Truncated Forms of a Novel Yeast Protein Suppress the Lethality of a G Protein α Subunit Deficiency by Interacting with the β Subunit

In Saccharomyces cerevisiae, the mating pheromone-initiated signal is transduced by a heterotrimeric G protein and normally results in transient cell cycle arrest and differentiation. A null allele of the Gα (GPA1/SCG1) subunit results in cell death due to unchecked signaling from the Gβγ (STE4, STE18, respectively) heterodimer. We have identified three high copy suppressors of gpa1 lethality. Two of these genes encode known transcription factors, Matα2p and Mcm1p. The third is a truncated form of a novel gene, SYG1. Overexpressed wild type SYG1 is a weak suppressor of gpa1. In contrast, the isolated mutant allele SYG1-1 is a strong suppressor that completely blocks the cell cycle arrest and differentiation phenotypes of gpa1 cells of both mating types. One deletion mutant (SYG1Δ340) can suppress the cell cycle arrest associated with gpa1, but the cells retain a differentiated morphology. SYG1-1 can suppress the effects of overexpressed wild type Gβ but is not able to suppress the lethality of an activated Gβ mutant (STE4Hpl). Consistent with these genetic observations, the suppressing form of Syg1p can interact with the STE4 gene product, as determined by a two-hybrid assay. SYG1-1 is also capable of promoting pheromone recovery in wild type cells, as judged by halo assay. The sequence of SYG1 predicts eight membrane-spanning domains. Deletion mutants of SYG1 indicate that complete gpa1 suppression requires removal of all of these hydrophobic regions. Interestingly, this truncated protein localizes to the same plasma membrane-enriched subcellular fraction as does full-length Syg1p. Three hypothetical yeast proteins, identified by their similarity to the SYG1 primary sequence within the gpa1 suppression domain, also appear to have related structures. The properties of Syg1p are consistent with those of a transmembrane signaling component that can respond to, or transduce signals through, Gβ or Gβγ.

In Saccharomyces cerevisiae, the mating pheromoneinitiated signal is transduced by a heterotrimeric G protein and normally results in transient cell cycle arrest and differentiation. A null allele of the G␣ (GPA1/SCG1) subunit results in cell death due to unchecked signaling from the G␤␥ (STE4, STE18, respectively) heterodimer. We have identified three high copy suppressors of gpa1 lethality. Two of these genes encode known transcription factors, Mat␣2p and Mcm1p. The third is a truncated form of a novel gene, SYG1. Overexpressed wild type SYG1 is a weak suppressor of gpa1. In contrast, the isolated mutant allele SYG1-1 is a strong suppressor that completely blocks the cell cycle arrest and differentiation phenotypes of gpa1 cells of both mating types. One deletion mutant (SYG1⌬340) can suppress the cell cycle arrest associated with gpa1, but the cells retain a differentiated morphology. SYG1-1 can suppress the effects of overexpressed wild type G␤ but is not able to suppress the lethality of an activated G␤ mutant (STE4 Hpl ). Consistent with these genetic observations, the suppressing form of Syg1p can interact with the STE4 gene product, as determined by a two-hybrid assay. SYG1-1 is also capable of promoting pheromone recovery in wild type cells, as judged by halo assay. The sequence of SYG1 predicts eight membrane-spanning domains. Deletion mutants of SYG1 indicate that complete gpa1 suppression requires removal of all of these hydrophobic regions. Interestingly, this truncated protein localizes to the same plasma membrane-enriched subcellular fraction as does full-length Syg1p. Three hypothetical yeast proteins, identified by their similarity to the SYG1 primary sequence within the gpa1 suppression domain, also appear to have related structures. The properties of Syg1p are consistent with those of a transmembrane signaling component that can respond to, or transduce signals through, G␤ or G␤␥.
The budding yeast Saccharomyces cerevisiae has two haploid cell types, a and ␣, which are defined by the gene cassette expressed from the MAT locus. Haploids of opposite mating type can conjugate to yield a/␣ diploids. Expression of the genes that determine cell identity (a, ␣, or a/␣) is controlled in large part through the transcription factor encoding genes MATa1, MAT␣1, MAT␣2, and MCM1 (reviewed in Refs. 1 and 2).
Mating between haploids of opposite mating type is initiated by the binding of pheromone, secreted from each cell type, to specific receptors expressed on cells of the opposite mating type. Pheromone binding triggers G 1 cell cycle arrest, a differentiation program that leads to morphologically altered cells (shmoos) that are enlarged and elongated, and induction of various genes required to consummate cell fusion (reviewed in Ref. 3). In this state, cells of opposite mating type fuse at their projection points. Subsequent nuclear fusion leads to mating resolution and eventual resumption of mitotic growth. Cell cycle arrest and differentiation can also be brought about by exogenously supplying the appropriate pheromone in the absence of opposite mating type cells. After some time, however, the arrested cells adapt to the pheromone and resume vegetative growth.
Mating signal transduction (reviewed in Refs. [3][4][5] begins with the binding of pheromone to its cognate receptor, ␣-factor receptor (Ste2p) on a cells or a-factor receptor (Ste3p) on ␣ cells. Pheromone binding triggers activation of a heterotrimeric G protein by releasing the G␣ subunit (encoded by the GPA1 gene, also called SCG1) from the signal transducing G␤␥ subunits (encoded by the STE4 and STE18 genes, respectively). Even in the absence of pheromone, cell cycle arrest, differentiation, and ultimately cell death can result from (i) a deficiency of the G␣ subunit (6,7), (ii) a G␤ subunit mutant that is insensitive to repression by G␣ (8,9), or (iii) overexpression of the wild type G␤ subunit (10). These data demonstrate that G␤␥ is the principal transducer of the signal for cell cycle arrest and differentiation and that G␣ acts as a repressor that is necessary for signal attenuation. It is unclear, however, whether G␣ works solely through G␤␥ repression or if it can actively stimulate an independent adaptation pathway.
Many components of the arrest-differentiation pathway downstream of G␤␥ have been identified through the analysis of recessive mutants, many of which are sterile (ste). These include the protein kinases Ste20p, Ste7p, Ste11p, Fus3p, and Kss1p. Cell cycle arrest is potentiated by Far1p through a direct interaction with a cyclin-dependent kinase (11), while pervasive changes in gene expression are mediated by the action of the Ste12p transcription factor (12). Less is known, however, about the immediate effector of G␤␥.
In this study we describe the isolation of three high copy gpa1 suppressors. Two of these are known to control the expression of components of this pathway. The third is a novel gene with truncated forms that can suppress either cell cycle arrest alone or both cell cycle arrest and differentiation. We demonstrate that this suppressor protein is tightly associated with the plasma membrane and that it acts to block G protein signaling at the level of G␤␥.

MATERIALS AND METHODS
Strains, Transformation, Selection, and Propagation Procedures-Strains are described in Table I. Strains SP1 and RS22-6C were mated to make diploid strain SR. Strain SRgHU was created by sequentially transforming strain SR with a gpa1::HIS3 EcoRI fragment from pgH and a gpa1::URA3 EcoRI fragment from pgU in order to consecutively eliminate both copies of GPA1. Strain SRgHU was then transformed with the pLGC plasmid. Following sporulation and tetrad dissection, strains GU1 and GU2 were obtained. Strain DB6 was generously provided by Duane Jenness, University of Massachusetts, Worcester. Strain DBC was created by first plating DB6 cultures on canavaninecontaining media to select for can1 mutants. After introducing pU␣2C, transformants were screened on the basis of colony color and a Leu Ϫ phenotype for isolates which had lost pDJ117. YDM400 was generously provided by Doreen Ma and Jeremy Thorner, University of California, Berkeley. GPY74 -15Ca was generously provided by Greg Payne, UCLA. Standard formulations were used for YPD, synthetic complete (SC), 1 and sporulation media, and all yeast cultures were grown at 30°C. All yeast transformations were done as described previously (13), and all plasmids were propagated in DH5␣ cells.
Suppressors of gpa1 lethality were isolated by first transforming GU1 cells with a library of yeast genomic fragments cloned into pUV1, a 2based plasmid carrying the URA3 marker (generous gift of Junichi Nikawa and Michael Wigler). 1% of the transformed cells were plated on SC-uracil to determine the total number of transformants. The remainder were grown in 20 ml of SC-uracil media for 2 days, and a sample of these cells (200 l) was plated on SC-uracil-arginine with canavanine (60 g/ml) added. Individual colonies were picked and grown in SC-uracil for preparation and analysis of plasmid DNA.
Plasmids-The GPA1 knock out constructs (pgH and pgU) were created by inserting blunt-ended HIS3 and URA3 fragments between the internal SspI sites of GPA1. pLGC was created by inserting the Klenow-blunted EcoRI fragment of GPA1 into the PvuII site of pTLC (14), which carries the LEU2 and CAN1 genes. pUV1-SYG1-1 is the library-derived clone. Vector pUV2 differs from pUV1 only in the orientation of URA3. pU␣2C was created by inserting a 4-kb CAN1 gene fragment (from pYeCAN1-2, provided by Kelly Tatchell, North Carolina State University) between the KpnI and SacI sites of pUV1-␣2, the library isolate of MAT␣ 2. pUG carries the GPA1 gene on a 2-based URA3 vector. pUGC is a version of pUG that includes the CAN1 gene. The p416-SYG1-1 centromeric version of SYG1-1 was created by cloning the KpnI/XbaI SYG1-1 fragment from pUV1-SYG1-1 into the KpnI/ XbaI sites of pRS416 (15).
The SYG1 disruption construct using the LEU2 marker was made as follows. The KpnI/XbaI SYG1-1 fragment was transferred into pBluescript KS (pKS, Stratagene) to form pKS-SYG1-1. To introduce the LEU2 marker, the HindIII/EcoRV fragment of SYG1-1 was removed from pKS-SYG1-1 and replaced by LEU2 on a HindIII/SmaI fragment. A NotI/SmaI syg1::LEU2 disruption fragment was used to transform strains SP1 and FY250. Southern and Northern analyses were used to confirm syg1::LEU2 disruptions.
Full-length pUV1-SYG1 and the deletion pUV1-SYG1⌬554 were created using the cDNA clones pKS-SYG1A and pKS-SYG1D, respectively. For both constructs, the polylinker EcoRI site of pUV1-SYG1-1 was changed to a NotI site using the adaptor 5Ј-AATTGCGGCCGC-3Ј (all oligos were obtained from Integrated DNA Technologies) to form pUV1-SYG1-1N. To make pUV1-SYG1, the polylinker XhoI site of pKS-SYG1A was changed to a NotI site using the oligomeric adapter 5Ј-TCGAAGCGGCCGC-3Ј to make pKS-SYG1AN. pKS-SYG1D already contained a polylinker NotI site at the 3Ј end of the clone. The 3Ј-SYG1 NheI/NotI fragments from pKS-SYG1AN and pKS-SYG1D were used to replace the NheI/NotI fragment of pUV1-SYG1-1N. The construct pUV1-SYG1 contains its own natural stop codon. A second pUV1-SYG1 construct was created from an independent cDNA isolate, and the two pUV1-SYG1 constructs behaved identically.
Three COOH-terminal deletion constructs were made using PCRgenerated fragments. pUV1-SYG1⌬519, pUV1-SYG1⌬464, and pUV1-SYG1⌬400 were created by replacing the SacI/EcoRI fragment of pUV1-SYG1⌬554 with PCR-generated fragments whose SYG1 sequence ended at the indicated codon and introduced an EcoRI site for cloning purposes. For pUV1-SYG1⌬464 and pUV1-SYG1⌬400, a TAA stop codon was introduced after the last SYG1 codon, whereas for pUV1-SYG1⌬519, the TAG stop codon was supplied by the vector. Each PCR used the upstream primer YS5 (5Ј-TTCATGTCGTACGCCAGG-3Ј) which is 5Ј of the SacI site. The downstream primers were 5Ј-TAGAAT-TCTTAAATCGATCTGTTGTTTCT-3Ј (SYG1⌬400), 5Ј-TAGAATTCT-TAGGTTCTATGCCAGATAAA-3Ј (SYG1⌬464), and 5Ј-TAGAATTCT-TATCCAAAGCGAAACTTAAC-3Ј (SYG1⌬519). For cloning, the PCRgenerated fragments were digested with SacI and EcoRI prior to gel purification and ligation. PCRs were done using three cycles of (94°C for 1 min, 40°C for 1 min, 72°C for 1.5 min) followed by 25 cycles of (93°C for 0.5 min, 55°C for 0.5 min, 72°C for 2 min) followed by 72°C for 5 min. Multiple independent PCR-derived clones of each construct were assayed with the same result. pUV1-SYG1⌬340 was constructed by removing the 5Ј-SacI restriction fragment from pUV1-SYG1-1 using the internal SacI site and a polylinker SacI site. pUV1-SYG1⌬199 was constructed by removing the 5Ј-NheI/NotI fragment from pUV1-SYG1-1N by digesting with NotI and NheI, blunting the termini with Klenow, and then ligating. For pUV1-SYG1⌬340 and pUV1-SYG1⌬199, the stop codon signal was provided by vector sequence. pADNS-SYG1-1 was constructed by first moving the HindIII/NotI fragment of pUV1-SYG1-1N into pADNS (16). The HindIII site was changed to a SalI site using the adaptor 5Ј-AGCTGTCGAC-3Ј. This construct was designated pADNS-SYG1-1SN. The most NH 2 -terminal portion of SYG1-1 was moved as a PCR-amplified fragment which introduced a SalI site and an in-frame start ATG codon. The NH 2terminal PCR primer was 5Ј-AGTCGTCGACAATGAAGTTTGCTGAC-CAT-3Ј. The downstream amplification primer 5Ј-GCTGTCGATGC-TATTGAC-3Ј was positioned 3Ј of the SYG1-1 PstI site. The amplified fragment was digested with SalI and PstI prior to being ligated into pADNS-SYG1-1SN. The PCR-amplified portion of pADNS-SYG1-1 was sequenced to confirm that no errors had been introduced. This construct was tested and shown to have gpa1 suppression activity in LG1TG and LG2TG, gpa1 strains that carry the leu2 mutation, although the level of suppression (number of survivors on canavanine-containing media) was a few fold below what is seen for pUV1-SYG1-1 in GU1 or GU2 cells. pADNS-SYG1 was constructed by substituting the SacI/NotI fragment of cDNA clone pKS-SYG1E for the incomplete pADNS-SYG1-1 SacI/ NotI fragment. pKS-SYG1E differs from pKS-SYG1A only in that pKS-SYG1E contains a poly(A) tail. To create the COOH-terminal hemagglutinin (HA)-tagged expression constructs, the SalI/NotI fragment from pADNS-SYG1 was cloned into pADCLX, which was derived from pAD54 (13) and has an in-frame NotI site preceding, and a unique KpnI site following, the HA epitope. This intermediate construct was called pCLXSF. To create pADCLX-SYG1⌬400, the SacI/NotI fragment of pCLXSF was replaced by the PCR fragment generated using YS5 as the upstream primer and 5Ј-GCTAGCGGCCGCCCAATCGATCTGTT-GTTTCT-3Ј as the downstream primer. To create pADCLX-SYG1, a fragment with a 3Ј in-frame NotI site was generated by PCR using the upstream primer 5Ј-TTGTGCGGTCTGTTCCAT-3Ј and the downstream primer 5Ј-GCTAGCGGCCGCCCCATAATACTTTCCACTTC-3Ј. This PCR product was cloned as an EcoRI/NotI fragment into pUV1-SYG1 and then moved as a SacI/NotI fragment into pCLXSF. To create pCLA-SYG1⌬400, the expression construct which fused the LexA coding sequence to the COOH terminus of SYG1⌬400, the following steps were taken. The vector pKSN was made by changing the XhoI site in the vector pKS to an NcoI site using the adaptor 5Ј-TCGACCATGG-3Ј. LexA was PCR amplified using the upstream primer 5Ј-GCATGCGGC-CGCGGATCCTTATGAAAGCGTTAACGGCC-3Ј which introduced inframe NotI and BamHI restriction sites and the downstream primer 5Ј-TACGCCATGGTTACAGCCAGTCGCCGTT-3Ј which added a stop codon and NcoI restriction site. The resulting LexA amplicon was cloned into pKSN. From pKSN-LexA, a NotI/KpnI LexA fragment was used to replace the HA epitope of p9CL-SYG1⌬400, creating pCLA-SYG1⌬400. p9CL-SYG1⌬400 has the SphI/SphI fragment of pADCLX-SYG1⌬400 inserted into the SphI/SphI backbone of pGBT9 (17).
YEp-␣2 was constructed by ligating the SacI/SmaI fragment of the original MAT␣2 genomic isolate into the SacI/SmaI sites of YEp13 (14). YEp-MCM1 was constructed by ligating the SacI fragment of the original MCM1 genomic isolate into YEp13. YCpGAL-STE4 for overexpressing STE4 under galactose-inducing conditions was generously provided by Steven Reed, Scripps Research Clinic.
Cell Extract Preparation, Fractionation, Membrane Stripping, and Immunoblot Analysis-Crude cell extract preparation was done as described (13) except that the lysis buffer was made of 20 mM 2-MOPS, pH 6.0, 1 mM MgCl 2 , 100 M EDTA, and 1 mM dithiothreitol. Crude cell extract fractionation by differential centrifugation was done as described (19). Immunoblot analysis using anti-HA monoclonal antibody was done as described (13). Marker proteins were detected similarly using the following polyclonal anitbodies: anti-Na ϩ -K ϩ -ATPase (1:500 dilution) for the plasma membrane, anti-glucose-6-phosphate dehydrogenase (1:3,000 dilution) for the cytosol, and anti-Sec63p (1:2,500 dilution) for the endoplasmic reticulum. These antibodies were generous gifts from the laboratories of Nathan Nelson (Roche Institute, Nutley, NJ), Greg Payne (UCLA), and David Meyer (UCLA), respectively. Immunoblots treated with polyclonal primary antibody were incubated with alkaline phosphatase-conjugated goat anti-rabbit antibody (Bio-Rad, 1:2,500 dilution). For the membrane stripping experiments, aliquots of the 10,000 ϫ g pellets were diluted into 0.5 ml of stripping buffer (0.5 M NaCl, 2 M urea, 10 mM Tris-HCl, pH 10.5, 0.5% ␤-mercaptoethanol, 0.5% Triton X-100, or 1% SDS) and incubated for at least 1 h on ice with occasional mixing. The samples were then centrifuged at 10,000 ϫ g for 10 min at 4°C. The pellets were resuspended in lysis buffer, and the supernatants were either concentrated at 4°C using Microcon-10 microconcentrators (Amicon) or acetone-precipitated.
Mating, Halo, Two-hybrid, STE4 Hpl , and Overexpressed Wild Type STE4 Suppression Assays-Quantitative mating assays were conducted essentially as described (20) except that approximately 10 6 cells of the tester strains DC14 or DC17 were used, and the experimental strains were cultured in plasmid maintenance synthetic medium before being switched into YPD medium. Mating efficiencies were calculated as the ratio of diploid colonies formed to the titer of experimental haploid cells used.
Halo assays were performed essentially as described (21) except that the experimental strains were cultured as for the quantitative mating assay and approximately 5 ϫ 10 5 cells were inoculated into either SC or YPD soft agar which was poured onto the corresponding media. Results shown used the YPD conditions which facilitated a slightly enhanced response to ␣-factor over the synthetic medium conditions. Synthetic ␣-factor (Sigma) was added to the filter discs in a volume of 10 l of water.
FUS1 induction analyses were done using the bar1/sst1 strain GPY74 -15Ca which was treated with ␣-factor at a concentration of 1 g/ml. After 2 h of incubation, Ͼ80% of the cells were unbudded. Time points for RNA analyses were taken prior to addition of pheromone (T ϭ 0) and 15, 60, 120, and 180 min after addition of pheromone. FUS1 induction as determined by reverse transcribed-PCR occurred as described (22). FUS1 induction was also measured from a 2-based FUS1-LacZ reporter construct, pSB234 (23). This was transformed into GPY74 -15Ca cells, followed by pADNS, pADNS-SYG1-1, YEp-␣2, and YEp-MCM1. Cells were grown overnight in selective media, switched to YPD, treated with ␣-factor for 1 h, lysed, and assayed for ␤-galactosidase activity (24). The two-hybrid ␤-galactosidase liquid assays were done similarly except that lysis was accomplished by two cycles of freeze-thaw using liquid nitrogen. To test for His ϩ complementation, individual transformants were patched onto media selecting for each of the two introduced plasmids and then replica plated to media that also selected for HIS3 expression. The patches on His ϩ selection media were replica plated a second time onto the same media to eliminate background growth. Filter lift ␤-galactosidase assays were done as described (25).
STE4 Hpl suppression was assayed by transforming DBC cells, culturing them for 1-2 days in SC-leucine media, and then plating 1-10 l of culture onto SC-leucine-arginineϩcanavanine media to select for loss of the pU␣2C maintenance plasmid. Suppression of overexpressed wild type STE4 was done essentially as described (10) except that individual transformants were first patched onto media containing sucrose and subsequently replica plated onto media containing galactose (3%). To enhance the growth difference, these were replica plated onto the same media a second time.
RNA Analysis, Hybridizations, Phage Library Screening, Sequencing, and PCR-Total RNA was prepared using the hot phenol method (26). For Northern analysis, total RNA was fractionated on a 0.8% agarose, 2% formaldehyde gel prior to transfer to reinforced nitrocellulose. Reverse transcribed-PCR studies employed conditions as described (27) except that 2 g of total RNA was used for cDNA synthesis and an annealing temperature of 54°C was used for all PCR reactions. The number of PCR amplification cycles and agarose percentage required for fragment detection by ethidium-stained agarose gel analysis depended on the fragment. The upstream and downstream primers, respectively, and the fragment sizes were as follows: ACT1 (5Ј-TT The S. cerevisiae cDNA library used was created by Jeff Kuret (Cold Spring Harbor Laboratory) (28). Standard screening and hybridization conditions were used. Sequencing was performed using dideoxy termination reactions.
Data base searches were performed using the BLAST algorithm (29).

RESULTS
Isolation of gpa1 Suppressors-To identify genes involved in regulating the pheromone response in S. cerevisiae, a genetic selection for yeast sequences capable of rescuing gpa1 cells was undertaken using a plasmid exchange protocol. Strain GU1 has a gpa1 null allele but is viable because it has an extrachromosomal copy of GPA1 carried on the plasmid pLGC, which also encodes a positive selectable marker (LEU2) and a negative selectable marker (CAN1). GU1 cells were transformed with a library of yeast genomic fragments carried on a 2based high copy plasmid. Approximately 5 ϫ 10 5 transformants were obtained from seven separate transformations. These cells were grown in synthetic media which selected for the library plasmid but allowed for the loss of pLGC. Transformants which lost the pLGC plasmid were identified using the negative selectable marker. Plasmid DNA was isolated from individual colonies and used to transform fresh GU1 cells which were then tested for suppression of gpa1 lethality.
A total of seven independent plasmids that could suppress gpa1 were isolated. Following restriction mapping, cross-hybridization and sequence analysis, it was determined that five clones contained the MAT␣2 gene and one clone carried MCM1. When tested in ␣ cells (GU2), MAT␣2, but not MCM1, was able to give gpa1 suppression. These genes encode proteins that act together and with other factors to control transcription of mating type-specific genes, including some that are required for the pheromone signaling pathway (1, 2).
The remaining clone was found to contain a portion of a novel gene, SYG1 (suppressor of yeast gpa1). The cloned form of the gene, SYG1-1, was also capable of suppressing gpa1 in an ␣ strain (GU2) that is otherwise isogenic with GU1. This demonstrated that suppression is likely to involve a generalized block of cell cycle arrest and not a mating type-specific interference.
Sequence analysis revealed that SYG1-1 had a 417-amino acid open reading frame that is a carboxyl-terminal truncation of SYG1 (Fig. 1). A portion of this sequence was used as a probe to isolate the remainder of the SYG1 coding region from a S. cerevisiae cDNA library. Four distinct SYG1 cDNAs were isolated and sequenced. All were collinear with the original SYG1-1 clone and contained additional 3Ј sequences that extended the open reading frame to 902 amino acids (Fig. 1B). The cDNA sequences were used to reconstruct a full-length SYG1 coding region attached to the SYG1 promoter. Fulllength SYG1, expressed from the same high copy expression plasmid as SYG1-1, was an extremely weak suppressor of gpa1. Relative to SYG1-1, suppression by SYG1 resulted in fewer colonies (Fig. 2), and these took twice as long to appear.
A striking feature of the SYG1 primary sequence was revealed by hydropathy and hydrophobic moment analyses. As shown (Figs. 1 and 3), several distinct hydrophobic domains are located within the full-length protein, and eight of these are predicted to span the membrane. The eighth domain was noted to be amphophilic, raising the possibility that it may form a pore. The truncation giving rise to SYG1-1 resulted in disruption of the first hydrophobic domain and deletion of the others. This suggested that removal of these domains conferred upon SYG1-1 its potent gpa1 suppression activity. FIG. 1. A, restriction map of SYG1. The SYG1 clone is shown as a 3-kb fragment extending from the upstream genomic XbaI site to the end of the SYG1A cDNA. The hatched box represents the open reading frame. Restriction sites are labeled as follows: X, XbaI; H, HindIII; P, PstI; N, NheI; RV, EcoRV; S, SacI, and RI, EcoRI. B, predicted amino acid sequence of SYG1. The termination codon is marked by a dot. The filled diamond positioned above the sequence indicates the end of the SYG1 sequence obtained in the original SYG1-1 clone. Delta symbols above the sequence indicate the positions of the deletion constructs tested for gpa1 suppression. Predicted transmembrane domains are underlined. SYG1 was recently determined to be located on chromosome IX (accession number Z46861). C, alignment of two regions of predicted protein sequence for Syg1p and four related yeast sequences; N2052, J0336, YCR524, and Pho81p (accession numbers X77395, X77688, X56909, and X52482, respectively). Identical residues are boxed, conserved residues (Ն0.3 from Dayhoff table (60)) are shown, and amino acid positions are given at the right. There are no gaps introduced.
Northern analysis (Fig. 4) showed that SYG1 is expressed as a 2.8-kb transcript in both a and ␣ haploids as well as in a/␣ diploids. Expression in GU1 and GU2 cells which overexpress GPA1 (lanes 1 and 2), appeared to be slightly repressed compared to levels in a wild type haploid cell (lane 4) or a diploid cell (lane 3), suggesting possible regulation by some component(s) of this pathway. Two smaller transcripts also appeared on the blot. These are likely to be SYG1 derived since they were eliminated in a syg1 deletion strain (lane 5). No change in SYG1 expression level was seen following treatment with ␣-factor (data not shown).
Strains that carry a syg1::LEU2 mutation showed no noticeable change in mating efficiency. These cells did not display any generalized defect in cell growth under normal conditions nor when exposed to stresses including growth at 37°C, heat shock treatment at 55°C, and hyperosmotic media. They were able to switch to alternate carbon sources, undergo meiosis, and give rise to viable spores. We considered the possible existence of genes that are functionally redundant with SYG1 and can thereby mask the effects of the syg1 null mutation. Although we have not detected any related sequences using low stringency hybridizations, four sequences with significant, though limited, similarity to SYG1 were identified through a data base search (Fig. 1C). The two regions of similarity with SYG1 lie within the sequence of SYG1-1. In addition, three of these sequences (N2052, J0336, and YCR524) encode predicted proteins that, like SYG1, have amino-terminal spans of about 400 hydrophilic residues followed by multiple strongly hydro-phobic domains. Overexpression of the amino termini of these proteins does not confer any detectable gpa1 suppression, however (data not shown).
We performed a carboxyl-terminal deletion analysis of SYG1 to further examine structure-function relationships (Fig. 5). SYG1⌬554, SYG1⌬519, and SYG1⌬464 showed no gpa1 suppression activity in either GU1 or GU2 cells. These constructs, then, did not display even the very weak suppression seen with full-length SYG1, perhaps due to altered protein conformation. The SYG1⌬400 mutant, in which all hydrophobic domains were eliminated, rescued gpa1 in both GU1 and GU2 as effectively as SYG1-1 (SYG1⌬417). The SYG1-1 and SYG1⌬400 rescued cells showed normal morphologies (Fig. 6, data not shown for SYG1⌬400). In contrast, deletion of 60 amino acids from SYG1⌬400 produced a surprising result. SYG1⌬340 was capable of suppressing the growth arrest associated with gpa1, but not the differentiation. These cells were able to undergo mitosis while maintaining a shmoo morphology (Fig. 6). Deletion of another 141 amino acids (SYG1⌬199) prevented any rescue.
Subcellular Localization of Full-length and Truncated Syg1p-Initially, we hypothesized that the dramatic increase in gpa1 suppressor activity of SYG1⌬400 compared to fulllength SYG1 was attributable to altered subcellular localization of the truncated protein. To test this model, crude extracts containing epitope-tagged Syg1⌬400p and Syg1p were subjected to differential centrifugation. It should be noted that the epitope fusion product of SYG1⌬400 retained gpa1 suppression activity (data not shown). Immunological analysis of marker proteins confirmed organellar separation. Surprisingly, both Syg1⌬400p and Syg1p localized to the particulate fraction, which was greatly enriched for plasma membrane (Fig. 7A). Therefore, altered subcellular localization of Syg1⌬400p does not account for its potent gpa1 suppression.
This subcellular localization result suggested that Syg1⌬400p may associate with another membrane-localized protein or contain its own membrane targeting signal, even though no such signal is predicted by the sequence. Treatment of the plasma membrane-enriched fraction with reagents (high salt, high pH, urea and ␤-mercaptoethanol) that typically disrupt peripheral protein-protein interactions did not release Syg1⌬400p (Fig. 7B). Triton X-100 and SDS were used to detect the presence of protein-membrane interactions. 1% SDS completely solubilized Syg1⌬400p, but 2% Triton X-100 was unable to do so for both Syg1⌬400p (Fig. 7B) and Syg1p (data not shown).

SYG1-1 Is a High Copy Suppressor in Mutant and Wild
Type Cells-The ability to rescue gpa1 requires high level expression of SYG1-1. This was demonstrated by expressing SYG1-1 on a centromere-based plasmid that is maintained as a low copy episome. This construct did not suppress gpa1, demonstrating that high levels of Syg1-1p are needed.
To address the possibility that SYG1-1 was interfering with signal transduction by directly or indirectly altering expression of signaling components, we performed quantitative RT-PCR. High copy SYG1-1 expressed in the wild type strains SP1 and FY250, as well as in the gpa1 pUV1-SYG1-1 strains GU1S and GU2S, had no effect on mating type-specific expression of pheromone receptors (in a MATa background STE2 is expressed at normal levels and STE3 is not expressed, as in wild type cells) nor on the normal abundance of STE4 message (data not shown). In addition, overexpression of SYG1-1 did not appear to block the induction of FUS1 after pheromone treatment. This result was confirmed using direct detection of a FUS1 promoter-driven LacZ construct (23). As seen in Table II, expression of SYG1-1 only slightly dampens FUS1 induction. As expected, expression of MAT␣2 had a strong inhibitory effect. Thus, for all of these signaling components, we found no evidence that message regulation is significantly altered by SYG1-1.
The ability of SYG1 and SYG1-1 to modulate response to pheromone was tested using a halo assay. Application of pheromone was used to create a zone of growth inhibition that reflects the normal cell cycle arrest response following activation of the mating pathway. Although wild type cells with SYG1 gave normal halos, cells overexpressing SYG1-1 gave rise to halos that became partially filled with colonies (Fig. 8A). Colonies within the halo were visible soon after those that formed the lawn. A similar result was seen when GPA1 was overexpressed in wild type cells (Fig. 8B), probably due to the sequestering of Ste4p into an inactive complex with Gpa1p. In both cases, the diameter of the halo, representing the initial sensitivity to pheromone, was unchanged from wild type cells. These data are consistent with the FUS1-lacZ data in which SYG1-1 had little effect on initial FUS1 induction. Simultaneous overexpression of GPA1 and SYG1-1 did not further enhance the growth of cells within the halo (Fig. 8B). This indicated that Gpa1p and Syg1-1p may compete for the same target (Ste4p). It also suggested that there may be a limit to the level FIG. 5. Deletion mutants of SYG1. All constructs have a wild type amino terminus and are deleted from the carboxyl terminus. The deletion number indicates the extent of SYG1 amino acid sequence present. The ⌬417 mutant is marked with an * to indicate that this is the SYG1-1 construct originally isolated. For full-length SYG1, ϩw indicates very weak suppression (see text). The dotted region must be removed for full gpa1 suppression. The black region is needed for suppression of differentiation (shmooing), and the diagonally striped region is needed for suppression of cell cycle arrest.
FIG. 6. Proliferation of differentiated cells. GU1 cells were transformed with pUV1-SYG1-1 and pUV1-SYG1⌬340, which are designated pSYG1-1 and pSYG1⌬340, respectively. Cells which lost the maintenance plasmid (pLGC) were selected and then grown on YPD for 1 or 2 days. Photos were taken using a Nikon FXA microscope using a ϫ60 objective and DIC (Nomarski) optics. Similar results were obtained using strain GU2. , and pADCLX-SYG1 (lane 3) were subjected to differential centrifugation. Approximately 35 g of the 10,000 ϫ g pellet and supernatant were analyzed for expression of the epitope-tagged gene products as well as the organellar marker proteins (PM, plasma membrane, Na ϩ -K ϩ -ATPase; cytosol, glucose-6-phosphate dehydrogenase; ER, endoplasmic reticulum, Sec63p). Indicated to the left are the apparent sizes of the molecular mass standards in kDa (Bio-Rad, low range, prestained). Syg1p migrated consistent with its predicted molecular mass of 104 kDa. Depicted here, Syg1⌬400p migrates slower than its predicted molecular mass of 47 kDa. However, using Sigma prestained standards, Syg1⌬400p migrates consistent with its predicted molecular mass. B, solubilization of Syg1⌬400p from the particulate 10,000 ϫ g fraction. The immunoblot shows that only 1% SDS was capable of releasing Syg1⌬400p from the particulate fraction, whereas 0.5 M NaCl, 0.5% Triton X-100, 2 M urea, 10 mM Tris-HCl, pH 10.5, and 0.5% ␤-mercaptoethanol did not. The Triton X-100 insolubility was also observed using 2% Triton X-100 (data not shown). After treatment with the solubilizing reagant, the samples were centrifuged at 10,000 ϫ g for 10 min to yield pellet (P) and supernatant (S) fractions. b Modified Miller units (24). Induction procedures are described under "Materials and Methods." Each activity value is the average of four or six independent assays. The entire experiment was repeated with similar results. of signal repression and/or adaptation enhancement that can be achieved in this way.

TABLE II Pheromone induction of FUS1-LacZ
The turbid halo results indicated that SYG1-1 may function to relieve cell cycle arrest by stimulating adaptation. Sst1p and Sst2p mediate separate adaptation pathways which, when disrupted (sst1 or sst2), result in a supersensitive response to pheromone. The Bar1p/Sst1p protease normally contributes to adaptation by degrading ␣-factor (30) while Sst2p functions in an independent pathway (21,31,32). Overexpression of SYG1-1 in sst1 or sst2 cells gave turbid halos with diameters that were unchanged (Fig. 8, C and D). These data indicated that SYG1-1 promotion of pheromone recovery is independent of Sst1p and Sst2p.
For GPA1, the co-overexpression of full-length SYG1 did not alter the halo fill-in effect (Fig. 8B). Similarly, overexpression of non-suppressing truncations of SYG1 did not modulate pheromone recovery mediated by SYG1-1 (data not shown). In addition, syg1::LEU2 mutants showed no alteration in halo formation from wild type cells (Fig. 8E). In cells expressing SYG1-1 that are also gpa1 no halo is ever visible (Fig. 8G).
Expression of SYG1-1 had very little effect on mating efficiency. Both wild type cells (SP1) and GU1 cells transformed with pUV1-SYG1-1 mated at levels that were essentially unchanged from untransformed controls (within 2-fold, data not shown). gpa1 cells that carried the SYG1-1 suppressor did show reduced mating efficiency (about 20-fold, data not shown). Given that these cells showed no pheromone response in a halo assay, it was surprising that they mated at this level of efficiency (no mating with cells of the same mating type was observed). This is, however, similar to the level of mating seen in gpa1 ts receptorless strains (33). Thus, our data confirm that gpa1 cells are indeed mating competent.
SYG1-1 Suppresses Excess Wild Type STE4 but Not a Dominant Active STE4 -To genetically localize the effects of SYG1-1 action, we examined the ability of this allele to suppress the effects of high levels of wild type STE4. The overexpression of STE4 is known to give a phenotype similar to that of the gpa1 mutation, and it can be suppressed by the overexpression of GPA1 (10). We used an inducible construct in which STE4 overexpresssion and the resulting cell cycle arrest were triggered by galactose. As seen in Fig. 9, SYG1-1 was able to suppress the cell cycle arrest caused by STE4 overexpression, although the full-length SYG1 clone was not.
To test whether Syg1-1p might be suppressing through a direct interaction with Ste4p, as is the case for Gpa1p, we examined the ability of this allele to suppress STE4 Hpl , a dominant mutant in the G␤ subunit (8). The STE4 Hpl mutant is lethal because the mutant protein is less able to bind Gpa1p and thereby escapes negative regulation, resulting in a constitutive signal for cell cycle arrest (9). This mutation is suppressed by the overexpression of GPA1 (34) or by the expression of the MAT␣ locus when an a strain is used (8). Strain DBC was used to test for suppression of STE4 Hpl . Although both MAT␣2 and MCM1 constructs allowed for plasmid exchange, neither SYG1-1 nor SYG1 overexpression did (Fig. 10), indicating that STE4 Hpl is epistatic to SYG1-1 overexpression. The inability of SYG1-1 to block the effects of STE4 Hpl is likely due to the activation mutation itself (Gly to Asp at residue 124). This alteration appears to affect directly the interaction of Ste4p with Gpa1p (9) and might also prevent protein-protein interactions critical for SYG1-1-mediated suppression. The STE4 Hpl result, in conjunction with the overexpressed STE4 data, suggested that Syg1-1p suppresses gpa1 by interacting with Ste4p.
Suppressing Forms of Syg1p Can Physically Interact with Ste4p-The ability of Syg1-1p to bind Ste4p was examined using the two-hybrid reporter system (35). The LexA-coding sequence was fused to the carboxyl terminus of Syg1⌬400p, and the resulting hybrid construct was shown to retain gpa1 suppression activity (amino-terminal fusion constructs lost this function). Fig. 11 shows that hybrid fusions of Syg1⌬400p or Gpa1p can interact with Ste4p to induce expression of a HIS3 reporter gene. The same constructs also induced lacZ expression as judged by a filter lift ␤-galactosidase assay (data not shown).
The interaction between Syg1⌬400p and Ste4p was quantitatively analyzed (Table III). This combination gave rise to signals that were consistently 10-fold above all combinations of vectors and non-interacting controls. However, signals for Gpa1p with Ste4p, which were similar to what has been reported (18), were significantly higher than those for Syg1⌬400p with Ste4p. This presumably results in part from Syg1⌬400p-LexA, like Syg1⌬400p, having a strong propensity for membrane attachment and thereby being resistant to nuclear localization. It may also reflect an inherently lower binding affinity. Also, although Syg1⌬400p does not appear to interact with Ste18p (Fig. 11, Table III), this or some other protein may strengthen Ste4p/Syg1⌬400p binding. DISCUSSION We have shown that the gpa1 mutation, which leads to constitutive G␤␥ signaling, can be suppressed by overexpression of MCM1, MAT␣2, and SYG1-1. MCM1 and MAT␣2 are both known to encode transcription factors that control the FIG. 8. Halo assays. The strains used were SP1 (wt), GPY74 -15Ca (sst1), YDM400 (sst2), SP1-SN (syg1), GU1 (gpa1 with pLGC) and GU1S (gpa1 with pSYG1-1). pSYG1-1 and pSYG1 designate that expression constructs pUV1-SYG1-1 and pUV1-SYG1, respectively, were transformed into these strains. 20 g of ␣-factor was used for all halo assays except sst1 and sst2 cells, for which 0.2 g was used (see "Materials and Methods" for details).
FIG. 9. Suppression of STE4 overexpression. FY251 cells carrying the galactose inducible STE4 plasmid (YCpGAL-STE4) were transformed with plasmids expressing the indicated genes (vector ϭ pUV2, GPA1 ϭ pUG, SYG1-1 ϭ pUV1-SYG1-1, SYG1 ϭ pUV1-SYG1). Four independent colonies were patched onto sucrose-containing media and replica plated to sucrose or galactose-containing media. expression of haploid specific genes such as GPA1, STE4, and STE18. Their ability to suppress STE4 Hpl is consistent with their mode of action being at the level of altered gene expression. In the case of MAT␣2 expression in a cells, gpa1 suppression has been described previously (36) and can be explained by the action of Mata1p/Mat␣2p heterodimers which are known to repress transcription of haploid-specific genes (1). For MAT␣2 expression in ␣ cells and for MCM1 expression in a cells, however, the basis for gpa1 suppression is not entirely clear. Both MCM1 and MAT␣2 have the capacity to work as heterodimeric partners and can act through interactions with other transcription regulating proteins (37). Therefore, increased amounts of these proteins may result in a stoichiometric imbalance resulting in blocked expression of signaling components or induced expression of factors that suppress the mating pathway.
Other cases of high copy or dominant suppressors of gpa1 have been reported, and all seem to function downstream of G␤␥. The MSG5 gene encodes a protein phosphatase that, when overexpressed, is apparently able to counteract some protein phosphorylation events that propagate the mating signal (38). Expression of a mutant a-factor receptor (STE3 DAF2-2 ) in a cells also suppresses the gpa1 mutation (39). The gpa1 suppressing action of STE3 DAF2-2 is carried out downstream of G␤␥ and results in altered FUS1 expression (36). Other, uncharacterized gpa1 suppressors have also been reported (38,40).
We have described the isolation and characterization of SYG1-1 (a truncated form of the SYG1 gene) that is a novel high copy suppressor of gpa1. SYG1 itself does not have expression characteristics common to many genes directly involved in the mating pathway; its transcription is not induced by pheromone nor is it restricted to haploids. Although SYG1 may not normally be involved in the pheromone response pathway, we present data consistent with its participation in G protein-mediated signaling.
The ability of SYG1-1 and SYG1⌬400 to suppress the gpa1 mutation and overexpressed STE4 is most easily explained by their gene products' binding and sequestering Ste4p (G␤). This interaction is at least partially dependent on residue 124 of Ste4p since the dominant STE4 Hpl mutant allele is not suppressible. Interestingly, this residue is also involved either directly or indirectly in Gpa1p interactions (9). Although fulllength Syg1p probably has Ste4p binding capability (high copy SYG1 is a weak suppressor of gpa1), the removal of putative transmembrane domains from Syg1p is required for its strong suppression of the pheromone pathway.
Despite lacking predicted transmembrane domains and a signal sequence, Syg1⌬400p localizes to the same plasma membrane-enriched subcellular fraction as does Syg1p. That Syg1⌬400p can not be released by a panel of reagents that disrupt protein-protein interactions suggested that it is tightly associated with the plasma membrane. Furthermore, both Syg1⌬400p and Syg1p were insoluble in the nonionic detergent Triton X-100, which disrupts hydrophobic but neither polar, protein-protein, nor protein-lipid interactions (41), thereby suggesting that the NH 2 -terminal portion of Syg1p is sufficient to confer this characteristic upon Syg1p. The overall solubility profiles for Syg1⌬400p and Syg1p may be indicative of proteinprotein interactions involving the cytoskeleton or membrane skeleton (42). Any interacting partner would, of course, need to be sufficiently abundant to accommodate the high levels of overexpressed Syg1⌬400p and Syg1p. Alternatively, the Triton X-100 insolubility of Syg1⌬400p and Syg1p may be an inherent characteristic, perhaps occurring by self-aggregation or by complexing with other insoluble material. Further biochemical studies should reveal the basis of Syg1⌬400p association with the plasma membrane and may help determine both the normal function of Syg1p and how G␤␥ relates to that function.
b Values are in modified Miller units (24) and represent the mean derived from three independent transformants except in the case of Gpa1p/Ste4p (two transformants only). For the Gpa1p/Ste4p combination, some transformants gave significantly lower values. All of these results were confirmed in a separate experiment (data not shown). downstream effectors including some forms of adenylyl cyclase (43), ion channels (44,45), phosducin (46), PI3 kinase (47), and ␤-adrenergic receptor kinase (48,49). As in mammalian cells, there may be multiple effectors responding to multiple G␤␥ heterodimers in yeast. Data from studies of Gpa2p, a Gpa1related G␣ protein known to activate adenylyl cyclase in yeast (50), support the existence of other G␤␥ complexes. GPA2 is expressed not only in haploid cells but also in diploids which are devoid of Ste4p and Ste18p. In addition, overexpression of GPA2 does not suppress gpa1 (51). These findings suggest that another G␤␥ may indeed exist in yeast, a premise which should be settled by the S. cerevisiae genome sequencing project. We are currently using synthetic lethal and yeast two-hybrid screens to elucidate the normal function of Syg1p, to identify functionally redundant gene products, and to test whether interactions with Ste4p, or any other G␤ or G␤␥ proteins, play a role in Syg1p function.
We have demonstrated that in otherwise wild type cells, high level expression of SYG1-1 stimulates recovery from pheromone, as judged by halo assay. Many factors required for adaptation and resumption of vegetative growth have been identified. These include pheromone receptors (21,52), Sst2p (21,31,32), and Ste4p (53). In addition, mutant forms of Ste4p that are impaired for transducing the mating signal still retain their ability to stimulate adaptation (54). The binding of Syg1-1p may promote pheromone recovery not only by sequestering Ste4p but also by accentuating its adaptation function. Alternatively, high levels of Syg1-1p may directly activate this, or an as yet unidentified, adaptation pathway.
Furthermore, as with GPA1 overexpression, SYG1-1 overexpression appears to dampen but not block the mating signal cascade in wild type cells (pheromone recovery and mating results). There are other examples of potent constitutive signaling mutation suppressors which do not block normal signaling in wild type cells. For example, high copy suppressors of activated mutant RAS2 V19 do not significantly disrupt signaling by RAS2 in wild type yeast cells (55,56) and, yet, have proven directly relavent to normal Ras signaling (57).
One striking outcome of the SYG1 deletion analysis was the ability of SYG1⌬340 to suppress the growth arrest phenotype of gpa1 cells, but leave them in a morphologically altered state resembling that induced by pheromone. These cells exhibited the general features associated with shmoos, including enlargement and elongation, but they were budded and continued to undergo mitosis. They showed some of the morphological characteristics seen in fus3 or far1 cells that have been treated with pheromone (58,59). This phenotype may have resulted because the truncated protein binds Ste4p less efficiently and therefore suppresses less well, falling below the threshold for blocking differentiation. It is also possible that Syg1⌬340p has the capacity to promote cell division in an otherwise differentiated cell. These SYG1-1 and SYG1⌬340 suppressed gpa1 cells should be useful tools for studying the components of the differentiation pathway and G protein signaling and may relate to differentiation processes in higher eucaryotes.
The ability of Syg1-1p to suppress constitutive pheromone signaling, stimulate adaptation in wild type cells, and physically interact with Ste4p suggests that Syg1p may normally respond to or transduce a signal via G␤ or G␤␥. Indeed, our observations are of particular note because the identity of possible G␤␥ effectors has remained elusive and because this is the first demonstration of an interaction between Ste4p (G␤) and a yeast protein which is not a G protein subunit. The tight localization of Syg1p to a membrane fraction involves additional features beyond the eight identified transmembrane sequences, and this localization is consistent with a role in G protein interactions. In addition, the identification of genes structurally related to SYG1 suggests the existence of a family of genes with shared functional characteristics. Finally, truncation mutants of SYG1 should continue to prove useful in the analysis of signals leading to cell cycle arrest and differentiation and may also help reveal other G protein-mediated pathways.