Interactions of Flavonoids and Other Phytochemicals with Adenosine Receptors

Introduction

Adenosine receptors mediate many of the physiological actions of adenosine in the periphery^1,2 and in the central nervous system.³ There are four subtypes of adenosine receptors: A₁, A_2A, A_2B, and A₃.⁴ Activation of A₁ receptors results in the bradycardiac,⁵ cerebroprotective,³ and antilipolytic effects of adenosine.⁷ Activation of A_2A receptors results in the hypotensive and antiplatelet aggregatory effects of adenosine.⁸ There are not yet any selective ligands with which to characterize physiological effects of selective activation of A_2B receptors. Several years ago a new subtype, A₃ receptors, was discovered⁹ while screening cDNA libraries for protein sequences resembling the G-protein-coupled receptors through the use of PCR (polymerase chain reaction). Activation of this receptor results in hypotension and promotion of release of inflammatory mediators from mast cells.¹⁰ We have introduced the first selective agonist ligands for the A₃ subtype, including the agonist IB-MECA (N⁶-(3-iodobenzyl)adenosine-5′-(N-methyluronamide).^11,12 In general adenosine antagonists are sought as renal protective,⁶ cognition enhancing,³ cerebroprotective,³ antiasthmatic,¹³ and antiinflammatory agents.¹³

Xanthines are the classical adenosine antagonists;² however, it is highly desirable to identify additional classes of antagonists. One reason for the current search for non-xanthine antagonists is that most xanthines bind only very weakly to A₃ receptors,¹⁴ for which no selective antagonists have yet been reported. A number of classes of non-xanthine adenosine antagonists (Figure 1) have already been reported. These include the nitrogen heterocycles¹¹ triazoloquinazolines,¹⁶ 9-methyladenines,¹⁷ pyrazolotriazolepyrimidines,¹⁸ and triazolotriazines.¹⁹ Many of the non-xanthine antagonists are relatively nonselective,^11,20 although selectivity for A₁ receptors¹⁷ or A₂ receptors^18,19 has been achieved. Several non-nitrogen heterocycles have also been reported to bind to adenosine receptors (Figure 1), including the protein tyrosine kinase inhibitor genistein (5,7-dihydroxy-4′-hydroxyisoflavone), 1,²¹ a natural product benzofurancarbaldehyde derivative 5-(3-hydroxypropyl)-7-methoxy-2-(3′-methoxy-4′-hydroxyphenyl)-3-benzo[b]furancarbaldehyde, 2,²² and the recently reported tetrahydrobenzothiophenes, e.g. BTH₄, 3.²³

A broad screening effort of phytochemicals in our laboratory, similar to that recently reported,²⁰ turned the focus to the flavones, members of the larger class of flavonoids, as a result of their relatively high affinity at adenosine receptors. Several of these derivatives even displayed significant affinity for A₃ receptors. Flavonoids are phenolic natural products derived largely from fruits, vegetables, bark, and flowers and are consumed in large doses in the human diet.²⁴ They have biological properties ranging from vascular protective and antioxidative properties²⁵ to antiinflammatory²⁶ and antineoplastic.²⁷ Since a related compound, genistein, 1, an isoflavone derivative, was shown to be a competitive adenosine antagonist at A₁ receptors in FRTL (thyroid) cells,²¹ we investigated the interactions of flavonoids in general as leads for developing novel adenosine antagonists.

Results

Affinity at Adenosine Receptors. K_i values at A₁ and A_2A receptors were determined in radioligand binding assays in brain membranes vs [³H]PIA or [³H]CGS 21680, respectively.^31,32 Affinity at human brain A₃ receptors expressed in HEK-293 cells⁴³ and at rat brain A₃ receptors stably expressed in Chinese hamster ovary (CHO) cells³⁰ was determined using [¹²⁵I]-AB-MECA [N⁶-(4-amino-3-iodobenzyl)adenosine-5′-(N-methyluronamide)].³⁰ A typical saturation curve for this radioligand at human A₃ receptors is shown in Figure 2. A K_d value of 0.59 nM was obtained, and the B_max was 0.159 pmol/mg of protein. Comparable curves at rat A₃ receptors are shown in ref 30, in which the K_d value was determined to be 1.48 nM. Thus, [¹²⁵I]AB-MECA is of slightly greater affinity at human vs rat A₃ receptors. In contrast to the rat receptor, binding at human A₃ receptors is optimal at 25 °C.

Flavone and its hydroxylated derivatives (compounds 4-21) bound to rat A₁ and A_2A adenosine receptors in the low micromolar range (Table 1). A structure–activity analysis indicated that the hydroxyl groups of naturally occurring flavones are not essential for affinity at adenosine receptors, since unsubstituted flavone, 4, displayed K_i values of 3.3 and 3.8 μM at A₁ and A_2A receptors, respectively. In general, there was a tendency toward lack of selectivity among the flavones between A₁ and A_2A receptors, except for compound 9, which had A₁ selectivity of 8-fold. The flavonol galangin (2-phenyl-3,5,7-trihydroxy-4H-1-benzopyran-4-one), 14, displayed the highest affinity at A₁ and A_2A receptors (≤1 μM). Tetramethylscutallarein, 11, circimaritin,^28b 13, galangin, 14, its trimethyl derivative, 15, and tetramethylkaempferol, 17, also had relatively high affinity among the analogues at the A₁ subtype (0.8–1.3 μM).

Table 1.

Affinities of Flavone Analogues Determined in Radioligand Binding Assays at A₁, A_2A, and A₃ Receptors^a-e

	Ki (μM) or % inhibition at the indicated concn (M)
compound	R1	R2	R3	(rA1)a	(rA2Δ)b	(hA3)c
4 (flavone)	H	H	H	3.28 ± 0.92	3.45 ± 1.16	16.9 ± 3.8
5	5-OH	H	H	2.17 ± 0.06	6.20 ± 1.24	47 ± 12% (10−4)
6	7-OH	H	H	3.03 ± 0.49	2.68 ± 0.68	41 ± 7% (10−4)
7	7-OH	3′,4′-(MeO)2	H	18.8 ± 3.6	35.4 ± 7.3	32 ± 4% (10−4)
8 (apigenin)	5,7-(OH)2	4′-OH	H	3.00 ± 0.29	7.58 ± 1.23	63 ± 1% (10−5)
9	5-OH-7-Me	4′-OMe	H	3.40 ± 0.35	28.0 ± 7.35	6.70 ± 1.78
10	5-OH-7-MeO	4′-OMe	H	d (10−4)	d (10−4)	69 ± 7% (10−5)
11 (tetramethylscutallarein)	5,6,7-(Me)3	4′-Me	H	1.29 ± 0.08	nd	4.48 ± 0.14
12 (hispidulin)	5,7-(OH)2-6-MeO	4′-OH	H	1.61 ± 0.29	6.48 ± 0.65	63 ± 2% (10−5)
13 (cirsimaritin)	5-OH-6,7-(MeO)2	4′-OH	H	1.20 ± 0.36	3.00 ± 0.70	1.72 ± 0.19
14 (galangin)	5,7-(OH)2	H	OH	0.863 ± 0.092	0.966 ± 0.164	3.15 ± 0.85
15	5,7-(Me)2O	H	OMe	0.509 ± 0.049	6.45 ± 1.48	1.21 ± 0.30
16	5,7-(AcO)2	H	OAc	11.6 ± 4.4	56.5 ± 9.5	17.5 ± 2.0
17 (tetramethylkaempferol)	5,7-(MeO)2	4′-OMe	OMe	1.07 ± 0.56	nd	3.37 ± 1.83
18 (rhamnetin)	7-OMe	3′,4′-(OH)2	OH	nd	nd	1.38 ± 0.18
19 (quercetin)	5,7-(OH)2	3′,4′-(OH)2	OH	2.47 ± 0.64	6.99 ± 0.89	see text
20 (pentamethylmorin)	5,7-(MeO)2	2′,4′-(OMe)2	OMe	27.6 ± 7.5	46.7 ± 2.7	2.65 ± 0.72
21 (hexamethylmyricetin)	5,7-(MeO)2	3′,4′,5′-(OMe)3	OMe	d (10−6)	nd	16.2 ± 2.2

Open in a new tab

^a

Displacement of specific [³H]PIA binding in rat brain membranes, expressed as K_i ± SEM in μM (n = 3–5).

^b

Displacement of specific [³H]CGS 21680 binding in rat striatal membranes, expressed as K_i ± SEM in μM (n = 3–6).

^c

Displacement of specific [¹²⁵I]AB-MECA binding at human A₃ receptors expressed in HEK-293 cells, in membranes, expressed as K_i ± SEM in μM (n = 2–3), or as a percentage of specific binding displaced at the specified concentration (M).

^d

Displacement of <10% of specific binding at the specified concentration (M). nd = not determined.

Typical inhibition curves for displacement of [¹²⁵I]AB-MECA binding at human A₃ receptors are shown in Figure 3A. The K_i values of many of the flavone derivatives at human A₃ adenosine receptors was in the range of 1–10 μM, with the most potent being the trimethyl derivative of galangin, 15, rhamnetin, 18, and circimaritin, 13. Thus, methylation of the hydroxyl groups of galangin enhanced affinity at A₃ receptors, while acetylation of the hydroxyl groups of galangin (e.g. 16) reduced affinity at all receptor subtypes. Unlike at A₁ and A_2A receptors, at A₃ receptors affinity of flavone derivatives was often (e.g. 13-15, 17-19) enhanced by the presence of multiple hydroxyl and/or methoxy substitution. Pentamethylmorin, 20, appeared to bind with some A₃ receptor selectivity (10- and 18-fold for human A₃ vs rat A₁ and A_2A receptors, respectively), with a K_i value of 2.65 μM. Substitution at the 3-position was not required (e.g. 13) to achieve micromolar affinity at adenosine receptors. The more selective A₃ ligands tended to contain multiple substitutions of the flavone, especially methoxy, of the phenyl ring, as in 7 and 20. In these compounds, affinity was diminished only at A₁ and A_2A receptors, thus enhancing selectivity for A₃ receptors.

Quercetin, 19, had an unexpected effect on the binding of [¹²⁵I]AB-MECA binding at human A₃ receptors. Rather than inhibiting the specific binding as in Figure 3A, it raised the level of binding of the radioligand. As shown in Figure 3B, at 30 μM quercetin the binding of the A₃ receptor radioligand was dramatically enhanced, with >10× the level of control binding in membranes of A₃-transfected HEK-293 cells. Curiously, cyanidin chloride, 25, although a very weak displacer in A₁ receptor binding (Table 2), also enhanced [¹²⁵I]ABMECA binding (Figure 3C). Since the additional binding occurred also in the presence of 200 μM NECA (540 ± 110% of control binding at 30 μM 25), it probably does not represent selective binding enhancement at A₃ receptors. Instead, cyanidin likely causes enhanced binding of [¹²⁵I]AB-MECA to a nonreceptor site on the membranes. It is not clear which structural features are necessary for this [¹²⁵I]AB-MECA effect. Rhamnetin, 18, a flavone also bearing the 3′,4′-dihydroxy groups, did not behave similarly (Table 1).

At rat A₃ receptors the affinity of flavone derivatives, like that of many xanthines,¹⁴ was weaker than at human A₃ receptors. The K_i values of 14 and 15 were 43.1 and 17.4 μM, respectively, at rat A₃ receptors (Figure 4). This represents a human A₃ vs rat A₃ ratio of 14-fold in both cases. Thus, within the same species (rat) compound 20 is likely nonselective; however, it may indeed prove to be selective when comparison is made between human A₃ and A₁/A_2A subtypes.

Reduction of the 2,3-olefinic bond of a flavonol, as in the trans-dihydroflavonol (±)-dihydroquercetin (Figure 5, Table 2), 22, greatly diminished affinity at adenosine receptors. However a combination of reduction and removal of the 3-hydroxy group, as in the flavanone sakuranetin, 23, restored affinity at A₁ and A₃ receptors (Table 2). Other stereoisomers were not studied for comparison to 22. Glycosidation, as in the 3-monoglycoside rutin, 26 (derived from quercetin, 19), and the 3,7-diglycoside robinin, 27 (derived from kaempferol), greatly diminished affinity at A₁ receptors. Similarly, catechin (a flavan-3-ol), 24, and cyanidin (a flavylium salt), 25, did not appreciably displace binding from A₃ receptors. The effects of fused aromatic substitution of the flavone ring system on receptor binding affinity were studied. α-Naphthoflavone, 28, had greater affinity at A₁ (0.8 μM), A_2A, and A₃ receptors than the corresponding β-isomer, 29. The isoflavone, genistein, 1, had very low affinity at human A₃ receptors.

A library of natural products of plant origin and related intermediates was screened for affinity at adenosine receptors. The compounds found to have considerable affinity for at least one subtype of adenosine receptors (Table 2) are shown in Figure 6. Included among the hits are alkaloids of the amaryllidaceae, 30 and 31,³⁴ oxogalanthine lactam, 32,³⁵ acetylhaemanthamine methiodide, 33,³⁶ hematoxylin, 35,³⁷ and arborinine, 36.³⁸ Hematoxylin, 35, a pigment isolated from logwood, was relatively potent at A₁ receptors, with a K_i value of 3.1 μM, and thus was highly selective for A₁ vs A_2A receptors. A synthetic intermediate (2,3-methylenedioxyfluoren-9-one), 34, with a ring system slightly similar to hematoxylin, bound to A₁ receptors, with a K_i value of 8.9 μM. Compound 33, was the weakest in binding among these derivatives. Affinities of the non-flavones at A₃ adenosine receptors (Table 2) were generally weak. There were several compounds that showed some degree of A₁ selectivity. In addition to 35, mentioned above, 32 was specific, and 34 was 10-fold selective for A₁ receptors. Only compound 36 was slightly selective for A_2A receptors (2-fold).

Discussion

The flavonoids are a ubiquitous family of phytochemicals that serve to protect plants from environmental stress. The biological effects of flavonoids in mammals have been studied for over 60 years,³⁹ and currently more than 100 preparations containing flavonoids are marketed in Europe,⁴² mainly for their action of decreasing the permeability and fragility of peripheral blood vessels.

The biochemical pathways with which flavonoids interact are numerous. Active flavonoids tend to inhibit cell activation at various levels. Many flavonoids, such as quercetin, 19,⁴⁰ inhibit protein kinase C in the micromolar range. Inhibition by flavonoids of many other mammalian enzymes, including phosphodiesterases⁴⁶ and transport ATPases, has been studied. At very high concentrations, certain flavonoids, such as quercetin, 19, but not apigenin, 8, have been found to inhibit adenosine deaminase.⁴⁷ Some flavonoids are carcinogenic and turn on genes or activate cells via fatty acid mobilization. It is also clear that some of the biological effects of flavonoids, such as inhibition of platelet function,⁴¹ are not fully explicable by any previously identified mechanism. Quercetin, 19, has been found to act synergisitically with β-adrenergic agonists to promote cAMP accumulation and lipolysis in isolated rat adipocytes.⁴⁶

We have shown that many flavones have considerable affinity for adenosine receptors. It is an historical coincidence that Albert Szent-Gyorgi,¹ the first investigator to detect the biological effects of adenosine, was also one of the first to investigate the biological effects of flavonoids.³⁹ The fact that many of the flavones have micromolar K_i values at adenosine receptors suggests that perhaps these receptors may be important in the spectrum of biological activities reported for flavones. By comparison, caffeine has a K_i value of 29 μM at A₁ receptors.² Middleton⁴⁰ has commented: “It is estimated that the flavonoids have been persistent in nature for somewhat over one billion years and, therefore, it seems entirely likely that they have been interacting with evolving animal forms/species over the eons”. Purinoceptors are also thought to be among the earliest G-protein-coupled receptors to appear in evolution.⁴⁵ Thus, just as the biological stimulant properties of the naturally occurring methylxanthines, such as caffeine and theophylline, were known long before their interaction with adenosine was suspected, we raise the possibility that the flavone derivatives are another class of natural products with considerable activity as a result of interaction with adenosine receptors. Moreover, the degree of activation of the receptors by endogenous adenosine may be more dependent on dietary factors than previously realized.

Most curious is the relatively high affinity among many of the flavones and flavonols at A₃ receptors. Unlike xanthines and nearly all other classes of non-xanthine adenosine antagonists,¹⁴ the affinity of flavone derivatives at the newly cloned A₃ receptor is generally in the same range as the affinity at A₁ and A_2A receptors. The A₃ receptors have already been implicated in vascular effects, inflammation, and cancer,⁴⁴ three areas in which flavonoids are also biologically active.^25–27 For example, flavone derivatives such as 8 and 19 have been shown to diminish the release of histamines from mast cells.²⁶ Adenosine agonists, presumably acting via A₃ receptors, have been shown to promote such a release.¹³ Thus, the unexpected finding of considerable affinity of flavone derivatives at both rat and human A₃ receptors may explain some of the previously observed biological effects of these compounds. Further pharmacological experiments will be required to determine to which degree these known effects of flavones are related to A₃ or other adenosine receptor action.

The structure-activity analysis of the flavonoids indicates that a carbonyl group at the 4-position promotes affinity at adenosine receptors. A comparison of α- and β-naphthoflavone also suggests that to achieve optimal receptor interaction the side of the molecule bearing the carbonyl group not be sterically crowded, as in the less active β-isomer. The various phytochemicals in Figure 5 were selected for their moderate adenosine receptor affinity by screening a library of several hundred phytochemicals. A feature common to all of the more potent adenosine receptors ligands is a phenyl ring that is fused to a five- or six-membered ring, which contains an α-carbonyl group. Compound 33, which is lacking the α-carbonyl group, was the weakest in this series in binding to A₁ receptors. This structural pattern, shared also with the flavones, suggests that there is a common pharmacophore among all of these phytochemicals. More detailed structure-activity studies are in progress in our laboratory.

The presence of multiple hydroxyl groups on the flavones tended to increase A₃ affinity and either had no effect on or decreased A₁ and A_2A affinity. The best A₃ selectivity was achieved with pentamethylmorin, 20, which contained a 2′-methoxy group on the phenyl ring that may impede free rotation of the ring.

In conclusion, this series of flavone derivatives may provide important leads for the development of potent and selective A₃ antagonists, which are as yet unknown.¹⁴ The present findings indicate that the chemical modification of flavonoids and other phytochemicals, in general, may lead to the development of novel adenosine antagonists.

Experimental Section