1,3,5-triazaspiro[5.5]undeca-2,4-dienes as selective Mycobacterium tuberculosis dihydrofolate reductase inhibitors with potent whole cell activity

Article abstract of DOI:10.1016/j.ejmech.2017.12.017

The emergence of multi- and extensively-drug resistant tubercular (MDR- and XDR-TB) strains of mycobacteria has limited the use of existing therapies, therefore new drugs are needed. Dihydrofolate reductase (DHFR) has recently attracted much attention as

Full text of DOI:10.1016/j.ejmech.2017.12.017

European Journal of Medicinal Chemistry 144 (2018) 262e276
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Research paper
1,3,5-triazaspiro[5.5]undeca-2,4-dienes as selective Mycobacterium
tuberculosis dihydrofolate reductase inhibitors with potent whole cell
activity
,
,
Xuan Yang a 1, Wassihun Wedajo b 1, Yoshiyuki Yamada b, Sue-Li Dahlroth c,
Jason Jun-Long Neo c, Thomas Dick b *, Wai-Keung Chui a
,
d,
, **
a Department of Pharmacy, Faculty of Science, National University of Singapore, 18 Science Drive 4, 117543, Singapore
b Department of Microbiology and Immunology, Yong Loo Lin School of Medicine, National University of Singapore, 5 Science Drive 2, 117545, Singapore
c Protein Production Platform, School of Biological Sciences, Nanyang Technological University, 138673, Singapore
d Public Health Research Institute, New Jersey Medical School, Rutgers University, 225 Warren Street, Newark, NJ07103, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The emergence of multi- and extensively-drug resistant tubercular (MDR- and XDR-TB) strains of
mycobacteria has limited the use of existing therapies, therefore new drugs are needed. Dihydrofolate
reductase (DHFR) has recently attracted much attention as a target for the development of anti-TB
agents. This study aimed to develop selective M. tuberculosis DHFR inhibitors using rationale scaf-
folding design and synthesis, phenotype-oriented screening, enzymatic inhibitory study, whole cell on-
target validation, molecular modeling, and in vitro DMPK determination to derive new anti-TB agents.
2,4-diamino-1-phenyl-1,3,5-triazaspiro[5.5]undeca-2,4-dienes 20b and 20c were identified as selective
Received 29 September 2017
Received in revised form
17 November 2017
Accepted 5 December 2017
Available online 7 December 2017
Keywords:
M. tuberculosis DHFR inhibitors, showing promising antimycobacterial activities (MIC50: 0.01
mM and
Dihydrofolate reductase (DHFR)
Antimycobacterial agent
Tuberculosis (TB)
MIC90: 0.025 M on M. tuberculosis H37Rv). This study provided compelling evidence that compound 20b
m
and 20c exerted whole cell antimycobacterial activity through DHFR inhibition. In addition, these two
compounds exhibited low cytotoxicity and low hemolytic activity. The in vitro DMPK and physiochemical
properties suggested their potential in vivo efficacy.
1,3,5-triazaspiro[5.5]undeca-2,4-diene
© 2017 Elsevier Masson SAS. All rights reserved.
1. Introduction
reported that 1.8 million deaths were TB related [1]. The statistics
from recent reports have shown that the prevalence of TB has
Tuberculosis (TB) is one of the leading causes of death world-
wide and this has become a global public health issue. In 2015,
WHO estimated 10.4 million cases of new TB infections and
escalated to an epidemic proportion. The emergence of multi and
extensively drug-resistant TB (MDR-TB and XDR-TB) strains of
Mycobacterium tuberculosis has limited the use of existing thera-
pies, thus exacerbating the epidemic further. In 2015, the mortality
cases due to MDR-TB were estimated at 0.25 million while among
the people who were diagnosed with MDR-TB, an estimated 9.5% of
them were suffering from XDR-TB [2]. In addition, within the HIV
positive population of patients, opportunistic infection due to TB is
contributing to increasing morbidity and mortality [3]. The epide-
miology of the disease has raised deep concerns among clinicians
and scientists therefore active efforts are underway to address this
public health related issue.
The current treatment regimen for drug-sensitive TB consists of
a combination of four first-line drugs namely, isoniazid, pyr-
azinamide, rifampicin and ethambutol [4]. The patients are
required to adhere to the combination therapy for at least 6
months. In case of drug-resistant strains, the treatment period can
Abbreviations: DHFR, dihydrofolate reductase; DMPK, drug metabolism and
pharmacokinetics; FRLM, female rat liver microsomes; HBA, hydrogen bond ac-
ceptors; HBD, hydrogen bond donors; MDR, multi-drug resistant; MRLM, male rat
liver microsomes; Mtb, M. tuberculosis; MTX, methotrexate; NMR, nuclear magnetic
resonance; SI, selectivity index; TB, tuberculosis; TPSA, topological polar surface
area; XDR, extensively-drug resistant.
* Corresponding author. Department of Microbiology and Immunology, Yong Loo
Lin School of Medicine, National University of Singapore, 5 Science Drive 2, 117545,
Singapore.
** Corresponding author. Department of Pharmacy, Faculty of Science, National
University of Singapore, 18 Science Drive 4, 117543, Singapore.
E-mail addresses: td367@njms.rutgers.edu (T. Dick), phacwk@nus.edu.sg
(W.-K. Chui).
1
These authors contribute equally to this work.
https://doi.org/10.1016/j.ejmech.2017.12.017
0223-5234/© 2017 Elsevier Masson SAS. All rights reserved.

X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
263
be extended to 18e24 months with the inclusion of some second-
line drugs such as fluoroquinolones, streptomycin, linezolid, para-
aminosalicylic acid and others [5]. Apart from the high pill burden
and the lengthy treatment, the therapies for TB can also be costly.
Many second-line drugs are known to be toxic and less efficient,
posing a significant barrier to patient compliance [6]. Hence, more
efforts are needed to broaden the availability of treatment options
by designing and synthesizing new anti-TB agents that exert
different mechanism of action to circumvent the current drug
resistance issue.
the distal end of the side chain, there are amino acid residues that
may offer hydrophobic and hydrogen-bonding interactions [18]. It
was therefore proposed to have phenyl alkyl group, benzoyl methyl
group, and coumarin alkyl group inserted into the distal end in an
attempt to interact with this part of the enzyme (Fig. 2).
2. Results and discussion
2.1. Chemistry
Dihydrofolate reductase (DHFR), a key enzyme in the folate
pathway, is responsible for reducing dihydrofolic acid to tetrahy-
drofolic acid. Tetrahydrofolate is involved in one carbon transfer
reactions that lead to the biosynthesis of nucleic acids and amino
acids. Thus, the inhibition of DHFR is able to arrest cell growth by
depleting the cellular pool of DNA, RNA and protein synthesis
precursors [7,8]. DHFR has been recognized as a validated drug
target for antibacterial agents [9]. However, few potent and selec-
tive DHFR inhibitors have so far been reported that exhibits anti-
mycobacterial activity despite on-going research work in the
scientific community [10]. For example, by using a well-known
DHFR inhibitor, methotrexate (MTX) as lead compound, two es-
ters of methotrexate (JSF-1187 and JSF-1183) have been reported to
demonstrate potent pathogenic DHFR inhibition activity (Fig. 1) by
Nixon et al. [11]. Nevertheless, these two compounds have not
shown significant selectivity for M. tuberculosis DHFR (Mtb DHFR)
due to their close similarity with MTX. Besides the pteridine scaf-
fold, three hits (PDP-1, PQD-1 and Dpt-1, Fig. 1) representing other
chemical scaffolds were identified by Kumar et al. through a
phenotypic screening against live H37Rv strain of Mtb, and DHFR
inhibitory activity evaluation [12]. However, only Dpt-1 exhibited
selectivity against pathogenic DHFR. Recently, Nilesh and col-
leagues have developed two 2,4-diamino-1,3,5-triazine derivatives
(Fig. 1) that showed selective Mtb DHFR inhibition, and these two
compounds were found to possess promising antitubercular ac-
tivity and lower cytotoxicity in human [13]. Collectively, these re-
ports suggested that there is potential pharmacotherapeutic
significance in using DHFR inhibitors for the treatment of tuber-
culosis. The core structural feature for the inhibition of DHFR is
known to be a diamino-substituted N-containing single or fused
polycyclic heterocycle such as pyrimidine, quinazoline or pteridine.
Many of these scaffolds have been extensively studied and
demonstrated to possess antibacterial activity against various mi-
croorganisms [7,14,15]. Nonetheless, few studies were carried out
on diamino-substituted N-containing spiro scaffold [16].
Over the years, in the main author's laboratory, the spiro het-
erocycle 1,3,5-triazaspiro[5.5]undeca-2,4-diene has demonstrated
behaviours that can be exploited for selective DHFR inhibitory ac-
tivity against pathogenic bacteria. For instance, 7s (Fig. 2) was
demonstrated to inhibit the growth of Mycobacterium smegmatis
[16], while P-I-23 (Fig. 2) was found to have no inhibitory action on
recombinant human DHFR, and was also found to be non-toxic to
A549 cell line human lung carcinoma [17]. Taken together, it was
hypothesized that with appropriate substitution on the saturated
nitrogen of this spiro heterocycle, selective mycobacterial DHFR
inhibitor with low human cytotoxicity could be developed. The
saturated nitrogen was selected as a point for side chain extension
because it was believed that the side chain would be accommo-
dated in a channel at the active site [13,18,19]. This paper reports a
study that entailed the synthesis of new compounds having
extension made to the saturated nitrogen (N1) of the 1,3,5-
triazaspiro[5.5]undeca-2,4-diene by inserting either a methylene
or a phenyl attachment. The side chain was elongated using click
chemistry. This conjunctive approach was applied to search for new
compounds with DHFR inhibitory activity. It is also reported that in
All the compounds were designed and synthesized using stra-
tegies that extended the side chain at position N1 of the spiro
heterocycle (Fig. 2). Based on compound 7s, one strategy was to link
N1 of 1,3,5-triazaspiro[5.5]undeca-2,4-diene and the 1,2,3-triazole
with a methylene group to evaluate if a rotatable alkyl linker
would be favour DHFR inhibition that might lead to anti-
mycobacterial activity. This strategy resulted in the design of
compounds 7a-7c, 11a-11c and 14 (Scheme 1). The second strategy
was derived from compound P-I-23 by attaching 4-methyleneoxy
phenyl ring at N1 of 1,3,5-triazaspiro[5.5]undeca-2,4-diene and
linking it to the triazole ring. This strategy afforded compounds
19a-19c, 20a-20c and 21 (Scheme 2A, 2B and 2C). In these two
strategies, the three-component one-pot synthesis described by
Modest's was used to generate the alkyne intermediates 4 and 18 as
hydrochloric salt [20]. The intermediates were reacted with
selected azido derivatives to give the target compounds.
2.1.1. Synthesis of target compounds 7a-7c, 11a-11c and 14
As shown in Scheme 1, the modified three-component one pot
synthetic strategy was adopted to generate intermediate 4 by
refluxing propargylamine hydrochloride 1, cyanoguanidine 2, and
cyclohexanone 3 in the presence of catalytic amount (0.2 equiva-
lent) of concentrated hydrochloric acid in ethanol [20]. In the
following step, the transformation of benzylbromides 5a-5c into
the corresponding azides 6a-6c was achieved via a nucleophilic
substitution of the bromide with three equivalents of sodium azide
under reflux condition. The azide intermediate was used in the next
step without further purification. The freshly prepared azides 6a-6c
were coupled to intermediate 4 via a 1,3-dipolar addition reaction
in the presence of catalytic amount (0.1 equivalent) of copper sul-
fate. The reaction of intermediate 4 and azide 6a-6c was monitored
by TLC (MeOH: DCM ¼ 1: 9) using UV as detector. The target
products 7a-7c were formed in reasonably good yield (61e68%)
when the reflux was conducted in a solvent system of THF and
water at volume ratio of 1:1. For the synthesis of compounds 9a-9c,
the ethereal spacer was furnished via Williamson synthesis by
refluxing the corresponding dibromoalkanes and compound 8 in
acetonitrile under basic condition. Potassium carbonate was used
to deprotonate the phenolic group on umbelliferon 8. Azide de-
rivatives 10a-10c were formed when compounds 9a-9c were
treated with sodium azide under reflux. The coupling reaction of
the azide derivatives 10a-10c and intermediate 4 produced the
desired target products 11a-11c. The synthesis of product 14 was
similar to the preparation of products 7 and 11. The acetophenone
12 was treated with sodium azide to afford the corresponding azide
13, which was followed by coupling with intermediate 4 to furnish
the desired target product 14. The yield for the final-step reaction
was further improved by dissolving the azide derivative in tetra-
hydrofuran and followed by dropwise addition of this azide solu-
tion into the reaction mixture with constant stirring.
2.1.2. Synthesis of target compounds 19a-19c, 20a-20c, and 21
The synthetic schemes for compounds 19a-19c, 20a-20c and 21
are illustrated in Scheme 2A, 2B and 2C respectively. Prop-2-
ynyloxy aniline 17 was prepared by a two-step reaction. Firstly, 4-

264
X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
Fig. 1. The structures of reported Mtb DHFR inhibitors that exhibit antimycobacterial activity [10ꢀ12].
nitrophenol 15 was refluxed with propargyl bromide under basic
condition to furnish 4-nitrophenyl-2-propynyl ether 16 via Wil-
liamson synthesis. The crude product was separated by filtering out
the reaction mixture, and purified by column chromatography us-
ing ethyl acetate/hexane (1:5 to 1:1, v/v) as the mobile phase. Then
the purified 4-nitrophenyl-2-propynyl ether 16 was reduced to 4-
(prop-2-ynyloxy)aniline 17 by an excess of iron powder under
acidic environment in excellent yields and purity. The crude
product 17 was used directly in the next step without further pu-
rification. The synthesis of intermediate 18 was conducted through
an “one-pot three components” synthesis employing 4-(prop-2-
ynyloxy)aniline 17, cyanoguanidine and cyclohexanone. This reac-
tion consisted of a trimerization between nitrile and cyanide
groups to provide biguanide in situ. The formation of the spiro ring
was achieved through the condensation reaction between the
biguanide intermediate and a cyclic ketone. Intermediate 18 was
purified by recrystallization from 95% ethanol in moderate yield. By
applying click chemistry, intermediate 18a was coupled with the
corresponding azido derivatives 6a-6c,10a-10c and 13 separately to
afford the desired target products 19a-19c, 20a-20c and 21,
respectively.
All the final products were characterized and confirmed by NMR
and MS. The purity of the target compounds was determined by
HPLC analysis, and all the products showed purity of 95% or above.
2.2. Phenotype-oriented screening and cytotoxicity study
Recent works from literature have highlighted the higher suc-
cess of phenotypic whole cell screening for identifying first-in-class
antibacterial agents when compared to biochemical target-based
approaches [21]. Therefore, a phenotype-oriented screening strat-
egy was adopted to evaluate antimycobacterial activity of the
synthesized compounds, followed by cytotoxicity detemination.
2.2.1. Antimycobacterial activity against M. bovis BCG
For
a
preliminary antimycobacterial study, single point
M. M. bovis
screening was conducted against M. bovis BCG at 25
m
BCG is widely used as an in vitro screening model for antitubercular
drug discovery as the strain represents a non-pathogenic close
relative to M. tuberculosis [22]. As shown in Table 1, compounds
19a-19c, 20a-20c, and 21 exhibited stronger inhibition in compar-
ison to compounds 7a-7c, 11a-11c, and 14. It was found that

X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
265
2.3. Evaluation of in vitro Mtb DHFR and human DHFR inhibitory
activities of compounds 20b, 20c, 19c and 11c
The two most potent compounds 20b and 20c were tested for
their ability to inhibit recombinant Mtb DHFR as well as the re-
combinant human (rh) DHFR in an attempt to investigate their
selectivity for inter-species enzymatic inhibition. The selectivity
index (SI) was calculated as a ratio of the IC50 values of rh DHFR to
the IC50 values of Mtb DHFR. Methotrexate, a very potent non-
selective DHFR inhibitor, was used as control. As shown in
Table 3, the IC50 values of the two compounds 20b and 20c indi-
cated that they were less potent than methotrexate in inhibiting rh
DHFR. However, they were 44e93 times more selective for rMtb
DHFR than methotrexate. Besides, it was found that the selectivity
increased with the elongation of the ethereal linker from 3 to 4
methylene groups.
In order to explore the role of coumarin in DHFR binding,
compound 19c was selected for the enzyme inhibition assays.
Through the comparison with compound 20b, it was found that the
incorporation of coumarin at the distal side chain increased po-
tency as well as selectivity. In addition, the attachment of 4-
methyleneoxy phenyl linker at N1 of 1,3,5-triazaspiro[5.5]undeca-
2,4-diene was found to be essential for recombinant Mtb DHFR
inhibition as well, by testing compound 11c on Mtb DHFR.
However, it was found that the enzymatic inhibitory activities
were not as potent as the whole cell antimycobacterial activities.
This could be for instance due to the influence of the inclusion of
His-tag on the recombinant Mtb DHFR or to accumulation of the
test compounds in the bacterial cell. To exclude the possibility that
the compounds may exert their whole cell antibacterial growth
inhibition activity off target, on-target verification was conducted.
Fig. 2. Design of compounds by side chain extension at the saturated nitrogen of 1,3,5-
triazaspiro[5.5]undeca-2,4-diene and 6,8,10-triazaspiro[4.5]deca-7,9-diene via click
chemistry.
compounds having the 4-methyleneoxy phenyl linker at N1 of
1,3,5-triazaspiro[5.5]undeca-2,4-diene exhibited relatively higher
antimycobacterial activity at 25 mM. Compounds that demonstrated
at least 60% inhibition were selected for MIC50 value determination.
The five selected compounds namely 19a, 19c, 20a, 20b, and 20c
were tested against M. bovis BCG by the broth microdilution
method as described previously [23e25]. As shown in Table 1,
compounds 20a, 20b, and 20c exhibited more potent anti-
2.4. On target activity verification through M. bovis BCG DHFR over-
expression and observation of MIC shift
In order to determine whether whole cell antimycobacterial
activity was mediated indeed through DHFR inhibition, growth
inhibition dose response experiments were carried out with a BCG
strain that overexpressed DHFR. To achieve the overexpression, the
DHFR gene dfrA was expressed under the control of the strong
constitutive hsp60 promoter [26]. 10-fold dfrA overexpression was
confirmed at the mRNA level by qRT-PCR. p-Aminosalicylic acid
(PAS), a second-line anti-TB drug, was recently reported to be a
prodrug, which is metabolized by folate pathway and inhibits
DHFR. Hence, PAS was used as a positive control in this study
(Fig. 3B), and the MIC90 value was shifted by 16 fold to a higher
value. In terms of compounds 20b and 20c, the growth inhibition
curves illustrated in Fig. 3A, show an MIC90 shift to a higher value in
the DHFR-overexpressed (DHFR-oe) strain. Both compounds 20b
and 20c exhibited a 25e30 fold shift, respectively, in the MIC90
(DHFR-oe BCG MIC90/wt-BCG MIC90) to higher values in the DHFR-
oe strain compared with wild type M. bovis BCG. Taken together,
these results suggested that these compounds exert their whole
cell antimycobacterial activity via the inhibition of DHFR.
mycobacterial activity (MIC50: 0.01e0.2 mM) against M. bovis BCG,
as compared to compounds 19a and 19c. The potency of com-
pounds 20a-20c may be attributed to the incorporation of the
coumarin fragment, which can form potential hydrogen bonds with
amino acid residues at the distal binding site. In particular, com-
pounds 20b and 20c demonstrated both MIC50 and MIC90 values at
sub-micro molar ranges. Besides, it can be seen that the anti-
mycobacterial activity increased with the elongation of ethereal
linker from 2 to 4 methylene groups.
2.2.2. Antimycobacterial activity against M. tuberculosis and
cytotoxicity evaluation
To confirm the inhibitory activity against pathogenic
M. tuberculosis, the most promising compounds 20b and 20c were
tested against M. tuberculosis H37Rv. Both compounds showed
2.5. In silico molecular modeling of compound 20c in Mtb and
antitubercular activity at MIC50 value of 0.01
mM and MIC90 value of
human DHFRs
0.025 M (Table 2). The excellent antitubercular activity of the two
m
compounds prompted the need to evaluate their cytotoxic and
hemolysis effects, by using the HepG2 cell line and red blood cells
for CC50 and LC50 determination, respectively. Both compounds
demonstrated low liver cell cytotoxicity and low hemolytic activity
up to 100 mM. Consequently, both compounds 20b and 20c, with
over 4000-fold selectivity, were subjected to DHFR inhibition
determination.
To better understand the binding interactions at the active site
of both Mtb DHFR and human DHFR, in silico molecular modeling
was conducted on compound 20c due to its on-target inhibitory
activity and better selectivity for rMtb DHFR. Crystal strcutures of
Mtb DHFR (PDB ID: 1DF7) and human DHFR (PDB ID: 2w3m) were
obtained from the protein data bank (PDB) and modified for the
docking experiments. Subsequently, the partial atomic charge

266
X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
Scheme 1. Synthesis of triazaspiroalkenes 7a-7c (A), 11a-11c (B) and 14 (C). Reagents and conditions: (a) EtOH, Conc. HCl (0.2 equiv), reflux; (b) NaN3 (3 equiv), DMF, reflux; (c)
CuSO$45H2O (0.1 equiv), Sodium ascorbate (0.3 equiv), THF/H2O (1:1, v/v), rt; (d) dibromoalkane, K2CO3 (6.0 equiv), acetonitrile, reflux.
distribution calculation and energy minimization of the ligands
were performed before conducting the docking experiments. The
docking protocol was optimized and validated by reproducing the
observed co-crystallized ligands in Mtb DHFR (RMSD value of
1.06 Å) and human DHFR (RMSD value of 0.85 Å) respectively. The
scoring function for docking is expressed as -log Kd which repre-
sents binding affinity in terms of hydrophobicity, polarity, repulsive
forces, entropy, solvation and crash score. A higher docking score
indicates tighter binding between the ligand and the receptor. The
pose of compound 20c with the highest docking score was selected,
and its docking scores for both enzymes are summarized in Table 4.
It was found that compound 20c exhibited stronger binding affinity
to Mtb DHFR as compared to its binding in human DHFR. This
observation was consistent with the selectivity observed in the
enzyme inhibition study.
As shown in Fig. 4A, compound 20c binds in the active site of
DHFR through the formation of hydrogen bonds with Ile94 (N-
H/O: 1.84 Å), Ile5 (N-H/O: 1.82 Å), Tyr100 (N-H/O: 2.86 Å) and
Asp 27 (N-H/O: 1.90 Å and 2.56 Å). Also, hydrogen bonds between
1,2,3-triazole and Leu24 (N-H/N: 2.19 Å), as well as between the
oxygen from 1-(4-methoxy)phenyl ring and Trp22 (N-H/O:
2.52 Å) are also observed. Furthermore, another three favourable

X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
267
Scheme 2. (A) Synthesis of triazaspiroalkenes 19a-19c (A), 20a-20c (B), and 21 (C). Reagents and conditions: (a) propargyl bromide, K2CO3 (1.05 equiv.), acetonitrile, reflux; (b) Fe
(6.5 equiv.), NH4Cl (0.9 equiv), ethanol/water (5:1), reflux; (c) cyclohexanone, dicyandiamide, Conc. HCl (1.2 equiv.), EtOH, reflux; (d) CuSO$45H2O (0.1 equiv), Sodium ascorbate (0.3
equiv), THF/H2O (1:1, v/v), rt.
hydrogen bonds were established between the distal ethereal ox-
ygen and Gln28 (N-H/O: 2.85 Å), as well as between the carbonyl
oxygen on coumarin and Arg32 (N-H/O: 2.19 Å), and Arg60 (N-
H/O: 2.05 Å), respectively. In addition, hydrophobic amino resi-
dues Leu57, Val54, Pro51, Phe31, Leu24, and Ile20 make close con-
tacts with the side chain (Fig. 4B). As a consequence, the insertion of
4-methyleneoxy phenyl linker between the DHFR pharmacophoric
scaffold and 1,2,3-triazole has allowed conducive orientation of the
DHFR pharmacophoric scaffold and 1,2,3-triazole side chain at the
active site.
As compound 20c demonstrated better inhibition activity for
rMtb DHFR over rh DHFR, the molecular modeling of compound 20c
in human DHFR was also conducted. As shown in Fig. 4C, the 2-
amino group forms hydrogen bonds with Tyr121 (N-H/O:
2.62 Å) and Ile7 (N-H/O: 2.22 Å), while the 4-amino group forms
hydrogen bond with Glu30 (N-H/O: 2.09 Å) at the DHFR anchoring
point. Besides, a salt bridge (N-H/O: 2.06 Å) is established be-
tween protonated N5 of 1,3,5-triazine and carboxylate side chain of
Glu30. Two additional hydrogen bonds are formed between the
distal ethereal oxygen on coumarin and Arg70 (N-H/O: 2.10 Å),
and between the lactone oxygen and Lys66 (N-H/O: 2.06 Å). Hy-
drophobic interactions (Fig. 4D) are involved with Trp24, Leu22,
Phe31, Phe34, Val115, Leu67 and Ile60.
Through the comparison of the corresponding interactions at
the different binding pockets, more favourable hydrogen bonding
interactions were observed in Mtb DHFR active site by the incor-
poration of 1,2,3-triazole and the coumarin scaffold. In addition, as
shown in Fig. 4D, though Phe31 and Leu22 could offer hydrophobic
interaction with the cyclohexyl spiro ring, they might also restrict
the entry of the inhibitor into the human DHFR anchoring point. In
contrast, the entrance to Mtb DHFR anchoring pocket appeared
wider and was able to accommodate the larger spiro ring. This
observation was also supported by Elaine F and colleagues' finding
that the distance between the side chains of Leu22 and Phe31 in
human DHFR is at least 2-fold shorter than the distance between
corresponding residues in Mtb DHFR [19]. It was also reported by
Elaine F et al. that a water molecule interacted with Trp22, Asp27
and Gln28 that collectively formed a pocket in Mtb DHFR. In
contrast, the corresponding binding pocket created by Leu20,
Arg23, and Gln28 in human DHFR restricted the accessibility of
methothrexate to the solvent. As shown in Fig. 4A, the 4-
methyleneoxy phenyl linker of compound 20c interacted with
Trp22 via hydrogen bonding, while the coumarin scaffold formed
hydrogen bonding with Gln28. Recent study highlighted the exis-
tence of glycerol binding site in the Mtb DHFR; however, the glyc-
erol binding site is essentially absent in human DHFR [27]. Besides,

268
X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
Table 1
Antimycobacterial activity of synthesized compounds against M. bovis BCG.
Cpd
Structure
Inhibition (%) at 25
M. bovis BCG
4.5
m
M
MIC50
(
m
M)
MIC90 (mM)
M. bovis BCG
M. bovis BCG
7a
N.D
N.D
7b
4.7
N.D
N.D
7c
5.4
5.8
N.D
N.D
N.D
N.D
11a
11b
7.8
N.D
N.D
11c
14
6.7
3.4
N.D
N.D
N.D
N.D
19a
19b
19c
20a
61.7
25.6
67.8
84.8
19.2
N.D
20.7
0.2
>100
N.D
>100
1.56

X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
269
Table 1 (continued )
Cpd
Structure
Inhibition (%) at 25
M. bovis BCG
88.4
m
M
MIC50
(
m
M)
MIC90 (mM)
M. bovis BCG
M. bovis BCG
20b
20c
21
0.02
0.2
92.1
12.8
0.01e0.02
0.1
N.D
N.D
Data reported are mean values taken from 3 independent experiments.
N.D: Not determined.
Table 2
Antitubercular activity and cytotoxicity of compound 20b and 20c.
Cpd
MIC50 H37Rv (
mM)
MIC90 H37Rv (
m
M)
CC50 HepG2 (
mM)
LC50 RBC (
m
M)
SI (CC50/MIC90)
20b
20c
0.01
0.01
0.025
0.025
>100
>100
>100
>100
>4000
>4000
Data reported are mean values taken from 3 independent experiments.
Table 3
Inhibitory activities of selected compounds on recombinant Mtb DHFR and human DHFR.
Cpd
IC50 Mtb DHFR (
mM)
IC50 rh DHFR
M)
SI (rh DHFR IC50/Mtb DHFR IC50)
(
m
20b
20c
19c
11c
1.34 0.12
0.82 0.047
5.82 0.068
>100
11.99 1.22
15.32 0.45
3.67 0.15
N.D
~8.95
~18.68
~0.63
N.A
Methotrexate
0.035 0.0024
0.0071 0.0003
~0.2
N.D: Not determined.
N.A: Not available.
Fig. 3. (A) Dose dependent growth inhibition curves of the two most active triazaspiroalkene derivatives identified in the screen. Compounds were assayed against wild type
M. bovis BCG (shown as clear symbols) and recombinant M. bovis BCG engineered to overexpress M. bovis BCG DHFR (shown as solid symbols). (B) Para-aminosalicylic acid (PAS)
MIC shift. DHFR-oe BCG, DHFR overexpressing M. bovis BCG; wt-BCG, Wild type M. bovis BCG. OD values are the means for three replicates and error bars indicate standard error of
the mean (SEM).

270
X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
Table 4
demonstrated reasonable in vitro metabolic stability in rat liver
microsomes. However, compound 20b was relatively more stable
with longer half-life and lower clearance in rat microsomes, as
compared to compound 20c.
Docking score of compound 20c in Mtb DHFR and human DHFR.
Compound
Docking score (-log Kd)
Mtb DHFR
Human DHFR
13.56
In terms of drug-likeness, a majority of marketed antibacterial
drugs possess different physicochemical properties in dis-
accordance with the Lipinski Rule [28]. The marketed gram-
positive drugs possess an average TPSA of 243 Å2 (range:
29e764 Å2), MW of 813 (range: 290e1880), HBD of 7.1 (range:
1e29) and HBA of 16.3 (range: 2e46) [29]. The physicochemical
properties of compounds 20b and 20c fall within these ranges.
Besides, in terms of a few anti-TB agents such as rifampicin,
streptomycin, amikacin, vancomycin, bedaquiline and delamanid,
they also possess higher molecular weight and TPSA.
20c
15.37
glycerol binding was proven to be weak and could be expelled from
the binding site to expose Arg32 and Arg60 for interaction with
inhibitors containing appropriate side chain. As can be seen in
Fig. 4A, the coumarin ring of compound 20c formed hydrogen
bonds with Arg32 and Arg60. As a consequence, these observations
and assumptions could account for the observed stronger binding
affinity of compound 20c on Mtb DHFR, therefore leading to the
DHFR selectivity.
3. Conclusion
2.6. Preliminary in vitro DMPK study on compounds 20b and 20c
In this work, using the side chain extension approach, the 2,4-
diamino-1-phenyl-1,3,5-triazaspiro[5.5]undeca-2,4-diene scaffold
has been developed into novel DHFR inhibitors with attractive
selectivity and potent antimycobacterial property. In terms of the
potency, the 4-methyleneoxy phenyl linker when directly attached
to N1 of the 1,3,5-triazaspiro[5.5]undeca-2,4-diene was found to be
essential for both antimycobacterial activity and Mtb DHFR inhi-
bition. The molecular docking study further corroborated the DHFR
selectivity of compound 20c for the enzyme of the pathogen and
Finally, compounds 20b and 20c were also subjected to in vitro
metabolic and pharmacokinetic profiling. Calculated parameters,
such as molecular weight and the numbers of hydrogen bond do-
nors and acceptors, are summarized in Table 5. The higher values of
topological polar surface area (TPSA) of compounds 20b and 20c
supported the good aqueous solubility of these compounds. With
regard to the DMPK study, both compounds 20b and 20c
Fig. 4. Molecular docking of compound 20c. (A) Optimized pose of compound 20c at the active site of Mtb DHFR. (B) Visualization of compound 20c interacting with the hy-
drophobic residues at the active site of Mtb DHFR. (C) Optimized pose of compound 20c at the active site of human DHFR. (B) Visualization of compound 20c interacting with the
hydrophobic residues at the active site of human DHFR. (Brown: binding site with higher hydrophobicity; Blue: binding site with lower hydrophobicity). (For interpretation of the
references to colour in this figure legend, the reader is referred to the Web version of this article.)

X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
271
Table 5
In vitro characterization of compounds 20b and 20c.
4.1.1. 5-(prop-2-yn-1-yl)-1,3,5-triazaspiro[5.5]undeca-1,3-diene-
2,4-diamine hydrochloride (4)
A
mixture of prop-2-yn-1-amine hydrochloride (3.0 g,
Property
20b
20c
32.77 mmol), cyanoguanidine (3.03 g, 36.04 mmol), cyclohexanone
(3.54 g, 36.07 mmol) and concentrated HCl (0.2 mL) was refluxed
with stirring under N2 gas until a negative biguanide test was ob-
tained. Precipitation of the product occurred either during the re-
action process, or when the reaction mixture was cool down to
room temperature. The solid was collected by filtration, washed
with ethanol and recrystallized from 80% ethanol in H2O. Yield:
Mol wt
593.08
2
6
9
2.73
154.33
>100
83.2 2.9
270.7 16.4
27.8 1.0
8.6 0.5
607.1
2
6
10
No. of HBDa
No. of HBAb
No. of rotatable bonds
cLogDc at pH 7.4
TPSA (Å2)d
3.25
154.33
95.83 1.41
59.0 2.3
113.5 9.2
39.2 1.6
20.4 1.7
Aq solubility (mM) at pH 7.4
RLM t1/2 (min, male)e
RLM t1/2 (min, female)e
53.7%. 1H NMR (400 MHz, DMSO-d6):
d 8.80 (1H, br, NH), 7.62 (2H,
CLint, in vitro
CLint, in vitro
(
(
m
m
L/min/mg, male)
L/min/mg, female)
br, NH), 7.43 (2H, br, NH), 4.32 (2H, s, CH2), 3.42 (1H, s, alkynyl-H),
1.94e1.82 (2H, m, cyclohexyl), 1.80e1.54 (7H, m, cyclohexyl),
1.20e1.03 (1H, m, cyclohexyl) ppm. 13C NMR (100 MHz, DMSO-d6):
157.77, 157.75, 79.65, 76.15, 71.95, 33.85, 32.77, 24.26, 21.26 ppm.
a
HBD: hydrogen bond donors.
HBA: hydrogen bond acceptors.
cLogD: calculated by MarvinSketch 6.1.0.
TPSA: topological polar surface area (http://www.molinspiration.com).
RLM: rat liver microsomes.
b
c
d
e
4.1.2. 1-nitro-4-(prop-2-yn-1-yloxy)benzene (16)
To a solution of 4-nitrophenol (1.00 g, 7.19 mmol) and K2CO3
(0.99 g, 7.21 mmol) in acetonitrile (50 mL) was added 3-
bromoprop-1-yne (0.90 g, 7.5 mmol, 80% wt in Toluene). The re-
action mixture was heated under reflux overnight. After cooling
down, the reaction mixture was filtered and acetonitrile in the
filtrate was removed in vacuo. The residue was extracted with ethyl
acetate and washed with saturated NH4Cl solution and brine. The
organic layers were combined and dried over Na2SO4 and filtered.
Removal of the solvent in vacuo afforded the crude product which
was purified by silica gel chromatography (EA: Hexane ¼ 1:5) to
afford a white solid (1.20 g) in yield 94.6%. 1H NMR (400 MHz,
unveiled the role of 1,2,3-triazole and coumarin ring in offering
favourable interactions with amino acid residues at Mtb DHFR
binding site. As mentioned above, the disconnect between enzy-
matic and cellular level activities might have been the result of the
influence of the inclusion of His-tag on recombinant Mtb DHFR. In
addition, further study on the role of coumarin moiety is underway
to elucidate a possible secondary synergistic mechanism that may
contribute to the observed cellular activity. Based on the results
obtained from the biological activities, docking studies and the
DMPK properties determination, compounds 20b and 20c are two
selective Mtb DHFR inhibitors which have shown promise for
further development into potential antitubercular agents. These
two 2,4-diamino-1-phenyl-1,3,5-triazaspiro[5.5]undeca-2,4-dienes
are hitherto believed to be first in demonstrating such anitmyoc-
bacterial property through DHFR inhibition.
DMSO-d6):
d 8.30e8.19 (2H, m, ArH), 7.28e7.16 (2H, m, ArH), 4.99
(2H, d, J ¼ 2.4 Hz, OCH2), 3.68 (1H, t, J ¼ 2.4 Hz, alkyne-CH) ppm. 13C
NMR (100 MHz, DMSO-d6): 162.83, 141.85, 126.26, 115.94, 79.64,
78.67, 56.81 ppm.
4.1.3. 4-(prop-2-yn-1-yloxy)aniline (17)
A mixture of 16 (1.05 g, 5.65 mmol), iron powder (2.16 g,
38.4 mmol), NH4Cl (0.29 g, 5.08 mmol) in ethanol/water (5: 1) was
stirred at reflux for 1 h. After cooling down, the reaction mixture
was filtered through celite and concentrated. The residue was
extracted with ethyl acetate, washed with saturated Na2CO3 solu-
tion and brine, and dried over Na2SO4 and filtered. Removal of the
solvent in vacuo afforded the crude product which was purified by
silica gel chromatography (EA: Hexane ¼ 1:3) to afford yellow oil
4. Experimental section
4.1. Chemistry
All reagents were purchased from Sigma-Aldrich (Hercules, CA,
USA), Tokyo Chemical Industry (Tokyo, Japan), Alfa Aesar (Ward
Hill, MA, USA), or Merck (Darmstadt, Germany) and were used
directly without any purification. Thin layer chromatography (TLC)
was performed using silica gel-coated aluminum plates with fluo-
rescent indicator and visualized with UV light at 254 nm. Column
chromatography was carried out on silica gel (pore size 60 Å,
230e400 mesh). 1H NMR and 13C NMR spectra were recorded on
Bruker Ultrashield™ 400 Plus NMR spectrometer. Chemical shift
(0.81 g) in yield 92.9%. 1H NMR (400 MHz, DMSO-d6):
d 6.74e6.65
(2H, m, ArH), 6.55e6.46 (2H, m, ArH), 4.66 (2H, s, NH2), 4.59 (2H, d,
J ¼ 2.4 Hz, OCH2), 3.47 (1H, t, J ¼ 2.4 Hz, alkyne-CH) ppm. 13C NMR
(100 MHz, DMSO-d6): 148.99, 143.63, 116.52, 115.18, 80.48, 78.04,
56.63 ppm.
4.1.4. 5-(4-(prop-2-yn-1-yloxy)phenyl)-1,3,5-triazaspiro[5.5]
undeca-1,3-diene-2,4-diamine hydrochloride (18)
values (d) are expressed as parts per million (ppm) relative to tet-
ramethylsilane as the internal standard. Coupling constants (J) are
expressed in Hz, and splitting patterns are described as follows:
s ¼ singlet, d ¼ doublet, t ¼ triplet, m ¼ multiplet, br ¼ broad
singlet, dd ¼ doublet of doublets, dt ¼ doublet of triplets,
td ¼ triplet of doublets, tt ¼ triplet of triplets, ddd ¼ doublet of
doublet of doublets. ESI-MS was recorded on an AB SCIEX API 2000
Q Trap mass spectrometer. High performance liquid chromatog-
raphy (HPLC) was performed for purity checking using Agilent 1100
A mixture of 17 (3.11 g, 21.14 mmol), cyanoguanidine (1.96 g,
23 mmol), ketone (24.0 mmol) and concentrated HCl (2 mL) was
refluxed with stirring under N2 gas until a negative biguanide test
was obtained (18 h for cyclohexanone and 36 h for cyclo-
pentanone). Precipitation of the product occurred either during the
reaction process, or when the reaction mixture was cool down to
room temperature. The solid was collected by filtration, washed
with ethanol and recrystallized from 80% ethanol in H2O. Yield
Series HPLC on a Lichrosorb 10 RP-18 250 ꢁ 4.0 mm 10
mm column.
62.8%. 1H NMR (400 MHz, DMSO-d6):
d 9.03 (1H, s, NH), 7.57 (3H, br,
HPLC separations were carried out using gradient elution. The
initial mobile phase consisting of 80% of 0.1% w/v trifluoroacetic
acid in water and 20% acetonitrile was gradually changed to 100%
acetonitrile over 20 min and maintained at 100% acetonitrile for
additional 2 min. All the final compounds were found to be pure up
to 95% or higher.
NH), 7.43e7.25 (2H, m, ArH), 7.22e7.02 (2H, m, ArH), 6.26 (1H, br,
NH), 4.85 (2H, d, J ¼ 2.4 Hz, OCH2), 3.66 (1H, t, J ¼ 2.4 Hz, alkyne-
CH), 1.95e1.80 (2H, m, cyclohexyl), 1.78e1.44 (5H, m, cyclohexyl),
1.32e1.23 (2H, m, cyclohexyl), 1.00e0.84 (1H, m, cyclohexyl) ppm.
13C NMR (100 MHz, DMSO-d6): 158.34, 158.33, 158.19, 131.80,
128.14, 116.46, 79.44, 79.11, 71.82, 56.48, 35.16, 24.42, 21.20 ppm.

272
X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
4.1.5. General procedure for the synthesis of compound 9a-9c
A mixture of coumarin derivative (1 eq), dibromoalkane (3 eq)
and K2CO3 (6 eq) in 50 mL acetonitrile was refluxed overnight. After
the completion of reaction, excess K2CO3 was removed by filtration
of celite and the filtrate was concentrated, washed with DCM and
dried with Na2SO4. The residue was purified by column chroma-
tography (EA: DCM: HEX ¼ 1:25:25) to afford the white solid.
ESI-MS m/z 353.7 (Mþ1)þ.
4.1.7.2. 5-((1-Phenethyl-1H-1,2,3-triazol-4-yl)methyl)-1,3,5-
triazaspiro[5.5]undeca-1,3-diene-2,4-diamine hydrochloride (7b).
Yield: 66.2%. Purity: 99.1%. 1H-NMR (400 MHz, DMSO-d6):
d 8.62
(1H, s, NH), 7.95 (1H, s, triazolyl-H), 7.54 (2H, br, NH), 7.34 (2H, br,
NH), 7.30e7.09 (5H, m, ArH), 4.70 (2H, s, CH2-triazine), 4.62 (2H, t,
J ¼ 7.1 Hz, CH2-triazole), 3.14 (2H, t, J ¼ 7.1 Hz, CH2Ar), 1.83e1.47
(9H, m, cyclohexyl), 1.13 (1H, br, cyclohexyl) ppm. 13C NMR
4.1.5.1. 7-(2-bromoethoxy)-2H-chromen-2-one (9a). Yield: 39.7%.
1H NMR (400 MHz, DMSO-d6):
d
8.00 (1H, dd, J ¼ 9.5, 0.6 Hz, C¼CH),
(100 MHz, DMSO-d6): d 158.40, 157.71, 143.28, 137.98, 129.12,
7.64 (1H, d, J ¼ 8.6 Hz, ArH), 7.03 (1H, d, J ¼ 2.4 Hz, ArH), 6.98 (1H,
dd, J ¼ 8.6, 2.5 Hz, ArH), 6.30 (1H, d, J ¼ 9.5 Hz, C¼CH), 4.47e4.39
(2H, m, CH2), 3.88e3.79 (2H, m, CH2) ppm. 13C NMR (100 MHz,
DMSO-d6): 161.46, 160.67, 155.79, 144.71, 130.06, 113.25, 113.22,
113.19, 101.93, 68.85, 31.49 ppm.
128.78, 127.01, 123.82, 71.85, 50.95, 38.45, 36.11, 33.80, 24.16,
21.32 ppm. ESI-MS m/z 367.7 (Mþ1)þ.
4.1.7.3. 5-((1-(3-Phenylpropyl)-1H-1,2,3-triazol-4-yl)methyl)-1,3,5-
triazaspiro[5.5]undeca-1,3-diene-2,4-diamine hydrochloride (7c).
Yield: 68.3%. Purity: 98.9%. 1H-NMR (400 MHz, DMSO-d6):
d 8.70
4.1.5.2. 7-(3-bromopropoxy)-2H-chromen-2-one (9b). Yield: 86.9%.
(1H, s, NH), 8.14 (1H, s, triazolyl-H), 7.57 (2H, br, NH), 7.38 (2H, br,
NH), 7.34e7.26 (2H, m, ArH), 7.19 (2H, m, ArH), 4.75 (2H, s, CH2-
triazine), 4.37 (2H, t, J ¼ 7.0 Hz, CH2-triazole), 2.53 (2H, t, J ¼ 7.0 Hz,
CH2Ar), 2.11 (2H, dq, J ¼ 9.6, 7.2 Hz, CH2), 1.86e1.51 (9H, m, cyclo-
hexyl), 1.23e1.07 (1H, m, cyclohexyl) ppm. 13C NMR (100 MHz,
1H NMR (400 MHz, DMSO-d6):
d
7.98 (1H, dd, J ¼ 9.5, 0.6 Hz, C¼CH),
7.63 (1H, d, J ¼ 8.5 Hz, ArH), 7.00 (1H, d, J ¼ 2.4 Hz, 1 ¼ ArH), 6.96
(1H, dd, J ¼ 8.6, 2.4 Hz, ArH), 6.29 (1H, d, J ¼ 9.4 Hz, C¼CH), 4.19 (2H,
t, J ¼ 6.0 Hz, CH2), 3.67 (2H, t, J ¼ 6.6 Hz, CH2), 2.27 (2H, p, J ¼ 6.3 Hz,
CH2) ppm. 13C NMR (100 MHz, DMSO-d6): 161.95, 160.70, 155.82,
144.73, 130.00, 113.10, 113.05, 112.96, 101.75, 66.58, 32.07,
31.44 ppm.
DMSO-d6):
d 158.45, 157.76, 143.47, 141.16, 128.86, 128.73, 126.48,
123.82, 71.92, 49.45, 38.53, 33.91, 32.27, 31.82, 24.21, 21.33 ppm.
ESI-MS m/z 381.5 (Mþ1)þ.
4.1.5.3. 7-(4-bromobutoxy)-2H-chromen-2-one (9c). Yield: 70.6%.
4.1.7.4. 7-(2-(4-((2,4-diamino-1,3,5-triazaspiro[5.5]undeca-2,4-dien-
1-yl)methyl)-1H-1,2,3-triazol-1-yl)ethoxy)-2H-chromen-2-one hy-
drochloride (11a). Yield: 35.5%. Purity: 99.7%. 1H NMR (400 MHz,
1H NMR (400 MHz, DMSO-d6):
d
7.98 (1H, d, J ¼ 9.5 Hz, C¼CH), 7.62
(1H, d, J ¼ 8.5 Hz, ArH), 6.99 (1H, d, J ¼ 2.4 Hz, ArH), 6.94 (1H, dd,
J ¼ 8.6, 2.4 Hz, ArH), 6.28 (1H, d, J ¼ 9.5 Hz, C¼CH), 4.12 (2H, t,
J ¼ 6.3 Hz, CH2), 3.61 (2H, t, J ¼ 6.6 Hz, CH2), 2.01e1.93 (2H, m, CH2),
1.91e1.81 (2H, m, CH2) ppm. 13C NMR (100 MHz, DMSO-d6): 162.18,
160.74, 155.86, 144.77, 129.94, 113.19, 112.90, 112.79, 101.64, 67.91,
35.18, 29.43, 27.61 ppm.
DMSO-d6):
d 8.61 (1H, s, NH), 8.18 (1H, s, triazolyl-H), 7.99 (1H, d,
J ¼ 9.5 Hz, C¼CH), 7.62 (1H, d, J ¼ 8.6 Hz, ArH), 7.34 (3H, d, m, NH),
6.99 (1H, d, J ¼ 2.4 Hz, Ar’H), 6.88 (1H, dd, J ¼ 8.7, 2.4 Hz, ArH), 6.30
(1H, d, J ¼ 9.5 Hz, C¼CH), 4.81 (2H, t, J ¼ 4.9 Hz, NCH2), 4.75 (2H, s,
triazolyl-H), 4.52 (2H, t, J ¼ 5.0 Hz, OCH2), 1.80e1.46 (9H, m,
cyclohexyl), 1.14e1.00 (1H, m, cyclohexyl) ppm. 13C NMR (100 MHz,
4.1.6. General procedure for the synthesis of azide derivatives 6a-
6c, 10a-10c and 13
DMSO-d6): d 160.83, 160.13, 157.93, 157.22, 155.23, 144.20, 143.00,
129.58, 124.05, 112.81, 112.75, 112.58, 101.41, 71.38, 66.84, 48.48,
To the corresponding bromides 5, 9a-9i, or 12 (1.0 equivalent) in
DMF, sodium azide (3.0 equivalent) was added. The resulting re-
action mixture was reflux for 5 h. After cooling down to room
temperature, the reaction mixture extracted with ethyl acetate
(3 ꢁ 20 mL). The combined organic layers were washed with brine
(1 ꢁ 10 mL) and dried over anhydrous Na2SO4. The crude com-
pound was obtained in vacuo, and directly used in next step
without further purification.
37.89, 33.34, 23.66, 20.82 ppm. ESI-MS m/z 451.4 (M þ H)þ.
4.1.7.5. 7-(3-(4-((2,4-Diamino-1,3,5-triazaspiro[5.5]undeca-2,4-
dien-1-yl)methyl)-1H-1,2,3-triazol-1-yl)propoxy)-2H-chromen-2-
one hydrochloride (11b). Yield: 38.3%. Purity: 99.9%. 1H NMR
(400 MHz, DMSO-d6):
d 8.75 (1H, s, NH), 8.16 (1H, s, triazolyl-H),
8.00 (1H, d, J ¼ 9.5 Hz, C¼CH), 7.64 (1H, d, J ¼ 8.6 Hz, ArH),
7.61e7.48 (2H, br, NH), 7.42 (2H, br, NH), 6.96 (1H, d, J ¼ 2.4 Hz,
ArH), 6.90 (1H, dd, J ¼ 8.6, 2.4 Hz, ArH), 6.29 (1H, d, J ¼ 9.5 Hz,
C¼CH), 4.74 (2H, s, CH2-triazine), 4.55 (2H, t, J ¼ 6.9 Hz, NCH2), 4.07
(2H, t, J ¼ 6.0 Hz, OCH2), 2.30 (2H, p, J ¼ 6.5 Hz, CH2), 1.82e1.50 (9H,
m, cyclohexyl), 1.12 (1H, m, cyclohexyl) ppm. 13C NMR (100 MHz,
4.1.7. General procedure for the synthesis of compound 7a-7c, 11a-
11c, 14, 19a-19c, 20a-20c and 21
The corresponding alkyne (1.0 equivalent) and azide derivative
(1.2 equivalent) were added into THF (5 mL). Sodium ascorbate
(0.06 eq) was dissolved in water (4 mL), followed by copper (II)
sulfate pentahydrate (0.03 eq). The reaction mixture was stirred
overnight at room temperature and the reaction was monitored by
TLC. After the completion of reaction, the mixture was concentrated
in vacuum and dissolved in MeOH/DCM (1:9). The transparent
filtrate was obtained by filtration and purified by column chro-
matography (MeOH/DCM ¼ 1:50 to 1:10).
DMSO-d6):
d 161.90, 160.70, 158.41, 155.80, 144.78, 143.48, 130.01,
124.05, 113.13, 113.04, 112.96, 101.65, 71.95, 65.81, 47.03, 38.48,
33.91, 29.66, 24.20, 21.28 ppm. ESI-MS m/z 465.4 (M þ H)þ.
4.1.7.6. 6-((5-(4-((2,4-Diamino-1,3,5-triazaspiro[5.5]undeca-2,4-
dien-1-yl)methyl)-1H-1,2,3-triazol-1-yl)pentyl)oxy)-2H-chromen-2-
one hydrochloride (11c). Yield: 46.9%. Purity: 99.2%. 1H NMR
(400 MHz, DMSO-d6):
d 8.68 (1H, br, NH), 8.14 (1H, s, triazolyl-H),
7.99 (1H, d, J ¼ 9.5 Hz, C¼CH), 7.62 (1H, d, J ¼ 8.6 Hz, ArH), 7.58
(2H, br, NH), 7.37 (2H, br, NH), 6.97 (1H, d, J ¼ 2.4 Hz, ArH), 6.92 (1H,
dd, J ¼ 8.6, 2.4 Hz, ArH), 6.28 (1H, d, J ¼ 9.5 Hz, C¼CH), 4.74 (2H, s,
CH2-triazine), 4.45 (2H, t, J ¼ 7.0 Hz, NCH2), 4.09 (2H, t, J ¼ 6.4 Hz,
OCH2), 1.97 (2H, p, J ¼ 7.1 Hz, CH2), 1.84e1.50 (11H, m, cyclohexyl,
CH2), 1.13 (1H, m, cyclohexyl) ppm. 13C NMR (100 MHz, DMSO-d6):
4.1.7.1. 5-((1-benzyl-1H-1,2,3-triazol-4-yl)methyl)-1,3,5-triazaspiro
[5.5]undeca-1,3-diene-2,4-diamine hydrochloride (7a). Yield: 60.9%.
Purity: 98.6%. 1H-NMR (400 MHz, DMSO-d6):
d 8.66 (1H, s, NH), 8.16
(1H, s, triazolyl-H), 7.55 (2H, br, NH), 7.42e7.23 (7H, m, ArH, NH),
5.61 (2H, s, CH2-benzyl), 4.73 (2H, s, CH2-triazine), 1.82e1.50 (9H,
m, cyclohexyl), 1.20e1.03 (1H, m, cyclohexyl) ppm. 13C NMR
d
162.14, 160.74, 158.43, 157.74, 155.85, 144.79, 143.40, 129.97,
(100 MHz, DMSO-d6):
128.60, 128.30, 124.08, 71.91, 53.31, 38.46, 33.89, 24.19, 21.32 ppm.
d
158.42, 157.74, 143.72, 136.52, 129.19,
123.88, 113.15, 112.92, 112.81, 101.63, 71.91, 68.07, 49.53, 38.50,
33.89, 26.94, 25.87, 24.20, 21.31 ppm. ESI-MS m/z 479.4 (M þ H)þ.

X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
273
4.1.7.7. 2-[4-(2,4-Diamino-1,3,5-triaza-spiro[5.5]undeca-2,4-dien-1-
ylmethyl)-[1,2,3]triazol-1-yl]-1-phenyl-ethanone hydrochloride (14).
Yield: 63.4%. Purity: 95.9%. 1H NMR (400 MHz, DMSO-d6):
8.70
(1H, s, NH), 8.09 (1H, s, triazolyl-H), 8.06 (2H, d, J ¼ 8.0, 0.9 Hz, ArH),
7.74 (1H, t, J ¼ 6.8 Hz, ArH), 7.63 (2H, s, NH), 7.61 (2H, t, J ¼ 7.7 Hz,
ArH), 7.38 (2H, s, NH), 6.19 (2H, s, COCH2), 4.82 (2H, s, triazolyl-H),
1.97e1.89 (2H, m, cyclohexyl), 1.87e1.54 (7H, m, cyclohexyl),
1.20e1.03 (1H, m, cyclohexyl) ppm. 13C NMR (100 MHz, DMSO-d6):
ESI-MS m/z 543.4 (Mþ1)þ.
d
4.1.7.12. 7-(3-(4-((4-(2,4-diamino-1,3,5-triazaspiro[5.5]undeca-2,4-
dien-1-yl)phenoxy)methyl)-1H-1,2,3-triazol-1-yl)propoxy)-2H-chro-
men-2-one hydrochloride (20b). Yield: 41.2%. Purity: 99.1%. 1H NMR
(400 MHz, DMSO-d6):
d 9.04 (1H, s, NH), 8.34 (1H, s, triazolyl-H),
8.00 (1H, d, J ¼ 9.5 Hz, C¼CH), 7.63 (1H, d, J ¼ 8.6 Hz, ArH), 7.56
(2H, s, NH), 7.26 (2H, d, J ¼ 8.9 Hz, ArH), 7.17 (2H, d, J ¼ 9.0 Hz, Ar),
7.01e6.88 (2H, m, Ar’H), 6.29 (1H, d, J ¼ 9.5 Hz, C¼CH), 6.22 (1H, s,
NH), 5.18 (2H, s, OCH2), 4.58 (2H, t, J ¼ 6.9 Hz, CH2), 4.11 (2H, t,
J ¼ 6.0 Hz, CH2), 2.33 (2H, p, J ¼ 6.4 Hz, CH2), 1.87e1.84 (2H, m,
cyclohexyl), 1.74e1.47 (5H, m, cyclohexyl), 1.29e1.24 (2H, m,
cyclohexyl), 0.94e0.84 (1H, m, cyclohexyl) ppm. 3C NMR (100 MHz,
DMSO-d6): 161.92, 160.71, 159.18, 158.32, 158.21, 155.79, 144.77,
142.81, 131.79, 129.99, 127.76, 125.36, 116.43, 113.12, 113.04, 112.96,
101.74, 71.83, 65.93, 61.92, 47.10, 35.18, 29.70, 24.43, 21.19 ppm. ESI-
MS m/z 557.4 (Mþ1)þ.
d
192.61, 158.44, 157.80, 143.38, 134.68, 134.62, 129.44, 128.63,
125.70, 71.95, 56.41, 38.40, 33.87, 24.21, 21.36 ppm. ESI-MS m/z
381.4 (M þ H)þ.
4.1.7.8. 5-(4-((1-benzyl-1H-1,2,3-triazol-4-yl)methoxy)phenyl)-
1,3,5-triazaspiro[5.5]undeca-1,3-diene-2,4-diamine
(19a). Yield: 62.7%. Purity: 99.3%. 1H NMR (400 MHz, DMSO-d6):
8.99 (1H, s, NH, exchangeable proton), 8.33 (1H, s, triazoyl-H), 7.54
hydrochloride
d
(2H, br, NH), 7.41e7.32 (5H, m, Ar’H), 7.26 (2H, d, J ¼ 8.9 Hz, ArH),
7.16 (2H, d, J ¼ 8.9 Hz, ArH), 6.22 (1H, s, NH), 5.63 (2H, s, OCH2), 5.18
(2H, s, CH2), 1.87e1.84 (2H, m, cyclohexyl), 1.66e1.50 (5H, m,
cyclohexyl), 1.31e1.23 (2H, m, cyclohexyl), 0.96e0.86 (1H, m,
4.1.7.13. 7-(4-(4-((4-(2,4-diamino-1,3,5-triazaspiro[5.5]undeca-2,4-
dien-1-yl)phenoxy)methyl)-1H-1,2,3-triazol-1-yl)butoxy)-2H-chro-
men-2-one hydrochloride (20c). Yield: 66.6%. Purity: 99.0%. 1H NMR
cyclohexyl) ppm. 13C NMR (100 MHz, DMSO-d6):
d 159.18, 158.31,
158.24, 143.11, 136.44, 131.79, 129.23, 128.65, 128.45, 127.78, 125.29,
116.43, 71.83, 61.88, 53.35, 35.17, 24.42, 21.21 ppm. ESI-MS m/z
445.7 (M þ H)þ.
(400 MHz, DMSO-d6):
d 8.98 (1H, br, NH), 8.31 (1H, s, triazolyl-H),
7.99 (1H, d, J ¼ 9.5 Hz, C¼CH), 7.63 (1H, d, J ¼ 8.6 Hz, Ar’H), 7.53
(3H, br, NH), 7.27 (2H, d, J ¼ 8.9 Hz, ArH), 7.17 (2H, d, J ¼ 9.0 Hz, ArH),
7.03e6.91 (2H, m, Ar’H), 6.28 (1H, d, J ¼ 9.5 Hz, C¼CH), 6.22 (1H, br,
NH), 5.18 (2H, s, OCH2), 4.47 (2H, t, J ¼ 7.0 Hz, CH2), 4.11 (2H, t,
J ¼ 6.4 Hz, CH2), 2.07e1.93 (2H, m, CH2), 1.88e1.84 (2H, m, cyclo-
hexyl), 1.79e1.65 (2H, m, CH2), 1.68e1.47 (5H, m, cyclohexyl),
1.29e1.25 (2H, m, cyclohexyl), 1.00e0.83 (1H, m, cyclohexyl) ppm.
13C NMR (100 MHz, DMSO-d6): 162.15, 160.74, 159.19, 158.29,
155.85, 144.79, 142.77, 131.80, 129.96, 127.76, 125.14, 116.43, 113.17,
112.93, 112.81, 101.66, 71.81, 68.07, 61.93, 49.55, 36.16, 26.92, 25.93,
24.41, 21.21 ppm. ESI-MS m/z 571.4 (Mþ1)þ.
4.1.7.9. 5-(4-((1-(2-Phenylethyl)-1H-1,2,3-triazol-4-yl)methoxy)
phenyl)-1,3,5-triazaspiro[5.5]undeca-1,3-diene-2,4-diamine hydro-
chloride (19b). Yield: 57.0%. Purity: 95.6%. 1H NMR (400 MHz,
DMSO-d6):
d 8.99 (1H, s, NH), 8.20 (1H, s, triazoyl-H), 7.54 (3H, br,
NH), 7.29e7.14 (9H, m, ArH and Ar’H), 7.23 (1H, s, Ar’H), 7.18 (4H, dt,
J ¼ 15.2, 6.2 Hz, Ar’H), 6.21 (1H, s, NH), 5.16 (2H, s, OCH2), 4.64 (2H, t,
J ¼ 7.3 Hz, NCH2), 3.17 (2H, t, J ¼ 7.3 Hz, CH2Ar’H), 1.86e1.85 (2H, m,
cyclohexyl), 1.67e1.50 (5H, m, cyclohexyl), 1.32e1.23 (2H, m, 2H,
cyclohexyl), 0.95e0.95 (1H, m, cyclohexyl) ppm. 13C NMR
(100 MHz, DMSO-d6): d 159.17,158.31, 158.24, 142.64, 138.03, 131.79,
4.1.7.14. 2-(4-((4-(2,4-diamino-1,3,5-triazaspiro[5.5]undeca-2,4-
dien-1-yl)phenoxy)methyl)-1H-1,2,3-triazol-1-yl)-1-phenylethanone
hydrochloride (21). Yield: 43.8%. Purity: 98.2%. 1H NMR (400 MHz,
129.12, 128.87, 127.75, 127.04, 125.06, 116.47, 71.83, 61.93, 50.94,
36.12, 35.18, 24.41, 21.21 ppm. ESI-MS m/z 459.7 (M þ H)þ.
DMSO-d6):
d 9.00 (1H, s, NH), 8.24 (1H, s, triazolyl-H), 8.08 (2H, d,
4.1.7.10. 5-(4-((1-(3-Phenylpropyl)-1H-1,2,3-triazol-4-yl)methoxy)
phenyl)-1,3,5-triazaspiro[5.5]undeca-1,3-diene-2,4-diamine hydro-
chloride (19c). Yield: 47.1%. Purity: 99.4%. 1H NMR (400 MHz,
J ¼ 7.1 Hz, Ar’H), 7.75 (1H, t, J ¼ 7.4 Hz, Ar’H), 7.62 (2H, t, J ¼ 7.6 Hz,
Ar’H), 7.52 (3H, br, NH), 7.28 (2H, d, J ¼ 8.9 Hz, Ar), 7.21 (2H, d,
J ¼ 9.0 Hz, ArH), 6.23 (2H, s, NCH2C¼O), 5.25 (2H, s, OCH2),
1.89e1.84 (2H, m, cyclohexyl), 1.65e1.57 (5H, m, cyclohexyl),
1.39e1.21 (2H, m, cyclohexyl), 0.99e0.85 (1H, m, cyclohexyl) ppm.
13C NMR (100 MHz, DMSO-d6): 192.61, 159.24, 158.26, 158.19,
142.72, 134.74, 134.56, 131.81, 129.46, 128.64, 127.78, 126.94, 116.46,
71.77, 61.90, 56.38, 35.11, 24.39, 21.25 ppm. ESI-MS m/z 473.5
(Mþ1)þ.
DMSO-d6):
d 9.02 (1H, s, NH), 8.31 (1H, s, triazolyl-H), 7.55 (3H, br,
NH), 7.31e7.16 (9H, m, ArH and Ar’H), 6.22 (1H, br, NH), 5.19 (1H, s,
OCH2), 4.39 (2H, J ¼ 7.1 Hz, NCH2), 2.59e2.55 (2H, m, Ar’CH2),
2.19e2.11 (2H, m, Ar’CH2), 1.87e1.84 (2H, m, cyclohexyl), 1.70e1.52
(5H, m, cyclohexyl), 1.31e1.23 (2H, m, cyclohexyl), 0.94e0.84 (1H,
m, cyclohexyl) ppm. 13C NMR (100 MHz, DMSO-d6):
d 159.18,
158.33, 158.23, 142.81, 141.15, 131.79, 128.86, 128.76, 127.77, 126.48,
125.07, 116.47, 71.83, 62.96, 49.43, 35.18, 32.33, 31.77, 24.42,
21.19 ppm. ESI-MS m/z 473.4 (M þ H)þ.
4.2. Strains and growth condition
4.1.7.11. 7-(2-(4-((4-(2,4-diamino-1,3,5-triazaspiro[5.5]undeca-2,4-
dien-1-yl)phenoxy)methyl)-1H-1,2,3-triazol-1-yl)ethoxy)-2H-chro-
men-2-one hydrochloride (20a). Yield: 66.5%. Purity: 99.5%. 1H
Liquid and solid cultures of M. bovis BCG (Pasteur 1173P2) and
Mycobacterium tuberculosis H37Rv (ATCC 27294) were grown as
described previously [25]. Liquid cultures were grown on rollers
using filter sterilized Middlebrook 7H9 broth (BD Difco) supple-
mented with 0.5% albumin, 0.2% dextrose, 0.0003% catalase, 0.085%
sodium chloride, 0.2% glycerol and 0.05% Tween 80. Solid cultures
were grown on Middlebrook 7H11 agar (BD Difco) supplemented
with 0.006% oleic acid, 0.5% albumin, 0.2% dextrose, 0.0003%
catalase, 0.085% sodium chloride and 0.5% glycerol. HepG2 (ATCC
HB.8065) cells were cultured at 37 ꢂC with 5% CO2 atmosphere in
DMEM media (Gibco) complemented with 10% heat-inactivated
FBS (Gibco), penicillin (100 U/mL, Gibco) and streptomycin
NMR (400 MHz, DMSO-d6): d 8.93 (1H, s, NH), 8.37 (1H, s, triazolyl-
H), 8.00 (1H, d, J ¼ 9.5 Hz, C¼CH), 7.63 (1H, d, J ¼ 8.6 Hz, Ar’H), 7.44
(2H, br, NH), 7.26 (2H, d, J ¼ 8.9 Hz, ArH), 7.17 (2H, d, J ¼ 9.0 Hz,
ArH), 7.03 (1H, d, J ¼ 2.4 Hz, Ar’H), 6.94 (1H, dd, J ¼ 8.6, 2.5 Hz, Ar’H),
6.30 (1H, d, J ¼ 9.5 Hz, C¼CH), 6.23 (1H, br, NH), 5.19 (2H, s, OCH2),
4.84 (2H, t, J ¼ 5.0 Hz, CH2), 4.55 (2H, t, J ¼ 5.0 Hz, CH2), 1.87e1.84
(2H, m, cyclohexyl), 1.70e1.52 (5H, m, cyclohexyl), 1.30e1.24 (2H,
m, cyclohexyl), 0.94e0.84 (1H, m, cyclohexyl) ppm. 13C NMR
(100 MHz, DMSO-d6): 161.32, 160.64, 159.18, 158.27, 158.22, 155.73,
144.70, 142.88, 131.78, 130.06, 127.75, 125.81, 116.42, 113.31, 113.26,
113.14, 101.99, 71.79, 67.30, 61.83, 49.36, 35.14, 24.39, 21.21 ppm.
(100
mg/mL, Gibco). For hemolysis assay, red blood cells were ob-
tained from Interstate Blood Bank, Inc. Laboratory, USA.

274
X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
4.3. MIC determination
DMSO was added instead of the test compound. A blank reading
was also taken, in which the DHFR enzyme was omitted and 10
mL
Minimum inhibitory concentrations (MICs) were determined by
the broth microdilution method as described previously [23e25].
Briefly, mycobacterial cells were grown in roller bottles until OD600
reached mid log phase (0.4e0.6). The cultures were then diluted to
a starting bacterial OD600 of 0.01. Compounds (dissolved in 90%
of DMSO was added instead of the test compound. The assay was
carried out at room temperature, and the rate of consumption of
NADPH during the conversion of DHF to tetrahydrofolate (THF) was
monitored by taking absorbance readings at 340 nm every 30 s over
6 min using a Hitachi U-1900 UV/visible spectrophotometer. A
graph of absorbance readings versus time was plotted and the slope
of the graph over the linear range from 30 s to 6 min was taken to
represent the rate of reaction. The percentage inhibition at each
concentration of inhibitor was calculated using the following
formulae:
DMSO) were serially diluted 2 fold in 100
mL volume in 96 well
microtiter plates (Flat bottom polystyrene, Corning Inc.). Then,
bacterial suspensions were dispensed to each well in 100
mL vol-
ume making the final volume and cell density 200
mL and
5 ꢁ 105 CFU/mL, respectively. Plates were then aseptically sealed
with Breathe-Easy sealing membrane (Sigma), placed in a humid-
ified lock-lock box and incubated at 37OC for 10 days with shaking
at 80 rpm. Growth inhibition was measured by reading optical
density (OD600) using TECAN infinite M200Pro microplate reader
(TECAN, Salzburg, Austria) and the inhibitory activities were
interpreted based on the MIC50 and MIC90 values, the minimum
concentrations that inhibit 50% and 90% of bacterial growth.
Rifampicin and untreated culture containing DMSO were included
as inhibition and growth controls, respectively.
Slope of inhibited enzyme
Activity ¼
Slope of uninhibited enzyme
Inhibition (%) ¼ 100 ꢀ Activity
Graphs of percentage inhibition against logarithmic concentra-
tion were plotted for each compound and IC50 values were calcu-
lated using GraphPad Prism Version 5.01.
4.4. Cytotoxicity and hemolysis assay
4.6. Construction of DHFR overexpressing M. bovis BCG strain
The in vitro cytotoxicity effect was determined using CellTiter
96® AQueous One Solution Cell Proliferation Assay (MTS) kit
(Promega). In the assay, human hepatic cell lines (HepG2, ATCC HB-
8065) were exposed to two fold serially diluted compound con-
centrations and viability was assessed by measuring the decrease in
the absorbance at 490 nm with a TECAN M200Pro microplate
reader. Absorbance was plotted as a function of compound con-
centrations and the minimum cidal concentration corresponding to
50% reduction in cell viability was determined. Triton X-100 was
used as a cytotoxic positive control.
Lytic effect on red blood cells was also evaluated following a
modified procedure described previously [30]. Briefly, cells were
washed with PBS, re-suspended to a final cell number of 106/mL
and incubated with test compounds for 24 h. After exposure, the
red blood cells were subsequently centrifuged (1000 g for 10 min)
and absorbance of the supernatant was read at 540 nm. PBS and
0.1% Triton X-100 were included as negative and positive controls,
respectively.
We used mycobacterial shuttle vector pMV262 (conferring
kanamycin resistance) to overexpress DHFR in M. bovis BCG under
the control of a constitutive promoter hsp60 [26]. The coding re-
gion of dfrA was first PCR amplified (KOD polymerase, Toyobo Life
Science) with primers containing BamHI and EcoRI restriction sites
(Forward BamHI; ccgggatccATGGTGGGGCTGATCTGGG and Reverse
EcoRI; ccggaattcTCATGAGCGGTGGTAGCTGT) and cloned inframe
(NotI-BamHI). Purified plasmid DNA from E. coli DH5a was elec-
troporated to M. Bovis BCG as previously described [31]. The
amount of DHFR overexpression was quantified at the transcript
level by using qRT-PCR (SuperScript® III) [32]. RNA from M. Bovis
BCG wild type was isolated from the equivalent of 20 mL of cells at
OD60nm ¼ 0.4. Cultures were centrifuged, re-suspended in TRIzol
(Invitrogen) and subjected to bead beating by using FastPrep-24™
5G Instrument (MP Biomedicals, 2 ꢁ 45 s each, 5 min on ice be-
tween pulses). RNA was purified using PureLink™ RNA Mini Kit
with TURBO DNA-free kit (Invitrogen). cDNA was created from 1 mg
of total RNA with the SuperScript III First Strand Synthesis System
(Invitrogen) by using random primers. Quantitative PCR was per-
formed with the FastStart Essential DNA Green Master (Roche)
using the LightCycler 96 Real-Time PCR System (Roche) by using
primers for dfrA (Forward; GTCGACCTCGGTAACCTCAC, Reverse;
4.5. DHFR inhibition assay
Potassium phosphate buffer (0.15 M, pH 7) was used for all the
DHFR assays. A solution of NADPH (2 mg/mL, 2 mM) in potassium
phosphate buffer (0.15 M, pH 7) was prepared immediately before
use and kept on ice. A solution of dihydrofolate (DHF) (1 mg/mL,
2 mM) was prepared immediately before use by suspending DHF in
deionized water and adding a drop of sodium hydroxide (1 M)
solution to aid dissolution. The DHF solution was protected from
light and kept on ice. Recombinant human DHFR was purchased
from Sigma-Aldrich (St. Louis, MO, USA). Recombinant Mtb DHFR
was supplied by Protein Production Platform, School of Biological
Sciences, Nanyang Technological University. The test compounds
were dissolved in DMSO and diluted to the appropriate
concentrations.
TGAGGTTGTCGGTTCACTCG)
ATGACGGCCTTCGGGTTGTAA,
and
Reverse;
16S
RNA
(Forward;
CGGCTGCTGGCACG-
TAGTTG), respectively. Relative expression (DDCq) was calculated
as described by using 16S RNA as reference [33]. Para-
aminosalicylic acid (PAS) was used as a positive control [34].
4.7. Molecular docking study
Molecular docking is another commonly used virtual screening
method for drug discovery. Unlike machine learning, it is a target-
based method to estimate the binding affinity between ligand and
target. In this study, molecular docking screening was conducted
using the Surflex-Dock module in SYBYLeX 2.1 (Tripos associate
Inc., St. Louis, MO, USA).
The assay was conducted by mixing the appropriate volumes of
phosphate buffer, NADPH, DHF, recombinant human or Mtb DHFR
and test compound in a 1 mL quartz cuvette to obtain a solution
containing 50
m
M DHF, 60
m
M NADPH, 0.15 M phosphate buffer (pH
For small molecule preparation, the 3D structure of the com-
pounds was minimized in SYBYL-X 2.1. The partial atomic charge
was calculated by the Gasteiger Huckel method and energy mini-
mization was performed using the Tripos force field with a
7), 1.5 ꢁ 10ꢀ3 units of recombinant human DHFR or 0.3 ng/mL Mtb
DHFR and various concentrations of the assayed compounds. For
each experiment, a control reading was taken in which 10 mL of

X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
275
distance-dependent dielectric and the Powell conjugate gradient
algorithm convergence criteria.
to Treatment of Tuberculosis (SPRINT-TB; www.sprinttb.org) led by
Nick Paton.
The 3D structure of Mtb DHFR (PDB: 1DF7) and human DHFR
(PDB: 2W3M) was obtained from protein data bank (PDB). Gener-
ally, During protein preparation, the co-crystallized ligand was first
removed from the structure together with the water molecules. H
atoms were then added and side chains were fixed. Protomol
generation parameters were threshold 0.5 and boat 0 for both en-
zymes. Surflex-Dock used an empirically scoring function based on
the binding affinity, which was expressed in -log Kd involving hy-
drophobicity, polarity, repulsion, entropy, solvation and crash
terms. The higher the score is, the more tightly the ligand binds to
receptor. The top 20 poses of each ligand were obtained and
analyzed for their interactions with the protein.
(iii) The NUS Medicine BSL-3 Core Facility, National University of
Singapore, National University Health System and the Singapore
Ministry of Health, National Medical Research Council, Center Grant
‘MINE’, Research Core #4 (NMRC/CG/013/2013).
(iv) Bill Jacobs from the Albert Einstein College of Medicine for
sharing his plasmid pMV262.
References
[1] M. Raviglione, G. Sulis, Tuberculosis, 2015: burden, challenges and strategy for
control and elimination, Infect. Dis. Rep. 8 (2016) 6570.
[2] World Health Organization (WHO), Global Tuberculosis Report, 2015. http://
www.who.int/tb/publications/global_report/en/.
[3] M. Pieroni, D. Machado, E. Azzali, S. Santos Costa, I. Couto, G. Costantino,
M. Viveiros, Rational design and synthesis of thioridazine analogues as en-
hancers of the antituberculosis therapy, J. Med. Chem. 58 (2015) 5842e5853.
[4] World Health Organization (WHO), Global Tuberculosis Report, 2015. http://
www.who.int/tb/publications/global_report/en/.
4.8. DMPK study
The rat liver microsome (RLM) stability and solubility assays
were carried out at Drug Development Unit (DDU), National Uni-
versity of Singapore (NUS).
[5] V. Dartois, The path of anti-tuberculosis drugs: from blood to lesions to
mycobacterial cells, Nat. Rev. Microbiol. 12 (2014) 159e167.
[6] E.W. Orenstein, S. Basu, N.S. Shah, J.R. Andrews, G.H. Friedland, A.P. Moll, N.R.
Gandhi, A.P. Galvani, Treatment outcomes among patients with multidrug-
resistant tuberculosis: systematic review and meta-analysis, Lancet Infect.
Dis., 9 153e161.
[7] I.M. Kompis, K. Islam, R.L. Then, DNA and RNA synthesis: antifolates, Chem.
Rev. 105 (2005) 593e620.
[8] D.L. Wright, A.C. Anderson, Antifolate agents: a patent review (2006 e 2010),
Expert Opinion on Therapeutic Patents, 21 (2011) 1293e1308.
[9] S. Hawser, S. Lociuro, K. Islam, Dihydrofolate reductase inhibitors as anti-
bacterial agents, Biochem. Pharmacol. 71 (2006) 941e948.
[10] L. Arundhati Chandrashekhar, M. Deepak Amarnath, K. Tengku Karmila,
B. Sanjib, D. Mariam Sohel, Repositioning of DHFR inhibitors, Curr. Top. Med.
Chem. 16 (2016) 2125e2143.
For determining solubility, the compound was serially diluted
from a 10 mM stock in DMSO to a series of final concentrations,
consisting of 100 mM, 50 mM, 25 mM, 12.5 mM, 6.25 mM and 3.13 mM
in an aqueous buffer (45 mM ethanolamine, 45 mM potassium
dihydrogen phosphate, 45 mM potassium acetate, pH 7.4).
Following an incubation of 90 min at room temperature, the
aqueous dilution was filtrated into a 96-well UV plate. After filtra-
tion, 80
mL of filtrate was transferred from each well to another well
in a 96-well UV plate. 20
mL of acetonitrile was added to each well.
This will give the 80:20 buffer:acetonitrile composition of the
sample solutions. After that, the sample solutions were diluted by
[11] Molly R. Nixon, Kurt W. Saionz, M.-S. Koo, Michael J. Szymonifka, H. Jung,
Justin P. Roberts, M. Nandakumar, A. Kumar, R. Liao, T. Rustad, James
C. Sacchettini, Kyu Y. Rhee, Joel S. Freundlich, David R. Sherman, Folate
pathway disruption leads to critical disruption of methionine derivatives in
Mycobacterium tuberculosis, Chem. Biol. 21 (2014) 819e830.
adding another 200
was read at 330 nm.
mL of buffer. The absorbance of the solutions
Metabolic stability was measured in liver microsomes from
male and female rats by LC-MS-based measuring 3 M over time up
to 45 min at 37 ꢂC. Incubation mixtures consisted of 7.5
L of 20 mg/
mL FRLM and MRLM (final: 0.3 mg microsome protein/mL), 2.5
of 600 M compound in acetonitrile (final: 3 M), 440 L of 0.1 M
ꢀ
[12] A. Kumar, A. Guardia, G. Colmenarejo, E. Perez, R.R. Gonzalez, P. Torres,
m
ꢀ
ꢀ
ꢀ
D. Calvo, R.M. Gomez, F. Ortega, E. Jimenez, R.C. Gabarro, J. Rullas, L. Ballell,
D.R. Sherman, A Focused screen identifies antifolates with activity on Myco-
bacterium tuberculosis, ACS Infect. Dis. 1 (2015) 604e614.
m
mL
[13] N.R. Tawari, S. Bag, A. Raju, A.C. Lele, R. Bairwa, M.K. Ray, M.G.R. Rajan,
L.U. Nawale, D. Sarkar, M.S. Degani, Rational drug design, synthesis and bio-
logical evaluation of dihydrofolate reductase inhibitors as antituberculosis
agents, Future Med. Chem. 7 (2015) 979e988.
[14] W. Zhou, E.W. Scocchera, D.L. Wright, A.C. Anderson, Antifolates as effective
antimicrobial agents: new generations of trimethoprim analogs, MedChem-
Comm 4 (2013) 908e915.
m
m
m
phosphate buffer (pH 7.4). The mixture was first shaken for 5 min
for pre-incubation in a shaking water bath at 37 ꢂC. Reaction was
initiated by adding 50
centration of 1 mM NADPH in the mixture. The total volume of the
reaction mixture was 500 L. For metabolic stability studies, ali-
quots of 50 L of the incubation sample mixture were collected at 0,
5, 10, 15, 30, and 45 min. After collection of samples, the reaction
was terminated with 100 L of chilled acetonitrile containing the
mL of 10 mM NADPH to obtain a final con-
[15] A.C. Anderson, D.L. Wright, Antifolate agents: a patent review (2010 e 2013),
Expert Opin. Ther. Pat. 24 (2014) 687e697.
m
m
[16] X. Ma, S.-T. Tan, C.-L. Khoo, H.-M. Sim, L.-W. Chan, W.-K. Chui, Synthesis and
antimicrobial activity of N1-benzyl or N1-benzyloxy-1,6-dihydro-1,3,5-
triazine-2,4-diamines, Bioorg. Med. Chem. Lett. 21 (2011) 5428e5431.
[17] P. Ong, Triazaspiroalkenes as Potential Inhibitors of Dihydrofolate Reductase,
Singapore, 2009.
[18] R. Li, R. Sirawaraporn, P. Chitnumsub, W. Sirawaraporn, J. Wooden,
F. Athappilly, S. Turley, W.G.J. Hol, Three-dimensional structure of
M. tuberculosis dihydrofolate reductase reveals opportunities for the design of
novel tuberculosis drugs, J. Mol. Biol. 295 (2000) 307e323.
m
internal standard. The mixture was then centrifuged at 10,000 ꢁ g
to remove the protein and the supernatant was subsequently
applied to LC-MS/MS analysis. Based on compound half-life, in vitro
intrinsic clearance CLint was calculated.
[19] E.F.F. da Cunha, T.d.C. Ramalho, R.B. de Alencastro, E.R. Maia, Interactions of 5-
deazapteridine derivatives with Mycobacterium tuberculosis and with human
dihydrofolate reductases, J. Biomol. Struct. Dyn. 22 (2004) 119e130.
[20] E.J. Modest, Chemical and biological studies on 1,2-dihydro-s-triazines. II.
Three-component synthesis, J. Org. Chem. 21 (1956) 1e13.
[21] U.H. Manjunatha, P.W. Smith, Perspective: challenges and opportunities in TB
drug discovery from phenotypic screening, Bioorg. Med. Chem. 23 (2015)
5087e5097.
Conflicts of interest
All authors have given approval to the final version of the
manuscript and declare that they have no conflict of interest.
Acknowledgments
[22] J. Magdalena, P. Supply, C. Locht, Specific differentiation between Mycobac-
terium bovis BCG and virulent strains of the Mycobacterium tuberculosis
complex, J. Clin. Microbiol. 36 (1998) 2471e2476.
This work was supported by:
[23] J.M. Andrews, Determination of minimum inhibitory concentrations,
J. Antimicrob. Chemother. 48 (2001) 5e16.
(i) The National University of Singapore Academic Research
Fund Tier 1 R-148-000-205-112, which was awarded to W.K. Chui.
(ii) The Singapore Ministry of Health's National Medical
Research Council's TCR Flagship grant NMRC/TCR/011-NUHS/2014
which was awarded to T. Dick. The grant contributes to part of
Singapore Programme of Research Investigating New Approaches
ꢀ
[24] D. Blanco-Ruano, D.M. Roberts, R. Gonzalez-Del-Rio, D. Alvarez, M.J. Rebollo,
ꢀ
ꢀ
E. Perez-Herran, A. Mendoza, Antimicrobial susceptibility testing for Myco-
bacterium sp, Mycobact. Protoc. (2015) 257e268.
[25] S.B. Lakshminarayana, T.B. Huat, P.C. Ho, U.H. Manjunatha, V. Dartois, T. Dick,
S.P. Rao, Comprehensive physicochemical, pharmacokinetic and activity
profiling of anti-TB agents, J. Antimicrob. Chemother. 70 (3) (2014) 857e867.

276
X. Yang et al. / European Journal of Medicinal Chemistry 144 (2018) 262e276
[26] C. Stover, V. De La Cruz, New uses of BCG for recombinant vaccines, Nature
351 (1991) 456.
[27] W. Hong, Z. Chang, Y. Wang, T. Sun, J. Pu, I.C. Paterson, H. Wang, The influence
of glycerol on the binding of methotrexate to Mycobacterium tuberculosis
dihydrofolate reductase: a molecular modelling study, Mol. Simul. 41 (2015)
1540e1545.
[30] W. Moreira, D.B. Aziz, T. Dick, Boromycin kills mycobacterial persisters
without detectable resistance, Front. Microbiol. 7 (2016) 199.
[31] T. Parish, N.G. Stoker, Electroporation of mycobacteria, Mycobact. Protoc.
(1998) 129e144.
[32] Y. Yamada, T. Dick, Mycobacterial caseinolytic protease gene Regulator ClgR is
a substrate of caseinolytic protease, mSphere 2 (2017).
[28] J.-P. Ebejer, M.H. Charlton, P.W. Finn, Are the physicochemical properties of
[33] P. MW, A new mathematical model for relative quantification in real-time RT-
PCR, Nucleic Acids Res. 29 (2001).
antibacterial
compounds
really
different
from
other
drugs?
J. Cheminformatics 8 (2016) 30.
[34] J. Zheng, E.J. Rubin, P. Bifani, V. Mathys, V. Lim, M. Au, J. Jang, J. Nam, T. Dick,
J.R. Walker, para-Aminosalicylic acid is a prodrug targeting dihydrofolate
reductase in Mycobacterium tuberculosis, J. Biol. Chem. 288 (2013)
23447e23456.
[29] S.B. Lakshminarayana, T.B. Huat, P.C. Ho, U.H. Manjunatha, V. Dartois, T. Dick,
S.P.S. Rao, Comprehensive physicochemical, pharmacokinetic and activity
profiling of anti-TB agents, J. Antimicrob. Chemother. 70 (2015) 857e867.