Antiproliferative activity of chalcones with basic functionalities

Article abstract of DOI:10.1016/j.bmc.2007.07.042

A library of chalcones with different basic groups were synthesized and evaluated for antiproliferative activities against the human breast cancer (MCF 7) and colon cancer (HCT 116) cell lines. Structure-activity relationships were analyzed by projection

Full text of DOI:10.1016/j.bmc.2007.07.042

Bioorganic & Medicinal Chemistry 15 (2007) 7021–7034
Antiproliferative activity of chalcones with basic functionalities
Xiaoling Liu and Mei-Lin Go*
Department of Pharmacy, National University of Singapore, 18 Science Drive 4, Singapore 117543, Singapore
Received 30 May 2007; revised 18 July 2007; accepted 20 July 2007
Available online 21 August 2007
Abstract—A library of chalcones with different basic groups were synthesized and evaluated for antiproliferative activities against
the human breast cancer (MCF 7) and colon cancer (HCT 116) cell lines. Structure–activity relationships were analyzed by projec-
tion methods (PCA/PLS) and multiple linear regression. Polar volume, hydrogen bonding features, HOMO energies, and charge on
the b carbon were found to be important factors. A basic group on either ring A or B of the chalcone led to a favourable increase in
polar volume, but when present on ring B, it increased HOMO energies and decreased the positive charge on the b carbon, both of
which led to lower activity. Several examples showed that final activity of the chalcone was influenced by compensatory interactions
among these parameters. In general, a single basic group on ring A was associated with good activity. A notable exception was com-
pound 1-123 which had basic groups on both rings A and B but still maintained a good activity profile with IC50 < 10 lM and selec-
tivity ratios >2.5. There was some evidence to show that structural differences in chalcones influenced not only activity but
mechanism of action. Compounds 6-130 and 7-140 which had basic groups on ring A interfered with cell cycle progression, but
the dibasic chalcone 1-123 had no effect.
ꢀ 2007 Elsevier Ltd. All rights reserved.
1. Introduction
the piperidinyl chalcones had promising antiprolifera-
tive activities (IC50 6 5 lM) and disrupted the cell cycle
at G1 and G2/M phases at these concentrations. In con-
trast, chalcones without the basic substituent had no ef-
fect on the cell cycle when tested at their IC50. In view of
this finding, we proceeded to synthesize a larger library
of chalcones bearing different basic substituents. The
compounds were evaluated for antiproliferative activi-
ties on two human cancer cell lines. Our objectives were
to assess how the presence of the basic moiety influenced
the physicochemical profile of the chalcone, and if this
would in turn affect antiproliferative activity. To address
these issues, the compounds were characterized by a
wide range of physicochemical descriptors and their
structure–activity relationships were analyzed by projec-
tion methods and multiple linear regression.
There are several reports on the antiproliferative proper-
ties of chalcones substituted with basic groups. Xia and
coworkers1 were among the first to demonstrate the im-
proved antiproliferative activity of chalcones with ami-
no groups. Pati et al.2 showed that methoxylated
chalcones with a 30-amino group had submicromolar
IC50 values against murine melanoma B16 cells. Dim-
mock and coworkers3 proposed that the presence of
amino function would increase the reactivity of chal-
cones as Michael acceptors and thus their antiprolifera-
tive activity. They postulated that the amino function
would be protonated at the low pH environment nor-
mally encountered in tumors. The electron withdrawing
effect of the protonated ammonium function would en-
hance the electrophilicity of the b carbon in the enone
linkage, hence increasing its reactivity as a Michael
acceptor. In our earlier investigation,4 we focused on
piperidinyl chalcones as antiproliferative agents, partly
because heterocycles like piperidine were less likely to
form reactive metabolites compared to the primary ami-
no function favored by other investigators.1–3 Many of
2. Design and synthesis
The chalcones were classified into groups according to
their substitution patterns on ring A (Table 1). Group
1 chalcones had ring A substituted with methoxy groups
at positions 2 and 4, and 1-methylpiperidin-4-yl at posi-
tion 5. Groups 6, 7 and 8 were variants of Group 1 in
which 1-methylpiperidin-4-yl had been replaced by groups
of comparable basicities, namely 1-ethylpiperidine
(Group 6), 1-methylpiperazine (Group 7) and 4-(1-pipe-
Keywords: Chalcones with basic functionalities; Antiproliferative act-
ivity; Structure–activity relationship.
*
Corresponding author. Tel.: +65 65162654; fax: +65 67791554;
e-mail: phagoml@nus.edu.sg
0968-0896/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2007.07.042

7022
X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
Table 1. IC50 (lM) values and selectivity ratios (SR) of synthesized chalcones
b
c
b
c
b
Code
Ring Aa
R0 on ring Ba
IC50 MCF7
SRMCF
IC50 HCT116
SRHCT
IC50 CCL186
1-5
1-6
20-Cl
2.8 0.4
3.4 0.7
2.5 0.6
7.2 0.3
4.3 1.1
3.9 0.6
8.5 1.9
6.6 0.7
5.4 0.5
7.0 0.8
7.0 1.6
3.8 0.4
3.9 0.5
2.8 0.3
2.7
2.3
1.4
1.9
3.0
2.7
2.7
1.6
1.6
1.8
2.4
3.5
2.3
2.1
3.4 0.0
3.9 0.2
2.9 0.5
9.9 2.6
7.4 2.1
12.4 1.6
7.8 1.8
10.2 1.3
10.4 1.8
9.4 1.1
17.7 6.7
10.3 2.9
8.3 3.9
8.7 3.1
2.2
2.0
1.2
1.4
1.7
0.9
2.9
1.0
0.9
1.3
0.9
1.3
1.1
0.7
7.6 0.9
7.8 1.7
3.6 0.4
H
40-Cl
OCH3
1-7
1-120
1-121
1-122
1-123
1-60
1-61
1-62
1-63
1-64
1-65
1-66
Pyridin-4-yld
Pyridin-3-yld
Pyridin-2-yld
40-MPZe
20-CH3
40-CH3
20-CH3O
40-CH3O
20-F
13.9 0.5
12.9 1.6
10.6 2.5
22.7 3.7
10.3 1.0
8.8 1.5
H3CO
N
12.5 1.4
16.7 5.0
13.2 3.0
CH3
GROUP 1
40-F
40-CF3
9
6
2.7
0.5
2-17
2-2
20-Cl
3.1 0.3
6.7 0.3
5.7 1.1
3.4 0.0
6.4 0.8
1.0
0.9
1.3
1.4
1.4
0.5
0.4
0.7
0.4
1.6
0.4
0.8
2.7
1.3
0.7
0.6
3.1 0.9
2.6 1.3
4.2 1.6
3.0 0.5
8.6 2.6
18.3 7.9
7.0 2.6
10.2 3.0
16 2.6g
50f,g
1.0
2.4
1.7
1.5
1.0
0.6
0.4
0.7
0.6
1.0
0.7
0.3
1.5
1.0
0.7
0.3
3.1 1.5
6.1 2.7
7.3 0.0
4.6 1.0
8.9 1.6
10.4 1.3
2.9 0.6
6.6 0.6
9.7 3.8
50f
40-Cl
2-3
2-4
H
OCH3
20-F
2-50
2-51
2-52
2-53
2-54
2-55
2-56
2-57
2-58
2-59
2-124
2-59a
40-F
20-CH3
40-CH3
20-CH3O
40-CH3O
30-CH3O l
20,40-(CH3O)2
30-Cl
22.3 1.4g
H3CO
OH
6.6 0.2
10.1 0.7
25.5 2.7g
31.2 5.1g
12.1 1.3g
2.7 0.6
13.8 0.9g
14.3 0.3g
14.4 1.4g
13.9 0.8g
N
6.6 1.7
7.4 0.5
25.8 0.6g
17.8 2.1g
13.9 1.6g
26.6 1.2g
4.4 0.9
2.2 0.6
37.7 3.7
18 3.4
9.4 1.9
7.9 0.8
CH3
40-CN
GROUP 2
40-CF3
40-MPZe
40-(CH3)2N
3-100
3-101
3-102
3-103
20 ,40-(CH3O)2
16.3 3.0g
21.4 0.5g
8.5 1.4
50g
1.4
1.2
1.3
1.0
13.4 3.0
10.5 7.8
4.9 1.9
50f,g
1.7
2.5
2.3
1.0
22.2 5.2
26.6 2.8
11.3 3.4
50f
20-Cl
40-Cl
H
N
HN
GROUP 3
4-104
4-105
4-106
4-107
20-Cl
47.4 4.9g
25.1 0.7g
50f,g
1.1
2.0
1.0
1.0
23.9 6.8g
50f,g
2.1
1.0
1.0
1.0
50f
50f
50f
50f
40-Cl
N
H
50f,g
20,40-(CH3O)2
50f,g
50f,g
GROUP 4
5-14
5-15
40-Cl
6.5 1.3
ND
ND
1.7
—
6.2 1.2
5.2 0.6
4.4 0.7
12 9.1
1.8
2.5
3.1
1.7
1.7
0.8
2.6
11.3 0.7
12.8 0.6
13.7 0.8
20.5 5.1
36.9 8.9
14.1 2.2
44.8 5.1
20-Cl
OCH3
5-16
H
—
5-110
5-111
5-112
5-113
Pyridin-4-yld
Pyridin-3-yld
Pyridin-2-yld
40-MPZe
7.9 0.6
13.6 2.8
6.4 2.0
13.7 1.8g
2.6
2.7
2.2
3.3
22.1 12.1
18.9 6.6
17 4.6
H3CO
GROUP 5
H3CO
6-130
6-131
6-132
6-133
6-134
20-Cl
2.6 0.5
4.8 1.4
3.4 0.3
2.9 1.3
8.8 3.5
3.9
3.7
2.3
2.3
2.3
6.7 2.1
7.7 1.6
6.8 1.6
7.5 3.0
10 2.7
1.5
2.3
1.2
0.9
2.0
10 0.9
17.5 1.8
7.8 2.3
6.6 0.3
20.4 1.2
H
40-Cl
Pyridin-4-yld
40-MPZe
OCH3
C2H5
N
GROUP 6
H3CO
N
7-140
7-141
7-142
7-143
7-144
20-Cl
6.2 0.6
7.2 0.4
6.0 1.4
2.5
2.3
2.6
3.4
2.2
14.1 0.7g
16.7 1.8g
19.4 2.0g
40.4 6.1g
29.8 0.8g
1.1
1.0
0.8
1.2
0.9
15.4 0.7
16.6 1.7
15.7 1.7
47 2.6
20-F
OCH3
N
H3C
H
40-MPZe
Pyridin-4-yld
13.8 2.4g
12.4 2.1
27.1 0.9
GROUP 7

X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
7023
Table 1 (continued)
b
c
b
c
b
Code
Ring Aa
R0 on ring Ba
IC50 MCF7
SRMCF
IC50 HCT116
SRHCT
IC50 CCL186
H3CO
N
8-160
8-161
20-Cl
20-F
3.7 0.3
3.5 0.2
1.8
2.0
19.3 8.6
16.8 4.9
0.3
0.4
6.5 0.7
6.9 0.2
OCH3
N
GROUP 8
H3CO
9-150
9-151
9-152
20-Cl
20-CH3O
H
22.4 5.7g
11.8 0.1g
25.5 0.9g
0.9
1.0
0.7
50f,g
0.4
0.2
0.5
19.2 1.0
11.8 3.8
18.6 0.5
OH
50f,g
H3C
37.3 2.4g
GROUP 9
10-8
10-9
10-10
20-Cl
40-Cl
H
2.9 0.2
3.2 0.1
4.2 0.3
2.1
1.6
2.3
2.3 0.7
3.3 0.7
3.6 1.2
2.6
1.6
2.7
6
0.1
H3C
N
OCH3
5.2 0.1
9.8 1.0
GROUP 10
GROUP 11
OH
H3C
N
11-11
20-Cl
ND
ND
—
—
6.9 0.6
2.2
1.2
15.3 0.6
13
2,3,4-Trimethoxy phenyl
20,40-(CH3O)2
25.1 2.6g
30.0 2.8
ND, not determined.
a Rings A and B refer to the rings in the chalcone template:
O
A
B
R'
.
b IC50 values were determined against the human breast cancer cells MCF 7, transformed human colon cancer cells HCT 116, and normal human
diploid embryonic lung fibroblasts CCL 186. Values represent the mean of at least two determinations.
c Selectivity ratio = IC50 CCL186/IC50 MCF7 or IC50 CCL186/IC50 HCT116
d Ring B of chalcone is replaced by pyridine.
.
e 40-MPZ = 40-(4-methylpiperazin-1-yl).
f Values were not accurately determined and IC50 is likely to exceed 50 lM.
g Significantly greater than 10 lM, independent samples t-test, p < 0.05.
ridinyl) piperidine (Group 8). Addition of a phenolic
OH group to ring A of Group 1 gave the Group 2 chal-
cones and removal of 4-methoxy from ring A of Group
1 gave the Group 10 chalcones. Groups 3 and 4 had only
basic functionalities on ring A and served to highlight
the importance of the omitted methoxy groups on this
ring. In contrast, Groups 5 and 9 with no basic substitu-
ent on ring A would provide insight on the role of the
basic moiety. Demethylation of the methoxy group in
Group 10 gave rise to Group 11.
dehyde and a substituted acetophenone (Fig. 1). Mech-
anistically, the reaction involves formation of
a
carbanion from the acetophenone in the presence of
the base NaOH, followed by nucleophilic attack by
the carbanion on the carbonyl carbon of the benzalde-
hyde and subsequent loss of water to give the chalcone.
The benzaldehydes were purchased from commercial
sources, except for 4-(4-methylpiperazin-1-yl) benzalde-
hyde which was synthesized by a palladium-catalyzed
reaction between 2-(4-bromophenyl)-1,3-dioxane and
1-methylpiperazine (Fig. 2).6 The aldehyde functionality
of 4-bromobenzaldehyde was first protected by conver-
sion to an acetal before reaction with 1-methylpiper-
azine. The product was subsequently deprotected in
aqueous acid and condensed with the desired acetophe-
none to give the chalcone.
The main substituents on ring B were the halogens (F and
Cl) which were associated with good activity in our earlier
study.4 Other substituents were selected from the four
quadrants of the Craig Plot,5 so as to cover a range of p
(lipophilicity) and r (electron donating/withdrawing) val-
ues. Most of these substituents were introduced to ring B
of Groups 1 and 2. Ring B was also replaced by pyridine, a
weakly basic heteroaromatic ring in some compounds.
The syntheses of the acetophenones 1-(2,4-dimethoxy-5-
(1-methylpiperidin-4-yl)phenyl)ethanone (1-1), 1-(2-hy-
droxy-4,6-dimethoxy-3-(1-methylpiperidin-4-yl)phenyl)
ethanone (2-1), and 1-(2-methoxy-5-(1-methylpiperidin-
4-yl)phenyl)ethanone (10-1) had been described earlier.4
The chalcones were synthesized by a base-catalyzed
Claisen Schmidt reaction between a substituted benzal-

7024
X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
O
H
O
B
R2
OH
O
O
a
b
CH2
CH3
A
B
R2
A
A
R1
R1
R1
c
O
A
B
R2
R1
Figure 1. Scheme for synthesis of chalcones, 3% w/v NaOH in methanol, 12 h, rt (28 ꢁC). (a) Formation of carbanion in presence of alkali;
(b) reaction with benzaldehyde; (c) loss of water.
CHO
Compounds 9-1 and 11-1 were not intentionally targeted
for synthesis but were obtained during the syntheses of
other acetophenones. Compound 9-1 was obtained
when the synthesis of 1-[5-(diethylaminomethyl)-2,4-
dimethoxy phenyl]ethanone was attempted. As shown
in Figure 5, the reaction was to proceed via the reduc-
tion of N,N-diethyl-2,4-dimethoxybenzamide to give
N-(2,4-dimethoxybenzyl)-N-ethylethanamine, followed
by acetylation with acetic anhydride and boron triflou-
ride to give the desired product. However, the Friedal
Crafts reaction resulted in the demethylation of one of
the methoxy groups and loss of the diethylamino func-
CHO
O
O
a
b
N
Br
N
Br
CH3
Figure 2. Reagents and conditions: (a) 1,3-propandiol, p-TsOH,
toluene, reflux, 24 h; (b) (i) Pd2(dba)3 sodium tertiary butoxide,
toluene, reflux 10 h; (ii) aq 1 M HCl.
tion
phenyl)ethanone (9-1). It was subsequently used as the
to
give
1-(2-hydroxy-4-methoxy-5-methyl-
These acetophenones were used for the syntheses of
Groups 1, 2, and 10 chalcones, respectively. 1-(2,4-
Dimethoxy-5-(1-ethylpiperidin-4-yl)phenyl)ethanone
(6-1), the starting acetophenone for Group 6, was
synthesized by a similar route (Fig. 3). In the case of
Groups 3, 4, and 5, their starting acetophenones were
purchased from commercial sources.
starting acetophenone for the Group 9 chalcones.
1-(2-Hydroxy-5-(1-methylpiperidin-4-yl)phenyl)ethanone
(11-1) was obtained during the Friedal Crafts acetylation
of 1-methyl-4-(40methoxyphenyl)piperidine using AlCl3
as the Lewis acid catalyst (Fig. 6). During this reaction,
demethylation of the methoxy group occurred. When
the catalyst was changed to boron trifluoride, demethyla-
tion was not observed and compound 10-1 was obtained4
(Fig. 6).
1-(2,4-Dimethoxy-5-(4-methylpiperazin-1-yl)phenyl)eth-
anone (7-1) and 1-[5-(1,40-dipiperidin-10-yl)-2,4-dimeth-
oxyphenyl]ethanone (8-1) were synthesized by Pd/
BINAP-catalyzed amination. Briefly, 1-bromo-2,4-
dimethoxybenzene was acylated to give the acetophe-
none which was then reacted with 1-methylpiperazine
or 1,40-bipiperidine in the presence of cesium carbonate
as base, Pd2(DBA)3, and rac-BINAP to give the desired
product (Fig. 4).7 These acetophenones were used for
the syntheses of Groups 7 and 8 chalcones.
3. Results and discussion
3.1. Structure–activity relationships
Table 1 lists the IC50 values of the chalcones against hu-
man cancer cell lines (breast cancer cells MCF 7 and
OCH3
OCH3
OCH3
O
OCH3
CH3
b
c
a
H3CO
H3CO
H3CO
6-1
H3CO
N
N
N
C2H5
C2H5
C2H5
Figure 3. Reagents and conditions: (a) 1-ethylpiperidin-4-one, HCl gas, acetic acid, 25 ꢁC, 24 h; (b) 10% Pd/C, H2, acetic acid–water, 25 ꢁC, 24 h;
(c) Boron trifluoroetherate, acetic anhydride, DCM.

X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
7025
OCH3
O
OCH3
O
OCH3
OCH3
O
CH3
b
a
CH3
CH3
H3CO
N
N
H3CO
H3CO
H3CO
N
Br
Br
8-1
7-1
N
CH3
Figure 4. Reagents and conditions: (a) boron trifluoroetherate, acetic anhydride, DCM; (b) Pd2(dba)3, Cs2CO3, rac-BINAP, toluene, reflux, 10 h,
1-methylpiperazine (to give 7-1) or 1,40-bipiperidine (to give 8-1).
OCH3
OCH3
OCH3
OCH3
OH
O
C
c
b
d
a
CH3
H3CO
H3CO
H3CO
H3CO
CON(C2H5)2
H3CO
CH2N(C2H5)2
9-1
COOH
COCl
CH3
Figure 5. Reagents and conditions: (a) dry benzene, SOCl2, reflux, 2 h; (b) dry THF, diethylamine; (c) borane, dry THF; (d) boron trifluoroetherate,
acetic anhydride, DCM.
tant in determining the effect on activity. Selectivities
OCH3
O
OCH3
OH
O
were similarly affected as seen from the large falls in
selectivities in some groups (7 and 8) and the absence
of a specific trend in other groups (1 and 6). The basic
group could be introduced into ring B, as shown in Step
B of Figure 7. The lone example (5-16 and 5-113)
showed a fall in activity but more examples would be
needed to establish a trend.
CH3
CH3
a
b
11-1
10-1
N
N
N
CH3
CH3
CH3
Second, comparisons were made to determine if having
more than one basic entity made a difference to activity.
Only a few compounds were available for this compari-
son (Step C of Fig. 7), namely, those with basic groups
on both rings (1-123, 2-124, 6-134, and 7-143) and their
counterparts with only a basic group on ring A (1-6, 2-3,
6-131, and 7-142). These compounds are listed in Table 2.
We found that in 5 out of 8 cases, chalcones with basic
groups on both rings had significantly higher IC50values
(poorer activity) than those with a single basic group
on ring A. No clear trend was observed for selectivity
ratios. Step D shows another approach which was to
start with compound 5-113, a monobasic chalcone with
the basic moiety on ring B (not ring A) and to compare
it with the dibasic chalcones 1-123, 6-134, and 7-143
(Table 2). Of the six comparisons made on two cell lines,
the IC50 values of five were not significantly different
from 5-113 but there were more instances (4 out of 6)
of a fall in selectivity on going from 5-113 to the dibasic
chalcone.
Figure 6. Reagents and conditions: (a) boron trifluoroetherate, acetic
anhydride, DCM; (b) AlCl3, acetyl chloride, DCM.
transformed human colon cancer cells HCT 116) and
normal human diploid embryonic lung fibroblasts
CCL 186. Selectivity ratios (IC50 CCL186/IC50 MCF7 and
IC50 CCL186/IC50 HCT) were calculated from mean values
for each compound. Arbitrary thresholds of
IC50 6 10 lM and ratio P2.5 were used to identify
promising compounds. Nearly 2/3rd of the compounds
were found to have IC50 values that were not signifi-
cantly greater than 10 lM (independent samples t-test,
p < 0.05). In terms of selectivity ratios, more compounds
had selectivity ratios P2.5 when evaluated against MCF
7 cells (13) than HCT 116 cells (6).
The contribution of the basic groups to antiproliferative
activity was assessed by making the following compari-
sons which are depicted in Figure 7. First, we compared
the activities (based on IC50 HCT116 values) of the non-
basic chalcones 5-14, 5-15, 5-16 with their counterparts
in Groups 1, 6, 7, and 8 (basic group on ring A) (Step
A). There was no significant difference in IC50 between
the non-basic chalcone and its Group 1 or 6 analog,
but the Group 7 chalcones (7-140, 7-142) were less active
than the non-basic chalcones (5-15, 5-16). This shows
that the nature of the basic group on ring A was impor-
Thus, the nature of the basic group and where it was lo-
cated (ring A or B) determined its effect on antiprolifer-
ative activity. As shown above, 1-methylpiperidine and
1-ethylpiperidine on ring A maintained IC50 values
while others like 1-methylpiperazine adversely affected
activity. In terms of IC50 values, there was little advan-
tage in having basic groups on both rings of the chal-
cone. Based on the limited compound pairs listed in

7026
X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
OCH3
O
A
C
H3CO
OCH3
O
A
BASIC GROUP
O
OCH3
B
A
Chalcone with basic functionality on
ring A or ring B
H3CO
N
B
A
NCH3
H3CO
BASIC GROUP
OCH3
O
Non-basic chalcone
Chalcone with
B
basic functionalities on
rings A + B
D
B
H3CO
N
NCH3
Figure 7. Scheme to summarize effects of basic groups on chalcone rings A and B. Step A: introducing a basic functionality resulted in either no
change or a fall in antiproliferative activity, depending on the nature of the basic group. Step B: comparing 5-16 and 5-113 showed a fall in activity.
More examples are needed for confirmation. Step C: most examples showed a significant drop in activity and no clear-cut effect on selectivity ratios.
Step D: most examples showed no significant change in antiproliferative activity and decrease in selectivity ratios.
Table 2. IC50 and selectivity ratios (SR) of chalcones selected to show relative contributions of the basic group to activity and selectivity
Basic group on ring Aa
Basic group on ring Bb
IC50 MCF7 (lM)
IC50 HCT116 (lM)
SRMCF7
SRHCT116
c
c
Compound
5-113
1-123
1-6
None
MP
MP
40-MPZ
40-MPZ
None
13.7 1.8
8.5 1.9
3.4 0.7d
8.8 3.5
4.8 1.4
13.8 2.4
6.0 1.4d
14.4 1.4
5.7 1.1d
17 4.6
7.8 1.8
2.9 0.5
3.3
2.7
2.3
2.3
3.7
3.4
2.6
0.7
1.3
2.6
2.9
2.0
2.0
2.3
1.2
0.8
0.7
1.7
6-134
6-131
7-143
7-142
2-124
2-3
EP
EP
40-MPZ
10.0 2.7
7.7 1.6
40.4 6.1e
19.4 2.0d
13.9 1.6
4.2 1.6d
None
40-MPZ
None
40-MPZ
None
MPZ
MPZ
MP
MP
a Structures of compounds are given in Table 1. MP, 1-methylpiperidin-4-yl; EP, 1-ethylpiperidin-4-yl; MPZ, 4-methylpiperazin-1-yl.
b 40-MPZ = 40-(4-methylpiperazin-1-yl).
c SR = Selectivity ratios (IC50 normal cell/IC50 cancer cell). Collated from Table 1.
d Significant difference between members of same Group (e.g., 1-123 and 1-6). Independent samples t-test, p < 0.05. For Step C of Figure 7.
e Significantly different from 5-113 (Independent samples t-test, p < 0.05). For Step D of Figure 7.
Table 2, monobasic chalcones fared better than their
dibasic counterparts in terms of IC50 values. It was also
more advantageous to have the basic moiety on ring A
than on ring B, but a basic group on ring B may be bet-
ter for selectivity. More examples are required to con-
firm the latter. Interesting, in spite of these general
conclusions, the only compound to comply with the
threshold limits for IC50 and selectivity on both cell lines
was the dibasic chalcone 1-123 with 1-methylpiperidinyl
on ring A and 4-methylpiperazinyl on ring B. This com-
pound had IC50 values of 7.8–8.5 lM and good selectiv-
ities (>2.5) for the cancer cell lines. Its Group 6 analog
(6-134) had comparable activity but poorer selectivity,
while its Group 2 and 7 analogs fared poorly in
comparison.
larger libraries. Moreover, such comparisons do not give
insight as to why certain variations were observed. A
QSAR analysis would provide a better understanding
and this approach was adopted here. The compounds
were drawn in their predominant ionized forms at pH
7.0 and a range of descriptors were determined in silico
from their minimized low energy conformations. The
parameters were shape descriptors (molecular length,
depth, and width), topological indices (various connec-
tivity and shape indices), steric parameters (area, vol-
ume), polarity measures (polar surface area, polar
volume, % hydrophilicity, partial positive and negative
surface areas), hydrogen bonding parameters (hydrogen
bond acceptors and donors, Hansen hydrogen bonding),
electronic parameters (molar refractivity, dipole mo-
ment, HOMO, LUMO, charge on the a and b carbons
on the enone linker), and lipophilicity descriptors
(ClogP, CNDO logP, Crippen logP). They are tabu-
lated in Supplementary Information Table 1. Projection
methods (principal component analysis PCA and its
regression extension, projection to latent structures by
3.2. Quantitative structure–activity analysis
Qualitative comparisons like those described in Section
3.1 are generally useful when applied to a small number
of compounds but become untenable when analyzing

X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
7027
partial least squares analysis PLS)6 and multiple linear
regression (MLR) with stepwise removal of correlated
parameters were employed in the analyses.
the dependent variable.10 In spite of the clustering of
certain groups in the score plot (Fig. 8), PLS was applied
to the entire library of compounds. This decision was
made because the PLS models derived from the clus-
tered groups (not shown) did not give additional insight
into SAR, although they were statistically more robust.
In addition, the small number of compounds in the clus-
tered groups may lead to over-interpretation of the SAR
trends.
3.2.1. Projection methods. We found a high degree of
correlation among the descriptors. Nearly 50% of the
parameters correlated at the level of r P 0.75.9 Pear-
son’s correlation coefficients for all descriptors are given
in Table 2 of Supplementary Information. In view of the
high level of correlation, this database would be well sui-
ted for analysis by projection methods.10 The 29 descrip-
tors were first analyzed by principal component analysis
(PCA) to reduce them to relevant components that rep-
resent the main variations in the data set. A significant
two-component model (Model 1) was obtained, with
cumulative r2 (reflecting variability) and cumulative q2
(reflecting level of predictability) of 0.67 and 0.61,
respectively. The 1st component summarized the topo-
logical and size parameters, whereas the 2nd component
had inputs from lipophilicity and hydrogen bonding
parameters. The score plot showed the distribution of
the test compounds based on these components
(Fig. 8) and it revealed distinct clustering among certain
Groups. These were Groups 1, 7 that had different basic
groups on ring A, and Groups 5, 9 that had no basic
group on ring A. This implied that the clustered groups
shared similar physicochemical profiles and were differ-
ent from the non-clustered groups. It would be advisable
to analyze their QSAR separately from other groups.
Fig. 8 also showed that some members in Groups 1, 2,
6, and 7 were isolated from the other compounds in
the group. Interestingly, all the outlying compounds
were found to have a 1-methylpiperazine group on ring
B.
A PLS model was obtained by the regression of the com-
bined pIC50 values of the compounds against the
descriptor set. A significant two-component model
(Model 2) with satisfactory cumulative r2 (0.443) and
q2 (0.283) was obtained. The loadings plot (Fig. 9)
showed that the two pIC50 values were close but did
not overlap. Thus, these values could be modeled sepa-
rately. Figure 9 also showed that the IC50 values were
largely influenced by the same parameters: a large dipole
moment and polar volume coupled with an electron
poor b carbon were favourable for activity, whereas
the presence of hydrogen bonding groups and a large
(more negative) HOMO energy had an inverse effect
on activity. When PLS was applied to the selectivity ra-
tios as dependent variables, no significant model was ob-
tained (r2 (0.082), q2 (0.032)).
In view of the good outcome of Model 2, PLS for each
pIC50 was separately determined and the model statistics
summarized in Table 3. It was noted that a statistically
more robust PLS model was obtained with pIC50 MCF7
(Model 4) than pIC50 HCT116 (Model 3). Polar volume,
HOMO, and the charge on b carbon were important
parameters influencing both activities. Dipole moment
strongly influenced pIC50 HCT116 but was not among
the top four descriptors for Model 4 (pIC50 MCF7). H do-
nor properties were rated prominently for pIC50 MCF7
but not for pIC50 HCT116. Some comments on these
parameters are given in the following paragraphs.
PLS is a regression extension of principal component
analysis. In PLS, principal components are calculated
to explain most of the variance in the descriptor set
and to concurrently maximize their correlations with
6
4
2
0
-2
-4
-6
-10 -9
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
9
10
t[1]
SIMCA-P 11 - 5/17/2007 12:55:41 PM
Figure 8. PCA score plot of t[1] (1st principal component) versus t[2] (2nd principal component) for chalcones (n = 65, 29 descriptors). Group (G) 1,
black h; G2, red s; G3, blue }; G4, green ; G5, yellow n; G6, pink h; G7, green s; G8, blue n; G9, brown h; G10, yellow }; G11, red };
*
Compound 13, purple }. G5 (yellow n) and G9 (brown h) do not have basic groups on ring A and are clustered on the left of quadrant. G1 (black
h) and G7 (green s) are also clustered along the midline. Outliers in G1, G2, G6 and G7 are enclosed in circle on right quadrant. These outliers are
1-123, 2-124, 6-134, 7-143. The eclipse corresponds to the confidence region based on Hotelling T2 (0.05).

7028
X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
ß-uncharge
DM
0.6
0.4
Polar Vol
PNSA1(S)
PIC50-MCF
0.2
a-uncharge
PIC50mAol-HwCidTth
Log P
CLOGP (S)
-0.0
-0.2
-0.4
kappa 2
i i
mol lengMtohl Vol
CMR (S)
connectivi
connectiivii
connectivi
AREA (S)
mol de th
LogPCrippe
CNDO LogP
% Hydrophi
Polar surf
PPSA 1(S)
LUMO
Hansen H b
H acceptor
H donor
HOMO
-0.2
-0.1
-0.0
0.1
w *c[1]
0.2
0.3
0.4
SIMCA-P 11 - 5/17/2007 1:19:08 PM
Figure 9. PLS Loading plot for pIC50 MCF (n = 61, 29 descriptors) and pIC50 HCT (n = 65, 29 descriptors). Parameters that were displaced far away
from origin contributed significantly to activity. The important contributors (encircled) were dipole moment DM, charge on b carbon (b uncharged),
polar volume (top right quadrant), and HOMO, H donor (lower left quadrant).
Polar volume was identified in both Models 3 and 4 as
an important factor directly influencing antiproliferative
activity. When the polar volumes of the various groups
were examined, a general trend was that chalcones with
more substituents on ring A had larger polar volumes.
This is shown in Table 4 where compounds from differ-
ent groups but with the same ring B substituent (20-Cl in
this case) were compared. Polar volume was largest for
the highly substituted Group 2 compound (2-17) and
decreased as substituents on ring A were progressively
removed. Thus polar volume decreased in the order
2-17 > 1-5 > 10-8 > 11-11. However, IC50 of these com-
pounds did not always change in the expected manner,
suggesting that other factors besides polar volume con-
tributed to the final activity profile. Besides the degree
of substitution on ring A, within each group, larger polar
volumes were found in compounds with 4-methylpipera-
zin-1-yl and dimethylamino groups on ring B. Pyridine as
ring B was also associated with a larger polar volume than
the phenyl ring B. For example, the polar volume of 1-6
b carbon was associated with better activity (Models 3
and 4, Table 3), which was in keeping with proposal
by Dimmock et al.3 The magnitude of the positive
charge on the b carbon varied with the nature of ring
B and was generally greater than 0.04 esu (Table 2, Sup-
plementary Information). One notable exception was
when ring B had a 40-(4-methylpiperazin-1-yl) substitu-
ent, which caused the positive charge on the b carbon
to be reduced to 0.031–0.037 esu. This may be due to
the aniline-like character of the proximal nitrogen atom
of the piperazine ring. The lone pair of electrons on the
proximal nitrogen were delocalized into ring B and this
may lead to the reduction in positive charge.
The PLS models identified low (more negative) HOMO
energies and poor H bonding capabilities as important
factors for good antiproliferative activity. The HOMO
energies were strongly influenced by the substitution
pattern on rings A and B. In particular, electron with-
drawing groups (F, Cl, CF3, pyridine) on ring B resulted
in low (favorable) HOMO energies while electron donat-
ing groups on ring B like 40-dimethylamino and 40-(4-
methylpiperazin-1-yl) had the opposite (unfavorable)
effect. H bond donor capability was identified as an
undesirable feature affecting pIC50 MCF7 (Model 4), but
not pIC50 HCT116 (Model 3). This difference may not be
important as there were significant correlations among
the parameters. HOMO and H bond acceptor were cor-
related at p < 0.01 (r = 0.324, two-tailed). Similarly, the
descriptors H bond acceptor and H bond donor were
correlated (r = 0.761, p < 0.01, two-tailed). Hence, we
concluded that both pIC50 values were adversely affected
by H bonding features. These features were most prom-
inent for Groups 2 and 9 that had the OH group on ring
A. Thus, several members in these groups had poor
pIC50 values.
3
3
˚
˚
(no substituent on ring B, 387 A ) increased to 431 A
3
0
˚
(1-123) and 535 A (1-121) on substitution with 4 -(4-
methylpiperazin-1-yl) and 30-pyridinyl (in place of ring
B), respectively. As before, an increase in polar volume
was not always accompanied by greater activity.
Dipole moment was identified as an important contrib-
utor to pIC50 HCT116 but not pIC50 MCF7. Since dipole
moment and polar volume were significantly correlated
to each other (Pearson’s correlation coefficient
r = 0.387, p 0.001, two-tailed), this difference may not
be important. Interestingly, within each group, com-
pounds with 20-Cl or 20-F on ring B had larger dipole
moments than compounds with 40-Cl or 20-OCH3.
The b carbon of all the compounds had a positive charge
when evaluated in their non-protonated states. This is
due to the transmission of the electron withdrawing ef-
fect of the carbonyl oxygen. A more electron deficient
As stated earlier, the key parameters interacted with
each other to influence the final activity of the com-

X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
7029
Table 3. Summary of model statistics from PLS and MLR
cause of the ring A substitution pattern but this defi-
ciency was adequately compensated by other
advantageous features (dipole moment, HOMO, charge
on b carbon). Another example is compound 2-17
which had IC50 values of 3 lM against both cell lines.
Compound 2-17 had 20-Cl on ring B and ring A substi-
tuted with two methoxy groups, 1-methylpiperidine
and a phenolic OH. In this case, good activity was re-
tained in spite of the H bonding features of the pheno-
lic OH. It may have been compensated in part by the
favorable HOMO energy and electron deficiency at
the b carbon (due to the electron withdrawing 20-Cl
on ring B) and its large polar volume (highly substi-
tuted ring A).
Model 3 (projection method, PLS)a
Dependent variable: pIC50 HCT116
Number of components: 3
Cumulative r2 0.49
Cumulative q2 0.23
Top four descriptors contributing to activityc
Dipole moment (+), polar volume (+), HOMO (ꢀ), charge on b
carbon (+)
Model 4 (projection method, PLS)a
Dependent variable: pIC50 MCF7
Number of components: 3
Cumulative r2 0.66
Cumulative q2 0.50
Top four descriptors contributing to activityc
Charge on b carbon (+); HOMO (ꢀ); Polar Volume (+); H bond
donor (ꢀ)
3.2.2. Multiple linear regression. Besides projection
methods, multiple linear regression (MLR) with step-
wise elimination of correlated parameters was also em-
ployed for model construction. The results are
summarized in Table 3. MLR gave rise to models with
poorer r2 values but shared several similarities to the
PLS models. As before, better model statistics were ob-
served with pIC50 MCF7 as compared to pIC50 HCT116. In
addition, the critical parameters identified by MLR
coincided to some extent with those found for the pro-
jection methods. In the case of pIC50 MCF7, polar vol-
ume and H bonding were ranked highly by both
methods. For pIC50 HCF116, both methods highlighted
the contribution of dipole moment, but H bond donor
was identified by MLR (not PLS) as an important
feature.
Model 5 (MLR, stepwise)b
Dependent variable: pIC50 HCT116
r2 0.27
F 11.49
Standard error of estimate 0.32
Descriptorsc: dipole moment (+), H bond donor (ꢀ)
Model 6 (MLR, stepwise)b
Dependent variable: pIC50 MCF7 r2 0.42
F 22.74
Standard error of estimate 0.28
Descriptorsc polar volume (+), H bond acceptor (ꢀ)
Model 7 (MLR, stepwise)b
Dependent variable: selectivity ratio against MCF 7 cells (SRMCF7
r2 0.49
F 55.78
)
Standard error of estimate 0.64
Descriptor: H bond donor (ꢀ)
As mentioned earlier, projection methods failed to iden-
tify statistically significant models for selectivity ratios
against HCT (SRHCT) and MCF (SRMCF). MLR did
not give a credible model for SRHCT, but a fairly prom-
ising model was obtained for SRMCF (Model 7, Table 3).
In this model, the presence of H bond donors was iden-
tified as the only factor that adversely affected selective
activity. This is an interesting finding because it implied
that omitting or reducing the presence of H bond donors
in compounds may lead to good activity and selectivity
against MCF 7 cells.
a PLS models were obtained with Simca-P 11.0, Umetrics (Umea,
Sweden).
b Multiple linear regression (MLR) with stepwise omission of param-
eters were run on SPSS for Windows, Version 14.0 (Chicago, IL).
c Descriptors are cited in order of decreasing contribution to model.
(+) Directly correlated to activity; (ꢀ) Inversely correlated to activity.
pound. The following two examples served to illustrate
this point. Compound 10-8 had good activity against
HCT 116 cells (IC50 2.3 lM). It had a 20-Cl substituent
on ring B and its ring A was substituted with 2-meth-
oxy and 5-(1-methylpiperidin-4-yl). It had the necessary
attributes for good activity, namely, a large dipole mo-
ment, low HOMO energy, and reasonable electron defi-
ciency at the b carbon. These properties were traced to
the presence of the electron withdrawing 20-Cl on ring
B. On the other hand, it had a small polar volume be-
3.3. Effect of selected chalcones on cell cycle of HCT 116
cells
We showed previously that compound 1-5 at 5 lM, a
concentration close to its IC50, caused HCT16 cells to
be arrested at the G1 phase, but not at the G2/M phase.4
Table 4. Polar volumes of representative chalcones with 20-Cl on ring B
Compounda
IC50 (lM)
Compounda
IC50 (lM)
b
c
Polar volume (A )
3
b
c
Polar volume (A )
3
˚
˚
1-5
2.8, 3.4
3.1, 3.1
331.6
495.1
132.7
33.1
7-140
8-160
9-150
10-8
6.2, 14.1
3.7, 19.3
22.4, 43.9
2.9, 2.3
361.6
333.2
45.9
2-17
3-101
4-104
5-15
6-130
21.4, 10.5
47.4, 23.9
ND, 5.2
2.6, 6.7
289.2
255.1
296.8
369.2
11-11
ND, 6.9
a Structures of compounds are given in Table 1.
b The two values are the IC50 MCF7 and IC50 HCT116, cited in that order. ND, not determined.
c Polar volume was determined from Sybyl 7.0 (Tripos Associates, St. Louis, MO).

7030
X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
At higher concentrations (10 and 20 lM), both phases
of the cell cycle were affected. We also found that com-
pounds 5-14 and 13, with no basic moieties on ring A,
did not affect the cell cycle at concentrations close to
their IC50 values.4 This led us to propose that the pres-
ence of a basic group played a special role in targeting
the G1 phase of the cell cycle. To test our hypothesis,
we extended our investigations to three analogs of com-
pound 1-5, namely 1-123, 6-130, and 7-140. Compound
1-123 had 4-methylpiperazin-1-yl on ring B and 1-meth-
ylpiperidin-4-yl on ring A and was identified as a prom-
ising compound based on its IC50 and selectivity profile.
Compounds 6-130 (IC50 HCT116 6.7 lM) and 7-140
(IC50 HCT116 14.1 lM) had 1-ethylpiperidin-4-yl and 4-
methylpiperazin-1-yl groups on ring A, respectively.
Both compounds had in common 2’-Cl on ring B. The
three compounds were tested at 5 and 0.5 lM for their
ability to interfere with the transition of HCT 116 cells
from G1 to G2/M, as well as cells released from a G2/
M block and moving on to G1. The results are summa-
rized in Table 5.
Based on these criteria, compound 1-123 did not cause
G1 arrest or G2 arrest at 5 and 0.5 lM. The proportion
of cells in G1 and G2/M were quite similar to that of
control (untreated) cells in each case. On the other hand,
compound 6-130 arrested cells at both G1 and G2/M
phases at 5 lM but not at 0.5 lM. It caused a sharp rise
in cells at the G1 phase (65% compared to 35% for con-
trol cells, Table 5A). In its presence, the proportion of
cells at the G2/M phase was 47%, compared to 32%
for control cells (Table 5B). In the case of compound
7-140, it arrested the cells at the G1 phase only at
5 lM (52% compared to 35% for control cells, Table
5A). It had no effect on cells released from G2/M block.
The results showed that changing the ring A basic group
affected the phase of the cell cycle transition that was
interrupted. Compounds 1-5 and 7-140 only interfered
with the G1 phase, suggesting that the 1-methylpiperid-
inyl and 4-methylpiperazinyl groups had equivalent ef-
fects. In contrast, changing the 1-methylpiperidinyl
ring of 1-5 to 1-ethylpiperidinyl ring (6-130) caused both
G1 and G2/M phases to be affected. Despite its good
activity and selectivity, 1-123 did not interfere with the
cell cycle, indicating that an extra basic group on ring
B had a significant effect on the mechanism of action.
Our preliminary results showed that 1-123 did not inter-
fere with tubulin binding which was an effect widely
associated with antiproliferative chalcones.11 Further
investigations into its mode of action are in progress.
The score plot (Fig. 8) showed that dibasic chalcones
like 1-123 were outliers from their respective groups.
Interesting, 1-123 had no effect on the cell cycle, in con-
The cell cycle progression experiments were designed to
examine synchronized cell populations at G1 or G2/M
phase though one transit of the cell cycle. When cells
were arrested at the G1 phase, the proportion of G1
cells, 24 h after their release from G1 block, would re-
main high and there would be fewer cells in the G2/M
phase. When cells were arrested at the G2/M phase,
the proportion of cells in this phase, 6–8 h after their re-
lease from the G2/M block, would remain high, and
there would be few cells in the G1 phase.
Table 5. Effect of chalcones 1-123, 6-130, and 7-140 on distribution of HCT 116 cells in various phases of the cell cycle
Time Concentration
1-123c 6-130c
(h)b
(lM)
(A) Cells released from G1 blocka
7-140c
G1
S
G2/M
G1
S
G2/M
G1
S
G2/M
8
5
69.4 (65.6)
3.6 (3.5)
27.1 (30.7) 70.7 (65.6)
49.4 (35.1) 18.7 (16.7) 33.4 (49.1) 65.0 (35.1)
3.2 (3.5)
26.1 (30.7) 65.3 (65.6)
6.4 (16.7) 28.8 (49.1) 52.2 (35.1) 17.5 (16.7) 31.7 (49.1)
3.8 (3.5)
30.8 (30.7)
24
48
52.4 (59.0) 12.8 (7.7) 35.6 (34.0) 59.4 (59.0)
68.2 (65.6) 3.6 (3.5) 28.3 (30.7) 66.3 (65.6)
36.6 (35.1) 16.5 (16.7) 47.9 (49.1) 34.1 (35.1) 18.2 (16.7) 49.3 (49.1) 35.6 (35.1) 18.2 (16.7) 47.6 (49.1)
6.8 (7.7)
32.8 (34.0) 63.4 (59.0)
6.5 (7.7)
30.1 (34.0)
8
0.5
3.6 (3.5)
30.2 (30.7) 68.9 (65.6)
3.3 (3.5)
27.8 (30.7)
24
48
58.2 (59.0)
9.0 (7.7)
33.2 (34.0) 59.9 (59.0)
8.9 (7.7)
31.8 (34.0) 59.9 (59.0)
7.4 (7.7)
33.3 (34.0)
(B) Cells released from G2/M blockd
Time Concentration
(h)e
1-123c
6-130c
7-140c
(lM)
G1
S
G2/M
G1
S
G2/M
G1
S
G2/M
6
5
55.9 (59.5) 10.0 (9.2)
53.7 (59.0)
54.5 (63.9)
34.9 (32.3) 45.3 (59.5)
38.7 (34.0) 50.7 (59.0)
37.3 (30.7) 63.0 (63.9)
8.2 (9.2)
9.9 (7.7)
4.9 (5.3)
47.1 (32.3) 53.2 (59.5)
39.8 (34.0) 54.3 (59.0)
32.0 (30.7) 59.2 (63.9)
9.0 (9.2)
7.1 (7.7)
4.3 (5.3)
38.7 (32.3)
38.8 (34.0)
34.6 (30.7)
24
48
8.0 (7.7)
8.5 (5.3)
6
0.5
53.9 (59.5) 12.7 (9.2)
50.9 (59.0) 8.2 (7.7)
54.8 (63.9) 10.8 (5.3)
34.6 (32.3) 51.4 (59.5) 12.5 (9.2)
41.2 (34.0) 53.8 (59.0)
34.2 (30.7) 50.5 (63.9)
36.9 (32.3) 54.2 (59.5) 11.6 (9.2)
37.6 (34.0) 53.8 (59.0)
43.9 (30.7) 56.5 (63.9)
35.2 (32.3)
39.2 (34.0)
38.4 (30.7)
24
48
9.1 (7.7)
5.5 (5.3)
7.3 (7.7)
4.6 (5.3)
a Proportion of cells at start of experiment (0 h) were G1 59.9; S 3.4; G2/M 35.6.
b Time (h) after cells were released from the G1 block. Proportion of cells at the 24-h time point would indicate if test compound blocked G1 ! G2
phase transition.
c % Cells in each phase after exposure to test compound at the stated concentration. Values in parentheses represent control values in concurrent
experiments.
d Proportion of cells at start of experiment (0 h) were G1, 35.1; S, 16.7; G2/M, 49.1.
e Time (h) after cells were released from G2/M block. Proportion of cells at the 6-h time point would indicate if test compound blocked G2 ! G1
phase transition.

X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
7031
trast to another compound in the same group (1-5)
which was shown in an earlier study4 to cause arrest
of the G1 phase of proliferating HCT 116 cells.
5.1.1. General procedure for the synthesis of chalcones.4A
solution of the aldehyde (1.2 mmol) in methanol (5 ml)
was added dropwise to a stirred solution of the aceto-
phenone (1 mmol) dissolved in 3% (w/v) NaOH in
methanol (20 ml). The solution was stirred at room
temperature (28 ꢁC) for 12 h. Alternatively, the re-
agents were reacted under microwave at 130 ꢁC,
3 min in a Biotage InitiatorTM (Biotage AB, Uppsala,
Sweden) microwave reactor. The solvent was removed
under reduced pressure. The resulting residue was dis-
solved in 1 M HCl and extracted with ether. The aque-
ous layer was rendered alkaline with saturated aqueous
Na2CO3 solution to give a precipitate that was filtered,
washed with water, and dried. Some products were ob-
tained by recrystallization from methanol–water,
whereas others were purified by MPLC. Yields, mp,
NMR chemical shifts, and MS data of synthesized
chalcones are given in Supplementary Information,
Appendix 1.
4. Conclusion
The objective of this study was to evaluate the SAR of
the antiproliferative activity of chalcones and to deter-
mine if the critical factors that affected activity were
influenced by the basic group(s) in the chalcone tem-
plate. QSAR analyses identified four parameters as ma-
jor contributors to activity, namely polar volume, H
bonding features, HOMO energies, and charge on b car-
bon. Good antiproliferative activity was associated with
large polar volumes, the absence of H bonding groups,
low HOMO energies, and an electron deficient b carbon.
The basic group contributed to a desirable increase in
polar volume but adversely affected HOMO and charge
on b carbon when located on ring B. It did not signifi-
cantly affect H bonding parameters. Thus the final activ-
ity profile of the basic chalcone depended on a close
interplay among these factors.
5.1.2. Syntheses of acetophenones. Syntheses of 1-1, 1-2,
and 1-10 were described earlier.4 1-(4-Piperazin-1-yl-
phenyl)ethanone and 1-(4-piperidin-1-yl-phenyl)etha-
none, starting acetophenones for Groups 3 and 4, were
purchased from Sigma–Aldrich Corporation (St. Louis,
Missouri, USA).
The nature of the basic entity was also important. In gen-
eral, chalcones with a single basic functionality had bet-
ter antiproliferative activities than those with more than
one basic group. It was also preferable to locate the basic
entity on ring A rather than ring B. In spite of these gen-
eralizations, the most promising compound in the library
was a dibasic chalcone (1-123) which had 1-methylpipe-
ridinyl on ring A and 4-methylpiperazinyl on ring B. Fi-
nally, it was noted that that structural differences in the
chalcones also influenced their downstream effects. Com-
pounds 1-5,4 6-130, and 7-140, characterized by a single
basic entity on ring A, arrested the cell cycle at either G1
or G1 and G2/M phases. On the other hand, compound
1-123 with two basic groups did not interfere with the
cell cycle. Thus a different mechanism accounted for its
good antiproliferative activity.
5.1.2.1. 1-Ethyl-4-(2,4-dimethoxyphenyl)-1,2,3,6-tetra-
hydropyridine. 1-Ethylpiperidin-4-one (0.15 mol) was
added to a stirred solution of 1,3,5-trimethoxybenzene
(0.15 mol) in glacial acetic acid (200 ml) at 15–20 ꢁC.
HCl gas was bubbled through the reaction mixture
(1 h), followed by stirring at rt 24 h, and heating for
3 h at 95–100 ꢁC. The solvent was removed under re-
duced pressure and the residue was diluted with water.
The aqueous solution was extracted with ether and the
aqueous layer made alkaline with 1 M NaOH. The
crude product was extracted with ethyl acetate and puri-
fied by flash column. Yield: 65%. MS-APCI: [M+H]+
1
248.2 (247.2). H NMR (CDCl3): 7.10–7.07 (1H, mb),
6.45–6.43 (2H, mb), 5.74 (1H, mb), 3.80 (3H, s), 3.78
(3H, s), 3.13–3.12 (2H, m), 2.68–2.51 (6H, m), 1.18–
1.13 (3H, t). 13C NMR (CDCl3): 159.8, 157.8, 135.1,
129.8, 124.7, 123.3, 103.9, 98.7, 55.3, 52.8, 52.2, 50.1,
29.7, 12.2.
5. Experimental
5.1. Chemistry
Melting points were determined in open glass capillary
tubes on a Gallenkamp melting point apparatus and
were uncorrected. 1H NMR and 13C NMR spectra were
recorded on a Bruker DPX 300 MHz spectrometer and
chemical shifts were reported in d (ppm) relative to the
internal standard TMS. Mass spectra were collected
on LcQ Finnigan MAT mass spectrometer with chemi-
cal ionization (APCI) or electron spray ionization
(ESI) as probes. Analytical thin-layer chromatography
was performed on TLC silica gel 60 F254 (230–400 mesh)
and MKC18F silica gel 60 plates (size 2.5 · 7.5 cm). Pre-
parative MPLC was done using a C18 column (Merck
Lichropret RP-18, 15–25 lm) on a Shimadzu Promi-
nence instrument. The purity of the final compound
was confirmed by reverse-phase HPLC (X-Bridge C-
18, 150 mm · 3.0 mm, 3 lM particle size) using two mo-
bile phases, methanol–water and acetonitrile–water.
5.1.2.2. 1-Ethyl-4-(2,4-dimethoxyphenyl)-piperidine. 1-
Ethyl-4-(2,4-dimethoxyphenyl)-1,2,3,6-tetrahydropyri-
dine (5 g) was hydrogenated in acetic acid/water (10:1,
150 ml) with 10% palladium on carbon (0.5 g) as catalyst
at 30–40 psi, rt, 24 h. The reaction mixture was filtered
over Celite and the solvent removed in vacuo to give a
residue which was diluted with water, made alkaline
with 1 M NaOH, and extracted with ethyl acetate. The
residue left on evaporation of solvent at reduced pres-
sure was a liquid. It was dried under vacuum and used
without further purification. Yield: 94%. 1H NMR
(CDCl3): 7.13–7.10 (1H, m), 6.47–6.45 (2H, m), 3.799
(3H, s), 3.791 (3H, s), 3.08–3.04 (2H, m), 2.86 (1H,
mb), 2.46–2.43 (3H, q), 2.05–2.03 (2H, m), 1.78–1.75
(4H, m), 1.14–1.09 (3H, t). 13C NMR (CDCl3): 158.8,
157.7, 127.0, 126.8, 103.9, 98.5, 55.2, 55.1, 54.2, 52.7,
34.7, 32.3, 12.1.

7032
X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
5.1.2.3. 1-(2,4-Dimethoxy-5-(1-ethylpiperidin-4-yl)phen-
5.1.2.7. 1-(2-Hydroxy-5-(1-methylpiperidin-4-yl)phenyl)-
ethanone (11-1). A stirred solution of 4-(4-methoxyphe-
nyl)-1-methylpiperidine4 (1.8 g) in CH2Cl2 (50 ml) was
cooled in an ice bath. Powdered AlCl3 (1.8 g) was added
in small portions, followed by dropwise addition of acet-
yl chloride (0.8 ml). Stirring was continued for 24 h at
room temperature. The reaction mixture was diluted
with water, made alkaline with Na2CO3, and extracted
with CH2Cl2. Removal of solvent in vacuo gave a solid
residue that was recrystallized with methanol/water (2:1)
to give the desired product in 75% yield. Mp 83–85 ꢁC.
1H NMR (CDCl3): 12.10 (1H, s, OH), 7.55–7.54 (1H,
d, J = 1.9 Hz), 7.35–7.32 (1H, dd, J1 = 8.6 Hz,
J2 = 1.9 Hz), 6.92–6.89 (1H, d, J = 8.6 Hz), 2.99–2.95
(2H, mb), 2.60 (3H, s), 2.45–2.40 (1H, m), 2.32 (3H,
s), 2.09–2.01 (2H, m), 1.82–1.74 (4H, m). 13C NMR
(CDCl3): 204.4, 160.7, 136.7, 135.4, 128.3, 119.4,
118.3, 56.1, 46.3, 41.1, 33.5, 26.6.
yl)ethanone (6-1). The procedure for the synthesis of 1-1
was followed as described earlier.4 6-1 was obtained in
83% yield after recrystallization from methanol/water
(2:1). Mp 128–129 ꢁC, MS-APCI: [M+H]+ 292.5
1
(291.2). H NMR (CDCl3) 7.75 (1H, s), 6.41 (1H, s),
3.92 (3H, s), 3.89 (3H, s), 3.10–3.07 (2H, m), 2.86–2.77
(1H, mb), 2.56 (3H, s), 2.51–2.44 (2H, q), 2.10–2.02
(2H, m), 1.79–1.71 (4H, m), 1.16–1.11 (3H, t). 13C
NMR (CDCl3): 197.6, 161.7, 159.6, 129.4, 126.7,
120.0, 94.4, 55.5, 55.4, 54.1, 52.6, 35.1, 31.9, 31.8, 12.1.
5.1.2.4.
1-(5-Bromo-2,4-dimethoxyphenyl)ethanone.
Boron trifluoride dimethyletherate (46 ml, 0.5 mol)
was added dropwise to a stirred solution of 1-bro-
mo-2,4-dimethoxybenzene (7.2 ml, 0.05 mol) in CH2Cl2
(150 ml) which had been cooled in an ice bath. Acetic
anhydride (37.8 ml) was added dropwise to the solu-
tion and stirring was continued for 24 h at room tem-
perature. The reaction mixture was diluted with water,
rendered alkaline with Na2CO3, and extracted with
CH2Cl2. Removal of solvent in vacuo gave a solid res-
idue that was recrystallized with ethyl acetate/hexane
to give the product in 85% yield. Mp 149–150 ꢁC.
MS-APCI: [M+H]+ 259.5:261.2 (1:1) (258.0). 1H
NMR (CDCl3): 8.04 (1H, s), 6.45 (1H, s), 3.96 (3H,
s), 3.94 (3H, s), 2.56 (3H, s). 13C NMR (CDCl3):
196.3, 160.4, 160.0, 135.1, 121.4, 102.7, 95.7, 56.3,
55.7, 31.7.
5.1.2.8. N,N-Diethyl-2,4-dimethoxybenzamide.8 2,4-
Dimethoxybenzoic acid (3.2 g, 1.8 mmol), thionyl
chloride (HAZARD) (7.8 ml), and dry benzene (HAZ-
ARD) (100 ml) were heated at reflux under stirring for
2 h. Solvent and excess thionyl chloride were removed
under vacuum and the crude acid chloride (3.6 g,
1.8 mmol) was dissolved in benzene (40 ml). This
solution was cooled to 0 ꢁC and diethylamine (5.6 ml,
5.4 mmol) in dry THF (15 ml) was added dropwise to
the stirred solution. After the addition was completed,
the reaction mixture was stirred for 2 h at 0 ꢁC and then
for 8 h at room temperature. The solvents were removed
under reduced pressure and the residue was extracted
with CH2Cl2. The organic layer was thoroughly washed
with 5% Na2CO3, 5% HCl, and water. Removal of
solvent after drying the organic layer afforded the prod-
uct which was used without further purification. 1H
NMR (CDCl3, 300 MHz, d, ppm): 7.13–7.10 (1H, d,
J = 8.3 Hz), 6.50–6.44 (2H, m), 3.81 (3H, s), 3.79 (3H,
s), 3.53 (2H, br), 3.14 (2H, br), 1.22–1.20 (3H, mb),
1.02 (3H, br).
5.1.2.5.
1-[2,4-Dimethoxy-5-(4-methylpiperazin-1-
yl)phenyl]ethanone (7-1). Cesium carbonate (1.4 equiv)
was finely ground in a nitrogen-filled glovebox and
weighed into an oven-dried Schlenk flask. The flask
was quickly capped with a rubber septum and purged
with argon. Pd2 (DBA)3 (tris(dibenzylidene ace-
tone)dipalladium (0), 0.0025 mmol, 0.5% Pd) and
rac-BINAP (rac-2,20-bis(diphenylphosphino)-1,10-binaph-
thyl) (0.0075 mmol) were added into the flask, followed
by 1-(5-bromo-2,4-dimethoxyphenyl)ethanone (1.0 mmol),
1-methylpiperazine (1.1–1.2 mmol), and toluene (4 ml).
The flask was immersed in an oil bath (100 ꢁC), stirred
for 10 h until the starting materials had been completely
consumed as judged by TLC. The solution was then
cooled to room temperature, diluted with ether, filtered,
and concentrated in vacuo. The crude product was puri-
fied by flash chromatography on silica gel to give a yield
of 76%. Mp 75–76 ꢁC, MS-APCI [M+H]+: 279.5 (278.2).
1H NMR (CDCl3): 7.46 (1H, s), 6.44 (1H, s), 3.91 (3H,
s), 3.89 (3H, s), 3.03 (4H, br), 2.61 (4H, br), 2.55 (3H, s),
2.33 (3H, s). 13C NMR (CDCl3): 197.5, 157.3, 157.1,
134.6, 120.1, 119.4, 95.8, 55.75, 55.71, 55.1, 50.7, 45.9,
31.9.
5.1.2.9. N-(2,4-Dimethoxybenzyl)-N-ethylethanamine.
Borane (1 M solution in THF, 15 ml) in THF (15 ml)
was slowly added to a stirred solution of N,N-
diethyl-2,4-dimethoxybenzamide (1.2 g, 5 mmol) in
THF (10 ml) in an ice bath. The reaction was stirred
for 5 h under an atmosphere of argon followed by
dropwise addition of methanol. Evaporation in vacuo
followed by addition of 6 M HCl gave a suspension
that was stirred for 18 h. The mixture was made alka-
line using 2 M NaOH and extracted with ether. The
product was used for the next step of the reaction with-
1
out further purification. H NMR (CDCl3): 7.35–7.33
(1H, d, J = 8.3 Hz), 6.50–6.44 (2H, m), 3.96 (2H, s),
3.81 (3H, s), 3.79 (3H, s), 2.75–2.68 (4H, q), 1.28–
1.23 (6H, t).
5.1.2.6. 1-(2,4-Dimethoxy-5-(4-(piperidin-1-yl)piperi-
din-1-yl)phenyl)ethanone (8-1). The method for 7-1 was
followed using 1,40 –bipiperidine. Compound 8-1 was
obtained in 49% yield, mp 257–258 ꢁC (HCl salt), MS-
5.1.2.10. 1-(2-Hydroxy-4-methoxy-5-methylphenyl)eth-
anone (9-1). N-(2,4-Dimethoxybenzyl)-N-ethylethan-
amine (0.05 mol) was reacted with boron trifluoride
dimethyletherate (0.5 mol) and acetic anhydride
(38 ml) in CH2Cl2 as described in 5.1.2.2. The product
was found to be 9.1, indicating that loss of the diethyl-
1
APCI [M+H]+: 345.2 (346.2). H NMR (MeOD): 8.11
(1H, s), 6.90 (1H, s), 4.12 (3H, s), 4.02 (3H, s), 3.82–
1.55 (22H, m). 13C NMR (MeOD): 197.3, 163.1, 157.2,
124.6, 122.2, 120.7, 98.4, 60.5, 58.0, 56.9, 54.1, 51.4,
48.4, 24.9, 23.7, 22.2.

X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
7033
amino function and demethylation had occurred during
the reaction. Yield: 52%, mp 64–65 ꢁC, MS-APCI
again drawn and minimized using the Molecular Mod-
eling Pro Plus Version 6.2.3 (ChemSW, Fairfield, CA)
forcefield (MM2) for the determination of width,
length, breadth, area, polar surface area, molecular
volume, dipole moment, hydrogen bonding parameters
(H bond donor, H bond acceptor, Hansen H bonding),
% hydrophilic surface, CNDO logP, Crippen logP,
logP, HOMO, LUMO, connectivity indices (0, 1, 2,
3, 4), kappa 2 shape index. The charge on the a and
b carbons was determined from Chem 3D Ultra 10.0
from chalcones drawn and minimized in their un-
charged states. The projection methods (PCA and
PLS) were carried out on SIMCA-P 11 (Umetrics
AB, Umea, Sweden) with default settings. MLR and
correlation analyses were run on SPSS 15.0 (SPSS
Inc., Chicago, IL).
1
[M+H]+: 181.4 (180.1). H NMR (CDCl3): 12.69 (1H,
OH), 7.42 (1H, s), 6.38 (1H, s), 3.85 (3H, s), 2.55 (3H,
s), 2.14 (3H, s). 13C NMR (CDCl3): 202.4, 164.4,
163.8, 131.7, 118.0, 113.0, 98.8, 55.6, 26.2, 15.5.
5.1.3. Syntheses of benzaldehydes. Only 40-(4-methylpi-
perazin-1-yl)benzaldehyde was synthesized. The other
benzaldehydes were purchased from commercial sources
(Sigma–Aldrich Chemical Company).
5.1.3.1. 2-(4-Bromophenyl)-1,3-dioxane. A solution of
4-bromobenzaldehyde (4.3 g, 23 mmol), 1,3-propanediol
(16.8 ml, 0.23 mmol), and p-toluenesulphonic acid
(0.05 g) in toluene (150 ml) was refluxed for 28 h. Water
was removed using a Dean-Stark apparatus. The reac-
tion mixture was poured into 5% Na2CO3 (100 ml)
and washed thoroughly. The organic layer was dried
over K2CO3 and removed in vacuo to give a white solid.
Yield 92%, mp 72–73 ꢁC. MS-APCI: [M+H]+ 243.3
(242.0). 1H NMR (CDCl3): 7.51–7.48 (2H, d,
J = 8.3 Hz), 7.37–7.34 (2H, d, J = 8.3 Hz), 5.46 (1H, s),
4.25–4.23 (2H, m), 4.02–3.93 (2H, m), 2.22–2.15 (1H,
m), 1.58–1.42 (1H, m). 13C NMR (CDCl3): 122.7,
117.5, 114.7, 110.6, 92.8, 65.7, 32.0.
5.3. Microculture tetrazolium assay for determination of
antiproliferative activity
Growth inhibitory activities were determined on the
following cell lines: HCT 116 (transformed human
colon cancer cells, gift from Dr. Balram Chowbay,
National Cancer Centre, Singapore), MCF 7 (human
breast cancer cells, American Type Culture Collection,
Rockville, MD, USA) and CCL 186 (normal human
diploid embryonic lung fibroblasts, American Type
Culture Collection, Rockville, MD, USA). Microcul-
ture tetrazolium assay was based on the ability of met-
abolically viable cells to reduce the yellow tetrazolium
salt (MTT) to colored formazan crystals.12 The proce-
dure was carried out as described earlier.4 Each
compound was tested at no less than eight concentra-
tions, with at least three runs for each concentration.
IC50 values were determined using Prism GraphPad
(San Diego, CA, USA).
5.1.3.2. 40-(4-Methylpiperazin-1-yl)benzaldehyde.6
A
Schlenk flask was charged with 2-(4-bromophenyl)-1,3-
dioxane (1.0 mmol), 1-methylpiperazine (1.1–1.2 mmol),
sodium tert-butoxide (1.4 mmol), tris(dibenzylidene ace-
tone)dipalladium (0) (0.0025 mmol, 0.5% Pd), rac-BIN-
AP (rac-2,20-bis(diphenylphosphino)-1,10-binaphthyl)
(0.0075 mmol), and toluene (2 ml) under argon. The
flask was immersed in an oil bath (80 ꢁC) and stirred un-
til the starting materials had been completely consumed
as judged by TLC (10 h). The solution was then allowed
to cool to room temperature, extracted with CH2Cl2,
and the crude product purified by flash chromatography
on silica gel. Yield: 81 %. Mp 58–60 ꢁC. MS-APCI:
5.4. Cell cycle analysis
Chalcones 1-123, 6-130, and 7-140 at 0.5 and 5 lM were
tested for their effects on the cell cycle of HCT 116. The
procedure described in Ref. 4 was followed.
1
[M+H]+ 205.4 (204.1). H NMR (CDCl3): 9.79 (1H, s,
CHO), 7.77–7.74 (2H, d, J = 8.6 Hz), 6.94–6.91 (2H, d,
J = 8.6 Hz), 3.44–3.40 (2H, t), 2.57–2.54 (2H, t), 1.65
(3H, s). 13C NMR (CDCl3): 190.4, 155.0, 131.8, 127.1,
113.5, 54.6, 47.0, 46.1.
5.5. Statistical analysis
Independent sample t-test run (SPSS 15.0, SPSS Inc.,
Chicago IL) was used to test for significant differences
in IC50 values.
5.2. Molecular modeling and collation of physicochemical
parameters
The pKa values of the various ionizable groups in the
chalcone library were estimated with ACD/pKaDB
found in ACD/I-Lab Version 6.0 for Microsoft Win-
dows. With this information, the chalcones were con-
structed in their predominant ionized states at pH 7.0
using the fragments found in the Sybyl 7.0 (Tripos
Inc., St. Louis, MO). They were geometry optimized
using the standard Tripos force field with a distance
dependent dielectric function until a root mean square
Acknowledgments
Liu XL gratefully acknowledges National University of
Singapore (NUS) for granting her a research scholarship
for graduate studies. This work was supported by Grant
RP 148 000042112 from NUS.
˚
deviation of 0.001 kcal/ mol A was achieved.
Supplementary data
Area, polar volume, ClogP, molar refractivity, partial
positive and partial negative surface areas were com-
puted for these compounds. The compounds were
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.bmc.
2007.07.042.

7034
X. Liu, M.-L. Go / Bioorg. Med. Chem. 15 (2007) 7021–7034
7. Wolfe, J. P.; Buchwald, S. L. J. Org. Chem. 2000, 65, 1144.
References and notes
8. Kamila, S.; Mukherjee, C.; Mondal, S. S.; De, A.
Tetrahedron 2003, 59, 1339.
9. Pharmacokinetics In Silico (phakiso) software, http://
www.phakiso.com/.
1. Xia, Y.; Yang, Z. Y.; Xia, P.; Bastow, K. F.;
Nakanishi, Y.; Lee, K. H. Bioorg. Med. Chem. Lett.
2000, 10, 699.
10. Livingstone, D. In Data Analysis for Chemists;
Oxford University Press: New York, 1995, Chapters
4 and 8.
11. Lawrence, N. J.; McGown, A. T.; Ducki, S.; Hadfield, J.
A. Anti-Cancer Drug Des. 2000, 15, 135.
2. Pati, H. N.; Holt, H. L.; LeBlanc, R.; Dickson, J.; Stewart,
M.; Brown, T.; Lee, M. Med. Chem. Res. 2005, 14, 19.
3. Dimmock, J. R.; Jha, A.; Zello, G. A.; Allen, T. M.;
Santos, C. L.; Balzarini, J.; De Clercq, E.; Manavathu, E.
K.; Stables, J. P. Pharmazie 2003, 58, 227.
12. Alley, M. C.; Scudiero, D. A.; Monks, A.; Hursey, M.
L.; Czerwinski, M. J.; Fine, D. L.; Abbott, B. J.;
Mayo, J. G.; Shoemaker, R. H.; Boyd, M. R. Cancer
Res. 1988, 48, 589.
4. Liu, X. L.; Go, M. L. Bioorg. Med. Chem. 2006, 14, 153.
5. Craig, P. N. J. Med. Chem. 1971, 15, 680.
6. Nielsen, S. F.; Larsen, M.; Boesen, T.; Schonning, K.;
Kromann, H. J. Med. Chem. 2005, 48, 2667.