Receptor-Mediated Activation of Gsα: Evidence for Intramolecular Signal Transduction
- Departments of 1Cellular and Molecular Physiology (G.G., C.H.B.) and2Pharmacology (S.R.M.), Yale University School of Medicine, New Haven, Connecticut 06520-8026, and 3Department of Psychiatry and Center for Neurobiology and Psychiatry (P.T.W.), University of California, San Francisco, California 94143
Abstract
To investigate the mechanism by which cell surface receptors activate heterotrimeric G proteins, we applied a scanning mutagenesis approach to the carboxyl-terminal 40% of αs (residues 236–394) to identify residues that play a role in receptor-mediated activation. We identified four regions of sequence in which mutations significantly impaired receptor-dependent stimulation of cAMP synthesis in transiently transfected cyc− S49 lymphoma cells, which lack endogenous αs. Residues at the carboxyl terminus are likely to be receptor contact sites. Buried residues near the bound GDP are connected to the carboxyl terminus by an α helix and may regulate GDP affinity. Residues in two adjacent loops of the GTPase domain at the interface with the helical domain, one of which includes a region, switch III, that changes conformation on GTP binding, are positioned to relay the receptor-initiated signal across the domain interface to facilitate GDP release. Consistent with this hypothesis, replacing the helical domain of αs with that of αi2 in an αs/αi2/αs chimera corrects the defect in receptor-mediated activation caused by αi2substitutions on the GTPase side of the interface. Thus, complementary interactions between residues across the domain interface seem to play a role in receptor-catalyzed activation.
Heterotrimeric G proteins transmit hormonal and sensory signals received by cell surface receptors to effector proteins that produce a wide variety of cellular responses (Neer, 1995). Theα, β, and γ subunits of G proteins are associated in the inactive GDP-bound form. Receptors activate G proteins by catalyzing replacement of GDP by GTP on the α subunit, resulting in dissociation of α·GTP from βγ, each of which can transmit signals to effectors. Hydrolysis of GTP by the α subunit regulates the timing of deactivation and reassociation of α with βγ. As intermediaries between receptors and effectors, G proteins play a crucial role in determining the specificity, nature, and degree of amplification of transmitted signals. For example, Gs mediates stimulation of adenylyl cyclase by β-adrenergic receptors. However, the molecular determinants that specify receptor/G protein interactions and the mechanism by which these interactions lead to G protein activation are not well understood.
Studies of G protein function can be interpreted in the context of the x-ray crystal structures of GTPγS-bound (active) (Noel et al., 1993; Coleman et al., 1994) and GDP-bound (inactive) (Lambright et al., 1994; Mixon et al., 1995) αsubunits and of αβγ heterotrimers (Wall et al., 1995; Lambright et al., 1996). The α subunits consist of two domains, a GTPase domain that resembles the oncogene protein, p21ras, and a helical domain consisting of α helices and connecting loops. Because the bound nucleotide is buried in the cleft between these domains, receptor-mediated nucleotide exchange presumably involves a conformational change that opens the cleft. Comparison of the structures of GTPγS-bound and GDP-bound α subunits reveals three regions in the GTPase domain (switches I–III) that change conformation, which could be involved in the activation process.
All α subunit residues involved in associating with receptors and with βγ, which is required for receptor-mediated activation (Fung, 1983), have not been identified. X-ray crystal structures of the αβγ complex (Wall et al., 1995; Lambright et al., 1996) showed that two α subunit regions contact the β subunit, the amino-terminal α helix and a region that includes switches I and II. The functional importance of these regions has been demonstrated using proteolysis (Navon and Fung, 1987), mutagenesis (Miller et al., 1988; Journot et al., 1991), and cross-linking (Garcia-Higuera et al., 1996). Numerous genetic and biochemical studies, reviewed by Neer (1995), have implicated the carboxyl-terminal region of the α subunit in interaction with receptors. However, the locations of these βγ and receptor-interacting residues, which are distant from the domain cleft, have not revealed the mechanism of receptor-mediated G protein activation.
We exploited the differences in receptor specificities of αs and αi2, which are relatively divergent members of the α subunit family, sharing ∼40% amino acid identity, to identify additional α subunit residues that mediate a response to receptor stimulation. Measurements of receptor-stimulated guanine nucleotide exchange in reconstituted phospholipid vesicles have demonstrated that the efficiency with which the β-adrenergic receptor regulates αi2 is ∼10% of that seen for αs (Rubensteinet al., 1991). The αs residues that specify interaction with the β-adrenergic receptor have been localized to the carboxyl-terminal 40% of αsby means of an αi2/αschimera (Masters et al., 1988).
By examining a panel of αs mutants in which clusters of residues were replaced by αi2homologs or alanines, we identified four regions of sequence that are specifically required for receptor-mediated activation. Residues at the extreme carboxyl terminus are the most likely receptor contact residues. Buried residues near the guanine ring of the bound GDP, connected to the carboxyl terminus by an α helix, may transmit the signal from the receptor to modulate GDP affinity. Residues in two adjacent loops of the GTPase domain at the interface with the helical domain, one of which includes switch III, are positioned to relay the receptor-initiated signal across the domain interface to facilitate GDP release. Consistent with this hypothesis, replacing the helical domain of αs with that of αi2in an αs/αi2/αschimera corrects the defect in receptor-mediated activation caused by αi2 substitutions on the GTPase side of the interface. Thus, complementary interactions between residues across the domain interface seem to play a role in receptor-catalyzed activation.
Experimental Procedures
Materials.
The expression vector pcDNA I/Amp was obtained from InVitrogen (Carlsbad, CA). Plasmids used for electroporation were prepared using Qiagen Plasmid Maxi Kits (Santa Clarita, CA). Isoproterenol, 1-methyl-3-isobutylxanthine, cAMP, and ATP were obtained from Sigma Chemical (St. Louis, MO). Dulbecco’s modified Eagle’s medium, minimal essential medium with Earle’s salts, and geneticin were obtained from GIBCO BRL (Grand Island, NY). Horse serum was obtained from Hyclone (Logan, UT). [3H]Adenine was obtained from Amersham (Little Chalfont, UK).
Construction of α subunit mutants and chimeras.
The αs mutant constructs were generated from rat αs cDNA (Jones and Reed, 1987). Chimeric α subunits were constructed from rat αs cDNA and mouse αi2 cDNA (Sullivan et al., 1986). Subcloning and mutagenesis procedures were verified by restriction enzyme analysis and DNA sequencing. All α subunit constructs produced in this study contain an epitope, referred to as the EE epitope (Grussenmeyer et al., 1985), which was generated by mutating αs residuesDYVPSD (189–194) toEYMPTE and αi2 residuesSDYIPTQ (166–172) toEEYMPTE (single-letter amino acid code; mutated residues are underlined). This epitope does not affect the ability of αs to activate adenylyl cyclase in response to stimulation by the β-adrenergic receptor (Wilson and Bourne, 1995).
The amino acid substitutions in the αs mutant constructs produced in this study are shown in Figs.1 and 5. To generate these constructs, the αs cDNA was subcloned into the expression vector pcDNA I/Amp as a HindIII fragment. The mutations in αs(1), αs(2), αs(10), αs(11), αs(12), αs(13), αs(14), αs(15), αs(16), and αs(17) were introduced into the αs cDNA by oligonucleotide-directed in vitro mutagenesis (Kunkelet al., 1987) using the BioRad Muta-Gene kit (Hercules, CA). The mutations in αs(3), αs(4), αs(5), αs(6), αs(7), αs(8), and αs(9) were introduced into the αs cDNA by ligatingBamHI fragments from previously generated constructs (Berlot and Bourne, 1992) that contained these mutations into αs in place of the analogous fragment. To produce αs(2 + 6), αs(6) was digested with BglII andEcoRV to yield a fragment containing the αs(6) mutations, which was ligated into αs(2) in place of the analogous fragment to produce an αs cDNA containing both the αs(2) and αs(6) mutations.
Panel of αs mutant constructs. All mutations are within the carboxyl-terminal 40% of the rat αs sequence (Jones and Reed, 1987), which is depicted in two sections (residues 236–314 and 315–394). Top lines, sequence of αs. Second lines, mouse αi2 (Sullivan et al., 1986) in the corresponding region. Dashes, residues identical to those of αs. Numbered sequences, individual mutant constructs. For each αs mutant construct, mutated residues are shown by the single-letter amino acid code. Dashes, residues identical to those of αs. Underlined sequences (located in regions 1–4), mutations that disrupted receptor-mediated activation of αs. Elements of secondary structure, determined from the structure of αs·GTPγS (Sunahara et al., 1997), are indicated:a, αhelices; b, βstrands; anddashes, turns and loops. Regions that switch conformation between the GDP-bound and GTPγS-bound forms of αt (Lambright et al., 1994) and αi1 (Mixon et al., 1995) (switches II and III) are indicated. The alignment shown, which is based on the recently solved structure of αs·GTPγS (Sunahara et al., 1997), differs from our previous alignment (Berlot and Bourne, 1992) in the location of an insertion of αssequence relative to that of αi2. In the previous alignment, αs residues 324–336 were inserted between αi2 residues 299 and 300.
Substitution of buried but not solvent-exposed residues in region 3 impairs receptor-mediated activation. A, Representation of αs mutants as in Fig. 1. B, cAMP accumulation in cyc− cells containing the indicated mutants in the αs context. Cells were electroporated with 30 μg of vector alone and of vector containing αs and αs(16) and with 45 μg of vector containing αs(17). Dark gray, cAMP values from unstimulated cells. Light gray, cAMP values from cells stimulated with 0.1 mm isoproterenol. C, Receptor-independent cAMP accumulation incyc− cells containing the indicated mutants in the αsRC context. For each mutant, the same amount of plasmid was used as is indicated in B for the corresponding αs mutant. cAMP levels in [3H]adenine-labeled cells were determined as described in Experimental Procedures. All values represent the mean ± standard error of at least three independent experiments.
Because receptor-dependent stimulation of cAMP synthesis was used to measure receptor-mediated activation of the mutant αs constructs, any effects the mutations might have on receptor-independent cAMP synthesis were controlled for by measuring basal cAMP accumulation in response to parallel constructs (the αsRC versions), in which substitution of cysteine for the arginine at position 201 (Landis et al., 1989) inhibits GTPase activity and causes constitutive activation. αsRC versions of the constructs were produced by ligating BamHI fragments containing the mutations into αsRC in place of the analogous fragment.
An αsis chimera, in which αs residues 62–235 are replaced by the homologous αi2 residues, was produced from αs and an αis chimera, in which αs residues 1–235 are replaced by the homologous αi2 residues. The αi2 cDNA was subcloned into pcDNA I/Amp as anEcoRI fragment. To generate αis, the αs cDNA was digested with BamHI and the fragment encoding αs residues 236–394 and the 3′ untranslated region of αs was ligated into αi2 in place of the analogous fragment. Then, αsis was generated using polymerase chain reactions that produced DNA fragments with overlapping ends that were combined subsequently in a fusion polymerase chain reaction (Hortonet al., 1989). An RC version of αsis, αsisRC, was produced by substituting cysteine for Arg179, which causes constitutive activation of αi2 by inhibiting GTPase activity (Wong et al., 1991). αsisRC was derived from αisRC, which was generated by ligating the BamHI αs fragment encoding αs residues 236–394 and the 3′ untranslated region of αs into αi2RC in place of the analogous fragment. To produce αsisRC, αi2RC was digested with DraIII to yield a fragment containing the RC mutation, which was ligated into αsis in place of the analogous fragment.
To introduce the αs(1), αs(2), αs(6), and αs(2 + 6) mutations into αsisRC and αsis, these mutations were first subcloned as BamHI fragments into αi2RC and αi2 in place of the analogous fragments. Digestion of these αi2RC and αi2constructs with DraIII yielded fragments containing the αs(1), αs(2), αs(6), and αs(2 + 6) mutations with or without the RC mutation, respectively, which were ligated into αsispcDNA I/Amp in place of the analogous fragments to produce αsisRC and αsis constructs, respectively, containing the desired mutations.
cAMP accumulation assay.
Transient transfections were performed using a subclone of cyc− S49 lymphoma cells (Bourne et al., 1975) that stably expresses Simian virus 40 large T antigen. These cells were maintained in Dulbecco’s modified Eagle’s medium containing 10% heat-inactivated horse serum and 0.6 mg/ml geneticin. Transient transfection of cells expressing TAg with vectors containing a Simian virus 40 origin of replication has been shown to maximize expression levels (Clipstone and Crabtree, 1992). Therefore, we used the expression vector, pcDNA I/Amp, which contains a Simian virus 40 origin of replication, as well as the cytomegalovirus promoter, to electroporate the cyc− cells.
The α subunit constructs were introduced intocyc− cells (2 × 107 cells in 1.0 ml of 20 mmHEPES-buffered minimal essential medium with Earle’s salts without bicarbonate) by electroporation at room temperature using a GIBCO BRL Cell-Porator (capacitance setting, 1600 μF; voltage setting, 250 V; Grand Island, NY). After electroporation, the cells were added to 4.0 ml of Dulbecco’s modified Eagle’s medium containing 10% heat-inactivated horse serum in 60-mm tissue culture dishes. At 24 hr after electroporation, the cells were labeled with 12 μCi/ml [3H]adenine. Then, 24 hr later, cAMP accumulation was measured. The cells first were washed in assay medium (20 mm HEPES-buffered Dulbecco’s modified Eagle’s medium without bicarbonate). The cells were transferred to 24-well plates and incubated at 37° for 30 min in the same medium containing 1 mm concentration of the phosphodiesterase inhibitor 1-methyl-3-isobutylxanthine, with or without the addition of 0.1 mm isoproterenol (a saturating stimulus). During this incubation, the cells attached to the wells. Reactions were terminated by aspiration and the immediate addition of 5% trichloroacetic acid plus 1 mm concentration each of ATP and cAMP. Nucleotides were separated on ion exchange columns (Salomon et al., 1974). cAMP accumulation was expressed as [3H]cAMP/([3H]ATP + [3H]cAMP) × 1000.
Results
Panel of αs mutant constructs for studying receptor-mediated activation.
The 159-residue carboxyl-terminal segment of αs (residues 236–394), which specifies interaction with the β-adrenergic receptor (Masterset al., 1988), contains 59 amino acids that are identical in the sequence of αi2 and therefore do not specify interaction with this receptor. We previously demonstrated that mutations of residues in three adjacent regions of the α subunit structure, the α2/β4, α3/β5, and α4/β6 loops (Noel et al., 1993), disrupt the ability of αs to activate adenylyl cyclase (Berlot and Bourne, 1992). Because the current study of receptor-mediated activation of αs mutants used a cAMP accumulation assay, we did not test 16 residues in these loops. Of the remaining 84 nonidentical residues, 61 were changed in small clusters to αi2 homologs or to alanines using 15 αs mutant constructs (Fig. 1).
Transient transfection assay for receptor-mediated activation of αs.
To test the abilities of mutant αs proteins to be activated by the β-adrenergic receptor, we measured receptor-dependent stimulation of cAMP synthesis after transient transfection ofcyc− S49 lymphoma cells (Bourne et al., 1975), which lack endogenous αs (Harris et al., 1985). Basal cAMP levels in cells transfected with 10–90 μg of vector containing αs varied linearly in proportion to the plasmid dose (Fig. 2A). Stimulation of these αs-transfected cells with the β-adrenergic agonist isoproterenol produced increased cAMP levels that also exhibited a linear relationship to the amount of transfected plasmid (Fig. 2A). We also determined receptor-independent cAMP accumulation by measuring basal cAMP levels in cells transfected with versions of the αs mutants in which Arg201 is mutated to cysteine. The αs containing this mutation, αsRC, exhibits constitutive activation due to inhibited GTPase activity (Landis et al., 1989). As with αs-transfected cells, basal cAMP levels incyc− cells transfected with 10–90 μg of vector containing αsRC varied linearly in proportion to the plasmid dose (Fig. 2B).
Transient transfection assay for receptor-mediated activation of αs. A, cAMP accumulation incyc− cells electroporated with the indicated doses of vector containing αs. cAMP levels were measured in the presence and absence of 0.1 mmisoproterenol. B, Receptor-independent cAMP accumulation incyc− cells electroporated with the indicated doses of vector containing αsRC. For the 0-μg points, 30 μg of vector was used. cAMP levels in [3H]adenine-labeled cells were determined as described in Experimental Procedures. All values represent the mean ± standard error of three independent experiments.
We initially measured receptor-independent cAMP accumulation due to the αsRC mutants using 30 μg of plasmid. At this plasmid dose, the activities of some of the αsRC mutants were reduced compared with that of αsRC. The expression levels of transiently expressed αs proteins incyc− cells were not high enough to be detected using an immunoblot but could be determined in transiently transfected HEK 293 cells. αsRC mutants with reduced activities in cyc− cells had similarly reduced activities in HEK 293 cells. The activities of these αsRC mutants directly correlated with their expression levels as determined by immunoblotting of HEK 293 cell membranes (data not shown). Because the activities of both αs and αsRC incyc− cells were directly proportional to the amount of transfected plasmid (Fig. 2), it was possible to normalize the expression levels of these αs and αsRC mutant constructs to that of αs and αsRC by transfecting with increased amounts of plasmid.
To compare receptor-dependent activation of αsmutants with that of αs, we identified plasmid doses for which the activities of the αsRC mutants were similar to those for 30 μg of the αsRC-containing plasmid (Fig.3B). At these plasmid doses, we compared receptor-dependent cAMP accumulation due to the corresponding αs mutants with that for 30 μg of the αs-containing plasmid (Fig. 3A). An assumption underlying this normalization procedure is that the substitutions in the mutant constructs have similar effects on the expression levels of αs and αsRC. This assumption is supported by the observation that the basal activity of αsRC is ∼10-fold greater than that of αs (Figs. 2 and 3) and the basal activities of each of the αsRC mutants also are ∼10-fold greater than those of the corresponding αsmutants (Fig. 3).
Receptor-mediated activation of mutant αs proteins. A, cAMP accumulation incyc− cells containing the indicated mutants in the αs context. Cells were electroporated with 20 μg of vector containing αs (2), 30 μg of vector alone, 30 μg of vector containing αs and αs (3 and 10–15), 60 μg of vector containing αs (5), 90 μg of vector containing αs (1, 4, 7, and 8), 120 μg of vector containing αs (9), and 180 μg of vector containing αs (6).Dark gray, cAMP values from unstimulated cells.Light gray, cAMP values from cells stimulated with 0.1 mm isoproterenol. B, Receptor-independent cAMP accumulation in cyc− cells containing the indicated mutants in the αsRC context. For each mutant, the same amount of plasmid was used as is indicated in A for the corresponding αs mutant. cAMP levels in [3H]adenine-labeled cells were determined as described in Experimental Procedures. All values represent the mean ± standard error of at least three independent experiments.
Of the four regions of sequence in which mutations disrupted receptor-mediated activation (see below), all except one of them (region 2) included at least one cluster of residues that did not decrease expression level. Studies using stably transfected cells confirmed that receptor-mediated activation of the region 2 mutant was decreased (see below). Thus, although a 9-fold range of plasmid doses was used for the transient transfection assay, our conclusions do not depend on the activities in this assay of the constructs with low expression levels.
Receptor-mediated activation of αs mutant constructs.
Receptor-stimulated cAMP accumulation due to 9 of the 15 αs mutant constructs was similar to that of αs (Fig. 3A). The other 6 constructs produced reduced receptor-dependent increases in cAMP levels and delineated four regions of sequence containing seven or fewer αs mutations that disrupt the ability of αs to be activated by the β-adrenergic receptor (Fig. 1). Region 1, defined by αs(1) and αs(2), contains V247, S250, S252, N254, M255, I257, and R258. Region 2, defined by αs(6), contains G304, K305, and K307-Y311. Region 3, defined by αs(11) and αs(12), contains V367, E370, and I372-R374. Region 4, defined by αs(15), contains R389-E392 and L394.
Because αs(6) in region 2 was poorly expressed in transiently transfected cells, we established lines ofcyc− cells stably transfected with this αs mutant construct to investigate further the role of region 2 in receptor-mediated activation. As expected from the results of the transient transfection assay, the expression levels obtained in αs(6)-expressing lines, as determined by immunoblotting, were lower than those in αs-expressing lines. In addition, a defect in receptor-mediated activation was seen in that isoproterenol-stimulated adenylyl cyclase activity was reduced relative to that stimulated by GTPγS (Grishina G and Berlot CH, unpublished observations). Thus, the mutations in region 2 impair receptor-dependent activation but also seem to decrease the stability of αs.
Mapping of mutations that block receptor-mediated activation onto the structure of a heterotrimeric G protein.
Because receptors interact with αβγ heterotrimers, we mapped the αs residues in which mutations disrupted activation by the β-adrenergic receptor onto the x-ray crystal structure of an αt/αi1chimera complexed with βtγt (Lambrightet al., 1996) to visualize their positions in three dimensions (Fig. 4). The recently solved structure of αs·GTPγS (Sunahara et al., 1997) is very similar to the structures of αt·GTPγS (Noel et al., 1993) and αi1·GTPγS (Coleman et al., 1994), indicating that the structure of αt/αi1βtγtis a good model for the Gs heterotrimer. Structural features unique to αs that are relevant to the mutations that blocked receptor-mediated activation are discussed. Some of the mutations that blocked receptor-mediated activation map onto solvent-exposed residues that could potentially interact directly with the receptor, whereas others map onto residues that are buried and are more likely to mediate nucleotide exchange by propagating conformational changes within αs.
Mapping of αs mutations that disrupt receptor-mediated activation onto the structure of an αβγ heterotrimer. A, View of the α subunit. The βγ subunits have been omitted for clarity. Left of GDP (yellow), helical domain. Right, GTPase domain. Blue spheres, residues mutated in αs(1) in region 1.Red spheres, residues mutated in αs(2) in region 1. Green spheres, residues mutated in αs(6) in region 2. Orange spheres, buried residues in region 3 mutated in αs(16). Residues mutated in αs(15) in region 4 are not seen in this structure, but the most carboxyl-terminal αt/αi1 residue visualized (residue 343 at the end of α5) is a magenta sphere. The amino-terminal portion of the αsischimera, consisting of αs residues (light blue), extends from the amino terminus to the end of α1. The middle portion of the chimera, consisting of αi2 residues (pink), extends from the αs/αi2 junction (s/i) to the αi2/αs junction (i/s) at the end of α2. The carboxyl-terminal portion of the chimera consists of αs residues (light blue).Gold, switches I–III. Numberson the spheres, αs residue numbers. B, View of the heterotrimer. The model is rotated 180 degrees around the vertical axis relative to the view in A. Gold, βstrands in the β subunit. Yellow, amino-terminal α helix and the connecting loops. White, γsubunit. Other colors, as in A. X-ray coordinates of an αt/αi1 chimera complexed with βtγt are from Lambright et al. (1996). These figures were drawn using MidasPlus, developed by the Computer Graphics Laboratory at UCSF.
The residues in regions 1 and 2 are located in the GTPase domain at the interface between the GTPase and helical domains. Region 1 extends from the middle of β4 to the middle of the β4/α3 loop and overlaps with switch III, which assumes different conformations in the structures of GTPγS-bound and GDP-bound α subunits (Noel et al., 1993; Coleman et al., 1994; Lambright et al., 1994; Mixon et al., 1995). Region 2 is located in the adjacent αG/α4 loop. Residues in the amino-terminal part of region 1, defined by αs(1), are buried within the interior of the molecule, making contacts with other residues in the GTPase domain. In the carboxyl-terminal part of region 1, the side chains of N254, M255, and R258 in αs(2) are in close proximity to residues in the helical domain. The residues in region 2 immediately precede a 12-residue insertion of sequence in αs relative to αt and αi. However, comparison of the structures of αs·GTPγS (Sunahara et al., 1997), αt·GTPγS (Noel et al., 1993), and αi1·GTPγS (Colemanet al., 1994) reveals that the orientation of region 2 with respect to the helical domain is the same in all of the α subunits. In the structure of αs·GTPγS (Sunaharaet al., 1997), the 12-residue insertion is located farther from the interface than the location of region 2 (Fig. 4A, to the right of Region 2). Of the region 2 residues, K305 and Y311 are closest to the interface, and all of the residues except for I308 are surface-exposed.
Region 3 is located near the guanine nucleotide binding pocket and includes residues in the β6/α5 loop and the beginning of α5. Three of the residues, E370, R373, and R374, are solvent-exposed, whereas two, V367 and I372 (shown in Fig. 4), are buried. V367 contacts the guanine ring of the bound nucleotide (Sunahara et al., 1997). The residues in region 4, located at the extreme carboxyl terminus, were not visualized in the αt/αi1βtγtstructure (Lambright et al., 1996) and occupy different positions in the structures in which they were visualized. In the structure of αt·GTPγS (Noel et al., 1993), this region contacts the α2/β4 loop, whereas in the structures of αs·GTPγS (Sunaharaet al., 1997), αi1·GDP·AlF4−·RGS4 (Tesmer et al., 1997), and αi1β1γ2 (Wall et al., 1995), the extreme carboxyl terminus is distant from the rest of the α subunit. Region 4 is linked to region 3 by the α5 helix.
Mutations of buried, but not surface-exposed, residues in region 3 disrupt receptor-mediated activation.
To determine the role of the buried and surface-exposed residues in region 3, we mutated separately each class of residues. We found that substitution of the buried residues with the homologous αi2 residues in αs(16) specifically reduced receptor-mediated increases in cAMP production, whereas substitution of the surface-exposed residues with alanine residues in αs(17) had no effect (Fig.5). Therefore, this region does not seem to be a receptor contact site but instead probably is important for transmitting the receptor signal to the bound GDP. Of the two residues mutated in αs(16), V367 is located in the β6/α5 loop, presumably in contact with the GDP, whereas I372 is near the beginning of α5. Because each of these residues was tested separately in αs(11) and αs(12), both regions seem to be involved in responding to receptor stimulation.
Defects caused by αi2 substitutions in regions 1 and 2 can be complemented by replacing the helical domain with αi2 residues.
Based on the locations of the region 1 and 2 residues at the interface between the GTPase and helical domains, we hypothesized that they mediate receptor-dependent activation via interactions with helical domain residues. Mismatches across the domain interface between αi2 residues in the GTPase domain and αs residues in the helical domain thus would be the cause of the reduced abilities of the region 1 and 2 αs mutant constructs to be activated by receptor stimulation. According to this hypothesis, replacing the helical domain of αs with that of αi2 would be expected to reverse the defect in receptor-mediated activation caused by the αi2for αs substitutions in regions 1 and 2. To test this prediction, we produced an α subunit chimera, αsis, in which αi2homologs are substituted for αs residues 62–235, extending from the end of α1 to the end of α2 (see Fig.4). The helical domain of this chimera is composed of αi2 residues, and the GTPase domain contains αs residues.
The properties of αsis were similar to those of αs, with the exception of an elevation in basal activity. The expression levels of αsisRC (the GTPase-inhibited version of αsis) and αsRC in membranes of transiently transfected HEK 293 cells were similar (data not shown). Receptor-independent cAMP accumulation due to αsisRC also was similar to that of αsRC (compare Fig.6B with Fig. 2B). Stimulation of αsis-transfected cells with isoproterenol resulted in cAMP levels similar to those in stimulated αs-transfected cells (compare Fig. 6A with Fig.2A). However, in the absence of β-adrenergic receptor stimulation, αsis produced significantly elevated cAMP levels in transiently transfected cells relative to αs (compare Fig. 6A with Fig. 2A). Adenylyl cyclase assays on membranes of cyc− cells stably transfected with αs or αsis confirmed that at equivalent expression levels, the basal activity of αsis was elevated relative to that of αs (Grishina G and Berlot CH, unpublished observations).
Receptor-mediated activation of αsis. A, cAMP accumulation incyc− cells electroporated with the indicated doses of vector containing αsis. cAMP levels were measured in the presence and absence of 0.1 mmisoproterenol. B, Receptor-independent cAMP accumulation incyc− cells electroporated with the indicated doses of vector containing αsisRC. For the 0-μg points, 30 μg of vector was used. cAMP levels in [3H]adenine-labeled cells were determined as described in Experimental Procedures. Each value represents the mean ± standard error of three independent experiments.
To determine whether the activation defects of the region 1 and 2 αs mutant constructs, αs(1), αs(2), and αs(6), could be due to mismatches between αs and αi2 residues across the domain interface, we determined the effects of introducing the αi2 substitutions of these mutant constructs into αsis. We also tested the effect of introducing the αs(2) and αs(6) mutations together, αs(2 + 6), in both αsand αsis. We first identified plasmid doses of the GTPase-inhibited RC versions of these constructs that produced receptor-independent cAMP stimulation in transiently transfectedcyc− cells comparable to that of 30 μg of αsRC and of αsisRC (Fig. 7B). Immunoblots of membranes from transiently transfected HEK 293 cells demonstrated that the expression levels of these constructs correlated with their activities (data not shown).
Receptor-mediated activation of αsisconstructs containing αi2 substitutions at the domain interface. A, cAMP accumulation in cyc−cells containing the indicated clusters of mutations in the αs or αsis context. Numbers in parentheses, construct numbers in Fig. 1. Cells were electroporated with 20 μg of vector containing αsis(2), 30 μg of vector alone and of vector containing αs and αsis, 40 μg of vector containing αsis(2 + 6), 60 μg of vector containing αsis(1) and αsis(6), and 180 μg of vector containing αs(2 + 6). Dark gray, cAMP values from unstimulated cells. Light gray, cAMP values from cells stimulated with 0.1 mm isoproterenol. B, Receptor-independent cAMP accumulation incyc− cells containing the indicated clusters of mutations in the αsRC or αsisRC context. For each mutant, the same amount of plasmid was used as is indicated in A for the corresponding αs or αsis mutant. cAMP levels in [3H]adenine-labeled cells were determined as described in Experimental Procedures. All values represent the mean ± standard error of at least three independent experiments.
The impairment of receptor-mediated activation caused by two of the three clusters of αi2 substitutions at the domain interface was reversed in the context of αsis (Fig. 7A). The isoproterenol-stimulated activities of αsis(2) and αsis(6) were similar to those of αs and αsis. Thus, matching αi2 residues across the domain interface seems to have corrected the decreased isoproterenol-stimulated activity of αs(2) and αs(6). Combining the αs(2) and αs(6) substitutions in αs(2 + 6) resulted in a larger decrease in isoproterenol-stimulated activity than was observed for αs(2) and αs(6), suggesting that the defects in αs caused by the two clusters of mutations are additive (compare Fig. 7 with Fig. 3). However, the isoproterenol-stimulated activity of αsis(2 + 6) was similar to those of αsis(2), αsis(6), and αs. In contrast, as was observed for αs(1), the isoproterenol-stimulated activity of αsis(1) was disrupted. The basal activities of αsis(2) and αsis(2+6) were not significantly different from that of αsis (p < 0.05), but they were consistently intermediate between those of αs and αsis, suggesting that matching αi2 residues across the domain interface partially corrects the elevated basal activity of αsis.
Discussion
In the analysis of αs mutants presented here, we identified four regions of sequence that mediate activation by the β-adrenergic receptor. Two of these regions, the extreme carboxyl terminus (region 4) and the β6/α5 loop (region 3), have been implicated previously in receptor/G protein interactions. However, current models of receptor-mediated activation based on these regions have not addressed the issue of how the bound GDP is released from its buried position between the GTPase and helical domains. The effects of mutations in the two other αs regions identified in the current study (regions 1 and 2), located in the β4/α3 and αG/α4 loops at the interface between these domains, suggest that interactions between residues across the domain interface may play a role in the response to receptor stimulation.
Region 4 (the extreme carboxyl terminus) is the most likely candidate for being a receptor contact site. This region was not visualized in the αt/αi1βtγtstructure (Lambright et al., 1996) and occupies different positions in the structures of αt·GTPγS (Noel et al., 1993), αs·GTPγS (Sunahara et al., 1997), αi1·GDP·AlF4−·RGS4 (Tesmer et al., 1997), and αi1β1γ2 (Wall et al., 1995), which could indicate a high degree of mobility. In the structure of αt·GTPγS (Noel et al., 1993), this region contacts the α2/β4 loop, which interacts with the β subunit in the heterotrimer structures (Wall et al., 1995;Lambright et al., 1996). Because receptors interact only with α subunits that are associated with βγ (Fung, 1983), the carboxyl terminus of the α subunit may be located near the interface between α2/β4 and the β subunit in the heterotrimer/receptor complex, so that receptors interact with both α and β in this region.
Residues at the extreme carboxyl termini of α subunits are sometimes, but not always, sufficient to confer receptor specificity (Conklinet al., 1993, 1996; Lee et al., 1995). For instance, although both the V2 vasopressin and β2-adrenergic receptors couple to αs, only the V2vasopressin receptor can stimulate a chimeric α subunit in which the last five residues of αq are replaced with the corresponding residues of αs (Conklin et al., 1996). Differences in the βγ specificities of α subunits and receptors may dictate whether carboxyl-terminal swaps between α subunits will switch receptor specificities. Also, other regions of αs, not identified in the current study, probably are receptor contact sites. For example, synthetic peptides corresponding to αs residues 354–372 and 384–394, which extend from the α4/β6 loop to the beginning of α5 and from the end of α5 to the α subunit carboxyl terminus, respectively, can mimic the effects of αs on the β2-adrenergic receptor (Rasenick et al., 1994). Our study of αs, using homologous substitutions, does not rule out these regions as being receptor contact sites because residues that are identical in αs and αi2 were not changed. Substitutions of surface-exposed αsresidues with alanine residues would be required to determine whether they are receptor contact sites. Furthermore, evaluation of receptor-mediated activation of αs mutants with substitutions in regions important for adenylyl cyclase activation, which includes residues in the α4/β6 loop (Berlot and Bourne, 1992), would require an assay that is independent of adenylyl cyclase activation.
The results of our study suggest that the role of region 3 (the β6/α5 loop and the amino terminal portion of the α5 helix) in αs is to regulate GDP affinity because substitutions of buried residues near the GDP, but not of adjacent solvent-exposed residues, disrupt receptor-mediated activation. GDP release by αs constructs with these substitutions is not entirely blocked because receptor-independent activation of adenylyl cyclase by the GTPase-inhibited versions of the constructs is normal. Of the two region 3 residues in which substitutions decreased receptor-mediated activation, V367, which contacts the guanine ring of the GDP, also is near the helical domain. However, the importance of this proximity to the helical domain is unclear because the helical domain residues closest to V367 are conserved among α subunits and are superimposable on the structures of αs, αi1, and αt. Substitutions of other buried residues in this region have been shown to affect guanine nucleotide exchange in several α subunits. The replacement of serine by alanine at position 366 of αs causes constitutive activation by increasing GDP release (Iiri et al., 1994). Also, the substitution of alanine for cysteine at position 325 of αo decreases affinity for GDP (Thomas et al., 1993). A mutagenesis study of αt(Onrust et al., 1997) identified two residues in the β6/α5 loop in which mutations reduced receptor-mediated activation, of which one, T323, corresponds to V367 in αs.
Receptors activate G proteins by promoting GTP binding, which involves both accelerating GDP release and increasing the relative affinity for GTP compared with GDP (Florio and Sternweis, 1989). Our results suggest that the role of regions 1 and 2 in this process may be to relay conformational changes initiated by receptor/G protein binding across the domain interface rather than to specify interaction with the β-adrenergic receptor. Altered contacts between the two domains might facilitate nucleotide exchange by opening the cleft in which GDP is buried. This hypothesis is based on our observation that replacing αs residues with αi2homologs on the GTPase side of the domain interface impairs receptor-mediated activation of αs, but not of a chimera, αsis, in which the helical domain consists of αi2 residues. Thus, matching αi2 residues on both sides of the domain interface of αs seems to restore receptor-initiated activation. As is sometimes the case with second-site suppressors of mutations, substitution of the helical domain of αi2 for that of αs corrects the defects of the region 1 and 2 mutations but does not, on its own, cause a defect in receptor-mediated activation.
In support of the concept of communication between regions 1 and 2 and the helical domain, comparison of the structures of GTPγS-bound and GDP-bound α subunits (Noel et al., 1993; Coleman et al., 1994; Lambright et al., 1994; Mixon et al., 1995) shows that conformational changes in switch III (located in region 1) are associated with changes in the positions of helical domain residues in the αD/αE loop, which it contacts in the GTPγS-bound form. Although it is not clear how receptor/G protein interaction might lead to conformational changes in switch III, communication between switches II and III, which are closer together in the structures of GTPγS-bound versus GDP-bound α subunits, could be involved. In the structure of αs·GTPγS (Sunahara et al., 1997), the side chain of the switch III residue E259 is hydrogen-bonded to the backbone nitrogen of the switch II residue R228, and the side chain of the switch II residue R231 is hydrogen-bonded to the side chain of E268 in the α3 helix, adjacent to switch III. Substitution of a histidine for R231 in switch II of αs leads to a decrease in receptor-mediated activation (Iiri et al., 1997).
Our results also suggest that interdomain interactions may regulate the basal activity of αs, because the basal activity of αsis is elevated. From the experiments presented here, we cannot determine whether the increased basal activity of αsis is due to increased nucleotide exchange or decreased GTPase activity. However, because the helical domain of αs, when expressed on its own and reconstituted with the GTPase domain of αs, stimulates GTPase activity and promotes binding of GTPγS (Markbyet al., 1993), interactions across the interface probably are important for both aspects of guanine nucleotide handling.
Other studies support the idea that interdomain communication plays a role in receptor-catalyzed G protein activation. A salt bridge interaction between D173 in the helical domain and K293 in the GTPase domain of αs is required for receptor-mediated activation (Codina and Birnbaumer, 1994). A mutation within region 1 that substitutes tryptophan for arginine at position 258 in αs, R258W, was found in a patient with pseudopseudohypoparathyroidism, and biochemical analysis revealed that although the mutant activated adenylyl cyclase normally in response to GTPγS, it exhibited attenuated responses to AlF4− and the β-adrenergic receptor (Warner DR and Weinstein LS, personal communication). Regions 1 and 2 may be important for receptor-mediated activation of other α subunits as well because a segment of α16 (residues 220–240), which overlaps with regions 1 and 2, contributes to the specificity of activation by the C5a receptor (Lee et al., 1995).
Our study suggests that complementary interactions between α subunit residues across the interface between the GTPase and helical domains play a role in facilitating receptor-mediated activation. Biochemical studies using purified αs mutant proteins with substitutions in these regions and cell lines that stably express these αs mutants will be required to more fully understand the nature of these intramolecular interactions and the mechanism by which receptor-catalyzed conformational changes lead to G protein activation.
Acknowledgments
We thank David Lambright and Paul Sigler for the coordinates of the αt/αi1 chimera complexed with βtγt, Stephen Sprang for the coordinates of αs·GTPγS complexed with the catalytic domains of adenylyl cyclase, Dennis Warner and Lee Weinstein for sharing data before publication, and Thomas Hynes, Rolando Medina, and Henry Bourne for helpful discussions and critical reading of the text.
Footnotes
-
Send reprint requests to: Dr. Catherine H. Berlot, Department of Cellular and Molecular Physiology, Yale University School of Medicine, 333 Cedar Street, New Haven, CT 06520-8026. E-mail:cathy_berlot{at}qm.yale.edu
-
This research was supported by National Institutes of Health Grant GM50369 to C.H.B.
- Abbreviations:
- GTPγS
- guanosine-5′-O-(3-thio)triphosphate
- HEPES
- 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid
- HEK
- human embryonic kidney
-
- Received November 25, 1997.
- Accepted February 18, 1998.
- The American Society for Pharmacology and Experimental Therapeutics
