A switch 3 point mutation in the alpha subunit of transducin yields a unique dominant-negative inhibitor.

The rhodopsin/transducin-coupled vertebrate vision system has served as a paradigm for G protein-coupled signaling. We have taken advantage of this system to identify new types of constitutively active, transducin-alpha (alphaT) subunits. Here we have described a novel dominant-negative mutation, made in the background of a chimera consisting of alphaT and the alpha subunit of G(i1) (designated alphaT*), which involves the substitution of a conserved arginine residue in the conformationally sensitive Switch 3 region. Changing Arg-238 to either lysine or alanine had little or no effect on the ability of alphaT* to undergo rhodopsin-stimulated GDP-GTP exchange, whereas substituting glutamic acid for arginine at this position yielded an alphaT* subunit (alphaT*(R238E)) that was incapable of undergoing rhodopsin-dependent nucleotide exchange and was unable to bind or stimulate the target/effector enzyme (cyclic GMP phosphodiesterase). Moreover, unlike the GDP-bound forms of alphaT*, alphaT*(R238A) and alphaT*(R238K), the alphaT*(R238E) mutant did not respond to aluminum fluoride (AlF4(-)), as read out by changes in Trp-207 fluorescence. However, surprisingly, we found that alphaT*(R238E) effectively blocked rhodopsin-catalyzed GDP-GTP exchange on alphaT*, as well as rhodopsin-stimulated phosphodiesterase activity. Analysis by high pressure liquid chromatography indicated that the alphaT*(R238E) mutant exists in a nucleotide-free state. Nucleotide-free forms of G alpha subunits were typically very sensitive to proteolytic degradation, but alphaT*(R238E) exhibited a resistance to trypsin-proteolysis similar to that observed with activated forms of alphaT*. Overall, these findings indicated that by mutating a single residue in Switch 3, it is possible to generate a unique type of dominant-negative G alpha subunit that can effectively block signaling by G protein-coupled receptors.

the hydrolysis of cyclic GMP. The reduction in the levels of cyclic GMP results in the closure of cation-specific cyclic GMP-gated ion channels in rod outer segments, leading to membrane hyperpolarization and the generation of a signal that is sent to the optic nerve. Significant amplification of the signal occurs at the level of G protein activation, so that several hundred ␣T subunits can be activated upon the absorption of a single photon by a rhodopsin molecule. The signaling system then shuts off upon the hydrolysis of GTP by ␣T, with this deactivation step being accelerated through the actions of regulator of G protein signaling proteins (2)(3)(4).
Given the central role that G proteins play as molecular switches in GPCR-coupled signal transduction, it is of fundamental importance to understand how they are activated by their upstream receptors to undergo GDP-GTP exchange. This is an especially intriguing question, given the indications that the binding site for GPCRs on their G␣ subunit targets is a significant distance from the guanine nucleotide-binding domain (5)(6)(7). It is generally agreed that the rate-limiting step for G protein activation is the dissociation of GDP from the G␣ subunit. The retinal G protein transducin exhibits an especially slow rate of intrinsic GDP dissociation and basal GDP-GTP exchange (8) and therefore is dependent on light-activated rhodopsin to catalyze these events. Unlike their small G protein counterparts, the ␣ subunits of the large G protein family contain a helical domain in addition to their guanine nucleotidebinding (Ras-like) domain (9 -14). Thus, it has been suspected that the role of GPCRs (e.g. rhodopsin) in stimulating GDP-GTP exchange involves changing the juxtaposition of the helical domain relative to the Ras-like domain, as well as perturbing the interactions that normally hold GDP tightly in place (via residues that stabilize the guanine ring and phosphate moieties). Recently, we have tested the importance of altering the relative positions of the helical and Ras-like domains (15). Substitutions were made for the conserved residues that comprise the linker regions connecting these two major domains. Indeed, we found that mutations in the linker regions of ␣T yielded molecules capable of constitutive GDP-GTP exchange in the absence of rhodopsin.
We have tried to identify mutations that yield constitutively active G␣ subunits that fully mimic the functional outcome of GPCR-mediated interactions, just as we have done for the small G protein Cdc42 (16,17). Along these lines, we have examined several different regions on ␣T, with one of these being the conformationally sensitive Switch 3 domain. The Switch 3 domain has been shown to play a critical role in G protein activation by linking the binding of target/effectors at Switch 2 to the stimulation of effector activity (18,19) and to be important for regulator of G protein signaling-stimulated GTP hydrolysis (20). Switch 3 might also help to ensure that ␣T reaches the appropriate activated conformation to engage downstream target/effectors and, in doing so, perhaps to influence nucleotide binding and the nucleotide exchange reaction.
To examine such possibilities, we have changed conserved residues in Switch 3 in the background of an ␣T/␣i1 chimera (␣T*, see "Experimental Procedures") and determined the consequences of these substitu-tions on ␣T* activation and ␣T*-target/effector interactions. Here we have reported the surprising finding that specific substitutions for one of these residues, Arg-238, can influence the ability of ␣T* to reach the activated (GTP-bound) state and to ultimately dissociate from the receptor (rhodopsin). Although certain changes at position 238 showed no apparent effect on the ability of ␣T* to functionally couple to rhodopsin, we found that when glutamic acid was substituted for the conserved arginine residue, an ␣T* subunit was generated that was capable of blocking the functional coupling between rhodopsin and transducin. The ␣T*(R238E) mutant was isolated in a nucleotide-free state but exhibited a reduced sensitivity to trypsin proteolysis, when compared with GDP-bound ␣T*. Thus, we believe that the R238E substitution yields a G␣ mutant that remains stable in the absence of bound nucleotides and therefore can serve as a dominant-negative inhibitor of the interactions of a GPCR with its G protein substrate. Mutagenesis-Degenerate primers from Invitrogen were used to generate different mutants of ␣T*(Arg-238). A plasmid containing the ␣T* gene, which codes for an ␣T-␣i1 chimera in which the corresponding region from ␣i1 was inserted between residues 215 and 294 of ␣T, was prepared as described previously (15). This served as a template to perform site-directed mutagenesis by PCR using the QuikChange sitedirected mutagenesis kit (Stratagene). The parental DNA template was digested using DpnI endonuclease, and the nicked vector DNA with the desired mutation was transformed into Escherichia coli XL-1 Blue competent cells. DNA was purified from single colonies using the Qiagen plasmid miniprep kit and sequenced at the BioResource Center of Cornell University.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-The recombinant ␣T* and the ␣T*(Arg-238) mutants were expressed and prepared using procedures similar to those described by Skiba et al. (21). Briefly, the ␣T* subunits were expressed in BL21(DE3) competent cells and purified in the presence of 50 M GDP. The proteins were eluted from a nickel-nitrilotriacetic acid column with 200 mM imidazole and then eluted from a Q column with a NaCl gradient. The ␣T* subunits were further purified by gel filtration chromatography on a HiLoad Superdex G75 HR 26/60 column equilibrated with a buffer containing 20 mM HEPES, pH 7.5, and 10% glycerol (Buffer G). The samples were aliquoted, snap-frozen, and stored at Ϫ80°C. The final yield of ␣T*, ␣T*(R238A), and ␣T*(R238K) ranged from 1 to 1.5 mg of pure protein/liter of bacterial culture. Some aggregation was noticed during the purification of the ␣T*(R238E) mutant, and the final yield of this mutant ranged from 0.1 to 0.2 mg of pure protein/liter of culture. However, gel filtration chromatography of purified ␣T* and ␣T*(R238E) yielded similar profiles with roughly 50% of the recombinant ␣T* subunits eluting at the expected size of ϳ40 kDa and ϳ50% of the total protein eluting at the size of ϳ80 kDa due to dimerization.
Purification of Retinal Proteins-Rod outer segment membranes were isolated using a sucrose gradient as described in Papermaster and Dreyer (22). Holotransducin and PDE6 were purified from rod outer segment membranes using previously described procedures (23). The PDE was further purified by gel filtration chromatography on a HiLoad Superdex G200 HR26/60 column equilibrated with 20 mM HEPES, pH 7.5, and 10% glycerol. Urea-washed rod outer segment membranes, prepared as described (24), provided the rhodopsin used in all experiments.
The ␣T and G␤␥ subunit components of holotransducin were separated on a Blue-Sepharose column equilibrated with 10 mM HEPES, pH 7.5, 6 mM MgCl 2 , 1 mM dithiothreitol, and 25% glycerol. The G␤␥ complex was eluted from the Blue-Sepharose column with 250 ml of low salt buffer (i.e. the equilibration buffer containing 100 mM KCl) and further purified by gel filtration chromatography on a HiLoad Superdex G75 HR 26/60 column equilibrated with buffer G. The final yield of G␤␥ ranged from 3 to 5 mg of pure protein/300 retina. The purified subunit complex was aliquoted, snap-frozen, and stored at Ϫ80°C.
HPLC Analysis-A Sunfire C-18 reversed phase column (0.46 ϫ 25 cm) filled with 5 M silica was obtained from Waters. The system consisted of a Waters 1525 binary HPLC pump machine and a Waters 2487 dual absorbance UV detector. Chromatography was performed in phosphate buffer (100 mM, pH 6.5) containing 10 mM tetrabutylammonium bromide, 7.5% (v/v) acetonitrile, and 0.2 mM NaN 3 (HPLC buffer) at ambient temperature with a flow rate of 1 ml/min. In this system, the order of elution of guanine nucleotides is GMP, GDP, and GTP (retention times 4.8, 8, and 14 min, respectively). The amount and identity of the guanine nucleotide bound to ␣T* was determined by adding ϳ400 g of protein sample to HPLC buffer and then centrifuging for 10 min at 12,000 ϫ g to remove denatured protein. The supernatant was added directly to the column, and guanine nucleotides were chromatographed using the same buffer. The column was calibrated with solutions of the different guanine nucleotides.
Trypsin Proteolysis Assay-The patterns of tryptic proteolysis were determined as described previously (25). Briefly, 20-l reactions containing 5 g of either ␣T* or ␣T*(R238E) were performed for 10 min at 25°C in 20 mM HEPES (pH 7.5) containing 2 mM MgSO 4 and 100 M GDP. Where indicated, 20 mM NaF and 600 M AlCl 3 were added to the buffer to form the aluminum fluoride (AlF 4 Ϫ ) complex and incubated at 25°C for 10 min. Trypsin digestions were performed by adding 2 l of trypsin (100 g of trypsin/ml) and incubating for 10 min at 25°C. The reactions were stopped with the addition of SDS sample buffer and heat treatment (95°C, 5 min). Proteolytic fragments were resolved on 15% SDS-gel electrophoresis and stained with Coomassie Blue. Fluorescence Measurements-Fluorescence measurements were made on a Varian Eclipse fluorescence spectrophotometer in 1 ml of HMN buffer (20 mM HEPES, pH 7.5, 5 mM MgCl 2 , and 100 mM NaCl) at 25°C. ␣T*-GDP (400 nM) was preincubated in HMN buffer for 5 min at 25°C. The AlF 4 Ϫ complex was formed by adding 10 mM NaF and 50 M AlCl 3 . The binding of AlF 4 Ϫ was measured by monitoring the enhancement of the intrinsic tryptophan fluorescence of ␣T* upon excitation at 300 nm and emission at 345 nm (15).  (20 l) were removed at the times indicated and applied directly to prewetted nitrocellulose filters (pore size ϭ 0.45 m) on a suction manifold. The filters were washed twice with HMN buffer and counted in a scintillation counter (LS6500 Multipurpose Scintillation counter) after the addition of 3 ml of scintillation liquid (Scintisafe 30% cocktail). The k app values for the binding reactions were obtained by fitting the data to the equation Y ϭ B max (1 Ϫ exp (Ϫkt) ) using GraphPad Prism software.
Measurement of cGMP PDE Activity-A pH microelectrode was used to measure the change in pH (in millivolts) that results from the release of one proton for each molecule of cGMP hydrolyzed (26). In a typical assay, purified ␣T*-GDP (1 M) was preincubated with 500 nM G␤␥, 20 nM rhodopsin, and 100 M GTP␥S in HMN buffer for 20 min at 25°C, in the presence of light. PDE (100 nM) was added, and the reaction was incubated for an additional 10 min at 25°C. The substrate, cGMP (5 mM), was added to the reaction mix, and the relative cGMP hydrolysis rate was calculated based on the change in pH over time.
Additional Procedures-Protein concentrations were measured as described by Bradford (27) using bovine serum albumin as a standard. SDS-PAGE was performed by the method of Laemmli (28) in 12 or 15% acrylamide gels. GraphPad Prizm (version 4) software was used to fit experimental data.

RESULTS
The Conformationally Sensitive Switch 3 Region of ␣T Influences Rhodopsin-stimulated Activation of Transducin-Comparisons of the x-ray crystal structures for the GDP-bound (signaling-inactive) and GTP␥Sbound (signaling-competent) forms of ␣T (12) indicated that there were three regions that undergo conformational changes in response to GDP-GTP exchange. Two of these regions correspond to the conformationally sensitive segments initially identified in Ras (29) and Ef-Tu (30), designated as Switch 1 and Switch 2. The third region, designated as Switch 3, undergoes changes in ␣T and other G␣ subunits but not in Ras nor in related small G proteins. Switch 3 was previously shown to play a key role in linking the binding of Switch 2 to target/effector proteins, with the stimulation of effector activity (19). Thus, removal of the entire Switch 3 loop, as well as mutation of the conserved Glu-232 in this segment of ␣T, yields an ␣T subunit that is capable of binding to the PDE in a GTP-dependent manner but is unable to stimulate PDE activity (cyclic GMP hydrolysis). These findings raised the possibility that Switch 3 may be communicating with another region of the ␣T subunit that was directly responsible for effector regulation. Fig. 1a shows the x-ray crystal structure for the GDP-bound form of retinal ␣T (10), with the Switch 3 segment highlighted in magenta. Arginine 238 is a conserved residue within Switch 3. In GDP-bound ␣T, Arg-238 contacts Glu-39 (Fig. 1b), which is in the phosphate-binding P-loop lying just upstream from the beginning of the large helical domain, whereas in the activated (GTP␥S-bound) ␣T subunit, Arg-238 contacts both Glu-39 and Gln-143 of the helical domain. It has been suggested that the helical domain of G␣ subunits plays a role in mediating the regulation of target/effectors, and in the particular case of ␣T, may be essential for the stimulation of PDE activity (31)(32)(33). Thus, we wanted to see whether disruption of the apparent links between the conserved Arg-238 in Switch 3 and Glu-39 and/or Gln-143 might compromise ␣T activation and/or its ability to bind and regulate the PDE. To facilitate the expression of different ␣T mutants in E. coli, point mutations were prepared in the background of an ␣T/␣i1 chimera referred to as ␣T* (see "Experimental Procedures"). Unlike wild-type ␣T, the ␣T* subunit is expressed in a soluble form in E. coli and (like retinal ␣T) is fully capable of functionally coupling to rhodopsin (15).
The results presented in Fig. 2 show that changing the arginine at position 238 to either alanine or lysine, within the ␣T* background, had only modest effects on the ability of ␣T* to respond to rhodopsin and G␤␥ and to undergo GDP-[ 35 S]GTP␥S exchange. The same appears to be true when Arg-238 is changed to glutamine (20). However, when Arg-238 was changed to glutamic acid, the resulting ␣T* mutant was completely ineffective in undergoing rhodopsin-stimulated activation.
The ␣T*(R238E) Mutant Is Unable to Activate the Cyclic GMP PDE-We also examined the abilities of different position 238 mutations within an ␣T* background to stimulate PDE activity. Both the ␣T*(R238A) and the ␣T*(R238K) mutants, upon binding to GTP␥S in a rhodopsin-dependent manner, were compromised in their ability to stimulate PDE activity (Fig. 3a). Their partial ability to stimulate the target/effector may reflect the importance in maintaining the linkage between Arg-238 and Glu-143 of the helical domain for proper effector regulation. However, the ␣T*(R238E) mutant was completely incapable of stimulating PDE activity, consistent with its inability to become activated in response to rhodopsin. Still, we wondered whether the ␣T*(R238E) mutant might be able to bind to PDE and prevent wild-type ␣T* from binding and stimulating the target/effector. However, as shown in Fig. 3b, this was not the case, as there was no detectable inhibition of the ␣T*-mediated stimulation of PDE activity when excess ␣T*(Arg-238) was preincubated with PDE prior to the addition of activated (GTP␥S-bound) ␣T* to the assay.
The ␣T*(R238E) Mutant Is Unresponsive to Aluminum Fluoride Treatment-We also examined whether substitutions for the conserved Arg 238 in Switch 3 had any effect on the ability of ␣T* to respond to aluminum fluoride (AlF 4 Ϫ ) and undergo activating conformational changes (34,35). Substituting either alanine or lysine for arginine at position 238 had no effect on the ability of ␣T* to respond to AlF 4 Ϫ when monitoring Trp-207 fluorescence (Fig. 4). However, the ␣T*(R238E) mutant was unresponsive to AlF 4 Ϫ , showing no detectable change in its intrinsic tryptophan fluorescence.
The ␣T*(R238E) Mutant Is Nucleotide-depleted-The inability of ␣T*(R238E) to respond to AlF 4 Ϫ , together with the fact that AlF 4 Ϫ only binds to the GDP-bound form of G␣ subunits, prompted us to examine the nucleotide-bound state of R238E mutant by HPLC. Fig. 5, a and b, show the HPLC profiles obtained following treatment with acetonitrile to release any prebound GDP or GTP from ␣T* and ␣T*(R238E), respectively. Fig. 5c shows the control elution profiles for GDP and GTP. It was clear that although ␣T* was purified in the GDP-bound state, thus accounting for its ability to respond to AlF 4 Ϫ , ␣T*(R238E) was isolated in a nucleotide-free state. This explained the inability of the R238E mutant to bind AlF 4 Ϫ . However, although it has been generally assumed that nucleotide-depleted G␣ subunits are highly susceptible to proteolysis and degradation, somewhat surprisingly, the ␣T*(R238E) mutant did not show an enhanced sensitivity to trypsin, and in fact, showed a greater degree of resistance than did the GDP-bound form of retinal ␣T or the recombinant ␣T* protein. For example, it has been well established that when the GDP-bound form of retinal ␣T is treated with trypsin, a sequence of proteolytic events occurs that culminates in the cleavage of arginine 204 in Switch 2, yielding fragments of ϳ23 and ϳ10 kDa (36,37). When activated forms of ␣T (either bound to AlF 4 Ϫ or bound to GTP␥S) are exposed to trypsin, the proteolytic site in Switch 2 is protected, as an outcome of the activating conformational change, thus resulting in the production of a stable fragment of ϳ34 kDa. The same was true when the recombinant ␣T* subunit was treated with trypsin. Fig. 6 shows that the 23-kDa fragment was generated when proteolysis of ␣T* was performed in the absence of AlF 4 Ϫ , whereas complete proteolysis was blocked when ␣T* was treated with the activating agent, yielding first an ϳ38-kDa fragment and then the 34-kDa fragment. When the same experiments were performed with the ␣T*(R238E) mutant, there was a slight shift in the mobility of the two core fragments of ␣T*(R238E), perhaps because of the point mutation; however, there was little or no generation of a fragment in the range of 20 -23 kDa. These results suggested that the conformation of the Switch 2 domain of ␣T*(Arg-238) at least partially mimics that of the activated   state, thus protecting the trypsin-sensitive Arg-204 residue and enabling this nucleotide-free mutant to exhibit increased resistance to proteolytic degradation.
The ␣T*(R238E) Mutant Blocks Rhodopsin-stimulated Nucleotide Exchange on ␣T*-Given that nucleotide-binding-defective mutants of Ras and related small G proteins tightly associate with their upstream activators (guanine nucleotide exchange factors or GEFs) and act as dominant-negative inhibitors (38,39), we then asked whether the ␣T*(R238E) mutant might block the functional coupling of rhodopsin to transducin. Indeed, we found that ␣T*(R238E) was a potent inhibitor of rhodopsin-stimulated GDP-[ 35 S]GTP␥S exchange on ␣T*. Fig. 7a shows that when ␣T*(R238E) was added together with ␣T* to an assay incubation containing rhodopsin and G␤␥, the presence of the R238E mutant caused a significant reduction in the rhodopsin-dependent binding of [ 35 S]GTP␥S by ␣T*. The inhibitory effects by ␣T*(R238E) were enhanced when the mutant was preincubated with rhodopsin and G␤␥, prior to adding ␣T* (Fig. 7a). Likewise, preincubation of the ␣T*(R238E) mutant with rhodopsin and G␤␥ potently inhibited the rhodopsin-dependent stimulation of PDE activity by ␣T* (Fig. 7b). However, consistent with our previous findings (Fig. 3b), when ␣T* was first activated by rhodopsin in the absence of ␣T*(R238E), the (R238E) mutant was no longer able to inhibit PDE activity. Fig. 8a shows that the inhibition of the rhodopsin-dependent activation of ␣T* by ␣T*(R238E) is dose-dependent and requires that ␣T*(R238E) is present in excess over ␣T*. However, the ability of ␣T*(R238E) to inhibit the rhodopsin-stimulated activation of ␣T* was not dependent on G␤␥. In fact, when the ␣T*(R238E) mutant was preincubated with rhodopsin in the absence of G␤␥, it was still able to effect an essentially complete inhibition of the rhodopsin-dependent GDP-[ 35 S]GTP␥S exchange activity of ␣T* (i.e. as assayed upon the addition of ␣T* together with excess G␤␥, Fig. 8b). Overall, these results indicated that although ␣T*(R238E) was incapable of undergoing rhodopsin-stimulated GDP-GTP␥S, it nonetheless was able to block rhodopsin-catalyzed G protein activation.

DISCUSSION
G proteins act as molecular switches in a variety of signaling pathways by linking the initial activation of cell surface GPCRs to the regulation of effector enzymes or ion channels. The two most fundamentally important steps in the actions of G proteins are the GTP binding event, which occurs as an outcome of receptor-stimulated GDP-GTP exchange, and GTP hydrolysis, which serves to terminate the signal and return the G protein to its basal, inactive state. The activation step needs to be tightly regulated, in the cases of both large and small G proteins, as the loss of regulation of GDP-GTP exchange results in constitutively active G proteins that have been implicated in a number of disease states (5,40). A good deal of information is now available regarding the fundamental mechanisms underlying the activation of small G proteins by their upstream activators (GEFs) (41,42). Based on x-ray crystal structures for different small G protein-GEF complexes, as well as various lines of biochemical study, it appears that the principle actions of GEFs are the perturbation of Mg 2ϩ binding, as Mg 2ϩ strongly influences the affinity of the G protein for GDP and the destabilization of the binding of the guanine ring moiety and the phosphate residues (43,44). In the case of the G␣ subunits, the underlying mechanism of activation by GPCRs is less well understood, owing to a lack of structural information for GPCR-G protein complexes.  The G␣ subunits hold a special challenge for the activation event, given that they are comprised of a conserved guanine nucleotide-binding domain, which is shared by Ras and other small G proteins and accordingly referred to as the GTPase domain, and a second large helical domain that has been described to fit over the GTPase domain like a "clam shell" (10 -14). This implies that as a first step in the activation of large G proteins, GPCRs need to alter the juxtaposition of the helical domain relative to the GTPase domain. We have shown that mutations of conserved residues in ␣T* that are part of the linker regions connecting the helical and GTPase domains result in constitutive GDP-GTP exchange (15). However, the rate of nucleotide exchange measured for the ␣T* linker mutants is still slow relative to the rate of rhodopsinstimulated nucleotide exchange. Thus, we have examined additional types of mutations in attempting to understand more about the regions on ␣T involved in regulating the activation event, with the ultimate aim being to identify a constitutively active G␣ mutant that fully mimics the functional capability of a GPCR-activated G protein. This led us to examining the possible interactions between the conformationally sensitive Switch 3 loop and the helical domain.
The Switch 3 loop undergoes changes upon GDP-GTP exchange exclusively in the G␣ subunits of large G proteins, thereby distinguishing it from the conformationally sensitive Switch 1 and 2 regions that are found in both small G proteins as well as in the ␣ subunits of large G proteins. We had originally found that the Switch 3 loop plays a key role in linking the binding of PDE by activated ␣T (via its Switch 2 domain) to the stimulation of PDE activity (19). Given that Switch 3 can make contacts with the helical domain and because it had been proposed that the helical domain may be involved in the stimulation of PDE activity FIGURE 7. ␣T*(R238E) blocks the rhodopsin-stimulated activation of ␣T*. a, either ␣T*(f) (500 nM) alone or ␣T* (500 nM) plus ␣T*(R238E) (3 M) (OE) was incubated together with rhodopsin (4 nM) and G␤␥ (500 nM) at 25°C for 20 min in the presence of light before [ 35 S]GTP␥S was added to start the reaction. In some cases, ␣T*(R238E) (3 M) () was first preincubated (preinc.) with rhodopsin and G␤␥ for 20 min at 25°C, and then ␣T* (500 nM) was added and incubated for an additional 20 min at 25°C, before initiating the assay. ␣T* was omitted in the control reaction (ࡗ). Aliquots were withdrawn at the times indicated, filtered, and counted. Binding of GTP␥S to proteins is expressed as the amount of GTP␥S bound (in picomoles) as a function of time (minutes). The results are representative of three experiments. b, PDE6 activity was measured as described in the legend for Fig. 3a under conditions in which rhodopsin (20 nM) and G␤␥ (500 nM) were incubated with ␣T* (500 nM), alone (blue histogram) or conditions in which rhodopsin and G␤␥ were preincubated with ␣T*(R238E) (3 M) prior to the addition of ␣T* (red histogram). In another experiment, ␣T* (500 nM) was first activated by rhodopsin and G␤␥ (designated ␣T*-pre-activated) prior to the addition of ␣T*(R238E) (3 M) and assaying for PDE activity (green histogram). ␣T* was omitted in the blank reaction. The results are representative of three experiments. when G␤␥ was absent from the preincubation). In other cases, ␣T* was assayed in the presence of rhodopsin and G␤␥ (f) or in the absence of rhodopsin (E), and ␣T*(R238E) was assayed in the presence of rhodopsin and G␤␥ (OE). The results are representative of three experiments (31)(32)(33), it was attractive to consider that interactions between Switch 3 and the helical domain might be essential for effector regulation. Since we already had shown that changing the position of the helical domain relative to the GTPase domain can influence GDP-GTP exchange, we also wondered whether disrupting interactions between Switch 3 and the helical domain might impact G␣ activation. In the present study, we have looked at this in some detail by mutating a conserved residue in Switch 3, Arg-238, which, based on the available x-ray structures for ␣T, appeared to make contacts with Glu-39 from the P-loop within the amino-terminal portion of the ␣T subunit, as well as with Gln-143 (the latter contact occurring when ␣T is in an activated conformation).
Certain changes at position 238, when made in the background of the recombinant ␣T* subunit, appeared to be largely without effect when assaying rhodopsin-stimulated nucleotide exchange; these included a conserved substitution (i.e. lysine for arginine) or when changing the arginine to alanine or glutamine (20). Changes at position 238 did impact the ability of the recombinant ␣T* subunit to stimulate PDE activity to varying extents. The R238A and R238K mutants reduced the extent of activation by about 50%, whereas an R238Q mutant showed an even greater impairment (20). These findings may reflect the importance of maintaining proper Switch 3 domain-helical domain interactions for full effector regulation. However, when we changed Arg-238 to glutamic acid, we completely eliminated the ability of the ␣T* subunit to undergo rhodopsin-stimulated GDP-GTP exchange. Consequently, this mutant ␣T* subunit was also unable to stimulate PDE activity.
At the moment, we do not fully understand why this single point mutation has such a dramatic effect on rhodopsin-dependent nucleotide exchange. It may have something to do with disrupting the normal interaction of Arg-238 with Glu-39, which lies in the P-loop. The x-ray crystal structures for ␣T also suggested that Arg-238 comes into proximity of the conserved NKXD motif, which participates in binding the guanine ring moiety. Thus, perhaps reversing the side-chain charge at position 238 (i.e. Arg to Glu) caused some destabilization of the guanine ring. Still, what was particularly striking and unexpected was the finding that the ␣T*(R238E) mutant inhibited the ability of rhodopsin to activate wild-type ␣T* subunits. Thus, as depicted in Fig. 9, the R238E mutant has the ability to act as a dominant-negative inhibitor. This carries a number of interesting implications. One concerns the traditional difficulty in generating nucleotide-depleted G␣ mutants. Although the nucleotide-depleted forms of Ras-like G proteins work extremely well as dominant-negative inhibitors by forming stable complexes with their respective GEFs and blocking GEF-stimulated activation of the endogenous G protein counterparts (38), nucleotide-depleted forms of G␣ subunits can be extremely unstable. This then raises the question of how ␣T*(R238E) was able to remain sufficiently stable so as to effectively act as a dominant-negative inhibitor. The answer may lie in its ability to assume what appears to be a partially activated conformation, as judged from its relative insensitivity to trypsin treatment. Specifically, the ␣T*(R238E) mutant yielded a stable core fragment after exposure to trypsin, similar to what is observed with activated forms of both retinal ␣T and recombinant ␣T* and distinct from what one sees when treating GDP-bound forms of the ␣T subunits with the protease (resulting in further degradation and the production of a 23-kDa fragment). Thus, although nucleotide-free forms of G␣ subunits are thought to be especially sensitive to proteolytic degradation, the ␣T*(R238E) mutant, apparently by virtue of disrupting Switch 3, adopted a conformation that exhibits resistance to proteolysis and therefore represents a relatively stable G␣ subunit.
In conclusion, we have shown that a single point mutation of a conserved residue in the Switch 3 domain of a G␣ subunit significantly impacts its ability to bind guanine nucleotides and yields a unique type of dominant-negative inhibitor. Recently, a mutation in the ␣ subunit of the G s protein (␣ s (S54N)), which corresponds to the well known Asn-17 mutations in Ras (38) and related small G proteins, has been shown to act as a dominant-negative inhibitor of signaling via the thyroid-stimulating hormone receptor (45). However, it seems unlikely that the ␣ s (S54N) mutant exists as a nucleotide-depleted protein because it behaves as a conditional dominant-negative inhibitor, showing some stimulation of basal adenylyl cyclase activity (46). A dominant-negative mutant of G s ␣ that contains three sets of distinct mutations, influencing different functions of the G protein, has also been reported (47). However, to our knowledge, the R238E mutant represents the first demonstration that a single point-mutated, nucleotide-binding-defective G␣ subunit can be generated and used in a manner analogous to the Asn-17 Ras mutants as an inhibitor of GPCR-coupled signaling.