2,2′-Dihydroxybenzophenones and their carbonyl N-analogues as inhibitor scaffolds for MDR-involved human glutathione transferase isoenzyme A1-1

Article abstract of DOI:10.1016/j.bmc.2014.06.007

The MDR-involved human GSTA1-1, an important isoenzyme overexpressed in several tumors leading to chemotherapeutic-resistant tumour cells, has been targeted by 2,2′-dihydroxybenzophenones and some of their carbonyl N-analogues, as its potential inhibitors. A structure-based library of the latter was built-up by a nucleophilic cleavage of suitably substituted xanthones to 2,2′-dihydroxy-benzophenones (5-9) and subsequent formation of their N-derivatives (oximes 11-13 and N-acyl hydrazones 14-16). Screening against hGSTA1-1 led to benzophenones 6 and 8, and hydrazones 14 and 16, having the highest inhibition potency (IC₅₀ values in the range 0.18 ± 0.02 to 1.77 ± 0.10 μM). Enzyme inhibition kinetics, molecular modeling and docking studies showed that they interact primarily at the CDNB-binding catalytic site of the enzyme. In addition, the results from cytotoxicity studies with human colon adenocarcinoma cells showed low LC ₅₀ values for benzophenone 6 and its N-acyl hydrazone analogue 14 (31.4 ± 0.4 μM and 87 ± 1.9 μM, respectively), in addition to the strong enzyme inhibition profile (IC_{50(6 )} = 1,77 ± 0.10 μM; IC_{50(14 )} = 0.33 ± 0.05 μM). These structures may serve as leads for the design of new potent mono- and bi-functional inhibitors and pro-drugs against human GTSs.

Full text of DOI:10.1016/j.bmc.2014.06.007

Bioorganic & Medicinal Chemistry 22 (2014) 3957–3970
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
2,20-Dihydroxybenzophenones and their carbonyl N-analogues as
inhibitor scaffolds for MDR-involved human glutathione transferase
isoenzyme A1-1
Fereniki D. Perperopoulou a, Petros G. Tsoungas b, Trias N. Thireou c, Vagelis E. Rinotas d, Eleni K. Douni c,d
,
Elias E. Eliopoulos c, Nikolaos E. Labrou a, Yannis D. Clonis a,
⇑
a Laboratory of Enzyme Technology, Department of Biotechnology, Agricultural University of Athens, 75 Iera Odos Street, GR-118 55 Athens, Greece
b Department of Biochemistry, Hellenic Pasteur Institute, Athens, Greece
c Laboratory of Genetics, Department of Biotechnology, Agricultural University of Athens, Athens, Greece
d Division of Immunology, Biomedical Sciences Research Center ‘Alexander Fleming’, Vari, Greece
a r t i c l e i n f o
a b s t r a c t
Article history:
The MDR-involved human GSTA1-1, an important isoenzyme overexpressed in several tumors leading to
chemotherapeutic-resistant tumour cells, has been targeted by 2,20-dihydroxybenzophenones and some
of their carbonyl N-analogues, as its potential inhibitors. A structure-based library of the latter was
built-up by a nucleophilic cleavage of suitably substituted xanthones to 2,20-dihydroxy-benzophenones
(5–9) and subsequent formation of their N-derivatives (oximes 11–13 and N-acyl hydrazones 14–16).
Screening against hGSTA1-1 led to benzophenones 6 and 8, and hydrazones 14 and 16, having the highest
Received 19 March 2014
Revised 20 May 2014
Accepted 4 June 2014
Available online 16 June 2014
Keywords:
Benzophenone
Enzyme inhibition
Glutathione transferase
Ketoxime
inhibition potency (IC50 values in the range 0.18 0.02 to 1.77 0.10 lM). Enzyme inhibition kinetics,
molecular modeling and docking studies showed that they interact primarily at the CDNB-binding cata-
lytic site of the enzyme. In addition, the results from cytotoxicity studies with human colon adenocarci-
noma cells showed low LC50 values for benzophenone 6 and its N-acyl hydrazone analogue 14
(31.4 0.4
(IC50(6) = 1,77 0.10
of new potent mono- and bi-functional inhibitors and pro-drugs against human GTSs.
Ó 2014 Elsevier Ltd. All rights reserved.
l
M and 87 1.9
l
M, respectively), in addition to the strong enzyme inhibition profile
Multiple drug resistance
N-Acyl hydrazone
lM; IC50(14) = 0.33 0.05
l
M). These structures may serve as leads for the design
1. Introduction
used.3 Inverse-electron-demand Diels–Alder reactions have been
described, furnishing xanthones and 2-hydroxybenzophenones in
low to moderate yields.4 A catalyst-free cascade sequence has been
reported for the synthesis of multi-functionalized 2-hydroxybenz-
Benzophenones, constitute a major class of compounds found in
the Clusiaceae (or Guttiferae) family of plants, along with xanthon-
es, coumarins and biflavonoids, having multiple biological activity.1
o-Hydroxybenzophenone derivatives, in particular, are ubiquitous
in naturally occurring and synthetic compounds.1 The presence of
the ortho-hydroxy diaryl ketone entity in many biologically active
compounds and natural products makes it a privileged structure
in medicinal chemistry and a synthesis target. Well-known and
important members are the combretastatins and phenstatins.2
Functionalized 2-hydroxybenzophenones have been obtained
from chromones, 3-formyl derivatives being most frequently
5
ophenones from electron deficient chromones.
Unsymmetrically substituted and congested hydroxybenzophe-
none derivatives have been reported, PKA inhibitor balanol6 or
G6Pase inhibitor mumbaistatin7 being two prominent examples.
It is known that the reactivity and biological activity of hydroxy-
benzophenones is linked to their acid-base and metal chelating
properties.8 It is also known that their pharmacology is usually
exerted through direct interaction with metal-bearing active
enzyme sites.9 It is reasonable to assume that the carbonyl and
the o-hydroxyl groups are major determinants of this activity.
Our recently reported interest in utilizing the reactivity profile of
xanthone10 in synthesis,11,12 as well as its inhibitory potential
towards GST,13 prompted us to investigate its ring-opened ana-
logue, substituted 2,20-o-dihydroxybenzophenones, towards GST,
taking advantage of their structure similarities, in pursuit of a
potent inhibitor against hGSTA1-1 involved in multiple drug
Abbreviations: Caco-2, human colon adenocarcinoma cell line; CDNB, 1-chloro-
2,4-dinitrobenzene; DMSO, dimethyl sulfoxide; GSH, glutathione; GST, glutathione
S-transferase; hGSTA1-1, human glutathione S-transferase isoenzyme A1-1; IPTG,
isopropyl-b-D-thiogalactopyranoside; MDR, multiple drug resistance; SM, supple-
mentary material.
⇑
Corresponding author. Tel.: +30 (210) 5294311; fax: +30 (210) 5294307.
E-mail address: clonis@aua.gr (Y.D. Clonis).
http://dx.doi.org/10.1016/j.bmc.2014.06.007
0968-0896/Ó 2014 Elsevier Ltd. All rights reserved.

3958
F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
O
resistance (MDR). To this end, GSTs (EC 2.5.1.18), a family of med-
ically important isoenzymes that catalyse the conjugation of gluta-
thione (GSH) to a variety of hydrophobic xenobiotic compounds,
have been drafted. This enzyme family renders hydrophilicity to
these xenobiotics, facilitating their metabolic processing and even-
tual secretion from the cell.14 Cytosolic GSTs are found as homodi-
O
1
mers or heterodimers.15 Each monomer has an
large -helical domain. The former domain contains the GSH bind-
ing site (G-site) on top of the large domain. Between the two
a/b domain and a
(i)
a
X
X
a
domains lies a hydrophobic pocket (H-site) in which the hydropho-
bic substrate (e.g., xenobiotic) binds and reacts with GSH. Since the
produced conjugates are susceptible to further modification and
eventual secretion from the cell, the GSTs are involved in major
detoxification mechanisms of the cell from several xenobiotics
and drugs. Based on exactly the same detoxification mechanisms,
cancer cells may acquire resistance by overexpressing GST activi-
ties,16 thus, hampering the effectiveness of certain chemothera-
peutic drugs. Therefore, several drugs and prodrugs, acting as
inhibitors against GSTs, have been proposed to overcome MDR
attributed to GST overexpression.17 GST-inhibiting strategies,
O
2
3
X = Cl
X = OMe
(ii)
O
R2
R1
2
3
1
4
8
5
7
6
O
focusing
on
ethacrynic
acid
analogues,18
individual
4
compounds17,19 and prodrug molecules20,21 have been employed.
Several GSH analogues have also been proposed as more specific
GST inhibitors,22 exploiting the high affinity of GSTs for the tripep-
tide substrate GSH. An alternative concept exploits the susceptibil-
ity of GSH conjugates (products of GST catalysis) against the GSH-
(iii)
R1/2 = Ph, Br
R1/2 = Br
R2
degrading enzyme c-glutamyltranspeptidase (cGT) and eventually
R
1 = R2
Br
OH
R1 = R2
certain peptidase-stable GSH analogues have been put to the test
/
as GST inhibitors.23,24
OH
O
R1
OH
To the best of our knowledge, this is the first report on 2,20-
dihydroxybenzophenones and their carbonyl N-analogues as
inhibitors against the MDR-involved human GSTs. Following GST
inhibition screening, in silico molecular docking and enzyme inhi-
bition kinetics, analogues exhibiting satisfactory inhibitory
potency would be regarded as ‘leads’ in designing new inhibitors
and respective prodrugs for human GSTs of medical importance.
X
O
OH
and/or
+
O
Br
OH
some decomposition
and polymerization
O
O
5-10 X = O
11-13 X = NOH
2. Results and discussion
2.1. Chemistry
14-16 X = NNHCOMe(Ar)
probable coupling modes
Scheme 1.
desired 5. Indeed, a recently reported protocol10 addressed this
issue, taking advantage of the reactivity profile of 1. Consequently,
the direct cleavage of 4, is a synthetically useful route in that it
allows access to regioselectively and diversely substituted diaryl
ketones 5. The forcing conditions can be offset by the potential of
any further desired functionalization on the regioselectively
substituted 5–9, either on the aryl rings or on the carbonyl moiety.
Eventually, this has been adopted as the method of choice and
derivatives 5–16 have been prepared for the objectives of the pres-
ent work (Table 1).
The cleavage of (di)bromo-substituted 4 merits a special com-
mentary (Scheme 1). The cleaving KOH has been used as a nucleo-
phile in the conversion of aryl halides to phenols through various
metal-catalyzed protocols.26 Furthermore, KOH/DMSO, acting as a
superbase, has been recently found27 to effect the cross-coupling
of a phenol with an aromatic halide, under mild conditions. In
our case, however, using this reagent for the ring opening of 4,
no coupling product of any kind has been isolated. Instead, the
reaction gave the expected benzophenone of type 2 in 73% yield,
along with some intractable material. It is assumed that an intra-
molecular H bonding, engaging one of the phenol OH groups of 8
or 9 and the distorted conformation of these structures hamper
any homo-(with 8 or 9) or hetero-coupling (with their precursor
4) through their bromine-bearing sites.
2.1.1. Synthesis
Benzophenones, carrying hydroxyl groups ortho-disposed to the
carbonyl group, interfere in various transformations, thus, necessi-
tating their protection and subsequent deprotection. Clearly, this is
a serious drawback in any synthetic scheme diminishing their
potential. The nucleophilic cleavage of xanthone of type
4
(Scheme 1) serves as a useful alternative route. The two benzene
rings of 1 are identical, imparting symmetry elements to the
structure. It is, thus, the substitution pattern on 1 that dictates
the corresponding one of 5. Unsubstituted or symmetrically and
asymmetrically substituted derivatives of 5, exist as a single
regio-isomer. Apparently, this is so, because the incorporated
cleaving alkali can occupy either one of the alternative ring open-
ing sites, ending up ortho-disposed to the carbonyl group.
The generation of the core structure 5 from the pyran CAO bond
cleavage of 4 has long been known.25 Indeed, the cleavage to 5 may
be accomplished, under mild conditions, using coordination com-
plexes of 1 with transition metals (Cr, Fe, Ir)25 or using strong
alkali, under either mild25 or forcing conditions.11,12 The cleavage,
under mild conditions, suffers from complications, such as genera-
tion of the coordination complex of 1 and ultimate removal of the
metal part in the former method or removal of the activating NO2
group in the latter one. On the other hand, cleavage, under forcing
conditions, despite its obvious drawback, does lead directly to the

F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
3959
Table 1
The 2,20-dihydroxybenzophenones and their N-carbonyl analogues of the present study and their hGSTA1-1 inhibition potency; ketones 5–10; ketoximes 11–13; N-acyl
hydrazones 14–16
Compound numbera
Structure
OH
Molecular formula
Molecular weight
Inhibition potency against hGSTA1-1b (%)
O
OH
5
C13H10O3
214
0
N
CO
NH
N
15
C19H16N3O3
334
31.8
OH
OH
OH
OH
O
7
C25H18O3
366
33.9
OH NOH OH
11
12
C13 H11 N O3
229
308
40.5
52.3
OH NOH OH
C13H10BrNO3
Br
OH
Br
O
OH
9
C13H8Br2O3
372
305
58.6
67.7
Br
OH NOH OH
13
C19H15NO3
CH2
O
CH2
O
O
10
C27H21BrO3
473
68.8
Br
OH
O
OH
6
C19H14O3
290
86.1
(continued on next page)

3960
F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
Table 1 (continued)
Compound numbera
Structure
Molecular formula
Molecular weight
Inhibition potency against hGSTA1-1b (%)
CO
NH
N
14
C20H16N2O3
332
87.4
OH
OH
OH
O
OH
8
C13H9BrO3
293
87.7
Br
CH3
CO
NH
N
OH
OH
Br
16
C15H13BrN2O3
349
96.1
a
In ascending order of inhibition potency against hGSTA1-1.
Mean value of three enzyme assays (25 lM analogue; error 6 3%).
b
Worth noting is the direct derivatization of 5–9 to the oximes
11–13 and the hydrazones 14–16, without the need of a protec-
tion–deprotection protocol. Intramolecular bonding, partly
masking the OH groups while increasing the electrophilicity of
the carbonyl site, may serve as a satisfactory rationale for this
outcome.
All structures take up a twisted conformation (a twist angle of
ca. 50° of the aryl rings around the carbonyl centre), in which the
two rings are not coplanar while the carbonyl or its N-derivatives
lie outside an obtuse valency angle in the range of 125–135°, indic-
ative of a strained conformation. Both OH groups are, thus, not
aligned for a notable intramolecular OHꢀ ꢀ ꢀO H-bonding interaction
with the carbonyl centre. In fact, that of the unsubstituted ring
forms a weak intramolecular H bond (Table 2) whereas the other
H
2.1.2. Structure
The structure of the parent 5 has been investigated by X-ray
crystallography,28 NMR (1H, 13C and 17O) and FT-IR spectros-
copy29,30 as well as DFT-B3LYP/6-31G⁄ calculations.30 Compound
5 exhibits a lower than C2 symmetry having its two phenol rings
twisted out of the carbonyl plane by ca. 38°.28 The deviation from
planarity is sterically triggered by the relative orientation of the
aryl rings.
one cannot be engaged in any at all.
0
Indeed, a H bonding, in the range of 1.992–1.934 ÅA, is found in 5,
8 and 9 and none in 6 and 7. The former is found to be ca. 0.2 Å
longer (weaker) than earlier calculations on 528 or its singly substi-
tuted 2-hydroxy-analogue.31 A C@O length of ca. 1.216 Å, remains
unaffected by H bonding and is virtually that of benzophenone.33
Furthermore, there seems to be no change on this bond length
X-ray and neutron-based multipolar and topological analyses
have measured electron distribution in the Resonance-Assisted
H-Bonded (RAHB) pseudo-ring conformation of 2-hydroxybenzo-
phenone analogues.31 Formal charges detected on the donor and
acceptor O and N atoms as well as on the H bond-engaged H atoms
have been rationalized by electrostatic and covalent interactions.
Substitution on the rings of 5 (one or both) will, expectedly,
cause a further distortion of the structure. Indeed, this has been
observed by geometry optimized calculations (MAGE v.6.44
programme) on the ketones 5–9, oximes 11–13 and hydrazones
14–16 (parent 5 is included as the reference structure). The novel
structures 6–16 (Table 1) have a molecular framework composed
of a C = X moiety [5–10, X = O; 11–13, X = NOH; 14–16, X = HNNC-
OAr (Ar: Ph, Py, Me)] bridging two phenols with their OH groups
ortho-disposed to the bridge. The carbonyl and the phenol hydro-
xyl groups, are set to develop intramolecular or intermolecular
upon blocking the OH groups. The bonds linking the carbonyl with
0
the aryl rings appear to have the same length of ca. 1.475 ÅA. This
bond length also reflects a virtually similar extent of conjugation
of each of the rings with the carbonyl. Larger torsion angles in
5–9 are those of the substituted ring to relieve strain.
Analogous H bonding-related geometry features are demon-
strated by oximes 11–13 and hydrazones 14–16. NOHꢀ ꢀ ꢀO and
OHꢀ ꢀ ꢀNN H bonding of magnitude similar to that of ketones is
found in 12, 15 and 16 while a much weaker in 14 and none in
11 and 13. A C@N length of 1.304 Å for 11–13 and 1.307 Å for
14–16, marginally affected by H bonding or substitution, indicates
an elongation of ca. 0.15–0.25Å. A N–O elongation of ca. 0.15 Å in
11–13 or an N–N one of 0.30 Å in 14–16, is also observed. A
N–HN. . .H bonding of 2.487 Å in 15, engaging the pyridine N,
‘locks’ the pyridine orientation, thus, the conformation of the
whole structure. In addition, the OAH length in both phenol rings
remains unchanged throughout the series. The bonds linking the
imine centre with the aryl rings appear to be of ca. 1.479–
1.480 Å, slightly longer than their precursors 5–8. Torsion angles
appear to be affected by substitution but rather more significantly
by the N–O repulsion of the oxime or hydrazone N lone pair and
phenol OH groups.
H-bonding or other non-covalent bonding (e.g.,
p–p stacking,
hydrophobic or halogen) interactions with the surroundings. A
substantial torsion has been found in 5 (angles of ca. 35–55°) to
minimize repulsion among the aromatic rings leading to a non
planar conformation.32 In our case, this is reflected on the intramo-
lecular H bonding with only one of the OH groups.

F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
3961
Table 2
Bond lengths (Å) and dihedral angles
x (degrees) of 5–16
a
Compound number
OAH
(C)Oꢀ ꢀ ꢀH
dO-O
x
NAO-H
O-Hꢀ ꢀ ꢀN
NAOꢀ ꢀ ꢀHAO
C@O
C@N
NHꢀ ꢀ ꢀN
5b,c
0.988d
1.711d
1.992
1.768e
>3
2.587d
2.742
34.2/52.6f
1.216
0.966/7
0.984e
6
7
8
9
0.966/7
0.966/7
0.966/7
0.966/7
>3
>3
41.6/42.7
38.2/33.8
36.1/53.2
32.7/53.2
1.217
1.215
1.216
1.216
1.216
>3
1.934
1.976
>3
2.701
2.718
2.642
dN-O
>3
10
11
12
13
14
15
16
0.966/7
0.966/7
0.966/7
0.966/7
0.966/7
0.966/7
43.3/43.5
55.5/51.3
41.1/44.4
43.6/52.8
47.6/55.0
37.6/54.7
0.969
0.968
0.968
2.960
1.933
>3
2.067
1.936
1.944
1.944
1.304
1.304
1.304
1.307
1.307
1.307
2.689
>3
>3
2.698
2.716
2.487
a
b
c
d
e
f
Shortest distance is recorded.
B3LYP/6-31G⁄ calculations.
HF/6-31G⁄ calculations.30
Dávalos, J. Z.; Guerrero, A; Herrero, R.; Jimenez, P.; Chana, A.; Abboud, J.L.M.; Lima, C. F.R.A.C.; Santos, L.M.N.B.F.; Lago, A.F. J. Org. Chem. 2010, 75, 2564.
2-Hydroxybenzophenone [Krygowski, T.M.; Zachara-Horeglad, J.E.; Paluciak, M. J. Org. Chem. 2010, 75, 4944].
x
values refer to unsubstituted (left) or substituted (right) aryl rings, respectively.
Consistent with the nature and strength of the H bonding are
screening and IC50 calculations. Having initially determined the
Michael is constant, Km, for the CDNB-hGSTA1-1 couple as
0.32 0.08 mM, an experimental [CDNB] <<Km would have given
low values for [CDNB]/Km. This assay condition would increase
their ir, 1H and 13C NMR spectra. Ir absorptions in the range of
1620–1615 cmꢁ1 for the C@O group and 3450–3200 cmꢁ1 for the
OH groups, were observed. Further, lowfield distinct 1H signals at
d = 10.64, 10.50 and 10.48 ppm for the OH groups and 13C signals
at d = 201–199 ppm for the C@O group appeared. The facile direct
derivatization of ketones 5–9 to 11–16 is of interest, as it lends
further support to the H bonding mode (see earlier section). The
experimentally and computationally derived values are in good
agreement with earlier results on 528,30 or its singly substituted
2-hydroxy-analogue.31 They also fall within the ranges commonly
reported in functional group databases.
the apparent inhibition caused by a fixed concentration (25 lM
in our hands) of a competitive inhibitor; hence, a low [CDNB], rel-
ative to Km, favours the identification of competitive inhibitors.35
In contrast, if we were to run the enzyme assays at relatively high
substrate concentrations (i.e., high [CDNB]/Km values), we would
bias the screening assay against competitive inhibitors, in favour
of uncompetitive ones, whereas the inhibition potency of non-
competitive inhibitors would not be affected by the ratio out-
come.35 We have chosen to perform the enzyme inhibition assays
under the so-called ‘balanced assay conditions’ (BAC) as the opti-
mum choice, ensuring an experimental [CDNB] = 0.3 mM ꢂ Km.
In order to reveal their enzyme inhibitory potency all com-
pounds were subjected to experimental screening against
hGSTA1-1 activity. From the data obtained (Table 1), one distin-
guishes three groups of inhibitory potency: (a) low inhibition (up
to ca. 34%; compounds 5, 15 and 7), (b) medium inhibition (ca.
34–69%; compounds 11, 12, 9, 13 and 10) and (c) high inhibition
(above 86%; compounds 6, 14, 8 and 16). BSP was also put to the
test as a control inhibitor, under the same conditions and it was
found to inhibit hGSTA1-1 by approximately 63%.19 The encourag-
ing behaviour of the ‘high inhibition’ compounds (Table 1; 6, 14, 8
and 16) was confirmed experimentally by the low IC50 values
obtained from concentration–response curves (Table 3; as an
example, see Fig. 1) and predicted by in silico studies. Molecular
modeling and docking analysis of 6, 14, 8 and 16 with hGSTA1-1,
in particular, provided an insight into salient structural features
unveiled upon interaction. By inspecting the location of the most
favourable conformations (i.e., low energy ones) of these com-
pounds, docked in the hGSTA1-1 binding site, the following obser-
vations are evident. A clustering occurs on two locations in the
binding area (Fig. 2), one in the proximity of the a-helix 155–169
(internal secondary pocket) and one in the proximity of the a-helix
210–220, where CDNB also binds (external catalytic pocket).
Upon generation of the enzyme complex, the geometry of the
structures adjusts to achieving the best fit. It is, thus, expected that
the weakly RAHB-stabilized conformers (‘closed’ or ‘pseudo ring’
form), that is, 5, 8, 9, 12 and 14–16, hiding polarity from the
surroundings, render the structures weakly lipophilic. It follows
that their acidic (OH and NH) groups take precedence over
Based on the presented data, the following features emerge: (a)
bromo substitution (whether mono-8 or di-9) has a negligible
effect on the conformation (5, 8 and 9 assume virtually the same
conformation), (b) a marked distortion by a substantial twist is
observed in the corresponding phenyl-substituted derivatives 6
and 7, particularly so in the latter, (c) the phenyl mono-substituted
derivatives 6 and 13 also assume a similar conformation, (d) sub-
stitution pattern and H bonding have an effect on the flexibility
and shape of the structures, (e) lone pair N–O repulsion in 11–13
or in 14–16, only to a lesser extent, is probably a determinant for
the observed conformations though a weak one (dN–O 2.7–3.4 Å).
2.2. Screening of the compounds and selection of ‘lead
structures’ as hGSTA1-1 inhibitors
Before embarking into the inhibition studies we performed con-
trol experiments with our enzyme preparation using bromosulf-
ophthalein (BSP) as a known hGSTA1-1 inhibitor.34 In silico
molecular modeling and docking analysis predicted that BSP binds
to a non-catalytic site, allowing simultaneous binding of the
substrate CDNB to the catalytic primary site (SM-1). This is in
agreement with earlier observations34 and has been confirmed
by kinetic studies with our enzyme preparation, using BSP as an
inhibitor and CDNB as a variable substrate, demonstrating a
non-competitive modality of inhibition.13,19
In designing the enzyme assay protocol for screening the com-
pounds as potential hGSTA1-1 inhibitors, the concentration of
25 lM, falling within the 1–30 lM range, suggested in bibliogra-
phy as an appropriate one for inhibitor screening,35 has been
chosen. A more crucial factor to be decided has been the substrate
concentration, [CDNB], in the enzyme inhibition assay for the

3962
F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
Table 3
Behaviour of compounds selected from screening experiments (Table 1) against hGSTA1-1 activity (IC50) and Caco2 cells (LC50
)
Compound number and structure
Modality of inhibitiona
IC50 against hGSTA1-1 (
lM)
LC50 against Caco2 cells (lM)
5
OH
OH
O
O
OH
OH
—
—
>400
6
8
Competitive, linear
1.77 0.10
31.4 0.4
OH
O
OH
Br
Mixed, linear
0.24 0.04
120 1.9
11
OH NOH OH
—
—
315 1.4
14
CO
NH
N
Competitive, linear
0.33 0.05
87 1.9
OH
OH
16
CH3
CO
NH
N
OH
OH
Br
Mixed, hyperbolic
0.18 0.02
>400
a
Compounds 6, 8, 14 and 16 showed mixed inhibition modality with the co-substrate GSH.
intermolecular interactions in the enzyme environment through
an accentuated hydrophilicity. Ring-substituent-triggered interac-
tions should eventually rest upon and facilitate charge transfer
pathways.36 To that end, singly substituted ketones 6 and 8 or
hydrazones 14 and 16 are highly efficient binders. Of these, the for-
mer, being unsymmetrically substituted, have an unequal charge
distribution and greater flexibility, in contrast to their symmetri-
cally disubstituted and sterically congested relatively poor binders
7 and 9. Compounds 14 and 16 incorporate an acidic NH site and a
hydrophobic N-acyl terminal group (Ph or Me). These features
allow interactions, through a charge relay, within the protein cav-
ity, determined by the polar side of helix 155–169, dispersed with
charged (Glu162, Glu169) and polar (Tyr166) residues. The lower
inhibitory performance of 15 may be attributed to its ‘locked’ con-
formation (see earlier), reducing its flexibility and eventually bind-
ing efficiency. Of the oximes, regardless of their E and/or Z
conformation, 13 shows the highest inhibitory potency, followed
by 12 and 11, a trend similar to that of 5, 8 and 9. 10, the
benzylated derivative of 5, has been found to have a variable inhib-
itory potency. Masking both hydroxyl groups removes the intra-
molecular
thus, facilitate mainly
H
bonding. Any potential hydrophilic interactions,
(arene–arene) or intermolecular H
p–p
bonding interactions, engaging the C@O group with the enzyme
surroundings. The variability of its inhibitory potency has been
tentatively attributed to its facile debenzylation in the enzyme cav-
ity. From the clustering of the inhibitors on docking (Fig. 2), steric
congestion (e.g., 7), apparently enhanced by symmetrical substitu-
tion, as a parameter impeding proper fit, may serve as a selection
rule for lead candidates.
2.3. Studying the modality of interaction between the selected
inhibitor lead structures and hGSTA1-1
On the basis of the ‘cherry picking’ (screening) experiments and
the low IC50 values observed, enzyme inhibition kinetics on com-
pounds 6, 8, 14 and 16 were performed in order to clarify their

F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
3963
(a)
(b)
100
80
60
40
20
0
100
80
60
40
20
0
0,001
0,01
0,1
1
0,1
1
10
[Inhibitor 16] (μΜ)
[Inhibitor 6] (μΜ)
Figure 1. Concentration–response graphs for the determination of the IC50 values for 6 (a) and 16 (b) against hGSTA1-1. The ‘concentration’ values (
lM) are presented on
logarithmic scale, whereas the ‘response’ values (as% ratios of inhibited over uninhibited rates) are presented on the ‘Remaining activity’ axis. The graphs were produced
using the GraFit3 v.3 computer program.
(a)
(b)
(c)
(d)
Figure 2. Clustering of the four selected inhibitors, 6 (a), 8 (b), 14 (c) and 16 (d) at the most probable binding positions on hGSTA1-1 as predicted by in silico molecular
docking. It is evident that clustering occurs on two locations in the binding site, one in the proximity of the a-helix 155–169 (internal secondary pocket; upper) shown in
purple and one in the proximity of the a-helix 210–220 where CDNB also binds (external catalytic pocket; down) shown in green. All ligands are depicted in sticks
representation. The position where the substrate CDNB would bind in the absence of inhibitor is shown as space filling dot model. The co-substrate GSH is depicted in
magenta, the S atom in yellow, N atoms in blue and O atoms in red. The figure is created using the PYMOL v1.5 program.

3964
F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
(a)
(b)
1200
1000
800
600
400
200
0
[Ιnhibitor 6] = 1.7 μΜ
60
40
20
0
[Ιnhibitor 6] = 0.5 μΜ
[Inhibitor 6] [= 0 μΜ
5
10
1/[CDNB] (1/μΜ)
15
20
-2 -1,5 -1 -0,5
0
0,5
1
1,5
2
[Inhibitor 6] (μΜ)
(c)
(d)
50
40
30
20
10
0
600
400
200
0
[IInhibitor 14] = 0 μΜ
[Inhibitor 14] = 0,37 μΜ
[Inhibitor 14] = 0,74 μΜ
0
10
20
30
-0,4 -0,2
0
0,2 0,4 0,6 0,8
1
1/[CDNB] (1/μΜ)
[Inhibitor 14] (μΜ)
Figure 3. Purely competitive inhibition kinetics of hGSTA1-1 with inhibitors 6 (a and b) and 14 (c,d) using CDNB as a variable substrate. Lineweaver–Burk graphs of initial
velocities of hGSTA1-1 versus [CDNB] (37.5–980 M) at different concentrations of inhibitor 6 (a) and inhibitor 14 (c). Secondary graphs for 6 (b) and 14 (d) derived from data
l
of respective primary graphs (a) for 6 and (c) for 14. The inhibition constants Ki(6) for 6 and Ki(14) for 14 are the intercepts on the basis axes of graphs (b) and (d), respectively.
Points are average of three enzyme assays. The graphs are created using the GraFit v.3 program.
(a)
(b)
Figure 4. Low energy conformations of substrates CDNB, GSH and inhibitors 6 (a) and 14 (b) at the most probable binding sites of hGSTA1-1 as predicted by in silico
molecular docking. All ligands are shown as balls-and-sticks, except for CDNB which is shown as space filling dot model. Both inhibitors (green ligands) partly occupy the
catalytic site and clash with CDNB when bound at the same site. GSH is depicted in magenta, the S atom in yellow, N atoms in blue and O atoms in red. The figure is created
using the PyMOL v1.4 program.
binding modality towards the target hGSTA1-1, a fundamental
knowledge useful in inhibitor design. In all four cases, two sets of
experiments were implemented, each employing either CDNB
and Figure 3d for 14).37,38 This behaviour suggests that these two
inhibitors compete with CDNB for the same binding site of the
enzyme; calculated inhibition constants
(from Fig. 3b) and Ki(14) = 0.38 0.05 M (from Fig. 3d). The
described kinetic model is in concert with the in silico molecular
docking analysis. The latter predicts that both inhibitors,
Ki(6) = 1.47 0.15 lM
(37.5–0 980
l
M) or GSH (100–2500
lM) as a variable substrate,
l
in the presence of different steady inhibitor concentrations.
6
2.3.1. Study of inhibitors 6 and 14
(Fig. 4a) and 14 (Fig. 4b), despite featuring different core struc-
tures, in their low energy most favoured position, they clash with
CDNB, if trying to accommodate them at the catalytic site of
hGSTA1-1 where CDNB binds. In this case, it appears that the bind-
ing modality (competitive or mixed) is not determined primarily
by the inhibitor core structure (i.e., benzophenone, as in 6 and 8
When using CDNB as a variable substrate, 6 and 14 displayed
purely competitive inhibition kinetics on the basis of the linearity
observed for both the double reciprocal Lineweaver–Burk graphs
(Fig. 3a for 6 and Figure 3c for 14), at various steady concentrations
of 6 and 14 and their respective secondary derivatives (Fig. 3b for 6

F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
3965
or its N-carbonyl hydrazone, as in 14 and 16), but rather by the
presence of bulky substituents (i.e., an aromatic group present on
6 and 14, absent in 8 and 16). This extra added volume forces 6
and 14 to adopt new orientations upon binding to hGTA1-1
(Fig. 4), eventually leaving not enough space for a simultaneous
binding of CDNB at the same binding site.
With GSH as a variable substrate, both 6 and 14 showed mixed
inhibition kinetics, manifested by the lines of the double reciprocal
Lineweaver–Burk graphs of hGSTA1-1 versus [GSH] initial veloci-
ties, at various steady concentrations of 6 and 14, not intersecting
the reciprocal velocity or [GSH] axes (SM-2a).37,38 Furthermore, the
linear correlation of the respective secondary graphs, depicting
slope versus [inhibitor] (SM-2b), is supportive of a purely mixed
type of inhibition.37,38 The equilibrium model for this type bears
the assumption that the inhibitor binds to both the free enzyme
and its enzyme–GSH complex, with no possibility for product for-
mation37,38 (the respective complexes are unreactive, ‘dead-end’).
The model suggests that inhibitors 6 and 14 may interact at a site
other than the GSH-binding site of hGSTA1-1, that being partly the
catalytic CDNB-binding site, as described earlier (Fig. 3).
enzyme–CDNB–8 complexes are the unreactive (‘dead-end’)
ones.37,38 However, at [8] >0.5
M (e.g., 0.8 M), the secondary plot
l
l
curves upwards (SM-4), suggesting the binding of a second mole-
cule of 8, thus, intensifying its inhibitory effect.38 This view is sup-
ported by earlier works13,39,19,34 on the existence of multiple
binding sites with GSTs for a single compound, often with varying
affinity and inhibitory potency. Kinetic studies and isothermal
titration calorimetry with the non-competitive inhibitor BSP and
hGSTA1-1 pointed to two binding site types for BSP per enzyme
subunit.34,40 Furthermore, since BSP and CDNB bind the enzyme
at different sites (SM-1), it has been proposed that the inhibition
by BSP could be attributed to conformational/structural changes
of the enzyme,40 a modality of inhibition similar to that observed
with 8.
The parabolic mixed inhibition modality with a high [8],
described above, assumes no catalytic activity for the enzyme–
inhibitor and enzyme–substrate–inhibitor complexes,37,38 and this
is confirmed by molecular modeling and docking. The in silico
models predict close proximity and interaction between the sub-
strate CDNB and 8 (two H-bonds, 2.56 and 2.76 Å) when both bind
the catalytic area of the enzyme (Fig. 5a and b). In this case, it is
reasonable to anticipate interference of 8 with the enzyme’s cata-
lytic function involving CDNB. However, at higher [8], a second
molecule of 8 is predicted to bind at the distant internal secondary
site (SM-5a), intensifying the inhibition effect.
2.3.2. Study of inhibitors 8 and 16
With CDNB as a variable substrate, 8 has shown mixed inhibi-
tion kinetics (cf 6),37,38 as manifested by the lines of the double
reciprocal Lineweaver–Burk graph not intersecting the reciprocal
velocity or [CDNB] axes (SM-3a) and the fair linearity observed
for the respective secondary graph (SM-3b); Ki(8) = 0.36 0.11 lM.
This inhibition modality rests upon an equilibrium model, which
Turning to inhibitor 16, with the CDNB as a variable substrate,
one observes mixed inhibition kinetics, again, as shown by the
lines of the Lineweaver–Burk graph intersecting left of the
reciprocal velocity axis (Fig. 6a). However, the points of the
foresees no product formation, since the predicted enzyme–8 and
(a)
(b)
(c)
(d)
Figure 5. Low energy conformations of substrates CDNB, GSH and inhibitor 8 (a), (b) and inhibitor 16 (c), (d) at the most probable binding sites of hGSTA1-1 as predicted by in
silico molecular docking in the absence and presence of CDNB. All ligands are shown as balls-and-sticks, except for CDNB which is shown as space filling dot model. (a) In the
absence of CDNB, inhibitor 8 (green ligand) is bound close to CDNB-binding region. (b) In the presence of CDNB, inhibitor 8 (yellow ligand) is bound close to CDNB, developing
H-bonds (2.56 and 2.76 Å). (c) In the absence of CDNB, inhibitor 16 (green ligand) is bound close to CDNB-binding region. (d) In the presence of CDNB, inhibitor 16 (green
ligand) is bound far enough from CDNB (space filling dot model) permitting catalytic function, though at a lower rate. The co-substrate GSH is depicted in magenta, the S atom
in yellow, N atoms in blue and O atoms in red. The figure is created using the PyMOL v1.4 program.

3966
F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
50
(a)
(b)
800
600
400
200
0
40
30
20
10
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0
10
20
30
[Inhibitor 16] (μΜ)
1/[CDNB] (1/μΜ)
Figure 6. Mixed inhibition kinetics of hGSTA1-1 with inhibitor 16 using CDNB as a variable substrate. (a) Lineweaver–Burk graph of initial velocities of hGSTA1-1 versus
[CDNB] (37.5–980
lM) at different concentrations of inhibitor 16 (s 0, d 0.05, h 0.20 and j 0.60
lM). (b) Secondary graph derived from data of graph (a). Points are average
of three enzyme assays. The graphs are created using the GraFit v.3 program.
(a)
(b)
150
LC50 > 400 μM
150
100
50
100
50
0
LC50 = 31.36 0.40 μM
0
0
0
0
200
400
400
400
50 100
300
0
10
20
30
40
[Compound 5] (µM)
[Compound 6] (µM)
(c)
(d)
LC50 = 120 1.87 μΜ
100
50
0
100
50
0
LC50 = 315 1.41 μΜ
100
200
300
50
0
100
200
300
400
[Compound 8] (µM)
[Compound 11] (µM)
(e)
(f)
150
100
50
LC50 > 400 μΜ
100
50
0
LC50 = 87 1.91 μM
0
100
200
300
0
100
200
300
400
[Compound 14] (µM)
[Compound 16] (µM)
Figure 7. The effect of 5 (a), 6 (b), 8 (c), 11 (d), 14 (e) and 16 (f) on the viability of human colon adenocarcinoma (Caco2) cells after 24 h treatment. Cytotoxicity was assessed
using a microplate MTT colorimetric assay. Survival (cell viability) was expressed as a percentage of the negative control without treatment with compounds. LC50 values are
given as mean + SEM from three independent experiments performed in triplicate. The graphs were produced using the GraphPad PRISM v.5 computer program.

F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
3967
secondary graph, derived from data of Figure 6a, curve downwards
3. Experimental
to a limiting rate (Fig. 6b), suggesting a hyperbolic mixed inhibition
modality.38 These findings predict37,38 that 16 binds to both the
free enzyme and the enzyme–CDNB complex, leading to formation
of at least two complexes, enzyme–16 and enzyme–CDNB–16,
respectively. A GSH molecule should be present onto both com-
plexes (not shown) due to using an enzyme-saturating GSH con-
centration in the respective assays. In contrast to the modality
described for 8, the present model predicts a breakdown of the
enzyme–GSH–CDNB–16 complex to products, at a rate slower than
that without inhibitor.37,38 On the basis of these experimental find-
ings, it is reasonable to assume that, in the presence of CDNB, 16
binds to a site where there can be no direct and detrimental to
the catalytic function interaction between them, thus, allowing a
reduced catalytic function. This is in concert with in silico molecu-
lar docking, showing CDNB at the catalytic primary site (Fig. 5d,
left) and 16 at a distant secondary site (Fig. 5d, right) of hGSTA1-
1, as the two most probable binding ones, respectively, producing
a reactive quadruple complex, enzyme–GSH–CDNB–16. Apparently,
these locations are not close enough (SM-5b) for 16 to abolish the
enzyme’s catalytic function on CDNB, as observed with 8.
In summary, to accommodate 8 or 16 in hGSTA1-1, simulta-
neously with CDNB while 8 should first be fixed at the most prob-
able position in the catalytic (primary) site, followed by a second
molecule, taking up a position in the distant (internal) secondary
site,19 located on the a-helix 86–109 side (SM-5a), the larger 16
binds only as a single molecule at the distant (internal) secondary
site (SM-5b). This is, indeed, demonstrated by a very narrow clus-
tering of probable positions for the inhibitors, indicating the vol-
ume limitations and shape restrains for the available protein cavity.
Because of the non-linearity observed with the secondary graph
(Fig. 6b), the Ki(16) was calculated from linear double reciprocal
3.1. Materials and instrumentation
Reagents were used as commercially purchased, while solvents
were purified and dried according to standard procedures. Melting
points were measured on an Electrothermal IA9000 Series appara-
tus and are uncorrected. Infrared spectra were recorded on a JASCO
FT/IR-5300 spectrometer as KBr discs. Elemental analyses were
performed on a Carlo Erba 1106 analyser. NMR spectra were mea-
sured on a Bruker Avance 400 MHz and a Varian 600 MHz spec-
trometers, in CDCl3 or DMSO-d6 solutions. Mass spectra were
recorded by Micromass—Platform LC or JEOL JMS-AX505 W low
or high resolution instruments. Analytical TLC was run on Fluka
Silica Gel F254. Preparative Flash Chromatography was run on Car-
lo Erba Reactifis-SDS SILICE 60 A C.C 40–63 lm Chromagel.
3.2. Synthesis of substituted 2,20-bis-hydroxybenzophenones
Details on the synthesis of the title hydroxybenzophenone
derivatives used in the present work have been described earlier
by Tsoungas et al.11,12 The methodology followed is summarized
in Scheme 1 whereas the derivatives tested are laid out in Table 1.
Briefly, the established10 reactivity profile of xanthone core struc-
ture 1 has been suitably exploited to effect its regioselective substi-
tution. Nucleophilically triggered ring-opening of 4 by alkali, then,
generated the corresponding, also regioselectively substituted, tar-
get benzophenones 5–10. If an alkoxide is used, as the cleaving
nucleophile, one of the phenolic OH groups in 5 is protected and
masked as an alkyl ether. This approach provides a means to differ-
entiate between two otherwise identical aromatic rings and OH
groups in 5 and, thus, allow the synthesis of a diverse array of use-
ful derivatives through further transformations. The general
method used for ring opening of 4 and spectral data of most active
of the tested ketones and previously unreported 2,20-bis-hydroxy-
benzophenones 6 and 8 are described herein (FT-IR, 1H NMR and
Mass Spectra are given as Supplementary material).
graphs, depicting 1/
versus 1/[16] (SM-6b), constructed from data of Figure 6;37,38
i(16) = 1.75 0.25 M. Using GSH as a variable substrate, both 8
DSlope versus 1/[16] (SM-6a) and 1/DIntercept
K
l
and 16 showed, predictably, mixed inhibition kinetics, since the
lines of the Lineweaver–Burk graph intersected the left of the reci-
procal velocity axis (SM-7a for 8 and SM-7b for 16).
3.2.1. Synthesis of benzophenones 5–10 (general method)
To a solution of xanthone (1 mol equiv) in DMSO, an aqueous
solution of 12 N KOH (1.4 mol equiv) is added and the reaction mix-
ture is refluxed in a preheated bath for 12 h. The reaction mixture is
then concentrated in vacuo and the residue is treated with ice-
water, slowly acidified with concentrated HCl to pH 3 and exhaus-
tively extracted with dichloromethane. The combined extracts are
repeatedly washed with water and brine, dried over sodium sulfate,
concentrated and the residue is either directly chromatographed
(silica, petroleum ether/dichloromethane 6:1) or triturated with
an ether/petroleum ether mixture prior to chromatography.
2.4. Studying the cytotoxic activity of the selected inhibitor lead
structures with human colon adenocarcinoma cell line
In the course of lead structure studies, it is useful to evaluate
compounds not only on the basis of target enzyme activity, but
also on cell-based assays. For the latter application, the human
colon adenocarcinoma cell line (Caco2) is a good choice, particu-
larly for this study, because it expresses predominantly the
hGSTA1-1 isoenzyme of interest.13,41,42 Therefore, the four selected
compounds, 6, 8, 14 and 16, along with two control structures,
benzophenone 5 and ketoxime 11, were evaluated for their cyto-
toxicity against Caco2 cells. The results obtained on cell viability
(Table 3) indicated that 5 and 11, the former with respect to 6
and 8 and the latter with respect to 14 and 16, showed low cyto-
0
3.2.1.1. 2-Hydroxy-4-phenyl-2 -hydroxybenzophenone (6). Yield:
72%, Rf = 0.64. IR
mmax: 3422 (OH), 1615 (C@O), 1594, 1509,
1479 cmꢁ1 1H NMR (400 MHz, CDCl3): d (ppm) 7.01–6.92 (3H, m,
.
toxicity (LC50 >400
Fig. 7d), whereas 6, 8 and 14 showed significant increase of cyto-
toxicity (respective LC50 values 31.4 0.4 (Fig. 7b),
120.0 1.9 M (Fig. 7c) and 87.0 1.9 M (Fig. 7e)). Interestingly,
of the four inhibitors tested, only 16 displayed very low cytotoxic
effect (LC50 >400 M; Fig. 7f), even lower than control structure
11 (LC50 315.0 1.4 M; Fig. 7d). Therefore, taking into consider-
lM for 5; Figure 7a) and 315 1.4 lM for 11;
Ar-H), 7.14–7.11(1H, d, Ar-H J = 8.4 Hz), 7.20–7.17 (1H, d, Ar-H
J = 8.4 Hz), 7.38–7.34 (1H, d, Ar-H J = 7.2 Hz), 7.45–7.41 (1H, d, Ar-
H, J = 7.6 Hz), 7.57–7.48 (2H, m, Ar-H), 7.70–7.66 (1H, dd, Ar-H,
J = 8.0 Hz, J = 1.6 Hz), 7.87–7.74 (1H, dd, Ar-H, J = 8.8 Hz, J = 2.4 Hz),
7.83 (1H, d, Ar-H, J = 2.4 Hz), 10.50 (1H, s, ArOH), 10.64 (1H, s, ArOH).
13C NMR (75.4 MHz, CDCl3): d (ppm) 199.2, 162.5, 161.5, 141.2,
134.6, 134.2, 134.1, 133.2, 131.8, 129.5, 128.9, 128.2, 127.6, 127.5,
123.5, 120.5, 120.1, 119.5, 118.1. HRMS-ES [MꢁH+] m/z: found
289.08630, calcd for C19H14O3 290.1790.
l
M
l
l
l
l
ation both the cytotoxicity and inhibition profiles (Table 3), one
would regard benzophenone 6 and its N-carbonyl hydrazone ana-
logue 14, as an overall better balanced choice for lead structures,
since they exhibit satisfactory cytotoxicity (Table 3;
0
3.2.1.2. 2-Hydroxy-4-bromo-2 -hydroxybenzophenone (8). Yield:
LC50(6) = 31.4 0.4
tory potency (IC50(6) = 1.77 0.10
l
M; LC50(14) = 87 1.9
l
M) and enzyme inhibi-
M).
81%, m.p.128 °C, Rf = 0.59. IR vmax: 3450–3200 (OH), 1621 (C@O), 1609,
lM; IC50(14) = 0.33 0.05
l

3968
F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
1
1585, 1479 cmꢁ1. H NMR (CDCl3): d (ppm) 7.1–6.9 (2H, m, Ar-H), 7.08
(1H, d, Ar-H), 7.6–7.5 (3H, m, Ar-H), 7.69 (1H, d, Ar-H J = 2.4 Hz), 10.49
(1H, s, Ar-H), 10.41 (1H, s, Ar-H). 13C NMR (75.4 MHz, CDCl3): d (ppm)
201.1, 162.3, 161.8, 136.8, 135.7, 133.9, 133.4, 123.8, 121.6, 120.8, 118.9,
117.5, 115.5. HRMS-ES [M+] m/z: found 293.030, calcd for C13H9BrO3
293.0220.
assays of inhibition experiments with the test compounds (see
below). The mixture was incubated at 25 °C for 5 min, prior to add-
ing the enzyme sample. Initial velocities were determined in trip-
licate and were corrected for spontaneous reaction rates, when
necessary. One unit of enzyme activity is defined as the amount
of enzyme that produces 1.0
the assay conditions.
lmol of product per minute under
3.2.2. Synthesis of oximes 11–13 and hydrazones 14–16
These were prepared by literature methods.11,12 Spectral data of
the most active of the oximes 11 and hydrazones 14 are described
herein (FT-IR, 1H NMR and Mass Spectra are given as Supplemen-
tary material, SM-8).
3.4.2. Screening the compounds as hGSTA1-1 inhibitors
The screening assay for the test compounds of Table 1 as possi-
ble GST inhibitors was implemented by adding the ingredients in
the following order (1 mL assay volume): potassium phosphate
buffer (100 mM, pH 6.5), 0.75
25 nmol test compound (5 L from a 5 mM solution in DMSO)
and enzyme (up to 20 L of purified GST, typically producing
0.15 A340 per min). The mixture was incubated at 25 °C for
1 min, prior to adding 0.3 mol CDNB (prepared in ethanol). The
observed rate was used to calculate the remaining activity (%), tak-
ing as 100% initial activity value the rate observed after replacing
the test compound by an equal volume of DMSO (5 lL).
lmol GSH (prepared in water),
3.2.2.1. 2,20-Bis-hydroxybenzophenone oxime (11).
76%, mp: 113 °C, Rf = 0.58. IR max: 3539 (NOH), 3362 (OH), 1620
(C@N) cmꢁ1 1H NMR (600 MHz, DMSO-d6): d (ppm) 7.03–6.71
(4H, m, Ar), 7.28–7.19 (4H, m, Ar), 9.67 (1H, br, OH), 11.60 (2H,
br s, NOH and OH). 13C NMR (75.4 MHz, DMSO-d6): d (ppm)
158.8, 158.7, 157.4, 130.1, 130.3, 129.7, 129.6, 128.4, 128.2,
128.1, 119.4, 118.6, 116.4. HRMS-ES [M+] m/z: found 228.0666
calcd for C13H11NO3 228.0666.
Yield:
l
m
l
.
D
l
3.5. Inhibition studies with purified hGSTA1-1
3.2.2.2. 2,20-Bis-hydroxybenzophenone N-benzoylhydrazone
(14).
Yield: 72%, mp: 236–7 °C, Rf = 0.43. IR
m
max: 3317 (NH),
The GraFit3 version 3 computer program (Erithacus Software,
Ltd., U.K.) was used for producing kinetic graphs, determining
apparent kinetic parameters/constants and IC50 values.
3280 (OH), 1649 (C@O) cmꢁ1
.
1H NMR (400 MHz, DMSO-d6): d
(ppm) 6.81–6.72 (1H, m, ArH), 6.84 (1H, dd, ArH, J = 7.6 Hz),
7.05–6.95 (3H, m, ArH), 7.10 (1H, d, ArH, J = 8 Hz), 7.2 (1H, dd,
ArH, J = 7.6 Hz, J = 1.6 Hz), 7.35–7.25 (1H, m, ArH), 7.68–7.38 (5H,
m, Ar), 10.50 (1H, s, ArOH), 10.18 (1H, s, ArOH), 13.00 (1H, s,
NH). 13C NMR (75.4 MHz, DMSO-d6): d (ppm) 163.4, 161.2, 161.1,
155.7, 132.9, 132.7, 132.6, 132.5, 130.7, 130.4, 128.9, 128.8,
127.8, 127.6, 121.5, 121.3, 118.5, 118.4, 117.9, 117.8. HRMS-ES
[M+] m/z: found 332.1089 calcd for C20H16N2O3 332.1088.
3.5.1. Determination of IC50 values for inhibitors 6, 8, 14 & 16
Initial velocities for the GST-catalysed reaction with CDNB and
GSH as substrates were measured at 25 °C, in the presence of var-
ious concentrations of inhibitors 6, 8, 14 & 16. The assay employed
was the same as that for the screening of the test compounds as
GST inhibitors (see previous paragraph). Different inhibitor quanti-
ties were introduced in the assay mixture in 5 lL DMSO. The
3.3. Expression and purification of hGSTA1-1
observed rate was used to calculate the remaining activity (%), tak-
ing as 100% initial activity value the observed rate (approx. 0.15
This is based on a published method.19 Briefly, the expression of
GST was induced from Escherichia coli BL21 (DE3) cells harbouring
the plasmid pET101/D by addition of IPTG. The cells were har-
vested by centrifugation (845 mg cell paste), resuspended in phos-
phate buffer, disrupted by sonication and the liquid phase
(‘supernatant’), containing the enzyme was collected by centrifu-
gation. The GST, from the supernatant, was purified on an affinity
chromatography adsorbent bearing the tripeptide glutathione
immobilized to cross-linked agarose, previously epoxy-activated
with bis-epoxirane (1,4-butanediol diglycidyl ether). Non adsorbed
protein was washed off with equilibration buffer, prior to desorb-
ing bound GST in equilibration buffer containing 10 mM glutathi-
one. Fractions with enzyme activity were polled (specific activity
ꢂ 83 enzyme units per mg protein), concentrated (nitrocellulose
filter, cutoff 10 kDa) and diluted by dropwise addition of glycerol
to 50% (v/v) final concentration (typically 445 enzyme units per
mL stock solution). The enzyme solution can be stored at –20 °C
for several months without appreciable loss of activity.
D
A340/min) after replacing the inhibitor by an equal volume of
DMSO (5 L). The IC50 values were determined from a graph
l
depicting remaining GST activity (%) versus inhibitor
concentration.
3.5.2. Kinetic analysis of inhibitors 6, 8, 14 & 16 using CDNB as a
variable substrate
Initial velocities for the hGSTA1-1-catalysed reaction with
CDNB as variable substrate were determined in reaction mixtures
of a total volume of 1 mL (25 °C) containing potassium phosphate
buffer (100 mM, pH 6.5), 2.5 mM GSH and different concentrations
of CDNB (typically 37.5–980
inhibitor 6 (0, 0.50 and 1.70
0.50 M) or inhibitor 14 (0, 0.37 and 0.74
0.05, 0.20 and 0.60 M).
l
M) in the absence and presence of
M) or inhibitor 8 (0, 0.25 and
M) or inhibitor 16 (0,
l
l
l
l
3.5.3. Kinetic analysis of inhibitors 6, 8, 14 & 16 using GSH as a
variable substrate
Initial velocities for the hGSTA1-1-catalysed reaction with GSH
as variable substrate were determined in reaction mixtures of a
total volume of 1 mL (25 °C) containing potassium phosphate buf-
fer (100 mM, pH 6.5), 1 mM CDNB and different concentrations of
3.4. Enzyme assays for testing the compounds as inhibitors for
hGSTA1-1
3.4.1. Routine enzyme assay for determining hGSTA1-1 activity
Determination of GST activity was performed by monitoring the
formation of the conjugate between CDNB and GSH at 340 nm
GSH (100–2500
1.70 and 5.10 M) or inhibitor 8 (0, 0.5, 0.8 and 1.3
tor 14 (0, 0.74 and 1.11 M) or inhibitor 16 (0, 0.4, 0.8 and 1.2
l
M) in the absence and presence of inhibitor 6 (0,
M) or inhibi-
M).
l
l
l
l
(e
= 9600 L moleꢁ1 cmꢁ1) at 25 °C. An assay volume of 1 mL con-
tained potassium phosphate buffer (100 mM, pH 6.5), 1
CDNB (33 L from a 30 mM solution in ethanol) and 2.5
L from a 75 mM aqueous solution). DMSO was also
L, in place of equal volume of buffer) only for control
l
mol of
3.6. Caco-2 cell line culture
l
l
l
lmol of
GSH (33
added (5
Caco-2 cells42 were grown as monolayer cultures in Dulbecco’s
Modified Eagle Medium (DMEM) from BIOCHROM supplemented

F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
3969
with 10% v/v fetal bovine serum (GIBCO) 1% v/v penicillin–strepto-
mycin solution (GIBCO) and 1% v/v -glutamine (GIBCO). The cells
were grown in standard conditions until 60–70% confluency and
and green biotechnology’ falling under the Operational Programme
‘Education and Lifelong Learning’ and is co-financed by the Euro-
pean Social Fund and National Resources.
L
maintained at 37 °C in an incubator with 5% CO2.
Supplementary data
3.7. Cytotoxicity experiments for determining LC50 values for
Caco-2 cells with compounds 5, 6, 8, 11, 14 & 16
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmc.2014.06.007.
Cytotoxicity was evaluated in Caco-2 cells using the MTT assay,
which measures the ability of viable cells to reduce a soluble tetra-
zolium salt to an insoluble purple formazan precipipate.43 Caco-2
cells used for MTT assay were seeded at a density 1.0 ꢃ 104 cells/
well in 96-well plates and pre-incubated for 48 h in DMEM con-
taining 10% FBS before the addition of the inhibitors. These were
dissolved in 100% DMSO and then diluted with serum-free DMEM
as culture medium to different concentrations and added to Caco-2
cells cultured in serum-free medium for an additional 24 h.44 After
removal of the medium, each well was incubated with 0.5 mg/mL
MTT (Sigma–Aldrich) in DMEM serum-free medium at 37 °C for
3 h. At the end of the incubation period, the medium was removed
References and notes
1. Acuna, U. M.; Jancovski, N.; Kenelly, E. J. Curr. Top. Med. Chem. 2013, 9, 1560.
2. Alvarez, C.; Alvarez, R.; Corchete, P.; Pirez-Melero, C.; Pelaez, R.; Medarde, M.
Bioorg. Med. Chem. 2008, 16, 8999. and references cited therein.
3. Terzidis, M. A.; Tsiaras, V. G.; Stephanidou-Stephanatou, J.; Tsoleridis, C. A.
Synthesis 2011, 97.
4. Dang, A.-T.; Miller, D. O.; Dawe, L. N.; Bodwell, G. J. Org. Lett. 2008, 10, 233. and
references cited therein.
5. Chen, H.; Xie, F.; Gong, J.; Hu, Y. J. Org. Chem. 2011, 76, 8495. and references
cited therein.
6. Storm, J. P.; Andersson, C.-M. J. Org. Chem. 2000, 65, 5264. and references cited
therein.
7. Kaiser, F.; Schwink, L.; Velder, J.; Schmalz, H.-G. J. Org. Chem. 2002, 67, 9248.
8. Martin, R. Aromatic Hydroxyketones: Preparation and Physical Properties, 3rd
ed., vol. 1, 2011, Springer.
9. Romão, M. J.; Knäblein, J.; Huber, R.; Moura, J. J. G. Prog. Biophys. Mol. Biol. 1997,
68, 121.
and the intracellular formazan was solubilised with 200 lL DMSO
and quantified by reading the absorbance at 550 nm on a micro-
plate reader (Optimax, Molecular Devices). Percentage of cell
viability was calculated based on the absorbance measured relative
to the absorbance of cells exposed to the negative control. The
GraphPad PRISM version 5 computer program was used for
producing cytotoxicity graphs and determining LC50 values.
10. Odrowaz-Sypniewski, M. R.; Tsoungas, P. G.; Varvounis, G.; Cordopatis, P.
Tetrahedron Lett. 2009, 50, 5981.
11. Gardikis, Y.; Tsoungas, P. G. .; Potamitis, C.; Zervou, M.; Cordopatis, P.; Hornby,
J. A.; Hornby, J. A. Heterocycles 2011, 83, 1077.
12. Gardikis, Y.; Tsoungas, P. G.; Potamitis, C.; Zervou, M.; Pairas, G.; Cordopatis, P.
Heterocycles 2011, 83, 1291.
13. Zoi, O. G.; Thireou, T. N.; Rinotas, V. E.; Tsoungas, P. G.; Eliopoulos, E. E.; Douni,
E. K.; Labrou, N. E.; Clonis, Y. D. J. Biomol. Screen. 2013, 18, 1092.
14. Frova, C. Biomol. Eng. 2006, 23, 149.
3.8. Modeling and docking: the in silico structure of hGSTA1-1
and docking of the 2,20-dihydroxybenzophenones and their N-
carbonyl analogues
15. Oakley, A. J. Drug Metab. Rev. 2011, 43, 138.
16. Kodera, Y.; Isobe, K.; Yamauchi, M.; Kondo, K.; Akiyama, S.; Ito, K.; Nakashima,
I.; Takagi, H. Cancer Chemother. Pharmacol. 1994, 34, 203.
17. Sau, A.; Trengo, F. P.; Valentino, F.; Federici, G.; Caccuri, A. M. Arch. Biochem.
Biophys. 2010, 500, 116.
18. Yang, X.; Liu, G.; Li, H.; Zhang, Y.; Song, D.; Li, C.; Wang, R.; Li, B.; Liang, W.; Jing,
Y.; Zhao, G. J. Med. Chem. 2010, 53, 1015.
19. Koutsoumpli, G. E.; Dimaki, V. D.; Thireou, T. N.; Eliopoulos, E. E.; Labrou, N. E.;
Varvounis, G. I.; Clonis, Y. D. J. Med. Chem. 2012, 55, 6802.
20. Axarli, I.; Labrou, N. E.; Petrou, C.; Rassias, N.; Cordopatis, P.; Clonis, Y. D. Eur. J.
Med. Chem. 2009, 44, 2009.
21. Johansson, K.; Ito, M.; Schophuizen, C. M. S.; Thengumtharayil, S. M.; Heuse, V.
D.; Zhang, J.; Shimoji, M.; Ang, W. H.; Dyson, P. J.; Shibata, A.; Shuto, S.; Ito, Y.;
Abe, H.; Morgenstern, R. Mol. Pharm. 2011, 8, 1698.
22. Cacciatore, I.; Caccuri, A. M.; Di Stefano, A.; Luisi, G.; Nalli, M.; Pinnen, F.; Ricci,
G.; Sozio, P. Il Farmaco 2003, 58, 787.
23. Adang, A. E.; Brussee, J.; van der Gen, A.; Mulder, G. J. J. Biol. Chem. 1991, 266,
830.
24. Razza, A.; Galili, N.; Smith, S.; Godwin, J.; Lancet, J.; Melchert, M.; Jones, M.;
Keck, J. G.; Meng, L.; Brown, G. L.; List, A. Blood 2009, 113, 6533.
25. Abd-El-Aziz, A. S.; Bernardin, S. Coord. Chem. Rev. 2000, 203, 219.
26. Tlili, A.; Xia, N.; Monnier, F.; Taillefer, M. Angew. Chem., Int. Ed. 2009, 48, 8725.
27. Yuan, Y.; Thomé, I.; Kim, S. H.; Chen, D.; Beyer, A.; Bonnamour, J.; Zuidema, E.;
Chang, S.; Bolm, C. Adv. Synth. Catal. 2010, 352, 2892.
The structure of hGSTA1-1 in complex with ethacrynic acid and
its glutathione conjugate was downloaded from the Protein Data
Bank (PDB code 1GSE) and prepared with the Protein Preparation
Wizard45 in Maestro (Schrodinger, LLC, New York, NY). A grid includ-
ing the tripeptide substrate glutathione was set up centered on
ethacrynic acid. The synthetic analogues were docked flexibly using
Glide SP.46,47 Docking results were both visually inspected and
quantitatively evaluated based on docking score. In order to study
analogue binding in the presence of CDNB, another grid was set up
including both the tripeptide substrate glutathione and CDNB, and
in silico molecular docking was repeated. Forty different global
molecular properties have been predicted for the compounds using
QikProp (Schrodinger, LLC, New York, NY). All figures depicting 3D
models were created using PyMOL, version 1.4 (Schrodinger, LLC).
4. Conclusions
28. Cox, P. J.; Kechagias, D.; Kelly, O. Acta Cryst. 2008, B64, 206.
29. Jaccard, G.; Lauterwein, J. Helv. Chim. Acta 1986, 69, 1469.
30. Andersen, K. B.; Langgard, M.; Spanget-Larsen, J. J. Mol. Struct. 1999, 509, 153.
31. Gilli, G.; Gilli, P. The Nature of the hydrogen bond. Outline of a comprehensive
H bond theory, Monograph, 2009; Oxford Scholarship online Monographs.
32. Lu, L.; He, L. J. Mol. Struct. 2012, 1010, 79.
33. Kolev, T.; Stambolyiska, B.; Yancheva, D. Chem. Phys. 2005, 324, 489.
34. Kolobe, D.; Sayed, Y.; Dirr, H. W. Biochem. J. 2004, 382, 703.
35. Copeland, R. A. Anal. Biochem. 2003, 320, 1.
36. Reaction mechanisms incorporating coupled proton and electron transfers
(PCET) concertedly, through proton relay, is currently a particularly active field
(see Bonin, J.; Constantin, C.; Robert, M.; Savéant, J.-M.; Tard, C. Acc. Chem. Res.
2012, 45, 372).
37. Dixon, M.; Webb, E. Enzymes (3rd ed.); Longman Group Ltd: London, UK, 1979;
pp 332–381.
2,20-Benzophenones and N-carbonyl analogues have been
investigated as inhibitors for the MDR-involved human GST isoen-
zyme A1–1. 2,20-Dihydroxybenzophenones 6 and 8 and the N-acy-
lhydrazone analogues 14 and 16 stood out after screening a
structure-based library of candidate inhibitors. All four structures
showed strong hGSTA1-1 inhibition potency (IC50 values in the
lower micromolar to sub-micromolar range), interacting at the
CDNB-binding site of the enzyme. Furthermore, on account of their
cytotoxicity (LC50(6) = 31.4 0.4
enzyme inhibition (IC50(6) = 1.77 0.10
M) profiles, benzophenone 6 and its N-acyl hydrazone analogue
l
M; LC50(14) = 87 1.9
lM) and
lM; IC50(14) = 0.33 0.05
l
38. Leskovac, V. Comprehensive Enzyme Kinetics; Kluwer Academic Publishers: New
York, USA, 2003; pp 73–94. 95–110, 139–170.
14 appear to be promising lead structures.
39. Mahajan, S.; Atkins, W. M. Cell. Mol. Life Sci. 2005, 62, 1221.
40. Sayed, Y.; Hornby, J. A.; Lopez, M.; Dirr, H. Biochem. J. 2002, 363, 341.
41. Beaumont, P. O.; Moore, M. J.; Ahmad, K.; Payne, M. M.; Lee, C.; Riddick, D. S.
Cancer Res. 1988, 58, 947.
42. Sambuy, Y.; De Angelis, I.; Ranaldi, G.; Scarino, M. L.; Stammati, A.; Zucco, F. Cell
Biol. Toxicol. 2005, 21, 1.
Acknowledgments
The present work was partly supported by the action THALES:
‘Glutathione transferases, multifunctional molecular tools in red

3970
F. D. Perperopoulou et al. / Bioorg. Med. Chem. 22 (2014) 3957–3970
43. Scudiero, D. A.; Shoemaker, R. H.; Paull, K. D.; Monks, A.; Tierne, S.; Nofziger, T.
H.; Currens, M. J.; Seniff, D.; Boyd, M. R. Cancer Res. 1988, 48, 4827.
44. Kowapradit,J.;Opanasopit,P.;Ngawhirunpat, T.;Apirakaramwong,A.;Rojanarata,
T.; Ruktanonchai, U.; Sajomsang, W. AAPS Pharm. Sci. Tech. 2010, 11, 497.
45. Sastry, G. M.; Adzhigirey, M.; Day, T.; Annabhimoju, R.; Sherman, W. J. Comput.
Aid. Mol. Des. 2013, 27, 221.
46. Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic, J. J.; Mainz, D. T.;
Repasky, M. P.; Knoll, E. H.; Shaw, D. E.; Shelley, M.; Perry, J. K.; Francis, P.;
Shenkin, P. S. J. Med. Chem. 2004, 47, 1739.
47. Halgren, T. A.; Murphy, R. B.; Friesner, R. A.; Beard, H. S.; Frye, L. L.; Pollard, W.
T.; Banks, J. L. J. Med. Chem. 2004, 47, 1750.