A Built-in Arginine Finger Triggers the Self-stimulatory GTPase-activating Activity of Rho Family GTPases*

Signal transduction through the Rho family GTPases requires regulated cycling of the GTPases between the active GTP-bound state and the inactive GDP-bound state. Rho family members containing an arginine residue at position 186 in the C-terminal polybasic region were found to possess a self-stimulatory GTPase-activating protein (GAP) activity through homophilic interaction, resulting in significantly enhanced intrinsic GTPase activities. This arginine residue functions effectively as an “arginine finger” in the GTPase activating reaction to confer the catalytic GAP activity but is not essential for the homophilic binding interactions of Rho family proteins. The arginine 186-mediated negative regulation seems to be absent from Cdc42, a Rho family member important for cell-division cycle regulation, of lower eukaryotes, yet appears to be a part of the turn-off machinery of Cdc42 from higher eukaryotes. Introduction of the arginine 186 mutation into S. cerevisiae CDC42 led to phenotypes consistent with down-regulatedCDC42 function. Thus, specific Rho family GTPases may utilize a built-in arginine finger, in addition to RhoGAPs, for negative regulation.

The Rho family small GTPases of the Ras superfamily are molecular switches controlling a variety of intracellular signaling events (1,2).
To understand how these molecular switches are turned off, much effort has been focused on the elucidation of the mechanism of RhoGAP 1 actions (3). Recent x-ray crystallography and mutagenesis studies have helped to establish that, like the case of Ras-RasGAP interaction (4), and to certain extent the case of heterotrimeric G-protein-RGS interaction (5,6), Rho GTPases undergo a transition state mimicked by a combination of AlF 4 Ϫ and RhoGAP in the GTPase-activating reaction catalyzed by RhoGAP (7)(8)(9). Furthermore, Rho GTPases appear to utilize a conserved arginine residue, termed "arginine finger", of RhoGAP to stabilize the transition state and to catalyze the cleavage of ␥P i from the bound GTP (3,8). This arginine finger mechanism of GTPase activation is apparently shared by both Ras and G␣i, since critical arginine residues in RasGAP and G␣i itself (Arg 789 in RasGAP and Arg 178 in a built-in domain of G␣i) have been demonstrated to play essential roles in the GTPase-activating process (10,11). Here we report an unexpected finding that Arg 186 in Rho GTPases functions effectively as a built-in "arginine finger" to confer a self-stimulatory GAP activity through homophilic interaction.
GTPase Activity Assay-The GTPase activities of the small GTPases were measured by the MESG/phosphorylase system monitoring the free ␥P i release from the GTP-bound G-proteins as described for the cases of Cdc42, Rac1, and RhoA (12)(13)(14)(15) and by the nitrocellulose filter-binding method (9).
Gel Filtration Chromatography-A Superdex 200HR 10/30 gel filtration column (Amersham Pharmacia Biotech) coupled to a Bio-Rad Biologic liquid chromatography system was used to analyze the homophilic interactions of the small GTPases as described (13).
Transition-state Complex Formation Assay-The fluorescence change of small G-protein bound mantGDP in the presence of AlF 4 Ϫ and the activated form of the G-protein was used to detect a GAP-reaction transition-state complex formation (13).

RESULTS AND DISCUSSION
The Ras superfamily small GTP-binding proteins are monomeric GTP-hydrolyzing enzymes that contain slow intrinsic GTPase activities (20). Certain members of the Rho family GTPases of the Ras superfamily, e.g. RhoA, Rac2, and Cdc42Hs, form reversible homodimers under physiological buffer conditions in vitro (13). Others such as RhoB and RhoC were found in either monomeric or oligomeric form (data not shown). Although sharing a very high degree of sequence homology, RhoA and RhoB behaved like Ras proteins showing a slow intrinsic GTPase reaction rate at ϳ0.02 min Ϫ1 that was GTPase dose-independent, whereas the rate of GTP hydrolysis by RhoC was found to increase significantly with increasing concentrations of the G-protein bound to GTP (Fig. 1A). Addition of GTP␥S-bound RhoC to RhoC-GTP resulted in a further enhanced rate of ␥P i release similar to that seen with the addition of RhoGAP, whereas GTP␥S-bound RhoA or RhoB had no detectable effect on the respective G-proteins bound to GTP (Fig. 1B), indicating that the activated form of RhoC, but not RhoA or RhoB, can provide a specific GAP activity toward itself. This self-stimulatory GAP activity may account for the faster intrinsic GTP-hydrolysis rate of RhoC compared with RhoA and RhoB (Fig. 1, A and B). Previous studies of the Rho family members Cdc42Hs and Rac2 also showed that these two GTPases contain a self-stimulatory GAP activity when in the activated GTP-bound form (13). Sequence alignment analysis revealed that the presence of a C-terminal polybasic motif, located immediately N-terminal to the CAAX isoprenylation sequences (Fig. 1C), correlated with the homophilic interaction of Rho family members. All Rho family proteins containing the polybasic sequences, including the additional mammalian members Rac1, Rac2, Cdc42Hs, and RhoG, have been found in reversible homodimer or higher oligomer and monomer states, whereas others lacking the polybasic residues, e.g. TC10 and RhoB, are exclusively monomeric like Ras 2 (13). Together with the finding that removal or substitution of the C-terminal polybasic residues led to the monomeric form only (see below) and a loss of self-stimulatory GAP activity 2 (13), these results suggest that the polybasic nature of the C-terminal domain is critical in mediating the homophilic interactions of specific Rho family GTPases. However, the homophilic interaction itself apparently is not sufficient for the enhanced GTP hydrolysis rate because RhoA does not display detectable GAP activity toward itself albeit containing C-terminal polybasic residues (Figs. 1B and 1C); rather, additional unique structural determinant(s) of the Rho GTPases are required for the observed self-stimulatory GAP activity.
Specific arginine residues of RasGAP, RhoGAP, and a built-in GTPase-activating domain of G␣i1, known as "arginine fingers," have been shown to facilitate the effective cleavage of phosphodiester bond linking ␥P i of the G-protein-bound GTP in their respective GTPase-activating reactions (3). In addition, sequence motifs that contain invariant arginine residues suspected to play a role in catalyzing GTP-hydrolysis have also been identified in the GAPs for the Rap, Ypt, Ran, and Arf proteins (4), and an invariant arginine residue in the built-in ArfGAP domain of ARD1, an Arf family GTPase, has been shown to be critical for the ARD1 GAP activity (28). An examination of the polybasic region of Cdc42Hs, RhoC, RhoG, Rac1, and Rac2 (Fig. 1C), which all displayed a faster intrinsic GTPase activity, suggested that a highly conserved arginine residue, arginine 186 (numbered by the sequences of Cdc42Hs), might be involved in the observed self-stimulatory GAP activity and the enhanced intrinsic GTPase activity of the Rho proteins. Mutation of the corresponding residue in RhoA, serine 188, to arginine (RhoAS(R)), resulted in a significant increase of the intrinsic rate of GTP hydrolysis ( Fig.  2A) and a gain of self-stimulatory GAP activity (Fig. 2B). Substitution of the arginine 186 residue of Cdc42Hs by lysine (Cdc42HsR(K)) effectively decreased the intrinsic rate of GTPase activity of Cdc42Hs to that of the wild-type RhoA ( Fig. 2A) and led to a loss of the self-stimulatory GAP activity (Fig. 2B). Both RhoAS(R) and Cdc42HsR(K) behaved indistinguishably in their GTP-binding and GDP/GTP-exchange properties from the wildtype proteins (data not shown). Moreover, posttranslational modification of the Rho proteins by C-terminal geranylgeranylation did not change the GTPase-activating effect of the arginine residue (data not shown). These results indicate that arginine 186 of the Rho family GTPases constitutes a critical determinant for the self-stimulatory GAP activity that controls their intrinsic 2 B. Zhang and Y. Zheng, unpublished observations. GTPase activities.
To determine whether arginine 186 plays a role primarily in the catalytic or homophilic binding interaction of the G-proteins, the RhoAS(R) and Cdc42HsR(K) mutants were further examined by a fluorescence assay originally designed to detect the formation of a transition-state complex for the Ras-RasGAP and Rho-RhoGAP interactions (19,21) and by gel-filtration analysis. In the case of the small G-protein-GAP interaction, addition of GAP together with AlF 4 Ϫ causes a change of the emission spectrum of the G-protein bound to the fluorescent GDP-analog, mantGDP, both in maximum absorption wavelength and in intensity (19,21). Similarly, this was observed when GTP␥S-bound RhoAS(R) mutant (as GAP) together with AlF 4 Ϫ was added to RhoAS(R)-mantGDP (Fig. 3). This result is indicative of the formation of an analog of the GTP-bound transition state of the GTPase-activating reaction involving RhoAS(R)-mantGDP, AlF 4 Ϫ , and RhoAS(R)-GTP␥S. Wild-type RhoA, in contrast, failed to form a transition-state complex with AlF 4 Ϫ when RhoA-mantGDP and RhoA-GTP␥S were incubated together (Fig. 3, insert). This result is consistent with its inability to act as a self-stimulatory GAP. When Cdc42Hs and Cdc42HsR(K) were examined in similar experiments, it was found that the Cdc42Hs mutant had lost the ability to form the transition-state complex with Cdc42HsR(K)-mantGDP and AlF 4 Ϫ when bound to GTP␥S, whereas wild-type Cdc42Hs behaved like the RhoAS(R) mutant (Fig. 3, insert). However, no change in the gel-filtration profiles was detected for Cdc42HsR(K) and RhoAS(R), with each mutant protein showing a similar monomer/dimer distribution pattern as the respective wild-type proteins (data not shown). Truncation of the last seven and eight amino acids from Cdc42Hs (C-7) and RhoA (Rho-8), respectively, resulted in only the monomeric form of the proteins (13). These results provide evidence that the selfstimulatory GAP reaction of Rho family GTPases employs an arginine finger-like mechanism similar to the RhoGAP-catalyzed reaction, i.e. the arginine 186 residue is essential for the formation of a GAP-reaction intermediate, but is not required for homophilic binding interaction.
To determine the potential physiological relevance of the arginine finger-mediated self-regulation of Rho family GT-Pases, we examined the role of arginine 186 in the Cdc42 subfamily. The eleven members of the Cdc42 subfamily that have been identified from S. cerevisiae to Homo sapiens share over 80% amino acid sequence identify, and they can functionally complement each other under defined conditions (2,22,23). Interestingly, arginine 186 is found in all the Cdc42 sequences from D. melanogaster and higher organisms, whereas a lysine or serine residue is present at the 186 position of Cdc42 from C. elegans and lower eukaryotes (Fig. 4A). The arginine finger theory would predict that Cdc42 from higher eukaryotes than D. melanogaster should contain a faster intrinsic GTP hydrolysis rate than that of lower eukaryotes because of the catalytic effect of arginine 186. Indeed, as shown in Fig. 4B, the Cdc42Sc and Cdc42Ce proteins displayed a slow intrinsic GTPase activity similar to that of RhoA and Ras with a GTP hydrolysis rate of 0.02 min Ϫ1 , whereas Cdc42Dm and Cdc42Hs demonstrated a marked higher intrinsic GTPase activity that can be further stimulated by the respective Cdc42 bound to GTP␥S. These results raised the possibility that possession of Arg 186 contributes to the mechanism of regulation of the Cdc42 GTPases of higher eukaryotes. To determine the in vivo effect of the built-in arginine finger, a lysine to arginine mutation was introduced into Cdc42Sc at position 186. Cdc42Sc is an essential gene product in S. cerevisiae and plays a critical role in polarized cell growth and cell division cycle regulation (24,25). Purified wild-type Cdc42Sc showed an intrinsic GTP hydrolysis ability similar to that of RhoA, whereas the Cdc42Sc K186R mutant protein displayed a significantly enhanced GTPase activity similar to that of Cdc42Hs (Figs. 5A and 2B). Genomic integration of cdc42 K186R mutation into a S. cerevisiae ⌬cdc42 strain resulted in a temperature-sensitive lethal phenotype at 37°C (Fig. 5B). At the permissive temperature of 23°C, the cdc42 K186R mutant led to a pleiotropic phenotype of elongated, multibudded, multinucleated cells (68%) and large, round unbudded cells (13%) (Fig. 5C, and data not shown), which is similar to phenotypes observed with S. cerevisiae CDC42 effector cla4 and ste20 mutants (26,27), and indicative of a G 2 cell cycle delay and/or a cytokinesis defect. Given that the cdc42 K186R mutant seems to be able to interact with regulators of Cdc42 such as Cdc24 and Bem3 like the wild-type protein, 2 it is likely that the introduced arginine finger is involved in an abnormal negative regulation of Cdc42Sc function, contributing to the temperature-sensitive phenotype and morphological abnormalities.
Rho family GTPases utilize a conserved GAP-stimulated GTP-hydrolyzing machinery similar to that of Ras for negative regulation (2)(3)(4). We demonstrate here that specific Rho family proteins may employ an additional built-in arginine finger-like mechanism through homophilic interaction to effectively accelerate the basal intrinsic GTPase activity. This additional negative regulation mechanism seems to be absent from many Rho GTPases from lower eukaryotes, and introduction of this arginine finger into such GTPases results in abnormal negative regulatory effects leading to morphological defects and lethality. Because the currently available three-dimensional structures of Rho family GTPases were all derived from various C-terminal truncated forms of the proteins, it remains to be seen how the homodimers are configured so that the C-terminal polybasic domain containing the critical arginine residue of one molecule may interact with the GTP binding core of another molecule. To this end, we have started to map the residues on Rho GTPases that may serve as sites for Arg 186 action and have identified Tyr 32 of Cdc42Hs at the GTP hydrolytic center as one of such sites. 2 It will be important in the near future to obtain a complete picture of the structural configurations of the Rho family homodimers and to compare it with that of the Rho GTPase-GAP complexes. We propose that the built-in arginine finger mechanism may provide an alternative to RhoGAP-mediated negative regulation for Rho GTPases from higher eukaryotes, thereby increasing the flexibility for the regulatory circuit of these small GTPases. The discovery of this novel type of negative regulation may provide valuable information concerning the many diverse roles of Rho GTPases and may generate an interesting paradigm for possible negative therapeutics involving these GTPases. P]GTP hydrolysis were compared between Cdc42Sc and Cdc42Sc K186R mutant protein at ϳ2 M concentration at 23°C. B, cdc42 K186R mutation in S. cerevisiae causes temperature-sensitive lethality. The cdc42 K186R mutant strain CCY3-3B and wild-type strain C276 -4A were streaked onto yeast extract-peptone-dextrose plates and incubated at 23 or 37°C for 3 days. C, the phenotype of cdc42 K186R mutant at the permissive temperature of 23°C.