Phe310 in Transmembrane VI of the α1B-Adrenergic Receptor Is a Key Switch Residue Involved in Activation and Catecholamine Ring Aromatic Bonding*

Pharmacophore mapping of adrenergic receptors indicates that the phenyl ring of catecholamine agonists is involved in receptor binding and activation. Here we evaluated Phe310, Phe311, and Phe303 in transmembrane VI (TMVI), as well as Tyr348 in TMVII of the α1B-adrenergic receptor (α1B-AR), which have been implicated in a catechol-ring interaction. Neither catecholamine docking studies nor mutagenesis studies of Phe311, Phe303, or Tyr348 supported a role for these residues in catechol-ring binding. By contrast, docking studies indicated that the Phe310 side chain is well positioned to interact with the catechol-ring, and substituted cysteine accessibility method studies revealed that the side chain of the 310, but not 311 residue, is both solvent accessible and directed into the agonist-binding pocket. Also, saturation mutagenesis of both Phe310 and Phe311 revealed for the former, but not for the latter, a direct relationship between side chain volume and agonist affinity, and that aromaticity is essential for wild-type agonist binding, and for both wild-type agonist potency and efficacy. Moreover, studies of Phe310 mutants combined with a previously described constitutively active α1B-AR mutant, A293E, indicated that although not required for spontaneous receptor isomerization from the basal state, R, to a partially activated conformation R′, interaction of Phe310 with catecholamine agonists is essential for isomerization from R′ to the fully activated state,R*.

␣ 1 -Adrenergic receptors (␣ 1 -AR) 1 are members of the heptahelical superfamily that share a common structural motif of seven putative ␣-helical transmembrane spanning regions linked by three extra-and three intracellular loops, an extracellular N terminus and intracellular C-terminal tail. Transmembrane signaling by all ␣ 1 -AR subtypes (␣ 1A , ␣ 1B , and ␣ 1D ) in response to the natural catecholamine agonists, norepinephrine and epinephrine, is mediated by G-proteins of the G q/11 family or in some instances, the G h family of tissue transglutaminases (1,2). Based on certain key structural features, ␣ 1 -ARs are more closely related to rhodopsin or the type A subfamily of GPCRs that includes ␤-ARs, than to the calcitonin (type B) or metabotropic (type C) subfamilies.
Binding of catecholamines by both ␣ 1 -and ␤-ARs involves the formation of a salt bridge between the basic aliphatic nitrogen atom common to all sympathomimetic amines and an aspartate (Asp 125 in the hamster ␣ 1B -AR; Asp 113 in the hamster ␤ 2 -AR) in the third transmembrane segment (TMIII) (3,4). With rhodopsin, light induced isomerization of the retinal chromophore leads to deprotonation of a Schiff base linking the chromophore to Lys 296 in TMVII (5). In the ground state the protonated Schiff base is ionically bonded to a TMIII acidic residue (Glu 113 ) that is equivalent to Asp 125 and Asp 113 in the ␣ 1B and ␤ 2 -ARs, respectively (5). With the ␣ 1B -AR there is evidence that receptor activation also is due to disruption of an ionic interaction between the TMIII aspartate and a TMVII lysine (Lys 331 ), due to competition between the catechol protonated amine and the TMVII lysine, for binding to the TMIII aspartate (3). The TMIII aspartate thus serves as a counterion and most likely is an important residue for agonist binding and activation of all adrenergic receptors.
For the ␤ 2 -AR, two serine residues, Ser 204 and Ser 207 , which are conserved in most adrenergic receptors, have been proposed to hydrogen bond with the meta-and para-hydroxyls, respectively, of the catechol ring (6). Mutation of either serine to an alanine results in a 30-fold decrease in affinity for catecholamine agonists, and each serine contributes about 50% to receptor activation. Thus, binding of both catechol hydroxyls is required for full agonist activity. By contrast, agonist binding to the ␣ 1A -AR involves an interaction between the meta-hydroxyl and Ser 188 (equivalent to Ser 203 , not Ser 204 in the ␤ 2 -AR) that plays a major role in receptor activation, being responsible for 70 -90% of the wild-type response. An interaction between the para-hydroxyl and Ser 192 (equivalent to Ser 207 in the ␤ 2 -AR), on the other hand, contributes minimally to receptor activation (7). Moreover, since the interacting serines in the ␣ 1A -AR are separated by four residues, whereas those in the ␤ 2 -AR are separated by only three residues, docking of the catecholamine ring is in a more planar orientation in the ␣ 1A -AR, and is rotated by about 120°to that in the ␤ 2 -AR.
This altered catechol ring orientation may also contribute to other agonist docking differences between ␣ 1 -and ␤-ARs. For example, stereoselectivity of binding and activation has been attributed, in part, to a hydrogen bond interaction between the chiral benzylic hydroxyl group attached to the ␤-carbon atom of catecholamines, and Asn 293 in TMVI of the ␤ 2 -AR (8). Although stereoselectivity of catecholamine binding and activation is preserved with ␣ 1 -ARs, the determinants of stereoselectivity have not been defined, and the Asn 293 equivalent is replaced by a residue (leucine or methionine) lacking hydrogen-bonding potential. This finding again provides evidence that some of the catecholamine-binding and activation residues in ␣ 1 -ARs are distinct from those in ␤-ARs.
Previous studies of the hamster ␤ 2 -AR suggested that a phenylalanine in TMVI (Phe 290 , equivalent to Phe 311 in the ␣ 1B -AR, see Fig. 1), which is conserved only in biogenic aminebinding GPCRs, is involved in forming an aromatic-aromatic interaction with the phenyl ring of catecholamines (9). This conclusion was based on the finding of a 10-fold decrease in agonist, but not antagonist, binding with mutation of Phe 290 to methionine. However, no additional studies were performed to exclude a nonspecific global or local change in receptor structure with this Phe 290 mutation, or to evaluate the role of the potential aromatic-aromatic interaction in receptor activation. In addition, substitution of an adjacent TMVI phenylalanine (Phe 289 ; equivalent to Phe 310 in the ␣ 1B -AR) to alanine, resulted in a 1000-fold decrease in agonist affinity with no change in antagonist binding. Finally, Tyr 326 in TMVII was also suggested to potentially be involved in a catechol ring interaction, since substitution of this residue with leucine decreased agonist, but not antagonist binding by 10-fold (9).
Here we show, based on macromolecular modeling studies, in which the planar orientation of the phenyl ring in ␣ 1 -ARs was taken into consideration when docking catecholamines, that interaction with the phenyl ring involves Phe 310 in TMVI and not Phe 311 in TMVI, or Tyr 348 in TMVII (equivalent to Tyr 326 in the ␤ 2 -AR). Furthermore, based on mutagenesis studies coupled with the evaluation of group-specific catecholamine analogs, as well as SCAM (substituted cysteine accessibility method) studies, we provide evidence that Phe 310 is critically involved both in forming an aromatic-aromatic interaction with the catecholamine phenyl ring and in receptor activation.

EXPERIMENTAL PROCEDURES
Materials-(Ϫ)-Epinephrine, (Ϫ)-norepinephrine, (Ϯ)-synephrine, (Ϯ)-halostachine, dopamine, phenethylamine, prazosin, phentolamine hydrochloride, lithium chloride, and dl-propranolol were purchased from Sigma. 5 Site-directed Mutagenesis-The construct used was the hamster ␣ 1B -AR cDNA with an octapeptide tag (1D4) at the end of coding region in the modified eukaryotic expressing vector, pMT2Ј (10). The presence of the 1D4 epitope at the C-terminal of ␣ 1B -AR does not affect its expression and function (data not shown). Site-directed mutagenesis was performed as described previously (2,11). Briefly, two primers, one carrying the nucleotide change(s) to produce the desired amino acid mutation in the ␣ 1B -AR sequence, the other carrying the nucleotide changes to convert a unique restriction site (ClaI) to another restriction site (NarI) in a non-essential region of the vector, pMT2Ј, were simultaneously annealed to the denatured template and a new second strand DNA containing both primers was synthesized by treatment with T4 polymerase. The DNA was then digested with ClaI to linearize reannealed parental plasmid and the reaction mixture used to transform Escherichia coli cells (BMH 71-18 mutS cells). Transformants were grown en mass in liquid media and used to isolate plasmid DNA. The resulting DNA was digested with ClaI again to linearize remaining parental plasmid, and transformed into DH5␣-cells. Transformants were plated and plasmid DNA prepared and sequenced to confirm the presence of the desired mutation.
Western Blotting-Membranes (50 g of protein) were dissolved in 1% CHAPS and SDS sample buffer overnight at 4°C, and then subjected to SDS-polyacrylamide gel electrophoresis, as described previously (2,12). The resolved proteins were electroblotted onto Immobilon-P membranes and then immunostained for detection using the ECL chemiluminescence system (Amersham), as described previously (2). ␣ 1B -AR was detected using a monoclonal antibody against the 1D4 epitope (12).
Ligand Binding-The ligand binding characteristics of the membrane expressed receptors were determined in a series of radioligand binding studies performed exactly as described previously (2, 10), using [ 125 I]HEAT, an ␣ 1 -specific antagonist, as the radioligand. For saturation binding studies, the concentrations of [ 125 I]HEAT used ranged from 10-fold below to 10-fold above the K d value, whereas for competition studies, a concentration near the K d of the radioligand value was used. K i values were determined using the Cheng-Prusoff equation (15). The membrane concentration used in these studies was selected to allow binding of less than 10% of the total radioligand added. To avoid interassay differences replicate studies with the wild-type ␣ 1B -AR were performed with the analysis of each mutant. Binding data were analyzed using the iterative, nonlinear, curve-fitting program, Prism. For comparison of the fit to a one-site or two-site model, the F test was used, p Ͻ 0.05 was considered to be statistically significant.
Reaction with MTSEA and Binding Assays in Intact Cells-Transfected COS-1 cells were harvested by trypsinization. After washing with phosphate-buffered saline, cells were resuspended in 1.4 ml of HEPES buffer (140 mM NaCl, 5.4 mM KCl, 1 mM EDTA, 0.006% bovine serum albumin, 25 mM HEPES, pH 7.4) as described previously (16). Aliquots FIG. 1. Sequence alignment of TM VI residues of hamster ␣ 1B and other G protein-coupled receptors. Sequences were aligned to maximize homology within this region using the GCG program "Pileup." The conserved phenylalanines corresponding to Phe 310 and Phe 311 in the hamster ␣ 1B -AR are shaded and in bold type, while Phe 303 , which is highly conserved among all G proteincoupled receptors is boxed. The dashed line at the top delineates the transmembrane residues of helix VI.
(80 l) of the cell suspension were incubated with 20 l of freshly prepared MTSEA at the stated concentrations at room temperature for 2 min. Cell suspensions were then diluted 20-fold, and 100-l aliquots were used to assay for [ 125 I]HEAT (600 pM) binding in the presence or absence of 0.1 mM phentolamine in a total volume of 250 l in triplicate. The result was analyzed as described above.
Construction of an ␣ 1B -AR Molecular Model and Catecholamine Docking-The coordinates of the ␣-carbon positions were determined by overlay of putative ␣ 1 -AR transmembrane residues with the transmembrane coordinates of bacteriorhodopsin (17), with data files generated using the Insight II molecular modeling software from Biosym Technologies. The boundaries of the putative transmembrane domains were determined using an algorithm based on the weighted pairwise comparisons of adjacent residues (18). The positioning of each helix with respect to the adjacent helices was based where possible on data from ␣ 1B -AR mutagenesis studies (10,18,19). The projections of the helices proposed by Baldwin (21) were used to determine the tilt of each helix. The model was minimized and conflicts adjusted to remove steric clashes based on dynamic runs, as described previously (22).
Phosphatidylinositol Hydrolysis in Intact Cells-Phosphatidylinositol (PI) hydrolysis in intact, transfected COS-1 cells was determined exactly as described previously (2), except that for determining basal activation of PI hydrolysis, the cells were seeded onto 6-well plates 1 day after transfection. Results are expressed as the mean Ϯ S.E. (error bars). An analysis of variance and the Student's t test were used to determine significant differences (p Ͻ 0.05).

RESULTS AND DISCUSSION
A model of the ␣ 1B -AR developed previously (22) was refined to accommodate the findings of recent mutagenesis studies and used to dock the agonist, (Ϫ)-epinephrine, into the binding pocket. In particular the refined model takes into account an interhelical stacking interaction we have identified between Ala 204 in TMV and Leu 314 in TMVI (10,20), as well as the planar orientation of the catechol ring. As shown in Fig. 2A, from this model it is evident that the TMVI Phe 310 side chain is well positioned to interact with the catechol ring in a parallel stacked and displaced conformation, which is an energetically favored structure for benzene dimers (23). The TMVI Phe 311 side chain, on the other hand, is directed toward TMV, or is projecting into the lipid bilayer, and Tyr 348 is located toward the intracellular end of TMVII, well below the plane of the catecholamine ligand.
To more directly evaluate the involvement of Phe 310 in catechol ring binding, it was mutated to alanine or leucine, and the resulting mutants (F310A and F310L), as well as the wildtype ␣ 1B -AR, were then evaluated in terms of membrane expression, ligand binding, and stimulation of PI hydrolysis. As controls for the potential Phe 310 -catechol ring interaction, alanine and leucine mutants of two other TMVI phenylalanines, Phe 311 and Phe 303 , and a leucine mutant of Tyr 348 in TMVII, were also constructed and similarly evaluated (Fig. 2B). Saturation binding and immunoblotting studies indicated that the F303A, F303L, and Y348L mutants were expressed in the plasma membrane and processed (glycosylation) at levels almost equal to the wild-type ␣ 1B -AR (Table I and Fig. 3). All three mutants showed only small decreases (1.7-2.2-fold) in affinity for the antagonist radioligand, [ 125 I]HEAT (Table I) and no change (Y348L) or an increase (5-20-fold, F303A and F303L) in affinity for the agonists, norepinephrine, epinephrine, and phenylephrine (Fig. 4). As will be reported in a subsequent paper, 2 the increased agonist affinities observed with the Phe 303 mutants are due to the fact that these substitutions result in constitutive receptor activation. Taken together, therefore, these findings do not support involvement of Phe 303 or Tyr 348 in an interaction with the catechol ring.
As shown in Table II, the Phe 310 mutants showed binding affinities for the antagonists, phentolamine, prazosin, and 5-methylurapidil that were 1.4 -44-fold less than observed with the wild-type receptor. However, their affinity for the radioligand [ 125 I]HEAT was largely unchanged (Table I) and their plasma membrane expression was equivalent to that of wild-type receptor (Table I and Fig. 3).
The Phe 311 mutants also displayed decreased antagonist binding (5-94-fold, Table II). However, in contrast to the Phe 310 mutants, plasma membrane expression was markedly impaired with both F311A and F311L, and could not be enhanced by increasing the amounts of plasmid transfected (Table I and Fig. 3). These findings suggest that whereas substitution of Phe 310 with alanine or leucine results in the loss of receptor interactions involved in the binding of some but not all antagonists, mutation of Phe 311 produces a global change in receptor conformation that impairs not only ligand binding, but also protein folding and membrane expression.
To further characterize the effects of alanine or leucine substitution at the Phe 310 and Phe 311 positions, the binding of group-specific agonists was evaluated. The agonists evaluated included the natural catecholamines, (Ϫ)-norepinephrine, and (Ϫ)-epinephrine, as well as (Ϫ)-phenylephrine, which is also a full agonist but lacks the para-hydroxyl at the 4-position of the catechol ring, and a group of partial agonists: (Ϯ)-synephrine, (Ϯ)-halostachine, phenethylamine, and dopamine, which lack a meta-hydroxyl at the 3-position of the catechol ring; both catechol ring hydroxyls; both catechol ring hydroxyls and the chiral hydroxyl on the ␤-carbon; or only the chiral ␤-hydroxyl, respectively. As shown in Fig. 4A, the F310A mutant showed affinity losses of up to 1000-fold for the full agonists, but smaller losses (10 -50-fold) for the partial agonists. Consistent with the involvement of Phe 310 in ligand binding, the F310L mutant in which the hydrophobicity but not the aromaticity of the wild-type phenylalanine residue is preserved, displayed lesser decreases in binding of both full agonists (30 -40-fold) and partial agonists (3-10-fold), than the F310A mutant (Fig. 4B).
As shown in Table III, the free energy change (⌬⌬G) observed for the full agonists, with substitution of Phe 310 with alanine, was close to 4 kcal/mol, but only about 2 kcal/mol for the partial agonists. For the leucine substitution the free energy change in agonist binding was approximately 2 kcal/mol for the full agonists, and Ͻ1 kcal/mol for the partial agonists. Given that the theoretical bond energy of an aromatic-aromatic interaction is approximately 2 kcal/mol (23), one interpretation of these findings is that the leucine substitution, because of the hydrophobic character of its side chain, but not the alanine substitution, which lacks both hydrophobicity and aromaticity, partially compensates for the wild-type phenylalanine. The greater free energy loss observed for full agonists with the alanine substitution suggests that the aromatic component of the Phe 310 interaction with the catechol ring is not only essential for ligand binding, but for critical positioning of the catechol ring to allow optimal interaction of other ligand moieties with the receptor, such as the catecholamine meta-hydroxyl and Ser 207 . With partial agonists, which lack at least one of the critical moieties of full agonists and, thus, most likely have a less constrained binding geometry, positioning of the catechol ring by an aromatic interaction with the Phe 310 side chain may be less critical. As a result, the free energy losses for partial agonists are less than those for full agonists, and almost exactly those anticipated (2 kcal/mol) for the loss of a single aromatic-aromatic interaction.
Like the Phe 310 mutants, the Phe 311 mutants also displayed decreased affinity for both full and partial agonists (Fig. 4). However, unlike the Phe 310 mutants, the loss of agonist affinity with leucine substitution of Phe 311 was greater than with ala- The receptor is modeled as it would appear looking down onto the membrane from the extracellular side. White circles with Roman numerals indicate the respective transmembrane helices. The projection of the helices from the extra-to intracellular surface of the membrane is indicated by the yellow cylinders. Epinephrine is shown interacting via its protonated amine with Asp 125 in TMIII; via the catechol ring meta-hydroxyl with Ser 207 in TMV, and via the catechol ring with Phe 310 in TMVI. Previously identified interactions between Asp 125 in TMIII and Lys 331 in TMVII (3), and Ala 204 in TMV and Leu 314 in TMVI (10,19), as well as the orientation of Phe 311 are shown. B, secondary structure of the hamster ␣ 1B -AR indicating the location of the native cysteine residues, including the putative, solvent inaccessible disulfide-linked extracellular pair, Cys 119 and Cys 195 (38), and the residues (Phe 303 , Phe 310 , Phe 311 , and Tyr 348 ) evaluated in this study. nine for the full agonists (up to 1000-fold versus 50 -100-fold), but the same or less for partial agonists (5-10-fold versus 8 -25-fold). Thus, the loss of both aromaticity and hydrophobicity with alanine substitution of Phe 311 could not be compensated by a residue with a hydrophobic side chain (leucine) and, in fact, such a residue caused a greater impairment of agonist binding. Together with the markedly decreased membrane expression observed for the Phe 311 mutants, these findings indicate a strict requirement for the phenylalanine side chain to allow proper folding into the wild-type receptor conformation, perhaps due to a critical interhelical stacking interaction between Phe 311 and residues in the TMV helix.
To further evaluate the postulate that Phe 310 forms a critical aromatic-aromatic interaction with the catechol ring, whereas Phe 311 is not directly involved in ligand binding but rather in global receptor structure, additional Phe 310 and Phe 311 mutants were constructed and their agonist-binding properties evaluated. As seen in Fig. 5, the Phe 310 mutants produced graded decreases in agonist affinity. Whereas the affinity for epinephrine, phenylephrine, and halostachine was up to 1000fold lower with the F310A mutant than with the wild-type receptor, substitution of Phe 310 with a tryptophan, which although slightly larger than the native phenylalanine, preserves both its hydrophobicity and aromaticity, resulted in a receptor protein with near wild-type agonist affinities. Substitution with a tyrosine, which also has an aromatic side chain, however, resulted in decreases in agonist affinity (25-30-fold) that were comparable to those observed with the F310L mutation. This may be due to the fact that the tyrosine ring contains an hydroxyl moiety. As a result, its side chain, unlike those of phenylalanine or tryptophan, is unlikely to be planar, and also has H-bonding potential. Thus, a tyrosine at the 310 position may perturb optimal positioning of the catechol ring due either to a steric clash or to loss of a favorable planar stacking interaction. Other substitutions at the 310 position with smaller hydrophobic or ␤-branched residues (valine and isoleucine) or polar residues (asparagine) resulted in significantly greater reductions in agonist affinity than did tyrosine, leucine, or phenylalanine. Not surprisingly, therefore, a positive and highly significant correlation was evident between the volume of the substituent side chain and agonist affinity (Fig. 5). This correlation indicates a van der Waals component to the bonding between the catechol ring and the side chain at the 310 posi-  tion. This is not inconsistent with an aromatic-aromatic interaction, which involves both a dipole-dipole and a van der Waals component (23). In keeping with an aromatic-aromatic interaction is the finding that the wild-type receptor and the F310W mutant, which both have a planar aromatic side chain at the 310 position, bind with considerably higher affinity than the F310L mutant. In the case of this latter mutant the 310 side chain is of similar volume to phenylalanine, but rather than being aromatic and planar, is aliphatic and bulky.
In contrast to the Phe 310 mutations, no correlation was observed between the volume of the substituent side chain at the 311 position and agonist affinity (Fig. 5). Thus, small polar residues, such as asparagine, produced lesser decreases in affinity than larger hydrophobic residues, such as leucine or isoleucine.
If Phe 310 indeed interacts with the catechol ring, as suggested by the above findings, it should be solvent accessible with its side chain projecting into the agonist-binding pocket. By contrast, Phe 311 , which we speculate projects toward TMV, should be much less solvent accessible. To directly evaluate the solvent accessibility of Phe 310 and Phe 311 , we constructed the cysteine mutants, F310C and F311C, and used the SCAM (16,24) to test their sensitivity to derivatization with the polar cysteine-modifying reagent, MTSEA. This compound is 2500 times more soluble in water than in n-octanol, fits into a cylinder of about 0.6 nm in diameter by 1 nm in length, and specifically adds its -SCH 2 CH 2 NH 3 ϩ moiety to reduced sulfhydryls to form mixed disulfides (24). Given the size of the -SCH 2 CH 2 NH 3 ϩ moiety, derivitization of a residue projecting into the binding pocket should sterically hinder ligand binding, whereas derivitization of a residue projecting toward an adjacent helix or into the lipid bilayer will occur much more slowly, and will either not influence ligand binding or will produce a more global change in receptor structure.
To evaluate the accessibility of Phe 310 and Phe 311 to MTSEA derivitization, we initially evaluated the accessibility of the 16 native cysteines (Fig. 2B) in the wild-type ␣ 1B -AR. As shown in Fig. 6, MTSEA modification of intact COS-1 cells expressing the wild-type ␣ 1B -AR irreversibly inhibited binding of the ra-dioligand [ 125 I]HEAT. This inhibition could be rescued by treatment with the reducing agent, dithiothreitol, and could be prevented by pretreatment with the ␣ 1 -AR antagonist, phentolamine (data not shown). This indicates that MTSEA specifically modified cysteine residues rather than producing a nonspecific disruption of receptor structure. As will be detailed in a subsequent paper, 2 mutagenesis of three native cysteines (Cys 128 , Cys 129 , and Cys 137 ) to serines, significantly reduced both the sensitivity and reactivity of the ␣ 1B -AR to MTSEA modification (Fig. 6, A and B). Importantly, the triple mutant (C128S/C129S/C137S) displayed similar binding affinities for various ␣ 1 -AR antagonists ([ 125 I]HEAT and prazosin) and agonists (epinephrine and cirazoline), and similar activity in stimulating PI hydrolysis, as the wild-type receptor (Table IV). This triple mutant was thus used as a template for SCAM studies of F310C and F311C.
Substitution of Phe 310 or Phe 311 with cysteine produced only a small decrease (3-15-fold) in [ 125 I]HEAT affinity, which was not further impaired when either F310C or F311C was combined with the cysteine triple mutant to produce the quadruple mutants F310C/C128S/C129S/C137S and F311C/C128S/ C129S/C137S (Table V). By contrast with the cysteine triple mutant (IC 50 for MTSEA-induced inhibition of [ 125 I]HEAT binding, Ͼ10 mM) additional substitution of Phe 310 to cysteine in the mutant, F310C/C128S/C129S/C137S, restored sensitivity to MTSEA (IC 50 0.98 Ϯ 0.23 mM, p Ͻ 0.001 as compared with the triple mutant) toward that observed with the wildtype receptor (0.33 Ϯ 0.06 mM) (Fig. 6A). However, sensitivity to MTSEA was not restored with additional substitution of Phe 311 to cysteine in the mutant, F311C/C128S/C129S/C137S (IC 50 Ͼ 10 mM) (Fig. 6A). In agreement with these findings, the reactivity of Phe 310 to modification with MTSEA in the F310C/ C128S/C129S/C137S mutant was rapid and identical to that observed with the wild-type receptor (t1 ⁄2 ϭ 1-3 s; Fig. 6B). By contrast, the reaction kinetics for MTSEA modification of Phe 311 in the F311C/C128S/C129S/C137S mutant were extremely slow and identical to those observed with the triple mutant, C128S/C1295/C137S (t1 ⁄2 Ͼ 600 s) (Fig. 6B). Nevertheless, given that MTSEA-induced receptor inactivation was evaluated with the antagonist, [ 125 I]HEAT, and given that the residues involved in antagonist and agonist binding may not be identical, these findings do not exclude the possibility that Phe 310 forms part of the antagonist, but not agonist-binding pocket. To address this issue, receptor-protection studies were performed to evaluate if agonist could protect against MTSEA inactivation. As shown in Fig. 6C, (Ϫ)-epinephrine fully protected both the wild-type receptor and the F310C/C128S/ C129S/C137S mutant. In contrast, the small amount of inhibition of [ 125 I]HEAT binding observed with MTSEA treatment of both the F311C/C128S/C129S/C137S and C128S/C129S/C137S mutants was unaltered by (Ϫ)-epinephrine pretreatment (Fig.  6C). Thus, the effect of MTSEA on [ 125 I]HEAT binding to the F311C/C128S/C129S/137S mutant is not due to derivatization   Fig. 5, D-F) than that of the charged MTSEA moiety, -SCH 2 CH 2 NH 3 ϩ (volume ϭ 283 Å 3 ), markedly perturbs [ 125 I]HEAT binding (Fig. 5, D-F). Taken together therefore, these data indicate that the side chain of Phe 310 , but not Phe 311 , is both solvent accessible and directed toward the agonist-binding pocket.
To determine the role of Phe 310 and Phe 311 in receptor activation, the ability of substitution mutants to mediate agonist-stimulated PI hydrolysis was compared with that of the wildtype ␣ 1B -AR. With the F310L mutant, the maximal response was unaltered but the potency of the epinephrine-stimulated PI response was decreased by about 10-fold (EC 50 ϭ 0.57 M versus 0.08 M for the wild-type ␣ 1B -AR), which is comparable to its decrease in epinephrine binding affinity (Fig. 7A). However, the F310A mutant showed a significant decrease not only in potency (EC 50 ϭ 120 M versus 0.08 M for the wild-type ␣ 1B -AR), but also in efficacy of epinephrine-stimulated PI hydrolysis (E max ϭ 60% of the wild-type response) (Fig. 7A). This decrease in agonist efficacy was also observed with two other catecholamine analogues, halostachine and phenethylamine, which lack either the catechol para-hydroxyl, or both catechol hydroxyls and the chiral hydroxyl on the ␤-carbon, respectively (Fig. 7, B and C). Like the Phe 310 mutants, alanine and leucine substitutions at the 311 position also affected (Ϫ)-epinephrinestimulated PI hydrolysis. However, in this case the leucine substituent produced a greater pertubation of receptor signaling (decreased potency, EC 50 ϭ 219 M versus 0.08 M for the wild-type ␣ 1B -AR, and efficacy, E max ϭ 40% of the wild-type response) than the alanine substituent (decreased potency only, EC 50 ϭ 4.9 M). Together with the alterations in agonist binding observed above with these Phe 310 and Phe 311 mutants, the data can be interpreted to indicate that at the 310 position, substitution of the native phenylalanine with a hydrophobic residue, such as leucine, can partially compensate for the loss of both aromaticity and hydrophobicity associated with an alanine substitution. At the 311 position, however, the greater perturbation of receptor signaling with leucine than with alanine may be due to the bulky, non-planar leucine side chain disrupting interhelical packing. In keeping with the requirement of an aromatic side chain at the 310 position for both agonist binding and receptor activation, is the finding that a mutant in which Phe 310 was replaced with another aromatic residue, tryptophan, displayed not only near wild-type agonist affinities, but also wild-type PI hydrolysis (Fig. 7, A-C).
Finally, to confirm that the effects of the F310A and F310L mutations on PI signaling are not due to a global change in receptor structure, we combined them with a previously described constitutively active mutant, A293E (25,26), to produce the double mutants, F310A/A293E and F310L/A293E. Since Ala 293 is about four helical turns below Phe 310 , it is possible that a structural change caused by the alanine or leucine substitutions will be transmitted to Ala 293 , thus altering its structure and constitutive signaling activity. As shown in Fig.  8, signaling in the absence of agonist by the F310A or F310L mutants was not different from that observed with the wildtype ␣ 1B -AR. Thus, F310A and F310L are not constitutively active. Moreover, increased basal signaling by the A293E mutant, which is dependent on its level of expression, was not altered with the double mutants, F310A/A293E or F310L/ A293E (Fig. 8, inset). As seen in Fig. 9A, compared with the wild-type ␣ 1B -AR, the A293E mutant showed all the hallmarks of a constitutively active receptor, viz, increased basal PI signaling, and a decreased EC 50 and increased E max for epinephrine-stimulated PI hydrolysis. Both double mutants showed similar increases in basal PI signaling to that observed with the A293E mutant, alone. However, like the single mutant,  Consistent with these changes in PI signaling, the increased epinephrine affinity observed with the A293E mutant (K i ϭ 0.017 M) compared with the wild-type ␣ 1B -AR (K i ϭ 0.75 M), was perturbed by additional substitution of Phe 310 with leucine in the F310L/A293E double mutant (K i ϭ 1.84 M), and further decreased in the F310A/A293E double mutant (K i ϭ 64 M) (Fig. 9B). Thus, the effects of alanine or leucine substitution of Phe 310 on PI signaling do not suggest distortion of global receptor structure, but are entirely consistent with the involvement of Phe 310 as a key switch residue involved in ligand binding and receptor activation.
Pharmacophore mapping of adrenergic agonists suggests that the catechol ring of ligands is important both for binding and receptor activity (27). In this study, we provide several lines of evidence to support the contention that Phe 310 in TMVI of the ␣ 1B -AR mediates such effects through the formation of an aromatic-aromatic interaction between its side chain and the catechol ring. First, computerized modeling of the ␣ 1B -AR structure, and docking of catecholamines into the ligand-binding pocket, indicate that the Phe 310 side chain is well positioned to interact with the catechol ring. Second, as required for a residue directly involved in ligand binding, SCAM studies reveal that the Phe 310 side chain is not only solvent accessible but is directed into the agonist-binding pocket. Third, since substitution of Phe 310 with a variety of other amino acids does not perturb membrane expression or post-translational processing (glycosylation) of the receptor, its effects on ligand bind-ing and receptor activation are due to local interactions, and not to alterations in global receptor structure. In addition, even loss of all bonding potential with alanine substitution of Phe 310 does not impair spontaneous isomerization of the receptor to a partially activated conformation, as is evident with the F310A/ A293E double mutant. Fourth, and most importantly, the aromatic character of the Phe 310 side chain is essential for wildtype agonist binding and, in terms of receptor signaling, for both wild-type agonist potency and efficacy. Thus, although substitution of Phe 310 with a residue that has a hydrophobic side chain, such as leucine, partially compensates for the loss of phenylalanine at the 310 position, only an aromatic residue, e.g. tryptophan, can fully restore ligand-binding and -receptor activation.
Our results do not support the direct involvement of Phe 311 in agonist binding as proposed by Strader et al. (9) for the equivalent residue (Phe 290 ) in the ␤ 2 -AR, or alternatively, indicate that the residues interacting with the catecholamine ring in the ␣ 1B and ␤ 2 -AR, differ. Thus, substitution of Phe 311 with a variety of different residues markedly impaired receptor expression and post-translation processing, indicating that phenylalanine at the 311 position is essential for proper folding of the receptor protein. In addition, neither the volume nor the hydrophobicity of Phe 311 substituent side chains could be directly related to agonist binding or receptor activation. Finally, and most importantly, SCAM studies indicated that the Phe 311 side chain is neither solvent accessible nor directed into the ligand-binding pocket. Thus, effects of Phe 311 substitutions on ligand binding and receptor activation are likely secondary to global changes in receptor structure, rather than to loss of an interaction directly involved in ligand binding.
Based on SCAM studies of the dopamine D2 receptor, it has been suggested recently that all of the aromatic residues in TMVI of biogenic amine receptors are solvent accessible (28). Furthermore, it was postulated that these aromatic residues relayed information from the site of agonist-binding in the extracellular half of the TMVI helix to produce a conformational change at the intracellular end of the helix, resulting in receptor activation. While an attractive hypothesis, the SCAM conclusions were based entirely on antagonist-inactivation data, and were not confirmed by evaluating an effect on agonist binding.
Apart from defining a critical residue (Phe 310 ) involved in ligand binding and signaling by the ␣ 1B -AR, the findings of this study have potentially important ramifications for our understanding of the mechanisms of GPCR activation. For rhodopsin, activation involves initial disruption of an ionic bond between the protonated Schiff base formed by the interaction of 11-cis-retinal with Lys 296 in TMVII and Glu 113 in TMIII (5). Similarly, activation of the ␣ 1B -AR involves disruption of an ionic bond linking Asp 125 in TMIII and Lys 331 in TMVII (3), and partial receptor activation can be induced by a moiety mimic (triethylamine) of the catecholamine protonated amine (29). For rhodopsin, activation has been shown to result in rigid body movement of the TMVI helix (30), and movement of this helix is also a feature of ␤ 2 -AR (31) and ␣ 1B -AR activation. 2 Thus, it is likely that movement of TMIII, due to disruption of a constraining interaction with TMVII, and movement of TMVI, are common features of GPCR receptor activation. Movement of these helices, which are contiguous with the second and third intracellular loops, is also consistent with the known involvement of these loops in G-protein activation (12).
Given these considerations, and based on the findings of this study that implicate Phe 310 in TMVI as a key switch residue in ␣ 1B -AR activation, it seems reasonable to postulate that activation of adrenergic receptors involves initial disruption of the FIG. 8. Basal PI hydrolysis. Total inositol phosphates generated in the absence of agonist were determined in intact COS-1 cells transfected with vector alone (mock), or plasmid encoding the wild-type (WT), or F310A or F310L mutant ␣ 1B -ARs. Total inositol phosphates were quantitated as described under "Experimental Procedures" in cells incubated for 30 min in the presence of 10 mM LiCl. The expression levels of the wild-type, F310A, and F310L mutants were 6.2 Ϯ 0.2, 5.7 Ϯ 0.1, and 5.9 Ϯ 0.3 pmol/mg protein, respectively. Inset, basal PI hydrolysis determined as described above in cells transfected with various amounts of plasmid encoding the wild-type (f) or mutant (Ⅺ, A293E; E, F310A/A293E; q, F310L/A293E) ␣ 1B -ARs, to produce the receptor expression levels (densities) shown. Receptor densities were determined from parallel saturation binding studies. Data points for each receptor construct did not differ from a linear relationship, as determined from test runs (p Ͼ 0.05). The slopes of the regression lines (0.68 Ϯ 0.28, wild-type; 3.79 Ϯ 1.65, A293E; 3.2 Ϯ 1.31, F310A/A293E; 2.53 Ϯ 0.86, F310L/A293E) provide an index of the amount of inositol phosphates generated per pmol of receptor.
Asp 125 /Lys 331 ionic interaction followed by Phe 310 /catechol ring-induced movement of TMVI. Indeed, the findings of our studies with the F310A or L/A293E double mutants support this model based on the following considerations: (i) central to the recently revised ternary complex model of GPCR activation is the finding that mutant receptors can exist in a constitutively active state that allows signaling in the absence of agonist; (ii) in support of this model is the finding that overexpression of wild-type receptors can also initiate biochemical responses in the absence of agonist; (iii) accordingly, it has been proposed that receptors spontaneously resonate between a basal state, R, and an active state, R*, with only the latter being able to productively interact with G-protein, and (iv) as a corollary, constitutively active mutants, which partially mimic the active state, represent an intermediate receptor conformation, RЈ (26). Based on these considerations and on the finding that agonists bind to constitutively active receptors with higher affinity than to wild-type receptors, it has been suggested that rather than inducing the active conformation upon binding, agonists merely select or "trap" the active conformation that results from spontaneous isomerization of R to R* (32). Nevertheless, recent structure-function studies of the angiotensin II (AT 1 ) receptor provide compelling evidence for an RЈ conformation that is a distinct intermediate between R and R*, and that isomerization from the RЈ conformation to the active state involves an inductive step that requires agonist binding (33).
In the present study we demonstrate that little, if any, PI hydrolysis is observed with the wild-type ␣ 1B -AR in the absence of agonist. Moreover, signaling in the absence of agonist was similar with the F310A and F310L mutants, and, importantly, was not reduced below that observed with the unliganded wildtype ␣ 1B -AR, even though these mutants markedly impaired agonist-induced signaling. Thus the unliganded wild-type ␣ 1B -AR receptor likely represents the true basal or R conformation, whereas the receptor maximally activated by high agonist concentrations represents the R* state. By contrast, and consistent with an RЈ conformation, basal signaling was readily apparent with the unliganded A293E mutant. Since the double mutants F310A/A293E and F310L/A293E also showed similar basal signaling in the absence of agonist, but impaired agonist-induced signaling, the Phe 310 -catechol ring interaction is likely not required for isomerization between R and RЈ, but is critical for full receptor activation. Consistent with the above considerations, it seems reasonable to speculate that some other activation process, e.g. disruption of the Asp 125 /Lys 331 ionic interaction, is required for transition from R to RЈ, whereas the Phe 310 -catechol ring interaction mediates isomerization from RЈ to R*. As with the AT 1 receptor (33), this latter requirement of a specific agonist-receptor interaction also supports the notion that isomerization from RЈ to R* is an agonistdependent inductive process. Clearly additional studies, particularly aimed at directly evaluating receptor conformational states, based, for example, on electron spin resonance (30), NMR (34), fluorescence spectroscopy studies (35), Fourier transform infrared resonance spectroscopy (36), and surface plasmon resonance spectroscopy (37), will be required to more fully define the activation process at atomic resolution.