P-glycoprotein regulates blood–testis barrier dynamics via its effects on the occludin/zonula occludens 1 (ZO-1) protein complex mediated by focal adhesion kinase (FAK)

Abstract

The blood–testis barrier (BTB), one of the tightest blood–tissue barriers in the mammalian body, creates an immune-privileged site for postmeiotic spermatid development to avoid the production of antibodies against spermatid-specific antigens, many of which express transiently during spermiogenesis and spermiation. However, the BTB undergoes extensive restructuring at stage VIII of the epithelial cycle to facilitate the transit of preleptotene spermatocytes and to prepare for meiosis. This action thus prompted us to investigate whether this stage can be a physiological window for the delivery of therapeutic and/or contraceptive drugs across the BTB to exert their effects at the immune-privileged site. Herein, we report findings that P-glycoprotein, an ATP-dependent efflux drug transporter and an integrated component of the occludin/zonula occludens 1 (ZO-1) adhesion complex at the BTB, structurally interacted with focal adhesion kinase (FAK), creating the occludin/ZO-1/FAK/P-glycoprotein regulatory complex. Interestingly, a knockdown of P-glycoprotein by RNAi was found to impede Sertoli cell BTB function, making the tight junction (TJ) barrier “leaky.” This effect was mediated by changes in the protein phosphorylation status of occludin via the action of FAK, thereby affecting the endocytic vesicle-mediated protein trafficking events that destabilized the TJ barrier. However, the silencing of P-glycoprotein, although capable of impeding drug transport across the BTB and TJ permeability barrier function, was not able to induce the BTB to be “freely” permeable to adjudin. These findings indicate that P-glycoprotein is involved in BTB restructuring during spermatogenesis but that P-glycoprotein–mediated restructuring does not “open up” the BTB to make it freely permeable to drugs.

In the mammalian testis, the blood–testis barrier (BTB) is a unique ultrastructure that physically divides the seminiferous epithelium into the adluminal (apical) and the basal compartments (1). As such, meiosis I and II, spermiogenesis, and spermiation all take place in an immune-privileged site, namely the apical compartment, which is segregated from the host immune system (1, 2). This separation thus avoids, at least in part, the production of antibodies against many germ cell-specific antigens that express transiently during spermiogenesis and spermiation. Although the BTB is one of the tightest blood–tissue barriers (1), it is a dynamic ultrastructure, undergoing extensive “restructuring” to accommodate the transit of preleptotene spermatocytes at the site at stage VIII of the epithelial cycle, so that zygotene spermatocytes differentiate to diplotene spermatocytes, and to prepare for meiosis I and meiosis II, in the apical compartment at stage XIV of the epithelial cycle (3, 4), without compromising the immunological barrier function (1). Although recent studies have demonstrated that the coexisting junctions—such as basal ectoplasmic specialization (basal ES; a testis-specific adherens junction type), gap junction, and tight junction (TJ), which are present side by side with desmosome to constitute the BTB—are working in concert to confer and regulate the BTB integrity (4, 5), it is not known whether this stage-specific restructuring of the BTB would also facilitate drug entry into the apical compartment. This information, if known, is critical for contraceptive development because many nonhormonal contraceptives presently under investigation that exert their effects in the apical compartment have poor bioavailability. For instance, adjudin [1-(2,4-dichlorobenzyl)-1H-indazole-3-carbohydrazide] (6) is a potential male contraceptive that exerts its effects primarily at the apical ES to disrupt spermatid adhesion in the apical compartment, causing reversible male infertility. However, less than 0.5% of adjudin administered by gavage in adult rats could reach the apical compartment because of the barrier function imposed by the BTB (6, 7).

Studies have shown that drug transport across a blood–tissue barrier is regulated by either ATP-dependent drug transporters known as ATP-binding cassette (ABC) transporters (e.g., P-glycoprotein, an efflux pump) or solute carrier (SLC) transporters, which do not require ATP consumption [e.g., organic anion transporting polypeptide 3 (Oatp3), an influx pump] (8, 9). At the Sertoli cell BTB, P-glycoprotein is structurally associated with occludin, claudin-11, and junctional adhesion molecule A (JAM-A) (10), whereas Oatp3 is an integrated component of the N-cadherin/β-catenin protein complex (a basal ES protein complex) (11). The transport of adjudin at the BTB was found to be regulated by the intricate actions of influx and efflux pumps. For instance, a knockdown of Oatp3 either alone or in combination with other solute carrier transporters, Na⁺-dependent organic cation/carnitine transporter 2 (OCTN2) and gonad-specific transporters 1 and 2 (GST-1 and -2), failed to perturb the TJ permeability barrier function at the BTB (11). Herein, we sought to examine whether a knockdown of an ATP-dependent efflux pump (e.g., P-glycoprotein) would impede TJ barrier function and the bioavailability of drugs in the apical compartment.

Results

Changes on the Steady-State Level of P-Glycoprotein and Its Cellular Localization at the Sertoli Cell BTB in Vitro in Response to Drug Exposure.

Sertoli cells cultured in vitro in ∼2–3 d are known to establish a functional TJ permeability barrier with ultrastructures of TJ, basal ES, gap junction, and desmosome that mimic the BTB in vivo, and this system has been widely used by investigators in the field to study Sertoli cell BTB function (12). This system was used in the present study as shown in Fig. 1 A–H, in which Sertoli cells cultured for 3 d [residual germ cells were removed by a hypotonic treatment (13) at ∼48 h after Sertoli cell plating and Sertoli cell purity was >98%] were treated with adjudin to assess changes in P-glycoprotein, an efflux pump, in two different regimens (Fig. 1 A and E). When the Sertoli cell epithelium was treated with a single dose of adjudin (Fig. 1A), P-glycoprotein level was found to be transiently induced (Fig. 1 B and C), and this observation was confirmed by confocal microscopy (Fig. 1D). However, when Sertoli cell epithelium was treated with multiple doses of adjudin, the P-glycoprotein level was significantly and persistently induced (Fig. 1 E–G), also consistent with the cellular localization (Fig. 1 H–J) and image analysis results (Fig. S1). These results thus demonstrate that adjudin induces the overexpression of P-glycoprotein and affects its cellular distribution at the Sertoli cell BTB.

Fig. 1.

The steady-state protein level of P-glycoprotein and its cellular localization at the Sertoli–Sertoli cell interface is up-regulated by adjudin in vitro. Sertoli cells were cultured on Matrigel-coated dishes or polyester membrane-based culture inserts as described in SI Materials and Methods. On day 3 (designated as time 0), cells were treated with a single dose (A–D) or multiple doses (E–H) of adjudin, and cells at different time points were harvested for lysate preparation or confocal microscopy. Controls were Sertoli cells cultured with vehicle (ethanol) alone. (B and F) Immunoblots assessing changes in the steady-state protein level of P-glycoprotein level after treatment of Sertoli cells with adjudin (1 μg/mL). Actin served as a protein loading control. (C and G) Histograms summarizing results shown in B and F. Each data point is the mean ± SD of three independent experiments and is normalized against actin, with the control at time 0 h arbitrarily set at 1. *P < 0.05; **P < 0.01 (ANOVA followed by Dunnett's test). (D and H) (Left and Center) Horizontal views of the Sertoli cell epithelium from control (at time 0) vs. adjudin-treated cells (at 1 d or 3 d) immunostained for P-glycoprotein. Each column shows an identical optical slice from the x–y plane [i.e., parallel to the plane of cell attachment (I)]. (Right) Vertical views of the Sertoli cell epithelium, which are reconstructed optical slices from the x–z plane [i.e., perpendicular to the plane of cell attachment (J)] corresponding to the sliced positions on the x–y plane marked by white dotted lines and white brackets (Middle). White brackets in Middle correspond to black brackets in Bottom. Adjudin treatment was found to induce a thickening of P-glycoprotein at the Sertoli–Sertoli cell interface (see white and black brackets) (Fig. S1). These findings are consistent with the corresponding immunoblotting data shown in B and F. These findings are representative results of ∼200 cells in each experimental group from three independent experiments, and the analyzed results are summarized in Fig. S1. Sertoli cell nuclei were visualized with DAPI (blue). (Scale bars: 12 μm.) H, hour; D, day.

Knockdown of P-Glycoprotein in Sertoli Cells by RNAi Disrupts the TJ Barrier and Distribution of Proteins at the BTB.

To investigate whether P-glycoprotein plays a role in maintaining BTB integrity, P-glycoprotein in Sertoli cells was silenced by RNAi in Sertoli cells with siRNA duplexes specific to Abcb1a (mdr1a, also known as mdr3) and Abcb1b (mdr1b) because P-glycoprotein is encoded by both genes in the rat (9). Abcb1a is the predominant gene expressed mostly by Sertoli cells and late spermatids, to a lesser extent in Leydig cells and peritubular myoid cells, but not at all in spermatogonia, spermatocytes, or early spermatids (10, 14). Sertoli cells were cultured for 3 d, and the cell epithelium was transfected with either nontargeting control or (Abcb1a + Abcb1b)-specific siRNA duplexes for 24 h. Thereafter, the transfection mixture was removed, and cells were washed twice, cultured for an additional 24 h, and terminated for immunoblot and immunofluorescence analysis (Fig. 2). When ∼70% of P-glycoprotein was silenced by RNAi, a concomitant decline of 20% was noted for claudin-11 (Fig. 2 B and C), but no decline was present for other BTB-associated proteins that were examined, including Oatp3 (a drug influx transporter), TJ proteins [e.g., occludin, tricellulin, claudin-5, JAM-A, coxsackievirus and adenovirus receptor (CAR), zonula occludens 1 (ZO-1)], TJ regulators [e.g., focal adhesion kinase (FAK), c-Src], and basal ES proteins (e.g., N-cadherin, α-catenin, β-catenin) (Fig. 2B), demonstrating that there was no off-target effect. A knockdown of P-glycoprotein in Sertoli cells was found to induce a disruption of the TJ barrier function vs. cells transfected with control siRNA duplexes (Fig. 2D). In Sertoli cells cotransfected with nontargeting control siRNA duplexes and siGLO Red transfection indicator, P-glycoprotein, occludin, CAR, ZO-1, N-cadherin, and β-catenin were localized almost exclusively to the Sertoli–Sertoli cell interface (Fig. 2 E and F, Left, green), except that FAK was found at both the cell–cell interface and in cytosol (Fig. 2E vii, green). After its knockdown, P-glycoprotein almost vanished at the cell–cell interface except for some staining in the cell cytosol (Fig. 2E ii, green). Other TJ proteins displayed broad disruptive patterns by redistributing from the cell–cell interface to the cell cytosol (Fig. 2E, Right, green); however, basal ES proteins were not affected (Fig. 2F, Right, green). These findings demonstrate a regulatory role of P-glycoprotein by affecting distribution of TJ, but not basal ES, proteins at the BTB.

Fig. 2.

A study to assess the effects of P-glycoprotein knockdown by RNAi on Sertoli cell BTB function in vitro. (A) The regimen used in this study. (B) Representative immunoblots illustrating an ∼70% knockdown of P-glycoprotein in Sertoli cells cultured at 0.5 × 10⁶ cells per cm² after transfection with nontargeting (control) vs. (Abcb1a + Abcb1b)-specific siRNA duplexes (SI Materials and Methods). In addition to P-glycoprotein, only the level of claudin-11 was reduced (by 20%) after its knockdown by RNAi on day 2 (RNAi-2D), but no decline was detected for Oatp3 (an influx pump), TJ, basal ES, or TJ regulatory proteins, indicating there was no off-target effect. (C) Densitometric analyses of P-glycoprotein and claudin-11 immunoblotting data normalized against actin with the control arbitrarily set at 1. Each bar is the mean ± SD of three experiments. (D) The knockdown of P-glycoprotein was found to transiently perturb the Sertoli cell TJ permeability barrier in vitro. (E and F) To investigate changes in protein distribution at the cell–cell interface, Sertoli cells cultured at 0.05 × 10⁶ cells per cm² on Matrigel-coated coverslips were cotransfected with siGLO Red (a transfection indicator) with either nontargeting or (Abcb1a + Abcb1b) siRNA duplexes and stained for P-glycoprotein, occludin, CAR, FAK, ZO-1, N-cadherin, and β-catenin (green). Although P-glycoprotein RNAi had no apparent effects on the steady-state levels of BTB proteins, its knockdown was found to induce mislocalization of occludin, CAR, FAK, and ZO-1, but not N-cadherin or β-catenin, at the cell–cell interface. (Scale bars: 20 μm.) *P < 0.05; **P < 0.01.

Knockdown of P-Glycoprotein by RNAi Accelerates the Influx of [³H]Adjudin from the Basal to the Apical Compartment Across the Sertoli Cell Epithelium.

On day 3, Sertoli cells cultured in vitro were transfected with nontargeting (control) or (Abcb1a + Abcb1b)-specific siRNA duplexes for 24 h for RNAi (Fig. 2A). At 2 d after transfection, when P-glycoprotein was silenced by ∼70% (Fig. 2 B and C), [³H]adjudin was added to either the apical (A) or the basal (B) compartment of the bicameral units (and designated as time 0) to assess the A-to-B or B-to-A flux to evaluate the apparent permeability (P_app) (Fig. 3 and Fig. S2), efflux ratio (Fig. S3A), and corrected flux ratio (Fig. S3B). It was noted that the knockdown of P-glycoprotein by ∼70% increased the flux of [³H]adjudin significantly from B-to-A but not from A-to-B nor in controls (Fig. 3). This observation was consistent with an increase in both the efflux ratio and corrected flux ratio after its RNAi (Fig. S3 A and B). However, [³H]adjudin was not “freely” permeable at the BTB after the knockdown, possibly because of other drug efflux transporters at the BTB (9).

Fig. 3.

Assessment of the effects of P-glycoprotein knockdown on the apparent permeability (P_app) of [³H]adjudin across the Sertoli cell epithelium. Sertoli cells were cultured on Matrigel-coated bicameral units at 1.2 × 10⁶ cells per cm² for 3 d to establish an intact epithelium with a functional TJ barrier. Thereafter, cells were transfected with (Abcb1a + Abcb1b)-specific vs. nontargeting siRNA duplexes for 24 h. At 2 d after transfection, when P-glycoprotein was knocked down by ∼70% (Fig. 2 A and B), [³H]adjudin (∼0.4 × 10⁶ cpm) was added to the apical (A) or basal (B) compartment of the bicameral units, and this time was designated as time 0 to assess the A-to-B or B-to-A flux. At selected time points thereafter, 50 μL of F12/DMEM was withdrawn from both compartments of the bicameral units in both A-to-B and B-to-A groups for radioactivity determination, and these data were plotted in Fig. S2 and used to compute P_app as shown herein. Each data point is the mean ± SD of triplicate determinations, and this experiment was repeated three times. *P < 0.05. Refer to Fig. S3 for additional directional flux data.

P-Glycoprotein Knockdown Alters Protein–Protein Interactions at the BTB.

Both P-glycoprotein and FAK are localized to the Sertoli cell BTB (10, 15), and FAK is an integrated component of the occludin/ZO-1 complex that regulates cell adhesion by determining the phosphorylation status of occludin at the BTB (15). We speculated that P-glycoprotein might structurally interact with FAK to form a regulatory protein complex. To test this hypothesis, coimmunoprecipitation (Co-IP) was performed with lysates from Sertoli cells cultured for 3 d with an established functional TJ barrier. Indeed, P-glycoprotein was found to structurally interact with FAK (Fig. 4A). More importantly, after the knockdown of P-glycoprotein by RNAi, occludin virtually lost its association with ZO-1 but had an increase in its interaction with FAK (Fig. 4 B and C); P-glycoprotein was also found to have a reduced association with occludin but an increase in its interaction with FAK (Fig. 4 B and D). However, the interaction between two basal ES proteins N-cadherin and β-catenin did not change, and P-glycoprotein and CAR did not show any interactions (Fig. 4 B–D). Because P-glycoprotein was found to structurally interact with FAK and occludin, and it is known that the phosphorylation status of occludin determines whether it is assembled into the TJ fibrils at blood–tissue barriers (16), we next examined whether P-glycoprotein knockdown would affect the phosphorylation status of occludin. We used an anti-occludin antibody as the precipitating antibody to extract occludin from cell lysates from nontargeting control vs. P-glycoprotein knockdown groups, and occludins recovered from the immunocomplexes were subjected to SDS/PAGE followed by immunoblotting with corresponding anti–phospho-Ser, -Thr, or -Tyr antibodies (Table S1) to assess changes in the phosphorylation status (Fig. 4 E and F). These findings indicate a decrease and an increase in occludin phosphorylation at Ser and Thr residues, respectively, but no change at Tyr (Fig. 4 E and F).

Fig. 4.

Co-IP assessment of the effects of P-glycoprotein knockdown on protein–protein interactions and the phosphorylation status of occludin at the Sertoli cell BTB. On day 3, Sertoli cells cultured at 0.5 × 10⁶ cells per cm² on Matrigel-coated dishes were transfected with (Abcb1a + Abcb1b)-specific vs. nontargeting siRNA duplexes for 24 h and were harvested 24 h thereafter [i.e., at 2 d after transfection for RNAi (RNAi-2D); Fig. 2A]. (A) Co-IP was performed with normal Sertoli cell lysate (∼300 μg of protein), and P-glycoprotein (P-gp) was found to structurally interact with FAK. Sertoli cell lysate (20 μg of protein, no Co-IP) and Co-IP performed with normal rabbit (Rb) or mouse (Ms) IgG served as positive and negative controls, respectively. (B) Sertoli cell lysates (∼300 μg of protein) were subjected to Co-IP with anti–ZO-1, anti-occludin, anti–P-glycoprotein, or anti–N-cadherin antibodies, and blots were probed with anti-occludin, anti-FAK, anti–P-glycoprotein, anti-CAR, or anti–β-catenin antibodies. Normal Sertoli cell lysates (50 μg of protein) without Co-IP served as positive controls. (C and D) Densitometric analyses of Co-IP data illustrating protein–protein interactions in which cells transfected with nontargeting siRNA duplexes (control) were arbitrarily set at 1, against which comparison was performed. (E) Sertoli cell lysates (∼300 μg of protein) were first subjected to Co-IP with an anti-occludin antibody to immunoprecipitate occludin, and the levels of occludin in the samples between the nontargeting (control) and the P-glycoprotein RNAi groups were similar (bottom blot), consistent with data shown in Fig. 2B. This finding also confirmed equal protein loading. These blots were then probed with anti–phospho-Ser, -Thr, or -Tyr antibodies to assess changes in phosphorylated content of occludin. (F) Densitometric analyses of Co-IP data shown in E normalized against occludin level with controls arbitrarily set at 1. Each bar is the mean ± SD of three to four experiments. *P < 0.05; **P < 0.01.

P-Glycoprotein Knockdown Accelerates the Kinetics of Protein Endocytosis at the BTB.

In Sertoli cell epithelium with an established functional TJ barrier, the knockdown of P-glycoprotein was found to significantly enhance the kinetics of internalization of TJ (e.g., occludin, CAR), but not basal ES (e.g., N-cadherin), integral membrane proteins (Fig. 5). These results thus demonstrate that P-glycoprotein knockdown induces TJ protein endocytosis, which, in turn, impedes barrier integrity.

Fig. 5.

A knockdown of P-glycoprotein by RNAi enhanced the kinetics of endocytosis of TJ (occludin and CAR) but not basal ES (N-cadherin) integral membrane proteins at the BTB. (A) On day 3, Sertoli cells cultured at 0.5 × 10⁶ cells per cm² on Matrigel-coated dishes were transfected with (Abcb1a + Abcb1b)-specific vs. nontargeting siRNA duplexes for 24 h. At 2 d after transfection, Sertoli cells were biotinylated at 4 °C. Thereafter, cells were incubated at 35 °C for various time points (0, 5, 15, 30, 45, and 60 min) to allow endocytosis and terminated to obtain lysates for immunoblotting. “Total” represents the amount of total labeled cell-surface proteins (after the 30-min biotinylation) without stripping. “-ve” stands for cell-surface proteins without biotinylation and serves as the negative control. Biotinylated proteins were recovered with UltraLink-immobilized NeutrAvidin Plus beads with ∼300 μg of protein from samples at each time point and were subjected to immunoblotting with the corresponding antibody (Table S1). Actin served as a protein loading control. (B–D) To assess the kinetics of endocytic vesicle-mediated endocytosis of occludin, CAR, and N-cadherin after P-glycoprotein knockdown vs. nontargeting controls, data were plotted by using the percentage of internalized proteins compared with total biotinylated proteins on the y axis against their changes over time on the x axis. Each bar is the mean ± SD of three experiments. *P < 0.05; **P < 0.01.

P-Glycoprotein Knockdown in Sertoli Cell Epithelium Modulates Endocytic Vesicle-Mediated Protein Trafficking.

The steady-state levels of early endosome antigen 1 (EEA-1, an endosome marker), caveolin-1 (a transcytosis marker), and ubiquitin-conjugating enzyme E2J1 (Ube2j1, an intracellular protein degradation marker) did not alter significantly after P-glycoprotein knockdown by ∼70% in Sertoli cells (Fig. 6 A and B). Dual-labeled immunofluorescence analysis was used to investigate the localization of N-cadherin/occludin (red), EEA-1/caveolin-1/Ube2j1 (green), and their colocalization in Sertoli cells after P-glycoprotein knockdown (Fig. 6 C and D). Increases in association between occludin and EEA-1 as well as Ube2j1 (Fig. 6D, white arrowheads indicating merged signals appearing in yellow-orange) were detected because of an increase in internalization of occludin induced by P-glycoprotein RNAi. However, there were no changes in the association between N-cadherin and EEA-1/caveolin-1/Ube2j1 (Fig. 6 C vs. D). These findings establish that P-glycoprotein knockdown in Sertoli cells induces EEA-1– and Ube2j1-mediated TJ protein (e.g., occludin) internalization and degradation, respectively, destabilizing BTB integrity.

Fig. 6.

Assessment of the effects of P-glycoprotein knockdown on the endocytic vesicle-mediated protein trafficking in Sertoli cell epithelium. (A and B) Sertoli cells cultured for 3 d were transfected with the Abcb1a + Abcb1b siRNA vs. the nontargeting siRNA duplexes (control) for 24 h. Cells were rinsed and cultured with fresh F12/DMEM for an additional 24 h and terminated thereafter and used for immunoblotting [i.e., at 2 d after transfection for RNAi (RNAi-2D)]. The knockdown of P-glycoprotein by ∼70% (Fig. 2B) had no apparent effects on the steady-state levels of EEA-1 (an endosome marker), caveolin-1 (a transcytosis marker), and Ube2j1 (an intracellular protein degradation marker) in Sertoli cells. (C and D) Cellular localization of these markers with N-cadherin (a basal ES protein) and occludin (a TJ protein) at the Sertoli cell BTB was examined by dual-labeled immunofluorescence analysis to assess changes in the distribution of N-cadherin/occludin (red) and EEA-1/caveolin-1/Ube2j1 (green) after P-glycoprotein knockdown. White arrowheads denote an increase in the colocalization of occludin with EEA-1 and Ube2j1, but not with caveolin-1, after P-glycoprotein silencing. Also, there were no changes in the colocalization of N-cadherin with any endocytic vesicle-mediated protein trafficking marker. DAPI (blue) was used to visualize nuclei. (Scale bars: 30 μm.)

Discussion

P-Glycoprotein Is an Integrated Component of the Occludin/ZO-1/FAK Protein Complex That Regulates Cell Adhesion and TJ Barrier Function at the BTB.

FAK is an integrated component of the occludin/ZO-1 complex in the testis (17). FAK exerts its effects on cell adhesion at the BTB by conferring proper phosphorylation to occludin (15) so that occludin can be assembled to the TJ fibrils (16), regulating paracellular transport at the Sertoli–Sertoli cell interfaces that constitute the BTB in the mammalian testis (1). Although P-glycoprotein was shown to be structurally associated with occludin at the BTB (10) and localized to the same sites as occludin, claudins, JAMs, and ZO-1 at blood–tissue barriers such as the blood–brain barrier (BBB) (18) and the BTB (10), P-glycoprotein is a known regulator of transcellular transport (e.g., drugs that are substrates of P-glycoprotein) via a receptor-mediated and ATP-dependent mechanism at blood–tissue barriers such as the intestinal barrier and the BBB (8, 19, 20). It is not known, however, whether P-glycoprotein would affect paracellular transport at the BTB. Herein, P-glycoprotein was shown to structurally interact with FAK and occludin. More importantly, a knockdown of P-glycoprotein by ∼70% with RNAi was found to significantly enhance occludin–FAK interaction. This increase in FAK association with occludin was accompanied by a significant but concomitant decline and increase in phosphorylation of Ser and Thr residues in occludin, respectively, which, in turn, destabilized the occludin/ZO-1 adhesion complex that led to a loss of occludin/ZO-1 protein–protein interaction. In short, the net result of an increase in FAK and occludin association after P-glycoprotein RNAi alters the proper phosphorylation status of occludins at the BTB is a reduction in Sertoli–Sertoli cell adhesion at the BTB, causing a transient disruption of the Sertoli cell TJ permeability barrier. This postulate was further supported by findings using dual-labeled immunofluorescence analysis that demonstrated that a knockdown of P-glycoprotein by RNAi led to a redistribution of occludin, CAR, and ZO-1 but not N-cadherin and β-catenin at the cell–cell interface, with these TJ proteins moving from the cell surface into the cell cytosol. This redistribution of proteins thus destabilizes cell adhesion at the Sertoli cell BTB, causing a transient disruption of the TJ permeability barrier. These findings are also supported by earlier studies with mdr1a^−/− (or Abcb1a^−/−) mice in which the knockout of P-glycoprotein led to a dysfunction of blood–inner ear barrier (21). In this context, it is of interest to note that mdr1a^−/− mice are fertile and viable, and they do not display obvious phenotypic abnormalities other than they are hypersensitive to drugs and the BBBs of these mice were disrupted (22, 23). Also, mdr1a^−/−/mdr1b^−/− (double-KO) mice displayed a significant increase in amyloid-β deposition in an Alzheimer's disease mouse model as a result of the BBB disruption (24). However, it is not known whether the BTBs of these mdr1a^−/− or mdr1a^−/−/mdr1b^−/− mice were compromised. Based on the findings reported herein, although the TJ permeability barrier at the BTB after P-glycoprotein knockdown was transiently disrupted as a result of TJ protein redistribution at the BTB, the basal ES proteins N-cadherin and β-catenin, which are capable of conferring the immunological barrier, were unaffected (25). This latter finding explains the maintenance of fertility in mdr1^−/− mice (22, 23), and it is also consistent with a recent study showing that an intact BTB is necessary for spermatogonial stem cell differentiation beyond type A spermatogonia to initiate spermatogenesis (26). Additionally, other efflux drug transporters that are also present in Sertoli cells (9) (SI Discussion) can likely supersede the lost function of P-glycoprotein at the BTB in the mdr1^−/− mice. Nonetheless, the findings establish that P-glycoprotein is crucial to BTB dynamics, possibly via its ability to recruit FAK to the occludin/ZO-1 protein complex to modulate the phosphorylation status and adhesive function of occludin during the epithelial cycle of spermatogenesis.

Changes in Protein–Protein Interactions Among Different Components of the Occludin/ZO-1/FAK/P-Glycoprotein Complex and the Phosphorylation Status of Occludin After P-Glycoprotein Knockdown by RNAi.

FAK was shown to structurally interact with occludin at the BTB in the rat testis (17), and it was also found to modulate the occludin-based cell adhesion protein complex at the BTB (15). Because P-glycoprotein was found to structurally interact with FAK and occludin (but not CAR) as shown herein, we speculate that the knockdown of P-glycoprotein that impeded the Sertoli cell TJ barrier function might be related to changes in protein adhesion at the BTB caused by changes in phosphorylation status of integral membrane proteins (e.g., occludin) at the BTB. Data presented in Fig. 4 support this postulate. For instance, there was a tighter association between occludin and FAK (but a reduced association between occludin and ZO-1) after P-glycoprotein knockdown (Fig. 4 B and C), which likely induced changes in the phosphorylation status of occludin (Fig. 4F); this, in turn, induced a loss of occludin/ZO-1 association, leading to a loss of Sertoli cell adhesion and impeded the Sertoli cell TJ barrier function as noted in Fig. 2D. However, we note that, because we only achieved an ∼70% knockdown of P-glycoprotein by RNAi with siRNA duplexes, changes in protein–protein interactions, such as an increase in P-glycoprotein–FAK, are likely underestimates by at least 30%. Nonetheless, these observations need to be expanded to include the use of shRNA and/or microRNA to obtain better knockdown efficacy, to include other efflux drug transporters using double or triple knockdowns as described (11), and to assess changes in phosphorylation status of integral membrane proteins at the BTB (e.g., JAM-A or nectin-2 in addition to occludin) and protein–protein interactions. In short, these findings support the notion that the entry of adjudin into the adluminal compartment is not mediated by paracellular transport at the Sertoli cell BTB but by transcellular transport with drug transporters.

Knockdown of P-Glycoprotein by RNAi Impedes Sertoli Cell TJ Barrier Function and Basal-to-Apical Directional Flux of Drug Transport at the BTB.

The knockdown of P-glycoprotein in the Sertoli cell epithelium was found to render the BTB significantly permeable to adjudin, an experimental nonhormonal male contraceptive (6), and this knockdown also impeded the TJ permeability barrier function; however, the disrupted BTB was not freely permeable to [³H]adjudin. These findings thus suggest that a disruption of the Sertoli cell TJ barrier that impedes the paracellular transport does not necessarily lead to an influx of [³H]adjudin across the BTB at the cell–cell interface (SI Discussion). Instead, [³H]adjudin enters the apical compartment via transcellular transport, such as by using influx pump transporters after the knockdown of P-glycoprotein (an efflux pump transporter). However, the striking observation in this study is that a knockdown of P-glycoprotein impedes Sertoli cell TJ barrier function via changes in distribution of TJ, but not basal ES, integral membrane proteins at the BTB, which are likely the result of an increase in protein endocytosis. Collectively, these findings demonstrate that P-glycoprotein may take part in BTB restructuring during the epithelial cycle in addition to serving as an active drug pump. In this context, it is of interest to note that the combined knockdown of four other Sertoli cell ATP-independent influx pump transporters {Oatp3 (Slco1a5), Slc22a5 (OCTN2), Slco6b1 [testis-specific transporter 1 (TST-1), and Slco6c1 (TST-2)]} by ∼70–90% failed to perturb the Sertoli cell TJ barrier, even though this knockdown rendered the Sertoli BTB significantly less permeable to [³H]adjudin because of the loss of the function of four influx pumps (11). These findings have demonstrated unequivocally that active ATP-dependent P-glycoprotein is a unique regulator of BTB dynamics via its effects on the FAK/occludin/ZO-1 adhesion protein complex at the BTB.

P-Glycoprotein Regulates BTB Function via Its Effects on the Endocytic Vesicle-Mediated Protein Trafficking.

Endocytic vesicle-mediated protein trafficking is known to regulate cell adhesion in multiple epithelia (1, 4). The striking observation reported herein is that the knockdown of P-glycoprotein was found to enhance the kinetics of endocytosis of integral membrane proteins of the TJ (e.g., occludin), but not the basal ES (e.g., N-cadherin), at the BTB, which is facilitated by the involvement of EEA-1 and Ube2j1, perhaps accelerating endosome/ubiquitin-mediated degradation of the internalized proteins. This finding thus establishes that an alteration of the phosphorylation status of occludin after the knockdown of P-glycoprotein led to its redistribution, which, in turn, impeded the kinetics of protein endocytosis, and perhaps enhanced its intracellular degradation, destabilizing the TJ barrier that contributes to its disruption.

In summary, P-glycoprotein is involved in regulating BTB dynamics, likely via its interaction with FAK, which modulates the phosphorylation status of the occludin/ZO-1 protein complex. This alteration, in turn, affects endocytic vesicle-mediated trafficking at the BTB.

Materials and Methods

Animals.

The use of Sprague–Dawley rats for all experiments reported herein was approved by the Rockefeller University Animal Care and Use Committee with protocol no. 09016. Animals were kept at the Rockefeller University Comparative Bioscience Center with free access to water and standard rat chow ad libitum at 22 °C and a light:dark cycle of 12h:12h.

Additional Materials and Methods.

Detailed information on antibodies, primary Sertoli cell cultures, transfection of Sertoli cells with siRNA duplexes for RNAi, administration of adjudin to Sertoli cells in vitro, functional assessment of the Sertoli cell TJ permeability barrier, immunoblot analysis and Co-IP, dual-labeled immunofluorescence analysis, confocal microscopy and image analysis, directional flux across the Sertoli cell BTB in vitro, endocytosis assay, and statistical analysis pertinent to experiments reported in this paper can be found in SI Materials and Methods.