Synthesis, characterization and interaction mechanism of new oxadiazolo-phthalimides as peripheral analgesics. IV

Article abstract of DOI:10.1016/S0022-2860(03)00418-6

The synthesis, characterization and spectroscopic studies of compounds 6a-g with analgesic activity is described. A new model of interaction between the drug and the enzyme is suggested. Application of the Resonance Valence Bond theory led us to propose, for the first time, an entirely new mechanism involving an electron transfer from the amino acid residue of the enzyme to the drug. Theoretical studies of various transition states involved in the interaction mechanism employing the semi-empirical molecular orbital calculations (AM1 method) have been carried out. This article also deals with an extensive study of the structure-activity relationships of seven oxadiazolo-phthalimides 6a-g.

Full text of DOI:10.1016/S0022-2860(03)00418-6

Journal of Molecular Structure 660 (2003) 1–13
www.elsevier.com/locate/molstruc
Synthesis, characterization and interaction mechanism of new
oxadiazolo-phthalimides as peripheral analgesics. IVq
a
a,
Roberto Antunesa,1,2, Hildson Batista , Rajendra M. Srivastava ,
*
b
George Thomas , Clidenor C. Araujo , Ricardo L. Longo , Helio Magalha˜es ,
b
a
c
´
Marcelo B.C. Leao , Antonio C. Pavao
´
a2
a
ˆ
˜
˜
a
´
Departamento de Quımica Fundamental, CCEN, Universidade Federal de Pernambuco, 50740-540 Recife, PE, Brazil
b
´
ˆ
´
Laboratorio de Tecnologia Farmaceutica, Universidade Federal da Paraıba, Joa˜o Pessoa, PB, Brazil
cDepartamento de Engenharia de Sistemas, Universidade Federal de Pernambuco, 50740-540 Recife, PE, Brazil
Received 5 December 2002; revised 2 June 2003; accepted 10 June 2003
Abstract
The synthesis, characterization and spectroscopic studies of compounds 6a–g with analgesic activity is described. A new
model of interaction between the drug and the enzyme is suggested. Application of the Resonance Valence Bond theory led us
to propose, for the first time, an entirely new mechanism involving an electron transfer from the amino acid residue of the
enzyme to the drug. Theoretical studies of various transition states involved in the interaction mechanism employing the semi-
empirical molecular orbital calculations (AM1 method) have been carried out. This article also deals with an extensive study of
the structure–activity relationships of seven oxadiazolo-phthalimides 6a–g.
q 2003 Elsevier B.V. All rights reserved.
Keywords: Phthalimide derivatives; 1,2,4-Oxadiazoles; Analgesic; Antiinflammatory
1. Introduction
analgesic drugs are administered, which can be
classified as narcotic, when acting on the central
nervous system, and non-narcotic or peripheral, when
acting directly on the damaged tissue [1]. Peripheral
analgesics are drugs that, independent of their mode
of action, show a therapeutic effect by supressing the
production of prostaglandins, which are lipids respon-
sible for inflammation and pain. Non-steroidal anti-
inflammatory drugs (NSAIDs) [2] are examples of
peripheral analgesics, since they block both isoforms
COX-1 and COX-2 of the enzyme cyclooxygenase
(COX). Acetylsalicylic acid (aspirin), for instance,
binds irreversibly to the COX enzyme via an
acetylation reaction [3,4], such that the production
Signaling molecules produced in the vicinity of the
damaged tissues usually cause the inflammation
process. To control pain due to inflammation,
q For analgesic activity of phthalimide derivatives, see part II,
Ref. [14].
*
Corresponding author. Tel.: þ55-81-3271-8443; fax: þ55-81-
3271-8442.
E-mail address: rms@ufpe.br (R.M. Srivastava).
Taken in part from the Ph.D. thesis of R. Antunes, 1998,
1
Universidade Federal de Pernambuco, Recife.
2
´
Present Address: Departamento de Quımica, Universidade
Federal Rural de Pernambuco, Recife.
0022-2860/$ - see front matter q 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0022-2860(03)00418-6

2
R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
Fig. 1. Seven new oxadiazolo-phthalimides with analgesic and antiinflammatory properties.
of pain-inducing mediators that act on nerve cells is
halted. Other NSAIDs such as indomethacin or
ibuprophen produce a reversible inhibition [5],
competing with arachidonic acid for the active site
of the enzyme.
evaluated the analgesic properties of several new
oxadiazolo-phthalimides (Fig. 1, 6a–g), and inter-
estingly, all of them have shown significant
analgesic and antiinflammatory properties [14].
This is a strong indication that these substances
act like NSAIDs, and therefore are able to inhibit
the PGHS or COX enzyme. On the other hand, the
present compounds do not have a carboxylate
group, which is present in most of the classic
NSAIDs. Thus, we propose that these new
substances can inhibit COX-2, and the present
work describes a qualitative chemical model for the
interaction between COX-2 and oxadiazolo-phtha-
limides, based upon molecular orbital calculations.
This pain-inducing mechanism has been actively
studied and, recently, it has been shown [6] that it is
not a nerve signal but a signal transported in the
bloodstream that triggers the central nervous system
to produce molecules such as interleukin-1b and
release them into the cerebrospinal fluid. Interleu-
kin-1b causes nerve cells to start producing COX,
thereby initiating pain signal within the nervous
system. When NSAIDs are injected directly into the
cerebrospinal fluid, they inhibit COX there and
make the inflammation less painful. Accordingly,
this class of painkillers had heretofore been thought
to work exclusively outside the brain and should
now be specifically designed to penetrate into the
brain.
2. Experimental procedure
2.1. General experimental methods
Although many of the NSAIDs possess analgesic
and antiinflammatory properties, some of them
have been used for the treatment of Alzheimer’s
disease [7] and also as anti-cancer agents [8,9].
Given the importance of these drugs, we have
been dedicated in synthesizing new oxadiazole
derivatives with analgesic and antiinflammatory
activities, for instance, 3-[3-(phenyl)-1,2,4-oxadia-
zol-5-yl]propionic acid (POPA) has shown to
possess both properties [10]. We have also been
interested in phthalimide derivatives, since some
studies revealed that thalidomide was efficient in
the treatment of erythema nodosum leprosum
(ENL), a sharp manifestation of the Hansen’s
disease [11,12]. In 1996, thalidomide has been
reported to be efficient in another inflammatory
process [13]. As a result, we have synthesized and
Melting points were obtained with a Digital
Electrothermal IA 9100 equipment and are uncor-
1
rected. H (300 MHz) and 13C (75.43 MHz) NMR
spectra were recorded with a Varian Unity Plus
300 MHz spectrophotometer using an appropriate
solvent and tetramethylsilane (TMS) as an internal
1
reference. The numbering system used for H and
13C spectral assignments of compounds 6a–g is
given in Fig. 1. Infrared spectra were recorded with
a Brucker IFS 66 instrument. Sanyo domestic
microwave oven (model EM-3500B, 1350W/
2450 MHz) was used for the synthesis of compound
3. Silica gel G plates (Merck) with fluorescent
indicator PF254 were employed for thin-layer
chromatography, using chloroform/ethyl acetate
(85%/15%) and the plates was revealed under
ultraviolet light.

R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
3
2.2. Chemistry
C, 67.69; H, 4.11; N, 13.17. Found: C, 67.40; H,
4.04; N, 12.98.
2.2.1. N-[3-(Phenyl)-1,2,4-oxadiazol-5-yl-methyl]
phthalimide (6a)
2.2.3. N-[3-(m-Tolyl)-1,2,4-oxadiazol-5-yl-methyl]
phthalimide (6c)
To compound 3 (9.3 mmol) dissolved in DME
(8.0 ml) was added DCC (7.4 mmol) and an
appropriate arylamidoxime 4a (9.3 mmol) in DME
(8.0 mL) and the reaction mixture was stirred for
30 min at room temperature and then refluxed for 16 h
[15]. The solvent was removed under vacuum and the
residue chromatographed on a silica gel column using
chloroform/n-hexane (70%/30%, v/v) as eluent,
which provided 6a as a colorless solid. Chromato-
graphically pure material was crystallized and recrys-
tallized from chloroform until the melting point
became constant: mp 189–190 8C, yield 51%; IR
Crystallized from chloroform, yield 56%; mp
144–145 8C; IR (KBr): 1773, 1723, 1578, 1521,
1
1109 and 948 cm21; H NMR (CDCl3) d 7.81–7.84
(m, 2H, H-20 and H-60), 7.94 (AA0BB0 system, 2H,
J10;11 ¼ 5:2 Hz; J10;12 ¼ 3:2 Hz; H-10, H-13); 7.79
(AA0BB0 system, 2H, J11;12 ¼ 5:2 Hz; J11;13 ¼ 3:2 Hz;
H-11, H-12); 7.28–7.35 (m, 2H, H-40 and H-50), 5.17
(s, 2H, H-6 and H-6 ), 2.38 (s, 3H, CH3-Ar); 13C
NMR (CDCl3) d 173.54 (C-5), 168.65 (C-3), 166.88
(C-15), 138.52 (C-50), 134.46 (C-10), 132.07 (C-40),
131.75 (C-9), 128.63 (C-30), 127.99 (C-60), 126.00 (C-
10), 124.61 (C-20), 123.84 (C-11), 33.54 (C-6), 21.23
(CH3-Ar). Anal. calcd for C18H13N3O3 (319.30): C,
67.69; H, 4.11; N, 13.17. Found: C, 67.40; H, 4.03; N,
13.37.
(KBr) 1776, 1723, 1598, 1572, 1107 and 949 cm21
;
1H NMR (CDCl3) d 8.01–8.03 (m, 2H, H-20 and H-
60), 7.95 (AA0BB0 system, 2H, J10;11 ¼ 5:6 Hz;
J10;12 ¼ 3:2 Hz; H-10, H-13); 7.79 (AA0BB0 system,
2H, J11;12 ¼ 5:6 Hz; J11;13 ¼ 3:2 Hz; H-11, H-12)
7.41–7.48 (m, 3H, H-30, H-40 and H-50), 5.18 (s, 2H,
H-6 and H-6 ; 13C NMR (CDCl3) d 173.67 (C-5),
168.59 (C-3), 166.93 (C-15), 134.51 (C-10), 131.79
(C-9), 131.31 (C-40), 128.75 (C-30,C-50) 127.52 (C-20,
C-60) 126.19 (C-10),123.89 (C-11) 33.59 (C-6). Anal.
calcd for C17H11N3O3 (305.08): C, 66.87; H, 3.63; N,
13.77. Found: C, 66.67; H, 3.70; N, 13.74. Other 1,2,4-
oxadiazoles were prepared, purified and characterized
similarly.
2.2.4. N-[3-( p-Tolyl)-1,2,4-oxadiazol-5-yl-methyl]
phthalimide (6d)
Crystallized from chloroform, yield 55%; mp
168.0–169 8C; IR (KBr): 1772, 1724, 1594, 1570,
1
1112 and 944 cm21; H NMR (CDCl3) d 7.88–790
(m, 2H, H-20 and H-60), 7.92 (AA0BB0 system, 2H,
J10;11 ¼ 5:4 Hz; J10;12 ¼ 3:0 Hz; H-10, H-13); 7.79
(AA0BB0 system, 2H, J11;12 ¼ 5:7 Hz; J10;12 ¼ 3:0 Hz;
H-11, H-12); 7.21–7.24 (m, 2H, H-30 and H-50), 5.16
(s, 2H, H-6 and H-6 ), 2.38 (s, 3H, CH3-Ar); 13C
NMR (CDCl3) d 173.46 (C-5), 168.62 (C-3), 166.90
(C-15), 141.64 (C-40), 134.47 (C-10), 131.85 (C-9),
129.45 (C-30, C-50), 127.46 (C-20, C-60), 123.86 (C-
11), 123.45 (C-10), 33.60 (C-6), 21.52 (C-16). Anal.
calcd for C18H13N3O3·1/2 H2O (328.29): C, 65.85; H,
4.30; N, 12.79. Found: C, 65.53; H, 4.01; N, 12.57.
2.2.2. N-[3-(o-Tolyl)-1,2,4-oxadiazol-5-yl-methyl]
phthalimide (6b)
Crystallized from chloroform, yield 49%; mp
132 8C; IR (KBr): 1772, 1718, 1602, 1569, 1107
and 947 cm21; 1H NMR (CDCl3) d 7.88–7.91 (m, 1H,
H-60), 7.94 (AA0BB0 system, 2H, J10;11 ¼ 5:4 Hz;
J11;13 ¼ 3:3 Hz; H-10, H-13); 7.79 (AA0BB0 system,
2H, J11;12 ¼ 5:4 Hz; J11;13 ¼ 3:0 Hz; H-11, H-12);
7.21–7.38 (m, 3H, H-30, H-40 and H-50), 5.18 (s, 2H,
H-6 and H-6 ), 2.57 (s, 3H, CH3-Ar); 13C NMR
(CDCl3) d 172.66 (C-5), 169.13 (C-3), 166.94 (C-15),
138.26 (C-10), 134, 49 (C-11), 131.86 (C-9), 131.28
(C-30), 130.68 (C-5), 130.17 (C-60) 125.87 (C-4),
125.52 (C-20), 123.86 (C-11), 33.62 (C-6), 22.01
(CH3-Ar). Anal. calcd for C18H13N3O3 (319.30):
2.2.5. N-[3-( p-Nitrophenyl)-1,2,4-oxadiazol-5yl-
methyl]phthalimide (6e)
Crystallized from acetone, yield 20%; mp 218–
219 8C; IR (KBr): 1777, 1721, 1608, 1572, 1108 and
943 cm21
;
1H NMR (CDCl3) d 8.28 (AA0BB0
system, 4H, H-20, H-30, H-50 and H-60), 7.98
(AA0BB0 system, 2H, J10;11 ¼ 5:7 Hz; J10;12 ¼ 3:0 Â
Hz; H-10, H-13); 7.84 (AA0BB0 system, 2H, J11;12
3:2 Hz; H-11, H-12); 5.22 (s, 2H, H-6 and H-6 );
¼

4
R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
13C NMR (CDCl3) d 174.65 (C-5), 167.09 (C-3),
166.88 (C-15), 149.51 (C-40), 134.66 (C-10), 131.74
(C-10), 131.08 (C-9), 128.57 (C-20, C-60), 124.03 (C-
11), 124.00 (C-30, C-50), 33.55 (C-6). Anal. calcd for
C17H10N4O5 (350.16): C, 58.29; H, 2.88; N, 15.99.
Found: C, 58.25; H, 2.91; N, 15.81.
geometry optimizations were performed in Cartesian
coordinates without any restriction or constraints,
except in the reaction coordinate calculations. The
convergence threshold for all geometry optimization
calculations was 5 £ 1023 kJ mol21 pm21. All
extreme points in the potential energy surface with
chemical relevance (reactants, products, intermediates
and transition states) have been characterized by the
calculation of the Hessian eigenvalues [18,19]. The
ORTEP-III program [20] was used for visualization of
the molecular structures. The partition coefficient
octanol/water, in its logarithmic form, log P has been
calculated with the CLOGP program [21], which uses
the additive properties of a data base containing nearly
8,000 fragments to estimate the partition coefficient of
a given organic molecule. All statistical analyses have
been performed with the Ein*Sight program [22],
including the correlation, regression and principal
component analysis (PCA).
2.2.6. N-[3-( p-Anisyl)-1,2,4-oxadiazol-5-yl-methyl]
phthalimide (6f)
Crystallized from methanol, yield 82%; mp 157–
158 8C; IR (KBr): 1774, 1729, 1614, 1593, 1113
and 946 cm21; 1H NMR (CDCl3) d 7.97–7.98 (m, 2H,
H-30 and H-50), 7.96 (AA0BB0 system, 2H, J10;11 ¼ 5:6
Hz; J10;12 ¼ 3:3 Hz; H-10, H-13); 7.80 (AA0BB0
system, 2H, J11;12 ¼ 5:4 Hz; J11;13 ¼ 3:2 Hz; H-11,
H-12); 6.94 (tt, 2H, J ¼ 9:04 and 2.2 Hz, H-30 and H-
50), 7.97–7.98 (m, 2H, H-20 and H-60), 5.16 (s, 2H, H-6
and H-6 ), 3.85 (s, 3H, CH3O); 13C NMR d 173.33 (C-
5), 168.28 (C-3), 166.92 (C-15), 161.96 (C-40), 134.47
(C-10), 131.81 (C-9), 129.15 (C-20, C-60), 123.86 (C-
11), 118.66 (C-10), 114.13 (C-30, C-50), 55.32 (CH3O),
33.58 (C-6). Anal. calcd for C18H13N3O4 (335.17): C,
64.48: H, 3.91; N, 12.53. Found: C, 64.97; H, 3.86;
N, 12.47.
3. Results and discussion
3.1. Synthesis
Reaction of phthalic anhydride 1 with glycine 2 in
a domestic microwave oven provided N-phthaloyl-
glycine 3 in about 60% yield. The preparation of 3 by
conventional heating has been reported earlier [23].
Its microwave-induced synthesis from 1 and 2 in a
mixture of DMF and N-methylmorpholine has been
described in the literature [24a]. Tetrachlorophthali-
midoacetic acid has also been obtained from
tetrachlorophthalic anhydride and glycine by micro-
wave-assisted technique using diglyme as solvent
[24b]. However, the literature does not describe the
synthesis of phthaloylglycine from phthalic anhydride
and glycine under solvent-free conditions. Our
approach to synthesize 3 avoided the use of any
solvent, thus reducing the work-up problem and is
friendly to the environment. Reaction of 3 with
arylamidoximes 4a–g in the presence of dicyclohex-
ylcarbodiimide (DCC) at room temperature affords
intermediates 5a–g, which on heating for an extended
period of time furnishes 6a–g (Fig. 2) [14].
2.2.7. N-[3-( p-Chlorophenyl)-1,2,4-oxadiazol-5-yl-
methyl]phthalimide (6g)
Crystallized from chloroform, yield 75%;
mp 174 8C; IR (KBr): 1774, 1716, 1595, 1568, 1115
1
and 945 cm21; H NMR (CDCl3) d 7.99–8.00 (m,
2H, H-20 and H-60), 7.98 (AA0BB0 system, 2H, J10;11
¼
5:2 Hz; J10;12 ¼ 3:0 Hz; H-10, H-13); 7.81 (AA0BB0
system, 2H, J11;12 ¼ 5:4 Hz; J11;13 ¼ 3:0 Hz; H-11,
H-12); 7.42 (tt, 2H, J ¼ 8:70 and 2.10 Hz, H-30
and H-50), 5.18 (s, 2H, H-6 and H-6 ); 13C
NMR (CDCl3) d 173.91 (C-5), 167.80 (C-3), 166.89
(C-15), 137.48 (C-40), 134.55 (C-10), 131.76 (C-9),
129.09 (C-30), 128.84 (C-20, C-60), 1 24.70 (C-10),
123.91 (C-11), 33.54 (C-6). Anal. calcd for C17H10-
N3O3Cl·1/2H2O (344.20): C, 59.31; H, 3.07; N, 12.20.
Found: C, 59.44; H, 2.91; N, 12.56.
2.3. Computational and other procedures
All quantum chemical calculations were performed
with the standard AM1 [16] Hamiltonian implemented
into the MOPAC 93 (revision 2) program [17]. All
The structures of compounds 6a–g were deter-
mined with the help of infrared, NMR spectra and
elemental analyses. Further, heating of 5a–g lost water

R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
5
Fig. 2. Synthesis of new oxadiazolo-phthalimides.
and cyclized to provide 1,2,4-oxadiazoles without any
other rearranged product. It is well known that heating
of O-acylamidoximes furnish 1,2,4-oxadiazoles exclu-
sively [25a,b]. Compounds 6a–g possess interesting
conformation. This is discussed in the section dealing
with Molecular Conformation.
parameter of the substituents, p(R1) and p(R2). In
addition, a topological descriptor, IR1, indicating the
presence (1) or absence (0) of the phthaloyl group has
been considered, since this group is known to be
important for the analgesic and antiinflammatory
activities [13]. The descriptors and their calculated
values are presented in Table 3. The atomic charges of
13 compounds are compiled in Table 4.
3.2. Pharmacology
Table 1
Phthalimido-1,2,4-oxadiazoles are a new class of
compounds and therefore 6a–g, as first step, were
tested for analgesic and antiinflammatory properties
and found to possess such activities. The results are
shown in Table 1 and have been published by us
earlier [14].
Inhibition of acetic acid induced writhing in mice
Compound Dose
No. of Inhibition of ID50 (with 95%
writhing (%) confidence limit)
(mg/kg) tests
Aspirin
6a
200
100
50
3
3
3
38.2*
61.7*
74.0
80.8 (50.8–103.0)
3.3. Structure–activity relationships (SAR)
6.25
4
3
3
73.8*
67.1*
38.9*
3.12
1.56
2.2 (1.4–3.5)
3.1 (2.1–4.6)
In order to better understand the biological activity
of the newly synthesized oxadiazolo-phthalimides, we
have attempted to establish some relationships
between the activity, that is, ID50, and molecular
descriptors. A total of 14 substances have been tested
and also calculated with quantum chemical method
(AM1) and fragment (CLOGP) programs. These
substances are shown in Table 2 and are grouped
according to the substituents at positions 3 and 5 of the
oxadiazole moiety.
6b
6.25
3.12
1.56
4
3
3
80.5*
76.5*
12.1N.S.
6c
6.25
3.12
1.56
4
3
3
79.2*
67.1*
25.5*
2.9 (1.9–4.4)
7.2 (4.8–10.8)
6d
12.5
4
4
3
72.5*
6.25
3.12
62.4*
16.1N.S.
As for the descriptors, we have chosen the
following electronic and molecular properties calcu-
lated with the AM1 method: (i) lowest unoccupied
molecular orbital energy, 1LUMO; (ii) highest
occupied molecular orbital energy, 1HOMO; (iii)
total dipole moment, m; (iv) standard enthalpy of
formation, DHf; and (v) the net atomic charges, q: In
order to take into account, at least partially, the effects
of transport and migration through membranes, we
have also calculated partition coefficients for the
molecules, log P; as well as the hydrophobicity
6e
6.25
3.12
1.56
3
3
3
62.9*
4.1 (2.4–7.0)
47.4*
29.8N.S.
6f
25.0
12.5
6.25
3.12
3
3
3
3
62.6*
53.3*
41.9*
20.7N.S.
14.9 (7.5–29.8)
6g
25.0
12.5
6.25
3
3
3
59.6*
35.1N.S.
16.3N.S.
20.3 (11.3–36.5)
N.S. ¼ not significant; * ¼ significant.

6
R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
Table 2
yielded the following values:
Molecular structures of the oxadiazole derivatives 6a–g, 7–13
ID50 ¼ 22:3ð^0:72ÞPC1 2 12:9ð^3:67Þlog P
2
þ ð^0:59Þðlog PÞ þ 40:2ð^5:37Þ
with n ¼ 13; r ¼ 0:93; s ¼ 3:18 and F3:9 ¼ 18:1:
This new model has a very high statistical
significance. It also shows the importance of the
partition coefficient, which has been taken into
account according to the quadratic model of Hansch
[26]. This model also presents a normal distribution of
the residues, with the standard deviation of the
residues within the [^2] range. Despite this model
having a large statistical significance, and thus
allowing for quantitative prediction, even though the
number of molecules is limited, it does not yield direct
information about the action mechanism of these
compounds. This is due to the fact that all electronic
topological information was compacted into the PC1.
As a result, we have also performed some correlation
analysis between ID50 and electronic descriptors and
found that the net charge on the C-5 atom, qðC-5Þ;
seems to be relevant. For instance, a clustering of four
groups has been identified as follows: (12; 6a; 6b; 6c;
6d; 6e), (7; 8; 9; 10; 11), (6f; 6g), (13). This clustering
has also been observed for other descriptors, but not as
evident as for qðC-5Þ: These results suggest that the
compounds with the best analgesic activity are those
which have an electron-withdrawing group at the
position 5 of the oxadiazole moiety and a weak
electron-donating group at C-3. These correlations
might be significant for a qualitative model for
explaining the action mechanism of these compounds.
Regarding partition coefficients, it should be
emphasized that the literature [27] reports that log P
should be in the 0.7–2.0 range for an adequate
transport through biological membranes. These trends
were obtained from the studies with inhibitors of
COX. However, if we check the calculated log P
values for the present series (Table 2), we find that
the most active compounds have log P values closest
to the upper end of this optimal interval, that is, 2.0.
Substance
R1
R2
6a
6b
6c
6d
6e
6f
6g
7
Ar
Ph
Ar
o-CH3Ph
m-CH3Ph
p-CH3Ph
p-NO2Ph
p-CH3Oph
p-ClPh
Ar0
Ar
Ar
Ar
Ar
Ar
2NH2
8
Ar0
9
Ar0
Ar0
Ar0
Ar0
–NH2
10
11
12
–NHCHO
–NyCHOEt
13
-CH2CH2CO2H
Ph
A correlation matrix between the experimental
ID50 and all descriptors have been built (Table 5),
which yielded the following largest correlation
coefficients in decreasing order for each descriptor,
log P : 0.506; DHf : 0.394; m : 0.268; qðC-5Þ : 0.239;
and qðN-2Þ : 0.238. An initial multiple regression with
these descriptors yielded a poor correlation with r ¼
0:8 and F ¼ 3:11: As a result, we have performed
a PCA of all compounds except 6g on the data matrix
containing all electronic and topological descriptors
(Table 6). The PCA shows that only with three
principal components (PC1, PC2 and PC3) it is
possible to explain more than 82% of the original data.
Also, PC1 alone explains nearly 53% of the original
information, so that it would be appropriate to
consider PC1 as a new descriptor, which contains
all the electronic and topological information. As
a result, the scores of PC1 with the physicochemical
3.4. Interaction mechanism model
Although, we would like to carry out the
experimental work involving phthalimides and
COX-2 to confirm the inhibitory effect of this enzyme,
2
parameters log P and ðlog PÞ were used to construct a
new statistical model for a multiple regression, which

R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
7
Table 3
Electronics, physicochemical and topological descriptors and ID50
Compound
ID50
1LUMO (eV)
1HOMO (eV)
m (D)
DHf (kcal/mol)
IR1
log P
p(R1)
p(R2)
6a
6b
6c
6d
6e
6f
6g
7
2.2
3.1
21.24
21.25
21.21
21.23
21.49
21.22
21.28
20.34
20.54
20.73
20.88
20.65
20.59
20.61
29.45
29.38
29.35
29.19
210.21
28.94
29.41
29.12
29.25
29.27
29.20
29.02
29.00
29.56
3.13
3.38
4.51
2.85
8.99
3.30
4.44
3.40
3.17
2.97
5.47
3.99
3.57
2.32
64.72
59.20
56.07
57.26
69.16
26.48
57.75
7.40
1
1
1
1
1
1
1
0
0
0
0
0
0
0
2.70
3.20
3.20
3.20
2.45
2.74
3.42
4.72
5.81
4.72
5.06
5.34
4.61
0.90
0.50
0.50
0.50
0.50
0.50
0.50
0.50
20.42
0
2.15
2.65
2.9
2.65
7.2
2.65
4.1
1.2
14.9
20.3
1.83
2.75
4.70
5.70
8
4.10
7.6
6.3
8.2
6.9
41.20
8.60
5.70
9
5.70
5.70
5.70
5.70
20.64
20.42
20.45
20.52
20.15
2.15
10
11
12
13
225.50
215.30
9.60
29.6
226.20
it is not possible at the moment because of the lack of
the facility. It would also be interesting to carry out
computational modeling studies to determine the
proximity of the arginine residue with phthalimides.
However, this is also difficult in our Department
because no such program is available. Under the
circumstance, we suggest the following theoretical
work, which includes the electron-transfer mechanism
and the transition state calculations of the enzyme–
drug complex.
Therefore, the present model consistently considers
an electron transfer from the CyN of the enzyme to
the carbonyl carbon as shown below (Scheme 1).
The above information leads to an intermediate
state with charge separation which presumably
forms an unstable four-membered ring. Consequently
a N–C covalent bond, is formed as shown in
intermediate A. Once a covalent bond is established
between the nitrogen and carbonyl carbon atoms, the
enzyme will suffer structural modification, thus
These models are based on the calculated molecu-
lar electronic parameters which permit to define the
centers involved in the interaction of the phthalimide
moiety of the molecule with the enzyme. AM1
calculations show that these phthalimide derivatives
(6a–g) have high electron affinity (1.39–1.66 eV),
indicating their capabilities to act as electrophiles.
The low LUMO (lowest unoccupied molecular
orbital) energy (21.21 to 21.49 eV) confirms the
eletrophilic character of carbonyl carbon. On the other
hand, the active sites of both isoforms of COX contain
arginine (Arg120) [28] besides other amino acid
residues. Since arginine is an amino acid containing a
guanidine group in its side-chain, guanidine has been
selected as the appropriate electron donor. Indeed, the
calculations show that it possesses a high HOMO
(highest occupied molecular orbital) energy. The
HOMO components are distributed in pz of N1
(0.54), px of N in R2 (0.47) and py of N3 (0.35).
Table 4
Net atomic charges on the oxadiazole moiety calculated via AM1
method. See Table 2 for numbering
Substance qðO-1Þ
qðN-2Þ
qðC-3Þ
qðN-4Þ
qðC-5Þ
6a
6b
6c
6d
6e
6f
6g
7
20.073 20.032 20.039 20.150 20.012
20.075 20.025 20.043 20.143 20.019
20.080 20.032 20.036 20.150 20.016
20.075 20.034 20.038 20.144 20.017
20.065 20.016 20.058 20.153 20.003
20.073 20.039 20.034 20.144 20.019
20.070 20.028 20.044 20.147 20.019
20.088 20.098 20.013 20.187 20.088
20.100 20.053 20.010 20.203 20.133
20.106 20.059 20.014 20.212 20.125
20.086 20.086 20.044 20.198 20.109
20.093 20.011 20.007 20.211 20.091
20.088 20.043 20.032 20.155 20.020
20.080 20.031 20.037 20.160 20.013
8
9
10
11
12
13

8
R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
Table 5
Correlation matrix between ID50 and the descriptors
ID50
1LUMO 1HOMO
m
DHf
log P IR1
pR1
pR2
qðO-1Þ qðN-2Þ qðC-3Þ qðN-4Þ qðC-5Þ
ID50
1.000
1LUMO 0.108 1.000
1HOMO 0.020 0.488
1.000
0.670
0.423
0.520
0.354
0.376
0.061
0.482
0.359
0.607
0.269
0.366
m
DHf
0.267 0.455
0.394 0.735
0.505 0.531
0.120 0.933
0.185 0.305
0.069 0.117
0.165 0.789
0.177 0.556
0.115 0.778
0.163 0.690
0.239 0.654
1.000
0.256 1.000
log P
IR1
0.044 0.263 1.000
0.255 0.838 0.558 1.000
pR1
pR2
0.063 0.537 0.579 0.523 1.000
0.210 0.391 0.033 0.150 0.817 1.000
qðO-1Þ
qðN-2Þ
qðC-3Þ
qðN-4Þ
qðC-5Þ
0.398 0.546 0.735 0.816 0.546 0.095 1.000
0.164 0.418 0.467 0.522 0.141 0.237 0.454
0.530 0.428 0.662 0.657 0.273 0.164 0.845
0.069 0.608 0.747 0.804 0.557 0.156 0.862
0.156 0.490 0.820 0.729 0.457 0.013 0.868
1.000
0.317
0.455
0.597
1.000
0.741
0.734
1.000
0.947
1.000
leading to the analgesic effect. Scheme 1 shows the
possibility of the formation of final two products, C
and D, where D represents the lower enthalpy of
formation (47.03 kcal/mol). This kind of unsynchro-
nized electron transfer mechanism is being suggested
for the first time in this system. Such electron transfer
reaction in others systems has been discussed by
Pava˜o et al. [29].
bonds [30a–c], with other amino acid residues
leading to enzyme modification. However, even so,
TS1 would be favored because two hydrogen bonds
would be formed: an intermolecular and other
intramolecular. For the TS4 state it would just remain
the possibility of intermolecular bonds, and the
probability of formation of two hydrogen bonds with
the active site is very small.
To arrive to these products, two paths are proposed
with their respective transition states TS‡n ðn ¼ 1–5Þ;
as shown in Scheme 2 for the cyclic and Scheme 3 for
the acyclic ones.
On the other hand, the TS3 and TS5 states present
large atomic charge variations when compared with
the reagents and TS1 (Table 7 and Schemes 2 and 3).
The cyclic path, forming a six-atom transition state
(TS1) is preferred because of hydrogen bonding
Table 6
ꢀ
(d1 ¼ 2:04 A; Scheme 2 and Table 5). The acyclic
Scores of the first three principal components analysis
route (TS4) is unable to form the hydrogen bond
ꢀ
Compound
PC1
PC2
PC3
0.426
(d ¼ 4:06 A; Scheme 3 and Table 5). The hydrogen
bond in TS1 can also justify the lowest energy in that
state (22.1 kcal/mol) as compared to the TS5
(30.6 kcal/mol), according to Fig. 3.
6a
6b
6c
6d
6e
6f
7
2.280
0.498
2.342
1.928
0.526
0.408
0.491
0.078
1.879
0.905
1.068
In addition, it can be noted that in the cyclic path
(Scheme 2), even after the opening of the phtaloyl ring
with N1 of arginine, the TS3 presents lower energy
(28.6 kcal/mol) than the TS5 state (30.6 kcal/mol), as
it can be seen in the Fig. 3. Once again, an
intramolecular hydrogen bond seems to favor the
TS3 state.
4.430
21.871
0.551
22.478
1.379
1.447
22.442
22.486
23.371
22.025
22.796
21.092
20.092
2.420
21.206
21.430
20.146
20.788
0.820
8
2.113
9
21.136
22.119
21.608
21.114
0.426
10
11
12
13
1.432
Considering that the carbonyl oxygen of the
phthaloyl ring can form two intermolecular hydrogen
0.354

R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
9
Scheme 1. Enthalpies of reagents and products.
The atomic charges can be stabilized by saline bridge
and other electrostatic interactions with other amino
acid residues of the active site, and once again TS3
will continue being the state more indicated to arrive
to the product D. That is, besides being stabilized by
the N15 and C2 interactions with the amino acid
residues, a hydrogen intramolecular bond still exists
reaction (2.2 kcal/mol), while TS3 produces D for an
exothermic reaction (218.9 kcal/mol). Therefore, the
higher energy in TS2 could be due to the rigidity of
the five-membered ring, with those atoms in the same
plane, together with the instability of C5, which is a
tetrahedral carbon bound to three heteroatoms. Thus,
the Hammond’s postulate can be applied here, which
says that: ‘in a reaction, with the same reagents and
being arrived to two final products with different
energy, the most stable product will also have a more
stable transition state’. In that way, the path for the
proposed nucleophilic reaction between arginine and
ꢀ
ðd1 ¼ 2:02 AÞ:
In the cyclic path (Scheme 2), the TS2 presents
higher energy (31.4 kcal/mol) than TS3 (28.6 kcal/
mol), as shown in Fig. 4. It can therefore be noted that
TS2 leads to the product C through a endothermic

10
R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
Fig. 3. Relative enthalpies of the transition states.
tives. The most stable conformations are those where
the oxadiazol and phthalimide rings are almost
perpendicular. The torsion angle C(8)–N(7)–C(6)–
C(5) of 6a is 84.358 (Fig. 5). On the other hand, the
oxadiazol and the phenyl rings are almost coplanar,
the N(2)–C(3)–C(16)–C17) torsion angle being
15.098 (Fig. 5). This is also observed for the meta
and para substituents in the phenyl ring. However, the
substituent in ortho position, as was expected, makes
the phenyl ring out of the plane.
Scheme 2. Cyclic transition states.
In order to verify the conformation of such
compounds, we compared the X-ray work of the
known compound, N-[3-( p-bromophenyl)-1,2,4-oxa-
diazol-5yl]methylphthalimide. This compound was
prepared by us exactly in a similar manner as the
others of this paper and the crystallographic data of
carbonyl of the phthaloyl ring, taking into account the
opening of the ring with the formation of the product
D, goes by TS1 ! TS3 ! D.
3.5. Molecular conformation
The AM1 calculations also permit to predict the
structures and conformations of phthalimide deriva-
Table 7
Enthalpies, atomic charges ðqÞ and distances ðdÞ for the transition
states
TS1
TS2
TS3
TS4
TS5
Hf0 (Kcal/mol)
88.04 97.36 94.57 91.41 96.51
˚
d1 (A)
˚
d2 (A)
2.04
1.72
1.36
1.51
2.02
1.49
4.06
1.72
3.98
1.49
qðN1Þ
qðC2Þ
qðN3Þ
qðH4Þ
qðC5Þ
qðO6Þ
qðN15Þ
20.36 20.32 20.31 20.35 20.31
þ0.30 þ0.29 þ0.36 þ0.28 þ0.38
20.33 20.38 20.30 20.35 20.33
þ0.27 þ0.37 þ0.30 þ0.25 þ0.27
þ0.46 þ0.38 þ0.46 þ0.45 þ0.43
20.51 20.53 20.44 20.46 20.41
20.38 20.38 20.52 20.37 20.52
Relative Hf0(Kcal/mol) 22.1
31.4
28.6
25.5
30.6
Scheme 3. Acyclic transition states.

R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
11
Fig. 4. Relative enthalpies leading to C and D products in the cyclic path.
a single crystal were collected and published [32]. The
ORTEP diagram of this compound is shown in Fig. 6.
Here, the torsion angle C(5)–C(6)–N(7)–C(8) is
104.938 showing that the oxadiazole ring is somewhat
perpendicular to the phthalimide ring. The X-ray
result also shows that the phenyl and oxadiazole
rings are coplanar. Semi-empirical and ab initio
calculations of this compound showed similarity
with X-ray results [31]. In summary, the calculated
and experimental results show similar conformations
for compound N-[3-( p-bromophenyl)1,2,4-oxadiazol-
5-yl]methylpthalimide.
Classic inhibitors of COX-1 (ibuprophen, indo-
methacin, naproxen, sulindac, etc.) and even the
specific inhibitors for COX-2 [SC-581, SC-558, SC-
58635 (CELECOXIB), Meloxicam, Dup 697, etc.]
more recently studied [32–34], act through mechan-
isms which could be based on electrons transfer or
Fig. 5. The most stable conformation for the oxadiazolo-phthalimide.

12
R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
the Personnel at Higher Level (CAPES) for financial
assistance.
References
´
ˆ
[1] A. Korolkovas, J.H. Burckhater, Quımica Farmaceutica,
Guanabara Koogan, Rio de Janeiro, 1988, pp. 159–217.
[2] S.H. Ferreira, Nat. New Biol. 240 (1972) 200.
[3] J.R. Vane, Nat. New Biol. 231 (1971) 232.
[4] F.J. Van gives Ouderaa, M. Buytenhek, D.H. Nugteren, D.A.
Van Dorp, Eur. J. Biochem. 109 (1980) 1.
[5] J.R. Vane, R.J. Flower, R.M. Botting, Stroke 21 (1990) 12.
[6] T.A. Samad, K.A. Moore, A. Sapirstein, S. Billet, A.
Allchorne, S. Poole, J.V. Bonventre, C.J. Woolf, Nature 410
(2001) 471.
[7] G.M. Pasinetti, Arch. Gerontol. Geriatr. (2001) 3.
[8] W. Dempke, C. Rie, A. Grothey, H.J. Schmoll, J.Cancer Res.
Clin. Oncol. 127, p. 411, 2001.
Fig. 6. ORTEP diagram.
[9] A.S. Kalgutkar, Z. Zhao, Curr. Drug Target. 2 (2001) 79.
[10] P. Afiatpour, R.M. Srivastava, M.L. de Oliveira, E.J. Barreiro,
Braz. J. Med. Biol. 27 (1994) 1403.
through hydrogen bond interactions. It is important to
highlight that until the present moment, aspirin is the
only analgesic capable of covalently binding to the
enzyme, and therefore, capable to inhibit it in an
irreversible way. All other analgesics, including the
classical and the specific inhibitors for COX-2, act in a
reversible way, through the interactions mentioned
above.
[11] J. Sheskin, J. Clin. Pharmacol. Ther. 6 (1965) 303.
[12] C.L. Crawford, Adv. Drug React. Toxicol. Rev. 13 (1994) 177.
[13] G.W. Muller, L.G. Corral, M.G. Shire, H. Wang, A. Moreira,
G. Kaplan, D.I. Stirling, J. Med. Chem. 39 (1996) 3238.
[14] R. Antunes, H. Batista, R.M. Srivastava, G. Thomas, C.C.
´
Araujo, Bioorg. Med. Chem. Lett. 8 (1998) 3071.
[15] R. Antunes, R.M. Srivastava, Heterocycl. Commun. 2 (1996)
247.
In conclusion, we would like to say that we have
explained the characterization of seven compounds
6a–g [14]. We are also suggesting the interaction
mechanism between these compounds and the
enzyme COX-2. Also, the Resonance Valence Bond
theory led us to propose, for the first time, an entirely
new mechanism involving an electron transfer from
the amino acid residue of the enzyme to the drug. It
was also possible to calculate the transition states
involving the drug and the enzyme and we found that
TS1 has lower energy than other transition states.
Besides, we also calculated the approximate volume
of the synthesized compounds and found them to be
[16] M.J.S. Dewar, E.G. Zoebisch, E.F. Healy, J.J.P. Stewart,
J. Am. Chem. Soc. 107 (1985) 3902.
[17] J.J.P. Stewart, MOPAC 93.00 Manual, Fujitsu Limited,
Tokyo, Japan, 1993.
[18] W.J. Hehre, L. Radom, P.R. Schleyer, J.A. Pople, Ab initio
Molecular Orbital Theory, Wiley, New York, 1986.
[19] T. Clark, A Handbook of Computational Chemistry, Wiley,
New York, 1985.
[20] M.N. Burnett, C.K. Johnson, ORTEP-III, Oak Ridge National
Laboratory, Oak Ridge, TN, USA, 1996.
[21] A. Leo, Chem. Rev. 93 (1993) 1281.
[22] Ein*Sight 3.0 Infometrix, 2200 Sixth Avenue, Suite 833,
Seattle, WA 98121.
[23] J.H. Bilman, W.F. Harting, J. Am. Chem. Soc. 70 (1948)
1473.
[24] (a) S.S. Bari, A.K. Bose, A.G. Chaudhary, M.S. Manhas, V.S.
Raju, E.W. Robb, J. Chem. Educ. 69 (1992) 938. (b) A.K.
Bose, M. Jayaraman, A. Okawa, S.S. Bari, E.W. Robb, M.S.
Manhas, Tetrahedron Lett. 37 (1996) 6989.
3
˚
385 A , which makes it possible to fit in the active site
cavity of COX-2 cavity which has a volume [29] of
3
˚
394 A .
[25] (a) R.M. Srivastava, F.J.S. Oliveira, D.S. Machado, R.M.
Souto-Maior, Synth. Commun. 29 (1999) 1437. (b) L.B.
Clapp, A.R. Katritzky,Advanced in Heterocyclic Chemistry
vol. 20Academic PressNew York (1976) 65–116.
[26] C. Hansch, T. Fujita, J. Am.Chem. Soc. 86 (1964) 1616.
[27] P. Moser, A. Sallmann, I. Wiesenberg, J. Med. Chem. 33
(1990) 2358.
Acknowledgements
We are grateful to the Brazilian National research
Council (CNPq), and to Program for Improvement of

R. Antunes et al. / Journal of Molecular Structure 660 (2003) 1–13
13
[28] R.G. Kurumbail, M.A. Stevens, J. Gierse, J.J. McDonald, R.A.
Stegeman, J.Y. Pak, D. Gildehans, J.M. Miyshiro, T.O.
Penning, K. Seibert, P.C. Isackson, W.C. Stallings, Nature
384 (1996) 644.
Soc. 109 (1987) 3856. (c) W.L. Jorgensen, J. Am. Chem. Soc.
111 (1989) 3770.
[31] H. Batista, G.B. Carpenter, R.M. Srivastava, J. Chem.
Cristallog. 30 (2000) 131.
[29] A.C. Pava˜o, C.A. Taft, T.C.F. Guimara˜es, M.B.C. Lea˜o, J.R.
Mohallem, W.A. Lester Jr., J. Phys. Chem. A 105 (2001) 5 and
the references cited therein.
[32] D. Picot, P.J. Loll, R.M. Garavito, Nature 367 (1994) 243.
[33] P.J.Loll,D.Picot,R.M.Garavito,Nat.Struct.Biol.2(1995)637.
[34] L.S. Simon, F.L. Lanza, P.E. Lipsky, R.C. Hubbard, S.
Talwalker, B.D. Schwartz, P.C. Isackson, G.S. Geis, Arthritis
Rheum. 41 (1998) 1591.
[30] (a) J.D. Mature, W.L. Jorgensen, J. Am. Chem. Soc. 108
(1986) 2517. (b) J.F. Blake, W.L. Jorgensen, J. Am. Chem.