Glucocorticoid Down-regulation of RhoA Is Required for the Steroid-induced Organization of the Junctional Complex and Tight Junction Formation in Rat Mammary Epithelial Tumor Cells*

In Con8 mammary epithelial tumor cells, we have documented previously that the synthetic glucocorticoid dexamethasone induces the reorganization of the tight junction and adherens junction (apical junction) and stimulates the monolayer transepithelial electrical resistance (TER), which is a reliable in vitro measurement of tight junction sealing. Western blots demonstrated that dexamethasone treatment down-regulated the level of the RhoA small GTPase prior to the stimulation of the monolayer TER. To test the role of RhoA in the steroid regulation of apical junction dynamics functionally, RhoA levels were altered in Con8 cells by transfection of either constitutively active (RhoA.V14) or dominant negative (RhoA.DN19) forms of RhoA. Ectopic expression of constitutively active RhoA disrupted the dexa-methasone-stimulated localization of zonula occlu-dens-1 and (cid:1) -catenin to sites of cell-cell contact, inhibited tight junction sealing, and prevented the complete formation of the F-actin ring structure at the apical side of the cell monolayer. In a complementary manner, dominant negative RhoA caused a precocious organization of the tight junction, adherens junction, and the F-actin rings in the absence of steroid, whereas the monolayer TER remained glucocorticoid-respon-sive. Taken together, our results demonstrate that the glucocorticoid

The apical junctional complex, which consists of the tight junction, the adherens junction, and desmosomes, is located at the points of contact between cells and controls intercellular adhesion and the permeability properties involved in epithelial cell-cell interactions (1)(2)(3)(4)(5). The tight junction, or zonula occludens (ZO), 1 is the most apical structure of the junctional complex and forms a continuous "gasket-like" seal around the lateral circumference of adjacent cells which controls the selective diffusion of solutes through a paracellular pathway. Tight junc-tions also establish and maintain cellular polarity by restricting the lateral diffusion of lipids and membrane proteins between the compositionally and functionally distinct apical and basolateral membrane domains (6 -13).
Tight junction assembly is generally thought to require the initial engagement of cell-cell contacts at the adherens junction, which lies immediately basal to the tight junction and is responsible for intercellular adhesion between neighboring cells (14). Adherens junction formation is a process that is mediated by the calcium-dependent intercellular adhesion between E-cadherin molecules and the formation of an intracellular protein complex that includes E-cadherin, ␤-catenin, ␣-catenin, plakoglobin, and the actin cytoskeleton (8,14,15). These events trigger a series of molecular processes leading to the recruitment of tight junction components to the points of cell-cell contact and formation of a tight paracellular seal of the epithelium (13). The tight junction complex includes several classes of transmembrane proteins, including occludin (8), the claudin protein family (16 -19), intracellular peripheral membrane proteins, and potential cell signaling molecules (3,5). The cytoplasmic tail of occludin interacts with and tethers a complex of related peripheral membrane proteins ZO-1 (20) and its two related family members ZO-2 (21) and ZO-3 (22), which are members of the MAGUK (membrane-associated guanylate kinase) protein family. The MAGUK protein family has been proposed to recruit a variety of proteins to the membrane at the tight junction to form a tight extracellular seal (3,5). Several potential cell signaling molecules, such as atypical protein kinase C (23), the heterotrimeric G protein ␣ subunit (24,25), as well as the Ras target AF-6/Afadin, which binds to ZO-1 (26,27), have been localized to the tight junction.
Both the tight junction and the adherens junction have been proposed to associate with the perijunctional actin cytoskeleton through multiprotein complexes (14,28,29) and form an integrated functional unit. The architecture of the apical junctional complex, the cytoskeleton, and sealing of tight junctions at the sites of cell-cell contact can be regulated in a dynamic manner depending on the tissue origin, physiological state, and availability of specific sets of extracellular cues (3, 30 -32). These observations implicate the potential involvement of distinct signal transduction cascades that functionally target the tight junction and/or adherens junctions in epithelial cells.
We have discovered that glucocorticoids, one class of steroid hormones, induce tight junction formation and cell polarity through a multistep cascade in cultured mammary epithelial cells (33)(34)(35)(36)(37). An early glucocorticoid-regulated step in the overall cascade is the rapid stimulation in expression of the helix-loop-helix transcriptional regulatory molecule Id-1, which is required for the steroid control of cell tight junction dynamics (38). In a later set of events, glucocorticoids induce the recruitment of the tight junction proteins (ZO-1 and occludin), adherens junction proteins (E-cadherin and ␤-catenin), as well as the Ras and phosphatidylinositol 3-kinase cell signaling proteins to the cell periphery at the sites of cell-cell contact (38). This junctional reorganization process is followed by a distinct Ras-and phosphatidylinositol 3-kinase-dependent step that results in formation of highly sealed tight junctions in a polarized cell monolayer (38). Glucocorticoids down-regulate the level of fascin protein (37), an actin-bundling protein that binds to the adherens junction protein ␤-catenin. Fascin acts as inhibitor of tight junction formation, and we have recently shown that transforming growth factor-␣ disrupts junctional organization by preventing the steroid down-regulation of fascin (35).
Given the complex nature of the steroid-regulated cascade leading to tight junction formation, additional classes of regulatory proteins may functionally link the cytoskeleton to the control of tight junction dynamics in mammary epithelial cells. One such potential category of cell signaling molecules is the Rho family of small GTPases which includes RhoA, Rac, and Cdc42 (39 -45). In several cell systems, RhoA has been shown to be involved in cytoskeletal reorganization and has been implicated in regulating the spatial organization of adherens and/or tight junction proteins (41,42,46,47). However, an unambiguous role for RhoA in controlling cell-cell interactions has not been defined because of tissue/cell type-specific differences in RhoA-activating cellular cascades and in the experimental manipulation of tight junction formation. A unique feature of the Con8 mammary epithelial tumor cells is the ability to control acutely and reversibly the organization and function of the apical junctional complex with glucocorticoid treatment without having to first destroy and then rebuild tight junctions. In this study, we demonstrate that glucocorticoids down-regulate RhoA levels and that this regulatory process is required for the steroid-induced remodeling of the apical junction architecture that leads to functional tight junctions.
Cell Culture and Measurement of Transepithelial Electrical Resistance-Con8 rat mammary epithelial tumor cells have been described previously (37,49). To generate control, dominant active, and dominant negative Rho cell lines, the neomycin-resistance gene containing plasmids pcEXV-Myc-RhoA.V14, pcEXV-Myc-RhoA.DN19, and the pcEXV empty vector were transfected into Con8 cells. After selection with the neomycin analog G418 for 2 weeks, 50 clones were selected, expanded, and tested for induction of Myc-tagged RhoA. All cell lines were routinely grown to 100% confluence on Nunc permeable supports in Dulbecco's modified Eagle's medium/Ham's F-12 supplemented with 10% calf-serum and penicillin/streptomycin and maintained at 37°C in a humid atmosphere of air/CO 2 (95:5). For culturing of cells transfected with either the RhoA.V14 or RhoA.N19 expression vector, G418 was added to the culture medium to a final concentration of 600 g/ml. The cell culture medium was changed every day during the experimental treatment. For the glucocorticoid treatment of cells, the synthetic glucocorticoid agonist dexamethasone (Sigma) was added to the normal growth medium at a final concentration of 1 M (prepared as a 1,000 M stock in ethanol). Formation of tight junction was monitored by daily measurements of transepithelial electrical resistance (TER) using an EVOM Epithelial Voltohmmeter (World Precision Instruments, Sarasota, FL), as we have described previously (33-35, 37, 49). Calculations for ohms⅐cm 2 were determined by subtracting the resistance measurement of a blank filter and multiplying by the area of the monolayer.
Immunofluorescence and Confocal Microscopy-Con8-and RhoAtransfected rat mammary cells were grown on Nunc filters and incubated with or without 1 M dexamethasone for several days. Cells were treated every 24 h. For fluorescent analysis, the cell layers were washed three times with Dulbecco's phosphate-buffered saline (BioWhittaker) and fixed with 1.75% formaldehyde in phosphate-buffered saline for 15 min at room temperature. After three additional washes with phosphate-buffered saline, the plasma membrane was permeabilized with 0.5% Triton X-100 extraction buffer (0.5% Triton X-100, 100 mM Tris-HCl, pH 7.5, 120 mM NaCl, 20 mM Hepes, and 5 mM EDTA) for 15 min at room temperature. After blocking in 5% non-fat dry milk, filters were stained for 1 h with anti-ZO-1 antibodies followed by a 1-h incubation with fluorescein isothiocyanate-conjugated secondary antibodies. For costaining, the cells were incubated with either ␤-catenin antibodies followed by Texas Red-conjugated secondary antibodies or with Texas Red-conjugated actin-phalloidin. Fluorescent images were obtained from a Zeiss axiograph, and confocal fluorescent images were taken with a Leica confocal microscope and processed with the corresponding Leica software on Microsoft Windows NT.
Western Blots-For Western blot analyses, cells were rinsed with ice-cold phosphate-buffered saline (BioWhittaker) and extracted in lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 20 mM MgCl 2 , 1% Nonidet P-40) containing protease inhibitors. Samples were normalized for protein content with the Bio-Rad Bradford protein assay. Cell lysates were fractionated on a 12% SDS-polyacrylamide gel. Proteins were transferred electrophoretically from the gel to a nitrocellulose membrane (Micron Separations, Inc., Westboro, MA). Blots were blocked in 5% non-fat dry milk in washing buffer (0.1 M Tris-HCl, pH 8.0, 150 mM NaCl, 0.05% Tween 20) for at least 30 min at room temperature and then incubated with anti-RhoA or anti-␣-occludin antibodies overnight at 4°C. After three washes for 10 min each in 1% non-fat dry milk in washing buffer, the blots were incubated with horseradish peroxidaseconjugated anti-mouse antibodies (Bio-Rad) for 1 h. The blots were then washed twice with 1% non-fat dry milk in washing buffer and once with washing buffer. The blots were developed with a PerkinElmer Life Sciences chemiluminescence reagent kit. The level of RhoA protein observed in the Western blots was quantified with the Scion image program (Scion Corp., Frederick, MD).

Glucocorticoid Down-regulation of RhoA Protein Levels and Stimulation of Monolayer Transepithelial Electrical
Resistance-Our previous studies (33)(34)(35)(36)(37) demonstrated that in Con8 mammary epithelial tumor cells, glucocorticoids induce tight junction formation through a multistep process that results in a final intercellular membrane sealing event. As a first step to determine whether the small GTPase RhoA plays a regulatory role in this cascade, we examined the effects of glucocorticoids on the level of RhoA protein. After reaching confluence, Con8 cells were serum starved for 24 h, treated with or without 1 M synthetic glucocorticoid dexamethasone for up to 4 days, and RhoA protein levels were determined by Western blot analysis using monoclonal anti-RhoA-specific antibodies. As shown in Fig. 1A, dexamethasone caused a significant down-regulation of RhoA protein compared with the untreated control cells. This down-regulation was typically detectable by 8 h of steroid treatment and reached statistically significant levels by 48 h, which continued throughout the time course. Western blots probed with anti-actin antibodies (Fig. 1A, upper blot) and anti-␣occludin antibodies (Fig. 1A, lower blot) show that these protein levels remained generally constant during the time course, with some fluctuations in ␣-occludin protein. These protein levels did not change in response to dexamethasone treatment, thus providing a control for the specificity of the effects of glucocorticoids on RhoA protein levels. In untreated cells, RhoA protein levels at different time points can vary slightly from experiment to experiment, without a statistically significant change (p ϭ 0.05) overall during the period of the time course (data not shown).
Parallel sets of time course experiments examined the monolayer TER in confluent Con8 cells cultured on Nunc filter supports in the presence or in the absence of dexamethasone. The percent steroid-induced increase in TER was determined at each time point as the ratio of TER observed in steroid treated versus nontreated cells. Consistent with our previous studies (33)(34)(35)(36), after an approximate 24-h time lag, dexamethasone-treated cell monolayers displayed a significant increase in monolayer resistance, and the absolute TER plateaued at about 1,000 ohms⅐cm 2 by 4 days in steroid (Fig. 1B). The results from several RhoA Western blots were also averaged and plotted as the percentage of RhoA protein detected in dexamethasone-treated cells compared with untreated control cells. As also shown in Fig. 1B, the dexamethasone downregulation of RhoA protein correlates with the stimulation of TER and becomes statistically significant (p ϭ 0.05) at 48 h. These results suggest a potential causative relationship between the glucocorticoid regulation of RhoA protein levels and tight junction formation.

Establishment of Mammary Tumor Cell Lines That Express Constitutively Active or Dominant Negative RhoA and Remain
Glucocorticoid-responsive-To assess functionally the requirement of the glucocorticoid down-regulation of RhoA for tight junction formation and/or sealing, stable Con8-derived cell lines expressing either the constitutively active or dominant negative forms of RhoA were established. The pcEXV-Myc-RhoA.V14 expression plasmid encodes a constitutively active RhoA protein, which maintains the GTP-bound form of the protein, and an N-terminal Myc epitope tag. The pcEXV-Myc-RhoA.N19 expression plasmid also contains an N-terminal Myc epitope tag and encodes a dominant negative RhoA that binds more tightly to Rho-GEFs than the wild type GTPase, and thus preventing activation of endogenous RhoA, without binding to effector proteins (50). After selection of transfected Con8-derived cells with the neomycin analog G418 for 2 weeks, individual cell clones were selected, expanded, and tested for their expression of Myc-tagged RhoA mutant proteins using a monoclonal Myc antibody (Fig. 2). Cells expressing the empty vector behaved identically to the wild type, untransfected Con8 cells, therefore we chose to use wild type cells as controls for the remainder of the experiments. The Con8-DN25 (expressing the dominant negative RhoA) and the Con8-V6 (expressing the constitutively active RhoA) cells were utilized for further characterization of the role of RhoA in the glucocorticoid-induced tight junction formation.
To determine whether the Con8-derived cell lines expressing either the dominant negative or constitutively active RhoA remain responsive to glucocorticoids, Con8-V6 and Con8-DN25 cells, along with control Con8 cells, were grown on tissue culture dishes and treated either with or without 1 M dexamethasone for 5 days. This relatively long time point was utilized because it represents the glucocorticoid treatment duration needed to observe the maximal stimulation in TER in the parental Con8 cells (see Fig. 1). Total cell extracts were fractionated by SDS-polyacrylamide gels, and Western blots were probed with antibodies specific for RhoA to detect total RhoA protein. Western blots were also probed with antibodies for the Myc tag on the RhoA constructs to detect the exogenous RhoA proteins, as well as antibodies to the Sgk, which we originally identified as a primary glucocorticoid-responsive gene in Con8 mammary tumor cells (51). As a control for protein loading, the Western blots were probed with antibodies against ␣-actin.
As expected, Con8 cells show a strong down-regulation of endogenous RhoA protein upon dexamethasone treatment (Fig.  3, RhoA). Con8-V6 and Con8-DN25 cells produce a high level of RhoA immunoreactive protein in the presence of dexamethasone because of the expression of exogenous RhoA sequences. In these transfected cells, we also observed a significant reduction in the down-regulation of total RhoA in dexamethasonetreated cells compared with untreated cells (Fig. 3, RhoA). Probing the Western blots with monoclonal anti-Myc antibodies revealed that the level of exogenous RhoA protein, either the constitutive active or dominant negative form, is partially down-regulated upon dexamethasone treatment (Fig. 3, Myc-RhoA). This down-regulation of exogenous RhoA protein suggests that one of the effects of dexamethasone may be to alter RhoA protein stability. Importantly, the impact of constitutively active and dominant negative RhoA signaling is maintained in the transfected cells because of the relatively high level of RhoA protein which continues to be produced in dexamethasone-treated cells (Fig. 3, Myc-RhoA).
As also shown in Fig. 3, dexamethasone strongly induced Sgk in both Con8-V6 and Con8-DN25 cells to approximately the same extent, demonstrating that both transfected cell lines retained their overall responsiveness to glucocorticoids. Similar results were obtained when the cells were treated with glucocorticoids for short durations, including 1 h (data not shown). For unknown reasons, the induction of Sgk protein upon dexamethasone treatment appears higher in Con8-V6 and Con8-DN25 cells compared with Con8 cells.
The Down-regulation of RhoA Is Required for the Glucocorticoid-induced Organization of the Junctional Complex-The Con8-V6 and Con8-DN25 cells were used to characterize functionally the role of RhoA protein in the glucocorticoid-induced formation of tight junctions. We documented previously that tight junction sealing is preceded by a distinct steroid-regulated membrane organization step that controls remodeling of both the adherens and tight junctions (33). The glucocorticoidinduced organization of the junctional complex was examined in Con8-V6, Con8-DN25, and control Con8 cells by immunofluorescence visualization of the ZO-1 tight junction protein and the ␤-catenin adherens junction protein. Cells were grown to confluence on filter supports, treated with or without dexamethasone for 5 days, and the localization of ZO-1 and ␤-catenin was examined by indirect immunofluorescence microscopy using specific primary antibodies. As expected for Con8 cells (33)(34)(35), dexamethasone induced an overall rearrangement of the apical junction in which the staining of both ZO-1 and ␤-catenin changes from a relatively disorganized pattern with protein localized to the cytoplasm and/or nucleus to a honeycomb-like staining pattern (Fig. 4, top panels). We have shown previously that this staining pattern results from the redistribution of the tight junction and adherens junction proteins to contact points along the cell periphery (34). In contrast, dexamethasone treatment failed to induce completely the junctional complex organization in Con8-V6 cells, which produce constitutively active RhoA, showing significant cytoplasmic staining for ZO-1 and mostly nuclear/cytoplasmic staining for ␤-catenin (Fig. 4, middle panels). This result demonstrates that ectopic expression of active RhoA, which overrides the dexamethasone down-regulation of endogenous RhoA, interferes with proper glucocorticoid-induced organization of the apical junction, thereby implicating the requirement to lower RhoA levels to induce tight junction formation.
In a complementary manner, in Con8-DN25 cells both ␤-catenin and ZO-1 localized to cell junctions in the absence of dexamethasone (Fig. 4, bottom panels), showing a distinct honeycomb-like staining pattern. These results suggest that expression of dominant negative RhoA protein causes a precocious membrane organization in the absence of steroid. After glucocorticoid treatment, Con8-DN25 cells show a slightly enhanced localization of ZO-1 to the cell periphery, whereas ␤-catenin staining remained unchanged compared with untreated cells. Taken together, our results imply that the downregulation of RhoA protein is required for the regulated membrane reorganization that leads to the localization of adherens junction and tight junction proteins to the apical junctions.

Effects of Expressing Constitutively Active RhoA or Dominant Negative RhoA on Dexamethasone-induced Tight Junction
Sealing-To examine the potential effects of RhoA on the regulated sealing of tight junctions which occurs after the remodeling of the apical junction, Con8-V6, Con8-DN25 cells, and Con8 cells were grown to confluence on Nunc filter supports, and the monolayer TER was monitored in cells treated with and without 1 M dexamethasone for 5 days. Dexamethasone treatment significantly increased TER in control Con8 cells and Con8-DN25 cells, compared with untreated Con8 cells (p ϭ 0.05; Fig. 5). In the absence of steroids, the TER was ϳ200  ohms⅐cm 2 , whereas in glucocorticoid-treated cells the TER reached ϳ950 ohms⅐cm 2 (Fig. 5, black bars). The TER obtained for glucocorticoid-treated Con8 cells is used as the 100% level, with which all of the TER measurements were compared. In contrast to the untransfected Con8 cells, in Con8-V6 cells, which express constitutively active RhoA, dexamethasone failed to induce an increase in the monolayer TER, which remained at the basal level of ϳ200 ohms⅐cm 2 in the presence or in the absence of steroid (Fig. 5, Con8-V6 bars). The TER in cells expressing the dominant negative Con8-DN25 was partially inducible, with the TER reaching ϳ500 ohms⅐cm 2 in steroid-treated cells compared with the 360 ohms⅐cm 2 , TER observed in untreated cells (Fig. 5, gray bars). The basal level TER for the Con8-DN25 cells was not significantly different from either the Con8 or Con8-V6 cells; however, the dexamethasone-induced TER was statistically significant (p ϭ 0.05) compared with the basal TER. Thus, tight junction sealing remains glucocorticoid-inducible in cells expressing the dominant negative form of RhoA, although the magnitude of this response was less than that observed in the control Con8 cells. Interestingly, in the absence of glucocorticoids, Con8-DN25 cells displayed an organized junctional complex (see Fig. 4) without tight junction sealing. This result suggests that the inhibitory effects of RhoA protein likely disrupt membrane organization at the apical junction, whereas other glucocorticoid-responsive signaling pathways are responsible for the induction of tight junction sealing.

Effects of RhoA on the Glucocorticoid-regulated Organization of F-actin at the Apical Junction and on Actin Stress Fiber
Formation-We observed previously that glucocorticoids stimulate perijunctional F-actin ring formation at adherens junctions of Con8 mammary tumor cells (33). Several studies have shown that F-actin is physically associated with the adherens junction by its tethering to ␤-catenin via ␣-catenin (52,53) and to the tight junction through association with ZO-1 (28), ZO-2, ZO-3, occludin (54), and cingulin (55). Furthermore, the absence of an intact actin cytoskeleton has been associated with ZO-1 (56,57) and occludin disorganization at the tight junction (57). Although RhoA has been implicated in stress fiber formation in different cell lines (44, 58 -60), RhoA has also been demonstrated to lose its regulatory control over stress fiber formation when up-regulated as an indirect consequence of Ras transformation (61). Given the apparent involvement of RhoA in actin cytoskeleton reorganization, Con8-V6 and Con8-DN25 cells were utilized to determine the potential effects of RhoA on the glucocorticoid control of F-actin ring formation and on stress fiber formation in Con8 mammary tumor cells.
After a 5-day treatment of cells cultured on filter supports in the presence or absence of dexamethasone, the localization pattern of F-actin was examined by confocal laser immunofluorescence microscopy. Apical staining visualized the localization of F-actin at the adherens and tight junctions (Apical), whereas actin stress fibers were detected at the basolateral Con8-V6 and Con8-DN25 cells along with control Con8 cells were grown to confluence on filter inserts and cultured in the presence (ϩDex) or absence (ϪDex) of 1 M dexamethasone for 5 days. The monolayer TER was measured and the ohms⅐cm 2 calculated as described under "Experimental Procedures." The average of the TER data for dexamethasone-treated Con8 cells was set at 100%, and the average TER data for all other cell lines is represented as a fraction thereof. Significance (p Ͻ 0.005) was established by conducting a two-sample t test analysis. Data shown represent an average of at least three independent experiments. surface (Basal). As shown in Fig. 6 (top panels, Basal), staining for F-actin showed that actin stress fiber formation was relatively unaffected by dexamethasone in Con8 cells under conditions in which the steroid induces F-actin ring formation (Fig.  6, top panels, Apical). In contrast to the control cells, in Con8-V6 cells expressing constitutively active RhoA, the formation of F-actin rings was less strongly induced by steroid treatment, and actin staining could be found throughout the cytoplasm (Fig. 6, middle panels, Apical). The actin stress fibers remained unchanged in Con8-V6 cells in the presence or absence of dexamethasone (Fig. 6, left panels, Basal), suggesting that the general cellular adhesion properties remained unaffected. In Con8-DN25 cells expressing dominant negative RhoA, we observed an induced perijunctional F-actin ring formation independent of dexamethasone treatment (Fig. 6, bottom panels, Apical). These results are consistent with the organized localization of ␤-catenin and ZO-1 in the junctional complex (shown in Fig. 4). Thus, in mammary tumor cells, the down-regulation of RhoA appears to be required for the steroid induced F-actin ring formation, whereas there are no significant effects on actin stress fiber formation.
RhoA Does Not Affect the Dexamethasone-induced Formation of Mammary Tumor Cell Monolayers-We observed previously that dexamethasone treatment of multilayered Con8 cells induces the cells to form a monolayer under conditions in which tight junctions are formed (35). To determine whether the observed effects of RhoA on apical junction remodeling and F-actin ring formation may be a fortuitous consequence of a disruption in cell monolayer formation, the XZ planes of actinstained cell samples were analyzed by confocal laser microscopy. As also shown in Fig. 6 (Z-planes), 5 days treatment with dexamethasone induced cell monolayer formation in the control Con8 cells, Con8-V6 cells, and Con8-DN25 cells. Thus, the disrupted formation of tight junctions and F-actin rings observed in cells expressing constitutively active RhoA and the precocious membrane and cytoskeletal remodeling observed in cells expressing dominant negative RhoA cannot be explained by any secondary effects on monolayer formation. DISCUSSION A key characteristic of the Con8 mammary tumor epithelial cells used in our study is that physiologically relevant hormonal cues stimulate the formation of tight junctions and a polarized cell monolayer (33)(34)(35)38). After an early transcriptional event (38), glucocorticoids induce the organization of the apical junctional complex which results in the recruitment of tight junction proteins, adherens junction proteins, and several intracellular signaling proteins to the cell periphery at the sites of cell-cell contact (33). This junctional reorganization, which is reversed by growth factor exposure (34 -36), is followed by a distinct Ras-dependent step that results in formation of highly sealed tight junctions (33). In the current study, we now establish that the steroid downregulation of the RhoA small GTPase is required for junctional reorganization. Ectopic expression of constitutively active RhoA protein interferes with glucocorticoid-induced localization of both the ZO-1 tight junction protein and the ␤-catenin adherens junction protein to the cell periphery. The subsequent sealing of the tight junction, as monitored by the monolayer TER, is also disrupted. In a complementary manner, expression of a dominant negative RhoA protein results in the precocious organization of the apical junctional complex, although sealing of the tight junction still required glucocorticoids. As diagrammed in Fig. 7, our results implicate RhoA as an inhibitor of the glucocorticoid-regulated cascade leading to tight junction formation in mammary epithelial tumor cells, which needs to be down-regulated prior to or during the reorganization of the overall apical junction. Our previous studies demonstrated that Ras, another small GTPase family member, is required for the steroid-inducible tight junction sealing in rat mammary epithelial cells, suggesting distinct cellular functions for RhoA and Ras in regulating apical junctional complex formation and function.
A precise role for RhoA in tight junction dynamics has not been defined previously, in part because of apparent cell type differences in RhoA function and regulation. MDCK cells are widely used to examine cell-cell interactions because these cells constitutively organize their apical junctional complex with sealed tight junctions (42,(62)(63)(64)(65)(66)(67)(68). However, in these cells nonphysiological conditions, such as the removal and readdition of calcium or ATP, are generally used to study tight junction dynamics. With an ATP depletion protocol, overexpression of a dominant negative RhoA in transiently transfected MDCK cells was shown to induce the accumulation of the ZO-1 and occludin tight junction proteins at the apical cell junctions (43). Using calcium switch assays, several studies demonstrated that ectopically expressed RhoA increased the monolayer TER (44,69), although paradoxically, in one of these studies overexpression of constitutively active RhoA caused ZO-1 localization to be highly disorganized despite an increase in barrier function (69). Interestingly, ectopic expression of either the constitutively active RhoA or dominant negative RhoA in MDCK cells was shown to perturb tight junction sealing, suggesting that RhoA inhibits tight junction formation but only at an optimal cellular level (42). Furthermore, expression of active RhoA, but not the dominant negative RhoA, disrupted the tight junction strand morphology and localization of occludin and ZO-1 tight junction proteins (42). Thus, in combination with our hormonal studies using mammary epithelial cells, the emerging evidence generally favors an inhibitory role for RhoA in tight junction formation. Although it is not known how downstream effectors of RhoA mediate their effects on the apical junctional complex, a family of Rho-activated serine/ threonine protein kinases, p160ROCK (ROCK-I) and ROK␣/ Rho kinase (ROCK-II), has been shown to be required for tight junction assembly and function in polarized intestinal epithelial cells (70) and MDCK cells (44). We are currently examining the role of individual downstream RhoA effectors in glucocorticoid-induced formation of tight junctions in mammary epithelial cells.
The tight junction, the adherens junction, and the cytoskeleton form an integrated functional unit through multiprotein complexes. For example, the cadherin proteins of the adherens junction are structurally connected to F-actin polymers through ␤-catenin and ␣-catenin (53). Also, the actin cytoskeleton interacts with the ZO-1 tight junction protein (31) or potentially indirectly through ZO-1-binding proteins such as spectrin (71) and cortactin (72). Furthermore, the actin cytoskeleton interacts with additional proteins localized to the tight junction, such as ZO-2, ZO-3, occludin (54), and cingulin (55). Therefore, one mechanism by which cellular RhoA activity could potentially disrupt the junctional complex is by selectively altering the cytoskeleton. Consistent with this viewpoint, we observed that ectopic expression of constitutively active RhoA in mammary epithelial cells prevented the complete steroid-induced reorganization of apical F-actin structures and ␤-catenin localization under conditions that ablate tight junction formation. Moreover, expression of a dominant negative RhoA protein precociously organized F-actin structures and ␤-catenin localization at the lateral apical surface in the absence of steroid, with almost no cytoplasmic staining of actin or ␤-catenin. We propose that the steroid-induced down-regulation of RhoA or expression of dominant negative RhoA allows an apical F-actin organization that permits ␤-catenin accumulation at the lateral plasma membranes, for example by generating a cytoskeletal scaffold to which ␤-catenin, and possibly ZO-1, can bind at points of cell-cell contacts. Alternatively, the dominant negative RhoA may cause the adherens and tight junction proteins to organize at the lateral plasma membrane and thus provide anchorage for actin filaments.
In the mammary epithelial tumor cells used in our study, ectopic expression of the constitutively active RhoA had no effect on stress fiber formation already produced by Con8 cells. This result suggests that the hormonal manipulation of tight junction dynamics in mammary epithelial tumor cells does not alter the general cellular adhesion properties. In MDCK cells, overexpression of either the dominant negative or the constitutively active RhoA increased stress fibers (42). In another study, ectopically expressed RhoA had a biphasic effect in that low levels stimulated actin stress fibers, whereas high levels of RhoA reduced stress fiber formation and caused further disruption of the overall actin cytoskeleton (73). These observations suggest that differences in the role of RhoA in stress fiber formation may be the result of either cell type-specific differences between MDCK cells and the mammary tumor cells used in our study and/or the nonhormonal approaches used to manipulate tight junction formation in MDCK cells.
Increased RhoA signaling has been reported in several malignant carcinomas such as breast cancer (74), testicular germ cell tumors (75,76), and in human pancreatic cancer cells (77). Conversely, monoterpenes, which are being tested in clinical FIG. 7. Model for the multistep glucocorticoid-regulated cascade that controls apical junction remodeling and tight junction sealing. We propose that the steroid-induced down-regulation of RhoA protein is a critical step in the glucocorticoid-induced remodeling of the junctional complex organization that includes both the adherens junction and the tight junction. The subsequent glucocorticoid-regulated tight junction sealing occurs in a Ras-and phosphatidylinositol 3-kinase (PI3K)-dependent fashion. Ectopic expression of a constitutively active RhoA (RhoA.V14) interferes with apical junction remodeling, resulting in a loss of tight junction sealing. In a complementary manner, ectopic expression of dominant negative RhoA (RhoA.N19) stimulates junctional remodeling in the absence of steroids. trials as breast cancer therapy, have been shown to inhibit the isoprenylation of RhoA and likely RhoA functionality (78). The Con8 cells used in our study are a tumorigenic mammary epithelial cell line (79), suggesting that utilization of RhoA is distinct from nontumorigenic cells such as the MDCK cell lines. We have documented that glucocorticoids induce an overall "differentiated-like" phenotype in the Con8 mammary tumor cells, which includes the remodeling of the apical junctional complex and formation of tight junctions, a G 1 cell cycle arrest, the down-regulation of transforming growth factor-␣ production, and the stimulated formation of a polarized cell monolayer (34 -36, 79, 80). Our study demonstrates that the glucocorticoid down-regulation of RhoA is an integral event in the emergence of this cellular phenotype. Furthermore, our results imply that small GTPases, such as RhoA and Ras, are involved in the in vivo regulation of the tight junctions in the mammary gland, for example during late pregnancy and lactation when glucocorticoid levels are high (81,82). We are currently examining the mechanism by which glucocorticoids down-regulate RhoA protein and are determining the potential in vivo significance of the down-regulation of RhoA and effects on its signaling targets.