A Dominant-negative Gα Mutant That Traps a Stable Rhodopsin-Gα-GTP-βγ Complex*

Residues comprising the guanine nucleotide-binding sites of the α subunits of heterotrimeric (large) G-proteins (Gα subunits), as well as the Ras-related (small) G-proteins, are highly conserved. This is especially the case for the phosphate-binding loop (P-loop) where both Gα subunits and Ras-related G-proteins have a conserved serine or threonine residue. Substitutions for this residue in Ras and related (small) G-proteins yield nucleotide-depleted, dominant-negative mutants. Here we have examined the consequences of changing the conserved serine residue in the P-loop to asparagine, within a chimeric Gα subunit (designated αT*) that is mainly comprised of the α subunit of the retinal G-protein transducin and a limited region from the α subunit of Gi1. The αT*(S43N) mutant exhibits a significantly higher rate of intrinsic GDP-GTP exchange compared with wild-type αT*, with light-activated rhodopsin (R*) causing only a moderate increase in the kinetics of nucleotide exchange on αT*(S43N). The αT*(S43N) mutant, when bound to either GDP or GTP, was able to significantly slow the rate of R*-catalyzed GDP-GTP exchange on wild-type αT*. Thus, GTP-bound αT*(S43N), as well as the GDP-bound mutant, is capable of forming a stable complex with R*. αT*(S43N) activated the cGMP phosphodiesterase (PDE) with a dose-response similar to wild-type αT*. Activation of the PDE by αT*(S43N) was unaffected if either R* or β1γ1 alone was present, whereas it was inhibited when R* and the β1γ1 subunit were added together. Overall, our studies suggest that the S43N substitution on αT* stabilizes an intermediate on the G-protein activation pathway consisting of an activated G-protein-coupled receptor, a GTP-bound Gα subunit, and the β1γ1 complex.

G protein-coupled receptors (GPCR) 2 are one of the largest families of membrane proteins and are involved in various physiological functions. In the past several years, significant advances have been made in the determination of structures at an atomic level of GPCRs (1,2), their cognate G-proteins, and their downstream targets (3). In addition, structures have been solved for complexes of G-proteins with their downstream tar-gets as well as their regulators (e.g. the regulators of G-protein signaling (RGS) proteins) (3). However, one of the central unresolved questions in this field involves the mechanism utilized by a GPCR to catalyze the release of GDP from its cognate G-protein (4,5).
The x-ray crystal structures of various G␣ subunits have shown that they are composed of two distinct domains: one that highly resembles the GTPase domain of the small G-protein Ras, and a second domain which is mainly ␣-helical in content and thus referred to as the helical domain. The guanine nucleotide is nestled between these two domains (4,5). The binding of GTP to ␣ T induces structural changes within 3 regions of the GTPase domain, designated as Switches 1, 2, and 3. The vertebrate visual system in rod cells has provided an excellent model system for understanding how GPCRs activate heterotrimeric G-proteins and subsequently their downstream targets (6,7). In the visual transduction pathway, the absorption of a photon leads to the isomerization of the covalently-bound chromophore, 11-cis-retinal, to its all-trans configuration. This converts rhodopsin to its functionally active conformation referred to as metarhodopsin II (R*). The heterotrimeric G-protein transducin is made up of a GDP-bound ␣ subunit (␣ T -GDP), as well as the ␤1 and ␥1 subunits, which are noncovalently complexed to each other and can be dissociated only by denaturation. The R* species binds to heterotrimeric transducin (␣ T -GDP/␤1␥1) with an extremely high affinity. This results in a weakening of the affinity of ␣ T for GDP such that it dissociates from the G-protein. GTP then binds to ␣ T yielding a species (␣ T -GTP) that has a reduced affinity for both R* and ␤1␥1. The structural changes in the switch regions of ␣ T -GTP allow it to bind and activate the downstream effector enzyme, the cGMP phosphodiesterase (PDE). The PDE is made up of two catalytic subunits (␣ PDE and ␤ PDE ) and two smaller regulatory subunits (␥ PDE ) and catalyzes the rapid hydrolysis of cGMP to GMP. The ␣ T -GTP species binds to ␥ PDE and relieves its inhibition of the catalytic activity of the ␣ PDE and ␤ PDE subunits. The activation of PDE is terminated by the hydrolysis of GTP by ␣ T , which is catalyzed by RGS9. The conversion of cGMP to GMP leads to a closure of cGMPgated channels on the rod cell membrane. This hyperpolarization results in an inhibition of neurotransmitter release, which represents the signal that is conveyed to the optic nerve.
Our laboratory has been interested in identifying and biochemically characterizing ␣ T mutants that might help to shed light on the mechanism by which GDP is released from a G␣ subunit, thus leading to G-protein activation (8 -12). These efforts have resulted in the identification of various G␣ mutants with novel biochemical properties. Owing to the difficulties in overexpressing native ␣ T in Escherichia coli, we have used a chimera of ␣ T and ␣ i1 , referred to as ␣ T *, as a starting G␣-backbone for these studies (13).
G-proteins belonging to both the monomeric (small) and heterotrimeric (large) G-protein families have a conserved serine or threonine residue in the P-loop. Mutation of this residue in the small G-protein Ras results in the stabilization of its nucleotide-free form relative to both its GDP-and GTP-bound forms (14). This yields a Ras protein that has an enhanced affinity for Ras-guanine nucleotide exchange factors (Ras-GEFs) and thereby has been used as a dominant-negative inhibitor of signaling by endogenous Ras in cells. Mutations of this residue have also been performed in some G␣ subunits. In the case of G␣ o or G␣ i2 , such changes resulted in their having an enhanced affinity for the G␤␥ subunit complex (15,16). A similar mutation in G␣ S gave rise to an enhanced affinity for the ␤-adrenergic receptor (17).
In this study, we have changed the conserved serine residue in the P-loop of ␣ T * to an asparagine residue (S43N). This mutant, unlike wild-type ␣ T *, is able to undergo GDP-GTP exchange at a significant rate even in the absence of R*. Similar to the phenotype shown by the G␣ S mutant toward the ␤-adrenergic receptor, the ␣ T *(S43N) mutant, when in the GDPbound state, slows R*-stimulated GDP-GTP exchange upon wild-type ␣ T *. In addition, we show that the GTP␥S-bound form of ␣ T *(S43N) is capable of activating the PDE and of slowing R*-stimulated GDP-GTP exchange upon wild-type ␣ T *. Moreover, using the activation of PDE as a probe, we have been able to show that GTP␥S-bound ␣ T *(S43N) stabilizes what may represent an intermediate complex that forms along the G-protein activation pathway and consists of both R* and the ␤1␥1 subunit complex, together with the ␣ T subunit.
During the course of our studies, Artemyev and co-workers (18) reported the characterization of the same mutation within the chimeric ␣ T * subunit. However, they reported that ␣ T *(S43N) was not capable of undergoing an exchange of GTP for GDP. We believe that this was likely due to a step in their purification protocol that was carried out at room temperature that would have resulted in the loss of the bound nucleotide. We have indeed observed that prolonged incubation of the ␣ T *(S43N) mutant at room temperature renders it unable to bind GTP␥S, and while it is still capable of binding to R* and preventing it from activating wild-type ␣ T *, it no longer exhibits many of the features that we describe below.

EXPERIMENTAL PROCEDURES
Materials-Frozen dark-adapted bovine retina were obtained from Lawson (Lincoln, NE). All other chemicals were from Sigma.
Purification of Visual Transduction Proteins-Rod outer segment (ROS) membranes were isolated as described (19). Holotransducin and PDE were obtained from ROS membranes essentially as described (20). The PDE was further purified by gel filtration chromatography on a HiLoad Superdex 200 HR26/60 column (GE Healthcare) equilibrated with HMAG buffer (20 mM Na-HEPES, pH 7.5, 5 mM MgCl 2 , 1 mM Na-azide, and 10% glycerol). Urea-washed disc membranes were prepared as described and served as the source of R* (21).
Expression and Purification of ␣ T *-We have used a construct designated pHis6Chi8, which was obtained from Dr. Heidi Hamm (Vanderbilt University). This chimera has the corresponding region from ␣ i1 inserted between residues 215 and 295 of ␣ T . In addition, residues 244 and 247 were changed back to the original amino acids in ␣ T , yielding what we refer to as ␣ T *. The recombinant ␣ T * subunits, both wild-type and S43N, were expressed in BL21 (DE3) supercompetent cells and purified as described (11,13). The proteins were further purified by ion exchange chromatography as outlined for the ␤1␥1 complex. The proteins were concentrated, aliquoted, snap-frozen, and stored at Ϫ80°C. The final yield of chimeric ␣ T * was typically ϳ1-2 mg of pure protein/liter of bacterial culture. Nucleotide occupancy of the purified recombinant ␣ T * and ␣ T *(S43N) was determined using HPLC (11).
Determination of Active Concentrations of ␣ T * Subunits-The amount of active, functional ␣ T * was determined by assaying [ 35 S]GTP␥S binding activity. Rhodopsin was light activated (R*) by incubation in ambient light for 5 min on ice. The ␣ T * subunits (500 nM) were incubated in the presence of 50 nM R*, 500 nM ␤1␥1, and 50 M [ 35 S]GTP␥S, in HMDM buffer (20 mM HEPES, pH 7.5, 5 mM MgCl 2 , and 0.01% (w/v) dodecylmaltoside) at a total volume of 100 l, for 1-2 h at room temperature. Subsequently, 40-l aliquots were applied, in duplicate, to prewet nitrocellulose filters (Schleicher & Schuell, pore size, 0.45 m) on a suction manifold. The filters were washed twice with HM buffer (20 mM HEPES, pH 7.5, 5 mM MgCl 2 ), added to scintillation liquid (30% LSC Scintisafe Mixture), and counted in a scintillation counter (LS6500 multipurpose scintillation counter). All protein concentrations of ␣ T * were determined by this procedure.
Time Course of GTP␥S Binding-R* and ␤1␥1 were mixed with [ 35 S]GTP␥S to give a final GTP␥S concentration of 50 M in HMDM buffer. The GDP-GTP exchange reaction was initiated by the addition of ␣ T *, with the final volume of the reaction mixture being 300 l. At different times, 20-l aliquots were added to prewetted nitrocellulose filters and processed as described in the previous section.
Fluorescence Measurements-Fluorescence measurements were carried out using a Varian eclipse spectrofluorimeter. The binding of aluminum fluoride (AlF 4 Ϫ ) was measured by monitoring the intrinsic tryptophan fluorescence of ␣ T * (excitation: 280 nm; emission: 340 nm). ␣ T * (300 nM) was mixed with 1 ml of HMDM buffer at room temperature and the tryptophan fluorescence emission was monitored continuously for 1 min. Subsequently, AlF 4 Ϫ (final concentrations: 5 mM NaF and 100 M AlCl 3 ) was added while monitoring the tryptophan fluorescence emission in real-time.
The GTP␥S-bound form of ␣ T *(S43N) was prepared by incubating the protein (5 M) with 100 M GTP␥S in HMDM buffer at room temperature for 2 h. The protein was subsequently placed on ice. ␣ T *(S43N)-GTP␥S was added at varying concentrations to a cuvette containing a mixture of R* (5 nM), ␤1␥1 (300 nM), and GTP␥S (50 M). Tryptophan fluorescence was monitored for 1 min and ␣ T *(wild-type) was added, and the kinetics of GTP␥S binding were monitored. All kinetic data were fitted to a single exponential Equation 1, where F ϭ fluorescence signal at any time t, F 0 ϭ fluorescence signal at time t ϭ 0, F ∞ ϭ fluorescence signal at t ϭ ∞, k obs ϭ observed rate constant.
Labeling the ␤1␥1 Subunit Complex-The ␤1␥1 complex (10 M) was incubated in a buffer containing 20 mM Na-HEPES (pH 7.5) and 5 mM MgCl 2 with a 0.5 mM solution of 5-((((2-iodoacetyl)amino)ethyl)amino)naphthalene-1-sulfonic acid (IAEDANS), dissolved in dimethyl formamide, at room temperature in the dark for 2 h. The reaction was quenched by the addition of ␤-mercaptoethanol at 11.2 mM. The labeled ␤1␥1 was separated from free probe and exchanged into HMAG buffer on a PD-10 desalting column (Amersham Biosciences). The extent of incorporation of the IAEDANS moiety was calculated using an extinction coefficient of 5600 M Ϫ1 cm Ϫ1 at 336 nm. The stoichiometry of labeling was determined by correcting the absorbance at 280 nm for IAEDANS absorption and measuring the concentration of ␤1␥1 using the calculated extinction coefficient of 53,600 M Ϫ1 cm Ϫ1 . The stoichiometry of labeling was determined to be 1.2 Ϯ 0.1 mol per mol ␤1␥1.
Fluorescence Anisotropy Measurements-Anisotropy measurements of IAEDANS-labeled ␤1␥1 complexes were carried out on a Cary eclipse spectrofluorimeter in the L-format. All reported anisotropy values have been corrected by determination of G-factors using horizontally polarized light. The excitation and emission wavelengths were set at 336 nm (band-width ϭ 10 nm) and 490 nm (bandwidth ϭ 20 nm), respectively. An integration time of 10 s was used for all measurements.
Measurements of cGMP PDE Activity-The analysis of cGMP hydrolysis by the retinal PDE was carried out as described (22). Briefly, a pH microelectrode was used to measure the decrease in pH resulting from the production of a proton for each molecule of cGMP hydrolyzed by PDE. All assays were carried out at 22°C in a final volume of 300 l containing 5 mM HEPES (pH 7.4), 100 mM NaCl, and 2 mM MgCl 2 (AB buffer). The dose response for the activation of PDE by ␣ T * was determined by mixing R* (118 nM), ␤1␥1 (370 nM), 100 M GTP␥S, and ␣ T * in AB buffer. This mixture was incubated at room temperature for 1-2 h. PDE (70 nM) was added and incubated at room temperature for 5 min. The reaction was initiated by the addition of 2 mM cGMP. At the end of each assay period, the buffering capacity (in mV/nmol) was determined by adding 500 nmol of sodium hydroxide. The rate of hydrolysis of cGMP (nmol/s) was determined from the ratio of the initial slope of the pH record (mV/nmol) and the buffering capacity of the assay buffer (mV/nmol).
GTP␥S-bound ␣ T *(S43N) was prepared by incubating the ␣ T * subunit (5 M) with GTP␥S (100 M) in AB buffer for 2 h at room temperature. To determine the effects of R* and the ␤1␥1 subunit complex on PDE activation, R* and/or ␤1␥1 were mixed with GTP␥S-bound ␣ T *(S43N), diluted in AB buffer containing 100 M GTP␥S to a final concentration of 250 nM, and incubated at room temperature for 5 min. PDE and cGMP were added to this mixture and the PDE activity measured as described above.

The Purified Recombinant ␣ T *(S43N) Mutant Is Isolated with
Bound GDP-Both the G␣ subunits of large G-proteins and the Ras-related small G-proteins bind to GDP with a very high affinity. This explains the requirement for an activated GPCR or a GEF to catalyze the dissociation of GDP from G␣ subunits and Ras-related G-proteins, respectively, in order for GDP-GTP exchange to occur. In the case of the small G-proteins, the highly conserved serine/threonine residues in the P-loop help to coordinate Mg 2ϩ , which is essential for the high affinity binding of GDP, such that substitutions at this site yield nucleotide-depleted, dominant-negative mutants (14). Magnesium is not necessary for the high affinity binding of GDP to the G␣ subunits of large G-proteins, possibly because of their helical domain, which is lacking in the small G-proteins (23). However, the x-ray crystal structure of the GDP-bound form of ␣ T shows that the main-chain nitrogen of the serine residue at position 43 makes a hydrogen bond with one of the non-bridging oxygens on the ␤-phosphate of GDP, thereby suggesting a potentially important role in high-affinity binding (24). In addition, there have been suggestions that substitutions for the conserved P-loop serine/threonine residue in G␣ subunits yield dominant-negative inhibitors, raising the question of whether these mutants are also nucleotide-depleted (16 -18, 25-27). Thus, we performed HPLC analysis to determine whether the ␣ T *(S43N) mutant, following its purification, still contained bound guanine nucleotide. We found that as in the case for the wild-type ␣ T * subunit, the ␣ T *(S43N) mutant was purified with stoichio-metric amounts of bound GDP (Fig. 1), i.e. with stoichiometries of 1 Ϯ 0.1 mol GDP per mol ␣ T * and 0.8 Ϯ 0.1 mol GDP per mol ␣ T * (S43N).
The x-ray crystal structure for the aluminum fluoride (AlF 4 Ϫ )/GDP-bound form of ␣ T shows that AlF 4 Ϫ occupies the ␥-phosphate-binding site of GTP (28). This structure has been postulated to resemble the transition state for GTP hydrolysis. When AlF 4 Ϫ binds to the GDP-bound form of ␣ T, there is an increase in the intrinsic tryptophan fluorescence due to a conformational change in the Switch 2 region, one of three regions of ␣ T that change conformation upon binding GTP (29). As expected, a robust increase in tryptophan fluorescence was observed upon addition of AlF 4 Ϫ to the wild-type ␣ T * subunit (Fig. 2). However, the addition of AlF 4 Ϫ to the ␣ T *(S43N) mutant did not cause a detectable change in tryptophan fluorescence. This was not entirely unexpected since the presence of Mg 2ϩ is required for the AlF 4 Ϫ -induced enhancement in tryptophan fluorescence. The x-ray crystal structures of both the GTP␥S-and AlF 4 Ϫ -bound forms of ␣ T show that the side-chain hydroxyl oxygen of Ser-43 is one of the coordination sites for the bound divalent cation (Mg 2ϩ in the GTP␥S-bound ␣ T structure and Ca 2ϩ in the AlF 4 Ϫ -bound ␣ T structure) (28,30). Thus, the ␣ T *(S43N) mutant would have a reduced affinity for Mg 2ϩ and, consequently, might not be able to undergo the AlF 4 Ϫ -induced conformational change that alters the position of Switch 2 and enhances its intrinsic tryptophan fluorescence.
R* Has Only Minor Effects on GDP-GTP Exchange within the ␣ T *(S43N) Mutant-Wild-type ␣ T * undergoes very slow guanine nucleotide exchange when assayed in the absence of R* and ␤1␥1, as monitored by either the enhancement in the intrinsic tryptophan fluorescence that accompanies the ex-change of GDP for GTP␥S (k obs ϭ 0.01 min Ϫ1 ), or by using a filter binding assay to measure the exchange of GDP for [ 35 S]GTP␥S (0.02 min Ϫ1 ) (Fig. 3A, open squares). This is similar to the spontaneous nucleotide exchange rate reported earlier (31). In contrast, we found that the ␣ T *(S43N) mutant undergoes nucleotide exchange with a rate constant that is about 4 -8-fold faster than that for the wild-type ␣ T * subunit (k obs ϭ 0.09 min Ϫ1 ) as monitored using intrinsic tryptophan fluorescence (Fig. 3A, dotted line). A similar rate constant was obtained when using a filter binding assay to measure the exchange of GDP for [ 35 S]GTP␥S (k obs ϭ 0.08 min Ϫ1 ). A filter binding assay was used to measure the rate of dissociation of GDP using wild-type ␣ T * or ␣ T *(S43N) that had been preloaded with [ 3 H]GDP. This experiment showed that the rate of release of GDP from ␣ T *(S43N) was 4-fold faster than that of wild-type ␣ T * (Fig. 3A, inset). Thus, changing the serine to asparagine at position 43 within the ␣ T * subunit apparently causes a change in the conformation of the P-loop such that there is an increase in the rate of GDP dissociation, which has been suggested to be the rate-limiting step for GDP-GTP exchange (32).
When R* was added together with the ␤1␥1 subunit complex to wild-type ␣ T *, there was the expected marked increase in the rate of GDP-GTP exchange. For example, the addition of 12 nM R*, along with 300 nM ␤1␥1, to 300 nM wild-type ␣ T * increased the rate constant for guanine nucleotide exchange by at least 500-fold (Fig. 3B, inset, open triangles). This rate enhancement due to light-activated rhodopsin is similar to what has been reported earlier (10). However, the addition of even relatively high levels of R* to the ␣ T *(S43N) mutant had only minimal effects on the rate of nucleotide exchange (Fig. 3B). Specifically, when the ␣ T *(S43N) mutant was assayed in the presence of 112 nM R* and 400 nM ␤1␥1, the rate constant for GDP-GTP  S43N). Wild-type ␣ T * (red) and ␣ T *(S43N) (green) at concentrations of 100 M were incubated in the presence of 7.5% acetonitrile for 5 min at room temperature and centrifuged at 16,000 ϫ g for 10 min at room temperature. An aliquot of the supernatant (20 l) was injected into a C18 reverse phase HPLC column equilibrated with 100 mM KH 2 PO 4 , pH 6.5, 10 mM tetrabutylammonium bromide, 0.2% NaN 3 , and 7.5% acetonitrile, and then isocratic elution was performed. The absorbance at 253 nm was monitored. The elution times of GDP and GTP standards are indicated by arrows. exchange was increased by only ϳ3-fold. The rate of GDP-GTP exchange upon wild-type ␣ T * was too fast to be measured under these conditions (Fig. 3B, inset, open squares). These findings suggested either that the ␣ T *(S43N) mutant is impaired in its ability to interact with R* or that the release of R* from the activated, GTP␥S-bound mutant is slow relative to its rate of release from wild-type ␣ T *, thus compromising the ability of R* to act catalytically.
The ␤1␥1 Subunit Complex Has No Effect on the GDP-GTP Exchange Activity of the ␣ T * (S43N) Mutant-The ␤1␥1 subunit complex has been shown to increase the affinity of GPCRs like R* for their G␣-signaling partners, as well as to help GPCRs to further increase the rate of G-protein activation (33)(34)(35)(36)(37). The stimulatory actions of the ␤1␥1 subunit complex were clearly evident when assaying nucleotide exchange on the wild-type ␣ T * subunit, as the rate constant for R*-stimulated GDP-GTP exchange increased from 0.02 min Ϫ1 in the absence of ␤1␥1 to 0.78 min Ϫ1 in the presence of 400 nM ␤1␥1 (Fig. 4A, inset). This rate enhancement, due to the ␤1␥1 complex, is similar to that reported in earlier studies (10). However, this was not the case when assaying the ␣ T *(S43N) mutant, as the rate of nucleotide exchange upon the addition of 400 nM ␤1␥1 to ␣ T *(S43N) in the presence of 14 nM R* was 0.11 min Ϫ1 (Fig. 4A), similar to the rate constant obtained in the absence of ␤1␥1 (0.11 min Ϫ1 ). Similarly, the ␤1␥1 subunit complex had no effect on the GDP-GTP exchange activity of the S43N mutant in the absence of R* (Fig. 4B). These findings raised the possibility that the ␣ T *(S43N) mutant might not be able to effectively interact with the ␤1␥1 subunit complex. We then used a fluorescence anisot-ropy assay that allows for the direct monitoring of the binding of G␣ subunits to the ␤1␥1 subunit complex labeled with IAEDANS, and found that the ␣ T *(S43N) mutant bound to ␤1␥1 with an affinity that was similar to that for the binding of wild-type ␣ T * to ␤1␥1 (Fig. 4C). This indicates that the ␤1␥1 subunit complex is able to bind to the ␣ T *(S43N) mutant but it is unable to help accelerate the nucleotide exchange reaction in the presence of R*.
As expected, the rate of R*-stimulated GDP-GTP exchange on wild-type ␣ T * was very slow in the absence of any added ␤1␥1. However, the rate of GDP-GTP exchange was enhanced in the presence of 60 nM ␤1␥1. Importantly, a similar rate of GDP-GTP exchange was observed when the IAEDANS-labeled ␤1␥1 was used instead of unlabeled ␤1␥1 (Fig. 4C, inset), demonstrating that the IAEDANS-labeled ␤1␥1 was functionally equivalent to its unlabeled counterpart.
␣ T * (S43N) Slows the Rate of R*-dependent GDP-GTP Exchange on Wild-type ␣ T *-It was first shown for the small G-protein Ras, that changing the serine residue at position 17, corresponding to Ser-43 in ␣ T *, to asparagine (S17N) yielded a potent dominant-negative inhibitor of Ras signaling by sequestering upstream GEFs (38,39). Likewise, the ␣ T *(S43N) mutant, even at relatively low concentrations in comparison to wild-type ␣ T *, was able to inhibit R*-stimulated GDP-GTP exchange on the wild-type ␣ T * subunit (Fig. 5A, inset). Specifically, the addition of ϳ20 nM ␣ T *(S43N) slowed the rate of R*-stimulated nucleotide exchange on wild-type ␣ T * (300 nM) by ϳ70-fold (Fig. 5A), indicating that the GDP-bound form of ␣ T *(S43N) was able to bind to R* with high affinity and prevent it from binding and activating the wild-type ␣ T * subunit.
Interestingly, we found that the GTP␥S-bound form of the ␣ T *(S43N) mutant was also capable of forming a stable complex with R*. As shown in Fig. 5B, increasing concentrations of the GTP␥S-bound form of this ␣ T * mutant were able to reduce the rate of R*-stimulated GDP-GTP exchange on wild-type ␣ T *. For example, the addition of ϳ50 nM GTP␥S-bound ␣ T *(S43N) slowed the rate of GDP-GTP exchange on wild-type ␣ T * by ϳ10-fold (Fig. 5B, inset).
Addition of light-activated rhodopsin at 115 nM was able to relieve the inhibition of nucleotide exchange on wild-type ␣ T * caused by the addition of either GDP-bound (Fig. 5C) or GTP␥S-bound (Fig. 5D) ␣ T * (S43N). This suggests the idea that the ␣ T *(S43N) mutant is acting by forming a stable complex with R* and blocking its interaction with wild-type ␣ T *.

R* and the ␤1␥1 Subunit Complex Together Inhibit PDE Activation by ␣ T * (S43N)-Given the indications that the GTP␥S-
bound ␣ T *(S43N) mutant can form a stable complex with R*, we were interested in seeing whether the ␣ T *(S43N)-GTP␥S species was also capable of engaging and regulating the cyclic GMP PDE. We found that like GTP␥S-bound wild-type ␣ T *, the GTP␥S-bound ␣ T *(S43N) mutant was able to bind to the PDE and stimulate its activity in a dose-dependent manner (Fig.  6A). The activation of the PDE by wild-type ␣ T * was unaffected when R* and the ␤1␥1 subunit complex were added separately or together. This is because ␣ T *-GTP␥S has a very low affinity for R* and ␤1␥1. Similarly, the ability of ␣ T *(S43N)-GTP␥S to activate PDE was unaffected when R* and the ␤1␥1 complex were added individually to ␣ T * (S43N). However, surprisingly, the addition of R* and the ␤1␥1 subunit complex together to ␣ T *(S43N) inhibited its ability to activate the PDE (Fig. 6B). The  Fig. 3A. B, 50 M [ 35 S]GTP␥S was added to 300 nM ␣ T *(S43N) in the absence (E) or presence of 400 nM ␤1␥1 (ƒ). The samples were filtered and counted as described in the legend for Fig. 3A. The y-axis shows the number of pmol of GTP␥S in 20 l of the reaction mix. The radioactivity data is shown as the mean Ϯ S.E. of three independent experiments. The lines through the data show a simultaneous single exponential fit of all 3 data sets. Data from experiments carried out in the absence of ␣ T *(S43N) are shown as corresponding closed symbols. C, aliquots of wild-type ␣ T * (E) and ␣ T *(S43N) (F) were added successively to the ␤1␥1 complex (400 nM) labeled with IAEDANS in HMDM buffer and the fluorescence anisotropy was monitored (excitation ϭ 336 nm, emission ϭ 490 nm). The data were fit to a bimolecular binding model. Inset, 50 M GTP␥S was added to 300 nM wild-type ␣ T * and 10 nM R* in the absence (red line) or the presence of either 60 nM unlabeled ␤1␥1 complex (green), or the ␤1␥1 complex labeled with IAEDANS (blue line) in HMDM buffer, and the tryptophan fluorescence emission was monitored in real-time with the excitation wavelength set at 300 nm and the emission wavelength at 345 nm.
inhibition of the ␣ T *(S43N)-stimulated PDE activity occurred in a dose-dependent manner with respect to either R* (Fig. 7A) or the ␤1␥1 subunit complex (Fig. 7B). This suggests that R*, the activated form of ␣ T *(S43N), and the ␤1␥1 subunit complex are able to form a ternary complex that prevents ␣ T *(S43N) from stimulating PDE activity.

DISCUSSION
The visual phototransduction pathway has been proven to be a valuable system for understanding how a heptahelical GPCR can lead to the activation of its cognate G-protein and to the stimulation of an intracellular effector activity (40). However, despite an extensive number of biochemical and structural studies, the exact sequence of events that occurs during this process is not clear. According to the traditional model for G-protein activation, a GPCR that is activated by an external stimulus like light or hormones binds to a heterotrimeric G-protein. This binding triggers the release of GDP from the G␣ subunit and its replacement by GTP. The binding of GTP has been shown to cause structural changes in the G␣ subunit (24). These structural changes have been proposed to cause the dissociation of the G␣ subunit from both the ␤␥ subunit complex and the activated GPCR. The GTP-bound G␣ subunit can then bind to and regulate its downstream targets.
Our strategy for gaining a better understanding of the mechanism of G-protein activation has been to identify and characterize G␣ mutants that might provide insight into different stages of the activation process (10 -12). In particular, we have generated mutations within a chimeric ␣ T subunit, referred to as ␣ T *, that contains an insertion between residues 215 and 295 of the corresponding region from G␣ i1 . Although the insertion comprises 79 residues, there are only 21 amino acid substitu- tions, of which 12 are conservative. This insertion includes the substitution of residues in the conformationally sensitive Switch 3 region of ␣ T . Despite these alterations, ␣ T * is similar to retinal ␣ T in undergoing GDP-GTP exchange in a R*-dependent manner and activating the cGMP phosphodiesterase.
In this study, we have changed the conserved serine residue within the region referred to as the P-loop to an asparagine residue. In the x-ray crystal structures of ␣ T , in complex with either GDP or GTP␥S, the main-chain nitrogen of Ser43 forms a hydrogen bond with the non-bridging oxygen of the ␤-phosphate of the nucleotide (24,30). In addition, the side-chain hydroxyl oxygen of Ser-43 forms hydrogen bonds with the carboxyl oxygen of Asp-196, located in the Switch 2 region, in both the GDP-and GTP␥S-bound forms of ␣ T , and with the side-chain hydroxyl of Thr-177 in the GTP␥S-bound structure of ␣ T .
One of the first observations we made regarding the ␣ T *(S43N) mutant was its ability to undergo spontaneous GDP-GTP exchange, in the absence of R*. The rate-limiting step of the GDP-GTP exchange reaction is the release of GDP from the ␣ T subunit (32). Thus, the S43N mutation apparently causes a change in the conformation of the P-loop to enhance the rate of GDP release. Still despite its weakened affinity, a substantial fraction of the purified ␣ T *(S43N) mutant contains bound GDP as indicated by reverse-phase HPLC.
Surprisingly, R* was much less effective at stimulating GDP-GTP exchange on ␣ T *(S43N) compared with the wild-type ␣ T * subunit. The main binding site for R* on the ␣ T subunit is  (S43N). A, wild-type ␣ T * (F) or ␣ T *(S43N) (E) at various concentrations was preincubated with 100 M GTP␥S in the presence of 118 nM R* and 370 nM ␤1␥1 for 1 h in a buffer containing 5 mM HEPES, pH 7.4 (AB buffer), at room temperature. PDE (70 nM) was added to this mixture and the pH of the sample was monitored. Subsequently, cGMP (2 mM) was added to initiate the reaction and the change in pH was monitored over time. At the end of the assay period, the buffering capacity (mV/nmol) was determined by the addition of 600 nmol of NaOH to the reaction mixture. The rate of hydrolysis of cGMP (nmol/s) was determined from the ratio of the slope of the pH record (mV/s) and the buffering capacity of the assay buffer (mV/nmol) and plotted on the Y-axis. The data are shown as the mean Ϯ S.E. from three independent experiments. B, GTP␥S-bound ␣ T *(S43N) or wild-type ␣ T * was prepared by incubating either ␣ T * subunit with 100 M GTP␥S at room temperature for 2 h and then placed on ice. The incubation mixture contained 1 M R* and 1 M ␤1␥1 in the case of wild-type ␣ T *. Subsequently, the GTP␥S-bound ␣ T * subunits were diluted into AB buffer to a final concentration of 250 nM in the presence or absence of R* (4.8 M) and ␤1␥1 (400 nM) and incubated at room temperature for 5 min. PDE was added to a final concentration of 160 nM, and its activation was measured as described for panel A. The data are presented as a bar graph with the y-axis showing the PDE activity in nmol/s. The red bars show data in the absence of ␤1␥1, and the green bars show data in the presence of ␤1␥1. The presence or absence of R* and the identity of the ␣ T * subunits are shown on the plots. FIGURE 7. Effects of R* and ␤1␥1 on PDE activation by ␣ T * (S43N). ␣ T *(S43N) bound to GTP␥S was prepared as described in Fig. 6B. This was diluted in AB buffer to a final concentration of 250 nM. Subsequently, the ␤1␥1 subunit complex (500 nM, panel A) or R* (4.8 M, panel B) was added to the sample in the presence of varying concentrations of R* (panel A) or ␤1␥1 (panel B). PDE was added to a final concentration of 160 nM, and its activation was measured as described for Fig. 6B. located at the C terminus, which is a considerable distance from the P-loop (4). This suggested that the impaired ability of R* to catalyze GDP-GTP exchange on ␣ T *(S43N) might not be due to R* having a reduced affinity for the GDP-bound form of ␣ T * (S43N). Instead, the ␣ T *(S43N) mutant may be impaired in its ability to undergo the necessary conformational change following GTP binding that enables a rapid release of R*, thereby making R* a less efficient catalyst of the nucleotide exchange reaction.
Mutation of the homologous serine residue in the monomeric G-protein Ras to an asparagine (S17N) yields a protein whose nucleotide-free form has an extremely high affinity for its GEFs (14,38). Introduction of this mutant into cells inhibits the activation of wild-type Ras, such that it behaves as a dominant-negative inhibitor (38). Similar to the S17N mutant of Ras, the ␣ T *(S43N) mutant inhibits the activation of wild-type ␣ T *. Because GTP␥S is able to bind to ␣ T *(S43N) in the absence of R*, we were able to prepare GTP␥S-bound ␣ T *(S43N) and examine its effects on the GDP-GTP exchange activity of wildtype ␣ T *. We found that the GTP␥S-bound form of ␣ T *(S43N) was also able to inhibit R*-dependent GDP-GTP exchange on wild-type ␣ T *, in a dose-dependent manner. These findings suggest that upon the binding of GTP␥S, the ␣ T *(S43N) mutant was still capable of associating with the activated receptor.
GTP␥S-bound wild-type ␣ T *, that was prepared in the presence of low concentrations of R* and ␤1␥1, activated the cGMP PDE very effectively. The presence of high concentrations of R* and ␤1␥1, either separately or together, had no effect on the ability of GTP␥S-bound ␣ T * to activate the PDE. This is consistent with the observation that upon binding GTP␥S, wildtype ␣ T * dissociates from R* and ␤1␥1 (41). The GTP␥S-bound form of ␣ T *(S43N) is able to activate the cGMP PDE with a dose-response that is similar to that for wild-type ␣ T *. Neither the addition of R* alone, nor ␤1␥1 alone, had any effect on the activation of the PDE by GTP␥S-bound ␣ T * (S43N). However, the combined addition of R* and ␤1␥1 strongly inhibited the activation of the PDE by ␣ T *(S43N). This provides evidence for the existence of a ternary complex consisting of R*, GTP␥Sbound ␣ T *(S43N), and ␤1␥1. The existence of such a ternary complex was initially postulated based on the enhancement of the cholera toxin-catalyzed ADP-ribosylation of retinal ␣ T that was induced in the presence of R*, the ␤1␥1 subunit complex, and the non-hydrolyzable GTP analog Gpp(NH)p (42,43). The ability of a ternary complex to form consisting of an activated receptor, an activated G␣ subunit, and a G␤␥ subunit complex carries interesting implications for the mechanisms that describe the reaction pathway for the activation of heterotrimeric G-proteins, as it raises important questions concerning whether G␤␥ has a role in influencing the binding of GTP and the ensuing GTP-induced activating conformational change within G␣, as well as when the activated GTP-bound G␣ subunit dissociates from G␤␥ and the receptor, as this has been been traditionally assumed to be necessary for G␣ to engage and regulate its effector activity.
Based on these findings, we propose the following model (Fig. 8). The exchange of GTP for GDP within a G␣ subunit complexed to ␤1␥1 and R* leads to the formation of a ternary complex. We postulate that the ␣ T *(S43N) mutant stabilizes the ternary complex, such that it persists even following GDP-GTP exchange on ␣ T *(S43N). It has been shown that the binding of Mg 2ϩ is essential for the enhancement in intrinsic trytophan fluorescence that accompanies the conformational change in Switch 2 that occurs upon the activation of retinal ␣ T (44). In the case of the ␣ T *(S43N) mutant, we observe changes in its intrinsic tryptophan fluorescence upon binding GTP␥S, but these changes are substantially smaller than those for wildtype ␣ T *. The Switch 2 region is one of two regions on the ␣ T subunit, with the other being the N-terminal helix, that is involved in binding to the ␤1␥1 complex (45). This subunit complex has been shown to increase the affinity of ␣ T for R* (34). Thus, in order for GTP-bound ␣ T to dissociate from R*, the affinity of ␤1␥1 for ␣ T has to be reduced and this might be an outcome of a Mg 2ϩ -induced conformational change that occurs in Switch 2. The inability of the GTP-bound ␣ T *(S43N) mutant to undergo this Mg 2ϩ -dependent conformational change would then stabilize a ternary complex consisting of R*, ␣ T *(S43N)-GTP␥S, and ␤1␥1. However, the free GTP␥Sbound ␣ T *(S43N) species (i.e. when not associated with either R* or ␤1␥1) is able to activate the cGMP PDE as effectively as wild-type ␣ T *. This indicates that upon GDP-GTP exchange, ␣ T *(S43N) is able to couple to its biological effector (i.e. the PDE), provided that neither R* or ␤1␥1 is bound to this mutant so as to interfere with effector binding. The activation of a GPCR (e.g. R*) promotes its interaction with a heterotrimeric G-protein (␣ T -GDP-␤1␥1). This results in GDP-GTP exchange on ␣ T , leading to a conformational change (i.e. depicted by the change from a red circle to a red square) that enables ␣ T -GTP to dissociate from ␤1␥1 and R*. Apparently, the binding of Mg 2ϩ by ␣ T is necessary for this conformational change. Thus, the GTP-bound ␣ T *(S43N) mutant, that is incapable of binding Mg 2ϩ and undergoing such a conformational change, is able to form a stable complex with R* and ␤1␥1.
In summary, we have characterized an S43N mutant of a G␣ subunit and shown that it can stabilize a ternary complex that also includes a GPCR (R*) and a G␤␥ complex (␤1␥1). Thus, the ␣ T *(S43N) acts as a dominant negative inhibitor of wild-type ␣ T by sequestering R* and ␤1␥1, even when it is bound to GTP. Perhaps more importantly, this ␣ T * mutant makes it possible to trap a previously uncharacterized intermediate along the G-protein activation pathway (i.e. consisting of the receptor, the GTP-bound G␣ subunit, and the G␤␥ complex).