regulation of sertoli cell tight junction dynamics in the ... · dynamics in the testis even though...

Regulation of Sertoli Cell Tight Junction Dynamics inthe Rat Testis via the Nitric Oxide Synthase/SolubleGuanylate Cyclase/3�,5�-Cyclic GuanosineMonophosphate/Protein Kinase G Signaling Pathway:an in Vitro Study

NIKKI P. Y. LEE AND C. YAN CHENG

Population Council, New York, New York 10021

Nitric oxide (NO) synthase (NOS) catalyzes the oxidation ofL-arginine to NO. NO plays a crucial role in regulating variousphysiological functions, possibly including junction dynamicsvia its effects on cAMP and cGMP, which are known modu-lators of tight junction (TJ) dynamics. Although inducibleNOS (iNOS) and endothelial NOS (eNOS) are found in thetestis and have been implicated in the regulation of spermat-ogenesis, their role(s) in TJ dynamics, if any, is not known.When Sertoli cells were cultured at 0.5–1.2 � 106 cells/cm2 onMatrigel-coated dishes or bicameral units, functional TJ bar-rier was formed when the barrier function was assessed byquantifying transepithelial electrical resistance across thecell epithelium. The assembly of the TJ barrier was shown toassociate with a significant plummeting in the levels of iNOSand eNOS, seemingly suggesting that their presence by pro-ducing NO might perturb TJ assembly. To further confirm therole of NOS on the TJ barrier function in vitro, zinc (II) pro-toporphyrin-IX (ZnPP), an NOS inhibitor and a soluble guan-ylate cyclase inhibitor, was added to the Sertoli cell culturesduring TJ assembly. Indeed, ZnPP was found to facilitate theassembly and maintenance of the Sertoli cell TJ barrier, pos-

sibly by inducing the production of TJ-associated proteins,such as occludin. Subsequent studies by immunoprecipitationand immunoblotting have shown that iNOS and eNOS arestructurally linked to TJ-integral membrane proteins, such asoccludin, and cytoskeletal proteins, such as actin, vimentin,and �-tubulin. When the cAMP and cGMP levels in these ZnPP-treated samples were quantified, a ZnPP-induced reduction ofintracellular cGMP, but not cAMP, was indeed detected. Fur-thermore, 8-bromo-cGMP, a cell membrane-permeable analogof cGMP, could also perturb the TJ barrier dose dependentlysimilar to the effects of 8-bromo-cAMP. KT-5823, a specificinhibitor of protein kinase G, was shown to facilitate the Ser-toli cell TJ barrier assembly. Cytokines, such as TGF-� andTNF-�, known to perturb the Sertoli cell TJ barrier, were alsoshown to stimulate Sertoli cell iNOS and eNOS expressiondose dependently in vitro. Collectively, these results illustrateNOS is an important physiological regulator of TJ dynamicsin the testis, exerting its effects via the NO/soluble guanylatecyclase/cGMP/protein kinase G signaling pathway. (Endocri-nology 144: 3114–3129, 2003)

NITRIC OXIDE (NO) SYNTHASE (NOS) catalyzes theoxidative conversion of l-arginine to NO and l-

citrulline. In mammals, three types of NOS are found,namely neuronal NOS (nNOS), inducible NOS (iNOS), andendothelial NOS (eNOS). These NOSs differ in molecularsize, and physicochemical properties, cellular distribution,and each is regulated differently (for a review, see Ref. 1).Both iNOS and eNOS are found in the testis and areimplicated in spermatogenesis, infertility, sperm matura-tion, and programmed cell death of Sertoli and germ cells(2–10). Yet it is not known if NOSs can regulate junctiondynamics in the testis even though two of their down-stream effector molecules, such as cAMP and cGMP, are

known regulators of junction dynamics for almost a de-cade (for reviews, see Refs. 11–13).

Tight junctions (TJs) in the testis are constituted by severalTJ-integral membrane proteins, such as occludins, claudins,and junctional adhesion molecules (JAMs), and peripheralproteins, such as zonula occludens (ZO)-1, ZO-2, ZO-3, cin-gulin, and AF-6 (for a review, Ref. 11). In the testis, TJsbetween adjacent Sertoli cells at the basal compartment of theseminiferous epithelium create the blood-testis barrier (BTB)(for a review, see Ref. 11), which provides a unique micro-environment for germ cell development by sequestering pro-teins in the systemic circulation residing in the interstitiumfrom entering the seminiferous epithelium (for review, seeRef. 11). As such, the precise nature of TJ regulation in thetestis not only is important to the study of spermatogenesis,a thorough understanding on the biology of TJ dynamics andtheir regulation will yield new insights in developing novelapproaches for male contraception because a disruption of TJdynamics per se will affect spermatogenesis such as migra-tion of preleptotene and leptotene spermatocytes across theBTB from the basal to adluminal compartment for furtherdevelopment. Recent studies have shown that Sertoli cell TJ

Abbreviations: AC, Adenylate cyclase; AF-6, s-afadin; AJ, adherensjunction; BTB, blood-testis barrier; Ca2�, calcium ions; COX-2, cycloox-ygenase 2; d, deoxy; eNOS, endothelial NOS; HO-1, heme-oxygenase 1;iNOS, inducible NOS; IP, immunoprecipitation; JAM, junctional adhe-sion molecule; NO, nitric oxide; NOS, NO synthase; nNOS, neuronalNOS; PKA, cAMP-dependent protein kinase (protein kinase A); PKG,cGMP-dependent protein kinase (protein kinase G); SDS, sodium do-decyl sulfate; sGC, soluble guanylate cyclase; TJ, tight junction; TER,transepithelial electrical resistance; ZnPP, zinc (II) protoporphyrin IX;ZO, zonula occludens.

0013-7227/03/$15.00/0 Endocrinology 144(7):3114–3129Printed in U.S.A. Copyright © 2003 by The Endocrine Society

doi: 10.1210/en.2002-0167

3114

dynamics are regulated by an array of molecules, whichinclude proteases, protease inhibitors, cytokines, cAMP, pro-tein phosphorylation and/or dephosphorylation, and intra-cellular Ca2� level (for a review, see Ref. 11). Because NO hasrecently been implicated in the regulation of the junctionpermeability in the blood-retina TJ barrier and TJ barrier ofblood vessels (for a review, see Ref. 13), it is of interest toinvestigate whether NO plays a role in TJ dynamics in thetestis.

Two of the immediate downstream signaling molecules ofNO are adenylate cyclase (AC) and soluble guanylate cyclase(sGC), which generate cAMP and cGMP, respectively. Thesetwo molecules in turn activate cAMP-dependent protein ki-nase (protein kinase A, PKA) and cGMP-dependent proteinkinase (protein kinase G, PKG) (for reviews, see Refs. 12 and14). Whereas the precise nature of the NO downstream reg-ulatory pathways via cAMP, PKA, cGMP, and PKG thataffect junction dynamics is virtually unknown, emergingevidence from studies of other epithelia have clearly impli-cated these molecules in TJ dynamics (13, 15). For instance,it is known that cAMP at 4–20 �m or 100–500 �m can fa-cilitate or perturb the Sertoli cell TJ barrier in vitro, suggestingthe regulatory effects of cAMP on Sertoli cell TJ dynamics areconcentration dependent (16). Herein, we report our findingsusing primary Sertoli cells cultured in vitro as a model toelucidate the roles of iNOS and eNOS in the TJ assembly invitro. Collectively, these results illustrate the biological sig-nificance of NOS in TJ regulation in the testis.

Materials and MethodsAnimals and antibodies

Sprague Dawley (outbred) rats were obtained from Charles RiverLaboratories, Inc. (Kingston, NY). The use of animals in this report wasapproved by The Rockefeller University Animal Care and Use Com-mittee with Protocol nos. 00111, 97117, and 95129-R. The polyclonal andmonoclonal antibodies used in this study were raised in either rabbits,mice, or goats and were purchased from Santa Cruz Biotechnology, Inc.(Santa Cruz, CA) as follows: these include iNOS (M-19; catalog no.sc-650, lot J151), eNOS (C-20; catalog no. sc-654, lot K291), occludin(H-279; catalog no. sc-5562, lot E031), actin (H-196; catalog no. sc-7210,lot C222), �-tubulin (H-300; catalog no. sc-5546, lot D042), vimentin (V9;catalog no. sc-6260, lot B252) and nectin-3 (C-19; catalog no. sc-14806, lotK261). Afadin (catalog no. 610732, lot 1) was purchased from BD Trans-duction Laboratories, Inc. (San Diego, CA). Bovine antirabbit IgG (cat-alog no. sc-2370), bovine antimouse IgG (catalog no. sc-2371), and bovineantigoat IgG (catalog no. sc-2350) conjugated to horseradish peroxidasefrom Santa Cruz Biotechnology, Inc. were used as secondary antibodies.Both the catalog and lot numbers are listed herein because severalantibodies from other vendors failed to yield satisfactory results inpreliminary experiments.

Primary testicular cell cultures

Sertoli cells were isolated from 20-d-old rat testes as described (17, 18).Freshly isolated Sertoli cells were suspended in serum-free Ham’s F-12Nutrient Mixture and DMEM (F12/DMEM, 1:1, vol/vol, Sigma, St.Louis, MO) supplemented with 15 mm HEPES, 1.2 g/liter sodium bi-carbonate, 10 �g/ml bovine insulin, 5 �g/ml human transferrin, 2.5ng/ml epidermal growth factor, 20 mg/liter gentamicin, and 10 �g/mlbacitracin. For the isolation of Sertoli cells from 45- and 90-d-old rats, anearlier protocol (19) modified in this laboratory (18) was used. Sertolicells isolated from 20-d-old rat testes were plated at high cell density at0.5 � 106 cells/cm2 on Matrigel (Collaborative Research, Inc., Bedford,MA)-coated 12-well dishes (effective surface area, �3.8 cm2 per well)with 3 ml F12/DMEM per well. Cells were hypotonically treated with

20 mm Tris, pH 7.4, for 2.5 min to lyse the contaminating germ cells (20)on d 2 (�36 h after plating), and cultures were washed twice to removelysed cells and debris. The purity of Sertoli cells isolated from 20-d-oldand adult rat testes was approximately 95% and 85%, respectively. Forgerm cell preparation, cells were isolated from 20-, 45-, 60-, and 90-d-oldrats by a mechanical method with sequential filtrations using miracloth(Calbiochem Corp., La Jolla, CA), 100-�m and 20-�m nylon meshes(Spectrum, Houston, TX) and glass wool as detailed elsewhere (21, 22).Germ cell preparations with purity of greater than 90% were routinelyobtained. The purity of Sertoli and germ cell preparations were moni-tored using markers specific for germ, Leydig, Sertoli, and myoid cells,such as c-Kit receptor, 3�-hydroxysteroid dehydrogenase, testin, andfibronectin, respectively, as detailed elsewhere (18). Seminiferous tu-bules were isolated from adult rat (�300 g body weight) testes as pre-viously described (18). The final tubule preparations were nonrespon-sive to human chorionic gonadotropin (10 ng/ml) treatment when thelevels of testosterone in spent media were quantified, suggesting thatthey had negligible Leydig cell contamination (18).

RNA extraction

Testes were removed from 10-, 20-, 60-, and 90-d-old rats and ho-mogenized in RNA STAT-60 for RNA isolation using a Tissumizer(Tekmar, Cincinnati, OH). Total RNA was also isolated from cells usingRNA STAT-60 according to the manufacturer’s protocol as described (18,23). RNA concentration was quantified by using a GeneQuant II spec-trophotometer (Pharmacia, Uppsala, Sweden) at 260 nm. Two micro-grams of RNA from each sample were routinely used for RT reaction.

Electron microscopy

To assess the presence of TJs between Sertoli cells cultured in vitro inaddition to monitoring the transepithelial electrical resistance (TER)across the cell epithelium on bicameral units, electron microscopy wasperformed. In brief, Sertoli cells were isolated from 20-d-old rat testesand cultured at 0.5 � 106 cells/cm2 on Matrigel-coated 60-mm culturedishes. Sertoli cells were subjected to hypotonic treatment approxi-mately 36 h after plating to remove residual germ cells (20). On d 5, thesecells were fixed and processed for electron microscopy. Briefly, cellswere fixed in 0.1 m cacodylate (pH 7.5 at 22 C) containing 2.5% (vol/vol)glutaraldehyde and 2.5% (wt/vol) paraformaldehyde for 1 h after a briefrinse with F12/DMEM. Cells were post-fixed with 1% OsO4 (vol/vol)in 0.1 m cacodylate for 1 h on ice and stained with 2% uranyl acetate(wt/vol) at room temperature. After dehydration through 70%, 95%, and100% ethanol, cells were then detached from culture dishes with pro-pylene oxide treatment and embedded in EPON (Electron MicroscopySciences, Fort Washington, PA) blocks (24). Silver sections were cutusing a Reichert Ultracut II ultramicrotome (Reichert Inc., Depew, NY)and examined with a JEOL 100CXII electron microscope (JEOL USA Inc.,Peabody, MA) at 80 kV. Electron microscopy was carried out in theBio-Imaging Resource Center at the Rockefeller University.

Treatment of Sertoli cells with cytokines

For cytokine treatments, recombinant human TGF-�1 (Calbiochem)at 0.1–3 ng/ml and recombinant TNF-� (Calbiochem) at 0.5–20 ng/mlwere added to the Sertoli cell cultures (0.5 � 106 cells/cm2) on d 3 (i.e.�24 h after hypotonic treatment). Control experiments included cellsreceiving no treatment (Ctrl) or cells exposed to vehicle (Ctrl/V). TNF-�and TGF-�1 were dissolved in PBS (10 mm sodium phosphate and 0.15m NaCl, pH 7.4, at 22 C) with 0.1% BSA (wt/vol), and 4 mm HCl in PBScontaining 0.1% BSA (wt/vol), respectively. Cells were terminated after6 h of incubation, and total RNA was extracted as described above.

Quantitative analysis on the Sertoli cell TJ-permeabilitybarrier by TER measurement across the cell epithelium

Freshly isolated Sertoli cells (1.2 � 106 cells/cm2) from 20-d-old rattestes were plated on Matrigel-coated (1:5 dilution with F12/DMEM,vol/vol) bicameral units (effective surface area, �0.6 cm2) (MillicellHA filters, Millipore Corp., Bedford, MA) as described (23) with0.5 ml of F12/DMEM each in the apical and basal compartment of the

Lee and Cheng • NOS and TJ Dynamics Endocrinology, July 2003, 144(7):3114–3129 3115

bicameral unit to allow the assembly of the TJ barrier. The TJ barrierwas quantified by recording TER across the Sertoli cell epithelium ona daily basis for up to a period of 5 d. Preliminary studies wereperformed using different cell densities to determine the optimal celldensity for TER measurement. Zinc (II) protoporphyrin-IX (ZnPP,BIOMOL Research Laboratories, Inc., Plymouth Meeting, PA), an NOS (25)and sGC (26) inhibitor, KT-5823 (C29H25N3O5) (BIOMOL), a specific inhib-itor of PKG, 8-bromo-cGMP (BIOMOL), a cell membrane-permeable cGMPanalog, 8-bromo-cAMP (BIOMOL), a cell membrane-permeable cAMPanalog, 9-(tetrahydro-2�-furyl) adenine (SQ-22536) (BIOMOL), a cell mem-brane-permeable adenylate cyclase inhibitor, were added to the basal andapical compartments of the bicameral units on d 1 at the desired concen-trations as shown in corresponding figure legends. The concentrations ofthe inhibitors used in our studies were selected as based on previouslypublished reports (16, 27–29). These compounds, except ZnPP, which wasreplenished daily, were incubated with Sertoli cells for only 24 h and wereremoved by rinsing the cell epithelium with F12/DMEM (3�) thereafter.TER readings from triplicate bicameral units in a single experiment weretaken daily and each experiment was repeated at least three times usingdifferent batches of Sertoli cells. The cytotoxicity of the compounds wastested by DNA assay (30) monitoring total DNA content in replicate cul-tures before and after treatment with specified inhibitors.

Semiquantitative RT-PCR

Semiquantitative RT-PCR was performed essentially as previouslydescribed (18, 23). Briefly, 2 �g of total RNA were reverse transcribedinto cDNAs using 1 �g oligo(deoxythymidine)15 with a Moloney murineleukemia virus RT kit (Promega Corp., Madison, WI) in a final reactionvolume of 25 �l. Each PCR was performed by combining 2 �l of RTproduct, 0.6 �g each of sense and antisense primers of a target gene(coamplified with �0.01 �g each of sense and antisense primers of ratribosomal S-16) (Table 1), 5 �l of 10� PCR buffer, 3 �l MgCl2 (25 mm),8 �l deoxy (d)-nucleotide triphosphates (200 �m each of dATP, dCTP,dGTP, and dTTP), 1.25 U Taq DNA polymerase, and sterile water to afinal reaction volume of 50 �l. The cycling parameters were as follows:denaturation at 94 C for 1 min, annealing at 56–61 C for 2 min (seeTable 1), and extension at 72 C for 3 min, for a total of 26–29 cycles, whichwas followed by a final extension of 15 min in a Techgene (Techne,Cambridge, UK) DNA Thermal Cycler. To obtain data suitable for semi-quantitative densitometric analysis, radioactive (i.e. hot) PCR was per-formed by using [�-32P]-labeled sense primers as described (18, 23). Therelative ratio of [32P]-sense primer of a target gene to [32P]-sense S-16 wasthe same as the unlabeled primers. To ensure that the production of aspecific target gene, such as eNOS and iNOS, and S-16 were in theirlinear phase, preliminary experiments were performed using differentconcentrations of template cDNAs (i.e. RT products) and primer pairs,and at different annealing temperatures. Autoradiography of PCR prod-ucts was performed using Kodak X-OMAT AR films (Eastman KodakCo., Rochester, NY). Autoradiograms were densitometrically scannedand normalized against S-16. In most experiments, results of RT-PCRwere verified by immunoblotting, which also ensured that any changesin the mRNA levels were indeed transmitted to functional proteins thatcould eventually affect cellular functions and junction dynamics.

Immunoprecipitation (IP) and immunoblotting

Cell lysates were obtained by incubating Sertoli and germ cells, andseminiferous tubules in an IP buffer [10 mm Tris, 0.15 m NaCl, 2 mmphenylmethylsulfonyl fluoride, 2 mm EDTA, 2 mm N-ethylmaleimide,1% Nonidet P-40 (vol/vol), 1 mm sodium orthovanadate (a proteintyrosine phosphatase inhibitor), 0.1 �m sodium okadate (a protein Ser/Thr phosphatase inhibitor), 10% glycerol (vol/vol), pH 7.4 at 22 C], tobe followed by a brief period of sonication (twice for 5 sec each, withsample tubes kept in melting ice). Cell lysates (supernatant) were ob-tained by centrifugation at 12,000 � g for 15 min at 4 C. Protein con-centration was quantified using BSA as a standard (31). For immuno-blotting without IP, lysates were denatured and resolved by SDS-PAGEunder reducing condition using approximately 100 �g protein per lane(18, 32), and all samples within a given experiment were processedsimultaneously to eliminate inter-experimental variations. Followingthe identification of a specific target protein by immunoblotting (18),blots were reprobed with a second target protein antibody by firstincubating the blot in a stripping buffer [62.5 mm Tris-HCl, pH 6.7, at22 C containing 100 mm 2-mercaptoethanol and 2% sodium dodecylsulfate (SDS) (wt/vol)] at 55 C for 30 min in a reciprocating water bathat 80 rpm to remove the initial primary and secondary antibodies, to befollowed by blocking in 5% milk in PBS-Tris buffer (10 mm sodiumphosphate, 0.15 m NaCl, 10 mm Tris, pH 7.4 at 22 C) (wt.vol) and thena second primary antibody. For immunoprecipitation, 400 �g of totalproteins from seminiferous tubule lysates (in �200-�l sample size) werefirst precleared by incubating the sample with normal rabbit serum (ornormal mouse or goat serum, depending on the source of precipitatingantibody to be used in subsequent steps) at 1:150 for 1 h at 4 C, followedby 10 �l protein A/G-PLUS agarose (Santa Cruz Biotechnology, Inc.) for1 h at 4 C, centrifuged at 1000 � g for 5 min to obtain the clear super-natant. Samples were then incubated with the corresponding antibodiesat 1:150 dilution in microfuge tubes and mounted onto a rotator (Glas-Col, Glass Tech Supplies Inc., Fullerton, CA) at room temperature for 3 hat approximately 15–20 rpm. Thereafter, 20 �l protein A/G-PLUS aga-rose (Santa Cruz Biotechnology, Inc.) was added to the samples toprecipitate the immunocomplexes. After 3 h of incubation at room tem-perature in a rotator, immunocomplexes bound to protein A/G-PLUSagarose were washed twice (IP buffer, 1000 � g, 5 min) and wereextracted in SDS-sample buffer [0.625 m Tris, pH 6.8 at 22 C, containing1.6% 2-mercaptoethanol, 1% SDS (wt/vol), 1 mm EDTA, 10% glycerol(vol/vol)] (32) and resolved by SDS-PAGE under reducing condition asdescribed (18). It is important that the protein A/G-PLUS agarose shouldbe resuspended by gentle agitation without vortexing so as not to dam-age the agarose beads.

Measurement of intracellular levels of cAMP and cGMP

ZnPP-treated and control Sertoli cell cultures were terminated usingan extraction buffer (50 mm Tris, pH 7.5, at 22 C, containing 5 mm EDTA,2 mm phenylmethylsulfonyl fluoride, 1 mm N-ethylmaleimide, and 1mm caffeine) by incubating cultures at 100 C for 5 min to coagulateproteins. Cells were then sonicated and centrifuged at 12,000 � g at 4 Cfor 15 min to remove cellular debris. The amount of cellular cAMP andcGMP in the supernatant was quantified using a [3H]cAMP assay system(with a detection limit of approximately 0.05 pmol/assay tube and the

TABLE 1. Primers used for semiquantitative RT-PCR to quantify mRNAs encoding S-16, iNOS, eNOS, and occludin

Target gene Primer sequence Orientation Position Length (bp) Annealingtemperature Refs.

S-16 5�-TCCGCTGCAGTCCGTTCAAGTCTT-3� Sense 15–38 385 a 575�-GCCAAACTTCTTGGATTCGCAGCG-3� Antisense 376–399

iNOS 5�-GCCTCCCTCTGGAAAGA-3� Sense 1210–1226 500 56 C 585�-TCCATGCAGACAACCTT-3� Antisense 1693–1709

eNOS 5�-TACGAAGAATGGAAGTGGTTC-3� Sense 433–453 249 59 C GenBank accession5�-TTGGCTCATCCATGTGGAACA-3� Antisense 661–681 no. AJ249546

Occludin 5�-CTGTCTATGCTCGTCATCG-3� Sense 770–788 294 61 C GenBank accession5�-CATTCCCGATCTAATGACGC-3� Antisense 1044–1063 no. AB016425

a The annealing temperature used for coamplifying S-16 with a target gene in each RT-PCR experiment was the same as used for the targetgene.

3116 Endocrinology, July 2003, 144(7):3114–3129 Lee and Cheng • NOS and TJ Dynamics

50% displacement was at 6.6 pmol) (Amersham Pharmacia Biotech,Piscataway, NJ) and a [3H]cGMP assay system (with a detection limit of�0.04 pmol/assay tube and the 50% displacement was at 3.2 pmol)(Amersham Pharmacia Biotech) according to the manufacturer’s in-structions. The interassay and intraassay coefficient of variation was at8–10% and 6–8% for both assays. All samples within a given experimentwere assayed simultaneously in a single experimental session to elim-inate interassay variations.

Statistical analysis

Student’s t test was performed by comparing treatment group at eachtime point with the corresponding control using the GB-STAT StatisticalAnalysis Software Package (version 7.0, Dynamic Microsystems, Inc.,Silver Spring, MD). Each experiment was repeated at least three timesusing different batches of cells, and each time point (or treatment group)had duplicate or triplicate cultures.

ResultsChanges in the steady-state mRNA levels of iNOS andeNOS during maturation of Sertoli cells, germ cells,and testes

To investigate the steady-state mRNA levels of NOSs intesticular cells and testes during maturation, we had isolatedSertoli cells, germ cells, and testes from rats at different ages.In 20-, 45-, and 90-d-old Sertoli cells, both NOSs rose steadilyand peaked at 90 d of age (Fig. 1, A–D). A much more drasticincrease in iNOS expression (Fig. 1, A and B) in Sertoli cellsduring maturation was detected when compared with eNOS(Fig. 1, C and D), 40-fold vs. 7-fold (Fig. 1, A and B vs. C andD). In contrast, the iNOS steady-state mRNA in germ cellsplunged steadily during maturation (Fig. 1, E and F),whereas the eNOS mRNA level rose steadily in germ cellsduring maturation (Fig. 1, G and H). It was noted that thelevel of eNOS in 20-d-old germ cells was too low to bedetected (Fig. 1, G and H). In the testes, iNOS expression leveltumbled steadily from 10–90 d of age (Fig. 1, I and J). Incontrast, eNOS increased steadily with testicular maturation(Fig. 1, K and L). Taken together, these data illustrate thedifferential response of both NOSs in Sertoli cells, germ cells,and testes during maturation, suggesting that although theseNOSs are performing similar biological function of produc-ing NO, they are regulated differently in the testis.

Changes in the levels of iNOS and eNOS in Sertoli cellcultures during TJ assembly

To assess the participation of NOSs in TJ assembly in vitro,Sertoli cells were cultured at high cell density (0.5 � 106

cells/cm2) for a period up to 7 d, and the steady-state mRNAand protein levels of iNOS (Fig. 2, A and E) and eNOS (Fig.2, B and F) were quantified. The iNOS level stayed highduring the initial 1–2 d in culture and tumbled rapidly there-after (Fig. 2, A and E). This pattern of changes in expressionis similar to eNOS (Fig. 2, B and F vs. A and E), suggestingthe assembly of junctions between Sertoli cells may requireneither NOSs. Yet Sertoli cells cultured at low cell density(2.5 � 104 cells/cm2) where TJ barrier failed to form due tolow cell number, the levels of iNOS and eNOS proteinsremained relatively stable throughout the entire culture pe-riod (Fig. 2C). Figure 2D is the same blot of Fig. 2, A and B(lower panel), but stained with an antiactin antibody illus-trating equal protein loading in this SDS-polyacrylamide gel.

The assembly of functional TJ barrier between Sertoli cellsin vitro

Earlier studies from this laboratory have shown that func-tional Sertoli cell TJ barrier was formed in vitro when Sertolicells were cultured at 0.5–1.2 � 106 cells/cm2 on Matrigel-coated dishes or bicameral units (23). This information wasreached based on physiological analysis, such as restricteddiffusion of [3H]inulin, [125I]BSA (17, 33), and fluoresceinisothiocyanate-labeled dextran (34) from the apical to thebasal compartment of the bicameral unit across the Sertolicell epithelium, and the establishment of an electrical resis-tance barrier (23) as reported herein. We have also examinedthe presence of TJs between Sertoli cells by electron micros-copy as described in Materials and Methods. TJs betweenadjacent Sertoli cells in vitro were shown in Fig. 3A, as de-noted by arrowheads. The basal and apical sides of the Sertolicell epithelium could be distinguished by the presence of themicrovilli (Fig. 3B) on the apical side of Sertoli cells, whichwere not found in the basal side, consistent with results ofearlier reports (for a review, see Ref. 35). The magnifiedinter-Sertoli cell TJ barrier as shown in Fig. 3C (indicated byarrowheads) was found near the basal side of the Sertoli cellepithelium. In parallel to this morphological analysis, theTER across the cell epithelium was quantified to confirm theassembly of an electrical resistance between Sertoli cells invitro (Fig. 3D). For instance, when Sertoli cells were plated onMatrigel-coated bicameral units at 1.2 � 106 cells/cm2 andTER across the cell epithelium was quantified daily, a steadyrise in TER was detected (Fig. 3D), correlating with the for-mation of a functional TJ barrier. The TER reached its peakby d 3 and remained steady when the TJ-barrier was estab-lished (Fig. 3D vs. 3, A–C).

ZnPP, an NOS inhibitor, facilitates the Sertoli cell TJbarrier assembly in vitro

As shown in Fig. 2, a significant decline in both iNOS (Fig.2, A and E) and eNOS (Fig. 2, B and F) was detected at thetime Sertoli cell TJ barrier was assembled in vitro, suggestingthe presence of NO might perturb the TJ barrier. Indeed,ZnPP, an NOS inhibitor, at 1 and 10 �m, when added toSertoli cell cultures during TJ assembly facilitated the TJbarrier assembly dose dependently (Fig. 3D). Earlier studieshave shown that the assembly of Sertoli cell TJ barrier in vitrowas associated with a transient induction of occludin (36), wehad investigated if ZnPP would affect occludin expression bySertoli cells. Consistent with several earlier reports (23, 36),the assembly of the TJ barrier was shown to associate witha transient occludin induction (Fig. 3, E and G), yet thepresence of ZnPP stimulated the expression and protein pro-duction of occludin by Sertoli cells (Fig. 3, F and H).

iNOS and eNOS are structurally linked to the occludin-based TJ multiprotein complexes

Because ZnPP, an NOS inhibitor, can facilitate the Ser-toli cell TJ assembly, suggesting NOS is a putative regu-lator of TJ dynamics in vitro, we sought to investigatewhether NOS physically interacts with the TJ multi-pro-tein complexes. By immunoprecipitation using seminifer-


FIG. 1. A–L, The steady-state mRNA levels of iNOS and eNOS in Sertoli cells, germ cells, and testes during maturation. Sertoli cells wereisolated from 20-, 45-, and 90-d-old rats and cultured at 0.5 � 106 cells/cm2 on Matrigel-coated dishes for at least 3 d with purity ranging between85 and 95% before their termination for RNA extraction. Germ cells were isolated from 20-, 45-, 60-, and 90-d-old rat testes and terminatedwithin 3 h after their isolation. Testes obtained from 10-, 20-, 60-, and 90-d-old rats were used immediately as described in Materials and Methodsfor RNA extraction. The steady-state iNOS mRNA level in Sertoli cells (A and B), germ cells (E and F), and testes (I and J), and the correspondingeNOS mRNA level in Sertoli cells (C and D), germ cells (G and H), and testes (K and L) are shown herein. Results were densitometrically scannedusing autoradiograms, such as those shown in A, C, E, G, I, and K, and were normalized against S-16, and are shown in B, F, and J for iNOSand D, H, and L for eNOS. Testes were obtained from three different rats, and Sertoli and germ cells were terminated from at least three separateculture experiments. Each bar represents a mean � SD of at least three separate experiments. ns, Not significantly different from cells/testesat 10- or 20-d-old rats by Student’s t test; *, significantly different from cells/testes at 10- or 20-d-old rats by Student’s t test (P � 0.05); **,significantly different from cells/testes at 10- or 20-d-old rats by Student’s t test (P � 0.01); nd, not detectable.


ous tubule lysates as the starting material, iNOS and eNOSwere found to structurally link to occludin, a TJ-integralmembrane protein, but not afadin and nectin-3, both ofwhich are adherens junction (AJ)-associated integral mem-brane proteins (Fig. 4). iNOS and eNOS also associatedwith actin, vimentin, and �-tubulin (Fig. 4). Taken collec-tively, these results illustrate iNOS and eNOS interactstructurally and functionally with the occludin/actin TJ,the intermediate filament vimentin-based desmosomesand hemidesmosomes, and the tubulin cytoskeleton.

ZnPP-induced facilitation of the Sertoli cell TJ barrierassembly was mediated apparently via the cGMP, but notthe cAMP, pathway

To further investigate the downstream effector(s) ofNOS in the regulation of the Sertoli cell TJ barrier, the

intracellular cAMP and cGMP levels in the ZnPP-treatedvs. control Sertoli cell cultures were quantified. It wasshown that the assembly of Sertoli cell TJ barrier wasassociated with a transient increase in both intracellularcAMP (Fig. 5A) and cGMP (Fig. 5B) levels at 1–2 d (see Fig.5 vs. Fig. 3D) when TJ was being formed, suggesting lowlevels of these cyclic nucleotides are indeed needed for TJbarrier assembly, consistent with an earlier report (16). Yetthe presence of ZnPP failed to induce any changes inintracellular cAMP level (Fig. 5A). However, ZnPP in-duced a significant reduction in cGMP level (Fig. 5B),which in turn facilitated the TJ barrier assembly (see Fig.3D). Thus, it is likely that ZnPP facilitates the assembly ofthe Sertoli cell TJ barrier as shown in Fig. 3D by plungingthe intracellular cGMP level. This postulate was furtherconfirmed with the study when Sertoli cells were incu-

FIG. 2. A–F, Changes in the levels of steady-state mRNA and protein of iNOS and eNOS in Sertoli cells when cultured at different densitiesin vitro. Cultures were terminated at specified time points for RNA and protein extraction. Semiquantitative RT-PCR and immunoblottingwere performed as described in Materials and Methods. Day 0 indicates cultures terminated within 3 h after cells were plated on culturedishes. Representative autoradiograms (upper panel) and chemiluminigrams (lower panel) are shown in A and B for iNOS and eNOS,respectively. E and F are the corresponding densitometrically scanned data using films, such as those shown in A and B, respectively.C shows that when Sertoli cells were cultured at 2.5 � 104 cells/cm2 without the formation of the Sertoli cell TJs, there is no change inthe level of both NOSs. D is the actin immunoblot representing equal protein loading among different samples using the same samplesshown in A and B. For data analysis, the mRNA level at each time point was normalized against S-16, whereas for protein level, eachtime point was normalized against d 0. Each bar represents a mean � SD of at least three separate experiments. Each time point hadreplicate cultures. ns, Not significantly different by Student’s t test; *, significantly different by Student’s t test (P � 0.05); **, significantlydifferent by Student’s t test (P � 0.01).


FIG. 3. A–H, Morphological analysis on the Sertoli cell tight junction in vitro and the effects of ZnPP, an NOS inhibitor, on the Sertoli cell TJbarrier and the level of occludin in vitro. Sertoli cells were cultured at 0.5 � 106 cells/cm2 on Matrigel-coated dishes to allow the establishmentof the Sertoli cell TJ barrier. On d 5, Sertoli cells were fixed and processed for electron microscopy as described in Materials and Methods. A,This electron micrograph shows the Sertoli cell TJs (indicated by arrowheads) between adjacent Sertoli cells. Magnification, �6750. B, Thisfigure demonstrates the presence of microvilli on the apical side, but not on the basal side, of the Sertoli cell epithelium, which is consistentwith an earlier report (35), showing the typical morphology of Sertoli cells having intact TJs when cultured in vitro. Magnification, �6750. C,This is a magnified view of the TJ barrier between two adjacent Sertoli cells (arrowheads), such as the one shown in (A), which appears as anelectron-dense material near the basal portion of the Sertoli cell epithelium. Magnification, �16,750. L, Lipid droplet; N, Sertoli cell nucleus;V, vacuole. (D) The assembly of functional TJ barrier in Sertoli cells cultured in vitro, cells were cultured at 1.2 � 106 cells/cm2 on Matrigel-coatedbicameral unit or at 0.5 � 106 cells/cm2 on Matrigel-coated 12-well dishes for 5 d to allow the assembly of Sertoli cell TJ barrier in vitro. TERwas quantified across the Sertoli cell epithelium in bicameral units at specified time points to assess the integrity of the TJ barrier. ZnPP at1 and 10 �M were added to both the apical and basal compartment of the bicameral unit on d 1 (indicated by an arrow), which was also presentin subsequent daily replacement media. The autoradiograms (upper panel) and chemiluminigrams (middle panel) showing the RNA and proteinlevel of occludin in control (E) and ZnPP-treated (F) Sertoli cell cultures. The actin immunoblots (lower panel) in E and F demonstrate equalprotein loading among samples at different time points. G and H are the corresponding densitometrically scanned data, using blots such asthose shown in E and F, normalized against S-16 for RNA level or protein level at d 0. Each time point had triplicate cultures in a singleexperiment, and each experiment was repeated at least three times using different patches of Sertoli cells. For TER experiment, each time pointis expressed as a mean � SD. ns, Not significantly different by Student’s t test; *, significantly different by Student’s t test (P � 0.05), **,significantly different by Student’s t test (P � 0.01).


bated with 8-bromo-cGMP, a cell membrane-permeablecGMP analog, at 0.1–1 mm for 24 h during TJ assembly.8-Bromo-cGMP was shown to significantly perturb theSertoli cell TJ barrier dose dependently (Fig. 6A). Becauselow concentrations of cAMP at 4 –20 �m were shown tofacilitate Sertoli cell TJ barrier assembly by enhancing TERacross the Sertoli cell epithelium in vitro (16), we sought toexamine whether cGMP had a similar biphasic effect onthe Sertoli cell TJ barrier. At 4 �m, 8-bromo-cGMP indeedfacilitated the TJ barrier assembly, enhancing the TERacross the cell epithelium (Fig. 6B). In contrast, the pres-ence of KT-5823, a specific PKG inhibitor (see Fig. 8 for the

signaling pathway involving cGMP and PKG), facilitatedthe Sertoli cell TJ barrier in vitro (Fig. 6C). Taken collec-tively, these data illustrate that the Sertoli cell TJ dynamicswas regulated, at least in part, via the NO/sGC/cGMP/PKG pathway. Even though the effects of ZnPP on TJdynamics as shown in Fig. 5 are likely mediated via thecGMP pathway because its presence in Sertoli cell cultureshad no effects on the cAMP level (Fig. 5, A vs. B), it is ofinterest to verify and expand results of earlier studiesillustrating the biphasic effects of cAMP on the Sertoli cellTJ dynamics (16, 27); in particular, results reported in Fig.5A have demonstrated the crucial role of cAMP in the

FIG. 3. Continued.


assembly of Sertoli cell TJs in vitro. Indeed, 8-bromo-cAMPat high concentration, such as 1 mm (but not 0.1 mm), couldsignificantly perturb the Sertoli-TJ barrier (Fig. 6D). Fur-thermore, SQ-22536, an adenylate cyclase inhibitor, facil-itated the Sertoli cell-TJ barrier assembly (Fig. 6E). Col-lectively, these findings indicate that high levels of cAMPcan perturb the Sertoli cell TJ barrier. Although much work

is needed to define how the intricate intracellular levels ofcAMP and cGMP are being regulated, which in turn mod-ulate the dynamics of the Sertoli cell TJ, these results haveclearly demonstrated that the interplay of cAMP andcGMP plays a crucial role in TJ regulation in the testis.

Regulation of iNOS and eNOS by TGF-�1 and TNF-�

Recent studies have shown that both TGF-� (36) and TNF�(37) can perturb the assembly of Sertoli cell TJ barrier in vitro;we thus sought to examine whether TGF-�1 can affect iNOSand eNOS expression in primary Sertoli cell cultures in vitro.When TGF-�1 was added to Sertoli cell cultures (0.5 � 106

cells/cm2) on d 3 at 0.1–3 ng/ml and incubated for 6 h, itsignificantly and dose dependently induced the expressionof iNOS and eNOS (Fig. 7, A and B) by Sertoli cells. Likewise,TNF-� at 0.5–20 ng/ml also significantly induced the Sertolicell steady-state iNOS mRNA level dose dependently, butnot eNOS (Fig. 7, C and D).

A current molecular model of TJ regulation

Collectively, the results presented herein have led us topostulate that the Sertoli cell TJ dynamics are regulated, atleast in part, by NOS, which determines the intracellularlevels of cGMP (Fig. 8). The cGMP level, along with intra-cellular cAMP and Ca2� levels, and phosphorylation anddephosphorylation of target proteins, such as occludin andother TJ-associated signaling molecules (for review, see Ref.11), are likely the major players that modulate the TJ dy-namics in the testis (Fig. 8).

DiscussionIs NOS/NO a crucial regulatory system of Sertoli cell TJdynamics in the testis?

Recent studies have shown that the Sertoli cell TJ dy-namics in the testis are regulated by an array of moleculesand ions, which include proteases, protease inhibitors,cAMP, [Ca2�], cytokines (e.g. TGF-�3, TNF�), protein ki-nases, and protein phosphatases (for a review, see Ref. 11).Data presented herein report yet another potential regu-latory system of TJ dynamics in the testis namely theNOS/NO system via the cGMP and protein kinase G sig-naling pathway. This conclusion was reached based onseveral lines of evidence. First, the assembly of Sertoli cellTJs in vitro was shown to associate with a significant re-duction on the steady-state mRNA and protein levels ofNOS, suggesting that NO, the enzymatic product of NOS,if present at high levels, it can perturb the Sertoli cell TJbarrier. Indeed, the presence of ZnPP, an NOS inhibitor(25), which blocks the production of endogenous NO canfacilitate the Sertoli cell TJ barrier in vitro dose depen-dently. Second, sGC (the intracellular putative NO recep-tor), cGMP (the second messenger), and PKG (the down-stream effector protein of NO) are largely localized to thebasal compartment of the seminiferous epithelium, resid-ing mostly in Sertoli cells (38 – 40) consistent with theirlocalization at the site of the BTB. Such physical intimacybetween the downstream effector molecules of NO and theSertoli cell TJs has further implicated the significance of

FIG. 4. A study by immunoprecipitation to investigate the structuralassociation of iNOS and eNOS with TJ-associated and cytoskeletalproteins. Immunoprecipitation was performed using either an anti-iNOS or eNOS antibody and lysates of the seminiferous tubule asdescribed in Materials and Methods. Immunocomplexes were dena-tured and the associated proteins were examined by immunoblotsafter SDS-PAGE. The resulting immunoblots were probed with ananti-iNOS, eNOS, occludin, actin, vimentin, and �-tubulin, nectin-3,or afadin antibody. It was found that iNOS and eNOS interact struc-turally with occludin, actin, vimentin, and �-tubulin. Immunoblotsstained with either nectin-3 and afadin antibodies indicated that NOSdid not associate with the nectin/afadin protein complex, which alsoserved as negative controls, implicating the specificity of the inter-actions. Testis lysates were used as a positive control to verify thecommercially available antibodies indeed react with the correspond-ing target proteins. An additional negative control was shown in thebottom lane where immunoprecipitation was performed without theuse of a primary antibody, such as anti-eNOS or anti-iNOS, whichfailed to precipitate the iNOS protein. This experiment was repeatedthree times using seminiferous tubules obtained from different rattestes.


NO in TJ dynamics. Third, both iNOS and eNOS wereshown to structurally associate with occludin, a putativeTJ-integral membrane protein largely confined to Sertolicells in the testis (note: occludin also confers cell adhesionfunction at the site of TJs) (for a review, see Ref. 11), andthe underlying actin network. Furthermore, an inhibitionof NOS by ZnPP can stimulate the steady-state mRNA andprotein levels of occludin, suggesting NO produced bySertoli and germ cells may play a role in regulating thehomeostasis of TJ-integral membrane proteins at the siteof the BTB, thereby affecting TJ dynamics. Other studieshave shown that NO is an important intrinsic regulator ofmicrovessel permeability, which may either increase ordecrease the TJ permeability barrier of endothelial cells inblood vessels (for a review, see Ref. 41). For instance, NOSinhibitors, such as Nw-monomethyl-l-arginine and l-Nw-nitro-l-arginine methyl ester, are known to attenuate theincreased microvessel permeability in response to iono-mycin and ATP, that is, making the TJ barrier tighter (42).Needless to say, our results presented herein using ZnPPare consistent with this earlier report illustrating NO canperturb the TJ barrier in endothelial cells in blood vessel,similar to its effects in Sertoli cells, whose action can beblocked by an NOS inhibitor. Furthermore, using mi-crovessel hydraulic conductivity as the means to assess theTJ barrier integrity between endothelial cells in perifusedfrog mesenteric venular microvessels, both NOS inhibi-tors, l-Nw-nitro-l-arginine methyl ester and Nw-mono-methyl-l-arginine, could dose dependently increase thetightness of the endothelial TJ barrier via a calcium-inde-pendent pathway because no changes of intracellular cal-cium concentration ([Ca2�]i) in these vessels were detected(43). NO apparently mediates its effects on TJ barrier func-tion by perturbing the cell adhesion function in the mi-crovessel because NO was shown to inhibit leukocyte-endothelial cell-adhesive interactions (44), suggesting

cross-talks are present between AJs and TJs. Also, NO wasshown to perturb the epithelial TJ barrier function in in-testinal cells in vitro (45). Yet 3-morpholino-sydnonimine,an NO donor, was shown to increase TER (i.e. promotingthe TJ barrier making the junction tighter) (46) across theconfluent retinal pigment epithelial cells isolated from ratretinas and cultured in vitro (note: this TJ barrier confersthe blood-ocular barrier function in vivo), suggesting thatNO can also facilitate TJ assembly. Taken collectively,these results illustrate that the diverging effects of NO onTJ function, whose effect is possibly cell and tissue specificand may be concentration dependent, similar to the effectsof cAMP on Sertoli cell TJ barrier (16). For instance, at 4 –20�m, dibutyryl cAMP facilitates the assembly of Sertoli cellTJ barrier, yet at 100 –500 �m, it inhibits the TJ barrierfunction (16). As reported herein, 8-bromo-cAMP at 0.1mm can also perturb the Sertoli cell TJ barrier, reducing theTER across the cell epithelium by as much as 80%, con-sistent with this earlier report (16). In this context, it isnoteworthy to mention that cytokines can also have abiphasic effects on TJ barrier in different epithelia, eitherfacilitating or inhibiting TJ barrier in vitro (for review, seeRef. 47). Nonetheless, these data illustrate the pivotal roleof the NOS/NO system in the regulation of TJ-permeabil-ity barrier. And the fact that both Sertoli and germ cells areactively expressing mRNAs encoding both iNOS andeNOS suggests that, although the BTB is constituted by theinter-Sertoli cell TJs, germ cells likely contribute to themaking and maintenance of the BTB integrity in the testis.

In this connection, it is of interest to note that, duringmaturation, an age-dependent increase in the Sertolisteady-state mRNAs of iNOS and eNOS is detected, andiNOS and eNOS display an age-dependent decline andinduction in expression by germ cells, respectively. Yet theexpression of iNOS and eNOS plummets and increases inthe testis, respectively, during maturation. Because an in-

FIG. 5. A and B, The intracellular cAMP and cGMP levels in Sertoli cell cultures (0.5 � 106 cells/cm2) with and without ZnPP. Sertoli cells werecultured at 0.5 � 106 cells/cm2 for 5 d to allow the assembly of Sertoli cell TJ barrier in vitro. On d 1, Sertoli cells were treated with ZnPP at10 �M (indicated by arrows). ZnPP was also included in the daily replacement of F12/DMEM, and cultures were terminated at specified timepoints for quantification of intracellular cAMP (A) and cGMP (B). The amount of cAMP and cGMP were expressed as pmol/mg total protein.The experiments were repeated for at least three times using different batches of Sertoli cells and each time point had triplicate cultures. Eachtime point is expressed as a mean � SD. ns, Not significantly different from control by Student’s t test; **, significantly different from controlby Student’s t test (P � 0.01).


duction of testicular eNOS at 60 –90 d of age (note: almosta 10-fold increase vs. immature rat testes at 10 –20 d of age)favors TJ disruption, whereas a decline of iNOS at these

ages presumably facilitates TJ assembly, coinciding withthe timing of rapid spermiogenesis and junction restruc-turing in the seminiferous epithelium at these ages. Based

FIG. 6. A–E, Effects of 8-bromo-cGMP, a cell membrane-permeable cGMP analog, KT-5823, an inhibitor of PKG, 8-bromo-cAMP, a cellmembrane-permeable cAMP analog, and SQ-22536, an adenylate cyclase inhibitor, on the Sertoli cell TJ-permeability barrier in vitro. Sertolicells were cultured at 1.2 � 106 cells/cm2 on Matrigel-coated bicameral unit for 5 d to allow the establishment of Sertoli cell-TJ permeabilitybarrier. Media with different concentrations of 8-bromo-cGMP (A and B), KT-5823 (C), 8-bromo-cAMP (D), or SQ-22536 (E) was added to theSertoli cell cultures on d 1 (indicated by closed arrows). After 24 h, cells were rinsed with fresh media and replaced with media without8-bromo-cGMP (A and B), KT-5823 (C), 8-bromo-cAMP (D), or SQ-22536 (E) (indicated by open arrows). Thereafter, TER across the cellepithelium was measured daily to monitor the integrity of the TJ barrier. Each experiment was repeated at least twice using different batchesof Sertoli cells with triplicate cultures in each experiment. Each time point is expressed as a mean � SD. ns, Not significantly different fromthe corresponding control by Student’s t test; *, significantly different from the corresponding control by Student’s t test (P � 0.05); **,significantly different from the corresponding control by Student’s t test (P � 0.01).


on these data, it is tempting to speculate that these twoNOSs may play a crucial role, alike a yin-yang relation-ship, in regulating the opening and closing of the BTB withiNOS facilitates the closing and eNOS the opening of theBTB. Needless to say, this concept must be vigorouslytested in future experiments because NOS is known toregulate an array of physiological function, including Ser-toli and germ cell apoptosis, sperm motility, and spermmaturation (2–10), and these effects must also be taken intoconsiderations for data interpretation. Nonetheless, thesedata clearly suggest that these two NOSs are regulateddifferentially between Sertoli and germ cells. Yet the in-timate coexistence of both cell types in the seminiferous

epithelium further complicates the issue regarding theirroles in TJ dynamics.

What is the downstream mechanistic pathway used byNOS/NO to affect Sertoli cell TJ dynamics in the testis?

Among the various downstream signaling pathways ofNOS/NO, the two best studied signal transducers of NO arecAMP and cGMP (for reviews, see Refs. 12 and 14). Indeed,recent studies of TJ regulation in multiple epithelia includingSertoli cells have shown that TJ dynamics are regulated bythese cyclic nucleotides (for a review, see Ref. 11). For in-stance, dibutyryl cAMP, an analog of cAMP that is non-cleavable by cAMP phosphodiesterase, was shown to have

FIG. 7. A–D, Changes in the steady-state mRNA levels of iNOS and eNOS in Sertoli cells cultured in vitro after treatment with TGF-�1 orTNF-�. Sertoli cells were cultured at 0.5 � 106 cells/cm2 for 3 d, thereafter TGF-�1 at 0.1–3 ng/ml or TNF-� at 0.5–20 ng/ml were added ontothe Sertoli cell epithelium. After 6 h of incubation, cells were terminated for RNA extraction and for semi-quantitative RT-PCR. Autoradiogramsshowing changes in the iNOS and eNOS expression after TGF-�1 (A and B) and TNF-� (C and D) treatment; B and D are the correspondingdensitometrically scanned data using autoradiograms, such as those shown in A and C. Data were normalized against S-16. Ctrl designatesSertoli cell cultures on d 3 without cytokine treatment and Ctrl/V represents cultures incubated with equivalent amount of vehicle to prepareeither the TFG-�1 or TNF-� stock solution. Each bar represents a mean � SD of at least three experiments and each time point had triplicatecultures in each experiment. ns, Not significantly different from controls by Student’s t test; *, significantly different from control by Student’st test (P � 0.05); **, significantly different from control by Student’s t test (P � 0.01).


a biphasic effect on Sertoli cell TJ barrier in vitro. DibutyrylcAMP at 4–20 �m or 100–500 �m can either promote orperturb the assembly and maintenance of the Sertoli cell TJbarrier in vitro, respectively (16). As such, we have investi-gated if the reported effects of NOS/NO on Sertoli cell TJdynamics are mediated via these cyclic nucleotides (see Fig.8). Interestingly, a transient but drastic increase in the intra-cellular cAMP and cGMP levels in Sertoli cells is detected atthe time the TJ barrier is being formed in vitro. Yet once theTJ barrier is formed, the levels of both cAMP and cGMPtumbled rapidly and reduced by as much as 2-fold, making

them similar to the basal level. These results seemingly sug-gest that, although cGMP and cAMP may not be required forthe maintenance of the Sertoli cell TJ barrier, they are neededfor TJ assembly, supporting the reported biphasic effects ofcAMP on Sertoli cell TJ barrier function as earlier reported(16). Furthermore, the presence of ZnPP, an inhibitor of NOSand an inhibitor of NO-dependent sGC (26) that can effec-tively and significantly inhibit the intracellular cGMP level,can facilitate the Sertoli cell TJ assembly, implicating thesignificance of cGMP in TJ dynamics. Indeed, when Sertolicells were exposed to 8-bromo-cGMP at 0.1–1 mm, a cell

FIG. 8. A schematic drawing that illustrates the potential pathway by which NOS/NO used to affect the Sertoli cell TJ dynamics. This figuredepicts the possible upstream and downstream regulatory pathways used by NOS/NO, shaded in gray, as reported herein to regulate Sertolicell TJ dynamics in the testis. Apart from the effects of NOS/NO on TJ dynamics, NOS/NO also mediate other diverse physiological functionsvia these pathways (for reviews, see Refs. 53–55). NO synthesized by NOS activates AC and sGC. AC and sGC mediate the conversion of ATPto cAMP and GTP to cGMP, respectively. cAMP activates PKA; cGMP activates either PKA, phosphodiesterase (PDE), PKG, and cyclicnucleotide-gated channels (CNG) (for reviews, see Refs. 12, 14, and 56). Collectively, NO mediates its effects using the signaling pathways viacAMP or cGMP to affect junction dynamics. Based on the results of this report and other studies, it is possible that NOS plays a crucial rolein the regulation of TJ dynamics via the NO/sGC/cGMP/PKG pathway. In addition, both iNOS and eNOS interact structurally with occludin.Also, TJ dynamics in the testis are regulated by intracellular cAMP and calcium levels [Ca2�]i, and phosphorylation/dephosphorylation ofTJ-integral membrane and/or signaling proteins as earlier reported (for a review, see Ref. 11).


membrane-permeable cGMP analog, on d 1 for a period of24 h, it perturbed the Sertoli cell TJ barrier dose dependently.Taken collectively, these data suggest that cGMP at muchlower concentrations, such as less than 5 �m (note: assumingcGMP has a similar biphasic effect as of cAMP on TJ dy-namics; see Ref. 16), can possibly promote TJ barrier assem-bly. As reported herein, 8-bromo-cGMP at 4 �m can indeedfacilitate the TJ barrier function. Furthermore, the use ofKT-5823, a PKG inhibitor, can also modulate the Sertoli cellTJ barrier function. It is of interest to note that the effect ofthe PKG inhibitor is comparably smaller than a cGMP ago-nist analog, such as 8-bromo-cGMP, suggesting that cGMPmay exert its effects via other effectors other than PKG asshown in Fig. 8. Yet this issue will require further investi-gation in future studies. In summary, these data clearly il-lustrate that NO is using the sGC/cGMP/PKG pathway tomediate its effects in the regulation of TJ dynamics (see Fig.8). Whereas the downstream effector(s) of PKG remains to beinvestigated, this effect, at least in part, is likely mediated viaTJ-integral membrane proteins, such as occludin, because thepresence of ZnPP that promotes the Sertoli cell TJ barrier,making it tighter, was shown to induce the production andexpression of occludin by Sertoli cells. These results are alsosupported by a recent report demonstrating NO can induceredistribution of occludin and ZO-1, moving them awayfrom the site of TJs to cell cytoplasm, perturbing the TJ barrier(48). Alternatively, the downstream target molecule of PKGcan be actin per se. For instance, a recent study using humancervical epithelial cells cultured in vitro has shown that theNO-induced disruption of the TJ barrier is mediated viachanges in cGMP and PKG (49). And 8-bromo-cGMP wasshown to perturb the TJ barrier in cervical epithelial cells aswell (49). Whereas 8-bromo-cGMP had no effects on the totalcellular actin content, yet it significantly induced the intra-cellular G-actin level disrupting the homeostasis of F- andG-actin levels (49), thereby fragmenting the cytoskeleton.This in turn destabilizes the epithelium and hence increasescell permeability.

What are the upstream signaling transducers that associatewith the NOS/NO system in the testis?

The putative signaling molecules that regulate the NOSfunction in the testis are not entirely known. Yet studiesin other epithelia, such as the blood-ocular barrier, and thepathogenesis of ocular inflammation, such as uveitis, haveimplicated the significance of cytokines in TJ regulation(for a review, see Ref. 50). Furthermore, recent studieshave shown that the presence of either TGF-�3 or TNF-�in Sertoli cells cultured in vitro, each cytokine can dosedependently perturb the TJ barrier function (36, 37). It istherefore logical to speculate that cytokines may mediatetheir effects, at least in part, via the NOS/NO pathway.Indeed, TNF-� and TGF-�1 were shown to activate thesteady-state mRNA levels of iNOS and eNOS as reportedherein, suggesting that the subsequent increase in NOlevel was used to perturb the TJ barrier. This argument isconsistent with the findings that the assembly of the Sertolicell TJ barrier in vitro associates with a plunge in NOS,illustrating an inverse relationship between the assembly

of Sertoli cell TJ barrier and the endogenous levels of NOS.Taken collectively, these data suggest that cytokines, suchas TNF�, that promote the expression of NOS will likelyreduce the production of occludin, the building block ofthe TJ fibrils. Indeed, a recent report has shown that TNF�can significantly inhibit Sertoli cell occludin production atthe time the Sertoli cell TJ barrier is disrupted (37). Thisconclusion is also supported by results of studies usingZnPP (an NOS inhibitor), illustrating this inhibitor canstimulate Sertoli cell occludin expression and production.These data are also consistent with a recent paper report-ing that the disruptive effects of interferon-� and lipo-polysaccharide on the TJ barrier in rat retinal pigmentepithelial cells in vitro could be modulated by using 3-mor-pholino-sydnonimine, an NO donor (46), confirming thefunctional relationship between cytokines and theNOS/NO system.

Does NOS/NO play a role in other junction dynamics otherthan TJ?

Earlier studies by immunohistochemistry have local-ized both eNOS and iNOS to the human and rat seminif-erous epithelium (2, 3, 9). For eNOS, most of the immu-nostaining is associated with Leydig and Sertoli cells, butnot normal germ cells except those undergoing degener-ation and apoptosis, and is not a stage-specific protein (3,9). Yet the pattern of eNOS staining is consistent with itslocalization at the inter-Sertoli TJ at the basal compartmentof the epithelium (3, 9). For iNOS, it is a stage-specificprotein in the rat testis, being highest at stages IX–XII (2).iNOS associates with elongating spermatids (but not elon-gate spermatids), pachytene spermatocytes, Leydig cells,Sertoli cells, and peritubular myoid cells; and its patternof localization in the seminiferous epithelium of the testesis also consistent with its localization at the inter-Sertoli TJ(2, 9). At present, the precise functional significance re-garding the intimate physical association of NOS with TJproteins, such as occludin, is not known. But this doesimplicate the role of NOS in TJ function. Furthermore,these immunohistochemical data also illustrate the asso-ciation of NOS with the site of desmosome-like junctions(cell-cell intermediate filament-based anchoring junctiontype) and adherens junctions (cell-cell actin-based anchor-ing junction type) between Sertoli cells as well as betweenSertoli and spermatids (also known as ectoplasmic spe-cialization, ES) (2, 3, 9). The demonstration of the associ-ation of NOS with vimentin, an intermediate filamentconstituent protein (for a review, see Ref. 11), by immu-noprecipitation as reported herein seemingly suggests thatNO can also regulate anchoring junction dynamics, suchas desmosome. Indeed, PKG, one of the downstream ef-fectors of NOS/NO (for reviews, see Refs. 12 and 14) wasshown to colocalize with vimentin (also a putative sub-strate of PKG) (51) in human neutrophils cultured in vitrowith functional anchoring junctions such as desmosomes,implicating the role of NO in desmosome function (andpossibly hemidesmosome function). A more recent studyby subcellular fractionation of cellular organelles and im-munofluorescent microcopy has shown that eNOS was


associated with platelet endothelial cell adhesion molecule1, a putative AJ-associated cell adhesion molecule, but notvascular endothelial-cadherin and plakoglobin (52). Takencollectively, these data have strengthened the notion thatNOS and its downstream effector proteins are also foundat the site of desmosomes and ES implicating their role incell adhesion function and AJ dynamics.

Acknowledgments

We thank Dr. Meng-yun Mo for his assistance in performing nucle-otide sequence analyses to verify the authenticity of the PCR productsfor S-16, iNOS, eNOS, and occludin. We are also grateful to Ms. EleanaSphicas for her excellent technical assistance in the Bio-Imaging Re-source Center at the Rockefeller University in studies using electronmicroscope. We are also indebted to Dr. Dolores D. Mruk for her helpfuldiscussion and interest throughout the course of this work.

Received December 18, 2002. Accepted March 17, 2003.Address all correspondence and requests for reprints to: C. Yan

Cheng, Ph.D., Population Council, Center for Biomedical Research, 1230York Avenue, New York, New York 10021. E-mail: [email protected].

This work was supported in part by grants from the ContraceptiveResearch and Development Program (Consortium for Industrial Col-laboration in Contraceptive Research CIG-96-05A, CIG-01-72) (toC.Y.C.), the Noopolis Foundation (to C.Y.C.), NIH (National Institute ofChild Health and Human Development, U54-HD-29990 Project 3; toC.Y.C.).

References

1. Alderton WK, Cooper CE, Knowles RG 2001 Nitric oxide synthases: structure,function and inhibition. Biochem J 357:593–615

2. O’Bryan MK, Schlatt S, Gerdprasert O, Phillips DJ, de Kretser DM, HedgerMP 2000 Inducible nitric oxide synthase in the rat testis: evidence for potentialroles in both normal function and inflammation-mediated infertility. BiolReprod 63:1285–1293

3. Zini A, O’Bryan MK, Magid MS, Schlegel PN 1996 Immunohistochemicallocalization of endothelial nitric oxide synthase in human testis, epididy-mis, and vas deferens suggests a possible role for nitric oxide in spermat-ogenesis, sperm maturation, and programmed cell death. Biol Reprod 55:935–941

4. Fujisawa M, Tatsumi N, Fujioka H, Kanzaki M, Okuda Y, ArakawaS, Kamidono S 2000 Nitric oxide production of rat Leydig and Sertolicells is stimulated by round spermatid factor(s). Mol Cell Endocrinol 160:99 –105

5. O’Bryan MK, Zini A, Cheng CY, Schlegel PN 1998 Human sperm endothelialnitric oxide synthase expression: correlation with sperm motility. Fertil Steril70:1143–1147

6. Tatsumi N, Fujisawa M, Kanzaki M, Okuda Y, Okada H, Arakawa S, Ka-midono S 1997 Nitric oxide production by cultured rat Leydig cells. Endo-crinology 138:994–998

7. Stephan JP, Guillemois C, Jegou B, Bauche F 1995 Nitric oxide production bySertoli cells in response to cytokines and lipopolysaccharide. Biochem BiophysRes Commun 213:218–224

8. Lissbrant E, Lofmark U, Collin O, Bergh A 1997 Is nitric oxide involved in theregulation of the rat testicular vasculature? Biol Reprod 56:1221–1227

9. Middendorff R, Muller D, Wichers S, Holstein AF, Davidoff MS 1997 Ev-idence for production and functional activity of nitric oxide in seminiferoustubules and blood vessels of the human testis. J Clin Endocrinol Metab 82:4154–4161

10. Bauche F, Stephan JP, Touzalin AM, Jegou B 1998 In vitro regulation of aninducible-type NO synthase in the rat seminiferous tubule cells. Biol Reprod58:431–438

11. Cheng CY, Mruk DD 2002 Cell junction dynamics in the testis: Sertoli-germcell interactions and male contraceptive development. Physiol Rev 82:825–874

12. Hofmann F, Ammendola A, Schlossmann J 2000 Rising behind NO: cGMP-dependent protein kinases. J Cell Sci 113:1671–1676

13. Michel CC, Curry FE 1999 Microvascular permeability. Physiol Rev 79:703–76114. Lucas KA, Pitari GM, Kazerounian S, Ruiz-Stewart I, Park J, Schulz S,

Chepenik KP, Waldman SA 2000 Guanylyl cyclases and signaling by cyclicGMP. Pharmacol Rev 52:375–413

15. Park JH, Okayama N, Gute D, Krsmanovic A, Battarbee H, Alexander JS 1999Hypoxia/aglycemia increases endothelial permeability: role of second mes-sengers and cytoskeleton. Am J Physiol 277:C1066–C1074

16. Janecki A, Jakubowiak A, Steinberger A 1991 Effects of cyclic AMP andphorbol ester on transepithelial electrical resistance of Sertoli cell monolayersin two-compartment culture. Mol Cell Endocrinol 82:61–69

17. Grima J, Pineau C, Bardin CW, Cheng CY 1992 Rat Sertoli cell clusterin,�2-macroglobulin, and testins: biosynthesis and differential regulation bygerm cells. Mol Cell Endocrinol 89:127–140

18. Lee NPY, Mruk DD, Lee WM, Cheng CY 2003 Is the cadherin/catenin com-plex a functional unit of cell-cell actin-based adherens junctions in the rat testis?Biol Reprod 68:489–508

19. Wright WW, Zabludoff SD, Erickson-Lawrence M, Karzai AW 1989 Germcell-Sertoli cell interactions. Studies of cyclic protein-2 in the seminiferoustubule. Ann NY Acad Sci 564:173–185

20. Galdieri M, Ziparo E, Palombi F, Russo MA, Stefanini M 1981 Pure Sertolicell cultures: a new model for the study of somatic-germ cell interactions. JAndrol 5:249–259

21. Aravindan GR, Pineau CP, Bardin CW, Cheng CY 1996 Ability of trypsin inmimicking germ cell factors that affect Sertoli cell secretory function. J CellPhysiol 168:123–133

22. Mruk DD, Zhu LJ, Silvestrini B, Lee WM, Cheng CY 1997 Interactions ofproteases and protease inhibitors in Sertoli-germ cell cocultures preceding theformation of specialized Sertoli-germ cell junctions in vitro. J Androl 18:612–622

23. Chung NPY, Cheng CY 2001 Is cadmium chloride-induced inter-Sertoli tightjunction permeability barrier disruption a suitable in vitro model to study theevents of junction disassembly during spermatogenesis in the rat testis? En-docrinology 142:1878–1888

24. Brown WJ, Farquhar MG 1989 Immunoperoxidase methods for the localiza-tion of antigens in cultured cells and tissue sections by electron microscopy.Methods Cell Biol 31:553–569

25. Wolff DJ, Naddelman RA, Lubeskie A, Saks DA 1996 Inhibition of nitricoxide synthase isoforms by porphyrins. Arch Biochem Biophys 333:27–34

26. Luo D, Vincent SR 1994 Metalloporphyrins inhibit nitric oxide-dependentcGMP formation in vivo. Eur J Pharmacol 267:263–267

27. Li JCH, Mruk DD, Cheng CY 2001 The inter-Sertoli tight junction permeabilitybarrier is regulated by the interplay of protein phosphatases and kinases: anin vitro study. J Androl 22:847–856

28. El-Mowafy AM, Biggs DF 2001 ETB receptor activates adenylyl cyclase via ac-PLA2-dependent mechanism: a novel counterregulatory mechanism of ET-induced contraction in airway smooth muscle. Biochem Biophys Res Commun286:388–393

29. Burkhardt M, Glazova M, Gambaryan S, Vollkommer T, Butt E, Bader B,Heermeier K, Lincoln TM, Walter U, Palmetshofer A 2000 KT5823 inhibitscGMP-dependent protein kinase activity in vitro but not in intact humanplatelets and rat mesangial cells. J Biol Chem 275:33536–33541

30. Gold DV, Shochat D 1980 A rapid colorimetric assay for the estimation ofmicrogram quantities of DNA. Anal Biochem 105:121–125

31. Bradford MM 1976 A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilizing the principle of protein-dye binding.Anal Biochem 72:248–254

32. Cheng CY, Musto NA, Gunsalus GL, Frick J, Bardin CW 1985 There are twoforms of androgen binding protein in human testes. Comparison of theirprotomeric variants with serum testosterone-estradiol binding globulin. J BiolChem 260:5631–5640

33. Grima J, Cheng CY 2000 Testin induction: the role of cyclic 3�, 5�-adenosinemonophosphate/protein kinase A signaling in the regulation of basal andlonidamine-induced testin expression by rat Sertoli cells. Biol Reprod 63:1648–1660

34. Mruk DD, Siu MKY, Conway AM, Lee NPY, Lau ASN, Cheng CY, Role oftissue inhibitor of metalloproteases-1 in junction dynamics in the testis.J Androl, in press

35. Steinberger A, Jakubowiak A 1993 Sertoli cell culture: historical perspectiveand review of methods. In: Russell LD, Griswold MD, eds. The Sertoli cells.Clearwater, FL: Cache River; 155–179

36. Lui WY, Lee WM, Cheng CY 2001 Transforming growth factor-�3 perturbs theinter-Sertoli tight junction permeability barrier in vitro possibly mediated viaits effects on occludin, zonula occludens-1, and claudin-11. Endocrinology142:1865–1877

37. Siu MKY, Lee WM, Cheng CY 2003 The interplay of collagen IV, tumornecrosis factor-�, gelatinase B (matrix metalloprotease-9), and tissue inhibitorof metalloproteases-1 in the basal lamina regulates Sertoli cell-tight junctiondynamics in the rat testis. Endocrinology 144:371–387

38. Middendorff R, Kumm M, Davidoff MS, Holstein AF, Muller D 2000 Gen-eration of cyclic guanosine monophosphate by heme oxygenases in the humantestis-a regulatory role for carbon monoxide in Sertoli cells? Biol Reprod63:651–657

39. Spruill A, Steiner A 1976 Immunohistochemical localization of cyclic nucle-otides during testicular development. J Cyclic Nucleotide Res 2:225–239

40. Spruill WA, Koide Y, Huang HL, Levine SN, Ong SH, Steiner AL, BeavoJA 1981 Immunocytochemical localization of cyclic guanosine monophos-phate-dependent protein kinase in endocrine tissues. Endocrinology 109:2239 –2248


41. Kubes P 1995 Nitric oxide affects microvascular permeability in the intact andinflamed vasculature. Microcirculation 2:235–244

42. He P, Liu B, Curry FE 1997 Effect of nitric oxide synthase inhibitors onendothelial [Ca2�]i and microvessel permeability. Am J Physiol 272:H176–H185

43. He P, Zeng M, Curry FE 1997 Effect of nitric oxide synthase inhibitors on basalmicrovessel permeability and endothelial cell [Ca2�]i. Am J Physiol 273:H747–H755

44. Kubes P, Suzuki M, Granger DN 1991 Nitric oxide: an endogenous modulatorof leukocyte adhesion. Proc Natl Acad Sci USA 88:4651–4655

45. Unno N, Menconi MJ, Smith M, Aguirre DE, Fink MP 1997 Hyperperme-ability of intestinal epithelial monolayers is induced by NO: effect of lowextracellular pH. Am J Physiol 272:G923–G934

46. Zech JC, Pouvreau I, Cotinet A, Goureau O, Le Varlet B, de Kozak Y 1998Effect of cytokines and nitric oxide on tight junctions in cultured rat retinalpigment epithelium. Invest Ophthalmol Vis Sci 39:1600–1608

47. Walsh SV, Hopkins AM, Nusrat A 2000 Modulation of tight junction structureand function by cytokines. Adv Drug Delivery Rev 41:303–313

48. Cuzzocrea S, Mazzon E, De Sarro A, Caputi AP 2000 Role of free radicals andpoly(ADP-ribose) synthetase in intestinal tight junction permeability. Mol Med6:766–778

49. Gorodeski GI 2000 NO increases permeability of cultured human cervical

epithelia by cGMP-mediated increase in G-actin. Am J Physiol Cell Physiol278:C942–C952

50. De Vos AF, Hoekzema R, Kijlstra A 1992 Cytokines and uveitis, a review. CurrEye Res 11:581–597

51. Wyatt TA, Lincoln TM, Pryzwansky KB 1991 Vimentin is transiently co-localized with and phosphorylated by cyclic GMP-dependent protein kinasein formyl-peptide-stimulated neutrophils. J Biol Chem 266:21274–21280

52. Govers R, Bevers L, De Bree P, Rabelink TJ 2002 Endothelial nitric oxide synthaseactivity is linked to its presence at cell-cell contacts. Biochem J 361:193–201

53. Droge W 2002 Free radicals in the physiological control of cell function. PhysiolRev 82:47–95

54. Davis KL, Martin E, Turko IV, Murad F 2001 Novel effects of nitric oxide.Annu Rev Pharmacol Toxicol 41:203–236

55. Bogdan C 2001 Nitric oxide and the regulation of gene expression. Trends CellBiol 11:66–75

56. Denninger JW, Marletta MA 1999 Guanylate cyclase and the NO/cGMPsignaling pathway. Biochim Biophys Acta 1411:334–350

57. Chan YL, Paz V, Olvera J, Wool IG 1990 The primary structure of rat ribosomalprotein S16. FEBS Lett 263:85–88

58. Nunokawa Y, Ishida N, Tanaka S 1993 Cloning of inducible nitric oxide synthasein rat vascular smooth muscle cells. Biochem Biophys Res Commun 191:89–94


regulation of sertoli cell tight junction dynamics in the ... · dynamics in the testis even though...

Documents