Improvement of the novel inhibitor for Mycobacterium enoyl-acyl carrier protein reductase (InhA): a structure–activity relationship study of KES4 assisted by in silico structure-based drug screening

Article abstract of DOI:10.1038/s41429-020-0293-6

InhA or enoyl-acyl carrier protein reductase of Mycobacterium tuberculosis (mtInhA), which controls mycobacterial cell wall construction, has been targeted in the development of antituberculosis drugs. Previously, our in silico structure-based drug screening study identified a novel class of compounds (designated KES4), which is capable of inhibiting the enzymatic activity of mtInhA, as well as mycobacterial growth. The compounds are composed of four ring structures (A–D), and the MD simulation predicted specific interactions with mtInhA of the D-ring and methylene group between the B-ring and C-ring; however, there is still room for improvement in the A-ring structure. In this study, a structure–activity relationship study of the A-ring was attempted with the assistance of in silico docking simulations. In brief, the virtual chemical library of A-ring-modified KES4 was constructed and subjected to in silico docking simulation against mtInhA using the GOLD program. Among the selected candidates, we achieved synthesis of seven compounds, and the bioactivities (effects on InhA activity and mycobacterial growth and cytotoxicity) of the synthesized molecules were evaluated. Among the compounds tested, two candidates (compounds 3d and 3f) exhibited superior properties as mtInhA-targeted anti-infectives for mycobacteria than the lead compound KES4.

Full text of DOI:10.1038/s41429-020-0293-6

The Journal of Antibiotics
https://doi.org/10.1038/s41429-020-0293-6
ARTICLE
Improvement of the novel inhibitor for Mycobacterium enoyl-acyl
carrier protein reductase (InhA): a structure–activity relationship
study of KES4 assisted by in silico structure-based drug screening
3
Junichi Taira1 Tomohiro Umei1 Keitaro Inoue1 Mitsuru Kitamura2 Francois Berenger1 James C. Sacchettini
●
●
●
●
●
●
Hiroshi Sakamoto1 Shunsuke Aoki1
●
Received: 28 November 2019 / Revised: 9 January 2020 / Accepted: 10 February 2020
© The Author(s), under exclusive licence to the Japan Antibiotics Research Association 2020
Abstract
InhA or enoyl-acyl carrier protein reductase of Mycobacterium tuberculosis (mtInhA), which controls mycobacterial cell
wall construction, has been targeted in the development of antituberculosis drugs. Previously, our in silico structure-based
drug screening study identified a novel class of compounds (designated KES4), which is capable of inhibiting the enzymatic
activity of mtInhA, as well as mycobacterial growth. The compounds are composed of four ring structures (A–D), and the MD
simulation predicted specific interactions with mtInhA of the D-ring and methylene group between the B-ring and C-ring;
however, there is still room for improvement in the A-ring structure. In this study, a structure–activity relationship study of the A-
ring was attempted with the assistance of in silico docking simulations. In brief, the virtual chemical library of A-ring-modified
KES4 was constructed and subjected to in silico docking simulation against mtInhA using the GOLD program. Among the
selected candidates, we achieved synthesis of seven compounds, and the bioactivities (effects on InhA activity and mycobacterial
growth and cytotoxicity) of the synthesized molecules were evaluated. Among the compounds tested, two candidates (compounds
3d and 3f) exhibited superior properties as mtInhA-targeted anti-infectives for mycobacteria than the lead compound KES4.
Introduction
however, inappropriate and/or long-term drug treatment has
allowed the emergence of multidrug-resistant TB (MDR-
TB) and extensively drug-resistant TB (XDR-TB). Treat-
ment for MDR-TB and XDR-TB is long, expensive, and
characterized by a high rate of adverse events. The main
difficulty is the identification of at least four active drugs in
order to design an effective regimen [5]. It is therefore
imperative to develop a novel class of anti-infectives cap-
able of sterilizing drug-resistant mycobacteria.
Tuberculosis (TB), which is caused by Mycobacterium
tuberculosis, is one of the most common and serious
infectious diseases in the world [1–4]. Multiple drug com-
bination therapy (e.g., isoniazid (INH), rifampicin, etham-
butol, and pyrazinamide) has generally been applied in TB
treatment to prevent the development of drug resistance;
The enoyl-acyl carrier protein reductase (InhA, EC
1.3.1.9) is known to be involved in the biosynthesis of
mycolic acids, a major component of mycobacterial cell
walls. The compounds capable of inhibiting enzymatic
activity of M. tuberculosis InhA (mtInhA) have been devel-
oped as anti-TB drugs [6–9]. Indeed, mtInhA is the primary
target of the first-line antitubercular drug INH and of the
second-line drug ethionamide (ETH). The enzyme catalyzes
the NADH-dependent reduction of the double bond of 2-
trans-enoyl-acyl carrier protein, an essential step in the fatty
acid elongation cycle of the type II fatty acid synthase (FAS-
II) system, which is significantly different to the mammalian
fatty acid synthesis pathway [10–13]. INH and ETH inhibit
InhA by covalently bonding to NAD at the active site of
Supplementary information The online version of this article (https://
doi.org/10.1038/s41429-020-0293-6) contains supplementary
material, which is available to authorized users.
* Shunsuke Aoki
aokis@bio.kyutech.ac.jp
1
Department of Bioscience and Bioinformatics, Graduate School of
Computer Science and Systems Engineering, Kyushu Institute of
Technology, Iizuka 820-8502, Japan
2
Department of Applied Chemistry, Kyushu Institute of
Technology, 1-1 Sensui-cho, Tobata, Kitakyushu 804-8550, Japan
3
Department of Biochemistry & Biophysics, Texas A&M
University, College Station, TX 77843-2128, USA

J. Taira et al.
mtInhA. Although many INH- and ETH-resistant clinical
isolates contain mutations within the inhA locus, resistance
can be conferred by mutation(s) at Ile21, Ile47, or Ser94 [14].
We have previously demonstrated that in silico structure-
based drug screening (SBDS) is suitable for screening lead
compounds of novel enzyme inhibitors and is a promising
tool that offers cost- and labor-saving advantages over the
use of random high-throughput screening methods [15–19].
A series of KES compounds, which contain 3-(phenoxy)
benzylpiperazine pharmacophore, has been identified, with
the assistance of in silico SBDS, as being inhibitory to
mtInhA [17]. Among the compounds tested, KES4: 1-(2-
furoyl)-4-{3-(phenoxy)benzyl}piperazine exhibited a potent
inhibition of mycobacterial growth, which appears to be
associated with InhA inhibition. As shown in Fig. 1, KES4
is composed of two phenyl, piperazinyl, and furoyl groups
(the ring structures are referred to as the A–D-rings). Our
previous simulation presumed that the 2-furoyl group and its
oxygen atom (D-ring) were involved in hydrophobic inter-
actions with Leu218 and hydrogen bonding with Tyr158,
respectively [17]. The hydrogen atom on methylene,
between the piperazinyl group (C-ring) and the 3-
phenoxybenzyl group (B-ring), interacts with Phe149 by a
CH-π interaction. Since Phe149, Tyr158, and Leu218 are
located in the active site of mtInhA, the B–D-rings play
important roles in mtInhA inhibition, although specific
interactions were not predicted on the A-rings. The simu-
lation suggested that the compound can still be improved.
A conventional structure–activity relationship (SAR)
study, based on minor modifications of the lead com-
pounds, is an orthodox strategy for obtaining KES4-based
novel effective InhA inhibitors, although this requires
deep insights into protein–molecule interactions, as well
as synthesis of many compounds. The in silico SBDS
technique is one of the powerful tool to find lead com-
pounds with novel pharmacophore; however, related
molecular selection, on the basis of the matched molecular
pair (MMP) method [19], is dependent on the scale of the
chemical library used, i.e., commercially unavailable
compounds are excluded from the selection because
chemical libraries have generally been constructed based
on commercially available (known) compounds. Thus, to
overcome these issues, a tailored chemical library of A-
ring-modified KES4 derivatives was prepared and sub-
jected to in silico docking simulation in this study.
Fig. 1 Chemical structures of KES4 and predicted interaction with
mtInhA. a Chemical structure of KES4. The phenyl and heterocyclic
rings are referred to as A–D-rings. b Predicted interactions of KES4
(light blue) to active site of mtInhA (gray): where the blue lines
between the methylene (between the C-ring and B-ring) and the
Phe149 indicates CH–π interaction; the orange dots surrounding the 2-
furoyl group of indicates hydrophobic interactions with the Leu218;
the green line between the D-ring proximal carbonyl group and the
Tyr158 indicates hydrogen bond; the red shaded area represents gap
between the A-ring and the enzyme
PubChem database (https://pubchem.ncbi.nlm.nih.gov/),
and a chemical library of KES4 derivatives was generated
using the canonical SMILES notation. Conversion to
MOL2-formatted three-dimensional chemical structures, the
addition of hydrogen atoms and electric charges, and
structural optimization were performed using the Energy
Minimize, Protonate 3D, and Energy Minimize modules,
respectively, which have been implemented in the Mole-
cular Operating Environment (MOE) version 2011.10
(Chemical Computing Group, Montreal, Canada). In gen-
eral, to screen for a competitive inhibitor using the SBDS
strategy, the utilization of substrate- or inhibitor-bound
forms of protein structures is preferable to apo-form struc-
tures. The protein data bank (PDB) structural data of
mtInhA in complex with pyrrolidine carboxamide deriva-
tive (PDB-id 4U0J) [20] was therefore employed in the
present SBDS after following structural adjustments. In
brief, hydrogen atoms and electric charges were added
using the Protonate 3D module in the MOE software after
the removal of water, oxygen atoms, and the pyrrolidine
carboxamide-derived inhibitor from the structure. Then,
hydrogen atoms were optimized with the Energy Minimize
module in MOE software.
Experimental procedure
Preparation of the compound library and protein
structural data
A total of 243 commercially available phenol analogs (with
>70% structural similarity to phenol) were selected from the

Improvement of the novel inhibitor for Mycobacterium enoyl-acyl carrier protein reductase (InhA): a. . .
In silico docking simulation
were trained using all data for each Tox21 target. The
logistic curve was used for the toxicity probability model
(a and b are the two parameters adjusted upon fitting).
This method is called Platt scaling [27]. The “x” is the raw
toxicity score given by a classification model for a given
candidate molecule.
Binding affinity between the 243 compounds in the che-
mical library and the mtInhA substrate-binding cavity was
predicted using tandem (two-step) GOLD screening with
Cambridge Crystallographic Data Center GOLD suite ver-
sion 5.3 [15–18, 21, 22]. The primary screening of the
compounds with single conformations identified 209 com-
pounds with a GOLD score above that of KES4 (81.80).
Then, ten conformations of each compound were generated
by the Conformation Search module of MOE and were
subjected to secondary screening. As the result of the sec-
ondary screening, 199 compounds with a GOLD score
above that of KES4 (80.36 1.57) were identified. Binding
between the compounds and amino acids, constituting the
mtInhA substrate-binding cavity, was presumed by ligand
interaction (LI) and protein–ligand interaction fingerprint
(PLIF) modules in MOE software. According to the LI and
PLIF results, the compounds that could interact with
Phe149, Thr158, and Leu218 of mtInhA were kept as
candidates, although those which could interact with Ile21,
Ile47, or Ser94 were excluded. Sixteen candidate com-
pounds were selected and designated as KET1-16 (Fig. S1).
We also predicted the toxicity of the compounds (Fig. S2).
Classification Quantitative Structure–Activity Relationship
(classification QSAR) models were trained on the Tox-
icology in the 21st Century (Tox21) dataset [23].
Tox21 contains toxicity data for 12 toxicity end-points (NR-
AR, NR-AR-LBD, NR-AhR, NR-Aromatase, NR-ER,
NR-ER-LBD, NR-PPAR-gamma, SR-ARE, SR-ATAD5,
SR-HSE, SR-MMP, and SR-p53) and ~8000 ligands. We
standardized all molecules using Francis Atkinson’s
standardizer (https://github.com/flatkinson/standardiser).
Molecules were then encoded using Faulon’s signature
molecular descriptor [24], with height one bond and
parameterized over MOL2 atom types. Our molecular
encoder is released as open-source [25]. For each target,
only molecules tested as toxic or nontoxic on that target
were retained. Not all Tox21 molecules have a toxicity
outcome against all targets. We then trained a classifica-
tion QSAR model on each target. Each model consists in a
bag [26] of 11 linear support vector machines (R package
svmpath; dot product kernel) trained on balanced boot-
straps. Each model had an AUC between 0.75 and 0.85.
However, the models for SR-ATAD5 and NR-Aromatase
did not show a good separation of the predicted scores
between toxic and nontoxic molecules, so their predic-
tions might be inaccurate. Finally, for each target, a
toxicity probability model (Eq. (1)) was fitted with gnu-
plot using five folds cross validation when training each
QSAR model. This probability model allows the trans-
formation of raw scores from each QSAR model into
toxicity probabilities. The final production QSAR models
p ftoxicgðxÞ ¼ 1:0=ð1:0 þ exp ða  x þ bÞÞ:
ð1Þ
Regarding Fig. S2, several compounds look quite safe
across all Tox21 toxicity end-points: KES4, KES3, KET1,
KET12, and KET8. Some other compounds have a warning
regarding a possible effect on the mitochondrial function
(stress response pathway mitochondrial membrane potential
—SR-MMP in Tox21). Those compounds are: KES5,
KET13, KET4, KET5, and KET7.
Chemistry and chemical methods
The general synthetic scheme employed in this study is
illustrated in Fig. 2. All reagents and solvents were com-
mercially available and required no further purification. All
reactions were carried out under nitrogen or argon. All
compounds were purified by silica gel column chromato-
graphy (Fuji Silysia Silica gel PSQ-100B or Kanto Che-
mical Silica Gel 60N). NMR spectra were recorded on a
Bruker Avance 500 system (Bruker) in CDCl3 [TMS
(for 1H, δ = 0) or CDCl3 (for 13C, δ = 77.0) was used as an
internal standard]. Mass spectra were obtained with a
JEOL JMS-SX102A mass spectrometer (JEOL) using an
electron ionization method with a double focusing sector
detector.
General procedure for synthesis of intermediate
3-phenoxybenzaldehyde derivatives 2a–g
Potassium hydroxide (1.2 eq) was added to a solution of
phenol derivatives 1a–g (1.8 eq) in m-xylene. The solution
was refluxed at 130 °C for 2 h and was then cooled to room
temperature. Copper (I) chloride (0.013 eq) and 2-(3-bro-
mophenyl)-1,3-dioxolane (1 eq) were added to the reaction
mixture, and stirring continued overnight at 130 °C. The
reaction was quenched by the addition of water and then
extracted with dichloromethane. The organic layer was
washed with brine, and the organic solvent was evaporated
under reduced pressure. The residue was redissolved in 1:1
THF/H2O and 2,2,2-trifluoroacetic acid (1.5 eq). The reac-
tion mixture was then stirred at room temperature for 2 h.
The reaction was quenched by an aqueous solution of
sodium hydroxide and was then extracted with ethyl
acetate. The organic layer was washed with brine and
dried with anhydrous Na2SO4. The organic solvent was
evaporated under reduced pressure to obtain the crude
products 2a–g.

J. Taira et al.
Fig. 2 Synthetic scheme of KES4 related compounds. (i) KOH, m-xylene, CuCl (I), 140 °C. (ii) TFA, THF/water (1:1) r.t. (iii) NaBH(OAc)3,
CH2Cl2, r.t
General procedure for synthesis of the target 1-(2-
furoyl)-4-(3-phenoxybenzyl)piperazine derivatives
3a–g
Compound 3b (KET3)
(1-(2-furoyl)-4-{3-[3-(tert-butyl)phenoxy]benzyl}piper-
azine): Yield 39%; 1H NMR (500 MHz, CDCl3) δ 7.49 (1H,
s, Ar–H), 7.26 (2H, q, J = 8.2 Hz, Ar–H), 7.14 (1H, d, J =
7.9 Hz, Ar–H), 7.08 (1H, s, Ar–H), 7.05 (1H, d, J = 3.8 Hz,
Ar–H), 6.97 (1H, d J = 3.5 Hz, Ar–H), 6.90 (1H, dd, J =
8.1, 1.7 Hz, Ar–H), 6.79 (1H, dd, J = 8.0, 2.4 Hz, Ar–H),
6.47 (1H, s, Ar–H), 3.79 (4H, s, CH2 × 2), 3.51 (2H, s,
CH2), 2.50 (4H, t, J = 4.8 Hz, CH2 × 2), 1.30 (9H, s, CH3 ×
3); 13C NMR (126 MHz, CDCl3) δ (ppm) 159.0, 157.5,
153.5, 148.0, 143.6, 139.7, 129.6, 129.2, 123.7, 120.3,
119.2, 117.4, 116.4, 116.3, 115.7, 111.2, 62.5, 34.8, 31.3;
MS m/z (M + H, C26H30N2O3), found 419.
1-(2-Furoyl)piperazine (1.1 eq) and sodium acetoxybor-
ohydride (1.2 eq) were added to a solution of 3-
phenoxybenzaldehyde derivatives 2a–g (1 eq) in dichlor-
omethane. The reaction mixture was stirred at room tem-
perature. After overnight stirring, the reaction was quenched
by the addition of water and was then extracted with
dichloromethane. The organic layer was washed with brine
and dried with anhydrous Na2SO4. The organic solvent was
evaporated under reduced pressure to obtain the crude
products 3a–g. The spectral data of compounds 3a–g are
reported herein in detail.
Compound 3c (KET6)
Compound 3a (KET1)
(1-(2-furoyl)-4-{3-[2-(sec-butyl)phenoxy]benzyl}piper-
azine): Yield 25%; 1H NMR (500 MHz, CDCl3) δ 7.46 (1H,
s, Ar–H), 7.29 (1H, dd, J = 7.1, 2.1 Hz, Ar–H), 7.24 (1H, t,
J = 7.9 Hz, Ar–H), 7.15–7.12 (2H, m, Ar–H), 7.00 (1H, d,
J = 7.7 Hz, Ar–H), 6.97 (1H, d, J = 3.5 Hz, Ar–H), 6.92
(1H, s, Ar–H), 6.88 (1H, dd, J = 6.8, 2.2 Hz, Ar–H), 6.81
(1H, dd, J = 8.2, 2.5 Hz, Ar–H), 6.47 (1H, s, Ar–H), 3.79
(4H, s, CH2 × 2), 3.50 (2H, s, CH2), 2.49 (4H, t, J = 4.9 Hz,
CH2 × 2), 1.68–1.53 (3H, m, CH, CH2), 1.20 (3H, d, J =
7.0, CH3), 0.82 (3H, t, J = 7.4, CH3); 13C NMR (126 MHz,
CDCl3) δ (ppm) 159.1, 158.5, 153.9, 148.0, 143.6, 139.7,
139.1, 129.4, 127.6, 126.8, 124.2, 123.0, 119.8, 118.1,
116.4, 116.2, 111.2, 62.6, 34.0, 30.1, 21.0, 12.3; MS m/z
(M + H, C26H30N2O3), found 419.
(1-(2-furoyl)-4-{3-[4-(methoxycarbonylmethyl)phenoxy]
benzyl}piperazine): Yield 92%; 1H NMR (500 MHz,
CDCl3) δ 7.46 (1H, s, Ar–H), 7.29–7.23 (3H, m, Ar–H),
7.07 (1H, d, J = 7.7 Hz, Ar–H), 7.03 (1H, s, Ar–H),
6.98–6.94 (3H, m, Ar–H), 6.90 (1H, dd, J = 7.6, 2.4 Hz,
Ar–H), 6.47 (1H, s, Ar–H), 3.80 (4H, s, CH2 × 2), 3.71 (3H,
s, CH3), 3.61 (2H, s, CH2), 3.52 (2H, s, CH2), 2.50 (4H, t,
J = 4.9 Hz, CH2 × 2); 13C NMR (126 MHz, CDCl3) δ (ppm)
172.1, 159.0, 157.2, 156.4, 148.0, 143.6, 139.9, 130.6,
129.6, 128.8, 124.0, 119.4, 118.8, 117.7, 116.2, 111.2,
62.5, 53.1, 52.0, 40.4; MS m/z (M + H, C25H26N2O5),
found 435.

Improvement of the novel inhibitor for Mycobacterium enoyl-acyl carrier protein reductase (InhA): a. . .
Compound 3d (KET8)
Hz, Ar–H), 6.47 (1H, s, Ar–H), 3.80 (4H, s, CH2 × 2), 3.52
(2H, s, CH2), 2.50 (4H, t, J = 4.7 Hz, CH2 × 2), 1.42 (9H, s,
CH3 × 3); 13C NMR (126 MHz, CDCl3) δ (ppm) 159.1,
157.9, 155.8, 148.0, 143.6, 140.9, 139.7, 129.5, 127.2,
127.1, 123.3, 123.2, 120.1, 119.2, 117.5, 116.3, 111.2,
62.5, 53.1, 34.8, 30.1; MS m/z (M + H, C26H30N2O3),
found 419.
(1-(2-furoyl)-4-[3-(2-ethylphenoxy)benzyl]piperazine):
Yield 43%; H NMR (500 MHz, CDCl3,) δ 7.46 (1H, s,
1
Ar–H), 7.29–7.19 (2H, m, Ar–H), 7.16 (1H, dt, J = 7.7, 1.6
Hz, Ar–H), 7.10 (1H, dt, J = 7.4, 1.3 Hz, Ar–H), 7.01 (1H,
d, J = 7.6 Hz, Ar–H), 6.97 (1H, d, J = 3.5 Hz, Ar–H), 6.94
(1H, s, Ar–H), 6.88 (1H, dd, J = 8.0, 1.2 Hz, Ar–H), 6.80
(1H, dd, J = 8.1, 2.1 Hz, Ar–H), 6.46 (1H, s, Ar–H), 3.80
(4H, s, CH2 × 2), 3.50 (2H, s, CH2), 2.65 (2H, q, J = 2.5 Hz,
CH2), 2.49 (4H, t, J = 4.7 Hz, CH2 × 2), 1.20 (3H, t, J =
7.6, CH3); 13C NMR (126 MHz, CDCl3) δ (ppm) 159.1,
158.2, 154.2, 148.0, 143.6, 139.7, 135.8, 131.9, 130.4,
129.9, 129.5, 127.6, 127.1, 124.0, 123.1, 119.6, 118.2,
116.2, 111.2, 62.5, 53.1, 23.2, 14.5; MS m/z (M + H,
C24H26N2O3), found 391.
Antimycobacterial assay
Mycobacterium smegmatis (IAM 12065 strain, RIKEN
BioResource Center, Saitama, Japan) was grown at 37 °C
for 24 h in 3.7% brain–heart infusion broth (Sigma, St.
Louise, MO, USA). Cultures were then diluted 70-fold with
broth that contained the candidate compounds in 96-well
flat-bottom clear plates (Corning, NY, USA). Each well was
inoculated with 150 μl of culture. Isoniazid (INH) (LKT
Laboratories, St. Paul, MN, USA) and 0.3% DMSO were
used as positive and negative controls, respectively. The
plates were incubated at 37 °C for 24 h, and then the cell
cultures were subjected to growth inhibition assays. The
inhibition of bacterial growth was determined by measuring
OD595 using a Bio-Rad Model 680 microplate reader (Bio-
Rad, Hercules, CA, USA).
Compound 3e (KET10)
(1-(2-furoyl)-4-{3-[3-(trifluoromethyl)phenoxy]benzyl}
piperazine): Yield 42%; 1H NMR (500 MHz, CDCl3) δ 7.46
(1H, s, Ar–H), 7.44 (1H, t, J = 8.1, Ar–H), 7.35–7.31 (2H,
m, Ar–H), 7.22 (1H, s, Ar–H), 7.17 (1H, dd, J = 8.2, 2.2
Hz, Ar–H), 7.13 (1H, d, J = 7.6 Hz, Ar–H), 7.06 (1H, s,
Ar–H), 6.98 (1H, d, J = 3.5 Hz, Ar–H), 6.93 (1H, dd, J =
7.8, 1.8 Hz, Ar–H), 6.47 (1H, s, Ar–H), 3.80 (4H, s, CH2 ×
Cytotoxic assay
2), 3.54 (2H, s, CH2), 2.50 (4H, t, J = 4.6 Hz, CH2 × 2); 13
C
NMR (126 MHz, CDCl3) δ (ppm) 159.1, 157.8, 156.3,
148.0, 143.6, 140.3, 130.3, 129.9, 124.8, 121.6, 119.9,
119.6, 118.2, 116.3, 115.1, 111.2, 62.4, 53.1; MS m/z (M +
H, C23H21F3N2O3), found 431.
The Madin–Darby canine kidney (MDCK) cell lines were
maintained in high-glucose Dulbecco’s Modified Eagle
Medium (DMEM) supplemented with 10% fetal bovine
serum (FBS), 2 mM L-glutamine, 100 U ml−1 of penicillin,
and 100 μg ml−1 of streptomycin and cultured at 37 °C in a
humidified atmosphere containing 5% CO2. The cells
(1.5 × 104 cells/well) were seeded into 96-well plates 6 h
before serum starvation with 0.25% FBS containing
DMEM. After 24 h of serum starvation, the compounds (50
μM) were added. Triclosan (TCS) (50 μM) and DMSO
(0.3%) were used as positive and negative controls,
respectively. The cultures were incubated for an additional
24 h, and then the cytotoxicity of the compounds was
analyzed with a Cell Counting Kit-8 (Dojin, Kumamoto,
Japan), as previously described [17].
Compound 3f (KET12)
(1-(2-furoyl)-4-[3-(2-methylphenoxy)benzyl]piperazine):
1
Yield 46%; H NMR (500 MHz, CDCl3) δ 7.46 (1H, s,
Ar–H), 7.26–7.19 (2H, m, Ar–H), 7.16 (1H, t, J = 7.2 Hz,
Ar–H), 7.06 (1H, t, J = 7.4 Hz, Ar–H), 7.01 (1H, d, J = 7.5
Hz, Ar–H), 6.98 (1H, t, J = 1.8 Hz, Ar–H), 6.93 (1H, s,
Ar–H), 6.89 (1 H, d, J = 8.0 Hz, Ar–H), 6.78 (1H, dd, J =
8.1, 2.4 Hz, Ar–H), 6.46 (1H, s, Ar–H), 3.80 (4H, s, CH2 ×
2), 3.50 (2H, s, CH2), 2.49 (4H, t, J = 4.7 Hz, CH2 × 2),
2.23 (3H, s, CH3); MS m/z (M + H, C23H24N2O3),
found 377.
Assay of mtInhA activity
Compound 3g (KET15)
The detail of cloning, expression, and purification of
mtInhA have been reported by Molle et al. [28]. We pre-
viously reported the methods not only for synthesizing of
trans-2-decenoyl CoA as InhA substrate, but also for
determining mtInhA activity based on the technique by
Delaine et al. [17, 29]. The recombinant mtInhA briefly, the
preincubation of enzyme, and inhibitors was performed in
50 μl (total volume) of 30 mM PIPES (pH 6.8), 150 mM
(1-(2-furoyl)-4-{3-[2-(tert-butyl)phenoxy]benzyl}piper-
azine): Yield 66%; 1H NMR (500 MHz, CDCl3) δ 7.46 (1H,
s, Ar–H), 7.40 (1H, dd, J = 8.3, 1.7 Hz, Ar–H), 7.26 (1H, t,
J = 7.9 Hz, Ar–H), 7.13 (1H, td, J = 7.7, 1.7 Hz, Ar–H),
7.07–7.03 (2H, m, Ar–H), 6.99–6.97 (2H, m, Ar–H), 6.86
(1H, dd, J = 7.6, 1.8 Hz, Ar–H), 6.80 (1H, dd, J = 8.0, 1.4

J. Taira et al.
NaCl, 200 μM NADH, 700 nM mtInhA, and compounds
3a, 3b, 3c, 3d, 3e, 3f, or 3g (50 μM). DMSO was used as
cosolvent, and its final concentration was <0.5%. After
1 min of preincubation at 25.5 °C, the addition of 200 μM
substrate (trans-2-decenoyl-CoA) initiated the reaction
which was followed at 340 nm (oxidation of NADH) and at
25 °C using an Infinite M200 PRO microplate reader
(TECAN, Männedorf, Switzerland). The inhibitory activity
of each compound tested was determined based on the
absorbances 5 min after the addition of substrate (trans-2-
decenoyl-CoA) and was expressed as the percentage of
mtInhA activity recorded in the absence of inhibitor.
OH/NH groups were protected by appropriate protecting
groups. Although the reasons for the failures to synthesize
the compounds are of interest, we did not investigate these
further in this study.
Growth inhibition of M. smegmatis by compounds
3a–g
As a primary evaluation of bioactivity, the effect of com-
pounds 3a–g on mycobacterial growth inhibition was
examined. M. smegmatis (biosafety level 1) was utilized as
a model mycobacteria in the antimycobacterial experiments.
M. smegmatis InhA shares 87% similarity with mtInhA in
amino acid sequence. We also previously confirmed that
tertiary structures of these inhibitor binding cavity were
comparable [17]. The inhibitory effects on mycobacterial
growth of the 50 μM compound solutions are presented in
Fig. 3. Compound 3a showed only ca. 40% growth inhi-
bition, while the other compounds (3b–g) almost com-
pletely inhibited mycobacterial growth. None of the
compounds exhibited any significant antimicrobial activity
against Escherichia coli at a concentration of 100 μM
(Fig. S3), suggesting that the compounds had mycobacteria-
selective growth inhibitory properties.
Next, we examined the dose dependency of the inhibi-
tory effect (Fig. 4). The concentrations of compounds that
exerted a 50% inhibition of M. smegmatis growth (IC50
values) were as follows: KES4 (lead compound), 9.6 μM;
compound 3a, 26.8 mM; compound 3b, 1.8 μM; compound
3c, 2.2 μM; compound 3d, 3.4 μM; compound 3e, 22.8 μM;
compound 3f, 7.8 μM; and compound 3g, 5.0 μM.
According to a previous study, the IC50 value of isoniazid
(the positive control) is 5.4 μM [16].
Results
Identification of candidate compounds by SBDS
We previously reported the identification of KES4, a novel
inhibitor of M. tuberculosis InhA [17]. Herein, a chemical
library composed of 243 A-ring modified KES4 derivatives
was prepared and subjected to in silico SBDS to assist in the
structural refinement of the KES4 A-ring for mtInhA inhi-
bition. The GOLD program evaluates the binding affinity of
compounds to enzyme active sites and provides a GOLD
score by calculating protein–ligand hydrogen bonding
energy, van der Waals contact energy, and molecule torsion
energy based on a genetic algorithm [30, 31]. The tandem
GOLD screening strategy, comprised of primary and sec-
ondary screenings with single and multiple conformations
of compounds, respectively, resulted in the identification of
199 compounds with GOLD scores above than that of
KES4 (80.36 1.57). The candidate compounds were fur-
ther selected on the basis of having specific interactions
with amino acids in mtInhA, which were determined by LI
and PLIF simulations. Compounds that could interact with
Phe149, Tyr158, and Leu218 of mtInhA were presumed to
be involved in interactions with KES4 and were kept as
candidates, while compounds that could interact with Ile21,
Ile47, or Ser94 of mtInhA, which are known to be asso-
ciated with the acquisition of drug resistance in M. tuber-
culosis, were excluded. The final 16 candidate compounds
were referred to as KET1-16, as shown in Fig. S1.
Synthesis of compounds
The synthetic strategy and structures of the KES4 deriva-
tives is shown in Fig. 2. Among the candidate compounds
KET1-16, we succeeded in synthesizing KET1, 3, 6, 8 10,
12, and 15. Hereafter, these synthesized compounds are
referred to as compounds 3a–g. The other compounds were
not obtained in sufficient quantities or adequate levels of
purity. It might be better that some phenols with additional
Fig. 3 Antimicrobial activity of the compounds on the growth of
M. smegmatis. The inhibitory effect on bacterial growth was monitored
by OD595 at 24 h after treatment with compounds (50 µM). Isoniazid
(INH, 50 μM) and 0.3% DMSO were used as positive and negative
controls, respectively. All experiments were performed in quad-
ruplicate, and data with error bars are expressed as mean SD. Sig-
nificance was evaluated by Dunnett’s test: **p < 0.001

Improvement of the novel inhibitor for Mycobacterium enoyl-acyl carrier protein reductase (InhA): a. . .
Fig. 4 Dose dependence of the compounds on the growth of M.
smegmatis: a KES4 (as lead compound); b–h Compounds 3a–g. The
inhibitory effect on bacterial growth was monitored by OD595 at 24 h
after treatment with the compounds. All experiments were performed
in quadruplicate, and data with error bars are expressed as mean SD.
IC50 values were determined based on a nonlinear regression method
Cytotoxicity of the compounds
To probe the damaging effects of the compounds on
mammalian cells, cytotoxicity was measured using the
MDCK cell line (Fig. 5). TCS, a representative InhA inhi-
bitor with a diphenyl ether pharmacophore [7, 8], was used
as a positive control. Compounds 3a and 3d–f did not show
a significant cytotoxicity. Compounds 3b and 3c showed
modest cytotoxicity at 50 μM, and this was comparable with
that of TCS.
Effects of compounds 3a–g on the enzymatic
activity of mtInhA
The effects of compounds 3a–g on the enzymatic activity of
recombinant mtInhA were examined to evaluate mtInhA
inhibition. Parikh et al. reported that TCS inhibits mtInhA
with a Ki value of 0.2 μM and, as shown in Fig. 6, mtInhA
was almost completely inhibited in the presence of 50 μM
TCS [7]. The enzymatic activity of mtInhA was suppressed
Fig. 5 Cytotoxicity of the compounds against the MDCK cell line. The
concentration of the compounds was 50 μM. Triclosan (TCS, 50 μM)
and 0.3% DMSO were used as positive and negative controls,
respectively. All experiments were performed in quadruplicate, and
data with error bars were expressed as mean S.D. Cell survival rates
were compared using Dunnett’s test for significance: **p < 0.001

J. Taira et al.
Table 1 Effect on mtInhA activity and lipophilicity (tPSA and LogP)
of the compounds
Compounds
mtInhA activity (%)a
tPSA
LogP
DMSO
KES4
3a
100 3.1
68.0 0.5
83.4 1.2
63.9 0.7
48.8 2.4
39.5 0.6
45.4 1.2
41.2 0.6
66.5 2.1
–
–
47.1
73.4
47.1
47.1
47.1
47.1
47.1
47.1
2.88
2.59
4.18
4.34
3.44
4.21
4.18
2.88
3b
3c
3d
3e
3f
3g
aMtInhA (0.7 μM) was preincubated with each compound (50 μM) at
room temperature for 1 min before its activity was measured. Vehicle
(0.3% DMSO) was used as the negative control. All experiments were
performed in triplicate, and data are expressed as mean SD
Fig. 6 Effect of compounds on enzymatic activity of mtInhA. The
enzymatic activity was measured as described in the experimental
procedures. MtInhA (0.7 μM) was preincubated with each compound
(50 μM) at room temperature for 1 min before its activity was mea-
sured. Triclosan (TCS) and 0.3% DMSO were used as positive and
negative controls, respectively. All experiments were performed in
quadruplicate, and data with error bars were expressed as mean SD
compared with the lead compound (KES4), while the
introduction of a methoxycarbonylmethyl group (compound
3a or KET1) on the A-ring abolished the antimycobacterial
activity (Figs. 3 and 4). As shown in Table 1, compound 3a
showed highest topological polar surface area and lowest
LogP values in comparison with KES4 and other candi-
dates. Since the cell wall of Mycobacterium is rich in
hydrophobic mycolic acid [10, 13], the hydrophilicity of
methoxycarbonylmethyl group can exert disadvantageous
effect on mycobacterial cell wall permeation among the
tested compounds. In addition, mtInhA inhibitory activity
of compound 3a was also lower than that of KES4 (Fig. 6).
Combined, both an antimycobacterial and mtInhA inhibi-
tory activity of KES4 could be spoiled by introduction of
electronegative atoms to the A-rings. Enzymatic inhibitory
potency of the KES4 derivatives also seems to correspond
with the size of hydrophobic substituents on the A-ring for
the following reasons (Fig. 6 and Table 1): (i) as described
in the previous section, mtInhA inhibition was moderate but
improved by the introduce of methyl and ethyl groups
(compounds 3d and 3f); (ii) mtInhA inhibitory activity by
the derivatives with the tert-butyl groups (compounds 3b
and 3g) were comparable to that of KES4; (iii) in spite of
trifluorometyl group is tert-butyl derivative, compound 3e
inhibited 55% of the enzymatic activity. These observations
suggested that the space around KES4 A-ring observed in
mtInhA binding cavity (see Fig. 1b) could be buried by
small and hydrophobic substituents and the interaction with
the enzyme could be facilitated, while bulky substituents
hindered the interaction.
to 68% (32% of the InhA activity was inhibited) in the
presence of 50 μM KES4 (Fig. 6 and Table 1). Among the
tested compounds, mtInhA activity was suppressed to
39.5% and 41.2% in the presence of compounds 3d and 3f,
respectively (these compounds inhibited ca. 60% of InhA
activity). Inhibitory activity of these compounds was only
moderately but more potent than that of KES4.
Discussion
Structural modification of KES4 could provide a novel class
of compounds with potent antimycobacterial activity. As
described in the previous section, KES4 seems to interact
with Leu218, Tyr158, and Phe149 in mtInhA through its
B–D-rings [17], while specific interaction was not predicted
for the A-ring (Fig. 1b). In this study, to increase the affinity
of KES4 for mtInhA, the chemical library constituted by a
series of A-ring-modified KES4 was prepared and subjected
to docking simulation. We tried to synthesize top 16 com-
pounds selected by the in silico docking simulations, though
the synthetic method for KES4 failed to introduce electro-
negative atoms on the A-rings excepting KET1 (i.e., KET2,
KET4, KET5, KET7, KET9, KET13, and KET16). Other
synthetic methods have not been tested at present stage due
to achievement of synthesis on all compounds was not
important issues. Resulting from the in silico docking
simulation and successive organic syntheses, we succeeded
in obtaining seven candidate compounds.
Compounds 3d and 3f showed preferable features in
terms of antimycobacterial activity, cytotoxicity toward
mammalian cells, inhibitory activity against mtInhA, and
antibacterial effects on Escherichia coli as indigenous model
bacteria. The properties of KES4 as antimycobacterial
Compounds 3b–g showed comparable or higher levels of
activity in terms of inhibiting Mycobacterium growth when

Improvement of the novel inhibitor for Mycobacterium enoyl-acyl carrier protein reductase (InhA): a. . .
agents might be improved by introduction of small alkyl
groups (e.g., ethyl, methyl) to the A-ring. Nevertheless, the
KES derivatives excepting compound 3a showed severe
growth inhibitions of M. smegmatis (Figs. 3 and 4),
whereas these compounds showed only moderate activity
on mtInhA inhibition (Fig. 6). These IC50 values were
estimated as 1.8–22.8 μM, though the enzymatic activity
was inhibited <50% under the presence of 50 μM of each
compound. These observations and discrepancy imply that
KES4 and its derivatives could inhibit other lethal targets in
Mycobacterium. As another possibility, metabolites from
KES4 and its derivatives might exert InhA inhibitory
effect.
In summary, we constructed virtual compound library of
A-ring-modified KES4 compounds using commercially
available phenol derivatives as building blocks for the A-
ring. We succeeded in synthesizing seven KES4 derivatives
(compounds 3a–g) from 16 candidates (KET1-16), which
were identified by in silico docking simulation. Two can-
didates (compounds 3d and 3f) exhibited superior proper-
ties to KES4 as mtInhA-targeted anti-infectives for
mycobacteria. Together, present study demonstrated that
custom-made compound library screening combined with
organic synthesis could be a cost- and labor-saving tech-
nique for the improvement of lead compounds. Experi-
mental validation of bioactivity should also be necessary
not only for acquisition of compounds with preferable
bioactivities, but also for the SAR study.
6. Banerjee A, Dubnau E, Quemard A, Balasubramanian V, Um KS,
Wilson T, et al. InhA, a gene encoding a target for isoniazid and
ethionamide in Mycobacterium tuberculosis. Science. 1994;263:
227–30.
7. Parikh SL, Xiao G, Tonge PJ. Inhibition of InhA, the enoyl
reductase from Mycobacterium tuberculosis, by triclosan and
isoniazid. Biochemistry. 2000;39:7645–50.
8. McMurry LM, McDermott PF, Levy SB. Genetic evidence that
InhA of Mycobacterium smegmatis is a target for triclosan.
Antimicrob Agents Chemother. 1999;43:711–3.
9. Quemard A, Sacchettini JC, Dessen A, Vilcheze C, Bittman R,
Jacobs WR Jr, et al. Enzymatic characterization of the target for
isoniazid in Mycobacterium tuberculosis. Biochemistry.
1995;34:8235–41.
10. Brennan PJ, Nikaido H. The envelope of mycobacteria. Annu Rev
Biochem. 1995;64:29–63.
11. Tonge PJ, Kisker C, Slayden RA. Development of modern InhA
inhibitors to combat drug resistant strains of Mycobacterium
tuberculosis. Curr Top Med Chem. 2007;7:489–98.
12. Rotta M, Pissinate K, Villela AD, Back DF, Timmers LF, et al.
Piperazine derivatives: synthesis, inhibition of the Mycobacterium
tuberculosis enoyl-acyl carrier protein reductase and SAR studies.
Eur J Med Chem. 2015;90:36–447.
13. Vilcheze C, Morbidoni HR, Weisbrod TR, Iwamoto H, Kuo M,
et al. Inactivation of the inhA-encoded fatty acid synthase II
(FASII) enoyl-acyl carrier protein reductase induces accumulation
of the FASI end products and cell lysis of Mycobacterium
smegmatis. J Bacteriol. 2000;182:59–4067.
14. Oliveira JS, Pereira JH, Canduri F, Rodrigues NC, de Souza ON,
de Azevedo WF Jr, et al. Crystallographic and pre-steady-state
kinetics studies on binding of NADH to wild-type and isoniazid-
resistant enoyl-ACP(CoA) reductase enzymes from Mycobacter-
ium tuberculosis. J Mol Biol. 2006;359:646–66.
15. Taira J, Ito T, Nakatani H, Umei T, Baba H, Kawashima S, et al.
In silico structure-based drug screening of novel antimycobacterial
pharmacophores by DOCK-GOLD tandem screening. Int
J
Acknowledgements This work was supported in part by a Takeda
Science Foundation to JT, and a Grant-in-Aid for Scientific Research
(C) (26460145) to SA and (18K05358) to HS from the Ministry of
Education, Culture, Sports, Science, and Technology of Japan.
Mycobacteriol. 2017;6:142–8.
16. Taira J, Morita K, Kawashima S, Umei T, Baba H, Maruoka T,
et al. Identification of a novel class of small compounds with anti-
tuberculosis activity by in silico structure-based drug screening. J
Antibiot. 2017;70:1057–64.
17. Kinjo T, Koseki Y, Kobayashi M, Yamada A, Morita K, Yama-
guchi K, et al. Identification of compounds with potential anti-
bacterial activity against Mycobacterium through structure-based
drug screening. J Chem Inf Model. 2013;53:1200–12.
18. Koseki Y, Kinjo T, Kobayashi M, Aoki S. Identification of novel
antimycobacterial chemical agents through the in silico multi-
conformational structure-based drug screening of a large-scale
chemical library. Eur J Med Chem. 2013;60:333–39.
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of
interest.
Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
19. Kanetaka H, Koseki Y, Taira J, Umei T, Komatsu H, Sakamoto H,
et al. Discovery of InhA inhibitors with anti-mycobacterial
activity through a matched molecular pair approach. Eur J Med
Chem. 2015;94:378–85.
References
20. He X, Alian A, Stroud R, Ortiz, de Montellano PR. Pyrrolidine
carboxamides as a novel class of inhibitors of enoyl acyl carrier
protein reductase from Mycobacterium tuberculosis. J Med Chem.
2006;49:6308–23.
1. Dye C, Williams BG. The population dynamics and control of
tuberculosis. Science. 2010;328:856–61.
2. Das P, Horton R. Tuberculosis-time to accelerate progress. Lan-
cet. 2010;375:1755–7.
21. Kobayashi M, Kinjo T, Koseki Y, Bourne CR, Barrow WW, Aoki
S. Identification of novel potential antibiotics against Staphylo-
coccus using structure-based drug screening targeting dihy-
drofolate reductase. J Chem Inf Model. 2014;54:1242–53.
22. Koseki Y, Aoki S. Computational medicinal chemistry for rational
drug design: Identification of novel chemical structures with
potential anti-tuberculosis activity. Curr Top Med Chem.
2014;14:176–88.
3. Das P, Horton R. Tuberculosis-getting to zero. Lancet.
2015;386:2231–2.
4. Lönnroth K, Castro KG, Chakaya JM, Chauhan LS, Floyd K,
Glaziou P, et al. Tuberculosis control and elimination 2010-50:
cure, care, and social development. Lancet. 2010;375:1814–29.
5. Sotgiu G, Centis R, D’Ambrosio L, Migliori GB. Tuberculosis
treatment and drug regimens. Cold Spring Harb Perspect Med.
2015;5:a017822.

J. Taira et al.
23. Schmidt CW. TOX 21: new dimensions of toxicity testing.
Environ Health Perspect. 2009;117:A348.
acid biosynthesis and growth of Mycobacterium tuberculosis. Mol
Microbiol. 2010;78:1591–605.
́
́
24. Faulon JL, Visco DP, Pophale RS. The signature molecular
descriptor. 1. Using extended valence sequences in QSAR and
QSPR studies. J Chem Info Comput Sci. 2003;43:707–20.
25. Berenger F. MolEnc: a molecular encoder using rdkit and OCaml.
2019. https://github.com/UnixJunkie/molenc. Accessed 19 Nov 2019.
26. Breiman L. Bagging predictors. Mach Learn. 1996;24:123–40.
27. Platt J. Probabilistic outputs for support vector machines and
comparisons to regularized likelihood methods. Adv Large Mar-
gin Classif. 1999;10:61–74.
29. Delaine T, Bernardes-Genisson V, Quemard A, Constant P,
Meunier B, et al. Development of isoniazide NAD truncated
adducts embedding a lipophilic fragment as potential bi-substrate
InhA inhibitors and antimycobacterial agents. Eur J Med Chem.
2010;45:4554–61.
30. Jones G, Willett P, Glen RC. Molecular recognition of receptor
sites using a genetic algorithm with a description of desolvation. J
Mol Biol. 1995;245:43–53.
31. Jones G, Willett P, Glen RC, Leach AR, Taylor R. Development
and validation of a genetic algorithm for flexible docking. J Mol
Biol. 1997;267:727–48.
28. Molle V, Gulten G, Vilcheze C, Veyron-Churlet R, Zanella-Cleon
I, Sacchettini JC, et al. Phosphorylation of InhA inhibits mycolic