Rho GTPases as Modulators of the Estrogen Receptor Transcriptional Response*

The estrogen receptor a (ER) is a ligand-dependent transcription factor that plays a critical role in the development and progression of breast cancer, in part, by regulating target genes involved in cellular proliferation. To identify novel components that affect the ER transcriptional response, we performed a genetic screen in yeast and identified RDI1, a Rho guanine nucleotide dissociation inhibitor (Rho GDI), as a positive regulator of ER transactivation. Overexpression of the human homologue of RDI1, Rho GDI a , increases ER a , ER b , androgen receptor, and glucocorticoid receptor transcriptional activation in mammalian cells but not activation by the unrelated transcription factors serum response factor and Sp1. In contrast, expression of constitutively active forms of RhoA, Rac1, and Cdc42 decrease ER transcriptional activity, suggesting that Rho GDI increases ER transactivation by antagonizing Rho function. Inhibition of RhoA by expression of either the Clostridium botulinum C3 transferase or a dominant negative RhoA resulted in enhanced ER transcriptional activation, thus phenocopying the effect of Rho GDI expression on ER transactivation. Together, these findings establish the Rho GTPases as important modulators of ER transcriptional activation. Since Rho GTPases regulate actin polymerization, our findings suggest a link between the major regulators of cellular architecture and steroid receptor transcriptional response.

The estrogen receptor ␣ (ER) 1 is a ligand-dependent transcription factor that transduces the estrogen signal (1). Activation of ER is responsible for female sexual development and maintenance of bone density (2,3). In addition, ER plays a critical role in the development and progression of breast can-cer by regulating genes and signaling pathways involved in cellular proliferation (4). Regulation of gene expression by the ER requires the coordinate activity of ligand binding, phosphorylation, and cofactor interactions, with particular combinations probably resulting in the tissue-specific responses elicited by the receptor (5)(6)(7). However, the extracellular cues and intracellular signaling pathways modulating these components and regulating ER transcriptional activation are not fully understood.
To identify novel proteins that modulate ER transcriptional activation, we have carried out a genetic screen in the yeast Saccharomyces cerevisiae. The ability of the human ER to function within yeast allows a wide variety of genetic approaches to be taken toward further defining the mechanism of ER transcriptional activation given the ease of genetic manipulation and simplicity of gene identification in yeast. In addition, with the large number of orthologous proteins carrying out the same biological functions in both S. cerevisiae and metazoans (8 -10), it is likely that the yeast factors affecting ER transactivation will have mammalian counterparts, which can be examined in vertebrate systems.
The genetic approach we have used to identify factors that affect ER transcriptional activation is dosage suppression analysis. In this technique, a mutant ER protein with a reduced ability to activate transcription is used as a substrate to isolate yeast gene product(s) that are capable of overcoming this mutant phenotype, thus restoring receptor transcriptional activity. The mutant ER derivative used in this screen is defective in transactivation by virtue of serine to alanine mutations in the three major N-terminal phosphorylation sites, serines 104, 106, and 118 (ER AAA ). This mutant was selected as the substrate because it has only a modest effect on ER transactivation and therefore has the potential to isolate a broad range of factors that affect receptor activity. We expect to isolate yeast factors that enhance ER transcriptional activity and, importantly, have human homologues that can then be examined in mammalian cells for effects on ER transcriptional response. This approach has proven successful for investigating various aspects of ER signal transduction (11). Using this system, we have isolated RDI1, the yeast Rho guanine nucleotide dissociation inhibitor (Rho GDI), as a gene product that is capable of increasing both ER AAA and WT ER transcriptional activation when overexpressed. This gene product is the yeast homologue of the mammalian Rho GDI␣, a cytoplasmic protein originally identified as a negative regulator of the Rho family of GTPbinding proteins (12)(13)(14)(15). The Rho family of GTPases, which include RhoA, Rac1, and Cdc42, are best known for their ability to regulate actin cytoskeletal remodeling in response to extracellular signals, thereby promoting changes in cell morphology, adhesion, and motility (16). In addition, by affecting multiple signaling pathways, Rho family members regulate gene transcription and cell cycle progression and have been implicated in cellular transformation and metastasis (17)(18)(19)(20)(21)(22). The Rho family members possess intrinsic GTPase activity and cycle between the inactive cytoplasmic GDP-bound and the active membrane-associated GTP-bound state. The exchange of GDP for GTP induces a conformational change in the G protein that allows effector molecules, such as protein kinases, to bind and initiate downstream signaling events (23). This GTP/GDP cycle is tightly regulated in response to extracellular signals by three different classes of proteins. Guanine nucleotide exchange factors (GEFs) catalyze the exchange of GDP for GTP, GTPaseactivating proteins (GAPs) accelerate the intrinsic GTPase activity of the Rho GTPases, and GDIs antagonize their activity by blocking GEFs and GAPs (12,24). However, since the cytoplasmic GDP-bound Rho GTPases predominate under physiological conditions, Rho GDI acts as a negative regulator of Rho GTPases mainly by blocking the dissociation of GDP. In addition, Rho GDI controls the subcellular localization of the GT-Pases, stimulating their release from the plasma membrane (12,25). Interestingly, Rubino et al. (26) identified an ERinteracting protein, termed Brx, which contains a domain virtually identical to the Rho GEF Lbc and was first to demonstrate a link between ER signaling and the Rho GTPases.
In this report, we extend the findings of Rubino et al. (26) and examine the effects of human Rho GDI␣ as well as the Rho GTPases, RhoA, Rac1, Cdc42, on transcriptional activation by ER and other members of the steroid receptor family in mammalian cells. Our findings indicate that Rho GDI␣ specifically increases the transcriptional activity of ER␣ and ER␤ as well as the glucocorticoid receptor (GR) and androgen receptor (AR), but not of the unrelated transcription factors serum response factor (SRF) and Sp1, and that this activation is mediated via repression of Rho GTPases. These results further establish the Rho-mediated signaling pathway as an important regulator of ER, GR, and AR transcriptional activity.

Plasmids
Yeast-The reporter plasmid ERE-CYC1-LacZ contains a single estrogen response element (ERE) upstream of a truncated CYC1 promoter linked to the ␤-galactosidase gene (27). The yeast high copy genomic library was described by Engebrecht et al. (28) and was generated by subcloning Sau3A partially digested yeast genomic DNA into the BamHI site of the YEP351 plasmid. WT ER and ER AAA were expressed from the Gal1-10 promoter in Trp1, 2 M plasmid (p2T-GAL) (29). p2T-GAL-ER AAA was constructed by subcloning the BamHI fragment containing ER AAA sequence from pcDNA3 (30) plasmid into p2T-GAL (29).
Mammalian Cells-The ER reporter plasmid contains one ERE from the Xenopus vitellogenin A2 gene, upstream of the herpes simplex virus thymidine kinase promoter (Ϫ109) linked to the firefly luciferase coding sequence (XETL) (31). The GR reporter plasmid (XG 46 TL) is identical to XETL, except two consensus GREs are substituted for the ERE (31). The steroid receptor expression plasmids are pcDNA3-human ER␣, pCMV5-human ER␤ (32), pcDNA3-rat GR (33), and pcDNA3-human AR (34). A BamHI/EcoRI fragment of human Rho GDI␣ from pGEX2T (35) was subcloned into pcDNA3 to create a Rho GDI␣ mammalian expression construct. 2 Expression plasmids for N-terminally Myctagged Rac1.L61 and Cdc42.L61 have been described previously (36). pRK5-Myc-RhoA.V14 was subcloned as an EcoRI fragment containing Myc-tagged RhoA.V14 from EXV plasmid (18). Both EXV.RhoA.V14 and EFC3-expressing Myc-tagged C3 transferase under the EF1␣ promoter have been described elsewhere (37,38). The dominant negative form of RhoA, RhoA.N19, was made by site-directed mutagenesis using the oligonucleotide 5Ј-GGAGCCTGTGGAAAGAACTGCTTGCT-CATAGTC-3Ј and the QuickChange mutagenesis kit (Stratagene) with pRK5-Myc-RhoA as the template. The entire RhoA.N19 coding region was sequenced to verify the base changed and to ensure that no other mutations were introduced. The Sp1 reporter contains six Sp1 binding sites upstream of the adenovirus major late promoter in front of the luciferase gene (39), and SRF reporter contains a fragment of the c-Fos promoter upstream of luciferase (18).

Yeast Strains, Growth Conditions, and ␤-Galactosidase Assay
The yeast strain W303a (a ade2 leu2 his3 trp1 ura3) was used to screen for ER activators. Yeast transformation was performed by the lithium acetate/polyethylene glycol method (40). To assay ER transcriptional activation, cells were cultured overnight in the appropriate selective medium containing 2% glucose and subcultured 1:20 in selective minimal medium containing 2% galactose, 1% raffinose to induce receptor expression and treated with 17␤-estradiol for 12 h. Quantitative liquid ␤-galactosidase assays were performed as described previously (11). Plate assays were performed by replica-plating colonies from glucose plates onto galactose X-gal indicator plates containing 0.1 nM 17␤-estradiol.

Cell Culture, Transfection, and Luciferase Assays
Human osteosarcoma U2OS (HTB 96) and human breast cancer MCF-7 (HTB-22) cell lines were obtained from the American Type Culture Collection (Manassas, VA), and the Ishikawa human uterine cancer cell line was obtained from Dr. Seth Guller (NYU School of Medicine). Cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (HyClone), 10 units/ml each of penicillin and streptomycin (Cellgro), and 2 mM L-glutamine (Cellgro). Between 1.2 and 1.3 ϫ 10 5 cells were seeded onto 35-mm plates in phenol red-free Dulbecco's modified Eagle's medium (Cellgro) supplemented with 10% charcoal-stripped fetal bovine serum and 2 mM L-glutamine. Transfections using Lipo-fectAMINE Plus reagent (Life Technologies) were performed according to the manufacturer's recommendation. Cells were treated with hormone agonists (100 nM 17␤-estradiol, 100 nM dexamethasone, and 100 nM R1881 for ER, GR, and AR, respectively), the ER antagonist ICI 182,780 (41) (100 nM), or ethanol vehicle 12 h post-transfection for 24 h. Transfected cells were washed once in phosphate-buffered saline and harvested in 1ϫ reporter lysis buffer (Promega) as per the manufacturer's instructions. Luciferase activity was quantified in a reaction mixture containing 25 mM glycylglycine, pH 7.8, 15 mM MgSO 4 , 1 mM ATP, 0.1 mg/ml bovine serum albumin, 1 mM dithiothreitol, using an LB 9507 luminometer (EG & G Berthold) and 1 mM D-luciferin as substrate. The steroid receptor transcriptional activity is normalized to reporter activity in the absence of transfected steroid receptors and to protein concentration as determined by the Bradford protein assay (Bio-Rad). Since MCF-7 cells contain endogenous ER, transcriptional activity of the receptor is normalized to XETL activity in the presence of ER antagonist ICI 182,780. The data presented represent the average of two experimental values, with error bars representing the range of the data points.

Immunoblotting
To prepare protein extracts from transfected cells, whole cell extracts prepared for luciferase assay in 1ϫ reporter lysis buffer were normalized for total protein and boiled for 3 min in SDS sample buffer. Protein extracts were fractionated by 12% SDS-polyacrylamide gel electrophoresis, transferred to Immobilon membrane (Millipore Corp.), and probed with anti-ER␣ polyclonal antibody (HC-20; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-Rho GDI polyclonal antibody (A-20; Santa Cruz Biotechnology), or anti-c-Myc monoclonal antibody (9E10; Santa Cruz Biotechnology). The blots were developed using horseradish peroxidase-coupled sheep anti-mouse or goat anti-rabbit antibodies and the ECL substrate as per the manufacturer's instructions (Amersham Pharmacia Biotech).

A Genetic Screen for Activators of ER␣ Transcriptional
Enhancement-Concomitant serine to alanine mutations at Nterminal phosphorylation sites 104, 106, and 118 (ER AAA ) result in a ϳ50% reduction in ER transcriptional activity in mammalian cells (42,43). To determine whether the transcriptional activity of ER AAA is also reduced in yeast, strains were constructed containing a galactose-inducible expression vector encoding either WT ER or ER AAA and an ER-responsive reporter plasmid. The transcriptional activities of WT ER and ER AAA were measured as a function of hormone concentration. transcriptional activity at all hormone concentrations tested, suggesting that the ER AAA is less efficient at engaging in the interactions necessary for transcriptional activation (Fig. 1A).
The ER AAA phenotype is most striking at 0.1 nM 17␤-estradiol. Under these conditions, yeast colonies expressing WT ER are blue, while ER AAA -expressing colonies appear white (Fig.  1B). To screen for ER activators, yeasts expressing ER AAA , along with an estrogen-responsive reporter gene, were transformed with a high copy yeast genomic library and assayed for receptor transcriptional activation on X-gal indicator plates containing 0.1 nM 17␤-estradiol. Candidate high copy suppressors changing the ER AAA -expressing yeast from white to blue were selected for further analysis (Fig. 1B). Of the 29,000 colonies screened, which represents approximately 3 times the size of the yeast genome, we identified six yeast open reading frames that enhance ER transcriptional activation (Table I).
A search of the yeast genome data base revealed that two of the candidate suppressors were yeast homologues of mammalian proteins previously shown to affect ER transactivation. These include 1) CKA1 (44), a homologue of the mammalian ␣-subunit of casein kinase II that phosphorylates ER at serine 167 in vitro (45) and 2) CAD1 (46), a member of the Jun transcription factor family that synergizes with ER in mammalian cells (47). In addition to genes known to regulate ER activity, several genes not known to affect ER were identified. YAK1 (48), a serine/threonine kinase with homology to ANPK, a protein kinase that interacts with the zinc finger region of the AR and increases AR-dependent transcriptional activation, was isolated once (49). In addition, MCK1 (50), a protein kinase with homology to glycogen synthase kinase-3 (51), was identified once. We also isolated RDI1 (14), the yeast Rho guanine nucleotide dissociation inhibitor (Rho GDI), three times, and LRG1 (52), a yeast protein that contains a GTPase-activating protein (GAP) homology domain, once. Although LRG1 is presently linked to GAP merely through sequence homology, it is interesting to note that GAP and RDI1 are both negative regulators of Rho GTPases. The recovery of known ER regulators together with the repeated isolation of certain genes indicates that the approach was sound and that the library was probably screened to saturation. Since Rho GDI negatively regulates Rho GTPases, this result suggests that the Rho GTPases may modulate ER transcriptional activation and is the focus of this report.
Rho GDI Expression Increases ER Transactivation-Among the human Rho GDIs, RDI1 is most similar to human Rho GDI␣, which negatively regulates the best studied Rho GT-Pases, RhoA, Rac1, and Cdc42. To examine whether the mammalian Rho GDI affects ER transcription in mammalian cells, we tested the ability of human Rho GDI␣ to enhance ER transcriptional activity in the human osteosarcoma cell line U2OS. ER-negative U2OS cells were transiently transfected with ER␣, an ER-responsive reporter plasmid, along with increasing amounts of Rho GDI␣. As shown in Fig. 2A, Rho GDI␣ stimulates ER transactivation in a dose-dependent manner. Enhancement of ER transcriptional activation by Rho GDI␣ was also observed for ER AAA mutant (not shown). To ensure that this enhanced transcriptional activity was not a result of increased ER protein production, we monitored protein expression in whole cell extracts using Western blot analysis. Fig. 2B illustrates that ER levels are not increased by Rho GDI expression. Indeed, the steady state concentration of ER decreased slightly with increasing Rho GDI expression, indicating that the effect of Rho GDI␣ on ER activity is greater than that observed. The effect of Rho GDI on ER transactivation is not restricted to single cell type, since Rho GDI␣ also enhanced ER transactivation in MCF-7 and Ishikawa cells (Fig. 2C). Thus, Rho GDI␣ can act as a positive regulator of ER-dependent transcriptional activation in a variety of mammalian cell lines.
Rho GDI Specifically Activates Steroid Hormone Receptors-We next tested the ability of Rho GDI␣ to affect transactivation by other members of the steroid receptor family, ER␤, GR, and AR, using transient transfection assays. Our results indicate that Rho GDI␣ also increased the transcriptional activity of ER␤, GR, and AR in a dose-dependent manner (Fig. 3, A-C). To determine whether Rho GDI-mediated activation is specific to steroid receptors, we tested the effect of Rho GDI␣ on Sp1-and SRF-dependent transactivation. Rho GTPase signaling has been previously shown to enhance transcriptional activation by SRF (18); thus, we would expect Rho GDI␣, as a negative regulator of Rho GTPases, to decrease SRF transcriptional activity. Consistent with this idea, Rho GDI␣ expression decreased SRF activity from a reporter plasmid containing the c-Fos SRF element (Fig. 4A). Similarly, Sp1 transcriptional activity using an Sp1-responsive reporter also decreased in response to Rho GDI overexpression (Fig. 4B). Taken together, these results strongly suggest that Rho GDI specifically increases transactivation by steroid hormone receptors, perhaps through a mechanism involving suppression of Rho GTPase signaling.
Rho GTPases Inhibit ER Transactivation-The GTPases known to interact with Rho GDI␣ include RhoA, Rac1, and Cdc42. To determine whether Rho GDI increases ER transactivation by inhibiting the Rho GTPases, we expressed constitutively active forms of Rho GTPases (RhoA.V14, Rac1.L61, and Cdc42.L61) in U2OS cells and examined ER transcriptional activation. As shown in Fig. 5, expression of RhoA.V14, Rac1.L61, and Cdc42.L61 decreased ER transcriptional enhancement. Active forms of Rho GTPases also decreased ER transactivation in MCF-7 and Ishikawa cells (Fig. 6). In all three cell types, expression of the constitutively active forms of RhoA, Rac1, and Cdc42 resulted in an accumulation of filamentous actin, as determined by fluorescent phalloidin staining. 3 These results are consistent with the model that Rho GDI 3 L. F. Su and M. J. Garabedian, unpublished observations.

FIG. 1. Isolation of yeast factors that increase transcriptional activation by ER␣.
A, transcriptional activation of WT ER and ER AAA as a function of 17␤-estradiol concentration. Yeast strains were transformed with either a galactose-inducible WT ER or ER AAA , along with an ERE-containing ␤-galactosidase reporter plasmid. Transcriptional activation by the WT ER (dotted line) and ER AAA (solid line) in response to increasing 17␤-estradiol concentration was determined by liquid ␤-galactosidase assay as described under "Experimental Procedures." Note that the ER AAA in yeast exhibits ϳ40% of the WT ER transcriptional activity at each estradiol concentration tested. The dosage suppression screen was carried out in the presence of 1 ϫ 10 Ϫ10 M 17␤estradiol, conditions under which the ER AAA phenotype is the most pronounced. B, the relative activity of WT ER, ER AAA , and ER AAA with an ER activator. Three independent colonies on X-gal indicator plates in the presence of 1 ϫ 10 Ϫ10 M 17␤-estradiol are shown and represent WT ER with an empty expression vector (WT ER), ER AAA plus an empty expression vector (ER (AAA)), and ER AAA plus the RDI1 suppressor plasmid (ER (AAA) ϩ ER activator). Under these conditions, colonies expressing WT ER are blue, ER AAA -expressing colonies appear white, and ER AAA -expressing the ER activator RDI1 are blue.
activates ER transcriptional enhancement by antagonizing Rho GTPases.
As an independent means of examining the effect of RhoA inhibition on ER transcriptional activation, we ectopically expressed the Clostridium botulinum C3 transferase, a protein toxin that ADP-ribosylates and inhibits RhoA but not Rac1 or Cdc42 (37,38). As with Rho GDI, expression of C3 transferase results in an enhancement of ER transcriptional activity but decreases SRF transcriptional activation in U2OS cells (Fig. 7A and data not shown). Inhibition of ER transcriptional activation by activated RhoA, but not Rac1 or Cdc42, was also relieved by C3 coexpression (not shown). Ectopic expression of C3 Plus reagent with 0.1 g of ER␣ expression construct or empty vector, 0.2 g of the ERE-containing reporter gene XETL, and increasing amounts of Rho GDI␣, as indicated. 12 h after the transfection, cells were treated with 100 nM 17␤-estradiol (E2) (dark bars) or the ethanol vehicle (light bars) for 24 h, harvested, and assayed for luciferase activity. ER␣ transcriptional activity is normalized to XETL reporter activity in the absence of ER. The data represent the mean of an experiment done in duplicate, which was repeated three times. B, expression of ER␣ does not increase by Rho GDI␣ coexpression. Whole cell extracts were prepared from transfected cells as described under "Experimental Procedures," and the expression of ER␣ and Rho GDI␣ was analyzed by Western blotting. C, MCF-7 and Ishikawa cells were transfected as above and assayed for luciferase activity. For Ishikawa cells, ER␣ transcriptional activity is normalized to XETL reporter activity in the absence of ER. For MCF-7 cells that contain endogenous ER, transcriptional activity of the receptor is normalized to XETL activity in the presence of ER antagonist ICI 182,780. The data represents the mean of experiments done in duplicate, which were repeated two times.
FIG. 3. Rho GDI␣ enhances the transcriptional activation by ER␤, GR, and AR. U2OS cells were transfected as described in Fig. 2 with paired expression and reporter plasmids for ER␤ ϩ XETL (A), GR ϩ XG 46 TL (B), or AR ϩ XG 46 TL (C) and, along with the indicated amount of Rho GDI␣, were treated with 100 nM 17␤-estradiol (E2), dexamethasone (Dex), and R1881, respectively, and harvested. In each case, receptor transcriptional activity shown is normalized to reporter activity in the absence of the receptor. The data shown represent experiments done in duplicate that have been repeated at least twice with similar results. transferase also increased ER transcriptional activation in MCF-7 and Ishikawa cells (not shown). Likewise, inhibition of endogenous RhoA by expression of a dominant negative form of RhoA (RhoA.N19) also results in greater ER transactivation (Fig. 7B). Thus, inhibition of RhoA results in enhanced ER transcriptional activation, indicating that Rho-mediated signaling events suppress ER transactivation. DISCUSSION We have demonstrated that Rho GDI␣ enhances the transcriptional activity of the ER␣ as well as ER␤, GR, and AR but not SRF or Sp1. We also show that activated mutant forms of RhoA, Rac1, and Cdc42 decrease, whereas inhibition of endogenous RhoA by C3 transferase or dominant negative RhoA increases ER transcriptional activation. From these results, we conclude that the enhanced ER transactivation observed upon Rho GDI␣ overexpression is mediated by antagonism of Rho GTPases and implicates the Rho family proteins RhoA, Rac1, and Cdc42 in signaling to ER.
What is the mechanism underlying the modulation of ER transactivation by RhoA? Since Rho GTPases mediate actin cytoskeleton reorganization as well as the activation of multiple signaling pathways, such as c-Jun N-terminal kinase (JNK) and p38, Rho-mediated inhibition of ER may result from either of these events. The ability of the Rho GTPase family members to repress ER transcriptional activity suggests that RhoA-, Rac1-, and Cdc42-mediated signaling to ER may converge at some common point through a shared signaling molecule. An attractive candidate for such a common regulator is LIM kinase (53). The GTP-bound forms of RhoA and Rac1/Cdc42 activate LIM kinase via phosphorylation through effector kinases ROCK and Pak, respectively (54,55). The activated LIM kinase phosphorylates cofilin, an actin-binding protein, thereby inhibiting its actin-depolymerizing activity and leading to the accumulation of actin filaments. Recently, it has been shown that changes in the actin cytoskeleton can affect transcriptional activation by SRF (56). In a model reminiscent of that proposed for regulation of SRF by actin (56), we speculate that suppression of ER transactivation could result either from releasing an ER corepressor that is associated with free G-actin or from binding a coactivator to actin filaments, thus preventing its interaction with the ER. One such putative actin-regulated factor is the SWI/SNF complex, which has previously been shown to be a coactivator for steroid receptors, including ER, in both yeast and mammalian cells (57)(58)(59) and contains ␤-actin as well as two actin-related protein subunits (60 -62). We are currently testing whether SWI/SNF and/or LIM kinase mediate the modulatory effects of Rho GTPases on ER transactivation. While the effect of actin cytoskeletal changes on ER remains unknown, actin dynamics may provide a means of modulating ER transcriptional activity during normal development or in pathological settings, such as tumor progression, when cells undergo extensive actin reorganization.
Alternatively, changes in ER transcriptional regulatory properties may result from the activation of signal transduction pathways by Rho GTPases. For example, Rac1/Cdc42 activate JNK and p38 mitogen-activated protein kinase pathways, which may affect ER or its coregulatory factors via phosphorylation. Unlike Rac1/Cdc42, RhoA is not thought to activate the JNK and p38 pathways; therefore, it is unlikely that the activated Rho GTPases are effecting ER transactivation via these pathways. Nevertheless, since we have not excluded the possibility that JNK and p38 are mediating the effect of the Rho GTPases on ER transcriptional activation, we are currently testing the impact of activation and inhibition of JNK and p38 on receptor transactivation.
A cellular activity induced by activated RhoA, Rac1, and Cdc42 is NF-B, which has been shown to inhibit steroid receptor transactivation by forming inhibitory heterocomplexes (63,64). It is tempting to speculate that the inhibition of ER by the Rho GTPases is mediated by NF-B. However, our preliminary findings suggest that inhibition of NF-B by overexpressing IB does not relieve the repressive effects of Rho GTPases on ER transactivation (not shown), suggesting that Rho GTPases regulate ER independent of NF-B.
Recently, an ER-interacting protein, termed Brx, was identified and shown to contain a domain virtually identical to the Rho GEF Lbc, although its enzymatic activity has not been demonstrated (26). Overexpression of Brx in Ishikawa cells increases ER transcriptional activation, and a dominant nega- tive form of Cdc42, but not RhoA or Rac1, reduces its coactivator function (26). Our results differ from this report in assessing the effect of Rho GTPases on ER transactivation and showing that RhoA, Rac1, and Cdc42 negatively regulate ER transcriptional activity. This apparent discrepancy between what would be predicted from Rubino et al. (26), that Cdc42 increases ER transactivation, and our results showing that Cdc42, RhoA, and Rac1 decreased ER transcriptional activation may be attributed to methodological or cell-specific differences. Alternatively, since Brx coactivator function is probably mediated by direct ER binding, the inhibition of ER activity by the dominant negative form of Cdc42 may have resulted from competition between ER and dominant negative Cdc42 for Brx binding, rather than from blocking the signaling pathway downstream of Brx. In contrast, the effect of Rho GDI and the Rho GTPases on ER appears to be indirect. Localization studies indicate that ER and Rho GDI are found in distinct subcellular compartments, with Rho GDI residing in the cytoplasm, whereas ER is confined to the nucleus. 3 In addition, GST-Rho GDI is unable to associate with estradiol-bound ER, although it is capable of binding Rho A in vitro. 3 While Brx and Rho GTPases may modulate ER activity through distinct mechanisms, the identification of different components in the Rho signaling pathway as modulators of ER transactivation underscores their importance in receptor regulation.
Our findings suggest that the Rho GTPases decrease transcriptional activation by ER␣, thus establishing a novel pathway of cross-talk between cell surface receptors that regulate Rho GTPase signaling and steroid receptor transcriptional activation. Another example of cross-talk between the cell surface and ER is the modulation of ER ligand-independent transcriptional activation by the epidermal growth factor/Ras/mitogenactivated protein kinase signaling pathway (65,66). It has been shown that treatment of cells with epidermal growth factor results in ER ligand-independent activation and phosphorylation by the mitogen-activated protein kinase, Erk1 (31,67). Although the mechanism of this increased ER transcriptional activation remains to be elucidated, it probably involves phosphorylation-dependent cofactor recruitment (68). Thus, Ras acts as a positive regulator of ER transcriptional enhancement (67), 3 whereas Rho GTPases suppress receptor transactivation. We speculate that the opposing actions of Ras and Rho GTPases on ER-mediated transcriptional activation provide a means of fine tuning the ER transcriptional response to changes in the extracellular environment.