Characterization of the Interactions between the Small GTPase Cdc42 and Its GTPase-activating Proteins and Putative Effectors

The small GTPase Cdc42 interacts with multiple factors to transduce diverse intracellular signals. The factors that preferentially recognize the GTP-bound, active state of Cdc42 include a panel of GTPase-activating proteins (GAPs), the Cdc42/Rac interactive binding (CRIB) motif-containing molecules, and the RasGAP domain containing IQGAP1 and IQGAP2. In the present study, we have determined the kinetic parameters underlying the functional interactions between the Cdc42-binding domains of some of these factors and Cdc42 by monitoring the continuous release of γPi and have compared the ability of the domains to bind to Cdc42. The catalytic efficiencies (K cat/K m ) of the GAP domains of Bcr, 3BP-1, and p190 on Cdc42 are found to be 60-, 160-, and over 500-fold less than that of Cdc42GAP, respectively, and the differences are due, to a large part, to differences inK m . The K m values of the GAP domains compare well to the binding affinity to the guanylyl imidodiphosphate-bound Cdc42, suggesting a rapid equilibrium reaction mechanism. The affinity of the Cdc42-binding domains of the CRIB motif of Wiskott-Aldrich Syndrome protein and p21 cdc42/rac -activated kinase 1, and the RasGAP-related domain of IQGAP1, which all inhibit the intrinsic rate of GTP hydrolysis of Cdc42, are found to be 4, 0.7, and 0.08 μm, respectively. These quantitative analysis provide insight that Cdc42GAP functions as an effective negative regulator of Cdc42 by fast, relatively tight binding to the GTP-bound Cdc42, whereas IQGAP1 interacts with Cdc42 as a putative effector with over 10-fold higher affinity than the CRIB domains and GAPs, and suggest that various GAPs and effectors employ distinct mechanism to play roles in Cdc42-mediated signaling pathways.

The Rho family small GTP-binding protein Cdc42 appears to be an essential component in the transduction of intracellular signals that induce actin-cytoskeleton reorganization and gene activation (1)(2)(3). Genetic analysis of bud-site mutants revealed that Cdc42 is a key component in an actin-dependent polarization process during budding (4), as well as in mating (5) in Saccharomyces cerevisiae. The mammalian homolog of yeast Cdc42 has been shown to regulate the formation of the actin structure filopodia in Swiss 3T3 cells (6,7) and cell polarity in T cells (8). Mutations of cdc42 in yeast (4) and introduction of Cdc42 mutants to mammalian cells (9) and to Drosophila (10) render large multinucleated cells, suggesting a possible role for Cdc42 in cytokinesis. Furthermore, constitutively activated form of Cdc42, like some of the other members of Rho family proteins (1)(2)(3), has been found to activate the Jun NH 2 -terminal kinase (11)(12)(13)(14), the p70 S6 kinase (15), serum response factors (16), and DNA synthesis pathway (17).
In cells receiving the appropriate stimuli, Cdc42 GTPase is converted to the active GTP-bound state by guanine nucleotide exchange factors (18), whereas the GTP-bound form is in turn rendered inactive due to its intrinsic GTPase activity that is further stimulated by the GTPase-activating proteins (GAPs) 1 (19). The RhoGAP family of proteins represents one of the major groups of regulators for the small GTPase, of which over 17 members have been identified to date (19). Primary sequence analysis revealed an ϳ170-amino acid homology region, designated as RhoGAP domain, in these proteins (20). Biochemical analysis of the RhoGAP activity indicated that this domain contains the minimum structural unit that is necessary and sufficient for the GAP activity (21). The RhoGAP domains share about 20 -40% amino acid identity, and the proteins in which they are contained are generally large and multifunctional. The RhoGAP family members include Cdc42GAP (also known as p50RhoGAP), which may form a complex with SH3 domain-containing signaling molecules such as PI-3 kinase and Src through an additional proline-rich motif (22,23); the breakpoint cluster region gene product (Bcr), which also possesses serine/threonine kinase (24) and guanine nucleotide exchange factor activities in its amino-terminal sequences (25); the Ras-GAP-binding phosphoprotein p190, which binds to GTP at its amino terminus (26,27); and the SH3 domain-binding protein 3BP-1, which was identified by a screen for molecules interacting with the Abl oncoprotein with high affinity (28,29). Recent x-ray crystallography studies of the RhoGAP domains of Cdc42GAP and p85, the regulatory subunit of PI-3 kinase, have provided further evidence that these critical domains involved in interaction with Rho GTPases adopt a highly conserved structural folding in three dimensions (30,31).
The incoming signals of Cdc42 may bifurcate at the activated small G-protein itself through multiple effector targets to lead to the diverse biological effects. A panel of candidate effectors, identified by an overlay assay using the [␥-32 P]GTP-bound Cdc42 (32), turned out to be a class of novel protein serine/ threonine kinases termed p21-activated kinases (PAKs) (33)(34)(35). Molecular cloning and data base searches led to the discovery of a family of proteins sharing a conserved Cdc42/Rac interactive binding (CRIB) motif of PAK (36), many of which have since been shown to be candidate effectors for Cdc42. The CRIB motif-containing putative effectors include PAK1, the kinase activity of which is activated upon binding to the active form of the GTPase (12,34), and the Wiskott-Aldrich Syndrome protein (WASP), which binds to Cdc42-GTP and has been implicated in Cdc42-mediated actin polymerization process (37)(38)(39). Additional putative Cdc42 effectors have recently been identified by Cdc42-GTP affinity binding approach, and these include the IQGAP1 and IQGAP2, two RasGAP domain-containing proteins with selective affinity to the GTP-bound Cdc42 (40,41). Moreover, in analogy to that suggested by the studies of p120RasGAP (42), the RhoGAPs may also serve as effectors of the small GTPases. Direct evidence of this comes from the findings that the RhoGAP domain of p85 can bind to the GTPbound states of Cdc42 and Rac1 resulting in PI-3 kinase activation (43), and that the brain specific RhoGAP member nchimaerin acts synergistically with Cdc42 and Rac1 to induce the formation of filopodia and lamellipodia (44).
The Cdc42-GAP and Cdc42-effector interactions have not been investigated in detail. In particular, little quantitative information of these seemingly important interactions has been made available and been compared. In the present work, we have developed a real-time spectroscopic method to study the kinetics of Cdc42-GTP hydrolysis in the context of its coupling to GAP domains of Cdc42GAP, Bcr, 3BP-1, and p190, to the CRIB motifs of PAK1 and WASP, and to the Ras-GAP related domain of IQGAP1. By an enzyme-coupled phosphorolysis reaction that allows direct measurement of ␥P i release from Cdc42, we show that the differences of catalytic efficiency of GAPs reside largely in their different K m values, which compare well with their dissociation constants (K d ) for Cdc42-GTP. In addition, we have also obtained the affinity of Cdc42-GTP for the Cdc42-interactive domains of WASP, PAK1, and IQGAP1 based upon the property of the effectors to inhibit the intrinsic rate of GTP hydrolysis of Cdc42. The results indicate that Cdc42GAP functions as an effective negative regulator of Cdc42 by fast, relatively high affinity binding to the GTPbound Cdc42, whereas IQGAP1 interacts with Cdc42 as a putative effector with over 10-fold higher affinity than the CRIB domains and GAPs, and suggest that various GAPs and effectors employ distinct mechanism to play roles in Cdc42-mediated signaling pathways.

EXPERIMENTAL PROCEDURES
Materials-GMP-PNP was from Boehringer Mannheim. GDP, GTP, bacterial purine nucleoside phosphorylase, and the ingredients to generate the phosphorylase substrate, 2-amino-6-mercapto-7-methylpurine ribonucleoside (MESG), were purchased from Sigma. MESG was synthesized following the published protocol (45) and was dried to a yellow solid in a lyophilizer for storage at Ϫ20°C. An estimated ϳ60% product of the final synthesis procedure was MESG, and the linearity of its absorbance response at 360 nm for measuring a range of P i concentrations from 1 M to 70 M was ensured by using the coupling reaction of purine nucleoside phosphorylase in a buffer containing standard phosphate solution (pH 7.6) (45).
Expression and Purification of Recombinant Proteins-Human Cdc42 protein was expressed in Escherichia coli as His 6 -tagged fusion by the pET expression system 28 (Novagen). Briefly, the cDNA of Cdc42 was cloned into pET28 vector at the in-frame BamHI-EcoRI sites, and the resulting construct was transformed into competent BL21 cells. The induction and purification of His 6 -tagged Cdc42 by isopropyl-1-thio-␤-D-galactopyranoside and Ni 2ϩ -charged agarose beads were carried out following the instructions provided by Novagen. The glutathione Stransferase (GST) fusion of GAP domains and the Cdc42-interacting effector domains of Cdc42GAP, Bcr, 3BP-1, p190, PAK1, WASP, and IQGAP1 were expressed in E. coli using the pGEX system (21). The GAP domains of Cdc42GAP, Bcr, 3BP-1, and p190 contain the amino acid residues 205-439, 1010 -1271, 185-410, and 1249 -1513 of the native proteins, respectively, and the Cdc42-interacting domains of human PAK1, WASP, and IQGAP1 contain amino acid residues 51-135, 215-295, and 864 -1657, respectively. The GST-Cdc42GAP and GST-3BP-1 were generated as described previously (22,29). The GST-Bcr, GST-p190, GST-PAK1, and GST-WASP constructs were made by polymerase chain reaction amplification of cDNAs corresponding to the indicated amino acid regions with primers that contain the BamHI and EcoRI sites, followed by digestion and ligation into the pGEX-KG vector. The GST-IQGAP1 was generated by cloning the BamHI-EcoRI fragment of cDNA (nucleotides 2703-5670) derived from pJ⍀-IQGAP1 (a kind gift of A. Bernards) into pGEX-KG. Production and purification of the GST-fusions from E. coli were carried out as described by glutathione affinity chromotography (21). If necessary, the GST moiety of the fusions was cleaved by thrombin digestion, followed by incubation with p-aminobenzamidine immobilized on agarose beads (Sigma) to remove the thrombin (46).
All proteins prepared for measurements were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Coomassie Blue staining analysis, and the contents of each were judged at least 90% pure. Concentrations of the recombinant GAP and effector Cdc42-binding domains were determined using the BCA protein assay reagents from Pierce with bovine serum albumin (BSA) as standard, and the effective concentration of Cdc42 was measured using the MESG system under single-turnover conditions as described below.
Spectroscopic Measurements of GTPase Activity-The hydrolysis of GTP by Cdc42 was measured by the MESG system monitoring the absorbance increase of the reaction mixture at 360 nm, based on the method described by Webb and Hunter (47). Briefly, a 0.8-ml solution containing 50 mM HEPES, pH 7.6, 0.1 mM EDTA, 0.2 mM MESG, 10 units of purine nucleoside phosphorylase, 200 M GTP, and the indicated amount of recombinant Cdc42 (Ͼ1 M) was mixed in a 4-mm width, 10-mm pathlength cuvette, and the time courses of absorbance change at 360 nm were recorded using a Pharmacia Ultraspec III spectrometer equipped with compatible computer software for data manipulations. Single-turnover GTPase reactions were initiated by addition of MgCl 2 to a final concentration of 5 mM. For measurements of GAP-catalyzed reactions, 5-50 l of stock solution containing indicated amounts of GAP were added together with MgCl 2 to the reaction mixtures. A control experiment in which Cdc42 was left out was carried out to provide a background of absorbance in each independent measurement to be subtracted from the sample signals. Moreover, experiments at a broad range of concentrations of GAPs were performed to minimize the possibility that the release of the reaction product, P i , from Cdc42-GDP, rather than the GTPase reaction, is the rate-limiting step.
The concentrations of P i in the reaction solution are proportional to the net absorbance change by a factor of extinction coefficient ⑀ 360 nm ϭ 11,000 M Ϫ1 cm Ϫ1 at pH 7.6 (47), and were used to determine the concentration of Cdc42 after one round of single-turnover reaction. Because the phosphorylase coupling reaction is extremely fast with a rate constant of 40 s Ϫ1 (45), the slope of the absorbance in the time course is treated as proportional to the rate of the GTPase activity of Cdc42: rate of Cdc42-GTP hydrolysis ϭ 1/⑀ ϫ ⌬A cm Ϫ1 min Ϫ1 .
Fluorescence Measurements of GTPase Activity-The fluorescence measurements were made using an SLM-Aminco Series 2 Luminescence Spectrometer. To monitor the tryptophan fluorescence emission of Cdc42, samples containing the indicated amount of Cdc42-GTP in the absence or presence (0.94 nM) of Cdc42GAP in a buffer with 50 mM HEPES, pH 7.6, and 10 mM MgCl 2 were mixed and thermostated at 20°C. The excitation was at 295 nm, and emission was at 330 nm. The increase of fluorescence intensity of a unique tryptophan residue (Trp-97) of Cdc42 due to the shift to a higher quantum yield of the GDPbound state (48) was followed during the time course of GTP hydrolysis. The difference in fluorescence intensities (in defined arbitrary units) between the GTP-and GDP-bound forms of Cdc42 was calibrated by a factor of 48,000 M Ϫ1 cm Ϫ1 . The background fluorescence as determined by omitting Cdc42 in the reaction mixture (which includes GAP and free GTP) was negligible comparing to the sample signal of Cdc42.
Filter-binding Assay of GTPase Activity-Recombinant Cdc42 was preloaded with [␥-32 P]GTP (10 Ci, 6000 Ci/mmol, NEN Life Science Products) in a 100-l buffer containing 50 mM HEPES, pH 7.6, 0.2 mg/ml BSA, and 0.5 mM EDTA for 10 min at ambient temperature before the addition of MgCl 2 to a final concentration of 5 mM. An aliquot of the [␥-32 P]GTP-loaded Cdc42 was mixed with a reaction buffer containing 50 mM HEPES, pH 7.6, 0.2 mg/ml BSA, and 5 mM MgCl 2 in the presence or absence of GAP. At different time points the reaction was terminated as described (21) by filtering the reaction mixture through nitrocellulose filters, followed by washing with 10 ml of ice-cold buffer with 50 mM HEPES, pH 7.6, and 10 mM MgCl 2 . The radioactivity retained on the filters were then subjected to quantitation by scintillation counting. In the competition assays to determine the binding affinity of GAP domains to Cdc42, GST-GAP domains at the indicated concentrations together with the indicated amount of Cdc42 preloaded with GMP-PNP or GDP were present in the reaction mixture in addition to the components described including 5 nM Cdc42-[␥-32 P]GTP, and the reactions were terminated after a 5-min incubation at 20°C.
Data Analysis-Kinetic data were analyzed by nonlinear regression using equations derived below with the program Enzfitter (Elsevier Biosoft). The rate constant (K c ) of intrinsic GTP hydrolysis by Cdc42 was determined by fitting data obtained by the MESG system, by tryptophan fluorescence, or by filter-binding assay to a single exponential function. Since the amount of Cdc42 present in the GAP reaction is in great excess of GAPs, a modified Michaelis-Menten equation was used to derive kinetic parameters assuming GAP acting as the enzyme catalyst, Cdc42-GTP as the substrate, and the Cdc42-GDP and P i as the products.
V 0 is the initial rate of hydrolysis, ⑀ is the extinction coefficient at 360 nm for the phosphorylase reaction product, [Cdc42] 0 is the total Cdc42 concentration, K c is the rate constant of intrinsic GTPase activity, and the term ⑀K c [Cdc42] 0 is a correction for the rate of intrinsic GTP hydrolysis by Cdc42. K cat is derived by V max /[GAP] 0 , with [GAP] 0 representing total GAP concentration.
To determine the affinity of GAPs to Cdc42, the competition assay involving Cdc42-[␥-32 P]GTP, GST-GAP, and competitor C (GMP-PNPor GDP-bound Cdc42) was described by two simultaneous reactions.
The dissociation constants (K d ) were derived by fitting data to the following derived equation under the conditions that the total concentration of competitor [C] 0 and reaction K m value are orders of magnitude larger than the concentration of Cdc42-[␥-32 P]GTP.
Inhibition of GAP-catalyzed GTP-hydrolysis % ϭ To determine the affinity of effector domains to Cdc42-GTP, the GTPase-inhibitory reaction involving Cdc42-GTP and an effector domain (E) was treated as two simultaneous steps.
The binding constants of the Cdc42-binding domain of effectors to Cdc42-GTP were extracted by fitting data of the initial rate of GTP hydrolysis by Cdc42 at various concentrations of the effector domain to the derived equation, assuming that the rate of hydrolysis of Cdc42-GTP in complex with E is much slower than Cdc42-GTP alone.
V 0 and [E] 0 are the initial rate of GTP hydrolysis and the total concentration of effector domain added to the assay, respectively, and K i represents the binding affinity (dissociation constant) of the effector domain to Cdc42-GTP.

RESULTS
A Spectroscopic Assay for ␥P i Release from Cdc42-GTP-The MESG/phosphorylase assay was developed to measure the ki-netics of P i release from GTPases and ATPases (45), and has been applied to the studies of Ras and RasGAP interaction (47).
To determine the suitability of this assay system for measuring the Cdc42-GTP hydrolysis, the time course of intrinsic Cdc42 GTPase reaction as reflected by the absorbance trace of ␥P i release was monitored in the presence of phosphorylase and MESG substrate (Fig. 1A). In the presence of 0.1 mM EDTA when nucleotide exchange occurs fast, Cdc42 undergoes multiple turnover resulting in a linear increase of P i release. Upon the addition of 5 mM final concentration of MgCl 2 , Cdc42 is stabilized at the GTP-bound state due to the presence of excess GTP (200 M) and starts single-turnover GTP hydrolysis. The response of the absorbance increase is linear up to 70 M Cdc42 (data not shown), and the conditions used here for the loading of GTP and the initiation of single-turnover GTPase reaction were adopted for all subsequent measurements.
To ensure that the observed rate of Cdc42-GTP hydrolysis measured by the MESG/phosphorylase method is valid, the time course for ␥P i -elicited absorbance change was compared with that obtained by two other established GTPase assays for Cdc42, the [␥-32 P]GTP filter binding and the Cdc42 tryptophan fluorescence methods (Fig. 1B). The time courses of Cdc42-GTP hydrolysis at 20°C measured by the three different assays show excellent agreement, and fittings of the data to a single exponential function yield intrinsic rate constants (K c ) of 0.073, 0.075, and 0.065 min Ϫ1 by the MESG/phosphorylase, the fluorescence, and the filter binding methods, respectively (Table I). These data validify the MESG/phosphorylase assay as a reliable means for measurement of the GTPase activity of Cdc42.
To see how Cdc42 would behave at more physiological temperatures, the rate of GTPase activity of Cdc42 was determined at 20, 30, and 37°C (Fig. 1C). A ϳ1.6-fold increase in was observed when temperature was raised from 20°C to 30°C, or from 30°to 37°C, and the GTP hydrolysis occurs ϳ2.6-fold faster at 37°C than at 20°C (Table I). To allow accurate assessment of the initial rates of GAP-stimulated GTP hydrolysis, all subsequent GAP-catalyzed GTPase assays were performed at 20°C, which provides an optimal window of absorbance increase.
Kinetics of Interaction of Cdc42-GTP with Cdc42GAP-To determine the kinetics of Cdc42GAP-stimulated hydrolysis by Cdc42, we first made sure that the amount of Cdc42GAP employed is within the linear range with regard of the GAPcatalyzed rate increase. Fig. 2A shows the absorbance traces for three different reaction with the same amount of Cdc42 but by addition of different amounts of Cdc42GAP to initiate the reactions. The initial rates of the reactions were linear with Cdc42GAP concentrations up to 4 nM ( Fig. 2A, insert). In the following experiments to investigate the interaction of Cdc42GAP with Cdc42-GTP, the initial rate of P i release was measured as a function of Cdc42 concentration at a fixed concentration of 3.6 nM Cdc42GAP (Fig. 2B). Since the amount of Cdc42 is in far excess of Cdc42GAP, the reaction can be treated by Michaelis-Menten kinetics with the adjustment for the intinsic rate of GTP hydrolysis by Cdc42 (see Equation 2 in the Experimental Procedures). The data were fitted to give a K m of 3.08 M and a V max 2.02 M/min (Fig. 2C), and the K cat value of 2103.9 min Ϫ1 was derived thereon (Table II). The assumption that all GAPs purified were active was applied when calculating K cat values in the current study, although this may lead to an underestimation if a fraction of the GAP was rendered inactive during the preparation procedures; there is currently no established method to determine this. Additional experiments were carried out with GST-Cdc42GAP fusion protein to see if the GST moiety in GST-GAP fusions may affect the kinetics of the reaction. Essentially identical results were ob-tained (K m value at 2.96 Ϯ 0.98 M, and at 2151 Ϯ 198 min Ϫ1 ), indicating that the GST portion of the fusion does not interfere with the GAP-Cdc42 interaction. Thereafter we used GSTfused GAPs for further investigation.
To compare the kinetic parameters obtained by the MESG/ phosphorylase assay with that by the fluorescence method, we also determined the initial rate of tryptophan fluorescence increase, which correlates with the rate of GTP hydrolysis at increasing concentrations of Cdc42-GTP (Fig. 3). The best fit of the data yielded a K m of 2.61 M and a V max of 2.38 M/min, and K cat was calculated to be 2578 min Ϫ1 . These are in good agreement with the kinetic constants determined by the MESG/ phosphorylase system (Table II), thus further legitimize the use of the MESG/phosphorylase assay to examine other Cdc42-GAP or Cdc42-effector interactions described below.
Kinetics of Interaction of Cdc42-GTP with the GAP Domains of Bcr, 3BP-1, and p190 -The interactions of the GAP domains of Bcr, 3BP-1, and p190 with Cdc42-GTP were examined by the MESG/phosphorylase system and compared with that of the Cdc42GAP. Similar to that shown in Fig. 2, Fig. 4 shows the initial rates of P i release measured as a function of concentrations of Cdc42-GTP at fixed concentrations of GST-GAP domains. Data analysis revealed the K m values of Bcr, 3BP-1, and p190 to be 23.72, 50.47, and 59.19 M, and the V max values to be 3.02, 7.55, and 2.52 M/min, respectively (Table II). The catalytic efficiencies of the GAP domains of Bcr, 3BP-1, and p190 are at least 60, 160, and over 500-fold less than that of the GAP domain of Cdc42GAP, respectively, and the difference in K m seems to be a major factor for the varied ability of the GAPs to stimulate GTP hydrolysis of Cdc42.
Cdc42 Binding to GAPs-The interaction of GAPs with Cdc42 at steady state was investigated by measuring the ability of GMP-PNP-bound (a nonhydrolyzable GTP anolog) or GDP-bound Cdc42 to inhibit competitively the GAP-stimulated hydrolysis of [␥-32 P]GTP-bound Cdc42. The concentration range in which Cdc42-GMP-PNP or Cdc42-GDP inhibits GAPmediated GTP hydrolysis is an indication of its affinity (K d ) for MgCl 2 . Arrow indicates the time point at which a final concentration of 5 mM MgCl 2 was added to the reaction mixture. The dashed line represents the period of addition and mixing. B, comparison of the time courses for single-turnover GTP hydrolysis by Cdc42 measured by the MESG system (squares), the tryptophan fluorescence change (diamonds), and the [␥-32 P]GTP filter-binding assay (triangles). Experimental conditions for each assay were as described under "Experimental Procedures." Data of GTP hydrolysis were best fitted into a single exponential to derive the intrinsic rate constant of Cdc42 (K c ). C, temperature dependence of the intrinsic GTPase activity of Cdc42. ␥P i release from 8.0 M Cdc42 was monitored by the MESG system during the time couse at 20°C, 30°C, and 37°C, respectively, upon the addition of 5 mM MgCl 2 as indicated by the arrow.   Fig. 1B were fitted into a single exponential function to derive the intrinsic rate constant K c . GTP-hydrolysis of Cdc42 measured at different temperatures by the P i /MESG method as depicted in Fig. 1C were also analyzed similarly. ND, not determined. Results are representative of three independent experiments.  (Table III). These K d values compare well with the K m values obtained for the GAPs (Table II), suggesting a rapid equilibrium binding of GAPs to Cdc42-GTP, which is not a rate-limiting step in the GAP-catalyzed reactions. The correlation between catalytic efficiency and binding affinity of the GAPs examined suggests that tight binding of a GAP domain may contribute to the stabilization of a transition state of Cdc42 in facilitating GTP hydrolysis. To see if there is an effect of product inhibition in the time courses of GAP-stimulated GTP hydrolysis, we measured the affinity of Cdc42-GDP to various GAPs in a similar assay, and found that both Cdc42GAP and 3BP-1 bound to Cdc42-GDP with negligible affinity, while Bcr and p190 recognize Cdc42-GDP with ϳ100-fold and ϳ2-fold lower affinity, respectively, than the GMP-PNP-bound Cdc42 (Table III). We therefore conclude that there were minimal, if any, Cdc42-GDP-inhibitory effect on the GAP reactions of Cdc42. This is in contrast with the GAP-catalyzed reaction of RhoA in which the reaction product RhoA-GDP demonstrates a comparable affinity to GAP as the reactant Rho-GTP, 2 suggesting a diversity of the mechanism of GAP-stimulated GTP hydrolysis among Rho proteins.
Interaction of Cdc42-GTP with Cdc42-binding Domains of Putative Effectors-We are interested in examining the interactions of putative effectors with Cdc42 and making comparisons with that of the GAPs, given that some RhoGAPs have been speculated to act as effectors for the G-protein function (43,44). The CRIB domain of rat ␣-PAK and the RasGAPrelated domain of IQGAP1 were previously observed to inhibit both the intrinsic and GAP-stimulated GTP hydrolysis of Cdc42 (33,40), and recently, it was reported that the CRIB domain of mouse PAK3 has a similar effect on Cdc42 (49). To determine the direct binding affinity of the effector domains to Cdc42-GTP, we conducted GTPase assays of Cdc42 in the presence of various doses of purified Cdc42-binding domains of IQGAP1, PAK1, and WASP (Fig. 6). As shown in Fig. 6A, the intrinsic GTPase activity is inhibited by the addition of IQGAP1, and this inhibitory effect is dependent on the concentrations of the effector. The initial rates of GTP hydrolysis as a function of the concentrations of the inhibitory effector domains were fitted by a nonlinear regression (Equation 3) to extract the binding constants (K i ) to Cdc42-GTP (Fig. 6B), and K i values of 0.082, 0.78, and 4.15 M for IQGAP1, PAK1, and WASP, respectively, were obtained (Table IV). Thus, IQGAP1 seems to bind to Cdc42-GTP with ϳ10-fold higher affinity than PAK1, and ϳ50-fold higher affinity than WASP. When the interaction of GAP domains with Cdc42 are compared with these effector domains, Cdc42GAP is the only GAP tested to fall in the range of effector interaction affinity (approximately micromolar range), while Bcr, 3BP-1, and p190 interact with  Fig. 1A at 20°C, and the reactions were initiated by adding Cdc42GAP at nM concentrations indicated together with MgCl 2 (arrow). Inset, initial rate of P i release from the Cdc42-GTP complex as a func-Cdc42 at least 5-fold less tightly than any of the effectors examined.

DISCUSSION
Increasing evidence has emerged implicating Cdc42 as an essential component in a variety of cellular transduction pathways; this prominent member of Rho family small GTPases mediates the reorganization of actin polymerization process in mammalian and yeast cells and regulates critical cellular processes such as mitogen-activated kinase kinase (Fus3/Kss1 in yeast and JNK/p38 in mammals) activation, p70 S6 kinase activation, and cell cycle progression (1)(2)(3), and may also have a role in mammalian development (50). Studies aimed at understanding the biochemical mechanisms underlying the biological effects of Cdc42 have led to the discovery of numerous regulators and potential effector targets, which specifically interact with the activated GTP-bound form of Cdc42 (19,36) and may serve to down-regulate or initiate one or more pathways bifurcating at the GTPase itself. In this paper we describe the quantitations of some of the interactions involving Cdc42 that have been implicated in the regulation or transduction of relevant signals, and provide for the first time a direct comparison of the kinetic and steady state parameters of the interactions with four mammalian RhoGAPs (Cdc42GAP, Bcr, 3BP-1, and p190) and three putative effectors (IQGAP1, PAK1, and WASP). Our data show that both the GAPs and the effectors interact with Cdc42-GTP by a fast equilibrium mechanism, and the affinities of the interaction correlate with the catalytic GAP activities of the GAPs or the GTPase-inhibitory activities of the effectors. The broad spectrum of binding affinities of these cellular factors to Cdc42 (K d varying from 80 nM to 50 M) suggest that they employ distinct mechanism to play roles, if any, in Cdc42-mediated signaling pathways.
Due to the relatively fast intrinsic GTP hydrolysis rate of Cdc42 (10-fold faster than Ras), the interactions of Cdc42 with GAPs and with putative effectors, which display GTPase-inhibitory protein function on Cdc42 similar to that by the GDP dissociation inhibitors in this respect (51), can be readily measured by the change in ␥P i release. The MESG/phosphorylase assay allows us to monitor in real time the GTP hydrolysis process by converting the P i signal to an absorbance change at 360 nm, and offers a few advantages over two other available methods to measure GTPase activity of Cdc42; it provides faster, continuous, and more quantitative trace of GTP hydrol-

TABLE II
Kinetic parameters of GAPs that regulate Cdc42 under single turnover conditions GAP activities of Cdc42GAP, Bcr, 3BP-1, and p190 analyzed in Figs. 2-4 by nonlinear regression yielded the V max and K m values listed, and the K cat and K cat /K m were derived thereon. Data represent results obtained by the MESG/phosphorylase method unless otherwise indicated. GAP reactions were carried out at 20°C under single turnover conditions under which MgCl 2 (5 mM) was in great excess of EDTA (0.1 mM). Concentrations of the GAP domains used in the measurements were 0.96 nM, 11.4 nM, 34.7 nM, and 36.3 nM for Cdc42GAP, Bcr, 3BP-1, and p190, respectively, by the MESG/phosphorylase method, and 0.94 nM Cdc42GAP by the fluorescence method. These concentrations of GAPs fell within the linear range of the absorbance or fluorescence increase in response to increasing doses of GAPs. Results are representative of at least three independent measurements.  ysis compared with the [␥-32 P]GTP filter binding method, and it does not suffer from the possible interference by inner filter effect at high concentrations of fluorophore (i.e. tryptophan residue of Cdc42) or by nonphysiological artifact of fluorescent GTP-analogs such as the N-methylanthraniloyl-GMP-PNP of the fluorescence method (48,49). As demonstrated in Figs. 1 and 3, this assay yielded similar kinetic parameters of the intrinsic GTPase activity as well as the GAP-stimulated GTPase activity of Cdc42 as that measured by the radioactive filter-binding and/or the tryptophan fluorescence methods (when the Cdc42 concentration is less than 10 M), thus establishing it as a reliable, easy-to-perform, and quantitive method to study the molecular interactions involving Cdc42 and other Rho family small GTPases. More attractively, the MESG/phosphorylase assay can be utilized to study the molecular interplays of multiple regulatory proteins of Rho family GTPases under multiple turnover conditions, e.g. when the activation of the small G-proteins is provided by guanine nucleotide exchange factors (Dbl-like proteins) and the deactivation or transduction of signals is mediated by GAPs or effectors. 3 Consistent with previous qualitative estimations (21-23), our quantitative analysis of Cdc42-GAP interactions indicates that Cdc42GAP represents the most active GAP specie toward Cdc42, with a catalytic efficiency (K cat /K m ) at least over 60-fold higher than the other GAPs been examined. Given its wide spectrum of tissue distributions, it is likely that Cdc42GAP may serve as one of the major negative regulators of Cdc42 signaling pathway, whereas the significantly lower catalytic efficiencies and binding affinities of 3BP-1 and p190 (ϳ160and ϳ600-fold lower, and ϳ16and ϳ19-fold less tight, respectively) suggest that they are probably not involved in direct regulation of Cdc42. So far, in vivo evidence of cellular functions of the GAPs has been lacking. Microinjections performed using the GAP domains suggest that while both Bcr and 3BP-1 3 Baolin Zhang and Yi Zheng, unpublished observation.

TABLE III Affinity of GAPs for Cdc42
The competition reactions were carried out under conditions described in Fig. 5. The dissociation constants (K d ) were derived by fitting data such as shown in Fig. 5  can down-regulate Rac-mediated membrane ruffling, Cdc42GAP and p190 seem to be able to inhibit the Rho-regulated actin-stress fiber formation in fibroblast cells (29,52). Our results that the GAP domains of Bcr and 3BP-1 contain supreme catalytic efficiency toward Rac with K m values at 3-5 M, and p190 and Cdc42GAP have K m values of 2-3 M for Rho 2 are consistent with these observations in cells. But it remains unclear what effect these GAPs may bear on the Cdc42-regulated processes such as filopodia formation and p70 S6 kinase and JNK/p38 kinase activations.
The mechanism of GAP activation of Cdc42 GTPase activity appears to be similar to that of Ras-RasGAP interaction, i.e. a fast equilibrium binding of GAP to the GTPase followed by GTP hydrolysis (53). The K m values of GAPs examined are in excellent agreement with the K d values observed under steady state conditions, indicating that GAP binding is not the rate-limiting step. However, the current data do not indicate further which of the consequent steps of ␥P i cleavage and product P i release is rate-limiting. Although experimental conditions have been carefully chosen to ensure that the reaction kinetics favor the GTP hydrolysis step, it remains possible that P i release may become rate-limiting at very high GAP concentrations. The role of GAP in the activation of GTPase reaction may be to stabilize a particular transition state conformation to await for the ␥-phosphate cleavage by intrinsic reactive groups of Cdc42, or to supply active site residues important for efficient catalysis, both of which have supporting evidence in the case of RasGAP action (53)(54)(55). Our data do not distinguish between these possibilities.
The binding of putative effectors to Cdc42, including PAK1, WASP, and IQGAP1, invariably led to the inhibition of the relatively fast intrinsic rate of GTP hydrolysis. It is speculated that this GTPase-inhibitory protein activity accompanying the binding and activation of downstream effectors may bear physiological relevance by prolonging the time at which Cdc42 remains active (33,40). We have taken advantage of this property of effector-Cdc42 interaction to derive a direct binding constant for the Cdc42-binding domains of PAK1, WASP, and IQGAP1, and found that the RasGAP-like domain of IQGAP1 contains over 10-fold higher affinity toward Cdc42-GTP than the CRIB domains of PAK1 and WASP (Table IV). The significant difference (over 7-fold) in binding affinity of the two CRIB domains also suggest that structural variability in the CRIB motif and the residues surrounding it may have an effect on its Cdc42binding function. Given the demonstrated role of WASP in Cdc42-mediated cytoskeleton regulation (39), it may also be inferred that either functional coupling can occur between Cdc42-GTP and its effector at M affinity, or WASP may be further regulated under physiological conditions to achieve higher affinity for Cdc42. In light of the recent finding that certain RhoGAPs may contribute to the effector functions of small GTPases (43, 44), a direct comparison of the binding affinities of these effector domains with that of GAP domains measured in this study seems to indicate that, while it is possible for Cdc42GAP to act as a putative effector for Cdc42 with a K d similar to that of WASP at M range, it is less likely that 3BP-1, or p190 functions to transduce signals from Cdc42 with a K d value at over 50 M. However, the relatively low binding affinity of some GAPs such as Bcr does not rule out their role in down-regulation or effector function of Cdc42 if the case for RasGAP, which has a K d of ϳ20 M for Ras, can be brought up as a comparison (56,57).
It is important to note that current studies utilizing the limited Cdc42-interactive domains of GAPs and effectors provide only a simplified model which may differ from the situations in which full-length molecules are likely to behave. Recently, the amino-terminal sequences of p120RasGAP containing the SH2 and SH3 domains were found to be required for the full catalytic GAP activity toward Ras (58). Along these lines, many of the RhoGAPs and effectors are large and multifunctional, therefore a much more complicated intra-or intermolecular interactions involving additional structural motifs of the molecules may contribute to the regulation of binding and/or catalytic interaction with Cdc42, in analogy to RasGAP. For example, we have previously observed that the carboxylterminal fragment of Bcr containing the GAP domain is a much more active GAP than the full-length Bcr immunoprecipitated from cell lysates (21), and the GAP activity of the RhoGAP member n-chimaerin has been reported to be subjected to regulation by lipid-binding (59). Thus, to measure and compare the binding and catalytic activities of intact RhoGAPs and effectors to Rho family GTPases will be of considerable interests in the future.
In summary, the data presented here provide direct comparison of the binding and catalytic activities of a panel of mammalian RhoGAP domains and putative effector domains to Cdc42. The superb catalytic efficiency of Cdc42GAP and high binding affinity of IQGAP1 constitute them as potentially important players in the regulation or transduction of Cdc42mediated signals. The GTPase-inhibitory reactions of the effector domains were carried out as in Fig. 6A. The initial rate of GTP hydrolysis as a function of the effector domain was treated by nonlinear regression fitting to Equation 3 under "Experimental Procedures," as shown in Fig. 6B to derive the dissociation constants (K i ) for the interaction between CDC42-GTP and the respective efector domains.