In vitro pharmacological evaluation of multitarget agents for thromboxane prostanoid receptor antagonism and COX-2 inhibition

Article abstract of DOI:10.1016/j.phrs.2015.11.012

Purpose Patients with high cardiovascular risk due to ageing and/or comorbidity (diabetes, atherosclerosis) that require effective management of chronic pain may take advantage from new non-steroidal anti-inflammatory drugs (NSAIDs) that at clinical dosag

Full text of DOI:10.1016/j.phrs.2015.11.012

Pharmacological Research 103 (2016) 132–143
Contents lists available at ScienceDirect
Pharmacological Research
journal homepage: www.elsevier.com/locate/yphrs
In vitro pharmacological evaluation of multitarget agents for
thromboxane prostanoid receptor antagonism and COX-2 inhibition
Malvina Hoxhaa,1, Carola Buccellatia,1, Valérie Caprac, Davide Garellab, Clara Cenab,
Barbara Rolandob, Roberta Frutterob, Silvia Carnevalia, Angelo Salaa, G.Enrico Rovatia,∗
Massimo Bertinariab
,
a Dipartimento di Scienze Farmacologiche e Biomolecolari, Università degli Studi di Milano, via Balzaretti 9, 20133 Milano, Italy
b Dipartimento di Scienza e Tecnologia del Farmaco, Università degli Studi di Torino, Via P. Giuria 9, 10125 Torino, Ital
c Dipartimento di Scienze della Salute, Università degli Studi di Milano, via di Rudinì 8, 20142 Milano, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
Purpose: Patients with high cardiovascular risk due to ageing and/or comorbidity (diabetes, atheroscle-
rosis) that require effective management of chronic pain may take advantage from new non-steroidal
anti-inflammatory drugs (NSAIDs) that at clinical dosages may integrate the anti-inflammatory activity
and reduced gastrointestinal side effects of selective cyclooxygenase-2 (COX-2) inhibitor (coxib) with a
cardioprotective component involving antagonism of thromboxane A2 prostanoid (TP) receptor.
Methods: New compounds were obtained modulating the structure of the most potent coxib, lumiracoxib,
to obtain novel multitarget NSAIDs endowed with balanced coxib and TP receptor antagonist properties.
Antagonist activity at TP receptor (pA2) was evaluated for all compounds in human platelets and in an
heterologous expression system by measuring prevention of aggregation and Gq-dependent production
of intracellular inositol phosphate induced by the stable thromboxane A2 (TXA2) agonist U46619. COX-
1 and COX-2 inhibitory activities were assessed in human washed platelets and lympho-monocytes
suspension, respectively. COX selectivity was determined from dose–response curves by calculating a
ratio (COX-2/COX-1) of IC50 values.
Received 30 October 2015
Accepted 15 November 2015
Available online 24 November 2015
Keywords:
Cyclooxygenase
Thromboxane
Inflammation
NSAID
Coxib
Multitarget drugs
Results: The tetrazole derivative 18 and the trifluoromethan sulfonamido-isoster 20 were the more active
antagonists at TP receptor, preventing human platelet aggregation and intracellular signalling, with pA2
values statistically higher from that of lumiracoxib. Comparative data regarding COX-2/COX-1 selectivity
showed that while compounds 18 and 7 were rather potent and selective COX-2 inhibitor, compound 20
was somehow less potent and selective for COX-2.
Conclusion: These results indicate that compounds 18 and 20 are two novel combined TP receptor antag-
onists and COX-2 inhibitors characterized by a fairly balanced COX-2 inhibitor activity and TP receptor
antagonism and that they may represent a first optimization of the original structure to improve their
multitarget activity.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
(COX) inhibition. COX is responsible for prostanoid production from
arachidonic acid (AA) and can be inhibited reversibly by non-aspirin
Non-steroidal anti-inflammatory drugs (NSAIDs) provide anal-
gesic and anti-inflammatory properties by virtue of cyclooxygenase
NSAIDs and irreversibly by aspirin. COX exists in two isoforms, the
housekeeping enzyme COX-1 responsible for the gastric cytopro-
tection and haemostatic integrity, and the inducible isoform COX-2,
mostly expressed in response to inflammatory stimuli and consti-
tutively present in some specific tissue such as endothelial cells,
brain and kidney [1,2].
Severe gastric problems such as bleeding, gastric erosion and
ulcers are the main side effect of chronic use of conventional
NSAIDs and aspirin, mainly due to the inhibition of the COX-
1-derived gastroprotective prostaglandin (PG) E2 production [3].
Celecoxib (Celebrex) and rofecoxib (Vioxx) were the first COX-2
Abbreviations: AA, arachidonic acid; CV, cardiovascular; COX, cyclooxygenase
coxibs; COX-2, selective inhibitors; DMEM, Dulbecco’s modified eagle’s medium;
EIA, enzyme immunoassay; IP, inositol phosphate; HEK293, human embryonic kid-
ney 293; NSAID, nonsteroidal anti-inflammatory drug; PG, prostaglandin; TXA2,
thromboxane A2; TP, thromboxane prostanoid receptor.
∗
Corresponding author.
E-mail address: genrico.rovati@unimi.it (G.Enrico Rovati).
These authors equally contributed to the work.
1
http://dx.doi.org/10.1016/j.phrs.2015.11.012
1043-6618/© 2015 Elsevier Ltd. All rights reserved.

M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
133
selective inhibitors (coxibs) to enter the market as second gener-
ation NSAIDs to be used in symptomatic treatments of patients
with osteoarthritis and rheumatoid arthritis with the promise of
being antiinflammatory while minimizing gastrointestinal (GI) tox-
icity [4]. However, in 2004 Vioxx was withdrawn from the market
and concern over potential cardiovascular (CV) toxicity and risk of
myocardial infarction and stroke associated with the extended use
of coxibs [5,6] and traditional NSAIDs in general [7–9] widespread
rapidly, leading the official medicine agencies to issue CV safety
warnings for coxibs still on the market and successively also for
non-selective NSAIDs.
Early explanation for the potential thrombotic risk included an
imbalance in the biosynthesis of thromboxane A2 (TXA2, a potent
platelet aggregator and vasoconstrictor) and prostacyclin (PGI2,
which has opposing actions) [10] as a result of the observation
that urinary excretion of the principal PGI2 metabolite, 2,3-dinor
6-keto PGF1␣, was reduced in patients treated with celecoxib and
rofecoxib, while TXB2, urinary metabolite of TXA2, was unaltered
[11,12]. Indeed, PGI2 is the major end product of COX-2 in vascular
endothelium and reduced prostacyclin receptor signalling has been
suggested to contribute to the adverse CV outcomes observed with
coxibs [13]. Other explanations have been also proposed to clarify
the effect of coxibs (and conventional NSAIDs) that do not involve
the isoform of COX present on endothelial cells. For example, it was
hypothesized that the hazard could depend upon differences in the
levels of lipid peroxides or in the supply of AA substrate between
platelets and endothelial cells, such that PGI2 synthesis is inhibited
by NSAIDs more readily than platelet-derived TXA2 [14]. Another
report suggests that the CV toxicity of rofecoxib could be due to its
intrinsic physico-chemical properties and primary metabolism that
increase Low Density Lipoproteins and membrane lipids oxidation
thus promoting formation of isoprostanes, a characteristic feature
of atherogenesis [15]. The involvement of isoprostanes is of partic-
ular interest considering that they are nonenzymatic products of
fatty acid oxidation, therefore insensitive to the action of aspirin
and NSAIDs, they are chemically stable and are produced in vivo
in quantities exceeding those of TXA2 and, finally act through the
TXA2 prostanoid (TP) receptor [16].
inhibition, and its derivative lumiracoxib was described: the com-
petitive antagonism at the TP receptor [17]. While it is true that
increase in CV risk has been reported not only for selective coxibs,
but also for conventional NSAIDs [18–20] including diclofenac [21],
its potency as TP receptor antagonist is certainly not sufficient to
have an impact in therapy at the prescribed clinical doses [17]. In
addition, despite it has been withdrawn from the market due to
hepatotoxicity problems (Novartis News (2007) Prexige® receives
“not approvable” letter in the US despite being one of the most stud-
ied COX-2 inhibitors), a recent meta-analysis of eighteen clinical
trials in patients with osteoarthritis taking lumiracoxib concluded
that no significant differences in CV outcomes was evident between
lumiracoxib and placebo or between lumiracoxib and other NSAIDs
[22]. Thus, these findings seems to corroborate our choice to use
lumiracoxib as a starting structure for chemical modification and
to reinforce the hypothesis that cardiotoxicity associated to the
different NSAIDs and coxibs might not depend on COX selectiv-
ity per se, but rather on distinctive characteristics of each single
molecule, including its pharmacokinetic, that might affect differ-
ently the intricate inter-eicosanoid network of biosynthetic and
signaling pathways leading to multiple events that may synergize
or be functionally opposed, as it is the case for platelet function
[23].
In the present study we report the physico-chemical profile and
the full pharmacological characterization of four different com-
pounds 18, 20, 7 and 32 (Fig. 1) endowed with dual COX-2 inhibitor
activity and TP receptor antagonism out of a large series of com-
pounds previously reported [24]. TP antagonism of these newly
synthesized compounds has been evaluated calculating their pA2
for anti-aggregating activity in human platelets as well as for their
activity in inhibiting phospholipase C induced inositol phosphate
production in Human Embryonic Kidney 293 (HEK293) cells tran-
siently transfected with the TP␣ receptor in response to the TXA2
stable analogue U46619. Moreover, we determined their COX-1
and COX-2 activity and selectivity in washed platelets and isolated
human monocytes, respectively. The same was performed with
reference molecules, namely the traditional NSAIDs naproxen, the
coxib lumiracoxib, as well as the potent and selective TP antagonist
terutroban [25].
For these reasons, we consider that the addition of a TP antag-
onist component to a coxib may provide protection against all the
harmful activities mediated through the activation of the TP recep-
tor by mediators sensitive and insensitive to aspirin/NSAIDs such as
the unopposed platelet-derived TXA2 and the nonenzymatic prod-
uct isoprostanes.
2. Material and methods
2.1. Reagents
Recently, an unexpected mechanism of action for diclofenac,
traditional NSAID with a non-selective profile of COX
Animal serum, antibiotics, other supplements, Lipofectamine
2000, Opti-MEM I and molecular biology reagents were purchased
a
Fig. 1. Chemical structures of reference compounds naproxen, lumiracoxib and terutroban, as well as of the newly synthetized compounds 7, 18, 20, and 32.

134
M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
from Invitrogen (Carlsbad, CA). Inositol-free Dulbecco’s modified
Eagle’s medium (DMEM) was obtained from ICN Pharmaceuti-
cals Inc. (Costa Mesa, CA). Ultima Gold was from PerkinElmer
Life and Analytical Sciences (Boston, MA), as were myo-[2-
3H]inositol. U46619 ([1R-[1␣␣␤(Z)␣(1E, 3S*)][1␣,4␣,5␤(Z), 6␣(1E,
3S*)][1␣␣,5␤(Z), 6␣(1E, 3S*)][1␣,4␣,5␤(Z), 6␣(1E, 3S*)][1␣␣␤(Z),
6␣(1E, 3S*)][1␣,4␣,5␤(Z), 6␣(1E, 3S*)][1␣␣,5␤(Z), 6␣(1E,
3S*)][1␣,4␣,5␤(Z), 6␣(1E, 3S*)]]-7-[6-(3-hydroxy-1-octenyl)-
2-oxabicyclo[2.2.1]hept-5-yl]- 5-heptenoic acid), SQ29,548
([1S-[1␣␣(Z)␣,4␣][1␣,2␣(Z), 3␣,4␣][1␣␣(Z), 3␣,4␣][1␣,2␣(Z),
3␣,4␣]]-7-[3-[[2-[(phenylamino) carbonyl]-hydrazino]methyl]-
COX-2 inhibition was evaluated quantifying PGE2 production in
24 h lipopolysaccharide (LPS) challenged preparations pretreated
(30 min, 37 ◦C) with increasing concentration of the tested com-
pound. PGE2 determination was carried out by EIA, according to
the manufacturer’s instructions, or mass spectrometry as described
below.
2.4. COX-1 inhibitory activity (human platelets)
Human platelets were recovered from PRP after centrifugation
at 650 × g for 15 min at room temperature. Their concentration was
adjusted at 2 × 108 cells/mL. Platelets were treated with increasing
concentration of the tested compounds, and incubated at 37 ◦C in
a Dubnoff bath for 30 min. In order to stimulate the TXB2 produc-
tion by platelet degranulation, 2 ␮M calcium ionophore A23187
was added to each test tube sample for 10 min at 37◦ C. Following
centrifugation at 1500 × g for 5 min, TXB2 production was evaluated
in the supernatant by mass spectrometry as described below.
7-oxabicyclo[2.2.1]hept-2-yl]-5-heptenoic
acid),
deuterated
standards ([d4]PGE2 and [d4]TXB2) and PGE2 enzyme
immunoassay- (EIA) kit were from Cayman Chemical (Ann Arbor,
MI). Anion exchange resin AG 1X-8 (formate form, 200–400 mesh)
and Lowry dye-binding protein reagents were from Bio-Rad (Her-
cules, CA). All other reagents of the highest purity were available
from Sigma–Aldrich (St. Louis, MO).
2.2. Isolation of human platelets and analysis of platelet
aggregation
2.5. Mass spectrometry determination of eicosanoids
PGE2 and TXB2 concentrations were evaluated by liquid
chromatography–tandem mass spectrometry using the isotopic
dilution of the deuterated internal standards [d4]PGE2 and
[d4]TXB2. Briefly, samples were spiked with internal standards
and an aliquot injected into a liquid chromatograph Agilent
1100 (Agilent Technologies, Santa Clara, CA). Chromatography
was carried out using a reverse phase column (Synergi 4 ␮m
Hydro-RP, 150 × 2 mm; Phenomenex, Torrance, CA). The column
was eluted with a linear gradient from 25 to 100% solvent B
(Methanol/Acetonitrile, 65/35) over 10 min (Solvent A:0.05% acetic
acid pH 6 with ammonia). The effluent from the high-performance
liquid chromatography (HPLC) column was directly infused into an
API4000 triple quadrupole operated in negative ion mode, mon-
itoring the following specific transitions: m/z 351 > 271 for PGE2,
m/z 355 > 275 for [d4]PGE2, m/z 369 > 169 TXB2 and m/z 373 > 173
for [d4]TXB2. Quantitation was carried out using standard curves
obtained with synthetic standards (Cayman Chemical, Ann Arbor,
MI).
CPD-anticoagulated human blood (Citrate Phosphate Dextrose
solution: sodium citrate, dihydrate, 26.3 g/L; dextrose, mono-
hydrate, 25.5 g/L; citric acid, anhydrous 3.27 g/L; monobasic
sodium phosphate, monohydrate, 2.22 g/L) was sampled follow-
ing informed consent from healthy volunteers of both genders
aged from18 to 60 years that had no history of CV disease.
Blood was treated with 100 ␮M acetylsalicylic acid and 25 mL buffy
coat was centrifuged at 280 × g for 15 min at room temperature
to obtain platelet-rich plasma (PRP), which was further cen-
trifuged at 650 × g for 10 min at room temperature. The pelleted
platelets were suspended in 8 mL washing buffer (mM compo-
sition: citric acid monohydrate 39, glucose monohydrate 5, KCl
5, CaCl2 2, MgCl2 × 6H2O 1, NaCl 103, pH 6.5), recentrifuged at
650 × g for 15 min at room temperature, and finally resuspended in
15 mL of Hank’s Balance Salt Solution (HBSS): CaCl2·2H2O 0.185 g/L;
KCl 0.40 g/L; KH2PO4 0.06 g/L; MgCl2·6H2O 0.10 g/L; MgSO4·7H2O
0.10 g/L; NaCl 8.00 g/L; NaHCO3 0.35 g/L; Na2HPO4 0.048 g/L; d-
glucose 1.00 g/L). The concentration was adjusted at approximately
2 × 108 cell/mL and platelet aggregation was assessed with a
Chrono-Log aggregometer (Mascia Brunelli, Milano, Italy), using
the Born turbidimetric assay at 37 ◦C in a 0.5 mL sample. After
incubation with drug or vehicle dimethyl sulfoxide (DMSO, maxi-
mum 0.2%, v/v) for 5 min at 37 ◦C, platelet aggregation was induced
by U46619 (0.1 ␮M) under continuous stirring and monitored
for 6 min. Experiments were repeated at least in triplicate using
platelets from different subjects. The anti-aggregating activity of
each compound was compared with its corresponding control
aggregation, recorded immediately before and after drug testing
due to the inter-subject variability of the platelet response to the
agonist challenge.
2.6. Culture and transfection of HEK293 cells
HEK293 cells (ATCC, Manassas, VA) were transiently transfected
as previously reported [17,26–30]. Briefly, cells were cultured in
DMEM and supplemented with 10% fetal bovine serum (FBS),
2 mM glutamine, 50 U/mL penicillin, 100 ␮g/mL streptomycin, and
20 mM HEPES buffer, pH 7.4, at 37 ◦C in a humidified atmosphere
of 95% air and 5% CO2. Cells were plated onto 12-well dishes pre-
viously coated with poly-d-lysine, in order to obtain a 50–60%
confluence at the time of transfection. DNA constructs of TP␣ wild
type receptor were previously obtained in our laboratory [26].
Ultrapure plasmids for cell transfection were obtained using the
QIAfilter Plasmid Kits by Qiagen (Hilden, Germany). Lipofectamine
2000 (Invitrogen, Carlsbad, CA) was used as a transfectant agent
and the transfection mix Lipofectamine 2000/DNA was prepared
in Opti-MEM I Medium, which was optimized at 2:1 ratio and later
added to the cells.
2.3. COX-2 inhibitory activity (lympho-monocytes)
The study of COX-2 activity was carried out in a lympho-
monocytes suspension, in order to avoid eventual compound
binding to plasma-protein. Lympho-monocytes were isolated from
buffy coat (diluted in NaCl 0.9% 1:1) Ficoll–Paque gradient density
centrifugation (400 × g for 30 min at 10 ◦C); enriched cell ring was
collected and twice saline washing (280 × g for 15 min at 10 ◦C)
performed to remove the remaining suspended platelets. Soon
after, a lysis buffer (NaCl 0.2% weight/volume, w/v) was added to
remove the remaining erythrocytes, immediately balanced with
an equal volume of equilibrating solution (NaCl 1.6% + saccarose
0.2%, w/v). Lympho-monocytes were then resuspended in HBSS and
2.7. Total inositol phosphate determination in HEK293 cells
The functional activity of the TP␣ receptor was assessed 48 h
after transfection by measuring the accumulation of total labelled
IPs through a column ion exchange chromatography as previously
described [26]. Briefly, the day before the assay, transfected HEK293
cells were labelled with 0.5–1 ␮Ci/mL of myo-[2-3H]inositol for
24 h in DMEM, free of inositol and serum, containing 20 mM HEPES

M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
135
Table 1
Topological polar surface area (tPSA), solubility in simulated gastric fluid (SGF) and phosphate buffered solution (PBS), dissociation constants and lipophilicity descriptors of
compounds under study.
Compound
tPSAa
(Å2)
Solubility
SGF
(␮M) SD
pKa S.D.b
clog Pc
log D 7.4 S.D.d
PBS
Lumiracoxib
7
18
20
32
49.3
49.3
61.1
58.2
83.5
23.4 0.7
8.7 0.1
54.1 5.4
13.9 1.6
19.3 4.9
442.1 21.4
4.15 0.03e
4.31 0.05
4.85 0.01e
6.70 0.01e
4.33 0.01
4.66
5.84
4.85
6.16
4.54
1.19 0.05
1.86 0.03
1.60 0.03
>3.5
440.2 3.9
533.9 16.2
60.7 2.3
553.1 15.0
0.87 0.04
a
Topological polar surface area values calculated using ChemBioDraw v.12 CambridgeSoft.
Data obtained by potentiometric titration (SIRIUS GlpKa).
Values calculated by using Bio-Loom for Windows v.1.5 BioByte Corp.
Data obtained by shake-flask method.
b
c
d
e
According to [24].
buffer pH 7.4, 0.5% (w/v) Albumax, 2 mM glutamine, and 100 U/mL
2.10. Solubility assessment in phosphate buffered saline (PBS)
and simulated gastric fluid (SGF)
penicillin streptomycine. On the day of the assay, cells were incu-
bated with 25 mM LiCl for 10 min. Soon after they were pretreated
with antagonists (diclofenac, lumiracoxib, and tested compounds)
for 30 min at 37 ◦C, and stimulated with either vehicle or 0.1 ␮M
U46619. The reaction was stopped by lysing the cells with 10 mM
formic acid for 30 min at 4 ◦C, and acidic phase transferred in 5 mM
NH4OH pH 8–9. Finally, total IPs were extracted by chromatography
AG 1X-8 columns formate form, 200–400 mesh size.
The solubility of compounds 7, 18, 20, 32 and lumiracoxib was
studied in simulated gastric fluid (SGF-without pepsin) and phos-
phate buffered saline (PBS) 0.05 M to evaluate the solubility of
compounds in acid (pH 1.5 for SGF) and neutral conditions (pH
7.4 for PBS) to simulate the gastric and the body fluid environment
respectively (Table 1) [32].
Stock solutions of compounds lumiracoxib, 7, 18, 20 and 32
(10 mM) were prepared in DMSO. Eight point calibration standards
(1, 5, 10, 20, 50, 100, 200 and 500 ␮M) were prepared from each
10 mM solution stock by dilution in HPLC mobile phase and ana-
lyzed by HPLC. 100 ␮L of each test compound (stock solution 10 mM
in DMSO) were added to 1900 ␮L of PBS and SGF in glass tubes in
triplicate and shaken for 90 min at 100 rpm at room temperature.
The samples were filtered using 0.45 ␮L PTFE filters and analysed
by HPLC.
HPLC analysis were performed with a HP 1100 chromato-
graph system (Agilent Technologies, Palo Alto, CA, USA) equipped
with a quaternary pump (model G1311A), a membrane degasser
(G1379A), a diode-array detector (DAD) (model G1315B) integrated
in the HP1100 system. Data analysis were processed using a HP
ChemStation system (Agilent Technologies). The analytical column
was a Tracer Excel C18 (250 × 4.6 mm, 5 ␮m; Teknokroma). The
mobile phase consisting of acetonitrile/HCOOH 0.1% 70/30 (v/v)
at flow-rate = 1.0 mL/min. The injection volume was 20 mL (Rheo-
dyne, Cotati, CA). The column effluent was monitored at 254 nm
and 280 nm referenced against a 800 nm wavelength. Quantitation
of compounds was done using calibration curves of compounds;
the linearity of the calibration curves was determined in a concen-
tration range of 1–500 ␮M (r2 > 0.99).
2.8. Determination of dissociation constants
The ionization constants of compounds were determined by
potentiometric titration with the GLpKa apparatus (Sirius Analyt-
ical Instruments Ltd., Forest Row, East Sussex, UK) as previously
described [24]. Brefly, because of the low aqueous solubility, com-
pounds required titrations in the presence of MeOH as co-solvent:
at least five different hydro-organic solutions (ionic strength
adjusted to 0.15 m with KCl) of the compounds (20 mL, ca. 0.5 mM
in 20–60 w% MeOH) were initially acidified to pH 1.8 with 0.5N HCl;
the solutions were then titrated with standardized 0.5 N KOH to pH
12.2 at 25 ◦C under N2.
The apparent ionization constants in the H2O–MeOH mixtures
(psK␣) were obtained and aqueous pKa values were calculated
by extrapolation to zero content of the co-solvent, following the
Yasuda–Shedlovsky procedure [31].
2.9. Determination of lipophilicity descriptors
Calculated partition coefficients of compounds in neutral form
(clog P) were obtained by using Bio-Loom for Windows v.1.5
(BioByte Corp. Claremont, CA, USA).
The distribution coefficient at pH 7.4 (log D7.4) of the com-
pounds between n-octanol and water was experimentally obtained
by shake-flask technique at room temperature. In the shake-flask
experiments phosphate 50 mM buffer with ionic strength adjusted
to 0.15 M with KCl, was used as aqueous phase; the organic (n-
octanol) and aqueous phase were mutually saturated by shaking for
4 h. The compounds were solubilised in the buffered aqueous phase
at a concentration of about 0.1 mM and an appropriate amount
of n-octanol was added. The two phases were shaken for about
20 min, by which time the partitioning equilibrium of solutes is
reached, and then centrifuged (1100 × g, 10 min). The concentra-
tion of the solutes was measured in the aqueous phase by HPLC.
Each log D value is an average of at least six measurements. All the
experiments were performed avoiding exposure to light.
2.11. Stability in human serum
A solution of each compound (10 mM) in DMSO was added
to human serum (from human male AB plasma, USA origin,
sterile-filtered, Sigma–Aldrich) preheated at 37 ◦C; the final con-
centration of the compound was 100 ␮M. Resulting solution were
incubated at 37 0.5 ◦C and at appropriate time intervals (2, 6,
24 h) 300 ␮L of the reaction mixture was withdrawn and added
to 300 ␮L of acetonitrile containing 0.1% trifluoroacetic acid in
order to deproteinize the proteins. Sample was sonicated, vor-
texed and then centrifuged for 10 min at 2150 × g. The clear
supernatant was filtered by 0.45 ␮m PTFE filters (Alltech) and ana-
lyzed by RP-HPLC with method previously described for solubility
assessment.

136
M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
Table 2
TP receptor antagonism at human washed platelet aggregation and total IP production in transfected HEK 293 cells. pA2 values were determined by measuring inhibition of
aggregation response to the stable agonist U46619.
Compound
pA2 %CV
Human washed platelet aggregation
Total IP production in HEK293 cells
Lumiracoxib
Diclofenac
Naproxen
Terutroban
18
20
7
5.0 2.5
5.4 4.9
4.1 2.5
9.4 4.1
5.6 3.5
5.9a 4.1
5.0 1.8
4.8 2.4
4.6 5.2
5.3 4.7
3.9 16.3
9.3 2.8
5.5a 2.2
5.7a 2.3
4.9 5.1
4.8 2.5
32
a95% CI vs lumiracoxib, see results.
2.12. Statistical analysis
3. Results
pA2 were calculated accordingly to the following set of equa-
tions as described in Prism 5 (GraphPad Software Inc., San Diego,
CA):
3.1. Chemistry
For the synthesis of compounds 7 (2-[(2-chloro-6-fluorophenyl)
amino]-5-methyl-benzoic acid) and 32 (2-(((4-chlorophenyl) sul-
fonyl) amino)-5-methyl-benzoic acid) the procedure reported
ˆ
(1)Agonistresponse = Bottom + (Top–Bottom)/(1 + 10((Log EC50
-
X)*HillSlope))
(2) Antagonist response = Bottom + (Top–Bottom)/(1 + (Antag/
in Scheme 1 (see Supplementary materials) was used. The
ˆ
FixedAg) HillSlope)
synthesis of lumiracoxib analogue 7 was accomplished as pre-
viously described [24]. Briefly, the Chan–Lam coupling was
used by reacting 2-amino-5-methylbenzoic acid with 2-chloro-
6-fluorophenylboronic acid in the presence of 1,8-diazabicylo-
ˆ
ˆ
ˆ
ˆ
(3) Antag = (10Log EC50)*(1 + ((10X)/(10(−1*pA2))) SchildSlope)
where X is the Log concentration of the agonist, Bottom is the
response when X = 0, Top is the response for an infinite concentra-
tion of X, EC50 is the concentration of the agonist that produce half
of the response, Hill Slope is the slopes of the curves, Fixed Ag is
the initial fixed concentration of the agonist used in the determina-
tion of the antagonist inhibition curve. The concentration-response
curves of platelet aggregation were analysed by Prism-5 software
utilizing the four-parameter logistic model as described in the
ALLFIT program [33]. Parameter errors are all expressed in percent-
age coefficient of variation (% CV) and calculated by simultaneous
analysing at least three different independent experiments per-
formed in duplicate or triplicate. P < 0.05 was set as the statistical
level of significance. All curves shown are computer generated.
[5,4,0]undec-7-ene (DBU) and
a stoichiometric amount of
copper acetate in dioxane solution. Compound 32 was syn-
thesized by reacting 2-amino-5-methylbenzoic acid with 4-
chlorobenzensulfonyl chloride in the presence of excess Na2CO3
in water at 60–80 ◦C. The product was isolated and recrystal-
lized from ethanol. Compounds 18 (N-(2-chloro-6-fluorophenyl)-
4-methyl-2-(1H-tetrazol-5-ylmethyl)-benzenamine) and 20 (N-
[[2-[(2-chloro-6-fluorophenyl) amino]-5-methylphenyl]methyl]-
1,1,1-trifluoromethanesulfonamide) were obtained as previously
reported [24] (Fig. 1).
Fig. 2. Representative original traces of acetylsalicylic acid-treated human platelets demonstrating the effect of compounds 18 and 20 on the TP-dependent aggregation.
Washed platelets were challenged with 0.1 ␮M U46619 (arrow) in the presence of vehicle (0.2% DMSO) or the indicated concentrations of compounds 18 (a) and 20 (b).

M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
137
a)
b)
c)
U46619
Naproxen
U46619
Lumiracoxib
U46619
Cp 18
100
50
0
100
50
0
100
50
0
-9
-8
-7
-6
-5
-4
-3
-9
-8
-7
-6
-5
-4
-9
-8
-7
-6
-5
-4
Log, [M]
Log, [M]
Log, [M]
d)
e)
f)
U46619
Cp 20
U46619
Cp 7
U46619
Cp 32
100
50
0
100
50
0
100
50
0
-9
-8
-7
-6
-5
-4
-9
-8
-7
-6
-5
-4
-9
-8
-7
-6
-5
-4
Log, [M]
Log, [M]
Log, [M]
Fig. 3. Antagonism of human platelet aggregation induced by U46619 by the indicated compounds. Concentration–response curves of U46619-induced washed platelet
aggregation from human blood and inhibition curves of: naproxen (a), lumiracoxib (b), compound 18 (c) compound 20 (d), compound 7 (e) and compound 32 (f). EC50’s were
calculated using a four parameters logistic model, while pA2’s were calculated accordingly to the set of Eqs. (1)–(3) as described in materials and methods. Values shown
represent platelet aggregation (mean SE) expressed as % maximal aggregation induced by 0.1 ␮M U46619. Blood was treated with 100 ␮M acetylsalicylic acid. Experiments
have been performed at least three times in duplicates. All curves shown were computer generated.
3.2. Physico-chemical characterization of compounds
32 also show a similar solubility with respect to lumiracoxib at
physiological pH, their solubility is slightly lowered when the pH is
brought down to 1.5. Compound 20 is the least soluble derivative
of the series at pH 7.4, nevertheless it retains a fair solubility in SGF
medium (0.59 fold that of lumiracoxib).
The pKa values of 7, 18 and 32 are slightly higher than that of
lumiracoxib. Derivative 20 presents a significantly lower acidity
with respect to that of the reference drug, this property reflects
in a distribution coefficient measured at pH 7.4 greater than 3.5,
consequently 20 is more lipophilic than lumiracoxib in physiolog-
ical condition. Compounds 7 and 18 are 1.56 and 1.34 fold more
lipophilic than lumiracoxib and this may favor cellular permeabil-
ity. All the compounds were also shown to be stable in human
serum with >98% unchanged form detected after 24 h incubation
(data not shown). From this preliminary characterization we could
assume that the tetrazole derivative 18 shows the best drug-like
properties among the compounds of this series.
The compounds selected for this study present most of the struc-
tural characteristics predicted for lumiracoxib analogs suitable as
COX-2 inhibitors from a recent in silico study. [34]. Their structures
contain: (1) two aromatic rings; (2) at least two oxygen atoms (with
the exception of compound 18); (3) at least one carboxyl group or
a carboxyl isosteric group; (4) at least one −OH or −NHn group.
Inspection of data reported in Table 1 evidences that the topolog-
ical polar surface area (tPSA) of the studied compounds is similar
or higher than that of lumiracoxib; in particular, the value of tPSA
for the tetrazole analog 18 (61.1) is very close to the mean value
found for a set of 36 NSAIDs (63.2) and of lumiracoxib analogs (70.5)
[34,35]. The measured solubility indicates that 18 is more soluble
than lumiracoxib both in simulated gastric fluid and in phosphate
buffered saline. Its solubility double that of the reference drug when
the simulated gastric fluid is considered as the medium. This may
be an advantage following oral administration. Derivatives 7 and
Table 3
COX-2 and COX-1 inhibitory activities determined by in vitro assay in lympho-monocytes and washed human platelets.
Compound
COX-2 inhibitionIC50 (␮M) % CV
COX-1 inhibitionIC50 (␮M) % CV
COX-2/COX-1selectivity
Lumiracoxib
Diclofenac
Naproxen
Terutroban
18
20
7
0.0035 26
0.0011 30
0.19 66
inactive at 10 ␮M
0.014 23
0.42 32
3.22 22
0.0083 6.2
0.11 10
N.D.
910
7.6
0.58
N.D.
942
38
13.2 22
16.1
6
0.025 46
1.20 45
25.5 10
inactive at 60 ␮M
1020
N.D.
32
N.D.: not determined.

138
M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
a)
b)
c)
10
8
6
4
2
0
8
6
4
2
0
U46619
Naproxen
U46619
U46619
Cp 18
Lumiracoxib
6
4
2
0
-12
-10
-8
-6
-4
-2
-12
-10
-8
-6
-4
-2
-12
-10
-8
-6
-4
-2
Log, [M]
Log, [M]
Log, [M]
e)
d)
f)
6
4
2
0
6
4
2
0
6
4
2
0
U46619
Cp 32
U46619
Cp 7
U46619
Cp 20
-12
-10
-8
Log, [M]
-6
-4
-2
-12
-10
-8
-6
-4
-2
-12
-10
-8
Log, [M]
-6
-4
-2
Log, [M]
Fig. 4. Antagonism by the indicated compounds of total IP production induced by the TXA2 analog U46619 in HEK cells transiently transfected with the alpha isoform
of the human TP receptor. Concentration–response curves of U46619-induced total IP production and inhibition curves of naproxen (a), lumiracoxib (b), compound 18 (c),
compound 20 (d), compound 7 (e), and compound 32 (f), obtained in the presence of 0.1 ␮M U46619. EC50’s were calculated using a four parameters logistic model, while pA2’s
were calculated accordingly to the set of Eqs. (1)–(3) as described in materials and methods. Values shown represent the mean percentage of fold increase over basal SE.
Experiments were performed at least three times in duplicates. All curves shown were computer generated.
3.3. Inhibition of TP receptor functional activity in human
platelets
95% CI-confidence interval 5.4–6.4 for compound 20 −pA2 = 5.0,
95% CI-confidence interval 4.7–5.2 for lumiracoxib).
A series of newly synthetized compounds, i.e., compound 18,
20, 7, and 32, as well as a non-selective COX inhibitor (naproxen)
and a potent and selective COX-2 inhibitor (lumiracoxib) (Fig. 1)
were studied for TP receptor antagonism in platelets from healthy
human volunteers. The extent of aggregation was detected by Born-
turbidimetric assay. Blood was collected in the presence of 100 ␮M
acetylsalicylic acid to render the platelets unresponsive to the chal-
lenge with arachidonic acid (1–3 ␮M), but fully responsive to the
calcium ionophore A-23187 (3 ␮M; data not shown). Representa-
tive traces of washed platelet aggregation obtained with 0.1 ␮M of
the stable TXA2 analogue U46619 in the presence of increasing con-
centrations (0.3–20 ␮M) of compounds 18 and 20 are portrayed in
Fig. 2. When platelets were challenged with increasing concentra-
tions of U46619, a concentration-dependent platelet aggregation
occurred, which revealed a potency value of 59 nM 19% CV (Fig. 3),
in perfect agreement with previous results [17]. This response was
thus truly independent of endogenous TXA2 formation. The sen-
sitivity of platelets to U46619 did not change during the time
required for the experiment and none of the tested compounds
caused any aggregator response by itself. Fig. 3 also shows the inhi-
bition curves of U46619-induced (0.1 ␮M) platelet aggregation due
to increasing concentrations of the newly synthetized compounds
as well as the reference compounds. Table 2 reports their respective
pA2 values calculated accordingly to Eqs. (1)–(3) (see methods sec-
tion). Among the different molecules tested, compounds 18 and 20
were found to be the most potent in terms of TXA2 antagonism with
pA2 values comparable to that of diclofenac, but, at least for com-
pound 20, statistically different from that of lumiracoxib (pA2 = 5.9,
3.4. Inhibition of TP˛ functional activity in HEK293
All compounds were also tested for their ability to inhibit
the total inositol phosphate (IP) production following classic TP
receptor coupling with Gq. The human TP␣ receptor transiently
expressed in HEK293 cells was activated by the stable TXA2
analogue U46619 (0.1 ␮M, 30 min) in the absence and presence
of 30 min pretreatment with increasing concentrations of the
reported antagonists (Fig. 4). TP receptor activation by U46619
resulted in a robust increase in total IP production with a calcu-
lated EC50 of 29.3 nM 10% CV, as previously reported [28–30].
No response has been obtained from mock-transfected cells (Fig.
1S). The pA2 values calculated for each compound are presented
in Table 2. It is worth notice here that the results obtained in
total IP production inhibition are in full agreement with those
obtained in aggregation studies. Compounds 18 and 20 were again
the most potent molecules, with pA2 values similar to that of
diclofenac and both statistically different from that of lumiracoxib
(pA2 = 5.5, 95% CI, 5.2–5.8 for compound 18 −pA2 = 5.7, 95% CI,
5.4–6.0 for compound 20 −pA2 = 4.6 95% CI, 4.1–5.1 for lumira-
coxib), in good agreement with affinity binding data previously
obtained in HEK293 cells [24].
3.5. COX-2/COX-1 selectivity
Newly synthesized compounds should maintain the COX-2
selectivity of the parent lumiracoxib and therefore their capac-
ity to act as COX-2 inhibitors was determined on isolated human

M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
139
a)
b)
TXB2 Naproxen
PGE2 Naproxen
100
90
80
70
60
50
40
30
20
10
0
TXB2 Cp 18
PGE2 Cp 18
100
90
80
70
60
50
TXB2
Lumiracoxib
PGE2
Lumiracoxib
40
30
20
10
0
-12 -11 -10 -9
-8
-7
-6
-5
-4
-11 -10 -9
-8
-7
-6
-5
-4
Log, [M]
Log, [M]
d)
c)
TXB2 Cp 7
PGE2 Cp 7
TXB2 Cp 20
100
100
90
80
70
60
50
40
30
20
10
0
90
80
70
60
50
40
30
20
10
0
PGE2 Cp 20
-11 -10 -9
-8
-7
-6
-5
-4
-11 -10 -9
-8
-7
-6
-5
-4
Log, [M]
Log, [M]
Fig. 5. Inhibition of COX-1 and COX-2 activity by the indicated compounds in comparison to the reference compounds naproxen and lumiracoxib (a). COX-1 activity was
assessed in terms of inhibition of TXB2 production induced by calcium ionophore in human washed platelets; COX-2 activity was assessed in terms of inhibition of PGE2
production induced by LPS in isolated human monocytes. Data are expressed as percent inhibition of TXB2 or PGE2 release versus untreated controls. Error bars represent
mean SE of at least three independent experiments, each performed in duplicate.
lympho-monocytes following treatment with acetylsalicylic acid.
COX-2 expression was stimulated overnight with 10 ␮g/mL of LPS,
and the PGE2 produced was determined by enzyme immunoas-
say and mass spectrometry, whereas COX-1 inhibitory activity was
determined in washed human platelets as TXB2 production by mass
spectrometry.
the very potent TP receptor antagonist terutroban [25], being the
less potent molecule of the series (Table 3). Among the other
molecules, the tetrazole derivative 18 and compound 7 inhibited
COX-2 in a very similar way with potencies of 0.014 and 0.025 ␮M,
respectively, while compound 20 shows a potency similar to that
of naproxen (Fig. 5 and Table 3).
All the compounds tested inhibit the COX-2 enzyme in a
concentration-dependent manner with lumiracoxib and diclofenac
displaying the highest absolute potency (Table 3), while compound
32, containing the 4-chlorobenzensulfonamide moiety present in
Diclofenac displayed by far the highest potency as COX-1
inhibitor in washed platelets, with values in the nM range, followed
by naproxen (Table 3). Once again the tetrazole derivative 18 and
compound 7 behaved very similarly with IC50 values of 13.2 ␮M

140
M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
GI complications but lower risk of CV events as an alternative to
-4
-5
conventional NSAIDs associated to gastroprotective therapy. How-
ever, just because taking a coxib, these patients have an increase
in CV risk, and therefore must undergo wash-out periods from the
drug, and even if treated with low-dose aspirin they are not pro-
tected from non-enzymatic AA products such as isoprostanes. On
the other hand, there is still an unmet need for an adequate pain
therapy combined with minimal GI damage and CV toxicity. Any-
one who is at risk for or who is experiencing a vascular thrombotic
diseases (diabetes, venous thromboembolism, etc.), may still suffer
from a further increase in vascular risk even treated with a tra-
ditional NSAIDs [9], with the possible exception of naproxen [8].
In addition, traditional NSAIDs have also been observed to inter-
fere with the cardioprotective effect of low-dose aspirin, somehow
endangering the benefit of their association [37].
It is worth mentioning here that a CV safer coxib may also be
a valuable drug in chemoprevention, considering that all the colon
adenoma prevention trials yielded encouraging results on cancer
prevention, in spite of the divergent incidence of CV events in their
long term use [5,6,38]. Indeed, cardiac toxicity during colon chemo-
prevention trials was the main reason for withdrawing rofecoxib
from the market, despite the threshold of the risk/benefit ratio
should be particularly low in this particular case, considering the
risk of developing a particularly aggressive form of cancer. More
to this, recent finding have highlighted a role for TXA2 synthesis in
colon tumorigenesis, corroborating the use of a molecule endowed
with TP antagonist properties in selected form of cancer [39]. Thus,
a dual COX-2 inhibitor and a TP receptor antagonist might be a
valuable solution to both the problem of long-term pain control
and cancer colon prevention.
-6
Compound 32
Compound 20
Naproxen
-7
Compound 7
Compound 18
-8
Lumiracoxib
Diclofenac
-9
-10
-10 -9 -8 -7 -6 -5 -4 -3 -2
COX-1, IC50 ,log [M]
Fig. 6. COX-2/COX-1 selectivity. IC50 values obtained for the various compounds
are plotted to appreciate their selectivity: the diagonal line indicates equivalence,
therefore compounds with high selectivity for COX-2 over COX-1 are plotted below
the line.
and 25.5 ␮M, respectively, while this time compound 20 showed
an IC50 of 16.1 ␮M, close to that of 18 and 7; compound 32, resulted
inactive in terms of COX-1 inhibition (Fig. 5 and Table 3).
Naproxen and lumiracoxib were expected to be, respectively,
the less and the most selective COX-2 inhibitors (Fig. 5 and Table 3)
in perfect agreement with previously published data summarizing
the selectivity profiles of various NSAIDs, including coxibs (Fig. 6)
[4]. Furthermore, Fig. 5 and Table 3 clearly indicate that compounds
18 and 7 are the most selective coxibs among the multitarget
molecules synthetized of notice; the sulfonamide derivative 20,
despite being only about 40 times more selective for COX-2 with
respect to COX-1 inhibition at calculated IC50, presented a very
steep slope of the concentration-response curve (Hill coefficient
»1). This characteristic has no direct impact on the Therapeutic
Index of the molecule, and, thus, on its safety, but rather suggest
that it is possible to find a dose of compound 20 that is selective for
COX-2. Thus, at concentrations ten times its IC50 for COX-2 inhibi-
tion, about 95% of COX-2 is inhibited, while no inhibition of COX-1
are observed (Fig. 5c).
Indeed,
a number of TXA2 biosynthesis inhibitors and/or
receptor antagonists have been developed and studied providing
evidence of their efficacy on top of aspirin [40,41], even though
they have always suffered from being perceived as too similar to
the economical and efficacious aspirin. In particular, the potent
and selective TP antagonist terutroban [42] has been demonstrated
to be an effective antithrombotic agent in peripheral arterial dis-
ease [43] and to improve endothelial function in atherosclerotic
patients [44]. Somehow unexpectedly, the PERFORM Phase III clin-
ical trial with terutroban in patients with cerebral ischemic events
did not meet the predefined criteria for non-inferiority to aspirin,
while showing similar rates of primary endpoint between the two
drugs [45]. Of notice, however, patients with a history of ischemic
stroke before the qualifying event had a statistically significant
lower recurrence rate with terutroban than with aspirin. Consider-
ing that some of these patients might have been receiving aspirin
before their qualifying event, these data seems to suggest an aspirin
treatment failure and support the use of a TP receptor antagonist as
an alternative strategy to pursue in selected subgroups of patients,
particularly those unresponsive to aspirin, such as diabetics [46].
Furthermore, it has been demonstrated that dimerization of
the prostacyclin receptor (IP) with TP␣ promotes a strong TP-
induced cAMP generation in a mechanism of control over TP
pro-aggregating activity [47]. However, when the naturally occur-
ring IP receptor mutant R212C is co-expressed with TP␣, TP
response is normalized [48] and patients carrying this polymor-
phism seem to develop more intimal hyperplasia than control
individuals [49]. Therefore, we are confident that our approach to
solve the CV issue of coxibs by designing a multitarget drug [23] is
an effective way to retain inhibition of the COX-2 dependent syn-
thesis of the inflammatory PG and extraplatelets TXA2 combining
the block of all TP agonists, both aspirin-sensitive and aspirin-
insensitive (isoprostanes), thus minimizing risks to the individuals.
The data presented here have shown that compounds 7 and 32
did not possess better TP antagonist properties than the reference
compound lumiracoxib, while losing most of their potency as COX
4. Discussion
In this contribution we report a full in vitro pharmacological
characterization of selected molecules from a series of compounds
recently synthetized [24] and two additional molecules with dual
COX-2 inhibitor activity and TP receptor antagonism. As ref-
erence compounds we have considered the non selective first
generation NSAID naproxen, the most selective COX-2 inhibitor
lumiracoxib and the TP antagonist terutroban. This work directs
to a new generation of multitarget compounds among which two
of the lumiracoxib derivatives, the tetrazole compound 18 and the
trifluoromethansulfonamido-isoster compound 20, are the most
active and with a fairly balanced COX-2 inhibitor activity and
TP receptor antagonism. However, TP receptor inhibitory potency
remain in the micromoloar range, still far from optimal values to be
considered around the nanomolar range, preventing, at this stage,
a real in vivo relevance for these molecules. However, these data
will be used as the base to adopt a new and different strategy to
design more potent molecules that might change the way we treat
chronic inflammatory diseases.
Today, there is no safe NSAIDs, their main side effects including
GI and renal toxicity [36]. Thus, coxib therapy is currently used
in chronic pain control in selected patients with higher risks of

M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
141
inhibitors albeit preserving COX-2 selectivity. The replacement of
5. Conclusions
2-fluoro-6-chloro-phenyl substituent present on the amino nitro-
gen in compound 7 with the p-chlorophenylsulfonyl substructure,
typical of terutroban, generated compound 32 where two phenyl
rings are linked by a sulfonamide group instead of the classic amino
group. However, this modulation did not accomplish our expecta-
tions, as compound 32 did not show better properties with respect
to compounds 18, 20 and 7. These data suggest that COX-2 selectiv-
ity might be determined by the presence of the 2-amino-5-methyl
benzen carboxylic acid moiety, while potency could be regulated
by the nature of the substituent present on the amino nitrogen. The
first observation is in agreement with literature data highlighting
the importance of methyl substitution for COX-2 selectivity [50].
In turn, compounds 18 and 20 have gained in TP receptor
antagonist activity, with compound 20 statistically different from
lumiracoxib in both inhibition of human platelets aggregation and
IP generation in a transfected system, despite, as mentioned before,
still far from required values for an in vivo application. These
functional results are in good agreement with affinity binding
data previously reported for these compounds [24], and support
the strategy of isosteric replacement of the carboxylic function
of lumiracoxib, either with linear or cyclic substructures for the
efficacious generation of molecules endowed with improved TP
antagonism. Compound 18 and particularly compound 20 dis-
played better balanced activities diverging less from the ideal value
of 1 when considering the ratio between the calculated poten-
cies at both pharmacological targets (pA2 TP/IC50 COX-2 = 179 and
3, respectively) with respect to lumiracoxib or diclofenac (pA2
TP/IC50 COX-2 = 2269 and 3619, respectively). However, compound
20 has lost most of its COX-2 inhibitory activity and COX selectiv-
ity comparing to compound 18 suggesting that a higher acidity of
carboxylic acid isoster is needed to maintain high level of COX-2
inhibition. Concerning COX-2 activity, it is interesting to note that
shortening of the alkyl chain connecting the acidic moiety to the
benzene ring reflects in a loss of activity (compare derivative 7
to lumiracoxib). This observation clearly indicates that the acidic
moiety should be separated from the benzene ring by one carbon
atom to reach optimal binding to COX-2 pocket, in agreement with
the postulated binding pose of diclofenac and lumiracoxib at the
COX-2 binding site [51]. Collectively our data indicate that both
pKa values and correct positioning of the acidic group are impor-
tant to maintain a high degree of COX-2 inhibition in this class of
compounds. Interestingly the distance of the acidic unit from the
benzene ring might also influence the antagonism at the TP recep-
tor (compare derivatives 18 and 20 vs 7 and 32 in Table 2). The
stringent structural requirements for COX-2 inhibition are appar-
ently in contrast with the need of compounds endowed with more
flexible and extended structures, which proved more efficient as TP
receptor antagonists. In this context, the use of different carboxylic
acid bioisosters, together with docking studies on both COX-2 and
TP receptor models could be envisaged for the development of new
molecules [52–55].
We show here that structure-activity analysis allowed us to
obtain prototypes of new chemical entities endowed with higher
TP antagonist potencies and a more balanced COX-2 selectivity.
The modification of existing drugs targeted against COX-2 (lumira-
coxib) can lead to new bivalent molecules with obvious advantage
in the pharmacological properties. The main goal of a multitarget
drug is to have both activities at the same concentration range with
a consistent pharmacodynamic and single pharmacokinetic profile
with respect to the co-administration of two different molecules.
Further studies will be certainly necessary to improve the pharma-
codinamic profile of these molecules before a careful evaluation of
our hypothesis could take place in an in vivo animal model.
We believe that it is more and more evident that treatment of
inflammation by a generalized inhibition of eicosanoid synthesis is
a strategy far to be optimal. These new compounds will be critical
for the innovation of the pharmacological treatment of patients that
require long-term therapy with anti-inflammatory drugs and are
at high CV risk for the primary importance of the TP receptor in
the CV setting (platelets aggregation, endothelial dysfunction and
plaque stabilization) [57]. This ‘third generation’ NSAID could be
safely used also in other chronic diseases such as Alzheimer [58,59]
or even in selected form of colon cancer where NSAIDs have already
been proven to have a clinical efficacy.
Acknowledgements
We gratefully thank Dr. Alain Rupin (Institut de Recherches
Servier, France) for generously providing us with terutroban. This
work was supported in part by a grant from Regione Lombardia
(SAL-02) and Fondazione Banca del Monte di Lombardia (GER, VC)
and by University of Turin (Ricerca locale 2013 to MB).
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.phrs.2015.11.
012.
References
[1] W.L. Smith, D.L. DeWitt, R.M. Garavito, Cyclooxygenases: structural, cellular,
and molecular biology, Annu. Rev. Biochem. 69 (2000) 145–182.
[2] Morita I. Distinct functions of COX-1 and COX-2. Prostaglandins Other Lipid
Mediat. 2002, 68–69, 165–175.
[3] L. Laine, K. Takeuchi, A. Tarnawski, Gastric mucosal defense and
cytoprotection: bench to bedside, Gastroenterology 135 (1) (2008) 41–60.
[4] G.A. FitzGerald, C. Patrono, The coxibs, selective inhibitors of
cyclooxygenase-2, N. Engl. J. Med. 345 (6) (2001) 433–442.
[5] R.S. Bresalier, R.S. Sandler, H. Quan, J.A. Bolognese, B. Oxenius, K. Horgan, C.
Lines, R. Riddell, D. Morton, A. Lanas, M.A. Konstam, J.A. Baron, Adenomatous
polyp prevention on vioxx trial I. Cardiovascular events associated with
rofecoxib in a colorectal adenoma chemoprevention trial, N. Engl. J. Med. 352
(11) (2005) 1092–1102.
[6] S.D. Solomon, J.J. McMurray, M.A. Pfeffer, J. Wittes, R. Fowler, P. Finn, W.F.
Anderson, A. Zauber, E. Hawk, M. Bertagnolli, Adenoma prevention with
celecoxib study I. Cardiovascular risk associated with celecoxib in a clinical
trial for colorectal adenoma prevention, N. Engl. J. Med. 352 (11) (2005)
1071–1080.
[7] J. Hippisley-Cox, C. Coupland, Risk of myocardial infarction in patients taking
cyclo-oxygenase-2 inhibitors or conventional non-steroidal
anti-inflammatory drugs: population based nested case-control analysis, BMJ
330 (7504) (2005) 1366.
Naproxen is a non-selective COX inhibitor (Fig. 6) and it appears
to have a better CV risk profile than other NSAIDs [8,19,20] despite
it is a very weak TP receptor antagonist. Interestingly, naproxen
has been demonstrated to inhibit not only platelet-derived TXA2
but also systemic PGI2 biosynthesis [56], thus, not disrupting the
balance between the opposing actions of TXA2 and PGI2. This bal-
ance, part of the intricate cross-talk of lipid mediators generated
from AA seems to be a fundamental feature when designing an anti-
inflammatory drug with a safe CV profile, and should be the target
of any therapeutical intervention that might affect eicosanoid
metabolism. Our strategy is precisely intended to correct the unbal-
ance between PGI2 and TXA2 synthesis that has been observed in
patients exposed to a coxib [12].
[8] P.M. Kearney, C. Baigent, J. Godwin, H. Halls, J.R. Emberson, C. Patrono, Do
selective cyclo-oxygenase-2 inhibitors and traditional non-steroidal
anti-inflammatory drugs increase the risk of atherothrombosis?
Meta-analysis of randomised trials, BMJ 332 (7553) (2006) 1302–1308.
[9] T.D. Warner, J.A. Mitchell, COX-2 selectivity alone does not define the
cardiovascular risks associated with non-steroidal anti-inflammatory drugs,
Lancet 371 (9608) (2008) 270–273.
[10] G.A. Fitzgerald, Coxibs and cardiovascular disease, N. Engl. J. Med. 351 (17)
(2004) 1709–1711.

142
M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
[11] F. Catella-Lawson, B. McAdam, B.W. Morrison, S. Kapoor, D. Kujubu, L. Antes,
K.C. Lasseter, H. Quan, B.J. Gertz, G.A. FitzGerald, Effects of specific inhibition
of cyclooxygenase-2 on sodium balance, hemodynamics, and vasoactive
eicosanoids, J. Pharmacol. Exp. Ther. 289 (2) (1999) 735–741.
[12] B.F. McAdam, F. Catella-Lawson, I.A. Mardini, S. Kapoor, J.A. Lawson, G.A.
FitzGerald, Systemic biosynthesis of prostacyclin by cyclooxygenase (COX)-2:
the human pharmacology of a selective inhibitor of COX-2, Proc. Natl. Acad.
Sci. U. S. A. 96 (1) (1999) 272–277.
[13] E. Arehart, J. Stitham, F.W. Asselbergs, K. Douville, T. MacKenzie, K.M.
Fetalvero, S. Gleim, Z. Kasza, Y. Rao, L. Martel, S. Segel, J. Robb, A. Kaplan, M.
Simons, R.J. Powell, J.H. Moore, E.B. Rimm, K.A. Martin, J. Hwa, Acceleration of
cardiovascular disease by a dysfunctional prostacyclin receptor mutation:
potential implications for cyclooxygenase-2 inhibition, Circ. Res. 102 (8)
(2008) 986–993.
[31] A. Avdeef, J.E.A. Comer, S.J. Thomson, pH-metric log P. 3. Glass electrode
calibration in methanol–water, applied to pKa determination of
water-insoluble substances, Anal. Chem. 65 (1) (1993) 42–49.
[32] E.H. Kerns, L. Di, G.T. Carter, In vitro solubility assays in drug discovery, Curr.
Drug Metabol. 9 (9) (2008) 879–885.
[33] A. De Lean, P.J. Munson, D. Rodbard, Simultaneous analysis of families of
sigmoidal curves: application to bioassay, radioligand assay, and
physiological dose–response curves, Am. J. Physiol. 235 (1978) E97–E102.
[34] R. Bartzatt, Drug analogs of COX-2 selective inhibitors lumiracoxib and
valdecoxib derived from in silico search and optimization, Antiinflamm.
Antiallergy Agents Med. Chem. 13 (1) (2014) 17–28.
[35] R. Bartzatt, Anti-inflammatory drugs and prediction of new structures by
comparative analysis, Antiinflamm. Antiallergy Agents Med. Chem. 11 (2)
(2012) 151–160.
[14] J.A. Mitchell, R. Lucas, I. Vojnovic, K. Hasan, J.R. Pepper, T.D. Warner, Stronger
inhibition by nonsteroid anti-inflammatory drugs of cyclooxygenase-1 in
endothelial cells than platelets offers an explanation for increased risk of
thrombotic events, FASEB J. 20 (14) (2006) 2468–2475.
[15] R.P. Mason, M.F. Walter, C.A. Day, R.F. Jacob, A biological rationale for the
cardiotoxic effects of rofecoxib: comparative analysis with other COX-2
selective agents and NSAIDS, Subcell. Biochem. 42 (2007) 175–190.
[16] L.P. Audoly, B. Rocca, J.E. Fabre, B.H. Koller, D. Thomas, A.L. Loeb, T.M. Coffman,
G.A. FitzGerald, Cardiovascular responses to the isoprostanes iPF(2alpha)-III
and iPE(2)-III are mediated via the thromboxane A(2) receptor in vivo,
Circulation 101 (24) (2000) 2833–2840.
[17] E. Selg, C. Buccellati, M. Andersson, G.E. Rovati, M. Ezinga, A. Sala, A.K. Larsson,
M. Ambrosio, L. Lastbom, V. Capra, B. Dahlen, A. Ryrfeldt, G.C. Folco, S.E.
Dahlen, Antagonism of thromboxane receptors by diclofenac and
lumiracoxib, Br. J. Pharmacol. 152 (8) (2007) 1185–1195.
[18] G.H. Gislason, J.N. Rasmussen, S.Z. Abildstrom, T.K. Schramm, M.L. Hansen, E.L.
Fosbol, R. Sorensen, F. Folke, P. Buch, N. Gadsboll, S. Rasmussen, H.E. Poulsen,
L. Kober, M. Madsen, C. Torp-Pedersen, Increased mortality and cardiovascular
morbidity associated with use of nonsteroidal anti-inflammatory drugs in
chronic heart failure, Arch. Intern. Med. 169 (2) (2009) 141–149.
[19] E.L. Fosbol, G.H. Gislason, S. Jacobsen, F. Folke, M.L. Hansen, T.K. Schramm, R.
Sorensen, J.N. Rasmussen, S.S. Andersen, S.Z. Abildstrom, J. Traerup, H.E.
Poulsen, S. Rasmussen, L. Kober, C. Torp-Pedersen, Risk of myocardial
infarction and death associated with the use of nonsteroidal
[36] F. Salvo, A. Fourrier-Reglat, F. Bazin, P. Robinson, N. Riera-Guardia, M. Haag,
A.P. Caputi, N. Moore, M.C. Sturkenboom, A. Pariente, Investigators of safety of
non-steroidal anti-inflammatory drugs SOSP. Cardiovascular and
gastrointestinal safety of NSAIDs: a systematic review of meta-analyses of
randomized clinical trials, Clin. Pharmacol. Ther. 89 (6) (2011) 855–866.
[37] F. Catella-Lawson, M.P. Reilly, S.C. Kapoor, A.J. Cucchiara, S. DeMarco, B.
Tournier, S.N. Vyas, G.A. FitzGerald, Cyclooxygenase inhibitors and the
antiplatelet effects of aspirin, N. Engl. J. Med. 345 (25) (2001) 1809–1817.
[38] L.J. Marnett, The COXIB experience: a look in the rearview mirror, Annu. Rev.
Pharmacol. Toxicol. 49 (2009) 265–290.
[39] H. Sakai, T. Suzuki, Y. Takahashi, M. Ukai, K. Tauchi, T. Fujii, N. Horikawa, T.
Minamimura, Y. Tabuchi, M. Morii, K. Tsukada, N. Takeguchi, Upregulation of
thromboxane synthase in human colorectal carcinoma and the cancer cell
proliferation by thromboxane A2, FEBS Lett. 580 (14) (2006) 3368–3374.
[40] A.J. Cayatte, Y. Du, J. Oliver-Krasinski, G. Lavielle, T.J. Verbeuren, R.A. Cohen,
The thromboxane receptor antagonist S18886 but not aspirin inhibits
atherogenesis in apo E-deficient mice: evidence that eicosanoids other than
thromboxane contribute to atherosclerosis, Arterioscler. Thromb. Vasc. Biol.
20 (7) (2000) 1724–1728.
[41] K.S. Sakariassen, E.A. Femia, F.M. Daray, G.M. Podda, C. Razzari, M. Pugliano,
A.E. Errasti, A.R. Armesto, W. Nowak, P. Alberts, J.P. Meyer, A.S. Sorensen, M.
Cattaneo, R.P. Rothlin, EV-077 in vitro inhibits platelet aggregation in type-2
diabetics on aspirin, Thromb. Res. 130 (5) (2012) 746–752.
[42] P. Gaussem, J.L. Reny, C. Thalamas, N. Chatelain, M. Kroumova, B. Jude, B.
Boneu, J.N. Fiessinger, The specific thromboxane receptor antagonist S18886:
pharmacokinetic and pharmacodynamic studies, J. Thromb. Haemost. 3 (7)
(2005) 1437–1445.
anti-inflammatory drugs (NSAIDs) among healthy individuals: a nationwide
cohort study, Clin. Pharmacol. Ther. 85 (2) (2009) 190–197.
[20] W.A. Ray, C. Varas-Lorenzo, C.P. Chung, J. Castellsague, K.T. Murray, C.M. Stein,
J.R. Daugherty, P.G. Arbogast, L.A. Garcia-Rodriguez, Cardiovascular risks of
nonsteroidal antiinflammatory drugs in patients after hospitalization for
serious coronary heart disease, Circ. Cardiovasc. Qual. Outcomes 2 (3) (2009)
155–163.
[43] J.N. Fiessinger, H. Bounameaux, M.A. Cairols, D.L. Clement, S. Coccheri, J.P.
Fletcher, U. Hoffmann, A.G. Turpie, Thromboxane antagonism with terutroban
in peripheral arterial disease: the TAIPAD study, J. Thromb. Haemost. 8 (11)
(2010) 2369–2376.
[21] P. McGettigan, D. Henry, Use of non-steroidal anti-inflammatory drugs that
elevate cardiovascular risk: an examination of sales and essential medicines
lists in low-, middle-, and high-income countries, PLoS Med. 10 (2) (2013)
e1001388.
[22] I.S. Mackenzie, L. Wei, T.M. Macdonald, Cardiovascular safety of lumiracoxib:
a meta-analysis of randomised controlled trials in patients with
osteoarthritis, Eur. J. Clin. Pharmacol. 69 (2) (2013) 133–141.
[23] G.E. Rovati, A. Sala, V. Capra, S.E. Dahlen, Folco G. Dual, COXIB/TP antagonists:
a possible new twist in NSAID pharmacology and cardiovascular risk, Trends
Pharmacol. Sci. 31 (3) (2010) 102–107.
[24] M. Bertinaria, M.A. Shaikh, C. Buccellati, C. Cena, B. Rolando, L. Lazzarato, R.
Fruttero, A. Gasco, M. Hoxha, V. Capra, A. Sala, G.E. Rovati, Designing
multitarget anti-inflammatory agents: chemical modulation of the
lumiracoxib structure toward dual thromboxane antagonists-COX-2
inhibitors, ChemMedChem 7 (9) (2012) 1647–1660.
[44] P.F. Lesault, L. Boyer, G. Pelle, A. Covali-Noroc, D. Rideau, S. Akakpo, E. Teiger,
J.L. Dubois-Rande, S. Adnot, Daily administration of the TP receptor antagonist
terutroban improved endothelial function in high-cardiovascular-risk
patients with atherosclerosis, Br. J. Clin. Pharmacol. 71 (6) (2011) 844–851.
[45] M.G. Bousser, P. Amarenco, A. Chamorro, M. Fisher, I. Ford, K.M. Fox, M.G.
Hennerici, H.P. Mattle, P.M. Rothwell, A. de Cordoue, M.D. Fratacci, Terutroban
versus aspirin in patients with cerebral ischaemic events (PERFORM): a
randomised, double-blind, parallel-group trial, Lancet 377 (9782) (2011)
2013–2022.
[46] M. Lee, B. Ovbiagele, Vascular events after stroke: terutroban fails to
PERFORM, Lancet 377 (9782) (2011) 1980–1982.
[47] S.J. Wilson, A.M. Roche, E. Kostetskaia, E.M. Smyth, Dimerization of the human
receptors for prostacyclin and thromboxane facilitates thromboxane
receptor-mediated cAMP generation, J. Biol. Chem. 279 (51) (2004)
53036–53047.
[25] S. Simonet, J.J. Descombes, M.O. Vallez, T. Dubuffet, G. Lavielle, T.J. Verbeuren,
S 18886, a new thromboxane (TP)-receptor antagonist is the active isomer of
S 18204 in all species, except in the guinea-pig, Adv. Exp. Med. Biol. 433
(1997) 173–176.
[48] S. Ibrahim, M. Tetruashvily, A.J. Frey, S.J. Wilson, J. Stitham, J. Hwa, E.M.
Smyth, Dominant negative actions of human prostacyclin receptor variant
through dimerization: implications for cardiovascular disease, Arterioscler.
Thromb. Vasc. Biol. 30 (9) (2010) 1802–1809.
[26] V. Capra, A. Veltri, C. Foglia, L. Crimaldi, A. Habib, M. Parenti, G.E. Rovati,
Mutational analysis of the highly conserved ERY motif of the thromboxane A2
receptor: alternative role in G protein-coupled receptor signaling, Mol.
Pharmacol. 66 (4) (2004) 880–889.
[49] P. Patrignani, C. Di Febbo, S. Tacconelli, K. Douville, M.D. Guglielmi, R.J.
Horvath, M. Ding, K. Sierra, J. Stitham, S. Gleim, G. Baccante, V. Moretta, L. Di
Francesco, M.L. Capone, E. Porreca, J. Hwa, Differential association between
human prostacyclin receptor polymorphisms and the development of venous
thrombosis and intimal hyperplasia: a clinical biomarker study,
Pharmacogenet. Genomics 18 (7) (2008) 611–620.
[50] A.L. Blobaum, L.J. Marnett, Molecular determinants for the selective inhibition
of cyclooxygenase-2 by lumiracoxib, J. Biol. Chem. 282 (22) (2007)
16379–16390.
[51] 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 of
cyclooxygenase-2 inhibition involving interactions with Ser-530 and Tyr-385,
J. Biol. Chem. 278 (46) (2003) 45763–45769.
[52] Y. Yamamoto, K. Kamiya, S. Terao, Modeling of human thromboxane A2
receptor and analysis of the receptor-ligand interaction, J. Med. Chem. 36 (7)
(1993) 820–825.
[27] C. Buccellati, A. Sala, G. Rossoni, V. Capra, G.E. Rovati, A. Di Gennaro, G. Folco,
S. Colli, C. Casagrande, Pharmacological characterization of 2NTX-99
[4-methoxy-N1-(4-trans-nitrooxycyclohexyl)-N3-(3-pyridinylmethyl)-1,3-
benzenedicarboxamide], a potential antiatherothrombotic agent with
anti-thromboxane and Nitric Oxide donor activity in platelet and vascular
preparations, J. Pharmacol. Exp. Ther. 317 (2) (2006) 830–837.
[28] M. Ambrosio, F. Fanelli, S. Brocchetti, F. Raimondi, M. Mauri, G.E. Rovati, V.
Capra, Superactive mutants of thromboxane prostanoid receptor: functional
and computational analysis of an active form alternative to constitutively
active mutants, Cell. Mol. Life Sci. 67 (17) (2010) 2979–2989.
[29] F. Fanelli, M. Mauri, V. Capra, F. Raimondi, F. Guzzi, M. Ambrosio, G.E. Rovati,
M. Parenti, Light on the structure of thromboxane A(2) receptor
heterodimers, Cell. Mol. Life Sci. 68 (18) (2011) 3109–3120.
[30] V. Capra, M. Busnelli, A. Perenna, M. Ambrosio, M.R. Accomazzo, C. Gales, B.
Chini, G.E. Rovati, Full and partial agonists of thromboxane prostanoid
receptor unveil fine tuning of receptor superactive conformation and G
protein activation, PLoS One 8 (3) (2013) e60475.
[53] A. Krishna, A. Yadav, Lead compound design for TPR/COX dual inhibition, J.
Mol. Model 18 (9) (2012) 4397–4408.
[54] X. Wang, L. Liu, L. Huang, K. Herbst-Robinson, A.S. Cornec, M.J. James, S.
Sugiyama, M. Bassetto, A. Brancale, J.Q. Trojanowski, V.M. Lee, A.B. Smith 3rd,
K.R. Brunden, C. Ballatore, Potent long-acting cyclopentane-1,3-dione

M. Hoxha et al. / Pharmacological Research 103 (2016) 132–143
143
thromboxane (A2)-receptor antagonists, ACS Med. Chem. Lett. 5 (9) (2014)
1015–1020.
[55] C. Ballatore, J.H. Soper, F. Piscitelli, M. James, L. Huang, O. Atasoylu, D.M.
Huryn, J.Q. Trojanowski, V.M. Lee, K.R. Brunden, A.B. Smith,
[57] C. Giannarelli, M.U. Zafar, J.J. Badimon, Prostanoid and TP-receptors in
atherothrombosis: is there a role for their antagonism? Thromb. Haemost.
104 (5) (2010) 949–954.
[58] J.C. Breitner, P.P. Zandi, Do nonsteroidal antiinflammatory drugs reduce the
risk of Alzheimer’s disease? N. Engl. J. Med. 345 (21) (2001) 1567–1568.
[59] B.A. Veld, A. Ruitenberg, A. Hofman, L.J. Launer, C.M. van Duijn, T. Stijnen,
M.M. Breteler, B.H. Stricker, Nonsteroidal antiinflammatory drugs and the risk
of Alzheimer’s disease, N. Engl. J. Med. 345 (21) (2001) 1515–1521.
Cyclopentane-1,3-dione: a novel isostere for the carboxylic acid functional
group. Application to the design of potent thromboxane (A2) receptor
antagonists, J. Med. Chem. 54 (19) (2011) 6969–6983.
[56] M.L. Capone, S. Tacconelli, M.G. Sciulli, M. Grana, E. Ricciotti, P. Minuz, P. Di
Gregorio, G. Merciaro, C. Patrono, P. Patrignani, Clinical pharmacology of
platelet, monocyte, and vascular cyclooxygenase inhibition by naproxen and
low–dose aspirin in healthy subjects, Circulation 109 (12) (2004) 1468–1471.