Functions for Cdc42p BEM adaptors in regulating a differentiation-type MAP kinase pathway

Abstract

Ras homology (Rho) GTPases regulate cell polarity and signal transduction pathways to control morphogenetic responses in different settings. In yeast, the Rho GTPase Cdc42p regulates cell polarity, and through the p21-activated kinase Ste20p, Cdc42p also regulates mitogen-activated protein kinase (MAPK) pathways (mating, filamentous growth or fMAPK, and HOG). Although much is known about how Cdc42p regulates cell polarity and the mating pathway, how Cdc42p regulates the fMAPK pathway is not clear. To address this question, Cdc42p-dependent MAPK pathways were compared in the filamentous (Σ1278b) strain background. Each MAPK pathway showed a unique activation profile, with the fMAPK pathway exhibiting slow activation kinetics compared with the mating and HOG pathways. A previously characterized version of Cdc42p, Cdc42p^E100A, that is specifically defective for fMAPK pathway signaling, was defective for interaction with Bem4p, the pathway-specific adaptor for the fMAPK pathway. Corresponding residues in Bem4p were identified that were required for interaction with Cdc42p and fMAPK pathway signaling. The polarity adaptor Bem1p also regulated the fMAPK pathway. Versions of Bem1p defective for recruitment of Ste20p to the plasma membrane, intramolecular interactions, and interaction with the GEF, Cdc24p, were defective for fMAPK pathway signaling. Bem1p also regulated effector pathways in different ways. In some pathways, multiple domains of the protein were required for its function, whereas in other pathways, a single domain or function was needed. Genetic suppression tests showed that Bem4p and Bem1p regulate the fMAPK pathway in an ordered sequence. Collectively, the study demonstrates unique and sequential functions for Rho GTPase adaptors in regulating MAPK pathways.

INTRODUCTION

Mitogen-activated protein kinase (MAPK) pathways are evolutionarily conserved signaling modules that control cell proliferation, cell differentiation, and the response to stress in eukaryotes (Keshet and Seger, 2010). In the budding yeast Saccharomyces cerevisiae, MAPK pathways control cell differentiation to specific cell types, such as the mating type or shmoo, which is regulated by the pheromone response pathway, the filamentous cell type by the fMAPK pathway, and the formation of meiotic progeny by the sporulation pathway. MAPK pathways also control the response to stress, such as osmotic stress, which is regulated by the high osmolarity glycerol response (HOG) pathway, and the response to compromised cell integrity by the protein kinase C pathway. Like many signaling pathways, MAPK pathways in yeast can share components (Schwartz and Madhani, 2004; Bardwell, 2005; Chen and Thorner, 2007; Hohmann et al., 2007; Saito, 2010; Levin, 2011). The Ras homology (Rho) GTPase Cdc42p regulates three MAPK pathways (mating, fMAPK, and HOG) and is an essential protein that controls the establishment of cell polarity, polarized secretion, and protein trafficking (Park and Bi, 2007). How Cdc42p and other proteins induce a pathway-specific response in a particular setting is not clear. This question is relevant because in higher organisms, cross-talk and misregulation of GTPase pathways, like those requiring CDC42 and RAS, are leading causes of cancer and other diseases. Understanding how common proteins function in pathway-specific settings remains at the forefront of studies of signaling pathway regulation (Giehl, 2005; Dhillon et al., 2007; Maik-Rachline and Seger, 2016; Rauch et al., 2016).

In response to carbon or nitrogen limitation, yeast and other fungal species can undergo a microbial differentiation response called filamentous or invasive/pseudohyphal growth (Gimeno et al., 1992; Cullen and Sprague, 2012). In some plant and animal pathogens, like the major human pathogen Candida albicans, filamentous growth is required for virulence (Lo et al., 1997; Whiteway and Bachewich, 2007; Finkel and Mitchell, 2011; Desai et al., 2014). In yeast, filamentous growth occurs in wild strain backgrounds (Σ1278b is used here) because the properties that are associated with filamentous growth have been lost due to genetic manipulation of laboratory strains (Liu et al., 1996; Dowell et al., 2010; Chin et al., 2012; Ryan et al., 2012). The Cdc42p-dependent fMAPK pathway is one of the major signaling pathways that regulates filamentous growth (Roberts and Fink, 1994; Cullen et al., 2004; Cullen and Sprague, 2012). Studies of the fMAPK pathway have provided insights into how ERK-type MAPK pathways regulate eukaryotic cell differentiation and fungal pathogenesis.

The fMAPK pathway shares components with the HOG and mating pathways, yet each pathway induces a specific set of target genes and a unique response (Brewster et al., 1993; Peter et al., 1996; Leberer et al., 1997; Roberts et al., 2000; Lamson et al., 2002; Schwartz and Madhani, 2004; Bardwell, 2005; Chen and Thorner, 2007; Cullen and Sprague, 2012). A core module composed of Cdc42p, the p21-activated kinase (PAK) Ste20p (Peter et al., 1996; Leberer et al., 1997), and the MAPKKK Ste11p regulates all three MAPK pathways. This core module regulates different MAP kinases for the three pathways. Kss1p is the main regulator of the fMAPK pathway (Madhani et al., 1997). Kss1p regulates a suite of transcriptional activators (Ste12p, Tec1p, Msa1p, and Msa2p) and repressors (Dig1p and Dig2p) (Cook et al., 1996, 1997; Madhani and Fink, 1997; Madhani et al., 1997; Bardwell et al., 1998a,b; Cullen and Sprague, 2000b; van der Felden et al., 2014) to activate transcription at filamentation-specific promoters (Madhani and Fink, 1997; Zeitlinger et al., 2003). Tec1p in particular functions exclusively in the fMAPK pathway and is degraded during the mating response (Bao et al., 2004; Chou et al., 2004).

Several proteins have been identified that function at or above the level of Cdc42p and regulate the fMAPK pathway. The mucin-type glycoprotein Msb2p interacts with the GTP-bound conformation of Cdc42p and is the main sensor that regulates the fMAPK pathway (Cullen et al., 2004). Msb2p functions with two other transmembrane proteins, Sho1p and Opy2p (Wu et al., 2006; Tatebayashi et al., 2007; Ekiel et al., 2009; Pitoniak et al., 2009; Yang et al., 2009; Yamamoto et al., 2010, 2016; Karunanithi and Cullen, 2012; Tatebayashi et al., 2015). Msb2p, Sho1p, and Opy2p regulate the fMAPK pathway and also the HOG pathway but do not control the activity of the mating pathway. Bem4p is a cytosolic adaptor and Cdc42p-interacting protein (Hirano et al., 1996; Mack et al., 1996) that regulates the fMAPK pathway but not the mating or HOG pathways (Pitoniak et al., 2015). Bem4p associates with Cdc42p, the guanine nucleotide exchange factor (GEF) Cdc24p (Johnson, 1999; Erickson and Cerione, 2001; Etienne-Manneville, 2004), the MAPKKK Ste11p, and the MAPK Kss1p (Pitoniak et al., 2015). Bem4p may direct Cdc42p, in some manner, to a pathway-specific complex containing Kss1p. We have also shown that bud-site-selection proteins, culminating at the Ras-type GTPase Rsr1p (Chant and Pringle, 1995; Bi and Park, 2012), also regulate the fMAPK pathway but not other Cdc42p-dependent MAPK pathways (Basu et al., 2016). Therefore, Msb2p, Bem4p, and Rsr1p function at the level of Cdc42p and regulate the fMAPK pathway.

Despite the utility of studying MAPK pathways in a genetically amenable system, Cdc42p-dependent MAPK pathways have not been directly compared in yeast. One reason is that most studies of the mating and HOG pathways have been carried out in laboratory strains, which have lost the ability to undergo filamentous growth. Another reason is that the HOG pathway has redundant branches that converge on the MAPKK Pbs2p, bypassing Cdc42p. Only one branch is dependent on Cdc42p, the Ste11p branch (Posas et al., 1996; Posas and Saito, 1997, 1998). Here, we developed a strain and assays that allowed us to directly compare Cdc42p-dependent MAPK pathways in yeast. By exploring this signaling landscape, we found that the pathways showed different kinetics, with the fMAPK pathway being activated more slowly than the mating and HOG pathways. We also found that the interaction between Bem4p and Cdc42p is critical for fMAPK pathway signaling. The major polarity adaptor for Cdc42p, Bem1p, also regulated the fMAPK pathway. Bem1p also regulates the mating (Leeuw et al., 1995) and HOG pathways (Tanaka et al., 2014), so its role in regulating the fMAPK pathway might be expected. What was surprising was that Bem1p regulated different effector pathways in different ways. We propose a model for how the Cdc42p module is activated in the context of the fMAPK pathway by interactions with Cdc24p (by Bem4p, Bem1p, and Rsr1p), Cdc42p (by Bem4p and Bem1p), and Ste20p (by Bem1p). Our study provides insights into how a Rho GTPase is directed to a differentiation-type MAPK pathway.

RESULTS

Signaling landscape of Cdc42p-dependent MAPK pathways

Three MAPK pathways in yeast (fMAPK, HOG, and mating) require common components yet induce different responses (Figure 1A; Roberts et al., 2000; Maleri et al., 2004; Bhattacharyya et al., 2006). The Rho-type GTPase Cdc42p, PAK Ste20p, and MAPKKK Ste11p are central to the regulation of the three pathways (Figure 1A, gray box). To directly compare Cdc42p-dependent MAPK pathways, experiments were performed in filamentous (Σ1278b) strains lacking the Sln1p-branch of the HOG pathway (ssk1Δ). Growth of cells in the nonpreferred carbon source galactose (GAL) induced filamentous growth (Figure 1B, left) and phosphorylation of the MAP kinase Kss1p (Figure 1B, right, P∼Kss1p and to a minor extent P∼Fus3p). Osmotic shock (0.4 M KCl) did not induce a morphogenetic response (Figure 1C, left) but induced rapid phosphorylation of Hog1p (Figure 1C, right). The mating pheromone α-factor induced mating projections (Figure 1D, left) and phosphorylation of Fus3p and Kss1p (Figure 1D, right, note low levels of control protein at the 6-h time point). Kss1p is normally phosphorylated in response to α-factor and functions to modulate the mating response (Ma et al., 1995; Sabbagh et al., 2001).

FIGURE 1:

Comparison of MAP kinase pathways that require Cdc42p and other proteins. (A) Three Cdc42p-dependent MAP kinase pathways. Colored text refers to pathway-specific factors fMAPK (red), mating (blue), and HOG (green). Black text refers to common or shared factors. The gray box denotes factors that are common to all three pathways. (B) To assess fMAPK pathway activity, wild-type cells (PC6810) were grown for 8 h in YEP-GAL and photographed (left, DIC 100X, bar, 10 microns). At right, immunoblots were probed with antibodies (p44/p42) to detect P∼Kss1p and P∼Fus3p. Immunoblots were also probed with α-Kss1p and α-Pgk1p antibodies. (C) The same cells as in B were grown in YPED + 0.4 M KCl (HOG) for 8 h and photographed (left, DIC 100×, bar, 10 microns). At right, immunoblots were probed with antibodies to detect P∼Hog1p (p38), Hog1p, and Pgk1p. (D) The same cells in B were grown for 8 h in YEPD + 6 µM α-factor to induce mating (left, DIC 100×, bar, 10 microns). At right, immunoblots were probed with α-p44/p42, α-Kss1p, α-Fus3p, and α-Pgk1p antibodies. In B–D, numbers indicate the relative band intensity of P∼MAPK to total MAPK levels, which was set to 1 for basal conditions. Asterisk, background band. Phosphoblots were performed in multiple replicates (n = 3 experiments), and representative images are shown. (E) b-Galactosidase assays of MAPK pathway-dependent reporters. For the x-axis, values are expressed as fold induction compared with the uninduced condition (YEPD) for each reporter y-axis, time (h). Each data point is represents the average of three biological replicates (n = 3). Error bars show the SD between trials. *p < 0.05 for values compared between the FUS1-lacZ data points or the 8XCRE-lacZ data points to the FRE-lacZ for the times indicated.

The above experiments looked at phosphorylated MAP kinases as a readout of pathway activity. As an independent test, the expression of MAPK pathway reporter genes was evaluated. GAL induced expression of an fMAPK pathway-dependent reporter (Figure 1E, red, pFRE-lacZ). Osmotic shock induced expression of a HOG pathway reporter (Figure 1E, green, p8XCRE-lacZ), and α-factor induced expression of a mating pathway reporter (Figure 1E, blue, pFUS1-lacZ). The phosphoblot analysis and transcriptional reporter assays showed that the fMAPK pathway was activated more slowly than the mating and HOG pathways. By P∼blot analysis, the fMAPK pathway was maximally activated by 6 h, compared with 5 min for HOG, and 1 h for the mating pathway (Figure 1, B–D). The mating and fMAPK pathways were directly comparable because P∼Fus3p and P∼Kss1p are recognized by the same antibodies (Supplemental Figure S1). By transcriptional reporter assays, the fMAPK pathway showed maximal activity by 7.5 h compared with 5 min for the HOG and 20 min for the mating pathways (Figure 1E). The fMAPK pathway also showed less transcriptional reporter induction (fourfold for the FRE-lacZ reporter) than the HOG (20-fold) and mating pathways (20-fold). This observation is consistent with previous studies from our lab of the FRE-lacZ reporter (Chavel et al., 2014) and other reporters (e.g., KSS1, YLRO42c, MSB2; Pitoniak et al., 2009) that show moderate induction in GAL. Therefore, two separate tests show that the fMAPK pathway has slow activation kinetics compared with the HOG and mating pathways.

The activity of the Cdc42p-dependent MAPK pathways was also examined in response to exposure to their noncognate stimuli (Supplemental Figures S1 and S2). With two exceptions, the pathways functioned in an insulated manner: 1) GAL induced the HOG pathway at a rate far less than that observed with KCl induction (Supplemental Figure S1A, p38 blot), which occurs during the filamentation response (Adhikari and Cullen, 2014), and 2) α-factor induced phosphorylation of Hog1p (Supplemental Figure S1A, p38 blot), which did not result in induction of the 8XCRE–lacZ reporter (Supplemental Figure S2B). These results support the prevailing view that Cdc42p-dependent pathways induce unique transcriptional and morphogenetic responses to specific stimuli.

Characterization of mutant isoforms of Cdc42p with MAPK pathway-specific phenotypes

MAPK pathways were also assessed by tests that were less labor intensive than phosphoblot analysis and provided a first approximation of MAPK pathway activity. The plate-washing assay (PWA) measures invasive growth and the activity of the fMAPK pathway (Roberts and Fink, 1994). Salt sensitivity was used to measure the activity of the HOG pathway (Maeda et al., 1994, 1995), and growth arrest by α-factor (halo assays) was used to measure the activity of the mating pathway (Sprague et al., 1983). Data from the functional tests were quantified by ImageJ analysis and color coded to facilitate interpretation (Figure 2A). Mutants lacking an intact fMAPK pathway were examined by these tests. As expected, loss of common amplifiers of the fMAPK pathway showed defects in all three pathways (Figure 2A; Supplemental Figure S3A, ste20Δ and ste11Δ). Also as expected, several mutants were defective for two of the three pathways (Figure 2A; Supplemental Figure S3A, sho1Δ, opy2Δ, and ste50Δ). Ste50p interacts with Ste11p and is thought to regulate all three pathways (Wu et al., 1999, 2006; Jansen et al., 2001; Truckses et al., 2006) but did not regulate the mating pathway in this context. We also found, in line with our previous work (Pitoniak et al., 2015), that loss of the adaptor protein Bem4p only impacted in the fMAPK pathway (Figure 2A; Supplemental Figure S3A, bem4Δ). Two proteins that regulate the HOG pathway, the PAK Cla4p (Cvrckova et al., 1995; Tatebayashi et al., 2006) and the adaptor protein Ahk1p (Nishimura et al., 2016), were also tested and were not found to regulate the HOG or fMAPK pathways (Figure 2A; Supplemental Figure S3, B–F). Thus, Ahk1p and Cla4p do not appear to regulate the fMAPK pathway. These results support and extend our understanding of the proteins that regulate the fMAPK pathway.

FIGURE 2:

Interaction between Bem4p and Cdc42p is required for fMAPK pathway activity. (A) CDC42 alleles that show phenotypes in MAP kinase-dependent assays (for raw data, see Supplemental Figures S3, S5, and S6). Color represents a phenotype in the PWA on YEPD (fMAPK, F, red), growth of cells on YEPD + 1 M KCl (HOG, H, green), or growth arrest by halo formation in response to α-factor (6 and 1.8 µM concentrations were tested; Mating, M, blue). Color intensity represents the severity of the defect based on quantitation by ImageJ analysis; gray, no difference from wild type. (B) Wild-type cells (PC6810), a collection of cdc42 alleles, and the ste11Δ mutant control strain (PC6604), were grown in YEP-GAL for 5.5 h to examine P∼Kss1p levels. See Figure 1B for details. (C) Two-hybrid analysis of the indicated versions of Cdc42p and Bem4p. Two-hybrid analysis was performed in multiple replicates (n = 2), and representative images are shown. For the complete data set, see Supplemental Figure S8. (D) Phenotypes of Cdc42p^E100A and the bem4∆ mutant in tests for MAP kinase pathway activity. See panel A for details. (E) Ribbon diagram of the GDP-bound conformation of Cdc42p, based on the structure of the mammalian Cdc42p and DOCK9 complex (PDB ID#2wmn). The switch I and II regions are highlighted in light green (Johnson, 1999). The glutamate residue at position 100 in the α3 helix is shown in yellow. The K328 residue in Bem4p is shown in magenta. (F) P∼Kss1p levels in strains carrying Bem4p^{K323A K328A}-HA and controls. Immunoblots were performed as in Figure 1B, except blots were also probed with α-HA antibodies to detect Bem4p-HA. (G) Wild-type cells (PC6810) and cells expressing cdc42^V36T (see Supplemental Table S1) containing pTEF2 (pCTL, PC6365) or pTEF2-BEM4-GFP (PC6367). Bar, 10 microns. Numbers indicate the circularity index (length-to-width ratio) of cells (n > 100) measured by ImageJ analysis; error, SD (H) The cdc42::NAT strain carrying the pRS316-GFP-CDC42 plasmid and indicated CDC42 alleles were examined for growth on SD-URA and 5-FOA media (n = 2).

Despite the fact that Cdc42p binds to the same PAK, Ste20p, to regulate the three MAPK pathways, versions of Cdc42p have been identified with MAPK pathway-specific phenotypes (Mosch et al., 2001; Tatebayashi et al., 2006; Truckses et al., 2006). An N-terminal green fluorescent protein (GFP) fusion to Cdc42p, expressed as the sole copy of CDC42 in the cell, was defective in the fMAPK pathway but not the HOG or mating pathways (Figure 2A, pGFP-Cdc42; Supplemental Figure S4). To follow up on this observation, a collection of cdc42 alleles that targeted charged, often evolutionarily conserved amino acid residues and functional domains (Kozminski et al., 2000), was introduced into Σ1278b, which is capable of filamentous growth. Twenty-four nonlethal alleles at 25°C were viable and were examined by functional tests (Supplemental Table S1; Supplemental Figures S5–S7). Most alleles of CDC42 did not show a mutant phenotype (Supplemental Figures S5–S7). However, several alleles showed pathway-specific phenotypes. Five alleles were mainly defective in the fMAPK pathway based on invasive growth (Figure 2A, K5A, D170A E171A, E100A, H102-104A, and F37Y) and P∼Kss1p analysis (Figure 2B). Note that the integrated version of CDC42 (INT) had a defect in invasive growth. E100A was previously shown to be specifically defective for fMAPK pathway activity (Mosch et al., 2001; Truckses et al., 2006). Another allele of CDC42 was identified that was primarily defective for growth on salt (Figure 2, A and B, R144A). Several alleles contained mutations in the Switch I domain, which binds PAKs and, as one might expect, were defective for all three pathways (Figure 2, A and B, V36T).

Interaction between Bem4p and Cdc42p stimulates fMAPK pathway signaling

Versions of Cdc42p that are specifically defective for fMAPK pathway activity might fail to interact with proteins that regulate the fMAPK pathway. In particular, we considered Bem4p because it is a pathway-specific regulator of the fMAPK pathway (Figure 2A; Pitoniak et al., 2015) and a Cdc42p-interacting protein (Mack et al., 1996; Drees et al., 2001). Bem4p interacts with Cdc42p by two-­hybrid analysis (Drees et al., 2001; Pitoniak et al., 2015). In this study, two-hybrid analysis was examined using a version that mimicked the GDP-bound conformation and did not associate with membranes (CDC42^{D118A C188S}). By two-hybrid analysis, Bem4p showed reduced interaction with Cdc42p^E100A (Figure 2C), which is present at the same level as wild-type Cdc42 (Supplemental Figure S4G). Bem4p was also partially defective for interaction with Cdc42p^K5A and Cdc42p^{D170A E171A}, but not for interaction with Cdc42p^V36T, Cdc42p^H102-104A, or Cdc42p^R144A (Supplemental Figure S8A). Thus, the signaling defect of Cdc42p^E100A in the fMAPK pathway might result from its inability to interact with Bem4p. Consistent with this possibility, the phenotypes of cells carrying Cdc42p^E100A resembled the phenotypes of the bem4Δ mutant. Like the bem4Δ mutant, Cdc42p^E100A was defective for invasive growth but not for growth on salt or halo formation (Figure 2D). Cdc42p^E100A was also defective for invasive growth by another commonly used test, the single cell invasive growth assay (Cullen and Sprague, 2000b), but not shmoo formation in response to α-factor (Supplemental Figure S7). Similar phenotypes have been reported for the bem4Δ mutant (Pitoniak et al., 2015).

Like other monomeric GTPases, Cdc42p undergoes a conformational change upon GTP binding that allows the protein to interact with effector proteins. Most proteins that interact with Cdc42p bind to the active or GTP-bound conformation of the protein. An unusual feature of Bem4p is that it interacts with the GDP-bound (inactive) and GTP-bound (active) conformations of Cdc42p (Drees et al., 2001; Pitoniak et al., 2015). We modeled the yeast Cdc42p protein onto the crystal structure of human Cdc42p (Rittinger et al., 1997; Nassar et al., 1998; Hoffman et al., 2000), which are >80% identical, and found that the glutamic acid at position 100, which is conserved between human and yeast Cdc42p, lies in the third alpha (α3) helix of the protein (Figure 2E, Cdc42 is in green, E100 is marked in yellow). The α3 helix is on the opposite side of the protein as the effector-binding domain (Figure 2E, Switch I and II) and is not thought to undergo a conformational change upon nucleotide binding. Therefore, Bem4p may interact with a region of Cdc42p that does not undergo a conformational change. This finding is consistent with Bem4’s ability to associate with the GDP and GTP-bound conformations of Cdc42p (Pitoniak et al., 2015).

We previously mapped a region of Bem4p from 300 to 400 amino acid residues as being required for interaction with Cdc42p (Pitoniak et al., 2015). This region exhibits sequence similarity to members of the SmgGDS family of proteins. SmgGDS proteins contain Armadillo (ARM) repeats (Peifer et al., 1994; Shimizu et al., 2017) that mediate interactions with Rho GTPases (Vithalani et al., 1998; Vikis et al., 2002; Hamel et al., 2011; Shimizu et al., 2017, 2018; Jennings et al., 2018) and other proteins. The predicted structure of Bem4p was determined by bioinformatics algorithms Phyre 2 and iTASSER, which generated similar structures with > 90% confidence and theoretical root mean square deviation values above cutoff. In both structures, Bem4p contained alpha helical motifs that resembled ARM repeats (Figure 2E, Bem4p is in blue, and the 300–400 region is marked in light blue).

Basic residues in the Bem4p^300-400 region might mediate electrostatic interactions with the glutamate at position 100 in Cdc42p. Ten basic residues are present in the Bem4p^300-400 region. Site-directed mutagenesis was used to generate versions of Bem4p that lacked basic residues based on the predicted minimal free energy between Cdc42p^E100A and Bem4p^300-400 using the protein-docking server Haddock (van Zundert et al., 2016). Bem4p^{K323A K328A K351A R352A}, Bem4p^{K323A, K328A}, Bem4p^{K351A R352A}, Bem4p^K323A, and Bem4p^K328A were constructed and tested. Bem4p^{K323A K328A K351A R352A}, Bem4p^{K323A K328A}, and Bem4p^K328A were defective for interaction with Cdc42p (Figure 2C; Supplemental Figure S8). The Bem4p^300-400 region is not thought to be important for the stability of the protein (Pitoniak et al., 2015). Bem4p^{K351A R352A} was partially defective, and Bem4p^K323A was not defective (Supplemental Figure S8). The residue K328 in Bem4p had the lowest free energy based on Haddock and may be a key residue that mediates interaction with Cdc42p (Figure 2E, red residue; Supplemental Movie S1). Compensatory mutations did not restore the interaction between Bem4p and Cdc42p (Supplemental Figure S8, Bem4p^K328E and Cdc42p^E100A), which indicates that, like for other Cdc42p-interacting proteins (Hoffman et al., 2000), the interaction between Bem4p and Cdc42p may involve multiple electrostatic contacts. Therefore, basic residues in the ARM-type repeat region of Bem4p are required for interaction with Cdc42p.

Movie S1.

Download video file ^{(2.7MB, mov)}

Modeling the interaction between Bem4p and Cdc42p.

Versions of Bem4p that showed reduced interaction with Cdc42p were tested for fMAPK pathway activity. Bem4p-HA^{K323A K328A} was expressed from the BEM4 locus at levels similar to the Bem4p-HA protein (Figure 2F, anti-HA blot). Bem4p-HA^{K323A K328A} was defective for fMAPK pathway activity based on P∼Kss1p levels (Figure 2F). Bem4p-HA^{K323A K328A} was also defective for invasive growth and expression of the FUS1-HIS3 reporter (Supplemental Figure S9), which in strains lacking an intact mating pathway (ste4Δ) provides a readout of fMAPK pathway activity (Cullen et al., 2004; Chavel et al., 2010; Adhikari et al., 2015b; Pitoniak et al., 2015; Basu et al., 2016). Bem4p-HA^K328A was also tested but did not show the same defects in MAPK signaling and invasive growth as Bem4p-HA^{K323A K328A}. Thus, both residues of Bem4p are probably important for the regulation of Cdc42p in the fMAPK pathway.

Bem4p can also regulate cell polarity. In particular, high-copy plasmids carrying BEM4 can suppress the temperature-sensitive growth defects of cdc42 alleles (Mack et al., 1996). A high-copy plasmid expressing BEM4 (pTEF2-BEM4) suppressed the morphological defect of cells carrying a version of Cdc42p with a single amino acid change in its effector binding domain (EFF), Cdc42p^V36T (Figure 2G). To test whether suppression occurred through the interaction between Bem4p and Cdc42p, the E100A mutation was introduced into cdc42^V36T by site-directed mutagenesis. Based on the above results, the E100A mutation would be expected to interfere with the interaction between Bem4p and Cdc42p. Cells expressing Cdc42p^{V36T E100A} had a viability defect compared with cells carrying versions of Cdc42p with either single change, based on plasmid loss with the drug 5-fluoroorotic acid (Figure 2H, 5-FOA; Boeke et al., 1984). Thus, Cdc42p^V36T and Cdc42p^E100A may function synergistically to regulate viability. Therefore, the interaction between Bem4p and Cdc42p may also be required for viability when Cdc42p’s interaction with effector proteins and its cell polarity function are compromised.

The polarity adaptor Bem1p also regulates the fMAPK pathway

Bem4p is a pathway-specific adaptor for the fMAPK pathway. Bem1p is another adaptor for Cdc42p, which functions to regulate the establishment of cell polarity (Bender and Pringle, 1991; Zheng et al., 1995; Park et al., 1997). Bem1p is critical for symmetry breaking by generating positive feedback (Butty et al., 1998; Bose et al., 2001; Gladfelter et al., 2001; Yamaguchi et al., 2007). Bem1p also regulates the mating (Leeuw et al., 1995; Lyons et al., 1996) and HOG pathways (Tanaka et al., 2014). Bem1p has not been tested for a role in regulating the fMAPK pathway because it is essential for viability in the Σ1278b background but is not essential in laboratory strains like S288c (Dowell et al., 2010).

To determine whether Bem1p regulates the fMAPK pathway, the BEM1 gene was deleted in a strain containing the BEM1 gene on a plasmid (pBEM1, provided by D. Lew). Plasmid-loss experiments with 5-FOA showed that the bem1∆ mutant was viable under some conditions (yeast extract [YEP]-GAL media at 30°C and synthetic media) but not others (yeast extract, peptone, and dextrose [YEPD] media at 30°C or YEP-GAL media at 25°C). The conditional viability of the bem1∆ mutant allowed assessment of Bem1p’s role in regulating fMAPK. The bem1∆ mutant showed reduced fMAPK pathway activity based on phosphorylation of Kss1p (Figure 3A). The signaling defect of the bem1∆ mutant was less severe than that of the ste11Δ mutant and resembled the phenotype of cells lacking other upstream pathway components like Msb2p (Cullen et al., 2004), Sho1p (Adhikari et al., 2015a), and Rsr1p (Basu et al., 2016). The bem1∆ mutant was also defective for invasive growth by the PWA (Figure 3B, washed). The bem1∆ mutant had a defect in colony ruffling (Figure 3B, Magnified), which occurs due to cell-to-cell contacts during biofilm/mat formation, a microbial process that is also controlled by the fMAPK pathway (Reynolds and Fink, 2001; Granek and Magwene, 2010). The bem1∆ mutant was defective in the formation of filamentous chains of cells (Figure 3B, Scraped; Supplemental Figure S10). By the single cell assay, the bem1∆ mutant exhibited polarity and bud-site-selection defects (Figure 3B, Single cell; Supplemental Figure S10). Therefore, Bem1p regulates the fMAPK pathway and is required for invasive growth and biofilm/mat formation.

FIGURE 3:

Bem1p regulates the fMAPK pathway. (A) P∼Kss1p levels in wild-type cells (PC538), the bem1Δ mutant (PC6509), and the ste11Δ mutant (PC3861) grown for 5.5 h in SD + AA media. See Figure 1B for details. (B) Filamentous growth and colony patterning of the same strains as in A grown in YP-GAL for 3 d. Bar, 20 microns. The experiment was performed in multiple biological replicates (n = 3). Representative images are shown.

Bem1p regulates the fMAPK pathway by recruitment of Ste20p to the plasma membrane (PM)

Bem1p is composed of domains that perform specific functions during polarity establishment (Park et al., 1997; Endo et al., 2003; Irazoqui et al., 2003). These include two SH3 domains, a PX domain, and a PB1 domain (Figure 4A). Point mutations in these domains selectively cripple individual aspects of Bem1p function (Irazoqui et al., 2003). Versions of Bem1p lacking several of these functional domains were tested to ascertain Bem1p’s role in regulating the fMAPK pathway. The fMAPK pathway operates in basal (Figure 4B, SD-LEU) and activated states (Supplemental Figure S11A, YEP-GAL) depending on the condition (Cullen et al., 2004; Adhikari et al., 2015a; Basu et al., 2016). Bem1p^P208L, which is defective for binding to effector proteins, like Ste20p (Leeuw et al., 1995; Bose et al., 2001; Winters and Pryciak, 2005; Takaku et al., 2010), was defective for fMAPK pathway signaling (Figure 4B). Bem1p^P355A, which prevents an intramolecular interaction that results in a constitutively open conformation of the protein (Irazoqui et al., 2003), was also defective for fMAPK pathway signaling (Figure 4B). Bem1p^R369A, which is defective for localization to the PM and binding to the phosphatidyl inositol polyphosphate (PIP) PI(4,5)P₂ at the PM (Ago et al., 2001; Yu and Lemmon, 2001; Irazoqui et al., 2003), was also defective for fMAPK pathway activity (Figure 4B). Bem1p^K482A, which is defective for interaction with Cdc24p (Peterson et al., 1994; Zheng et al., 1995; Park et al., 1997; Toenjes et al., 1999; Elion, 2000; Gulli et al., 2000; Nern and Arkowitz, 2000; Shimada et al., 2000; Ito et al., 2001; Butty et al., 2002), did not show a defect in fMAPK signaling (Figure 4B) However, a minor defect was seen in YEP-GAL (see Figure 5E and Supplemental Figure S11A). One interpretation of this result is that Bem1p interacts with effector proteins and Cdc24p at the PM to regulate the fMAPK pathway. Bem1p may also require cycling between its open and closed conformation to regulate the fMAPK pathway.

FIGURE 4:

Role of Bem1p in regulating the fMAPK pathway. (A) Diagram of the Bem1p protein, which shows the effector-binding domain (EFF), intramolecular binding domain (IM), plasma membrane (PM) binding domain, and Cdc24p-interacting domain (Cdc24). Alleles used in the study come from Irazoqui et al. (2003). (B) Immunoblot analysis of wild-type cells (PC6810) and the bem1Δ mutant (PC6680) containing wild-type BEM1 (PC6514), a control plasmid pRS315 (CTL), or BEM1 alleles on plasmids (PC6515-6518), and the ste11Δ mutant (PC6604) grown on SD-LEU media. Immunoblots were performed as in Figure 1B, except blots were also probed with α−Myc antibodies to detect Bem1p-Myc. Axial budding (%) was determined for cells grown to mid-log phase in SD-LEU medium. Budding pattern was determined by CFW staining (n > 200). FRE-lacZ assays were performed under the same conditions (n = 3) and the average values are shown. A <10% difference by SD was observed between trials. (C) Localization of GFP-Ste20p expressed from plasmid (PC4394) described in Leberer et al. (1997) in wild-type cells (PC6810) and the bem1Δ mutant (PC6680). Cells were grown in S-GAL-URA medium to mid log phase (4 h). Bar, 10 microns. (D) Immunoblot of GFP-Ste20p in the indicated strains grown under the same conditons. Antibodies to Pgk1p were used as a control for protein levels.

FIGURE 5:

Bem1p regulates different pathways in different ways. (A) Wild-type cells (PC6810), the bem1Δ mutant (PC6680), and the ste11Δ mutant (PC6604) were grown for 5.5 h in YEP-GAL media and extracts were probed by immunoblot analysis as described in Figure 1B. (B) Cells in A were grown in SD + AA containing 0.4 M KCl (5 min). Immunoblots were performed as described in Figure 1C. Cells in A were also grown in SD + AA media with 6 µM of α-factor. (D) Cells in A were examined by phenotypic tests. (E) Cells in Figure 4B were examined by phosphoblot analysis under fMAPK pathway-activating conditions (YEP-GAL). FRE-lacZ assays were performed in YEP-GAL (n = 3). Average values are reported with <10% SD between trials. Phosphoblots were also performed as indicated during growth on SD-LEU + 0.4 M KCl (to evaluate P∼Hog1p) and on SD-LEU + 6 µM α-factor (to evaluate P∼Fus3p). (F) Growth of bem1Δ mutants in the ste4 SSK1+ (PC6509) or ssk1Δ background (PC6680), and the ssk1Δ rsr1Δ bem1Δ triple mutant (PC7343) containing the indicated plasmids was assessed by spotting cells onto the indicated media. On SD-LEU-HIS, the FUS1-HIS3 growth reporter was used to examine the fMAPK pathway activity. (G) The localization of Sho1p-GFP and pGFP-2XPH-PLCδ in wild-type cells and the bem1Δ mutant. Two representative examples are shown (n > 50; Bar, 10 microns). (H) Immunoblot analysis of pGFP-2XPH-PLCδ and Sho1p-GFP in wild-type cells and the bem1Δ mutant. Blots for α-GFP and α-Pgk1p as a control for protein loading.

To test whether Bem1p is required for the recruitment of Ste20p to the PM, the localization of GFP-Ste20p was examined. In wild-type cells, GFP-Ste20p localized to the cell cortex in small buds and the distal pole in large cells (Figure 4C, seen in 5–10% of cells). In cells lacking Bem1p, GFP-Ste20p showed a diffuse pattern (Figure 4C, seen in >200 cells examined). By immunoblot analysis, the level of GFP-Ste20p was the same in wild-type and bem1Δ mutant cells (Figure 4D). These data fit with the idea that Ste20p is the major effector of Cdc42p in the fMAPK pathway (Peter et al., 1996; Leberer et al., 1997; Truckses et al., 2006) and with the idea that Bem1p recruits Ste20p to GTP-Cdc42p at the PM during mating (Peter et al., 1996; Leberer et al., 1997; Winters and Pryciak, 2005; Truckses et al., 2006) and is required for plasma-membrane localization of Ste20p during filamentous growth (Mosch et al., 2001). Therefore, Bem1p is required for the PM localization of Ste20p in the fMAPK pathway. This may be due to a direct interaction between Bem1p and Ste20p or involve other proteins.

Bem1p regulates effector pathways by different mechanisms

To compare Bem1p function in Cdc42p-dependent MAPK pathways, phosphoblot and functional assays were performed. As shown above (Figures 3 and 4), Bem1p regulated the fMAPK pathway (Figure 5, A and D). Bem1p did not play a major role in regulating the HOG pathway (Figure 5B, at 5 min) although the bem1Δ ssk1Δ mutant was sensitive to growth on high-osmolarity medium (Figure 5D, KCl). Bem1p also regulated the mating pathway (Figure 5, C and D). Therefore, Bem1p is a general regulator of Cdc42p-dependent MAPK pathways in yeast.

Does Bem1p regulate Cdc42p-dependent MAPK pathways in the same way? As shown above for basal conditions (Figure 4), all of the domains of Bem1p tested (P208L, P355A, R369A, and K482A) were required for Bem1p to regulate the fMAPK pathway under activating conditions (Figure 5E; for the blot plus controls, see Supplemental Figure S11A). A different pattern was seen in response to salt or pheromone (Figure 5E; Supplemental Figure S11, B and C). Therefore, Bem1p regulates Cdc42p-dependent MAPK pathways by different mechanisms. In line with this idea, bem1 alleles showed different patterns of fMAPK pathway reporter activity (Figure 5F, SD-HIS) and salt sensitivity (Figure 5F, KCl). The requirement of Bem1p alleles in production of P∼Fus3p levels did not match its requirement for shmoo formation (Supplemental Figure S11D), which might be explained because Bem1p has multiple functions in mating that include direction sensing (Moore et al., 2013) and the formation of mating projections (Ito et al., 2001; Butty et al., 2002).

Bem1p is required for viability in the Σ1278b background under some conditions (Dowell et al., 2010). The viability defect of the bem1∆ mutant was rescued by all of the versions of Bem1p tested (Figure 5F, YEP-GAL 25°C). The bem1∆ mutant also had a bud-site-selection defect that was rescued by all of the versions of Bem1p tested (see Figure 4B, Axial %). Versions of Bem1p that rescued the bud-site-selection defect failed to rescue the fMAPK pathway signaling defect (see Figure 4B), which addressed the potential concern that Bem1p regulates the fMAPK pathway by its role in regulating bud-site selection (Basu et al., 2016). In some strains, Bem1p requires Rsr1p for viability, which defines a function for Bem1p in symmetry breaking during polarity establishment (Irazoqui et al., 2003). Bem1p also required Rsr1p for viability in the Σ1278b background (Figure 5F, 5-FOA). Here, a unique pattern of rescue by bem1 alleles was seen. The pattern we saw matched the pattern shown in Irazoqui et al. (2003), except that R369A rescued the viability defect in this study but not in Irazoqui et al. (2003). The fact that bem1 alleles have different phenotypes in symmetry breaking in the two studies might be explained by differences in strain backgrounds. Bem1p is required for viability in some strain backgrounds but not others.

Bem1p was also required for the localization of Sho1p-GFP to the PM (Figure 5G). This was not due to a change in the levels of the Sho1p protein (Figure 5H). Sho1p is an integral membrane tetraspan protein that regulates the fMAPK and HOG pathways (Maeda et al., 1995; O’Rourke and Herskowitz, 1998; Cullen et al., 2004; Tatebayashi et al., 2015). At first, we thought this phenotype might explain how Bem1p regulates the fMAPK pathway. However, all versions of Bem1p tested rescued the localization defect of Sho1p (Supplemental Figure S12A). Like Cdc42p (Zhang et al., 2001; Menon, 2007), Bem1p regulates exocytosis, the polarized delivery of vesicles to the PM (Zajac et al., 2005; France et al., 2006b). Bem1p was not required for regulating the distribution of PIPs at the PM, which might also impact protein localization, based on the localization (Figure 5G) and levels of a PI(4,5)P₂ binding reporter protein (Figure 5H, pGFP-2XPH-PLCδ;  Stefan et al., 2002). The localization defect of Sho1p-GFP in the bem1Δ mutant might result from a defect in exocytosis. Indeed, mutants defective for exocytosis were also defective for the plasma-membrane distribution of Sho1p-GFP (Supplemental Figure S12B). Therefore, Bem1p regulates different effector pathways in different ways. In some settings, Bem1p acts as a multifunctional adaptor where one or more domain(s) is required for function. In other settings, a single domain of Bem1p is required.

Ordering the activation sequence of the Cdc42p module in the fMAPK pathway

Bem4p and Bem1p are adaptor proteins that function at the level of Cdc42p and regulate the fMAPK pathway. As their names suggest (Bud emergence defect), BEM1 (Bender and Pringle, 1991; Chant et al., 1991) and BEM4 (Hirano et al., 1996; Mack et al., 1996) were initially identified as high-copy suppressors of the growth defects of conditional alleles of cdc24 and/or cdc42. Bem1p and Bem4p are not homologues, and the proteins have unrelated amino acid sequences and protein-interaction domains. Loss of either protein results in a defect in the fMAPK pathway (e.g., Figures 2A and 3–5). Thus, Bem1p and Bem4p play nonredundant roles in regulating the fMAPK pathway. In support of this idea, a high-copy plasmid containing BEM1, which rescues the signaling defect of the bem1∆ mutant, did not rescue the fMAPK signaling defect of the bem4Δ mutant (Supplemental Figure S13). Likewise, a high-copy plasmid containing Bem4p, which rescues the signaling defect of the bem4Δ mutant, did not rescue the signaling defect of the bem1Δ mutant (Supplemental Figure S13). Therefore, Bem4p and Bem1p have nonoverlapping functions in regulating the fMAPK pathway.

As adaptor proteins, Bem1p and Bem4p interact with multiple proteins, particularly those that function at the level of the Cdc42p module. Bem1p and Bem4p both interact with Rsr1p, Cdc24p, and Cdc42p (Peterson et al., 1994; Zheng et al., 1995; Hirano et al., 1996; Drees et al., 2001). Rsr1p is the Ras-type GTPase that regulates bud-site selection (Park et al., 1997) and the fMAPK pathway (Basu et al., 2016). Rsr1p can also interact with Cdc42p (Kozminski et al., 2003). To define how Rsr1p, Bem4p, and Bem1p regulate the fMAPK pathway, genetic suppression analysis was used. Genetic suppression analysis allows the ordering of components in a pathway by gain- and loss-of-function alleles. GFP-Msb2p (Adhikari et al., 2015b), Sho1p^P120L (Vadaie et al., 2008a), and Ste11-4p (Stevenson et al., 1992) hyperactivate the fMAPK pathway and were tested for bypass of the signaling defects of the rsr1Δ, bem4Δ, and bem1Δ mutants. Hyperactive versions of Msb2p, Sho1p, and Ste11p partially rescued the fMAPK signaling defect of the rsr1∆ mutant (Figure 6A). This finding is consistent with previous results (Basu et al., 2016), which suggest that Msb2p/Sho1p and Rsr1p converge on the Cdc42p module. For the bem4∆ mutant, a hyperactive version of Msb2p did not bypass the signaling defect, whereas hyperactive versions of Sho1p and Ste11p did (Figure 6B). This result indicates that Bem4p functions between Msb2p and Sho1p in the fMAPK pathway. For the bem1∆ mutant, hyperactive versions of Msb2p and Sho1p did not bypass the signaling defect, whereas a hyperactive version of Ste11p did (Figure 6C). This result indicates that Bem1p functions between Sho1p and Ste11p in the fMAPK pathway. Collectively, the data indicate that Bem4p functions upstream of Bem1p in the fMAPK pathway. This conclusion is probably an oversimplification, because the proteins exist in multi-protein complexes and have multiple functions in regulating the fMAPK pathway. However, this order is supported by the fact that Bem4p associates with the GDP-bound conformation of Cdc42p (Drees et al., 2001; Pitoniak et al., 2015), whereas Bem1p preferentially associates with the GTP-bound conformation of Cdc42p (Butty et al., 1998; Kozminski et al., 2003). Two-hybrid analysis showed that the interaction between Bem1p and GDP-Cdc42p required Rsr1p (Figure 6D, Cdc42 refers to CDC42^{D118A C188S}), whereas the interaction between Bem4p and GDP-Cdc42p did not require Rsr1p or Bem1p (Figure 6E).

FIGURE 6:

Bem1p and Bem4p regulate the fMAPK pathway at different points in the signaling cascade. (A) Wild-type cells (PC538) and the rsr1Δ mutant (PC4256) were examined in YEPD (Glu) or YEP-GAL (Gal) media by immunoblot analysis. The rsr1Δ mutant contained either a control plasmid (pRS316, Ctl) or plasmids harboring hyperactive versions of Msb2p (Msb2*, pGFP-MSB2; Sho1*, pSHO1-P120L-GFP, or Ste11*, pSTE11-4, see Table 2 for details). Cells were grown in YEPD or YEP-GAL media for 5.5 h. See Figure 1B for details. (B) Same as in A, except the bem4Δ mutant (PC3551) was examined. (C) Same as in A, except the bem1Δ mutant (PC6509) was examined. (D) Two-hybrid analysis of Bem1p and Cdc42p in wild-type cells and the indicated mutants. (E) Two-hybrid analysis of Bem4p and Cdc42p in wild-type cells and the indicated mutants. (F) Two-hybrid analysis of the interactions of Rsr1p and Ste4p with full-length Cdc24p and versions that lack the CH (CHΔ) or PH-like and PB1 (PH-likeΔ PB1Δ) domains. See Table 2 for details about plasmids.

The above results led us to ask how the activation of the fMAPK pathway is initiated at the level of Cdc24p. To address this question, we turned to the mating pathway, which is known to recruit Cdc24p by Ste4p (Zhao et al., 1995). By two-hybrid analysis, Ste4p, which regulates the mating pathway and binds the CH domain of Cdc24p, bound to full-length Cdc24p, whereas Rsr1p, which regulates the fMAPK pathway and also binds the CH domain of Cdc24p (Shimada et al., 2004), did not (Figure 6F). Rather, Rsr1p interacted with a version of Cdc24p lacking an auto-inhibitory domain (PH-likeΔ-PB1Δ). Given that Bem4p binds to the PH-like domain of Cdc24p to relieve auto-inhibition, it is possible that Ste4p initiates recruitment of Cdc24p in the mating pathway, whereas in the fMAPK pathway, Bem4p may bind Cdc24p to relieve auto-inhibition (Pitoniak et al., 2015), which permits subsequent binding by Rsr1p. Bem1p binds the PB1 domain of Cdc24p and may also initiate recruitment of Cdc24p in the fMAPK pathway. However, Bem1p acts at a later step in the pathway than Bem4p and does not require binding to Cdc24p to activate the fMAPK pathway under some conditions, which argues against this possibility.

Exploring the roles of Bem4p and Bem1p on fMAPK pathway kinetics

Given that the fMAPK pathway has a unique induction profile, we became interested in understanding the relative contribution of the Bem adaptors in regulating the kinetics of fMAPK pathway signaling. Phosphorylation of the MAP kinase Kss1p was examined in cells carrying the Cdc42^E100A allele, and in the bem1Δ and bem4Δ mutants over the time. Similar to the results shown in Figure 1, the phosphorylation of Kss1p was induced by Gal after 1 h and peaked around 6 h (Figure 7). The Cdc42^E100A allele showed a similar activation profile but levels of P∼Kss1p were reduced compared with wild type (Figure 7A). This result indicates that this residue of Cdc42p is involved in apmlifying the induction of the fMAPK pathway. E100 is in a region of Cdc42p that does not undergo a conformation change, suggesting that this is not due to a defective interaction with the effector PAK Ste20p. A similar pattern was observed in cells lacking Bem4p (Figure 7B), except that the levels of P∼Kss1p were even lower than seen in cells with Cdc42^E100A. These data support the idea that the signaling defect of Cdc42^E100A is through a defect in interaction with Bem4p. In cells lacking Bem1p, no induction of P∼Kss1p was observed (Figure 7C, compare 3- and 6-h time points), even when P∼Kss1p levels were monitored for extended periods (8 h). This result indicates that Bem1p is critical for induction of the fMAPK pathway, since the lack of this protein completely ablated pathway activity under this condition. These results support the idea that Bem4p and Bem1p regulate the fMAPK pathway by different mechanisms.

FIGURE 7:

Time-course phosphoblot analysis of P∼Kss1 induction in response to growth in GAL. (A) Wild-type cells (PC6810) were compared with the cdc42^E100A mutant (Supplemental Table S1). Cells were grown in YEP-GAL media to induce fMAPK activity at the time points indicated. See Figure 1B for details. (B) Wild-type cells (PC538) were compared with the bem4Δ mutant (PC3551). Cells were grown as described in A. See Figure 1B for details. (C) Wild-type cells (PC6810) were compared with the bem1Δ mutant (PC6680). Cells were grown as described in A. See Figure 1B for details.

DISCUSSION

Rho GTPases are key regulators of cell polarity and signal transduction in eukaryotes. Many proteins interact with Rho GTPases and impact their functions and activities, including adaptor proteins that link GTPases to their activators and effectors. By examining two adaptors of Cdc42p in yeast, Bem4p and Bem1p, we further define how Cdc42p regulates one of the many pathways in which it functions. Bem1p and Bem4p, as well as Rsr1p, the GTPase that regulates bud-site-selection and the fMAPK pathway, exhibit some degree of pathway selectivity. Bem4p and Rsr1p regulate the fMAPK pathway but not the mating or HOG pathways, while Bem1p regulates the fMAPK pathway by a specific mechanism.

Cdc42p-dependent MAPK pathways show different kinetic profiles

The way that signaling pathways are activated can shape the output response, as well as signal specificity. MAPK pathway dynamics can be complex, involving fast and slow responses (English et al., 2015), oscillations (Hilioti et al., 2008), and feedback control (Madhani and Fink, 1997; Breitkreutz et al., 2001). Moreover, activation of one pathway before another can stimulate or attenuate the activation of the other pathway (Baltanas et al., 2013). It is also known that filamentation morphogenetic responses do not always correlate with MAP-kinase cascade-dependent transcriptional activation (Mosch et al., 1999). By directly comparing Cdc42p-dependent MAPK pathways in yeast, we show that the MAPK pathways have different activation profiles. The mating and HOG pathways are rapidly induced, whereas the fMAPK pathway has slow activation kinetics. The slow activation kinetics of the fMAPK pathway might mean that gene expression is required for MAPK pathway activation. Given that the fMAPK pathway is induced under nutrient-limiting conditions, one possibility is that derepression of glucose-repressed genes may be required for fMAPK function (Carlson, 1999). In a related study from our lab (Prabhakar et al., 2019), we show that the fMAPK pathway is regulated throughout the cell cycle and peaks in G₂/M phase. Cell synchronization experiments also showed that the fMAPK pathway is activated (in G₂/M) to the same levels as the mating and HOG pathways. Future studies will be required to determine whether the different activation kinetics of Cdc42p-dependent MAPK pathways contributes to pathway-specific responses.

Interaction between Bem4p and Cdc42p is critical for fMAPK pathway activity

One challenge in studying scaffold-type adaptors that interact with multiple proteins is to define which protein interactions are critical for pathway activity. Bem4p interacts with Cdc24p, Cdc42p, Ste11p, and Kss1p (Pitoniak et al., 2015), yet the critical interactions relevant for Bem4p function have not been explored. Here, we identify a version of Cdc42p (E100A) that is defective for interaction with Bem4p and the fMAPK pathway. Similarly, specific amino acid residues on Bem4p were identified that are defective for interaction with Cdc42p and fMAPK pathway activity. Together these results define the interaction between Cdc42p and Bem4p as being critical for fMAPK pathway activity. This may be the first example in yeast where a protein binds directly to Cdc42p to impart pathway selectivity.

A model for Cdc42p regulation in the fMAPK pathway

We also show that the major polarity adaptor of Cdc42p, Bem1p, regulates the fMAPK pathway. Bem1p is required for the localization of Ste20p, the effector of Cdc42p in the fMAPK pathway, to the PM. Bem1p regulates the mating pathway through a similar mechanism (Yamaguchi et al., 2007). This conclusion addresses a longstanding problem of how Ste20p is recruited to the PM during filamentous growth. Recently, Bem1p has been shown to interact with Msb2p in the HOG pathway (Tanaka et al., 2014). Like Bem1p, Msb2p preferentially associates with the GTP-bound conformation of Cdc42p (Cullen et al., 2004). Therefore, Bem1p may function in a protein complex that connects Msb2p, the Cdc42p module, and Ste20p.

Four proteins interact with the Cdc42p module and regulate the fMAPK pathway: Msb2p (Cullen et al., 2004), Bem4p (Pitoniak et al., 2015), Rsr1p (Basu et al., 2016), and Bem1p (this study). Each of these proteins has a nonredundant function in the fMAPK pathway. A current challenge is to define the specific functions for each of these proteins in regulating the GTPase module. We have previously shown that Msb2p associates with the GTP-bound conformation of Cdc42p (Cullen et al., 2004) and Rsr1p regulates the fMAPK pathway by interacting with the CH domain of Cdc24p (Basu et al., 2016). Here, we show that Bem4p interacts with Cdc42p to regulate the fMAPK pathway, and Bem1p is required for recruitment of Ste20p to the PM to GTP-Cdc42p. Collectively, these findings may begin to explain how Cdc42p is directed to the fMAPK pathway (Figure 8). Bem4p and Bem1p may affect each other’s localization and/or stability, and future experiments are needed to explore this issue.

FIGURE 8:

Model for proteins that regulate the Cdc42p module in the fMAPK pathway. Bem4p interacts with the PH-like domain of Cdc24p (Pitoniak et al., 2015) and the α3 helix of Cdc42p (this study) and may initiate recruitment of Cdc24p to the fMAPK pathway. Rsr1p interacts with the CH domain of Cdc24p to regulate the fMAPK pathway (Basu et al., 2016), and Bem1p interacts with the PB1 domain of Cdc24p and with Ste20p to regulate the fMAPK pathway (this study). Each protein is associated with the PM: Rsr1p and Cdc42p by lipid modification (not shown for Cdc42p), Bem1p by the PX domain, and Bem4p in an unknown manner (question mark). Black arrows refer to the order for recruitment and activation of the GTPase module into the fMAPK pathway, which is based on genetic suppression analysis.

One way to order proteins in a pathway is by genetic suppression analysis. Genetic suppression analysis using hyperactive versions of fMAPK pathway components showed that Bem4p functions above Bem1p in the fMAPK pathway. This conclusion is consistent with the fact that Bem4p can interact with the inactive (GDP-bound) conformation of Cdc42p, whereas Bem1p primarily interacts with the active (GTP-bound) conformation of the protein. Our results further indicate that Rsr1p does not initiate Cdc24p recruitment to the pathway, because it does not interact with the full-length version of Cdc24p. Bem4p or Bem1p may initiate the recruitment of Cdc24p; however, because the interaction between Bem1p and Cdc24p is not necessary under all conditions, we favor the possibility that Bem4p may initiate the recruitment of Cdc24p to the fMAPK pathway.

It will be interesting to define whether the mechanism surrounding Cdc42p regulation in the fMAPK pathway postulated here is conserved among fungal species. Experiments were performed in a haploid S. cerevisiae of the Σ1278b background. Future work will be needed to determine how broadly our findings overlap with other cell types and systems. It has previously been shown that Cdc42p and its regulators, Rsr1p and Bem1p, contribute to hyphal formation and virulence in the major human fungal pathogen C. albicans (Bassilana et al., 2003). Perhaps Rsr1p, Bem1p, and Bem4p regulate the orthologous MAP kinase pathway in C. albicans and other pathogens.

Comparison of Cdc42p regulation in the mating and fMAPK pathways

Proteins that regulate Cdc42p in the fMAPK pathway characterized in this study can be compared with proteins that regulate Cdc42p in the mating pathway, which has been characterized previously (Sprague et al., 1983, 1991; Errede et al., 1995). A potentially analogous activation mechanism comes from comparing Ste4p, the ß-subunit of the heterotrimeric G-protein in the mating pathway, to Rsr1p in the fMAPK pathway. Both proteins interact with the CH domain of Cdc24p (Zhao et al., 1995; Basu et al., 2016), and both proteins are tied to the PM. Ste4p is anchored to the PM by Ste18p, which is a palmitoylated and farnesylated protein (Zhao et al., 1995; Butty et al., 1998; Nern and Arkowitz, 1998, 1999; Pryciak and Huntress, 1998; Wiget et al., 2004), and Rsr1p is modified by a lipid moiety (Park et al., 2002). Thus, Cdc24p may be activated by Ste4p in the mating pathway and by Rsr1p in the fMAPK pathway. Ste4p can bind to full-length Cdc24p, perhaps to initiate signaling, whereas Rsr1p binds to versions of Cdc24p that lack auto-inhibitory domains. During mating, Ste4p may be critical for initiating recruitment of Cdc24p into the mating pathway, whereas during filamentous growth, Rsr1p is not likely to initiate Cdc24p recruitment to the fMAPK pathway.

During mating, the interaction between Ste4p and Cdc24p is stabilized by Far1p, an adaptor protein in the mating pathway that is also required for cell-cycle arrest in response to pheromone (Simon et al., 1995; Zhao et al., 1995; Nern and Arkowitz, 1998). Bem4p may play an analogous role in the fMAPK pathway. Bem4p interacts with Rsr1p (Drees et al., 2001) and Cdc24p (Chavel et al., 2014) and might stabilize the Rsr1p–Cdc24p interaction. Bem4p also interacts with Cdc42p to regulate the fMAPK pathway. Finally, Bem1p regulates the mating and fMAPK pathways by plasma-­membrane recruitment of Ste20p (Figure 8), although this may be an oversimplification, as there are differences in the ways that Bem1p regulates the two pathways. Therefore, the mating and fMAPK pathways utilize different proteins to regulate the same GTPase module.

Bem1p regulates different pathways in different ways

We also show that Bem1p regulates different effector pathways in different ways. Bem1p regulates multiple Cdc42p-dependent processes, including polarity establishment, exocytosis, and MAPK signaling. Remarkably, different domains of Bem1p play different roles in regulating these different pathways. In some pathways, Bem1p acted as a multifunctional adaptor because multiple domains were required for Bem1p function. In other pathways, only one or none of the functional domains of Bem1p were required. Two general conclusions can be reached from these observations. First, context is critically important for studying adaptor protein function. Assessing Bem1p function was sensitive to the particular strain, assay, and pathway being tested. The second conclusion is that adaptors may operate in different ways in different contexts.

MATERIALS AND METHODS

Media and growth conditions

Experiments were performed at 30°C unless otherwise indicated. YEPD (2% glucose [D]), YEP-GAL (2% GAL), and synthetic complete media containing yeast nitrogen and 2% glucose (SD) or 2% galactose (S-GAL) were used. Amino acids were added to the synthetic media as required. For some experiments, 0.2% glucose was used as indicated.

Strains and plasmids

Strains are listed in Table 1 and plasmids in Table 2. Escherichia coli and S. cerevisiae strains were manipulated using standard methods (Rose et al., 1990). Gene disruptions and GAL1 promoter fusions were made by PCR-based methods (Longtine et al., 1998). Some gene deletions were constructed using cassettes that contained antibiotic resistance markers (Goldstein et al., 1999). Gene disruptions were confirmed by PCR Southern analysis and phenotype.

TABLE 1:

Yeast strains used in the study.

Name	Genotypea,b	Reference
PC313	MATa ura3-52	Liu et al., 1993
PC344	MATa ura3-52/MATa ura3-52	Cullen and Sprague, 2000a
PC538	MATa ste4 FUS1-lacZ FUS1-HIS3 ura3-52	Cullen et al., 2004
PC859c	MATa ura3-52 ura3- his3	Cullen and Sprague, 2002
PC1660	MATa ura3-52 sec10-2	Grote et al., 2000
PC1661	MATa ura3-52 sec15-1	Grote et al., 2000
PC2738	MATa ste4 FUS1-lacZ FUS1-HIS3 ura3-52 bem4::URA3	Pitoniak et al., 2015
PC3861	MATa ste4 FUS1-lacZ FUS1-HIS3 ura3-52 ste11::NAT	Karunanithi et al., 2010
PC4256	MATa ste4 FUS1-lacZ FUS1-HIS3 ura3-52 rsr1::NAT	Basu et al., 2016
PC4988d	PJ69-4a MATa trpl-901 leu2-3,112 ura3-52 his3-200 ga14A ga18OA LYS2::GALl-HIS3 GAL2-ADE2 met2::GAL7-lacZ	James et al., 1996
PC5024	MATa ura3-52 ste11::NAT	Pitoniak et al., 2015
PC6472	MATa ste4 FUS1-lacZ FUS1-HIS3 ura3-52 cyk3::KlURA	This study
PC6509	MATa ura3-52 leu2 bem1::NAT pRS316 BEM1	This study
PC6539	MATa ste4 FUS1-lacZ FUS1-HIS3 ura3-52 cdc42::NAT pRS316-GFP-linker-CDC42	This study
PC6545	MATa ste4 FUS1-lacZ FUS1-HIS3 ura3-52 cdc42::NAT pRS316-GFP-linker-CDC42 rga1::HYG	This study
PC6548	MATa ste4 FUS1-lacZ FUS1-HIS3 ura3-52 cdc42::NAT pRS316-GFP-linker-CDC42 rsr1::HYG	This study
PC6555	MATa ste4 FUS1-lacZ FUS1-HIS3 ura3-52 ssk1 cdc42:: HYG pRS316-GFP-linker-CDC42	This study
PC6557	MATa ura3-52 cdc42::HYG pRS316-GFP-linker-CDC42	This study
PC6591	MATa ura3-5 leu2	This study
PC6810	MATa ura3-52 leu2 ssk1	This study
PC6604	MATa ura3-52 leu2 ssk1 ste11::NAT	This study
PC6614	MATa ura3-5 leu2 ahk1::NAT	This study
PC6622	MATa ura3-52 cdc42:: HYG pRS316-GFP-linker-CDC42 rdi1::NAT	This study
PC6624	MATa ura3-52 leu2 ssk1 cla4::NAT	This study
PC6630	MATa ura3-52 GAL rga1::NAT	This study
PC6632	MATa ura3-52 cdc42:: HYG pRS316-GFP-linker-CDC42 rga1::NAT	This study
PC6634	MATa ura3-52 GAL-RDI1:: KanMX6	This study
PC6636	MATa ura3-52 cdc42:: HYG pRS316-GFP-linker-CDC42 GAL RDI1:: KanMX6	This study
PC6680	MATa ura3-52 ssk1 bem1::NAT pRS316-BEM1	This study
PC6685	MATa ura3-52 ssk1 cdc42::NAT pGFP-Cdc42	This study
PC6701	MATa ura3-52 leu2 ssk1 GAL RDI1:: KanMX6	This study
PC6724	MATa ura3-52 leu2 ssk1 HOG1-GFP	This study
PC6726	MATa ura3-52 leu2 ssk1 KSS1-GFP	This study
PC6729	MATa ura3-52 leu2 ssk1 FUS3-GFP	This study
PC6883d	PJ69-4a MATa trpl-901 leu2-3,112 ura3-52 his3-200 ga14A ga18OA LYS2::GALl-HIS3 GAL2-ADE2 met2::GAL7-lacZ bem4::KlURA3	This study
PC7064	MATa ura3-52 leu2 ssk1 bem4::KlURA3	This study
PC7085d	PJ69-4a MATa trpl-901 leu2-3,112 ura3-52 his3-200 ga14A ga18O LYS2::GALl-HIS3 GAL2-ADE2 met2::GAL7-lacZ rsr1:: KlURA3	This study
PC7159d	PJ69-4a MATa trpl-901 leu2-3,112 ura3-52 his3-200 ga14A ga18O LYS2::GALl-HIS3 GAL2-ADE2 met2::GAL7-lacZ bem1:: KlURA3	This study
PC7214	MATa ste4 FUS1-lacZ FUS1-HIS3 ura3-52 BEM4-HA K323A 328A:: KanMX6	This study
PC7343	MATa ura3-52 leu2 ssk1 bem1 rsr1::KanMX6 pRs315-BEM1	This study

^aAll strains are in the Σ1278b background unless otherwise indicated.

^bStrains containing 26 alleles were also constructed in the Σ1278b ssk1 background (Supplemental Table S1).

^cStrains designated ura3- were generated by selection for resistance to 5-FOA.

^dStrain from M. Johnston lab, University of Colorado Anschutz Medical Campus.

TABLE 2:

Plasmids used in the study. Continued

Name	Description	Reference
PC1405	pFRE-lacZ	Madhani et al., 1997
PC1422	pRS315	Sikorski and Hieter, 1989
PC1441	YCp50-STE11-4	Stevenson et al. 1992
PC1601	pRS316-SHO1-GFP	Marles et al., 2004
PC1614	pRS316-SHO1(D16H)-GFP	Marles et al., 2004
PC1696	pGFP-MSB2	Vadaie et al., 2008b
PC1715	pRS316-SHO1(P120L)-GFP	Vadaie et al., 2008b
PC1882	pRS316-GAL-SHO1(D16H)-GFP::KanMX6	Vadaie et al., 2008b
PC2205	pNAT	Goldstein and McCusker, 1999
PC2206	pHYG	Goldstein and McCusker, 1999
PC2207	pRS316	Sikorski and Hieter, 1989
PC2560	pRS426 GFP-2XPH-PLC	Stefan et al., 2002
PC4190	pOAD-BEM4	McCraith et al., 2000
PC4205	pOAD	McCraith et al., 2000
PC4206	pOBD-2	McCraith et al., 2000
PC4232	pOBD-2 CDC42(D57Y, C188S)	Pitoniak et al., 2015
PC4394	pRL116 CEN/URA3 GFP-Ste20	Leberer et al., 1997
PC4590	pOAD-RSR1	This study
PC4878	pV84-FUS1-lacZ	This study
PC6025	pYM111 pRS416 8XCRE-LacZ Amp Ura	Yamamoto et al., 2010
PC6365	pYEp352-TEF2	52
PC6367	pYEp352-TEF2-BEM4-GFP	Pitoniak et al., 2015
PC6455	pRS306-GFP-linker-CDC42 (pDLB3609)	Irazoqui et al., 2003
PC6454	pRS316-GFP-linker-CDC42	This study
PC6514	BEM1p-BEM1-12XMYC-SWE1t (pDLB2374)	Irazoqui et al., 2003
PC6515	BEM1 p-bem1P208L-12XMYC-SWE1t (pDLB2375)	Irazoqui et al., 2003
PC6516	BEM1 p-bem1P355A-12XMYC-SWE1t (pDLB2377)	Irazoqui et al., 2003
PC6517	BEM1p-bem1R369A-12XMYC-SWE1t (pDLB2378)	Irazoqui et al., 2003
PC6518	BEM1p-bem1K482A-12XMYC-SWE1t (pDLB2379)	Irazoqui et al., 2003
PC6519	pRS316-BEM1	D. Lew
PC6892	2μ SEC4: URA3; pNB142	France et al., 2006a
PC6893	2μ SEC15: URA3; pNB192	France et al., 2006a
PC6894	EXO70, 2u, URA3	France et al., 2006a
PC6797	pOBD-2-CDC42(K5A, D57Y, C188S)	This study
PC6798	pOBD-2-CDC42(R144A, D57Y, C188S)	This study
PC6799	pOBD-2-CDC42(E100A, D57Y, C188S)	This study
PC6800	pOBD-2-CDC42(V36T, D57Y, C188S)	This study
PC6802	pOBD-2-CDC42(H102A, H103A, H104A D57Y, C188S)	This study
PC6805	pOBD2-CDC42(F37Y, D57Y, C188S)	This study
PC6831	pOAD-BEM1	This study
PC6833	pOBD-2 CDC42(C188S)	This study
PC6834	pOBD-2 CDC42(C188S, G12V)	This study
PC6835	pOBD-2 CDC42(C188S, T17N)	This study
PC6836	pOBD-2 CDC42(C188S, D118A)	This study
PC7025	pRS316-HIS-CDC42	This study
PC7028	pYIp352-TEF2-RDI1	This study
PC7096	pOAD-BEM4(K323A, K328A, K351A, R352A)	This study
PC7097	pOAD-BEM4(K351A, R352A)	This study
PC7112	pOAD-BEM4(K323A, K328A)	This study
PC7160	pOAD-BEM4(K323A)	This study
PC7161	pOAD-BEM4(K328A)	This study
PC7162	pOAD-BEM4(K328E)	This study
PC7163	pOBD-CDC42(E100K, D57Y, C188S)	This study

Plasmids pRS315 and pRS316 have been described (Sikorski and Hieter, 1989). Plasmid pRL116 CEN/URA4 pGFP-Ste20 has been described and was provided by P. Pryciak (Leberer et al., 1997). Plasmids carrying BEM1p-BEM1-12XMYC-SWE1t (pDLB2374), BEM1p-bem1^P208L-12XMYC-SWE1t (pDLB2375), BEM1p-bem1^P355A-12XMYC-SWE1t (pDLB2377), BEM1p-bem1^R369A-12XMYC-SWE1t (pDLB2378), and BEM1p-bem1^K482A-12XMYC-SWE1t (pDLB2379) have been described (Irazoqui et al., 2003). These plasmids and pRS316-BEM1 were generously provided by Daniel Lew at Duke University School of Medicine.

To generate PC6810, the LEU2 gene was disrupted by NAT in PC313 by homologous recombination. The NAT gene was replaced with URA3 cassette (HA-URA3-HA). The strain was treated on 5-FOA to force out the URA3 selection marker. In this strain, the SSK1 gene was disrupted using the NAT gene, which was again replaced with URA3 cassette. The resulting strain PC6810 is a derivative of PC313, which is MATa ura3-52 carrying unmarked ssk1 and leu2 gene deletions.

To generate pRS316-GFP-Linker-Cdc42p, the insert was subcloned from pDLB3609 (Irazoqui et al., 2003, also shared by D. Lew) using EcoR1 (NEB Cat #R0101, Ipswich, MA) and Sal1 (NEB Cat #R3138, Ipswich, MA). To generate pRS315-GFP-Linker-Cdc42p, the insert was subcloned from pDLB 3609 (Irazoqui et al., 2003) using PstI (NEB Cat #R3140, Ipswich, MA) and SalI (NEB Cat #R3138, Ipswich, MA). To generate pYEp352-TEF2-RDI1 (PC7028), the RDI1 gene was amplified from the genomic DNA cloned in plasmid carrying TEF2 promoter using Xba1 (NEB Cat #R0145, Ipswich, MA) and Sal1 (NEB Cat #R3138, Ipswich, MA). To create the Bem4p alleles for two-hybrid analysis PC4190 with pOAD BEM4 was utilized. pV84-FUS1-lacZ was constructed in plasmid V84 by homologous recombination.

Functional tests to evaluate budding pattern and MAP kinase activity

The budding pattern was determined as described (Basu et al., 2016). Budding pattern was determined by bud position and bud scar position to the proximal, equatorial, or distal parts of the cell. For some experiments, cells were grown to mid-log phase and resuspended in 1 ml of water. Cells were stained with Fluorescent brightener 28 (Calcofluor White [CFW], Sigma-Aldrich Life Science and Biochemicals, St. Louis, MO) using a final concentration of 0.01% for 10 min at 25°C. Cells were washed one time in water and examined by microscopy.

The PWA was performed as described (Roberts and Fink, 1994). The single cell invasive growth assay was performed as described (Cullen and Sprague, 2000b). To evaluate the HOG pathway, cells were spotted onto media supplemented with 1 M KCl. Halo assays were performed as described (Sprague et al., 1983). Cells were grown for 16 h to saturation. Cell density was measured by adding 20 μl overnight culture to 180 μl water (OD_600nm ∼ 0.2). Approximately 5 μl overnight culture was added to 300 μl water and top spread on agar plates; 3 μl (1.8 µM) and 10 μl (6 µM) alpha factor (1 mg/ml) was added on the dried plates and incubated at 30°C for 2 d. ImageJ analysis was used to quantitate colony growth or the degree of agar invasion by measuring signal intensity compared with background.

Experiments to evaluate the morphogenetic response to pheromone (shmoos) were performed by microscopy. The FUS1-HIS3 reporter in cells lacking STE4 was used to evaluate the activity of the fMAPK pathway (Cullen et al., 2004; Chavel et al., 2010; Adhikari et al., 2015b; Pitoniak et al., 2015; Basu et al., 2016) and was measured by spotting cells onto SD-HIS or SD-HIS medium with 3-amino-1,2,4 triazole. Tests for viability were performed by plasmid loss experiments of the URA3 gene on 5-FOA (Boeke et al., 1984).

Construction and evaluation of a collection of CDC42 alleles in the Σ1278b background

A collection of CDC42 alleles that was previously generated (Kozminski et al., 2000) was introduced into a filamentous (Σ1278b) strain lacking SSK1 (PC6810). Plasmids from E. coli strains harboring CDC42 alleles linked to the LEU2 selection marker and flanking sequence required for homologous recombination were digested with BanII (NEB Cat #R0119S, Ipswich, MA) and Xba1 (NEB#R0145S, Ipswich, MA) and transformed into yeast lacking a genomic copy of CDC42 (cdc42::NAT) and containing pRS316 GFP-CDC42 (PC6454). Selection for LEU2+/NAT– colonies favored integration of the alleles at the CDC42 locus. Selection on 5-FOA was used to force out the pGFP-CDC42 plasmid and determine whether the alleles were viable. To confirm the DNA sequence of each allele, genomic DNA was extracted. The CDC42 gene and flanking region was amplified by PCR. PCR products were purified and the entire CDC42 gene was evaluated by DNA sequencing analysis (Roswell Park Cancer Research Center, Buffalo, NY). In addition to the mutations mentioned in Supplemental Table S1, the library also carries a polymorphic variation at A190T. Twenty-four of 40 alleles were obtained by this method that were viable without the cover plasmid (pRS316-GFP-CDC42). Seven alleles were unable to lose the pGFP-CDC42 plasmid, indicating that they produced versions of Cdc42p that compromised its essential function, and one allele was not obtained. Introduction of the wild-type version of CDC42 resulted in an invasive growth defect that was taken into account when examining the other alleles.

β-Galactosidase assays

Wild-type cells (PC313), or cells lacking the redundant branch of the HOG pathway (ssk1∆ PC6810), were transformed with plasmids pFRE-lacZ/URA3 2µ AMP (PC1405) provided by H. Madhani (Madhani et al., 1997), pRS416 8XCRE-LacZ AMP URA3 (pYM111 or PC6025) provided by Haruo Saito (Tatebayashi et al., 2006), or pV84-FUS1-lacZ (PC4878, this study). Cells harboring plasmids were grown to saturation at 30°C in SD-URA medium to maintain selection for the plasmids. Cells in mid-log phase in YEPD were collected, washed three times in distilled water, and added to YEP-GAL media, YEPD medium supplemented with 0.4 M KCl, or YEPD medium supplemented with 1.2 µM alpha-factor (5 µl of 1 mg/ml α-factor [0.6 µM] in 5 ml) or 6 µM alpha-factor unless specified. One milliliter of culture aliquots was collected at the indicated time points, assessed for cell density at A₆₀₀, and stored at –80°C.

β-Galactosidase assays were performed as described (Cullen et al., 2004). Cells were resuspended in 100 µl Z-buffer (44.32 ml H₂O with 5 ml phosphate buffer [0.6 M Na₂HPO₄ + 0.4 M NaH₂PO₄], 0.5 ml 1 M KCl, 50 µl 1 M MgSO₄, 135 µl ß-mercaptoethanol) containing 2 µl of 5% sarkosyl and 2 µl toluene. Tubes were incubated at 37°C for 30 min. Z-buffer (650 µl) containing ortho-nitrophenyl-β-galactoside was added to each tube. Reactions were stopped with 250 µl of 1 M Na₂CO₃, and the time was recorded. Cell extracts were removed by centrifugation (13,000 rpm for 3 min), and 200 µl of the supernatant was used to determine optical density at A₄₂₀. Miller units were calculated as 1000 × A₄₂₀/(A₆₀₀ X time). Experiments were performed in three independent replicates. The two-tailed paired Student’s t test was used to measure the statistical significance between samples from multiple trials using ProStat (Poly Software International).

Phosphoblot and immunoblot analysis

Cells were induced for immunoblots according to standard conditions and protocols. For basal conditions, cells grown for 16 h (750 µl) were inoculated into fresh YEPD media (15 ml) and incubated for 3 h to mid-log phase. For inductions, mid-log phase cultures were harvested, washed three times with water, and resuspended in YEP-Gal (15 ml), YEPD + 0.4 M KCl (15 ml), or YEPD + α-factor (15 ml) media over a designated time series.

Immunoblots were performed as described (Basu et al., 2016) with the following exceptions in antibody dilution and blocking buffer. The p44/p42 antibodies (Cell Signaling Technology, Danvers, MA; Cat #4370) and Phospho p38 antibodies (Cell Signaling Technology, Danvers MA; #9211) were used to detect P∼MAP kinases and diluted 1:10,000 in 5% bovine serum albumin (BSA). Anti-Kss1p antibodies (Santa Cruz Biotechnology, Santa Cruz, CA; Cat #6775) and anti-Hog1p antibodies (Santa Cruz Biotechnology, Santa Cruz, CA; #yC-20) were used at dilutions 1:10000 and 1:5000, respectively, in 5% nonfat dried milk. Anti-Fus3p antiserum (Santa Cruz Biotechnology, Santa Cruz, CA; #6773) was used at a 1:10,000 dilution in 5% BSA. Anti-Pgk1p antibodies (Life Technologies; Camarillo, CA; Cat #459250) were used at 1:10,000 dilution. Secondary antibodies goat anti-mouse IgG–HRP (Bio-Rad Laboratories, Hercules, CA; Cat #170-6516), goat anti-rabbit IgG-HRP (Jackson ImmunoResearch Laboratories, West Grove, PA; Cat #111-035-144), and donkey anti-goat IgG-HRP (Santa Cruz Biotechnology, Santa Cruz, CA; Cat #sc-2020) were used. Quantitation of band intensities for immunoblot analysis was performed with Image Lab Software (Bio-Rad) at an exposure with no saturated pixel in any lane. Background subtraction was performed according to the software user guide provided by the manufacturer. Band intensities of phosphoproteins were measured against total protein levels based on Pgk1p band intensity.

Site-directed mutagenesis

To insert point mutations in plasmids carrying the CDC42 or BEM4 genes, the GeneArt Site-Directed Mutagenesis Kit was used (Thermo Fisher Cat #A13282, Grand Island, NY) according to the manufacturer’s protocol. Briefly, an overlapping ∼40 nucleotide primer set was designed with the desired point mutations designed in the center of the primers. The template was amplified using AccuPrime Pfx DNA polymerase (Thermo Fisher Cat #12344024, Grand Island, NY). The template was methylated by adding DNA methylase and S-adenosyl methionine in the PCR, which was provided by the manufacturer. Following PCR, the free ends of the PCR product were recombined in vitro by the addition of recombinase enzyme mix. The recombined PCR product was transformed into E. coli cells (One Shot MAX Efficiency DH5α-T1^R). The specially designed host E. coli strain that contains McrBc endonuclease digested the methylated template, leaving the nonmethylated PCR amplified mutated product intact.

Differential interference contrast (DIC) and fluorescence microscopy

DIC and fluorescence microscopy of cells were performed using DIC and fluorescence filter sets on an Axioplan 2 fluorescence microscope (Zeiss) with a PLAN-APOCHROMAT 100×/1.4 (oil) objective (N.A. 0.17) (Zeiss). For most experiments, proteins were visualized by resuspending cells in water at 25°C. Digital images were obtained with the Axiocam MRm camera (Zeiss). Axiovision 4.4 software (Zeiss) was used for image acquisition and analysis.

Protein modeling

To visualize the putative structure of Bem4p, two web-based protein fold recognition and function prediction servers were utilized: Phyre2 (www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index) and iTASSER (https://zhanglab.ccmb.med.umich.edu/I-TASSER/). Phyre2 is designed to predict structure based on primary amino acid sequence (Kelley et al., 2015) and was used to model the structure of yeast Cdc42p and alleles using information from the crystal structure of Homo sapiens Cdc42p in Protein Data Bank (PDB; www.rcsb.org/pdb/home/home.do). The protein-docking server Haddock (van Zundert et al., 2016) was also used and is publicly available (http://milou.science.uu.nl/services/HADDOCK2.2/).

Abbreviations used:

ARM

Armadillo

BSA

bovine serum albumin

CFW

Calcofluor White

D

dextrose

DIC

differential-interference contrast

EFF

effector-binding domain

5-FOA

5-fluoroorotic acid

GAL

galactose

GEF

guanine nucleotide exchange factor

GFP

green fluorescent protein

HOG

high osmolarity glycerol response

MAPK

mitogen-activated protein kinase

PAK

p21-activated kinase

PDB

Protein Data Bank

PIP

phosphatidylinositol phosphate

PM

plasma membrane

PWA

plate-washing assay

Rho

Ras homology

YEP

yeast extract and peptone

YEPD

yeast extract, peptone, and dextrose.