Chimeras of (cid:97) 1 -Adrenergic Receptor Subtypes Identify Critical Residues That Modulate Active State Isomerization*

We have identified previously two amino acids, one in each of the fifth and sixth transmembrane segments of both the (cid:97) 1a -adrenergic receptor and the (cid:97) 1b -adrenergic receptor (AR), that account almost entirely for the selectivity of agonist binding by these receptor subtypes (Hwa, J., Graham, R. M., and Perez, D. M. (1995) J. Biol. Chem. 270, 23189–23195). Thus reversal of these two res- idues, from those found in the native receptor of one subtype to those in the other subtype, produces comple- mentary changes in subtype selectivity of agonist binding. Here we show that mutating only one of these resi- dues in either the (cid:97) 1b -AR or the (cid:97) 1a -AR to the corresponding residue in the other subtype (Ala 204 3 Val for the (cid:97) 1b ; Met 292 3 Leu for the (cid:97) 1a -AR) results in chimeras that are constitutively active for signaling by both the phospholipase C and phospholipase A 2 path- ways. This is evident by an increased affinity for agonists, increased basal phospholipase C and phospho- lipase A 2 activation, and increased agonist potency. Although mutation of the other residue involved in agonist binding selectivity, to the corresponding residue in the other subtype (Leu (40:20:0.5; v/v/v) was added. The tubes were vortexed and centrifuged to remove The supernatant, which contained released arachidonic acid, was added to 20 ml of scintillation mixture, dark-adapted, and counted for radioac- tivity. The counts indicated the relative amount of arachidonic acid released. Previous experiments with thin-layer chromatography have indicated the identity of the radioactivity as arachidonic acid (1). -AR normal packing and sixth helical type valine position methionine position valine leucine position 292, Met 292 3 Leu valine at position 185 constitutive activation, normal interhelical packing is prevented. this later steric constraint be relieved with the substitution of Val 185 by alanine. Thus the valine at this position in both the (cid:97) 1a -AR and (cid:97) 1b -AR acts as a molecular inducer by perturbing conformation, whereas a methionine in transmembrane VI functions as a repressor, reverting the valine-induced conformational change back to normal.

Recently a number of point mutations have been described in both the ␣ 1 -AR and other adrenergic receptors that result in partial activation in the absence of agonist stimulation (3,4). Such mutations are not unique to the adrenergic receptor superfamily, since constitutively active somatic mutations of the human thyrotropin receptor that result in adenomas and hyperthyroidism have also been identified (5). In addition, mutant hyperfunctioning luteinizing hormone receptors have been reported that lead to precocious puberty in young males (6). In general, these mutations are characterized by increased agonist binding, increased agonist independent receptor activation, and increased agonist potency (EC 50 ). The increased agonist binding is independent of G-protein binding and thus is an intrinsic property of the receptor. Additionally, constitutive activity of these mutants can be reversed by some antagonists. Based on these findings, as well as the observation that overexpression of wild type receptors can also initiate biochemical responses in the absence of agonist, it has been proposed that receptors spontaneously isomerize between a basal state, R, and an active state, R*. Furthermore, agonists bind with higher affinities to the active state and "trap" the receptor in the R* conformation.
With regard to the hamster ␣ 1b -AR, two noteworthy mutations have been identified. Substitution of Ala 293 in the third intracellular loop to any of the other amino acids results in varying degrees of constitutive activity (3). In addition, in the third transmembrane domain, replacement of the natural cysteine residue, Cys 128 , by phenylalanine also results in constitutive activity (4). However, by contrast with the Ala 293 mutants, the increased basal effector activation and increased agonist potency observed with the Cys 128 3 Phe mutation are pathway-specific, since they are confined almost entirely to the PLC pathway. Here we show that when one of two residues in either the ␣ 1a -or the ␣ 1b -ARs, which we have shown previously are involved in determining subtype selectivity for agonists (7), is mutated to the corresponding residue in the other subtype (Ala 204 3 Val, ␣ 1b -AR; Met 292 3 Leu, 2 ␣ 1a -AR), the resulting chimeras are constitutively active for both the PLC and PLA 2 pathways. Mutations in Val 185 in the ␣ 1a -AR (the residue corresponding to Ala 204 in the ␣ 1b -AR) or Leu 314 in the ␣ 1b -AR (the residue corresponding to Met 292 in the ␣ 1a -AR) when replaced by the corresponding residue in the other subtype, are not constitutively active. However, when combined with the constitutively activating mutants, the resulting double mutants, Ala 204 3 Val/Leu 314 3 Met and Val 185 3 Ala/Met 292 3 Leu, show a restoration of native receptor signaling. This paper documents that active state isomerization of receptors can be modulated by the composition of specific amino acids in adjacent transmembrane domains.

Site-directed Mutagenesis
The construct used was the hamster ␣ 1b -AR cDNA and rat ␣ 1a -AR cDNA described previously (7). Site-directed mutagenesis was performed as described previously (7), utilizing the oligonucleotide-mediated double primer M13 method (8). Mutations were confirmed by DNA sequencing.

Transfection of COS-1 Cells
COS-1 cells (ATCC) were transfected with the various constructs as described previously (7) using the DEAE-dextran method. Initial transfections for ligand binding analysis were performed with 5 g of DNA/ml of transfection solution. To achieve equal receptor numbers for the functional assays, the Ala 204 3 Val construct was reduced to 2 g of DNA/ml of transfection solution while the wild type receptor was reduced to 1 g of DNA/ml of transfection solution. Transfection efficiencies for the DEAE-dextran method ranges from 10 to 20%.

Ligand Binding
The ligand binding characteristics of the expressed receptors were determined in a series of radioligand binding studies using [ 125 I]HEAT, an ␣ 1 -specific antagonist (9). Membranes were prepared as described previously (9). For competition studies, the procedure involved duplicate tubes containing 200 pM [ 125 I]HEAT, HEM buffer (20 mM HEPES, pH 7.5, 1.4 mM EGTA, 12.5 mM MgCl 2 ), and competing ligand (at 10 or more different dilutions). Nonspecific binding was determined by the addition of phentolamine (10 Ϫ4 M). For saturation binding studies, 80 -1600 pM [ 125 I]HEAT was used. After 1 h of incubation at room temperature, the reactions were stopped by the addition of ice-cold HEM buffer and were filtered onto Whatman GF/C glass filters with a Brandel Cell Harvestor. The filters were washed five times with ice-cold HEM. They were then analyzed for bound radioactivity using a Packard Auto-gamma 500 Counter. Data were then analyzed with the interactive program LIGAND. IC 50 values were converted to K i values using the Cheng-Prusoff equation and a K D for [ 125 I]HEAT of 90 pM. K i values were expressed as the mean Ϯ standard error of the mean. An ANOVA and Student's t test were used to determine significant differences (p Ͻ 0.05). To detect small but significant differences, sets of mutations along with the two control wild type receptors were evaluated simultaneously (i.e. in the same assay). 3 H Radio-receptor Assay (DuPont)-We used this technique to accurately measure the total basal production of IP 3 . In brief it involved washing the cells in Hanks' balanced salt solution, followed by a 30-min incubation in 10 mM LiCl. Trichloroacetic acid (100%) was then used to disrupt the cells and the cell debris removed with ultracentrifugation. The trichloroacetic acid was removed by 1,1,2-trichloro-1,2,2-trifluoroethane:tri-n-octylamine (3:1) extraction. The resultant supernatant was then used in competition against a known concentration of prelabeled ([ 3 H]IP 3 ) IP 3 receptors. A standard curve was performed. This enabled us to assess the total amount of IP 3 in our cells.

Inositol Phosphate Determination
PI Hydrolysis Assay-For dose-response studies, PI hydrolysis was determined as described previously (9). Briefly, cells expressing the receptor constructs following transient transfection were grown in 60-mm dishes (3 ϫ 10 6 cells/dish). They were then washed and labeled for 16 -24 h with [ 3 H]inositol (DuPont NEN) at 1 Ci/ml in DMEM supplemented with 5% fetal bovine serum. After 24 h, cells were washed three times with DMEM (no serum) and incubated in the serum-free medium for 30 min, followed by a 30-min incubation in DMEM containing 10 mM LiCl. Agonists were then added, incubated for 30 min, after which the medium was removed and the cells lysed with 1 ml of ice-cold 0.4 M perchloric acid. One-half volume of 0.72 N KOH, 0.6 M KHCO 3 was added and the sample centrifuged. The supernatant was applied to 1-ml packed AG1-X8 (Bio-Rad) columns (100 -200-mesh, formate form). The column was washed with 16 ml of 0.1 M formic acid, and total inositol phosphates were eluted with 1 M ammonium formate, 0.1 M formic acid in 2-ml fractions. Scintillate was added and the fractions counted. Functional assays and equilibrium binding studies (to determine receptor densities) were performed on sequential transfections in which equal quantities of DNA were used.

Arachidonic Acid Assay
This was performed according to a previously published protocol (1). Briefly, 24 h prior to assay the transiently transfected cells were washed once in serum-free DMEM followed by the addition of [ 3 H]arachidonic acid (1 Ci/ml) and serum-free DMEM. To inhibit receptor signaling, phentolamine, an ␣ 1 -AR-specific antagonist, was added to a final concentration of 100 M. On the day of assay, the cells were washed twice with serum-free DMEM and incubated for 30 min. Phenidone, a dual cyclooxygenase and lipooxygenase inhibitor, was added to a final concentration of 100 M for 30 min. Agonists were added, and after another 30 min, the medium was removed and the cells were disrupted with 1 ml of 0.4 M perchloric acid. The fluid and debris were scraped into tubes, and 1 ml of MEOH:CHCl 3 :HCl (40:20:0.5; v/v/v) was added. The tubes were vortexed and centrifuged to remove debris. The supernatant, which contained released arachidonic acid, was added to 20 ml of scintillation mixture, dark-adapted, and counted for radioactivity. The counts indicated the relative amount of arachidonic acid released. Previous experiments with thin-layer chromatography have indicated the identity of the radioactivity as arachidonic acid (1).

Pharmacological Characterization of Ala 204 3 Val and
Leu 314 3 Met (␣ 1b -AR)-Ala 204 and Leu 314 in the ␣ 1b -AR are located in the fifth and sixth transmembrane domains, respectively, and are believed to face the ligand binding pocket ( Fig.  1) (7,10). We have shown previously that these residues are critical in defining an ␣ 1b -AR agonist-selective pharmacology. The double mutant, Ala 204 3 Val/Leu 314 3 Met, has an ␣ 1a -AR agonist pharmacology (7). For the purpose of clarity, some of these data are repeated in Table I. Full pharmacological characterization of these mutants are described in detail elsewhere (7). As shown in Table I, the single mutant Ala 204 3 Val exhibits an increased binding affinity (3-10-fold) for the catecholamines epinephrine (p Ͻ 0.001) and norepinephrine (p Ͻ 0.001). This mutant also has a 10-fold increase in affinity for oxymetazoline (p Ͻ 0.001) as well as for other imidazoline agonists, a phenomenon not seen with two other constitutively active mutations of this receptor (4). This indicates that the enhanced affinity is the result of a shift in the pharmacological specificity of the ligand binding pocket due specifically to the Ala 204 3 Val mutation. This increased binding affinity does not arise from an effect due to enhanced G-protein coupling, since pertussis toxin pretreatment (1 g/ml) or Gpp(NH)p, which uncouples receptor/G-protein interactions, did not alter the increase in agonist binding affinity (data not shown). There was no significant change in antagonist binding (7) (data not shown), including the affinity of Ala 204 3 Val for [ 125 I]HEAT (K D 61 Ϯ 10 pM), as compared to the published affinity of the wild type ␣ 1b -AR (K D 78 pM) (11).
Leu 314 3 Met alone exhibited no significant increase in the binding of either the nonselective agonists, (Ϫ)-epinephrine and (Ϫ)-norepinephrine (Table I), or antagonists (data not shown). There was, however, an increased binding affinity for oxymetazoline and methoxamine, two ␣ 1a -selective agonists, as described previously (7). When Leu 314 3 Met was combined with Ala 204 3 Val in the same ␣ 1b -AR construct, there was an overall decrease in binding affinity for nonselective agonists as compared to the Ala 204 3 Val mutant alone. The additional mutation of Leu 314 to methionine, therefore, had an inhibitory effect on the binding affinity for the two natural ligands. This same trend was not apparent for the synthetic subtype-selective agonists.
Equilibrium binding studies revealed expression levels (3.0 -3.9 pmol/mg protein, respectively) for both the Ala 204 3 Val and the Leu 314 3 Met single mutants which were similar to that observed with the wild type ␣ 1b -AR (Table I). When combined, the expression level of the double mutant (Ala 204 3 Val/Leu 314 3 Met) dropped to 1.2 pmol/mg protein, a value not dissimilar to that observed for the wild type ␣ 1a -AR (1.0 pmol/mg protein).
Inositol Phosphate Production-The ability of each construct to activate PLC was determined by the two techniques described under "Experimental Procedures." To measure total basal levels of IP 3 , a [ 3 H]IP 3 radio-receptor assay (DuPont) was used on the transfected COS-1 cells. From this we were able to accurately measure total IP 3 produced by the receptors in the absence of agonist. The Ala 204 3 Val mutant showed almost a 3-fold increase in basal activity over that of the wild type ␣ 1b -AR (p Ͻ 0.05) ( Fig. 2A). As indicated above, our binding data indicated that the wild type ␣ 1b -AR, and the single mutants, Ala 204 3 Val and Leu 314 3 Met, are expressed more abundantly than the wild type ␣ 1a -AR (Table I). Therefore, to eliminate the possibility that any increase in activity may be related to receptor density alone, we expressed each construct at multiple levels: 0.6, 2.8, and 3.0 pmol/mg protein for the wild type ␣ 1b -AR and 0.6, 2.0, and 3.9 pmol/mg protein for the Ala 204 3 Val mutant. This was accomplished by changing the amount of cDNA transfected. This indicated that the increased basal activity observed with Ala 204 3 Val is directly related to receptor density. Graphs of basal activity versus receptor number for the mutants produced slopes that were 3 times greater than the wild type receptor (Fig. 2C), consistent with the 3-fold increased basal activity of Ala 204 3 Val. In the presence of the antagonist, phentolamine (100 M) (added 24 h prior to assay), there was a significant decrease (p Ͻ 0.05) in IP 3 production ( Fig. 4A), indicating negative agonist activity.
To determine agonist potency for the PLC pathway, total inositol phosphates were determined by adding [ 3 H]inositol 24 h before the assay and then measuring epinephrine-stimulated inositol phosphate release. The Ala 204 3 Val mutation showed a leftward shift in the (Ϫ)-epinephrine concentration-response curve, as compared to the wild type ␣ 1b -AR or the ␣ 1a -AR. The EC 50 (concentration of agonist producing 50% of maximal activation) values Ϯ S.E. were 11 Ϯ 1 nM, 38 Ϯ 4 nM, and 77 Ϯ 8 nM, respectively (Fig. 5A). This change correlated with the ϳ3-fold decrease in the K i value for epinephrine and norepinephrine with this mutant. Maximum (Ϫ)-epinephrine-stimulated inositol phosphate turnover by the Ala 204 3 Val mutant was not different from that of the wild type receptor (Fig. 6C).
Arachidonic Acid Production-The ability of each construct to activate the PLA 2 pathway was determined by prelabeling the transfected COS-1 cells with [ 3 H]arachidonic acid (1 Ci/ ml) for 16 -24 h and measuring agonist-stimulated [ 3 H]arachidonic acid release. As shown in Fig. 3A, with the Ala 204 3 Val mutation, basal arachidonic acid release was increased above that observed with the wild type ␣ 1b -AR. However, there was no significant difference in arachidonic acid release between the wild type ␣ 1b -AR, and either the wild type ␣ 1a -AR or the Leu 314 3 Met mutation. As with IP 3 production, to eliminate the possibility that any increase in activity may be related to receptor density alone, we expressed each construct at multiple FIG. 1. Schematic diagram of the fifth and sixth transmembrane domains (TMV and TMVI) showing the locations of the targeted residues. Ala 204 (bold, italics) in the wild type ␣ 1b -AR is positionally equivalent to Val 185 (bold) in the wild type ␣ 1a -AR. Similarly Leu 314 (bold, italics) in the wild type ␣ 1b -AR is positionally equivalent to Met 292 (bold) in the wild type ␣ 1a -AR. These residues are postulated to be at equivalent positions on transmembrane domains V and VI, both being one helical turn away from the extracellular surface.

TABLE I
Agonist binding by wild type ␣ 1a -AR, wild type ␣ 1b -AR, and mutant ␣ 1 -ARs Competition binding studies were used to determine K i values (Ϫlog Ϯ S.E.) using membranes prepared from COS-1 cells expressing the wild type ␣ 1b -AR, and the single or double ␣ 1b -AR mutants or the wild type ␣ 1a -AR and the single or double ␣ 1a -AR mutants. All competition binding isotherms were best fit to a single-site model. Mean receptor densities (B max ) were determined on the same membranes from equilibrium binding studies. Each mutant receptor was compared to its respective wild type control. One-way analysis of variance (ANOVA) was followed by a Student-Newman-Keuls multiple comparison test. P-values indicate significant differences from the ␣ 1b -AR (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001) or from the ␣ 1a -AR ( ϩ , p Ͻ 0.05, ϩϩ , p Ͻ 0.01, ϩϩϩ , p Ͻ 0.001). These data have been taken from previously published studies (7).  (Fig. 3C). The addition of phentolamine (100 M) significantly inhibited the basal levels of arachidonic acid release observed with the Ala 204 3 Val mutation (p Ͻ 0.05) (Fig. 4B).
With this mutant, the EC 50 (1.0 Ϯ 0.6 nM) for (Ϫ)-epinephrine was decreased by approximately 10-fold compared to the wild type ␣ 1b (6.7 Ϯ 0.9 nM) or ␣ 1a -AR (11 Ϯ 1 nM) (Fig. 5B). Val showed a significant difference from the wild type ␣ 1b -AR (*, p Ͻ 0.05). Receptor expression levels were the same as for Fig. 2. Panel B, an identical assay to the above. However, in this experiment the mutant ␣ 1a -AR constructs were used. The Met 292 3 Leu was the only mutant that showed a statistical difference from the wild type ␣ 1a -AR (*, p Ͻ 0.05). Panel C, relationship of receptor density to basal arachidonic acid release. Wild type (ç) or the Ala 204 3 Val mutant (f) were expressed in COS-1 cells at different receptor densities by varying the amount of cDNA used in each transfection. The slope of the wild type receptor was 5.7 cpm/fmol receptor, while that of the Ala 204 3 Val was 23.5 cpm/fmol receptor.
ing were now no longer increased over those observed with the wild type ␣ 1b -AR ( Figs. 2A and 3A). The EC 50 of the double mutant for (Ϫ)-epinephrine-stimulated IP production was 148 Ϯ 16 nM compared to 11 Ϯ 1 nM for the Ala 204 3 Val mutant alone (Fig. 5C). For arachidonic acid release, the EC 50 for epinephrine was 6 Ϯ 1 nM as compared to 1 Ϯ 1 nM for the Ala 204 3 Val mutant alone (Fig. 5D). The maximal stimulation of the double mutant with (Ϫ)-epinephrine was similar to wild type receptors for both IP and arachidonic acid production (Fig. 6, C and D).
Inositol Phosphate and Arachidonic Acid Release by the ␣ 1a -AR Chimeras-Since the ␣ 1a AR chimeras, Val 185 3 Ala and Met 292 3 Leu, influence agonist binding in a comparable but reciprocal fashion to the respective ␣ 1b -AR chimeras, Ala 204 3 Val and Leu 314 3 Met, we hypothesized that reversing Met 292 to Leu in the ␣ 1a -AR should induce constitutive activity and that combining this mutation with the Val 185 3 Ala mutation should restore native signaling activity for both the PLC and PLA 2 pathways. Consistent with this hypothesis, the EC 50 values for epinephrine-stimulated IP and arachidonic acid release were 11 Ϯ 1 and 2 Ϯ 0.2 nM, respectively, for Met 292 3 Leu (␣ 1a -AR chimera) as compared to 77 Ϯ 4 and 11 Ϯ 1 nM for the wild type ␣ 1a -AR (Fig. 6, A and B). The Met 292 3 Leu chimera also showed increased basal IP 3 (Fig. 2B) and arachidonic acid release (Fig. 3B) that could be inhibited by phentolamine (100 M) (Fig. 4). However, when combined with the Val 185 3 Ala mutation, not only the increased agonist binding but also the increased signaling observed with Met 292 3 Leu mutation alone were now suppressed back to the levels observed for the wild type ␣ 1a -AR. The antagonist binding affinity for Met 292 3 Leu remained the same as the wild type ␣ 1a -AR, retaining the 10 -100 fold selectivity for phentolamine, 5-methylurapidil, and WB4101 compared with the wild type ␣ 1b -AR (data not shown). The EC 50 values for epinephrinestimulated IP 3 and arachidonic acid release by the double mutant were 93 Ϯ 15 nM and 13 Ϯ 2 nM, respectively: values that did not differ significantly from those for the wild type ␣ 1a -AR (Fig. 5, C and D). Likewise, maximal IP 3 and AA stimulation by (Ϫ)-epinephrine with the double mutants did not significantly change from that observed with wild type receptors (Fig. 6, C and D). DISCUSSION Like other G-protein-coupled receptors, the molecular mechanism involved in ␣ 1 -AR activation remains unclear. Some insights into receptor activation have come recently from the discovery of mutations that constitutively activate G-proteincoupled receptors. These studies suggest that the native receptors spontaneously isomerize between a basal (R) and activated (R*) conformation, with only the R* conformation being able to productively interact with the receptor-coupled G-protein and, thus, to result in effector activation. Furthermore, agonists bind to the R* state with higher affinity than to the R state and "trap" the R* conformation. For wild type receptors, the equilibrium between the R and R* states markedly favors the basal conformation, and, in the absence of agonist, effector activation is minimal, unless the receptors are highly overexpressed. With marked overexpression, the number of receptor molecules in the R* state increases sufficiently to allow effector activation in the absence of agonist, even though the ratio of molecules in the R versus R* state is unchanged. By contrast to the wild type receptors, with mutations that constitutively activate the receptor, the energetic requirements for spontaneous isomerization from the R to R* are reduced and thus at any instance more receptors are in the R* conformation. This allows effector activation in the absence of agonist and explains the other hallmarks of constitutive activity, i.e. an increase in both agonist binding and in agonist potency.
In this paper we have characterized novel ␣ 1b and ␣ 1a chimeras that involve residues previously shown to be major determinants of the selectivity of these subtypes for agonist binding. From our earlier experiments (7), we demonstrated that one of these chimeras, which involves a change of Ala 204 in the ␣ 1b -AR to the equivalent residue in the ␣ 1a -AR (Val 185 ), resulted in a mutant ␣ 1b -receptor that recognized a number of agonists including the natural ligands, (Ϫ)-epinephrine and (Ϫ)-norepinephrine, with higher affinity (Table I). Because both the wild type ␣ 1b -and ␣ 1a -ARs bind both (Ϫ)-epinephrine and (Ϫ)-norepinephrine with similar affinity, we questioned whether the higher agonist affinity of the Ala 204 3 Val mutant signified that it was constitutively active. The results show that in addition to increased agonist binding, this mutant displayed all the properties of a constitutively activated receptor, including agonist-independent effector activation and increased agonist potency. What was also intriguing was that when this Ala 204 3 Val mutation was combined with a change in Leu 314 (the other residue identified previously to be involved in subtype selectivity for agonists) to methionine (the equivalent residue in the ␣ 1a -AR), all three parameters of constitutive activity were suppressed. Thus, the phenotype of the double mutant was indistinguishable from the wild type ␣ 1b -AR, except for a reversal of its agonist-binding profile to that of the ␣ 1a -AR.
To confirm that this phenomenon was not merely serendipitous or unique to the ␣ 1b -AR structure, we reversed the mutations in the ␣ 1a -AR. Characterization of these chimeras revealed analogous but complementary effects to those observed with the ␣ 1b chimeras. Mutation at Met 292 to leucine (the equivalent residue in the ␣ 1b -AR) resulted in constitutive activity. Mutation of Val 185 to alanine (the reverse of the Ala 204 3 Val mutation in the ␣ 1b -AR) did not change receptor binding or signaling from that observed with the wild type ␣ 1a -AR. However, when combined with the Met 292 3 Leu mutation all parameters of constitutive activity were suppressed, so that binding of (Ϫ)-epinephrine and (Ϫ)-norepinephrine, and signaling by the double mutant did not differ from the wild type ␣ 1a -AR.
To further characterize the constitutively active ␣ 1 chimeras, we evaluated antagonist binding and the effects of antagonists on basal effector activation. Phentolamine suppressed basal effector activation by both the ␣ 1b -AR constitutively active mutant, Ala 204 3 Val, and the ␣ 1a -AR constitutively active mutant, Met 292 3 Leu, indicating that phentolamine is a negative agonist. However, there was no change in the binding affinity of the mutants for phentolamine or other ␣ 1 -antagonists, including the ␣ 1a -selective compounds, 5-methylurapidil and WB4101. One would predict that these constitutively active mutants would recognize these compounds with lower affinity, since fewer receptor molecules would be in the R (basal) conformation that binds negative agonists with higher affinity than the R* conformation. Failure to detect a change in affinity, however, can be explained by the fact that the Ala 204 3 Val (␣ 1b -AR) and Met 292 3 Leu (␣ 1a -AR) mutations only result in partial activation. Thus, with these mutations the proportion of molecules in the R* state is presumably not large enough to detect a change in affinity. In support of this notion, combination of the Ala 204 3 Val mutation with two previously defined constitutively active mutations, Cys 128 3 Phe and Ala 293 3 Glu (4), resulted in not only increased constitutive activity, compared to either single mutation alone, but also a lower affinity (ϳ10 -20-fold) for a number of antagonists including phentolamine. 3 Based on these findings, the question arises as to the mechanism(s) involved in the altered binding and signaling properties of these ␣ 1 chimeras. Differences based simply on alterations in agonist/receptor contacts due to altered properties of the substituting residues, in terms of hydrophobicity or size, as proposed previously to explain the changes in agonist binding alone (7), cannot simply explain the increased agonist-independ-FIG. 5. Dose-response curves for IP and arachidonic acid release using (؊)-epinephrine as the agonist. Panel A, an IP assay was used to measure the potency of second messenger production upon addition of varying concentrations of (Ϫ)-epinephrine for the two wild type and Ala 204 3 Val mutant receptor. Five to 10 separate dose-response experiments were performed to determine each response. Each point represents the mean Ϯ S.E. for each concentration. There was no significant difference for the two wild type receptors. However, there was a leftward shift for the mutation Ala 204 3 Val by approximately 1 log fold. Panel B, an arachidonic acid assay was used to measure the potency of second messenger production with different concentrations of (Ϫ)-epinephrine, for the two wild type receptors and for Ala 204 3 Val mutant. Five to ten separate assays were performed to establish each response. Each point represents the mean Ϯ S.E. for each concentration. The two wild type receptors did not differ significantly. However, there was a 1 log fold leftward shift with the Ala 204 3 Val mutation. Panel C, dose-response curve for IP production using varying concentrations of (Ϫ)-epinephrine for the wild type ␣ 1b -AR, the Ala 204 3 Val, and the combined Ala 204 3 Val/Leu 314 3 Met mutations. Each data point represents the mean Ϯ S.E. derived from eight separate experiments. With the Ala 204 3 Val mutation, there was a shift to the left in the dose-response curve. However, with the double mutation, Ala 204 3 Val/Leu 314 3 Met, the dose-response curve was now similar to that observed for the wild type receptor. Panel D, dose-response curve for arachidonic acid production using varying concentrations of (Ϫ)-epinephrine for the wild type ␣ 1b -AR, the Ala 204 3 Val, and the combined Ala 204 3 Val/Leu 314 3 Met mutations. Each data point represents the mean Ϯ S.E. derived from eight separate experiments. Similar to panel C, with the Ala 204 3 Val mutation there was a shift to the left in the dose-response curve. However, with the double mutation, Ala 204 3 Val/Leu 314 3 Met, the dose-response curve was now similar to that observed for the wild type receptor. ent effector activation observed with the Ala 204 3 Val and Met 292 3 Leu mutations, or the restoration of wild type signaling with the addition of the Leu 314 3 Met and Val 185 3 Ala mutations, respectively. Rather, we speculate that the observed changes in agonist binding and receptor signaling are due to perturbations in the helical structure of the involved transmembrane segments. The amino acid composition of an ␣-helix, for example, has been found to be critical for determining conformation (12)(13)(14)(15). Also, changing the amino acids in an ␣-helix can perturb interhelical interactions (16). In this regard, the complementarity of the binding and signaling effects observed with the chimeras indicates there are critical interactions between the fifth and the sixth transmembrane segments. Thus the residues identified here may influence helical conformation, either by direct intrahelical effects or indirectly by an effect at the level of the side-chain packing between adjacent helices.
Although structural resolution at the atomic level will be required eventually to discern effects on helical conformation, based on the following considerations, we propose a model involving predominant effects on interhelical packing, which accounts for the phenotypes observed with the chimeras, and provide insights into the mechanism(s) of G-protein-coupled receptor activation. First, it is of interest that the Ala 204 3 Val constitutively activating mutation, as well as a previously characterized constitutively activating ␣ 1b -AR mutation, Cys 128 3 Phe, involve transmembrane residues that are located approximately one helical turn above or below residues putatively involved in forming critical interactions with agonists. Thus, in the fifth transmembrane domain the Ala 204 is one helical turn above Ser 207 , which hydrogen-bonds with the catechol hydroxyl of phenethylamines; and in the third transmembrane domain Cys 128 is one helical turn below Asp 125 , which forms a salt bridge with the protonated amine of agonists. This suggests that these mutations alter the conformation of the third and fifth transmembrane segments, or the relative positioning of the helices, leading to an enhanced propensity for the receptor to isomerize spontaneously to the R* state, as evidenced by increased basal signal transduction, an agonist-independent manifestation of constitutive activity. However, this conformational change may also facilitate the bonding of critical residues such as Ser 207 and Asp 125 with agonist, thus leading to agonist-dependent manifestations of constitutive activity such as high affinity binding and increased potency. Second, the constitutive activity of the Met 292 3 Leu ␣ 1a chimera, suggests that a conformational change in the sixth transmembrane segment may also be involved in receptor signaling. However, this may be indirect through an interaction with the fifth helical segment, which then also influences agonist bonding with Ser 207 , since mutation of Val 185 in the fifth transmembrane segment to Ala relieves the constitutive activity of Met 292 3 Leu.
Based on these considerations, we propose that the constitutive activity of the Ala 204 3 Val is not due to the loss of alanine, which in the native receptor structure may be postulated to act as a repressor of active state isomerization. Rather, the valine, either because of its bulk or ␤-substituent, prevents normal side-chain packing between the fifth and the sixth helical segments. As a result there is a conformational distortion of the fifth helical segments that mimics the active state conformation and allows enhanced interactions between Ser 207 and the agonist (Fig. 7). With the additional substitution of Leu 314 by a methionine residue, this conformational distortion is relieved and the receptor is no longer constitutively active. Similarly, with the native ␣ 1a -AR there is normal packing between the fifth and the sixth helical segments, despite the presence of the wild type valine at position 185, since the native methionine at position 292 allows the valine to be accommodated. However, with a leucine at position 292, as occurs in the Met 292 3 Leu mutation, valine at position 185 now induces constitutive activation, since normal interhelical packing is prevented. Finally, FIG. 6. Dose-response curves of IP and arachidonic release for the single and double ␣ 1a -mutants using (؊)-epinephrine as the agonist. Panel A, dose-response curve for IP production using varying concentrations of (Ϫ)-epinephrine. Each data point represents the mean Ϯ S.E. derived from eight separate experiments. With the Met 292 3 Leu mutation, there was a shift to the left in the dose-response curve. However, with the double mutation, Met 292 3 Leu/Val 185 3 Ala, the dose-response curve was now similar to that observed for the wild type ␣ 1a -AR. Panel B, dose-response curve for arachidonic acid production using varying concentrations of (Ϫ)-epinephrine. Each data point represents the mean Ϯ S.E. derived from eight separate experiments. Similar to panel A, with the Met 292 3 Leu mutation there was a shift to the left in the dose-response curve. However, with the double mutation, Met 292 3 Leu/Val 185 3 Ala, the dose-response curve was now similar to that observed for the wild type receptor. Panels C and D, maximal IP (C) and arachidonic acid (D) release observed with (Ϫ)-epinephrine stimulation of the wild type ␣ 1a and ␣ 1b receptors and their corresponding single and double mutants. Data were generated as described in panels A and B and in Fig. 5. The maximum responses for the mutants were not significantly different (using an ANOVA) from that observed with their respective wild type receptors. this latter steric constraint can be relieved with the substitution of Val 185 by alanine.
Recently, mutational analysis of the tachykinin NK-1 receptor revealed that replacement of two residues located at the extracellular surface of the fifth and sixth transmembrane domains (Glu 193 and Tyr 272 , respectively) with histidine, enabled Zn 2ϩ ions to be coordinated and to act as an antagonist (17). Although the replacement of the wild type residues with histidine may have resulted in altered helical conformations allowing Zn 2ϩ coordination, it is of interest that Glu 193 and Tyr 272 are exactly one helical turn above Ala 204 (␣ 1b ) or Val 185 (␣ 1a ), and Leu 314 (␣ 1b ) or Met 292 (␣ 1a ), respectively. Thus this finding with the NK-1 receptor provides support for our contention of an important interhelical interaction between the fifth and sixth transmembrane domains. In addition, if the rotation of helix V is transfixed such that Ser 207 of the ␣ 1b -AR (or Ser 188 of the ␣ 1a -AR) faces the ligand binding pocket to allow coordination with agonist, then, as suggested by Liu et al. (18), the helices, including helix V and VI, have to be arranged in a counterclockwise orientation, as viewed from the extracellular surface. Arrangement in a clockwise orientation is precluded, since if Ala 204 and Leu 314 (␣ 1b -AR) or Val 185 or Met 292 (␣ 1a -AR) have to interact, this would result in Ser 207 facing away from the ligand-binding pocket toward the lipid bilayer. If one now considers the interhelical connectivity suggested by this study (helix V with VI), as well as the NK-1 receptor study (17), in addition to the connectivity between helices II and VII, I and VII, and III and VII, demonstrated from studies of the gonadotrophin-releasing hormone receptor (19), the muscarinic acetylcholine receptors (18), and rhodopsin (20), respectively, then it follows that the helices must be arranged, not only in a counterclockwise direction, but also such that the helices are juxtaposed in the order I, II, III, etc., to VII, rather than in some other order, as suggested by Pardo et al. (21) and Zhang and Weinstein (22).
At present the only available structural data on receptor activation (23), which is of low resolution, comes from studies of bacteriorhodopsin. These data implicate conformational movement of the sixth and seventh transmembrane segments. However, this receptor is not G-protein-coupled. In support of our notion that the receptor activation involves conformational movement of the third, fifth, and possibly the sixth transmembrane domains is the finding that both the third and second intracellular loops, which are contiguous with these transmembrane segments, are involved in G-protein activation. In addition, recent spin labeling data from Khorana and Hubbell (24) indicate that photoexcitation of rhodopsin, a member of the G-protein-coupled receptor family, involves rigid movement of the third helix relative to the others in the ligand binding helical bundle.
The findings of this study also provide potential insights into the evolution of receptor structure. This is evident from the fact that the wild type receptors show little, if any, agonist-independent effector activation, yet mutation of a single residue can result in constitutive activation. This suggests that the native structures have evolved to select residues that repress active state isomerization. Furthermore, if a mutation did occur that introduced such a residue, the resulting constitutively active receptor may have been deleterious to the organism. Propagation of this receptor structure would then only be possible if a simultaneous mutation occurred in the complementary residue in the adjacent transmembrane segment, which could serve as a surrogate "molecular brake" to suppress constitutive activity. Although such a double mutant would have an altered agonist binding profile when evaluated with a variety of synthetic ligands, as observed here with the ␣ 1 chimeras, binding of the natural ligands, (Ϫ)-epinephrine and (Ϫ)-norepinephrine, would be unchanged. This finding, as well as the fact that the ␣ 1b -and ␣ 1a -AR appear to activate all effector pathways similarly, suggests that from the point of view of the organism, such a double mutant would be phenotypically silent. Indeed, based on such considerations, one can postulate that such an evolutionary mechanism may underlie subtype diversity for some classes of G-protein-coupled receptors, such as the ␣ 1 -AR. Our ability to discriminate these subtypes may, thus, be due to the development of synthetic ligands that can detect subtle differences between the ligand recognition sites of these subtypes, which can be exploited to develop subtype selective drugs, particularly as different subtypes have been shown to possess different functions, even when expressed in the same tissue. However, further studies will be required, particularly with other classes of G-protein-coupled receptors to confirm or refute this postulate. FIG. 7. A schematic diagram outlining the concept of "molecular repressors and inducers." In the wild type ␣ 1b -AR at position 204, there is an alanine that does not modulate activity. When replaced to a valine, the valine either because of its bulk or ␤-substituent, prevents normal side-chain packing between the fifth and the sixth helical segments. As a result there is a conformational distortion of the fifth helical segments that mimics the active state conformation. With the additional substitution of Leu 314 by a methionine residue, this conformational distortion is relieved and the receptor is no longer constitutively active. Similarly, with the native ␣ 1a -AR there is normal packing between the fifth and the sixth helical segments, despite the presence of the wild type valine at position 185, since the native methionine at position 292 allows the valine to be accommodated. However, with a leucine at position 292, as occurs in the Met 292 3 Leu mutation, valine at position 185 now induces constitutive activation, since normal interhelical packing is prevented. Finally, this later steric constraint can be relieved with the substitution of Val 185 by alanine. Thus the valine at this position in both the ␣ 1a -AR and ␣ 1b -AR acts as a molecular inducer by perturbing conformation, whereas a methionine in transmembrane VI functions as a repressor, reverting the valineinduced conformational change back to normal. TMV, transmembrane V; TMVI, transmembrane VI.