Localized feedback phosphorylation of Ste5p scaffold by associated MAPK cascade.

Scaffold proteins play pivotal roles during signal transduction. In Saccharomyces cerevisiae, the Ste5p scaffold protein is required for activation of the mating MAPK cascade in response to mating pheromone and assembles a G protein-MAPK cascade complex at the plasma membrane. To serve this function, Ste5p undergoes a regulated localization event involving nuclear shuttling and recruitment to the cell cortex. Here, we show that Ste5p is also subject to two types of phosphorylation and increases in abundance as a result of MAPK activation. During vegetative growth, Ste5p is basally phosphorylated through a process regulated by the CDK Cdc28p. During mating pheromone signaling, Ste5p undergoes increased phosphorylation by the mating MAPK cascade. Multiple kinases of the mating MAPK cascade contribute to pheromone-induced phosphorylation of Ste5p, with the mating MAPKs contributing the most. Pheromone induction or overexpression of the Ste4p Gbeta subunit increases the abundance of Ste5p at a post-translational step, as long as the mating MAPKs are present. Increasing the level of MAPK activation increases the amount of Ste5p at the cell cortex. Analysis of Ste5p localization mutants reveals a strict requirement for Ste5p recruitment to the plasma membrane for the pheromone-induced phosphorylation. These results suggest that the pool of Ste5p that is recruited to the plasma membrane selectively undergoes feedback phosphorylation by the associated MAPKs, leading to an increased pool of Ste5p at the site of polarized growth. These findings provide evidence of a spatially regulated mechanism for post-activation control of a signaling scaffold that potentiates pathway activation.

Scaffold proteins play pivotal roles during signal transduction. In Saccharomyces cerevisiae, the Ste5p scaffold protein is required for activation of the mating MAPK cascade in response to mating pheromone and assembles a G protein-MAPK cascade complex at the plasma membrane. To serve this function, Ste5p undergoes a regulated localization event involving nuclear shuttling and recruitment to the cell cortex. Here, we show that Ste5p is also subject to two types of phosphorylation and increases in abundance as a result of MAPK activation. During vegetative growth, Ste5p is basally phosphorylated through a process regulated by the CDK Cdc28p. During mating pheromone signaling, Ste5p undergoes increased phosphorylation by the mating MAPK cascade. Multiple kinases of the mating MAPK cascade contribute to pheromone-induced phosphorylation of Ste5p, with the mating MAPKs contributing the most. Pheromone induction or overexpression of the Ste4p G␤ subunit increases the abundance of Ste5p at a post-translational step, as long as the mating MAPKs are present. Increasing the level of MAPK activation increases the amount of Ste5p at the cell cortex. Analysis of Ste5p localization mutants reveals a strict requirement for Ste5p recruitment to the plasma membrane for the pheromone-induced phosphorylation. These results suggest that the pool of Ste5p that is recruited to the plasma membrane selectively undergoes feedback phosphorylation by the associated MAPKs, leading to an increased pool of Ste5p at the site of polarized growth. These findings provide evidence of a spatially regulated mechanism for post-activation control of a signaling scaffold that potentiates pathway activation.
Cells employ complex signal transduction networks to properly adapt to environmental stimuli and integrate different external cues with the physiological state of the cell. Scaffold and adapter proteins play crucial roles in mediating the temporal and spatial organization of the networks of signal transduction enzymes that mediate responses to stimuli (1)(2)(3)(4)(5)(6)(7)(8). To fulfill these tasks, scaffold and adapter proteins are regulated at the level of phosphorylation, oligomerization, and cellular distribution, which may influence the activity and location of the scaffold and its binding partners (9 -14).
The mating response of haploid Saccharomyces cerevisiae cells provides one of the best studied examples of a signaling pathway that is regulated by a scaffold protein (15)(16)(17). Upon binding of mating pheromone to a G protein-coupled receptor of the serpentine family, the G␤␥ (Ste4p/Ste18p) dimer of the G protein is released from an inhibitory G␣ subunit (Gpa1p) and activates a mitogen-activated protein kinase (MAPK) 1 cascade. The MAPK cascade consists of a MAPKKK Ste11p, a MAPKK Ste7p, and two MAPKs, Fus3p and Kss1p, of which Fus3p is the major MAPK. The relay of the signal through the MAPK cascade is achieved through sequential phosphorylation of each kinase. Previous work has established that the Ste5p scaffold is essential for this signal relay and plays two distinct roles: Ste5p binds to Ste11p, Ste7p, and Fus3p and tethers them into an active complex. In addition, Ste5p binds to the G␤ subunit of the activated G protein and enables Ste11p to be activated by Ste20p, a p21-activated protein kinase that is enriched at the plasma membrane through its association with Cdc42p, a Rhotype GTPase.
A variety of evidence argues that it is the interaction between a RING-H2 domain in Ste5p and the G␤ subunit (Ste4p) that allows for the assembly of the associated MAPK cascade near Ste20p at the plasma membrane (18 -21). Localization studies indicate that Ste5p undergoes an elaborate recruitment process to be functional (9). During vegetative growth, Ste5p continuously shuttles between cytoplasm and nucleus. In response to mating pheromone, a pool of Ste5p that is derived from the nucleus is recruited to Ste4p at the plasma membrane.
Despite the pivotal role of Ste5p in regulating the mating MAPK cascade, little is known about how it is regulated at a molecular level. The active form of Ste5p is an oligomer, which may also shuttle and be recruited to the plasma membrane (13,22). Ste5p may also undergo conformational changes to mediate activation of the MAPK cascade (13,23). In vitro evidence suggests that Fus3p phosphorylates Ste5p (24), however, in vivo evidence in support of such a feedback regulatory mechanism has been lacking. Previous work suggests that the bulk pool of Ste5p is phosphorylated on at least 15 serine and threonine residues during vegetative growth (25), however, phosphorylation as a result of mating pheromone was not detected. To better understand how Ste5p phosphorylation is regulated, we devised a methodology that allows reproducible detection of phosphorylated forms of Ste5p expressed at native levels in vivo both during vegetative growth and pheromone signaling. Using this methodology, we find that Ste5p is phosphorylated by two distinct sets of kinases during vegetative growth and in response to mating pheromone. Pheromone-induced phosphorylation requires plasma membrane localization of Ste5p and is primarily regulated by the mating MAPKs with additional input by upstream kinases. Moreover, the mating MAPKs positively regulate the abundance of Ste5p at a post-translational step during pheromone stimulation, suggesting a potential level of feedback control that could be regulated by phosphorylation. Table I for a list of yeast strains and plasmids used in this study. Yeast strains were grown in standard selective synthetic complete (SC) media. Strains transformed with P GAL -driven genes were pre-grown in 2% raffinose medium, and then switched to 2% galactose medium to induce transcription. MAPK cascade kinase deletion strains carrying an additional copy of the STE12 gene under control of a leaky GAL1 promoter (pNC252) were grown in 2% dextrose medium, which permitted low-level transcription of STE12. The leaky GAL1 promoter was confirmed in a growth test using a P FUS1 -HIS3 reporter gene. Pheromone induction was performed at a cell density of ϳA 600 1.0 for 1 h with 250 nM ␣-factor (C. Dahl, Harvard Medical School, Boston, MA) for bar1⌬ cells and 5 M for BAR1 cells, unless indicated otherwise. In some instances, mating pathway activation employed overexpression of the STE4 genes as follows: cells carrying P GAL -STE4 (pL19) were grown in 2% raffinose medium to a cell density of ϳA 600 0.8, then switched to 2% galactose medium for 4 h prior to a 1-h ␣-factor exposure. Cells carrying temperature-sensitive mutations were grown at room temperature, and then shifted to 37°C for 4 h. Transformation of yeast was performed as described (26) with the addition of 40 mM dithiothreitol upon plasmid DNA incubation. Standard cloning techniques were used to construct all plasmids. pCU-NLS K128T -S5-M9 was made by swapping a 1.2-kb AFLII-SphI fragment of pSKM96 (S. Mahanty) into pSKM12 (9). pCL-S5C180A-M9 was made by swapping a 4.1-kb SacI-SphI fragment of pSKM88 (9) into pSKM49 (S. Mahanty). Gene replacements were carried out by homologous recombination using EY957/pNC113 to create AFY112 and EY1110/ pSURE11 to create AFY335. Gene replacements were confirmed by mating assays. AFY49, AFY104, and AFY274 are ura3 Ϫ derivatives of EY1881 (E. Elion), K4580 (27) and EY1883 (28), respectively; obtained by selection with 5-fluoro-orotic acid.

Strains and Plasmids-See
Assessment of Ste5p Abundance-Cells harboring GAL1-STE4 (pL19) and STE5-MYC9 (pSKM49 or pSKM92) were grown in SC selective medium containing 2% raffinose to an A 600 of ϳ0.75, then pelleted and resuspended in fresh medium containing 2% galactose and induced for 4 h with shaking at 30°C, followed by treatment with ␣ factor for the indicated times. The cycloheximide experiments were done by growing cells in SC selective medium containing 2% dextrose or 2% galactose to an A 600 ϳ 0.75, then treated with 10 mg/ml cycloheximide and 50 nM ␣ factor in the indicated order for the indicated lengths of time. Whole cell extracts were prepared as previously described by glass bead breakage (29) in the described buffer with addition of 150 mM NaCl, 2 mM benzamidine, 4 mM 1,10-phenanthroline, 50 mM NaF, 1:100 dilution of phosphatase inhibitor mixture (Sigma P2850). Decreased total protein recovered from cycloheximide-treated cells provided evidence that translation had been inhibited. The pulse expression experiments were done by expressing STE5-MYC9 from the GAL1 promoter for 1.5 h in SC selective medium containing 2% galactose to an A 600 ϳ ste7::LEU2 B. Errede 0.5, removing an aliquot, then pelleting the cells at room temperature and resuspending them in prewarmed SC selective medium at pH 4 containing 2% glucose, with or without 2 M ␣ factor. The inclusion of 150 mM NaCl or greater in the breaking buffer was essential for efficient recovery of Ste5 protein and detection of increases in abundance after ␣ factor treatment. Northern Analysis-Northern analysis was performed as described (30). STE5 mRNA was detected with an internal 1.5-kb KpnI-SalI fragment from STE5, and FUS3 mRNA was detected with a 3.0-kb EcoRI-EcoRI fragment from pYEE93 (31). DNA fragments were purified from low melting agarose gels, radiolabeled by the hexamer labeling method, then purified by ethanol precipitation with carrier tRNA.
Gel-shift Assays-A Myc-tagged Ste5p (Ste5-Myc9p (9)) that is fully functional and expressed at native levels was used for the analysis. For whole cell extractions, equal numbers of cells were normalized by cell density, washed in ice-cold H 2 O containing 1 mM phenylmethylsulfonyl fluoride, resuspended in SDS-loading buffer containing 60 mM Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, 5% ␤-mercaptoethanol and bromphenol blue, and then broken by vortexing with acid-washed glass beads (Sigma). The lysates were boiled then electrophoresed on 7.5% SDS-PAGE gels. Immunoblots were probed with 9E10 mAb (Harvard University Monoclonal Antibody facility; cell supernatant, 1:25 dilution) and ␣Tcm1 mAb (J. Warner, Albert Einstein College of Medicine, Bronx, NY; 1:10,000 dilution). Immunoreactivity was detected with horseradish peroxidase-conjugated secondary antibody (Bio-Rad) using the enhanced chemiluminescence system (Amersham Biosciences).
Phosphatase Treatment of Ste5-MYC9p Immune Complexes-Whole cell extracts were prepared exactly as described above except the extraction buffer contained 0.14% ␤-mercaptoethanol and no bromphenol blue. 150 l of protein extract from 20 A 600 units worth of cells was diluted 1:20 in modified H buffer (29) that also contained 150 mM NaCl, 50 mM NaF, 5 mM benzamidine, and 1 mM EDTA. Immunoprecipitations of Ste5-MYC9p were carried out as described (29) using 60 l of 9E10 cell supernatant and 50 l of Protein A-Sepharose beads (Sigma). Ste5-MYC9p immunoprecipitates were washed in dephosphorylation buffer (40 mM PIPES, pH 6.0, 1 mM dithiothreitol, 2 mM phenylmethylsulfonyl fluoride and protease inhibitors (leustatin, pepstatin, chymostatin, papain, 10 g/ml each)). Each reaction used 1 ⁄3 of the immune complexes and 0.35 unit of acid phosphatase from potato (Fluka) dissolved in storage buffer containing 100 mM HEPES, pH 7.4, 0.5 mM MgCl 2 , 0.5 mM dithiothreitol, 50% glycerol. In addition, the phosphatase inhibitors, 50 mM NaF and 10 mM orthovanadate, were added where indicated. The reactions were performed at 4°C and inactivated after 1 h by boiling the immune complexes in SDS loading buffer at 95°C for 10 min. Immunoblot analysis was performed as described above. Bacterial alkaline phosphatase and protein phosphatase 2 were less effective at dephosphorylating Ste5-MYC9p in other experiments. 2 Kinase Assays-Fus3-HAp, Fus3K42R-HAp, and GST-Ste5p immune complexes were prepared from whole cell extracts as described (24,28). Kinase assays were performed as described (29). Duplicate immune complexes were analyzed by immunoblot analysis using 12CA5 monoclonal antibody to detect Fus3-HAp and Fus3K42R-HAp and anti-GST antiserum to detect GST and GST-Ste5p.

The Mating MAPKs Positively Regulate the Abundance of Ste5p
Post-translationally-To study potential post-translational modification of Ste5p we expressed a fully functional tagged derivative of Ste5p (Ste5-MYC9p) (9) at native levels from a centromeric plasmid in a ste5⌬ strain. Immunoblot analysis showed that the abundance of Ste5p increases as a result of mating pheromone stimulation, with the increase most apparent 1 h after ␣ factor addition, as confirmed by reprobing for the ribosomal protein Tcm1p (Fig. 1A). Overexpression of the Ste4p G␤ subunit, which binds Ste5p, further increased the abundance of Ste5p in the presence of ␣ factor (Fig. 1A) and was sufficient to increase the abundance of Ste5p in the absence of ␣ factor (Fig. 1B). Similar increases in abundance were found for Fus3p (Fig. 1, A and B). The increase in Ste5p abundance was dependent on the mating MAPKs and was blocked in a fus3⌬ kss1⌬ double mutant (Fig. 1B). In contrast, the level of STE5 mRNA did not increase as a result of activation of the mating pathway, and was not affected by mutations in the mating MAPKs or STE12, whereas the level of FUS3 mRNA was increased by mating pheromone and decreased by mutations in STE5 and STE12 (Fig. 1C). These findings suggested that the increase in Ste5p protein was post-transcriptional.
To determine whether the increase in Ste5p abundance might be post-translational, we treated cells with cycloheximide to block translation of Ste5p and then added ␣ factor. Prior analysis has shown that ␣ factor activation of the mating MAPKs still occurs in the presence of cycloheximide (32). Immunoblot analysis of the cycloheximide-treated cells revealed an increase in Ste5p abundance after ␣ factor treatment ( Fig.  1D), demonstrating that the increase is post-translational. To circumvent secondary effects of cycloheximide on the level of components in the pathway that activate the MAPKs, we also induced with ␣ factor for 1 h before adding cycloheximide for another hour in wild type and fus3⌬ kss1⌬ cells. The level of Ste5p was still greater as a result of ␣ factor induction with no increase in the absence of Fus3p and Kss1p (Fig. 1E). A pulse expression method was next used to determine whether Ste5p is stabilized during ␣ factor induction (33). The transcription of the STE5-MYC9 gene was induced for 90 min with the GAL1 promoter. Further expression was repressed by the addition of dextrose and the abundance of Ste5-Myc9p was monitored for 150 min in cells treated with or without ␣ factor. The level of Ste5-Myc9p fell to 24% of the initial level in the cells that had not been treated with ␣ factor compared with a much smaller decline to ϳ85% in the ␣ factor-treated cells (Fig. 1, F and G). Thus, Ste5p abundance increases at a post-translational step as a result of activation of the mating MAPKs, possibly as a result of stabilization of Ste5p from degradation. However, these findings do not rule out the possibility that the ␣ factorinduced increase in Ste5p abundance may also involve enhanced translation of STE5 mRNA, because the level of Ste5p increases at the earliest (10 min) time point after the shift to glucose in ␣ factor-treated cells (Fig. 1F).
Ste5p Is Rapidly and Specifically Modified in Response to Pheromone Signaling-Previous unpublished work in this laboratory suggested that Ste5p is modified in vivo in response to the ␣ factor mating pheromone, but it was difficult to reproducibly detect the modified forms using conventional methods of extract preparation (34) (note broad mobility of Ste5p in Fig.  1). To better capture the modification status of Ste5p we lysed cells directly in SDS-loading buffer and separated these lysates on SDS-PAGE gels. Under these conditions, we detected a pattern of at least two differently migrating species of Ste5p during vegetative growth, and this pattern shifted toward a slower migrating species in the presence of mating pheromone in addition to greater abundance ( Fig. 2A). The difference in the migration pattern in the absence and presence of mating pheromone was highly reproducible and not attributed to differences in loading, as shown by the relative levels of Tcm1p.
To determine whether the modification on Ste5p was the result of an initial signaling event, we compared its mobility in an ␣ factor time course of cells that had been pretreated with cycloheximide. An optimal shift in the Ste5p migration pattern was detected as early as 5 min after pheromone exposure, and this migration pattern was independent of the presence of cycloheximide (Fig. 2B, only ϩCHX data shown). 15 to 30 min after induction, the level of slower migrating Ste5p species declined slightly, but still remained detectably elevated after 60 min of ␣-factor exposure. These findings indicate that Ste5p modification occurs early in the signaling process and is not dependent upon transcriptional induction.
To determine whether the Ste5p modification was specific to the mating pheromone stimulus, we tested whether stimuli that activate signaling pathways other than the mating MAPK cascade would affect Ste5p modification. When cells were exposed to osmotic stress conditions known to activate the high osmolarity growth pathway (i.e. 0.4 M NaCl), Ste5p did not undergo a mobility shift (Fig. 2C, although the abundance declined slightly at the 60-min time point). Similar results were obtained when cells were exposed to heat shock, which activates the protein kinase C pathway (data not shown). Therefore, the modification of Ste5p that occurs in the presence of mating pheromone is a specific consequence of activation of the mating pathway.
Finally, to establish a basis for a potential general relevance of the modification on Ste5p, we compared its behavior in two different strain backgrounds, W303a (Fig. 2, A-C) and S288C (Fig. 2D). Ste5p underwent a ␣ factor-induced modification in S288C cells. The ability to detect the modification in S288C was more difficult compared with W303a and more dependent on a brief ␣ factor exposure, consistent with the enhanced ability of this strain background to attenuate the activation of the mating pathway (35). Thus, the Ste5p modification is a general feature of pheromone signaling in different strain backgrounds and is likely to be physiologically relevant.
Ste5p Is Basally Phosphorylated during Vegetative Growth and Hyperphosphorylated in the Presence of Mating Pheromone-To determine whether the modifications were the result of phosphorylation, Ste5p was purified by immunoprecipitation from both uninduced and ␣ factor-induced cells and subjected to phosphatase treatment. Incubation of the Ste5p immune complexes with the nonspecific acid phosphatase from potato caused a complete loss of slower migrating, modified forms of Ste5p compared with mock-treated Ste5p (Fig. 3) and this de-modification was inhibited by phosphatase inhibitors. An effect of phosphatase on Ste5p mobility was observed for both the constitutively modified (i.e. Ϫ␣F) and hypermodified (i.e. ϩ␣F) forms of Ste5p. Therefore, the Ste5p modification revealed by changes in mobility is because of phosphorylation during vegetative growth and enhanced phosphorylation during pheromone signaling. Furthermore, this experiment establishes that the mobility shift provides a reliable way to assess Ste5p phosphorylation.
Basal Phosphorylation of Ste5p Is Independent of the MAPK Cascade Kinases-We next tested whether the basal phosphorylation of Ste5p during vegetative growth was regulated by EY705 (ste5⌬H3). ϩ␣F indicates that cells were treated with 5 mM ␣ factor for 150 min. Note that ste5⌬H3 is a partial deletion of the STE5 gene that generates a truncated mRNA. D, Fus3p and Kss1p increase Ste5p abundance post-translationally. Strains (as in A and B) were grown either in YEP medium containing 2% glucose or in selective medium containing 2% galactose to logarithmic phase, incubated for 10 min in 10 g/ml cycloheximide, then treated with (or without) 50 nM ␣ factor for 15 min. The abundance of Ste5p is less in 2% galactose medium than in 2% glucose medium (R. McCully, D. Simpson, and E. Elion, unpublished results). E, treatment of cells with cycloheximide after ␣ factor induction. Strains were grown in selective medium containing 2% glucose to logarithmic phase, incubated for 60 min in 50 nM ␣ factor, then an additional 60 min after addition of 10 g/ml cycloheximide. F, Ste5-MYC9p is more stable in ␣ factor-treated cells than in vegetatively growing cells. MATa BAR1 cells harboring GAL1-STE5-MYC9 were incubated for 90 min in 2% galactose medium (T ϭ 0), then pelleted at room temperature and resuspended in 2% glucose medium for the indicated times in the absence or presence of 2 M ␣ factor. G, graph of relative abundance of Ste5-Myc9 in the immunoblots in F. The level of Ste5-Myc9p was normalized to the level of Tcm1p for each time point. Quantitation was done with the Scion Image 1.62c densitometry program of the public domain software NIH image (rsb.info.nih.gov/nih-image). Ste5-MYC9p (pSKM12) immune complexes were prepared from uninduced (Ϫ␣F) and ␣ factor-treated cells (ϩ␣F) and then incubated either with mock buffer (lanes 1 and 4), buffer containing potato acid phosphatase alone (lanes 2 and 5), or both potato acid phosphatase and phosphatase inhibitors (lanes 3 and 6).
the mating pathway kinases including Ste20p, which activates Ste11p, and the kinases that bind Ste5p (i.e. Ste11p, Ste7p and Fus3p and Kss1p). All of these kinases exert regulatory functions during vegetative growth as well as during mating (35)(36)(37). In addition, Ste20p serves redundant functions with the p21-activated protein kinase Cla4p for budding (27). The electrophoretic mobility of Ste5p was analyzed in vegetatively dividing cells with null mutations in each of the mating pathway kinases and in a ste20⌬ cla4-75 ts double mutant that harbors a temperature-sensitive mutation in the CLA4 gene. The migration pattern of Ste5p was largely unchanged in the absence of Ste20p and Cla4p with slightly less basal phosphorylation at nonpermissive temperature (Fig. 4A, lanes 1-4, compare relative intensities of the doublet bands). In contrast, loss of Ste11p, Ste7p or Fus3p, and Kss1p had no effect on basal phosphorylation of Ste5p (Fig. 4A, lanes 5-8). The apparent increase in basal phosphorylation of Ste5p in the fus3⌬ kss1⌬ double mutant was a consequence of the lower abundance of Ste5p. Longer exposure of the immunoblot and comparative analysis of other experiments revealed no obvious increase in basal phosphorylation. Therefore, the mating pathway kinases and Cla4p are not responsible for basal phosphorylation of Ste5p.
Basal Phosphorylation of Ste5p Is Regulated by CDK Cdc28p-Another candidate kinase that has previously been linked to regulating components of the mating pathway is the CDK Cdc28p (38). Cdc28p is active in vegetatively growing cells and is inactivated in the presence of ␣ factor, causing cell cycle arrest in G 1 phase. Ste5p accumulates in the nuclei of vegetatively growing wild type G 1 phase cells and undergoes strong nuclear accumulation upon inactivation of Cdc28p (9), raising the possibility that Cdc28p could directly or indirectly affect its localization. The mobility of Ste5p was compared in wild type and cdc28-4 mutant cells before and after a temperature shift. Strikingly, a larger fraction of Ste5p migrated as the faster migrating species in the cdc28-4 mutant at nonpermissive temperature (Fig. 4B), indicating that Cdc28p directly or indirectly regulates Ste5p phosphorylation.
We tested the possibility that the decrease in basal phosphorylation of Ste5p after Cdc28-4p inactivation was a secondary effect of an increase in the pool of Ste5p that resides in the nucleus. If this were the case, then increasing the pool of Ste5p that is nuclear should decrease basal phosphorylation of Ste5p. The nuclear pool of Ste5p was greatly increased throughout the cell division cycle by mutating the Msn5p exportin that is responsible for nuclear export of Ste5p during vegetative growth (9). However, the msn5⌬ mutation had no effect on the Ste5p migration pattern (Fig. 4C), indicating that the reduction in Ste5p basal phosphorylation in the cdc28-4 mutant is not the result of Ste5p being redistributed to the nucleus.
Cdc28p Regulates Basal Phosphorylation of the Cytoplasmic Pool of Ste5p-We next determined whether Ste5p is basally modified in the cytoplasm, by monitoring the mobility of Ste5⌬49 -66p, a derivative of Ste5p that is cytoplasmic and excluded from the nucleus (9). Ste5⌬49 -66p lacks the major NLS and is unable to be imported into the nucleus and remains cytoplasmic even in the absence of the Msn5p exportin. If Ste5p were phosphorylated in the nucleus during nuclear shuttling, then Ste5⌬49 -66p should not be basally modified. However, Ste5⌬49 -66p was still modified, indicating that basal phosphorylation occurs in the cytoplasm (Fig. 4D, lane 1). Furthermore, inactivation of Cdc28-4p still caused loss of the slower migrating Ste5⌬49 -66p species (Fig. 4D). Therefore, Cdc28p regulates basal phosphorylation of Ste5p in the cytoplasm. (24). We tested whether Ste5p could be phosphorylated by kinases other than Fus3p in two in vitro assays. GST-Ste5p was co-immunoprecipitated with either Fus3-HAp or catalytically inactive Fus3K42R-HAp along with other associated substrates in the whole cell extracts and incubated in an in vitro kinase assay that has been established for Fus3p (Fig. 5A) (29). Duplicate immune complexes were assessed for the relative amount of Fus3-HAp, Fus3K42R-HAp, and GST-Ste5p by immunoblot analysis (Fig. 5, A and B). In the Fus3-HAp immune complex kinase assay, the associated GST-Ste5p fusion protein was phosphorylated, as expected (Fig. 5A, lanes 2 and 3). The additional phosphorylated proteins are other physiologically relevant substrates that co-precipitate with Fus3p, including Ste12p, Far1p, Ste11p, Ste7p, Dig1p, and Dig2p (15,17). When the kinase assay was performed on immune complexes of kinase-inactive Fus3R42-HAp, most, but not all, of the GST-Ste5p phosphorylation was abrogated (Fig. 5A, lane 4). These results recapitulate the capacity of Fus3p to phosphorylate Ste5p in vitro and suggest that additional associated kinase(s) also phosphorylate Ste5p.
Mutations in Mating Pathway Kinases Block Hyperphosphorylation of Ste5p in Vivo-To determine whether mating pathway kinases regulate Ste5p phosphorylation during pheromone signaling in vivo, we looked at the migration of Ste5p in ste20⌬, ste11⌬, ste7⌬, and fus3⌬ kss1⌬ mutant cells after they had been treated with mating pheromone. To circumvent the secondary consequence of decreased activation of Ste12p-dependent transcription and the attendant decrease in expression of signaling components (e.g. STE2, STE4, and FUS3), we expressed additional low levels of Ste12p. Under these conditions, null mutations in STE20, STE11, STE7, or FUS3/KSS1 blocked the pheromone-induced shift in the Ste5p migration pattern (Fig.   6A, lanes 1-10; note that longer exposure indicated that basal phosphorylation was not elevated in the ste11⌬ and fus3⌬ kss1⌬ strains compared with wild type). In contrast, single mutation of FUS3 or KSS1 did not interfere with the pheromone-induced Ste5p modification (lanes 11-14). The inhibitory effect of MAPK cascade mutants on ␣ factor-induced Ste5p phosphorylation was highly specific. By comparison, mutations in a variety of other kinases that have been linked to the mating pathway (e. g. prr1⌬, prr2⌬, and cbk1⌬), did not block basal or induced phosphorylation of Ste5p (data not shown). These findings demonstrate that the mating MAPK cascade is responsible for hyperphosphorylation of Ste5p during pheromone signaling and are consistent with the in vitro results (Fig.  5). However, they do not distinguish the relative contributions of individual kinases to total phosphorylation of Ste5p.
Hyperphosphorylation of Ste5p Is Regulated by Fus3p, Kss1p, and Ste11p-In a second approach, we enhanced the activation of the mating pathway by overexpressing Ste4p, which is sufficient to activate the pathway (42). Increasing the level of Ste4p increases the amount of free G␤␥ dimers, thereby increasing the pool of Ste5p that is recruited to the plasma membrane and enhancing activation of the associated MAPK cascade by Ste20p. Overexpression of Ste4p stimulated the modification of Ste5p based on a very pronounced mobility shift (Fig. 6B, lanes 1 and 2). The overexpression of Ste4p was sufficient to induce a mobility shift in Ste5p (data not shown) in addition to an increase in abundance (Fig. 1, A and B), however, a further increase was detected in the presence of ␣ factor. Phosphatase treatment confirmed that the mobility shift in the presence of ␣ factor and Ste4p was because of phosphorylation (data not shown).
Under these conditions, a partial shift in Ste5p migration after ␣ factor induction could still be observed in all ste mutant strains tested. However, the magnitude of the residual mobility shift was not equivalent in the different kinase mutants after correcting for differences in abundance by comparing long exposures of the immunoblots. Representative mutants showing the effect of an early (ste20⌬) and a late block (ste7⌬) in the mating MAPK cascade are shown in Fig. 6B. The ste20⌬ mutation partially decreased hyperphosphorylation of Ste5p, but did not block the increase in Ste5p accumulation. In contrast, the ste7⌬ mutations caused a greater decrease in the amount of Ste5p hyperphosphorylation and also blocked accumulation of Ste5p. Similarly, the ste11⌬ mutation caused a partial defect in Ste5p hyperphosphorylation and did not block Ste5p accumulation, whereas the fus3⌬ kss1⌬ double mutant was most severely defective in both Ste5p hyperphosphorylation and accumulation (Fig. 6C, lanes 3-6). These comparisons demonstrate that Fus3p and Kss1p are most responsible for the total amount of hyperphosphorylation and accumulation of Ste5p and suggest that the upstream kinases play a lesser role.
To test the possibility that additional MAPK cascade kinases besides Fus3p and Kss1p were responsible for hyperphosphorylation of Ste5p, we examined the mobility of Ste5p in a ste11⌬ fus3⌬ kss1⌬ triple mutant devoid of mating MAPK activity. Strikingly, the Ste5p mobility shift induced by ␣ factor was completely abolished in the ste11⌬ fus3⌬ kss1⌬ triple mutant (Fig. 6C, lanes 7 and 8; longer exposure of immunoblots do not reveal hypermodification in the triple mutant). Together these findings suggest that the MAPKs Fus3p and Kss1p are the major contributors of the pheromone-induced phosphorylation and accumulation of Ste5p, but that Ste11p and/or Ste7p also contribute to a lesser degree.
Pheromone-induced Ste5p Phosphorylation Requires Nuclear Shuttling and Recruitment of Ste5p to the Plasma Membrane-Because Ste5p shuttles continuously between the nucleus and cytoplasm and is recruited to the plasma membrane in response to pheromone signaling, we wondered whether Ste5p phosphorylation during pheromone stimulation occurs in a specific subcellular compartment. To address this question, the migration pattern of two mutant derivatives of Ste5p with altered localization, Ste5⌬49 -66p and TAgNLS-Ste5p, was compared (Fig. 7A, cartoon) (9). Ste5⌬49 -66p does not shuttle through the nucleus and as a consequence is not recruited to the plasma membrane. In contrast, TagNLS-Ste5p shuttles through the nucleus, but is predominantly nuclear both in the absence and presence of mating pheromone because of efficient reimport and is poorly recruited to the plasma membrane. To circumvent secondary effects of the mutations on signal transduction, we analyzed the modification status of the mutant derivatives of Ste5p in a strain that also expressed wild type Ste5p. Basal phosphorylation of both Ste5⌬49 -66p and TAgNLS-Ste5p occurred as efficiently as with wild type Ste5p, consistent with it occurring in the cytoplasm prior to nuclear import. In contrast, the pheromone-induced shift of both Ste5⌬49 -66p and TAgNLS-Ste5p was blocked (Fig. 7A, lanes  3-6), suggesting that the pheromone-dependent Ste5p phosphorylation does not occur in the cytoplasm or the nucleus, but may occur at the plasma membrane.
We further tested the possibility that the pheromone-induced modification occurs at the plasma membrane, by comparing the ability of TAgNLS K128T -Ste5p to be modified with that of TAgNLS-Ste5p. Like TAgNLS-Ste5p, TAgNLS K128T -Ste5p shuttles through the nucleus and localizes predominantly in the nucleus both in the absence and presence of pheromone. However, in contrast to TAgNLS-Ste5p it is less efficiently reimported into the nucleus and more efficiently recruited to the plasma membrane than wild type Ste5p in the presence of mating pheromone (9). Strikingly, TAgNLS K128T -Ste5p underwent a pheromone-induced mobility shift comparable with wild type Ste5p (Fig. 7A, lanes 7 and 8), suggesting that pheromone-induced phosphorylation of Ste5p occurs at the plasma membrane.
A second observation supported the possibility that phosphorylation of Ste5p occurs at the plasma membrane. Ste5-CTMp localizes to the plasma membrane through a transmembrane domain and activates the mating pathway, presumably because of its recruitment of the MAPK cascade to Ste20p (20,21). Overexpression of Ste5-CTMp blocked Ste5-MYC9p modification induced by ␣ factor (Fig. 7B, lanes 1 and 3), even though expression of Ste5-CTMp greatly activates the mating MAPK cascade. This observation is consistent with the possibility that Ste5-CTMp sequesters the MAPK cascade kinases at the plasma membrane away from Ste5-MYC9p.
Hyperactivation of the Mating MAPK Cascade in the Absence of Ste5p Recruitment to Ste4p Is Not Sufficient to Hyperphosphorylate Ste5p-To further determine whether plasma membrane recruitment of Ste5p is required for its hyperphosphorylation in response to pheromone, we tested whether interfering with the ability of Ste5p to bind to Ste4p would block hyperphosphorylation. The Ste5C180Ap mutant harbors a single amino acid exchange in a cysteine predicted to be one of the Zn 2ϩ -coordinating cysteines of the RING-H2 domain of Ste5p. Previous work has shown that the C180A mutation blocks binding of Ste5p to Ste4p and plasma membrane recruitment in vivo, even when Ste4p is overexpressed or wild type Ste5p is present (9,19). Ste5C180Ap underwent basal modification, but not ␣ factor-induced modification in a STE5 background (Fig.  7C, top panel). To further confirm the necessity of recruitment, the experiment was repeated in the presence of increased levels of Ste4p. However, overexpression of Ste4p failed to restore the pheromone-dependent shift (Fig. 7C, bottom panel). Thus, pheromone-induced phosphorylation of Ste5p requires its recruitment to the plasma membrane, an event that requires binding of Ste5p to Ste4p.
If recruitment of Ste5p to Ste4p is needed for Ste5p hyperphosphorylation in response to pheromone signaling, then hyperactivation of the mating MAPK cascade in the absence of Ste5p recruitment should not be sufficient to induce hyperphosphorylation of Ste5p. To test this possibility, we monitored the mobility of Ste5p from a strain expressing a constitutively active form of MAPKKK Ste11p, Ste11-4p (43). As predicted, Ste11-4p did not substantially enhance Ste5p phosphorylation in the absence of pheromone (Fig. 7D, lanes 1, 3, and 5). Strikingly, however, the addition of ␣ factor caused an even greater shift in Ste5p mobility in the STE11-4 strain than in wild type (Fig. 7D, lanes 2 and 4), consistent with the expected hyperactivation of the MAPK cascade. Furthermore, this shift required the presence of Ste4p (Fig. 7D, lane 6). Activation of the MAPK cascade kinases is therefore not sufficient to cause Ste5p hyperphosphorylation in the absence of mating pheromone and Ste4p.
Hyperphosphorylation of the Mating MAPK Cascade Increases the Amount of Ste5p That Localizes at the Cell Cortex-We determined whether the enhanced phosphorylation of Ste5p that occurs in the presence of Ste11-4p correlates with changes in the level of Ste5p that is found at the cell cortex during mating pheromone stimulation. The localization of Ste5 was determined in wild type and STE11-4 cells expressing either STE5-GFP or Ste5-Myc9p. The strength of the Ste5-GFP and Ste5-Myc9p cortical signals were stronger in STE11-4 cells than in wild type cells that were treated with ␣ factor (Fig. 8,  A and B). Furthermore, quantification of Ste5-Myc9p in STE11-4 cells revealed a greater number of cells displaying cortical recruitment of Ste5-Myc9p when Ste11-4p was present and the degree of polarized growth was noticeably greater at the 30-min time point (Fig. 8B). These findings suggest that activation of the mating MAPK cascade increases the amount of Ste5p that is recruited to sites of polarized growth. DISCUSSION Our results show that the phosphorylation of Ste5p during vegetative growth and in response to pheromone signaling are two differently regulated phosphorylation events in terms of both the kinases conferring the phosphorylation and the subcellular compartment where the events take place. Basal phosphorylation of Ste5p occurs independently of the mating MAPK cascade components; even a ste20 cla4 double mutant, deleted for the two topmost kinases that together are essential during vegetative growth, is not defective in basal phosphorylation of Ste5p. Strikingly, however, inactivation of Cdc28-4p greatly decreases Ste5p phosphorylation during vegetative growth, indicating that Cdc28 directly or indirectly regulates Ste5p phosphorylation. The decrease in Ste5p phosphorylation in the cdc28-4 strain is not a secondary effect of greater accumulation of Ste5p in G 1 phase cells upon inactivation of Cdc28-4p, be- cause increasing the pool of Ste5p that is nuclear in a msn5⌬ mutant does not affect the level of basal Ste5p phosphorylation. In addition, it is unlikely to be a secondary effect of G 1 phase arrest because basal phosphorylation of Ste5p is also reduced in a cdc28-1N mutant that arrests in G 2 /M phase (data not shown). Two lines of evidence suggest that Cdc28p regulates basal phosphorylation of Ste5p in the cytoplasm. First, restricting Ste5p to the cytoplasm by deleting its NLS sequence (i.e. Ste5⌬49 -66p) has no effect on the ability of Ste5p to be basally phosphorylated. Second, basal phosphorylation of the cytoplasmic-restricted form of Ste5p is decreased in a cdc28-4 mutant. Thus, the available evidence argues that Ste5p is basally phosphorylated in the cytoplasm.
The ␣ factor-induced phosphorylation of Ste5p is rapid and occurs in the presence of cycloheximide, indicating that it is a primary event that is the result of pheromone signaling. Moreover, it is highly specific to the mating pheromone stimulus and does not occur as a result of stimuli known to activate the high osmolarity growth and PKC MAPK cascades. A deletion in any individual kinase in the mating MAPK cascade partially inhibits pheromone-induced phosphorylation. In contrast, a double deletion of both FUS3 and KSS1 causes a large decrease in the pheromone-induced phosphorylation event, arguing that Fus3p and Kss1p are the major kinases phosphorylating Ste5p in vivo. This conclusion is supported by the ability of Fus3p to phosphorylate Ste5p in vitro (24).
Additional analysis suggests that other MAPK cascade kinases contribute to the pheromone-induced phosphorylation of Ste5p. Under conditions in which the pheromone-induced phosphorylation is enhanced by overexpression of the G␤ subunit Ste4p, it is possible to detect residual phosphorylation in the fus3⌬ kss1⌬ double mutant. The residual phosphorylation is completely abrogated in a ste11⌬ fus3⌬ kss1⌬ triple mutant, suggesting that Ste11p and/or Ste7p could act in addition to Fus3p and Kss1p to phosphorylate Ste5p. Although our experiments do not distinguish between Ste11 and Ste7, the fact that a ste7⌬ single mutant has an equivalent defect in Ste5p phosphorylation as a fus3⌬ kss1⌬ double mutant suggests that Ste7 does not contribute significantly to total Ste5p phosphorylation apart from activation of Fus3p and Kss1p. In addition, precedent is consistent with the MAPKKK (Ste11) class of kinases having targets besides MAPKKs (44), whereas MAPKKs are thought to be highly specific for MAPKs.
Interestingly, the pheromone-induced phosphorylation of Ste5p is strictly dependent on its recruitment to the plasma membrane, even though Ste5p is distributed throughout the cytoplasm and nucleus. Multiple lines of evidence support this interpretation. First, overexpression of membrane-associated Ste5-CTMp interferes with pheromone-induced phosphorylation of wild type Ste5p. Second, restoring plasma membrane recruitment of the nuclear-localized TAgNLS-Ste5p restores pheromone-induced phosphorylation. Third, Ste5C180Ap, which is only defective in binding to G␤, is unable to undergo enhanced phosphorylation in response to pheromone signaling, in a cell that expresses wild type Ste5p and is competent for signaling. Most strikingly, hyperphosphorylation of Ste5p in the presence of constitutively active Ste11-4p still requires pheromone induction, which presumably allows for plasma membrane recruitment of the Ste5p/Ste11-4p signaling complex and activation of the mating MAPK cascade kinases at the Ste11p step alone was not sufficient. Together these results indicate that the pheromone-induced phosphorylation event occurs at the plasma membrane within a Ste5p signaling complex, possibly upon binding of Ste5p to Ste4p. The obvious advantage of this mechanism is that it allows for tightly regulated feedback control of Ste5p at the site of activation.
Further work is needed to determine the biological function of both basal and induced phosphorylation of Ste5p. Given the fact that Ste5p is phosphorylated on a minimum of 15 serine and threonine residues during vegetative growth (25) and the multiple aspects of Ste5p function, it will not be trivial to sort out the relevance of individual phosphorylations. Cdc28p and the mating MAPKs are both proline-dependent kinases with potential for overlap if they recognize minimal S/TP sites in Ste5p, although their idealized consensus recognition sites are likely to be different. Overlap in recognition sites has been noted for the Fus3p and Pcl/Pho85 class of cyclin-dependent kinases (45). Possible levels of regulation of Ste5p include influencing the binding properties of the kinases or other components or influencing the stability of the Ste5p signaling complex at the plasma membrane.
Interestingly, we found that the activation of the mating MAPKs increases the abundance of Ste5p. Additional analysis suggests that this increase is post-transcriptional and occurs at a post-translational step that involves stabilization of Ste5p. These findings support the possibility that feedback phosphorylation of Ste5p by the MAPKs stabilizes Ste5p, perhaps by preventing it from being degraded either at the plasma membrane or after it dissociates. Fus3p is more likely to play the critical role in feedback phosphorylation at the plasma membrane, because Kss1p is not found associated with Ste5p at cortical sites (17). This interpretation is strongly supported by the observation that a fus3⌬ null mutation reduces the level of Ste5p that accumulates at cortical sites. 3 Given that the level of Fus3p also increases in the presence of ␣ factor, this type of regulatory loop might ensure that a larger pool of Ste5p is present to activate the increased pool of Fus3p and further potentiate pathway activation. Interestingly, greater recruitment of Ste5p is observed in a STE11-4 strain after 15 and 30 min of ␣ factor stimulation. This finding correlates with the greater phosphorylation of Ste5p and raises the possibility that feedback phosphorylation of Ste5p by the MAPKs potentiates pathway activation. Such a regulatory device might be particularly important for polarized growth, which may be linked to cortical recruitment of Ste5p and requires greater pathway activation than other outputs (such as cell cycle arrest or transcriptional activation; Ref. 32).