Advertisement

Functional characterization of the Cdc42p binding domain of yeast Ste20p protein kinase

Ekkehard Leberer, Cunle Wu, Thomas Leeuw, Anne Fourest‐Lieuvin, Jeffrey E Segall, David Y Thomas

Author Affiliations

  1. Ekkehard Leberer1,
  2. Cunle Wu1,
  3. Thomas Leeuw1,
  4. Anne Fourest‐Lieuvin1,
  5. Jeffrey E Segall2 and
  6. David Y Thomas1
  1. 1 Eukaryotic Genetics Group, Biotechnology Research Institute, National Research Council of Canada, 6100 Royalmount Avenue, Montreal, Quebec, H4P 2R2, Canada
  2. 2 Department of Anatomy and Structural Biology, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY, 10461, USA
View Full Text

Abstract

Ste20p from Saccharomyces cerevisiae belongs to the Ste20p/p65PAK family of protein kinases which are highly conserved from yeast to man and regulate conserved mitogen‐activated protein kinase pathways. Ste20p fulfills multiple roles in pheromone signaling, morphological switching and vegetative growth and binds Cdc42p, a Rho‐like small GTP binding protein required for polarized morphogenesis. We have analyzed the functional consequences of mutations that prevent binding of Cdc42p to Ste20p. The complete amino‐terminal, non‐catalytic half of Ste20p, including the conserved Cdc42p binding domain, was dispensable for heterotrimeric G‐protein‐mediated pheromone signaling. However, the Cdc42p binding domain was necessary for filamentous growth in response to nitrogen starvation and for an essential function that Ste20p shares with its isoform Cla4p during vegetative growth. Moreover, the Cdc42p binding domain was required for cell–cell adhesion during conjugation. Subcellular localization of wild‐type and mutant Ste20p fused to green fluorescent protein showed that the Cdc42p binding domain is needed to direct localization of Ste20p to regions of polarized growth. These results suggest that Ste20p is regulated in different developmental pathways by different mechanisms which involve heterotrimeric and small GTP binding proteins.

Introduction

The mating response of the haploid yeast Saccharomyces cerevisiae provides a genetically tractable model system to study a G‐protein regulated mitogen‐activated protein (MAP) kinase cascade. This response is initiated by the pheromones a and α‐factor released from MATa and MATα cells, respectively. These short polypeptides bind to cell type‐specific receptors and thereby trigger the activation of a heterotrimeric G‐protein that is common to both cell types (for a review, see Herskowitz, 1995). The β and γ subunits of the G‐protein then activate a MAP kinase cascade consisting of Ste7p (a MAP kinase kinase or MEK homolog), Ste11p (a MEK kinase homolog) and the partially redundant MAP kinase homologs Fus3p and Kss1p (Herskowitz, 1995). The Ste20p protein kinase (a MEK kinase kinase) (Leberer et al., 1992a; Wu et al., 1995), Ste5p (Leberer et al., 1993; Whiteway et al., 1995) and Ste50p (Ramezani Rad et al., 1992) play crucial roles in the activation of this MAP kinase cascade which ultimately regulates the Ste12p transcription factor (Dolan et al., 1989) and Far1p, a negative regulator of cyclin/Cdc28 (Chang and Herskowitz, 1990).

This signaling pathway induces differentiation processes required to initiate conjugation of cells with opposite mating types. These cellular processes, which are typical of differentiating cells, include growth arrest in G1 of the cell cycle, morphological and biochemical alterations in the cell surface leading to the mating‐specific ‘shmoo’ morphology and transcriptional activation of genes whose products facilitate cellular and nuclear fusion (Herskowitz, 1995).

Some of the components of this signaling cascade are not only involved in mating but also in inducing morphological transitions of yeast cells. The switch to pseudohyphal and invasive growth requires Ste20p, Ste11p, Ste7p and Ste12p (Liu et al., 1993; Roberts and Fink, 1994). The MAP kinases involved in this response are as yet unidentified (Liu et al., 1993; Roberts and Fink, 1994).

Ste20p shares an essential function with the closely related isoform Cla4p in polarized cell growth during budding and cytokinesis (Cvrckova et al., 1995). Both enzymes belong to the Ste20p/p65PAK family of protein kinases believed to be involved in the regulation of conserved MAP kinase pathways conveying signals from the cell surface to the nucleus (Bagrodia et al., 1995; Polverino et al., 1995; Brown et al., 1996; Frost et al., 1996). Members of this protein kinase family bind the Rho‐like small GTP binding proteins Cdc42 or Rac at a conserved binding site which includes a conserved sequence motif of 16 amino acids that appears to be sufficient for binding of Cdc42 or Rac (Burbelo et al., 1995). This sequence motif is also present in the non‐receptor tyrosine kinases p120ACK from mammalian cells and DPR2 from Drosophila, the mixed lineage kinase (MLK) family of serine/threonine protein kinases and in numerous non‐kinase proteins without defined biochemical activity, including WASP, a protein implicated in the immunodeficiency disorder Wiskott–Aldrich syndrome (Burbelo et al., 1995).

In mammalian cells, Cdc42 and Rac regulate morphological responses such as membrane ruffling and the formation of filopodia in fibroblasts in response to numerous mitogenic stimuli (for a review, see Hall, 1994). It is now generally believed that at least some of these effects are mediated through kinases of the Ste20p/p65PAK family and the conserved c‐Jun N‐terminal kinase/stress‐activated protein kinase (JNK/SAPK) pathway (Bagrodia et al., 1995; Polverino et al., 1995; Pombo et al., 1995; Brown et al., 1996; Frost et al., 1996). Isoforms of p65PAK can be activated not only by small GTPases but also by heterotrimeric G‐proteins in various signaling pathways (Knaus et al., 1995; Teo et al., 1995).

Yeast Cdc42p (Johnson and Pringle, 1990), its GDP–GTP exchange factor Cdc24p (Zheng et al., 1994) and the SH3 domain protein Bem1p (Chenevert et al., 1992), which associates with Cdc24p (Peterson et al., 1994) and Ste20p (Leeuw et al., 1995), are involved in controlling cell polarity during budding and in response to pheromone (for a review, see Chant and Stowers, 1995). Recent evidence suggests that Cdc42p and Cdc24p are also involved in pheromone signaling, and it has been proposed that the Cdc42p–GTPase cycle might play a role in linking Ste20p to the heterotrimeric mating response G‐protein (Simon et al., 1995; Zhao et al., 1995). Moreover, Cdc42p has been shown to act upstream of Ste20p in the pathway that regulates filamentous growth (Mösch et al., 1996).

Here, we show in a mutational analysis of Ste20p that the Cdc42p binding domain is dispensable for pheromone signaling. However, the Cdc42p binding domain is needed for cell–cell adhesion during conjugation, for morphological switching in response to a nutritional signal and for an essential function shared by Ste20p and Cla4p during vegetative growth. We also demonstrate that the Cdc42p binding domain is required to direct localization of Ste20p to the actin cytoskeleton in regions of polarized growth during budding. These results do not support the view that Ste20p acts as a mediator of Cdc42p in pheromone signaling, but are consistent with the model that Ste20p represents an essential target of Cdc42p during morphological switching and vegetative growth.

Results

Interaction of Cdc42p with Ste20p

Ste20p has a tripartite structure with a large non‐catalytic region in the amino‐terminal half, a kinase domain in the carboxy‐terminal half and a short non‐catalytic sequence carboxy‐terminal to the kinase domain (Leberer et al., 1992a). We used the two‐hybrid system and an in vitro binding assay to confirm that Cdc42p binding to Ste20p is restricted to a short domain within the amino‐terminal half. This domain is conserved among the members of the Ste20p protein kinase family and has been shown biochemically to bind Cdc42p in both p65PAK and Ste20p (Manser et al., 1994; Simon et al., 1995; Zhao et al., 1995).

As summarized in Figure 1A, two‐hybrid interactions of either wild‐type Cdc42p or its activated G12V mutant version were only found with the non‐catalytic, amino‐terminal half of Ste20p, and these interactions were abolished by deleting a stretch of 36 amino acids between amino acid residues 333 and 370, corresponding to the conserved Cdc42p binding domain first identified in p65PAK(Manser et al., 1994). We did not observe interactions with full‐length Ste20p, corroborating our previous observations that protein fragments might reveal two‐hybrid interactions not seen between full‐length proteins (Leeuw et al., 1995; Whiteway et al., 1995).

Figure 1.Figure 1.
Figure 1.

Interactions of Ste20p with Cdc42p. (A) Two‐hybrid interactions with wild‐type Cdc42p and its activated G12V mutant version. Interactions were measured by using fusions with the DNA binding domain (DBD) of LexA and the transcriptional activation domain (AD) of Gal4p. Data (Miller units) are means ± SD of three independent experiments. Growth on ‐His medium was evaluated as (−) no growth, (+) normal growth. The Cdc42p binding domain (CBD) of Ste20p is indicated by the hatched bar. The plasmids were pDH37 (Ste20p1−939), pRL27 (Ste20p1−497), pRL7 (Ste20p497−939), pAF3 (Ste20p1−435), pAF2 (Ste20p1−298), pRL99 (Ste20p1−435/Δ334−369), pRL39 (Cdc42p wild‐type) and pRL58 (Cdc42pG12V). All plasmids were also tested in combination with lamin fusion constructs as controls. These transformants did not grow on ‐His medium and had β‐galactosidase activities of 0.05–0.11 units. (B) Resin binding assay to determine interactions between Ste20p and Cdc42p. GST–Ste20pΔ334−369, GST–Ste20p and GST were bound to glutathione–Sepharose beads and incubated with GTPγS‐loaded Cdc42p fused to maltose binding protein (MBP). After washing, the beads were subjected to SDS–PAGE in 7.0% (upper panel) and 10.0% (lower panel) gels and immunoblotted for the presence of Ste20p–GST (upper panel) and Cdc42p–MBP (lower panel) with antibodies to Ste20p and MBP, respectively. GTPγS‐loaded Cdc42p bound to wild‐type Ste20p but not to its mutant version. No binding was observed with GDPβS‐loaded Cdc42p (data not shown).

We then used a resin binding assay to demonstrate that a Ste20p mutant in which the Cdc42p binding domain had been deleted was unable to bind Cdc42p. Cdc42p fused to maltose binding protein (MBP) was affinity purified from Escherichia coli, activated with GTPγS and analyzed for its ability to bind to fusions of wild‐type Ste20p and the Ste20pΔ334−369 mutant with glutathione S–transferase (GST) immobilized on glutathione–Sepharose beads. The results indicated that MBP–Cdc42p bound to wild‐type Ste20p but not to its mutant version (Figure 1B).

Role of the Cdc42p binding domain of Ste20p during mating

To characterize the functional role of non‐catalytic sequences in the amino‐terminal half of Ste20p, we constructed several amino‐terminal truncation mutants which were expressed under the control of the galactose‐inducible GAL1 promoter (Figure 2A). Cells expressing truncations of Ste20p up to amino acid residue 495 (ΔN221, ΔN355 and ΔN494) were not affected by their ability to mate with a STE20 wild‐type strain (Figure 2A). The induction of a pheromone‐inducible reporter gene, FUS1::lacZ, and the pheromone‐induced formation of mating‐specific cell shapes, as indicated by the formation of ‘shmoo’ morphologies, were also normal (Figure 2A). Cells expressing a truncation of Ste20p up to residue 610 (ΔN609) were reduced in their mating efficiencies (Figure 2A). These cells were also reduced in FUS1::lacZ induction but produced typical mating‐specific morphologies at normal frequencies in response to pheromone (Figure 2A). Western blot analysis indicated that these cells produced less mutant protein when compared with cells producing wild‐type Ste20p or the ΔN494 mutant (data not shown). Thus, it is conceivable that decreases in mating efficiencies and FUS1::lacZ induction in these cells are caused by reduced mutant protein levels.

Figure 2.

Mating functions of Ste20p deletion mutants. (A) Wild‐type Ste20p and amino‐terminal truncation mutants were placed under control of the GAL1 promoter and transformed into ste20‐deleted cells (YEL206 to measure mating efficiencies, or supersensitive sst1 YEL120 to measure shmoo formation and FUS1::lacZ induction). Cells were grown in selective (2%) raffinose medium to mid‐exponential phase. Mating efficiencies were then determined in galactose (4%) medium at 30°C with the tester strain A281‐4C (mean values ± SD of three independent experiments). To examine pheromone‐induced shmoo formation, cells were switched to galactose (4%) medium containing 0.1 μM α‐factor and grown for 6 h at 30°C. Cells were sonicated briefly and observed microscopically to determine cell shape. ‘Shmoo’ formation is given as percentage of cells producing one or more projections (means ± SD, n = 3; for each experiment, 200 cells were counted). To analyze FUS1::lacZ induction, cells were transformed with the reporter plasmid pSB234 and grown in selective raffinose (2%) medium to mid‐exponential phase. After addition of 4% galactose, cells were incubated for 2 h at 30°C. Half of the cells were then treated with 0.1 μM α‐factor for 6 h at 30°C and β‐galactosidase activities were measured (expressed in Miller units). (B) Internal Ste20p deletion mutants were integrated into the genome. Mating efficiencies (strains YEL276 and YEL285 transformed with plasmid pRS313) were determined with the tester strain A‐281‐4C (mean values ± SD of three independent experiments). Pheromone‐induced projection formation of the supersensitive strains YEL286 and YEL288 was examined microscopically as described above after 6 h of growth in YEPD medium containing 0.1 μM α‐factor. Values are given as percentage of cells producing one or more projections (means ± SD, n = 3; for each experiment, 200 cells were counted). To analyze FUS1::lacZ induction, cells were transformed with the reporter plasmid pSB234 and grown in selective raffinose (2%) medium to mid‐exponential phase. Half of the cells were then treated with 0.1 μM α‐factor for 6 h at 30°C and β‐galactosidase activity was measured (expressed in Miller units). In cells deleted for STE20, the mating efficiency was <0.00001%, basal and induced FUS1::lacZ expressions were <0.1 and projection formation was 0.

Basal levels of FUS1::lacZ expression were slightly increased in cells overexpressing the ΔN494 and ΔN609 truncation mutants (Figure 2A). These elevated levels were not observed in ste4‐deleted cells and hence were dependent on the function of Gβ (data not shown). Moreover, the truncation mutants failed to suppress defects in mating and FUS1::lacZ induction in ste4‐deleted cells (data not shown).

To study further the function of the Cdc42p binding domain, we used homologous recombination to replace the wild‐type STE20 gene with mutant alleles in which the sequences between amino acid residues 333 and 370 (Δ334–369) or 257 and 583 (Δ258–582) were deleted, respectively (Figure 2B). As shown by immunoblot analyses (Figure 3A), the mutant proteins were expressed at wild‐type levels. We investigated the in vitro protein kinase activity of the Δ334–369 mutant protein immunopurified from yeast cells. Autophosphorylation and the ability to phosphorylate myelin basic protein were unaffected by the deletion (Figure 3B).

Figure 3.

Biochemical characterization of Ste20p deletion mutants. (A) Immunoblot analysis of extracts prepared from STE20 wild‐type (lane 1; strain W303‐1A), ste20Δ258−582 (lane 2; strain YEL285) and ste20Δ334−369 (lane 3; strain YEL276) cells. (B) In vitro protein kinase activities of Ste20p in immune complexes isolated from STE20 wild‐type (lane 1) and ste20Δ334−369 mutant (lane 2) cells. The immune complexes were analyzed for autophosphorylation (upper panel), myelin basic protein (MBP) phosphorylation (middle panel) and immunodetectable Ste20p protein (lower panel). The autoradiographs and immunoblots were analyzed densitometrically with a laser scanner. The relative phosphorylation levels were calculated in relation to the relative protein amounts and are shown on the bottom of the autoradiographs.

Cells carrying the Δ334–369 and Δ258–582 alleles, respectively, mated at wild‐type levels with STE20 wild‐type tester cells (Figure 2B). As indicated by the formation of clear zones in α‐factor halo assays, these cells responded to pheromone with growth arrest at the same pheromone levels as wild‐type cells (Figure 4). Moreover, pheromone concentration‐dependent FUS1::lacZ induction of mutant cells was indistinguishable from wild‐type cells (Figure 5).

Figure 4.

Growth arrest assays of (A) and (C) STE20 wild‐type (supersensitive sst1 strain YEL106) and (B) and (D) ste20Δ258−582 mutant (supersensitive sst1 strain YEL288) cells. Exponentially growing cells (106) were embedded into molten agar medium, and 50, 5 or 2 (A and B; counterclockwise from top) and 0.5, 0.05 or 0.005 μg (C and D; counterclockwise from top) of α‐factor (solubilized in 5 μl of 90% methanol) were added onto the surface of the plates. Photographs were taken after 1 day of incubation at 30°C. Identical results were obtained with cells carrying the ste20Δ334−369 allele (strain YEL286; data not shown).

Figure 5.

Pheromone concentration dependence of FUS1::lacZ induction. The reporter plasmid pSB234 (Trueheart et al., 1987) was introduced into isogenic supersensitive (sst1) STE20 wild‐type (strain YEL106), ste20Δ334−369 mutant (strain YEL286) and ste20Δ258−582 mutant (strain YEL288) cells. Transformants were grown in selective medium at 30°C to mid‐exponential phase and then incubated for 6 h at 30°C with the indicated α‐factor concentrations. β‐Galactosidase activities were then measured and are given as mean values ± SD of three independent experiments.

The mating efficiencies of cells carrying the Δ334–369 and Δ258–582 alleles, respectively, were reduced when the mating partner of the opposite mating type also carried the same allele (Table I). This bilateral defect is similar to far1 mutant cells where mating is only attenuated slightly as long as one mating partner carries the FAR1 wild‐type gene but is reduced drastically when both mating partners carry far1 mutations (Valtz et al., 1995). This mating defect of far1 mutant cells has been ascribed to a defect in oriented projection formation toward the mating partner (Dorer et al., 1995; Valtz et al., 1995).

View this table:
Table 1. Mating and zygote formation defects of ste20 mutant cells

Indeed, we found that cells carrying either the Δ334–369 or Δ258–582 alleles of STE20 were defective in mating with far1‐c mutant cells (Table I). We therefore investigated the possibility that cells carrying the ste20 mutant alleles, like far1‐c mutant cells, might be defective in oriented morphogenesis. We analyzed the ste20 mutant cells in a direct microscope assay for orientation in a spatial gradient of α‐factor (Segall, 1993). As illustrated in Figure 6, cells carrying the Δ334–369 allele were able to produce projections growing in the direction of the tip of a pheromone‐releasing micropipet. The degree of orientation was quantified by determining the cosine of the angle between the direction of projection formation and the direction of the micropipet (Segall, 1993). Perfect orientation toward the pipet is indicated by a cosine of 1.0, random orientation by a cosine of 0 and perfect orientation away from the pipet by a cosine of −1.0. The STE20 wild‐type cells were found to orient with a cosine of 0.73 ± 0.029 (n = 12), whereas the ste20 mutant cells oriented with a value of 0.73 ± 0.036 (n = 23). Similar values were obtained for cells carrying the ste20Δ258−582 allele. These results indicate a high degree of orientation for both STE20 wild‐type and ste20 mutant cells and exclude the possibility that mating defects between either ste20 mutant cells or ste20 and far1 mutant cells are caused by defects in oriented morphogenesis.

Figure 6.

Responses of (A) STE20 wild‐type (strain YEL106) and (B) ste20Δ334−369 mutant (strainYEL286) cells in a spatial gradient of α‐factor. Exponentially growing cells were analyzed with 67 nM α‐factor in the pipet at 30°C. Images were taken 5 h after induction with pheromone. The out‐of‐focus micropipet is coming from the upper right corner and is directed to the center of the image.

The microscopic inspection of mating assay mixtures revealed the presence of a high percentage of zygotes when at least one mating partner carried wild‐type genes of both STE20 and FAR1 (Table I). However, only few zygotes were found when mating was performed between two ste20 mutant strains or between ste20 and far1 mutant strains (Table I). The percentage of cells producing projections was high in these mating mixtures (Table I). However, these projection‐forming cells were not attached to each other. These results suggest that ste20 mutant cells were able to respond morphologically to cells of opposite mating type but were defective in cell–cell adhesion and, therefore, unable to form zygotes when the opposite mating partner carried mutations in either STE20 or FAR1.

Perturbation of the actin cytoskeleton by overexpression of amino‐terminally truncated Ste20p mutant proteins

Galactose‐induced overexpression of an amino‐terminal truncation allele of STE20 has been reported to be deleterious to growth (Ramer and Davis, 1993). We found that cells overexpressing the ΔN221 or ΔN355 alleles were viable (Figure 7A). In contrast, prolonged galactose‐induced overexpression of either the ΔN494 or ΔN609 alleles was lethal (Figure 7A). This result was also obtained in cells deleted for either STE4, STE5, FAR1 or STE12, or in diploid cells (data not shown), suggesting that lethality was not dependent on the pheromone response pathway.

Figure 7.Figure 7.
Figure 7.

Galactose‐induced overexpression of amino‐terminally truncated Ste20p mutants. (A) Effect of truncations on cellular viability and on the ability to complement ste20 cla4 mutations. Cells deleted for both STE20 and CLA4 (strain YEL257‐1A‐1) contained plasmid pRL21 carrying URA3 for selection and CLA4 to support viability. These cells were transformed with pRS313GAL plasmids carrying HIS3 as selectable marker and either no insert, wild‐type STE20 or the indicated truncation alleles under control of the GAL1 promoter. Patches of transformants were grown on ‐His medium containing 2% glucose (left panel). Patches were then replica plated onto ‐His media containing either 4% galactose (middle panel) or 4% galactose and 0.1% 5‐fluoroorotic acid (5‐FOA; right panel), grown overnight at 30°C, replica plated again on the same media and grown for another day. Cells were forced by 5‐FOA to lose the CLA4 plasmid and were therefore dependent on the function of the STE20 alleles to support viability. Galactose‐induced expression of the ΔN494 and ΔN609 alleles was detrimental for viability. This effect was also seen in other strains including those that were wild‐type for STE20 or CLA4 (data not shown). Expression of the ΔN355 allele was unable to support viability in the absence of CLA4. (B) Effect of overexpression of the ΔN494 allele of STE20 on the actin cytoskeleton. Cells deleted for STE20 (strain YEL206) were transformed with pRS313GAL plasmids carrying either the ste20ΔN494 allele or wild‐type (wt) STE20 under control of the GAL1 promoter. Transformants were grown in selective raffinose (2%) medium to mid‐exponential phase at 30°C. After addition of 4% galactose, cells were grown for 4 h in the absence or presence of 1 μM α‐factor. Fixed cells were then stained with rhodamine–phalloidin to visualize actin. Photomicrographs of representative cells were then taken by Nomarski optics and rhodamine fluorescence with a 100× objective (bar = 5 μm). Similar effects on actin were seen in cells expressing the ΔN609 allele of STE20 (data not shown).

The morphology of cells overexpressing the ΔN494 or ΔN609 alleles was enlarged and rounded (Figure 7B). Cell ghosts, indicative of cell lysis, could be observed after prolonged expression (data not shown). Since we previously found that a fraction of Ste20p co‐purified with actin (Leeuw et al., 1995), we investigated whether the truncation mutants might interfere with the reorganization of the actin cytoskeleton during budding. Cells overproducing wild‐type Ste20p showed normal actin staining with rhodamine–phalloidin (Figure 7B). In proliferating cells, actin was polarized toward the developing bud, and, in pheromone‐induced cells, actin was polarized toward the ‘shmoo’ tip. Cells overproducing the ΔN494 allele revealed an abnormal actin staining pattern with small patches distributed all over the cell (Figure 7B). However, pheromone treatment of these cells induced normal orientation of actin toward the ‘shmoo’ tips (Figure 7B). We conclude from these experiments that overexpression of truncated Ste20p mutants perturbs the polarized reorganization of the actin cytoskeleton during budding but not during the formation of mating‐specific morphologies.

The Cdc42p binding domain of Ste20p is necessary for cellular viability in the absence of Cla4p

Ste20p and Cla4p share a redundant function that is essential for cellular viability (Cvrckova et al., 1995). Cells deleted for CLA4 alone are viable but have defects in cytokinesis (Cvrckova et al., 1995). Deletions of both CLA4 and STE20 cause lethality (Cvrckova et al., 1995). To examine whether the non‐catalytic sequences of Ste20p, including the Cdc42p binding domain, are necessary for vegetative growth in the absence of Cla4p, we analyzed the various Ste20p mutants in a strain that was deleted for both CLA4 and STE20.

Plasmids carrying the various amino‐terminal ste20 truncation mutants under control of the GAL1 promoter were transformed into a cla4Δ ste20Δ strain whose viability was supported by a plasmid carrying CLA4 and URA3 as selectable marker. Expression of the ste20 mutants was then induced by galactose, and loss of the CLA4 plasmid was enforced by the presence of 5‐fluoroorotic acid (Boeke et al., 1987) (Figure 7A). Cells expressing the ΔN221 allele of STE20 were fully viable, whereas cells expressing the ΔN355, ΔN494 or ΔN609 alleles were unable to grow. A microscopic analysis of these cells showed a pleiotropic population of large and irregularly shaped cells and revealed many cell ghosts (data not shown). Thus, the amino‐terminal region up to amino acid 222 was not essential for viability in the absence of Cla4p. However, the ability to support viability in the absence of Cla4p was lost after partial removal of the Cdc42p binding domain in the ΔN355 mutant.

To investigate further the role of the Cdc42p binding domain of Ste20p in vegetative growth, we constructed cla4Δ strains that were either deleted for genomic STE20 or in which genomic STE20 was replaced with the Δ334–369 allele by homologous recombination. Viability of the double mutant cells was supported by galactose‐induced expression of CLA4 carried on a plasmid. CLA4 expression was then suppressed by replica plating patches of yeast cells from galactose medium to glucose medium. As illustrated in Figure 8, Cla4p‐depleted cells were viable when they carried the wild‐type STE20 gene but were unable to grow when they were completely deleted for STE20 or when STE20 was replaced with the Δ334–369 allele. The microscopic inspection revealed a heterogeneous population of large and aberrantly shaped cla4Δ ste20Δ334−369 cells that were similar to cla4Δ ste20Δ cells. These results indicate that the Cdc42p binding domain of Ste20p plays an essential role in vegetative growth in the absence of Cla4p.

Figure 8.

Requirement of the Cdc42p binding site of Ste20p for cellular viability in the absence of Cla4p. Cells that were deleted for CLA4 and carried either wild‐type STE20(strain YEL257‐12C), a deletion of STE20 (strain YEL257‐1A‐2) or the ste20Δ334−369 allele (strain YEL303‐1B), contained plasmid pDH129 carrying CLA4 under control of the GAL1 promoter. Patches of cells were grown on selective galactose (4%) medium overnight at 30°C, then replica plated to selective glucose (2%) medium and grown for 24 h at 30°C. Viability of cla4Δ ste20Δ or cla4Δ ste20Δ334−369 double mutant cells was supported by galactose‐induced expression of CLA4 but was prevented by glucose‐dependent suppression of CLA4.

Essential role of the Cdc42p binding domain of Ste20p in pseudomycelial growth

Ste20p is required for the induction of filamentous growth in response to nitrogen starvation (Liu et al., 1993). To determine whether the non‐catalytic region of Ste20p is necessary for this response, we transformed diploid cells deleted for both alleles of STE20 with plasmids expressing the ste20Δ334−369 and ste20Δ258−582 alleles and induced the transformants by nitrogen starvation. As shown in Figure 9, transformants with the mutant alleles were unable to form pseudohyphae, although transformants expressing wild‐type STE20 were able to do so. These findings suggest that the Cdc42p binding domain of Ste20p is necessary for mediating the starvation signal.

Figure 9.

Requirement of the Cdc42p binding site of Ste20p for pseudomycelial growth. The diploid ste20/ste20 strain HLY492 was transformed with plasmids pSTE20‐5 carrying wild‐type STE20 (A), pRS316 carrying no insert (B), pRS316‐STE20(Δ334–369) carrying the ste20Δ334−369 allele (C) and pRS316‐STE20(Δ258–582) carrying the ste20Δ258−582 allele (D), and colonies were grown on selective nitrogen starvation medium (Gimeno et al., 1992) for 5 days at 30°C. Photomicrographs were taken with a 4× objective (bar = 1 mm).

The Cdc42p binding domain is required for subcellular localization of Ste20p

Finally, we asked whether the Cdc42p binding site is necessary for the subcellular localization of Ste20p. The sequence encoding green fluorescent protein (GFP) (Chalfie et al., 1994) was fused to the amino‐terminal sequence of STE20 and the ste20Δ334−369 allele. The chimeric proteins were fully functional, as shown by complementation of the mating defect of cells deleted for STE20 (data not shown). Examination of cells deleted for STE20 but expressing the GFP–Ste20p wild‐type fusion protein under the STE20 promoter on a centromere‐based plasmid showed that the fusion protein localized at low concentrations to the plasma membrane and to cortical actin patches, and was concentrated in the growing bud (Figure 10b). This asymmetric distribution pattern was similar to the rhodamine–phalloidin staining pattern of actin (Figure 10c). GFP alone was found to be distributed uniformly in the cell (data not shown).

Figure 10.

Subcellular localization of GFP–Ste20p. Cells deleted for STE20 (strain YEL206) were transformed with plasmids pRL116 and pBTL72 carrying fusions of GFP with wild‐type STE20 (af) and the ste20Δ334−369 (ste20‐8) allele (gl), respectively. Transformants were grown in selective medium at room temperature to mid‐exponential phase (a–c and g–i), and half of the cultures were incubated for 90 min with 1 μM α‐factor at room temperature (d–f and j–l). Fixed cells were then stained with rhodamine–phalloidin, viewed by Nomarski optics (a, d, g and j) and analyzed for GFP fluorescence to visualize GFP–Ste20p chimeras (b, e, h and k) and rhodamine fluorescence to visualize actin (c, f, i and l). Identical distributions of GFP–Ste20p chimeras were observed in viable cells (data not shown). Photomicrographs were taken with a 100× objective (bar = 5 μm) and are representative for a large number of cells investigated.

In pheromone‐induced cells, the fusion protein was concentrated intracellularly in the ‘shmoo’ tip, giving rise to a crescent‐like staining pattern (Figure 10e). Again, this asymmetric distribution pattern was similar to the rhodamine–phalloidin staining pattern of actin (Figure 10c and f).

Cells expressing the GFP–Ste20pΔ334−369 fusion protein showed a normal pattern of actin staining during both budding and ‘shmooing’ (Figure 10i and l). However, the GFP–Ste20pΔ334−369 fusion protein showed a uniform intracellular distribution in the budding cells (Figure 10h). In pheromone‐induced cells, the mutant fusion protein was also distributed all over the cell but was found at slightly higher concentrations in the ‘shmoo’ tip (Figure 10k). Based on these observations, we conclude that the Cdc42p binding domain plays an important role in directing subcellular localization of Ste20p to sites of polarized growth.

Discussion

The Cdc42p binding domain of Ste20p is required for interaction with Cdc42p but not for in vitro kinase activity

Ste20p contains within its amino‐terminal non‐catalytic half, between amino acid residues 333 and 370, a sequence which is homologous to the Cdc42p binding sites of mammalian p65PAK or p120ACK (Manser et al., 1993, 1994). The sequence LRISTPYNAKHIHHVGVD from amino acid residues 335–352 corresponds to a sequence motif known to be sufficient to bind Cdc42 or Rac in a variety of Cdc42/Rac binding proteins (Burbelo et al., 1995).

Consistent with the view that the Cdc42p binding domain may be the only region of Ste20p to bind Cdc42p, mutational removal of the sequence between residues 333 and 370 prevented the interaction with Cdc42p in vivo and in vitro, as shown by the two‐hybrid interaction system and by a resin binding assay, respectively (Figure 1). Although these results do not completely exclude the possibility that Cdc42p might bind under in vivo conditions to other regions of Ste20p not detected by these methods, the defects of the Ste20p truncation mutants in morphological switching and cytokinesis strongly argue, as discussed below, for the view that these mutants have lost the ability to bind Cdc42p.

Removal of the Cdc42p binding site blocked neither the ability of Ste20p to autophosphorylate nor to phosphorylate a generic substrate, myelin basic protein, in an in vitro kinase assay (Figure 3). We found, in contrast to a previous study (Simon et al., 1995), that the in vitro kinase activity of Ste20p was not activated by addition of GTPγS‐Cdc42p (data not shown). This failure of activation was also observed for the Ste20p mutant proteins, although we found, in agreement with previous studies (Manser et al., 1994; Martin et al., 1995) that a GST–p65PAK fusion protein could be activated under the same conditions (data not shown). One possible explanation for the discrepancy with the results from Simon et al. (1995) might be that these authors purified recombinant GST–Ste20p from insect cells, whereas in our case the native protein was isolated from yeast cells and might have been already in an activated state.

Cdc42p binding to Ste20p is dispensable for pheromone signaling

Nearly all of the amino‐terminal non‐catalytic half of Ste20p (up to amino acid residue 583), including the Cdc42p binding site, could be removed without affecting mating‐specific responses (Figures 2 and 4, 5, 6). Cdc42p binding‐deficient Ste20p mutants were normally regulated in a pheromone‐dependent manner in vivo. These findings contradict the view that Cdc42p binding to Ste20p provides a functional link between the Ste20p/MAP kinase module and the heterotrimeric G‐protein (Simon et al., 1995; Zhao et al., 1995), and argue strongly for a direct regulation of Ste20p by the activated β‐ and γ‐subunits of the mating response G‐protein (Figure 11). This view is also supported by genetic interactions between STE20 and STE4 (encoding Gβ) (Leberer et al., 1992a) and our recent observation that Gβ directly binds to Ste20p in vivo and in vitro (T.Leeuw and E.Leberer, unpublished results). Our results also imply that molecular determinants which specify the pheromone signaling function of Ste20p must lie within either the kinase domain or the non‐catalytic sequence carboxy‐terminal to the kinase domain. This conclusion is supported by findings that sequences carboxy‐terminal to the kinase domain are essential for the mating function of Ste20p (E.Leberer, unpublished results). Activation of Ste20p may also involve the adaptor protein Ste5p (Figure 11) which has been shown to interact with the kinases of the MAP kinase module (for a review, see Elion, 1995), to co‐immunoprecipitate with Ste20p (Leeuw et al., 1995) and to associate directly with Gβ (Whiteway et al., 1995).

Figure 11.

Proposed model for multiple mechanisms of Ste20p regulation. Cdc42p binding to Ste20p is not required for activation of the pheromone‐responsive MAP kinase pathway, but is essential for the induction of pseudohyphal growth and the control of polarized growth during budding. We propose that Ste20p is activated by Gβγ during pheromone signaling by a mechanism that might involve the adaptor protein Ste5p which binds to Gβ and the protein kinases acting downstream of Ste20p (Elion, 1995; Whiteway et al., 1995). We also propose that Cdc42p is involved in pheromone signaling and mating through mechanisms that do not require binding to Ste20p. Our results suggest that Cdc42p binding to Ste20p plays a role in cell–cell adhesion during conjugation. In these pheromone‐induced processes, Cdc42p is likely to be activated through its GDP–GTP exchange factor Cdc24p which has been shown to interact with Gβ (Zhao et al., 1995). Our model does not exclude the involvement of as yet unidentified components.

Cdc42p plays a role during mating. It is localized at both the bud site and the tips of mating projections and is required for the formation of these projections during mating (Johnson and Pringle, 1990; Ziman et al., 1993). This function of Cdc42p is most likely under control of the GDP–GTP exchange factor Cdc24p which binds to the β‐subunit of the mating response G‐protein (Zhao et al., 1995). Temperature‐sensitive mutations in CDC42 and CDC24 affect pheromone signaling (Simon et al., 1995; Zhao et al., 1995), and both overexpression of a constitutively activated form of Cdc42p (Simon et al., 1995; Zhao et al., 1995; Akada et al., 1996) and deletion of RGA1 encoding a putative GTPase‐activating protein for Cdc42p (Stevenson et al., 1995) weakly activate the pheromone response pathway. However, the data presented here suggest that any effects of Cdc42p on the pheromone signaling pathway and on pheromone‐induced morphological changes are mediated by mechanisms that do not involve a physical interaction between Cdc42p and Ste20p (Figure 11). This conclusion is corroborated by the finding that overexpression of Cdc42p can improve mating in cells deleted for STE20 (Leberer et al., 1996).

Potential role of Cdc42p binding to Ste20p in cell–cell adhesion

A surprising result of this study was that Ste20p and Cdc42p might play roles in the initial phase of zygote formation when cells attach to each other (Figure 11). Although the formation of zygotes occurred at a normal frequency as long as one mating partner expressed wild‐type Ste20p, zygote formation was reduced when both mating partners expressed Cdc42p binding‐deficient Ste20p mutants (Table I). This bilateral mating defect is reminiscent of cells carrying mutations in the FUS1, FUS2 or FAR1 genes, where mating is nearly normal when at least one mating partner carries the corresponding wild‐type gene, but is drastically attenuated when both mating partners carry mutations in these genes (McCaffrey et al., 1987; Trueheart et al., 1987; Dorer et al., 1995; Elion et al., 1995; Valtz et al., 1995). However, in contrast to Cdc42p binding‐deficient Ste20p mutant cells, fus1 and fus2 mutant cells are able to attach to each other but are delayed in cellular fusion (McCaffrey et al., 1987; Trueheart et al., 1987; Elion et al., 1995). Interestingly, we also observed reduced mating and defective zygote formation when one mating partner expressed a Cdc42p binding‐deficient Ste20p mutant and the opposite mating partner carried a far1‐c mutant allele (Table I).

Cell–cell adhesion during zygote formation involves the combined effects of cell surface agglutinins (Lipke and Kurjan, 1992), orientation of a mating protrusion toward a neighboring cell with opposite mating type (Segall, 1993), a partner selection system (Jackson et al., 1991) and the deposition of proteins required for the fusion of cells in the tips of the mating protrusions (McCaffrey et al., 1987; Trueheart et al., 1987; Elion et al., 1995). The reduced mating efficiencies of Cdc42p binding‐deficient Ste20p mutant cells could be explained if both mating partners are defective in orienting mating‐specific protrusions toward the mating partner, a defect which has been ascribed to far1‐c mutant cells (Dorer et al., 1995; Valtz et al., 1995). However, Cdc42p binding‐deficient Ste20p mutant cells were able to respond morphologically to the mating partner by attaining a typical ‘shmoo’ morphology (Table I) and orienting mating protrusions toward the pheromone source (Figure 6).

A more likely interpretation assumes that Cdc42p binding to Ste20p plays a role in polarized transport of components required for cell–cell adhesion in the tips of mating projections. It is tempting to speculate that Cdc42p binding to Ste20p might provide a link between the pheromone signaling pathway and the Pkc1p/Mpk1p pathway which is stimulated in response to pheromone through Ste20p (Zarzov et al., 1996). It has been speculated that this stimulation might be necessary for polarized cell wall synthesis during the formation of mating protrusions (Zarzov et al., 1996). Interestingly, in mammalian cells, the Rho family members Cdc42, Rac1 and RhoA induce the formation of focal complexes, filopodia and lamellipodia which might be involved in cell attachment and hence functionally equivalent to mating protrusions (Kozma et al., 1995; Nobes and Hall, 1995). Moreover, the Ste20p homolog of Drosophila has been shown recently to localize to focal adhesions in epidermal cells (Harden et al., 1996). More research will be required to study further the roles of Ste20p and Cdc42p in cell–cell adhesion and zygote formation during conjugation.

Cdc42p binding to Ste20p is required to induce filamentous growth

Removal of the Cdc42p binding domain of Ste20p completely prevented the induction of pseudomycelial growth of diploid cells in response to nitrogen starvation (Figure 9). Moreover, haploid cells expressing a Cdc42p binding‐deficient Ste20p mutant have been shown to be blocked in invasive growth (M.Peter, personal communication). These results are consistent with the recent demonstration that Cdc42p acts as a regulator of the pseudohyphal Ste20p/MAP kinase pathway which does not require the function of the heterotrimeric G‐protein but involves the Ras homolog Ras2p (Mösch et al., 1996). Thus, differential regulation of Ste20p by Gβγ in response to pheromone and by GTP‐Cdc42p in response to a nutritional signal might explain how the same protein kinase can regulate different developmental pathways within the same cell (Figure 11).

The Cdc42p binding domain of Ste20p is indispensable for vegetative growth in the absence of Cla4p

Ste20p and its isoform Cla4p share an as yet undefined essential role in vegetative growth (Cvrckova et al., 1995). Both protein kinases are necessary for cytokinesis, proper formation of the septin ring and proper localization of cell surface growth with respect to the septin ring (Cvrckova et al., 1995). Deletion of both genes, CLA4 and STE20, causes lethality (Cvrckova et al., 1995). This phenotype was not rescued by Cdc42p binding‐deficient Ste20p mutants, indicating that the essential function of Ste20p requires Cdc42p binding (Figures 7A and 8). This result strongly suggests that Ste20p and Cla4p are essential targets of Cdc42p during vegetative growth (Figure 11). A role for Ste20p downstream of Cdc42p during budding is also suggested by the finding that multiple copies of STE20 cure the growth defect of temperature‐sensitive cdc42 mutant cells at the restrictive temperature (Leberer et al., 1996).

The Cdc42p binding domain is needed for proper intracellular localization of Ste20p

We have used fusions with GFP to determine the subcellular localization of Ste20p (Figure 10). As will be described elsewhere in more detail, GFP–Ste20p showed a cell cycle‐dependent intracellular distribution. In G1 of the cell cycle or in late mitosis, GFP–Ste20p localized to the plasma membrane and to cortical actin patches. In cells undergoing budding, GFP–Ste20p was accumulated in the growing bud together with cortical actin bundles (Figure 10b and c) and at the mother‐bud neck in late mitosis (data not shown). This asymmetric distribution was lost after removal of the Cdc42p binding site although the polarized actin staining pattern remained unaffected (Figure 10h and i).

In pheromone‐induced cells, GFP–Ste20p was concentrated with cortical actin bundles at the tips of mating protrusions, giving rise to a crescent‐like staining pattern (Figure 10e and f). This kind of staining pattern has also been observed for Cdc42p (Ziman et al., 1993). After removal of the Cdc42p binding site, Ste20p was distributed all over the cell but was still found at slightly higher concentrations in the projection tips where cortical actin bundles were oriented normally (Figure 10k and l). These results argue strongly for a role for Cdc42p in localizing Ste20p to regions of polarized growth where cortical actin bundles reorient, and are in agreement with our previous findings that a fraction of Ste20p co‐purified with actin in sucrose density gradients (Leeuw et al., 1995). Interestingly, Cdc42 binding to WASP, a protein implicated in the immunodeficiency disorder Wiskott–Aldrich syndrome, has been shown to direct localization of WASP to clusters of polymerized actin (Symons et al., 1996).

Overexpression of truncated Ste20p mutants

Overexpression of the ΔN494 and ΔN609 alleles of STE20 was deleterious to growth (Figure 7A) probably by preventing the normal reorientation of cortical actin into the emerging bud (Figure 7B). This lethality could be explained if the truncation mutants interfere with the functions of Ste20p and Cla4p during bud emergence, e.g. by preventing phosphorylation of a common substrate that regulates actin reorganization. Alternatively, the truncation mutants might be hyperactive and therefore phosphorylate substrates in an uncoordinated mode. Interestingly, destruction of the Cdc42p binding site was not sufficient to generate this lethal effect since overexpression of the ΔN355 (Figure 7A) and Δ334–369 (data not shown) mutant proteins, in which the conserved Cdc42p binding motif from amino acid residues 335–352 had been removed, was not detrimental (Figure 7A). Thus, to generate a deleterious truncation mutant, additional removal of sequences carboxy‐terminal to the Cdc42p binding motif was required.

Overexpression of the ΔN494 and ΔN609 truncation mutants stimulated basal FUS1::lacZ expression (Figure 2A). This stimulation required the function of Gβ(encoded by STE4), suggesting that a regulatory input from the heterotrimeric G‐protein is required. Moreover, the truncation mutants did not complement the pheromone signaling defects in ste4 mutant cells (data not shown). Stimulation of basal FUS1::lacZ expression was not observed in cells overexpressing the ΔN355 (Figure 2A) and Δ334–369 (data not shown) mutants, suggesting that additional removal of sequences carboxy‐terminal to the conserved Cdc42p binding motif is required to engender this phenotype. We found that removal of these sequences did not affect the in vitro kinase activity of Ste20p (data not shown). However, stimulation of basal pheromone signaling by the ΔN494 and ΔN609 truncation mutants could be explained if removal of amino‐terminal sequences improves the in vivo stimulation of Ste11p, which has been shown to be a substrate of Ste20p in vitro (Wu et al., 1995).

Multiple mechanisms of Ste20p regulation

Taken together, our results demonstrate that Ste20p is regulated by multiple mechanisms during different developmental pathways (Figure 11). Cdc42p‐dependent mechanisms are involved in morphological switching and the control of polarized growth and cytokinesis during cell proliferation, whereas a Cdc42p‐independent mechanism is involved in G‐protein‐coupled receptor signaling. The high degree of conservation of the Ste20p/p65PAK family protein kinases suggests that similar multiple regulatory mechanisms may exist in other organisms including mammalian cells. It is conceivable that Cdc42/Rac‐dependent and ‐independent mechanisms regulate mammalian p65PAK isoforms in different signaling pathways, and some of these mechanisms may depend on heterotrimeric G‐proteins. Interestingly, some members of the Ste20p/p65PAK family are lacking a Cdc42/Rac binding site, including GC kinase which, like p65PAK, is able to activate the JNK/SAPK pathway (Pombo et al., 1995). Thus the same MAP kinase cascade may be regulated by Cdc42/Rac‐dependent and ‐independent members of the same protein kinase family. An interplay with heterotrimeric and small G‐proteins may provide a sensitive mode for fine tuning specificity of Ste20p/p65PAK protein kinases in various signaling pathways.

Materials and methods

Materials

Restriction endonucleases and DNA‐modifying enzymes were obtained from Boehringer Mannheim, Bethesda Research Laboratories, Pharmacia LKB Biotechnology Inc. and New England Biolabs Inc. Ultima Taqthermostable DNA polymerase was purchased from Perkin Elmer. [γ‐32P]ATP was obtained from ICN. GTPγS and GDPβS were from Boehringer Mannheim. Rhodamine‐labeled phalloidin was obtained from Molecular Probes. Glutathione–Sepharose beads and glutathione were obtained from Pharmacia Biotech Inc. Amylose–agarose beads and antibodies to maltose binding protein were purchased from New England Biolabs Inc. Synthetic α‐factor was purchased from Sigma, dissolved in 90% methanol at a concentration of 10 mg/ml and stored at −20°C. Horseradish peroxidase‐conjugated goat anti‐rabbit IgG was obtained from Bio‐Rad. All other reagents were of the highest purity grade commercially available.

Yeast manipulations

Yeast media, culture conditions and manipulations of yeast strains were as described (Rose et al., 1990). Yeast transformations with circular or linearized plasmid DNA were carried out after treatment of yeast cells with lithium acetate (Rose et al., 1990). Gene disruptions were performed by the one‐step gene replacement method (Rose et al., 1990).

Construction of plasmids

Plasmid pRS316‐STE20 is a centromere E.coli–yeast shuttle vector containing URA3 for selection in S.cerevisiae (Sikorski and Hieter, 1989) and the STE20 gene (Leberer et al., 1992a). Plasmid pVTU‐STE20 is pVTU102‐U carrying STE20 under control of the ADH1 promoter and URA3 as selectable marker (Leberer et al., 1992a).

To construct plasmid pVTU‐STE20(Δ334–369), a fragment encoding the Cdc42p binding domain of Ste20p from amino acids 334 to 369 was removed by site‐directed mutagenesis (Kunkel et al., 1987) using single‐stranded pVTU‐STE20 as template and the oligodeoxynucleotide 5′‐TCCTCATCTTCTATAACCACCACTTCTAGTGGTATTTCCAAAAG‐3′ as primer. Plasmid pRS316‐STE20(Δ334–369) was then constructed by replacing the BamHI–XhoI fragment of plasmid pRS316‐STE20 with the BamHI– XhoI fragment of pVTU‐STE20(Δ334–369). Plasmid pRS316‐STE20(Δ258–582) was obtained by deleting the HindIII fragment from nucleotide positions 771–1746 of STE20 from pRS316‐STE20.

Plasmids carrying full‐length and truncated versions of STE20 under control of the GAL1 promoter were constructed by cloning PCR (Saiki et al., 1988) fragments of STE20 into the SpeI and SacII sites of plasmid pRS313GAL (Leberer et al., 1992a). Plasmid pSTE20‐5 (Leberer et al., 1992a) was used as a template. Ultima Taq thermostable DNA polymerase was used in all PCR reactions. To construct plasmid pDH166 carrying full‐length STE20, the oligodeoxynucleotide primers 5′‐GACTAGTCATGAGCAATGATCCATCTGCT‐3′ (ODH72) and 5′‐TCCCCGCGGCGATAATAAGGTGTACCCTGC‐3′ (ODH76; the newly created SpeI and SacII sites, respectively, are underlined) were used. The ΔN221, ΔN355, ΔN494 and ΔN609 truncation fragments were amplified by using the reverse primer ODH76 and the following forward primers (the newly created SpeI sites are underlined): 5′‐GACTAGTCATGATCAATTCAGCTTCCCATTCG‐3′ (ODH81), 5′‐GACTAGTCATGGGTGAGTACACAGGTTTGCCG‐3′ (ODH73), 5′‐GACTAGTCATGAATTCTGCCGCCAATGTT‐3′ (ODH74), and 5′‐GACTAGTCATGATTTGCTCAGACGGTGACCCA‐3′ (ODH75).

To construct fusions of STE20 with the cDNA encoding the S65T mutant version of GFP (Heim et al., 1995), a BamHI–SpeI fragment of the GFP cDNA was amplified by PCR using the oligodeoxynucleotides 5′‐CGGGATCCGCTAGCATGAGTAAAGGAGAAGAAC‐3′ and 5′‐GGACTAGTTTGTATAGTTCATCCATGCC‐3′ (the newly created BamHI and SpeI sites, respectively, are underlined) as primers, and pRSET‐S65T (kindly provided by R.Y.Tsien) as a template. The fragment was cloned into pRS316GAL (plasmid pRS316 carrying an EcoRI–BamHI fragment of the GAL1 promoter; Leberer et al., 1992b) to create plasmid pBTL41. Plasmid pBTL72 carrying an in‐frame fusion of the GFP–S65T cDNA with the amino‐terminal sequence of full‐length STE20 was then constructed by cloning a SpeI–SacII fragment of STE20 (obtained from plasmid pDH166) into the respective sites of pBTL41. To construct plasmid pBTL73 carrying an in‐frame fusion of the GFP–S65T cDNA with the amino‐terminal sequence of the ste20Δ334−369 mutant allele, a SpeI–SacII fragment of ste20Δ334−369 was amplified by PCR using the oligodeoxynucleotides 5′‐GACTAGTCATGAGCAATGATCCATCTGCT‐3′ and 5′‐TCCCCGCGGTTACCCGGGCTTTTGTTTATCATCTTC‐3′ as primers (the newly created SpeI and SacII sites, respectively, are underlined) and pRS316‐STE20(Δ334–369) as a template, and subsequently subcloned into pBTL41. Plasmids pRL116 and pBTL56, carrying the GFP::STE20 chimeras under control of the STE20 promoter, were then created by replacing the GAL1 promoter in plasmids pBTL72 and pBTL73, respectively, with the STE20 promotor amplified by PCR using the deoxynucleotides 5′‐CCGCTCGAGGAATTCTATCCAGAACCGCCT‐3′ and 5′‐GTTGCTAGCTTCTTGTATTAGTCGAGGATC‐3′ as primers (the newly created XhoI and NheI sites, respectively, are underlined) and plasmid pSTE20‐5 as a template. These plasmids were found to fully complement the mating defect of cells deleted for STE20 (data not shown).

Plasmid pRL21 is the yeast multicopy shuttle vector pRS426 (Christianson et al., 1992) carrying the CLA4 gene in a HindIII–EcoRI fragment (isolated from plasmid pC2537; Cvrckova et al., 1995) and URA3 as selectable marker. Plasmid pDH129 is pRS313GAL carrying CLA4 under control of the GAL1 promoter. To construct this plasmid, a XbaI–SacII fragment of CLA4 was amplified by PCR using the oligodeoxynucleotides 5′‐GCTCTAGAGCATGTCTCTTTCAGCTGCAGCG‐3′ and 5′‐TCCCCGCGGAATAGTTGTGTGCTTCATTCC‐3′ (the newly created XbaI and SacII sites, respectively, are underlined) as primers, and plasmid pC2537 as a template, and subcloned behind the GAL1 promoter into pRS313GAL.

Yeast strains

The S.cerevisiae yeast strains used in this study are listed in Table II. Strains containing the ste20Δ334−369 and ste20Δ258−582 mutant alleles were constructed by a two‐step procedure. The SphI–XhoI fragments of plasmids pRS316‐STE20(Δ334–369) and pRS316‐STE20(Δ258–582) were subcloned into the integrating plasmid pRS306 (Sikorski and Hieter, 1989). The resulting plasmids were then linearized with BamHI, transformed into strain W303‐1A and URA3+ transformants were selected. Subsequently ura3 derivatives were selected by growth in the presence of 5‐fluoroorotic acid (Boeke et al., 1987), and colonies that had become simultaneously ura3 and mating defective with the MATα far1‐c strain JC31‐7D were selected. To ensure that wild‐type STE20 was replaced with the mutant versions, the STE20 region was amplified by PCR using the oligodeoxynucleotides 5′‐GCTGCCCATGAAGCTGGTGGA‐3′ and 5′‐TACTTGGGTCACCGTCTGAGC‐3′ as primers. Wild‐type STE20 produced a 1.7 kb fragment, whereas the mutant versions produced fragments of 1.6 and 1.3 kb in size, respectively.

View this table:
Table 2. Yeast strains

Replacement of the sequence of the CLA4 gene encoding the catalytic domain of Cla4p with TRP1 was achieved as follows. The HindIII–EcoRI fragment of pC2537 (Cvrckova et al., 1995) was subcloned into the Bluescript KS(+) vector (Stratagene) to yield plasmid pDH122. The NsiI fragment from nucleotide positions 1084–2596 of the CLA4 gene (Cvrckova et al., 1995) was then replaced with a 0.9 kb PstI–PstI fragment derived from plasmid pJJ246 (Jones and Prakash, 1990) to yield plasmid pDH124. This plasmid was linearized with SmaI and KpnI and transformed into yeast to replace 1.5 kb of the chromosomal CLA4 gene with TRP1 by homologous recombination (Rose et al., 1990). The replacement was confirmed by PCR using oligodeoxynucleotide primers corresponding to the nucleotide sequences from positions 1012–1032 and 2742–2762 of the CLA4 gene (Cvrckova et al., 1995) (data not shown).

Yeast mating and orientation assays

Patch mating tests were carried out as described (Leberer et al., 1992b). Quantitative mating assays were performed by a filter mating assay (Leberer et al., 1993). Briefly, exponentially growing experimental cells were mixed with a 4‐fold excess of exponentially growing tester cells and filtered through nitrocellulose filters. The filters were placed onto YEP‐glucose (2%) or YEP‐galactose (4%) plates and mating was allowed to proceed for 4 h at 30°C. The cells were then washed off the filters and titered on selective media to determine the number of diploid cells and the number of diploid cells plus experimental haploid cells. The mating efficiency was defined as the number of diploid cells determined on the first selective medium divided by the number of diploid plus experimental haploid cells determined on the second selective medium.

Orientation of yeast cells in a gradient of α‐factor was measured as described (Segall, 1993).

Two‐hybrid protein interaction assays

The assays were performed as described (Leeuw et al., 1995) in strain M364‐2C which is a far1‐1 derivative strain of L40. Plasmids pRL39 and pRL58 encoding fusions of the transciptional activation domain of Gal4p with the amino‐terminus of Cdc42p and Cdc42pG12V, respectively, were constructed by cloning PCR fragments into plasmid pGAD424 (Chien et al., 1991). Wild‐type CDC42 was amplified by using the oligodeoxynucleotide primers 5′‐CCGGAATTCATGCAAACGCTAAAGTGTGTT‐3′ (ORL9) and 5′‐CGCGGATCCCTACAAAATTGTACATTTTTTACT‐3′ (ORL10; the newly created EcoRI and BamHI sites, respectively, are underlined) and plasmid pGAL‐CDC42 (Ziman et al., 1991) as a template. CDC42G12V was amplified by using the oligodeoxynucleotide primers 5′‐CCGGAATTCATGCAAACGCTAAAGTGTGTTGTTGTCGT‐3′ (ORL14; the newly created EcoRI site is underlined) and ORL10, and plasmid pGAL‐CDC42G12V (Ziman et al., 1991) as a template. Plasmids encoding fusions of the LexA DNA binding domain with the amino‐terminus of Ste20p fragments were constructed by cloning PCR fragments of STE20 into plasmid pBTM116 (Chien et al., 1991). Plasmid pDH37 encoding a full‐length Ste20p fusion was constructed by amplifying a BglII–XhoI fragment of STE20 as described (Leberer et al., 1992a), and cloning this fragment into the BamHI and SalI sites of pBTM116. A 1.5 kb EcoRI fragment of this plasmid was then cloned into the EcoRI site of pBTM116 to create plasmid pRL27 encoding a Ste20p1−497fusion. Plasmid pRL7 encoding a Ste20p497−939 fusion was made by amplifying a BamHI–XhoI fragment of STE20 using the oligodeoxynucleotide primers 5′‐CGGGATCCTGAATTCTGCCGCCAATGTTTC‐3′ (OAF7; the BamHI site used for cloning is underlined) and 5′‐CCGCCGCTCGAGCTATCTTAGTGTACCACTATTTGA‐3′ (OAF1; the newly created XhoI site is underlined), and pDH37 as a template. Plasmid pAF2 encoding a Ste20p1−298 fusion was constructed by amplification of a BglII–XhoI fragment of STE20 using the oligodeoxynucleotides 5′‐GATCCTCGACTAAAGATCTTAATGAGCAATGATCC‐3′ (OKC42; the BglII site used for cloning is underlined) and OAF1 as primers, and pDH37 as a template. Plasmid pAF3 encoding a Ste20p1−435 fusion was created by deletion of the SalI–SalI fragment from plasmid pRL27. Plasmid pRL99 encoding the Ste20p1−435/Δ334−369 fusion was constructed by replacing the BamHI–SalI fragment of plasmid pAF3 with the BamHI–SalI fragment of plasmid pRS316‐STE20(Δ334–369).

Resin binding assays

To create plasmid pAN117, an EcoRI–BamHI fragment of CDC42 was removed from plasmid pRL39 and cloned into plasmid pMAL‐p2 (New England Biolabs Inc.) in‐frame with the gene encoding maltose binding protein. To create plasmid pGEX‐STE20(K649A)encoding a catalytically inactive full‐length GST–Ste20p fusion protein, a BglII–XhoI fragment of STE20K649A was amplified by PCR using the oligodeoxynucleotides 5′‐TCAGATCTATGAGCAATGATCCATCT‐3′ and 5′‐CCGCTCGAGTTTACTTTTATCATC‐3′ (the newly created BglII and XhoI sites, respectively, are underlined) as primers, and pVTU‐STE20(K649A) (Wu et al., 1995) as template. The fragment was then cloned into the BamHI and XhoI sites of pGEX‐4T‐3 (Pharmacia Biotech Inc.). To create plasmid pGEX‐STE20(Δ334–369), a BglII–XhoI fragment of STE20Δ334−369 was amplified by PCR using the oligodeoxynucleotides 5′‐TCAGATCTATGAGCAATGATCCATCT‐3′ and 5′‐CCGCTCGAGTTTACTTTTATCATC‐3′ (the newly created BglII and XhoI sites, respectively, are underlined) as primers, and pVTU‐STE20(Δ334–369) as template, and cloned into the BamHI and XhoI sites of pGEX‐4T‐3.

The fusion proteins were expressed in the protease‐deficient E.coli strain UT5600 (New England Biolabs Inc.). Maltose binding protein–Cdc42p was affinity purified by using amylose–agarose beads in buffer A (20 mM Tris–HCl buffer, pH 7.5, containing 200 mM NaCl, 5 mM EDTA, 1 mM β‐mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 1 mM pefabloc, 2 μg/ml pepstatin, 2 μg/ml leupeptin, and 2 μg/ml aprotinin), eluted from the beads by incubation in buffer B (50 mM Tris–HCl buffer, pH 7.5, containing 150 mM NaCl, 1.5 mM MgCl2, 1 mM β‐mercaptoethanol and 10 mM maltose) for 5 min at room temperature, and stored in aliquots at −70°C. The fusion protein was then loaded with GTPγS or GDPβS by incubating 3 μg of purified protein in 30 μl of buffer C [50 mM Tris–HCl buffer, pH 7.5, containing 4 mM EDTA, 0.5 mM dithiothreitol (DTT), 0.5% bovine serum albumin, 0.1 mM GTPγS or GDPβS] for 15 min at 30°C. The reaction was stopped by adding 10 mM MgCl2 and cooling on ice.

GST–Ste20p fusions were bound to glutathione–Sepharose beads and purified as described (Wu et al., 1995). Five μg of GST–Ste20p bound to the resin were incubated with 3 μg of GTPγS‐loaded maltose binding protein–Cdc42p in buffer D (50 mM Tris–HCl buffer, pH 7.5, containing 100 mM NaCl, 5 mM MgCl2, 0.1 mM DTT, 0.2 mM GTP and 0.07% Triton X‐100) for 20 min at room temperature. The beads were then washed three times with buffer D, resuspended in Laemmli buffer and separated by SDS–PAGE (Laemmli, 1970). Immunoblots to detect maltose binding protein and Ste20p were then performed as described (Leeuw et al., 1995).

Microscopical analyses

For actin staining, cells were fixed in 3.7% formaldehyde for 20 min at room temperature, washed twice in 10 mM phosphate buffer, pH 7.4, containing 150 mM NaCl (phosphate‐buffered saline, PBS), incubated for 45 min at room temperature with 0.5 μM rhodamine–phalloidin in PBS and then washed twice in PBS. For microscopic analyses, a Leica Aristoplan epifluorescence microscope was used with differential interference contrast (Nomarski) optics. Fluorescence of rhodamine and GFP was observed with N2.1 and I3 filters, respectively. Pictures were taken with a Kodak TMAX 400 ASA film.

Assays

Western blot analyses and in vitro protein kinase assays were performed as described previously (Wu et al., 1995). β‐Galactosidase activities were measured as described, and Miller units were defined as (OD420×1000)/(OD600×t×v) (Leberer et al., 1992b). Growth inhibition in response to α‐factor was determined by ‘halo’ assays as described previously (Leberer et al., 1994).

Acknowledgements

We wish to thank Doreen Harcus, Robert Larocque and Viktoria Lytvyn for expert technical assistance, and Janet Chenevert, Fatima Cvrckova, Douglas Johnson, Haoping Liu, André Nantel, Roger Y.Tsien and Nicole Valtz for providing yeast strains and plasmids. We also wish to thank Matthias Peter for communicating results prior to publication, and Doreen Harcus and Malcolm Whiteway for critical reading of the manuscript. Work from J.E.S. was supported by a research grant (MCB 9304992) from the US National Science Foundation. The National Research Council of Canada publication number for this work is 39932.

References

View Abstract