Mutation of Peptide Binding Site in Transmembrane Region of a G Protein-coupled Receptor Accounts for Endothelin Receptor Subtype Selectivity*

The molecular basis for endothelin (ET) isopeptide se- lectivity between ET, and ET, receptors was studied by examining ligand binding to several site-specific mu- tants of the human ET, receptor. Based on a computer-built three-dimensional model of the ET, receptor, five non-conserved amino acids, clustered around the putative ligand binding site, were targeted for mutation to alanine. Expression of the wild-type and mutant ET, receptors in COS-7 cells revealed that the binding profile of one of the ET, mutants, Tyr'*' -.* Ala, was character- istic of the ET, receptor. In the 'I?yrla" - Ala ET, receptor mutant the affinity of two ET,-selective agonists, endo- thelin-3 and sarafotoxin S ~ C , was increased 10-200-fold, whereas that for two ET,-selective antagonists, BQ-123 and BMS-182874, was reduced 360-2,OOO-fold. Thus, mutation of a single amino acid in the second transmem- brane region of the wild-type

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked indicate this fact. The abbreviation used is: ET, endothelin. vasoconstriction and vasodilation of smooth muscle (4-61, pressor and depressor (2, 7) effects, positive myocardial inotropy and chronotropy (81, and mitogenicity (9). These diverse actions are widely attributed to the existence of multiple endothelin receptor subtypes with discrete and regulated cellular distributions and functions.
Due largely to the potent and long acting vasoconstrictor effects of the endothelin isopeptides on vascular and non-vascular smooth muscle, the endothelin receptors have been proposed as targets for therapeutic intervention in numerous diseases (10, 11). Two receptor subtypes for mammalian endothelins have been identified on the basis of molecular (12, 13) and pharmacological (1, 14) evidence. Comparison of deduced amino acid sequences for the ET, and ET, subtypes reveals that these proteins are 59% identical and are members of the putative heptahelical receptor family that is G proteincoupled. The level of conservation is greater in the intracellular loops and transmembrane regions where the sequence identity is 75%. Despite this level of homology between subtypes, ET, and ET, receptors are distinguished by differential affinities for peptidic and nonpeptidic ligands. The ET, receptor subtype binds ET-3 and sarafotoxin S6c with low affinity and BQ-123 and BMS-182874 with high affinity. Conversely, the ET, receptor subtype binds ET-3 and sarafotoxin S6c with high affinity and BQ-123 and BMS-182874 with low affinity.
Recent reports describing chimeric ET, and ET, receptors have attempted to address the molecular basis of subtype-selective ligand binding. These studies demonstrated that transmembrane regions 1, 2, 3, and 7 (151, as well as the the first extracellular loop (161, compose a subdomain for the ET, antagonist, BQ-123, whereas critical binding determinants for ET,-selective agonists appear to reside in transmembrane regions 4-6 (15). However, the precise amino acid residues involved in the binding site for peptidic or nonpeptidic ligands remain to be identified. Elucidation of the structural determinants involved in ligand binding is critical to understanding the mechanism of ET receptor-ligand interactions and to rational drug development. To this end, we developed a computergenerated three-dimensional model of the ET, receptor to guide site-directed mutagenesis aimed at probing ligand-selective interactions with this receptor. Here we report that Tyr12' in the second helix of the ET, receptor is a critical determinant of subtype A-selective ligand binding.
MATERIALS AND METHODS Model Building-Sequences of the human ET, and ET, receptor subtypes were aligned, and pairwise comparisons of amino acid hydrophobicity values were made using the Goldman, Engleman, Steitz hydrophobicity scale (17), which produced the conservation value as: hydrophobicity conservation value = scale range -(H, -HJ, where the scale ranges from moat hydrophobic (F = -3.7) to most hydrophilic ( R = 12.3) and HI and H2 are values for the corresponding ET, and ET, amino acid pair being compared (18). The identified transmembrane sequences were then threaded through the seven a-helices of bacteriorhodopsin (19). Because proline residues introduce kinks into straight helices with an average kink angle, 0, of 26" (55") (20) and helix kinking is important for shaping the size of the putative ligand binding cavity (20,21), helical segments in bacteriorhodopsin containing kinks caused by prolines were brought to standard a-helical +,$ values (-57,-47) by constrained minimization. Helices in the ET, model containing prolines were then kinked by constrained minimization of selected torsional angles until the kink angle 0 was within average values (see above). Molecular modeling was conducted with INSIGHT and DISCOVER (Biosym Technologies, San Diego, CAI and GRASP (A. Nicholls, Colum-

Endothelin Receptor Subtype Selectivity
bia University) with interactive graphics display and calculations conducted on a Silicon Graphics 4D/440 work station. The geometry of the seven-helix bundle was further optimized using molecular mechanics calculations as implemented in DISCOVER, followed by placement of side chains using the confomtional search procedure CONGEN (22).

Cloning and Mutagenesis of Mutant Receptors-Site-directed mu-
tagenesis of the human placental ET, receptor cDNA (23) was conducted using the method described by Deng and Nickoloff (24). Briefly, the wild-type E T , receptor cDNA was subcloned into a truncated version of vector pACYC184 designed to maximize the number of unique restriction sites. Oligonucleotide(s) that introduced the desired mutation, created an analytical restriction site, and destroyed a unique restriction site were synthesized and p d e d (Applied Biosystems 391 synthesizer, Palo Alto, CA). The modified DNA was used to transform mutS Escherichia coli, and plasmid DNA was prepared and digested with the appropriate restriction enzyme. The resulting digestion linearized the m u t a t e d DNA while leaving the mutated DNA circular.
Transformation of K-12 E. coli yielded colonies, the majority of which contained plasmid DNA with the desired mutation. Mutant plasmids were identified by restriction analysis and subsequently verified by sequencing the entire cDNA clone (25). The wild-type and mutant receptor CDNAE were then subcloned into the mammalian expression vector, pCDM8.
Expression of Receptors and Radioligand Binding-Wild-type and mutant ET, and ET, receptor cDNAs were transfected into COS7 cells using the polycationic lipid Lipofedamine (Life Technologies Inc.) according to the manufacturer's instructions. Cells were harvested 48-72 h after transfection in buffer A (Dulbecco's modifted Eagle's medium containing 20 m~ Hepes pH 7.4 at 37 "C, 0.1 m~ phenylmethylsulfonyl fluoride, 10 &d soybean trypsin inhibitor), Polytron-homogenized, and centrifuged at 100,000 x g for 1 h at 4 "C. The Supernatant was discarded and the membrane pellet was resuspended in buffer A. Membranes were homogenized and stored in aliquota at -80 "C. Radioligand binding wa8 conducted as previously described (26). COS7 cell membranes (0.610 pg of protein) were incubated with 30-50 PM mI-ET-l in the presence of increasing concentrations of competitor for 2 h at 37 "C. Nonspecific binding was defined in the presence of 100 1 1~ ET-1. Data were analyzed by iterative curve fitting to a 1-or 2-binding site model, and Ki values were calculated from IC, values (27).

RESULTS AND DISCUSSION
In previous model building of various heptahelical receptors, the seven transmembrane a-helices were typically predicted from amino acid sequence-derived hydrophobicity based on alignment to bacteriorhodopsin (28)(29)(30)(31)(32)(33)(34)(35)(36). In an attempt to more accurately determine helical boundaries of transmembrane regions, we have used a hydrophobicity conservation matrix (17,18) in coqjunction with sequence similarity, In a blind test, this procedure correctly identified the transmembrane regions in baderiorhodopsin to within 1 amino acid residue. Following construction, the ET, receptor model was analyzed for amino acid residues which 1) were predicted to be in the transmembrane regions (170 total residues), 2) differed between ET, and ET, subtypes (32 residues), 3) represented non-conservative amino acid pairs (side chain pairs considered to be conservative and discarded from selection were VakIle, Val:Leu, Metne, Leu:Met, and Ala:Ser) between ET, and ET, subtypes (12 residues), 4) were predicted to be at the extracellular surface (5 residues), and 5 ) were predicted to be in the putative binding cavity (2 residues; see Fig. L4). These two amino acid residues, Qru9 and Serle7 in the second and third transmembrane regions, respectively (see Fig. m), were targeted for alanine replacement by site-directed mutagenesis (23,24). In addition, the three other amino acids (Valzs, Gly6l, and Phem) predicted to be near the extracellular surface were mutated as a test of the ET, model.
Mutant and wild-type ET, receptors and wild-type ET, receptors were transiently expressed in COS-7 cells, and the af- Transmembrane helices are numbered Z through Vn. Amino acids displayed in space-filling mode and colored in green are the 5 residues targeted for site-directed mutation. Other residues colored in purple are those exposed in the extracellular cavity. C, cross-sectional view of the ET, receptor model with residues colored as described for B with the extracellular surface oriented toward the top.

TABLE I Summary of inhibition constants (Ki in nM) for the wild-type and mutant ET, receptors
Ki values are shown for the wild-type ET, receptor for comparison. Binding was conducted in membrane preparations from transiently transfected COS7 cells. Values are means 2 S.E. from three to six competition curves with each concentration in duplicate.

Agent
Wild-type ET, Y129A ET, S167A ET, finity of peptidic and nonpeptidic ligands was determined in 1251-endothelin-l competition binding experiments. Only binding to T y P 9 + Ala ET, was substantially affected (Table I). The affinity of two ET,-selective peptides, endothelin-3 and sarafotoxin S~C , increased. In contrast, the affinity of two ET,,-selective agents, the cyclic pentapeptide (37) BQ-123 (D-Asp-L-Pro-D-val-L-hU-D-Trp) and BMS-182874, a novel naphthalenesulfonamide ET, receptor-selective antagonist (381, decreased (Fig. 2). Binding of the non-selective N-pyrimidinyl benzenesulfonamide, Ro 46-2005 (39), was not altered by the Tyr"' + Ala mutation. Thus, the K, values of both ET, (BQ-123 and BMS-182874) and ET, (endothelin-3 and sarafotoxin S6c) subtype-selective ligands were altered to that of the opposing subtype. Importantly, neither the affinity of endothelin-1 nor the affinity of endothelin-2 was altered by the Tyr'" + Ala mutation, suggesting that the receptor has not undergone structural rearrangement. These data indicate that Tyr"' + Ala mutation of the human ET, receptor confers ETB-like receptor binding to a n ET, receptor.
Confirmation of the role of Tyr"' in ET, binding was provided by replacing this amino acid with histidine, the corresponding residue in the ET, receptor. Binding to the T y r 1 ' ' + His mutant ET, receptor resulted in a profile similar to the wild-type ET, receptor (K, values in nM: endothelin-1 = 0.3 2 0.0; endothelin-3 = 0.4 2 0.1; sarafotoxin S6c = 39 2 7; BQ-123 2 6).
The above data are consistent with the computer-built receptor model in that, of the 5 amino acid residues of the wild-type ET, receptor replaced with alanine, only those located in the putative ligand binding cavity produced a significant effect on binding (Table I). The Val225 + Ala and €' he3' ' 4 Ala mutants were not expected to affect ligand binding as they were pre-= 2,600 2 500; BMS-182874 = 74,100 2 9,900; RO 46-2005 = 400 Subtype Selectivity 12385 dicted to be oriented toward the phospholipid environment. Similarly, mutation of the glycine at position 261 to alanine should do little to affect receptor-ligand interactions.
The results with the T y r l " + Ala ET, mutant are consistent with the hypothesis that helix 2 contributes to "fine selectivity" of ligand binding. Although transmembrane regions 1,2,3, and 7 (15) as well as the first extracellular loop (16) of the ET, receptor were previously suggested to be involved in the selective high affinity binding of BQ-123 to these receptors, our data are the first to indicate that Tyr"' in transmembrane region 2 is critical to ET receptor ligand selectivity. Interestingly, preliminary experimentation suggests that the Tyr'" + Ala ET, mutant does not display an altered functional responsiveness Numerous reports have detailed precise contacts between small nonpeptide ligands and amino acid residues in transmembrane regions of aminergic and serotonergic (40)(41)(42)(43) receptors. Similarly, nonpeptidic ligands for G protein-coupled receptors for which the natural ligand is a peptide have been shown to interact with amino acids in transmembrane regions (44)(45)(46). In contrast, peptidic ligands often contact residues in the extracellular loop regions of their receptors (46,47). The alterations in binding affinity of endothelin-3, sarafotoxin S~C , and BQ-123 for the Tyr'" ET,, mutant suggest that these endothelin isopeptides and peptidic antagonist interact with Tyr"' in the second transmembrane region of the ET, receptor.
Indeed, because the affinity of BMS-182874 is also affected by the Tyr'" + Ala mutation, these data indicate that the naphthalene sulfonamide antagonist also interacts with Tyr'''. This represents the first example of a peptide interaction with a transmembrane region of a G protein-coupled receptor.
In summary, this is the first report describing a specific component of the ET, receptor binding site. To the extent that all the surface-exposed, receptor subtype-variable amino acids were evaluated by alanine-scanning mutagenesis, the data reported here indicate that in transmembrane region 2 is involved in determining subtype-selective peptidic and nonpeptidic (agonist and antagonist) ligand binding and contributes to fine selectivity in binding to endothelin receptors. to