Filamin A Is a Regulator of Blood-Testis Barrier Assembly during Postnatal Development in the Rat Testis

Abstract

The blood-testis barrier (BTB) is an important ultrastructure in the testis. A delay in its assembly during postnatal development leads to meiotic arrest. Also, a disruption of the BTB by toxicants in adult rats leads to a failure in spermatogonial differentiation. However, the regulation of BTB assembly remains unknown. Herein, filamin A, an actin filament cross-linker that is known to maintain and regulate cytoskeleton structure and function in other epithelia, was shown to be highly expressed during the assembly of Sertoli cell BTB in vitro and postnatal development of BTB in vivo, perhaps being used to maintain the actin filament network at the BTB. A knockdown of filamin A by RNA interference was found to partially perturb the Sertoli cell tight junction (TJ) permeability barrier both in vitro and in vivo. Interestingly, this down-regulating effect on the TJ barrier function after the knockdown of filamin A was associated with a mis-localization of both TJ and basal ectoplasmic specialization proteins. Filamin A knockdown also induced a disorganization of the actin filament network in Sertoli cells in vitro and in vivo. Collectively, these findings illustrate that filamin A regulates BTB assembly by recruiting these proteins to the microenvironment in the seminiferous epithelium to serve as the building blocks. In short, filamin A participates in BTB assembly by regulating protein recruitment during postnatal development in the rat testis.

In the rat testis, initiation of spermatogenesis, namely differentiation of preleptotene/leptotene to zygotene and late spermatocytes, plus the initiation of meiosis I and II all take place by approximately 15–26 d postpartum (dpp) (1), coinciding with the assembly of the blood-testis barrier (BTB) that occurs by approximately 15–19 dpp (2, 3) and that is completed and fully functional by approximately 25 dpp (4). Interestingly, treatment of neonatal rats with diethylstilbestrol (a synthetic nonsteroidal estrogen) that delayed the BTB assembly by 4 wk also delayed the first wave of spermiation by 4 wk because spermatocytes failed to enter meiosis I/II but underwent degeneration (5), illustrating meiotic arrest in the absence of a functional BTB. Additionally, rats treated with an acute high dose of adjudin [1-(2,4-dichlorobenzyl)-1H-indazole-3-carbohydrazide, a potential male contraceptive (6) that is known to induce reversible male infertility at low doses in rats (7)] failed to reinitiate spermatogenesis to regain fertility long after the drug was metabolically cleared vs. controls and other low-dose groups, because the BTB was found to be irreversibly disrupted even though the spermatogonial stem cell and spermatogonia population in the testes remained unaffected (8). Collectively, these findings demonstrate unequivocally the significance of the BTB on the initiation of spermatogenesis, in particular differentiation of spermatogonia beyond type A and meiosis. In fact, BTB dysfunction leads to infertility in men (9, 10). Interestingly, there is no report in the literature investigating the mechanisms that regulate BTB assembly during postnatal development, and the participating molecules that regulate BTB assembly are also not known.

We thought it pertinent to examine the role of actin regulatory proteins in BTB assembly because the most distinctive ultrastructure of the BTB is the bundles of actin filaments that line perpendicularly to the apposing plasma membranes of adjacent Sertoli cells in the basal compartment of the seminiferous epithelium near the basement membrane. These actin filament bundles are also the dominant structural component of the basal ectoplasmic specialization (ES), a testis-specific actin-based adherens junction, coexisting with tight junctions (TJ) and gap junctions, and these junctions together with desmosomes constitute the BTB and confer its unusual adhesive strength vs. other blood-tissue barriers (11). In a survey to assess changes in the expression of several actin regulatory proteins during postnatal development including actin-related protein 3 (Arp3) of the Arp2/3 complex that confers branched actin polymerization (12), epidermal growth factor receptor pathway substrate 8 (an actin barbed-end capping and bundling protein) (13), drebrin E (an actin-binding protein that recruits Arp3 to the ES in the testis) (14), and filamin A (a nonmuscle actin filament cross-linker that confers actin filament network and cell adhesion) (15, 16), only filamin A was found to be predominantly expressed in the testis at the time when BTB begins to assemble at age approximately 15–17 dpp. We thus performed studies to delineate the role of filamin A on the assembly and maintenance of BTB during post-natal development. Some of these findings are highly unexpected, yet they illustrate some unexplored functional role of this actin cross-linking protein, in particular its ability to recruit TJ and basal ES proteins to the BTB for its assembly during postnatal development.

Materials and Methods

Animals and antibodies

Sprague Dawley rats were obtained from Charles River Laboratories (Kingston, NY). The use of these animals was approved by the Institutional Animal Use and Care Committee of the Rockefeller University (protocol numbers 06018 and 12506).

Primary cultures of germ cells

Germ cells were isolated from testes of adult rat (∼275–300 g body weight) as detailed elsewhere (17). Using this mechanical procedure without trypsinization, DNA flow cytometric analysis of germ cell populations isolated from these approximately 90- to 100-d-old rat testes performed as earlier described (17) showed that the relative percentages of spermatogonia (2C, 6.78%), spermatogonia synthesizing DNA/preleptotene spermatocytes (S-phase, 1.41%), primary spermatocytes (4C, 9.89%), round spermatids (1C, 48.02%), and hypercondensed elongating/elongated spermatids (H, 33.90%) were similar to rat testes in vivo. Total germ cells were cultured in serum-free F12/DMEM (Ham's F12 nutrient mixture/DMEM; Sigma-Aldrich, St. Louis, MO) supplement with 2 mm sodium pyruvate and 6 mm sodium dl-lactate at a density of 2.5 × 10⁶ cells/ml in 100-mm petri dishes at 35 C and terminated within 16–20 h (17). These germ cells had a viability of more than 97% based on erythrosine red dye exclusion test (18).

Primary cultures of Sertoli cells and assessment of TJ permeability barrier in vitro

Sertoli cells were isolated from 20-d-old rat testes (19). For different experiments, freshly isolated Sertoli cells were plated on Matrigel-coated (BD Biosciences, San Jose, CA) (diluted 1:7 in F12/DMEM) 1) 12-well culture dishes at a density of 0.5 × 10⁶ cells/cm² for subsequent lysate preparation, 2) coverslips at a density of 0.05 × 10⁶ cells/cm² for subsequent dual-labeled immunofluorescence analysis (so that cell nuclei were spatially well separated and cell-cell interface could be easily visualized), or 3) Millicell bicameral units (Millipore, Billerica, MA) at a density of 1.2 × 10⁶ cells/cm² (to form an intact cell epithelium on the bicameral unit as earlier characterized) (20), which were then placed in 24-well dishes (with 0.5 ml F12/DMEM in the apical and basal compartment) for subsequent transepithelial electrical resistance (TER) measurement to assess the Sertoli cell TJ barrier function. Sertoli cells were cultured in F12/DMEM supplemented with epidermal growth factor, insulin, bacitracin, and transferrin at 35 C in a humidified atmosphere with 95% air/5% CO₂ (vol/vol) (19). About 36 h thereafter, cultures were subjected to a hypotonic treatment using 20 mm Tris (pH 7.4) at 22 C for 2 min (21) to lyse residual germ cells. Thereafter, cells were washed twice in F12/DMEM. Thus, the purity of our Sertoli cell preparations was more than 98% with negligible contaminations of either germ, peritubular myoid, or Leydig cells when assessed by RT-PCR using corresponding cell markers (22). It is noted that Sertoli cells from 20-d-old rats are fully differentiated without mitotic activity (23). Also, a functional TJ barrier was established by these Sertoli cells when TER was quantified across the cell epithelium in bicameral units using a Millipore Millicell electrical resistance system (24–26), and ultrastructures of TJ, basal ES, gap junction, and desmosome that constitute the BTB in vivo were visible by electron microscopy (27–30). In short, the Sertoli cell TJ barrier that was established in vitro mimicked the BTB in vivo, and this in vitro culture system has been extensively used by investigators in the field to study BTB dynamics (24, 25, 31–35). In some experiments, Sertoli cell cultures on d 4 were treated with vehicle control, TNFα (10 ng/ml), TGF-β3 (3 ng/ml), testosterone (2 × 10⁻⁷ m), or estradiol-17β (2 × 10⁻⁹ m) and terminated on d 6 for dual-labeled immunofluorescence analysis. These concentrations of cytokines and steroids were selected based on earlier studies (29, 35–37). In all experiments, Sertoli cells plated on Matrigel-coated bicameral units, dishes, or coverslips were prepared in triplicates including both control and treatment groups. Each experiment was repeated at least three times using different batches of Sertoli cells excluding pilot experiments which were used to assess optimal experimental conditions as reported herein.

Transfection of Sertoli cells with small interfering RNA (siRNA) duplexes

For immunoblot analysis, Sertoli cells cultured on Matrigel-coated 12-well dishes on d 4 were transfected with 100 nm filamin A-specific siRNA duplexes (J-098267-12; Thermo Scientific Dharmacon, Lafayette, CO) or nontargeting siCONTROL pool (D-001810-10; Thermo Scientific Dharmacon), using 4 μl RiboJuice siRNA transfection reagent (Novagen; EMD Biosciences, Billerica, MA) in a final volume of 1 ml F12/DMEM for 24 h. Thereafter, cells were washed, and media were replenished with fresh F12/DMEM containing growth factors for another 72 h before being harvested for lysate preparation. For dual-labeled immunofluorescence analysis, Sertoli cells cultured on Matrigel-coated coverslips were transfected on d 3 with 80 nm filamin siRNA duplexes vs. nontargeting control duplexes, together with 1 nm siGLO red transfection indicator (D-001630-02; Dharmacon) for 24 h. Cells were then subjected to fixation by methanol or paraformaldehyde for staining 72 h after transfection. For TER measurement, cells cultured on Millicell bicameral units were transfected with 150 nm filamin A-specific siRNA duplexes vs. control duplexes twice (24 h per transfection), on d 2 and 3 with a 12-h interval for recovery, using 4 μl RiboJuice siRNA transfection reagent in a final volume of 0.5 ml F12/DMEM. TER across the cell epithelium was assessed on each day as described above to assess changes in TJ barrier function.

Filamin A silencing in rat testes in vivo

Intratesticular injection of siRNA duplexes for RNA interference (RNAi) was performed with a 28-gauge needle as described (13). In short, rats (n = 3–4 rats per treatment group vs. controls) were subjected to RNAi via intratesticular injection of filamin A-specific vs. nontargeting siRNA duplexes by administering each testis with a transfection solution of 12.5 μl containing 100 nm siRNA duplexes, 0.5 μl RiboJuice, and 11.5 μl Opti-MEM (Invitrogen, Carlsbad, CA) on 18, 19, and 20 dpp, respectively, with a total of three administrations per rat. This treatment protocol and the selected concentration of siRNA duplexes were based on pilot experiments. In short, nontargeting control and filamin A-specific siRNA duplexes were administered to one of the two testes in the same animal. Rats were terminated by CO₂ asphyxiation on 21, 25, 30, and 35 dpp. Testes were collected in liquid nitrogen, and frozen sections (7 μm thickness) were obtained in a cryostat at −22 C, either for dual-labeled immunofluorescence analysis (n = 3 rats for each time point) or BTB integrity assay in vivo (n = 3 rats for each time point). Both negative and positive control [treated with cadmium chloride at 5 mg/kg body weight (BW) for 3 d] groups also have n = 3 rats.

RNA extraction and RT-PCR

Testes from 90-d-old rats, cultured Sertoli cells, and germ cells were lysed in TRIzol reagent (Invitrogen) for RNA extraction (38). Contaminating genomic DNA was removed in RNA samples by treatment with ribonuclease-free deoxyribonuclease I (Invitrogen). Total RNA was then reverse transcribed into cDNA with Moloney murine leukemia virus reverse transcriptase (Promega, Madison, WI). Primers specific to filamin A (sense, 5′-CCAAGTAGACTGCTCAAGTG-3′, nucleotides 3659–3678; antisense, 5′-TGTATGTGCCATCACCACAG-3′, nucleotides 4049–4068) (accession number NM_001134599.1) and S-16 (sense, 5′-TCCGCTGCAGTCCGTTCAAGTCTT-3′, nucleotides 87–110; antisense, 5′-GCCAAACTTCTTGGATTCGCAGCG-3′, nucleotides 448–471) (accession number NM_001169146.1) were used for PCR with Go Taq DNA polymerase (Promega) (38) (Table 1).

Table 1.

Primers used for PCR

Gene	GenBank accession number	Primer orientation	Primer sequence (5′–3′)	Nucleotides position	Expected size (bp)
FLNa	NM_001134599.1	Sense	CCAAGTAGACTGCTCAAGTG	3659–3678	410
		Antisense	TGTATGTGCCATCACCACAG	4049–4068
S-16	NM_001169146.1	Sense	TCCGCTGCAGTCCGTTCAAGTCTT	87–110	385
		Antisense	GCCAAACTTCTTGGATTCGCAGCG	448–471

Immunoblot analysis and coimmunoprecipitation (co-IP)

Lysates of testes and Sertoli and germ cells were obtained as described (30, 38). Protein concentrations in these samples were quantified by spectrophotometry using the DC protein assay kit (Bio-Rad Laboratories, Hercules, CA). For immunoblot analysis, approximately 80 μg protein lysates from testes and 30 and 100 μg protein lysates from Sertoli and germ cells, respectively, were used for each lane. For co-IP, 500 μg Sertoli cell protein lysate was used, and co-IP was performed as described (12, 30, 38). Antibodies used for immunoblot analysis are listed in Table 2. Chemiluminescence was performed using reagents prepared in our laboratory as described (39). Densitometric analysis was performed using Scion Image software (version 4.0.3). To avoid inter-experimental variations, all samples within an experiment group were simultaneously analyzed by immunoblotting.

Table 2.

Antibodies used for different experiments in this report

Antibody	Host species	Vendor	Catalog number	Working dilution			Conjugation
				IB	IF	IP
Filamin A	Mouse	Abcam, Cambridge, MA	ab80837	1:500	1:100
β1-Integrin	Rabbit	BD Biosciences	610468	1:1000
Vimentin	Mouse	Santa Cruz Biotechnology, Santa Cruz, CA	sc-6260	1:300	1:50
Arp3	Mouse	Sigma-Aldrich	A5979	1:3000
Occludin	Rabbit	Zymed/Invitrogen	71-1500	1:250	1:50	1:40
ZO-1	Rabbit	Zymed/Invitrogen	61-7300	1:250	1:50
JAM-A	Rabbit	Zymed/Invitrogen	36-1700	1:300	1:50	1:40
N-Cadherin	Rabbit	Santa Cruz Biotechnology	sc-7939	1:300	1:50	1:40
α-Catenin	Rabbit	Santa Cruz Biotechnology	sc-7894	1:250
β-Catenin	Mouse	Zymed/Invitrogen, Grand Island, NY	71-2700	1:250	1:50	1:40
γ-Catenin	Rabbit	BD Transduction Laboratories, San Jose, CA	610254	1:1000
Erk1/2	Rabbit	Cell Signaling Technology, Danvers, MA	4695	1:1000
Phospho-Erk1/2	Rabbit	Cell Signaling Technology	4370	1:2000
Actin	Goat	Santa Cruz Biotechnology	sc-1616	1:250
Mouse IgG	Bovine	Santa Cruz Biotechnology	sc-2371	1:3000			HRP
Rabbit IgG	Bovine	Santa Cruz Biotechnology	sc-2370	1:3000			HRP
Goat IgG	Bovine	Santa Cruz Biotechnology	sc-2350	1:3000			HRP
Mouse IgG	Donkey	Invitrogen	A21202		1:100		Alexa Fluor 488
Rabbit IgG	Donkey	Invitrogen	A21206		1:100		Alexa Fluor 488
Rabbit IgG	Donkey	Invitrogen	A31572		1:100		Alexa Fluor 555

Antibodies used in the experiments reported herein were found to cross-react with the corresponding proteins in the rat as indicated by the manufacturers. HRP, Horseradish peroxidase; IB, immunoblotting; IF, immunofluorescence microscopy; IP, immunoprecipitation.

Dual-labeled immunofluorescence analysis and filamentous actin (F-actin) staining

Frozen sections of testes of approximately 7 μm thickness were obtained in a cryostat at −22 C and placed on microscopic glass slides or Sertoli cells plated on Matrigel-coated coverslips (in controls or after treatments) at 0.05 × 10⁶ cells/cm² were fixed in 4% paraformaldehyde (wt/vol, in PBS) at room temperature (20 ± 2 C) or in methanol at −20 C for 10 min. Cells or sections were permeabilized with 0.1% Triton X-100 (vol/vol, in PBS) for 4 min. For blocking, fixed testis sections or Sertoli cells were treated with 5% BSA (wt/vol, in PBS) for 30 min. Primary antibodies (Table 2) were diluted in 1% BSA (wt/vol, in PBS) at 1:50 or 1:100 and incubated overnight (at 4 C for frozen sections, and at room temperature for cultured Sertoli cells). Secondary antibodies (Invitrogen) conjugated with CY3–555 (red) or fluorescein isothiocyanate (FITC)-488 (green) were diluted at 1:150 in 1% BSA (wt/vol, in PBS) and incubated at room temperature for 1 h. For F-actin staining, testis sections or cells were incubated with rhodamine phalloidin (red) (Invitrogen) or FITC-conjugated phalloidin (green) (Invitrogen) at the same time with the secondary antibodies. Sections or cells were mounted with Prolong Gold Antifade reagent with 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen). Images were visualized with an Olympus BX61 fluorescence microscope and captured using an Olympus DP71 digital camera at 12.5 megapixels in TIFF format. Fluorescence images were analyzed and adjusted for brightness/contrast and for image overlay using Photoshop in Adobe Creative Suite Design Premium (version 3.0). In in vivo experiments of filamin A knockdown, some images were subjected to semiquantitative analysis by quantifying signal intensity using ImageJ version 1.44I software to assess the efficacy of filamin A knockdown. All micrographs reported herein are representative results of at least three independent experiments using different sets of animals or batches of cells, which yielded similar results.

BTB integrity assay

For rats subjected to BTB integrity assay after filamin A knockdown in vivo, rats were under anesthesia by ketamine HCl (60 mg/kg BW, im) with xylazine (10 mg/kg BW, im) as an analgesic. Thereafter, the jugular vein was exposed by making a small incision on the skin. Approximately 1 mg inulin-FITC (4.6 kDa; Sigma-Aldrich) in 200 μl PBS was administered into the jugular vein using a 28-gauge needle. About 1 h later, rats were euthanized by CO₂ asphyxiation, testes were removed, snap-frozen in liquid nitrogen for at least 5 min, and frozen cross-sections were obtained in a cryostat at −20 C. Using fluorescence microscopy, the BTB integrity was evaluated by its ability to block the diffusion of inulin-FITC from the basal to the adluminal compartment as described (8, 40). For positive controls, rats were treated with one dose of cadmium chloride (5 mg/kg BW) via ip 72 h before BTB assay (n = 3 rats for each time point), which is known to disrupt the BTB (41, 42). Negative controls were testes received nontargeting siRNA treatment from the same rats with n = 3 rats. Data from this in vivo assay were also semiquantitatively analyzed by plotting the fluorescence signal that traveled from the basement membrane (adjacent to the BTB) vs. the radius of the tubule as described (8, 40).

Statistical analysis

Data from treatment groups vs. the corresponding controls were statistically analyzed by two-way ANOVA with Dunnett's test using GB-STAT software (version 7.0; Dynamic Microsystems Inc., Silver Spring, MD).

Results

Filamin A is a Sertoli cell protein in rat testes involved in BTB assembly in vitro and in vivo

Filamin A was expressed exclusively by Sertoli cells in the seminiferous epithelium (Supplemental Fig. 1, A and B, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org). Its protein level was significantly induced during the assembly of the TJ permeability barrier (Fig. 1, A and B, and Supplemental Fig. 1C), and the ultrastructures of TJ, basal ES, gap junction, and desmosome were also detected in these cells in vitro (Fig. 1B), which mimicked the BTB in vivo as reported earlier (9, 33). Using an antibody specific to filamin A (Table 2 and Fig. 1Ci), filamin A was distributed in Sertoli cell cytoplasm in filamentous forms consistent with its function as an actin cross-linker (16) (Fig. 1Cii). Filamin A was also found to structurally interact with junctional adhesion molecule-A (JAM-A) (a TJ integral membrane protein at the BTB) (9) but not occludin or β-catenin (Fig. 1D). In the testes, the expression of filamin A was high at ages approximately 12–17 dpp (Supplemental Fig. 1B and Fig. 1E) when the BTB was being assembled (4, 9). A similar trend of expression was found in β1-integrin level, a marker of spermatogonial stem cells (43) and of the apical ES at the Sertoli-spermatid interface (44) but not vimentin (Fig. 1E). β1-Integrin and vimentin were selected because they are known binding ligands of filamin A involved in cell adhesion (45, 46). The expression of filamin A and its colocalization with F-actin were also examined by dual-labeled immunofluorescence analysis. Filamin A was found to distribute around spermatogonia and earlier spermatocytes, and at the tunica propria of the tubule, and colocalized with F-actin on 12 and 17 dpp (Fig. 1F). By 20 dpp, the signal of filamin A and its colocalization with F-actin in the epithelium began to decline, and very weak staining of filamin A was detected in the testis by 90 dpp and limited mostly to the basal compartment consistent with its localization at the BTB (Fig. 1F).

Fig. 1.

Expression and characterization of filamin A during the assembly of Sertoli cell BTB in vitro and in vivo. A, Immunoblot analysis of filamin A in lysates (30 μg protein) of Sertoli cells (SC) cultured in vitro with a functional TJ barrier established by approximately d 2–3. Actin served as a loading control. B, Electron micrographs of Sertoli cells cultured in vitro for 4 d with ultrastructures found at the BTB in vivo. Microvilli, which are typical features of Sertoli cells cultured in vitro, were visible in these cells (green arrowheads in Bi and Bii). Two adjacent Sertoli cells (SC) and the distinctive nucleus (Nu) with the specialized junctions annotated by apposing white arrowheads were seen (Bi). Gap junctions (blue arrowheads in Bii) and TJ (red arrowheads in Biii, illustrating the kisses between apposing Sertoli cells, see apposing white arrowheads) coexisted with basal ES (see yellow bracket), typified by the presence of actin filament bundles (yellow arrowheads in Bii and Biii) sandwiched between cisternae of endoplasmic reticulum (ER) and the apposing plasma membranes of two adjacent Sertoli cells (apposing white arrowheads), which together with desmosome (purple arrowheads, Bii) constitute the BTB. Scale bars, 2.5 μm (Bi); 0.2 μm (Bii), and 0.25 μm (Biii). C, Immunoblotting to illustrate the specificity of the anti-filamin A antibody using lysates of SC (Ci). This antibody was used to visualize filamin A (green) in Sertoli cells (0.05 × 10⁶ cells/cm²) cultured for 4 d. Nuclei were visualized with DAPI (blue) (Cii). Scale bar, 25 μm (Cii). D, Using co-IP, filamin A was found to structurally interact with JAM-A, but not occludin or β-catenin, using 500 μg protein lysates of Sertoli cells cultured for 4 d with an established TJ permeability barrier. Sertoli cell lysate (30 μg protein) alone without co-IP and co-IP using normal rabbit IgG to substitute for the precipitating antibody served as the corresponding positive and negative control, respectively. The findings shown here are representative data of an experiment that was repeated three times and yielded identical results. E, Immunoblot analysis of filamin A, β1-integrin, and vimentin using testis lysate (30 μg protein) from rats at different dpp, with actin serving as a protein loading control (Ei). The histogram summarizes results of Ei after each data point was normalized against actin. Protein levels at 12 dpp were arbitrarily set as 1 against which statistical comparison was performed (Eii). Each bar is a mean ± sd of n = 3 rats. **, P < 0.01. F, Colocalization of filamin A (green) and F-actin (red) in frozen sections of testes from rats at different ages. Nuclei were visualized with DAPI (blue). It was noted that the expression of filamin A and its colocalization with F-actin were considerably high at approximately 12–17 dpp at the time of BTB assembly but declined considerably by 20–30 dpp when the BTB was established (4). Scale bar in F, 25 μm, which applies to all micrographs in F.

Effects of cytokines and steroids on the localization of filamin A vs. F-actin and vimentin in Sertoli cell epithelium

Cytokines (e.g. TGF-β3 and TNFα) and steroids (e.g. testosterone and estradiol-17β) were shown to regulate Sertoli cell BTB dynamics in vitro and in vivo (9, 47). To examine whether filamin A is involved in these events, Sertoli cells cultured for 4 d with an established TJ barrier were treated with either cytokines or steroids and terminated 48 h later to examine any alterations in cellular distribution of filamin A vs. F-actin and vimentin (Fig. 2). In control cultures, filamin A, F-actin, and vimentin were evenly distributed in Sertoli cell cytosol as shown in Fig. 2, A, B, and C, respectively. Interestingly, in Sertoli cells treated with TNFα, but not TGF-β3, filamin A displayed a remarkably concentrated signal around cell nuclei instead of a uniform distribution in control cells (Fig. 2D vs. 2A). Although in testosterone-treated cells, filamin A localized more predominantly at the cell cortex near the plasma membrane (Fig. 2J vs. 2A), and this pattern of changes in filamin A was consistent with F-actin (Fig. 2, E and K). The distribution of the intermediate filament protein vimentin, however, exhibited no obvious change after cytokine or steroid treatment (Fig. 2, F, I, L, and O vs. C). However, it is of interest to note that treatment of Sertoli cells with TNFα (10 ng/ml), TGF-β3 (3 ng/ml), testosterone (2 × 10⁻⁷ m), estradiol-17β (2 × 10⁻⁹ m) vs. controls failed to induce any significant changes in the steady-state protein levels of filamin A, vimentin, and actin when cells were harvested by 48 h for immunoblot analysis similar to the Sertoli cells shown in Fig. 2 except that a density of 0.5 × 10⁶ cells/cm² was used so that sufficient protein could be obtained for analysis, illustrating changes reported in Fig. 2 were the results of alterations in the localization and/or distribution of filamin A and F-actin in Sertoli cell cytosol rather than an up- or down-regulation on the expression of these proteins.

Fig. 2.

Effects of cytokines and steroids on the cellular localization of filamin A vs. F-actin and vimentin in Sertoli cells cultured in vitro. Sertoli cells (0.05 × 10⁶ cells/cm²) were cultured alone for 4 d. Thereafter, cells were treated with TNFα (10 ng/ml), TGF-β3 (3 ng/ml), testosterone (2 × 10⁻⁷ m), or estradiol-17β (2 × 10⁻⁹ m) vs. controls for 48 h as shown in A–O. Cells were fixed and processed for immunofluorescence microscopy as described in Materials and Methods and stained for filamin A (green), F-actin (red), and vimentin (green). Nuclei were visualized with DAPI (blue). TNFα treatment led to an accumulation of filamin A in cytoplasm closer to cell nuclei (arrowheads in D), whereas testosterone treatment led to a more intense signal of filamin A at the cell cortex (arrowheads in J). Changes of F-actin were similar to filamin A in the same treatment group (arrowheads in E and K) vs. controls; however, vimentin in all groups were unaffected. Scale bar in A, 25 μm, which applies to micrographs in A–O.

Knockdown of filamin A by RNAi in Sertoli cells perturbs the TJ permeability barrier via changes in the distribution of BTB-associated adhesion proteins and protein-protein interactions at the cell-cell interface

To investigate whether filamin A participates in maintaining a functional TJ permeability barrier, filamin A was knocked down by transfecting Sertoli cells with specific filamin A siRNA duplexes vs. nontargeting siRNA duplexes. A study by immunoblot analysis showed that when filamin A was silenced by approximately 70% vs. nontargeting control, no off-target effect was detected when the following proteins were monitored: known binding ligands of filamin A β1-integrin and vimentin; actin regulatory protein Arp3, known to be involved in actin nucleation/branching when it forms a complex with Arp2 (48); TJ proteins occludin, zona occludens 1 (ZO-1), and JAM-A; basal ES proteins N-cadherin and α/β/γ-catenin; and cell signaling MAPK ERK1/2 and phospho-ERK1/2 (Fig. 3, A and B, and Supplemental Fig. 2, A–D). However, the knockdown of filamin A by approximately 70% via two cycles of transfection was found to partially perturb the Sertoli cell TJ barrier function by approximately 35–40% vs. the corresponding control (Fig. 3C). Using dual-labeled immunofluorescence analysis in which Sertoli cells plated at a lower density for optimal visualization of cell-cell contacts (note that these cells possessed ultrastructures of TJ, basal ES, gap junction, and desmosome when examined by electron microscopy, see Fig. 1B) were transfected with filamin A siRNA together with the siGLO transfection indicator (Fig. 3D, red), the signal of filamin A (green fluorescence) was considerably reduced (Fig. 3D) in filamin A-silenced cells, consistent with immunoblotting data shown in Fig. 3A (Fig. 3D). Interestingly, F-actin became truncated and mislocalized in filamin A-silenced cells, forming some knot-like filament ultrastructures (Fig. 3D). Furthermore, occludin, ZO-1, and β-catenin, but not N-cadherin, were also found to be mislocalized with these proteins redistributed from the cell-cell interface into cell cytosol (Fig. 3D). Because filamin A knockdown by RNAi was found to interfere with the Sertoli cell TJ barrier in vitro without affecting the steady-state levels of junction proteins, we next performed co-IP to examine any alterations of protein-protein interactions in protein adhesion complexes at the Sertoli cell BTB. As anticipated, there was a significant reduction in protein-protein interactions between basal ES proteins N-cadherin and β-catenin at the Sertoli cell BTB after filamin A knockdown (Fig. 3, E and F), consistent with dual-labeled immunofluorescence data shown in Fig. 3D in which β-catenin was found to be mislocalized, moving away from cell-cell interface into cell cytosol, whereas N-cadherin remained near the cell surface (Fig. 3D vs. E and F).

Fig. 3.

Knockdown of filamin A by RNAi in Sertoli cell epithelium perturbs the TJ barrier function via changes in F-actin organization, protein distribution, and protein-protein interactions at the BTB. A, Sertoli cells (0.5 × 10⁶ cells/cm²) were cultured alone for 4 d; thereafter, cells were transfected on d 4 with specific filamin A (FLNa) siRNA vs. nontargeting control siRNA duplexes for 24 h. Cultures were terminated 72 h later to obtain lysates for immunoblotting. When filamin A was knocked down by approximately 70% (B), no off-target effect was detected when several BTB proteins were assessed by immunoblot analysis (see Supplemental Fig. 2). pERK, Phospho-ERK. B, Histogram based on immunoblot results of filamin A shown in A after normalizing the data against actin to illustrate the efficacy of RNAi. Filamin A level in the nontargeting control was arbitrarily set at 1. Bar is mean ± sd of two independent experiments, but three additional pilot experiments yielded similar results. **, P < 0.01. C, Effects of filamin A knockdown by RNAi on the Sertoli cell TJ barrier function was monitored by quantifying TER across the cell epithelium when Sertoli cells (1.2 × 10⁶ cells/cm²) were transfected with the corresponding siRNA duplexes on d 2 and 3 for 24 h, respectively, with a 12-h interval for recovery, at a final concentration of 150 nm. Filamin A silencing was found to perturb the Sertoli cell TJ barrier function. *, P < 0.05; **, P < 0.01. This experiment was repeated three times using different Sertoli cell preparations and yielded similar results. D, On d 3, Sertoli cells (0.05 × 10⁶ cells/cm²) were transfected with the corresponding siRNA duplexes for 24 h at a final concentration of 80 nm, together with 1 nm siGLO Red (Cy3), which served as a transfection indicator. Immunostaining of filamin A, F-actin, occludin, ZO-1, N-cadherin, and β-catenin (all in green) was performed on d 6 to investigate the effects of filamin A RNAi on protein distribution in Sertoli cells. Red staining surrounding nuclei indicates positive transfection. After transfection, the signal of filamin A was considerably declined in cells treated with filamin A-specific siRNA duplexes vs. control (Ctrl) groups. F-actin became less organized in treated cells vs. controls. A considerable change in protein distribution in occludin, ZO-1, and β-catenin, but not N-cadherin, at the cell-cell interface was detected after filamin A knockdown when these proteins moved from the cell surface into the cell cytosol. Scale bar in D, 25 μm, which applies to all micrographs. E, Co-IP was performed using Sertoli cell lysates (∼500 μg protein) after filamin A silencing vs. nontargeting control with an anti-N-cadherin antibody (see Table 2) as the precipitating antibody. The blots were probed with an anti-β-catenin to assess changes in protein-protein interactions between N-cadherin and β-catenin. Sertoli cell lysates without co-IP (∼35 μg protein) served as the positive control. F, Histogram summarizing co-IP results shown in C from n = 3 experiments. The protein-protein interactions in the nontargeting control group were arbitrarily set at 1. **, P < 0.01.

Filamin A knockdown by RNAi in the testis interferes with the BTB assembly during postnatal development

Data in Fig. 1 illustrate the possible involvement of filamin A in BTB assembly during postnatal development, and the knockdown of filamin A in Sertoli cells was found to partially affect the TJ permeability barrier function via mislocalization of TJ and basal ES protein at the BTB (Fig. 3). We next sought to examine the BTB phenotype when filamin A was knocked down in the testis by transfecting testes with specific filamin A siRNA duplexes vs. nontargeting control duplexes via intratesticular administration in immature rats at the time that BTB was being assembled by 18–20 dpp (4, 9) using a regimen shown in Fig. 4A. Thereafter, an in vivo functional assay was used to assess the BTB integrity by monitoring the barrier function that blocked the transit of FITC-inulin administered at the jugular vein from entering the apical compartment (Fig. 4, B and C). In positive controls, rats were treated with CdCl₂ (5 mg/kg BW) 72 h before the BTB assay, which is known to induce BTB disruption (41, 42). It was found that on 21 and 25 dpp, the time when a functional BTB was being assembled in rats (4, 9), the knockdown of filamin A significantly delayed the assembly of BTB (Fig. 4B), and the BTB integrity in the knockdown testes was significantly more leaky vs. the control testes (Fig. 4C). Interestingly, no significant difference was detected between filamin A-silenced testes and controls on 30 and 35 dpp (Fig. 4, B and C) when the basal expression of filamin A was considerably low vs. younger rats (Fig. 1E). Results shown in Fig. 4 and Fig. 1E thus demonstrate filamin A is involved in the assembly of a functional BTB during postnatal development.

Fig. 4.

Effects of filamin A knockdown on BTB assembly during postnatal development in the rat testis. A, A schematic drawing illustrating the treatment regimen. Testes of rats (n = 3–4 rats for each termination time point in treatment vs. control groups) at age 18, 19, and 20 dpp received either nontargeting control or specific filamin A siRNA duplexes daily (in ∼10–12.5 μl per testis to a desired concentration of 100 nm siRNA duplexes, assuming the volume of the testis from rats at 18, 19, and 20 dpp to be 0.058, 0.072, and 0.085 ml, corresponding to a testis weight at 0.058, 0.072, and 0.085 g, respectively) with three consecutive doses. After treatment, in vivo BTB integrity assays were performed on 21, 25, 30, and 35 dpp in the treatment vs. the two control groups. For positive control, rats were treated with a single dose of CdCl₂ (5 mg/kg BW, ip) without any siRNA duplexes and terminated 3 d later because CdCl₂ is known to induce irreversible BTB damage in rats within 48 h (41). B, Localization of inulin-FITC (green) in sections of frozen testes after administration of the inulin-FITC via the jugular vein in the BTB integrity assay. BTB integrity was assessed by its ability to block the influx of inulin-FITC into the adluminal compartment from the basal compartment, which lay behind the BTB in the epithelium near the basement membrane as denoted by the white broken circle in a tubule. White brackets in the micrographs illustrate the distance of inulin-FITC diffused into the seminiferous epithelium from BTB. Representative damaged tubules after filamin A RNAi at different time points were shown. Scale bar in B, 25 μm, which applies to all micrographs. Ctrl, Control. C, Histogram illustrating semiquantitative BTB integrity assay data by comparing the distance traveled by inulin-FITC from the BTB (D_Signal) vs. the radius of a tubule (D_Radius) (for a tubule that was obliquely sectioned, tubule radius was obtained by averaging the shortest and the longest radius from the basement membrane). Each bar is mean ± sd of approximately 120–180 tubules that were randomly selected from three rat testes. **, P < 0.01. FLNa, Filamin A.

Knockdown of filamin A by RNAi in vivo that disrupted postnatal BTB assembly is mediated by disorganization of the F-actin filament network and mislocalization of TJ and basal ES proteins at the BTB

The effects of filamin A knockdown on F-actin (red) organization in rat testes in vivo during postnatal development were further assessed by dual-labeled immunofluorescence analysis (Fig. 5). After filamin A RNAi (see regimen in Fig. 4A), rat testes of all ages examined herein showed a considerably weakened fluorescence signal of filamin A (green) vs. testes that received the nontargeting control siRNA duplexes (Fig. 5, A and B). The findings shown in Fig. 5B thus support that an effective transient knockdown of filamin A had occurred in vivo. However, on 21 dpp after filamin A knockdown, F-actin staining using rhodamine-conjugated phalloidin revealed a disorganization (e.g. F-actin was fragmented) of F-actin in testes (annotated by white arrowheads, Fig. 5) when compared with the corresponding control where F-actin filaments formed contiguous bundles (annotated by yellow arrowheads, Fig. 5). Similar phenotypes were found in the filamin A-silenced rat testes vs. control rats by 25 dpp, but this became less obvious by 30 dpp (Fig. 5). By 35 dpp, no significant difference in F-actin organization was observed in filamin A-silenced testes and control testes (Fig. 5). We next assessed changes in localization of TJ (e.g. JAM-A and ZO-1) and basal ES (e.g. β-catenin and N-cadherin) proteins in filamin A knockdown vs. control testes during postnatal development at the time of BTB assembly. On 21 and 25 dpp after filamin A silencing, JAM-A (red) and ZO-1 (green) were found to fail to be recruited to the BTB and exhibited a wider (see white brackets in Fig. 6A) and/or diffused (see white arrowheads in Fig. 6A) staining at BTB in the filamin A-silenced testes vs. control testes (Fig. 6A), consistent with in vitro findings shown in Fig. 3D. These changes were not noticeable and less obvious by 30 and 35 dpp. On 21, 25, and 30 dpp, N-cadherin was also found to fail to be recruited to the BTB, adjacent to the basement membrane, but less obvious for β-catenin (Fig. 6B) in the filamin A-silenced testes; however, the signals of both N-cadherin and β-catenin in the filamin A-silenced testes failed to surround spermatogonia and/or spermatocytes distinctively in the basal compartment near the BTB (see white arrowheads in Fig. 6B). These findings illustrate that the knockdown of filamin A in vivo affected the BTB assembly during postnatal development, which is mediated via changes in F-actin organization and proper distribution of adhesion protein complexes (e.g. JAM-A-ZO-1 and N-cadherin-β-catenin) at the BTB.

Fig. 5.

Changes in F-actin vs. filamin A organization in the seminiferous epithelium of rat testes during postnatal development after filamin A knockdown. A, Sections of control (Ctrl, treated with nontargeting siRNA duplexes) and filamin A-silenced testes from rats at specified ages (see Fig. 4A for the treatment regimen) were costained for filamin A (green) and F-actin (red) with cell nuclei visualized by DAPI (blue). In filamin A-silenced tubules, the signal of filamin A in the epithelium was considerably weakened vs. the corresponding control, which indicated an effective knockdown of filamin A in vivo. F-actin filaments were found to become disorganized, failed to form a continuous belt-like ultrastructure surrounding the seminiferous epithelium at the BTB in filamin A-silenced tubules (white arrowheads) vs. the corresponding control (yellow arrowheads), which was considerably more obvious on 21 and 25 dpp but became less obvious by 30 dpp and almost indistinguishable from control by 35 dpp. Scale bar, 15 μm, which applies to all micrographs. B, Imaging analysis of fluorescence signals of filamin A in the seminiferous epithelium of rat testes after filamin A knockdown in vivo. The histogram summarized findings shown in A by quantifying the fluorescence intensity of filamin A using ImageJ version 1.44I software package. At least 50 tubules were randomly selected from each testis and quantified. Each bar represents mean ± sd of n = 3 rats. **, P < 0.01.

Fig. 6.

The knockdown of filamin A (FLNa) by RNAi in the testis during postnatal development induces changes in the recruitment of TJ and basal ES proteins to the BTB. Sections of control (Ctrl, treated with nontargeting siRNA duplexes, panels a–d) and filamin A-silenced testes from rats (panels e–h) at specified ages (see Fig. 4A for the treatment regimen) were used for dual-labeled immunofluorescence analysis to colocalize TJ proteins [e.g. JAM-A (red) and ZO-1 (green)] in panel A or basal ES proteins [e.g. β-catenin (red) and N-cadherin (green)] in panel B at the BTB with cell nuclei visualized by DAPI (blue). The white broken ring indicates approximate location of the basement membrane in each tubule, which is adjacent to the BTB. In filamin A-silenced tubules, especially on 21 and 25 dpp (D), the signals of JAM-A (and ZO-1 in rats at 21 dpp) were found to be considerably mislocalized; instead of being recruited to the site of BTB, they were diffusely localized from the BTB (white brackets and white arrowheads in panel A) vs. the corresponding control. The signal of N-cadherin and β-catenin also became diffusely localized, in particular N-cadherin by 30 dpp in filamin A-silenced tubules vs. control (white arrowheads in panel B). It is noted that the changes depicted herein are rather uniform in all the tubules examined, because at age 21 and 25 dpp, stage-specific changes regarding the cellular composition of the tubules are not obvious. Even by age 35 dpp, only round spermatids were detected in the tubules without any condensing spermatids because elongating spermatids are not found in tubules until age 38 dpp in rats. Scale bar in panel Aa (21 dpp), 25 μm, which applies to for all micrographs at 21 and 25 dpp in panels A and B; scale bar in Aa (30 dpp), 50 μm, which applies to all micrographs at 30 and 35 dpp in panels A and B.

Discussion

Filamin A recruits integral membrane proteins and their adaptors to the BTB microenvironment for the assembly of TJ and basal ES during postnatal development

Filamin A is a dimeric protein composed of two identical polypeptide chains of 280 kDa that self-associate noncovalently at the C terminus via the dimerizing domain to form a V-shaped functional protein. Because each monomer of filamin A contains an F-actin-binding domain at its N terminus, each V-shaped dimeric filamin A binds to two F-actin, which favors perpendicular (i.e. at 90°) branching of F-actin (15, 49), illustrating this actin cross-linker is crucial to maintain the unique network of F-actin by causing the bundling of filament bundles at the basal ES that coexists with TJ and gap junction, which, in turn, constitute the BTB. In short, this intrinsic cross-linking activity of filamin A confers the typical configuration of actin filament bundles that lie perpendicular to the plasma membrane of Sertoli cells at the basal ES. Indeed, using an in vitro system of Sertoli cell cultures that mimics the BTB in vivo, which has been widely used by investigators to study BTB dynamics (27, 29, 30, 33), we found the expression of filamin A to be significantly induced during the assembly of the functional TJ barrier in vitro as well as during the BTB assembly in vivo at approximately 15–17 dpp during postnatal development. This correlative evidence was further confirmed by the knockdown of filamin A by RNAi to approximately 70% in Sertoli cells. It was found that the Sertoli cell TJ barrier was partially disrupted, which is accompanied by a mislocalization of TJ (e.g. occludin and ZO-1) and basal ES (e.g. β-catenin) proteins, in which these proteins moved away from cell-cell interface into cytosol, thereby destabilizing the TJ barrier function. These findings are consistent with a recent report using human coronary artery endothelial cells in which a knockdown of filamin A by RNAi was found to reduce the vascular permeability in vitro (50). More importantly, these in vitro findings were confirmed in studies in vivo, in which the knockdown of filamin A by RNAi indeed reduced the BTB integrity in vivo in postnatal rat testes, and this was also accompanied by a mislocalization of the actin network at the BTB and TJ (e.g. JAM-1 and ZO-1) and basal ES (e.g. N-cadherin and β-catenin) proteins. For instance, the recruitment of these integral membrane proteins and their corresponding adaptors that serve as the building blocks to the BTB site was partially affected in the seminiferous epithelium of the filamin A-silenced testes that were transfected with filamin A-specific siRNA duplexes vs. control testes transfected with nontargeting control siRNA duplexes. These findings also illustrate that filamin A, besides serving as the organizer of actin-based cytoskeleton and cell structure (15, 16), is involved in regulating cell adhesion function by recruiting adhesion protein complexes (e.g. JAM-A-ZO-1 and N-cadherin-β-catenin) to the BTB microenvironment for its assembly. Because filamins are known to serve as scaffolds for a wide range of proteins by binding them via various interacting domains (51), we have examined whether filamin A regulates the assembly of TJ and basal ES during postnatal development by interacting with these adhesion proteins. Indeed, the result of co-IP revealed direct structural interaction between filamin A and JAM-A. More important, knockdown of filamin A by RNAi was found to significantly perturb the structural interactions between adhesion protein complex N-cadherin and β-catenin at the Sertoli cell BTB, which contributes, at least in part, to the disruption of the Sertoli cell TJ permeability barrier. Moreover, filamin A is likely to work in concert with other actin-binding and regulatory proteins, such as epidermal growth factor receptor pathway substrate 8 (13), Arp2/3 complex (12), and drebrin E (14), to regulate BTB assembly during postnatal development and during spermatogenesis (9, 52), when the adhesions between differentiated Sertoli cells begin to form and the F-actin at the cell cortex begins to reshape, resulting in the change of local cell membrane shape at the cell-cell contacts. Based on the findings reported herein, it is noted that changes in the BTB integrity during postnatal development after filamin A knockdown was better detected at 21–25 dpp. After this period, even though the effects of filamin A knockdown on its expression was still detectable by immunofluorescence analysis, little change on the BTB integrity, F-actin organization, and recruitment of adhesion proteins to the BTB was observed. These findings, together with the age-dependent down-regulation of filamin A expression in rat testes, seemingly suggest that filamin A exerts its role on BTB function only during BTB postnatal development or its initial stages of development. It is likely that its function can be substituted or superseded by other actin cross-linking and regulatory proteins during development. This possibility needs additional future investigations.

TNFα and testosterone regulate BTB function via their effects on filamin A localization in the Sertoli BTB microenvironment

Cytokines (e.g. TNFα and TGF-β3) and steroids (e.g. testosterone) are known to perturb and promote Sertoli cell TJ barrier function, respectively (9, 11), in rat testes, illustrating these molecules have contrasting effects on the BTB function. Recent studies have shown that although both cytokines and testosterone promote endocytosis of integral membrane proteins (e.g. occludin and N-cadherin) at the BTB, cytokines target endocytosed proteins to endosome-/ubiquitin-mediated degradation, whereas testosterone induces transcytosis and recycling of endocytosed proteins back to the Sertoli cell surface (9, 11). Thus, it was postulated that cytokines disrupted “old” TJ-fibrils above the preleptotene spermatocytes in transit at the BTB after testosterone promoted the assembly of “new” TJ-fibrils behind the spermatocytes, such that the BTB integrity could be maintained during the passage of spermatocytes at the site (9, 11, 53). Interestingly, as reported herein, treatment of Sertoli cell epithelium in vitro with an established TJ barrier with either TNFα (but not TGF-β3) or testosterone (but not estradiol-17β) leads to differences in the distribution of filamin A and F-actin. For instance, TNFα induces filamin A and actin filaments to be clustered in cell cytosol instead of uniformly distributed in the cell, perhaps being used to facilitate endocytosis and degradation of integral membrane proteins at the BTB, thereby destabilizing the Sertoli cell TJ barrier (53). But testosterone causes filamin A and actin filaments to be more localized to the cell cortex near the plasma membrane, perhaps strengthening the cell-cell interface at the TJ barrier, perhaps via protein transcytosis and recycling as reported (53, 54). Although these possibilities will need to be addressed and further investigated in future studies, the findings reported herein have demonstrated filamin A can serve as a mediator of TNFα- and testosterone-induced regulation on the BTB function, which is supported by recent findings that filamins can mediate interferon signaling (55).

Conclusion

Filamin A is a crucial regulator of BTB assembly during postnatal development in the mammalian testis. It mediates its effects by inducing reorganization of actin filaments at the basal ES, recruiting cell adhesion protein complexes (e.g. JAM-A-ZO-1 and N-cadherin-β-catenin) and supporting protein-protein interactions in cell adhesion protein complexes (e.g. N-cadherin-β-catenin) at the developing BTB.