Feedback Phosphorylation of an RGS Protein by MAP Kinase in Yeast*

Regulators of G protein signaling (RGS proteins) are well known to accelerate G protein GTPase activity in vitro and to promote G protein desensitization in vivo. Less is known about how RGS proteins are themselves regulated. To address this question we purified the RGS in yeast, Sst2, and used electrospray ionization mass spectrometry to identify post-translational modifications. This analysis revealed that Sst2 is phosphorylated at Ser-539 and that phosphorylation occurs in response to pheromone stimulation. Ser-539 lies within a consensus mitogen-activated protein (MAP) kinase phosphorylation site, Pro-X-Ser-Pro. Phosphorylation is blocked by mutations in the MAP kinase genes (FUS3, KSS1), as well as by mutations in components needed for MAP kinase activation (STE11, STE7, STE4, STE18). Phosphorylation is also blocked by replacing Ser-539 with Ala, Asp, or Glu (but not Thr). These point mutations do not alter pheromone sensitivity, as determined by growth arrest and reporter transcription assays. However, phosphorylation appears to slow the rate of Sst2 degradation. These findings indicate that the G protein-regulated MAP kinase in yeast can act as a feedback regulator of Sst2, itself a regulator of G protein signaling.

All eukaryotic cells respond to external signals through cell surface receptors linked to heterotrimeric G proteins. In humans, G protein-linked receptors can detect hormones, neurotransmitters, and sensory stimuli (odors, taste, light). In the yeast Saccharomyces cerevisiae, G protein linked pheromone receptors mediate events required for mating and cell fusion. Upon receptor stimulation, the G protein binds GTP and undergoes subunit dissociation. The G␣ subunit or the G␤␥ subunit dimer can then propagate the signal through a variety of effector enzymes or ion channels. Upon GTP hydrolysis, the G protein subunits reassociate and signaling stops. Additional proteins can modulate this cycle of G protein activation and inactivation. For instance, G protein-coupled receptor kinases (GRKs) 1 and arrestins promote desensitization through phosphorylation, uncoupling, and internalization of the receptor (1).
It is now evident that G proteins are also subject to desensitization and that this process involves members of the RGS protein family. The contribution of RGS proteins in vivo has been established through genetic analysis in yeast and nematodes. In yeast, disruption of the RGS gene SST2 can increase pheromone sensitivity by 100 -300-fold (2). The mechanism of RGS action has also been well characterized through detailed biochemical and x-ray crystallographic analysis of purified components (3,4). Stop flow fluorescence measurements reveal that RGS proteins can accelerate G␣ GTPase activity by up to 1000-fold (5).
How are RGS proteins themselves regulated? Some RGS genes, including SST2, are transcriptionally induced by G protein activation (6). Virtually nothing is known about how these proteins are regulated post-translationally, however. In particular, it is not known if RGS proteins are phosphorylated or otherwise modified in response to cell stimulation. To address this question, we examined whether Sst2 undergoes any chemical modifications in vivo. Our analysis reveals that Sst2 is stoichiometrically phosphorylated at Ser-539, in response to pheromone stimulation, and in a Fus3-dependent manner. Phosphorylation appears to stabilize the protein, thereby augmenting the transcriptional induction that occurs in response to pheromone stimulation.
14. pRS315-SST2 contains SST2 under the control of its own promoter (14). Plasmid pGA1905 contains a myc epitope-tagged FUS3 K42R "ki-* This work was supported by the National Science Foundation Science and Technology Center for Molecular Biotechnology, National Institutes of Health Research Resource Grant RR11823, and National Institutes of Health Grants GM55316 and GM59167 (to H. G. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  nase dead" mutant, as described in Ref. 15. SST2 mutants were obtained using the QuickChange mutagenesis kit (Stratagene) and confirmed by DNA sequencing.
Purification-BJ2168 cells transformed with pAD4M-SST2-his were grown to A 600 nm ϳ 0.8 and treated with 2.5 M ␣-factor, as indicated. Cells were chilled, harvested by centrifugation, rapidly frozen in liquid nitrogen, and thawed in urea buffer (6 M urea, 100 mM Na 2 H 2 PO 4 , 10 mM Tris, 10 mM 2-mercaptoethanol, pH 8.0), 250 mM NaCl, 15 mM imidazole (at 250 ml/6 liter of medium) at room temperature. Cells were further disrupted using a stainless steel beadbeater (Biospec) packed in ice and salt, with 10 ϫ 30-s pulses, once every 90 s. The remaining procedures were carried out at room temperature. The disrupted cells were rocked for 60 -90 min, then clarified by centrifugation 3840 ϫ g, 15 min and paper filtration (Whatman No. 1). The soluble material was mixed with 3 ml of equilibrated Superflow Ni 2ϩ -NTA resin (Qiagen) for 60 -90 min, then packed into a HR 10/10 (Amersham Pharmacia Biotech) column and washed using 10 column volumes (CV) of urea buffer, 250 mM NaCl, 15 mM imidazole at 1.5 ml/min, followed by 10 CV urea buffer at 1 ml/min. Sst2 was eluted in 10 CV of urea buffer, 75 mM imidazole at 1 ml/min. The eluate was mixed with 2 ml of equilibrated Mono Q resin (Amersham Pharmacia Biotech) for 60 min, then packed into a HR 10/10 column and washed with 5 CV of urea buffer at 1 ml/min, then 7.5 CV of urea buffer and a linear gradient (0 -60 mM) of NaCl at 0.5 ml/min and 7.5 CV 60 mM NaCl urea buffer at 0.5 ml/min. Sst2 was eluted with 15 CV urea buffer, 1 M NaCl at 0.5 ml/min. Peak fractions were pooled, concentrated, and desalted using an Ultrafree-30 (Millipore) filter. The final purified product (500 l) was resolved by sodium dodecyl sulfide-polyacrylamide gel electrophoresis (SDS-PAGE) and silver staining (Silver Stain Plus, Bio-Rad). Except where indicated, 16-cm 8% acrylamide gels were used to resolve the 82-and 84-kDa species of Sst2. Antibodies and conditions for immunoblot detection of Sst2 are described elsewhere (7).
Sample Preparation-Each silver-stained protein band (ϳ1 g) was excised (20-cm gel) and transferred to a 0.5-ml microcentrifuge tube and destained with a 1:1 solution of 30 mM potassium ferricyanide and 100 mM sodium thiosulfate for 2 min. The gel was washed four times with Milli-Q water, cut into small pieces, washed three times for 8 min each with 12.5 mM ammonium bicarbonate, 50% acetonitrile and Speedvaclyophilized. Samples were reconstituted with 40 l of 12.5 ng/ml endoproteinase Asp-N or pepsin (Roche Molecular Biochemicals), incubated on ice for 40 min, then at 37°C overnight. Peptides were extracted three times with 40 l of 5% acetic acid, 50% acetonitrile. All four supernatants were pooled, and the solution was dried under vacuum and reconstituted in 10 l of capillary electrophoresis buffer (10 mM acetic acid, 10% MeOH).
Mass Spectrometry-Solid phase extraction (SPE) capillary electrophoresis electrospray ionization tandem mass spectrometry was performed essentially as described (16). The SPE cartridge consisted of Spherisorb C-18 material between two Teflon membranes inside a Teflon tube (17). The reconstituted peptide mixture was loaded on the SPE cartridge at 1 l/min. The concentrated peptides were washed for 10 min and eluted by applying a small plug of 67% acetonitrile, 3 mM acetic acid across the SPE cartridge. Separation was achieved by capillary electrophoresis in a 50 m inner diameter and 150 m outer diameter fused silica capillary 40 cm in length. Separated analytes were electrosprayed into a Finnigan MAT LCQ ion trap mass spectrometer and detected in the MS mode. Ion signals above a predetermined threshold automatically triggered the instrument to switch from MS to MS/MS mode for generating collision-induced dissociated spectra (datadependent MS/MS). The ion trap was run with automatic gain, and the MS scan was performed in three segments from 400 -900, 900 -1400, and 1400 -2000 atomic mass units, respectively. To determine the phosphorylation site in the protein, the generated collision-induced dissociated spectra were searched against a yeast genomic sequence data base using the SEQUEST search program (18).

RESULTS
To determine whether Sst2 undergoes post-translational modification, we isolated a His-tagged version of the protein from both pheromone-stimulated and -unstimulated cells. Cells were disrupted in a denaturing buffer containing 6 M urea, so as to fully solubilize Sst2 and to preserve any modifications. Sst2 was then purified by sequential Ni 2ϩ -NTA affinity and Mono Q-Sepharose ion exchange chromatography, and resolved by gel electrophoresis. As shown in Figs. 1 and 2, Sst2 from control (unstimulated) cells migrates at the predicted size of ϳ82 kDa, while Sst2 from pheromone-stimulated cells migrates as a doublet of ϳ82 and 84 kDa. Both products were confirmed by immunoblotting with anti-Sst2 antibodies (Fig.  1A, bottom) and by mass spectrometry sequencing (Fig. 1B).
The appearance of a new higher molecular weight species suggested that Sst2 undergoes some form of stimulus-depend-FIG. 1. Purification and mass spectrometry of Sst2. His-tagged Sst2 (in plasmid pAD4M) was expressed in BJ2168 cells and purified by Ni 2ϩ -NTA ("first column") and Mono Q-Sepharose ("second column") chromatography. A, pooled and concentrated protein fractions were subjected to SDS-PAGE (8% acrylamide, 7-cm gel) and silver-stained. Sst2 (arrows) purified from ␣-factor-stimulated cells migrates as a doublet of molecular mass ϳ82 and 84 kDa (top panel, "silver stain"). An identical gel was immunoblotted using anti-Sst2 antibodies, to confirm that the two bands represent Sst2 (bottom panel, "Sst2 Ab"). B, purified protein was excised from a 20-cm silver-stained gel, subjected to limited proteolysis, and analyzed by SPE capillary electrophoresis electrospray ionization tandem mass spectrometry. Using either Asp-N (shown) or pepsin (not shown), the pheromone-treated 84-kDa band yielded a phospho-Ser that was absent in the treated and untreated 82-kDa bands. MS analysis also revealed a second phospho-Ser present in both the 82-and 84-kDa bands, indicating that Sst2 undergoes constitutive as well as pheromone-dependent phosphorylation. 3 ent post-translational modification. To characterize this further, each band was excised from the gel, subjected to limited proteolysis, and analyzed by electrospray ionization tandem mass spectrometry. As indicated in Fig. 1B, a phospho-Ser at position 539 was present in the 84-kDa band. The corresponding nonphosphorylated peptide was obtained from the 82-kDa bands, from either stimulated or unstimulated cells (data not shown).
To confirm that phosphorylation occurs on Ser-539, we replaced this amino acid with Ala, Glu, Asp, or Thr. With the exception of Thr, these residues cannot be phosphorylated by protein kinases. The acidic residues Asp and Glu can often substitute for the negatively charged phospho-Ser (19). As expected, the Ser-539 3 Thr mutation preserved the pheromonedependent mobility shift, while the Ala, Asp, and Glu mutations blocked the shift completely ( Fig. 2A). These data indicate that Sst2 undergoes a stimulus-dependent phosphorylation at Ser-539 and that Ser-539 phosphorylation causes the gel mobility shift.
Ser-539 lies within an ideal consensus sequence (Pro-X-Ser-Pro) for phosphorylation by MAP kinases (19). Indeed, the pheromone signaling pathway involves at least one MAP kinase family member, Fus3 (20). To determine whether Sst2 is a substrate for Fus3 in vivo, we examined whether Sst2 phosphorylation requires Fus3 expression or activity. We first established the time course of Sst2 phosphorylation (Fig. 2B). We then tested whether Sst2 is phosphorylated in cells lacking the FUS3 gene or any of the other known components of the signaling cascade (Fig. 2C). As predicted, Sst2 failed to undergo phosphorylation in cells lacking FUS3, provided that KSS1 (another MAP kinase that can partially rescue a fus3 mutant) was also deleted (21,22). Phosphorylation was not restored by a Fus3 mutant (Lys-42 3 Arg) that lacks kinase activity. Phosphorylation was also abolished by mutations that block expression of the upstream MAP kinase kinase (MAPKK, Ste7) or MAPKK kinase (Ste11). This was expected, since both Ste7 and Ste11 are required for Fus3 activity (23). Phosphorylation was preserved in a Gpa1 Gly-302 3 Ser mutant that cannot bind Sst2 (24), as well as in mutants lacking the MAPKKK kinase Ste20 or the transcription factor (and Fus3 substrate) Ste12. Deletion of the pheromone receptor Ste2 blocks the pheromonemediated response, but also leads to elevated basal signaling (9) and Sst2 phosphorylation (Fig. 2C). Thus, the only components of the pathway necessary for Sst2 phosphorylation are those needed to activate MAP kinase and the MAP kinase itself. Taken together, these findings indicate that Sst2 is phosphorylated in vivo by Fus3.
Loss-of-function mutations in Sst2 are well known to increase pheromone sensitivity (25). To determine whether MAP kinase phosphorylation has any effect on pheromone signaling in vivo, we tested the ability of the Ser-539 mutants to regulate the pheromone response pathway, using a reporter transcription assay. As shown in Fig. 3A, pheromone-induced gene transcription (using the FUS1 promoter linked to the essential HIS3 gene) was unaffected by any of the Ser-539 substitutions. Likewise, no difference was seen using an alternative pheromone bioassay, the growth inhibition plate assay (halo assay, Fig. 3B) (26,27).
Like many cell regulators, Sst2 has a short half-life in vivo (7,28). Some regulatory factors (e.g. cyclins, cyclin inhibitors) are targeted for destruction by phosphorylation and, in some cases, by ubiquitination (28). Thus, we considered whether phosphorylation of Ser-539 contributes to Sst2 proteolysis. Cells were transformed with plasmids containing wild-type or the 539 mutant forms of Sst2, under the control of a constitutive promoter (ADH1, to prevent pheromone-mediated changes in SST2 transcription). Cells were treated with cycloheximide to block new protein synthesis and sampled at various times. As shown in Fig. 3C, the overall levels of Sst2 drop rapidly when translation is blocked, but the decrease is most pronounced for the nonphosphorylated (82-kDa) species. An alternative explanation, that loss of the 82-kDa species is due to increased phosphorylation, is improbable, since there is no concomitant increase in the 84-kDa form of the protein. Surprisingly, all of the 539 mutants degraded at rates comparable with the phosphorylated form of Sst2 (data not shown). These data suggest that Sst2 is rapidly degraded, except when Ser-539 is phosphorylated or replaced with another amino acid. DISCUSSION One of the defining characteristics of desensitization is feedback inhibition. One way this is accomplished is through increased expression of factors that can attenuate the signal. Indeed it is well known that increased expression of Sst2 will inhibit the pheromone response (7). It is also known that pheromone stimulation leads to increased transcription of SST2 (6). Our findings indicate that pheromone stimulation promotes increased stabilization of the protein, and this is achieved in part through phosphorylation at Ser-539. The combined effect of transcriptional induction and protein stabilization should allow for sustained expression of Sst2 and at a lower energy expenditure than transcriptional induction alone. Consistent with this interpretation, we have found that expression of endogenous Sst2 is induced by pheromone (7), and the induced . BJ2168 cells expressing wild-type or mutant Sst2 (pAD4M-SST2-His) were treated with ␣-factor for 1 h ("ϩ"). Cells were lysed in SDS-PAGE sample buffer, and the conversion of Sst2 to the phosphorylated species was monitored by 8% SDS-PAGE and immunoblotting, using anti-Sst2 antibodies (7). B, to determine the time course of Sst2 phosphorylation in vivo, BJSST2 cells containing pRS316-ADH-SST2 were exposed to 2.5 M ␣-factor for the indicated times, lysed in SDS-PAGE sample buffer, and monitored by immunoblotting. C, to determine which signaling components are needed for Sst2 phosphorylation, extracts were prepared from YPH499 or YPH499-based mutant strains, containing pRS316-ADH-SST2, and lacking the ␣-factor receptor (STE2), the G protein ␤ subunit (STE4), the G␥ subunit (STE18), the MAP kinase kinase kinase kinase (STE20), the MAP kinase kinase kinase (STE11), the MAP kinase kinase (STE7), the functionally redundant MAP kinases (FUS3, KSS1), or the downstream transcription factor (STE12). The gpa1 G302S product cannot bind to Sst2, but is otherwise fully functional (24). Fus3 K42R lacks kinase activity (31). Cells were treated with 2.5 M ␣-factor for 1 h as indicated ("ϩ"), and subjected to immunoblotting. protein exists predominantly in the 84-kDa phosphorylated form (data not shown). The Ser-539 mutants do not alter pheromone sensitivity, presumably because they are stabilized even without being phosphorylated.
How do these findings fit with our understanding of RGS structure-function relationships? The crystal structure determination of rat G i1 ␣ complexed with RGS4 provides some important insights (3). The RGS core region contains nine ␣-helices that form two subdomains. The majority of residues that contact G␣ are on the "bottom" of the subdomain comprised of helices 4, 5, 6, and 7. A structurally based alignment of RGS4 and Sst2 reveals that Ser-539 lies between helices 6 and 7, on "top" of this subdomain. Thus it appears that Sst2 is phosphorylated at a site distal from the G protein binding interface and therefore would not be expected to have any direct effect on Sst2-Gpa1 interaction. Correspondingly, Sst2 could undergo phosphorylation even in the Gpa1-bound state. This model is consistent with the lack of any G protein modulatory effects of the Ser-539 mutants. There is also the possibility that phosphorylation modulates some other (G protein-independent) function of Sst2. The adjoining region of Sst2 has a 118-amino acid insert (residues 539 -657). The function of this region is unknown, but it is probably dispensable for GTPase accelerating activity, since it is absent in all other RGS proteins. Intriguingly, the same insert contains at least two PEST motifs (544 -581 and 606 -626) (7), which are often found in proteins that are rapidly degraded (29). Perhaps this insert serves as a proteolytic signal, but one which is regulated by pheromonedependent MAP kinase activity. How do these findings fit with our understanding of MAP kinase function? To our knowledge, this work represents the first direct identification of a MAP kinase phosphorylation site in yeast. It is also a rare example of a MAP kinase substrate outside of the nucleus (30 -32). In yeast, Fus3 is known to phosphorylate a transcriptional activator (Ste12), transcriptional repressors (Dig1, Dig2), and cyclin inhibitors (30,33,34). Even in mammalian cells only a handful of non-nuclear substrates for MAP kinases have been identified. However, this list includes members of the GRK and arrestin families, two proteins that promote desensitization of G protein coupled receptors (35). 4 Thus the regulation of receptors and G proteins may be coordinated in part by the ability of MAP kinases to phosphorylate a variety of desensitization factors acting throughout the signaling pathway.
In summary, RGS proteins are well known to promote G protein GTPase activity and desensitization. Our objective was to determine whether RGS proteins are themselves regulated, through post-translational modification. Our demonstration that Sst2 is phosphorylated in response to pheromone stimulation reveals one way that G protein signaling pathways might undergo feedback regulation in vivo. Given the striking similarities in yeast and human RGS proteins, it is likely that the regulatory mechanisms described here will also be recapitulated in more complex organisms.