Leishmanicidal and cytotoxic activities and 4D-QSAR of 2-arylidene indan-1,3-diones

Article abstract of DOI:10.1002/ardp.202100081

The indan-1,3-dione and its derivatives are important building blocks in organic synthesis and present important biological activities. Herein, the leishmanicidal and cytotoxicity evaluation of 16 2-arylidene indan-1,3-diones is described. The compounds w

Full text of DOI:10.1002/ardp.202100081

Received: 1 March 2021
Revised: 23 June 2021
Accepted: 25 June 2021
|
|
DOI: 10.1002/ardp.202100081
F U L L P A P E R
Leishmanicidal and cytotoxic activities and 4D‐QSAR
of 2‐arylidene indan‐1,3‐diones
Ana P. M. de Souza1 | Maria C. A. Costa2 | Alex R. de Aguiar1
Gustavo C. Bressan3 | Graziela D. de Almeida Lima4 | Wallace P. Lima3,5
|
|
Maria P. G. Borsodi5 | Bartira R. Bergmann5 | Márcia M. C. Ferreira2
|
Róbson R. Teixeira1
1Departamento de Química, Universidade
Federal de Viçosa, Viçosa, Brazil
Abstract
2Theoretical and Applied Chemometrics
Laboratory (LQTA), Institute of Chemistry,
University of Campinas, Campinas, Brazil
The indan‐1,3‐dione and its derivatives are important building blocks in organic
synthesis and present important biological activities. Herein, the leishmanicidal and
cytotoxicity evaluation of 16 2‐arylidene indan‐1,3‐diones is described. The com-
pounds were evaluated against the leukemia cell lines HL60 and Nalm6, and the
most effective ones were 2‐(4‐nitrobenzylidene)‐1H‐indene‐1,3(2H)‐dione (4) and
4‐[(1,3‐dioxo‐1H‐inden‐2(3H)‐ylidene)methyl]benzonitrile (10), presenting IC50 va-
lues of around 30 µmol/L against Nalm6. The leishmanicidal activity was assessed on
Leishmania amazonensis, with derivative 4 (IC50 = 16.6 µmol/L) being the most active.
A four‐dimensional quantitative structure–activity analysis (4D‐QSAR) was applied
to the indandione derivatives, through partial least‐squares regression. The statis-
tics presented by the regression models built with the selected field descriptors of
Coulomb (C) and Lennard‐Jones (L) nature, considering the activities against
L. amazonensis, HL60, and Nalm6 leukemia cells, were, respectively, R2 = 0.88, 0.92,
and 0.98; Q2 = 0.83, 0.88, and 0.97. The presence of positive Coulomb descriptors
near the carbonyl groups indicates that these polar groups are related to the ac-
tivities. Besides, the presence of positive Lennard‐Jones descriptors close to sub-
stituents R3 or R1 indicates that bulky nonpolar substituents in these positions tend
to increase the activities. This study provides useful insights into the mode of action
of indandione derivatives for each biological activity involved.
3Escola de Ciências da Saúde, Universidade do
Grande Rio, Duque de Caxias, Brazil
4Departamento de Bioquímica e Biologia
Molecular, Universidade Federal de Viçosa,
Viçosa, Brazil
5Instituto de Biofísica Carlos Chagas Filho,
Universidade Federal do Rio de Janeiro, Rio
de Janeiro, Brazil
Correspondence
Róbson R. Teixeira, Departamento de
Química, Universidade Federal de Viçosa, Av.
P. H. Rolfs, S/N, Viçosa‐MG 36570‐900, Brazil.
Email: robsonr.teixeira@ufv.br
K E Y W O R D S
2‐arylidene indan‐1,3‐diones, 4D‐QSAR, cytotoxic activity, indan‐1,3‐dione, leishmanicidal
activity
least 20 species of the genus Leishmania.[1] It is estimated that
|
1
INTRODUCTION
leishmaniases affect 350 million people in 98 countries, with a global
Leishmaniases and cancer are serious diseases that affect humans
and are of great importance in terms of public health. Leishmaniases
are a group of parasitic infections in which the etiologic agents are at
incidence of 0.9–1.6 million cases per year.[2] The parasites are
transmitted to humans through the bite of female mosquitoes in
the sand fly subfamily. The vast majority of leishmaniasis cases are
© 2021 Deutsche Pharmazeutische Gesellschaft
|
Arch. Pharm. 2021;e2100081.
wileyonlinelibrary.com/journal/ardp
1 of 16
https://doi.org/10.1002/ardp.202100081

2 of 16
DE SOUZA ET AL.
|
associated with the poorest human population, which is in a state of
malnutrition and presents a weak immune system. In addition to
these, other aggravating factors can be considered such as pre-
carious housing, lack of financial resources, displacement of popu-
lation, and environmental changes.[3] The clinical manifestations of
leishmaniases include cutaneous, visceral, and mucocutaneous.[3] The
first‐line drugs for the treatment of leishmaniases are pentavalent
antimonials.[4] Nevertheless, there are disadvantages related to
them, such as the occurrence of serious side effects and a high in-
cidence of disease recurrence.[5] There are alternative medicines,
such as pentamidine, miltefosine, amphotericin B, and paromomycin.
However, they present, among others, high toxicity, high resistance,
teratogenicity, and ototoxicity.[6–10]
Within this scenario and with the aim of expanding the knowl-
edge about the biological activities of indandione derivatives and
their possible therapeutic potential, herein, we describe the evalua-
tion of a series of 2‐arylidene indandiones on Leishmania amazonensis
as well as against two leukemia cancer cell lines. Besides, a four‐
dimensional quantitative structure–activity (4D‐QSAR) analysis was
applied to the indandione derivatives (and their activities) and the
results are discussed.
|
2
RESULTS AND DISCUSSION
|
2.1
Preparation of 2‐arylidene indan‐1,3‐diones
Cancer is the term given to more than 277 related diseases in
which cells that have lost the ability to self‐regulate proliferate without
control, can invade adjacent tissues and organs, or can even spread to
other parts of the body through the blood and lymphatic systems, giving
rise to new tumor sites (i.e., metastasis).[11] It is the second leading
cause of death worldwide (https://www.cancer.gov/about-cancer/
understanding/what-is-cancer) and can have several causes including
lifestyle habits, genetics, carcinogens, and environmental factors.[12]
Cancer treatment involves radiotherapy, surgery, chemotherapy, im-
munotherapy, targeted therapy, hormone therapy, stem cell
transplant, and precision medicine (https://www.cancer.gov/about-
cancer/treatment/types). In terms of chemotherapy, there are several
available drugs for cancer treatment. However, these drugs present
several side effects (https://www.cancer.org/treatment/treatments-
and-side-effects/treatment-types/chemotherapy/chemotherapy-side-
effects.html), and there is also the development of cancer cell line re-
sistance to the available drugs.[13] The resistance is related to the
complex cell signaling pathways modulating the proliferation and ability
of cancer cell lines to escape from apoptotic processes.
The compounds 1, 3–16 (Table 1) investigated herein were prepared,
in one step, from the zirconium‐catalyzed Knoevenagel condensation
reactions between indan‐1,3‐dione and different aromatic alde-
hydes.[30] The general reaction involved in the preparation of the
compounds is shown in Scheme 1.
The indandione derivatives were prepared with yields ranging
from 64% to 95%. It should be mentioned that compound 17 was
obtained via demethylation of compound 13 with BBr3.
|
2.2
Cytotoxic and leishmanicidal activities of
2‐arylidene indan‐1,3‐diones
Considering our interest in the discovery of alternative therapeutic
options for the treatment of cancer and leishmaniases, once syn-
thesized, the compounds 1, 3–17 were subjected to biological assays
to evaluate their leishmanicidal and cytotoxic activities against two
lines of leukemia (Table 1).
Given the problems aforementioned related to the available
chemotherapy for leishmaniasis and cancer, the necessity for the
search and development of new agents that could overcome these
disadvantages is clear.
To examine the cytotoxic effect of 2‐arylidene indan‐1,3‐diones
(1, 3–17), we performed the MTT assay on the leukemia cell lines
HL60 and Nalm6. As a general trend, the compounds showed mod-
erate cytotoxic activity (Table 1). The inspection of IC50 values shows
that the cytotoxic activity depends on the substitution pattern of the
aromatic ring of the arylidene moiety. Considering the HL60 cell line,
five of the 16 synthesized derivatives had IC50 values below
50 µmol/l (compounds 1, 3, 4, 10, and 12). For the Nalm6, five pre-
sented IC50 values below 45 µmol/l (compounds 1, 4, 10, 12, and 15).
Considering these most active derivatives (1, 3, 4, 10, 12, and 15),
while compounds 1, 3, 4, and 10 present as a common structural
feature an arylidene moiety with one group at the para‐position
(1 (–Cl); 3 (–Br); 4 (–NO2); and 10 (–CN)), in compounds 12 and 15,
the aromatic ring of the arylidene portion has three groups. Another
aspect to be noticed is that while derivatives 4 and 10 presented
electron‐withdrawing groups at the para‐position in the arylidene
moiety, compounds 12 and 15 displayed methoxy and hydroxyl
groups, respectively, attached to the same position. Comparing the
cytotoxic activity of 12 and 15, the replacement of a methoxy group
at the R2 position (see Scheme 1) at the arylidene portion by a hy-
droxyl resulted in better activity for derivative 15. Taking into
The indan‐1,3‐dione is an aromatic bicyclic β‐diketone that was
first synthesized more than a century ago.[14] This compound and its
derivatives are valuable synthetic precursors that have been widely
applied for the production of dyes,[15] semiconductors,[16] hetero-
cycles,[17,18] and pharmaceuticals.[19] Indan‐1,3‐dione derivatives
present several biological activities such as antitumoral,[20] antic-
oagulant,[21] anti‐inflammatory,[22,23] neuroprotective,[24] and anti-
microbial.[25] Besides, compounds presenting the indan‐1,3‐dione
core have been isolated from nature.[26,27] For instance, the freder-
icamycin A is a natural spiro indan‐1,3‐dione presenting antitumor
antibiotic activity.[28,29] Figure 1 shows the structures of indan‐
1,3‐dione, some of its derivatives, and related properties.
We have been interested in the biological profile of indandiones.
In this sense, we recently demonstrated the antiviral effect of
2‐arylidene indan‐1,3‐diones.[30] We have also been involved in the
search and development of new compounds that can be possible
alternatives for cancer and leishmaniasis treatment.[31–46]

DE SOUZA ET AL.
3 of 16
|
FIGURE 1 Indan‐1,3‐dione derivatives and related properties. In the structures of the derivatives, the indan‐1,3‐dione core is highlighted
in red
account the compounds presenting methoxy and hydroxyl groups,
the best substitution pattern in terms of improving cytotoxicity is
related to the presence of three methoxy groups at R1, R2, and R3
positions regarding the HL60 cell line and two methoxy groups at R1
and R3 and one hydroxy at R2 for Nalm6. In terms of the halogenated
compounds, the introduction of a fluorine at the R2 position resulted
in compounds with lower activity as compared to chlorine and bro-
mine groups.
In the investigation performed by Pati et al.,[47] indandione de-
rivatives presenting electron‐withdrawing groups attached to the
aromatic ring were prepared and evaluated on the leukemia cell line
Molt4/C8, displaying IC50 values within 7–8 µmol/l. In the present
investigation, the IC50 values of 2‐arylidene derivatives 4 and 10,
which contain electron‐withdrawing groups at the para‐position of
the arylidene portion, as evaluated against the Nalm6 leukemia cell
line were around 30 µmol/l. Liu et al.[48] also biologically evaluated

4 of 16
DE SOUZA ET AL.
|
SCHEME 1 Zirconium‐catalyzed Knoevenagel condensation involved in the preparation of 2‐arylidene‐indan‐1,3‐diones (1, 3–16)
FIGURE 2 Plot of reference versus predicted values for 2‐arylidene indan‐1,3‐diones tested against (a) Leishmania amazonensis,
(b) HL60 leukemia cells, and (c) Nalm6 leukemia cells
indandione derivatives against the K562 leukemia cell line. The IC50
values described in the aforementioned study were lower than the
IC50 described herein.
values of 16.6 and 24.8 µmol/l. To the best of our knowledge, this is
the first investigation of the leishmanicidal activity of 2‐arylidene
indan‐1,3‐diones.
In terms of the effect of the compounds against the promasti-
gote form of L. amazonensis, one of the etiologic agents of cutaneous
leishmaniasis, it was found that, in general, the compounds presented
IC50 values higher than 40 µmol/l. Exceptions to this trend are the
most active derivatives 4 and 15, which presented, respectively, IC50
Indandiones are a relatively new group of compounds that ex-
hibit an interesting variety of biological activities. Unfortunately,
little is known about the mechanisms of action of compounds derived
from this group. However, different research groups that perform
the synthesis and studies of the biological potential of indandiones

DE SOUZA ET AL.
5 of 16
|
TABLE 1 Structures and effects of 2‐arylidene indan‐1,3‐dione derivatives on promastigote forms of Leishmania amazonensis and on HL60
and Nalm6 leukemia cell lines
b
c
IC50 (µmol/l)/pIC50
Compounda
1
Structure
L. amazonensis
HL60
Nalm6
49.0 0.2/4.31
47.88/4.32
42.76/4.36
3
4
5
44.9 0.2/4.35
16.6 0.2/4.78
80.9 0.2/4.09
45.08/4.35
41.99/4.38
71.85/4.14
46.58/4.33
33.92/4.47
66.76/4.18
6
7
80.7 0.1/4.09
125.0 0.1/3.9
111.60/3.95
126.90/3.9
88.90/4.05
>200/3.70
8
242.0 0.3/3.62
197.0 0.1/3.71
63.0 0.1/4.2
>200/3.7
Inactive
9
189.30/3.72
45.10/4.35
>200/3.70
27.30/4.56
10
11
1618.0 0.1/2.79
Inactive
Inactive
(Continues)

6 of 16
DE SOUZA ET AL.
|
TABLE 1 (Continued)
b
c
IC50 (µmol/l)/pIC50
Compounda
12
Structure
L. amazonensis
HL60
Nalm6
81.1 0.1/4.09
43.97/4.36
36.51/4.44
13
14
227.0 0.2/3.64
>200/3.7
Inactive
67.6 0.2/4.17
103.30/3.99
>200/3.70
15
16
24.8 0.3/4.61
83.1 0.1/4.08
>200/3.70
31.08/4.51
73.89/4.13
133.00/3.88
17
147.0 0.2/3.83
110.20/3.96
120.70/3.92
aThe compounds are numbered as previously reported.[16]
bThe concentration of the compound required for 50% inhibition.
cpIC50 means –logIC50 (the pIC50 values were calculated by converting µmol/l into mol/l).
report that these derivatives showed promising antioxidant activ-
ity.[24,49] This activity is important, as studies attribute the antileu-
kemia and leishmanicidal capacity of different substances to their
antioxidative potential.[50–52] However, as already mentioned, little is
known about the mechanisms of action of indandiones, and to better
clarify the relationship between the activity and structure of the
synthesized compounds investigated herein, QSAR models were
performed and the results are presented in sequence.
selected molecular field descriptors are of both Coulomb (C) and
Lennard‐Jones (L) nature.
From the results obtained for 30–40 randomizations
(y‐randomization graphs a–c in Figures 3–5), one can conclude that
the models are free of chance correlation. The intercepts for R2
versus R(yrand, y) and Q2 versus R(yrand, y) must be lower than 0.3 and
0.05, respectively, according to Eriksson et al.[55] The values found in
graphs (a) and (b) of Figures 3–5 were 0.11; 0.15; and 0.13 for R2
versus R(yrand, y) and −1.16; −0.87; and −1.09 for Q2 versus R(yrand
,
y), respectively, for the three models. Besides, the Q2 and R2 values
for randomized y (graphs c in Figures 3–5) were below or close to
0.00 and 0.40, respectively, confirming that the randomized models
were of poor quality, as expected.
|
2.3
QSAR models for the studied activities of
2‐arylidene indan‐1,3‐dione derivatives
The best regression models for 2‐arylidene indan‐1,3‐dione deriva-
tives tested against L. amazonensis, HL60, and Nalm6 leukemia cells
(Figure 2) presented statistics (Table 2) that fulfill the minimal re-
quirements for QSAR studies (R2 > 0.6 and Q2LOO > 0.5).[53,54] The
To assess the robustness of the models, LNO cross‐validation
(graphs d of Figures 3–5) was repeated 30–40 times, leaving one to
five compounds out from the training sets one at a time for partial
least‐squares (PLS) models of the derivatives tested against

DE SOUZA ET AL.
7 of 16
|
TABLE 2 Parameters for the evaluation and validation of the best PLS obtained models
Parameters
Leishmania amazonensis
HL60
15
Nalm6
13
Number of compounds
Number of factors
Descriptorsa
16
2
1
1
̶
L3650, L3632, L2606, C541
L2600, L521, C540, L516,
L663, C709, C2035
C1326, C2630, C2644,
C2839, L2150 L2151,
L2421, L2631, L3705
Total variance percent
54.18
0.88
0.83
0.17
0.18
2.40
0.81
57.17
0.92
0.88
0.08
0.09
1.48
0.86
57.09
0.98
0.97
0.05
0.06
0.85
0.96
R2b
Q2c
SECd
SEVe
RE (%)f
g
Q2
LNO
Abbreviation: PLS, partial least squares.
aMolecular field descriptors of Coulomb (C) and Lennard‐Jones (L) nature.
bCoefficient of multiple determination.
cCross‐validated correlation coefficient.
dStandard error of calibration.
eStandard error of cross‐validation.
fMean relative error.
gFive‐fold cross‐validated correlation coefficient for L. amazonensis and HL60 PLS models and four‐fold cross‐validated correlation coefficient for the
Nalm6 PLS model.
L. amazonensis and HL60 leukemia cells. For the set tested against
Nalm6 leukemia cells, there are a smaller number of compounds and,
in this case, LNO cross‐validation was performed, leaving one to four
compounds out from the training set. For the three PLS models,
and negatively correlated to the L3650 descriptor. This means that
pIC50 increases when the Coulomb and van der Waals interactions
(described by C541, L2606, and L3632) increase, and decreases
when L3650 increases.
the average Q2
values are close to the value of Q2LOO, and the
LNO
pIC50 = 0.34C541 + 0.26L2606 + 0.19L3632 − 0.50L3650
standard deviations for each N value are small, indicating that the
QSAR models are robust.
(1)
To evaluate the consistency of the model, the signals of
The predictive ability of the models can be assessed by the re-
sults obtained for the five‐ or four‐fold cross‐validation (graphs d in
Figures 3–5), once the total sets of the studied compounds are small
(16, 15, and 13 compounds evaluated against L. amazonensis, HL60,
and Nalm6 leukemia cells, respectively) to select external test sets.
The residues and calculated relative error (Table S1), the respective
the coefficients for the correlation between pIC50 and the
descriptors were also investigated.[56] From the coincidence of
the signals for the regression coefficients in the model
(Equation 1) and the signals of correlation coefficients between
each respective descriptor and pIC50 (+0.67 C541; +0.68 L2606;
+0.66 L3632; −0.71 L3650), this model proved to be self‐
consistent.
mean relative errors (RE), and the Q2
(N = 5 or N = 4) values in
LNO
Table 2 indicated that the final models can be used for the prediction
of new bioactive compounds.
The position of the field descriptors used to build the PLS
model, considering compound 15, which is one of the most active
and bulky derivatives, can be visualized in Figure 6, where − and
+ are the signals of the regression coefficients in the model. The
negative Lennard‐Jones (L) descriptor L3650 is located close to
the R2 substituent, meaning that great van der Waals interac-
tions on this position are unfavorable for leishmanicidal activity.
On the other hand, two positive L descriptors are located near R3
(Figure 9), suggesting that bulky substituents attached to this
position favor the biological activity. Comparing compounds 12
and 15 (Figure 9 and Table 1), the activity reduced drastically
|
2.4
Descriptor discussion for regression models
|
2.4.1
Leishmanicidal activity
According to the autoscaled regression coefficients signals in
Equation (1), the leishmanicidal activity is positively correlated to
Coulomb and Lennard‐Jones descriptors C541, L2606, and L3632,

8 of 16
DE SOUZA ET AL.
|
(a)
(b)
(c)
(d)
FIGURE 3 Plots of the y‐randomization test (30 repetitions) for the partial least‐squares model of 2‐arylidene indan‐1,3‐dione derivatives
tested against Leishmania amazonensis, where R(yrand, y) means R (pIC50randomized, pIC50). (a) R2 versus R(yrand, y); (b) Q2 versus R(yrand, y); and
(c) Q2 versus R2, where Q2 is the coefficient of determination for LOO cross‐validation. (d) LNO cross‐validation (N = 1–5) for 30 repetitions
when the substituent in R2 was changed from hydroxyl
in 15 (pIC50 = 4.61) to the bulkier group, –OCH3 in 12
(pIC50 = 4.09).
Also, the positive Coulomb descriptor close to the carbonyl
groups confirms the importance of these groups to the biological
activity of 2‐arylidene indan‐1,3‐dione derivatives, which is corro-
borated by the literature.[47,57,58]
From Equation (1) and the location of the descriptors in Figure 6,
polar groups at position R2 do not seem to be favorable for the
leishmanicidal activity, in agreement with the experimental results
shown in Table 1. Considering the substituents at R2 (Table 1) for
compounds 5 (–F), 14 (–OH), 1 (–Cl), and 3 (–Br), and the respective
values of pIC50 (4.09; 4.17; 4.31; 4.35), one can see that the activity
decreases with the increase in electronegativity of the substituents
in this position.
It is known that the bioactivity of α,β‐unsaturated ketones is
related to the conjugated double bond with the carbonyl function-
ality (–CO–CH═CH–), as removal of the enone group in chalcones,
for example, renders them inactive.[57,58] This α,β‐unsaturated ke-
tone fragment is also present in the series of 2‐arylidene‐1,
3‐indandiones under investigation. Pati et al.[47] performed mole-
cular modifications in a series of 2‐arylidene‐1‐indanones previously
evaluated for cytotoxic properties, generating a new series of
2‐arylidene‐1,3‐indandiones and another of chalcones. These mod-
ifications were made to find explanations for the variation in
bioactivities. The authors found that cytotoxicity and the fractional
positive charge on the olefinic carbon atom were increased by
The PLS regression model suggests that the interaction between
the ring where substitutions were made and the receptor occurs
mainly by van der Waals interactions. Thus, the activity would be
favored by large nonpolar substituents at position R3 and small
nonpolar substituents at R2.

DE SOUZA ET AL.
9 of 16
|
(a)
(b)
(c)
(d)
FIGURE 4 Plots of the y‐randomization test (40 repetitions) for the partial least‐squares model of 2‐arylidene indan‐1,3‐dione derivatives
tested against HL60 leukemia cells, where R(yrand, y) means R (pIC50randomized, pIC50). (a) R2 versus R(yrand, y); (b) Q2 versus R(yrand, y); and (c) Q2
versus R2, where Q2 is the coefficient of determination for LOO cross‐validation. (d) LNO cross‐validation (N = 1–5) for 40 repetitions
placing the additional electron‐attracting oxygen atom into the in-
dane scaffold,[47] reinforcing the importance of the carbonyl groups
for the bioactivity of these compounds.
correlation coefficients[56] between each respective descriptor and
pIC50 (+0.63 C1326; −0.71 C2630; −0.71 C2644; −0.80 C2839;
+0.72 L2150; +0.79 L2151; +0.79 L2421; +0.83 L2631; −0.73
L3705). The coincidence of the signals proved that this model is self‐
consistent.
The results of the 4D‐QSAR analysis provided some insights into
the mode of interaction with the receptor. However, it was not
possible to fully explain the variations in activity with the substitu-
tions at positions R1 to R3. Although the regression model works for
the majority of this series of compounds, some of them do not fit the
presented analysis, as the most active compound 4, suggesting the
necessity of further investigations.
pIC50 = 0.12C1326 − 0.14C2630 − 0.14C2644 − 0.16C2839
+ 0.14L2150 + 0.15L2151 + 0.15L2421 + 0.16L2631
− 0.14L3705
(2)
The location of the field descriptors selected to build the PLS
model, around the bulky and active compound 15, can be visua-
lized in Figure 7. From Equation (2), the biological activity against
Nalm6 leukemia cells is positively correlated with the Coulomb
descriptor C1326, located close to one of the carbonyl groups
that are crucial for the bioactivity of this class of com-
pounds.[57,58] Lennard‐Jones descriptors, positively correlated
with pIC50, are located close to position R1 of the compounds
|
2.4.2
Cytotoxic activity against the Nalm6 cell line
As the first step to this analysis, the consistency of the model pre-
sented in Equation (2) was evaluated by comparison of the auto-
scaled regression coefficients signals (Equation 2) and the signals of

10 of 16
DE SOUZA ET AL.
|
(a)
(b)
(c)
(d)
FIGURE 5 Plots of the y‐randomization test (30 repetitions) for the partial least‐squares model of 2‐arylidene indan‐1,3‐dione derivatives
tested against Nalm6 leukemia cells, where R(yrand, y) means R (pIC50randomized, pIC50). (a) R2 versus R(yrand, y); (b) Q2 versus R(yrand, y); and
(c) Q2 versus R2, where Q2 is the coefficient of determination for LOO cross‐validation. (d) LNO cross‐validation (N = 1–4) for 30 repetitions
(Figures 7 and 9), indicating that the biological activity is favored
by van der Waals interactions in that region, so that bulky non-
polar substituents in R1 tend to increase pIC50. Besides this, the
presence of three negative Coulomb descriptors between R1 and
R2 positions is indicative that polar substituents in these posi-
tions are unfavorable to biological activity, corroborating the
assumption that pIC50 tends to increase with bulky and nonpolar
substituents in R1.
studied compounds. However, the model could not explain the
high activity of compounds 4 and 10, indicating that other factors
related to bioactivity, besides those found in this analysis, need
to be explored.
|
2.4.3
Cytotoxicity against the HL60 cell line
The negative Lennard‐Jones descriptor (−L3705) is located
close to substituents R2 and R3, where bulky groups would be
unfavorable for the considered biological activity. From these
results, one can assume that the interaction of the compounds
with the receptor occurs mainly by substituents in the R1 posi-
tion and by the carbonyl group located close to the R1 position.
The model also indicated that the biological activity under in-
vestigation is favored by bulky and nonpolar substituents in the
R1 position, which explains the activity of the majority of the
The regression model in Equation (3) is self‐consistent, as can be
observed by the coincidence of the autoscaled regression
coefficients signals (Equation 3) and the signals of correlation
coefficients[56] between each respective descriptor and pIC50 (+0.83
C540; +0.85 C709; +0.68 C2035; −0.72 L516; −0.71 L521; −0.62
L663; +0.63 L2600), as explained before.
pIC50 = 0.21C540 + 0.21C709 + 0.17C2035 − 0.18L516
(3)
− 0.17L521 − 0.16L663 + 0.16L2600

DE SOUZA ET AL.
11 of 16
|
FIGURE 6 Compound 15 and 4D descriptors
used to build the partial least‐squares models for
2‐arylidene indan‐1,3‐dione derivatives tested
against Leishmania amazonensis. C and L are the
Coulomb and Lennard‐Jones descriptors, and
− and + are the signals of the regression
coefficients in the model
FIGURE 7 Compound 15 and 4D descriptors
used to build the partial least‐squares models for
2‐arylidene indan‐1,3‐dione derivatives tested
against Nalm6 leukemia cells. C and L indicate the
Coulomb and Lennard‐Jones descriptors, and
− and + are the signals of the regression
coefficients in the model
The location of the field descriptors (Figure 8) around com-
pounds 12 (active; Table 1) and 13 (very low activity; Table 1) sug-
gests a mode of interaction with the receptor similar to that
presented for the analysis of the series tested against the Nalm6
leukemia cells. One can observe a positive Lennard‐Jones descriptor
close to the R1 position, indicating that pIC50 increases with great
van der Waals interaction in this region. The positive Coulomb de-
scriptors close to the carbonyl moieties provide evidence, again, of
the importance of these groups for the bioactivity of this class of
compounds and suggest that this region could also be involved in the
interactions with the receptor. The presence of negative Lennard‐
Jones descriptors near the ring without substituents means that
bulky and nonpolar groups in this region of the molecules induce a
decrease in the activity.
descriptors close to carbonyl moieties, positively correlated
with pIC50 series of
Previous 4D‐QSAR analysis[59] of
.
a
1,4‐naphthoquinones tested against HL60 leukemia cells also
presented positive Coulomb descriptors located close to the car-
bonyl groups (quinone oxygens) involved in the production of ra-
dical anions (O2–•). Quinones can generate reactive oxygen
species (ROS) through the activation by the cytochrome P450 and
P450 reductase enzymes acting as anticancer agents.[60] Com-
paring the results of both 4D‐QSAR analysis for compounds tested
against HL60 leukemia cells, one can suggest that ROS could also
be generated by the indan‐1,3‐dione moiety.
|
3
CONCLUSIONS
These results suggest that the interaction with the receptor in
HL60 leukemia cells may occur mainly by bulky and nonpolar
groups located in the R1 position and by the carbonyl moieties
present in the indan‐1,3‐dione structure. It is interesting to notice
that at least one positive Coulomb descriptor is present in the
vicinities of carbonyl groups in the models discussed here, but for
the activity against HL60 leukemia cells, there are three Coulomb
The present investigation contributed toward broadening the
pharmacological profile of indan‐1,3‐dione derivatives. Sixteen
compounds prepared from indan‐1,3‐dione and bearing arylidene
fragments had two biological activities that were evaluated. For
the first time, the leishmanicidal effect of 2‐arylidene indan‐1,3‐
diones on Leishmania amazonensis, one of the most important

12 of 16
DE SOUZA ET AL.
|
|
4
EXPERIMENTAL
|
4.1
Preparation of indan‐1,3‐dione derivatives
The compounds investigated herein, 1, 3–17 (Table 1) (the InChI
codes are listed in the Supporting Information, together with biolo-
gical activity data), were prepared as previously described.[30]
|
4.2
Antileishmanial activity of 2‐arylidene
indan‐1,3‐diones
Promastigotes of L. amazonensis (WHOM/BR/75/Josefa) were cul-
tured in M199 medium, supplemented with 50 IU/ml of penicillin,
50 μg/ml of streptomycin, 10% (v/v) of heat‐inactivated fetal calf
serum, and 2% (v/v) of human urine at 26°C. For evaluation of the
antipromastigote activity, promastigotes of L. amazonensis were
plated in triplicate at 5 × 105 parasites/ml with varying concentra-
tions of the tested compound (0, 0.1, 1, 10, and 100 µmol/l) in a final
volume of 200 µl of medium M199 containing 5% (v/v) of HIFCS and
1% (v/v) of dimethyl sulfoxide (DMSO). After 72 h of incubation,
parasite viability was assessed by adding resazurin (50 µmol/l) for an
additional 3 h. The fluorescence was quantified (excitation
λ = 560 nm; emission λ = 590 nm), and the data obtained from three
experiments were expressed as the mean standard error of the
FIGURE 8 Superimposed compounds 12 (active) and 13 (very
low activity), and 4D descriptors used to build the partial least‐
squares models for 2‐arylidene indan‐1,3‐dione derivatives tested
against HL60 leukemia cells. C and L indicate the Coulomb and
Lennard‐Jones descriptors, and − and + are the signals of the
regression coefficients in the model
etiological agents of cutaneous leishmaniasis, was demonstrated.
It was possible to conclude that the leishmanicidal potency of the
derivatives depends on the substitution pattern of the aromatic
ring of the arylidene moiety. The most active derivative displayed
an IC50 value of around 15 µmol/l. The cytotoxic effect of the
2‐arylidene indan‐1,3‐dione derivatives was also assessed
against HL60 and Nalm6 leukemia cell lines, and the most active
compounds presented IC50 approximately equal to 30 µmol/l. As
noticed with the leishmanicidal activity evaluation, the cyto-
toxicity is dependent on the substitution pattern of the arylidene
moiety. Considering that the cytotoxicity profile of indan‐1,
3‐dione derivatives has been little explored, the results described
in this investigation contribute toward increasing the knowledge
in this field. A 4D‐QSAR analysis was performed considering the
three investigated biological activities. It was found that the
carbonyl groups of indan‐1,3‐diones are very important in terms
of the evaluated bioactivities. With the field descriptors selected
by the regression models, it was possible to gain some insights
into the mode of action of this series for each investigated bio-
logical activity. The activity of the 2‐arylidene indan‐1,3‐dione
derivatives against L. amazonensis is favored by bulky nonpolar
substituents at the R3 position, while the cytotoxic effects of the
compounds against HL60 and Nalm6 leukemic cells are favored
by the same type of substituents, but at the R1 position. Bulky
nonpolar groups in the R2 position do not favor the activity against
L. amazonensis and on Nalm6 leukemic cells. The positive Coulomb
descriptors close to the carbonyl moieties suggest that these groups
can participate in the interactions with the receptor. It is believed that
the results found in the present study open new possibilities for the
design of more potent indandione derivatives and may result in the
discovery of new pharmaceuticals that can be helpful in the treatment
of leishmaniasis and cancer. Further investigations in this regard are
underway in our group.
mean (mean SEM). The half‐maximal inhibitory concentration (IC50
)
was determined by logarithmic nonlinear regression analysis using
GraphPrism software (version 5, GraphPad).
|
4.3
Cytotoxicity evaluation of 2‐arylidene
indan‐1,3‐diones
Human leukemia cell lines HL60 (acute myelogenous leukemia) and
Nalm6 (acute lymphoblastic leukemia) were kindly provided by Dr. Jose
Andrés Yunes (Centro Infantil Boldrini, Campinas, São Paulo, Brazil).
Cell lines were grown in Roswell Park Memorial Institute (RPMI)‐1640
medium (Sigma) supplemented with 10% (v/v) fetal bovine serum (FBS;
LGC Biotecnologia), 100 μg/ml of streptomycin, and 100 units/ml of
penicillin (Sigma) at pH 7.2 and 37°C under a 5% CO2 atmosphere. Cell
viability was evaluated using the MTT (3‐(4,5‐dimethylthiazol‐2‐yl)‐2,5‐
diphenyltetrazolium bromide) method.[61] HL60 and Nalm6 cells were
seeded onto 96‐well plates at a concentration of 1 × 104 cells/well. Each
well contained 100 μl of complete RPMI medium and 100 μl of each
compound solution at different concentrations (200, 150, 100, 50, 25,
12.5, and 6.25 µmol/L). The cells were incubated at 37°C under 5% of
CO2 for 48 h. All the compounds 1 and 3–17 were diluted in RPMI
medium with 10% FBS plus DMSO (0.4% v/v) (Sigma). After 48 h of
culture, MTT (5 mg/ml; Sigma) was added to each well and incubated
for 4 h at 37 °C. The MTT solution was then removed and DMSO
(100 μl/well) was added to solubilize the formazan. After 20 min at
37°C, absorbance was measured at 540 nm in a microplate reader

DE SOUZA ET AL.
13 of 16
|
(SpectraMax M5, Molecular Devices). Analyses were carried out in
triplicate and the results were normalized considering the cells treated
with only DMSO (0.4% v/v). The half‐maximal inhibitory concentration
(IC50) values were determined using GraphPadPrism version 6.1.
Compounds presenting an inhibitory effect lower than 10% at
200 µmol/l were considered inactive.
FIGURE 9 Basic structure of 2‐arylidene indan‐1,3‐dione
derivatives. The numbering of the atoms indicated in this figure is the
|
4.4
Geometry optimization of 2‐arylidene indan‐
1,3‐diones
same as that from the GaussView program
In the absence of the crystallized binder–receptor complex, the
three‐dimensional geometry of the 2‐arylidene indan‐1,3‐dione de-
rivatives (Table 1) was prepared based on the crystallographic data
simulation for each compound, which incorporates conformational
freedom into the development of 3D‐QSAR models. This conforma-
tional freedom would be the fourth dimension.
from compounds
1 (2‐(4‐chlorobenzylidene)‐1H‐indene‐1,3(2H)‐
dione) and 15 (2‐(4‐hydroxy‐3,5‐dimethoxybenzylidene)‐1H‐indene‐
1,3(2H)‐dione), found in the Cambridge Structural Database (CSD),[62]
entries XICLIH (https://www.ccdc.cam.ac.uk/structures/search?id=
doi:10.5517/ccdc.csd.cc1lsgc6%26sid=DataCite) and XICLED (https://
www.ccdc.cam.ac.uk/structures/search?id=doi:10.5517/ccdc.csd.
cc1lsgb5%26sid=DataCite), respectively. The geometries of these two
compounds were, posteriorly, optimized by the DFT/B3LYP method,
with the def2‐TZVPP basis set,[63] using Gaussian 9.0.[64] This triple zeta
valence basis set is a high‐quality Gaussian basis set optimized for
atoms from H to Rn and was chosen mainly due to the presence of
bromine in one of the studied compounds (Table 1).
MD simulations, performed using the GROMACS‐4.6.5[67]
computational package, were applied to the optimized molecular
geometries using LQTA‐QSAR software.[66] An explicit aqueous
medium was considered. Each compound was placed in a cubic
box with a minimum distance of 10 Å from the molecule to the
edge of the box, which was then filled with water molecules.
Atomic positions were optimized using the steepest descent and
conjugate gradient algorithm with 50 N of maximum force ap-
plied to the atoms, as the convergence criterion. The system was
heated following the scheme of 50, 100, 200, and 350 K for a
10 ps simulation time performed in a 2‐fs step size. Then, the
system was cooled to 300 K and simulated for 500 ps. The con-
formations obtained from each compound were recorded every
10 ps for 500 ps and then they were organized in *. gro files for
the construction of the CEP.
Starting from the optimized geometries of compounds 1 and
15, the three‐dimensional geometries of all other derivatives were
constructed using GaussView 3.0 software,[65] by changing or
adding the substituents R1, R2, and R3 (Figure 9 and Table 1). The
conformational search of each substituent was performed using
the PM3 semiempirical method using the keyword scan from
Gaussian 9.0.[64] To determine the local minima conformations, the
axes a, b, and c (and d for compound 9) depicted in Figure 9 were
rotated with a 15° increment. The conformational analysis in-
volves the search for the biologically active conformation, which is
not necessarily the most stable conformation (global minimum).
Flexible molecules can adopt a large number of stable conforma-
tions that can better fit the target than the global minimum.
Sometimes, the global minimum is stabilized by intramolecular
interactions, which may not be favorable to a binder–receptor
interaction. Based on this argument, stable conformations visually
presenting few intramolecular interactions were chosen. Then, the
selected local minimum of each compound was optimized using
the DFT/B3LYP method as described above.
To build the CEP with all conformations of all compounds
(Figure S1), the resulting conformations from MD simulations were
aligned by atoms 10, 11, 16, and 18 (Figure 9). After that, a virtual
cubic grid was built, large enough to contain the CEP of all molecules,
whose dimensions were 17 × 14 × 13 Å. Then, the LQTAgrid module
from the LQTA‐QSAR program[66] was used to calculate the field
+
descriptors, selecting as a probe the fragment NH3 that mimics the
amino‐terminal portion of peptides. The interaction energies were
calculated using the atomic charges from electrostatic potentials
(ChelpG), obtained with Gaussian 9.0 during the optimization of the
geometries. Each point of the virtual grid, with 1 Å resolution, was
explored by the probe, and 7560 descriptors were generated. The
field descriptors are the contributions of electrostatic and van der
Waals energies (Coulomb and Lennard‐Jones potentials) obtained by
+
the interaction between the probe NH3 and each point of the grid,
according to Equations (4) and (5), respectively.
|
4.5
4D‐QSAR analysis
i
qi qj
1
EC
=
,
(4)
(5)
∑
n
4πε0r
ij
i=1
The methodology chosen for the QSAR analysis was the LQTA‐QSAR
approach[66] that is based on the generation of a conformational
ensemble profile (CEP), instead of considering a single conformation.
The CEPs were generated through molecular dynamics (MD)
1/2
12
ij
12
12
i
12
C
= (C × C
ii
C
Ci6j
⎤
jj
)
1
ij
⎡
⎢
⎣
ELJ
=
−
→
∑
⎥
⎦
r12
ri6j
1/2.
Ci6j = (Ci6i × C6jj
n
ij
i=1
)

14 of 16
DE SOUZA ET AL.
|
In Equations (4) and (5), qi is the charge of the ith atom from
CEP, qj is the charge of the probe, ε0 is the vacuum permittivity,
Cii(12), Cii(6), Cjj(12), and Cjj(6) are parameters adapted from the ffG43a1
Gromos force field[66] for atoms in CEP and the probe, respectively, n
indicates the number of conformations aligned in CEP, and rij is the
distance between the j probe and the ith atom of CEP.
(Figure S2b,c), compounds 4, 9, and 14 presented somewhat large
values of leverage. Nevertheless, no compounds were removed from
the data sets, which contain a small number of compounds.
The quality of the final regression models was assessed by
analyzing the coefficient of multiple determination (R2), the standard
error of calibration (SEC), the cross‐validated correlation coefficient
(Q2), and the standard error of cross‐validation (SEV).
The y‐randomization test[53,54] was applied to investigate the
presence of chance correlation between the dependent variable and
descriptors, that is, descriptors that are statistically well correlated
to y, although in reality not related to the problem under in-
vestigation. For this test, only the vector y is randomized (yrand),
while the matrix X is left untouched. New parallel models were de-
veloped with the values of the original descriptors kept untouched
and the values of the dependent variable, y, permuted between the
compounds. In this study, 30–40 randomization runs were carried
out. It is expected that the statistical parameters from the rando-
mized models (Qy2rand and Ry2rand) should be significantly lower than
|
4.6
Descriptor extraction
As the first step in descriptor extraction, those with variance below
0.02 were excluded, as indicated by Kubinyi.[68] The resulting de-
scriptors were filtered, in the second step, using a correlation coef-
ficient cut‐off, where those presenting
a Pearson correlation
coefficient with y (biological activity) lower than 0.3 were eliminated.
The next and last step was to eliminate descriptors with poor dis-
tribution profiles concerning y utilizing the digital filter Comparative
Distribution Detection Algorithm (CDDA). According to Barbosa and
Ferreira,[69] CDDA provides a way to quantify how similarly dis-
tributed y and a given descriptor are, enabling the removal of those
not well distributed. At the end of these procedures, the field de-
scriptors were reduced from 7560 to 546 for the leishmanicidal
activity, from which 537 were Coulomb (C) and nine were Lennard‐
Jones (L) descriptors. For the activities against HL60 and Nalm6
leukemia cells, the numbers of descriptors were reduced to 720 (644
C and 76 L) and 139 (135 C and 4 L), respectively.
those obtained for nonrandomized data (QL2OO and 2). Another ap-
R
proach to judge whether the real model is characterized by chance
correlation is based on the absolute value of the Pearson correlation
coefficient R(y,yrand), between the original vector y and the rando-
mized vector yrand. Two y randomization plots, R(y,yrand) versus Q2
and R(y,yrand) versus R2, were drawn for all randomized and real
models. Two linear equations of R(y,yrand) versus Q2 and R(y,yrand
versus R2 were obtained for each model. It has been recommended
)
that for a model free of chance correlation, the intercepts are aQ
0.05 and aR < 0.3.[64]
<
|
4.7
PLS regression
To test the robustness of the model, leave‐N‐out (LNO) cross‐
validation was performed. In this test, X and y are simultaneously
randomized and divided into blocks of N samples. Then, each block is
excluded once and a new model is built on the reduced data set. LNO
was repeated 30 times for analysis against leishmaniasis and 40
times for analyses against both types of leukemia cells, for N varying
from 1 to 5 for leishmaniasis and the HL60 cell line, and from 1 to 3
for Nalm6 leukemia cells; the average QL2NO, with its standard de-
viation, was calculated for each value of N. The critical N is the
maximum value for which QL2NO is still stable and high. For a good
model, the average QL2NO should remain close to QL2OO, with small
variations at all values for N up to the critical N. From our experi-
ence, 2 SD should not be greater than 0.1 (QL2OO ± 0. 05) for N = 2, 3,
and so forth, including the critical value of N.
The remaining descriptors were organized in matrices X, one for
each subset of compounds (Table 1), and correlated with the corre-
sponding −logIC50 values arranged in column vectors y, using QSAR
modeling software.[70] The Ordered Predictors Selection (OPS) Al-
gorithm[71] was applied to the autoscaled data, for a further selection
of descriptors. The aim of this method is to obtain a vector that
contains information about the location of the best variables for
prediction. The columns of matrix X are reordered such that the
most important descriptors are placed in the first columns. Then, PLS
regressions are built successively to find the best model. The number
of factors is determined using the leave‐one‐out (LOO) cross‐
validation method. In the end, only four, seven, and five descriptors
for leishmanicidal, HL60 cytotoxic activity, and Nalm6 cytotoxicity,
respectively, were selected.
ACKNOWLEDGMENTS
In this study, the applicability domain is defined by the leverage
and the studentized residuals. The presence of outliers was analyzed
by observing the plot of leverage versus studentized residuals
(Figure S2). For the PLS model of the derivatives tested against L.
amazonensis, compound 15 (Figure S2a) presented a studentized
residual somewhat beyond the critical value of 2.0,[53,54] while
compounds 4 and 11 presented large leverages. For PLS models of
the derivatives tested against HL60 and Nalm6 leukemia cells
We are grateful to CNPq, CAPES, FAPEMIG, and FAPESP for fi-
nancial support.
CONFLICT OF INTERESTS
The authors declare that there are no conflicts of interest.
ORCID
Róbson R. Teixeira
http://orcid.org/0000-0003-3181-1108

DE SOUZA ET AL.
15 of 16
|
REFERENCES
[34] A. M. A. Lima, R. R. Teixeira, B. F. da Silva, R. P. Siqueira,
I. E. P. da Silva, E. G. Santos, M. C. Fernandes, V. H. S. Gonçalves,
G. C. Bressan, T. A. O. Mendes, S. O. de Paula, A. V. Costa,
M. H. dos Santos, Quim. Nova 2019, 42, 473.
[1] S. Burza, S. L. Croft, M. Boelaert, Lancet 2018, 392, 951.
[2] J. Alvar, I. D. Velez, C. Bern, M. Herrero, P. Desjeux, J. Cano,
J. Jannin, M. den Boer, PLoS One 2012, 7, e35671.
[35] M. C. F. Dias, B. L. de Sousa, M. Ionta, R. R. Teixeira, T. Q. Goulart,
G. A. Ferreira‐Silva, E. J. Pilau, M. H. dos Santos, J. Braz. Chem. Soc.
2021, 32, 572.
[3] WHO. Leishmaniasis Fact Sheet No. 375. WHO, Geneva,
Switzerland 2015. http://www.who.int/mediacentre/factsheets/
fs375/en/ (accessed: December 2020).
[36] P. A. R. Gazolla, R. R. Teixeira, A. M. da Silva, B. G. Vaz,
G. A. Vasconcelos, R. P. Siqueira, V. H. S. Gonçalves, H. S. Pereira,
G. C. Bressan, Quim. Nova 2018, 41, 497.
[4] A. K. Haldar, P. Sen, S. Roy, Mol. Biol. Int. 2011, 2011, 571242.
[5] F. Frézard, C. Demicheli, R. R. Ribeiro, Molecules 2009, 14, 2317.
[6] B. S. McGwire, A. R. Satoskar, QJM 2014, 107, 7.
[7] S. L. Croft, S. Sundar, A. H. Fairlamb, Clin. Microbiol. Rev. 2006, 19,
111.
[37] G. D. A. Lima, M. P. Rodrigues, T. A. O. Mendes, G. A. Moreira,
R. P. Siqueira, A. M. da Silva, B. G. Vaz, J. L. R. Fietto, G. C. Bressan,
M. Machado‐Neves, R. R. Teixeira, Toxicol. In Vitro 2018, 53, 1.
[38] G. A. Moreira, G. D. A. Lima, R. P. Siqueira, M. V. A. Barros,
A. L. M. Adjanohoun, V. C. Santos, E. A. A. Barbosa, R. K. Loterio,
J. C. de Paiva, V. H. S. Gonçalves, L. C. S. Viol, M. R. Almeida,
J. L. R. Fietto, M. Machado‐Neves, R. S. Ferreira, R. R. Teixeira,
G. C. Bressan, Toxicol. Appl. Pharmacol. 2018, 356, 214.
[39] R. P. Siqueira, M. V. A. Barros, E. A. A. Barbosa, T. S. Onofre,
V. H. S. Gonçalves, H. S. Pereira, A. S. Júnior, L. L. de Oliveira,
M. R. Almeida, J. L. R. Fietto, R. R. Teixeira, G. C. Bressan, Eur.
J. Med. Chem. 2017, 134, 97.
[8] S. L. Croft, G. H. Coombs, Trends Parasitol. 2003, 19, 502.
[9] M. Ouellette, J. Drummelsmith, B. Papadopoulou, Drug Resist.
Updates 2004, 7, 257.
[10] L. Monzote, Open Antimicrob. Agents J. 2009, 1, 9.
[11] S. H. Hassanpour, M. Dehghani, J. Cancer Res. Pract. 2017, 4, 127.
[12] C. B. Blackadar, World J. Clin. Oncol. 2016, 7, 54.
[13] B. Mansoori, A. Mohammadi, S. Davudian, S. Shirjang, B. Baradaran,
Adv. Pharm. Bull. 2017, 7, 339.
[14] W. Wislicenus, Ber. Dtsch. Chem. Ges. 1887, 20, 589.
[15] M. Matusi, R. Kimura, Y. Kubota, K. Funabiki, K. Manseki, J. Jin,
A. Hansuebsai, Y. Inoue, S. Higashijima, Dyes Pigm. 2017, 147, 50.
[16] T. Du, R. Gao, Y. Deng, C. Wang, Q. Zhou, Y. Geng, Angew. Chem. Int.
Ed. 2020, 59, 221.
[40] A. F. S. Maia, R. P. Siqueira, F. M. O. de Oliveira, J. G. Ferreira,
S. F. da Silva, C. A. D. Caiuby, L. L. de Oliveira, S. O. de Paula,
R. A. C. Souza, S. Guilardi, G. C. Bressan, R. R. Teixeira, Bioorg. Med.
Chem. Lett. 2016, 26, 2810.
[17] K. Shing, Curr. Org. Synth. 2016, 13, 385.
[41] R. P. Siqueira, E. A. A. Barbosa, M. D. Polêto, G. L. Righetto,
T. V. Seraphim, R. L. Salgado, J. G. Ferreira, M. V. A. Barros,
L. L. de Oliveira, A. B. A. Laranjeira, M. R. Almeida, A. Silva‐Júnior,
J. L. R. Fietto, J. Kobarg, E. B. de Oliveira, R. R. Teixeira, J. C. Borges,
J. A. Yunes, G. C. Bressan, PLoS One 2015, 10, e0134882.
[42] R. R. Teixeira, G. C. Bressan, W. L. Pereira, J. G. Ferreira,
F. M. de Oliveira, D. C. Thomaz, Molecules 2013, 18, 1881.
[43] R. R. Teixeira, L. C. A. Barbosa, C. R. A. Maltha, M. E. Rocha,
D. P. Bezerra, L. V. Costa‐Lotufo, C. Pessoa, M. O. Moraes,
Molecules 2007, 12, 1101.
[18] S. Asadi, G. M. Ziarani, Mol. Divers. 2016, 20, 111.
[19] R. Pluskota, M. Koba, Mini‐Rev. Org. Chem. 2018, 18, 1321.
[20] M. K. P. Jayatunga, S. Thompson, T. C. McKee, M. C. Chan,
K. M. Reece, A. P. Hardy, R. Sekirnik, P. T. Seden, K. M. Cook,
J. B. McMahon, W. D. Figg, C. J. Schofield, A. D. Hamilton, Eur.
J. Med. Chem. 2015, 94, 509.
[21] B. E. Watt, A. T. Proudfoot, S. M. Bradberry, J. A. Vale, Toxicol. Rev.
2005, 24, 259.
[22] S. Amidi, F. Kobarfard, A. B. Moghaddam, K. Tabib, Z. Soleymani,
Iran. J. Pharm. Sci. 2013, 12, 91.
[44] M. P. Rodrigues, D. C. Tomaz, L. A. de Souza, T. S. Onofre,
W. A. de Menezes, J. Almeida‐Silva, A. M. Suarez‐Fontes,
M. R. de Almeida, A. M. da Silva, G. C. Bressan, M. A. Vannier‐
Santos, J. L. R. Fietto, R. R. Teixeira, Eur. J. Med. Chem. 2019, 183,
111688.
[23] D. Leblois, S. Piessard, G. Le Baut, P. Kumar, J.‐D. Brion, L. Sparfel,
R.‐Y. Sanches, M. Juge, J.‐Y. Petit, L. Welin, Eur. J. Med. Chem. 1987,
22, 229.
[24] C. B. Mishra, A. Manral, S. Kumari, V. Saini, M. Tiwari, Bioorg. Med.
Chem. 2016, 24, 3829.
[45] R. R. Teixeira, P. A. R. Gazolla, A. M. da Silva, M. P. G. Borsodi,
B. R. Bergmann, R. S. Ferreira, B. G. Vaz, G. A. Vasconcelos,
W. P. Lima, Eur. J. Med. Chem. 2018, 146, 274.
[25] M. Jeyachandran, P. Ramesh, Org. Chem. Int. 2011, 360810.
[26] R. Misra, R. C. Pandey, B. D. Hilton, J. Antibiot. 1987, 40, 786.
[27] Y. Wang, H.‐X. Liu, Y.‐C. Chen, Z.‐H. Sun, H.‐H. Li, S.‐N. Li, M.‐L. Yan,
W.‐M. Zhang, Molecules 2017, 22, 765.
[28] R. C. Pandey, M. W. Toussaint, R. M. Stroshane, C. C. Kalita,
A. A. Aszalos, A. L. Garretson, T. T. Wei, K. M. Byrne,
R. F. Geoghegan Jr., R. J. White, J. Antibiot. 1981, 34, 1389.
[29] D. J. Warnick‐Pickle, K. M. Byrne, R. C. Pandey, R. J. White,
J. Antibiot. 1981, 34, 1402.
[46] W. L. Pereira, R. S. Vasconcellos, C. Mariotini‐Moura, R. S. Gomes,
R. C. Firmino, A. M. da Silva, A. Silva‐Júnior, G. C. Bressan,
M. R. Almeida, L. C. C. Afonso, R. R. Teixeira, J. L. R. Fietto,
Molecules 2015, 14, 22435.
[47] H. N. Pati, U. Das, E. Clercq, J. Balzarini, J. R. Dimmock, J. Enzyme
Inhib. Med. Chem. 2007, 22, 37.
[48] X. L. Liu, B. W. Pan, W. H. Zhang, C. Yang, J. Yang, Y. Shi, T. T. Feng,
Y. Zhou, W. C. Yuan, Org. Biomol. Chem. 2014, 13, 601.
[49] O. Kouzi, E. Pontiki, D. Hadjipavlou‐Litina, Molecules 2019, 24,
4411.
[30] A. F. C. S. Oliveira, A. P. M. Souza, A. S. de Oliveira, M. L. da Silva,
F. M. de Oliveira, E. G. Santos, I. E. P. da Silva, R. S. Ferreira,
F. S. Villela, F. T. Martins, D. H. S. Leal, B. G. Vaz, R. R. Teixeira,
S. O. de Paula, Eur. J. Med. Chem. 2018, 149, 98.
[50] H. Fadel, I. Sifaoui, A. López‐Arencibia, M. Reyes‐Batle,
I. A. Jiménez, J. Lorenzo‐Morales, N. Ghedadba, S. Benayache,
J. E. Piñero, I. L. Bazzocchi, Pathogens 2019, 8, 201.
[31] R. P. Siqueira, M. M. M. Caetano, L. A. de Souza, P. M. S. dos Passos,
N. B. Simaroli, M. V. A. Barros, A. P. M. de Souza, L. L. de Oliveira,
A. Silva‐Júnior, J. L. R. Fietto, R. R. Teixeira, F. R. Teixeira,
G. C. Bressan, Toxicol. In Vitro 2020, 65, 104777.
[51] S. B. Jannet, N. Hymery, S. Bourgou, A. Jdey, M. Lachaal, C. Magné,
R. Ksouri, Biomed. Pharmacother. 2017, 90, 375.
[32] J. A. do Vale, A. P. M. de Souza, G. D. A. Lima, V. H. S. Gonçalves,
G. A. Moreira, M. V. A. Barros, W. L. Pereira, N. C. L. Merchid,
J. L. R. Fietto, G. C. Bressan, R. R. Teixeira, M. Machado‐Neves,
Anticancer Drugs 2020, 31, 718.
[52] D. Lamorte, I. Faraone, I. Laurenzana, S. Trino, D. Russo, D. K. Rai,
M. F. Armentano, P. Musto, A. Sgambato, L. D. De Luca, L. Milella,
A. Caivano, Molecules 2020, 25, 4890.
[53] M. M. C. Ferreira, J. Braz. Chem. Soc. 2002, 13, 742.
[54] Guidance Document on the Validation of (Quantitative) Structure‐
Activity Relationship [(Q)SAR] Models. OECD Environment Health
[33] R. R. Teixeira, A. M. da Silva, R. P. Siqueira, V. H. S. Gonçalves,
H. S. Pereira, R. S. Ferreira, A. V. Costa, E. B. de Melo, F. R. Paula,
M. M. C. Ferreira, G. C. Bressan, J. Braz. Chem. Soc. 2019, 30, 541.

16 of 16
DE SOUZA ET AL.
|
and Safety Publications Series on Testing and Assessment No. 69,
OECD, Paris 2007. http://www.oecd.org/dataoecd/55/35/
38130292.pdf (accessed: March 2017).
S. S. Iyengar, J. Tomasi, M. Cossi, J. M. Millam, M. Klene, C. Adamo,
R. Cammi, J. W. Ochterski, R. L. Martin, K. Morokuma, O. Farkas,
J. B. Foresma, D. J. Fox, Gaussian Inc., Wallingford, CT 2016.
[65] A. B. Nielsen, A. J. Holder, GaussView 3.0 User's Reference, Gaussian
Inc., Pittsburgh, PA 2000–2003.
[55] L. Eriksson, J. Jaworska, A. P. Worth, M. T. Cronin, R. M. Mcdowell,
P. Gramatica, Environ. Health Persp. 2003, 111, 1361.
[56] R. Kiralj, M. M. C. Ferreira, J. Chemom. 2010, 24, 681.
[57] K. Olympia, P. Eleni, H.‐L. Dimitra, Molecules 2019, 24, 4411.
[58] S. Verma, A. K. Srivastava, O. P. Pandey, PharmaTutor 2018, 6, 22.
[59] M. C. A. Costa, P. O. M. Carvalho, M. M. C. Ferreira, J. Chemom.
2020, 34, e3131.
[66] J. P. A. Martins, E. G. Barbosa, K. F. M. Pasqualato,
M. M. C. Ferreira, J. Chem. Inf. Model. 2009, 49, 1428.
[67] E. Lindahl, B. Hess, D. van der Spoel, J. Mol. Mod. 2001, 7, 306.
[68] H. Kubinyi, Drug Discov. Today 1997, 2, 457.
[69] E. G. Barbosa, M. M. C. Ferreira, Mol. Inf. 2012, 31, 75.
[70] J. P. A. Martins, M. M. C. Ferreira, Quim. Nova 2013, 36, 554.
[71] R. F. Teófilo, J. P. A. Martins, M. M. C. Ferreira, J. Chemom. 2009,
23, 32.
[60] K. W. Wellington, Chem. Soc. Rev. 2015, 5, 20309.
[61] T. Mosmann, J. Immunol. Methods 1983, 65, 55.
[62] Cambridge Crystallographic Data Centre, Cambridge Structural Da-
tabase Software Version 5.29‐2007 + 1 Update, Cambridge Crystal-
lographic Data Centre, UK 2007.
SUPPORTING INFORMATION
[63] F. Weigend, R. Ahlrich, Phys. Chem. Chem. Phys. 2005, 7, 3297.
[64] Gaussian 09, Revision A.02, M. J. Frisch, G. W. Trucks,
H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
G. Scalmani, V. Barone, G. A. Petersson, H. Nakatsuji, X. Li,
M. Caricato, A. Marenich, J. Bloino, B. G. Janesko, R. Gomperts,
B. Mennucci, H. P. Hratchian, J. V. Ortiz, A. F. Izmaylov,
J. L. Sonnenberg, D. Williams‐Young, F. Ding, F. Lipparini, F. Egidi,
J. Goings, B. Peng, A. Petrone, T. Henderson, D. Ranasinghe,
V. G. Zakrzewski, J. Gao, N. Rega, G. Zheng, W. Liang, M. Hada,
M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima,
Y. Honda, O. Kitao, H. Nakai, T. Vreven, K. Throssell,
J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark,
J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, T. Keith,
R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant,
Additional Supporting Information may be found online in the sup-
porting information tab for this article.
How to cite this article: A. P. M. de Souza, M. C. A. Costa,
A. R. de Aguiar, G. C. Bressan, G. D. de Almeida Lima,
W. P. Lima, M. P. G. Borsodi, B. R. Bergmann, M. M. C.
Ferreira, R. R. Teixeira, Arch. Pharm. 2021; e2100081.
https://doi.org/10.1002/ardp.202100081