Transforming Activity of the Rho Family GTPase, Wrch-1, a Wnt-regulated Cdc42 Homolog, Is Dependent on a Novel Carboxyl-terminal Palmitoylation Motif*

Wrch-1 is a Rho family GTPase that shares strong sequence and functional similarity with Cdc42. Like Cdc42, Wrch-1 can promote anchorage-independent growth transformation. We determined that activated Wrch-1 also promoted anchorage-dependent growth transformation of NIH 3T3 fibroblasts. Wrch-1 contains a distinct carboxyl-terminal extension not found in Cdc42, suggesting potential differences in subcellular location and function. Consistent with this, we found that Wrch-1 associated extensively with plasma membrane and endosomes, rather than with cytosol and perinuclear membranes like Cdc42. Like Cdc42, Wrch-1 terminates in a CAAX tetrapeptide (where C is cysteine, A is aliphatic amino acid, and X is any amino acid) motif (CCFV), suggesting that Wrch-1 may be prenylated similarly to Cdc42. Most surprisingly, unlike Cdc42, Wrch-1 did not incorporate isoprenoid moieties, and Wrch-1 membrane localization was not altered by inhibitors of protein prenylation. Instead, we showed that Wrch-1 is modified by the fatty acid palmitate, and pharmacologic inhibition of protein palmitoylation caused mislocalization of Wrch-1. Most interestingly, mutation of the second cysteine of the CCFV motif (CCFV > CSFV), but not the first, abrogated both Wrch-1 membrane localization and transformation. These results suggest that Wrch-1 membrane association, subcellular localization, and biological activity are mediated by a novel membrane-targeting mechanism distinct from that of Cdc42 and other isoprenylated Rho family GTPases.

factors activate Rho proteins by promoting GDP dissociation in exchange for GTP (6,7), whereas GTPase-activating proteins downregulate Rho protein function by stimulating their intrinsic GTPase activity to hydrolyze GTP to GDP (8). A third regulatory class of proteins includes the Rho guanine nucleotide dissociation inhibitors (RhoGDIs) 4 that bind the carboxyl terminus of Rho GTPases and sequester them in the cytosol (9,10). Missense mutations within the switch regions of Rho proteins lock them in a GTP-bound conformation and render these proteins GTPase-deficient and constitutively activated. Activated forms of some Rho family GTPases cause growth transformation of NIH 3T3 mouse fibroblasts, and aberrant activity of both regulatory proteins and effectors of the Rho signaling pathways have been linked to human cancers (11)(12)(13)(14).
Wrch-1 (Wnt-regulated Cdc42 homolog-1) is a novel member of the Rho subfamily, whose transcription is up-regulated in Wnt-1 transformation of mouse mammary epithelial cells (15). Like many other Rho family members, Wrch-1 activation is regulated by its nucleotide state, and a single missense mutation at residue 107 (analogous to Q61L activating mutation in Cdc42) rendered Wrch-1 more active in signaling (15). Ectopic expression of a constitutively active form of Wrch-1(107L) caused a Wnt1-like change in the cellular morphology of mammary epithelial cells, suggesting a contribution for Wrch-1 in Wnt transformation (15). Additionally, like other Rho family proteins, Wrch-1 activation can promote growth transformation (16).
The correct subcellular localization and function of Ras and Rho family members are dictated by post-translational modification of the carboxyl-terminal hypervariable domain, including the last four amino acids known as the CAAX motif (17,18). The canonical CAAX motif consists of a cysteine residue, two aliphatic residues (AA), and at the last position any amino acid (X). The conserved cysteine residue serves as the site for post-translational modification by either farnesyltransferase (FTase) or geranylgeranyltransferase I (GGTase I), which irreversibly attaches an isoprenoid moiety (19). Although CAAX-signaled prenyla-tion is necessary (20 -25), a second targeting signal immediately upstream of the CAAX motif is also required for proper membrane association and subcellular localization. Some small GTPases, such as H-Ras, N-Ras, and the Cdc42-related protein TC10, are covalently modified by addition of a palmitoyl fatty acid on cysteine residues, whereas others, such as K-Ras4B and Cdc42, contain several basic residues that serve as their second targeting signal (18,23,26). Finally, additional carboxyl-terminal sequences provide further specificity in targeting Ras and Rho GTPases to distinct plasma membrane microdomains or to endomembrane compartments. Thus, functionally highly related Rho GTPase isoforms (e.g. RhoA, RhoB, and RhoC), by virtue of divergent carboxyl-terminal sequences and modifications, can exhibit strikingly different biological functions dependent on distinct subcellular locations (22,27,28).
Although Wrch-1 has been shown to localize to plasma membrane and internal membranes (15,29), the role of the carboxyl terminus in mediating Wrch-1 biological function has not been determined. Like Cdc42, Wrch-1 induces actin reorganization and formation of filopodia and causes activation of the PAK and c-Jun NH 2 -terminal kinase serine/ threonine kinases (15). However, Wrch-1 contains amino-terminal sequences not found in Cdc42, and we and others recently showed that this amino-terminal extension serves as a negative modulator of Wrch-1 effector interaction (16,30). In addition, Wrch-1 and Cdc42 differ considerably at their carboxyl termini, sharing only 25% sequence identity. Furthermore, Wrch-1 has an extended carboxyl terminus ending in an unusual CAAX sequence (CCFV) not characteristic of known substrates for farnesyl-and geranylgeranyltransferases (31,32). Therefore, in the present study we assessed the mechanism and role of the unique carboxyl terminus in regulation of Wrch-1 function. As with Cdc42, we found that Wrch-1 activation caused anchorage-dependent growth transformation of NIH 3T3 cells. However, although the carboxyl terminus was critical for Wrch-1 subcellular localization and transforming activity, Wrch-1 function was dependent not on modification by prenylation but on modification by palmitoylation. Thus, although Wrch-1 and Cdc42 share significant functional properties, they exhibit considerable divergence in lipid modifications and subcellular distribution and, consequently, may have divergent roles in cell physiology.

EXPERIMENTAL PROCEDURES
Molecular Constructs-pcDNA3 expression constructs encoding wild type (WT) and GTPase-deficient (Q107L) human Wrch-1 was obtained from Dr. A. Levine (15). PCR-mediated DNA amplification was used to introduce 5Ј and 3Ј BamHI sites flanking Wrch-1(WT) for subcloning into various epitope-tagged expression vector constructs. To create other constitutively activated mutants of Wrch-1, a glutamate to leucine mutation was generated at residue 107 using the Quick-Change site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions. Site-directed mutagenesis was also used to create the carboxyl-terminal CCFV motif mutants by changing cysteine residues at positions 255 and 256 to generate CCFV Ͼ SCFV (C255S), CCFV Ͼ CSFV (C256S), and CCFV Ͼ SSFV (C255S and C256S) point mutants in the activated Wrch-1(Q107L) background. To generate glutathione S-transferase (GST), green fluorescence protein (GFP), and hemagglutinin (HA) epitope-tagged Wrch-1 proteins for prenylation, transformation, and localization assays, Wrch-1 coding regions were digested with BamHI and ligated into the BamHI site of pGEX-2T multiple cloning site, the 5Ј BglII and 3Ј BamHI sites of the pEGFP-C1 (multiple cloning site), and the BamHI site of pCGN-hygro, respectively (33). All sequences were verified by the Genome Analysis Facility at the University of North Carolina, Chapel Hill.
Cell Culture and Transfections-NIH 3T3 mouse fibroblasts were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% calf serum (Invitrogen) and 1% penicillin/streptomycin (Invitrogen). Stable NIH cell lines were created by transfection of pCGN-wrch-1 constructs with FuGENE 6 (Roche Applied Science) per the manufacturer's instructions. After 48 h, cells were split into DMEM containing 200 g/ml hygromycin B (Roche Applied Science) and maintained in antibiotic selection until colonies formed. Colonies (Ͼ50) were pooled for use in soft agar assays.
Immunofluorescence-NIH 3T3 cells were seeded onto glass coverslips in 35-mm dishes. After 24 h, cells were transfected with HA-tagged pCGN-wrch-1(107L) constructs using FuGENE 6. After 24 h, cells were fixed with 3.7% formaldehyde in PBS and permeabilized with 0.5% Triton X-100 in Tris-buffered saline (TBS). Cells were then incubated with anti-HA antibody (Covance) for 1 h at room temperature. After three washes in 0.1% Triton X-100 in TBS, cells were incubated with Alexa Fluor 488-conjugated secondary antibody for 30 min (Molecular Probes) and washed three times with 0.1% Triton X-100 in TBS. Coverslips were mounted onto glass microslides with Vectashield Hardset mounting medium (Vector Laboratories, Burlingame, CA) and analyzed on the fluorescent microscope as described above.
Purified Protein Preparation and in Vitro Prenylation Assay-Recombinant GST-tagged proteins were produced from Escherichia coli BL21 strain as described previously (16). Protein induction was confirmed by SDS-PAGE and Coomassie staining.
For isoprenylation of purified proteins, 5 g of each purified protein was added to the prenylation reaction mixture containing 30 g of bovine brain high speed supernatant and 1 M [ 3 H]farnesylpyrophosphate (FPP) or [ 3 H]geranylgeranylpyrophosphate (GGPP) (8 -10 Ci/mmol; American Radiochemical, Inc.). Reaction mixture was incubated for 30 min at 30°C. Reaction was stopped with SDS protein sample buffer. Samples were boiled briefly, run on 4 -20% SDS-PAGE, prepared for fluorography, and exposed to preflashed film for 4 days.
Western Blot Analysis-NIH 3T3 cells transiently expressing GFPtagged or HA-tagged Wrch-1 proteins were lysed in 1% Triton X-100 lysis buffer containing protease inhibitors (5 g/ml aprotinin, 10 M leupeptin, 20 nM ␤-glycerophosphate, 12 mM p-nitrophenyl phosphate, 0.5 mM Pefabloc, and 0.1 mM sodium vanadate) or magnesium lysis buffer (25 mM HEPES, 150 mM NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, 10% glycerol, 10 mM MgCl 2 , and 1 mM EDTA, pH 8.0, containing Complete protease inhibitor tablet). Whole cell lysates were cleared, and protein concentration was determined using DC Lowry protein assay. Fifteen g of protein lysates were prepared in sample buffer, resolved on SDS-PAGE, and transferred to Immobilon-P membrane. Membranes were then blocked in 5% nonfat dry milk and probed for HA-tagged Wrch-1 proteins using mouse anti-HA antibody, for ␤-actin as a loading control using mouse anti-␤-actin (Sigma), for endogenous phosphorylated PAK using rabbit anti-phospho-PAK1 (Ser-144)/PAK2 (Ser-141), for total endogenous PAK using rabbit anti-PAK1/2/3, or for GFP-tagged proteins using mouse anti-GFP antibody (Clontech), followed by anti-mouse HRP-conjugated antibody or antirabbit HRP-conjugated antibody and SuperSignal West Dura extended duration substrate as above. Fig.  1A, the carboxyl termini of Wrch-1 and Cdc42 exhibit sequence differences that may result in different functional roles for Wrch-1. First, Wrch-1 terminates in an atypical CAAX motif. Second, Wrch-1 contains an additional 21 residues that have no counterpart in the Cdc42 carboxyl terminus. To determine the role of Wrch-1 carboxyl-terminal sequences in dictating Wrch-1 membrane association and subcellular location, and to compare these properties with those of Cdc42, we expressed GFP-tagged Wrch-1 and Cdc42 in NIH 3T3 mouse fibroblasts and performed live cell imaging analyses (Fig. 1B). Whereas GFP alone localized to the cytosol and prominently in the nucleus, both GFP-Cdc42 and GFP-Wrch-1 were excluded from the nucleus. Consistent with previous observations (18), we detected Cdc42 localization strongly in the cytosol, with its predominant membrane staining in the perinuclear region. Wrch-1, however, distributed mainly to the plasma membrane and to internal membranes reminiscent of endosomes and polarized, perinuclear Golgi, demonstrating an only partially overlapping localization pattern with that of Cdc42 ( Fig. 1B) (15,29).

The Carboxyl-terminal Nine Residues of Wrch-1 Promote a Subcellular Distribution Distinct from That Seen for Cdc42-As shown in
We next wanted to determine which of the unique carboxyl-terminal sequences of Wrch-1 were sufficient to dictate its unique localization. Philips and colleagues (18) demonstrated previously that the carboxylterminal CAAX-and hypervariable domain-containing 20 amino acids of Cdc42 and other Rho family GTPases were sufficient to determine their membrane targeting. As described previously, a GFP fusion pro-tein terminating in the 20 carboxyl-terminal residues of Cdc42 exhibited the same subcellular distribution as authentic Cdc42 (data not shown) (18,22). Despite the extended length of the Wrch-1 carboxyl terminus, we found that even fewer residues were sufficient for authentic Wrch-1 localization, as a GFP fusion protein terminating in the nine most carboxyl-terminal residues of Wrch-1 displayed the same plasma membrane and endosomal localization as authentic Wrch-1 (Fig. 1B). To confirm that the internal membrane structures were indeed endosomes, we compared directly the localization of GFP-Wrch-1 and Texas Red-transferrin, which marks early endosomes. As shown in Fig. 1C, merging these images demonstrated that Wrch-1 on internal membranes colocalizes with transferrin. Thus, although Wrch-1 and Cdc42 regulate similar downstream effector functions (15,29), they exhibit significant differences in their subcellular distribution.
Wrch-1 Membrane Localization Is Not Dependent Upon a CAAX Motif-By analogy to Cdc42 and other CAAX-terminating small GTPases, the existence of a putative CAAX motif (CCFV) in Wrch-1 suggested that the cysteine at residue 255 (CCFV) is likely to be a target for prenylation and critical for membrane localization. To evaluate this possibility, we generated HA-tagged Wrch-1 proteins with missense mutations in the CAAX motif. Cysteine to serine substitutions of the cysteine residue of the CAAX motif abolishes the prenylation of Ras and Rho GTPases, resulting in cytosolic and inactive proteins (20 -23, 25). Therefore, we generated the analogous mutant of Wrch-1 (C255S, SCFV). Most surprisingly, this mutant did not localize exclusively to the cytosol as seen in similarly mutated Ras and Rho GTPases (20 -24). Instead, it exhibited similar subcellular localization as wild type Wrch-1 (Fig. 2). However, it did show reduced asymmetrical "Golgi-like" localization near the nucleus, suggesting that the first cysteine may function as a Golgi targeting or retention signal (Fig. 2). Thus, in contrast to Cdc42 and other Rho GTPases, an intact CAAX motif is not required for membrane localization.
The Wrch-1 CAAX motif contains a second cysteine residue at position 255 and is similar to the carboxyl-terminal CCXX motif present on proteins modified by geranylgeranyltransferase II (GGTase II). Presently, all known substrates of GGTase II are members of the Rab small GTPase family. Thus, we speculated that Wrch-1 may possess an atypical prenylation signal sequence that is dependent instead on this second cysteine residue, alone or in combination with the first cysteine. Therefore, we generated a mutant of this cysteine residue, either alone or together with mutation of Cys-255. Most surprisingly, both Wrch-1 (C256S, CSFV) and Wrch-1 (C255S/C256S, SSFV) were mislocalized to the cytosol in a pattern similar to that of unprocessed small GTPases (Fig. 2). However, the CSFV mutant did retain minimal plasma membrane localization. Of particular note, mutation of the second or both cysteines resulted in nucleoplasm, but not nucleosome, accumulation of the HA-tagged Wrch-1 protein, implicating a potential nuclear localization signal within the carboxyl terminus of Wrch-1. These results suggested that Wrch-1 may be prenylated at Cys-256 rather than at Cys-255 and that, like other Ras and Rho family proteins, additional sequences within the carboxyl terminus may function as a secondary membrane targeting signal and/or nuclear localization signal in conjugation with modification of these cysteine residues.
Wrch-1 Localization Is Not Dependent on Isoprenoid Modification-The importance of Cys-256 suggested that perhaps Wrch-1 terminates in an atypical prenylation signal sequence. To address this possibility, we first determined whether Wrch-1 is a substrate for GGTase I or for FTase. We treated cells expressing GFP-tagged Wrch-1 with pharmacological inhibitors of GGTase I (GGTI) and of FTase (FTI). We have shown previously that inhibition of prenylation of GFP-tagged small GTPases results in nuclear accumulation because of the loss of the membrane targeting lipid and to the nuclear localization signal present within the GFP protein (20,25). Therefore, sensitivity of the GFPtagged GTPase to drug treatment was measured by redistribution of nuclear excluded, membrane-bound proteins to the cytosol and nucleus. Efficacy and specificity of drug treatment were confirmed by their ability to cause relocalization of GFP-tagged H-Ras (farnesylated), Cdc42 (geranylgeranylated), and K-Ras (alternatively prenylated) proteins to the nucleus and cytosol upon treatment with FTI, GGTI, and FTI plus GGTI, respectively (Fig. 3). As expected, FTI treatment caused mislocalization of H-Ras but not K-Ras4B or Cdc42, whereas GGTI treatment caused mislocalization of Cdc42 but not H-Ras or K-Ras4B. Most surprisingly, Wrch-1 localization was completely unaffected by treatment with either GGTI or FTI (Fig. 3). To eliminate the possibility that  Wrch-1, like K-Ras, is alternatively prenylated by geranylgeranylation when farnesylation is blocked, cells were also treated with a combination of both FTI and GGTI. Although H-Ras, Cdc42, and K-Ras were all sensitive to the combination treatment, Wrch-1 still did not mislocalize (Fig. 3). These results suggested that, unlike most Rho proteins, Wrch-1 is not post-translationally modified by either FTase or GGTase I and may, therefore, be a novel substrate for prenylation by the Rab GGTase, GGTase II.
We next evaluated whether Wrch-1 is a substrate for GGTase II or for an unknown prenyltransferase. Because there are currently no pharmacological inhibitors available that specifically target GGTase II, we used compactin to treat NIH 3T3 cells transiently transfected with pEGFP-Wrch-1, -Rab5, -Cdc42, or empty vector. Compactin, an inhibitor of hydroxymethylglutaryl-CoA reductase, prevents the formation of all isoprenoid precursors, thereby preventing the formation of the farnesyl and geranylgeranyl isoprenoid moieties used by all prenyltransferases, including GGTase II. Vehicle-treated cells showed nuclear exclusion and membrane and cytosolic localization of each GFP-tagged GTPase, as expected (Fig. 4). Compactin treatment caused both Cdc42 and the GGTase II substrate Rab5 (CAAX ϭ CCSN) to mislocalize to the nucleus and cytosol (Fig. 4). In contrast, Wrch-1 was entirely resistant to the treatment and remained excluded from the nucleus and targeted to the membranes and cytosol (Fig. 4). Given that both Rab5 and Cdc42 were susceptible to compactin treatment, these results clearly demonstrate that Wrch-1 membrane association is not dependent on modification with any isoprenoid moiety.
To confirm that Wrch-1 does not utilize isoprenoid lipid groups for its membrane targeting, we performed in vitro prenylation assays on purified, un-lipidated Wrch-1 protein to directly label Wrch-  (Fig. 5). To confirm further the absence of isoprenyl modification of purified Wrch-1 protein, radioactivity of each [ 3 H]FPP-or [ 3 H]GGPP-bound protein reaction was determined by counting in a scintillation counter. As expected, no radioactivity was detected with Wrch-1, whereas [ 3 H]FPP radioactivity counts were present for H-Ras WT and [ 3 H]GGPP counts for H-Ras CVLL and Rab5 WT (data not shown). Taken together with the above prenyltransferase inhibitor data, these data indicate that, unlike most Ras and Rho family proteins, Wrch-1 is not a substrate of FTase, GGTase I, or GGTase II, and its membrane localization is independent of isoprenylation.
Wrch-1 Subcellular Localization Is Dependent on Palmitoylation-The data above indicate that, unlike those of Cdc42, the last four amino acids of Wrch-1 do not function as a canonical "CAAX" motif to specify prenylation. However, because a carboxyl-terminal cysteine is clearly important for Wrch-1 localization, we then investigated other potential post-translational modifications that might occur at cysteine residues. Palmitoylation is the reversible attachment of a palmitoyl fatty acid to cysteines via a thioester bond (35,36). Although no consensus signal sequence exists to aid in the prediction of which cysteine residues are likely to be palmitoylated (35,36), palmitoylation of cysteines in the hypervariable domains of prenylated small GTPases is common, although CAAX-signaled prenylation is a critical prerequisite for this fatty acid modification (18,26,37,38). To determine whether the carboxyl-terminal cysteine residues of Wrch-1 are susceptible to thioester linkage to acyl groups like palmitates, we performed a recently described nonradioactive method for determining protein acylation (34). This method utilizes the ability of hydroxylamine to cleave thioester bonds resulting in free sulfhydryl groups that can then interact with biotin-conjugated 1-biotinamido-4-[4Ј-(maleimidomethyl)cyclohexanecarboxamido] butane (Btn-BMCC) sulfhydryl-specific reagent, effectively labeling acylated cysteine residues. Transiently expressed GFP-tagged Wrch-1 protein immunoprecipitated from human embryonic kidney 293 cells, was sus-  pended in N-ethylmaleimide to block any free, nonacylated cysteine residues, treated with hydroxylamine to cleave thioester bonds, and incubated with Btn-BMCC reagent to label newly exposed sulfhydryl groups, followed by Western blot analysis with streptavidin to detect any fatty acid modifications. Like H-Ras (palmitoylated on Cys-181 and Cys-184), Wrch-1 was labeled with the Btn-BMCC reagent, suggesting that Wrch-1 contains one or more cysteine residues that are acylated (Fig. 6A, Wrch-1 CCFV lane, top  band). The nonspecific lower band seen in Fig. 6A, Wrch-1 lanes, is also observed at the same molecular weight in the H-Ras lane. As expected, the Btn-BMCC signal for K-Ras4B (no palmitoylated cysteines) was notably absent (Fig. 6A).
Because Wrch-1 contains two cysteine residues at its carboxyl terminus that may be palmitoylated, we wanted to determine whether one or both cysteine residues were susceptible to acylation. We anticipated that the second cysteine, Cys-256, would be the preferred site for acylation because of the cytosolic and nuclear subcellular distribution when mutated to a serine residue (CSFV). However, the CSFV mutant did retain some membrane localization, and it is possible that the first cysteine (Cys-255) may also be lipid-modified. As seen with parental Wrch-1 (CCFV), Wrch-1 (C255S, SCFV) labeled with Btn-BMCC, albeit at a lower efficiency, suggesting that the first cysteine residue (Cys-255) may regulate acylation of the second cysteine (Cys-256) (Fig.  6A, Wrch-1 SCFV lane, top band). As expected, neither the Wrch-1 (C256S, CSFV) nor (C255S/C256S, SSFV) mutants were labeled with Btn-BMCC (Fig. 6A, Wrch-1 CSFV, -SSFV lanes, top band). Additionally, treatment of Wrch-1 with the palmitate analog 2-BP, a potent inhibitor of small GTPase palmitoylation (37), caused a dramatic decrease in Wrch-1 fatty acid modification, suggesting that palmitate lipids are the fatty acid moieties utilized by Wrch-1 (Fig. 6B, top panel). Expression of each protein was confirmed by Western blot analysis (Fig.  6A, bottom panel). These results indicate that the second cysteine residue of Wrch-1 (Cys-256) is the primary cysteine residue for palmitate fatty acid modification.
To investigate directly whether the fatty acid modification of Wrch-1 detected by the above Btn-BMCC method is due to incorporation of palmitoyl groups at its carboxyl-terminal cysteines, similar to H-Ras, we used [ 3 H]palmitate to metabolically label NIH 3T3 cells transiently expressing Wrch-1 (38). Correct incorporation of [ 3 H]palmitate was confirmed using cells expressing H-Ras (palmitoylated on Cys-181 and Cys-184) and K-Ras (no palmitoylatable cysteines) as positive and negative controls, respectively. Immunoprecipitation of the exogenous HA-tagged Wrch-1 protein from labeled cells, followed by autoradiography, showed that Wrch-1, like H-Ras, incorporated [ 3 H]palmitate (Fig. 6C, top panel). The reduced level of incorporation seen in Wrch-1 compared with H-Ras may be due in part to potential differences in palmitoyl turnover and, therefore, steady-state levels of the palmitoylated GTPases and in part to the single palmitoylated cysteine residue in Wrch-1 compared with the two palmitoylatable cysteines in H-Ras. Western blot analysis of the total lysate input and immunoprecipitated K-Ras4B protein confirm that the absence of a [ 3 H]palmitate signal in the K-Ras-expressing cells was not due simply to a lack of protein expression or immunoprecipitation (Fig. 6C, middle and bottom  panel). These results directly demonstrate that Wrch-1 is modified by palmitoylation.
To determine whether palmitoylation influences Wrch-1 localization, we treated NIH 3T3 cells transiently expressing GFP-tagged Wrch-1, H-Ras, K-Ras, or empty vector with 2-BP. 2-BP has been shown previously to cause the redistribution of palmitoylated GTPases, most recently the Cdc42 homologous protein Chp/Wrch-2, and other proteins within the cell (18,37,39). As expected, the lack of palmitoylatable sites within the K-Ras carboxyl terminus rendered its localization insensitive to the effects of 2-BP, whereas H-Ras was mislocalized from plasma membrane to endomembranes, thereby demonstrating substrate specificity of 2-BP for carboxyl-terminal cysteine-containing proteins (Fig. 7). Upon treatment with 2-BP, GFP-Wrch-1 relocalized dramatically to the cytosol and accumulated in the nucleus (Fig. 7). That Wrch-1 localization is sensitive to 2-BP treatment is consistent with our Btn-BMCC and metabolic labeling studies showing that Wrch-1 is palmitoylated. These results suggest that palmitate(s) could be the sole lipid modification normally causing retention of Wrch-1 in the cytosol and membranes.
The Carboxyl-terminal Cysteines Are Necessary for Wrch-1 Signaling to PAK-Because inhibition of palmitoylation and mutation of the carboxyl-terminal cysteines to serines rendered Wrch-1 improperly distributed, we predicted that loss of the palmitate modification would also affect Wrch-1 downstream signaling pathways. Previous studies have shown that activated, constitutively GTP-bound Wrch-1 results in increased auto-phosphorylation of the serine/threonine kinase PAK (15,16). PAK1 is a well known effector of several Rho family proteins, including Cdc42 and Rac (3). To determine whether mutation of the carboxyl-terminal cysteines of Wrch-1 would affect its ability to promote auto-phosphorylation of PAK in NIH 3T3 fibroblasts, we transiently expressed the cysteine to serine mutants in an activated Wrch-1(Q107L) background and looked for differences in the PAK phosphorylation status. As shown previously by us and others (15,16), the Wrch-1(Q107L) CCFV parent resulted in an increase in phosphorylated PAK levels when compared with vector-expressing cells (Fig. 8). Consistent with the loss of membrane association, the ability of the Wrch-1 SCFV, -CSFV, and SSFV mutants to induce phosphorylation of PAK was considerably reduced when compared with the CCFV parent. This suggests that correct localization of Wrch-1 via its palmitate modification is necessary for downstream signaling molecules.
The Carboxyl-terminal Cysteine 256 (CCFV), but Not 255 (CCFV), Is Required for Wrch-1 Transformation-Given that loss of palmitate modification affected Wrch-1 downstream signaling, we predicted that loss of the palmitate modification would also affect Wrch-1 biological activity. The transforming activity of prenylated Rho family proteins is impaired when palmitate modification is blocked (25). We sought to determine whether loss of palmitoylation would interfere with Wrch-1 transformation.
Previous studies showed that activated Raf can cooperate with constitutively activated Rho family members such as Cdc42 to cause synergistic transformation of NIH 3T3 cells (40 -43). To determine whether Wrch-1, like Cdc42, could also cooperate with Raf to cause focus for-mation, we transiently cotransfected into NIH 3T3 cells pCGN-hygro constructs expressing activated Wrch-1(Q107L), Cdc42(Q61L), or Rac1(Q61L), together with either pZIP-Raf22W or the empty pZIP-NeoSV(x)1 vector. We then evaluated the dishes for the appearance of foci of transformed cells. Raf-induced foci of transformed cells are indistinguishable in appearance from those caused by activated Ras and are characterized by large, spreading foci of highly refractile, spindleshaped, morphologically transformed cells. In contrast, activated Rho GTPases cause foci of transformed cells that are very distinct from those of Raf, and are characterized by tight clusters of rounded, refractile cells that are frequently multinucleated. We anticipated that the appearance of Wrch-1 foci might resemble those of Cdc42 more than those of Raf-1. As expected, activated Wrch-1 alone was unable to induce focus formation (Fig. 9A). In cooperation with Raf, however, activated Wrch-1 formed foci that were similar to those of Cdc42 and Rac1 (Fig. 9A).
Next, we evaluated whether loss of the carboxyl-terminal cysteines, individually or together, impaired Wrch-1 focus formation in cooperation with Raf. The ability of activated Wrch-1-CSFV and -SSFV to form foci of transformed cells was greatly reduced compared with that of the parental Wrch-1 (Fig. 9B). These results are consistent with a critical role for the second cysteine (CCFV), rather than the canonical CAAX cysteine (CCFV), in mediating Wrch-1 membrane association. Most unexpectedly, mutation of the first cysteine residue, SCFV, did not impair Wrch-1 focus forming ability but rather enhanced it. This result suggests that the roles of the two cysteines are distinct and that the upstream cysteine plays a negative regulatory role in Wrch-1 biological activity.
To determine whether these results also applied to other aspects of the transformed phenotype, we evaluated the ability of activated Wrch-1 carboxyl-terminal mutants to promote anchorage-independent growth in soft agar. In contrast to their activity in focus formation assays, Rho proteins alone, including Wrch-1 (16), are sufficient to confer anchorage-independent growth and do not require cooperation with Raf in soft agar assays. Therefore, NIH 3T3 cells stably expressing activated, HA-tagged Wrch-1(Q107L) were seeded into agar and analyzed for colony forming activity. Comparable expression of each Wrch-1 protein was seen (data not shown). Activated Wrch-1 potently induced colony formation in soft agar (Fig. 9, C and D). Consistent with the focus formation data, mutation of either the second cysteine (C256S, CSFV) or both cysteines (C255S/C256S, SSFV) resulted in a strong reduction of colony formation, whereas the cysteine to serine mutation at residue 255 (SCFV) led to a significant increase in Wrch-1-induced colony formation (Fig. 9, C and D). Taken together, our focus formation and soft agar analyses suggest that the second cysteine residue of the CCFV motif is required for Wrch-1 transforming activity, whereas the first cysteine may function instead as a negative regulator.

DISCUSSION
Wrch-1 gene expression is up-regulated in Wnt-1-transformed cells, and Wrch-1 activation can phenocopy the changes in cellular morphology caused by Wnt-1 (15). Wrch-1 shares significant amino acid sequence and functional identity with Cdc42 but exhibits significant divergence in carboxyl-terminal sequences. Wrch-1 terminates in an atypical CAAX tetrapeptide motif, and its hypervariable domain possesses an additional 21 amino acid residues not found in Cdc42. In the present study, we evaluated the role of these unique carboxyl-terminal features in Wrch-1 membrane association and biological activity. Although we found that the carboxyl-terminal nine residues of Wrch-1 alone were sufficient to promote Wrch-1 membrane association, Wrch-1 and Cdc42 exhibited very distinct patterns of subcellular localization, with significant amounts of Wrch-1 found at the plasma membrane and early endosomes. Most surprisingly, we found that an intact CAAX motif was not required for Wrch-1 membrane association, but instead, mutation of a second carboxyl-terminal cysteine significantly reduced Wrch-1 membrane association. Furthermore, Wrch-1 membrane association was not dependent on isoprenoid modification but was instead dependent on palmitoylation of the second cysteine residue. Finally, we found that Wrch-1, like Cdc42, can also promote growth transformation of NIH 3T3 cells and that the palmitoylated cysteine was critical for this activity.
Highly related Rho and Ras GTPases exhibit distinct cellular functions that can be attributed in part to subcellular localizations dictated by their distinct hypervariable domains (22,27,28). For example, RhoA shares 90% identity with RhoB and RhoC, and these three proteins share common regulators and effectors (3). However, despite these strong similarities, whereas RhoA can promote growth transformation, there is evidence that RhoB may function in an opposite fashion and exhibit tumor suppressor function (20,44). RhoC but not RhoA has been associated with tumor cell invasion (45,46). These three related Rho GTPases show the greatest sequence divergence in their carboxyl-terminal sequences, and this divergence results in differences in subcellular localization that in turn promote different cellular functions (22,27,28). Philips and co-workers (18) eloquently demonstrated that the last 20 amino acids of several Rho and Ras proteins, including Cdc42, mimic the subcellular localization of the full-length proteins. We have shown here that at least the last 9 amino acids of Wrch-1 are sufficient to confer proper subcellular distribution. This finding is consistent with other studies that illustrate that all of the membrane targeting information is located in the carboxyl terminus.
Thus, although Wrch-1 and Cdc42 share significant sequence identity and functional overlap, their divergent carboxyl-terminal sequences may also impart different biological roles to these biochemically related proteins. Mutation of the cysteine residue of the CAAX motifs of Cdc42 and other Rho GTPases to prevent prenylation results in loss of membrane association and biological activity (20 -24). Therefore, we were surprised that the analogous mutation of the Wrch-1 CAAX motif (C255S, SCFV) did not cause complete mislocalization or loss of transforming activity. Analogously to Ras family GTPases, the membrane localization of conventional Rho family proteins generally requires either a geranylgeranyl (e.g. Cdc42) or a farnesyl lipid group (e.g. Rnd3/ RhoE) attached to the cysteine residue of the CAAX motif and a "second  signal" consisting of either several basic residues or palmitoylated cysteine residues in the upstream hypervariable domain (3). In direct contrast, we show here that the Cdc42 homolog Wrch-1 does not utilize either a geranylgeranyl or farnesyl isoprenoid moiety for membrane targeting. Instead, its localization is regulated by a palmitoyl fatty acid, demonstrating that the Wrch-1 CAAX-like motif, CCFV, is not a canonical, prenylated CAAX.
It is unclear at this juncture whether Wrch-1 also requires a second signal for proper localization to plasma membranes. However, the carboxyl terminus of Wrch-1 contains several basic residues that could form a polybasic second signal to complement the palmitate modification (47,48), and are included in the short stretch of nine amino acids that constitutes a minimal targeting sequence. Alternatively, by analogy to CAAX-containing palmitoylatable small GTPases, Wrch-1 may require other poorly defined but essential motifs surrounding palmitoylatable cysteines (38,49). The Wrch-1 carboxyl terminus also contains uncommon but conserved residues such as tandem tryptophans and a tyrosine. The tandem tryptophan residues may represent a di-aromatic motif of the kind frequently associated with endosomal sorting (50) and may help to direct Wrch-1 to endosomes. The contribution of these residues to Wrch-1 membrane targeting and function is currently under investigation.
More distantly related Rho and Ras proteins also target to membranes but do not depend on carboxyl-terminal lipid modification. For example, the Rho-related proteins, RhoBTB-1/2 and Miro-1/2, as well as the Ras-related proteins, Rit, Rin, Gem, and Rem2, are not known to undergo lipid modification, yet display distinct membrane associations (29,(51)(52)(53). Conversely, although Rab proteins are prenylated, they lack a conventional CAAX motif and, instead, terminate in CCXX, CXC, and XXCC sequences that, in combination with upstream residues, serve as targeting motifs for GGTase II modification (54,55). We have demonstrated here that the CCXX motif of Wrch-1 is not a target for GGTase II but rather for palmitoylation. Most interestingly, Chp/Wrch-2, the closest relative of Wrch-1, lacks a CAAX motif and should, therefore, not be modified by prenyltransferases (56). However, it shares with Wrch-1 a CFV (CXX) motif, incorporates a fatty acid modification at its carboxyl-terminal cysteine residue, and Chp membrane association is also disrupted by 2-BP treatment (39). It is interesting to speculate that the CFV (CXX) motif may be a novel recognition site for post-translational modification by palmitoyltransferases. Although other mammalian Rho GTPases are also palmitoylated, their palmitate modification is dependent on prior modification by prenylation. Therefore, Wrch-1 and Chp undergo unique lipid modification-dependent membrane targeting not seen with other known mammalian Ras family GTPases. Most interestingly, atypical Rho-like proteins have been described in the plant Arabidopsis that also undergo a prenyl-independent, palmitoyl modification; however, these small GTPases terminate neither in conventional CAAX nor in CXX motifs (57,58). In addition, their carboxylterminal sequences lack the basic residues found in Wrch-1 and Chp, and multiple palmitoylated cysteines appear to be required for full membrane association.
Our finding that the Wrch-1 carboxyl-terminal Cys to Ser mutants have differential effects on Wrch-1 localization and function is unusual for small GTPases and suggests that each cysteine has a distinct contri- FIGURE 9. Carboxyl-terminal cysteines of the hypervariable domain differentially modulate Wrch-1 transforming activity. NIH 3T3 cells were transiently cotransfected with either empty vector pZIPneo or activated pZIP-Raf22W, along with pCGN constructs encoding activated Rac1(Q61L), Cdc42(Q61L), or Wrch-1(Q107L). After 14 days, dishes were fixed and stained with crystal violet, and foci of transformed cells were counted. Images are representative of at least three separate experiments carried out in duplicate. A, activated Wrch-1 cooperates with activated Raf to form Cdc42-like foci of transformed cells. B, cooperation with Raf by Wrch-1 containing carboxylterminal mutations at cysteine residues 255 (CCFV Ͼ SCFV), 256 (CCFV Ͼ CSFV), or both (CCFV Ͼ SSFV). C, NIH 3T3 cells stably expressing the above-mentioned pCGN Wrch-1 constructs were seeded into soft agar and analyzed for their ability to induce anchorage-independent growth. Colonies formed after 14 -21 days were stained and scanned (D) and quantified (C). Images and bar graphs are representative of at least two separate experiments carried out in triplicate.