Mutagenesis of the Conserved Residue Glu259 of Gsα Demonstrates the Importance of Interactions between Switches 2 and 3 for Activation*

We previously reported that substitution of Arg258 within the switch 3 region of Gsα impaired activation and increased basal GDP release due to loss of an interaction between the helical and GTPase domains (Warner, D. R., Weng, G., Yu, S., Matalon, R., and Weinstein, L. S. (1998) J Biol. Chem. 273, 23976–23983). The adjacent residue (Glu259) is strictly conserved in G protein α-subunits and is predicted to be important in activation. To determine the importance of Glu259, this residue was mutated to Ala (Gsα-E259A), Gln (Gsα-E259Q), Asp (Gsα-E259D), or Val (Gsα-E259V), and the properties of in vitrotranslation products were examined. The Gsα-E259V was studied because this mutation was identified in a patient with Albright hereditary osteodystrophy. S49 cyc reconstitution assays demonstrated that Gsα-E259D stimulated adenylyl cyclase normally in the presence of GTPγS but was less efficient with isoproterenol or AlF4 −. The other mutants had more severely impaired effector activation, particularly in response to AlF4 −. In trypsin protection assays, GTPγS was a more effective activator than AlF4 − for all mutants, with Gsα-E259D being the least severely impaired. For Gsα-E259D, the AlF4 −-induced activation defect was more pronounced at low Mg2+ concentrations. Gsα-E259D and Gsα-E259A purified fromEscherichia coli had normal rates of GDP release (as assessed by the rate GTPγS binding). However, for both mutants, the ability of AlF4 − to decrease the rate of GTPγS binding was impaired, suggesting that they bound AlF4 − more poorly. GTPγS bound to purified Gsα-E259D irreversibly in the presence of 1 mm free Mg2+, but dissociated readily at micromolar concentrations. Sucrose density gradient analysis ofin vitro translates demonstrated that all mutants except Gsα-E259V bind to βγ at 0 °C and were stable at higher temperatures. In the active conformation Glu259interacts with conserved residues in the switch 2 region that are important in maintaining both the active state and AlF4 − in the guanine nucleotide binding pocket. Although both Gsα Arg258 and Glu259 are critical for activation, the mechanisms by which these residues affect Gsα protein activation are distinct.

Heterotrimeric guanine nucleotide-binding proteins (G proteins) 1 couple heptahelical receptors to intracellular effectors and are composed of three subunits (␣, ␤, and ␥) (reviewed in Refs. [1][2][3]. The ␣-subunits, which are distinct for each G protein, bind guanine nucleotide and modulate the activity of specific downstream effectors. For G s , these include the stimulation of adenylyl cyclase and modulation of ion channels (4,5). In the inactive state, GDP-bound ␣-subunit is associated with a ␤␥-dimer. Upon receptor activation, the ␣-subunit undergoes a conformational change resulting in the exchange of GTP for GDP and dissociation from ␤␥. While GTP is bound, the ␣-subunit interacts with and regulates specific effectors. An intrinsic GTPase activity within the ␣-subunit hydrolyzes bound GTP to GDP, returning the G protein to the inactive state. Analogs of GTP, such as GTP␥S and GDP-AlF 4 Ϫ , lock the G protein in the active state.
X-ray crystal structures reveal that G protein ␣-subunits have two domains, a ras-like GTPase domain, which includes the regions for guanine nucleotide binding and effector interaction, and a helical domain, which may prevent release of GDP in the inactive state (6 -12). Comparison of the crystal structures of inactive (GDP-bound) and activated (GTP␥S-or AlF 4 Ϫ -bound) ␣-subunits demonstrates three regions (named switches 1, 2, and 3), the conformation of which changes upon switching from the inactive to active state. The movement of switches 1 and 2 is directly related to the presence of the ␥-phosphate group, whereas switch 3 has no direct contact with bound guanine nucleotide. Upon activation, switches 2 and 3 move toward each other, and the two regions form multiple interactions that presumably stabilize the active state (7,10). Switch 3 residues also make contacts with the helical domain that are important for high affinity guanine nucleotide binding (10,15). At least for transducin, this region may also be important in effector activation (13).
We have previously shown that substitutions of the switch 3 residue Arg 258 impairs activation by receptor or AlF 4 Ϫ (15). 2 The latter effect was the direct result of decreased GDP binding due to loss of contacts between the Arg 258 side chain and residues within the helical domain. The adjacent residue (Glu 259 ) is invariant in all known G protein ␣-subunits and is predicted to be important in activation, because it makes interactions with switch 2 residues in the active state (7,12). Moreover, this residue is mutated to a valine in a patient with Albright hereditary osteodystrophy (16). In the present report, we provide evidence that substitution of Glu 259 also leads to impaired activation, particularly by receptor or AlF 4 Ϫ . How-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ To whom correspondence should be addressed: Bldg. 49, Rm. 2A28, National Institutes of Health, Bethesda, MD 20892-4440. Tel.: 301-496-2007; Fax: 301-496-8244; E-mail: dwarner@helix.nih.gov. 1 The abbreviations used are: G protein, guanine nucleotide-binding protein; G s , stimulatory G protein; G s ␣, G s ␣-subunit; G s ␣-E259D, -E259A, -E259Q, and -E259V, G s ␣ mutant with Glu 259 substituted to aspartate, alanine, glutamine, and valine, respectively; AlF 4 Ϫ , mixture of 10 M AlCl 3 and 10 mM NaF; GTP␥S, guanosine-5Ј-O-(3-thiotriphosphate); WT, wild type. 2 All numbering is based on the G s ␣-1 sequence reported by Kozasa et al. (17). ever, impaired activation of these mutants by AlF 4 Ϫ is not the result of decreased GDP binding (as is the case for the Arg 258 mutants) but rather is the result of a decreased ability to bind the AlF 4 Ϫ moiety. The crystal structure of GTP␥S-bound G s ␣ reveals interactions between the acidic side chain of Glu 259 and basic residues within switch 2 that are important in maintaining the active state and in binding of AlF 4 Ϫ (12). Although adjacent switch 3 residues in G s ␣ (Arg 258 and Glu 259 ) are both critical for activation, the mechanisms by which mutations of these residues result in defective activation are distinct.

EXPERIMENTAL PROCEDURES
Construction of G s ␣ Plasmids and in Vitro Transcription/Translation-To generate G s ␣ Glu 259 mutants, polymerase chain reaction was performed as described previously (15) using linearized vector containing wild type G s ␣ cDNA as template. The upstream primer was 5Ј-G-ACAAAGTCAACTTCCACATGTTTGACGTGGGTGGCCAGCGCGAT-GAACG-3Ј, and the downstream mutagenic primers were as follows: 5Ј-GAGCCTCCTGCAGGCGGTTGGTCTGGTTGTCCACCCGGATGA-CCATGTTG-3Ј for E259V, 5Ј-GAGCCTCCTGCAGGCGGTTGGTCTG-GTTGTCCGCCCGGATGACCATGTTG-3Ј for E259A, 5Ј-GAGCCTCCT-GCAGGCGGTTGGTCTGGTTGTCCTGCCGGATGACCATGTTG-3Ј for E259Q, and 5Ј-GAGCCTCCTGCAGGCGGTTGGTCTGGTTGTCGTCC-CGGATGACCATGTTG-3Ј for E259D. Each polymerase chain reaction product was digested with HincII andSse8387I and ligated into the transcription vector pBluescript II SK (Stratagene, La Jolla, CA) that contained wild type human G s ␣ cDNA (splice variant G s ␣-1, Ref. 17) in which the same HincII-Sse8387I restriction fragment had been removed. Mutations were verified by DNA sequencing, and synthesis of full-length G s ␣ from each construct was confirmed by immune precipitation of in vitro translated products with RM antibody, directed against the carboxyl-terminal decapeptide of G s ␣ (18). In vitro transcription/translation was performed on G s ␣ plasmids as described previously (15,19) using the TNT-coupled transcription/translation system from Promega, with the exception that in most experiments, no RNase inhibitor was added.
Adenylyl Cyclase Assays-Wild type and mutant G s ␣ in vitro transcription/translation products (10 l of translation medium) were reconstituted into 25 g of purified S49 cyc plasma membranes and tested for stimulation of adenylyl cyclase in the presence of various agents as indicated in Table I (15,19,20). Reactions were incubated for 15 min at 30°C, and the amount of [ 32 P]cAMP produced was measured as described previously (21).
Trypsin Protection Assays-Limited trypsin digestion of in vitro translated G s ␣ was performed as described previously (15,19). Briefly, 1 l of in vitro translated [ 35 S]methionine-labeled G s ␣ was incubated in incubation buffer (20 mM HEPES, pH 8.0, 10 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol) with or without 100 M GTP␥S or 10 mM NaF/10 M AlCl 3 at various temperatures for 1 h and then digested with 200 g/ml tosyl-L-phenylyalanine chloromethyl ketone-trypsin for 5 min at 20°C. In some experiments, GDP was also included in the preincubation, and in other experiments the MgCl 2 concentration was varied. Reactions were terminated by boiling in Laemmli buffer. Digestion products were separated on 10% SDS-polyacrylamide gels, and the amount of 38-kDa protected fragment was measured by phosphorimag-ing. The percentage of protection is the signal in 38-kDa protected band divided by the signal in the undigested full-length G s ␣ band ϫ 100.
Sucrose Density Gradient Centrifugation-[ 35 S]Methionine-labeled G s ␣ was synthesized, and rate zonal centrifugation was performed on linear 5-20% sucrose gradients (200 l) as described previously (19,22). Gradients were prepared in 20 mM HEPES, pH 8.0, 1 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, 100 mM NaCl, 0.1% Lubrol-PX. Six-l fractions were obtained and analyzed by SDS-polyacrylamide gel electrophoresis, and the relative amount of G s ␣ in each fraction was quantified as described previously (19). To assess the ability of G s ␣ to bind to G␤␥, in vitro translation products were preincubated for 1 h at 0°C in the presence or absence of G␤␥ (20 g/ml) prior to centrifugation. In order to optimize separation between free ␣-subunit and heterotrimer, Ϫ . The smaller products in the left lane are due to initiation of protein synthesis at downstream methionine codons. Quantitation of trypsin protection assays for G s ␣-E259D is presented in Table II.
a 10 mM NaF, 10 M AlCl 3 , and 100 M GDP. b The results for G s ␣-WT are the same as previously published (15) because these were generated simultaneously with those obtained for the G s ␣-Arg 258 mutants. 0.15% (w/v) CHAPS was substituted for Lubrol-PX in the preincubations and gradients, and the samples were centrifuged at 120,000 rpm (627,000 ϫ g at the maximum radial distance from the center of rotation (R max ) in a TLA-120.2 rotor (Beckman). G␤␥ was isolated from bovine brain (23).
Expression and Purification of G s ␣ from Escherichia coli-Plasmid pQE60, containing the long form of bovine G s ␣ cDNA with a hexahistidine extension at the carboxyl terminus, was a generous gift of A. G. Gilman and R. K. Sunahara. The Glu 259 residue was mutated by site-directed mutagenesis using the Quickchange kit (Statagene). Each mutated cDNA was sequenced to confirm the presence of the desired mutation and to rule out polymerase chain reaction artifacts. After transformation into E. coli strain JM109, cultures were grown, G s ␣ expression was induced, and G s ␣ proteins were purified as described previ-ously (15,24), except that [GDP] was only 10 M in the storage buffer.
Guanine Nucleotide Binding Assays-Assays measuring the rate of binding of GTP␥S were performed as described previously (15,25). Briefly, 1-2 pmol of purified G s ␣ was incubated at 37°C in a final volume of 2 ml containing 1 M [ 35 S]GTP␥S (5,000 -10,000 cpm/pmol) in 25 mM HEPES, pH 8.0, 1 mM EDTA, 100 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol, and 0.01% Lubrol-PX with or without 10 mM NaF/10 M AlCl 3 . At various times, 50-l aliquots were removed and diluted with 2 ml of ice-cold stop solution (25 mM Tris-HCl, 100 mM NaCl, 25 mM MgCl 2 , and 100 M GTP) and maintained on ice until all samples were collected. Samples were then filtered under vacuum through nitrocellulose filters (Millipore) and washed twice with 10 ml of stop solution without GTP, and filters were dissolved in 10 ml of scintillation mixture. To determine the effect of Mg 2ϩ on the rate of GTP␥S dissociation, ϳ2.5 pmol of purified G s ␣ was loaded with [ 35 S]GTP␥S at 30°C for 45 min in the presence of various free Mg 2ϩ concentrations. After addition of 100 M cold GTP␥S, bound [ 35 S]GTP␥S was determined at various time points as described above. k off for GTP␥S dissociation was determined by fitting the data to the function y ϭ ae Ϫkt ϩ b using the software GraphPad Prism, version 2.01. Free Mg 2ϩ concentrations were calculated as described (26).

Substitution of G s ␣ Glu 259
Leads to Decreased Activation-G s ␣ Glu 259 substitution mutants were cloned into the transcription vector pBluescript, and the in vitro transcription/ translation products were compared with those of G s ␣-WT in various biochemical assays. We substituted Glu 259 with valine (G s ␣-E259V) because a mutation encoding this substitution has been identified in a patient with Albright hereditary osteodystrophy (16), a human disorder associated with heterozygous loss-of-function mutations of G s ␣ (27,28). Because the presence of an amino acid with a bulky and branched side chain (valine) may introduce nonspecific steric effects, we also generated and analyzed additional mutants in which Glu 259 was replaced by alanine (G s ␣-E259A), glutamine (G s ␣-E259Q), or aspartate (G s ␣-E259D). In G s ␣-E259A, the acidic side chain was removed, whereas in G s ␣-E259Q it is converted to a residue in which the carboxyl group is replaced by a neutral amide group. In G s ␣-E259D, the charge of the residue at position Glu 259 is Ϫ . G s ␣-WT (f and Ⅺ), G s ␣-E259A (q and E), and G s ␣-E259D (OE and ‚) were incubated with 1 M [ 35 S]GTP␥S (ϳ10,000 cpm/pmol) at 37°C for varying times, and the amount of bound GTP␥S was determined as described under "Experimental Procedures." For each G s ␣, each data point (with or without AlF 4 Ϫ ) was normalized to maximal binding at 10 min in the absence of AlF 4 Ϫ . Each data point is the mean Ϯ S.D. of triplicate determinations. This experiment was representative of three experiments. The B max values in the absence of AlF 4 Ϫ were as follows: G s ␣-WT, 3 pmol; G s ␣-E259A, 2 pmol; and G s ␣-E259D, 1 pmol. Ϫ -induced trypsin protection These data were obtained from experiments of the type presented in Fig. 1. The amount of the 38-kDa trypsin-stable G s ␣ fragment was determined by phosphorimaging, and for G s ␣-WT, it is expressed as a percent of undigested G s ␣ (mean Ϯ S.E.). No protection was observed when AlF 4 Ϫ and GTP␥S were excluded. Maximum trypsin protection has a theoretical limit of 71%, based on the removal of 2 of 7 total methionine residues by trypsin. For G s ␣-E259D, the data are expressed as percentage of wild type at each condition (mean Ϯ S.E.). The number of experiments performed for each condition is shown in the right column.
The percentage of protection of G s ␣-E259D was significantly less than that of G s ␣-WT at all conditions except at 30°C in the presence of maintained, but the length of the side chain is shortened by one methylene group. After reconstitution of translation products into purified S49 cyc membranes (which lack endogenous G s ␣), G s ␣-E259V had markedly decreased ability to stimulate adenylyl cyclase in the presence of GTP␥S, AlF 4 Ϫ , or activated receptor (isoproterenol ϩ GTP) (Table I). For G s ␣-E259A and -E259Q, the ability to stimulate adenylyl cyclase was moderately reduced in the presence of GTP␥S (ϳ40% of G s ␣-WT) and more markedly reduced in the presence of AlF 4 Ϫ or activated receptor. Stimulation of adenylyl cyclase by G s ␣-E259D was normal in the presence of GTP␥S but moderately reduced in the presence of AlF 4 Ϫ or activated receptor. Although the severity of the defect varied depending on which specific residue replaced Glu 259 , for each G s ␣-Glu 259 mutant, GTP␥S was the most effective activator and AlF 4 Ϫ the least effective activator. We next examined the ability of AlF 4 Ϫ or GTP␥S to protect each mutant from trypsin digestion, which measures the ability of each agent to bind to G s ␣ and induce the active conformation (29). In the inactive, GDP-bound state, two arginine residues within switch 2 (most likely Arg 228 and Arg 231 , based upon sequence homology with transducin) are sensitive to trypsin digestion, leading to the generation of low molecular weight fragments. When G s ␣ attains the active conformation, these residues are inaccessible to trypsin digestion (7) and therefore trypsinization of activated G s ␣ generates a partially protected 38-kDa product. G s ␣-WT was well protected by AlF 4 Ϫ or GTP␥S at temperatures up to 37°C (Fig. 1, Table II). At 30°C, G s ␣-E259V, -E259A, and -E259Q showed little protection by GTP␥S and no protection by AlF 4 Ϫ (Fig. 1). In contrast, both GTP␥S and AlF 4 Ϫ were able to protect G s ␣-E259D, with GTP␥S being a more efficient activator than AlF 4 Ϫ (Fig. 1, Table II). Consistent with the results of the cyc reconstitution assays, AlF 4 Ϫ was less effective than GTP␥S in protecting all G s ␣-E259 mutants from Because the G s ␣-E259D encoded the most subtle structural change and had the smallest activation defect, we studied the ability of this mutant to be protected by GTP␥S and AlF 4 Ϫ at various temperatures and in the presence or absence of excess GDP (Table II). For G s ␣-R258 mutants, the activation defect in the presence of AlF 4 Ϫ was more severe at higher temperatures and was reversible in the presence of excess GDP (15). Although raising the temperature had little effect on the ability of GTP␥S to protect G s ␣-E259D from trypsin protection, temperature had a profound effect on protection by AlF 4 Ϫ , being 85, 49, and 7% of G s ␣-WT at 25, 30, and 37°C, respectively. At 37°C, addition of 2 mM GDP was able to somewhat reverse the defect in activation by AlF 4 Ϫ , although not to the extent that it was able to reverse the defect in the G s ␣-R258 mutants (15). Interestingly, addition of GDP lowered the ability of AlF 4 Ϫ to protect G s ␣-E259D at 25 and 30°C (Table II). Although this effect was consistently observed, we have no good explanation for this observation.

Substitution of G s ␣ Glu 259 Has Little Effect on the Rate of GDP Release in the Basal State-The impaired activation of G s ␣-Glu 259 mutants by AlF 4
Ϫ could result from decreased affinity for AlF 4 Ϫ , decreased ability for the GDP-AlF 4 Ϫ complex to activate the mutant G s ␣s, or decreased ability of the mutant G s ␣s to maintain the GDP-bound state because GDP binding is a prerequisite for AlF 4 Ϫ binding and activation. For the G s ␣-Arg 258 mutants, impaired activation by AlF 4 Ϫ is primarily the result of impaired GDP binding (15). The inability of GDP to significantly reverse the AlF 4 Ϫ -induced activation defect in G s ␣-E259D suggests that this defect is not due to defective GDP binding.
To directly evaluate the rate of GDP release in the basal state, we expressed and purified bovine G s ␣-WT, -E259A, and -E259D, each with a carboxyl-terminal hexahistidine tag, from E. coli and examined the time course of GTP␥S binding. The rate of GTP␥S binding has been shown to be limited by the rate of GDP dissociation, and the experimentally determined values of these two rates are essentially identical (30,31). This assay has also been previously used as a measure of the GDP disso-ciation rate in other G s ␣ mutants (15,32). Substitution of Glu 259 had little effect on the rate of GDP release in the basal state, as the time course of GTP␥S binding at 37°C (in the absence of AlF 4 Ϫ ) is essentially identical for G s ␣-WT and -E259D, whereas the rate of GTP␥S binding for G s ␣-E259A is only increased minimally (Fig. 2). Consistent with these results, the rate of increase of trypsin protection of G s ␣-WT and -E259D in vitro translation products in the presence of GTP␥S was also identical (data not shown). These results demonstrate that unlike substitutions of Arg 258 , the rate of GDP release is not significantly altered by substitution of Glu 259 , and therefore the impaired activation by AlF 4 Ϫ is not primarily due to decreased GDP binding.
Substitution of G s ␣ Glu 259 Decreases AlF 4 Ϫ Binding-We next examined the ability of AlF 4 Ϫ to interact with mutant G s ␣s in the GDP-bound state to determine whether the decreased activation of G s ␣-E259 mutants by AlF 4 Ϫ is due to impaired AlF 4 Ϫ binding. It has been shown previously that the rate and extent of GTP␥S binding to G ␣-subunits is markedly reduced in the presence of AlF 4 Ϫ , presumably because the GDP-AlF 4 Ϫ complex bound to G␣ is more stable than GDP alone (8). Because G s ␣-WT, -E259D, and -E259A have similar rates of GTP␥S binding in the absence of AlF 4 Ϫ , the time course of GTP␥S binding in the presence of AlF 4 Ϫ should reflect the ability of each form of G s ␣ to interact with AlF 4 Ϫ . Similar to previously reported observations (8), the rate and extent of GTP␥S binding to G s ␣-WT was markedly reduced in the presence of AlF 4 Ϫ (Fig. 2). In contrast, AlF 4 Ϫ only partially reduced the rate and extent of GTP␥S binding to G s ␣-E259D and had a minimal effect on the GTP␥S binding curve for G s ␣-E259A (Fig.  2). These results are consistent with the results of adenylyl cyclase and trypsin protection assays, which demonstrate that AlF 4 Ϫ -induced activation is severely impaired in G s ␣-E259A but only partially impaired in G s ␣-E259D and suggest that the decreased ability of AlF 4 Ϫ to activate G s ␣-E259 mutants is primarily due to decreased ability of the mutants to maintain AlF 4 Ϫ in the guanine nucleotide binding pocket.

Effect of Mg 2ϩ Concentration on Activation by AlF 4
Ϫ and GTP␥S Binding-Substitution of G s ␣ Arg 231 , a residue in switch 2 that interacts with switch 3 residues in the active state, leads to a defect in activation by AlF 4 Ϫ that is more pronounced at low Mg 2ϩ concentrations (33). We therefore examined the effect of varying Mg 2ϩ concentration on the ability of AlF 4 Ϫ to protect G s ␣-E259D from trypsin digestion. In the trypsin protection experiments shown in Fig. 1 and Table II, the MgCl 2 concentration was 10 mM (ϳ9 mM free Mg 2ϩ ). Lowering the MgCl 2 concentration to 2 mM (ϳ1 mM free Mg 2ϩ ) had no effect on the ability of AlF 4 Ϫ to protect G s ␣-WT at 30°C (Fig.  3). In contrast, lowering the MgCl 2 concentration below 8 mM (ϳ7 mM free Mg 2ϩ ) further impaired the ability of AlF 4 Ϫ to protect G s ␣-E259D in a concentration-dependent manner. Increasing the MgCl 2 concentration up to 100 mM did not reverse the defect at 37°C (data not shown). These results are similar to those observed for the G s ␣-R231 mutant (33) and demonstrate that, like this mutant, the GDP-AlF 4 Ϫ -bound form of G s ␣-E259D has a lower apparent affinity for Mg 2ϩ than G s ␣-WT.
We next examined the effect of lowering the Mg 2ϩ concentration on the dissociation of GTP␥S from G s ␣-E259D to determine whether or not the Mg 2ϩ dependence was specific for the GDP-AlF 4 Ϫ -bound form. The apparent K d of GTP␥S-G s ␣-WT for Mg 2ϩ is very low (5-10 nM), and binding of GTP␥S is essentially irreversible in the presence of micromolar concentrations of Mg 2ϩ (34). Consistent with previously published results (34), no dissociation of GTP␥S from G s ␣-WT was observed at free Mg 2ϩ concentrations of 30 M or higher (Fig. 4 and data not shown), although GTP␥S dissociated rapidly (k off ϭ 2.5 min Ϫ1 ) in the absence of Mg 2ϩ (5 mM EDTA). For G s ␣-E259D, GTP␥S binding was essentially irreversible in the presence of 1 mM free Mg 2ϩ , but in contrast to G s ␣-WT, GTP␥S clearly dissociated from G s ␣-E259D in the presence of 30 M free Mg 2ϩ (Fig.  4, k off ϭ 0.05 min Ϫ1 ). Dissociation of GTP␥S from G s ␣-E259D (k off ϭ 3.7 min Ϫ1 ) was similar to that of G s ␣-WT in the absence of Mg 2ϩ (5 mM EDTA). Therefore, like GDP-AlF 4 Ϫ -G s ␣-E259D, GTP␥S-G s ␣-E259D appears to have decreased affinity for Mg 2ϩ , although the defects are apparent in the millimolar range for the former and micromolar range for the latter.
In contrast to G s ␣-E259D, there is a slow rate of dissociation of GTP␥S from G s ␣-R231H in the presence of high Mg 2ϩ concentrations (33). Another G s ␣ mutant (G s ␣-G226A) also displays an abnormally high apparent K d for Mg 2ϩ to prevent GTP␥S dissociation (34). Similar to G s ␣-E259D, GTP␥S dissociates from G s ␣-G226A in the presence of micromolar concentrations of Mg 2ϩ . There is also considerable dissociation of GTP␥S from G s ␣-G226A even in the presence of maximal (millimolar) concentrations of Mg 2ϩ , because this mutant cannot FIG. 5. Sucrose density gradient centrifugation of G s ␣ in vitro translation products. A, [ 35 S]methionine-labeled in vitro translates of both G s ␣-WT and -E259 mutants were preincubated for 1 h at 0°C (E) or 30°C (q) and subjected to sucrose density gradient centrifugation as described under "Experimental Procedures." Fractions (6 l each) were collected, and odd-numbered fractions were analyzed by SDS-polyacrylamide gel electrophoresis and phosphorimaging (15,19). The data are expressed as the percentage of total G s ␣ present in each fraction. Fraction 1 represents the top of the gradient. The position and S value of standard proteins are indicated at the top of the G s ␣-WT gradients. B, sucrose density gradient profiles of G s ␣-WT (q), G s ␣-E259A (f),G s ␣-E259Q (OE),G s ␣-E259D (Ⅺ), and G s ␣-E259V (‚) after preincubation for 1 h at 0°C in the presence of purified bovine brain G␤␥ (20 g/ml). The profile for G s ␣-WT in the absence of G␤␥ is also shown (E). All G s ␣-E259 (except G s ␣-E259V) mutants held at 0°C in the absence of G␤␥ had sucrose density gradient profiles similar to that of G s ␣-WT (data not shown). G s ␣-E259V had a somewhat broader peak at 0°C that was unaltered in the presence of G␤␥. Conditions were modified to optimize separation between free ␣-subunit and heterotrimer as outlined under "Experimental Procedures." Similar results were obtained with the detergent octyl-␤-glucoside (0.3% w/v).
attain the active conformation that stabilizes the Mg 2ϩ -GTP␥S complex. The ability of G s ␣-E259D to irreversibly bind GTP␥S in the presence of 1 mM Mg 2ϩ suggests that this mutant can attain the active conformation necessary to stabilize Mg 2ϩ -GTP␥S, consistent with the results obtained in the adenylyl cyclase and trypsin protection assays (Table I and Fig. 1).
G s ␣-E259Q, E259A, and E259D, but not G s ␣-E259V, Maintain Normal Overall Conformation and G␤␥ Interaction-We examined the ability of each G s ␣-E259 mutant to interact with ␤␥ by subjecting in vitro translates to sucrose density gradient centrifugation in the presence or absence of purified bovine brain ␤␥. We previously showed that G s ␣ has a sedimentation coefficient of ϳ3.7 S (15,19). When in vitro translates of each G s ␣-E259 mutant was held on ice, the gradient profiles of all mutants were virtually the same as G s ␣-WT and consistent with the overall proper conformation (sedimentation coefficient, ϳ3.7 S) (Fig. 5A). When preincubated on ice with purified bovine brain ␤␥, G s ␣-WT, -E259Q, -E259A, and -E259D formed heterotrimers, as demonstrated by significant shifting of the peak toward the bottom of the gradient (Fig. 5B). In contrast, ␤␥ had no effect on the sedimentation profile of G s ␣-E259V, indicating that this mutant does not interact with ␤␥. After preincubation at 30°C, gradient profiles demonstrate that all mutants except G s ␣-E259V maintain an the normal 3.7 S conformation, whereas for G s ␣-E259V, the majority of the protein is a higher S value material and is presumably denatured (19). Therefore, the valine substitution probably alters the overall conformation and stability of the protein due to nonspecific steric effects of its bulky hydrophobic side chain. In contrast, the activation defect in G s ␣-E259A, E259Q, and E259D is not secondary to defects in thermostability or ␤␥ binding. DISCUSSION We previously reported that substitution of the G s ␣ switch 3 residue Arg 258 leads to impaired activation in the presence of AlF 4 Ϫ or activated receptor (isoproterenol ϩ GTP) but normal activation in the presence of GTP␥S (15). The impaired activation by AlF 4 Ϫ was reversible in the presence of excess GDP, and further characterization demonstrated a defect in GDP binding, presumably due to loss of direct contact between Arg 258 and a residue(s) in the helical domain that would open the cleft through which guanine nucleotide must exit. In this study, we examined the effect of substituting the adjacent switch 3 residue (Glu 259 ) on G s ␣ function for the following reasons: 1) this residue is strictly conserved among G protein ␣-subunits and therefore might have an important role in the biochemical function of these proteins; 2) upon activation, the Glu 259 side chain interacts with several residues in the switch 2 region (7,12) and therefore substitutions of this residue might be predicted to directly impair G protein activation; 3) this G s ␣ residue is mutated to a valine in a patient with Albright hereditary osteodystrophy (16), a human disorder associated with heterozygous inactivating mutations within the G s ␣ gene (27,28).
Substitution of G s ␣ Glu 259 to valine had a marked effect on the conformation and stability of the protein. This mutant was unable to interact with ␤␥, even though Glu 259 is not within the ␤␥ interaction site (11). This mutant was also more thermolabile. Presumably, the presence of a bulky and branched side chain provided by valine introduces nonspecific steric effects that severely affect the conformation and stability of the protein. We would predict that the primary biochemical defect in the patient harboring this mutation is lack of expression of G s ␣-E259V in the membrane at physiological temperatures, similar to what is observed in other patients with mutants encoding unstable forms of G s ␣ protein (15,19,32).
In order to determine whether residue Glu 259 is critical in maintaining either the basal or activated state, we generated mutants with more subtle alterations of Glu 259 side chain. The most subtle mutation was G s ␣-E259D, in which the charge of the residue is maintained but the length of the side chain is shortened by one methylene group. We also made two mutants in which the side chain was either removed (G s ␣-E259A) or converted from an acidic to neutral amino acid (G s ␣-E259Q). In all three of these mutants, the overall conformation and stability, as well as the ability to interact with ␤␥, was maintained, as determined by sucrose density gradient experiments. Based upon adenylyl cyclase and trypsin protection assays, activation of G s ␣-E259D by GTP␥S was normal, demonstrating that this mutant has not lost its intrinsic ability to attain the active conformation and activate adenylyl cyclase. However, this mutant had decreased ability of to be activated by AlF 4 Ϫ or recep- 6. Crystal structure of G s ␣-GTP␥S. Detailed view of interactions between Glu 259 in switch 3, Glu 268 in ␣3, and residues in switch 2 and between Gly 226 and the ␥ phosphate of GTP␥S. Hydrogen bonds are shown as dotted lines. The atom coloring scheme is as follows: black, carbon; red, oxygen; blue, nitrogen; and yellow, sulfur. Mg 2ϩ and water molecules are shown as magenta and cyan spheres, respectively. The figure was generated with MOLSCRIPT (39) and rendered with RASTER3D (37) using coordinates for the short form of bovine G s ␣-GTP␥S (Protein Data Bank accession code 1AZT (12)), although the numbering on the figure corresponds to the long form of G s ␣ (17). This view is similar to that previously shown for transducin (6).
tor. G s ␣-E259Q and -E259A showed a more severe phenotype, with decreased activation in the presence of all agents. In all three mutants, GTP␥S was the most efficient activator whereas AlF 4 Ϫ was the least efficient. Mutation of the analogous residue in transducin (Glu 232 ) to leucine had no effect on the ability of the G protein to interact with ␤␥ or its receptor (rhodopsin), but it did appear to decrease the ability of GTP␥S to mediate trypsin protection and effector activation (13).
One possible mechanism for impaired activation by AlF 4 Ϫ is decreased ability to maintain the GDP-bound state, because binding of GDP is a prerequisite for AlF 4 Ϫ binding and activation. This is the primary mechanism by which substitutions of G s ␣ Arg 258 lead to impaired activation by AlF 4 Ϫ (15). However, the ability of G s ␣-E259 mutants to maintain the GDP-bound state was similar to that of G s ␣-WT, as demonstrated by both G s ␣-E259A and -E259D having a rate of GDP release that was similar to G s ␣-WT, as well as an inability for excess GDP to significantly reverse the AlF 4 Ϫ -induced activation defect. Consistent with normal guanine nucleotide binding, both G s ␣-E259A and -E259D were thermostable. Binding of AlF 4 Ϫ to the GDP-bound ␣-subunit results in formation of a stable and activated GDP-AlF 4 Ϫ -protein complex that mimics the transition state of the GTPase reaction and will slow the rate of GTP␥S binding, probably by inhibiting GDP release (8). The ability of AlF 4 Ϫ to inhibit the rate and extent of GTP␥S binding to both G s ␣-E259A and -E259D was significantly reduced, suggesting that in these mutants the activation defect in response to AlF 4 Ϫ is due at least in part to impaired AlF 4 Ϫ binding. The fact that the activation defect is greater for AlF 4 Ϫ than GTP␥S suggests that mutation of Glu 259 has a more dramatic effect on stabilizing the transition (AlF 4 Ϫ -bound) state than the activated (GTP␥S-bound) state.
It is of interest that the biochemical phenotype of our G s ␣-Glu 259 mutants is quite similar to that previously described for another G s ␣ mutant present in a patient with Albright hereditary osteodystrophy, in which the switch 2 residue Arg 231 is mutated to histidine (G s ␣-R231H) (33,35). Similar to the G s ␣-Glu 259 mutants, this mutation leads to normal GTP␥S-mediated but decreased AlF 4 Ϫ -and receptor-mediated activation. Moreover, similar to the G s ␣-R231H mutant, the AlF 4 Ϫ -induced activation defect in G s ␣-E259D was more pronounced at low Mg 2ϩ concentrations (33). This is not surprising, based upon mutual interactions between Glu 259 and Arg 231 present in the active (GTP␥S-bound) conformation of G s ␣ (Fig. 6). Upon activation, interactions between switches 2 and 3 stabilize the GTP-bound form of the G protein. Specifically, Arg 231 in switch 2 interacts with Glu 259 in switch 3 through a water molecule and directly with Glu 268 in the ␣3 helix (Fig. 6). Conversely, Glu 259 interacts with two basic switch 2 residues, Arg 228 and Arg 231 . Both Arg 231 and Glu 259 interact with Gly 226 , a residue that is critical for both AlF 4 Ϫ binding (36) and conformational switching of switch 2 upon binding of GTP or AlF 4 Ϫ (29). Therefore, the impaired activation and AlF 4 Ϫ binding observed in G s ␣-Glu 259 mutants might be the direct result of loss of contacts with Gly 226 . Loss of these contacts may also result in the apparent decreased affinity of G s ␣-Glu 259 and R231H mutants for Mg 2ϩ because mutation of Gly 226 to alanine also lowers the apparent affinity of G s ␣ for Mg 2ϩ (34).
Mutation of Glu 259 leads to a subtle defect in receptor-mediated activation (at least when compared with activation by GTP␥S). G s ␣-E259 mutants are able to bind ␤␥, and mutation of the analogous residue in transducin (Glu 232 ) has no effect on interactions with ␤␥ or receptor (13). It has been proposed that decreased receptor activation of G s ␣-R231H is due to a conditional defect in GTP binding, which is more pronounced in states in which guanine nucleotide binding is destabilized (such as interaction with activated receptor (33)). Our results are consistent with those observed with G s ␣-R231H and support this hypothesis.
In conclusion, this study provides further evidence for the role of switch 3 in the activation mechanism and demonstrates the importance of interactions between Glu 259 and switch 2 residues. Taken together with the prior studies on Arg 258 mutants (15,38), the present results demonstrate the importance of switch 3 in maintaining both the basal and active states.