Assessment of mycobacterium tuberculosis pantothenate kinase vulnerability through target knockdown and mechanistically diverse inhibitors

Article abstract of DOI:10.1128/AAC.00140-14

Pantothenate kinase (PanK) catalyzes the phosphorylation of pantothenate, the first committed and rate-limiting step toward coenzyme A (CoA) biosynthesis. In our earlier reports, we had established that the type I isoform encoded by the coaA gene is an essential pantothenate kinase in Mycobacterium tuberculosis, and this vital information was then exploited to screen large libraries for identification of mechanistically different classes of PanK inhibitors. The present report summarizes the synthesis and expansion efforts to understand the structure-Activity relationships leading to the optimization of enzyme inhibition along with antimycobacterial activity. Additionally, we report the progression of two distinct classes of inhibitors, the triazoles, which are ATP competitors, and the biaryl acetic acids, with a mixed mode of inhibition. Cocrystallization studies provided evidence of these inhibitors binding to the enzyme. This was further substantiated with the biaryl acids having MIC against the wild-type M. tuberculosis strain and the subsequent establishment of a target link with an upshift in MIC in a strain overexpressing PanK. On the other hand, the ATP competitors had cellular activity only in a M. tuberculosis knockdown strain with reduced PanK expression levels. Additionally, in vitro and in vivo survival kinetic studies performed with a M. tuberculosis PanK (MtPanK) knockdown strain indicated that the target levels have to be significantly reduced to bring in growth inhibition. The dual approaches employed here thus established the poor vulnerability of PanK in M. tuberculosis.

Full text of DOI:10.1128/AAC.00140-14

Assessment of Mycobacterium
tuberculosis Pantothenate Kinase
Vulnerability through Target Knockdown
and Mechanistically Diverse Inhibitors
B. K. Kishore Reddy, Sudhir Landge, Sudha Ravishankar,
Vikas Patil, Vikas Shinde, Subramanyam Tantry, Manoj
Kale, Anandkumar Raichurkar, Sreenivasaiah Menasinakai,
Naina Vinay Mudugal, Anisha Ambady, Anirban Ghosh,
Ragadeepthi Tunduguru, Parvinder Kaur, Ragini Singh,
Naveen Kumar, Sowmya Bharath, Aishwarya Sundaram,
Jyothi Bhat, Vasan K. Sambandamurthy, Christofer
Björkelid, T. Alwyn Jones, Kaveri Das, Balachandra
Bandodkar, Krishnan Malolanarasimhan, Kakoli Mukherjee
and Vasanthi Ramachandran
Antimicrob. Agents Chemother. 2014, 58(6):3312. DOI:
10.1128/AAC.00140-14.
Published Ahead of Print 31 March 2014.
Updated information and services can be found at:
http://aac.asm.org/content/58/6/3312
These include:
REFERENCES
This article cites 29 articles, 13 of which can be accessed free
at: http://aac.asm.org/content/58/6/3312#ref-list-1
CONTENT ALERTS
Receive: RSS Feeds, eTOCs, free email alerts (when new
articles cite this article), more»
Information about commercial reprint orders: http://journals.asm.org/site/misc/reprints.xhtml
To subscribe to to another ASM Journal go to: http://journals.asm.org/site/subscriptions/

Assessment of Mycobacterium tuberculosis Pantothenate Kinase
Vulnerability through Target Knockdown and Mechanistically Diverse
Inhibitors
B. K. Kishore Reddy,a Sudhir Landge,a Sudha Ravishankar,b Vikas Patil,a Vikas Shinde,a Subramanyam Tantry,a Manoj Kale,a
Anandkumar Raichurkar,a Sreenivasaiah Menasinakai,a Naina Vinay Mudugal,b Anisha Ambady,b Anirban Ghosh,b*
Ragadeepthi Tunduguru,b* Parvinder Kaur,b Ragini Singh,e Naveen Kumar,c Sowmya Bharath,c Aishwarya Sundaram,b* Jyothi Bhat,b
Vasan K. Sambandamurthy,b Christofer Björkelid,d T. Alwyn Jones,d Kaveri Das,b Balachandra Bandodkar,a*
Krishnan Malolanarasimhan,a* Kakoli Mukherjee,b* Vasanthi Ramachandranb
Department of Chemistry, iMED infection, AstraZeneca India Pvt. Ltd., Bangalore, Indiaa; Department of Bioscience, iMED infection, AstraZeneca India Pvt. Ltd., Bangalore,
Indiab; Department of DMPK and Animal Sciences, iMED infection, AstraZeneca India Pvt. Ltd., Bangalore, Indiac; Department of Cell and Molecular Biology, Uppsala
University, Uppsala, Swedend; Cellworks Research India Pvt. Ltd., Bangalore, Indiae
Pantothenate kinase (PanK) catalyzes the phosphorylation of pantothenate, the first committed and rate-limiting step toward
coenzyme A (CoA) biosynthesis. In our earlier reports, we had established that the type I isoform encoded by the coaA gene is an
essential pantothenate kinase in Mycobacterium tuberculosis, and this vital information was then exploited to screen large li-
braries for identification of mechanistically different classes of PanK inhibitors. The present report summarizes the synthesis
and expansion efforts to understand the structure-activity relationships leading to the optimization of enzyme inhibition along
with antimycobacterial activity. Additionally, we report the progression of two distinct classes of inhibitors, the triazoles, which
are ATP competitors, and the biaryl acetic acids, with a mixed mode of inhibition. Cocrystallization studies provided evidence of
these inhibitors binding to the enzyme. This was further substantiated with the biaryl acids having MIC against the wild-type M.
tuberculosis strain and the subsequent establishment of a target link with an upshift in MIC in a strain overexpressing PanK. On
the other hand, the ATP competitors had cellular activity only in a M. tuberculosis knockdown strain with reduced PanK expres-
sion levels. Additionally, in vitro and in vivo survival kinetic studies performed with a M. tuberculosis PanK (MtPanK) knock-
down strain indicated that the target levels have to be significantly reduced to bring in growth inhibition. The dual approaches
employed here thus established the poor vulnerability of PanK in M. tuberculosis.
he burden of tuberculosis (TB) has reached unprecedented olite central to many vital functions. The current report presents
T
levels worldwide (1), and the rapid emergence of multi-
an evaluation of the use of one of the key enzymes of the coenzyme
drug-resistant Mycobacterium tuberculosis strains has now been A (CoA) biosynthetic pathway as an option for designing a new
recognized as a major challenge for global TB control measures anti-TB drug. The CoA cofactor is an essential acyl group carrier
(2). To eliminate TB as a public health problem by 2050, the inci- indispensable for respiration and lipid metabolism in various or-
dence of infection has to fall by 16% per year for the next 30 to 40 ganisms. Loss of intracellular CoA levels either through the use of
years (3); however, the rates are falling by only 2% (1). The diffi- inhibitors or gene knockouts of the enzymes involved in the bio-
culties encountered by all efforts to address this concern have been
compounded with additional impediments, including coinfection
of HIV and M. tuberculosis, emergence and spread of multidrug-
resistant (MDR) and extensively drug-resistant (XDR) M. tuber-
culosis strains, and inadequate diagnostic and control measures
(2). There is an urgent need for defined TB regimens that are
shorter, efficacious, and, most importantly, safer. Such therapy
should be affordable and practical for use in low-resource settings
and free of interactions, especially with antiretroviral regimens
(4). Among the different drug exploration strategies available, the
key to the success of a target-based anti-TB program rests on iden-
tification of multiple chemical scaffolds with mechanistically di-
verse modes of inhibition. Such a strategy could overcome some of
the challenges associated with poor druggability and, possibly, the
often-encountered issue of potent inhibitors lacking the ability to
translate to antibacterial activity. In pursuit of identifying targets
for novel antimycobacterials, it is very important to limit the
choice to those targets that impact viability with the slightest per-
turbation in protein levels. This adverse effect on survival or per-
sistence normally happens owing to the loss of an essential metab-
synthetic pathway have demonstrated that inhibition of CoA syn-
Received 22 January 2014 Returned for modification 26 February 2014
Accepted 24 March 2014
Published ahead of print 31 March 2014
Address correspondence to Vasanthi Ramachandran,
vasanthi.ramachandran@astrazeneca.com.
* Present address: Anirban Ghosh, Department of Microbial and Molecular
Systems, Faculty of Bioscience Engineering, KU Leuven, Leuven, Belgium;
Ragadeepthi Tunduguru, Department of Biochemistry and Molecular Biology,
Indiana University School of Medicine, Indianapolis, Indiana, USA; Aishwarya
Sundaram, Joint Research Division, Molecular Metabolic Control, German Cancer
Research Center Heidelberg, Center for Molecular Biology Heidelberg, Heidelberg
University Hospital, Heidelberg, Germany; Balachandra Bandodkar, Alkem
Laboratories Ltd., Bangalore, India; Krishnan Malolanarasimhan, Fermentation
Technology Development Centre, Dr. Reddy’s Laboratories Ltd., Hyderabad, India;
Kakoli Mukherjee, Alkem Laboratories Ltd., Bangalore, India.
B.K.K.R., S.L., and S.R. contributed equally to this article.
Copyright © 2014, American Society for Microbiology. All Rights Reserved.
doi:10.1128/AAC.00140-14
3312 aac.asm.org
Antimicrobial Agents and Chemotherapy p. 3312–3326
June 2014 Volume 58 Number 6

Elucidation of PanK Vulnerability in M. tuberculosis
TABLE 1 Bacterial strains and plasmids used in this study
Strain or plasmid
Descriptiona
Source or reference
Strains
E. coli pMOS Blue
F= endA1 hsdR17 (rKϪ mKϩ), supE44 thi-1 recA1 gyrA96 relA1 lac [F= lacIqZ⌬M15
proABϩ Tn10 (Tetr)
GE Healthcare
M. tuberculosis H37Rv ATCC 27294
M. bovis BCG—CoaA OE strain
Virulent laboratory strain of M. tuberculosis
MtPanK-overexpressing M. bovis BCG
AZI collection
This study
Plasmids
pBAN0477
pAZI9018b
M. tuberculosis coaA gene cloned at NdeI and BamHI cloning site, Hygr
E. coli-mycobacterial IPTG-inducible shuttle vector—a replicating vector with
lacZ, Hygr, NdeI, and BamHI as cloning sites
This study
12
pBAN0317
pAZI9479
M. tuberculosis coaA conditional expression vector with truncated, mutant coaA
gene cloned downstream of pristinamycin-inducible promoter (pPTR)
Mycobacterial conditional-expression (integrating) vector with pPTR system
This study
13
a Tetr, tetracycline resistance; Hyg, hygromycin resistance.
thesis is a viable option to discover new antimicrobials (5–7). inhibitors with nanomolar enzyme potency and (ii) assessment of
Significant differences between the bacterial and human counter- vulnerability by conditional expression of MtPanK.
parts enable selective inhibition of the targets on this pathway.
MATERIALS AND METHODS
Bacterial strains, media, chemicals, and reagents. E. coli MOS Blue cells
However, for slow-growing pathogens such as M. tuberculosis,
which are known to persist within the host for extended periods,
there is an added requirement of complete killing of the pathogen
and its elimination from the host to ensure a cure.
CoA is assembled in five steps from pantothenate, and the
first reaction in this pathway, the phosphorylation of pantothe-
(Amersham) were used for gene cloning and plasmid propagation. The M.
bovis BCG-Pasteur Merieux wild-type strain and M. smegmatis mc2155
were used as expression hosts. M. tuberculosis H37Rv ATCC 27294 was
used to generate a coaA conditional-expression strain. Electrocompetent
E. coli and mycobacterial cells were prepared as described earlier (12). All
the enzymes used for cloning were from either New England BioLabs or
nate to yield 4=-phoshopantothenate, is catalyzed by the en-
zyme pantothenate kinase (PanK). PanK, the key rate-deter-
GE Healthcare. Isopropyl-beta-thiogalactopyranoside (IPTG), rifampin,
mining enzyme of the CoA biosynthesis pathway (8), exists in and isoniazid were obtained from Sigma, hygromycin was from Roche,
and pristinamycin (Synercid) was from Sanofi Aventis. Zirconia beads
(0.1 mm diameter) and a Mini-BeadBeater were purchased from Biospec
Products. In general, LB medium was used for E. coli growth and 7H9
medium (Middlebrook 7H9 medium supplemented with 1% albumin-
dextrose-catalase [ADC], 0.2% glycerol, and 0.05% Tween 80) for myco-
bacterial growth. Unless otherwise mentioned, the coaA conditional-ex-
pression strain was grown in 7H9 medium supplemented with 50 ␮g/ml
hygromycin and 10 ng/ml pristinamycin 1 (P1). Details of the strains used
in this study are summarized in Table 1.
three different isoforms. Type I PanK, encoded by the coaA
gene, is present in many bacterial species, including Escherichia
coli. Type II PanK is found mostly in eukaryotes but also in some
bacteria, including Staphylococcus aureus. Other bacteria, such as
Bacillus anthracis, utilize a type III PanK, encoded by the gene
coaX. Recent work with a conditional coaX mutant has demon-
strated that B. anthracis type III PanK is an essential enzyme,
thereby validating this enzyme as an antimicrobial target (30). The
same conclusion could not be extended to and established for the
Protein expression, purification, and inhibitor screening for IC50
M. tuberculosis coaX gene, since a deletion mutant showed no determinations in MtPanK assay. MtPanK expression, purification, and
activity assays were carried out as described earlier (10). Routine screening
and 50% inhibitory concentration (IC50) estimations for the MtPanK
inhibitors were carried out by keeping both substrate and cofactor con-
centrations at their respective Km values, i.e., pantothenate at 122 ␮M
and ATP at 395 ␮M. For selected inhibitors, competition with ATP was
checked by measuring IC50 at a 50ϫ Km ATP concentration, i.e., 6 mM.
Generation of recombinant M. bovis BCG overexpressing MtPanK.
The full-length coaA gene was PCR amplified from M. tuberculosis
genomic DNA using primers MtbcoaAF (5=-ATTCCAACATATGATGTC
GCGGCTTAGCGAGCC-3=) and MtbcoaAR (5=AGGCCTAGGATCCTT
growth defects in vitro, in macrophages or in mice, thereby ruling
out the essentiality of coaX in this organism (9). In contrast, coaA
could be inactivated only in the presence of an extra copy of the gene,
thus proving its essentiality for the survival of M. tuberculosis (9).
We had earlier reported the identification of M. tuberculosis
PanK (MtPanK) inhibitors with multiple modes of inhibition ob-
tained through screening of large libraries (10), an approach that
could pave the path for evaluating targets such as MtPanK. Among
these, the triazoles were found to be competitive with ATP, while
the biaryl acids were mixed noncompetitive. Specific interaction ACAGCTTGCGCAGCCGCA-3=). Recombinant plasmid pBAN0477 was
generated by cloning the amplicon at BamHI and NdeI sites of
pAZI9018b, a replicating E. coli-mycobacterium shuttle vector, with an
IPTG-inducible promoter (12). Following transformation into M. bovis
BCG, recombinants were identified by PCR and the transcript levels were
monitored by reverse transcription-PCR (RT-PCR).
MIC tests were set up per CLSI (Clinical and Laboratory Standards
Institute) guidelines (14). Modes of action of the triazoles and the
biaryls were evaluated by MIC modulation using a M. tuberculosis coaA
knockdown strain (coaAKD) (described below) and a BCG strain overex-
pressing MtPanK, respectively. Wild-type M. tuberculosis and M. bovis
with the target was shown by determining the crystal structures
with both types of inhibitors bound in the enzyme (11). These
inhibitors were then evaluated for their ability to translate to cel-
lular growth inhibition with in silico predictive tools as well as in
vitro experimentation. In order to confirm the link with the target
for compounds that had cellular activity, MIC modulation studies
were carried out with MtPanK overexpression and knockdown
strains. However, the lack of cellular inhibition with some of the
inhibitors prompted us to investigate this further through a two-
pronged strategy: (i) exploration of chemical space to get more BCG were used as control strains. Isoniazid and rifampin served as refer-
June 2014 Volume 58 Number 6
aac.asm.org 3313

Reddy et al.
ence drugs. MICs were determined in the presence of 100 ␮M IPTG for tion. A virtual platform was built for mapping a network of pathways of
the MtPanK overexpression strain and either 0 or 10 ng/ml P1 for the M. M. tuberculosis with Cellworks’ proprietary technology (iC-PHYS) using
ordinary differential equations along lines similar to those used for E. coli
experiments (16). In this mathematical model, more than 10 key path-
ways (glycolysis, tricarboxylic acid [TCA], pentose phosphate, glyoxylate
shunt, branched-chain amino acid synthesis, and mycolic acid biosynthe-
sis and other cell wall component biosynthetic pathways, including
arabinogalactan, peptidoglycan, PIM-LAM, etc.) were simulated. Ad-
ditionally, this platform could predict the outcome of any type of
inhibitor-target interaction (competitive, noncompetitive, mixed
mode, and uncompetitive) as well as the resultant modulation in me-
tabolite levels within the cells. This in silico kinetic platform is also
equipped to evaluate and differentiate between genetic and chemical
vulnerability. In the case of the genetic vulnerability assessment, the
target level is sequentially reduced and the impact on cell viability
monitored, whereas in the case of the chemical vulnerability assess-
ment, cell viability is evaluated by incorporating the kinetic parame-
ters (rate constants, Km, Vmax) of the target as well as its substrate and
products into the platform. In both cases, biomass levels served as
indicators of cell growth. The inhibitors were ranked with respect to
their [I]/Ki values so as to gauge the lowest MIC for each type of
inhibition (16). The Ki value used for this simulation was 0.1 ␮M.
Chemical synthesis. (i) Triazoles. Triazoles were synthesized starting
with coupling of thiosemicarbazide 1 with N-BOC (N-butoxycarbonyl)
amino acid 2 in the presence of HATU {1-[bis(dimethylamino)methyl-
ene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxid hexafluorophosphate}
to get corresponding amide 3, which underwent cyclization in the pres-
ence of sodium hydroxide to afford triazole 4. The free thiol group of 4 was
alkylated with halide 5 to give triazole 6, which, upon deprotection fol-
lowed by coupling with acid chloride, gave triazole 7 (Fig. 1). Diversity in
the scaffold was introduced by thiosemicarbazides 1 or amino acids 2 or
alkyl halides 5 and acid chlorides.
(ii) Quinolones. Quinolone ureas were synthesized starting from sub-
stituted anilines as given in Fig. 2. Acetylation of substituted aniline 8
resulted in corresponding acetanilide 9, which under Vilsmeier-Haack
reaction conditions afforded chloroquinoline aldehyde 10. Aldehyde 10
was converted into quinolone aldehyde 11 using sodium acetate and ace-
tic acid. Compound 11, upon reductive amination with various amines
12, afforded secondary amines 13, which on reaction with isocyanates 14
afforded ureas 15.
(iii) BAA. Biaryl acetic acids (BAA) were synthesized starting from
3-hydroxy benzaldehyde as given in Fig. 3. Bromination of benzaldehyde
16 at 0°C resulted in 2-bromo-5-hydroxybenzaldehyde 17. Reductive al-
kylation of benzaldehyde 17 with piperazine 18 gave intermediate 19,
which, upon subsequent arylation with various boronic acids following
the Suzuki coupling protocol, led to benzaldehyde 20. Alkylation of ben-
zaldehyde 20 with alkyl bromoacetates using Na2CO3-dimethylforma-
mide (DMF) resulted in compounds 21 to 23, which, upon subsequent
hydrolysis, afforded the desired biaryl acetic acids 24 in good yields. Biaryl
acetamides 25 were synthesized starting from biaryl acetic acids 24 using
the HATU/DIEA (diisopropylethylamine) protocol.
tuberculosis coaAKD strain. These MIC tests were performed in 96-well
plates using the Microplate alamar blue method (15) for a 7-day measure-
ment and the turbidometric method for a 14-day measurement.
Generation of M. tuberculosis coaAKD. pAZI9479 is a conditional-
expression vector (13) derived from the pAZI9018b replicating plasmid.
Removal of the mycobacterial origin of replication and replacement of
the IPTG-inducible promoter system with a pristinamycin-inducible
promoter system (unpublished data) yielded pAZI9479. This integrat-
ing vector is particularly useful for generating single-crossover knock-
down strains through homologous recombination. M. tuberculosis coaA
conditional-expression vector pBAN0317 was generated by cloning a
truncated synthetic gene of coaA (bp 1 to 798) carrying an in-frame stop
codon at codon 117 into pAZI9479 at NcoI-SphI sites. Transformants
were selected on 7H10 plates supplemented with 50 ␮g/ml hygromycin
and 100 ng/ml of P1. A set of PCRs was performed to confirm the genotype
of the recombinant coaA conditional-expression strain. The minimum
inducer required for the growth of this strain was determined by plating
dilutions of this culture on 7H10 plates supplemented with 0 to 100 ng/ml
of P1.
In vitro growth kinetics of the M. tuberculosis coaAKD strain. A
culture of the M. tuberculosis coaAKD strain was grown in 7H9 broth
supplemented with 50 ␮g/ml hygromycin and 10 ng/ml P1 up to the
mid-log phase. Harvested cells were washed three times with plain 7H9
medium, resuspended in fresh medium, and used as an inoculum for both
in vitro growth kinetics and in vivo phenotyping experiments. In brief, the
culture concentrate was used as a master stock to inoculate 7H9 broth
supplemented with 50 ␮g/ml hygromycin to get an optical density at A600
of ϳ0.01. This master culture was further split into 3 parts and supple-
mented with 0, 10, and 25 ng/ml of P1. The growth kinetics of these
cultures were monitored by measuring A600 and enumerating CFU at
regular intervals. At each time point, dilutions of these cultures were as-
sessed for CFU on 7H10 agar plates without (for revertants) and with (for
survivors) 10 ng/ml of P1.
Measurement of intracellular PanK levels by Western blotting. Cul-
tures (10 ml) of wild-type M. tuberculosis H37Rv and M. tuberculosis
coaAKD strains were harvested 6 days post-P1 treatments. The cell
pellets were washed twice in phosphate-buffered saline (PBS) and re-
suspended in about 500 ␮l of PBS supplemented with a cocktail of
protease inhibitors. Cell suspensions were transferred to 2-ml screw-
cap tubes (Biospec Products) containing about 0.1 g of 0.1-mm-diam-
eter Zirconia beads (Biospec Products). Cell lysates were prepared by
bead beating the cell suspension twice at 4,000 ϫ g for 20 s each time.
Protein estimation using Bradford reagent was performed on the clar-
ified lysates obtained after centrifuging the cell lysates at 12,000 ϫ g for
5 min. Two micrograms of total protein of each sample was analyzed
by SDS-PAGE, after which the proteins were blotted onto a Hybond
nitrocellulose membrane (GE Healthcare). Rabbit polyclonal anti-
PanK serum (diluted 1:400,000) was used for probing the blot and
developed using an ECL kit (GE Healthcare). Densitometric analysis of
the bands on the autoradiogram was done using the Quantity One tool
on the Bio-Rad gel documentation system.
RESULTS
In vivo phenotyping. The AstraZeneca Animal Ethics Committee,
registered with the government of India (registration no. CPCSEA 99/5),
approved all animal experimental protocols and usage. The culture stocks
were prepared as described above. BALB/c mice (8 to 10 weeks old) were
infected with about 106 CFU of either the M. tuberculosis H37Rv strain or
the M. tuberculosis coaAKD strain. Three mice per group per time point
were sacrificed at designated days postinfection to monitor the course of
infection. Lung and spleen homogenates from these infected animals were
tested for viable bacteria on 7H10 plates for wild-type M. tuberculosis
H37Rv and on 7H10 agar plates supplemented with 50 ␮g/ml hygromycin
and 50 ng/ml of P1 for M. tuberculosis coaAKD.
Identification of diverse chemical series targeting MtPanK. We
conducted two high-throughput screening (HTS) campaigns for
identification of inhibitors of the MtPanK enzyme. The first effort
(campaign 1) screened a diverse library of ϳ70,000 compounds,
while the second effort (campaign 2) screened a larger library of
ϳ1,000,000 compounds. Campaign 1 chemistry focused on tria-
zole and quinolone scaffolds which were shown to be competitive
inhibitors of ATP. In order to increase the diversity of the inhibi-
tor classes, series which were prioritized from campaign 2 were
either uncompetitive with ATP (thiazole) or mixed noncompeti-
Prediction of the most ideal type of PanK inhibition using simula- tive with ATP (biaryl acetic acid and quinoline carboxamide). The
3314 aac.asm.org
Antimicrobial Agents and Chemotherapy

Elucidation of PanK Vulnerability in M. tuberculosis
FIG 1 Synthesis of triazoles. Reagents and conditions were as follows: i, HATU, DIEA, dichloromethane (DCM), room temperature, 12 h, 60% to 70%; ii, 15%
aqueous NaOH, MeOH:1,4 dioxane (1:1), room temperature, 3 h, 60% to 70%; iii, K2CO3, tetrahydrofuran (THF), 40°C, 16 h, 80% to 90%; iv, trifluoroethanoic
acid (TFA), DCM, 0°C to room temperature, 3 h, 90%; v, ArCoCl, triethylamine (TEA), DCM, 0°C to room temperature, 2 to 3 h, 40% to 50%; vi, MeI, NaH,
THF, 0°C to room temperature, 40%.
kinetic mechanism of each scaffold was characterized in detail as yl¡N-benzyl¡N-Me. Replacement of the trifluoro methyl
described earlier (10). The detailed mode of binding of a subset of group (7d) on the R3 phenyl ring with chloro (7j) reduced the
triazole and biaryl acid inhibitors was determined by crystallo- enzyme potency to 0.9 ␮M from 0.2 ␮M. Introduction of a
graphic studies on enzyme-inhibitor complexes. These inhibitors second substitution in the ring brought back the potency; thus,
were found to occupy overlapping positions with CoA and each 7c showed an IC50 of 0.19 ␮M. The most potent molecule, 7a,
other (11).
with an IC50 of 0.08 ␮M at 50ϫ Km ATP, was obtained when the
SAR for triazoles and quinolone ureas. The two series, 4-fluoro benzyl group replaced the three-atom linker on the
namely, triazoles and quinolones, were identified as hits from a left side of triazole.
70K library screening. Both the series had a molecular mass of
Quinolones. Compound 15d (Table 3) was one of the early
about 430 Da and multiple handles to develop the structure-ac- hits, with an IC50 of 1.6 ␮M at 50ϫ Km ATP. Substitution on the
tivity relationship (SAR). Inhibitors from both series were primar- R3 ring with 2-fluoro (15f) resulted in a loss of activity. The sub-
ily competitive with ATP. The initial IC50 values of representatives stitution requirement was exactly opposite in the case of the
of both series were about 100 nM at Km ATP, and the potency was chloro group on the ring. The compound (15g) with a 4-chloro
reduced by 3.5- to Ͼ100-fold when the ATP concentration was group was inactive, while the 2-chloro derivative (15c, with IC50 ϭ
increased to 50ϫ Km ATP. The SAR of the two series is described 0.5 ␮M) was three times more potent than 15d. The 15e com-
in detail below.
pound that had a methoxy group replaced by a fluorine had activ-
Triazoles. The SAR for the triazole series is summarized in ity comparable to that of 15d. Regarding substitutions on the ter-
Table 2. Compound 7m was the initial hit in this series, with an tiary nitrogen of the urea, the cyclopropyl substitution (15h) was
IC50 of 87 nM at Km ATP which then increased to 5.5 ␮M at 50ϫ inactive. When a cyclohexyl group of 15d was changed to cyclo-
Km ATP. All subsequent optimization was done by following the heptyl, the 15a compound showed the best IC50 at 0.2 ␮M. On the
IC50 at 50ϫ Km ATP. Substitution on the R2 ring to p-chloro (7l) other hand, the difluoro derivative (15b) showed a slight loss in
and o-methyl (7k) groups resulted in potencies of 3.1 and 2.8 ␮M, potency (IC50 at 0.6 ␮M). When the cycloalkyl groups were
respectively. When the R1 group on triazole nitrogen was changed changed to aryl or heteroaryl, loss of activity was observed (data
to N-4-fluorobenzyl (7i), potency improved to 0.7 ␮M and not shown).
further improved to 0.32 ␮M upon N-methylation of the
Triazoles and quinolones were screened for MICs against wild-
amide (7h).
type M. tuberculosis along with the M. tuberculosis coaAKD strain
With a 3-pyridyl methyl substitution at triazole nitrogen, R1 in the absence and presence of 10 ng/ml P1. The wild-type and M.
was found to be the best in terms of potency (Table 2). The tuberculosis coaAKD strains did not show any growth inhibition in
order of potency for N substitutions was N-3-pyridyl meth- the presence of 10 ng/ml P1, while MICs were observed under
FIG 2 Synthesis of quinolone ureas. Reagents and conditions were as follows: i, AcCl, TEA, DCM, 0 to 25°C, 3 h, 99%; ii, POCl3, DMF, 0 to 80°C, 16 h, 80%; iii,
NaOAc·3H2O, AcOH, 110°C, 3 h, 80%; iv, NaCNBH3, AcOH, 2-propanol, 40°C, 1 h, 50% to 60%; v, DMAP (4-dimethylaminopyridine), DCM, 2 h, 40% to 50%.
June 2014 Volume 58 Number 6
aac.asm.org 3315

Reddy et al.
FIG 3 Synthesis of biaryl acetic acids. Reagents and conditions were as follows: i, Br2, DCM, 0 to 2°C, 17 h, 76%; ii, NaCNBH3, MeOH, room temperature, 2 h,
68%; iii, boronic acids, Pd(PPh3)4, dimethyl ether (DME):1 M Na2CO3 (2:1), reflux, 5 h, 60%; iv, alkyl bromoacetate, K2CO3, acetone, 70%; v, 5 N HCl:CH3CN
(1:3), 70% to 90%; vi, HATU, DIEA, aq. NH3, 80%.
knockdown conditions (without P1) for a few compounds, indi- potency was lost when the aryl ring was moved farther away by
cating specific target engagement. The improvement in 14-day
MIC values over the 7-day values suggested a requirement for
longer incubation period so as to allow sufficient inhibition of the
target and, consequently, arrest of cell growth. Isoniazid, the
known anti-TB drug, served as a negative control in these experi-
ments (Table 4).
Biaryl acetic acids. Initial hits from the biaryl acetic acid series
(see the representative structure in Fig. 4) were weak inhibitors of
the enzyme (IC50 of ϳ8 to 10 ␮M); hence, several analogs were
synthesized to improve the potency. In order to track the SAR,
three major modifications were made.
(i) Substitutions at the aryl position of the biphenyl ring. In a
traditional medicinal chemistry approach, the phenyl ring was
replaced with electron-withdrawing groups (EWG) or electron-
donating groups (EDG) or replaced with heteroaryl groups. The
SAR data of biraryl acids (Table 5) suggested that the 4= position is
the preferred position for substitutions improving enzyme po-
tency and, notably, that EWG members were preferred to EDG
members. Further, a model experiment with a chlorosubstitution
at the ortho-, meta-, and para- position indicated that the para-
substituent was approximately 15-fold more potent than its ortho-
and meta- counterparts (24b to 24d). The trend was similar with
other EWG members (24i, 24j, 24l, and 24n). We explored the
possibility of replacing the aryl with heteroaryl rings such as pyr-
idine (24g to 24h). The position of an aza group on the heteroaryl
did not have any bearing on enzyme potency. However, introduc-
ing a cyano group at the 4= position of pyridine (24k) brought back
potency. It is certain that the EWG members have a critical role to
play in enhancing the enzyme potency.
converting it into either a benzylic or an amidic linker (data not
shown).
(iii) Oxyacetic acid at the 4= position. It is clear from Table 5
that although the carboxylic acids had very good enzyme inhibi-
tion, it did not translate to potent antimycobacterial activity. This
could be attributed to the lower level of permeation of biaryl acetic
acids across the mycobacterial cell membrane. In order to find a
suitable replacement for acids, we synthesized its bioisostere—
tetrazole (26). This compound was as potent as the parent acid in
terms of enzyme potency; however, this too did not translate into
cellular potency, with the MIC remaining at 64 ␮g/ml (Table 6).
Esters are known to act as prodrugs for acids, as they get
hydrolyzed by various esterases inside the bacterial cell. Keep-
ing this in mind, we synthesized methyl and ethyl esters (Table
6, 21a and 22a) of the parent compound (Table 5, 24j). We
observed that these esters have cellular potency with MICs
ranging from 4 to 64 ␮g/ml in addition to retaining a reason-
able potency on the enzyme. Ethyl ester 22b (Table 6) was a
weak inhibitor of the enzyme with an IC50 of 50 ␮M; however,
for the same compound, the cellular potency remained at 4
␮g/ml. This clearly indicated that the enzyme potency of the
acids correlated well with the cellular potency of the esters.
Similar trends were observed with compounds 21b and 21d
(Table 6). Compound 23a (Table 6) was screened against M.
bovis BCG strains with overexpressed PanK. MIC values in
wild-type M. tuberculosis H37Rv and in M. bovis BCG were sim-
ilar, whereas the BCG strain with increased PanK levels clearly
showed higher MIC values (Ͼ64 ␮g/ml), thus linking cellular po-
tency to inhibition of the target enzyme.
(ii) The aryl piperazine substitutions. An attempt to add chi-
rality by introducing a methyl group on the methylene between
the biaryl and the piperazine rings (24t) did not provide improve-
ments in potency. It was observed that the optimal potency could
be achieved with the 2=-pyridyl piperazines (24j). Any deviations
made by either changing the aza position to 4=-pyridyl piperazine
or substituting pyridyl with other aryl groups proved to be detri-
Crystallographic studies. Crystal structures of protein-inhib-
itor complexes of MtPanK with triazole and biaryl classes of com-
pounds have previously been described (11). The typical binding
modes of the triazole and biaryl compounds are exemplified by
compound 7i (PDB identification no. [ID] 4BFW) in Fig. 5A and
compound 24j (PDB ID 4BFY) in Fig. 5B, respectively. Despite
mental to enzyme inhibition (24r and 24s). In addition, enzyme differences in the mode of inhibition, compounds of the two
3316 aac.asm.org
Antimicrobial Agents and Chemotherapy

Elucidation of PanK Vulnerability in M. tuberculosis
TABLE 2 SAR of triazoles
classes share a binding site which overlaps that of pantothenate
and CoA (Fig. 5C).
Binding mode of the triazole compounds. Members of the
triazole class of compounds form a U-shaped conformation sim-
ilar to the shape of CoA, where the substituted phenyl rings are
roughly coplanar (Fig. 5C). One of the two nitrogens in the tria-
zole ring forms a hydrogen bond (HB) with the hydroxyl group of
Tyr235, whereas the other makes a HB with the amide side chain
of Asn277. The carbonyl group of the amide linker makes a HB
with a water molecule, and the trifluromethyl at the ortho position
of phenyl ring is in close proximity to His179. No HB interactions
were observed for the oxygen in the thioether side chain, and this
could explain the retention of activity with compounds lacking
oxygen in the thioether side chain (7a, 7b, 7e, and 7g). The
4-fluoro phenyl ring attached to the thioether linker binds in a
hydrophobic pocket composed of the side chains of Val199,
Ala100, Tyr235, Phe239, and Met242. The ring is also parallel to
the plane of the guanidinium group of Arg238, with the fluorine
atom above the plane (11).
We have not been able to cocrystallize the inhibitor-PanK
complex with ATP or ATP analogues. Therefore, we have com-
pared our enzyme-inhibitor complexes with the E. coli PanK
(EcPanK) structure in complex with the ATP analogue AMPPNP
(17) to help evaluate the effect of ATP on enzyme-inhibitor inter-
actions. The loss of potency in the presence of ATP (Table 2)
correlates with potential contacts of the R3 halogen substitution
with the gamma-phosphate of the ATP analogue (11). In our
modeling, 7m (the inhibitor showing the largest drop in potency)
has a fluorine-phosphate oxygen contact of 2.3 Å, while for 7i (the
least-affected triazole for which we have an enzyme-inhibitor
complex), the closest contact is increased to 3.2 Å. Note that the
difference between 7i and 7m is the R1 substitution (a large fluor-
benzyl-to-methyl change discussed in more detail below) and is
accommodated but results in small conformational changes in the
active site that affect the detailed set of interactions between the
inhibitor and enzyme. Removing the ortho substitution at R3 is
likely to result in a loss of potency in the absence of ATP but may
increase potency at higher ATP levels.
The structure of the enzyme-7i complex illustrates how substi-
tutions are accommodated at R1. The fluor-benzyl group fits into
a preformed tunnel into the active site that is not involved in the
binding of substrates or CoA. The tunnel surface is mostly hydro-
phobic and lined by side chains from four segments of the protein,
including Tyr182, Phe254, Tyr257, Ile272, and Ile276. The struc-
ture suggests that the increased potency of N-3-pyridyl methyl
substitutions arises from a hydrogen bond interaction to the hy-
droxyl group of Tyr257.
Three pairs of compounds (compounds 7a and 7c, compounds
7b and 7d, and compounds 7e and 7h) demonstrated the effect of
modifying the length of the R2 group. In all cases, the short-chain
4-fluoro thiobenzyl variant was the most potent. Most of our en-
zyme-inhibitor complexes had been determined with long-chain
variants. However, we had one complex where the inhibitor was a
4-fluoro benzyl variant of 7m. In this complex, the side chain of
Tyr182 had a different rotamer conformation, bringing it closer to
the 4-fluoro thiobenzyl group. The increased potency associated
with the 4-fluoro thiobenzyl variants was presumably a conse-
quence of these changes.
Binding mode of biaryl class. The mode of binding of the
biaryl class of compounds is exemplified by the crystal structures
June 2014 Volume 58 Number 6
aac.asm.org 3317

Reddy et al.
TABLE 3 SAR of quinolone amide
in complex with 24j and 25b (PDB ID 4BFY and 4BFZ), represent- group interacts with the guanidinium group of Arg274 from a
ing acid and amide functionality, respectively, at the ether linkage 2-fold related dimer. However, we have no evidence that the as-
(Fig. 5B). Except for these changes, the enzyme-inhibitor interac- sembly we observed in the crystal is also relevant in solution. The
tions are very similar. The para-nitrile phenyl group occupies the reduced potencies of ortho- and meta- substitutions are equally
hydrophobic tunnel formed by different segments of the enzyme. difficult to explain. The increase in potency observed by the intro-
One face of the ring packs against the side chain of residue Ile276, duction of a methyl group at the ortho position (24o) indicates an
while the other has no interactions with the surface residues of the interaction with a side chain of Ile272.
tunnel and would be exposed to solvent in the molecular dimer.
The lack of potency as a result of introducing a methyl group
The para-nitrile group is located 3.2 Å above the peptide plane on the methylene between the biaryl and the piperazine rings (24t)
connecting Met144 and His145. Whether these interactions are arises due to potential close contacts to the side chains of Ile272
sufficient to explain the increased potency of 24j relative to 24a and Ile276. One of the nitrogens of the piperazine ring makes
requires more-detailed computational analysis. In the crystal, water-mediated HBs with side chains of Tyr235 and Asn277,
however, we have a PanK dimer in the asymmetric unit that com- whereas the oxygen of the ether linkage makes a HB with the
bines with a crystallographic 2-fold axis to create an assembly with hydroxyl of Tyr182. The nitrogen of the pyridine ring makes a
222 symmetry. Within this assembly, the para-nitrile phenyl hydrogen bond to a water molecule. The introduction of a methyl
3318 aac.asm.org
Antimicrobial Agents and Chemotherapy

Elucidation of PanK Vulnerability in M. tuberculosis
TABLE 4 MIC modulation studies with the coaAKD strain
MIC (␮g/ml)a
animals (9). However, there are no reports on the vulnerability of
MtPanK to indicate the extent to which its levels need to be re-
duced to achieve bacterial growth inhibition.
A M. tuberculosis coaA conditional-expression strain was gen-
erated in which the coaA expression is regulated by the pristina-
mycin-inducible promoter. A set of PCRs confirmed the genotype
of the recombinant strain, i.e., the presence of a hygromycin se-
lection marker and the presence of a full-length coaA gene down-
stream of the pPTR promoter and the truncated gene downstream
of the wild-type promoter (data not shown).
The minimum inducer concentration required to support
growth of M. tuberculosis coaAKD was determined by plating for
survivors on 7H10 media supplemented with 0, 2, 10, 25, 50, and
100 ng/ml of P1. While the growth of M. tuberculosis H37Rv cul-
ture remained unaffected at these concentrations of P1 (data not
shown), the M. tuberculosis coaAKD strain could not grow on agar
plates unless supplemented with at least 10 ng/ml of P1 (Fig. 6),
thus confirming its essentiality in vitro. In order to understand the
vulnerability of MtPanK, a time-to-death study combined with
intracellular target level measurements was performed. Viability
was monitored through measurements of A600 and CFU (on agar
M.
coaAKD
with 10 ng/
ml P1
tuberculosis
H37Rv
coaAKD
without P1
MtPanK IC50
Compound (50ϫ Km
ID
ATP) (␮M)
D7
D14
D7
D14
D7
D14
16
7a
7b
7c
7g
7h
7i
7j
15a
15b
15d
15e
Isoniazid
0.08
0.18
0.19
0.28
0.33
0.68
0.90
0.21
0.60
1.67
3.19
ND
Ͼ64 Ͼ64 Ͼ64 Ͼ64 16
Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64 64
Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64 64
Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64
Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64
Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64
Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64
Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64
Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64
Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64 32
Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64 Ͼ64
0.03
0.03
0.03
0.03
0.03
0.06
a D7 and D14, incubation for 7 and 14 days, respectively, at 37°C; ND, not done.
group at the 3= position (24p) in our model would cause a close plates with 10 ng/ml P1) at regular intervals for up to 15 days.
contact to the ring of Tyr177 but may be better accommodated by Surprisingly, the M. tuberculosis coaAKD culture grew well in liq-
flipping the pyridine ring. The introduction of groups at the 2= uid cultures even in the absence of P1 (Fig. 7A). This was in con-
position (24r and 24s) would cause intramolecular close contacts. trast to our earlier observation, wherein no survivors could be
Changing the aza position to 4=-pyridyl (24d and 24q) produces a isolated in the absence of P1 on solid media. This growth trend
100-fold loss of potency. The HB interactions differ at the acid/ continued until the end of day 15, when all cultures were termi-
amide group substitution but have a minimal effect on potency. In nated due to appearance of clumps. This observation suggested
the case of 24j, the carboxylate group makes direct HB interactions that within the approximately 15 generations that the coaAKD
with side chains of Tyr182. In the 25b complex, the interaction culture went through in the absence of inducer, the PanK was not
with Arg238 is absent. The side chains of Asn277 and Tyr235, titrated down to levels incompatible with growth. In order to sub-
which are important for triazole binding, make only solvent-me- stantiate this hypothesis, the coaAKD culture was subjected to
diated hydrogen bonds with the biaryls.
dilutions in fresh media and incubated further to provide more
Comparison of the triazole and biaryl binding modes. The generations, which would help in reducing the protein levels.
triazole and biaryl compounds occupy similar binding sites However, these experiments were not conclusive owing to clump-
(Fig. 5D). The thioether and alkylamide side chains of the tria- ing of cultures within 48 h of dilution. Leaky expression of Mt-
zole scaffold overlap the ether and aryl-piperazine side chains PanK from the inducible promoter was ruled out by Western blot-
of the biaryl scaffold, respectively. The triazole extensions are ting of the culture samples (Fig. 7B). While hardly any MtPanK
longer and extend closer to the ATP binding site. The potential could be detected in culture without P1, the M. tuberculosis sigma
contact with the gamma-phosphate of ATP may explain why 70 (MtSigA) internal control could be detected to the same levels
the triazoles are competitive ATP inhibitors whereas the biaryl as seen with wild-type M. tuberculosis H37Rv (Fig. 7B). The results
compounds are not. The substituted benzyl group extending detailed above suggest that MtPanK is not a vulnerable target in
from the triazole core overlaps the para-nitrilephenyl group of vitro, as even a very small amount of the enzyme seemed to be
biaryls. Although both compound classes form extensive net- sufficient for bacterial growth.
works of hydrogen bond interactions, no PanK partner forms a
hydrogen bond with both inhibitors. A structurally conserved
water molecule, however, makes a HB interaction with either
the pyridine of the biaryls or the carbonyl oxygen of the amide
linker of the triazoles.
Target vulnerability studies. (i) In vitro vulnerability of
MtPanK. Despite having nanomolar potency on the enzyme, the
triazole class did not yield MICs for the wild-type M. tuberculosis
H37Rv strain, whereas the biaryls had MICs, although they were
severalfold higher than the IC50. This prompted us to investigate
the vulnerability of PanK both in vitro and in vivo using a MtPanK
knockdown strain.
M. tuberculosis possesses two isoforms of pantothenate kinase
encoded by the coaA and coaX genes. While there is essentiality
information available for both, coaX has been shown to be nones-
sential for the survival of M. tuberculosis in macrophages and in
FIG 4 Generic structure of biaryl acetic acid.
June 2014 Volume 58 Number 6
aac.asm.org 3319

Reddy et al.
TABLE 5 Illustration of SAR of biaryl acetic acids
(ii) In vivo essentiality of MtPanK. The phenotype of coaAKD ference in the growth rate compared to that of the control M.
M. tuberculosis strain in mice was studied in parallel with that of tuberculosis H37Rv strain, albeit at a statistically insignificant level
the M. tuberculosis H37Rv wild-type strain. BALB/c mice were (Fig. 7C), suggesting poor vulnerability of MtPanK in vivo as well.
infected via the intravenous route with the M. tuberculosis H37Rv
(iii) In silico-based vulnerability test of MtPanK. In this
strain or the M. tuberculosis coaAKD strain grown in the presence study, chemical vulnerability of MtPanK was modeled and evalu-
of 10 ng/ml P1. Owing to the poor pharmacokinetic properties of ated using the in silico M. tuberculosis platform. This model has the
P1 in mice (unpublished data), the course of infection had to be unique capability of predicting both the differential levels of effec-
monitored under inducer-depleted conditions only. The bacterial tiveness among the three types of inhibition (competitive, uncom-
burdens in lungs and spleen of infected mice were monitored over petitive, and noncompetitive) (16, 18) and the ensuing growth
150 days on plates supplemented with 50 ng/ml of P1. At the end of inhibition. Towards this goal, PanK inhibition and the resultant
the 150-day study, the coaAKD strain exhibited a very small dif- flux following different types of inhibition were simulated in the
3320 aac.asm.org
Antimicrobial Agents and Chemotherapy

Elucidation of PanK Vulnerability in M. tuberculosis
TABLE 6 SAR at the aryloxyacetic acid position of compound 8
platform and the [I]/Ki ratio was deduced with the aim of ranking ATP binding first and pantothenate next (19). Within the various
them based on analysis of the type which yields early growth ar- classes of inhibitors, in mixed-mode inhibition the inhibitor binds
rest. However, in order to simulate various kinds of inhibitions of sequentially to the free enzyme and to the enzyme-substrate com-
the reaction, it was important to know the M. tuberculosis PanK plex. Therefore, in this mode of inhibition, the inhibitor, when
reaction mechanism. In M. tuberculosis, PanK catalyzes a bisub- modeled in the platform, is both competitive and uncompetitive
strate reaction, the two substrates being pantothenate and ATP. In with ATP. Since such an inhibitor binds both forms of the enzyme,
the absence of any clear verification of the mode of the actual it could be more potent than those which target only one form of
reaction mechanism for PanK in M. tuberculosis, the mechanism the enzyme.
simulated in the in silico platform was as reported earlier for E. coli
In the current study, following simulation, the predicted [I]/Ki
(16) wherein this enzyme follows a sequential mechanism, with ratios that were required to reduce the CoA pathway flux to below
June 2014 Volume 58 Number 6
aac.asm.org 3321

Reddy et al.
FIG 5 (A) The typical binding mode of triazole class is represented by compound 7i (gray carbons) in the MtPanK active site (PDB ID 4BFW). Key residues (as
lines) involved in the interactions are shown in yellow. (B) Showing the typical binding mode of the biaryl class exemplified by compound 24j (green carbons)
and compound 25b (pink carbons) representing acid and amide functionality at ether linkage, respectively, in the MtPanK active site (PDB ID 4BFY and 4BFZ).
Key residues (as lines) involved in the interactions are also shown in the respective ligand carbon colors. In the case of the amide structure, Arg238 has a different
rotamer, whereas Tyr182 has shown a tendency of having two different rotamers. (C) Superposition of the active site of MtPanK crystal structures showing
triazole compound 7i (green), biaryl compound 24j (gray), phosphopantothenate (yellow, PDB ID 2ZSA), and CoA (cyan, PDB ID 2GES). (D) Superposition of
triazole as represented by compound 7i (yellow carbons, PDB ID 4BFW) and biaryl as represented by compound 24j (green carbons, PDB ID 4BFY) in the
MtPanK active site. The active-site residues (as lines) are also shown in the respective ligand carbon colors.
a critical value of 0.07 ␮mol/s (equivalent to the MIC) were found
to be 80, 12, 400, and 10,000 ␮M for competitive, mixed, noncom-
petitive, and uncompetitive modes of inhibition, respectively.
These observations demonstrated that mixed inhibition would be
the preferred mode that would require the lowest inhibitor con-
centration for an early biomass arrest (Fig. 8).
DISCUSSION
Although the essentiality of a gene is considered to be one of the
key attributes for choosing an antibacterial target, chemical
moieties against vulnerable targets are preferable in an anti-
infective discovery program. Vulnerability is defined as the ex-
tent of inhibition of a target required to have a negative impact
different concentrations of P1. Data are representative of the results of 3 indepen-
on growth, leading to cell death (20). As this parameter varies dent experiments. LOD, limit of detection; *, below the LOD.
FIG 6 Dependence of the M. tuberculosis coaAKD strain on P1. The conditional-
expression strain was grown with 100 ng/ml of P1 until mid-log phase. Washed
cells were diluted in fresh medium and plated on 7H10 plates supplemented with
3322 aac.asm.org
Antimicrobial Agents and Chemotherapy

Elucidation of PanK Vulnerability in M. tuberculosis
FIG 7 (A) Growth kinetics of M. tuberculosis coaAKD strain. A growth kinetics study was performed to assess the time required to achieve a bactericidal effect
in the absence of inducer P1. The M. tuberculosis coaAKD strain was grown with 10 ng/ml of P1 until an A600 of ϳ0.2 was reached. Harvested and washed cells were
used to inoculate cultures supplemented with 0, 10, or 25 ng/ml of P1. Culture growth was monitored by plating for survivors on plates containing P1. The data
are a representation of the results of two such experiments. (B) Effect of P1 concentration on MtPanK expression in the M. tuberculosis coaAKD strain. Protein
samples prepared from day 6 postexposure to P1 of the growth kinetics study were used for Western blot analyses. Two micrograms of total protein from the
lysates were loaded per lane, blotted, and probed with anti-MtPanK antibody (top panel) or MtSigA antibody (bottom panel). 0P1, 10P1, and 25P1 represent the
P1 concentrations (in ng/ml) used in the growth kinetics study. Rv data represent a sample from the M. tuberculosis H37Rv control. Std data represent His(6)-PanK
in the top panel and His(6)-SigA in the bottom panel. (C) In vivo phenotype of M. tuberculosis coaAKD strain. Mice were infected by the intravenous (i.v.) route
with the M. tuberculosis H37Rv strain and M. tuberculosis coaAKD strain with 3 mice per time point. The course of infection was monitored over 150 days
postinfection in the absence of P1 supplementation. Bacterial loads in lungs and spleen were measured by plating organ homogenates for CFU on 7H10 plates for
the control strain and 7H10 plates supplemented with 50 ng/ml of P1 for the conditional-expression strain.
from one target to another, the best target would be the one the vulnerability of a target in M. tuberculosis, one would have to
that requires minimal perturbation to kill the pathogen, monitor growth inhibition by either using compounds with dif-
whereas nonvulnerable targets may require more than 90% ferent modes of inhibition against the target or gradually reducing
inhibition to adversely affect growth (21–23). In order to assess the levels of target gene expression in the bacterial cell. However,
FIG 8 Mixed mode of inhibition of MtPanK and the resultant flux through the pathway. The predicted [I]/Ki ratio required to reduce the CoA pathway flux to
below a critical value of 0.07 ␮mol/s (indicated as the MIC) was the least for the mixed mode of inhibition (11).
June 2014 Volume 58 Number 6
aac.asm.org 3323

Reddy et al.
the latter is reflective of only one type of chemical inhibition and acid derivatives (esters) with IC50 Ͻ 1 ␮M had M. tuberculosis
hence is not appropriate for direct extrapolation of the effect of MICs ranging between 4 and 16 ␮g/ml. The mode of action of a
one to estimate the effect of the other. Additionally, a chemical few of these active esters was confirmed by the reduced cellular
inhibitor could bring in effects that are more pleiotropic (12, 23), activity against M. bovis BCG overexpressing MtPanK (MIC Ͼ 64
whereas a genetic knockdown is very specific to the intended tar- ␮g/ml). This observation helped to link the enzyme inhibition
get. An in silico platform would enable prediction of the extent of with the MIC and thus provided an additional proof of the chem-
target depletion or the type of inhibitor needed to evaluate novel ical validation of PanK as a possible drug target in M. tuberculosis.
targets whose vulnerability is not understood. From the simula-
All classes of MtPanK inhibitors discussed in this report could
tion profiles observed, it was evident that for MtPanK, a mixed be optimized to nanomolar potencies regardless of their mode of
mode of inhibition would most likely result in growth inhibition inhibition (IC50 Յ 50 nM when both substrates for MtPanK were
at a much lower inhibitor concentration.
maintained at Km). However, only inhibitors that did not have a
In the present study, MtPanK, an essential bacterial enzyme, competitive mechanism of inhibition had antimycobacterial ac-
was extensively evaluated as a putative target for the discovery tivity. This differential response was earlier demonstrated using in
of novel antituberculosis drugs. Results reported here indicate silico-based simulation studies in E. coli (16). In that report, for
that targets such as MtPanK are amenable to a range of bio- MtPanK, [I]/Ki was lowest with mixed noncompetitive inhibitors.
chemical and biophysical studies, thus offering opportunities Growth inhibition in the presence of a mixed noncompetitive
to invest in HTS to identify inhibitors with maximum struc- inhibitor was preceded by a steady decline in PanK levels to below
tural and mechanistic diversity, ATP competitive and ATP the critical level of 5% (16). Cross-species applicability of this E.
noncompetitive, which could possibly serve as start points for a coli platform was tested and confirmed by comparing predictions
lead identification program(s). PanK inhibition within M. tu- with the experimental data generated using a mycobacterial spe-
berculosis through the use of ATP competitors (triazoles) may re- cies (16). Since both the triazoles and the biaryl esters showed
quire very high potency to inhibit the enzyme inside the cell, since modulation of MICs in M. tuberculosis strains expressing different
the intracellular ATP concentration of around 1 mM (24) is in levels of PanK, a direct and specific link to target could be estab-
10-fold excess of its Km for ATP. Per the Cheng-Prusoff relation- lished.
ship (25) between IC50 and Ki, a competitive inhibitor may require
Unlike targets that are involved in essential housekeeping pro-
a much (Ͼ10-fold) higher concentration to achieve the same de- cesses, for targets such as PanK, for which there is no clinical
gree of inhibition, as obtained by a noncompetitive or uncompeti- precedent, an evaluation of their vulnerability is necessary. To this
tive inhibitor with an equivalent Ki value. Hence, IC50s were de- end, a parallel effort involving evaluation of M. tuberculosis
termined at 50ϫ Km of ATP to closely mimic the physiological coaAKD with regard to its growth kinetics in vitro and in vivo was
condition inside the cell.
In the case of triazoles and quinolones, even with potencies of on 7H10 plates with P1 and, for revertants, without P1.
about 100 nM and 200 nM, respectively, at 50ϫ Km of ATP, cel- The M. tuberculosis coaAKD strain demonstrated complete in-
carried out. In all of these experiments, the survivors were checked
lular inhibition could not be achieved. These results suggested that ducer dependence for growth on solid agar medium, thereby
either a more potent competitive inhibitor or an inhibitor with a establishing its essentiality. The in vitro survival kinetic studies
different mode of inhibition is required. However, these inhibi- in the absence of inducer indicated growth rates similar to
tors demonstrated growth inhibition of M. tuberculosis in cells those of wild-type strains as observed by CFU recovered from
with reduced MtPanK levels (coaAKD) along similar lines as re- the plates with P1. In the case of in vivo phenotyping experi-
ported earlier for another target from the same pathway (31). ments, however, small differences in growth rates were seen,
These data clearly indicate that with lowered expression levels, the albeit the differences was insignificant. This observation sug-
inhibitor is able to engage the target and bring about M. tubercu- gested either that the M. tuberculosis coaAKD strain had re-
losis growth inhibition, suggesting that MtPanK is poorly or verted through accumulation of genetic mutation or that there
weakly vulnerable. The observations reported here match earlier was leaky expression of MtPanK. The possibility of strain re-
reports in E. coli (26–28), where it was shown that an adverse effect version was ruled out, as the very same samples failed to grow
on bacterial growth required Ͼ95% enzyme inhibition against a on 7H10 plates without P1. With regard to the leaky expression,
few enzymes of the CoA biosynthetic pathway, sustained over sev- MtPanK could not be detected on day 6 postwithdrawal of
eral generations.
inducer in Western blots. However, the recovery of viable cells
Since ATP-competitive inhibitors did not demonstrate cel- from broth could be attributed to the presence of trace
lular activity against wild-type M. tuberculosis, as an alternative, amounts of MtPanK undetectable by Western blots. If such low
the mixed noncompetitive inhibitors (modified biaryl acids) were levels of MtPanK can keep the cells viable, the current studies
also evaluated and optimized for lead identification. While the indicate that a nearly 100% depletion of the target is required to
original hits inhibited the enzyme at a micromolar concentration, achieve growth inhibition. As M. tuberculosis is known to have a
the IC50 could be improved to 22 nM by systematic changes. True long generation time, a single cell may need to go through nearly
binding of these compounds was confirmed by midpoint (Tm) 30 generations to form a visible colony on solid medium, enough
shift experiments followed by elucidation of cocrystal structures for CoA to drop below critical levels.
with inhibitors of both triazole and biaryl classes (11). The crystal
According to earlier reports, a reduction in PanK activity of
structures not only gave information about key residues for bind- more than 95% was required to achieve growth inhibition since
ing of both classes of inhibitors but also hinted at the observed the steady-state levels of CoA are far in excess of what is critical
difference in the modes of inhibition. Free carboxylic acids with for cell survival (24, 27, 28). Results from the current study are
potent IC50s failed to show cellular activity, indicating the inability in concurrence with these earlier observations and were further
to permeate the M. tuberculosis cell wall. However, biaryl acetic substantiated by the inability to recover colonies on solid me-
3324 aac.asm.org
Antimicrobial Agents and Chemotherapy

Elucidation of PanK Vulnerability in M. tuberculosis
V, Mukherjee K, Sambandamurthy V, Sharma UK. 2010. Essentiality
and functional analysis of type I and type III pantothenate kinases of
Mycobacterium tuberculosis. Microbiology 156:2691–2701. http://dx.doi
.org/10.1099/mic.0.040717-0.
dium. However, the broth experiments could be conducted
only for a maximum of 15 days (for reasons of cell clumping),
during which the PanK levels may not have become depleted
below critical levels. In addition, under conditions where the
strain exists in a nearly nonreplicating phase in an animal
model, reduction in CoA levels could be very minimal and
hence would explain the recovery of M. tuberculosis even after
long periods in vivo. The vulnerability studies, both genetic and
chemical, reported here demonstrate that PanK is not an attractive
antimycobacterial target. This is similar to the data observed in
earlier studies with a ⌬panCD M. tuberculosis strain (29), wherein
the auxotrophic mutant persisted in the lungs and spleens of in-
fected mice for over 6 months following intravenous infection.
Since PanCD and PanK are both enzymes from the CoA biosyn-
thetic pathway, the observations from this study and the earlier
⌬panCD report suggest the possible poor vulnerability of targets
of this pathway.
10. Venkatraman J, Bhat J, Solapure SM, Sandesh J, Sarkar D, Aishwarya S,
Mukherjee K, Datta S, Malolanarasimhan K, Bandodkar B, Das K.
2012. Identification and characterization of mechanistically diverse inhib-
itors of the mycobacterial tuberculosis enzyme, pantothenate kinase
[CoaA]. J. Biomol. Screen. 17:293–302. http://dx.doi.org/10.1177/108705
7111423069.
11. Björkelid C, Bergfors T, Raichurkar AK, Mukherjee K, Malola-
narasimhan K, Bandodkar B, Jones TA. 2013. Structural and biochem-
ical characterization of compounds inhibiting Mycobacterium tuberculosis
PanK. J. Biol. Chem. 288:18260–18270. http://dx.doi.org/10.1074/jbc
.M113.476473.
12. Kaur P, Agarwal S, Datta S. 2009. Delineating bacteriostatic and bacte-
ricidal targets in Mycobacteria using IPTG inducible antisense expression.
PLoS One 4:e5923. http://dx.doi.org/10.1371/journal.pone.0005923.
13. Forti F, Costa A, Ghisotti D. 2009. Pristinamycin inducible gene regu-
lation in mycobacteria. J. Biotechnol. 140:270–277. http://dx.doi.org/10
.1016/j.jbiotec.2009.02.001.
14. Clinical and Laboratory Standards Institute (CLSI). 2011. Susceptibility
testing of mycobacteria, nocardiae, and other aerobic actinomycetes; ap-
proved standard—2nd ed. CLSI, Wayne, PA.
15. Collins L, Franzblau SG. 1997. Microplate alamar blue assay versus
BACTEC 460 system for high-throughput screening of compounds
against Mycobacterium tuberculosis and Mycobacterium avium. Antimi-
crob. Agents Chemother. 41:1004–1009.
16. Barve A, Gupta A, Solapure SM, Kumar A, Ramachandran V, Seshadri
K, Vali S, Datta S. 2010. A kinetic platform for in silico modelling of the
metabolic dynamics in Escherichia coli. Adv. Appl. Bioinform. Chem.
2010:1–14. http://dx.doi.org/10.2147/AABC.S14368.
17. Yun M, Park CG, Kim JY, Rock CO, Jackowski S, Park HW. 2000.
Structural basis for the feedback regulation of Escherichia coli pantothe-
nate kinase by coenzyme A. J. Biol. Chem. 275:28093–28099.
18. Ramachandran V, Singh R, Yang X, Tunduguru R, Mohapatra S,
Khandelwal S, Patel S, Datta S. 2013. Genetic and chemical knock-
down: a complementary strategy for evaluating an anti-infective target.
Adv. Appl. Bioinform. Chem. 6:1–13. http://dx.doi.org/10.2147
/AABC.S39198.
19. Song WJ, Jackowski S. 1994. Kinetics and regulation of pantothenate
kinase from Escherichia coli. J. Biol. Chem. 269:27051–27056.
20. Datta S. 1999. Understanding the characteristics of a good anti-infective
drug target. J. Parasit. Dis. 23:139–140.
21. Kohanski MA, Dwyer DJ, Hayete B, Lawrence CA, Collins JJ. 2007. A
common mechanism of cellular death induced by bactericidal antibiotics.
Cell 130:797–810. http://dx.doi.org/10.1016/j.cell.2007.06.049.
22. Jørgensen CM, Hammer K, Jensen PR, Martinussen J. 2004. Expression
of the pyrG gene determines the pool sizes of CTP and dCTP in Lactococcus
lactis. Eur. J. Biochem. 271:2438–2445. http://dx.doi.org/10.1111/j.1432
-1033.2004.04168.x.
23. Wei J, Krishnamoorthy V, Murphy K, Kim J, Schnappinger D, Alberd
T, Sassetti CM, Rhee KY, Rubin EJ. 2011. Depletion of antibiotic targets
has widely varying effects on growth. Proc. Natl. Acad. Sci. U. S. A. 108:
4176–4181. http://dx.doi.org/10.1073/pnas.1018301108.
24. Buchholz A, Takors R, Wandrey C. 2001. Quantification of intracel-
lular metabolites in Escherichia coli K12 using liquid chromatographic-
electrospray ionization tandem mass spectrometric techniques. Anal.
Biochem. 295:129–137. http://dx.doi.org/10.1006/abio.2001.5183.
25. Cheng Y, Prusoff WH. 1973. Relationship between the inhibition con-
stant (KI) and the concentration of inhibitor which causes 50 per cent
inhibition (I50) of an enzymatic reaction. Biochem. Pharmacol. 22:3099–
3108. http://dx.doi.org/10.1016/0006-2952(73)90196-2.
The results described here have thus unraveled the importance
of understanding the vulnerability of a novel target either through
the use of conditional-expression strains and/or via inhibitors
with diverse mechanistic activities before embarking on a drug
discovery program.
ACKNOWLEDGMENTS
We acknowledge the guidance and support provided by Tanjore Balga-
nesh and Santanu Datta for this study. Our sincere thanks to Debasmita
Sarkar for protein supply, Jateendranath Sandesh for analytical support,
and Suresh Solapure and Sunita DeSousa and Janani Venkatraman for
critical input.
We acknowledge support from the Foundation for Strategic Research,
the Swedish Research Council, and Uppsala University.
REFERENCES
1. WHO. 2012. Global tuberculosis report WHO. World Health Organiza-
tion, Geneva, Switzerland.
2. Raviglione M, Marais B, Lönnroth FK, Getahun H, Migliori GB, Har-
ries AD, Nunn P, Lienhardt C, Graham S, Chakaya J, Weyer K, Cole S,
Kaufmann SH, Zumla A. 2012. Scaling up interventions to achieve global
tuberculosis control: progress and new developments. Lancet 379:1902–
1913. http://dx.doi.org/10.1016/S0140-6736(12)60727-2.
3. Lönnroth K, Jaramillo E, Williams BG, Dye C, Raviglione M. 2009.
Drivers of tuberculosis epidemics: the role of risk factors and social
determinants. Soc. Sci. Med. 68:2240–2246. http://dx.doi.org/10.1016/j
.socscimed.2009.03.041.
4. Zumla A, Abubakar I, Raviglione M, Hoelscher M, Ditiu L, Mchugh
TD, Squire SB, Cox H, Ford N, McNerney R, Marais B, Grobusch M,
Lawn SD, Migliori GB, Mwaba P, O’Grady J, Pletschette M, Ramsay A,
Chakaya J, Schito M, Swaminathan S, Memish Z, Maeurer M, Rifat
Atun R. 2012. Drug-resistant tuberculosis - current dilemmas, unan-
swered questions, challenges, and priority needs. J. Infect. Dis. 205:S228–
S240. http://dx.doi.org/10.1093/infdis/jir858.
5. Spry C, Kirk K, Saliba KJ. 2008. Coenzyme biosynthesis, an antibacterial
drug target. FEMS Microbiol. Rev. 32:56–106. http://dx.doi.org/10.1111
/j.1574-6976.2007.00093.x.
6. Martinelli LK, Aldrich CC. 2012. Antimetabolite poisoning of cofactor
biosynthesis. Chem. Biol. 19:543–544. http://dx.doi.org/10.1016/j
.chembiol.2012.05.004.
26. Song WJ, Jackowski S. 1992. Cloning, sequencing, and expression of the
pantothenate kinase (coaA) gene of Escherichia coli. J. Bacteriol. 174:6411–
6417.
27. Zhang YM, Frank MW, Virga KG, Lee RE, Rock CO. 2004. Acyl carrier
protein is a cellular target for the antibacterial action of the pantothena-
mide class of pantothenate antimetabolites. J. Biol. Chem. 279:50969–
50975. http://dx.doi.org/10.1074/jbc.M409607200.
7. Gerdes SY, Scholle MD, D’Souza M, Bernal A, Baev MV, Farrell M,
Kurnasov OV, Daugherty MD, Mseeh F, Polanuyer BM, Campbell JW,
Anantha S, Shatalin KY, Chowdhury SA, Fonstein MY, Osterman AL.
2002. From genetic footprinting to antimicrobial drug targets: examples
in cofactor biosynthetic pathways. J. Bacteriol. 184:4555–4572. http://dx
.doi.org/10.1128/JB.184.16.4555-4572.2002.
8. Vallari DS, Jackowski S, Rock CO. 1987. Regulation of pantothenate
kinase by coenzyme A and its thioesters. J. Biol. Chem. 262:2468–2471.
9. Awasthy D, Ambady A, Bhat J, Sheikh G, Ravishankar S, Subbulakshmi
28. Christopherson RI, Duggleby RG. 1983. Metabolic resistance: the pro-
tection of enzymes against drugs which are tight-binding inhibitors by the
June 2014 Volume 58 Number 6
aac.asm.org 3325

Reddy et al.
accumulation of substrate. Eur. J. Biochem. 134:331–335. http://dx.doi 30. Leonardi R, Zhang YM, Charles OR, Jackowski S. 2005. Coenzyme A:
.org/10.1111/j.1432-1033.1983.tb07571.x.
29. Sambandamurthy VK, Derrick SC, Jalapathy KV, Chen B, Russell
back in action. Prog. Lipid Res. 44:125–153. http://dx.doi.org/10.1016/j
.plipres.2005.04.001.
RG, Morris SL, Jacobs WR, Jr. 2005. Long-term protection against 31. Abrahams GL, Kumar A, Savvi S, Hung AW, Wen S, Abell C, Barry CE,
tuberculosis following vaccination with a severely attenuated double
lysine and pantothenate auxotroph of Mycobacterium tuberculosis. In-
fect. Immun. 73:1196–1203. http://dx.doi.org/10.1128/IAI.73.2.1196
-1203.2005.
III, Sherman DR, Boshoff HI, Mizrahi V. 2012. Pathway-selective sen-
sitization of Mycobacterium tuberculosis for target-based whole-cell
screening. Chem. Biol. 19:844–854. http://dx.doi.org/10.1016/j.chembiol
.2012.05.020.
3326 aac.asm.org
Antimicrobial Agents and Chemotherapy