Substitutions in the Pheromone-Responsive G_ß Protein of Saccharomyces cerevisiae Confer a Defect in Recovery from Pheromone Treatment

Corresponding author: David E. Stone, Laboratory for Molecular Biology (M/C 567), Department of Biological Sciences, University of Illinois at Chicago, 900 South Ashland Avenue, Chicago, IL 60607, U62248{at}uicvm.uic.edu (E-mail).

Communicating editor: M. D. ROSE

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The pheromone-responsive G protein of Saccharomyces cerevisiae, Gpa1p, stimulates an adaptive mechanism that downregulates the mating signal. In a genetic screen designed to identify signaling elements required for Gpa1p-mediated adaptation, a large collection of adaptive-defective (Adp^-) mutants were recovered. Of the 49 mutants characterized thus far, approximately three-quarters exhibit a dominant defect in the negative regulation of the pheromone response. Eight of the dominant Adp^- mutations showed tight linkage to the gene encoding the pheromone-responsive G_ß, STE4. Sequence analysis of the STE4 locus in the relevant mutant strains revealed seven novel STE4 alleles, each of which was shown to disrupt proper regulation of the pheromone response. Although the STE4 mutations had only minor effects on basal mating pathway activity, the mutant forms of G_ß dramatically affected the ability of the cell to turn off the mating response after exposure to pheromone. Moreover, the signaling activity of the aberrant G_ß subunits was suppressed by G322E, a mutant form of Gpa1p that blocks the pheromone response by sequestering G_ß, but not by E364K, a hyperadaptive form of Gpa1p. On the basis of these observations, we propose that Gpa1p-mediated adaptation involves the binding of an unknown negative regulator to G_ß.

THE differentiation of proliferating haploid yeast cells into mating competent cells is triggered by the exchange of peptide-mating pheromones. Each of the two yeast mating types, MATa and MAT, expresses a seven-transmembrane domain receptor that binds the pheromone secreted by cells of the opposite type. By analogy to mammalian systems, the pheromone-bound receptor is thought to activate its associated G protein by catalyzing the exchange of GTP for GDP on G, the result of which is a conformational shift in G and subsequent release of G_ß. The signal is then transmitted by G_ß to a MAP kinase cascade, ultimately resulting in the activation of a mating-specific transcription factor and the degradation of G₁ cyclins (for reviews see KURJAN 1992 ; SPRAGUE and THORNER 1992 ). Thus, haploid yeast cells arrest in the G₁ phase of the cell cycle, induce mating-specific genes, and elongate toward their mating partners in preparation for cytoplasmic and nuclear fusion.

The basal activity of the pheromone signaling pathway and the duration of the response after stimulus are regulated in numerous ways. The known negative regulatory mechanisms involve both types of pheromones (MACKAY et al. 1988 ; MARCUS et al. 1991 ) and their receptors (KONOPKA et al. 1988 ; RENEKE et al. 1988 ), the G protein (COLE and REED 1991 ; STRATTON et al. 1996 ), and one of the MAP kinases (DOI et al. 1994 ). These apparently redundant mechanisms may serve distinct purposes, such as promoting desensitization to a chronic stimulus, rapidly terminating the pheromone response upon diploid formation, and balancing sensitivity to pheromone against inappropriate induction of the mating signal. The need for complex and subtle modulation of this signaling pathway may be satisfied in part by the pheromone-responsive G protein Gpa1p. Gpa1p has been postulated to oppose the activity of its ß subunit (encoded by STE4 and STE18) in at least two ways. In what is presumed to be its inactive (GDP-bound) state, Gpa1p sequesters G_ß. In what is presumed to be its active (GTP-bound) form, Gpa1p stimulates an adaptive mechanism that is independent of G_ß sequestration (STRATTON et al. 1996 ). Whether Gpa1p inhibits the mating signal by direct interaction with G_ß or by indirect means presumably depends on the degree to which the receptor is stimulated, as well as on other unknown factors.

It has recently become clear that yeast is not exceptional in its use of G_ß as a signaling molecule. G_ß dimers are now known to interact with effector molecules and induce signals in many systems (CLAPHAM and NEER 1993 ; INIGUEZ-LLUHI et al. 1993 ). How the signaling activity of G_ß is controlled and how the signals generated by a given G and its ß are related are, therefore, questions of increasing interest. In yeast, the relationship between G and G_ß is clearly antagonistic. Pheromone–receptor binding results in a bifurcated signal: the heterotrimeric G protein separates into its two functional subunits and free G_ß induces changes in cellular physiology that are, after a delay, counteracted by activated G. An analogous antagonism between activated G and free G_ß has recently been found in three metazoan systems (CRESPO et al. 1995 ; LIU and SIMON 1996 ; SCHREIBMAYER et al. 1996 ).

In an effort to better understand how G protein signals are regulated and, specifically, to identify downstream components in the Gpa1p-mediated adaptive pathway, we undertook a screen for genomic mutations that result in supersensitive/adaptive defects. Here we describe the characterization of seven novel alleles of STE4 (G_ß), all of which confer a dominant Adp^- phenotype. We show that although some of these mutant forms of G_ß have little effect on basal signaling, they dramatically compromise the ability of the cell to recover from pheromone treatment. Our data also suggest that contrary to previous conclusions, Gpa1p-mediated adaptation does not stimulate Ste4p phosphorylation (COLE and REED 1991 ). In fact, our results suggest that the phosphorylation of Ste4p plays no role in adaptation.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast strains, media, and microbiological techniques:
Yeast growth media were prepared as described by SHERMAN et al. 1986 . Tryptophan, adenine, histidine, and uracil were omitted from synthetic media as necessary to maintain plasmids or select diploids. For experiments requiring the induction of the GAL1-regulated genes, cells were grown to midlog phase in selective sucrose medium, and galactose was added to a final concentration of 2%, or the cells were pelleted and resuspended in selective galactose medium. Yeast transformations were carried out according to the method of ITO et al. 1983 .

All strains constructed for this study were derived from strain 15Dau bar1 (MATa bar1 ade1 his2 leu2-3,-112 trp1 ura3), which is congenic with strain BF264-15D, as described previously (REED et al. 1985 ). Unless otherwise noted, the designation of cells as wild type, GPA1-E364K , or GPA1-G322E refers to strain 15Dau bar1 transformed with the centromeric vectors YCplac22 or YCplac33 (GIETZ and SUGINO 1988 ) containing the wild-type, E364K, or G322E alleles of GPA1, respectively (STONE and REED 1990 ). Thus, these strains express a plasmid-borne allele of GPA1 as well as the native copy of GPA1.

The parent strain for the mutant hunt, DSY278, was created by sequentially integrating the P_GAL1-GPA1-E364K (previously denoted GAL1^EG28-E364K in STRATTON et al. 1996 ) and FUS1-LEU2 constructs at the TRP1 and HIS2 loci, respectively, and then transforming the resulting strain with the GAL1-CLN3-2 centromeric vector YCp50/GAL1-CLN3-2 (CROSS 1988 ).

The ste4::LEU2 deletion strain ELY100, which was used in the linkage analysis and for the measurement of growth rates, was created by integrating the FUS1-LEU2 construct at the HIS2 locus and transplacing the wild-type STE4 allele with a ste4::LEU2 deletion/disruption construct.

To assess the effects of the adaptive-defective STE4 (STE4 Adp^-) alleles in a clean genetic background, the wild-type STE4 allele was replaced by the various mutant alleles in a two-step process. The ste4::URA3 sequence was excised as a 4.0-kb EcoRI-SphI fragment from the ste4::URA3/YCplac33 plasmid, gel purified, and used to transform strain 15Dau bar1. Disruption and deletion of the chromosomal STE4 locus was confirmed by PCR analysis and mating assays, and the resulting strain was named ELY104. To replace the chromosomal ste4::URA3 construct with the various STE4 Adp^- alleles, the mutant STE4 sequences were excised as 3.4 -kb EcoRI-SphI fragments from the various YCplac33/STE4 plasmids and were transformed into strain ELY104. Transformants were selected by growth on medium containing 5-fluoroorotic acid (FOA), which counterselects the URA3 marker. All gene replacements were confirmed by PCR analysis and halo tests.

Plasmids:
Recombinant DNA techniques were essentially as described by SAMBROOK et al. 1989 and AUSUBEL et al. 1994 . Plasmids created for this study were constructed as follows:

YCplac33/STE4^Hd: Plasmid YCplac33/STE4 (provided by Dr. G. COLE) contains three HindIII sites, two in the coding region of STE4 and one in the polylinker. This plasmid was partially digested with HindIII, and the ends were reclosed by treatment with Klenow and T4 ligase. The desired clone, a YCplac33/STE4 plasmid lacking only the target HindIII site, was isolated by screening bacterial transformants. Digestion of this plasmid (YCplac33/STE4^Hd) with HindIII deletes the sequence encoding residues 63–398 of Ste4p.

ste4::URA3/YCplac33: The sequence encoding residues 63–398 of Ste4p in YCplac33/STE4 was replaced by a HindIII fragment containing the URA3 gene.

YCplac33/GPA1-E364K , YCplac33/GPA1-G322E, and YCplac33/GPA1: The XbaI-SacI fragment containing the coding region of GPA1 and 1.75 kb of the 5' flanking sequence was moved from YCplac111 plasmids containing GPA1-E364K, GPA1-G322E, and wild-type GPA1 (STONE and REED 1990 ) to YCplac33 (GIETZ and SUGINO 1988 ).

FUS1-lacZ::ADE1: The ADE1 gene was amplified by PCR using PUC19-ADE1(-B) (provided by Dr. P. HAGLEY) as the template and Pfu (Stratagene, La Jolla, CA) as the thermophilic polymerase. The product was then ligated to the TA cloning vector pCRII (Invitrogen, San Diego, CA) to yield pCRII/ADE1. The fragment containing ADE1 was isolated from pCRII/ADE1 by digestion with ApaI, followed by gel purification and ligation to Apa1 cut and phosphatased pSB231 (TRUEHEART et al. 1987 ), thereby disrupting the URA3 gene.

Pheromone response and growth assays:
Strains were tested for pheromone-induced growth inhibition in standard halo assays, as described previously (COLE et al. 1990 ). After growth to midlog phase, ~10⁵ cells were diluted into 7 ml top agar (0.7%) and spread onto plates. Four-microgram doses of synthetic factor (Multiple Peptide Systems, San Diego, CA) were then dotted onto the surface of the plates in 4 -microliter aliquots. The plates were incubated for 2 days at 30° before photographing.

For determination of growth rates (Table 1), the STE4 wild-type and mutant transplacement strains were grown to midlog phase in rich medium (YEPD) at 30°, and the A₆₀₀ of each culture was measured over three doublings. Throughout the experiment, the cells were maintained in exponential growth by periodic dilution with warm medium.

Immunoblots:
Crude cell lysates were prepared as follows: 2 x 10⁸ midlog cells were harvested and boiled in 200 µg of 2x sample buffer [0.24 M Tris-HCl (pH 6.8), 2% SDS, 2% ß-mercaptoethanol, 20% glycerol] for 4 min. Acid-washed glass beads (0.45 mm) were added to the suspension, and the mixture was vortexed vigorously at 4° until lysis was complete and then boiled again for 2 min. The lysates were spun at 2000 rpm to remove the glass beads and then for 10 min at 12,000 g to pellet the cell debris. The resulting lysates were normalized by measuring absorbance at 280 nm. Samples containing ~0.4 A₂₈₀ unit were electrophoresed on a discontinuous SDS-polyacrylamide gel (4%/10%) and electroblotted to PVDF-Plus transfer membrane (Micron Separations, Inc., Westboro, MA) according to the protocol supplied with the semidry blotter (Fisher Scientific, Pittsburgh, PA). Blots were then blocked with 5% nonfat dry milk in TBS overnight. The blots were subsequently incubated with diluted, strip-purified antisera raised against full-length Ste4p (45) in 0.2x blocking solution in TBS for 4 hr, and then processed with the ECL immunoblot detection kit (Amersham, Arlington Heights, IL) according to the manufacturer's instructions. Strip purification of the Ste4p antisera was performed according to the method described by HARLOW and LANE 1988 using GST-Ste4p fusion protein purified from Escherichia coli.

Genetic screen and characterization:
Mutant selection: 2 x 10⁷ cells of strain DSY278 were spread on YEPD plates and mutagenized with UV radiation to 50% lethality by inverting the plates for ~3 sec on a short-wave UV transilluminator. After incubation in the dark for 3 days, the cells were replica-plated to synthetic sucrose medium lacking uracil, incubated for an additional 2 days, and then replica plated to the selection media (synthetic galactose medium containing 20 ng, 0.5 µg, or 2 µg factor and lacking leucine). After 3 days of growth, colonies of all sizes were streaked onto fresh selection plates containing the same amount of factor (20 ng, 0.5 µg, or 2 µg), and were allowed to grow for an additional 2 days. The mutants were then grown on rich medium lacking pheromone for 2 days before being tested for growth on synthetic glucose medium lacking leucine and on YEPD plates containing 10, 25, 50, 75, 100, and 125 ng of factor. Finally, the putative mutants were scored for leucine prototrophy and projection formation on medium lacking factor. All incubations were at 30°.

Backcrosses: Putative Adp^- mutants were crossed to 15Dau bar1 carrying YCplac33 (GIETZ and SUGINO 1988 ), and the mating mixtures were spread onto synthetic galactose medium lacking histidine, uracil, and tryptophan (SG -HUT) to select diploids. After sporulation on potassium acetate plates, asci were dissected or subjected to random spore analysis.

Dominance test: All mutants to be tested for dominance were crossed to a MATa deletion strain, (KT23x8:EG123: ura3 leu2 trp1 his4 BAR1; TATCHELL et al. 1981 ), transformed with pJM9, a YCp50-based plasmid carrying the MAT locus (provided by HAY-OAK PARK), and mated on rich galactose-containing medium (YEPG) for 8 hr at 30°. Diploids were then selected from the mating mixture by sequentially spreading onto SG -HUT plates twice. The resulting diploid cells were grown overnight on rich medium at 30° before being replica plated to medium containing galactose and 0.5 µg/ml FOA. Ura^- cells were picked from FOA-resistant patches, uracil auxotrophy was confirmed, and halo tests were performed on YEPD and YEPG.

Linkage analysis: The Adp^- mutant strains were tested for linkage to STE4 as follows: The mutant strains were crossed to strain ELY100, which is a MAT ste4::LEU2 FUS1-LEU2 strain isogenic with the parent strain, DSY278. Diploid cells were sporulated, asci were dissected, and segregants were scored for leucine prototrophy. At least 12 asci were dissected for each mutant strain. Mutations linked to STE4 were expected to segregate 4 Leu⁺:0 Leu^-.

Gap repair: Mutant STE4 alleles were retrieved from their genomic loci by gap repair (ORR-WEAVER et al. 1983 ; ROTHSTEIN 1991 ). YCplac33/STE4^Hd was digested with HindIII and XhoI, which removes the sequence-encoding residues 63–398 of Ste4p, gel purified, and used to transform the various STE4 mutant strains. Ura⁺ transformants were selected. Plasmids recovered from these yeast transformants were amplified in E. coli and were used to transform the parent strain DSY278. Plasmids that caused DSY278 cells to exhibit a defect in recovery from pheromone treatment were subjected to sequence analysis.

Sequence analysis: Dideoxy sequence reactions were performed using the Sequenase Version 2.0 kit (United States Biochemical, Cleveland, OH).

Measurement of FUS1-lacZ activity:
FUS1 promoter activity was assayed in cells transformed with a centromeric reporter plasmid, FUS1-lacZ::ADE1, derived from pSB231 (TRUEHEART et al. 1987 ). ß-galactosidase activity was measured as described previously (SLATER and CRAIG 1987 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation of Adp^- mutants:
As a first step toward understanding how the activated form of Gpa1p promotes adaptation, we sought genomic mutations that would disrupt the ability of E364K, a hyperadaptive allele of GPA1, to confer resistance to pheromone. To create a strain in which viability would depend on such a mutation, three constructs were introduced into the MATa leu2-3,-112 strain 15Dau: FUS1-LEU2, P_GAL1 -GPA1-E364K, and GAL1-CLN3-2. Our rationale for the construction of this strain, DSY278, was as follows: The hybrid gene composed of the pheromone-inducible FUS1 promoter fused to the coding region of LEU2 allows the parent strain to grow in leucine-deficient media if the pheromone signaling pathway is activated. Activation of the pathway does not inhibit growth because the CLN3-2 allele, a dominant mutation that stabilizes one of the G₁ cyclins in yeast, uncouples pheromone-induced transcription from pheromone-induced G₁ arrest (CROSS 1988 ). Cells carrying this allele are able to proliferate even when the pheromone pathway is stimulated. The parent strain, however, also contains P_GAL1 -GPA1-E364K, the hyperadaptive GPA1 allele E364K, under the control of an attenuated GAL1 promoter, GAL1-EG28 (STRATTON et al. 1996 ). Because Gpa1-E364Kp blocks pheromone-induced activity of the FUS1 promoter (STONE and REED 1990 ), cells are unable to grow on galactose-based medium containing pheromone and lacking leucine. The basis for the selection of Adp^- mutants, then, is that mutations that interfere with the ability of Gpa1p to stimulate adaptation should allow FUS1-LEU2 induction.

Approximately 2 x 10⁷ cells were mutagenized on plates, and with the P_GAL1-GPA1-E364K construct induced, they were replica plated to medium containing pheromone and lacking leucine. To sort out the desired mutations from an unwanted background (e.g., mutations that knock out the expression or function of P_GAL1-GPA1-E364K ), colonies recovered in this selection were screened for a heightened response to pheromone with the expression of GPA1-E364K repressed. Mutations that blocked Gpa1-E364Kp–mediated adaptation and that conferred supersensitivity by themselves were considered likely to be in genes encoding its effector or downstream elements in its signaling pathway.

Colonies picked from the primary selection plates were cured of the GAL1-CLN3-2 vector and then tested for their ability to grow at various concentrations of factor in single-colony formation and halo assays under both P_GAL1-GPA1-E364K–expressing and –repressing conditions. Of the roughly 1000 mutants isolated in this selection that were able to induce FUS1-LEU2 despite expression of GPA1-E364K , about one-quarter exhibited a phenotype that was dramatic enough to warrant further analysis. These 253 mutants fell into two classes. The 183 class I mutants showed an extreme sensitiv-ity to pheromone when GPA1-E364K expression was turned off, as manifested by very large halos, but were uncompromised in their ability to adapt to pheromone treatment when the mutant GPA1 allele was turned on, as evidenced by very turbid halos (see Figure 1). Many of these mutants showed an increased expression of FUS1-LEU2 and displayed mating-specific morphological changes in the absence of pheromone, indicating an elevation in the basal pheromone pathway activity. Presumably, class I mutations augment the mating signal to such a degree that the Gpa1p-mediated adaptive signal is partially overridden, and the FUS1-LEU2 construct is induced sufficiently to generate Leu⁺ colonies. Characterization of these mutants will be described elsewhere. In contrast to the highly supersensitive/Adp⁺ phenotype of class I mutants, the 70 class II mutants were slightly supersensitive and Adp^- (Figure 1): They were clearly compromised in their ability to recover from pheromone treatment when GPA1-E364K was expressed (as evidenced by clear or lightly filled halos), and they showed a slightly greater than normal response when the hyperadaptive GPA1 allele was repressed (as evidenced by large halos). Like the class I mutants, the class II mutants fell into two subgroups: those that were leucine prototrophs, even in the absence of pheromone (i.e., they exhibited a high basal pathway activity), and those that required pheromone to induce the FUS1-LEU2 construct.

View larger version (75K): In this window In a new window Download PPT slide   Figure 1. —Halo tests showing representative class I and II mutants. The parent strain and all the mutants contain the PGAL1-GPA1-E364K fusion that is repressed by glucose and induced by galactose. Cells were grown in rich glucose medium (top row) or rich galactose medium (bottom row), and halo tests were performed on the corresponding medium using 4 µg of factor. Class I mutants are supersensitive/adaptive+. Class II mutants are supersensitive/Adp-.

To test the Adp^- mutations for dominance, the class II mutants were crossed to a specialized strain designed to allow recovery of MATa/mat diploids (see MATERIALS AND METHODS). The dominance relationships of mutations affecting the yeast pheromone response must be tested in diploids expressing only MATa information because MATa/MAT strains do not express the mating response pathway and are therefore sterile. For those mutations that conferred a constitutive Leu⁺ phenotype, dominance was assessed by simply scoring the diploids for leucine prototrophy. Additionally, all MATa/mat diploids carrying Adp^- mutations were subjected to halo tests on P_GAL1-GPA1-E364K–inducing medium. Failure of a given diploid to form colonies within the halo zone was taken as an indication that the mutation in question was dominant. Because we expected the Adp^- phenotypes to result from loss-of-function mutations, it was surprising to find that of the 49 class II mutants analyzed in this study, 37 were dominant and only 12 were recessive.

Dominant missense mutations in STE4 block Gpa1p-mediated adaptation:
The pheromone-responsive G_ß of yeast, Ste4p, is rapidly phosphorylated when cells are treated with factor, and this modification is thought to be an adaptive mechanism (COLE and REED 1991 ). Deletion of the putative phosphorylation domain of Ste4p, residues 310–346, confers supersensitivity (COLE and REED 1991 ) and blocks Gpa1p-mediated adaptation (D. E. STONE and H. F. Stratton, unpublished results). In other words, the STE4^310-346 allele is epistatic to GPA1-E364K. Deletion of this domain also prevents certain signaling-defective forms of Ste4p from promoting adaptation (GRISHIN et al. 1994 ). The recovery of such a high proportion of dominant Adp^- mutants in our screen led us to speculate that novel alleles of STE4 were represented in the collection. In particular, we expected to find lesions in the 310–346 region of Ste4p. Therefore, we asked whether dominant Adp^- mutations could be separated from STE4 in linkage analysis. Ten strains carrying such mutations, each of which conferred a constitutively active FUS1-LEU2 phenotype that resulted in leucine prototrophy and which segregated as single loci, were crossed to a congenic MAT tester strain. Like the mutant strains, the tester strain contained the FUS1-LEU2 reporter construct. In addition, the STE4 locus of this strain was marked with the LEU2 gene. Diploids resulting from crosses between the MATa FUS1-LEU2 leu2^- Adp^- mutants and the MAT FUS1-LEU2 leu2^-ste4::LEU2 tester were sporulated and dissected. A meiotic segregant is expected to be prototrophic for leucine if it inherits either the ste4::LEU2 construct or the Adp^- mutation (which activates FUS1-LEU2). If the Adp^- mutation is in STE4, then all segregants are expected to be Leu⁺ because the ste4::LEU2 construct will always segregate away from the STE4 Adp^- allele. Of the 10 mutants analyzed in this manner, eight yielded 100% Leu⁺ segregants in 12 tetrads. These results strongly suggest that mutations in STE4 are responsible for the dominant Adp^- phenotype in the majority of these mutant strains.

To recover the STE4 alleles from the strains showing tight linkage between the adaptive mutation and the STE4 locus, the gap repair method described by ORR-WEAVER et al. 1983 and ROTHSTEIN 1991 was used. The mutant strains were transformed with a gapped STE4 plasmid, and plasmids repaired in vivo (presumably by recombination with the mutant STE4 allele) were extracted. Rescued plasmids that could confer the original dominant Adp^- phenotype upon transformation of the unmutagenized parent strain were sequenced across the STE4 coding region. The results of this analysis are shown in Figure 2 and Table 1. In each case, the rescued STE4 allele differed from the published STE4 sequence (WHITEWAY et al. 1989 ) in two positions. Nucleotide 709 was invariably found to be C instead of T, which is predicted to result in the substitution of serine for leucine at residue 236. This apparent polymorphism was confirmed by sequencing the wild-type allele of STE4 from strain 15Dau. Each rescued allele was also found to differ from the wild-type sequence at one additional site. The seven novel STE4 mutations cluster in two regions: a C-terminal group includes substitutions in residues 405, 409, 410, and 411; an N-terminal group includes substitutions at residues 115 and 138. The STE4-S410L allele was recovered independently three times, suggesting that many of the STE4 mutations generated in this screen have been identified.

View larger version (7K): In this window In a new window Download PPT slide   Figure 2. —Positions of the mutations and the resulting substitutions in the STE4 Adp- alleles. The open boxes represent the WD40 repeat units found in all Gß subunits. The shaded boxes represent the domains unique to Ste4p.

G_ß subunits containing the dominant Adp^- forms of Ste4p can be sequestered by Gpa1p:
All seven of the Adp^- strains listed in Table 1 are leucine prototrophs, presumably because the STE4 Adp^- alleles that they carry raise the activity of the basal mating signal enough to induce the expression of FUS1-LEU2. None of these strains, however, exhibit the changes in morphology and growth rate that are expected to result from a high basal mating signal. Because even a slight increase in the basal signal could create a selective pressure for sterile mutations or result in physiological adaptation, such mutant phenotypes could be lost over time. A derivative of wild-type strain 15Dau in which the native STE4 allele had been deleted was therefore transformed with the gap-repaired yeast centromeric vectors containing the STE4 Adp^- alleles. With the exception of the STE4-F115S cells, each of the newly created mutant strains grew normally (Table 1), forming visible colonies at a rate indistinguishable from 15Dau ste4 cells transformed with a plasmid-borne copy of wild-type STE4 (Figure 3). Moreover, examination of the mutant cells under the light microscope revealed no sign of projection formation (shmooing), the morphological change particular to cells responding to pheromone (data not shown). Assuming that Ste4p reaches its normal steady-state level and is localized within the time it takes the transformed cells to go through one or two generations (2–4 hr), the growth rate of the transformed colonies provides a sensitive means to assess the ability of Gpa1p to bind the mutant forms of G_ß. A defect in –ß binding caused by a mutation in STE4 should lead to activation of the mating signal and, hence, projection formation and inhibition of growth, as has been observed in cells expressing STE4^Hpl alleles (BLINDER et al. 1989 ; WHITEWAY et al. 1994 ). Although the original mutant strains may have undergone genetic changes that normalized the basal signal during their isolation, introduction of the STE4 Adp^- alleles into a clean genetic background and immediate assessment of the transformed cells eliminates this variable.

View larger version (71K): In this window In a new window Download PPT slide   Figure 3. —Plating efficiency and colony size of the STE4-W410L Adp- mutant in comparison to wild-type cells. Centromeric plasmids containing either the wild-type or a mutant allele of STE4 were transformed into strain 15Dau bar1 ste4::LEU2. The resulting transformants were spread at an equal density on plates and incubated at 30° for 3 days. The colony sizes and densities were indistinguishable for all the mutants and the wild-type allele of STE4, except for STE4-F115S cells, which formed noticeably smaller colonies.

Cells expressing STE4 Adp^- alleles grow normally in the absence of pheromone, but their growth is arrested by doses of factor that wild-type cells are able to overcome (Table 1 and Figure 4). Is this supersensitive phenotype caused by a defect in –ß affinity? As a second means of assessing the ability of Gpa1p to bind the mutant forms of G_ß, halo tests were performed on STE4 Adp^- cells expressing either E364K or G322E, two mutant alleles of GPA1. Although GPA1-E364K and GPA1-G322E both confer resistance to pheromone, they act by different mechanisms (STRATTON et al. 1996 ). Whereas Gpa1-E364Kp stimulates adaptation (recovery from pheromone treatment), Gpa1-G322Ep confers insensitivity (an inability to respond to pheromone). Based on the biochemical characterization of an analogous mutant form of G_s (LEE et al. 1992 ), as well as on genetic (STRATTON et al. 1996 ) and biochemical (M. CISMOWSKI and D. STONE, unpublished results) evidence, Gpa1-G322Ep is thought to confer resistance to pheromone by sequestering G_ß. Thus, if Gpa1p can bind G_ß in spite of the mutations in STE4, then GPA1-G322E should be epistatic to the STE4 Adp^- alleles. As shown in Figure 4, cells expressing Gpa1-G322Ep and wild-type Ste4p were almost completely resistant to factor. With the exception of the cells transformed with the STE4-F115S allele, which formed a very turbid halo, the strains expressing GPA1-G322E and the STE4 Adp^- alleles were similarly unresponsive. In contrast, the hyperadaptive phenotype conferred by GPA1-E364K in cells expressing wild-type STE4, as evidenced by turbid halos, was much less apparent in cells expressing the STE4 Adp^- alleles.

View larger version (62K): In this window In a new window Download PPT slide   Figure 4. —Epistasis analysis. Halo tests showing the ability of GPA1-E364K and GPA1-G322E to confer pheromone resistance in the various STE4 Adp- mutant backgrounds. Cells were grown in synthetic medium lacking uracil to select for the GPA1 vectors, and halo tests were performed on the corresponding medium using 4 µg of factor. The area around the point source of pheromone was photographed after 2 days of incubation at 30°. All magnifications are equivalent.

Effects of the dominant Adp^- STE4 mutations on mating-specific transcription:
In addition to its effects on growth and cellular morphology, pheromone stimulates the transcription of genes that are required for modulation of the mating signal and for cell and nuclear fusion. In fact, measuring the expression of mating-specific genes is a more sensitive means to assess mating pathway activity than are the cellular responses to pheromone (CROSS et al. 1988 ). Transcriptional assays are also more informative over a much shorter time period than are colony forming assays, such as halo tests. To determine how the STE4 Adp^- mutations affect the transcriptional activity of mating-specific genes, we transformed six of the seven original mutant strains with FUS1-lacZ::ADE1, a centromeric vector containing a transcriptional fusion between the FUS1 promoter region and the lacZ coding sequence. Like the native FUS1 gene, the expression level of the FUS1-lacZ reporter construct is proportional to the activity of the pheromone signaling pathway: its expression is dependent on an intact signaling pathway and is strongly induced by pheromone. The mutant strains and a wild-type control strain, each carrying the reporter vector and a centromeric plasmid containing GPA1, were grown to midlog phase and sampled for basal pathway activity. They were then treated with a submaximal dose of factor for 4 hr, and samples were harvested for determination of ß-galac-tosidase activity. As shown in Figure 5A, the basal activity of the mating signal was variously affected by the STE4 Adp^- mutations. The STE4-F115S allele caused the greatest increase in basal FUS1-lacZ activity (~23% of full induction), the L138F substitution and the three-most C-terminal mutations in STE4 caused a moderate increase in the constitutive signal (6–11% of full induction), and the STE4-A405V cells showed the smallest increase in the steady-state lacZ expression (~3% of full induction). After treatment with a low concentration of -factor, the mutant strains all displayed supersensitivity, producing two to 10 times more ß-galactosidase than the wild-type control cells (Figure 5B). Induction of FUS1-lacZ was similar in all the strains after a saturating dose of factor (data not shown). In comparing the mutant strains, no correlation between the degree of basal signaling and the degree of supersensitivity was apparent. Identical experiments performed with strains in which the STE4 Adp^- alleles had been introduced into a clean genetic background by transplacement of wild-type STE4 produced similar results, although the basal activities of the mutant strains were not as highly induced (data not shown).

View larger version (57K): In this window In a new window Download PPT slide   Figure 5. —Effects of the STE4 Adp- alleles when expressed with various forms of Gpa1p on the activity of the FUS1 promoter. The mutant strains were transformed with a FUS1-lacZ centromeric plasmid and either YCplac33/GPA1, YCplac33/GPA1-E364K, or YCplac33/GPA1-G322E. Transformants were grown to midlogarithmic phase and were either treated with 20 ng/ml of factor for 4 hr or grown an additional 4 hr in the absence of pheromone. ß-Galactosidase levels were determined as described in ma-terials and methods. The values shown are the mean of duplicate determinations for each strain. The data are typical of numerous experiments. (A) The value indicated for each mutant is the percent of full FUS1-lacZ induction, as determined by treating wild-type cells with 100 ng/ml factor for 4 hr. (B–D) The induced levels of ß-galactosidase in the mutant cells are expressed relative to the induced ß-galactosidase activity in the wild-type control cells. (A) Uninduced cells expressing wild-type GPA1. (B) Induced cells expressing wild-type GPA1. (C) Induced cells expressing GPA1-E364K. (D) Induced cells expressing GAL1-G322E.

Because the Adp^- mutants were originally selected as cells that could induce FUS1-LEU2 in spite of GPA1-E364K expression, we assessed the ability of Gpa1-E364Kp to repress mating-specific transcription in the STE4 Adp^- backgrounds. The STE4 Adp^- mutant strains were transformed with FUS1-lacZ::ADE1, and a centromeric plasmid containing the GPA1-E364K coding region under the control of the GPA1 promoter (YCplac33/GPA1-E364K). The resulting transformants were then grown to midlog phase, and ß-galactosidase activity was measured before and 4 hr after treatment with a submaximal dose of factor. As shown in Figure 5C, the adaptive function of Gpa1-E364Kp was severely compromised in all six of the STE4 Adp^- mutants tested: the induced ß-galactosidase rose to levels 11.4 - to 15.9-fold higher than in the wild-type control strain. As we observed in cells expressing the wild-type allele of GPA1, the induced FUS1-lacZ activities in the mutant strains expressing GPA1-E364K showed no correlation with the increases in basal FUS1-lacZ activity.

The superinduction of the FUS1 promoter could be caused either by the inability of Gpa1p to sequester the mutant forms of G_ß after pheromone treatment or by the failure of a protein other than Gpa1p to downregulate the aberrant G_ß subunits. To distinguish these possibilities, the ability of the G322E mutant form of Gpa1p to inhibit FUS1-lacZ activity in the STE4 Adp^- strains was assessed. Gpa1-G322Ep negatively regulates the mating signal by sequestering G_ß. Mutant cells transformed with both the FUS1-lacZ reporter vector and a centromeric plasmid containing the GPA1-G322E coding region under the control of the GPA1 promoter (YCplac33/GPA1-G322E) were treated with factor, and the induced ß-galactosidase levels were measured, as described above. The results of this experiment, shown in Figure 5D, are striking. The induced ß-galactosidase levels in the various mutants expressing Gpa1-G322Ep varied over a 10-fold range, just as the basal FUS1-lacZ activities did (Figure 5A). Moreover, a comparison of the mutants to one another revealed the same relationships that were observed when the basal activities were measured (Figure 5A). For example, the Adp^- form of Ste4p that caused the highest basal FUS1 activity (Ste4 -F115Sp) was least affected by Gpa1-G322Ep expression, whereas the mutant G_ß that caused the smallest increase in basal FUS1 activity (Ste4-A405Vp) was most effectively inhibited by Gpa1-G322Ep.

Pheromone-induced phosphorylation of the dominant Adp^- forms of Ste4p is normal:
Because pheromone-induced phosphorylation of Ste4p is thought to be an adaptive mechanism that is dependent upon Gpa1p (COLE and REED 1991 ), and because the STE4 Adp^- mutants block Gpa1p-mediated adaptation, it was of interest to determine whether the adaptive defects are caused by an inability to phosphorylate Ste4p. The effect of the Adp^- mutations on Ste4p phosphorylation was assessed by monitoring the migration of Ste4p on denaturing polyacrylamide gels, taking advantage of the observation that pheromone-induced phosphorylation of Ste4p leads to a characteristic reduction in its electrophoretic mobility (COLE and REED 1991 ). The results of an experiment in which wild-type and mutant cells were treated with a submaximal dose of pheromone for a period of 2 hr are shown in Figure 6. Proteins extracted from each culture were electrophoresed, blotted to nitrocellulose, and probed with an affinity-purified anti-Ste4p polyclonal antibody. Consistent with previously reported results, the phosphorylation state of Ste4p changed dramatically when wild-type cells were exposed to factor. The bulk of the protein was chased from the fastest migrating (relatively unphosphorylated) species of Ste4p in the untreated control to the slowest migrating (fully phosphorylated) species of Ste4p at 20 ng/ml factor, half the dose that is required to block cell cycle progression under these culture conditions. Identical results were obtained when cells were challenged with 100 ng/ml of factor (data not shown). Interestingly, all six of the STE4 Adp^- strains tested in this assay were fully responsive, exhibiting no defect in Ste4p phosphorylation. Therefore, the Adp^- phenotypes cannot be attributed to a failure to modify the mutant forms of G_ß. Before pheromone stimulation, however, the six STE4 Adp^- strains did show a detectable increase in the basal level of Ste4p phosphorylation. By comparing the intensity of the uppermost band to that of the lowermost band in lanes 1–7 of Figure 6, it is apparent that there is some variation in the mutant phenotypes: Ste4-F115Sp exhibited the greatest basal phosphorylation, whereas Ste4 -A405Vp appeared most similar to wild type. The remaining mutants displayed an intermediate increase in the ratio of phosphorylated to unphosphorylated Ste4p. These data are consistent with the effects that the STE4 Adp^- alleles had on FUS1-lacZ activity (Figure 5).

View larger version (29K): In this window In a new window Download PPT slide   Figure 6. —Phosphorylation state of Ste4p in STE4 Adp- mutant and wild-type cells. Midlogarithmic phase mutant and wild-type cells were grown in the absence of pheromone or treated with various doses of -factor for 2 hr, as indicated. Crude lysates were prepared as described in materials and methods, and electrophoresed on denaturing polyacrylamide gels. Proteins were then blotted to polyvinyldifluoride membranes and probed with an affinity-purified antibody raised against Ste4p. The rate of migration of the Ste4p protein is a function of its phosphorylation state: the unphosphorylated form (dash) migrates fastest; the fully phosphorylated form (P*) migrates slowest.

Gpa1p does not stimulate phosphorylation of G_ß:
Based on the observation that Ste4p phosphorylation is dramatically impaired in a gpa1 strain (COLE and REED 1991 ), it has been suggested that Gpa1p stimulates this modification of Ste4p, thereby promoting an adaptive mechanism. The identification of mutations in STE4 that block Gpa1p-mediated adaptation, while having no effect on pheromone-induced phosphorylation of Ste4p, calls this hypothesis into question. To test this idea in a different way, wild-type cells were transformed with single-copy plasmids containing either GPA1-E364K or wild-type GPA1, and were treated with various concentrations of pheromone for a period of 2 hr. If Gpa1p downregulates the mating response by inducing Ste4p phosphorylation, a hyperadaptive allele of GPA1 such as E364K would be expected to augment pheromone-induced phosphorylation of Ste4p. As shown in Figure 7, Gpa1-E364Kp did not stimulate Ste4p phosphorylation. In fact, close examination of Figure 7A reveals a slight inhibitory effect on this phenomenon (compare the ratio of the highest and lowest bands in the wild-type and GPA1-E364K cells, especially at doses >50 ng/ml). Although the magnitude of this effect is small, the result is significant. A similar effect was found in cells expressing wild-type levels of Gpa1-E364Kp when Ste4p phosphorylation was examined as a function of time (data not shown). Moreover, the phosphorylation of Ste4p was almost completely blocked when the GAL1 promoter was used to drive overexpression of GPA1-E364K (Figure 7B).

View larger version (53K): In this window In a new window Download PPT slide   Figure 7. —Pheromone-induced phosphorylation of Ste4p in wild-type and cells expressing GPA1-E364K. Midlogarithmic phase cells were treated with the indicated doses of factor for 2 hr. Immunoblots were prepared as described in materials and methods and probed with an affinity-purified antibody raised against Ste4p. The rate of migration of the Ste4p protein is a function of its phosphorylation state: the unphosphorylated form (arrow) migrates fastest; the fully phosphorylated form (star) migrates slowest. Molecular weights are in kilodaltons. (A) Wild-type cells were transformed with either an GPA1-E364K centromeric plasmid (YCplac111/GPA1-E364K ) or an identical plasmid containing wild-type GPA1 (YCplac111/GPA1). These strains express approximately the wild-type level of Gpa1p. (B) Wild-type cells transformed either with the GAL1 centromeric expression vector YCpG2 or with YCpG2 carrying GAL1-E364K or GAL1-GPA1. The latter two strains overexpress Gpa1p by a factor of at least 10, as determined by scanning immunoblots (data not shown) probed with strip-purified antisera raised against Gpa1p (STONE et al. 1991 ).

Since the experimental cells used in these assays were constitutively expressing GPA1-E364K, we considered the possibility that the Gpa1p-mediated adaptive pathway might itself be downregulated as a consequence of chronic activation, thus minimizing Gpa1-E364Kp–in-duced phosphorylation of Ste4p. To prevent such long-term physiological changes, the P_GAL1-GPA1 cells and P_GAL1-GPA1-E364K cells were cultured under noninducing conditions, and the GAL1 promoter was induced concomitantly with the addition of pheromone. Using this protocol, dose–response curves and time courses were generated, but in no case was GPA1-E364K expression correlated with increased phosphorylation of Ste4p (data not shown).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The realization that the ß as well as the subunits of heterotrimeric G proteins are active signaling elements in numerous systems (CLAPHAM and NEER 1993 ; INIGUEZ-LLUHI et al. 1993 ) raises the question as to how the independent branches of G protein–mediated signaling pathways are functionally related, and how they are regulated. A priori, one can imagine systems in which G and G_ß regulate different effectors and systems in which they act on the same effector. In principle, the effects of G and G_ß stimulation could be additive, synergistic, or antagonistic. In fact, examples of each type are known (CLAPHAM and NEER 1993 ).

The yeast mating response, which is driven by the ß subunit of a G protein, provides an example of antagonistic regulation. We have shown that the pheromone-inducible G protein Gpa1p stimulates an adaptive signal that downregulates the G_ß-induced mating signal independently of G_ß sequestration (STRATTON et al. 1996 ). To identify elements downstream of Gpa1p in this pathway, we conducted a genetic screen for adaptive defects in a strain expressing a hyperadaptive allele of GPA1. Both recessive and dominant Adp^- mutations were recovered. Eight out of the 10 dominant mutations tested showed tight linkage to STE4. Sequence analysis of the STE4 locus in these mutant strains revealed seven novel STE4 alleles—F115S, L138F, A405V, G409D, S410L, W411L, and W411S—each of which were shown to disrupt proper regulation of the pheromone response. The genetic characterization of these mutant forms of G_ß led to two conclusions: (1) G_ß is a target of G-mediated adaptation and (2) the adaptive mechanism or mechanisms impaired by the lesions in STE4 do not involve G_ß phosphorylation.

G_ß is a target of Gpa1p-mediated adaptation:
The mutant alleles of STE4 isolated in this work confer a defect in regulation of the mating signal, especially in recovery from exposure to pheromone. It has been suggested that Ste4p negatively regulates the pheromone response in addition to stimulating it (GRISHIN et al. 1994 ). If Ste4p works in concert with Gpa1p to stimu-late adaptation, mutations resulting in the loss of this function should be recessive. The discovery of mutations in STE4 that disrupt Gpa1p-mediated adaptation and that are dominant suggests that Ste4p is a target of, rather than a cofactor in, the G-induced desensitization mechanism. The lesions in question presumably render G_ß refractory to this adaptive mechanism.

Could the Adp^- mutations in STE4 simply disrupt –ß binding? Two pieces of evidence argue against this possibility. First, the mutant forms of Ste4p have relatively large effects on the induced signal in comparison to their effects on the basal signal. With the exception of the F115S mutation, the alterations in STE4 had virtually no significant impact on the growth and morphology of cells cultured in the absence of pheromone, and only small effects on the steady-state levels of mating-specific transcription and Ste4p phosphorylation (Table 1 and Figure 3, Figure 5, and Figure 6). One STE4 Adp^- strain, STE4-A405V, was almost indistinguishable from wild type when grown in the absence of pheromone. After stimulation with pheromone, however, the effects of the STE4 Adp^- alleles were dramatic: cells coexpressing a STE4 Adp^- allele with wild-type Gpa1p formed larger than normal halos (Table 1 and Figure 4, top row); those expressing a STE4 Adp^- allele with the hyperadaptive allele of GPA1, E364K , showed significantly less colony formation within the halos than cells expressing GPA1-E364K and wild-type STE4 (Figure 4, middle row). Thus, it appears that inactivation of the aberrant G_ß subunits is normal or nearly normal in the absence of pheromone, but is defective after pheromone treatment.

The results of the experiments designed to measure the effects of the STE4 Adp^- alleles on mating-specific transcription, which is more sensitive to pheromone stimulation than are the cellular responses, also suggest that the mutations primarily affect downregulation of the induced mating signal rather than basal pathway activity. In examining the transcriptional data, it is helpful to rank the mutants according to their relative ß-galactosidase levels. The effects of the various STE4 mutations on basal FUS1-lacZ transcription are widely variable: the STE4-F115S allele induces the highest constitutive activity, STE4-A405V has the least effect on transcription, and the other alleles are intermediate in this regard (Figure 5A). After treatment with a low dose of pheromone, the mutant strains expressing wild-type Gpa1p exhibited between 2- and 10-fold supersensitivity (Figure 5B); those expressing GPA1-E364K all hyperinduced FUS1-lacZ to a similar level, 11–16-fold higher than the control (Figure 5C). Moreover, the rank order of the induced and basal FUS1-lacZ activities did not correlate in cells expressing either wild-type Gpa1p or Gpa1-E364Kp. In the wild-type GPA1 background, for example, the allele that showed the greatest supersensitivity, STE4-S410L, exhibited one of the lowest basal activities. Similarly, the mutants that exhibited the highest basal activities were most effectively downregulated by Gpa1-E364Kp, whereas FUS1-lacZ was not downregulated as well by Gpa1-E364Kp in those mutants with the lowest basal activities. The lack of correspondence between the effect of a given STE4 allele on basal and induced mating-specific transcription is significant. If the regulatory defect were simply caused by poor –ß binding, then the mutations that most dramatically affect G_ß sequestration in the absence of pheromone might be expected to have the most adverse effect on recovery.

The second argument against the idea that the STE4 Adp^- mutations merely cause dissociation of the G protein subunits relies on experiments with G322E, a mutant form of Gpa1p that confers insensitivity to pheromone by sequestering G_ß (STRATTON et al. 1996 ; M. CISMOWSKI and D. STONE, unpublished results). As discussed above, the hyperadaptive form of Gpa1p, E364K, cannot fully stimulate recovery from pheromone treatment in cells expressing any of the STE4 Adp^- alleles (Figure 4, middle row, and 5). The STE4 Adp^- mutations are epistatic to GPA1-E364K. In contrast, the ability of Gpa1-G322Ep to sequester G_ß and block the mating response, as assayed in halo tests, is unaffected in all but one of the mutant strains (Figure 4, bottom row). The STE4-F115S mutant cells were the exception, forming detectable halos in spite of Gpa1-G322Ep expression. The difference in the effects of the STE4 Adp^- alleles on cells expressing these two functionally dissimilar forms of Gpa1p is consistent with the idea that the Adp^- mutations affect the mating signal primarily through a defect in Gpa1p-mediated adaptation rather than by disrupting the sequestration of G_ß by G. The transcriptional and Ste4p phosphorylation assays, however, demonstrate that the mutant forms of G_ß do have some impact on basal signaling (Figure 5 and Figure 6). The levels of FUS1-lacZ activity and Ste4p phosphorylation are elevated to a variable degree, depending on which allele is tested. Presumably, the small increases in constitutive FUS1-lacZ activity result from the release of G_ß from G. Consistent with this inference, Gpa1-G322Ep suppressed the various forms of G_ß in an allele-specific manner. When the effect of coexpressing GPA1-G322E and the STE4 Adp^- alleles in cells treated with pheromone was examined at the transcriptional level, a striking similarity to the basal signaling activities of the STE4 Adp^- mutants expressing wild-type Gpa1p was seen (Figure 5A and Figure C). The STE4 Adp^- mutations that manifested the highest constitutive FUS1-lacZ activity, presumably those that are most disruptive to –ß binding, were also least well suppressed by Gpa1-G322Ep in cells exposed to pheromone. In fact, the bar graphs representing these two data sets are almost superimposable. This correlation between basal FUS1 activity in wild-type cells and induced FUS1 activity in cells expressing Gpa1-G322Ep supports the idea that these measurements reflect –ß affinity. Thus, the A405V form of Ste4p is the least disruptive, and the F115S form of Ste4p is most disruptive to –ß binding. Note that although STE4-A405V has almost no impact on basal signaling, it confers as severe a defect in recovery from pheromone treatment as do any of the other alleles.

Taken together, the data discussed in this section indicate that one target of Gpa1p-mediated adaptation is G_ß. The observed defect in downregulation conferred by the STE4 Adp^- alleles can be explained in two ways. Gpa1p might stimulate the binding of an unknown regulator to G_ß, and the Adp^- mutations might disrupt this interaction. Candidates for this regulator include Akr1 (KAO et al. 1996 ; PRYCIAK and HARTWELL 1996 ), Syg1 (SPAIN et al. 1995 ), Cdc24 (SIMON et al. 1995 ; ZHAO et al. 1995 ), and Ste5 (WHITEWAY et al. 1995 )—proteins that are all thought to interact with Ste4p in vivo. Alternatively, the Adp^- forms of G_ß may simply be refractory to sequestration by Gpa1p after pheromone treatment, but more easily bound by Gpa1p during vegetative growth. Although our data are inconsistent with the idea that the STE4 Adp^- mutations merely cause the release of G_ß from Gpa1p, we cannot eliminate the possibility that they differentially affect the ability of Gpa1p to bind G_ß in vegetative and pheromone-treated cells. Slight defects in –ß affinity could also be overcome by the induction of downstream adaptive mechanisms in dividing cells. In other words, our failure to observe projection formation (shmooing) and poor growth (Table 1 and Figure 3) in cells newly transformed with the various STE4 Adp^- alleles could conceivably be caused by a rapid change in cellular physiology.

Ste4p phosphorylation and Gpa1p-mediated adaptation:
Like metazoan G_ß subunits, Ste4p is made up of a repeating amino acid sequence motif of ~40 residues. In addition to the seven repeat motifs found in metazoan G_ß subunits, however, Ste4p contains two unique insertions, one at the amino terminus and another between repeat motifs five and six. When cells are exposed to pheromone, Ste4p is rapidly phosphorylated within the internal domain, residues 310–346 (COLE and REED 1991 ; and E LI and D. STONE, unpublished results). Because this domain has been implicated in adaptation, we asked whether the STE4 Adp^- mutations affect the phosphorylation state of Ste4p. A priori, the Adp^- phenotypes could be caused by an inability to phosphorylate Ste4p. As shown in Figure 6, this is clearly not the case. After treatment with a submaximal dose of pheromone, the mutant forms of Ste4p were all fully phosphorylated. Our data are also inconsistent with the idea that Gpa1p stimulates Ste4p phosphorylation. Rather than augmenting phosphorylation of G_ß, a hyperadaptive form of Gpa1p actually inhibits it (Figure 7). Thus, although these data are strictly correlative, our failure to observe a correspondence between adaptation and Ste4p phosphorylation indicates a need to reexamine the role of G_ß modification in pheromone signaling.

Location of the Adp^- mutations in the three-dimensional structure of G_ß:
The crystal structures of the G_ß dimer from mammalian retinal cells (transducin ß) in its free form (SONDEK et al. 1996 ), complexed with a chimeric form of (LAMBRIGHT et al. 1996 ), and complexed with retinal phosducin (GAUDET et al. 1996 ) have recently been solved. Because most of the contacts between and ß involve residues that are conserved in both proteins, the interactions observed in the –ß crystal are likely to occur in other members of the heterotrimeric G protein family, including Gpa1p and Ste4p (LAMBRIGHT et al. 1996 ). Structural information about the Ste4p residues implicated in adaptation is summarized in Table 2, and the locations of the corresponding residues in the transducin ß crystal are shown in Figure 8.

View larger version (43K): In this window In a new window Download PPT slide   Figure 8. —Ribbon diagram of the transducin ß crystal structure. The N-terminal helix of transducin ß and the blades of the propeller are distinguished by color. Transducin gamma is shown in purple. The transducin ß residues that correspond to the substituted residues in the Adp- forms of Ste4p are shown in red and are labeled. The analogous residues of Ste4p are indicated in parentheses. The label 330-332 refers to G330, S331, and W332 of transducin ß; 409-411 refers to G409, S410, and W411 of Ste4p.

The transducin ß dimer is primarily a seven-bladed ß-propeller. Each of the blades is a ß sheet made of four antiparallel strands. The N-terminus of G_ß forms an helix that interacts with the first helix of the G subunit, and the second helix and extended portions of G partially encircle the ß propeller. Interaction between G and G_ß occurs at two distinct interfaces. Residues in the loops and turns at the top of the ß propeller domain of G_ß (blades 2–5, and 7) interact with residues in or adjacent to the switch I and switch II region of G. This is called the switch interface. The second –ß interface is formed by interaction between the N-terminal helix of G and the side of the first blade of the ß propeller. This is called the N-terminal interface. The surface area of the N-terminal interface is about half that of the switch interface.

Because the structures of the free and -bound forms of transducin ß are not significantly different, it is unlikely that the signaling activity of G_ß depends on a change in its conformation. Rather, key contact sites for G_ß effectors and regulators may be unmasked when the G protein subunits dissociate. By analogy to the mammalian –ß crystal structure, three of the Ste4p residues identified in this study are located in the regions of –ß interaction and are predicted to lie on the exposed surface of the free Ste4p-Ste18p dimer. Residues L138 (in blade 2) and W411 (in blade 7) are predicted to contact the switch region of Gpa1p; residue F115 (in blade 1) is predicted to contact the N-terminal helix of Gpa1p. In addition to interacting with Gpa1p, all three of these residues are readily accessible to effector and regulatory molecules. This raises the possibility that the Adp^- forms of Ste4p prevent proper downregulation of the mating signal by hyperstimulating the mating effector. It is not necessary, however, to invoke gain-of-function mutations to explain the data. It is more plausible that F115S, L138F, W411L, and W411S are loss-of-function mutations that prevent downregulation of G_ß by Gpa1p or by an unknown negative regulator. A precedent for such a molecule is phosducin, which downregulates G_ß moieties in a variety of metazoan systems. The recently solved crystal structure of the transducin ß-retinal phosducin complex (GAUDET et al. 1996 ) reveals direct contacts between the N-terminal domain of phosducin (helices 1 and 3) and the residues of transducin G_ß that are analogous to residues L138 and W411 of Ste4p. The structure of the transducin ß–phosducin complex also suggests that phosducin induces a conformational shift in G_ß. The positions of three loops at the top of blades 6 and 7 of the ß propeller are changed upon G_ß–phosducin binding. Interestingly, the most C-terminal of these loops corresponds to residues 408–417 of Ste4p. This suggests a mechanism by which the A405V, G409D, and S410L substitutions in Ste4p might prevent proper downregulation of the mating signal. Although the analogous residues of transducin G_ß contact neither G or phosducin directly, substitutions at these positions might affect a local change in the conformation of the ß propeller such that the binding of a negative regulator is inhibited.

In conclusion, although it is possible that the STE4 Adp^- mutations preferentially disrupt sequestration of G_ß by G after pheromone treatment, we do not favor this idea. We propose that the pheromone-responsive yeast G protein Gpa1p stimulates an unknown molecule to bind and downregulate its ß (Ste4p-Ste18p). The interaction between this putative downregulator and the Ste4p-Ste18p dimer is adversely affected by changes in Ste4p residues that are involved in direct contact (F115, L138, and W411) and by C-terminal mutations that impair an essential conformational shift (A405, G409, and S410). The F115S substitution most likely affects –ß binding as well as downregulation of G_ß by the unknown regulator.

	FOOTNOTES

¹ Present address: Glaxo Institute for Molecular Biology S. A., 14, chemin des Auix—Case Postale 674, 1228 Plan-les-Quates, Geneva, Switzerland.

	ACKNOWLEDGMENTS

The authors would like to thank HEIDI HAMM for helpful discussions and for creating Figure 8; FRED CROSS, GARY COLE, PHILLIP NAGLEY, and HAY-OAK PARK for plasmids; and MARY CISMOWSKI and STEVE REED for helpful discussions and for critical reading of the manuscript. This work was supported by an American Cancer Society Research grant (VM-92A) to D.E.S.

Manuscript received April 1, 1997; Accepted for publication November 21, 1997.

	LITERATURE CITED