Connexin 43 and Plakophilin-2 As a Protein Complex That Regulates Blood–Testis Barrier Dynamics

Connexin 43 and plakophilin-2 as a protein complex that regulates blood–testis barrier dynamics

Michelle W. M. Lia, Dolores D. Mruka, Will M. Leeb, and C. Yan Chenga,1

aCenter for Biomedical Research, Population Council, New York, NY 10065; and bSchool of Biological Sciences, The University of Hong Kong, Hong Kong, China

Edited by Ryuzo Yanagimachi, University of Hawaii, Honolulu, HI, and approved April 24, 2009 (received for review February 13, 2009) The blood–testis barrier (BTB) formed by adjacent Sertoli cells is located behind the TJ fibrils, followed by the desmosome, forming composed of coexisting tight junction (TJ), basal ectoplasmic special- the junction complexes (7). Interestingly, the BTB is constituted by ization (ES), and desmosome-like junction. Desmosome-like junctions coexisting TJ, basal ectoplasmic specialization (basal ES, an atyp- display structural features of desmosome and gap junctions, but its ical testis-specific AJ), and desmosome-like junction (1). We envi- function at the BTB remains unknown. Herein, we demonstrate that sioned that these coexisting junctions must have a physiological connexin 43 (Cx43), a gap junction integral membrane protein, struc- significance at the BTB other than their structural role in contrib- turally interacts with desmosomal protein plakophilin-2 (PKP2), basal uting to the barrier integrity. The desmosome-like junction in the ES proteins N-cadherin and ␤-catenin, and signaling molecule c-Src, seminiferous epithelium was reported to possess the ultrastructural but not with the TJ proteins occludin and ZO-1 in the seminiferous features of both the desmosome and gap junction (GJ) when epithelium of adult rats. The localization of Cx43 in the seminiferous examined by electron microscopy (8). There is accumulating evi- epithelium during (i) the normal epithelial cycle of spermatogenesis dence that desmosome can serve as a platform for signal transduc- and (ii) anchoring junction restructuring at the Sertoli–spermatid tion (9) and that connexin 43 (Cx43) can mediate cell–cell com- interface induced by adjudin which mimics junction restructuring munication (10). We thus sought to examine the physiological events during spermatogenesis have suggested that Cx43 is involved significance of the desmosome-like junction at the BTB. We report in cell adhesion. The knockdown of Cx43 by RNAi technique using herein a protein complex consisting of Cx43 and plakophilin-2

speciﬁc siRNA duplexes was performed in primary Sertoli cell cultures (PKP2) at the desmosome-like junction and describe how it regu- BIOLOGY

with an established TJ permeability barrier that mimicked the BTB in lates BTB dynamics through its effects on the TJ protein complex DEVELOPMENTAL vivo. This knockdown of Cx43 affected neither the TJ barrier function occludin/ZO-1. nor the steady-state levels of junction proteins of TJ, basal ES, and desmosome-like junction. However, after the knockdown of both Results Cx43 and PKP2, the Sertoli cell TJ barrier function was perturbed Stage-Specific Expression of Cx43 at the Blood–Testis Barrier in Rat transiently. This perturbation was concomitant with a mislocalization Testes. The specificity of the anti-Cx43 antibody is shown in Fig. of occludin and ZO-1 from the cell–cell interface. In summary, Cx43 1A. A single band of Ϸ39 kDa was detected by immunoblot and PKP2 form a protein complex within the desmosome-like junction analysis using lysates of seminiferous tubules, and an additional to regulate cell adhesion at the BTB, partly through its effects on the phosphorylated isoform of Ϸ41 kDa was detected using lysates occludin/ZO-1 complex, so as to facilitate the transit of primary of Sertoli cells (see also supporting information (SI) Fig S1). This preleptotene spermatocytes. antibody was also used to localize Cx43 in the seminiferous epithelium of adult rat testes (Figs. 1 B and C and Table S1). The desmosome-like junction ͉ seminiferous epithelial cycle ͉ stage-specific expression of Cx43 is shown in Fig. 1B, compared Sertoli cell ͉ spermatogenesis with the control shown in Fig. 1D. In various stages of the seminiferous epithelial cycle, Cx43 was detected near the apical n adult mammals, including rodents and humans, the blood–testis ES at the Sertoli cell–elongating spermatid interface and pre- Ibarrier (BTB) is a unique and important ultrastructure in the dominantly at the BTB (Fig. 1C). Its expression at the apical ES seminiferous epithelium for maintaining spermatogenesis. Unlike and the BTB was highest from stage V to early stage VIII, but other blood–tissue barriers, such as the blood–brain barrier, the it diminished greatly at both sites at late stage VIII (Fig. 1C). The BTB is constituted by adjacent Sertoli cells near the basement decline in Cx43 at late stage VIII coincided with spermiation, membrane in the seminiferous epithelium, instead of at the endo- which involves the disassembly of the apical ES to release the thelial tight junction (TJ) barrier of microvessels in the interstitium fully developed spermatids into the tubule lumen, and with the (1, 2). In the rat testis, the BTB confers the ‘fence’ function that restructuring at the BTB that facilitates the transit of preleptotene spermatocytes. These findings implicate the likely involve- monitors the paracellular influx of water, electrolytes, hormones, ment of Cx43 in cell adhesion at the apical ES and the BTB. and biomolecules between adjacent Sertoli cells and maintains cell polarity (3). Furthermore, the BTB creates a specialized microen- The Decline in Cx43 During Adjudin-Induced Anchoring Junction Restruc- vironment in the apical compartment of the seminiferous epithe- turing in the Seminiferous Epithelium. To study the restructuring of lium and segregates the entire event of postmeiotic germ cell anchoring junctions in the seminiferous epithelium, an estab- development from the systemic circulation (3). Although the BTB lished in vivo model was used. Adult rats were fed a single dose is considered to be one of the tightest blood–tissue barriers in of adjudin (50 mg/kg body weight) to induce germ cell loss, most mammals (4), it must undergo extensive restructuring at stages VIII and IX of the seminiferous epithelial cycle in adult rat testes. As

such, junction restructuring facilitates the transit of primary prelep- Author contributions: C.Y.C designed research; M.W.M.L. and C.Y.C. performed research; totene spermatocytes through the BTB while they differentiate into D.D.M., W.M.L., and C.Y.C. contributed new reagents/analytic tools; M.W.M.L., D.D.M., and leptotene and zygotene spermatocytes. However, the immunolog- C.Y.C. analyzed data; and M.W.M.L. and C.Y.C. wrote the paper. ical barrier must be maintained and remain intact (5). The mech- The authors declare no conﬂict of interest. anisms by which this timely BTB restructuring event is regulated This article is a PNAS Direct Submission. have remained obscure for almost half a century (3, 6). 1To whom correspondence should be addressed. E-mail: [email protected]. In other epithelia, the TJ barrier that confers the barrier function This article contains supporting information online at www.pnas.org/cgi/content/full/ is supported structurally by discrete adherens junction (AJ) plaques 0901700106/DCSupplemental.

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0901700106 PNAS ͉ June 23, 2009 ͉ vol. 106 ͉ no. 25 ͉ 10213–10218 Downloaded by guest on September 27, 2021 Time after treatment with A B A Adjudin (50 mg/kg b.w.) 95 - 72 - - Connexin 43, 39 kDa 55 - -Actin,42kDa 43 - Cx43, 34 - 39-41 kDa 1.5 26 - B

17 - 1 - Actin, 42 kDa

0.5 * ** ** C II IV V VI ** ** 0

Time after treatment with Adjudin

C a: Ctrl i b: i c: ii VII lateVII VIII lateVIII

IX X XI XII d: 11 H e: iii f: 11 H

iii

XIII XIV D Rabbit IgG g: 1 D h: iv i: v

v iv

Fig. 1. Stage-dependent expression of Cx43 in the seminiferous epithelium of j: 2 D k: 4 D l: Rabbit IgG adult rat testes. (A) Immunoblot analysis illustrates the speciﬁcity of the rabbit anti-Cx43 antibody using lysates of seminiferous tubules (ST) and Sertoli cells (SC). (B) Immunohistochemical localization of Cx43 in the seminiferous epithelium of adult rat testes. The expression of Cx43, which appears as reddish-brown precip- itates, in the epithelium is stage dependent. (C) The localization of Cx43 in the epithelium at different stages of the epithelial cycle. Cx43 is detected at the BTB Fig. 2. The expression and localization of Cx43 during adjudin-induced and the apical ES, but its expression is more predominant at the basal region. The restructuring of anchoring junctions in adult rat testes. Adult rats were expression of Cx43 at both the BTB and apical ES is highest from stage VI to early treated with adjudin, and testes were obtained for immunoblot analysis stage VIII; it diminishes drastically at both sites at late stage VIII. (D) In the negative (using lysates) (A, B) and immunohistochemistry (frozen sections) (C). The control, rabbit IgG was substituted for the primary antibody. (Scale bar, 150 ␮m steady-state protein level of Cx43 decreased after adjudin treatment (A, in B and D and 30 ␮minC.) Upper Lane), with actin serving as the protein loading control (A, Lower Lane). (B) The bar graph shows the results of 3 experiments. *, P Ͻ 0.05; **, P Ͻ 0.01. notably elongating spermatids, from the epithelium beginning at (C) Changes in the localization of Cx43 in the seminiferous epithelium were Ϸ1 day posttreatment (3). The BTB remained intact despite the studied in testes collected 11 h (H) and 1, 2, and 4 days (D) after adjudin disruption of adhesion between Sertoli and germ cells (3) when treatment. The boxed areas in a, d, and g are magniﬁed and shown in b, c, e, assessed by an in vivo integrity assay (11). A time-dependent h, and i. At 11 h and 1 day after adjudin treatment, immunoreactive Cx43 at decline in Cx43 in the testes was detected following the disrup- the basal region remained at the BTB but diminished gradually, and its tion of anchoring junctions induced by adjudin (Fig. 2 A and B). expression at the apical ES also was considerably reduced compared with controls (C d–f, and g–i versus a– c). In the negative control (Cl), normal rabbit Results from a representative immunohistochemical experiment IgG replaced the primary antibody. (Scale bar, 200 ␮mina, d, g, j, k, and l, 100 are shown in Fig. 2C. In a normal stage VII tubule, the staining ␮minf;50␮minb, c, e, h, and i.) of Cx43 was very strong near elongating spermatids at the apical ES (Fig. 2Cb). At the time of adjudin-induced depletion of elongating spermatids, Cx43 staining near these areas dimin- co-localize with both ZO-1, a TJ adaptor (Fig. 3Aa–d), and ished (compare Fig. 2Cband h). At the BTB, the intensity of N-cadherin, a basal ES integral membrane protein (Fig. 3Ae–h), Cx43 staining diminished in some damaged tubules by 11 h and at the BTB of the seminiferous epithelium. The structural 1 day posttreatment (compare Fig. 2C d–i and a–c). Cx43 interaction of Cx43 with some selected junction proteins was staining also was more diffuse and was less restricted to the BTB. detected by co-immunoprecipitation. These proteins include When the drug was metabolized by 2 to 4 days (12) and more N-cadherin and ␤-catenin from the basal ES, the GJ protein germ cells had left the epithelium, Cx43 staining considerably connexin 26 (Cx26), desmosomal protein PKP2, and c-Src, a weakened in the seminiferous epithelium (Fig. 2Cjand k). These non-receptor protein kinase that was shown to be a component findings further support the notion that Cx43 is involved in the of the BTB (Fig. 3B) (13). Even though Cx43 co-localized with germ cell adhesion in the seminiferous epithelium. ZO-1 and N-cadherin at the BTB (Fig. 3A), Cx43 was shown to interact structurally with the N-cadherin/␤-catenin protein Cx43 as an Integral Component of the Blood–Testis Barrier. Using complex of the basal ES instead of with the TJ-based occludin/ dual-labeled immunofluorescent analysis, Cx43 was shown to ZO-1 protein complex (Fig. 3B).

10214 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0901700106 Li et al. Downloaded by guest on September 27, 2021 ZO-1 N-cadherin RNAi A A a Ctrl Cx43 a e Connexin 43, GJ 39 - 41 kDa - Connexin 26, 26 kDa b - Occludin, 65 kDa Connexin 43 b f TJ - JAM-A, 36 kDa 1.5

Connexin 43 - ZO-1, 210 kDa 1 - N-cadherin, 130 kDa Basal ES - α-catenin, 102 kDa 0.5 *

c g - β-catenin, 92 kDa (Arbitrary unit) 0

Merge DJ - γ-catenin, 82 kDa Relative protein level Ctrl siRNA Cx 43 siRNA HD - β1-integrin, 140 kDa Protein kinase - c-Src p60, 60 kDa d h Cytoskeleton - Actin, 42 kDa DAPI B b a Cell surface protein level RNAi 1.5 IP antibody -ve Ctrl Cx43 Ctrl siRNA B Cx43 siRNA - Occludin, 65 kDa 1 - JAM-A, 36 kDa - N-cadherin, 130 kDa 0.5

β (Arbitrary unit)

surface protein - 1-integrin, 140 kDa Biotinylated cell IB: connexin 43 Relative protein level 0 OccludinJAM-A N-cadherin β1-integrin -++--++--+

Fig. 3. The co-localization and interaction of Cx43 with other known Fig. 4. A functional study to assess the role of Cx43 in BTB function by RNAi ϫ 6 2 constituent proteins at the BTB. (A) Partial co-localization of Cx43 (FITC) with in primary Sertoli cell cultures. Sertoli cells cultured at 0.5 10 cells/cm on either ZO-1 (CY3) (Aa–d) or N-cadherin (CY3) (Ae–h) at the BTB was detected, Matrigel-coated dishes were used on day 4 when the functional TJ perme- showing that Cx43 is a component of the BTB. Nuclei were stained with DAPI. ability barrier was formed. Cells were transfected with 50 nM of either

(Scale bar, 50 ␮m.) (B) Interactions of Cx43 with other junction proteins, non-targeting control duplexes or specific Cx43 siRNA duplexes, using Ribo- BIOLOGY including ␤-catenin, c-Src, ZO-1, occludin, and N-cadherin, which were re- Juice transfection reagent. Cells were washed 24 h later, and cultures were DEVELOPMENTAL ported to be present at the BTB, were examined by co-immunoprecipitation terminated after 48 h to obtain cell lysates. The steady-state protein levels of using lysates of seminiferous tubules from adult rats. ␤-Catenin, c-Src, N- selected markers were investigated (A). Besides a knockdown of Cx43 by Ϸ cadherin, connexin 26, and PKP2, but not ZO-1, occludin, and laminin ␥3, 50% (Ab), none of the proteins examined showed significant changes in interact with Cx43. Rabbit IgG and goat IgG served as negative controls. These their steady-state levels. The levels of selected integral membrane proteins findings are representative data from 3 independent experiments. at the cell surface were then examined to see if the knockdown of Cx43 would affect their levels (B). Immunoblots shown in (Aa) and (Ba) are representative results of 3 independent culture experiments (each in trip- Ͻ A Knockdown of Cx43 by RNAi Did Not Affect Sertoli Cell Adhesion at licate), which are summarized in (Ab) and (Bb). *, P 0.01 versus control siRNA. the Blood–Testis Barrier. The significance of Cx43 in Sertoli cell adhesion at the BTB was assessed by RNAi in primary Sertoli cell cultures having an established TJ permeability barrier that mimics respectively, was detected in primary Sertoli cell culture (Fig. 5A the BTB in vivo. A knockdown of Cx43 by Ϸ50% in Sertoli cells a and b and Fig. 5Baand b). This knockdown of Cx43 and PKP2 using specific siRNA duplexes versus non-targeting control siRNA also caused a decline in the steady-state level of ZO-1 (Fig. 5A duplexes showed that a reduction of Cx43 did not affect the a and b) and a mild but significant decline in the levels of steady-state levels of various integral membrane or regulatory N-cadherin and Coxsackie virus and adenovirus receptor (CAR) proteins at the BTB (Fig. 4Aaand b). Dual-labeled immunoflu- at the Sertoli cell surface using the biotinylation technique (Fig. orescence analysis revealed no observable changes in the distribu- 5Caand b), suggesting a possible increase in endocytosis of tion of TJ proteins occludin and ZO-1 at the Sertoli–Sertoli cell N-cadherin and CAR. A surge in the protein level of the TJ protein claudin-11 was observed, but its level at the cell surface interface (see later sections). Using a biotinylation technique to remained unchanged (Fig. 5Aaand b and Fig. 5Caand b). The assess changes in proteins at the cell surface following RNAi of findings shown in Fig. 6C are consistent with results of Fig. 6A, Cx43 versus controls, we found that a knockdown of Cx43 did not wherein the Sertoli cell TJ permeability barrier was disrupted affect the levels of 4 integral membrane proteins at the cell surface with simultaneous knockdown of both Cx43 and PKP2 but not (Fig. 4B). These proteins were the TJ proteins occludin and when the expression of either Cx43 or PKP2 alone was sup- JAM-A, the basal ES-protein N-cadherin, and the hemidesmosome pressed (Fig. 6A). A redistribution of ZO-1 and occludin at the protein ␤1-integrin. Furthermore, a knockdown of Cx43 failed to cell–cell interface was also observed following knockdown of perturb the Sertoli cell TJ permeability barrier (see later discus- both Cx43 and PKP2 but not with the knockdown of Cx43 alone sion). Collectively, a knockdown of Cx43 alone, and thus the (Fig. 6 B and C). However, knockdown of PKP2 alone was suppression of connexons comprising Cx43, had an insignificant sufficient to perturb the distribution of ZO-1, but not occludin, effect on the BTB integrity. In short, Cx43 alone may not be at the Sertoli–Sertoli cell interface. These findings show that the required for the maintenance of the BTB between Sertoli cells. desmosome-like junction is functionally constituted by the GJ protein Cx43 and desmosomal protein PKP2. Furthermore, this A Knockdown of Cx43 and PKP2 Perturbs TJ Permeability Barrier Cx43/PKP2 complex is linked functionally to the occludin/ZO-1 complex. A disruption of the Cx43/PKP2 can affect the integrity Function and Cell Adhesion in Sertoli Cells. Besides existing by itself of the TJ permeability barrier via a mislocalization of occludin in the testis, the GJ is also a known integral component of the and ZO-1, perhaps mediated by enhanced endocytosis of occlu- desmosome-like junction, which has the ultrastructural features din at the BTB. of both the desmosome and GJ (14). Because Cx43 interacted with the desmosomal protein PKP2 (Fig. 3B), we sought to Discussion examine if a concurrent knockdown of Cx43 and PKP2 would Cx43 Alone Does Not Confer Cell Adhesion in the Seminiferous affect the BTB function. A knockdown of Cx43 and PKP2, by Epithelium. Cx43, a GJ protein, was shown to express stage speci- Ϸ40% at the protein level and by Ϸ60% at the RNA level, ficity in the seminiferous epithelium of rat testes. Based on its

Li et al. PNAS ͉ June 23, 2009 ͉ vol. 106 ͉ no. 25 ͉ 10215 Downloaded by guest on September 27, 2021 A a RNAi 2 D 3 D Ctrl Cx43 + Ctrl Cx43 + PKP2 PKP2 Connexin 43, GJ 39 - 41 kDa b 1.5 - Connexin 26, 26 kDa - Occludin, 65 kDa * 1 - Claudin-11, 22 kDa * * TJ - CAR, 46 kDa ** 0.5

- ZO-1, 210 kDa (Arbitrary unit)

- N-cadherin, 130 kDa Relative protein level 0 Basal ES - α-catenin, 102 kDa Connexin 43 Claudin-11 ZO-1 β - -catenin, 92 kDa 2D - Ctrl siRNA Fig. 5. A functional study to assess the role of the HD - β1-integrin, 140 kDa 2D - (Cx43 + PKP2) siRNA Cx43/PKP2 complex in the BTB by RNAi in primary - ERK1, 44 kDa 3D - Ctrl siRNA Sertoli cell cultures. On day 4, when the Sertoli cell TJ Signaling - ERK2, 42 kDa 3D - Cx43 + PKP2 siRNA barrier was established, cells were transfected with kinases - pERK1, 44 kDa siRNA duplexes for 24 h, either non-targeting control - pERK2, 42 kDa or speciﬁc Cx43 and PKP2 siRNA pool, at a ﬁnal con- - Vimentin, 55 kDa Cytoskeleton centration of 100 nM using RiboJuice siRNA transfec- - Actin, 42 kDa tion reagent. Cultures were terminated 48 or 72 h thereafter. Lysates were obtained for immunoblot- BCa RNAi - 2D RNAi - 2D ting, and RNA was extracted for RT-PCR. (A) A knock- a down of Cx43 by Ϸ40% on day 2 (2 D) and by Ϸ30% on day 3 (3 D) after transfection was detected. An increase Ctrl Cx43 + PKP2 Ctrl Cx43 + PKP2 PKP2 Cx43 + PKP2 Ctrl Cx43 + 653 bp Ctrl -ve in TJ integral membrane protein claudin-11 of Ϸ10% 517 bp - Plakophilin-2, 561 bp - Claudin-11, 22 kDa 453 bp and a decline in the TJ adaptor protein ZO-1 of Ϸ10% 394 bp - S16, 385 bp 298 bp - CAR, 46 kDa were detected 3 days (3 D) after transfection. Other - N-cadherin, 130 kDa selected target proteins at the BTB remained un- surface protein b Biotinylated cell changed after RNAi of Cx43 and PKP2. (B)Onday2(2 1 D) after RNAi of Cx43 and PKP2, there was a decline in Ϸ b 1 PKP2 at the RNA level of 60% when assessed by * Ctrl siRNA 0.5 * * RT-PCR. (C) Cell surface proteins were biotinylated on (Cx43 + PKP2) day 2 (2 D) after transfection with the corresponding siRNA (Arbitrary unit) 0 0.5 siRNA duplexes to assess the effect of RNAi of Cx43 and Relative RNA level Relative RNA Plakophilin-2 PKP2 on the steady-state level of integral membrane Ctrl siRNA proteins versus control siRNA duplexes. Results shown (Cx43 + PKP2) siRNA protein (Arbitrary unit) 0 here are representative results of 3 experiments. *, P Ͻ

Relative level of cell surface Claudin-11CAR N-cadherin 0.05; **, P Ͻ 0.01.

localization in normal and adjudin-treated testes, it seems to be single knockdown of either Cx43 or PKP2, was shown to disrupt the associated with the restructuring of junctions. At stage VIII of the Sertoli cell TJ permeability function. The knockdown of both Cx43 epithelial cycle, fully developed spermatids leave the epithelium at and PKP2 also was shown to induce a redistribution of the TJ spermiation. The BTB undergoes restructuring to facilitate the proteins occludin and ZO-1 from TJ fibrils at the Sertoli cell transit of preleptotene spermatocytes, and considerable loss of surface. As compared with controls, a significant decline in the level Cx43 at the apical ES and the BTB was detected immunohisto- of biotinylated cell surface proteins, including N-cadherin and chemically. Similarly, when elongating spermatids are induced to CAR, was detected at the BTB following the knockdown of Cx43 depart from the epithelium prematurely by adjudin in tubules other and PKP2. These findings show that the Cx43/PKP2 protein com- than those at stage VIII, the staining of Cx43 is reduced drastically plex is a putative functional GJ/desmosome protein complex that at the apical ES. Collectively, these findings suggest that Cx43 is regulates Sertoli cell TJ barrier function through its effects on the involved in conferring cell adhesion in the seminiferous epithelium. occludin/ZO-1 protein complex at the BTB. This observation also We further investigated the role of Cx43 in cell–cell adhesion using is in agreement with an emerging concept that the desmosome can a loss-of-function approach by RNAi using primary Sertoli cells serve as the platform for signal transduction (9), so its disruption at cultured in vitro with an established TJ permeability barrier that the desmosome-like junction can perturb the TJ barrier function at mimicked the BTB in vivo. The knockdown of only Cx43 in Sertoli the BTB. cells did not cause any changes in the integrity of the TJ barrier or in the distribution and levels of junction proteins, such as occludin A Possible Mechanism by Which the Cx43/PKP2 Complex Induces and ZO-1, at or near the cell surface. These results suggest that the Blood–Testis Barrier Restructuring. The desmosome is an interme- loss of Cx43 function at the GJ and the desmosome-like junction diate filament-based anchoring junction, whereas the GJ is a does not affect the BTB integrity and show that Cx43 alone is not communication junction (3, 7). Thus, the desmosome-like junc- necessary for the maintenance of the BTB integrity. It is likely that tion is expected to confer cell adhesion (3). In this study, a other connexins (15) can supersede the transient loss of Cx43 at the transient loss of the functional desmosome-like junction via a BTB conferred by Sertoli cells. knockdown of Cx43 and PKP2 was shown to cause a disruption of the Sertoli cell TJ barrier partly via a redistribution of occludin Regulation of the Cell Adhesion by the Cx43/PKP2 Protein Complex at and ZO-1 at the cell–cell interface. The ERK MAPK pathway, the Blood–Testis Barrier. Because Cx43 was shown to interact which becomes activated during various junction restructuring structurally with the desmosomal protein PKP2, as demonstrated by events (16), is not involved in this event. The non-receptor co-immunoprecipitation, the functional role of this Cx43/PKP2 protein tyrosine kinase c-Src is likely to be involved. complex in the desmosome-like junction was examined. The simul- c-Src interacts with Cx43, and it has been reported to be a taneous knockdown of both Cx43 and PKP2 by RNAi, but not the component of the BTB in the rat testis (13). Studies from other

10216 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0901700106 Li et al. Downloaded by guest on September 27, 2021 A Cx43 and PKP2, c-Src may fail to maintain the phosphorylation Transfection status of occludin at the TJ fibrils at the BTB, leading to a decline in occludin at the Sertoli cell interface. This change in distribution of occludin, such as moving from the Sertoli–Sertoli cell interface ** * ** to cell cytosol, thus perturbs the Sertoli cell TJ barrier. ** * Is the Desmosome-Like Junction Involved in the Preleptotene Sper- matocyte Transit at the Blood–Testis Barrier? It was postulated that junction proteins, such as CAR, residing on primary spermatocytes Time in culture (Day) may interact with those on Sertoli cells to facilitate cell movement at the BTB while maintaining the immunological barrier during ZO-1/ Occludin/ B ZO-1 C Occludin spermatogenesis (21). There is a unidirectional movement of siRNA/DAPI siRNA/DAPI primary spermatocytes from the basal to the apical compartment of ab ab the epithelium while differentiating into late-stage spermatocytes (3). Desmosome-like junctions, as well as those formed between Sertoli cells, also are formed between Sertoli and spermatocytes/ round spermatids but not elongating and elongated spermatids (8). We propose herein that primary preleptotene spermatocytes in transit at the BTB probably use the desmosome-like junction to cd cd facilitate cell movement while maintaining the immunological barrier. At stages VIII and IX of the epithelial cycle, the interaction of a desmosome-like junction protein on the Sertoli cell with one on a primary spermatocyte in transit, instead of with one on another Sertoli cell, may cause a localized reduction in the number of functional desmosome-like junctions between Sertoli cells. It has

ef ef been reported that GJ communication from germ cells to Sertoli BIOLOGY

cells differs from that between Sertoli cells (15, 22). A change of DEVELOPMENTAL Cx43–Cx43 association from 2 Sertoli cells to a Sertoli cell– spermatocyte may destabilize the occludin/ZO-1-based TJ fibrils at the BTB, possibly at the apical region of the spermatocyte, facili- tating cell migration at the BTB. Other studies have shown that the gh gh formation of an N-cadherin-based AJ is a prerequisite for the assembly of GJ (23, 24) and desmosome (25). Because the knockdown of Cx43 and PKP2 led to a loss of the basal ES protein N-cadherin at the cell surface (Fig. 7), the disruption of basal ES might prohibit the assembly of a desmosome-like junction at the apical region of a spermatocyte in transit. There probably is a feedback loop that promotes the assembly of TJ, basal ES, and/or Fig. 6. A functional study to assess the role of Cx43 and PKP2 in BTB function desmosome-like junction in the basal region of a spermatocyte in by RNAi in primary Sertoli cell cultures. (A) The TJ barrier function was monitored transit to assemble new TJ fibrils to maintain the immunological by quantifying TER across the cell epithelium. Sertoli cells were transfected for barrier. This feed-back loop might be mediated by paracrine factors 24 h with the corresponding siRNA duplexes at a final concentration of 80 nM. *, produced locally at the BTB, which are known to regulate BTB Ͻ Ͻ P 0.05; **, P 0.01. (B,C) Fluorescence staining of ZO-1 (CY3) and occludin (CY3) dynamics (3, 11, 26). was performed 48 h after transfection to investigate the distribution of the target proteins after the knockdown of Cx43 and/or PKP2. In the transfection step of the Materials and Methods experiment, 52.5 nM of control (Ctrl), Cx43, and/or PKP2 siRNA was co-transfected with 17.5 nM of siGLO Green (FITC), acting as a transfection indicator, at a final Animals. Sprague-Dawley rats were obtained from Charles River Laboratory. concentration of 70 nM. The green staining surrounding nuclei indicates positive The use of animals was approved by the Animal Use and Care Committee of transfection. After knockdown of both Cx43 and PKP2, there was less ZO-1 and the Rockefeller University (protocol number 06018). occludin at the Sertoli–Sertoli cell interface than in controls (Baand b and Caand b vs. Bgand h and Cgand h, respectively). A decline in ZO-1 at the cell–cell Primary Sertoli Cell Culture. Sertoli cells were isolated from testes of 20-day-old interface was observed after the knockdown of PKP2 only, but not of Cx43 (Be Sprague-Dawley rats and were cultured in F12/DMEM supplemented with and f and c and d vs.aand b). There was no change in the distribution of occludin growth factors and bacitracin (27). The functional TJ permeability barrier, in the single knockdown of Cx43 or PKP2 (Cc–fvs.C a and b). Results shown are which mimicked the BTB in vivo morphologically and functionally, was estab- the representative data of 3 experiments. (Scale bars, 40 ␮m.) lished on day 3 or 4 when assessed by transepithelial electrical resistance (TER) across the cell epithelium (27) and by electron microscopy (28). Ultrastructures of TJ, ES, and desmosome-like junction were observed. In this in vitro system, epithelia have shown that occludins localized at TJ fibrils (17) are virtually no apical ES was present, because the cultured Sertoli cells were contaminated with negligible germ cells, including elongating and elongated phosphorylated at both Ser/Thr and Tyr residues (18, 19). A minor spermatids. pool of less phosphorylated occludins is not assembled into TJ fibrils and is restricted to the basolateral region (18, 20). The Transient Transfection of siRNA Duplexes and TJ Permeability Barrier Assess- modulation of the phosphorylation status of occludins thus pro- ment. Sertoli cells were cultured on Matrigel (BD Biosciences)-coated culture vides a unique mechanism to assemble and disassemble TJ fibrils plates at 0.5 ϫ 106 cells/cm2, Millicell bicameral units at 1.0 ϫ 106 cells/cm2,or in epithelia rapidly in response to changes in the cellular environ- coverglasses at 0.08–0.1 ϫ 106 cells/cm2. Cells were transfected with siRNA ment, such as during the transit of preleptotene spermatocytes at duplexes (Dharmacon, Thermo Fisher Scientific), namely the ON-TARGET plus non-targeting siRNA control pool (D-001810–10), the siRNA pool specifically the BTB. The c-Src in the Cx43/PKP2/c-Src complex is in proximity targeting Cx43 (J-100614–09, -10, -11) or PKP2 (J-087525–09 -10, -11, -12), and to the TJ-based occludin and ZO-1, as demonstrated by their siGLO green transfection indicator (D-001630–01), at a final concentration of co-localization with Cx43. Following a transient knockdown of 70–100 nM using RiboJuice siRNA transfection reagent (Novagen, EMD Bio-

Li et al. PNAS ͉ June 23, 2009 ͉ vol. 106 ͉ no. 25 ͉ 10217 Downloaded by guest on September 27, 2021 Fig. 7. A schematic drawing illustrating the regulation of BTB dynamics by desmosome-like junction. (A) A sim- pliﬁed overview of different junctions at the Sertoli– Sertoli interface at the BTB in normal adult rat testes. (B) Following the knockdown of Cx43 alone, there is a decline in the number of connexons constituted by Cx43, possibly in both the GJ and desmosome-like junction, but the cell adhesion conferred by TJ and AJ is unaffected, perhaps because connexons constituted by other GJ proteins maintain the GJ communication and the BTB integrity. As such, the BTB integrity still remains intact. (C) The knockdown of both Cx43 and PKP2 leads to a decline in the levels of TJ proteins occludin, CAR, and ZO-1 and basal ES protein N-cadherin through a yet-to-be deﬁned signaling pathway(s). Nonetheless, this leads to BTB disruption. This mechanism, as depicted here, is likely uti- lized by the testis at stage VIII of the seminiferous epithelial cycle of spermatogenesis to induce BTB restructuring to facilitate the transit of primary preleptotene spermatocytes at the BTB. AJ refers to the basal ES at the BTB.

sciences) on day 3 or 4 after an intact cell epithelium was established. In the group endosome-mediated protein degradation were inhibited. Thus, the level of cell transfected with both Cx43 and PKP2 siRNA, equal amounts of each siRNA duplex surface proteins following RNAi in treatment versus control groups could be pool were used so that the final concentration was equivalent to the non- assessed quantitatively. Cell lysates were prepared in Nonidet P-40 lysis buffer targeting control siRNA duplex. For treatment groups transfected with either (26). Biotinylated cell surface proteins were isolated from cell lysates using Ultra- Cx43 or PKP2 siRNA alone, an equal amount of Ctrl siRNA was used together with Link Immobilized NeutrAvidin Protein Plus (Pierce/Thermo-Fisher Scientific) and Cx43 or PKP2 siRNA to obtain the same final siRNA concentration. Cells then were were resolved in SDS/PAGE for immunoblot analysis using the corresponding incubated in the transfection mixture for 24 h and replenished with fresh DMEM/ specific antibodies. F12 thereafter. In selected experiments where Sertoli cells were plated on bicameral units, the TJ barrier across the cell epithelium was monitored by TER as Statistical Analyses. GB-STAT statistical analysis software package (Version 7.0, described (27). Dynamic Microsystems) was used to perform Student’s t test and 1-way ANOVA, followed by the Tukey/Kramer procedure. Estimation of the Cell Surface Protein Levels Following RNAi by the Biotinylation General Methods. Information on the lysate preparation, RT-PCR, immunoblot Technique. For Sertoli cell cultures transfected with non-targeting control siRNA, analysis, immunohistochemistry, and dual-labeled immunofluorescence anal- specific siRNA duplexes targeting Cx43, PKP2, or Cx43 and PKP2 were used to ysis can be found in the SI Text. assess whether there were any changes in the level of junction proteins at the cell surface following RNAi. In short, proteins on the Sertoli cell surface following ACKNOWLEDGMENTS. This study was supported in part by grants from the transfection were biotinylated using sulfo-NHS-SS-biotin (Pierce/Thermo-Fisher National Institutes of Health (NICHD, R01 HD056034, R03 HD051512, and U54 Scientific) at 4 °C for 30 min, and non-bound biotins were quenched with 50 mM HD029990 Project 5) to C.Y.C. and Hong Kong Research Grants Council (HKU ammonium chloride solution (26, 29). At this temperature, endocytosis and 7693/07M) to W.M.L.

1. Wong CH, Cheng CY (2005) The blood–testis barrier: Its biology, regulation, and 16. Wong CH, Cheng CY (2005) Mitogen-activated protein kinases, adherens junction physiological role in spermatogenesis. Current Topics in Developmental Biology dynamics, and spermatogenesis: A review of recent data. Dev Biol 286:1–15. 71:263–296. 17. Fujimoto K (1995) Freeze-fracture replica electronic microscopy combined with SDS 2. Setchell BP (2008) Blood–testis barrier, junctional and transport proteins and spermat- digestion for cytochemical labeling of integral membrane proteins. Application to the ogenesis in Molecular Mechanisms in Spermatogenesis, ed Cheng CY (Springer/Landes immunogold labeling of intercellular junctional complexes. J Cell Sci 108:3443–3449. Bioscience, Austin, Texas), pp 212–233. 18. Sakakibara A, Furuse M, Saitou M, Ando-Akatsuka Y, Tsukita S (1997) Possible involve- 3. Mruk DD, Bruno S, Cheng CY (2008) Anchoring junctions as drug targets: Role in ment of phosphorylation of occludin in tight junction formation. J Cell Biol 137:1393– contraceptive development. Pharmacol Rev 60:146–180. 1401. 4. Setchell BP, Waites GMB (1975) The blood–testis barrier in The Handbook of Physiol- 19. Tsukamoto T, Nigam SK (1999) Role of tyrosine phosphorylation in the reassembly of ogy. Section 7, Vol. V. Male Reproductive System, eds Hamilton DW, Greep RO occludin and other tight junction proteins. Am J Physiol 276:F737–F750. (American Physiological Society, Washington, DC), pp 143–172. 20. Cordenonsi M, et al. (1997) Occludin dephosphorylation in early development of 5. Russell LD (1977) Movement of spermatocytes from the basal to the adluminal com- Xenopus laevis. J Cell Sci 110:3131–3139. partment of the rat testis. Am J Anat 148:313–328. 21. Wang CQF, Cheng CY (2007) A seamless trespass: Germ cell migration across the 6. Clermont Y (1972) Kinetics of spermatogenesis in mammals: Seminiferous epithelium seminiferous epithelium during spermatogenesis. J Cell Biol 178:549–556. cycle and spermatogonial renewal. Physiol Rev 52:198–235. 22. Decrouy X, Gasc JM, Pointis G, Segretain D (2004) Functional characterization of cx43 7. Alberts B, et al. (2002) Mol Biol Cell (Garland Science, New York) 4th Ed. based gap junctions during spermatogenesis. J Cell Physiol 200:146–154. 8. Russell L (1977) Desmosome-like junctions between Sertoli and germ cells in the rat 23. Meyer RA, Laird DW, Revel J-P, Johnson RG (1992) Inhibition of gap junction and testis. Am J Anat 148:301–312. adherens junction assembly by connexin and A-CAM antibodies. J Cell Biol 119:179– 9. Green KJ, Simpson CL (2007) Desmosomes: New perspectives on a classic. J Invest 189. Dermatol 127:2499–2515. 24. Shaw RM, Fay AJ, Puthenveedu MA, von Zastrow M, Jan Y-N (2007) Microtubule 10. Mese G, Richard G, White TW (2007) Gap junctions: Basic structure and function. plus-end-tracking proteins target gap junctions directly from the cell interior to J Invest Dermatol 127:2516–2524. adherens junctions. Cell 128:547–560. 11. Li MWM, et al. (2006) Tumor necrosis factor ␣ reversibly disrupts the blood–testis 25. Amagai M, et al. (1995) Delayed assembly of desmosomes in keratinocytes with barrier integrity and impairs Sertoli-germ cell adhesion in the seminiferous epithelium disrupted classic-cadherin-mediated cell adhesion by a dominant negative mutant. of adult rat testes. J Endocrinol 190:313–329. J Invest Dermatol 104:27–32. 12. Cheng CY, et al. (2005) AF-2364 [1-(2,4-dichlorobenzyl)-1H-indazole-3-carbohydra- 26. Yan HHN, Mruk DD, Lee WM, Cheng CY (2008) Blood–testis barrier dynamics are zide] is a potential male contraceptive: A review of recent data. Contraception 72:251– regulated by testosterone and cytokines via their differential effects on the kinetics of 261. protein endocytosis and recycling in Sertoli cells. FASEB J 22:1945–1959. 13. Lee NPY, Cheng CY (2005) Protein kinases and adherens junction dynamics in the 27. Yan HHN, Mruk DD, Wong EWP, Lee WM, Cheng CY (2008) An autocrine axis in the seminiferous epithelium of the rat testis. J Cell Physiol 202:344–360. testis that coordinates spermiation and blood–testis barrier restructuring during sper- 14. Vogl AW, Vaid KS, Guttman JA (2008) The Sertoli cell cytoskeleton. Molecular Mech- matogenesis. Proc Natl Acad Sci USA 105:8950–8955. anism in Spermatogenesis, ed Cheng CY (Springer/Landes Bioscience, Austin, Texas), 28. Siu MKY, Wong CH, Lee WM, Cheng CY (2005) Sertoli-germ cell anchoring junction pp 186–211. dynamics in the testis are regulated by an interplay of lipid and protein kinases. J Biol 15. Risley MS, Tan IP, Farrell J (2002) Gap junctions with varied permeability properties Chem 280:25029–25047. establish cell-type speciﬁc communication pathways in the rat seminiferous epithe- 29. Le TL, Yap AS, Stow JL (1999) Recycling of E-cadherin: A potential mechanism for lium. Biol Reprod 67:945–952. regulating cadherin dynamics. J Cell Biol 146:219–232.

10218 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0901700106 Li et al. Downloaded by guest on September 27, 2021