Mutation of the Fourth Cytoplasmic Loop of Rhodopsin Affects Binding of Transducin and Peptides Derived from the Carboxyl-terminal Sequences of Transducin α and γ Subunits*

The role of the putative fourth cytoplasmic loop of rhodopsin in the binding and catalytic activation of the heterotrimeric G protein, transducin (Gt), is not well defined. We developed a novel assay to measure the ability of Gt, or Gt-derived peptides, to inhibit the photoregeneration of rhodopsin from its active metarhodopsin II state. We show that a peptide corresponding to residues 340–350 of the α subunit of Gt, or a cysteinyl-thioetherfarnesyl peptide corresponding to residues 50–71 of the γ subunit of Gt, are able to interact with metarhodopsin II and inhibit its photoconversion to rhodopsin. Alteration of the amino acid sequence of either peptide, or removal of the farnesyl group from the γ-derived peptide, prevents inhibition. Mutation of the amino-terminal region of the fourth cytoplasmic loop of rhodopsin affects interaction with Gt (Marin, E. P., Krishna, A. G., Zvyaga T. A., Isele, J., Siebert, F., and Sakmar, T. P. (2000) J. Biol. Chem. 275, 1930–1936). Here, we provide evidence that this segment of rhodopsin interacts with the carboxyl-terminal peptide of the α subunit of Gt. We propose that the amino-terminal region of the fourth cytoplasmic loop of rhodopsin is part of the binding site for the carboxyl terminus of the α subunit of Gt and plays a role in the regulation of βγ subunit binding.

G protein-coupled receptors transmit extracellular signals to the cell's interior via heterotrimeric G proteins and effector enzymes or ion channels (1,2). Rhodopsin is one of the archetypes of the G protein-coupled receptor superfamily. It triggers the biochemical amplification machinery of the visual cascade in the rod photoreceptor cell, which comprises the G protein transducin (G t ) 1 and the effector, a cyclic GMP-specific phosphodiesterase (3,4). The transduction of a light signal begins with the photochemical cis-trans isomerization of the chromophore, 11-cis-retinal. Protein conformational changes are transmitted from the ligand-binding site to the cytoplasmic surface of the receptor where catalytic activation of G t occurs. This intramolecular conversion from inactive (rhodopsin) to active (R*) states mediated by chromophore isomerization has been termed "signal transmission" (5). Key structural correlates of the transition to the active state include deprotonation of the retinylidene Schiff base (6) with concomitant protonation of the Glu 113 counterion (7,8) and the protonation of the cytoplasmic surface of rhodopsin (9, 10) mediated by the highly conserved Glu 134 residue at the cytoplasmic border of transmembrane (TM) helix 3. Movements of TM helices have been proposed to accompany the signal transmission process, with a change in the orientation of TM helices 3 and 6 relative to each other as the most prominent feature (11)(12)(13).
The cytoplasmic surface of rhodopsin comprises four loops and a carboxyl-terminal tail. The first (C1), second (C2), and third (C3) cytoplasmic loops connect adjacent TM helices. The fourth cytoplasmic loop (C4) is bounded by TM helix 7 at its amino terminus and two palmitoyl groups inserted into the membrane bilayer at its carboxyl terminus. The palmitoyl groups are attached to Cys 322 and Cys 323 via thioester linkages. A schematic of the structure of rhodopsin is presented in the preceding paper (14). A variety of experimental approaches, including proteolysis, chemical modification, peptide competition, and site-directed mutagenesis in combination with biochemical and biophysical assays, have been employed to map the sites of rhodopsin responsible for the binding and activation of G t . The salient results have indicated that loops C2 and C3 are involved in G t binding and activation (15,16). In addition, recent studies indicate a role for the loop C4 in G t activation (14,17,18). Despite these studies and the availability of a high-resolution crystal structure for the G t holoprotein (19), there is little information concerning: 1) the key functional intramolecular interactions on the cytoplasmic surface of rhodopsin that form and regulate the catalytic site for G t , 2) G t subunit specificity for binding to particular cytoplasmic regions of rhodopsin, 3) the molecular mechanism of rhodopsin-catalyzed nucleotide release by G t .
In the preceding paper (14), we identified a region at the amino terminus of loop C4 of rhodopsin that most likely interacts with the ␣ subunit of G t , G␣ t . Here, we studied the interaction of site-directed mutants of rhodopsin with G t and peptides derived from the carboxyl-terminal sequences of G␣ t and G␥ t (20 -22). We used a novel assay in which the all-transretinal in metarhodopsin II (MII) is photoconverted to the cis configuration using blue light. The flash-induced photoregeneration of rhodopsin from MII can be followed spectroscopically with millisecond time resolution (20). Since stabilization of R* by G t or G t -derived peptides inhibits the rate of photoregeneration, the assay can be used to monitor the interaction of R* with G t (21,22). We show that peptide ␣(340 -350), corresponding to the carboxyl-terminal undecapeptide of G␣ t , and peptide ␥(50 -71)-far, corresponding to the carboxyl-terminal cysteinylthioetherfarnesyl peptide of G␥ t , stabilized R*. Alteration of the amino acid sequence of either peptide, or removal of the farnesyl group of the G␥ t -derived peptide prevented stabilization of R*. G t failed to stabilize mutant rhodopsins with alterations of the amino terminus of loop C4 near the TM helix 7 border. The G␣ t -derived peptide also failed to stabilize these mutants, suggesting that loop C4 comprises part of a binding site for the carboxyl-terminal tail of G␣ t .

EXPERIMENTAL PROCEDURES
Preparation of Rhodopsin and G t -Purified bovine rhodopsin was prepared from hypotonically washed rod outer segment membranes essentially as described (23). The recombinant mutant pigments used in this study are shown in Fig. 1A. The construction, expression, and purification of these samples were reported (14). Following purification, the samples were concentrated ϳ10-fold using Centricon-30 filtration devices (Amicon). G t holoprotein was purified from rod outer segment preparations essentially as described previously (24).
Peptide Synthesis-The synthetic peptides used in this study are listed in Fig. 1B. Peptides were synthesized using the Fmoc (N-(9fluorenyl)methoxycarbonyl) strategy with HBTU activation (Fastmoc, 0.1 mmol small-scale cycles) on an ABI Model 433A peptide synthesizer. Farnesylation of G␥ t -derived peptides was carried out by dissolving pure peptide (60 mg) in 5 ml of a solution of 50% (v/v) 1-propanol containing 35 mmol of sodium carbonate. The resulting solution was saturated with nitrogen and 0.6 ml of a freshly prepared 10% (v/v) farnesyl bromide solution in 1-propanol was slowly added under vigorous stirring while pH was adjusted to Ͼ9. The solution was flushed with nitrogen again and incubated for 24 -48 h with shaking. The farnesyl peptides were purified by reverse phase high performance liquid chromatography.
Instrumentation-Time-resolved absorption traces were recorded on a custom-built single-wavelength absorption photometer. Light from a 150-W halogen light source passes through a Jobin-Yvon HR460 monochromator (focal length 460 mm, 1200 lines/mm, slit width set to 1 mm) tuned to 543 nm, and from there through the cuvette (4-mm optical path length) and a band-pass interference filter onto a large surface PIN photodetector. The output is nulled and amplified twice, filtered with a 500-s electronic low-pass filter and recorded using a modified Nicolet 2090-IIIA digital oscilloscope.
Photoregeneration Experiments-The rhodopsin photoregeneration assay was performed as reported (20) with adaptation for recombinant pigments as follows. All samples contained 2 M pigment in a volume of 0.26 ml of 200 mM Na 2 HPO 4 (pH 8.0), 10 mM NaCl, and approximately 0.03% (w/v) dodecyl maltoside (DM). Due to the concentration procedure required for recombinant samples, the final DM concentration was not precisely known, but was estimated not to exceed 0.035% (w/v). After equilibrating the sample cuvette to 13°C, the sample was illuminated for 30 s with a green HeNe laser (543.5 nm, 5 mW, Melles Griot) to cause quantitative formation of MII. Absorption at 543 nm was recorded continuously. After 50 ms, a flash of blue light (412 Ϯ 7 nm, about 20-s duration) was applied to the sample and formation of photo-regenerated pigment was measured at 543 nm for an additional 200 ms. Discharge of the flashlamp affected the sensitive electronics of the detector, causing a brief artifactual negative deflection. Four records induced by four separate flashes were collected from each sample with 30-s intervals between the recordings. Starting with the initial illumination, each experiment took approximately 140 s. The four records were averaged to produce experimental data traces as presented in Fig. 2. The experimental photoregeneration signal traces are depicted as absorbance changes at 543 nm versus time (i.e. a rising signal indicates a proportional increase of absorbance due to reprotonation of the Schiff base).
Numerical Fitting Procedures and Determination of Initial Slope Values-The photoregeneration signal comprises a fast phase, which is not resolved, and a slow phase, which is monitored for 200 ms (see Fig.  2). Data points obtained 4.5-7.0 ms after the blue flash were averaged and used as an estimate for the amplitude of the fast phase. The relative amplitude of the fast phase of the photoregeneration of the recombinant pigments was the same as that of rhodopsin. The initial slope of the slow phase of a photoregeneration trace was determined from the numerical fit of a simple exponential-rise function offset by the amplitude of the fast phase. Values for relative slope are presented (Table I) to demonstrate the effect of G t or G t -derived peptides on the initial slope of the photoregeneration signal. Relative slope is defined as the ratio of the initial slope of the slow phase of the photoregeneration trace in the presence of G t or G t -derived peptides versus the initial slope in their absence. A relative slope of 1.0 indicates no effect, whereas a slope of Ͻ1.0 indicates inhibition of photoregeneration. The relative slope for an experiment with rhodopsin and G t (3 M) was typically about 0.7.

RESULTS
The Effect of G t and G t -derived Peptides on the Photoregeneration of Rhodopsin-A photoregeneration assay was employed that measures the kinetics of photoconversion of MII to rhodopsin (20). The assay takes advantage of the conformational coupling of the cytoplasmic surface of the active state of rhodopsin, R*, to the chromophore-binding pocket in the membrane-embedded domain of the receptor. The slow phase of the photoconversion kinetics essentially monitors the reprotonation of the retinylidene Schiff base as a function of time after a sample of R* is subjected to a blue flash. Photoconversion of R* ( max ϭ 380 nm) to rhodopsin ( max ϭ 500 nm) is inhibited if R* is bound to G t or certain G t -derived peptides. R* that is not bound in a stabilizing complex with G t or G t -derived peptides is FIG. 1. Amino acid sequences of recombinant rhodopsins and G t -derived peptides. A, amino acid sequence of the loop C4 of wildtype rhodopsin (WT Rho) and mutant opsins (CTr1, CTr2, CTr4, and CysXV). In bovine rhodopsin, this region extends from position 310 at the intracellular junction of the TM helix 7 to Cys 322 and Cys 323 , which are palmitoylated and inserted into the membrane (40 -42). Changes from the wild-type sequence are highlighted in gray. In CTr1, CTr2, and CTr4, portions of the C4 loop have been replaced with analogous sequences from the ␤ 2 -AR. In CysXV, the palmitoylation sites are replaced by serine residues. B, amino acid sequences of peptides were derived from the carboxyl termini of bovine G␣ t and G␥ t subunits. The ␣(340 -350) peptide corresponds to the native sequence of G␣ t , and the ␥(50 -71)-far corresponds to the native sequence of G␥ t , which is post-translationally modified by cysteinyl thioether farnesylation. Peptides with alterations of primary structure, or a lack of carboxyl-terminal farnesylation, were used as controls. The carboxyl termini of the G␥ t -derived peptides were amidated. The carboxyl termini of the G␣ t -derived peptides, and the amino termini of all the peptides, were unmodified. more readily photoconverted.
The effect of G t on the photoconversion of R* was studied first ( Fig. 2A). The change in absorbance at 543 nm is plotted as a function of time. The blue flash is applied to the sample at 50 ms. Superimposed upon an initial rapid change in amplitude, the slow phase of the trace represents the back conversion of R* to rhodopsin. The experiment was repeated with identical rhodopsin samples in the presence of increasing concentrations of G t (0, 0.5, 1.0, 2.0, 3.0, and 5.0 M). The traces are superimposed to show a clear dose-dependent inhibition of the photoregeneration reaction by G t . The experimental traces were fit to an exponential function in order to calculate values for initial slopes. The calculated fits are shown as dashed lines in Fig. 2. The initial slopes of the photoregeneration traces are plotted as a function of G t concentration in the inset. A satisfactory hyperbolic fit yielded an effective concentration at 50% inhibition (IC 50 ) value of 2.56 Ϯ 2.0 M. This value effectively represents a binding constant for the interaction between R* and G t under the conditions of the assay.
The effects of two peptides corresponding to the carboxylterminal regions of G␣ t and G␥ t on the photoconversion of R* were studied next (Fig. 2, B  Specificity of the Effects of ␣(340 -350) and ␥(50 -71)-far Peptides-The specificity of the effect of the G t -derived peptides was studied by performing control experiments with altered peptides. The peptide sequences are shown in Fig. 1B. One control peptide, ␣(340 -350) K341R/L349A, was studied to evaluate the specificity of the carboxyl-terminal sequence of G␣ t in R* interaction. This peptide failed to show peptide-R* interaction (25,26) and the substitution of Leu 349 by alanine in G␣ t was reported to abolish coupling to active rhodopsin (27,28). As shown in Fig. 3, the ␣(340 -350) K341R/L349A peptide failed to inhibit the photoregeneration of R*. Similarly, the requirements for length, primary structure, and farnesylation of the G␥ t -derived peptide were tested. Peptides ␥(60 -71)-far and ␥(50 -71)-far both inhibited photoregeneration similarly (Fig.  3). Positions 64 and 67 in G␥ t have been reported to be critical for interaction with MII, as observed with the altered peptide ␥(60 -71)-far F64T/L67S (29). The ␥(60 -71)-far F64A/L67A peptide did not inhibit photoregeneration (Fig. 3). In addition, the ␥(60 -71) peptide, which lacked cysteinyl farnesylation, did not inhibit photoregeneration (Fig. 3). This finding is consistent with earlier results showing that lack of farnesylation prevented MII stabilization by G␥ t -derived peptides (30,31). In  other control experiments, the 1D4 peptide, corresponding to the carboxyl-terminal 18 amino acids of rhodopsin, did not affect photoregeneration, nor did it affect the inhibition of photoregeneration by G t and the G t -derived peptides (data not shown).
It has been reported that detergent concentration has an influence on the activation rate of G t by R* (16,32). The effect of varying concentrations of DM on the photoregeneration kinetics of rhodopsin and on the inhibition of photoregeneration by G t (3 M), ␣(340 -350) (200 M), or ␥(50 -71)-far (500 M) was studied. Varying DM concentrations from 0.01 to 0.10% (w/v) had no effect on the photoregeneration kinetics of rhodopsin, and no effect on the inhibition of photoregeneration by ␣(340 -350) (data not shown). However, the inhibition of photoregeneration by G t and ␥(50 -71)-far was reduced by increasing DM concentrations from 0.01% to 0.10% (data not shown). This effect mirrors the reduction in the rate of G t activation by R* in the presence of increasing [DM], as previously reported (16,32). The final DM concentration under the standard conditions of the photoregeneration assay using heterologously expressed and purified mutant pigments is estimated to be 0.01-0.035%, a range in which detergent effects were found to be modest. Furthermore, the final DM concentration in assays of each of the recombinant samples is virtually identical.
Photoregeneration Assay of Recombinant Rhodopsin and Loop C4 Mutants-Photoregeneration assays were carried out on wild-type recombinant rhodopsin and four mutant rhodopsins (Fig. 1A). Representative photoregeneration traces are presented in Fig. 4 Fig. 2. These results confirm that the photoregeneration assay can be employed to study recombinant pigments prepared in relatively small quantities in a heterologous expression system.
Four mutant pigments with alterations of the amino acid sequence of the C4 loop were prepared (Fig. 1A). Mutants CTr1, CTr2, and CTr4 are essentially chimeric receptors in which parts of the C4 loop of rhodopsin are replaced by sequences from the ␤ 2 -adrenergic receptor (␤ 2 -AR). Mutant CysXV (C322S/C323S) was designed to evaluate the effect of receptor palmitoylation on the photoregeneration kinetics. The mutant opsin genes were expressed in COS cells, treated with 11-cisretinal and purified in DM detergent solution. The levels of palmitoylation of the expressed mutant pigments CTr2 and CTr4 were similar to that of the wild-type receptor expressed in parallel (14); CysXV was not palmitoylated (14,33). The ability of each of the mutant pigments to activate G t was evaluated using a fluorescence G t activation assay (Table I) , and CysXV show distinct effects of G t or G t -derived peptides. Photoregeneration of CTr4 is not influenced by G t or G tderived peptides, and CTr2 shows an effect with ␥(50 -71)-far, a minor effect with G t holoprotein and no effect with ␣(340 -350).  (14). b The number given for n refers to the number of independent samples studied. Each kinetic trace (Fig. 4) resulted from four separate photoregeneration experiments per sample.
c The relative slope is defined as the initial slope of the slow phase of the photoregeneration trace in the presence of G t , ␣(340 -350), or ␥(50 -71)-far divided by the initial slope of the trace in the absence of G t or G t -derived peptide. The initial slope of the slow phase of the photoregeneration trace was determined from the numerical fit of a simple exponential-rise function offset by the amplitude of the fast phase. The data are presented as mean Ϯ S.E. d The normalized inhibition is defined as: (1 Ϫ mean relative slope) mutant /(1 Ϫ mean relative slope) rhodopsin . The propagated S.E. values are also presented. ability to activate G t (14).
In the absence of G t or G t -derived peptides, the photoregeneration kinetics of rhodopsin and the four C4 mutants were essentially identical (Fig. 4, black traces). This result suggests that the C4 loop mutations do not affect the photoregeneration reaction. The effects of G t or G t -derived peptides on the photoregeneration of the mutant pigments are shown in Fig. 4 (red  traces). The effects can be conveniently evaluated by comparing relative slopes (Fig. 5A). The relative slope is defined as the initial slope of the slow phase of the photoregeneration trace in the presence of G t , ␣(340 -350), or ␥(50 -71)-far divided by the initial slope of the trace in the absence of G t or G t -derived peptide. The relative slope provides a quantitative measure of the effect of G t or a G t -derived peptide on the slow phase of the photoregeneration kinetics. A relative slope of 1.0 indicates no inhibition of photoregeneration, and relative slopes of Ͻ1.0 indicate a progressive inhibition of photoregeneration. Average values for relative slopes are presented in Table I.
Photoregeneration of mutant CTr1 was inhibited by G t and the G t -derived peptides. The degrees of inhibition were identical to those seen with wild-type rhodopsin. The behavior of mutants CTr2 and CTr4 was different from that of rhodopsin. G t and the peptide ␣(340 -350) did not inhibit photoregeneration of CTr2. This result is best appreciated in Fig. 5A, where the relative slopes for CTr2 in the presence of G t and ␣(340 -350) are ϳ1.0. However, the ␥(50 -71)-far peptide was able to inhibit photoregeneration of CTr2 to the same extent observed with rhodopsin. The photoregeneration of mutant CTr4 was not affected by G t , ␣(340 -350), or ␥(50 -71)-far. This result is best seen in Fig. 5A, where the relative slopes for CTr4 in the presence of G t and the G t -derived peptides are ϳ1.0. Photoregeneration of mutant CysXV was inhibited by G t and the G tderived peptides. However, the inhibition by peptide ␥(50 -71)far was relatively more pronounced for CysXV than for rhodopsin.
Each of the peptides and G t inhibit the photoregeneration of rhodopsin to different degrees. Therefore, the ability of each peptide and G t to affect a particular mutant cannot be compared directly using relative slopes. However, by normalizing the relative slope of a mutant to that of rhodopsin, such a comparison can be made. This expression, termed the normalized inhibition, is obtained from the following equation: (1 Ϫ mean relative slope) mutant /(1 Ϫ mean relative slope) rhodopsin . The normalized inhibition for rhodopsin is defined to be 1.0. A value of zero indicates no inhibition of the photoregeneration of the mutant pigment by a particular ligand. Data are plotted in Fig. 5B and listed in Table I. Mutant CTr1 is similar to rhodopsin with respect to inhibition of photoregeneration by G t , ␣(340 -350), and ␥(50 -71)-far. Mutant CTr2 shows essentially normal interaction with ␥(50 -71)-far, but fails to be affected by G t and ␣(340 -350). Mutant CTr4 is unaffected by G t and both G t -derived peptides. The photoregeneration of mutant CysXV is inhibited by ␣(340 -350) and G t normally, but displays an enhanced sensitivity to ␥(50 -71)-far. DISCUSSION Several lines of evidence suggest that the conformation of the cytoplasmic surface of the active state of rhodopsin, R*, is coupled to the conformation of the chromophore-binding pocket in the membrane-embedded domain of the receptor (3,4). In analogy to G protein-coupled receptors with diffusible ligands in which G protein binding stabilizes a receptor conformation with a high affinity ligand-binding site, MII is stabilized at the expense of its tautomeric forms by the binding of G t or G tderived peptides. This stabilization of MII is the basis of the "extra-MII" assay (34,35). This assay, however, can only be applied under conditions of a dynamic equilibrium between metarhodopsin I and MII, which is exquisitely sensitive to membrane environment, pH, temperature, ionic strength, etc. Therefore, an assay was developed that measures the kinetics of photoconversion of MII to rhodopsin in detergent solution (20 -22). The assay uses the fact that photoconversion of MII ( max ϭ 380 nm) to rhodopsin ( max ϭ 500 nm) following a blue actinic flash is inhibited if the MII molecule is bound to G t (20), or certain G t -derived peptides (21,22) as a result of the coupling between the conformation of the cytoplasmic surface and that of the chromophore-binding pocket.
The initial step in photoregeneration, the photoisomerization of the retinal to its cis conformation, may be compared with loading a spring that subsequently drives the protein, including its cytoplasmic domain, back to the ground state conformation (20). The product formed in this initial step, termed "reverted meta (RM)," is characterized by a MII-like protein conformation and a cis-retinal with a deprotonated Schiff base; it is spectrally indistinguishable from MII. RM rapidly converts FIG . 5. Quantitation of the effects of G t , ␣(340 -350), and ␥(50 -71)-far on the photoregeneration of recombinant rhodopsin and rhodopsin mutants. The slow phases of the photoregeneration traces were fit with exponential-rise equations, and the initial slopes were determined. A, relative slopes. Bars represent the average ratio of the initial slope of the slow phase of photoregeneration of pigment in the presence of G t (Transducin), ␣(340 -350), or ␥(50 -71)-far to the initial slope with pigment alone. The error bars display the standard errors. The photoregeneration of mutants CTr2 and CTr4 is unaffected by G t and ␣(340 -350). Only mutant CTr4 photoregeneration is insensitive to ␥(50 -71)-far. B, normalized inhibition. The vertical bars indicate the ability of G t and G t -derived peptides to inhibit photoregeneration of mutant pigments relative to their effect on wild-type rhodopsin (WT). The normalized inhibition values were determined from the following equation: (1 Ϫ mean relative slope) mutant /(1 Ϫ mean relative slope) rhodopsin. The error bars depict the propagated standard errors from the determination of relative slope. Numerical values are given in Table I. This analysis allows for direct comparison of the effects of each peptide and G t on each mutant. The effect of G t and G t -derived peptides on CTr1 and CysXV are similar to their effect on rhodopsin. The photoregeneration of CTr4 does not show an effect of G t or either G t -derived peptide. This behavior is consistent with a failure of CTr4 to bind G t , ␣(340 -350), and ␥(50 -71)-far. Replacement of the entire C4 loop in CTr2 shows a graded effect: CTr2 is essentially insensitive to inhibition by ␣(340 -350) and G t , but shows effects with ␥(50 -71)-far nearly identical to those of wild-type rhodopsin.
to a rhodopsin-like species characterized by a rhodopsin-like protein conformation and a cis-retinal with a protonated Schiff base. The presence of G t does not affect RM formation, indicating that the isomerization of the retinal itself is unaffected. However, bound G t prevents RM from converting to rhodopsin, by stabilizing the MII-like conformation of RM. Dissociation of G t from RM by GTP␥S treatment allows RM to revert to rhodopsin quantitatively (20). The effects, and presumably the mechanism of action, of certain G t -derived peptides on photoregeneration are similar to those of G t itself (22).
Photoregeneration Is Sensitive to Interactions with G t and Certain G t -derived Peptides-G t interacts with R* to stabilize the active signaling state such that photoregeneration to rhodopsin is effectively blocked (20). Recently, a peptide corresponding to the carboxyl terminus of G␥ t and a peptide analogue related to the carboxyl terminus of G␣ t were demonstrated to mimic the effect of G t by inhibiting photoregeneration of R* (22). Here we showed that synthetic peptide ␣(340 -350) could cause the same effect as G t (Fig. 2). In addition, the effect of ␣(340 -350) was specific to its primary structure since a mutant peptide had no effect (Fig. 3). Synthetic peptide ␥(50 -71)-far also inhibited photoregeneration of R* (Fig. 2). The effect was specific to its primary structure and to the presence of cysteinyl thioether farnesylation (Fig. 3). Using single peptides that represent small regions of G t provides a powerful probe of subunit-and domain-specific interactions.
The potencies of the ␣(340 -350) and ␥(50 -71)-far peptides are about 20-and 100-fold less than that of G t , respectively (Fig. 2). This finding is reasonable considering that the tertiary structure of a short peptide is less defined, so that a higher binding energy, and thus concentration, is needed for the "induced fit." Also, the cytoplasmic surface domain of R* comprises multiple interaction sites for G t binding, including the loops C2 and C3 (15,16,36). G t also has at least two, and probably more, sites that interact with the receptor during binding and activation. Peptide ␣(340 -350) showed a clear inhibition of photoregeneration with an almost complete suppression at saturating concentrations (Fig. 2B). The peptide ␥(50 -71)-far showed a lower efficacy to inhibit photoregeneration. Although the inhibitory effect saturated at high concentrations with a normal first-order binding isotherm, there was not a complete suppression of the photoregeneration effect (Fig.  2C). The binding of ␥(50 -71)-far to R* is likely to be quite complex due to specificity which arises from both the farnesyl moiety and the peptide sequence (Fig. 3). The carboxyl-terminal region of G␥ t was also studied using a MII difference spectroscopy assay with similar findings (30,31).
The Role of a Conserved Region at the Amino Terminus of Loop C4 of Rhodopsin in G t Binding-We used the photoregeneration assay to probe the effects of G t and G t -derived peptides on expressed rhodopsin and rhodopsin mutants. The rhodopsin loop C4 mutations did not significantly affect the signal transmission path itself, as is seen from the similar kinetics of the photoregeneration signals in the absence of G t and G t -derived peptides (Fig. 4). The results in Figs. 2 and 3 show that G t , and peptides ␣(340 -350) and ␥(50 -71)-far are specific probes of rhodopsin signaling. Therefore, inhibition of photoregeneration by G t or G t -derived peptides is interpreted as the specific interaction of these reagents with the intracellular surface. A lack of inhibition due to alteration of either the peptide or the C4 loop is interpreted as a disruption of interaction. An advantage of the photoregeneration assay is that only productive binding interactions that stabilize specific conformations of the protein are reported.
In theory, a particular mutation might have the effect of uncoupling the conformation of the cytoplasmic surface from that of the chromophore binding pocket. For example, an E134Q mutant has been shown to assume a partially activated conformation at the cytoplasmic surface while the chromophore and surrounding structures remain in the dark, inactive conformation (12). This type of mutant might give misleading results, as photoregeneration (monitored by structural rearrangements surrounding the chromophore) could proceed unhindered, even as G t or peptides bound normally to the cytoplasmic surface. The rhodopsin loop C4 mutants showed no evidence of any uncoupling between the chromophore-binding pocket and cytoplasmic surface conformations. All of the mutants showed similar photoregeneration kinetics in the absence of peptide (Fig. 4, black traces). This suggests that the effects of the mutations are localized to the cytoplasmic surface, and do not affect the photoregeneration process itself. However, the mutant E134R/R135E photoregenerated with kinetics that were distinctly different from that of rhodopsin (data not shown), and therefore was not considered further in this study. Of course, mutants could exist that would foil all assays that rely on the detection of binding events at the intracellular surface resulting from conformational changes that are induced elsewhere in the protein. The extra-MII assay commonly used for rhodopsin, and the GTP-induced agonist affinity shift assay extensively used for other G protein-coupled receptors have the same potential limitations and are much less sensitive and specific.
Taken together, the results in the preceding paper (14) and the biophysical analysis of selected rhodopsin mutants herein strongly suggest that the amino terminus of C4 plays an important role in G t binding and activation. Both G t and ␣(340 -350) binding are disrupted when a tripeptide in this region is replaced by a sequence from the ␤ 2 -AR (Fig. 5). The most straightforward interpretation of the data is that the aminoterminal region of loop C4 directly influences or is part of a direct binding site for the carboxyl-terminal tail of G␣ t .
It is surprising that CTr4, in which residues 310 -312 are replaced with analogous sequence from the ␤ 2 -AR, binds neither ␣(340 -350) nor ␥(50 -71)-far. Can the two peptides bind to overlapping sites on the receptor, each of which includes the 310 -312 region? This seems unlikely, because the carboxyl termini of G␣ t and G␥ t are located at a significant distance apart from each other in the structure of the heterotrimer (19). Two potential explanations, which are not mutually exclusive, arise: 1) the peptides bind to different sites on the receptor and the sites are allosterically coupled; 2) G t undergoes a large conformational change on contact with R* to bring the carboxyl termini of G␣ t and G␥ t into close proximity with the interaction domain near residues 310 -312. A conformational switch in G t , induced by the contact with R*, is an element of the "sequential fit" model (22), and it may be identical to the switch in G␤␥ t that was suggested earlier (31).
Possible Role of G␥ t -farnesyl in Docking of G t to the Active Receptor-The relevance of the data to the binding site of G␤␥ t is more subtle. The observation that CTr2, but not CTr4, binds ␥(50 -71)-far highlights the complexity of the binding interaction between ␥(50 -71)-far and rhodopsin. Further evidence of this complexity, as noted above, is that ␥(50 -71)-far fails to fully inhibit the photoregeneration reaction, even at saturating concentrations. This behavior contrasts with that of ␣(340 -350) (Fig. 2). In addition, both the farnesyl moiety and the peptide itself are required for binding to R* (Fig. 3). Each is likely to have a distinct binding site that may be differently altered in the mutants studied. The binding of ␥(50 -71)-far to CTr2, but not CTr4, suggests that the structural integrity of the fourth loop is disrupted by substitution of 310 -312 with ␤ 2 -AR sequence, but that the substitution of the entire loop restores the structural determinants required for ␥(50 -71)-far binding. In this scenario, the tertiary, but not necessarily the primary structure of C4 would be critical for ␥(50 -71)-far binding. We have recently provided direct evidence, based on monolayer expansion measurements, that both the farnesylated carboxyl terminus of G␥ t and the myristoylated amino terminus of G␣ t are involved in the membrane interaction of G t (37). G␥ tfar plays a role in both membrane and receptor interactions of G t . We hypothesize that opening of the fourth loop structure could guide the farnesylated carboxyl terminus of G␥ t toward the receptor, thus ensuring the docking of G t . By definition, such a process proceeds through predominantly hydrophobic interactions and avoids the need for G t to "jump" onto the receptor via a transiently soluble intermediate state, which would slow the catalytic interaction (38,39). Assuming a fundamentally similar G protein activation mechanism for rhodopsin and the ␤ 2 -AR, it is conceivable that the intact tertiary fourth loop structure in the CTr2 mutant has the capability to form a docking site for the ␥-peptide from G t .
In summary, we developed a novel biophysical assay to probe the G t -binding domain of rhodopsin and expressed rhodopsin. G t and peptides corresponding to the carboxyl-terminal regions of G␣ t and G␥ t specifically bind to R* and stabilize the active state of the receptor. We conclude that the amino-terminal region of loop C4 acts as part of the binding site for G␣ t and modulates the G t -binding domain of R*. Future work is underway to reconcile the various models for allosteric regulation of the G t -binding surface of R*, especially concerning the binding of G␤␥ t . This work will require the use of additional rhodopsin mutants and expressed G protein subunits.