Cell Tissue Res (1998) 294:279±287 � Springer-Verlag 1998

REGULAR ARTICLE

Abdesselam Lablack ´ VØronique BourdonNorah Defamie ´ Catherine Batias ´ Marc MesnilPatrick Fenichel ´ Georges PointisDominique Segretain

Ultrastructural and biochemical evidence for gap junctionand connexin 43 expression in a clonal Sertoli cell line: a potential modelin the study of junctional complex formation

Received: 26 November 1997/Accepted: 26 May 1998

This research was supported in part by ARC grant no. 67 08. A. Lwas supported by the Fondation pour la Recherche MØdicale.

A. Lablack ´ C. Batias ´ D. SegretainGroupe d'Etude des Communications Cellulaires,Histologie-Embryologie, CHU-Paris Ouest FacultØ de MØdecine,45 Rue des St P�res, F-75006 Paris 2, France

V. Bourdon ´ N. Defamie ´ C. Batias ´ P. Fenichel ´ G. Pointis ())Groupe de Recherche sur l'Interaction GamØtique,INSERM CJF 95/04, FacultØ de MØdecine, Avenue de Valombrose,F-06107 Nice Cedex 02, FranceFax: +33 04 93 37 77 37; e-mail: pointis@unice.fr

M. MesnilUnit of Multistage Carcinogenesis,International Agency for Research on Cancer,150 cours A. Thomas, F-69372 Lyon Cedex 08, France

Abstract To clarify the exact role of Sertoli cells in tes-ticular intercellular communications, a murine Sertoli cellline (42GPA9) has recently been established. Electron-microscopy studies indicate that the morphology of theseimmortalized cells strongly resembles that of mouseSertoli cells in vivo with an indentend nucleus, elongatedmitochondria and numerous lysosome-like structures. Ul-trastructure analysis has also revealed that 42GPA9 cellsform gap junctions as demonstrated by the presence ofsmall electron-dense bridges that connect the plasmamembranes of adjacent cells. The gap junction proteinconnexin 43 (Cx43) has been identified in cultured42GPA9 cells by immunofluorescence and Western blotanalysis. No immunostaining is detected in the absenceof apparent intercellular contact. The anti-Cx43 antibodylabels the contacts between 42GPA9 cells at confluency.This specific staining appears as small dots forming iso-lated rows of dots or surrounding the entire cell, suggest-ing that Cx43 is assembled into membrane plaques. Thegap junctional communication capacity of the 42GPA9cell line has been demonstrated by the dye-transfer tech-nique. Exposure of 42GPA9 cells for 24 h to cAMP and12-O-tetradecanoylphorbol-13-acetate greatly reducesthe Cx43 staining at cell-cell contacts and concomitantlyincreases the cytoplasmic staining, suggesting that these

agents alter the trafficking of Cx43 to the plasma mem-brane. Thus, the 42GPA9 line may provide a useful in vit-ro model for studying gap junction communication be-tween Sertoli cells.

Key words Gap junctions ´ Connexin 43 ´ Sertoli cellline ´ Regulation ´ Mouse

Introduction

In the mamalian testis, Sertoli cells are strategically locat-ed to create a specific environment that promotes germcell differentiation. There is now strong evidence thatthe various intercellular junctions of the seminiferous epi-thelium play an essential role in spermatogenesis (Russeland Peterson 1985). Tight junctions, which form the maincomponent of the blood-testis barrier, have been well de-scribed (for a review, see Pelletier and Byers 1992).

In addition to tight junctions, gap junctions are presentin the blood-testis barrier as demonstrated by ultrastruc-tural analysis in rats (Dym and Fawcett 1970; Gilula etal. 1976; Meyer et al. 1977) and mice (Nagano and Susuki1976). These observations have been further substantiatedby the characterization of the constitutive proteins of gapjunctions, the connexins (Cx). One of them, Cx43, hasbeen identified in the rat testis by immunoblot analysis(Kadle et al. 1991). Within the seminiferous epithelium,Cx43 co-localizes with the gap junctions of the Sertolicell junctional blood barrier (Risley et al. 1992; Pelletier1995). Other members of the connexin family, viz. Cx26,Cx32, Cx33 and Cx37, have been also found to be ex-pressed in the testis of several species but their precisedistributions within the seminiferous tubules have notbeen determined (Zhang and Nicholson 1989; Haefligeret al. 1992).

There is now strong evidence that gap junctions playan important role in gametogenesis. Indeed, a lack ofthe Cx37 gene reduces fertility in females (Simon et al.1997). Cx43-knock-out mice exhibit a 50% reduction of

280

germ cells in fetal testes and ovaries (Juneja et al. 1996).However, since the newborns of these mice die because ofa heart defect (Reaume et al. 1995), it is difficult to clarifythe involvement of Cx43 in spermatogenesis. In the testis,gap-junctional communication between adjacent Sertolicells has long been postulated as being neccessary toco-ordinate the activity of adjacent Sertoli cells duringthe first wave of spermatogenesis and during specificstages of the adult seminiferous epithelium (Gilula et al.1976; Russel and Peterson 1985). However, the molecularevents that trigger the formation, maintenance and regula-tion of gap junctions are difficult to examine in the testis,because of complex organization of the seminiferous epi-thelium. Thus, an in vitro model would be a valuable toolfor studying the regulation of gap junctions in Sertolicells. Primary cultures of immature Sertoli cells havebeen used but these cells rapidly lose their functionalcharacteristics, including gap junction communication(Solari and Fritz 1978).

Recently, we have characterized a clonal Sertoli cellline, 42GPA9 (Bourdon et al. 1998), established from ma-ture transgenic mice and expressing the large T antigen ofpolyoma virus (PyLT), a viral oncoprotein with immortal-izing but not transforming properties (Paquis-Flucklingeret al. 1993). The immortalized cells express a number ofSertoli-cell-specific proteins, including transferrin, clus-terin, the ligand of c-Kit and a functional follicle-stimulat-ing hormone (FSH) receptor and are able to form tightjunctions when cultured at confluency. The purpose ofthe present study has been to investigate further theSertoli cell origin of this cell line by ultrastructural anal-ysis and to study the junctional property of this cell lineby evaluating the presence of gap junctions and ofCx43. The effect of agents that activate protein kinaseA (PKA) and protein kinase C (PKC) on Cx43 expressionhas also been investigated in this clonal Sertoli cell line.

Materials and methods

Materials

Dulbecco's modified Eagle's medium (DMEM), Ham's F-12 medi-um, L-glutamine, penicillin/streptomycin, trypsin-EDTA and fetalcalf serum (FCS) were obtained from Gibco-BRL (Cergy Pontoise,France). Insulin, transferrin, testosterone, epidermal growth factor(EGF), Lucifer yellow CH, 8-bromo-cAMP, glutaraldehyde, Eponand Tween 20 were purchased from Sigma (St. Louis, Mo.) and12-O-tetradecanoylphorbol-13-acetate (TPA) from Alexis (San Die-go, Calif.). Immature (14-day-old) male mice were obtained fromIFFA-CREDO (l'Arbresle, France) and maintained under stablelighting conditions (14L:10D).

Cell cultures

The clonal Sertoli cell line 42GPA9 was established from maturePyLT transgenic mice as previously described (Bourdon et al.1998). Cells were maintained in DMEM containing 10% FCS.Sertoli cells from 14-day-old mice were prepared according to amethod previously reported (Fillion et al. 1993) and cultured inDMEM containing insulin (10 �g/ml), transferrin (5 �g/ml) and gen-tamycin (4 �m/ml) at a density of 0.5�106/well in 8-well Labtek

chamber slides (Nunc, Copenhagen, Denmark). Both the Sertoli cellline and freshly isolated Sertoli cells were maintained at 32�C in asaturated atmosphere of 5% CO2. To study the effect of protein ki-nase modulators, 42GPA9 cells were cultured in presence of 0.5 mM8-bromo-cAMP or 10±7 M TPA before being assayed for Cx 43 ex-pression.

Morphological analysis

For electron-microscopic (EM) studies, cells were fixed with 3.5%glutaraldehyde in phosphate-buffered saline (PBS; 0.1 M, pH 7.4)for 1 h at room temperature and then removed from the Petri dishwith a rubber policeman. The cells were centrifuged and the super-natant was discarded, whereas the cell pellet was washed in PBSovernight at 4�C. Cells were impregnated with reduced osmiumfor 1 h at room temperature as previously described (Karnovsky1971). Finally, they were dehydrated and embedded in Epon. Thinsections of the cell pellet were deposited on copper grids and count-erstained with lead citrate. EM observations were made at 60 kVwith a JEOL or a Philips CM 10 (Institut A. Fessard, CNRS, Gifsur Yvette, France). Photographs were taken on the rotating stageof the electron microscope.

Immunological studies

After being washed in PBS, cells were fixed in cold methanol at±20�C for 5 min and then washed with PBS containing 0.1% Tween20. Cryostat sections (6 �m in thickness) of frozen mouse testis em-bedded in OCT compound (Miles, Elkhart, Ind.) were subjected tothe same protocol. Fixed Sertoli cells and testicular cryostat sectionswere incubated with a mouse monoclonal anti-Cx43 antibody (Affi-niti, Mamhead, UK) diluted in PBS (1:100) containing 3% bovineserum albumine (BSA) for 1 h at room temperature in a humidchamber or overnight at 4�C. After a quick rinse in PBS-Tween,the fluorescein-isothiocyanate (FITC)-coupled secondary antibody(Biosys, Compi�gne, France), diluted in PBS with 3% BSA, was de-posited on the slides for 45 min, followed by a final rinse in PBS.Slides were then mounted with Citifluor and examined with a NikonLM equiped with an FITC filter. Photographs were taken with Ko-dak T Max 400 film at 800 ASA. Immunofluorescence controls con-sisted of the secondary antibody fluorochrome conjugate withoutprimary antibody.

Gel electrophoresis

Cultured cells were directly solubilized by adding 0.5 ml sodium do-decyl sulfate (SDS) sample buffer and proteins were resolved bySDS-polyacrylamide gel electrophoresis (SDS-PAGE; 12% acryl-amide) by the method previously described (Laemmli 1970). Aftertransfer from gels onto nitrocellulose membrane (BA 85; Schleicherand Schuell), proteins were immunostained with anti-Cx43 anti-body, by using non-fat dry milk as a non-specific blocking agent.The immunoreactive bands were visualized by the enhanced che-mo-luminescence system (ECL, Amsterham, UK). Rat brain lysate(a gift from Affiniti) was used as a positive control for Cx43 expres-sion.

Estimation of gap junctional communication by dye transfer

The gap junctional communication capacity of the 42GPA9 cell linewas assayed by direct microinjection of a fluorescent dye, Luciferyellow CH, into individual cells and by following the transfer ofthe dye into neighbouring cells as previously described (Mesnil etal. 1994a). The number of fluorescent cells was counted 5 min aftereach microinjection. At least 15 microinjections in three differentculture dishes were performed and the average number of fluores-cent cells was estimated. Texas red dextran was microinjected asa control, since this agent cannot pass through gap junctions.

281

Fig. 1 Electron micrographs ofthe clonal 42GPA9 cell line(A, B) and of adult mouse Ser-toli cells (C). A The nucleus (N)of the clonal 42GPA9 cell line isinvaginated and contains nucle-oli. A large Golgi apparatuscomposed of stacks of saccules(arrow) and a few mitochondria(M) are present in the cyto-plasm. One pole of the cell,lacking organelles, is filled withfilamentous structures (open ar-row). B A portion of the cyto-plasm of the clonal 42GPA9 cellline. Numerous anastomosed ERcisternae are interspersedthroughout the cytoplasm (dou-ble arrowhead). Several lamel-lar dark bodies of a lysosomalnature are also present (arrows).Vesiculo-tubular profiles (openarrow) are located close to theplasma membrane. C The basalportion of an adult mouse Sero-til cell (stage III±IV). Largevacuolated mitochondria (M) fillthe cytoplasm. Numerous lyso-some-like granules (arrows) arescattered between ER profiles(double arrowhead) (Go Golgiapparatus, BL basal lamina).A �12 500; B, C �22 000

282

Results

Ultrastructural analysis of the 42GPA9 cell line

Cells of the 42GPA9 line were elongated or rounded inshape with large nuclei. At the ultrastructural level, thenuclear membrane presented numerous infoldings andseveral nucleoli were observed (Fig. 1A). The culturedcells exhibited signs of polarization as suggested by thepresence of numerous vesicles on one side of the cell cy-toplasm and of developed cytoskeletal components on theopposite side (Fig. 1A, B). In agreement with light-micro-scope observations, the Golgi apparatus was located atone pole of the nucleus. Numerous stacks of sacculescomposed the Golgi region (Fig. 1A) and many small ves-icles filled the Golgi medulla. Small plates of endoplas-mic reticulum (ER) cisternae were interspersed within thisregion and throughout the cytoplasm (Fig. 1A, B). They

also formed small clusters of elongated anastomosed rib-bons. ER profiles appeared in adult mouse Sertoli cells assmall distended vacuoles of various sizes and shapes(Fig. 1C), whereas these cisternae were more continuousin cultured immortalized cells, although distended areaswere also observed (Fig. 1B). Numerous small mitochon-dria with a rod-like structure were scattered in the cyto-plasm (Fig. 1B). The inner mitochondrial cristae of the42GPA9 cell line were composed of parallel sheets,whereas these structures were poorly organized in Sertolicells (compare Fig. 1B, C). The most important internalfeatures of the cell line were the presence of numerous la-mellar dark bodies, which could be lysosomes (Fig. 1B).These granules, which contained membranous residueswith a dark appearance, filled the cytoplasm. In adult Ser-toli cells, the basal cytoplasm contained numerous lyso-some-like structures with a dark content related to thestage of the cycle of the seminiferous tubules (Fig. 1C).These structures are presumably associated with the pha-gocytotic activity of these cells. Finally, both cell types,viz. Sertoli and immortalized cells, contained a well-de-veloped endocytotic apparatus represented by numerousvesiculo-tubular structures and endosomes surroundedby vesicles (Fig. 1B, C).

Fig. 2A±C Electron micrographs of intercellular junctions in the42GPA9 cell line. A Presence of a gap junction (arrow) betweentwo adjacent cells. �56 000. B High-power magnification of thesame area showing small electron-dense bridges (arrow) �285 000.C ER cisternae (arrows) underline adjacent cell contiguity. �80 000

283

EM analysis of gap junctions in the 42GAP9 cell line

Studied at the EM level, classical gap junctions were vis-ible near reduced intercellular spaces (Fig 2A). At highermagnification, the gap junctions were found to be com-posed of small electron-dense bridges that connected theplasma membranes of adjacent cells (Fig. 2B). Two majortypes of junctions could be observed at the EM level: (1)small focal junctions located at the extremity of thin cy-toplasmic invaginations in close contact with neighbour-

ing cells and (2) junctional complexes located along theinterface with neighbouring cells. In this latter membra-nous association, numerous ER cisternae were found inclose proximity to the plasma membrane on one or bothsides of some junctions (Fig. 2C).

Immunohistochemical expression of Cx43 in the clonal42GPA9 cell line, in Sertoli cells in primary cultureand in testis

As shown in Fig. 3A, specific staining for Cx43 markedthe connection of adjacent cells in the clonal Sertoli cellline. This staining appeared as small dots between thecells, forming isolated rows of dots or surrounding the en-tire cell. In most cases, the Cx43 antibody strongly la-belled the contact between cells. Occasionally, a faint im-munofluorescent signal was observed within the cells. Inthe absence of contacts between cells, no specific immu-nostaining was observed (data not shown). The immuno-reactivity for Cx43, localized between adjacent cells, wasreduced in primary culture of Sertoli cells as comparedwith 42GPA9 Sertoli cells (Fig. 3B). A strong signal forCx43 was observed in frozen sections of mouse testis(Fig. 3C). Immunoreactivity was mainly detected betweenadjacent Sertoli cells and formed a thin line that wassometimes anastomosed. Small positive dots were alsopresent near the lumen of the tubule. Some variation inthe intensity of the signal appeared in relation to the stageof the cycle (data not shown).

Western blot analysis

The presence of Cx43 in the clonal 42GPA9 cell line wasconfirmed by Western blot analysis for Cx43. An immu-

Fig. 3 Immunofluorescence staining of the 42GPA9 cell line (A),mouse Sertoli cells in primary culture (B), and adult mouse testissections (C) with an anti-Cx43 antibody. A, B �359; C �264

45

31

Cx43

1 2

Fig. 4 Western blot analysis for Cx43 in the 42GPA9 cell line. Cel-lular proteins were separated by SDS-PAGE and immunoreactedwith an antibody to Cx43. Lane 1 42GPA9 cell lysates, lane 2 lysatefrom rat brain used as a positive control, numbers molecular mass(kDa) based on protein standards (representative of three indepen-dent experiments)

284

Fig. 5A, B Gap junctional communication capacity of the 42GPA9 cell line determined by the dye-transfer method. Lucifer yellow micro-injected into a cell (A, asterisk; phase-contrast microscopy) is tranferred to adjoining cells (B, fluorescence microscopy). �400

Fig. 6A±C Immunofluorescence staining for Cx43 in 42GPA9 cells treated with 8-bromo-cAMP and TPA. Cells were cultured for 24 hwithout (A) or with 0.5 mM 8-bromo-cAMP (B) or 10±7 M TPA (C) before immunostaining. �359

285

noreactive band was observed near the 43±45 kDa posi-tion of the SDS gel in the 42GPA9 cell line lysate(Fig. 4, lane 1). This diffuse immunoreactive band, whichappeared as a doublet, was also present in the rat brain ly-sate used as a positive control (Fig. 4, lane 2).

Estimation of gap junctional communication capacityby dye-transfer

Under our culture conditions, cells of the 42GPA9 linewere capable of intercellular communication via gapjunctions, as demonstrated by the diffusion of Lucifer yel-low from one cell to neighbouring cells (Fig. 5). This ca-pacity to communicate ranged from 8±10 cells per micro-injection and was not attributable to the presence of inter-cytoplasmic bridges since the high molecular weight Tex-as red dextran remained localized inside the microinjectedcells (data not shown).

Effects of 8-bromo-cAMP and TPA on theimmunolocalization of Cx43 in the 42GPA9 cell line

As indicated in Fig. 6, treatment of the 42GPA9 cells with0.5 mM 8-bromo-cAMP for 24 h altered the immunola-belling for Cx43 (Fig. 6B) as compared with cells cul-tured in the absence of the nucleotide (Fig. 6A). The spot-ty reaction located on the cell plasma membrane at theboundary between adjacent cells decreased in 8-bromo-cAMP-treated cells, whereas an intense cytoplasmic im-munoreactivity appeared concomitantly. In 42GPA9 cellstreated for 24 h with 10±7 M TPA, the immunoreactivesignal was mainly cytoplasmic (Fig. 6C), whereas the pe-ripheral labelling decreased in comparison with the con-trol cells (Fig. 6A). In order to analyse further the cyto-plasmic localization of immunoreactivity found in42GPA9-treated cells, a specific antibody against the Gol-gi complex (Jasmin et al. 1989) was used in double-im-munolabelling experiments. A similar location of the im-munoreactive signals was observed with the two antibod-ies used, suggesting that the cytoplasmic Cx43 labellingwas mainly located at the level of the Golgi apparatus (da-ta not shown).

Discussion

There is now strong evidence that testicular cell-cell inter-actions are important in the regulation of mammalianspermatogenesis (for reviews, see Griswold 1988; JØgou1992). This cross-talk between testicular cells may be me-diated via several mechanisms, including paracrine path-ways and direct contact-dependent gap junctional path-ways. Primary culture of Sertoli cells has been largelyused to study one aspect of this intercellular communica-tion mediated through Sertoli cell factors. Although thepathway of communication involving cellular junctionshas been less well studied, this system of communication

via gap junctions seems to play a crucial role in sperma-togenesis. Indeed, it has been reported that, in the maturetestis, gap junctions between Sertoli cells are intermingledwith tight junctions (Dym and Fawcett 1970) and that al-teration of this junctional complex may be associated, insome cases, with abnormal germinal development (JØgou1992; Byers et al. 1993; Meyer et al. 1996).

The present study shows that cells of the clonal Sertolicell line 42GPA9, recently characterized by our laborato-ry (Bourdon et al. 1998), form a high level of junctionalcomplex in vitro at confluency. This junctional complexincludes both tight (Bourdon et al. 1998) and gap (thisstudy) junctions as demonstrated by electron microscopy.In addition, the immunofluorescence studies and Westernblot analysis support these data and implicate Cx43 in thecell-cell communication between adjacent 42GPA9 cells.

The high expression of Cx43 by this cell line, which isderived from adult mouse Sertoli cells, is in agreementwith a recent study suggesting that most Sertoli-Sertoligap junctions in testes include Cx43 (Risley et al.1992). The results of the immunoblot indicate that theCx43 antibody recognizes more than one polypeptide atthe 43-kDa position. As suggested for other tissues andcell lines, these bands may reflect differentially phosphor-ylated forms of Cx43 (Musil et al. 1990). In the immuno-fluorescence studies, the signals are mainly punctate andlocalized to the cell-cell contact areas of the cells, sug-gesting that Cx43 is assembled into membrane plaques.However, in some cells of the 42GPA9 cell line, a diffusespecific signal is found within the cells, indicating thatCx43 can also be located within the cytoplasm. Cells fromthe 42GPA9 cell line continue to divide in vitro and thecell monolayer continuously remodels its morphology.These structural changes are probably the result of exten-sive degradation and continual renewal of the cellularjunction components. Thus, this secondary immunofluo-rescent staining probably reflects different steps of gapjunction formation.

A second interesting aspect of the present data is theobservation that the 42GPA9 cells are communication-competent as demonstrated by their ability to transferLucifer yellow from one cell to another. These resultssuggest that the gap junctions that occur between cellsof the 42GPA9 cell line are probably functional. In addi-tion to Cx43, other connexins designated Cx26, Cx32,Cx33 and Cx37 have been characterized in the testis(Zhang and Nicholson 1989; Haefliger et al. 1992).Whether the communication competence of the 42GPA9cell line is solely dependent upon the presence of Cx43,as described in the present study, or involves the expres-sion of other types of connexin is currently unknown. Thispossibility needs further examination.

Our results also show that Sertoli cells isolated from14-day-old mouse testis and maintained in primary cul-ture for 3 days express lower levels of Cx43 immunoreac-tivity compared with the 42GPA9 Sertoli cells. Similarobservations have been reported for the tight-junction-as-sociated protein ZO-1 (Bourdon et al. 1998). It is wellknown that Sertoli cells rapidly lose their specific differ-

286

entiated functions during primary culture (for a review,see Steinberger and Jakubowiak 1992), including the ex-pression of cell-cell junctions (Solari and Fritz 1978).Thus, the changes in Cx43 immunoreactivity betweenSertoli cells in primary culture and the 42GPA9 Sertolicell line may be a consequence of such an alteration.

Specific changes in Sertoli cell Cx43 expression, relat-ed to the age of animals and the stage of spermatogenesis,have been observed (Risley et al. 1992; Pelletier 1995;Tan et al. 1996). These changes are associated with mod-ifications of both the density of gap junctions and theirelectrical coupling (Eusebi et al. 1983). Although these da-ta provide some information on the physiological eventsthat influence Sertoli-Sertoli cell communication, the mo-lecular and cellular events that trigger junctional complexformation and dynamics are still unknown. The formationof gap junctions is the result of a series of intracellularevents: Cx gene transcription, Cx mRNA translation inthe rough endoplasmic reticulum, Cx trafficking to theGolgi and assembly into hexameric connexons, Cx trans-port to the plasma membrane, insertion of connexons intothe plasma membrane, and assembly of connexons intofunctional gap junctions (Kumar and Gilula 1986). Thesteps that are potential sites of the control of gap junctionsin Sertoli cells and the physiological factors capable of al-tering Sertoli cell communications are not yet defined. Onthe basis of previous data, FSH and cAMP have been pro-posed as controlling Sertoli-Sertoli cell communication(Grassi et al. 1986; Pluciennik et al. 1994). The present re-sults suggest that (1) endocrine and paracrine factors thatactivate PKA or PKC are potential regulators of gap junc-tion communication in the 42GPA9 line and (2) under ourexperimental conditions, cAMP and TPA are able to de-crease the trafficking of Cx43 to the plasma membraneand probably to reduce gap junctional communication.Whether these agents affect transcriptional or translationalevents in the 42GPA9 cell line or only alter the traffickingof Cx43 from the Golgi to gap junctions, as recently re-ported in rat myometrium (Hendrix et al. 1995), needs tobe determined by further studies. Post-translational phos-phorylation of Cx43 is another point for the potential con-trol of gap junction by PKA and PKC (for a review, seeStagg and Fletcher 1990). In addition, it has been shownthat the aberrant localization of Cx43 may be the resultof a different phosphorylation state of Cx43 in other celltypes (Puranam et al. 1993; Mesnil et al. 1994b). Studiesare underway to determine whether a different state ofphosphorylation may (1) explain the cytoplasmic localiza-tion of Cx43 in 42GPA9 cells exposed for 24 h to cAMPand TPA and (2) alter the ability of the cells to communi-cate.

In conclusion, the 42GPA9 cell line, in addition to ex-pressing specific markers of Sertoli cells (Bourdon et al.1998), appears to have the ability to form junctional com-plexes between cells; these complexes comprise tight andgap junctions that mimic the blood-testis barrier. The invitro system described here may provide a useful modelfor the investigation of the complex phenomena that con-trol the proliferation and maturation of germ cells within

the seminiferous tubule via gap junctions and for the iden-tification of the physiological and pharmacological agentscapable of interfering with Sertoli-Sertoli cell communi-cation in the testis.

Acknowledgements The authors wish to express their thanks to B.Deleau (Noesis, Saclay, France) for expert help with the densitomet-ric program analysis, B. Decrossas and N. Parseghian for their tech-nical assistance, F. Carpentier and J.M. Lepecq for photographicwork, and D. Cunningham for revising the English.

References

Bourdon V, Lablack A, Abbe P, Segretain D, Pointis G (1998) Char-acterization of a clonal Sertoli cell line using adult PyLT trans-genic mice. Biol Reprod 58:591±599

Byers S, Pelletier RM, Suarez-Quian C (1993) Sertoli-Sertoli celljunctions and the seminiferous epithelium barrier. In: RusselLD, Griswold MD (eds) The Sertoli cell. Cache River, Clearwa-ter, Fla, pp 431±446

Dym M, Fawcett DW (1970) The blood-testis barrier in rat and thephysiological compartmentation of the seminiferous epithelium.Biol Reprod 3:308±326

Eusebi F, Ziparo E, Fratamico G, Russo M, Stefanini M (1983) In-tercellular communication in rat seminiferous tubules. Dev Biol100:249±255

Fillion C, Malassine A, Tahri-Joutei A, Allevard AM, Bedin M,Gharib C, Hugues JN, Pointis G (1993) Immunoreactive argi-nine vasopressin in the testis: immunocytochemical localizationand testicular content in normal and in experimental cryptorchidmouse. Biol Reprod 48:786±792

Gilula NB, Fawcett DW, Aoki A (1976) The Sertoli cell occludingjunctions and gap junctions in mature and developing mamma-lian testes. Dev Biol 50:142±168

Grassi F, Monaco L, Fratamico G, Dolci S, Iannini E, Conti M, Eu-sebi F, Stefanini M (1986) Putative second messengers affectcell coupling in the seminiferous tubules. Cell Biol Int Rep10:631±639

Griswold MD (1988) Protein secretions of Sertoli cells. Int Rev Cy-tol 100:133±156

Haefliger JA, Bruzzone R, Jenkins NA, Gilbert DJ, Copeland NG,Paul DL (1992) Four novel members of the family of gap junc-tion proteins. J Biol Chem 267:2057±2064

Hendrix EM, Myatt L, Sellers S, Russel PT, Larsen WJ (1995) Ste-roid hormone regulation of rat myometrial gap junction forma-tion: effects on Cx43 levels and trafficking. Biol Reprod52:547±560

Jasmin BJ, Cartaud J, Bornens M, Changeux JP (1989) The Golgiapparatus in chick skeletal muscle: change in its distributionduring endplate development and after denervation. Proc NatlAcad Sci USA 86:7218±7222

JØgou B (1992) The Sertoli cell. Bailli�re's Clin Endocrinol Metab6:273±311

Juneja SC, Enders GC, Kidder GM (1996) Developmental defects inmouse gonads resulting from connexin 43 null mutation (ab-stract). Biol Reprod 54 [Suppl 1]:195

Kadle R, Zhang JT, Nicholson BJ (1991) Tissue specific distributionof differentially phosphorylated forms of Cx43. Mol Cell Biol11:363±369

Karnovsky MJ (1971) Use of ferrocyanide reduced osmium tetrox-ide in electron microscopy (abstract). J Cell Biol 51:284

Kumar NM, Gilula NB (1986) The gap junction communicationchannel. Cell 84:381±388

Laemmli UK (1970) Cleavage of structural proteins during the as-sembly of the head of bacteriophage T4. Nature 227:680±685

Mesnil M, Asamoto M, Piccoli C, Yamasaki H (1994a) Possible mo-lecular mechanism of loss of homologous and heterologous gapjunctional intercellular communication in rat liver epithelial celllines. Cell Adhes Commun 2:377±384

287

Mesnil M, Rideout D, Kumar NM, Gilula NB (1994b) Non-commu-nicating human and murine carcinoma cells produce a1 gapjunction mRNA. Carcinogenesis 15:1541±1547

Meyer JM, Mezrahid P, Vignon F, Chabrier G, Reiss D, Rumpler Y(1996) Sertoli cell barrier dysfunction and spermatogenic cyclebreakdown in the human testis: a lanthanum tracer investigation.Int J Androl 19:190±198

Meyer R, Posalky Z, McGinley D (1977) Intercellular junction de-velopment in maturing rat seminiferous tubules. J UltrastructRes 61:271±283

Musil LS, Cunningham BA, Edelman GM, Goodenough DA (1990)Differential phosphorylation of the gap junction protein connex-in 43 in junctional communication-competent and -deficient celllines. J Cell Biol 111:2077±2088

Nagano T, Susuki F (1976) The postnatal development of junctionalcomplexes of the mouse Sertoli cells as revealed by freeze-frac-ture. Anat Rec 185:403±418

Paquis-Flucklinger V, Michiels JF, Vidal F, Alquier C, Pointis G,Bourdon V, Cuzin F, Rassoulzadegan M (1993) Expression intransgenic mice of the large T antigen of polyoma virus inducesSertoli cell tumors and allows the establishment of differentiatedcell lines. Oncogene 8:2087±2094

Pelletier RM (1995) The distribution of connexin 43 is associatedwith the germ cell differentiation and with the modulation ofthe Sertoli cell junctional barrier in continual (guinea pig) andseasonal breeder (mink) testes. J Androl 16:400±409

Pelletier RM, Byers SW (1992) The blood testis barrier and Sertolicell junctions: structural considerations. Microsc Res Tech20:3±33

Pluciennik F, Joffre M, DØl�ze J (1994) Follicle-stimulating hor-mone increases gap junction communication in Sertoli cellsfrom immature rat testis in primary culture. J Membrane Biol139:81±96

Puranam KL, Laird DW, Revel JP (1993) Trapping an intermediateform of connexin 43 in the Golgi. Exp Cell Res 206:85±92

Reaume AG, Sousa PA de, Kulkarni S, Langille LB, Zhu D, DaviesTC, Juneja SC, Kidder GM, Rossant J (1995) Cardiac malforma-tion in neonatal mice lacking connexin 43. Science 267:1831±1834

Risley MS, Tan IP, Roy C, Saez JC (1992) Cell-, age- and stage-de-pendent distribution of connexin 43 gap junctions in testes. JCell Sci 103:81±96

Russel LD, Peterson RN (1985) Sertoli cell junctions: morphologi-cal and functional correlates. Int Rev Cytol 94:177±211

Simon AM, Goodenough DH, Li E, Paul DL (1997) Female infertil-ity in mice lacking connexin 37. Nature 385:525±529

Solari AJ, Fritz IB (1978) The ultrastructure of immature Sertolicells. Maturation-like changes during culture and maintenanceof mitotic potentiality. Biol Reprod 18:329±345

Stagg RB, Fletcher WH (1990) The hormone-induced regulation ofcontact-dependent cell-cell communication by phsophorylation.Endocr Rev 11:302±325

Steinberger A, Jakubowiak A (1992) Sertoli cell culture: historicalperspective and review of methods. In: Russel LD, GriswoldMD (eds) The Sertoli cell. Cache River, Clearwater, Fla, pp155±179

Tan IP, Roy C, Saez JC, Saez CG, Paul DL, Risley MS (1996) Reg-ulated assembly of connexin 33 and connexin 43 into rat Sertolicell gap junctions. Biol Reprod 54:1300±1310

Zhang JT, Nicholson EJ (1989) Sequence and tissue distribution of asecond protein of hepatic gap junctions, Cx26, as deduced fromits cDNA. J Cell Biol 109:3391±3401