Design and Synthesis of Novel Dual-Action Compounds Targeting the Adenosine A2A Receptor and Adenosine Transporter for Neuroprotection

Article abstract of DOI:10.1002/cmdc.201100126

A novel compound, N⁶-(4-hydroxybenzyl)adenosine, isolated from Gastrodia elata and which has been shown to be a potential therapeutic agent for preventing and treating neurodegenerative disease, was found to target both the adenosine A_2A

Full text of DOI:10.1002/cmdc.201100126

MED
DOI: 10.1002/cmdc.201100126
Design and Synthesis of Novel Dual-Action Compounds
Targeting the Adenosine A2A Receptor and Adenosine
Transporter for Neuroprotection
Jhih-Bin Chen,[a] Eric Minwei Liu,[b] Ting-Rong Chern,[b] Chieh-Wen Yang,[a] Chia-I Lin,[a] Nai-
Kuei Huang,[e] Yun-Lian Lin,[e] Yijuang Chern,[d] Jung-Hsin Lin,*[b, c, d] and Jim-Min Fang*[a]
A novel compound, N6-(4-hydroxybenzyl)adenosine, isolated
from Gastrodia elata and which has been shown to be a po-
tential therapeutic agent for preventing and treating neurode-
generative disease, was found to target both the adenosine
A2A receptor (A2AR) and the equilibrative nucleoside transport-
er 1 (ENT1). As A2AR and ENT1 are proximal in the synaptic
crevice of striatum, where the mutant huntingtin aggregate is
located, the dual-action compounds that concomitantly target
these two membrane proteins may be beneficial for the thera-
py of Huntington’s disease. To design the desired dual-action
compounds, pharmacophore models of the A2AR agonists and
the ENT1 inhibitors were constructed. Accordingly, potentially
active compounds were designed and synthesized by chemical
modification of adenosine, particularly at the N6 and C5’ posi-
tions, if the predicted activity was within an acceptable range.
Indeed, some of the designed compounds exhibit significant
dual-action properties toward both A2AR and ENT1. Both phar-
macophore models exhibit good statistical correlation between
predicted and measured activities. In agreement with competi-
tive ligand binding assay results, these compounds also pre-
vent apoptosis in serum-deprived PC12 cells, rendering a cru-
cial function in neuroprotection and potential utility in the
treatment of neurodegenerative diseases.
Introduction
Huntington’s disease (HD) is an autosomal dominant neurode-
generative disease caused by a CAG trinucleotide expansion in
the Huntingtin (Htt) gene, which shows chorea, dementia, and
psychiatric symptoms.[1] An effective treatment for HD has not
yet been developed, although a few therapeutic agents with
moderate effects have been reported.[2] It has been demon-
strated that the selective A2A adenosine receptor (A2AR) agonist,
CGS21680, can attenuate HD symptoms in a transgenic mouse
model,[3] and this compound has been shown to be able to
rescue the urea cycle deficiency of HD by enhancing the activi-
ty of the ubiquitin–proteasome system.[4] However, CGS21680
is known to exert strong immunosuppression[5] among other
side effects, and is therefore unsuitable for clinical use. Recent-
ly, an adenosine analogue designated T1-11 (compound 1),
also an A2AR agonist, was isolated from Gastordia elata and
shown to prevent serum-deprived PC12 cell apoptosis, sug-
gesting its therapeutic potential in treating neural degenera-
tive diseases.[6]
preferable to stably binding compounds,[9] especially in target-
ing the A2AR signaling system, given the aforementioned rea-
sons.
Interestingly, compound 1 was found to be an ENT1 inhibi-
tor as well, i.e., a dual-action compound. It is therefore of con-
siderable interest to modify compound 1 to generate more
similar compounds that retain the dual-action property of 1 in
[a] J.-B. Chen,+ C.-W. Yang, C.-I. Lin, Prof. J.-M. Fang
Department of Chemistry, National Taiwan University
No. 1, Sec. 4, Roosevelt Rd., Taipei 106 (Taiwan)
Fax: (+886)2-23637812
E-mail: jmfang@ntu.edu.tw
[b] E. M. Liu,+ T.-R. Chern, Prof. J.-H. Lin
School of Pharmacy, National Taiwan University
No. 1, Sec. 1, Ren-Ai Rd., Taipei 100 (Taiwan)
Fax: (+886)2-27823060
E-mail: jlin@ntu.edu.tw
[c] Prof. J.-H. Lin
Division of Mechanics, Research Center for Applied Sciences
Academia Sinica
It has been recognized that A2AR and the adenosine trans-
porters (such as the equilibrative nucleoside transporter ENT1)
are both localized in the striatum,[7] where mutant Htt aggre-
gates.[8] Inhibition of the adenosine transporter would elevate
local adenosine concentrations, thereby enhancing the proba-
bility of A2AR agonism relative to A2AR agonists alone. On the
other hand, the potent immunosuppressive effect of
CGS21680 is a consequence of A2AR signaling,[5] which indicates
that extremely strong binding affinity toward A2AR may not be
a desirable property for clinically useful therapeutic agents.
Therefore, transient or weakly binding compounds may be
No. 128, Sec. 2, Academia Rd., Taipei 115 (Taiwan)
E-mail: jhlin@gate.sinica.edu.tw
[d] Prof. Y. Chern, Prof. J.-H. Lin
Institute of Biomedical Sciences
Academia Sinica, Nankang, Taipei 115 (Taiwan)
[e] Prof. N.-K. Huang, Y.-L. Lin
Division of Medicinal Chemistry
National Research Institute of Chinese Medicine
No. 155-1, Sec. 2, LiNung St., Taipei 112 (Taiwan)
[+] These authors contributed equally to this work.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201100126.
1390
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1390 – 1400

Dual-Action Neuroprotective Compounds
order to determine whether such analogues exhibit similar
neuroprotection effects. Such designed multiple ligands
(DML)[10] are conceptually distinct from promiscuous com-
pounds discovered by random screening because they are ra-
tionally designed to optimize the desired properties. It has
been proposed that DMLs may be advantageous over single-
target compounds for the treatment of diseases with complex
etiologies such as asthma, obesity, cancer, and psychiatric dis-
eases.[10,11] It is thought that the intrinsic redundancy and ro-
bustness of complex biological networks may be responsible
for the failure of highly selective drugs to deliver the intended
therapeutic effects.[12] A recent assessment indicates that A2AR
pharmacology is indeed rather complex,[13] and therefore DMLs
may be especially suitable for targeting the A2AR signaling
system. However, the challenge in designing multiple-target li-
gands[14] is partly due to difficulty in the appropriate construc-
tion of computational models that describe the interactions of
the ligands with several targets, and is also due in part to the
increased restraints of chemical synthesis and satisfaction of
physicochemical properties of compounds. Perhaps associated
with the increased complexity as the number of targets in-
creases, currently most of the reported DMLs are dual-function
ligands.[14b]
Figure 1. Competitive ligand binding assays were performed to
determine if the designed compounds indeed bind A2AR and
ENT1 with good affinity. Finally, these compounds were as-
sayed for their ability to prevent apoptosis in serum-deprived
PC12 cells, which is a crucial indication of their potential for
treating neurodegenerative diseases.[18]
To rationally design the proper dual-action ligands as thera-
peutics for neurodegenerative diseases such as Huntington’s
disease, we first constructed two pharmacophore models: one
for A2AR agonists, and the other for ENT1 inhibitors. Three-di-
mensional pharmacophores are specific spatial distributions of
chemical functional features of a series of compounds that
target the same active site of a biomolecule and exert the
same function;[15] these are particularly useful if high-resolution
X-ray or NMR structures of the target biomolecules are not yet
available. In the past, pharmacophore analysis has been suc-
cessfully applied to numerous drug discovery tasks.[16] In the
current work, pharmacophore modeling is not used for scaf-
fold hopping, but for the search for dual-action compounds,
and the prediction of binding affinity to individual targets be-
Figure 1. Structures of CGS21680, NBTI, and designed adenosine derivatives
with modification at the N6- and C5’-positions with acceptable predicted ac-
tivities from pharmacophore mapping.
comes crucial for this. The statistical performance of three-di- Results and Discussion
mensional quantitative structure–activity relationship (3D-
Pharmacophore model of human adenosine A2AR agonists
QSAR) models derived from the pharmacophore analysis is
therefore of great importance. The outliers of the QSAR
models need to be removed in order to obtain models of
better statistical quality. Although the high-resolution crystal
structure of human adenosine A2AR bound with the agonist
UK-432097 was recently reported,[17] it is not clear whether this
structure with partial attenuation at the outward tilt of helix VI
due to the fused T4 lysozyme is suitable for constructing QSAR
models of general A2AR agonists. On the other hand, the crystal
structure of human ENT1 (hENT1) is not yet available, and
there is also no suitable structural template for homology
modeling.
As part of the dual-pharmacophore drug design approach, a
3D pharmacophore model of human A2AR (hA2AR) agonists was
first constructed to aid in the design of compounds that could
function as hA2AR agonists. The final training set, after iterative
removal of outliers in the 3D-QSAR model, includes 25 com-
pounds spanning a wide range of structural diversity and
hA2AR activity (Ki values from 1.2 nm to 187 mm) selected from
published sources (see Supporting Information). A potent
hA2AR agonist, CGS21680,[19] is also included in this training set.
The HypoGen module of the Catalyst software package by Ac-
celrys[20] was used to construct the pharmacophore model of
these ligands. The constructed pharmacophore is illustrated in
Figure 2A, which shows four geometric features including: hy-
drophobic (HP, cyan), ring aromatic (RA, gold), hydrogen bond
donor (HBD, magenta), and hydrogen bond acceptor (HBA,
green). It is not surprising to see that for CGS21680 all the four
features of the constructed pharmacophore can be fitted
Based on the scaffold of compound 1, we set out to design
a series of adenosine derivatives. Chemical modifications of ad-
enosine were carried out if the pharmacophore mapping of
the designed compound predicted acceptable activity (predict-
ed Ki <10 mm) for both pharmacophore models. The com-
pounds with acceptable predicted activities are shown in
ChemMedChem 2011, 6, 1390 – 1400
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1391

J.-H. Lin, J.-M. Fang, et al.
MED
Figure 2. The 3D pharmacophore model of adenosine A2A receptor agonists.
A) Geometric relationships of the pharmacophore features. The pharmaco-
phore features are represented by meshed spheres with the following color
code: cyan=hydrophobic (HP), gold=ring aromatic (RA), magenta=hydro-
gen bond donor (HBD), green=hydrogen bond acceptor (HBA). The dis-
tance between the centers of the HP sphere and the smaller sphere of HBA:
d
(HP–HBA) =6.767 ꢁ; with the same notation: d(RA–HBA) =5.789 ꢁ, d(RA–
HP) =6.844 ꢁ, d(HP–HBD) =8.873 ꢁ, d(HBA–HBD) =9.174 ꢁ, and d(RA–HBD) =3.563 ꢁ. The
centers of smaller spheres represent the positions of the ligand atoms for
pharmacophore features with directionality (e.g., HBA, HBD, RA, etc.). Best
alignment of B) CGS21680, C) 1, D) NBTI, E) 6, and F) 11 to the pharmaco-
phore model.
Figure 3. The 3D pharmacophore model of human equilibrative nucleoside
transporter (hENT1) inhibitors. A) Geometric relationships of the pharmaco-
phore features. The pharmacophore features are represented by meshed
spheres with the following color code: gold=ring aromatic (RA), green=hy-
drogen bond acceptor (HBA). There are two HBA features in the constructed
pharmacophore model that are designated as HBA1 and HBA2. With analo-
gous notation as described for Figure 2, the distances between the pharma-
cophore features are d(RA–HBA1) =3.433 ꢁ, d(RA–HBA2) =10.388 ꢁ, and d(HBA1–
HBA2) =9.688 ꢁ. Best alignment of B) NBTI, C) 1, D) CGS21680, E) 6, and F) 11
to the pharmacophore model.
nicely (Figure 2B). In contrast, S-(4-nitrobenzyl)-6-thioinosine
(NBTI)[21] lacks a ring aromatic fitting (Figure 2D), in agreement
with its weak affinity as a hA2AR ligand, although it exhibits
strong binding with the adenosine transporter. Due to the lack
of an exocyclic amine as an H-bond donor moiety, NBTI shows
only a suboptimal fit to the pharmacophore model. If the
sulfur atom of NBTI is replaced by nitrogen (i.e., compound 5),
then the fitting orientation is similar to that of compounds 1,
6, and 11 (see figure S1 in Supporting Information). The de-
signed dual-action ligands 1, 6, and 11 fit at least three fea-
tures in this pharmacophore model for A2AR agonists (Fig-
ure 2C, E, and F).
The constructed pharmacophore model of the hENT1 inhibi-
tors consists of only three features, including two hydrogen
bond acceptors and one ring aromatic. All five compounds
(NBTI, 1, CGS21680, 6, and 11) can fit into all these three fea-
tures (Figure 3B–F). The difference in the number of features
between the pharmacophores of hA2AR agonists and hENT1 in-
hibitors can be attributed to the nature of the training set
compounds.
Zhu and Buolamwini[22] constructed an ENT1 pharmacophore
with the software PHASE.[23] In comparison, their pharmaco-
phore consists of five features, and the predicted conforma-
tions of the training set compounds are rather similar.[22] The
Pharmacophore model of human ENT1 inhibitors
To design dual-function compounds that act as concomitant
hA2AR agonists and hENT1 inhibitors, the pharmacophore of
hENT1 inhibitors was also constructed (Figure 3A). The training
set includes 25 compounds with hENT1 inhibitory activity
ranging from IC50 values of 0.29 nm to 32 mm, which were se-
lected from published sources (see Supporting Information).
1392
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1390 – 1400

Dual-Action Neuroprotective Compounds
inherent methodological difference between PHASE and Cata-
lyst should be responsible for the differences of the two phar-
macophores. For example, the pharmacophore constructed by
PHASE was based only on the three most active aligned com-
pounds, followed by establishing a 3D-QSAR with the rest of
the compounds. This method was based on an assumption
that the middle or lower potency compounds share a common
binding pose. In contrast, the pharmacophore and the 3D-
QSAR model constructed by the HypoGen module in Catalyst
were built simultaneously and were based on all training set
compounds. Therefore, the middle or low potency compounds
in the training set also contributed to the alignment during
the pharmacophore generation process. Despite the differen-
ces in modeling methodology, the statistical quality of our 3D-
QSAR model is similar (r2 =0.962 and RMSE=0.927) to that of
the previous study by PHASE (r2 =0.916 and RMSE=0.405),
and our model was capable of predicting the activity of our
synthesized compounds.
salt) in the presence of N,N-diisopropylethylamine. As 4-hy-
droxybenzylamine was not commercially available, it was pre-
pared in two steps from 4-hydroxybenzaldehyde via condensa-
tion with hydroxylamine to give an intermediate oxime, which
was subsequently hydrogenated by Pd/C catalysis and HCl. By
procedures similar to that used for compound 1, a series of N6-
substituted adenosine derivatives 2–6 were prepared by sub-
stitution reactions of 6-chloropurine ribofuranoside with ap-
propriately substituted benzylamines. In lieu of the conven-
tional heating method, microwave irradiation was also applied
to shorten the reaction time in the preparation of N6-(3-indoly-
lethyl)adenosine (6).
Treatment of N6-(4-methoxybenzyl)adenosine 3 with 2,2-di-
methoxypropane in anhydrous acetone afforded the corre-
sponding 3-acetonide, which reacted with p-toluenesulfonyl
chloride in the presence of pyridine to give a mixture of tosy-
late and chloride derivatives (Scheme 1). The tosylate deriva-
tive was unstable, whereas the chloride compound (7-aceto-
nide) could be isolated by chromatography and subsequently
hydrolyzed to give 7. Without separation, this mixture was
treated with sodium azide, followed by acid-catalyzed hydroly-
sis, to give an azido compound 8. Alternatively, Staudinger re-
duction of 8-acetonide gave an intermediate amine, which was
converted into the corresponding acetamide 9 and sulfona-
mide 10 after removal of the 2’,3’-isopropylidene group in
acid.
By carefully scrutinizing the structures in the hA2AR agonists
training set (see Supporting Information), we found that many
of them possess a hydrophobic group at the 5’ nucleoside
end, especially those compounds with higher potency, includ-
ing CGS21680. Therefore, the constructed pharmacophore
must include this important feature. In contrast, for the hENT1
inhibitors training set, nearly all 5’ nucleoside ends possess a
polar hydroxy group.
The pharmacophore differences among the two investigat-
ing targets also shed some light on the design of dual-function
compounds. For example, compound 6 does not fit well into
the hydrophobic feature in the A2AR pharmacophore model,
and it is indeed less potent than CGS21680, which fits well to
all features. However, compound 6 could fit well into
all three features in the hENT1 model.
In another approach (Scheme 2), the Cu2+-catalyzed 1,3-di-
polar cycloaddition (click reaction)[24] of azido compound 8
with 1-hexyne, 1-octyne, and 3-phenyl-1-propyne afforded tria-
zole derivatives 12, 13, and 14, respectively. Likewise, the click
reaction of 8-acetonide with propargyl alcohol gave a triazole
Compound 11 fits well into all features including
the hydrophobic part of the A2AR pharmacophore
model. Nevertheless, it still fits well to the hENT1
pharmacophore model, indicating the higher toler-
ance of ENT1 pharmacophore model. In summary, by
comparing the features and compound-fitting quali-
ties of these two pharmacophore models, we hy-
pothesize that the hA2AR binding pocket has an im-
portant hydrophobic site, and the hENT1 binding
pocket may be more flexible in accommodating nu-
cleosides with hydrophobic moieties at the 5’ posi-
tion in this compound series. Pharmacophore analy-
ses provide insight into the design and understand-
ing of dual-function compounds in the absence of
structural information for hENT1.
Synthesis of adenosine derivatives
A
representative library of adenosine analogues
(Figure 1) was developed based on the pharmaco-
phore models. Compound 1, originally isolated from
G. elata,[6] was synthesized in high yield by substitu-
tion reaction of 6-chloropurine ribofuranoside (17)
Scheme 1. Synthesis of N6-(4-methoxybenzyl)adenosine (3) and substitution of its 5’-hy-
with 4-hydroxybenzylamine (as the hydrochloride droxy group with chlorine, azide, acetamide, and p-toluenesulfonamide.
ChemMedChem 2011, 6, 1390 – 1400
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1393

J.-H. Lin, J.-M. Fang, et al.
MED
4-methoxybenzylamine, a 1,3-dipolar cycloaddition of the
cyano group with sodium azide introduced the desired tetra-
zole moiety at the C5’ position,[27] giving 16 after removal of
the 2’,3’-isopropylidene group under acidic conditions.
Biological evaluation of N6- and C5’-modified adenosine
derivatives
The pharmacological properties of the prepared adenosine an-
alogues were characterized by MDS Pharma Services (Taipei,
Taiwan) using radioligand binding assays. The binding con-
stants (Ki) of some representative compounds are listed in
Table 1. The potent A2AR agonist CGS21680 appears to lack ac-
tivity against ENT1, whereas the ENT1 inhibitor NBTI shows no
binding affinity for A2AR. Neither CGS21680 nor NBTI is a dual-
function drug. Some prepared adenosine analogues exhibit
dual actions toward A2AR and ENT1; in particular, compounds
1, 4, and 6 show Ki values in the low- and sub-micromolar
range with A2AR and ENT1, respectively. Except for 11 and 15,
the adenosine derivatives with modification at the C5’ position
appeared to lose their binding affinity for A2AR, yet they main-
tained high affinity for ENT1. It should be stressed that al-
though the dual-action compounds developed in this work do
not seem to be more potent in terms of binding affinity, they
could be more site-specific, and the dosage in the final formu-
lation could be decreased, therefore decreasing the probability
of inducing side effects including the reported immunosup-
pressive effect of very potent A2AR agonists such as CGS21680.
We previously reported that compound 1 isolated from an
aqueous methanolic extract of G. elata prevented
apoptosis of serum-deprived PC12 cells by suppress-
ing c-Jun N-terminal kinase (JNK) activity.[9] In this
study, serum-deprived PC12 cells were treated with
test compound at the indicated dose for 24 h. Cell vi-
ability was monitored by MTT assay, and is expressed
as a percentage of the MTT activity measured in the
serum-containing group, as shown in Table 2. At a
concentration of 1 mm, compounds 4 and 6 also res-
cued PC12 cells from apoptosis evoked by serum
withdrawal equally as well as 1, according to the cell
viability of MTT assays. Collectively, the dual function
of these compounds in the activation of the adeno-
sine receptor and in blockade of adenosine transport-
er activity might increase the effective concentration
of adenosine, especially when these two proteins are
in proximity.
Scheme 2. Elaboration of the 5-azido group by click reaction to give triazole
compounds.
compound, the hydroxy group of which was activated as a
mesylate and then reduced by sodium borohydride to give a
methyl group. Compound 11 was obtained after removal of
the 2’,3’-isopropylidene group.
The acetonide of 6-chloropurine ribofuranoside was oxidized
by (diacetoxyiodo)benzene with catalysis by (2,2,6,6-tetrame-
thylpiperidinyl-1-yl)oxyl (TEMPO)[25] to give carboxylic acid 18
(Scheme 3). The coupling reactions of acid 18 with ethylamine
and ammonia gave amides 19 and 20, respectively. The chlor-
ine atom in 19 was replaced by 4-hydroxybenzylamine, and
subsequent hydrolysis of the acetonide furnished amide 15.
On the other hand, amide 20 was converted into nitrile 21
upon treatment with DMSO, oxalyl chloride, and N,N-diisopro-
pylethylamine.[26] After the substitution of the chlorine atom by
Statistical assessment of pharmacophore models
Figure 4 shows the scatter plot of the experimental
pKi versus predicted pKi values from the pharmaco-
phore model of A2AR agonists. The r2 value of the
predicted Ki values versus the experimental Ki values
is 0.962, and the root-mean-square of error (RMSE) is
0.658 kcalmolꢀ1. This pharmacophore model was fur-
ther evaluated by using Fisher’s randomization test
for statistical significance, as implemented in the Cat-
Scheme 3. Synthesis of carboxamide and tetrazole derivatives.
1394
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1390 – 1400

Dual-Action Neuroprotective Compounds
Table 1. Binding activity of the N6- and C5’-modified adenosine deriva-
tives with adenosine receptor and transporter.[a]
Compound
Ki [mm]
[b]
A
R
ENT1[c]
2A
CGS21680
NBTI
7.77ꢂ10ꢀ2
>10
2.62
>100
14.4
30.1
3.21
27.7
4.39
>100
–
2.9ꢂ10ꢀ4
5.38ꢂ10ꢀ1
1
1-acetonide
>100
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
1.44ꢂ10ꢀ2
3.18ꢂ10ꢀ1
3.72
6.51ꢂ10ꢀ3
3.47
2.98
–
5.81ꢂ10ꢀ1
1.43
>100
>100
2ꢂ10ꢀ1
41.8
9.60ꢂ10ꢀ1
5.11ꢂ10ꢀ1
5.2ꢂ10ꢀ2
1.16ꢂ10ꢀ1
Figure 4. Scatter plot of the predicted pKi values of A2AR agonists versus
>100
>100
>100
20.3
>100
*
*
measured pKi values; : training compounds, : synthesized compounds.
>10
1.17
Table 3. Comparison of activities of compounds with the fitted number
[a] Radioligand binding assays were performed by MDS Pharma Services
Taiwan (Taipei, Taiwan) using standard binding protocols. [b] Human ade-
nosine A2A receptor. [c] Guinea pig equilibrium transporter 1.
of features of the A2AR agonist pharmacophore model.[a]
Compound
HBD
HBA
RA
HP
Fit value
CGS21680
1 (T1-11)
NBTI
0.166
0.121
0.422
0.377
1.181
0.125
0.135
0.304
0.359
0.445
0.201
0.248
ꢂ
0.403
0.434
0.436
ꢂ
0.901
ꢂ
10.6482
8.68049
8.62848
8.52907
9.59155
Table 2. Cell viability of the N6- and C5’-modified adenosine derivatives.[a]
6
11
0.544
Compound
Cell viability [%]
[a] All values are reported in ꢁ.
CGS21680
NBTI
1
88.6ꢁ9.6
29.1ꢁ2.1
81.5ꢁ1.8
63.7ꢁ2.9
42.3ꢁ1.8
83.9ꢁ4.5
48.2ꢁ1.3
118.8ꢁ3.9
36.8ꢁ5.3
29.9ꢁ1.5
24.1ꢁ4.5
36.4ꢁ3.9
36.7ꢁ0.6
28.9ꢁ2.6
27.6ꢁ0.4
30.6ꢁ4.7
87.0ꢁ8.4
30.0ꢁ3.5
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
tion of the feature on the compound from the center of the
feature in the pharmacophore model. To reiterate, Table 3 is a
quantitative representation of Figure 4. When a ligand is fitted
into a pharmacophore, the quality of fitting (or mapping) is in-
dicated by the “fit value.” A higher fit value represents a better
fit, and the computer fit values depend on two factors: the
weights assigned to the pharmacophore features and how
close the features in the molecules are to the exact locations
of the features in the pharmacophore model.
The potent A2AR agonist CGS21680 fits all the four features
of the constructed pharmacophore, and the deviations of the
fitted feature locations from the exact locations of the features
of the pharmacophore model are also very small. Compared
with CGS21680, compound 1 lacks a hydrophobic moiety to fit
into the feature of the pharmacophore model, and therefore
exhibited decreased affinity. In contrast, NBTI lacks a ring aro-
matic moiety, which completely abolishes the activity to ago-
nize the adenosine A2A receptor. This indicates that the ring ar-
omatic feature may be more important than the hydrophobic
moiety in this pharmacophore model. Compound 6 does not
fit the hydrophobic feature of the pharmacophore model,
whereas compound 11 does show a higher fit value. In con-
trast, the binding assays indicate that compound 6 (Ki =
4.39 mm) exhibits ~10-fold stronger binding affinity than com-
pound 11 (Ki =41.8 mm). This can be rationalized by the fact
[a] Serum-deprived PC12 cells were treated with or without compound at
1 mm for 24 h. Cell viability was monitored by the MTT assay, and is ex-
pressed as a percentage of the MTT activity measured in the serum-con-
taining group (100%). Serum deprivation resulted cell survival rate to
33.0ꢁ1.9. Data points represent the mean ꢁSEM of at least three inde-
pendent experiments (n=3–6).
Scramble module. The CatScramble module scrambles pKi
values randomly 19 times to generate new hypotheses (i.e.,
pharmacophore models). None of the 19 hypotheses from the
scrambled data had a cost lower than the reported hypothesis.
Table 3 summarizes the fitted features of the compounds in
Figure 4, along with the distance deviation of the fitted loca-
ChemMedChem 2011, 6, 1390 – 1400
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1395

J.-H. Lin, J.-M. Fang, et al.
MED
that all the fitted locations of the features of compound 6
have fewer deviations from the exact locations of the features.
The constructed pharmacophore model of the hENT1 inhibi-
tors consists of only three features, namely, a ring aromatic fea-
ture and two hydrogen bond acceptors. The r2 value of the
predicted versus experimental Ki values is 0.927, and the RMSE
is 0.85 kcalmolꢀ1 (Figure 5). This pharmacophore model was
Conclusions
We have adopted a dual-pharmacophore modeling approach
in the design of dual-action compounds that target the A2AR
signaling system. Based on the structural scaffold of 1, we de-
signed and synthesized a series of adenosine derivatives and
carried out chemical modifications of adenosine if the pharma-
cophore fitting of the modified compound predicts acceptable
activity. The competitive ligand binding assays verified that the
designed compounds indeed bind to both A2AR and ENT1 with
moderate affinity. Finally, these compounds were shown to
prevent apoptosis in serum-deprived PC12 cells, which is a cru-
cial indication for their neuroprotection and potential for treat-
ing neurodegenerative diseases.
Experimental Section
General
All reagents and solvents were of reagent grade and were used
without further purification unless otherwise specified. THF and
Et2O were distilled from Na/benzophenone, and CH2Cl2 was dis-
tilled from CaH2. All air- or moisture-sensitive experiments were
performed under argon. All glassware was oven-dried for >2 h
and used after cooling to room temperature (RT) in desiccators. Mi-
crowave reactions were conducted with a focused single-mode mi-
crowave unit (CEM Discover). The instrument consists of a continu-
ously focused microwave power delivery system with operator-se-
lectable power output.
Figure 5. Scatter plot of the predicted pKi values of ENT1 inhibitors versus
*
*
measured pKi values; : training compounds, : synthesized compounds.
further evaluated by the CatScramble module. All five com-
pounds (NBTI, 1, CGS21680, 6, and 11) can fit into all these
three features (Figure 3B–F), and therefore the deviations from
the exact locations of the features need to be compared
(Table 4). Apparently, the most potent inhibitor, NBTI, has the
Melting points were recorded on a Yanaco micro apparatus. Opti-
cal rotations were measured on a digital polarimeter from Japan
JASCO (DIP-1000); [a]D values are given in units of 10ꢀ1 degcm2 gꢀ1
.
IR spectra were recorded on a Nicolet Magna 550-II instrument.
NMR spectra were obtained on a Varian Unity Plus-400 (400 MHz),
and chemical shifts (d) are reported in ppm relative to dH =7.24/
dC =77.0 (central line of t) for CHCl3/CDCl3, dH =2.05/dC =29.92 for
(CH3)2CO/(CD3)2CO, dH =3.31/dC =49.0 for CH3OH/CD3OD, and dH =
2.49 (m)/dC =39.5 (m) for (CH3)2SO/(CD3)2SO. The splitting patterns
are reported as s (singlet), d (doublet), t (triplet), q (quartet), m
(multiplet) and br (broad). Coupling constants (J) are given in Hz.
ESI-MS experiments were conducted on a Bruker Daltonics BioTOF
III high-resolution mass spectrometer. Analytical thin-layer chroma-
tography (TLC) was performed on E. Merck silica gel 60 F254 plates
(0.25 mm). Compounds were visualized by UV, anisaldehyde, or
ninhydrin spray. Column chromatography was carried out on col-
umns packed with 70–230 mesh silica gel. The purity of com-
pounds tested on A2AR and ENT1 was assessed to be ꢂ95% by
HPLC (Agilent HP-1100) with detection at l=280 nm.
Table 4. Comparison of activities of compounds with the fitted number
of features of the ENT1 inhibitor pharmacophore model.[a]
Compound
HBD
HBA
RA
Fit value
NBTI
0.421
0.544
0.567
0.421
0.647
0.619
1.276
0.647
0.793
0.586
0.358
0.493
5.92748
4.94133
5.37611
5.40138
1
CGS21680
6
[a] All values are reported in ꢁ.
highest fit value and smallest deviation of all three features.
The fit values of compounds 1, 11, and 6 are 6.61, 6.4, and
5.86, respectively, which are consistent with the ranking of
their measured activity. CGS21680 (with a high fit value of 7.1)
is clearly an outlier of this model, as this compound has no in-
hibitory activity toward hENT1. However, CGS21680 is the only
compound with a ring aromatic feature fitted onto the nucleo-
side moiety (Figure 3). It is therefore important to carefully ex-
amine whether the fitted functional group is indeed the same
as the functional groups of the training set compounds that
define the consensus feature in the pharmacophore analysis.
The “fit value” alone cannot be considered the measure of fit-
ness.
Construction of pharmacophore models
The HypoGen module of Catalyst (Accelrys) was used to construct
the pharmacophore model of human A2AR agonists and human
ENT1 inhibitors. The chemical structures of the training set com-
pounds and their binding affinities to the hA2AR (or hENT1) were
collected from published sources (see Supporting Information).[27,28]
It is important to ensure that the activity of the training com-
pounds cover at least four orders of magnitude, with at least three
compounds in each log scale.[15,29] It is also recommended to select
compounds with greater chemical diversity for constructing the
training set.[29] Each compound was sketched with ChemDraw and
1396
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1390 – 1400

Dual-Action Neuroprotective Compounds
then imported into Catalyst 4.11. The “Best” option was used for
conformation generation. In Catalyst, the Poling algorithm was
used to generate 250 conformations, the energies of which are
<20.0 kcalmolꢀ1 from the lowest energy of all the conformations.
Five molecular features were selected: hydrophobic (HP), hydrogen
bond acceptor (HBA), hydrogen bond donor (HBD), positively ioniz-
able atom (PI), and negatively ionizable atom (NI). All these com-
pounds were loaded into the Catalyst spreadsheet, and the default
uncertainty value of 3 was assigned. All other parameters were set
at default.
J=7.2 Hz), 7.32 (1H, d, J=9.2 Hz), 7.18 (1H, s), 7.05 (1H, t, J=
8.0 Hz), 5.80 (1H, d, J=6.0 Hz), 5.47–5.44 (2H, m), 5.20 (1H, d, J=
4.4 Hz), 4.61 (1H, d, J=5.6 Hz), 4.14 (1H, d, J=2.8 Hz), 3.96 (1H, d,
J=3.2 Hz), 3.77 (1H, brs), 3.69–3.52 (2H, m), 3.01 ppm (2H, t, J=
7.2 Hz); 13C NMR ([D6]DMSO, 100 MHz): d=154.3, 152.2, 148.0,
139.5, 136.0, 127.1, 122.4, 120.8, 119.6, 118.3, 118.1, 111.7, 111.2,
87.9, 85.8, 73.4, 70.6, 61.6, 40.5, 25.1 ppm; ESI-HRMS calcd for
C20H23N6O4: 411.1775, found: m/z 411.1750 [M+H]+.
5’-Azido-5’-deoxy-2’,3’-O-isopropylidene-N6-(4-methoxybenzyl)a-
denosine (8-acetonide): A solution of p-toluenesulfonyl chloride
(6.3 g, 34.6 mmol) in anhydrous pyridine (6.0 mL) was added drop-
wise via syringe to the acetonide of N6-(4-methoxybenzyl)adeno-
sine (3-acetonide, 2.96 g, 6.9 mmol) in anhydrous pyridine (36 mL).
The mixture was stirred at RT for 6 h. Pyridine was removed under
reduced pressure, and the residue was extracted with CH2Cl2 and
H2O. The organic layer was dried over MgSO4, filtered, and concen-
trated to give a mixture of sulfonate and the chloride derivatives
N6-(4-Hydroxybenzyl)adenosine (1):[6] Hydroxylamine hydrochlo-
ride (1.29 g, 18.6 mmol) and NaOAc (1.67 g, 20.4 mmol) were
added to a solution of 4-hydroxybenzaldehyde (1.25 g, 10.2 mmol)
in EtOH (20 mL). The reaction mixture was stirred at RT for 6 h.
EtOH was removed under reduced pressure. H2O was added to the
residue, followed by extraction with Et2O (3ꢂ20 mL). The com-
bined organic layer was dried over MgSO4. After the volatiles were
removed by rotary evaporation under reduced pressure, the resi-
due was recrystallized from CH2Cl2 to give the oxime of 4-hydroxy-
benzaldehyde (1.3 g, 93%). C7H7NO2: light-yellow solid; mp: 92.0–
93.68C.
1
(5:1) as shown by the H NMR spectrum.
The above-prepared mixture was dissolved in anhydrous DMF
(70 mL), and NaN3 (1.34 g, 20.6 mmol) was added. The mixture was
stirred at 808C for 6 h, and then concentrated under reduced pres-
sure. The residue was extracted with CH2Cl2 and H2O. The organic
layer was dried over MgSO4, filtered, and concentrated to give a
pale-yellow oil, which was purified by flash chromatography (silica
gel, CH2Cl2/MeOH (100:1)) to give 8-acetonide (653 mg, 21% over-
all yield). C21H24N8O4: colorless oil; TLC (EtOAc/Hexane (6:4)) Rf =
0.39; ½aꢃ2D3 =+5.0 (EtOAc, c=1.0); IR (neat): n˜max =3280, 2987, 2931,
2101, 1618, 1512, 1478, 1375, 1330, 1296, 1248, 1211, 1154,
A solution of the above-prepared oxime (342 mg, 2.5 mmol) and
concentrated HCl (1 mL) in EtOH (20 mL) was subjected to hydro-
genation at atmospheric pressure in the presence of 10% Pd/C
(80 mg) for 4 h. The reaction mixture was filtered through Celite.
The filtrate was concentrated to yield the HCl salt of 4-hydroxyben-
zylamine as light-yellow solids, which were used for the next step
without further purification.
1
1091 cmꢀ1; H NMR (CDCl3, 400 MHz): d=8.38 (1H, brs), 7.72 (1H,
A mixture of 4-hydroxybenzylamine (395 mg, as the HCl salt), 6-
chloropurine riboside (143 mg, 0.5 mmol), and iPr2NEt (2 mL,
12 mmol) in 1-propanol (25 mL) was heated at 708C for 6 h. The
mixture was concentrated under reduced pressure, and triturated
with H2O to give white precipitates, which were filtered to yield
the title compound 1 (151 mg, 81%). The purity of product was
>99% as shown by HPLC on an Inertsil ODS-3 column (4.6ꢂ
brs), 7.26 (2H, d, J=8.8 Hz), 6.84 (2H, d, J=8.8 Hz), 6.37 (1H, brs),
6.06 (1H, d, J=2.0 Hz), 5.46 ꢀ5.44 (1H, m), 5.07 ꢀ5.05 (1H, m),
4.77 (2H, brs), 4.38 ꢀ4.35 (1H, m), 3.77 (3H, s), 3.51–3.62 (2H, m),
1.61 (3H, s), 1.39 ppm (3H, s); 13C NMR (CDCl3, 100 MHz): d=
158.7, 154.5, 153.1, 147.9, 139.0, 130.2, 128.8 (2ꢂ), 120.1, 114.4,
113.8 (2ꢂ), 90.4, 85.6, 83.9, 82.0, 55.2, 52.2, 43.7, 29.7, 27.1,
25.3 ppm; ESI-HRMS calcd for C21H24N8O4: 453.1999, found: m/z
453.1999 [M+H]+.
250 mm, 5 mm) with elution of 0.1% TFA/MeOH (6:4). C17H19N5O5:
20
white powder; mp: 208.7–209.28C (lit.[6] mp: 216–2198C); ½aꢃD
=
5’-Acetamido-5’-deoxy-N6-(4-methoxybenzyl)adenosine (9): The
azido compound 8-acetonide (95 mg, 0.21 mmol) was stirred with
PPh3 (66 mg, 0.24 mmol) in THF/H2O (10:1, 2 mL) at RT for 4.5 h.
The mixture was concentrated under reduced pressure. The resi-
due was taken up with CH2Cl2 and H2O, and acidified to pH 2 with
aqueous HCl (1m). The aqueous phase was separated, neutralized
with saturated aqueous NaHCO3, and extracted with CH2Cl2. The
organic extract was dried over MgSO4, filtered, and concentrated
to yield a crude amine product.
ꢀ64.5 (DMSO, c=1) (lit.[6] ½aꢃ2D5 =ꢀ87 (MeOH, c=0.1)); TLC (MeOH/
EtOAc (1:9)) Rf =0.3; 1H NMR ([D6]DMSO, 400 MHz): d=9.22 (1H,
s), 8.34 (1H, s), 8.30 (1H, brs), 8.18 (1H, s), 7.12 (2H, d, J=8.0 Hz),
6.65 (2H, d, J=8.0 Hz), 5.86 (1H, d, J=5.6 Hz), 5.41 (2H, m), 5.18
(1H, d, J=5.6 Hz), 4.60 (2H, m), 4.13 (1H, q, J=4.6, 7.4 Hz), 3.95
(1H, q, J=3.4, 6.2 Hz), 3.66 (1H, m), 3.53 ppm (1H, m); 13C NMR
([D6]DMSO, 400 MHz): d=155.3, 153.6, 151.6, 147.6, 139.1, 129.5,
127.9 (2ꢂ), 119.2, 114.4 (2ꢂ), 87.6, 85.6, 73.3, 70.5, 61.5, 42.4 ppm;
ESI-HRMS calcd for C17H20N5O5: 374.1459, found: m/z 374.1412
[M+H]+.
The crude amine was treated with acetic anhydride (98.6 mL,
1.05 mmol) in anhydrous pyridine (0.2 mL). The mixture was stirred
at RT for 1.5 h, and then concentrated under reduced pressure.
The residue was extracted with CH2Cl2 and H2O. The organic layer
was dried over MgSO4, filtered, concentrated, and purified by flash
chromatography (silica gel, CH2Cl2/MeOH (98:2)) to give the aceto-
nide of compound 9 (56 mg, 57% yield for two steps). The purity
of product was >99% as shown by HPLC on an HC-C18 column
(Agilent, 4.6ꢂ250 mm, 5 mm) with elution of gradients of 30–60%
aqueous CH3CN. C23H28N6O5: colorless oil; TLC (CH2Cl2/MeOH (98:2))
Rf =0.2; ½aꢃ2D5 =ꢀ146.6 (CHCl3, c=1.0); IR (neat): n˜max =3280, 3062,
2989, 2930, 2835, 2358, 1667, 1620, 1513, 1376, 1336, 1296, 1246,
N6-(3-Indolylethyl)adenosine (6): A solution of 6-chloropurine ri-
bonucleoside (71 mg, 0.25 mmol), tryptamine (101 mg, 0.63 mmol),
and iPr2NEt (0.24 mL, 2.88 mmol) in EtOH (3 mL) was placed in a
round-bottom flask (10 mL). The flask was placed into the cavity of
a focused monomode microwave reactor, and irradiated at 150 W
for 10 min in EtOH at reflux. The mixture was concentrated by
rotary evaporation, and the residue was purified by flash chroma-
tography (silica gel, MeOH/EtOAc (1:9)) to give the title compound
6 (85 mg, 83%). The purity of product was >99% as shown by
HPLC on an HC-C18 column (Agilent, 4.6ꢂ250 mm, 5 mm) by elution
with gradients of 30–60% aqueous CH3CN. C20H22N6O4: white
powder; mp: 187.0–187.28C; ½aꢃ2D0 =ꢀ55.7 (CH3OH, c=1.0); TLC
(MeOH/EtOAc (1:9)) Rf =0.41; 1H NMR ([D6]DMSO, 400 MHz): d=
10.78 (1H, s), 8.33 (1H, s), 8.25 (1H, s), 7.96 (1H, brs), 7.61 (1H, d,
1215, 1096, 1034 cmꢀ1 1H NMR (CDCl3, 400 MHz): d=8.36–8.38
;
(2H, m), 7.73 (1H, s), 7.28 (2H, d, J=8.8 Hz), 6.86 (2H, d, J=8.8 Hz),
6.15 (1H, brs), 5.77 (1H, d, J=4.8 Hz), 5.26 (1H, t, J=4.8 Hz), 4.81
ChemMedChem 2011, 6, 1390 – 1400
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1397

J.-H. Lin, J.-M. Fang, et al.
MED
(1H, dd, J=4.0, 2.4 Hz), 4.76 (2H, brs), 4.47–4.48 (1H, m), 4.11–4.17
(1H, m), 3.79 (3H, s), 3.24 (1H, d, J=14.4 Hz), 2.15 (3H, s), 1.61 (3H,
s), 1.34 ppm (3H, s); 13C NMR (CDCl3, 100 MHz): d=170.5, 158.8,
154.8, 152.7, 147.7, 139.7, 130.0, 128.9 (2ꢂ), 121.1, 114.6, 113.9 (2ꢂ
), 92.5, 83.3, 82.2, 81.3, 55.3, 43.9, 41.1, 27.6, 25.4, 23.2 ppm; ESI-
HRMS calcd for C23H28N6O5: 469.2190, found: m/z 469.2193 [M+
H]+.
1244, 1176, 1111, 1058 cmꢀ1
;
1H NMR (CD3OD, 400 MHz): d=8.22
(1H, s), 7.99 (1H, s), 7.45 (1H, s), 7.31 (2H, d, J=8.8 Hz), 6.87 (2H,
d, J=8.8 Hz), 5.96 (1H, d, J=4.0 Hz), 4.82–4.68 (5H, m), 4.46 (1H, t,
J=4.0 Hz), 4.34 (1H, q, J=4.0 Hz), 3.77 (3H, s), 2.15 ppm (3H, s);
13C NMR (CDCl3, 100 MHz): d=158.7, 154.2, 152.9, 148.0, 142.8,
138.8, 130.1, 128.8 (2ꢂ), 123.1, 119.6, 113.8 (2ꢂ), 89.3, 82.2, 73.4,
70.8, 55.2, 50.9, 43.9, 10.5 ppm; ESI-HRMS (negative mode) m/z
[MꢀH]ꢀ calcd for C21H24N8O4: 451.1842, found: 451.1843.
3,4-Dihydroxy-5-[6-(4-hydroxybenzylamino)-purin-9-yl]tetrahy-
drofuran-2-carboxylic acid ethylamide (15): The acetonide de-
rived from 6-chloropurine ribofuranoside (17-acetonide, 158 mg,
0.48 mmol) was stirred with PhI(OAc)2 (509 mg, 1.56 mmol) and
The acetonide of 9 (17.2 mg, 0.037 mmol) was stirred in 3m HCl/
THF (1:1, 0.1 mL) at RT for 14 h, and then neutralized with saturat-
ed aqueous NaHCO3. The mixture was concentrated under reduced
pressure, and the residue was extracted with CH2Cl2 and H2O. The
organic layer was dried over MgSO4, filtered, and concentrated to
give the title compound 9 (11 mg, 70%). The purity of product 9
was 99% as shown by HPLC on an HC-C18 column (Agilent, 4.6ꢂ
250 mm, 5 mm) with elution of gradients of 30–60% aqueous
CH3CN in 20 min. C20H24N6O5: white powder; mp: 121.1–121.68C;
TLC (CH2Cl2/MeOH (9:1)) Rf =0.5; ½aꢃ2D5 =ꢀ108.7 (THF, c=0.89); IR
(neat): n˜max =3275, 3071, 2923, 2852, 2360, 1621, 1512, 1375, 1339,
(2,2,6,6-tetramethylpiperidinyl-1-yl)oxyl
(TEMPO,
15.4 mg,
0.1 mmol) in a degassed solution of CH3CN/H2O (1:1, 2.6 mL) at
408C for 4 h. The mixture was concentrated under reduced pres-
sure to yield a crude acid product 18.
The crude acid was treated with EtNH2 (117 mg, as the HCl salt), O-
(benzotriazol-1-yl)-N,N,N’,N’-tetramethyluronium hexafluorophos-
phate (HBTU, 375 mg, 0.72 mmol) and iPr2NEt (0.5 mL, 2.89 mmol)
in anhydrous DMF (11.6 mL) at RT for 14 h. The mixture was con-
centrated under reduced pressure. The residue was extracted with
CH2Cl2 and H2O. The organic layer was dried over MgSO4, filtered,
concentrated, and purified by flash chromatography (silica gel,
EtOAc/hexane (1:1)) to yield an amide product 19 as colorless oil.
1
1297, 1245, 1175, 1126, 1076 cmꢀ1; H NMR (CDCl3, 400 MHz): d=
8.76 (1H, s), 8.27 (1H, s), 7.24 (2H, d, J=8.4 Hz), 6.81 (2H, d, J=
8.4 Hz), 6.54 (1H, s), 5.70 (1H, d, J=5.6 Hz), 4.72 (3H, d, J=5.6 Hz),
4.23 (1H, s), 4.18 (1H, s), 3.98–4.03 (1H, m), 3.75 (3H, s), 3.13 (1H,
d, J=14.0 Hz), 2.02 ppm (3H, s); 13C NMR (DMSO, 100 MHz): d=
169.4, 157.9, 154.3, 152.3, 148.3, 140.2, 131.8, 128.4 (2ꢂ), 119.8,
113.5 (2ꢂ), 87.9, 83.6, 72.6, 71.3, 55.1, 42.4, 41.1, 22.7; ESI-HRMS
calcd for C20H24N6O5: 427.1730, found: m/z 427.1727 [M+H]+.
The amide product was treated with 4-hydroxybenzylamine
(385 mg, 2.4 mmol as the HCl salt) and iPr2NEt (2.8 mL, 16.9 mmol)
in 1-propanol (28 mL) at 708C for 2 h. The mixture was concentrat-
ed under reduced pressure, and the residue was extracted with
CH2Cl2 and H2O. The organic layer was dried over MgSO4, filtered,
concentrated, and purified by flash chromatography (silica gel,
5’-Deoxy-5’-(4-methyl-1,2,3-triazol-1-yl)-N6-(4-methoxybenzyl)a-
denosine (11): A mixture of azido compound 8-acetonide (313 mg,
0.69 mmol), CuSO4·5H2O (24.9 mg), sodium ascorbate (61.4 mg)
and propargyl alcohol in H2O/tBuOH (1:1, 7 mL) was stirred at RT
for 12 h, and then concentrated under reduced pressure. The resi-
due was extracted with CH2Cl2 and H2O. The organic layer was
dried over MgSO4, filtered, and concentrated to yield a triazole ace-
tonide (~220 mg) as a colorless oil. TLC (CH2Cl2/MeOH (9:1)) Rf =
0.5; ESI-HRMS calcd for C24H28N8O5: 509.2261, found: m/z 509.2267
[M+H]+.
CH2Cl2/MeOH (97:3)) to give 15-acetonide (179 mg, 82%).
22
C23H28N6O5: colorless oil; TLC (EtOAc/hexane (4:1)) Rf =0.13; ½aꢃD
=
ꢀ32.0 (EtOAc, c=1.0); IR (neat): n˜max =3347, 3103, 2982, 2933,
1732, 1667, 1615, 1516, 1479, 1461, 1376, 1332, 1295, 1245, 1212,
1
1154, 1088 cmꢀ1; H NMR (CDCl3, 400 MHz): d=9.01 (1H, brs), 8.35
(1H, brs), 7.81 (1H, s), 7.14 (1H, brs), 7.04 (2H, d, J=8.0 Hz), 6.68
(2H, d, J=8.0 Hz), 6.49 (1H, t, 4.8 Hz), 6.03 (1H, d, J=2.4 Hz), 5.33–
5.38 (2H, m), 4.70 (3H, s), 3.09–3.16 (2H, m), 1.62 (3H, s), 1.37 (3H,
s), 0.90 (3H, t, J=7.2 Hz); 13C NMR (CDCl3, 100 MHz): d=168.7,
155.8, 154.2, 153.1, 147.7, 139.1, 128.8 (3ꢂ), 119.6, 115.4 (2ꢂ),
114.3, 91.6, 85.6, 83.3, 82.4, 43.9, 34.0, 27.0, 25.1, 14.2; ESI-HRMS m/
z [M+H]+ calcd for C22H26N6O5: 455.2043, found: 455.2037.
The above-prepared triazole compound was stirred with Et3N
(0.15 mL, 1.08 mmol) and methylsulfonyl chloride (0.08 mL,
1.08 mmol) in anhydrous CH2Cl2 (4.3 mL) at RT for 2 h. The mixture
was concentrated under reduced pressure, and the residue was ex-
tracted with CH2Cl2 and H2O. The organic layer was dried over
MgSO4, filtered, and concentrated to yield a mesylate compound
as a colorless oil. TLC (EtOAc/Hex (4:1)) Rf =0.45; ESI-HRMS: m/z
[M+Na]+ calcd for C25H30N8O7SNa: 609.1856, found: 609.1876.
The acetonide of 15 (26 mg, 0.057 mmol) was stirred in 1m HCl/
THF (1:1, 0.3 mL) at RT for 16 h, and then neutralized with saturat-
ed aqueous NaHCO3. After concentration, the residue was triturat-
ed with H2O to give the title compound 15, which was then recrys-
tallized from MeOH (14.65 mg, 62%). C19H22N6O5: white powder,
mp: 179.7–180.58C; TLC (EtOAc) Rf =0.04; ½aꢃ2D3 =ꢀ27.7 (MeOH, c=
1.0); IR (neat): n˜max =3256, 2688, 2360, 1618, 1515, 1335, 1294,
The mesylate was treated with NaBH4 (24.5 mg, 0.65 mmol) at 08C
in DMF, and then heated at 608C for 6 h. The mixture was concen-
trated under reduced pressure, and the residue was extracted with
CH2Cl2 and H2O. The organic layer was dried over MgSO4, filtered,
and concentrated to give 11-acetonide as colorless oil. TLC (EtOAc/
Hex (4:1)) Rf =0.25; ESI-HRMS: m/z [M+H]+ calcd for C24H28N8O4:
493.2312, found: 493.2312.
1
1232, 1128, 1052 cmꢀ1; H NMR (CD3OD, 400 MHz): d=8.29 (1H, s),
8.22 (1H, s), 7.20 (2H, d, J=8.4 Hz), 6.73 (2H, d, J=8.4 Hz), 6.00
(1H, d, J=7.6 Hz), 4.76–4.73 (1H, m), 4.70 (2H, brs), 4.46 (1H, s),
4.30–4.31 (1H, m), 3.36 (2H, q, 7.2 Hz), 1.21 ppm (3H, t, 7.2 Hz);
13C NMR (CD3OD, 100 MHz): d=171.8, 157.5, 155.8, 153.6, 149.1,
141.9, 130.5, 129.9 (2ꢂ), 121.3, 116.2 (2ꢂ), 90.5, 86.3, 74.9, 73.4,
44.9, 35.2, 15.3 ppm; ESI-HRMS m/z [MꢀH]+ calcd for C19H21N6O5:
413.1573, found: 413.1573; Anal. Calcd for C19H22N6O5·H2O: C 52.77,
H 5.59, N 19.43, found: C 52.88, H 5.40, N 19.44.
The acetonide of 11 was stirred in 3m HCl/THF (1:1, 0.33 mL) at RT
for 14 h, and then neutralized with saturated aqueous NaHCO3.
The mixture was concentrated under reduced pressure, and the
residue was dissolved in THF, filtered, and concentrated to give the
title compound 11 (48.1 mg, 25% overall yield). The purity of prod-
uct was 98% as shown by HPLC on an HC-C18 column (Agilent,
4.6ꢂ250 mm, 5 mm) with elution of gradients of 30–60% aqueous
CH3CN. C21H24N8O4: white powder; mp: 183.0–183.28C; TLC
(CH2Cl2/MeOH (9:1)) Rf =0.12; ½aꢃ2D7 =+20.3 (CH3OH, c=0.45); IR
(neat): n˜max =3217, 2921, 2850, 2685, 1620, 1513, 1470, 1337, 1297,
2-[6-(4-Methoxybenzylamino)purin-9-yl]-5-(1H-tetrazol-5-yl)tetra-
hydrofuran-3,4-diol (16): The crude acid 18 obtained from oxida-
tion of 17-acetonide (~3.98 mmol) with PhI(OAc)2/TEMPO was
1398
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1390 – 1400

Dual-Action Neuroprotective Compounds
treated with NH4Cl(426 mg, 7.96 mmol), benzotriazol-1-yl-oxytripyr-
Compound 16-acetonide (460 mg, 0.99 mmol) was stirred in 3m
HCl/THF (1:1, 0.1 mL) at RT for 14 h, and then neutralized with sa-
turated aqueous NaHCO3. The mixture was concentrated under re-
duced pressure; the residue was taken up with THF, filtered, and
concentrated to give the title compound 16 (320 mg, 76%). The
purity of product was 99% as shown by HPLC on an HC-C18
column (Agilent, 4.6ꢂ250 mm, 5 mm) with elution of gradients of
30–60% aqueous CH3CN in 20 min. C18H19N9O4: white powder; mp:
210.0–210.68C; TLC (CH2Cl2/MeOH (9:1)) Rf =0.05; ½aꢃ2D6 =ꢀ25.8
(THF, c=1.0); IR (neat): n˜max =3397, 2841, 2692, 1623, 1511, 1475,
rolidinophosphonium
hexafluorophosphate
(PyBOP,
3.07 g,
5.97 mmol), hydroxybenzotriazole (HOBt, 807 mg, 5.97 mmol), and
iPr2NEt (2.5 mL, 15.9 mmol) in anhydrous DMF (40 mL) at 508C for
14 h. The mixture was concentrated under reduced pressure. The
residue was extracted with CH2Cl2 and H2O. The organic layer was
dried over MgSO4, filtered, concentrated, and purified by flash
chromatography (silica gel, EtOAc/hexane (1:1 to 4:1)) to yield an
amide product 20 as a colorless oil.
A solution of DMSO (0.85 mL, 11.9 mmol) in CH2Cl2 (10 mL) was
added to a solution of oxalyl chloride (0.7 mL, 7.96 mmol) in CH2Cl2
(10 mL) at ꢀ788C. The mixture was stirred for 30 min, and a solu-
tion of amide 20 (~3.98 mmol) in CH2Cl2 (20 mL) was added. The
mixture was stirred at ꢀ788C for another 30 min, and iPr2NEt
(2.6 mL, 15.9 mmol) was added. After 1 h stirring, formation of ni-
trile 21 was monitored by TLC analysis. The mixture was extracted
with CH2Cl2 and H2O. The organic phase was dried over MgSO4, fil-
tered, concentrated, and purified by flash chromatography (silica
gel, EtOAc/hexane (2:3)) to yield nitrile 21 as a colorless oil
(863 mg) contaminated with a small amount of HOBt.
1
1419, 1339, 1302, 1236, 1180, 1124, 1045 cmꢀ1; H NMR ([D6]DMSO,
400 MHz): d=8.82 (1H, s), 8.30 (1H, brs), 8.20 (1H, s), 7.26 (2H, d,
J=8.0 Hz), 6.83 (2H, d, J=8.0 Hz), 6.08 (1H, d, J=5.6 Hz), 5.53 (1H,
d, J=6.0 Hz), 5.46 (1H, d, J=2.8 Hz), 5.18 (1H, s), 4.91 (1H, d, J=
5.2 Hz), 4.62 (2H, brs), 4.20 (1H, s), 3.69 ppm (3H, d, J=2.0 Hz);
13C NMR ([D6]DMSO, 100 MHz): d=159.9, 157.4, 153.7, 151.9, 138.8,
131.5, 127.9 (2ꢂ), 113.1 (2ꢂ), 85.9, 79.1, 75.4, 74.3, 54.6, 41.9 ppm;
ESI-HRMS (negative mode) m/z [MꢀH]ꢀ calcd for C18H19N9O4:
427.1730, found: 427.1727.
Radioligand binding assays. Radioligand binding assays were per-
formed by MDS Pharma Services Taiwan (Taipei, Taiwan) using stan-
dard binding protocols. For the binding assay of the A2AR,[30] mem-
brane proteins collected from HEK293 cells overexpressing the
human A2AR were incubated in reaction buffer [50 mm Tris-HCl
(pH 7.4), 10 mm MgCl2, 1 mm EDTA, and 2 UmLꢀ1 adenosine deami-
nase] containing [3H]CGS21680 (50 nm) for 90 min at 258C. Non-
specific binding was assessed in the presence of 50 mm adenosine-
5’-N-ethylcarboxamide (NECA).
The above-prepared nitrile product (863 mg, 2.68 mmol) was treat-
ed with 4-methoxybenzylamine (1.84 g, 13.4 mmol) and iPr2NEt
(15.5 mL) in 1-propanol (26 mL) at 708C for 4 h. The mixture was
concentrated under reduced pressure, and the residue was extract-
ed with CH2Cl2 and H2O. The organic layer was dried over MgSO4,
filtered, concentrated, and purified by flash chromatography (silica
gel, CH2Cl2/MeOH (300:1 to 150:1)) to give compound 22 (905 mg,
54% overall yield). C21H22N6O4: colorless oil; TLC (EtOAc/hexane
(4:1)) Rf =0.55; ½aꢃ2D6 =+25.8 (EtOAc, c=1.0); IR (neat): n˜max =3373,
3282, 2990, 2925, 2853, 1679, 1618, 1512, 1465, 1376, 1331, 1295,
Binding assays for adenosine transporters were conducted as de-
scribed earlier.[31] Membrane fractions collected from the cerebral
cortex of Duncan Hartley derived guinea pigs were incubated with
[3H]NBTI (0.5 nm) for 30 min at 258C in an incubation buffer con-
taining 50 mm Tris-HCl (pH 7.4). Nonspecific binding was assessed
in the presence of 5 mm NBTI, a high-affinity inhibitor of ENT1,
which inhibits only ENT1 at 0.5 nm.[32] Reactions were terminated
by filtration over GF/B glass fibers and washing with the corre-
sponding reaction buffer.
1249, 1212, 1135, 1086 cmꢀ1 1H NMR (CDCl3, 400 MHz): d=8.39
;
(1H, brs), 7.64 (1H, brs), 7.26 (2H, d, J=10.4 Hz), 6.83 (2H, d, J=
10.4 Hz), 6.54 (1H, t, J=5.6 Hz), 6.13 (1H, s), 5.77 (1H, d, J=4.0 Hz),
5.68 (1H, dd, J=1.6, 4.0 Hz), 4.95 (1H, d, J=1.6 Hz), 4.75 (2H, brs),
3.79 (3H, s), 1.57 (3H, s), 1.42 ppm (3H, s); 13C NMR (CDCl3,
100 MHz): d=158.8, 154.5, 153.2, 148.2, 138.9, 130.2, 128.9 (2ꢂ),
119.7, 115.9, 114.5, 113.9 (2ꢂ), 91.6, 84.6, 83.9, 75.1, 55.3, 44.0, 26.6,
25.1 ppm; ESI-HRMS (negative mode) m/z [MꢀH]ꢀ calcd for
C21H22N6O4: 421.1624, found: 421.1612.
MTT metabolism assay. PC12 cells purchased from ATCC (Mana-
ssas, VA, USA) were maintained in Dulbecco’s modified Eagle’s
medium (DMEM) supplemented with 10% horse serum and 5%
fetal bovine serum and incubated in a CO2 incubator (5%) at 378C.
Survival was assessed by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphe-
nyltetrazolium bromide (MTT) metabolism assay as described else-
where.[33] In brief, cells grown on 150 mm plates were washed
three times with phosphate-buffered saline (PBS) and resuspended
in DMEM. Suspended cells (1ꢂ104 cells) were plated on 96-well
plates and treated with or without the indicated reagent. After in-
cubation for 24 h, MTT (0.5 mgmLꢀ1) was added to the medium
and incubated for 3 h. After discarding the medium, DMSO
(100 mL) was then applied to the well to dissolve the formazan
crystals derived from the mitochondrial cleavage of the tetrazolium
ring by live cells. The absorbance at l 570/630 nm in each well was
measured on a micro-enzyme-linked immunosorbent assay (ELISA)
reader.
A solution of nitrile 22 (905 mg, 2.14 mmol) and NH4Cl (429 mg,
8.04 mmol) in DMF (20 mL) was cooled to 08C, and NaN3 (523 mg,
8.04 mmol) was added. The ice bath was removed; the mixture
was heated at 408C for 1 h, slowly to 908C, and stirring was main-
tained at 908C for 9 h. The mixture was cooled, concentrated
under reduced pressure, dissolved in EtOAc, and extracted with
aqueous NaHCO3 (pH 8). The combined aqueous phase was acidi-
fied to pH 2 by the addition of aqueous HCl (1m), and extracted
with CH2Cl2. The organic layer was dried over MgSO4, filtered, and
concentrated to give a practically pure tetrazole product 16-aceto-
nide as a colorless oil (460 mg, 46% yield). C21H23N9O4: TLC (CH2Cl2/
MeOH (9:1)) Rf =0.25; ½aꢃ2D7 =+13.2 (EtOAc, c=1.0); IR (neat):
n˜max =3361, 2926, 2852, 1613, 1513, 1481, 1375, 1333, 1293, 1249,
1
1210, 1176, 1154, 1101, 1034 cmꢀ1; H NMR (CDCl3, 400 MHz): d=
7.90 (1H, brs), 7.68 (1H, brs), 7.35 (2H, d, J=8.4 Hz), 6.86 (2H, d,
J=8.4 Hz), 6.83 (1H, brs), 6.18 (1H, s), 5.85 (1H, s), 5.73 (1H, d, J=
6.0 Hz), 5.49 (1H, d, J=6.0 Hz), 4.92 (1H, dd, J=6.8, 7.6 Hz), 4.39
(1H, dd, J=4.0, 10.4 Hz), 3.77 (3H, s), 1.69 (3H, s), 1.43 ppm (3H,
s); 13C NMR (CDCl3, 100 MHz): d=158.4, 154.9, 152.9, 152.6, 146.2,
138.5, 129.5, 129.2 (2ꢂ), 118.4, 114.2, 113.7 (2ꢂ), 93.4, 85.9, 83.7,
82.3, 55.4, 44.1, 27.1, 25.2 ppm; ESI-HRMS (negative mode) m/z
[MꢀH]ꢀ calcd for C21H23N9O4: 464.1795, found: 464.1786.
Acknowledgements
We thank the National Science Council (NSC) of Taiwan for finan-
cial support. Y.C.C. was supported by grant NSC98-2321-B-001-
008. Y.L.L. was supported by grants NSC95-2323-B-077-002, NSC
96-2323-B-077-001, NSC 97-2323-B-077-001, and NSC 98-2323-B-
ChemMedChem 2011, 6, 1390 – 1400
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1399

J.-H. Lin, J.-M. Fang, et al.
MED
[13] P. Popoli, D. Blum, M. R. Domenici, S. Burnouf, Y. J. Chern, Curr. Pharm.
Des. 2008, 14, 1500–1511.
077-001. J.H.L. was supported by grants NSC 98-2627-B-001-002,
NSC 98-2323-B-002-001, and by the Research Center for Applied
Sciences, Academia Sinica. J.M.F was supported by grant NSC 97-
2628M-002-013-MY3.
[14] a) R. Morphy, Z. Rankovic, J. Med. Chem. 2006, 49, 4961–4970; b) R.
Morphy, Z. Rankovic, Curr. Pharm. Des. 2009, 15, 587–600.
[15] Y. Kurogi, O. F. Guner, Curr. Med. Chem. 2001, 8, 1035–1055.
[16] a) J. J. Kaminski, D. F. Rane, M. E. Snow, L. Weber, M. L. Rothofsky, S. D.
Anderson, S. L. Lin, J. Med. Chem. 1997, 40, 4103–4112; b) J. Singh, H.
van Vlijmen, Y. S. Liao, W. C. Lee, M. Cornebise, M. Harris, I. H. Shu, A.
Gill, J. H. Cuervo, W. M. Abraham, S. P. Adams, J. Med. Chem. 2002, 45,
2988–2993; c) E. Kotsikorou, E. Oldfield, J. Med. Chem. 2003, 46, 2932–
2944; d) G. S. Chen, C. S. Chang, W. M. Kan, C. L. Chang, K. C. Wang, J. W.
Chern, J. Med. Chem. 2001, 44, 3759–3763; e) D. Zampieri, M. G.
Mamolo, E. Laurini, C. Florio, C. Zanette, M. Fermeglia, P. Posocco, M. S.
Paneni, S. Pricl, L. Vio, J. Med. Chem. 2009, 52, 5380–5393; f) D. G. Wei,
X. L. Jiang, L. Zhou, J. Chen, Z. Chen, C. He, K. Yang, Y. Liu, J. F. Pei, L. H.
Lai, J. Med. Chem. 2008, 51, 7882–7888.
[17] F. Xu, H. Wu, V. Katritch, G. W. Han, K. A. Jackobson, Z.-G. Gao, V. Chere-
zov, R. C. Stevens, Science 2011, 332, 322–327.
[18] a) G. Walkinshaw, C. M. Waters, Neuroscience 1994, 63, 975–987; b) M. C.
Chiang, Y. C. Lee, C. L. Huang, Y. J. Chern, J. Biol. Chem. 2005, 280,
14331–14340.
[19] M. F. Jarvis, R. Schulz, A. J. Hutchison, U. H. Do, M. A. Sills, M. Williams, J.
Pharmacol. Exp. Ther. 1989, 251, 888–893.
[20] Catalyst, Accelrys, 9685 North Scranton Road, San Diego, CA 92121
(USA): http://www.accelrys.com (accessed May 17, 2011).
[21] C. E. Cass, L. A. Gaudette, A. R. Paterson, Biochim. Biophys. Acta Bio-
membr. 1974, 345, 1–10.
[22] Z. X. Zhu, J. K. Buolamwini, Bioorg. Med. Chem. 2008, 16, 3848–3865.
[23] S. L. Dixon, A. M. Smondyrev, E. H. Knoll, S. N. Rao, D. E. Shaw, R. A. Fries-
ner, J. Comput.-Aided Mol. Des. 2006, 20, 647–671.
[24] H. C. Kolb, M. G. Finn, K. B. Sharpless, Angew. Chem. 2001, 113, 2056;
Angew. Chem. Int. Ed. 2001, 40, 2004.
[25] J. B. Epp, T. S. Widlanski, J. Org. Chem. 1999, 64, 293–295.
[26] N. Nakajima, M. Saito, M. Ubukata, Tetrahedron 2002, 58, 3561–3577.
[27] M. P. Bosch, F. Campos, I. Niubo, G. Rosell, J. L. Diaz, J. Brea, M. I. Loza,
A. Guerrero, J. Med. Chem. 2004, 47, 4041–4053.
[28] a) P. L. Martin, R. J. Barrett, J. Linden, W. M. Abraham, Drug Dev. Res.
1997, 40, 313–324; b) R. Volpini, E. Camaioni, S. Costanzi, S. Vittori, K. N.
Klotz, G. Cristalli, Nucleosides Nucleotides 1999, 18, 2511–2520; c) R. Vol-
pini, S. Costanzi, C. Lambertucci, F. R. Portino, S. Taffi, S. Vittori, J. A. Za-
blocki, K. N. Klotz, G. Cristalli, ARKIVOC 2004, 301–311.
[29] O. Guner, O. Clement, Y. Kurogi, Curr. Med. Chem. 2004, 11, 2991–3005.
[30] K. Varani, S. Gessi, A. Dalpiaz, P. A. Borea, Br. J. Pharmacol. 1996, 117,
1693–1701.
[31] A. Verma, P. J. Marangos, Life Sci. 1985, 36, 283–290.
[32] J. L. Ward, A. Sherali, Z. P. Mo, C. M. Tse, J. Biol. Chem. 2000, 275, 8375–
8381.
[33] a) T. Mosmann, J. Immunol. Methods 1983, 65, 55–63; b) N. K. Huang,
Y. W. Lin, C. L. Huang, R. O. Messing, Y. Chern, J. Biol. Chem. 2001, 276,
13838–13846.
Keywords: adenosine signaling · designed multiple ligand ·
neuroprotection · pharmacophore analysis · structure–activity
relationships
[1] a) S. E. Andrew, Y. P. Goldberg, B. Kremer, H. Telenius, J. Theilmann, S.
Adam, E. Starr, F. Squitieri, B. Y. Lin, M. A. Kalchman, R. K. Graham, M. R.
Hayden, Nat. Genet. 1993, 4, 398–403; b) M. DiFiglia, E. Sapp, K. O.
Chase, S. W. Davies, G. P. Bates, J. P. Vonsattel, N. Aronin, Science 1997,
277, 1990–1993; c) M. E. Macdonald, C. M. Ambrose, M. P. Duyao, R. H.
Myers, C. Lin, L. Srinidhi, G. Barnes, S. A. Taylor, M. James, N. Groot, H.
Macfarlane, B. Jenkins, M. A. Anderson, N. S. Wexler, J. F. Gusella, G. P.
Bates, S. Baxendale, H. Hummerich, S. Kirby, M. North, S. Youngman, R.
Mott, G. Zehetner, Z. Sedlacek, A. Poustka, A. M. Frischauf, H. Lehrach,
A. J. Buckler, D. Church, L. Doucettestamm, M. C. Odonovan, L. Ribara-
mirez, M. Shah, V. P. Stanton, S. A. Strobel, K. M. Draths, J. L. Wales, P.
Dervan, D. E. Housman, M. Altherr, R. Shiang, L. Thompson, T. Fielder,
J. J. Wasmuth, D. Tagle, J. Valdes, L. Elmer, M. Allard, L. Castilla, M. Swar-
oop, K. Blanchard, F. S. Collins, R. Snell, T. Holloway, K. Gillespie, N.
Datson, D. Shaw, P. S. Harper, Cell 1993, 72, 971–983.
[2] a) M. F. Beal, R. J. Ferrante, Nat. Rev. Neurosci. 2004, 5, 373-384; b) S. I.
Okamoto, M. A. Pouladi, M. Talantova, D. D. Yao, P. Xia, D. E. Ehrnhoefer,
R. Zaidi, A. Clemente, M. Kaul, R. K. Graham, D. X. Zhang, H. S. V. Chen,
G. Tong, M. R. Hayden, S. A. Lipton, Nat. Med. 2009, 15, 1407–1408.
[3] S. Y. Chou, Y. C. Lee, H. M. Chen, M. C. Chiang, H. L. Lai, H. H. Chang, Y. C.
Wu, C. N. Sun, C. L. Chien, Y. S. Lin, S. C. Wang, Y. Y. Tung, C. Chang, Y. J.
Chern, J. Neurochem. 2005, 93, 310–320.
[4] M. C. Chiang, H. M. Chen, H. L. Lai, H. W. Chen, S. Y. Chou, C. M. Chen,
F. J. Tsai, Y. Chern, Hum. Mol. Genet. 2009, 18, 2929–2942.
[5] A. A. Link, T. Kino, J. A. Worth, J. L. McGuire, M. L. Crane, G. P. Chrousos,
R. L. Wilder, I. J. Elenkov, J. Immunol. 2000, 164, 436–442.
[6] N. K. Huang, Y. J. Chern, J. M. Fang, C. I. Lin, W. P. Chen, Y. L. Lin, J. Nat.
Prod. 2007, 70, 571–574.
[7] a) J. S. Fink, D. R. Weaver, S. A. Rivkees, R. A. Peterfreund, A. E. Pollack,
E. M. Adler, S. M. Reppert, Mol. Brain Res. 1992, 14, 186–195; b) A. K.
Dixon, A. K. Gubitz, D. J. S. Sirinathsinghji, P. J. Richardson, T. C. Freeman,
Br. J. Pharmacol. 1996, 118, 1461–1468; c) D. L. Rosin, A. Robeva, R. L.
Woodard, P. G. Guyenet, J. Linden, J. Comp. Neurol. 1998, 401, 163–186;
d) C. M. Anderson, W. Xiong, J. D. Geiger, J. D. Young, C. E. Cass, S. A.
Baldwin, F. E. Parkinson, J. Neurochem. 1999, 73, 867–873.
[8] E. Sapp, C. Schwarz, K. Chase, P. G. Bhide, A. B. Young, J. Penney, J. P.
Vonsattel, N. Aronin, M. DiFiglia, Ann. Neurol. 1997, 42, 604–612.
[9] S. Ohlson, Drug Discovery Today 2008, 13, 433–439.
[10] a) R. Morphy, C. Kay, Z. Rankovic, Drug Discovery Today 2004, 9, 641–
651; b) R. Morphy, Z. Rankovic, J. Med. Chem. 2005, 48, 6523–6543.
[11] G. R. Zimmermann, J. Lehar, C. T. Keith, Drug Discovery Today 2007, 12,
34–42.
Received: March 4, 2011
Revised: May 8, 2011
[12] A. L. Hopkins, Nat. Chem. Biol. 2008, 4, 682–690.
Published online on June 20, 2011
1400
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1390 – 1400