RhoA-GDP Regulates RhoB Protein Stability

RhoA plays a significant role in actin stress fibers formation. However, silencing RhoA alone or RhoA and RhoC did not completely suppress the stress fibers suggesting a residual “Rho-like” activity. RhoB, the third member of the Rho subclass, is a shortlived protein barely detectable in basal conditions. In various cell types, the silencing of RhoA induced a strong up-regulation of both total and active RhoB protein levels that were rescued by re-expressing RhoA and related to an enhanced half-life of the protein. The RhoA-dependent regulation of RhoB does not depend on the activity of RhoA but is mediated by its GDP-bound form. The stabilization of RhoB was not dependent on isoprenoid biosynthesis, Rho kinase, extracellular signal-regulated kinase, p38 mitogen-activated kinase, or phosphatidylinositol 3′-OH kinase pathways but required RhoGDIα. The forced expression of RhoGDIα increased RhoB half-life, whereas its knock-down antagonized the induction of RhoB following RhoA silencing. Moreover, a RhoA mutant (RhoAR68E) unable to bind RhoGDIα was significantly less efficient as compared with wild-type RhoA in reversing RhoB up-regulation upon RhoA silencing. These results suggest that, in basal conditions, RhoGDIα is rate-limiting and the suppression of RhoA makes it available to stabilize RhoB. Our results highlight RhoGDIα-dependent cross-talks that regulate the stability of RhoGTPases.

The small GTPases of the Rho family are at the cross-roads of signaling pathways initiated by receptors to diffusible biological mediators and those depending on cell-adhesion receptors. They are key signaling molecules regulating a plethora of biological pathways (1). The Rho GTPases shuttle between an inactive GDP-bound state and an active GTP-bound state. Their level of activation is regulated by three classes of factors: the guanine-nucleotide exchange factors that catalyze the exchange of GDP to GTP, the GAPs that increase the intrinsic GTPase activity of the RhoGTPase, and the RhoGDIs that inhibit the exchange of GDP to GTP. However, their mechanism of action may not be solely restricted to activation of downstream signaling cascades when GTP-loaded (2).
In mammals, RhoGDIs constitute a family encompassing three members: RhoGDI␣, the ubiquitously expressed archetypal member of the family; Ly/D4-GDI or RhoGDI␤, which has a hematopoietic tissue-specific expression pattern; and RhoGDI-3 or -␥, which is membrane-anchored and preferentially expressed in brain, pancreas, lung, kidney, and testis. RhoGDIs are usually perceived as "static" inhibitors preventing the activation of the downstream effectors by the RhoGTPases. Accumulative evidences suggest that RhoGTPase-RhoGDI complexes are highly dynamic. Phosphorylation of RhoGDI␣ by various kinases that decreases its affinity for RhoGTPases is one mechanism used by receptors to activate specific Rho-GTPases (3,4). By contrast, phosphorylation of the RhoGTPases themselves seems to increase their affinity for RhoGDI␣ thus leading to signal termination (5)(6)(7). Furthermore, a novel function has been attributed to RhoGDI␣ in the activation of NADPH oxidase by its capacity to present the RhoGTPase to the appropriate effectors in a way that potentiates efficient activation (8,9). More recently, it was shown that a Rac1 mutant displaying an increased affinity for RhoGDI␣ stimulated the activity of RhoA suggesting that RhoGDI is a key intermediate in the cross-talks between RhoA and Rac1 (10).
Among the RhoGTPase family, the Rho subclass includes RhoA, its closely related homolog RhoC, and RhoB. By contrast to RhoA and RhoC, RhoB is a short-lived protein displaying antitumorigenic properties (11). In various solid tumors, RhoB expression diminished in parallel with tumor progression (12,13). Its overexpression antagonizes cell migration, tumor growth, and metastasis (14,15). Furthermore, RhoB was reported to be an essential component of the anti-tumoral response triggered by farnesyl transferase inhibitors as well as of the apoptotic response of transformed cells to DNA damaging agents (16,17). Thus, a better understanding of the mechanisms regulating RhoB expression and stability is of great importance for the optimization of potential anti-cancer strategies.
In this report, we targeted RhoGTPases by using a siRNA 4based approach. This technology that recently allowed us to highlight the regulation operated by Cdc42 on matrix metalloprotease-1 (18) has several advantages as compared with classical methods. Beside a higher specificity, siRNA suppresses both GDP-and GTP-bound forms allowing to better evaluate the contribution of either form in the global function of individual RhoGTPase. Silencing RhoA led to an increased RhoB expression by a post-transcriptional mechanism extending the half-life of the protein. Our results demonstrated that this regulation is mediated by RhoA-GDP. Investigation of various intracellular signaling pathways suggested a role of a RhoGDI␣dependent mechanism in this process.
RT-PCR Analysis-The RT-PCR amplifications were performed in an automated thermocycler (GeneAmp PCR system 9600) using a GeneAmp Thermostable rTth Reverse Transcriptase RNA PCR kit (PerkinElmer Life Sciences) with pairs of primers amplifying mRNA coding for human RhoA (5Ј-GTA-CATGGAGTGTTCAGCAAAGACC-3Ј and 5Ј-GGTGGGCCA-GACGGGTTGGACA-3Ј), human mutated RhoA (mRhoA) (5Ј-GTACATGGAGTGTTCAGCAAAGACC-3Ј and 5Ј-AGAAGG-CACAGTCGAGGCTGATCA-3Ј), and 28 S rRNA (5Ј-GTTCA-CCCACTAATAGGGAACGTGA-3Ј and 5Ј-GGATTCTGACT-TAGAGGCGTTCAGT-3Ј). For the 28 S rRNA, the efficiency of the RT-PCR was controlled by a synthetic RNA co-transcribed and co-amplified with the same primers as the endogenous RNA to yield an amplification product of slightly larger size. The RT-PCR conditions were described elsewhere (18). Briefly, 10 ng of total RNA and a known copy number of the standard synthetic RNA were reverse transcribed (70°C for 15 min). Then, RNA-DNA heteroduplexes were denatured for 2 min at 94°C and amplification was carried out for 22 cycles (RhoA and mRhoA) or for 17 cycles (28 S rRNA) at 94°C for 15 s, 66°C for 20 s, and 72°C for 20 s (10 s for 28 S rRNA). The RT-PCR products were quantified after electrophoresis on a 10% polyacrylamide gel and staining (Gelstar, FMC BioProducts) using a Fluor-S TM MultiImager (Bio-Rad).
Real Time Quantitative PCR-Total RNA was isolated from cells 48 h after transfection with siRNA using the High Pure RNA isolation kit (Roche Applied Science). 100 ng of total RNA were reversed transcribed using SuperScript II Reverse Transcriptase (Invitrogen). Real time PCR was performed in a final volume of 20 l containing 2 l of cDNA (corresponding to 10 ng of total RNA for RhoB amplification and corresponding to 0.1 ng of total RNA for glyceraldehyde-3-phosphate dehydrogenase amplification), 300 nM of each primer and 10 l of the qPCR MasterMix Plus for SYBR Green (Eurogentec) in the Abi Prism 7000 Sequence Detection system (Applied Biosystems). The results were normalized to the glyceraldehyde-3phosphate dehydrogenase transcript. PCR was performed with the following primers: RhoB, forward, 5Ј-GCCACGCGCGCC-GCGCTGCA-3Ј, reverse, 5Ј-CCGGCAGGGGCAGGCGC-GAC-3Ј; and glyceraldehyde-3-phosphate dehydrogenase: forward, 5Ј-CCTGGCCAAGGTCATCCATGACA-3Ј, reverse, 5Ј-GGGATGACCTTGCCCACAGCCTT-3Ј.
Creation of the Various Expression Vectors and Cell Transfection-The entire coding sequence of human RhoA was amplified by RT-PCR. The amplification product was mutated by means of a PCR-based approach with mutated primers. Five silent mutations were introduced in the sequence targeted by the 1st siRhoA to make it resistant to this siRNA. The mutated RhoA cDNA (mRhoA) was cloned into pcDNA3 and pcDNA4/TO (Invitrogen). Sequencing confirmed that the five expected mutations were introduced into the cDNA. Rescue experiments with HS578T cells were carried out as previously described (18). Briefly, cells were first transfected with 6 nM siScr or 6 nM siRhoA for 14 -16 h following the protocol described above. Immediately after washing, each pool of cells was trypsinized and seeded in two wells of 6-well plates. Three hours later, 1 g of pcDNA4/TO or pcDNA4/TO/mRhoA were transfected into cells for 20 -24 h with 3 l of Genejuice TM (Novagen). Cells were washed and cultured for a further 24 h before being processed for immunoblotting analysis. Rescue experiments were also carried out with PC-3 cells engineered to express mRhoA in a tetracycline-dependent way. For this purpose, PC-3 cells were first transfected with pcDNA6/TR (Invitrogen) and selected in medium supplemented with 1 g/ml blasticidin. A clone expressing a high level of tetracycline repressor (PC-3/TR) was isolated. PC-3/TR cells were transfected with either the empty pcDNA4/TO or the pcDNA4/TO/mRhoA and selected in medium supplemented with 1 g/ml blasticidin ϩ 200 g/ml Zeocin. 3 clones of cells transfected with pcDNA4/TO (PC-3/TR/control) and 3 clones of cells transfected with pcDNA4/TO/mRhoA (PC-3/TR/ mRhoA) were isolated and amplified. For rescue experiments, the PC-3/TR/control and the PC-3/TR/mRhoA clones were seeded at subconfluence in medium without blasticidin and Zeocin. 24 h later, the cells were transfected with 6 nM siScr or 6 nM siRhoA for 14 -16 h following the procedure described above. Immediately after washing, each pool of transfected cells was trypsinized and separated in two parts, the first was cultured with normal medium and the second with medium supplemented with 1 g/ml tetracycline. 48 h later, cells were processed for immunoblotting analysis and a measure of the expression level of both endogenous and mutated RhoA mRNA by RT-PCR. The entire coding sequence of human RhoA was amplified by RT-PCR. A point mutation was introduced in the amplification product by means of a PCR-based approach with mutated primers to generate a cDNA encoding RhoAN19, the dominant negative form of RhoA. The entire sequence was cloned into pcDNA4/TO (Invitrogen). Sequencing confirmed that the expected point mutation was introduced into the cDNA. PC-3/TR cells were transfected with the pcDNA4/TO/RhoAN19 and selected in medium supplemented with 1 g/ml blasticidin ϩ 200 g/ml Zeocin. 3 clones (PC-3/TR/RhoAN19) expressing in a tetracycline-dependent way a dominant-negative RhoA were isolated and amplified. The functionality of the dominant-negative RhoA was tested with a GTPase assay as described below. To generate the RhoAR68E mutant, 4 mutations were introduced in the mRhoA cDNA. A HA-tagged RhoB was generated by amplifying the whole human RhoB cDNA with a forward primer including the sequence coding for the HA-FLAG plus a start codon (MYPYDVPDYA) upstream of the first 22 nucleotides of the human RhoB coding sequence and a classical reverse primer hybridizing at the end of the coding sequence. To generate the HA-tagged RhoBR68E, 3 mutations were introduced in the HA-RhoB sequence by a PCR-based approach. These three cDNAs were cloned into pcDNA4/TO. The entire coding sequence of human RhoGDI␣ was amplified by RT-PCR and cloned into pcDNA3 (pcDNA3_RhoGDI␣). The integrity of the four cDNAs was confirmed by sequencing. For plasmid transfection, HS578T cells were seeded in 6-well plates. 3 h after seeding, a total amount of 1 g of plasmid was transfected into cells for 20 -24 h with 3 l of Genejuice TM (Novagen).
Rho Translocation Assay-A Rho translocation assay was performed as described previously (22,23). HS578T cells were incubated with lysis buffer containing 50 mM Hepes, pH 7.4, 50 mM NaCl, 1 mM MgCl 2 , 2 mM EDTA, 5 mM sodium fluoride, 0.1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride, 4 g/ml aprotinin, 1 mM dithiothreitol, and 0.1% Triton X-100 for 5 min on ice. The cell lysates were centrifuged at 24,000 ϫ g for 15 min at 4°C. The supernatant corresponding to the cytosol fraction was collected and the pellet resuspended in ice-cold lysis buffer containing 1% Triton X-100 and centrifuged at 24,000 ϫ g for 15 min at 4°C. The supernatant corresponding to the membrane fraction was collected.
GTPase Assays-The assay was carried out as previously described (24,25). Briefly, cells were chilled on ice and lysed in ice-cold buffer containing 0.5% Triton X-100, 25 mM Hepes, pH 7.3, 150 mM NaCl, 4% glycerol, 10 mM NaF, 20 mM ␤-glycerophosphate, 0.1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride, 4 g/ml aprotinin. Lysates were centrifuged for 8 min at 13,000 ϫ g. Supernatants were immediately frozen in liquid nitrogen and stored at Ϫ80°C until use. An aliquot of each supernatant collected before freezing was denatured in SDS-PAGE lysis buffer to measure the total RhoGTPase content by Western blotting. For pull-down assays, supernatants were incubated for 30 min with 30 g of GST-PBD protein containing the Cdc42 and Rac binding region of PAK-1B, or GST-RBD protein containing the Rho binding region of rhotekin affinity linked to glutathione-Sepharose beads. The beads were washed 4 times in lysis buffer and boiled in 60 l of SDS-PAGE lysis buffer.
Determination of RhoB Protein Stability-48 h after transfection with siRNA or 24 h after transfection with plasmids cells were incubated with 20 g/ml cycloheximide, at a concentration that effectively blocked synthesis of the RhoB protein (26), and the RhoB protein levels were analyzed by Western blot at various time points after cycloheximide addition.
Cytoskeleton Labeling-For fibrillar actin labeling, HS578T cells were fixed with 3% paraformaldehyde in PBS for 15 min and permeabilized with 0.2% Triton X-100 in PBS for 5 min. The samples were blocked with 0.2% bovine serum albumin in PBS for 30 min and incubated with 50 ng/ml of phalloidin-TRITC and with 50 ng/ml of 4Ј,6-diamidino-2-phenylindole for 30 min. Fluorescence was analyzed with a Leica DMRB microscope. A minimum of 100 cells in each condition tested were analyzed to determine the percentage of cells displaying stress fibers.

Silencing RhoA and RhoC Does Not Completely Suppress the Actin Stress
Fibers-RhoA is a key determinant in the organization of actin cytoskeleton and cell shape. However, its specific repression, up to 95%, in various cell types with two different sets of siRNA did not alter the cell morphology of either actin stress fibers formation (Refs. 18 and 21 and Fig. 1). According to the data published by Simpson and co-workers (27), this lack of morphological alteration could be due to a compensatory mechanism between RhoA and RhoC. We tested this hypothesis by a double silencing of RhoA and RhoC. Although the morphology of human breast adenocarcinoma cell HS578T was somewhat altered by this double knock-down, stress fibers were still visible, suggesting a residual "Rho-like" activity ( Fig. 1). Therefore, we hypothesized that it might be related to the activity of the third member of the Rho subclass, RhoB. This RhoGTPase is indeed able to activate effectors such as ROCK and to induce the formation of actin stress fibers.
RhoA-dependent Regulation of RhoB-RhoB is a short-lived and inducible protein that is barely detectable as observed by Western blot analysis of whole lysate of mock-transfected cells or of cells transfected with a control siRNA (Fig. 2). The knockdown of RhoA in HS578T cells induced a significant up-regulation of the RhoB protein level (Fig. 2). RhoB concentration was also slightly increased upon RhoC silencing, whereas the double silencing of RhoA and RhoC induced its dramatic increase as compared with the single silencing of RhoA (Fig. 2). The negative regulation of RhoB operated by RhoA and RhoC was also observed in human HT1080 fibrosarcoma cells, A2058 melanoma cells, in primary HSF (Fig. 2), and in human prostate PC-3 adenocarcinoma cells (Fig. 4B) suggesting that it represents a wide spread mechanism. Similar results were observed by using a second siRNA targeting another sequence of RhoA or RhoC mRNA (Fig. 3A). To further validate our observations, we designed control siRNA with two nucleotide changes from the target sequence (CtrRhoA and CtrRhoC) as recommended in Ref. 28. As illustrated in supplementary Fig.  S1, the use of these controls confirmed the negative regulations operated by RhoA and RhoC on RhoB but also the specificity of the cross-regulations between RhoA and RhoC. By contrast, the silencing of Rac1 or Cdc42 with our previously published siRNA sequences (18) did not affect RhoB protein level (Fig.  3A). The induction of RhoB following RhoA silencing was observed at a concentration of siRNA as low as 0.2 nM (Fig. 3B). To definitively ascertain the specificity of the negative regulation operated by RhoA on RhoB expression, the silencing of RhoA was rescued by re-expressing a RhoA mRNA resistant to  the first siRNA targeting RhoA by introducing five neutral mutations impairing the silencing by this siRNA (mRhoA) (Fig.  4). HS578T cells were first transfected with scrambled siRNA or the first siRNA targeting RhoA. Immediately after washing and trypsinization, each pool of transfected cells was separated in two halves and transfected either with empty pcDNA3 or pcDNA3/mRhoA as described under "Experimental Procedures." The transient re-expression of RhoA reversed at least partly the overexpression of RhoB (Fig. 4A). The residual RhoB (40%) could be due to the presence of some cells that were not transfected by pcDNA3/mRhoA. The strong (more than 10 times the physiological level) and transient (ϳ24 h) re-expression of RhoA is obviously not the most appropriate manner to rescue silencing. To reverse more adequately RhoA silencing, we transfected PC-3 cells stably expressing a repressor sensitive to tetracycline (PC-3/TR) with an empty inducible vector (pcDNA4/TO) or with the same vector encoding mRhoA (pcDNA4/TO/mRhoA) to generate, respectively, control clones (PC-3/TR/control) or clones expressing mRhoA in a tetracycline-dependent way (PC-3/TR/mRhoA). Three clones of PC-3/TR/control cells and 3 clones of PC-3/TR/mRhoA cells were isolated. These clones were transfected with 6 nM scrambled siRNA or 6 nM siRhoA. Immediately after washing and trypsinization, each pool of transfected cells was separated in two halves and cultured for 2 days either in the absence or presence of 1 g/ml tetracycline. Cells were then processed for Western blot or RT-PCR analysis (Fig. 4B). RT-PCR analysis revealed that endogenous RhoA was similarly repressed in the absence or presence of tetracycline, whereas the mutated RhoA was significantly induced by tetracycline. The re-expression of the RhoA protein in these conditions nearly completely reversed the overexpression of RhoB, whereas treatment of PC-3/TR/control cells with tetracycline did not affect the RhoB protein level. These results definitively confirm the negative regulation operated by RhoA on RhoB. It should be noted that the re-expression of RhoA also reversed the up-regulation of RhoC upon RhoA silencing (supplementary Fig. S2).
The Up-regulated RhoB Is Biologically Active-As measured by pull-down assay, the increased RhoB level observed in RhoA and/or RhoC silenced cells is paralleled by an enhanced level of the RhoB active form (Fig. 5A). RhoB-GTP was estimated to represent between 2 and 3% of the total up-regulated RhoB. However, an efficient activation of the downstream effectors also requires translocation of the RhoGTPase to the membrane. To evaluate the subcellular localization of RhoB, a differential extraction procedure was used. It showed that up-regulated RhoB is mainly associated with the membrane fraction (Fig.  5B). These observations suggested that the induced RhoB was functional and could be responsible for the residual Rho-like activity observed in Fig. 1. To address this issue, a simultaneous repression of RhoA, RhoC, and RhoB was performed in HS578T cells. Western blot analysis and pull-down assays collected 48 h after transfection supported the efficient repression of total and active RhoB (Fig. 5A)   the double knockdown RhoA ϩ RhoC (Fig. 5C). The percentage of HS578T cells displaying stress fiber was 89 Ϯ 3% in cells transfected with siScr, 69 Ϯ 2% in cells transfected with siRhoA ϩ siRhoC, and 25 Ϯ 2% in cells transfected with siRhoA ϩ siRhoC ϩ siRhoB. RhoB Up-regulation Is Related to a Prolonged Half-life of the Protein-As assessed by real time quantitative PCR measurements following RhoA or RhoC silencing, the induction of RhoB was not related to an increase of its mRNA level (Fig. 6A). It was, however, clearly up-regulated by the double silencing RhoA ϩ RhoC. The stability of the RhoB protein was determined in HS578T cells by blocking protein synthesis with cycloheximide. As compared with cells transfected with the scrambled siRNA, the half-life of RhoB was slightly increased in RhoC-silenced cells. By contrast, RhoA silencing as well as the double silencing RhoA ϩ RhoC induced a dramatic increase of RhoB stability from 3.0 to 16.5 h (Fig. 6B and Table 1). In the double knock-down RhoA ϩ RhoC, the RhoB protein level is therefore up-regulated both by transcriptional and post-translational mechanisms. Similar results were also obtained in HSF (not shown).
The Induction of RhoB Is Neither Dependent on Isoprenoid Biosynthesis, nor ERK1,2, p38MAPK, or PI3K Pathways, nor ROCK or RhoA Activity-Inhibitors of the mevalonate pathway such as simvastatin have been reported to increase the stability of RhoB by reducing the availability of isoprenoid intermediates required for translocation to the membrane (26). It was also observed in our models (Fig. 5B). Addition of geranylgeranylpyrophosphate and/or farnesyl-pyrophosphate rescued the simvastatin-induced up-regulation of RhoB (26). A similar procedure used in RhoA-and/or RhoC-silenced cells did not alter the up-regulation of RhoB (supplementary Fig. S3). Moreover, by using a differential extraction procedure, we observed that up-regulated RhoB by silencing RhoA was associated with the membrane fraction, whereas the RhoB induced upon simvastatin treatment was cytoplasmic (Fig. 5B). These results rule out the implication of isoprenoid biosynthesis in the up-regulation of RhoB reported here. In parallel, we tested the involvement of the ERK1,2, p38MAPK, and PI3K pathways by means of pharmacological inhibitors. Used at concentrations allowing an efficient inhibition of their specific target, none of these inhibitors affected the RhoA-dependent regulation of RhoB (supplemental Fig. S3) suggesting that neither ERK1,2, p38MAPK, nor PI3K pathways were involved in the up-regulation of RhoB reported here. Alternatively, the knockdown of RhoA and/or RhoC could affect the activation level of a common downstream effector such as ROCK. To test this hypothesis, ROCK was specifically inhibited in HS578T cells with daily renewed Y-27632 during 48 h of culture. This treatment barely modulated the RhoB protein level in HS578T cells (Fig. 7A), suggesting that the RhoA-mediated RhoB regulation did not depend on the classical activation of downstream signaling cascades upon GTP loading. A siRNA approach as we used depletes both GTPand GDP-bound RhoA. To discriminate between the effects of both forms, PC-3 cell lines expressing, in an inducible way, the dominant-negative form of RhoA were created by transfecting PC-3 cells stably expressing a repressor sensitive to tetracycline (PC-3/TR) with a vector encoding RhoAN19 (pcDNA4/TO/RhoAN19) that prevents GDP-GTP exchange. Three clones of PC-3/TR/RhoAN19 cells were isolated on the basis of their ability to express upon tetracycline addition a functional RhoAN19 by using a pull-down assay (Fig.  7). RhoAN19 is visible as a slower migrating band in Fig. 7. As observed in Fig. 7B, the repression of RhoA-GTP alone was not sufficient to increase RhoB expression. By contrast, the silencing of RhoA in the three clones of PC-3/TR/RhoAN19 induced RhoB expression as efficiently with (ϩtet) or without (Ϫtet) prior depletion of RhoA-GTP (Fig. 7C). These results suggest that the regulation of RhoB expression is mediated by RhoA-GDP and not by RhoA-GTP.
RhoGDI␣ Is Involved in the Induction of RhoB-Several studies suggested that in addition to regulating RhoGTPase activation, RhoGDIs protect some RhoGTPases from degradation  (7). To address the role of RhoGDI␣ in RhoB protein stability we transfected HS578T cells with an expression vector encoding the whole coding sequence of human RhoGDI␣. Its forced expression strongly enhanced the RhoB protein level as compared with cells transfected with an empty vector where RhoB was barely detectable (Fig. 8A). Upon RhoGDI␣ overexpression, RhoB was mainly membrane-bound (supplementary Fig.   S4B). The GTP-bound RhoB form was also increased and represented between 2 and 3% of total RhoB (supplementary Fig. S4A). This upregulation of RhoB was associated with a strong increase of its half-life (Table 1) suggesting that RhoGDI␣ may protect RhoB from degradation (Fig. 8B). This potential role was tested by performing simultaneous knockdown of RhoA and RhoGDI␣ in HS578T cells. The level of RhoB protein 48 h after transfection with the two different siRNA targeting RhoA alone was reduced by more than 70% if RhoGDI␣ was simultaneously knocked down (Fig. 8C). To investigate more directly the involvement of RhoGDI␣ in the regulation of RhoB, we generated by a PCR-based approach a cDNA encoding a RhoA mutant (mRho-AR68E) unable to bind RhoGDI␣ but still able to undergo GDP-GTP exchange and to bind downstream effectors as assessed by a pull-down assay (supplementary Fig. S5). The construct further contains five silent mutations making the encoded mRNA resistant to the first siRhoA as shown in rescue experiments in Fig. 8D. Even though this mRhoAR68E mutant was strongly expressed it was significantly less efficient as compared with wildtype RhoA in rescuing RhoB up-regulation upon RhoA silencing (Fig.  8D). To further investigate the molecular mechanism driving RhoB stabilization, we tested its interaction with RhoGDI␣. Coimmunoprecipitation experiments in siRhoA-transfected cells did not reveal any association between endogenous RhoB and RhoGDI␣. A direct interaction between RhoB and RhoGDI␣ could be observed by co-immunoprecipitation in lysates of HS578T cells co-transfected with expression vectors encoding a HAtagged RhoB and RhoGDI␣ (supplemental Fig. S6). The generation of the R68E mutation in RhoB completely abolished its interaction with RhoGDI␣ (supplemental Fig. S6). Co-transfection of RhoGDI␣ increased significantly the half-life of HA-tagged RhoB but not that of HAtagged RhoBR68E (Table 1)  can be observed in overexpression experiments, a direct role for RhoGDI␣ in the stabilization of endogenous RhoB remains elusive.

DISCUSSION
Deciphering the functions and regulations of closely related members of the RhoGTPase family like RhoA, RhoB, and RhoC requires a highly specific approach. In this study, we took advantage of the high precision afforded by the siRNA technology to underscore the role played by mainly RhoA, and to a lesser extent by RhoC, in the regulation of RhoB. The control operated by RhoA on RhoB appears to be a widespread mechanism as it was observed in many cell types of various lineages. The induction of RhoB protein with two different siRNA targeting RhoA and at concentrations as low as 0.2 nM are significant arguments supporting the specificity of our observations. Rescue experiments with a mutated RhoA mRNA (mRhoA) resistant to the first siRNA targeting RhoA definitely validate our results. To appropriately rescue the effect of RhoA silencing, we generated clones of PC-3 cells expressing the mRhoA in a tetracycline dependent way. The concentration of tetracycline used here (1 g/ml) did not affect the RhoA-dependent regulation of RhoB as observed in the control clones. The inducible re-expression of mRhoA was close to the physiological range in the 3 PC-3/TR/ mRhoA clones tested and was maintained as long as tetracycline was present. By using this procedure, the rescue nearly completely suppressed the up-regulation of RhoB. This technology also opens the way to in vivo rescue experiments by feeding animals, in which these cells could be injected to  No modulation of the RhoB mRNA level in RhoA-silenced cells was observed by real time quantitative PCR analysis. Moreover, two independent comparisons of the gene expression profiles by microarrays of PC-3 cells transfected either with an irrelevant siRNA or the first siRNA targeting RhoA did not reveal any modulation of RhoB mRNA expression. 5 We demonstrated that the increase of RhoB occurs at a post-transcriptional level through stabilization of the protein as assessed by half-life measurements. Fritz and Kaina (29) reported an up-regulation of the RhoB gene by overexpressing the dominant-negative mutant RhoAN19. The results reported here demonstrate that silencing of a RhoGTPase does not simply mimic the overexpression of a dominant-negative mutant that targets guanine-nucleotide exchange factors. Whereas some Rho guanine-nucleotide exchange factors are highly specific, others activate multiple Rho GTPases (30). Our data suggest that the double silencing of RhoA ϩ RhoC is an experimental condition closer to the overexpression of RhoAN19 than the single silencing of RhoA. This is likely because the RhoA dominant-negative mutant depletes the pool of guanine-nucleotide exchange factor(s) that activate both RhoA and RhoC. Our results clearly demonstrate that, under physiological conditions, RhoA is the main determinant to set the low steady-state level of RhoB through (a) post-transcriptional(s) mechanism(s) that strongly decrease(s) the half-life of the protein. The RhoAmediated regulation of RhoB has a physiological significance and can be operational in cells exposed to bacterial toxins such as Clostridium difficile toxin B (TcdB). Genth et al. (31) reported that, in cells treated with 1 ng/ml of TcdB, the RhoA level was decreased with a parallel increase in the RhoB level. More recently, Huelsenbeck et al. (32) demonstrated that the apoptotic effect of TcdB was mediated by RhoB.
Our results strongly suggest that the RhoA-dependent regulation of RhoB does not depend on the activity of RhoA but is mediated by its GDP-bound form explaining why such a regulation is only visible upon depletion of the inactive pool of RhoA. Using a siRNA approach, Simpson et al. (27) recently reported a cross-regulation between RhoA and RhoC that was also observed in our models. Thus, our results suggest that such regulations, uncovered by a siRNA-based approach and undetectable by conventional tools, are common in the RhoGTPase family.
By contrast to RhoA and RhoC reported to be up-regulated in various cancers, RhoB displays properties that might participate in tumor suppression (33). As assessed by the pull-down assay and indicated by the presence of actin stress fibers the up-regulated RhoB is indeed actually biologically active. It could contribute to the anti-tumoral effect of siRNA targeting RhoA or RhoC recently reported in vitro and in vivo (34). Furthermore, we observed that the induction of the cell cycle inhibitor p21 CIP1 following RhoC silencing in PC-3 cells is RhoB-dependent. 5 It is well documented that alterations of post-translational prenylation of RhoB by inhibitors of the mevalonate pathways or farnesyl transferase can increase the expression of RhoB by acting at a transcriptional level (35) but also through modulation of the protein stability (26). However, the RhoB protein induced by treatment with one of these inhibitors, simvastatin as we showed here, is cytosolic, whereas the RhoB induced by repressing RhoA is associated with the membrane fraction similarly to the RhoB expressed in basal conditions. This suggests that silencing of RhoA did not affect post-translational lipid modifications required for RhoB subcellular localization. Moreover, the increased half-life of RhoB following RhoA  transfected with an empty vector (pcDNA3), or transfected with an expression vector encoding the whole coding cDNA sequence of RhoGDI␣ (pcDNA3_RhoGDI␣). 24 h after transfection, cells were processed for Western blot analysis with specific antibodies to RhoB, RhoGDI␣, and ERK1,2. B, HS578T cells were transfected with an empty vector (pcDNA3) or with an expression vector encoding the whole coding DNA sequence of RhoGDI␣ (pcDNA3_RhoGDI␣). 24 h after transfection, cells were cultured with 20 g/ml cycloheximide for the indicated time before being processed for Western blot analysis. Representative blots of three independent experiments are shown. The bottom panel illustrates the densitometric analysis of the illustrated blots. Blots were loaded with 30 (pcDNA3) or 5 g (pcDNA3_RhoGDI␣) of proteins to obtain similar initial signal intensities. C, HS578T cells were transfected with calcium phosphate alone (CaP), 20 nM of an irrelevant siRNA (siScr), 20 nM of the first or of the second siRNA targeting RhoA (siRhoA or siRhoA#2) alone, or in combination with 20 nM siRhoGDI␣ (ϩsiRhoGDI␣). 48 h after transfection, cells were processed for Western blot analysis with specific antibodies to RhoB, RhoA, RhoGDI␣, and ERK1,2. D, representative Western blot analysis with specific antibodies to RhoA, RhoB, or ERK1,2 of whole cell lysates of HS578T transfected with 10 nM of the first siRNA targeting RhoA (siRhoA) and 1 g of empty pcDNA4/TO (vector), pcDNA4/TO/mRhoA (mRhoA), or pcDNA4/TO/mRhoAR68E (mRhoAR68E). The lower panel illustrates the densitometric analysis of RhoB signal intensity. Results are the mean Ϯ S.D. of three independent experiments. silencing was not reversed by supplementation of prenylation precursors such as geranylgeranyl-pyrophosphate or farnesylpyrophosphate demonstrating that the mevalonate pathway is not involved in the regulation of RhoB in our experimental model. Nevertheless, isoprenylation might constitute a signal for RhoB degradation (26). In physiological conditions, the isoprenyl moiety is masked upon binding to RhoGDIs (36) thus explaining, at least in part, how RhoGDI can protect the RhoG-TPases from ubiquitin-mediated degradation as reported for RhoA (7).
RhoGDI␥ has been reported to be the preferred RhoGDI partner of RhoB (37,38). However, similarly to RhoGDI␤, its pattern of expression is tissue specific and is unlikely to account for the RhoA-dependent regulation of RhoB, a process observed in many cell types derived from various tissues (37). We considered that the ubiquitously expressed RhoGDI␣ was a better candidate. Moreover, it was described as interacting with RhoB in some studies (39). Our data strongly suggest that RhoGDI␣ can stabilize RhoB and is operational in the up-regulation of RhoB following RhoA silencing. Moreover, we observed that the ability of RhoA to bind RhoGDI␣ is involved, at least partly, in the regulation of RhoB reported here. The affinity of RhoGDI␣ is likely higher for RhoA and RhoC than for RhoB because the latter lacks Ser-188, a residue affecting positively the binding to RhoGDI␣ upon phosphorylation by PKA (7,40). In overexpression experiments, an interaction between RhoB and RhoGDI␣ was actually observed. Moreover, transfected RhoGDI␣ can stabilize wild-type RhoB and not RhoBR68E that has lost the ability to bind RhoGDI␣. However, a direct stabilization of endogenous RhoB by RhoGDI␣ remains elusive. Alternatively, it could involve the interaction of RhoGDI␣ with some proteins other than RhoB. Quantitation of the RhoGDI␣ level in various cell types shows that its molar amount is roughly equal to the molar amount of the three GTPases: RhoA, Rac1, and Cdc42 (38). In physiological conditions RhoGDI␣ is thus likely rate-limiting and its overexpression or the silencing of RhoA are two means for making it available to stabilize RhoB. The freed RhoGDI␣ by RhoA silencing could also contribute to the up-regulation of the RhoC protein level that we and others observed upon RhoA depletion ( Fig. 2 and Ref. 27). In agreement with our hypothesis, it should be noticed that TcdB prevents RhoA from interacting with RhoGDI (32) and, as mentioned above, increases in parallel the RhoB protein level. Recently, Wong and co-workers (10) reported that the alteration of RhoGDI-dependent cross-talk between RhoA and Rac1 suppresses integrin-mediated bacterial uptake. Altogether these data suggest that such interplays should be a widespread mechanism to control the stability and the activity of the RhoGTPases.
Our study demonstrates that silencing a RhoGTPase does not simply recapitulate the effects of a dominant-negative mutant but reveal novel mechanisms of regulation. The identification of these mechanisms is related to the highest specificity of the siRNA approach but also to the strategy of depletion of both GTP-and GDP-bound forms thus unraveling an unexpected role for RhoA-GDP. Mechanisms implicating the GDPbound form of RhoGTPase are not restricted to RhoA. Arozarena and co-workers (2) previously reported the involve-ment of Cdc42-GDP in Ras signaling. Di-Poi and co-workers (9) observed that Rac1-GDP in complex with RhoGDI can efficiently activate the NADPH oxidase. Our results also suggest that RhoGDI␣ is a necessary component for stabilizing RhoB that could act directly, indirectly, or as part of a multiproteic complex. The potential interplays between members of the RhoGTPase family should be taken into account for analysis of loss-of-function experiments as well as for the efficient design of therapeutic strategies based on a siRNA approach.