Hydrophobic amino acid in the i2 loop plays a key role in receptor-G protein coupling.

Signal transduction of the heptahelical G protein-coupled receptors (GPCRs) involves multiple receptor domains, but a universal consensus domain for coupling has not yet been defined. Alanine mutagenesis scanning was performed on the intracellular loops and the COOH tail of the human muscarinic cholinergic receptor (Hm1) to identify coupling domains. Stimulation of phosphatidylinositol (PI) turnover was determined after transfection of the alanine mutants into U293 human embryonic kidney cells. Alanine substitutions in four regions (loops i1, i2, and NH2 and COOH junctions of i3) impaired coupling efficiency by approximately 50% or more, but the strongest reduction (> 80%) resulted from alanine replacement of a single amino acid, leucine 131. This residue is located in the middle of the second intracellular loop (i2), within the highly conserved GPCR motif (DRYXXV(I)XXPL). The position equivalent to Leu-131 in Hm1 contains a bulky hydrophobic amino acid (L, I, V, M, or F) in nearly all cloned GPCRs. Substitution of Leu-131 with polar amino acids (aspartate and asparagine) also resulted in strongly defective coupling, whereas phenylalanine (found in the equivalent position in the beta 2 adrenoceptor) can replace leucine without losing PI coupling ability of Hm1. Alanine substitution of the corresponding amino acid in the Hm3 receptor (L174A) also inhibited agonist-stimulated PI turnover, while replacing Phe-139 with alanine in the beta 2 adrenoceptor suppressed stimulation of adenylyl cyclase. We propose that a bulky hydrophobic amino acid in the middle of the i2 loop serves as a general site relevant to G protein coupling, whereas coupling selectivity is governed by other receptor domains.


Signal transduction of the heptahelical G proteincoupled receptors (GPCRs) involves multiple receptor domains, but a universal consensus domain for coupling
has not yet been defined. Alanine mutagenesis scanning was performed on the intracellular loops and the COOH tail of the human muscarinic cholinergic receptor (Hml) to identify coupling domains. Stimulation of phosphatidylinositol (PI) turnover was determined after transfection of the alanine mutants into U293 human embryonic kidney cells. Alanine substitutions in four regions (loops il, i2, and N H 2 and COOH junctions of i3) impaired coupling efficiency by -6W' or more, but the strongest reduction (>80%) resulted from alanine replacement of a single amino acid, leucine 131. This residue is located in the middle of the second intracellular loop (i2), within the highly conserved GPCR motif (DRYXXV(1)XYPL). The position equivalent to Leu-131 in Hml contains a bulky hydrophobic amino acid (L, I, V, M, or F) in nearly all cloned GPCRs. Substitution of Leu-131 with polar amino acids (aspartate and asparagine) also resulted in strongly defective coupling, whereas phenylalanine (found in the equivalent position in the p2 adrenoceptor) can replace leucine without losing PI coupling ability of Hml. Alanine substitution of the corresponding amino acid in the Hm3 receptor (L174A) also inhibited agonist-stimulated PI turnover, while replacing Phe-139 with alanine in the p2 adrenoceptor suppressed stimulation of adenylyl cyclase. We propose that a bulky hydrophobic amino acid in the middle of the i2 loop serves as a general site relevant to G protein coupling, whereas coupling selectivity is governed by other receptor domains.
The H m l muscarinic cholinergic receptor is a member of the large family of G protein-coupled receptors (GPCRs).l The structure of GPCRs is thought to consist of seven transmembrane helices, with three intracellular loops (il-i3) and the cytoplasmic COOH tail (1). Studies with mutagenesis (2, 3), receptor-derived peptides (443, and receptor antibodies (7) have shown that multiple domains of the intracellular receptor portion contribute to functional G protein coupling. However, a * This work was supported by United States Public Health Service Grants GM43102 and MHO0996 from the National Institute ofGenera1 Medical Sciences, National Institutes of Health. The costs ofpublication 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.
universal consensus domain has yet to be identified, and suggested coupling sites show greatly divergent sequences such as the highly variable i3 loop of catechol and peptide neurotransmitter receptors. On the basis of previously proposed coupling domains, we introduced mutations into the i3 loop of H m l expected to uncouple the receptor from its G protein (81, but the ability of the mutant receptors to stimulate phosphatidylinosi-to1 (PI) turnover was barely affected. This result prompted us to undertake alanine mutagenesis scanning of the intracellular surface of Hml. Several domains were identified where alanine substitutions interfered with carbachol-stimulated coupling to PI turnover, and a conserved lipophilic residue in loop i2 (Leu-131 of Hml) was found to play a major role in receptor coupling.

EXPERIMENTAL PROCEDURES
Material~-[~H]NMS (specific activity of 80 Ci/mmol) and [3H]CGP12177 (specific activity of 46 Ci/mmol) were obtained from Amersham Corp. All other reagents were of analytical grade quality.
Construction of Vectors Expressing Hml, Hm3, and p2 Point Mutants -The construction of wild-type Hml and Hm3 in vector pSG5 was described previously (9-11). The hamster p2 adrenoceptor, in pCD-NAIneo and in Bluescript SK+, was obtained from C. D. Strader (12). Mutagenesis was performed with the Bluescript vector before transferring the mutated p2 gene, cut with XbaI, back to pCDNAIneo. All point mutations were introduced using the "unique site elimination" method (TransformerTM site-directed mutagenesis kit, Clontech) (13). All mutants were analyzed by restriction mapping and sequencing of the mutated region before use.
Dansfection of Human Embryonic Kidney Cells fU293)"The cells were transfected with the use of the calcium precipitation method as described before (9,lO). Transient expression yields were -900 fmoYmg protein for Hml, whereas some Hml mutants yielded considerably lower expression. Maximal stimulation of PI turnover by 1 m M carbachol appears to be depressed if the expression level drops below 250 fmoYmg protein; therefore, receptor expression was always measured in parallel with PI turnover. Experiments that yielded less than 250 fmoYmg protein were excluded from the data analysis. Expression of Hm3 and 62 wild-type and their respective mutant receptors gave comparable and very high yields (see "Results").
Receptor Binding Assays-Receptor expression of Hml, Hm3 and their respective mutants was determined in intact cell monolayers as described previously (9, 10) using 2 n M f3H1NMS. Binding competition curves with carbachol, to determine the carbachol IC5,, binding value, were also camed out in intact cells using 0.1 n M L3H1NMS. Receptor expression of the p2 receptor and its mutant was determined in cell homogenates using 3 n M CGP12177 (14) as described. Competition curves with isoproterenol were obtained with 0.5 n M CGP12177.
PI nrnouer-PI turnover was measured after labeling the cells with [3Hlmyoinositol for 24 h (9-111. For the assay of inositol monophosphate, which accounts for most of the inositol phosphate 3H activities in the presence of 1 m M LiCl, six-well cell culture dishes (well diameter of 3.5 cm) were used. Results were expressed as percent of total 3H activity, and the percent values were compared between carbachol-treated and untreated cells.
C A M P Assay-Isoproterenol dose response for AMP production was measured in the absence of a phosphodiesterase inhibitor. Intact attached monolayer cells were rinsed once with 1 ml of serum-free Dulbecco's modified Eagle's medium supplemented with 25 m M HEPES. Serum-free medium was added to each well (final volume 300 pl), followed by 10 1. 11 of the appropriate concentration of isoproterenol. The cells were incubated at 37 "C for 15 min. The reaction was terminated by the addition of 100 p1 of 0.4 M HCI. The cells were then incubated at 95 "C for 5-10 min, scraped off the wells, transferred into 1.5-ml polypropylene microcentrifuge tubes, and centrifuged at 4 "C for 5 min. The supernatant was collected and assayed for CAMP content with a competitive protein binding assay kit (Amersham Corp.). Protein content was determined by the Bradford method using the U. S. Biochemical Corp. reagent; it ranged from 0.25 to 0.3 mg/well. The data were fitted to the equation E = (Emm x L)/(Emax + L), where E is the stimulation of 22273 CAMP production over control values and E,,, the maximum stimulation. The results were fitted with the Minim 1.2 program (g-ll), after deletion of the data at M isoproterenol, which may have resulted in lower CAMP levels by p2 stimulation of Gi (15).

RESULTS AND DISCUSSION
Receptor portions of H m l shown by deletion mutations not to participate in coupling (9, 10) were excluded from the alanine scanning analysis (Fig. 1, arrows). Additionally, mutations of the following residues of H m l have been previously shown by us not to affect coupling significantly: Glu-214, Thr-215, Glu-216, Lys-359, Lys-361, Ser-368, Thr-428 (Fig. 1, solid circles) (8,10). Of the remaining intracellular residues, both lipophilic and poladcharged amino acids were selected to cover receptor regions with possible involvement in coupling. Assuming that each coupling domain may entail several amino acids, this strategy was felt to cover most or all potential coupling sites. were selected to avoid any overlap with previous mutations. Point mutations previously shown not to affect significantly G protein coupling (8, 1 0 ) are shown as filled circles, whereas the largest deletions of the i3 loop that still yielded full coupling efficiency (9,10) are shown by the arrows.
Nine multiple alanine point mutants and one COOH tail truncation mutant were constructed (Fig. 1). Three out of 10 mutants (mutants 3, 9, and 10; Fig. 1) did not yield productive tracer binding; therefore, carbachol-stimulated PI turnover was assayed for only seven mutants. Mutant 5 was modestly deficient in coupling to PI turnover; lower receptor expression relative to the wild-type in each individual experiment may have contributed to this result (Table I). Mutants 6 (W209N 12llAiY212A) and 7 (E360M362A5366A) yielded tracer binding only slightly lower than the wild type but were defective in coupling (Table I). Cheung et al. (3) previously demonstrated that specific hydrophobic, but not hydrophilic residues, in the NHz-terminal junction of the i3 loop of the p2 adrenoceptor are involved in G protein coupling. The coupling-deficient mutant 6 also implicates lipophilic residues of the NHzterminal i3 junction in coupling of Hml. Similarly, mutant 7 involves changes in the COOH-terminal i3 loop junction, which has been implicated in coupling (2,3). These results support a functional coupling role for the COOH-and NH2-terminal junctions of the i3 loop. Furthermore, mutant 1 (K51AN52A554A) displayed impaired coupling, also implicating the i l loop in this process. The receptor expression yield in each of the reported cases in Table I was above 250 fmoVmg protein, which we consider a threshold for allowing near-maximal stimulation of .PI turnover by Hml, on the basis of numerous previous transfection experiments with H m l and its mutants (8)(9)(10)(11). However, in the absence of an accurate molecular model at atomic resolution, the interpretation of partially defective coupling of mutant receptors is difficult because the loop and COOH tail junctions must form a tightly packed structure that responds to agonist-induced activation.
Only mutant 4 (L131A/Y133A/K136A, located in the middle of the i2 loop) was profoundly deficient in mediating PI turnover (Table I). Since this mutant contains three alanine point mutations, we constructed the individual alanine point mutants. Unexpectedly, mutant L131A was again strongly defective, while receptor expression was similar to that of the wildtype. In contrast, Y133A and K136A were as efficient as the Statistically significant compared with wild-type, p 5 0,0001 (one factorial ANOVA and Fisher PLSD).
Receptor yields for l j~l 3 3 and mutants 4 and 5 were consistently severalfold lower than those of the Hml wild-type in individual experiments.
~~~ ~ wild-type receptor in stimulating PI turnover. These results suggest that Leu-131 represents a novel site with relevance to G protein coupling. Replacement of Tyr-133 with alanine lowered receptor expression significantly, indicating a possible role of this residue in internal folding. Indeed, only a single experiment yielded receptor expression above 250 fmoVmg protein (Table I), whereas the yield was below 150 fmovmg protein in additional experiments, preventing quantitative analysis of maximal PI turnover. Hence, impaired coupling but normal expression of L131A suggests that this residue interact directly with the G protein, rather than affecting internal folding.
A dose-response curve with carbachol shows the profound effect of the L131A substitution in H m l on PI turnover (Fig.  2 A ) . The binding affinity of the Hml mutant L131A to carbachol (ICso = 0.28 k 0.03 mM) was largely unchanged from that of the wild-type receptor (ICso = 0.34 2 0.03 mM), indicating that loss of coupling efficiency was not caused by a change in agonist affinity.
Sequence comparison among 70 cloned mammalian GPCRs (16) indicates that a lipophilic amino acid is well conserved a t  lipophilic amino acid). The i2 loop may extend by one or more residues the site corresponding to Leu-131 in Hml. A few selected sequences are shown in Table 11. This site is located at the 3' end of a highly conserved i2 loop motif with the following most common residues: DRYXXV(1)XXPL where X i s any amino acid. Substitution of aspartate with asparagine in the DRY motif of the muscarinic m l receptor resulted in reduced G protein COUpling (17). The residue immediately adjacent to DRY is somewhat conserved, consisting of lipophilic aromatic or aliphatic amino acids in many GPCR sequences. However, substitution of Phe-125 with alanine had no effect on H m l expression and coupling (Table I). In contrast, position Val-127 is very highly conserved and contains either valine or isoleucine in 67 out of the 70 mammalian GPCRs listed in Ref. 16. Alanine substitution of Val-127 in H m l indeed caused a significant loss in coupling efficiency, although the effect was smaller than for the L131A mutant (Table I). Mutant V127A also showed some impairment of overall receptor expression. These results indicate that position Val-127 contributes to t h e i l loop coupling domain either directly or indirectly, by affecting Hml/i2 folding. One amino acid 5' to Leu-131 is predominantly occupied by proline (in 54 out of 70 sequences) (161, which has been shown previously to play a small but measurable role in p2 adrenoceptor coupling (18). Substituting Pro-130 with alanine (Hml mutant P131A) did not affect carbachol-stimulated PI turnover relative to wild-type HM1 (Table I). Alanine residues are found in a few receptors at the position equivalent to Pro-130, such as the a2 adrenoceptor subtypes and the glycoprotein hormone receptors (161, suggesting that alanine substitution is permissive for G protein coupling. The Leu-131 site contains bulky lipophilic amino acids in 64 out of 70 mammalinan GPCR sequences (16); leucine occurs 27 times, while isoleucine (12 times), valine (4 times), methionine (11 times), and phenylalanine (10 times) are also found. We therefore hypothesized that a bulky hydrophobic amino acid is required in the well defined position Leu-131. Indeed, substitution of Leu-131 with phenylalanine (e.g. p2 adrenoceptor) yielded a mutant Hml receptor with productive PI coupling. Substitution with methionine to give L131M still supported measurable, but clearly reduced PI turnover, whereas point mutants with the polar amino acids aspartate or asparagine were highly deficient (Table I). In each case, overall receptor expression was similar to that of the wild-type receptor. Hence, a lipophilic amino acid in position 131 appears to be required for H m l coupling to PI turnover.
Since the muscarinic Hm3 receptor is closely related to the H m l receptor (19), the equivalent Leu-174 residue was replaced with alanine. Again, PI coupling was inhibited without strongly affecting receptor expression, measured with 2 nM L3H1NMS (Hm3 wild-type, 1974 2 110 fmoVmg protein; L174A, 1240 f 288 fmoVmg protein). A dose-response curve with carbachol shows the profound effect of alanine substitution in Hm3 Leu-174 (Fig. 2 B ) . Because of the low coupling efficiency of the mutants, EC50 values could not be determined.
As both Hml and Hm3 couple to G proteins that activate phospholipase C (19), we chose the p2 adrenoceptor receptor to test whether this site is also crucial for coupling to G,, which activates adenylyl cyclase. Introduction of the equivalent point mutation F139A caused a significant loss in isoproterenol-induced CAMP accumulation at an equivalent level of p2 receptor expression (wild-type, 4300 fmoVmg protein; F139A, 4100 fmoVmg protein). A dose-response curve (Fig. 2C) shows that maximum stimulation of CAMP accumulation is reduced by -75% and the EC50 is shifted from 1 n~ to 7 nM. In contrast, isoproterenol binding curves measured with [3H]CGP12177 indicated only a minimal change of affinity of the mutant receptor (wild-type ICso 3.3 * 0.3 p~; F139A, 5.3 * 0.8 VM). Therefore a nonpolar amino acid residue in the middle of the i2 loop, i.e. Phe-139, also plays an important role in signal transduction to adenylyl cyclase via G,.
Involvement of the i2 loop in G protein coupling was suggested previously, although a single amino acid had not been identified as a coupling site. A synthetic peptide corresponding to the entire i2 loop of the turkey erythrocyte p l adrenoceptor suppressed adenylyl cyclase activation by 90% (20). Furthermore, transducin prevented the rhodopsin binding of an antipeptide antibody generated against the i2 loop of rhodopsin (7). Substitution of the entire i2 loop with the the analogous region of t h e m l muscarinic receptor has been shown not to affect coupling to adenylyl cyclase of the p2 adrenoceptor (21). These results implicated the i2 domain as a coupling site for all G proteins; however, since the i2 loops of m l and p2 are interchangeable, coupling specificity appears to reside at other receptor domains, e.g. the i3 loop (22). These combined results support the view that receptor-G protein coupling involves multiple domains.
In conclusion, the i2 loop appears to be central to receptor folding, activation, and G protein coupling. The lipophilic amino acid in the middle of i2 (e.g. Leu-131 of m l , Leu-174 of m3, and Phe-139 of p2) in the consensus domain DRYXX-V(1)XXPL (L = leucine or other lipophilic amino acid) could represent a major coupling site. This site may interact with conserved residues of the COOH-terminal domain of G, subunits, which also contain conserved lipophilic amino acids (23). However, we cannot exclude the possibility that this hydrophobic residue is essential to the internal folding of the i2 loop itself to elicit productive coupling. Measuring second messenger activation does not distinguish between receptor-G protein binding and activation, whereas these processes can be readily differentiated for rhodopsin by optical measurement (24). Nevertheless, the results document the importance of the central portion of the i2 loop in G protein coupling within a motif that is highly conserved even among distantly related GPCRS. The crucial relevance of the lipophilic anchor residue in i2 to receptor function is further documented by our recent finding that the Hml mutant L131A is also deficient in agonist-induced internalization and down-regulation.2 Hence, this residue may also contribute to the elusive receptor domains mediating cellular trafficking of GPCRs. In a small number of cloned GPCR genes, such as the VIP (25) and secretin (26) receptors, the i2 loop sequence deviates considerably from the motif consisting of DRYXXV(1)XXF'L. It remains to be determined which role the i2 loop plays in the G protein coupling of those receptors.