Identification of Destabilizing and Stabilizing Mutations of Ste2p, a G Protein Coupled Receptor in Saccharomyces cerevisiae

Abstract

The isolation of mutations affecting the stabilities of transmembrane proteins is useful for enhancing the suitability of proteins for structural characterization and for identification of determinants of membrane protein stability. We have pursued a strategy for identification of stabilized variants of the yeast α-factor receptor Ste2p. Because it was not possible to screen directly for mutations providing thermal stabilization, we first isolated a battery of destabilized temperature sensitive variants, based on loss of signaling function and decreased binding of fluorescent ligand, then screened for intragenic second-site suppressors of these phenotypes. The initial screens recovered singly and multiply substituted mutations conferring temperature sensitivity throughout the predicted transmembrane helices of the receptor. All of the singly-substituted variants exhibit decreases in cell-surface expression. We then screened randomly mutagenized libraries of clones expressing temperature sensitive variants for second-site suppressors that restore elevated levels of binding sites for fluorescent ligand. To determine whether any of these were global suppressors, and thus likely stabilizing mutations, they were combined with different temperature sensitive mutations. Eight of the suppressors exhibited the ability to reverse the defect in ligand binding of multiple temperature sensitive mutations. Combining certain of the suppressors into a single allele resulted in greater levels of suppression than was seen with either suppressor alone. Solubilized receptors containing suppressor mutations in the absence of temperature sensitive mutations exhibit a reduced tendency to aggregate during immobilization on an affinity matrix. Several of the suppressors also exhibit allele-specific behavior indicative of specific intramolecular interactions in the receptor.

Transmembrane proteins (TMPs) are encoded by 20-30% of the open reading frames in most genomes ¹, carry out critical physiological roles in signaling, transmembrane transport, and catalysis, and are the targets of a large fraction of clinically useful drugs ². However, determination of the structures of TMPs, particularly of eukaryotic TMPs, remains challenging and such structures are grossly under-represented in the protein structure database ³. Reasons for the difficulty of structure determination for TMPs include low levels of expression (both in native and heterologous expression systems), the structure-perturbing effects of detergents used for the extraction and purification of TMPs, the need to establish intermolecular contacts involving both hydrophobic and hydrophilic surfaces, and the instability and flexibility of TMPs, which can hinder packing into well-ordered crystals (see ⁴.

The growing number of examples of successful structure determination of TMPs can be attributed, in part, to the implementation of strategies for stabilization of TMP folding. Strategies for stabilization also appear to be of particular importance in protecting the native conformations of TMPs against the denaturing effects of detergents used in their solubilization and purification ^{5, 6}. However, current understanding of the underlying basis of folding of transmembrane proteins is not sufficient provide a reliable basis for rational design of amino acid substitutions that will provide TMP stabilization. Successful strategies for TMP stabilization have included the use of proteins derived from thermophilic organisms ⁷, the use of TMP variants that contain an insertion of a stable soluble protein in place of a flexible loop or fused to the N-terminal of the protein^8-38, the use of variant forms of TMPs containing point mutations that confer stability or conformational homogeneity ^{24, 39-45}, and co-crystallization with anti-TMP antibodies ^{16, 17, 46, 47}. These strategies, while providing important advances for structure determination, have been sub-optimal in several respects: 1) Development of appropriate antibodies and strategies for co-purification and co-crystallization is time-consuming and antibody binding can, in some cases, perturb protein structures. 2) Thermophilic orthologs of membrane proteins are not always available and may be difficult to express in common expression systems. 3) Fusion and truncation of membrane proteins can lead to substantial loss or alteration of the protein’s function ⁹. 4) Generation and assay of large numbers of individual site-directed mutations is time consuming and allows sampling of only a limited subset of sequence space.

We report here the development of a protocol for the identification of stabilizing mutations in TMPs. The protocol is based on the use of fluorescence activated cell sorting (FACS) to screen a library of randomly mutated TMPs expressed in yeast cells in order to identify variants with enhanced thermal stabilities. This approach is distinct from previous protocols for stabilization based on one-at-a-time testing of large numbers of site-directed substitutions ^{24, 40-44} in that it provides more direct and complete screening of the full range of possible amino acid substitutions. However, related random screening approaches using bacterial expression systems have been successfully applied to the development of thermostabilized variants of a soluble protein ⁴⁸ as well as several different examples of G protein Coupled Receptors (GPCRs) ^{45, 49-54}.

Among TMPs, G Protein Coupled Receptors (GPCRs) are an important class of receptors involved in the transduction of diverse extracellular signals. The first structure of a GPCR was obtained by crystallography of the native protein (rhodopsin) derived from native tissue ⁵⁵, but all subsequent crystal structures of other GPCRs have been acquired using heterologously expressed protein that has been modified to enhance stability and homogeneity. In many cases, this has resulted in loss ⁹ or modification of the signaling activity ^{5, 56}. As a test system for implementation of the mutagenesis and screening approach, we have focused on the pheromone receptor Ste2p, a GPCR that is normally expressed in Mata haploid cells of the yeast Saccharomyces cerevisiae. Ste2p is activated by binding of α-factor, a 13-residue peptide secreted by cells of the opposite Matα mating type, resulting in G protein-dependent signaling via a MAP kinase cascade with downstream effects that include the transcriptional activation of a number of genes, as well as the arrest of the cell cycle ⁵⁷. Although Ste2p exhibits very little sequence similarity to mammalian receptors, it activates a trimeric G protein that is closely related to trimeric mammalian G proteins. Furthermore, some mammalian GPCRs can activate the yeast mating pathway, indicating that yeast and mammalian GPCRs share similar mechanisms of action ^58-60. Purification of the Ste2p receptor has been reported by several groups ^61-63. Determination of the structure of Ste2p could allow identification of conserved structural elements that underlie signaling function across the range of widely diverged GPCR sequences. Moreover, techniques developed to identify stabilized Ste2p may be applied to mammalian GPCRs and other TMPs.

Initial experiments examining the stability of Ste2p indicated that the native protein, once it is incorporated into the plasma membrane of cells, is stable to temperatures well above those at which yeast can survive. Furthermore, we, and others ⁶⁴, have found that high temperature treatment induces a transition in intact yeast cells that results in high levels of nonspecific binding of ligand that does not involve Ste2p receptors. These observations preclude the use of assays of ligand binding to yeast cells for directly screening for receptors variants exhibiting enhanced thermal stability. Consequently, we have pursued an approach that involves initial identification of variant receptors with decreased thermal stability (temperature sensitive receptors) followed by isolation of intragenic second-site suppressor mutations that suppress the initial temperature sensitive alleles. Our goal was to identify “global suppressors,” that provide overall stabilization of the protein rather than specifically interacting with particular destabilizing mutations in the original temperature sensitive alleles ⁶⁵. Such suppressor mutations are expected to provide enhanced stability when transferred from the temperature sensitive background into otherwise wild-type receptors. Since a hallmark of “global suppression” is the ability to suppress diverse starting temperature sensitive alleles, pursuit of this approach requires isolation of a collection of different temperature sensitive starting alleles.

We describe herein the isolation of a collection of temperature sensitive alleles of the gene encoding the yeast pheromone receptor Ste2p and the identification and characterization of intragenic second-site suppressors of these initial destabilizing mutations. Analysis of an array of combinations of temperature sensitive and suppressor mutations confirmed the global nature of some of the suppressors, but also provided evidence for certain allele-specific interactions between starting and suppressing mutations. When transferred to an otherwise normal receptor, some of the suppressor mutations enhanced the ability of the detergent-solubilized Ste2p receptor to survive initial purification.

MATERIALS AND METHODS

Strains and Plasmids

The strains and plasmids used in this study are listed in Tables S1-S5. All yeast strains used in this study were generated by transforming plasmids into the host strain A232 (MATa ste2-Δ bar1⁻ cry1^R ade2-1 his4-580 lys2_oc tyr1_oc SUP4-3^ts leu2 ura3 FUS1::p[FUS1-lacZ TRP1]), originally derived from strain DJ213-7-3⁶⁷. Since strain A232 lacks the normal chromosomal copy of STE2, plasmid borne alleles provide the only Ste2p in the cells.

Plasmids encoding full-length STE2 were derived from plasmid pMD1483, which was constructed by ligating a SacI-SphI fragment of pMD240 ⁶⁶ containing STE2 fused to a C-terminal triple HA epitope into a similarly digested pMD1383 ⁶⁸. The plasmids encoding C-terminally truncated STE2 were derived from plasmid pMD1422 ⁶⁹. This is a multicopy plasmid encoding Ste2p truncated after amino acid 303, which is C-terminal to the end of the seventh transmembrane helix. A triple repeat of the c-myc epitope tag is fused to the C-terminal end of this truncated Ste2p. In some cases, temperature sensitive mutations identified in the genetic screens were transferred from plasmids encoding full-length Ste2p into plasmid encoding truncated Ste2p by digesting with SphI and KpnI and ligating the STE2-containing fragment into SphI-KpnI-cut pMD1422.

Mutagenesis

Site-directed mutagenesis was performed essentially as described previously ^{70, 71}. Oligonucleotides used for site-directed mutagenesis and PCR are listed in Table S6 in the Supplemental Information. For the construction of double mutants, two oligonucleotides, each encoding a separate mutation, were used in a single reaction, with the exception of cases where the two oligonucleotides would overlap, in which case a longer single oligonucleotide encoding both mutations was used.

Mutagenized libraries of full-length receptor were generated by error-prone PCR as described previously ^{71, 72}. PCR reactions were performed with skewed dNTP ratios (0.2 mM dATP and dGTP vs. 1mM dCTP and dTTP) and elevated Mg²⁺ and Mn²⁺ concentrations (7 mM and 0.5 mM, respectively) to introduce mutations into the PCR product. The 30 cycles of PCR were conducted with Taq DNA polymerase, 200 pM template, and 300 nM each of the primers ON915 and ON 918. The region amplified by error-prone PCR spanned residues 4-304 of Ste2p, covering the entire membrane-spanning region. The mutagenized PCR product was co-transformed into yeast with plasmid pMD1483 that had been doubly digested with HpaI and KpnI (which cut STE2 at positions corresponding to amino acids V45 and G273). Homologous recombination during transformation re-circularizes the plasmid with the mutagenized PCR product, resulting in the incorporation of random mutations into the plasmid-encoded STE2.

Mutagenized libraries of truncated receptor were generated by mutagenic PCR using the Agilent Genemorph II random mutagenesis kit with 500 ng of template DNA per reaction. 25 cycles of PCR were performed using the primers ON915 and ON918. The coding region of the C-terminally truncated STE2 gene from residue 4 to residue 303 was targeted for mutagenesis. As performed for construction of libraries of full-length receptors, the PCR product was transformed into yeast along with HpaI-KpnI doubly digested plasmid pMD1422, the ends of which overlapped the mutagenized PCR product by 126 bases on the 5′ side of STE2 and 112 bases on the 3′ side. After transformation into yeast, plasmids were extracted from randomly selected colonies and sequenced to determine the mutation rate, which averaged 3.6 mutations per sequenced clone.

Screen for temperature sensitive STE2 alleles based on loss of pheromone-dependent growth arrest

To identify clones expressing temperature sensitive receptors, we developed a screen for receptors that lose signaling function at high temperatures, preventing α-factor-dependent growth arrest. Libraries of mutagenized plasmids encoding full-length STE2 were transformed into the ste2-Δ yeast strain A232. Freshly transformed cells were replica plated onto one set of plates that contained 300nM α-factor and another set that lacked α-factor. One plate of each type was incubated at 24°C and another was incubated at 34°C. Cells from colonies that grew at 34°C but not at 24°C were collected, streaked to single colonies, and then inoculated into liquid culture. Plasmids were isolated from yeast cells using a Wizard Miniprep Kit (Promega Corp.): Cells were disrupted by vortexing with zirconia beads (Biospec Products Inc.) in the lysis buffer from the kit. The lysate was centrifuged for two minutes at 100 × g to pellet the beads and the supernatant was mixed with the kit’s neutralization solution. All other steps (clearing the lysate, binding to resin, washing the resin, elution of the DNA) followed the manufacturer’s standard protocol.

Growth Arrest Assay

To perform the growth arrest assay, sets of plates were prepared by casting either 25 ml of media in 90mm round plates or 50 ml in 128mm × 86mm rectangular plates. After cooling, α-factor was spread on each plate to the desired concentration and equilibrated for at least 24 hours. The strains to be tested were cultured to an OD₆₀₀ of 1 in synthetic media lacking α-factor and uracil (SD-ura), and then diluted with the same liquid media to an OD₆₀₀ of 0.3 in a 96-well plate. The cultures were then 10-fold serially diluted into neighboring wells. 2 μL from each dilution were spotted onto each plate containing different concentrations of α-factor, allowed to dry, incubated at the appropriate temperatures, and photographed after 2 days of growth.

Flow Cytometry and Cell Sorting

The fluorescent ligand used in this project, [K⁷(NBD),Nle¹²] α-factor, contains the fluorophore NBD (7-nitrobenz-2-oxa-1,3-diazol-4-yl) conjugated to the lysine at the seventh position in α-factor. In addition, the methionine normally present at position 12 of α-factor is replaced by norleucine ⁷³. (Norleucine is isosteric to the native methionine residue that it replaces and does not affect biological activity or binding.) The fluorescent ligand has a measured K_D of ~27 nM compared to ~8 nM for unlabeled α-factor ⁷³. The fluorescent ligand was also shown to be able to trigger growth arrest in exposed cells with an EC₅₀ similar to that of unlabeled α-factor⁷⁴.

Ligand binding assays were performed as previously reported ⁷⁴. The strains to be assayed were grown to an OD₆₀₀ of ~1 in SD-ura medium. Samples consisted of 1.5 ×10⁶ cells and the relevant concentration of ligand in 20 mM sodium acetate pH 4.6 in a final sample volume of 500 μL. Samples were maintained at 0°C during and after addition of ligand to prevent endocytosis of receptors. Cellular fluorescence was analyzed using a BD FACSCanto flow cytometer. The fluorescent signal due to binding was calculated by subtracting the mean autofluorescence of a control without ligand from the mean fluorescence of the sample with ligand. Ligand binding parameters were determined by fitting the fluorescence measurements of samples that had been incubated with varying concentrations of fluorescent α-factor to a single-site binding equation that included a non-specific binding component using the ligand-binding module of the non-linear least squares function of SigmaPlot (Systat Software Inc). The listed binding parameters for variants of Ste2p represent the averages of independent fits of data from the three genetically independent yeast transformants. Due to day-to-day variability in the assay, strains expressing mutant alleles were always compared against strains expressing comparable wild-type alleles assayed in parallel.

For cell sorting, samples were prepared that contained cells cultured at the desired temperature and 300 nM [K⁷(NBD),Nle¹²] α-factor in 20 mM acetate buffer (pH 4.6). The two-stage screen for temperature sensitive alleles was conducted using selection gates defined with reference to the fluorescence channel histogram so as to minimize the number of cells from the unmutagenized control and to maximize the number of events from the randomly mutagenized library (Figure S1 in the Supplemental Information). The selected population was split evenly between two FACS sorting gates, with each sorting gate collecting 15% of the total population, as shown in Figure S1A in the Supplemental Information. The pool of clones that had been selected for highest levels of ligand binding after growth at 24 °C (P4 in Figure S1A) was cultured at 34 °C overnight, and then subjected to a second screen that selected for cells with low levels of ligand binding (the lowest 30% of fluorescence) when cultured at the elevated temperature (P5 in Figure S1B in the Supplemental Information). In the screens for suppressors, selection gates were chosen with reference to two dimensional dot plots of fluorescence channel vs. forward scatter, so as to minimize the number of cells from the unmutagenized control and to maximize the number of events from the randomly mutagenized library (Figure S2). Since large cells are expected to bind more fluorescent ligand than small cells, and forward scatter in flow cytometry is related to size of cells, a diagonal orientation of the selection gate was used to eliminate cells that exhibit high fluorescence because they are large rather than because they express more stable receptors.

Preparation of cell lysates and membrane fractions

Tagged alleles of STE2 were used in all experiments to allow immunoblotting with anti-tag antibodies. Full-length alleles of STE2 consisted of fusions to three copies of the influenza HA epitope tag. C-terminally truncated alleles of STE2 were fused to a triple c-myc epitope tag, as described previously ⁷⁵. For immunoblotting of cell lysates, strains were cultured in SD-ura to an OD₆₀₀ of 1. 5 OD × mls of culture (~1 × 10⁸ cells) were pelleted by centrifugation at 14,000 × g for 2 minutes and resuspended in 20 mM sodium acetate pH 4.6 with 1 mM freshly prepared phenylmethanesulfonyl fluoride (PMSF) (from a 100mM stock in ethanol). The cells were disrupted by adding zirconia beads (Biospec Products Inc.) and conducting ten cycles of 20s of disruption by vortexing performed at 8 °C, interspersed with incubations of 1 minute on ice.

To isolate membrane fractions, the cell lysates were centrifuged for two minutes at 4 °C at 200 × g to pellet the zirconia beads. Another two minute spin at 200 × g further separated unbroken cells and cell wall fragments from the lysate. The supernatant from this spin was then centrifuged at 18,000 × g for 45 minutes at 4 °C to pellet the cell membranes.

Membranes were solubilized by resuspending them in 1% n-dodecyl-β-D-maltopyranoside (DDM) (Affymetrix). The samples were then nutated at room temperature for 2 hours. To remove any unsolubilized fraction, the samples were centrifuged for 30 minutes at 4 °C at 18,000 × g.

For immunoblotting, cell lysates, resuspended membrane fractions, or solubilized membrane were mixed with gel loading buffer (100 mM Tris pH 6.8, 0.1 mM EDTA, 9 M urea, 5% SDS, and 0.02 mg/ml bromophenol blue) then applied to Bio-Rad Criterion 8-15% Tris-HCl gels. Each well was loaded with the lysate prepared from 0.04 OD₆₀₀ × ml of cells. After electrophoresis, the gels were transferred to 0.2 μm nitrocellulose membranes by overnight electroblotting. The membrane was blocked by incubation in 2.5% Carnation powdered milk for two hours, then incubated for two hours with a 1:5000 dilution of anti-HA antibody (Roche Inc) or anti-c-myc antibody (Roche Inc) followed by another two hour incubation with 1:15000 dilution of goat anti-mouse antibody conjugated to horseradish peroxidase. The membrane was then incubated with SuperSignal West Dura Chemiluminescent Substrate (Thermo Scientific), followed by imaging on CL-X Posure film (Thermo Scientific).

Immobilization of solubilized receptors

For immobilization of solubilized receptors, we used 8 μm polystyrene beads that were pre-conjugated to goat anti-mouse antibodies (Spherotech). The beads were washed with phosphate buffered saline pH 7.4 (PBS) and resuspended in fresh PBS. They were then incubated for two hours at room temperature with agitation with mouse anti-myc antibodies (Roche Inc) at a concentration of 20 μg antibody per ml of beads. The beads were then washed with fresh PBS and incubated with solubilized receptor in PBS with 0.1% DDM and 1 mM PMSF with agitation for 2 hours. After protein binding, the beads were pelleted by centrifugation and washed with PBS containing 0.1% DDM and 1 mM PMSF.

RESULTS

We report here the development of an approach for identifying mutations that enhance the thermal stability of the yeast α-factor receptor, Ste2p. The approach is based on the use of an assay for high affinity binding of a cell-impermeant fluorescent ligand to monitor the abundance of correctly folded receptors at the cell surface ⁷⁴. However, this strategy could not be used directly to stabilize wild-type receptors because of the loss of yeast viability and high non-specific ligand binding of ligand to yeast cells observed at temperatures needed to denature native Ste2p. We also explored the use of an alternative approach for assaying receptor stability based on loss of signaling function at elevated temperatures, monitored via transcriptional induction of a pheromone-responsive FUS1-lacZ reporter gene ⁷¹. While loss of signaling was observed over the range 30-40 °C, most of this loss, apparently, does not result from destabilization of receptors because ligand binding by wild-type Ste2p is unaffected by these treatments and because similar temperature-dependence of signaling is observed in strains in which the pheromone response pathway is activated in a ligand-independent fashion by deletion of GPA1 encoding the G protein α-subunit (results not shown). Thus, the screen for stabilizing mutations was conducted by first isolating temperature-sensitive mutations conferring loss of receptor binding and signaling function at 34 °C and then isolating intragenic second-site suppressors that restore high levels of ligand binding to the temperature sensitive alleles.

Isolation of temperature sensitive mutants of STE2 based on loss of signaling function

To isolate temperature sensitive alleles of STE2, we used error-prone PCR to randomly mutagenize a region of the full-length receptor that included the sequences encoding all seven predicted transmembrane segments, then screened yeast transformants for alleles conferring temperature sensitive signaling based on resistance to α-factor-dependent growth arrest at 34 °C but not at 24 °C. From a library containing approximately 10,000 clones, we isolated six temperature sensitive alleles containing single amino acid substitutions, and seven additional alleles containing multiple amino acid substitutions (Figure S3 in the Supplemental Information and Table 1; a tabulation of the phenotypes of individual mutations discussed in this manuscript is presented in Table S7 in the Supplemental Information).

Table 1.

Temperature sensitive alleles identified from the growth arrest and binding-based screens

Mutations	Screen
S95Y	Growth Arrest
Q135H, L164P, S207I	Growth Arrest
I260K	Growth Arrest
G31R, M69K, S213F	Growth Arrest
H94Y	Growth Arrest
I53V, L113R	Growth Arrest
S104Y	Growth Arrest
V45A, V152L, T179I, Q253K	Growth Arrest
G31R, A52T	Growth Arrest
V49D	Growth Arrest
F55L, F116S, N158D	Growth Arrest
Q51P, G174W, G237S	Growth Arrest
L211R	Growth Arrest
F38Y	Binding

To determine whether the missense mutations in the STE2 gene that were recovered in this screen were responsible for the observed phenotypes, the single mutations that elicited growth arrest phenotypes were reconstructed by re-cloning the sequenced regions of the mutated STE2 genes into a fresh plasmid and transforming a fresh yeast host. The growth arrest assay performed on these reconstructed strains, shown in Figure 1, verified that cells expressing receptors containing each of the six tested single amino acid substitutions exhibit temperature-dependent growth phenotypes on α-factor. (In this assay, growth of cells on medium containing α-factor indicates loss of signaling function such that growth at high temperature but not low temperature is characteristic of temperature sensitivity of Ste2p). Some of the single mutations also result in partial defects in growth arrest at 24°C.

Figure 1.

Temperature sensitive receptors with single amino acid substitutions exhibit a temperature dependent growth phenotype. The figure shows the α-factor-dependent growth inhibition of clones that were reconstructed by transforming plasmids bearing missense mutations in either full-length STE2 or C-terminally truncated STE2 into the yeast host strain. Each panel shows the growth of four 10-fold serial dilutions of the relevant yeast strain in the presence of the indicated concentration of α-factor. The clones were scored as Functional (F), Nonfunctional (NF), Temperature Sensitive (TS), or Partially Temperature Sensitive (pTS).

The only previously identified temperature sensitive mutation in STE2 is the A52T substitution, which was reported to result in a temperature dependent loss of responsiveness to α-factor ^{64, 76}. This substitution was recovered as part of a multiply-substituted STE2 allele exhibiting temperature sensitive α-factor-dependent growth arrest. However, neither A52T nor the other mutation recovered in this allele, G31R, result in any detectable defect in growth arrest when tested individually by plating dilutions of culture on plates containing α-factor. This discrepancy may reflect that fact that the previous study of A52T ⁷⁶ assayed growth arrest based on the sizes of inhibitory halos surrounding α-factor aliquots applied to culture plates, in contrast to the dilution assay reported here. The A52T substitution, when present as the only mutation in Ste2p, did result in decreased numbers of cell surface binding sites compared to wild type receptors (with a greater decrease at 34 °C than 24 °C)(Figure S4 in the Supplemental Information).

In addition to the missense mutations conferring temperature sensitivity, we also recovered nine STE2 alleles that displayed temperature sensitive signaling apparently resulting from the presence of premature stop codons at various positions in the coding sequence for the receptor. Temperature sensitive growth of these clones can be explained by the fact that the host strain used for these studies contains the temperature sensitive termination suppressor SUP4-3^ts, which allows read-through of amber stop codons (UAG) at the low, permissive temperature ⁷⁷. This class of mutations was not studied further.

Mutated receptors exhibiting temperature-dependent loss of signaling function are present at reduced abundances in cells

Saturation binding experiments using the fluorescent α-factor analog [K⁷(NBD),Nle¹²] α-factor were used to compare the relative numbers of binding sites present on the external surface of cells expressing full-length normal and mutant receptors (Figure 2 and Table 2). When cultured at high temperature (34 °C), cells expressing most of the identified temperature sensitive receptor variants exhibited very low levels of ligand binding that precluded measurement of the abundance and affinities of binding competent receptors (Figures 2B and 2C and Table 2). Among the mutant full-length receptors isolated from the screen for loss of function, only the variant containing the V49D substitution retained a measurable number of cell surface binding sites at 34 °C. It was present at 40% of the level of similarly expressed full-length wild-type Ste2p and exhibited a 2-fold reduction in binding affinity for [K⁷(NBD),Nle¹²] α-factor compared with wild type full-length receptors at this temperature.

Figure 2.

Full-length temperature-sensitive receptors isolated based on loss of signaling function exhibit reduced numbers of binding sites at the cell surface. Strains expressing the indicated full-length STE2 alleles were grown in liquid cultures at 24 °C (A) or 34 °C (B and C). The error bars represent the standard error based on assay of binding to three independent isolates of each strain. Binding was determined by flow cytometry as described in Materials and Methods. (Parameters derived from fitting are listed in Table 2.)

Table 2.

Ligand Binding Parameters of Temperature Sensitive Receptors.

	Full-length receptors				C-terminally truncated receptors
	34 °C		24 °		34 °C		24 °C
Variant	Relative Bmax (% wild-type)	KD (nM)	Relative Bmax (% wild-type)	KD (nM)	Relative Bmaxa (% wild-type)	KD (nM)	Relative Bmax (% wild-type)	KD (nM)
Wild-type	100 ± 24	28 ± 4.7	100 ± 14	7.3 ± 0.9	1250 ± 240	45 ± 1.7	1250 ± 370	19 ± 3.8
F38Y	− b	− b	− b	− b	330 ± 62	64 ± 7.9	1100 ± 120	59 ± 11
V49D	37 ± 8	49 ± 20	80 ± 29	5.3 ± 2.3	100 ± 20	35 ± 4.0	290 ± 150	23 ± 3.3
H94Y	NDc	ND	30 ± 4.2	5.2 ± 2.1	280 ± 53	90 ± 12	360 ± 100	47 ± 8.1
S95Y	ND	ND	3.2 ± 1.6	2.1 ± 1.5	ND	ND	7.0 ± 3.2	8.1 ± 7.5
S104Y	ND	ND	26 ± 3.0	2.0 ± 0.7	ND	ND	66 ± 22	12 ± 3.4
L211R	ND	ND	10 ± 2.1	0.8 ± 0.1	ND	ND	ND	ND
I260K	ND	ND	ND	ND	ND	ND	ND	ND

^aLevels are relative to full-length wild-type receptor levels using a conversion factor of 12.5 ± 2.3, determined from the ratio of B_max values that were fit to strains expressing either full-length or truncated wild-type receptors (Figure S2 in the Supplemental Information). The conversion factor was only measured at 24 °C and was applied to both temperatures.

^bNot tested

^cND; Not Detectable

Reductions in numbers of cell surface binding were also observed in cells expressing mutant receptors cultured at the permissive temperature of 24 °C (Figure 2A and Table 2). Cells expressing the full-length H94Y and S104Y variants retained only ~30% of the wild type number of sites, and greater reductions in numbers of sites were detected for most of the other temperature sensitive variants. Only receptors containing the full-length V49D substitution were expressed at nearly wild-type levels at 24 °C. Affinities of the mutated receptors for the fluorescent ligand were similar to, or higher than, the affinity of the normal full-length receptor assayed at this temperature.

The screen used to identify these temperature sensitive mutants is based on loss of signaling function at high temperature, which could occur either through a temperature sensitive defect in the signaling capacity of pre-existing receptors, or through a temperature-dependent reduction in receptor numbers at the cell surface. However, responses to pheromone in yeast are not dependent on the abundance of Ste2p, as long as the number of receptors expressed at the cell surface exceeds a minimal threshold. Similar signaling responses are obtained for cells expressing STE2 from a multicopy plasmid (at more than 10 times the level from the normal chromosomal STE2 allele) and cells expressing STE2 from a galactose-inducible promoter under repressing conditions (less than 20% of the normal level for chromosomal STE2) ^{68, 78-80}. Thus, for a mutation to result in a loss of signaling function that is purely due a reduction in receptor numbers at the cell surface, the mutation would have to severely reduce cell surface expression to less than the threshold level. In fact, we observe only a weak correlation between the numbers of receptor variants present at the cell surface and retention of signaling function (Figure 1 and Table 2). The strongest signaling response among the full-length variants at 34 °C was observed for the receptor containing the H94Y substitution, which is present at the cell surface at low abundance. At the same time, the V49D variant, which is present at the highest abundance among tested full-length variants at 34 °C, exhibits a severe defect in α-factor-dependent growth arrest (Figure 1). The correlation between signaling and expression was also weak for cells cultured at 24 °C, where cells expressing the most abundant variant containing the V49D substitution exhibited robust pheromone-dependent growth arrest, but cells expressing the second most abundant variant (H94Y), exhibited weaker responses to α-factor than some of the variants with much lower expression at the cell surface, such as those containing the L211R and I260K substitutions.

Effects of C-terminal truncation of temperature sensitive receptors

The C-terminal tail of Ste2p contains multiple sites implicated in endocytosis of receptors and down regulation of signaling. Thus, truncation of this tail results in increased levels of receptor cell surface expression and hypersensitive signaling responses ^{78, 79}. While our initial screen for temperature sensitive signaling function was conducted using full-length STE2 alleles, the identified temperature sensitive single amino acid substitutions were subsequently transferred into Ste2p alleles that were C-terminally truncated at K304 to facilitate assays of ligand binding (see Figure 3 and Table 2). C-terminally truncated receptors that were otherwise wild-type were expressed at the cell surface at levels approximately ten times higher than full-length wild type receptors (Figure S5 in the Supplemental Information). At 34 °C, cell surface levels of two variants (V49D and H94Y), were considerably higher than the levels observed for full-length receptors with the same substitutions. At 24 °C, the cell-surface abundances of the V49D and H94Y truncated variants were approximately four and twelve times higher, respectively, than the levels of the full-length form with the same mutation. However, the abundances of some mutants (S95Y and S104Y) increased to lesser extents upon truncation. Binding to the truncated form of the L211R variant could not be detected at either 24 or 34 °C, despite the fact that binding to the full-length variant was detected at 24 °C, indicating that C-terminal truncation of this allele actually results in a decrease in cell surface abundance. The affinities of the truncated forms of Ste2p for ligand were generally somewhat lower than the affinities of full-length forms with the same substitutions (but differed from the truncated wild-type by less than a factor of 3).

Figure 3.

C-terminally truncated temperature-sensitive receptors express reduced numbers of binding sites at the cell surface. Strains expressing the indicated truncated STE2 alleles were grown in liquid cultures at 24 °C (A and B) or 34 °C (C). The error bars represent the standard error based on assay of binding to three independent isolates of each strain. Binding was determined by flow cytometry as described in Materials and Methods. (Parameters derived from fitting are listed in Table 2.)

In cases where decreased ligand binding to variant forms of Ste2p is seen at elevated temperatures, the decrease in binding could result either from decreased delivery of newly-synthesized receptors to the plasma membrane during cell growth at high temperature or from temperature-induced unfolding or degradation of receptors that are already present at the cell surface. However, when cells expressing C-terminally truncated Ste2p variants that were initially cultured at 24 °C were treated for 30 minutes at increasing temperatures (up to 52 °C), no progressive loss of binding sites was observed (Figure S6 in the Supplemental Information). This indicates that the observed reductions in numbers of sites are due to reduced synthesis or trafficking rather than unfolding or degradation of pre-existing receptors.

Isolation of temperature sensitive alleles of STE2 based on temperature sensitive loss of ligand binding

To directly identify temperature sensitive mutations resulting in loss of binding sites at high temperature while maintaining high abundances of sites at low temperature, we conducted a second screen of randomly mutated STE2 alleles based only on ligand binding considerations. This screen was conducted in two stages: The first stage used Fluorescence Activated Cell Sorting (FACS) to identify clones that express high numbers of receptors competent for ligand binding at the cell surface at 24 °C, confirming that the selected population is capable of folding to form a native ligand-binding site at the low temperature. The second stage used FACS to sort for clones with reduced numbers of ligand binding sites at 34 °C. This screen was conducted using cells expressing C-terminally truncated receptors in order to enhance the strength of the signal from bound fluorescent ligand binding (see Figure S5 in the Supplemental Information).

From a library of ~115,000 strains expressing randomly mutagenized truncated receptor, we identified three different alleles with the desired phenotype: 1) One of the alleles encoded a single amino acid substitution, F38Y, in the N-terminal extracellular portion of Ste2p. 2) A second allele contained two substitutions, N46I and V49D. The V49D substitution had already been identified as resulting in temperature sensitivity of signaling function (Figures 1 and 2). 3) The third allele included three substitutions, one of which was N158Y at the same position as a mutation that had been classified as having a medium effect on ligand binding (N158D), but whose effects had not been tested in truncated receptors. The temperature sensitive ligand-binding phenotypes of these three alleles were confirmed following retransformation into fresh yeast hosts (Figure 4). All three variants exhibit fewer binding sites at 34 °C than at 24 °C. Furthermore, consistent with the goal of screening for mutations with less severe ligand binding phenotypes that may not be sufficient to interfere with signaling responses, the F38Y substitution exhibits no detectable loss of growth arrest (results not shown).

Figure 4.

Temperature-dependence of ligand binding to cells expressing temperature sensitive receptors isolated based on the screen for loss of ligand binding. The graph shows the fluorescence of bound [K⁷(NBD),Nle¹²] α-factor determined by flow cytometry following growth at 24 °C and 34 °C and incubation with 300 nM [K⁷(NBD),Nle¹²] α-factor. The assay was performed using cells that had been re-transformed with plasmids isolated from clones derived from the original screen. The error bars represent the standard error based on assay of binding to three independent isolates of each strain.

Additivity of effects of destabilizing mutations identified in temperature sensitive mutants containing multiple amino acid substitution

To determine the phenotypes of individual mutations recovered in the seven multiply substituted alleles recovered in the screen for temperature sensitive loss of signaling function, all of the individual substitutions recovered in these alleles were recreated by site-directed mutagenesis and tested individually for their effects on growth arrest and ligand binding of full-length receptors. Surprisingly, although they exhibited various effects on ligand binding, only one out of the 17 tested mutations, L113R, exhibited a growth phenotype that differed from wild-type when introduced individually into Ste2p (Figure S7 in the Supplemental Information). Thus, the observed signaling defects of the remainder of the multiply substituted alleles must result from additive effects of multiple mutations acting together.

Additivity of mutational effects could arise through specific interactions among mutated residues or it could be the result of incremental effects on global protein stability, where the effects of each individual mutation are too small to elicit a growth phenotype. To distinguish between these possibilities, we tested the effects of combining individual mutations derived from different multiply-substituted temperature sensitive alleles into new doubly-substituted alleles. In deciding which mutations to incorporate into these new alleles, each characterized individual mutation was categorized as “severe”, “medium”, or “weak,” based on its effect on ligand binding (Figure S8 in the Supplemental Information): Mutations classified as “weak” resulted in high levels of cell surface receptors that were statistically indistinguishable from levels of wild-type Ste2p. Mutations classified as “severe” resulted in levels of ligand binding that were reduced to less than ~30% of wild-type levels. The category of mutations classified as “medium” included all alleles that resulted in numbers of binding sites that were intermediate between the severe and weak categories. A sampling of the possible combinations of each type of pairing (severe-severe, severe-medium, severe-weak, etc) was constructed by site-directed mutagenesis, transformed into yeast, and assayed for temperature-dependent growth arrest (Figure S9 in the Supplemental Information and Figure 5). The results can be summarized as follows: 1) Combination of two individual weak mutations, a weak with a medium mutation, or two medium mutations all resulted in receptor variants with α-factor-dependent growth indistinguishable from that conferred by wild-type Ste2p. 2) Combinations of two individual severe mutations resulted in receptors that exhibited no detectable function at any temperature in the growth arrest assay. 3) Combinations of medium mutations with severe mutations always yielded receptors with at least partial function, ranging from fully functional at all temperatures (S207I + M69K) to strongly temperature sensitive (S213F+ F55L and S213F + M69K).

Figure 5.

Pattern of additivity of individual temperature-sensitive mutations that did not yield temperature sensitive growth arrest as single substitutions. On the axes, the color of the circles indicates the effect of the mutation alone on the number of cell surface ligand binding sites. The fill pattern of the squares at the intersection of two mutations indicates the effect of the combination of the two mutations, as measured by growth arrest assay.

A possible explanation for the additive effects of the mutations recovered from multiply-substituted alleles could be that the individual substitutions cause modest changes in the levels of receptors at the cell surface such that the population of receptors remains higher than the threshold required for full signaling response ^{68, 78-80}, whereas combinations of substitutions that exhibit measurable phenotypes reduce the numbers of receptors below this threshold. As a test of this idea, we examined the effects of reducing the expression of the receptors containing individual mutations by encoding the mutant genes on centromere-based CEN plasmids, expected to provide a 5-10 fold reduction in cell surface expression compared to the multicopy plasmids used elsewhere in this study ⁷⁴. When 21 strains with CEN plasmids expressing individual mutations derived from multiply-substituted STE2 were assayed for growth arrest in the presence of α-factor at different temperatures (Figure S10 in the Supplemental Information), most showed phenotypes indistinguishable from what was observed with multicopy plasmids. However, three variants (containing the substitutions L113R, Q135H and S207I) that exhibited no signaling phenotypes when expressed from multicopy plasmids did, in fact, exhibit diminished growth arrest at high temperature when expressed from CEN plasmids. The variant containing the S207I substitution is expressed at relatively low levels when encoded on a multicopy plasmid (Figure S8 in the Supplemental Information), thus reduction of gene copy number may further reduce the number of cell surface receptors below a threshold required for signaling. However, this explanation does not apply to the observed signaling defect of the Q135H variant, since this Ste2p containing the Q135H substitution is present at the cell surface at nearly wild-type levels when encoded on a multicopy plasmid. Thus, the basis for the signaling defect of CEN plasmid-encoded Q135H variant remains unclear.

Identification of suppressors of temperature sensitive mutations

To identify mutations likely to enhance the thermal stability of the yeast α-factor receptor Ste2p, we conducted random mutagenesis of temperature sensitive STE2 alleles and screened for second site intragenic suppressors of temperature sensitivity using FACS to identify clones that maintain high numbers of cell surface binding sites for [K⁷(NBD),Nle¹²] α-factor following treatment at elevated temperature. Libraries of mutagenized C-terminally truncated STE2 genes were constructed by error-prone PCR of templates encoding four different temperature sensitive alleles (V49D, S95Y, S104Y, and I260K). Each of the four libraries contained approximately 100,000 clones. A gate for sorting was established using a two dimensional analysis in which the fluorescence of bound ligand is plotted against the forward scattering, an indication of cell size (Figure S2 in the Supplemental Information). The gate was set so as to collect the 1% of cells exhibiting the highest levels of bound fluorescent ligand (out of 200,000 sorted cells).

A total of 72 clones collected from each library were re-cultured and assayed for fluorescent ligand binding levels at a saturating ligand concentration. Of the 288 clones tested, 101 exhibited increased levels of ligand binding compared to the levels of fluorescence for the relevant starting allele. (When ten randomly selected plasmids encoding examples of these alleles were transferred to a fresh yeast host and tested for binding of [K⁷(NBD),Nle¹²] α-factor, nine of them resulted in increased numbers of cell-surface ligand binding sites compared with the control temperature sensitive allele (data not shown), indicating low levels of false positives in this pool of suppressed alleles.) The sequence changes observed in the STE2 genes from all the confirmed clones isolated in the suppressor screen (omitting duplicate clones, clones with reversions, and clones containing only silent mutations) are listed in Table S8 in the Supplemental Information. Mutations that were recovered from multiple mutagenized libraries, mutations that were recovered from multiple independent clones within a single library, and mutations that were recovered from alleles containing only a single amino acid substitution (other than the temperature sensitive starting mutation) were selected for further analysis. The allele with the highest level of suppression (R58G, S95Y) restored ~65% of the wild-type number of receptors; most of the suppressed alleles restored levels to less than 35% of wild type levels. Several of the recovered second-site mutations also suppressed temperature sensitive defects in pheromone-dependent cell cycle arrest (Figure S11 in the Supplemental Information).

Testing for global suppression in an array of site-directed double mutants

To determine the extent to which the identified suppressor mutations can suppress different temperature sensitive mutations, site-directed mutagenesis was used to create a double mutant array consisting of all possible combinations of seven different temperature sensitive mutations with nineteen different suppressor mutations isolated from the mutagenic libraries. To provide additional breadth for the testing of the “global” nature of the suppressors, the temperature sensitive mutations F38Y, A52T, and H94Y were included in the double mutant array even though these mutations had not served as templates for any of the libraries that were screened for suppressors. Cells expressing each component of the double mutant array were assayed for ligand binding under saturating conditions after growth at restrictive temperatures (Figure 6) and for the ability to elicit growth arrest in response to α-factor, assayed at permissive and restrictive temperatures, by serial dilution onto solid medium containing different concentrations of α-factor (Figures S12-S15 in the Supplemental Information). The tested restrictive temperature for the double mutations including the substitutions F38Y or A52T was 34 °C, the temperature at which a temperature dependent loss of ligand binding is observed ⁶⁴. The tested restrictive temperature for the double mutations including the substitutions V49D, H94Y, S95Y, S104Y, or I260K was 30 °C, the minimum temperature at which a temperature-dependent loss of growth arrest is observed. The quantitation of percent recovery, calculated as “R-values”, is displayed as a heat map in Figure 7A and the statistical significance of the observed suppression for each combination is displayed as a second heat map in Figure 7B.

Figure 6.

Ligand binding assays performed on yeast strains in the double mutant array. Strains were cultured overnight at 30 °C (34 °C for F38Y and A52T mutants) and assayed for ligand binding with 300 nM fluorescent α-factor. The error bars represent the standard error based on assay of binding to three independent isolates of each strain.

Figure 7.

Summary of the effects of combinations of suppressor and temperatures sensitive mutations on numbers of ligand binding sites at the cell surface. The temperature-sensitive mutations are listed across the top of the tables. Suppressor mutations are listed along the sides. A) Each mutation pair was scored using the metric R, defined as the difference between the fluorescence of the double mutant and the temperature-sensitive single mutant divided by the difference between wild-type and the temperature-sensitive single mutant, as a measure of recovery of ligand binding levels due to the suppressor. Magenta boxes indicate positive effect of the suppressor mutation on ligand binding levels, while cyan indicates that the suppressor mutation had a negative effect on ligand binding levels. Note that the scale is clipped to better highlight the intermediate effects. B) The indicated p-values refer to the comparison of the ligand binding of the double mutant to that of the temperature sensitive mutant alone. Only p-values for double mutants with an R-value > 5 or < −5 were mapped. Magenta boxes indicate the significance of the p values for alleles showing enhanced ligand binding resulting from suppression. Blue boxes indicate the significance of the p-values for alleles showing decreased ligand binding resulting from combining the two mutations.

Of the 19 original suppressor mutations, 14 exhibited significant suppression of the loss of cell-surface binding sites for at least one temperature sensitive starting allele (Figure 6). Seven of these suppressed more than one starting allele (Figure 7; R > 10, p < 0.01). The five mutations R58S, R58G, F119L, F119S and M218T provided the most general global suppression, each enhancing ligand binding to at least three different temperature sensitive mutations. In addition to their effects on receptor abundance, several suppressor mutations also at least partially reversed the temperature sensitive signaling defects of starting alleles (Figures S12-S15) as indicated by reduced growth of the double mutant strain on α-factor-containing plates at 34 °C compared with cells expressing the starting allele alone (referred to as “Temperature Sensitive” in Figures S12-S15 in the Supplemental Information). By this criterion, the R58S substitution improved the signaling of the temperature sensitive S95Y, I260K, and V49D Ste2p variants and the F119L and F119S substitutions both improve the signaling of the S95Y and S104Y variants. In accordance with previous results ⁸¹, the R58G substitution retains full signaling function in a quantitative assay of transcriptional induction of the FUS1-lacZ reporter gene ⁸¹. Similar confirmation of signaling function was obtained for R58S and F119L substitutions (results not shown). No signaling assays are shown for suppressors of temperature sensitive starting mutations that do not exhibit defects in signaling at any temperature (H94Y and F38Y).

Interactions between mutations of R58 and S95 exhibit partial allele specificity

The substitutions R58G and R58S provide suppression of multiple temperature sensitive mutations, but their suppression of the starting mutation S95Y is particularly strong (restoring >90% of wild-type levels of binding for the doubly substituted receptors compared to <1% for S95Y alone; see Figures 3 and 6). Furthermore, the two mutations at R58 were the only recovered substitutions that suppress the temperature sensitive allele S95Y. This suggested that genetic interactions between positions 58 and 95 might include a specific component, reflecting direct interaction between amino acids at those two positions. Based on a homology model of Ste2p that was generated through comparison to the rhodopsin structure ⁸², S95 and R58 are predicted to be in close proximity (Figure S16 in the Supplemental Information). A simple explanation of the observed interaction is that replacing R58 with a residue with a smaller side chain (R58G and R58S) relieves steric clash resulting from replacing S95 with a residue with a larger side chain (S95Y). To test this hypothesis, seven different R58 substitutions (R58A, R58G, R58L, R58K, R58Q, R58S, and R58W) involving replacement of R58 with side chains of different sizes, polarities, and charges, were assayed for their effects on ligand binding at restrictive temperatures (either 30 °C or 34 °C) when combined with four different temperature sensitive starting mutations (F38Y, A52T, S95Y, and S104Y). Consistent with the idea that suppression arises, at least in part, from relief of steric clash, introduction of alanine at position 58 also suppressed the temperature sensitive loss of binding sites resulting from the S95Y mutation, increasing cell-surface binding sites to ~55% of wild-type levels (Figure 8). However, the presence of alanine at position 58, (a substitution that has previously been reported to reduce ligand binding ^{83, 84} also provides an element of global suppression, since this substitution also partially suppresses the temperature sensitive mutations A52T and S104Y. Substitution of tryptophan or lysine for R58 provided only very weak (2.8% and 1.2% of wild-type levels, respectively) suppression of S95Y, while substitution of glutamine or leucine provided intermediate (16.8% and 13.7% of wild-type levels, respectively) levels of suppression. None of these substitutions of larger residues for R58 provided any suppression of the temperature sensitive mutations other than S95Y.

Figure 8.

Suppression of different temperature sensitive mutations by site-direct substitutions at position R58. Binding of [K⁷(NBD),Nle¹²] α-factor to cells was conducted as described in Materials and Methods at a concentration of 300 nM ligand. For the clones containing the S95Y and S104Y mutations, ligand binding was measured after growth at 30 °C. For the clones containing the A52T and F38Y mutations, ligand binding was measured after growth at 34 °C. The error bars represent the standard error based on assay of binding to three independent isolates of each strain.

The effects of some combinations of suppressors are additive

As a test of whether the effects of the identified suppressor mutations are additive, STE2 alleles encoding various combinations of the global suppressors R58S, F119L, and M218T with the temperature sensitive mutations A52T and F38Y were created by site-directed mutagenesis. Strains expressing these alleles were assayed for fluorescent ligand binding under saturating conditions after overnight growth at 34 °C (Figure 9). For one of the temperature sensitive starting alleles (A52T), all combinations of suppressor mutations exhibit greater effects than individual mutations in restoring ligand binding (Figure 9). The particular combination of R58S and F119L restores the full level of wild type ligand binding. For another starting allele, F38Y, one suppressor, F119L provides restoration of ligand binding for cells cultured at high temperature (to about 70% of the wild type abundance), but no combination of F119L with additional suppressors provided any improvement over F119L alone. However, the combination of the weaker suppressors R58S and M218T together with F38Y did result in slightly higher levels of ligand binding than either of these two suppressors alone.

Figure 9.

Global suppressors identified from the double mutant array exhibit additive effects on ligand binding. Combinations of the global suppressor mutations R58S, F119L, and M218T with the temperature sensitive mutations A52T (A) and F38Y (B) were constructed by site-directed mutagenesis. Assays of binding of [K⁷(NBD),Nle¹²] α-factor were performed using 300 nM ligand on cells cultured at 34 °C. The error bars represent the standard error based on assay of binding to three independent isolates of each strain.

Suppressor mutations stabilize receptor in the absence of temperature sensitive starting mutations

To determine whether the identified suppressor mutations provide stabilization of otherwise wild type receptors that will facilitate purification and crystallization of Ste2p, it will be necessary to examine the stabilities of mutant receptors in a solubilized state useful for crystallization trials. However, our laboratory has, to date, been able to achieve only very poor yields of purified solubilized Ste2p receptors, despite relatively high expression levels in cells. Furthermore, while C-terminal truncation generally leads to increased levels of Ste2p expression at the cell surface, previous attempts to purify the C-terminally truncated forms of the receptor have been less successful than the application of similar procedures to full-length Ste2p (Nadia Fedoriw, Kathleen Clark, and M. Dumont, unpublished results). In view of these difficulties, we have focused on the development of initial stages of a purification procedure involving immobilization of receptors on small (8 μm) polystyrene beads, so that purification of correctly folded receptors can ultimately be monitored by quantitation of binding of fluorescent ligand using flow cytometry. Immobilization via the triple c-myc epitope tag fused to the C-terminal of the truncated Ste2p is achieved through the use of mouse anti-myc antibodies in conjunction with beads that are coated with goat anti-mouse antibodies. Dodecylmaltoside was used as a detergent for solubilization, based on its effective solubilization of yeast proteins in general, and on its use in previously reported purifications of Ste2p ^{61-63, 85}.

In initial experiments, only very low yields of bead-bound receptors could be detected by immunoblotting of material eluted from the beads using SDS and urea. However, as shown in Figure 10, the presence of 300 nM α-factor during solubilization and immobilization dramatically improved the yield of Ste2p bound to beads for all the Ste2p variants tested. (Immunoblotting of Ste2p typically gives rise to multiple closely spaced bands, corresponding to differentially glycosylated and phosphorylated forms of the protein ⁸⁶.) Stabilization of solubilized Ste2p by ligand binding has been reported previously ⁶¹. Since the conformational state of a protein would not necessarily be expected to affect accessibility of an appended epitope tag, the observed failure of Ste2p to bind to bead-conjugated anti-tag antibodies in the absence of ligand indicates that instability of solubilized Ste2p causes the protein to adopt a state that leads to aggregation or precipitation. This observed ligand-dependence of bead binding demonstrates the existence of a population of solubilized receptors that retain the capability to bind ligand with high affinity, since the presence of a low concentration of ligand is sufficient to maintain Ste2p in a state that allows the attached epitope tag to bind immobilized antibody.

Figure 10.

Effects of the presence of α-factor and suppressor mutations on the yield of immobilization of Ste2p. The indicated C-terminally truncated Ste2p variants were solubilized using dodecylmaltoside, either in the presence or absence of 300nM α-factor then incubated for 2 hours at 24°C with beads coated with anti-c-myc antibodies. To determine the amount bound, the Ste2p was then stripped from the beads using SDS/urea buffer and subjected to immunoblotting, probed with anti-c-myc antibodies.

We compared immobilization of receptors containing four different suppressor mutations (in the absence of temperature sensitive mutations) with that of normal receptors in the presence and absence of α-factor (Figure 10). Solubilized forms of two variant receptors containing the suppressors, R58G and R58S, bind to antibody-coated beads in the presence of α-factor at approximately the same yield as wild-type receptors. However, in the absence of α-factor, receptors containing either of these substitutions bind to beads at levels that, while still low, are considerably greater than the yield of wild-type Ste2p. Even in the presence of α-factor, the binding yields of variants containing two other suppressor mutations, F119L and M218T, are less than for wild-type receptors, either because these variants are less stable or because they are produced in cells at lower levels than wild-type receptors. In either case, these results fail to provide evidence for stabilization of otherwise wild-type receptors by the F119L and M218T substitutions.

To examine the basis for the loss of Ste2p during purification, the fates of wild-type and R58G variant forms of the protein were monitored by immunoblotting of the samples based on the c-myc tag following incubation of solubilized receptors for different amounts of time in the presence of 0.1% DDM at room temperature (Figure S17 in the Supplemental Information). While both the wild type and mutant proteins migrated predominantly as monomers with a component of dimer, an additional population of higher order oligomers was detected as a function of incubation time for the normal Ste2p receptors but not for the R58G variants.

DISCUSSION

Destabilizing mutations of Ste2p receptors

High levels of non-specific binding and the inviability of yeast cells treated at high temperatures prevented the use of ligand binding as a basis for directly screening for mutations enhancing the thermostability of the Ste2p receptor. Thus, we pursued an approach involving the isolation of temperature sensitive mutations in the receptor that could be used as starting alleles for the isolation of stabilizing second-site intragenic suppressor mutations. Two screens, one for receptors with reduced signaling function and the other for receptors present at reduced levels at the cell surface, yielded a total of seven temperature sensitive alleles encoding single amino acid substitutions. These substitutions are localized to the extracellular N-terminal of the receptor as well as to the extracellular side of predicted transmembrane helices 1, 2, 5, and 6 as shown in Figure 11. Their distribution on different helices and different predicted helical faces suggests that there is no single core of the receptor that provides structural stability, but is generally consistent with the previously observed predominance of mutations affecting stability in extracellular regions of other GPCRs ⁵¹ and the apparent lower tolerance of the extracellular surface of GPCRs to mutation, compared with intracellular regions ⁵². Based on the low frequencies at which these substitutions were recovered, it is likely that the alleles we recovered represent only a small fraction of the range of destabilizing mutations that are possible in the STE2 gene.

Figure 11.

Distribution of temperature-sensitive mutation in Ste2p. The figure shows the predicted topology of Ste2p ⁸¹. Residues in circles with black fills are sites at which single substitutions cause temperature-sensitive phenotypes. The other highlighted residues have only been found to confer temperature sensitivity when they are present in alleles with multiple amino acid substitutions. The residues in enlarged circles had substitutions that caused large decreases in ligand binding. The residues in hexagons had substitutions that caused moderate decreases in ligand binding, while the residues in diamonds had substitutions that caused little decreases in ligand binding.

Seven temperature sensitive alleles containing multiple amino acid substitutions were recovered from the screens that we conducted for mutations conferring temperature sensitive α-factor-dependent growth arrest. We were surprised to find that, with one exception (L113R), the 17 individual mutations from these multiply substituted alleles did not, by themselves, exhibit any temperature sensitive effects on receptor signaling although, some did result in decreased numbers of cell surface binding sites. Attempts to enhance the signaling phenotypes of these individual substitutions by decreasing plasmid copy number (from multicopy, at ~40 copies per cell to CEN, at 1-3 copies per cell) were not generally successful. Thus, either receptor expression level is not the limiting factor in signaling by the multiply-substituted alleles or cell surface levels of receptors containing these individual mutations, even when expressed at low copy number, are above the threshold required for effective signaling, so that combinations of mutations are required to reduce cell surface expression below this threshold. The diversity of this type of multiply substituted temperature sensitive alleles that we recovered indicates that a large number of amino acid substitutions with no obvious phenotype as single mutations are capable of rendering Ste2p temperature sensitive when present in combination with additional amino acid changes.

To test the specificities of interactions among mutations, we assayed the growth arrest phenotypes of 15 different reshuffled combinations of amino acid substitutions originally recovered in the multiply substituted alleles: Eight of the new combinations resulted in temperature sensitive growth arrest phenotypes, two resulted in non-functional receptors at high and low temperatures, and five resulted in a functional receptor that was not temperature sensitive. This indicates that the observed phenotypes of the combined mutations do not generally result from specific interactions between mutated sites, but instead reflect additive or synergistic effects of incremental destabilization by the individual substitutions. Such additivity or synergy of effects on protein stabilization is consistent with the idea that stabilized proteins can be constructed by combining single mutations providing small individual enhancements of stability ^{5, 87}.

All of the temperature sensitive receptor variants that we isolated, including those identified based on loss of signaling function, are present at the cell surface at reduced levels compared to equivalent (full-length or truncated) wild-type receptors. Furthermore, in all cases where the levels at the cell surface were high enough to be measured, the numbers of cell surface binding sites on cells expressing the variant receptors were lower at 34°C than at 24°C. Some temperature sensitive variant receptors were present at such low levels on cells at 24°C that it is likely that the further reduction in numbers of resulting from growth at 34°C was sufficient to reduce receptor abundance below a threshold required in the yeast system for effective signaling. However, reduced receptor abundance does not provide an explanation for all the observed effects of the identified temperature sensitive mutations. For-example, C-terminally truncated forms of receptors containing the amino acid substitutions V49D and H94Y exhibit defects in signaling at 34 °C (Figure 1) even when they are expressed at the cell surface at levels that are 10-30 times greater than the number of binding sites on cells expressing chromosomally-encoded normal full-length Ste2p ^{74, 88} (see Table 2). Thus, the observed defects in signaling by these variants at high temperature must be due to a loss of receptor signaling function that is not accompanied by loss of ligand binding capabilities.

Decreased abundances of mutant receptors can arise either through decreased synthesis and trafficking, or enhanced unfolding or degradation of receptors. However, no progressive loss of binding sites is seen in variant-expressing cells subjected to short-term treatment at increased temperatures (Figure S6 in the Supplemental Information). Thus, the observed decreases in receptor numbers observed upon culturing variant-expressing cells at 34 °C most likely reflect decreased flux of newly synthesized receptors to the plasma membrane, perhaps due to recognition and destruction of misfolded proteins by the quality control system of the endoplasmic secretory pathway ⁸⁹.

The fact that we observe both increases and decreases in cell surface abundance in comparisons between full-length and C-terminally truncated temperature sensitive receptors indicates that two disparate types of effects result from Ste2p truncation. In most cases, C-terminal truncation resulted in increased cell surface abundances of receptors, as expected if the C-terminal tail is involved in endocytosis and degradation of receptors. However, for one mutation (L211R), C-terminal truncation resulted in decreased numbers of receptors at the cell surface. In such cases, removal of the C-terminal tail may lead to loss of stabilizing interactions between the C-terminal tail and the rest of the receptor, as has been observed previously for certain mutant forms of Ste2p ⁹⁰. Since many constructs used for crystallographic structure determination of GPCRs involve the use of C-terminal truncation to remove portions of the proteins that are expected to be flexible, possible destabilization of receptors by truncation is a factor that merits further consideration.

Mutations that suppress temperature sensitivity

By screening randomly mutagenized temperature sensitive alleles, we identified 19 intragenic second-site suppressors that result in elevated levels of cell surface binding sites for fluorescent ligand. Suppressors that act by providing overall thermodynamic stabilization of receptors would be expected to be “global”, meaning that they would be capable of suppressing the effects of multiple temperature sensitive mutations at different positions in the protein ⁶⁵. To examine the full spectrum of interactions between different combinations of suppressors and temperature sensitive starting alleles, we tested all possible pairings of sets of candidate suppressor and temperature sensitive mutations to determine numbers of ligand binding sites at the cell surface after overnight growth at restrictive temperatures (either 30 °C or 34 °C). Seven mutations (R58G, R58S, F119L, F119S, M218T, L248R, and T279A) of the original 19 candidate suppressors provide “global suppression” in that they are capable of suppressing more than one of the tested temperature sensitive starting alleles (Figure 7). The locations of these seven suppressor mutations, along with the single amino acid temperature sensitive mutations for which suppression was tested are shown in Figure 12. Both the M218T and the R58G substitutions had previously been recovered in a prior screen performed by our laboratory, selecting for mutations that suppress the loss-of-signaling function of multiple mutations in the third transmembrane helix of Ste2p ⁸¹. The conclusion that mutations providing global suppression act by providing overall stabilization of the receptor is reinforced by the additive effects seen in combining multiple suppressors with different temperature sensitive mutations. Enhancements of protein stability achieved by combination of multiple individual stabilizing mutations have been reported for other membrane proteins ^{5, 51, 91}.

Figure 12.

Distribution of suppressor mutations in Ste2p mapped on to the predicted transmembrane topology of Ste2p. The residues highlighted in black are the locations of the temperature sensitive mutations. The other highlighted residues indicate the positions of suppressors. Large grey circles show the locations of global suppressors, defined as mutations that restore at least 10% of ligand binding to a 99% confidence level for more than one temperature-sensitive mutation. Residues depicted as smaller dark grey circles are allele-specific suppressors, defined as being able to restore at least 10% of ligand binding to a 99% confidence level for no more than one of the tested temperature-sensitive mutations.

The suppressor mutations shown in Figure 12 are widely distributed, indicating that determinants of stability are not restricted to one particular region of the sequence. The diversity of the recovered suppressor mutations suggests that stabilizing mutations may be relatively common, implying that there is little evolutionary pressure to maximize Ste2p stability, or even that there may be pressure to maintain instability ⁹². Substitutions that enhance stability also appear to constitute a common class of mutation in other membrane proteins ⁹³.

In addition to providing increased numbers of receptors at the cell surface, some of the isolated suppressor mutations, such as substitutions at R58 and F119, also restore enhanced signaling in the growth arrest assay. Suppression of signaling defects of some of the temperature-sensitive starting alleles that result in low cell surface receptor levels, such as S95Y and S104Y, is likely to be the result of restoration of these levels to the threshold required for effective signal transduction. However, for other temperature sensitive alleles, such as V49D, that do not greatly reduce numbers of cell surface binding sites, suppression must be occurring through a mechanism other than simple augmentation of the numbers of receptors. Instead, the suppressing mutations may be compensating for mechanistic deficiencies in ligand-induced activation of the receptor.

The extent to which the stabilities of membrane proteins in native membranes correspond to stabilities in the detergent-solubilized states of such proteins used for purification and crystallization remains unknown. As an initial step toward characterizing the behavior of potentially stabilized Ste2p variants in detergent, we investigated the antibody-mediated immobilization on polystyrene beads of variants containing suppressor mutations introduced into C-terminally truncated forms of Ste2p. While previous efforts in our laboratory to immobilize and purify truncated Ste2p in the absence of ligand have been unsuccessful, we were able to immobilize the wild-type truncated receptors in the presence of the receptor ligand α-factor, consistent with stabilization of the receptor by ligand ⁶¹ and demonstrating that the solubilized receptor remains in a state competent for high affinity ligand binding. In addition, we were able to achieve immobilization of truncated receptors containing the suppressor mutations R58G and R58S even in the absence of ligand, indicating that these mutations can provide stabilization of receptors. (Despite the evidence that solubilized Ste2p retains the capacity for ligand binding, we have not, to date, been able to directly detect binding of fluorescent ligands to the immobilized receptors under any of the conditions tested. This difficulty stems, in part from the complications of attempting to determine ligand binding to receptors immobilized in the presence of ligand.)

In addition to the observed patterns of global suppression, several of the identified suppressor mutations exhibit partial specificity for particular temperature sensitive starting alleles, indicative of specific interactions between the sites of the temperature sensitive and suppressing mutations. The most striking case of this is the suppression of mutations at several sites on the predicted second transmembrane helix of Ste2p by mutations at R58 in the predicted first transmembrane helix. Mutations at position R58 were unique among the suppressor substitutions in their ability to provide substantial suppression of the temperature sensitive phenotype of the substitution S95Y. Furthermore, while global suppression of multiple temperature sensitive mutations was only observed when R58 was replaced by amino acids with small side chains, some specific suppression of S95Y was observed when bulky residues were substituted for R58. This specific suppression may depend on the loss of a positive charge at position 58, since the substitution R58K, which preserves the charge at this position, was the only tested substitution that did not result in significant suppression of S95Y. An additional example of allele-specific suppression involves the temperature sensitive mutation I260K, which is suppressed by four nearby amino acid substitutions (I256N, P258A, P258L, and G273D) that do not suppress any of the other temperature sensitive mutations in the tested array (see Figure 7). Substitution of lysine for I260 may result in a distortion of the sixth transmembrane helix that can be reversed by diverse suppressors in this region. This is consistent with previous observations that any replacement of the proline at position 258 (the site of two substitutions that suppress I260K), as well as a variety of other substitutions in neighboring regions of the sixth transmembrane segment, result in constitutive activation of the receptor ^94-96.

In summary, we have developed and tested genetic screens for mutant α-factor receptors that display temperature sensitive signaling function and temperature sensitive ligand binding. All of the recovered singly-substituted temperature sensitive alleles exhibit decreases in total and cell-surface expression both at low and high temperatures, reflecting either impaired synthesis and targeting or unfolding and degradation by cellular quality control mechanisms. However, temperature sensitivity appears to result from effects both on receptor expression levels and on signaling function. We then identified intragenic second-site suppressors of diverse temperature sensitive alleles from randomly mutagenized libraries, including seven substitutions that were able to suppress multiple temperature sensitive alleles, indicating that they act by providing overall stabilization of receptors. Detergent-solubilized Ste2p receptors containing substitutions at R58 exhibited improved immobilization onto antibody-conjugated beads and reduced levels of aggregation compared to wild-type receptors.

ABBREVIATIONS

TMP

transmembrane protein

GPCR

G protein coupled receptor

NBD

(7-nitrobenz-2-oxa-1,3-diazol-4-yl)

FACS

Fluorescence Activated Cell Sorting