Interaction between transmembrane domains five and six of the α-factor receptor

The α-factor pheromone receptor (STE2) activates a G protein signal pathway that induces conjugation of the yeast Saccharomyces cerevisiae. Previous studies implicated the third intracellular loop of this receptor in G protein activation. Therefore, the roles of transmembrane domains five and six (TMD5 and -6) that bracket the third intracellular loop were analyzed by scanning mutagenesis in which each residue was substituted with cysteine. Out of 42 mutants examined, four constitutive mutants and two strong loss-of-function mutants were identified. Double mutants combining Cys substitutions in TMD5 and TMD6 gave a broader range of phenotypes. Interestingly, a V223C mutation in TMD5 caused constitutive activity when combined with the L247C, L248C, or S251C mutations in TMD6. Also, the L226C mutation in TMD5 caused constitutive activity when combined with either the M250C or S251C mutations in TMD6. The residues affected by these mutations are predicted to fall on one side of their respective helices, suggesting that they may interact. In support of this, cysteines substituted at position 223 in TMD5 and position 247 in TMD6 formed a disulfide bond, providing the first direct evidence of an interaction between these transmembrane domains in the α-factor receptor. Altogether, these results identify an important region of interaction between conserved hydrophobic regions at the base of TMD5 and TMD6 that is required for the proper regulation of receptor signaling.

The ␣-factor pheromone receptor (STE2) activates a G protein signal pathway that induces conjugation of the yeast Saccharomyces cerevisiae. Previous studies implicated the third intracellular loop of this receptor in G protein activation. Therefore, the roles of transmembrane domains five and six (TMD5 and -6) that bracket the third intracellular loop were analyzed by scanning mutagenesis in which each residue was substituted with cysteine. Out of 42 mutants examined, four constitutive mutants and two strong loss-of-function mutants were identified. Double mutants combining Cys substitutions in TMD5 and TMD6 gave a broader range of phenotypes. Interestingly, a V223C mutation in TMD5 caused constitutive activity when combined with the L247C, L248C, or S251C mutations in TMD6. Also, the L226C mutation in TMD5 caused constitutive activity when combined with either the M250C or S251C mutations in TMD6. The residues affected by these mutations are predicted to fall on one side of their respective helices, suggesting that they may interact. In support of this, cysteines substituted at position 223 in TMD5 and position 247 in TMD6 formed a disulfide bond, providing the first direct evidence of an interaction between these transmembrane domains in the ␣-factor receptor. Altogether, these results identify an important region of interaction between conserved hydrophobic regions at the base of TMD5 and TMD6 that is required for the proper regulation of receptor signaling.
The ␣-factor receptor (STE2) binds a peptide-mating pheromone ligand and transduces a signal that promotes the conjugation of the yeast Saccharomyces cerevisiae (1). Like other members of the G protein-coupled receptor (GPCR) 1 family, the ␣-factor receptor functions by stimulating the ␣ subunit of a heterotrimeric guanine-nucleotide binding protein to bind GTP (2). The binding of GTP to the G␣ subunit promotes its release from the G␤␥ subunits. Either the ␣ or the ␤␥ subunits then go on to stimulate the next step in the signal pathway. During pheromone signaling, the ␤␥ subunits activate a mitogen-activated protein kinase cascade that leads to transcriptional activation of pheromone-responsive genes and to cell division arrest in G 1 (3)(4)(5). The ␤␥ subunits also lead to activation of Cdc42p, which promotes polarized growth to form the conjugation bridge that connects the mating cells (6,7).
The organization of the functional domains in the ␣-factor receptor is similar to that of many other GPCRs. Receptors in this family are composed of seven transmembrane domains (TMDs) that are connected by intracellular and extracellular loops. The core region of the receptor containing the seven TMDs has been found to be generally responsible for ligand binding and G protein activation (8 -10). Mutational analysis of the ␣-factor receptor indicates that residues near the extracellular ends of the TMDs are involved in ligand binding (11). These residues also appear to play an important role in promoting the activated receptor conformation upon ligand binding. Interestingly, mutations affecting this domain confer a dominant-negative phenotype, apparently because the mutant receptors sequester G proteins and thereby interfere with the ability of wild-type receptors to signal (11,12). Analysis of the intracellular domains of the ␣-factor receptor has demonstrated that the third intracellular loop plays a key role in G protein activation (13)(14)(15)(16). Consistent with this, the third loop becomes hypersensitive to proteolytic digestion with trypsin in response to ligand binding, indicating that this region undergoes a conformational change that is likely to be important for G protein activation (17). The cytoplasmic C-terminal tail is not required for signaling, but is instead a target for phosphorylation that promotes receptor desensitization (18) and downregulation of receptors by endocytosis (14,19).
The TMDs are thought to play an important role in GPCR activation by propagating a conformational change from the ligand-binding domain to the intracellular domains that promote G protein activation. In view of the important role of the third intracellular loop in this process, it seems likely that TMD5 and TMD6 will also play an important role in signaling, because they bracket the third intracellular loop. Interaction between TMD5 and TMD6 may play an important role in signaling, because biophysical studies (20) indicate that TMD5 and TMD6 are in close proximity to one another in rhodopsin and modeling studies predict that they are adjacent in other members of the GPCR family (21). TMD6 of the ␣-factor receptor has been implicated as playing a special role in signaling, because mutations affecting Gln 253 , Ser 254 , and Pro 258 in this domain cause constitutive receptor activity, indicating that these residues function to restrain the receptor in the inactive conformation (22,23). Therefore, the role of the residues in TMD5 and TMD6 was investigated in this study by a scanning mutagenesis approach in which each residue was substituted with Cys. The advantage of substituting with Cys is that it provides opportunities to take advantage of the chemical reactivity of the sulfur in the Cys side chain to carry out chemical cross-linking studies with the mutant proteins (24,25). The results of this study identify a site of interaction between conserved hydrophobic regions at the base of TMD5 and TMD6 that is important for the proper regulation of ␣-factor receptor signaling.
Cysteine Scanning Mutagenesis-Plasmid pPD225 (YEp-URA3-STE2-3XHA) was constructed to facilitate cysteine scanning mutagenesis of TMDs 5 and 6. The plasmid is based on pLG59 (11), which carries a modified STE2 gene in which a triple hemagglutinin epitope tag was introduced at the C-terminal end of the receptor coding sequences. Site-directed mutagenesis was carried out by PCR amplification using mutagenic oligonucleotides and either the QUICK CHANGE mutagenesis kit (Stratagene) or PFU DNA polymerase (Promega). To prevent interference from the two endogenous cysteine codons, Cys 59 was changed to Ile, the corresponding amino acid in the homologous receptor from S. kluyveri (28) and Cys 252 was changed to Ala. These substitutions did not affect the function of the receptor. Silent mutations were introduced by PCR to add unique restriction sites for BglII, NheI, and KpnI to facilitate subcloning of mutagenized DNA fragments. Relative to the ATG initiation codon, the NheI site was introduced at position 616, the BglII at position 695, and the KpnI site was added at position 814. The resulting STE2 gene was fully functional as judged by its ability to complement a ste2⌬ mutation in the functional assays described below and to produce the expected protein product by Western blot analysis. The modified STE2 gene in pPD225 was then used as a starting vector to create a set of Cys substitution mutants in which the codons for each amino acid in TMDs 5 and 6 was changed to TGT to encode Cys. The various TMD5/TMD6 double mutants were created by subcloning the 78-base pair NheI-BglII fragment encoding one of the TMD5 Cys mutations into the NheI-BglII site of a plasmid carrying the appropriate Cys substitution in TMD6. Plasmid pPD225-T7 and its derivatives were constructed by using PCR to add the coding sequence for the T7 epitope (MASMTGGQQMG) in-frame just prior to the trypsin site at Lys 304 of Ste2p. All mutations created for this study were confirmed by dideoxy sequencing of the double-stranded DNA using a Sequenase kit from United States Biochemical.
␣-Factor Receptor Analysis-Immunoblot analysis of Ste2p was carried out essentially as described previously (8). 2.5 ϫ 10 8 mid-logarithmic phase cells were harvested and lysed by agitation with glass beads in 250 l of lysis buffer (2% SDS, 100 mM Tris, pH 7.5, 8 M urea). 100 g of protein extract, as determined by using the BCA Protein Assay kit (Pierce), was separated by electrophoresis on a 10% SDS-polyacrylamide gel, electrophoretically transferred to nitrocellulose, and then probed with rabbit anti-Ste2p antibodies (8), anti-hemagglutinin antibody 12CA5 (Roche Molecular Biochemicals), or anti-T7 antibodies (Novagen, Inc.) as indicated. Immunoreactive proteins were detected by enhanced chemiluminescence using an ECL kit (Amersham Pharmacia Biotech). To assay disulfide bond formation, cells were grown to log phase and then 1 ϫ 10 8 cells were harvested by centrifugation and lysed by agitation with glass beads in a lysis buffer containing 10% glycerol, 100 mM NaCl, 20 mM Tris, pH 7.5, 1 mM EDTA, 10 g/ml phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, 2 g/ml leupeptin, 100 g/ml TPCK, and 50 g/ml benzamidine. The lysate was cleared by centrifugation at 400 ϫ g for 5 min, and then membranes were harvested by centrifugation at 100,000 ϫ g for 45 min. The membrane pellet was washed and then resuspended in 100 l of a buffer containing 100 mM NaCl, 20 mM Tris, pH 7.5, 1 mM EDTA, 100 g of TPCK. The protein concentration was determined by the Bradford assay (Bio-Rad), and then aliquots containing either 50 or 100 g of membrane protein were then digested with TPCK-treated trypsin (Worthington Biochemicals Inc.) at 37°C for 1 h. The TPCK-treated trypsin was freshly dissolved in sterile water at a concentration of 10 mg/ml prior to each experiment and then added to membranes at the indicated final concentration. Reactions were stopped by addition of N-ethylmaleimide to 10 mM and 2ϫ SDS-gel sample buffer. Where indicated, samples were reduced by addition of dithiothreitol to a final concentration of 200 mM. The gel samples were incubated at room temperature, heated at 37°C for 15 min, and then separated on 15% SDS-polyacrylamide gel and electrophoretically transferred to 0.05-m nitrocellulose. The blots were probed with anti-T7 antibodies at 200 ng/ml (Novagen Inc.), and then the immunoreactive bands were detected using a Super Signal Ultra ECL kit (Pierce).
␣-Factor-induced Responses-Mating assays were carried out with yeast strain yLG123 (ste2⌬) containing Cys substitution alleles of STE2 on plasmid pPD225. Cells were replica plated to YPD plates containing a lawn of MAT␣ cells (lys1␣), incubated at 30°C for 12 h, replica plated to minimal plates, and then incubated at 30°C for 48 h to select for the growth of diploids. Halo assays for cell division arrest, summarized in Tables I and II, were performed by spreading on solid medium plates 1.5 ϫ 10 5 yLG123 yeast cells carrying either a wild-type or the indicated mutant receptor gene on a plasmid, and then placing sterile filter disks containing 1500, 750, 375, 187.5, or 93.8 pmol of ␣-factor were placed on the lawn of cells. Cells were grown at 30°C for 72 h, and then the diameters of the zones of ␣-factor-induced cell division arrest were measured. Relative sensitivity to ␣-factor-induced cell division arrest was determined by plotting the average halo size from two independent experiments versus the log of the amount of ␣-factor added, which gives a linear relationship. Basal levels of the pheromone-responsive reporter gene FUS1-lacZ were assayed on synthetic medium plates containing 20 g/ml X-gal (5-bromo-4-chloro-3-indoyl ␤-D-galactoside) and buffered to pH 7. To assay ␣-factor induction of FUS1-lacZ, cultures were grown overnight to logarithmic phase in selective medium, diluted to 4 ϫ 10 6 cells/ml, and incubated with the indicated concentrations of synthetic ␣-factor (Bachem) for 2 h. The cells were permeabilized with 0.05% SDS and CHCl 3 , and then ␤-galactosidase assays were performed using the colorimetric substrate O-nitrophenyl-␤-D-galactopyranoside (29). The results represent the average of two to three independent experiments, each done in duplicate. The standard deviation was usually less than 10% and was always less than 15%.

Cysteine Scanning Mutagenesis of TMD5 and TMD6 -
The role of each residue in TMD5 and TMD6 of the ␣-factor receptor (STE2) was analyzed by mutating the corresponding codons to encode Cys. The advantage of substituting with Cys is that the chemical reactivity of the sulfhydryl group in the Cys side chain can later be exploited in cross-linking studies. A potential drawback of Cys is that deleterious effects can occur due to formation of improper disulfide bonds with endogenous cysteines elsewhere in the protein. To circumvent this potential problem, the two endogenous Cys codons at positions 59 and 252 were changed to encode Ile and Ala, respectively, in the STE2 gene on plasmid pPD225 (see "Experimental Procedures"). Mutation of the endogenous Cys residues did not affect receptor activity, as observed previously (30). This version of the receptor lacking Cys residues will be designated as the wild-type for this study. To identify the residues to target for mutagenesis, hydropathy analysis was used to predict that TMD5 includes residues 209 -229 and TMD6 includes residues 246 -266 ( Fig. 1) (31,32). The 42 different Cys substitution mutant versions of the STE2 gene were constructed as derivatives of plasmid pPD225, and then the series of plasmids was introduced into a ste2⌬ strain that lacks the chromosomal STE2 gene for analysis. Western immunoblot analysis demonstrated that all of the Cys substitution mutant strains produced receptor protein (Ste2p) at a level that was equivalent to the wild-type level and appeared to be properly glycosylated (data not shown) Phenotypes of TMD5 Mutants-Mating assays were performed to screen the Cys substitution mutants to determine whether they produced functional cell surface receptors (Table  I). All but one of the TMD5 receptor mutants were able to mate, indicating that most of the mutants retained at least partial ␣-Factor Receptor Transmembrane Domain Organization receptor activity. In contrast, the substitution of Lys 225 with Cys (K225C) caused a strong defect and appeared to be completely defective in mating ( Fig. 2A). The TMD5 mutants were next examined for ability to promote ␣-factor-induced cell division arrest, which is a more stringent test of receptor function that can be used to compare the relative sensitivity to ␣-factor. The cell division arrest response was quantitated by measuring the zone of growth inhibition (halo) surrounding filter disks containing ␣-factor that are added to a lawn of cells on solid medium. This assay also monitors the ability of the mutant strains to maintain the pheromone response for a longer time period (2 days). Most of the Cys substitution mutants produced halos that were equivalent to that of the wild-type (Table I).
The only TMD5 mutant that failed to respond in this assay carried the K225C substitution (Fig. 2B). The K225C substitution also caused a nearly complete defect in activation of the pheromone-responsive FUS1-lacZ reporter gene, a very sensitive assay for pheromone response, indicating that the ste2-K225C mutant is strongly defective in signal transduction (Fig.  2C). In contrast, the A229C mutant showed about 3-fold increased sensitivity to ␣-factor (Table I and Fig. 2). The close proximity of Lys 225 and Ala 229 to each other at the junction between TMD5 and the third cytoplasmic loop suggests that this region plays an important role in signaling.
The TMD5 mutants were next examined for the ability to activate signaling in a ligand-independent manner using methods similar to those used previously to characterize constitutively active ␣-factor receptors mutated in TMD6 (22,23). For this analysis, yeast carrying a Cys substitution mutant receptor plasmid were replica-plated to solid medium containing X-gal, a chromogenic substrate for ␤-galactosidase, that was used to compare the basal activity of the FUS1-lacZ reporter gene. Plates were monitored for the appearance of cells displaying a blue color, indicating elevated basal expression of FUS1-lacZ. From the set of TMD5 mutants, only cells carrying the S214C substitution consistently displayed a darker blue color in the X-gal plates. Quantitative analysis demonstrated that the basal FUS1-lacZ reporter gene activity of the ste2-S214C mutant was increased 2-fold over the wild-type. Altogether, these data demonstrate that, with the exception of K225C, Cys FIG. 1. Residues targeted for mutagenesis in TMD5 and TMD6 of the ␣-factor receptor. A partial sequence of the ␣-factor receptor highlighting the residues that were mutated in this study is shown. Residues enclosed in boxes were initially predicted by hydropathy analysis to correspond to TMD5 and TMD6 and were substituted with Cys residues by site-directed mutagenesis.
a Yeast strain yLG123 carrying the indicated STE2 allele on plasmid pPD225.
b Ability to mate with lys1␣ cells in a patch mating assay. c Relative ␣-factor sensitivity was determined by halo assays for cell division arrest. ϩ indicates essentially wild-type response. Defective indicates no detectable halo after 2 days. Partial response indicates a turbid zone of cell division arrest. d Basal levels of the FUS1-lacZ reporter gene activity were determined by comparing the relative intensity of blue color for colonies on X-gal plates of strain yLG123 carrying the indicated receptor plasmid. To quantify the -fold increase in basal signaling, receptor plasmids were introduced into strain PMY1 and assayed as described under "Experimental Procedures."

FIG. 2. Analysis of mutant receptors.
Yeast strain yLG123 carrying the wild-type receptor plasmid pPD225, or a mutant version carrying the indicated Cys substitution were assayed for ability to respond to ␣-factor in the following assays: mating test (A), halo assay for cell division arrest (B), and induction of a FUS1-lacZ reporter gene (C). Mating tests were carried out by mating patches of MATa cells to a lawn of MAT␣ strain lys1␣. Diploid progeny resulting from mating were detected by growth on selective minimal medium as shown in A. Halo assays were performed by adding 450 or 150 pmol of ␣-factor to a filter disc that was placed on a lawn of the indicated cells. The size of the zone of cell division arrest surrounding the source of ␣-factor was recorded as shown in B. Activation of the FUS1-lacZ reporter gene was determined by assaying ␤-galactosidase activity in cells that were treated with the indicated concentrations of ␣-factor for 2 h as shown in C.
substitutions were very well tolerated in TMD5.
Phenotypes of TMD6 Mutants-All of the receptor mutants carrying Cys substitutions in TMD6 (Table II) were able to promote a high degree of mating with the exception of the Y266C substitution affecting Tyr 266 . STE2-Y266C was identified previously as a dominant-negative mutant of the ␣-factor receptor (11). The Y266C protein was shown to be produced at levels equivalent to the wild-type receptor and to be stable at the cell surface but defective for signaling. The STE2-Y266C mutant was also the only TMD6 mutant that was completely defective for responding to pheromone in the halo assay (Fig.  2B). Four TMD6 mutants (Q253C, S254C, P258C, and S259C) showed significant growth in the zone of division arrest, indicating that these mutants were impaired in maintaining the cell division arrest response (Fig. 2B and Table II). The A265C mutant displayed a lightly turbid zone of growth arrest indicating a weak defect in cell division arrest (Fig. 2B). The TMD6 mutants showed corresponding defects in their ability to induce a FUS1-lacZ reporter gene ( Fig. 2C and data not shown). The Y266C mutant showed the strongest defect in FUS1-lacZ activation: it could only be induced to 75% of the wild-type maximum, and the dose-response curve was shifted about 10-fold to the right.
Analysis of the basal levels of signaling of the TMD6 mutants on X-gal indicator plates identified four substitution mutants in TMD6 with elevated basal signaling (Q253C, S254C, P258C, and S259C). Quantitative analysis of the FUS1-lacZ reporter gene in strain PMY1 showed that Q253C and P258C substitutions caused strong 4.5-and 47-fold elevated basal signaling, respectively. The S254C and S259C substitutions caused only relatively weak constitutive activity (ϳ2-fold). The identification of these mutants was consistent with the previous identification of other mutations affecting these residues that cause constitutive signaling (22,23,33). The failure to find any new mutations in either TMD5 or TMD6 that cause strong constitutive activity further highlights the significance of Gln 253 and Pro 258 in receptor function.
Double Cysteine Substitution Mutants-Because so few of the single Cys substitution mutations resulted in mutant phenotypes, double mutants were examined to determine if they would reveal a greater range of phenotypes. In particular, double mutant analysis was carried out by combining a mutation in TMD5 with one in TMD6 with the goal of identifying combinations causing synergistic defects that may be indicative of an interaction between TMD5 and TMD6. Double mutant combinations were targeted to residues 222-229 in TMD5 and residues 246 -252 in TMD6 (Fig. 3). These residues were selected for analysis, because they were predicted to be near the cytoplasmic ends of the transmembrane segments and close to the third intracellular loop, a key region for G protein activation. At least some of the corresponding residues in TMD5 and TMD6 are likely to be in close proximity, because each of these regions corresponds to about two turns of an ␣-helix. In addition, another advantage of targeting these regions was that the corresponding single mutants, aside from the ste2-K225C mutant, were similar to the wild-type in their ability to respond to ␣-factor.
Most double Cys substitution mutants were capable of responding to pheromone in both short term mating assays and a long term halo assays (Fig. 3). This indicates that most double mutant combinations of two Cys substitutions that are each Defective Ϫ a Yeast strain yLG123 carrying the indicated STE2 allele on plasmid pPD225.
b Ability to mate with lys1␣ cells in a patch mating assay. c Relative ␣-factor sensitivity was determined by halo assays for cell division arrest. ϩ indicates essentially wild-type response. Defective indicates no detectable halo after 2 days. Partial response indicates a turbid zone of cell division arrest. d Basal levels of the FUS1-lacZ reporter gene activity were determined by comparing the relative intensity of blue color for colonies on X-gal plates of strain yLG123 carrying the indicated receptor plasmid. To quantify the -fold increase in basal signaling, receptor plasmids were introduced into strain PMY1 and assayed as described under "Experimental Procedures."

FIG. 3. Double Cys substitution mutant phenotypes.
A, summary of mutant phenotypes. Double mutants were constructed by combining a Cys substitution in TMD5 with one in TMD6 as indicated in the matrix. The substitutions in TMD5 are listed to the left and the substitutions in TMD6 are listed across the top. B, patch mating tests and halo assays (C) for ␣-factor-induced cell division arrest for a wildtype and a selected set of double mutants carrying the V223C substitution. Yeast strain yLG123 carrying the indicated wild-type STE2 or double mutant version of plasmid pPD225 was analyzed for mating, for ␣-factor-induced cell division arrest by halo assay, and for activation of the FUS1-lacZ reporter gene as described under "Experimental Procedures." Constitutive indicates elevated basal expression of the FUS1-lacZ reporter gene. Filled Halo indicates no detectable cell division arrest in response to ␣-factor. Turbid Halo indicates that cells formed a partial zone of growth inhibition in response to ␣-factor but were not able to maintain the zone of inhibition. Sterile indicates that cells did not mate detectably in a patch mating assay. All single mutants show wild-type function except for the ste2-K225C mutant (Fig. 2). Double mutants were also constructed with the S254C mutation but are not shown because the single ste2-S254C mutant displays strong phenotypes on its own (Table I).

␣-Factor Receptor Transmembrane Domain Organization
tolerated individually did not alter receptor function. In contrast, the ste2-V223C-M250C mutant showed no detectable mating and the ste2-L226C-L247C mutant was partially impaired in its ability to mate (Fig. 3B and Table III). These mutants also failed to form clear halos, indicating strong defects in the cell division arrest response. A few of the double mutant combinations that could mate well also showed defects in the halo assay even though the corresponding single mutants were not significantly impaired (L222C/L248C, L222C/ I249C, L222C/M250C, and V224C/L248C). Thus, certain double mutant combinations caused defects in receptor function that were not observed for either single substitution.
The ligand-independent signaling activity of the double mutant strains was examined next to identify those displaying elevated basal levels of signaling as compared with the corresponding single Cys mutants. Two sets of double mutants were identified by this criterion (Fig. 3). Double mutant combinations involving V223C with L247C, L248C, or S251C and the combinations involving L226C with M250C or S251C displayed constitutive activity in the plate assay. Quantitative assays showed that the mutants displayed 1.9-to 3.7-fold increased basal activity (Table III). The constitutive phenotype was not due to specific effects of the cysteine residues, such as disulfide bond formation, because double mutations in which either Val 223 or Leu 226 were changed to Ser also caused similar levels of ligand-independent signaling (not shown).
The constitutive signaling phenotype associated with certain double mutants could be indicative of an important physical interaction between TMD5 and TMD6, because other residues that cause constitutive activity when mutated have been implicated in inter-helix interactions in the ␣-factor receptor (22). To examine this possibility, helical wheel analysis was used to predict the orientation of the residues in their respective ␣-helices. Interestingly, Val 223 and Leu 226 are expected to fall on one side of TMD5, and Leu 247 , Leu 248 , Met 250 , and Ser 251 are predicted to cluster on one side of TMD6 (Fig. 4). Furthermore, all of the residues implicated by the constitutive double mutants are expected to be within a little more than one helical turn from each other at the cytoplasmic ends of TMD5 and TMD6. Thus, it is possible that these residues may interact.
Intramolecular Cross-linking-To determine whether the residues identified by genetic analysis may interact physically, the spatial relationship of the substituted cysteines was examined by testing their ability to form an intramolecular disulfide bond. The initial experiments were patterned after similar studies with other membrane proteins such as the bacterial chemotaxis receptors, lac permease, and rhodopsin in which samples were oxidized by treatment with I 2 or Cu(II)-1,10phenanthroline to promote disulfide bond formation (24,25,34). However, treatment of yeast membranes with these strong oxidants led to a high degree of nonspecific aggregation of the ␣-factor receptor. Total yeast membrane proteins were also aggregated as judged by Coomassie Blue staining of proteins separated by gel electrophoresis. The aggregation was due, at least in part, to nonspecific effects rather than to disulfide bond formation, because a receptor protein lacking Cys residues also aggregated under these conditions. In contrast, experiments carried out in the absence of strong oxidants showed that incubation with ambient oxygen was sufficient to promote disulfide bond formation in some mutant receptors without causing aggregation (see below). Interestingly, in the course of these studies, Oprian et al. (35) reported that disulfide bonds that formed after exposure of rhodopsin to ambient oxygen were more specific and were more likely to indicate a close spatial relationship of the residues involved than were disulfide bonds formed after treatment with strong oxidants. Therefore, all further experiments were carried out with ambient oxygen treatment.
Disulfide bond formation between TMD5 and TMD6 was initially assayed by comparing the partial trypsin digestion products of wild-type and mutant receptors on Western blots probed with anti-Ste2p antibodies. This approach attempted to take advantage of the ability of trypsin to cleave preferentially the third loop between TMD5 and -6 (17), but the complex pattern of the partial trypsin digestion made the results difficult to interpret. Therefore, to specifically detect disulfide bonds involving TMD6 after complete digestions with trypsin, a T7 epitope tag was introduced between residues 303 and 304 at the C-terminal end of the tryptic fragment that includes TMD6 and TMD7 (Fig. 5A). The T7 tag was selected for these FIG. 4. Helical wheel plots of transmembrane helices 5, 6, and 7. The relative position of the amino acids in TMDs 5, 6, and 7 are displayed in helical wheel format as viewed from the cytoplasmic side of the helix bundle. Sites of Cys substitution mutations that caused constitutive activity when combined with V223C are indicated in boxes, and those that caused constitutive activity when combined with L226C are indicated with ovals.
a Yeast strain yLG123 carrying the indicated STE2 allele on plasmid pPD225.
b Ability to mate with lys1␣ cells in a patch mating assay. c Relative ␣-factor sensitivity was determined by halo assays for cell division arrest. ϩ indicates essentially wild-type response. Defective indicates no detectable halo after 2 days. Partial response indicates a turbid zone of cell division arrest. d Basal levels of the FUS1-lacZ reporter gene activity were determined by comparing the relative intensity of blue color for colonies on X-gal plates of strain yLG123 carrying the indicated receptor plasmid. To quantify the -fold increase in basal signaling, receptor plasmids were introduced into strain PMY1 and assayed as described under "Experimental Procedures." e All double mutant combinations tested involving K225C were defective. studies, because it lacks a site for trypsin digestion. The epitope-tagged receptor gene (STE2-T7) fully complemented a ste2⌬ mutation for pheromone-induced responses, including mating, cell division arrest, and FUS1-lacZ induction (Fig. 5B and data not shown). The expected Ste2p-T7 protein was detected by immunoblotting at levels similar to the untagged control receptor (Fig. 5C). The T7 tag thus permitted a strategy where complete digestion with trypsin could be used to assay disulfide bond formation. As diagrammed in Fig. 5A, the T7tagged tryptic peptide should be approximately 10 kDa, whereas cross-linking between TMDs 5 and 6 should generate a larger 13-kDa peptide.
Trypsin digestion of the Ste2-T7 protein devoid of cysteine resulted in the detection of the expected 10-kDa peptide containing TMD6 and TMD7 on a Western blot probed with anti-T7 antibody (Fig. 6A). Interestingly, analysis of the T7tagged versions of the double mutants involving V223C showed that one of the mutant receptors (V223C/L247C) consistently yielded a 13-kDa tryptic peptide as expected for cross-linking (Fig. 6B). This 13-kDa band did not shift down in mobility even when 33-fold more trypsin was used than was required to detect the lower 10-kDa band from the control receptor protein lacking cysteine residues. That the 13-kDa band was due to a disulfide bond was confirmed by showing that it shifted to a 10-kDa band after treatment with the reducing agent dithiothreitol. In six independent experiments, we consistently observed a 50% to Ͼ90% shift of the 13-kDa band to a 10-kDa band. As an additional specificity test, we constructed a V223S/ L247C mutant to examine the effects of mutating Val 223 to Ser instead of Cys. The V223S/L247C protein yielded a 10-kDa band after trypsin digestion similar to the cysteine-less control receptor (Fig. 6C). These data provide, to our knowledge, the first direct demonstration of the physical proximity between any of the TMDs in the ␣-factor receptor and indicate that Val 223 and Leu 247 are likely to interact in the wild-type ␣-factor receptor.

DISCUSSION
Mutant Phenotypes-Scanning mutagenesis of TMD5 and TMD6 of the ␣-factor receptor revealed that only two substitutions out of 42 tested showed strongly impaired signaling (K225C and Y266C). The ability of the transmembrane domains of the ␣-factor receptor to tolerate substitutions with Cys is consistent with what has been reported for other membrane proteins, including the bacterial chemotaxis receptors (24,36), the dopamine receptor (37,38), and lac permease (25,39). Cys is likely to be tolerated in transmembrane domains, because its side chain is small and has a neutral hydrophobicity profile. Nonetheless, the fact that the majority of the Cys substitutions do not interfere with function indicates that few amino acid side chains in TMD5 and TMD6 of the ␣-factor receptor are specifically required for signaling. Although this could suggest that TMD5 and TMD6 are not important for signaling, the observation that double mutations resulted in a much higher frequency of mutant phenotypes indicates otherwise. Interestingly, similar results were obtained for a scanning mutagenesis of the third intracellular loop of the ␣-factor receptor. Few single substitution mutations in the third loop revealed mutant phenotypes, whereas double mutations more readily resulted in mutant phenotypes (13). This suggests that signal transduction through these critical domains is mediated by a series of contacts, rather than the participation of a limited number of essential residues.
FIG. 5. Properties of T7-epitope-tagged receptor. The T7 epitope tag was introduced just prior to the trypsin site at Lys 304 in STE2 to create pPD225-T7. A, diagram of Ste2 protein with the position of the T7 tag indicated by an arrow and the sequence of the T7 epitope shown in the box. Sites of trypsin digestion are indicated with an asterisk. The tryptic fragment containing TMD5 is shown with diagonal hatching, and the tryptic fragment corresponding to TMD6 is shaded gray. B, halo assay for ␣-factor-induced cell division arrest. yLG123 cells carrying plasmids encoding the untagged (pPD225) or the T7 epitope-tagged receptor (pPD225-T7) were spread on solid medium, and then filter disks containing 187.5, 375, or 750 pmol of ␣-factor were applied to induce the observed zones of cell division arrest. C, Western blot analysis of Ste2-T7 proteins produced by ste2⌬ cells (yLG123) carrying the indicated STE2 allele on a plasmid. The blot was probed with anti-T7 monoclonal antibody. The Cys scanning mutagenesis did, however, identify two mutants that indicate that the junction between TMD5 and the third intracellular loop is important for ␣-factor receptor signaling. The K225C mutation caused loss of function, and A229C caused 3-fold hypersensitivity to pheromone. Interestingly, double mutant analysis carried out by combining Cys substitutions in TMD5 and TMD6 identified a neighboring cluster of residues that are also important for receptor function. This cluster of residues was identified, because certain double mutant combinations caused synergistic defects that were not seen in the corresponding single mutants. For example, the V223C/M250C double mutant was strongly defective in signaling and other double mutants that combined V223C in TMD5 with L247C, I248C, or S251C substitutions in TMD6 caused constitutive activity. Constitutive phenotypes were also identified when L226C in TMD5 was combined with the M250C or S251C substitutions in TMD6. The residues mutated in all of these double mutant combinations show an interesting relationship in that they are predicted to fall on one face of their respective helices (Fig. 4). Thus, the genetic interactions displayed by these mutants suggest that the residues involved are important for receptor function and may define regions of TMD5 and TMD6 that interact.
Interaction between TMD5 and TMD6 -The spatial relationship of the residues identified in the double mutants analysis was examined by determining if the residues were close enough to form a disulfide bond. Screening the mutant proteins identified a cross-link between Cys residues substituted for Val 223 and Leu 247 . Interestingly, detection of this disulfide bond did not require the addition of strong oxidizing agents that have been used in most previous Cys cross-linking studies. This is significant because studies on rhodopsin indicated that disulfide bonds promoted by ambient oxygen resulted in cross-link-ing that was more specific and was indicative of a close spatial relationship between the Cys residues (35). Another difference in our approach relative to previous cross-linking studies was that we assayed disulfide bond formation by analyzing the gel mobility of an epitope-tagged tryptic fragment. The ␣-factor receptor could not be analyzed by the previously used protease digestion methods (35,40,41), because it lacks a unique protease site between the Cys residues that can be used to assess whether they are disulfide-linked. The ␣-factor receptor also did not tolerate the introduction of a novel protease site (Factor X) in the third loop 2 to use for this analysis. Thus, introduction of an epitope tag should be applicable to other receptors that, like the ␣-factor receptor, cannot be analyzed by previously used protease digestion approaches. In addition, the epitope tag strategy may also be useful for confirming the results of cross-linking studies that detect disulfide bonds formed between coexpressed halves of split receptors (34,35), because the split receptors may not always efficiently associate to form a native receptor structure.
The detection of a cross-link between Val 223 and Leu 247 provides an important point of reference for the orientation of the residues in TMD5 and TMD6 (Fig. 4). For example, it is interesting that Ser 254 in TMD6 is predicted to face in the general direction of TMD5, because substitution of Ser 254 with residues of large van der Waals radius (i.e. S254F) caused higher constitutive activity than did substitution with smaller residues, suggesting that the constitutive activity is a consequence of altered helix packing (22). This orientation also predicts that Ser 219 in TMD5 faces toward TMD6, consistent with the interpretation that Ser 219 substitution mutants respond to a broader range of agonists because of altered helix packing (42,43). In addition, this model indicates that TMD5 interacts with the side of TMD6 that is opposite from Gln 253 , a residue that is predicted to interact with TMD7 (Fig. 4). Gln 253 is thought to interact with Ser 288 and Ser 292 in TMD7 based on the analysis of constitutively active receptor mutants (22). The overall arrangement of TMD5, -6, and -7 predicted by these studies is significant, because it is consistent with the organization of the helix bundle predicted for members of the rhodopsin family of GPCRs (20,44). Thus, although the ␣-factor receptor does not share significant sequence homology with the large rhodopsin family of GPCRs, there is underlying structural and functional similarity.
The identification of a cross-link between Cys residues substituted for Val 223 and Leu 247 also has interesting implications for models of the membrane topology of TMD5. Based on hydropathy, TMD5 was predicted to encompass residues 209 -229 (31,32). However, for residues 223 and 247 to be in close juxtaposition, the core of TMD5 is now predicted to span residues 203-225 (Fig. 7). This new positioning of TMD5 has two interesting consequences. One is that this alignment places Lys 225 at the cytoplasmic interface, instead of being buried in the membrane. As there is no obvious counterion for Lys 225 in the other TMDs, positioning this residue nearer the polar lipid head groups and cytoplasm is more energetically favorable. The other interesting aspect of this topology is that it places the TMD5 residues affected by dominant-negative mutations near the extracellular end of the transmembrane segment (Fig. 7). The four dominant-negative mutations in TMD5 are now in better juxtaposition with the 10 other sites of dominant-negative mutations found in TMD2, -3, -4, -6, and -7 that are all predicted to map near the extracellular ends of membranespanning regions (11,12). Thus, shifting the predicted topology of TMD5 also helps to better define the domain affected by the 2 P. Dube, A. DeCostanzo, and J. B. Konopka, unpublished data. . The central amino acids of the 431-residue long ␣-factor receptor are indicated by the one-letter code. The black line between TMD5 and TMD6 highlights the positions of Val 223 and Leu 247 that are predicted by the results of this study to be in close proximity. A set of lines between Gln 253 in TMD6 and Ser 288 and Ser 292 in TMD7 mark the predicted intramolecular contact that was identified by analysis of constitutive mutants (22). Black circles indicate the residues affected by dominant-negative mutations (11,12). The open circle indicates the position of Lys 225 , and the triangle indicates the position of Ala 229 ; these are sites of Cys substitutions that caused loss of function and supersensitivity, respectively. dominant-negative mutations that plays an important role in ligand binding and receptor activation. Altogether, these results significantly improve the two-dimensional models of receptor structure and provide an important frame of reference for the development of a three-dimensional model of the ␣-factor receptor.
Implications for Signal Transduction-The identification of an interaction between a cluster of nonpolar residues at the base of TMD5 and -6 in the ␣-factor receptor has interesting implications for the function of other GPCRs. This may be a conserved feature of GPCRs, because disruption of intramolecular contacts between the base of TMD5 and TMD6 is thought to constitutively activate the Lutropin receptor and cause precocious puberty (45). Similarly, the interaction of a hydrophobic cluster of amino acids at the base of TMD5 and TMD6 appears to be important for restraining the thyrotropin-releasing hormone receptor in the inactive state (46). Although, the particular amino acids identified in the thyrotropin-releasing hormone receptor and Lutropin receptor are different than those in the ␣-factor receptor, the domains of TMD5 and TMD6 identified in these studies may function in an analogous way. In this regard, it is interesting that the corresponding regions of TMD5 and TMD6 of rhodopsin are closely packed (20). Collectively, these data indicate that, despite a lack of sequence similarity among the GPCR family, there is a remarkable similarity in the structure and function of these receptors.