CHARACTERIZATION OF HUMAN STRIATAL A₂-ADENOSINE RECEPTORS USING RADIOLIGAND BINDING AND PHOTOAFFINITY LABELING

Abstract

The adenosine agonist [³H]CGS21680 (2-[4-[[2-carboxyethyl]phenyl]ethylamino]-5'-N-ethylcarboxamidoadenosine) bound to A₂ receptors in human striatal membranes with a K_d of 17.8±1.1 nM and a B_max of 313± 10 fmol/mg protein. The addition of 100 μM GTP diminished both the affinity of agonist radioligand for A₂ adenosine binding sites and the total binding, resulting in K_d and B_max values of 28.6±1.0 nM and 185± 22 fmol/mg of protein. Adenosine ligands competed for [³H]CGS21680 with the expected potency order. The adenosine antagonist [³H]XAC (8-[4-[[[[(2-aminoethyl)-amino]carbonyl]methyl]oxy]phenyl]-1,3-dipropylxanthine), although A₁-selective in the rat, binds to human striatal A₂ receptors with high affinity. 25 nM CPX (8-cyclopentyl-1,3-dipropylxanthine), an A₁-selective antagonist, was added to the incubation medium and effectively eliminated 91% of [³H]XAC (1 nM) binding to human A₁ receptors, yet preserved 90% of binding to A₂ receptors. [³H]XAC exhibited saturable, specific binding (50% of total) to A₂ sites with a K_d of 2.98±0.54 nM and a B_max of 0.71±0.23 pmol/mg protein (25°C, non-specific binding defined with 100 μM NECA). The potency order for antagonists against 1 nM [³H]XAC was CGS15943A > XAC ≈ PD115,199 > PAPA-XAC > CPX > HTQZ ≈ XCC ≈ CP-66,713 > theophylline ≈ caffeine, indicative of an A₂-type binding site. A_2a-receptors were found to be present in the human cortex, albeit at a much lower density than in the striatum. Photoaffinity labeling using ¹²⁵I-PAPA-APEC revealed a molecular weight of 45K, but proteolytic cleavage was observed, resulting in fragments of MW 43K and 37K. In the absence of proteolytic inhibitors the 37K fragment, which still bound ¹²⁵I-PAPA-APEC, was predominant.

INTRODUCTION

A₂-adenosine receptors mediate the anti-platelet-aggregatory effects (1) and vasodilatory effects (2,3) of adenosine. Two putative subtypes of A₂-adenosine receptors have been distinguished: A high affinity A_2a receptor and a low affinity A_2b receptor (23). In the central nervous system, A_2a-adenosine receptors (4), are localized mainly in the striatum and olfactory tubicle (5). Both A₁ and A₂-adenosine receptors mediate the depression of neuronal firing elicited by adenosine. A selective A₂-adenosine agonist was found to act as a locomotor depressant, through a centrally-mediated mechanism (6).

Recently, two selective agonist radioligands with high affinity for A₂-receptors, [³H]CGS21680 and ¹²⁵I-PAPA-APEC (2-[4-[2-[2-(4-aminophenylacetyl)aminoethyl]-aminocarbonyl]ethyl]phenyl]ethylamino]-5'-N-ethylcarboxamidoadenosine), have been reported (7,17). The development of antagonist radioligands for A₂-receptors has been impeded by the lack of truly selective agents. In this study, we have taken advantage of the unusual high affinity (but not selectivity) at human A₂-receptors of the antagonist [³H]XAC. This is consistent with previously noted species differences (27,28) in affinity of xanthine derivatives for adenosine receptors.

Recently, a G-protein linked receptor for which the amino acid sequence was determined by recombinant DNA methodology using mRNA from the dog thyroid, was identified as an A₂-receptor (10). When expressed in COS cells, the protein was found to resemble the high affinity A_2a-receptor in radioligand binding and in stimulation of adenylate cyclase. The A_2a-receptor has been characterized previously by photoaffinity labeling methods (7–9) and found to be a glycoprotein of molecular weight 45K Daltons (bovine and rat). The A_2a-adenosine receptor in rabbit striatum has a molecular weight of 47K (9), and in the absence of proteolytic inhibitors undergoes proteolytic cleavage to yield a 38K fragment, still capable of binding radioligands with the appropriate pharmacology. In this work, it is shown that the human A_2a-receptor also undergoes proteolytic cleavage to yield a 37K fragment. Because of the difficulty of obtaining fresh human brain samples, some proteolytic cleavage will be unavoidable, even in the presence of proteolytic inhibitors. Therefore, we chose to study the 37K proteolytic product, and protease inihbitors were deliberately left out after the membrane preparations.

MATERIALS AND METHODS

XAC, CPA (N⁶-cyclopentyladenosine), ADAC (N⁶-[4[[[4-[[[(2-aminoethyl)amino]carbonyl]methyl]anilino]carbonyl]-methyl]phenyl]adenosine), DPMA (N⁶-[2-(3,5-dimethoxyphenyl)-2-(2-methylphenyl)ethyl]adenosine), CGS21680, NECA (5'-N-ethylcarboxamidoadenosine), 2-chloroadenosine, 8-p-sulfophenyltheophylline and CPX were obtained from Research Biochemicals, Inc. (Natick, MA). The A₂-adenosine agonists APEC (2-[4-[2-[2-aminoethyl]aminocarbonyl]ethyl]phenyl]ethylamino]-5'-N-ethylcarbox-amidoadenosine), its N^α-p-aminophenylacetyl derivative, PAPA-APEC, and other derivatives were prepared as described (12). Ac-ADAC, the N-acetyl amide of ADAC, was prepared as described (11). m-DITC (m-phenylenediisothiocyanate) derivatives of ADAC and APEC were prepared as described (12). PD115,199 (8-[4-[[(2-dimethylaminoethyl)N-methylamino]sulfonyl]phenyl]-1,3-dipropylxanthine) (13) and HTQZ (3-(3-hydroxyphenyl)-5H-thiatolo[2,3-b]quinazoline) (14) were the gift of Dr. J. Bristol, Parke-Davis Warner Lambert, Ann Arbor MI. CGS15943A (9-chloro-2-(2-furyl)-(1,2,4]triazolo[1,5-c]quinazolin-5-amine) (15) was the gift of Dr. J. Francis of CIBA-Geigy, Summit, NJ. [³H]XAC and [³H]CGS21680 were obtained from Dr. E. Do of Dupont NEN (Boston, MA). CP-66,713 (4-amino-8-chloro-1-phenyl-[1,2,4]triazolo(4,3-a]quinoxaline) (16) was the gift of Dr. R. Sarges of Pfizer, Groton, CT.

Preparation of striatal membranes

Human striatal tissue was isolated by dissection from 1 cm thick coronal sections obtained from the Brain Bank at the University of California at lrvine School of Medicine. The age ranged from 45 to 85 years (median age 73), both male and female, and the samples had been frozen at −70°C between 2.5 and 9 hours post-mortem. The tissue was pooled at the time of preparation of membranes. A subject diagnosed with Alzheimer's disease was an 89 year-old male, and this sample was processed separately. Although none of the human subjects tested positive for infectious diseases, precautions against contact with infectious agents were observed. Caudate and putamen were processed together for most experiments, in the absence of protease inhibitors. Striatal membranes were homogenized in 20 volumes of ice cold 50 mM Tris (tris(hydroxymethyl)aminomethane) HCI at pH 7.4 using a polytron (Kinematica, Gmbh, Luzerne, Switzerland) at a setting of 2–3 for 10 sec. The membrane suspension was then centrifuged at 37,000 × g for 20 min at 4°C. The pellet was resuspended in the above buffer solution, and the membranes were again homogenized and centrifuged. Finally the pellet was suspended in buffer (100 mg wet weight per ml) and stored frozen at −70°C until used. Protein was determined using the BCA protein assay reagents (Pierce Chemical Co., Rockford, III).

Radioligand binding

[³H]CGS 21680 binding was carried out as described (17) using 20 μM 2-chloroadenosine to determine non-specific binding.

The binding of [³H]XAC (18–20) to human striatal membranes was measured in a total volume of 1 ml. An unlabeled competing ligand or NECA (at a final concentration of 100 μM, for determination of non-specific binding) was dissolved in 25 μl of dimethylsulfoxide. To this solution was added, 25 μl of 1 μl CPX to eliminate binding to A₁-adenosine receptors, 750 μl of 50 mM Tris at pH 7.4 at room temperature, and 100 μl of radioligand to achieve a final concentration of 1 nM. Finally, 100 μl of a striatal tissue suspension was added to the assay medium, which contained 3 IU/ml adenosine deaminase, from calf intestine (Boehringer Mannheim GmbH, Germany). The mixture was incubated with shaking for 90 min at 24°C. All assays were done in duplicate or triplicate. The binding reaction was terminated by addition of 4 ml of ice cold 50 mM Tris, pH 7.4, followed by rapid filtration using a Brandel Cell Harvester (Brandel, Gaithersburg, MD). The contents of each tube were filtered through a Whatman GF/B filter which had been soaked in a flat pan in buffer containing 0.3% polyethyleneimine for 90 min. The addition of 0.01% CHAPS (3-[(3-cholamidopropyl)dimethylammnio]-1-propane sulfonate) to the TRIS buffer solution used for washing the membranes after adsorption to the glass fiber filters was found to enhance specific [³H]XAC binding. The filters were washed twice with 4 ml of ice cold 50 mM Tris, pH 7.4. Each filter disc was added to 7 ml of scintillation fluid, vortexed, and counted after at least 6 hours.

The minimum quantity of tissue required for reproducible binding was investigated, based on the ratio of specific to total binding. Specific [³H]XAC binding to A₂-receptors relative to total binding increased linearly with increases in protein concentration over a range of 15–100 μg of protein per ml, suggesting that in this range non-specific binding represents in large part binding of [³H]XAC to the glass fiber filters. The degree of specific binding reached a plateau at quantities of protein in the range of 100–400 μg per ml. Thus, [³H]XAC binding experiments were carried out using 150–200 μg protein per ml.

IC₅₀ values were computer-generated from competition binding data, using a non-linear regression formula on the GraphPAD program (Institute for Scientific Information), and were converted to K_i values using a K_d value for [³H]XAC of 2.98 nM, as determined in saturation experiments (see below) and the Cheng-Prusoff equation (21).

Photoaffinity labeling

For photoaffinity labeling, ¹²⁵I-azido-PAPA-APEC was synthesized and purified by HPLC as described (22). Frozen human brain membranes were thawed, washed and suspended in 50 mM HEPES (N-2-hydroxyethylpiperazin-N'-2-ethanesulfonic acid) buffer (pH 6.8), containing 10 mM MgCl₂ in either the absence or presence of protease inhibitors (5 mM EDTA, 0.1 mM phenylmethanesulfonyl fluoride, 0.1 mg/ml soybean trypsin inhibitor, 5 μg/ml leupeptin, 1 μg/ml pepstatin A). After pretreatment with 0.5 unit/ml adenosine deaminase, the membranes (approximately 1 mg protein/ ml / tube) were incubated in the dark at 37°C for 1 hour in the presence of ¹²⁵I-azido-PAPA-APEC. Ice cold incubation buffer was then added followed by centrifugation at 43,000 μ g. The pellet was resuspended in 1 ml incubation buffer and irradiated for 4 min at a distance of 1 cm from the light source (UVCG-25 mineral light, 254 nm, UVP Inc., San Gabriel, CA). Membranes were then washed again and solubilized in 10% SDS/glycerol buffer prior to electrophoresis on a 10% acylamide gel. Radioactive bands were visualized by autoradiography.

RESULTS

Radioligand binding

Membranes prepared from pooled human striatal samples were subjected to saturation binding experiments at 25°C with [³H]CGS 21680, a high affinity and A₂-selective adenosine agonist (3,17). A representative saturation isotherm for specific [³H]CGS21680 binding is shown in Figure 1. The amount of specific binding for 5 nM [³H]CGS21680 was approximately 65% of the total counts. Non-specific binding was defined with 20 μM 2-chloroadenosine. The same amount of nonspecific binding was obtained in the presence of 40 μM 2-chloroadenosine, 200 μM NECA, or 40 μM CGS21680. The Scatchard plot of the data is linear indicating a homogeneous class of non-cooperative binding sites. A K_d value of 17.8±1.1 nM and a B_max value of 313±10 fmol/mg protein were determined.

Figure 1.

Saturation of [³H]CGS21680 binding to human striatal A₂-adenosine receptors. Speciiic (o), non-specific (•), and total (Δ) binding were determined for 90 min at 24°C. Values are means of a typical experiment done in triplicate. The calculated K_d-value of 17.8±1.0 nM represents the mean of three separate experiments on membranes from pooled striatal tissue.

[³H]CGS21680 binding was also carried out in the presence of 100 μM GTP. The presence of the guanine nucleotide diminished the binding of the agonist radioligand at A₂ adenosine binding sites, with a K_d value of 28.6±0.95 nM and a B_max value of 185±22 fmol/mg protein observed.

Known adenosine receptor agonists competitively displaced specific [³H]CGS21680 binding to human striatal membranes, as shown in Table 1. A rank order of potency consistent with binding to human A₂-adenosine receptors (12,23) was observed: DPMA = m-DITC-APEC > NECA > APEC > m-DITC-ADAC = CV1808 = CGS21680 = Biotinyl-APEC > R-PIA (N⁶-phenylisopropyladenosine) > ADAC > CCPA (2-chloro-N⁶-cyclopentyladenosine) > N-Acetyl-ADAC = CPA. CGS15943A, a potent non-xanthine adenosine receptor antagonist, was found to be the most potent compound to inhibit [³H]CGS21680 binding, with a K_i value of 0.96 nM. We found CGS21680 to inhibit the binding of [³H]CGS21680 with a K_i-value of 35 nM and to inhibit the binding of [³H]PIA to adenosine A₁ receptors with a K_i-value of 4580 nM (both in human striatum), thus CGS21680 is approximately 132-fold selective. All compounds were found to interact with one binding component (i.e., Hill coefficients not significantly different from unity). A K_i for R-PIA of 84 nM in human striatal membranes was close to the value of 124 nM reported for rat striatum using [³H]NECA in the presence of 50 nM CPA (23).

TABLE 1.

Inhibition of Agonist Radioligand Binding at Human A₂-Adenosine Receptors.

Compound:	Ki (nM), Human A2-Receotors	Ki (nM), Rat A2-Receptorsb	Ratiod
agonists
DPMA	5.46±1.97	4.4c	1.2
m-DITC-APEC	6.78 ± 0.94	22	0.31
NECA	12.9 ± 3.7	12	1.1
APEC	22.3 ± 7.2	12	1.9
m-DITC-ADAC	28.2 ± 6.8	203	0.14
CV1808	29.5±6.5	105	0.28
CGS21680	34.7 ± 5.3	14	2.5
Biotinyl-APEC	38.8 ± 6.4	55	0.71
R-PIA	84.3±28.1	410	0.21
ADAC	151 ± 18	218	0.69
CCPA	212 ± 2	3900c	0.05
Ac-ADAC	339 ± 63	700c	0.48
CPA	362 ± 39	890	0.41
antagonists
CGS15943	0.960±0.291	3	0.3
XAC	8.55±2.65	50	0.17
theophylline	4420± 150	21,000	0.21

^aAffinity, expressed as K_i values in nM, of adenosine receptor agonists and antagonists at human striatal A₂-adenosine receptors (vs. [³H]CGS21680) and in rat (vs. [³H]NECA or [³H]CGS21680) striatum, unless noted. Binding of [³H]CGS21680 (5 nM) was carried out for 90 min at 24°C, in the presence of 0.05 M TRIS buffer at pH 7.4, 10 mM MgCl₂. K_i values are the mean ± standard deviation of at least 3 experiments done in triplicate.

^bversus binding of [³H]CGS21680 in rat striatum, from refs 12,17, unless noted.

^cK_i values versus binding of [³H]NECA in the presence of 50 nM N⁶-cyclopentyladenosine are from refs 23, 31 or determined here (Ac-ADAC).

^dratio of K_i-value at human A₂-receptors versus K_i value at rat A₂-receptors.

[³H]XAC (18–20) was examined as an antagonist radioligand at A₂-adenosine receptors in human striatal membranes. Since XAC has a high affinity for A₁-receptors, indeed in the rat (18) and mouse (24) it is A₁-selective, it was necessary to add a sufficient quantity of an A₁-selective antagonist to selectively block the A₁ component. As in Bruns et al. (13) and in our previous study of [³H]XAC binding to A₂-receptors in rabbit striatal membranes (20), CPX (8-cyclopentyl-1,3-dipropylxanthine) was selected for this purpose and added to the incubation medium. To determine the appropriate concentration of CPX to add to the assay medium, CPX inhibition of [³H]XAC binding to both A₁ - and A₂-receptors in human striatum was studied. The resulting inhibition curve (Figure 2) fit a 2-site model when analyzed by a computer-assisted curve-fitting program. The K_i-value for CPX at the high affinity (assumed to be A₁) site was found to be 0.46±0.14 nM and at the low affinity (assumed to be A₂) site 113±2.2 nM. The high affinity component represented 37% of the combined, specific [³H]XAC binding. In the presence of 25 nM CPX, 91% of [³H]XAC (1 nM) binding to human A₁ receptors was eliminated, and 90% of binding to A₂ receptors was preserved. Thus 25 nM CPX was used for subsequent A₂ receptor binding experiments with [³H]XAC. The incubation was carried out at 25°C with non-specific binding defined in the presence of 100 μM NECA.

Figure 2.

Inhibition of binding of [³H]XAC to human striatal adenosine receptors by CPX. The biphasic curve represents binding of CPX to high affinity (A₁) and low affinity (A₂) sites. The binding was carried out at 24°C for 90 min. [³H]XAC was present at a concentration of 1.0 nM in 0.05 M Tris HCI buffer at pH 7.4.

1 nM [³H]XAC bound to human striatal membranes, incubated for 90 min at 24°C, with specific A₂-binding amounting to approximately 50% of total binding. Scatchard analysis of the saturation curve showed that [³H]XAC exhibited saturable, specific binding to A₂ binding sites with a K_d of 2.98±0.54 nM and a B_max of 0.71±0.23 pmol/mg protein (Figure 3).

Figure 3.

Saturation curve of [³H]XAC binding to human striatal A₂-adenosine receptors. The binding was determined for 90 min at 24°C. Values are means of a typical experiment done in triplicate. The calculated K_d-value of 2.98 nM represents the mean of three separate experiments using membranes from pooled striatal tissue.

Previously characterized adenosine receptor antagonists competitively displaced specific [³H]XAC binding to human striatal A₂-adenosine receptors as shown in Figure 4. Inhibition of [³H]XAC binding by adenosine antagonists was consistent with affinities previously measured at rabbit A₂-adenosine receptors (20) with a rank order of potency against 1 nM [³H]XAC: CGS15943A > XAC = PD115,199 > PAPA-XAC > CPX > HTQZ = XCC (8-[4-[(carboxymethyl)oxy]phenyl]-1,3-dipropylxanthine) = CP-66,713 > theophylline = caffeine, indicative of A₂ receptors (Table 2). The antagonist inhibition curves also fit monophasic analysis with Hill coefficients in the range 1.0. The non-xanthine heterocycle HTQZ was not very potent at human A₂ receptors, compared to its affinity at rat A₂-adenosine receptors, at which the K_i value is 124 nM (14). In contrast, the non-xanthine heterocycle CGS15943 was more potent at human A₂ receptors than at rat A₂-adenosine receptors, at which the K_i value is 3 nM (17).

Figure 4.

Inhibition of binding of [³H]XAC to human striatal A₂-adenosine receptors by various adenosine receptor ligands. The binding was carried out at 24°C for 90 min. [³H]XAC was present at a concentration of 1.0 nM in 0.05 M Tris HCI buffer at pH 7.4 in the presence of 25 nM CPX.

TABLE 2.

Inhibition of Antagonist Radioligand Binding at A₂-Receptors from Several Species.^a

Compound:	Ki (nM), Human A2-Receptors	Ki (nM), Rabbit A2-Receptors	Ki (nM), Rat A2-Receptorsb
CGS15943	0.609±0.100	n. d.	3
XAC	5.93±2.78	7.5	50
PD115,199	8.66±3.46	3.28	16c
PAPA-XAC	57.5±7.9	n. d.	n. d.
CPX	110±35	179	490
HTQZ	742±55	n. d.	124c
XCC	772±81	196	1130
CP-66,713	832±62	365	12
theophylline	4720±1225	38,250	21,000
caffeine	5780±950	n. d.	37,000

^aAffinity, expressed as K_i values in nM, of adenosine receptor antagonists at human and rabbit striatal A₂-adenosine receptors vs. [³H]XAC and in rat striatum vs. [³H]NECA in the presence of 50 nM N⁶-cyclopentyladenosine, unless noted. Binding of [³H]XAC (1.0 nM) was carried out for 90 min at 24°C, in the presence of 0.05 M Tris hydrochloride buffer at pH 7.4, 10 mM MgCl₂ and 25 nM CPX. K_i values are the mean ± standard deviation of at least 3 experiments done in triplicate. Rabbit data from ref 20.

^bversus binding of [³H]CGS21680 in rat striatum, from ref 17 or determined here, (CP-66,713)unless noted.

^cK_i values vs. binding of [³H]NECA in the presence of 50 nM N⁶-cyclopentyladenosine are from refs 14,23.

Adenosine agonists inhibited binding of [³H]XAC with Hill coefficients less than unity, indicative of complex binding to several affinity states. Thus, IC₅₀ values only are reported in Table 3. The curves for inhibition by the agonist CGS21680 was analysed for conformity to a two site fit, but the results were variable, and no conclusion could be drawn.

TABLE 3.

Inhibition of [³H]XAC binding at human striatal A₂-adenosine receptors y adenosine receptor agonists.^a

Compound:	IC50 (nM), Human A2-Receptors	Hill coefficient	Ki (nM), Rat A2-Receptorsa
NECA	186±40	0.53	10.3
R-PIA	1410±94	0.40	124
CV-1808	146±30	0.71	97
CGS21680	992±551	0.48	15

^aK_i values versus binding of [³H]NECA in the presence of 50 nM N⁶-cyclopentyladenosine are from references 17, 23.

The effects of adding guanine nucleotides or cations to the incubation medium with single concentrations of agonist and antagonist radioligands were compared (Table 4). GTP (0.1 mM) had a significant effect on [³H]CGS21680 binding, resulting in a 37% decrease in binding, whereas Mg⁺⁺ increased [³H]CGS21680 binding by 71%. Neither GTP nor Mg⁺⁺ altered [³H]XAC binding significantly. [³H]XAC binding was sensitive to the presence of Na⁺ ions, with a decrease of 28%. The presence of 1 mM GTP did not have a clearly discernable effect on the curve for inhibition of [³H]XAC binding by CGS21680. The IC₅₀ value was 2194±1363 nM in the presence and 992±551 (Table 3) in the absence of GTP.

TABLE 4.

Effect of GTP and cations on [³H]CGS21680 binding or [³H]XAC binding.^a

	% of control
Agent	[3H]CGS21680(5 nM)	[3H]XAC (1 nM)
GTP (100 μM)	63±7.7	102±4
Mg++ (10 mM)	171±14	103±6.5
Na+ (150 mM)	n.d.	72±4.5

^aMembranes were incubated with [³H]XAC (1.0 nM) and the appropriate concentration of the adenosine derivative for 60 min at 24°C, in 0.05 M TRIS buffer at pH 7.4 in the presence of 10 mM MgCl₂ and 25 nM CPX. K_i values are the mean ± standard deviation of at least4 experiments done in triplicate, n.d. = not determined.

A comparison of A₂-adenosine receptor density in caudate and putamen showed similar levels in these striatal regions (Table 5). Curiously, a distinct, low level of specific binding to A₂-adenosine receptors in the cerebral cortex was found, as reported by Wan et al. (26). 5 nM [³H]CGS21680 binding to cerebral cortical A₂-adenosine receptors revealed roughly one tenth of the level of specific binding found under the same conditions in the striatum, and this was unaffected by the addition of 50 nM CPX. In the cortex, [³H]XAC, in the presence of 25 nM CPX, also displayed one tenth of the level of specific binding relative to the striatum. In the absence of CPX, the level of specific [³H]XAC binding (1 nM) was 186±17 fmoVmg protein, as expected for a region rich in A₁ adenosine receptors.

TABLE 5.

Regional distribution of [³H]CGS21680 binding sites in human brain.^a

Human brain region	[3H]CGS21680 (fmol/mg protein)	[3H]XAC binding (fmol/mg protein)
putamen	67±5.5 (n=6)	100±14 (n=8)
caudate	39±0.5 (n=4)	84±2.5 (n=4)
cortexa	4.4±0.19 (n=4)	9.5±4.8 (n=3)

^aSpecific binding was determined using 5 nM [³H]CGS21680 or 1 nM [³H]XAC, nonspecific binding was determined in the presence of 20 μM 2-chloroadenosine or 100 μM NECA. Each experiment was conducted in triplicate. Frontal and temporal cortex were combined.

Photoaffinitv labeling

Human striatal membranes prepared in the above manner, except that caudate and putamen were processed separately, were subjected to photoaffinity labeling experiments, using a method developed by us (7,22). PAPA-APEC was synthesized as reported (12) and iodinated by the chloramine-T/¹²⁵I method. ¹²⁵I-PAPA-APEC was purified using reverse phase HPLC (12). The aryl amine of ¹²⁵I-PAPA-APEC was converted to an aryl azide through diazotization, and a second HPLC purification step was carried out. The resulting photoaffinity label was used to study the A₂-receptor binding subunit and derived protein fragments separated on SDS polyacrylamide gels.

Photoaffinity labeling revealed several selectively labeled bands, the largest of which had a molecular weight of 45K (Figure 5), in agreement with that found for the bovine A₂-receptor (7). In this receptor preparation, proteolytic cleavage was observed. resulting in fragments of MW 43K and 37K. In the absence of proteolytic inhibitors, the 37K fragment, which still bound ¹²⁵I-PAPA-APEC, was predominant.

Figure 5.

Photoaffinity labeling of A₂-adenosine receptors in membranes (protease inhibitors added) from normal human striatum and from the striatum of an Alzheimer's patient. The control lanes each contain 105 μg protein, and the Alzheimers lanes each contain 173 μg protein. Membranes were incubated with the agonist photoaffinity probe ¹²⁵I-azido-PAPA-APEC (1 nM) in the absence and presence of 5 mM theophylline. The membranes were irradiated with UV light and solubilized for sodium dodecylsulfate - polyacrylamide gel electrophoresis and autoradiography.

A preliminary comparison of A2 receptor photoaffinity labeling in striatal membranes (processed with protease inhibitors) obtained from normal human subjects and from a person diagnosed with Alzheimer's disease (Figure 5) showed no apparent difference in mobility of the receptor and fragments.

DISCUSSION

The human central A₂ adenosine receptor has been studied using radioligand binding (25,26) with [³H]NECA or (³H]CGS21680. In spite of recent advances in characterization of A2 adenosine receptors through agonist binding, photoaffinity labeling, and cloning, a generally selective antagonist for this receptor is still lacking. A number of heterocyclic compounds (14–16) have been reported to have high affinity and/or modest selectivity for A₂-adenosine receptors, yet none of these compounds are suitable as general A₂-adenosine radioligands.

In this study we have explored the usefulness of [³H]XAC as an antagonist radioligand for the human A₂-adenosine receptor. In other species, such as the rat, [³H]XAC is A₁-selective and has successfully been used to label A₁-receptors (18). However, considerable species differences exist, and 8-substituted xanthines have been shown to have particularly high affinity at human A₂-receptors (27). Thus, it was reasonable to expect that XAC, structurally also an 8-phenyl substituted xanthine, might display high affinity for human A₂-receptors, as was found to be the case for the rabbit (K_d 3.8 nM). Indeed, we found a K_d of 3 nM for [³H]XAC binding to human striatal A₂-receptors. Since XAC also has considerable A₁-affinity (nanomolar), and A₁ -receptors are present in human striatum, it was necessary to add a sufficient quantity of an A₁-selective ligand in order to selectivity block the A₁ component. We have shown that in rabbit striatum [³H]XAC in the presence of 50 nM CPX is useful in an A₂-binding assay (20). Similarly, we now show that this method is applicable to human A₂-adenosine receptors, in the presence of 25 nM CPX. The adenosine agonist CCPA was reported to be more highly A₁-selective than CPA in the rat (32). However, CCPA was not satisfactory for this purpose of blocking 1³H]XAC to human striatal A₁ receptors (data not shown), due to its unexpectedly high affinity at human striatal A₂ receptors (Table I). Competition for [³H]XAC binding sites in the human striatum by the antagonist CPX fit a 2-site model. Nearly all of the [³H]XAC binding to A₁ receptors was eliminated at 25 nM CPX. Thus, CPX was selected as more valuable for this purpose and added to the incubation buffer in subsequent binding experiments.

The rabbit A₂-receptor was recently shown to undergo a specific proteolytic degradation, particularly in solubilized fractions, resulting in a change in molecular weight from 47,000 to 38,000. Both native and partially hydrolyzed rabbit A₂-receptors bind the radioligand ¹²⁵I-PAPA-APEC with high affinity. Similarly, the human A₂-adenosine receptor forms proteolytic fragments that retain the radioligand binding properties. It is not known what enzyme is responsible for the cleavage or whether this proteolysis is relevant to the action of adenosine receptors in vivo. The proteolysis is diminished but not completely in the presence of proteolytic inhibitors (see Materials and Methods). Since significant proteolysis of the autopsy-derived human A₂-receptor occurs, we treated the membranes in the absence of proteolytic inhibitors, resulting in a predominance of a 37K fragment that bound [³H]XAC and [³H]CGS21680 with high affinity. It is uncertain if this proteolytic fragment is identical to the native receptor in all pharmacological properties.

Inhibition of [³H]CGS21680 binding by agonists and antagonists (Table 1) and [³H]XAC binding by antagonists (Table 2) was as predicted in order of potency for A₂-adenosine receptors. Bruns et al. (13) determined K_i values for inhibition of binding of [³H]PD115,199 in rat striatum by the A₁-selective antagonist CPX and the non-selective PD115,199 to be 190 and 4.1 nM, respectively. This is in close agreement with our findings of K_i values against [³H]XAC binding in human striatum of 110 and 8.7 nM for CPX and PD115,199, respectively. In comparing [³H]XAC binding in human striatum to [³H]NECA or [³H]CGS21680 binding in rat striatum, CPX appeared to be 3-fold more potent at human A₂-adenosine receptors. Theophylline and caffeine were considerably more potent at human A₂-adenosine receptors than at rat A₂-adenosine receptors as indicated in both [³H]CGS21680 and [³H]NECA binding. The apparently higher affinity of theophylline in human striatum raises the question whether A₂-adenosine receptors may be particularly relevant to the therapeutic use of theophylline in man. A previous study of [³H]XAC binding to A₂ adenosine receptors of human platelets (19) reported somewhat higher K_i values (nM) for antagonist competitors: XAC, 17; XCC, 1800; theophylline, 29,000; and caffeine, 51,000.

Among non-xanthine adenosine antagonists assayed versus [³H]XAC binding in human striatum, CGS15943 was at least as potent at human A₂-adenosine receptors as at rat A₂-adenosine receptors (Table 2). Thus, CGS15943 is the most potent reported A₂ adenosine antagonist, although it is not very selective. Disappointingly, two recently reported putative A₂-selective antagonists, CP-66,713 and HTQZ, were less potent than predicted from K_i values at rat A₂-adenosine receptors.

The adenosine agonist DPMA was reported by the group at Parke Davis Warner-Lambert (31) as a moderately selective and high affinity agonist at A₂-adenosine receptors. We found that DPMA is equally highly potent at human A₂-adenosine receptors. CV1808, a slightly selective A₂-adenosine receptor agonist, was moderately potent in human A₂-adenosine receptors. CCPA (32), highly A₁-selective agonist in the rat, was more potent at human A₂-adenosine receptors (even more potent than CPA) than expected from data at rat A₂-adenosine receptors. m-DITC-APEC and m-DITC-ADAC were designed as isothiocyanate-bearing affinity labels selective for A₂-and A₁-adenosine receptors, respectively (12). m-DITC-ADAC was more potent at human A₂-adenosine receptors than in the rat. Biotinyl-APEC (12) is highly selective for A₂-adenosine receptors (>90-fold, in rat brain). The high affinity of biotinyl-APEC in human brain suggests its use in conjunction with avidin complexation in affinity chromatography and histochemical studies of human A₂-adenosine receptors.

The high affinity of XAC and its derivatives at human A₂-adenosine receptors raises the possibility of their use as molecular probes for this receptor, for example in affinity labeling or affinity Chromatography. For example, XAC immobilized to an agarose chromatography matrix has been used for the affinity purification to homogeneity of A₁-adenosine receptors from rat (29) and bovine brain (30). Such a column may be of use in purifying human A₂-adenosine receptors.

Specific binding to A₂-receptors in human cerebral cortex was detected using [³H]CGS21680. This result was also obtained by Wan et al, although the relative level in cortex compared to the striatum was somewhat higher in their study. In contrast measurable levels of A₂-receptors in the rat cortex were not observed in previous studies (5) of adenosine receptor localization using [³H]CGS21680. The observed specific binding of [³H]XAC in the cortex is not as conclusive as is the [³H]CGS21680 binding, because of the low receptor density. The [³H]XAC binding may reflect the presence of A₂ receptors in the cortex, but binding to residual A₁ receptors cannot be ruled out. Similarly, Bruns et al (13) found [³H]PD115,199 in the rat cortex displayed 7% of the level of specific binding relative to the striatum. It is not known whether the apparent presence of the A_2a receptor in a region in which it is commonly thought to be absent reflects differences in species or the detection limits of the assay methods used. There is evidence for physiological effects of A₂-selective agonists when applied directly to brain regions other than those in which the receptor occurs in high density. In rat cortical neurons, iontophoretically applied CGS21680 depressed spontaneous and acetylcholine-evoked firing (4).

In drug screening, especially in the development of therapeutic entities to act at adenosine receptors, it is advisable to be cognizant of the species differences in affinities observed for adenosine antagonists. The differences are so great (up to 70-fold observed in this study) that a complete reversal of selectivity between the rat and human for a given drug is possible. For example, in Stone et al. (27), PACPX (8-(2-amino-4-chlorophenyl)-1,3-dipropylxanthine) was found to be 73-fold A₁ selective in rat brain binding and nearly 2-fold A₁ selective in human brain.

It has been shown that GTP regulates the binding of agonist radioligands to A₁-receptors (reviewed in ref. 33). In the absence of GTP, two affinity states are detected, and in the presence of GTP most of the high affinity sites are converted to low affinity. At A₂-receptors, GTP added to the assay medium affected the diminished the binding of the agonist radioligand [³H]CGS21680, but due to variability it was not possible to draw conclusions about the relative percentage of high and low affinity states. The enhancement of binding of [³H]CGS21680 by magnesium ions (Table 4) is consistent with the regulation of agonist binding by GTP-binding proteins, since for A₁ receptors magnesium ions shift the distribution predominantly towards the high affinity form (33).

¹²⁵I-PAPA-APEC photoaffinity labeling of A₂-receptors was previously applied to bovine (7) and rabbit (9) striatum, the DDT₁ MF-2 smooth muscle cell line (22), and PC12 pheochromocytoma cells (9). We have now shown that the human A₂-receptor is nearly identical in molecular weight to the receptor in rat, calf, and other species (9).