green or red was magnified and shown in inset. Bar, 120 μm; bar, 25 μm in inset. E, Hematoxylin and eosin staining of paraffin sections of testes. Cross-sections of representative stage VIII and IX tubules in which insets of different colors are enlarged images of the corresponding boxed and colored areas that illustrate the entrapment of step 19 spermatids in the epithelium (annotated by black arrows), spermatids with lost polarity (yellow arrows), and phagosomes that failed to be transported to the basal compartment at stage IX (blue arrows). Scale bar, 320 μm (first panel); 80 μm (middle and bottom panel); 30 μm (inset). Data shown in panels D and E are representative findings of four rats.

For the remaining ezrin that was expressed at the apical ES, it was mislocalized, no longer restricted to the convex side of the spermatid head (Figure 4C), and defects in spermatid polarity (polarity defect is defined by a spermatid head no longer pointing toward the basement membrane; instead, it deviated by at least 90° from its intended orientation) was also detected in stage VIII tubules (Figure 4, C and D, and Table 1). More important, defects in spermatid transport were detected in which step 19 spermatids were embedded deep inside the epithelium in either stage VIII or IX tubules, near the basement membrane when they should have been aligned near the edge of the tubule lumen or be released to the tubule lumen (Figure 4E). Also, residual bodies that were engulfed by Sertoli cells at stage VIII that should have been transported to the basal compartment for degradation at stage IX (41) were found in the adluminal compartment in stage IX tubules (Figure 4E). Defects in spermatid polarity were also noted in these tubules (Figure 4E).

Table 1.

Phenotypic Changes in the Seminiferous Epithelium of Rat Testes After Silencing of Ezrin Using Specific siRNA Duplexes vs Nontargeting Controls^a

Characteristic	Change
Defects in spermatid transport/spermiation (with more than five elongated spermatids trapped in the seminiferous epithelium per cross-section of stage VIII–IX tubules)	16.5% ± 8.0%
Defects in spermatid polarity (more than five elongated spermatids loss polarity in stage VII–VIII tubules in which their heads were pointing at least 90° from the intended orientation)	8.2% ± 1.58%
Defects in phagocytosis (with more than three phagosomes found in the adluminal compartment in late stage IX tubules)	7.5% ± 2.77%

^aEach data point is a mean ± SD after ezrin RNAi of three rats in which data were normalized against control rats treated with nontargeting control siRNA duplexes.

Ezrin knockdown in the testis in vivo perturbs basal ES protein distribution and BTB integrity

We next examined any changes in the BTB after the ezrin knockdown in the testis in vivo because ezrin was expressed at the BTB during the epithelial cycle (Figure 1). After the knockdown of ezrin at the BTB by almost 60% (Figure 5, A and B), the fluorescence signals of F-actin, tight junction proteins occludin and ZO-1, and basal ES protein N-cadherin were found unaffected; however, the localization of N-cadherin was disrupted (Figure 5, A and B). For instance, N-cadherin was found to be diffused away from the BTB, no longer confined to the BTB near the basement membrane (Figure 5A), and such changes were statistically significant (Figure 5B). These findings are consistent with a loss of BTB integrity in which the Sertoli cell barrier in ezrin knockdown testes failed to block the passage of FITC-inulin from entering the adluminal compartment (Figure 5C, red arrows, right vs left panel) vs control testes (white arrowheads, left panel; and white or yellow arrowheads in right panel, which annotate the relative location of the basement membrane near the BTB). It is noted that in approximately 30% of the tubules examined, ezrin knockdown induced a loss of BTB integrity, mostly in stage VIII and IX tubules as noted in Figure 5C. For instance, about two of the five tubules shown in the right panel in the ezrin RNAi rat testis had disrupted BTB vs five tubules with intact BTB in the left panel (Figure 5C, Ctrl RNAi).

Figure 5.

A and B, A study to assess the effects of ezrin knockdown in the testis in vivo on the expression, localization, and distribution of proteins at the BTB. A, After the knockdown of ezrin, ezrin (green) signals in the seminiferous epithelium near the basement membrane (annotated by a dashed white line) consistent with its localization at the BTB were considerably diminished. F-actin at the BTB appeared to fail to form a continuous band at the BTB, and the distribution/localization of occludin and ZO-1 in the stage VIII tubules was not affected. However, N-cadherin no longer tightly localized to the BTB but became diffusely distributed (annotated by the white bracket). Scale bar, 20 μm, which applies to other micrographs. Findings are representative data of four experiments. B, Data shown in panel A was summarized in this histogram by assessing the following: 1) the intensity of the fluorescence signals for ezrin, F-actin, occludin, and ZO-1 in in stage VIII tubules; or 2) distance of N-cadherin fluorescence signal from the BTB (annotated by a dashed broken line) in both groups. Each bar is a mean ± SD of four rats. *, P < .05. C, BTB integrity was assessed by the ability of an intact Sertoli cell barrier to block the passage of FITC-insulin (green fluorescence) into the adluminal compartment by crossing the BTB (annotated by the white arrowheads in Ctrl RNAi, left panel; and also by white and yellow arrowheads in Ezrin RNAi for the two tubules in right panel). However, after the knockdown of ezrin, two of the tubules shown in the right panel had a disrupted BTB in which the fluorescence tag was found in the adluminal compartment (annotated by red arrows). These findings are representative data of three rats in each group.

A possible mechanism by which ezrin knockdown perturbs spermatid adhesion, impeding spermatid transport and polarity

Ezrin knockdown in the testis in vivo was found to down-regulate the expression of ezrin at the apical ES in stage VIII tubules by approximately 60% (Figure 6, A and B). Although the overall fluorescence intensity of F-actin at the apical ES after ezrin knockdown was not considerably different from the control, some step 19 spermatids were found to have considerably less F-actin (Figure 6A). The most striking differences detected were the expression and/or distribution of two apical ES regulatory proteins p-FAK-Tyr³⁹⁷ (39) and actin-bundling protein palladin (47). First, although the expression p-FAK-Tyr³⁹⁷ at the apical ES was not considerably diminished after the ezrin knockdown (Figure 6, A and B), the silencing of ezrin by 60% induced mislocalization of p-FAK-Tyr³⁹⁷ because p-FAK-Tyr³⁹⁷ was no longer limited to the convex side and the tip of spermatid heads as noted in control testes; instead it moved away from these sites (Figure 6, A and B), and some of these spermatids also lost their polarity (Figure 6A, white arrows). Second, the expression of actin bundling protein palladin was down-regulated by as much as approximately 8-fold (Figure 6, A and B); thus, many step 19 spermatids in stage VIII tubules lost their polarity because actin filament bundles could no longer maintain their proper organization at the apical ES. These changes in turn impeded the localization and/or expression of two apical ES adhesion proteins: nectin-3 and β1-integrin (Figure 7, A and B). Nectin-3 no longer restricted to the convex side of the spermatid head but encircled the tips of spermatid heads with abnormal polarity, and its expression was down-regulated (Figure 7, A and B). The expression of β1-integrin was also down-regulated in step 19 spermatids, in particular spermatids with defects in polarity (Figure 7, A and B).

Figure 6.

A and B, A study to assess the effects of ezrin knockdown in the testis in vivo on the expression, localization, and distribution of proteins at the apical ES, impeding spermatid transport and polarity in stage VIII tubules. A, After the knockdown of ezrin in the testis in vivo, ezrin signal (red fluorescence) in stage VIII tubules as shown herein were considerably diminished, consistent with data shown in Figure 4C. Many step 19 spermatids had lost their polarity (white arrows). Although the signal for F-actin was not grossly diminished in early-stage VIII tubules after ezrin knockdown, F-actin was diminished or not found in some step 19 spermatids (blue inset). p-FAK-Tyr³⁹⁷, the activated/phosphorylated form of FAK known to confer spermatid adhesion, thereby regulating spermatid transport and also polarity (39), was found to be mildly, but not statistically significantly, diminished (see panel B). However, p-FAK-Tyr³⁹⁷ was found to be mislocalized because it was no longer tightly restricted to the tip of the spermatid head as noted in control testes; instead, it moved away from the spermatid head (see green and yellow insets), and misoriented spermatids were annotated (white arrows). Palladin, an actin-bundling protein at the apical ES, was considerably diminished; this thus impeded the organization of the actin filament bundles at the apical ES, perturbing spermatid polarity (white arrow). It is noted that there were mild differences in DAPI intensity in ezrin-silenced testes vs control in the palladin panel, but such differences could not account for the approximately 8-fold reduction in palladin expression. Insets are enlarged images of the corresponding colored/boxed area. Scale bar, 50 μm (applies to other micrographs); scale bar in inset, 20 μm, which applies to other insets. Data are representative findings of three experiments. B, Image analysis of fluorescence signals shown in panel A. Each bar is a mean ± SD of four rats. *, P < .05; **, P < .01.

Figure 7.

A and B, A study to investigate changes in the expression and localization of apical ES proteins in stage VIII tubules after the in vivo knockdown of ezrin in the testis. A, Apical ES-specific adhesion proteins nectin-3 (66) and β1-integrin (67, 68) were found to be down-regulated in the seminiferous epithelium of testes in rats after the ezrin knockdown. More important, nectin-3 was mislocalized in which it no longer restricted to the dorsal (convex) side of step 19 spermatids; it formed a semicircle surrounding the tip of spermatid heads (yellow arrows), and many of these spermatids had lost their polarity (white arrows; see top panel). The middle panel is the enlarged images of nectin-3 from another testis that supports findings shown in the top panel. The expression of β1-integrin (bottom panel) was also down-regulated and considerably diminished, in particular in step 19 spermatids that had lost the polarity (annotated by white arrows). Insets are enlarged images of boxed/colored areas. Scale bar, 50 μm, which applies to other micrograph; scale bar in inset and also middle panel, 20 μm. B, Image analysis that illustrates a loss of fluorescence intensity in nectin-3 and β1-integrin after the ezrin knockdown with three rats. Fluorescence in spermatids of control rats was arbitrarily set at 1. *, P < .01.

Discussion

Ezrin is an actin-binding protein that supports the organization of actin microfilaments at cell cortex such as cell adhesion complexes at the TJ and AJ as well as microvilli of apical membrane domain in mammalian epithelia (9, 10, 48, 49), most notably in the small intestine (14). Thus, although neonatal ezrin^−/− mice were normal at birth, they failed to survive past weaning at approximately 21 days postpartum due to defects in the morphogenesis of intestinal villi and the organization of apical membrane proteins in the small intestine (14). As such, its role in spermatogenesis remains unknown. In the testis, ES is an F-actin-rich, testis-specific AJ. The integrity of the ES relies on the actin filament bundles that lie between Sertoli cell plasma membrane and the cisternae of endoplasmic reticulum, which undergo intermittent reorganization via changes between their bundled and unbundled/branched configuration to facilitate preleptotene spermatocyte and spermatid transport, as well as spermiation at stage VIII of the epithelial cycle (3, 5). Thus, ezrin, being highly expressed at the apical and basal ES in stage VIII tubules, is important to the organization of microfilaments at the ES. This reasoning is supported by the findings herein that a knockdown of ezrin in vitro indeed impeded the TJ-barrier via changes in actin filament organization in Sertoli cells. Furthermore, ezrin knockdown in vivo also disrupted the Sertoli cell BTB integrity, validating data observed in vitro. Also, its knockdown in vivo disrupted spermatid transport at stage VIII of the cycle, leading to defects in spermiation in which step 19 spermatids were trapped inside into the epithelium in stage VIII–IX tubules. Its knockdown also perturbed the transport of residual bodies transformed into phagosomes at stage IX of the cycle because phagocytosis and the transport of phagosomes are actin-dependent cellular events (41).

It is of interest to note that ezrin is expressed at the basal ES/BTB in virtually all stages of the epithelial cycle except that it is considerably diminished at stage IX. Ezrin, however, is restrictively expressed at the apical ES between Sertoli cells and step 19 spermatids at stage VIII but not in other stages. These findings thus illustrate that although basal and apical ES share common structural features, their functions are differentially regulated, such as by ezrin via its stage-specific expression. Also, the up-regulation and the spatiotemporal expression of ezrin at the apical ES in stage VIII tubules illustrate that the transport of spermatids in the adluminal compartment to prepare for spermiation at this stage is different from other stages, and ezrin is tightly involved in the final step of spermatid transport and spermiation. Our findings using the in vivo knockdown model indeed support this possibility. First, a knockdown of ezrin in the testis in vivo impeded step 19 spermatid transport and spermatid polarity in which spermatids remained deep inside the epithelium with defects in polarity, illustrating the apical ES function had been disrupted. Second, the transport of phagocytosed residual bodies, namely phagosomes, that requires functional polarized actin microfilaments was also disrupted. For instance, these organelles should have been located near the basal compartment in stage IX tubules (41, 50), yet they were still found near the tubule lumen, illustrating the organization of F-actin microfilaments had been disrupted. A disruption of F-actin filament bundles at the apical ES was supported by findings in which the expression of actin bundling protein palladin was considerably down-regulated. Furthermore, a disruption of actin filament bundles at the basal ES had occurred in ezrin knockdown testis is supported by findings in which basal ES protein N-cadherin failed to localize tightly to the BTB, but diffusing away from the site.

Our findings are also consistent with an earlier report in which ezrin was shown to localize to the residual bodies and phagosomes in the mouse testis at stage IX as well as apical ES at stage VIII by IHC (20). Herein we further expand findings of this earlier report, illustrating ezrin is critically involved in actin-based cytoskeletal function in Sertoli cells and also involved in phagosome transport. This finding is not entirely unexpected because phagocytosis of foreign substances including bacteria, cellular debris, and residual bodies by macrophages and Sertoli cells are actin-based cytoskeleton-dependent events (41, 42). In this context, it is of interest to note that ezrin is also expressed by peritubular myoid cells in the tunica propria, and myoid cell layer is known to confer seminiferous epithelial barrier function in rodents (51, 52). Thus, the reported effects on the Sertoli cell barrier function after ezrin knockdown in vivo might be contributed by a loss of myoid cell function. This possibility must be carefully evaluated in future studies.

Consistent with an earlier report that ezrin is an integrated component of TNT in MSTO-211H cells (a human lung fibroblast-like cancer cell line) (25, 53), ezrin was shown to colocalize with F-actin to construct the TNT in Sertoli cells. A knockdown of ezrin in vitro was found to disrupt the organization of TNT in Sertoli cells in which F-actin failed to assemble intact TNT. These findings suggest that ezrin that is highly expressed at stage VIII at the apical ES can be used to organize TNT to coordinate cross talk between (and among) step 19 spermatids and the Sertoli cell so that the release of sperms at spermiation can be tightly coordinated and synchronized. Although gap junctions are the ultrastructure known to communicate signals between cells, gap junctions, unlike those at the BTB, which coexist with basal ES (8), are not found at the Sertoli-spermatid (8–19 spermatids) interface at the apical ES. Obviously hemichannel gap junctions that are capable of providing intercellular communications between Sertoli cell and spermatids can still be present at the apical ES due to the abundant presence of connexins (such as connexin 43 and connexin 33) at the apical ES (54–56). Nonetheless, the pore size of gap junction is limited to small molecules (54, 55) to transfer small chemical signals between cells.

It is known that small regulatory RNAs, such as microRNA and Piwi-interacting RNA, are found in rodent testes that are abundant in spermatocytes and spermatids (57–59), and they can likely be used to regulate Sertoli cell function in response to changes of the epithelial cycle, which can be transferred via gap junctions and TNT. In fact, small RNAs transported between mammalian cells via gap junctions (60–63) are known to regulate cellular events. Studies have also shown that TNT can transfer large molecules between cells, such as H-Ras from B-to-T lymphocytes (64), and polyglutamine aggregates that cause Huntington's disease between neuronal cells (65). Thus, TNT at the ES may coordinate Sertoli cell function across the seminiferous epithelium during the epithelial cycle. We thus speculate that siRNA duplexes administered to the testis in vivo reach the intratubular spermatids via TNT (or hemichannel gap junctions). This possibility must be carefully evaluated in future studies.

In summary, ezrin is involved in the organization of actin microfilaments at the ES such as the basal ES at the BTB; however, its function at the apical ES is limited to stage VIII of the epithelial cycle, tightly involved in the final step of spermatid transport, polarity, and spermiation as well as the transport of residual bodies/phagosomes.