Discovery of a Novel Mycobacterial F-ATP Synthase Inhibitor and its Potency in Combination with Diarylquinolines

Article abstract of DOI:10.1002/anie.202002546

The F₁F_O-ATP synthase is required for growth and viability of Mycobacterium tuberculosis and is a validated clinical target. A mycobacterium-specific loop of the enzyme's rotary γ subunit plays a role in the coupling of ATP synthesis within the enzyme complex. We report the discovery of a novel antimycobacterial, termed GaMF1, that targets this γ subunit loop. Biochemical and NMR studies show that GaMF1 inhibits ATP synthase activity by binding to the loop. GaMF1 is bactericidal and is active against multidrug- as well as bedaquiline-resistant strains. Chemistry efforts on the scaffold revealed a dynamic structure activity relationship and delivered analogues with nanomolar potencies. Combining GaMF1 with bedaquiline or novel diarylquinoline analogues showed potentiation without inducing genotoxicity or phenotypic changes in a human embryonic stem cell reporter assay. These results suggest that GaMF1 presents an attractive lead for the discovery of a novel class of anti-tuberculosis F-ATP synthase inhibitors.

Full text of DOI:10.1002/anie.202002546

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Discovery of a novel Mycobacterial F-ATP synthase inhibitor and
its potency in combination with diarylquinolines
Authors: Adam Hotra, Priya Ragunathan, Pearly Shuyi Ng,
Pattarakiat Seankongsuk, Amaravadhi Harikishore, Jickky
Palmae Sarathy, Saw Wuan Geok, Umayal Lakshmanan,
Patcharaporn Sae-Lao, Nitin Pal Kalia, Joon Shin,
Revathy Kalyanasundaram, Sivaraj Anbarasu, Krupakar
Parthasarathy, Chaudhari Namrata Pradeep, Harshyaa
Makhija, Peter Dröge, Anders Poulsen, Jocelyn Hui Ling Tan,
Kevin Pethe, Thomas Dick, Roderick W. Bates, and Gerhard
Grüber
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202002546
Link to VoR: https://doi.org/10.1002/anie.202002546

10.1002/anie.202002546
Angewandte Chemie International Edition
RESEARCH ARTICLE
Discovery of a novel Mycobacterial F-ATP synthase inhibitor and
its potency in combination with diarylquinolines
Adam Hotra1,2,3, Priya Ragunathan2, Pearly Shuyi Ng4, Pattarakiat Seankongsuk1, Amaravadhi
Harikishore1,2, Jickky Palmae Sarathy5, Wuan-Geok Saw2, Umayal Lakshmanan4, Patcharaporn Sae-
Lao1, Nitin Pal Kalia6, Joon Shin2, Revathy Kalyanasundaram7, Sivaraj Anbarasu7, Krupakar
Parthasarathy7, Chaudhari Namrata Pradeep2, Harshyaa Makhija2, Peter Dröge2, Anders Poulsen4,
Jocelyn Hui Ling Tan4, Kevin Pethe2,6, Thomas Dick5,8,9, Roderick W. Bates1,*, and Gerhard Grüber2,*
disease.[1] During infection, Mtb inhabits a wide range of
Abstract: The F1FO-ATP synthase is required for growth and
intracellular and extracellular environments. Many of these
environments are highly dynamic, and mycobacteria face with the
viability of Mycobacterium tuberculosis. Bedaquiline validated this
target clinically. We showed recently that a mycobacterium-specific
constant challenge of redirecting their metabolic activity
commensurate with either growth or non-replicating
persistence.[2] A fundamental feature in this adaptation is the
ability of mycobacteria to respire, regenerate reducing
equivalents, and generate the biological energy currency
adenosine triphosphate (ATP) via oxidative phosphorylation.
During the latter, an electrochemical gradient of protons is
generated by the electron transport chain (ETC; Fig. 1A). The
electrochemical gradient drives the energy converter F1FO ATP
synthase and is required for viability of growing and non-growing
mycobacteria.[2] Phenotypic screening identified the first-in-class
diarylquinoline drug bedaquiline (BDQ, Sirturo®) that inhibits the
mycobacterial F-ATP synthase (Fig. 1A-B).[3] However, the
loop of the enzyme’s rotary γ subunit plays a role in coupling of ATP
synthesis within the enzyme complex. Here, we report the discovery
of a novel antimycobacterial, termed GaMF1, targeting this γ subunit
loop. Biochemical and Nuclear Magnetic Resonance studies show
that GaMF1 inhibits ATP synthase activity by binding to the loop.
GaMF1 is bactericidal and is active against multidrug- as well as
Bedaquiline-resistant strains. Chemistry efforts on the scaffold
revealed a dynamic structure activity relationship and delivered
analogues with nanomolar potencies. Combining GaMF1 with
bedaquiline or novel diarylquinoline analogues showed potentiation
without inducing genotoxicity or phenotypic changes in a human
embryonic stem cell reporter assay. These results suggest that
GaMF1 presents an attractive lead for the discovery of a novel class
of anti-tuberculosis F-ATP synthase inhibitors.
successful advance of this drug has been overshadowed by the
[4]
development of clinical resistance
along with a very long
1School of Physical and Mathematical Sciences, Nanyang
Technological University, 21 Nanyang Link, Singapore 637371,
Republic of Singapore; E-mail: roderick@ntu.edu.sg
2School of Biological Sciences, Nanyang Technological University, 60
Nanyang Drive, Singapore 637551, Republic of Singapore; E-mail:
ggrueber@ntu.edu.sg
terminal half-life leading to concerns regarding tissue
accumulation.[5] Recent medicinal chemistry campaigns resulted
in a new generation of 3,5-dialkoxypyridine (DARQ) analogues of
BDQ that have the potential to address these issues.[6] TBAJ-587
and -876 are new development compounds of this series (Fig.
1B), showing improved physio-chemical properties.[6] The
3Nanyang Institute of Technology in Health and Medicine,
Interdisciplinary Graduate School, Nanyang Technological University,
Republic of Singapore
discovery of squaramides (SQA) [7], compound 5228485 [8], and
[9]
epigallocatechin gallate (EGCG)
as potent ATP synthesis
inhibitors underlined the relevance of the mycobacterial F-ATP
synthase as a drug target.
4Experimental Drug Development Centre, Agency for Science
Technology and Research, A*STAR, 10 Biopolis Road, Singapore
138670
Nevertheless, infections with MDR- and XDR Mtb strains and
coinfection with HIV have worsened the prospect of controlling
TB.[10] In addition, the prolonged treatment regimen for six months
with a combination of four drugs has resulted in poor patient
compliance, leading to the emergence of MDR and XDR strains
of Mtb.[11] Therefore, in order to improve the effectiveness of TB
treatment and to improve control of MDR and XDR TB, new
antitubercular targets and agents are required. Anti-TB targets
should be a) essential for viability in Mtb, b) novel to avoid of cross
resistance with existing drugs, c) absent in humans to avoid
mechanism-based toxicity issues.
5Department of Microbiology and Immunology, Yong Loo Lin School
of Medicine, National University of Singapore, 14 Medical Drive,
Singapore 117599, Republic of Singapore
6Lee Kong Chian School of Medicine, Nanyang Technological
University, Experimental Medicine Building, Singapore
7Centre for Drug Discovery and Development, Sathyabama Institute
of Science and Technology, Chennai 600119, India
8Center for Discovery and Innovation, Hackensack Meridian Health,
340 Kingsland Street, Nutley, NJ 07110, USA
9Center for Discovery and Innovation, Hackensack Meridian Health,
340 Kingsland Street, Nutley, NJ 07110, USA
The F1FO ATP synthase is essential for growth and viability
in mycobacteria.[12] In contrast, in other prokaryotes, the enzyme
is dispensable for growth on fermentable carbon sources.[13] The
mycobacterial enzyme is composed of a membrane-embedded
FO complex (subunits a:b:b’:c9), responsible for H+-translocation
from the periplasmic space to the cytoplasmic side, and the F1
part (subunits α3:β3:γ:ε) [14] which uses the proton-motive force to
synthesize ATP within the three catalytic αβ-pairs (Fig. 1A). This
catalytic α3β3-headpiece is linked with the FO part via the two
rotating central stalk subunits γ and ε, as well as the peripheral
Introduction
Tuberculosis (TB) is a devastating disease that results in 1.4
million deaths every year.[1] Multidrug resistant- (MDR) and
extensively drug-resistant (XDR) bacterial strains of
Mycobacterium tuberculosis (Mtb), which causes TB, are
spreading worldwide, causing major global health issues.
Furthermore, about one third of the world population is latently
infected by Koch’s bacillus and susceptible to develop the
stalk subunits b, b’ and δ.[14c, 14e]
.
We recently demonstrated the specific role of the unique loop
of mycobacterial subunit γ (γ166-179-loop; M. smegmatis
1
This article is protected by copyright. All rights reserved.

10.1002/anie.202002546
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 1. (A) Enzyme composition and inhibitors of the electron transport chain and F-ATP synthase of Mtb. (B) Structures of BDQ, TBAJ-587 and TBAJ-876.
numbering) effecting ATP hydrolysis, -driven H+-pumping and -
synthesis.[15] As the mycobacterial γ-loop is unique, absent in the
human homologue and other prokaryotes, as well as important for
the regulation of ATP synthesis, -hydrolysis and proton pumping,
it fulfils fundamental criteria as a potential TB-drug target.
We here conducted an in silico compound screening
of BDQ and other DARQ analogues, opening a new avenue for
an efficient multi-drug combination. In addition, initial structure-
activity relationship studies (SAR) revealed a path to optimize the
potency of the series.
Results and Discussion
exercise based on our structural and biochemical data of the
[15, 16]
mycobacterial enzyme F-ATP synthase
that led to the
Homology and Pharmacophore Modelling
discovery of the mycobacterial F-ATP synthase inhibitor GaMF1.
The variety of biochemical assays, the use of a BDQ-resistant F-
ATP synthase mutant and structural experiments validated that
GaMF1 exerts its anti-mycobacterial activity through inhibition of
the F-ATP synthase. Furthermore, GaMF1 increased the potency
The γ166-179 loop was exploited as a binding epitope for de
[15, 16]
novo drug design.
As high-resolution crystallographic
structures were lacking when the project started, homology
Figure 2. Receptor-based pharmacophore modeling and pharmacophore mapping. (A) Generation of Ludi-based interaction sites using acceptor, donor, and
hydrophobic features as probes at γ166-179 loop-c ring interface residues. (B) Five features comprising of one hydrogen bond acceptor (green sphered arrows) in
vicinity of R172 sidechain atoms, one hydrogen bond donor targeting Q44 residue of the c-ring (magenta sphered arrows), three hydrophobic features (cyan spheres)
in the vicinity of P74, Y209, R172 residues were included in the receptor based pharmacophore model.
2
This article is protected by copyright. All rights reserved.

10.1002/anie.202002546
Angewandte Chemie International Edition
RESEARCH ARTICLE
modelling enabled us to build 3D structures with good accuracy
(see Supplementary Information). Our receptor-based pharmaco-
phore modelling studies highlighted the possible interaction sites
at the γ166-179 loop interface (Fig. 2A). Mainly, one hydrogen
bond acceptor, one hydrogen bond donor and three hydrophobic
features were included in the pharmacophore model at the γ166-
179 loop-c ring interface (Fig. 2B). Further, the exclusion volumes
constraints around protein atoms near the ligand (not shown for
clarity) were also incorporated to delineate the shape of the
binding cavity as well as to prevent clashes of ligand with protein
atoms. Database screening with a 5-feature mapping resulted in
library of about 30,000 molecules. This library was further refined
by a series of standard precision, extra precision docking
protocols with glide score, XP scores and consensus scoring
themes. Together with visual inspection of molecular interactions
at the γ166-179 loop and c-ring interface resulted in 81 molecules
for experimental screening.
Antimycobacterial activity and Minimum inhibitory
concentration determination
We identified ligands that bear complementarity to the γ166-
179 loop and c-ring interface and provided a focused library of
which GaMF1 (Fig. 3A) exhibited a minimal inhibitory
concentration (MIC50) of 11 μM in the fast growing M. smegmatis
mc2 155 strain, 17 μM in M. bovis bacillus Calmette–Guérin (BCG),
and 33 μM in Mtb H37Rv (Fig. 3B). A synthesis protocol for
GaMF1 was established (see Supplementary Information:
Synthesis of GaMF1) to produce the compound in bulk to study
the bactericidal activity of the compound and its mechanism of
action. GaMF1 was bactericidal against M. smegmatis mc2 155
at 20-fold its MIC50, indicated by the decreasing cell number of
about 99% in 3 days (Fig. 3C) In contrast, BDQ had delayed
bactericidal activity, as reported before [17], followed by a reduction
in bacterial viability (Fig. 3C). GaMF1 was also bactericidal
Figure 3: (A) Structure of GaMF1. (B) Growth inhibition of M. smegmatis (blue), M bovis BCG (green) and Mtb (red) by GaMF1 with MIC50 of 11, 17 and 33 µM,
respectively. (C) Initial five days of GaMF1-, BDQ kill kinetics against M. smegmatis mc2 155. The bacteria were grown in liquid culture (LBT) in the presence of the
indicated concentrations of GaMF1 and BDQ. The experiments were repeated twice, and the profiles were identical. (D) Images (a, b), HR resistant Mtb colonies
observed after day 7 incubation in no-compound containing Middlebrook 7H9 broth (4.188 log10 CFU/ml); Image (c), no colonies of HR resistant Mtb observed after
day 7 incubation in compound GaMF1 containing Middlebrook 7H9 broth. (E) Antimycobacterial activity of GaMF1 against Mtb isolate resistant to Isoniazid (H) and
Rifampicin (R) as determined by Colony forming units (CFU) based method in Middlebrook 7H11 agar plates.
against three drug resistant clinical isolates (isoniazid (H),
rifampicin (R) monoresistant- and HR-resistant Mtb strains),
causing a significant reduction in bacterial viability in 3 to 7 days
(Fig. 3D- E). Preliminary in vitro ADME profiling revealed that that
is in line with previous reports [15, 18], and an IC50 of 7.1 ± 1.3 nM
M. bovis BCG (Fig. 4B). In comparison, GaMF1 revealed a potent
ATP synthesis inhibition at an IC50 of 0.5 ± 0.1 μM in M.
smegmatis IMVs (Fig. 4A) and 5.2 ± 1.1 μM in M. bovis BCG (Fig.
4B). GaMF1 inhibited ATP synthesis of M. smegmatis (Fig. 4A)
and M. bovis BCG (Fig. 4B) IMVs in the presence of succinate,
indicating that GaMF1 does not interfere with the NADH-
dehydrogenases. In parallel, a possible effect of GaMF1 on
oxygen consumption and thereby the cytochrome oxidases of the
ETC was tested using methylene blue as an oxygen probe. As
visualized for the fast- and slow grower M. smegmatis (Fig. 4C)
and M. bovis BCG (Fig. 4D), oxygen consumption was unaffected
in both strains at µM concentrations of GaMF1. In comparison,
nM- and µM concentrations of BDQ revealed a slight and
moderate effect on oxygen respiration in both strains (Fig 4C-D),
reflecting that BDQ does also affect oxygen consumption as
described by Kalia et al. [19]
GaMF1, with a clogP value of 4.37, is less lipophilic than BDQ [6c]
and has a good metabolic stability in mouse liver microsomes (T
,
½
with or without NADPH of 47.5 min and 68 min, respectively; Clint
of 14.6 µl/min/mg protein (Supplementary Table 1A-B)).
Insights into the inhibition of mycobacterial F-ATP synthase
by GaMF1
To confirm that GaMF1 exerts its anti-mycobacterial activity
via inhibition of the mycobacterial F-ATP synthase, ATP synthesis
was measured using inverted membrane vesicles (IMVs) of M.
smegmatis and M. bovis BCG. Figure 4A shows a typical profile
of M. smegmatis IMVs ATP synthesis inhibition by BDQ with a
half-maximal inhibitory concentration (IC50) of 1.1 ± 0.2 nM, which
3
This article is protected by copyright. All rights reserved.

10.1002/anie.202002546
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 4: Inhibition of ATP synthesis by GaMF1 in IMVs from M. smegmatis (A) and M. bovis (B) using the electron donors NADH (dark green) or succinate (light
green). BDQ was used as a control BDQ (black/grey). (C-D) Test of possible effects of GaMF1 and BDQ on oxygen consumption in M. smegmatis IMVs (C) and M.
bovis BCG cells (D). BDQ shows partial inhibition of oxygen consumption at nanomolar range (500 nM) and significant inhibition at micro-molar level (6 μM). In
comparison, GaMF1 did not inhibit oxygen consumption even at high micro-molar level (6- or 11 μM). Thioridazine (THZ) was used as positive control and 7H9
media /drug free (DF) served as a blank/negative control (C). GaMF1 did also not affect oxygen consumption in the slow grower M. bovis over a 96-h period, while
BDQ at 500 nM completely inhibited oxygen consumption (D). (E) Inhibition of ATP synthesis by BDQ (black) and GaMF1 (green) on M. smegmatis I66M c subunit
mutant IMVs with NADH as substrate. The M. smegmatis I66M mutant IMVs (●) reveal a 10-fold sensitivity compared to WT (▲), while the inhibitory profile of
GaMF1 on M. smegmatis I66M mutant IMVs (♦) did not alter. (F) Additive effect of GaMF1 and BDQ (black) on the ATP synthesis inhibition in IMV’s of M. smegmatis
I66M mutant. In the I66M mutant, BDQ cannot fully inhibit the ATP synthesis. However, when GaMF1 (green) was added in a concentration dependent manner,
ATP synthesis significantly decreased to almost full inhibition.
We next tested the potency of GaMF1 against an F-ATP
synthase bearing an I66M substitution in the M. smegmatis
subunit c, which is associated with resistance to BDQ. [20] While
the mutant displayed a 10-fold shift in sensitivity to BDQ (Fig. 4E),
the mutation did not affect the potency of GaMF1 (Fig. 4E).
Interestingly, addition of GaMF1 reduced ATP synthesis of the
I66M mutant IMVs significantly in a concentration dependent
manner (Fig. 4F), demonstrating that the novel F-ATP synthase
inhibitor GaMF1 shows no cross resistance to BDQ.
peptide, which according to its NMR solution structure, forms a
loop of polar residues with a surface distribution of negatively
charged residues on one side and an amphiphilic charge
distribution on the other side of the loop.[16] The chemical shift
perturbations (CSPs) for NH and Hα between free peptide and the
peptide-GaMF1 complex derived from the 2D-TOCSY spectra of
the Mtγ166-179-loop with GaMF1 suggest that the amino acids
N167, G168, E169, R172, and S173 became perturbed upon
ligand binding (Fig. 5A-B). The data reflect the involvement of
γ166-179-loop residues in GaMF1-binding.
To obtain further insights into the molecular interactions of
GaMF1, we docked the ligand into subunit Mtγ using the NMR
perturbed residues as ambiguous interaction restraints (AIR) or
active site residues for ligand docking (Fig. 5C). GaMF1
Target specificity of GaMF1
The in vitro assay data revealed that GaMF1 inhibits ATP
synthesis and the growth of M. smegmatis and M. bovis BCG
strains. Next, we performed NMR titration experiments to assess
whether GaMF1 does bind to the M. tuberculosis γ166-179
comprises of
a 3-bromo-benzoyl moiety (ring A), a 1,4
diaminobenzene moiety (ring B) and a substituted pyrimidine (ring
C) and was predicted to bind to a region in the vicinity of the Mtγ-
4
This article is protected by copyright. All rights reserved.

10.1002/anie.202002546
Angewandte Chemie International Edition
RESEARCH ARTICLE
loop residues with a glide score of 5.58 (and a GOLD PLP fitness
of 46.5). Both ring A and -B of GaMF1 are involved in interactions
with the γ166-179 loop residues. On one hand, the carbonyl and
the amide atoms on ring A were engaged in close contacts with
γQ171 side-chain amide atoms and γD64 main-chain amide
atoms. On the other hand, the 3-bromo benzene group on ring A
was also involved in π-alkyl interactions with amino acids γH65
and alkyl atoms of γ R172 (highlighted by pink lines). This moiety
is positioned in the vicinity of the c-ring residues such as cR52,
cQ46, cP47, cE48 (beige coloured) via close contacts, and
thereby chaperone the stability of ligand binding at the Mtγ-c
interface. The phenyl-ring (ring B) and pyrimidine (ring C) of
GaMF1 were engaged in π-anion interactions (orange arrows)
with γE169 and γE70 residues, respectively. The 6-methyl on ring
C maintained aliphatic hydrophobic interactions with γI214 (pink
line, (α2 / β5 loop, light red)). In addition, the amide atoms which
link the 3-N-ethyl fragment on ring C mediate the hydrogen
bonding interactions with γR71 residues from α1/β1 loop (blue),
while the N-ethyl fragment was engaged in van der Waal’s
contacts with γP72 (light green line) and γD174 residues (Fig.
5C).
respectively (Fig. 5D). While the antibiotic Moxiflocaxin, which is
active against both Gram-positive and Gram-negative bacteria by
inhibiting DNA gyrase, inhibits both the growth of S. aureus and
E. coli, no growth inhibition of the two bacteria strains was
observed in the presence of increasing concentration of GaMF1.
Since the extra γ-loop is specific to mycobacterial F-ATP
synthases, GaMF1 is not expected to bind to the E. coli or S.
aureus F-ATP synthase and consequently should not inhibit the
major energy converter of these prokaryotic cells. NMR titration
experiments and docking studies are consistent with a direct
interaction between GaMF1 and the subunit Mtγ. However, we
cannot exclude the possibility that GaMF1 may interfere with
additional targets in the pathogen. Future studies could focus on
the selection of escape mutants resistant to GaMF1 to confirm
target engagement genetically, or alternatively to identify indirect
mechanism of resistance such as those described for BDQ of
clofazimine [21]
.
GaMF1 synergizes with BDQ and TBAJ-876 analogs
The in vitro ATP synthesis experiments of the I66M mutant
IMVs and GaMF1 indicated a synergistic effect but no cross
resistance to the diarylquinoline BDQ (Fig. 4F). To assess a
synergistic effect of both compounds in more detail, the effect of
drug combinations on the growth inhibition of M. smegmatis was
examined. BDQ inhibited growth at 31.3 nM. However, in
combination with GaMF1, the IC50 of BDQ reduced with the
concentration of GaMF1 added, thus confirming a synergistic
effect of both compounds (Fig. 6A). Next, the synergistic effect on
ATP synthesis was studied in M. bovis BCG. In the whole cell ATP
synthesis assay, an additive effect of BDQ and GaMF1 was
observed (Fig. 6B). BDQ alone at 200 nM reduced ATP synthesis
from 3.7x104 RLU to 8x103 RLU. After addition of GaMF1, ATP
formation dropped in a concentration dependent manner of
GaMF1 and lowered the IC50 of BDQ significantly (Fig. 6B).
Most recently, diarylquinoline analogs with improved potency
and physio-chemical properties have been described. [6a-6c] These
improved analogues include 5307, 5316, and 5377 [6c] that interact
with the subunit c-ring and subunit ε of the mycobacterial F-ATP
[6d]
synthase
(Fig. 6C). Using a modified synthesis protocol (see
Supplementary Information) the racemates of 5307 ((±)-5307),
5316 ((±)-5316), and 5366 ((±)-5366) were synthesized (CCDC
deposition numbers: 1967685, 1967683, 1967684, respectively),
which inhibited M. smegmatis mc2 155 growth at an MIC50 of 1.6
nM, 4.4 nM and 1.0 nM, respectively (Fig. 6D). The improved
MIC50 values of the racemates compared to BDQ were in line with
improved IC50 in ATP synthesis inhibition of M. smegmatis mc2
155 IMVs with values of 0.1 ± 0.01 nM, 0.2 ± 0.08 nM, and 0.6 ±
0.33 nM for the racemates 5307, 5316, and 5366, respectively
(Fig. 6E). Importantly, combining (±)-5307 with 300 nM of GaMF1
revealed a significant reduction in cell growth (Fig. 6F),
demonstrating that the potency of the already improved DARQ
derivative increased by the synergistic effect of GaMF1. These
results open a new avenue for the development of a novel
combinatory drug regimen. Interestingly, the combination of 300
nM GaMF1 with either (±)-5316 or (±)-5366 did not cause a
synergistic effect, reflecting a specific combinatorial potency of
(±)-5307 and GaMF1.
Figure 5: (A) 2D-TOCSY NMR spectral data of the free Mtb γ166-179 peptide
and complexed with GaMF1 highlights the chemical shift changes for NH
resonances (absolute values of CSP) between free and complex peptide. (B)
Absolute values of CSP for HA resonances between free and the GaMF1-bound
γ166-179 peptide. (C) Predicted binding interaction of GaMF1 on the
mycobacterial γ-c ring interface. The three loop regions of β1/α1 (blue), β5/α2
(light red) and β4/α4 (magenta) and together with residues of the c-ring loops
(beige) form the ligand binding pocket. (D) A screen of GaMF1 against S. aureus
and E. coli, as representatives of the human microbiome, confirm the specificity
of GaMF1 for Mtb as well as that the compound does not affect the human
microbiome.
Viability and pluripotency of human embryonic stem cells
remain unaffected by BDQ, (±)-5307 and GaMF1
Human stem cell technologies are becoming increasingly
important as valuable ex vivo test models during the drug
discovery process (reviewed in [22]). We therefore decided to
employ a human embryonic stem cell (hESC) line (E3) that we
have recently genetically engineered into a highly sensitive
pluripotency reporter [23], and examined both potential drug-
induced genotoxicity and perturbations of the hESC
transcriptional program that lead to hESC differentiation. The
assays revealed that neither the tested compounds (±)-5307, (±)-
To confirm its target specificity and to demonstrate that
GaMF1 is not a broad spectrum antibiotic, the compound was
tested against Staphylococcus aureus and Escherichia coli as
representatives of Gram-positive and Gram-negative bacteria,
5
This article is protected by copyright. All rights reserved.

10.1002/anie.202002546
Angewandte Chemie International Edition
RESEARCH ARTICLE
5307 in combination with 300 nM GaMF1 nor the control BDQ
elicited substantial genotoxic effects or induced major alterations
in the global transcriptional program (Supplementary Information;
Supplementary Fig. 2A-G).
Figure 6: (A) Synergistic effect of the TB drug BDQ and GaMF1 on cell growth of M. smegmatis. Combination of different concentrations of BDQ and GaMF1 was
done by two subsequent serial dilutions in the 96 well plate containing 100 µl of 7H9 media. Suspension of M. smegmatis mc2 155 in logarithmic growth phase was
added to obtain starting OD600 = 0.05 in 200 µl. Plates were incubated 24 h in 37 °C. (B) Whole cell ATP synthesis assay with M. bovis BCG in the presence of BDQ
and GaMF1. (C) Structures of (±)-5307, (±)-5316, and (±)-5366. (D) ATP synthesis inhibition of M. smegmatis IMVs by (±)-5307, (±)-5316, and (±)-5366, respectively.
(E) M. smegmatis growth inhibition affected by (±)-5307 (●) and in combination with 300 nM of GaMF1 (□).
SAR studies of GaMF1 series
In order to improve the efficacy of GaMF1, a series of
analogs was synthesized (see Supplementary Information) and
tested in mycobacterial ATP synthesis assay (IC50) and in the
mycobacterial assay (M. smegmatis and/or Mtb MIC50). Key SAR
findings are summarized in Table 1.
It was established that ring C was essential for maintaining
activity as truncated analog 1 (Table 1) did not inhibit ATP
synthesis (IC50 > 100 μM). The amide functionality linking rings A
and B was essential for activity as switching to the reverse amide
2 led to a loss in activity. This is consistent with the docking pose
presented previously (Fig. 5C), where the carbonyl and amide
atoms on ring A were predicted to be in close contacts with γQ171
and γD64.
While the removal of substitution on the phenyl ring (3) led to
a modest drop in activity, replacement of the 3-bromo with either
electron-withdrawing CF3 (4) or electron-donating pyrrolidine (5)
substituents gave compounds with comparable potencies. This
result suggests that lipophilic substitutions are preferred in this
Figure 7: Inhibition of ATP synthesis by GaMF1 analog 8 in IMVs from M.
part of the molecule and that the electronics of the aromatic ring
smegmatis with an IC50 of 90 nM on M. smegmatis (A) and 8.7 nM on M. bovis
have little contribution towards potency. This observation is also
consistent with the docking pose (Fig. 5C), which shows (i) space
to tolerate substituents on the phenyl ring and (ii) no key
interactions between the neighbouring residues and the phenyl
ring. Hence, we concluded that the region around phenyl ring A
would perhaps be the most amendable to structural modifications.
BCG IMVs (B). (C) 300 nM of GaMF1 analog 8 increase the potency of (±)-5307
in ATP synthesis inhibition (∆) when compared to the inhibitory effect of (±)-
5307 alone (▲). (D) Addition of 500 nM (▲) and 9 μM (■) of GaMF1 analog 8
shifted the inhibition curve to the left, showing a 2-fold (0.7 nM) and 10-fold (0.17
nM) decrease in MIC50 thereby revealing the synergistic effect on the same
target enzyme. The effect of racemate ((±)-5307) on growth of M. smegmatis
mc2 155 is shown in curve (●).
6
This article is protected by copyright. All rights reserved.

10.1002/anie.202002546
Angewandte Chemie International Edition
RESEARCH ARTICLE
Table 1: SAR of selected GaMF1 analogs
nM on M. bovis BCG IMVs (Fig. 7B) and inhibited Mtb growth
culture at a concentration of 3 µM, while the m-chloro phenyl
derivative 7 maintained an activity profile similar to the parent
molecule. Interestingly, when 300 nM of analog 8 was combined
in an ATP synthesis assay of M. smegmatis with (±)-5307 (Fig.
7C) a clear synergistic effect could be observed, reflecting the
potency of this unique TB-compound combination. Addition of
GaMF1 analog 8 shifted the growth curve to the left, thereby
revealing the synergistic effect on the same target enzyme (Fig.
7D).
IC50
MIC50
MIC50
Cpd No.
Structure
(μm)a
(μM)b
(μM)c
0.5
11
33
GaMF1
1
2
>100
>100
>100
-
-
Conclusion
Respiration is a promising target for the development of new
antimycobacterial agents, including the inhibitors clofazimine,
Telacebec (Q203), and BDQ which inhibit the mycobacterial
complex II (Ndh2; [24]), the cytochrome bc1 complex [25], and the
F-ATP synthase.[3] The latter was approved by FDA for the
treatment of MDR-TB in 2012. [4a] An increased understanding of
the function and structure of different respiratory complexes and
F-ATP synthase in mycobacterial biology is becoming key to
-
3
4
5
6
7
8
1.7
27
-
[11-13, 14a,
advancing cellular bioenergetics as a new target space.
26] Here, we identified a novel TB-compound GaMF1 targeting the
unique mycobacterial F-ATP synthase γ-loop and being specific.
It retained ATP synthesis inhibition against a BDQ resistant
mutant and killing potency against H- and R monoresistant Mtb
strains as well as the HR-resistant Mtb strains. NMR- and docking
results of GaMF1 support the mode of action, describing that
binding of the compound to the mycobacterial γ-loop interferes
with the smooth rotation of the γ subunit along the loop residues
of the rotating c-ring turbine, thereby interrupting proper coupling
of proton conduction in the a-c subunits and ATP synthesis in the
ɑ3:β3 headpiece via the central stalk subunit γ. In comparison,
0.23
0.21
0.28
0.29
0.09
13.4
9.7
10.1
13.9
8.8
5.2
3
BDQ, TBAJ-876 and its derivatives 5307, 5316 and 5377 bind to
[14b]
both subunit c
and ε [6d, 14d, 27], and drug binding presumably
[14b]
results in stalling of c-ring rotation relative to subunit a.
Combining GaMF1 with BDQ or (±)-5307 increased the MIC- and
IC potency, while not affecting viability and pluripotency of human
embryonic stem cells, which is essential for further clinical
applications. Medicinal chemistry efforts resulted in GaMF1
derivatives with increased potency regarding ATP synthesis
inhibition and Mtb cell viability. Future studies will focus on the
utility of the new lead for the treatment of TB, considering that in
patients, Mtb grows extra- as well as intra-cellular. Here, we
demonstrated activity of GaMF1 against Mtb growing in broth, i.e.
against the extracellular form of the bacterium. Therefore, activity
of the new F-ATP synthase inhibitor against intracellular Mtb will
be determined in macrophage infection models.
23.5
14.6
9.9
Acknowledgements
a ATP synthesis Inhibitory Concentration on M. smegmatis IMVs
b M. smegmatis Minimum Inhibitory Concentration
This research was supported by the National Research
Foundation (NRF) Singapore, NRF Competitive Research
Programme (CRP), Grant Award Number NRF–CRP18–2017–
01. A. Hotra is grateful to receive an IGS Premium Scholarship,
Institute of Technology in Health and Medicine at NTU, and J. P.
Sarathy received a PhD scholarship from the School of Medicine,
NUS, Singapore. We thank Dr. Yongxin Li for the X-ray
crystallographic structure determinations.
c M. tuberculosis Minimum Inhibitory Concentration
In an attempt to retain the key amide functionality while
addressing potential concerns about toxicity arising due to
oxidation of the 1,4-phenylenediamine moiety (ring B), analogs (6
– 8) bearing a central benzimidazole motif with a phenyl
substituent were synthesized. Such a modification would restrict
rotational freedom between the A and B ring while retaining the
phenyl substituent. All three analogs displayed good IC50 and
MIC50 values, with one (8) showing a 10-fold improvement
compared to the original compound, GaMF1. The p-chloro phenyl
derivative 8 afforded a 18-fold enhancement in ATP synthesis
inhibition of about 90 nM on M. smegmatis (Fig. 7A) and even 8.7
Conflict of Interest
G.G., R.W.B., A.H., T.D., K.P. are inventors on the
patent PCT/SG2018/050075, and G.G., R.W.B., P.S.N. are
inventors on the patent 10201911205R which are related to the
chemical series described in this article.
7
This article is protected by copyright. All rights reserved.

10.1002/anie.202002546
Angewandte Chemie International Edition
RESEARCH ARTICLE
Langer, O. Yildiz, L. Eckhardt-Strelau, J. E. Guillemont, A. Koul,
T. Meier, Sci Adv 2015, 1, e1500106; c) N. Kamariah, R. G.
Huber, W. Nartey, S. Bhushan, P. J. Bond, G. Grüber, J Struct
Biol 2019, 207, 199-208; d) S. Joon, P. Ragunathan, L.
Sundararaman, W. Nartey, S. Kundu, M. S. S. Manimekalai, N.
Bogdanović, T. Dick, G. Grüber, FEBS J 2018, 285, 1111-1128;
e) N. Kamariah, P. Ragunathan, J. Shin, W. G. Saw, C. F.
Wong, T. Dick, G. Grüber, Progr. Biophysi. Mol. Biol. 2019, 152,
64-73.
Keywords: ATP synthesis • Bioenergetics • Drug Discovery • F-
ATP synthase • OXPHOS • Tuberculosis
References
[1]
[2]
World Health Organization (2019) Global tuberculosis report
2019. https://www.who.int/tb/publications/global_report/en/.
a) G. M. Cook, K. Hards, C. Vilcheze, T. Hartman, M. Berney,
Microbiol Spectr 2014, 2 doi: 10.1128/microbiolspec; b) S. P.
Rao, S. Alonso, L. Rand, T. Dick, K. Pethe, Proc Natl Acad Sci
U S A 2008, 105, 11945-11950.
[15] A. Hotra, M. Suter, G. Biuković, P. Ragunathan, S. Kundu, T.
Dick, G. Grüber, FEBS J 2016, 283, 1947-1961.
[16] R. Priya, G. Biuković, M. S. Manimekalai, J. Lim, S. P. Rao, G.
Grüber, J Bioenerg Biomembr 2013, 45, 121-129.
[17] A. Koul, L. Vranckx, N. Dhar, H. W. Gohlmann, E. Ozdemir, J.
M. Neefs, M. Schulz, P. Lu, E. Mortz, J. D. McKinney, K.
Andries, D. Bald, Nat Commun 2014, 5, 3369.
[18] P. Ragunathan, H. Sielaff, L. Sundararaman, G. Biuković, M. S.
Subramanian Manimekalai, D. Singh, S. Kundu, T. Wohland,
W. Frasch, T. Dick, G. Grüber, J Biol Chem 2017, 292, 11262-
11279.
[19] N. P. Kalia, E. J. Hasenoehrl, N. B. Ab Rahman, V. H. Koh, M.
L. T. Ang, D. R. Sajorda, K. Hards, G. Grüber, S. Alonso, G. M.
Cook, M. Berney, K. Pethe, Proc Natl Acad Sci U S A 2017,
114, 7426-7431.
[20] a) S. Petrella, E. Cambau, A. Chauffour, K. Andries, V. Jarlier,
W. Sougakoff, Antimicrob Agents Chemother 2006, 50, 2853-
2856; b) S. Kundu, G. Biuković, G. Grüber, T. Dick, Antimicrob
Agents Chemother 2016, 60, 6977-6979.
[3]
K. Andries, P. Verhasselt, J. Guillemont, H. W. Gohlmann, J.
M. Neefs, H. Winkler, J. Van Gestel, P. Timmerman, M. Zhu, E.
Lee, P. Williams, D. de Chaffoy, E. Huitric, S. Hoffner, E.
Cambau, C. Truffot-Pernot, N. Lounis, V. Jarlier, Science 2005,
307, 223-227.
[4]
[5]
a) K. Andries, C. Villellas, N. Coeck, K. Thys, T. Gevers, L.
Vranckx, N. Lounis, B. C. de Jong, A. Koul, PLoS One 2014, 9,
e102135; b) R. C. Hartkoorn, S. Uplekar, S. T. Cole, Antimicrob
Agents Chemother 2014, 58, 2979-2981.
a) A. H. Diacon, P. R. Donald, A. Pym, M. Grobusch, R. F.
Patientia, R. Mahanyele, N. Bantubani, R. Narasimooloo, T. De
Marez, R. van Heeswijk, N. Lounis, P. Meyvisch, K. Andries, D.
F. McNeeley, Antimicrob Agents Chemother 2012, 56, 3271-
3276; b) J. Guillemont, C. Meyer, A. Poncelet, X. Bourdrez, K.
Andries, Future medicinal chemistry 2011, 3, 1345-1360;
cFDA, 2012.
[21] R. C. Hartkoorn, S. Uplekar, S. T. Cole, Antimicrob Agents
Chemother 2014, 58, 2979-2981.
[6]
a) A. S. T. Tong, P. J. Choi, A. Blaser, H. S. Sutherland, S. K.
Y. Tsang, J. Guillemont, M. Motte, C. B. Cooper, K. Andries, W.
Van den Broeck, S. G. Franzblau, A. M. Upton, W. A. Denny,
B. D. Palmer, D. Conole, ACS Med Chem Lett 2017, 8, 1019-
1024; b) P. J. Choi, H. S. Sutherland, A. S. T. Tong, A. Blaser,
S. G. Franzblau, C. B. Cooper, M. U. Lotlikar, A. M. Upton, J.
Guillemont, M. Motte, L. Queguiner, K. Andries, W. Van den
Broeck, W. A. Denny, B. D. Palmer, Bioorg Med Chem Lett
2017, 27, 5190-5196; c) H. S. Sutherland, A. S. T. Tong, P. J.
Choi, D. Conole, A. Blaser, S. G. Franzblau, C. B. Cooper, A.
M. Upton, M. U. Lotlikar, W. A. Denny, B. D. Palmer, Bioorganic
& medicinal chemistry 2018, 26, 1797-1809; d) J. P. Sarathy,
P. Ragunathan, J. Shin, C. B. Cooper, A. M. Upton, G. Grüber,
T. Dick, Antimicrob Agents Chemother 2019, 63 pii: e01191-19.
S. J. Tantry, S. D. Markad, V. Shinde, J. Bhat, G. Balakrishnan,
A. K. Gupta, A. Ambady, A. Raichurkar, C. Kedari, S. Sharma,
N. V. Mudugal, A. Narayan, C. N. Naveen Kumar, R. Nanduri,
S. Bharath, J. Reddy, V. Panduga, K. R. Prabhakar, K.
Kandaswamy, R. Saralaya, P. Kaur, N. Dinesh, S. Guptha, K.
Rich, D. Murray, H. Plant, M. Preston, H. Ashton, D. Plant, J.
Walsh, P. Alcock, K. Naylor, M. Collier, J. Whiteaker, R. E.
McLaughlin, M. Mallya, M. Panda, S. Rudrapatna, V.
Ramachandran, R. Shandil, V. K. Sambandamurthy, K. Mdluli,
C. B. Cooper, H. Rubin, T. Yano, P. Iyer, S. Narayanan, S.
Kavanagh, K. Mukherjee, V. Balasubramanian, V. P.
Hosagrahara, S. Solapure, S. Ravishankar, P. S. Hameed, J
Med Chem 2017, 60, 1379-1399.
[22] a) K. G. Chen, B. S. Mallon, K. Park, P. G. Robey, R. D. G.
McKay, M. M. Gottesman, W. Zheng, Trends in molecular
medicine 2018, 24, 805-820; b) Y. Avior, I. Sagi, N. Benvenisty,
Nature reviews. Molecular cell biology 2016, 17, 170-182.
[23] S. H. Vijaya Chandra, H. Makhija, S. Peter, C. M. Myint Wai, J.
Li, J. Zhu, Z. Ren, M. S. D'Alcontres, J. W. Siau, S. Chee, F. J.
Ghadessy, P. Dröge, Nucleic Acids Res 2016, 44, e55.
[24] T. Yano, S. Kassovska-Bratinova, J. S. Teh, J. Winkler, K.
Sullivan, A. Isaacs, N. M. Schechter, H. Rubin, J Biol Chem
2011, 286, 10276-10287.
[25] K. Pethe, P. Bifani, J. Jang, S. Kang, S. Park, S. Ahn, J. Jiricek,
J. Jung, H. K. Jeon, J. Cechetto, T. Christophe, H. Lee, M.
Kempf, M. Jackson, A. J. Lenaerts, H. Pham, V. Jones, M. J.
Seo, Y. M. Kim, M. Seo, J. J. Seo, D. Park, Y. Ko, I. Choi, R.
Kim, S. Y. Kim, S. Lim, S. A. Yim, J. Nam, H. Kang, H. Kwon,
C. T. Oh, Y. Cho, Y. Jang, J. Kim, A. Chua, B. H. Tan, M. B.
Nanjundappa, S. P. Rao, W. S. Barnes, R. Wintjens, J. R.
Walker, S. Alonso, S. Lee, J. Kim, S. Oh, T. Oh, U. Nehrbass,
S. J. Han, Z. No, J. Lee, P. Brodin, S. N. Cho, K. Nam, J. Kim,
Nature medicine 2013, 19, 1157-1160.
[26] a) B. Wiseman, R. G. Nitharwal, O. Fedotovskaya, J. Schafer,
H. Guo, Q. Kuang, S. Benlekbir, D. Sjostrand, P. Adelroth, J. L.
Rubinstein, P. Brzezinski, M. Hogbom, Nat Struct Mol Biol
2018, 25, 1128-1136; b) H. Gong, J. Li, A. Xu, Y. Tang, W. Ji,
R. Gao, S. Wang, L. Yu, C. Tian, J. Li, H. Y. Yen, S. Man Lam,
G. Shui, X. Yang, Y. Sun, X. Li, M. Jia, C. Yang, B. Jiang, Z.
Lou, C. V. Robinson, L. L. Wong, L. W. Guddat, F. Sun, Q.
Wang, Z. Rao, Science 2018, 362; c) W. Moreira, D. B. Aziz, T.
Dick, Frontiers in microbiology 2016, 7, 199.
[7]
[8]
[9]
S. Kumar, R. Mehra, S. Sharma, N. P. Bokolia, D. Raina, A.
Nargotra, P. P. Singh, I. A. Khan, Tuberculosis 2018, 108, 56-
63.
W. G. Saw, M. L. Wu, P. Ragunathan, G. Biuković, A. M. Lau,
J. Shin, A. Harikishore, C. Y. Cheung, K. Hards, J. P. Sarathy,
R. W. Bates, G. M. Cook, T. Dick, G. Grüber, Scientific Rep.
2019, 9, 16759.
[27] G. Biuković, S. Basak, M. S. Manimekalai, S. Rishikesan, M.
Roessle, T. Dick, S. P. Rao, C. Hunke, G. Grüber, Antimicrob
Agents Chemother 2013, 57, 168-176.
[10] a) A. Mandavilli, Nature medicine 2007, 13, 271; b) D. G.
Russell, C. E. Barry, 3rd, J. L. Flynn, Science 2010, 328, 852-
856.
[11] A. Koul, E. Arnoult, N. Lounis, J. Guillemont, K. Andries, Nature
2011, 469, 483-490.
[12] C. M. Sassetti, D. H. Boyd, E. J. Rubin, Mol Microbiol 2003, 48,
77-84.
[13] G. M. Cook, K. Hards, E. Dunn, A. Heikal, Y. Nakatani, C.
Greening, D. C. Crick, F. L. Fontes, K. Pethe, E. Hasenoehrl,
M. Berney, Microbiol Spectr 2017, 5.
[14] a) A. T. Zhang, M. G. Montgomery, A. G. W. Leslie, G. M. Cook,
J. E. Walker, Proc Natl Acad Sci U S A 2019; b) L. Preiss, J. D.
8
This article is protected by copyright. All rights reserved.

10.1002/anie.202002546
Angewandte Chemie International Edition
RESEARCH ARTICLE
Entry for the Table of Contents
The antimycobacterial GaMF1 inhibits the mycobacterial F-ATP
synthase by binding to a specific loop of subunit γ, preventing ATP
synthesis via oxidative phosphorylation. The bactericidal
compound is active against multidrug-resistant strains, increased
the potency of the TB-drug Bedaquiline and other DARQ
analogues, opening a new avenue for an efficient multi-drug
combination. Chemistry efforts resulted in analogues with
nanomolar potencies.
9
This article is protected by copyright. All rights reserved.