Phenoxybenzamine binding reveals the helical orientation of the third transmembrane domain of adrenergic receptors.

Phenoxybenzamine (PB), a classical alpha-adrenergic antagonist, binds irreversibly to the alpha-adrenergic receptors (ARs). Amino acid sequence alignments and the predicted helical arrangement of the seven transmembrane (TM) domains suggested an accessible cysteine residue in transmembrane 3 of the alpha(2)-ARs, in position C(3.36) (in subtypes A, B, and C corresponding to amino acid residue numbers 117/96/135, respectively), as a possible site for the PB interaction. Irreversible binding of PB to recombinant human alpha(2)-ARs (90 nm, 30 min) reduced the ligand binding capacity of alpha(2A)-, alpha(2B)-, and alpha(2C)-AR by 81, 96, and 77%. When the TM3 cysteine, Cys(117), of alpha(2A)-AR was mutated to valine (alpha(2A)-C117V), the receptor became resistant to PB (inactivation, 10%). The beta(2)-AR contains a valine in this position (V(3.36); position number 117) and a cysteine in the preceding position (Cys(116)) and was not inactivated by PB (10 microm, 30 min) (inactivation 26%). The helical orientation of TM3 was tested by exchanging the amino acids at positions 116 and 117 of the alpha(2A)-AR and beta(2)-AR. The alpha(2A)-F116C/C117V mutant was resistant to PB (inactivation, 7%), whereas beta(2)-V117C was irreversibly inactivated (inactivation, 93%), confirming that position 3.36 is exposed to receptor ligands, and position 3.35 is not exposed in the binding pocket.

In the adrenergic receptor subfamily, the seven ␣-helical transmembrane (TM) 1 domains form a crevice for the recognition and binding of ligands (1). The amino acid sequences are highly conserved within the seven hydrophobic TM domains of the three human ␣ 2 -adrenoreceptor (␣ 2 -AR) subtypes (ϳ 75% amino acid identity) (2)(3)(4)(5). The ␣ 2 -AR subtypes also share significant structural identity within the TM domains with other members of the adrenoreceptor family (2,6). For example, of the 182 amino acids comprising the TM domains of the human ␣ 2A -AR, about 40% are identical with the human ␤ 2 -AR. The conserved regions are known to contain structural determinants responsible for recognizing and binding the endogenous hormones/neurotransmitters adrenaline and noradrenaline and other receptor ligands (2,(7)(8)(9).
Phenoxybenzamine (PB) is an irreversible, subtype-nonselective ␣-AR antagonist. PB has been used as a pharmacological tool to study ␣-AR subpopulations in tissue preparations. PB was also the first ␣-AR antagonist to be therapeutically evaluated in humans. It produces long lasting ␣-AR blockade and reduces blood pressure, but its clinical use was limited by severe side effects (10 -12). Although the pharmacology of PB has been studied quite extensively, the molecular basis of its interaction with ␣-ARs has not been examined in detail. It is known that ␤-haloalkylamines, such as PB, cyclize in aqueous solution to form an unstable aziridinium ion, which can bind to target proteins with a strong ionic bond. The aziridinium ion then opens to create a reactive intermediate, with the consequence that a covalent bond between the drug molecule and the binding site can be formed. Side chains of amino acid residues that can be alkylated by haloalkylamines include -SH, -OH, ϭNH, and -COOH (13). Of the susceptible amino acid residues, cysteine (-SH) is the most reactive (14).
The helical arrangement of the TM domains indicated by receptor modeling (15,16), in conjunction with analysis of sequence alignments (see Fig. 1), suggested to us an interaction between PB and the TM3 region of the ␣-ARs. The amino acid sequence alignment of TM3 of adrenergic receptors presented in Fig. 1 is validated by the position of the conserved aspartate residue (D 3.32 according to the nomenclature of Ballesteros and Weinstein (17) or position 113 in the ␣ 2A -AR), known to be crucial for binding the charged nitrogen present in adrenergic phenethylamine ligands (7,9). According to our hypothesis, the reactive aziridinium derivative of PB forms a covalent bond with C 3.36 in TM3 of the ␣ 2A -, ␣ 2B -and ␣ 2C -AR ( Fig. 2) (corresponding to amino acid residues 117, 96 and 135, respectively). Also the three ␣ 1 -AR subtypes have a cysteine in this position, but the three ␤-AR subtypes have a valine residue in its place (Fig. 1). To test our hypothesis, we determined the irreversible binding of PB to the three human ␣ 2 -AR subtypes and constructed and tested an ␣ 2A -AR mutant lacking the Cys 117 (Cys 117 was substituted with valine; ␣ 2A -C117V). We also tested the effect of PB on the human ␤ 2 -AR, which has a valine (V 3.36 ) in the position corresponding to Cys 117 in ␣ 2A -AR and a cysteine (C 3.35 ) in the preceding position (Fig. 1). The wild-type recombinant ␤ 2 -AR was resistant to the alkylating effect of PB, which indicated that this position is not exposed in the cavity. To confirm the structural orientation of TM3, ␣ 2A -AR was mutated to resemble ␤ 2 -AR (␣ 2A -F116C/C117V) and vice versa (␤ 2 -V117C). The ␣ 2A -F116C/C117V mutant was resistant to the alkylating effect of PB, whereas ␤ 2 -V117C was irreversibly inactivated by PB treatment, confirming that a cysteine in position 3.36 is required for alkylation of adrenergic receptors by PB and that position 3.35 is unreachable by ligands in the binding crevice and probably points toward another TM helix or the lipid bilayer.
Mutagenesis and Receptor Production-The cDNA encoding the human ␣ 2A -AR (2) was inserted into the SmaI site, and the cDNA of the human ␤ 2 -AR (18) was inserted into the HindIII/XbaI sites of the vector pALTER-1 (Promega, Madison, WI). Site-directed mutagenesis was performed utilizing the Altered Sites II in vitro mutagenesis system (Promega). Mutated ␣ 2 -AR cDNAs were subcloned into the KpnI/ BamHI sites of the expression vector pREP4 (Invitrogen). The wild-type ␤ 2 -and mutated ␤ 2 -AR cDNAs were subcloned into the PvuII site of the vector pREP4.
Chinese hamster ovary (CHO) cell lines expressing wild-type ␣ 2A -, ␣ 2B -, and ␣ 2C -AR were established as described earlier (19). Adherent CHO cells (American Type Culture Collection, Manassas, VA) were cultured as reported previously (16). The pREP4-based expression constructs were transfected into cells using the Lipofectin reagent kit (Life Technologies, Inc.). Hygromycin B (Roche Molecular Biochemicals)resistant (450 g/ml) cell cultures expressing wild-type and mutated ␣ 2A -and ␤ 2 -ARs were tested for their ability to bind the radioligands Receptor Inactivation and Ligand Binding-Cells were harvested into chilled phosphate-buffered saline, pelleted, washed, suspended in ice-cold 50 mM potassium phosphate buffer (pH 7.4 at 25°C), and homogenized with an Ultra-Turrax homogenizer (model T25; Janke & Kunkel, Staufen, Germany; 9500 rpm, twice for 10 s). The homogenate was used for saturation and competition binding assays or receptor inactivation experiments.
Saturation binding assays were performed in 50 mM potassium phosphate buffer as described previously (20). Whole-cell homogenates containing 20 -50 g of protein were incubated in 250 l of buffer with radioligand (0.125-8 nM) at 25°C. Specificity of binding was defined with 10 M phentolamine or 10 M propranolol (for ␣ 2 -or ␤ 2 -ARs, respectively). Competition binding assays were carried out essentially as reported previously (20)  For receptor inactivation assays, cell homogenates were incubated with PB (10 pM-10 M) in 2.5 ml of potassium phosphate buffer for 30 or 60 min at 37°C. The protein concentration was 0.6 -0.9 mg/ml during PB treatment. Next, membranes were pelleted at 40,000 ϫ g for 15 min at 4°C, washed twice with 2.5 ml of ice-cold K ϩ -phosphate buffer, and rehomogenized. Control samples were treated in an identical manner but without PB. Residual receptor binding capacity in the membranes was assessed by incubation (0.2-0.4 mg protein/assay tube) with 2 nM [ethyl-3 H]RS79948 -197 or [ 3 H]CGP-12177. Nonspecific radioligand binding was again determined by including 10 M phentolamine or propranolol in parallel assays.
Molecular Modeling-A model of the human ␣ 2A -AR was built as described earlier (15). Phenoxybenzamine was manually docked into the binding cavity. A covalent bond was introduced between the terminal carbon of the N-chloroethylene group of the ligand and the sulfur atom of the Cys 117 side chain.
Reaction of PB with Oligopeptides and Mass Spectroscopic Analysis-Two 9-mer oligopeptides, Val-Leu-Phe-Cys-Thr-Ser-Ser-Ile-Val corresponding to residues 114 -122 of the wild-type ␣ 2A -AR and Val-Leu-Phe-Val-Thr-Ser-Ser-Ile-Val representing the same region of the ␣ 2A -C117V mutant receptor, were synthesized with an Applied Biosystems 431A peptide synthesizer using standard Fmoc (N-(9-fluorenyl) methoxycarbonyl) techniques. To 0.2 mM oligopeptide in 50 mM ammonium acetate buffer (pH 7.4, at 24°C), 1.2 molar equivalents of 5 mM PB were added. After 2 h of incubation at room temperature, 1 volume of acetonitrile was added to dissolve the precipitated product, and the reaction mixture was analyzed with high pressure liquid chromatography-mass spectrometry (PE Sciex API 365).  (Table I) (Table I).

Site-directed Mutagenesis and Transfections-To
Receptor Inactivation Studies-To validate the experimental conditions in the PB inactivation assay, we first compared the effect of PB treatment (10 pM-10 M) with two incubation times (30 or 60 min) on the capacity of [ethyl-3 H]RS79948 -197 and [ 3 H]CGP-12177 binding in CHO cell homogenates expressing wild-type ␣ 2 -and ␤ 2 -ARs (data not shown). In subsequent experiments, PB was used at concentrations of 30, 60, and 90 nM for ␣ 2 -ARs and 1 and 10 M for ␤ 2 -ARs. The incubation of cell homogenates for 30 min at 37°C in the absence (control) and presence of PB was chosen as optimal for inactivation assays.
We first tested the alkylating effect of PB on the three human ␣ 2 -AR subtypes (␣ 2A , ␣ 2B , and ␣ 2C ) expressed in CHO cells. PB treatment (90 nM, 30 min) reduced the binding capacity of ␣ 2A -, ␣ 2B -, and ␣ 2C -AR by 81, 96, and 77%, respectively (Fig. 3). This was in agreement with the involvement of an exposed cysteine in TM3 of the ␣ 2 -ARs in the alkylating effect To further characterize the interaction of PB with TM3 cysteines, we compared the effects of PB treatment on ␣ 2A -and ␤ 2 -ARs. Instead of a cysteine, the human ␤ 2 -AR contains a valine in the corresponding position (V 3.36 ). Incubation with PB (90 nM, 30 min) inactivated 81% of ␣ 2A -AR, but ␤ 2 -AR binding was not affected (inactivation 0%). When C 3.36 of ␣ 2A -AR was mutated to valine, the mutated receptor ␣ 2A -C117V became relatively resistant to the alkylating effect of PB (inactivation 10% at 90 nM, 30 min). Also an ␣ 2A -AR mutant containing a cysteine in the preceding position (␣ 2A -F116C/C117V), resembling the ␤ 2 -AR, was resistant to alkylation by PB (inactivation 7%) (Fig. 4). As the expression level of the wild-type ␣ 2A -AR in these experiments was about 10-fold higher than the expression levels of the ␣ 2A -C117V and ␣ 2A -F116C/C117V receptor mutants (Table I), we later performed PB inactivation experiments with another batch of cells expressing wild-type ␣ 2A -AR at only 284 Ϯ 5 fmol/mg total cellular protein. These receptor inactivation results were similar to those in Table I (inactivation 73%). However, the inactivation was dependent on the concentration of PB. We tested the ␣ 2A -AR wild-type (Fig. 5) and mutated receptors also with 1 and 10 M PB. All three receptors were potently inactivated by these high concentrations of PB (inactivation with 1/10 M PB was 89.3/99.5%, 41.5/98.3%, and 62.8/97.5% for wild-type ␣ 2A -AR, ␣ 2A -C117V, and ␣ 2A -F116C/C117V, respectively).
After the opposite mutation in ␤ 2 -AR (V 3.36 to cysteine, to resemble the ␣ 2A -AR), this receptor became susceptible to the alkylating effect of 10 M PB (inactivation 93%). The wild-type ␤ 2 -AR was relatively resistant to PB alkylation even with such a high PB concentration (inactivation 26%) (Fig. 5). This confirmed our hypothesis of a structure-activity relationship between the alkylating effect of PB and an exposed cysteine residue in position 3.36 (117 in ␣ 2A -AR and ␤ 2 -V117C) in TM3 of adrenergic receptors. This also indicated that the preceding position (F 3.35 in ␣ 2A -and C 3.35 in ␤ 2 -AR) is not exposed in the binding pocket and thus not accessible for PB and other receptor ligands. Covalent Binding of PB to Cysteine-containing Receptor Peptide-To confirm the covalent bonding of PB to Cys 117 in TM3 of ␣ 2A -AR, PB was reacted in aqueous solution with a synthetic 9-mer peptide corresponding to residues 114 -122 of TM3 of the receptor. After 2 h of incubation, mass spectroscopic analysis revealed that all PB had hydrolyzed (MϩH ϩ m/z 286.4), acetylated in the employed acetate buffer (MϩH ϩ m/z 328.5), or reacted with the thiol group of the peptide (Mϩ2(H ϩ ) m/z 619.0, MϩH ϩ m/z 1236.1). The reference peptide with a valine substituted for the cysteine in position 117 did not react with PB. The identified reaction products are illustrated in Fig. 6.
Competition Binding Assays-All wild-type and mutated receptors were tested with PB also under competition binding assay conditions (Table I). At all receptor variants, PB inhibited specific binding of 2 nM [ethyl-3 H]RS79948 -197 or [ 3 H]CGP-12177 in the competitive assays with steep monophasic competition curves, indicating that the lack of alkylation by PB was not because of lack of binding affinity but rather because of the absence of an accessible cysteine in the binding cavity. The three PB-resistant receptors, ␣ 2A -C117V, ␣ 2A -F116C/C117V and ␤ 2 -AR, were also capable of binding PB, although with lower apparent affinity than the PB-sensitive receptors (Table I). Competition binding assays were also per-  formed with phentolamine, another subtype-nonselective ␣ 2 -AR antagonist. The substitution of Cys 117 of ␣ 2A -AR with valine resulted in 28 -45-fold decreases in the affinity of the receptor for phentolamine (Table I). Phentolamine was bound with low affinity to both wild-type ␤ 2 -AR and the ␤ 2 -V117C mutant receptor. DISCUSSION We have here demonstrated that an exposed cysteine residue, C 3.36 , in position 117/96/135 in the binding cavity of ␣ 2A / ␣ 2B /␣ 2C -AR is required for the alkylating effect of the irreversible ␣-AR antagonist, PB. The present work also shows that the depth of the ligand binding cavity extends to at least one helical turn below the conserved D 3.32 in the adrenergic receptors. Introduction of a valine into position 3.36 makes the ␣ 2A -AR resistant to PB (␣ 2A -C117V). The ␤ 2 -AR has a valine in this position and a cysteine in the preceding position and is resistant to the alkylating effect of PB. However, when V 3.36 was substituted by cysteine, the ␤ 2 -V117C receptor mutant became susceptible to irreversible inactivation by PB. It has been reported previously that position 116 in ␤ 2 -AR, i.e. C 3.35 , is not involved in recognition and binding of ␤-AR antagonists (21). Site-directed mutagenesis and modeling of the ␣ 1B -AR has also clearly indicated that the residue in position 3.35 points toward TM2 and is not exposed in the cavity (22,23). Our second PB-resistant ␣ 2A -AR mutant, ␣ 2A -F116C/C117V, also supports a hypothetical structural orientation of TM3, where position 116 (3.35) is not exposed in the binding cavity and is unreachable by PB. Position 3.36 in the middle of the TM3 segment has also been shown to be exposed in the receptor binding-site crevice of other non-adrenergic monoamine receptors, such as the dopamine D2 receptor (Cys 118 ) (25)(26)(27) and the 5-HT 2A receptor (Ser 159 ) (28).
Inactivation of the PB-resistant ␣ 2A -AR mutants with high PB concentrations indicated that there must be other sites in the receptor protein in addition to the investigated TM3 cysteine that are susceptible to alkylation by PB. Mass spectroscopic analysis of the reaction products obtained from incubation of PB with a synthetic peptide corresponding to this region of TM3 verified that PB indeed forms a covalent bond with the cysteine residue. No reaction product was generated when PB was incubated with a control peptide with a valine in place of the cysteine.
In the inactivation assay with 10 M PB, [ 3 H]CGP-12177 binding to the wild-type ␤ 2 -AR was inhibited by 26%, indicating that PB might have interactions with other sites in the ␤ 2 -AR binding cavity, lacking a cysteine in position 3.36. However, Gether et al. (21) have shown previously that none of the five cysteines located in the TM domains of the ␤ 2 -AR (Cys 77 , Cys 116 , Cys 125 , Cys 285 , and Cys 327 ) are essential for ␤ 2 -antagonist binding. Indeed, studies with the cysteine-reactive reagent 2-aminoethyl methane thiosulfnate indicated that none of these cysteines is exposed in the water-accessible binding site crevice (29). The observed effect of the employed high PB concentration may thus represent interactions with other reactive amino acid side chains than -SH of an exposed cysteine.
The mutant receptors bound the employed radioligands with similar affinities compared with the wild-type receptors, which verified the proper expression and folding of the modified proteins. Competition binding assays showed that the PB-resistant receptors ␣ 2A -C117V, ␣ 2A -F116C/C117V, and ␤ 2 -AR were also capable of binding PB, although with reduced apparent affinity, indicating that the lack of alkylation by PB was not because of lack of binding affinity. It should be noted that the true affinity of an irreversible ligand cannot be determined reliably in competition binding assays, and the apparent affinity of PB determined in this way actually represents both reversible and irreversible binding and grossly overestimates the affinity of the PB-susceptible receptors (16,24).
Cysteine substitutions in the binding cavity can, however, have structure-dependent selective effects on ligand binding affinities. That this indeed is the case for ␣ 2A -AR was indicated by our preliminary experiments with another ␣ 2 -AR radioligand, [ 3 H]RX821002. No specific binding was detected for this radioligand at ␣ 2 -C117V and ␣ 2 -F116C/C117V (results not shown), despite adequate receptor expression levels as determined in assays with [ 3 H]RS79948 -197. RX821002 is an imidazoline derivative that interacts with TM3 of ␣ 2 -ARs, and we therefore wanted to test another widely used imidazoline derivative, phentolamine, as a reference compound. Phentolamine does not bind irreversibly to adrenoreceptors, and its true affinity can be determined in competition binding assays. The affinity of phentolamine at mutated ␣ 2A -ARs lacking C 3.36 was clearly lower than at the wild-type receptor. This indicates that C 3.36 is important for binding of imidazoline derivatives and further supports the location of this residue in the binding cavity. PB is not an imidazoline, and its binding mode in the ␣ 2A -AR binding cavity is different from phentolamine and RX821002. Our subsequent competition binding assays with structurally diverse ␣ 2 -AR ligands indicate that the effects of the C 3.36 substitution are specific for imidazoline and imidazole derivatives. 2 Using irreversible binding of the ␣-AR antagonist PB as a criterion for identifying a sulfhydryl side chain of an endogenous or introduced cysteine as being exposed in the ligandaccessible binding cavity of the receptor, we have demonstrated that C 3.36 (Cys 117 ) in the TM3 domain of the ␣ 2A -AR is responsible for binding PB. This cysteine also appears to be important in interactions with other receptor antagonists, as indicated by the observed decrease in the affinity for the imidazoline derivative phentolamine when C 3.36 was substituted with valine. The structural orientation of the TM3 domain of ␣ 2A -AR, where amino acid residues D 3.32 and C 3.36 point into the binding cavity, and F 3.35 is not exposed in the cavity, may, however, represent only one receptor conformation. All G-protein-coupled receptors are thought to exist in an equilibrium between two or more conformations or allosteric states, R and R* (29 -31). Being antagonists, the ligands tested in this study mainly interact with the receptor in its predominant, inactive conformation (R) (30). The orientation of the TM domains may change upon receptor activation and agonist binding (1,21,30). Nevertheless, results obtained with catecholamine agonists (9,22,32) indicate that D 3.32 indeed is accessible also in the active conformation (R*), and if TM3 has typical ␣-helical periodicity, then C 3.36 would be expected to be exposed also in the active conformation (R*).