STUDIES OF THE INTERACTION BETWEEN THE RHODOPSIN CYTOPLASMIC DOMAIN AND TRANSDUCIN*

Structural requirements for the activation of transducin by rhodopsin have been studied by site-specific mutagenesis of bovine rhodopsin. A variety of single amino acid replacements and amino acid insertions and deletions of varying sizes were carried out in the two cytoplasmic loops CD (amino acids 134-151) and EF (amino acids 231-252). Except for deletion mutant A137-150, all the mutants bound 11-cis-retinal and displayed normal spectral characteristics. Deletion mutant A236-239 in loop EF caused a 50% reduction of transducin activation, whereas deletion mutant A244-249 and the larger deletions in loop EF abolished transducin activation. An 8-amino acid deletion in the cytoplasmic loop CD as well as a replacement of 13 amino acids with an unrelated sequence showed no transducin activation. Several single amino acid substitutions also caused significant reduction in transducin activation. The conserved charged pair Glu-1341 Arg-135 in the cytoplasmic loop CD was required for transducin activation; its reversal or neutralization abolished transducin activation. Three amino acid replacements in loop EF (S240A, T243V, and K248L) resulted in significant reduction in transducin activation. We conclude that 1) both the cytoplasmic loops CD and EF are required for transducin activation, and 2) effective functional interaction between rhodopsin and transducin involves relatively large peptide sequences in the cytoplasmic loops.


STUDIES OF THE INTERACTION BETWEEN THE RHODOPSIN CYTOPLASMIC DOMAIN AND TRANSDUCIN*
Structural requirements for the activation of transducin by rhodopsin have been studied by site-specific mutagenesis of bovine rhodopsin. A variety of single amino acid replacements and amino acid insertions and deletions of varying sizes were carried out in the two cytoplasmic loops CD (amino acids 134-151) and EF (amino acids 231-252). Except for deletion mutant A137-150, all the mutants bound 11-cis-retinal and displayed normal spectral characteristics. Deletion mutant A236-239 in loop EF caused a 50% reduction of transducin activation, whereas deletion mutant A244-249 and the larger deletions in loop EF abolished transducin activation. An 8-amino acid deletion in the cytoplasmic loop CD as well as a replacement of 13 amino acids with an unrelated sequence showed no transducin activation. Several single amino acid substitutions also caused significant reduction in transducin activation. The conserved charged pair Glu-1341 Arg-135 in the cytoplasmic loop CD was required for transducin activation; its reversal or neutralization abolished transducin activation. Three amino acid replacements in loop EF (S240A, T243V, and K248L) resulted in significant reduction in transducin activation. We conclude that 1) both the cytoplasmic loops CD and EF are required for transducin activation, and 2) effective functional interaction between rhodopsin and transducin involves relatively large peptide sequences in the cytoplasmic loops.
Rhodopsin is the photoreceptor in the retinal rod cell. Its primary structure has been established through both peptide and DNA sequencing (1)(2)(3). The apoprotein consists of a single polypeptide chain of 348 amino acids. Hydropathy plots, proteolysis experiments, and binding studies with monoclonal antibodies suggest that rhodopsin contains seven transmembrane segments (A-G) with water-exposed polypeptide domains on the cytoplasmic and intradiscal sides (2,(4)(5)(6)(7). A secondary structure model identifying the seven putative a-helical transmembrane segments is shown in Fig. 1. The chromophore 11-cis-retinal, covalently linked to Lys-296 as a Schiff base, is embedded in the hydrophobic core of rhodopsin (8). Absorption of light by rhodopsin causes isomerization of the chromophore from 114s-to all-trans-retinal and drives rhodopsin through a series of structural changes leading to the photointermediate metarhodopsin I1 (MU).' MI1 activates transducin and also mediates interactions with other proteins on the cytoplasmic surface. The interaction between MI1 and transducin is an important early step in signal transduction in the visual process. Because of its fundamental importance, we are interested in studying this interaction at the molecular level. Insights into rhodopsin-transducin interaction should be of general signi~cance for the study of the superfamily of receptors that are coupled to G proteins.
Some information on the rhodopsin-transducin interaction has been obtained from previous studies. Proteolysis of rhodopsin suggested the involvement of loop EF in transducin binding (9). In another approach, antipeptide antibodies directed against specific sites on the cytoplasmic domain of rhodopsin were tested for their ability to block transducin binding (10). More recently, peptides corresponding to portions of the cytoplasmic loops in rhodopsin were used to test for inhibition of rhodopsin-transducin interaction (11). It was concluded that as many as three c~oplasmic loops of rhodopsin may be involved in interaction with transducin. However, specific questions regarding the binding and activation of transducin by rhodopsin remain unanswered. What is the nature of the structural interaction between transducin and rhodopsin? What are the conformational changes in the two molecules during this interaction? What are the mechanisms of transducin-mediated GTP-GDP exchange, the subsequent release of Ta-GTP, and the separation of the 8--y subunits as a complex that occur on the cytoplasmic face of rhodopsin?
Recently, we reported on rhodopsin mutants with deletions in the c~oplasmic loops (1'2). These mutants bound but failed to activate transducin. A conserved charge pair (Glu-l34/Arg-135) in rhodopsin was suggested to be a part of the transducinbinding site. Furthermore, deletion and replacements of large peptides showed that substantial portions of loops CD and EF were necessary for functional interaction of rhodopsin with transducin. We have now continued to investigate rhodopsin-transducin interaction by extensive application of sitespecific mutagenesis in the cytoplasmic loops CD and EF in rhodopsin (Fig. 1). The mutations carried out include deletions of peptide sequences of varying sizes (Table I) and single amino acid replacements as well as amino acid insertions (Tables I1 and 111). Our results indicate the importance of specific peptide sequences in rhodopsin in the above interaction. Thus, whereas a mutant with a deletion of an 8-amino acid sequence from loop CD showed a normal UV-visible spectrum, it failed to activate transducin. Another loop CD ' The abbreviations used are: MII, metarhodopsin 11; G protein, guanine nucleotide-binding regulatory protein; dATPyS, deoxyadenosine 5'-O-(thiotriphosphate). " 14767 mutant with 13 amino acids replaced by an unrelated amino acid sequence formed a normal chromophore, but also failed to activate transducin. A deletion of 14 amino acids in the same loop caused loss of retinal binding capability, presumably because of structural constraints introduced by the deletion. In loop EF, the carboxyl-terminal region (amino acids 244-249) and the region containing potential phosphorylation sites near the center of the loop (amino acids 236-243) are evidently important for transducin activation. Of the presumed 22 amino acids in loop EF, 19 could be deleted without affecting retinal binding, although transducin activation was lost. Finally, systematic analysis of single and multiple amino acid substitutions in loop E F showed that 3 amino acids  were particularly important. We conclude that rhodopsin (meta-rhodopsin 11)-transducin interaction involves both loops CD and EF in rhodopsin and that the interaction involves large portions of these loops.

EXPERIMENTAL PROCEDURES
Materials-The expression vector pMT2, a 6-lactamase derivative of p91023 (13) A-G. One-letter abbreviations are used for amino acids. The dashed line between Cys-110 and Cys-187 on the intradiscal side shows a disulfide bond. Cys-322 and Cys-323 on the cytoplasmic side are palmitoylated. The amino acids in the cytoplasmic loops CD and EF that are the focus of this work are in boldface type. of the buffers and media have been described (14,15).
Monoclonal Antibody-The 1D4 hybridoma cell line was generously provided by Dr. R. S. Molday (16). The hybridomas were grown in large scale in the tissue culture facilities of the Massachusetts Institute of Technology Cancer Center. 1D4 antibodies were harvested from the hybridoma media by (NH4)2S04 precipitation. The coupling of the antibody to Sepharose 2B has been described (14). The octadecapeptide (positions -1' to -18' from the carboxyl-terminal end of rhodopsin) used to elute rhodopsin from the immunoaffinity resin was the gift of Dr. P. Kim. Preparation of Oligonucleotides-Oligonucleotides were synthesized on an Applied Biosystems Model 380A DNA synthesizer. The purification and characterization of the oligonucleotides were performed according to Ferretti et al. (17).
Construction of Rhodopsin Genes with Mutations in Loop CD: Mutants CD-1 to CD-&"he mutant opsin genes were constructed by restriction fragment replacement in the synthetic rhodopsin gene (17). For the construction of the mutants targeting the charged pair Glu-134/Arg-135 (mutants CD-1 to CD-5) (Table II), two new unique restriction sites (RsrII and SpeI) were introduced into the synthetic gene in the expression plasmid (18). After digestion with RsrII and SpeI, the large fragment was separated on an agarose gel and was ligated with the synthetic DNA duplexes with the desired codon alteration(s).
Deletion Mutations in Loop CD-The two deletion mutations CD-A1 and CD-A2 and the loop replacement mutation CD-7 were introduced between the restriction sites PuuI and Ah11 (Fig. 2). Because these sites were not unique within the plasmid, digestions with ApaI and AurII were used to generate a fragment in which PuuI and AhaII sites were unique. The purified ApaI-AurII fragment was digested with PuuI and AhaII to generate three fragments. The fragments were separated by agarose gel electrophoresis, and the two large fragments were purified. Because the ligation of PuuI cleavage sites is sensitive to deoxyadenosine methylation, the plasmid used for PuuI digests was isolated from a dam-strain of Escherichia coli. The large fragment from the ApaI-AurII digestion was prepared separately from a plasmid isolated from E. coli strain DH1. Fragments 1-111 and the synthetic DNA duplex were ligated (Fig. 2). Ligation mixtures were used to transform CaClz-treated E. coli strain DHl. Plasmid DNA was prepared from ampicillin-resistant colonies.
Rhodopsin Genes with Mutations in Loop EF-Deletion mutants EF-A1 to EF-A4 (Table I) and substitution mutants EF-2 to EF-15 (Table 111) were assembled in two-component ligations using the large MluI-PstI restriction fragment and synthetic duplexes containing the desired codon alteration(s).
Mutant EF-1 (Table ZZZ)-Mutant EF-1 was constructed in a threecomponent ligation using an additional AuaII-PstI restriction fragment from the synthetic gene. For the preparation of mutant EF-A5, the pMT4 vector was digested with MluI and PstI. The large restriction fragment was purified from an agarose gel, and the singlestranded overhangs were removed by digestion with mungbean nuclease. The blunt ends were ligated to recircularize the plasmid and to yield an in-frame deletion. Ligation mixtures were used to transform CaClZ-treated E. coli strain DH1. Plasmid DNA was prepared from ampicillin-resistant colonies. A small restriction fragment containing the mutation was subcloned into pMT4 to minimize the chance of spurious mutations caused by nuclease treatment. The ligation mixture consisted of an EcoRI-HinfI fragment (613 base    Expression of Rhodopsin Mutants in COS-1 Cells-The procedure for the transient transfection of COS-1 cells has been reported (14, 15). COS-1 cells were plated at a density of -5 X lo6 cells/lO-cm culture plate and transfected within 14-18 h with 8 pg of CsCIpurified plasmid DNA/plate. Cells were harvested 72 h after transfection.
Binding of 11-&-Retinal Chromophore and Purification of Rhodopsin Mutants-Freshly harvested COS cells (six plates in 6 ml of phosphate-buffered saline) were incubated with 35 pl of 11-cis-retinal in ethanol (1 mM) at 4 "C in the dark. A second aliquot of retinal solution (35 p l ) was added after 45 min and incubated for an additional 45 min. The cells were collected by centrifugation, and the supernatant fraction was discarded. The cell pellet was resuspended in 10 ml of solubilization buffer (50 mM Tris-HCI, pH 6.8, 100 mM NaCI, 1 mM CaCI2, 1 mM MgCI,, 1% dodecyl maltoside, 0.1 mM phenylmethylsulfonyl fluoride). After 30 min a t 4 "C with gentle mixing, the insoluble material was pelleted by centrifugation a t 100,000 X g for 30 min at 4 "C. The supernatant fraction, containing the solubilized rhodopsin, was incubated for 3 h a t 4 "C with 150 pl of 1D4-Sepharose. The resin was collected by centrifugation, the 5 ml of wash buffer (50 mM Tris-HCI, pH 6.8, 100 mM NaCI, 1 mM supernatant fraction was removed, and the resin was resuspended in CaCL, 0.1% dodecyl maltoside) and washed for 5 min a t 4 "C with gentle mixing. The wash procedure was repeated a total of five times. Elution of the rhodopsin from the immunoaffinity resin was carried out in 1 h in the presence of carboxyl-terminal peptide 1'-18' (180 pg of peptide/ml of buffer) as previously described (14) Synthetic DNA duplexes containing the desired codon alterations were used to replace a PuuI-Aha11 restriction fragment in the synthetic rhodopsin gene. These restriction sites were not unique within the plasmid. Therefore, multiple restriction digests and restriction fragment purifications were required as described under "Experimental Procedures." In summary, each mutant was constructed in a fourcomponent ligation consisting of the following: 1) a 5412-base pair ApaI-AurII fragment, 2) a 415-base pair AurII-PuuI fragment, 3) a 380-base pair AhaII-ApaI fragment, and 4) a 78-base pair synthetic duplex containing the desired codon alterations. bp, base pairs. containing eluted rhodopsin was centrifuged at 100,000 X g for 30 min.
Rhodospin in Digitonin-Transfected COS-1 cells (12 dishes, 10mm plates) were harvested, regenerated with 11-cis-retinal, solubilized, and incubated with the resin as described above. The resin was split into two fractions. One was treated as described above in dodecyl maltoside. The other fraction was washed and eluted in buffer containing 0.1% digitonin instead of dodecyl maltoside.
Characterization of Rhodopsin Mutants-Purified rhodopsin mutants were characterized in three ways: ( a ) UV-visible spectroscopy (rhodopsin concentrations were based on the absorbance difference a t 500 nm before and after illumination assuming a molar absorption coefficient of c = 42,700 M" cm"), (b) sodium dodecyl sulfatepolyacrylamide gel electrophoresis and visualization of the protein bands by silver staining, and (e) transducin activation assay as described below.
Transduein Actiuation Assay-The rhodopsin mutants were assayed for their ability to stimulate GTPase activity of transducin in a light-dependent manner. The assay mixture (100 p l ) contained 2.5 nM rhodopsin, 2.5 ~L M purified transducin, 20 p~ [Y-~*P]GTP, 0.01% dodecyl maltoside, 10 mM Tris maleate, pH 7.2, 100 mM NaC1, 2 mM MgCI,, and 1 mM dithiothreitol. All the components except G T P were mixed in the dark, and the solution was equilibrated a t 25 "C. Illumination was performed with a 150-watt fiber optic light source and a 495-nm cutoff filter in tandem with an IR filter. After continuous illumination for 1 min, the reaction was started by the addition of GTP. Aliquots (20 pl) were removed at 2,4,6, and 8 min and added to 200 pl of molybdic acid solution (6.25 g of MOO:, dissolved in 35 ml of concentrated H,SO, and diluted to 500 ml with H20). 0.1 ml of a reducing solution (5.7 g of Na2Sz05, 0.2 g of Na2SO:,, and 0.1 g of 1amino-2-naphthol-4-sulfonic acid dissolved in 100 rnl of water) was added, and the solution was mixed. The mixture was extracted by vortexing with 700 pl of isoamyl alcohol. After phase separation by  FIG. 3. Comparison of UV-visible spectra of COS cell rhodopsin solubilized in digitonin or dodecyl maltoside. Left, UV-visible absorption spectrum in digitonin. The procedure for the purification of rhodopsin and solubilization in digitonin has been described under "Experimental Procedures." Spectral ratios of C2. 4 were not obtained using this procedure. Right, UV-visible absorption spectroscopy of rhodopsin in dode-cy1 maltoside. The procedure used is as described under "Experimental Procedures.'' The A280/A5m spectral ratio is indicative of the purity of the preparation. Spectral ratios in the range of 1.6-1.8 were routinely obtained. UV-visible spectra were taken before and immediately after a 3-s illumination with light >495 nm. At 4 "C, only MI (480 nm) was present.
At 10 "C, MI and a small amount of MI1 (380 nm) was formed. At 15 "C, the amount of MI1 increased. In dodecyl maltoside detergent buffer, illumination of rhodopsin (Rho) at 4 "C or above resulted in complete conversion to MI1 (data not shown).
centrifugation, 0.6 ml of the organic layer was analyzed for Pi by scintillation counting.

Characterization of Rhodopsin Expressed in COS Cells:
Influence of Detergents UV-visible Spectral Characteristics-Rhodopsin prepared from COS cells using dodecyl maltoside showed an absorption ratio a t 280 nm/500 nm of 1.6-1.7 (Fig. 3). Bovine rhodopsin, purified in parallel, gave the same spectral ratio. COS cell rhodopsin purified using digitonin gave A280/A500 ratios of 2.4-4 for different preparations (Fig. 3). Photoactiuation and Stability of Intermediates-Illumination of rhodopsin in dodecyl maltoside with light >495 nm for 10 s quantitatively converted all of the pigment to MI1 (380 nm). No temperature effect was seen for this conversion between 4 and 25 "C. Illumination of rhodopsin in digitonin with a 495-nm cutoff filter gave a mixture of MI (480 nm) and MI1 species (Fig. 4), whose composition was temperaturedependent.
GTPase Activity in Transducin-The linear range for rhodopsin activation of the GTPase activity in transducin was determined by assaying rhodopsin at concentrations ranging from 200 pM to 5 nM (Fig. 5). The precision of this assay was estimated to be +lo%. COS cell rhodopsin was assayed in parallel with each rhodopsin mutant as an internal standard. To determine the pH optimum for the rhodopsin-transducin   Table I. GTPase assays with purified mutants and transducin were carried out as described under "Experimental Procedures." The extinction coefficients of rhodopsin and the mutant pigments were assumed to be 42,700 M" cm" (29).
Activities were not adjusted for any decay of metarhodopsin I1 that might have occurred during the 8-min course of the assay. Results are presented as mean f S.D. averaged from two to four separate experiments as indicated by the numbers in parentheses. Purified rhodopsin mutants and COS cell rhodopsin were assayed in parallel. The mutant activities were normalized to the COS cell rhodopsin activities.

Deletions and Sequence Replacements in Loops CD and EF
Two questions were addressed in considering the mutations that were to be introduced. First, does the interaction between rhodopsin and transducin require participation of peptide sequences in one or both cytoplasmic loops of rhodopsin? Second, are there electrostatic or hydrogen bond interactions between specific amino acids in rhodopsin and transducin?
Loop CD-The deletions introduced in loop CD are shown in Table I. Mutant CD-Al, with a deletion of 8 amino acids (amino acids 143-1-50), bound 11-cis-retinal and formed the characteristic Amax at 500 nm. However, it showed no activation of transducin (Table IV). Mutant CD-A2, with a 14amino acid deletion, was expressed at normal levels in COS cells, but it failed to bind 11-cis-retinal. To remove possible constraint in the packing of helices, a 13-amino acid segment in loop CD was replaced by an amino acid sequence derived from the first intradiscal loop (mutant CD-7) (Table 11). This mutant bound 11-cis-retinal and displayed a normal UVvisible spectrum, but showed no transducin activation (Table  VI. Loop EF-Mutant EF-Al, with 6 amino acids deleted close to the beginning of helix F (amino acids 244-249), failed to show transducin activation. Deletion EF-A2, closer to the end of helix E, activated transducin a t 54% of the wild-type level, whereas EF-A3, with a deletion 4 amino acids larger than that in EF-A2, showed only slight (3% of the wild type) transducin activity (Table IV). Mutants EF-A4 and EF-A5, which contained 13-and 19-amino acid deletions (Table I), respectively, both failed to stimulate the transducin GTPase activity, although they formed a normal chromophore with ll-cis-retinal.
Peptide Sequence Replacement in Loop EF-In mutant EF-15 (Table 111), amino acid sequence 235-250 was replaced with an amino acid sequence from loop BC (positions 97-112, except for the change Cys-110 + Ser). This mutant was designed to remove sequence-specific interaction with transducin as in the other deletion mutants, but without affecting packing of the helices. The mutation retained a Thr at position 243. Mutant EF-15 activated transducin at a very low level (7%) ( Table VI).
Loop CD-To investigate the role of the conserved charged pair Glu-134/Arg-135, three mutants with single amino acid substitutions and two mutants with double substitutions were constructed and characterized (Table  11). Light and dark spectra of the purified retinal-regenerated mutants are shown in Fig. 7. Mutant E134Q (CD-2) showed 1.45 times higher activity than wild-type rhodopsin in the GTPase assay. Mutant E134D (CD-1) stimulated GTPase activity to 56% of the wild type. Mutant R135Q (CD-4) showed 8% of the wild-type GTPase activity. Both double mutants E134A/R135A (CD-5) and E134R/R135E (CD-6) failed to activate transducin.
Loop EF-Point mutants were introduced in the Ser and Thr residues in loop EF (Table 111). Mutant S240A (EF-3) showed 60% GTPase activity compared to the wild type. Mutant T243V (EF-4) showed 40% activity in the GTPase assay. Mutant EF-6, with the 3 Ser and Thr residues in loop EF replaced (S240A, T242G, and T243G), showed only 8% of the wild-type GTPase activity. Mutant EF-5, which also had the 3 Ser and Thr residues replaced but with Val substituting for Thr (S240A, T242V, and T243V), showed higher GTPase activity (46%) than mutant EF-6.
To investigate the role of the charged amino acid residues,

TABLE V GTPase activity of amino acid replacements in loop CD
The amino acid sequences of the loop region in wild-type rhodopsin and rhodopsin mutants are shown in Table 11. Results are presented as mean f S.D. averaged from two to three separate experiments as indicated by the numbers in parentheses. All the mutant activity values were normalized to COS cell rhodopsin controls assayed in parallel.   (2) 0.60 k 0.18 (2) 0.40 k 0.21 (2) 0.34 f 0.30 (3) 0.10 f 0.08 (2) 0.98 ? 0.04 (2) 0.72 f 0.09 (2) 0.61 f 0.20 (2) 1.10 ? 0.10 (2) 1.01 f 0.16 (2) 1.01 f 0.05 (2) 0.99 f 0.26 (2) 0.56 f 0.04 (2) 0.07 f 0.08 (2) wt, wild-type bovine rhodopsin amino acid sequence (1)(2)(3). value (500 nm), indicating that there was no influence of these amino acids on the spectral properties (18). Upon illumination, each pigment was completely converted to the MI1 form (380 nm). The capacity of these mutants to activate transducin is shown in Table V. a set of nine mutants was prepared as shown in Table 111. The mutations aimed at replacing the charged amino acids by neutral isosteric amino acids. Thus, mutant EF-1 contained two replacements, K231T and E232Q. This mutant showed nearly wild-type activity in the GTPase assay (87%) ( Table  VI). Mutant EF-2 (E239Q) ( Table 111) was previously reported to have a normal UV-visible spectrum and to display wildtype transducin activation in digitonin (15). This mutant was re-examined after purification in parallel in digitonin and dodecyl maltoside. In both detergents, mutant EF-2 showed a normal spectrum and transducin activation. The triple mutant EF-13 (E247Q/K248L/E249Q) also showed wild-type phenotype in both detergents.  was previously reported to be inactive in the GTPase assay in digitonin. By using [y-"PIGTP of higher specific activity than that used in earlier experiments (15) and filtered light rather than white light, a residual activity of -15% was detected in digitonin (Fig. 8). When assayed in dodecyl maltoside, mutant EF-8 displayed 72% of the wild-type activity.
Double and triple mutants were constructed to further investigate the role of the charged amino acids Glu-247, Lys-248, and Glu-249 (Table 111), which are conserved in most visual pigments.
Loop EF Insertion Mutant-The middle region of loop EF has the potential for an a-helical secondary structure. An insertion mutation, EF-14 (Table 111), was made that introduced 2 additional amino acids (Ser-Thr) after Ala-241 so as to extend the putative a-helix and to change its potential amphipathic character. The resulting mutant, EF-14, displayed -55% GTPase activity.

DISCUSSION
By using site-specific mutagenesis, we have investigated the structural requirements for the interaction between transducin and the cytoplasmic loops CD and EF of rhodopsin. Of the two classes of mutations studied, one comprised deletions of varying lengths in the loop segments. These mutations were designed to identify particular peptide sequences that were important for the rhodopsin-transducin interaction. The second group of mutations consisted of substitutions of polar or charged amino acids by neutral or hydrophobic residues. These replacements were designed to evaluate the contributions of electrostatic interactions or hydrogen bonding in the rhodopsin-transducin association. Transducin activation by the rhodopsin mutants was measured throughout by the GTPase assay.
Retinal Binding and Chromophore Formation by Mutant

Rhodopsin or Mutant (nM)
FIG. 8. Determination of light-dependent GTPase activity versus rhodopsin concentration for mutant EF-8 and COS cell rhodopsin prepared in digitonin. Assays were carried out with pigment concentrations ranging from 10 to 120 nM. The rate of P, release was determined from the time course for each pigment concentration shown. The rate of Pi release was plotted against the corresponding pigment concentration. Rhodopsin (0) was significantly more active than mutant EF-8 (0) at all pigment concentrations assayed.
Opsins-Bovine rod opsin as expressed in COS-1 cells binds 11-cis-retinal and forms the UV-visible spectrum characteristic of native bovine rhodopsin. Binding of the retinal and formation of the characteristic chromophore provide a sensitive assay for the formation of correctly folded rhodopsin. Retinal binding occurred in all but one of the mutants studied, including those with large deletions of 15 and 19 amino acids in loop EF, even though in the 19-amino acid deletion, the extent of regeneration of the chromophore was low (Fig. 9).
In the secondary structure model (Fig. l), loop EF contains 21 amino acids. The finding that deletion of 19 amino acids allows chromophore regeneration suggests that either helixes E and F are very close to each other in the tertiary structure or that the membrane boundaries shown in Fig. 1 for helixes E and F are incorrect.
A deletion of 8 amino acids in the cytoplasmic loop CD formed an opsin with normal retinal binding properties. However, a deletion of 14 amino acids in the same loop caused inability to bind retinal. Replacement of the deleted sequence by an unrelated amino acid sequence restored retinal binding and formation of the correct chromophore, indicating that there was no sequence specificity in the loop CD region for the formation of the retinal-binding pocket. Thus, in general, the deletion mutations in the cytoplasmic loops do not impair opsin folding and chromophore formation. This made possible the study of the cytoplasmic mutations now reported. In contrast, the mutations in the intradiscal domain generally affect retinal binding (21).
Solubilization in Digitonin and Dodecyl Maltoside-Integral membrane proteins differ greatly in their behavior toward different detergents. Digitonin has been commonly used in the past for solubilization of rhodopsin (22,23). In this work, the use of digitonin for the purification and characterization of rhodopsin and the mutants caused misinterpretations. As an example, mutant K248L, when assayed in digitonin for light-dependent transducin activation, was inactive (15). Reassay of this mutant in dodecyl maltoside showed 72% of the wild-type activity. GTPase activity assays in dodecyl maltoside, in general, gave 10 times higher activity than those in digitonin. Several possibilities could account for the large differences in activity between the two detergents. Since rhodopsin seems to be stable in both detergents, one possibility could be the use of white light in the earlier experiments (14, 15) with digitonin-solubilized samples. It has been reported (24) that constant illumination of rhodopsin with white light can cause the formation of photoisomers, thereby reducing the amount of MI1 in the pool of photoproducts. In fact,  (Table I). This deletion mutant had 19 amino acids removed from loop EF. According to the secondary structure model (Fig. l ) , only a 3-amino acid long linker would remain to connect helices E and F. The mutant bound 11-cis-retinal to give a Amax of 500 nm and an A280/A500 spectral ratio of 5.5. This low 11-cis-retinal regeneration could be due to structural constraints caused by the large deletion. Upon illumination, the pigment was converted to the characteristic MI1 form (380 nm) indistinguishable from that of wildtype rhodopsin. mutant K248L, which is inactive in digitonin under constant white light illumination, was partially active under light passed through a long-pass filter (Fig. 8). It is also possible that digitonin inhibits the formation of MII. As seen in Fig.  4, rhodopsin in digitonin formed a mixture of MI and MII. The ratio of MI to MI1 intermediates of rhodopsin is influenced dramatically by temperature. Increasing temperature favors the formation of MII. Under the same conditions, in dodecyl maltoside, rhodopsin is converted instantaneously to MI1 upon illumination. Finally, there is the possibility that digitonin inhibits rhodopsin-transducin interaction through an effect on rhodopsin or transducin or both. In summary, in our work, dodecyl maltoside has uniformly been superior to digitonin for rhodopsin purification and functional studies as judged by A280/A500 ratios in absorption spectra, and by GTPase assays.
Deletions in Loops CD and EF-Previously, Konig et al. (11) showed that synthetic peptides corresponding in sequence to the cytoplasmic rhodopsin loops competed for transducin binding to photolyzed rhodopsin. The results indicated that the cytoplasmic loops CD and EF were involved in interaction with transducin. The involvement of loop EF was also supported by an earlier study by Kuhn and Hargrave (9), where limited proteolysis of loop EF caused the loss of light-induced transducin binding. Our earlier (12,15) and present results support the involvement of loops CD and EF in transducin binding and activation. Thus, an 8-amino acid deletion mutant in loop CD (A143-150) failed to activate transducin, indicating that loop CD is essential for transducin activation. Furthermore, when 13 amino acids in loop CD were replaced by an unrelated sequence (CD-7) (Table 11), the mutant bound 11-cis-retinal to form the normal chromophore, but it failed to activate transducin. This result further demonstrates the requirement of a specific sequence in loop CD for interaction with transducin.
The requirement of loop EF was also clearly shown by our results. All of the loop EF deletion mutations now described affected the ability of these mutants to activate transducin in uitro. A 4-amino acid deletion in the amino-terminal part of the loop (A236-239) caused a 50% reduction of transducin activation. Furthermore, a 6-amino acid deletion in the carboxyl-terminal portion of loop EF (A244-249) caused a complete inability to activate transducin. Similarly, all the other deletion mutations involving the carboxyl terminus of loop EF abolished transducin activation (Table IV).
Single and Multiple Amino Acid Substitutions in Loops CD and EF-A relatively large number of mutants with amino acid substitutions in loop EF were prepared to probe the role of polar and charged amino acids. There is a cluster of 2 threonine residues and 1 serine residue in the center of this loop. These residues are involved in light-inducedphosphorylation (25). The role of these residues was investigated by carrying out a series of single and multiple amino acid substitutions. The GTPase activity results obtained with these mutants showed that Ser-240, Thr-243, and Lys-248 were important for transducin activation (Tables I11 and VI). Mutations that neutralized negatively charged amino acids (Table 111) had no influence on spectral properties and no significant effect on transducin activation (Tables I11 and VI). The above results show a direct involvement of loop EF in transducin activation and are in agreement with peptide competition studies (11) in which a peptide corresponding to loop EF could compete with rhodopsin for transducin binding and with a previous study of rhodopsin loop mutants (15).
Other members of the seven-helical receptor family have also been shown to have an active involvement of loop EF in  (Tables IV and V). This experiment tested whether activity could be recovered by increasing the concentration of the pigment in the assay mixture. When assayed at a pigment concentration of 10 nM, no detectable transducin activation was observed for either of the mutants. The Pi release observed under these conditions is identical to the intrinsic GTPase activity of transducin observed in the absence of pigment. The activity of wild-type rhodopsin is 12.5 pmol of Pi released per pmol of rhodopsin/min at this concentration (10 nM). This pigment concentration for wild-type rhodopsin is beyond the linear range of the assay (0.2-5 nM). A pigment concentration of 20 nM is saturating (Fig. 5). Even at a pigment concentration of 30 nM, no transducin activation was observed for either of the mutants.
G protein coupling. Kobilka et al. (26) suggested that with chimeric aP-and &adrenergic receptors, the specificity for coupling to the G protein is within loop EF. Kubo et al. (27) concluded that in the muscarinic acetylcholine receptor, the selective coupling of receptor subtypes I and I1 with different effector systems is due to the loop EF region in these receptors. Again, Wong et al. (28) indicated with chimeric muscarinic cholinergic/P-adrenergic receptors that the third cytoplasmic loop determines G protein specificity of the ligandactivated receptor. The charged pair Glu-134/Arg-135 located at the cytoplasmic border of helix C is found in nearly every G proteincoupled receptor characterized to date. We investigated the role of this charged pair with a series of five mutants (Table  11). All five mutants displayed normal UV-visible spectra. The amino acid substitutions had a drastic influence on rhodopsintransducin interaction (Table  V). Arg-135 is clearly very important for transducin interaction since substitution R135Q decreased transducin activation by more than an order of magnitude. Charge reversal and exchange of both residues with Ala caused complete inactivation of the resulting pigments in the transducin assay. In the case of the charge reversal mutant, we could not detect stimulation of the transducin GTPase activity even when the pigment concentration was increased 4-and 12-fold over the standard assay concentration (Fig. 10). We found previously that the charged pair plays an essential role in MII-transducin interactions that lead to binding of transducin (12).