Conversion of A₃ Adenosine Receptor Agonists into Selective Antagonists by Modification of the 5′-Ribofuran-uronamide Moiety

Abstract

The highly selective agonists of the A₃ adenosine receptor (AR), Cl-IB-MECA (2-chloro-N⁶-(3-iodobenzyl)-5′-N-methylcarboxamidoadenosine) and its 4′-thio analogue, were successfully converted into selective antagonists simply by appending a second N-methyl group on the 5′-uronamide position. The 2-chloro-5′-(N,N-dimethyl)uronamido analogues bound to, but did not activate the human A₃AR, with K_i values of 29 nM (4′-O) and 15 (4′-S) nM, showing >100-fold selectivity over A₁, A_2A, and A_2BARs. Competitive antagonism was demonstrated by Schild analysis. The 2-(dimethylamino)-5′-(N,N-dimethyl)uronamido substitution also retained A₃AR selectivity but lowered affinity.

Antagonists of the A₃ adenosine receptor (AR) are of potential use for clinical targets, including the treatment of glaucoma, allergic conditions, and inflammation.¹ Potent and selective antagonists for the human A₃AR have recently been synthesized.^2–6 These human A₃AR antagonists were found to be weak or ineffective at the rat A₃AR^7,8 and were unsuitable for evaluation in small animal models or for further development as drugs. Thus, A₃AR antagonists of which the affinity and selectivity are independent of species are sought as drug candidates. In previous studies, it was found that antagonists derived from adenosine analogs, in contrast to nonpurine heterocyclic antagonists, could be species-independent, potent, and selective A₃AR antagonists.⁹

Among the four subtypes of receptors for adenosine 1, the intrinsic efficacy of nucleosides acting at the A₃ subtype is known to be especially sensitive to structural changes. For example, substitution by a benzyl or a 3-iodobenzyl group at the N⁶ position of adenosine was demonstrated to increase affinity but decrease efficacy at the human A₃AR, in the inhibition of forskolin-stimulated adenylyl cyclase.^9–11 Additional substitution by a 2-Cl substituent further increased affinity and decreased efficacy.^9,10 As a result, N⁶-(3-iodobenzyl)adenosine 2 was a partial agonist achieving only 46% of the maximal effect, while 2-chloro-N⁶-(3-iodobenzyl)adenosine 3 was a potent antagonist for the A₃AR, albeit nonselective. A 5′-methyluronamide substitution of 3 restored its efficacy and rendered it selective for the A₃AR. Thus, Cl-IB-MECA 4 and its 4′-thio analogue LJ568 5 are selective agonists for the A₃AR.^9,12,13 Here, we report that 4 and 5 were successfully converted into selective antagonists by appending an additional methyl group on the 5′-uronamide nitrogen. The finding that removing H-binding ability in the region of the 5′-uronamide lowers efficacy is consistent with expectations derived from rhodopsin-based, dynamic molecular modeling and ligand docking.¹⁴

Synthetic routes to N,N-dimethylamide derivatives 6 – 8 are shown in Scheme 1. The 5′-ester group was aminolyzed with dimethylamine, which also led to a side product substituted at the 2-position.¹⁵ The 4′-S analogue was prepared from an acetoxy intermediate,¹⁶ similar to a reported series of derivatives.¹²

Scheme 1^a.

^aReagents & conditions: a) K₂CO₃, MeOH, r.t.; b) HOAc; c) 40% aq. Me₂NH, r.t.; d) Me₂NH, EDC, HOBT, DIPEA; e) TBAF, THF.

Binding assays were carried out using standard radioligands in Chinese hamster ovary (CHO) cells stably expressing a human AR subtype.¹⁷ The binding affinity at the human A₃AR of the 2-Cl antagonist derivatives 6 and 7 was shown to be 29 and 15 nM, respectively (Table 1). Thus, the compounds were demonstrated to be over 100-fold selective in binding to the human A₃AR in comparison to other AR subtypes. In addition, the 2-(dimethylamino) substitution in 8 resulted in A₃AR selectivity but with lower affinity.

Table 1.

Potency of a series of adenosine derivatives for four subtypes of human ARs and the efficacy at the A₃AR.

	Potency (Ki or EC50, nM)				Efficacyc
Compound	A1d	A2Ad	A2Be	A3d	A3
1a	310e	700e	24,000	290e	100
2b	7.4 ± 1.7	132 ± 22	~10,000	5.8 ± 0.4	46 ± 8
3b	16.8 ± 2.2	197 ± 34	>10,000	1.8 ± 0.1	0
4b	222 ± 22	5360 ± 2470	>10,000	1.4 ± 0.3	100
5f	193 ± 46	223 ± 36	ND	0.38± 0.07	114 ± 9
6g	5870 ± 930	>10,000	>10,000	29.0 ± 4.9	0
7g	6220 ± 640	>10,000	>10,000	15.5 ± 3.1	0
8	>10,000	>10,000	>10,000	315 ± 19	0

All experiments were done on CHO cells stably expressing one of four subtypes of human ARs. The binding affinity for A₁, A_2A and A₃ARs was expressed as K_i values and was determined by using agonist radioligands ([³H]R-PIA), ([³H]CGS21680), [¹²⁵I]I-AB-MECA, respectively. The potency at the A_2BAR was expressed as EC₅₀ values and was determined by stimulation of cAMP production in AR-transfected CHO cells. The efficacy at A₃ARs was determined by inhibition of forskolin-stimulated cAMP production in AR-transfected CHO cells, as described in the text. Data are expressed as mean ± standard error.

^aValues from reference ²¹.

^bValues from references 9, 21.

^cAt a concentration of 10 μM, in comparison to the maximal effect of a full agonist NECA at 10 μM.

^dK_i in binding, unless noted.

^ecAMP assay.

^fK_i values from reference ¹³.

^g6, MRS3771; 7, LJ-1256.

ND not determined.

To probe species differences, the affinity of 6 and 7 was also measured at the rat A₃AR. Although the affinity decreased with respect to the affinity at the human A₃AR, these two compounds showed moderate affinity at the rat A₃AR. The K_i values were 286 ± 31 and 321 ± 74 nM for 6 and 7, respectively.

In functional assays consisting of measuring inhibition of forskolin-stimulated production of 3′,5′-cyclic-adenosine monophosphate (cAMP) in intact CHO cells heterologously expressing ARs,¹⁸ single concentration determinations (Table 1) indicated that A₃AR agonism was absent in compounds 6 – 8. In contrast, the concentration-response curve for compound 4 indicated full agonism, as previously reported, with an EC₅₀ of 1.2 nM at the human A₃AR (Figure 1A). A Schild analysis¹⁹ indicated that 6 concentration-dependently antagonized the A₃ agonist 4 to inhibit forskolin-stimulated cAMP accumulation in CHO cells stably expressing the human A₃AR with a K_B value of 48 nM (Figure 1B,C). In contrast, 6, at a concentration of 1 μM, had no significant effect on cAMP accumulation inhibited by the A₁ agonist CPA (N⁶-cyclopentyladenosine) in CHO cells expressing the human A₁AR under the similar conditions (data not shown).

Figure 1.

A) Inhibition by nucleoside derivatives of forskolin-stimulated cAMP accumulation in CHO cells stably expressing the human A₃AR. Cl-IB-MECA 4 concentration-dependently inhibited with an EC7₅₀ of 1.2±0.3 nM (n=3), while 6 and 7 alone had no effect. B) Right-shift of the concentration-response curves and C) Schild analysis for the concentration-dependent inhibition by 6 of the effect of Cl-IB-MECA 4 on forskolin-stimulated cAMP accumulation in CHO cells stably expressing the human A₃AR. D) A₃AR-induced changes in intracellular [Ca²⁺] in A₃AR-expressing CHO cells (proportional to relative fluorescence units, RFU) upon activation by nucleosides. In control CHO cells there was no response to NECA.

Similar results were obtained in an assay of A₃AR-induced changes in intracellular [Ca²⁺].²⁰ The nonselective AR agonist NECA (5′-N-ethyl-carboxamidoadenosine) activated the human A₃AR expressed in CHO cells to induce a concentration-dependent rise in intracellular [Ca²⁺] with a potency corresponding to an EC₅₀ of 58 ± 16 nM (n=4), while 6 and 7 alone did not induce calcium transients (Figure 1D). Compound 8 also did not induce a Ca²⁺ response (data not shown). Increasing concentrations of 6 right-shifted the activation curves for NECA (not shown), and a K_B value for 6 of 20 nM was determined. The A₃AR antagonist 7 similarly shifted the NECA-induced Ca²⁺ response to the right (not shown).

In this study, two selective A₃AR agonists, Cl-IB-MECA and its 4′-thio analogue, have been successfully transformed into antagonists selective for the A₃AR by appending an additional N-methyl group on the 5′-uronamide position. Other related N,N-dialkyl substitution in the 4′-thio nucleoside series of adenosine agonists had more complex changes in affinity and efficacy.¹² Thus, it appears that the 5′-(N,N-dimethyl)uronamido group especially tends to preserve affinity and selectivity in N⁶-3-iodobenzyladenosine derivatives, while entirely abolishing activation of the human A₃AR. There was also an interdependence of the effects of an N,N-dimethyl moiety and the N⁶-substituent, since the N⁶-methyl analogue corresponding to 7 was a partial agonist.¹² A further advantage of the N⁶-benzyl substitution of 6 and 7 over many other smaller or larger groups is that the high A₃AR affinity tends to be retained in murine species,²¹ which was confirmed in the present study.

Molecular modeling of the A₃AR indicated that flexibility of the adenosine derivative in the 5′-region correlated with putative conformational changes of the receptor associated with activation. Although there is no global conformational model of the activated state of the receptor, local conformational changes have been proposed. One such change is the rotation, anticlockwise from the extracellular perspective of the receptor, of the conserved Trp243 in TM6.^9,21 We have proposed that both flexibility of the 5′-uronamide and its ability to make and break multiple H-bonds as this conformational change occurs are needed for receptor activation. The low efficacy of 5′-thioether derivatives is also consistent with the need for H-bonding in this region in order to activate the A₃AR.²²

The present findings are consistent, in that we have removed the H-bond-donating ability of the 5′-uronamide with a relatively subtle structural alteration, resulting in the loss of ability to activate the A₃AR. From a comparison of the A_2A and A₃AR models,²³ the main difference in the docking complex was the preference of the χ angle; the A₃AR preferred nucleosides bound in an anti-form, while the A_2AAR preferred a high-anti conformation (approximately −70°) about the glycosidic bond. Another difference was the binding at the 5′-position. In the A_2AAR, the 5′-NH formed a H-bond with an important T88 residue in TM3, but in the model of the A₃AR complex, this was replaced by a stronger interaction proposed between the 5′-carbonyl group and S7.42. Accordingly, the 5′-dimethylamides displayed dramatically diminished binding affinity at the A_2AAR.

A₃AR antagonists are potentially useful therapeutically for a number of disorders.^1,24,25 However, the A₃ antagonists have not been widely used in animal models due to their extremely weak potency in murine species.⁸ Thus, A₃AR antagonists independent of species are of high priority to be developed. For this reason, a nucleoside analog IB-MECA (2-H analogue of 4) was converted to an antagonist by cyclization of the 5′-uronamide moiety into a spirolactam, resulting in a selective A₃AR antagonist of moderate affinity for both human and rat A₃ARs.⁹ Another advantage of nucleoside A₃AR antagonists over other heterocycles⁷ is increased aqueous compatibility. For example, the clog p values of 6 and 7 are 1.69 and 1.73, respectively, in comparison to 6.86 for MRS1191, a dihydropyridine antagonist of the A₃AR.⁷ Here, two additional selective antagonists with reasonable affinity for both human and rat A₃ARs were introduced, which increased the diversity of rat A₃AR antagonists.

It is clear that more potent and selective antagonists are needed for eventual therapeutic purposes. The newly synthesized A₃AR antagonists could be evaluated in models of a number of disorders related to the A₃AR, such as glaucoma and inflammation. Currently, existing nucleoside analogs should be good templates for further modification and development of potent and selective antagonists for the A₃ARs in diverse species.

Chart 1.