3D-QSAR studies of heterocyclic quinones with inhibitory activity on vascular smooth muscle cell proliferation using pharmacophore-based alignment

Article abstract of DOI:10.1016/j.bmc.2008.09.062

The abnormal proliferation and migration of vascular smooth muscle cells (SMCs) play an important role in the pathology of coronary artery atherosclerosis and restenosis following angioplasty. It was reported that some heterocyclic quinone derivatives such as 6-arylamino-quinoxaline-5,8-diones and 6-arylamino-1H-benzo[d]imidazole-4,7-diones have inhibitory activity on rat aortic smooth muscle cell (RAoSMC) proliferation. To understand the structural basis for antiproliferative activity to design more potent agents, we generated pharmacophore models of representative molecules with high activity using Genetic Algorithm with Linear Assignment of Hypermolecular Alignment of Database (GALAHAD) and aligned a series of compounds to the selected pharmacophore model, then performed three-dimensional quantitative structure-activity relationship (3D-QSAR) studies using Comparative Molecular Field Analysis (CoMFA) and Comparative Molecular Similarity Indices Analysis (CoMSIA). Good cross-validated correlations were obtained with CoMFA (resulting in q² of 0.734 and r² of 0.947) and CoMSIA (resulting in q² of 0.736 and r² of 0.913). The IC₅₀ values of the heterocyclic quinone derivatives on RAoSMC exhibited a strong correlation with steric and hydrophobic fields of the 3D structure of the molecules, resulting in the reliable prediction of inhibitory activity of the series of compounds.

Full text of DOI:10.1016/j.bmc.2008.09.062

Bioorganic & Medicinal Chemistry 16 (2008) 9772–9779
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
3D-QSAR studies of heterocyclic quinones with inhibitory activity on
vascular smooth muscle cell proliferation using pharmacophore-based
alignment
*
*
Chung-Kyu Ryu , Yoonji Lee, Seul-gi Park, Hea-Jung You, Ra-Young Lee, Seung-Yon Lee, Sun Choi
College of Pharmacy and Division of Life & Pharmaceutical Sciences, Ewha Womans University, Seoul 120-750, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 22 July 2008
Revised 26 September 2008
Accepted 26 September 2008
Available online 30 September 2008
The abnormal proliferation and migration of vascular smooth muscle cells (SMCs) play an important role
in the pathology of coronary artery atherosclerosis and restenosis following angioplasty. It was reported
that some heterocyclic quinone derivatives such as 6-arylamino-quinoxaline-5,8-diones and 6-arylami-
no-1H-benzo[d]imidazole-4,7-diones have inhibitory activity on rat aortic smooth muscle cell (RAoSMC)
proliferation. To understand the structural basis for antiproliferative activity to design more potent
agents, we generated pharmacophore models of representative molecules with high activity using
Genetic Algorithm with Linear Assignment of Hypermolecular Alignment of Database (GALAHAD) and
aligned a series of compounds to the selected pharmacophore model, then performed three-dimensional
quantitative structure–activity relationship (3D-QSAR) studies using Comparative Molecular Field Anal-
ysis (CoMFA) and Comparative Molecular Similarity Indices Analysis (CoMSIA).
Keywords:
Antiproliferative
Vascular smooth muscle cell
QSAR
CoMFA
CoMSIA
Good cross-validated correlations were obtained with CoMFA (resulting in q2 of 0.734 and r2 of 0.947)
and CoMSIA (resulting in q2 of 0.736 and r2 of 0.913). The IC50 values of the heterocyclic quinone deriv-
atives on RAoSMC exhibited a strong correlation with steric and hydrophobic fields of the 3D structure of
the molecules, resulting in the reliable prediction of inhibitory activity of the series of compounds.
Ó 2008 Elsevier Ltd. All rights reserved.
Quinoxaline-5,8-dione
1H-benzo[d]imidazole-4,7-dione
1. Introduction
2-(2-pyridyl)-1H-benzo[d]imidazole-4,7-diones 57 were evaluated
in vitro for their antiproliferative activity on the SMCs.
The abnormal proliferation and migration of vascular smooth
muscle cells (SMCs) play a pivotal role in progression of coronary
artery atherosclerosis and restenosis following angioplasty.1 Arte-
rial injury results in the migration of SMCs into the intimal layer
of the arterial wall, where they proliferate and synthesize extracel-
lular matrix components. Heterocyclic quinonoid compounds are
an attractive class of biologically active molecules.2 Therefore, we
synthesized and tested various quinone derivatives to elucidate
their contribution to the antiproliferative effects on PDGF-stimu-
lated SMC proliferation. Among the tested quinones, 1H-
benzo[d]imidazole-4,7-dione and quinoxaline-5,8-dione deriva-
tives 1–10 showed potent antiproliferative activity (Fig. 1).
It was reported that some quinonoid derivatives such as 1H-
benzo[d]imidazole-4,7-diones 1–4,3,4 6–8,5 and quinoxaline-5,8-
diones 96 have their antiproliferative activity on rat aortic smooth
muscle cells. We newly synthesized quinoxaline-5,8-diones 10 and
tested them to elucidate their contribution to the antiproliferative
effects on PDGF-stimulated SMC proliferation. Also, 6-arylamino-
To understand the structural basis for antiproliferative activity
to design more potent agents, we performed three-dimensional
quantitative structure–activity relationship (3D-QSAR) studies for
a series of compounds using Comparative Molecular Field Analysis
(CoMFA) and Comparative Molecular Similarity Indices Analysis
(CoMSIA). First, we generated pharmacophore model of represen-
tative molecules with high activity using Genetic Algorithm with
Linear Assignment of Hypermolecular Alignment of Database
(GALAHAD) and aligned all the molecules to the model. Then,
3D-QSAR analyses were performed. The antiproliferative activities
of heterocyclic quinone derivatives on rat aortic smooth muscle
cell (RAoSMC) exhibited a strong relationship with the molecular
interaction fields from the 3D-QSAR studies, resulting in reliable
prediction of inhibitory activity of a series of compounds.
2. Chemistry and biological activities
The IC50 values of compounds, 6-arylamino-1H-benzo[d]imidaz-
ole-4,7-diones 1–4,3,4 6-arylamino-2-(2-pyridyl)-1H-benzo[d]imid-
azole-4,7-diones 5, 6-arylamino-5-chloro-2-(2-pyridyl)-1H-benzo-
[d]imidazole-4,7-diones 6–8,5 6-arylamino-2,3-bis(pyridin-2-yl)-
7-chloro-quinoxaline-5,8-diones 9,6 and 6-arylamino-2,3-bis
(methyl)-quinoxaline-5,8-diones 10 were used for this analysis
* Corresponding authors. Tel.: +82 2 3277 3027; fax: +82 2 3277 3051 (C.-K.R.);
tel.: +82 2 3277 4503; fax: +82 2 3277 2851 (S.C.).
E-mail addresses: ckryu@mm.ewha.ac.kr (C.-K. Ryu), sunchoi@ewha.ac.kr (S.
Choi).
0968-0896/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2008.09.062

C.-K. Ryu et al. / Bioorg. Med. Chem. 16 (2008) 9772–9779
9773
according to theknownmethod8 withminormodifications. 2,3-Bis(-
methyl)-quinoxaline-5,8-dione (12) was synthesized by oxidizing
compound 11 with Ag2O in 79% yields. 6-Arylamino-2,3-bis(-
methyl)-quinoxaline-5,8-diones 10a–f were prepared by substitu-
tion on compound 12 with appropriate arylamines. Most of these
substitutions went as expected.
The compounds 5a–e were prepared by nucleophilic substitu-
tion on each 2-(2-pyridyl)-4,7-1H-benzo[d]imidazole-4,7-dione
13 with arylamines according to the reported method7 to evaluate
the antiproliferative effects on PDGF-stimulated SMC proliferation.
The newly prepared compounds 5a–e and 10a–f were evalu-
ated in vitro for their antiproliferative activity on the RAoSMCs.
Inhibition of proliferation of these cells was determined by pub-
lished assay method6 and the IC50 values were determined and
compared to the positive control mycophenolic acid (MPA). Their
structures and activities are summarized in Tables 1 and 2.
O
O
R1
O
R1
H
N
H
N
H
N
H
N
R2
R3
R2
R3
R4
W
N
N
Y Z
X
X
O
R1, R2, R3 = H, F, Cl ..
5a-e: W=Y=CH, Z=N, X=H
6a-f: W=Y=CH, Z=N, X=Cl
7a-j: Y=Z=CH, W=N, X=Cl
R1, R2, R3 = H, F, Cl ...
1a-n: R4=CH3, X=Cl
2a-f: R4=CF3, X=Cl
3a-n: R4=H, X=Cl
8a-d: W=Z=CH, Y=N, X=Cl
4a-i: R4=H, X=Br
R1
R4
N
N
O
O
R1
R4
O
O
H
N
H
N
R2
R3
N
N
R2
R3
H3C
H3C
N
N
Cl
R1, R2, R3, R4 = H, F, Cl ...
R1, R2, R3, R4 = H, F, Cl ...
3. 3D-QSAR modeling
9a-h
10a-f
The training (65 compounds) and test sets (17 compounds)
used in these studies comprise a series of the 6-phenylamino-
1H-benzo[d]imidazole-4,7-diones and 6-phenylamino-quinoxa-
line-5,8-diones. The IC50 values were converted to pIC50 (i.e.,
ÀlogIC50) and used as dependent variables in CoMFA and CoMSIA
QSAR analyses. The distribution of biological activities of training
set and test set versus % of compounds is shown in Figure 2.
Figure 1. Quinoxaline-5,8-dione and 1H-benzo[d]imidazole-4,7-dione derivatives.
(Tables 1 and 2). Among them, compounds 10 were newly synthe-
sized and tested to elucidate their contribution to the antiprolifera-
tive effects on PDGF-stimulated SMC proliferation. The method used
to synthesize the quinoxaline-5,8-diones 10a–f is shown in Scheme
1. 2,3-bis(Methyl)-5,8-hydroxyquinoxaline (11) was prepared
Table 1
Structures and IC50 values of the 6-phenylamino-1H-benzo[d]imidazole-4,7-diones for inhibition of SMC proliferation
O
R1
H
N
H
N
R2
R3
R4
N
X
O
b
b
Compound
X
R1
R2
R3
R4
IC50
(lM)
Compound
X
R1
R2
R3
R4
IC50 (lM)
1a3
1b4
1ca,4
1d3
1e3
1fa,4
1g4
1h4
1i4
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
H
H
H
H
H
H
H
H
H
H
F
H
H
H
H
H
H
H
H
Cl
Br
H
H
Cl
F
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Cl
Br
I
H
F
Cl
I
CH3
CH2CH3
CF3
OCF3
H
H
F
Cl
H
F
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CF3
CF3
CF3
CF3
CF3
CF3
H
H
H
H
H
H
H
H
H
H
H
H
H
0.8
3.0
2.8
1.0
0.8
3.0
0.6
3.0
0.9
1.0
1.2
0.8
0.6
0.8
50.0
12.0
15.0
25.0
40.0
20.0
5.5
1.2
2.0
1.3
1.4
2.5
2.5
2.7
1.4
1.6
1.3
1.1
1.7
2.7
4a4
4ba,4
4c4
4d3
4e3
4f
Br
Br
Br
Br
Br
Br
Br
Br
Br
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
F
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Cl
Br
I
CF3
H
H
H
H
H
H
H
H
Cl
Br
H
H
H
H
H
F
H
F
Cl
CH3
CF3
H
H
H
H
Cl
Br
I
OCH2CH3
OCF3
Cl
Br
OCF3
H
H
H
H
H
H
H
H
H
H
H
1.1
1.0
4.0
1.0
3.1
3.0
5.3
2.5
3.0
13
12
20
6.5
20
12
6.5
6
3.5
20
9.4
50
60
50
75
19
31
20
60
25
13
21
20
45
42
4g4
4h4
4i4
H
1j3
5a5
5b5
5c5
5d5
5e5
6a5
6b5
6c5
6d5
6e5
6f
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
p-Pyridine
p-Pyridine
p-Pyridine
p-Pyridine
p-Pyridine
p-Pyridine
p-Pyridine
p-Pyridine
p-Pyridine
p-Pyridine
m-Pyridine
m-Pyridine
m-Pyridine
m-Pyridine
1k4
1l3
H
H
H
H
Cl
Cl
F
1m4
1n4
2a3
2b3
2ca,3
2d3
2e3
2f3
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
F
H
F
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
I
CH3
CF3
H
F
Br
I
CH3
CF3
OCF3
CN
H
3aa3
3b4
3c4
7a5
7b5
7ca,5
7d5
7e5
7f
F
Br
OCH3
OCH2CH3
H
H
H
H
Cl
Cl
3da,3
3e3
3fa,3
3g4
3h4
3i4
Cl
Br
CH3
F
Cl
H
H
H
H
7ga,5
7h5
7ia,5
7j5
3j3
H
H
F
H
3k3
3l4
8a5
8b5
8ca,5
8d5
F
Cl
F
H
H
3ma,4
3n4
CeHo
OCH2CH3
F
F
H
a
Compounds in test set.
b
The inhibitory activity against the PDGF-induced proliferation of the SMCs isolated from rat thoracic aorta.

9774
C.-K. Ryu et al. / Bioorg. Med. Chem. 16 (2008) 9772–9779
Table 2
Structures and IC50 values of the 6-phenylamino-quinoxaline-5,8-diones for inhibition of SMC proliferation
R1
O
H
N
R2
R3
R5
R6
N
N
X
O
R4
b
Compound
X
R1
R2
R3
R4
R5
R6
IC50 (lM)
9a6
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Cl
F
H
H
H
H
F
H
F
Cl
OH
OCH3
OCF3
Cl
F
F
H
H
H
H
H
H
H
F
H
H
H
H
F
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
CH3
CH3
CH3
CH3
CH3
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
o-Pyridine
CH3
CH3
CH3
CH3
CH3
3.1
1.0
1.5
5.5
3.5
1.0
1.0
1.2
3.1
3.3
1.1
1.5
4.5
3.8
9b6
9ca,6
9da,6
9ea,6
9f6
9g6
9h6
10aa
10b
10c
10d
10e
10f
Cl
Br
OCH2CH3
F
Cl
H
Cl
H
CH3
CH3
a
Compounds in test set.
b
The inhibitory activity against the PDGF-induced proliferation of the SMCs isolated from rat thoracic aorta.
R1
R4
O
N
O
O
OH
OH
H
N
R2
R3
H3C
H3C
H3C
H3C
N
N
H3C
H3C
N
N
b
a
N
O
12
11
1
2
3
4
R , R , R , R = H, F, Cl ...
10a-f
O
O
R1
O
O
H
N
H
N
R2
R3
H
N
c
N
N
N
N
13
1
2
3
R , R , R = H, F, Cl ..
5a-e
Scheme 1. Synthesis of 6-arylamino-2,3-bis(methyl)-quinoxaline-5,8-diones and 2-aryl-6-arylamino-1H-benzo[d]imidazole-4,7-diones. Reagents and conditions: (a) Ag2O/
rt/4 h/79%; (b) arylamine (1 equiv)/EtOH/reflux/10 h/31–53%; (c) arylamine (1 equiv)/EtOH/reflux/5 h/66–84%.
1m, 1n, 1g, 5d, 6d, 9g, and 10c), and all the other molecules were
aligned to the best pharmacophore model. The alignment of the
molecules is shown in Figure 3.
50
40
30
20
10
0
Two different 3D-QSAR methods, CoMFA and CoMSIA were ap-
plied to find a correlation between molecular property fields of the
aligned molecules and their activities. Leave-one-out (LOO) cross-
validated PLS analyses were performed to determine the optimal
number of components to be used in the final QSAR models and
to check the predictive ability of the models. The resulting contour
maps from the CoMFA and CoMSIA models showed the steric,
hydrophobic, and hydrogen-bond acceptor requirements of the
series of compounds for inhibitory activity and these findings
would be used for designing more potent inhibitors against SMC
proliferation.
pIC50
Figure 2. Distribution of biological activities for training set and test set versus % of
compounds. (White bars: training set; black bars: test set.)
Since the major weakness of the traditional 3D-QSAR has been
the alignment of the molecules, we applied pharmacophore mod-
eling using GALAHAD in order to get the more reasonable align-
ment results. First, the representative molecules with high
activities were selected among the training set. Then, pharmaco-
phore models were generated using the selected molecules (i.e.,
4. Results and discussion
4.1. CoMFA and CoMSIA statistical results
The statistical results for 3D-QSAR modeling are summarized in
Table 3. PLS analyses gave cross-validated q2 of 0.734 for CoMFA,

C.-K. Ryu et al. / Bioorg. Med. Chem. 16 (2008) 9772–9779
9775
Figure 3. Selected pharmacophore model and molecular alignments using that model. Cyan and green spheres are represented for hydrophobes and hydrogen-bond
acceptors, respectively.
For comparison, we also performed 3D-QSAR studies based on
atom-by-atom alignment. The resulting q2 and r2 values were sim-
ilar to those of pharmacophore-alignment based QSAR studies;
q2 = 0.748 and r2 = 0.942 for CoMFA, q2 = 0.743 and r2 = 0.907 for
CoMSIA. But in this case, the predictability of the test set (i.e.,
r2pred) was decreased to 0.704 for CoMFA and 0.700 for CoMSIA.
Although atom-by-atom alignment would result in more per-
fect fit in a congeneric series of compounds, whether that align-
ment represents what is happening inside the biological system
would be questionable. As shown in some reported co-crystallized
structures of a congeneric series of compounds, the alignment of
active conformers do not necessarily display perfect match as
shown in atom-by-atom alignment. That might be because each
compound would change its conformation, even in congeneric
part, to achieve the best activity in the biological system. There-
fore, using pharmacophore-based alignment could be a reasonable
way for more realistic 3D-QSAR studies when the active conforma-
tions of tested compounds are not available as in crystal structures.
Table 3
Summary of CoMFA and CoMSIA results
CoMFA
CoMSIA
q2a
0.734
0.947
0.147
174.153
0.814
65
0.736
0.913
0.189
101.639
0.793
65
r2b
SEEc
F value
d
r2
pred
No. of compounds
NOCe
6
6
Contribution
Steric
Electrostatic
Hydrophobic
H-bond donor
H-bond acceptor
0.586
0.414
—
—
—
0.130
0.447
0.233
0.060
0.130
a
Cross-validated correlation coefficient after leave-one-out procedure.
Correlation coefficient.
Standard error of estimate.
Predicted correlation coefficient for test set of compounds.
Optimal number of components.
b
c
d
e
4.2. CoMFA contour maps
and q2 of 0.736 for CoMSIA with six components in both models.
Figure 4 shows the strong correlations between the actual versus
predicted pIC50 values for 65 compounds in the training set by
CoMFA and CoMSIA models. The correlation coefficients for CoMFA
and CoMSIA were 0.947 and 0.913, respectively.
A set of 17 compounds, not included in the model generation,
was prepared to determine the predictive abilities of the CoMFA
and CoMSIA models. Both models well predicted the activities of
the test set compounds. The predicted r2 values from the CoMFA
and CoMSIA models were found to be 0.814 and 0.793, respec-
tively. The comparison between the actual versus predicted pIC50
values for the test set based on the CoMFA and CoMSIA models
are presented in Table 4 and Figure 5. These data indicate that reli-
able CoMFA and CoMSIA models were successfully constructed.
According to the 3D-QSAR results, the contour maps were cre-
ated using the data from PLS analyses. These maps help us to ex-
plain the chemical property features of the compounds which
influence the biological activities.
Figure 6(a) and (b) shows the steric and electrostatic maps for
the CoMFA model on compounds 1g and 9g, which are the most ac-
tive compounds in each series, respectively. In the steric CoMFA
map, the green contour indicates the areas in which steric bulk
on an inhibitor is favored, and the yellow contour represents the
regions in which steric bulk is disfavored to overall inhibitory
activity. The electrostatic contour maps are shown as red and blue
contours in Figure 6(c) and (d). The red contour means the regions
in which the electronegative groups in the ligands are associated
with increased antiproliferative activities, while the blue contour
6.5
6.0
5.5
6.5
6.0
5.5
5.0
5.0
y = 0.9474x + 0.2829
y = 0.9133x +0.4663
R2 = 0.9474
4.5
R2 = 0.9132
4.5
4.0
4.0
4.0
4.0
4.5
5.0
5.5
6.0
6.5
4.5
5.0
5.5
6.0
6.5
Actual pIC50
Actual pIC50
Figure 4. Plots of predicted versus actual pIC50 values for the training set molecules based on (a) CoMFA and (b) CoMSIA models.

9776
C.-K. Ryu et al. / Bioorg. Med. Chem. 16 (2008) 9772–9779
Table 4
hand, R5 position of the 6-phenylamino-quinoxaline-5,8-dione ser-
ies may be favorable in bulky groups. As in the case of compounds
5a–d, 6a–f, 7a–i, and 8a–d, the substitution of bulky pyridine ring
on the R4 position decreases the overall biological activity. On the
other hand, in case of compound 9a–9h, the substitution of pyri-
dine ring on R5 position does not negatively affect their activities.
Observed and predicted activities for the test set by CoMFA and CoMSIA models
Compound
pIC50
CoMFA
Residual
CoMSIA
Residual
Predict
Predict
1c4
1f4
5.55
5.52
4.82
5.26
5.89
5.60
5.77
6.00
4.92
4.30
4.70
4.60
4.35
5.82
5.26
5.46
5.51
5.913
5.993
4.749
5.676
5.616
5.629
5.609
5.588
4.938
4.164
4.554
4.654
4.630
5.817
5.229
5.693
5.609
À0.36
À0.47
0.07
5.945
6.068
4.521
5.648
5.776
5.627
5.692
5.778
5.014
4.463
4.289
4.571
4.651
5.750
4.964
5.738
5.604
À0.39
À0.55
0.30
2c3
3a3
3d3
3f3
À0.42
À0.39
4.3. CoMSIA contour maps
0.27
0.11
À0.03
0.16
0.41
À0.03
0.08
0.22
À0.09
À0.16
0.41
The steric and electrostatic maps of CoMSIA are quite similar to
those of CoMFA. Figure 7(a) and (b) show the steric maps for CoM-
SIA model for compounds 1g and 9g, respectively. According to
CoMSIA steric fields, the sterically bulky groups on R4 position of
the 6-phenylamino-1H-benzo[d]imidazole-4,7-dione series have
a negative effect on the activity, as shown in the CoMFA steric
map. On the other hand, R5 and R6 position of the 6-phenylami-
no-quinoxaline-5,8-dione series may be favorable in bulky groups.
The hydrogen-bond acceptor map of the CoMSIA model on com-
pound 9g is presented in Figure 7(e). The magenta and green-blue
indicate the areas occupied by hydrogen-bond acceptor groups
exerting favorable and unfavorable effects on SMC antiprolifera-
tion, respectively. A large magenta polyhedron, around the R4 posi-
tion of the 6-phenylamino-1H-benzo[d]imidazole-4,7-dione series
and R5 or R6 positions of the 6-phenylamino-quinoxaline-5,8-dione
series, indicates the location of hydrogen-bond acceptor groups
having an enhancing effect on SMC antiproliferation. This result
3m4
4b4
6f5
À0.02
7c5
7g5
7i5
0.14
0.15
À0.05
À0.28
0.00
0.03
8c5
9c6
9d6
9e6
10a
À0.30
0.07
0.03
0.30
À0.23
À0.28
À0.10
À0.09
displays the areas that positively charged groups have improved
biological activity and thus lower the IC50 values.
According to the CoMFA contour maps, the sterically bulky
groups on R4 position of the 6-phenylamino-1H-benzo[d]imidaz-
ole-4,7-dione series would be detrimental to activity. On the other
6.5
6.0
5.5
6.5
6.0
5.5
5.0
5.0
y = 0.9359x + 0.3798
y = 0.99x + 0.098
R2 = 0.7925
R2 = 0.8138
4.5
4.5
4.0
4.0
4.0
4.5
5.0
5.5
6.0
6.5
4.0
4.5
5.0
5.5
6.0
6.5
Actual pIC50
Actual pIC50
Figure 5. Plots of predicted versus actual pIC50 values for the test set molecules based on (a) CoMFA and (b) CoMSIA models.
Figure 6. CoMFA contour maps for steric and electrostatic fields. (a) Steric map displayed with compound 1g; (b) steric map with compound 9g; (c) electrostatic map with
compound 1g; (d) electrostatic map with compound 9g. Sterics: green (favored)/yellow (disfavored); electrostatics: blue (positive-charge favored)/red (negative-charge
favored).

C.-K. Ryu et al. / Bioorg. Med. Chem. 16 (2008) 9772–9779
9777
Figure 7. CoMSIA contour maps for steric, hydrophobic and H-bond acceptor fields. (a) Steric map displayed with compound 1g; (b) steric map with compound 9g; (c)
hydrophobic map with compound 1g; (d) hydrophobic map with compound 9g; (e) H-bond acceptor map with compound 9g. Sterics: green (favored)/yellow (disfavored);
hydrophobic: brown (favored)/white (disfavored); hydrogen bonding acceptor: magenta (favored)/green-blue (disfavored).
is consistent with the fact that compounds with o-pyridine substi-
tuent in those positions showed much better activities than those
with m- or p-pyridine.
with electron-donating and less hydrophobic groups in R4 position
is in progress.
The hydrophobic maps of the CoMSIA model on compounds 1g
and 9g are shown in Figure 7(c) and (d), respectively. The brown
and white represents the areas where hydrophobic properties are
favored and disfavored, respectively. R4 position of the 6-phenyla-
mino-1H-benzo[d]imidazole-4,7-dione series is hydrophobic-fa-
vored region, thus the compounds substituted with the
hydrophobic groups on that position were more potent than
unsubstituted compounds. The hydrophobic-favored region in R4
position coincided with the fact that the methyl-substituted com-
pound 1 series was more active than non-substituted compounds.
Although trifluoromethyl group is more hydrophobic than methyl
group, it is an electron-withdrawing group and that might nega-
tively contribute to the biological activity.
To compare the deviation of electron density (of CF3- and CH3-
containing compounds), we calculated their atomic charges by
Gasteiger–Hückel method. We found that the electron density of
the whole molecule changed when it was substituted with CF3 in
R4 position. As Figure 8 shows, electron density in the quinone ring
decreased after substituted with CF3 in compound 2a. The quinone
ring is thought to be critical to the biological activity, and the
changes of its electron density might have caused the decrease of
the activity.
5. Conclusions
A
series of the 6-phenylamino-1H-benzo[d]imidazole-4,7-
diones and 6-phenylamino-quinoxaline-5,8-diones were synthe-
sized and their 3D-QSAR studies were performed, resulting in the
good cross-validated correlations obtained by CoMFA (q2 of 0.734
and r2 of 0.947) and CoMSIA (q2 of 0.736 and r2 of 0.913). The pre-
dictive ability of CoMFA and CoMSIA, determined by a test set, gave
good predictive correlation coefficients of 0.814 and 0.793,
respectively.
The steric, hydrophobic, and hydrogen-bond acceptor require-
ments for the inhibitory activity are well described from the con-
tour maps of these studies. It appears that the pharmacophore-
based alignment provided reasonable superimposition of the mol-
ecules for our 3D-QSAR studies. Along with the pharmacophore
model, more precise pharmacophore features were suggested by
the contour maps, and these results would be further utilized to
design more potent inhibitors of smooth muscle cell proliferation.
6. Experimental
All melting points were measured in open capillary tubes with a
Büchi melting point B-545 and were uncorrected. TLC was per-
formed on precoated silica gel (60G 254, Merck) using chloroform
as a solvent. The compounds were detected under UV light
(254 nm) or by heating to 110 °C after spraying with a 30%
H2SO4–vanillin solution. Column chromatography was performed
on silica gel G60 (70–230 mesh, ASTM, Merck). 1H NMR spectra
were recorded on Unity Varian INOVA 400 MHz FT-NMR spectrom-
According to the electrostatic contour map (Fig. 6(c) and (d)),
the quinone ring is (À)-charge favored region for better activity.
It is consistent with the result that compound 2a (with the elec-
tron-withdrawing CF3 group) is less active compared with others.
The low activity of CF3 substituted compounds may also be influ-
enced by the decrease of H-bonding ability of hetero atoms in
the quinone ring. To clarify this point, synthesis of the compounds

9778
C.-K. Ryu et al. / Bioorg. Med. Chem. 16 (2008) 9772–9779
Figure 8. Molecular surfaces of compound 1a, 2a, and 3a colored by electrostatic potential (EP). Van der Waals surfaces were generated by MOLCAD and displayed in
transparent style. The electrostatic potential at each surface was computed with a global range for the property from red (most positive) to purple (most negative), so
identical colors on the surfaces of different molecules reflect the identical property values.
eter using DMSO-d6 as a solvent, and chemical shifts were given in
ppm with TMS as a standard. High-resolution mass spectra (HRMS
EI) were obtained on a Jeol JMS AX505 WA. 2,3-Bis(methyl)-5,8-
hydroxy-5,8-dihydroquinoxaline (11) was prepared according to
the known method.8 Other reagents were purchased from Aldrich
Chemical Co.
6H, CH3). MS (m/z): 313 (M+), 298, 278, 236, 199, 97, 84, 57. HRMS
(m/z): Calcd for C16H12O2N3Cl 313.0618 Found: 313.0615.
6.2.3. 6-(4-Bromophenyl)amino-2,3-dimethyl-5,8-quinoxalin-
edione (10c)
Dark red powder (37%). mp: 255–257 °C. 1H NMR (DMSO-d6): d
9.4 (s, 1H, NH), 7.4–7.6 (m, 4H, benzene), 6.3 (s, 1H, H7), 2.6 (s, 6H,
2CH3). MS (m/z): 357 (M+), 278, 222, 108, 96, 83, 55. HRMS (m/z):
Calcd for C16H12O2N3Br 357.0110 Found: 357.0113.
6.1. Synthesis of the 2,3-bis(methyl)-quinoxaline-5,8-dione
(12)
Compounds 11 1.26 g (6.6 mmol) was suspended in 500 mL of
1,2-diethoxyethane and treated Ag2O 1.9 g (7.7 mmol) for 4 h at
rt in the dark. The solvent was removed under vacuum, and the
crude product was purified by silica gel column chromatography
with CHCl3. Crystallization from aq. EtOH afforded compound 12;
79%, mp 179–181 °C (lit.8 mp 182 °C dec).
6.2.4. 2,3-Dimethyl-6-(4-ethoxyphenyl)amino-5,8-quinoxalin-
edione (10d)
Dark red needle (35%). mp: 217–219 °C. 1H NMR (DMSO-d6): d
9.3 (s, 1H, NH), 6.9–7.2 (m, 4H, benzene), 6.0 (s, 1H, H7), 4.1 (q,
2H, J = 7.2, CH2), 2.6 (s, 6H, 2CH3), 1.3 (t, 3H, J = 7.2, CH3). HRMS
(m/z): Calcd for C18H17N3O3 323.1270 Found: 323.1295.
6.2. General procedure for synthesis of the 6-arylamino-2,3-
bis(methyl)-quinoxaline-5,8-diones 10
6.2.5. 6-(3,5-Difluorophenyl)amino-2,3-dimethyl-5,8-quinoxal-
inedione (10e)
Dark red powder (31%). mp: 234–236 °C. 1H NMR (DMSO-d6): d
9.5 (s, 1H, NH), 7.16–7.18 (m, 1H, benzene), 7.03–7.08 (m, 2H, ben-
zene), 6.4 (s, 1H, H7), 2.5 (s, 6H, 2CH3), HRMS (m/z): Calcd for
C16H11F2N3O2 315.0819 Found: 315.0814.
A solution of compound 12 0.063 g (0.34 mmol) in 15 mL of 95%
EtOH was added to the solution of the arylamine (0.35 mmol) in
5 mL of 95% EtOH and stirred at rt for 1 h and then refluxed for
10 h. After the mixture was kept overnight in the refrigerator or
poured into 20 mL of ice water, the precipitate was collected by fil-
tration. The crude product was purified by silica gel column chro-
matography with CHCl3 and MeOH. Crystallization from aq. EtOH
afforded the compounds 10a–f.
6.2.6. 6-(3,4-Dichlorophenyl)amino-2,3-dimethyl-5,8-quinox-
alinedione (10f)
Dark red powder (51%). mp: 213 °C (dec). 1H NMR (DMSO-d6): d
9.5 (s, 1H, NH), 7.68–7.69 (m, 2H, benzene), 7.42–7.45 (m, 2H, ben-
zene), 6.2 (s, 1H, H7), 2.6 (s, 6H, 2CH3), HRMS (m/z): Calcd for
C16H11Cl2N3O2 347.0229 Found: 347.0225.
6.2.1. 2,3-Dimethyl-6-(4-fluorophenyl)amino-5,8-quinoxal-
inedione (10a)
Dark purple powder (53%). mp: 248–249 °C 1H NMR (DMSO-
d6): d 9.4 (s, 1H, NH), 7.2–7.4 (m, 4H, benzene), 6.0 (s, 1H, H7),
2.7 (s, 6H, 2CH3). MS (m/z): 297 (M+), 282, 278, 136, 162, 96, 84,
57. HRMS (m/z): Calcd for C16H12O2N3F 297.0914 Found: 297.0914.
6.3. Molecular modeling
All computational studies were performed using Tripos SYBYL
version 8.0.3 software package based on a Linux (RHEL 4.0 Intel
Xeon processor 5050) workstation. The structures of the tested
compounds were built with Concord and energy minimized using
the Tripos force field and Gasteiger–Hückel charges (method: pow-
ell, termination: gradient 0.05 kcal/mol Å, and max iterations:
1,000,000) implemented in the SYBYL.
6.2.2. 6-(4-Chlorophenyl)amino-2,3-dimethyl-5,8-quinoxalin-
edione (10b)
Dark viloet powder (41%). mp: 253–254 °C 1H NMR (DMSO-d6):
d 9.4 (s, 1H, NH), 7.4–7.5 (m, 4H, benzene), 6.2 (s, 1H, H7), 2.6 (s,

C.-K. Ryu et al. / Bioorg. Med. Chem. 16 (2008) 9772–9779
9779
6.3.1. Molecular alignment
Acknowledgments
Pharmacophore models of the representative molecules with
high activities were generated using Genetic Algorithm with Linear
Assignment of Hypermolecular Alignment of Database (GALAHAD)
and a series of molecules was aligned to the selected pharmaco-
phore model for 3D-QSAR studies.
This work was supported by the second stage of the Brain
Korea 21 Project and grants from the Basic Research Program
(R01-2006-000-10020-0) of the Korea Science and Engineering
Foundation (KOSEF) to C.-K. Ryu and the National Core Research
Center (NCRC) program (R15-2006-020) of the Ministry of Edu-
cation, Science and Technology (MEST) and KOSEF through the
Center for Cell Signaling & Drug Discovery Research at Ewha
Womans University and the Ewha Womans University Research
Grant of 2006 to S. Choi.
6.3.2. 3D-QSAR studies using CoMFA and CoMSIA
The steric and electrostatic field energies for CoMFA were calcu-
lated using the SYBYL default parameters: 2.0 Å grid points spac-
ing, an sp3 carbon probe atom with +1 charge and a van der
Waals radius of 1.52 Å, and column filtering of 2.0 kcal/mol. In
CoMSIA, the steric, electrostatic, hydrophobic, hydrogen-bond do-
nor and acceptor potential fields were calculated at each lattice
intersection with the SYBYL default parameters and the attenua-
tion factor of 0.3 was used.
References and notes
1. Ross, R. Nature 1993, 362, 801.
2. Middleton, R. W.; Parrick, J. In The Chemistry of the Quinonoid Compounds;
Patai, S., Rappoport, Z., Eds.; John Wiley & Sons: London, 1988; pp 1019–
1066. Part 2.
3. Hong, S.-Y.; Chung, K.-H.; You, H.-J.; Choi, I. H.; Chae, M. J.; Han, J.-Y.; Jung, O.-J.;
Kang, S.-J.; Ryu, C.-K. Bioorg. Med. Chem. Lett. 2004, 14, 3563.
4. Chung, K.-H.; Hong, S.-Y.; You, H.-J.; Park, R.-E.; Ryu, C.-K. Bioorg. Med. Chem.
2006, 14, 5795.
5. Ryu, C.-K.; Lee, R-. Y.; Lee, S.-Y.; Chung, H.-J.; Lee, S. K.; Chung, K.-H. Bioorg. Med.
Chem. Lett. 2008, 18, 2948.
6. Chung, H.-J.; Jung, O.-J.; Chae, M. J.; Hong, S.-Y.; Chung, K.-H.; Lee, S. K.; Ryu, C.-
K. Bioorg. Med. Chem. Lett. 2005, 15, 3380.
The CoMFA and CoMSIA descriptors were used as independent
variables, and pIC50 values were used as dependent variables in par-
tial least-squares (PLS) regression analyses to derive 3D-QSAR mod-
els. Theoptimumnumberof principalcomponentsisobtainedbythe
leave-one-out (LOO) cross-validation procedure. Using the optimal
number of principal components, the final PLS analyses were carried
outtogeneratethepredictiveQSARmodelswithoutcross-validation
process. To visualize the 3D-QSAR results in term of field
contributions, isocontour maps were generated using the field type
7. Ryu, C. -K.; Song, E. H.; Shim, J.-Y.; You, H.-J.; Choi, K. U.; Choi, I. H.; Lee, E. Y.;
Chae, M. J. Bioorg. Med. Chem. Lett. 2003, 13, 17.
8. Gum, W. F.; Joullie, M. M. J. Org. Chem. 1967, 32, 53.
*
‘stdev coeff’ and the contour levels were set to default values.