Antagonistic regulation of tight junction dynamics by glucocorticoids and transforming growth factor-beta in mouse mammary epithelial cells.

The synthetic glucocorticoid, dexamethasone, stimulated the transepithelial electrical resistance and suppressed the DNA synthesis of 31EG4 nontransformed mouse mammary epithelial cells. The addition of transforming growth factor-beta 1 (TGF-beta) to mammary cells simultaneously with or up to 24 h after dexamethasone treatment prevented the steroid induction of transepithelial electrical resistance and stimulated the incorporation of [3H]thymidine. However, the TGF-beta inhibition of tight junction formation did not require de novo DNA synthesis. Confocal microscopy revealed that the organized immunostaining pattern of the tight junction protein, ZO-1, and F-actin at the cell periphery was disrupted by TGF-beta, resulting in disorganized and diffuse staining patterns throughout the cell. Western blot analysis demonstrated that TGF-beta did not alter the protein levels of ZO-1. In contrast to cells not treated or pretreated with steroid for up to 24 h, TGF-beta had no effect on cells pretreated with dexamethasone for 48 h. Transfection of chimeric reporter genes containing promoters responsive to either glucocorticoid or TGF-beta demonstrated that the mutual antagonism of tight junction dynamics by dexamethasone and TGF-beta occurs in the presence of intact signaling pathways. Taken together, our results establish for the first time that glucocorticoids and TGF-beta can antagonistically regulate tight junction formation in a nontransformed mammary cell line.

The synthetic glucocorticoid, dexamethasone, stimulated the transepithelial electrical resistance and suppressed the DNA synthesis of 31EG4 nontransformed mouse mammary epithelial cells. The addition of transforming growth factor-␤ 1 (TGF-␤) to mammary cells simultaneously with or up to 24 h after dexamethasone treatment prevented the steroid induction of transepithelial electrical resistance and stimulated the incorporation of [ 3 H]thymidine. However, the TGF-␤ inhibition of tight junction formation did not require de novo DNA synthesis. Confocal microscopy revealed that the organized immunostaining pattern of the tight junction protein, ZO-1, and F-actin at the cell periphery was disrupted by TGF-␤, resulting in disorganized and diffuse staining patterns throughout the cell. Western blot analysis demonstrated that TGF-␤ did not alter the protein levels of ZO-1. In contrast to cells not treated or pretreated with steroid for up to 24 h, TGF-␤ had no effect on cells pretreated with dexamethasone for 48 h. Transfection of chimeric reporter genes containing promoters responsive to either glucocorticoid or TGF-␤ demonstrated that the mutual antagonism of tight junction dynamics by dexamethasone and TGF-␤ occurs in the presence of intact signaling pathways. Taken together, our results establish for the first time that glucocorticoids and TGF-␤ can antagonistically regulate tight junction formation in a nontransformed mammary cell line.
The precise regulation of cell-cell interactions is an essential feature of the development and function of the mammary gland. Three differentiated cell types, stromal, myoepithelial, and epithelial cells, exist in the mammary epithelium and/or mesenchyme, and reciprocal communication between these cell types occurs as part of mammary morphogenesis during postnatal development and puberty, as well as during pregnancy and lactation (1). Radioactive tracer studies and freeze fracture electron microscopy revealed that an increase in tight junction structural organization and a decrease in permeability of mammary epithelium correlates with the differentiation state of the mammary gland during the onset and establishment of lactation (2)(3)(4). Environmental cues such as systemic ovarian steroids and lactogenic hormones, locally acting growth factors, and the extracellular matrix control the normal differentiation and proliferation of mammary cells (5)(6)(7) and collectively have been proposed to regulate the dynamics of cell-cell interactions. Part of the complexity of understanding the hormonal pathways that control mammary cell-cell interactions is the likelihood that combinations of steroids, protein hormones, and growth factors contribute to this process in an additive, synergistic, or antagonistic fashion. Also, cellular targets of hormonal pathways that potentially regulate mammary cell-cell interactions are generally uncharacterized, although our in vitro evidence has shown that glucocorticoids regulate tight junctions in transformed and nontransformed mammary epithelial cells (8 -10).
The tight junction or zonula occludens is the most apical member of a series of intercellular junctions known as the junctional complex with the adherens junction immediately below it and desmosomes further basal. Tight junctions form a continuous seal around the lateral membrane of adjacent cells and have a highly dynamic structure whose permeability, assembly, and/or disassembly can be altered by a variety of cellular and metabolic regulators (11)(12)(13). At intercellular contact points, where the membranes of adjacent cells come into close proximity, tight junctions serve as a selective barrier to the paracellular diffusion of solutes on the basis of size and charge across epithelial and endothelial cell monolayer (14,15). Tight junctions also contribute to the maintenance of cellular polarity by physically defining the border between the apical and basolateral plasma membrane surfaces (16). To date, one transmembrane protein, occludin (17), and five cytoplasmic-residing peripheral membrane proteins, ZO-1 (18), , cingulin (20), the 7H6 antigen (21), and Rab13 (22) have been identified as being localized at tight junctions. Actin, which forms the characteristic perijunctional ring at the underlying adherens junction, has also been shown to associate with the tight junction, suggesting a direct interaction between the cytoskeleton and tight junctions (13,(23)(24)(25)(26). The 220-kDa phosphoprotein ZO-1 (27), which has been shown to specifically bind to the related cytoplasmic protein, ZO-2 (19), and to the cytoskeletal protein, spectrin (28), is the only known cytoplasmic link to the transmembrane protein occludin (29). Moreover, ZO-1 is a member of a family of membrane-associated proteins containing potential SH3 and guanylate kinase domains, the first of which is the lethal(1)discs-large-1 (dlg) tumor suppressor gene located at the septate junction in Drosophila (30). This homology implicates ZO-1 as a potential component of signal transduction cascades through the plasma membrane (31).
Conceivably, the availability and combinatorial actions of systemic steroids and locally acting growth factors, which regulate particular stages of mammary gland growth and differentiation, concomitantly control tight junction dynamics. In vivo studies suggest that one or more of the lactogenic hor-mones, such as prolactin, insulin, and/or glucocorticoids (32,33), may play an important role in establishing and regulating appropriate intercellular contacts. Our recent in vitro studies using cultured 31EG4 mouse mammary epithelial cells demonstrated that glucocorticoids but not the other lactogenic hormones stimulated an increase in monolayer transepithelial electrical resistance and a decrease in the paracellular permeability (10), implicating this steroid as a key regulator of mammary tight junctions. The glucocorticoid-dependent stimulation of 31EG4 tight junctions occurs by a receptor-dependent process requiring normal levels of extracellular calcium (10) and functioning serine/threonine phosphosphorylation-dephosphorylation cascades (9). However, the role of other classes of extracellular signals on tight junction dynamics is unknown, and perhaps other mammogenic factors may concurrently modulate tight junction function and/or structure. One candidate mammogenic factor is transforming growth factor-␤ (TGF-␤), 1 which is produced within morphologically distinct areas of the mammary gland (34,35) and signals through its cognate cell surface serine/threonine kinase receptors. TGF-␤ is a regulator of mammary branching morphogenesis and cell growth (36), yet relatively little is known about its effects on mammary cell-cell interactions (7,37). In this study, we establish that glucocorticoids and TGF-␤ mutually antagonize each other's actions to regulate tight junction dynamics, localization of the ZO-1 tight junction protein, and cell growth in nontransformed mammary epithelial cells. Cell Culture and Transepithelial Electrical Resistance Measurements-31EG4 mammary cells were cultured on permeable filter supports in Dulbecco's modified Eagle's medium/Ham's F-12 supplemented with 2% fetal bovine serum, 50 g/ml gentamicin sulfate, and 5 g/ml insulin with daily changes of medium as described previously (9,10). In appropriate experiments, dexamethasone was added to a final concentration of 1 M, and TGF-␤ was added to a final concentration of 10 ng/ml. In order to inhibit DNA synthesis, cell cultures were exposed to 1 mM hydroxyurea. TER was measured on filter grown cells using the EVOM Epithelial Voltohmmeter (World Precision Instruments) as described previously (8 -10). The EVOM provides an alternating square wave current of Ϯ 20 A through the monolayer to measure the tightness of the tight junction. Daily resistance measurements were taken at room temperature with the electrode after alcohol sterilization. Calculations for ohms⅐cm 2 were made by subtracting a blank filter (159 ⍀) and multiplying by the area of the monolayer (0.49 cm 2 ).

Assay of DNA Synthesis by [ 3 H]Thymidine
Incorporation-DNA synthesis was quantitated by determining the incorporation of [ 3 H]thymidine. Triplicate samples of cells grown on permeable filter supports under the indicated hormonal conditions were incubated with 6 Ci/ml of [ 3 H]thymidine (82.9 Ci/mmol) for 2 h at 37°C in a humid atmosphere of air/CO 2 (95:5). The cells were washed three times with cold 10% trichloroacetic acid, and the filters were placed directly into scintillation vials with 300 l of 0.3 N NaOH to lyse the cells. Radioactivity was quantified on a Beckman LS 1801 liquid scintillation counter.
Immunofluorescence Microscopy-31EG4 mammary cells were plated at 100% confluency on 24-well filters and incubated with the indicated combinations and times of exposure to dexamethasone and/or TGF-␤ for 48 or 72 h. The cell monolayers were washed three times with PBS and then fixed with acetone/methanol (50:50) at Ϫ20°C for 5 min. Cells were allowed to air dry and then washed with TBST-3% NFDM (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.05% Tween 20, and 3% nonfat dry milk). All subsequent incubations with antibodies and washes were performed with this buffer. Cells were incubated with ZO-1 monoclonal antibody R40.76 (1:400 dilution) at room temperature for 1 h and then washed three times. Fluorescein isothiocyanate-conjugated goat IgG anti-rat IgG was added at a 1:100 dilution for 1 h at room temperature and then washed three times. The filters were punched out and mounted on glass slides in 50% glycerol, 50 mM Tris, pH 8 -9, and 0.4% n-propyl gallate. Immunofluorescent images were photographed at 1600 ASA with Kodak Ektachrome 400 ASA film.
Confocal Immunofluorescence Microscopy-For confocal microscopy, filter-grown monolayers were incubated with combinations of dexamethasone and TGF-␤ for 72 h. After three washes, the monolayers were fixed with 1.75% formaldehyde in PBS for 15 min at room temperature. The cells were washed three more times and permeabilized with 0.5% Triton X-100 in PBS for 10 min. Following three washes, cells were then blocked with TBST-3% NFDM. ZO-1 staining was performed as described above. After the ZO-1 immunostaining, cells were stained for F-actin by incubating monolayers with 165 nM rhodamine phalloidin in PBS for 0.5 h. After three washes, filters were mounted on glass slides as described above.
Confocal images were obtained from a Zeiss Axioplan epifluorescence microscope using a Zeiss 40ϫ Plan-Neofluar multi-immersion objective (0.9 numerical aperture) and analyzed with a Bio-Rad MRC 600 system. A split screen of the double label images was obtained with a dual filter set for fluorescein and Texas red. A series of optical sections was collected for each specimen at ascending z levels beginning at the basal surface in 1-m increments. Due to the increased thickness of the cell monolayer, the number of optical sections collected in the presence of TGF-␤ was approximately twice the amount of that collected from monolayers cultured in the absence of TGF-␤. In Fig. 2, only the most apical regions of ZO-1 are exhibited in order to investigate whether actin can colocalize with ZO-1 at the apical level. After contrast and zoom enhancements, images were assigned pseudocolors representing the original color of the label used for immunofluorescence and printed with a Kodak Colorease PS printer.
Western Blot Analysis for ZO-1 and E-Cadherin Protein Production-Cell lysates were subjected to 6% SDS-polyacrylamide gel electrophoresis and then transferred to nitrocellulose membranes (Micron Separations Inc., Westboro, MA). The membranes were blocked overnight at 4°C with TBST-5% NFDM and incubated with a mixture of primary rat monoclonal antibodies directed against ZO-1, R40.76 (1:1000), and primary polyclonal E-cadherin antibodies (1:1600) in TBST-1% NFDM overnight at 4°C; secondary antibody directed against rat IgG conjugated to HRPO (Cappel) was diluted 1:10,000 in TBST-1% NFDM and incubated for 1 h. The signal was detected by enhanced chemiluminescence on Hyperfilm-ECL from Amersham Corp. Parallel cell samples were electrophoretically fractionated, and equivalent protein loading was demonstrated by Coomassie Blue staining of the protein gel.
Transfection Procedures-31EG4 mammary cells from a logarithmically growing culture were transfected by electroporation. Briefly, cells were harvested with trypsin-EDTA, washed twice with sterile Ca 2ϩand Mg 2ϩ -free PBS, and resuspended in sucrose buffer containing 270 mM sucrose, 7 mM sodium phosphate buffer, pH 7.4, and 1 mM MgCl 2 . 250 l of cell suspension and plasmid DNA (15-25 g total including 10 -15 g of either GRE-CAT or p3TP-lux) were mixed, and 5 electric pulses (400 V square wave pulse for 99 s) were delivered to the sample using a BTX 800 Transfector apparatus (BTX Inc., San Diego, CA). The samples were incubated for 10 min on ice and cultured at 37°C. 16 h after transfection, the cells were treated for the indicated times with TGF-␤ and/or dexamethasone in fresh media and harvested for CAT or luciferase assays. Cells harvested for CAT assays were washed three times in PBS, resuspended in 0.1 M Tris-HCl, pH 7.8, and prepared by four cycles of freeze-thawing. The cell lysates were heated at 68°C for 15 min and centrifuged at 12,000 ϫ g for 10 min, and the supernatants were recovered. Cells for the luciferase assay were washed three times with PBS and lysed with 1 ml of reporter lysis buffer (Promega) according to the manufacturer's instructions.
CAT and Luciferase Assays-CAT activity in the cell extracts con-taining 30 -50 g of lysate protein was measured by the nonchromatographic assay of Neumann and co-workers (38).

TGF-␤ Reversibly Inhibits Glucocorticoid Stimulation in
Transepithelial Electrical Resistance of 31EG4 Mammary Epithelial Cells-We have previously shown that the synthetic glucocorticoid, dexamethasone, stimulates formation of tight junctions in cultured 31EG4 cells by measuring an increase in TER and a decrease in apical to basolateral movement of small radiolabeled tracers (10). To test the effects of TGF-␤ on glucocorticoid stimulated tight junction function, confluent monolayers of mammary cells were grown on permeable supports and treated with or without 1 M dexamethasone (Dex) in the presence or the absence of 10 ng/ml TGF-␤. After an approximately 24-h time lag, dexamethasone induced a significant increase in monolayer TER, whereas in untreated cells, the basal TER remained low (Fig. 1). When added simultaneously with dexamethasone, TGF-␤ completely repressed the ability of glucocorticoids to stimulate TER (Fig. 1). The apical to basal leakage of [ 14 C]mannitol was significantly reduced in dexamethasone-treated cells compared with cells treated with steroid and TGF-␤ (data not shown), confirming that the failure to increase TER in the presence of TGF-␤ correlated with a decrease in paracellular permeability through the monolayer.
Consistent with its mode of action on mammary gland differentiation (35), the inhibitory effects of TGF-␤ on steroidregulated tight junction formation in 31EG4 mammary cells were completely reversible. As also shown in Fig. 1, TGF-␤ withdrawal from cells treated for 48 h with both dexamethasone and TGF-␤ led to a rapid stimulation in TER after a 24 h time lag. The observed time course of tight junction formation after TGF-␤ withdrawal was similar to that in cells initially treated with steroid alone, suggesting that TGF-␤ may be preventing tight junction function at the earlier steps in the glucocorticoid signal transduction pathway. The reversible nature of the TGF-␤-mediated inhibition of tight junction formation demonstrates that TGF-␤ does not indirectly prevent a stimulation in TER as a result of cytotoxic effects on the mammary cells.

TGF-␤ Disrupts Cellular Distribution of ZO-1 and Alters Cell
Morphology-To further investigate the mechanism of hormone-regulated tight junction dynamics, we tested whether TGF-␤ can alter the characteristic tight junction staining and/or expression of ZO-1. As part of this study, F-actin localization was also evaluated because the actin-based cytoskeleton is essential for tight junction assembly and maintenance (24). 31EG4 cells were treated with combinations of dexamethasone and TGF-␤ for 72 h and doubly stained with primary ZO-1 monoclonal antibodies and fluorescein isothiocyanate-conjugated goat anti-rat IgG as well as with rhodaminephalloidin for F-actin. Confocal microscopy was utilized to investigate the spatial relationship between ZO-1 and actin under conditions where tight junctions are tightly sealed by dexamethasone and disrupted by TGF-␤.
To precisely compare the distribution of ZO-1 with that of actin, optical sections were taken in the x-y plane of ZO-1 and pseudocolors were assigned to ZO-1 as green and actin as red. The images shown in Fig. 2 illustrate sections at the apical plane of the tight junction (top panels) and the plane 2 m lower (bottom panels). Cells that were not treated with steroid or growth factor (Fig. 2, a and b) displayed ZO-1 and actin colocalization as a fine ring at the cell periphery at the most apical cell-cell border. Consistent with our previous results (9), cells grown in the presence of dexamethasone were larger and had a greater linear junction length per cell (Fig. 2, c and d  versus a and b). In dexamethasone-treated cells, ZO-1 and actin colocalizes in the lower plane (Fig. 2, d), but actin signal is significantly decreased in the upper plane (Fig. 2, c). Therefore, under conditions where TER is greatly enhanced by dexamethasone, actin appears to be absent from the most apical region of ZO-1 localization.
TGF-␤ drastically disrupted the overall pattern of ZO-1 and actin staining in glucocorticoid-treated (Fig. 2, e and f) and untreated (Fig. 2, g and h) cells, causing a redistribution from points of cell-cell contacts to a more broad and cytoplasmic staining. Immunostaining at the upper plane of TGF-␤-treated cells revealed a complete loss of ZO-1 junctional staining with both ZO-1 and actin displaying a dispersed colocalization in this region (Fig. 2, e versus a). At the more basolateral plane of TGF-␤-treated cells (Fig. 2, f), some ZO-1 is seen to localize at the cell periphery, but the staining lacks the complete tight rings typical of junctional associated plaque proteins. When the mammary cells were cultured in the presence of both dexa- Throughout the 120-h time course, the TER was determined at the indicated times, and the ohms⅐cm 2 were calculated as described in the text. TER measurements were performed in triplicate, and the results are the averages of three separate experiments. methasone and TGF-␤ (Fig. 2, g and h), there is slightly more specific junctional staining of ZO-1 and actin. In the upper left portion, ZO-1 and actin staining is detected only in the apical plane (Fig. 2, g), indicating that certain areas of cells are raised higher than others. ZO-1 and actin staining in the mammary cells treated with glucocorticoids and TGF-␤ appear to be heterogeneous in that although most cells have no tight perijunctional staining, some cells have both specific junctional staining as well as diffuse cytoplasmic staining and still others have only the characteristic cell peripheral staining. Similar to cells treated with TGF-␤ alone, actin colocalizes with ZO-1 even in the most apical plane (Fig. 2, g). Thus, under conditions in which TGF-␤ prevents dexamethasone from stimulating TER, a remodeling of cell morphology occurs with a cellular redistribution of junctional associated proteins.
TGF-␤ Does Not Regulate ZO-1 Protein Levels-To determine whether TGF-␤ treatment alters the level of ZO-1 protein expression, 31EG4 mammary cells were cultured on permeable supports and treated in the presence or the absence of the indicated combinations of dexamethasone and/or TGF-␤ for 72 h. Western blots of electrophoretically fractionated whole cell extracts were incubated simultaneously with monoclonal antibodies to ZO-1 and the adherens junction protein, E-cadherin. As shown in Fig. 3, TGF-␤ treatment did not affect either the basal levels or the dexamethasone-induced expression of ZO-1 previously characterized in these cells (9). Approximately equivalent gel loading was shown by similar amounts of Ecadherin in each sample. Importantly, under all hormonal conditions, the monoclonal antibodies to ZO-1 recognized a single protein band of approximately 220-kDa proteins. Thus, under conditions in which the transepithelial electrical resistance fails to be induced by glucocorticoids, the pattern of ZO-1 distribution in mammary cells but not its expression is significantly altered by exposure to TGF-␤.
Pretreatment with Dexamethasone Precludes TGF-␤ from Disrupting Tight Junction Integrity or Altering ZO-1 Distribution-When added simultaneously with dexamethasone, TGF-␤ precludes glucocorticoids from inducing the sealing of tight junctions. To determine whether TGF-␤ can disrupt tight junctions once they are well sealed due to glucocorticoid treatment, TGF-␤ was added for 24 h to 24 or 48 h dexamethasonepretreated cells, and the monolayers were assayed for changes in TER or ZO-1 localization. As previously mentioned, dexamethasone rapidly induces the TER of 31EG4 mammary cells after an initial time lag of 24 h. As shown in Fig. 4 Cell lysates were normalized for total protein (that was confirmed by Coomassie Blue staining of a parallel polyacrylamide gel), electrophoretically fractionated, and blotted onto nitrocellulose filters. The nitrocellulose blot was incubated simultaneously with ZO-1 and Ecadherin primary antibodies, and protein signals were detected by chemiluminescence after incubation of secondary antibodies as described in the text. The protein molecular mass standards are myosin (200 kDa) and phosphorylase b (97.4 kDa). B, cell extracts from dexamethasone-treated (Dex) and untreated (No Addition) 31EG4 mammary epithelial cells as well as NIH 3T3 fibroblasts (NIH-3T3) were electrophoretically fractionated, and nitrocellulose blots were analyzed for cytokeratin type II protein as described in the text. The molecular mass markers are bovine serum albumin (66.2 kDa) and ovalbumin (45 kDa). methasone-treated cells after the stimulation in TER had no effect on the maintenance of well established tight junctions.
Indirect immunofluorescence microscopy revealed that the addition of TGF-␤ to steroid-treated cells with an induced TER had no effect on the cellular distribution of ZO-1 protein. TGF-␤ was added to mammary cells at various times (0, 24, or 48 h or no addition) during exposure to 1 M dexamethasone and then incubated for up to 48 or 72 total hours. A control set of cultures were treated with TGF-␤ for the same time points in the absence of steroid. Under conditions in which the monolayer TER is stimulated by dexamethasone for 72 h (no TGF-␤ or TGF-␤ treatment 48 -72 h), ZO-1 protein was localized to the tight junction as a sharp continuous band of immunostaining at the cell periphery (Fig. 5). The 48 h dexamethasone-treated and untreated cells displayed the same peripheral ZO-1 staining as mammary cells exposed to steroid for 72 h. In contrast, when TGF-␤ was added to 24 h steroid-treated cells (TGF-␤ 24 -48 h) or simultaneously with dexamethasone (0 -72 h), TGF-␤ caused a disruption of the overall ZO-1 staining pattern with most of the immunostaining being diffused throughout the cell (Fig. 5). A similar TGF-␤ effect on ZO-1 immunostaining was also observed in the absence of glucocorticoids (Fig. 5). Taken together, these observations demonstrate that glucocorticoid or TGF-␤ receptor signaling pathways can regulate tight junction dynamics of 31EG4 cells.

Mutual Antagonism of Tight Junction Integrity by Dexamethasone and TGF-␤ Is Not Due to Inhibition of Receptor
Signaling-One possible mechanism through which dexamethasone and TGF-␤ treatment can mutually exclude the actions of the other signal is by directly inhibiting receptor function. To test this possibility, 31EG4 mammary cells were incubated for 72 h with or without combinations of dexamethasone and TGF-␤ after transfection with reporter genes containing promoter elements responsive to either glucocorticoid receptor function (GRE-CAT) or TGF-␤ receptor signaling (p3TP-Lux). GRE-CAT contains six glucocorticoid response elements and the minimal promoter of the alcohol dehydrogenase gene linked to the bacterial CAT reporter gene. p3TP-Lux, which has been shown to be TGF-␤-responsive (39), contains three consecutive TPA response elements and a portion of the plasminogen activator inhibitor gene promoter linked to the luciferase gene. As shown in Fig. 6 (top panel) A control culture was maintained without the added steroid (No Addition). At 0, 24, or 48 h of dexamethasone treatment (large arrows), the cells were exposed to 10 ng/ml TGF-␤, and the incubations were continued for up to 48 or 72 h. The monolayer transepithelial electrical resistance was determined at the indicated times, and the ohms⅐cm 2 were calculated as described in the text. The TER measurements were performed in triplicate, and the results are the averages of three separate experiments. dexamethasone and TGF-␤ or in cells treated with TGF-␤ during the last 24 h of a 72-h dexamethasone treatment (Fig. 6,  lower panel). Taken together, these data show that the glucocorticoid receptor and the TGF-␤ receptor are both functional in the presence of both hormones and suggest that the mutual antagonism regulating tight junction dynamics are specific postreceptor events.
TGF-␤ Stimulates DNA Synthesis of 31EG4 Mammary Epithelial Cells-Our previous results with both nontransformed and transformed rodent mammary cells showed that the glucocorticoid stimulation of tight junction formation is a differentiated property that is accompanied by and perhaps related to an inhibition of cell proliferation (9). This observation suggests that TGF-␤ might reverse or prevent the growth suppression effects of glucocorticoids under conditions in which the formation of tight junctions is inhibited. Depending on the cell type, TGF-␤ can either suppress or stimulate cell proliferation. TGF-␤ has been shown generally to inhibit or not effect epithelial cell growth (40,41), although several studies have documented growth stimulatory effects of TGF-␤ on a small number of epithelial-derived cells (42,43). Conflicting evidence exists on the growth effects of TGF-␤ on mammary epithelial cells because in certain mammary epithelial tumors, TGF-␤ production correlates with tumor progression and growth (44,45), whereas in many mammary tumor cell lines, TGF-␤ is growth inhibitory (46,47). 31EG4 mammary cells are epithelial in nature because they express the cytokeratin type II epithelial cell marker protein that is not produced in murine fibroblasts (Fig. 3B). We therefore investigated TGF-␤ effects on the growth of 31EG4 cells cultured on permeable supports in the presence of combinations of dexamethasone and TGF-␤ during 48-or 72-h time courses. As shown in Fig. 7, dexamethasone significantly inhibited the incorporation of To determine whether the disruption of glucocorticoid-induced tight junction formation by TGF-␤ requires the stimulation of [ 3 H]thymidine incorporation, the effects of TGF-␤ on tight junction dynamics was monitored in the presence or the absence of a DNA synthesis inhibitor, hydroxyurea. After a 24-h dexamethasone treatment, 31EG4 cells were incubated with combinations of TGF-␤ and/or hydroxyurea for an additional 24 h and assayed for monolayer TER and incorporation of [ 3 H]thymidine. The control set of cultures had no additions of hormone or metabolic inhibitor. As shown in Fig. 8, TGF-␤ inhibited the dexamethasone-stimulated monolayer TER in the presence or the absence of hydroxyurea. Dexamethasone induced a significant increase in monolayer TER in the presence of hydroxyurea, although with a somewhat reduced response. Under the conditions of this experiment, hydroxyurea abolished the incorporation of [ 3 H]thymidine in the presence or the absence of TGF-␤, demonstrating that this metabolic agent effectively inhibited DNA synthesis in cells cultured on permeable supports. These results demonstrate that dexamethasone induction of and TGF-␤ inhibition of tight junction formation do not require de novo DNA synthesis and, because hydroxyurea acts at or near the G 1 /S border (48), cell cycle progression into and past the S phase is not needed in order for TGF-␤ to disrupt the function of intercellular junctional complexes.

DISCUSSION
Our results using a nontransformed mammary epithelial cell line represent the first evidence that glucocorticoids, an important systemic lactogenic steroid (49), and TGF-␤, a locally produced factor, can antagonistically regulate tight junction dynamics. This observation implicates these two distinct extracellular signals as playing important roles in controlling cell-cell interactions in vivo. The mammary gland undergoes a progression of morphological and functional changes during pregnancy, lactation, and involution, including temporal regulation of tight junction permeability. For instance, colostrum, the milk secreted by the mammary gland a few days before or after parturition, contains more protein, immunoglobulins, so- dium, and chloride and less potassium and lactose than the milk secreted during established lactation (2). These differences indicate an alteration in paracellular movement that is principally due to the state of the tight junction. During late pregnancy, freeze fracture studies have shown that the tight junctional network was diffuse with relatively scant ridges, depicting a leaky epithelia, whereas the network at 1 day post-partum and the duration of lactation were condensed with abundant ridges between the lumen and the intercellular space, representing a tight epithelia (3,50). Thus, at a stage when glucocorticoids function in the lactation process, milk components are strictly secreted into the lumen of the ducts via apically directed secretory pathways, and highly developed tight junctions are required to prevent paracellular leakage of blood and milk components from opposite sides of the mammary epithelium. Tight junction permeability increases concomitantly with a decline in mammary blood flow and milk secretion as part of the involution process after lactation (51,52). It is likely that the local factors responsible for maintaining a relatively unrestricted paracellular pathway is developmentally regulated depending on the stage of differentiation of the mammary gland. In this regard, TGF-␤ 1 transcript levels were detected in all stages (5 week, mature, pregnant) of the mammary gland development except during lactation (35). In addition, TGF-␤ can suppress the synthesis and secretion of milk caseins from mammary explants in pregnant mice (53).
During functional differentiation of the mammary gland, the key hormonal regulators of mammary cell-cell interactions must be selective and reversible in their actions and in some instances, counteract each other depending on the stage of differentiation. Consistent with these biological properties, glucocorticoids and TGF-␤ antagonistically control tight junction dynamics in vitro in a temporally regulated manner. In our studies of mammary cells pretreated with glucocorticoids for 48 h, TGF-␤ failed to reduce the monolayer TER or alter ZO-1 localization. In contrast, the addition of TGF-␤ simultaneously with or up to 24 h after glucocorticoid treatment disrupted the structural organization and function of tight junctions. Activation of glucocorticoid-responsive or TGF-␤-responsive reporter plasmids demonstrated that the mutual antagonism displayed by dexamethasone and TGF-␤ targets signaling pathways that regulate cell-cell interactions and growth control rather than having direct effects on receptor function. The time lag required for TER induction suggests that this process is mediated by steroid-regulated events of about 24 h. Because exposure to TGF-␤ during but not after this lag period precludes the ability of dexamethasone to induce tight junction formation, the timing of glucocorticoid-induced gene expression and/or function is critical for regulating TGF-␤ responsiveness. Given the known transcriptional mechanism of action of glucocorticoid receptors (54,55), it is tempting to consider that expression and/or activity of the immediate early steroid-regulated gene products that initiate the tight junction response can be reversed by TGF-␤ receptor signaling. Conceivably, the delayed dexamethasone-responsive gene products, which are responsible for the increase in TER and development of well sealed tight junctions, are relatively unaffected by exposure to TGF-␤. Alternatively, components of the TGF-␤ receptor signaling pathway that directly or indirectly target the tight junction may not be functional or adequately expressed in glucocorticoid-treated mammary cells. Identification of the signaling components that regulate tight junction assembly, disassembly, and integrity will be important to clarify the precise mechanism of hormonal control.
Dexamethasone coordinately induced tight junction formation and suppressed DNA synthesis, whereas the TGF-␤ disruption of ZO-1 localization and prevention of steroid-induced TER were accompanied by a stimulation in [ 3 H]thymidine incorporation. Flow cytometry analysis revealed that TGF-␤ also induced a shift in cellular DNA content to a profile consistent with a growing population of cells. 2 The ability of TGF-␤ to preclude the steroid induction of tight junction formation occurred in the presence of hydroxyurea, a DNA synthesis inhibitor, demonstrating that this disruptive effect on tight junction integrity is not an indirect consequence of an increase in DNA synthesis. This observation suggests that the tight junction machinery is a selective target of TGF-␤ receptor signaling. However, this evidence does not exclude the possibility that TGF-␤-mediated stimulation in [ 3 H]thymidine incorporation and cell growth may be a consequence of the disruption in tight junction structure and ZO-1 redistribution.
It has been proposed that junctional plaque proteins, such as ZO-1, help orchestrate the interactions of integral membrane proteins with cytoplasmic signaling components and, as a result, may serve as regulators of cell proliferation, cell adhesion, and cell-cell interactions. In this context, ZO-1 is homologous to the lethal(1)discs-large-1 (dlg) tumor suppressor gene of Drosophila, whose mutation resulted in neoplastic overgrowth of 2 H. Cha and G. Firestone, unpublished data. the imaginal discs as well as a loss of cell polarity and adhesion. Although a role for ZO-1 in intercellular signaling is unknown, several studies have attempted to ascertain the nature of its regulation. For example, ZO-1 was shown to be phosphorylated only on serine residues under normal conditions in MDCK cells (27) and tyrosine residues during the formation of tight junctions in the slit diaphragms of glomerular epithelial cells (56). We have shown that glucocorticoid stimulation of TER in 31EG4 cells did not alter the localization, phosphate content, or phosphopeptide digest pattern of ZO-1 but did increase protein levels (9). Recent evidence has shown that the PKC agonist 1,2-dioctanoylglycerol can promote the assembly of tight junctions, as evidenced by the translocation of ZO-1 and actin filaments to the cell periphery (57), and PKC can in vitro phosphorylate ZO-1, which may play a role in the formation of tight junctions (58). Previous observations demonstrating that ZO-1 is detected at the tight junction as well as adherens junction zone only in cells with leaky tight junctions (59) may explain the high colocalization of actin and ZO-1 in 31EG4 cells treated with TGF-␤. Clearly, the modification and regulation of ZO-1 are complicated processes that are dependent on the physiological environment, cell type, and integrity of the tight junction. Conceivably, the TGF-␤-mediated stimulation in [ 3 H]thymidine incorporation and tight junction disassembly may involve modification and redistribution of the ZO-1 protein.
TGF-␤ is a potent growth inhibitor of many epithelial-derived cells but generally stimulates proliferation of many fibroblast cells (60). The 31EG4 mammary cells used in our study are epithelial in nature because they are polarized and express the epithelial-specific marker gene cytokeratin type II, which is not produced in mouse fibroblasts. Thus, the untransformed cells are one of only a few characterized epithelial cell lines in which TGF-␤ stimulates their growth (42,43). Perhaps one reason for this relatively rare effect of TGF-␤ on epithelial cells is that the mammary cells are maintained on filters at confluency, which exposes the entire surface of the cells to nutrients in the extracellular environment. In addition, several studies have shown a general correlation between TGF-␤ expression and tumor cell growth (61,62). For example, highly proliferative mammary tumors contain an increased level of TGF-␤ transcripts compared with their normal human mammary cell counterparts (62). The mechanism by which TGF-␤ promotes cell growth is not well understood, although the stimulation in proliferation in a fibroblast cell line is accompanied by an activation of cyclin E-Cdk2 kinase and down-regulation of the p27/Kip1 cell cycle inhibitor (63). Another mechanism by which TGF-␤ may be stimulating mammary epithelial cell growth is by inducing transdifferentiation to a nonepithelial, mesenchymal-like phenotype (37). Whatever the precise mechanism of growth regulation of mammary epithelial cells, the novel antagonistic regulation of tight junction dynamics by TGF-␤ and glucocorticoids may have important implications for understanding hormonal contributions to controlling differentiation processes of the mammary gland, as well as cell-cell interactions involved with invasiveness and metastasis of mammary tumors.
It is tempting to consider that the selective antagonism between glucocorticoid and TGF-␤ signaling observed in vitro parallels that of in vivo cellular events associated with the control of mammary cell-cell interactions. Transgenic mice expressing TGF-␤ 1 targeted to the pregnant mammary gland showed inhibited alveolar development and lactation (64). We propose that TGF-␤, which is developmentally regulated during mammary gland differentiation, plays a critical role in maintaining a relatively leaky paracellular pathway, and that during lactation, dexamethasone is the predominant regulator of tight junction integrity. In an analogous manner, TGF-␤ has been shown to alter the synthesis of extracellular matrix components (65)(66)(67), which can potentially affect extracellular signaling and cell function. For instance, tenascin-C, which inhibits fibronectin-mediated adhesion, is stimulated by TGF-␤ 1 and down-regulated by glucocorticoids (68,69). The downstream targets of TGF-␤and/or glucocorticoid receptor-activated signaling cascades in mammary epithelial cells are mostly unknown. Conceivably, ZO-1 or other tight junction proteins could serve as potential targets for steroid or growth factor signaling pathways. We are currently attempting to elucidate the cellular events underlying the mutual antagonism of tight junction dynamics by glucocorticoids and TGF-␤ that operate in mammary epithelial cells. Such pathways could potentially represent an important cross-talk between growth factor and steroid receptor signal transduction cascades that is necessary to guide the functional relationships between particular sets of environmental cues that control mammary cell-cell interactions.