Identification of Effector Residues on Photoreceptor G Protein, Transducin*

Transducin is a photoreceptor-specific heterotrimeric GTP-binding protein that plays a key role in the vertebrate visual transduction cascade. Here, using scanning site-directed mutagenesis of the chimeric Gαt/Gαi1 α-subunit (Gαt/i), we identified Gαt residues critical for interaction with the effector enzyme, rod cGMP phosphodiesterase (PDE). Our evidence suggests that residue Ile208 in the switch II region directly interacts with the effector in the active GTP-bound conformation of Gαt. Residues Arg201, Arg204, and Trp207are essential for the conformation-dependent Gαt/effector interaction either via direct contacts with the inhibitory PDE γ-subunit or by forming an effector-competent conformation through the communication network between switch II and the switch III/α3-helix domain of Gαt. Residues His244 and Asn247 in the α3 helix of Gαt are responsible for the conformation-independent effector-specific interaction. Insertion of these residues rendered the Gαt/i chimera with the ability to bind PDE γ-subunit and stimulate PDE activity approaching that of native Gαt. Comparative analysis of the interactions of Gαt/i mutants with PDE and RGS16 revealed two adjacent but distinct interfaces on transducin. This indicates a possibility for a functional trimeric complex, RGS/Gα/effector, that may play a central role in turn-off mechanisms of G protein signaling systems, particularly in phototransduction.

The visual transduction cascade in vertebrate photoreceptors represents a classical example of a G protein signaling system. In rod photoreceptor cells, light-activated rhodopsin stimulates GTP-GDP exchange on the retinal G protein, transducin, resulting in dissociation of G␣ t GTP 1 from G␤␥ t and rhodopsin. Liberated G␣ t in active GTP-bound conformation stimulates the effector enzyme, cGMP phosphodiesterase (PDE), by displacing the inhibitory ␥-subunits (P␥) from the PDE catalytic core (P␣␤). cGMP hydrolysis by active PDE results in closure of cGMP-gated channels in the plasma mem-brane (1)(2)(3). As in other G protein cascades, the lifetime of a transducin-mediated signal is linked to the intrinsic GTPase activity of G␣ t . Hydrolysis of GTP converts the G␣ t molecule to the inactive GDP-bound conformation allowing release of P␥ for re-inhibition of P␣␤. A member of the RGS family (4 -6), RGS9, and perhaps other retina-specific RGS proteins serve as GTPase-activating proteins (GAPs) for transducin (7)(8)(9). They target a transitional intermediate conformation of G␣ t during GTP hydrolysis to accelerate GTP hydrolysis and expedite the signal termination (8,10,11). The P␥ subunit assists RGS9 in its GAP function, thus providing an elegant feedback mechanism for the effector participation in quenching the visual excitation (7).
Recent progress in understanding molecular mechanisms of G protein action has been advanced by solutions of crystal structures of several G protein ␣-subunits in active GTPbound, inactive GDP-bound, and transitional, AlF 4 Ϫ -complexed conformations (12)(13)(14)(15)(16). Three regions of G␣ subunits called switches I-III change their conformation upon GTP-GDP exchange (13,16). One or more of these switches is likely to participate in effector interaction. A wealth of biochemical data on G protein interaction with effectors combined with the structural information allowed a detailed mapping of effector surfaces of G␣ subunits. The two well characterized G protein/ effector systems are G␣ s /adenylyl cyclase and G␣ t /PDE. Initially, several potential effector binding sites corresponding to the ␣2 helix (switch II), the ␣3 helix/␣3-␤5 loop, and the ␣4-␤6 loop were identified on G␣ s using a scanning mutagenesis approach (17,18). Similar regions of G␣ t were implicated in the effector interaction. A synthetic peptide, G␣ t -(293-314) corresponding to the ␣4-␤6 region was shown to activate PDE in vitro (19,20). Sites of the chemical cross-linking of the P␥ subunit to G␣ t were also localized to within the ␣4-␤6 loop (21,22). Analysis of the interaction of G␣ t /G␣ i1 chimeras with P␥ strongly implicated the G␣ t ␣3 helix with the ␣3/␤5 loop as a major effector binding region in both basal and activated states (23). The study concluded that the ␣3-␤5 region provides for the specificity of the G␣ t -P␥ interaction. The second composite P␥ binding site is dependent on the active conformation of G␣ t and conserved between G␣ t and G␣ i1 (23). Two switch regions of G␣ t , II and III, have been proposed to participate in the interaction with P␥ (12,24) and might therefore form such a conserved domain. The switch III region contains a cluster of acidic residues that represent a speculative target for the polycationic region of P␥ which is known to interact with G␣ t . However, mutation of the acidic cluster did not affect the ability of mutant G␣ t to stimulate PDE (25). Evidence for the involvement of the switch II region of G␣ t in interaction with PDE is based on the inability of the G␣ t mutant W207F to activate the enzyme (24). The involvement of the switch II region of G␣ subunits in effector recognition has been supported recently by identification of conserved switch II residues in both G␣ i and G␣ s required for interaction with adenylyl cyclase (26). The role of the switch II domain and the ␣3-␤5 region as major regions for G␣/effector interactions that emerged from biochemical studies has just been confirmed by the crystal structure of G␣ s complexed with catalytic domains of adenylyl cyclase (27,28).
To define roles of the switch II and ␣3-␤5 regions in the G␣ t /PDE interaction and identify specific residues required for the effector binding we carried out scanning mutational analysis of these regions in the G␣ t /G␣ i1 chimera (G␣ t/i ) containing G␣ t switch II and G␣ i1 ␣3-␤5. Conserved surface-exposed residues of the G␣ t/i switch II region were replaced by Ala residues to delineate conserved conformation-dependent effector interactions. G␣ i1 residues that are different between G␣ i1 and G␣ t within the ␣3-␤5 region of G␣ t/i were replaced by corresponding G␣ t residues to identify effector-specifying residues of G␣ t .

EXPERIMENTAL PROCEDURES
Preparation of ROS Membranes, G␣ t GDP, G␤␥ t , and P␥BC-Bovine ROS membranes were prepared as described previously (29). Ureawashed ROS membranes (uROS) were prepared according to protocol in Yamanaka et al. (30). G␣ t GDP was prepared and purified by chromatography on Blue Sepharose CL-6B as described in Ref. 31. G␤␥ t was purified according to Kleuss et al. (32). P␥BC was obtained and purified as described previously (33). The purified proteins were stored in 50% glycerol at Ϫ20°C or without glycerol at Ϫ80°C.
Site-directed Mutagenesis of Chimeric G␣ t/i -Mutagenesis of G␣ t/i switch II residues was performed using the vector for expression of His 6 -tagged G␣ t /G␣ i1 chimera 8 (G␣ t/i ) as a template for PCR amplifications (23). The R201A, E203A, R204A, K205A, and K206A substitutions were introduced using 5Ј-primer 1 and 3Ј-primers 2, 3, 4, 5, and 6, respectively, for PCR amplification (see below). The PCR products were digested with NcoI and BamHI and subcloned into the NcoI/BamHI digested pHis 6 G␣ t/i . The 5Ј-primer 1 and 3Ј-primers 7 and 8 were used to obtain the PCR products carrying the I208A and W207F mutations, respectively. The PCR products were blunt-ended with Klenow fragment, digested with NcoI, and ligated into pHis 6 G␣ t/i . Prior to the ligation pHis 6 G␣ t/i was digested with BamHI, blunt-ended with mung bean nuclease, and then cut with NcoI. The E212A and E212N/G213D substitutions were introduced by PCR using 5Ј-primers 9 and 10, respectively, and a 3Ј-primer 11. The PCR products were cut with BamHI and HindIII and subcloned into the BamHI/HindIII-digested pHis 6 G␣ t/i . The Met 247 -Lys 248 -Asp 251 3 Leu 243 -His 244 -Asn 247 (LHN), Met 247 3 Leu 243 (L), Lys 248 3 His 244 (H), and Asp 251 3 Asn 247 (N) mutations were encoded by 5Ј-primers 12, 13, 14, and 15, respectively, and used for PCR with 3Ј-primer 11. The Lys 248 -Asp 251 3 His 244 -Asn 247 (HN) mutant was generated by PCR with primers 14 and 11 using the LHN template. The PCR products were digested with SphI and HindIII and inserted into pHis 6 G␣ t/i . A silent SpeI site was created in pHis 6 G␣ t/i using QuikChange kit (Stratagene). pHis 6 G␣ t/i was a template for PCR reaction using oligonucleotides 16 and 17, and Pfu DNA polymerase. The PCR products were treated with DpnI specific for methylated and hemimethylated DNA and then transformed into E. coli DH5␣. pHis 6 G␣ t/i -SpeI was used for PCR-directed mutagenesis with 5Ј-primer 1 and 3Ј-primers 18 and 19 to generate a triple mutant Asn 256 -Lys 257 - The sequences of all mutants were verified by automated DNA sequencing at the University of Iowa DNA Core Facility. G␣ t/i and all mutants were expressed and purified as described previously (23). The purified proteins were tested in the trypsin protection assay as described previously (34).
Binding of G␣ t/i and Its Mutants to GST-RGS16 -G␣ t/i or its mutants (1 M final concentration) were mixed with glutathione-agarose retaining ϳ10 g of GST-RGS16 in 200 l of 20 mM HEPES buffer (pH 7.6), 100 mM NaCl, 4 mM MgCl 2 (buffer A) containing 30 M AlCl 3 and 10 mM NaF. After incubation for 20 min at 25°C, the agarose beads were spun down and washed three times with 1 ml of buffer A, and the bound proteins were eluted with a sample buffer for SDS-PAGE.
Single Turnover GTPase Assay-Single turnover GTPase activity measurements were carried out in suspensions of uROS membranes (5 M rhodopsin) reconstituted with G␣ t/i or its mutants (2 M) and G␤␥ t (1 M) essentially as described in Refs. 11 and 35. Bleached uROS membranes were mixed with RGS16 (1 M) and preincubated for 5 min at 25°C. The GTPase reaction was initiated by addition of 100 nM [␥-32 P]GTP (ϳ4 ϫ 10 5 dpm/pmol). The GTPase rate constants were calculated by fitting the experimental data to an exponential function: %GTP hydrolyzed ϭ 100(1 Ϫ e Ϫkt ), where k is a rate constant for GTP hydrolysis.
Fluorescence Assays-Fluorescence assays of interaction between G␣ t and P␥BC were performed on a F-2000 fluorescence spectrophotometer (Hitachi) in 1 ml of buffer A essentially as described in Ref. 33. Where indicated, the buffer A contained 30 M AlCl 3 and 10 mM NaF. Fluorescence of P␥BC was monitored with excitation at 445 nm and emission at 495 nm. Concentration of P␥BC was determined using ⑀ 445 ϭ 53,000. Time traces of AlF 4 Ϫ -induced increase in the tryptophan fluorescence of G␣ t/i and its mutants (200 nM) were recorded on an AB2 fluorescence spectrophotometer (Spectronic Instruments) in a stirred 1-ml cuvette with excitation at 280 nm and emission at 340 nm. Stock solutions of AlCl 3 and NaF were mixed immediately prior to injection of 10 l of the mixture into a cuvette (final concentrations 30 M AlCl 3 and 10 mM NaF).
PDE Activation Assay-HoloPDE was extracted from ROS membranes and purified as described earlier (36). PDE (0.2 nM) was reconstituted with 2 M G␣ t GDP or G␣ t/i mutants and 2 M G␤␥ t in suspensions of urea-washed ROS membranes containing 10 M rhodopsin. GTP␥S (10 M) was added to the reaction mixture, and PDE activity was measured using [ 3 H]cGMP similarly as described (37).
Miscellaneous Procedures-Protein concentrations were determined by the method of Bradford (38) using IgG as a standard or using calculated extinction coefficients at 280 nm. SDS-polyacrylamide gel electrophoresis was performed by the method of Laemmli (39) in 12% acrylamide gels. Rhodopsin concentrations were measured using the difference in absorbance at 500 nm between "dark" and bleached ROS preparations. Fitting of the experimental data was performed with nonlinear least squares criteria using GraphPad Prizm (version 2) software. The results are expressed as the mean Ϯ S.E. of triplicate measurements. Examination of the crystal structures of G␣ t , G␣ i , and RGS4 was performed using RasMol (version 2.6) software.

Expression and Characterization of Mutant G␣ t /G␣ i1 Proteins with Substitutions in the G␣ t Switch II Region-
The following amino acid residues of G␣ t switch II that are surfaceexposed based on the crystal structure of G␣ t GTP␥S, were selected for the scanning mutagenesis (12): Arg 201 , Ser 202 , Arg 204 , Lys 205 , Lys 206 , Trp 207 , Ile 208 , His 209 , Glu 212 , and Gly 213 . The surface exposure of Trp 207 is limited, but this residue had been implicated in the effector interaction previously (24). In addition, we included in our analysis a G␣ t mutant E203A as its GDP-bound form was reported to be constitutively active toward PDE (40). Conserved G␣ t residues Arg 201 , Arg 204 , Lys 205 , Lys 206 , and Ile 208 were replaced by Ala residues. Trp 207 was substituted by Phe because the G␣ t W207F mutant was properly folded when characterized previously (24). Nonhomologous substitutions of the G␣ t residues involved in the interactions with RGS proteins, S202D and H209Q, have been generated earlier (41) and utilized in this study. Residues Glu 212 and Gly 213 of G␣ t are conserved between G␣ t and G␣ i , but G␣ s has the Asn-Asp pair instead. Two mutants, E212A and a double mutant E212N/G213D, were made to assess the role of these residues. All mutations of G␣ t residues were introduced into G␣ t /G␣ i1 chimeric protein, Chi8, which is efficiently expressed in E. coli. (23). Chi8 contains more than 80% of the G␣ t sequence including all three G␣ t switch regions and is fully competent to interact with activated rhodopsin and G␤␥ t (23,41). For clarity, we refer to Chi8 as G␣ t/i . Expression of G␣ t/i and all switch II mutants yielded similar amounts of soluble protein (ϳ5 mg/liter of culture).
To critically evaluate the potential role of mutations within the switch II region, it is essential to confirm that these mutations do not interfere with the ability of G␣ to undergo conformational change to the active state. Activation of G␣ t GDP upon exchange with GTP␥S or binding of AlF 4 Ϫ leads to an increase in Trp 207 fluorescence and to protection of the switch II region from cleavage with trypsin (24,42). Time-traces of AlF 4 Ϫ -induced tryptophan fluorescence increase for G␣ t , G␣ i1 , G␣ t/i , and the G␣ t/i switch II mutants are shown in Fig. 1. The fluorescence changes were ϳ40 -45% for G␣ t , and G␣ i1 and ϳ35% for G␣ t/i , with similar fast kinetics for G␣ i1 and G␣ t/i , whereas the kinetics for G␣ t were slower (Fig. 1). Except for R204A, all G␣ t/i mutants had comparable (ϳ25-40%) enhancement in tryptophan fluorescence upon addition of AlF 4 Ϫ . However, the kinetics of the fluorescence change varied substantially for different mutants from slow, G␣ t -like, to fast, analogous to G␣ i . The R204A mutant had a very low degree of fluorescent enhancement (Ͻ5%). In the crystal structure of G␣ t GDP⅐AlF 4 Ϫ (G␣ t ⅐AlF 4 Ϫ ) side chains of Arg 204 and Trp 207 contact one another (15). Perhaps loss of the Arg 204 /Trp 207 contact and the Arg 204 / Glu 241 salt bridge (43) in the R204A mutant are responsible for the low fluorescence enhancement. Four mutants, R201A, E203A, R204A, and W207F, have displayed a low degree of proteolytic protection (10% or less) in the trypsin protection assay in the presence of AlF 4 Ϫ (not shown). Similar reduction in trypsin resistance was observed for analogous mutants of G␣ t translated in vitro or expressed in COS-7 cells (44). The most stringent criterion for proper folding of G␣ t/i mutants is the ability to hydrolyze GTP. GTPase activity of G␣ t/i mutants attests to their ability to interact with G␤␥ t and rhodopsin to exchange GTP for GDP and to undergo activation-inactivation conformational changes. All tested G␣ t/i switch II mutants had basal GTPase rates ranging from 0.010 s Ϫ1 to 0.025 s Ϫ1 , which are comparable to those of G␣ t/i (0.015 s Ϫ1 ) (Table I) and native G␣ t (0.019 s Ϫ1 ) measured under similar conditions (11). Several of the mutated residues, Arg 201 , Glu 203 , Lys 206 , Trp 207 , and His 209 , contact G␤␥ t (45), which may account for some variability in observed GTPase rates. A somewhat lower GTPase rate for the E203A mutant is consistent with the catalytic properties of the G␣ i E207A mutant (34). Appropriate GTPase activities of R201A, E203A, R204A, and W207F mutants indicate that reduced trypsin protection of these mutants may result from the differences in primary structures rather than distortion of the switch II conformation. Side chains of G␣ t Arg 201 , Arg 204 , and Trp 207 may be involved in preventing tryptic cleavage of G␣ t ⅐AlF 4 Ϫ by forming ordered interactions with the switch III/␣3 region (13).
Interaction of G␣ t/i Switch II Mutants with the P␥ Subunit-To identify switch II residues of G␣ t essential for effector binding we employed an assay that utilizes the P␥ subunit labeled with the environmentally sensitive fluorescent probe, 3-(bromoacetyl)-7-diethyl aminocoumarin (P␥BC) (33). Binding of G␣ t GTP␥S or G␣ t ⅐AlF 4 Ϫ to P␥BC causes ϳ7-fold maximal increase in the probe fluorescence. Using this assay, affinities of G␣ t GTP␥S or G␣ t ⅐AlF 4 Ϫ for P␥BC are similar (K d values 2-4 nM), whereas G␣ t GDP affinity for P␥BC drops by ϳ20-fold to a K d of 75 nM (11,33). Addition of G␣ t/i ⅐AlF 4 Ϫ to P␥BC led to a maximal fluorescence enhancement F/F 0 of 3-fold. From the binding curve, a K d for this G␣ t/i ⅐AlF 4 Ϫ /P␥BC is 48 nM (Fig. 2A,  Table I). As expected, an affinity of G␣ t/i ⅐AlF 4 Ϫ for P␥BC was lower than that of G␣ t ⅐AlF 4 Ϫ , since in G␣ t/i the G␣ t ␣3-␤5 region is replaced by the G␣ i1 region. The interaction between G␣ t/i and P␥BC was conformation-dependent. The binding curve of G␣ t GDP to P␥BC as measured by the fluorescence increase exhibited a K d of 0.98 M and F/F 0 of 4.0 (Fig. 2B). Similar relative increases in affinity of G␣ t/i and G␣ t for P␥BC upon adoption of an active conformation suggest that G␣ t/i represents a well suited model protein for mutational analysis to identify residues involved in the GTP-dependent G␣ t /effector interaction. Four mutants of G␣ t/i switch II, R201A, R204A, W207F, and I208A in the AlF 4 Ϫ conformation had dramatically reduced affinity for P␥BC (Table I and Fig. 2A). The calculated K d values for R201A and R204A were 1.3 M and 1.5 M, respectively, whereas the K d values for W207F (3.4 M) and I208A (3.0 M) were even higher. Interestingly, the estimated affinities of these four mutants for P␥BC did not change significantly in the GDP-bound conformations (Table I, Fig. 2B), suggesting that the mutants have lost the conformation-dependent interaction between the G␣ t/i switch II domain and P␥BC. All other G␣ t/i switch II mutants in both the active and inactive conformations had affinities for P␥BC comparable to the respective affinities of G␣ t/i ⅐AlF 4 Ϫ or G␣ t/i GDP for the effector protein. Despite a low level of protection in the trypsin sensitivity test, the E203A mutant was fully capable of the conformation-dependent interaction with the effector. Its binding affinity for P␥BC was enhanced by ϳ30-fold in the presence of AlF 4 Ϫ , similarly as it was seen for G␣ t/i . These results do not support previously reported constitutive activity of G␣ t E203AGDP toward PDE (40). Residues Ser 202 , Glu 203 , Lys 205 , Lys 206 , and His 209 of G␣ t participate in the interaction with RGS proteins based on the crystal structure of RGS4 with G␣ i1 ⅐AlF 4 Ϫ (46). Substitutions of these residues in G␣ t/i did not significantly alter its affinity for P␥BC (Table I).
Interaction of G␣ t/i Switch II Mutants with RGS16 -The switch II region of G␣ subunits is an important domain for binding to RGS proteins (46). To further delineate spatial relationships between the effector and RGS interacting surfaces on G␣ t , we examined the ability of a human homologue of mouse retina-specific mRGSr (hRGSr or RGS16) (8,11,47) to bind the G␣ t/i switch II mutants and stimulate their GTPase activity. In addition, the competence of G␣ t/i mutants to interact with RGS may also serve as an indication of the mutant's ability to assume a proper AlF 4 Ϫ -bound conformation. RGS16 as with most other RGS proteins binds preferentially to the AlF 4 Ϫ conformation of G protein ␣-subunits (8,10,11,48). Binding between the G␣ t/i switch II mutants and RGS16 was evaluated using precipitation of mutants by glutathione-agarose beads containing immobilized GST-RGS16. The binding assay demonstrated that mutants R201A and R204A had modestly decreased affinity for RGS16, whereas E203A, K205A, and W207F had significant impairment of RGS16 binding (Fig. 3). Two mutants, S202D (41) and K206A (not shown), did not bind to RGS16 using this assay. The binding properties of the mutants correlated well with the capacity of RGS16 to stimulate their GTPase activity in a single turnover GTPase assay (Fig.  4, Table I). GTPase activity of the majority of G␣ t/i mutants was enhanced by ϳ4.5-7-fold in the presence of 1 M RGS16, whereas the response of E203A, K205A, and W207F mutants was notably lower (ϳ2-3-fold) (Fig. 4, Table I). RGS16 failed to stimulate GTPase activity of the G␣ t/i S202D and K206A mutants (Table I). Explaining our results, side chains of residues Ser 202 , Glu 203 , Lys 205 , and Lys 206 directly interact with RGS (46). W207F is the only substitution of a non-RGS contact residue that significantly diminished G␣/RGS interaction.
Interaction between G␣ t/i Mutants with Substitutions within the ␣3-␤5 Region and P␥BC-Evidence suggests that the region G␣ t -(237-270) (the ␣3 helix and the ␣3-␤5 loop of G␣ t ) participates in a specific interaction between P␥ and both GDP and GTP-bound conformations of transducin (23). Overall structures of G␣ t and G␣ i including the ␣3-␤5 regions are very similar (12,14). Therefore, the specificity of G␣ t for P␥ within this region is based on the primary structure differences. A limited number of G␣ t residues within region G␣ t -(237-270) are replaced by nonhomologous residues of G␣ i . Initially we generated the following three mutants of G␣ t/i in which G␣ i1 residues were substituted by the corresponding G␣ t residues: a triple mutant Met 247 -Lys 248 -Asp 251 3 Leu 243 -His 244 -Asn 247 (LHN), a triple mutant Asn 256 -Lys 257 -Trp 258 3 His 252 -Arg 253 -Tyr 254 (HRY), and a double mutant Thr 260 -Asp 261 3 Ala 256 -Thr 257 (AT). The yields of soluble HRY and AT mutants expressed in E. coli were similar to that of G␣ t/i (ϳ5 mg/liter of culture). Expression levels for the LHN mutant were 4 -5-fold lower. However, all three mutants had GTPase activity, trypsin protection, and tryptophan fluorescence enhancement properties similar to G␣ t/i (not shown). The prediction is that substitutions of the G␣ t effector-specific residues for G␣ i1 residues in Ϫ , are shown.

FIG. 3. Binding of G␣ t/i switch II mutants to GST-RGS16. SDSpolyacrylamide gel (12%) stained with Coomassie Blue. Binding of G␣ t/i and its mutants complexed with GDP and AlF 4
Ϫ to GST-RGS16 immobilized on glutathione-agarose was performed as described under "Experimental Procedures." G␣ t/i would lead to an increased affinity of such mutant(s) for P␥BC. Fig. 5 shows binding of LHN, HRY, and AT mutants in AlF 4 Ϫ and GDP-bound conformations to P␥BC. The LHN mutant in both AlF 4 Ϫ (K d 6 nM) and GDP-bound (K d 35 nM) conformations demonstrated a strong increase in affinity for P␥BC (Fig. 5, Table I). Furthermore, the K d value and the larger maximal fluorescence enhancement (F/F 0 ϳ6-fold) are comparable to those measured for interaction of native G␣ t with P␥BC (11,33). Next, we generated three mutants with single substitutions Met 247 3 Leu 243 (L), Lys 248 3 His 244 (H), and Asp 251 3 Asn 247 (N). Expression levels of the L mutant were similar to the yields of the LHN mutant and ϳ4 -5-fold lower then those for the H and N mutants suggesting that Leu 243 of G␣ t is responsible for decreased expression of G␣ t /G␣ i chimeras containing the G␣ t -(237-270) segment (23). Mutants H and N in both AlF 4 Ϫ and GDP-bound conformations showed improved interaction with P␥BC (Fig. 5, Table I). Based on the results obtained with the single substitutions, a double mutant, Lys 248 -Asp 251 3 His 244 -Asn 247 (HN) was made and tested for effector binding. The HN mutation enabled G␣ t/i to bind P␥BC (K d of 8 nM) with the affinity only ϳ2-fold lower than that for native G␣ t (Fig. 5, Table I).
Activation of Rod HoloPDE by G␣ t/i Mutants-Binding between G␣ t and P␥ represents a simple and informative model assay frequently used to assess the transducin/effector interaction. However, to activate PDE in vivo transducin ultimately interacts with P␥ subunits that are not free but complexed with PDE catalytic subunits (1). We tested the ability of G␣ t/i mutants to stimulate activity of holoPDE reconstituted with ureawashed rod outer segment (uROS) membranes and G␤␥ t in the presence of GTP␥S. G␣ t/i as well as the HRY, AT, and L mu-tants activated holoPDE very weakly (Fig. 6). Mutants H and N stimulated PDE activity by ϳ3-4 fold. The triple substitution LHN and the double substitution HN conferred to G␣ t/i the ability to stimulate basal PDE activity by ϳ5-6-fold, which was approaching the effects of native G␣ t under the same conditions (Fig. 6).

DISCUSSION
Heterotrimeric G protein G␣ subunits upon GTP-GDP exchange assume an active GTP-bound conformation that has significantly enhanced affinity for effector proteins. Three regions of G␣ subunits undergo the conformational change (switches I-III) and might therefore account for conformationdependent effector activation (13,16). Experimental evidence has been accumulated supporting the role of the switch II region of G␣ i and G␣ s in interaction with adenylyl cyclase. Common and distinct effector residues within switch II regions of G␣ i and G␣ s have been identified (26). The crystal structure of G␣ s complexed with the catalytic domains of adenylyl cyclase has confirmed the G␣ s switch II effector-binding residues identified using Ala scanning mutagenesis (27,28). The data on G␣ t switch II participation in effector activation are more limited. The W207F G␣ t mutant was shown to have a 100-fold lower affinity for P␥ and was unable to activate PDE (24). The GTP␥S-induced conformational change of G␣ t brings the exposed side chain of Trp 207 into contact with the ␣3 helix (13) and as a result, Trp 207 is significantly less solvent-exposed in the active conformation of G␣ t . Therefore, the possibility remained that failure of the W207F mutant to form important contacts with the ␣3 helix and/or a general distortion of switch II, rather than the loss of the PDE contact residue, is respon- sible for the mutant inability to activate PDE. The recently reported efficient functional expression of G␣ t /G␣ i1 chimeric proteins in E. coli provided an opportunity for scanning mutational analysis of G␣ t residues (23). Our examination revealed four switch II residues, Arg 201 , Arg 204 , Trp 207 , and Ile 208 , critical for the conformation-dependent interaction between G␣ t and P␥. Most mutations of switch II had very little or no effect on the ability of G␣ t/i GDP to interact with P␥. However, mutants R201A, R204A, W207F, and I208A in the AlF 4 Ϫ conformation had dramatically reduced affinity for P␥BC compared with G␣ t/i ⅐AlF 4 Ϫ . In fact, the affinities of all four mutants for P␥ in the AlF 4 Ϫ and GDP-bound forms were similar, suggesting that they lack conformation-dependent interactions with the effector protein.
Do all four residues directly participate in the effector interaction, and how does a single substitution at any of the four positions totally disrupt this interaction? Several scenarios are plausible. Substitution of one of the effector residues could lead to the altered positioning of other residue(s). Side chains of Arg 204 and Trp 207 contact each other in the crystal structure of G␣ t ⅐AlF 4 Ϫ (15), and both residues may be reciprocally affected by a single substitution. In addition, the W207F mutation significantly diminished the G␣ t/i /RGS16 interaction, pointing out to a potential distortion of the overall conformation of the mutant switch II region. Finally, and most importantly, the crystal structure of active G␣ t shows that side chains of Arg 201 , Arg 204 , and Trp 207 form ordered interactions with residues Glu 232 , Glu 241 , Leu 245 , and Ile 249 from ␣3 and switch III (12,13). Residue Glu 232 in switch III is critical for coupling switch II to switch III upon G␣ t activation (13,25). The water-mediated contact between Glu 232 and the guanidino NH group of Arg 204 (13) might be severed in the R204A mutant. Therefore, residues Arg 201 , Arg 204 , and Trp 207 are crucial for the G␣ t / effector interaction either by means of direct interaction with P␥ or by forming a communication network between switch II and the switch III/␣3 helix domain. In contrast, residue Ile 208 is well surface-exposed and does not appear to make new GTPdependent contacts with switches II, III, or the ␣3 helix. Our data strongly argue in favor of Ile 208 directly interacting with P␥ in the active conformation of G␣ t . Comparison of effector residues of G␣ t , G␣ i2 , and G␣ s within switch II offers insight into common and individual features of G␣/effector interactions. Although the majority of effector residues are conserved between G␣ subunits, distinct conserved residues participate in binding effector molecules. Only the position corresponding to Arg 204 of G␣ t appears to be important in all three G␣ subunits (Fig. 7A). Arg 209 in G␣ i2 is the effector residue based on the biochemical evidence (26). Arg 231 of G␣ s makes a backbone contact with adenylyl cyclase (27) (Fig. 7A). Ile 208 in G␣ t corresponds to the effector-interacting Ile 213 in G␣ i2 . G␣ s exclusively has two non-conserved effector residues, Gln 236 and Asn 239 (27,28). Parallel residues, His 209 and Glu 212 of G␣ t and His 214 and Glu 217 of G␣ i2 are not involved in interactions with effector proteins (Fig. 7A).
The conserved switch II region of G␣-subunits is not sufficient to provide specificity to a particular effector among a diverse group of G␣ targets. Other G␣ regions have been implicated in effector interaction in various G protein signaling systems. The effector role of G␣ s ␣3 helix and ␣3-␤5 loop was reinforced by the crystal structure of the complex between G␣ s and catalytic domains of adenylyl cyclase (27). The G␣ t /G␣ i chimera study provided compelling evidence for the effectorspecific conformation-independent interaction domain spanning G␣ t -(237-270) (G␣ t ␣3 helix and ␣3-␤5 loop) (23). G␣ t/i utilized in this study represented an excellent tool to map effector residues within this region. Substitutions of the nonconserved G␣ i1 residues by the effector residues of G␣ t were expected to produce a "gain-of-function" effect. His 244 and Asn 247 of G␣ t have been identified as effector-interacting residues. Incorporation of these residues instead of G␣ i residues conferred upon G␣ t/i the P␥ binding potency and ability to stimulate PDE only 2-fold lower than that of native G␣ t . These residues are located in the ␣3 helix and are not directly involved in the network of interactions linking ␣2, ␣3, and ␣4 helices of G␣ t . Most likely, His 244 and Asn 247 directly interact with P␥. Interestingly, His 244 was one of three residues iden-tified as sites of cross-linking between G␣ t and P␥ (22). Also, Asn 247 aligns with Arg 256 , which was identified as an effector residue in G␣ q (49). Comparison of the ␣3-␤5 effector residues of G␣ t and G␣ s suggests that two different mechanisms of achieving effector specificity are employed by these G␣ subunits (Fig. 7A). The effector segment 279 NRWLRT 284 of G␣ s is relatively conserved with the corresponding G␣ i sequence, but it is less homologous to the aligned G␣ t segment. The G␣ s ability to stimulate adenylyl cyclase is likely due to the shift of its ␣3-␤5 loop from the switch II helix rather than due to the differences in the primary structures of G␣ s and G␣ i ␣3-␤5 loops (28). Functional properties of the G␣ t/i HRY and AT mutants indicate that the ␣3-␤5 loop is not involved in the G␣ t specific interaction with P␥. The later interaction is ensured by His 244 and Asn 247 in the ␣3 helix of G␣ t . Mapping the effectorinteracting residues in switch II and ␣3 regions on the crystal structure of active G␣ t shows they are positioned on the same "effector face" of transducin (Fig. 7B). The effector interface may also include the ␣4 helix and ␣4-␤6 loop of G␣ t earlier implicated in biochemical studies (19 -22, 50). However, an insertion of the G␣ t -(295-314) segment of G␣ t into G␣ i1 only marginally improved the latter's ability to bind P␥ (23). Several residues within the ␣4-␤6 loop of G␣ i have been implicated in inhibition of adenylyl cyclase (26). Nonetheless, it is possible that this loop is important for stabilization of the G␣ effector interface rather than in direct interactions with effector proteins (28).
In addition to its role as a conformation-sensitive effector binding domain, the switch II region of G␣ is also an important site for binding RGS proteins (46). Besides their role as GAPs (48,51,52) for the G i and G q families, RGS proteins may act as antagonists for some G protein effectors. RGS4 has been shown to block activation of phospholipase C␤ by G␣ q GTP␥S (53) and to inhibit inositol phosphate synthesis activated by AlF 4 Ϫ in COS-7 cells overexpressing G q (54). However, we demonstrated FIG. 7. A, effector-interacting residues of G␣ subunits. Boxed are the residues essential for interacton of G␣ t , G␣ i (26), and G␣ s (27,28) with effectors. Underlined are the residues of G␣ t that correspond to RGS4-contact residues of G␣ i1 (46). B, effector and RGS binding interfaces of G␣ t . Effector-interacting residues from the switch II region and the ␣3 helix are shown in magenta and green, respectively. The G␣ t switch II residues corresponding to the RGS4 contacts of G␣ i1 (46) are yellow. All other switch II residues are in light blue. Switches I and III are shown in cyan and blue, respectively. The space-filling model was generated using RasMol (version 2.6) software and the coordinates of G␣ t ⅐AlF 4 Ϫ (15).
earlier that P␥ and RGS16 interact with G␣ t ⅐AlF 4 Ϫ noncompetitively and that RGS16 acts on the transducin/PDE signaling system to accelerate G protein and effector inactivation rather than to block effector activation (11). The G␣ t GAP activity of RGS9, a major photoreceptor RGS protein, has been enhanced by P␥, supporting simultaneous binding of RGS and P␥ to G␣ t (7). To gain more information on spatial relationships between RGS and P␥ binding surfaces on G␣ t , all G␣ t/i switch II mutants were examined for their ability to interact with RGS16. Most of the G␣ t mutations that impaired interaction with RGS16 coincided with replacements of the RGS contact residues (46). Moreover, none of the mutants with substituted RGS contact residues had notably altered binding to P␥. A spacefilling model shows that despite a large number of neighbors between RGS and P␥ interacting residues in the primary structure of G␣ t there is no significant overlap between the two surfaces (Fig. 7B). This finding is in noteworthy agreement with the similar conclusion made based on the G␣ s /effector crystal structure (28).
A more detailed mapping on effector surfaces of G␣ t than biochemical tools can provide is to be attained from a crystal structure of transducin complexed with P␥ or holoPDE. The potential for simultaneous binding of effector and RGS proteins to G␣ subunits brings up an intriguing possibility of a new interface between RGS and G␣ effector. Trimeric complexes between G␣ subunits, their effectors, and RGS proteins may play an important role in rapid termination of G protein-mediated signals.