The effect of carboxyl-terminal mutagenesis of Gt alpha on rhodopsin and guanine nucleotide binding.

The carboxyl terminus of G protein alpha subunits plays an important role in receptor recognition. To identify the amino acids that participate in this interaction, COOH-terminal mutants of alpha t (the transducin alpha subunit) were expressed in vitro and analyzed for their ability to interact with rhodopsin and to bind guanine nucleotide. Gly-348, the reported site of a beta turn, was replaced with other neutral amino acids without severely affecting rhodopsin binding. However, proline substitution abolished rhodopsin interaction, suggesting that flexibility is important at this site. A comparison between C347Y, which lost both rhodopsin and guanine nucleotide binding, and a mutant substituted with alpha q sequence (D346E/C347Y/G348N/F350V), in which guanine nucleotide binding was restored, implies that distinct motifs maintain the structure of the alpha subunit and are necessary for selective interaction with receptors. Surprisingly, mutants L344A, L349A, F350stop, and stop351A demonstrated a parallel loss of rhodopsin and guanine nucleotide binding. Altered profiles of L344A and F350stop on sucrose density gradients indicate that these mutants may undergo denaturation. The equivalent of alpha tL344A generated in alpha s and alpha i did not show such a severe loss of guanine nucleotide binding, revealing that the alpha t carboxyl terminus is unique in its susceptibility to changes in amino acid sequence.

Transducin (G t ), 1 a heterotrimeric G protein activated by rhodopsin in response to light, regulates visual signal transduction pathways in the vertebrate rod cell. Because of their relative abundance, the G t ␣, ␤, and ␥ subunits and rhodopsin have been used extensively as structural models for the study of receptor-G protein interaction. Several investigations have indicated that the COOH terminus of the ␣ subunit plays an important role in receptor recognition. For example, the interaction of ␣ t with rhodopsin can be disrupted by ADP-ribosylation of the COOH-terminal cysteine (Cys-347) with pertussis toxin (1). A synthetic peptide corresponding to the COOH-terminal 11 amino acids of the ␣ subunit can also block the binding of ␣ t to rhodopsin. When added alone, this peptide can stabilize the active form of rhodopsin, metarhodopsin II, a property of the G protein heterotrimer (2). NMR studies have demonstrated that this peptide exhibits dramatic conformational changes when incubated with light-activated rhodopsin, strongly suggesting that the COOH terminus of ␣ t binds to rhodopsin and undergoes conformational changes necessary for the promotion of guanine nucleotide exchange (3). A number of studies (for review, see Ref. 4) have verified the participation of the ␣ subunit COOH terminus in receptor coupling for other G proteins (5)(6)(7)(8)(9)(10)(11)(12). Despite the extensive literature on this subject, the mechanism by which these amino acids interact with G protein-coupled receptors is not well understood, although it has been proposed that a conserved glycine in the ␣ i/o/t family plays a central role in regulating the conformation of the COOH terminus (3,11).
In this report, we undertake a mutational strategy to define the COOH-terminal amino acids of ␣ t that participate in receptor interaction. Efforts to express the full-length ␣ t polypeptide have so far been unsuccessful using either bacterial or baculovirus expression, although recently an amino-terminal truncation of ␣ t has been expressed (13). We have chosen the method of in vitro translation, which has allowed us to generate fulllength, radiolabeled ␣ t for rhodopsin interaction and guanine nucleotide binding studies. This system has been used successfully by several laboratories to investigate both the receptor and effector coupling properties of ␣ s (14,15) and guanine nucleotide binding properties of ␣ o (16). Our studies have revealed the selective importance of specific amino acids in the interaction between the ␣ t and rhodopsin. We have also observed that certain mutations in the COOH terminus destroy both rhodopsin binding and guanine nucleotide binding, perhaps by significantly altering their tertiary structure. Equivalent mutants in ␣ i and ␣ s were not as severely affected, suggesting that the ␣ t COOH terminus is unique in its structural properties compared to other G protein ␣ subunits.

EXPERIMENTAL PROCEDURES
Isolation of Rod Outer Segments-Urea-stripped rod outer segments (ROS) containing rhodopsin were isolated in the dark from frozen, dark-adapted retina (W. H. Lawson, Lincoln, NE) on 25/30% (w/w) sucrose gradients as described (17,18). The concentration of functional rhodopsin, quantified by absorbance at 500 nm and using an extinction coefficient of 42,700 M Ϫ1 cm Ϫ1 (19), was 4 g/l. The purity was approximately 90%, estimated by Coomassie Blue staining on SDS-polyacrylamide gels. ROS membranes were analyzed for the presence of ␣ t and ␤␥ by immunoblot analysis (20). The membranes were chromatographed on 10% polyacrylamide gels, transferred to nitrocellulose, and incubated with AS/7, an anti-COOH-terminal antibody that recognizes both ␣ t and ␣ i (DuPont NEN), and with an antibody that recognizes G t ␤␥ (a gift from Dr. Gary L. Johnson, National Jewish Center for Immunology and Respiratory Medicine, Denver, CO). The binding of the antibodies was visualized by incubation with 125 I-protein A and autoradiography.
Mutagenesis of ␣ t -The EcoRI-PstI fragment from the ␣ t cDNA (a gift from Dr. Henry Bourne, University of California at San Francisco) (21) was inserted into pSelect (Promega) at the HindIII site by the addition of HindIII linkers. Site-directed mutagenesis was performed using the "Altered Sites" mutagenesis system according to the manufacturer's protocols (Promega).
Preparation of in Vitro Translated Subunits-The double-stranded pSelect-␣ t and mutant constructs were linearized with EcoRV and incubated with T7 polymerase to synthesize sense RNA (22). The RNAs were incubated at 30°C for 1 h with rabbit reticulocyte lysate (Promega) in the presence of 2 l of [ 35 S]methionine (1200 Ci/mmol; Amersham Life Sciences) in a final volume of 25 l according to the manufacturer's directions. The radioactivity incorporated into protein was measured as the insoluble fraction from a hot trichloroacetic acid precipitation (23). The mutant and wild-type ␣ s and ␣ i2 cDNAs (24) were inserted into the HindIII site of pSP72 (Promega). All procedures for in vitro transcription and translation were carried out as described for ␣ t .
Rhodopsin Binding Assay-The in vitro translation products were diluted 2-fold and centrifuged through Bio-Spin P-6 columns (Bio-Rad) to remove salts and free nucleotides. The eluates were equilibrated in buffer A (20 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM MgCl 2 , and 1 mM dithiothreitol). Urea-stripped ROS membranes containing 7 g (0.18 nmol) of rhodopsin were mixed with approximately 50,000 cpm (0.02 pmol) of in vitro translated mutant or wild-type ␣ t in buffer A and incubated for 30 min at 4°C. Experiments were performed in the light or the dark and in the presence or absence of 200 M GTP␥S. To terminate the assay, the reaction mixtures were centrifuged in a TLA45 rotor (Optima TLX centrifuge, Beckman) at 45,000 rpm through a cushion of 0.2 M sucrose in buffer A. After rinsing with buffer A, the pellets were resuspended in SDS-Laemmli sample buffer (25) and chromatographed on 10% SDS-polyacrylamide gels. The gels were treated with Amplify (Amersham) and autoradiographed. The amount of mutant or wild-type ␣ t bound to ROS membranes was quantified by phosphorimage analysis (Molecular Dynamics). The specific binding of ␣ t to rhodopsin was calculated as the amount bound in the absence of GTP␥S minus the amount bound in its presence. For each construct, an aliquot of the eluate from the Bio-Spin P-6 columns was chromatographed on SDS-polyacrylamide gels and quantified by phosphorimage analysis to determine the amount of radioactivity specifically incorporated into the ␣ t band. These values were used to correct for differences in expression of the various in vitro translated ␣ t subunits in the rhodopsin binding assay. Under these conditions, approximately 15% of the added wildtype ␣ t binds to ROS membranes.
Trypsin Resistance Assay-This assay was used to measure GTP␥S binding to the G protein ␣ subunits. Approximately 30,000 cpm (0.012 pmol) of the in vitro translation reaction were diluted into 25 l of buffer B (50 mM Tris-HCl, pH 7.5, 6 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol) and incubated for 2 h at 30°C with 100 M GTP␥S. The trypsin resistance assay was initiated by incubation with 0.3-0.6 g of trypsin (sequencing grade modified trypsin, Promega) for 30 min at 30°C. The reaction was terminated by the addition of SDS-Laemmli sample buffer (25). The samples were boiled and chromatographed on 10% SDS-polyacrylamide gels. The gels were treated with Amplify and analyzed as described above. For each construct, an aliquot of the in vitro translation product was chromatographed on SDS-polyacrylamide gels and quantified by phosphorimage analysis to correct for differences in expression levels as described above.
Sucrose Density Gradient Sedimentation-In vitro translated products of ␣ t and ␣ t mutants (80,000 cpm, 0.032 pmol) were mixed with 100 l of buffer A and overlaid on a 5-20% linear sucrose gradient in buffer A (16). After centrifugation at 4°C for 16 h at 55,000 rpm in an SW55Ti rotor, the gradients were fractionated into 250-l aliquots per tube and chromatographed on SDS-polyacrylamide gels followed by phosphorimage analysis. The two peak fractions from each sample were selected and reapplied to SDS-polyacrylamide gels, followed again by phosphorimage analysis for estimation of the sizes of the different peaks. Fig. 1 shows the COOH-terminal 11 amino acids of ␣ t and 5 other members of the G protein ␣ subunit family. Two of these amino acids, Leu-344 and Leu-349 in ␣ t , are conserved among all G protein ␣ subunits. A positively charged amino acid (Lys-345 in ␣ t ) and a hydrophobic amino acid at the COOH terminus is conserved in most ␣ subunits. Gly-348, previously shown in ␣ t to be critical for the interaction of a synthetic peptide corre-sponding to this sequence with rhodopsin (3), is preserved in ␣ i and ␣ o but is replaced by other amino acids outside the ␣ i/o/t family.

RESULTS
A binding assay for measuring the interaction of ␣ t with rhodopsin-containing membranes was developed using 35 S-labeled, in vitro translated protein. This assay is based on the formation of a stable complex between rhodopsin and G t in the absence of guanine nucleotide (26 -28). Fig. 2A shows the binding of in vitro translated ␣ t and ␣ s to bovine ROS membranes. The binding of ␣ t to rhodopsin in the light is approximately twice the level of binding in the dark. The addition of GTP␥S, a nonhydrolyzable GTP analog, releases ␣ t from the membranes, reflecting the promotion of guanine nucleotide exchange by rhodopsin in this system. In contrast, ␣ s , which does not interact with rhodopsin, showed no significant increase in binding in the light compared to samples kept in the dark and was not released by incubation with GTP␥S. The observed GTP␥S-dependent release of ␣ t from ROS membranes incubated in the dark suggests that some of the rhodopsin in these preparations is already bleached. For this reason, GTP␥S-dependent binding (the difference between the binding of ␣ t to photolyzed rhodopsin in the presence and absence of GTP␥S) rather than light-dependent binding (the difference between ␣ t binding to rhodopsin in the light and in the dark) was used in all subsequent experiments as a measure of specific interaction with rhodopsin. It has been established by several laboratories that ␣ t requires the presence of ␤␥ to form a stable complex with rhodopsin (29 -31), suggesting that ␤␥ is present in our assay system. Immunoblots (Fig. 2B) performed with subunitspecific antibodies (32) confirmed the presence of low levels of ␣ t , ␤, and ␥ (data not shown) in these ROS preparations. Therefore, the in vitro translated ␣ t and ROS membranes were incubated together without further addition of ␤␥ to the rhodopsin binding assay.
The COOH-terminal seven amino acids of ␣ t were mutated individually to alanines, expressed by in vitro translation, and assayed for rhodopsin binding. Fig. 3A shows an autoradiogram of the rhodopsin binding assay. The results were quantified by phosphorimage analysis and are shown in Fig. 3B. Mutants L344A and L349A demonstrated the most severe (96 and 93%, respectively) loss of binding. C347A, the site of ADPribosylation by pertussis toxin (33), G348A, the site of a proposed ␤ turn in the COOH terminus (3), and F350A were moderately affected, demonstrating a 32, 36, and 30% loss of binding to rhodopsin, respectively. All of the alanine mutants that bound to rhodopsin (K345A, D346A, C347A, G348A, and F350A) were released by the addition of GTP␥S, indicating that they are able to undergo receptor-mediated guanine nucleotide exchange. The sensitivity of the two leucines to disruption by alanine mutagenesis raised questions concerning the structural integrity of these mutants. G protein-coupled receptors function as ligand exchange enzymes, meaning that they promote the ex-change of a ligand, in this case guanine nucleotide bound to the G protein ␣ subunit (34). Basal guanine nucleotide exchange occurs in ␣ t at a slow but measurable rate that is enhanced approximately 70-fold in the presence of photolyzed rhodopsin (32). The loss of basal GTP␥S binding in the ␣ t mutants would suggest more extensive changes in the overall structure of the subunit. The binding of guanine nucleotide can be measured by examining the resistance of ␣ t to digestion by trypsin in the presence of GTP␥S (35)(36)(37). Fig. 4A demonstrates that in vitro translated ␣ t is digested almost entirely by high concentrations of trypsin in the absence of guanine nucleotide. In contrast, a 32-kDa band is protected from digestion in the presence of GTP␥S. A 23-kDa band observed in the trypsin-digested samples is due to protection by GDP present in the in vitro translation mixture (16,37). The amount of the trypsin-resistant 32-kDa band increased to 40% of the total in vitro translated ␣ t  (25) and analyzed by SDS-PAGE as described under "Experimental Procedures." In addition to the full-length 39-kDa (wild-type ␣ t ) band, other bands appear in the absence of trypsin and GTP␥S that represent translation initiation at methionines downstream from the primary initiation codon. A 38-kDa band, appearing at lower concentrations of trypsin, is a partial digest of ␣ t , described previously (37). The 32-and 23-kDa bands are GTP␥Sand GDPprotected fragments of ␣ t , respectively. Numbers on the left side of the gel indicate approximate molecular weights of prestained molecular markers (Bio-Rad). B, time course of GTP␥S binding to ␣ t using the trypsin resistance assay. [ 35 S]Methionine-labeled ␣ t was incubated with 100 M GTP␥S for the indicated duration, followed by incubation with 0.5 g of trypsin for 30 min. The samples were subjected to SDS-PAGE followed by phosphorimage analysis. The results are presented as a percent of trypsin-resistant 32-kDa band compared to undigested ␣ t . during 8 h of incubation with GTP␥S (Fig. 4B). In contrast, more than 80% of ␣ s and ␣ i is protected after only a 2-h incubation in GTP␥S (data not shown), suggesting that these ␣ subunits undergo basal guanine nucleotide exchange at a much higher rate than ␣ t in our assay mixture, which contains approximately 6 mM Mg 2ϩ ion.
The COOH-terminal alanine mutants were tested for their ability to resist digestion by trypsin in the presence of GTP␥S (Fig. 5). C347A and F350A showed a partial decrease in trypsin resistance of 56 and 41%, respectively. The most dramatically affected mutants were L344A and L349A. Very little of the protected 32-kDa band was detected in these samples. Therefore, these mutants appear to be unable to bind GTP␥S. Neither L344A nor L349A showed significant amounts of the 23-kDa protected fragment, indicating that they may not bind GDP. Incubation with very high concentrations (2 mM) of either GTP or GDP also did not protect these mutants from trypsin digestion, although both the 32-and the 23-kDa fragments were apparent in samples of wild-type ␣ t (data not shown). Therefore, these mutants are incapable of binding any form of guanine nucleotide. These results demonstrate a parallel loss of both rhodopsin binding and guanine nucleotide binding for L344A and L349A.
To investigate in more detail the influence of mutation on the properties of ␣ t , the seven COOH-terminal residues were replaced with different amino acids and assayed for both rhodopsin binding and trypsin resistance (Fig. 6). When isoleucine, a hydrophobic amino acid, was introduced at Leu-344 and Leu-349, the resulting proteins demonstrated trypsin resistance at 69 and 50% of wild-type levels, respectively, and rhodopsin binding at 43 and 27% of wild-type levels, respectively, suggesting that these sites are more critical for rhodopsin binding than for guanine nucleotide binding. Interestingly, mutation to phenylalanine at these positions resulted in significantly reduced activity in both assays, particularly for L344F, perhaps due to steric hindrance caused by the larger size of this amino acid. Rhodopsin binding and guanine nucleotide binding are both abolished by the substitution of proline for Lys-345 or Asp-346. These data suggest that fixing the angle of the COOH terminus with a proline at residue 345 or 346 disrupts both activities of the protein. In contrast, the introduction of a proline at Gly-348 dramatically inhibited rhodopsin interaction but caused only a 39% decrease in intrinsic guanine nucleotide binding. Mutation of Gly-348 to asparagine (G348N) or leucine (G348L) caused significant (53 and 57%, respectively) losses in rhodopsin binding but only small (18 and 27%, respectively) losses in guanine nucleotide binding. Mutation of Cys-347, also conserved in the ␣ i/o/t family, to alanine, caused a partial decrease in both rhodopsin binding (to 68%) and guanine nucleotide binding (to 44%). In contrast, both receptor binding and guanine nucleotide binding were lost when this residue was mutated to tyrosine, the amino acid found at the equivalent position in both ␣ s and ␣ q . The COOH-terminal residue, Phe-350, is tolerant to substitutions, retaining activity when changed to alanine (weakly hydrophobic) and glutamine (hydrophilic), and exhibiting significantly reduced activity (a 72 and a 65% loss in rhodopsin binding and trypsin resistance, respectively) only when mutated to glutamic acid, a negatively charged amino acid.
Additional point mutations and deletions at the amino and COOH termini of ␣ t were compared for rhodopsin and guanine nucleotide binding activities (Fig. 7). The double mutation K345A/D346A, in which two charged residues are eliminated, and a mutation equivalent to the ␣ o sequence, D346G/F350Y, bound guanine nucleotide and interacted with rhodopsin as well as wild-type ␣ t . In contrast, the mutant D346E/C347Y/ G348N/F350V, which represents the COOH-terminal sequence of ␣ q , could not bind rhodopsin but bound guanine nucleotide normally. Interestingly, removal of a single amino acid from the COOH terminus (F350stop) was sufficient to destroy both rhodopsin and guanine nucleotide binding, despite the tolerance of Phe-350 to point mutagenesis shown in Fig. 6. Extension of the COOH terminus by a single alanine (stop351A) also caused a complete loss of both rhodopsin and guanine nucleotide binding. Mutations at the amino terminus, such as G2A (in which the myristoylation site is removed), and the deletions delN6/A7M and delN10/H11M resulted in a loss of receptor binding (data not shown) but no loss of guanine nucleotide binding, suggesting that the loss of guanine nucleotide binding and rhodopsin interaction simultaneously is specific to the COOH terminus.
This parallel loss of rhodopsin and guanine nucleotide exhibited by the COOH-terminal mutants could be due to denaturation of the polypeptides. To investigate this possibility, the wild-type protein and mutants L344A, G348A, and F350stop were analyzed by sucrose density gradient sedimentation (Fig. 8).
The peak fractions of all of these proteins were at the same position in the gradient. Similar amounts of G348A were recovered from the gradients compared to wild-type ␣ t . However, substantially less L344A and F350stop were recovered, despite equal loading of these proteins on the gradients. No peaks were observed at any other position on the sucrose gradients, which could account for the lost radioactivity. A similar result was observed by Denker et al. (16) with a 14-amino acid COOHterminal truncation of ␣ o , which, when stripped of guanine nucleotide, formed no discrete peaks on sucrose density gradients. These data imply that L344A and F350stop are less stable than wild-type ␣ t and are likely to be denatured in our assays.
There have been no reports in the literature of such sensitivity to mutagenesis in the COOH terminus of other G proteins, raising the prospect that our observations are specific for ␣ t . To test this possibility, the equivalent of the mutant ␣ t L344A was also generated in ␣ s (L388A) and ␣ i2 (L349A). Neither ␣ s L388A nor ␣ i2 L349A demonstrated a severe decrease in GTP␥S-dependent trypsin resistance compared to their respective wild-type proteins, indicating that these mutants are better able to bind guanine nucleotide than the equivalent ␣ t mutant (Figs. 9 and 10). Because ␣ i is closely related in sequence to ␣ t and can be activated by rhodopsin (38), the ability of ␣ i L349A to bind rhodopsin was measured (Fig. 10). This mutant showed only a modest decrease in rhodopsin binding (35%) compared to wild-type ␣ i . Therefore, the extreme sensitivity of the ␣ t COOH terminus to mutagenesis does not appear to be shared with other members of the G protein ␣ subunit family. DISCUSSION A number of laboratories have suggested that information for receptor interaction resides in the COOH terminus of G protein ␣ subunits (1-3, 5-12). Using NMR, Dratz et al. (3) described the structure of a synthetic peptide corresponding to the COOH-terminal 11 amino acids of ␣ t (amino acids 340 -350) bound to rhodopsin. Gly-348 was shown to be essential to the formation of a ␤ turn, which was necessary for the binding of the COOH-terminal peptide to rhodopsin. In these studies, substitution of this glycine with leucine abolished the binding of the synthetic peptide. In our experiments, replacement of Gly-348 with alanine, leucine, or asparagine partially reduced rhodopsin binding, whereas substitution with proline abolished this interaction. All of these mutations had a less severe effect on guanine nucleotide binding, suggesting that this residue is critical for the interaction of ␣ t with rhodopsin. However, the effect of mutation of Gly-348 on rhodopsin binding was less severe than would be predicted from the synthetic peptide studies described above. Perhaps other regions in the fulllength protein also interact with rhodopsin and partially compensate for the loss of this COOH-terminal residue. For example, residues 311-329 have been proposed to interact with rhodopsin and to participate in receptor-catalyzed guanine nu-FIG. 6. Rhodopsin binding and trypsin resistance for additional ␣ t COOHterminal point mutants. The seven COOH-terminal amino acid residues of ␣ t were subjected to further mutagenesis. These mutants were expressed by in vitro translation, radiolabeled with [ 35 S]methionine, and used for rhodopsin binding and trypsin resistance assays as described under "Experimental Procedures" and the legends to Figs FIG. 7. Rhodopsin binding and trypsin resistance of aminoand COOH-terminal mutants of ␣ t . A, amino-and COOH-terminal mutants of ␣ t were assayed for their ability to bind rhodopsin and for trypsin resistance in the presence of GTP␥S, as described in the legends to Figs. 3 and 5 and under "Experimental Procedures." Group I, rhodopsin binding was determined from duplicate samples; group II, rhodopsin binding was determined from a single sample. *, trypsin resistance was determined from four samples for D346E/C347Y/G348N/ F350V. All error bars represent the range of duplicate samples except D346E/C347Y/G348N/F350V, which is shown as S.E. The groups I, II, and III are described in Fig. 7B. B, sequences of the mutations used for rhodopsin binding and trypsin resistance assays shown in A.
cleotide exchange (2,39). Dratz et al. (3) also reported that the ␤ turn in the rhodopsin-bound conformation of the peptide disappeared when rhodopsin was converted to metarhodopsin II by exposure to light. From this result, a requirement for flexibility at Gly-348 for binding to light-exposed rhodopsin may be inferred, which is consistent with our observation that proline is more effective at disrupting rhodopsin binding than other amino acids at this position. Proline substitutions at the neighboring residues, Lys-345 and Asp-346, abolished both rhodopsin binding and guanine nucleotide binding. Therefore, these two residues are more critical for guanine nucleotide binding than is Gly-348.
Conklin et al. (11) reported that replacement of the last three amino acids of ␣ q with the sequence found in ␣ i (amino acids NLV to GLF) was sufficient to allow the activation of this chimera by the D 2 dopamine receptor, a G i -coupled receptor. Therefore, not only is the COOH terminus responsible for binding to receptor, it also contains information that determines the selectivity between G proteins and their receptors. In our experiments, C347Y, a substitution in ␣ t with the ␣ q amino acid, exhibited a dramatic loss of both rhodopsin and guanine nucleotide binding. Substitution of this amino acid along with other amino acids corresponding to ␣ q sequence (D346E/C347Y/ G348N/F350V) resulted in a rescue of guanine nucleotide binding but no restoration of rhodopsin binding. These data indicate that the tertiary structure of the surrounding domains may be drastically affected by a single mutation but is preserved when the sequence of another ␣ subunit representing a complete structural motif is introduced. Understanding the role of C347Y and its surrounding structural motif in receptor interaction will require further study.
The removal or addition of a single amino acid at Phe-350 but not its substitution resulted in a loss of both receptor coupling and guanine nucleotide binding, suggesting that the length of the COOH terminus is critical for the function of ␣ t . A dramatic loss of both guanine nucleotide binding and rhodopsin binding was observed for the COOH-terminal mutants L344A and L349A as well. Mutations in Lys-345 and Asp-346 (to proline), Cys-347 (to tyrosine), and Phe-350 (to glutamic acid) also exhibited a parallel loss, either partial or complete, of both rhodopsin and guanine nucleotide binding. In contrast, the aminoterminal mutants (G2A, delN6/A7M, and delN10/H11M) showed only a loss of rhodopsin binding. This unexpected sensitivity of the COOH terminus for changes in both activities was shown to be specific for ␣ t . For example, mutations equivalent to L344A in ␣ s (L388A) and ␣ i (L349A) do not exhibit a large decrease in guanine nucleotide binding. Moreover, ␣ i L349A showed only a small decrease in rhodopsin binding compared to the wild-type protein. In ␣ o , removal of 5, 10, or even 14 amino acids from the COOH terminus was reported to have no effect on the binding of GTP␥S, although the 14-amino acid deletion mutant lost affinity for GDP (16). Therefore, ␣ s , ␣ i , and ␣ o are clearly less susceptible than ␣ t to alteration of the COOH terminus. The reduced recovery of L344A and F350stop from sucrose density gradients suggest that the ␣ t COOHterminal mutants are unstable and susceptible to denaturation, which may account for the loss of both rhodopsin and guanine nucleotide binding. FIG. 8. Sedimentation profiles of ␣ t and ␣ t mutants on sucrose density gradients. 80,000 cpm of in vitro translated products were overlaid on 5-20% linear sucrose gradients and centrifuged for 16 h at 55,000 rpm in an SW55Ti rotor at 4°C. After centrifugation, the gradients were fractionated into 250-l aliquots per tube and chromatographed by SDS-PAGE followed by phosphorimage analysis as described under "Experimental Procedures." The data are representative of three independent experiments.
FIG. 9. Trypsin resistance of ␣ s and ␣ i mutants. Mutants at the equivalent position with ␣ t L344A in ␣ s (L388A) and ␣ i (L349A) were assayed for resistance to trypsin in the presence of GTP␥S. The ␣ s L388A and ␣ i L349A mutants produced trypsin-resistant fragments of 36 and 37 kDa, respectively. Several bands that appear below the major in vitro translation product in the lanes without trypsin are peptides translated from ATG sites downstream from the ATG initiation site. The ␣ t subunit is known to possess several structural and functional properties that are distinct from other closely related G protein ␣ subunits. Unlike ␣ i or ␣ o (40,41), the rate of receptor-independent guanine nucleotide exchange in ␣ t is not affected by magnesium ion (32). X-ray crystallographic analysis also shows differences between ␣ t and ␣ i at the sites of contact with guanine nucleotide (42). Other differences are apparent in the amino terminus. When counted from the common ␤1 strand, ␣ t has the shortest amino terminus and is not modified by palmitic acid like other ␣ subunits of this family (␣ i , ␣ o ). In addition, ␣ t is heterogeneously acylated with myristate or one of three other fatty acids (43,44), whereas ␣ i is acylated only with myristic acid. This heterogeneous modification is so far unique to retinal photoreceptor cells (45). Therefore, it is presumed that in vitro translated ␣ t (synthesized in a rabbit reticulocyte lysate) is predominantly myristoylated and similar in modification to ␣ i , ruling out differences in myristoylation, at least, as an explanation for our observations. Many laboratories using a variety of techniques have concluded that the amino and COOH termini are closely associated, allowing for coordinate regulation of ␤␥ and receptor binding, as well as control of guanine nucleotide binding (37, 46 -48). At present, the possibility that the unique properties of the ␣ t COOH terminus are due to association with a unique amino terminus is speculative because the extreme amino termini of ␣ t (amino acids 1-26) and ␣ i (amino acids 1-32) have not been resolved by x-ray crystallography. However, the orientation of the adjacent regions is consistent with their close approximation (39,42,49,50). Future studies in which the amino terminus from ␣ i is substituted for that of ␣ t in the L344A, L349A, and F350stop mutants, for example, will help to clarify the relationship between these two domains and their role in rhodopsin and guanine nucleotide binding.
The present report describes the use of mutagenesis to understand the role of the COOH-terminal amino acids of ␣ t in rhodopsin binding. Of the alanine mutants, L344A and L349A showed the most dramatic decrease in rhodopsin binding and guanine nucleotide binding. Alternative substitutions suggest a preference for hydrophobic amino acids at these positions. Mutations in Gly-348 demonstrated that this residue is more important for rhodopsin binding than for guanine nucleotide binding and that flexibility at this position is critical for the interaction of the COOH terminus with rhodopsin. A comparison between C347Y and a mutant substituted with the ␣ q sequence (including C347Y) implies that structural motifs exist in the COOH terminus that maintain the structure of the ␣ subunit and are necessary for selective interaction with Gprotein-coupled receptors. Mutations at nearly every position in the COOH terminus of ␣ t but not in the amino terminus caused an unexpected parallel loss of rhodopsin binding and guanine nucleotide binding. Since mutations equivalent to ␣ t L344A in ␣ s and ␣ i did not exhibit a dramatic decrease in these activities, we conclude that the ␣ t COOH terminus is uniquely susceptible to amino acid alterations compared to other closely related G-protein ␣ subunits, perhaps reflecting distinct biochemical properties necessary for function of this rod cell-specific G protein.