European Journal of Medicinal Chemistry 60 (2013) 333e339

Contents lists available at SciVerse ScienceDirect

European Journal of Medicinal Chemistry

journal homepage: http://www.elsevier.com/locate/ejmech

Original article

Identification of novel antimycobacterial chemical agents through the

in silico multi-conformational structure-based drug screening of
a large-scale chemical library
Yuji Koseki a, Tomohiro Kinjo a, Maiko Kobayashi a, Shunsuke Aoki a, b, *
a
Department of Bioscience and Bioinformatics, Graduate School of Computer Science and Systems Engineering, Kyushu Institute of Technology, 680-4 Kawazu, Iizuka-shi,
Fukuoka 820-8502, Japan
b
Biomedical Informatics Research and Development Center (BMIRC), Kyushu Institute of Technology, 680-4 Kawazu, Iizuka-shi, Fukuoka 820-8502, Japan

a r t i c l e i n f o a b s t r a c t

Article history: The increasing prevalence of drug-resistant tuberculosis, which is resistant to effective multiple
Received 1 October 2012 antibiotic, presents a major global health threat. The thymidine monophosphate kinase (TMPK) of
Received in revised form Mycobacterium tuberculosis (M. tuberculosis), which is an essential enzyme for the maintenance of the
3 December 2012
thymidine triphosphate pools, is considered an attractive target for the development of effective anti-
Accepted 7 December 2012
Available online 20 December 2012
biotics against tuberculosis. In this study, we attempted to identify novel chemical compounds that
specifically target the M. tuberculosis TMPK (mtTMPK). We performed in silico structure-based drug
screening using the crystal structure data of mtTMPK and a large-scale virtual compound library, which
Keywords:
Antibiotics
is composed of 6,192,930 chemicals. Through a three-step screening method using the DOCK and GOLD,
Mycobacterium we identified ten chemical compounds that were predicted to have high binding affinity to the active site
In silico structure-based drug screening cleft of the mtTMPK. We then evaluated the antibiotic effects of these chemical compounds on model
Thymidine monophosphate kinase (TMPK) mycobacteria strains. As a result, we found that a chemical compound, K10, completely inhibited the
growth of Mycobacterium vanbaalenii (M. vanbaalenii) and Mycobacterium smegmatis (M. smegmatis).
Moreover, K10 does not exhibit any toxic effects on the growth of enterobacteria and mammalian cells.
The structural and experimental information regarding this novel chemical compound, K10, is likely to
be useful for the hit-to-lead optimization of new antibiotics for the treatment of tuberculosis.
Ó 2012 Elsevier Masson SAS. All rights reserved.

1. Introduction is often incomplete, which has resulted in increasing prevalences of

Mycobacterium tuberculosis is the causative agent of human
tuberculosis (TB), which is one of the major infectious diseases and
causes approximately 1.8 million deaths every year worldwide.
These deaths also include a large number of human immunodefi-
ciency virus (HIV) co-infected patients [1], which makes sense
because HIV infection is known to be a deadly synergistic factor for
TB [2]. The Bacille Calmette-Guérin (BCG) vaccine shows protective
effects against childhood TB. However, the protective efficacy of the
BCG vaccine is limited and incomplete against adult TB [3e5]. In
most cases, TB is currently treated with a cocktail of four first-line
drugs, which include isoniazid, rifampin, ethambutol and strepto-
mycin, for at least six months [6e8]. However, the drug compliance
* Corresponding author. Department of Bioscience and Bioinformatics, Graduate
School of Computer Science and Systems Engineering, Kyushu Institute of Tech-
nology, 680-4 Kawazu, Iizuka-shi, Fukuoka 820-8502, Japan. Tel.: þ81 948 29 7819;
fax: þ81 948 29 7801.
E-mail address: aokis@bio.kyutech.ac.jp (S. Aoki).
multidrug-resistant TB (MDR-TB), extensively drug-resistant TB
(XDR-TB) and totally drug-resistant TB (TDR-TB) in many areas
around the world [9e11]. Therefore, the discovery of new anti-TB
drugs is important for the treatment against the increasing cases
of MDR-, XDR- and TDR-TB.
The anti-TB drugs in current development target various biolog-
ical pathways, such as mycobacterial cell wall synthesis, DNA/RNA
synthesis and protein synthesis [12,13]. The M. tuberculosis thymi-
dine monophosphate kinase (mtTMPK, EC 2.7.4.9) is proposed as an
attractive target for the treatment of TB [14]. mtTMPK catalyses the
reversible phosphorylation of deoxythymidine 50 -monophosphate
(dTMP) to deoxythymidine 50 -diphosphate in the presence of Mg2þ
and adenosine triphosphate (ATP). It is a crucial enzyme in both the
de novo and the salvage pathways of DNA synthesis for bacterial
growth [15e17]. Previously, several research groups have reported
the successful development of nucleotide analogue inhibitors of
mtTMPK [18,19]. Therefore, mtTMPK is considered one of the
significant targets for the development of novel TB therapeutic
chemical compounds.

0223-5234/$ e see front matter Ó 2012 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ejmech.2012.12.012
334 Y. Koseki et al. / European Journal of Medicinal Chemistry 60 (2013) 333e339
Fig. 1. The three-dimensional structure of M. tuberculosis TMPK (mtTMPK). The
secondary structures are indicated by ribbon representations (a-helix: orange, b-sheet:
green). The amino acid residues of the active site and Mg2þ ion are shown as a stick
model and a sphere model, respectively. (For interpretation of the references to colour
in this figure legend, the reader is referred to the web version of this article.)
In this study, we used a three-step hierarchical in silico structure-
based drug screening (SBDS) to screen for chemical compounds that
are potentially able to bind to the active site of mtTMPK. After
making careful predictions of proteineligand interactions using
multi-conformational data of the chemical compounds, we selected
ten compounds that exhibited average GOLD scores over 70 from the
large-scale virtual chemical compounds library (6,192,938 chem-
icals) that was analysed. Moreover, we experimentally demonstrated
that four compounds have antibiotic effects against Mycobacterium
vanbaalenii that are similar to those of isoniazid. Among them, three
chemical compounds had strong inhibitory effects on the growth of
the model mycobacteria strains. In addition, two of the active
compounds did not exhibit any significant toxic effects on the
growth of Escherichia coli (E.coli) and the viability of cultured
Table 1
List of the ten selected chemical compounds that were identified by the in silico screen.
No. Compound name
K1 N-{2-[4-(aminosulfonyl)phenyl]ethyl}-2-(3-isopropoxyphenyl)-4-quinolinecarboxamide
K2 N,N0 -bis(4-methylphenyl)-6-[(5-methyl-1,3,4-thiadiazol-2-yl)thio]-1,3,5-triazine-2,4-diamine
K3 N-[4-(benzyloxy)-phenyl]-1-(4-nitrobenzoyl)-4-piperidinecarboxamide
K4 N-[2({2-[(4-ethoxyphenyl)amino]-2-oxoethyl}thio)-1,3-benzothiazol-6-yl]-2-methylbenzamide
K5 benzyl [(6-chloro-2-oxo-4-propyl-2H-chromen-7-yl)oxy]acetate
K6 2-(4-chloro-2-methylphenoxy)-N-[2-methoxy-5-(2-oxo-2H-chromen-3-yl)phenyl]acetamide
K7 benzyl [(6-chloro-4-ethyl-2-oxo-2H-chromen-7-yl)oxy]acetate
K8 methyl {4-[(4-tert-butylbenzoyl)amino]phenoxy}acetate
K9 ethyl [(4-butyl-6-chloro-2-oxo-2H-chromen-7-yl)-oxy]acetate
K10 8-quinolinyl 4-[(4-chlorophenyl)sulfonyl]benzoate
a
Each value represents the mean  SEM.
mammalian cells. The structural and experimental information of
the active compound K10 will contribute to the development of
anti-TB medical drugs.
2. Results
2.1. Molecular docking and scoring of putative binding affinities
We performed a three-step screening process using DOCK (first
screening) and GOLD (single- and multi-conformation screening)
to identify chemicals that potentially bind to the active site cleft of
mtTMPK. The three-dimensional structure of mtTMPK is charac-
terized by nine a-helices (H1eH9) and five b-sheets (S1eS5), which
is also called the Rossmann fold [20,21]. We focused on the
following 14 amino acid residues: Asp 9, Arg 14, Phe 36, Pro 37, Tyr
39, His 53, Phe 70, Asp 94, Arg 95, Asn 100, Tyr 103, Arg 153, Ala 161
and Asp 163 (Fig. 1) [17,18,21]. All of these residues are located in
the active site cleft of mtTMPK and comprise the phosphate binding
loop (P-loop) and phosphate donor binding site (LID region) [21].
From a constructed structural database of 6,192,930 chemicals, the
DOCK screening identified 43,932 chemicals that are potentially
capable of binding to the active site cleft with an estimated binding
free energy of less than 45 kcal/mol. This cut-off was chosen
because nucleotide analogue inhibitors of mtTMPK [18,22] have
estimated binding free energies of approximately 45 kcal/mol.
Following the DOCK screening, we performed a GOLD screening
using a single conformation per chemical and identified the top
3100 chemical compounds. During the first GOLD screening
(single-conformation GOLD screening), we initially identified the
top 1000 ranked chemical compounds with GOLD scores of more
than 70 because our previous research demonstrated that
compounds with GOLD scores of over 60 can affect enzyme activity
at concentrations lower than several hundred mM [23,24]. A
previous study has suggested that the accuracy of docking simu-
lations with multiple conformers is improved relative to the
accuracy of simulations with single conformers [25]. We then
performed a second GOLD screening using multiple conformations
per chemical structure (five structural conformations for each
compound). To remove the already evaluated chemicals, we elim-
inated several chemical compounds from the previously identified
1000 compounds according to the bioactivity assay results in the
PubChem database [26]. The elimination of chemical compounds
was based on the following three criteria: (1) the compound was
not identified through high-throughput screening, (2) the
compound did not have potentially toxic effects (i.e., cytotoxicity),
and (3) the compound had one or more direct hydrogen bond
(H-bond) interactions with the active site. Among the remaining
chemical compounds with average GOLD scores of over 70, we
selected ten chemical compounds, K1eK10 (Table 1), that form
GOLD score
Single conf. Multiple conf.a
86.33 88.17  0.95
78.63 85.98  0.21
75.58 84.56  0.28
76.28 84.23  0.11
80.23 79.85  0.16
73.00 79.14  0.49
70.91 76.00  0.33
74.31 74.79  0.18
72.67 74.53  0.45
74.00 71.92  0.47
 Y. Koseki et al. / European Journal of Medicinal Chemistry 60 (2013) 333e339 335

Fig. 2. General flowchart for the three-step in silico SBDS used to identify novel
antimycobacterial compounds targeting mtTMPK.

H-bond interactions with at least one of the target amino acids

(Fig. 2). Furthermore, the relationship between the kinetic param-
eter of mtTMPK inhibitors (Ki values ¼ 7.2e69 mM) [18,22] and the
GOLD score was verified by the in silico docking study. As a result,
the correlation coefficients (r2) for the correlations between the Ki
values and the single- and multi-conformational GOLD scores were
0.76 and 0.80, respectively (Table 2).
2.2. In vitro antimycobacterial activity assay
We then performed antimycobacterial activity assays to confirm
that the ten chemicals (K1eK10) are actually able to exhibit
inhibitory effects on the growth of two model mycobacteria:
M. vanbaalenii and Mycobacterium smegmatis. These model bacteria
were utilized because of their amino acid similarities to
M. tuberculosis [TMPK of M. vanbaalenii (96%) and M. smegmatis
(97%)] according to the NCBI protein blast search [27] and because
these strains are harmless (biosafety level 1) [28]. The inhibitory
effects of the ten candidate chemicals that were identified through
the hierarchical in silico SBDS on the growth of M. vanbaalenii were
tested. K1, K2, K7 and K10 (all at concentrations of 50 mM) signifi-
cantly inhibited the growth of M. vanbaalenii compared with the
negative control (DMSO). In addition, after 18 h, these effects were
similar to or higher than those obtained with the positive control
(isoniazid). In contrast, the six other compounds (K3eK6, K8 and
K9) had no significant or weak inhibitory effects on M. vanbaalenii
growth compared with the negative and positive controls (Fig. 3A).
Fig. 3. Inhibitory effects of the candidate chemicals (K1eK10) on the growth of the
model mycobacteria: (A) M. vanbaalenii and (B) M. smegmatis. The concentrations of all
of the chemicals were (A) 50 mM or (B) 100 mM. DMSO (0.3%) and isoniazid (50/
100 mM) were used as the negative and positive controls, respectively. Each value
represents the mean  SEM of four independent experiments. Dunnett’s multiple
comparison test was performed. The statistical significance is indicated in all of the
figures (**p < 0.01, n.s. not significant).
We also examined the effects of the four chemicals (K1, K2, K7 and
K10) on M. smegmatis growth. After 24 h, K1 and K2 (100 mM)
moderately inhibited the growth of M. smegmatis compared with
the positive control (isoniazid). Similarly to isoniazid, K10 (100 mM)
completely inhibited the growth of M. smegmatis, whereas K7
(100 mM) has no inhibitory effect on the growth of M. smegmatis
(Fig. 3B). The chemical structure and physicochemical properties of
the active hits are indicated in Fig. 4A and Table 3, respectively.
Furthermore, we investigated the dose-dependent effects of the
active hits on the growth of M. vanbaalenii to determine the IC50
values. The experimentally determined IC50 values of K1, K2 and
K10 for M. vanbaalenii were 15 mM, 22 mM and 20 mM, respectively
(Fig. 4BeD). However, these values are higher than the IC50 value of
isoniazid for M. vanbaalenii (2.6 mM; data not shown).

Table 2
Evaluation of the relationship between the GOLD score and the kinetic parameter (Ki) for nucleotide analogue inhibitors of mtTMPK and isoniazid-NAD.

Compound name Ki (mM)a GOLD score

Single conf. Multiple conf.b

N-[(30 -deoxythymidine-30 -yl)methyl]-N0 -phenylthiourea 69.0 73.81 78.78  0.33
N-[(30 -deoxythymidine-30 -yl)methyl]-N0 -(4-methylphenyl)thiourea 36.0 75.57 82.16  0.36
N-(4-chlorophenyl)-N0 -[(30 -deoxythymidine-30 -yl)methyl]thiourea 21.0 76.89 82.63  0.77
1-[3-(azidomethyl)-2,3-dideoxy-5-O-phosphoryl-b-D-erythro-pentofuranosyl]thymine 12.0 78.41 83.94  0.35
1-[3-(aminomethyl)-2,3-dideoxy-5-O-phosphoryl-b-D-erythro-pentofuranosyl]thymine 10.5 76.67 85.30  0.17
30 -azide-30 -deoxythymidine monophosphate 10.0 80.05 84.62  0.52
N-(3,4-dichlorophenyl)-N0 -[(30 -deoxythymidine-30 -yl)methyl]thiourea 7.2 78.77 84.72  0.53
Isoniazid-NAD (INH-NAD)c 7.5  104 107.43 127.10  2.99
a
The Ki values were based on previously published data [18,22,49].
b
Each value represents the mean  SEM.
c
INH-NAD inhibits enoyleacyl carrier protein reductase (mtInhA) [6,42].
336 Y. Koseki et al. / European Journal of Medicinal Chemistry 60 (2013) 333e339
Fig. 4. Chemical structural formula of the active hits (A) and dose-dependent effects of the active hits (K1, K2 and K10) on the growth of M. vanbaalenii: (B) K1, (C) K2 and (D) K10.
The bacterial growth rate (%) was determined through a 24 h cultivation of M. vanbaalenii. Each plotted value represents the mean  SD of four independent experiments.
2.3. Enterobacteria and mammalian cell toxicity assay
We performed chemical compound toxicity assays to confirm
that the chemical compounds do not have toxic effects on both
a model enterobacteria (E. coli) and mammalian, MadineDarby
Canine Kidney (MDCK) and human neuroblastoma (SH-SY5Y)
cells. The active hit chemicals identified by the antimycobacterial
assays were investigated using E. coli growth and mammalian cell
cytotoxicity assays. All of the compounds showed no toxic effect on
the growth of E. coli BL21 (data not shown) and JM 109 (Fig. 5A). K1
and K10 showed no significant toxic effects on cultured MDCK and
SH-SY5Y cells, whereas compound K2 had significant toxic effects
on these mammalian cells (Fig. 5B and C).
2.4. Predicted binding modes of active hit chemical compounds
A proteineligand interaction fingerprint (PLIF) analysis was per-
formed to better understand the putative interaction between
mtTMPK and the active chemicals. The predicted best binding modes
of the three active chemicals (K1, K2 and K10) are presented in Fig. 6.
All of the chemicals are predicted to be located near the active site
cleft of mtTMPK. The 3-isopropoxyphenyl ring in K1 forms an arenee
cation interaction with Arg 153 and a H-bond interaction with the
main chain of Ala 161. Moreover, the 4-quinolinecarboxamide in K1
forms H-bond interactions with Arg 95 and Arg 153. In addition, the
4-aminosulfonyl ring in K1 forms a hydrophobic interaction with Phe
70 and a H-bond interaction with Asn 100 (Fig. 6A). The 1,3,5-triazine
and the (5-methyl-1,3,4-thiadiazol-2-yl)thio rings in K2 form
Table 3
Physicochemical properties of the active hit compounds.
Chemical Molecular log P H-bond H-bond
name weight acceptor donor
K1 490 4.71 5 2
K2 422 5.68 5 2
K10 424 5.12 5 0
a
Topological molecular polar surface area.
H-bond interactions with Asp 9, Arg 14, Asp 94 and Arg 95. Moreover,
the 4-methylphenyl ring in K2 forms an areneecation interaction
with Arg 153 (Fig. 6B). The sulfonyl and benzoate in K10 form H-bond
interactions with His 53 and Arg 95. The 4-chlorophenyl and
8-quinolinyl rings in K10 additionally form areneearene interactions
with Tyr 39 and Phe 70, respectively (Fig. 6C).
3. Discussion
Using a three-step in silico SBDS (DOCK, single-conformation
GOLD and multi-conformation GOLD) of 6,192,930 chemical
compounds, we identified ten candidate compounds that could
potentially inhibit mtTMPK activity. We then evaluated the actual
inhibitory effects of the compounds on the growth of model
mycobacteria and identified four active compounds (compound K1,
K2, K7 and K10) that were able to inhibit M. vanbaalenii growth.
Two chemicals (K1 and K2) also exhibited moderate inhibitory
effects on the growth of M. smegmatis. Similarly to isoniazid, K10
completely inhibited the growth of both M. vanbaalenii and
M. smegmatis and exhibited no significant toxic effects on E. coli,
MDCK and SH-SY5Y cells. A PLIF analysis of these active chemicals
showed that all of the chemicals highlighted a key H-bond inter-
action with Arg 95, which is found in the active site of mtTMPK [21].
Recently, a study that used live-cell imaging with a microfluidic
device to culture M. smegmatis demonstrated that the cell division
of M. smegmatis is similar to that of M. tuberculosis, which suggests
that M. smegmatis is an appropriate model mycobacteria and will
contribute to the development of antibiotics for TB [29]. These
results suggested that the structural information of K10 is useful in
the research and development of novel TB therapeutic chemicals.
Several nucleotide analogue inhibitors for mtTMPK have been
developed [18,19]. However, these chemicals have low permeability
tPSAa to the mycobacterial cell wall and are cleaved by phosphatase or
phosphorylase [19,30]. In contrast, the chemical K10 is an artificially
111.4 synthesised chemical that is composed of hydrophobic chemical
88.5 groups, 8-quinolinyl, (4-chlorophenly)sulfonyl and benzoate. Thus,
73.3 K10 is expected to have a higher permeability and stability than
the nucleotide analogues. According to the structureeactivity
 Y. Koseki et al. / European Journal of Medicinal Chemistry 60 (2013) 333e339 337

Fig. 5. Toxic effects of the chemicals (K1, K2 and K10) on the model enterobacteria and mammalian cells: (A) E. coli JM 109, (B) MDCK cells and (C) SHeSY5Y cells. The concentration
of all of the chemicals was 50 mM (IC90 value for mycobacteria growth). DMSO (0.3%) and isoniazid (50 mM) were used as the negative and positive controls, respectively. Each value
represents the mean  SEM of four independent experiments. Dunnett’s multiple comparison test was performed (**p < 0.01, n.s. not significant).

Fig. 6. Putative intermolecular interactions between mtTMPK and the hit chemicals by PLIF analysis: (A) K1, (B) K2 and (C) K10. The H-bond, areneearene and areneecation
interactions are indicated by green, blue dotted and blue dashed lines, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)

relationships (SAR), it is hypothesized that an areneearene
interaction with Tyr 39, a H-bond interaction with His 53 and an
areneearene/hydrophobic interaction with Phe 70 are important for
the antibiotic effects (Fig. 6). The targeted amino acid residues are
consistent with a previous structural analysis of the complexes of
inhibitors with mtTMPK [17]. Thus, the structural information of the
chemical compounds identified in this study is likely to be useful in
the design of new antibiotics against TB.
Many medical drugs have been developed through the typical
processes, such as the identification of active chemicals using a HTS
method and the optimization of their physicochemical properties
through the synthesis of related analogues. However, these
processes are costly and require much time [31]. In contrast,
computer-aided drug design, which is based on molecular docking,
pharmacophore modelling, three-dimensional quantitative
structure-activity (3D-QSAR) [32] and molecular dynamics (MD)
[33], is a shortcut technique that enables a more rapid hit identi-
fication than HTS [34]. Several research groups have suggested that
a multi-step in silico SBDS is very effective and exhibits reasonable
accuracy [35,36]. In addition, the docking of multiple conforma-
tions of chemicals into an active site demonstrated that the docking
accuracy was improved compared to the docking of a single
conformation per chemical [25]. Previously, two novel anti-TB
agents were identified through in silico SBDS using a single
conformation of the chemicals (2.5% hit rate) [37]. In this study, we
developed a novel hierarchical multistep SBDS method, performed
biological experiments with ten candidate chemicals that were
predicted by the three-step in silico SBDS and successfully identi-
fied three novel potential anti-mycobacteria chemicals (30% hit
rate). Thus, our screening methods could contribute to the further
identification of hit compounds for the development of therapeutic
chemicals against not only TB but also other diseases.
4. Conclusion
In the present study, we used a three-step in silico SBDS
approach with a large-scale chemical library to identify ten
chemicals that were predicted to have a high binding affinity to
mtTMPK. The results of the in vitro biological assays indicated
that one of these chemical compounds, K10, exhibits a potent
antimycobacterial effect and no toxicity. Thus, the structural and
predicted proteineligand interaction information for this novel
chemical compound, K10, will be helpful for further hit-to-lead
optimization of new antibiotics against drug-resistant TB.
5. Experimental protocols
5.1. Chemical compounds library
For the in silico SBDS, we used a chemical compounds library
(6,192,930 commercially available compounds) that was derived
from the web-based database ZINC [38]. This compounds library is
338 Y. Koseki et al. / European Journal of Medicinal Chemistry 60 (2013) 333e339
ADME/Tox-Drug-Like filtered (Absorption, Distribution, Metabo-
lism, Excretion and Toxicity) [39]. The three-dimensional structures
of mtTMPK inhibitors [18,22] were generated using the Builder
module in the Molecular Operating Environment (MOE) version
2010.10 [40]. The multi-conformation structural data of the
compounds was generated using the LowMode MD module in the
MOE 2010.10 with the default search parameters: rejection
limit ¼ 100, iteration limit ¼ 10,000, RMS gradient ¼ 0.005, MM
iteration limit ¼ 500, RMSD limit ¼ 0.25 and conformation
limit ¼ 10,000 [40,41].
5.2. Preparation of protein structure
The crystal structure data for mtTMPK (PDB ID: 1MRN) [17]
and mtInhA (PDB ID: 2IDZ) [42] were obtained from the Research
Collaboratory for Structural Bioinformatics Protein Data Bank [43].
These two structure files were based on mtTMPK and mtInhA
complexed with inhibitors, P1-(adenosine-50 )-P5-(thymidine-50 )-
pentaphosphate (Ap5T) and INH-NAD, respectively. Before using
the mtTMPK and mtInhA crystal structures in the in silico SBDS, all
of the crystal waters and the inhibitor (Ap5T, INH-NAD) were
removed. In addition, hydrogen atoms were added to mtTMPK and
mtInhA using the MMFF94x force field, and the energy was mini-
mized using the Protonate 3D and Energy Minimize modules of
MOE 2010.10. As part this process, all of the heavy atoms, except the
hydrogen atoms, were tethered to relieve any short contacts [40].
5.3. Hierarchical in silico structure-based drug screening
The hierarchical in silico SBDS was performed using UCSF DOCK
version 6.4 [44] and CCDC GOLD suite version 5.0.1 [45]. The three-
step in silico SBDS was conducted using the energy-minimized
structure. The proteineligand interaction space was restricted to
the active site cleft of mtTMPK. For the first screening with DOCK,
we constructed the molecular surface and generated the interac-
tion cleft using the DMS and SPHGEN programs, respectively [44].
In the rigid ligand-based DOCK screening, the grid-based energy
scoring function, which is composed of van der Waals and elec-
trostatic interaction energies, was applied to estimate the potential
binding affinities [44]. Following the screening with rigid condi-
tions in DOCK, the protein-chemical binding affinities were pre-
dicted with the GOLD program, which uses a genetic algorithm
(GA) method for the docking and conformation search [45,46]. The
docking studies were performed using the default evolutionary
parameters for GOLD (single- and multi-conformation screening):
operations ¼ 100,000, island ¼ 5, migration ¼ 15, mutation ¼ 95
and crossover ¼ 95. The ten docking poses for each chemical
compound were saved for use in the proteineligand interaction
analyses. The analysis of the in silico SBDS results was performed
using the proteineligand interaction fingerprint (PLIF) and Ligand
Interactions module in MOE 2010.10 [40]. All of the calculations
were performed on clusters of personal computers running the
Linux operating system.
5.4. Chemical compounds
All of the chemical compounds were purchased from the
ChemBridge Corporation (San Diego, CA) [47] and dissolved in
dimethyl sulfoxide (DMSO, SIGMA) for the in vitro assays. The
chemical compounds (ZINC IDs given in parentheses) used were
the following:
K1: N-{2-[4-(aminosulfonyl)phenyl]ethyl}-2-(3-isopropoxyphenyl)-
4-quinolinecarboxamide (02867600);
K2: N,N0 -bis(4-methylphenyl)-6-[(5-methyl-1,3,4-thiadiazol-2-yl)
thio]-1,3,5-triazine-2,4-diamine (00708281);
K3: N-[4-(benzyloxy)phenyl]-1-(4-nitrobenzoyl)-4-piperidine
carboxamide (16267193);
K4: N-[2-({2-[(4-ethoxyphenyl)amino]-2-oxoethyl}thio)-1,3-
benzothiazol-6-yl] -2-methylbenzamide (00680597);
K5: benzyl [(6-chloro-2-oxo-4-propyl-2H-chromen-7-yl)oxy]
acetate (01210682);
K6: 2-(4-chloro-2-methylphenoxy)-N-[2-methoxy-5-(2-oxo-
2H-chromen-3-yl)phenyl] acetamide (01065313);
K7: benzyl [(6-chloro-4-ethyl-2-oxo-2H-chromen-7-yl)oxy]
acetate (01142847);
K8: methyl {4-[(4-tert-butylbenzoyl)amino]phenoxy}acetate
(00619622);
K9: ethyl [(4-butyl-6-chroro-2-oxo-2H-chromen-7-yl)oxy]
acetate (02479771); and.
K10: 8-quinolinyl 4-[(4-chlorophenyl)sulfonyl]benzoate
(01138659).
5.5. Bacterial strains
M. vanbaalenii and M. smegmatis were obtained from the RIKEN
BioResource Center (Saitama, Japan) [28]. E. coli JM 109 and BL 21
were kindly gifted from Dr. S. Sueda (Kyushu Institute of Technology).
5.6. Bacterial growth assay
The bacteria were cultured using modified methods that were
previously described [48]. Briefly, M. vanbaalenii and M. smegmatis
were grown overnight or for 2 days at 300 rpm and 37  C in 3 ml of
culture medium [3.7% Brain heart infusion broth (Difco)]. E. coli was
grown overnight at 300 rpm and 37  C in 3 ml of liquid culture [1%
Tryptone (BD), 0.5% Yeast extract (BD) and 0.5% NaCl (Wako)]. We
used a 6 x dilution of the M. vanbaalenii culture, an 8 x dilution of
the M. smegmatis culture and a 10 x dilution of the E. coli culture for
the respective growth assays. In the assays, the diluted bacteria
were seeded in 96-well assay plates (NUNC) and incubated in
culture medium containing one of the chemicals. In the
M. vanbaalenii and M. smegmatis growth assays, 0.3% DMSO was
used as the negative control and isoniazid (LKT) was used as the
positive control. Ampicillin (SIGMA) was used as the positive
control in the E. coli growth assay. After 4, 8, 18 and 24 h, we
measured the absorbance of the culture media at 595 nm with
a micro plate reader (BioRad).
5.7. Mammalian cell toxicity assay
The cytotoxic effect of the chemical compounds on mammalian
cells was assayed using the Cell Counting Kit-8 (DOJIN), which was
used to measure the number of living cells. The MDCK and SH-SY5Y
cells were seeded in 100-mm dishes and maintained in DMEM
medium (Wako) supplemented with 10% FBS (GIBCO), 100 units/ml
penicillin, 100 mg/ml streptomycin (GIBCO) and 2 mM L-Glutamine
(GIBCO). The MDCK and SH-SY5Y cells were seeded in 96-well
assay plates (CORNING) at 5.0  103 and 1.5  104 cells/well,
respectively, and were incubated 6 h or overnight at 37  C in 5%
CO2. For cell starvation, the culture medium of the MDCK and
SH-SY5Y cells was replaced by fresh DMEM medium containing
0.25% FBS, and the cells were then incubated overnight and for 2
days, respectively. After cell starvation, the culture medium of the
MDCK and SH-SY5Y cells was replaced by fresh medium (0.25%
FBS) containing one of the chemical compounds or isoniazid
(50 mM) as the negative control. The MDCK and SH-SY5Y cells were
then incubated for 1 day and 2 days, respectively. We then added
10 ml/well of Cell Counting Kit-8, and, after 3 h incubation, we
measured the absorbance of WST-8 formazan at 450 nm with
a micro plate reader (BioRad).
 Y. Koseki et al. / European Journal of Medicinal Chemistry 60 (2013) 333e339
5.8. Statistical analysis
All of the statistical analyses were performed using R version
2.15.0 (The R Foundation for Statistical Computing, Vienna,
Austria) and GraphPad Prism version 4 (GraphPad Software, Inc.,
San Diego, CA).
Acknowledgements
The authors are very thankful to Dr. S. Fujii (Kyushu Institute of
Technology) for the valuable comments and Mr. K. Tsuruta (Kyushu
Institute of Technology) for the excellent technical support.
References
[1] World Health Organization, Global Tuberculosis Control: WHO Report 2011,
ISBN 978 92 4 156438 0.
[2] K.H. Mayer, C.D. Hamilton, Synergistic pandemics: confronting the global HIV
and tuberculosis epidemics, Clin. Infect. Dis. 50 (2010) S67eS70.
[3] G.A. Colditz, T.F. Brewer, C.S. Berkey, M.E. Wilson, E. Burdick, H.V. Fineberg,
F. Mosteller, Efficacy of BCG vaccine in the prevention of tuberculosis meta-
analysis of the published literature, JAMA 271 (1994) 698e702.
[4] S.H.E. Kaufmann, A.J. McMichael, Annulling a dangerous liaison: vaccination
strategies against AIDS and tuberculosis, Nat. Med. 11 (2005) S33eS44.
[5] T.H.M. Ottenhoff, S.H.E. Kaufmann, Vaccines against tuberculosis: where are
we and where do we need to go? PLoS Pathog. 8 (2012) 1e12.
[6] B.R. Bloom, C.J.L. Murray, Tuberculosis: commentary on a reemergent killer,
Science 257 (1992) 1055e1064.
[7] A.E. Belanger, G.S. Besra, M.E. Ford, K. Mikusova, J.T. Belisle, P.J. Brennan,
J.M. Inamine, The embAB genes of Mycobacterium avium encode an arabinosyl
transferase involved in cell wall arabinan biosynthesis that is the target for the
antimycobacterial drug ethambutol, Proc. Natl. Acad. Sci.U. S. A. 93 (1996)
11919e11924.
[8] A.P. Carter, W.M. Clemons, D.E. Brodersen, R.J. Morgan-Warren, B.T. Wimberly,
V. Ramakrishnan, Functional insights from the structure of the 30S ribosomal
subunit and its interactions with antibiotics, Nature 407 (2000) 340e348.
[9] C. Dye, Global epidemiology of tuberculosis, Lancet 367 (2006) 938e940.
[10] A.A. Velayati, M.R. Masjedi, P. Farnia, P. Tabarsi, J. Ghanavi, A.H. ZiaZarifi,
S.E. Hoffner, Emergence of new forms of totally drug-resistant tuberculosis
Bacilli Super extensively drug-resistant tuberculosis or totally drug-resistant
strains in Iran, Chest 136 (2009) 420e425.
[11] N.R. Gandhi, P. Nunn, K. Dheda, H.S. Schaaf, M. Zignol, D. van Soolingen,
P. Jensen, J. Bayona, Multidrug-resistant and extensively drug-resistant
tuberculosis: a threat to global control of tuberculosis, Lancet 375 (2010)
1830e1843.
[12] Z. Ying, The magic bullets and tuberculosis drug targets, Annu. Rev. Phar-
macol. Toxicol 45 (2005) 529e564.
[13] B. Villemagne, C. Crauste, M. Flipo, A.R. Baulard, B. Deprez, N. Willand,
Tuberculosis: the drug development pipeline at a glance, Eur. J. Med. Chem. 51
(2012) 1e16.
[14] S. Pochet, L. Dugue, G. Labesse, M. Delepierre, H. Munier-Lehmann, Compar-
ative study of purine and pyrimidine nucleoside analogues acting on the
thymidylate kinases of Mycobacterium tuberculosis and of humans, Chem-
biochem 4 (2003) 742e747.
[15] M.A. Moors, D.A. Portnoy, Identification of bacterial genes that contribute to
survival and growth in an intracellular environment, Trends Microbiol. (1995)
83e85.
[16] H. Munier-Lehmann, A. Chaffotte, S. Pochet, G. Labesse, Thymidylate kinase of
Mycobacterium tuberculosis: a chimera sharing properties common to
eukaryotic and bacterial enzymes, Protein Sci. 10 (2001) 1195e1205.
[17] A. Haouz, V. Vanheusden, H. Munier-Lehmann, M. Froeyen, P. Herdewijn, S. Van
Calenbergh, M. Delarue, Enzymatic and structural analysis of inhibitors designed
against Mycobacterium tuberculosis thymidylate kinase - new insights into the
phosphoryl transfer mechanism, J. Biol. Chem. 278 (2003) 4963e4971.
[18] I. Van Daele, H. Munier-Lehmann, M. Froeyen, J. Balzarini, S. Van Calenbergh,
Rational design of 5’-thiourea-substituted alpha-thymidine analogues as
thymidine monophosphate kinase inhibitors capable of inhibiting mycobac-
terial growth, J. Med. Chem. 50 (2007) 5281e5292.
[19] O. Familiar, H. Munier-Lehmann, J. Antonio Ainsa, M.-J. Camarasa, M.-J. Perez-
Perez, Design, synthesis and inhibitory activity against Mycobacterium tuber-
culosis thymidine monophosphate kinase of acyclic nucleoside analogues with
a distal imidazoquinolinone, Eur. J. Med. Chem. 45 (2010) 5910e5918.
[20] A. Lavie, I.R. Vetter, M. Konrad, R.S. Goody, J. Reinstein, I. Schlichting, Structure
of thymidylate kinase reveals the cause behind the limiting step in AZT
activation, Nat. Struct. Mol. Biol. 4 (1997) 601e604.
339
[21] I.L. de la Sierra, H. Munier-Lehmann, A.M. Gilles, O. Barzu, M. Delarue, X-ray
structure of TMP kinase from Mycobacterium tuberculosis complexed with
TMP at 1.95 angstrom resolution, J. Mol. Biol. 311 (2001) 87e100.
[22] V. Vanheusden, H. Munier-Lehmann, M. Froeyen, L. Dugue, A. Heyerick, D. De
Keukeleire, S. Pochet, R. Busson, P. Herdewijn, S. Van Calenbergh, 3 ’-C-
branched-chain-substituted nucleosides and nucleotides as potent inhibitors
of Mycobacterium tuberculosis thymidine monophosphate kinase, J. Med.
Chem. 46 (2003) 3811e3821.
[23] K. Hirayama, S. Aoki, K. Nishikawa, T. Matsumoto, K. Wada, Identification of
novel chemical inhibitors for ubiquitin C-terminal hydrolase-L3 by virtual
screening, Bioorg. Med. Chem. 15 (2007) 6810e6818.
[24] T. Mitsui, K. Hirayama, S. Aoki, K. Nishikawa, K. Uchida, T. Matsumoto, T. Kabuta,
K. Wada, Identification of a novel chemical potentiator and inhibitors of UCH-L1
by in silico drug screening, Neurochem. Int. 56 (2010) 679e686.
[25] N. Sauton, D. Lagorce, B.O. Villoutreix, M.A. Miteva, MS-DOCK: accurate
multiple conformation generator and rigid docking protocol for multi-step
virtual ligand screening, BMC Bioinf. 9 (2008) 184.
[26] Q. Li, T. Chen, Y. Wang, S.H. Bryant, PubChem as a public resource for drug
discovery, Drug Discovery Today 15 (2010) 1052e1057.
[27] M. Johnson, I. Zaretskaya, Y. Raytselis, Y. Merezhuk, S. McGinnis, T.L. Madden,
NCBIBLAST: a better web interface, Nucleic Acids Res. 36 (2008) W5eW9.
[28] RIKEN, RIKEN BioResource Center, Japan, <http://www.jcm.riken.go.jp>.
[29] B.B. Aldridge, M. Fernandez-Suarez, D. Heller, V. Ambravaneswaran, D. Irimia,
M. Toner, S.M. Fortune, Asymmetry and aging of mycobacterial cells lead to
variable growth and antibiotic susceptibility, Science 335 (2012) 100e104.
[30] M.J. Perez-Perez, E.M. Priego, A.I. Hernandez, M.J. Camarasa, J. Balzarini,
S. Liekens, Thymidine phosphorylase inhibitors: recent developments and
potential therapeutic applications, Mini Rev. Med. Chem. 5 (2005) 1113e1123.
[31] G.M. Keseru, G.M. Makara, Hit discovery and hit-to-lead approaches, Drug
Discovery Today 11 (2006) 831e839.
[32] X.-Y. Lu, Y.-D. Chen, Y.-J. Jiang, Q.-D. You, Discovery of potential new InhA
direct inhibitors based on pharmacophore and 3D-QSAR analysis followed by
in silico screening, Eur. J. Med. Chem. 44 (2009) 3718e3730.
[33] Y. Koseki, T. Kinjo, M. Kuroki, S. Aoki, Aberrant structures of Parkinson’s
disease-associated ubiquitin C-terminal hydrolase L1 predicted by molecular
dynamics, Chem. Phys. Lett. 535 (2012) 163e168.
[34] M. Miteva, Hierarchical structure-based virtual screening for drug design,
Biotechnol. Biotechnol. Equip. 22 (2008) 634e638.
[35] T. Pencheva, O.S. Soumana, I. Pajeva, M.A. Miteval, Post-docking virtual
screening of diverse binding pockets: comparative study using DOCK,
AMMOS, X-Score and FRED scoring functions, Eur. J. Med. Chem. 45 (2010)
2622e2628.
[36] S. Kalyaanamoorthy, Y.-P.P. Chen, Structure-based drug design to augment hit
discovery, Drug Discovery Today 16 (2011) 741e748.
[37] M. Muddassar, J.W. Jang, H.S. Gon, Y.S. Cho, E.E. Kim, K.C. Keum, T. Oh,
S.-N. Cho, A.N. Pae, Identification of novel antitubercular compounds through
hybrid virtual screening approach, Bioorg. Med. Chem. 18 (2010) 6914e6921.
[38] J.J. Irwin, B.K. Shoichet, ZINC e a free database of commercially available
compounds for virtual screening, J. Chem. Inf. Model. 45 (2005) 177e182.
[39] C.A. Lipinski, Drug-like properties and the causes of poor solubility and poor
permeability, J. Pharmacol. Toxicological Methods 44 (2000) 235e249.
[40] Molecular Operating Environment (MOE), 2010. 10, Chemical Computing
Group Inc, 1010 Sherbooke St. West, Suite #910, Montreal, QC, Canada, H3A
2R7, 2010.
[41] P. Labute, LowModeMD-implicit low-mode velocity filtering applied to
conformational search of macrocycles and protein loops, J. Chem. Inf. Model.
50 (2010) 792e800.
[42] M.V.B. Dias, I.B. Vasconcelos, A.M.X. Prado, V. Fadel, L.A. Basso, W.F. De
Azevedo Jr., D.S. Santos, Crystallographic studies on the binding of isonicotinyl-
NAD adduct to wild-type and isoniazid resistant 2-trans-enoyl-ACP (CoA)
reductase from Mycobacterium tuberculosis, J. Struct. Biol. 159 (2007) 369e380.
[43] H.M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T.N. Bhat, H. Weissig,
I.N. Shindyalov, P.E. Bourne, The protein data bank, Nucl. Acids Res. 28 (2000)
235e242.
[44] P.T. Lang, S.R. Brozell, S. Mukherjee, E.F. Pettersen, E.C. Meng, V. Thomas,
R.C. Rizzo, D.A. Case, T.L. James, I.D. Kuntz, DOCK 6: combining techniques to
model RNA-small molecule complexes, RNA 15 (2009) 1219e1230.
[45] G. Jones, P. Willett, R.C. Glen, A.R. Leach, R. Taylor, Development and validation
of a genetic algorithm for flexible docking, J. Mol. Biol. 267 (1997) 727e748.
[46] G. Jones, P. Willett, R.C. Glen, Molecular recognition of receptor site using
a genetic algorithm with description of desolvation, J. Mol. Biol. 245 (1995)
43e53.
[47] ChemBridge, San Deigo, CA, <http://www.chembridge.com>.
[48] Y. Izumizono, S. Arevalo, Y. Koseki, M. Kuroki, S. Aoki, Identification of novel
potential antibiotics for tuberculosis by in silico structure-based drug
screening, Eur. J. Med. Chem. 46 (2011) 1849e1856.
[49] R. Rawat, A. Whitty, P.J. Tonge, The isoniazid-NAD adduct is a slow, tight-
binding inhibitor of InhA, the Mycobacterium tuberculosis enoyl reductase:
adduct affinity and drug resistance, Proc. Natl. Acad. Sci.U. S. A. 100 (2003)
13881e13886.