A, Adenosine Receptors TWO AMINO ACIDS ARE RESPONSIBLE FOR SPECIES DIFFERENCES IN LIGAND RECOGNITION*

Species differences in ligand binding to A, adenosine receptors were localized to the seventh transmembrane (TM7) region based on the binding of [8-SH]cyclopentyl- 1,3-dipropylxanthine and three other ligands to wild type and six bovinelcanine interspecies receptor chime- ras expressed in COS-1 cells. Subsequent site-directed mutagenesis experiments identified amino acid 270 (isoleucine/methionine, bovinelcanine) as being primarily responsible for species differences in the binding of NB-adenine-substituted compounds, R-W-phenylisopro-pyladenosine (R-PIA) and (S)-A@-endonorbornan-2-yl-S- methyladenine, and the C-8-substituted xanthine, [SHlcyclopentyl-1,3-dipropylxanthine. These data are consistent with the hypothesis that the NB region of adenines and the C-&region of xanthines bind to the same region of the receptor. A second TM7 amino acid, 277 (serinelthreonine, bovinehanine), selectively influences the binding of the ribose-substituted adenosine analog, 5’-N-ethylcarboxamidoadenosine to a variable extent, depending on the nature of amino acid 270. We hypoth-esize

its intracellular metabolic role, adenosine also acts extracellularly via cell surface receptors in the G protein-coupled superfamily, Receptor-mediated effects are numerous and include inhibition of neurotransmitter release in the central nervous system (1); negative inotropic, chronotropic, and dromotropic effects in the heart (2); dilatation of arterial smooth muscle, including coronary arteries (3); effects on immunologic functions (4), and mast cell degranulation (5). Four subtypes of adenosine receptors have been cloned to date, A,, ha, 4b and A,, each from several species (6)(7)(8)(9)(10) (reviewed in Refs. 11 and 12). Additional subtypes may also exist. The 4, and hb receptor subtypes stimulate adenylyl cyclase upon activation (11). Additionally, the recombinant 42b receptor activates a C1-conductance in Xenopus oocytes (13). A, and A, receptors are inhibitory to adenylyl cyclase and couple to phospholipase C (5, 14).
R01-HL37942. The costs of publication of this article were defrayed in * This work was supported by National Institutes of Health Grant part by the payment of page charges. This article must therefore be hereby marked "aduertisenent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
I( To whom correspondence should be addressed: Box 158, Dept. Cardiology, Health Sciences Center, University of Virginia, Charlottesville, VA 22908. Tel.: 804-924-5600 Fax: 804-982-3183. The A, adenosine receptor has been cloned from five species, dog, cow, rat, rabbit, and human (10)(11)(12). Despite > 90% receptor identity at the amino acid level, the pharmacology of adenosine A, receptors differs markedly between species (15)(16)(17)(18)(19). There are differences, not only in the affinities for different ligands, but in rank order potencies as well. Bovine and canine receptors differ the most, with bovine A, adenosine receptors having higher affinity for NG-substituted adenine analogs such as the agonist R-PIA' and the antagonist N-0861; bovine A, receptors also have higher affinity for C-8-substituted xanthine antagonists such as CPX (11,15,20). Rat and human receptors are intermediate in their binding characteristics (15,17,18). These binding properties are similar for native brain membrane receptors and for recombinant receptors expressed in COS-1 cells.' The goal of this study was to identify regions of the A, adenosine receptors responsible for conferring species specificity in binding. Our approach was to construct chimeric caninehovine receptors to identify the domaids) of the receptor responsible for imparting species-dependent characteristics. This was followed by site-directed mutagenesis to target individual amino acids. The results implicate two amino acids in TM7, 270 and 277 (isoleucine/methionine (I/M) and serinehhreonine (SI") bovinekanine) as conferring species specificity in ligand binding. The contribution of these two amino acids varies depending on the nature of the ligand.
EXPERIMENTAL PROCEDURES Materials-Restriction enzymes were obtained from Promega (HindIII, KpnZ, NcoI) and Life Technologies, Inc. (Spel, HpaI, SspI). DNA Sequenase I1 kits were obtained from United States Biolabs. 33P used in sequencing was obtained from Amersham Corp. L3HICPX was from DuPont-NEN. N-0861 was a g&t from Whitby Research, Inc. R-PIA, NECA, and b&er reagents were purchased from Sigma. Mutagenesis was performed using the Altered Sites Kit from Promega. Media for bacterial culture were from Life Technologies, Inc. Reagents used in the electrophoresis of sequencing reactions were from Kodak IBI. The canine A, adenosine receptor cDNA (RDC7) was a gift from G. Vassart, Brussels, Belgium. The bovine A, adenosine receptor was cloned as described previously (20).
Construction of Chimera-Canine adenosine A, cDNA in Bluescript was digested withKpnI and HindIII (37 "C, 2 h) and the released insert subcloned into the mutagenesis vector PALTER which had been digested with the same two restriction enzymes. The bovine A, receptor pBOV13 in Bluescript was digested with SpeI (37 "C, 2 h) to remove a large portion of the noncoding region that includes a unique HpaI site. The plasmid was religated and the insert subcloned into PALTER using KpnI andXbaI sites (37 "C, 2 h). Oligonucleotides complementary to the noncoding strand were designed to introduce unique, silent restriction The abbreviations used are: R-PIA, R-h@-phenylisopropyladenosine; N-0861, (S)-h@-endonorbornan-2-yl-9-methyladenine; CPX, 8-cyclopentyl-1,3-dipropylxanthine; NECA, 5'-N-ethylcarboxamidoadenosine; TM, transmembrane.
A. Tucker, unpublished results. * kbp, kilobase pairs. oligonucleotides 5'-TCCTTCGTGGTGGGGTTAACGCCCATGTTCGG-sites into TM regions 4 and 5 , which flank the second exofacial loop. The 3' and 5'-TCC'ITCGTGGTGGGGTTAACCCCGCTGTTCGG-3' were used as primers to introduce silent HpaI sites into the fourth transmembrane region of the bovine and canine clones, respectively. NcoI sites were placed into TM5 of both receptors using 5"TCGAGAAGGT-CATCTCCATGGAGTACATGG-3'. The mutated cDNAs were used to construct six bovine-canine adenosine A, receptor chimeras. pALT-RDC7 and pALT-BOV13S were digested with restriction endonucleases as defined in Table I and the resulting DNA fragments purified on a low melt agarose gel. The recovered DNA fragments were ligated together in the combinations depicted in Table I. The resulting chimera were subcloned into the expression vector CLDNlOB (a gift from M. Reff, Smith Kline Becham Laboratories) using KpnI and HindIII for DBB, DDB, DBD, and HindIII alone for BDD, BBD, and BDB. The CLDNlOB vector contains a cytomegalovirus early promoter and an SV40 ori.
DNA Sequencing-The nucleotide sequences of all chimeras and mutants were verified by DNA sequencing on both strands with a Sequenase I1 kit (United States Biolabs).
Membrane Preparation and Radioligand Binding-Forty-eight h after transfection, using the DEAE-dextran method (21), COS-1 cells were washed in phosphate-buffered saline and homogenized in 10 volumes of buffer A (10 rm NaHEPES, pH 7.4, 10 m M EDTA, 1 m M benzamidine). The membranes were washed twice by centrifugation at 20,000 x g for 30 min in 10 volumes of buffer B (10 rm NaHEPES, pH 7.4, 1 rm EDTA, 1 m M benzamidine), and the final pellet was resuspended at a membrane protein concentration of 1 mg/ml in buffer B supplemented with 10% (w/v) sucrose and frozen in aliquots at -20 "C. Protein concentrations were determined using fluorescarnine (22) with bovine serum albumin standards. It is notable that treating membranes with 10 m~ EDTA and conducting subsequent radioligand binding assays in the absence of added divalent cation results in the absence of GTP sensitive high affinity agonist binding (23). These conditions were chosen to avoid the complication of two af€inity states for agonists that are found in the presence of divalent cations. Thus, under the conditions used in the study, only a single low affinity agonist binding site is detected. For equilibrium binding assays six concentrations of [3H]CPX were used in triplicate in tubes containing membranes with 25 pg of membrane protein at 21 "C for 2-3 h in a volume of 100 p1. Experiments were repeated two to six times. Specific binding was fit to a single site binding model using Marquardt's nonlinear least squares interpolation (24). For graphical comparison of changes in the affinity of r3H1CPX among receptor chimera and mutants, plots were constructed of B /Bmm versus B/B,Jfree, where free refers to [free [3HlCPXl in nanomoles. The slopes of these plots are proportional to the ligand affinity (l/Kd, n"' ) and are independent of differences in expression (BmJ that occur with COS cell transfection.
To calculate the KI of competing drugs, [3H]CPX was added to tubes at about 0.5 x its K, for a given receptor, and competing ligand was added over a range of concentrations to tubes containing 25 pg of protein in a final volume of 100 pl. The experiments were incubated at 21 "C for 2 h. The IC, values were calculated using a three parameter . KI values were precisely calculated from IC,,, B,, and the concentration of r3H1CPX and its K,, as described by Linden (25).

RESULTS
To identify regions of the adenosine A, receptor responsible for conferring differences in ligand binding between bovine and canine adenosine A, receptors, caninehovine chimeric receptors were constructed by ligating the appropriate fragments summarized in Table I. Each chimera consists of three segments and is named according to the species from which each segment was derived. The wild type bovine and canine receptors are referred to as BBB and DDD, respectively. Chimera are named by three-letter codes indicating the species of the corresponding segment of the receptors. Eight receptors were characterized in this study (two wild type and six chimeras). Splice sites for the chimeras occur at Leu-140 in TM4 and Ser-176 in TM5 and flank the second exofacial loop containing a hypervariable region (Fig. 1). With the exception of the BDD chimeric receptor (Bm, 200 fmoVmg of protein), all of the receptors, including wild type, chimeric, and mutant (see below), showed good expression (>1 pmoVmg of protein); however, because of the variability in B,, among the different receptors, the Scatchard plots are modified such that they are normalized to Bmu. The slopes of these modified Scatchard plots are proportional to ligand affinity. Pharmacologic characterization of the expressed receptors was done using four ligands. The two antagonists chosen were from different classes of compounds, L3HlCPX is a C-8-substituted xanthine, while N-0861 contains an h@-substituted adenine ring, but does not have a ribose at the position of the adenine as would adenosine. The ribose is required for agonist activity (26). The two agonists chosen were R-PIA, an NG-substituted ligand, and NECA, a ribose 5'-substituted ligand.
Wild type bovine (BBB) A, adenosine receptors bind L3H1CPX with a 20-fold higher affinity than canine (DDD) receptors ( Fig.  2A, Table 11). The NG-substituted ligands, R-PIA and N-0861 also bind to bovine receptors with 54-and 26-fold a higher affinity, respectively; conversely, NECA binds to the bovine receptor with a 20-fold lesser affinity than the canine receptor ( Fig. 2 B , Table 11). Similar patterns of relative binding affinities of these ligands have been noted using brain membranes.
Ligand Binding to A, Receptor Species Chimeras-Representative experiments showing equilibrium binding of L3H]CPX to the six bovinelcanine chimeric receptors are shown in Fig. 3 and data from multiple experiments are summarized in Table  11. All of the receptor constructs ending in bovine sequence, containing TMs 5-7, bind L3H1CPX with relatively high affinity, with KD values between 0.56 to 1.05 m, while all of those ending in canine sequence have lower affinity, with KD values between 11 and 15 m. A similar pattern is seen in competition binding assays with the other ligands. Table I1 shows the K, values of R-PIA, N-0861, and NECA for wild type and chimeric receptors. The binding affinities of these ligands depend predominantly on the carboxyl third of the chimeric receptors. This point is illustrated graphically in Fig. 4. Thus, a consistent finding for all of the ligands tested, whether they bind preferentially to bovine or canine receptors, is that species dependent binding properties are conferred primarily by the carboxylterminal third of the receptors.

A , Adenosine Receptor Chimeras and Mutants
Ligand Binding to Zkansmemhrane 7 Mutants of Adenosine A , Receptors-Prior studies on G-protein linked receptors for ligands with low molecular masses have implicated the transmembrane regions as forming the ligand binding pocket (27)(28)(29). There are two amino acid differences between bovine and canine adenosine A, receptors in putative TMs 5-7, amino acids 270 and 277, both in TM7. These were targeted for site-directed mutagenesis. Mutant receptors were constructed in which amino acids 270 and 277 were switched, individually or together, from bovine to canine or vice versa. Table 111 shows the K, (["HlCPX) and K, values (R-PIA, NECA, N-0861) of various ligands for wild type and mutant receptors. Both amino acids contribute to species differences in ligand binding to an extent that depends on the nature of the ligand. This point is illustrated graphically in Fig. 5. For the C-8-substituted xanthine, CPX, and the W-substituted adenine compounds, R-PIA and N-0861, amino acid 270 is the major determinant of species differences in receptor affinity. Of the eight receptors examined (two wild type and six mutated), the four receptors containing Ile in position 270 invariably bound these three ligands with higher affinity than the four receptors with Met in position 270. For example, the affinities of ['HICPX for receptors with Ile-270 are on average 5.9-fold higher than receptors with Met-270. In the cases of R-PIA and N-0861 this factor is 7.3-and 15-fold, respectively. By comparison, changing the amino acid in position 277 (S or T) produced small, directionally inconsistent changes in the affinity of receptors for the PIC-8 ligands.
The affinity ratio of I:'HICPX, R-PIA, and N-0861 for receptors with Ser-277 versus Thr-277 is 2.13, 1.53, and 0.83, respectively. In sum these data indicate that the high relative affinity of WlC-8 ligands for bovine versus canine receptors is attributable primarily to amino acid 270, Ile versus Met.
Amino acid 277 takes on greater significance in determining the affinity of adenosine A, receptors for NECA, an agonist substituted on ribose, and not on adenine. This is the only ligand examined in this study that hinds with higher affinity to wild type canine than to wild type bovine receptors. The hinding of data to the eight receptors ( Fig. 5 and Table 111) suggests an interesting interaction hetween amino acid 270 and 277 in the regulation of NECA binding. A change in amino acid 277 from Thr to Ser always reduces NECA binding affinity, hut the reduction in affinity is small (1.4-1.9 fold when amino acid 270 is Met, and large (4.R10.9 fold) when amino acid 270 is Ile. Thus, the relatively low affinity of the bovine receptor can he attributed to both Ile-270 and Ser-277. and c h a n~n g either of these amino acids to the corresponding canine sequence suhstantially increases NECA binding affinity.

DISCUSSlON
This study identifies amino acids in TM7 responsihle for confemng species selectivity in ligand binding to A, adenosine receptors. The influence of these two amino acids on binding varies depending on the nature of the ligand. Adenine and xanthine compounds substituted with aryl or cycloalkyl suhstituents at the W or the C-8 positions, respectively, hind preferentially to bovine as opposed to canine receptors. primarily due to amino acid 270, isoleucine in the hovine receptor and methionine in the canine receptor. This is entirely consistent with the data of Peet et al. (30) which suggests that the Mi position of adenines and the C-8 position of xanthines occupy the same position in the binding pocket of the A , adenosine receptor. This P I C -8 model also is supported hy an analysis of steric and electrostatic properties of ligands (31 and hy binding studies with various ligands to membranes expressing recornhinant dog, rat, and bovine A, adenosine receptors. The K'lC-8 orientation appears also to hold for A, adenosine receptors (9,

321.
NECA, which like adenosine is unsuhstituted in the N" position, binds preferentially to receptors that contain Thr-277 (canine) as opposed to Ser-277 (bovine). Changing amino acid 277 from threonine to serine consistently reduces the affinity of receptors for NECA. It is striking that this effect is relatively large if amino acid 270 is isoleucine rather than methionine. Consistent with these results is the recent finding that mutation of threonine 277 in the human A, adenosine receptor to serine or alanine causes a relatively selective decrease in NECA binding affinity (33). Moreover, interaction of A, adenosine receptor ligands with amino acid 277 is predicted in the binding model of Dudley et al. (34). Our data indicate some interplay between amino acids in positions 270 and 277 in determining affinities for ligands without bulky N6 or C-8 substitutions. One possibility is that the relatively small isoleucine side chain promotes docking of aryl or cycloalkyl iWC-8 substituent, while the slightly larger methionine group repels these, but interacts favorably with the N6 region of unsubstituted agonists. In the absence of the methionine at 270 (such as in the bovine receptor which has an isoleucine, or the human receptor containing a threonine), the amino acid at position 277, via interactions with a different region of the ligand, has a larger influence on ligand affinity for 5' or C2-substituted ligands without N6 substitutions. A threonine at position 277 favors high affinity binding to such ligands. The rat adenosine A, receptor shows a ligand binding profile that is consistent with this theory; this receptor displays high affinity for N6and C-8-substituted ligands and for ligands with 5'or C2-substitutions. As would be predicted, it has an isoleucine at position 270 as does the bovine A, receptor, but a threonine at position 277 like the canine A, receptor.

A , Adenosine Receptor Chimeras and
It is possible that mutations in receptor amino acids alter binding affinity either because ligands interact directly with mutated amino acids, or because the mutations change receptor conformation to indirectly influence binding to remote domains. We have attempted to minimize the latter possibility in this study by making conservative interspecies mutations that are not likely to produce changes in receptor structure. The mutations we have introduced apparently do not produce major changes in receptor structure or expression. Thus, we hypothesize that amino acid 270 is directly involved in the docking of the N6 portion of adenines and the C-8 portion of xanthines.
However, in the absence of structural data it is not possible to exclude the possibility of indirect effects of changing amino acids 270 and 277 on a remote ligand recognition domain.
Our data support a model for ligand binding in which the N6 or C-8 substituents of the ligand interact with a region of the receptor containing amino acid 270, while a different region of the ligand, perhaps on the ribose, interacts with a hydroxyl on amino acid 277. Because the antagonists, CPX and N-0861, do not have ribose moieties and their binding is not significantly affected by the T/S mutations at position 277, the interaction of the ribose at this position is an attractive hypothesis. Either the 5' N-ethylcarboxyl substituent or some other portion of the ribose moiety of NECA may interact at position 277. Binding studies on mutant receptors using ligands substituted at other positions, such as 2-chloroadenosine, which has a chlorine at the C2 position on the adenine ring, will help to clarify this.
This study exploits unique species pharmacology in order to better understand ligand orientation in the adenosine A, receptor binding site(s). While there are computer-generated models for the A, binding domain, supportive structure-function analysis of adenosine receptors has been limited to date. with the N6 region of the adenine and ribose hydroxyl groups, respectively, in part based on early chemical modification studies of the adenosine A, and ha receptors using the histidineselective agent diethylpyrocarbonate. These studies showed that alkylating histidine residues modified ligand binding characteristics (36-38). Examination of the sequences for the putative transmembrane amino acids of the adenosine A, receptor reveals two histidines, one in each of transmembranes 6 and 7. Data from mutational analyses of the two transmembrane histidines in the bovineh, receptor have been difficult to interpret.

TBLE I1 Summary of binding affinities of the antagonists PHICPX and N-0861 and the agonists R-PIA and NECA for wild type and chimeric A, adenosine receptors
Amnities are expressed as the means * S.E. of two to six independent experiments performed in triplicate. Also indicated for each ligand tested is the fold change from the wild type canine affinity. Mutation of the histidine residue in TM7 (His-278) to leucine dramatically decreased both agonist and antagonist binding by 90% (39); however, this may have been due to poor receptor expression rather than to the loss of a selective receptor-ligand interaction. Mutation of the histidine in TM6 effected no change in agonist affinity, but caused a 3.8-fold decrease in antagonist affinity. Again, there was a large (74%) decrease in receptor number. Interestingly, histidine 278 is adjacent to the threonine 277 mutated in this study. Given the dramatic decrease in receptor number and the nonspecific effects on ligand binding seen with mutation of histidines 251 and 278, it is possible that these histidines are important for receptor processing or configuration of the ligand binding domain without Despite their different binding profiles, the amino acid sequences of the canine and bovine adenosine receptors are very similar over most of their length, especially in the transmembrane regions, which typically are thought to form the ligand binding domain. Exceptions to this are seen among the glycoprotein hormone receptors, including the lutropin (LH), follitropin (FSH), and thyrotropin (TSH) receptors, which have large extracellular domains that bind ligands with high affinity (41,42). The A, adenosine receptors differ most between species in a hypervariable region of the second exofacial loop. We have speculated previously (20) that this region is variable either because it does not have an important role in receptor structure-function, so it accommodates frequent mutations, or that this region is responsible for species differences in ligand binding. The results of this study support the former possibility.
Previous studies of G protein-coupled receptors, most notably adrenergic and muscarinic receptors, using site-directed mutagenesis and intersubtype chimeric receptors have demonstrated the importance of transmembranes 5, 6, and 7 in agonist and antagonist binding (43)(44)(45)(46)(47). In several receptors, TM7 has been shown to be important for antagonist binding (43,45,46). For several G protein-coupled receptors, including the aZadrenergic, the 5-HT2, and the 5-HT,, receptors, marked pharmacological differences between homologues from different   Table   111. Error bars show the -log of the means * S.E. of binding constants. species can be attributed to single amino acid differences (48)(49)(50). For the qadrenergic and the 5-HT2 receptor, the critical amino acid is in TM5, but for 5-HT1, it is in TM7. Likewise, the results of this study of the A, adenosine receptor implicate TM7 as being largely responsible for interspecies variation in pharmacology.
We have identified two amino acids responsible for conferring species selectivity for different ligands to adenosine A, receptors. Other amino acids in the TM regions would also be expected to be involved in ligand interaction. Only six amino acids differ between the bovine and canine adenosine A, receptors in the transmembrane domains, the most divergent TMs being 4 and 7. Among subtypes of adenosine receptors, TM2 and TM3 are most highly conserved and have been shown to be important for ligand binding in the cationic amine receptors (44, 51). The lack of variability in these regions raises the possibility that these domains may be important in binding to invariant regions of agonist ligands, such as the 2' or 3' OH groups of the ribose moiety or unsubstituted portions of the adenine ring. These identical regions should not contribute to receptor subtype or species selectivity for ligands. Neither of the published models of adenosine A, receptor-ligand binding pocket (34,35) is completely consistent with the data from our mutational analysis of the adenosine A, receptor; however, both the models of Peet and of Ijzerman propose interactions between adenosine and amino acids in TM6 and TM7. Results from mutational analyses, including our own, should lead to the development of more refined and accurate models for the adenosine receptor ligand binding domain.