The Cdc42 Binding and Scaffolding Activities of the Fission Yeast Adaptor Protein Scd2

The small GTP-binding protein Cdc42, the guanine nucleotide exchange factor Scd1, the p21-activated kinase Shk1, and the adaptor protein Scd2 are involved in the Cdc42-dependent signaling cascade in fission yeast. In the present study, we analyzed the Cdc42 binding and scaffolding activities of Scd2 by co-precipitation assays. We found that two SH3-containing regions, amino acid residues 1-87 (CB1 (Cdc42-binding region 1)) and 110-266 (CB2), of Scd2 can bind to the GTP-bound form of Cdc42. CB2 is cryptic because of the intramolecular binding between the SH3 domain in CB2 (SH3(C)) and the PX domain and binds to Cdc42 only when the Scd2 PB1 domain binds to the PC motif-containing region (residues 760-872) of Scd1. This CB2.Cdc42 association, which would stabilize the open configuration of Scd2, enables the SH3(C) domain to bind to the polyproline motif of Shk1. We also found that the GTP-bound form of Cdc42 binds to the CRIB motif of Shk1 more strongly than to Scd2. Thus, Scd2 functions as a scaffold to form a protein complex, and the GTP-bound Cdc42 might be transferred effectively from the upstream activator Scd1 to the downstream effector Shk1 via Scd2.

The Cdc42/Rac protein is a member of the Rho family of small GTP-binding proteins (1)(2)(3)(4). The Cdc42/Rac protein is involved in eukaryotic cellular signaling pathways regulating cell morphology, cell adhesion, actin dynamics, cell cycle progression, and kinase signaling (1)(2)(3)(4), as a molecular switch cycling between the GDP-bound, inactive form and the GTPbound, active form (4 -8). Guanine nucleotide exchange factors for Cdc42/Rac, containing Dbl homology domains, promote GDP dissociation and GTP association. The GTP-bound form of Cdc42/Rac interacts with a variety of target proteins to initiate downstream responses (4 -8). The Cdc42 protein of the fission yeast Schizosaccharomyces pombe is involved in controlling polarized cell growth (9 -11). The disruption of the cdc42 gene is lethal to fission yeast, and the cdc42 mutants exhibit abnormal morphological phenotypes (9). Genetic and molecular studies suggest that a Dbl homology domain-containing Scd1 protein functions as the guanine nucleotide exchange factor for the fission yeast Cdc42 protein and that the GTP-bound Cdc42 activates Ste20 homologous kinase 1 (Shk1) 1 (12)(13)(14)(15).
The fission yeast Shk1 belongs to the family of the p21activated kinase (PAK), which is a highly conserved serine/ threonine kinase activated by the Cdc42/Rac protein. PAK consists of an amino-terminal regulatory region and a carboxylterminal kinase domain. Most of the PAKs contain a Cdc42 and Rac interactive binding (CRIB) motif in the regulatory region, which binds to the GTP-bound Cdc42/Rac protein (16,17). Intramolecular binding of the CRIB motif of PAK to the kinase domain inhibits the enzyme activity (18 -22). Recent studies indicate that the Cdc42/Rac protein relieves this autoinhibition, thereby allosterically inducing the activation of the kinase activity (18 -22). The fission yeast PAK family kinase, Shk1, is positively regulated by the adaptor protein, Scd2 (shape and conjugation deficiency 2, also called Ral3), as well as by Cdc42 (12,13). Co-expression of Scd2 enhances the binding of Shk1 to Cdc42 in a yeast two-hybrid system (12,13). Furthermore, the overexpression of Scd2 stimulates the autophosphorylation activity of Shk1 in vivo, whereas the recombinant Scd2 by itself is not able to stimulate it in vitro (12). These results indicate that Scd2 participates in the activation of Shk1 but is not sufficient for it.
In the present study on the fission yeast system, we analyzed the interactions of Scd2 with Scd1, Cdc42, and Shk1. We found that the four domains of Scd2 have important roles in the interactions with these proteins. The SH3(N)-and SH3(C)containing regions (CB1 and CB2 (Cdc42-binding regions 1 and 2)) bind to the switch regions of the GTP-bound form of Cdc42.
The SH3(C) domain intramolecularly binds to the PX domain, which prevents CB2 from binding to Cdc42. CB2 is able to bind to the GTP-bound form of Cdc42 when the PB1 domain of Scd2 is associated with the PC motif-containing region of Scd1 (Scd1-PC). Furthermore, the binding of the polyproline-binding site of the SH3(C) domain to Shk1 is enhanced when Scd2 binds to Scd1-PC and the GTP-bound form of Cdc42. Thus, these four proteins can form a quaternary complex that includes two Cdc42-binding proteins. On the other hand, we also found that the CRIB motif of Shk1 binds preferentially to Cdc42, even in the presence of Scd2.

EXPERIMENTAL PROCEDURES
Plasmid Construction-The DNA fragments encoding Scd1 and Scd2 were amplified by PCRs from the scd1 and scd2 genes, which were kindly provided by Dr. M. Yamamoto (University of Tokyo). The DNA fragments encoding Cdc42 and Shk1 were amplified by reverse transcription-PCR from mRNA extracted from fission yeast cells. We also used PCR to amplify the BamHI/XhoI fragments encoding the Scd2 and Cdc42 proteins and the BamHI/EcoRI fragments encoding the Shk1 proteins. To obtain the XhoI/SalI fragment encoding Scd1(760 -872), a SalI site was introduced after the stop codon. Point mutations of the Scd2 proteins were introduced into the DNA fragments encoding Scd2 by site-directed mutagenesis using two subsequent PCRs. Point mutations of the Cdc42 proteins were introduced into the DNA fragments encoding Cdc42, using the QuikChange site-directed mutagenesis kit (Stratagene). The DNA fragments were subcloned into pGEX 4T-3 (Amersham Biosciences) for the preparation of the glutathione S-transferase (GST) fusion proteins. The His 6 -tagged forms of the proteins were prepared by subcloning into pRSET A (Invitrogen). The integrity of all of the PCR products was confirmed by sequencing.
Protein Preparation-GST fusion and His 6 -tagged proteins were expressed in Escherichia coli at 30°C and were purified using glutathione-Sepharose 4B (Amersham Biosciences) and Talon resins (Clontech), respectively, according to the manufacturer's instructions. The His 6 -Shk1(1-658) protein was purified as described previously (12). Scd1(760 -872) was obtained by thrombin cleavage of the purified GST-Scd1(760 -872). The His 6 -tagged proteins and Scd1(760 -872) were further purified by ion exchange and/or gel filtration chromatography. These proteins were stored in 20 mM Tris-HCl buffer (pH 7.5) containing 100 mM NaCl, 1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride in 50% glycerol at Ϫ20°C or in 15% glycerol at Ϫ80°C. The purified proteins were fractionated on SDS-polyacrylamide gels that were stained with SYPRO orange (Molecular Probes). The SYPRO orange-stained gels were photographed with a luminescent image analyzer, LAS-1000plus (FUJIFILM), and the protein concentrations were determined from the relative band intensities with bovine serum albumin (Bio-Rad) as the standard. The Cdc42 protein was complexed with GDP or GTP␥S before the binding assays, as previously described (34).
Co-precipitation Assay-The GST fusion proteins (ϳ1 g in 100 l of phosphate-buffered saline (PBS) containing 0.05% Triton X-100 and 5 mM MgCl 2 ) were each mixed with 5 l of glutathione-Sepharose 4B beads. The mixture was incubated at 4°C for 1 h with 10 pmol of His 6 -tagged proteins in the absence or presence of 10 pmol of Scd1(760 -872). After this incubation, the resin was washed twice with PBS containing 5 mM MgCl 2 . The bound proteins were eluted from the resin by boiling in Laemmli's buffer, resolved by SDS-PAGE, and transferred to a nitrocellulose membrane. The immunoblots were probed with the anti-His 6 -tag monoclonal antibody (Clontech), and the co-precipitating His 6 -tag proteins were detected by development or with the LAS-1000plus image analyzer (FUJIFILM), using the ECL immunodetection system (Amersham Biosciences). The developed Western blots were scanned, and the relative band intensities were determined with a Bioimage densitometer (MilliGen). The relative band intensities detected with LAS-1000plus were quantified with ImageGauge (FUJIFILM).

RESULTS AND DISCUSSION
GTP Dependence of the Scd2(1-536) Binding to Cdc42-Chang et al. (13) used a yeast two-hybrid system to show that Scd2 binds to the fission yeast Cdc42 protein, but it is not known whether this binding is GTP-dependent. The budding yeast homologue of Scd2, Bem1p, was shown to bind to the budding yeast Cdc42 protein (Cdc42p) in a GTP-dependent manner (33). To determine the GTP dependence of the Scd2 binding to the fission yeast Cdc42 protein, we performed a co-precipitation assay using recombinant GST fusion and His 6 tagged proteins. We incubated the GST fusion form of the full-length Scd2 (GST-Scd2 (1-536)) with the GDP-bound or GTP␥S-bound His 6 -tagged form of the fission yeast Cdc42 protein (His 6 -Cdc42). The protein complexes bound to the glutathione-Sepharose 4B beads were subjected to SDS-PAGE. The co-precipitating His 6 -Cdc42 was detected with an anti-His 6 -tag antibody. As shown in Fig. 2 (lanes 1 and 2), the full-length Scd2(1-536) binds to the GTP␥S-bound Cdc42 but not to the GDP-bound Cdc42.
Binding Activities of the Scd2 Fragments to Cdc42-Bose et al. (33) showed that the deletion of the amino-terminal 140 amino acid residues of Bem1p decreases the binding to the GTP-bound form of Cdc42p. Furthermore, their data showed that the deletion of either the amino-terminal 234 or the carboxyl-terminal 35 amino acid residues of Bem1p impaired the binding to the GTP-bound form of Cdc42p (33). In Bem1p, the amino-terminal 234-residue region contains two SH3 domains (residues 79 -127 and 160 -212), and the carboxyl-terminal 35 amino acid residues are included in the PB1 domain (residues 472-551) (24,33).
In our results, the amino-terminal half of Scd2, Scd2(1-266), was important to bind to the GTP-bound form of Cdc42, but Scd2(267-536) was not. Furthermore, the regions located adjacent to the SH3 domains, Scd2(1-24) and Scd2(186 -266), were necessary for the binding to Cdc42. These regions have no known binding motifs and do not have significant homology with each other. However, using homology search algorithms, we found that Scd2(1-24) and Scd2(181-228) have low sequence homology to the corresponding regions of Bem1p, within amino acid residues 52-78 and 213-260, respectively The wild-type and W62R mutant of GST-Scd2(1-87) (1 g) were tested for binding to the GDP-and GTP␥S-bound His 6 -Cdc42 (10 pmol). The co-precipitating His 6 -Cdc42 proteins were detected by Western blotting with ECL development, and the band intensities were determined by densitometry. Intensities relative to that obtained after the incubation of the GTP␥S-bound His 6 -Cdc42 with GST-Scd2(1-87) are shown. The data represent the averages of at least three independent experiments. B, association of the point mutants of GST-Scd2(1-266) with His 6 -Cdc42. Intensities relative to that obtained after the incubation of the GTP␥S-bound His 6 -Cdc42 with GST-Scd2(1-266) are shown. C, association of the point mutants of GST-Scd2(110 -536) with His 6 -Cdc42. Intensities relative to that obtained after the incubation of the GTP␥Sbound His 6 -Cdc42 with GST-Scd2(110 -536/W160R) are shown. D and T denote the GDP-and GTP␥S-bound His 6 -Cdc42, respectively.
FIG. 2. Association of the Scd2 fragments with Cdc42. The GST fusion forms of the Scd2 fragments (1 g) were tested for binding to the GDP-and GTP␥S-bound His 6 -tagged forms of the fission yeast Cdc42 protein (His 6 -Cdc42) (10 pmol). The co-precipitating His 6 -Cdc42 proteins were detected by Western blotting and development using the ECL immunodetection system (Amersham Biosciences), as described under "Experimental Procedures." D and T denote the GDP-and GTP␥S-bound His 6 -Cdc42, respectively.
(data not shown). This agrees with the result that the deletion of the amino-terminal 234 amino acid residues of Bem1p impairs the binding to Cdc42p (33). However, our results are inconsistent with the report that the deletion of the carboxylterminal 35 amino acid residues of Bem1p impairs Cdc42p binding (33). Considering that Scd2(1-450), lacking the PB1 domain, was insoluble, a possible explanation for this observation could be that the PB1 domain might be important for the conformational stability of the full-length Scd2.
The three-dimensional structures of many SH3 domains have been determined using NMR and x-ray crystallography, revealing that the well conserved tryptophan residues in the hydrophobic ligand-binding sites of the SH3 domains directly contact the ligands (35)(36)(37)(38). Therefore, we replaced the conserved Trp 62 of the SH3(N) domain with an arginine residue (Fig. 1A). As a result, the GTP␥S-bound His 6 -Cdc42 did not co-precipitate with GST-Scd2(1-87/W62R) (Fig. 3A). This indicates that Scd2(1-87), including the amino-terminal 24 amino acid residues and the SH3(N) domain, composes a functional region that binds to the GTP-bound form of Cdc42. Thus, we designated this region, Scd2(1-87), as CB1 (Fig. 1A).
In this context, it was reported that a subunit of the human NADPH oxidase, the p47 phox protein, exhibits an intramolecular binding between the PX and SH3 domains (23). The p47 phox protein is an adaptor protein containing one PX domain and two SH3 domains. The PX domain of the p47 phox protein intramolecularly binds to the second SH3 domain. We investi-gated whether the intramolecular binding between the SH3(C) and PX domains of Scd2 inhibits the second Cdc42-binding site. His 6 -tagged forms of Scd2(1-266) and Scd2(267-536) were prepared (Fig. 1C) and were tested for their bindings to the GST fusions of Scd2(267-536) and Scd2(1-266), respectively. We found that Scd2(1-266) and Scd2(267-536) bind to each other (Fig. 4, A, lane 1, and B, lane 1) and, furthermore, that the substitution of an Arg residue for either the conserved Trp 160 residue of the SH3(C) domain or the conserved Pro 364 residue of the polyproline motif in the PX domain impairs this binding (Fig. 4, A, lanes 2 and 3, and B, lane 2). These results clearly demonstrate that Scd2 has the intramolecular association between the ligand-binding site of the SH3(C) domain and the polyproline motif of the PX domain.
Based on these results, we next investigated how the mutations that disrupt the intramolecular binding affect the inter- Cdc42 Binding and Scaffolding of Scd2 molecular binding of Scd2 to Cdc42. The W160R mutation did not affect the Cdc42 binding activity of Scd2(1-266) (Fig. 3B). This means that the ligand-binding site, or at least the Trp 160 residue, in the SH3(C) domain is not directly involved in the binding to the GTP␥S-bound His 6 -Cdc42. On the other hand, the W160R mutation enabled Scd2(110 -536) to bind to the GTP␥S-bound His 6 -Cdc42 (Fig. 3C). We also found that the P364R mutation within the PX domain has the same effect on the binding activity of Scd2(110 -536) to Cdc42 (Fig. 3C). These results suggest that the intramolecular binding between the SH3(C) and PX domains prevents Scd2(110 -536) from binding to Cdc42. It should be noted here that GST-Scd2(110 -536/ W160R) and GST-Scd2(110 -536/P364R) bound to the GTP␥Sbound form, but not the GDP-bound form, of His 6 -Cdc42 (Fig.   3C). These results indicate that amino acid residues 110 -266, including the SH3(C) domain, composes a functional region that binds to the GTP-bound form of Cdc42, independently of CB1. Thus, we designated the region containing amino acid residues 110 -266 as CB2 (Fig. 1A).
The GTP-bound form of Cdc42 binds to regions containing CRIB motifs in many Cdc42 targets (2,16). On the other hand, no CRIB motif exists in several Cdc42 targets, such as phosphatidylinositol 3-kinase (39,40), phospholipase D (41), phospholipase C-␤2 (42), Cdc42-interacting protein 4 (43), IQGAP (3,44), and the ␥-subunit of the coatomer complex (45). In the present study, we identified two Cdc42-binding regions, each containing an SH3 domain but lacking a CRIB motif. Among other Cdc42 target proteins lacking the CRIB motif, however,  Shk1(141-216). The mutants of the GTP␥S-bound His 6 -Cdc42 (10 pmol) were tested for binding to GST-Shk1(141-216) (1 g). The co-precipitating His 6 -Cdc42 proteins were quantified with a luminescent image analyzer, LAS-1000plus. Intensities relative to that obtained after the incubation of the wild-type His 6 -Cdc42 with GST-Shk1(141-216) are shown. The data represent the averages of at least three independent experiments. E, association of the point mutants of His 6 -Cdc42 with GST-Scd2(1-87). Intensities relative to that obtained after the incubation of the wild-type His 6 -Cdc42 with GST-Scd2(1-87) are shown. F, association of the point mutants of His 6 -Cdc42 with GST-Scd2(110 -536). The point mutants of the GTP␥S-bound His 6 -Cdc42 were tested for binding to GST-Scd2(110 -536) in the presence of Scd1(760 -872) (10 pmol). Intensities relative to that obtained after the incubation of the wild-type His 6 -Cdc42 with GST-Scd2(110 -536) are shown. the p85 subunit of phosphatidylinositol 3-kinase and Cdc42interacting protein 4 both contain one SH3 domain (39,40,43). In the case of the p85 subunit, the Rho-GAP homology domain has been reported to be responsible for the binding to the GTP-bound Cdc42 (39). On the other hand, it is unclear whether the Cdc42-interacting protein 4 SH3 domain binds to Cdc42⅐GTP.
Cdc42 Mutants-To examine which regions of the fission yeast Cdc42 protein bind to CB1 and CB2 of Scd2, we introduced mutations into the His 6 -Cdc42 protein. Considering that the SH3(N) domain in CB1 is essential for the binding to Cdc42, we substituted Ala for the Pro residues that were supposed to be on the protein surface on the basis of the solution structures of the GTP-bound form of the human Cdc42 protein (46 -48). The Pro 26 residue is in the ␣1 helix (Fig. 6A). The Pro 34 residue is in switch I, and the Pro 69 and Pro 73 residues are in switch II. In addition to these common switch regions, the region containing the ␣3 helix and the adjacent ␤4 strand also exhibits significant changes between the solution structures of the GDP-and GTP-bound forms of the human Cdc42 protein and is therefore called "switch III" (48). Thus, the Pro 99 and Pro 106 residues in the ␣3 helix were mutated. The Pro 123 residue is in the ␣I helix, which is the insert helix specific to the Rho family of small GTP-binding proteins (48). The Pro 179 residue is in the carboxyl-terminal variable loop region. In addition to these proline residues, we replaced the Tyr 32 and Asp 38 residues in switch I with Ala residues, because many effector proteins bind to the switch I regions of small GTPbinding proteins (4,49). The P26A and P73A mutants of the His 6 -Cdc42 proteins were insoluble (data not shown). The expression levels of the P99A and P106A mutants were relatively low (Ͻ0.1 mg from 250 ml of cell culture), whereas those of the other mutants were as high as that of the wild-type His 6 -Cdc42 (0.4 -0.5 mg from 250 ml cell culture) (data not shown). We also constructed the P99A/P106A mutant of His 6 -Cdc42, and unexpectedly, the expression level of the P99A/P106A mutant was as high as that of the wild-type His 6 -Cdc42 (data not shown). Therefore, we used the P99A/P106A mutant, instead of the P99A and P106A mutants. Thus, these seven mutants of the His 6 -Cdc42 proteins were prepared for the experiments to examine the binding activities of the Cdc42 mutants to Scd2 as well as Shk1 (Fig. 6B).
Shk1 CRIB Binding Activities of the Cdc42 Mutants-The regulatory region of the Shk1 protein consists of several polyproline motifs and the CRIB motif (Fig. 6C). We first examined the activities of the Cdc42 mutants to bind to the CRIB motif-containing fragment of Shk1, Shk1(141-216) or Shk1-CRIB. Among the mutated residues, only the Asp 38 residue was expected to be on the interface with the CRIB motif on the basis of the complex structures of the CRIB motifs and the human Cdc42 protein (46,47,49,50). In fact, only the D38A mutation impaired the binding of Cdc42 to Shk1(141-216) (Fig. 6D). GST-Shk1(141-216) bound to the other mutants as much as the wild-type His 6 -Cdc42 (Fig. 6D), and these bindings were GTP␥S-dependent (data not shown). Our results are consistent with previous reports that the D38A mutation drastically impairs the binding of human and budding yeast Cdc42 proteins to CRIB motifs (49,51). Thus, the six novel Cdc42 mutants (Y32A, P34A, P69A, P99A/P106A, P123A, and P179A) retain the GTP-dependent Shk1-CRIB binding activity.
CB1 and CB2 Binding Activities of the Cdc42 Mutants-To examine the CB1 binding activities of the Cdc42 mutants, we incubated GST-Scd2(1-87) with the His 6 -tagged forms of the Cdc42 mutants and measured the amounts of the His 6 -Cdc42⅐GTP␥S complex that co-precipitated with GST-Scd2(1-87). The Y32A, P34A, D38A, P99A/P106A, and P123A mutants of His 6 -Cdc42 co-precipitated with GST-Scd2(1-87) as efficiently as the wild-type His 6 -Cdc42 (Fig. 6E) in the GTP␥S-dependent manner (data not shown). There is a candidate for the polyproline motif, which is a typical binding target of SH3 domains, in the carboxyl-terminal region of the fission yeast Cdc42 protein (amino acid residues 178 -192; DPPVPH-KKKSKCLVL). However, the CB1 binding to the P179A mutant was similar to that of the wild-type Cdc42 (Fig. 6E). We also generated Cdc42 (1-177), which lacks the entire prolinerich carboxyl-terminal region, as a recombinant His 6 -tagged protein, but the deletion did not significantly affect the CB1 binding of Cdc42 (data not shown). In contrast, the P69A mutation decreased the GST-Scd2(1-87) binding (Fig. 6E). The binding reduction by the P69A mutation in switch II is consistent with the GTP dependence of the binding. However, switch II of the fission yeast Cdc42 protein does not have the typical polyproline motif, although the SH3(N) domain is involved in the CB1⅐Cdc42 association. Accordingly, the CB1⅐Cdc42 binding is not the typical SH3⅐polyproline binding but an atypical ligand-binding of the SH3(N) domain. In contrast, the adaptor protein Nck was reported to bind to a small GTP-binding protein (human R-Ras) via the typical SH3⅐polyproline association; the SH3 domain of Nck binds to the polyproline motif in the carboxyl-terminal variable loop region of R-Ras, and this binding does not depend on GTP (52).
To examine the CB2 binding activities of the Cdc42 mutants, we next incubated GST-Scd2(110 -536) with the His 6 -Cdc42 mutant proteins in the presence of Scd1(760 -872). GST-Scd2(110 -536) co-precipitated the P34A, P69A, P123A, and P179A mutants as well as the wild-type His 6 -Cdc42 (Fig. 6F). In contrast, the Y32A, D38A, and P99A/P106A mutations decreased the amount of the co-precipitated His 6 -Cdc42 (Fig. 6F), although these bindings were still GTP␥S-dependent (data not FIG. 8. The Scd1(760 -872) and Cdc42 dependence on the association of Scd2 with Shk1. In the presence of Scd1(760 -872) and/or the GTP␥S-bound His 6 -Cdc42 (10 pmol), the wild-type and W160R mutant of GST-Scd2(1-536) (1 g) were tested for binding to His 6 -Shk1(1-658) (10 pmol). The co-precipitating His 6 -Shk1(1-658) was detected and quantified with the LAS-1000plus image analyzer. Intensities relative to that obtained after the incubation of shown). These results show that switch I of the GTP-bound form of Cdc42 is involved in the CB2 binding as well as in the Shk1-CRIB binding and are consistent with the report that mutations in switch I impair the binding of the budding yeast Cdc42 protein to Bem1p (51). On the other hand, our results indicate that switch III, containing the Pro 99 and Pro 106 residues, is also involved in the GTP-dependent binding of Cdc42 to CB2. Therefore, the binding mechanism of Cdc42 to CB2 differs from that to the CRIB motif of Shk1. Furthermore, the two Cdc42-binding regions, CB1 and CB2, of Scd2 recognize different switch regions of Cdc42.
We next incubated GST-Shk1(141-216) with the GTP␥Sbound His 6 -Cdc42, followed by the addition of His 6 -Scd2 . Unlike the case of the Scd2 fragments, the amount of His 6 -Cdc42 that co-precipitated with GST-Shk1(141-216) did not decrease with the addition of His 6 -Scd2(1-536) (Fig. 7D). Similar results were also obtained in the presence of Scd1(760 -872). These results suggest that Scd2 and Shk1-CRIB are not able to bind simultaneously to Cdc42 but rather that the GTPbound form of Cdc42 binds to Shk1-CRIB more strongly than to Scd2.
The results described above show that Scd2 unambiguously binds to Cdc42 in the ternary complex of Scd1, Scd2, and Cdc42. Meanwhile, the quaternary complex includes two proteins that can bind to the GTP-bound form of Cdc42. The isolated CRIB motif of Shk1 can bind preferentially to Cdc42, but it remains elusive whether the CRIB motif of the fulllength Shk1 actually binds to Cdc42 in the quaternary complex. We examined how the addition of His 6 -Shk1(141-216) affects the quaternary complex formation. The amount of Cdc42 in the ternary complex significantly decreased with the addition of His 6 -Shk1(141-216) (Fig. 10, A, lanes 1 and 2, and  B), whereas that in the quaternary complex only slightly decreased (Fig. 10, A, lanes 3 and 4, and B). This result suggests that the CRIB motif of the full-length Shk1, but not Scd2, binds to Cdc42 in the quaternary complex and protects it from the external Shk1-CRIB fragment.
Biological Importance of the Cdc42 Binding and Scaffolding Activities of Scd2-In the present study, we have shown that Scd2 functions as a scaffold to assemble Scd1, the GTP-bound form of Cdc42, and Shk1 together. Furthermore, our results suggest that Scd2 and Shk1 are unlikely to bind simultaneously to the same Cdc42 molecule and that the CRIB motif of the full-length Shk1, but not Scd2, binds to Cdc42 in the quaternary complex. The quaternary complex formation might induce the Cdc42-binding target to change from Scd2 to Shk1. Therefore, we propose the following model of the Scd2 activation of Shk1. First of all, Scd1 promotes the GDP dissociation and the GTP association of Cdc42 (Fig. 11, step 1). The GTPbound, active Cdc42 transiently binds to CB1 and CB2 of Scd2 that forms a binary complex with Scd1 (step 2). The Scd1-dependent CB2⅐Cdc42 association stabilizes the open configuration of Scd2, which enables the SH3(C) domain to bind to the polyproline-containing region of Shk1. Thus, Shk1 participates in the complex formation, and the CRIB motif of Shk1 can receive Cdc42 from Scd2 (steps 3 and 4). Consequently, the GTP-bound form of Cdc42 is effectively transferred from the upstream activator Scd1 to the downstream effector Shk1 via Scd2, leading to the activation of Shk1 (step 5).
The full-length human PAK homologue, ␣PAK, was reported to bind to Cdc42 with lower affinity than that of the isolated CRIB motif, indicating that the intramolecular interaction inhibits the Cdc42 binding activity as well as the kinase activity of ␣PAK (22). On the other hand, our results indicate that the Cdc42 binding of the CRIB motif in the full-length Shk1 is facilitated in the quaternary complex. Therefore, we suggest that the association of the full-length Shk1 with the scaffolding protein Scd2 promotes the dissociation of the CRIB motif from the kinase domain, possibly by keeping Cdc42⅐GTP in the proximity of the CRIB motif and/or changing the conformation of Shk1.