Investigating the structure-activity relationships of N’-[(5-nitrofuran-2-yl) methylene] substituted hydrazides against Trypanosoma cruzi to design novel active compounds

Article abstract of DOI:10.1016/j.ejmech.2017.12.011

Chagas disease, caused by the protozoan Trypanosoma cruzi, is a neglected chronic tropical infection endemic in Latin America. New and effective treatments are urgently needed because the two available drugs - benznidazole (BZD) and nifurtimox (NFX) - have limited curative power in the chronic phase of the disease. We have previously reported the design and synthesis of N’-[(5-nitrofuran-2-yl) methylene] substituted hydrazides that showed high trypanocidal activity against axenic epimastigote forms of three T. cruzi strains. Here we show that these compounds are also active against a BZD- and NFX-resistant strain. Herein, multivariate approaches (hierarchical cluster analysis and principal component analysis) were applied to a set of thirty-six formerly characterized compounds. Based on the findings from exploratory data analysis, novel compounds were designed and synthesized. These compounds showed two-to three-fold higher trypanocidal activity against epimastigote forms than the previous set and were 25–30-fold more active than BZD. Their activity was also evaluated against intracellular amastigotes by high content screening (HCS). The most active compounds (BSF-38 to BSF-40) showed a selective index (SI′) greater than 200, in contrast to the SI′ values of reference drugs (NFX, 16.45; BZD, > 3), and a 70-fold greater activity than BZD. These findings indicate that nitrofuran compounds designed based on the activity against epimastigote forms show promising trypanocidal activity against intracellular amastigotes, which correspond to the predominant parasite stage in the chronic phase of Chagas disease.

Full text of DOI:10.1016/j.ejmech.2017.12.011

Accepted Manuscript
Investigating the structure-activity relationships of N’-[(5-nitrofuran-2-yl) methylene]
substituted hydrazides against Trypanosoma cruzi to design novel active compounds
Fanny Palace-Berl, Kerly Fernanda Mesquita Pasqualoto, Bianca Zingales, Carolina
Borsoi Moraes, Mariana Bury, Caio Haddad Franco, Adelson Lopes da Silva Neto,
João Sussumu Murayama, Solange Lessa Nunes, Marcelo Nunes Silva, Leoberto
Costa Tavares
PII:
S0223-5234(17)31016-4
10.1016/j.ejmech.2017.12.011
EJMECH 9989
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 23 August 2017
Revised Date: 29 November 2017
Accepted Date: 2 December 2017
Please cite this article as: F. Palace-Berl, K.F.M. Pasqualoto, B. Zingales, C.B. Moraes, M. Bury,
C.H. Franco, A.L. da Silva Neto, Joã.Sussumu. Murayama, S.L. Nunes, M.N. Silva, L.C. Tavares,
Investigating the structure-activity relationships of N’-[(5-nitrofuran-2-yl) methylene] substituted
hydrazides against Trypanosoma cruzi to design novel active compounds, European Journal of
Medicinal Chemistry (2018), doi: 10.1016/j.ejmech.2017.12.011.
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

1
ACCEPTED MANUSCRIPT
Investigating the structure-activity relationships of N’-[(5-nitrofuran-2-yl)
methylene] substituted hydrazides against Trypanosoma cruzi to design novel active
compounds
Fanny Palace-Berl,a* Kerly Fernanda Mesquita Pasqualoto,b Bianca Zingales,c Carolina
Borsoi Moraes,d Mariana Bury,c Caio Haddad Franco,d Adelson Lopes da Silva Neto,a
João Sussumu Murayama,a Solange Lessa Nunes,c Marcelo Nunes Silva,c Leoberto
Costa Tavaresa
aDepartment of Biochemical and Pharmaceutical Technology, Faculty of Pharmaceutical
b
Sciences, University of São Paulo, SP, Brazil; Innovation and Industrial Development
c
Laboratory, Butantan Institute, SP, Brazil; Department of Biochemistry, Chemistry Institute,
d
University of São Paulo, SP, Brazil; Laboratório Nacional de Biociências (LNBio), Centro
Nacional de Pesquisa em Energia e Materiais (CNPEM), Campinas, Brazil.
ABSTRACT
Chagas disease, caused by the protozoan Trypanosoma cruzi, is a neglected chronic
tropical infection endemic in Latin America. New and effective treatments are urgently
needed because the two available drugs - benznidazole (BZD) and nifurtimox (NFX) -
have limited curative power in the chronic phase of the disease. We have previously
reported the design and synthesis of N’-[(5-nitrofuran-2-yl) methylene] substituted
hydrazides that showed high trypanocidal activity against axenic epimastigote forms of
three T. cruzi strains. Here we show that these compounds are also active against a
BZD- and NFX-resistant strain. Herein, multivariate approaches (hierarchical cluster
analysis and principal component analysis) were applied to a set of thirty-six formerly
characterized compounds. Based on the findings from exploratory data analysis, novel
compounds were designed and synthesized. These compounds showed two- to three-
fold higher trypanocidal activity against epimastigote forms than the previous set and
were 25 to 30-fold more active than BZD. Their activity was also evaluated against
intracellular amastigotes by high content screening (HCS). The most active compounds
*Corresponding author address: Av. Prof. Lineu Prestes, 580, São Paulo, SP 05508-000, Brazil. Phone:
+55 11 30913693; +55 11 3815 6386; Cel.: +1 519 830 6827; e-mail: palaceberlf@usp.br

2
ACCEPTED MANUSCRIPT
(BSF-38 to BSF-40) showed a selective index (SI’) greater than 200, in contrast to the
SI’ values of reference drugs (NFX, 16.45; BZD, > 3), and a 70-fold greater activity
than BZD. These findings indicate that nitrofuran compounds designed based on the
activity against epimastigote forms show promising trypanocidal activity against
intracellular amastigotes, which correspond to the predominant parasite stage in the
chronic phase of Chagas disease.
Keywords: Nitrofurans; Trypanosoma cruzi; amastigote intracellular forms;
exploratory data analysis; chemometric approaches; Chagas disease; structure-activity
relationships.
1 INTRODUCTION
Chagas disease (CD) is a potentially life-threatening illness to around eight
million infected people in 21 Latin American countries. The disease is emergent in non-
endemic countries introduced by extensive global migrations and perpetuated by means
of congenital transmission [1].
The protozoan Trypanosoma cruzi, which is the etiological agent of CD, has
three main developmental forms: the epimastigote, encountered in the gut of triatomine
vectors and easily cultured in the laboratory; the trypomastigote, encountered in the
blood of infected hosts and that has the capacity to invade mammalian cells; and the
amastigote, which multiplies in the cytoplasm of infected cells. T. cruzi strains show a
remarkable genotypic and phenotypic heterogeneity. At the present, the parasite strains
are partitioned into six lineages or discrete typing units (DTUs), TcI–TcVI [2], which
have distinct, but not exclusive ecological and epidemiological associations [3].
Current treatment options for CD are limited to two nitroheterocyclic drugs:
benznidazole (BZD) and nifurtimox (NFX). Although both drugs are quite effective in
curing patients in the acute phase and cases of congenital transmission, the efficacy of

3
ACCEPTED MANUSCRIPT
both drastically diminishes in the chronic phase, which is the prevalent clinical
presentation encountered [4]. The reasons for the marked difference in the drug activity
in the two phases of CD are unclear, but they may be related to unfavorable
pharmacokinetic properties, such as relatively short terminal half-life and reduced tissue
penetration, which might limit their action on the intracellular amastigotes, which
prevail in the chronic phase [5]. On the other hand, the acute phase is characterized by
high parasitemia and abundance of trypomastigote bloodstream forms. Drug
performance can also be affected by individual characteristics of the immunological
response of the patient [6]. One additional factor implicated in the low rates of cure of
BZD and/or NFX in some patients (acute or chronic phases) is the occurrence of T.
cruzi strains naturally resistant to these compounds [7, 8]. It is worth noting that the
resistance or sensitivity to the available drugs is an intrinsic characteristic of a given
strain, and is verified in the three developmental stages of the strain (epimastigote,
trypomastigote and amastigote) [9-11].
The information outlined above indicates the urgent need for efficacious new
drugs active against all phases of CD and all parasite strains. In a previous study, we
have reported the designing and synthesis of N’-[(5-nitrofuran-2-yl) methylene]
substituted hydrazides against T. cruzi, primarily based on molecular modifications of
nifuroxazide (NF), a nitroheterocyclic drug [12, 13] (Figure 1). In those studies, the
anti-T. cruzi activity was assessed against epimastigote forms of three parasite strains
representatives of DTUs TcI, TcII and TcV, which have high prevalence in patients of
different Latin American regions [3]. Most of the synthesized compounds showed
trypanocidal activity four- to ten-fold higher than BZD.

4
ACCEPTED MANUSCRIPT
Figure 1. Nifurtimox (NFX) and nifuroxazide (NF; lead compound) chemical structures
and structural scaffold of the designed compounds. Similar structural portions of NFX
and NF (rectangles) were maintained in the designed compounds. The molecular
modifications in R correspond to aryl, alkyl, cicloalkyl, and heterocyclic groups.
Herein, we have investigated the activity of the previously reported compounds
[12, 13] against a fourth strain, Colombiana, which is considered resistant to both BZD
and NFX [7]. Structure-activity relationships were established to a set of thirty-six N’-
[(5-nitrofuran-2-yl) methylene] analogues employing multivariate chemometric
techniques in order to design novel and more active compounds against T. cruzi. The
explored compounds have in common the (5-nitrofuran-2-yl)methanamine moiety, that
is similar to the pharmacophoric group of NFX (rectangles in Figure 1). The 5-
nitrofuran group was disclosed as an important structural fragment for anti-
trypanosome activity [14, 15]. In addition, nitroheterocyclic compounds have been
reported as pro-drugs due to the nitro group reduction process that is required to
undergo biological response [16, 17]. These information about the 5-nitrofuran group
led us to further investigate N’-[(5-nitrofuran-2-yl) methylene] analogues against
trypanosomatids.
Multivariate methods are often applied when more than one variable needs to be
observed simultaneously. As complementary methods to recognize patterns in the
investigated set of compounds, hierarchical cluster screening (HCS) and principal
component analysis (PCA) were employed [18, 19]. Based on the findings of these

5
ACCEPTED MANUSCRIPT
analyses, structure-activity relationships were established. This allowed the designing
and synthesis of novel compounds that showed a considerable increase in trypanocidal
activity against trypomastigotes and intracellular amastigote forms, the latter evaluated
by high content screening (HCS).
2 RESULTS AND DISCUSSION
2.1 Biological activity against epimastigote forms of T. cruzi
The biological activity of thirty-six N’-[(5-nitrofuran-2-yl) methylene] substituted
hydrazides against the Colombiana strain, which belongs to T. cruzi DTU TcI [2], was
determined, following protocols previously reported [13]. This activity is expressed as
the inhibitory concentration IC50, that corresponds to the drug concentration that inhibits
50% of the epimastigote growth (Table 1; See also Supplementary Material: Dose-
response curve of anti-T. cruzi activity – epimastigote forms, Colombiana strain, p.S3).
The activity of eighteen N’-[(5-nitrofuran-2-yl) methylene] substituted hydrazides
against Silvio X10 cl1, Y and Bug 2149 cl10 strains has been reported elsewhere [12,
13]. Compounds BSF-14 and BSF-31 showed low anti-epimastigote activity and their
IC50 values could not be determined even with the maximum concentration used in the
assay, dictated by the compound solubility.
According to Table 1, eighteen compounds out of thirty-six (50%) were more
active than NFX, and thirty-two compounds (89%) were more active than BZD for the
Colombiana strain. The activity of the same compounds was evaluated against
epimastigotes of the Silvio X10 cl1 strain (also DTU TcI), indicating that four
compounds out of thirty-six (11%) were more active than NFX and twenty-five
compounds out of thirty-six (70%), more active than BZD [13] (See Supplementary
Material: Table 1S, Biological activity of compounds against four strains of T. cruzi).

6
ACCEPTED MANUSCRIPT
Comparison of the data of both strains evidences that the Colombiana strain is more
susceptible to the compounds than the Silvio strain, in contrast to what is reported for
the reference drugs, BZD and NFX, for which Silvio is more sensitive than Colombiana
[7].
Furthermore, it is noteworthy that 4-lipophilic-substituted benzene compounds
show high anti-parasitic activity. This feature could be related to the presence of a
lipophilic pocket in a putative receptor or to facilitated membrane permeation. In
contrast, hydrophilic substituents did not show increment of activity. Interestingly,
compound BSF-23 [-C6H4-4-N(CH3)2] exhibited a huge loss of biological activity
when compared to its bioisostere BSF-28 [-C6H4-4-tert-C4H9]. This characteristic was
previously reported [13], and was mainly attributed to the solvent accessible surface
area of hydrophobic atoms (ASA_H), which showed high values for the most active
compound (BSF-28) and low values for less active (BSF-23). Thus, the hydrophobicity
of 4-substituted benzene seems to be directly related to the anti-T. cruzi activity. These
observations match with the computational analysis presented below.
In face of the promising anti-T. cruzi activity of the compounds and aiming at
predicting more active compounds, the experimental data were used as dependent
variables to establish qualitative structure-activity relationships.
Table 1. Biological activity of N’-[(5-nitrofuran-2-yl) methylene] substituted
hydrazides against epimastigote forms of T. cruzi Colombiana strain.
Colombiana
(TcI)
IC50 (µM)
(mean ±SD)
9.52 ±1.03
60.61 ±6.17
7.58 ±0.76
7.07 ±0.74
Colombiana
(TcI)
IC50 (µM)
(mean ±SD)
8.64 ±0.92
8.07 ±0.78
Substituents
(R)
Substituents
(R)
Cpd
Cpd
BSF-1a
BSF-2
BSF-3a
BSF-4
BSF-19a
BSF-20a
BSF-21a
BSF-22a
-CH3
-CH2CN
-C5H11
-C6H4-4-Cl
-C8H17
-C6H4-4-iso-C3H7 6.08 ±0.65
-C6H4-4-C3H7 7.25 ±0.75
-OC2H5

7
ACCEPTED MANUSCRIPT
2-furyl
cyclopentyl
-O-tert-C4H9
-C6H4
13.14 ±1.07
9.67 ±0.93
7.41 ±0.73
11.30 ±1.14
15.12 ±1.32
10.19 ±0.78
8.53 ±0.80
10.06 ±0.99
11.92 ±1.21
ND
-C6H4-4-N(CH3)2 85.53 ±9.36
-C6H4-4-OC2H5 11.55 ±1.15
BSF-5
BSF-23a
BSF-24
BSF-25
BSF-26
BSF-27a
BSF-28a
BSF-29a
BSF-30
BSF-31
BSF-32
BSF-33a
BSF-34a
BSF-35a
BSF-36a
BSF-6a
BSF-7a
BSF-8a
BSF-9
-C6H4-3-NO2
2-naphthyl
-C6H4-4-C4H9
40.64 ±4.24
7.58 ±0.28
2.45 ±0.27
2-thiophene
cyclohexyl
-C6H4-3-CH3
-CH2-C6H4
-C6H4-4-CH3
-C6H4-4-NH2
-5-4-CH3-1,2,3-
thiadiazole
-C6H4-4-CN
-C6H4-C2H5
BSF-10a
BSF-11a
BSF-12a
BSF-13a
BSF-14
-C6H4-4-tert-C4H9 6.10 ±0.64
-C6H4-4-OC3H7 5.26 ±0.55
-C6H4-3,4-OCH3 26.59 ±2.63
-C6H4-4-sulfamoyl
-C6H4-3CF3
ND
9.98 ±1.06
14.24 ±1.45
-C6H4-4-NHC4H9 3.94 ±0.43
BSF-15
BSF-16a
BSF-17
BSF-18a
NF
20.24 ±2.04
6.55 ±0.69
-C6H4-4-OC4H9
-C6H4-4-C6H4
-C6H4-2-OC6H4
2.86 ±0.31
3.06 ±0.23
5.79 ±0.43
-C6H4-4-OCH3 8.08 ±0.85
187.78 ±9.91
47.91 ±4.96
BZD
9.93 ±1.01
NFX
a: The anti-T. cruzi activity of these compounds was previously reported against Silvio X10 cl1, Y and
Bug 2149 cl10 strains [12, 13]. IC50: compound concentration that causes a reduction of 50% of parasite
growth, compared to controls. IC50 values presented as the mean and the standard deviation (SD),
correspond to triplicates from at least two independent experiments. Errors are in a range of 10%. ND:
not determined in the maximum concentration used in the assay, without precipitation of compound.
Cpd: compounds, NF: nifuroxazide, BZD: benznidazole, NFX: nifurtimox.
2.2 Molecular modelling approach, calculation and selection of descriptors
The crystallographic structure of NF was used as template to construct the three-
dimensional (3D) molecular models of each compound in their neutral form, as
described [12, 13]. The lowest-energy conformer of each compound was selected in the
energy equilibrium region of the conformational ensemble profile (CEP) from
molecular dynamics (MD) simulations. The root means square deviation (RMSD) was
computed as a criterion to verify whether the structural integrity was maintained after
simulations by comparing the atomic positions of the selected conformers from CEP
and those of the energy-minimized structures (See Figure 1S, Supplementary Material).
The RMSD values found for the lowest-energy conformer of all compounds and the NF
template were lower than 0.07 Å. The pharmacophore (5-nitro-2-furan) and the N-

8
ACCEPTED MANUSCRIPT
acylhydrazone moiety were considered in the structures overlay in order to emphasise
the pharmacophore moiety structural integrity for the investigated compounds.
Otherwise, higher RMSD values would indicate loss of structural integrity during MD
simulations regarding the crystallographic structure used as template [20]. The total
potential energy (ETOTAL) was calculated considering the summation of the following
energy contributions: contribution of axial strain energy (ESTRETCH), energy input of
angular deformation (EBEND), contribution of torsional strain energy (ETORS), energy
input of interactions of 1-4 type (E1-4), energy input of van der Waals energy (EvdW),
contribution of electrostatic energy (ECHARGE), including the intramolecular energy of
solvation (Esolv) and the intramolecular energy of hydrogen bonding (EHb). The values
obtained for each energy contribution and for total potential energy can be found in
Table 2S of Supplementary Material. The total potential energy values ranged from 4.22
to 67.86 kcal.mol-1.
Molecular properties, or descriptors, of different nature (thermodynamic,
hydrophobic, electronic, topological, geometric, spectrograph and steric properties)
were calculated for each compound, using the methods listed in Table 3S
(Supplementary Material). The values of the calculated properties can be found in Table
4S (Supplementary Material).
The activity values (expressed as log1/IC50) against epimastigote forms of the
four strains were considered as the dependent variable. Because compounds BSF-14
and BSF-31 did not show activity against T. cruzi, they were discarded. Thirty-four
compounds plus NF were considered to carry out the analysis (n = 35). The activity
ranges were evaluated for each strain and, then, common intervals were assigned as
follows: log1/IC50 values from 6.00 to 5.40 as highly active compounds (H); log1/IC50

9
ACCEPTED MANUSCRIPT
values from 5.40 to 4.60 as moderately active (M); log1/IC50 values from 4.60 to 3.50 as
low activity compounds (L).
A matrix composed by 35 rows (samples or compounds) and 56 columns
(descriptors plus biological activity values) was used as input for studies of structure-
property/activity relationships. The descriptors values were autoscaled [21] due to the
different orders of magnitude among the molecular properties considered in the
analysis. A preliminary variable selection was carried out using as criteria (i) the
Pearson's linear correlation coefficient between descriptors and biological activity
(cutoff = 0.35), and (ii) data distribution through scatter plots (biological activity, pIC50
values, versus each calculated molecular property or descriptor) [18]. These findings are
presented in Figure 2S of Supplementary Material.
2.3 Exploratory data analysis (Hierarchical cluster analysis, HCA, and principal
components analysis, PCA)
The unsupervised HCA aims at grouping a data set considering similar
attributes. Euclidean distance and complete linking method were used to systematically
cluster the samples [18, 22]. The similarity index (SI), generated by the data set, varies
between 0 and 1, in which 1 corresponds to the maximum similarity. The similarity
matrix generates the dendrogram, which is a map shaped tree constructed from the data
distances [18, 19, 21]. According to HCA, 35 samples (34 compounds and NF) were
grouped into two major clusters, A and B, sharing 33.9 % and 61.2 % similarity,
respectively (Figure 2A). Cluster A grouped more hydrophilic compounds with low
anti-trypanosomal activity. Compound BSF-23 [-C6H4-4-N(CH3)2] was grouped with
BSF-21 [-C6H4-4-iso-C3H7] into cluster B, probably due to topological similarities.
Compounds containing alkyl and heterocyclic substituents (BSF-1 to BSF-7; BSF-9,

10
ACCEPTED MANUSCRIPT
BSF-10, BSF-12 and BSF-15) grouped mostly in cluster A. Only one alkyl-substituted
compound, BSF-20 [-C8H17], was grouped into cluster B likely due to its bulky highly
lipophilic substituent. The highly active compounds were grouped in the sub-cluster B'
sharing 80.7 % similarity. Compound BSF-27 [-C6H4-4-C4H9] (highly active) was
grouped with BSF-28 [-C6H4-4-terc-C4H9] (moderately active) into sub-cluster B"
probably because they are constitutional isomers, but they differ regarding molecular
volume.
Figure 2. HCA findings: (A) Dendrogram of samples in which the compounds were
grouped into two main clusters: A (SI = 0.339; less bulky, hydrophilic, aliphatic,
heterocyclic, cycloalkyl substituents) and B (SI = 0.612; bulky and hydrophobic
substituents). SI: similarity index; H: highly active compounds against the T. cruzi

11
ACCEPTED MANUSCRIPT
epimastigote forms (in blue); M: moderately active compounds; L: low activity
compounds. PCA findings: (B) Factors selection table: the two first factors, or PCs,
explained 86.94 % of total variance from the original data; (C) Loadings’ table for the
two first factors, PC1 and PC2: high loadings values are in bold; (D) Outliers’
diagnosis: samples residual versus Mahalanobis distance; (E) Plot of scores for the two
first factors, PC1 versus PC2.
PCA has the purpose of correlating descriptors in order to find other variables,
not only to reduce the descriptors’ number, but also to group redundant descriptors in a
new set of variables, orthogonal, called principal components (PC) or factors, which
describe the information contained in the original data set [18, 19, 21]. In regard to our
data set, the first two factors have described approximately 87 % of the total variance
from the original data, being 77 % discriminated by PC1 (Figure 2B). In the table of
loadings (Figure 2C) are highlighted the descriptors which most influenced the
compounds’ discrimination. In PC1, the loading values for stereochemical and
geometric (ASA_H, AvdW and VvdW), topological (IPlatt, IRandic, IHyper Wiener, ISzeged, IWiener),
hydrophobic (ClogPWM), hydrophobic/stereochemistry (RM), and constitutional (no of
atoms and MM) properties were higher and similar ranging from 0.25 to 0.29 (absolute
values). For PC2, thermodynamic (ETOT; -0.57) and electronic (CHELPG_C6; 0.55)
properties showed the highest loading values.
Regarding the score plot (Figure 2E), PCA and HCA had complementary
findings. Compounds with higher values in PC1, or Factor 1 (77 %), correspond to the
same compounds grouped into cluster B (dendrogram of samples, HCA). Of note,
bulkier and more hydrophobic substituents seem to improve the activity against T. cruzi
epimastigotes.
The diagnosis of possible outliers was carried through plotting sample residual
values versus Mahalanobis distance [23]. The confidence interval used was 95 %, which

12
ACCEPTED MANUSCRIPT
is defined internally in Pirouette program [22], and the threshold lines are indicated in
red color (Figure 2D). Generally, samples are considered as outliers when they cross
both threshold lines. Compound BSF-6 (five-membered cycloalkyl substituent;
moderately active) and NF (similar to phenyl-substituted compounds; low activity) have
crossed only one threshold line and were not considered as outliers in this exploratory
analysis.
A panel with the activity profile of all compounds (n = 36) against the four T.
cruzi strains is provided in Figure 3 to better visualize the activity data, which were used
for grouping the compounds regarding the chemical nature of their substituents (Figure
3). The 4-substituted benzene compounds, mainly those presenting bulky and
hydrophobic substituents, showed better activity.
Figure 3. Activity profile of the 36 compounds against epimastigote forms of T. cruzi.
Ranges of IC50 values are indicated by the colors: dark-green, 0−2 µM; green, 3-5 µM;
light-green, 6-10 µM; yellow, 11-30 µM; orange, 31-60 µM; dark-orange, 61-99 µM;
red, >100 µM; a: heterocyclic; b: cycloalkyl; c: benzene; d: 2 or 3 or 3,4-substituted
benzene; e: naphthyl; f: reference drugs.
The molecular properties highlighted as important to the anti-epimastigote
activity in HCA/PCA approaches (Figure 2) and the data from the compounds’ activity
profile regarding the structural nature of substituents (Figure 3) were considered to
choose the representative descriptors (hydrophobic and steric/geometric) to be further
exploited. ClogPWM (calculated partition coefficient, water/n-octanol, applying the

13
ACCEPTED MANUSCRIPT
weighted method) [24-27] and ASA_H (solvent accessible surface area of all
hydrophobic atoms) properties can be visualized in Figure 4, comparing the more active
compounds (BSF-27, BSF-33, BSF-34, BSF-35) with the NF lead compound and NFX
drug (Figure 4).
Figure 4. Representation of the calculated hydrophobic and steric/geometric descriptors
for the most active compounds (BSF-27, BSF-33, BSF-34, BSF-35), NF (nifuroxazide)
and NFX (nifurtimox): (A) map of lipophilic potential, MLP, calculated using the
SYBYL 8.0 package. More hydrophobic regions are shown in brown ( 0.13) and more
hydrophilic regions in blue (-0.13); the ClogP values were calculated using the Marvin
program [27], considering equal weights to the methods of VISVANADHAN et al,
1989 Klopman et al, 1993 and PHYSPROP © database [24-27]. The compounds are
presented in the stick model, where carbon atoms are in light gray, oxygen in red,
nitrogen in blue, sulphur in yellow, and hydrogen in green. (B) ASA_H was calculated
using ViewerLite 5.0 software [28] employing a probe of 1.4 Å radii (water); the

14
ACCEPTED MANUSCRIPT
surfaces are displayed as transparent format. The black spots correspond to hydrophobic
atoms (|qi| < 0.125; |qi| is the absolute value of the partial charge of the atom). The
compounds are showed in stick model (carbon atoms in gray, oxygen in red, nitrogen in
blue, sulphur in yellow and hydrogen in white).
The lipophilicity property is widely investigated in SAR and QSAR approaches.
The descriptors of such nature are commonly related to biological activity, since they
may express the permeation of a molecule through biological tissues/membranes, as
well as interactions with transporter proteins and enzymes [29]. The biological activity
against T. cruzi has been related to hydrophobic descriptors in previous studies
considering compounds of similar structures [13, 30, 31]. In this regard, comparing
BSF-8 [-C6H4] and BSF-13 [-C6H4-4-CH3] with the novel 4-alkyl substituted benzene
compounds, we verified that the biological activity increases proportionally with the
size of the alkyl chain. This suggests the presence of a lipophilic pocket in a putative
receptor of the parasite that would better accommodate compounds with greater
hydrophobicity.
Maps of lipophilic potential (MLP) were calculated onto a molecular Connolly
surface and interpreted according to a color scheme, which ranged from brown (0.13,
more lipophilic region) to blue (-0.13; more hydrophilic region). Regarding the most
active compounds, 4-alkyl-substituted benzenes have contributed to increase compound
lipophilicity (brown region on the molecular surface of BSF-27, BSF-33, BSF-34, BSF-
35; see Figure 4A). The ASA_H descriptor indicated that the solvent accessible surface
area of hydrophobic atoms is bigger for the most active compounds (BSF-27, BSF-33,
BSF-34, BSF-35) than for NF and NFX.

15
ACCEPTED MANUSCRIPT
2.4 Designing of new compounds, synthesis and biological activity against
epimastigote forms
Based on the structure-activity/property findings, we designed four new
compounds (See Supplementary material, Figure 3S: Structure elucidation of 5-nitro-2-
1
furfuriliden derivatives - H and 13C NMR spectra of new compounds). Three
compounds (BSF-38 to BSF-40) were designed aiming at increasing the activity against
epimastigote forms, and the fourth compound (BSF-37) was designed to evaluate the
contribution of benzene ring to the activity. The compounds were synthesized as shown
in Scheme 1 (Figure 5A), and their biological activity was evaluated against
epimastigote forms of the four T. cruzi strains (Figure 5B).
A
Silvio X10 cl1
IC50 (µM)
(mean ±SD)
Y
Bug 2149 cl10
IC50 (µM)
(mean ±SD)
Colombiana
IC50 (µM)
(mean ±SD)
B
Substituents
(R)
IC50 (µM)
(mean ±SD)
2.30 ±0.25
3.77 ±0.12
1.98 ±0.19
1.99 ±0.19
2.50 ±0.22
2.55 ±0.20
1.37 ±0.13
1.47 ±0.14
1.27 ±0.13
1.99 ±0.18
0.67 ±0.06
0.85 ±0.07
1.47 ±0.16
1.76 ±0.19
1.05 ±0.12
1.09 ±0.11
BSF-37
BSF-38
BSF-39
BSF-40
nifuroxazide
benznidazole
nifurtimox
300 ±8.24
27.15 ±2.75
6.02 ±0.32
300 ±5.12
40.23 ±4.07
12.84 ±1.30
300 ±9.32 187.78 ±9.91
29.23 ±3.09
7.32 ±0.76
47.91 ±4.96
9.93 ±1.01
Figure 5. (A) Scheme 1: synthetic route of 5-nitro-2-furfuriliden derivatives. Reaction
conditions: (i) CH3OH, H2SO4/reflux/4 h; N2H4 80% in H2O/75 ºC/30 min; (ii) 5-Nitro-
2-furaldehyde, C2H5OH/r.t./6 - 8 h. (B) Biological activity of new compounds against
epimastigote forms of four T. cruzi strains. IC50 value: compound concentration causing
50% reduction in the parasite growth in comparison to controls. IC50 values, presented
as mean and standard deviation (SD), correspond to triplicates from at least two
independent experiments. Errors are in a range of 10%.

16
ACCEPTED MANUSCRIPT
The data indicate that all new compounds show increased trypanocidal activity
against epimastigote forms of the four strains as compared to the starting compounds
(Table 1S of Supplementary Material: Biological activity of compounds against four
strains of T. cruzi). Regardless the phenotypic differences among the strains [2], the
substitution of benzene ring (BSF-38) by cyclohexyl (BSF-37) has no impact on the
anti-epimastigote activity. Furthermore, the increase of the aliphatic chain of the 4-
substituted 6-membered rings (BSF-37 to BSF-40) improves the activity against the
epimastigote forms of the four strains.
2.5 Biological activity against trypomastigote forms of T. cruzi.
As mentioned above, the acute phase of CD is characterized by high parasitemia
due to circulating trypomastigote forms. Thus, we evaluated the activity of the new
compounds (BSF-37 to BSF-40) and the reference drug NFX against this developmental
stage. Trypomastigotes of the Colombiana strain (106/mL) were incubated in the
absence or presence of the drugs. Percentages of inhibition of trypomastigote viability
promoted by the compounds, in relation to the control incubated in the absence of drug
were calculated (Table 2).
Table 2. Anti-T. cruzi activity of the compounds and reference drug against
trypomastigote forms of the Colombiana strain.
Trypomastigotes
Inhibition of
(105/mL)a
0.1
viability (%)b
98.5
98.5
98.5
98.5
7.3
BSF-37
BSF-38
BSF-39
BSF-40
NFX
0.1
0.1
0.1
6.3
6.8
---
Control
Compounds were incubated at 1 µM final concentration for
20h incubation at 37 oC; a: values represent the mean of
two independent experiments; b: inhibition of viability in
relation to the control.

17
ACCEPTED MANUSCRIPT
After 20 h incubation in the absence of drug, but in the presence of 0.5%
DMSO, we observed 32% inhibition of viable trypomastigote in the control, regarding
the initial parasite concentration of 106 parasites/mL. This is due to the presence of
DMSO and to the fact that trypomastigotes are non-dividing cells. NFX at 1 µM
concentration inhibited only 7.3% of parasite viability, whereas the new compounds
promoted a drastic inhibition of 98.5%.
2.6 Biological activity against intracellular amastigote forms of T. cruzi.
Drug analysis against the intracellular amastigote form has been considered as
the most relevant assay to study compounds with anti-T. cruzi activity [32], since it
more closely represents the T. cruzi tissue infection that leads to the development of CD
in humans.
Exploratory data analysis (HCA method) was used to assist the selection of
compounds to be screened against the amastigote forms. Thereby, compounds were
selected to represent each cluster in the dendrogram (Figure 2A), including the novel
four compounds designed posteriorly. Compounds containing substituents of different
chemical nature were included in the sample to evaluate whether they have the same
tendency against both epimastigote and amastigote forms of T. cruzi.
The analysis against amastigote forms was carried out using a high content
screening (HCS). This approach is applied in studies of cell signalling, cell physiology,
in vitro toxicological tests, enzyme targets, among others [32]. HCS evaluates the
antiparasitic activity and cytotoxicity toward host cells based on the determination of
parameters such as number of host cells, ratio of infected cells and number of
amastigotes in infected cells.

18
ACCEPTED MANUSCRIPT
The assays were carried out with the Y strain and the human osteosarcoma cell
line U2OS as host cell [33]. The anti-amastigote activity (IC50), the activity on U2OS
host cells (cytotoxicity, CC50) and the selectivity toward amastigotes (selectivity index,
SI’) were determined (Table 3).
It was observed that the most active compounds were very selective, with SI’
values greater than 200, in contrast to the values of reference drugs: SI’ for NFX =16.45
and BZD > 3. Furthermore, it was observed high correlation between results obtained
for the same compounds against epimastigote (Table 1S of Supplementary Material:
Biological activity of compounds against four strains of T. cruzi) and amastigote forms
of the Y strain, reaching a correlation value of 0.786 (Spearman’s non-parametric
correlation method). Considering results obtained against amastigote forms, the same
behavior was observed for BSF-37 and BSF-38 compounds (P = 0.137; statistical
analysis – Student’s t-test, unpaired) meaning that they do not have significant
differences of activity on amastigote forms. The novel designed compounds having the
benzene ring (BSF-38 to BSF-40) had similar anti-amastigote activity values (P =
0.694, one-way variance analysis, one way-ANOVA, with 95% of confidence interval).
Table 3. Anti-T. cruzi activity and host cell toxicity of the compounds and reference
drugs against the amastigote form of the Y strain
Y
U2OS CC50
(µM)
(mean ±SD)
Max.
Activity
(%)
IC50 (µM)
(mean ±SD)
SI’
Compounds
BSF-1
98.68
4.62 ±1.48 39.21
2.61 ±0.15 19.29
1.06 ±0.11 57.18
2.90 ±0.09 25.43
1.77 ±0.27 21.39
1.87 ±1.54 64.20 ±12.32
1.33 ±0.19 19.55 ±7.47
1.43 ±0.40 >200
±0.66
±7.39
±9.12
±1,01
±3.90
8.49
7.39
53.94
8.77
12.08
34.33
14.70
>140
114.96
94.53
119.43
95.51
97.05
114.96
98.37
BSF-3
BSF-9
BSF-10
BSF-16
BSF-26
BSF-27
BSF-28

19
ACCEPTED MANUSCRIPT
101.80
114.96
97.80
102.60
104.24
103.06
98.26
91.94
85.16
96.60
0.35 ±0.20 12.04
±1.90
1.46 ±0.39 46.62 ±11.80
0.71 ±0.11 109.07 ±23.95
34.40
BSF-33
BSF-34
BSF-35
BSF-37
BSF-38
BSF-39
BSF-40
NF
31.93
153.62
> 120
>208
>244
>227
10.73
> 3
1.66 ±0.62
0.96 ±0.10
0.82 ±0.12
0.88 ±0.30
>200
>200
>200
>200
4.26 ±1.02 45.69 ±23.60
64.34 ±39.52 >200
5.18 ±1.15 85.19 ± 21.08
BZD
NFX
16.45
Y: amastigote form of the Y strain. IC50: concentration of compound which inhibits 50% of amastigotes
compared to the control. CC50: concentration of compound which inhibits 50% of U2OS cells, compared
to control. U2OS: Homo sapiens osteosarcoma cells. SI’: selectivity index [SI’ = CC50 U2OS / IC50 of Y].
Max. Activity: Maximum Activity of the inhibition of amastigotes, compared to controls. Data presented
as mean and standard deviation (SD) of triplicates and at least two independent assays. BZD:
benznidazole; NFX: nifurtimox; NF: nifuroxazide (lead compound).
Among the most active compounds, compound BSF-39 [-C6H4-4-C7H15] showed
the best selectivity to the amastigotes in relation to the host cell. Representative images
obtained for infected cells treated with this compound at the concentration of 1.56 µM
showed that the typical morphology of U2OS cell was maintained and no amastigote
forms were present (Figure 6). Few intracellular parasites were still observed when
BSF-39 was added at 0.39 µM concentration (Figure 6). The data obtained with the
reference drug BZD at 1.56 µM indicated the presence of abundant intracellular
parasites (Figure 6) (See also Figure 4S of Supplementary Material). The dose-response
curves obtained for each compound selected in the assays against the amastigote forms
are found in Figure 5S of Supplementary Material.

20
ACCEPTED MANUSCRIPT
Figure 6. Images obtained by high content screening (HCS) for BSF-39 [-C6H4-4-
C7H15]- and benznidazole-treated infected U2OS cells at 50, 12.5, 1.56 and 0.39 µM
concentrations. Assays performed with the Y strain, in U2OS cells and incubation for
72 h. Host cell and parasite DNA are stained with Draq5 (Biofocus).
3 CONCLUSION
Based on the findings from exploratory data analysis, four new compounds were
designed and synthesized. and their activity was evaluated against trypomastigotes and
intracellular amastigotes. Our findings indicate that the novel compounds are active and
selective for the three developmental forms of T. cruzi: epimastigotes, trypomastigotes
and intracellular amastigotes. These results stimulate further studies on N’-[(5-

21
ACCEPTED MANUSCRIPT
nitrofuran-2-yl) methylene] substituted hydrazides as potential drug candidates to treat
Chagas disease.
4 EXPERIMENTAL SECTIONS
4.1 Chemistry
NMR spectra were recorded on a Bruker ADPX Advanced (300 MHz)
spectrometer employing DMSO-d6 solutions with tetramethylsilane as internal standard.
Melting points were determined using Büchi M-560 apparatus and elemental analysis
was performed on a Perkin-Elmer 24013 CHN Elemental Analyzer. Synthesis and anti-
T. cruzi active of compounds BSF-13, BSF-18 [12]; BSF-1, BSF-3, BSF-6, BSF-7,
BSF-8, BSF-10, BSF-11, BSF-12, BSF-16, BSF-19, BSF-20, BSF-21, BSF-22, BSF-23,
BSF-27, BSF-28, BSF-29, BSF-33, BSF-34, BSF-35, BSF-36 [13], was previously
reported by our group.
4.1.1 General procedure for the synthesis of hydrazides (b): carboxylic acid (I)
(0.02 mol) was refluxed for 4 hours in 20.0 mL (0.49 mol) of anhydrous methanol and
0.5 mL (0.01 mol) of sulphuric acid. The reaction mixture was cooled down to room
temperature and the hydrazine hydrate 80% (v/v) (10.0 mL, 0.13 mol) was added. The
system was maintained by vigorously stirring for more 30 minutes in reflux. After this
period, the mixture was maintained at low temperature to give II and was purified from
ethyl acetate. The hydrazide intermediate of compounds 1 and 2 were commercially
obtained (Sigma-Aldrich, purity of 97%) [34, 35].
4.1.2 General procedure for the synthesis of nitrofurans compounds: were
synthesized by refluxing 5-nitro-2-furaldehyde 98% (5 mmol) and hydrazides II (5

22
ACCEPTED MANUSCRIPT
mmol) in water, sulphuric acid, acetic acid, and methanol (8:7:8:20 v/v) for 1 h. After
cooling, the mixture was poured into cold water to precipitate and purified by
recrystallization from acetonitrile [36]. Compounds with high water solubility were
synthesized in 10 mL of ethanol PA (5 mmol) with 5-nitro-2-furaldehyde 98% (1 mmol)
and hydrazides II (1 mmol) at room temperature and vigorous stirring for 6-8 h [37].
Compounds BSF-26, BSF-37, BSF-38, BSF-39, BSF-40 was identified as a new
chemical entity and is described following.
4.1.2.1 4-penpthyl-N'-[(5-nitrofuran-2-yl)methylene] cyclohexane carbohydrazide
(BSF-37): light yellow solid (91%); mp 169 – 170 ºC. 1H NMR (DMSO-d6, 300 MHz):
δ (ppm): 11.67/11.50 (s,1H, H8), 8.17/7.93 (s,1H, H6), 7.76 (d, 1H, J(4,3) = 3.8 Hz, H4),
7.18 (d, 1H, J(3,4) = 3.2 Hz, H3), 3.07/2.18 (t, 1H, J = 12 Hz, H10), 1.79 (d, 4H, J=10.5
Hz, H11, H15), 1.47-1.19 (m, 11H, H3, H18, H19 – Hax. e H16, H17 Hax./eq.), 0.95-
13
0.84 (m, 5H, H18, H19 (Heq.), H20 (CH3)); C NMR {H} (DMSO- d6, 75 MHz): δ
(ppm): 177.3/172.2 (C9), 152.0 (C2), 151.9 (C5), 133.8/130.5 (C6), 114.6 (C4), 114.0
(C3), 43.1/36.4 (C10), 36.7 (C16), 36.6 (C13), 32.0 (C12), 31.9 (C14), 31.6 (C18), 28.8
(C11), 28.3 (C15), 25.9 (C17), 22.1 (C19), 13.9 (C20); Anal. Calcd. for (C17H25N3O4):
C, 60.88; H, 7.51; N, 12.53. Found: C, 60.58; H, 7.20; N, 12.16.
4.1.2.2
4-penthyl-N'-[(5-nitrofuran-2-yl)methylene]benzohydrazide
(BSF-38):
Yellow solid (η 94%); mp 200-201 ºC. 1H NMR (DMSO-d6, 300 MHz): δ (ppm): 12.15
(s,1H, H8), 8.41 (s,1H, H6), 7.84 (d, 2H, J=7.7 Hz, H11, H15), 7.79 (d, 1H, J=3.0 Hz,
H4), 7.36 (d, 2H, J=7.9 Hz, H12, H14), 7.26 (d, 1H, J=3.3 Hz, H3), 2.65 (t, 2H, J=7.5
Hz, H16), 1.60 (t, 2H, J=6.9 Hz, H17), 1.30 (d, 4H, H18, H19), 0.86 (t, 3H, J=6.6 Hz,
H20); 13C NMR {H} (DMSO- d6, 75 MHz): δ (ppm): 163.3 (C9), 151.9 (C2, C5), 147.0
(C13), 135.1 (C6), 130.2 (C10), 128.4 (C11, C15), 127.8 (C12, C14), 115.0 (C4), 114.6

23
ACCEPTED MANUSCRIPT
(C3), 34.9 (C16), 30.8 (C17), 30.3 (C18), 21.9 (C19), 13.8 (C20); Anal. Calcd. for
(C17H19N3O4, 329.35 g): C, 62.0; H, 5.81; N, 12.76. Found: C, 61.94; H, 5.73; N, 12.45.
4.1.2.3 4-hepthyl-N'-[(5-nitrofuran-2-yl)methylene]benzohydrazide (BSF-39):
Yellow solid (η 92%); mp 178-179 ºC. 1H NMR (DMSO-d6, 300 MHz): δ (ppm): 12.15
(s,1H, H8), 8.42 (s,1H, H6), 7.85 (d, 2H, J=7.6 Hz, H11, H15), 7.79 (d, 1H, J=3.0 Hz,
H4), 7.35 (d, 2H, J=7.7 Hz, H12, H14), 7.26 (s, 1H, H3), 2.65 (t, 2H, J=7.2 Hz, H16),
13
1.59 (s, 2H, H17), 1.28 (s, 8H, H18-H21), 0.86 (s, 3H, H22); C NMR {H} (DMSO-
d6, 75 MHz): δ (ppm): 163.3 (C9), 151.9 (C2), 151.8 (C5), 147.0 (C13), 135.2 (C6),
130.2 (C10), 128.4 (C11, C15), 127.9 (C12), 127.8 (C14), 115.0 (C4), 114.6 (C3), 35.0
(C16), 31.2 (C20), 30.6 (C17), 28.5 (C18), 28.4 (C19), 22.0 (C21), 13.8 (C20); Anal.
Calcd. for (C19H23N3O4, 357.470 g): C, 63.85; H, 6.49; N, 11.76. Found: C, 63.58; H,
6.39; N, 11.38.
4.1.2.4 4-hepthyloxy-N'-[(5-nitrofuran-2-yl)methylene]benzohydrazide (BSF-40):
Yellow solid (η 88%); mp 172-173 ºC. 1H NMR (DMSO-d6, 300 MHz): δ (ppm):12.09
(s,1H, H8), 8.41 (s,1H, H6), 7.91 (d, 2H, J=8.8 Hz, H11, H15), 7.79 (d, 1H, J=3.9 Hz,
H4), 7.24 (d, 1H, J=3.9 Hz, H3),7.06 (d, 2H, J=8.8 Hz, H12, H14), 4.05 (t, 2H, J=6.3
Hz, H17), 1.78-1.71 (m, 2H, H18), 1.42-1.28 (m, 8H, H19-H22), 0.87 (s, 3H, J=6.9 Hz,
13
H23); C NMR {H} (DMSO- d6, 75 MHz): δ (ppm): 161.8 (C9), 160.0 (C13), 152.0
(C2), 151.8 (C5), 134.7 (C6), 129.9 (C11), 129.8 (C15), 124.5 (C10), 114.8 (C4), 114.6
(C3), 114.2 (C12, C14), 67.8 (C17), 31.2 (C21), 28.5 (C18), 28.4 (C20), 25.4 (C19),
22.0 (C22), 13.9 (C23); Anal. Calcd. for (C19H23N3O5, 373,40 g): C, 61.11; H, 6.21; N,
11.25. Found: C, 60.99; H, 5.96; N, 10.87.
4.2 Biological activity against epimastigote forms of T. cruzi strains

24
ACCEPTED MANUSCRIPT
In vitro anti-T. cruzi activity assays were performed against the epimastigote
forms of Silvio X10 cl1, Y, Bug 2149 cl10 and Colombiana strains, as described
elsewhere [13]. Exponentially growing epimastigotes (1.0*107 parasites/mL) were
incubated in 96-well, flat-bottom tissue culture plates in 200 µL LIT-FCS medium with
different drug concentrations for 72 h at 28 °C. The nitro derivatives, BZ and NFX
standard drugs, and the lead compound NF were dissolved in DMSO and diluted into
LIT-FCS medium to obtain a concentration from 0.4 to 300.0 µM. DMSO concentration
in the medium did not exceed 1.0%. After the incubation period, the amount of viable
parasites was determined from the absorbance measurements in a microplate reader
(Biochrom – EX Read 400, England), at 562 nm wavelength. The assays were
performed in triplicate and, at least, in two independent experiments. The anti-T. cruzi
activity was calculated by using the formula: PGI = {1-[(Ap - Apb) / (Ac – Acb)]} x 100,
where: PGI = percentage of growth inhibition; Ap = A562 of the culture at a given
compound concentration after 72 h incubation; Apb = A562 of the medium at a given
compound concentration without parasites (blank of compounds in each concentration);
Ac = A562 of the culture in the absence of any compound; ; Acb = A562 of the LIT-FCS
medium (blank). The inhibitory concentration IC50, corresponding to the drug
concentration that inhibited 50% parasite growth, was calculated from the percentage of
growth inhibition using nonlinear model growth/sigmoid dose-response for each case
using, at least, six concentration values [38, 39].
4.3 Molecular modelling approach, calculation and selection of descriptors
The 3D molecular models of the compounds were built up using the Cartesian
coordinates of crystallized structure of the lead compound NF (LEQTAC code, Rfator
=
0.11)[40] retrieved from the Cambridge Structural Database (CSD) as reference

25
ACCEPTED MANUSCRIPT
geometry [41]. Each molecular model was energy-minimized in the MM+ force field
[42], without any constrains. The partial atomic charges were calculated employing the
AM1 (Austin Model 1) semiempirical method [43]. The energy minimization of the 3D
molecular models was also performed employing the MOLSIM 3.2 program[44] by
applying the steepest descent and, subsequently, the conjugated gradient method, using
a convergence criterion of 0.01 kcal/mol. The energy-minimized structures were the
input to the MD simulations of 1ns (1,000,000 steps; step size of 1fs) at 301 K (28 °C),
the same temperature of the biological assay. Dielectric constant of 3.5 was used to
simulate the environment of the biological membranes. It was assigned a fictitious
atomic mass of 5000 u.m.a. to some atoms position in order to maintain the structural
integrity of the molecular models during simulation, based on the 1H and 13C NMR and
NF crystallized structure. Trajectory files were recorded every 20 simulation steps to
generate 50,000 conformations. The lowest-energy conformation was selected from the
equilibrium region of conformational ensemble profile (CEP) and compared to the
initial energy-minimized molecular model to verify whether the structural integrity was
maintained after simulation. The RMSD value was used as criterion employing the
Hyperchem 8.0 program.[41] The energy minimization procedure was made in
MOLSIM 3.2 program[44], applying the steepest descent and conjugated gradient
methods and the convergence criterion was 0.01 kcal/mol. At this time, some
thermodynamic properties were obtained as follows: the total potential energy from
minimization step (Etotal); the intramolecular energy contribution of solvation (Esolv),
applying the hydration shell model proposed by Forsythe and Hopfinger (1973) [45];
and, the intramolecular energy contribution of hydrogen bonding (EHb). The resulting
3D molecular model for each compound was the input structure to the calculation of
physicochemical and structural properties, which were explored in this study.

26
ACCEPTED MANUSCRIPT
The electronic properties as Charges from Electrostatic Potentials using a Grid
based method (CHELPG) [46], dipole moment (total and x, y, z) and frontier
molecular orbital energy (EHOMO and ELUMO) were calculated using the B3LYP
6.311++(d,p) method [47-49] implemented in the Gaussian 03W program [50].
The Marvin 5.2.1_1 program [27] was used to calculate steric,
topological/geometric, and lipophilic properties. ClogPWM was calculated by the
weighted method, assigning equal weight for each method [24-26].
Further detailed information regarding all descriptors, methods, and respective
software used to perform these calculations are listed in table 3S of Supplementary
Material.
The independent variables or descriptors or molecular properties of different
nature were calculated and generated a matrix containing 56 columns (independent
variables or descriptors plus biological activity (log1/C)) and 36 rows corresponding to
the number of compounds.
Procedure to select the most representative molecular properties: two filters were
used as criteria to previously select the independent variables: (i) the Pearson correlation
coefficient and (ii) a visual inspection of scatter plots for biological activity versus each
variable or descriptor or molecular property [18]. The Pearson correlation coefficient
value of 0.35 was formerly used as cut off for selecting the calculated independent
variables. Regarding the visual inspection, only variables presenting uniform
distribution and linear tendency with the biological data were selected to compose the
final matrix.
The final matrix was used as input for the exploratory data analysis. Due to the
distinct magnitude orders among the calculated variables, the autoscaling procedure was
applied as a preprocessing method [21].

27
ACCEPTED MANUSCRIPT
4.4 Exploratory data analysis[18, 19]
4.4.1 Hierarchical cluster analysis, HCA
HCA was performed using the Pirouette 3.11 program [22], employing the
complete linkage method and Euclidean distance. In HCA, distances between pairs of
samples (or variables) are calculated and compared. When distances between samples
are relatively small, this implies that the samples are similar. The calculated distances
between samples were set on a similarity matrix whose elements are called similarity
indices, ranging between 0 and 1, where 1 is equivalent to a maximum similarity. The
results are usually presented as a two-dimensional chart, called dendrogram [22].
The multivariate distance dab between two samples vectors, a and b, was
determined by computing differences at each of the m variables:
1/M
m
dab= ꢀꢁꢂxaj-xbjꢃMꢄ
i
M is the order of the distance, and here corresponds to the Euclidean distance (M
= 2).[22] Because of inter-sample distances can vary with the type and number of
measurements, it is customary to transform them onto a somewhat more standard scale
of similarity, where dmax is the largest distance in the data set:
dab
similarityab= 1-
dmax
HCA was carried out for samples and variables. In the first case, biological
activity was considered as dependent variable. In the second option, the biological
activity is considered as independent variable [18, 19, 22].
4.4.2 Principal Components Analysis, PCA

28
ACCEPTED MANUSCRIPT
The exploratory analysis of PCA was also carried out employing the Pirouette
3.11 software [22]. For PCA, the biological activity was considered as the dependent
variable. The data were decomposed into two matrices, one of scores related to the
samples, and another of loadings, related to the variables [18, 19].
The new set of axes generates the PCs, or factors, into which are the information
related to the original descriptors. Thus, the number of PCs that explain most of the
variability in the data set can be determined, considering that these PCs are uncorrelated
and mutually orthogonal variables built up as simple linear combinations from the
original data. In this exploratory data analysis, PCA was run up to ten factors or PCs.
The outliers’ diagnosis, implemented in Pirouette 3.11 software [22], was also
performed through the Mahalanobis distance [51].
4.5 Biological activity against trypomastigote forms of T. cruzi
Trypomastigote forms of the Colombiana strain were obtained from the
supernatant of infected LLC-MK2 cells. After centrifugation at 7,000 rpm at 4 °C, the
parasites (106 cells/mL) were resuspended in DME medium supplemented with 2% FCS
and added to 96-well, flat-bottom tissue culture plates in the absence or presence of 1
µM of different compounds for 20 h at 37 °C in a 5% CO2 atmosphere. DMSO
concentration in the medium was 0.5%. After incubation, the viability of the parasites
was examined by motility under a light microscope. The number of live parasites was
quantified in a Neubauer chamber. The percentage of inhibition of trypomastigote
viability promoted by the compounds was calculated in relation to the control incubated
in the absence of drug. The assay was performed in two independent experiments.
4.6 Biological activity against amastigote forms of T. cruzi