Fragment-based design, docking, synthesis, biological evaluation and structure-activity relationships of 2-benzo/benzisothiazolimino-5-aryliden-4- thiazolidinones as cycloxygenase/lipoxygenase inhibitors

Article abstract of DOI:10.1016/j.ejmech.2011.10.029

Balanced modulation of several targets is one of the current strategies for the treatment of multi-factorial diseases. Based on the knowledge of inflammation mechanisms, it was inferred that the balanced inhibition of cyclooxygenase-1/cyclooxygenase-2/lip

Full text of DOI:10.1016/j.ejmech.2011.10.029

European Journal of Medicinal Chemistry 47 (2012) 111e124
Contents lists available at SciVerse ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Fragment-based design, docking, synthesis, biological evaluation and
structureeactivity relationships of 2-benzo/benzisothiazolimino-5-aryliden-4-
thiazolidinones as cycloxygenase/lipoxygenase inhibitors
,
a
c
d,
Phaedra Eleftheriou b, Athina Geronikaki a , Dimitra Hadjipavlou-Litina , Paola Vicini , Olga Filz
*
**
,
Dmitry Filimonov d, Vladimir Poroikov d, Shailendra S. Chaudhaerye, Kuldeep K. Roye, Anil K. Saxena e
a School of Pharmacy, Department of Pharmaceutical Chemistry, Aristotle University, Thessaloniki 54124, Greece
b Department of Medical Laboratory Studies, School of Health and Medical Care, Alexander Technological Education Institute of Thessaloniki, Thessaloniki 57400, Greece
c University of Parma, Dipartimento Farmaceutico, Parma 43124, Italy
d Institute of Biomedical Chemistry of Rus. Acad. Med. Sci., Moscow 119121, Russia
e Central Drug Research Institute, Chattar Manzil Palace, Lucknow 226 001, India
a r t i c l e i n f o
a b s t r a c t
Article history:
Balanced modulation of several targets is one of the current strategies for the treatment of multi-factorial
diseases. Based on the knowledge of inflammation mechanisms, it was inferred that the balanced
inhibition of cyclooxygenase-1/cyclooxygenase-2/lipoxygenase might be a promising approach for
treatment of such a multifactorial disease state as inflammation. Detection of fragments responsible for
interaction with enzyme’s binding site provides the basis for designing new molecules with increased
affinity and selectivity. A new chemoinformatics approach was proposed and applied to create a frag-
ment library that was used to design novel inhibitors of cycloxygenase-1/cycloxygenase-2/lipoxygenase
enzymes. Potential binding sites were elucidated by docking. Synthesis of novel compounds, and the
in vitro/in vivo biological testing confirmed the results of computational studies. The benzothiazolyl
moiety was proved to be of great significance for developing more potent inhibitors.
Received 13 July 2011
Received in revised form
11 October 2011
Accepted 13 October 2011
Available online 25 October 2011
Keywords:
Multi-target drugs
Inflammation
COX-1/2 inhibitors
Lipoxygenase inhibitors
Benzo/benzisothiazolidinones
Fragment-based drug design
Ó 2011 Elsevier Masson SAS. All rights reserved.
1. Introduction
cyclooxygenase (COX), which mediates the production of prosta-
glandins, prostacyclins and thromboxanes from arachidonic acid.
The “one-molecule-one-target” concept led to the discovery of
many successful drugs in the XX century and will probably lead to
many more pharmaceutical agents in the future [1]. However,
a balanced modulation of several targets can provide advanced ther-
apeutic effects and a favourable side effects profile compared to the
action of a selective ligand, particularly for complex diseases (cancer,
diabetes, atherosclerosis, stroke, depression, and many others) [2e9].
Inflammation is a multifactorial process which reflects the
response of the organism to various stimuli and is related to many
disorders such as arthritis, asthma, psoriasis, etc. Non-steroidal
anti-inflammatory agents (NSAIDs) [10e16] act by inhibition of
Old generation of NSAIDs, mainly non-selective or selective COX-1
inhibitors, were associated with gastrointestinal problems [12],
renal disorders [13] and bleeding diathesis, since COX-1 isoenzyme
is needed for mucus formation and thromboxan production [14].
On the other hand, certain new generation NSAIDs, selective COX-2
inhibitors, were associated with serious cardiovascular problems
due to the increased platelet aggregation [13,15,16]. Compounds
that combine COX-1/2 and lipoxygenase (LOX) inhibition present
multiple advantages because they act on the two major arachidonic
acid metabolic pathways [15]. Since leucotrienes play a role in
blood coagulation and gastric tract irritation, LOX inhibition can
ameliorate GI tract irritation caused by COX-1 inhibition as well as
the prothrombotic tendency resulting from COX-2 inhibition [17].
Several derivatives with pyrazoline, thiophene, di-tert-butylphenol,
hydrazone, pyrrolidine, and pyrazole subunits have been found to
be potent dual COX/LOX inhibitors [11]. More recently, some other
compounds such as a number of 1,3-diarylprop-2-yn-1-ones,
rofecoxib [18] as well as a number of diaryl isoselenazoles [19],
Abbreviations: MNA, multilevelneighbourhoods of atoms; FBDD, fragment-
based drug design.
* Corresponding author. Tel.: þ30 2310 997616.
** Corresponding author. Tel.: þ30 2310 997616; fax: þ30 2310 998559.
E-mail addresses: geronik@pharm.auth.gr (A. Geronikaki), fioland@yandex.ru
(O. Filz).
0223-5234/$ e see front matter Ó 2011 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ejmech.2011.10.029

112
P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
novel acrylic acid [20] and nemesulide derivatives [21] were tested
for COX-1, COX-2, and LOX inhibitory activity. Licofelone, a COX-2/5-
LOX inhibitor, has been evaluated in phase III clinical trials [17].
In the present study PASS-based approach, appropriately
modified, was proposed for rational, fragment-based design of
pharmaceutical agents that simultaneously inhibit COX-1, COX-2
and LOX enzymes, involved in the inflammation process, in order to
develop anti-inflammatory agents with activity superior to the
earlier obtained substances [18,20,22]. FBDD can be applied for
finding multi-target ligands since it enables to (1) include in the
designed compounds fragments or functional groups that increase
affinity to the studied targets and bioavailability, and (2) exclude
fragments that increase the risk of side and toxic effects.
which are formed by common chemical reactions. Several methods
for splitting of molecules into a set of cyclic and acyclic fragments
had been developed [35e37]. In contrast to these methods our
approach for splitting the whole structure into fragments is based
on the probability of a chemical bond to break, which is inversely
proportional to the energy of the bond formation, available from
the literature [38]. The developed rules include some empirical
restrictions on splitting structures into linkers and rings. Such an
approach should provide splitting molecules, which are potentially
synthesizable but occur rarely in the set of common chemical
reactions’ products.
The rules are presented below:
Selection of fragments that can be used for the design of new
chemicals with the required biological activity profile is a key stage
in FBDD. Currently, both experimental (X-ray diffraction, NMR) and
computational methods (molecular modelling, similarity search,
(Q)SAR) are applied to fragment library preparation [23]. X-ray
determination of 3D structure of proteins is difficult in case of
membrane-spanning proteins, such as G-protein coupled receptors
that are known to be important drug targets [24]. On the other
hand, NMR is heavily used for determination of 3D structure of
large proteins. Molecular modelling also requires the accurate
determination of 3D structure of “ligandeprotein” complexes. In
the methods based on maximal common substructure comparison
or similarity with the structures from the training set [25,26]
typically relatively large substructures are used as predefined
scaffold for the search. This significantly reduces the chance to find
principally new compounds without such scaffold. The main
problem of QSAR application to fragment-based drug design
[27e29] is associated with necessity to collect large homogeneous
datasets of chemical structures with experimentally determined
self-consistent quantitative values of activity [30].
It was shown earlier that, using the Bayesian approach imple-
mented in computer program PASS (Prediction of Activity Spectra
for Substances) [31,32], it is possible to identify the multi-target
antihypertensive agents (dual angiotensin-converting enzyme/
neutral endopeptidase inhibitors) [33] and multi-target anti-
inflammatory agents (dual cyclooxygenase/lipoxygenase inhibi-
tors) among the commercially available substances and virtually
designed chemical structures [22]. Here we applied the FBDD
method, based on the PASS algorithm, to the preparation of a frag-
ment library, focused on the inhibition of COX-1, COX-2 and LOX
enzymes. We modified the standard PASS algorithm, with the aim
to allow for considering the mutual influence of atoms in the
particular fragment (contribution of atoms) and the occurrence of
entire fragment in the set of active compounds (contribution of
fragment).
- The following bond types e CeN, NeN, NeO, CeC, SeS, CeO e
as well as bonds between the ring atom and non-ring atom and
bonds connecting two cycles can be split.
- Double and triple bonds cannot be broken.
The example of splitting of the acetylsalicylic acid structure
(active pharmaceutical ingredient of Aspirin) is shown in Fig. 1.
To estimate the fragments’ contribution into the COX-1, COX-2
and LOX inhibition we have taken into account the mutual influ-
ence of atoms included in the structure of the considered fragment.
Such kind of influence could be estimated, using the representation
of each atom included in the fragment, as a set of MNA descriptors
(Multilevel Neighbourhoods of Atoms) [31].
MNA descriptors are calculated iteratively for each atom in the
structure and reflect the peculiarities of its neighbourhoods. MNA
descriptors of the 2nd level are currently used in PASS for biological
activity prediction [31]. The probability of a chemical compound to
be active (Pa) or inactive (Pi) is estimated using MNA descriptors’
frequencies in the sub-sets of active and inactive compounds from
the PASS training set. The accuracy of prediction is represented by
Independent Accuracy of Prediction (IAP) values [31], which is
calculated in leave-one-out cross validation (LOO CV) procedure for
the whole PASS training set (for details see the Supplementary data).
To determine the influence of a specific atom X on a particular bio-
logical activity, the partial contribution of this atom is calculated on the
basis of zero, the first and the second level’s MNA descriptors, obtained
for each atom positioned in the immediate vicinity of the considered
atom X. Such approach allows taking into account the influence of the
parts of molecule surrounding this atom (see Supplementary data).
For each atom X in the fragment under consideration and for the
analyzed biological activity we calculated the probabilities of
influence on the activity (Pa) and inactivity (Pi) on the basis of PASS
algorithm [31].
PaꢀPi values are used for estimation of fragments’ contribution
into the activity:
We prepared a fragment library focused on COX-1, COX-2 and
LOX inhibition, and, also, designed and synthesized novel
compounds comprising fragments of this library. All compounds
were tested in vitro for COX-1, COX-2 and LOX inhibitory activity
and in vivo for their anti-inflammatory activity in mice using the
carrageenan paw oedema model. Structureeactivity relationships
for the studied multi-target anti-inflammatory agents were
analyzed and compared with the predictions obtained using the
proposed FBDD method.
P
ðPa ꢀ PiÞ
j
Cf
¼
AN
Where AN is the number of atoms in the fragment.
Integral characteristic of the fragments’ contribution C into the
particular activity was calculated as a sum of Cf values while taking
into account each occurrence of a given fragment into the set of
active compounds from the training set (see Supplementary data).
2. Results and discussion
2.1. Computer-aided estimation of fragments’ importance for COX-
1, COX-2 and LOX inhibition
The earliest computer-aided methods employed decomposition
of molecules into a set of fragments (RECAP) [34] around bonds,
Fig. 1. An example of defragmentation of the acetylsalicylic acid structure.

P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
113
Table 1
The fragments, which have the highest C-values, are considered
as the most significant for a particular activity. Some of the most
significant fragments for COX-1, COX-2 and LOX inhibition are given
in the appropriate columns of Table 1. They are arranged in the
descending order of COX-1 C-values.
Influence of different fragments on the COX-1, COX-2 and LOX inhibition, obtained
by the estimation of fragments’ contribution.
Fragments, which have high C-values for two or three enzymes
simultaneously, can be selected for design, synthesis and testing of
novel potent anti-inflammatory agents with a multiple mechanism
of action. However organic synthesis is a complex field with many
subtleties and different restraints which should be taken into
consideration when chemical synthesis of certain molecules is
planned [39]. Based on the results of fragment library analysis and
taking into account the synthetic accessibility and SAR analysis of
the designed thiazole derivatives [22], seven fragments, designated
with dark grey background in Table 1, were selected to be incor-
porated into new anti-inflammatory candidates: benzothiazole,
benzol, phenol, fluorobenzol, chlorobenzol, anisole and 2-imino-5-
methylidene-1,3-thiazolidin-4-one.
During the process of structure design we took into account
previous investigations concerning the interactions between COX-
1/2 and thiazolidinones and LOX and thiazolidinones which were
based on the molecular docking results and were confirmed by
experimental studies [22]. According to these studies, the central
position of 2-thiazolylimino-5-methylidene-1,3-thiazolidin-4-one
plays a role in the interaction between COX-1, LOX and thiazolidi-
nones. Therefore we decided to save central position of 2-
thiazolylimino-5-methylidene-1,3-thiazolidin-4-one. Taking into
account previous results of experimental and theoretical studies on
thiazolidinones [22], we proposed to introduce fluoro-benzene,
benzole, phenole, anisole, chloro-benzene as substituents of the
5th (unoccupied) carbon atom of 2-thiazolylimino-5-methylidene-
1,3-hiazolidin-4-one. Following the similar logic we have decided to
replace thiazole by benzothiazole in de-novo designed molecule for
comparison with the previously investigated thiazolidinones. The
fragments-substituents of 2-thiazolylimino-5-methylidene-1,3-
thiazolidin-4-one (benzole, phenole, anisole, chlor-benzene) were
selected among other fragments with high C-values (for example,
phenyl acetate, di-fluor benzene, acetic acid) taking into account the
convenience of synthesis. The reason for linking 2-thiazolylimino-
5-methylidene-1,3-thiazolidin-4-one,
benzo(iso)thiazole
and
benzene-substituents into one molecule was based on the earlier
reported data on the biological activity of thiazolidinones [22].
Since the benzothiazole fragment was among the proposed ones
for incorporation, a series of twenty benzothiazolyl/benzisothia-
zolyl derivatives were selected from the bulk of already synthesized
compounds [40,41] (Fig. 2: Compounds 1e6, 8, 9,11e20, 22 and 23).
Three more compounds (Fig. 2: Compounds 7, 10 and 21) were
designed and synthesized according tothe method described below.
2.2. Chemistry
Compounds7,10 and21 (Fig. 2)weresynthesisedbythemulti-step
reaction protocol (Scheme 1), which we reported earlier [40e44].
All the new compounds were characterized with respect to mps,
elemental analyses and spectroscopic data (1H NMR, MS and IR).
Concerning the 1H NMR data, as it was shown by us earlier [41],
the low field NH signals of compounds account for the amino-
eimino tautomeric equilibrium. It is worthwhile mentioning that
in the biological systems the aminoeimino equilibrium is of great
importance in the binding process with the target biomolecule.
2.3. Biological results
The potential pharmacotherapeutic action of benzothiazolyl
(BT) and benzisothiazolyl derivatives (BIT) was evaluated by the

114
P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
In both cases (i.e. with benzothiazole and benzisothiazole
derivatives) the results confirm the findings that with increasing
the overall lipophilicity the biological response is also increased.
The opposite was observed for the thiazolylimino-5-aryliden-4-
thiazolidinones.
It should be mentioned that benzo[d]thiazolylimino-5-aryliden-
4-thiazolidinones exhibited more potent COX-1 inhibition than
their thiazole analogues and proved to be even more potent COX-1
inhibitors than naproxen.
The influence on the activity of different scaffold types
could be presented in the following order: benzothiazole >>
benzisothiazole ꢁ thiazole.
R
R
O
S
S
N
N
N
S
O
N
N
N
H
H
S
Benzothiazole series
Benzisothiazole series
1. R = 2-NO2
2. R = 3-NO2
3. R = 4-NO2
4. R = 2-Cl
5. R = 3-Cl
6. R = 4-Cl
7. R = 3-OH
8. R = 4-OH
13. R = 2-NO2
14. R = 3- NO2
15. R = 4- NO2
16. R = 2-Cl
17. R = 3-Cl
18. R = 4-Cl
19. R = 4-OH
20.R = 3-OMe, 4-OH
21.R =3,5-OMe,4-OH
22. R = H
9. R = 3-OMe, 4-OH
10.R=3,5-OMe, 4-OH
11. R = 4-OMe
2.3.1.2. COX-2 inhibition. Most of compounds added to the assay
23.R= 4-OMe
mixture in a concentration of 200
results shown in Table 2 were obtained with an arachidonic acid
substrate concentration of 0.1 M.
mM showed COX-2 inhibition. The
12. R = H
m
Fig. 2. Structure of designed compounds.
It was observed that among the benzothiazole derivatives
chloro substitution is more favourable. 5-(2-Chloro-benzylidene)-
2-(benzo[d]thiazol-2-ylimino) thiazolidin-4-one (4) exhibited the
highest COX-2 inhibition (58.8%) followed by 3- and 4-chloro
derivatives (compounds 5 and 6, 32.0% and 20.0%, respectively).
The replacement of the more lipophilic chloro substituent by the
less lipophilic nitro group, leads to a decrease of COX-2 inhibitory
activity in the case of 2- and 4-substituted derivatives (9.4% and 8%
respectively), while 3-NO2 derivative (compound 2, 32%) was the
most active. Replacement of the 4-nitro group with 4-methoxy and
4-hydroxy-group resulted in obtaining compounds with increased
activity (compounds 11 and 8, 26.4% and 20.6% respectively), while
introduction of a hydroxy-group in the 3-position led to an inactive
compound (compound 7). The same result was observed upon
introduction of a methoxy group in positions 3 and 5 of p-hydroxy
substituent (compounds 9 and 10).
Benzisothiazole-2-ylimino-5-aryliden-4-thiazolidinones were
also found to exhibit some COX-2 inhibitory activity (12.3e45.0%).
In this case, 2-(benzo[d]isothiazol-3-ylimino)-5-(4-chloro-benzy-
lidene) thiazolidin-4-one (18) exhibited the best COX-2 inhibition
(45.1%), followed by 3-Cl (29%) and 2-Cl (24.5%) derivatives.
Replacement of 4-Cl substituent with 4-nitro (15) resulted in
a decrease of inhibition (14.0%), while 2-nitro derivative had the
second most potent inhibitory activity after compound 18.
Although, no significant changes were observed within
different scaffold types and within different substituents, the
occurrence of slight variations in inhibitory activity of
compounds with different substituents points to the influence of
benzothiazole, benzisothiazole and thiazole scaffolds on the
COX-2 inhibition. This influence may be relatively ordered as:
benzothiazole z benzisothiazole > thiazole.
in vitro study of COX-1, COX-2 and LOX inhibitory activity and by the
in vivo study of anti-inflammatory activity using the carrageenan
mouse paw oedema model (for results see Table 2) [40,41]. Previ-
ously obtained data on thiazolyl derivatives (TH) are also shown for
comparison purposes [33].
2.3.1. In vitro experiments
2.3.1.1. COX-1 inhibition. With the exception of compounds 1, 10
and 11, all benzothiazol-2-yliminothiazolidin-4-ones showed good
inhibitory activity with IC50 values ranging between 0.018 and
39.8
an arachidonic acid concentration of 0.1
dene)-2-(benzo[d]thiazol-2-ylimino) thiazolidin-4-one (4) exhibi-
ted the highest inhibitory activity (IC50 0.018 M). The lower
inhibitory activity of derivatives substituted in position 3 (5) and 4
(6) (IC50 22.4 and 0.3 M respectively) shows that the addition of
mM (Table 2). The results shown in Table 2 were obtained using
m
M. 5-(2-Chloro-benzyli-
m
m
the Cl substituent at the 2-position of the phenyl ring is more
favourable. In case of nitro derivatives the positions 3- and 4- are
more favourable than position 2- (IC50, 0.31
>200 M (1), respectively). Addition of a hydroxy group at position
3 (7) and 4 (8), of the phenyl ring also resulted in less active
compounds (IC50, 5.6 M and 7.9 M, respectively). Moreover,
mM (2), 0.51 mM (3) and
m
m
m
replacement of the 4-hydroxy (8) with a 4-methoxy (11) substit-
uent diminished the inhibitory activity dramatically (IC50
200.5
methyl group. The introduction of a methoxy group at positions 3
and 3,5 of compound 8 led to less active compounds (IC50, 39.8
mM) probably due to the stereochemical hindrance of the
mM
and 200
m
M, respectively).
O
O
C
HN
a
b
Het
Het
Het
2.3.1.3. LOX inhibition. All benzothiazolylthiazolidinone derivatives
tested were more potent than their thiazolyl analogues. The pres-
ence of the OMe group at o- and m-positions of the molecule,
favours activity (compounds 9, 10 and 11). The p-OMe derivative
(11) is the most potent compound (IC50 28.2
mM). Derivative 10,
bearing an OH group at the p-position and two OMe groups at the
two m-positions, is the second more potent compound (IC50
NH2
N
H
CH2Cl
N
S
I
II
III
O
HN
Het
c
N
S
33.9
substituent in compound 9 resulted in reduced activity (IC50
35.5 M). On the other hand the p-OH derivative (8) was even less
active (IC50 36.3 M). The m-OH derivative (7) exhibited slightly
higher activity than the p-OH analogue (IC50 34.7 M). Compared to
compound 5 with an m-Cl group and compound 2 with an m-NO2
substituent compound 7 demonstrated slightly higher inhibition.
ortho-NO2 (1) substitution has a beneficial effect on LOX inhibitory
action. Interestingly, m- and p-NO2 derivatives (2 and 3) as well as
mM), while the presence of a p-OH and only one m-OMe
R
N
S
Het =
m
N
IV
m
S
m
IV, 1-12
IV, 13-23
Scheme 1. Synthesis of designed compounds. Reagents and conditions: (a) ClCOCH2Cl,
N,N-DMF, rt., h., (b) NH4SCN, EtOH, reflux, 1e3 h., (c) RC6H4CHO, MeCOOH,
MeCOONa, reflux, 2e4 h.
2

P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
115
Table 2
Experimental evaluation of LOX, COX-1, and COX-2 inhibition obtained for the set of benzothiazole- (BT), benzisothiazole-(BIT) and thiazole (TH) [25] derivatives.
R
O
R
O
R
S
S
N
S
N
N
N
S
O
N
N
N
N
H
S
N
H
H
S
TH
BT
BIT
N
R
CPEa
%
COX-1 inhibition
COX-2 inhibition %
LOX inhibition
(IC50
,
m
M)
(IC50, mM)
BT
BT
THb
BT
THb
BT
THb
1
2
3
4
5
6
7
8
2-NO2
3-NO2
4-NO2
2-Cl
3-Cl
4-Cl
3-OH
4-OH
3-OMe, 4-OH
3,5-OMe, 4-OH
4-OMe
H
e
>200
0.31
0.51
0.018
22.4
0.31
5.61
7.91
39.8
200.0
200.5
4.0
e
9.4
32.0
8.0
58.8
32.0
20.0
1.0
20.6
6.6
35.5
42.0
50.1
71.0
35.5
50.0
34.7
36.3
35.5
33.9
28.2
e
40.0
57.0
71.8
78.0
68.7
e
>200
141.3
>200
125.0
>200
e
12.1
4.51
2.11
30.4
6.2
89.1
251.2
114.6
125.9
120.0
e
e
68.7
72.7
36.5
37.8
e
158.0
0.0
6.21
e
116.0
99.5
122.2
e
9
10
11
12
>200
>200
>200
TH
0.0
26.4
6.7
2.51
0.0
TH
>100
BIT
e
BIT
BIT
BIT
TH
13
14
15
16
17
18
19
20
2-NO2
3-NO2
4-NO2
2-Cl
3-Cl
4-Cl
e
35.5
125.9
56.2
34.9
0.0
30.0
79.4
36.6
43.6
100.0
28.2
17.7
33.9
100
e
49.0
28.6
67.0
e
>200
141.3
>200
125.0
>200
158.0
e
12.1
4.5
2.1
30.4
6.2
0.0
6.2
e
89.1
251.2
114.6
125.9
120.0
116.0
122.2
99.5
e
14.0
24.5
29.0
45.0
12.3
0.0
>200
>200
52.4
>200
>200
>200
>200
e
40.3
69.8
80.0
47
73.5
37.8
47%
4-OH
3-OMe, 4-OH
3,5-OMe, 4-OH
H
21
22
23
>200
>200
e
3.0
0.0
0.0
e
4-OCH3
e
e
e
Indomethacin
Naproxen
Celecoxib
19.5
34.5
68%
0.1 (IC50
)
Nordihydroguaretic
acid
31.3
Values are means of three determinations and deviation from the mean is <10% of the mean value.
a
The percentage of inhibition of induction of oedema at the right hind paw in the carrageenan-induced paw oedema inflammation model.
Reference [25].
b
p- and o-Cl derivatives (6 and 4) exhibited distinctively low activity
with IC50 values ranging between 42.0 and 71.0 M. All derivatives
were more potent than non-substituted compound 12.
compared to the corresponding o-Cl (16, IC50 43.6
m
M) substituted
m
derivative. In both cases it seems that the lower
correlated with a higher inhibitory activity.
p values of R are
Among the benzisothiazolylthiazolidinone derivatives, the best
LOX inhibitory activity was observed for compound 19 (IC50
17.7 mM), followed by compounds 18, 13, 20 and 15. It seems that
the presence of p-OH group in this case is beneficial for LOX
inhibitory activity. The introduction of a methoxy group in m-
Taking into account differences in affinity to LOX (IC50)
between benzothiazole, benzisothiazole and thiazole deriva-
tives, the influences of benzothiazole, benzisothiazole and thia-
zole scaffolds on the LOX inhibition may be ordered as:
benzothiazole z benzisothiazole >> thiazole.
position of compound 19 led to compound 20 (IC50 33.9
the activity twice less than that of compound 19. The introduction
of a second methoxy group (compound 21, IC50100 M) led to
a dramatic decrease of activity, showing an opposite influence of
the OMe group than to observed in benzothiazolyl derivatives. The
replacement of a p-OH group with p-Cl (18) or p-NO2 (15) also leads
to compounds with decreased activity. Compound with an o-NO2
m
M) with
As seen from the inhibitory action of the above compounds on
the three enzymes, the majority of substituents exerted a similar
influence on the inhibition of COX-1 and COX-2 isoenzymes, while,
in most cases, LOX inhibition was differently affected by the
substituents. Thus, the p-OMe derivative (11) which is the most
potent LOX inhibitor among the benzothiazolyl thiazolidinones
tested is, at the same time, one of the less potent COX-1/2 inhibi-
tors. In general, the presence of OMe substituent has the opposite
m
group (13, IC50 30 mM) demonstrates a higher inhibitory activity

116
P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
effect on COX-1/2 to that on LOX inhibition. On the other hand, the
most potent COX-1 and COX-2 inhibitor (4) is one of the less potent
LOX inhibitors. However, the m-NO2 derivative, which is the second
most potent COX-1/2 inhibitor also exhibited good although not the
best LOX inhibitory action.
In a similar way, compound 19 which is the most potent LOX
inhibitor among benzisothiazolyl derivatives, is one of the less
active COX-1/2 inhibitors. However, the p-Cl derivative which is the
second more potent LOX inhibitor also exhibits good COX-1/2
inhibitory action.
Benzothiazolyl moiety strongly favoured COX-1/2 as well as LOX
inhibition compared to thiazolyl moiety. The differences were
statistically significant in all three cases (N ¼ 9, p ¼ 0.022 for COX-1,
N ¼ 9, p ¼ 0.014 for COX-2 and N ¼ 8, p ¼ 0.014 for LOX).
The presence of benzisothiazolyl moiety as opposed to thiazolyl
moiety improved inhibitory action towards all the three enzymes;
in case of LOX the improvement was statistically significant (N ¼ 8,
p ¼ 0.021).
Comparing the influence of benzothiazolyl to that of benziso-
thiazolyl moiety we conclude that statistically significant increased
inhibitory activity was observed for benzothiazoles in case of COX-1
(N ¼ 9, p ¼ 0.044).
2.3.2. In vivo experiment. Inhibition of the carrageenan-induced
oedema
The application of Wilcoxon criterion to benzothiazole/thiazole
derivatives clearly demonstrated that fragments’ contributions
estimated by PASS algorithm roughly agree with the experimental
data.
Benzothiazolyl moiety was among the fragments predicted to
have positive influence on all three enzymes inhibitory activities.
Benzisothiazole has not been identified as a significant fragment
for inhibition of COX-1, COX-2 and LOX, which could be explained
by the relative novelty of benzisothiazole derivatives (small
number of compounds from this chemical class is included in the
PASS training set).
To explain the influence of a particular fragment on the inhibi-
tion of COX-1, COX-2 and LOX depending on the chemical class (and
therefore depending on different combinations of fragments), the
splitting of the whole structure set into the sub-sets of benzothia-
zole, benzisothiazole and thiazole derivatives was performed.
These three sets were included into the PASS training set with the
activities “Lipoxygenase inhibitor, benzothiazole”, “Lipoxygenase
inhibitor, thiazole” and “Lipoxygenase inhibitor, benzisothiazole”.
The in vivo anti-inflammatory effect of the tested compounds
was assessed using the functional model of carrageenan-induced
mouse paw oedema and is presented as the percentage of inhibi-
tion of oedema induction at the right hind paw (Table 2). Oedema
was measured by the weight increase of the right hind paw
compared to uninjected left paw. Carrageenan-induced oedema is
a nonspecific inflammation that is highly sensitive to non-steroidal
anti-inflammatory drugs (NSAIDs), therefore, carrageenan has been
accepted as a suitable agent for the study of new compounds with
anti-inflammatory activity. As shown in Table 2, the majority of the
investigated compounds inhibited carrageenan-induced paw
oedema. The protection ranged up to 80%, while the reference drug,
indomethacin, induced 47% protection at an equivalent dose. As
apparent from the data presented in Table 2, chloro-derivatives are
the most active compounds compared to other benzothiazolyl
thiazolidinones.
It seems that high
inflammatory activity, e.g., for compounds 3 and 6, p-Cl > p-NO2 (
NO2 ¼ ꢀ0.28, -Cl ¼ 0.71). The most active among o-, m- and p-
p values of R are correlated with higher anti-
p
-
p
Besides, the structures with IC50 exceeding 100 mM, were included
chloro derivatives is compound 5 (m-derivative), indicating that the
substituted position in this case is important for biological activity.
Replacement of p-OH group (compound 8, 68.7%) with a methoxy
group led to less potent compound (11, 37.8%), while introduction of
OCH3 group at position meta of p-hydroxy derivative gave
compound 9 with increased activity (72.7%). Introduction of the
second methoxy group (position m) resulted in the derivative with
the much lower activity (36.5%).
to the PASS training set with indication of low activity, for instance,
“Lipoxygenase inhibitor, thiazole, low inhibition”. Similar proce-
dures were performed for COX-1 and COX-2 enzymes. Therefore,
for each enzyme and every chemical class three possible types of
influence on the interaction could be predicted: “no influence”,
“high influence” and “low influence”.
It can be inferred, that the major fragments that have a positive
influence on COX-1 inhibition are: benzothiazole and 2-imino-5-
methylidene-1,3-thiazolidine-4-one (Table 3). At the same time,
different substituents in the benzene ring, do not significantly
increase the affinity to COX-1/2 and LOX. Therefore, different
substituents play only a supplementary role in COX-1 inhibition.
This conclusion is in agreement with experimental data: differ-
ences in average IC50 values depending on the type of substituents
at the same positions are less than differences between the struc-
tures with altered scaffold type (thiazole/benzothiazole).
Although there are no significant changes in average values for
COX-2 (%) and LOX (IC50) in thiazole derivatives depending on the
type of substituent, some fragments including Cl, OH and OMe are
predicted as important in almost all cases. The assessments ob-
tained for these fragments correspond to experimental data. The
cases where the assessments are not in agreement with the
experimental data are marked by grey colour. For all the three
enzymes there are differences in assessments between the 1,3
thiazole-4-one (fragment 5) and 2-imino-5-methylidene-1,3-
thiazolidine-4-one (fragment 6).
In case of benzisothiazole derivatives compound 20 is the most
active one, being followed by compounds 22, 19 and 16. Here again
low
p
values of R are correlated with lower anti-inflammatory
p-Cl (p-
activity, e.g., for compounds 15 and 18, p-NO2
>
NO2 ¼ ꢀ0.28,
p
-Cl ¼ 0.71) - as in the case with benzothiazolyl
derivatives. Again, replacement of p-OH group with methoxy, as
well as introduction of two methoxy groups at positions 3,5
decreased the activity, while introduction of one methoxy group
led to the most potent compound (20).
No clear correlation between in vivo activity, expressed in CPE
(carrageenan paw oedema) % values, and in vitro activity, expressed
as COX-1/2 and LOX inhibition, has been established. This is not
surprising because of the several different mechanisms, operating
at the first step of oedema formation; of note, the lack of such
correlation is reported by other authors [20].
2.4. Influence of specific fragments onto the COX-1, COX-2 and LOX
inhibition
It can be concluded therefore that the nitrogen atom in 2-imino-
5-methylidene-1,3-thiazolidine-4-one (central ring of thiazole with
substituents) is the essential feature for ligand binding to COX-1,
COX-2 and LOX.
Due to the nature of MNA descriptors, the fragments important
for COX-1, COX-2 and LOX inhibition in the benzothiazole and
benzisothiazole series, appeared to be equally important for
To estimate the statistical significance of the differences in
inhibitory action of compounds containing thiazolyl, benzothia-
zolyl and benzisothiazolyl fragments, the Wilcoxon two tailed rank
test for paired samples was carried out (XLSTAT 2010, Addinsoft,
the results of which are given in Supplementary data).

P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
117
Table 3
Fragments’ contribution into the cyclooxygenase-1, cyclooxygenase-2 and lipoxygenase inhibition, depending on the chemical class where fragment occurs.a
Influence on COX-1 inhibition Influence on COX-2 inhibition Influence on LOX inhibition
TH
No
BIT
BT
TH
No
BIT
BT
TH
No
BIT
BT
Low
No
High
No
High
No
1
Low
Low
No
No
High
No
High
No
No
High
No
High
High
No
No
High
High
No
2
No
No
No
4
No
High
High
Low
High
High
High
No
No
No
No
5
Low
High
Low
Low
Low
High
High
High
High
Low
High
High
High
High
High
High
High
High
High
No
High
High
High
High
High
High
High
High
High
High
High
High
High
High
High
High
High
High
High
High
6
7
8
O
9
10
No
Low
Low
Low
Low
No
No
No
No
No
High
No
No
No
High
No
11
Low
Low
12
No
High
High
High
High
High
High
High
High
13
a
The influence of particular fragment on inhibition of COX-1, COX-2 and LOX depending on the specific chemical class is demonstrated in column TH (influence in classes of
thiazole derivatives), BIT (influence in the class of benzisothiazole derivatives).
thiazole series. This fact demonstrates the extrapolative abilities of
the proposed approach.
binding mode of chemical compounds, the docking analysis still
remains a helpful tool in SAR studies.
2.5.1. Docking to COX-1 active centre
2.5. Results of docking experiments
The binding mode of the most active compound, 2-(benzothia-
zol-2-ylimino)-5-(2-chloro-benzylidene)-thiazolidin-4-one,
4,
Docking studies were undertaken using the automated GOLD
docking program [45] in order to gain insight into the binding
mode of the most active compounds and elucidate the impact of
structural differences on their COX-1, COX-2 and LOX inhibitory
activities. Although, only crystallographic data can fully clarify the
within the binding site of COX-1 (PDB-ID: 1cqe) [46] is illustrated in
Fig. 3. The compound is oriented with its 2-chloro-benzylidene
moiety in proximity to the hydrophobic cavity containing the amino
acids Pro585 and His581. The thiazolidinone ring is in the vicinity of

118
P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
Fig. 3. Docking of 2-(benzothiazol-2-ylamino)-5-(2-chloro-benzylidene)-thiazol-4-
one (4) into the active site of COX-1 (blue dotted line represents hydrogen bonding).
(For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
the amino acid residues Gln192, Ser353, while the benzothiazolyl
moiety of the compound is oriented in the vicinity of the hydro-
phobic pocket comprised of the amino acids Pro514, His90, Phe91,
and His513, bothacting asa hydrophobic groupanda hydrogenbond
(Hb) acceptor. N atom of the benzothiazolyl ring is involved in
hydrogen bond formation with hydrogen of the side chain of Asn515
(distance ¼ 3.3 Å), while N atom of the thiazolidinone moiety
interacts with the hydrogenatomof the NHgroupof theside chainof
Gln192 (distance ¼ 2.93 Å). Two more Hbs are formed between O
atoms of the oxo group of thiazolidinone ring and H atoms of the NH
group of the side chain of the amino acids Gln192 (distance ¼ 3.45 Å)
and Gln351(distance ¼ 3.02 Å). The four hydrogen bonds and the
hydrophobic interactions assure stable complex formation and
explain the very high inhibitory activity of this compound.
Fig. 4. A. Docking of 2-(benzothiazol-2-ylamino)-5-(3-nitro-benzylidene)-thiazol-4-
one (2) (capped sticks) into the active site of COX-1 (green dotted line represents
hydrogen bonding). B. Docking of 2-(benzothiazol-2-ylamino)-5-(2-nitro-benzylidene)-
thiazol-4-one (1) into the active site of COX-1. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)
Replacement of 2-Cl substituent of compound 4 with a NO2
group in compound 1 leads to such changing of the orientation of
the compound in the active site of COX-1 which do not allow the
formation of any of the hydrogen bonds observed in case of
compound 4. The 2-nitro-benzylidene group of 2-(benzothiazol-2-
ylimino)-5-(2-nitro-benzylidene)-thiazolidin-4-one (1) lies in the
vicinity of the amino acids Pro585, His581, Gln192, Gln351, and
Phe356, in a way similar to 2-Cl-benzylidene group of compound 4,
while the benzothiazole ring is surrounded by the amino acid
His90, Thr94, Pro514 and Phe91, at a distance enabling hydrophobic
interaction (distance ¼ 2.74 Å). However, the lack of Hbs results in
reduced stability thus explaining the inactivity of this compound.
However, existence of a NO2 group at position 3 (compound 2,
Fig. 4A), has a totally different effect on the binding mode of the
compound. The m-substituted compound lacks steric obstruction of
the bulky NO2 group thus having the ability to bend and acquire
more favourable structural conformations. The orientation of the 3-
NO2 derivative now favours formation of three Hbs providing
a much more stable binding, which is in accordance with the rela-
tively high activity of the compound. Orientation of compound 2
(Fig. 4A) differs from that of compound 4. Benzothiazole moiety is
now placed in the vicinity of the amino acid residues Asn515, Thr94,
Phe356. However, N atom of the benzothiazole ring also participates
in hydrogen bond interaction with the hydrogen of the NH of the
side chain of Asn515 amino acid (distance ¼ 3.4 Å) in a way similar to
that of compound 4. The 4-oxo group of the thiazolidinone ring,
strongly contributes to complex stabilization by forming two
hydrogen bonds with hydrogen atoms offered now by the NH group
of the side chain of His90 (distance ¼ 2.5 Å) and the OH group of
Ser516 (distance ¼ 2.33 Å). On the other hand, the benzyl group of
the 3-nitro-benzylidene moiety, placed in immediate vicinity of the
amino acids Tyr355, Gln192, Leu352 and Ser516 at the mouth of the
active site, participates in hydrophobic interactions.
The number of Hbs formed, has the most crucial role in complex
stabilization although hydrophobic interactions also contribute to
stability. Asn515 seems toparticipatein Hb formationwith NH group
of the benzothiazole moiety of the most active derivatives in this
series. Interestingly, formation of one Hb between Asn515 and the
NH group of the thiazolidinone ring, was also observed in case of the
most active thiazole derivative, 2-(thiazole-2-ylimino)-5-(m-chlor-
ophenylidene)-4-thiazolidinone, studied previously [22]. In general,
orientation of the benzothiazole derivatives favours formation of
more Hbs compared to the previously tested thiazole derivatives,
explaining the best activity of these compounds. However, Hbs with
Arg120 andTyr355, observedin mostmonoaryl ordi(bi)aryl NSAIDS,
like Ibuprofen (active S-isomer) and flubriprofen [47,48], are absent
in case of benzothiazole as well as thiazol derivatives.
2.5.2. Docking to COX-2 active centre
For the docking analysis of the compounds, crystallographic
data on the mouse COX-2 enzyme (PDB-ID: 3ln1) were used. Since
the data on murine COX-2 have been reported in many studies
[49e51], the numbers of amino acid residues in murine COX-2
analogue are shown in brackets for easy comparison with crystal-
lographic data and docking results presented in the literature. The
binding site of COX-2 (PDB-ID: 3ln1) [52] is about 23 Å (between
Ser516:Og and Asp501:O); the deep cylindrical shaped pocket being
more expanded at the bottom. The affinity of selective COX-2

P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
119
ligands is mainly governed by the hydrophobic interactions within
the very confined COX-2 binding pocket as reported by Kurumbail
et al., studying the crystal structure of COX-2 bound with the
selective COX-2 inhibitor SC-558 [53].
entry of the active site (Fig. 5A). In all cases, the benzo[d]thiazole ring
is placed in the vicinity of His75(86), Tyr341(355), Ser331(345),
Leu338(352), Arg499(513) and Ile503(517) residues while the keto
group of thiazolidinone ring may now form a hydrogen bond with
Tyr341. Moreover, Hb interactions between N atom of the benzo[d]
thiazole moiety and hydrogens of the side chain keto and amidic
bond keto groups of Gln178(192) and Leu338(352) respectively
were observed with all the meta- and para-substituted compounds
in both benzo[d]thiazole and benzo[d]isothiazole series. Despite
such additional H-bonding, the failure of these compounds (meta-
and para-substituted) to be placed deep inside the hydrophobic
pocket and the lack of hydrophobic interactions lead to a formation
of a less stable complex, thus exhibiting lower inhibitory action.
In the benzo[d]isothiazole series, all derivatives attain (L-sha-
ped) conformation and are placed at the deep pocket of the COX-2,
in the vicinity of the amino acids Met99(113), Val102(116),
Ile331(345), Leu345(459), Ala512(527) and Leu517(531). Moreover,
Hb interaction with Ser516(530) is formed. Close similarity in
orientation between ortho- and para-substituted analogues of
benzo[d]isothiazole series and ortho-substituted analogues of
benzo[d]thiazole series was observed, supporting the particular
role of lipophilic interactions within the COX-2 binding pocket as an
Docking of the most active o-Cl derivative, 4, is shown in Fig. 5B.
Compound 4 is oriented deep in the cleft of the binding site
where it is stabilized in a bent (L-shaped) conformation. Hydro-
phobic interactions with the surrounding residues Leu345(359),
Met99(113), Val102(116), Ile331(345), Ala512(527) and Leu517(531)
strongly contribute to complex stabilization. Moreover, a H-bond
(2.331 Å) formed between oxygen of the keto-group of the thiazo-
lidinone ring and hydrogen of OH group of Ser516(530), a residue
being acetylated by aspirin, further stabilizes the complex. L-shaped
conformation and an overall orientation (very similar to that of
compound 4), are also attained by the o-NO2 derivative, 1. Although,
Ser516(530) is also involved in Hb formation, the presence of the
bulky NO2 group partly prevents hydrophobic interactions, thus
explaining the lower activity of the compound. Unlike ortho-deriv-
atives, the meta- and para-substituted analogues in this series
are unable to fit deep in the hydrophobic pocket due to the
steric perturbations of the meta- and para-substituents with the
surrounding residues and thus attain extended conformations at the
Fig. 5. Docking of selected derivatives in the COX-2 (PDB-ID: 3ln1) active site. The number of the amino acid analogue in murin COX-2 [1e3] is shown in brackets for easy
comparison with other crystallographic or docking results A. Docking of benzo[d]thiazole derivatives. Thick lines were used for ortho-derivatives and thin lines were used for meta-
and para-derivatives. B. Docking of compound 4. C. Docking of compound 17. D. Docking of compound 18.

120
P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
essential factor for potential inhibition of the COX-2 enzyme. In
case of benzo[d]isothiazole derivatives, para-substitution seems to
permit pretty stronger hydrophobic interactions compared to the
meta-substituted derivatives thus explaining the higher activity of
the former. Fig. 5D and C shows the docking of the most active p-Cl
benzo[d]isothiazole derivative, 18, and its less active m-Cl analogue,
17, respectively.
The bended conformation of the benzo[d]thiazole/isothiazole
derivatives is more close to the non-linear, tricyclic structure of the
diarylheterocycle class of COX-2-selective inhibitors (e.g. celecoxib
and rofecoxib). Hydrogen bond formation with Ser530 was
observed in many COX-2 inhibitors, among which diclofenac and
lumiracoxib. Ser530 as well as Tyr385 are placed deep in the cleft of
the active centre. These amino acids are normally reached mainly
by COX-2 inhibitors. Tyr-385/Ser-530 chelation of electron-rich
centres was found to be crucial for acetylation of COX-2 by
aspirin and was mainly associated with time-dependent, non-
covalent inhibition of COX-2 by inhibitors like diclofenac [54]. The
binding mode of benzo[d]thiazole/isothiazole derivatives to the
active centre strikingly differs from that of previously studied
thiazolyl derivatives which attain a more straight conformation.
Although they are placed relatively deep in the active site, they
form hydrogen bond interactions with Arg120 and are less potent
inhibitors than their benzo[d]thiazole/isothiazolyl analogues.
Interaction with Arg120 was observed in most carboxylic acid
inhibitors, among which indomethacin, a non-selective, time-
dependent, functionally irreversible NSAID [55].
2.5.3. Docking to LOX active centre
Lipoxygenases are enzymes containing a single polypeptide
chain with a molecular mass of 75e80 kDa in animals and
94e104 kDa in plants [56,57] which share a great identity with each
other. They all have two-domains: the small N-terminal b-barrel
domain and a larger catalytic domain, containing a non-haem iron
atom which participates in catalysis. The catalytic iron is ligated to
three conserved histidines, one His/Asn/Ser, and the C-terminal
isoleucine [58]. The highest sequence identity between LOXs is in
the catalytic domain near the iron atom [59]. A high identity (63%)
was observed between soybean-3 and potato 5-LOX isoenzymes,
according to NCBI protein BLAST [60]. Interestingly, higher identity
is observed between soybean LOX b and human 5-LOX than
between human 5-LOX and potato 5-LOX enzymes.
The binding site of LOX (PDB-ID: 1lox) [52] is a deep and wide
hydrophobic pocket formed mainly by the side chains of 23 amino
acid residues, namely Ile172, Ile173, Phe175, Ser178, Phe353,
Glu357, His361, Leu362, His366, Ile400, Arg403, and Ala404,
Leu408, Val409, Ile414, Phe415, Ile418, Met419, His545, Gln548,
Ile593, Val594, Gln596, and Leu597. The Leu597 located at the C-
terminal is a key determinant for controlling the shape and size of
the cavity. The non-haem iron is ligated to residues, His361, His366,
His545, and Ile663 forming the site for the regioselective and
stereospecific hydroperoxidation of polyunsaturated fatty acids.
Molecular docking of the known LOX inhibitor Nordihy-
droguaiaretic acid (NDGA), and compounds of both the benzo[d]
thiazole and benzo[d]isothiazole series into the active site of LOX
suggest that hydrophobic interactions play a major role in modu-
lating respective LOX inhibitory activities. Docking studies showed
that compounds of the benzo[d]isothiazole series were consider-
ably better fitted into the active site of LOX than compounds of the
benzo[d]thiazole series, due to the observed steric disturbances in
case of the the benzo[d]thiazole derivatives.
Fig. 6. Predicted binding mode of compound 19 at the active site of LOX. The ligands are
shown in stick form with green coloured carbon. Hydrogens of the protein are not
shown for better visualization of the binding poses. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
Phe415, Ile418, Met419, Ile593, Val594, and Leu597, while nitrogen
of the thiazolidinone ring is involved in H-bond interaction with
Gln548, an amino acid residue present in close proximity to the
non-haem iron. On the other hand, the phenyl ring of the benzy-
lidene moiety is in parallel disposition to the His366 residue
(3.31 Å) exhibiting aromatic pep interaction. The double bond C]
C attached to the thiazolidinone core is placed above the non-haem
iron (Fe) with FeeCH distance of 2.78 Å, while the 4-OH substituent
of the benzylidene moiety is in close proximity to His366, Leu362
and Asn401. It is obvious that, although hydrophobic interactions
play a major role in complex stabilization, H-bond formation,
aromatic pep interactions and iron coordination bonds participate
in stability. Compounds 13 and 18, bearing a 2-nitro and 4-chloro
group respectively, in place of the 4-OH substituent of compound
19, exhibited an overall orientation similar to compound 19.
However, the slight differences resulting in lack of certain inter-
actions like H-bond formation may be the reason behind compar-
atively lower LOX inhibitory activity of these compounds compared
to compound 19.
The predicted binding mode of the most active derivative of the
benzo[d]thiazole series, 11 (LOX IC50 ¼ 28.2
mM), into the active site
of LOX is shown in Fig. 7. This compound attains a bent-shaped
conformation with the benzo[d]thiazole moiety oriented in the
hydrophobic pocket formed by the amino acid residues Phe353,
Glu357, Leu358, Phe415, Ile418, Met419, Ile593, Val594, and Leu597,
and the thiazolidinone ring placed in parallel (4.64 Å) to the His361
with the keto group projected towards the plane of His366. Unlike
the most potent compound 19, the 4-methoxybenzylidene group is
now placed in the pocket formed by Phe175, Leu597, Arg403 and
Ile663 amino acid residues, exhibiting hydrophobic interactions,
Fig. 6 shows the optimal binding mode of the most potent
compound 19 (LOX IC50 ¼ 17.7
mM) into the active site of 15-LOX.
The benzo[d]isothiazole moiety is placed in the hydrophobic cavity
formed by the amino acid residues Phe353, Glu357, Leu358,

P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
121
influences of benzothiazole, benzisothiazole and thiazole scaffolds
on COX-1, COX-2 and LOX inhibition can be ordered as:
benzothiazole z benzisothiazole >> thiazole. Docking studies of
the most active compounds explain the observed activity.
The results of docking correspond well to the fragment-based
analysis on the basis of the PASS algorithm.
Tanimoto coefficients (Tc) were calculated for the pairs of thia-
zolidinones/benzothiazolylthiazolidinones and thiazolidinones/
benzisothiazolylthiazolidinones (see Supplementary data). Minimal
value of Tc is 0.48 for o-Cl, m-Cl and p-Cl substituents of thiazole/
benzisothiazole derivatives. Maximal Tc-value is 0.7 for 4-hydroxy-
3,5-dimethoxyphenyl-methylidene substituents of thiazole/ben-
zothiazole. As was shown in our earlier paper [61], if the estimated
Tc-values do not exceed 0.7, the synthesized compounds can be
considered as New Chemical Entities. Therefore, we have shown
that the proposed FBDD approach provides the possibility to design
the multi-target compounds with a considerable chemical novelty.
4. Experimental section
Melting points (mp ꢂC) were determined with a Buchi 512
apparatus or with a Boetius apparatus and are uncorrected.
Elemental analysis was performed on a ThermoQuest (Italia) Fla-
shEA 1112 Elemental Analyser, for C, H, N, and S. The values found
for C, H, N, S were within ꢃ0.4% of the theoretical ones. IR spectra,
such as KBr pellets, were recorded on a JASCO FT-IR 300E spec-
trophotometer (Jasco Ltd., Tokyo, Japan), while in Nujol, on a Perkin
Elmer Spectrum BX. Wave numbers in the IR spectra are given in
cmꢀ1 1H NMR spectra of the newly synthesized compounds, in
.
DMSO-d6 solutions, were recorded on a Bruker AC 300 instrument
(Bruker, Karlsruhe, Germany) at 298 K. Chemical shifts are reported
as d (ppm) relative to TMS as internal standard, coupling constants J
are expressed in Hz. The progress of the reactions was monitored by
thin layer chromatography with F254 silica-gel precoated sheets
(Merck, Darmstadt, Germany). UV light was used for detection.
Solvents, unless otherwise specified, were of analytical reagent
grade or of the highest quality commercially available. Synthetic
starting material, reagents and solvents were purchased from
Aldrich Chemie (Steinheim Germany) and from Fluka.
Fig. 7. Predicted binding mode of compound 11 at the active site of LOX. The ligands are
shown in stick form with green coloured carbon. Hydrogens of the protein are not
shown for better visualization of the binding poses. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)
while the C]C bond of the 4-methoxybenzylidene moiety is still in
close proximity to the non-haem iron. The relatively distinct
orientation of this compound (11) compared to the most potent
compound 19, may be one of the possible reasons behind its lower
LOX inhibitory activity.
The high importance of hydrophobic interactions and iron
coordination in complex stabilization of LOX inhibitors has been
previously reported in the literature when activity of novel acrylic
acid derivatives was discussed [20]. Better contribution to hydro-
phobic interactions of the benzo[d]thiazolyl/isothiazolyl moiety
compared to the thiazolyl ring may be responsible for the better
activity of the benzo[d]thiazolyl/isothiazolyl derivatives compared
to the previously studied thiazolyl compounds.
4.1. Materials
All chemicals were of analytical grade and commercially avail-
able. Soybean lipoxygenase 1b, linoleic acid sodium salt, and
indomethacin were obtained from Sigma Chemical, Co. (St. Louis,
MO). For the in vivo experiments, male and female mice (g) were
used. The kit for COX activity assay was purchased from Cayman.
3. Conclusions
4.2. Chemistry
In this work we have demonstrated that, using the proposed
FBDD method, one can identify the fragments important for
simultaneous COX-1, COX-2 and LOX inhibition. Based on this
knowledge, we designed new, more active derivatives.
4.2.1. General procedure for synthesis of 2-(heteroarylimino)
thiazolidin-4-ones (III)
A solution of 2-chloro-N-(heteroaryl)acetamide (5 mmol) and
ammonium thiocyanate (10 mmol) in 20 mL of 96% ethanol was
refluxed for 1e3 h and allowed to stand overnight. The precipitate
was filtered, washed with water and then recrystallised.
Using this approach a fragment library was prepared, including
about 4200 small fragments with MW less than 200 Da. Fragments
with the highest positive influence on inhibition of COX-1, COX-2
and LOX enzymes were selected. Benzothiazole was considered
as a fragment of great significance for the inhibition of all
three enzymes and was among the fragments proposed to be
incorporated into the new scaffold. Appropriate derivatives were
designed, synthesized or selected from the bulk of already
synthesized compounds and tested in in vitro and in vivo assays,
which confirmed the computational predictions. In general the
4.2.2. General procedure for synthesis of 2-heteroarylimino-5-
benzylidene-4-thiazolidinones (IV)
To a well stirred solution of 2-(heteroarylimino) thiazolidin-4-
one (4 mmol) in 35 mL of acetic acid, buffered with sodium
acetate (9 mmol), appropriate aldehyde (6 mmol) was added. The
solution was refluxed over different periods till the completion of
the reaction, monitored by TLC. After cooling the reaction mixture

122
P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
to the room temperature the precipitate formed, was filtered
abundantly washed with water and then recrystallized.
and human recombinant COX-2 enzymes included in the “COX
Inhibitor Screening Assay” kit provided by Cayman (Cayman
Chemical Co., Ann Arbor, MI). The assay directly measures PGF2a
produced by SnCl2 reduction of COX-derived PGH2. The prosta-
noids production was quantified via enzyme immunoassay using
a broadly specific antibody that binds to all the major prostaglandin
compounds. For better visualization of compounds’ differences in
a 0e100% inhibition scale and for straight comparison with results
obtained previously for their thiazolyl analogues, COX-1 inhibitory
4.2.3. 5-(3-Hydroxybenzylidene)-2-(benzo[d]thiazol-2-ylimino)
thiazolidin-4-one (7)
Reaction time: 6 h. Yield: 42%, mp 293e294 ꢂC (dioxane). TLC:
eluent ¼ toluene/dioxane/acetic acid 90/10/5. IR(KBr):
(OeH), 3129 (NeH), 1696 (C]O), 1564 (N]C) cmꢀ1
(DMSO-d6):
1H NMR (DMSO-d6):
benzothiazole), 7.72 (s, 1H, Ar2), 7.96e8.05 (m, 2H, benzothiazole),
9.95 (s, 1H, OH), 12.93 (s, 1H, NH). Elemental Anal. for C17H11N3O2S2
C, H, N.
n
¼ 3400
1H NMR
¼ 6.95e7.56 (m, 6H, Ar,
.
d
¼
d
activity was tested at an arachidonic acid concentration of 1
COX-2 inhibitory activity was tested at an arachidonic acid
concentration of 0.1 M. Both arachidonic acid concentrations are
mM and
m
much lower than the saturating concentration. The compounds
4.2.4. 5-(4-Hydroxy-3,5-dimethoxybenzylidene)-2-(benzo[d]
thiazol-2-ylimino)thiazolidin-4-one (10)
were added to the reaction mixture at a final concentration of
200
mM for the estimation of activity %. Different compounds’
Reaction time: 4 h. Yield: 73%, m.p. 272e273 ꢂC (dioxane). TLC:
concentrations were used for the calculation of IC50 values for the
most active compounds Naproxen was tested under the same
experimental conditions to be used as a positive control.
eluent ¼ toluene/dioxane/acetic acid 90/10/5. IR(KBr):
n
¼ 3441
(OeH), 3193 (NeH), 1698 (C]O), 1581(N]C) cmꢀ1
.
1H NMR
(DMSO-d6):
d
¼ 3.88 (s, 3H, CH3eO), 7.00 (d, 1H, J ¼ 8.4 H-50), 7.21
(d, 1H, J ¼ 8.1 H-60), 7.30e7.39 (m, 2H, H-20, H-60), 7.49 (t, 1H, J ¼ 7.9
H-5), 7.71 (s, 1H, CH), 7.89 (d, 1H, J ¼ 7.8 H-4), 8.01 (d, 1H, J ¼ 8.1 H-
7), 10.01 (s, 1H, OH), 12.79 (s, 1H, NH). Elemental Anal. for
C19H15N3O4S2 (413.48): C, H, N.
4.3.2. In vivo experiments. Inhibition of the carrageenan-induced
oedema
Oedema was induced in the right hind paw of mice (AKR) by the
intradermal injection of 0.05 mL of 2% carrageenan in water. Mice of
both sexes were used, but pregnant females were excluded.
Each group was composed of 6e10 animals. The animals, which
have been bred in our laboratory, were housed under standard
conditions and received a diet of commercial food pellets and water
ad libitum prior to experimentation, but they were fasted during
the experimental period. The tested compounds (0.01 mmol/kg
body weight) were suspended in water with a few drops of Tween-
80, ground in a mortar before use and were given intraperitoneally
simultaneously with the carrageenin injection. Indomethacin was
also administered under the same conditions to be used as a posi-
tive control. The mice were euthanized 3.5 h after the carrageenin
injection. The difference between the weight of the injected and
uninjected paws was calculated for each animal. It was compared
with that in control animals (treated with water) and expressed as
a percent inhibition of the oedema CPE% values (Table 2). Each
experiment was performed in duplicate, and the standard devia-
tion was less than 10%.
4.2.5. 2-(Benzo[d]isothiazol-3-ylimino)-5-(4-hydroxy-3,5-di
methoxybenzylidene)thiazolidin-4-one (21)
Reaction time
eluent ¼ toluene/dioxane/acetic acid 90/10/5. IR(KBr):
: 6
h. Yield: 78%, m.p.247e249ꢂ. TLC:
n
¼ 3392
(OH), 3171 (NH), 2965, 2856 (CH3),1706 (C]O), 1579 (C]N) cmꢀ1
.
1H NMR (DMSO-d6):
d
¼ 3.85 (s, 6H, 2 CH3O), 6.98 (s, 2H, 20þ60),
7.54 (t, 1H, J ¼ 7.61, 6), 7.67e7.62 (m, 2H, CHþ5), 8.07 (d, 1H, J ¼ 7.91,
7) 8.16 (d, 1H, J ¼ 8.09, 4), 9.36 (s, 1H, OH), 12.65 (s, 1H, NH).
Elemental Anal. for C19H15N3O4S2 (413.48): C, H, N.
4.3. Biological assays
4.3.1. In vitro experiments
In the in vitro assays each experiment was performed at least in
triplicate and the standard deviation of absorbance was less than
10% of the average values.
4.3.1.1. Soybean lipoxygenase inhibition study in vitro. Lipoxygenase
inhibition was evaluated using soybean LOX type 1b as reported
previously [20,43]. sLOX-1 is the soybean isoenzyme mostly used in
drug screening, although sLOX-3 has also been used [62]. Because
of structural and functional similarities with mammalian LOXs,
sLOX is commonly used for both mechanistic and inhibition studies
and is widely accepted as a model for LOXs from other sources [63].
It has been shown that inhibition of plant LOX activity by NSAIDs is
qualitatively similar to the inhibition of the rat mast cell LOX and
can be used as a simple screen for such activity [61]. Similarity in
inhibition behaviour between soybean LOX-1 and human
recombinant 5-LOX or soybean LOX-3 and human blood serum LOX
has been observed in many cases [62,63].
4.4. Docking analysis
The docking experiments were carried our using the GOLD [45]
docking program. For docking studies, the atomic co-ordinates of
COX-1 (PDB-ID: 1cqe) [46], COX-2 (PDB-ID: 3ln1) [49] and LOX
(PDB-ID: 1lox) [64] were downloaded from the Brookhaven Protein
Data Bank (www.rcsb.org) and prepared using Protein Preparation
Wizard where bond orders were assigned, water and other residues
except bound ligands were deleted and finally minimized using
OPLS-2005 force field implemented in Schrodinger software
package [65]. The bound conformations of flurbiprofen, celecoxib
and ADA were used as controls in order to define the active site in
COX-1, COX-2 and LOX, respectively, and optimized 3D-structures
of new ligands were docked within 10 Å radius by running 20
genetic algorithm (GA) steps for each. The docked poses of geom-
etry optimized ligands were ranked using the GoldScore (GS), and
ChemScore (CS) to find the most optimal binding pose of each
ligand. In the GOLD program, the default parameters: population
size (100); selection-pressure (1.1); number of operations (10,000);
number of islands (1); niche size (2); and operator weights for
migrate (0), mutate (100) and crossover (100) were applied. Finally,
the top five binding poses of each ligand were analyzed to select the
best binding pose of each ligand in order to untangle the essential
parameters in terms of direct- (H-bonds) and indirect
The tested compounds, dissolved in DMSO, were added to the
reaction mixture at a final concentration of 100
mM and were
preincubated for 4 min at 28 ꢂC with soybean lipoxygenase at
a concentration of 7ꢄ10ꢀ7 w/v. Enzyme reaction was initiated by
the addition of sodium linolate at a final concentration of 0.1 mM.
The conversion of sodium linoleate to 13-hydroperoxylinoleic acid
was measured at 234 nm. Nordihydroguaretic acid, an appropriate
standard inhibitor, was used as positive control.
4.3.1.2. COX inhibitor screening assay [22]. The COX-1 and COX-2
activities of the compounds were measured using ovine COX-1

P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
123
[25] Y. Cao, T. Jiang, T. Girke, A maximum common substructure-based algorithm
for searching and predicting drug-like compounds, Bioinformatics 24 (2008)
366e374.
(hydrophobic) interactions governing binding disparities among
the series of these compounds.
[26] J. Masciocchi, G. Frau1, M. Fanton, M. Sturlese, M. Floris, L. Pireddu, P. Palla,
F. Cedrati, P. Rodriguez-Tome, S. Moro, MMsINC:
a large-scale chemo-
Appendix. Supplementary data
informatics database, Nucleic Acids Res. 37 (2009) 284e290.
[27] Q.S. Du, R.B. Huang, Y.T. Wei, Z.W. Pang, L.Q. Du, K.C. Chou, Fragment-based
quantitative structure-activity relationship (FB-QSAR) for fragment-based
drug design, J. Comput. Chem. 30 (2009) 295e304.
[28] B.M. Spowage, C.L. Bruce, J.D. Hirst, Interpretable correlation descriptors for
quantitative structureeactivity relationships, J. Cheminf. 1 (2009).
[29] M.A.C. Pérez, M.B. Sanz, L.R. Torres, R.G. Ávalos, M. Pérez González,
H. González Díaz, A topological sub-structural approach for predicting human
intestinal absorption of drugs, Eur. J. Med. Chem. 39 (2004) 905e916.
[30] D. Livingstone, Data Analysis for Chemists. Oxford University Press, Oxford,
1995.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.ejmech.2011.10.029.
References
[1] C. Melchiorre, M.L. Bolognesi, A. Minarini, M. Rosini, V. Tumiatti, Polyamines
in drug discovery: from the universal template approach to the multitarget-
directed ligand design strategy, J. Med. Chem. 53 (2010) 5906e5914.
[2] R. Morphy, Z. Rankovic, Designed multiple ligands. An emerging drug
discovery paradigm, J. Med. Chem. 48 (2005) 6523e6543.
[31] D.A. Filimonov, V.V. Poroikov, Probabilistic approach in activity prediction. in:
A. Varnek, A. Tropsha (Eds.), Chemoinformatics Approaches to Virtual
Screening. RSC Publishing, Cambridge (UK), 2008, pp. 182e216.
[32] A. Geronikaki, D. Druzhilovsky, A. Zakharov, V. Poroikov, Computer-aided
predictions for medicinal chemistry via Internet, SAR QSAR Environ. Res. 19
(2008) 27e38.
[33] A.A. Lagunin, O.A. Gomazkov, D.A. Filimonov, T.A. Gureeva, E.A. Dilakyan,
E.V. Kugaevskaya, Y.E. Elisseeva, N.I. Solovyeva, V.V. Poroikov, Computer-
aided selection of potential antihypertensive compounds with dual mecha-
nisms of action, J. Med. Chem. 46 (2003) 3326e3332.
[34] X.Q. Lewell, D.B. Judd, S.P. Watson, M.M. Hann, RECAPeretrosynthetic
combinatorial analysis procedure: a powerful new technique for identifying
privileged molecular fragments with useful applications in combinatorial
chemistry, J. Chem. Inf. Comput. Sci. 38 (3) (1998) 511e522.
[35] J. Degen, M. Rarey, FlexNovo: structure-based searching in large fragment
spaces, Chem. Med. Chem. 1 (2006) 854e868.
[3] C. Wermuth, Multitargeted drugs: the end of the “one-target-one disease”
philosophy? Drug Disc. Today 9 (2004) 826e827.
[4] P. Csermely, V. Agoston, S. Pongor, The efficiency of multi-target drugs: the
network approach might help drug design, Trends Pharmacol. Sci. 26 (2005)
178e182.
[5] J. Lehar, A.S. Krueger, W. Avery, A.M. Heilbut, L.M. Johansen, E.R. Price,
R.J. Rickles, G.F. Short, J.E. Staunton, X. Jin, M.S. Lee, G.R. Zimmermann,
A.A. Borisy, Synergistic drug combinations tend to improve therapeutically
relevant selectivity, Nat. Biotechnol. 27 (2009) 659e666.
[6] J. Rodon, J. Perez, R. Kurzrock, Combining targeted therapies: practical issues
to consider at the bench and bedside, Oncologist 15 (2010) 37e50.
[7] J. Minnerup, W.-R. Schabitz, Multifunctional actions of approved and candi-
date stroke drugs, Neurotherapeutics 6 (2009) 43e52.
[36] D. Douguet, H. Munier-Lehmann, G. Labesse, S. Pochet, LEA3D: a computer-aided
liganddesignforstructure-baseddrugdesign,J. Med. Chem.48(2005)2457e2468.
[37] C.A. Nicolaou, J. Apostolakis, C.S. Pattichis, De novo drug design using multi-
objective evolutionary graphs, J. Chem. Inf. Model. 49 (2009) 295e307.
[38] A.S. Tarendash, Let’s Review Chemistry, the Physical Setting, fourth ed. Bar-
ron’s Educational Series, New York, 2006, 228e242.
[8] M.J. Milan, Dual- and triple-acting agents for treatment core and co-morbid
symptoms of major depression: novel concepts, new drugs, Neuro-
therapeutics 6 (2009) 53e77.
[9] C.R. Rodrigues, M.P. Veloso, H. Verli, C.A. Fraga, A.L. Miranda, E.J. Barreiro,
Selective PGHS-2 inhibitors:
a rational approach for treatment of the
inflammation, Curr. Med. Chem. 9 (2002) 849e867.
[39] P.S. Kutchukian, E.I. Shakhnovich, De novo design: balancing novelty and
confined chemical space, Expert Opin. Drug Discov. 5 (2010) 789e812.
[40] P. Vicini, A. Geronikaki, M. Incerti, F. Zani, J.C. Dearden, M. Hewitt, Hetero-
arylimino-5-benzylidene-4-thiazolidinones analogues of 2-thiazolylimino-5-
benzylidene-4-thiazolidinones with antimicrobial activity: synthesis and
structureeactivity relationship, Bioorg. Med. Chem. 16 (7) (2008) 3714e3724.
[41] P. Vicini, A. Geronikaki, A. Kitka, M. Incerti, F. Zani, Synthesis and antimicrobial
activity of novel 2-thiazolylimino-5-arylidene-4-thiazolidinones, Bioorg. Med.
Chem. 14 (2006) 3859e3864.
[10] R. Ottana, R. Maccari, M.T. Barreca, G. Bruno, A. Rotondo, A. Rossi,
G. Chiricosta, R. Paola, L. Sautebin, S. Cuzzocrea, M.G. Vigorita, 5-Arylidene-2-
imino-4-thiazolidinones: design and synthesis of novel anti-inflammatory
agents, Bioorg.Med. Chem. 13 (2005) 4243e4252.
[11] T.D. Venu, S. Shashikanth, S.A. Khanum, S. Naveen, A. Firdouse, M.A. Sridhar,
J.S. Prasad, Synthesis and crystallographic analysis of benzophenone deriva-
tivesdthe potential anti-inflammatory agents, Bioorg. Med.Chem. 15 (2007)
3505e3514.
[12] X. Leval, F. Julémont, J. Delarge, B. Pirotte, J.-M. Dogné, New trends in dual 5-
LOX/COX inhibition, Curr. Med. Chem. 9 (2002) 941e962.
[42] A. Geronikaki, G. Theophilidis, Synthesis of 2-(aminoacetylamino)thiazole
derivatives and comparison of their local anaesthetic activity by method of
action potential, Eur. J. Med. Chem. 27 (1992) 709e716.
[13] J.F. Penrose, K.F. Austen, B.K. Lam, Leukotrienes: biosynthetic pathways,
release and receptor-mediated actions with relevance to disease states. in:
J.L. Gallin, R. Snyderman (Eds.), Inflammation Basic Principles and Clinical
Correlates. Lippicort Williams & Wilkins, Philadelphia, 1999, pp. 361e372.
[14] J.R. Vane, Y.S. Bakhle, R.M. Botting, Cyclooxygenases 1 and 2, Annu. Rev.
Pharmacol. Toxicol. 38 (1998) 97e120.
[15] J. Martel-Pelletier, D. Lajeunesse, P. Reboul, J.-P. Pelletier, Therapeutic role of
dual inhibitors of 5-LOX and COX, selective and non-selective non-steroidal
anti-inflammatory drugs, Ann. Rheum. Dis. 62 (2003) 501e509.
[16] J.Y. Jouzeau, B. Terlain, A. Abid, E. Nedelec, P. Netter, Cyclo-oxygenase isoen-
zymes. How recent findings affect thinking about nonsteroidal anti-
inflammatory drugs, Drugs 53 (1997) 563e582.
[43] C. Papadopoulou, A. Geronikaki, D. Hadjipavlou-Litina, Synthesis and biolog-
ical evaluation of new thiazolyl/benzothiazolyl-amides, derivatives of 4-
phenyl-piperazine, Farmaco 60 (2005) 969e973.
ꢀ
ꢀ
ꢀ
ꢁ
[44] O. Kouatli, P. Zoumpoulakis, Ch. Camoutsis, M. Sokovic, A. Ciric, J. Glamoclija,
A. Geronikaki, Novel 4-thiazolidinone derivatives as potential antifungal and
antibacterial drugs, Bioorg. Med. Chem. 18 (2010) 426e432.
[45] GOLD, Version 2.2; Cambridge Crystallographic Data Centre: Cambridge, UK.
[46] D. Picot, P.J. Loll, R.M. Garavito, The X-ray crystal structure of the membrane
protein prostaglandin H2 synthase-1, Nature 367 (1994) 243e249.
[47] S. Mittal, A. Malde, C. Selvam, K.S. Arun, P.S. Johar, S.M. Jachak, P. Ramarao,
P.V. Bharatam, H.P.S. Chawla, Synthesis and evaluation of S-4-(3-thienyl)
[17] M. Bayes, X. Rabasseda, J.R. Prous, Gateways to clinical trials, Methods Find.
Exp. Clin. Pharmacol. 24 (8) (2002) 525e551.
phenyl-
a-methylacetic acid, Bioorg. Med. Chem. Lett. 14 (2004) 979e982.
[48] M. Lindner, W. Sippl, A. Radwan, Pharmacophore Elucidation and molecular
docking studies on 5-phenyl-1-(3-pyridyl)-1H-1,2,4-triazole-3-carboxylic
acid derivatives as COX-2 inhibitors, Sci. Pharm. 78 (2010) 195e214.
[49] Md. J. Uddin, P.N. Praveen Rao, E.E. Knaus, Design of acyclic triaryl olefins:
a new class of potent and selective cyclooxygenase-2 (COX-2) inhibitors,
Bioorg. Med. Chem. Lett. 14 (2004) 1953e1956.
[50] Md. J. Uddin, P.N. Praveen Rao, R. McDonald, E.E. Knaus, A new class of acyclic
2-alkyl-1,1,2-triaryl (Z)-olefins as selective cyclooxygenase-2 inhibitors,
J. Med. Chem. 47 (2004) 6108e6111.
[51] Hye-Jung Kim, Chong Hak Chae, Kyu Yang Yi, Kyung-Lae Park, Sung-eun Yoo,
Computational studies of COX-2 inhibitors: 3D-QSAR and docking, Bioorg.
Med. Chem. 12 (2004) 1629e1641.
[52] J.L. Wang, D. Limburg, M.J. Graneto, J. Springer, J.R.B. Hamper, S. Liao,
J.L. Pawlitz, R.G. Kurumbail, T. Maziasz, J.J. Talley, J.R. Kiefer, J. Carter, The
novel benzopyran class of selective cyclooxygenase-2 inhibitors. Part 2: the
second clinical candidate having a shorter and favorable human half-life,
Bioorg. Med. Chem. Lett. 20 (2010) 7159e7163.
[18] P.N. Rao, Q.H. Chen, E.E. Knaus, Synthesis and structure-activity relationship
studies of 1,3-diarylprop-2-yn-1-ones: dual inhibitors of cyclooxygenases and
lipoxygenases, J. Med. Chem. 49 (5) (2006) 1668e1683.
[19] M. Scholz, H.K. Ulbrich, G. Dannhardt, Investigations concerning the COX/5-
LOX inhibiting and hydroxyl radical scavenging potencies of novel 4,5-
diaryl isoselenazoles, Eur. J. Med. Chem. 43 (2008) 1152e1159.
[20] E. Pontiki, D. Hadjipavlou-Litina, K. Litinas, O. Nicolotti, A. Carotti, Design,
synthesis and pharmacobiological evaluation of novel acrylic acid deriva-
tives acting as lipoxygenase and cyclooxygenase-1 inhibitors with antioxi-
dant and anti-inflammatory activities, Eur. J. Med. Chem. 46 (2011)
191e200.
[21] Yue Li, Shu-Han Chen, Tian-Miao Ou, Jia-Heng Tan, Ding Li, Lian-Quan Gu, Zhi-
Shu Huang, Syntheses and characterization of nimesulide derivatives for dual
enzyme inhibitors of both cyclooxygenase-1/2 and 5-lipoxygenase, Bioorg.
Med.Chem. 19 (2011) 2074e2083.
[22] A.A. Geronikaki, A.A. Lagunin, D.I. Hadjipavlou-Litina, P.T. Elefteriou,
D.A. Filimonov, V.V. Poroikov, I. Alam, A.K. Saxena, Computer-aided discovery
of anti-inflammatory thiazolidinones with dual cyclooxygenase/lipoxygenase
inhibition, J. Med. Chem. 51 (2008) 1601e1609.
[23] M. Fischer, R.E. Hubbard, Fragment-based ligand discovery, Mol. Interven-
tions. 9 (2009) 23e30.
[53] R.G. Kurumbail, A.M. Stevens, J.K. Gierse, J.J. McDonald, R.A. Stegeman, J.Y. Pak,
D. Gildehaus, J.M. Miyashiro, T.D. Penning, K. Seibert, P.C. Isakson,
W.C. Stallings, Structural basis for selective inhibition of cyclooxygenase-2 by
anti-inflammatory agents, Nature 384 (1996) 644e648.
[54] S.W. Rowlinson, J.R. Kiefer, J.J. Prusakiewicz, J.L. Pawlitz, K.R. Kozak,
A.S. Kalgutkar, W.C. Stallings, R.G. Kurumbail, L.J. Marnett, A novel mechanism
[24] J.P. Overington, B. Al-Lazikani, A.L. Hopkins, How many drug targets are there?
Nat. Rev. Drug Discov. 5 (2006) 993e996.

124
P. Eleftheriou et al. / European Journal of Medicinal Chemistry 47 (2012) 111e124
of cyclooxygenase-2 inhibition involving interactions with Ser-530 and Tyr-
385, J. Biol. Chem. 278 (2003) 45763e45769.
[55] A.L. Blobaum, L.J. Marnett, Structural and functional basis of cyclooxygenase
inhibition, J. Med. Chem. 50 (2007) 1225e1441.
[56] AR Brash Lipoxygenases: occurrence, functions, catalysis, and acquisition of
substrate, J. Biol. Chem. 274 (1999) 23679e23682.
[57] D. Shibata, B. Axelrod, Plant lipoxygenases, J. Lipid Mediat Cell Signal. 12
(1995) 213e228.
[61] R.K. Somvanshi, A.K. Singh, M. Saxena, B. Mishra, S. Dey, Development of novel
peptide inhibitor of lipoxygenase based on biochemical and BIAcore
evidences, Biochim. Biophys. Acta 1784 (2008) 1812e1817.
[62] D. Lapenna, G. Ciofani, S.D. Pierdomenico, M. Neri, C. Cuccurullo,
M.A. Giamberardino, F. Cuccurullo, Inhibitory activity of salicylic acid on
lipoxygenase-dependent lipid peroxidation, Biochim. Biophys. Acta 1790
(2009) 25e30.
[63] I.B. Taraporewala, J.M. Kauffman, Synthesis and structureeactivity relation-
ships of anti-inflammatory 9,10-dihydro-9-oxo-2-acridinealkanoic acids and
4-(2-carboxyphenyl)aminobenzenealkanoic acids, J. Pharm. Sci. 79 (2) (1990)
173e178.
[58] W. Minor, J. Steczko, B. Stec, Z. Otwinowski, J.T. Bolin, R. Walter, B. Axelrod,
Crystal structure of soybean lipoxygenase L-1 at 1.4 Å resolution, Biochem-
istry 35 (1996) 10687e10701.
[59] S.T. Prigge, J.C. Boyington, M. Faig, K.S. Doctor, B.J. Gaffney, L.M. Amzel,
Structure and mechanism of lipoxygenases, Biochimie 79 (1997) 629e636.
[60] P. Aparoy, R.N. Reddy, Lalitha Guruprasad, M.R. Reddy, P. Reddanna,
Homology modeling of 5-lipoxygenase and hints for better inhibitor design,
J. Comput. Aided Mol. Des. 22 (2008) 611e619.
[64] S.A. Gillmor, A. Villaseñor, R. Fletterick, E. Sigal, M.F. Browner, The struc-
ture of mammalian 15-lipoxygenase reveals similarity to the lipases and
the determinants of substrate specificity, Nat. Struct. Biol.
1003e1009.
4 (1997)
[65] Schrödinger, Version 8.0. Schrödinger, LLC, New York, 2005.