Identification of structurally distinct alpha 2-adrenergic receptors.

Recent studies involving a variety of membrane receptors and ion channels indicate that diversity exists among these proteins as evidenced by tissue-specific and developmentally related expression of different isoforms. Alpha 2-Adrenergic receptors, plasma membrane proteins involved in sympathetic neurotransmission, may similarly represent a nonhomogeneous class of binding sites based on the following observations. First, their activation can elicit a wide variety of effector cell responses, which are apparently triggered by at least three different signal transduction mechanisms. Second, alpha 2-adrenergic receptors in various tissues and species exhibit marked differences in their ligand recognition properties. To determine if heterogeneity of the receptor protein itself is involved in generating this diversity, we structurally characterized the alpha 2-adrenergic receptor in two tissues that exhibit the greatest differences in ligand recognition properties, neonatal rat lung and human platelet. We report here that these differences in ligand recognition are maintained after partial receptor purification (50-100-fold) and are associated with distinct differences in the physical and structural properties of the receptor protein. The human platelet and neonatal rat lung receptor differ in the apparent molecular weight of their hormone-binding subunits (human platelet, Mr approximately 64,000 versus neonatal rat lung, Mr approximately 44,000) as well as in the number or type of their associated oligosaccharide moieties. The observed diversity is consistent with expression of isoforms of the alpha 2-adrenergic receptor and suggests the presence of more than one gene encoding similar but distinct receptor proteins.

Recent studies involving a variety of membrane receptors and ion channels indicate that diversity exists among these proteins as evidenced by tissue-specific and developmentally related expression of different isoforms. az-Adrenergic receptors, plasma membrane proteins involved in sympathetic neurotransmission, may similarly represent a nonhomogeneous class of binding sites based on the following observations. First, their activation can elicit a wide variety of effector cell responses, which are apparently triggered by at least three different signal transduction mechanisms. Second, a2-adrenergic receptors in various tissues and species exhibit marked differences in their ligand recognition properties. To determine if heterogeneity of the receptor protein itself is involved in generating this diversity, we structurally characterized the az-adrenergic receptor in two tissues that exhibit the greatest differences in ligand recognition properties, neonatal rat lung and human platelet. We report here that these differences in ligand recognition are maintained after partial receptor purification (50-100-fold) and are associated with distinct differences in the physical and structural properties of the receptor protein. The human platelet and neonatal rat lung receptor differ in the apparent molecular weight of their hormone-binding subunits (human platelet, M, -64,000 versus neonatal rat lung, M, -44,000) as well as in the number or type of their associated oligosaccharide moieties. The observed diversity is consistent with expression of isoforms of the az-adrenergic receptor and suggests the presence of more than one gene encoding similar but distinct receptor proteins.
a2-Adrenergic receptor (a2AR)l activation results in a variety of tissue-specific effects including smooth muscle contraction, epithelial cell chloride secretion, and platelet aggregation/secretion, as well as inhibition of insulin release, lipolysis, and depolarization-induced neurotransmitter release (1). Recent studies suggest that these effects may involve of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed.  1-3). Thus, many effects elicited by a2AR activation are apparently unrelated to inhibition of adenylate cyclase and may involve coupling to a Na+/H+ antiporter or to ion channels. In addition to these differences in effector cell responses, a2AR in various tissues and species exhibit differing affinities for a number of ligands including clonidine, idazoxan, prazosin, and oxymetazoline, but maintain similar high affinity for the a2-selective antagonist rauwolscine (1,(4)(5)(6). The present study was undertaken to determine if the observed differences in ligand recognition properties, which range from 10-to 100fold, are associated with distinct structural differences in the a2AR expressed in these tissues.

Materials-[3H]
Rauwolscine (80 Ci/mmol) was obtained from Du Pont-New England Nuclear. Heparin-agarose (Type 11) was purchased from Sigma. Wheat germ-lectin agarose was from Vector Laboratories (Burlingame, CA). N-Glycanase was obtained from Genzyme Corp. (Boston, MA). Prazosin was a gift from Pfizer, and oxymetazoline was provided by Schering Corp. (Bloomfield, NJ). Rats were obtained from Charles River Breeding Laboratories (Wilmington, MA). Outdated human platelets were purchased from the Massachusetts General Hospital blood bank. All other chemicals and materials were obtained or synthesized as described previously (7,8).
Membrane Preparation and Partial Receptor Purification-Rat lung and brain tissue were removed from 1-day-old rat pups and immediately frozen in liquid nitrogen. Frozen rat lung tissue obtained from 50-75 1-day-old pups was thawed and minced in ice-cold buffer containing 250 mM sucrose, 10 mM Tris, 5 mM EDTA, 2 mM EGTA, pH 7.4. Tissue was disrupted with a Polytron tissue homogenizer (2 X 10 s, setting 8; Brinkmann Instruments) and then with a Dounce homogenizer after which membrane aggregates and nuclear material were pelleted by centrifugation at 2,400 X g (Sorvall RCBB, type SS34 rotor). The supernatant was decanted and centrifuged at 24,000 X g for 10 min to obtain the membrane pellet, which was washed twice and resuspended by homogenization in 50 mM Hepes, 5 mM EDTA, 2 mM EGTA, 150 mM NaCl, pH 7.6.
Frozen rat brain tissue was thawed and minced in buffer containing 50 mM Tris, 5 mM EDTA, 2 mM EGTA, pH 7.5, followed by disruption with a Polytron tissue homogenizer. The disrupted tissue was pelleted by centrifugation at 39,000 X g for 10 min and rehomogenized into buffer containing 600 mM KCl, 20 mM imidazole, 5 mM EDTA, 2 mM EGTA, pH 7.5. The homogenate was then pelleted and washed twice with buffer containing 50 mM Hepes, 5 mM EDTA, 2 mM EGTA, 100 mM NaCl, pH 7.6, prior to final resuspension in the same buffer. Platelet membranes were prepared from outdated platelets (1-day postexpiration date) as described previously (8), except that the crude membrane fraction was not further purified by sucrose gradient centrifugation.
Membranes were solubilized at 4 "C for 15 min in buffer containing 50 mM Hepes, 5 mM EDTA, 2 mM EGTA, 100 mM NaCI, 1% digitonin, pH 7.4, at a detergent to protein ratio of 3:l. The soluble fraction was isolated by centrifugation (100,000 X g ) and subjected to heparinagarose affinity chromatography (10 volumes of detergent-solubilized preparation to 1 volume of heparin-agarose; flow rate, 2.5 column volumes h-'1. The column was then washed with 50 column volumes of 50 mM Hepes, 5 mM EDTA, 2 mM EGTA, 150 mM NaC1, 0.2% digitonin, pH 7.4, after which the absorbance of the flow-through was 0 at 280 nm. The receptor protein was then eluted with wash buffer containing 700 mM NaCl at a flow rate of 2 column volumes h-I, and the collected fractions were assayed with [3H]rauwolscine for receptor binding activity. This technique resulted in approximately a 50-fold purification of the receptor protein (peak specific activity, 1.5-3 pmol mg" of protein) with 75% recovery, based on receptor binding in the crude solubilized preparation. All procedures were conducted at 4 "C, and all buffers contained EDTA (5 mM), EGTA (2 mM), and the following protease inhibitors: pepstatin (2 pg ml-'1, leupeptin (2 pg ml-I), antipain (2 pg ml-I), soybean trypsin inhibitor (5 pg ml-I), benzamidine (10 pM), phenylmethylsulfonyl fluoride (100 pM).
In some experiments, human platelet membranes (80 ml, 560 mg of protein, -100 pmol of receptor) were mixed with frozen intact rat lungs. The mixed tissue preparation was then processed as described above for rat lung membranes. In these experiments the amount of platelet tissue utilized was approximately equal to the calculated yield of rat lung membranes. The mixed membrane preparation was then solubilized and subjected to heparin-agarose affinity chromatography.
For lectin affinity chromatography the elutions from the heparinagarose resin were pooled (10 ml), and magnesium was added to 10 mM prior to wheat germ-lectin affinity chromatography. One ml of resin was initially equilibrated with 20 column volumes of 50 mM Hepes, 5 mM EDTA, 2 mM EGTA, 700 mM NaCl, 0.2% digitonin, pH 7.4 (Buffer A) followed by 5 column volumes of the same buffer containing 10 mM MgC12. The solubilized receptor preparation was then loaded onto the resin by gravity flow and the resin subsequently washed with 10 column volumes of Buffer A containing 10 mM MgC12 followed by 5 column volumes of Buffer A containing 100 mM NaCl without MgC12. Bound glycoproteins were then eluted by batch technique with 3 column volumes of the second wash buffer containing 300 mM N-acetylglucosamine.
Binding Studies-Membrane binding assays were conducted at 24 "C as described previously (8). For solubilized receptor binding assays, increasing concentrations of competing ligand were incubated in duplicate with receptor (10-20 fmol) and [3H]rauwolscine (10 nM) for 2 h at 4 "C, and bound ligand was separated by precipitation with bovine y-globulin/polyethylene glycol followed by vacuum filtration. The filters were placed in 10 ml of Hydrofluor scintillation fluid and counted in a liquid scintillation spectrometer (Beckman, model LS 1800) with a counting efficiency of 50%. Nonspecific binding was determined in the presence of 10 p~ yohimbine. ICm values are defined as the concentration of competing ligand inhibiting 50% of specific [3H]rauwolscine binding and are presented as the mean f 1 S.E.
Photoaffinity hbeling and SDS-PAGE-Aliquots of receptor (-100 fmol) eluted from heparin-agarose or wheat germ-agarose chromatographic steps were photolabeled with 'Z51-rau-AZPC as described previously (7). The photolyzed samples were then desalted, lyophilized, and subsequently solubilized in sample buffer prior to SDS-PAGE (7). Gels were dried, and autoradiographs were obtained by exposing the dried gels to Kodak XAR-5 film at -70 "C in a cassette containing Du Pont Cronex Lightning Plus intensifying screens.

Ligand Recognition Properties of the Partially Purified
Receptor Protein-In membranes prepared from human platelets and neonatal rat lung, the a2AR exhibits markedly different affinity for the a-adrenergic receptor ligands, prazosin and oxymetazoline, as previously observed by Bylund (6). To characterize the binding properties of the receptor in these two tissues in more detail and to eliminate potential differences due to membrane environment, the a2AR from these two tissues was solubilized and the receptor protein purified -50-fold by heparin-agarose affinity chromatography. As shown in Table I, the differences in ligand recognition properties observed in membrane preparations were maintained after solubilization and partial purification of the receptor protein. Thus, the human platelet and neonatal rat lung a2AR exhibited similar high affinity for rauwolscine but differed by -100-fold in their affinity for prazosin and oxymetazoline. In similar preparations of rat brain cortex, the a2AR exhibited affinities for oxymetazoline and prazosin, which were intermediate between those observed in human platelet and neonatal rat lung receptor preparations (Table I). Attempts to  characterize the binding properties of the solubilized a2AR from adult rat lung were unsuccessful due to low levels of specific [3H]rauwolscine binding.
Identification of the Hormone Binding Subunit-To determine if the differences in ligand recognition properties of the human platelet and neonatal rat lung a2AR are associated with structural variations of the receptor protein, a photoaffinity probe specific for the a2AR, 17a-hydroxy-20a-

yohimban-16~-[N-(4-azido-3-['251]iodo)phenyl] carboxamide
('251-rau-AZPC) (7-lo), was used to label covalently the receptor's hormone-binding subunit. In neonatal rat lung preparations, '251-rau-AZPC labels a major peptide species with an apparent molecular weight of -44,000 (Mr = 44,000 & 750, n = 5 ) (Fig. 1A). The labeling is inhibited by the selective cy2receptor antagonist rauwolscine and in a stereoselective manner by the a-agonist epinephrine. In contrast, the photolabeled human platelet a2AR migrates as a broad species with an apparent molecular weight of -64,000 (MI = 64,000 & 1,200, n = 5 ) (Fig. lB), which is consistent with that reported for the purified and affinity-labeled receptor (11,12). Labeling of the human platelet a2AR is also stereoselectively inhibited by the a-agonist epinephrine. In some neonatal rat lung preparations, two specifically labeled peptides were identified  000) (Fig. 2A). It is not known if the two labeled species are distinct proteins or merely reflect minor differences in post-translational processing (e.g. glycosylation). A similar finding has been observed for other adrenergic receptor subtypes identified by photolabeling (13,14).
In additional experiments, we have shown that photolabeling of both the human platelet and neonatal rat lung a2AR is inhibited by adrenergic agonists and antagonists with a rank order of potency consistent with an a2AR binding site. However, labeling in the two tissues is differentially inhibited by prazosin and oxymetazoline as predicted by competition binding studies. Thus, in neonatal rat lung prazosin inhibits photoincorporation of '251-rau-AZPC with approximately 100fold greater potency than in human platelet, while the converse is true for oxymetazoline (Fig. 2).
Taken together, these experiments demonstrate that in addition to the differences in ligand-binding properties between human platelets and neonatal rat lung, the a2AR isolated from these two tissues also exhibit distinct biophysical properties. This may be due to differences in their primary structure or to differences in post-translational processing, or both. A trivial explanation for these findings is that the lower molecular weight species identified in neonatal rat lung is the result of tissue-specific proteolysis. However, the following considerations apparently exclude this possibility. All procedures (membrane preparations, receptor solubilization, and purification) were carried out at 4 "C, and all buffers contained several protease inhibitors. In addition, a specifically labeled M , = 44,000 species was still observed for neonatal rat lung when the concentration of protease inhibitors was increased 10-fold and additional protease inhibitors added (aprotinin, 20 milliunits/ml; 1,lO-phenanthroline, 0.1 mM). To address further the issue of proteolysis, platelet membranes were mixed with either intact rat lung tissue or neonatal rat lung membranes and were then processed as described under "Experimental Procedures." Competition binding studies using the resulting membrane preparation indicated that the mixed a2AR population recognized oxymetazoline and prazosin with affinities intermediate between those observed in either tissue alone. Furthermore, photo- affinity labeling of the receptor proteins purified after solubilization of the mixed membrane preparation revealed both a M , -64,000 (platelet) and the M , -44,000 and -49,000 species characteristic of neonatal rat lung (Fig. 3, lanes 1 and  2 ) . These experiments and the fact that a specifically labeled M , -64,000 species has not been observed in neonatal rat lung preparations under any conditions strongly suggest that proteolysis does not account for the heterogeneity revealed by photoaffinity labeling.

-
Analysis of Receptor Glycosylation-The human platelet and porcine brain azAR are apparently glycosylated proteins as they readily bind to lectin affinity resins and migrate as broad diffuse bands on SDS-PAGE (7, [9][10][11][12], a finding that is characteristic of several glycosylated membrane-bound hormone receptors. The differences in M , of the human platelet and neonatal rat lung receptor protein may reflect differential glycosylation as the number or type of oligosaccharide chains often influences the migration of glycoproteins in polyacrylamide gels. To address this issue, we determined the ability of the human platelet and neonatal rat lung a2AR to adsorb to a wheat germ-lectin resin. Whereas greater than 75% of the solubilized human platelet receptor adsorbed to the lectin resin, 80-85% of the neonatal rat lung a2AR was found in the fall-through (Fig. 4). The human platelet a2AR could be eluted with N-acetylglucosamine, and subsequent photolabeling revealed the expected M , -64,000 protein. In contrast, with neonatal rat lung preparations a M , -44,000 protein was visualized by photoaffinity labeling of the fall-through fractions from the lectin resin. Similarly, when mixed platelet and neonatal rat lung preparations were processed and subsequently adsorbed to and eluted from the lectin resin, only the MI -64,000 peptide was visualized in the eluted fractions (Fig. 3, lanes 3 and 4). After removal of N-linked oligosaccharides by cleavage of the glycopeptide bonds with N-glycanase (15), the photolabeled human platelet azAR exhibited a MI value (-44,000-48,000) similar to the native neonatal rat lung receptor (Fig. 5 A ) . The effect of N-glycanase is due solely to removal of oligosaccharides and not due to nonspecific proteolysis since the M , value of nonglycosylated proteins was  ( M , -44,000 and -49,000) azAR were visualized in the elution from the heparin-agarose resin, only the platelet a2AR ( M , -64,000) was apparent after lectin affinity chromatography. Lanes 1 and 3, no competing ligand; lanes 2 and 4,  rauwolscine (1 p~) . not altered by enzyme treatment. Furthermore, the shift in the M , of the photolabeled human platelet a2AR following treatment with N-glycanase was prevented (Fig. 5 A ) by (Man)5(GlcNAc)2-Asn-peptide, which is thought to act as a competitive inhibitor of the enzyme (16).

Heterogeneity of cyz-Adrenergic Receptors
To determine if the presence of N-linked oligosaccharides accounts for the observed differences in ligand recognition, the binding properties of the human platelet a2AR were reexamined after treatment with N-glycanase. As shown in Fig.  5B after deglycosylation of the human platelet a2AR, high affinity oxymetazoline and low affinity prazosin binding were still observed.
The difference in the glycosylation state of the neonatal rat lung and human platelet azAR is not due to a deficiency of the glycosylation machinery in the neonatal animal itself,' since photolabeling of the a2AR from a different tissue of the neonatal rat (cerebral cortex) revealed a M , -61,000 protein, which also migrated as a broad species characteristic of heavily glycosylated proteins. Similar results were obtained in cerebral cortex membranes isolated from adult rats (Fig. 6).

DISCUSSION
The a2AR is a member of a class of membrane-bound receptors which utilize guanine nucleotide binding proteins in their signal transduction pathway and share several structural features as determined by molecular cloning. Members of this receptor group include the muscarinic and P-adrenergic receptors as well as the visual pigment rhodopsin and the substance K receptor (17-24). These receptor proteins consist of seven stretches of hydrophobic amino acids, which are presumed to span the membrane bilayer with intervening hydrophilic sequences forming intracellular and extracellular loops. Glycosylation likely occurs at the extracellular amino terminus where a variable number of sites for N-linked glycosylation are found. Potential sites for phosphorylation by CAMP-dependent protein kinase or receptor-specific protein kinases are found in the carboxyl terminus and/or in the third intracellular loop.
Photoaffinity labeling and mutagenesis studies suggest that the ligand binding domain actually exists in a hydrophobic pocket formed by the membrane-spanning sequences rather B. I. Terman, R. M. Graham, and S. M. Lanier, unpublished data.
These data indicate that the photolabeled al-adrenergic receptor in neonatal rat lung exhibits a M , of -77,000 as expected for the glycosylated receptor protein (39). This observation suggests that the glycosylation machinery is intact in the neonatal rat lung.  ( l a n e 3). B, effect of deglycosylation on the binding properties of human platelet az-adrenergic receptor. The digitoninsolubilized receptor protein was purified -50-fold by heparin-agarose affinity chromatography as described in the legend to Fig. 1.One-ml aliquots of the solubilized receptor was exchanged into 50 mM sodium phosphate buffer (pH 7.4) containing 0.1% digitonin and incubated with or without N-glycanase as described in A . Increasing concentrations of competing ligand were then incubated in duplicate with solubilized receptor (10-20 fmol) and [3H]rauwolscine (10 nM) for 2 h at 4 "C, and bound ligand was separated by precipitation with bovine y-globulinlpolyethylene glycol followed by vacuum filtration. The results shown are representative of two different receptor preparations. than in extracellular hydrophilic regions (25)(26)(27)(28)(29)(30). Although these receptors share many structural properties, substantial sequence diversity exists among the various subgroups, particularly at the amino and carboxyl termini and in the third intracellular loop. This loop has been implicated as containing the domain that interacts with guanine nucleotide binding proteins, thereby providing specificity for eliciting effector responses. The highest degree of amino acid sequence identity among this class of receptors is found in the membranespanning regions, and pharmacological specificity is likely achieved by subtle changes within these regions.
As is the case for several membrane receptor and ion channels (31-33), there appears to be a family of related proteins within the various subgroups of G-protein-coupled receptors, which are expressed in a tissue-specific manner (18)(19)(20). The existence of distinct but related receptor proteins within these subgroups was initially suggested by differences in ligand recognition (&-versus &adrenergic receptors (34), MI-versus M2-muscarinic receptors (35)) and has recently been confirmed by cDNA cloning (19)(20)(21)(22). Indeed, the latter approach has revealed the existence of previously unidentified muscarinic receptor subtypes (M3 and MI), which are also expressed in a tissue-specific manner but are difficult to distinguish on the basis of their ligand recognition properties (18). The M1-M4 receptors differ primarily in their third intracellular loops, their carboxyl terminus, and in their amino-terminal regions where the number of consensus sequences for glycosylation varies from two to five (18).
The data presented in this paper suggest that a2-adrenergic receptors also consist of a family of related but distinct proteins that are expressed in a tissue-specific manner. The differences in ligand recognition properties among human platelet, neonatal rat lung, and rat brain cortex a2AR are maintained after solubilization and partial purification. This suggests that their binding properties are specific for the particular receptor protein and unrelated to the membrane environment. The human platelet and neonatal rat lung a2AR also exhibit distinct biophysical properties which are due to differences in glycosylation. Our observation that in contrast to the human platelet azAR, the neonatal rat lung a2AR does not adsorb to a wheat germ-lectin resin and that its M , is unaffected by N-glycanase suggests that the receptor protein is devoid of N-linked oligosaccharides. The differential gly-cosylation of the human platelet and rat lung a2AR is not due to a deficiency of the glycosylation pathway in the neonatal animal but may reflect differences in the primary structure of the receptor protein in the two tissues. The human platelet and rat neonatal lung a2AR may differ in the number of potential sites for N-linked glycosylation in a manner analogous to the four muscarinic subtypes recently identified by molecular cloning (18). Alternatively, differences in the primary sequence at the amino terminus may result in an altered secondary structure, which could preclude glycosylation of the rat lung anAR despite the presence of an appropriate consensus sequence.
It is clear from the present study that variability in glycosylation does not account for the observed differences in ligand recognition properties of the human platelet and neonatal rat lung a2AR. Thus, the high affinity for oxymetazoline and low affinity for prazosin exhibited by the human platelet a2AR is maintained after removal of N-linked oligosaccharides. Therefore, the differences in ligand recognition between the two receptor proteins must reflect differences in the amino acid composition of the ligand-binding pocket.
The apparent a2AR subtypes are likely products of different genes and not the result of alternative splicing since the gene encoding the human platelet a2AR lacks introns in the coding region as do the genes encoding muscarinic and 0-adrenergic receptors (18,19). The existence of different genes that could give rise to related but distinct a2AR proteins is also supported by the demonstration that at least two hybridizing species are observed after Southern analysis of a human genomic library with a restriction fragment of the human platelet a2AR gene (17). The a2AR isolated from rat cerebral cortex may represent yet another distinct receptor subtype as suggested by its apparent molecular weight and ligand recognition properties. This possibility is currently under investigation. Although the rat brain a2AR exhibits affinity for prazosin and oxymetazoline, which is intermediate between that observed in neonatal rat lung and human platelet a2AR, the rat brain a2AR does not appear to be a mixture of the two proteins since only a higher molecular weight species is observed by photolabeling using a photoadduct that binds to both isoforms with equal affinity.
Although a2-adrenergic receptors have been subclassified as ff2A and (Y2B based on their relative affinities for prazosin and oxymetazoline (6)) the wide variation in ligand recognition properties of the receptor protein in different tissues and species suggests that additional receptor subclasses exist (Refs. 1, 4-6, and present study), the pharmacological classification of which may require the use of as yet undeveloped ligands. This hypothesis is supported by the recent development of ligands capable of distinguishing between the preand postsynaptic a2AR (36-38). Determination of the precise differences between human platelet and rat lung a2AR and identification of additional subclasses will require isolation of the cDNA encoding the particular subtype. Such studies will be greatly facilitated by the availability of the recently described genomic clone for the human platelet a2AR (17). The characterization and localization of a2AR subtypes, as well as apparent isoforms of other G-protein-coupled receptors and ion channels, may eventually allow the tissue-specific targeting of novel therapeutic agents.