Structure-activity relationship study of antitrypanosomal chalcone derivatives using multivariate analysis

Article abstract of DOI:10.1016/j.bmcl.2019.04.020

Chagas disease represents one of several neglected diseases with a reduced number of chemotherapeutical drugs including the highly toxic compounds benznidazole and nifurtimox. In this sense, natural products represent an import scaffold for the discovery of new biologically active compounds, in which chalcones are promising representatives due to their antitrypanosomal potential. In this work, a series of 36 chalcone derivatives were synthesized and tested against trypomastigotes of Trypanosoma cruzi. In addition, a detailed investigation on their molecular features was performed. The obtained results suggest that certain molecular features are fundamental for an efficient antitrypanosomal potential of chalcones, such as allylic groups, α,β-unsaturated carbonyl system, and aromatic hydroxyl groups. These results were obtained based on the interpretation of machine-learning and multivariate statistical methods, which revealed the essential characteristics of chalcone prototypes against trypomastigotes of T. cruzi.

Full text of DOI:10.1016/j.bmcl.2019.04.020

Accepted Manuscript
Structure-activity relationship study of antitrypanosomal chalcone derivatives
using multivariate analysis
Kaio de S. Gomes, Thais A. da Costa-Silva, Igor H. Oliveira, Andrea M. Aguilar,
Diogo Oliveira-Silva, Miriam Uemi, Wender A. Silva, Lennine R. Melo, Carlos
Kleber Z. Andrade, Andre G. Tempone, João L. Baldim, João Henrique G. Lago
PII:
S0960-894X(19)30240-9
https://doi.org/10.1016/j.bmcl.2019.04.020
BMCL 26387
DOI:
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
8 March 2019
10 April 2019
11 April 2019
Please cite this article as: Gomes, K.d.S., da Costa-Silva, T.A., Oliveira, I.H., Aguilar, A.M., Oliveira-Silva, D.,
Uemi, M., Silva, W.A., Melo, L.R., Andrade, C.K.Z., Tempone, A.G., Baldim, J.L., Lago, J.H.G., Structure-activity
relationship study of antitrypanosomal chalcone derivatives using multivariate analysis, Bioorganic & Medicinal
Chemistry Letters (2019), doi: https://doi.org/10.1016/j.bmcl.2019.04.020
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.

Bioorganic & Medicinal Chemistry Letters
Structure-activity relationship study of antitrypanosomal chalcone derivatives
using multivariate analysis
Kaio de S. Gomesa, Thais A. da Costa-Silvaa, Igor H. Oliveirab, Andrea M. Aguilarb, Diogo Oliveira-Silvab,
Miriam Uemib, Wender A. Silvac, Lennine R. Meloc, Carlos Kleber Z. Andradec, Andre G. Temponed,
João L. Baldima*, João Henrique G. Lagoa*
aCentro de Ciências Naturais e Humanas, Universidade Federal do ABC, Santo André, SP, 09210-580, Brazil
bDepartamento de Química, Instituto de Ciências Ambientais, Químicas e Farmacêuticas, Universidade Federal de São Paulo, Diadema, SP, 09972-270, Brazil.
cInstituto de Química, Universidade de Brasília, Brasília, DF, 70910-900, Brazil
dCentro de Parasitologia e Micologia, Instituto Adolfo Lutz, São Paulo, SP, 01246-000, Brazil.
*corresponding authors, E-mails: joao.lago@ufabc.edu.br (J.H.G.L.) and jotaelebaldim@gmail.com (J.L.B.)
ARTICLE INFO
ABSTRACT
Article history:
Received
Revised
Accepted
Available online
Chagas disease represents one of several neglected diseases with reduced number of
chemotherapeutical drugs including the highly toxic compounds benznidazole and nifurtimox. In
this sense, natural products represent an import scaffold for discovery of new biologically active
compounds, in which chalcones are promising representatives due to their antitrypanosomal
potential. In this work, a series of 36 chalcone derivatives were synthesized and tested against
trypomastigotes of Trypanosoma cruzi. In addition, a detailed investigation on their molecular
features was performed. The obtained results suggest that certain molecular features are
fundamental for an efficient antitrypanosomal potential of chalcones, such as allylic groups,
unsaturated carbonyl system and aromatic hydroxyl groups. These results were obtained
based on the interpretation of machine-learning and multivariate statistical methods, which
revealed the essential characteristics of chalcone prototypes against trypomastigotes of T. cruzi.
Keywords:
Chalcone
Trypanocidal activity
Trypanosoma cruzi
Multivariate analysis
Drug discovery
2019 Elsevier Ltd. All rights reserved.
According to the World Health Organization (WHO), Chagas
Disease (CD) causes around 12000 deaths every year, and due to
emigration events, non-endemic areas are associated to
increasing numbers of detected cases.1-3 Currently, there is no
effective remediation for CD, which has its regulation based on
chemotherapy and vector control.4 The chemotherapy is long-
lasting, which increases the contact with the drug, consequently
intensifying the toxic effects and the development of mechanisms
of resistance by the parasite.4,5
for a systematic investigation of important molecular features
required for a leader compound against T. cruzi.
A series of chalcones (compounds 1 – 26 – figure 1) was
prepared by a Claisen-Schmidt reaction and their structural
identification was performed by NMR and ESI-HRMS.16
Sequentially, the EC50 and CC50 values to each compound were
determined against trypomastigotes forms of T. cruzi and NCTC
cells, respectively.17 In addition, the respective selectivity indexes
(SI) were calculated using the ratio between the CC50 and EC50
values (Table 1).
Drugs used in the treatment of CD, such as benznidazole and
nifurtimox, present high toxicity, low efficacy and numerous side
effects.6-8 Hence, it is essential to develop new alternatives and
methods for identification of novel candidates for the treatment
of CD. In this context, several studies using chalcones derivatives
report their effects as anti-inflammatory,9 anti-leishmania,10,11
antiviral,12 antitumoral13, and, especially, as antitrypanosomal.
Among all tested chalcones, the most actives were compounds
4 (EC50 = 3.3 M) and 5 (EC50 = 13.9 M). These compounds
present, respectively, phenyl and nitro groups at R3 position.
Despite of the effective activity against trypomastigotes of T.
cruzi, compound 4 exhibited cytotoxicity against NCTC cells
(CC50 = 29.0 M). Comparatively, compound 5 afforded a
reduced cytotoxicity (CC50 = 127.3 M), suggesting that this
chalcone could be considered as a molecular prototype for CD
treatment. To compound 7, with an allylic group (propenoxyl at
R1), an EC50 = 10.6 M was determined with intermediary
cytotoxicity (CC50 = 77.6 M). The selectivity index obtained for
compound 7 (SI = 7.3) suggesting that the allylic molecular
14,15
In this work, different new chalcones derivatives were
synthesized and evaluated against trypomastigotes of T. cruzi to
understand the structure-activity relationship of substituents in
the chalcone series. Furthermore, quantitative structure
relationship models based on multivariate statistical analysis
(MSA) and machine learning techniques (MLT) were developed

subunit is associated with improvements on the antitrypanosomal
activity. Compound 2, with an EC50 = 28.6 M and cytotoxicity
greater than 200 M, resulted in a SI = 7. This compound
presents a N,N-dimethyl amino group at R5, conferring an
average antitrypanosomal effect. Other chalcones synthesized in
this work displayed irrelevant SI values ranging from 1.0 to 4.3,
and were considered ineffective due to the cytotoxicity levels.
Compounds 9 – 11, 13, 15 – 20, 23 – 26 exhibited EC50 > 100
M and CC50 > 200 M, being considered inactive against
trypomastigotes of T. cruzi.
Table 1. Antitrypanosomal activity (trypomastigotes forms of T.
cruzi), toxicity (NCTC cells) and selectivity index (SI) to
chalcones and related derivatives 1 – 36 and benznidazole.
EC50 (µM)
CC50 (µM)
T. cruzi
compound
NCTC cells
(95% CI)
SI
trypomastigotes
(95% CI)
28.8 (26.0 - 32.0)
28.6 (21.9 - 37.3)
60.0 (40.3 - 89.5)
3.3 (1.9 - 5.6)
13.9 (11.6 - 16.6)
18.9 (15.0 – 24.0)
10.6 (7.6 - 14.7)
60.9 (44.4 - 83.6)
> 100
1
2
3
4
5
6
7
8
9
79.5 (67.8 - 93.2)
>200
2.8
7.0
3.3
8.9
9.1
4.3
7.3
1.6
>200
29.0 (24.8 - 34.0)
127.3 (118.2 - 137.1)
82.3 (54.3 - 124.8)
77.6 (69.9 - 86.2)
100.0 (83.3 - 119.9)
139.9 (120.6 - 162.3)
>200
-
-
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
benznidazol
> 100
> 100
26.9 (15.8 - 45.9)
> 100
73.2 (61.3 - 87.3)
> 100
>200
-
3.3
-
1.6
-
-
-
-
-
-
1.0
1.5
-
-
-
1.1
-
-
-
-
-
-
-
-
3.4
-
89.3 (62.49 - 127.6)
50.4 (43.5 - 58.4)
117.1 (100.3 - 136.8)
>200
> 100
> 100
>100
>100
>200
50.4 (43.5 - 58.4)
100.7 (82.2 - 123.2)
>200
>100
>200
26.5 (23.4 - 30.1)
18.4 (14.2 - 23.5)
> 100
> 100
> 100
26.0 (22.9 - 29.5)
> 100
24.5 (21.5 - 27.9)
27.3 (23.1 - 32.2)
>200
120.3 (109.1 - 132.5)
>200
29.9 (23.9 - 37.3)
>200
> 100
> 100
> 100
> 100
> 100
> 100
> 100
>200
>200
>200
>200
>200
>200
>200
28.3 (24.6 - 31.5)
> 100
16.5 (13.3 - 19.7)
95.6 (91.9 - 106.2)
>200
> 200
12.0
Figure 1 - Structures of compounds 1 - 36
Aiming to understand the antitrypanosomal effect of
unsaturated carbonyl group, compounds 1 – 5, with elevated
SI values, were submitted to a catalytic hydrogenation,18-20
affording their respective dihydrochalcones 27 – 31 and alcohols
32 - 36 (Figure 1). As demonstrated in Table 1, he reduction of
double bond resulted in inactive derivatives (EC50 > 100 M). In
addition, reduction of ketones 27 – 31 to alcohol derivatives 32 –
34 and 36 resulted in inactive compounds. On the other hand,
compound 35 was considered active as antitrypanosomal agent
(EC50 = 28.3 M), with average cytotoxicity (CC50 = 95.6 M).
The mechanism of action of benznidazole involves the
reduction of nitro groups to amines through the formation of
electrophilic metabolites and/or free radical intermediates. These
events are governed by the action of nitroreductases in the
cytochrome P450 of T. cruzi. These free radicals and
electrophilic
metabolites
damage
the
intracellular
macromolecules in the parasite. Glutathione (GSH) and
trypanothione [T(SH)2)] conjugate with the drugs metabolites
creating drug-thiol adducts that are further metabolized to
mercapturates in the mammal host. In addition, the free radicals

generated are neutralized by the reduced forms of GSH or
T(SH)2.21 Important molecular features for the stabilization of
free radicals could be identified in the chalcones such as
unsaturated carbonyl system (compounds 1 – 26), chelated
hydroxyl (compounds 1, 9, 17 and 27), and free phenolic
(compounds 2, 21, 26, 28, 32, and 33) groups. On the other hand,
prooxidant characteristics are associated with carbonyl and vinyl
conjugated system, as in the case of compounds 1 – 26. It is
important to highlight that a balance between prooxidant and
antioxidant features is essential for the antitrypanosomal effect of
chalcones, since the trypanosomatids present a rudimentary
antioxidant system compared to the mammalian host.22
Regarding this statement, the loss of the electrophilic moiety
after hydrogenation may justify the decrease of antitrypanosomal
activity, as observed for the compounds 27 – 36, with the
exception of compound 35. In this sense, a prototype against CD
must satisfy prooxidant and antioxidant features, in which Bors’
principles could be inspected in the compounds.23
Figure 2. The Loading Scatter Plot highlighting the distribution
of each attribute among classes A, B and C. The closest attributes
to each class are classified as important subunits for the class.
This plot is based on the PLS-DA model developed for the
studied chalcones.
Aiming to further explain the implications of the structural
modifications on the antitrypanosomal activity, three
discriminant in silico models were developed. The combination
between different substituents were evaluated by MSA and MLT
and QSAR analysis of chalcones were obtained based on the
most important groups selected from the loading scatter plot
(LSP), variable importance for predictions (VIPs), One-R, and
J48 experiments. The development of a Partial Least Square-
Discriminant Analysis (PLS-DA) method based on a binary
matrix with 36 instances was fitted in a model with six
components reaching a R2Y of 0.66. R2Y is the percent of
variation of the entire Y explained by the model. Values of R2 >
0.5 are considered good values due to the component complexity
of the instances.24
Furthermore, molecular features of chalcones were
investigated based on machine learning methods, including One-
R26 and J4824. These methods explain datasets using classes for
describing data variation based on a single rule, and decisions
trees for explaining the dataset, respectively.24,26 Results of One-
R suggested (with 75% percent of certainty) that presence of R1-
propenoxy is a crucial moiety for antitrypanosomal activity.
Likewise, J48 results suggested, with 72.2% of certainty, that the
presence of R1-propenoxy moiety is
antitrypanosomal activity (Class A, Figure 3).
a main feature to
The PLS-DA method afforded the identification of promising
subunits from the main dataset and their statistical weights. Since
PLS-DA is a statistical method based on linear regression,25 and
chalcones could act in several different targets in the parasite, the
classification model is better explained with the association of X
variables and their direct dependence of a class (Y variables,
based on the EC50 values). VIPs with values > 1 were selected
and their distribution analyzed in the LSP (Figure 2). The
distribution of X variables suggested that substituents such as R1-
propenoxy (VIP = 1.8, Coefficient = 0.59 to the Class A) and
methoxyl on carbons R4, R5 and R6 (all with VIPs = 1.22,
Coefficient = 0.17 to the Class A) were considered essential to
antitrypanosomal activity. These results also suggested that
presence of -unsaturated carbonyl system is an essential
feature for antitrypanosomal activity. Chalcones with reduced
vinyl system (compounds 5 and 7) are strongly associated with
decreasing the antitrypanosomal activity (89% of instances which
have double bond hydrogenated were inactive), with exception of
compound 35 (Class B, EC50 values ranging from 20 to 30 M).
Compound 35 presents molecular features important to class A,
with methoxyl groups at R3, R4 and R5 (VIPs: 1.22; Coeff.: 0.17
to class A). The LSP shows that most substituents are either in
the gray area or classified at C (inactive), therefore the statistical
model predicts diminished activity with such com substituents,
which is observed in the values of EC50.
Figure 3. Decision tree created by applying the binary matrix to
Weka 3.0. The decision tree correlates the existence and absence
of molecular subunits and they implication on the
antitrypanosomal activity.
The decision tree also points that -unsaturated system
plays an important role for antitrypanosomal activity. As
discussed above, the decision tree algorithm also suggested that a
-unsaturated carbonyl system is important to biological
activity, corroborating multivariate statistics results. On the other
hand, the absence of R1-propenoxy, the methoxyl group at R3
together with the -unsaturated system will classify chalcones
as active, similarly chalcones with conjugated carbonyl system
are reported to be more active than their hydrogenated
derivatives.26, 27

4. Kerboeuf, D.; Riou, M.; Guegnard, F. Mini Rev. Med. Chem.
2008, 8, 116.
These results corroborate our findings of multivariate analysis
and machine learning, in which a -unsaturated carbonyl
system is indispensable for antitrypanosomal properties of
chalcones. Furthermore, congruence was observed in all
statistical models, pointing to these three subunits as crucial for
antitrypanosomal activity (Figure 4).
5. Paes, M. C.; Cosentino-Gomes, D.; de Souza, C. F.; Nogueira, N.
P. D. A.; Meyer-Fernandes, J. R. J. Parasitol. Res. 2011, 2011, 1.
6. Oliveira, M.F.; Dias, A. T. N.; Pontes, V. M. O.; Souza Junior, A.
S.; Coelho, H. L. L.; Coelho, I. C. B. Rev. Patol. Trop. 2008, 37,
209.
7.
Soares Sobrinho, J. L.; Medeiros, F. P. M.; de La Roca, M. F.;
Silva, K. E. R.; Lima, L. N. A.; Rolim Neto, J. R. Rev. Patol.
Trop. 2007, 36, 103.
8. Marin Neto, J. A.; Rassi Junior, Avezum Junior, A.; Mattos.;
Rassi, A. Inst. Oswaldo Cruz. 2009, 104, 319.
9. Reddy, M. V. B.; Hwang, T.; Leu, Y.; Chiou, W.; Wu, T. Bioorg.
Med. Chem. 2011, 19, 2751.
10. Aponte, J. C.; Castillo, D.; Estevez, Y.; Gonzales, G.; Arevalo, J.;
Hammond, G. B.; Sauvain, M. Bioorg. Med. Chem. Lett. 2010, 20,
100.
11. Dal Picolo, C. R.; Bezerra, M. P.; Gomes, K. de S.; Passero, L. F.
D.; Laurenti, M. D.; Martins, E. G.; Sartorelli, P.; Lago, J. H. G.
Fitoterapia, 2014, 97, 28.
12. Cheenpracha,
S.;
Karalai,
C.;
Pongliamanont,
C.;
Subhadhirasakul, S.; Tewtrakal, S. Bioorg. Med. Chem. 2005, 14,
1710.
13. Watteberg, L. W.; Coccia, J. B.; Galhaith, A. R. Cancer Lett.
1994, 83, 165.
14. Passalacqua, T. G.; Dutra, L. A.; de Almeida, L.; Velásquez, A.
M.; Torres, F. A.; Yamasaki, P. R.; dos Santos, M. B.; Regasini,
L. O.; Michels, P. A.; Bolzani, V. da S.; Graminha, M. A. Bioorg.
Med. Chem. Lett. 2015, 25, 3342.
15. García, E.; Coa, J. C.; Otero, E.; Carda, M.; Vélez, I. D.; Robledo,
S. M.; Cardona, W. I. Med. Chem. Res. 2018, 27, 497.
16. Silva, W. A.; Andrade, C. K. Z.; Napolitano, H. B.; Vencato, I.;
Lariucci, C.; de Castro, R. C.; Camargo, A. J. J. Braz. Chem. Soc.
2013, 24, 133.
17. Morais, T. R.; da Costa-Silva, T. A.; Tempone, A. G.; Borborema,
S. E. T.; Scotti, M. T.; de Sousa, R. M. F.; Araujo, A. C. C.; de
Oliveira, A.; de Morais, S. A. L.; Sartorelli, P.; Lago, J. H. G.
Molecules, 2014, 19, 5761.
Figure 4. The network represents each subunit suggested by its
directed node (method). The inner subunits were suggested by
MSA and MLT. The results of the PLS-DA method, which is a
multivariate statistical analysis, are interpreted based on VIPs,
Coefficients and the statistical weights of LSP. The One-R and
J48 methods, which are machine learning methods are interpreted
based on their rules and trees. The network was created in
Cytoscape 3.3.0 using the methods as sources and the subunits as
target variables. The organization of the network was performed
using the algorithm Allegro Layout version 2.2.1.
18. Barradas, S.; Hernández-Torres, G.; Urbano, A.; Carreño, M. C.
Org. Lett. 2012, 14, 5952.
19. Cho, C. S.; Shim, S. C. J. Org. Chem.,2006, 691, 4329.
20. Martínez, R.; Ramón, D. J.; Yus, M. Tetrahedron, 2006, 8988
21. Maya, J. D.; Cassels, B. K.; Iturriaga-Vásquez, P.; Ferreira, J.;
Faúndez, M.; Galanti, N.; Ferreira, A.; Morello, A. Comp.
Biochem. Phys. Part A. 2007, 146, 601.
22. Bors, W.; Heller, W.; Michel, C.; Saran, M. Meth. Enzymol., 1990,
186, 343
23. Baldim, J. L.; Alcântara, B. G. V.; Domingos, O. S.; Soares, M.
G.; Caldas, I. S.; Novaes, R. D.; Oliveira, T. B.; Lago, J. H. G.;
Chagas-Paula, D. A.; Oxid. Med. Cell. Longev. 2017, 1
24. Chagas-Paula, D. A.; Zhang, T.; Da Costa, F. B.; Edrada-Ebel, R.
Metabolites. 2015, 5, 404.
25. Polishchuk, P.; Tinkov, O.; Khristova, T.; Ognichenko, L.;
Kosinskaya, A.; Varnek, A.; Kuz’min, V. J. Chem. Inf. Mod.
2016, 56, 1455.
In conclusion, chalcones and their derivatives provided
promising lead compounds as antitrypanosomal agents. The use
of multivariate analysis and machine learning methods were
fundamental to identify essential molecular features of chalcones
as antitrypanosomal agents. It is worth highlighting that -
unsaturated carbonyl system, methoxyl group at R4, R5 and R6, as
well as R1-propenoxy played an important role for improvement
of antitrypanosomal activity in studied chalcones derivatives.
26. Holte, R. C. Mach. Learn. 1993, 11, 63.
27. Aponte, J. C.; Verástegui, M.; Málaga, E.; Zimik, M.; Quiliano,
M.; Vaisberg, A. J.; Gilman, R. H.; Hammond, G. B. J. Med.
Chem., 2008, 51, 6230.
Acknowledgments
Supporting Information
The authors acknowledge FAPESP (Sao Paulo Research
Foundation - Projects 2018/07885-1 and 2016/19269-8), CAPES,
CNPq and Núcleo de Instrumentação para Pesquisa e Ensino
(NIPE) of ICAQF – UNIFESP. A.G.T. and J.H.G.L. thanks to
CNPq for the scientific research award.
Supplementary data associated with this article can be found,
in the online version, at XXXXXX.
References and notes
1. Izumi, E.; Ueda-Nakamura, T.; Dias Filho, B. P.; Veiga-Junior, V.
F.; Nakamura, C. V. Nat. Prod. Rep. 2011, 28, 809.
2. Beer, M.; German, E. O.; Giberti, G.; Susana, M. V.; Alonso, M.
R.; Patricia, S. V.; Beer, M. F.; Frank, F. M.; Ernesto, B. A.;
Cerny, N.; Luis, M. E.; Cazorla, S. I. Pharm. Biol. 2016, 54, 1.
3. Machado-Silva, A.; Cerqueira, P. G.; Grazielle-Silva, V.; Gadelha,
F. R.; Peloso, E. de F.; Teixeira S. M. R.; Machado, C. R. Mutat.
Res. Rev. 2016, 767, 8.

Structure-activity relationship study of antitrypanosomal
chalcone derivatives using multivariate analysis
Kaio de S. Gomes, Thais A. da Costa-Silva, Igor H. Oliveira, Andrea M. Aguilar, Diogo Oliveira-Silva,
Miriam Uemi, Wender A. Silva, Lennine R. Melo, Carlos Kleber Z. Andrade, Andre G. Tempone,
João L. Baldim, João Henrique G. Lago