A FUNCTIONAL SCREENING OF ADENOSINE ANALOGUES AT THE ADENOSINE A_2B RECEPTOR: A SEARCH FOR POTENT AGONISTS

Abstract

Various adenosine analogues were tested at the adenosine A_2B receptor. Agonist potencies were determined by measuring the cyclic AMP production in Chinese Hamster Ovary cells expressing human A_2B receptors. 5′-N-Substituted carboxamidoadenosines were most potent. 5′-N-Ethylcarboxamidoadenosine (NECA) was most active with an EC₅₀ value of 3.1 μM. Other ribose modified derivatives displayed low to negligible activity. Potency was reduced by substitution on the exocyclic amino function (N⁶) of the purine ring system. The most active N⁶-substituted derivative N⁶-methyl-NECA was 5 fold less potent than NECA. C8- and most C2-substituted analogues were virtually inactive. 1-Deaza-analogues had a reduced potency, 3- and 7-deazaanalogues were not active.

Introduction

The adenosine A_2B receptor is one of the four known subtypes of adenosine receptors, the others being named A₁, A_2A and A₃. The A₁ and A₃ receptor subtypes inhibit the enzyme adenylate cyclase, diminishing the production of the second messenger cyclic AMP. On the contrary, both A₂ subtypes stimulate adenylate cyclase to produce cyclic AMP. A subdivision of A₂ receptors was first proposed by Daly et al. based upon observations of lower EC₅₀ values on cyclic AMP accumulation in rat striatum than in rat brain slices¹. Bruns et al. designated the high affinity striatal A₂ receptor as the A_2A subtype, whereas they referred to the low affinity subtype, existing throughout the brain, as to A_2B². Molecular cloning and expression have been reported of all four subtypes of adenosine receptors of several species, including the A_2B receptor from both rat and human brain^3,4. By the cloning techniques it had also become possible to determine tissue distributions of the receptors. In the human body, low expression of the A_2B receptor was found in kidney, moderate levels were observed in heart, brain and lungs, and high densities were measured in the colon⁵.

Through A_2B receptors, adenosine inhibits the Tumour Necrosis Factor-α (TNF-α) production by monocytes^6,7. TNF-α is a pro-inflammatory cytokine that triggers the production cascade of other cytokines and acute-phase proteins⁸. When produced in large quantities—e.g., when the immune system is overactivated due to a challenge by pathogenic bacteria—TNF-α can lead to septic shock and eventually to death^9,10. A future therapeutic purpose for adenosine A_2B receptor agonists could thus be the treatment of septic shock.

However, no A_2B selective compounds have been reported yet. One of the most potent agonists on this subtype is 5′-N-ethylcarboxamidoadenosine (NECA). EC₅₀ values on the production of cyclic AMP are in the low micromolar range, considerably higher than on the other subtypes^11–14.

As a start of our search to find agonists with increased potency and selectivity, we screened adenosine analogues with a great variety of substituents and modifications on their ability to generate cyclic AMP in a CHO cell line stably transfected with human adenosine A_2B receptor cDNA⁴.

Results

Cyclic AMP production was measured at compound concentrations of 100 μM. Cyclic AMP production is shown as percentage of the production by 100 (μM NECA. A number of compounds were taken further into investigation by determining their EC₅₀ values.

NECA (1) and 5′-N-cyclopropylcarboxamidoadenosine (NCPCA; 2) were found to have EC₅₀ values on cyclic AMP production of 3.1 and 5.3 μM, respectively (TABLE 1).

TABLE 1.

Activities of NECA and NCPCA on cAMP production in CHO cells expressing human A_2B receptors

		R	EC50 (μM)
1	NECA	ethyl	3.1±2.4
2	NCPCA	cyclopropyl	5.3±1.6

More 5′-carboxamidoadenosines are listed in TABLE 2. Compounds 3–8 are N⁶-(3-pentyl)-substituted. In this series, the 5′-N-benzyl-substituted analogue (8) did not show activity. Cyclic AMP production decreased according to the following substituent order: cyclopropyl > isopropyl > methyl > allyl > 3-pentyl. NECA lost its activity when the ethyl substituent was modified into a 2-aminoethyl group (compound 9).

TABLE 2.

Activities of N⁶-(3-pentyl), 5′-N-disubstituted carboxamidoadenosines on cAMP production in CHO cells expressing human A_2B receptors. Compounds were tested at 100 μM. Activities are given as percentages of stimulation with 100 μM NECA.

	R	% stimulation
3	methyl	63%
4	allyl	41%
5	isopropyl	69%
6	cyclopropyl	78%
7	3-pentyl	6%
8	benzyl	0%
9	2-NH2-ethyl*	1%

^*N⁶-unsubstituted analogue

Adenosines substituted at the C2-position of the purine are shown in TABLE 3. Several 2-thio-, 2-amino- and 2-halo-analogues were evaluated. C2-substution of adenosine diminished potency. In the 2-halo series, 2-chloroadenosine (18) was the most potent compound with an EC₅₀ value of 24 μM; potency decreased when an iodo or a fluoro substituent was present instead of a chloro substituent. 2-Thio- and 2-amino-substituted adenosines, including the A_2A agonist CV 1808, displayed low potency.

TABLE 3.

Activities of C2-substituted adenosines on cAMP production in CHO cells expressing human A_2B receptors. Compounds were tested at 100 μM. Activities are given as percentages of stimulation with 100 μM NECA.

	R	% stimulation or EC50 (μM)
10	SH	1%
11	SCH2CH2CN	0%
12	S-cyclohexyl	3%
13	S-hexadecyl	5%
14	SBn	3%
15	NHPh (CV1808)	10%
16	NHPhEt	3%
17	F	23%
18	Cl	24±11μM
19	I	2%

In the C8-amino-substituted series (TABLE 4), only the 8-amino-analogue itself showed some activity, albeit marginal (9%).

TABLE 4.

Activities of C8-substituted NECA and MECA analogues on cAMP production in CHO cells expressing human A_2B receptors. Compounds were tested at 100 μM. Activities are given as percentages of stimulation with 100 μM NECA.

	R1	R2	% stimulation
20	Et	Br	0%
21	Et	NH2	9%
22	Et	EtNH	0%
23	Et	Me2N	0%
24	Me	MeNH	0%
25	Me	Me2N	0%

N⁶-substituted analogues were categorized into four groups as shown in TABLES 5–8. In TABLE 5, NECA analogues are listed; analogues of 5′-N-methylcarboxamidoadenosine (MECA) are shown in TABLE 6; adenosine derivatives without modifications in the ribose are displayed in TABLE 7; some N⁶-functionalized congeners are listed in TABLE 8.

TABLE 5.

Activities of N⁶-substituted NECA analogues on cAMP production in CHO cells expressing human A_2B receptors. Compounds were tested at 100 μM. Activities are given as percentages of stimulation with 100 μM NECA.

	R	EC50 (μM) or % stimulation
26	methyl	19±6μM
27	3-CH3O-phenyl	63%
28	3-CH3-benzyl	16%
29	4-CH3O-benzyl	14%
30	3-CH3O-benzyl	91%
31	3-F-benzyl	80%
32	3-Cl-benzyl	84%
33	3-Br-benzyl	95%
34	3-I-benzyl	71%
35	2-NO2-benzyl	87%
36	3-NO2-benzyl	57%
37	4-NO2-benzyl	41%
38	benzyl*	24%

^*5′-N-cyclopropyl analogue

TABLE 8.

Activities of N⁶ functionalized congeners on cAMP production in CHO cells expressing human A_2B receptors. Compounds were tested at 100 μM. Activities are given as percentages of stimulation with 100 μM NECA.

	R	% stimulation
80	NHAc	13%
81	N(CH3)3	17%
82	OH	19%

TABLE 6.

Activities of N⁶-substituted MECA analogues on cAMP production in CHO cells expressing human A_2B receptors. Compounds were tested at 100 μM. Activities are given as percentages of stimulation with 100 μM NECA.

	R1	R2	% stimulation
39	3-CH3-benzyl	H	13%
40	3-CF3-benzyl	H	0%
41	4-Cl-benzyl	H	1%
42	3-Br-benzyl	H	16%
43	4-Br-benzyl	H	3%
44	3-I-benzyl (IB-MECA)	H	10%
45	3-I-benzyl (Cl-IB-MECA)	Cl	3%
46	3-I-benzyl	SMe	0%
47	3-I-benzyl	NHMe	0%
48	4-SO3H-benzyl	H	0%
49	3-I-4-NH2-benzyl (I-AB-MECA)	H	18%
50	3-NO2-benzyl	H	4%
51	(R)-1-phenyl-ethyl	H	23%
52	(S)-1-phenyl-ethyl	H	15%
53		H	0%
54	benzyl	H	15%

TABLE 7.

Activities of N⁶-substituted adenosines on cAMP production in CHO cells expressing human A_2B receptors. Compounds were tested at 100 μM. Activities are given as percentages of stimulation with 100 μM NECA.

	R1	R2	EC50 (μM)or % stimulation
55	cyclopentyl (CPA)	H	203±97 μM
56	phenyl	H	63±19 μM
57	benzyl	H	96±32 μM
58	phenylethyl	H	67±4 μM
59	(R)-phenylisopropyl (R-PIA	H	53 ±8μM
60	4-SO3H-phenyl (SPA)	H	52 ±17μM
61		H	3%
62	4-NH2-benzyl	H	12%
63	2-CH3benzyl (metrifudil)	H	21%
64	2-Cl-benzyl	H	35%
65	3-I-benzyl	H	18%
66	3-I-benzyl	Cl	6%
67	3-I-benzyl	NH2	6%
68	4-SO3H-phenylpropyl	H	1%
69	4-SO3H-phenylbutyl	H	2%
70	4-SO3H-phenyldecyl	H	4%
71	n-butyl	H	23%
72	n-decyl	H	3%
73	4-NH2-butyl	H	10%
74	10-NH2-decyl	H	0%
75	N, N-dimethy*	H	0%
76	N, N-dipropy*	H	0%
77	NH2	H	0%
78	(adenosin-N6-yl)butyl	H	15%
79	(adenosin-N6-yl)decyl	H	13%

^*N⁶, N⁶-disubstituted analogue

In the series of the NECA analogues in TABLE 5, the N⁶-methyl-substituted member (26) was the most active one with an EC₅₀ value of 19 μM. The other analogues, most of them bearing N⁶-(substituted)-benzyl groups, had activities varying from 14–95%. The MECA analogues in TABLE 6 generally displayed a considerably lower activity than the NECA analogues (0–23%). In the adenosine series (TABLE 7), EC₅₀ values of compounds 55–60 were determined. The A₁ selective agonist N⁶-cyclopentyladenosine (CPA, 55) had an EC₅₀ value of 203 μM. The N⁶-phenyl-(56), N⁶-benzyl-(57), N⁶-phenylethyl-(58) and N⁶-(R)-phenylisopropyl-substituted (R-PIA, 59) adenosines were more active, with EC₅₀ values in the range of 53–96 μM. The activity of N⁶-(4-sulfophenyl)adenosine (SPA, 60) was in this same range (52 μM). Adenosine derivatives with other N⁶-substituents were also tested. N⁶-furyl adenosine (61) gave a 3% response. Analogues with substituted benzyl groups had activities of 12–35% (62–65). N⁶-alkyladenosines (71–74) were less potent than the N⁶-phenyl(alkyl)adenosines mentioned before. N⁶, N⁶-dialkyladenosines (75,76) were inactive. The N⁶-(4-sulfo-phenyl)alkyladenosines 68–70 appeared to be far less active than SPA (60). N⁶-amino adenosine (6-hydrazino purine riboside, 77) was not active.

The N⁶, C2-disubstituted compounds 45–47 and 66,67 were found to have reduced activities in comparison with the C2-unsubstituted analogues 44 and 72. Analogues with large substituents, such as the (adenosin-N⁶-yl)alky)-substituted compounds 78 and 79 in TABLE 7 and the functionalized congeners 80–82 in TABLE 8, had low activities (10–20%).

Deaza derivatives (TABLE 9) showed negligible potencies, except the 1-deaza analogues.

TABLE 9.

Activities of deazaadenosine analogues on cAMP production in CHO cells expressing human A_2B receptors. Compounds were tested at 100 μM. Activities are given as percentages of stimulation with 100 μM NECA.

	analogue	% stimulation or EC50 (μM)
83	1-deazaadenosine	10%
84	2-chloro-1-deazaadenosine	20%
85	3-deazaadenosine	1%
86	1, 3-dideazaadenosine	1%
87	7-deazaadenosine	0%
88	1, 7-dideazaadenosine	0%
89	1-deaza-NECA	16±5μM

7-Xanthinyl adenosines (FIG. 1) lacked potency, as well as the ribose modified adenosines in FIG. 2. Compounds that were not categorized, most of them having more than one modification, are depicted in FIG. 3. These all had low activities.

FIG. 1.

7-Xanthinyl analogues (no activities on cAMP production in CHO cells expressing human A_2B receptors). Compounds were tested at 100 μM.

FIG. 2.

Activities of ribose modified adenosines on cAMP production in CHO cells expressing human A_2B receptors. Compounds were tested at 100 μM. Activities are given as percentages of stimulation with 100 μM NECA.

FIG. 3.

Activities of variously modified analogues on cAMP production in CHO cells expressing human A_2B receptors. Compounds were tested at 100 μM. Activities are given as percentages of stimulation with 100 μM NECA.

Discussion

Our aim was to analyze compounds for A_2B agonism. Therefore we tested modified adenosine analogues. In this approach our focus was on stimulation of the receptor. We did not do experiments to determine antagonistic activities.

Until today the reference ligand for the adenosine A_2B receptor has been the nonselective adenosine agonist NECA. Pierce et al. and Alexander et al. reported NECA to have an EC₅₀ value of approximately 1 μM on the production of cyclic AMP in CHO-K1 cells^4,14. These cells were also used here as the test system. EC₅₀ values of 3.1 and 5.3 μM were determined for NEC A and its 5′-N-cyclopropyl analogue NCPCA, respectively (TABLE 1).

Regarding the relatively high potency of the carboxamido analogues NECA and NCPCA, the compounds in TABLE 2 are interesting. Although the N⁶-(3-pentyl) group reduced activity, still a trend can be observed. The 5′-N-cyclopropyl-substituted analogue was the most potent ligand in this series, followed by the isopropyl-, the methyl-, the allyl-, the (3-pentyl)- and the benzyl-substituted compounds. A (3-pentyl) and a benzyl group in this position are largely unfavourable, probably because of steric or electronic interactions.

Substitution at the C2-position (TABLE 3) with appropriate groups is known to enhance A_2A selectivity. The relatively A_2A selective agonist CV 1808 (15) showed only low activity in the A_2B assay. This was also observed with other 2-amino-substituted adenosines and with 2-thio-analogues. Of the halogen atoms in the 2-position, chloro (in compound 18) appeared to be a relatively good substituent. The activity (24 μM), however, did not equal that of NECA. Recently Alexander et al reported 2-chloroadenosine to have an EC₅₀ value of 5.3 μM on the same cell line we used here¹⁴; the A_2A reference compound 2-[p-(2-carboxyethyl)-phenylethylamino]-NECA (CGS 21680, not tested in our study) showed a 40% response at a concentration of 100 μM and it was inactive at concentrations of 10 μM. Results from both this and our study suggest that C2 is not a suitable site of modification in order to gain A_2B potency. Probably there is little space in the receptor for a substituent in this position.

Substitution at the C8-position (TABLE 4) led to virtually inactive compounds. Bruns reported responses for 8-methylamino- and 8-dimethylaminoadenosine of 33% and 5% respectively¹¹. These responses are higher than the activities we measured with the MECA and NECA analogues. The 5′-N-methylcarboxamido moiety in 8-methylamino-MECA (24) probably causes the lower activity in comparison with Bruns’ adenosine analogue. More recently it has been shown that C8-substitution yields partial agonists for the A₁ receptor¹⁵.

N⁶-Substitution with appropriate groups has led to A₁ and A₃ selective agonists^16,17. In the N⁶-substituent series of MECA (TABLE 6), compounds have reduced activities compared to their NECA analogues (TABLE 5). It is concluded that an ethyl substituent on the 5′-carboxamido group is favoured over a methyl substituent. The selective A₃ agonist 2-chloro-N⁶-(3-iodobenzyl)-MECA (Cl-IB-MECA, 45)¹⁸ was included in the MECA series. This compound and the other 2-substituted derivatives 46, 47 (TABLE 6), 66 and 67 (TABLE 7) displayed lower activities than their C2-unsubstituted analogues 44 and 72. These results also indicate that C2-substitution diminishes A_2B potency.

N⁶ should be monosubstituted, because the disubstituted analogues 75 and 76 were inactive. The potency order for the N⁶-benzyl-substituted derivatives (TABLES 5, 6 and 7) was: 2 ≥ 3 > 4 substitution. However, no trends in the type of benzyl substituents could be discovered. In the adenosine series (TABLE 7) optima in potency were found for N⁶-phenyl (56) and N⁶-(R)-phenylisopropyl (59) substitution (50–60 μM). The N⁶-benzyl substituted analogue (57) showed the lowest activity (96 (μM). Metrifudil (63, N⁶-[(2-methylphenyl)methyl]adenosine) has been reported to be a moderately A_2B selective agonist, albeit five times less potent than NECA on guinea pig aorta contraction¹⁹. Here we report a 21% response for metrifudil. These results suggest that substituted N⁶-phenyl and N⁶-phenylisopropyl groups might lead to more active agonists than substituted benzyl groups.

The NH₂ group in the adenosine molecule is even allowed to bear rather large substituents (TABLES 7 and 8), not causing a total loss of activity (diadenosines 78 and 79, functionalized congeners 80–82). There must be some space in the receptor in this position. A part of the N⁶-substituent might occupy a region outside the receptor, as has been hypothesized for the adenosine A₁ receptor²⁰.

From the results of the deazaadenosines in TABLE 9 it is obvious that the purine nitrogen atoms are essential for potency of adenosine A_2B receptor agonists, especially N3 and N7. When N1 was substituted for CH, a response was left, such as in 1-deazaadenosine (90), 2-chloro-1-deazaadenosine (91) and 1-deaza-NECA (96). 1-Deaza-NECA²¹ was moderately potent with an EC₅₀ value of 16 μM. When the other nitrogen atoms were substituted, the response was negligible. In Bruns’ study 7-deazaadenosine was found to be inactive, whereas 1- and 3-deazaadenosine were not evaluated. 2-Azaadenosine on the contrary (not tested in our assay) was slightly less active than NECA¹¹. All results together suggest that the purine skeleton should not be altered. Similar observations have been done for analogues in TABLE 9 in binding experiments on the rat brain A₁ receptor and on the platelet A_2A receptor²².

Derivatives in which the purine ring system was modified into a (substituted) xanthinyl group are depicted in FIG. 1. All three hybrides gave very low responses. This accounts again for the importance of an intact purine ring system for A_2B agonists.

Low responses were also found for adenosines with modified ribose parts (FIG. 2) and with various modifications (FIG. 3). It should be mentioned that adenosine deaminase was added to the assay to avoid false hits caused by endogenous adenosine. The use of adenosine deaminase, however, may have inactivated some of the derivatives in FIG. 2. Especially suspect are 94, 95, 97 and 98, all having (besides a free 6-amino function) an unsubstituted hydroxymethylene group which has been proven necessary for adenosine deaminase substrates to be hydrolysed²³. 5′-Deoxyasteromycin (99), not a substrate for adenosine deaminase, did not show activity, whereas Bruns found a fairly good agonistic activity for asteromycin 98¹¹. Apparently, removal of the 5′-hydroxyl group is detrimental for A_2B receptor activity. This is also the case when the hydroxymethylene group is removed (103, compare with 66), or when a methyl group is added on C-5′ (93). Adding a methyl group on C-4′ also lowers activity (94 and 105, compare with 54). Substituting the ribose ring oxygen for a methylene group does not cause an increase of activity (98 and 104). Summarizing, for enhanced potency it is not allowed to add methyl groups in various positions in the ribose part, neither to remove the 5′-hydroxyl group or the 5′-methylene group, nor substituting the ring oxygen for a methylene group.

Conclusions and evaluation

By the study presented here, we gained more insight in the pharmacologic profile of the human adenosine A_2B receptor. Of all compounds, NECA was most potent with an EC₅₀ value of 3.1 μM, and still the most active agonist reported yet. Concerning the medicinal chemistry of the human A_2B receptor we conclude that 1) 5′-N-carboxamidoadenosines show highest potencies; NECA analogues are most potent in this series, MECA derivatives are less active; the rest of the ribose moiety should not be changed; 2) C2- and C8-substitution lower agonist activity; 3) N⁶-substitution reduces potency, but less severely than C2 and C8 substitution; 4) the purine ring system should not be modified into deazapurines or xanthines.

Two sites of substitution might be promising in order to increase agonist potency. The first site is the 5′-carboxamido moiety. Analogues with novel substituents on the carboxamido group could be synthesized and tested. Secondly, there appears to be space in the N⁶-region of the receptor. Appropriate substituents in this part of the molecule, possibly combined with a 5′-carboxamido function, might yield more potent and selective agonists.

Experimental section

Materials

Test system

A Chinese Hamster Ovary (CHO.K1) cell line stably transfected with human adenosine A_2B receptor cDNA^4,14 was used to test the adenosine analogues.

Compounds

[³H]-cyclic AMP was obtained from NEN (Du Pont de Nemours, ‘s-Hertogenbosch, the Netherlands).

Test compounds came from the following persons or companies: NECA (1) and 2-chloroadenosine (18) were purchased from Sigma; NCPCA (2) and compounds 3–8 and 27 were a gift of Dr. R.A. Olsson; compounds 15 and 16 were a gift of Takeda Chem. Ind. Ltd., Japan; compounds 19, 56 and 58 were synthesized according to literature procedures, with minor modifications^24–26; CPA (55) and R-PIA (59) were purchased from Boehringer Mannheim, Mannheim, Germany; compound 57 was a gift from Dr. A. van Aerschot, Leuven, Belgium; SPA (60) was a gift from Research Biochemicals Inc., Massachusetts, USA; compound 62 was a gift from Dr. J. Linden; compound 64 was a gift from Dr. O. Saiko and Prof. H.P. Wolf, Merck, Darmstadt, Germany; compounds 71–76, 78 and 79 have been synthesized and published before by Van Galen et al.²⁷; one of the authors, G.C., provided the deaza compounds 83–89; one of the authors, K. A. J., provided compounds 9–14,17, 20–54, 61, 63–70,77, 80–82,90–109.

Cell culture

CHO-K1 cells were grown under 5% CO₂/95% O₂ humidified atmosphere at a temperature of 37°C in DMEM supplemented with Hams F12 nutrient mixture (1/1), 10% newborn calf serum, 2mM glutamine and containing 100 IU penicillin and streptomycin. Cells were cultured in 10 cm Ø round plates and subcultured when grown confluent (approximately after 72 hours). PBS/EDTA containing 0.25% trypsine was used for detaching the cells from the plates. Experimental cultures were grown overnight as a monolayer in 24 wells tissue culture plates (400 μL/well; 0.8×10⁶ cells/well).

Cyclic AMP generation

Cyclic AMP generation was performed in DMEM/HEPES buffer (DMEM containing 50mM HEPES, pH 7.4, 37°C). To each well, washed twice with DMEM/HEPES buffer, 125 μL adenosine deaminase (final concentration 10 U/mL) and 125 μL rolipram (final concentration 100 μM) were added. After incubation for 40 minutes at 37°C, 100 μL of the ligand solution (final concentration 100 μM; maximal DMSO concentration 3%) was added. After 15 minutes, incubation at 37°C was terminated by removing the medium and adding 200 μL of 0.1 M HCl. Wells were stored at −20°C until assay.

Cyclic AMP determination

The amounts of cyclic AMP were determined after a protocol with cAMP binding protein²⁸ with the following minor modifications. As a buffer was used 150 mM K₂HPO₄/10mM EDTA/0.2% BSA FV at pH 7.5. Samples (20 μL) were incubated for 90 minutes at 0°C. Incubates were filtered over GF/C glass microfibre filters in a Brandel M-24 Cell Harvester. The filters were additionally rinsed with 4 times 2 ml 150 mM K₂HPO₄/10mM EDTA (pH 7.5, 4°C). Punched filters Were counted in Packard Emulsifier Safe scintillation fluid after 2 hours of extraction.