Anti-T. cruzi activities and QSAR studies of 3-arylquinoxaline-2- carbonitrile di-N-oxides

Article abstract of DOI:10.1016/j.bmcl.2010.06.101

In a continuing effort to identify new active compounds for combating Chagas disease and other neglected diseases, our research group synthesized and evaluated 23 3-arylquinoxaline-2-carbonitrile di-N-oxides against Trypanosoma cruzi. Five of them presented IC₅₀ values of the same magnitude as the standard drug Nifurtimox, making them valid as new lead compounds. The optimized molecular structures of 23 derivatives represented by 1497 types of DRAGON descriptors were subjected to linear regression analysis, and the derived QSAR was shown to be predictive. In this way, we achieved a rational guide for the proposal of new candidate structures whose activities still remain unknown.

Full text of DOI:10.1016/j.bmcl.2010.06.101

Bioorganic & Medicinal Chemistry Letters 20 (2010) 4831–4835
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Anti-T. cruzi activities and QSAR studies of 3-arylquinoxaline-2-carbonitrile
di-N-oxides
b
c
b
c
Esther Vicente a,b, , Pablo R. Duchowicz , Diego Benítez , Eduardo A. Castro , Hugo Cerecetto ,
*
Mercedes González c, Antonio Monge a
a Unidad de Investigación y Desarrollo de Medicamentos, Centro de Investigación en Farmacobiología Aplicada (CIFA), University of Navarra, C/Irunlarrea s/n, 31008 Pamplona, Spain
b Instituto de Investigaciones Fisicoquímicas Teóricas y Aplicadas, INIFTA (UNLP, CCT La Plata-CONICET), Diag. 113 y 64, C.C. 16, Suc. 4, 1900 La Plata, Argentina
c Laboratorio de Química Orgánica, Facultad de Ciencias-Facultad de Química, Universidad de la República, Iguá 4225, 11400 Montevideo, Uruguay
a r t i c l e i n f o
a b s t r a c t
Article history:
In a continuing effort to identify new active compounds for combating Chagas disease and other
neglected diseases, our research group synthesized and evaluated 23 3-arylquinoxaline-2-carbonitrile
di-N-oxides against Trypanosoma cruzi. Five of them presented IC50 values of the same magnitude as
the standard drug Nifurtimox, making them valid as new lead compounds. The optimized molecular
structures of 23 derivatives represented by 1497 types of DRAGON descriptors were subjected to linear
regression analysis, and the derived QSAR was shown to be predictive. In this way, we achieved a rational
guide for the proposal of new candidate structures whose activities still remain unknown.
Ó 2010 Elsevier Ltd. All rights reserved.
Received 1 May 2010
Revised 17 June 2010
Accepted 19 June 2010
Available online 25 June 2010
Keywords:
Anti-trypanosomal
Chagas disease
Multivariable linear regression
Quinoxaline N-oxide
QSAR
Replacement method
Chagas disease, or American trypanosomiasis, is one of the most
important parasitic diseases in Latin America, mainly, in rural areas
where poverty is widespread. It is caused by the haemoflagellate
protozoan Trypanosoma cruzi, which is transmitted to humans by
blood-sucking reduviid bugs (Triatoma infestans and Triatoma
rubrovaria) that deposit their infective feces on the skin at the time
of biting. Cases have also been reported in the USA and Canada, as a
result of transfusion-related infection. Currently, there are an esti-
mated 16–18 million persons infected by T. cruzi. Approximately
2–3 million individuals develop the typical symptoms of Chagas
disease and some 100 million persons (25% of the Latin American
population) are at risk of acquiring this infection.1 It is estimated
that 14,000 people die from the disease every year; this number
of deaths in Latin America is greater than that of any other para-
site-born disease, including malaria.2
This disease represents a serious public health problem in the
countries and areas where it is endemic (21 countries in Central
and South America) because, in addition to the fact that no effec-
tive methods of immunoprophylaxis currently exist, chemother-
apy treatment for controlling this parasitic infection remains
undeveloped.1 Nifurtimox and Benznidazol are drugs that are
commonly used to treat this disease; however, these compounds
produce several adverse reactions and are not effective in the
chronic phase of the disease. Therefore, the design, synthesis, and
biological evaluation of new compounds with potential activity
against T. cruzi is considered to be of great importance.3
Chagas disease is included in the collective ‘neglected diseases’
because it principally affects poor people in developing countries
for which health interventions (as well as drug research and devel-
opment) are inadequate to the existing needs.4–6 In order to be
useful worldwide, antichagasic drugs must be inexpensive so that
they are routinely available to populations in need in developing
countries. Therefore, our research group carried out a search for
inexpensive, available reagents so as to be able to prepare new
and inexpensive anti-T. cruzi drug candidates.
Quinoxalines and their mono- and di-N-oxide derivatives dis-
play a broad range of biological activities,7 and quinoxaline di-N-
oxides are known to undergo bioreductions under hypoxia, causing
DNA damage.8–11 Given the known activity of other classes of bior-
eductive agents, such as Nifurtimox,12 it seemed logical to us to
evaluate our group of structures against T. cruzi. As a result of
the anti-Chagas research project, our group published several arti-
cles in which the synthesis and biological evaluation of a large
amount of quinoxaline and quinoxaline di-N-oxides have been de-
scribed. From these studies, several derivatives, with different pat-
terns of substituents at the quinoxaline nucleus, were prepared
with outstanding in vitro anti-trypanosomal activity (i.e., lead
* Corresponding author. Tel.: +34 948425653; fax: +34 948425652 (E.V.).
E-mail addresses: estherviccem@gmail.com, cifa@unav.es (E. Vicente).
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.06.101

4832
E. Vicente et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4831–4835
compounds I–IV, Fig. 1).13–18 Screening of the in vitro anti-trypan-
osomal activity of quinoxaline-2-carbonitrile di-N-oxide deriva-
tives14,18 indicated that their potency mainly depended on the
substituents in R6/R7 positions, with 6/7-halogenated derivatives
being the most active compounds (I–IV, Table 1), and on the com-
pounds lipophilicity, with the most active derivatives being the
less hydrophilic (Table 1). Recently, we have theoretically demon-
strated that quinoxaline di-N-oxide could be bioreduced into the
parasite, thereby producing reactive species which cause death of
the parasite.19
In a continuing effort to identify new active compounds to com-
bat Chagas disease, and other neglected diseases,20–22 our research
group prepared 23 3-arylquinoxaline di-N-oxide derivatives, 1–23.
Several structural modifications were introduced at the lead com-
pounds I–IV, by applying the isosteric and homologous strategies
(Fig. 1) attempting to increase the compounds lipophilic proper-
ties. Based on the above, we proposed, as a first approximation,
the elimination of nonaromatic moiety linked to C-3 of the quinox-
aline scaffold (piperazinyl ring in I–II or carbonylamino group in
III–IV), keeping the carbonitrile group linked to C-2, in order to ob-
tain a series of 3-arylquinoxaline-2-carbonitrile di-N-oxides, 1–23.
Variations on the electronic profile of the W para-substituent of the
phenyl moiety were carried out (1–12). On the other hand, the hal-
ogen atoms present in the prototypes I–IV were isosterically re-
placed by other monovalent groups. These modifications were
conducted in order to establish the contributions of electronic
and steric parameters for the optimization of the previous lead
compounds, I–IV.
Table 1
In vitro activity of previous lead compounds, I–IVa,b against epimastigote Tulahuen 2
strain of T. cruzi
PGIc,d (%)
IC50
(lM)
MLOGPf
e
Compound
I
II
III
IV
100
94
53.1
92.4
6.5
6.7
19.2
10.8
1.912
2.987
0.868
1.616
a
b
c
d
e
f
For structure see Figure. 1.
From Refs. 8,12.
Percentage of growth inhibition respect to untreated parasite.
Inhibition of epimastigotes growth of Tulahuen 2 strain, doses = 25
Fifty percentage inhibitory concentrations (
Moriguchi-octanol/water partition coefficient.
l
M.
lM).
correlation was observed between the anti-proliferative epimas-
tigote activity and the in vivo anti-T. cruzi activity.27–30 Table 2
shows the results expressed as percentage of epimastigotes T.
cruzi, Tulahuen 2 strain, growth inhibition (PGI) at 25 lM doses
and the corresponding IC50 values. As shown in Table 2, almost
all of the compounds were active, with PGI >50%, in the preli-
minary assay at 25 lM. Five of these derivatives (8, 11, 12, 21,
and 23) displayed IC50 values of the same order than the
standard drug Nifurtimox (Nfx), when tested in vitro against epi-
mastigote forms of T. cruzi, making them new, valid lead
compounds.
Some interesting structure–activity relationships can be ob-
served from these results. The most lipophilic compound, 8 (Table
2), according to Moriguchi-octanol/water partition coefficient,31
was one of the best anti-T. cruzi agents. However, in general, the
elimination of piperazinyl ring linked to C-3 of the quinoxaline
subunit, present in the prototypes I and II, leads to compounds
which do not improve the anti-trypanosomal activity. On the other
hand, the elimination of carbonylamino group linked to C-3 of the
quinoxaline scaffold, present in the prototypes III and IV, leads to
the most active derivative of both of these series of compounds
(21, Table 2).
The 3-arylquinoxaline-2-carbonitrile di-N-oxides were ob-
tained by the classical Beirut reaction.23 The aforementioned com-
pounds (1–23) were prepared, in good to excellent yields,
according to the synthetic process previously reported.20,21 Struc-
tures of the compounds were confirmed by infrared spectroscopy,
nuclear magnetic resonance and mass spectrometry; purity was
established by elemental analysis (Scheme 1).
Anti-Trypanosoma cruzi activities were determined according
to the previously described method.14,18,24 The existence of the
epimastigote form of T. cruzi as an obligate mammalian intracel-
lular stage has been revisited and confirmed.25,26 Moreover, good
With the aim of completing the analysis of these biological re-
sults, we resorted to the quantitative structure–activity relation-
O
N+
O
N+
N
N
R7
Cl
N+
O
N
N+
O
Z
N
Y
W
I: Y=OCH3;
1-12
II: Y=H ;
Z=CF3
Designed compounds
O
N+
O
N+
N
N
R7
F
F
X
N+
O
NH
N+
O
X
O
III: X=O
IV: X=S
13-23
Figure 1. Design of 3-arylquinoxaline-2-carbonitrile di-N-oxides, 1–23, as anti-trypanosomal drugs, obtained from previous lead compounds with structural
modifications.14,18

E. Vicente et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4831–4835
4833
O
N+
O
N+
N
R7
R7
R6
O
Et3N
N
+
O
N+
O
Ar
CH2Cl2
Ar
R6
N
Scheme 1. Synthetic route of 3-arylquinoxaline-2-carbonitrile di-N-oxides.20,21
Table 2
Structures and in vitro activity of evaluated 3-arylquinoxaline-2-carbonitrile di-N-oxides, 1–23 against T. cruzi
O
N+
R7
CN
N+
Ar
O
R7
Ar
PGIa,b,c (%)
IC50
15.4
(
lM)
MLOGPe
c,d
Compound
1
F
Phenyl
89.4
100
82.8
63.8
100
84.2
85.4
100
100
80.2
100
100
46.3
83.4
64.5
68.5
100
33.4
100
100
100
74.3
100
100
2.230
2.716
2.230
2.622
3.103
1.972
2.744
3.219
1.944
2.331
2.790
1.688
0.582
0.983
1.113
0.845
1.495
0.354
1.354
1.755
1.885
1.617
2.267
ꢀ0.746
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Nfxf
CF3
H
F
CF3
OCH3
F
CF3
H
Phenyl
15.1
15.2
14.4
14.9
14.6
13.4
9.2
14.5
13.1
9.2
40-Fluorophenyl
40-Fluorophenyl
40-Fluorophenyl
40-Fluorophenyl
40-Chlorophenyl
40-Chlorophenyl
40-Trifluoromethoxyphenyl
40-Trifluoromethoxyphenyl
40-Trifluoromethoxyphenyl
40-Trifluoromethoxyphenyl
20-Furyl
F
CF3
OCH3
H
F
Cl
CH3
CF3
OCH3
H
F
Cl
8.2
25.0
14.9
18.9
18.1
12.5
>25.0
12.7
13.0
7.1
20-Furyl
20-Furyl
20-Furyl
20-Furyl
20-Furyl
20-Thienyl
20-Thienyl
20-Thienyl
CH3
CF3
—
20-Thienyl
14.8
7.8
7.7
20-Thienyl
—
a
b
c
d
e
f
Percentage of growth inhibition with respect to untreated parasite.
Inhibition of epimastigotes growth of Tulahuen 2 strain, doses = 25
The results are the means of three different experiments with a SD less than 10% in all cases.
lM.
Fifty percentage inhibitory concentrations ( M).
l
Moriguchi-octanol/water partition coefficient.
Nifurtimox.
ships (QSAR) theory for elucidating the regularities present in
data.32,33 The ultimate role of this theory is the proposal of a model
capable of estimating the activities of compounds by relying on
the assumption that these resulting effects are a consequence of
the molecular structure. In our study, we further explored the
biological profile exhibited by 3-arylquinoxaline-2-carbonitrile
di-N-oxides, establishing predictive models based on the current
experimental data available. We started by analyzing the effect
of lipophilicity values on the activity of compounds by means of
QSAR models, expressing the lipophilicity contribution with the
Moriguchi-octanol/water partition coefficient ( MLOGP).31 Next,
the molecular structure of the studied 3-arylquinoxaline-2-carbo-
nitrile di-N-oxide derivatives, 1–23, was appropriately represented
by 1497 theoretical descriptors, calculated with DRAGON soft-
ware.34 These molecular descriptors included definitions of all
types, capturing the constitutional, topological, geometrical or
electronic aspects of the molecular structures being considered,
which were obtained through mathematical formulae obtained
from several theories, such as the Chemical Graph Theory, Informa-
tion Theory, Quantum Mechanics, etc.35–37 The best structure–
activity models were searched by means of the replacement meth-
od (RM) variable subset selection technique, an algorithm based on
multivariable linear regression analysis.38,39
In each reported structure–activity relationship, we developed a
model for a set of calibration compounds belonging to the training
set, and after that, we corroborated its predictive power by means
of predicting the activities on external molecules that were not
considered during the fitting of the model and which constitute
the test set. Selection of the members for both sets was carried
out in such a way that all of the members similar structural
characteristics. This table also includes the predictions achieved
by Eqs. 1–4, shown below.
We began the QSAR analysis by establishing a relationship be-
tween PGI and the lipophilicity of the quinoxaline derivatives, ex-
pressed by the Moriguchi-octanol/water partition coefficient
(MLOGP). The following linear regression was found:

4834
E. Vicente et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4831–4835
log10PGI ¼ 1:691ðꢁ0:07Þ þ 0:109ðꢁ0:03Þ ꢂ MLOGP
N ¼ 16; d ¼ 1; N=d ¼ 16; R ¼ 0:662;
ð1Þ
This QSAR had predictive capability and acceptably predicted
the test set compounds (see Table 3). RDF080u is the Radial Distri-
bution Function 8.0/unweighted, while G1e is the first component
symmetry directional WHIM index/weighted by atomic Sanderson
electronegativities.44 Both of them are highly conformational
dependent structural attributes and are suitable for complex-
behaving properties because they take into account the 3D-
arrangement of the atoms without ambiguities (as those appearing
when using chemical graphs), and also because they do not depend
on the molecular size, thus being applicable to a large series of
molecules posing great structural variance; the descriptors are
common characteristics to all of the molecules modeled. The Radial
Distribution Function RDF080u is a kind of molecular descriptor
defined for an ensemble of atoms, and may be interpreted as the
probability distribution for finding an atom in a spherical volume
of certain radius, incorporating different types of atomic properties
in order to differentiate the nature and contribution of atoms to the
property being modeled. In the case of RDF080u, the sphere radius
is 8.0 Å and no atomic property is used, thus solely characterizing
the molecular size. The WHIM descriptor G1e is calculated based
on statistical indices of the atoms projected onto three principal
axes (principal components) obtained from weighted covariance
matrices of the atomic coordinates. The weights used are again
the atomic properties that enable differentiation among atoms.
The aim of defining such indices is to capture 3D-information
regarding size, shape, symmetry, and atom distributions with re-
spect to invariant reference frames. Similarly, we have previously
stated that electronic characteristic of various groups of N-oxide
containing heterocycles were fundamental for explaining the
anti-T. cruzi activity.19
S ¼ 0:107; FIT ¼ 0:643; p < 10ꢀ5
range in PGI : 33:4—100%; outliersð> 2SÞ : 0
Rloo ¼ 0:465; Sloo ¼ 0:129; Srand < S;
N
¼ 7;
R
¼ 0:610;
S
val
¼ 0:097
v
al
v
al
In this equation, N corresponds to the number of compounds, R
is the coefficient of correlation, S stands for the model’s standard
deviation of calibration, FIT is the Kubinyi parameter,40 p is the sig-
nificance of the model, subindex loo stands for the Leave-One-Out
Cross Validation technique,41,42 Srand represents the standard devi-
ation obtained according to the Y-Randomization technique,43 and
val subindex corresponds to the external validation test set.
Whether taking into account calibration or considering validation
points, it is clearly observed that Eq. 1 was unable to describe
the PGI effect of quinoxaline compounds. It is clearly appreciated
that Eq. 1 was unable to describe the PGI effect of quinoxaline com-
pounds, either from the calibration or the validation points of
views. Therefore, as a next step, we searched the best one-, two-,
and three-variable linear regression models by using the replace-
ment method (RM) approach, which was able to select the most
relevant and representative structural descriptors for the training
set. This exploration led to the following two-variable model:
log10PGI ¼ 5:9ðꢁ0:5Þ ꢀ 0:040ðꢁ0:009Þ ꢂ RDF080u ꢀ 22ðꢁ3Þ ꢂ G1e ð2Þ
N ¼ 16; d ¼ 2; N=d ¼ 8; R ¼ 0:898;
S ¼ 0:065; FIT ¼ 2:720; p < 10ꢀ5
range in PGI : 33:4—100%; outliersð> 2:5SÞ : 0
Rloo ¼ 0:853; Sloo ¼ 0:086; Srand ¼ 0:079; Rmax ¼ 0:497
We continued the QSAR study by establishing a link between
IC50 values and the MLOGP partitioning coefficients. The following
linear regression was found:
N
¼ 7;
R
v
¼ 0:479;
S
val
¼ 0:083
v
al
al
where Rmax is the maximal intercorrelation between the descriptors
participating in Eq. 2, revealing that there is no serious duplication
of information of the descriptors.
log10IC50 ¼ 1:4ðꢁ0:1Þ ꢀ 0:14ðꢁ0:05Þ ꢂ MLOGP
ð3Þ
N ¼ 16; d ¼ 1; N=d ¼ 16; R ¼ 0:604; S ¼ 0:157;
FIT ¼ 0:472; p < 10ꢀ5
range in IC50 : 7:8—25; outliersð> 2:5SÞ : 0
Rloo ¼ 0:351; Sloo ¼ 0:086; Srand < S;
Table 3
Predicted biological data of 3-arylquinoxaline-2-carbonitrile di-N-oxides, 1–23,
according to QSAR equations (1)–(4)
N
¼ 7;
R
v
¼ 0:495;
S
val
¼ 0:156
v
al
al
PGIa (%)
PGIb (%)
IC50
(
lM)
IC50 (lM)
c
d
Compound
Eq. 3 also had poor quality on this quinoxaline data set. We also
1*
2
3
85.9
97.0
85.9
94.7
106.9
80.5
97.7
110.0
79.9
88.1
98.8
75.0
56.8
62.8
64.9
60.7
71.4
53.7
68.9
76.2
78.7
73.6
86.7
82.2
106.9
92.4
77.9
101.4
105.8
82.0
106.7
88.7
74.9
92.5
91.9
64.0
71.8
72.1
69.8
80.7
30.8
85.0
82.2
86.7
79.6
96.8
13.1
11.3
13.1
11.6
10.0
14.3
11.2
9.6
14.4
12.7
11.0
15.6
22.0
19.4
18.7
20.3
16.6
23.7
17.3
15.3
14.6
15.9
13.0
18.1
13.0
18.1
15.3
12.1
14.3
13.2
9.9
searched the best linear regressions through the RM in the pool of
1497 descriptors. The predictive linear QSAR reported in Eq. 4 gi-
ven by two descriptors was able to capture the essential structural
features of the quinoxaline di-N-oxide derivatives that related to
their anti-trypanosomal effect.
4
5
6*
7
8*
9*
10
11
12
13
14
15
16*
17*
18
19
20
21*
22
23
log10IC50 ¼ ꢀ0:9ðꢁ0:7Þ ꢀ 4:5ðꢁ1Þ ꢂ SIC0 þ 22ðꢁ3Þ ꢂ G1e
N ¼ 16; d ¼ 2; N=d ¼ 8; R ¼ 0:924; S ¼ 0:078;
ð4Þ
10.8
10.2
9.2
FIT ¼ 3:816; p < 10ꢀ5
9.8
range in IC50 : 7:8—25; outliersð> 2SÞ : 0
Rloo ¼ 0:888; Sloo ¼ 0:094; Srand ¼ 0:080; Rmax ¼ 0:097
21.8
15.3
15.3
20.9
11.5
47.4
17.3
12.2
12.2
17.2
9.4
N
¼ 7;
R
v
¼ 0:646;
S
val
¼ 0:144
v
al
al
Again, this QSAR resulted capable of predicting quinoxalines
that had not been contemplated during the model development.
The SIC0 topological descriptor corresponds to the structural infor-
mation content (neighborhood symmetry of 0-order).44 It mea-
sures the complexity of a compound as the diversity of elements
present in its molecular structure, such as atoms, bonds, cycles,
etc. In this case, SIC0 descriptor expresses the zero-order neighbor-
hood symmetry for all of the vertexes in the chemical graph.
Models from the QSAR Theory enable us to predict the activity for
substances that have yet not been tested due to the fact that they are
a
b
c
PGI predicted according to Eq. 1.
PGI predicted according to Eq. 2.
IC50 predicted according to Eq. 3.
IC50 predicted according to Eq. 4.
d
*
Data used as test set.

E. Vicente et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4831–4835
4835
4. Trouiller, P.; Olliaro, P.; Torreele, E.; Orbinski, J.; Laing, R.; Ford, N. Lancet 2002,
359, 2188.
5. Caines, K. In GHP Study Paper 4: Global Health Partnerships and Neglected
Diseases; DFID Health Resource: London, UK, 2004.
O
N+
N
R7
R6
6. Croft, S. L. Curr. Opin. Investig. Drugs 2007, 8, 103.
N+
O
7. Carta, A.; Corona, P.; Loriga, M. Curr. Med. Chem. 2005, 12, 2259.
8. Chowdhury, G.; Kotandeniya, D.; Daniels, J. S.; Barnes, C. L.; Gates, K. S. Chem.
Res. Toxicol. 2004, 17, 1399.
9. Ganley, B.; Chowdhury, G.; Bhansali, J.; Daniels, J. S.; Gates, K. S. Bioorg. Med.
Chem. 2001, 9, 2395.
OCF3
10. Azqueta, A.; Arbillaga, L.; Pachon, G.; Cascante, M.; Creppy, E. E.; Lopez de
Cerain, A. Chem. Biol. Interact. 2007, 168, 95.
11. Azqueta, A.; Pachon, G.; Cascante, M.; Creppy, E. E.; de Cerain, A. L. Mutagenesis
2005, 20, 165.
12. Boiani, M.; Piacenza, L.; Hernández, P.; Boiani, L.; Cerecetto, H.; González, M.;
Denicola, A. Biochem. Pharmacol. 2010, 79, 1736.
46, R6=H; R7=Cl, predicted IC50=8.3 µM
64, R6=F; R7=F, predicted IC50=9.9 µM
65, R6=Cl; R7=Cl, predicted IC50=7.2 µM
13. Cerecetto, H.; Di Maio, R.; González, M.; Risso, M.; Saenz, P.; Seoane, G.;
Denicola, A.; Peluffo, G.; Quijano, C.; Olea-Azar, C. J. Med. Chem. 1999, 42, 1941.
14. Aguirre, G.; Cerecetto, H.; Di Maio, R.; González, M.; Alfaro, M. E. M.; Jaso, A.;
Zarranz, B.; Ortega, M. A.; Aldana, I.; Monge-Vega, A. Bioorg. Med. Chem. Lett.
2004, 14, 3835.
Figure 2. Structures of the three best resulting anti-trypanosomal derivatives, as
predicted by QSAR equations
2 and 4, among the 49 newly proposed 3-
arylquinoxaline-2-carbonitrile di-N-oxides.45
15. Urquiola, C.; Vieites, M.; Aguirre, G.; Marín, A.; Solano, B.; Arrambide, G.;
Noblia, P.; Lavaggi, M. L.; Torre, M. H.; González, M.; Monge, A.; Gambino, D.;
Cerecetto, H. Bioorg. Med. Chem. 2006, 14, 5503.
16. Lavaggi, M. L.; Aguirre, G.; Boiani, L.; Orelli, L.; Garcia, B.; Cerecetto, H.;
González, M. Eur. J. Med. Chem. 2008, 43, 1737.
17. Romeiro, N. C.; Aguirre, G.; Hernández, P.; González, M.; Cerecetto, H.; Aldana,
I.; Pérez-Silanes, S.; Monge, A.; Barreiro, E. J.; Lima, L. M. Bioorg. Med. Chem.
2009, 17, 641.
18. Ancizu, S.; Moreno, E.; Torres, E.; Burguete, A.; Pérez-Silanes, S.; Benítez, D.;
Villar, R.; Solano, B.; Marín, A.; Aldana, I.; Cerecetto, H.; González, M.; Monge, A.
Molecules 2009, 14, 2256.
19. Boiani, M.;Cerecetto, H.;González,M.;Gasteiger,J.J.Chem.Inf.Model.2008,48, 213.
20. Vicente, E.; Charnaud, S.; Bongard, E.; Villar, R.; Burguete, A.; Solano, B.; Ancizu,
S.; Pérez-Silanes, S.; Aldana, I.; Vivas, L.; Monge, A. Molecules 2008, 13, 69.
21. Vicente, E.; Lima, L. M.; Bongard, E.; Charnaud, S.; Villar, R.; Solano, B.;
Burguete, A.; Pérez-Silanes, S.; Aldana, I.; Vivas, L.; Monge, A. Eur. J. Med. Chem.
2008, 43, 1903.
22. Vicente, E.; Pérez-Silanes, S.; Lima, L. M.; Ancizu, S.; Burguete, A.; Solano, B.;
Villar, R.; Aldana, I.; Monge, A. Bioorg. Med. Chem. 2009, 17, 385.
23. Haddadin, M. J.; Issidorides, C. H. Heterocycles 1993, 35, 1503.
24. Cerecetto, H.; Di Maio, R.; Ibarruri, G.; Seoane, G.; Denicola, A.; Peluffo, G.;
Quijano, C.; Paulino, M. Farmaco 1998, 53, 89.
25. Almeida-de-Faria, M.; Freymuller, E.; Colli, W.; Alves, M. J. M. Exp. Parasitol.
1999, 92, 263.
26. Faucher, J. F.; Baltz, T.; Petry, K. G. Parasitol. Res. 1995, 81, 441.
27. Boiani, L.; Davies, C.; Arredondo, C.; Porcal, W.; Merlino, A.; Gerpe, A.; Boiani,
M.; Pacheco, J. P.; Basombrio, M. A.; Cerecetto, H.; Gonzalez, M. Eur. J. Med.
Chem. 2008, 43, 2229.
28. Boiani, L.; Gerpe, A.; Aran, V. J.; de Ortiz, S. T.; Serna, E.; de Bilbao, N. V.;
Sanabria, L.; Yaluff, G.; Nakayama, H.; de Arias, A. R.; Maya, J. D.; Morello, J. A.;
Cerecetto, H.; Gonzalez, M. Eur. J. Med. Chem. 2009, 44, 1034.
29. Boiani, M.; Boiani, L.; Denicola, A.; de Ortiz, S. T.; Serna, E.; de Bilbao, N. V.;
Sanabria, L.; Yaluff, G.; Nakayama, H.; de Arias, A. R.; Vega, C.; Rolan, M.;
Gomez-Barrio, A.; Cerecetto, H.; Gonzalez, M. J. Med. Chem. 2006, 49, 3215.
30. Porcal, W.; Hernandez, P.; Boiani, M.; Aguirre, G.; Boiani, L.; Chidichimo, A.;
Cazzulo, J. J.; Campillo, N. E.; Paez, J. A.; Castro, A.; Krauth-Siegel, R. L.; Davies,
C.; Basombrio, M. A.; Gonzalez, M.; Cerecetto, H. J. Med. Chem. 2007, 50, 6004.
31. Moriguchi, I.; Hirono, S.; Liu, Q.; Nakagome, I.; Matshusita, Y. Chem. Pharm. Bull.
1992, 40, 127.
32. Hansch, C.; Leo, A. In Exploring QSAR Fundamentals and Applications in Chemistry
and Biology; American Chemical Society: Washington, DC, 1995.
33. Katritzky, A. R.; Lobanov, V. S.; Karelson, M. Chem. Soc. Rev. 1995, 24, 279.
34. DRAGON 5.0 Evaluation Version. Available from: <http://www.disat.unimib.it/
chm/>.
unstable or simply because their measurement requires too much
time or is costly. In terms of economical aspects, these studies per-
mit rational use of the resources available in the laboratory, avoiding
the need to perform expensive and unnecessary experimental deter-
minations. With respect to moral aspects, QSAR applied to the field
of Medicinal Chemistry have reached a great importance in the vir-
tual screening of new compounds before their synthesis, and thus
represent an effective alternative that reduces animal testing in bio-
logical assays. Application of the QSAR equations developed in this
work now enables one to propose new candidate structures that still
do not have experimentally assigned biological data.
In view of the results achieved and in a further attempt to im-
prove the biological profile of these quinoxaline di-N-oxide, we ap-
plied the QSAR models that we developed in order to estimate the
anti-trypanosomal parameters of new possible antichagasic candi-
dates that still do not present experimentally assigned biological
data. The new molecular structures were proposed on the basis
that these chemicals can be synthesized from inexpensive and
commercially available reagents.
A total of 49 different structures, 24–72, were predicted as anti-
trypanosomal candidates, using a model for each evaluated param-
eter (PGI and IC50). Only four of them showed PGIs less than 50%
and IC50 values greater than 25 lM, so they are considered to be
inactive compounds. Among the other 43 active derivatives, three
of them (46, 64, and 65) are highlighted, showing IC50 values less
than 10 lM, good enough to justify their synthesis. Figure 2 sum-
marizes the three best resulting anti-trypanosomal structures as
predicted by Eqs. 2 and 4. As can observed, these three derivatives
showed halogenated atoms at position R6 and/or R7 and the group
40-trifluoromethoxyphenyl linked to the R4 position of the quinox-
aline scaffold.45
Acknowledgments
35. Akaike, H. Information Theory and Extension of the Maximum Likelihood
Principle. In Second International Symposium on Information Theory; Petrov, B.
N., Csáki, F., Eds.; Akademiai Kiado: Budapest, 1973; p 267.
36. Bader, R. F. W. In Atoms in Molecules—A Quantum Theory; Clarendon Press:
Oxford, 1990.
37. Trinajstic, N. In Chemical Graph Theory; CRC Press: Boca Raton, FL, 1992.
38. Duchowicz, P. R.; Castro, E. A.; Fernandez, F. M.; Gonzalez, M. P. Chem. Phys.
Lett. 2005, 412, 376.
This work has been carried out with the financial support of the
RIDIMEDCHAG-CYTED. P.R.D. and E.A.C. are researchers from the
National Council of Scientific and Technological Research
(CONICET).
Supplementary data
39. Duchowicz, P. R.; Castro, E. A.; Fernández, F. M. MATCH Commun. Math. Comput.
Chem. 2006, 55, 179.
40. Kubinyi, H. Quant. Struct.-Act. Relat. 1994, 13, 393.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2010.06.101.
41. Golbraikh, A.; Tropsha, A. J. Mol. Graphics Modell. 2002, 20, 269.
42. Hawkins, D. M.; Basak, S. C.; Mills, D. J. Chem. Inf. Comput. Sci. 2003, 43, 579.
43. Wold, S.; Eriksson, L. Statistical Validation of QSAR Results. In Chemometrics
Methods in Molecular Design; van de Waterbeemd, H., Ed.; VCH: Weinheim,
1995; p 309.
References and notes
44. Todeschini, R.; Consonni, V.. In Molecular Descriptors for Chemoinformatics;
Wiley-VCH: Weinheim, 2009; Vol. 2,.
45. Table S1, including in Supporting Information, shows the predicted biological
data of new proposed 3-arylquinoxaline di-N-oxides, 24–72, according to QSAR
equations 2 and 4.
1. Cerecetto, H.; González, M. Mini-Rev. Med. Chem. 2008, 8, 1355.
2. Ribeiro, I.; Sevcsik, A.-M.; Alves, F.; Diap, G.; Don, R.; Harhay, M. O.; Chang, S.;
Pecoul, B. PLoS Negl. Trop. Dis. 2009, 3, e484.
3. Rivera, G.; Bocanegra-García, V.; Ordaz-Pichardo, C.; Nogueda-Torres, B.;
Monge, A. Curr. Med. Chem. 2009, 16, 3286.