Design and Development of Mycobacterium tuberculosis Lysine ε-Aminotransferase Inhibitors for Latent Tuberculosis Infection

Article abstract of DOI:10.1111/cbdd.12655

Lysine ε-aminotransferase (LAT) is a protein involved in lysine catabolism, and it plays a significant role during the persistent/latent phase of Mycobacterium tuberculosis (MTB), as observed by its up-regulation by ~40-fold during this stage. We have used the crystal structure of MTB LAT in external aldimine form in complex with its substrate lysine as a template to design and identify seven lead compounds with IC₅₀ ranging from 18.06 to > 90 μm. We have synthesized 21 compounds based on the identified lead, and compound 21 [2,2′-oxybis(N′-(4-fluorobenzylidene)acetohydrazide)] was found to be the most active with MTB LAT IC₅₀ of 0.81 ± 0.03 μm. Compound 21 also showed a 2.3 log reduction in the nutrient-starved MTB model and was more potent than standard isoniazid and rifampicin at the same dose level of 10 μg/mL.

Full text of DOI:10.1111/cbdd.12655

Chem Biol Drug Des 2016; 87: 265–274
Research Letter
Design and Development of Mycobacterium
tuberculosis Lysine ɛ-Aminotransferase Inhibitors for
Latent Tuberculosis Infection
Brindha Devi Parthiban, Shalini Saxena,
from human immunodeficiency virus infection, malnutrition,
administration of chemotherapy or steroids and antitumor
necrosis factor treatment predispose to developing TB dis-
ease in people latently infected with MTB (1). The lifetime
risk of developing active TB disease after primary infection
is often said to be about 10%, and it is stated that approx-
imately one half of those who develop the disease will do
so within 5 years after infection. Treatment of latent TB
infection seems to be essential for controlling and eliminat-
ing TB all over the world. Successful therapy of latent TB
reduces the risk of activation of dormant infection and
subsequent progression to active TB disease. The cur-
rently recommended treatment of latent TB is based on
the monotherapy with isoniazid (INH); however, the period
Manoj Chandran, Padma Sridevi Jonnalagadda,
Renu Yadav, Rudraraju Reshma Srilakshmi,
Yogeeswari Perumal and Sriram Dharmarajan*
Centre for Infectious Disease Research, Department of
Pharmacy, Birla Institute of Technology & Science-Pilani,
Hyderabad Campus, Hyderabad 500078, India
*Corresponding author: Sriram Dharmarajan,
dsriram@hyderabad.bits-pilani.ac.in
Lysine ɛ-aminotransferase (LAT) is a protein involved in
lysine catabolism, and it plays a significant role during
the persistent/latent phase of Mycobacterium tuber-
culosis (MTB), as observed by its up-regulation by
a
of treatment can vary from 6 to 9 months . The combined
therapy with INH and RMP is another option for combating
the LTBI. A series of clinical trials have demonstrated that
~
40-fold during this stage. We have used the crystal
structure of MTB LAT in external aldimine form in com-
plex with its substrate lysine as a template to design
and identify seven lead compounds with IC50 ranging
from 18.06 to > 90 lM. We have synthesized 21 com-
pounds based on the identified lead, and compound
3
- or 4-month INH + RMP treatment is equivalent to that
of 6- or 9-month INH usage, although adverse effects are
significantly more frequent (2).
0
0
2
1 [2,2 -oxybis(N -(4-fluorobenzylidene)acetohydrazide)]
was found to be the most active with MTB LAT IC50 of
.81 ± 0.03 lM. Compound 21 also showed a 2.3 log
The common belief is that in latent TB, due to the ‘potent’
immune response generated by the host, the bacillus
stops growing and enters a state of stationary phase,
eventually becoming non-replicating, while retaining the
ability to resume growth under favorable circumstances
within the granuloma (3). These MTB are often referred to
as dormant, persistent, and non-replicating (4). The persis-
tent bacilli are thought to encounter depletion of nutrients,
shifts in pH, production of growth-limiting products, and
reduced oxygen or increased nitric oxide within the granu-
loma; this has led to the development of various in vitro
models of latency (5). Bacterial culture models are subject
to various stress conditions like hypoxia (6), nutritional star-
vation (7), acidic pH (8), and reactive oxidative/nitrosative
species (9), which could mimic some of the microenviron-
ments encountered by the bacillus in vivo. MTB subject to
these culture conditions (stress) induce a distinct set of
gene expression profiles, different from normal optimum
growth conditions. In Bettes et al. nutrient starvation
model, one of the enzyme lysine ɛ-aminotransferase (LAT)
overexpressed by ~40-fold and makes attractive target for
0
reduction in the nutrient-starved MTB model and was
more potent than standard isoniazid and rifampicin at
the same dose level of 10 lg/mL.
Key words: antimycobacterial, lysine ɛ-aminotransferase,
Mycobacterium tuberculosis, Tuberculosis
Abbreviations: LAT, lysine ɛ-aminotransferase; MTB,
Mycobacterium tuberculosis; TB, tuberculosis.
Received 19 March 2015, revised 13 August 2015 and
accepted for publication 27 August 2015
One-third of the earth’s population is infected with
Mycobacterium tuberculosis (MTB), but only 5–10% of the
infected individuals develop active tuberculosis (TB) over
their lifetime. The remaining 90–95% stays healthy and are
called latently infected (LTBI) individuals. It is estimated
that approximately one-third of the world’s population is
infected with LTBI. LTBI may develop into active TB dis-
ease within a few years or decades after the primary infec-
tion with MTB. Defects in cell-mediated immunity resulting
LTBI. LAT is a vitamin B -dependent enzyme and converts
6
l-lysine to a-aminoadipate-d-semialdehyde, which is sub-
1
sequently cyclized to form D -piperideine-6-carboxylic
ª 2015 John Wiley & Sons A/S. doi: 10.1111/cbdd.12655
265

Parthiban et al.
acid. Crystal structure of MTB LAT is available with its
substrate lysine (10), and we have designed novel inhibi-
tors using e-pharmacophore approach. One of the most
active lead was derivatised to study SAR and its activity
against nutrient-starved MTB model.
erated and further used for virtual screening of a commer-
cial database Asinex containing five lakh molecules and
our in-house database of 3000 molecules, with must
match five features of six-point pharmacophore features.
Compounds retrieved by e-pharmacophore filter using
Phase with a fit value above 1.0 were regarded as poten-
tial hits and were carried forward for high-throughput vir-
tual screening. Compounds resulting with a score of
≥ ꢀ4.0 kcal/mol and with two or more hydrogen bonds
were subjected to another round of docking by Glide XP.
Glide XP scores combine accurately physics-based scor-
ing terms and thorough sampling and result in compounds
with docking scores between ꢀ7.649 and ꢀ6.222 kcal/
mol. The final short listing of the molecules was based on
the protein–ligand interaction in the active site cleft through
hydrogen bonding with Arg170, Gln274, Ala129, Gly128,
and p–p stacking interactions with Phe167. The 2D repre-
sentation of top hits identified from Asinex database has
been represented in Figure 2. Finally, top seven hits were
selected from the Asinex and our in-house database and
were evaluated biologically against MTB LAT enzyme inhi-
bition studies at various concentrations from 100 to 10 lM.
In this study, our effort is to identify novel small molecule
inhibitors that bind to the active site; this is the place
where the substrate binds and reaction occurs of MTB
LAT protein started with e-pharmacophore approach,
which combined both the aspects of structure-based and
ligand-based techniques using the crystal structure of LAT
from MTB in external aldimine form in complex with its
ꢀ
substrate lysine (PDB code: 2CJD) with 2.00 A resolution.
Pharmacophore hypotheses based on mapping of the
energetic terms from the extra precision (XP) Glide scoring
function onto atom centers were obtained using PHASE
(
PHASE, v3.3; Schr o€ dinger, LLC, New York, NY, USA) with
the default set of six chemical features: hydrogen bond
acceptor (A), hydrogen bond donor (D), hydrophobic (H),
negative ionizable (N), positive ionizable (P), and aromatic
ring (R). The Glide XP scoring terms were computed, and
the energies were mapped onto atoms. The pharma-
cophore sites were generated, and the Glide XP energies
from the atoms that comprised each pharmacophore sites
were summed up. These sites were then ranked based on
the individual energies, and the most favorable sites were
selected for the pharmacophore hypothesis. This pharma-
cophore model was then used as query for virtual screen-
ing. A six-point e-pharmacophore model was generated
with MTB LAT protein domain. The pharmacophoric sites
established included three hydrogen bond acceptor (A1,
A3, and A4), one ring aromatic (R14), and two negative
ionizable group (N9 and N11) (Figure 1), which were gen-
Among the seven compounds tested against MTB LAT,
four molecules showed IC50 of ≤ 25 lM. The most promis-
ing compounds from the in vitro MTB LAT assay, Lead 1
having IC50 of 18.06 lM, were then evaluated in more
detail to provide a structural basis for the further optimiza-
tion by chemical synthesis. In silico molecular docking
studies were carried out with the crystal structure of MTB
LAT in external aldimine form in complex with its substrate
L-lysine with bound PMP [PDB ID: 2CJD]. The reference
ligand (L-lysine with bound PMP) was re-docked with the
active site residues of the MTB LAT to validate the active
site cavity. The ligand exhibited
a Glide score of
ꢀ
5.81 kcal/mol and was found in the vicinity of Arg170,
Glu243, Arg422, Val63, Gly169, Gly127, Gly128, Ala129,
Val132, Phe167, Asn271, Val273, Glu238, Gly242, and
Gln274 amino acid residues. Re-docking results showed
that the compound exhibited similar interactions as that of
ꢀ
the original crystal structure with RMSD of 0.97 A sug-
gesting reliability of the docking method (Figure 3).
The docking orientation of the Lead 1, when compared
with the crystal ligand, the lead compound, revealed
important hydrogen bonding interaction with Arg170 amino
acid residue. The docking score of lead was found to be
ꢀ
7.06 kcal/mol, and a strong metal co-ordination bond
with the oxygen atom of Ash271 was observed as shown
in Figure 4. In addition to hydrogen bonding interactions,
the phenyl ring interacted via p–p interaction with phenyl
ring of Phe167 amino acid residue, analogues to the one
observed with the crystal ligand L-lysine. The orientation of
lead compound was found to be closely similar to that of
crystal ligand exhibiting a very good fitness in the active
site pocket, making this compound the most active
(IC50 = 18.06 lM) as shown in Table 1.
Figure 1: The energy based pharmacophoric features of the
reference ligand bound to the protein.
2
66
Chem Biol Drug Des 2016; 87: 265–274

Mycobacterium tuberculosis Lysine ɛ-Aminotransferase Inhibitors
Figure 2: Structures of identified hits.
Figure 3: (A) Interactions of reference inhibitor with the active site residues of MtbLAT protein; (B) Superimposition of docked pose of the
reference inhibitor (Green) to the original pose of the reference inhibitor (Magenta).
Further to investigate the potential of this Lead 1, a library
was designed with the goal of obtaining a lead series with
tractable SAR and potencies better than the identified ini-
tial screening hits. A library of 21 compounds has been
synthesized using various substituted phenyl group on
both ends and also replacing central oxygen atom with
methylene group as summarized in Table 1. The designed
target molecules were synthesized via three steps in a
classical way starting with commercially available and less
expensive diglycolic acid as shown in Figure 5. In the first
step, the diglycolic acid (1) was converted into its methyl
ester (2) by following the conventional sulfuric acid–metha-
nol method well established in literature. The obtained
methyl ester after refluxing with hydrazine hydrate yielded
hydrazide derivative (3) in good yield and purity. The
transformation of the resultant hydrazide into imines by
Chem Biol Drug Des 2016; 87: 265–274
267

Parthiban et al.
employing various substituted benzaldehydes afforded final
compounds (4–24). The purity of the synthesized com-
pounds was analyzed by LC–MS and elemental analyses,
and the structures were confirmed by spectral techniques
(
(
Appendix S1). In the nuclear magnetic resonance spectra
H-NMR and C-NMR), the signals of the respective pro-
1
13
tons of the prepared derivatives were verified on the basis
of their chemical shifts, multiplicities, and coupling con-
stants. The elemental analysis results were within ꢁ 0.4%
of the theoretical values.
All the synthesized compounds were further experimentally
screened for the MTB LAT inhibitory potency using the
protocol described in the Supportive Information section.
Out of the 21 molecules tested, ten molecules exhibited
MTB LAT inhibitory IC50 < 18 lM (Table 1) and were more
potent than Lead 1. Compound 21 emerged as the most
potent compound with IC50 of 0.81 ꢁ 0.03 lM.
With respect to the structure–activity relationship explo-
rations, in general, out of the various substitutions
attempted on the left as well as the right hand core at R
Figure 4: Interaction profile diagram of active lead compound
Lead 1.
Table 1: Structure, MTB LAT inhibition and cytotoxicity of synthesized compounds
NHN
NHN
NHN
NHN
R
R
R
R
O
-24
2
5-27
O
4
O
O
O
Compound
R
MTB LAT IC50 in lM
Cytotoxicity at 100 lM
Lead 1
–
18.06 ꢁ 1.63
8.47 ꢁ 0.13
42.3 ꢁ 0.18
41.68 ꢁ 0.31
54.01 ꢁ 1.02
58.61 ꢁ 1.12
95.3 ꢁ 2.29
5.46 ꢁ 0.12
1.96 ꢁ 0.17
7.55 ꢁ 0.82
11.5 ꢁ 0.20
2.78 ꢁ 0.52
39.33 ꢁ 0.68
6.14 ꢁ 0.19
12.40 ꢁ 0.66
65.4 ꢁ 1.46
1.02 ꢁ 0.18
44.55 ꢁ 1.13
0.81 ꢁ 0.03
> 50
48.12 ꢁ 2.12
4
5
6
7
8
9
2-Hydroxy phenyl
3-Hydroxy phenyl
4-Hydroxy phenyl
2-Methoxy phenyl
3-Methoxy phenyl
4-Methoxy phenyl
2-Benzyloxy phenyl
3-Benzyloxy phenyl
4-Benzyloxy phenyl
2-Methyl phenyl
3-Methyl phenyl
4-Methyl phenyl
2-Nitro phenyl
44.64 ꢁ 0.75
NT
NT
NT
NT
NT
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
9.20 ꢁ 0.90
26.24 ꢁ 0.10
31.56 ꢁ 2.07
43.85 ꢁ 0.80
44.08 ꢁ 0.19
NT
13.97 ꢁ 0.88
38.46 ꢁ 1.24
NT
3-Nitro phenyl
4-Nitro phenyl
2-Fluoro phenyl
3-Fluoro phenyl
4-Fluoro phenyl
2-Bromo phenyl
3-Bromo phenyl
4-Bromo phenyl
4-Fluoro phenyl
2-Fluoro phenyl
3-Benzyloxy phenyl
36.34 ꢁ 0.62
NT
25.57 ꢁ 0.12
NT
> 50
NT
NT
NT
NT
36.97 ꢁ 1.90
> 50
> 50
> 50
NT
NT = indicates not tested.
68
2
Chem Biol Drug Des 2016; 87: 265–274

Mycobacterium tuberculosis Lysine ɛ-Aminotransferase Inhibitors
Figure 5: Synthetic protocol of the compounds 4–24.
position, benzyloxyphenyl- and fluorophenyl-substituted
analogues turned out to be more promising leads. The
synthesized compounds were docked to MTB LAT in
order to support the structure–activity relationship. Com-
pounds 4–6, substituted at the R position with hydroxy
phenyl group at the ortho, meta, and para positions, differ
in their activity ranges (8.47 ꢁ 0.13–42.3 ꢁ 0.18 lM).
Here, hydroxy phenyl in the ortho position turned out to
be the most active compound (compound 4), while meta
and para hydroxy phenyl substituted at the R position
supported by the low docking score of ꢀ5.50 and
ꢀ5.47 kcal/mol (Figure 6).
Further replacement of 2, 3, and 4-methoxy phenyl on the
R position (compound 7, 8 and 9) in the lead molecule
resulted in loss of activity (IC50 54.01 ꢁ 1.02–
95.3 ꢁ 2.29 lM). This was well supported by the interaction
profiles in the docking studies, and the closer analysis of
all the compounds revealed that because of bulky substi-
tution at the R position, both the methoxy and phenyl
groups are bent in the active site, thus losing the interac-
tion with the important amino acids (Figure 7). The binding
energy of all the compounds was found in the range of
ꢀ5.21 to ꢀ5.54 kcal/mol.
(compound 5 and 6) were not active in MTB LAT enzyme
inhibition studies. The binding analysis of compound 4 in
the active site of protein revealed that it is stabilized by
three hydrogen bonding interactions. The first hydrogen
bond was observed between the hydroxyl group on the
phenyl ring with the oxygen atom of Gln303 amino acid,
another hydrogen bond was observed with the side chain
of NH2 group of Gly128 amino acid, and lastly, hydrogen
bonding was seen between O atom of carboxamide with
Ala129 amino acid residue. Apart from this, two strong
metal co-ordination bonds are also observed with NH
group of Lys300 and oxygen atom of Ash271. In addition
to these interactions, the phenyl ring interacted via p–p
interaction with Phe167, analogues to the one observed
with the crystal ligand. The binding energy of compound 4
is found to be ꢀ8.25 kcal/mol, while the predicted binding
pose of the less active compound (5 and 6) suggests that
both the compounds took different orientation because
the hydroxyl group of both the compounds failed to retain
the hydrogen bonding interaction with Gln303. This is also
Next, 2, 3, and 4-benzyloxy phenyl is substituted at the R1
position (compound 10, 11 and 12) Here, all the com-
pounds are showing good activity with IC50 of
1.96 ꢁ 0.17–7.55 ꢁ 0.82 lM. The binding analysis of all
the active compounds revealed that the they are involved
2
in hydrogen bonding interaction with the side of NH of
Arg170, which is considered to be crucial for enzyme inhi-
bition studies. Apart from this, the compounds were fur-
ther stabilized by p–p interaction with Phe167 (Figure 8).
With reference to compound 12 instead of Arg170, it is
interacting with the oxygen atom of Glu243 amino acid
residue where it reveals a Glu243 ‘switch’ through which
the enzyme changes its substrate specificities (10). Apart
from this, the benzyloxy phenyl group is inserted into the
hydrophobic cavity where it stabilized through various
Figure 6: Interacting pattern of the compound 4, 5 and 6 in the active site of the LAT protein.
Chem Biol Drug Des 2016; 87: 265–274
269

Parthiban et al.
Figure 7: Interacting pattern of the compound 7, 8 and 9 in the active site of the LAT protein.
Figure 8: Interacting pattern of the compound 10, 11 and 12 in the active site of the LAT protein.
hydrophobic interactions. The activity of all the compounds
were well correlated with the docking scores of the com-
pounds which were found in the range of ꢀ6.01 to
gous to those observed with the crystal ligand where
Arg170 and Phe167 were considered to be crucial for
bioactivity (Figure 9). Further, the 3-methyl phenyl at both
sides oriented deeply into the active site pocket and was
found to be stabilized by hydrophobic interactions with
Val32, Val273, Ala129, Pro415, and Ala146 residues.
However, it was surprising to note that replacement of the
para-substituted methyl phenyl at R (compound 15) led to
a reduction in the activity (IC50 = 39.33 ꢁ 0.68 lM) as
compared to compound 14. The binding analysis of this
compound in the active site of MtbLAT revealed that the
molecule was found to be in a slightly different orientation/
pose compared to that of compound 14, thereby losing
the hydrogen bonding interactions with important amino
acid residues (Figure 9). The ortho position of methyl
ꢀ
6.35 kcal/mol.
Further, the effect of methyl phenyl substitution at the R
position (13–15) revealed that the meta position of methyl
phenyl at R (compound 14) is more active as compared to
the ortho and para positions (compound 13 and 15). The
compound 14 is showing IC50 of 2.78 ꢁ 0.52 lM, which is
sixfold higher potent than the identified hit (IC50 18.06 lM).
This was well supported by interaction profile in their dock-
ing studies as well with a glide docking score of ꢀ6.91 ori-
enting in a similar manner to the crystal ligand retaining
hydrogen bonding interactions and p–p interaction, analo-
Figure 9: Interacting pattern of the
compound 13, 14 and 15 in the active site
of the LAT protein.
2
70
Chem Biol Drug Des 2016; 87: 265–274

Mycobacterium tuberculosis Lysine ɛ-Aminotransferase Inhibitors
phenyl group is showing moderately activity with IC50 of
the lead compound (Lead 1). Substitution with 2-fluo-
rophenyl group (compound 19) is also showing good
activity with 1.02 ꢁ 0.18 lM, while replacement of the 3-
fluorophenyl group at R resulted in compound 20, with
decreased activity (IC50 = 44.55 ꢁ 1.13 lM). The ligand-
binding analysis of the most active compound 21
revealed a good binding affinity against the protein with
the docking score of ꢀ7.06 kcal/mol. In compound 21,
4-fluorophenyl at both positions was found to be involved
in the polar contacts with Gln274, Glu238, Glu243, and
Thr180 as illustrated in Figure 11. In silico analysis of this
compound indicated that the molecules oriented in a
similar manner to that crystal ligand, maintaining hydro-
1
1.5 ꢁ 0.20 lM. The compound 13 is showing hydrogen
bonding interaction with Ala129 and metal co-ordination
bond with NH2 group of Lyn300; it is also stabilized by
p–p interaction with the phenyl ring of Phe167 amino acid
residue.
Next, variation in 2, 3, and 4-nitro phenyl substituent’s on
the lead molecule at the R position was investigated (com-
pound 16–18). Here, the 2-nitro phenyl is showing good
activity (IC50 = 6.14 ꢁ 0.19 lM), while the 4-nitro phenyl-
substituted compound led to a drastic reduction in the
activity (IC50 = 65.40 ꢁ 1.46 lM). Substitution with 3-nitro
phenyl is also showing good activity (IC50 = 12.40
gen bonding interaction with the side chain of the NH
2
ꢁ
0.66 lM). The binding analysis revealed that in com-
group of Arg170 and Asp271 amino acid residues. The
compounds were further stabilized by the various
hydrophobic interactions with Ile184, Val63, Ala416,
Pro415, Ala129, Val132, and Val273 amino acid residues.
Comparison of binding mode of compound 21 and the
crystal lysine substrate is illustrated in Figure 11. While
the predicted binding pose of the less active compound
pound 18, the 4-nitro phenyl is slightly away from the
active site and it does not fully occupy the active site
pocket, while compound 16 fully occupies the active site
pocket and it also maintains the crucial hydrogen bonding
as well as p–p interaction with Arg170 and Phe167 amino
acid residues (Figure 10).
(
20) suggests that the 3-fluoro group is slightly outside
Out of the various substitutions attempted on the R posi-
tion, the fluorophenyl-substituted analogues were also
turned out to be more promising leads. Among these,
the most active compound 21 with a 4-fluorophenyl
group at the R position showed good in vitro inhibitory
activity with an IC50 of 0.81 ꢁ 0.03 lM which showed a
the pocket, the orientation of the molecule is also chan-
ged which makes the compound loose interaction with
Arg170; here, oxygen atom is interacting with the side
chain of the NH
2
group of Ala129 instead of Arg170
(Figure 12). The less activity of compound 20 is further
supported by the low docking score of ꢀ5.15 kcal/mol
as compared to compound 21.
2
2 times more increment in activity when compared to
Figure 10: Interacting pattern of the compound 16, 17 and 18 in the active site of the LAT protein.
Figure 11: Comparison of binding mode of
compound 21 and crystal ligand.
Chem Biol Drug Des 2016; 87: 265–274
271

Parthiban et al.
Figure 12: Interacting pattern of the
compound 19 and 20 in the active site of
the LAT protein.
Lastly, the effect of 2, 3, and 4-bromo phenyl substituted
at the R position (compound 22–24) was checked. Here,
all the molecules did not show good activity in MTB LAT
enzyme inhibition studies which is also indicated in their
interaction profile where the molecule was found to be in a
slightly different orientation/pose; thus, they are outside
the binding pocket resulting in the loss of key hydrogen
bonding interactions in the active site of protein
pounds 21, 19 and 11. The docking scores were found in
the range of ꢀ4.12 to ꢀ4.90 kcal/mol. Further, the orien-
tation is these compounds was changed relative to that of
active compounds, which again explains the difference in
their activity profiles (Figure 14).
Betts et al. (7) designed a simple model where it can be
used to test novel antimycobacterial drugs active against
persistent bacteria, and they have developed a nutrient
starvation model, in which MTB is in a state of little or no
replication, exhibits a low respiration rate, and yet main-
tains long-term viability. In addition, cultures are resistant
to both isoniazid (INH) and rifampicin (RIF). Through pro-
teome analysis, they have been able to identify proteins
induced under these conditions and have then gone on to
characterize gene expression patterns occurring as the
(Figure 13). All the compounds showed low docking score
in the range of ꢀ4.98 to ꢀ5.00 kcal/mol as compared to
other active compounds. Further, replacing the central
oxygen atom of the most active compounds 21, 19 and
1
1 with methylene makes the compounds completely
inactive (Table 1). Binding analysis of these compounds
25–27) revealed that the docking scores of these com-
pounds were found very low as compared to active com-
(
Figure 13: Interacting pattern of the compound 22, 23 and 24 in the active site of the LAT protein.
Figure 14: Interacting pattern of the compound 25, 26 and 27 in the active site of the LAT protein.
72 Chem Biol Drug Des 2016; 87: 265–274
2

Mycobacterium tuberculosis Lysine ɛ-Aminotransferase Inhibitors
bacilli adapt to starvation using DNA microarrays. In their
10 lg/mL concentration, standard first-line anti-TB drugs
INH and RIF reduced ~1.1 and 2.0 log bacterial reduction,
respectively, whereas MOXI showed high activity with ~2.4
log bacterial reduction (Figure 15). Minimum inhibitory con-
centration (MIC) of INH was 0.05 lg/mL, Rif was 0.1 lg/
mL, and moxifloxacin was 0.78 lg/mL against active
MTB. When compared to INH, five compounds showed to
be more active and compound 21 was more potent than
RIF. We have also tested compound 22 which has the
IC50 of > 50 lM in MTB LAT enzyme study, which was not
active in dormant model. All these synthesized molecules
were not active log phase culture of MTB (MIC of > 25 lg/
mL), and it indicates that these molecules targeting latent
MTB target LAT.
model, LAT is up-regulated approximately 40 times and
we have used the model to screen the most potent LAT
inhibitors synthesized by us. We have screened eight
molecules which showed IC50 of < 10 lM, together with
standard compounds like INH, RIF, and moxifloxacin
(
MOXI). Cultures of in vitro-grown MTB bacteria were sub-
jected to nutrient depletion by growing the culture in PBS
phosphate-buffered saline) for 6 weeks. Nutrient starva-
(
tion using this method triggered a dormancy response in
the bacilli that were termed non-replicating persistence
(NRP), a physiological state thought to mimic the one
exhibited by MTB during various stages of persistent infec-
tion. After 6 weeks, the culture was treated with the syn-
thesized and standard drugs at a concentration of 10 lg/
mL in a tube and incubated at 37 °C for 7 days. The trea-
The safety profile of the most potent compounds was
assessed by in vitro cytotoxicity in RAW 264.7 cells at
100 lM concentration using (4,5-dimethylthiazol-2-yl)-2,5-
diphenyltetrazolium bromide (MTT) assay. As MTB bacteria
reside inside the macrophages, the monocyte macrophage
cell lines (RAW 264.7) were mostly used for anti-TB drug
discovery research, to check whether the test compounds
ꢀ6
ted cell suspensions were diluted 10-fold up to 10 using
Middlebrook 7H9 medium supplemented with OADC and
were plated in 48-well plates in triplicates. The plates were
incubated at 37 °C for 4 weeks, and the wells with visible
bacterial growth were counted as positive and MPN values
were calculated using standard statistical methods (11). At
Figure 15: In vitro antitubercular activity
against nutrient starved MTB at 10 lg/mL
Figure 16: Melt curve for the protein
complex with inhibitor 21 using differential
scanning flourimetry.
Chem Biol Drug Des 2016; 87: 265–274
273

Parthiban et al.
were non-toxic toward macrophages but toxic to the
M. tuberculosis. Almost all the tested compounds demon-
strated a good safety profile with very low inhibitory
potential. Compounds were inhibited within a range of
2. Kunst H., Khan K.S. (2010) Age-related risk of hepato-
toxicity in the treatment of latent tuberculosis infection: a
systematic review. Int J Tuberc Lung Dis;14:1374–1381.
3. Cardona P.J. (2007) New insights on the nature of
latent tuberculosis infection and its treatment. Inflamm
Allergy Drug Targets;6:27–39.
4. Stewart G.R., Robertson B.D., Young D.B. (2003) Tu-
berculosis: a problem with persistence. Nat Rev Micro-
biol;1:97–105.
9
.2–48.12%. Furthermore, the most active compound 21
showed 25.57% inhibition at 100 lM concentration of the
drug, reflecting its safety profile in the eukaryotes, and
cytotoxicity results of all the compounds are illustrated in
the Table 1.
5
. Zhang Y. (2004) Persistent and dormant tubercle bacilli
Furthermore, the binding affinity of the most potent ana-
logue was evaluated by measuring the thermal stability of
the protein–ligand complex using biophysical differential
scanning fluorimetry experiments. Compound 21 displayed
and latent tuberculosis. Front Biosci;9:1136–1156.
6. Wayne L.G., Sohaskey C.D. (2001) Nonreplicating per-
sistence of Mycobacterium tuberculosis. Ann Rev
Microbiol;55:139–163.
a T
m
shift of 2.1 °C (T
m
= 43.2 °C) compared with the
7. Betts J.C., Lukey P.T., Robb L.C., McAdam R.A., Dun-
can K. (2002) Evaluation of a nutrient starvation model of
Mycobacterium tuberculosis persistence by gene and
protein expression profiling. Mol Microbiol;43:717–731.
8. Roxas B.A., Li Q. (2009) Acid stress response of a
mycobacterial proteome: insight from a gene ontology
analysis. Int J Clin Exp Med;2:309–328.
native protein (Tm = 41.1 °C) (Figure 16), a repercussion of
strong binding of the ligand to the protein, and highly cor-
related with its LAT IC50 of 0.8 ꢁ 0.58 lM. A higher or
m
positive shift of melting temperature (T ) of protein–ligand
complex compared to the native protein Tm signified better
stabilization of the protein–ligand complex, which in turn
would reflect on the inhibitor binding (12).
9. Voskuil M.I., Bartek I.L., Visconti K., Schoolnik G.K.
(2011) Mycobacterium tuberculosis response to stress
In conclusion, we disclose a new class of compounds that
target the mycobacterial LAT enzyme which overexpressed
in dormant TB, with promising attributes of synthetic
accessibility and antitubercular activity. With new anti-TB
agents, activity against dormant TB was desperately
needed, and we believe that the present class of MTB
LAT inhibitors reported in this work, which would be inter-
esting as potential leads for further optimization.
from reactive oxygen and nitrogen species front. Micro-
biology;2:105.
10. Tripathi S.M., Ramachandran R. (2006) Direct evidence
for
a glutamate switch necessary for substrate
recognition: crystal structures of lysine e-aminotrans-
ferase (Rv3290c) from Mycobacterium tuberculosis
H37Rv. J Mol Biol;362:877–886.
11. De Man J.C. (1975) The probability of most probable
numbers. J Appl Microbiol;1:67–78.
1
2. Niesen F.H., Berglund H., Vedadi M. (2007) The use of
differential scanning fluorimetry to detect ligand inter-
actions that promote protein stability. Nat Pro-
toc;2:2212–2221.
Acknowledgments
DS acknowledges University Grants Commission, New
Delhi, India, for UGC Research Award.
Note
a
Conflict of Interest
WHO:
http://whqlibdoc.who.int/publications/2011/
9
789241500708_eng.pdf.
No conflict of interest to declare.
Supporting Information
References
Additional Supporting Information may be found in the
online version of this article:
1
. Druszczy nꢁ ska M., Kowalewicz -Kulbat M., Fol M., Wlo-
darczyk M., Rudnicka W. (2012) Latent M. tuberculosis
infection – pathogenesis, diagnosis, treatment and pre-
vention strategies. Polish J Microbiol;61:3–10.
Appendix S1. Methodology used for docking, spectral
details of all the synthesized compounds and the protocol
used for biological evaluation.
2
74
Chem Biol Drug Des 2016; 87: 265–274