Structural determinants of A3 adenosine receptor activation: Nucleoside ligands at the agonist/antagonist boundary

Article abstract of DOI:10.1021/jm020211+

Mutagenesis of the human A₃ adenosine receptor (AR) suggested that certain amino acid residues contributed differently to ligand binding and activation processes. Here we demonstrated that various adenosine modifications, including adenine subs

Full text of DOI:10.1021/jm020211+

J . Med. Chem. 2002, 45, 4471-4484
4471
Str u ctu r a l Deter m in a n ts of A3 Ad en osin e Recep tor Activa tion : Nu cleosid e
Liga n d s a t th e Agon ist/An ta gon ist Bou n d a r y
Zhan-Guo Gao,‡ Soo-Kyung Kim,‡ Thibaud Biadatti,† Wangzhong Chen,‡ Kyeong Lee,‡ Dov Barak,§
Seong Gon Kim,‡ Carl R. J ohnson,† and Kenneth A. J acobson*,‡
Molecular Recognition Section, Laboratory of Bioorganic Chemistry, National Institute of Diabetes and Digestive and Kidney
Diseases, National Institutes of Health, Bethesda, Maryland 20892, Department of Chemistry, Wayne State University, Detroit,
Michigan 48202-3489, and Israel Institute for Biological Research, Ness Ziona, Israel
Received May 15, 2002
Mutagenesis of the human A3 adenosine receptor (AR) suggested that certain amino acid
residues contributed differently to ligand binding and activation processes. Here we demon-
strated that various adenosine modifications, including adenine substitution and ribose ring
constraints, also contributed differentially to these processes. The ligand effects on cyclic AMP
production in intact CHO cells expressing the A3AR and in receptor binding were compared.
Notably, the simple 2-fluoro group alone or 2-chloro in combination with N6-substitution
dramatically diminished the efficacy of adenosine derivatives, even converting agonist into
antagonist. Other affinity-increasing substitutions, including N6-(3-iodobenzyl) 4 and the
(Northern)-methanocarba 15, also reduced efficacy, except in combination with a flexible 5′-
uronamide. 2-Cl-N6-(3-iodobenzyl) derivatives, both in the (N)-methanocarba (i.e., of the
Northern conformation) and riboside series 18 and 5, respectively, were potent antagonists
with little residual agonism. Ring-constrained 2′,3′-epoxide derivatives in both riboside and
(N)-methanocarba series 13 and 21, respectively, and a cyclized (spiral) 4′,5′-uronamide
derivative 14 were synthesized and found to be human A3AR antagonists. 14 bound potently
at both human (26 nM) and rat (49 nM) A3ARs. A rhodopsin-based A3AR model, containing all
domains except the C-terminal region, indicated separate structural requirements for receptor
binding and activation for these adenosine analogues. Ligand docking, taking into account
binding of selected derivatives at mutant A3ARs, featured interactions of TM3 (His95) with
the adenine moiety and TMs 6 and 7 with the ribose 5′-region. The 5′-OH group of antagonist
N6-(3-iodobenzyl)-2-chloroadenosine 5 formed a H-bond with N274 but not with S271. The 5′-
substituent of nucleoside antagonists moved toward TM7 and away from TM6. The conserved
Trp243 (6.48) side chain, involved in recognition of the classical (nonnucleoside) A3AR
antagonists but not adenosine-derived ligands, displayed a characteristic movement exclusively
upon docking of agonists. Thus, A3AR activation appeared to require flexibility at the 5′- and
3′-positions, which was diminished in (N)-methanocarba, spiro, and epoxide analogues, and
was characteristic of ribose interactions at TM6 and TM7.
In tr od u ction
in their potency enhancement.4 Such modifications also
enhance the stability of adenosine derivatives in biologi-
cal systems, principally by preventing the action of
adenosine deaminase at the 6-position and/or adenosine
kinase at the 5′-position. The ribose moiety has also
been replaced effectively with a ring-constrained (N)-
methanocarba group (Chart 2),5,6 which selects a con-
formation that favors binding at the A3AR versus the
other subtypes. Thus, (N)-methanocarba derivatives 16,
19, and 20, which also contain a 5′-uronamide group,
were shown to be potent and selective full agonists at
the A3AR.
Several studies7-9 have suggested that modification
of the adenosine structure may reduce efficacy at one
or more of the AR subtypes. For example, C-8 substitu-
tion was shown to reduce efficacy of 6 at the A1AR.8 C-2
substitution was shown to reduce efficacy of adenosine
derivatives acting at the A3AR.8,9 We recently reported
the unexpected result that the commonly used A1AR-
selective agonist 7 acted as an antagonist at the human
A3AR,9 and this theme is expanded in the present study.
These findings suggest that rational modification of
The structure-activity relationship (SAR) of nucleo-
sides acting as agonists at adenosine receptors (ARs)
has been studied mainly at the “classical” A1 and A2A
-
AR subtypes. Several review articles1,2 have sum-
marized known SAR of adenosine derivatives, based
primarily on radioligand binding experiments. Modifica-
tions of nucleosides (Chart 1) known to increase receptor
binding affinity among adenosine agonists include: 5′-
uronamides such as 1, substitutions at the 2- (2, 3) and
N6-positions (4-7), or the replacement of the adenine
moiety with a 1,3-substituted xanthine moiety (8). A
pair of N6-phenylisoproyl R- and S-diastereomers 9 and
10 has demonstrated stereoselectivity of binding at A1
and A2AARs.3 Combinations of these groups, as in the
A3AR-selective agonists 11 and 12, are often additive
* Corresponding author: Dr. K. A. J acobson, Chief, Molecular
Recognition Section, Bldg. 8A, Rm. B1A-19, NIH, NIDDK, LBC,
Bethesda, MD 20892-0810. Tel: 301-496-9024; fax: 301-480-8422;
e-mail: kajacobs@helix.nih.gov.
‡ National Institutes of Health, Bethesda.
† Wayne State University.
§ Israel Institute for Biological Research.
10.1021/jm020211+ CCC: $22.00 © 2002 American Chemical Society
Published on Web 08/27/2002

4472 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20
Gao et al.
Ch a r t 1. Chemical Structures of Adenosine Derivatives Used in the Present Study and Containing a 9-Ribose Moiety
are typically of high affinity across species.12 For other
Ch a r t 2. Chemical Structures of Adenosine Derivatives
Used in the Present Study and Containing an
(N)-Methanocarba Moiety
G protein-coupled receptors, there are examples of
minor modifications in the structure of agonists leading
to loss of efficacy, e.g., conversion of agonists to antago-
nists at P2Y1 nucleotide receptors,13 opiate receptors,14
and neurokinin receptors.15
In this study, we compared binding and functional
properties of representative ligand probes, i.e., both
previously reported2,4-6 and novel derivatives of adeno-
sine. To probe the effects of rigidifying the ribose moiety,
novel 2′-,3′-epoxide derivatives 13 and 21, in both
riboside and (N)-methanocarba series, respectively, and
a cyclized (spiro) 4′,5′-uronamide derivative 14 were
prepared. We have categorized modifications of adeno-
sine, including adenine substitution and ribose ring
constraints, according to their typically differential
effects on the binding and activation processes at the
A3AR, and explained the results based on a receptor
model. The proposed mode of docking nucleoside-based
agonists and antagonists the A3AR also took into
account binding characteristics of selected derivatives
at mutant A3ARs.
Resu lts
Ch em ica l Syn th esis. The 2′,3′-anhydroadenosine
derivative 21 was synthesized as shown in Scheme 1.
The (N)-methanocarba triol 18 was treated with R-ac-
etoxyisobutyryl bromide in wet acetonitrile,16,17 followed
by direct treatment of the mixture of both isomers of
trans-3′(2′)-bromo-2′(3′)-acetate with Amberlite (OH-)
resin in methanol.18 Similarly for the 5′-methylamide
12 epoxidation at the 2′,3′-positions was accomplished
by reaction with R-acetoxyisobutyryl bromide followed
by cyclization in base to yield 13.
adenosine agonists into antagonists may be possible.
There are many nonnucleoside classes of A3AR receptor
antagonists;1,10 however, the approach of converting
adenosine derivatives into pure antagonists of high
affinity has not previously succeeded.11 Moreover, an-
tagonists of high affinity and selectivity at nonprimate
species are still lacking.9,10 This study may remedy that
lack, since the lead compounds, i.e., adenosine agonists,
The spiral analogue 14 was synthesized in 10 steps
(Scheme 2) from a brominated cyclopentenone derivative

Structural Determinants of A3 Adenosine Receptor Activation
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20 4473
Sch em e 1. Synthesis of Ring-Constrained Adenosine Derivatives 21 and 13 Containing 2′,3′-Epoxide Groupsa
a
Reagents and conditions: (a) Acetoxyisobutyryl bromide, CH3CN. (b) Amberlite IRA-400 (OH-) resin, MeOH.
Sch em e 2. Synthesis of a Ring-Constrained Spiro Adenosine Derivative 14 Containing a 4′,5′-Lactam Groupa
a
Reagents and conditions: (a) allylmagnesium chloride, THF, -78 °C. (b) Br2, Et3N. (c) O3, MeOH/pyridine, -78 °C, then dimethyl
sulfide. (d) MeOH, TsOH, 65 °C. (e) Zn, MeOH, 65 °C. (f) (1) O3, CH2C12, then dimethyl sulfide; (2) p-MeOC6H4CH2NH2, NaCNBH3. (g)
ceric ammonium nitrate, MeCN/H2O. (h) K2CO3, MeOH, 18-crown-6. (i) (1) HC1 (3%), MeOH, 65 °C; (2) Ac2O, Et3N. (j) hexamethyldisilazane,
6-chloropurine, TMSOTf, MeCN, 85 °C. (k) (1) NH3, MeOH, 20 min; (2) 3-iodobenzylamine hydrochloride, Et3N, t-BuOH, 85 °C, 6 days.
24. The use of enantiopure 24 in the synthesis of
4-methyl ribose systems has been described.19,20 For the
synthesis of the target molecule 14, a more functional-
ized side chain has to be incorporated into the ribose
system. Allylmagnesium bromide was added to 24
providing the acid labile tertiary alcohol 25. Direct

4474 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20
Gao et al.
Ta ble 1. Binding and Function of a Series of Adenosine
Derivatives at Human A3 Receptors Expressed in CHO Cellsa
EC50 (nM)
efficacy (%)
affinity (Ki, nM)
1
2
3
4
5
6
7
8
26 ( 14
2660 ( 880
1630 ( 610
85 ( 13
NA
103 ( 6
31 ( 3
100 ( 7
46 ( 8
0
35 ( 12
99 ( 13
87 ( 24
5.8 ( 0.4
1.8 ( 0.1
72 ( 12b
38 ( 6b
816 ( 206
8.7 ( 0.9
68 ( 12
1.8 ( 0.7
1.4 ( 0.3
1100 ( 110
29.3 ( 1.6
353 ( 54
3.4 ( 0.6
9.2 ( 0.7
1.9 ( 0.7
1.3 ( 0.1
2.1 ( 0.4
572 ( 113
1.5 ( 0.2
114 ( 16
242 ( 47b
NA
97 ( 4
0
828 ( 236
27 ( 11
210 ( 56
3.6 ( 1.3
2.8 ( 1.4
NA
96 ( 4
102 ( 6
97 ( 3
100
99 ( 6
7 ( 1
0
41 ( 5
104 ( 7
13 ( 1
3 ( 2
101 ( 5
103 ( 9
0
9
10
11
12
13
14
15
16
17
18
19
20
21
36
37
F igu r e 1. The effects of a series of adenosine derivatives on
forskolin-stimulated cyclic AMP accumulation in intact CHO
cells expressing the human A3AR. Results shown are from one
representative experiment, and the EC50 values listed in Table
1 are from 3 to 5 experiments performed in duplicate.
NA
803 ( 329
1.4 ( 0.2
NA
NA
5.4 ( 0.9
3.3 ( 1.1
NA
1.6 ( 0.3
158 ( 28
include the A1AR selective agonists 6, 7, 9, and 10; an
A2A receptor selective agonist 2-[p-(carboxyethyl)phen-
103 ( 6
98 ( 5
ylethylamino]-5′-N-ethylcarboxamidoadenosine 37; the
A3 selective ligands 4, 5, 11, 12, 16, 17, 18, 19, and 20;
the ligands of low selectivity 1, 2, 3, 8, 15, and 36; and
the newly synthesized derivatives 13, 14, and 21. Some
of these adenosine derivatives, including N6-(3-iodoben-
zyl) derivatives 11 and 12, already demonstrated to be
potent A3AR agonists,2,4-6 were included as reference
compounds in the present study. Other adenosine
derivatives, especially those newly synthesized, had not
previously been evaluated in functional assays, and
might otherwise be presumed to be agonists being
adenosine derivatives.
In the derivatives A3AR binding assay, the 2-chloro
substituent and the N6-iodobenzyl group generally
increased the affinity of the adenosine, while the
substitution of the ribose moiety with an (N)-methano-
carba moiety5,6 slightly decreased or did not influence
the affinity significantly.
A recent report9 demonstrated that the potent A1AR
agonist N6-cyclopentyladenosine 6 is also a full agonist
at A3ARs, while the corresponding 2-chloro derivative
7 is a moderately potent, full antagonist at the A3AR.
However, it was not clear if other 2-chloro-substituted
analogues would behave like 6 or like 7. To explore
further the effects of the 2-chloro substituent in the
activation of the human A3AR, two pairs of N6-(3-
iodobenzyl) adenosine derivatives with or without chloro
substituent (the ribosides 4 versus 5 and the methano-
carba derivatives 17 versus 18) were compared. Figure
1 demonstrated that 4 was more efficacious than 5,
while 17 was more efficacious than 18. 5′-Alcohol
derivatives 5 and 18 and the spiral derivative 14 were
demonstrated to have little if any detectable agonist
activity (Table 1), suggesting possible antagonism. This
was further demonstrated by their effects on the con-
centration-response curve of the agonist-induced inhi-
bition of forskolin-stimulated cyclic AMP production in
intact CHO cells (Figure 2). They shifted the agonist
concentration-response curve to the right in a dose-
dependent manner, corresponding to KB values24 of 2.4
( 0.4, 3.0 ( 0.6, and 2.0 ( 0.3 nM for compounds 5, 14
(not shown), and 18, respectively.
a
The activation of the human A3 receptor by adenosine deriva-
tives was investigated by examining their effects on forskolin-
stiumulated cyclic AMP accumulation in intact CHO cells express-
ing the human A3AR. Abbreviations: 1 NECA, 5 MRS542, 8
DBXRM, 11 IB-MECA, 12 Cl-IB-MECA, 14 MRS1292, 17
MRS1743, 18 MRS1760, 20 MRS1898, 36 N6-(4-amino-3-iodo-
benzyl)adenosine-5′-N-methyluronamide, 37 2-[p-(carboxyethyl)-
phenylethylamino]-5′-N-ethylcarboxamidoadenosine. The binding
affinity was determined using [125I]36. Data were from 3 to 5
experiments. NA, not applicable. Data from Gao and J acobson.9
b
ozonolysis of 25 in methanol/pyridine led to ribose
derivative 27 but only in modest yield and on a small
scale and was unsuccessful when attempted on a larger
scale. Protection of the external alkene as the dibromide
followed by ozonolysis afforded the ribose framework 28
in 82% yield. Conversion of 28 to the â-methylglycoside
29 and reductive elimination of bromine smoothly
provided the 4-allylribose 30. Ozonolysis of 30 followed
by reductive amination with 4-methyoxybenzylamine/
NaBH3CN gave lactam 31. Oxidation of 31 with ceric
ammonium nitrate gave a mixture of deprotected 32 and
the benzoyl derivative 33. The latter was readily
converted to 32 upon treatment with potassium carbon-
ate in methanol. To proceed with Vorbru¨ggen nucleosi-
dation,21 the isopropylidene protection was exchanged
for acetate by treatment of 32 with HCl in methanol.
Treatment of 34 with silylated 6-chloropurine and TMS
triflate in acetonitrile at 65 °C followed by chromatog-
raphy of the product mixture gave nucleoside 35 in 81%
yield. Brief exposure of 35 to ammonia and then
3-iodobenzylamine gave the target 14 in 87% yield.
Biologica l Activity. (a ) In ter a ction of a Wid e
Ra n ge of Ad en osin e Der iva tives w ith th e Hu m a n
A3AR Sta bly Exp r essed in CHO Cells. The binding
and activation of the human A3AR by a wide range of
adenosine derivatives (Charts 1 and 2) were investi-
gated. Binding was studied using the radioiodinated
form of the agonist N6-(4-amino-3-iodobenzyl)adenosine-
5′-N-methyluronamide ([125I]36),22 and functional effects
on forskolin-stimulated cyclic AMP production were
examined in intact Chinese hamster ovary (CHO) cells
expressing the human A3AR (Table 1).23 The previously
reported adenosine derivatives1 used in this study
The observation that 17 was less efficacious than 4
(Figure 1) suggested that the replacement of the ribose

Structural Determinants of A3 Adenosine Receptor Activation
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20 4475
F igu r e 2. Shift of the dose-response curves inhibition by the agonist 12 of forskolin-stimulated cyclic AMP accumulation in
intact CHO cells expressing the human A3AR, in the presence of antagonists 7 (A) and 18 (C). Schild analysis of the data for 7
(B) and 18 (D).
ring by an (N)-methanocarba ring system may also
decrease agonist efficacy. Hence, we further compared
another pair of compounds, the 2-chloro riboside 3 and
the corresponding (N)-methanocarba derivative 15.
Compound 3 was a full agonist at the human A3AR,
while 15 was a partial agonist (Figure 3A) with a
maximal efficacy of ∼40% of the maximal efficacy of 3.
The 2-chloro substituent decreased the agonist ef-
ficacy for a number of multi-substituted adenosine
analogues; however, this was not a general phenom-
enon, since the 2-chloro substituted 12, and 20, and
2-chloroadenosine 3 itself, were full agonists at the
human A3AR. Hence, the efficacy-diminishing effect of
the chloro substituent also depended on which ad-
ditional structural features such as an N6-substituent
were present. Curiously, the smaller and more polar
2-fluoro substituent (see 2) was able to decrease the
efficacy as A3AR agonist, even in an absence of N6-
substituent. Compound 2 proved to be a partial agonist
at the A3AR, with ∼30% of the maximal efficacy of 3
(Table 1).
It has been suggested that the ribose 2′ and 3′-
hydroxyl groups are critical in the activation of A1-
ARs.25,26 Here we further examined the role of these two
hydroxy groups in the activation of A3ARs. The epoxi-
dation of the two hydroxyl groups of the potent agonist
F igu r e 3. Differential effects of nucleoside 5′-CH2OH deriva-
12 resulting in compound 13 led to a decrease of binding
affinity by more than 1000-fold, and the efficacy was
also dramatically decreased to only ∼7% the maximal
efficacy of 12 (Figure 3B). Similarly, the epoxidation of
the methanocarba derivative 18 resulting in compound
21 also induced a 300-fold affinity decrease, suggesting
the critical role of one or both of the hydroxyl groups in
binding and activation processes at the human A3AR.
To further analyze the structural determinants of
activation of the human A3AR, we tested compound 8,
tives 2, 3, and 15 (A) and 5′-uronamide derivatives 8, 12, and
13 (B) on forskolin-stimulated cyclic AMP accumulation in
CHO cells expressing the human A3AR. Results shown are
from a representative experiment, and the EC50 values are
listed in Table 1 are calculated from at least three experiments.
a 5′-uronamide derivative in which the adenine moiety
was replaced with a 1,3-dibutylxanthine moiety. Al-
though the 1,3-dialkylxanthine moiety alone is an AR
antagonist, 8 was found to be a full agonist at the
human A3AR (Figure 3B), similar to previous findings

4476 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20
Gao et al.
Ta ble 2. Effects of a Nucleotide GTPγS on Ligand Competition
Curves for a Tritiated Antagonist Binding to Membranes from
CHO Cells Expressing the Human A3ARa
Ta ble 3. Binding Affinity of a Series of Adenosine Derivatives
for Human Wild Type and H95A and W243A Mutant
Receptorsa
Ki (nM)
Ki (nM)
control
+GTPγS
shift (fold)
WT
H95A
W243A
5
7
12
14
18
38b
3.2 ( 0.3
38 ( 6
3.9 ( 0.4
34 ( 7
1.2
Adenosine Derivatives
0.9b
4.2
4
5
6
7
11
12b
15
1
17
18
19
20b
21
3.8 ( 0.8
1.9 ( 0.2
142 ( 23
114 ( 17
1.6 ( 0.2
2.3 ( 0.6
83 ( 9
0.8 ( 0.2
3.9 ( 0.4
1.7 ( 0.1
2.1 ( 0.3
1.4 ( 0.4
584 ( 130
85 ( 34
132 ( 26
615 ( 154
796 ( 113
ND
60 ( 15
ND
ND
99 ( 17
460 ( 95
ND
143 ( 36
ND
6.0 ( 1.3
2.4 ( 0.3
137 ( 24
100 ( 9
0.9 ( 0.1
2.9 ( 0.5
46 ( 19
0.9 ( 0.1
2.7 ( 0.2
1.3 ( 0.1
1.2 ( 0.4
2.1 ( 0.6
490 ( 107
1.5 ( 0.2
26 ( 7
6.3 ( 0.4
24 ( 3
0.92
1.5
1.0
2.7 ( 0.3
1.2 ( 0.1
4.0 ( 0.3
1.2 ( 0.2
a
Binding was carried out in membranes (80 µg of protein) of
CHO cells stably expressing recombinant the human A3AR. The
concentration of [3H]8-ethyl-4-methyl-2-phenyl-(8R)-4,5,7,8-tet-
rahydro-1H-imidazo[2,1-i]-purin-5-one was 5 nM. The concentra-
tion of GTPγS used was 100 µM. Data were expressed as mean (
b
SEM from three independent experiments. 38 is N-[9-chloro-2-
(2-furanyl)[1,2,4]triazolo[1,5-c]quinazolin-5-yl]benzene-aceta-
mide. Data are from Gao and J acobson.9
Nonnucleoside Antagonistsc
39
40
238 ( 39
ND
ND
2150 ( 230
818 ( 97
with the rat A3AR.12 It should also be noted that 1,3-
dibutylxanthine itself has extremely low affinity for the
A3AR, and the ribose attachment increased both the
affinity and the efficacy of the xanthine derivative (data
not shown).
78 ( 11
ND, not determined. [125I]I-AB-MECA, 1 nM. Protein concen-
a
tration (10-20 µg of protein). b Values from ref 29. c 39 is 5-propyl-
2-ethyl-4-propyl-3-(ethylsulfanylcarbonyl)-6-phenylpyridine-5-car-
boxylate; 40 is S-1,4-dihydro-2-methyl-6-phenyl-4-(phenylethynyl)-
3,5-pyridinedicarboxylic acid, 3-ethyl 5-(phenylmethyl) ester.61
(b) Effects of a Nu cleotid e GTP γS on th e Dis-
p la cem en t of a n An t a gon ist R a d ioliga n d by t h e
Ad en osin e Der iva tives. The effects of some of the
adenosine derivatives on the human A3AR were further
examined utilizing a functional assay based on the
method of guanine nucleotide-induced shift of agonist
competing for antagonist radioligand binding.8,22,27 [3H]8-
ethyl-4-methyl-2-phenyl-(8R)-4,5,7,8-tetrahydro-1H-imi-
dazo[2,1-i]-purin-5-one is a novel A3AR-selective radi-
oligand introduced by Mu¨ller et al.28 and has recently
been used to characterize the human A3AR.23,29 The
nonhydrolyzable GTP analogue, GTPγS (100 µM), caused
a 4-fold decrease in the affinity of the agonist 12 in
competition for [3H]8-ethyl-4-methyl-2-phenyl-(8R)-
4,5,7,8-tetrahydro-1H-imidazo[2,1-i]-purin-5-one bind-
ing (Table 2), but it only had a slight or no effect on the
affinity of adenosine-derived antagonists 5, 7, 14, and
18. This indicated that in this functional assay these
four nucleosides behaved essentially the same as the
classical A3AR antagonist 38 N-[9-chloro-2-(2-furanyl)-
[1,2,4]triazolo[1,5-c]quinazolin-5-yl]benzene-aceta-
mide.9,22 Thus, the unexpected loss of agonism among
various adenosine derivatives was confirmed using
several different methods5 of determining the degree of
receptor activation.
(c) Mod u la tion of th e Bin d in g of Ad en osin e
Der iva tives to th e Hu m a n A3AR by H95A a n d
W243A Mu ta tion s. We further29 examined the binding
of a number of adenosine derivatives with or without a
2-chloro substituent to the wild type and H95A mutant
ARs (Table 3). The H95A mutation induced a larger
rightward shift of the compounds having a 2-chloro
substituent, although the range of shift also depended
on additional substitution. Indeed, although a fairly
subtle change in structure, addition of a 2-chloro sub-
stituent had a major effect on the interaction of adeno-
sine derivatives with the A3AR.
wild type and W243A mutant receptors (Table 3). It was
found that this residue was involved in the binding of
only nonnucleoside antagonists such as 5-propyl-2-ethyl-
4-propyl-3-(ethylsulfanylcarbonyl)-6-phenylpyridine-5-
carboxylate 39 and the S-enantiomer of 1,4-dihydro-2-
m et h yl-6-ph en yl-4-(ph en ylet h yn yl)-3,5-pyr idin edi-
carboxylic acid, 3-ethyl 5-(phenylmethyl) ester 40.
However, mutation of this residue had no effect on the
affinity of either agonists or antagonists derived from
adenosine (Figure 4). Hence, this residue discriminated
antagonist ligands of different chemical classes.
(d ) Effects of th e Ad en osin e Der iva tives on Ra t
ARs. It is known that currently available A3AR antago-
nists, including 3822 and 5-[[(4-methoxyphenyl)amino]-
carbonyl]amino-8-propyl-2-(2-furyl)-pyrazolo[4,3-e]1,2,4-
triazolo[1,5-c]pyrimidine (MRE 3008F20),30 have very
high affinity for the human A3AR, but low affinity for
the rat A3AR. In contrast, typical A3AR agonists have
similar A3AR affinity in both species. Hence, potent rat
A3AR antagonists could be identified by modifying the
structure of the adenosine and adenosine derivatives.
As shown in Table 4, antagonists 5 and 18 displayed
high affinity for both the recombinant rat A3AR trans-
fected in CHO cells and the endogenous rat A3AR in
RBL-2H3 basophils.12 The potent human A3AR triaz-
oloquinazoline antagonist22 38 at 10 µM displaced less
than 50% of [125I]36 binding. Consistent with lack of
agonism neither 5 nor 18 at e 10 µM induced a
significant inhibition of forskolin-stimulated cyclic AMP
production in CHO cells expressing the rat A3AR (data
not shown).
The Ki values (nM) at rat brain ARs were measured
as described.6 The Ki values (µM) were determined to
be 3.92 ( 0.40 (A1) and >20 (A2A) for 13, 12.1 ( 2.4 (A1)
and 20.8 ( 0.4 (A2A) for 14, and 24.2 ( 3.8 (A1) and 47.9
( 7.0 (A2A) for 21.
(e) Molecu la r Mod elin g of th e Hu m a n A3AR a n d
Liga n d Dock in g. 3D-structures of all domains of the
A3AR except the C-terminal region were newly con-
A tryptophan residue (W243) in TM6 has been sug-
gested to be involved in antagonist binding to the
human A3AR.29 Here we further examined the interac-
tions of the present set of adenosine derivatives with

Structural Determinants of A3 Adenosine Receptor Activation
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20 4477
F igu r e 4. Differential modulation of binding at wild type and W243A mutant human A3ARs for adenosine-derived antagonists
(A, 5; B, 7) and nonnucleoside antagonists (C, pyridine derivative 39; D, dihydropyridine derivative 40). Results shown are from
a representative experiment, and the Ki values listed in Table 1 are calculated from three experiments. [125I]4-amino-3-iodobenzyl)-
adenosine-5′-N-methyluronamide, 1 nM. Protein concentration (10-20 µg of protein).
Ta ble 4. Binding Affinity of Adenosine Derivatives for
Endogenous Rat A3 Receptors Expressed on RBL-2H3
Basophilic Cells and Recombinant Rat A3 Receptors Stably
Å in the previous model.32 Both structures showed
Transfected in CHO Cellsa
in R-helix of the calculated A3AR model and the tem-
plate molecule was found to be 1.37 Å, compared to 2.69
overall similarities, especially the regions having de-
fined secondary structures.
Ki (nM)
5
12
14
18
38
39
We used an automated docking procedure to deter-
mine the most favorable binding locations and orienta-
tions for several ligands (1, 4, 5, 12, and 14) and
compared these results with our experimental results.
An energetically favorable model that correlated well
with the experimental result was selected.
The putative binding site surrounding the full agonist
12 docked in the A3AR is shown in Figure 5A. Residues
that were within 5 Å proximity to the ligand in this
putative binding site were: L91 (3.33), T94 (3.36), H95
(3.37), Q167 (EL2), F168 (EL2), F182 (5.43), F239 (6,-
44), W243 (6.48), L246 (6.51), N250 (6.55), S271 (7.42),
H272 (7.43), and N274 (7.45). Mutagenesis results for
several AR subtypes29,33-35 were consistent with molec-
ular modeling that featured direct interaction of ligands
with TMs 3, 6, 7, and EL2.
The purine ring was located in a hydrophobic pocket
defined by L91 (3.33) and H95 (3.37) and the amine of
the N6-substituent interacted with N250 (6.55) through
H-bonding, which was similar to our A2AAR docking
model.33 The inability of the N250A mutant A3AR29 or
the corresponding mutant A2AAR33 to bind either ra-
diolabeled agonist or antagonist was consistent with a
proposed direct interaction of this residue with 12 and
other ligands. Additional aromatic-aromatic stabiliza-
tion of the N6-benzyl group in compound 12 interacting
with surrounding F182 (5.43) and F168 (EL2) may be
responsible for the increase of binding affinity. The 2′-
OH and 3′-OH groups of the ribose ring were involved
CHO
1.6 ( 0.3 0.9 ( 0.2 49 ( 7 1.3 ( 0.1 >10 µM 234 ( 36
RBL-2H3 2.7 ( 1.2 ND
51 ( 17 2.9 ( 0.8 ND
357 ( 114
a
[
125I]4-amino-3-iodobenzyl)adenosine-5′-N-methyluronamide (1
nM) was used as a radioligand to determine the binding affinity.
ND, not determined. The Ki values (µM) for 14 at rat brain ARs,
measured as described:6 12.1 ( 2.4 (A1) and 20.8 ( 0.4 (A2A),
indicating selectivities of 250- and 420-fold, respectively.
structed using homology modeling from the X-ray
structure of rhodopsin.31 The inclusion of loop regions
reduced the large deviation of the end of the TM
bundles, since the previous A3AR model32 had a high
rmsd (root-mean-squared deviation) value between back-
bone atoms in the R-helix relative to the rhodopsin
structure. To validate the reliability of the calculated
model, stereochemical accuracy, packing quality, and
folding reliability were checked. Using a Ramachandran
plot, æ and ψ angles were compared with the rhodopsin
crystal structure. All amino acids in the R-helices were
located in the favored region of right-handed R-helix in
the Ramachandran plot. Only 3% of the residues within
loop segments were in the sterically disallowed region.
From calculated ω angles, there were no cis peptide
bonds in the calculated A3AR model as well as in the
rhodopsin structure. All CR atoms except Cys displayed
S-chirality. For the packing quality, there were no bump
regions in the calculated A3AR model. To check the
folding reliability, the rmsd between backbone atoms

4478 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20
Gao et al.
formed H-bonds with S271 (7.42) and T94 (3.36). An
additional hydrophobic interaction between the terminal
alkyl group at the 5′-position and F239 (6.44) stabilized
the complex.
A difference in comparison to our former model29 was
seen in the role of H95. Our previous human A3AR
model29 featured a direct interaction between the 2-chlo-
ro substituent of adenosine derivatives with a histidine
residue (H95) in the third transmembrane domain
(TM3) of the human A3AR, yet experimental evidence
did not support such an interaction.29 The present study
demonstrated the role of the 2-chloro substituent in
ligand recognition by the human A3AR, yet it did not
postulate a direct contact with H95, thus removing the
discrepancy. The previous model29 proposed a conforma-
tion about the glycosidic bond that was intermediate
between syn- and anti-conformations and also a direct
interaction of the 2-Cl substituent with the side chain
of H95. However, the relative energy difference between
the intermediate and anti-conformations of the purine
ring of 12 and other ligands in bound complex showed
that the anti-conformation, which did not show a direct
interaction with the H95 side chain, was more energeti-
cally favorable than the intermediate conformation in
the A3AR. In the anti-conformation, H95 was a compo-
nent of the hydrophobic pocket that accommodated the
purine ring through hydrophobic interaction. This model
featuring the anti-conformation as active conformation
of full agonists 1 and 12 was consistent with the binding
studies of methanocarba-adenosine analogues,5,6 since
the (N)-methanocarba modification of adenosine stabi-
lized the anti-conformation.6 The other evidence which
did not support a direct 2-Cl-H95 interaction was that
the H95A mutation reduced the binding affinity of both
2-Cl and 2-H adenosine analogues, although the effect
on 2-Cl-analogues was larger. Cl may be involved
indirectly by altering the hydrophobic interaction with
the purine ring by inducing a change of π-electron
density due to its electron-withdrawing effect. The
efficacy-reducing effect of 2-F would be consistent with
an electronic effect of 2-substitution.
Binding modes of nucleoside 5′-N-alkyluronamide-
containing agonists 1 and 12 compared to nucleoside
antagonists 5 and 14 were different with respect to the
5′-position of the ribose ring. The association with TMs
6 and 7 with the flexible alkylamide group of 5′-
uronamide agonists 1 and 12 was stabilized by hydro-
phobic interaction with F239 (6.44) as well as by
H-bonding with S271 (7.42) and T94 (3.36). In the case
of nucleoside antagonist binding, the 5′-OH group of N6-
(3-iodobenzyl)-2-chloroadenosine 5 formed a H-bond
with N274 but not with S271. This bonding occurred
as a result of a small movement of the 5′-substituent
toward TM7 and away from TM6. The carbonyl group
of the cyclic amide in the antagonist 14 still retained
H-bonding with S271 (7.42) but lost additional H-
bonding with T94 (3.36) due to the different geometry
of NH in flexible versus rigid alkylamide groups.
F igu r e 5. The optimized conformations of the human A3AR
following docking of agonist (A) 12 or antagonists (B) 5 and
(C) 14. The ligand molecules were colored according to atom
type.
in H-bonding with Q167 (EL2) and H272 (7.43), respec-
tively, which was supported by our A3AR neoceptor
model.32 The epoxidation of 2′- and 3′-OH groups
decreased the binding affinity by about 1000-fold (com-
pound 12 vs 13) supporting a possible direct interaction
of the hydroxyl group(s) with the receptor. The carbonyl
and amino groups of the 5′-uronamide substituent of 12
Another pronounced difference between nucleoside
agonist 5′-N-alkyluronamide derivatives 1 and 12 and
the antagonists 5 and 14 seen in the docked A3AR
complexes was the position of the conserved W243
(6.48). For antagonists 5 or 14 the model achieved
energetic convergence without the movement of the side

Structural Determinants of A3 Adenosine Receptor Activation
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20 4479
N6-cyclopentyl group alone did not reduce efficacy;
however, the combination of 2-chloro and N6-cyclopentyl
in the present pharmacological system appeared to
abolish agonism entirely.9 Other N6 substitution (e.g.,
diastereomers 9 and 10) had no effect on efficacy.
The (N)-methanocarba substitution of the ribose ring,
resulting in a rigidification of the pseudoribose moiety
in an A3AR-preferring conformation,6 while preserving
and/or enhancing binding selectivity, also appeared to
reduce A3AR efficacy. Thus, in the 2-H case, compound
17 was less efficacious than 4. Similarly, in the 2-Cl
case, 15 was less efficacious than 3.
The above-mentioned reduction of efficacy was subject
to reversal upon, at least certain, 5′-substitution of the
molecule. The 5′-alkylamide group (either N-ethyl as in
1, or N-methyl) appeared to restore efficacy. All of the
acyclic 5′-amide derivatives, except the 2′,3′-epoxide 13,
were full agonists. Comparison of pairs of compounds
differing only in the presence of the 5′-alkylamide group
demonstrated that the efficacy-preserving effect of this
substituent took precedence over other modifications,
i.e., N6-(3-iodobenzyl), 2-halo, and methanocarba, that
otherwise reduced efficacy. Thus, among N6-(3-iodoben-
zyl) analogues 11 was more efficacious than 4, and for
the corresponding 2-chloro analogues 12 was more
efficacious than 5. For N6-(3-iodobenzyl) and (N)-metha-
nocarba analogues, 19 was more efficacious than 17, and
for the 2-chloro analogues 20 was more efficacious than
18. The previously reported5 full agonism of 20 was
consistent with the above patterns, since although this
compound is a 2-chloro-(N)-methanocarba derivative, it
also contained the 5′-N-methylamide group.
F igu r e 6. The superimposition of the unoccupied human A3-
AR model and the complex after binding of full agonist 12.
The orientation of the side chain of W243 (6.48) within the
A3AR binding pocket shifted dramatically upon agonist binding
(final position circled). The indole rotation is counterclockwise
as viewed from the exofacial side. The colors of amino acid
side chains were: yellow in TM3 (H95), cyan in TM5 (F182),
blue in TM6 (F239, W243), purple in TM7 (H272), and
magenta in EL2 (Q167, F168). Upon binding of 12 the
orientation (CCO-CR-Câ-Cγ dihedral angle) of the W243 side
chain shifts from 156.5° to -59.3°. The corresponding angles
in Figure 5B,C are 146.5° and 148.1°, respectively.
chain of W (Figure 5B,C), while the model with full
agonist 12 (Figure 5A) could not optimize until the W
side chain was moved within the binding pocket. The
interaction between the W side chain and the 5′-alkyl
group was sterically unfavorable. The comparison of the
positions of side chains surrounding the putative bind-
ing site of 12 before versus after docking shows that only
W243 underwent substantial movement (Figure 6). This
movement, which was required for binding of the two
above-mentioned agonist nucleosides, may be one of the
components in the overall conformational change of ARs
induced by agonist binding. In the absence of this side
chain, i.e., the W243A mutant receptor, such agonist
such as 12 bound but failed to activate the A3AR.29
Thus, at least within the limited set of adenosine
analogues examined (i.e., N6-3-iodobenzyl analogues and
5′-uronamide analogues) this model distinguished the
structural bases for binding and activation processes.
Partial agonists of ARs are sought for potential
therapeutic use, and a number of strategies have been
pursued to develop them. A partial agonist of the A1
receptor may be useful to slow AV nodal conduction in
the heart and thereby ventricular rate without causing
AV block, bradycardia, atrial arrhythmias, or vasodi-
lation.36 8-Alkyl substitutions of 4 reduced both affinity
and efficacy at the A1AR.37 IJ zerman and co-workers
reported that 5′-modification in combination with N6-
benzyl substituents known to increase A3AR affinity
resulted in the first partial agonists with high affinity
at the human A3AR.8,38 In general, the 2′-and 3′-hydroxy
groups of adenosine and its derivatives are important
for high affinity binding to and activation of ARs.2,5,11,26
Modification of the ribose moiety of adenosine resulted
in a number of A1-selective compounds with reduced A1-
AR agonist activity. Removal of the 2′- or 3′-hydroxyl
group of full agonists resulted in a drastic reduction of
both affinity and efficacy.25 A similar result, i.e., reduc-
tion of efficacy, that was not overcome by the 5′-
uronamide modification, was seen at the A3AR with the
epoxide derivatives 13 and 21. The ribose-like moieties
of these epoxide derivatives lost the “Northern” confor-
mational character, which would be expected to reduce
affinity. Also, formation of this epoxide group precluded
the ability for hydrogen bond donation to the receptor
as seen in A3AR docking of agonists.
Discu ssion
We have demonstrated that certain substitutions of
adenosine and adenosine analogues contributed differ-
ently to binding and function of A3ARs. Specifically, a
2-chloro group in certain cases lowered efficacy. This
effect required the presence of an N6-substitution, such
as 3-iodobenzyl or cyclopentyl.9 The N6-unsubstituted
2-Cl analogue 3, however, was a full agonist.
The N6-(3-iodobenzyl) group itself may lead to reduc-
tion of efficacy, even in the absence of multiple substitu-
tions of adenosine. Thus, 4 was a low-efficacy, partial
agonist. The combination of 2-chloro and N6-substitution
appeared to reduce efficacy further than each modifica-
tion alone. Thus, 5 was demonstrated to be an antago-
nist with little if any detectable agonist activity. The
Recently, Hutchinson et al.39 described that 2-iodo-
N6-(S-endo-norborn-2-yl)adenosine, a potent, full agonist
at the A1AR, had no detectable agonist activity at the
A2AAR, despite weak, but significant binding affinity.
Similarly for the A3AR, the finding that 7 was an

4480 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20
Gao et al.
antagonist,9 together with the results of the present
study indicated that potent antagonists for both human
and rat A3ARs could be identified by modification of the
endogenous AR agonist, adenosine.
ground-state structures.43 Agonist binding disrupts the
intramolecular contact networks, replacing them with
a new set that incorporates the ligand itself. The
experimental data and crystal structure of rhodopsin
suggested that activation by light-induced conforma-
tional changes, including the separation of TMs 3 and
6 and increased exposure of the inner faces of TMs 2, 3,
6, and 7, and a decrease in exposure near the ends of
TMs 4 and 5.44 Specifically, activation of rhodopsin has
been predicted to cause a clockwise rotation viewed from
the cytoplasmic surface of the endofacial section of TM6,
accompanied by a significant outward movement rela-
tive to TM3.45 Although mutation of the conserved
residues in various 7TM GPCRs (G protein-coupled
receptors) had different effects on receptor function, the
conserved W6.48 was typically surrounded by hydro-
phobic residues from TM3, 6, and 7, as was observed
for the thyrotropin releasing hormone (TRH) receptor.46
This arrangement suggested the formation of an in-
tramolecular TM network to constrain the A3AR, as for
the TRH receptor, in an inactive conformation. We
speculate that the conserved P6.50 above W6.48 may
act as a flexible hinge, and that activation leads to a
straightening of TM6, as proposed for rhodopsin.47
In previous studies,11,19 we unsuccessfully sought to
convert the agonist 11 into an A3AR antagonist by
truncating the ribose moiety. In fact, the present study
accomplished that goal with the agonist adenosine,
using a more easily attainable, structural modification
(2-Cl) and also modifications requiring extensive syn-
thetic schemes, such as the spiro modification of 14. The
antagonism by 14 appears to be related to the rigidity
of the spiro-lactam, rather than merely the 4′-alkyl
substitution, since a related 4′-methyl analogue19 was
a full A3AR agonist.
Consistent with the present study, it has been dem-
onstrated that 8 is a full agonist for the rat A3AR,4 while
both related xanthine 7-riboside derivatives (the 3′-
deoxy and 5′-CH2OH analogues) were reported to be
partial agonists for the rat A3AR. However, a 3′-
deoxyadenosine derivative (2-chloro-N6-(3-iodobenzyl)-
5′-N-methylcarbamoyl-3′-deoxyadenosine) was still a
full agonist for the A3AR with no affinity change.11
Previously, it was found that 2-fluoroadenosine acted
as a partial agonist at the A2AAR.40 Our data were
consistent with and extended that observation to the
A3AR. It is interesting that 2-halo substitution is
generally thought to enhance only the affinity of the
ligands, while little attention has been paid on their
dramatic efficacy-diminishing effects. It remains to be
determined if most of the other adenosine derivatives
currently used are full agonists for one subtype but are
partial, full agonists or antagonists for other AR sub-
types. Also, it should be noted that the same compound
may behave as a nearly full agonist, a partial agonist,
or an antagonist in different tissues. Thus, this clas-
sification will also depend on the biological system in
which the measurement is made.41 Actually, a signifi-
cant number of competitive antagonists of GPCRs
currently used as therapeutic agents were first identi-
fied as partial agonists or evolved in a structural series
from an initially identified partial agonist.42
To gain insight into the distinct structural require-
ments for receptor binding and activation, an A3AR
model containing all domains except the C-terminal
region were constructed by homology from the X-ray
structure of rhodopsin. Ligand docking, taking into
account binding of selected derivatives at mutant A3-
ARs, featured interactions of TM3 (His95) with the
adenine moiety and TMs 6 and 7 with the ribose
5′-region. The conserved Trp243 residue in TM6 con-
tributed to the binding of the classical A3AR antago-
nists29 but not of adenosine-derived agonists and an-
tagonists; however, this side chain when present
displayed a characteristic movement seen exclusively
upon docking of agonists. Thus, this residue distin-
guished components of receptor binding and subsequent
activation of the A3AR and furthermore suggested
different modes of binding for the two classes of A3AR
antagonists. Whether the movement of W243 applies
also to other adenosine agonists, such as 5′-CH2OH
derivatives,9 remains to be examined.
Thus, our model proposes that movement of W243 and
consequently TM6 may be involved in the A3AR activa-
tion process,29 similar to previously postulated move-
ment of GPCRs upon activation.33,35 Flexibility of the
adenosine 5′-position, as well as the 3′-position, which
was diminished in (N)-methanocarba, spiral, and ep-
oxide analogues, and characteristic ribose binding at
TM6 and TM7 appeared to be required for activation of
the A3AR. We speculate that the introduction of rigidity
in the ribose ring-constrained analogues might be
responsible for the reduction of efficacy at the A3AR,
thus transforming agonists into antagonists. Other
analogues lacking ribose ring constraints, i.e., contain-
ing 2-Cl and N6-substitutions, also displayed reduced
efficacy, which could not be explained in terms of
reduced flexibility. A parallel hypothesis for reduced
efficacy is that these particular adenine modifications
increased the binding affinity in a way that conse-
quently made it energetically more difficult to twist
TM6. While A3AR antagonists were reported to display
a high correlation (r2 ) 0.99) between functional potency
and A3AR binding affinity,48 the correlation between
agonistic potency and binding affinity was weaker. A
correlation coefficient (r2) of 0.79 was calculated for the
present set of agonists. While there may be many factors
responsible for this variation, it is possible that in some
cases higher affinity of the complex may limit the ability
to activate the receptor through geometric constraint.
In addition, there was evidence for an electronic com-
ponent of the efficacy-reducing effects of adenine sub-
stitutions, since the electron-withdrawing 2-F substi-
tution40 of 2 produced a partial A3AR agonist.
In conclusion, the present study provided new insights
into the ligand-A3AR interaction, especially different
binding properties of nucleoside agonists and antago-
nists interacting with Trp243 and at the 5′-position with
TM6 and TM7. Furthermore, based on structural con-
siderations we have synthesized and characterized
adenosine-derived antagonists and partial agonists for
A3ARs, including the rat homologue. Thus, while most
In rhodopsin, both H-bonding networks and van der
Waals contacts link the TM helices, stabilizing the

Structural Determinants of A3 Adenosine Receptor Activation
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20 4481
) 4.1 Hz, 1H), 4.78 (b, 2H), 4.67 (m, 1H), 2.93 (d, J ) 4.7 Hz,
3H), 2.19 (s, 3H). FAB-MS: m/z (relative intensity) 649.0
(M+1, 2.2), 651.0 (M+3, 2.6), 391.3 (15), 73.1 (100), 57.1 (30),
43.0 (20).
A3AR antagonists reported are nonadenosine deriva-
tives of low aqueous solubility, we conclude that it is
possible to design potent and more hydrophilic nucleo-
side-based antagonists of both human and rat A3ARs,
such as the highly A3AR selective antagonist 14 or the
less selective, but more potent 5. Selective A3 antago-
nism independent of species has long been sought, and
partial agonism at this receptor may prove to be equally
important, especially in light of the therapeutic interest
in activation of A3 receptors (cancer, cardioprotection,
cerebroprotection).1,2,9,10,62 Additional structural modi-
fication of adenosine may provide further increases in
species-independent potency and selectivity. The exist-
ence of A3AR inverse agonists is yet to be demonstrated,
and should be sought among diverse antagonist struc-
tures, including adenosine derivatives.
Intermediate 23 was dissolved in methanol and treated with
Amberlite IRA-400 (OH-) resin. The product 13 was purified
using preparative TLC as a white amorphous solid. NMR
(CDCl3): d 7.78-7.70 (m, 2H), 7.64 (d, J ) 8.0 Hz, 1H), 7.35
(d, J ) 7.1 Hz, 1H), 7.09 (t, J ) 7.7 Hz, 1H), 6.36 (s, 1H), 6.24
(b, 1H), 6.07 (b, 1H), 4.78 (b, 1H), 4.69 (s, 1H), 4.59 (d, J ) 2.5
Hz, 1H), 4.07 (d, J ) 2.5 Hz, 1H), 2.64 (d, J ) 4.9 Hz, 3H).
FAB-MS: m/z (relative intensity) 527.1 (M+ + 1, 5), 119.0
(100), 85.0 (60). HRMS (FAB) for C19H18ClIN5O2 (M+H): calcd
527.0095, found 527.0091.
Molecu la r Mod elin g. All calculations were performed on
a
Silicon Graphics Octane workstation (300 MHz MIPS
R12000 (IP30) processor). All ligand structures were con-
structed using the “Sketch Molecule” of SYBYL 6.7.1.49
A
random search for compounds 5, 12, and 14 was performed to
obtain thermally stable conformations. The options of random
search for all rotatable bonds were 3,000 iteration, 3 kcal
energy cutoffs, and chirality checking. In all cases, MMFF force
field50 and charge were applied using distance-dependent
dielectric constants and conjugate gradient method until the
gradient reached to 0.05 kcal/mol/Å. After clustering the low
energy conformers from the result of the conformational
search, the representative ones from all groups were re-
optimized by semiempirical molecular orbital calculations
using the PM3 method in the MOPAC 6.0 package.51
A human A3AR model was built using homology modeling
from the recently published X-ray structure of bovine rhodop-
sin31 as we previously described.32 Multiple-sequence align-
ment data of selected G protein-coupled receptors were used
for the construction of human A3AR TM domains.52 For the
model of the second extracellular loop, EL2, two beta-sheet
domains in rhodopsin were first aligned including the disulfide
bond between Cys83 and Cys166 and then other parts were
added or deleted. To construct the N-terminal region, intra
and extracellular loops, each alignment was manually adjusted
preserving the overall shape of loop. Acetyl and N-methyl
amide groups blocked the N-terminal and C-terminal region,
respectively. All helices with backbone constraints and loops
were minimized separately. After combining together, all
structures were reminimized initially with backbone con-
straints in the secondary structure and then without any
constraints. The Amber all-atom force field53 and the conjugate
gradient method with a fixed dielectric constant of 4.0 were
used for all calculations, terminating when the gradient
reached 0.05 kcal/mol/Å.
Exp er im en ta l P r oced u r es
Ch em ica l Syn th esis. Compound 12 was purchased from
Sigma (St. Louis, MO), and 18 was prepared as reported.5,6
Other chemicals were from Aldrich (Milwaukee, WI) and of
analytical grade. 1H NMR spectra were recorded using a
Varian Gemini-300 spectrometer. High-resolution FAB (fast
atom bombardment) mass spectra were taken with a J EOL
SX102 spectrometer using nitrobenzoic acid as matrix.
Purity of compounds 13 (90%), 14 (99%), and 21 (93%) was
determined using a Hewlett-Packard 1090 HPLC apparatus
(detector at 260 nm) equipped with an SMT OD-5-60 RP-
C18 analytical column (250 × 4.6 mm; Separation Methods
Technologies, Inc., Newark, DE) in two linear gradient solvent
systems: System A was 0.1 M triethylammonium acetate/CH3-
CN from 95/5 to 40/60 in 20 min and the flow rate was of 1
mL/min; system B was 5 mM tetrabutylammonium phosphate/
CH3CN from 80/20 to 40/60 in 20 min and the flow rate was 1
mL/min.
{5-[2-Ch lor o-6-(3-iod o-ben zyla m in o)-p u r in -9-yl]-3-oxa -
tr icyclo[4.1.0.02,4]h ep t-1-yl}-m eth a n ol (21). A suspension
of 18 (100 mg, 0.189 mmol) in acetonitrile (4 mL) was treated
with acetonitrile/water (100:1, 0.4 mL) and R-acetoxyisobutyryl
bromide (0.11 mL, 0.758 mmol), and the resulting mixture was
stirred at room temperature for 1 h. A clear solution resulted
after 45 min, and a white precipitate was formed. Aqueous
sodium bicarbonate was added and the mixture was extracted
with ethyl acetate. Treatment of the crude mixture from the
combined organic phase with Amberlite IRA-400 (OH-) resin
in dry methanol (4 mL) gave 40.6 mg of 2′,3′-anhydroadenosine
derivative 21, in 42% yield after purification on silica gel
column chromatography(CHCl3/MeOH ) 20:1).
For the conformational refinement of the A3AR, the opti-
mized structures were then used as the starting point for
subsequent 50 ps Molecular Dynamics (MD), during which the
protein backbone atoms in the secondary structures were
constrained as in the previous step. The options of MD at 300
K with 0.2 ps coupling constant were a time step of 1 fs and a
nonbonded update every 25 fs. The lengths of bonds with
hydrogen atoms were constrained according to the SHAKE
algorithm.54 The average structure from the last 10 ps trajec-
tory of MD was re-minimized with backbone constraints in the
secondary structure and then without all constraints.
1H NMR (CD3OD-d6) δ 0.82 (m, 1H), 1.40 (t, 1H, J ) 4.8
Hz), 1.83 (m, 1H), 3.72 (d, 1H, J ) 2.4 Hz, H-3′), 3.79 (ps d,
1H, J ) 11.7 Hz, H-5′), 3.93 (d, 1H, J ) 2.4 Hz, H-2′), 4.06 (ps
d,1H, J ) 11.7 Hz, H-5′), 4.72 (s, 1H, H-1′), 4.78 (s, 2H,
NHCH2-), 7.10 (t, 1H, J ) 7.8 Hz, 5”-H), 7.41 (d, 1H, J ) 7.5
Hz, 4” or 6”-H), 7.62 (d, 1H, J ) 7.5 Hz, 4” or 6”-H), 7.79, (s,
1H, 2”-H), 8.27 (s, 1H, H-8). MS (FAB) m/z ) 510 (M+ + 1).
HRMS (FAB) for C18H17ClIN6O3 (M+H): calcd 510.0194, found
510.0185.
4-[2-Ch lor o-6-(3-iod o-ben zyla m in o)-p u r in -9-yl]-3,6-d i-
oxa -bicyclo[3.1.0]h exa n e-2-ca r boxylic a cid m eth yla m id e
(13). Compound 12 (4.6 mg, 0.008 mmol) was suspended in
acetonitrile (0.2 mL) and treated with 5 µL of R-acetoxyisobu-
tyryl bromide. After the sample was stirred for 1 h at room
temperature, the mixture was treated with 5% of sodium
bicarbonate (5 mL), and stirring was continued for 1 min. The
solution was extracted with ethyl acetate and the organic layer
was washed once with water and dried over magnesium
sulfate. After evaporation of the solvent, the crude intermedi-
ate 23 was purified on a preparative TLC plate (silica, 2000µ
thickness, eluting with methanol/chloroform in a ratio of 1:
4). NMR (CDCl3): d 8.24 (br, s, 1H), 7.75 (s, 1H), 7.64 (d, J )
8.0 Hz, 1H), 7.35 (d, J ) 7.7 Hz, 1H), 7.09 (t, J ) 7.7 Hz, 1H),
6.34 (b, 1H), 6.20 (d, J ) 2.2 Hz, 1H), 5.69 (m, 1H), 4.88 (d, J
For the accuracy of 3D A3AR model, the distribution of the
main chain torsion angles æ and ψ was examined in a
Ramachandran plot, and all ω angles for the peptide planarity
were measured. The chirality of all CR atoms, which in
naturally occurring amino acids is of the L-configuration, was
checked. RMS deviations between backbone atoms in all
helices were compared to the X-ray structure of rhodopsin as
a template.
Flexible docking was facilitated through the “FlexiDock”
utility in the Biopolymer module of SYBYL 6.7.1. During
flexible docking, only the ligand was defined with rotatable
bonds. After the hydrogens were added to the receptor, atomic
charges were assigned using Kollman All-atom for the protein
and calculated using Gasteiger-Hu¨ckel for the ligand. Hydro-

4482 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20
Gao et al.
gen bonding sites were marked for all residues in the active
site and for ligands that were able to act as hydrogen bond
donor or acceptor. Ligands were prepositioned in the cavity
as a starting point for FlexiDock. Default FlexiDock param-
eters were set at 3000-generation for genetic algorithms. To
increase the binding interaction, the torsion angles of the side
chains that directly interacted within 5 Å of the ligands
according to the results of FlexiDock were manually adjusted.
Finally, the complex structure was minimized using an Amber
force field with fixed dielectric constant (4.0), until the
conjugate gradient reached 0.1 kcal/mol/Å.
P h a r m a cologica l Meth od s. [125I]N6-(4-amino-3-iodoben-
zyl)adenosine-5′-N-methyluronamide (I-AB-MECA; 2000 Ci/
mmol), [3H]8-ethyl-4-methyl-2-phenyl-(8R)-4,5,7,8-tetrahydro-
1H-imidazo[2,1-i]-purin-5-one (PSB-11, 53 Ci/mmol) and
[3H]cyclic AMP (40 Ci/mmol) were from Amersham Pharmacia
Biotech (Buckinghamshire, UK).
removed and cells were washed three times with 1 mL of
DMEM, containing 50 mM HEPES, pH 7.4. Cells were then
treated with agonists and/or test compounds in the presence
of rolipram (10 µM) and adenosine deaminase (3 units/mL).
After 45 min, forskolin (10 µM) was added to the medium, and
incubation was continued an additional 15 min. The reaction
was terminated by removing the supernatant, and cells were
lysed upon the addition of 200 µL of 0.1 M ice-cold HCl. The
cell lysate was resuspended and stored at -20 °C. For
determination of cyclic AMP production, protein kinase A
(PKA) was incubated with [3H]cyclic AMP (2 nM) in K2HPO4/
EDTA buffer (K2HPO4, 150 mM; EDTA, 10 mM), 20 µL of the
cell lysate, and 30 µL 0.1 M HCl or 50 µL of cyclic AMP solution
(0-16 pmol/200 µL for standard curve). Bound radioactivity
was separated by rapid filtration through Whatman GF/C
filters and washed once with cold buffer. Bound radioactivity
was measured by liquid scintillation spectrometry.
Effects of a GTP An a logu e on th e Com p etition of
Tr itia ted An ta gon ist Bin d in g to th e Hu m a n A3AR by
Ad en osin e Der iva tives. The procedure used has been de-
scribed previously.23,58 Briefly, [3H]8-ethyl-4-methyl-2-phenyl-
(8R)-4,5,7,8-tetrahydro-1H-imidazo[2,1-i]-purin-5-one (5 nM)
was incubated membranes (80 µg of protein) from CHO cells
expressing the human A3AR in the absence and presence of
GTPγS (100 µM) and increasing concentrations of test com-
pounds in a total assay volume of 400 µL at 25 °C for 60 min.
Bound radioactivity was separated by rapid filtration through
Whatman GF/C filters and washed once with cold buffer.
Sta tistica l An a lysis. Binding and functional parameters
were estimated using GraphPAD Prism software (GraphPAD,
San Diego, CA). IC50 values obtained from competition curves
were converted to Ki values using the Cheng-Prusoff equa-
tion.59 Data were expressed as mean ( standard error.
Cell Cu ltu r e a n d Mem br a n e P r ep a r a tion . CHO (Chi-
nese hamster ovary) cells expressing recombinant the human
A3AR were cultured in DMEM supplemented with 10% fetal
bovine serum, 100 units/mL penicillin, 100 µg/mL streptomy-
cin, 2 µmol/mL glutamine, and 800 µg/mL Geneticin. The CHO
cells expressing rat A3ARs were cultured in DMEM and F12
(1:1). Cells were harvested by trypsinization. After homogeni-
zation and suspension, cells were centrifuged at 500 g for 10
min, and the pellet was resuspended in 50 mM Tris‚HCl buffer
(pH 8.0) containing 10 mM MgCl2, 1 mM EDTA and 0.1 mg/
mL CHAPS. The suspension was homogenized with an electric
homogenizer for 10 s, and was then recentrifuged at 20000g
for 20 min at 4 °C. The resultant pellets were resuspended in
buffer in the presence of 3 Units/mL adenosine deaminase,
and the suspension was stored at -80 °C until the binding
experiments. Striatal and forebrain tissues from Wistar rats
were homogenized in ice-cold 50 mM Tris‚HCl buffer, pH 7.4,
using an electric homogenizer. The homogenate was centri-
fuged at 20000g for 10 min at 4 °C, and the pellet was washed
in fresh buffer. The final pellet was stored at -80 °C until the
binding experiments. The protein concentration was measured
using the Bradford assay.60
Ack n ow led gm en t. We thank Dr. Bruce Liang (Uni-
versity of Connecticut) and Dr. J ohn Daly (NIDDK) for
helpful discussions, Dr. Gary Stiles (Duke University)
for the gifts of CHO cells expressing the human and rat
A3ARs, and Prof. Christa Mu¨ller (University of Bonn,
Germany) for providing [3H]8-ethyl-4-methyl-2-phenyl-
(8R)-4,5,7,8-tetrahydro-1H-imidazo[2,1-i]-purin-5-one.
Z.-G.G. thanks Gilead Sciences (Foster City, CA) for
financial support. We thank Dr. Neli Melman for
technical assistance.
Site-Dir ected Mu ta gen esis. The method used was de-
scribed previously.29 The protocols used were as described in
the QuikChange Site-Directed Mutagenesis kit (La J olla, CA).
Mutations were confirmed by DNA sequencing.
Tr a n sien t Exp r ession of Wild Typ e (WT) a n d Mu ta n t
A3ARs in COS-7 Cells. COS-7 cells (10-6) were grown in 100-
mm cell culture dishes containing Dulbecco’s modified Eagle’s
medium supplemented with 10% fetal bovine serum, 100 units/
mL penicillin, 100 µg/mL streptomycin, and 2 µmol/mL
glutamine. After 24 h, cells were washed with phosphate-
buffered saline (with calcium) and then transfected with
plasmid DNA (10 µg/dish) using the DEAE-dextran method55
for 1 h. The cells were then treated with 100 µM chloroquine
for 3 h in culture medium and cultured for an additional 48 h
at 37 °C and 5% CO2.
Su p p or tin g In for m a tion Ava ila ble: Synthetic methods
for preparation of compounds 25-35 and 14. This material is
available free of charge via the Internet at http://pubs.acs.org.
Refer en ces
(1) J acobson, K. A.; Knutsen, L. J . S. P1 and P2 purine and
pyrimidine receptors. In Purinergic and Pyrimidinergic Signal-
ing I; Handbook of Experimental Pharmacology, Vol. 151/I;
Abbracchio, M. P., Williams, M., Eds.; Springer-Verlag: Berlin,
Germany, 2001; pp 129-175.
Bin d in g Assa y u sin g [125I]4-a m in o-3-iod oben zyl)a d -
en osin e-5′-N-m eth ylu r on a m id e. For competitive binding
assay, each tube contained 50 µL of membrane suspension (20
µg of protein), 25 µL of [125I]N6-(4-amino-3-iodobenzyl)adeno-
sine-5′-N-methyluronamide (1.0 nM), and 25 µL of increasing
concentrations of the test ligands in Tris‚HCl buffer (50 mM,
pH 8.0) containing 10 mM MgCl2, 1 mM EDTA. Nonspecific
binding was determined using 10 µM of 12 in the buffer. The
mixtures were incubated at 37 °C for 60 min. Binding reactions
were terminated by filtration through Whatman GF/B filters
under reduced pressure using a MT-24 cell harvester (Bran-
dell, Gaithersburgh, MD). Filters were washed three times
with 9 mL of ice-cold buffer. Radioactivity was determined in
a Beckman 5500B γ-counter.
Cyclic AMP Accu m u la tion Assa y. Intracellular cyclic
AMP levels were measured with a competitive protein binding
method.56,57 CHO cell that expressed recombinant human and
rat A3ARs were harvested by trypsinization. After centrifuga-
tion and resupension in medium, cells were planted in 24-well
plates in 1.0 mL of medium. After 24 h, the medium was
(2) Mu¨ller, C. E. Adenosine receptor ligands-recent developments
part I. Agonists. Curr. Med. Chem. 2000, 7, 1269-1288.
(3) Daly, J . W.; Padgett, W.; Thompson, R. D.; Kusachi, S.; Bugni,
W. J .; Olsson, R. A. Structure-activity relationships for N6-
substituted adenosines at a brain A1-adenosine receptor with a
comparison to an A2-adenosine receptor regulating coronary
blood flow. Biochem. Pharmacol. 1986, 35, 2467-2481.
(4) Kim, H. O.; J i, X.-d.; Siddiqi, S. M.; Olah, M. E.; Stiles, G. L.;
J acobson, K. A. 2-Substitution of N6-benzyladenosine-5′-urona-
mides enhances selectivity for A3-adenosine receptors. J . Med.
Chem. 1994, 37, 3614-3621.
(5) Lee, K.; Ravi, R. G.; J i, X.-d.; Marquez, V. E.; J acobson, K. A.
Ring-constrained (N)methanocarba-nucleosides as adenosine
receptor agonists: Independent 5′-uronamide and 2′-deoxy
modifications. Biorg. Med. Chem. Lett. 2001, 11, 1333-1337.
(6) J acobson, K. A.; J i, X-d.; Li, A.-H.; Melman, N.; Siddiqui, M.
A.; Shin, K.-J .; Marquez, V. E.; Ravi, R. G. Methanocarba
Analogues of Purine Nucleosides as Potent and Selective Ad-
enosine Receptor Agonists. J . Med. Chem. 2000, 43, 2196-2203.
(7) van der Wenden, E. M.; Carnielli, M.; Roelen, H. C. P. F.;
Lorenzen, A.; von Frijtag Drabbe Ku¨nzel, J . K.; IJ zerman, A. P.
5′-substituted adenosine analogues as new high-affinity partial
agonists for the adenosine A1 receptor. J . Med. Chem. 1998, 41,
102-108.

Structural Determinants of A3 Adenosine Receptor Activation
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20 4483
(8) van Tilburg, E. W.; von Frijtag Drabbe Ku¨nzel, J .; de Groote,
M.; Vollinga, R. C.; Lorenzen, A.; IJ zerman, A. P. N6,5′-
Disubstituted adenosine derivatives as partial agonists for the
human adenosine A3 receptor. J . Med. Chem. 1999, 42, 1393-
1400.
(30) Varani, K.; Merighi, S.; Gessi, S.; Klotz, K. N.; Leung, E.;
Baraldi,.G.; Cacciari, B.; Romagnoli, R.; Spalluto, G.; Borea, P.
A. [
3H]MRE 3008F20: a novel antagonist radioligand for the
pharmacological and biochemical characterization of human A3
adenosine receptors. Mol. Pharmacol. 2000, 57, 968-975.
(31) Palczewski, K.; Kumasaka, T.; Hori, T.; Behnke, C. A.; Mo-
toshima, H.; Fox, B. A.; Le Trong, I.; Teller, D. C.; Okada, T.;
Stenkamp, T. E.; Yamamoto, M.; Miyano, M. Crystal Structure
of Rhodopsin: A G Protein-Coupled Receptor. Science 2000, 289,
739-745.
(32) J acobson, K. A.; Gao, Z. G.; Chen, A.; Barak, D.; Kim, S. A.; Lee,
K.; Link, A.; van Rompaey, P.; van Calenbergh, S.; Liang, B. T.
Neoceptor Concept Based on Molecular Complementarity in
GPCRs: A Mutant Adenosine A3 Receptor with Selectively
Enhanced Affinity for Amine-Modified Nucleosides. J . Med.
Chem. 2001, 44, 4125-4136.
(33) Kim, J .; Wess, J .; van Rhee, A. M.; Scho¨neberg, T.; J acobson, K.
Site-directed Mutagenesis Identifies Residues Involved in Ligand
Recognition in the Human A2a Adenosine Receptor. J . Biol.
Chem. 1995, 270, 13987-13997.
(34) Kim, J .; J iang, Q.; Glashofer, M.; Yehle, S.; Wess, J .; J acobson,
K. A. Glutamate Residues in the Second Extracellular Loop of
the Human A2a Adenosine Receptor Are Required for Ligand
Recognition. Mol. Pharmacol. 1996, 49, 683-691.
(35) J iang, Q.; van Rhee, M.; Kim. J .; Yehle, S.; Wess, J .; J acobson,
K. A. Hydrophilic Side Chains in the Third and Seventh
Transmembrane Helical Domains of Human A2A Adenosine
Receptors Are Required for Ligand Recognition. Mol. Pharmacol.
1996, 50, 512-521.
(36) Wu, L.; Belardinelli, L.; Zablocki, J . A.; Palle, V.; Shryock, J . C.
A partial agonist of the A1-adenosine receptor selectively slows
AV conduction in guinea pig hearts. Am. J . Physiol. Heart Circ.
Physiol. 2001, 280, H334-H343.
(37) Roelen, H.; Veldman, N.; Spek, A. L.; von Frijtag Drabbe Ku¨nzel,
J .; Mathot, R. A.; IJ zerman A. P. N6,C8-disubstituted adenosine
derivatives as partial agonists for adenosine A1 receptors. J .
Med. Chem. 1996, 39, 1463-1471.
(38) van Tilburg, E. W.; von Frijtag Drabbe Ku¨nzel, J .; de Groote,
M.; IJ zerman, AP. 2,5′-Disubstituted adenosine derivatives:
(9) Gao, Z. G.; J acobson, K. A, 2-Chloro-N6-cyclopentyladenosine,
adenosine A1 receptor agonist, antagonizes the adenosine A3
receptor, Eur. J . Pharmacol. 2002, 443, 39-42.
(10) Baraldi, P. G.; Cacciari, B.; Romagnoli, R.; Merighi, S.; Varani,
K.; Borea, P. A.; Spalluto G. A3 Adenosine receptor ligands;
history and perspectives. Med. Res. Rev. 2000, 20, 103-128.
(11) J acobson, KA.; Siddiqi, S. M.; Olah, M. E.; J i, X-d.; Melman,
N.; Bellamkonda, K.; Meshulam, Y.; Stiles, G. L.; Kim, H. O.
Structure-activity relationships of 9-alkyladenine and ribose-
modified adenosine derivatives at rat A3 adenosine receptors.
J . Med. Chem. 1995, 38, 1720-1735.
(12) Park, K. S.; Hoffmann, C.; Kim, H. O.; Padgett, W. L.; Daly, J .
W.; Brambilla, R.; Motta, C.; Abbracchio, M. P.; and J acobson,
K. A. Activation and desensitization of rat A3-adenosine recep-
tors by selective adenosine derivatives and xanthine-7-ribosides.
Drug Dev. Res. 1998, 44, 97-105.
(13) Nandanan, E.; J ang, S. Y.; Moro, S.; Kim, H.; Siddiqi, M. A.;
Russ, P.; Marquez, V. E.; Busson, R.; Herdewijn, P.; Harden, T.
K.; Boyer, J . L.; J acobson, K. A. Synthesis, biological activity,
and molecular modeling of ribose-modified adenosine bisphos-
phate analogues as P2Y1 receptor ligands. J . Med. Chem. 2000,
43, 829-842.
(14) Zimmerman, D. M.; Leander, J . D. Selective opioid receptor
agonists and antagonists: research tools and potential thera-
peutic agents. J . Med. Chem. 1990, 33, 895-902.
(15) Holst, B.; Zoffmann, S.; Elling, CE.; Hjorth, S. A.; Schwartz, T.
W. Steric hindrance mutagenesis versus alanine scan in map-
ping of ligand binding sites in the tachykinin NK1 receptor. Mol.
Pharmacol. 1998, 53, 166-175.
(16) Robins, M. J .; Hansske, F.; Low, N. H.; Park, J . I. A mild
conversion of vicinal diols to alkenes. Efficient transformation
of ribonucleosides into 2′-ene and 2′,3′-dideoxynucleosides. Tet-
rahedron Lett. 1984, 25, 367-370.
(17) Hansske, F.; Robins, M. J . Regiospecific and stereoselective
conversion of ribonucleosides to 3′-deoxynucleosides. A high yield
three-stage synthesis of cordycepin from adenosine. Tetrahedron
Lett. 1985, 26, 4295.
(18) Russell, A. F.; Greenberg, S.; Moffatt, J . G. Reaction of 2-acyl-
oxyisobutyryl halides with nucleosides. II. Reaction of adenosine.
J . Am. Chem. Soc. 1973, 95, 4025.
(19) Siddiqi, S. M.; J acobson, K. A.; Esker, J . L.; Melman, N.; Tiwari,
K. N.; Secrist, J . A.; Schneller, S. W.; Cristalli, G.; J ohnson, C.
A.; IJ zerman, A. P.Search for new purine- and ribose-modified
adenosine analogues as selective agonists and antagonists at
adenosine receptors. J . Med. Chem. 1995, 38, 1174-1188.
(20) J ohnson, C. R.; Esker, J . L.; Van Zandt, M. C. Chemoenzymic
synthesis of 4-substituted riboses. S-(4′-Methyladenosyl)-L-ho-
mocysteine. J . Org. Chem. 1994, 59, 5854-5855.
(21) Vorbru¨ggen H.; Ruh-Pohlenz, C. Synthesis of nucleosides. Org.
React. 2000, 55, 1-630.
(22) J acobson, K. A.; Park, K. S.; J iang, J . L.; Kim, Y.-C.; Olah, M.
E.; Stiles, G. L.; J i, X-d. Pharmacological characterization of
novel A3 adenosine receptor-selective antagonists. Neurophar-
macology 1997, 36, 1157-1165
(23) Gao, Z. G.; Van Muijlwijk-Koezen, J . E.; Chen, A.; Mu¨ller, C.
E.; IJ zerman, A. P.; J acobson, K. A. Allosteric modulation of A3
adenosine receptors by a series of 3-(2-pyridinyl)isoquinoline
derivatives. Mol. Pharmacol. 2001, 60, 1057-1063.
(24) Arunlakshana, O.; Schild, H. O. Some quantitative uses of drug
antagonists. Br. J . Pharmacol. Chemother. 1959, 14, 48-58.
(25) van der Wenden, E. M.; von Frijtag Drabbe Ku¨nzel, J . K.;
Mathot, R. A.; Danhof, M.; IJ zerman, A. P.; Soudijn, W. Ribose-
modified adenosine analogues as potential partial agonists for
the adenosine receptor. J . Med. Chem. 1995, 38, 4000-4006.
(26) Lohse, M. J .; Klotz, K. N.; Diekmann, E.; Friedrich, K.; Schwabe,
U.2′, 3′-Dideoxy-N6-cyclohexyladenosine: an adenosine deriva-
tive with antagonist properties at adenosine receptors. Eur. J .
Pharmacol. 1988, 156, 157-160.
evaluation of selectivity and efficacy for the adenosine A1, A2A
,
and A3 receptor. J . Med. Chem. 2002, 45, 420-429.
(39) Hutchinson, S. A.; Baker, S. P.; Scammells, P. J . New 2,N6-
Disubstituted adenosines: potent and selective A1 adenosine
receptor agonists. Bioorg. Med. Chem. 2002, 10, 1115-1122.
(40) Daly, J . W.; Padgett, W. L. Agonist activity of 2- and 5′-
substituted adenosine analogues and their N6-cycloalkyl deriva-
tives at A1- and A2-adenosine receptors coupled to adenylate
cyclase. Biochem. Pharmacol. 1992, 43, 1089-1093.
(41) Kenakin, T. P. The Pharmacologic Analysis of Drug-Receptor
Interactions; Lippincott-Raven, Philadelphia, New York, 1997.
(42) Harden, T. K.; Boyer, J . L.; Dougherty, R. W. Drug analysis
based on signaling responses to G-protein-coupled receptors. J .
Recept. Signal Transduct. Res. 2001, 2, 167-190.
(43) Teller, D. C.; Okada, T.; Behnke, C. A.; Palczewski, K.; Sten-
kamp, R. E. Advances in determination of a high-resolution
three-dimensional structure of rhodopsin, a model of G-protein-
coupled receptors (GPCRs). Biochemistry 2001, 40, 7761-7772.
(44) Meng, E. C.; Bourne, H. R. Receptor activation: what does the
rhodopsin structure tell us? Trends Pharmacol. Sci. 2001, 22,
587-593.
(45) Lu, Z.-L.; Saldanha, J . W.; Hulme, E. C. Seven-transmembrane
receptors: crystals clarify. Trends Pharmacol. Sci. 2002, 23,
140-146.
(46) Colson, A.-O.; Perlman, J . H.; J insi-Parimoo, A.; Nussenzveig,
D. R.; Osman, R.; Gershengorn, M. C. A hydrophobic cluster
between transmembrane helices 5 and 6 constrains the thy-
rotropin-releasing hormone receptor in an inactive conformation.
Mol. Pharmacol. 1998, 54, 968-978.
(47) Ballesteros, J . A.; Shi, L.; J avitch, J . A. Structural mimicry in
G protein-coupled receptors: Implications of the high-resolution
structure of rhodopsin for structure-function analysis of rhodop-
sin-like receptors. Mol. Pharmacol. 2001, 60, 1-19.
(48) Baraldi, P. G.; Cacciari, B.; Moro, S.; Spalluto, G.; Pastorin, G.;
Da Ros, T.; Klotz, K.-N.; Varani, K.; Gessi, S.; Borea, P. A.
Synthesis, biological activity, and molecular modeling investiga-
tion of new pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine deriva-
tives as human A3 adenosine receptor antagonists. J . Med.
Chem. 2002, 45, 770-780.
(27) Lorenzen, A.; Guerra, L.; Vogt, H.; Schwabe, U. Interaction of
full and partial agonists of the A1 adenosine receptor with
receptor/G protein complexes in rat brain membranes. Mol.
Pharmacol. 1996, 49, 915-926.
(28) Mu¨ller, C. E.; Diekmann, M.; Thorand, M.; Ozola, V. [3H]8-Ethyl-
4-methyl-2-phenyl-(8R)-4,5,7,8-tetrahydro-1H-imidazo[2,1-i]-pu-
rin-5-one ([3H]PSB-11), a Novel High-Affinity Antagonist Radi-
oligand for Human A3 Adenosine Receptors. Bioorg. Med. Chem.
Lett. 2002, 12, 501-503.
(49) Sybyl Molecular Modeling System, version 6.7.1; Tripos Inc.,
1969 South Hanley Rd., St. Louis, MO 63144, USA.
(50) Halgren, T. A.; MMFF VII. Characterization of MMFF94,
MMFF94s, and other widely available force fields for conforma-
tional energies and for intermolecular-interaction energies and
geometries. J . Comput. Chem. 1999, 20, 730-748.
(51) Stewart, J . J . P. MOPAC: A semiempirical molecular orbital
program. J . Comput.-Aided Mol. Des. 1990, 4, 1-105.
(52) van Rhee, A. M.; J acobson, K. A. Molecular architecture of G
protein-coupled receptors. Drug Dev. Res. 1996, 37, 1-38.
(29) Gao, Z. G.; Chen, A.; Barak, D.; Kim, S. K.; Mu¨ller, C. E.;
J acobson,K. A. Identification by site-directed mutagenesis of
residues involved in ligand recognition and activation of the
human A3 adenosine receptor. J . Biol. Chem. 2002, 277, 19056-
19063.

4484 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 20
Gao et al.
(53) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K.
M., J r.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell,
J . W.; Kollman, P. A. A second generation force field for the
simulation of proteins, nucleic acids and organic molecules. J .
Am. Chem. Soc. 1995, 117, 5179-5197.
(54) Ryckaert, J . P.; Ciccotti, G.; Berendsen, H. J . C. Numerical
integration of the Cartesian equations of motion for a system
with constraints: Molecular dynamics of n-alkanes.J . Comput.
Phys. 1977, 23, 327-333.
(55) Cullen, B. R. Use of eukaryotic expression technology in the
functional analysis of cloned genes. Methods Enzymol. 1987, 52,
684-704.
(56) Nordstedt, C.; Fredholm, B. B. A modification of a protein-
binding method for rapid quantification of cAMP in cell-culture
supernatants and body fluid. Anal. Biochem. 1990, 189, 231-
234.
(57) Post SR.; Ostrom RS.; Insel PA. Biochemical methods for
detection and measurement of cyclic AMP and adenylyl cyclase
activity. Methods Mol. Biol. 2000, 126, 363-374.
(58) Gao, Z. G.; J iang, Q.; J acobson, K. A.; IJ zerman, A. P. Site-
directed mutagenesis studies of human A2A adenosine recep-
tors: involvement of Glu13 and His278 in ligand binding and
sodium modulation. Biochem. Pharmacol. 2000, 60, 661-668.
(59) Cheng, Y.-C.; Prusoff, W. H. Relationship between the inhibition
constant (Ki) and the concentration of inhibitor which causes
50% inhibition (I50) of an enzymatic reaction. Biochem. Phar-
macol. 1973, 22, 3099-3108.
(60) Bradford, M. M. A rapid and sensitive method for the quanti-
tation of microgram quantities of protein utilizing the principle
of protein-dye binding. Anal. Biochem. 1976, 72, 248-254.
(61) J iang, J . L.; Li, A.-H.; J ang, S. Y.; Chang, L.; Melman, N.; Moro,
S.; J i, X.-d.; Lobkovsky, E. B.; Clardy, J . C.; J acobson, K. A.
Chiral resolution and stereospecificity of 4-phenylethynyl-6-
phenyl-1,4-dihydropyridine derivatives as selective A3 adenosine
receptor antagonists. J . Med. Chem. 1999, 42, 3055-3065.
(62) Fishman, P.; Madi, L.; Bar-Yehuda, S.; Barer, F.; Del Valle, L.;
Khalili, K. Evidence for involvement of Wnt signaling pathway
in IB-MECA mediated suppression of melanoma cells. Oncogene
2002, 21, 4060-4064.
J M020211+