1-(2-Hydroxybenzoyl)-thiosemicarbazides are promising antimicrobial agents targeting D-alanine-D-alanine ligase in bacterio

Article abstract of DOI:10.1016/j.ejmech.2018.09.067

The bacterial cell wall and the enzymes involved in peptidoglycan synthesis are privileged targets for the development of novel antibacterial agents. In this work, a series of 1-(2-hydroxybenzoyl)-thiosemicarbazides inhibitors of D-Ala-D-Ala ligase (Ddl) were designed and synthesized in order to target resistant strains of bacteria. Among these, the 4-(3,4-dichlorophenyl)-1-(2-hydroxybenzoyl)-3-thiosemicarbazide 29 was identified as a potent Ddl inhibitor with activity in the micromolar range. This compound, possessing strong antimicrobial activity including against multidrug resistant strains, was proven to act through a bactericidal mechanism and demonstrated very low cytotoxicity on THP-1 human monocytic cell line. Inhibition of Ddl activity by 29 was confirmed in bacterio using UPLC-MS/MS by demonstrating an increase in D-Ala intracellular pools accompanied by a commensurate decrease in D-Ala-D-Ala. Further structure-activity relationships (SARs) studies provided evidence that the hydroxyl substituent in the 2-position (R₁) of the benzoylthiosemicarbazide scaffold is essential for the enzymatic inhibition. This work thus highlights the 1-(2-hydroxybenzoyl)-thiosemicarbazide motif as a very promising tool for the development of novel antibacterial compounds acting through an interesting mechanism of action and low cytotoxicity.

Full text of DOI:10.1016/j.ejmech.2018.09.067

Accepted Manuscript
1-(2-Hydroxybenzoyl)-thiosemicarbazides are promising antimicrobial agents
targeting D-alanine-D-alanine ligase in bacterio
Alice Ameryckx, Léopold Thabault, Lionel Pochet, Serge Leimanis, Jacques H.
Poupaert, Johan Wouters, Bernard Joris, Françoise Van Bambeke, Raphaël
Frédérick
PII:
S0223-5234(18)30853-5
10.1016/j.ejmech.2018.09.067
EJMECH 10778
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 4 May 2018
Revised Date: 9 August 2018
Accepted Date: 26 September 2018
Please cite this article as: A. Ameryckx, Lé. Thabault, L. Pochet, S. Leimanis, J.H. Poupaert, J. Wouters,
B. Joris, Franç. Van Bambeke, Raphaë. Frédérick, 1-(2-Hydroxybenzoyl)-thiosemicarbazides are
promising antimicrobial agents targeting D-alanine-D-alanine ligase in bacterio, European Journal of
Medicinal Chemistry (2018), doi: https://doi.org/10.1016/j.ejmech.2018.09.067.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
1-(2-Hydroxybenzoyl)-thiosemicarbazides are
promising antimicrobial agents targeting
-alanine ligase in bacterio.
D-alanine-
D
a
a
b
c
a
Alice Ameryckx, ‡ Léopold Thabault, ‡ Lionel Pochet, Serge Leimanis, Jacques H. Poupaert,
d
c
e
a,*
Johan Wouters, Bernard Joris, Françoise Van Bambeke, Raphaël Frédérick
a
Medicinal Chemistry Research Group (CMFA), Louvain Drug Research Institute (LDRI),
Université catholique de Louvain (UCLouvain), 73 Avenue Mounier, B1.73.10, 1200 Bruxelles,
Belgium
b
Department of Pharmacy, Namur Medicine & Drug Innovation Center (NAMEDIC), Namur
Research Institute for Life Sciences (NARILIS), University of Namur, Namur, Belgium.
c
Centre d'Ingénierie des Protéines, Institut de Chimie B6A, Sart-Tilman, Université de Liège,
Liège, Belgium
d
Department of Chemistry, Namur Medicine & Drug Innovation Center (NAMEDIC), Namur
Research Institute for Life Sciences (NARILIS), University of Namur, Namur, Belgium.
e
Pharmacologie cellulaire et moléculaire (FACM), Louvain Drug Research Institute (LDRI),
Université catholique de Louvain (UCLouvain), 73 Avenue Mounier, B1.73.05, 1200 Bruxelles,
Belgium
1

ACCEPTED MANUSCRIPT
ABSTRACT. The bacterial cell wall and the enzymes involved in peptidoglycan synthesis are
privileged targets for the development of novel antibacterial agents. In this work, a series of 1-(2-
hydroxybenzoyl)-thiosemicarbazides inhibitors of
D
-Ala- -Ala ligase (Ddl) were designed and
D
synthesized in order to target resistant strains of bacteria. Among these, the 4-(3,4-
dichlorophenyl)-1-(2-hydroxybenzoyl)-3-thiosemicarbazide 29 was identified as a potent Ddl
inhibitor with activity in the micromolar range. This compound, possessing strong antimicrobial
activity including against multidrug resistant strains, was proven to act through a bactericidal
mechanism and demonstrated very low cytotoxicity on THP-1 human monocytic cell line.
Inhibition of Ddl activity by 29 was confirmed in bacterio using UPLC-MS/MS by
demonstrating an increase in
D-Ala intracellular pools accompanied by a commensurate decrease
in -Ala- -Ala. Further structure-activity relationships (SARs) studies provided evidence that the
D
D
hydroxyl substituent in the 2-position (R ) of the benzoylthiosemicarbazide scaffold is essential
1
for the enzymatic inhibition. This work thus highlights the 1-(2-hydroxybenzoyl)-
thiosemicarbazide motif as a very promising tool for the development of novel antibacterial
compounds acting through an interesting mechanism of action and low cytotoxicity.
Keywords:
D
-alanine- -alanine ligase inhibitors, Benzoylthiosemicarbazides, Antimicrobial
D
agents, Antibiotics, Structure-activity relationships.
1. INTRODUCTION
In an era of growing antibiotic resistance, the search for effective molecules with novel
1
mechanisms of action is a priority. Today, the bacterial cell wall and the enzymes involved in
peptidoglycan biosynthesis, a major cell wall structural component, constitute validated targets
2

ACCEPTED MANUSCRIPT
2
of many antimicrobial agents. The peptidoglycan is a polymeric network which alternates N-
acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) disaccharide units cross-
3
linked via peptide bonds thanks to the catalytic activity of
D,
D-transpeptidases. These enzymes
catalyze the reticulation of the peptide chains ending in
of the terminal
D
-Ala-
D
-Ala, with subsequent elimination
D
-Ala residue and transfer of the remaining peptide chain on the lateral amino
group of another peptide of an adjacent glycan chain. This last step of peptidoglycan biosynthetic
pathway is inhibited by β-lactams antibiotics (such as penicillins), which act as suicide inhibitors
4
of the transpeptidases, and by glycopeptides (vancomycin) that directly bind the terminal
D-Ala-
D-Ala moiety, creating a steric hindrance that prevents the access of
D D
,
-transpeptidases to their
5
substrate. However, the frequent resistance associated with the use of these antibiotics revealed
6
the need to target other enzymes acting on earlier steps of peptidoglycan synthesis. Among
these,
Ddl is present in both Gram-negative and -positive bacteria and catalyzes the formation of
Ala- -Ala dipeptide, an early step in peptidoglycan synthesis. Two isoforms exist for E. coli and
S. typhimurium, DdlA and DdlB. Alternative ligases, catalyzing the formation of
VanA, -B and -D types) or -Ala- -Ser (VanC, -E, -G and -L types), were also described in
vancomycin-resistant enterococci.
D-alanine-D-alanine ligase (Ddl) represents a very interesting target.
D
-
D
7
D
-Ala- -Lac
D
(
D
D
5
, 8
Therefore, the discovery of inhibitors that are active not
only on Ddl but also on these alternative ligases would increase their potential spectrum of
activity. Today, various crystallographic structures of DdlB in presence of inhibitors, or in the
apo form, are available and could thus contribute to the rational development of novel
9
-12
6
antibiotics. Over the past few years, four main classes of Ddl inhibitors were described. First,
1
3-17
18-21
analogues of the substrate
D-Ala were reported
including
D-cycloserine (DCS, Figure 1),
the only inhibitor used in the past as a therapeutic agent. However, its use is rather limited due to
3

ACCEPTED MANUSCRIPT
2
2
its known neurotoxicity. This cyclic analogue of
D-Ala acts as a competitive and reversible
19
inhibitor of Ddl (K = 25 µM against M. tuberculosis Ddl) and exhibits relatively high MIC
i
2
3
values (50 mg/L against M. tuberculosis). Secondly, compounds that mimic the transition state
of the enzymatic reaction were reported. These latter inhibit Ddl by acting as false substrates.
Interestingly, these compounds allowed crystallographic studies of the enzyme-inhibitor complex
2
4-25
thus affording important structural information for drug design.
reaction product -Ala- -Ala also showed interesting Ddl inhibition.
showing no structural resemblance with the substrate, the reaction intermediate or the product
Third, analogues of the
13, 26
D
D
Fourth, inhibitors
2
7-36
were discovered by screening chemical libraries or by rational drug design.
For instance,
thanks to a high-throughput screening against S. aureus Ddl (StaDDl), compound 1 (Figure 1)
was identified (K = 4 µM). Although modestly active, this compound was used in structural
i
1
2
studies and revealed a novel binding mode at an allosteric site. This study thus paves the way to
the development of original Ddl inhibitors acting via new mechanisms of inhibition. However,
even if the most potent molecules reported so far showed inhibitory potencies in the micromolar
range, only a few of them were evaluated for their antibacterial activity, which unfortunately
usually remained very low.
An initial de novo design study carried out previously in our group allowed identifying the 2-
phenylbenzoxazole series, exemplified by molecule 2 (Figure 1), as promising Ddl inhibitors.
3
3
This series, however, revealed to be deprived of any antibacterial activity. In the present work,
we performed a screening of our internal library of compounds and identified the 1-(2-
hydroxybenzoyl)-4-phenyl-3-thiosemicarbazide 3 (Figure 1) as an interesting tool compound
with an IC50 of 0.80 µM against Ddl (E. faecalis) and relatively weak but reasonable
antimicrobial potency on S. aureus ATCC 25923 and E. faecalis ATCC 29212 with MIC values
4

ACCEPTED MANUSCRIPT
of 128 mg/L and 512 mg/L respectively. We found from the literature that this compound was, in
fact, previously reported to be a very weak carbonic anhydrase inhibitor I and IX (K > 80 µM on
i
3
7
both isoforms) and to have some antibacterial activities evaluated against usual random strains,
3
8
although no information were given regarding its mode of action . These encouraging results
prompted us to further develop and study this series of compounds, particularly their mode of
action as antibacterials.
Figure 1. Chemical structure and affinity for Ddl or inhibitory potency of DCS (M. tuberculosis
1
9
12
Ddl) , 3-chloro-2,2-dimethyl-N-[4-(trifluoromethyl)phenyl]propanamide 1 (S. aureus Ddl) and
3
3
2
-phenylbenzoxazole 2 (E. faecalis Ddl) reported in the literature. The IC50 of 1-(2-
hydroxybenzoyl)-4-phenyl-3-thiosemicarbazide 3 (E. faecalis Ddl) is presented as the mean ±
SD of experiments realized in triplicate, n=2.
Various pharmacomodulations of compound 3 were investigated to improve the Ddl inhibition
and the antibacterial potency, and to establish qualitative structure-activity relationships. The
mechanism of action towards Ddl was also studied and the activity against Gram-positive and
negative, sensitive or resistant strains to current therapeutics was determined. Finally, the in vivo
biochemical mechanism was determined by UHPLC-MS/MS dosage of intracellular
L-Ala,
D-Ala
5

ACCEPTED MANUSCRIPT
and
D-Ala-D-Ala levels in response to 4-(3,4-dichlorophenyl)-1-(2-hydroxybenzoyl)-3-
thiosemicarbazide 29.
. CHEMISTRY
2
The synthesis of benzoylthiosemicarbazides 3 and 14-49 was carried out according to a two-
step procedure presented in Scheme 1. First, the benzohydrazides 4-13 were obtained according
to the literature by refluxing hydrazine hydrate and the appropriate methyl benzoate in ethanol or
3
9-40
were commercially available (4 and 8).
Then, these benzohydrazides were reacted with
4
1-42
various isothiocyanates in refluxing methanol or at room temperature.
The targeted
benzoylthiosemicarbazides 3 and 14-49 were collected by filtration and recrystallized from
ethanol if necessary.
Scheme 1. Synthetic route for the preparation of the benzoylthiosemicarbazides 3 and 14-49 and
a
their precursors 4-13.
a
Reagents and conditions: (i) 65% hydrazine hydrate (5 equiv), EtOH, reflux, 2.5 h-24 h (ii)
MeOH, reflux or r.t., 0.5 h-24 h.
6

ACCEPTED MANUSCRIPT
1
The three labile protons (NH) of the thiosemicarbazide linker were visible on H NMR in
DMSO at approximately 9.72 ppm, 9.94 ppm and 10.47 ppm depending on the substituents of
1
the thiosemicarbazide motif. The H NMR
spectrum of 1-(2-hydroxybenzoyl)-
thiosemicarbazides 3, 22-42 showed a duplication of two labile proton signals of the -NH-NH-
functionality, probably reflecting an equilibrium between two of the potential tautomeric forms
of this molecule (Scheme 2), as well as the 1-(2-methoxybenzoyl)-thiosemicarbazide 49 and the
1
-(2-hydroxynaphtoyl)-thiosemicarbazide 51. As this peak splitting can only be seen for the
benzoylthiosemicarbazides bearing a substituent containing an oxygen in the 2-position (see
Supporting Information), intramolecular H-bonding was suggested to explain this enol form
1
stabilization allowing to observe both peaks in H NMR. The existence of this keto-enol
equilibrium for 1-(2-hydroxybenzoyl)-thiosemicarbazides was already mentioned in the
4
3
literature.
Scheme 2: Potential keto-enol equilibrium for compounds bearing a substituent containing an
1
oxygen in the 2-position is visible in H NMR spectrum.
To better understand the structure-activity relationships (SARs) in this series, other analogues
were also prepared. First, the left part of the model compound 3 was rigidified by introducing a
quinazolinone moiety and extended with the replacement of the phenyl by a naphtyl group, as
depicted in Scheme 3. After obtaining the naphtohydrazide precursor 50, the syntheses of 51-52
were thus carried out following the previous procedure for thiosemicarbazides synthesis. We
then also investigated the replacement of the thiocarbonyl moiety by a carbonyl group to obtain
7

ACCEPTED MANUSCRIPT
the semicarbazide motif. To this end, the commercially available benzohydrazide 4 and 2-
hydroxybenzohydrazide 5 were added to 3,4-dichloro isocyanate in methanol at reflux to afford
respectively 2-benzoyl-N-(3,4-dichlorophenyl)hydrazine-1-carboxamide 53 and N-(3,4-
dichlorophenyl)-2-(2-hydroxybenzoyl)hydrazine-1-carboxamide 54 (Scheme 3) following the
4
4
adapted known procedure.
a
Scheme 3. Synthesis of new structural analogues.
O
O
O
NH2
N
i
H
OH
OH
5
0
O
Cl
Cl
O
H
N
H
NH2
NH2
N
Cl
Cl
+
N
N
N
H
H
ii
C
S
OH
S
OH
5
0
51
O
O
H
N
H
N
N
N
+
S
ii
N
N
N
N
O
C
C
5
2
S
O
H
N
H
N
Cl
Cl
Cl
Cl
NH2
N
H
N
+
R1
R1
H
iii
O
4
4
5
-5
O
53-54
53 R1 = H
54 R1 = 2-OH
R1 = H
R1 = 2-OH
a
Reagents and conditions: (i) 65% hydrazine hydrate (5 equiv), EtOH, reflux, 1 h (ii) EtOH,
reflux, 4 h- 20 h (iii) MeOH, reflux, 17 h.
1
13
The structures and purities were assessed by H NMR, C NMR, HRMS and HPLC. The
analysis of spectral data of the target compounds are provided in the experimental section. These
molecules were subsequently evaluated on Ddl using purified His-tagged Ddl and their
antimicrobial activity was assessed on sensitive and resistant bacteria.
3. PHARMACOLOGICAL EVALUATION AND DISCUSSION
8

ACCEPTED MANUSCRIPT
3.1. Enzymatic assay
3.1.1 Optimization. Prior to the evaluation of compounds 3, 14-49 and 51-54 towards
4
5
recombinant His-tagged Ddl from E. coli, the colorimetric malachite green assay was
optimized for the determination of inorganic phosphate generated during the reaction catalyzed
by Ddl. The linearity zone (Figure 2a), the optimal working conditions (Figure 2b), the
tolerance to DMSO (Figure 2c) and the K values for the two
D
-Ala sites (Figure 2d) were
determined. The linearity zone for the phosphate concentrations is from 0 to 40 µM. Working
conditions are 20 min of incubation time with -Ala, 20 mg/L of enzyme concentration and 10%
m
D
of DMSO. A detailed description of the assay optimization can be found in the Supporting
Information.
9

ACCEPTED MANUSCRIPT
Figure 2. Optimization of the enzymatic assay.
D-Ala and ATP concentrations were set
respectively at 1 mM and 500 µM. Buffer composition was 20 mM Tris.HCl (pH 7.4), 10 mM
MgCl , 10 mM KCl. (a) Evolution of the absorbance as a function of phosphate concentrations
2
in µM. Inorganic phosphate (0-80 µM) was incubated with malachite green for 25 min.
Absorbance was monitored at 650 nm. (b) Evolution of phosphate concentration (µM) in
relationship with the product of enzyme concentration (mg/L) and time (min). Concentrations of
Ddl were 5-20 mg/L. Incubation time with the substrate was 0-30 min. (c) Evaluation of DMSO
influence on the enzymatic activity. DMSO concentrations (v/v) were 1, 2, 3, 5 and 10%. (d)
-1
Velocity of the enzyme (µM.min ) as a function of substrate concentration (mM). Substrate
7
, 46
concentrations were 0.45-30 fold the theoretical Km2 (value reported in the literature).
Apparent K = 5.62 mM.
m
3
.1.2. Activity of inhibitors on the purified enzyme. A primary screening at 10 µM ([
mM) was performed in triplicate to select inhibitors that lowered the Ddl activity of 30% or more
Ddl inhibition ≥ 30%). Then, for the most active compounds, an IC50 was determined after 30
D
-Ala] = 1
(
min of incubation with Ddl (Table 1). DCS was used as a reference and it showed an IC50 of 262
6
µM in our hands, similar to the published data.
Table 1. Structures, Ddl inhibitory activities and some physicochemical parameters of
compounds 3, 14-49 and 51-54.
IC50 (µM) a LogDb TPSAb S (mmol/L)b
pH=7.4) (pH=7.4)
Compd
X
R
1
R
2
(
10

ACCEPTED MANUSCRIPT
3
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
2-OH
H
Phenyl
0.80 ± 0.12
n.a.
2.31
2.14
2.63
2.87
2.69
2.17
2.37
2.19
3.57
2.79
3.13
2.87
2.22
2.50
2.28
3.40
3.63
2.36
2.18
105.48
85.25
0.94
0.82
0.2
1
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
Phenyl
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
H
2-Chlorophenyl
3-Chlorophenyl
4-Chlorophenyl
2-Fluorophenyl
3-Fluorophenyl
4-Fluorophenyl
3,4-Dichlorophenyl
2-Chlorophenyl
3-Chlorophenyl
4-Chlorophenyl
2-Fluorophenyl
3-Fluorophenyl
4-Fluorophenyl
2,4-Dichlorophenyl
3,4-Dichlorophenyl
2-Methoxyphenyl
4-Methoxyphenyl
n.a.
85.25
H
n.a.
85.25
0.07
0.05
0.5
H
n.a.
85.25
H
n.a.
85.25
H
n.a.
85.25
0.3
H
n.a.
85.25
0.26
0.002
0.31
0.07
0.12
0.46
0.25
0.23
0.04
0.008
0.65
0.72
H
n.a.
85.25
2-OH
2-OH
2-OH
2-OH
2-OH
2-OH
2-OH
2-OH
2-OH
2-OH
0.79 ± 0.14
0.63 ± 0.13
0.83 ± 0.16
0.97 ± 0.18
0.10 ± 0.15
0.84 ± 0.17
0.58 ± 0.15
1.54 ± 0.30
0.55 ± 0.08
0.71 ± 0.13
105.48
105.48
105.48
105.48
105.48
105.48
105.48
105.48
114.71
114.71
11

ACCEPTED MANUSCRIPT
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
1
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
/
2-OH
2-OH
2-OH
2-OH
2-OH
2-OH
2-OH
2-OH
2-OH
2-OH-4-I
2-OH-4-NH
3-OH
4-OH
4-OH
2-F
4-Cyanophenyl
Pentyl
0.51 ± 0.11
0.93 ± 0.15
*
2.02
2.75
0.71
3.37
3.23
2.08
2.89
3.54
1.32
4.31
3.01
3.30
3.25
1.89
3.26
3.49
3.50
4.08
4.54
129.27
105.48
117.95
105.48
105.48
122.55
114.71
114.71
118.37
105.48
131.5
0.64
0.32
3-Morpholinopropyl
1-Naphthyl
13.4
0.52 ± 0.14
0.73 ± 0.12
0.37 ± 0.04
0.08
3-Iodophenyl
4-Acetylphenyl
0.03
0.47
4-Trifluoromethoxyphenyl 0.70 ± 0.16
0.08
4-Benzyloxyphenyl
Pyridin-3-yl
0.99 ± 0.99
*
0.04
3.82
3,4-Dichlorophenyl
3,4-Dichlorophenyl
3,4-Dichlorophenyl
3,4-Dichlorophenyl
4-Nitrophenyl
0.93 ± 0.12
0.99 ± 0.99
n.a.
0.007
0.009
0.01
2
105.48
105.48
154.31
85.25
n.a.
0.01
n.a.
0.45
3,4-Dichlorophenyl
3,4-Dichlorophenyl
3,4-Dichlorophenyl
3,4-Dichlorophenyl
3,4-Dichlorophenyl
n.a.
0.003
0.003
0.003
0.001
0.0001
3-F
n.a.
85.25
4-F
n.a.
85.25
2-OMe
/
n.a.
62.39
0.56 ± 0.07
105.48
12

ACCEPTED MANUSCRIPT
5
5
5
2
3
4
/
O
O
/
/
Phenyl
3,4-Dichlorophenyl
3,4-Dichlorophenyl
/
n.a.
n.a.
1.95
3.00
3.66
/
88.82
70.23
90.46
/
0.75
0.002
0.01
/
H
2-OH
/
n.a.
DCS
262 ± 43.4
a
IC values are presented as the (mean ± SD) of measures performed in triplicate (n≥2)
5
0
b
LogD, TPSA (topological polar surface area) and S (solubility) were calculated with
ACD/Labs® program
*
Ddl inhibition of 30% at 10 µM
n.a. = not active; 100% residual activity at 10 µM
Firstly, compared to the parent compound 3, it can be observed that removal of the hydroxyl
group in the 2-position always leads to inactive compounds (14-21), independently of the nature
of the substituent present on the right part (R ) of the molecule (Table 1). When this hydroxyl
2
group is replaced by a bioisosteric fluorine 46-48 (R = F) or a methoxyl group 49 (R = OMe),
1
1
no inhibition is found either. Further comparison of the hydroxy-benzoylthiosemicarbazides 29,
3 and 44 possessing the hydroxyl function in the 2-, 3- and 4-position, respectively, also clearly
4
reveals that only the 2-position is tolerated. However, it seems that an additional substituent on
the left phenyl (41-42), or its replacement by a naphtyl group (51), allows to keep the activity if
the 2-OH is still present. This feature indicates that there is still enough space in the inhibitor
binding pocket and could be important for further structural modifications.
It thus appears from these first SAR modifications that a hydroxyl group in this 2-position is
key for Ddl inhibition. Because its replacement by a fluorine leads to inactive compounds, it may
13

ACCEPTED MANUSCRIPT
be inferred that (i) it probably contributes to the stabilization of the inhibitor inside the Ddl
cavity through H-bonding with residues bearing H-bond acceptor groups and (ii) it could
stabilize a particular conformation by intramolecular H-bonding. This hypothesis was further
validated by the X-ray structure of a representative benzoylthiosemicarbazide 29. As observed
from Figure 3, this compound co-crystallized with one DMSO molecule from the crystallization
solvent. The analysis of the X-ray structure of 29 revealed that the hydroxyl group on the
aromatic ring was, as suggested, intramolecularly H-bonded to the thiosemicarbazide NH (N-
H•••O interaction) leaving the OH available for an additional intermolecular H-bond with the
DMSO sulfoxide function. The OH group in the 2-position thus probably plays a key role both as
an intramolecular H-bond acceptor and an intermolecular H-bond donor. The inactivity of
compound 49 bearing a 2-methoxy substituent reinforces this hypothesis.
14

ACCEPTED MANUSCRIPT
Figure 3. View of the molecular interaction (H-bonds represented as dashed lines) within the
crystal packing of compound 29 (ORTEP representation for each molecule, 50% probability
representation).
Thereafter, we investigated the effect of the modulation on the right part (R ) of the
2
benzoylthiosemicarbazide scaffold. In this position an electron-withdrawing (Cl, F, CN, I ...) or
electron-donating (MeO, BnO ...) substituent is introduced at the phenyl ring (22-32 and 36-39).
Subsequent Ddl inhibition is always in the same range to that of the parent compound 3 (in the
submicromolar range). Interestingly, the introduction of a pentyl (33) or a naphtyl (35) group in
the same position also leads to potent Ddl inhibitors, in similar range that previously tested
molecules. On the contrary, introduction of more hydrophilic substituents such as 3-
morpholinopropyl (34) or pyridin-3-yl (40) decreases the Ddl inhibition as this compound
exhibits only a maximum Ddl inhibition of 30% at the highest concentration. The data obtained
2
thanks to these pharmacomodulations on R suggest that no critical interaction is made between
this part of the molecule and the enzymatic cavity apart from hydrophobic interactions.
Finally, the importance of the thiosemicarbazide linker was assessed. Its replacement by a
semicarbazide function, with (54) and without (53) the hydroxyl substituent in the 2-position of
the aromatic ring (R ), as well as the complete replacement of the 4-phenylthiosemicarbazide in
1
3
with a 1-(4-oxoquinazolin-3(4H)-yl)thiourea scaffold (52), to evaluate a cyclized structure,
afford inactive derivatives.
Furthermore, according to the literature, the inhibition of Ddl by compound 3, chosen for its
good aqueous solubility, was tested in the presence and absence of 0.01% Triton X-100 to
4
7-48
double-check that the activity was not an artefact due, for instance, to drug aggregation.
No
15

ACCEPTED MANUSCRIPT
change in Ddl inhibition was noticed, therefore proving that the activity was the result of a
specific interaction to Ddl.
These results highlight the importance of the thiosemicarbazide template and the 2-OH
benzoyl group for the proper recognition inside the Ddl enzyme cavity.
3.1.3. Study of the binding mode. A residual activity of Ddl of about 25% was always observed
for active compounds after 30 min of incubation with the inhibitor. To gain a deeper
understanding of the mechanism of action in this series a detailed study of the binding mode was
investigated. Compound 3 was selected for its good aqueous solubility and adequate lipophilicity
(
Table 1). First, upon increasing the incubation time of the inhibitor with the enzyme from 10 to
20 min (Figure 4a), we first noticed that, although the IC50 of 3 was not modified, the
maximum inhibition percentage increased from 60% (10 min incubation) up to more than 90%
90 min incubation). Then, the reversibility of the inhibition was assessed using a rapid dilution
1
(
4
9
assay. To this end, 29 was selected as a model compound for subsequent studies given its
activity on Ddl and bacteria (see below). As observed from Figure 4b, the enzymatic activity
was recovered after a rapid and large dilution of the enzyme-inhibitor complex, thus proving the
reversible interaction of our inhibitors with Ddl.
16

ACCEPTED MANUSCRIPT
Figure 4. (a) IC50 of benzoylthiosemicarbazide 3 with increasing incubation time (10, 30, 60, 90
and 120 min) of Ddl with the inhibitor. (b) Rapid dilution method with a) 3 µM of 29 and 200
mg/L of Ddl-His ; b) 0.3 µM of 29 and 20 mg/L of Ddl-His and c) rapid dilution of a) allows for
6
6
the recovery of the activity compared to the control b). All values are presented as the (mean ±
SD) of measures performed in triplicate.
Next, competition studies were carried out with 29 against both
b). Regarding competition against -Ala, as Ddl binds two -Ala molecules in two distinct
binding sites, we assumed that, at concentrations where the substrate occupies both sites (1
D-Ala and ATP (Figure 5a and
5
D
D
5
0
1
5, 46
mM~K ), K is negligible compared to K .
When a saturating concentration of ATP (500
-Ala were found (Km1 = 487.2 µM and Km2 = 10.5 mM).
-Ala of 50 mM was used, a K value for ATP of 691 µM could be
m2
m1
m2
µM) was used, two K values for
D
m
When a concentration of
D
m
approximated. Detailed information regarding these measurements can be found in the available
Supporting Information.
Upon increasing concentrations of 29, a decrease of the Vmax value was observed on both
graphs (Figure 5). Moreover, we noticed that the V/K
concentrations of 29 (see Supporting Information) thus suggesting a mixed non-competitive
inhibition profile for this compound towards both -Ala (α >1, inhibitor preferentially binds the
m
value also changes with increasing
D
free enzyme) and ATP (α <1, inhibitor preferentially binds the enzyme-ATP complex). This
mechanism of action led us to hypothesize an interaction of 29 at an allosteric site rather than at
the active site. This assumption is further supported by the recent identification of such an
allosteric site on Ddl together with the identification of Ddl inhibitors with similar kinetic
1
2
profile.
17

ACCEPTED MANUSCRIPT
Figure 5. Competition studies of model inhibitor 29 at three concentrations (0 µM, 1 µM and 10
-
1
µM) with (a)
enzyme velocity (µM.min ) as a function of [ATP] (mM). All values are presented as the (mean
SD) of measures performed in triplicate.
D
-Ala: enzyme velocity (µM.min ) as a function of [ -Ala] (mM). (b) ATP:
D
-1
±
3.2. Biological activities
3.2.1. In vitro antibacterial activity. The MICs of compounds 3, 14-48 and 51-54 against two
Gram-positive bacterial strains, S. aureus ATCC 25923 and E. faecalis ATCC 29212 were
determined using maximal concentrations of 100 µM and 400 µM respectively (see details in the
Supporting Information available) with 1% DMSO (v:v). Unfortunately, compounds 46-48 were
not soluble at these concentrations. The most active compounds are 28 with a MIC of 50 µM
(17.81 mg/L) against both strains, 29 with a MIC of 100 µM (35.62 mg/L) and 50 µM (17.81
mg/L) against S. aureus ATCC 25923 and E. faecalis ATCC 29212 respectively, 37-38 with
MICs of 25 µM (8.23 and 9.28 mg/L) against S. aureus ATCC 25923 and 50 µM (16.47 and
1
8.56 mg/L) against E. faecalis ATCC 29212, and, finally, 40 with a MIC of 50 µM (14.42
mg/L) against both strains (Table 2). The benzoylthiosemicarbazides 22-27, 32, 39 and 41 have
MICs ranging from 100 µM to 400 µM. In fact all these molecules have better in vitro activities
18

ACCEPTED MANUSCRIPT
than DCS, the reference antibiotic targeting Ddl. The other compounds assayed showed no
biological activities at these concentrations.
Table 2. MIC values of active benzoylthiosemicarbazides against fully susceptible Gram-
positive reference strains.
MIC (mM/(mg/L))
S. aureus ATCC
5923a
E. faecalis ATCC
2
29212a
Compd
mM
> 0.1
0.1
mg/L
mM
0.4
mg/L
128.72
128.72
61.06
2
2
2
2
2
2
2
3
3
3
3
2
3
5
6
7
8
9
2
7
8
9
> 32.18
32.18
0.4
> 0.1
> 0.1
> 0.1
0.05
0.1
> 30.53
> 30.53
> 30.53
17.81
0.2
0.2
61.06
0.2
61.06
0.05
0.05
0.1
17.81
35.62
17.81
> 0.1
> 31.23
8.23
31.23
0.025
0.025
0.1
0.05
0.05
> 0.4
16.47
9.28
18.56
39.34
> 157.38
19

ACCEPTED MANUSCRIPT
4
4
0
1
0.05
0.1
14.42
48.21
32
0.05
> 0.4
1.25
14.42
> 192.85
128
DCS
0.313
a
Staphylococcus aureus, S. aureus; Enterococcus faecalis, E. faecalis.
The active compounds on Ddl, 3, 22-42 and 51 were also assayed against various Gram-
positive and negative bacteria, including clinical isolates resistant to current antibiotics and
51
52
53
strains expressing alternative ligases (E. faecalis JH2-2::C1, BM 4390, BM 4575 ). MICs are
shown in Table 3.
Table 3. MIC values of active compounds 3, 22-42 and 51 against Gram-positive resistant
strains.
MIC (mM/(mg/L)) of:
E. faecalisa
S. aureusb
Compd
BM 4390 JH2-2::C1 BM 4575
MU 50
> 0.1
NRS 119
SA 325
SA 481
VRS-1
0.1/32.18
> 0.1
2
2
2
2
3
2
3
8
9
8
> 0.1
> 0.1
> 0.1
> 0.1
> 0.1
> 0.1
0.1/32.18 0.1/32.18 0.1/32.18
> 0.1
> 0.1
0.1/32.18
> 0.1
0.025/8.90 0.05/17.81 0.025/8.90 0.05/17.81 0.025/8.90 0.05/17.81 0.05/17.81 0.025/8.90
0.05/17.81 0.1/35.62 0.05/17.81 0.1/35.62 0.05/17.81 0.05/17.81 0.1/35.62 0.05/17.81
> 0.1
0.1/37.12
> 0.1
N.D.
> 0.1
0.1/37.12 0.1/37.12
> 0.1
N.D.
> 0.1
c
VAN
0.7/1024
0.35/512
0.005/8
0.0007/1
N.D.
>0.175/>256
20

ACCEPTED MANUSCRIPT
c
LZD
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
0.003/1
0.19/64
N.D.
N.D.
N.D.
N.D.
N.D.
c
CIP
N.D.
N.D.
0.77/256
a
52
VRE (Vancomycin resistant enterococcus): E. faecalis BM 4390 (vanB genotype, in vitro
mutant of a clinical isolate with inactive Ddl; constitutive expression of -Ala- -Lac); E. faecalis
JH2-2::C1 (vanB genotype, engineered derivative of JH2-2; constitutive expression of Ddl and
D
D
51
5
3
of the VanB ligase); E. faecalis BM 4575 (vanE genotype, clinical isolate, constitutive
expression of Ddl and of the VanE
D
-Ala- -Ser ligase).
D
b
S. aureus MU 50: MRSA (methicillin resistant S. aureus) and VISA (vancomycin
intermediate resistant S. aureus); S. aureus NRS 119: MRSA resistant to linezolid (clinical
5
4
isolate) ; S. aureus SA 325: CA-MRSA (community-acquired MRSA) resistant to
erythromycin; S. aureus SA 481: HA-MRSA (hospital-acquired MRSA) resistant to
ciprofloxacin and moxifloxacin; VRS-1: VRSA (vancomycin resistant S. aureus, vanA genotype)
and HA-MRSA.
c
VAN, vancomycin; LZD, linezolid; CIP, ciprofloxacin.
All the tested compounds were inactive against Gram-negative bacteria (Klebsiella
pneumoniae ATCC 700603, Escherichia coli ATCC 25922 and Pseudomonas aeruginosa
PAO1). Compound 28 was equipotent against all Gram-positive strains, including those
expressing alternative ligases (E. faecalis BM 4390, BM 4575, JH2-2::C1; VRSA VRS-1) or
resistant to glycopeptides, oxazolidinones, or fluoroquinolones, with MICs of 25-50 µM. The
benzoylthiosemicarbazide 29 showed MICs of 50-100 µM, whatever the strain tested.
Compound 22 displayed MIC of 100 µM towards S. aureus NRS 119, SA 325, SA 481 and
VRS-1, and compound 23 showed MIC of 100 µM towards S. aureus SA 325, but these latter
21

ACCEPTED MANUSCRIPT
were not active on E. faecalis and the other staphylococci at 100 µM. Finally, compound 38
displayed MIC of 100 µM for E. faecalis JH2-2::C1, S. aureus NRS 119 and SA 325. These data
suggest that compounds 28, 29 and 38 may also be active on alternative ligases. Moreover, the
ability of 2-hydroxybenzoyl-thiosemicarbazides derivatives to prevent the growth of various
bacterial strains resistant to a wide range of antibiotics such as methicillin, linezolid,
erythromycin, ciprofloxacin and moxifloxacin highlights them as attractive molecules warranting
further development.
Interestingly, none of the benzoylthiosemicarbazides that are inactive on ligase exhibits
antibacterial activity (see data in the Supporting Information). This supports our hypothesis that
Ddl is the bacterial target of these compounds and could explain in part the antibacterial activity
3
8, 55-56
reported previously by other groups for benzoylthiosemicarbazide analogues.
Particularly,
Siwek et al. and Plech et al. recently reported that 4-benzoylthiosemicarbazides derivatives could
5
7, 58
act as topoisomerase IV and DNA gyrase inhibitors.
However, no convincing concordance
was found in their work between the antimicrobial activity and the inhibition of their presumed
targets. The present analysis may therefore suggest that Ddl could be the bacterial target of these
derivatives.
3.2.2. Killing curve for compound 29. To determine if the model compound 29 was
bactericidal, killing curves were performed over 24 hours for S. aureus ATCC 25923 exposed to
increasing concentrations of inhibitor. As shown in Figure 6, a bactericidal effect was reached
6
when using a 10 CFU/mL starting inoculum after 24 h of incubation with concentrations of 2 or
5
× the MIC (0.2 and 0.5 mM), suggesting that the compound is slowly bactericidal. A complete
eradication of bacteria was obtained at 5× MIC.
22

ACCEPTED MANUSCRIPT
Figure 6. Time-kill curves of S. aureus ATCC 25923 incubated with the model inhibitor 29. The
strain was incubated with growth media (●) as positive control, with 5% DMSO (■) as growth
control; with molecule 29 at 1× the MIC (×), 2× the MIC (
▼), and 5× the MIC (♦). All values
are presented as the (mean ± SD) of measures performed in triplicate.
3
.2.3. Determination of in vivo
compound 29. In order to demonstrate that Ddl is effectively the in vivo biochemical target of our
-(2-hydroxybenzoyl)-thiosemicarbazides, the effect of 29 on intracellular levels of -Ala, -Ala
and -Ala- -Ala was evaluated by UHPLC-MS/MS. To this end we used an LC-MS/MS
methodology adapted from the literature in order to quantify these metabolites in vivo.
Briefly, the strategy involves the derivatization of -Ala, -Ala and -Ala- -Ala with Marfey’s
L
-Ala,
D
-Ala and
D
-Ala-
D
-Ala levels in presence of model
1
L
D
D
D
1
7, 59-60
L
D
D
D
6
1
reagent on S. aureus extracts followed by the chromatographic separation and detection of
Marfey’s derivatives in positive electrospray ionization (ESI) mode with multiple reaction
+
monitoring (MRM) for the transition corresponding to a neutral loss of 45 Da [M+H-45] , the
Marfey’s terminal protonated amide. Based on preliminary experiments, we chose to expose
23

ACCEPTED MANUSCRIPT
bacteria to 2× MIC of 29 for limited periods of time, i.e. conditions where only a modest
decrease in bacterial counts was observed (∆ log CFU/mL = 1). The metabolite levels (
Ala and -Ala- -Ala) were analyzed over 30 min.
As shown in Figure 7, the [ -Ala]/[ -Ala- -Ala] ratio increased over time in these conditions.
L-Ala,
D-
D
D
D
D
D
This experiment thus proves that 1-(2-hydroxybenzoyl)-thiosemicarbazides exert their
antimicrobial activity by targeting Ddl in bacterio.
Figure 7. Evolution over time of S. aureus
×MIC of compound 29 compared to the control (DMSO 5%) and the reference antibiotic, DCS
2×MIC). All values are presented as the (mean ± SD) of measures performed in triplicate. A
L-Ala,
D-Ala and
D
-Ala- -Ala levels in presence of
D
2
(
multiple comparison (two-way ANOVA) led to the following statistical results: **** extremely
significant (P value < 0.0001), *** extremely significant (P value from 0.0001 to 0.001), ** very
significant (P value from 0.001 to 0.01), * significant (P value from 0.01 to 0.05), no asterisk
means not significant (P value ≥ 0.05).
24

ACCEPTED MANUSCRIPT
3.2.4. Cytotoxicity of model compound 29. The representative inhibitor 29 was then assayed for
its potential cytotoxicity on THP-1 human monocytic cell line. Figure 8 shows the evolution of
cell survival as a function of time and inhibitor concentration. A significant but moderate (12-
2
5%) decrease in viability was observed after 24 h of incubation over the range of concentrations
investigated. This suggests that the compound is well tolerated at the concentrations showing
antibacterial activity (1× MIC or 0.1 mM).
Figure 8. Cytotoxicity on THP-1 human monocytic cell line. Percentage of cell survival as a
function of time (2 h, 6 h and 24 h) and compound 29 concentration (0.1 mM - 1× MIC, 0.2 mM
-
2× MIC and 0.4 mM - 4× MIC) compared to the control (DMSO 1%). All values are presented
as the (mean ± SD) of measures performed in triplicate. A multiple comparison (two-way
ANOVA) led to the following statistical results: **** extremely significant (P value < 0.0001),
*
0
** extremely significant (P value from 0.0001 to 0.001), ** very significant (P value from
.001 to 0.01), * significant (P value from 0.01 to 0.05), no asterisk means not significant (P
value ≥ 0.05).
25

ACCEPTED MANUSCRIPT
3.2.5. Selectivity profile of 29. Finally, these encouraging results prompted us to investigate the
selectivity of 29. To this effect, we tested its inhibitory activity against various enzymes and
receptors at 10 µM, i.e. a concentration approx. 7 times higher than its IC towards Ddl (Table
50
4
). From these data we can conclude that the 4-(3,4-dichlorophenyl)-1-(2-hydroxybenzoyl)-3-
thiosemicarbazide 29 is highly selective for Ddl as none of the investigated enzymes or receptors
are impacted in these conditions apart from the angiotensin converting enzyme, ACE, that is
weakly inhibited (66%) by this compound. Despite that the literature describes the parent
benzoylthiosemicarbazide 3 as a CA I and CA IX inhibitor, carbonic anhydrase II was not
impacted by 29.
These results demonstrate that the benzoylthiosemicarbazide motif is a selective and efficient
pharmacological tool to target Ddl.
Table 4. Selectivity profile of compound 29.
%
Enzyme or receptor
Radioligand
Control inhibitor (IC , nM)
50
Inhibition
3
3
5
-HT1 (h)
[ H]8-OH-
[ H]8-OH-DPAT (0.6)
NI
A
DPAT
ACE (h)
-
Captopril (0.7)
66
NI
Acetylcholinesterase (h)
Arg1 (h)
-
-
-
-
Galanthamine (520)
S-(2-boronoethyl)-l-cysteine (13 µM)
Clavulanic acid (11)
Acetazolamide (29)
CP55940 (0.58)
NI
β-Lactamase (Bacillus cereus)
Carbonic anhydrase II (h)
NI
NI
3
CB (h)
[ H]CP 55940
NI
1
COX (h)
-
-
Diclofenac (8)
NI
1
NI a
DNA Gyrase (E. coli and S. aureus)
-
26

ACCEPTED MANUSCRIPT
3
GABAA1 (α1,β2,γ2) (h)
IDO (h)
[ H]Muscimol
Muscimol (75)
NI
NI
NI
NI
NI
NI
NI
NI
NI
NI
NI
NI
NI
NI a
-
-
-
1-L-Methyltryptophan (37 µM)
Staurosporine (31)
Oxamic acid (577 µM)
Pirenzepine (22)
Lck kinase (h)
LDH (h)
3
M (h)
1
[ H]Pirenzepine
3
MAG lipase (h)
MAO-A (h)
PDE3A (h)
[ H]-2-OG
Disulfiram (363)
-
-
-
-
Clorgiline (40)
Milrinone (490)
PHGDH (h)
1-(4-Chlorophenyl)-2-morpholino-2-
thioxoethanone (8.7 µM)
sPLA
2
(type V) (h)
Oleyloxyethyl phosphorylcholine (20 µM)
3
Potassium Channel hERG (h)
PPARα (h)
[ H]Dofetilide
Terfenadine (80)
GW7647 (52)
PPACK (0.103)
-
3
[ H]WY 14643
Thrombin (h)
-
-
Topo IV (E. coli and S. aureus)
a
57
from reference ; NI = inhibition ≤ 30% at 10 µM; Significance criteria (Eurofins Cerep): ≥
5
0% of inhibition
3
.3. Molecular docking
Because the binding mode of 29 was established to be non-competitive, we performed a
preliminary molecular docking study at the known allosteric site of Ddl. To this end, we used the
1
2
structure of StaDdl available in the Protein Databank (PDB code: 2I80) and docked 29 by mean
of the automated Gold program with the aim to understand at the molecular level the interactions
stabilizing the benzoylthiosemicarbazide in Ddl. To be consistent with the potential existence of
27

ACCEPTED MANUSCRIPT
tautomeric forms in this series of compounds, both tautomers of the model
benzoylthiosemicarbazide 29 were investigated.
As a result, the binding mode of 29 in both tautomeric forms (Figure 9a) supports the
hypothesis that the right part of the inhibitor (R ) could partially occupy a hydrophobic pocket
2
(color code in green) open to the substrate binding site, and that no crucial interaction is made by
this aromatic moiety apart from stabilizing hydrophobic interactions. At the bottom of the
allosteric site, a quadrupolar T-shape interaction between Phe313 and the phenol ring of 29
ensures an additional stabilization. Moreover, it can be observed that both tautomers fit very well
this pocket thanks to an almost planar geometry stabilized through H-bond interactions between
the thiosemicarbazide function and the 2-hydroxyphenyl moiety. This intramolecular H-bond
was already mentioned in the X-ray structure of 29 (Figure 3). In the “keto” form (Figure 9b),
the hydroxyl in the 2-position of 29 would be stabilized through H-bonding with the carbonyl
oxygen of Pro311 backbone, whereas in the “enol” form (Figure 9c), the compound would be
stabilized through H-bonds between the carbonyl oxygen of Pro311 and the thiol function, and
between the carbonyl oxygen of Leu94 and the enol function. Whether the “keto” or the “enol”
tautomeric form is being considered, this molecular modeling analysis strongly supports the SAR
reported earlier in this work, particularly regarding the key role of the hydroxyl function in the 2-
position of the aryl-thiosemicarbazide for the proper conformation and interaction with Ddl.
28