A Polymorphism of the Human fls-Adrenergic ReceDtor within the Fourth Transmembrane Domain Alters Ligand Binding and Functional Properties of the Receptor*

the the first characterization of a occurring mutation of a human adrenergic Our the hypothe-sized role of this region of the receptor for ligand and receptor activation, as well as for establishing critical interactions overall receptor con- polymorphism wild-type cDNA ligated pcDNA-Neo, transfected a phosphate precipitation technique (20,21). cell isolated by their resistance pg/ml G418. of for transfection, efficiency of transfection with the Ile" construct, as WT &AR. found transfections lZ6I-CYP bound Ile" with affinity, and binding assays using this were used to screen the permanent clonal cell for the of the transfected receptor. transfected stability sion transfections Ilela as compared In preliminary studies, we also found that diazerine lzsI-CYP photolabeled Ilela receptor at the same molecular weight as the WT &AR on SDS-polyacrylamide gel electrophoresis, suggesting that Ile" represents a mature form of the receptor.

We have recently identified several naturally occurring variants of the human &-adrenergic receptor (BaAR). One of these polymorphisms, which is relatively uncommon, is a mutation occurring in the fourth transmembrane spanning domain, with Ile substituted for Thr at amino acid 164 within the proposed ligand binding pocket. This mutation is adjacent to Ser's6 which has been predicted to interact with the @-carbon hydroxyl group of adrenergic ligands.
To determine the functional significance of this variant, we constructed by site-directed techniques a mutated @pAR (Ile'"4) with this substitution and expressed it in CHW-1102 cells. In the presence of guanine nucleotide, Ile'" displayed a lower binding affinity for epinephrine as compared with the wild-type BaAR (Kr = 1460 2 79 versus 368 2 39 nM; p e 0.001). A similarly decreased affinity was found with the catecholamines isoproterenol and norepinephrine, but not with dobutamine or dopamine which lack hydroxyl groups on their @-carbons. In addition, antagonists without aromatic ring polar substituents displayed a decreased affinity for the mutated receptor. In agonist competition experiments conducted in the absence of guanine nucleotide, Ile'" failed to exhibit detectable high affinity binding, suggesting an impairment in the formation of the agonist-receptor-Gs complex. Consistent with this finding, functional coupling to Gs as determined in adenylyl cyclase assays was significantly (-60%) depressed with Ile'84 under both basal and agonist-stimulated conditions. BaAR sequestration, which is also triggered by agonist binding, was also found to be -66% reduced in the IleIB4 polymorphism.
This study represents the first characterization of a naturally occurring mutation of a human adrenergic receptor. Our findings generally support the hypothesized role of this region of the receptor for ligand binding and receptor activation, as well as for establishing critical interactions for overall receptor conformation.
The human &adrenergic receptor (P2AR)' has been exten-* This work was supported by National Institutes of Health Grant R01-HL45967 (to S. B. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
With regard to the latter, several lines of evidence have shown that the individual transmembrane spanning domains contain key amino acids which are critical determinants of agonist and antagonist binding. Strader et al. (15,16) have shown that Asp113 in the third transmembrane region, and Se?w and Se1307 within the fifth transmembrane region of the &AR, appear to interact with the amino and aromatic hydroxyl groups of catecholamine agonists, respectively. In addition to these direct interactions, there also appear to be contributions from other transmembrane domains which are necessary for formation of the hydrophobic binding pocket. Studies utilizing chimeric 4 8 2 -and &/Pn-adrenergic receptors have suggested important roles for these other domains, presumably either as sites for direct interaction with ligands or as critical for maintenance of the overall receptor conformation necessary for ligand binding (2, 13). Computer modeling techniques have also predicted a number of interactions between the &AR and agonists or antagonists (17,18). Some of these interactions have been difficult to prove by mutagenesis. Of particular importance to this report is the prediction that Ser'" in the fourth transmembrane region forms a hydrogen bond with the &hydroxyl group of agonists (18). Mutagenesis of S e P to Ala results in abnormal protein folding into the membrane, which has made the role of S e P 5 in agonist binding untestable (15).
Recently, we identified within a cohort of 107 subjects two uncommon polymorphisms of the P2AR gene which result in mutations of the receptor within transmembrane-spanning domains, one in the putative first transmembrane region (VaP4 + Met) and one in the fourth transmembrane region (Thr" + Ile) (19). Based upon studies such as those cited above, we postulated that the mutation of Thr" to Ile might result in abnormal ligand binding. We report here the consequences of this natural polymorphism when stably expressed in a mammalian cell line.

Methods
Constructs and Transfections-We have previously identified polymorphisms of the human &AR gene using a combination of temper-ature gradient gel electrophoresis as a screening technique followed by direct sequencing (19). Here we report the consequences of a single polymorphism at nucleic acid 491, where a thymidine was found to be substituted for a cytidine, resulting in the substitution of Ile for Thr at amino acid residue 164. This mutation is denoted Ilela. We mimicked this mutation in the wild-type &AR cDNA by site directed techniques similar to those described (3). The wild-type and mutant constructs were then individually ligated into the expression vector pcDNA-Neo, and CHW-1102 cells were permanently transfected by a calcium phosphate precipitation technique (20,21). Clonal cell lines were isolated by their resistance to 300 pg/ml G418. The amount of DNA required for transfection, and the efficiency of transfection with the Ile" construct, were the same as that of WT &AR. We found in transient transfections that lZ6I-CYP bound Ile" with high affinity, and binding assays using this radioligand were used to screen the permanent clonal cell lines for the presence of the transfected receptor. Growth rates of the transfected cells and the stability of expression of &AR were the same for transfections of Ilela as compared with WT &AR. In preliminary studies, we also found that diazerine lzsI-CYP photolabeled Ilela receptor migrated at the same molecular weight as the WT &AR on SDS-polyacrylamide gel electrophoresis, suggesting that Ile" represents a mature form of the receptor. Clonal lines expressing the same receptor level were utilized for all studies except where indicated.
Radwligand Binding Assays-Radioligand saturation and competition assays were performed essentially as described previously (21). Nearly confluent cells were washed in calcium-free phosphate-buffered saline (PBS) and scraped in ice-cold 5 mM Tris, 2 mM EDTA (pH 7.4). Membrane suspensions were then centrifuged at 40,000 X g for 10 min at 4 "C and pellets resuspended in 25 mM glycylglycine buffer (pH 7.4) containing 12.5 mM MgC12. Competition studies were performed in a final volume of 250 p1 containing 30 pM '251-cYP, 100 p~ GppNHp, 1 mM ascorbic acid, and 0-1000 p~ competing ligand at 37 "C for 1.5 h. For delineation of high and low affinity agonist binding constants, competition studies were carried out in the absence of guanine nucleotide under similar conditions as above. Saturation assays were performed with 5-300 PM '261-CYP with propranolol (1 p~) used to determine nonspecific binding. Assays were stopped by dilution and rapid filtration over Whatman GF/C filters and subsequently washed with 10 volumes of cold 10 mM Tris (pH 7.4) buffer prior to counting in a gamma counter.
Adenylyl cycluse Assays-Adenylyl cyclase activities were determined as described previously (3,6) by a modification of the method of Salomon (22). Briefly, membranes were prepared as described above and incubated in 30 mM Tris, 2 mM MgCL, 0.8 mM EDTA, 0.12 mM ATP, 0.06 mM GTP, 2.8 mM phosphoenolpyruvate, 2.2 pg of myokinase, 0.1 mM CAMP, 1.0 pCi of [w3'P]ATP in a final volume of 50 pl for 45 min at 37 'C. Reactions were stopped by the addition of excess unlabeled ATP. The [32P]cAMP formed was separated by sequential chromatography over Dowex and alumina columns and counted in a scintillation counter. Column efficiency was determined by [3H]cAMP which was included in the stop buffer. Activities were measured in the presence of water (basal) or various concentrations of agonist as indicated.
Sequestration-Receptor sequestration was performed by a method similar to that described elsewhere (5). Whole cells in monolayers were exposed to 10 p~ epinephrine in 75-cm2 flasks containing serumfree media with 0.1 mM ascorbic acid at 37 "C for the indicated times. Control cells were prepared similarly but not exposed to epinephrine. At the end of the incubation, cells were washed five times with icecold PBS and detached with ice-cold EDTA. Greater than 95% of cells remained intact after detachment as assessed by trypan blue exclusion. Cells were washed three times by centrifugation and resuspension in cold PBS and then assayed in polypropylene tubes containing 300 pM '261-CYP with or without the hydrophilic antagonist CGP12177 (1.0 p M ) or the hydrophobic antagonist propranolol (0.3 pM), in a final volume of 250 pl of PBS. This incubation was carried out for 3.5 h at 13 "C, conditions which provide for equilibrium ligand binding while inhibiting the return of sequestered receptors to the cell surface (5, 7, 23). The number of intracellular plus extracellular &AR was defined as the difference between total '261-CYP binding tors were defined as the difference between total lZ6I-CYP binding and that occurring in the presence of propranolol. Cell surface recepand that occurring in the presence of CGP12177. Typically, a 10% decrease in cell surface receptors represented a decrease of -40,000 cpm of '261-CYP binding as defined above.
Data Analysis-Radioligand binding and adenylyl cyclase curves were analyzed by nonlinear least squares techniques (24) Sources for all other reagents were as described (8,19,21). and Ile", respectively). In addition, the Kd for '2sI-CYP as derived by Scatchard analysis was also equivalent between the receptors (25.12 f 5.97 uersus 27.68 f 4.07 pM, respectively). In competition studies (performed in the presence of GppNHp), each receptor displayed agonist rank order affinities typical for the PzAR subtype (i.e. isoproterenol > epinephrine > norepinephrine). However, Ile" consistently displayed decreased affinities toward these agonists ( Table I).

RESULTS
The magnitude of the increases in the inhibition constant K, (approximately 4-fold) was the same for all three catecholamines. Since the weight of evidence has suggested that the neighboring amino acid Ser'65 plays a key role in binding of the @-hydroxyl group of catecholamines (17), we considered that the substitution of Ile at position 164 may perturb this interaction. We therefore studied two agonists, dopamine and dobutamine, which do not have @-hydroxyl groups. As shown in the table, in contrast to what was observed for catecholamines, neither dopamine nor dobutamine distinguished between WT PzAR and Ile" receptors.
We also examined the ability of the two receptors to form the high affinity ternary complex by performing agonist competition studies with isoproterenol in the absence of guanine nucleotide (Fig. 1). In the case of the WT P2AR, shallow curves were obtained which were best fit to a two-site model with a high affinity component representing 17 f 1.6% of receptors with an affinity of 0.90 f 0.22 nM and a low affinity component representing 82.9 zk 1.7% of receptors with an affinity of 296 f 58 nM (n = 6). In contrast, curves for Ile" were steep (mean Hill coefficient = 0.91) and displayed only a single low affinity component (Fig. 1). In the presence of GppNHp, WT @ZAR binding underwent a classic shift to steep curves with single low affinity binding components, whereas the single-site Ile'64 curves displayed minor rightward shifts (K, = 524 f 44 nM to 772 f 60 nM, p = not significant).
We next examined the binding characteristics for a group of PAR antagonists with diverse structures. Propranol, ICI118551, and CGP 20712 all clearly showed lower affinities for IlelU as compared with WT &AR, and alprenolol displayed a similar trend (Table I, lower portion). On the other hand, pindolol and CGP12177 showed no differences in affinities for the two receptors. Given that all of these antagonists have @-hydroxyl groups, it is clear that other interactions have also been perturbed. One common feature of the antagonists (and also agonists) which demonstrate decreased affinity toward Ile" is the absence of polar substituents on the aromatic rings.
To determine whether the differences in catecholamine binding found between WT @zAR and Ile" receptors trans- lates into differences in receptor function, we examined two agonist-dependent parameters: activation of adenylyl cyclase and receptor sequestration. As shown in Fig. 2 7.49 f 1.02 pmol/min/mg protein, respectively, n = 6, p < 0.01). These differences in activity were not due to differences in receptor expression, which were equivalent (see above). It is of interest to note that the basal (i.e. agonist-independent) activity of Ile'@ was also reduced as compared with WT PzAR (1.88 k 0.38 uersus 5.06 & 0.91 pmol/min/mg protein, respectively, n = 6, p < 0.02). As shown in Fig. 3, the lower basal adenylyl cyclase activities of Ile16* as compared with WT BzAR was observed in multiple clonal cell lines having a broad range of receptor densities. With increasing expression of either receptor, basal levels of adenylyl cyclase increased, but the Ile'64 mutant consistently displayed lower basal levels as compared with wild-type BzAR expressed at a similar level.
However, the basal activity of lle" is higher than that of FIG. 1. Agonist competition with 'z61-CYP for WT &AR and Ile'" receptors in the absence or presence of guanine nucleotide. Membranes prepared from cells expressing WT P2AR or Ile" were prepared and competition studies carried out as described under "Methods." WT &AR binding curves in the absence of GppNHp were shallow and displayed high and low affinity components, whereas Ilela binding curves were steep and displayed a single low affinity binding site. In the presence of GppNHp, WT &AR binding curves displayed a shift to a single low affinity component, whereas Ile" curves were unchanged. Shown is a single experiment representative of six performed. Mean data from six such experiments are presented under "Results." nontransfected CHW cells (which was 0.71 f 0.20 pmol/min/ mg protein, p < 0.05). It should also be noted that forskolininduced activation of Ile"j4 is not different from that of nontransfected CHW cells (15.13 f 4.15 uers'sus 14.54 f 0.24 pmol/min/mg protein, p = not significant). Taken together, these data suggest that the mutant receptor displays perturbed nonugonist-dependent coupling to Gs in addition to that observed in the presence of agonist.
Although the mechanisms of receptor sequestration are not known with certainty, it is clear that the process is triggered by agonist binding (25). We therefore assessed the ability of Ile" to undergo sequestration in response to saturating concentrations of epinephrine. Fig. 4 shows the results of epinephrine (10 ")-induced sequestration experiments with W T &AR and Ile". WT P2AR expressed in CHW cells undergo -15-20% sequestration under these conditions (5, 7, 8). Exposure of Ile" to epinephrine resulted in -65% less maximal sequestration as compared with W T P2AR (5.8 f 2.0 versus 16.4 f 1.2% sequestered, respectively, n = 7, p < 0.001).
It should be noted that these findings are not artifacts of differences in ligand binding between the two receptors since lZ5I-CYP and CGP12177 display equal affinities for WT P2AR and Ile", and saturating concentrations of propranolol were used in the sequestration assay.
As delineated in the above studies, there appears to be multiple potential aberrancies in the ligand binding interactions that have been perturbed in Ile'64. One possible component of this distinction is the proposed interaction of S e P 6 with the @-hydroxyl group of some ligands (17). To further evaluate this, the binding of the (+)-enantiomers of isoproterenol and propranolol were assessed. Both compounds in CHW cells expressing W T &AR or Ile16r. Confluent cells were exposed to 10 pM epinephrine and 0.1 mM ascorbic acid in serum-free media for the indicated times. Cells were then extensively washed, detached, and whole cell '*'I-CYP binding studies were performed at 13 "C as described under "Methods." Results from six independent experiments are shown. The maximal agonist induced sequestration for Ile" was lower as compared with WT pzAR (5.8 k 2.0% uersus 16.4 k 1.2%, p < 0.001).

DISCUSSION
It is well accepted that the major determinants of ligand binding in G-protein coupled receptors in general, and adrenergic receptors in particular, reside in the transmembrane a-helices (reviewed in Refs. 26 and 27). Mutagenesis studies have identified transmembrane region 3 and transmembrane region 5 of the p2AR as being especially important regions of receptor/ligand interaction. For example, mutation of either Se?" or Serzo7 in the fifth transmembrane region results in receptors with an approximately 10-fold decreased affinity for catecholamines; these results were reproduced when catecholamine analogs lacking aromatic hydroxyl groups were studied (15). Mutation of Asp113 in transmembrane region 3 resulted in considerable loss of affinity for both agonists and antagonists (16, 28). Based on charge considerations, this group was predicted to form the counterion for the ligand amine group. Taken together, these and other studies have suggested a hydrophobic binding pocket in which the catechol moiety is oriented toward transmembrane region 5 and the amine head group toward transmembrane region 3. This interpretation has been supported by computer models which attempt to account for steric interactions among the various receptor and ligand groups (17,18).
Unlike what is known about the third and fifth transmembrane regions, evidence for a particular role of the interposed fourth transmembrane region has been limited. Frielle et al.
(13) noted a shift in relative affinities for epinephrine and norepinephrine when transmembrane region 4 was exchanged in plAR/B2AR chimeras. This suggests that this region possesses residues that may provide for subtype selectivity of these receptors for agonists. Similar results were also obtained with subtype selective antagonists. In contrast, Kobilka et al. (2), using a2AR/j32AR chimeras, noted little effect of trans-membrane region 4 on ligand specificity. Strader et al. (15) substituted Ala for Ser at position 165 of the fourth transmembrane domain in the hamster P2AR but were unable to detect the resulting protein in transfected cells by immunoprecipitation and noted very little lZ5I-CYP binding. They interpreted this finding as reflecting improper folding of the receptor in the cell membrane, due to a global perturbation of receptor conformation (15).
Computer modeling also supports the notion of the above interactions between agonists and regions of transmembrane domains 3, 4, and 5, but the results of these modeling results are not universal (17,18). In one such model (17), Ser"j5 in transmembrane region 4 interacts with the @-OH group of the ligand aliphatic backbone. This model would be consistent with a stereoselective interaction at SeP5, since the @-carbon of physiologic adrenergic agonists forms a chiral center. Accordingly, removal of the @-hydroxyl group, as seen with the ligands dopamine and dobutamine, or substitution with the (+)-stereoisomer would be expected to eliminate the proposed hydrogen bond and thus result in equal affinities between the wild-type and mutant receptors. As noted under "Results," competition studies with dopamine, dobutamine, and (+)propranolol all failed to distinguish between WT P2AR and Ile", consistent with this proposed interaction. Surprisingly, however, (+)-isoproterenol maintained a decreased affinity for the mutant as compared with the wild-type. Taken together, one interpretation of our results is that the Ile substitution at residue 164 has perturbed the role of Ser'65, resulting in both its inability to interact with the @-hydroxyl group of some catecholamines and its inability to participate in the maintenance of proper receptor conformation. As discussed below, this latter effect may be manifested by altered binding of some antagonists and by depressed agonist and nonagonistdependent coupling of the receptor to Gs. This proposed mechanism of the effects of the Thr to Ile" substitution (i.e. interference with the function of Ser'65) is based on the majority of our results being consistent with the predicted role of Ser'65 by computer modeling (17) and results of sitedirected mutagenesis of Serlm (15). We recognize, however, that Thr" may have a direct interaction with some ligands which is perturbed by the substitution of Ile. Based on our findings, though, such a substitution would also affect nonagonist coupling to Gs, independent of SeP5. Although there is little supportive evidence for such, we recognize this possibility.
Examination of the divergent structures of the antagonists tested may suggest mechanisms for the changes in affinities observed. Antagonists with polar nitrogen substituents within the catecholamine aromatic regions (CGP12177, pindolol) fail to differentiate between WT P2AR and Ile'64. At physiologic pH, these groups presumably have a net positive charge and thus potentially provide a stabilizing interaction with another, as of yet unidentified, side group within the receptor. This stabilizing interaction may be strong enough to effectively compensate for any distorting interaction at the proposed Ser'65-@-hydroxyl hydrogen bond. Although CGP20712 does have nitrogen substituents on the aromatic catecholamine ring, it did manifest altered binding to the Ilela mutant. However, this compound also has a bulky trifluoromethyl group which presumably negates any polar properties which may be contributed by the nitrogens (Table I).
In addition to the abnormalities in ligand binding noted above, substitution of Ile for Thr at position 164 also results in functional abnormalities of adenylyl cyclase activation. In agonist competition binding studies performed in the absence of guanine nucleotide, we were unable to detect a high affinity binding site in the Ilelm mutant, suggesting an impairment in the ability of this receptor to form the high affinity agonist ternary complex. In adenylyl cyclase assays, such an impairment of physical coupling was manifested as a significantly depressed maximal agonist-stimulated activity as compared with WT &AR, under conditions where receptor expressions were the same.
We suggest that in addition to impaired agonist binding, another abnormality is present in Ile" which further contributes to its depressed function. We found that the basal ( i e . nonagonist-dependent) adenylyl cyclase activity of Ile" was also depressed. We and others ( Fig. 3 and Ref. 29) have noted that the basal adenylyl cyclase activity of nontransfected cells can be lower than that of transfected cells expressing high levels of &AR. Thus although the tendency for the receptor to couple to Gs is markedly favored in the agonist-occupied form, with high receptor expression detectable basal coupling indeed occurs (29). In the current study, Ilela has a basal adenylyl cyclase activity which is between that of WT BzAR and nontransfected cells, suggesting that the nonugonist occupied form of Ile'& is in some way different than that of WT P2AR. Since it has been shown that mutating Ser'= causes the receptor to be in a markedly abnormal conformation (15), it is not unexpected that our mutating the adjacent Ilela also results in a global, but clearly less dramatic, alteration in receptor function. Additionally, since the known regions of interaction with Gs are intracellular and are all somewhat remote from the mutation in transmembrane region 4, a global rather than local distortion of the receptor structure should be implicated.
We also observed depressed agonist-promoted sequestration of Ilel&. Although the precise molecular determinants for this process have not yet been identified, several recent studies have suggested that sequestration requires that the B2AR undergo a conformational change after agonist binding. Neither physical or functional coupling to Gs appear to be necessary for sequestration. Campbell et al. (10) studied several mutations of the human B2AR in CHW cells; although impairment of adenylyl cyclase activity was associated with loss of ability to undergo down-regulation, no mutant exhibited impaired sequestration. Cheung et al. (11) noted loss of Gprotein coupling and sequestration when the N-terminal portion of the third intracellular loop of the hamster &AR was deleted; substitution of the analogous region of the M1 muscarinic receptor restored sequestration but not coupling. Taken together, these results suggest a conformational rather than functional dependence for the integrity of the sequestration response. The observed impairment of sequestration for the IlelB4 mutant presented here is therefore most consistent with a perturbation in agonist-promoted triggering of an appropriate conformational change, rather than an indirect result of impaired coupling to Gs.
The current study is the first to delineate the functional characteristics of a naturally occurring mutation of any ad-renergic receptor. The presence of a dysfunctional &AR variant may explain, in part, the variability in responses to catecholamines which have been noted in physiologic studies in man (30). Based on the structures of the compounds studied, we cannot offer a unified mechanism for all of the mutated receptor's dysfunctions that we observed. It is clear, however, that this region (possibly dictated by S e P ) is important for ligand binding, agonist-mediated sequestration and coupling to Gs, and establishing overall receptor conformation.