Transducin-α C-terminal Peptide Binding Site Consists of C-D and E-F Loops of Rhodopsin

The binding of heterotrimeric GTP-binding proteins (G-proteins) to serpentine receptors involves several independent contacts. We have deduced the points of interaction between mutant bovine rhodopsins and αt-(340-350), a peptide corresponding to the C terminus of the α subunit (αt) of bovine retinal G-protein, transducin. Direct binding of αt-(340-350) to rhodopsin stabilizes the activated metarhodopsin II state (M II), consequently uncoupling the rhodopsin-transducin interaction. This peptide action requires two segments on the cytoplasmic domain of rhodopsin: the Tyr136-Val137-Val138-Val139 sequence on the C-D loop and the Glu247-Lys248-Glu249-Val250-Thr251 sequence on the E-F loop. We propose that a tertiary interaction of these two loop regions forms a pocket for binding the αt C terminus of the transducin during light transduction in vivo In most G-proteins, the C termini of α subunits are important for interaction with receptors, and, in several serpentine receptors, regions similar to those in rhodopsin are essential for G-protein activation, indicating that the interaction described here may be a generally applicable mode of G-protein binding in signal transduction.

(M II). The M II then binds and activates the retinal G-protein, transducin (G t ). Evidence from peptide competition (6), mutational (7)(8)(9), and biochemical (10) studies have implicated three cytoplasmic regions of M II as being critical for G t interaction. Likewise, in transducin, the ␣ subunit residues 340 -350 at the C terminus, 311-323 at ␣4/␤6/␣5 regions, 8 -23 at the N terminus, and the farnesylated at the C-terminal tail of the ␥ 1 subunit have been shown to be specific contact sites for rhodopsin (11,12). Additional contact sites involving the ␤ subunit are anticipated but have not been mapped. Thus, several distinct contacts are involved in the signal transfer from rhodopsin to G t , but which segment of G t interacts specifically with a particular region of rhodopsin is not known.
This paper focuses on the identification of the residues of bovine rhodopsin that interact with the transducin ␣ subunit C-terminal residues 340 IKENLKDCGLF 350 , a region that is important in rhodopsin-transducin coupling (11,(13)(14)(15)(16). The ability of an 11-amino acid ␣ t -(340 -350) peptide to directly stabilize the M II state of rhodopsin mutants was employed. We report that the binding site consists of the residues Tyr 136 through Val 139 in the C-D loop and the residues Glu 247 through Thr 251 in the C-terminal portion of the E-F loop of bovine rhodopsin.

EXPERIMENTAL PROCEDURES
Expression, Purification, and Characterization of Mutant Rhodopsins-Procedures for the construction of mutants and expression of opsins have been described earlier (17,18). Wild-type and mutant opsin genes (Table I) were expressed in COS1 cells by transient transfection of corresponding gene. The rhodopsin chromophore was generated by adding 11-cis-retinal (40 M) to a cell suspension, the cells were solubilized in 1% dodecyl maltoside, and rhodopsin was purified by immunoaffinity chromatography (17,18). The pigment concentration was calculated from its absorbance at 500 nm based on E 500 ϭ 42,700 M Ϫ1 cm Ϫ1 . For rhodopsin samples prepared for M I 7 M M II equilibrium studies, the dodecyl maltoside was replaced with 1% digitonin in all washes and elution.
Transducin Activation-Transducin was isolated from the bovine rod outer segment as described by Fung et al. (19). Catalytic activation of transducin by wild-type and mutant rhodopsins was assayed by a (GTP␥S) binding assay as described by Wessling-Resnick and Johnson (20). The assay mixtures consisted of 1-5 nM purified rhodopsin, 2 M transducin, 20 M [ 35 S]GTP␥S (1130 Ci/mmol) in 10 mM Tris-HCl, pH 7.2, 100 mM NaCl, 5 mM dithiothreitol, and 0.012% dodecyl maltoside. The assay was initiated by illumination for 2 min at a wavelength greater than 495 nm. The reaction mixture then remained in the dark at 23°C for 60 min. The number of moles of [ 35 S]GTP␥S bound per mol of rhodopsin in 60 min was estimated from the [ 35 S]GTP␥S retained on the filter after filteration and washing.
Synthesis and Characterization of Peptides-The ␣ t -(340 -350) peptide, Ac-IKENLKDCGLF, and seven analogues ( Fig. 2A) were synthesized, purified, and characterized by the protein chemistry core services of the Research Institute of the Cleveland Clinic as described earlier (18). These peptides will be referred to as peptides 1 (the parent peptide) through 8.
The ␣ t -(340 -350)-induced M I 7 M II Equilibrium Assay-The principles and procedure for the M I 7 M II equilibrium assay have been described previously (16,18). In a typical experiment, ϳ5 to 8 ϫ 10 Ϫ8 M * This work was supported in part by NEI Grant 09704 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
rhodopsin was evaluated with 1 ϫ 10 Ϫ4 , 5 ϫ 10 Ϫ4 , and 1 ϫ 10 Ϫ3 M concentrations of each peptide. The ␣ t -(340 -350) peptide or its analogues were mixed with wild-type and mutant rhodopsins in 1% digitonin in the dark and kept on ice for 20 min. Dark spectra were recorded at 5°C. The samples were then exposed to light for 20 s using a 150-watt Fiber-Lite fitted with 490 nm cut-off filter; the sample was allowed to equilibrate in the dark at 5°C for 20 min, and then light spectra were recorded (see Figs Previous studies have found that bleaching rhodopsin in dodecyl maltoside in the absence of G t yields the active M II state with a max ϳ380 nm, an intermediate that stimulates [ 35 S]GTP␥S binding to G t (steps 2 through 7). In contrast, bleaching rhodopsin in 1% digitonin in the absence of G t yields predominantly the inactive M I ( max ϳ480 nm) intermediate (step 2) and a small amount of M II (7,17,18,21). Formation of the M II state in digitonin requires a stoichiometric mixture of rhodopsin and G t because instantaneous stabilization of M II by G t is necessary (steps 3 and 5). Adding the non-hydrolyzable GTP analogue, GTP␥S, destabilizes the M II:G t complex (step 7) and shifts the equilibrium in favor of the M I state (21). The formation of the M II state in digitonin can also be induced by ␣ t -(340 -350) rather than G t (steps 3 and 4). The stabilization of the M II state depends on the amino acid sequence and concentration of the ␣ t -(340 -350) peptide (11,18). Thus, ␣ t -(340 -350), through binding to a site on rhodopsin, competes with G t to shift the M I 7 M II equilibrium and thus inhibits [ 35 S]GTP␥S binding to G t (step 6).
To identify the site on rhodopsin required for binding ␣ t -(340 -350), wild-type and mutant rhodopsins purified from transfected COS1 cells were used in both assays. The mutant opsins (Table I) expressed well (expression was estimated by Western blot analysis, not shown), bound 11-cis-retinal and yielded a chromophore within 90% of that yielded by the wildtype, indicating that the mutant polypeptides folded normally to a native state. All mutant rhodopsins purified in dodecyl maltoside and digitonin yielded a chromophore with a max ϳ500 nm (data not shown). Light-activated rhodopsin samples in dodecyl maltoside were used for measuring [ 35 S]GTP␥S binding to G t ( Fig. 2A and Table I). The wild-type rhodopsin activated nearly 267 Ϯ 29 mol of G t /mol of rhodopsin. Unregenerated opsin and rhodopsin not exposed to light activated 15 Ϯ 2 mol of G t /mol of rhodopsin.
The synthetic 11-amino acid ␣ t -(340 -350) peptide 1 inhibited (50 Ϯ 10% of maximal) G t activation by bleached rhodopsin. The apparent K i for inhibition by peptide 1 was 80 M. In agreement with earlier studies using urea-washed rod outer segment disc membranes, an inhibition greater than 50% could not be achieved at a higher peptide 1 concentration (11). As shown in Figs and G t with the detergent-solubilized wild-type rhodopsin obtained from COS1 cells had properties identical to those reported for interaction with bovine retinal rhodopsin (11).
Nature of the Interaction between ␣ t -(340 -350) and M II-To determine which amino acid side chains of ␣ t -(340 -350) are important for M II stabilization, the synthetic analogues shown in Fig. 2A were used. The C-terminal 347 CGLF 350 sequence was not examined because this region has already been shown to form a ␤-turn structure, and the subtype of the ␤-turn is speculated to be important for receptor selectivity (16,22). Conservative single amino acid substitution of the remaining seven residues led to varying phenotypes (Fig. 2). Peptides 2 and 8 competed as effectively against G t as the parent peptide 1 in both assays. Peptide 5 was a slightly (Ϸ2-fold) more effective competitor of G t and also better shifted the M I 7 M II equilibrium in favor of M II (Fig. 2, A and B). Peptide 3 had a slightly lower potency (Ϸ2.5-fold less) than peptide 1. Peptides 4, 6, and 7 were very poor competitors in the G t activation assay and were completely ineffective in shifting the M I 7 M II equilibrium (Fig. 2B).
Studies using transferred nuclear Overhauser effect spectroscopy suggested that a salt bridge between Glu 342 and Lys 345 exists in ␣ t -(340 -350) bound to rhodopsin in the dark, that is broken during M II stabilization and replaced by a new salt bridge between Lys 345 and the free ␣-COO Ϫ group of the peptide (16). The Gln substitution can provide hydrogen bonding interactions in the place of a salt bridge. However, the Lys 345 3 Gln change (peptide 7) is likely to affect both conformations required for binding to rhodopsin, in the dark, as well as M II. But, the Glu 342 3 Gln change in peptide 4 is expected to favor the conformation that enables M II stabilization. The lack of peptide 4 binding suggests that Glu 342 is essential for stabilizing M II state. Lys 341 made a negligible contribution. Leu 344 is a critical residue. Substitution with shorter Ala (peptide 6) produced an inactive peptide indicating that hydrophobicity and the side chain size of Leu 344 side chain are stringent requirements for interaction with rhodopsin. Consistent with this observation, Martin et al. (23) discovered that combinatorial analogues of ␣ t -(340 -350) preserve the shape of the hydrophobic "face" with little variation, whereas larger changes in the hydrophilic face are tolerated. Mutagenesis studies suggest a binding preference for hydrophobic amino acids at the Leu 344 and Leu 349 positions of ␣ t (13,14). Thus, the site on rhodopsin that binds ␣ t -(340 -350) is expected to consist of charged and hydrophobic residues. Furthermore, the analogue and mutagenesis studies combined together indicate that the ␣ t -(340 -350) peptide and residues 340 -350 of G t are similar in their interactions with M II and therefore very likely bind to a common site on rhodopsin.
Localization of ␣ t -(340 -350) Binding Residues of Rhodop-sin-To identify the rhodopsin ␣ t -(340 -350) binding site, we created rhodopsin mutants in three distinct cytoplasmic regions involved in the G t interaction. Amidst these, it should be possible to identify mutants in which the ␣ t -(340 -350) binding is abolished even though the interaction with G t is not completely abolished. The abolished ␣ t -(340 -350) binding should be restored upon re-introduction of the wild-type amino acid sequence. Five C-D loop mutants and six E-F loop mutants were chosen for analysis. In these mutants, formation of rho- Transducin activation by mutant and wild-type rhodopsin were carried out as described under "Experimental Procedures." The relative mean GTP␥S binding values (S.E. Ͻ 5%) are equilibrium values obtained without any correction for the decay of metarhodopsin II that might have occurred during the 60-min assay period. The results presented are a mean of 3 to 5 separate experiments in which mutant and wild-type rhodopsin expressed in COS cells were assayed in parallel and the mutant activities were normalized to the wild-type activity. The concentration of opsin was estimated from the max assuming a molar extinction coefficient of 42,000 mol Ϫ1 cm Ϫ1 .

Mutant
Sequence a Transducin activation (fraction of wild-type) Peptide 1 binding b

150
Wild-type dopsin and M II-like states determined by spectral analysis was not altered, but the activation of G t by the mutant M II was altered to various degrees (Table I). The mutants of the remaining cytoplasmic region implicated in G t binding (residues 310 -322 at the membrane-proximal carboxyl tail region of rhodopsin) had a low yield of a rhodopsin-like chromophore and M II-like photo product. Hence, they were not examined further.
In a previous study we concluded that the residues Glu 134 and Arg 135 in the C-D loop are not essential for ␣ t -(340 -350) binding (18). When the highly conserved Tyr 136 residue was substituted with Gly, the resulting mutant bound retinal poorly (data not shown). The mutant CD1, in which the residues 137-140 were replaced by four alanines to preserve the ␣-helical potential but alter the amino acid sequence, transducin activation was reduced Ϸ30%. As shown in Fig. 3, the M II yield was Ϸ20% in the presence of Ϸ2000-fold excess of ␣ t -(340 -350). To specifically determine the role of the Tyr 136 residue in the context of the CD1 mutant sequence, the resi-dues 136 -140 (YVVVC of the wild-type) were replaced with LAAAA (mutant CD2). Transducin activation was reduced Ϸ80%. A nearly 6000-fold excess of ␣ t -(340 -350) did not shift the M II 7 M I equilibrium (Fig. 3). Thus, Tyr 136 is an important determinant for ␣ t -(340 -350) stabilization of the M II state. Previously, Ridge et al. (24) used cysteine scanning mutagenesis to demonstrate that individual replacement of Tyr 136 , Val 137 , Val 138 , and Val 139 led to partial loss of G t activation. The Cys 140 residue was found to be not essential (17,24). Therefore, the VVV sequence following Tyr 136 may contribute stabilizing interactions. In our study, the remaining substitution mutants (CD3, CD4, and CD5) caused partial loss of G t activation but showed normal affinity for the ␣ t -(340 -350) (data not shown). Thus, the residues 141-150 of the C-D loop do not participate in ␣ t -(340 -350) binding.
Franke et al. (7) found that replacing the E-F loop region between residues 231 and 252 with an amino acid sequence from the extracellular loop B-C produced a rhodopsin molecule that was normally activated by light but stimulated transducin very poorly. We constructed the same mutant (EF1 in Table I). The mutant exhibited normal photocycle properties as reported earlier and also activated transducin at only Ϸ9% of the wildtype control. This mutant produced an M I-like state when bleached in digitonin. The M II 7 M I equilibrium of the mutant was not shifted by ␣ t -(340 -350), suggesting that this mutant lacks the binding site for the peptide (Table I and Fig. 3).
We constructed mutants EF2 through EF5 by restoring the wild-type amino acid sequence in different parts of the E-F loop region. The mutant EF6 was constructed to examine the remaining two residues predicted to be the part of the E-F loop. As indicated in Fig. 3 and Table I, mutants EF2, EF3, and EF6 activated transducin at Ϸ40 -60% of the wild-type. The mutants EF2 and EF3 bound ␣ t -(340 -350) almost as well as the wild-type. The mutant EF6 exhibited an interesting phenotype. The ␣ t -(340 -350) binding was evident because the M I peak decreased. However, this decrease was not accompanied by a transition to a distinct M II peak but rather by an increase in light scattering at the spectral region below 380 nm. The EF6 mutation must either alter the affinity for ␣ t -(340 -350) or decrease the stability of the M II⅐␣ t -(340 -350) complex. Therefore, removing the Val 250 -Thr 251 side chains likely indirectly influences the ␣ t -(340 -350) interaction. The mutant EF4 was essentially inactive in both the peptide binding and G t stimulation assays. Examination of the residues in this mutant indicates that hydrophilic and charged residues present in the wild-type rhodopsin E-F loop are replaced by hydrophobic (Leu), shorter (2 Gly residues), and hydrogen-bonding (Asn, Ser, and Thr) residues.
On the basis of earlier mutagenesis studies, the Glu 247 -Lys-Glu-Val-Thr 251 sequence is thought to be essential for efficient activation of transducin and that the other residues play a relatively minor role (7,9). The EKE triad sequence was kept in the mutant EF5. The M II 7 M I mixture generated by bleaching this mutant was shifted toward M II formation by ␣ t -(340 -350), suggesting that the peptide was now able to bind and stabilize the M II intermediate. Transducin activation was partially restored (Ϸ30%). Therefore, it seems reasonable that the charged triad Glu 247 -Lys 248 -Glu 249 is necessary for the M II 7 M I equilibrium shift promoted by ␣ t -(340 -350) binding. We conclude that ␣ t -(340 -350)-mediated stabilization requires the hydrophobic residues Tyr 136 -Val-Val-Val 139 on the C-D loop and the hydrophilic charged residues Glu 247 -Lys-Glu-Val-Thr 251 on the E-F loop of rhodopsin. The type of analysis used here is not sensitive enough to determine which side chains of the ␣ t -(340 -350) interact with each of the regions. The molar ratio of wild-type rhodopsin to peptide 1 was ϳ1:1400, and the ratio of various mutant rhodopsins to peptide 1 varied from 1:3000 to 1:13,000. Fig. 4 depicts the location of the two sites required for ␣ t -(340 -350) binding in the cytoplasmic extensions of the C and F helices. In conventional models, the transmembrane helices of rhodopsin are terminated at the membrane-aqueous interface. However, recent site-directed spin-labeling studies suggest that Ϸ1 to 3 turns of the helices extend into the cytoplasm, with helix C having a close tertiary interaction with helices B, D, E, and F. In a revised model, the Tyr 136 and Val 139 of helix C faces helix F, and Lys 247 of helix F faces helix C (24,25). This observation supports our hypothesis that the tertiary interaction of Tyr 136 -Cys 140 and Glu 247 -Thr 251 regions forms a subsite that is stabilized by ␣ t -(340 -350). The spin-labeling studies suggest that both these segments are rigid relative to the helix E extension, which is more dynamic and not essential for binding the ␣ t -(340 -350). Based on these observations, some qualitative conclusions can be drawn regarding M II stabilization by ␣ t -(340 -350) and holotransducin in vivo. Perhaps the rigidity of the cytoplasmic helix C and helix F extensions is required to provide an optimal surface for binding. The M II stabilization may occur because entropy is lost after ␣ t -(340 -350) has bound to the rigid cytoplasmic extensions of helices C and F. This loss explains M II⅐G t complex stabilization by the G t-␣ residues 340 -350, which are currently believed to be disordered in the heterotrimer (26). It is noteworthy that these two rhodopsin helices contact the ionone ring of the 11-cis-retinal chromophore (27,28).
It is now generally assumed that the G-protein binding site of all GPCRs comprises regions from the C-D loop, E-F loop, and the membrane-proximal segment of the cytoplasmic tail (1)(2)(3)(4)(5)(6)(7)(8)(9). Various types of studies indicate that the E-F loop is preeminent in the G-protein activation process in GPCRs (3)(4)(5). A hydrophobic site near the N-terminal region of the E-F loop that is important for G-protein coupling in several GPCRs (4, 29) appears not to be crucial for ␣ t -(340 -350) interaction with rhodopsin. Instead, our results indicate that the ␣ t -(340 -350) binding involves a hydrophobic region of the C-D loop. The hydrophilic and charged portion of this pocket near the C terminus of the E-F loop corresponds to a site that is important for G-protein activation in several GPCRs (4). E-F loop regions of different GPCRs were found to cross-link to specific ␣-subunits (30), as well as to ␤-subunits of G-protein heterotrimers (31). All these evidences suggest that the E-F loop may wrap around the G-protein heterotrimer, establishing contacts with critical regions of the ␣-subunit, as well as with the ␤␥ complex. Our results are the first description of an interaction between a defined region on transducin and a specific site on the receptor. Our approach could be used to explore the three other subsites on rhodopsin for G t regions, ␣ t -(311-323), ␣ t - (8 -23), and the farnesylated ␥ t -(60 -71) residues.