Novel Aromatic Residues in Transmembrane Domains IV and V Involved in Agonist Binding at a 1a -Adrenergic Receptors*

We examined the role that aromatic residues located in the transmembrane helices of the a 1a -adrenergic re- ceptor play in promoting antagonist binding. Since a 1 antagonists display low affinity binding at b 2 -adrener- gic receptors, two phenylalanine residues, Phe-163 and Phe-187, of the a 1a -AR were mutated to the correspond- ing b 2 -residue. Neither F163Q nor F187A mutations of the a 1a had any effect on the affinity of the a 1 -antago- nists. However, the affinity of the endogenous agonist epinephrine was reduced 12.5- and 8-fold by the F163Q and F187A mutations, respectively. An additive loss in affinity (150-fold) for epinephrine was observed at an a 1a containing both mutations. The loss of agonist affin- ity scenario could be reversed by a gain of affinity with mutation of the corresponding residues in the b 2 to the phenylalanine residues in the a 1a . We propose that both Phe-163 and Phe-187 are involved in independent aromatic interactions with the catechol ring of agonists. The potency but not the efficacy of epinephrine in stimulating phosphatidylinositol hydrolysis was reduced 35-fold at the F163Q/F187A a 1a relative to the wild type receptor. Therefore, Phe-163 and Phe-187 represent novel binding contacts in the agonist binding pocket of the a 1a -AR, but are not involved directly in receptor activation. Site-directed Mutagenesis in the a 1a -AR— Site-directed mutagenesis of the pMT 2 9 rat a 1a -AR plasmid was performed using A of cDNA the mutation of the -AR a primer the Eco RI cloning site the start site of translation an antisense primer the Nae I site of the rat 1a -AR the mutation. The F187A mutation using a primer the mutation the endog- Nae I restriction an antisense primer the Not I restriction site the stop codon. the fidelity PCR RI-F163Q- Nae I fragment m g of pMT 2 rat a 1a -AR plasmid, n M each of and antisense primers, n M each of dNTPs, and 2.6 units of Taq and Pwo DNA polymerase a m M Tris-HCl, 7.5, m M KCl, m M M (w/v) Tween a final MgCl 2 of 1.5 m Nae I-F187A- Not I fragment the same protocol the performed with a final MgCl 2 concentration of 4 m M The amplification reactions, repeated for of denaturation at 95 °C for 3 min and an annealing and elongation phase at 72 °C for 3 min. pair fragments from each of the PCR purified followed a Eco RI/ Nae I or Nae I/ Not I restriction enzyme PCR ligated with each other or with WT subcloned pMT2 9 expression full-length rat a 1a -AR construct containing single or double mutations. full-length sum-of squares fit of the data to these equations. Functional data for cAMP stimulation or IP release in COS-1 cells was fit to a sigmoidal concen- tration-response curve. In functional experiments, the term potency applies to the concentration of agonist that stimulates the half-maximal response (EC 50 ). Statistically significant differences in the potency of agonists were determined by t test analysis.

We examined the role that aromatic residues located in the transmembrane helices of the ␣ 1a -adrenergic receptor play in promoting antagonist binding. Since ␣ 1antagonists display low affinity binding at ␤ 2 -adrenergic receptors, two phenylalanine residues, Phe-163 and Phe-187, of the ␣ 1a -AR were mutated to the corresponding ␤ 2 -residue. Neither F163Q nor F187A mutations of the ␣ 1a had any effect on the affinity of the ␣ 1 -antagonists. However, the affinity of the endogenous agonist epinephrine was reduced 12.5-and 8-fold by the F163Q and F187A mutations, respectively. An additive loss in affinity (150-fold) for epinephrine was observed at an ␣ 1a containing both mutations. The loss of agonist affinity scenario could be reversed by a gain of affinity with mutation of the corresponding residues in the ␤ 2 to the phenylalanine residues in the ␣ 1a . We propose that both Phe-163 and Phe-187 are involved in independent aromatic interactions with the catechol ring of agonists. The potency but not the efficacy of epinephrine in stimulating phosphatidylinositol hydrolysis was reduced 35fold at the F163Q/F187A ␣ 1a relative to the wild type receptor. Therefore, Phe-163 and Phe-187 represent novel binding contacts in the agonist binding pocket of the ␣ 1a -AR, but are not involved directly in receptor activation.
The adrenergic receptors (␣ 1a , ␣ 1b , ␣ 1d , ␣ 2a , ␣ 2b , ␣ 2c , ␤ 1 , ␤ 2 , and ␤ 3 ) are part of a larger family of membrane proteins commonly referred to as the G protein-coupled receptor family. All of these AR subtypes bind the endogenous catecholamines, epinephrine and norepinephrine and mediate the actions of the sympathetic nervous system (1). The proposed topography adopted by these G protein-coupled receptors in the plasma membrane is modeled to that of bacteriorhodopsin, in which the protein folds to a highly ordered structure comprised of a series of seven hydrophobic transmembrane (TM) 1 -spanning domains, linked by three intracellular and three extracellular loops (2). Comparison of the amino acid sequences of the cloned adrenergic receptors illustrates the greatest residue conservation is found within the transmembrane helix domains. Accordingly, the binding pocket for epinephrine and norepinephrine is localized to this region in all adrenergic receptors, within the circular array of TM helices in the rhodopsin-like core of the receptor (3). Although differences in the agonist binding interactions at different adrenergic receptor subtypes exist, in general, the catecholamine is stabilized in the binding pocket by an ionic interaction involving the protonated amine of epinephrine and an aspartic acid residue in TM3 (4), by hydrogen bonds between the catechol hydroxyl groups of the agonist and serine residues in TM5 (5), and by an aromatic/hydrophobic interaction involving the catechol ring of the agonist and a phenylalanine residue in TM6 (6). The ␤-hydroxyl of the agonist, which confers stereoselectivity, has been identified in the ␤ 2 -AR as interacting with Asn-293 in TM6 (7), but is not conserved in the ␣ 1 -and ␣ 2 -ARs, highlighting the potential differences in agonist binding among family members. For the endogenous agonists, epinephrine and norepinephrine, it is thought that all the point contacts with the receptor have been identified.
Our understanding of the molecular interactions promoting antagonist binding to adrenergic receptors is limited. We have proposed that the plane of the antagonist binding pocket on these receptors lies above that of the agonist binding pocket, since we have identified that the ␣ 1a -AR-selective antagonists phentolamine and WB4101 make point contacts with three residues in the second extracellular loop of the receptor (8). However, subsequent studies conducted on a series of ␤ 2 -/ ␣ 1a -AR chimeric receptors were unable to identify further point contacts between these antagonists and other extracellular loop residues of the ␣ 1a -AR (9). Two other studies have identified the importance of phenylalanine residues located close to the extracellular surface of the receptor in antagonist binding at adrenergic receptors. In the ␣ 2 -AR, a single phenylalanine residue (Phe-412) at the top of TM7 promotes high affinity binding of yohimbine (10) and mutagenesis of a phenylalanine residue (Phe-82) in the upper regions of TM2 in the ␣ 1a -AR accounts for the ␣ 1a selectivity of the dihydropyridine antagonists (11). These interactions on their own cannot account for the high affinity binding of all these antagonists to adrenergic receptors. Therefore, the adrenergic receptor antagonists must make additional contact with as yet unidentified residues located within the transmembrane helices of the receptor.
Given that residues in the second extracellular loop connecting TM4 and TM5 have been implicated in binding antagonists at ␣ 1 -ARs and that ␣ 1 -AR antagonists contain high aromatic character, we evaluated the role that aromatic residues located at the top of these TM helices may have in binding adrenergic receptor antagonists. Specifically, we identified two phenylalanine residues (Phe-163, Phe-187) located in the upper regions of TM4 and TM5, respectively, of the ␣ 1a -AR that were obvious candidates for mutagenesis. This selection was influenced by neither residue being conserved in the ␤ 2 -AR, a receptor at which these ␣ 1 -AR selective antagonists possess markedly reduced affinity (10,000-fold lower). Here we unexpectedly report that, although neither mutation altered the pharmacological phenotype with regard to antagonist binding at the ␣ 1a -AR, these mutations had significant effects on the binding of catecholamine and phenethylamine agonists at this receptor, agonists for which all receptor contacts were thought to have been identified. Since these phenylalanine residues are not conserved in ␣ 2 -ARs and ␤-ARs, these results stress the differences in the agonist binding pocket of adrenergic receptors despite the fact they all bind the same endogenous ligands. Site-directed Mutagenesis in the ␣ 1a -AR-Site-directed mutagenesis of the pMT 2 Ј rat ␣ 1a -AR plasmid was performed using polymerase chain reaction technology using commercially synthesized oligonucleotides (Life Technologies, Inc.) specifically designed to code for the desired mutation. A fragment of cDNA encoding the F163Q mutation of the rat ␣ 1a -AR was generated using a sense primer containing the unique EcoRI cloning site before the start site of translation and an antisense primer targeted to the unique NaeI site of the rat ␣ 1a -AR that also encoded the mutation. The F187A mutation was generated using a sense primer containing the mutation that was targeted to the endogenous NaeI restriction site and an antisense primer containing the NotI restriction site after the stop codon. Using the Expand high fidelity PCR protocol (Roche Molecular Biochemicals), the EcoRI-F163Q-NaeI fragment was generated using 1 g of pMT 2 Ј rat ␣ 1a -AR plasmid, 300 nM each of sense and antisense primers, 200 nM each of dNTPs, and 2.6 units of Taq and Pwo DNA polymerase in a 20 mM Tris-HCl, pH 7.5, buffer containing 100 mM KCl, 1 mM dithiothreitol, 0.1 mM EDTA, 0.05% (w/v) Tween 20, and a final MgCl 2 concentration of 1.5 mM. The NaeI-F187A-NotI fragment was generated using the same protocol as before except that the reaction was performed with a final MgCl 2 concentration of 4 mM. The amplification reactions, repeated for 40 cycles, consisted of denaturation at 95°C for 3 min and an annealing and elongation phase at 72°C for another 3 min. The resulting 550-and 900-base pair fragments from each of the PCR reactions were isolated and purified followed by a EcoRI/NaeI or NaeI/NotI restriction enzyme digest, respectively. The PCR products were ligated with each other or with their respective WT fragment and subcloned into the pMT2Ј expression vector to yield the full-length rat ␣ 1a -AR construct containing single or double mutations. Mutation(s) were confirmed by full-length sequence analysis of the construct by the dideoxy method (Cleveland Clinic Sequencing Core Facility).

Materials-Drugs
Site-directed Mutagenesis in the ␤ 2 -AR-Site-directed mutagenesis was performed using the synthetic ␤ 2 -AR gene construct, previously shown to code for a receptor protein with similar pharmacology to that of the hamster ␤ 2 -AR (12). This construct codes for a 5Ј EcoRI restriction site upstream of the start codon and terminates with a stop codon and a NotI restriction site, permitting its subcloning into the mammalian expression plasmid pMT 2 Ј. The coding region contains unique restriction sites spaced about 50 base pairs apart to facilitate mutagenic cassette replacement. Mutagenesis of the ␤ 2 -AR at positions Gln-170 and Ala-202 was performed by cassette replacement using synthetic oligonucleotides that introduce the phenylalanine codon at the desired location (Life Technologies, Inc.). Plasmid DNA was prepared following transformation into DH5␣ cells (Life Technologies, Inc.) and purified by RNase treatment and column chromatography using the Wizard maxiprep kit (Promega, Madison, WI). All mutations were confirmed by sequencing using the dideoxy chain termination method (Sequenase; Amersham Pharmacia Biotech).
Cell Culture and Transfection-COS-1 cells (American Type Culture Collection, Rockville, MD) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum, 1% (v/v) penicillin and streptomycin. Cells were maintained and passaged upon reaching confluence by standard cell culture techniques. Experiments were conducted on cells between passages 10 and 25. Cells were transiently transfected with either the wild type or mutated ␣ 1a -AR or ␤ 2 -AR cDNA subcloned into the eukaryotic expression plasmid pMT2Ј, using the DEAE-dextran method as described previously (13). Membrane Preparation-Transiently transfected COS-1 cells were scraped 72 h after transfection, collected, and washed in Hanks' balanced salt solution (HBSS), then pelleted under low speed centrifugation (1260 ϫ g for 5 min). The cell pellet was re-suspended in a 0.25 M sucrose solution and after an additional low speed centrifugation step, the pellet was resuspended in a 10-ml volume of water containing a mixture of protease inhibitors (leupeptin (40 g), phenylmethylsulfonyl fluoride (68 g), bacitracin (400 g), and benzamidine (400 g)) and frozen at Ϫ70°C for 30 min. Membranes were prepared from the cell suspension by 20 strokes of a "B" glass Dounce homogenizer. Nuclear debris was removed by a low speed centrifugation step. Membranes in the supernatant were washed with HEM buffer (20 mM HEPES, pH 7.4, 1.4 mM EGTA, and 12.5 mM MgCl 2 ) and pelleted by high speed centrifugation (30,000 ϫ g for 15 min). Two further washes of the membrane pellet with HEM buffer (20 ml) were performed, and the final pellet was reconstituted in a known volume of HEM buffer containing 10% (v/v) glycerol and stored at Ϫ70°C until use. The protein concentration of the membrane preparation was determined by performing a Bradford assay, using bovine serum albumin as the known standard (14).
Measurement of Ligand Binding Affinities-Saturation binding experiments to measure the affinity of [ 125 I]HEAT or [ 125 I]CYP at ␣ 1a -ARs and ␤ 2 -ARs, respectively, were performed as described previously (12,13). The binding affinities of various adrenergic receptor agonists and antagonists were determined in a series of competition binding experiments. For assays investigating the ligand binding properties of the wild type and mutant ␣ 1a -ARs, receptors were labeled with the ␣ 1 -AR selective antagonist, [ 125 I]HEAT. Studies conducted on the various ␤ 2 -ARs measured the displacement of the ␤ 2 -AR selective antagonist, [ 125 I]CYP. All binding assays were performed in duplicate in HEM buffer, in a total assay volume of 250 l. Aliquots of COS-1 cell membranes expressing the ␣ 1 -AR were incubated with 100 pM [ 125 I]HEAT and a range of 12 different concentrations of the competing agonist or antagonist. Nonspecific binding was defined as the amount of radioactivity that remained bound to the filters in the presence of 10 Ϫ4 M phentolamine. COS-1 cell membranes expressing the ␤ 2 -ARs were incubated with 30 pM [ 125 I]CYP and 12 competing concentrations of ligand. Nonspecific binding in these assays was determined by the radioactivity bound in the presence of 10 Ϫ6 M propranolol. All assays were incubated in a shaking water bath at 22°C for 60 min, after which unbound radioactivity was separated from membrane bound radioactivity by filtration through Whatman GF/C filter paper using a Brandel cell harvester. Filters were washed with 20 ml of ice-cold HEM buffer to remove further nonspecifically bound radioactivity. Bound radioactivity remaining on the filters was counted on a ICN ␥-counter operating at 79.8% efficiency.
Quantitation of Intracellular Inositol Phosphates (IP)-Studies to investigate the signaling capacity of the F163Q/F187A ␣ 1a -AR relative to the wild type ␣ 1a -AR were performed on COS-1 cells, plated on 60-mm culture plates, and maintained in DMEM supplemented with 5% fetal bovine serum. COS-1 cells were transiently transfected by the DEAE-dextran method, using concentrations of the two ␣ 1a -AR cDNA constructs that resulted in equal membrane expression levels of the receptor. [ 3 H]Myoinositol (3 Ci) was added to each plate 48 h after transfection and incubated for another 16 h prior to experimentation to permit uptake by the fibroblasts. Measurement of intracellular IP levels was performed under serum-free conditions by washing the cells with 10 ml of serum-free DMEM. To prevent complete hydrolysis of inositol phosphate moieties, assays were conducted in the presence of the phosphatase inhibitor, LiCl (10 mM), in a total assay volume of 3 ml. Bolus concentrations of epinephrine were added directly to the media and the cells were incubated at 37°C for 45 min in a 5% CO 2 atmosphere. A concentration-response curve for epinephrine was generated over a suitable range of concentrations performing each data point in triplicate. Incubations were terminated by removal of the medium containing the agonist and adding a 1-ml volume of a 0.4 M perchloric acid solution. The cell lysate was scraped, collected, and neutralized by the addition of a 0.5 ml volume of a 0.72 N KOH, 0.6 M KHCO 3 solution. Soluble inositol phosphates in the lysate were isolated by passage through a Bio-Rad AG 1X-8 resin column that was buffered with a 0.1 M formic acid solution. After washing the column with 0.1 M formic acid, bound [ 3 H]IPs were displaced from the column by eluting the column with a 0.1 M formic acid solution containing 1 M ammonium formate. The eluant was collected directly in scintillation vials, scintillant added and the radioactivity detected using a ␤-counter (Beckman, Irvine, CA).
Quantitation of Intracellular Cyclic 3Ј,5Ј-Adenosine Monophosphate (cAMP)-Studies to investigate the signaling capacity of the mutant ␤ 2 -ARs relative to the wild type receptor were conducted in COS-1 cells transiently transfected to express the wild type or point mutated ␤ 2 -ARs. Transfected conditions were titered so that the wild type and mutant receptors were expressed at equal receptor densities. Measurements of the intracellular cAMP levels were made under serum-free conditions in 3 ml of DMEM supplemented with 20 mM HEPES and 5 mM theophylline to prevent phosphodiesterase-mediated breakdown. Agonists (either isoproterenol or ephedrine) were added directly to the medium and incubated at 37°C for 30 min in a 5% CO 2 atmosphere. Concentration-response curves for each agonist were constructed over a suitable range of concentrations performing each data point in duplicate. Incubations were terminated by the removal of the medium containing the agonist and the addition of 200 l of a 0.1 M HCl solution. The cell lysate was scraped and transferred to microcentrifuge tubes. The residual concentration of cAMP in the lysate of each individual plate was determined by a radioimmunoassay kit, following the manufacturer's directions (Amersham Pharmacia Biotech). Radioactivity was counted using a ␤-counter (Beckman).
Data Analysis-Competition binding and functional data were analyzed using the non-linear regression functions of the non-iterative curve fitting program GraphPad Prism. Binding affinities (K i ) were determined by transformation of the program calculated IC 50 value using the Cheng-Prusoff equation. The binding data for each ligand was modeled to one-or two-site binding. The most suitable model was determined by performing an F test comparison of the least sum-of squares fit of the data to these equations. Functional data for cAMP stimulation or IP release in COS-1 cells was fit to a sigmoidal concentration-response curve. In functional experiments, the term potency applies to the concentration of agonist that stimulates the half-maximal response (EC 50 ). Statistically significant differences in the potency of agonists were determined by t test analysis.

RESULTS AND DISCUSSION
Structure-activity studies in our laboratory and others have enhanced our understanding of the molecular interactions involved in promoting the binding of both agonists and antagonists to the various subtypes of the ␣ 1 -adrenergic receptor family. Although agonists bind within the rhodopsin-like core of the receptor to well defined residues (15,16), our knowledge to date indicates that antagonists interact with residues in the receptor that are closer to or exposed on the extracellular face of the receptor. For example, phentolamine and WB4101 interact with three consecutive residues of the second extracellular loop of the ␣ 1a -AR (8). However, the interaction of the antagonists with these residues can only account for a fraction of the total energy required to promote their high affinity binding to the receptor.
Aromaticity has an important function in ligand binding at adrenergic receptors, as demonstrated by the direct involvement of a phenylalanine residue in TM6 in promoting agonist and antagonist binding at the ␤ 2 -AR (6) and ␣ 1b -AR (16). Furthermore, early pharmacophore studies indicated that the phe-nyl ring, conserved in all adrenergic receptor agonists, is a critical requirement for agonist properties (17). Since the structure of ␣ 1 -AR antagonists conserve the aromaticity, these antagonists may also potentially bind to ␣ 1 -ARs via interactions with accessible phenylalanine residues in the receptor. Consideration of the amino acid sequences in the transmembrane helix regions identified two phenylalanine residues as candidates that may be involved in promoting antagonist binding at ␣ 1 -ARs. The first of these residues is located in the first helical turn of TM helix 4, and the other is located in the upper regions of TM helix 5. These candidate residues were selected since neither of these phenylalanine residues is conserved in the corresponding position of the ␤ 2 -AR (Fig. 1), a receptor at which all of the ␣ 1a -AR selective antagonists display 10,000-fold lower affinity. Furthermore, both residues are located directly below the second extracellular loop of the receptor at a distance we expect would permit the antagonists to maintain their other previously determined interactions. Therefore, we performed site-directed mutagenesis of the Phe-163 and Phe-187 residues in the ␣ 1a -AR, substituting to the corresponding residue in the ␤ 2 -AR. The phenotype resulting from these mutations in the adrenergic receptors was assessed in terms of its binding profile for both antagonists and agonists.
The roles of Phe-163 and Phe-187 in promoting antagonist binding to ␣ 1a -ARs were evaluated directly in radioligand binding studies. However, contrary to our expectations, the results of our competition binding studies did not support our hypothesis. In saturation binding studies, the radiolabeled antagonist [ 125 I]HEAT bound to the wild type and each of the single as well as double mutant receptors with equal affinity (Table I).
The affinities of a panel of ␣ 1 -AR selective antagonists at either the F163Q ␣ 1a -AR or the F187A ␣ 1a -AR were not significantly different from their respective affinity at the wild type ␣ 1a -AR. Therefore, we conclude that neither phenylalanine residue at the top of the TM4 or TM5 helix is directly involved in binding antagonists at the ␣ 1a -AR.
Previously, we have suggested that the second extracellular loops in G protein-coupled receptors may fold over the top of the receptor. Since phentolamine and WB4101 make point contacts with residues on this corresponding loop of the ␣ 1a -AR (8), we speculated that the folding of the loop orients other pharmacophores in these antagonists such that they project toward and interact with phenylalanine residues at the top of the transmembrane helices. Our results would suggest that the orientation of the loop does not bring the antagonists into proximity with the TM4 or TM5 helices. Rather, the results of other mutagenesis studies in the ␣ 1b -AR identifying a phenylalanine residue in TM6 as a key contact (16), as well as other studies in the ␣ 2 -AR that suggest TM7 plays a role in antagonist binding (10), would indicate that the loop is oriented to bring the antagonists into close contact with these helices.
To complete the characterization of the F163Q-and F187A- ␣ 1a -AR mutations, the affinities of a panel of agonists (Fig. 2) was determined in additional competition binding studies. K i values for five ␣ 1 -AR agonists are shown in Table I. Unlike the results obtained with the antagonists, mutagenesis of either phenylalanine residue in the ␣ 1a -AR had considerable effects on the binding affinities of agonists at this receptor. In COS-1 cells, overexpressed ␣ 1 -ARs display single-site competition curves with Hill coefficients near unity. The addition of GTP analogues does not change the slope significantly and were not used in these competition studies. The affinity of the endogenous agonist epinephrine was reduced 12.5-fold by the F163Q mutation (p Ͻ 0.01) and by a factor of 8-fold by the F187A mutation (p Ͻ 0.01), suggesting that each phenylalanine residue is directly involved in stabilizing the catecholamine in the agonist binding pocket of the ␣ 1a -AR. Furthermore, we argue that the impaired binding of epinephrine at the mutated ␣ 1a -AR is not the result of changes in the global conformation of the receptor since the affinities of the antagonists were not effected by either mutation. Therefore, both the Phe-163 and Phe-187 residues of the ␣ 1a -AR constitute novel and specific agonist-receptor point contacts in the binding pocket of the ␣ 1 -ARs.
Like epinephrine, the affinities of the synthetic phenethylamines, phenylephrine and methoxamine (Fig. 2), were also reduced by each phenylalanine mutation. A 6-fold decrease in affinity for phenylephrine (p Ͻ 0.05) and a 12-fold decrease in the affinity of methoxamine (p Ͻ 0.01) were observed at ␣ 1a -ARs displaying the F163Q mutation (Table I). The F187A mutation had greater effects on the binding of the endogenous agonist epinephrine than on either of the phenethylamines. Phenylephrine and methoxamine binding was reduced by factors of just 3-and 4-fold, respectively, by the F187A mutation (p Ͻ 0.05), as opposed to the greater than 8-fold decrease in affinity for epinephrine induced by the same mutation.
Single mutagenesis of either the Phe-163 or Phe-187 residue of the ␣ 1a -AR illustrates that both amino acids provide significant contributions to the binding and stabilization of the agonist in the binding pocket of the receptor. To prove that these interactions between each phenylalanine residue and the catechol ring of the agonist are independent of one another, we constructed a mutant receptor expressing both mutations (F163Q/F187A ␣ 1a -AR). As observed with the single mutations, the binding affinity of the ␣ 1a -AR-selective antagonists was not altered by the combination of these mutations, indicating that the global conformation of the receptor remains comparable to the wild type ␣ 1a -AR (Table I). However, the affinity of epinephrine for the F163Q/F187A ␣ 1a -AR was found to be over 150-fold lower than its affinity at the wild type ␣ 1a -AR (p Ͻ 0.001). The additive loss of affinity observed upon combining both mutations in the receptor confirms that each phenylalanine residue makes an interaction with the agonist that is independent of the other residue. A similar additive decrease in affinity at the F163Q/F187A ␣ 1a -AR was observed for phenylephrine (32-fold; p Ͻ 0.01). We conclude that the effects observed on agonist binding by mutation of the Phe-163 and Phe-187 residues of the ␣ 1a -AR identify two distinct and independent interactions between the agonist and the receptor.
The influence of mutating each of these phenylalanine residues on the binding properties of imidazolines at ␣ 1a -ARs was also investigated. Imidazolines, e.g. oxymetazoline and clonidine, retain the aromatic characteristics of the phenethylamines; however, their chemical structure differs in that they possess an imidazoline ring containing the protonated amine (Fig. 2). Minimal effects on the binding affinity of oxymetazoline were observed at either the F163Q or the F187A mutation. Only in the ␣ 1a -AR exhibiting both phenylalanine mutations did the affinity of oxymetazoline decrease significantly versus wild type (8-fold; p Ͻ 0.05; Table I). With clonidine, a minimal decrease in affinity was observed with the F163Q mutation. However, the affinity of this drug for the ␣ 1a -AR was actually increased 5-fold by the F187A mutation (p Ͻ 0.05). Our observations indicate that these F163Q and F187A mutations have greater impact on the binding properties of catecholamines and phenethylamines than on the imidazoline class of agonists, suggesting that there are differences in the three-dimensional geometry of the agonist binding pocket recognized by these two drug classes. Indeed, previous studies have indicated that the imidazolines bind differently in the ␣ 1 -AR agonist pocket (18) and confer differential desensitization of the receptor (19), suggesting that the phenethylamines and imidazolines may promote agonist trafficking by coupling to different signaling pathways.
The F163Q-, F187A-, and F163Q/F187A-mutated ␣ 1a -ARs all displayed phenotypes showing a reduced affinity for the endogenous agonist epinephrine. To confirm and validate these phenotypes, we constructed mutations of the ␤ 2 -AR that now introduced phenylalanine residues to the corresponding positions of this receptor. The aim of this mutagenesis strategy was to TABLE I Agonist and antagonist binding affinities at wild type and phenylalanine mutations of the ␣ 1a -AR pK I values for the binding of agonists and antagonists were determined in competition binding studies on membranes transiently transfected to express wild type ␣ 1a -AR, F163Q, F187A, or F163Q/F187A mutations of the ␣ 1a -AR. All competition binding isotherms were best fit to a single-site model. The receptor densities and affinities for the antagonist [ 125 I]HEAT were determined from saturation binding isotherms performed on the same membranes. Values in parentheses represent the -fold changes in agonist affinity relative to the affinity at the wild type receptor. Statistically significant differences in affinities at the mutant receptors versus wild type values were determined by Student's t test; *, p Ͻ 0.05, **, p Ͻ 0.01, ***, p Ͻ 0.001. Data are reported as the mean Ϯ S.E. of three to six experiments. ND, not determined. demonstrate a gain of function manifested by an increased affinity of agonists at the phenylalanine-substituted ␤ 2 -ARs relative to the wild type receptor. Therefore, the Q170F and A202F mutations were individually inserted into the coding sequence of the ␤ 2 -AR by cassette oligonucleotide replacement and the mutant receptors expressed in COS-1 cells to study their agonist binding properties. Incorporation of either the Q170F or the A202F mutation to the ␤ 2 -AR did not appear to affect the global conformation of the receptor since the affinities of several antagonists were not found to be different from the observed wild type values (Table  II). However, as we predicted, both mutations did alter the binding properties of agonists to the ␤ 2 -AR. Small but statistically significant increases in the affinity of both epinephrine and isoproterenol were observed at both the Q170F or A202F mutations relative to their affinity at the WT ␤ 2 -AR (Table II). The affinity of isoproterenol was increased by 1.6-fold (p Ͻ 0.05) and by 2-fold (p Ͻ 0.05) at the ␤ 2 -ARs expressing the Q170F and A202F mutations, respectively. Likewise, the affinity of epinephrine was increased 2.4-fold at the Q170F mutation (p Ͻ 0.05) and by 1.65-fold at the A202F mutation. In order to validate these small increases in affinity at the individual mutations, a ␤ 2 -AR that coded for both phenylalanine mutations was constructed. Competition binding studies at this Q170F/A202F receptor revealed a larger 4.2-fold increase in affinity for both isoproterenol (p Ͻ 0.05) and epinephrine (p Ͻ 0.05) ( Table III). The observed additivity of the response at the double mutation validates the increased affinities of these agonists at each phenylalanine residue. Therefore, incorporation of the phenylalanine residues to the fourth and fifth transmembrane helices of the ␤ 2 -AR results in small gains in the affinity of agonists.
The ternary complex model of agonist interactions at these receptors predicts that agonists bind to the G protein-coupled and uncoupled forms of the receptor with different affinity (20). The small increases in agonist affinity observed at the Q170Fand A202F-␤ 2 -ARs may therefore be due to an enhanced coupling of the mutant receptors to a G protein. To address this concern, the affinities of these agonists at both of the single mutation constructs (Table II) and the double mutation construct (Table III) of the ␤ 2 -AR were determined in the presence of GTP␥S, a stable GTP analog that "locks" the receptor in the uncoupled state. Under these experimental conditions, both epinephrine and isoproterenol displayed affinity increases at the single and double mutations that were of similar magnitude to those observed in the absence of GTP␥S. The results of these binding experiments were confirmed by investigating the basal signaling properties of these mutant receptors. Measurement of the cAMP signal transduction pathway in COS-1 cells transiently transfected to express either the wild type or a mutant ␤ 2 -AR revealed that the basal cAMP stimulation associated with either the Q170F or A202F ␤ 2 -AR was not different from the wild type receptor (data not shown). We conclude that neither of these mutations to the ␤ 2 -AR induces conformational changes that render the receptor to be constitutively active. Rather, these binding and signaling studies indicate that the observed increases in agonist affinity at the phenylalaninesubstituted ␤ 2 -ARs are independent of G protein coupling effects and that the increases in affinity are due directly to interactions between the agonist and the phenylalanine residues.
Phenylalanine residues in TM6 of biogenic amine receptors have been shown to be important for agonist binding and activation in the ␤ 2 -AR (6), the dopamine D 2 receptor (21), serotonin receptors (22), and recently in the ␣ 1b -AR (16). We have now demonstrated that two additional and novel phenylalanine residues found in the TM4 and TM5 helices are also involved in agonist binding at the ␣ 1a -AR receptor. The ␣ 1b -AR and ␣ 1d -AR-subtypes also conserve TM5 aromaticity but conserve TM4 hydrophobicity. Therefore, these agonist point contacts may be conserved in the other ␣ 1 -AR subtypes, but with subtle differences in the characteristics of the interaction. In addition, these two aromatic residues are also conserved in the dopamine D 2 receptor, while only the TM5 phenylalanine residue is conserved in the serotonin receptor. Based on the other known ionic and hydrogen bond interactions between the agonist and receptor, we know that the catechol ring group of the agonist is projected toward the closed end of the receptor that is formed by the TM4, TM5, and TM6 helices, the same helices upon which the important phenylalanine residues that we and others (16) have identified. Consequently, these phenylalanine residues may assist in binding epinephrine at the ␣ 1a -AR as a result of stacking interactions formed between the respective -electron clouds of the agonists catechol ring and the aromatic side chain of the phenylalanine residue. Alternatively, the phenylalanine residues may interact with the catechol ring group of the agonist via edge to face aromatic interactions. These aromatic interactions occur as a result of dipole-dipole interactions that develop due to an unequal local electron distribution within the respective aromatic rings (23). The strength of these interactions is dependent upon the strength of the dipole moments in the aromatic rings, the distance between the two aromatic rings (1/r 5 ), and the angle of orientation between the rings.
To determine whether these TM4 and TM5 phenylalanine residues of the ␣ 1a -AR interact with the catechol ring of the agonist via aromatic interactions, a further series of radioligand binding experiments were conducted. In these experiments, a series of agonists with a variety of electron donating (i.e., CH 3 , OCH 3 , OH ranked in order of increasing electron-TABLE II Agonist and antagonist binding affinities at single mutations of the ␤ 2 -AR pK I values for the binding of agonists and antagonists were determined in competition binding studies on membranes transiently transfected to express wild type ␤ 2 -AR, Q170F, or A202F mutations of the ␤ 2 -AR. All competition binding isotherms best fit to a single-site model in the absence or presence of GTP␥S. The receptor densities and the affinity values of the radiolabeled antagonist [ 125 I]CYP were determined from saturation binding isotherms performed on the same membranes. Values in parentheses represent the -fold changes in agonist affinity relative to the affinity at the wild type receptor. Statistically significant differences in affinity values at mutant receptors relative to wild type were determined by Student's t test analysis; *, p Ͻ 0.05. Data are reported as the mean Ϯ S.E.M. of three to four experiments. ND, not determined.    donating strength) and/or electron withdrawing (i.e., Cl) substituents in the catechol ring of the agonist were used. Since the strength of an aromatic interaction is dependent upon the magnitude of the charge separation in the ring, we hypothesized that these substituents would influence the dipole strength sufficiently such that the magnitude of the affinity increase or decrease would be altered accordingly with electron-donating groups increasing aromaticity and electron-withdrawing groups decreasing aromaticity. The affinities of albuterol, metaproterenol, dichloroisoproterenol, nylidrin, and ephedrine at wild type and Q170F/A202F ␤ 2 -ARs are listed in Table III. As shown in Fig. 2, dichloroisoproterenol is a chemical analog of the agonist isoproterenol that differs only in respect that both catechol hydroxyl groups on the ring are substituted with chlorine groups. The strong electronegative character of the chlorine substituents when compared with ring hydroxyl groups, acts to draw the electron density out from the center of the catechol ring to the extremities, weakening the strength of the charge separation in the ring. Consistent with our hypothesis, the affinity of dicloroisoproterenol was only marginally increased by a factor of 1.4-fold while the affinity of isoproterenol was increased by 4.2-fold (p Ͻ 0.05) at the double mutant receptor (Table III). Albuterol, metaproterenol, and nylidrin, which all have intermediate electron-donating properties as compared with isoproterenol, displayed intermediate increases in affinity ranging from 2-to 3-fold.
Ephedrine was the only drug tested that displayed a greater gain in affinity (6-fold; p Ͻ 0.01) than that of isoproterenol at the Q170F/A202F ␤ 2 -AR. Relative to isoproterenol, the structure of ephedrine has no substituents in the ortho, meta, or para positions of its aromatic ring. As a consequence, the absence of anchoring substituents in the aromatic ring of ephedrine may permit greater rotational freedom that optimizes its interaction with the phenylalanine residues causing ephedrine to bind differently in the pocket as compared with epinephrine. Based on the binding profiles of dichloroisoproterenol (1.4-fold increase) and isoproterenol (4.2-fold increase) at the Q170F/ A202F ␤ 2 -AR being proportional to the magnitude of the theoretical dipole moment in the catechol ring, we conclude that the gain in affinity displayed for agonists at this receptor is provided by aromatic interactions between the catechol ring of the drug and the phenylalanine residues substituted into TM4 and TM5 of the ␤ 2 -AR. Analysis of aromaticity in the ␣ 1 -agonists is also consistent with the results in the ␤ 2 -AR, where a rank order of the loss of affinity (methoxamine Ͻ phenylephrine Ͻ epinephrine) correlates to the increasing aromatic character of the catechol ring (OCH 3 Ͻ OH) (Table I).
Based on the changes in drug affinity at the F163Q/F187A ␣ 1a -AR, we can calculate that the two aromatic interactions in the ␣ 1a -AR contribute a total free energy equivalent of 2.95 kcal/mol toward the binding of epinephrine. Since the theoretical bond energy of a single aromatic-aromatic interaction lies in the range of 1.5-2 kcal/mol (23), our reported change in free energy at the double mutant is consistent with each phenylalanine residue participating in an independent aromatic-aromatic interaction with the catechol ring of the agonist. Since neither of these phenylalanine residues are conserved in the TM4 and TM5 helices of the ␤ 2 -AR, we expected to observe increases of similar magnitude in the affinity of epinephrine affinity for the Q170F/A202F ␤ 2 -AR as a result of the receptors potential to bind the agonist by two novel aromatic interactions. Contrary to our expectations, the observed change in free energy resulting from epinephrine binding at the Q170F/ A202F ␤ 2 -AR was calculated to be only 0.75 kcal/mol. In theory, the greatest energy contribution between two aromatic rings occurs when the rings are aligned such that the ␦ ϩ edge of one ring is projected toward the ␦ Ϫ face of the other ring. As the angle of orientation moves away from this perpendicular edgeto-face alignment, the force of the interaction decreases exponentially. In a previous study, we have reported that the catechol ring of epinephrine may be aligned differently in the agonist binding pockets of the ␣ 1a -AR and the ␤ 2 -AR, a consequence of the unique hydrogen bond interactions between the meta and para catechol hydroxyl groups of the agonist and the serine side chains and the number of amino acids separating these serine residues in TM5 of these receptors (15). As a result, the relative orientation of the catechol ring in epinephrine between these receptors is altered by a angle of 120°. The binding data in this study suggest that the alignment of epinephrine at the ␣ 1a -AR is oriented so that the interaction of its catechol ring with the TM4 and TM5 phenylalanine residues is favorable; thus, the proposed planar orientation of epinephrine results in an edge to face orientation of the phenylalanines to the catechol ring. Previous structural studies on the dopamine D 2 receptor (24) and density maps in the rhodopsin receptor (2) suggest that this region of TM5 exists as a loop and not a true ␣-helix, therefore permitting Phe-187, Ser-188, and Ser-192 to remain accessible in the binding crevice of the receptor in order that contact with the agonist is maintained. Conversely, the angle of orientation of the catechol ring of epinephrine in the binding pocket of the ␤ 2 -AR is such that it minimizes the aromatic interactions with the Q170F or A202F mutations in this receptor, consistent with the proposed skewed orientation of epinephrine in the pocket and the resulting and less favorable edge to edge orientation of the phenylalanines to the catechol ring. Therefore, this study suggests that the unique nature of the serine interactions in the ␣ 1a -AR is also important in terms of the receptor's ability to enhance agonist binding by promoting aromatic-aromatic interactions.
Sequence alignment of the TM4 and TM5 helices reveals that the two phenylalanine residues in TM4 and TM5 involved in agonist binding at the ␣ 1a -AR receptor are also conserved in the dopamine D 2 receptor but only the TM5 phenylalanine residue is conserved in the serotonin receptor and the ␣ 1b -AR and ␣ 1d -AR subtypes. In the two additional ␣ 1 -AR subtypes, the corresponding TM4 residue is a leucine, suggesting that agonist binding to this site on TM4 is promoted by hydrophobicity rather than aromaticity in these subtypes. Our observations comparing ␣ 1a -AR and ␤ 2 -AR agonist binding interactions and the potential for hydrophobic interactions at TM4 at the ␣ 1b -AR and ␣ 1d -AR subtypes emphasizes the diversity of interactions in the agonist binding pocket among the adrenergic receptor subtypes, even though they all bind the endogenous agonist with similar affinities.
The conserved phenylalanine residue in TM6 is involved in both binding and the efficacy of signaling in biogenic amine receptors (6,16,21,22). Therefore, the importance of the Phe-163 and Phe-187 residues in facilitating agonist-induced receptor activation in the ␣ 1a -AR was studied by determining the ability of the mutant receptor to stimulate phosphatidylinositol hydrolysis in response to challenge with epinephrine. Experiments were conducted in COS-1 cells that were transiently transfected to express either the wild type ␣ 1a -AR or the F163Q/F187A ␣ 1a -AR at equal receptor densities. As shown in Fig. 3, the concentration-response curve for epinephrine in stimulating IP at the F163Q/F187A ␣ 1a -AR (EC 50 ϭ 4.89 Ϯ 1.1 M) is shifted dramatically to the right of the wild type receptor response (EC 50 ϭ 0.13 Ϯ 0.08 M). This shift represents a 36-fold decrease in the potency of epinephrine in stimulating IP at the F163Q/F187A ␣ 1a -AR (p Ͻ 0.01), consistent with its change in affinity. The efficacy (maximal response) of the epinephrine was not altered with respect to the wild type receptor by either of the F163Q and F187A mutations of the ␣ 1a -AR. In the ␤ 2 -AR, the equivalent paradigm was held. Ephedrine, which demonstrated the greatest increase in affinity (6-fold) at the double mutant, Q170F/A202F receptor, also produced a 4-fold leftward shift in potency (EC 50 ϭ 2.4 Ϯ 1.1 M) when compared with the WT receptor (EC 50 ϭ 8.8 Ϯ 1.4 M) with no change in the maximal response (Fig. 4). We conclude that the decreased potency of epinephrine at the F163Q/F187A ␣ 1a -AR results from the reduced occupancy of the receptor that is attributable to the phenylalanine residues acting as important binding contacts.
In conclusion, we have identified two novel aromatic-aromatic interactions that promote the high affinity binding of epinephrine and the phenethylamines to the ␣ 1a -AR. These interactions involve two phenylalanine residues located in the upper regions of TM helix 4 and TM helix 5 of the ␣ 1a -AR, respectively. Each phenylalanine residue interacts independently with the catechol ring of the agonist. The free energy change in binding affinities resulting from the mutagenesis of these residues is consistent with the reported energetic contributions provided by aromatic-aromatic interactions in proteins. Neither Phe-163 nor Phe-187 appear to be directly involved in promoting agonist induced stimulation of phosphatidylinositol hydrolysis. Rather, the results indicate the importance of aromaticity to agonist binding and may have important consequences for future therapeutic drug design of selective ␣ 1 -AR agonists. Differences in binding as observed upon incorporation of the corresponding aromatic residues in the ␤ 2 -AR stress the differences in the agonist binding pocket between the members of the adrenergic receptor family, even though they all bind the endogenous agonists with similar affinity.