Regulation of Stress Response Signaling by the N-terminal Dishevelled/EGL-10/Pleckstrin Domain of Sst2, a Regulator of G Protein Signaling in Saccharomyces cerevisiae

All members of the regulator of G protein signaling (RGS) family contain a conserved core domain that can accelerate G protein GTPase activity. The RGS in yeast, Sst2, can inhibit a G protein signal leading to mating. In addition, some RGS proteins contain an N-terminal domain of unknown function. Here we use complementary whole genome analysis methods to investigate the function of the N-terminal Sst2 domain. To identify a signaling pathway regulated by N-Sst2, we performed genome-wide transcription profiling of cells expressing this fragment alone and found differences in 53 transcripts. Of these, 40 are induced by N-Sst2, and nearly all contain a stress response element (STRE) in the promoter region. To identify components of a signaling pathway leading from N-Sst2 to STREs, we performed a genome-wide two-hybrid analysis using N-Sst2 as bait and found 17 interacting proteins. To identify the functionally rel-evant interacting proteins, we analyzed all of the available gene deletion mutants vps36 also repress VPS36 for further A vps36 (cid:1) diminishes by the Con-versely, overexpression of Vps36 enhances the pheromone response in sst2 (cid:1) cells but not in wild type. These findings indicate that Vps36 and Sst2 have opposite and opposing effects on the pheromone and stress response pathways, with Vps36 acting downstream of the G protein and independently of Sst2 activity.

All cells have the capacity to respond to chemical and sensory stimuli in their environment. In many cases, signal detection occurs through cell surface receptors coupled to G proteins. One particularly well characterized example is the pheromone response pathway in yeast (1). In this case, haploid a and ␣ cell types each secrete a peptide pheromone that binds to receptors on cells of the opposite type. Pheromone stimulation leads to activation of a G protein, which entails GTP binding and dissociation of the G␣ and ␤␥ subunits. The G␤␥ moiety activates downstream signaling events required for mating, including alterations in gene transcription, morphological and cytoskeletal changes, and cell cycle arrest in G 1 . Among the induced genes is SST2, which encodes a feedback regulator that stimulates G protein GTPase activity and G protein inactivation (1).
Over the past 5 years, an extensive family of Sst2-related proteins has been identified in higher eukaryotes (2). In every instance examined, the region of core-RGS 1 homology is both necessary and sufficient for G protein GTPase activating function (3). Some RGS proteins contain additional domains or motifs that may be recognized by proteins other than G␣ (4 -13). The RGS protein p115RhoGEF has one domain that acts as a GTPase-accelerating protein for G 13 ␣ and a second domain that acts as a GDP-GTP exchange factor for RhoA (14,15). Other RGS proteins including Egl-10, Eat-16, RGS6, RGS7, RGS9, RGS11, and FlbA, have large N-terminal segments containing a conserved Dishevelled, Egl-10, and pleckstrin (DEP) domain (16). Sst2 has two such DEP regions, composed of residues 50 -135 and 279 -358. The function of the RGS DEP domains is not known, but in at least two cases (Egl-10, Sst2) they appear necessary and sufficient for membrane localization (17,18). In the case of Sst2, the N-terminal domain can be expressed as a separate entity, the result of an endoproteolytic processing event in vivo (18).
Our objective here was to establish a signaling function for the DEP domain of Sst2, designated N-Sst2. With the completion of the yeast genome sequence, approaches to the identification of new signaling pathways have changed dramatically. Analysis of gene function has become more comprehensive and systematic and can occur at several levels. First, closely related protein isoforms can be identified through sequence similarity analysis or through complementation of gene mutations by functionally similar genes. Second, transcriptional changes can be monitored under various physiological conditions, through the use of RNA hybridization arrays (19). Third, signaling complexes can be identified through the use of two-hybrid screens (20,21) or through the isolation and sequencing of multiprotein complexes (22)(23)(24)(25)(26). Fourth, the functional significance of each signaling component can be determined through gene disruption mutations (27).
There are, however, limitations to each of these methods. For instance, transcription analysis can reveal how different physiological conditions affect a particular signaling pathway but cannot be used to identify the components of that pathway. Two-hybrid analysis can reveal the components of a pathway, but it cannot be used to determine how physiological changes affect the interactions of each component. Thus, a combined analysis, encompassing multiple whole genome approaches, can provide highly complementary information about any cellular process. For instance, a combination of two different high throughput methods, protein interaction mapping and large scale phenotypic analysis, was recently used to identify novel DNA repair and DNA damage checkpoint pathway components in Caenorhabditis elegans (28).
Here we have used a combination of transcription profiling, protein interaction mapping, and phenotypic analysis of gene disruption mutants to investigate signaling by the N-terminal Sst2 domain. Our findings indicate that N-Sst2 modulates the stress response and does so through proteins not previously recognized to participate in the mating or stress pathways.
Two-hybrid N-Sst2 Bait Plasmid Construction-Oligonucleotides were designed to PCR-amplify the N-terminal portion (amino acids 1-392) of Sst2, using pRS316-ADH-Sst2 as the template, the forward primer (5Ј-CCG GAA TTC ATG GTG GAT AAA AAT AGG ACG-3Ј, containing a synthetic EcoRI site, indicated in boldface type), and the reverse primer (5Ј-GTA CCC ATG GTT ACA TAT GAC CCC TTA ATG TGA A-3Ј, containing a synthetic NcoI site, indicated in boldface type). The 0.9-kbp product was cloned in frame downstream of the GAL4 DNA-binding domain contained in the yeast two-hybrid bait vector pOBD2 (35).
RNA Isolation and Hybridization-Total RNA was isolated from yeast strains using RNeasy columns (Qiagen, Valencia, CA), and stored at Ϫ80°C. Genomic DNA was removed by DNase digestion of 10 g of total RNA for 30 min at 37°C in a 100-l reaction containing DNase I (1.4 units; Invitrogen), RNase inhibitor (0.1 units; Invitrogen), and dithiothreitol (1 mM) in 1ϫ PCR buffer I (PerkinElmer Life Sciences). DNase was removed by passage through an RNeasy column (Qiagen).
Amplified, biotin-labeled cRNA was produced from total RNA as described (36). Briefly, 10 g of total RNA was incubated for 10 min at 70°C with a high pressure liquid chromatography-purified oligo(dT) primer containing a T7 RNA polymerase promoter site (5Ј-GGC CAG TGA ATT GTA ATA CGA CTC ACT ATA GGG AGG CGG T-3Ј; from GENSET Inc., La Jolla, CA). Following priming, cDNA was prepared using the SuperScript II cDNA synthesis kit (Invitrogen) with the following conditions: 65 min at 50°C for first strand synthesis with Superscript II reverse transcriptase, followed by 150 min at 16°C for second strand synthesis with Escherichia coli ligase, E. coli polymerase, and RNase H. cDNA was purified by phenol/chloroform extraction followed by removal of the organic fraction using Phase Lock Gel I tubes (5 Prime to 3 Prime Inc., Boulder, CO). Biotin-labeled cRNA was transcribed in an in vitro transcription reaction mixture containing T7 RNA polymerase (Epicenter, Madison, WI), bio-11-CTP, and bio-11-UTP (Enzo Laboratories, Farmingdale, NY) for 16 h at 37°C. The cRNA product was purified by RNeasy column and then quantitated by UV absorbance at 260 nm. 15 g of cRNA was fragmented for 35 min at 95°C and then added to a 300-l hybridization mixture containing bovine serum albumin (0.5 mg/ml) and herring sperm DNA (0.1 mg/ml; Promega, Madison, WI) in 1ϫ MES. To estimate the sensitivity of the oligonucleotide arrays, we included 11 in vitro synthesized transcripts (spiked transcripts) in each hybridization (37). 200 l of hybridization mixture was applied to a Ye6100 subA GeneChip (Affymetrix, Santa Clara, CA), and hybridization was allowed to proceed for 20 h at 45°C on a rotisserie. The sample was then hybridized sequentially to the Ye6100 subB, subC, and subD designs, comprising ϳ6400 yeast genes and open reading frames. When hybridization was complete, arrays were stained with streptavidin-conjugated phycoerythrin (Molecular Probes, Inc., Eugene, OR) as described (36). Fluorescence intensity was quantitated using the Affymetrix GeneChip laser scanner.
The resulting array images were captured in the GeneChip version 3.3 software package and reduced to relative expression values (average difference values) for each transcript. The spiked transcripts were used to generate a standard curve of concentration versus their average difference values. The abundance of each transcript (stated in terms of control transcripts per total transcripts) ranged from 1:300,000 to 1:1,000, calculated by assuming an average RNA size of 1,000 ribonucleotides. This standard curve was then used to determine the absolute expression level of the yeast transcripts and is presented as RNA copies per million total transcripts. Based on the signal response from these control transcripts, the sensitivity of the arrays ranged between 1:100,000 and 1:200,000. Consequently, expression values below 10 RNA copies per million total transcripts are considered below the limit of accuracy. Final data analysis was performed using Excel (Microsoft Corp., Redmond, WA). Pairwise comparisons generated -fold change values. -Fold change values of Ͼ4 were considered to be significant.
The arrays include probe sets representing the 5Ј and 3Ј regions of the ␤-actin transcript. The 5Ј to 3Ј signal ratios were Ͼ0.7 across all arrays, indicating that the source RNA was of suitable quality.
Two-hybrid Screening-Transformants containing the N-Sst2-Gal4 fusion plasmid were mated to a set of ϳ6,000 colonies, each expressing a unique full-length open reading frame fused to the Gal4 activation domain, as described previously (20,35). The resulting diploids were transferred to selective plates deficient in histidine and monitored after 10 days. Proteins identified in two independent screens were analyzed further.
Growth, Transcription, and Phosphorylation Bioassays-For NaCldependent growth inhibition, saturated cultures were diluted to A 600 ϳ 0.2 and grown to A 600 ϳ 0.8 and then treated with either water or 10 M ␣-factor (final concentration) for 2 h. 10 l of cells were spotted onto solid medium containing 0.75 M NaCl (where indicated) either without dilution or diluted 1:10, 1:100, 1:1,000, 1:10,000, and 1:100,000 with water or with 10 M ␣-factor (where indicated).
For the pheromone-dependent growth inhibition assay (halo assay), overnight cultures were grown in selective media, and 100 l was diluted with 2 ml of sterile water, followed by the addition of an equal volume of 1% (w/v) dissolved agar (55°C), and poured onto an agar plate containing the same medium. Sterile filter discs were spotted with synthetic ␣-factor pheromone and placed onto the nascent lawn to induce growth arrest. The resulting zone of growth-arrested cells was documented after 2 days.
For pheromone-dependent reporter transcription assays (29), strains were grown for 36 h in standard dextrose-selective medium and then diluted in selective medium containing galactose to induce expression of Vps36, Ste4, or Ste12. Mid-log phase cells were then aliquoted (90 l) to a 96-well plate and mixed with 10 l of ␣-factor for 90 min in quadruplicate. For HSP12 reporter transcription assays, strains were grown in selective medium to mid-log phase and then aliquoted (85 l) to a 96-well plate and mixed with 15 l of 5 M NaCl for 10 min in triplicate. Cells to be treated with NaCl were maintained at room temperature instead of 30°C to reduce basal activity of the stress response promoter. ␤-Galactosidase activity was measured by adding 20 l of a freshly prepared solution of 83 M fluorescein di-␤-D-galactopyranoside (Molecular Probes, Inc.; 10 mM stock in Me 2 SO), 137.5 mM PIPES, pH 7.2, 2.5% Triton X-100, and incubating for 90 min at 37°C. The reaction was stopped by the addition of 20 l of 1 M Na 2 CO 3 , and the resulting fluorescence activity was measured at 485-nm excitation, 530-nm emission.
Bioinformatics-For each gene demonstrated to be differentially expressed in the microarray analysis, a region upstream of the translation start site (to the nearest stop codon, up to 500 bp) was analyzed for sequence motifs representing possible promoter regulatory elements. Most regulatory elements in yeast are found within this region (38). Alignace was used to search for conserved motifs (39). Genespring (Silicon Genetics, Redwood City, CA) was used to identify predefined motifs. Genespring was also used for statistical analysis, comparing the frequency of a motif in a gene list with the upstream regions of all yeast genes. The Transfac data base (Biobase, Braunschweig, Germany) was used to search for previously identified transcription factor binding sites.
A number of Vps36 protein homologues were identified from the nonredundant GenBank TM data base using the advanced BLAST algorithm and the full-length Vps36 protein sequence as the query. PSI-BLAST was used to detect weaker, but nonetheless biologically significant, homologues to Vps36 (40). Multiple alignments of all the Vps36 homologues were performed with ClustalW (41). The alignment was formatted in Phylip and imported into TREEVIEW for visualization of the Vps36 family (42). Classical basic nuclear localization signals were identified using the World Wide Web version of PSORT II (43). The more sensitive hidden Markov model algorithm utilized in the SMART data base was used to detect additional signaling domains and motifs within Vps36 and its homologues (16).

RESULTS
All signaling pathways regulate gene expression. For instance, in yeast, G protein activation leads to the induction of genes with a pheromone response element (PRE) in their promoter region. One of the induced genes encodes Sst2, which is well known to attenuate G protein signaling through its GTPase accelerating function (44). Regulation of the G protein requires the C-terminal RGS domain, C-Sst2 (18,30). The function of the N-terminal domain is not known. Our aim was to determine whether N-Sst2 regulates a distinct signaling pathway, perhaps independently of the G protein. To this end, we sought to identify genes whose expression changed substantially in cells containing just the N-Sst2 domain. Our approach was to use oligonucleotide probe microarrays to monitor the mRNA levels of all yeast genes, comparing cells that express N-Sst2 with cells that lack the N-Sst2 domain (express C-Sst2 alone).
An sst2⌬ mutant strain was transformed with plasmids containing either the N-Sst2 segment (residues 1-392 plus a Myc epitope tag) or the C-Sst2 segment (residues 411-698), as described previously (18). Expression was verified by immunoblot analysis with anti-Myc and anti-Sst2 antibodies, as well as by in vivo complementation of the sst2⌬ mutation, which requires co-expression of both N-Sst2 and C-Sst2. To provide uniform expression under various growth conditions, a constitutive promoter from ADH1 was used in place of the native (PRE-containing) promoter from SST2. Cultures in mid-log phase were collected, and total RNA was isolated. Biotin-labeled cRNA was prepared and hybridized to an Affymetrix GeneChip set representing ϳ6,400 genes and open reading frames over four separate chips. The arrays were then treated with streptavidin-conjugated phycoerythrin, and fluorescence intensity was measured using the Affymetrix GeneChip laser scanner, as described under "Experimental Procedures." Expression profiles in cells containing N-Sst2 or C-Sst2 are shown are Table I. A comparison of N-Sst2 versus C-Sst2 revealed a Ͼ4-fold difference in 53 independent transcripts (0.82% of all genes), of which 40 increased and 13 decreased ("induced" and "repressed" by N-Sst2, respectively). Fewer differences were observed in cells treated with pheromone or when comparing N-Sst2 or C-Sst2 with full-length Sst2 (Fig.  1). 2 Thus, N-Sst2 is necessary for the transcriptional regulation of a discrete set of genes.
We then examined whether the 53 genes regulated by N-Sst2 share common elements in the promoter region. This analysis revealed a CCCCT motif in 32 of the 40 genes (80%) induced by N-Sst2 (Table II). The CCCCT motif is a core consensus sequence of the stress response element (STRE), also known as a UAS PDS (45)(46)(47). In addition, multiple copies of the motif were identified in 21 of the 40 N-Sst2-induced genes. In 15 of these genes, the CCCCT motifs were Ͻ60 bp apart. A similar analysis of the entire yeast genome revealed multiple copies of the motif in 515 of 6,144 genes (8.4%), 212 of which are within 60 bp of one another (3.4%). Previous gene profiling analysis has demonstrated that most genes induced upon treatment with NaCl contain the CCCCT sequence (48). The sequence CCCCT functions in both directions (49). Of the 32 genes identified here, the motif was found in both the sense (49%) and antisense (51%) orientations.
Microarray analysis provides a convenient measure of transcriptional regulation of a large number of genes. However, other methods are better suited to measure transcriptional regulation of specific genes, particularly under various physiological conditions or genetic backgrounds. To expand our analysis of N-Sst2 signaling, we turned to a reporter transcription assay composed of the HSP12 promoter and lacZ (␤-galactosidase) gene. This promoter was selected because the microarray data had indicated a substantial (8.6-fold) difference in HSP12 transcript levels in cells that express N-Sst2 versus C-Sst2. Moreover, the promoter region of HSP12 contains five STREs and has been previously used to monitor gene regulation in response to heat, high salt, and high osmolarity stress conditions (50).
We initially compared HSP12-lacZ induction in an sst2⌬ 2 P. Flanary and H. Dohlman, manuscript in preparation. Thus, the results of the reporter transcription assay corroborate the differences observed by microarray analysis, in which the basal level of expression was highest for N-Sst2 and lowest for C-Sst2. An intermediate basal activity was observed for full-length Sst2 (Fig. 1). We then examined if N-Sst2 could modulate HSP12-lacZ induction by exposure to high concentrations of salt (0.75 M NaCl), a known activator of the stress response pathway. As shown in Fig. 1, there was minimal salt induction of HSP12-lacZ in cells expressing N-Sst2. In contrast, there was a larger induction in cells expressing full-length Sst2 (2.6-fold), C-Sst2 (1.7-fold), or no Sst2 (vector control, 1.8-fold) (Fig. 1). These data indicate that N-Sst2 is a potent activator of the stress response pathway, but the high basal activity leads to a diminished salt induction.
High concentrations of NaCl are known to inhibit cell growth, in addition to stimulating expression of STREs. Moreover, a number of mutants with disrupted signaling to STRE genes will grow poorly in high osmolarity medium (50). Thus, we examined whether N-Sst2 (or C-Sst2) has any effect on growth in high salt. Saturated cultures were diluted and spotted onto solid medium, either in the absence or presence of 0.75 M NaCl. In the absence of salt, cells expressing full-length Sst2, N-Sst2, or vector grew equally well (Fig. 2, top left panel). The growth of these strains was impaired to a similar extent in salt-containing medium (Fig. 2, bottom left panel). In contrast, cells expressing C-Sst2 grew more poorly than the other transformed strains, in the absence or presence of salt. These results parallel the HSP12-lacZ reporter transcription data presented in Fig. 1. Since the C-Sst2-containing cells are unable to ex-

TABLE II Genes induced by N-Sst2 contain a stress-response element
Alignace was used to identify motifs of 8 nucleotides in length within the 5Ј intergenic region of each open reading frame. Genespring was used to search for highly conserved sequence elements, and for statistical analysis. The Transfac database was used to identify potential transcription factor binding sites. Starting positions of the CCCCT and CCCCTTAT motifs are provided, relative to the initiator ATG. N/C, ratios of transcript numbers for N-Sst2 and C-Sst2.

Stress Response Signaling by Sst2
press normal amounts of an STRE-containing gene, they are evidently unable to mount a full response to stress growth conditions. Even in normal medium, the growth of the C-Sst2 cells is impaired, as if they were exposed to salt. Conversely, N-Sst2 expression mimics the transcriptional induction observed with salt treatment, and these cells are able to grow well in the absence or presence of high salt concentrations.
We then examined if pheromone treatment would alter the growth of cells expressing N-Sst2 or C-Sst2, in the absence or presence of salt. Pheromone is known to impair growth, leading to cell cycle arrest in G 1 . Cells lacking SST2 are supersensitive to pheromone-induced growth arrest. However, we have previously shown that neither N-Sst2 nor C-Sst2 alone can rescue an sst2⌬ mutation (18). Consistent with these earlier observations, cells expressing N-Sst2, C-Sst2, or vector grew poorly in the presence of pheromone, compared with the full-length protein (Fig. 2, top right panel). Pheromone-treated cells expressing the gain-of-function mutant Sst2 P20L grew better than those with full-length Sst2, as previously reported (51). Cells expressing C-Sst2 grew poorly in the presence or absence of high salt, and even more poorly in the presence of salt plus pheromone (Fig. 2, bottom right panel). This pattern of additive growth inhibition was evident throughout and is consistent with separate and additive mechanisms of action.
Genetic studies have revealed at least two osmosensing pathways that converge on the MAP kinase kinase Pbs2, leading to tyrosine phosphorylation of the MAP kinase Hog1 (52). MAP kinases are the only tyrosine-phosphorylated proteins in yeast, and Hog1 is the only MAP kinase phosphorylated in response to salt stress (52). Thus, we examined if N-Sst2 or C-Sst2 have any effect on Hog1 phosphorylation, by immunoblotting whole cell extracts with anti-phosphotyrosine antibodies. As shown previously (52), salt treatment leads to a dramatic increase in Hog1 phosphorylation (Fig. 3). However, Hog1 phosphorylation is largely unaffected by expression of N-Sst2 or C-Sst2, in the absence or presence of added salt. These results are in contrast to the reporter transcription assay presented above, in which N-Sst2 stimulated, and C-Sst2 inhibited, basal expression of HSP12-lacZ. These data together with those presented in Fig.  2 suggest that N-Sst2 acts through another, Hog1-independent, pathway.
One way to determine the biological role of N-Sst2 is through the identification of associated proteins. To this end, we carried out a two-hybrid screen against an array of nearly all yeast open reading frames. A strain expressing N-Sst2 fused to the Gal4 DNA-binding domain was mated to a set of ϳ6,000 colonies, each expressing a unique full length open reading frame fused to the Gal4 activation domain. Any proteins identified in two independent screens were analyzed further. As shown in Table III, N-Sst2 reproducibly yielded 17 putative interactions. This compares to an average of 3.3 positives per protein obtained for an independent set of 192 DNA binding domain hybrids, described previously (20). Using an identical screening array, full-length Sst2 yielded no specific positives. Mpt5, a protein shown to bind Sst2 in a previous two-hybrid screen, was not identified in our screen. Genetic analysis revealed that Mpt5 can attenuate pheromone signaling downstream of the G protein and independently of the C-terminal RGS domain (53,54).
Of the 17 putative N-Sst2-binding proteins, five (29%) were listed by the Yeast Protein Data base as unclassified, having no known functional or structural homologues. This value is similar to the percentage of unclassified genes listed throughout the entire data base. At least three of these genes, PEP12, TLG2, and VPS36, are required for proper sorting of vacuolar proteases and normal vacuolar morphology (55)(56)(57)(58)(59)(60)(61)(62)(63)(64). Several other interacting proteins are protein kinases, including Iks1, and a member of the MAP kinase family Kss1 (KSS1 product, kinase suppressor of sst2). Kss1 can phosphorylate Sst2 at Ser 539 , which is located in the C-terminal domain of the protein (65).
We then examined whether any of the potential N-Sst2 binding partners are required to transmit a signal via the stress response pathway. Gene deletion mutants were obtained for 15 of the 17 interacting proteins and evaluated for changes in HSP12-lacZ activity. As shown in Fig. 4, all three of the vacuolar sorting mutants, pep12⌬, tlg2⌬, and vps36⌬, yielded a high basal activity. In addition, these mutants exhibited a diminished induction with salt treatment (1.4-, 1.8-, and 2.1-fold, respectively) as compared with the wild-type strain (2.6-fold induction). This pattern of activity (high basal, low induction) resembles that seen with N-Sst2 in Fig. 1. Our positive control for this assay was a deletion of the high osmolarity glycerol kinase gene HOG1 (52). Like N-Sst2 and the binding partner mutants, the hog1⌬ strain exhibited a diminished induction with salt (2.1-fold). In contrast to the other mutants, however, hog1⌬ exhibited a normal or slightly reduced basal activity. We also tested four other vacuolar sorting mutants, vps5⌬, vps20⌬, vps21⌬, and vps25⌬. Vps5 was chosen because it contains a phosphoinositide-binding Phox homology (PX) domain, which is found in the mammalian RGS protein RGS-PX1 (66). Vps20 and Vps25 were reported previously to interact with Vps36 in two-hybrid assays (20,21). Vps21 was chosen arbitrarily as a negative control to rule out the possibility that altered STRE signaling is a generalized consequence of impaired vacuolar function. As shown in Fig. 4, all four mutants exhibited normal basal and salt-induced activities. Taken together, these data indicate that PEP12, TLG2, and VPS36 are necessary for full activation of the stress response pathway but act in a manner distinct from the well characterized Hog1 kinase.
Sst2 is well known to regulate pheromone signaling. An sst2⌬ mutant can enhance pheromone sensitivity by Ͼ100-fold. Thus, we then examined if any of the 15 candidate binding partners could also regulate the pheromone response. For these experiments, two standard bioassays were used. In the halo assay, cells are spread onto solid media and exposed to ␣-factor pheromone spotted onto filter disks. The resulting zone of growth inhibition gives an indication of pheromone response (halo size) and recovery (halo turbidity). Of the 15 mutants tested, only vps36⌬, pep12⌬, and tlg2⌬ produced more turbid zones of growth inhibition, as compared with the wild-type control, indicating an enhanced ability to recover from pheromone-induced growth arrest. An additional mutant (kel1⌬) had the opposite effect, producing halos that were slightly larger and less turbid than the wild-type control (Fig. 5A).
A number of vacuolar sorting mutants have previously been reported to produce turbid halos (67). Halo turbidity in the vacuolar sorting mutants could result from missorting and FIG. 3. Stress-mediated Hog1 phosphorylation: N-Sst2 versus C-Sst2. An sst2⌬ mutant strain (YDM400) was transformed with a plasmid containing no insert (Vector; pRS316-ADH), N-SST2, C-SST2, or full-length SST2. Cells were grown to mid-log phase and treated with NaCl or water for 60 min. Cell lysates were prepared and subjected to 8% SDS-PAGE and immunoblotting with anti-phosphotyrosine antibodies to detect phosphorylated-Hog1 (P-Hog1). secretion of vacuolar proteases and the consequent proteolysis of pheromone or the pheromone receptors. Indeed, we found that each of the vps mutants tested (vps5⌬, vps20⌬, vps21⌬,  vps25⌬, vps35⌬, vps38⌬) also yielded slightly turbid halos (data not shown). To provide an independent assessment of pheromone sensitivity, we tested each of the candidate binding partners using a pheromone-responsive transcription reporter assay (FUS1 promoter, lacZ reporter). In agreement with the results of the halo assay, vps36⌬, pep12⌬, and tlg2⌬ exhibited a diminished transcription response (Fig. 5B, top panel). The effects were particularly dramatic for the vps36⌬ and tlg2⌬ mutants, with reductions of 30 and 50%, respectively. Deletion of two candidate Vps36-binding partners (vps20⌬, vps25⌬) also resulted in a diminished response, equal to or greater than that exhibited by vps36⌬ (Fig. 5B, middle). One of the control mutants (vps5⌬) responded like wild-type, while a second (vps21⌬) had a diminished response. Therefore, two additional control mutants (vps35⌬ and vps38⌬, selected arbitrarily) were tested and found to also respond like wild-type (Fig. 5B, bottom). Thus, in addition to their role in stress response signaling, VPS36, TLG2, and PEP12 are also necessary for full activation of the pheromone response pathway. The pheromone signaling phenotype is seen with some but not all vps mutants. The stress signaling phenotype is not shared by any of the other vps mutants. Taken together, these data indicate that Vps36, Tlg2, and Pep12 function in a manner similar to N-Sst2 but distinct from other vacuolar sorting factors.
Because it has an especially strong pheromone signaling phenotype and because it has not been well characterized previously, we selected Vps36 for further analysis. We first showed that a plasmid-borne copy of Vps36 could reverse the pheromone-resistant phenotype of the vps36⌬ mutation (Fig. 6A). Overexpression of Vps36 in a wild-type strain did not further enhance signaling and even had a modest inhibitory effect (Fig.  6, A and B). However, overexpression of Vps36 in cells lacking SST2 did result in a dramatic elevation of pheromone sensitivity (Fig. 6B). The signal-enhancing effects of Vps36 may be unmasked when Sst2 protein levels are low, as occurs in the absence of pheromone (30).
Having shown that Vps36 activity is diminished by Sst2, we next examined whether Sst2 activity is similarly dependent on Vps36. Overexpression of full-length SST2 is well known to inhibit the pheromone response (30). Overexpression of Sst2 also reduces the already diminished response of the vps36⌬ strain (Fig. 7), indicating that Sst2 can inhibit signaling in the absence of Vps36. In contrast, overexpression of N-Sst2 has no TABLE III Genome-wide two-hybrid analysis using N-Sst2 as bait A strain expressing N-Sst2 fused to the Gal4 DNA-binding domain was mated to a set of ϳ6,000 colonies, each expressing a unique full-length open reading frame fused to the Gal4 activation domain, as described. Proteins identified in two independent screens with the same bait are listed. Only genes identified in 10 or fewer screens are listed, since those identified at a higher frequency are likely to be nonspecific ("false") positives. Functional descriptions and number of amino acids (aa) are from the YPD database (note that MVP1 is erroneously listed in YPD as being required for vacuolar protein sorting). Predicted conserved sequence motifs are from the SMART data base (16,93 4. Loss of N-Sst2-binding proteins induces a stress-activated promoter. Wild-type and gene disruption mutants were transformed with a plasmid containing the stress-activated HSP12 promoter and lacZ reporter gene (pRS423-HSP12-lacZ). Cells were grown to mid-log phase and treated with 0.75 M NaCl or water for 10 min, and the resulting ␤-galactosidase activity was measured as described under "Experimental Procedures." Data shown are typical of three independent experiments performed in triplicate. Error bars, ϮS.E. effect on signaling in a wild-type strain (Fig. 2) (18). Presumably, this is because N-Sst2 lacks the GTPase accelerating activity associated with the C-terminal RGS domain. Remarkably, N-Sst2 is a potent inhibitor in the vps36⌬ mutant and is even more active than full-length Sst2 (Fig. 7). Thus, a signaldampening effect of Sst2 is unmasked when Vps36 is absent, and this effect resides in the N-Sst2 domain. These data suggest an antagonistic relationship between Vps36 and Sst2, in their ability to regulate the pheromone response pathway.
Sst2 is well known to attenuate signaling through its ability to accelerate G␣ GTPase activity. Thus, we examined whether Vps36 also acts through the G protein ␣ subunit or if it affects signaling downstream of the G protein. G␣-independent signaling was achieved through overexpression of STE4 (G␤), mutational activation of STE11 (effector kinase), and overexpression of STE12 (transcription factor). Overexpression of STE4 results in elevated levels of the G␤␥ complex, above that which can bind to G␣ (68 -70). The STE11-4 mutant encodes a constitutively active form of the mitogen-activated protein kinase kinase kinase, Ste11 (34). Overexpression of STE12 results in elevated transcription of PRE-containing genes (71). As shown in Fig. 8, the vps36⌬ mutant attenuated signaling by at least 25% in every case, whether or not ␣-factor was added. These data suggest that Vps36 can function downstream of the G protein, independently of Sst2 RGS activity, and most likely at the level of transcription. DISCUSSION The GTPase accelerating activity of RGS proteins is well established. However, many RGS family members are likely to have other functions as well. Our goal was to identify a possible signaling function for the N-terminal (non-RGS-homologous) domain of Sst2. Yeast has specific advantages for this type of investigation, since nearly every gene has been arrayed for RNA expression studies, subjected to two-hybrid analysis, and genetically disrupted. Here, a comprehensive analysis of transcription indicates that N-Sst2 regulates a pathway leading to STRE activation. Comprehensive two-hybrid analysis has revealed candidate targets of N-Sst2 action. The functional significance of some of these interactions was established using gene disruption mutations, in conjunction with functional assays of stress-and pheromone-mediated signaling in yeast. Of the 17 proteins identified as potential N-Sst2 interactors, at least one has previously been implicated in STRE activation. Sho1 is thought to act by recruiting the MAP kinase kinase Pbs2 to the plasma membrane (72)(73)(74). Pbs2 phosphorylates Hog1, which in turn activates the transcription factors Msn2 and Msn4 (48,75). Msn2/4 are known to bind to the CCCCT motif present in stress-activated genes (46,47,76,77). Not surprisingly, deletion of MSN2 and MSN4 results in poor growth and decreased induction of STRE-regulated genes upon exposure to high osmolarity media, heat shock, nutrient limitation, and oxidative stresses (46,47,77). However, our analysis indicates that binding of N-Sst2 to Sho1 is of little functional consequence, at least with respect to Hog1 phosphorylation (Fig. 3). Moreover, the pattern of HSP12-lacZ FIG. 5. Loss of N-Sst2-binding proteins inhibits the pheromone response pathway. A, wild-type and mutant strains were plated, and the nascent lawn was exposed to sterile filter discs spotted with ␣-factor (from bottom clockwise: 15, 25, 50, and 75 g for 48 h) and then photographed. The wild-type strain BY4741 is a negative control, and the isogenic fus3⌬ strains are a positive control. Other mutants tested but not shown (vps5⌬, vps20⌬, vps21⌬, vps25⌬, vps35⌬, and vps38⌬) also produced slightly turbid halos. B, wild-type and mutant strains were transformed with a plasmid containing the pheromone-responsive FUS1 promoter-lacZ reporter. Cells were then treated with the indicated concentrations of ␣-factor, and the resulting ␤-galactosidase activity was measured as described under "Experimental Procedures." Data shown are typical of three independent experiments performed in triplicate. Error bars, ϮS.E. activation by N-Sst2 resembles that of the N-Sst2 binding partner mutants (high basal, low salt induction) ( Fig. 1) but is substantially different from that of the hog1⌬ mutant (normal basal, low salt induction) (Fig. 4).
Another pathway suggested to contribute to the stress response involves Gpa2 (the only other G protein in yeast besides Gpa1) (78). Gpa2 is activated by a putative glucose receptor, Gpr1, and transmits a signal leading to activation of adenylyl cyclase and the cAMP-dependent protein kinase. Gene profiling studies have been conducted following activation of the cAMP-cAMP-dependent protein kinase pathway (79) and have revealed 17 genes that are repressed by cAMP, 10 of which are dependent on MSN2 and MSN4 expression. We noted that eight of the cAMP-regulated genes show at least a modest increase in cells expressing N-Sst2, and two of these increase by at least 2-fold: YAK1 (YJL141C) and YHR033. In comparison, over half the genes induced by N-Sst2 were previously identified as being induced after a shift to high osmolarity medium (48). Our data suggest that STREs may be activated by a distinct signaling pathway that is dependent on N-Sst2. Whole genome two-hybrid analysis revealed three components that also modulate transcription of an STRE-containing gene. Prior data have indicated that all three proteins are necessary for different aspects of protein trafficking and vacuolar function in yeast cells. Tlg2 is a syntaxin (t-SNARE) that functions in transport from the endosome to the late Golgi within the endocytic pathway (55,56). Pep12 is a syntaxin that is required for protein sorting between the Golgi and endosome (57,58). Vps36 is one of a diverse class of gene products needed for protein trafficking from the pre-vacuolar compartment to the vacuole (59 -62). Notably, an earlier screen yielded a number of mutants with altered vacuolar function and diminished growth in high salt (63,64). We have observed that this salt-sensitive growth phenotype is shared by the vps36⌬, tlg2⌬, and pep12⌬ mutants (data not shown). This could result from altered signaling to STREs or more likely from defects in the transport of proteins that mediate osmotic regulation. STRE activation is not a general phenotype of vacuolar sorting mutants, however (Fig. 4).
An important question is whether the Vps36, Tlg2, and Pep12 bind physically to N-Sst2 or if they exert their documented functional effects through some common bridging protein. So far, direct binding of N-Sst2 to Vps36 and Tlg2 has been demonstrated using synthetic peptide arrays (80). 3 This analysis revealed binding to three discontinuous peptide "epitopes" within Vps36 and a single segment within Tlg2. This approach is also currently being used to identify the epitope(s) within N-Sst2 recognized by Vps36 and Tlg2.
(pSTE11-4) (B), or overexpression of the transcription factor Ste12 (pSTE12) (C). Cells were treated with the indicated concentration of ␣-factor, and the resulting ␤-galactosidase activity was measured as detailed under "Experimental Procedures." Data shown are typical of three independent experiments performed in quadruplicate. Error bars, ϮS.E. membrane-restricted phosphoinositides (85). In this regard, sorting nexins interact directly with endocytosed receptors, such as receptor tyrosine kinases activated by epidermal growth factor, but also have more general effects on endosomal traffic. RGS-PX1 was shown to accelerate GTP hydrolysis and inhibit signaling by G s ␣ and also to delay lysosomal degradation and inactivation of the epidermal growth factor receptor (66). Because of its bifunctional role as both a GTPase-accelerating protein and as a sorting nexin, RGS-PX1 may link heterotrimeric G protein signaling and vesicular trafficking in mammals. Sst2 might similarly link G protein signaling and vacuolar sorting in yeast. Sst2 is expressed only in haploid cells, however, so any membrane trafficking function would probably occur only in conjunction with mating. Another possibility is that Sst2 and Vps36 regulate transcription. Recently, Vps36 was reported to bind to Snf8, Vps25 (Yjr102), and Vps20 (20,21). The human RNA polymerase II elongation factorassociated proteins EAP45 (or CGI-145, Fig. 9A), EAP30, and EAP20 appear to be homologous with Vps36, Snf8, and Vps25, respectively. The human EAPs form a complex that can inhibit elongation factor repression of RNA polymerase II activity (86 -88). Likewise, a similar complex of Vps36, Snf8, and Vps25 might derepress RNA polymerase II in yeast. This could explain the absence of SUC2 derepression in snf8⌬, vps36⌬, and vps25⌬ mutants (86,89). It could also explain how VPS36 can coordinately regulate vacuolar trafficking and pheromone signaling. Vps36 must regulate STRE transcription by a distinct pathway, however, since the effects on this promoter occur independently of putative binding partners Vps25 and Vps20. Snf8 was not tested because of technical difficulties with the reporter transcription assay in the available knockout strain. Thus, our genetic analysis in yeast and parallel studies in human cells suggest that Vps36 acts at the level of transcription. Notably, we have identified a simple pattern-7 nuclear localization signal (NLS; Fig. 9B) and two novel zinc finger Ran-GDP binding domains (RBZ domain; Fig. 9B). Residues 120 -186 could also form a RING finger domain, which is a specialized type of zinc finger involved in protein-protein interactions, including binding to E2 ubiquitin-conjugating enzymes (90). The RBZ domain may serve to recruit Vps36 to Ran-GDP. RanGDP is found nearly exclusively in the cytoplasm and the cytoplasmic face of the nuclear pore complex (91). The nuclear localization signal suggests that Vps36 may exist in the nucleus or shuttle between the cytoplasm and nucleus via the nuclear pore complex (92). However, while the RBZs and nuclear localization signal are present within the S. cerevisiae and Schizosaccharomyces pombe proteins, they are absent from Vps36 homologues in higher organisms (see Fig. 9A; data not shown).
In conclusion, Sst2 is one of a growing list of RGS proteins with at least two signaling functions. We have shown that Vps36 and the N-Sst2 domain can cooperatively regulate both the pheromone and stress response pathways. Whereas the pheromone response is inhibited by N-Sst2 and activated by Vps36, the stress response is activated by N-Sst2 and inhibited by Vps36 (Fig. 9C). The identification of candidate N-Sst2 binding proteins and their demonstrated role in stress signaling will be extremely useful in addressing the mechanism by which these proteins function within the cell. The approach used here also serves as a model for an integrated analysis of signaling pathways in other systems. Such an approach will be important for the identification and characterization of a large number of unknown gene products as they are identified through genome sequencing programs.