Analysis of the Mechanisms of Action of the Saccharomyces cerevisiae Dominant Lethal cdc42 G12V and Dominant Negative cdc42 D118A Mutations*

The Saccharomyces cerevisiae Cdc42p GTPase is localized to the plasma membrane and involved in signal transduction mechanisms controlling cell polarity. The mechanisms of action of the dominant negative cdc42 D118Amutant and the lethal, gain of functioncdc42 G12V mutant were examined. Cdc42D118A,C188Sp and its guanine-nucleotide exchange factor Cdc24p displayed a temperature-dependent interaction in the two-hybrid system, which correlated with the temperature dependence of the cdc42 D118A phenotype and supported a Cdc24p sequestration model for the mechanism ofcdc42 D118A action. Five cdc42mutations were isolated that led to decreased interactions with Cdc24p. The isolation of one mutation (V44A) correlated with the observations that the T35A effector domain mutation could interfere with Cdc42D118A,C188Sp-Cdc24p interactions and could suppress the cdc42 D118A mutation, suggesting that Cdc24p may interact with Cdc42p through its effector domain. Thecdc42 G12V mutant phenotypes were suppressed by the intragenic T35A and K183–187Q mutations and in skm1Δ and cla4Δ cells but not ste20Δ cells, suggesting that the mechanism of cdc42 G12Vaction is through the Skm1p and Cla4p protein kinases at the plasma membrane. Two intragenic suppressors ofcdc42 G12V were also identified that displayed a dominant negative phenotype at 16 °C, which was not suppressed by overexpression of Cdc24p, suggesting an alternate mechanism of action for these dominant negative mutations.

The Saccharomyces cerevisiae Cdc42p GTPase is localized to the plasma membrane and involved in signal transduction mechanisms controlling cell polarity. The mechanisms of action of the dominant negative cdc42 D118A mutant and the lethal, gain of function cdc42 G12V mutant were examined. Cdc42 D118A,C188S p and its guanine-nucleotide exchange factor Cdc24p displayed a temperature-dependent interaction in the twohybrid system, which correlated with the temperature dependence of the cdc42 D118A phenotype and supported a Cdc24p sequestration model for the mechanism of cdc42 D118A action. Five cdc42 mutations were isolated that led to decreased interactions with Cdc24p. The isolation of one mutation (V44A) correlated with the observations that the T35A effector domain mutation could interfere with Cdc42 D118A,C188S p-Cdc24p interactions and could suppress the cdc42 D118A mutation, suggesting that Cdc24p may interact with Cdc42p through its effector domain. The cdc42 G12V mutant phenotypes were suppressed by the intragenic T35A and K183-187Q mutations and in skm1⌬ and cla4⌬ cells but not ste20⌬ cells, suggesting that the mechanism of cdc42 G12V action is through the Skm1p and Cla4p protein kinases at the plasma membrane. Two intragenic suppressors of cdc42 G12V were also identified that displayed a dominant negative phenotype at 16°C, which was not suppressed by overexpression of Cdc24p, suggesting an alternate mechanism of action for these dominant negative mutations.
The establishment of cell polarity is crucial for the control of many cellular and developmental processes, such as the generation of cell shape, the intracellular movement of organelles, and the secretion and deposition of new cell surface constituents (1). Polarized growth in the yeast Saccharomyces cerevisiae occurs in response to both internal and external signals, resulting in different morphological structures (2)(3)(4)(5). The mechanics of cell polarity initiation during the mitotic cell cycle can be divided into three sequential phases: (i) nonrandom bud site selection; (ii) organization of proteins at the bud site; and (iii) bud emergence and polarized growth. Genetic and biochemical studies have identified over 25 proteins, including several GTPases and components of the actin cytoskeleton, that are involved in the regulation of the cell polarity pathway in S. cerevisiae (1,6,7).
At least six members of the Ras superfamily of GTPases (Rsr1p/Bud1p, Cdc42p, Rho1p, Rho2p, Rho3p, and Rho4p) are involved in controlling cell polarity in S. cerevisiae. These proteins are active when in the GTP-bound state and inactive in the GDP-bound state (8,9). The activity of these GTPases is controlled by regulatory proteins, such as guanine-nucleotide exchange factors, GTPase-activating proteins, and guanine-nucleotide dissociation inhibitors, as well as by the intracellular localization of the GTPase. Rsr1p/Bud1p is a member of the Ras subfamily and is responsible for bud site selection at one of the two cell poles, but it is not required for bud emergence or polarized cell growth (10 -12). Cdc42p is a member of the Rho/Rac subfamily and is involved in bud site selection, bud emergence, polarized growth, and cytokinesis (13)(14)(15)(16). The Rho proteins have been implicated in bud formation, actin reorganization, polarized growth, and activation of ␤-glucan synthesis (17)(18)(19)(20)(21)(22)(23).
Highly conserved (80 -85% identical) functional homologs of S. cerevisiae Cdc42p have been characterized in Schizosaccharomyces pombe (24,25), Caenorhabditis elegans (26), Drosophila melanogaster (27), and Homo sapiens (28,29), suggesting that Cdc42p may have conserved functions in these other eukaryotes. Analyses of the morphological phenotypes of dominant lethal S. cerevisiae cdc42 alleles indicated that Cdc42p functions in bud emergence and the subsequent polarized cell growth and cytokinesis (16). These data included the observation that the cdc42 G12V mutation resulted in dominant lethality and large, multibudded cells, suggesting that the mutant protein was activated (GTP-bound) and constitutively interacting with downstream effectors of the pathway. These effectors may include Cla4p, Ste20p, and/or Skm1p, three S. cerevisiae members of the Pak family of protein kinases that interact with GTP-bound Cdc42p (25, 30 -34). In contrast, the cdc42 D118A mutant exhibited a temperature-dependent, dominant negative phenotype, suggesting that Cdc42 D118A p was inactive (GDP-bound) but could bind and sequester a cellular factor necessary for the budding process (16,35). A candidate for this cellular factor was Cdc24p due to its ability to multicopysuppress the cdc42 D118A mutation and because a cdc24 ts cdc42 ts double mutant displayed synthetic lethality (35). In addition, Cdc24p showed limited amino acid sequence similarity with the Dbl proto-oncoprotein, which acts as a guaninenucleotide exchange factor for human Cdc42p (36), and biochemical evidence indicated that Cdc24p catalyzes guaninenucleotide exchange on Cdc42p in vitro (37).
In localization studies, S. cerevisiae Cdc42p was found to be targeted to the plasma membrane in the vicinity of secretory vesicles that are found at the site of bud emergence, to the tips and sides of enlarging buds, and to the tips of mating projections in ␣-factor arrested cells (38). Cdc42p contains the Cterminal Lys 183 -Lys-Ser-Lys-Lys-Cys-Thr-Ile-Leu sequence that is modified by geranylgeranylation at the Cys residue, which is necessary for its anchoring within the plasma membrane (38,39). This prenylation is deemed necessary because the cdc42 C188S mutation resulted in a nonfunctional protein that fractionated almost exclusively into soluble pools (16,38) and because the cdc42 C188S mutation can suppress the cdc42 G12V , cdc42 Q61L , and cdc42 D118A lethal mutations (16). However, whether geranylgeranylation is necessary and sufficient for Cdc42p targeting to the sites of polarized growth is unknown. The polybasic domain of four lysine residues that is next to the prenylated Cys residue is another possible localization determinant. Similar domains in the K-Ras protein are important for membrane targeting; altering these Lys residues to Gln results in delocalized K-Ras proteins (40,41).
To determine the mechanisms of action of the cdc42 D118A and cdc42 G12V mutations, the interactions between Cdc42 D118A p and Cdc24p were examined in the yeast two-hybrid protein system, and extragenic and intragenic suppressors of the cdc42 G12V allele were characterized. The data support the hypothesis that the cdc42 D118A dominant negative phenotype is due to sequestration of Cdc24p away from endogenous Cdc42p and suggest that the nature of the cdc42 G12V growth and morphological phenotypes is due to improper interactions with the Skm1p and Cla4p protein kinases at the plasma membrane. Two Cdc42 effector domain mutations were also identified that either suppressed the cdc42 D118A phenotype or disrupted Cdc42 D118A p-Cdc24p two-hybrid interactions, suggesting that Cdc24p may interact with Cdc42p through its effector domain.

EXPERIMENTAL PROCEDURES
Reagents, Media, and Strains-Enzymes, dideoxy sequencing, and polymerase chain reaction kits and other reagents were obtained from standard commercial sources and used as specified by the suppliers.
Protein determinations were performed using the Bio-Rad protein assay kit using bovine serum albumin as the standard, and immunoblots were developed using either the Enhanced Chemiluminescence (ECL) system (Amersham Corp.) or Renaissance system (NEN Life Science Products). Horseradish peroxidase-conjugated goat anti-rabbit IgG, protease inhibitors (phenylmethylsulfonyl fluoride, N-tosyl-L-phenylalanine chloromethyl ketone, aprotinin, leupeptin, and pepstatin), and glass beads (425-600 m) were obtained from Sigma. Cdc42p-specific antibodies were isolated and purified as described previously (16).
Conditions for the growth and maintenance of bacterial and yeast strains have been described (42,43). The S. cerevisiae strains used are listed in Table I. The S. cerevisiae strain HF7c (Ref. 44; provided by David Beach, Cold Spring Harbor Laboratories) was used in two-hybrid screens. Yeast transformations were performed as described (43), and transformants were selected on synthetic complete drop-out media lacking the appropriate amino acid(s) and containing 2% glucose as a carbon source (i.e. SC-Leu). Transformants were transferred to solid or liquid media containing 2% glucose or 2% raffinose plus 2% galactose (for induction of the GAL promoter) for growth analysis and photomicroscopy.
Plasmids and DNA Manipulations-Standard procedures were used for recombinant DNA manipulations (42) and plasmid isolation from Escherichia coli (45). Sequencing was either by the dideoxy chain termination method (46) with the U.S. Biochemical Corp. Sequenase sequencing kit or through automated sequencing at the Vermont Cancer Center DNA Sequencing Facility. Site-directed mutagenesis was performed with the MUTAGENE kit (Bio-Rad). Plasmids pBM272 (47), pRS306 and pRS315 (48), pRS425 (49), pJJ215 (50), pPGK (51), pAS1-CYH2 (52), pGAD2F (53), pRS315(CDC24-B) (35), and YEp351(CDC42), pGAL-CDC42, pRS315(CDC42), pRS315(cdc42 G12V ), pRS315(cdc42 D118A ), pRS315(cdc42 D118A,C188S ), pGAL-cdc42 G12V , and pGAL-cdc42 D118A (16) have been previously described. Plasmid pRS315(GAL1/10) was constructed by blunt-ending the 685-base pair EcoRI-HindIII fragment from pBM272 containing the divergent GAL1/10 promoters with the Klenow fragment of DNA polymerase I and inserting it into the unique SmaI site of pRS315. Plasmid pPGK2 was constructed by inserting the PGK promoter from plasmid pPGK on a XhoI plus SalI fragment into the unique SalI site of a derivative of pRS425, which had the BamHI to HindIII fragment from its multiple cloning site removed. 2 Plasmid pPGK2E, which has the unique EagI site of pPGK2 removed, was constructed by cleaving pPGK2 with EagI, blunt-ending with S1 nuclease, and religating with T4 DNA ligase. Plasmids pPGK2-CDC42 and pPGK2E-CDC42 were constructed by inserting a PCR-generated CDC42 gene contained on a BamHI plus HindIII fragment into either pPGK2 or pPGK2E that had been digested with BamHI plus HindIII. pGAD2F-CDC24 was constructed by inserting the ϳ4-kilobase pair BamHI plus HindIII fragment from pRS315(CDC24-B), which was blunt-ended with the Klenow fragment of DNA polymerase, into pGAD2F that had been digested with BamHI and blunt-ended with the Klenow fragment of DNA polymerase. This study a HD2 was generated by mating W303-1A with Y763. HD2-1 was generated by integrating the skm1::HIS3 fragment into HD2. TRY2 was generated by mating RAK63 with HD2-1-6D. TRY1 was generated by integrating the cla4::TRP1 fragment into TRY2. HD2-1-2B and HD2-1-6D are congenic strains derived from HD2-1. SKM1 (Ref. 32; GenBank TM accession number X69322) was isolated from W303-1A genomic DNA by PCR using the 5Ј-primer TCC-CCCGGGCATATGAAGGGCGTAAAAAAG (underlined sequence is a NdeI site and contains the SKM1 start codon; double underlined sequence is a SmaI site) and the 3Ј-primer GCTCTAGACTCGAGACATA-ACGCGAAGCAAACG (underlined sequence is a XhoI site; double underlined sequence is a XbaI site; nonunderlined sequence is the reverse complement of ϩ144 to ϩ163 downstream of the SKM1 stop codon). The resulting 2547-base pair PCR fragment was digested with SmaI plus XbaI and inserted into SmaI plus XbaI-digested pTZ18U (54). To generate a skm1::HIS3 disruption, the SmaI-XhoI fragment containing HIS3 from plasmid pJJ215 was blunt-ended with Klenow fragment and inserted into the blunt-ended unique StuI site at ϩ241 of the SKM1 coding region in pTZ18U(SKM1). The resulting plasmid was digested with SmaI plus XbaI, releasing a 3.86-kilobase pair skm1::HIS3 fragment that was used to transform the diploid strain HD2 to His ϩ . Sporulation and tetrad analysis of stable His ϩ transformants yielded His ϩ haploid cells in which the skm1::HIS3 allele had replaced a wildtype SKM1 allele at its chromosomal location, which was confirmed by DNA-DNA blot hybridization (data not shown).
PCR and Site-directed Mutagenesis-The PCR mutagenesis protocol was based on the Zhou et al. (55) protocol previously described. Plasmid pRS315(cdc42 D118A, C188S ) was amplified under essentially standard reaction conditions (reaction volume was 200 l; reaction conditions were 10 mM Tris-HCl, pH 8.8, 50 mM KCl, 50 M each dNTP, 2 fmol of template, 50 pmol of each primer, and 5 units of AmpliTaq DNA polymerase; cycle profile (30 cycles total) was 94°C for 1 min, 50°C for 2 min, and 72°C for 3 min) on a Perkin-Elmer DNA thermal cycler model 480. The 5Ј-and 3Ј-primers were, respectively, GAATTCAAGCT-TCGTATTAGGTCTTCC (underlined sequence is an EcoRI site; double underlined sequence is a HindIII site; nonunderlined sequence is Ϫ20 to Ϫ6 upstream of the CDC42 start codon), and CGCGGATCCGG-GCATATACTAATATG (underlined sequence is a BamHI site; nonunderlined sequence is the reverse complement of ϩ2 to ϩ18 downstream of the CDC42 stop codon). The pool of PCR fragments was digested with NdeI plus BamHI (the NdeI site is at the CDC42 start codon) and directionally inserted into NdeI plus BamHI-digested pAS1-CYH2. The pool of pAS1-CYH2(cdc42 D118A,C188S,X (X is any possible new mutation) plasmids was amplified in E. coli and transformed into S. cerevisiae HF7c cells already containing pGAD2F(CDC24). Plasmid pAS1-CYH2(cdc42 D118A,C188S ) was obtained using the same procedure; the entire coding region was sequenced to confirm the presence of only those two mutations and no other spurious mutations.
The introduction of the T35A mutation into the wild-type, cdc42 G12V , or cdc42 D118A mutant gene was accomplished by a modified Kunkel method of site-directed mutagenesis as described previously (16). The starting templates were uracil-containing single-stranded DNA isolated from E. coli CJ236 cells containing pRS315(CDC42), pRS315(cdc42 G12V ), or pRS315(cdc42 D118A ). The nucleotide sequence of the T35A mutagenic primer was GTTCCAGCAGTGTTCG (underlined G is A in the wildtype sequence). pGAL versions of the new double mutants were constructed using the same method as the original pGAL single mutants (16); the pRS315 double-mutant plasmids were digested with HpaI to delete CDC42 upstream sequences, and the 685-base pair EcoRI-Hin-dIII fragment from pBM272 containing the divergent GAL1/10 promoters was blunt-ended with the Klenow fragment of DNA polymerase I and inserted into this unique HpaI site. A PCR approach was employed to generate the Cdc42 T35A,D118A,C188S triple mutant for use in twohybrid protein studies. The starting template was 100 ng of pRS315-(cdc42 T35A,D118A ) and the nucleotide sequence of the C188S mutagenic primer was CGCGGATCCGACTACAAAATTGTAGATTTTTTACTTT-TCTTGATAACAGG (Cys 188 to Ser; underlined G is C in the wild-type sequence; the double underlined sequence is a BamHI restriction site). The mutagenic primer was used as the 3Ј-primer in the PCR reaction with the same 5Ј-primer used in the PCR mutagenesis reactions (see above). Primers were used at a final concentration of 0.1 nM in a final reaction volume of 50 l. The PCR cycling parameters were 30 cycles of 94°C for 1 min, 55°C for 2 min, and 72°C for 3 min, followed by a 5-min extension reaction at 72°C. The resulting PCR product was digested with NdeI plus BamHI (the NdeI site is at the CDC42 start codon) and directionally inserted into NdeI plus BamHI-digested pAS1-CYH2.
Introduction of the K183-187Q mutations into either the wild-type or cdc42 G12V mutant gene was accomplished using a PCR approach. The nucleotide sequence of the K183-187Q mutagenic primer was CGCG-GATCCGACTACAAAATTGTACATTGTTGACTTTGCTGGATAACA-GG (lysine 183, 184, 186, and 187 to Gln; underlined G bases are T in the wild-type sequence; the double underlined sequence is a BamHI restriction site). The mutagenic primer was used as the 3Ј-primer in the PCR reaction with the same 5Ј-primer used in the PCR mutagenesis reactions (see above). Template for the PCR reaction was 1 g of either YEp351(CDC42) to generate the cdc42 K183-187Q quadruple mutant or pRS315(cdc42 G12V ) to generate the cdc42 G12V,K183-187Q quintuple mutant. Primers were used at a final concentration of 0.1 M in a final reaction volume of 100 l, and the PCR cycling parameters were as described above. The mutant genes were placed under the control of the PGK promoter by digesting the resulting ϳ600-base pair PCR products with HindIII plus BamHI and inserting into HindIII plus BamHIcleaved pPGK2. For all mutant genes, the entire coding region for each mutant gene was sequenced to confirm the presence of the desired mutation(s) and the absence of any spurious mutations.
Selection of Intragenic Suppressors of the cdc42 G12V Mutant-A PCR mutagenesis approach was taken to identify intragenic mutations that suppress the cdc42 G12V dominant lethality. The starting template was pGAL-cdc42 G12V ; the 5Ј-and 3Ј-primers utilized were the same as used in the PCR mutagenesis of cdc42 D118A,C188S (see above). 30 pmol of each primer and 2 fmol of template were used in a reaction volume of 100 l. The PCR cycling parameters were 30 cycles of 94°C for 15 s, 50°C for 30 s, and 72°C for 2 min. The library of PCR products obtained was extracted with PCI (phenol:chloroform:isoamyl alcohol, 25:24:1), ethanol-precipitated, and resuspended in 20 l of sterile distilled H 2 O. The resulting library of fragments was digested with BamHI plus HindIII and ligated into BamHI plus HindIII-cleaved pPGK2. The resulting library of pPGK2 plasmids was transformed into E. coli SURE cells by electroporation; ϳ24,000 ampicillin-resistant transformants were pooled, and plasmid DNA was extracted and resuspended in 200 l of sterile distilled H 2 O.
Expression of the cdc42 G12V mutant gene on plasmids is lethal to wild-type W303-1A cells and does not give rise to viable transformants (16). Therefore, intragenic cdc42 G12V suppressors were identified by an increased transformation frequency of W303-1A cells with the pPGK2 plasmid library containing mutagenized cdc42 G12V genes (see above). A 5-l aliquot of the plasmid library was transformed into W303-1A cells, and half of the transformants were incubated on SC-Leu media at 30°C and half on SC-Leu at 35°C. A total of 1400 Leu ϩ transformants were obtained at 30°C and 856 Leu ϩ transformants at 35°C. Screens for secondary temperature-dependent growth phenotypes were performed on SC-Leu media at 23 and 16°C (see "Results"). Plasmid DNA from yeast transformants was recovered into E. coli SURE cells and subjected to dideoxy sequencing protocols to confirm the presence of the original G12V mutation and the appearance of new coding region mutations.
To separate the single mutations identified as cdc42 G12V suppressors from the G12V mutation, the EagI-BamHI restriction fragment from the double mutant genes, which only contained the new single mutations, was substituted for the wild-type EagI-BamHI fragment in pPGK2E-CDC42. The presence of only the single mutations was confirmed by automated DNA sequencing (Vermont Cancer Center DNA Sequencing Facility). To further analyze two of the single mutant genes identified in the screen (S86P and S89P), they were placed under the control of the wild-type CDC42 promoter and integrated into the genome. The single mutant genes were excised from plasmid pPGK2E by NdeI plus SalI digestion, which generates the mutant genes without a promoter, and then inserted into plasmid pRS315(CDC42) that had been digested with NdeI plus SalI, releasing the wild-type gene. The resulting mutant genes under CDC42 promoter control were excised from the plasmids by NotI plus SalI digestion and inserted into the integrating plasmid pRS306, which had been digested with NotI plus SalI. These plasmids were then integrated into the genome of the diploid strain DJD6 -11 (cdc42⌬::TRP1/CDC42 ura3-52/ura3-52) at the ura3-52 locus by cutting the pRS306 derivative plasmids with BsmI, which has a unique site in the URA3 gene, and transforming DJD6 -11 with the linearized plasmids, selecting for Ura ϩ transformants at 23°C. Stable Ura ϩ transformants were subjected to tetrad analysis to follow the ura3-52::cdc42 S86P ::URA3 or ura3-52::cdc42 S89P ::URA3 marked loci. In more than 10 tetrads each, Ura ϩ , Trp ϩ haploid cells could not be isolated (data not shown), indicating that the cdc42 S86P and cdc42 S89P mutant genes do not encode functional Cdc42 proteins.
Cell Fractionation and Immunoblot Analyses-Cell fractionation experiments were performed as described previously (38). Briefly, cells containing PGK promoter-driven cdc42 mutant genes on plasmids were grown in SC-Leu liquid media to midlog phase at 23°C. ϳ1 ϫ 10 8 cells were collected, washed with water, resuspended in 200 l of lysis buffer (0.3 M sorbitol, 140 mM NaCl, 50 mM Tris, pH 8.0) with protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride and 1:1000 dilutions of 1 mg/ml stock of aprotinin in water, 1 mg/ml stock of N-tosyl-Lphenylalanine chloromethyl ketone in 95% ethanol, 1 mg/ml stock of leupeptin in water, and 1 mg/ml stock of pepstatin in methanol), and lysed on ice by vortexing with 425-600 M acid-washed glass beads. Greater than 90% cell lysis was verified by light microscopy. Cells lysates were spun at 500 ϫ g for 4 min at 4°C; the 500 ϫ g supernatants were then spun at 10,000 ϫ g for 10 min at 4°C, and the pellets were resuspended in the same volume of lysis buffer. To assess the relative amount of Cdc42p in each fraction, equal volumes of each fraction were loaded onto an SDS-12.5% polyacrylamide gel for immunoblot analysis.
For immunoblots, protein samples were diluted 1:1 in SDS-lysis buffer (57) containing 40% ␤-mercaptoethanol, heated at 100°C for 5 min, and separated on an SDS-polyacrylamide gel, and protein was transferred to nitrocellulose paper (BA-S83, 0.2 m; Schleicher and Schuell). Affinity-purified anti-Cdc42p antibody was used at 1:500, and horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody was used at 1:1000. Antibody-antigen complexes were detected with the ECL system or Renaissance system.
Photomicroscopy-Cells were grown to log phase in the appropriate synthetic complete media, collected, sonicated briefly, and examined morphologically. To assess the percentage of dead cells in a population, 2 l of a 4 mg/ml solution of methylene blue in distilled H 2 O was added to 100 l of a sonicated cell suspension. Cells that took up the dye and were deep blue were scored as dead cells. Photomicroscopy using Hoffman modulation optics was performed using an Olympus BH-2 epifluorescence microscope. Photographs were obtained using Kodak TMAX 400 film. Digital images (Fig. 4)

Temperature-dependent Interaction between Cdc24p and
Cdc42 D118A,C188S p-The cdc42 D118A mutant displays a dominant negative phenotype at 23°C but not at 30°C or higher temperatures, and overexpression of Cdc24p can suppress this phenotype (35). These data suggested that the cdc42 D118A dominant negative phenotype may be due to the nonfunctional binding of Cdc24p by mutant Cdc42 D118A p such that Cdc24p could not interact with the endogenous wild-type Cdc42p. It also suggested that this interaction could occur at 23 but not 30°C. To test this hypothesis, the interaction between these proteins in the yeast two-hybrid protein system was examined. In frame fusion proteins between Cdc24p and the GAL4 transcriptional activation domain in plasmid pGAD2F and between Cdc42 D118A,C188S p and the GAL4 DNA binding domain in pAS1-CYH2 were generated. The C188S mutation was incorporated to bypass the normal plasma membrane localization of Cdc42p (16,38). A two-hybrid protein interaction between Cdc24p and Cdc42 D118A,C188S p was observed at 23 but not 30 or 34°C (Fig. 1). This interaction correlates with the temperature dependence of the cdc42 D118A mutant phenotype and further supports the hypothesis that the dominant negative phenotype was due to binding of Cdc24p by mutant Cdc42 D118A p. Our data are consistent with a model in which Cdc42 D118A p binds Cdc24p within the cell at 23°C, not allowing the endogenous wild-type Cdc42p to bind, and that interaction is lost at higher temperatures, but we cannot rule out the possibility that our data are the result of a unique behavior of mutant Cdc42 D118A p.
Isolation of Mutations in cdc42 D118A,C188S That Inhibit Interactions with Cdc24p-To define the domain(s) of Cdc42p that interact with Cdc24p, PCR-generated mutations that reduced the two-hybrid interactions with Cdc24p, using a blue-to-white colony color change at 23°C, were introduced into the cdc42 D118A,C188S gene in pAS1-CYH2. From nitrocellulose lifts of ϳ800 colonies, eight colonies of white or pale blue color were chosen for further characterization. The plasmid DNA from these eight colonies was recovered into E. coli, and three plasmids were found to have no CDC42 insert; these colonies were white in the assay, as would be expected. The remaining five plasmids displayed the characteristic cdc42 restriction enzyme banding pattern and contained single point mutations in the cdc42 D118A,C188S gene resulting in the following amino acid changes: V44A, S86P, I117S, T138A, and L165S. All of the new mutant proteins showed a reduced interaction with Cdc24p (Table II), and all were equally expressed in S. cerevisiae as shown by immunoblot analysis (data not shown).
The V44A mutation lies in the putative effector domain (see below), and the I117S is next to the starting D118A mutation in the conserved G-4 domain of GTPases, which has been implicated in the binding of guanine nucleotides (see Fig. 5 1. Two-hybrid protein interactions between Cdc24p and Cdc42 D118A,C188S p. HF7c cells containing pGAD2F-CDC24 and either pAS1-CYH2(cdc42 D118A,C188S ) or pAS1-CYH2-lamin were selected on SC-Leu-Trp media at 23°C and then tested at various temperatures.

TABLE II Two-hybrid protein interactions between Cdc42p and Cdc24p
The indicated Cdc42 proteins were fused in frame to the GAL4 DNA-binding domain in plasmid pAS1-CYH2 and Cdc24p was fused in frame to the GAL4 activation domain in plasmid pGAD2F. Plasmids were transformed into strain HF7c and selected on SC-Leu-Trp at 23°C. The ␤-galactosidase liquid assays were done in triplicate at 23°C, and these results are representative of at least three independent assays. Nitrocellulose lifts were done in Z buffer containing 1 mg/ml X-Gal. cdc42 G12V and cdc42 D118A Mutants-The phenotype of the cdc42 G12V allele suggested that the mutant protein constitutively interacted with a downstream effector that activated the cell polarity pathway. Therefore, mutations that disrupt the interaction between Cdc42p and downstream effectors should suppress this mutant phenotype. In addition, these effector domain mutations, when present as the only mutation, should result in a nonfunctional protein and large, round, unbudded cells. In contrast, the dominant negative phenotype of the cdc42 D118A allele is due to a nonfunctional interaction with an upstream component of the pathway, Cdc24p (see above), and therefore would not be predicted to be suppressed by effector domain mutations. The T35A effector domain mutation has been shown to interfere with the ability of Cdc42p to interact with the Pak family of protein kinases, which are downstream effectors of Cdc42 function (25,30,60,61).
The T35A mutation can suppress the dominant lethality ( Fig. 2A) and morphological abnormalities (Fig. 2B) of the cdc42 G12V mutation at 23°C. Neither the cdc42 T35A nor cdc42 G12V,T35A mutant gene can complement the cdc42-1 ts mutant at 37°C (Fig. 2A), indicating that these alleles encode nonfunctional proteins, presumably due to their inability to interact with downstream effectors of the pathway. This result is substantiated by the morphological phenotype of large, round unbudded cells seen in these mutant cell cultures at 37°C (data not shown). Surprisingly, the T35A mutation can also suppress the morphological phenotype (large, round unbudded cells) of the dominant negative cdc42 D118A allele (Fig.  2B). Cells expressing the cdc42 T35A,D118A double mutant gene for 9 h at 23°C displayed 44% budded cells (n ϭ 200), with 70% of the budded cells exhibiting abnormal bud shapes and/or multiple buds. Given that the cdc42 D118A dominant negative phenotype is due to sequestration of Cdc24p, the suppression of the cdc42 D118A phenotype may be due to an altered interaction of Cdc42 T35A,D118A p with Cdc24p. In fact, introduction of the T35A mutation into the Cdc42 D118A,C188S protein leads to a loss of interaction with Cdc24p in the two-hybrid protein assay (Table II). The T35A mutation cannot suppress the lethal growth defect associated with the cdc42 D118A mutation ( Fig.  2A), and neither the single nor double mutant gene could complement the cdc42-1 ts mutant at 37°C. Interestingly, the V44A effector domain mutation (see above) could suppress both the cdc42 D118A growth and morphological defects. 3 Taken together, these data suggest that the Cdc24p guanine-nucleotide exchange factor interacts with Cdc42p through its effector domain (see "Discussion").
Suppression of cdc42 G12V by the K183-187Q Mutation-To examine the role of the C-terminal polylysine region in targeting of Cdc42p to the plasma membrane, the four Lys residues were altered to uncharged Gln residues in either the wild-type or cdc42 G12V mutant gene. The intragenic K183-187Q mutation was able to suppress the cdc42 G12V lethality, but the quintuple mutant gene was unable to complement the cdc42-1 ts mutant (Fig. 3A). This result suggested that this polylysine region plays an important role in the function of Cdc42p. Interestingly, the quadruple K183-187Q mutant gene can complement the cdc42-1 ts allele at 37°C (Fig. 3A), suggesting that some of this mutant protein can be properly localized and, hence, functional.
To address the mechanism of K183-187Q suppression, the subcellular localization of the K183-187Q mutant protein was examined using cell fractionation protocols (Fig. 3B). Under these experimental conditions, the wild-type Cdc42 protein fractionated predominantly into the particulate pool (Fig. 3B). The Cdc42 K183-187Q protein fractionated into both soluble and particulate pools (Fig. 3B), indicating that the K183-187Q mutation leads to partial loss of membrane localization. The Cdc42 G12V,K183-187Q protein has a similar fractionation pattern as the Cdc42 K183-187Q protein (data not shown). The probability that this fractionation pattern is due to the K183-187Q mutation is bolstered by the fractionation patterns of two other suppressors of the G12V mutation (see below), which show a predominantly particulate fractionation pattern (Fig. 3B). This result is consistent with the K183-187Q mutant phenotypes described above and suggests that the K183-187Q mutation is affecting Cdc42p function by altering its subcellular localization.
Effects of Pak Kinase Deletions on cdc42 G12V Lethality-To test the hypothesis that cdc42 G12V lethality was due to an improper interaction with a downstream effector(s) at the plasma membrane, the effects of Pak kinase deletions on cdc42 G12V lethality were examined. The pGAL-cdc42 G12V plasmid was transformed into strains that had individual single deletions in either of the three Pak kinases, CLA4, STE20, or SKM1, as well as the corresponding double deletion mutants, and growth and morphological phenotypes on galactose-containing media at 23°C were assayed (Table III; Fig. 4). The cdc42 G12V lethality was still observed in the cla4⌬ and ste20⌬ 3 T. J. Richman and D. I. Johnson, manuscript in preparation.

FIG. 2.
Effects of the T35A mutation on the cdc42 G12V and cdc42 D118A phenotypes. A, pGAL plasmids containing the indicated CDC42 genes were transformed into DJTD2-16A cells (cdc42-1 ts ), and transformants were selected on SC-Leu media at 23°C. Individual transformants were streaked to SC-Leu plates containing either 2% glucose or 2% galactose plus 2% raffinose and incubated at 23 and 37°C. B, pGAL plasmids containing the indicated CDC42 genes were transformed into DJTD2-16A cells, and transformants were selected on SC-Leu media at 23°C. Individual transformants were subcultured in SC-Leu liquid media containing 2% raffinose and incubated at 23°C to early log phase when galactose was added to a 2% final concentration. Cells were further incubated at 23°C for 9 h and then sonicated and examined microscopically.
single deletion mutants as well as the cla4⌬skm1⌬ and ste20⌬skm1⌬ double deletion mutants, but cdc42 G12V expression was not lethal in the skm1⌬ mutant (HD2-1-2B), with 68% of the cells appearing normal in morphology (Table III). In addition, cdc42 G12V expression was still lethal in a rga1⌬ mutant (data not shown), which has a deletion in one of the Cdc42p GTPase-activating proteins Rga1p. These results suggested that cdc42 G12V lethality was due, in part, to an interaction with the Skm1p protein kinase. However, when the morphological phenotypes of the cla4⌬ cells were examined, a dramatic change from the typical cdc42 G12V morphological phenotype was observed ( Fig. 4; Table III). Instead of the large, multibudded cell phenotype observed when overexpressing cdc42 G12V in wild-type or ste20⌬ cells (Fig. 4), a new phenotype of large, round cells with one or more small buds (76%) as well as large, round unbudded cells (ϳ10%) was observed. This cellular morphology was also observed in skm1⌬ cells but at a lower frequency (26 versus 76%). The large, round unbudded phenotype is similar to the cdc42 loss of function or dominant negative phenotype; however, the presence of small buds on the large, round cells suggested that bud emergence had occurred in these cells, but growth was restricted to the mother cell and not directed into an enlarging bud. Taken together, these data suggest that the mechanism of cdc42 G12V action is through interactions with Skm1p and Cla4p but not Ste20p or Rga1p (see "Discussion").
Isolation of Temperature-dependent Intragenic Suppressors of cdc42 G12V -A PCR-generated mutant library was screened for intragenic suppressors of the cdc42 G12V mutant gene by increased transformation frequency of wild-type W303-1A cells (see "Experimental Procedures"). Based on the above-mentioned suppressor results, three types of mutations were envisioned arising from this screen: (i) mutations that affect the localization of Cdc42p, such as the C188S and K183-187Q mutations, (ii) mutations in the effector domain, such as the T35A mutation, and (iii) loss of function or null mutations as well as true revertants of the G12V mutation. Leu ϩ transfor- FIG. 3. Analysis of the K183-187Q mutation. A, wild-type and mutant CDC42 genes under the control of the PGK promoter were transformed into DJTD2-16A cells (cdc42-1 ts ), and transformants were selected on SC-Leu media at 23°C. Individual transformants were streaked to SC-Leu plates and incubated at 23 and 36°C. B, protein extracts of wild-type W303-1A cells transformed with plasmids containing CDC42 genes under the control of the PGK promoter were spun at 10,000 ϫ g to produce pellet (P) and supernatant (S) fractions. Equal volumes of protein fractions were run on a 12.5% polyacrylamide gel. Cdc42 protein was detected using affinity-purified anti-Cdc42p antibodies (1:500 dilution) and horseradish peroxidase-conjugated goat antirabbit IgG secondary antibody (1:1000 dilution) with the ECL detection system. a Cells containing the indicated deletions (⌬) were transformed with the pGAL-cdc42 G12V plasmid, selecting for Leu ϩ transformants on SC-Leu media at 23°C. Individual transformants were streaked with SC-Leu plus 2% raffinose plus 2% galactose. The deletion strains used are listed in Table I. The cla4⌬ ste20⌬ double deletion cells are inviable (30) and were not examined.
b Abbreviations for morphological phenotypes are as follows: NB, normal budded cells; NU, normal unbudded cells; LRU, large, round unbudded cells; LRSB, large, round cells with Ն1 small bud; AB, abnormal budded cells, which were either the cdc42 G12V multibudded phenotype (M) or the elongated bud phenotype seen in cla4⌬ strains (EB) (30). Numbers are reported as a percentage of 200 cells counted and are representative of at least two independent experiments.
mants were selected at both 30 and 35°C with the goal of isolating cdc42 G12V suppressors that had secondary temperature-dependent phenotypes, thereby eliminating the third type of mutations. A total of 1400 Leu ϩ transformants were obtained at 30°C and 856 Leu ϩ transformants at 35°C. Of these 2256 transformants, four had secondary temperature-dependent growth phenotypes (data not shown).
The plasmids from these four transformants were purified by passage through E. coli and retransformed into W303-1A. The resulting transformants displayed poor growth at 34, 30, and 23°C and no growth at 16°C (data not shown). These transformants were grown at 30°C and then shifted to 16°C for 18 h to examine their morphological phenotypes. At the semipermissive temperature of 30°C, the four mutants showed an increase in unbudded cells over wild type (Table IV, top), and all four showed a further increase in unbudded cells within the population when shifted to 16°C for 18 h (Table IV, top), suggesting that they were exerting a dominant negative effect over the endogenous wild-type CDC42 allele. There was also an increase in the percentage of dead (i.e. methylene blue-staining) cells in the mutant populations (Table IV, top). In addition, all four mutant genes were lethal in a cdc42-1 ts background (i.e. plasmids containing the mutant genes could not be transformed into DJTD2-16A cells), which is a result similar to that seen with plasmids containing the cdc42 D118A dominant negative mutant (16).
DNA sequence analysis of plasmids from three of these transformants revealed the presence of the original G12V mutation as well as a new single nucleotide point mutation in the CDC42 coding region. All three point mutations were T to C transitions resulting in the S86P, S89P, and C157R amino acid changes (Fig. 5). The fourth plasmid contained the original G12V mutation along with two coding region mutations, resulting in the S89P and K128E amino acid changes. Given the similarity in phenotype between the single S89P mutant and the double S89P,K128E mutant, it is likely that the K128E mutation is silent. Interestingly, the S89P mutation is analogous to the S89F and S89L dominant negative mutations identified in the C. elegans ras homolog let60 (62) and D. melanogaster Cdc42 (Ref. 59; see "Discussion").
To further analyze these mutations, they were separated away from the G12V mutation (see "Experimental Procedures"). Wild-type W303-1A cells containing the cdc42 S86P or cdc42 S89P single mutant gene in plasmid pPGK2E retained the cold-sensitive phenotype displayed in the G12V double mutants (data not shown) and had an increase in the percentage of unbudded cells in the population (Table IV, top). Cells containing the cdc42 C157R single mutant gene did not exhibit a strong cold-sensitive phenotype, but they did have an increase in unbudded cells in the population. Similar growth phenotypes were observed when the single mutant genes were expressed in the cdc42-1 ts mutant strain DJTD2-16A (Table IV, bottom). In addition, the single mutant genes could not complement the cdc42-1 ts mutation at 37°C (Table IV, bottom), indicating that they did not encode functional proteins at that temperature. The cdc42 S86P or cdc42 S89P single mutant gene could not serve as the sole copy of CDC42 in a cdc42⌬ background (see "Experimental Procedures"), further substantiating the theory that these alleles encode nonfunctional proteins.
For the S86P and S89P mutations, the mechanism of suppression of the cdc42 G12V mutation was not due to loss of membrane attachment, because both show a particulate fractionation pattern that was similar to the wild-type Cdc42p fractionation pattern (Fig. 3B). In addition, the 16°C dominant negative phenotype could not be suppressed by overexpression of Cdc24p or by overexpression of the Cdc42p effector Cla4p (data not shown), suggesting that the mechanism of action of these alleles is different from that of the dominant negative cdc42 D118A allele (see above) and not due to sequestration of the Cla4p downstream effector. Interestingly, the S86P mutation was also identified in the screen for mutations that reduced a two-hybrid protein interaction between Cdc24p and Cdc42 D118A,C188S p at 23°C (see above; Table II). Cdc42 D118A,C188S,S86P p does not interact with Cdc24p in the two-hybrid protein system at 23°C (Table II) or at 16°C (data not shown), indicating a different dominant negative mechanism for this allele. DISCUSSION In genetic and biochemical experiments with numerous Rasrelated proteins, a region between residues 26 and 48 has been identified as being required for interactions with downstream effectors (8,63,64). The first indication that this so-called "effector domain" may be an important region of Cdc42p came from a sequence comparison between functional homologs of Cdc42p and other closely related GTPases (14). Functional homologs of Cdc42p from S. pombe, C. elegans, and humans can complement the S. cerevisiae cdc42-1 ts mutation and are 80 -85% identical to Cdc42p, especially in the highly conserved region between residues 26 and 50 (Fig. 5). The human Rac1 protein is 74% identical to Cdc42p but cannot complement the cdc42-1 ts mutation, indicating that it is not a functional homolog. Interestingly, the only region of Rac1p that is significantly different from functional Cdc42p homologs is residues 41-52, the region in Ras-like proteins that interacts with effector proteins. The inability of Rac1p to complement the cdc42-1 ts mutant may therefore be due to its inability to interact with a Cdc42-specific effector.
The S. cerevisiae Cdc42 GTPase interacts with several pro- a Cells were grown at 30°C to midlog phase and sonicated prior to determining the budding index. n ϭ 100 cells.
b Cells were grown at 30°C to midlog phase and then shifted to 16°C for 20 h (for G12V double mutants) or 45 h (for single mutants) prior to determining the budding index. n ϭ 100 cells.
c Cells were grown at 30°C to midlog phase and then shifted to 16°C for 20 h. The percentage of dead cells was determined by staining with methylene blue. n ϭ 100 unbudded cells. teins, including the Ste20, Cla4, and Skm1 protein kinases (30,32,34,61) that are predicted to function downstream in the cell polarity pathway. This prediction is based, in part, on the inability of these proteins to interact with the Cdc42 T35A effector domain mutant protein. The Thr 35 residue lies within the G-2 domain of Ras-related GTPases (8), which is predicted to change conformation upon GTP binding. The results presented herein further define the cdc42 T35A mutation as an effector domain mutation, but the ability of the T35A mutation to suppress the dominant negative cdc42 D118A morphological phenotypes and to disrupt interactions with Cdc24p suggests that this region of Cdc42p may also interact with the Cdc24p exchange factor. Another mutation (V44A) in the effector domain was identified by its ability to disrupt the interaction between Cdc42 D118A,C188S p and Cdc24p, providing further support for this role of the effector domain, and recently, the T35A mutation in the human Cdc42p was found to disrupt responsiveness to Cdc24p-mediated nucleotide exchange activity (58). Taken together, these data suggest that the so-called "effector domain" plays multiple roles in the interactions of Ras-related GTPases with their regulatory and effector proteins.
The ability of the K183-187Q mutation to suppress the cdc42 G12V dominant lethality is due, in part, to the partial delocalization of the mutant protein from the plasma membrane. As opposed to the nonfunctional cdc42 C188S mutant gene that cannot complement the cdc42-1 ts mutant (16), the ability of the cdc42 K183-187Q mutant gene to complement the cdc42-1 ts mutant suggests that this mutation has an intermediate effect on Cdc42p function. In addition, these results suggest that the polylysine domain of Cdc42p is necessary but not sufficient for complete plasma membrane localization. This is an important point, because Cdc42p is targeted to a specific location on the plasma membrane at sites of polarized growth (38) as opposed to general plasma membrane localization of Ras proteins. This C-terminal polylysine region is not found in most Ras-like GTPases, and its positive charges may be functioning in interactions with negatively charged components, either protein or phospholipid, at the membrane site. Whether these interactions play a role in the specific targeting of Cdc42p or in enhancing membrane association is unclear at this point. Interestingly, deletion of this region in the mammalian Cdc42p led to loss of interaction with phosphatidylinositol 4,5-bisphosphate-containing vesicles (65); phosphatidylinositol 4,5bisphosphate has also been shown to enhance nucleotide exchange with Cdc42Hs (65). Further support for the importance of the polybasic region in Cdc42p function comes from the isolation of a new temperature-sensitive cdc42 mutation (K186R) within the polybasic region (15).
The ability of the skm1⌬ and cla4⌬ mutations to suppress the growth and/or morphological phenotypes of the cdc42 G12V mutation suggest that the Cdc42 G12V mutant protein is exerting its lethal effects through these two protein kinases but not through the Ste20p protein kinase. Interpretation of these phenotypes at the protein-protein interaction level is complicated, however, by the presence of endogenous wild-type Cdc42p in these cells. For instance, the lethality of the cdc42 G12V mutation could be due either to a direct effect of Cdc42 G12V p on a cellular process or to an indirect effect of Cdc42 G12V p on the interactions between wild-type Cdc42p and another protein in the cell. The reversal of cdc42 G12V lethality in a skm1⌬ mutant can be explained by postulating a role for Skm1p in either mediating Cdc42 G12V p lethality or in inhibiting the function of the endogenous wild-type Cdc42p in these cells. Skm1p may be functioning either through a direct interaction with Cdc42p or through an indirect interaction with other downstream effectors such as Cla4p. The mechanism by which cdc42 G12V lethal-ity is restored in cla4⌬skm1⌬ and ste20⌬skm1⌬ double deletion mutants is unclear at this point, but it could reflect the inability of Cdc42 G12V p-expressing cells to grow in the presence of only a single Pak-like kinase; again this could be due to a nonfunctional interaction between either Cdc42 G12V p or endogenous wild-type Cdc42p with the remaining Pak-like kinase. It should be noted that in two-hybrid protein studies, Cdc42 G12V p interacts comparably with Cla4p and Ste20p (30,61); twohybrid interactions between Cdc42 G12V p and Skm1p have not been reported.
The presence of Cla4p, either as the sole Pak kinase in the cell or in combination with Ste20p and/or Skm1p, does seem necessary for the cdc42 G12V -dependent generation of multibudded cells. In addition, the absence of Cla4p, and to a lesser extent Skm1p, shifts the Cdc42 G12V mutant phenotype from a multibudded morphology to a large, round cell with one or more small buds, suggesting that Cla4p and Skm1p are not necessary for bud emergence but are necessary for restricting growth to the enlarging bud. This phenotype was reminiscent of the phenotype observed when cdc24 -4 ts cells were arrested with hydroxyurea and then released into 37°C (restrictive temperature) media (66), the phenotype that first suggested Cdc24p also functioned later in the cell cycle beyond bud emergence. This phenotype was also reminiscent of Cdc24p overexpression phenotypes (35), of downstream mutants of the polarity pathway such as rho1 mutants or pkc1 mutants (67)(68)(69), and of Skm1p overexpression phenotypes (32). These data suggesting a role for Cla4p in restricting growth to the enlarging bud are also consistent with those previously obtained for cla4 mutations in wild-type backgrounds (30).
To identify other domains of Cdc42p that are important for function, intragenic suppressors of the cdc42 G12V dominant lethality were isolated. Given the different mechanisms of suppression observed for the T35A and K183-187Q mutations, it was important to assay both the ability of these suppressors to complement the cdc42-1 ts mutant and the subcellular fractionation of the mutant proteins to distinguish between loss of function mutations and loss of localization mutations. The C157R mutation lies in a domain implicated in the responsiveness of the human Cdc42p to Cdc24p-mediated nucleotide exchange activity (58) and within the G-5 domain of Ras-related GTPases, which functions in the binding of guanine nucleotides (8). The Cys 157 residue is unique to Rho/Rac/Cdc42 proteins in this domain, but we have not pursued the C157R mutation further at this time because it does not exhibit a phenotype on its own. The fractionation patterns of the S86P and S89P mutant proteins, as well as their inability to complement the cdc42-1 ts mutant or act as the sole copy of Cdc42p within the cell, suggested that these mutations suppressed the cdc42 G12V phenotype by generating a nonfunctional albeit properly localized protein. However, the dominant negative phenotype of these mutant genes at 16°C suggested that these mutant proteins were able to negatively interact with some component of the pathway, possibly sequestering it away from the endogenous Cdc42p. It is unlikely that the sequestered component was Cla4p or Cdc24p, since overexpression of either was unable to suppress the dominant negative phenotype. In addition, the S86P mutation disrupted the interaction between Cdc42 D118A,C188S p and Cdc24p in a two-hybrid protein assay at both 23 and 16°C, suggesting that the sequestered component was not Cdc24p. Interestingly, the paradigmatic cdc42 T17N dominant negative allele also could not be suppressed by overexpression of Cdc24p. 4 Taken together, these data suggest the mechanism of action of these dominant negative alleles is dif-ferent from that of the cdc42 D118A allele.
The S86P and S89P mutations are within a domain of Cdc42p (residues 82-120) and other GTPases in which dominant negative mutations have been recently isolated (59,62). In addition, mutations in this domain lead to diminished responses to Cdc24p-mediated nucleotide exchange activity (58), suggesting that this domain of Cdc42p plays an important role in its function. The analogous domain in S. cerevisiae Ras2p is involved in the interaction between Ras2p and its GTPaseactivating protein, Ira2p (70,71). This domain in the Ras crystal structure corresponds to the turn between loop 6 and the ␣3 helix (72), a region of the protein that is predicted to be in close proximity to bound nucleotide. Introduction of additional Pro residues into this region (Fig. 5) could have a profound effect on the conformation of the protein and/or its ability to bind nucleotide, thereby leading to loss of interactions with both guanine-nucleotide exchange factors and GTPase-activating proteins.
Overall, these studies have identified the effector domain of Cdc42p as being important for interactions with both downstream effectors and the upstream guanine-nucleotide exchange factor Cdc24p and have identified two new domains of Cdc42p as being important for function and/or membrane localization. Biochemical interaction and genetic suppressor studies in the future may further define the regions of the effector domain that are necessary for interactions with multiple Cdc42p effectors and regulators and may elucidate the mechanism of Cdc42p targeting to the plasma membrane.