Xanthine Functionalized Congeners as Potent Ligands at A₂-Adenosine Receptors

Abstract

Amide derivatives of a carboxylic acid congener of 1,3-dialkylxanthine, having a 4-[(carboxymethyl)oxy]phenyl substituent at the 8-position, have been synthesized in order to identify potent antagonists at A₂-adenosine receptors stimulatory to adenylate cyclase in platelets. Distal structural features of amide-linked chains and the size of the 1,3-dialkyl groups have been varied. 1,3-Diethyl groups, more than 1,3-dimethyl or 1,3-dipropyl groups, favor A₂ potency, even in the presence of extended chains attached at the 8-(p-substituted-phenyl) position. Polar groups, such as amines, on the chain simultaneously enhance water solubility and A₂ potency. Among the most potent A₂ ligands are an amine congener, 8-[4-[[[[(2-aminoethyl)amino]carbonyl]methyl]oxy]phenyl]-1,3-diethylxanthine, and its D-lysyl conjugate, which have K_B values of 21 and 23 nM, respectively, for the antagonism of N-ethyl-adenosine-5′-uronamide-stimulated adenylate cyclase activity in human platelet membranes. Strategies for the selection and tritiation of new radioligands for use in competitive binding assays at A₂-adenosine receptors have been considered.

The synthesis of potent and/or selective analogues of caffeine and theophylline as antagonists at extracellular adenosine receptors remains a challenge. Two adenosine receptor subtypes have been delineated on the basis of inhibition (A₁) or stimulation (A₂) of adenylate cyclase by adenosine and adenosine analogues. Numerous physiological functions have been shown to be regulated by either or both A₁ and A₂-adenosine receptors.¹ In many organs and cell lines, a single-receptor subtype has been found to occur. For example, in platelets and PC12 pheochromocytoma cells the existing adenosine receptors appear to be all of the A₂ subtype.² Fat cells are considered to contain only the A₁-receptor subtype.

Characterization of adenosine receptors with reversibly-binding radioligands has been carried out mainly with brain membranes, due to the high density of A₁ receptors in that organ. A number of radioligands for A₁ receptors have been reported. Tritiated N⁶-cycloalkyladenosines^3,4 and tritiated (R)-N⁶-(phenylisopropyl)adenosine⁵ (PIA) having high specific activity are used regularly in competitive-binding assays to determine affinity of new analogues at A₁ receptors. The choice of antagonist radioligands had been limited to [³H]-1,3-diethyl-8-phenylxanthine³ ([³H]-DPX, 1b), which has a K_i value at A₁ receptors of about 60 nM. The recently introduced 8-[4-[[[[(2-aminoethyl)-amino]carbonyl]methyl]oxy]phenyl]-1,3-dipropylxanthine ([³H]XAC, 6c), a xanthine amine congener,^6,7 has a higher affinity and specific activity than [³H]DPX, and its use overcomes a number of difficulties associated with the use of [³H]DPX as an antagonist radioligand for A₁ receptors.

Until recently the screening of analogues at A₂ receptors has relied on biological assays of adenylate cyclase, platelet aggregation, or in vivo models such as coronary vasodilation.^2,8,14 A competitive binding assay at A₂ receptors in striatal tissue using ³H-labeled N-ethyladenosine-5′-uronamide (NECA) as a radiotracer has been introduced.⁹ Although NECA is one of the most potent A₂ agonists reported, it is not selective for that subtype, and it in fact binds with approximately 5–50-fold greater affinity at A₁-adenosine receptors.¹⁰ Thus, in the striatum, where both adenosine-receptor subtypes occur, it is necessary to block the A₁-receptor component of [³H]NECA binding by use of a selective A₁ ligand^9b or by pretreatment of the membranes with N-ethylmaleimide.^9a However, there also is evidence that NECA can bind to sites that have characteristics different from those expected of A₂ receptors coupled to adenylate cyclase, particularly in peripheral tissue.^13,14 With the goal of finding a more convenient method for the study of A₂ receptors, we have applied the functionalized congener approach,^6,11 previously used in the design of A₁-receptor radioligands, such as [³H]XAC, to the development of potential antagonist A₂-receptor radioligands.

Results and Discussion

Although a number of moderately A₂-selective antagonists have been reported,¹⁵ none are of sufficiently high affinity to serve as radiotracer for competitive-binding studies. For this reason, we approached the development of an A₂-antagonist radioligand from the standpoint of maximizing A₂ potency regardless of the selectivity ratio. At that point a highly potent candidate should be radiolabeled and its binding properties examined in a membrane preparation such as human platelets in which A₂-but not A₁-adenosine receptors are present. Subsequently, the platelet assay is to be used to study other potential A₂-adenosine ligands by measuring their inhibition of binding of a well-characterized radioligand.

The classical approach to developing new drugs has been to explore the effect of structural modifications at or around a primary pharmacophore. We have developed a “functionalized congener” approach for adenosine-receptor ligands, by which a chemically functionalized chain is incorporated at a point that does not reduce biological activity. The resulting active congener can be joined covalently to a variety of moieties, including amino acids and peptides¹¹ or solid supports for affinity chromatography, through a functional group such as an amine or a carboxylic acid. For xanthines as adenosine antagonists it has been shown⁶ that distal structural changes on a chain attached through the 8-(p-substituted-phenyl) position can modulate the potency and selectivity of the drugs, perhaps through interactions at sites on the receptor distal to the binding site for the pharmacophore. We now show the validity of this approach for xanthines as antagonists at A₂-adenosine receptors.

The K_i values for reversal of 2-chloroadenosine-stimulated cyclic AMP production in guinea pig brain slices for a carboxylic acid congener, 2c, an ethyl ester, 3c, an amine congener, 6c (all R = Pr), and a p-toluidine, 7a (R = Me) were 34, 30, 49, and 20 nM, respectively. Furthermore, within a series of amino acid conjugates of 6c, the D-lysyl conjugate 11c displayed high potency, high water solubility, and stability to enzymatic hydrolysis. Since a dissociation constant in the range of 10⁻⁸ M would be acceptable for a radiotracer, these compounds constituted suitable lead compounds. As modifications of these xanthines designed to increases A₂ potency we have explored changing the size of the 1,3-dialkyl groups and incorporating distal amino groups in the functionalized 8-phenyl moiety (see Table I for synthesis and characterization).

Table I.

Synthesis and Characterization^e of New Xanthine Derivatives (compound suffixes a, b, and c refer to R = Me, Et, and Pr, respectively)

no.	% yield	mp, °C	formula	anal.
2b	67a	>310	C17H18N4O5·H2O	C, H, N.
3a	61	294–296	C17H18N4O5·1/4H2O	C, H, N
3b	86	267–269	C19H22N4O5	C, H, N
4b	71	275–277	C20H24N4O5·1/2DMF	C, H, N
5b	88	300–302	C19H23N5O4·1/2H2O	C, H, N
6a	99	255–258	C17H20N6O4·1/4H2O	C, H, N
6b	92	233–235	C19H24N6O4·H2O·DMF	C; H, Nb
7b	58	>310	C24H26N5O4·4/5H2O	C, H, N
8a	60	>310	C24H23N5O6	H, N; Cc
8b	92	310–313	C26H27N6O6·1/2H2O	C, H, N
9a	94	280 dec	C25H27N7O5	H, N; Cd
9b	77	273–278d	C27H31N7O5·3/4H2O	C, H, N
10b	79	215–219	C38H50N8O9·1/2H2O	C, H, N
10c	95	207–211	C40H54N8O9	C, H, N
11b	100	270–275	C25H36N8O5·2HB·2H2O	C, H, N
11c	83	dec begin 190	C27H40N8O5·3HBr	C, H, N

^aYield calculated from 6-amino-1,3-diethyl-5-nitrosouracil, which was treated with Na₂S₂O₄ and then condensed with [(4-formylphenyl)oxy]acetic acid, and the benzylidene adduct was oxidized with NaIO₄ as in ref 12.

^bH calcd 6.77, found 6.25; N calcd 19.95, found 18.97.

^cC calcd 60.37, found 52.82.

^dC calcd 59.40, found 51.49.

^eProton NMR resonances (300 MHz) in ppm from Me₄Si for selected compounds in (CD₃)₂SO. 6b: δ 8.15 (t, 1 H, amide NH), 8.05 and 7.05 (each d, J = 8.7 Hz, 2 H, Ar), 4.54 (s, 2 H, CH₂O), 4.08 and 3.94 (each q, 2 H, Et methylene), 3.21 (m, 2 H, CH₂NHCO), 2.68 (t, J = 6.2 Hz, 2 H, CH₂NH₂), 1.26 and 1.13 (each t, 2 H, J = 6.8 Hz, CH₃). 8a: δ 8.10 (d, 2 H, Ar, meta to O), 7.58 (d, 2 H, Ar, ortho to NH), 7.23 (d, 2 H, Ar, ortho to CH₂), 7.13 (d, 2 H, Ar, ortho to O), 4.79 (s, 2 H, CH₂O), 3.63 (s, 2 H, CH₂Ar), 3.61 (s, 3 H, OCH₃), 3.49 and 3.27 (each s, 3 H, NCH₃). 9a: δ 8.06 (d, 2 H, Ar, meta to O), 8.02 (1 H, amide NH), 7.56 (d, 2 H, Ar, ortho to NH), 7.21 (d, 2 H, Ar, ortho to CH₂), 7.13 (d, 2 H, Ar, ortho to O), 4.75 (s, 2 H, CH₂O), 3.48 and 3.24 (each s, 3 H, NCH₃), 3.36 (s, 2 H, CH₂Ar), 3.07 (m, 2 H, CH₂NHCO), 2.60 (t, J = 6.7 Hz, 2 H, CH₂NH₂).

As reported previously,² the affinity of XAC (6c) for A₂ receptors of human platelets was in close agreement with the value in brain slices, but the potency of 2c showed a discrepancy between these two systems. Since we intended to study binding to A₂ receptors in platelets, the effect on platelet adenylate cyclase was adopted as the sole criterion of A₂ potency in this study.

In each case the potency of the 1,3-diethyl analogue at platelet A₂ receptors was greater than the 1,3-dimethyl analogue and at least as great as that of the 1,3-dipropyl analogue (Table II). The enhancement of A₂ affinity by ethyl substituents was more marked for the simple 8-phenyl series, 1, than for the functionalized congeners. In contrast, Bruns et al.^9b reported 1,3-dipropyl-8-phenyl-xanthine to be 5-fold more potent than the diethyl analogue in inhibition of [³H]NECA binding to a striatal A₂-adenosine receptor.

Table II.

Potencies of Xanthine Derivatives at A₂- and A₁-Adenosine Receptors

no.	Ra	A2 receptor:b KB, nM	A1 receptor:c Ki, nM
1a	Me	1900	70
1b	Et	210	65
1c	Pr	2100	13
2c	Pr	2400 (1500–3900)	50
3b	Et	170 (75–370)	140
3c	Pr	135 (106–173)	13
4b	Et	73 (54–98)	90
5b	Et	48 (36–65)	57
6a	Me	40 (30–54)	13
6b	Et	21 (18–23)	12
6c	Pr	25 (21–30)	1.2
7a	Me	430 (260–720)	27
7b	Et	100 (60–170)	9.3
8b	Et	210 (190–230)	58
9a	Me	63 (57–68)	42
9b	Et	28 (15–51)	25
10b	Et	230 (100–540)	180
11b	Et	23 (17–30)	9.4
11c	Pr	37 (35–40)	0.87

^aSubstituent at the 1,3-positions (see Figure 1).

^bAntagonism of NECA-induced stimulation of adenylate cyclase activity in human platelet membranes. Values are means with 95% confidence limits from three separate experiments. Values for 2c, 6c, and 11c from ref 2; value for 1b from ref 14.

^cInhibition of [³H]PIA binding to rat cerebral cortex membranes. Values are means from two separate experiments.

While the carboxylic acid congener 2c was not particularly potent, esterification to give compound 3c increased potency markedly at A₂ receptors, but not at A₁ receptors. A comparison of ethyl and isopropyl esters showed preference of the branched isopropyl group at A₂ receptors.

Compounds 7a and 7b, containing the p-toluidine group, were not as potent at A₂ receptors as expected on the basis of previous studies. However, when the [(2-aminoethyl)-amino]carbonyl group was attached to the p-methyl group of the toluidine, as in 9, the potency rose. Thus, the potency-enhancing effect of a distal amino group, observed consistently with the A₁ receptor particularly with dipropyl analogues (e.g., 6c, 11c),^6,11 now appears to apply as well to the A₂ receptor. The enhancement of A₂ potency by an amino group is also evident in comparing the ethyl amide 5 and the 2-aminoethyl amide 6b.

The potencies of XAC (6c) and the diethyl analogue 6b were nearly identical at A₂ receptors. The specific binding of a tritiated 1,3-dipropylxanthine amine congener (XAC) to A₂-adenosine receptors in platelets has been reported.¹³ However, a high degree of nonspecific binding of [³H]XAC will limit the usefulness of this radioligand in screening assays. For an A₂ radioligand it would be preferable to use the diethyl analogue, due to the likely lower levels of nonspecific binding with 1,3-diethylxanthines, in comparison to the corresponding 1,3-dipropyl analogues. Although not strictly indicative of the level of potential filter binding, the octanol–water partition coefficients (log p) are as follows: −0.63 (6a), −0.06 (6b), 0.81 (6c), −0.05 (9a), and −1.20 (11b). Thus, there is a large difference in polarity between 1,3-diethyl (e.g., 6b) and 1,3-dipropyl (e.g., 6c) analogues, which is also reflected in maximum aqueous solubility (pH 7.2, 0.01 M phosphate) of 500 and 90 μM, respectively. The lipophilic contribution of an additional ring and amide bond in the chain, as in 9a, is roughly equivalent to substituting 1,3-dimethyl with 1,3-diethyl substituents. Another potential A₂ radioligand is 11b, the polar conjugate of D-lysine and diethyl-XAC.

The synthesis of tritiated analogues of 1,3-dipropylxanthines through the catalytic tritiation of 1,3-diallyl precursors has been demonstrated.¹² To obtain the corresponding tritiated diethyl analogues in similar fashion would require starting with vinyl substituents. As an alternative we have explored coupling of the diethyl carboxylic congener 2b to ethylenediamine using the water-soluble carbodiimide 1-ethyl-3-[3-(dimethylamino)-propyl]carbodiimide hydrochloride and 1-hydroxybenzotriazole as a catalyst. Coupling reactions using tritiated ethylenediamine to produce the desired radioligand are in progress.

Experimental Section

Xanthine analogues were synthesized as previously reported^6,11,12 or by routes shown in Figure 1. HPLC-grade dimethylformamide (Aldrich Chemical Co., Milwaukee, WI) was used without further purification. New compounds were characterized by 300-MHz proton NMR on a Varian 300XL and gave spectra consistent with their structures. Except for compounds 10 and 11, chemicalionization mass spectra (CIMS) were obtained with a Finnigan 1015D spectrometer using ammonia gas with a Model 6000 data collection system. Samples for elemental analysis and biological testing were shown to be homogeneous by thin-layer chromatography (silica, CHCl₃/MeOH/HOAc, 85:10:5 and 10:10:1) and, if necessary, recrystallized from solvent mixtures including DMF/ether, DMF/water, or Me₂SO/ether.

Figure 1.

Synthesis of xanthine derivatives. Reagents: (a) EtOH/HCl,⁷ 50 °C (3) or R’OH/DMAP/EDAC (4); (b) 70% ethylamine, aqueous (5) or ethylenediamine, neat⁶ (6, 9); (c) DCC/HOBt; (d) HBr/HOAc. R = Me, Et, or n-Pr, corresponding to compound suffixes a, b, and c, respectively.

Partition coefficients were determined by addition of a dimethyl sulfoxide solution (5 μL) of the xanthine to an equivolume mixture (2 mL) of 1-octanol and aqueous sodium phosphate, pH 7.2, 0.1 M. After thorough mixing of phases and separation, the ratio of concentrations was determined by comparing absorption at 310 nm of aliquots diluted in methanol (aqueous was filtered). The ±SEM for six determinations corresponded to 0.01 log p unit.

Measurements of the potencies of the xanthines as antagonists of NECA-induced stimulation of adenylate cyclase activity in human platelet membranes and as inhibitors of [³H]PIA binding to rat cerebral cortex membranes were carried out as reported previously.^2,3 The slopes of the inhibition curves had Hill coefficients of about 1.0, and all compounds completely inhibited the response to NECA or the specific binding of [³H]PIA.

8-[4-[(Carboxymethyl)oxy]phenyl]-1,3-diethylxanthine 2-Propyl Ester (4b)

Compound 2b (20 mg, 56 μmol) was suspended in a solution of 4-(dimethylamino)pyridine (7 mg) and 2-propanol (0.1 mL) in dimethylformamide (1 mL) and the mixture treated with 1-ethyl-3-[3-(dimethylamino)propyl]-carbodiimide hydrochloride (24 mg, 110 μmol). After several minutes a solution formed, followed by precipitation. After 2 h, water (3 mL) was added, and the white solid was collected, washed with water, and dried, giving 15.8 mg of compound 4; CIMS (NH₃), m/e 401 (M + 1)⁺.

8-[4-[[[[(2-Aminoethyl)amino]carbonyl]methyl]oxy]-phenyl]-1,3-diethylxanthine (6b)

Compound 3b (70 mg, 0.18 mmol) was dissolved in ethylenediamine (2 mL) with stirring. The solvent was evaporated under a stream of nitrogen. The oily residue was triturated with methanol and ether to give compound 6b (67 mg, 92% yield) as a solid; CIMS (NH₃), m/e 401 (M +1)⁺.

Alternately, the carboxylic acid 2b was preactivated with 1-ethyl-3-[3-(dimethylamino)propyl]carbodiimide hydrochloride and 1-hydroxybenzotriazole hydrate in dimethylformamide, and this mixture was added slowly to a solution of 1 equiv of ethylenediamine. The product, 6b, was isolated by thin-layer chromatography (CHCl₃/MeOH/HOAc, 10:10:1, on silica gel plates) in 44% yield (determined by UV).

8-[4-[[[[4-(Carboxymethyl)anilino]carbonyl]methyl]-oxy]phenyl]-1,3-diethylxanthine Methyl Ester (8b)

Compound 2b (80.8 mg, 0.23 mmol), methyl (p-aminophenyl)acetate hydrochloride (55 mg, 0.27 mmol), HOBt (31 mg, 0.23 mmol), and finally 1-ethyl-3-[3-(dimethylamino)propyl)carbodiimide hydrochloride (98 mg, 0.46 mmol) were combined with stirring in 8 mL of dimethylformamide. Diisopropylethylamine (39 μL, 0.23 mmol) was added, and the mixture was stirred overnight. Water was added, and the product (60 mg) was collected, washed with water, and dried; CIMS (NH₃), m/e 506 (M + 1)⁺.

8-[4-[(Carboxymethyl)oxy]phenyl]-1,3-diethylxanthine 2-(D-Lysylamino)ethylamide Dihydrobromide (11b)

Compound 6b (50 mg, 0.13 mmol) was suspended in 2 mL of dimethylformamide and treated with N^α-Boc-N^ε-Cbz-D-lysine (95 mg, 0.25 mmol), HOBt (17 mg, 0.13 mmol), and 1-ethyl-3-[3-(dimethylamino)propyl]carbodiimide hydrochloride (54 mg, 0.25 mmol). The mixture was stirred overnight. Aqueous workup, as above, followed by recrystallization from ethyl acetate/hexanes provided 75 mg of compound 10b. The protecting groups were removed with 30% HBr/acetic acid (2 mL), giving compound 11b (72 mg). An analytical sample was prepared by recrystallization from methanol/ether.