Machine learning prioritizes synthesis of primaquine ureidoamides with high antimalarial activity and attenuated cytotoxicity

Article abstract of DOI:10.1016/j.ejmech.2018.01.062

Primaquine (PQ) is a commonly used drug that can prevent the transmission of Plasmodium falciparum malaria, however toxicity limits its use. We prepared five groups of PQ derivatives: amides 1a-k, ureas 2a-k, semicarbazides 3a,b, acylsemicarbazides 4a-k a

Full text of DOI:10.1016/j.ejmech.2018.01.062

Accepted Manuscript
Machine learning prioritizes synthesis of primaquine ureidoamides with high
antimalarial activity and attenuated cytotoxicity
Jurica Levatić, Kristina Pavić, Ivana Perković, Lidija Uzelac, Katja Ester, Marijeta
Kralj, Marcel Kaiser, Matthias Rottmann, Fran Supek, Branka Zorc
PII:
S0223-5234(18)30075-8
10.1016/j.ejmech.2018.01.062
EJMECH 10139
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 16 November 2017
Revised Date: 16 January 2018
Accepted Date: 18 January 2018
Please cite this article as: J. Levatić, K. Pavić, I. Perković, L. Uzelac, K. Ester, M. Kralj, M. Kaiser, M.
Rottmann, F. Supek, B. Zorc, Machine learning prioritizes synthesis of primaquine ureidoamides with
high antimalarial activity and attenuated cytotoxicity, European Journal of Medicinal Chemistry (2018),
doi: 10.1016/j.ejmech.2018.01.062.
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
Machine learning prioritizes synthesis of primaquine ureidoamides with high
antimalarial activity and attenuated cytotoxicity
a,b
c
c
d
d
Jurica Levatić , Kristina Pavić , Ivana Perković , Lidija Uzelac , Katja Ester , Marijeta Kralj
d
e
e
a, *
c, *
,
Marcel Kaiser , Matthias Rottmann , Fran Supek , Branka Zorc
a
Genome Data Science, Institute for Research in Biomedicine (IRB Barcelona), The Barcelona
Institute of Science and Technology (BIST). Baldiri Reixac 10, 08028, Barcelona, Spain
b
Department of Knowledge Technologies, Jožef Stefan Institute, Jamova cesta 39, SI-1000,
c
Ljubljana, Slovenia Faculty of Pharmacy and Biochemistry, University of Zagreb, A. Kovačića
1
, HR-10 000 Zagreb, Croatia
d
Laboratory of Experimental Therapy, Division of Molecular Medicine, Rudjer Bošković
Institute, Bijenička cesta 54, HR-10 000 Zagreb, Croatia
e
Parasite Chemotherapy, Medical Parasitology & Infection Biology, Swiss Tropical and Public
Health Institute, 4051 Basel, Switzerland
*
Corresponding authors.
E-mail addresses: fran.supek@irbbarcelona.org (F. Supek); bzorc@pharma.hr (B. Zorc).
1

ACCEPTED MANUSCRIPT
A B S T R A C T
Primaquine (PQ) is a commonly used drug that can prevent the transmission of Plasmodium
falciparum malaria, however its toxicity limits its use. We prepared five groups of PQ
derivatives: amides 1a-k, ureas 2a-k, semicarbazides 3a,b, acylsemicarbazides 4a-k and bis-
ureas 5a-v, and evaluated them for antimalarial activity in vitro against the erythrocytic stage of
P. falciparum NF54. Particular substituents, such as trityl (in 2j and 5r) and methoxybenzhydryl
(in 3b and 5v) were associated with a favorable cytotoxicity-to-activity ratio. To systematically
link structural features of PQ derivatives to antiplasmodial activity, we performed a quantitative-
structure activity relationship (QSAR) study using the Support Vector Machines machine
2
learning method. This yielded a highly accurate statistical model (R = 0.776 in cross-validation),
which was used to prioritize novel candidate compounds. Seven novel PQ-ureidoamides 10a-g
were synthesized and evaluated for activity, highlighting the benzhydryl ureidoamides 10e and
1
1
0f derived from p-chlorophenylglycine. Further experiments on human cell lines revealed that
0e and 10f are an order of magnitude less toxic than PQ in vitro while having antimalarial
activity indistinguishable from PQ. The toxicity profile of novel compounds 10 toward human
cells was particularly favorable when the glucose-6-phosphate dehydrogenase (G6PD) was
inhibited, while toxicity of PQ was exacerbated by G6PD inhibition. Our work therefore
highlights promising lead compounds for the development of effective antimalarial drugs that
may also be safer for G6PD-deficient patients. In addition, we provide computational inference
of antimalarial activity and cytotoxicity for thousands of PQ-like molecular structures.
Keywords: QSAR, Primaquine, Ureidoamide, Antimalarial activity, Plasmodium falciparum,
Glucose-6-phosphate dehydrogenase
2

ACCEPTED MANUSCRIPT
1
. Introduction
Malaria is an infectious disease affecting humans and animals, caused by parasitic protozoans
belonging to the genus Plasmodium. Of the five species that cause malaria in humans, P.
falciparum is responsible for most malaria-associated mortality worldwide and is the
predominant species in tropical and subtropical countries. The World Health Organization tracks
a decline in the global malaria burden due to more efficient drugs and better mosquito control.
However, malaria is still a life-threatening disease: in 2015 alone, there were an estimated 212
million new cases of malaria and 429 000 deaths [1].
Currently approved drugs include the artemisinins (artemisinin, dihydroartemisin, artemether,
artesunate) and quinoline derivatives (chloroquine, amodiaquine, mefloquine, quinine and
primaquine) [2]. Primaquine (PQ) is an 8-aminoquinoline antimalarial drug active against all
species causing human malaria, including multi-resistant P. falciparum strains. PQ is the only
clinically approved drug known to eliminate dormant liver forms of P. ovale and P. vivax [3]. In
the case of P. falciparum malaria, PQ is uniquely effective in killing gametocytes produced
during the sexual life stage of the parasite in the blood, disrupting the transmission of infection to
mosquitoes and slowing the spread of the disease. This means that the biological activity profile
of PQ (an 8-aminoquinoline) is different from standard 4-aminoquinoline antimalarial drugs such
as chloroquine (CQ) or amodiaquine. In particular, CQ is active against the asexual blood stages
(trophozoites) of P. falciparum, which feed on erythrocyte contents and therefore need to
detoxify ingested heme, a process that CQ interferes with. In contrast, PQ is less effective against
asexual blood stages of P. falciparum, but clears the sexual stage very efficiently. Therefore the
3

ACCEPTED MANUSCRIPT
underlying molecular mechanism of action of PQ likely differs from CQ and related drugs, and is
not currently well understood.
Because of the unique biological activity profile, PQ is complementary to other antimalarials and
is commonly added to drug regimens, with the primary goal of containing the spread of malaria
(in P. falciparum) or preventing relapse (for species with dormant liver stages, P. ovale and P.
vivax). However, toxicity of PQ is an important concern [4], particularly in patients with a
deficiency in the glucose-6-phosphate dehydrogenase (G6PD) enzyme – a widespread genetic
trait, particulary in disease-endemic regions. Globally 400 million people are estimated to be
affected with G6PD deficiency [5]. The highest prevalence is found in tropical Africa and
tropical/subtropical Asia: approx. 5-30%, depending on the country and the method used to
estimate the deficiency, reviewed in [6]. This constitutes a serious limitation to the broad use of
PQ as an antimalarial agent. In practice, either the PQ dosage must be reduced in order to
ameliorate side effects [7] or G6PD deficiency has to be tested in each patient before applying
the drug [8], presenting a substantial logistical challenge in broad application of PQ. Toxicity to
G6PD-deficient patients is a drawback specific to 8-aminoquinolines such as PQ and is not
known to be pertinent for other quinoline drugs. In order to overcome this drawback and improve
efficacy, various PQ derivatives have been synthesized and evaluated.
Recently published work bears witness that PQ derivatization and hybridization are interesting
approaches in search of new antimalarials [9–22]. In particular, we have previously designed and
prepared nearly a hundred primaquine derivatives of amide, urea, bis-urea, semicarbazide and
acylsemicarbazide type, showing that they possess antiproliferative effects against human cell
lines [23–29], however the antimalarial potential of these compounds has not been evaluated
until now. Here we report the results of screening of selected PQ derivatives against P.
4

ACCEPTED MANUSCRIPT
falciparum and a detailed QSAR study which guided the synthesis of seven new compounds of
general formula 10. They were active against malaria while exhibiting a very favorable toxicity
profile towards human cells in vitro, which was retained even when the G6PD enzyme was
inhibited.
2
. Results and discussion
2
.1. Chemistry
Six series of primaquine (PQ) derivatives were prepared and screened for antimalarial activity:
amides 1, ureas 2, semicarbazides 3, acylsemicarbazides 4, bis-ureas 5 and ureidoamides 10.
Their general structures are shown in Figure 1.
Fig. 1. Structures of the primaquine derivatives: amides 1a-k, ureas 2a-k, semicarbazides 3a,b,
acylsemicarbazides 4a-k, bis-ureas 5a-v and ureidoamides 10a-g.
Syntheses of derivatives 1-5 were previously published [23–29], while ureidoamides 10 are
novel compounds, described here for the first time. The synthetic pathway leading to compounds
1
0 is outlined in Scheme 1.
5

ACCEPTED MANUSCRIPT
Scheme 1. Synthesis of ureidoamides 10.
Novel ureidoamides 10a-g with primaquine and amino acid moieties were prepared by
aminolysis of N-(1-benzotriazolecarbonyl)-amino acid amides 9a-g with primaquine, while
amides 9 were obtained from N-(1-benzotriazolecarbonyl)-amino acid chlorides 8a-c and the
corresponding amines or hydrazones. Three amino acids (
chlorophenylglycine), four amines (p-bromoaniline,
chlorophenyl)(phenyl)methanamine)), N-methyl-1,1,diphenylmethanamine and one hydrazone
4-methoxybenzophenone hydrazone) were used. Synthesis of similar ureidoamides with various
L-leucine,
D
-phenylglycine, DL-p-
diphenylmethanamine,
(4-
(
amines and amino acids was previously described by our group [30,31], as well as the synthesis
of the starting N-(1-benzotriazolecarbonyl)-amino acids 7 and their chlorides 8 [32].
The presence of the primaquine residue in compounds 10a-g was confirmed by NMR spectra:
1
hydrogen atoms close to pyridine nitrogen occurred in H NMR spectra between 8.54 and 8.52,
aromatic hydrogen in p-position from pyridine nitrogen atom at 8.09-8.06, hydrogen bound to
nitrogen close to quinoline ring from 6.13 to 5.97, methoxy group at 3.82-3.81, CH of PQ chiral
6

ACCEPTED MANUSCRIPT
carbon as a multiplet at 3.62-3.57, and finally methyl group between 1.21 and 1.19 ppm. All NH
signals were D O exchangeable.
2
The presence of carbonyl groups in N-(1-benzotriazolecarbonyl)-amino acid amides 9a-g was
indicated by the appearance of two strong stretching vibration bands in IR spectra, the first
–
1
–1
between 1750 and 1716 cm and the second between 1682 and 1649 cm , while two carbonyl
–
1
groups in products 10 were located at lower wavenumbers (between 1650 and 1509 cm ).
1
3
In C NMR spectra carbonyl groups close to benzotriazole in products 9 appeared between
1
48.84 and 147.40 ppm and amide carbonyls between 172.20 and 167.39 ppm. Urea carbonyl
groups in products 10 were shifted to lower ppm values (from 157.62 to 153.85), while amide
carbonyl appeared between 172.46 and 169.45 ppm. C=N group in product 10g appeared at
1
55.00 ppm. Phenyl moieties gave expected signals in aromatic region and appropriate carbon
count. Structures of compounds 9f and 9g were confirmed indirectly, after chemical
derivatization to products 10f and 10g, respectively.
Mass spectroscopy gave molecular ion peaks which corresponded to expected molecular
formulas for all prepared compounds from series 10. In series 9, molecular ion peaks were
visible only in spectra of 9e and 9g. Other compounds of general formula 9 gave only
benzotriazole fragment at m/z 120.1 or 117.9. Chemical structures of compounds 10a-g were also
supported by CHN analyses.
2
.2. Antimalarial screening
The compound series 1-5 were evaluated for their antimalarial activity against the erythrocytic
stage of the Plasmodium falciparum NF54 strain, while an initial estimate of their cytotoxicity
towards mammalian cells was obtained using the L6 cell line derived from rat skeletal
7

ACCEPTED MANUSCRIPT
myoblasts. The IC50 values (the concentration of the tested compound necessary for 50% growth
inhibition) and cytotoxicity data are given in Table 1. As a simple measure of a compound’s
potential utility, we use a selectivity index (SI), here defined as ratio of the cytotoxicity (towards
L6 cells) and the antimalarial activity IC50 values. For reference, the SI of primaquine in our
experimental system (see Materials and Methods) was 39.6.
While the majority of tested compounds exhibited a certain level of antimalarial activity, some
series appeared more promising than others. For instance, all PQ-cinnamic acid derivatives of the
amide type (1a-k) showed modest antimalarial activity, with IC50 roughly an order of magnitude
higher than PQ, ranging between 8 and 25 µM; we note a somewhat more favorable profile in
the trimethoxy derivative 1e. Similarly, the PQ-cinnamic acid acylsemicarbazides 4a-k showed
modest activity overall, with one potential exception being the trimethoxy derivative 4e with SI
=
30 (antimalarial IC 4.5 µM and cytotoxicity 138.2 µM), thereby approximately matching the
50
parent compound (PQ). Therefore, series 1 and 4 do not overall appear promising, with certain
exceptions.
In contrast, other PQ derivative series had multiple members with favorable activity and
cytotoxicity, which were dependant on the substituents. In particular, the urea series 2a-2k
contained lowly active hydroxyalkyl and cycloalkyl derivatives, however the chlorobenzhydryl
urea 2i, a compound with two PQ moieties 2k and especially the trityl urea 2j were active and
moreover had a high SI (33, 28 and 113, respectively). Concerning the large series of bis-urea
derivatives 5a-v, hydroxyalkyl and cycloalkyl derivatives were again lowly active. Halogenated
aryl and biaryl derivatives had modest activity (IC from 7.5 to 17.8 µM) and exhibited some
50
cytotoxicity. The most interesting compound was again the trityl derivative 5r, which exhibited
SI
>
80 (IC50 1.2 µM and cytotoxicity
>
100 µM). Furthermore, compound 5v bearing a
8

ACCEPTED MANUSCRIPT
dimethoxybenzhydryl substituent was highly active (at PQ-like levels, 2.2 µM). The two
semicarbazides in series 3 (3a and 3b), which also bear methoxybenzhydryl groups, showed high
activity and low cytotoxicity as well.
In summary, the trityl compounds 2j and 5r represent active PQ derivatives with apparently
improved selectivity over the parent compound, PQ (SI 113 and 80, respectively, compared to SI
of primaquine 39.6), as estimated in our experimental system. The dimethoxybenzhydryl
compounds 3b and 5v also appear highly promising, with SI 78.3 and 61.8, respectively.
Motivated by these examples where favorable biological properties appear linked to certain
substituents, we further developed a QSAR model to direct the synthesis of a novel series of
antimalarial lead compounds based on PQ.
Table 1
In vitro screening of primaquine derivatives against erythrocytic stage of P. falciparum and
cytotoxicity towards L6 rat cells.
Antimalarial Cytotox-
a
a
b
Synth.
Ref.
QSAR
train/test set
Compd.
Structural formula
activity
IC , µM
icity
IC , µM
50
SI
c
5
0
1
a
15.1
>24.8
12.4
18.5
>250
>240
>220
1.2
[29]
[29]
[29]
[29]
Test
Test
Test
Test
1
b
10.1
19.4
18.5
1
c
1
d
11.9
9

ACCEPTED MANUSCRIPT
1
e
8.1
>210
25.9
1.3
[29]
[29]
Test
Test
Test
Test
1
f
22.1
13.7
13.2
29
16.8
3.5
1
g
1.2
[29]
1
h
0.3
[29]
1
i
15.6
12.9
11.3
27.4
52.4
16.2
160
1.0
12.4
14.0
n.d.
n.d.
[29]
[29]
[29]
[24]
[24]
Test
Test
1
j
1
k
158.1
Test
d
2
a
n.d.
Train
Train
2
b
n.d.
n.d.
n.d.
2
c
65.9
42.1
8.1
n.d.
n.d.
n.d.
n.d.
n.d.
[23]
[23]
[23]
[23]
[26]
Train
Train
Train
Train
Train
2
d
2
e
n.d.
n.d.
n.d.
2f
47.3
17.1
2
g
10

ACCEPTED MANUSCRIPT
2
h
11.9
4.3
14.9
140
1.3
[28]
Test
–
2
i
32.6
[26,27]
2
j
1.5
6.4
4.1
>171
>180
>195
114.0
28.1
47.6
[26]
[26]
[26]
Train
Train
Train
2
k
3
a
3
b
2.3
>180
78.3
[26]
Train
4
a
17.2
22.9
13.9
36.5
3.6
0.8
1.6
0.2
1.9
[29]
[29]
[29]
[29]
Test
Test
Test
Test
4
b
4
c
22.2
4
d
>19.7
37
4
e
4.5
138.2
30.7
[29]
Test
4
f
20.4
24.2
15.5
7.3
0.8
0.3
1.2
[29]
[29]
[29]
Test
Test
Test
4
g
4
h
>21.4
25.1
11

ACCEPTED MANUSCRIPT
4
i
>19.4
>19.4
10.6
28.3
0.5
1.5
[29]
[29]
Test
Test
4
j
4
k
>17.1
>170
9.9
[29]
Test
5
a
58.8
43.7
n.d.
n.d.
n.d.
n.d.
[26]
[26]
Train
Train
5
b
[
26,27]
5
c
41.4
178.1
4.3
–
5
d
66.7
75.4
46.3
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
[26]
[25]
[25]
Train
Train
Train
5
e
5
f
5
g
>111.0
n.d.
n.d.
[25]
Train
5
h
27.1
14.9
n.d.
n.d.
3.9
[25]
[28]
Train
Test
5
i
58.6
5
j
17.8
17.8
11.8
91.9
38.7
32.1
5.2
2.2
2.7
[28]
[28]
[28]
Test
Test
Test
5
k
5
l
12

ACCEPTED MANUSCRIPT
5
m
10.3
15.2
10.3
10.0
7.5
14.9
n.d.
n.d.
10.4
8.4
1.4
n.d.
n.d.
1.0
[28]
[26]
Test
Train
Train
Test
–
5
n
5
o
[25]
5
5
p
q
[26]
1.1
[26,27]
5
r
s
1.2
>100
83.3
[26]
Test
O
N
NH
O
N
O
H
H
H
N
3.7
76.6
n.d.
20.7
n.d.
[26,27]
[25]
–
5
N
N
N
N
H
H
O
5
t
20.8
Train
5
u
5.6
2.2
16.0
2.9
[26]
[26]
Train
Train
5
v
136.1
61.9
1
0a
3.4
6.5
3.0
>170
>140
>160
50.0 this work
21.5 this work
53.3 this work
Test
Test
Test
1
0b
1
0c
1
0d
3.2
54.0
16.9 this work
Test
13

ACCEPTED MANUSCRIPT
1
0e
1.4
1.8
2.7
>160
>150
114.3 this work
83.3 this work
Test
Test
Test
1
0f
1
0g
>120
59.5
44.4 this work
39.7
PQ
diphosphate
1
.5
a
IC50 – the concentration of the compound necessary for 50% growth inhibition of P.
falciparum or L6 rat myoblast cells; the displayed numbers are averages of two or three
b
independent experiments. SI - selectivity index, defined as a ratio of the cytotoxicity and the
c
d
antimalarial activity. Reference for synthesis. n.d. – not determined.
2
.3. Quantitative structure-activity relationship modeling
The QSAR model for prediction of antimalarial activity (log IC against the P. falciparum NF54
5
0
strain) was trained using the Support Vector Machines (SVM) algorithm. SVMs are a machine
learning method known to have a high predictive accuracy and to generalize well to novel data
sets and are therefore increasingly used also for QSAR studies [33], including our previous work
for modeling biological activity of peptides, ionophores, and substrate specificity of a human
protein [34–36]. Here, the SVM regression model was trained on a set of 56 compounds
(
structures provided in Supporting Information), which consisted of 23 PQ derivatives from this
study (series 1-5; Table 1) and additional 33 compounds from published data by Kaur et al. [37–
9] that were sufficiently similar in structure to the original 23 compounds; see Materials and
3
Methods for how the QSAR applicability domain was defined. The SVM model had a high
2
accuracy in predicting activity against P. falciparum when tested in cross-validation: R = 0.776;
root-mean-square error (RMSE) = 0.294 when predicting log10 IC50 units (Figure 2a). In other
14

ACCEPTED MANUSCRIPT
words, we estimate that our SVM model can predict the IC of a PQ derivative within a 1.97-
5
0
0
.294
fold range of the actual value (10
). Moreover, the model retained much of the predictive
accuracy when tested on an independent test set of 37 PQ derivatives that are within the model’s
applicability domain (see Materials and Methods), yielding a RMSE = 0.403 log units of IC
50
1
0
(
Figure 2b), implying a good qualitative agreement of the model to observed data. If the
compounds are classed into two groups, lowly active (above-median IC50 ≥ 8.9 M) and highly
active (below-median IC50 < 8.9 M) compounds, our predictions would be 91% accurate on the
µ
µ
training set, and 95% accurate on the independent test set (measuring the fraction compounds
assigned to the correct class; Figure 2d). Therefore, this appears to be a useful statistical model
to prioritize the synthesis of further PQ derivatives with antiplasmodial activity.
15

ACCEPTED MANUSCRIPT
Fig. 2. Measured and predicted activity of 56 training set compounds (A) and 37 testing
compounds (B); Predicted activities of an in house library of 522 compounds, sorted from the
most active to the least active compound (C); Qualitative agreements of the model’s predictions
with the observed data (D). Novel ureidoamides 10a-g are marked with red dots. Predicted
activities for the training compounds were obtained in ten-fold cross-validation.
Next, the SVM model was applied to infer the antimalarial activity of an in-house virtual library
of 522 molecules; the structures and activity predictions are supplied in Supporting Information.
Based on their high predicted antimalarial activity (Figure 2c) and further considerations related
to chemical synthesis, seven PQ ureidoamides 10a-g bearing an amino acid residue were selected
for synthesis and evaluation against P. falciparum and for an initial cytotoxicity screen using the
L6 mammalian cell line.
2
.4. Antimalarial activity and cytotoxicity of novel PQ ureidoamides
Indeed, four out of seven compounds from the series 10 showed a favorable
cytotoxicity/antimalarial activity ratio, with SI > 50 (PQ has SI = 39.6). The benzhydryl
ureidoamide derived from p-chlorophenylglycine 10e was the most promising of all 64 tested
compounds, with SI
compounds 10e and 10f have very similar antiplasmodial activity as the parent compound, PQ
Table 1), they are at least 2.7 and 2.5-fold less cytotoxic relative to PQ in this particular in vitro
> 114, followed by chlorobenzhydryl derivative 10f with SI > 83. While the
(
model using L6 rat skeletal myoblasts. Of note, these are conservative estimates, given that the
L6 experimental assay reached the boundary of its detection range; in other words, the actual
toxicity of 10e and 10f may be even lower.
16

ACCEPTED MANUSCRIPT
To follow up on the initial observation of reduced cytotoxicity in rat L6 cells, we performed
growth inhibition screens on four human tumor cell lines from diverse tissues: H460 (lung),
MCF-7 (breast), SW620 (colon) and A2780 (ovary); Table 2. For the novel compounds in series
1
0, human cell lines suggest at least an order of magnitude reduced cytotoxicity compared to PQ
(Tables 2 and S1). Compound 10e in particular did not reach IC50 up to the maximum tested
concentration of 100 µM in three of the four cell lines, and similarly so for 10f. Therefore, 10e
and 10f are at least 13-fold less toxic than PQ, which has median IC50 across cell lines 7.5 µM
(range 5 to 14 µM) in this model system (Table 2). Compound 10c also exhibits a favorable
toxicity profile (non-toxic up to 100 µM in all four human cell lines; Table 2) and retains much
of the activity (P. falciparum IC = 3.0 µM). In this respect, the novel compounds 10 appear to
5
0
be a large improvement over other PQ derivative series tested here: while the above-reported
compounds 3b, 5r and 5v exhibited PQ-like antimalarial activity (Table 1), they are also
similarly cytotoxic to human cells as PQ (Table 2), in stark contrast to 10c, 10e and 10f which
are substantially less cytotoxic than PQ. Of note, our data for 10c, 10e and 10f were measured
across five independent experiments (biological replicates) for the H460 and MCF-7 cell lines
and at least two experiments in SW620 and A2780 cells (Table 2), suggesting robust results. To
further examine effects on a broader panel of human tissues, we also tested against the tumor-
derived cell lines HCT 116 (colon), HL-60 and K562 (blood/myeloid lineage) and Capan-1
(pancreatic). All compounds from series 10 had very low cytotoxicity also on all these additional
cell lines (IC >100 µM; single biological replicate), unlike PQ (Table S1).
5
0
Table 2
a
IC values (in µM; mean ± standard deviation) for human cell lines
5
0
17

ACCEPTED MANUSCRIPT
Cell lines
Compd.
H460
MCF-7 SW620
A2780
c
>
100b 3 ± 0.4
>100
1 ± 0.1
(n = 3)
3
b
(
n = 3)
(n = 2)
(n = 2)
c
3
± 0.3
3 ± 2
7 ± 4
1 ± 0.2
5
5
r
(
n = 2)
(n = 2)
(n = 2)
(n = 2)
c
5
(
± 0.3 1 ± 0.5
n = 2)
> 100
0.06 ± 0.01
(n = 3)
v
(n = 2)
(n = 2)
>
100
> 100
(n = 4)
> 100
>100
(n = 3)
>100
36±29
(n = 3)
1
0a
0b
(
6
(
n = 5)
4 ± 32
n = 2)
d
1
n. t.
(n = 2)
(n = 3)
>
100
> 100
(n = 5)
>100
(n = 3)
> 100
(n = 2)
1
0c
0d
0e
0f
0g
PQ
(
n = 5)
5
(
2 ± 31 16 ± 0.2 32 ± 8
1
n. t.
n = 2)
100
n = 5)
100
n = 5)
100
n = 2)
(n = 2)
(n = 3)
>
(
> 100
(n = 5)
6 ± 1
(n = 3)
> 100
(n = 2)
1
>
(
> 100
(n = 5)
> 100
(n = 3)
14 ± 4
(n = 2)
1
>
(
> 100
(n = 2)
> 100
(n = 3)
1
n. t.
1
0 ± 5
5 ± 2
(n = 6)
14 ± 2
(nb = 3)
5 ± 0.2
(n = 2)
(
n = 3)
a
IC50 – the concentration that causes 50% growth inhibition. Number of independent experiments (biological
replicates) given in parentheses. Values previously reported by Pavić et al. [23]. n. t. – not tested.
c
d
2
.5. Cytotoxicity in G6PD-deficient cells
One important mechanism of PQ toxicity in patients is related to the G6PD enzyme, wherein the
persons with an inherited deficiency of G6PD may suffer severe side effects, in particular
hemolytic anemia [40,41]. We therefore asked whether the favorable cytotoxicity profile of
novel compounds 10 is retained or abolished in G6PD-deficient cells, in comparison to PQ and
the most promising previous PQ derivatives from series 3 and 5. To investigate, we applied a
18

ACCEPTED MANUSCRIPT
G6PD inhibitor 6-aminonicotinamide (6-AN) to the human cell lines H460 and MCF-7, thereby
reducing G6PD activity in the cells. We observed that in highly active PQ derivatives 3b, 5r and
5
v, the cytotoxicity is aggravated upon G6PD inhibition (Figures 3c and 3d), similarly as for PQ
itself (Figures 3a and 3b).
In particular, applying 10 µM of PQ by itself lowers the MCF-7 cell growth rate to 25% of the
untreated cells (meaning, growth is reduced by three-quarters, but there is still a net increase in
number of cells; Figures 3a and 3b). However, the same PQ concentration in combination with
6
-AN (at 10 µM) results in a negative growth rate of approx. –45%, meaning there is much cell
death occurring caused by PQ in the G6PD-inhibited cells. In other words, there is a synergistic
effect on cytotoxicity by PQ and G6PD inhibition (Figures 3a and 3b). In contrast, a highly
cytotoxic cancer drug not known to act in a G6PD-dependent manner, paclitaxel, does not show
synergies with 6-AN (Figures 3a and 3b). The toxicity of 6-AN itself – always applied at 10 µM
–
for the H460 and MCF-7 cell lines is modest and thus unlikely to confound results (82%
growth rate for MCF-7 and 87% for H460 for 6-AN only).
We next turned to examine the effects of G6PD inhibition on the cytotoxicity of the novel
compounds 10a, 10c, 10e and 10f in human cell lines (Figures 3e and 3f). There was little
difference in growth inhibition of the H460 and MCF-7 cells regardless of if the G6PD inhibitor
was simultaneously applied or not. In particular, at 10 µM, compound 10e results in >50% MCF-
7
cell growth with or without 6-AN (80% vs. 65%) and equally so with the compound 10f (93%
vs. 76%). The observable difference is in line with the effect size due to applying 6-AN itself
approx. 15%) and therefore there does not appear to be a synergistic effect of 10e and 10f with
(
G6PD inhibition, unlike PQ (25% vs. –45%, which amounts to a 70% difference) and unlike the
previous compounds 3b, 5r and 5v (Figure 3c and 3d). The compounds 10a and 10c show a
19

ACCEPTED MANUSCRIPT
similar pattern, indicating that G6PD inhibition does not substantially aggravate their toxicity
(
Figures 3e and 3f). H460 cells broadly uphold these results for 10a, 10c and 10f. Overall, our
data suggest that the novel PQ derivatives synthetized herein are promising lead compounds for
antimalarial drugs with an improved safety profile for G6PD-deficient patients. Testing of
compounds 10 and their derivatives in other cell types and in model animals is therefore
warranted to further establish the medicinal potential of these novel PQ derivatives.
Fig. 3. The effects of G6PD inhibition on the cytotoxicity of the compounds to MCF-7 and H460
cell lines; controls (A and B), PQ-derivatives (C and D) and newly synthetized ureidoamides (E
and F).
2
.6. A comprehensive resource of predicted antimalarial activity and cytotoxicity of PQ-like
compounds
20

ACCEPTED MANUSCRIPT
Our efforts to computationally model the structure-activity relationship of PQ derivatives
resulted in a highly accurate predictor based on the SVM algoritm. We were thus further
motivated to employ computational modeling to provide a library of predictions for a broad set
of 13,401 PQ-like molecular structures extracted from PubChem; see Supporting Text S1 for a
description of the modeling, and see Supporting Information for the inferred biological activities.
In brief, we predicted 199 compounds to be simultaneously more active and less cytotoxic than
PQ itself (predictions listed in Supporting Information). While we took precautions to ensure that
the molecular structures of the >13,000 tested compounds are reasonably consistent with the
training set of PQ derivatives (see section 4.3.3), we advise caution in interpreting the
predictions, given that QSAR models tend to become less reliable as structures diverge. We
recovered individual examples of correct predictions in this data set, as suggested by a literature
search (see Supporting Text S1 and Figure S1). If further validated by independent data, this
database of predicted activities could serve as a useful resource to inspire future research on
antimalarial PQ derivatives.
3
. Conclusions
PQ and its derivatives are important drugs for treatment of various types of malaria, where in
particular PQ has a unique ability to prevent P. falciparum transmission by killing the
gametocytes of the parasite in the bloodstream. However, applying PQ at a broad scale is
complicated by severe toxicity towards a subset of patients with an inherited G6PD deficiency.
Therefore, there is a need for novel derivatives of PQ which would retain or improve its efficacy,
while presenting a more favorable toxicity profile. Here, we evaluated the activity of 64 diverse
PQ derivatives against the erythrocytic stage of the P. falciparum NF54 strain. PQ-cinnamic acid
amides 1d and 1e, chlorobenzhydryl 2i, trityl 2j and 2k urea with two PQ residues, PQ-cinnamic
21

ACCEPTED MANUSCRIPT
acid semicarbazide 4e, the trityl bis-urea 5r and the methoxybenzhydryl compounds 3b and 5v
showed high activities. Based on these experimental measurements, a QSAR model prioritized a
novel series of PQ ureidoamides 10a-g that we designed, prepared and evaluated against P.
falciparum and mammalian cells. The results confirmed high activity and low cytotoxicity of
ureidoamides 10, especially of benzhydryl and chlorobenzhydryl compounds 10e and 10f
derived from p-chlorophenylglycine. These showed antimalarial activity comparable with PQ,
but at least an order of magnitude lower cytotoxicity to various human cell types and in
particular their toxic effects were not notably aggravated in G6PD-inhibited cells, in contrast to
PQ. We suggest that 10a-g constitute promising lead compounds for development of more
effective and safer drugs for malaria treatment, particularly for G6PD-deficient patients.
Furthermore, our computational models predict hundreds of additional PQ-related compounds
with favorable antimalarial activity and cytotoxicity profiles, highlighting new avenues for future
research efforts.
4
4
4
. Materials and methods
.1. Synthesis
.1.1. General information
Melting points were measured on a Stuart Melting Point (SMP3) apparatus (Barloworld
Scientific, UK) in open capillaries with uncorrected values. IR spectra were recorded on FTIR
Perkin Elmer Spectrum One and UV-Vis spectra on Lambda 20 double beam spectrophotometer
1
13
(
Perkin-Elmer, UK). All NMR ( H and C) were recorded at 25
°C on NMR Avance 600
1
13
spectrometer (Bruker, Germany) at 300 and 150 MHz for H and C nuclei, respectively.
22

ACCEPTED MANUSCRIPT
Chemical shifts (
δ
) are reported in parts per million (ppm) using tetramethylsilane as reference in
1
13
the H and the DMSO residual peak as reference in the C spectra (39.51 ppm). Coupling
constants (J) are reported in Hertz (Hz). Mass spectra were recorded on HPLC-MS/MS (HPLC,
Agilent Technologies 1200 Series; MS, Agilent Technologies 6410 Triple Quad). Mass
determination was realized using electron spray ionization (ESI) in positive or negative mode.
CEM Discover microwave reactor was used for microwave reactions (P = 150 W, t = 70 °C).
Elemental analyses were performed on a CHNS LECO analyzer (LECO Corporation, USA).
Found values for carbon, hydrogen and nitrogen were within ± 0.4% of the theoretical values.
All compounds were routinely checked by TLC with Merck silica gel 60F-254 glass plates using
the following solvent systems: petrolether/ethyl acetate/methanol 30:10:5, cyclohexane/ethyl
acetate
1:1,
cyclohexane/
dichloromethane/methanol
10:18:2,
dichloromethane/dichloromethane/methanol 97:3 and 95:5. Spots were visualized by short-wave
UV light and iodine vapour. Column chromatography was performed on silica gel 0.063–0.200
mm (Kemika, Croatia) and 0.040–0.063 mm (Merck, Germany), with the same eluents used in
TLC.
1
H-benzo[d][1,2,3]triazole (BtH), triphosgene, triethylamine (TEA), hydrazine hydrate,
L-
leucine, -phenylglycine, DL-p-chlorophenylglycine, p-bromoaniline and N-methyl-1,1-
D
diphenylmethanamine were purchased from Sigma-Aldrich. Primaquine, diphenylmethanamine
and (4-chlorophenyl)(phenyl)methanamine were prepared from commercially available salts
(Sigma-Aldrich) prior the use. All reactions with primaquine were run light protected. 1-
Benzotriazole carboxylic acid chloride (BtcCl, 6) was prepared according to our procedures
published earlier [42,43]. 4-Methoxybenzophenone hydrazone was prepared from 4-
23

ACCEPTED MANUSCRIPT
methoxybenzophenone and hydrazine hydrate [44]. Primaquine derivatives 1-5 were prepared
according to the previously published procedures [23–29].
4
.1.2. Btc-amino acids (7a-c): general procedure
To a suspension of 20 mmol of amino acid in dry dioxane (40 ml) a solution of 1.81 g (10 mmol)
BtcCl (6) in dry dioxane (10 ml) was added dropwise. The reaction mixture was stirred overnight
at room temperature. Amino acid hydrochloride was filtered off, washed with dioxane and the
mother liquor was evaporated under reduced pressure. The crude products were triturated several
times with hot toluene. 2-[(1H-1,2,3-benzotriazole-1-carbonyl)amino]-4-methylpentanoic acid
(
Btc-
-phenylglycine, 7b) were described in our previous paper [32], while 2-[(1H-1,2,3-
benzotriazole-1-carbonyl)amino]-2-(4-chlorophenyl)acetic acid (Btc-DL-p-chlorophenylglycine,
L-leucine, 7a) and 2-[(1H-1,2,3-benzotriazole-1-carbonyl)amino]-2-phenylacetic acid (Btc-
D
7
c) is a new compound.
.1.2.1. 2-[(1H-1,2,3-benzotriazole-1-carbonyl)amino]-2-(4-chlorophenyl)acetic
4
acid
(Btc-DL-p-chlorophenylglycine (7c)
From the reaction of 3.712 g p-chlorophenylglycine and 1.81 g chloride 6, 3.274 g (99%) of 7c
o
was obtained; mp 127 C (decomp.); IR (KBr):
ν
max 3360, 2924, 2540, 1750, 1712, 1494, 1450,
1
5
8
8
7
1
380, 1290, 1184, 1136, 1064, 1018, 976, 932, 822, 786, 754, 732, 680, 662, 642, 624, 600, 558,
–
1 1
38, 516, 466 cm ; H NMR (DMSO-d ) δ 9.63-9.62 (d, J = 7.21 Hz, 1H), 8.24-8.22 (d, J =
6
.35 Hz, 1H), 8.16-8.14 (d, J = 8.30 Hz, 1H), 7.75-7.71 (t, J = 7.56 Hz, 1H), 7.63-7.61 (d, J =
.46 Hz, 2H), 7.59-7.55 (t, J = 7.72 Hz, 1H), 7.49-7.47 (d, J = 8.45 Hz, 2H), 7.18-7.17 (d, J =
1
3
.24 Hz, 1H), 5.71-5.69 (d, J = 7.20 Hz, 1H); C NMR (DMSO-d )
δ
170.66, 148.50, 145.48,
6
35.76, 132.81, 131.25, 130.22, 130.14, 128.41, 125.76, 119.94, 113.38, 56.69; ESI-MS; m/z
24

ACCEPTED MANUSCRIPT
+
calculated for C15
H11ClN
4
O
3
: 330.05, found: 353.2 (M+23) ; calculated for C
6
H
9
N
3
: 119.12,
+
found: 120.1 (BtH+1) .
4
.1.3. Btc-amino acid chlorides (8a-c): general procedure
A solution of 4 mmol Btc-amino acid in 20 ml thionyl chloride was stirred overnight at room
temperature. Thionyl chloride was evaporated under reduced pressure and the residue was
dissolved in dry toluene and evaporated again (several times). Crude products 8a-c were used
immediately in further reactions.
4
.1.4. Btc-amino acid amides (9a-g): general procedure
A solution of 3.4 mmol corresponding amine or hydrazone and 4 mmol TEA in dry toluene (20
ml) was added dropwise to a solution of 4 mmol chloride 8 in dry toluene (30 ml). The reaction
mixture was stirred 30 min at room temperature. TEA × HCl was filtered off and mother liquor
was evaporated. The residue was dissolved in ethyl acetate/1% HCl mixture (1:1). The organic
layer was extracted three times with HCl solution (w = 1%), washed two times with water, dried
over anhydrous sodium sulfate, filtered and evaporated under reduced pressure.
4
.1.4.1.
methylpentanamide (9a
From the reaction of 0.623 g diphenylmethanamine, 1.179 g chloride 8a and 0.404 g TEA, and
after trituration with ether, 1.081 g (72%) of 9a was obtained; mp 155–158 °C; IR (KBr):
2-[(1H-1,2,3-benzotriazole-1-carbonyl)amino]-N-(diphenylmethyl)-4-
)
ν
max
3
1
328, 3261, 3068, 2958, 2929, 1716, 1650, 1557, 1522, 1450, 1378, 1287, 1230, 1152, 1079,
–
1 1
050, 1013, 925, 855, 836, 753, 699, 633, 567, 521, 462 cm ; H NMR (DMSO-d )
δ 9.12-9.09
6
(2d, J = 3.95 Hz, 2H'), 8.23-8.18 (m, 2H), 7.76-7.72 (t, J = 7.71 Hz, 1H), 7.58-7.55 (t, J = 7.68
25

ACCEPTED MANUSCRIPT
Hz, 1H), 7.37-7.24 (m, 10H), 6.18-6.16 (d, J = 8.37 Hz, 1H), 4.71-4.65 (m, 1H), 1.94-1.88 (m,
1
3
1
1
1
H), 1.73-1.62 (m, 2H), 0.95-0.89 (d, J = 6.48 Hz, 6H); C NMR (DMSO-d ) δ 170.58, 148.84,
6
45.47, 142.19, 142.08, 131.32, 130.06, 128.40, 128.34, 127.49, 127.24, 127.05, 125.65, 119.84,
13.54, 56.08, 52.97, 38.89, 24.46, 23.08, 21.38 [45].
4
.1.4.2.
2-[(1H-1,2,3-benzotriazole-1-carbonyl)amino]-N-(diphenylmethyl)-N-methyl-2-
phenylacetamide (9b
)
From the reaction of 0.671 g N-methyl-1,1-diphenylmethanamine, 1.259 g chloride 8b and 0.404
g TEA, and after purification by column chromatography (mobile phase cyclohexane/ethyl
o
acetate/methanol 30:10:5), 0.857 g (53%) of 9b was obtained; mp 82–83.5 C; IR (KBr): νmax
3
6
7
1
1
1
4
377, 3032, 2939, 1735, 1649, 1490, 1405, 1294, 1231, 1081, 1033, 932, 867, 813, 753, 702,
–
1 1
11, 569, 519 cm ; H NMR (DMSO-d )
δ
9.16-9.14 (d, J = 5.97 Hz, 1H), 8.22-8.14 (m, 2H),
6
.75-7.70 (t, J = 8.27 Hz, 1H), 7.54 (m, 3H), 7.43-7.21 (m, 11H), 7.02-6.93 (m, 2H), 6.53 (s,
1
3
H), 6.16-6.14 (d, J = 6.23 Hz, 1H), 2.67 (s, 3H); C NMR (DMSO-d6)
δ 169.24, 147.40,
45.50, 138.47, 138.24, 135.92, 131.08, 130.26, 128.80, 128.60, 128.51, 128.48, 128.42, 128.28,
27.51, 125.71, 119.97, 113.28, 61.00, 55.58, 31.69.
.1.4.3.
2-[(1H-1,2,3-benzotriazole-1-carbonyl)amino]-N-[(4-
chlorophenyl)(phenyl)methyl]-2-phenylacetamide (9c
From the reaction of 0.740 g (4-chlorophenyl)(phenyl)methanamine, 1.259 g chloride 8b and
.404 g TEA, and after trituration with ether/petroleum ether and purification of mother liquid by
)
0
column chromatography (mobile phase cyclohexane/ethyl acetate/methanol 30:10:5), 1.400 g
o
(
83%) of 9c was obtained; mp 116–118.5 C; IR (KBr): νmax 3286, 3062, 3032, 1748, 1660,
1
5
502, 1499, 1448, 1380, 1232, 1182, 1126, 1094, 1036, 934, 852, 812, 752, 698, 612, 572, 554,
–
1 1
18, 496 cm ; H NMR (DMSO-d )
δ 9.41-9.39 (t, J = 7.41 Hz, 1H), 9.03-9.01 (d, J = 7.14 Hz,
6
26

ACCEPTED MANUSCRIPT
1
H), 8.23-8.21 (d, J = 8.29 Hz, 1H), 8.15-8.13 (dd, J = 2.6 Hz, 8.3 Hz, 1H), 7.74-7.71 (t, J = 7.60
Hz, 1H), 7.61-7.60 (d, J = 7.81 Hz, 2H), 7.87-7.55 (t, J = 7.91 Hz, 1H), 7.43-7.35 (m, 6H), 7.32-
7
5
1
1
.28 (m, 2H), 7.25-7.21 (m, 2H), 7.11-7.08 (t, J = 7.37 Hz, 2H), 6.19-6.17 (d, J = 8.10 Hz, 1H),
1
3
.84-5.81 (t, J = 6.46 Hz, 1H); C NMR (DMSO-d )
δ 172.20, 148.06, 145.50, 141.21, 140.76,
6
35.60, 132.20, 131.15, 130.21, 129.40, 128.79, 128.51, 128.26, 128.19, 128.05, 128.00, 126.72,
25.69, 119.91, 113.33, 59.56, 56.54.
4
.1.4.4.
2-[(1H-1,2,3-benzotriazole-1-carbonyl)amino]-N-(4-bromophenyl)-2-(4-
chlorophenyl)acetamide (9d
)
From the reaction of 0.585 g p-bromoaniline, 1.397 g chloride 8c and 0.404 g TEA, and after
purification by column chromatography (mobile phase cyclohexane/ethyl acetate/methanol
3
0:10:5) and trituration with ether/petroleum ether and toluene/petroleum ether, 0.989 g (60%) of
o
9
d was obtained; 168.5–169.5 C; IR (KBr):
ν
max 3406, 3232, 3062, 1732, 1656, 1600, 1498,
–
1
452, 1396, 1346, 1290, 1218, 1092, 1014, 930, 818, 778, 754, 724 ,656, 628, 594, 540, 502 cm
1
1
;
H NMR (DMSO-d6) δ 10.63 (s, 1H), 9.48-9.47 (d, J = 7.14 Hz, 1H), 8.24-8.22 (d, J = 8.31
Hz, 1H), 8.16-8.14 (d, J = 8.33 Hz, 1H), 7.75-7.71 (t, J = 7.55 Hz, 1H), 7.67-7.65 (d, J = 8.50
13
Hz, 2H), 7.60-7.57 (t, J = 7.38 Hz, 3H), 7.55-7.48 (m, 4H), 5.83-5.82 (d, J = 7.18 Hz, 1H); C
NMR (DMSO-d 167.39, 148.52, 145.48, 137.78, 135.92, 133.04, 131.72, 131.24, 130.22,
29.80, 128.62, 125.77, 121.47, 119.95, 115.62, 113.39, 57.58.
.1.4.5. 2-[(1H-1,2,3-benzotriazole-1-carbonyl)amino]-2-(4-chlorophenyl)-N-
6
) δ
1
4
(diphenylmethyl)acetamide (9e)
From the reaction of 0.623 g diphenylmethanamine, 1.397 g chloride 8c and 0.404 g TEA, and
after trituration several times with ether/petroleum ether, 0.640 g (38%) of 9e was obtained; mp
o
1
64.5–166.5 C; IR (KBr): νmax 3388, 3276, 3062, 1732, 1682, 1648, 1492, 1448, 1380, 1288,
27

ACCEPTED MANUSCRIPT
–
1 1
1
9
8
7
7
6
218, 1094, 1024, 930, 818, 750, 702, 648, 590, 566, 540 cm ; H NMR (DMSO-d ) δ 9.41-
.40 (d, J = 8.23 Hz, 1H), 9.10-9.09 (d, J = 7.29 Hz, 1H'), 8.23-8.21 (d, J = 8.36 Hz, 1H), 8.14-
.12 (d, J = 8.31 Hz, 1H), 7.74-7.71 (t, J = 7.85 Hz, 1H), 7.64-7.62 (d, J = 8.52 Hz, 2H), 7.57-
.54 (t, J = 8.59 Hz, 1H), 7.47-7.46 (d, J = 8.04 Hz, 2H), 7.37-7.35 (m, 2H), 7.32-7.31 (d, J =
.07 Hz, 2H), 7.29-7.28 (m, 1H), 7.27-7.20 (m, 3H), 7.10-7.08 (d, J = 7.01 Hz, 2H), 6.17-6.15
1
3
(
d, J = 8.20 Hz, 1H), 5.84-5.83 (d, J = 7.23 Hz, 1H); C NMR (DMSO-d )
δ
167.71, 148.10,
45.47, 141.62, 141.60, 136.63, 132.79, 131.15, 130.18, 129.27, 128.44, 128.42, 128.29, 127.55,
27.22, 127.04, 126.85, 125.69, 119.90, 113.31, 56.55, 56.34.
.1.4.6. 2-[(1H-1,2,3-benzotriazole-1-carbonyl)amino]-2-(4-chlorophenyl)-N-[(4-
6
1
1
4
chlorophenyl)(phenyl)methyl]acetamide (9f
From the reaction of 0.740 g (4-chlorophenyl)(phenyl)methanamine, 1.397 g chloride 8c and
.404 g TEA, and after purification by column chromatography (mobile phase cyclohexane/ethyl
)
0
acetate/methanol 30:10:5) and trituration with ether/petroleum ether, 0.307 g (17%) of 9f was
o
obtained; mp 158–159.5 C; IR (KBr): νmax 3394, 3279, 3062, 1730, 1651, 1492, 1450, 1410,
–
1
1
381, 1321, 1290, 1219, 1184, 1150, 1093, 1017, 930, 828, 753, 702, 623, 563, 537 cm . The
structure of product 9f was confirmed indirectly, by chemical derivatization to compound 10f.
.1.4.7. N-[(4-chlorophenyl)({N'-[(4-
methoxyphenyl)(phenyl)methylidene]hydrazinecarbonyl})methyl]-1H-1,2,3-benzotriazole-1-
carboxamide (9g
4
)
From the reaction of 0.769 g 4-methoxybenzophenone hydrazone, 1.397 g chloride 8c and 0.404
g TEA, and after purification by column chromatography (mobile phase cyclohexane/ethyl
acetate/methanol 30:10:5) and trituration with ether, 1.136 g (62%) of 9g was obtained; IR
(
KBr): νmax 3373, 3307, 3181, 3076, 3049, 2923, 2829, 1733, 1676, 1604, 1506, 1489, 1443,
28

ACCEPTED MANUSCRIPT
1
6
377, 1305, 1289, 1256, 1179, 1157, 1108, 1089, 1031, 960, 935, 836, 806, 779, 749, 696, 653,
–
1
28, 603, 570, 532 cm . The structure of product 9g was confirmed indirectly, by chemical
derivatization to compound 10g.
4
.1.5. PQ-ureidoamides (10a-g): general procedure
Method A: NaOH solution (w = 5%) was added to a solution of 1 mmol PQ diphosphate in water
until pH 9-10 was reached, and PQ base was extracted four times with dichloromethane and
washed with water until pH 7. The organic layer was dried over anhydrous sodium sulfate,
filtered and evaporated under reduced pressure. PQ base and 0.5 mmol Btc-amino acid amide 9
were mixed in dioxane and stirred overnight at room temperature, light protected. The solvent
was evaporated under reduced pressure.
Method B: A suspension of 0.4 mmol PQ diphosphate, 0.4 mmol Btc-amino acid amide 9b and
0
.8 mmol TEA in dichloromethane (3 mL) was heated at 70 °C in microwave reactor for 45 min.
The solvent was removed under reduced pressure. The residue was dissolved in ethyl acetate/5%
NaOH mixture (1:1). The organic layer was extracted with 5% NaOH three times, washed with
water two times, dried over anhydrous sodium sulfate, filtered and evaporated.
4
.1.5.1.
N-(diphenylmethyl)-2-[({4-[(6-methoxyquinolin-8-
yl)amino]pentyl}carbamoyl)amino]-4-methylpentanamide (10a
)
Method A, from the reaction of 0.221 g Btc-amino acid amide 9a and 0.259 g PQ, and after
purification by column chromatography (mobile phase cyclohexane/ethyl acetate/methanol
3
0:10:5) and trituration with ether, 0.253 g (87%) of 10a was obtained; mp 184.5–185.5 °C; IR
(
KBr): νmax 3316, 2958, 1655, 1624, 1569, 1519, 1493, 1454, 1422, 1388, 1336, 1290, 1201,
−1 1
1
6
167, 1155, 1052, 1028, 824, 791, 746, 697 cm ; H NMR (DMSO-d ) δ 8.89-8.87 (d, J = 8.53
29