Antiproliferative activity of aroylacrylic acids. Structure-activity study based on molecular interaction fields

Article abstract of DOI:10.1016/j.ejmech.2011.04.043

Antiproliferative activity of 27 phenyl-substituted 4-aryl-4-oxo-2-butenoic acids (aroylacrylic acids) toward Human cervix carcinoma (HeLa), Human chronic myelogenous leukemia (K562) and Human colon tumor (LS174) cell lines in vitro are reported. Compounds are active toward all examined cell lines. The most active compounds bear two or three branched alkyl or cycloalkyl substituents on phenyl moiety having potencies in low micromolar ranges. One of most potent derivatives arrests the cell cycle at S phase in HeLa cells. The 3D QSAR study, using molecular interaction fields (MIF) and derived alignment independent descriptors (GRIND-2), rationalize the structural characteristics correlated with potency of compounds. Covalent chemistry, most possibly involved in the mode of action of reported compounds, was quantitatively accounted using frontier molecular orbitals. Pharmacophoric pattern of most potent compounds are used as a template for virtual screening, to find similar ones in database of compounds screened against DTP-NCI 60 tumor cell lines. Potency of obtained hits is well predicted.

Full text of DOI:10.1016/j.ejmech.2011.04.043

European Journal of Medicinal Chemistry 46 (2011) 3265e3273
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Antiproliferative activity of aroylacrylic acids. Structure-activity study based
on molecular interaction fields
,
b
ꢀ a
ꢀ b
ꢀ a
ꢁ
ꢁ
ꢁ
*
Branko J. Drakulic , Tatjana P. Stanojkovic , Zeljko S. Zizak , Milan M. Dabovic
a
ꢁ
Department of Chemistry-Institue of Chemistry, Technology and Metallurgy, University of Belgrade, Njegoseva 12, 11000 Belgrade, Serbia
b Institute of Oncology and Radiology of Serbia, Pasterova 14, 11000 Belgrade, Serbia
a r t i c l e i n f o
a b s t r a c t
Article history:
Antiproliferative activity of 27 phenyl-substituted 4-aryl-4-oxo-2-butenoic acids (aroylacrylic acids)
toward Human cervix carcinoma (HeLa), Human chronic myelogenous leukemia (K562) and Human
colon tumor (LS174) cell lines in vitro are reported. Compounds are active toward all examined cell lines.
The most active compounds bear two or three branched alkyl or cycloalkyl substituents on phenyl
moiety having potencies in low micromolar ranges. One of most potent derivatives arrests the cell cycle
at S phase in HeLa cells. The 3D QSAR study, using molecular interaction fields (MIF) and derived
alignment independent descriptors (GRIND-2), rationalize the structural characteristics correlated with
potency of compounds. Covalent chemistry, most possibly involved in the mode of action of reported
compounds, was quantitatively accounted using frontier molecular orbitals. Pharmacophoric pattern of
most potent compounds are used as a template for virtual screening, to find similar ones in database of
compounds screened against DTP-NCI 60 tumor cell lines. Potency of obtained hits is well predicted.
Ó 2011 Elsevier Masson SAS. All rights reserved.
Received 14 February 2011
Received in revised form
9 April 2011
Accepted 13 April 2011
Available online 21 April 2011
Keywords:
4-aryl-4-oxo-2-butenoic acids
Antiproliferative activity
Human tumors
Molecular interaction fields
1. Introduction
covalent inhibition of diverse pharmacological targets trigger our
interest to further evaluate previous findings. Recently reported
First synthesis of parent 4-phenyl-4-oxo-2-butenoic acid (1,
benzoylacrylic acid) has been reported 1882 [1]. Since then number
of congeners was synthesized, and they’re antibacterial activity and
2D QSAR was described [2]. It was shown that ketovinyl part of
molecule (eC(O)eC]Ce) acts as Michel acceptor that covalently
bind reactive sulfhydryl (eSH) groups of biomolecules [3]. Cytem-
bena, 4-(4-methoxyphenyl)-4-oxo-3-bromo-butenoic acid (NSC
104801), structurally similar to aroylacrylic acids have been exten-
sively studied during late sixteen’s of the last century as cytostatic
agent [4], and was discontinued for further usage due to sever side-
effects. In previous article [5] we have reported antiproliferative
activity of nineteen aroylacrylic acids (1e6, 8e11,14, 20e26 as given
in Table 1, and 4eCH3C(O)NH- phenyl substituted derivative)
toward human cervix carcinoma cells (HeLa), after 42 h of exposure
to compounds. The 2D QSAR study shown that potency of
compounds is well correlated with estimated lipophilicity. Even
Michael acceptor comprising molecules are commonly flagged as
less than optimal leads, due to its potential promiscuity and
consequent toxicity, accumulated examples, reported recently [6],
of using ketovinyl moiety as a warhead for directed and selective
suicide inhibitor [7] that selectively target proteosome of HIV
infected cells, bear ketovinyl moiety as covalent warhead. Covalent
inhibitors of epidermal growth factor receptor (EGFR), as Neratinib
[8] and Pelitinib [9], incorporates ketovinyl moiety that target
conserved cystein residue in the ATP binding site, overcoming in this
way resistance to previous generation of non-covalent inhibitors.
Along with this, integrated bioinformatic analysis of National Cancer
Institute inhibition profile for the database [10] of compounds
screened across a panel of tumor cells (NCI60) [11] as possible
targets for ketovinyl containing compounds propose gluthathione
S-transferase, acting in cellular detoxification, as well as NFkB
(nuclear factor kappa-light-chain-enhancer of activated B cells)
involved in cellular response to stress. It should be noted that
number of chalcones and derivatives (suberones and similar),
structurally related to 4-aryl-4-oxo-2-butenoic acids, exert signifi-
cant antiproliferative potency. Majority of compounds within this
class bear HBD/HBA, or polar substituents on phenyl rings (MeOe,
OHe, eNO2, eOCH2Oe, or halogen) [12]. Within our set, alkyl
substituted compounds are most potent.
2. Results and discussion
In this article we report the synthesis, antiprolferative activity of
twenty-seven (E)-4-aryl-4-oxo-2-butenoic acid (1e27) toward
* Corresponding author. Tel.: þ381 113336738; fax: þ381 112636061.
ꢀ
E-mail address: bdrakuli@chem.bg.ac.rs (B.J. Drakulic).
0223-5234/$ e see front matter Ó 2011 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ejmech.2011.04.043

ꢀ
B.J. Drakulic et al. / European Journal of Medicinal Chemistry 46 (2011) 3265e3273
3266
Table 1
cycloalkyl, as in 18 and 19) substituents are more potent than the
Structures and IC50 (mM) values of antiproliferative potency for compounds 1e27,
determined under exactly same conditions.
rest. It should be noted that increases in the number of alkyl
substituents did not progressively contribute to potency. Overall,
within the subset of most potent compounds (12e19), those having
two substituents are more potent than the rest having three
substituents, and most potent compounds have substituents in
positions 2 and 5.
According to in vivo tests, reported by NCI-DTP, compound 9
(NSC 329366) is well tolerated in P388 Leukemia model in CD2F1
(CDF1) mice, by intraperitoneal administration, in concentration up
to 50 mg/kg; while compound 1 (NSC 143) is well tolerated in L1210
Leukemia model in CD2F1 (CDF1) mice by intraperitoneal admin-
istration in concentration up to 25 mg/kg.1
Compound R-
IC50
(
m
M)a
No
HeLa
LS174
K562
1
H-
30.64 (ꢀ4.14) 45.53 (ꢀ6.59) 30.19 (ꢀ0.77)
7.62 (ꢀ3.08) 28.42 (ꢀ2.32) 9.17 (ꢀ2.40)
27.79 (ꢀ3.25) 21.04 (ꢀ4.91) 5.53 (ꢀ0.95)
4.90 (ꢀ1.32) 25.65 (ꢀ2.55) 5.56 (ꢀ0.53)
2
4-Me-
3
4-Et-
Cell cycle on HeLa cells was monitored after 24 h, 48 h and 72 h
treatment of cells with compound 19. Results are given in Fig. 1 and
Table S1 in Supplementary material. Compound 19 arrests the cell
cycle at S phase, as can be seen from data obtained after 24 and
48 h. After 72 h, compound caused extensive damage of cells, as can
be seen from high subG1 band in (Fig. 1c). It is known that resis-
tance and sensitivity of cells toward radiation correlates with the
level of sulfhydryl compounds in the cell. Nonprotein sulfhydryls
(NPS), are natural radioprotector, as well as powerful antioxidative
protector, and tend to be at highest level in S phase of the cell cycle.
Because of that cells exert lowest sensitivity to radiation in this
phase. Ketovinyl containing compounds are powerful thiol
“quenchers”, so this fact is often used as possible explanation for
the potency of such compounds toward malignant cells; ketovinyl
containing compounds bind NPS, and make cells more sensitive to
oxidative stress [16]. Even such scenario coincide with our results,
in the light of knowledge of protein targets, spanning the range
from long time known gluthathione S-transferase to widespread
kinases in last decades, which are prompt to covalent inhibition by
ketovinyl derivatives, we can speculate that quenching of NPS can
only partially explain the mode of action of our compounds.
To obtain deeper insight in structural features that govern
differences in potency, we performed 3D quantitative structure-
activity relationships (3D QSAR). For the 3D QSAR, molecular
interaction fields (MIF) obtained by GRID programme were calcu-
lated using hydrogen bond acceptor (carbonyl oxygen O), hydrogen
bond donor (neutral flat amine, N1), hydrophobic (DRY), and shape
(TIP) [17] probes. Alignment independent descriptors (GRIND-2)
[18], derived from obtained MIF, are subsequently processed by
partial least square analysis (PLS), as applied in Pentacle software
[19], for details see experimental part.
All obtained models have a good statistics (Table 2). In PCA
models, five principal components explain 56e60% variance
(Table S2 in Supplementary material). Antiproliferative potency of
1e27 toward HeLa and LS174 cell lines were explained by 3 latent
variables (LV) models. For the action of compounds toward K562,
4LV model was used.
Three LV PLS coefficient plots for potency of compounds toward
HeLa and LS174 cells are given on Figs. 2a and 2b. Four LV PLS
coefficient plot for potency of compounds toward K562 cells is
given on (Fig. 2c). Associations of variables with compounds are
given in Tables S3e5 in Supplementary material. Experimental vs.
calculated p(IC50) values are given in Table S6 in Supplementary
material.
4
4-i-Pr-
5
6
7
8
4-n-Bu-
4-tert-Bu-
4-n-dodecyl-
2,4-di-Me-
2,5-di-Me-
3,4-di-Me-
4.78 (ꢀ1.45) 6.74 (ꢀ2.00)
5.10 (ꢀ0.82) 8.74 (ꢀ1.57)
5.82 (ꢀ0.66) 7.85 (ꢀ1.15)
6.06 (ꢀ1.18) 7.08 (ꢀ2.30)
4.95 (ꢀ0.59)
4.33 (ꢀ0.45)
4.01 (ꢀ0.56)
5.35 (ꢀ0.29)
9
7.86 (ꢀ0.84) 21.98 (ꢀ1.90) 5.90 (ꢀ0.28)
7.28 (ꢀ1.37) 12.11 (ꢀ5.74) 5.62 (ꢀ0.49)
7.01 (ꢀ0.90) 17.58 (ꢀ2.59) 5.60 (ꢀ0.51)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
b
-tetralinyl-
2-Et-5-i-Pr-
2,4-di-i-Pr-
2,5-di-i-Pr-
2,4,6-tri-Et-
2,4,6-tri-i-Pr-
2-Et-3,5-di-i-Pr-
4.89 (ꢀ0.46) 5.94 (0.594)
4.06 (ꢀ0.97) 5.36 (ꢀ0.56)
4.75 (ꢀ0.51) 6.23 (ꢀ0.33)
6.53 (ꢀ1.55) 7.28 (ꢀ1.77)
6.10 (ꢀ1.40) 8.28 (ꢀ1.54)
1.36 (ꢀ0.59)
4.68 (ꢀ1.61)
3.28 (ꢀ1.18)
4.01 (ꢀ0.89)
4.67 (ꢀ0.09)
7.61 (ꢀ1.87) 12.35 (ꢀ4.88) 5.10 (ꢀ0.14)
4-Ch-2,5-di-Me- 2.38 (ꢀ0.65) 6.64 (ꢀ1.20)
4.24 (ꢀ0.23)
3.32 (ꢀ0.63)
2,5-di-Ch-
2-Cl-4-Me-
4-F-
1.26 (ꢀ0.02) 4.28 (ꢀ0.60)
7.60 (ꢀ1.73) 27.69 (ꢀ6.99) 14.59 (ꢀ6.52)
28.41 (ꢀ5.29) 33.78 (ꢀ3.37) 16.90 (ꢀ5.12)
25.42 (ꢀ2.43) 28.65 (ꢀ6.12) 21.66 (ꢀ4.36)
18.68 (ꢀ4.20) 27.12 (ꢀ4.14) 11.19 (ꢀ2.15)
4-Cl-
4-Br-
2,4-di-Cl-
3,4-di-Cl-
4-MeO-
62.94 (ꢀ5.50) >100
58.47 (ꢀ12.23)
28.30 (ꢀ3.66) 32.33 (ꢀ3.65) 23.15 (ꢀ9.67)
27.36 (ꢀ7.15) 34.74 (ꢀ11.41) 12.32 (ꢀ9.21)
2,3,4-tri-MeO-
96.18 (ꢀ2.16) > 100
> 100
a
Mean of three measurements, each performed in triplicate.
three human tumor cell lines, and 3D QSAR, aimed to rationalize
structural characteristics needed for significant potency.
Compounds were obtained by Friedel-Crafts acylation of aromatic
substrates with maleic acid anhydride in good yields (Scheme 1).
Structures of compounds 1e27 are given in Table 1. Even known
from literature data, stereochemistry on ketovinyl double bond is
proved by single crystal x-ray structure data for compound 1 [13],
and 13 [14] (Figure S1 in Supplementary material), as well as from
the coupling constants of (keto)vinyl protons in 1H NMR spectra for
all derivatives. Vinyl protons of derivatives having E-stereochem-
istry show coupling constants of w16 Hz [15].
Antiproliferative activity of compounds was assessed toward
Human cervix carcinoma (HeLa), Human chronic myelogenous
leukemia (K562) and Human colon tumor (LS174) cell lines in vitro.
The majority of compounds exert potency toward all tested cell
lines in two-digit to low micromolar concentrations (Table 1).
Brief structure-activity for potency of 1e27 is as follows:
Halogen- and methoxy- substituted compounds are less potent than
alkyl-substituted congeners. Within group of alkyl-substituted
compounds, those having two, or three branched alkyl (or
1
In NCI-DTP database (http://dtp.nci.nih.gov/dtpstandard/dwindex/index.jsp)
compounds 1 (NSC 143), 7 (NSC 30659), 8 (NSC 31589), 9 (NSC 329366), 22 (NSC
32875), 23 (NSC 39971), and 26 (NSC 25331) were recorded. NCI60 cell line
screening data were not performed for those compounds. Only yeast screening
data, and in vivo screening data for 1 and 9 exists.
Scheme 1. Syntheses of 1e27, R is as given in Table 1.

ꢀ
B.J. Drakulic et al. / European Journal of Medicinal Chemistry 46 (2011) 3265e3273
3267
Fig. 1. Effect of 19 on cell cycle distribution in HeLa cells, after a) 24 h, b) 48 h, and c) 72 h treatment of cells with compound.
with the bulkier substituent in position 3- or 4- of the phenyl
ring, as is illustrated on Fig. 3b for compound 18, K562 model.
ꢁ In HeLa model, variable DRY-O 403, positively correlated with
potency, gives almost clear-cut between more potent alkyl
As can be seen from Table 1, the order of potency of 1e27 toward
tested cell lines are similar, but not the same, so appearance of
common variables having high positive or negative impact on
models, within all three PLS coefficient plots, are briefly outlined to
explain how structural variation govern differences of potency
among compounds.
ꢁ Compounds most potent toward all cell lines tested have
branched alkyl substituents in positions 2 and 5 of the phenyl
ring. Variables DRY-DRY 36, 37 and 39, for K562, LS174, and
HeLa model, respectively, positively correlated with the
potency, even haven’t highest impact on overall models, nicely
distinguishes most potent compounds from less potent ones.
Those variables connect nodes of the DRY field interacting with
the ketovinyl eC]Ce, and with the 5 alkyl (or cycloalkyl)
substituent (spatial distance of DRY nodes are 11.50e12.80 Å),
as illustrated on Fig. 3a, for compound 13, HeLa model.
ꢁ Variable TIP-TIP 311 in K562 and HeLa model, is expressed for
all compounds, except for 4-n-dodecyl derivative (7), dis-
tinguishing in this way compound that have a significantly
longer 4-alkyl substituent.
ꢁ Variables TIP-TIP 319 and 320, for K562 and HeLa model,
respectively, negatively correlated with potency are expressed
for majority of compounds having 4-alkyl substituent. This
emphasize that 2,5-di-alkyl substituted compounds are more
potent than those that bear substituent in position 4, yet this is
bulky alkyl substituent. Variable connect nodes of TYP probe on
distance 16.00e17.28 Å, associated with the carboxyl group and
Table 2
3LV PLS models for the potency of compounds toward HeLa and LS174 cells, and 4LV
PLS models for the potency of compounds toward K562 cells.
Component SSX SSXa SDEC SDEP R2 R2acc Q2acc LOO Qa2cc LTO Qa2cc RG
HeLa PLS model
1
2
3
23.95 23.95 0.27 0.34 0.61 0.61 0.41
25.58 49.53 0.18 0.29 0.22 0.83 0.56
8.12 57.65 0.13 0.28 0.08 0.91 0.61
LS174 PLS model
0.41
0.56
0.61
0.39
0.54
0.59
1
2
3
29.80 29.80 0.20 0.23 0.59 0.59 0.53
27.66 57.46 0.13 0.18 0.23 0.82 0.68
8.07 65.53 0.10 0.18 0.08 0.90 0.66
K562 PLS model
0.52
0.68
0.66
0.49
0.68
0.67
1
2
3
4
30.34 30.34 0.25 0.29 0.48 0.48 0.36
22.17 52.51 0.17 0.24 0.27 0.74 0.53
7.11 59.62 0.13 0.24 0.12 0.86 0.56
4.03 63.66 0.08 0.24 0.08 0.95 0.55
0.36
0.52
0.55
0.55
0.31
0.51
0.51
0.52
SSX - Percentage of the X sum of squares; SSXacc - accumulative percentage of the X
sum of squares; SDEP - standard deviation of error of the predictions; R2 - coefficient
of determination; R2acc - accumulative coefficient of determination; Q2acc - accu-
mulative squared predictive correlation coefficient; LOO - leave one out, LTO - leave
two out; RG - random groups (three random group of approximately same size were
used).
Fig. 2. 3LV PLS coefficient plots for the model for potency of compounds toward HeLa
a), and LS174 b) cells, and 4LV PLS coefficient plots for the model for potency
compounds toward K562 cells, c).

ꢀ
B.J. Drakulic et al. / European Journal of Medicinal Chemistry 46 (2011) 3265e3273
3268
Fig. 3. Examples of variables associated with compounds, as described in the text. a) Variable DRY-DRY 39 for compound 13, HeLa model. b) Variable TIP-TIP 320 for compound 18,
K562 model. c) Variable DRY-O 403 for compound 14, HeLa model. d) Variable DRY-N1 454 for compound 15, HeLa model. e) Variable DRY-TIP 586 for compound 19, LS174 model.
f) Variable O-TIP 768 for compound 18, HeLa model.
substituted compounds, bearing bulkier substituents, from
other, less potent. Node of O probe is always associated with
carboxyl eOH, while DRY probe nodes, on distance
13.76e14.08 Å, are associated with meta or para substituents.
This is illustrated on Fig. 3c, for compound 14, HeLa model. In
K562 model similar variable, DRY-O 404, appear.
ꢁ Variable DRY-N1 454, and 459, for HeLa and LS174 models,
respectively, has highest positive impact on model. Variable
connect nodes of the hydrophobic DRY probe above phenyl ring
and nodes of hydrogen bond donor probe (N1) interacting with
aroyl oxygen, on short distance of 1.28e2.24 Å, as is illustrated
on Fig. 3d, for compound 15, HeLa model. Variable is expressed
for most potent compounds, that have bulky alkyl substituents
in ortho position. Such substituents cause significant torsion
between phenyl moiety and aroyl keto group. Similarly, long
time ago in literature have been emphasized that aroylacrylic
acids and derivatives bearing ortho alkyl substituents are more
potent antibacterial agents than those which have different
substitution pattern on the phenyl ring [2b].
ꢁ Next variables positively correlated with potency, in all models,
are DRY-TIP 583, 585 and 586 for HeLa, K562 and LS174 models,
respectively. Those variables well separate more potent
compounds from less potent ones. TIP node is associated
with eCOOH group, common for all compounds, while DRY
nodes, on distance 13.76e15.04 Å, are associated with bulkier
substituents in positions 2, 3 or 4. This is illustrated on Fig. 3e,
for compound 19, LS174 model.
ꢁ Variables O-TIP 766 and 768, for K562 and HeLa models,
respectively, negatively correlated with potency, similar as
TIP-TIP 320, in majority are not expressed for compounds
having unsubstituted position 4. Variable is illustrated on
Fig. 3f, for compound 18, HeLa model. In HeLa model variable
O-TIP 766 distinguish 2,5-disubstituted derivatives having
medium-size alkyl substituents (9 and 12) from the rest of

ꢀ
B.J. Drakulic et al. / European Journal of Medicinal Chemistry 46 (2011) 3265e3273
3269
compounds. O node is associated with eCOOH group, while TIP
nodes, at distance 14.40e15.36 Å, are associated with substit-
uents in positions 3- or 4-.
ketovinyl moiety e target of nucleophilic attack in molecules, as is
exemplified on Fig. 4 for compound 13.
13C Chemical shifts of the carbon atom 2 of butenoic moiety, that
should be target of nucleophilic attack, or other carbons of keto-
vinyl moiety, extracted from HSQC spectra of most potent 12e19
did not show any correlation with potency data. This is not
surprising because compounds act on whole cells, involving
permeation in cell interior, reaching the potential intracellular
target(s), and interaction with those targets. It should be noted that
even on model system, introduction of ortho substituents in series
of compounds structurally related to 1e27, significantly deviate
regularity of the change of electron density on carbon atom that act
as a target for nucleophile attack, in respect to other compounds
having para substituents, and for which good structure-activity
relationship can be obtained using phenyl-substituents effect
only [22]. Cited example describe variation of substituents bearing
heteroatoms directly connected to phenyl ring, so having much
stronger resonance effect than alkyl groups in our compounds.
In the next step we search NCI-DTP database for compounds
active toward K562 cells that comprise ketovinyl moiety. Molecular
mass of hits is limited to 400 (about 20% higher than mass of largest
19), and chalcones and dibenzalacetones are excluded to preclude
appearance of trivial results in virtual screening. Sixty-three
compounds were collected in this way (Table S9 in Supplemen-
tary material), and database was made in Pentacle virtual screening
(VS) mode, 19 principal components (PC) explain 85% of variance.
Three compounds (13, 14 and 19) were used as templates for VS.
Database was “contaminated” with compound 14, to test which
method give best results for similarity search. Minimum distance
was found as most suitable method, this is in accordance with data
found in original article that describes Pentacle VS module [23]. To
all PC’s are ascribed same weights and VS were done with all 19
PC’s. First three compounds that are identified as most similar to
templates are: 1. NSC 145150 (CAS 70857-52-2, NCI 60 cell lines
mean pGI50 ¼ 4.99, K562 pGI50 ¼ 5.0); 2. NSC 672120 (NCI 60 cell
lines mean pGI50 ¼ 5.90, K562 pGI50 ¼ 6.8); and 3. NSC 321477 (CAS
28644-86-2, Zexbrevin, NCI 60 cell lines mean pGI50 ¼ 5.24, K562
pGI50 ¼ 5.4). Structures of these compounds are shown on Fig. 5.
All compounds are proved as active toward all NCI60 cell lines,
Along with the antiproliferative activity, NSC 145150 is described as
antibiotic [24]; while NSC 321477 is sesquiterpene lactone isolated
from Zexmenia brevifolia and also exerts antibacterial and antifungal
activity. As instant version of Pentacle does not offer visualization of
variables most important for obtained similarity, we rebuilt whole
model in Pentacle standard mode, to obtain 19 principal compo-
nents PCA model. Such model explains same variance as in VS mode,
so should be considered as valid. Compounds 13, 14 and 19 were
added to this set, and variables common to both NSC 145150, NSC
Other variables having significant impact on models (those
having significant positive or negative intensity on Fig. 2a, b and c,
but not assigned by numbers) were not commented, because those
variables are expressed for all compounds, or only distinguish
compound 7 from the rest, as is exemplified with the variable
TIP-TIP 311 in K562 and HeLa model.
In the next step we try to find out whether 3D-dependent whole
molecular descriptors show correlation with the potency of
compounds. Therefore, we calculated virtual logP [20], molecular
surfaces (solvent accessible, polar, apolar), volume, as well as iso-
surfaces obtained by GRID DRY, OH2 and H probe on visually
chosen isoenergetic values (Table S7 in Supplementary material).
The best correlation for potency of compounds toward all three cell
lines are obtained between p(IC50) values and ratio of apolar to total
surface area (r ¼ 0.74 (HeLa), 0.77 (K562); 0.86 (LS174)). Other
linear correlations that include up to 3 descriptors (MLR), yield
significantly inferior statistical results.
We speculate that our compounds exert biological activity by
Michael-type addition of thiol and amino group of enzymes amino
acid side chains. On the ground of literature data and experiments
performed in our laboratory [21], aroylacrylic acids and derivatives
make stable, non-reversible adducts with relatively simple thiols
and amines, acting exclusively as unsaturated ketones, not as
unsaturated acids. All descriptors described so far do not account
for “covalent chemistry”, most probably involved in the mode of
biological action, so we extracted highest occupied molecular
orbital (HOMO) and lowest unoccupied molecular orbital (LUMO)
energies from MOPAC output files (Table S8 in Supplementary
material), in attempt to include those data in our correlations. By
using whole-molecular 3D descriptors (as described in previous
paragraph), and HOMO and LUMO energies, fair, statistically
significant correlations, that include ASA/SA ratio and energy of
LUMO orbitals, appear for the potency of compounds toward HeLa
and K562 cells:
pðIC50ÞHeLa ¼ ꢂ 3:064ðꢀ1:21ÞðASA=SAÞ þ 2:034ðꢀ1:95ÞLUMO
þ 12:74ðꢀ3:30Þ
ð1Þ
(n ¼ 27; r ¼ 0.789; s ¼ 0.282; F ¼ 19.755; Q2 ¼ 0.529;
sPRESS ¼ 0.315)
Exclusion of compound 27 gives something better statistics
(r ¼ 0.835; sd ¼ 0.230; F ¼ 26.537; Q2 ¼ 0.595).
pðIC50ÞK562 ¼ ꢂ 2:509ðꢀ0:89ÞðASA=SAÞ þ 1:727ðꢀ1:44ÞLUMO
þ 11:56ðꢀ2:45Þ
ð2Þ
(n ¼ 26; r ¼ 0.825; s ¼ 0.208; F ¼ 24.560; Q2 ¼ 0.574;
sPRESS ¼ 0.240)
Exclusion of compounds 12 and 24 gives something better
statistics (r ¼ 0.905; sd ¼ 0.124; F ¼ 47.291; Q2 ¼ 0.746). For LS174
cells similar correlation could be obtained, but such correlation lack
statistical significance.
Even ASA/SA term in Equations (1) and (2) lack in-dept struc-
ture-activity trends, as were obtained by Pentacle models (obvious
that, within set of most potent 12e19, molecules having two
branched substituents exert better potency than those with three
such substituents, so potency was not collinear with increase of the
ratio between apolar and total surface area), it should be noted that
both equations include LUMO energies, which are often associated
with the susceptibility of molecules to nucleophilic attack. Along
with this, for each molecule LUMO orbital is associated with
Fig. 4. The LUMO of compound 13, ꢂ1.68 eV.

ꢀ
3270
B.J. Drakulic et al. / European Journal of Medicinal Chemistry 46 (2011) 3265e3273
Fig. 5. Structures of a) NSC 145150, b) NSC 672120, and c) NSC 321477.
321477, NSC 672120, and 13, 14 and 19 were found manually,
searching regions of variables space near the variables which have
highest impact on K562 model for 1e27, as described above. In this
way four variables, which reflects similarity among compounds,
were chosen:2 TIP-TIP covering distance 13.20e13.60 Å, illustrated
on Figure S2a and b in Supplementary material, comparable with
TIP-TIP 311 (13.12e13.44 Å); DRY-N1 covering distance
10.80e11.20 Å, illustrated on Figure S3a and b in Supplementary
material, comparable with DRY-N1 495 (14.40e14.72 Å); DRY-TIP
covering distance 11.20e11.60 Å, illustrated on Figure S4a and b in
Supplementary material, comparable with DRY-TIP (14.40e14.72 Å);
as well as DRY-DRY covering distance 4.80e5.20 Å, which cannot be
found in models for potency of 1e27 toward studied cell lines, but
nicely show fields associated with both activated double bond and
proximal hydrophobic region of molecules, common to all mole-
cules compared. This variable is illustrated on Figure S5a and b in
Supplementary material. To illustrate similarity between MIF’s of
hits and template molecule, hot spots, as obtained by Pentacle, for
14, NSC 145150, NSC 321477, and NSC 672120 are given in Supple-
mentary as 3D files. Finally, we projected hits of our similarity search
(NSC 145150, NSC 321477, NSC 672120) on the K562 model, using
Simple MLR, that include LUMO energies indicate “covalent chem-
istry” involved in the mode of action of compounds, as is expected for
ketovinyl containing derivatives. Similarity search performed by
pharmacophore pattern comparison, using molecular interaction
fields and derived GRIND-2 descriptors, identify similar compounds
in NCI-DTP database, that contain ketovinyl moiety and exert
potency toward K562 cells. The model obtained for 1e27 very well
predicted potency of two of those compounds.
4. Experimental
4.1. Chemistry
All chemicals were purchased from Aldrich, Fluka or Merck,
having >98% purity and were used as received. For synthesis of
1e27 the dry methylene-chloride was stored over molecular sieves.
Separation of 12 was done on ICN 12e26, 60 Å silica gel, using two
Büchi C-601 pumps supplied with C-615 pump manager. Melting
points were determinate on Büchi apparatus in open capillary
tubes and are uncorrected. The IR spectra are recorded on Thermo
Nicollet 6700 FT-IR spectrophotometer (KBr pallet or ATR). 1H and
13C NMR spectra are recorded on Varian Gemini 200 spectrometer
on 200/50 MHz, or Bruker AVANCE on 500/125 MHz in CDCl3, or
DMSO-d6, using tetramethylsilane (TMS) as internal standard.
HSQC (Heteronuclear Correlation spectra), of 12e19 were recorded
on Bruker AVANCE spectrometer. All chemical shifts are reported in
ppm downfield from TMS. The high-resolution mass spectra are
recorded on Agilent 6200 LC/ESI-MS TOF instrument in methanol,
with cone voltage of 70 V.
3
pGI50 values to describe potency for NSC compounds. Good
prediction were obtained with 4LV (r2 ¼ 0.937, SDEP ¼ 0.05), for NSC
145150 (experimental 5.00, predicted 5.004), and NSC 321477
(experimental 5.40, predicted 5.33); while prediction failed for NSC
672120, for which pGI50 lay out of the potency range of 1e27.
3. Conclusion
(E)-4-Aryl-4-oxo-2-butenoic acids exert antiproliferative activity
against human malignant cells in vitro. Twenty-seven congeners
tested toward Human cervix carcinma (HeLa), Human chronic
myelogenous leukemia (K562) and Human colon tumor (LS174) cells
are potent in two-digit micromolar to low micromolar concentra-
tions. Representative compound arrests cell in the S phase of the cell
cycle. Most potent compounds bear branched alkyl substituents in
positions 2 and 5 of the phenyl ring. 3D structure-activity analysis
rationalizes structural requirements for the significant potency.
4.1.1. Synthesis
Compounds (1e27) are synthesized as described in literature
[25], by Friedel-Crafts acylation of substituted benzenes with
maleic acid anhydride using anhydrous AlCl3 as the catalyst in dry
CH2Cl2. The 6.125 g (62.5 mmol) of maleic acid anhydride, 16.5 g
(125 mmol) of anhydrous AlCl3 and 62.5 mmol of aromatic
substrates were used. After 4e6 h of stirring on room temperature
the reaction mixture was captured with ice/conc. HCl. The solvent
was removed by steam distillation. The residuals was filtered on
vacuum, or extracted by appropriate solvent. In this way obtained
solids or oily substances are processed by solution of Na2CO3 to pH
8.5, filtered to remove traces of Al from catalyst, neutralized by
dropwise addition of conc. HCl, filtered, than washed thoroughly by
water, dried and crystallized from appropriate solvent. Yields of
pure products were from 40 to 92%. The aromatic hydrocarbons for
2
Please be aware that models for potency of 1e27 were made using grid reso-
lution of 0.4 Å, while VS was done using grid resolution of 0.5 Å.
3
NCI-DTP pGI50 was obtained after 48 h exposure of target cells to compounds,
while pIC50 was obtained after 72 h exposure of target cells to compounds. For
other differences see experimental, and http://dtp.nci.nih.gov/branches/btb/ivclsp.
html.

ꢀ
B.J. Drakulic et al. / European Journal of Medicinal Chemistry 46 (2011) 3265e3273
3271
synthesis of 12 and 17 were obtained as follows: Acylation of
commercially available alkyl-substituted benzenes by acetyl chlo-
ride gives ketones that were purified by distillation under reduced
pressure, and characterized by NMR spectroscopy. Pure ketones
were reduced by Huang-Minlon procedure [26], purified by distil-
lation, characterized by NMR spectroscopy and acylated as
described above. Synthesis of 12 and 17 were started from 30 mmol
of aromatic precursors. Compound 17 was isolated as a pure
product. Compound 12 was obtained as the mixture of w20% of 2-i-
Pr-5-Et- derivative and w80% 2-Et-5-i-Pr- derivative (estimated by
the ratio under 1H NMR signals) which is purified by medium
pressure column chromatography using hexanes/AcOEt/AcOH 4: 1:
0.2 until 2-Et-5-i-Pr- derivative gives clear NMR spectra (w45%
yield calculated on initial mixture). The minor derivative was dis-
carded. 1,4-di-Me-2-cyclohexylbenzene was prepared according to
described procedure [27], starting with 50 mmol of DCC. The pure
product was isolated after vacuum distillation in 65% yield. Acyla-
tion, as described above gives pure 18. Full characterization data of
compounds (1e27) are given in Supplementary material.
24 h, 48 h and 72 h (in concentration of two IC50 values, where IC50
was as obtained after 72 h of incubation with compound, as
described above), were fixed in 70% ethanol on ice for the one week,
and then centrifuged. The pellet was treated with RNase A
(100 m mg/mL
g/mL) at 37 ꢃC for 30 min, and then incubated with 40
of propidium iodide (PI) for at least 30 min. Cells were analyzed
using a FACSCalibur flow cytometer (BD Biosciences Franklin Lakes,
NJ, USA), equipped with a 15 mW, air-cooled 488 nm argon ion laser
for excitation of PI. The PI fluorescence (FL2) was collected after
passing a 585/42-nm band pass filter. FACSCalibur flow cytometer
equipped with an FL2 upgraded doublet discrimination module
(DDM), allows screening, and then excluding possible occurrence of
cell doublets, clumps and debris, by plotting FL2-area versus FL2-
width signals [30]. PI fluorescence data was collected using linear
amplification. For the each sample 20,000 events was collected.
Finally, data were analyzed using CELLQuest 3.2.1.f1 software (BD
Biosciences).
4.3. Modeling
4.2. Biological procedures
For modeling purposes IC50, in molar concentrations, were
converted to corresponding negative decade logarithms, p(IC50).
The SMILES notation of 1e27 were converted to 3D by OMEGA [31],
using MM94FF [32], and further optimized by MOPAC 2009 to root-
Stock solutions of 1e27 were made in dimethylsulfoxide (Fluka
Chemie AG Buchs, Switzerland) at a concentration of 10 mM,
filtered through Millipore filter (0.22
m
M), before use, and after-
mean square gradient of 0.001 kcal molꢂ1 ꢂ1, by semiempirical
A
wards diluted to various working concentrations with RPMI-1640
nutrient medium (Sigma Chemical Co. St Louis, MO) supple-
mented with 3 mmol/L L-glutamine, 100 mg/mL streptomycin,
100 IU/mL penicillin, 10% heat inactivated fetal bovine serum (FBS -
Sigma Chemical Co.), and 25 mM Hepes, adjusted to pH 7.2 by
bicarbonate solution.
MO PM6 method [33], using implicit solvation in water (COSMO)
[34]. During optimization constraint is imposed on ketovinyl
moiety. VegaZZ [35] was used as GUI. All optimized structures were
saved in mol2 format and submitted to Pentacle software for an
alignment-free 3D QSAR analysis. Molecular interaction fields are
computed using built-in GRID program [36], with grid resolution of
0.4 Å. AMANDA algorithm were used for extraction of hot spots
(nodes) from obtained MIFs’ (discretization); distances and relative
position of nodes were described by CLAC algorithm (encoding),
suitable for examination of the group of congeneric compounds.
Five principal components/latent variables were initially used for
both principal component analysis (PCA) and partial least square
(PLS) model. Final PLS models were reported with three latent
variables (LV) for the potency of compounds toward HeLa and LS174
cells, and with 4LV for potency of compounds toward K562 cells.
Compounds extracted from NCI-DTP database were saved in
SMILES format (including stereochemistry assignation), converted
to 3D by OMEGA, and optimized on semiempirical MO level, using
same settings as for 1e27. Database were built in Pentacle virtual
screening mode (VS), using default settings: grid resolution 0.5 Å,
DRY, O, N1 and TIP probes, AMANDA algorithm for extraction of the
hot spots, and MACC algorithm for encoding.
4.2.1. Treatment of HeLa, LS174 and K562 cells
HeLa and LS174 cells were seeded into 96-well microtiter plates
(2000 and 5000 cells, respectively, in 0.1 mL of nutrient medium per
well); K562 cells were maintained as suspension culture. After 20 h,
to wells with HeLa and LS174 cells, five different concentrations of
1e27 were applied, except to the control wells, the wells with cells
grown in a nutrient medium only. To K562 cells (3000 cells per
well), investigated compounds were added 2 h after the cell seeding.
All concentrations were set up in triplicate. Nutrient medium with
corresponding concentrations of investigated compounds, but
without cells, was used as a blank, also in triplicate.
4.2.2. Determination of cell survival
HeLa, LS174 and K562 cell survival, was determined by MTT test,
according to Mosmann [28], with modification by Ohno and Abe
[29], 72 h upon addition of the compound. Briefly, 20
m
L of MTT
The 3D dependent whole molecular properties, molecular
surfaces (solvent accessible, polar, apolar), volume and virtual logP
are calculated by VegaZZ 2.4.0 [35]. BICUBE [37] was used to extract
isosurfaces from molecular interaction fields, obtained by GRID
hydrophobic (DRY), hydrogen bond donor/hydrogen bond acceptor
(OH2), and H probes. The last probe is used to characterize
molecular shape, while DRY and OH2 probes are chosen to char-
acterize hydrophobic and polar part of molecules, respectively.
HOMO and LUMO energies were extracted from MOPAC output
files, and orbitals are visualized by Jmol [38]. Multiple linear
regression (MLR) analysis was performed by BILIN program [39]. All
calculations were done on AMD dual core ꢄ64 processor (5 GHz) in
Windows or Linux environments.
solution (5 mg/mL PBS) were added to each well. Samples were
incubated for further 4 h at 37 ꢃC in 5% CO2 and humidified air
atmosphere. Then, 100
mL of 10% SDS were added to the wells.
Absorbance at 570 nm was measured the next day. To get cell
survival (%), absorbance of a sample with cells grown in the pres-
ence of various concentration of 1e27 was divided by the control
absorbance (the absorbance of sample with cells grown in nutrient
medium only), and multiplied by 100. It was implied that absor-
bance of blank was always subtracted from absorbance of a corre-
sponding sample with target cells. IC50 concentration was given as
the concentration of agent that inhibits cell survival by 50%,
compared with vehicle-treated control. All IC50’s were reported as
a mean of three measurements, each done in triplicate.
4.3.1. GRIND methodology
4.2.3. Cell cycle determination
Program Pentacle uses alignment independent descriptors
derived from GRID molecular interaction fields (MIF). More
negative value of GRID MIF for any used probe corresponds to
Cell cycle determination was done on HeLa cells. Aliquots of
5 ꢄ 105 control cells, or cells treated with compound 19, during

ꢀ
B.J. Drakulic et al. / European Journal of Medicinal Chemistry 46 (2011) 3265e3273
3272
more favorable interaction between a probe (e.g. hydrogen bond
donor, hydrogen bond acceptor, hydrophobic) and a molecule for
which the GRID MIF is calculated. By calculating MIFs for
different GRID probes around a molecule and extracting most
relevant regions one can obtain a fingerprint of a receptor to
which small molecule could fit well. These regions show favor-
able energy of interaction and represent positions where groups
of a potential receptor would interact favorably with a ligand.
Such MIF pattern can be described as the virtual receptor site
(VRS). The each GRIND descriptor consists of two nodes extracted
from MIFs and encodes their energy product and the spatial
distance. GRIND variables represent geometrical relationships
between relevant pharmacophoric points around studied mole-
cules, which are entirely invariable to position of molecule(s) in
space and alignment of molecules. Derivation of GRIND
descriptors includes next steps: (i) computing a set of MIF around
studied molecules, (ii) filtering the MIF, to extract the most
relevant regions that define the VRS, and (iii) encoding the VRS
into the GRIND variables. In our models the CLAC algorithm,
suitable for examination of the set of congeneric compounds, is
used. GRIND variables can be used for comparison of molecules
and their classification within sets of structurally diverse entities,
and Pentacle program use principal component analysis (PCA) for
this type of analysis. A dependent variable (such is biological
activity of a certain type) can be correlated to GRIND descriptors
(as independent variables), obtained on a set of molecules, by
partial least square analysis (PLS). Most intensive bars in the PLS
plots have the highest impact on a model. Bars having positive
values on y scale represent variables positively correlated with
activity, while those having negative values on y scale are
negatively correlated with activity. Within the each block (auto-
or cross-correlograms, that correspond to pairs of nodes of
a same or a different probe, respectively) variables are arranged
from left to right on the x scale of the plot, according to
ascending distance between their nodes. In addition to the
spatial arrangement of molecules and nodes encoded in GRIND
variables, each node of each variable exert specific energy of
interaction with a target molecule. Therefore, the strength of
interaction between respective GRID probe in particular node
and molecules are accounted in addition to the spatial positions
of VRS regions. Pentacle use GRIND-2, the second generation of
GRID-based alignment independent descriptors. For more infor-
mation see reference [18b] and http://cadd.imim.es/grib-cadd/
projects/pentacle.
References
[1] H. von Peschman, Ueber Condensationsprodukte zweibasischer Fettsäuren,
Ber. Dtsch. Chem. Ges 15 (1882) 881e892.
[2] a) H. Rinderknecht, J.L. Ward, F. Bergel, A.L. Morrison, Studies on antibiotics. II.
Bacteriological activity and possible mode of action of certain nonnitrogenous
natural and synthetic antibiotics, Biochem. J. 41 (1947) 463e469;
b) B.J. Cramer, W. Schroeder, W.J. Moran, C.H. Nield, M. Edwards, C.I. Jarowski,
B. Puetzer, The antibacterial activity of some b-aroyl-acrylic acids, esters, and
amides, J. Am. Pharm. Assoc. Sci. 37 (1948) 439e449;
c) F.K. Kirchner, J.H. Bailey, C.J. Cavallito, Ring substituted benzoylacrylic acids
as antibacterial agents, J. Amer. Chem. Soc. 71 (1949) 1210e1213;
d) N.H. Cromwell, P.L. Creger, K.E. Cook, Studies with the amine adducts of
b-benzoylacrylic acid and its methyl ester, J. Amer. Chem. Soc. 78 (1956)
4412e4416;
e) A. Dal Pozzo, M. Acquasaliante, G. Donezzeli, P. De Maria, C. Nicoli, 3-Car-
bonylacrylic derivatives as potential antimicrobial agents. Correlations
between activity and reactivity toward cysteine, J. Med. Chem. 30 (1987)
1674e1677;
f) K. Bowden, M.S. Dixon, J.R. Runson, Structure-activity relations. Part 4.
Reactivity and anti-bacterial activity of 3-aroylacrylic acids and their methyl
esters, J. Chem. Res. Synopses (1979) 8;
g) K. Bowden, A. Dal Pozzo, C.K. Duah, Structure-activity relations. Part 5.
Antibacterial activity of a series of substituted (E)-3-(4-phenylbenzoyl)acrylic
acids, -chalcones, -2-hydroxychalcones and -a-bromochalcones; addition of
cysteine to substituted 3-benzoylacrylic acids and related compounds,
J. Chem. Res. Synopses (1990) 2801e2830;
h) K. Bowden, M.P. Henry, Structure-activity relations. II. Antibacterial activity
of 3-benzoylacrylic acids and esters, ACS Advances in Chemistry Series, Biol.
Correl.-Hansch Approach, Symp. 114 (1972) 130e140.
[3] a) B.M. Anderson, M.L. Tanhcoo, A. Dal Pozzo, Selective enzyme inactivation by
benzoylacrylic acid derivatives, Biochem. Biophys. Acta 703 (1982) 204e211;
b) B.M. Anderson, C.D. Anderson, G. Donzzeli, A. Dal Pozzo, Sulfhydryl enzyme
inactivation by nicotinoylacrylates, Biochem. Biophys. Acta 787 (1984) 215e220.
[4] a) O. Dvorak, J. Venta, M. Semonsky, Report on treatment of advanced carci-
noma of genitals with the preparation mbba, Neoplasma 12 (1965) 93e100;
b) O. Dvorak, Cytembena treatment of advanced gynaecological carcinomas,
Neoplasma 18 (1971) 461e464;
c) O. Dvorak, J. Bauer, Cytembena treatment of advances and relapsing uterine
cervix carcinoma, Neoplasma 18 (1971) 465e466;
d) Z. Matejovsky, Effects of Cytembena in the treatment of malignant
musculoskeletal tumours, Neoplasma 18 (1971) 473e480.
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
[5] Z. Juranic, Lj. Stevovic, B. Drakulic, T. Stanojkovic, S. Radulovic, I. Juranic,
Substituted (E)-
-(benzoyl)acrylic acids suppressed survival of neoplastic
human HeLa cells, J. Serb. Chem. Soc. 64 (1999) 505e512.
[6] S. Amslinger, The tunable functionality of -unsaturated carbonyl
b
a,b
compounds enables their differential application in biological systems,
ChemMedChem 5 (2010) 351e356.
[7] D.L. Buckley, T.W. Corson, N. Aberle, C.M. Crews, Hiv Protease-mediated
activation of sterically capped proteasome inhibitors and substrates, J. Am.
Chem. Soc. 133 (2011) 698e700.
[8] a) S.K. Rabindran, C.M. Discafani, E.C. Rosfjord, M. Baxter, M.B. Floyd, J. Golas,
W.A. Hallett, B.D. Johnson, R. Nilakantan, E. Overbeek, M.F. Reich, R. Shen,
X. Shi, H.-R. Tsou, Y.-F. Wang, A. Wissner, Antitumor Activity of HKI-272, an
orally active, irreversible inhibitor of the HER-2 tyrosine kinase, Cancer Res.
64 (2004) 3958e3965;
b) L.V. Sequist, B. Besse, T.J. Lynch, V.A. Miller, K.K. Wong, B. Gitlitz, K. Eaton,
C. Zacharchuk, A. Freyman, C. Powell, R. Ananthakrishnan, S. Quinn, J.-C. Soria,
Neratinib, an irreversible pan-ErbB receptor tyrosine kinase inhibitor: results
of a phase II trial in patients with advanced non-small-cell lung cancer, J. Clin.
Oncol. 28 (2010) 3076e3083.
5. Contributions
[9] A. Wissner, E. Overbeek, M.F. Reich, M.B. Floyd, B.D. Johnson, N. Mamuya,
E.C. Rosfjord, C. Discafani, R. Davis, X. Shi, S.K. Rabindran, B.C. Gruber, F. Ye,
W.A. Hallett, R. Nilakantan, R. Shen, Y.-F. Wang, L.M. Greenberger, H.-R. Tsou,
Synthesis and structure-activity relationships of 6,7-disubstituted 4-anilino-
quinoline-3-carbonitriles. The design of an orally active, irreversible inhibitor
of the tyrosine kinase activity of the epidermal growth factor receptor (EGFR)
and the human epidermal growth factor receptor-2 (HER-2), J. Med. Chem. 46
(2003) 49e63.
ꢀ
B.J. Drakulic design, prepares, purify, characterize compounds,
ꢀ
and done all modeling work. B.J. Drakulic write article from draft to
the final version. T.P. Stanojkovic make biological tests (IC50
determination). Z.S. Zizak make biological tests (IC50 and cell cycle
determinations) and commented results of the effect of compound
on cell cycle. M.M. Dabovic insightfully commented final version.
ꢀ
ꢁ
ꢁ
ꢁ
ꢀ
[10] D.G. Covell, A. Wallqvist, R. Huang, N. Thanki, A.A. Rabow, X-Jun Lu, Analysis of
gene expression, small-molecule screening and structural databases, Proteins
59 (2005) 403e433.
Authors declare no conflict of interest.
[11] R.H. Shoemaker, The NCI60 human tumour cell line anticancer drug screen,
Nat. Rev. Cancer 6 (2006) 813e823.
Acknowledgement
[12] a) A.-M. Katsori, D. Hadjipavlou-Litina, Chalcones in cancer: understanding
their role in terms of QSAR, Curr. Med. Chem. 16 (2009) 1062e1081;
b) J.R. Dimmock, G.A. Zello, E.O. Oloo, J.W. Quail, H.-B. Kraatz, P. Perjesi,
F. Aradi, K. Takacs-Novak, T.M. Allen, C.L. Santos, J. Balzarini, E. De Clercq,
J.P. Stables, Correlations between cytotoxicity and topography of some
2-arylidenebenzocycloalkanones determined by X-ray crystallography, J. Med.
Chem. 45 (2002) 3103e3111.
Ministry of Science and Technological Development of Serbia
support this work. Grant 172035.
Appendix. Supplementary material
ꢀ
ꢀ
ꢀ
ꢀ
[13] B.J. Drakulic, G.A. Bogdanovic, S.B. Novakovic, I.O. Juranic, Crystal structure of
(E)-4-phenyl-4-oxo-2-butenoic acid, 41st Conference of the Serbian Chemical
Society, Book of Abstracts, p. 191, Belgrade 2003.
Supplementary data associated with this article can be found in
the online version, at doi:10.1016/j.ejmech.2011.04.043.

ꢀ
B.J. Drakulic et al. / European Journal of Medicinal Chemistry 46 (2011) 3265e3273
3273
ꢀ
ꢀ
ꢀ
ꢀ
[14] S.B. Novakovic, G.A. Bogdanovic, B.J. Drakulic, A. Spasojevic-de Biré, I.O.
[25] D. Papa, E. Schwenk, F. Villani, E. Klingsberg,
b
-Aroylacrylic acids, J. Am. Chem.
ꢀ
Juranic, Electronic and topological properties of the non-covalent interactions
Soc. 70 (1948) 3356e3360.
in derivative of benzoylacrylic acid, Second Humboldt Conference on Non-
covalent Interactions, Book of abstracts, p. 45, Vrsac, Serbia, 2009. Full article
[26] Haung-Minlon, A simple modification of the Wolff-Kishner reduction, J. Am.
Chem. Soc. 68 (1946) 2487e2488.
ꢁ
in preparation.
[27] N.J. Kim, K.H. Chung, E.K. Ryu, Friedel-Crafts cyclohexylation of arenes with
1,3-dicyclohexylcarbodiimide (DCC), Tetrahedron Lett. 35 (1994) 903e904.
[28] T. Mosmann, Rapid colorimetric assay for cellular growth and survival:
application to proliferation and cytotoxicity assays, J. Immunol. Methods 65
(1983) 55e63.
[29] M. Ohno, T. Abe, Rapid colorimetric assay for the quantification of leukemia
inhibitory factor (LIF) and interleukin-6 (IL-6), J. Immunol. Methods 145
(1991) 199e203.
[15] a) N. Sugiyama, H. Kataoka, C. Kashima, K. Yamada, The photoreaction of
b
-benoylacrylic acid, Bull. Chem. Soc. Jpn. 42 (1969) 1353e1356;
ꢀ
b) B.J. Drakulic, Synthesis, Epoxidation and Photoisomerization of Ar-
aroylacrylic Acids, Graduate thesis, Faculty of Chemistry, University of Bel-
grade, February 1996, Belgrade.
[16] H.N. Patia, U. Dasa, R.K. Sharmab, J.R. Dimmock, Cytotoxic thiol alkylators,
Mini-Rev. Med. Chem. 7 (2007) 131e139.
[17] F. Fontaine, M. Pastor, F. Sanz, Incorporating molecular shape into the
alignment-free GRid-INdependent descriptors, J. Med. Chem. 47 (2004)
2805e2815.
[30] M. Carbonari, T. Tedesco, M. Fiorilli, A unified procedure for conservative
(morphology) and integral (DNA and immunophenotype) cell staining for
flow cytometry, Cytometry 44 (2001) 120e125.
[18] a) M. Pastor, G. Cruciani, I. McLay, S. Pickett, S. Clementi, GRid-INdependent
descriptors (GRIND): a novel class of alignment-independent three-dimen-
sional molecular descriptors, J. Med.Chem. 43 (2000) 3233e3243;
b) A. Durán, G. Comesaña, M. Pastor, Development and validation of AMANDA,
a new algorithm for selecting highly relevant regions in molecular interaction
fields, J. Chem. Inf. Mod. 48 (2008) 1813e1823.
[31] J. Boström, Reproducing the conformations of protein-bound ligands: a critical
evaluation of several popular conformational searching tools, J. Comput. Aid.
Mol. Des. 15 (2001) 1137e1152.
[32] T.A. Halgren, Merck molecular force field. I. Basis, Form, Scope, Parameteri-
zation, performance MMFF94, J. Comput. Chem. 17 (1996) 490e519.
[33] J.J.P. Stewart, Optimization of parameters for semiempirical methods
V: modification of NDDO approximations and application to 70 elements,
J. Mol. Mod. 13 (2007) 1173e1213.
[19] Pentacle v. 1.0.5, Molecular Discovery Ltd, http://www.moldiscovery.com
(accessed on 03.04.11).
[20] P. Gaillard, P.A. Carrupt, B. Testa, A. Boudon, Molecular lipophilicity potential, a tool
in 3D QSAR: method and applications, J. Comput. Aid. Mol. Des 8 (1994) 83e96.
[21] a) B.J. Drakulic, Z.D. Juranic, T.P. Stanojkovic, I.O. Juranic, 2-[(Carboxymethyl)
sulfanyl]-4-oxo-4-arylbutanoic acids selectively suppressed proliferation of
neoplastic human HeLa cells. A SAR/QSAR study, J. Med. Chem. 48 (2005)
5600e5603;
[34] A. Klamt, G. Schüümann, COSMO: a new approach to dielectric screening in
solvents with explicit expressions for the screening energy and its gradient,
J. Chem. Soc. Perkin Trans. 2 (1993) 799e805.
[35] A. Pedretti, L. Villa, G. Vistoli, VEGA - An open platform to develop chemo-bio-
informatics applications, using plug-in architecture and script programming,
J. Comput. Aid. Mol. Des 18 (2004) 167e173. http://www.ddl.unimi.it VegaZZ
2.4.0 (accessed on 03.04.11).
[36] a) P.J. Goodford, A computational procedure for determining energetically
favorable binding sites on biologically important macromolecules, J. Med.
Chem. 28 (1985) 849e857. http://www.moldiscovery.com GRID v.22b
Molecular Discovery Ltd. (assessed on 03.04.11);
b) G. Cruciani (Ed.), Molecular Interaction Fields. Applications in Drug
Discovery and ADME Prediction, Wiley-VCH, Zurich, Switzerland, 2006.
[37] G. Caron, A. Nurisso, G. Ermondi, How to extend the use of grid-based
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
b) M.D. Vitorovic-Todorovic, I.O. Juranic, L.M. Mandic, B.J. Drakulic, 4-Aryl-4-
oxo-N-phenyl-2-aminylbutyramides as acetyl- and butyrylcholinesterase
inhibitors. Preparation, anticholinesterase activity, docking study, and 3D
structureeactivity relationship based on molecular interaction fields, Bioorg.
Med. Chem. 18 (2010) 1181e1193;
ꢀ
c) B.J. Drakulic, unpublised results;
ꢀ
ꢀ
ꢀ
ꢀ
d) B.J. Drakulic, M.D. Vitorovic-Todorovic, I.N. Cvijetic, unpublised results.
[22] A.T. Dinkova-Kostova, M.A. Massiah, R.E. Bozak, R.J. Hicks, P. Talalay, Potency
of Michael reaction acceptors as inducers of enzymes that protect against
carcinogenesis depends on their reactivity with sulfhydryl groups, Proc. Natl.
Acad. Sci. USA 98 (2001) 3404e3409.
interaction energy maps from chemistry to biotopics, ChemMedChem
4
(2009) 29e36. www.casmedchem.unito.it (assessed on 03.04.11).
[38] Jmol v11.2.10, an open-source Java viewer for chemical structures in 3D.
http://www.jmol.org/ (assessed on 03.04.11).
[39] H. Kubinyi, Non-linear dependence of biological activity on hydrophobic
character: the bilinear model, J. Med. Chem. 20 (1977) 625e629. http://www.
kubinyi.de/bilin-program.html (assessed on 03.04.11).
[23] A. Duran, I. Zamora, M. Pastor, Suitability of GRIND-Based principal properties
for the description of molecular similarity and ligand-based virtual screening,
J. Chem. Inf. Model. 49 (2009) 2129e2138.
[24] W.J. Houlihan, J. Nadelson, for Sandoz Inc. patent, US 4083997, 1978.