Abstract
The α2-adrenergic receptors (α2-ARs) mediate signals to intracellular second messengers via guanine nucleotide binding proteins. Three human genes encoding α2-AR subtypes (α2A, α2B, α2C) have been cloned. Several chemical compounds display subtype differences in their binding and/or functional activity. Site-directed mutagenesis and molecular modeling are new tools with which to investigate the subtype selectivity of ligands. In this study, we introduce a new approach to mapping of the binding site crevice of the human α2A-AR. Based on a three-dimensional receptor model, we systematically mutated residues 197–201 and 204 in the fifth transmembrane domain of the human α2A-AR to cysteine. Chloroethylclonidine, an alkylating derivative of the α2-adrenergic agonist clonidine, binds irreversibly to α2A-ARs by forming a covalent bond with the sulfhydryl side chain of a cysteine residue exposed in the binding cavity, leading to inactivation of the receptor. Irreversible binding of chloroethylclonidine was used as a criterion for identifying introduced cysteine residues as being exposed in the binding cavity. The results supported a receptor model in which the fifth transmembrane domain is α-helical, with residues Val197, Ser200, Cys201, and Ser204 exposed in the binding pocket. Residues Ile198, Ser199, Ile202, and Gly203 face the lipid bilayer of the plasma membrane. This approach emerges as a powerful tool for structural characterization of the α2-ARs.
The α2-ARs mediate diverse physiological and pharmacological effects of the neurotransmitters/hormones norepinephrine and epinephrine and related synthetic molecules. Three genes encoding human α2-AR subtypes have been cloned, representing the pharmacologically defined subtypes α2A, α2B, and α2C (Kobilka et al., 1987; Reganet al., 1988; Lomasney et al., 1990). Related α2-AR genes also have been identified in other species, such as rat, mouse, pig, opossum, and fish (Guyer et al., 1990; Zeng et al., 1990; Lanier et al., 1991; Chen et al., 1992; Link et al., 1992;Svensson et al., 1993; Blaxall et al., 1994). α2-ARs, like all other members of the GPCR family, consist of a polypeptide chain that is predicted to span the cell membrane seven times. The amino acid sequences within the seven hydrophobic TMs are highly conserved in the three α2-AR subtypes. These TM regions are predicted to be α-helical and to form a pocket crucial for the identification and binding of ligand molecules. Binding of a receptor agonist in this binding cavity either leads to or stabilizes a conformational change in the receptor protein, promoting its coupling with G proteins. The resulting G protein activation initiates a cascade of intracellular biochemical events and physiological responses (Savarese and Fraser, 1992; Scheer et al., 1996).
Several α2-AR ligands, such as oxymetazoline, chlorpromazine, prazosin, UK 14,304, and dexmedetomidine, display some degree of subtype selectivity in either their binding affinity or functional activity (Marjamäki et al., 1993; Janssonet al., 1994). A comparison of the ligand binding properties of the human α2-AR subtypes with their species homologues also has revealed some differences. For example, Hα2A binds the antagonists yohimbine and rauwolscine with significantly higher affinity than its mouse homologue, Mα2A (Link et al., 1992). Analysis of the primary structures of these two receptors has identified a Cys201-to-Ser201 substitution in the TM5 of Mα2A. When Ser201 of the Mα2A was mutated to cysteine, the affinity of the mouse receptor for yohimbine was significantly increased. This suggested that the residue at position 201 in TM5 of α2A-ARs might be exposed in the binding cavity and directly participate in ligand recognition. Site-directed mutagenesis and computer-aided modeling can be used to explore the structural determinants of receptor/ligand interactions, including species differences and subtype selectivity. Mapping of residues exposed in the binding cavity may allow the subsequent synthesis of new therapeutic agents targeted to specific ligand recognition sites.
CEC, which often has been used to discriminate between α1-AR subtypes in functional assays (Hanet al., 1987; Tian et al., 1990), has been shown to inactivate irreversibly Hα2A and Hα2C, whereas Hα2B is relatively resistant to its alkylating effect (Michel et al., 1993). CEC is known to undergo intramolecular cyclization to a reactive aziridinium ion before irreversible receptor inactivation (Vargas et al., 1993). The aziridinium ion presumably forms a covalent bond with the free SH-group of an exposed cysteine residue. The primary structure of Hα2A has a cysteine in position 201; Hα2C also has a cysteine in the corresponding position (position 215), whereas the CEC-resistant subtype Hα2B has a serine in the corresponding position (position 177) (Fig.1). Such an amino acid substitution might explain the subtype-selective reactivity of CEC at the different human α2-AR subtypes. To test this hypothesis, we determined the irreversible binding of CEC to the three human α2-AR subtypes as well as the Mα2A and constructed and tested a series of mutant α2A-ARs with cysteines located at different positions in this region of TM5.
Computer-aided modeling was used to predict the three-dimensional structure of the Hα2A. The TM domains of GPCRs usually are presented as fixed α-helices, with one side exposed in the binding cavity (Savarese and Fraser, 1992; Baldwin, 1993; Schwartz, 1994). With site-directed mutagenesis, however, the pattern of exposure of residues in TM5 of the dopamine D2 receptor to a hydrophilic thiol-reactive alkylating agent was shown to be inconsistent with this prediction (Javitch et al., 1995). In our model of the Hα2A, TM5 was predicted to be α-helical, with residues Val197, Ser200, Cys201, and Ser204 forming part of the surface of the ligand-accessible binding site crevice and residues Ile198, Ser199, Ile202, and Gly203 facing the lipid bilayer of the plasma membrane. To map the structure and orientation of the TM5 in the Hα2A and to test this model, we mutated residues 197–201 and 204, one at a time, to a cysteine. Irreversible binding of CEC was used as a criterion for identifying a sulfhydryl side chain of an introduced cysteine as being exposed in the binding cavity and accessible to CEC. Our results confirmed the α-helical structure and predicted rotational orientation of TM5 in Hα2A.
Experimental Procedures
Materials.
[3H]RX821002 [2-(2-methoxy-1,4-benzodioxan-2-yl)-2-imidazoline] was from Amersham International (Buckinghamshire, UK; specific activity, 52 Ci/mmol). Phentolamine and CEC were from Research Biochemicals (Natick, MA). Cell culture reagents were supplied by GIBCO (Gaithersburg, MD). The 10-mer oligopeptide Tyr-Val-Ile-Ser-Ser-Cys-Ile-Gly-Ser-Phe (TM5 region of Hα2A: Tyr196 to Phe205) was supplied by the Center for Biotechnology (Turku, Finland). Other chemicals were of analytical grade and were purchased from commercial suppliers.
Reaction of CEC with oligopeptide and mass spectroscopic analysis.
The 10-mer oligopeptide Tyr-Val-Ile-Ser-Ser-Cys-Ile-Gly-Ser-Phe was dissolved in 50 mm K+-phosphate buffer, pH 7.4, at 21°, and one molar equivalent of CEC was added. The reaction mixture was incubated for 60 min at 37° and analyzed by matrix-assisted laser desorption mass spectrometry (Finnigan MAT, Hemel Hempstead, UK).
Mutagenesis and expression vectors.
The cDNA encoding Hα2A (Kobilka et al., 1987) was inserted into the SmaI site of the vector pALTER-1 (Promega, Madison, WI). Site-directed mutagenesis was performed using the Altered Sites II In Vitro Mutagenesis System (Promega). The mutated DNA fragments were sequenced manually by dideoxy sequencing of double-stranded DNA with Sequenase (United States Biochemical, Cleveland, OH) and confirmed with an ABI377 automated sequencer (Perkin-Elmer Cetus (Norwalk, CT). The WT Hα2A and the mutated receptor cDNAs were subcloned into theKpnI/BamHI sites of the expression vector pREP4 (InVitrogen, NV Leek, The Netherlands), which also contains the gene for hygromycin B resistance.
The cDNAs encoding Hα2B, Hα2C, Mα2A (Regan et al., 1988; Lomasney et al., 1990; Link et al., 1992) and the S201C mutant of Mα2A, created and confirmed as described, were similarly subcloned into the pREP4 expression vector for receptor production.
Cell culture and transfections.
Adherent CHO cells (American Type Culture Collection, Rockville, MD) were cultured in α-minimum essential medium supplemented with 2 mm glutamine, 20 mm NaHCO3, 5% heat-inactivated fetal calf serum, penicillin (50 units/ml), and streptomycin (50 μg/ml). Cells were grown in 5% CO2 at 37°. The pREP4-based expression constructs were transfected into CHO cells with use of the Lipofectin reagent kit (GIBCO, Paisley, UK). For each transfection, we used 3 μg of plasmid DNA/5 × 104 cells. Hygromycin B (Boehringer-Mannheim Biochemica, Mannheim, Germany)-resistant (550 μg/ml) cell cultures were examined for their ability to bind the α2-AR antagonist [3H]RX821002. The transfected cells chosen for further experiments were subsequently maintained in 200 μg/ml hygromycin B.
Receptor inactivation and ligand binding.
Cells were harvested into chilled phosphate-buffered saline, pelleted, washed, suspended in ice-cold 50 mmK+-phosphate buffer, pH 7.4, at 21°, and homogenized with an Ultra-Turrax homogenizer (model T25, Janke & Kunkel, Staufen, Germany; setting, 9500 rpm, twice for 10 sec). The homogenate was used for saturation and competition binding assays or receptor inactivation experiments.
Saturation studies were performed in K+-phosphate buffer as described previously (Halme et al., 1995). Whole-cell homogenates containing 40–80 μg of protein were incubated with [3H]RX821002 (0.125–8 nm). Specificity of binding was defined with 10 μmphentolamine. Competition studies were done as reported previously(Halme et al., 1995), using [3H]RX821002 concentrations close to its affinity constant (Kd ) at each receptor and 13–15 concentrations of the competitor CEC.
For receptor inactivation, cell homogenates first were incubated with CEC (1 and 10 μm) in 2.5 ml of K+-phosphate buffer for 15, 30, or 60 min at 37°. The protein content was 0.3–0.5 mg/ml during CEC treatment. Next, membranes were pelleted at 40,000 × g for 15 min at 4°, washed twice with 2.5 ml of ice-cold K+-phosphate buffer, and rehomogenized with the Ultra-Turrax homogenizer. Residual α2-AR binding was assessed by incubating the homogenate (0.1–0.2 mg/assay tube) with 2.5 nm [3H]RX821002. Nonspecific binding was determined by including 10 μmphentolamine in parallel assays.
Three-dimensional modeling of Hα2A binding cavity.
The molecular modeling of Hα2A and the binary complex with CEC will be described in complete detail (V. Cockcroft, A. Marjamäki, H. Frang, M. Pihlavisto, J.-M. Savola, and M. Scheinin, Ligand interaction of serine-cysteine amino acid exchanges in TM5 of αZ-adrenergic receptors, manuscript in preparation.). The structural coordinates of the high-resolution electron cryomicroscopy model of bacteriorhodopsin (Henderson et al., 1990) was used as a three-dimensional template for structural mapping of GPCR sequences.
Results
Site-directed mutagenesis and transfections.
To examine the structure of the TM5 domain of Hα2A, amino acid residues from Val197 to Cys201 and Ser204 were mutated to introduce or delete cysteines. The introduced mutations were confirmed and the absence of secondary mutations was verified by dideoxy sequencing of double-stranded DNA.
Mutated and WT receptors were expressed in CHO cells. Hygromycin B-resistant cell cultures were examined for their ability to bind the α2-AR antagonist radioligand [3H]RX821002. Three cell lines from each transfection expressing the expected receptor were isolated for preliminary experiments, and one cell line from each transfection was expanded for further experiments (Table1) and subsequently maintained in 200 μg/ml hygromycin B.
Receptor inactivation studies.
CEC, an alkylating derivative of clonidine, has been used previously to discriminate between α1-AR subtypes, but it also has been shown to inactivate α2-ARs in a subtype-selective manner. Based on our hypothesis, the reactive aziridinium ion derivative of CEC forms a covalent bond with an exposed SH-group of a cysteine residue in the receptor molecule (Fig.2) and inactivates the receptor by steric blockade of the binding cavity. Covalent bonding of CEC to protein was confirmed by allowing it to react with a synthetic 10-mer oligopeptide corresponding to residues 196–205 of the TM5 region of the Hα2A and then undergoing mass spectroscopic analysis (Fig.3). After 1 hr at 37°, the oligopeptide was totally alkylated in a manner consistent with our hypothesis presented in Fig. 2.
To validate our experimental conditions in the CEC inactivation assay, we first compared the effect of CEC treatment at 37° for 15, 30, or 60 min, followed by two washes, on the capacity of [3H]RX821002 binding in CHO cell homogenates expressing WT Hα2A (data not shown). The incubation of cell homogenates for 30 min at 37° in the absence (control) and presence of CEC was chosen as optimal for further experiments.
First, we tested the alkylating effect of CEC on the three human α2-AR subtypes (Hα2A, Hα2B, and Hα2C) expressed in CHO cells. CEC treatment reduced the binding capacity of Hα2A and Hα2C by 85%, whereas Hα2B was resistant to the alkylating effect of CEC (Fig. 4). This was in agreement with the involvement of a cysteine in position 201 or in a corresponding position in the alkylating effect of CEC (see amino acid sequence alignment in Fig. 1). To further characterize the interaction of TM5 cysteines and CEC, we compared the effects of CEC treatment on Hα2AWT and Mα2AWT. Instead of a cysteine, the Mα2AWT contains a serine in position 201 (Fig. 1). Incubation with CEC inactivated 75% of Hα2AWT but only 23% of Mα2A was irreversibly inactivated. When the Cys201 of Hα2A was mutated to a serine to resemble the Mα2A, it became resistant to the alkylating effect of CEC (inactivation, 15%). After the opposite mutation in Mα2A (Ser201 to cysteine), this receptor became susceptible to the irreversible effect of CEC (inactivation, 60%) (Fig.5). This confirms our hypothesis of a structure-activity relationship between the alkylating effect of CEC and a cysteine residue in this position of TM5.
In our three-dimensional receptor model (Fig.6), the amino acid residues Val197, Ser200, Cys201, and Ser204 were accessible and exposed in the binding cavity, whereas Ile198, Ser199, Ile202, and Gly203 were facing the lipid bilayer in an α-helical TM5 of the Hα2A. To map the surface of the ligand binding pocket, we systematically mutated residues from Val197 to Ser200 and Ser204 to cysteine. Before introducing new cysteine residues to the TM5 of the Hα2A, the WT Cys201 of Hα2A was substituted with serine. This Hα2ASer201 is resistant to the alkylating effect of CEC (Fig. 5) and was used as a negative control in these experiments. We investigated the capability of CEC to inactivate WT and mutated receptors expected to contain a cysteine residue exposed in the binding crevice (Fig.7). Relative to the WT Hα2A (inactivation, 75%), the extent of inactivation was smaller when the cysteine residue was deeper in the cavity (Ser201Cys204 mutant inactivation, 52%) and greater when the residue was closer to the extracellular surface of the plasma membrane (Ser201Cys197 mutant inactivation, 97%). This probably was due to different rates of alkylation of the receptors under our assay conditions. After a 60-min CEC treatment, the difference in the extent of receptor inactivation was minimal (Ser201Cys197, Cys201, and Ser201Cys204 inactivation, 96%, 92%, and 86% respectively), and it seems that all accessible cysteines ultimately would be alkylated, given sufficient time.
Although in our model Ser200 is pointing partly toward the TM4 domain, the aliphatic hydroxyl side chain of this residue can rotate toward the cavity and thus participate in ligand recognition. CEC treatment reduced the number of detectable α2-ARs in CHO cell homogenates expressing the Hα2ASer201Cys200 mutant by 61%, indicating the SH side chain of Cys200 also is exposed in the cavity. The difference in the orientation of the residues at positions 200 and 201 also might account for the difference in the extent of receptor inactivation between Hα2AWT and Hα2ASer201Cys200 (inactivation, 75% versus 61%) (Fig. 7).
Amino acids from Val197 to Cys201 represent one full turn in the α-helical model of TM5. The residues 197–200 of Hα2ASer201 were mutated to cysteine, one at a time, and the effect of CEC treatment on the binding activity of the WT Hα2A and the mutant receptors was examined (Fig. 7). Two cysteine residues at positions 198 and 199, expected to face the lipid bilayer of the cell membrane, were relatively resistant to the alkylating effect of CEC (Ser201Cys198 and Ser201Cys199 inactivation, 25% and 24%, respectively). The results obtained with site-directed mutagenesis thus support our three-dimensional model and confirm the α-helical structure of TM5 in Hα2A.
Ligand binding assays.
Saturation isotherms of [3H]RX821002 binding- and LIGAND- (McPherson, 1985) derived Kd (receptor affinity) and B max (receptor density) values were determined in three separate experiments for each cell line (Table 1). Three-point mutations of Hα2ASer201, Ser199, Ser200, and Ser204 to cysteine, resulted in 3–5-fold decreases in receptor affinity for the α2-AR antagonist [3H]RX821002. The expression level of the Hα2ASer201 mutant used in our experiments was only 156 ± 13 fmol/mg of total cellular protein. Similar results of receptor inactivation by CEC were, however, later obtained in experiments with another batch of Hα2ASer201 cells, expressing 4736 ± 234 fmol/mg of cell homogenate (inactivation, 7 ± 2%). This indicates that the weak alkylating effect of CEC on Hα2ASer201 presented in Fig. 5 is not dependent on the receptor expression level.
In all investigated cell lines expressing WT and mutant receptors, the addition of CEC to competition binding assays inhibited specific binding of 2.5 nm [3H]RX821002 with steep monophasic competition curves. The affinity of Hα2B for [3H]R821002 was relatively low (Kd = 6.12 ± 0.46 nm), and the receptor inactivation assays consequently were performed using 6 nm[3H]RX821002. Similar results were obtained with both radioligand concentrations (9 ± 2% and 13 ± 2% inactivation with 2.5 and 6 nm[3H]RX821002, respectively). We tested whether the lack of alkylating effects of CEC (Figs. 4, 5, and 7) would be due to low or absent binding affinity of CEC to Hα2B, Mα2A, or the Hα2A mutants Hα2ASer201, Hα2ASer201Cys198 ,and Hα2ASer201Cys199 (Table 1). The two CEC-resistant WT receptors Hα2B and Mα2A and the Hα2ASer201, Hα2ASer201Cys198, and Hα2ASer201Cys199 mutants also were capable of binding CEC (apparentKi = 1016 ± 115, 539 ± 46, 260 ± 32, 467 ± 49, and 2624 ± 96 nm, respectively). The lack of alkylation thus is not due to lack of binding affinity but rather to the absence of an accessible cysteine residue on the surface of the binding site crevice.
Discussion
In the current study, we were able to demonstrate that an exposed cysteine residue in the binding cavity of α2-AR is required for the alkylating effect of CEC. Although the apparent binding affinities (apparent Ki value) of CEC were comparable for the WT and mutated Hα2A-ARs, the alkylating effect of CEC treatment was dependent on the location of a reactive cysteine residue in TM5. True affinity of an irreversible ligand cannot be determined reliably in a conventional competition binding assay, and the apparent affinity of CEC determined in this way actually may represent both reversible and irreversible binding (Michelet al., 1993). Simultaneous inactivation and competitive binding should, however, overestimate the apparent affinity of CEC for α2-AR subtypes/mutants that become alkylated, indicating the lack of alkylating effects of CEC in our assays is not due to the lack of CEC binding.
We used a receptor model predicting the α-helical structure of TM5 in Hα2A-AR, in which the residues Val197, Ser200, Cys201, and Ser204 were pointing to the binding pocket, whereas the residues Ile198, Ser 199, Ile202, and Gly203 were facing the lipid bilayer. This model was supported by the results obtained through site-directed mutagenesis and CEC inactivation experiments. The primary structures of the TM5 regions of all α2-AR subtypes contain a cysteine in the position corresponding to Cys209 of Hα2A (Fig. 1). This cysteine is facing the lipid bilayer of the plasma membrane in our receptor model and thus was not expected to interfere with CEC inactivation experiments. This orientation of Cys209 was supported by the results obtained with Hα2B, Mα2A, Hα2ASer201, Hα2ASer201Cys198, and Hα2ASer201Cys199 not containing cysteine residues exposed in the binding cavity and shown to be resistant to the alkylating effect of CEC (Figs. 4, 5, and 7). These results are consistent with an α-helical structure of TM5 and provide constraints for the rotational orientation of this helix in relation to the binding cavity.
The β2-AR is one of the most extensively structurally characterized GPCRs. With site-directed mutagenesis, it has been possible to identify several amino acid residues that are critical for and probably directly involved in ligand binding. The catechol hydroxyl groups of epinephrine seem to interact with two serine residues present in TM5 of β2-AR (Strader et al., 1989). The two serine residues at corresponding positions of the human α2A-AR (Ser200 and Ser 204) have been shown to participate in hydrogen bond interactions with the catechol hydroxyl groups of catecholamine agonists (Wang et al., 1991). When Ser201 of the mouse α2A-AR was mutated to cysteine, the corresponding amino acid in Hα2A, this cysteine-to-serine substitution was shown to be critical for the low affinity of the mouse receptor for yohimbine and rauwolscine (Link et al., 1992). Site-directed mutagenesis and analysis of the three-dimensional model of the hamster α1B-AR also indicated that three serine residues, corresponding to positions 200, 201, and 204 in Hα2A, are important for agonist interactions (Cavalli et al., 1996). When the structural determinants of subtype-selective agonist binding of the hamster α1B-AR were identified, it was shown that mutation of Ala204 to valine (corresponding to position 197 in Hα2A) in TM5 conferred onto α1B-AR the binding properties of the α1A-AR (Hwa et al., 1995). These results obtained with computer-aided modeling and site-directed mutagenesis from different members of the same receptor family support our model and the location of residues 197, 200, 201, and 204 of Hα2A in the binding site cavity.
We introduced a new useful approach to mapping of the binding site crevice of human α2-ARs by using cysteine substitution mutagenesis and irreversible cysteine-specific covalent binding of CEC to the receptor. This method emerges as a very useful tool for structural characterization of the α2-ARs. Using this method together with three-dimensional modeling, we were able to confirm the predicted α-helical structure of TM5 and its orientation in Hα2A. One of the goals of the current study was to assess whether a so-called molecular yardstick approach could provide distance constraints for determining improved models of GPCRs. This technique, perhaps with some modifications, also could be applicable to structural studies on TM4 and TM6 of the same receptor.
Javitch et al. (1995) and Fu et al. (1996)previously introduced a cysteine-reactive approach using as reactive agents nonspecific polar methanethiosulfonate derivatives to probe another monoamine receptor, the D2 dopamine receptor. Our method can be seen as a simple development of this approach. We suggest our method has the advantage of introducing recognition specificity by using a thiol-reactive group incorporated into an affinity ligand of the target receptor. This allows the use of significantly lower reagent concentrations and lessens the possibility of indirectly blocking radioligand binding.
Conventional loss-of-function mutagenesis has not always produced definitive answers for the purpose of assigning amino acid residues as being inside the binding cavity. It has been shown that point mutations of positions expected to be outside the binding cavity can have a marked effect on ligand affinities, presumably through conformational control (Fong et al., 1992; Sachais et al., 1993). In addition, several reports exist of residue substitutions at sites at which ligand/receptor contact interaction would be expected on the basis of amino acid conservation patterns that have not shown expected disruption of ligand binding (Befort et al., 1996). Although the receptor modeling presented here is still rather crude, it has been used to devise an experimental approach, gain-of-function mutagenesis, to provide, through covalent bond formation, solid information that will lead to better models of not only the receptors but also binary complexes of ligand and receptor.
Footnotes
- Received July 21, 1997.
- Accepted October 8, 1997.
-
Send reprint requests to: Dr. Anne Marjamäki, MediCity Research Laboratory, City of Turku, Tykistökatu 6 A, FIN-20520, Turku, Finland. E-mail: anmarja{at}utu.fi
-
↵1 Current affiliation: Juvantia Pharma, Tykistökatu 6 A, FIN-20520 Turku, Finland.
-
This work was supported by the Academy of Finland and Technology Development Centre of Finland.
Abbreviations
- AR
- adrenergic receptor
- CEC
- chloroethylclonidine
- CHO
- Chinese hamster ovary
- GPCR
- G protein-coupled receptor
- SH
- sulfhydryl
- TM
- transmembrane domain
- Hα2A
- human α2A-adrenergic receptor
- Hα2B
- human α2B-adrenergic receptor
- Hα2C
- human α2C-adrenergic receptor
- Mα2A
- mouse α2A-adrenergic receptor
- WT
- wild-type
- The American Society for Pharmacology and Experimental Therapeutics