Rhodopsin Recognition by Mutant Gsα Containing C-terminal Residues of Transducin*

The C-terminal regions of the heterotrimeric G protein α-subunits play key roles in selective activation of G proteins by their cognate receptors. In this study, mutant Gsα proteins with substitutions by C-terminal residues of transducin (Gtα) were analyzed for their interaction with light-activated rhodopsin (R*) to delineate the critical determinants of the Gtα/R* coupling. In contrast to Gsα, a chimeric Gsα/Gtα protein containing only 11 C-terminal residues from transducin was capable of binding to and being potently activated by R*. Our results suggest that Cys347 and Gly348 are absolutely essential, whereas Asp346 is more modestly involved in the Gt activation by R*. In addition, the analysis of the intrinsic nucleotide exchange in mutant Gsα indicated an interaction between the C terminus and the switch II region in Gtα·GDP. Mutant Gsα containing the Gtα C terminus and substitutions of Asn239and Asp240 (switch II) by the corresponding Gtα residues, Glu212 and Gly213, displayed significant reductions in spontaneous guanosine 5′-O-(3-thiotriphosphate)-binding rates to the levels approaching those in Gtα. Communication between the C terminus and switch II of Gtα does not appear essential for the activational coupling between Gt and R*, but may represent one of the mechanisms by which Gα subunits control intrinsic nucleotide exchange.

In the visual transduction cascade, photolyzed rhodopsin (R*) 1 activates the photoreceptor G protein, transducin (G t ␣␤␥), by catalyzing GDP/GTP exchange on the G t ␣ subunit. Transducin sites of interaction with R* have been extensively investigated (1,2). Despite the fact that the G t ␤␥ subunit participates in the interaction with R* and enhances transducin affinity for R* (3,4), the G t ␣ subunit contains principal sites for specific recognition of the activated receptor. Abundant evidence points to the C-terminal domain of G t ␣ as the major R* contact site. An ADP-ribosylation of G t ␣ Cys 347 by pertussis toxin uncouples G t from R* (5,6). A synthetic peptide corresponding to the 11 C-terminal amino acid residues of G t ␣, G t ␣-(340 -350), is capable of competing with G t for binding to R* and stabilizing the metarhodopsin II conformation (7). Mu-tational analysis of the G t ␣ C-terminal tail (8,9) and screening of the combinatorial peptide library (10) have identified residues within G t ␣-(340 -350) that are required for interaction with R*. All studies have consistently implicated conserved Leu 344 and Leu 349 as being absolutely essential for G t ␣ binding to R* (8 -10). However, the receptor role of other C-terminal residues of G t ␣ has not been conclusively established. An NMR analysis of structural changes induced in the synthetic peptide G t ␣-(340 -350) upon its binding to R* suggested a unique role for Gly 348 in the transition from a disordered G t ␣-(340 -350) to the R*-binding conformation with an open reverse turn and a helix-terminating C-capping motif (11). In agreement with the role of Gly 348 , activation of the G t ␣ mutant Gly 348 3 Ala expressed in COS-7 cells was severely impaired (8). Furthermore, Cys 347 and Gly 348 were invariant in R*-binding peptides obtained with a combinatorial library (10). In contrast, Cys 347 3 Ala isolated from transfected COS-7 cells (8) or in vitro translated Gly 348 3 Ala (9) showed no significant defects in interaction with R*. Thus, results of the "loss-of-function" experiments (Ala mutagenesis of G t ␣) are not entirely consistent with a "gain-of-function" combinatorial selection of peptides.
To elucidate the role of individual C-terminal residues of G t ␣ in the specific interaction with R*, we initially generated a chimeric G s ␣/G t ␣ protein containing 11 C-terminal residues of G t ␣. The finding that, in contrast to G s ␣, this chimera was capable of effective and specific interaction with R* allowed us to utilize G s ␣ as a background for introduction of selected C-terminal residues of G t ␣. Analysis of the G s ␣ mutants identified minimal insertions of G t ␣ residues required for R* recognition and revealed a potential role of the C-terminal and switch II residues Glu 212 and Gly 213 in providing for a low intrinsic GDP/GTP exchange rate characteristic for G t ␣.

EXPERIMENTAL PROCEDURES
Materials-[ 35 S]GTP␥S (1160 Ci/mmol) was purchased from Amersham Pharmacia Biotech. Restriction enzymes were from New England Biolabs Inc. T4 DNA ligase was from Roche Molecular Biochemicals. Cloned Pfu DNA polymerase was from Stratagene. L-1-Tosylamido-2phenylethyl chloromethyl ketone-treated trypsin was from Worthington. All other chemicals were from Sigma or Fisher.
Site-directed Mutagenesis of G s ␣-The short splice form of bovine G s ␣ was subcloned into the vector pHis 6 -G s ␣ for bacterial expression using PCR-directed cloning and the NcoI and HindIII restriction sites (12). The long splice form G s ␣ numbering is used throughout the text for clarity. pHis 6 -G s ␣ was used as a template for PCR-based mutagenesis. G s ␣/G t ␣-(340 -350) was generated by replacing the C terminus of G s ␣ with the 11 C-terminal residues of G t ␣. The PCR included a forward primer carrying a BglII site overlapping codons for G s ␣ 294 QDL 296 and a reverse primer coding for the C terminus of G t ␣ and carrying a HindIII site overlapping the stop codon. The PCR product (ϳ300 base pairs) was digested with BglII and HindIII and subcloned into the pHis 6 vector. Similarly, all other mutations within the G s ␣ C terminus were introduced using reverse primers carrying the desired mutation(s). To substitute Asn 239 -Asp 240 in G s ␣ for the corresponding G t ␣ residues, Glu 212 -Gly 213 , a forward 5Ј-phosphorylated primer carrying these substitutions was paired with the appropriate reverse C-terminal primers and PCR-amplified using pHis 6 -G s ␣ or mutant pHis 6 -G s ␣ as a template.
The products were ligated to the upstream G s ␣ cDNA fragment, which was obtained by PCR amplification with the N-terminal primer and the reverse 5Ј-phosphorylated primer coding for the sequence adjacent to G s ␣ Asn 239 -Asp 240 . After the ligation, DNA fragments of the appropriate size (ϳ1200 base pairs) were separated on an agarose gel, excised, and then used as templates for PCR amplification with primers flanking the G s ␣ open reading frame. Amplified mutant cDNAs were subcloned into the pHis 6 vector as described above. Sequences of all G s ␣ mutants were confirmed by automated DNA sequencing at the University of Iowa DNA Core Facility. Typically, for protein expression, 500 ml of 2ϫ TY medium (16 gl liter tryptone, 10 gl liter yeast extract, 5 gl liter NaCl) was inoculated with 15 ml of overnight culture of BL21(DE3) cells transformed with mutant G s ␣ cDNAs. At A 600 ϭ 0.5, expression was induced by 30 M isopropyl-␤-D-thiogalactopyranoside for 18 h at 25-30°C. Cells were pelleted and kept frozen at Ϫ70°C. Purification of G s ␣ and its mutants was carried out as described (12,13). Yields for 80 -90% pure mutant G s ␣ were normally 3-5 mg.
Trypsin Protection Assay-G s ␣ mutants (20 M) were incubated for 30 min at 25°C in 20 mM HEPES (pH 8.0) containing 130 mM NaCl, 50 M GDP, and 5 mM MgSO 4 . Where indicated, 50 M GTP␥S or 10 mM NaF and 30 M AlCl 3 were included in the buffer. Trypsin digestions were performed with 50 g/ml trypsin for 20 min at 25°C and stopped by simultaneous SDS sample buffer addition and heat treatment (100°C, 5 min). The time courses of GTP␥S binding in the trypsin protection assay were followed by trypsinolysis of mutant G s ␣ with 100 g/ml trypsin for 10 min at 25°C. Binding and uROS membranes (10 M rhodopsin) were mixed in 100 l of 20 mM HEPES (pH 7.6) containing 150 mM NaCl and 4 mM MgCl 2 . The mixture was then illuminated with a white light transilluminator lamp (Fisher) for 5 min at 25°C. Rod outer segment membranes were centrifuged for 10 min at 20,000 ϫ g, and the pellets were washed twice with 1 ml of the buffer. Bound G proteins were then extracted using 50 l of 10 M GTP␥S in hypotonic buffer, and the supernatants (150,000 ϫ g, 15 min) were concentrated and analyzed by SDS-polyacrylamide gel electrophoresis.
GTP␥S Binding Assay-G s ␣ or its mutants (1 M) alone or mixed with 2 M G t ␤␥ or with 2 M G t ␤␥ and uROS membranes (300 nM rhodopsin if not otherwise indicated) were incubated for 3 min at 25°C. Binding reactions were started by addition of 5 M [ 35 S]GTP␥S (1 Ci/mmol). Aliquots of 20 l were withdrawn at the indicated times and passed through Whatman cellulose nitrate filters (0.45 m), which were then washed three times with 1 ml of ice-cold 20 mM Tris-HCl (pH 8.0) containing 130 mM NaCl and 10 mM MgSO 4 . Upon complete dissolution in a 3a70B mixture, the filters were counted in a liquid scintillation counter. The k app values for the binding reactions were calculated by fitting the data to the following equation: % GTP␥S bound ϭ 100%(1 Ϫ e Ϫkt ). A linear regression fit was used when Ͻ25% of total mutant G s ␣ bound GTP␥S during the time course of the binding reactions.

Expression and Characterization of the G s ␣/G t ␣ Chimera
Containing the G t ␣ C Terminus-To assess the suitability of G s ␣/G t ␣ chimeras and G s ␣ mutants as tools for mapping the G t ␣/R* interface, we first expressed G s ␣ and the G s ␣/G t ␣-(340 -350) chimera as His-tagged proteins in Escherichia coli and examined their interactions with R*. G s ␣ and G s ␣/G t ␣-(340 -350) were fully capable of undergoing an activating conformational change upon the binding of GTP␥S or GDP/AlF 4 Ϫ as tested in the trypsin protection assay (Fig. 1A). Direct binding experiments of G s ␣ and G s ␣/G t ␣-(340 -350) to R* were carried out in suspensions of uROS reconstituted with G t ␤␥ (15). For the purpose of comparison with native G t ␣, recombinant Histagged G i ␣ and the transducin-like chimeric G t ␣/G i ␣ protein G t ␣* (14) were included (Fig. 1B). Predictably, native myristoylated G t ␣ showed the strongest binding to R*, followed by recombinant non-myristoylated G t ␣* and G i ␣. Relatively potent binding of G i ␣ to R* is not surprising. G i ␣ has a C-terminal sequence almost identical to that of G t ␣, and activation of G i ␣ by R* has been reported (13). Under the conditions of this assay, G s ␣ showed no detectable binding to R*. However, the incorporation of the G t ␣ C terminus into G s ␣ resulted in nota-ble binding to R* (Fig. 1B). G s ␣/G t ␣-(340 -350) bound to R* significantly more weakly compared with G t ␣, in part likely due to the lack of the N-terminal myristoylation of the chimeric protein. Somewhat weaker binding of G s ␣/G t ␣-(340 -350) to R* in comparison with G t ␣* or G i ␣ probably reflects additional R*-interacting sites in these proteins, although a lower affinity of G s ␣/G t ␣-(340 -350) for G t ␤␥ may have contributed as well.
Further functional evaluation of G s ␣ and G s ␣/G t ␣-(340 -350) interactions with R* was performed using a GTP␥S binding assay. In control experiments, the very slow GTP␥S binding by G t ␣* due to intrinsic nucleotide exchange (k app ϭ 0.002 min Ϫ1 ) was greatly accelerated in the presence of R* and G t ␤␥ (k app ϭ 0.35 min Ϫ1 ) ( Fig. 2A). As expected, binding of GTP␥S to G s ␣ due to spontaneous nucleotide exchange (k app ϭ 0.030 min Ϫ1 ) was significantly faster than that to G t ␣* (Fig. 2B). Addition of G t ␤␥ inhibited the GTP␥S binding to G s ␣ (k app ϭ 0.003 min Ϫ1 ) (Fig. 2B). This effect is consistent with the inhibition of the G s ␣⅐GDP/GTP␥S exchange by G␤␥ demonstrated earlier (16). The presence of uROS membranes only slightly enhanced GTP␥S binding to G t ␤␥-complexed G s ␣ (k app ϭ 0.006 min Ϫ1 ) (Fig. 2B). It is not clear whether this small increase is due to very weak affinity of G s ␣ for R* or perhaps just to reduction of effective G t ␤␥ concentration caused by its partitioning to rod outer segment membranes. We favor the latter explanation because no interaction between G s ␣ and R* was observed in the direct binding experiments shown in Fig. 1B. Nonetheless, the rate of GTP␥S binding to G s ␣ in the presence of both uROS and G t ␤␥ was still considerably slower than the intrinsic rate. Similarly to G s ␣, G s ␣/G t ␣-(340 -350) had a high intrinsic GTP␥Sbinding rate (0.031 min Ϫ1 ), which was inhibited by G t ␤␥ (0.002 min Ϫ1 ) (Fig. 2C); yet addition of uROS to G s ␣/G t ␣-(340 -350) in the presence of G t ␤␥ had a profound effect. The GTP␥S-binding rate was accelerated to levels (k app ϭ 0.27 min Ϫ1 ) far exceeding the G s ␣/G t ␣-(340 -350) intrinsic GTP␥S-binding rate (Fig. 2C). Overall, analysis of the interaction of G s ␣ and G s ␣/G t ␣-(340 -350) with R* supports previous conclusions that the C terminus of G t ␣ is the major determinant of the G t ␣ interaction with R*. It also indicates that G s ␣ provides an appropriate background for introduction of specific G t ␣ C-terminal residues to test their role for G t ␣ activation by R*.
Characterization of G s ␣ Mutants Containing C-terminal Residues of G t ␣-The 11-residue C-terminal segments of G s ␣ and G t ␣ contain two conserved Leu residues at positions Ϫ7 and Ϫ2. The G s ␣/G t ␣ residues at several C-terminal positions, Ϫ10 (Arg/Lys), Ϫ6 (Arg/Lys), and Ϫ1 (Leu/Phe), are highly homologous (Table I), and these differences do not appear critical for receptor recognition (8 -10). Therefore, we focused on two clusters of non-homologous G s ␣/G t ␣ substitutions at positions Ϫ9 and Ϫ8 (Met-His/Glu-Asn) and positions Ϫ5, Ϫ4, and Ϫ3 (Gln- and G s ␣/G t ␣-(340 -350) (lanes 1-5, respectively) reconstituted with G t ␤␥ were bound to R* in uROS membranes, followed by extraction of bound G proteins with GTP␥S and analysis by SDS-polyacrylamide gel electrophoresis as described under "Experimental Procedures." Tyr-Glu/Asp-Cys-Gly). To limit the number of potential combinations of substituted residues, we decided to introduce mutations into mutant G s ␣ with Glu Ϫ3 replaced by Gly. The Gly residue is highly conserved in the G i /G t /G o family and is crucial to the ability of the G t ␣ C terminus to adopt the R*-binding conformation (11). Furthermore, Gly 348 is invariant in peptides selected by binding to R* (10), and its replacement with Ala leads to a severe impairment of G t ␣ activation by R* (8).
Surprisingly, a point mutation of G s ␣, Glu Ϫ3 3 Gly, led to a strong reduction in the expression levels and functional activity as judged by the lack of trypsin protection in the presence of AlF 4 Ϫ and severely impaired GTP␥S binding (data not shown). Thus, the conformational changes of the G s ␣ C terminus due to insertion of Gly Ϫ3 appear to influence the overall structural integrity of G s ␣. Interestingly, additional substitutions of G s ␣ residues by corresponding G t ␣ residues in the two triple G s ␣ mutants, Met Ϫ9 3 Glu/His Ϫ8 3 Asn/Glu Ϫ3 3 Gly (EN-G mutant) ( Table I) and Gln Ϫ5 3 Asp/Tyr Ϫ4 3 Cys/Glu Ϫ3 3 Gly (DCG mutant), rescued expression levels and the ability of mutant G s ␣ to undergo an activational conformational change. Both mutants EN-G and DCG had expression levels comparable to that of G s ␣ and were analogous to G s ␣ in the trypsin protection assay (data not shown). EN-G and DCG had similar intrinsic GTP␥S-binding rates, 0.033 and 0.029 min Ϫ1 , respectively (Fig. 3). Binding of G t ␤␥ strongly inhibited these rates (k app ϭ 0.003 min Ϫ1 ), as seen for G s ␣. The key difference between EN-G and DCG was revealed in the presence of R*. Whereas R* in the presence of G t ␤␥ did not meaningfully affect the GTP␥S binding by EN-G (k app ϭ 0.005 min Ϫ1 ), it significantly stimulated the rate of GTP␥S binding by DCG (k app ϭ 0.073 min Ϫ1 ) (Fig. 3). To more accurately assess the relative abilities of G t ␣*, G s ␣/G t ␣-(340 -350), and DCG to interact with R*, the kinetics of R*-induced GTP␥S binding were analyzed at varying concentrations of uROS membranes. Fig. 4 shows the apparent rates of GTP␥S binding to G t ␣*, G s ␣/G t ␣-(340 -350), and DCG plotted as functions of the photoreceptor membrane (rhodopsin) concentration. The apparent activation constants for G t ␣*, G s ␣/G t ␣-(340 -350), and DCG from the slopes in Fig.  4 are 1.33 ϫ 10 6 , 0.66 ϫ 10 6 , and 0.23 ϫ 10 6 M Ϫ1 ⅐min Ϫ1 , respectively. This analysis underscores the role of the G t ␣ C terminus and the residues Asp 346 , Cys 347 , and Gly 348 , in particular, in the interaction with R*.
The above results suggest that one or more of these G t ␣ residues, Asp 346 , Cys 347 , and Gly 348 , are crucial for R* recognition. Based on these findings, three double mutants of G s ␣, Gln Ϫ5 3 Asp/Tyr Ϫ4 3 Cys (DC), Tyr Ϫ4 3 Cys/Glu Ϫ3 3 Gly (CG), and Gln Ϫ5 3 Asp/Glu Ϫ3 3 Gly (D-G), were generated and analyzed (Table I). The DC mutant had reduced levels of expression and trypsin protection, and the GTP␥S binding by DC was not notably affected by R* (data not shown). The CG and D-G mutants were expressed and trypsin-protected in the presence of AlF 4 Ϫ similarly to G s ␣ (Fig. 5A). Unexpectedly, the CG mutant demonstrated a low degree of trypsin protection after incubation with GTP␥S, but not with AlF 4 Ϫ (Fig. 5A). Examination of the rates of intrinsic nucleotide exchange confirmed a relatively slow basal GTP␥S-binding rate by CG (k app ϭ 0.007 min Ϫ1 ) as compared with G s ␣ or other G s ␣ mutants (Fig. 5B). In contrast to the D-G mutant, which showed no significant enhancement of the GTP␥S-binding rate by R* in relation to G s ␣ (Fig. 5C), the GTP␥S-binding rate of CG was accelerated by the activated receptor (k app ϭ 0.047 min Ϫ1 ; best fit with B max ϭ 45%) (Fig. 5B). Although the R*-induced activation of CG was somewhat lower than that of DCG, the data seem to indicate the central role of G t ␣ Cys 347 and Gly 348 .
Intrinsic and R*-induced GTP␥S Binding of G s ␣ Mutants containing G t ␣ Glu 212 and Gly 213 -The lower rate of intrinsic GTP␥S binding for the CG mutant prompted us to further investigate this phenomenon. In one of the three molecules in the asymmetric unit of the G t ␣⅐GTP␥S crystal, the C-terminal residues 343-349 are seen to make contacts with residues 212-215 from the ␣ 2 /␤ 4 loop (switch II) of transducin (17). This interaction may contribute to the very slow spontaneous nucleotide exchange of G t ␣ and possibly provide a plausible explanation for the property of the CG mutant. The residues Glu 212 and Gly 213 of G t ␣ are substituted in G s ␣ by the non-homologous residues Asn 239 and Asp 240 . The latter residues may interfere significantly with the coupling between switch II and the C terminus in G s ␣/G t ␣-(340 -350) or the DCG mutant, but perhaps to a lesser extent in the CG mutant. To test this hypothesis, we introduced the Asn 239 3 Glu and Asp 240 3 Gly substitutions into G s ␣ (EG-G s ␣), G s ␣/G t ␣-(340 -350) (EG-G s ␣/G t ␣-(340 -350)), and the DCG mutant (EG-DCG). These mutants were fully functional as assessed by the levels of expression and trypsin protection in the presence of AlF 4 Ϫ (Fig. 6, A and B). EG-G s ␣ was functionally indistinguishable from G s ␣ (data not shown). The nucleotide-binding rates for EG-G s ␣/G t ␣-(340 -350) and EG-DCG were considerably slower than that for G s ␣, as shown by the trypsin protection test performed after time courses of GTP␥S binding (Fig. 6, A and B). Corroborating these observations, the rates of intrinsic GTP␥S binding by EG-G s ␣/G t ␣-(340 -350) and EG-DCG were 0.004 and 0.009 min Ϫ1 , respectively (Fig. 6, C and D). In fact, these rates were sufficiently slow to make the inhibition of nucleotide exchange by G t ␤␥ appear insignificant. R*-induced stimulations of GTP␥S binding to EG-G s ␣/G t ␣-(340 -350) (k app ϭ 0.20 min Ϫ1 ) and EG-DCG (k app ϭ 0.064 min Ϫ1 ) were comparable to those for G s ␣/G t ␣-(340 -350) and DCG, respectively (Fig. 6, C and D). DISCUSSION Seven-transmembrane domain receptors transduce extracellular signals across the plasma membrane to the intracellular effectors by stimulating GDP/GTP exchange on the ␣-subunits of heterotrimeric GTP-binding proteins. Following the activational interaction with receptors, G␣⅐GTP and G␤␥ are released to activate their targets. Although G protein-coupled receptors may activate more than one related G protein, typically G protein/receptor interactions are rather specific, and a particular receptor recognizes G proteins from one family (2). The G␤␥ subunits contribute to the selectivity of G protein/receptor interactions, which apparently resides to a large extent in G␣ subunits. Evidence suggests that 10 or fewer C-terminal residues of G␣ subunits compose a major receptor-binding and specificity domain (7)(8)(9)(10)(11). The C termini are generally conserved within each of the four families of G proteins (G s , G i , G q , and G 12,13 ), but are diverse between members of different families, except for the two absolutely invariant Leu residues at positions Ϫ2 and Ϫ7. The C-terminal diversity is essential for the G protein/receptor selectivity. An elegant study demonstrated that replacement of just three C-terminal residues of G q ␣ at positions Ϫ1, Ϫ3, and Ϫ4 by corresponding G i ␣ residues allowed G q to switch specificity to a new, G i ␣-coupled receptor (18). Similar findings were later reported using G␣ 13/i , G␣ q/s , and G␣ s/q C-terminal chimeras (19). Conservation of the C-terminal residues within one family requires participation of additional regions of G␣ to achieve fine receptor tuning. For example, Gln 304 and Glu 308 in the ␣ 4 helix of G i ␣ 1 are important to the selectivity of the 5-hydroxytryptamine receptor for G i ␣ 1 over G t ␣ (20). Chimeric G t ␣/G i ␣ proteins have been very instrumental in gaining insight into the G i ␣ 1 /5-hydroxytryptamine receptor selectivity (20), but would not be helpful in elucidating the C-terminal determinants of the G t ␣/R* interaction since the C termini of G t ␣ and G i ␣ are almost identical, and R* potently stimulates G i ␣ in vitro (13). In fact, the "receptor switch" or "gain-of-recognition" type of experiment for R* and chimeric G␣ has not until now been reported. Instead, in addition to Ala mutagenesis of the G t ␣ C terminus, which identifies impairments of function (8,9), a positive selection for R* binding was carried out using a com- binatorial peptide library (10). The mutagenesis and peptide selection studies provide a consensus on the key role of such residues as Leu 344 and Leu 349 , but disagree on the role of other residues, including Cys 347 and Gly 348 . This may be explained in part by different experimental approaches and by potentially distinct consequences of the same amino acid substitution in full-length G t ␣ and in a peptide.
Our approach to delineation of the C-terminal contacts of G t ␣ with R* involved using G s ␣ for introduction of nonconserved G t ␣ residues and examining activation of mutant G s ␣ by the receptor. The choice of G s ␣ was based on its apparent inability to bind and be activated by R*. In contrast, chimeric G s ␣/G t ␣ containing only 11 C-terminal residues from transducin was capable of efficient coupling to R*. The analysis of R*-induced rates of GTP␥S binding by mutant G s ␣ reconstituted with G t ␤␥ demonstrated the key role of transducin Cys 347 and Gly 348 in the G t ␣ interaction with the activated receptor. In addition, Asp 346 appears to be important, as the DCG mutant containing all three residues was stimulated by R* to a greater extent than the mutant with the Cys and Gly residues only. This result is in excellent agreement with the study on sequence selection based on R*/peptide interaction, which demonstrated an invariant presence of Cys and Gly residues at positions Ϫ4 and Ϫ3 in peptides with high affinity for R* (10). Interestingly, the Asp residue at position Ϫ5 was also highly conserved in R*binding peptides (10). Our study extends the previous findings by demonstrating the correlation of the ability of C-terminal peptides containing the key residues to bind R* (10) with the activation of G␣ subunits containing the same residues. The model of the mutant G s ␣ C-terminal sequence with Asp, Cys, and Gly at positions Ϫ5, Ϫ4, and Ϫ3, respectively, is shown in Fig. 7B. This model was generated using the R*-induced structure of G t ␣-(340 -350) as a template (Fig. 7A) (11). It indicates that the presence of the key G t ␣ residues would enable the C terminus of G s ␣, upon binding to R*, to adopt an overall conformation very similar to that of G t ␣-(340 -350).
Our data are also consistent with the study showing the critical role of the Cys Ϫ4 and Gly Ϫ3 residues of G i ␣ 2 in the selectivity of several G i -coupled receptors (19). Cys Ϫ4 and Gly Ϫ3 are conserved in the G i family (G i ␣ 1-3 , G t ␣, and G o ␣), but other G protein families have non-homologous substitutions at these positions. Therefore, residues at positions Ϫ3 and Ϫ4 are likely to serve as major determinants that discriminate G icoupled receptors from receptors signaling through other G proteins.
In the course of this study, we made an interesting observation that the CG mutant of G s ␣ had notably reduced intrinsic nucleotide exchange in comparison with G s ␣, DCG, and G s ␣/ G t ␣-(340 -350). We then hypothesized that this effect may have resulted from the interaction of the C-terminal residues in the CG mutant with the ␣ 2 /␤ 4 loop from the switch II region. Such an interaction is seen between residues 343-349 and 212-215 in one of the G t ␣⅐GTP␥S crystal structures (17) and is supported by analysis of the conformational changes using fluorescently labeled G t ␣ (22). The results of our study suggest that a similar interaction is present in the GDP-bound conformation of G t ␣. The residues Glu 212 and Gly 213 of G t ␣ are substituted by Asn 239 and Asp 240 in G s ␣, and we speculated that because of these differences, coupling between the ␣ 2 /␤ 4 loop and the C terminus may not be possible in DCG and G s ␣/G t ␣-(340 -350). This hypothesis has been confirmed by significant decreases in the rates of intrinsic GTP␥S binding in DCG and G s ␣/G t ␣-(340 -350) upon substitution of Asn 239 and Asp 240 by the Glu and Gly residues, respectively. We have modeled the substitutions Glu 212 3 Asn and Gly 213 3 Asp in G t ␣ using the FIG. 7. A, the "backbone" and "sticks" representations of the R*induced structure of G t ␣-(340 -350) (11); B, the model structure of the C-terminal sequence of DCG (mutant G s ␣ with Asp, Cys, and Gly at positions Ϫ5, Ϫ4, and Ϫ3, respectively). The model was generated using the Swiss-PDB Viewer (21) and coordinates of the R*-induced structure of G t ␣-(340 -350) as a template (11). The images in A and B were obtained using RasMol Version 2.6. FIG. 8. A, a Connolly diagram of the ␣ 2 /␤ 4 loop (in red) and the C terminus (in green) from the structure of G t ␣⅐GTP␥S (17). The linkage is formed primarily by van der Waals contacts. B, the substitutions Glu 212 3 Asn and Gly 213 3 Asp modeled into the G t ␣⅐GTP␥S structure. The mutation Gly 213 3 Asp leads to steric hindrances between the Asp side chain and Gly 348 and Leu 349 . The model and the images were generated using SYBYL Version 6.5.3 (Tripos Associates, St. Louis, MO). G t ␣⅐GTP␥S structure as a template (17). Fig. 8 demonstrates that introduction of Asp instead of G t ␣ Gly 213 creates multiple steric hindrances between the Asp side chain and Gly 348 and Leu 349 , thus providing a structural basis for the lack of interaction between the ␣ 2 /␤ 4 loop and the C terminus in DCG and G s ␣/G t ␣-(340 -350). Although this model does not reveal an explanation for the functional properties of the CG mutant, it is possible that the steric hindrances are reduced when Gln is present at position Ϫ5 instead of Asp.
Altogether, our results suggest that the interaction between the C terminus and the switch II region in G t ␣ contributes to the low intrinsic rate of nucleotide exchange of transducin. Furthermore, this interaction must be present not only in the active conformations of G t ␣ (17,22), but in the GDP-bound conformation as well because GDP release is the rate-limiting step. It is unlikely that an analogous interaction controls the GDP affinity in G o ␣ as it has an Asp residue, similarly to G s ␣, instead of Gly 213 in G t ␣. Contacts of the Ile-Ile residues at positions Ϫ12 and Ϫ11 from the C terminus with Val 34 and Leu 195 appear to be involved in regulating G o ␣ affinity for GDP (23). Lack of the C terminus/switch II contacts in G s ␣ and G o ␣ may in part account for the relatively high spontaneous GDP/ GTP exchange rate. The G␤␥ subunit binding is required to lower the basal exchange rate of these G␣ proteins. The C terminus/switch II coupling is likely to be just one of the mechanisms that control the rate of spontaneous nucleotide exchange. G i ␣ has a relatively high intrinsic GDP/GTP exchange rate despite the fact that the sequences corresponding to G t ␣ residues 343-349 and 212-215 are identical in G t ␣ and G i ␣. The lack of discrete interdomain contacts between the Ras-like and the ␣-helical domains in G i ␣⅐GDP may account for the rapid dissociation of GDP (24).
Both groups of mutant G␣ subunits with high (G s ␣/G t ␣-(340 -350) and DCG) and low (EG-G s ␣/G t ␣-(340 -350) and EG-DCG) intrinsic GTP␥S-binding rates were comparably activated by R* in the presence of G t ␤␥. This indicates either that binding of G t ␤␥ blocks the contacts between the G t ␣ C terminus and the switch II region or that these contacts do not hinder or enhance the coupling of the G t ␣⅐GDP C terminus to light-activated rhodopsin. Consequently, potential communication between the C terminus and the switch II region of G t ␣ in G t ␣␤␥ is not an essential component of R*-induced GDP release. The latter process likely involves transmission of the conformational change from the C terminus via the ␣ 5 helix to the ␤ 6 /␣ 5 loop that contains the guanine ring-binding residues (25).