Translational impact of novel widely pharmacological characterized mofezolac-derived COX-1 inhibitors combined with bortezomib on human multiple myeloma cell lines viability

Article abstract of DOI:10.1016/j.ejmech.2018.12.029

A set of novel diarylisoxazoles has been projected using mofezolac (1) as a lead compound to investigate structure-inhibitory activity relationships of new compounds and the cyclooxygenases (COXs) catalytic activity. Mofezolac was chosen because is the mo

Full text of DOI:10.1016/j.ejmech.2018.12.029

Accepted Manuscript
Translational impact of novel widely pharmacological characterized mofezolac-
derived COX-1 inhibitors combined with bortezomib on human multiple myeloma cell
lines viability
Maria Laura Pati, Paola Vitale, Savina Ferorelli, Mariaclara Iaselli, Morena Miciaccia,
Angelina Boccarelli, Giuseppe Davide Di Mauro, Cosimo G. Fortuna, Thaisa
Francielle Souza Domingos, Luiz Cláudio Rodrigues Pereira da Silva, Marcelo de
Pádula, Lucio Mendes Cabral, Plínio Cunha Sathler, Angelo Vacca, Antonio Scilimati,
Maria Grazia Perrone
PII:
S0223-5234(18)31066-3
DOI:
https://doi.org/10.1016/j.ejmech.2018.12.029
EJMECH 10962
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 21 September 2018
Revised Date: 10 December 2018
Accepted Date: 12 December 2018
Please cite this article as: M.L. Pati, P. Vitale, S. Ferorelli, M. Iaselli, M. Miciaccia, A. Boccarelli, G.D.
Di Mauro, C.G. Fortuna, T.F. Souza Domingos, Luiz.Clá. Rodrigues Pereira da Silva, M. de Pádula,
L.M. Cabral, Plí.Cunha. Sathler, A. Vacca, A. Scilimati, M.G. Perrone, Translational impact of novel
widely pharmacological characterized mofezolac-derived COX-1 inhibitors combined with bortezomib on
human multiple myeloma cell lines viability, European Journal of Medicinal Chemistry (2019), doi: https://
doi.org/10.1016/j.ejmech.2018.12.029.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Translational impact of novel widely pharmacological characterized mofezolac-derived COX-1
inhibitors combined with bortezomib on human multiple myeloma cell lines viability
Maria Laura Pati, Paola Vitale, Savina Ferorelli, Mariaclara Iaselli, Morena Miciaccia, Angelina Boccarelli,
Giuseppe Davide Di Mauro, Cosimo G. Fortuna, Thaisa Francielle Souza Domingos, Luiz Cláudio Rodrigues
Pereira da Silva, Marcelo de Pádula, Lucio Mendes Cabral, Plínio Cunha Sathler, Angelo Vacca, Antonio
Scilimati*, Maria Grazia Perrone*

ACCEPTED MANUSCRIPT
Translational impact of novel widely pharmacological characterized mofezolac-derived COX-
1
inhibitors combined with bortezomib on human multiple myeloma cell lines viability
a§
a§
a
a
a
Maria Laura Pati , Paola Vitale , Savina Ferorelli , Mariaclara Iaselli , Morena Miciaccia ,
b
c
c
Angelina Boccarelli , Giuseppe Davide Di Mauro , Cosimo G. Fortuna , Thaisa Francielle Souza
d
d
d
d
Domingos , Luiz Cláudio Rodrigues Pereira da Silva , Marcelo de Pádula , Lucio Mendes Cabral ,
d
e
a
a
Plínio Cunha Sathler , Angelo Vacca , Antonio Scilimati *, Maria Grazia Perrone *
a
Department of Pharmacy - Pharmaceutical Sciences, University of Bari “Aldo Moro”, Via E.
b
Orabona 4, 70125 Bari, Italy. Department of Biomedical Sciences and Human Oncology,
c
University of Bari “Aldo Moro”, Piazza Giulio Cesare 11, 70124 Bari, Italy. Department of
d
Chemical Science, University of Catania, Viale Andrea Doria 6, 95125 Catania, Italy. Faculty of
Pharmacy, Federal University of Rio de Janeiro, Center of Health Sciences, Carlos Chagas Filho
e
Avenue, 373, 21941599, Rio de Janeiro, Brazil. Department of Biomedical Sciences and Human
Oncology, Internal Medicine Unit G. Baccelli, University of Bari Aldo Moro Medical School, Bari,
Italy.
Keywords: cyclooxygenase, selective COX-1 inhibition, mofezolac, structure-inhibitory activity
relationships, bortezomib, multiple myeloma.
Corresponding authors:
Phone: +39 0805442747. Fax: +39 0805442724. E-mail address: mariagrazia.perrone@uniba.it
(M.G. Perrone).
Phone: +39 0805442753. Fax: +39 0805442724. E-mail address: antonio.scilimati@uniba.it (A.
Scilimati).
§
M. L. Pati and P. Vitale have equally contributed to this project.

ACCEPTED MANUSCRIPT
Abstract: A set of novel diarylisoxazoles has been projected using mofezolac (1) as a lead
compound to investigate structure-inhibitory activity relationships of such compounds and the
cyclooxygenases (COXs) catalytic activity. Mofezolac was chosen because is the most potent and
selective reversible COX-1 inhibitor [COX-1 IC = 0.0079 µM and COX-2 IC > 50 µM, with a
5
0
50
selectivity index (SI) in favor of COX-1 higher than 6,300]. Seventeen new compounds were
synthesized in fair to good yields and evaluated for their COXs inhibitory activity and selectivity.
SIs ranged between 1 and higher than 1,190. 3,4-Bis(4-methoxyphenyl)-5-vinylisoxazole (22) has
the highest SI with COX-1 IC = 0.042 µM and COX-2 IC > 50 µM. 1 and 22 were superior to
5
0
50
aspirin in inhibiting platelet aggregation (IC = 0.45, 0.63 and 1.11 µM, respectively) in hPRP
5
0
assay. They did not induce blood coagulation and hemolysis, and are neither genotoxic nor
mutagen. 1 and 22 slightly increase bortezomib cytotoxic effect on multiple myeloma (MM) cell
lines (NCI-H929 and RPMI-8226) and affects MM cell cycle and apoptosis when co-administered
with bortezomib, a drug clinically used to treat plasma cell neoplasms including MM. In addition,
structure-based binding mode of 1 and 22, through FLAG calculation, allowed to explain the one
order of magnitude difference between COX-1 IC values of the two compounds. Specifically, the
50
higher inhibitory potency seems due to the formation of a H-bond between COX-1 S530 and the
carboxyl, present in 1 and absent in 22.

ACCEPTED MANUSCRIPT
INTRODUCTION
Non-Steroidal Anti-inflammatory Drugs (NSAIDs) are the most used drugs to treat inflammatory
syndrome, fever, pain, and rheumatic pathologies. NSAIDs therapeutic and side effects are mainly
due to the inhibition of prostaglandins biosynthesis, that starts with the bis-oxygenation of
arachidonic acid (AA) catalyzed by the cyclooxygenase (COX) isoenzymes, in which the first
formed prostaglandin (PG)G is converted into the PGH , in turn precursor of the PGI , PGE ,
2
2
2
2
thromboxane (TX)A and other eicosanoids [1]. Prostaglandins are crucial mediators of
2
inflammation [2], regulators of platelet function and hemostasis [3], angiogenesis [4] and support
the growth of several solid tumors [5].
Two COXs isoforms are known, COX-1 and COX-2. Their expression is either constitutive or
inducible, depending on the organ considered and specific physio-pathological established
condition [6]. In particular, COX-2 expression is constitutive in liver and brain, and in almost all
body districts where it is induced in response to inflammatory stimuli and growth factors upon
various pathological conditions in which the inflammation is the committed earliest stage [7],
including cancer and, neurological and neurodegenerative diseases [8-12]. COX-1 plays
homeostatic function in most tissues (e.g. platelets, lung, prostate, brain, gastrointestinal tract,
kidney, liver and spleen), where it is constitutively expressed, and its overexpression is rarely
observed [1]. Experimental evidences proven COX-1 expressed and involved in different tumors
such as breast [13], ovarian [14], head and neck [15] cancer, renal cell carcinoma [16], and
hematological (i.e., multiple myeloma) tumors [17] for which it could be considered as a biomarker
of a diagnostic value.
Multiple myeloma (MM) is a tumor that affects the plasma cells, key components of the immune
system. MM is also characterized as a clonal proliferation of neoplastic plasma cells in the bone
marrow (BM) and accounts for 30 % of all hematological malignancies [18]. Nowadays,
bortezomib, a proteasome inhibitor, or immunomodulators such as lenalidomide and thalidomide
are used to treat MM. The effect of bortezomib, the first proteasome inhibitor to be approved by the

ACCEPTED MANUSCRIPT
Food and Drug Administration in 2003, is well recognized. At present, bortezomib is the first- and
second-line treatment in MM. However, the increasing administration of bortezomib is associated
with the development of drug resistance [19].
Due to the high risk of thromboembolism, as a heavy side effect, associated to the above mentioned
immunomodulators [20-21] or proteasome inhibitor treatment, MM patients preventively hire also
aspirin (Figure 1) [22].
Nearly all MM patients experience relapse and refractory disease [23]. Aspirin use, along with
reducing the incidence of thrombotic complications also improves overall survival, by producing
additive or synergistic effects in the treatment of MM because enhances bortezomib cytotoxicity via
suppression of Bcl-2, survivin and AKT phosphorylation [24].
Low dose-aspirin anti-thrombotic effect is mainly due to the irreversible inhibition of platelet COX-
1
catalytic activity by acetylation of COX-1 Ser530, that permanently blocks the AA binding inside
the enzyme active site.
Platelets are almost unable to replace inactive acetyl-COX-1 being anucleated. Thus, daily aspirin
hiring has a therapeutic advantage against thrombosis acting by inhibiting any COX-1 mediated
reaction. Unfortunately, variability in the sensitivity of patients to aspirin interferes with the
necessary complete and persistent suppression of TXA biosynthesis in cardio-protection, by
2
causing among other factors COX-1 dependent “aspirin resistance” [25].
The use of aspirin is also restricted, since it causes either patient-hypersensitivity or-intolerance in
up to 20 % of the worldwide population [26].
Aspirin resistance, hypersensitivity and intolerance limit the aspirin use. Therefore, COX-1
inhibitors alternative to aspirin are needed.
An increasing number of selective COX-1 inhibitors has been found, such as TFAP, ABEX-3TF,
FR122047, SC-560, P6, and mofezolac (Figure 1 and 2)]. Among them, mofezolac is the unique
marketed drug, sold in Japan as Disopain® to relieve rheumatoid arthritis pain [1].

ACCEPTED MANUSCRIPT
CF
CF3
3
COOH
OCOCH3
O
NH
O
NH
OCH CH3
2
H N
H N
2
2
aspirin
TFAP
ABEX-3TF
H CO
3
OCH3
F3C
N
N
S
N
N
O
Cl
N
H C
3
OCH3
FR122047
SC-560
Figure 1. Aspirin and COX-1 selective inhibitors.
Most of the highly selective COX-1 inhibitors belong to the diarylheterocycle chemical class, and
among them the isoxazole has particularly been used as central ring in the structure of mofezolac,
P6 [3-(5-chlorofuran-2-yl)-5-methyl-4-phenylisoxazole] and its analogues, and valdecoxib
(conversely, is a selective COX-2 inhibitor) (Figure 2) [27].
P6 and mofezolac are two isoxazole-scaffold based COX-1 inhibitors: P6 bears a methyl on C5, a
phenyl on the isoxazole-C4 and a 5-chlorofuranyl on isoxazole-C3, whereas two 4-methoxyphenyls
are linked to mofezolac central heterocyclic ring, that possesses an acidic moiety on C5. Valdecoxib
chemical structure has a phenyl on the isoxazole-C3, a methyl on isoxazole-C5, and differs mainly
from P6 and mofezolac for the presence of a sulfonamide group on the para position of the
isoxazole-C4 phenyl, that seems responsible of the reversal of the selectivity in favor of COX-2
(
Figure 2). The presence of an acidic moiety (-CH COOH) at mofezolac C5 seems to be important
2

ACCEPTED MANUSCRIPT
for the high inhibitory potency and COX-1 selectivity, mainly due to its strong interaction with
R120 located at entrance of the COX-1 catalytic site [28].
Figure 2. Isoxazole-scaffold based COXs inhibitors.
A number of diarylheterocycles with a different central ring have been previously prepared with the
aim to clarify the importance of the isoxazole for the selective COX-1 inhibition by P6 and its
analogues. In particular, the isoxazole replacement by isothiazole or pyrazole determines a
remarkable inhibitory activity reduction of the corresponding diarylisothiazoles and diarylpyrazoles
in the COX-1 catalyzed AA bis-oxygenation reaction [29].
In this regard, differently substituted 3,4-diarylisoxazoles were prepared and tested as COX-1 and
COX-2 isoenzymes inhibitors for an extensive structure-inhibitory activity relationship
investigation [30-31]. Based on the obtained results some compound key features appeared
determinant to selectively inhibit COX-1 (Figure 3).
NH2
R=CH , Cl, Br
F
3
O
R
F
Cl
COOCH , CH OH
3
2
Cl
OCH3
Ar
R1
R2
N
CH , CF , CH OH, CH CH
3 3 2 2 3
O
CH OCOCH
2
3
COOCH3
F, NH2
phenyl

ACCEPTED MANUSCRIPT
Figure 3. Key chemical moieties determinant isoxazole-based scaffold for selective COX-1
inhibitors.
Specifically, in the presence of a methyl, bromo or chloro-substituted furan linked to the isoxazole-
C3, the aromatic moiety at the isoxazole-C4 can be a phenyl or an aryl bearing electron-donating
group (EDG) capable of +M effect (o- or p-halophenyl, p-aminophenyl, or p-methoxyphenyl); a
variety of substituents are well-tolerated at isoxazole-C5 position such as short-chain alkyls, CF3,
hydroxymethyl, amino groups (Figure 3) [27,32].
As an attempt to still validate these findings, herein, we describe a new set of isoxazoles designed
taking into account the already identified determinants. The novel compounds were projected also
with further additional functionalities as an attempt to target diseases such as multiple myeloma
[
20-21, 33], thrombosis [34], and ovarian cancer [14].
RESULTS AND DISCUSSION
A first set of molecules was designed as diarylisoxazolyl-5-acetic acids, in analogy to mofezolac (1,
oCOX-1 IC = 0.0079 µM and hCOX-2 IC > 50 µM), whereas the second set derived from the 5-
5
0
50
amino-3-(5-chlorofuran-2-yl)-4-phenylisoxazole (27) used as a “lead compound” [oCOX-1 IC =
5
0
1
.1 µM and hCOX-2 IC > 50 µM, and with COX-1 IC = 0.58 µM inhibitory activity assessed in
50 50
the human ovarian cancer cell line hOVCAR-3 expressing only the COX-1 isoform [35]. The main
difference between mofezolac and 27 is the presence at isoxazole-C5 of an acid (acetic group) in
mofezolac and a basic (amino group) in 27.
The isoxazoles 8-12 have been prepared by "one-pot" 1,3-dipolar cycloaddition of substituted
arylnitrile oxides 6a-c and 6e (Scheme 1) and the arylpropanones sodium enolate 7 generated at 0
°
C by deprotonation of the corresponding ketones with NaH under thermodynamic conditions
(Scheme 2) [34-35]. Specifically, arylnitrile oxides were prepared by dehydrohalogenation of

ACCEPTED MANUSCRIPT
benzohydroximinoyl chlorides 5a-c and 5e in turn obtained from the corresponding oximes 4a-c
and 4e.
i
ii
iii
ArCHO
ArCH
a-e
N
OH
ArC(Cl) N OH
ArCNO
4
5a-c, 5e
6a-c, 6e
Ar =
a
: 4-CH OC H
b
: 4-NO C H
c
: 4-FC H
d: 4-CH SC H e: Ar = 4-CH SO C H
4 3 6 4 3 2 6 4
3
6
4
2
6
4
6
Scheme 1. Synthesis of the arylnitrile oxides 6a-c, 6e. Reagent and conditions: (i) NH OH HCl/aq.
2
NaOH, quantitative yield; (ii) N-chlorosuccinimide, DMF; (iii) Et N/THF, quantitative yield.
3
Scheme 2. Synthesis of diarylisoxazoles 8-12. Reagent and conditions: (i) NaH/THF, 0 °C; (ii)
arylnitrile oxide (ArCNO)/THF, from 0 °C to r.t.; (iii) aq. NH Cl.
4
Reduction of the nitro group of the isoxazole 12 by stannous chloride in hydrochloric acid provides
13 [35], that by reacting with acetic anhydride affords 14 (Scheme 3).
Scheme 3. Synthesis of 13-14. Reagents and conditions: (i) SnCl /HCl, EtOH, 24h; (ii) Ac O/Py,
2
2
r.t..

ACCEPTED MANUSCRIPT
3,4-Diaryl-5-methylisoxazoles 8, 10, 13-14 were treated with n-BuLi and kept at -78 °C for 1h.
Then, anhydrous CO was bubbled into the lithiated intermediates still at -78 °C, to afford the
2
corresponding diarylisoxazolyl acetic acids 1 (mofezolac) [28] and 15-17 (Scheme 4).
Scheme 4. Synthesis of diarylisoxazolyl acetic acids 1, 15-17. Reagents and conditions: (i) n-BuLi,
-
78 °C, 1h; (ii) CO (g), from -78 °C to r.t.; (iii) 0.1M HCl.
2
3,4-Bis(4-methoxyphenyl)-5-(bromomethyl)isoxazole (18) was prepared by bromination of 8 with
N-bromosuccinimide (NBS) in the presence of azobisisobutyronitrile (AIBN) as a radical initiator.
Then, 18 reacted with anhydrous tetrabutylammonium fluoride (TBAF) to afford 19 (Scheme 5).
Scheme 5. Preparation of 5-halomethylisoxazoles 18-19. Reagents and conditions: (i) N-
bromosuccinimide/azoisobutironitrile/CCl ; (ii) anhydrous tetrabutylamonium fluoride/THF.
4
The alcohol 20 was obtained by reducing the carboxylic moiety of mofezolac in the presence of
borane. 20 was transformed into its tosylate 21 by reaction with Ts O in the presence of Et N, and
2
3

ACCEPTED MANUSCRIPT
then the 5-vinylisoxazole 22 formed by beta-elimination of the toluensulfonic acid, induced by
TBAF (Scheme 6).
Scheme 6. Synthesis of the target 20 and 22. Reagents and conditions: (i) BMS/THF at 0 °C to r.t.;
(
ii) Ts O, Et N at 0 °C to r.t.; (iii) TBAF/THF, r.t..
2 3
5-Amino-3,4-diarylisoxazoles 23-26 were prepared by a one-pot methodology, through a
cycloaddition reaction between arylnitrile oxides and lithium arylacetonitriles [37] (Scheme 7):
Scheme 7. Synthesis of 23-26. Reagents and conditions: (i) LDA/THF, -78 °C to r.t.; (ii) arylnitrile
oxides (ArCNO)/THF, 0 °C to r.t.; (iii) aq. NH Cl.
4

ACCEPTED MANUSCRIPT
All the novel isoxazoles were evaluated in vitro against oCOX-1 and hCOX-2 to determine their
COXs IC values (Table 1).
5
0
Table 1. COX inhibitory activity of isoxazoles 8-13, 15-20, 22-28 and mofezolac (1) used as a
reference compound.
COX-1
COX-2
Compound
Ar
G
R
a
a
IC (µM)
IC (µM)
5
0
50
1
8
9
1
1
1
1
1
1
1
1
1
2
2
2
4-CH
3
3
OC
OC
6
6
H
4
OCH
OCH
F
3
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
2
3
3
3
3
3
3
2
2
2
2
2
2
COOH
0.0079
0.076
> 50
4-CH
H
4
0.35
3
4-NO
C
H
> 50
> 50
2
6
4
0
1
2
3
5
6
7
8
9
0
2
3
4-FC
6
H
4
OCH
3
3 (94)
> 50 (24)
> 50
4-CH
SO
C
H
F
0.07 (52)
3
2
6
4
4-NO
4-NH
C
C
H
H
OCH
OCH
OCH
OCH
OCH
OCH
OCH
OCH
OCH
OCH
36 % at 10 µM > 50
2
6
6
4
3
3
3
3
3
3
3
3
3
3
23 (53)
> 50
2
4
4-FC
H
4
COOH
COOH
COOH
Br
7.5 (92)
> 50 (33)
12 (79)
> 50
6
4-NH
C
H
> 50 (42)
> 50
2
6
4
4-CH
4-CH
4-CH
4-CH
4-CH
4-CH
CONHC H
6 4
3
OC
OC
OC
OC
OC
H
4
H
4
H
4
H
4
H
4
0.10 (93)
0.35 (100)
0.18 (63)
0.3 (85)
0.19 (82)
3
3
3
3
3
6
6
6
6
6
F
CH
CH=CH
NH
OH 0.12 (100)
2
b
b
0.042 (100)
> 50
2
8.3 (89)
> 50
2

ACCEPTED MANUSCRIPT
2
2
2
2
2
4
5
6
7
8
4-FC
4-NO
4-CH
6
H
4
OCH
3
NH
NH
NH
2
2
2
13 (83)
12 (71)
> 50
> 50
> 50
> 50
> 50
> 50
C H
6
F
2
4
SO
C
H
4
F
3
2
6
c
c
5-chlorofuran-2-yl
5-chlorofuran-2-yl
H
NH2
CH3
1.1
NH2
4.3
a) in round brackets is reported the inhibition percentage (%) of the isoenzyme activity at the
compound final concentration of 50 µM; b) Selective Index, SI = COX-2 IC / COX-1 IC >
5
0
50
1,190; c) [35].
Mofezolac (1), one of our “leads” used as reference compound for this SAR investigation has a
selective COX-1 inhibitory activity (IC = 0.0079 µM) at nanomolar level. The absence of the
5
0
acidic function on isoxazole-C5, as in compound 8, leads to a preferential but non-selective COX-1
inhibitor, and also less potent (COX-1 IC = 0.076 µM and COX-2 IC = 0.35 µM, selective index
5
0
50
SI = 4.6). The introduction of a nitro group on the phenyl linked to the isoxazole-C4 was found to
be detrimental to 9 and 12 inhibitory activity of both cyclooxygenase isoforms. Particularly
interesting is the introduction of a vinyl group to the isoxazole-C5, affording compound 22, found
to be selective and potent COX-1 inhibitor (SI > 1,190). The substitution of the carboxyl with a
hydroxyl determines a reduction of the COXs inhibitory activity of 20 (COX-1 IC = 0.12 µM,
5
0
COX-2 IC50 = 0.19 µM, SI = 1.6) compared to 1. Metabolically, 20 could be considered the
precursor of mofezolac (1). The coexistence of two methoxyphenyls and a -CH X [X = bromine
2
(
18) or fluorine (19)] on isoxazole-C5 led to a submicromolar COX-1 inhibitor potency (0.10 and
.35 µM, respectively) but also to a lost in selectivity. In fact, both halo-compounds inhibit COX-2
with an IC of 0.18 and 0.3 µM for compound 18 and 19, respectively. The presence of fluorine as
0
5
0
a substituent of the aryl group in position C3 of the isoxazole leads to selective COX-1 inhibitors
10, 15, and 24) with a partial loss of inhibitory potency. In fact, compound 15, which is different
from the lead compound (1) only for the substitution of a methoxy group with a fluorine atom
(

ACCEPTED MANUSCRIPT
remains a selective COX-1 inhibitor, but its IC = 7.5 µM is lower of three orders of magnitude
5
0
than 1. When the acid function is also replaced with a basic amino group (24), the inhibitory
potency is further reduced (IC = 13 µM), probably due to the simultaneous presence of a fluorine
5
0
atom and NH group leads to micromolar COX-1 inhibitors potency. COX-1 inhibitory activity
2
(
IC = 3.0 µM) of 10 is comparable to that of its acid derivative 15, indicating that the presence of
50
fluorine contributes to COX-1 selectivity but not to the potency of the novel synthesized
diarylisoxazoles.
On the other hand, the presence of NH group is determinant in the interaction with COX-1 active
2
site because it forms strong H-bond with COX-1 Ser 530, which adopts a “down” position during
the flexible docking simulation and it does not matter if the NH is bound to the isoxazole-C5 (23,
2
25 and 27) or to the phenyl at isoxazole-C4 (28) [35]. Comparing 25 and 9 behavior, there is a
further confirmation of our previous studies. Specifically, the absence of the amino group at
isoxazole-C5, compound 9, determines a total loss of COXs inhibitory activity with respect to the
compound 25. On the contrary, the introduction of the amino group on the phenyl to isoxazole-C3
is associated with a loss of inhibitory activity (13 and 16). In particular, 16 has been designed to
better understand if the position of the amino group has a role in the interaction with the active site
of COX-1 and in fact by simply replacing the methoxy group of 1 with an amino group, a COX-1
inhibitor with a nanomolar activity becomes an inactive compound.
By transforming the amino group in an amide, reverting the aminic to acidic characteristic of the
moiety (NH to NHCOCH , compound 17) COX-1 inhibitory activity was restored (IC = 12 µM).
2
3
50
On the other hand, the introduction of a 4-methylsulfamoyl substituent on the phenyl linked to
isoxazole-C3 reverts the selectivity, even if it is not a rule [26]. In fact, compound 11 has a COX-2
IC = 0.07 µM. As it is evident from the data listed in Table 1, among all the synthesized inhibitors,
5
0
the most potent and selective identified inhibitor is compound 22. To rationalize this outcome, an in
silico study using Fingerprints for Ligands and Proteins (FLAP) [38] by using COX-1 co-
crystallized with mofezolac (PDB code: 5HQK) has been performed, also to identify fruitful and

ACCEPTED MANUSCRIPT
eventually not yet identified interactions established between 22 and COX-1 aminoacids forming
the active site.
Structure-based binding mode. The X-ray structure of the complex COX-1:mofezolac (1) was
used as a validated model [28]. First, 1 was removed from the COX-1 template and without any
constrain FLAP (Fingerprints for Ligands and Proteins) [38] calculated by GRID molecular
interaction fields (a computational procedure for determining energetically favorable binding sites
on biological macromolecoles) of 22 in the cavity of COX-1. 22 was chosen because it was the best
COXs inhibitors of set above reported (COXs SI > 1,190 in favor of COX-1, Table 1). The results
of the best arrangement of 22 were analyzed (Figure 4).
Figure 4. Arrangement of the 3,4-bis(4-methoxyphenyl)-5-vinylisoxazole (22) in COX-1 active
site.

ACCEPTED MANUSCRIPT
Hydrophobic interactions have been observed between the isoxazole ring and A527, the vinyl group
and F518, the phenyl linked to the isoxazole-C3 with L359 and Y355; the phenyl on the isoxazole-
C4 with Y355, I523 and L352.
The strongest interaction (violet spot in the Figure 4) is observed between the methoxy of the
phenyl on the on the isoxazole-C4, as a H-bond acceptor, and Y355.
In Figure 5, the superimposition of mofezolac (1) and 22 in the COX-1 active site is reported.
Figure 5. Mofezolac (1) (green) and 22 (blue) in the COX-1 active site.
From the comparison of the best arrangement of both compounds in the active site, analyzing their
interactions with specific residues, it is possible to explain the difference between their COX-1
inhibitory activity values: mofezolac (1) (IC = 0.0079 µM) and 22 (IC = 0.042 µM). They show
5
0
50
a similar orientation in the active site, demonstrated by their common interactions with I523, L352
and Y355, whereas for mofezolac a strong H-bond donor interaction between S530 and the acid

ACCEPTED MANUSCRIPT
group is observed. Probably, such a H-bonding is one of the main factors responsible of mofezolac
higher inhibitory activity than 22.
In addition, a further docking investigation was performed to explain the inactivity of 9. Figure 6
shows as 9 interacts with the active site of the COX-1.
Figure 6. Compound 9 in COX-1 active site.
Hydrophobic interactions are observed between the isoxazole ring and L352, the methyl and L531,
the phenyl at isoxazole-C3 with F518, L352 and Y355 and, the phenyl on the isoxazole-C4 with
Y355 and L531.
The strongest interaction (violet spot in Figure 6) is observed between the methyl linked to the
isoxazole ring and L531.

ACCEPTED MANUSCRIPT
Figure 7. Mofezolac (1) (green) and 9 (blue) in the active site of COX-1.
The best arrangement of both compounds in the active site, allows to explain the difference between
their COX-1 inhibitory activity values (Table 1). They have similar orientation in the active site by
interacting with L352 and Y355. Mofezolac has a strong H-bond donor interaction between S530
and the acid group, absent in 9. Moreover, mofezolac interacts with I523 and this interaction has not
been observed for 9. Probably, such a H-bonding and the interaction with I523 are the main factors
responsible for mofezolac higher inhibitory activity than 9 (Figure 7).
The methoxy substituent is a “key group” in the interaction with COX-1 active site. In our previous
investigations, the para-methoxyphenyl of a diarylpyrazole was found oriented toward the apex of
the COX-1 active site and formed hydrophobic interactions with residues F518, M522, and I523,
whereas its methoxy substituent lies within van der Waals contact range of the side chains of F381,
L384, Y385, and W387. Notably, the oxygen atom of the para-methoxy forms two contact points to
the W387 side chain. W387 is located at the top of the COX-1 active site near the catalytic Y385

ACCEPTED MANUSCRIPT
residue that is critical for the proper positioning of arachidonic acid within the active site to yield
the cyclooxygenase-mediated PGG2.
Pharmacological and toxicological characterization of compounds 1 and 22. Mofezolac (1) and
22 were further characterized by evaluating their contribution in platelet aggregation and blood
coagulation, and their hemolytic and genotoxicity effects. In addition, their multiple myeloma cell
lines cytotoxicity, apoptosis and cell cycle modification effects has been investigated to verify if
these two highly selective COX-1 inhibitors have the same aspirin additive or synergistic effects in
the treatment of MM when co-administered with bortezomib, one of the drugs clinically used to
treat plasma cell neoplasms including MM [24].
Platelet aggregation assay. The antiplatelet profile of 1, 22 and aspirin (reference compound) have
been determined by using an in vitro platelet aggregation method in which platelet aggregation was
induced by arachidonic acid (AA) in human citrated platelet rich plasma. 1 and 22 have a
comparable potency in inhibiting platelet aggregation. Their IC are 0.45 ± 0.04 µM and 0.63 ±
5
0
0
.17 µM, respectively, and better than aspirin (IC =1.11±1.17 µM). Thus, 1 and 22, as aspirin,
50
reduce the TXA production by COX-1 inhibition as a part of the mechanism that impair normal
2
platelet aggregation [39].
Anticoagulant assays. Blood anticoagulant profile of 1 and 22 has been determined by activated
partial thromboplastin time (APTT) and prothrombin time (PT) using human plasma (Table 2).
a
Table 2. Evaluation of 1 and 22 (100 µM) in in vitro coagulation of human plasma (n = 3) by
activated partial thromboplastin time (APTT) and prothrombin time (PT).
Compounds
PT (seconds) ± SD
13.9 ± 0.8
APTT (seconds) ± SD
44.8 ± 1.5
C –
C +
1
300 ± 0.0
300 ± 0.0
14.2 ± 0.3
50.8 ± 0.23

ACCEPTED MANUSCRIPT
2
2
13.7 ± 0.5
51.2 ± 0.3
50.3 ± 0.3
aspirin
14.6 ± 1.0
(a) negative control (C-): 1 % DMSO; positive control (C+): 100 µM rivaroxaban; aspirin (100
µM). The APTT and PT values are expressed by mean (n = 3) and standard deviation (SD).
Neither the extrinsic nor intrinsic pathways of the coagulation cascade were affected by 1 and 22,
suggesting that their anti-hemostatic profile relies on the direct impairment of platelet aggregation,
thus differing from dual action molecules (Table 2). This result reveals a small bleeding risk in
comparison to dual acting molecules whose mechanism often leads to severe bleeding disorders
[
40-41].
Hemolytic activity. The hemolytic activity of 1 and 22 in comparison to aspirin was investigated
and results are listed in Table 3.
a
Table 3. Hemolytic profile of 1 and 22 (100 µM) by hemolysis assay: values below 10 % are
considered non-hemolytic.
Compounds
Hemolysis (%) ± SD
0.00 ± 0.22
100 ± 2.01
C–
C+
1
0.00 ± 0.02
0.0 ± 0.01
22
aspirin
0.0 ± 0.04
a
1
% DMSO is the negative control (C-) and 1 % Triton X-100 (C+) is the positive control; aspirin
(100 µM). The hemolysis assay values are expressed by mean (n = 3) and standard deviation (SD).

ACCEPTED MANUSCRIPT
The hemocompability profile for the compounds was determined based on the degree of erythrocyte
lysis observed in the 3h incubation period at 37°C. Hemolysis values below 10 % are considered
non-hemolytic, confirming an acceptable toxicity profile of the tested compounds [42].
Mutagenic and genotoxic activity. The safety profile of 1 and 22 is reinforced by genotoxicity
tests which revealed no mutagenicity against Salmonella typhimurium auxotroph mutant strains for
the two compounds through reverse mutagenesis and histidine prototrophy (Ames test), compared
to the positive control 4-nitro-quinoline 1-oxide (4-NQO), nor genotoxicity, evaluated against
Escherichia coli through the SOS chromotest. This low risk profile is maintained even at the
highest tested concentration (500 µM, Table 4), revealing the safety profile of 1 and 22 as
promising for “lead prototypes”.
Table 4. Mutagenic and genotoxic activity of 1 and 22 (500 µM) without metabolic activation.
Ames test
SOS chromotest
Compound
1
S. typhimurium
E. coli
TA97
TA98
TA100 TA102
PQ35
PQ37
-
-
-
-
-
-
-
-
-
-
-
-
22
a
4
-NQO
+
-
+
-
+
-
+
-
+
-
+
-
b
DMSO
aspirin
-
-
-
-
-
-
a) 4-NQO was used as a positive control; b) 1 % DMSO was used as a negative control.
Western blotting. Based on our best knowledge, the expression of COX-1 and COX-2 in different
myeloma cell lines has received little attention [17] and it is valuable any investigation aimed at
deepening the COXs link with MM. For this purpose, two cell lines derived of human multiple

ACCEPTED MANUSCRIPT
myeloma (NCI-H929 and RPMI-8226) were chosen for further investigations. NCI-H929 expresses
only COX-1 in a very low percentage which was quantified as 9 % with respect to HEK-293 COX-
1, whereas RPMI-8226 does not express both COX-1 and COX-2 (Figure 8). The engineered cell
lines HEK-293 COX-1 and HEK-293 COX-2 that express COX-1 or COX-2, respectively, upon
stimulation with 10µg/mL tetracycline, have been used as positive controls [43].
Figure 8. Expression of COX-1 and COX-2 isoenzymes in NCI-H929 and RPMI-8226 cell lines
[
cell homogenates (30 µg protein) were applied to western blotting]. HEK-293 COX-1 and HEK-
93 COX-2 were used as positive controls for COX-1 and COX-2 expression, respectively. The
2
protein level of β-actin was used as loading control. The image is representative of one out of three
experiments.
Cytotoxic activity. Since no data regarding COX-1 inhibitors in MM treatment are available, 1 and
22, as two potent and selective COX-1 inhibitors, were tested in two MM NCI-H929 and RPMI
8226 cell lines to verify their cytotoxicity, alone and in combination with bortezomib, a commonly
drug used in therapy. The effect on cell viability of 1 and 22 was compared with the effect of aspirin
used alone and in combination with bortezomib.
1
, 22 and bortezomib were preliminarily used at different concentrations chosen upon their
previously determined COXs EC or IC values. In both cell lines bortezomib was effective at
5
0
50
very low concentrations (2nM in NCI-H929 and 3nM in RPMI-8226), whereas 1 and 22 were not

ACCEPTED MANUSCRIPT
cytotoxic at the concentrations used. 7.9 nM for compound 1, 42 nM for compound 22 and 1.7 µM
for aspirin (COX-1 IC ). In both RPMI 8226 and NCI-H929 cell lines the antiproliferative activity
5
0
of bortezomib was improved not affected by the combined treatment with 1, 22 or aspirin (Figure
A and 9B).
A
9
RPMI 8226
100
80
60
40
20
0
B
NCI-H929
100
80
60
40
20
0

ACCEPTED MANUSCRIPT
Figure 9. Antiproliferative activity of bortezomib alone and in combination with 1, 22 or aspirin
after 48h incubation time with RPMI-8226 (A) and NCI-H929 (B) cell lines. Error bars represent
mean ± SD of three experiments in triplicate; one-way ANOVA followed by Bonferroni’s post-hoc
comparison test: P>0.05
In RPMI-8226, that does not express COXs, bortezomib has a 53 % cytotoxicity whereas aspirin, 1
and 22 have a lower toxicity of 7, 1 and 4 %, respectively. The cytotoxicity of the proteasome
inhibitor bortezomib does not change when administered in association with the COX-1 inhibitors.
In fact, its toxicity extent reaches 66 % if co-administered with aspirin, 56 % and 66 % in co-
administration with 1 or 22, respectively.
When the same experiment was performed with NCI-H929, that expresses only COX-1, the
percentages changed. In all cases, co-administration leads to an decrease in the cytotoxicity of
bortezomib compared to when it is administered alone.
These data demonstrate that, in our conditions, no-synergistic cytotoxic effect was observed by the
co-administration of bortezomib with aspirin [22] and with other COX-1 inhibitors as in the case of
1
and 22.
Cell cycle analysis. Cytofluorimetric analysis of NCI-H929 and RPMI-8226 cell cycle was
performed to evaluate the effects of bortezomib alone or in combination with COX-1 inhibitors 1,
22 and aspirin on cellular cycle. Cells were treated for 48 h with different concentrations 1, 22 or
aspirin, chosen based on their IC or EC values from cytostatic effect. In NCI-H929, bortezomib
5
0
50
alone displayed 10 % more cells in G0/G1 and S cell cycle phases compared to control cells, while
bortezomib in combination with COX-1 inhibitors showed a higher alteration of the normal phases
of the cycle (15 % more cells were registered in G0/G1 and S cell cycle phases) compared to
control and bortezomib treated cells (Figure 10).

ACCEPTED MANUSCRIPT
Figure 10. Percentage (%) of NCI-H929 cells in G1, S, and G2 /M phases of the cell cycle during
continuous exposure to different compounds concentrations: 7.9 nM for compound 1, 42 nM for
compound 22 and 1.7 µM for aspirin (COX-1 IC values) and 2 nM for bortezomib (EC cytotoxic
5
0
50
activity). Error bars represent mean ± SD of three experiments in triplicate; one-way ANOVA
followed by Bonferroni’s post-hoc comparison test: P>0.05
Interestingly, in RPMI-8226 cells no difference was detected between treated cells and control cells
(data not shown), while by treating NCI-H929 with bortezomib, aspirin, 1 and 22 most of the cells
are in phase G2/M like the control, while co-administration of bortezomib with all COX-1
inhibitors seems to have an effect on the cell cycle. In fact, a reduction in the number of cells in G2
phase in favor of G0/G1 and S phases is detected.
Apoptosis. The apoptotic effect exerted by bortezomib alone or in combination with COX-1
inhibitors were determined using Annexin V/PI in NCI-H929 and RPMI-8226 cells. In NCI-H929
no apoptosis was detected after treatment with COX-1 inhibitors (Annexin V/PI negative), while 50
%
of cells were registered in an early apoptotic phase after treatment with bortezomib (Annexin V

ACCEPTED MANUSCRIPT
positive/PI negative). There is a significant shift of cells treated with bortezomib in combination
with compound 1, 22 or aspirin from Annexin V positive/PI negative quadrant (early apoptotic
cells) compared to control cells (Figure 11).

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Figure 11. Flow cytometry dot plot of NCI-H929 cells treated with the different tested compounds
for 48h at different concentrations: 7.9, 42, 1.7 µM and 2 nM for 1, 22, aspirin and bortezomib,
respectively. The image is representative of one out of three experiments.
In NCI-H929 cell line, cells in apoptotic phase are 49 % when treated with bortezomib. Their
percentage increases to 64, 61 and 56 when co-administered with 1, 22 and aspirin, respectively.
In RPMI-8226 cells with bortezomib alone or in combination with COX inhibitors displayed an
apoptotic effect not so different from the control cells, with only 10 % of cells detected in an early
apoptotic phase (Annexin V positive/PI negative; Figure 12).
These results encourage to deepen the role of COX-1 inhibition in apoptosis, since a marked
difference is observed when treating RPMI-8226 or NCI-H929 cell line.

ACCEPTED MANUSCRIPT