RhoA Prenylation Is Required for Promotion of Cell Growth and Transformation and Cytoskeleton Organization but Not for Induction of Serum Response Element Transcription*

The importance of post-translational geranylgeranylation of the GTPase RhoA for its ability to induce cellular proliferation and malignant transformation is not well understood. In this manuscript we demonstrate that geranylgeranylation is required for the proper cellular localization of V14RhoA and for its ability to induce actin stress fiber and focal adhesion formation. Furthermore, V14RhoA geranylgeranylation was also required for suppressing p21WAF transcription, promoting cell cycle progression and cellular proliferation. The ability of V14RhoA to induce focus formation and enhance plating efficiency and oncogenic Ras anchorage-dependent growth was also dependent on its geranylgeranylation. The only biological activity of V14RhoA that was not dependent on its prenylation was its ability to induce serum response element transcriptional activity. Furthermore, we demonstrate that a farnesylated form of V14RhoA was also able to bind RhoGDI-1, was able to induce cytoskeleton organization, proliferation, and transformation, and was just as potent as geranylgeranylated V14RhoA at suppressing p21WAF transcriptional activity. These results demonstrate that RhoA geranylgeranylation is required for its biological activity and that the nature of the lipid modification is not critical.

Small G proteins of the Ras superfamily are regulatory proteins whose activity is controlled by a GDP/GTP cycle. Several members of the Ras superfamily are regulators of signaling pathways that control cell growth, differentiation, and oncogenic transformation as well as actin cytoskeletal organization (1). The Rho protein branch of this superfamily includes at least eight distinct Rho families (RhoA, B, C, D, and G, Rac1 and 2, TC10, Cdc42, and Rnd1, 2, and 3) (2) that are regulated by Rho-GTPase activating proteins and a large family of guanine nucleotide exchange factors of the Dbl family proteins. Moreover Rho guanine nucleotide dissociation inhibitors (RhoGDIs) 1 stabilize the inactive GDP-bound form of the Rho proteins. Rho proteins notably regulate signal transduction from cell surface receptors to intracellular molecules and are involved in a variety of cellular processes including cell morphology (3), motility (4), cytokinesis (5,6), cell proliferation (7,8), and tumor progression (9 -11).
Ras and Rho proteins are post-translationally modified by the isoprenoid lipids, farnesyl, and geranylgeranyl (12). Two prenyltransferases, farnesyltransferase (FTase) and geranylgeranyltransferase I (GGTase I), catalyze the covalent attachment of the farnesyl and geranylgeranyl groups, respectively, to the carboxyl-terminal cysteine of proteins ending in a CAAX motif (C is a cysteine, A usually aliphatic amino acid, and X any amino acid). FTase prefers CAAX sequences where X is a serine, methionine, cysteine, alanine, or glutamine, as in Ras or in nuclear lamins (13)(14)(15). When X is a leucine or isoleucine the protein, as in the Rho/Rac family of proteins, is geranylgeranylated by GGTase I (16,17). Protein prenylation is important in targeting proteins to cellular membranes but also in protein-protein interactions (18 -20). This process appeared to be critical for the oncoprotein Ras functions as observed with the dependence of its transforming activity on prenylation (21,22). Hence, over the past decade the functional role of protein prenylation has been intensively studied for Ras, while a few studies took interest in the other proteins. It has been shown that geranylgeranylation of RhoA is required for its correct subcellular localization (23) and for interaction with its GDP/GTP cycle regulators, guanine nucleotide dissociation inhibitor and guanine nucleotide exchange factor (24,25). Furthermore, RhoA prenylation is needed for phospholipase D activation (26,27) and potentiation of AP-1 transcription (28). However, although it was demonstrated that the prenylation of the highly homologous Rho protein, RhoB, is required for its cell transforming function but not its ability to activate serum response element-dependent transcription (29), whether RhoA geranylgeranylation is required for suppression of p21 WAF , induction of proliferation, and cytoskeleton organization was still not established. It was only shown that inhibition of protein geranylgeranylation by the GGTase I inhibitor, GGTI-298, resulted in G 0 /G 1 cell cycle arrest, induction of p21 WAF transcrip-tion, programmed cell death, and disorganization of actin cytoskeleton (30 -32). Similar effects observed with the Clostridum botulinum C3 exoenzyme, a specific inhibitor of Rho A, B, and C proteins (32)(33)(34), as well as with a dominant negative mutants of RhoA (7,32) have suggested a role of Rho proteins in these process.
Specific isoprenoids may facilitate distinct consequences for protein functions. Hence, normal Ras function is critically dependent on modification by a farnesyl group, and a geranylgeranylated normal Ras protein appeared to be a potent inhibitor of cellular proliferation (35). In contrast either a C15 or a C20 isoprenoid can promote membrane interaction of the oncoprotein Ras necessary for triggering the cell transformation (35)(36)(37). On the other hand, it was demonstrated that a specific prenylation of the ␥ subunits of G protein is needed for the membrane localization, interaction with the G␤ and their effectors (38 -42). Thus an evaluation of the role of specific isoprenoid modification in the function of other prenylated protein such as Rho GTPase would be needful for a better understanding of the functional role for specific isoprenoid modification of a protein.
In this manuscript, we determined whether RhoA prenylation is required for its activity on cytoskeleton organization, proliferation, transcription, and transformation. Furthermore, we also investigated whether the nature of the prenyl group (i.e. farnesyl versus geranylgeranyl) influences the biological activities of RhoA. To this end, we generated RhoA mutants by deleting the CAAX sequence to produce an unprenylated protein as well as by mutating the CAAX box to produce farnesylated RhoA.
The coding region of RhoGDI-1 was isolated by reverse transcriptionpolymerase chain reaction from human fibroblasts by using specific primers (forward primer, 5Ј-GCTAAGCTTGGGATCCGCTGAGCAGG-AG-3Ј, and reverse primer, 5Ј-CCGGAATTCGGCTCAGTCCTTCCAG-TCCTTCTTGATG-3Ј) and subcloned into the bacterial glutathione Stransferase (GST) expression vector pGST-Parallel2 (generously provided by P. Sheffield) as an BamHI/EcoRI fragment. pSRE was provided by Dr. R. Jove (H. Lee Moffitt Cancer Center, Tampa, FL) and p21P containing the full-length sequence of p21 WAF promoter was provided by Dr. X.-F. Wang (Duke University Medical Center, Durham, NC). pCMV-␤gal was used to normalize for transfection efficiency.
Cell Culture and Transfection-COS-7 and NIH-3T3 cells were grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% FCS, at 37°C in a humidified incubator containing 5% CO 2 . COS-7 cells and NIH-3T3 cell lines transiently and stably expressing, respectively, the RhoA mutants Ha-RasL61 or Ha-RasL61/V14RhoA were generated by transfection using LipofectAMINE TM Reagent Plus method as indicated by the supplier (Life Technologies, Inc.), followed by selection for NIH-3T3 cell lines with 200 g/ml of Zeocin TM (Cayla S.A.) for RhoA, 1 mg/ml of Geneticin® (Life Technologies, Inc.) for Ha-RasL61 and 200 g/ml of Zeocin TM plus 1 mg/ml of Geneticin® for Ha-RasL61/RhoA. Cell clones were expanded into mass culture, and RhoA and Ha-RasL61 protein expression were analyzed by Western blotting (as described below).
In Vitro Interaction of RhoA with RhoGDI-1-GST-RhoGDI-1 fusion protein expression and purification was done as described by Sheffield et al. (44). For in vitro interaction assay 30 l of glutathione-Sepharose beads (Amersham Pharmacia Biotech) were preincubated with 100 g of bacterial overexpressed GST-RhoGDI soluble extract for 2 h at 4°C, washed with interaction buffer (50 mM Tris-HCl pH 8.5, 150 mM NaCl), then incubated with 100 g of cell homogenates diluted in 500 l of interaction buffer, and incubated overnight at 4°C. Sepharose beads were next washed twice with interaction buffer and resuspended directly in Laemmli buffer. RhoA content was analyzed by SDS-polyacrylamide gel electrophoresis followed by Western blot with anti-Myc antibodies.
Fluorescence-Cells were seeded on glass coverslips into six-well plates (Nunc) at a density of 8 ϫ 10 4 cells/well in DMEM 10% FCS. 48 h later cells were serum-starved for 48 h. Then cells were fixed in 3% paraformaldehyde and permeabilized into 0.1% Triton X-100 in phosphate-buffered saline. Actin fibers were detected by incubation with tetramethylrhodamine isothiocyanate-labeled phalloidin (Molecular Probes). RhoA and vinculin were detected with anti-c-Myc (Calbiochem) or anti-vinculin antibodies (Sigma Immuno Chemical), respectively, and fluorescein isothiocyanate-conjugated anti-mouse IgG antibody (Sigma Immuno Chemical). Cells were viewed on a Zeiss Axiophot microscope, and pictures were taken with a Princeton camera.
Cell Growth Determination-NIH-3T3 cells were seeded at 2000 cells/well in a 96-well plates six times on day 0 in DMEM containing 10% or 2.5% FCS, and the amount of cells was evaluated on day 0 (6 h post-plating) and at the intervals indicated in the figure legends by a MTT test as described previously (45).
Focus Formation Assay-NIH-3T3 cell lines expressing the V14RhoA constructs were seeded at 25000 cells in a 25-cm 2 flask. Cell foci were scored 15 days after confluence after fixing with AFA (ethanol/formol/ acetic acid, 75/20/5) and staining with crystal violet.
Plating Efficiency Determination-For plating efficiency assays, NIH-3T3 cells were seeded at 400 cells/60-mm culture dishes in DMEM 10% or 2.5% FCS. Cell plating efficiency was scored 10 or 15 days later, respectively, by fixing with AFA and staining with crystal violet.
Anchorage-independent Growth Assays-Cells were seeded at 16,000 cells/well in 12-well plates in triplicate in 0.3% agar over a 0.6% agar layer as described elsewhere (46). Cells were fed twice weekly until colonies grew to a suitable size for observation (about 12 days). Colonies were photographed after 4 h of incubation with 1 mg/ml MTT in DMEM at 37°C. The growth of colonies of cell lines expressing V14RhoA or Ha-RasL61/V14RhoA constructs was compared with the control colonies expressing Ha-RasL61.
Regulation of SRE and p21 WAF Promoter Activities-NIH-3T3 cells were seeded at 2 ϫ 10 5 /well in six-well plates. 24 h later they were transfected with 0.5 g of pSRE or p21 WAF , 0.2 g of pCMV-␤gal and 1 g of pCMV-V14RhoA-CLVL, pCMV-V14RhoA-CVLS, or pCMV-V14RhoA-⌬ using LipofectAMINE TM Reagent Plus as indicated by the supplier. 15 h after transfection, the cells were replenished with fresh growth medium. For p21 WAF promoter activity analysis cells were har-vested 30 h later and lysed in 200 l of lysis buffer (Promega). For SRE activity analysis, 24 h after transfection cells were washed twice with phosphate-buffered saline and incubated in DMEM supplemented with 0.5% FCS during the next 24 h, before harvesting and lysis. Cell extracts were used for ␤-galactosidase (CLONTECH) and luciferase (Promega) assays.

Prenylation Status and Subcellular Localization of RhoA Mutants in NIH-3T3 and COS-7 Cell
Transfectants-To evaluate the role of prenyl group and of its nature (C15 farnesyl or C20 geranylgeranyl) in RhoA functions we generated RhoA mutants by altering the CAAX sequence to render RhoA a substrate either of geranylgeranyltransferase I or farnesyltransferase or not prenylated. Using standard polymerase chain reaction mutagenesis, the CLVL sequence of RhoA (Rho-CLVL) was either replaced by the CAAX box of the farnesylated Ha-Ras protein (RhoA-CVLS) or deleted (RhoA-⌬). The plasmids encoding these RhoA mutants were transfected in the murine NIH-3T3 fibroblasts. Expression of RhoA was controlled by Western blot in 10 different clones picked for each construction. Representative clones with similar expression level to each other were selected (Fig. 1).
The prenylation status of a protein being important for its proper subcellular localization (23), we checked the localization of the RhoA mutants by immunofluorescence. V14RhoA-CLVL cells displayed a diffuse staining throughout the cytoplasmic compartment ( Fig. 2A) with an increased staining in the perinuclear area of some cells. RhoA-CVLS cells showed a similar pattern of fluorescence ( Fig. 2A). In contrast RhoA-⌬ staining was not confined to a specific subcellular compartment but was rather spread throughout the whole cell ( Fig. 2A). This pattern was also observed when all of RhoA mutants were treated with lovastatin, a potent inhibitor of prenylation (data not shown), confirming that prenylation is essential for proper protein subcellular localization.
We next analyzed the subcellular localization of RhoA mutants by Western blot after separation of cell homogenates in cytosol and membrane proteins by Triton X114 partitioning as described under "Experimental Procedures." As shown in Fig.  2B unprenylated V14RhoA was essentially associated with the cytosolic fraction, whereas prenylated RhoA, either farnesylated or geranylgeranylated, was distributed in both fractions. About 40% of prenylated V14RhoA expressed in NIH-3T3 cells was observed in membrane fraction. When analyzed in COS-7 cells after transient expression, prenylated V14RhoA display a similar subcellular distribution, whereas wild type RhoA was essentially found in cytosol fraction (data not shown), illustrating the shift toward the membrane of RhoA GTPase after activation but only when prenylated. These results together confirmed that prenylation is essential for proper protein subcellular localization and also indicated that farnesyl can sub- stitute for geranylgeranyl without any dramatic effect on RhoA subcellular localization.
Interaction of RhoA with its Guanine Dissociation Inhibitor, RhoGDI-1-Although it was well established that isoprenoid modification of RhoA is required for interaction with RhoGDI (24,47), it is still not known whether prenylation of RhoA has to be specific. To assess the role of isoprenoid the in vitro interaction of RhoA mutants with GST-RhoGDI-1 was examined. Because it was described that RhoGDI-1 binds poorly RhoA-GTP (48,49), we used COS-7 cells transfected transiently with wild type RhoA bearing or deleted of the different CAAX boxes. We found as expected that nonprenylated RhoA was unable to bind GST-RhoGDI-1 (Fig. 3). In contrast geranylgeranylated as well as farnesylated RhoA strongly interacted with GST-RhoGDI-1. As illustrated by the quantitation of the signal of precipitated RhoA normalized by the signal of total RhoA on Western blot, GST-RhoGDI-1 appeared to bind equal levels of both farnesylated and geranylgeranylated RhoA (Fig. 3).
Role of RhoA Prenylation on Cytoskeleton Organization-The Rho protein family whose members include RhoA has been shown to influence a number of cellular processes including actin stress fiber organization, cell adhesion, cell proliferation, and transformation. We next determined whether prenylation and moreover the nature of the added prenyl group play a role on RhoA implication on actin stress fibers and focal adhesions.
We observed that whereas NIH-3T3 fibroblasts transfected with the empty vector (mock) or V14RhoA-⌬ have conserved a typical fibroblast morphology, V14RhoA-CLVL and V14RhoA-CLVS expressing cells were smaller and displayed a more epithelial-like morphology (Fig. 4A). Similar changes of cell size were observed in Swiss 3T3 cells microinjected with active RhoA (3) as well as in murine tumor cells (50).
To examine the effect of the V14RhoA mutants on stress fibers and focal adhesions, the cells were serum-starved for 48 h before analysis. Indeed removal of growth factors and LPA (a well known stimulator of RhoA functions) from culture medium led to the reduction of actin stress fiber content in mock and V14RhoA-⌬ cells (Fig. 4B). In contrast, under these conditions V14RhoA-CLVL and V14RhoA-CLVS cells displayed well preserved stress fibers (Fig. 4B). In parallel experiments, vinculin immunostaining showed that in V14RhoA-CLVL and V14RhoA-CLVS but not V14RhoA-⌬ expressing cells a retention of numerous focal adhesions in the absence of serum (Fig. 4B). These results would indicate that prenylation is required for RhoA implication in stress fibers and focal adhesions and that geranylgeranylation could be substituted for by farnesylation.
Role of Prenylation of RhoA on Cell Growth Properties-We next assessed the role of the prenylation of RhoA on cell growth. Mock-transfected NIH-3T3 cells and those expressing V14RhoA-CLVL, V14RhoA-CVLS, or V14RhoA-⌬ maintained in DMEM supplemented with 10% FCS displayed similar growth rate (Fig. 5A). However, in lower serum concentration (2.5%), V14RhoA-CLVL and -CVLS cells showed a significantly higher growth rate than mock and V14RhoA-⌬ cells (Fig. 5B), indicating that prenylated V14RhoA expression reduced the serum requirement of the NIH-3T3 cells as described by Perona et al. (9).
We next determined whether the ability of RhoA to transform cells depends on its geranylgeranylation and whether farnesylated RhoA remains transforming. To this end, we have compared various V14RhoA mutants in focus formation, plating efficiency, and anchorage-independent growth assays. For the focus formation assay, NIH-3T3 cells stably expressing V14RhoA mutants were seeded at 25,000 cells/25-cm 2 flask, and cell foci were scored 12 days later. Whereas mock and V14RhoA-⌬ cells grew in a monolayer, V14RhoA-CLVL or CVLS cells grew at higher density and formed numerous foci, indicating that expression of only prenylated forms of RhoA led to a loss of contact inhibition (Fig. 6A). Moreover, only prenylated V14RhoA were able to develop clones after plating in drastic conditions (400 cells in 60-mm dishes) (Fig. 6B). But we observed in each of these experiments that a fewer number of clones grew from cells expressing farnesylated RhoA than geranylgeranylated RhoA. When testing on anchorage-independent growth by soft agar cultures, V14RhoA-CLVL and V14RhoA-CVLS as well as V14RhoA-⌬ expression in NIH-3T3 cells did not induce growth (data not shown). However, prenylated V14RhoA but not V14RhoA-⌬ expression increased signifi-cantly the anchorage-independent growth of oncogenic Ha-RasL61 transformed NIH-3T3 cells (Fig. 7).
Role of RhoA Prenylation on Transcriptional Activity-RhoA was described to stimulate the SRE activity (51) as well as to repress p21 WAF transcription (32). Our recent results suggested indirectly that geranylgeranylation of RhoA was required for its activity on p21 WAF transcription (32). To confirm this hypothesis we compared the respective ability of V14RhoA-CLVL, -CVLS, and -⌬ to regulate luciferase transcription under the control of p21 WAF promoter. We initially cotransfected NIH-3T3 wild type cells with p21 WAF promoter and RhoA expression plasmids. Aliquots of cell homogenates were assayed 48 h after transfection for ␤galactosidase and luciferase activities. As we have shown previously RhoA-CLVL induced an important (4-fold) repression of p21 WAF promoter activity (Fig. 8A). Furthermore, the farnesylated RhoA-CVLS also displayed a similar effect on p21 WAF activity, but the unprenylated form of RhoA, RhoA-⌬, was not able to regulate the activity of the promoter (Fig. 8A). Thus, the prenylation of RhoA is critical to its ability to repress the p21 WAF transcriptional activity. Furthermore, farnesylated RhoA was just as potent as geranylgeranylated RhoA at repressing p21 WAF transcription.
To determine whether or not prenylation was required for the role of RhoA on SRE-mediated transcription, we have performed similar experiments co-transfecting NIH-3T3 wild type cells with luciferase under the control of SRE and RhoA expression plasmids. Cells were incubated 24 h post-transfection with DMEM supplemented with 0.5% FCS for the next 18 h. Then as a control of SRE activity FCS was added for 4 h. As expected, addition of FCS resulted in a 4 -5-fold stimulation of luciferase activity, whereas the expression of constitutively active geranylgeranylated RhoA induced a 3-fold increase of luciferase activity (Fig. 8B). Then to directly compare the capacity of the prenylated and unprenylated RhoA to regulate SRE transcriptional activity, the luciferase induction was normalized by RhoA protein expression determined in parallel by Western blot. As shown in Fig. 8C, the geranylgeranylated RhoA-CLVL and the farnesylated RhoA-CVLS displayed a similar effect on SRE activity. Surprinsingly RhoA-⌬ showed also a stimulatory activity on SRE dependent transcription; however, this unprenylated form of RhoA was less efficacious than the geranylgeranylated and the farnesylated forms.

DISCUSSION
Although a number of studies showed that functions of Ras proteins depend on its farnesylation little is known about the role of prenylation for geranylgeranylated proteins. Using CAAX box mutants of RhoA, we demonstrated that prenylation of RhoA is essential for most but not all of its cellular functions. Furthermore, replacing the geranylgeranyl group by a farnesyl had little effect on the biological activities of RhoA.
The carboxyl-terminal amino acid sequence of the protein, the CAAX box, appears to contain all the critical determinants for interaction with a specific prenyl-transferase (15). The only substitution of the last three carboxyl-terminal amino acids of prenylated proteins such as Ras (15), ␥ subunit of G protein (40) or of RhoB as we recently described (52), 2 which could be either geranylgeranylated or farnesylated, is sufficient to mod-ify the nature of their prenylation. Thus, to create RhoA prenylation mutants we used a similar strategy deleting the CAAX box or substituting the RhoA CLVL sequence by CVLS, the CAAX box of Ha-Ras, a protein exclusively farnesylated. This RhoA-CVLS mutant appeared to be a substrate for the FTase because FTI-277, a highly specific inhibitor of the enzyme, but not GGTI-298, blocked its prenylation.
Prenylated RhoA was detected in the cytoplasmic compartment in NIH-3T3 cells stably transfected with RhoA, as previously observed in Rat-2 cells or in Madin-Darby canine kidney cells (23,53). Adamson et al. (23) showed that only the cysteine within the CAAX box is important for the correct intracellular localization of RhoA protein. Indeed its substitution for serine lead to a marked reduction of the normal cytoplasmic fluorescence in Madin-Darby canine kidney cells in favor of a strong nuclear localization. We obtained similar results in NIH-3T3 transfected with RhoA deleted of its CAAX box. Whereas the prenylated RhoA localized preferentially in the cytosol, the RhoA-⌬ mutant accumulated in whole cells with a predominant nuclear localization as observed either with the unfarnesylated form of Ras (54) or RhoB (11,23). A similar subcellular redistribution of unprenylated protein to the nucleus was also reported for protein tyrosine phosphatases PRL-1, -2, and -3 in NIH-3T3 cells (55). Numerous cells displayed a marked staining of RhoA in the perinuclear area that might suggest that RhoA is associated with some vesicles. Indeed after separation of membrane and cytosolic proteins, a large proportion of V14RhoA appeared to be associated with membrane fractions. However, RhoA was not fully associated with plasma membrane as it was observed in NE-115 neuronal cells (56) or endothelial cells (57). The pattern of perinuclear fluorescence suggested that RhoA might accumulate in the endoplasmic reticulum or endosomal vesicles as described previously (58). Nevertheless, it is noteworthy that regardless of the prenyl group, a C20 geranylgeranyl or a C15 farnesyl, linked to the RhoA protein, NIH-3T3 transfected cells displayed similar RhoA immunofluorescence patterns, suggesting that the shorter farnesyl group is sufficient for directing the intracellular destination of RhoA.
It was well established that regulators of small G protein activity interact only with the post-translationally modified protein (4,24,25,59). For example RhoGDI inhibits the dissociation of the GDP bound form of RhoA only if the protein is prenylated (24). Similar prenyl-dependent protein interactions should be required for the association of the RhoA-GTP bound form with its effectors and thus for its biological roles such as actin cytoskeleton organization and cell proliferation and transformation control. Nevertheless the role of isoprenylation in the control of stress fibers formation was still contradictory. Although it was shown that HMG-CoA reductase inhibitors induce cell rounding and breakdown of the actin cytoskeleton in cultured cells (60 -62), suggesting a role of prenylated protein in stress fiber control, Kranenburg et al. (56) showed that isoprenylation is not required for stress fiber formation in N1E-115 neuronal cells. Herein we showed with a CAAX box mutant that prenylated RhoA ensured stress fiber and focal adhesion maintenance in NIH-3T3 cells, whereas unprenylated form of RhoA did not. Furthermore we demonstrated that the role of RhoA on NIH-3T3 cell proliferation and transformation was also dependent on the prenyl group. Hence, the unprenylated form of RhoA was not able to promote cell proliferation or anchorage-dependent cell growth, whereas farnesylated and geranylgeranylated RhoA acted similarly on cell cytoskeleton and proliferation.
RhoA protein has been suggested to regulate cell cycle progression by modulating the protein stability of cell cycle regu- lators such as p27 KIP1 (63) or transcription of specific genes such as cyclin D1 (64), c-fos (51), or p21 WAF (32). We found (32) that the expression of constitutively activated RhoA downregulated p21 WAF promoter. Similarly Marshall and co-workers (65) showed that induction of DNA synthesis by oncogenic Ras required Rho activity for the suppression of p21 WAF induction. As for its role on cell proliferation, we showed that overexpression of only prenylated RhoA was necessary to down-regulate p21 WAF promoter, whereas the unprenylated form was inert.
It has been shown that the small G proteins Cdc42, Rac1, RhoA, and RhoB can also activate the transcription of c-fos via the SRE site of its promoter in a manner dependent upon SRF (29,51). Whereas prenylation of RhoB is required for its cell transforming functions, its ability to activate SRE-mediated transcription is independent of prenylation (29). Although RhoA is ϳ85% identical to RhoB, it is not known whether unprenylated RhoA also would be able to induced SRF-mediated transcription. In this report, we showed that despite their different subcellular localization, prenylated and unprenylated forms of RhoA were capable to induce SRE-mediated transcriptional activity, as it has been shown for RhoB (29). But although Lebowitz et al. (29) observed that both prenylated and unprenylated forms of RhoB were equipotent at inducing transcription, we showed that the unprenylated form of RhoA was less efficacious that the prenylated form. Some possible explanations of this discrepancy between prenylation requirement for p21 WAF and SRE transcription would be that active RhoA may find effectors that trigger the SRF signaling pathway in many compartments of the cell or that because of its overexpression, low quantities of unprenylated RhoA could localized properly. However similar results should be expected for the down-regulation of p21 WAF promoter. By contrast the prenylation of RhoA is required for this later effect. This suggests that the signaling pathways triggering either p21 WAF or SRE transcription are separable and that the interaction with effectors for p21 WAF transcription requires prenylation of RhoA, whereas interaction with effectors for SRE-mediated transcription do not. Moreover, our results are in agreement with the hypothesis previously advanced by Lebowitz et al. (29) that SRF-mediated transcriptional activation by Rho proteins is insufficient for transformation.
Although the farnesylated and geranylgeranylated RhoA displayed a similar localization, the presence of the farnesyl group could impair some of RhoA functions and/or may dictate different biological functions as suggested for RhoB protein (66). But our data did not evidence a significative difference between the farnesylated and the geranylgeranylated RhoA behavior on cell cytoskeleton, growth properties, or transcription. The only faint differences we noticed were on cell plating efficiency and SRE transcription, where geranylgeranylated RhoA appeared a little more efficacious. Likely it was demonstrated that the farnesylated and geranylgeranylated forms of Ras or Rap1A/Krev-1 displayed similar functions on cell transformation (35). In contrast it appeared a marked distinction between farnesylation and geranylgeranylation in G␥ protein interactions and functions (38 -42), such as the more effective interaction of farnesylated than geranylgeranylated ␥1 with light-activated rhodopsin (38). It is thus possible that each isoprenoid imparts either a distinct or similar function to proteins as a result of the specificity of interaction between the prenylated proteins and their partners. For instance farnesylated or geranylgeranylated G␥ subunits did not display the same biological properties, whereas for other prenylated proteins such as small G proteins, Ras, or Rho, either a farnesyl or a geranylgeranyl moiety led to similar functions. Indeed the interaction of RhoGDI-1 with RhoA, while requiring the pres- FIG. 8. The prenylation of RhoA is required for suppressing on p21 WAF transcription but not for inducing SRE-mediated transcription. A, NIH-3T3 wild type cells were cotransfected with luciferase under the control of p21 WAF promoter, V14RhoA mutants, and ␤-galactosidase plasmids. 48 h later cell homogenates were prepared, and then ␤-galactosidase activity and luciferase activity were assayed. B and C, NIH-3T3 wild type cells were cotransfected with luciferase under the control of SRE, V14RhoA mutants, and ␤-galactosidase plasmids. 24 h post transfection cells were incubated in 0.5% FCS containing medium for additional 18 h. Where indicated 10% FCS was added 4 h before the end of the incubation. In C the induction factor of luciferase activity was normalized against the RhoA mutant expression analyzed by Western blotting. In each experiment the luciferase activity was normalized for transfection efficiency against ␤-galactosidase activity and expressed as relative luciferase. Bars represent standard deviation. The data are representative of three independent experiments. ence of a prenyl group, is not affected by its nature, i.e. farnesyl or geranylgeranyl, as illustrated by the equal amounts of farnesylated and geranylgeranylated RhoA precipitated by GST-RhoGDI-1 (Fig. 3). Moreover, it was shown that the binding affinity of RhoGDI-1 for the farnesylated and geranylgeranylated moiety is not really important (42).
We described previously a potent and selective GGTase I inhibitor GGTI-298 that blocks human tumors in G0/G1, induces apoptosis, and inhibits the tumor growth in nude mice xenografts (30,46,67). More recently we showed that GGTI-298 strongly induces p21 WAF accumulation in tumor cells (31) and suggested that RhoA is a target for GGTI-298 (32). Here we demonstrated that the prenylation of RhoA is necessary for its transforming activity, enforcing the idea that RhoA is an important target of GGTase I inhibitors.