Design, synthesis and biological evaluation of antimicrobial diarylimine and –amine compounds targeting the interaction between the bacterial NusB and NusE proteins

Article abstract of DOI:10.1016/j.ejmech.2019.05.090

Discovery of antimicrobial agents with a novel model of action is in urgent need for the clinical management of multidrug-resistant bacterial infections. Recently, we reported the identification of a first-in-class bacterial ribosomal RNA synthesis inhibitor, which interrupted the interaction between the bacterial transcription factor NusB and NusE. In this study, a series of diaryl derivatives were rationally designed and synthesized based on the previously established pharmacophore model. Inhibitory activity against the NusB-NusE binding, circular dichroism of compound treated NusB, antimicrobial activity, cytotoxicity, hemolytic property and cell permeability using Caco-2 cells were measured. Structure-activity relationship and quantitative structure–activity relationship were also concluded and discussed. Some of the derivatives demonstrated improved antimicrobial activity than the hit compound against a panel of clinically important pathogens, lowering the minimum inhibition concentration to 1–2 μg/mL against Staphylococcus aureus, including clinical strains of methicillin-resistant Staphylococcus aureus at a level comparable to some of the marketed antibiotics. Given the improved antimicrobial activity, specific inhibition of target protein-protein interaction and promising pharmacokinetic properties without significant cytotoxicity, this series of diaryl compounds have high potentials and deserve for further studies towards a new class of antimicrobial agents in the future.

Full text of DOI:10.1016/j.ejmech.2019.05.090

Accepted Manuscript
Design, synthesis and biological evaluation of antimicrobial diarylimine and –amine
compounds targeting the interaction between the bacterial NusB and NusE proteins
Yangyi Qiu, Shu Ting Chan, Lin Lin, Tsun Lam Shek, Tsz Fung Tsang, Nilakshi
Barua, Yufeng Zhang, Margaret Ip, Paul Kay-sheung Chan, Nicolas Blanchard, Gilles
Hanquet, Zhong Zuo, Xiao Yang, Cong Ma
PII:
S0223-5234(19)30514-8
DOI:
https://doi.org/10.1016/j.ejmech.2019.05.090
EJMECH 11400
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 13 April 2019
Revised Date: 30 May 2019
Accepted Date: 31 May 2019
Please cite this article as: Y. Qiu, S.T. Chan, L. Lin, T.L. Shek, T.F. Tsang, N. Barua, Y. Zhang, M.
Ip, P.K.-s. Chan, N. Blanchard, G. Hanquet, Z. Zuo, X. Yang, C. Ma, Design, synthesis and biological
evaluation of antimicrobial diarylimine and –amine compounds targeting the interaction between the
bacterial NusB and NusE proteins, European Journal of Medicinal Chemistry (2019), doi: https://
doi.org/10.1016/j.ejmech.2019.05.090.
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
Design, Synthesis and Biological Evaluation of
Antimicrobial Diarylimine and –amine Compounds
Targeting the Interaction between the Bacterial NusB
and NusE Proteins
a,b
a,b
c
d
c
Yangyi Qiu , Shu Ting Chan , Lin Lin , Tsun Lam Shek , Tsz Fung Tsang , Nilakshi
c
d
c
c,e
f
Barua , Yufeng Zhang , Margaret Ip , Paul Kay-sheung Chan , Nicolas Blanchard ,
g
d,
c,
a,b,
Gilles Hanquet , Zhong Zuo ***, Xiao Yang **, Cong Ma
*
a
The Hong Kong Polytechnic University Shenzhen Research Institute, Shenzhen, China
b
State Key Laboratory of Chemical Biology and Drug Discovery, and Department of Applied
Biology and Chemical Technology, The Hong Kong Polytechnic University, Kowloon, Hong
Kong SAR
c
Department of Microbiology, The Chinese University of Hong Kong, Prince of Wales
Hospital, Shatin, Hong Kong SAR
d
School of Pharmacy, Faculty of Medicine, The Chinese University of Hong Kong, Shatin,
New Territories, Hong Kong SAR
e
Stanley Ho Centre for Emerging Infectious Diseases, The Chinese University of Hong Kong,
Shatin, Hong Kong SAR
1

ACCEPTED MANUSCRIPT
f
Université de Haute-Alsace, Université de Strasbourg, CNRS, LIMA, UMR 7042,
Mulhouse, 68000, France
g
Laboratoire d’Innovation Moléculaire et Applications, ECPM, UMR 7042, Université de
Strasbourg/Université de Haute-Alsace, CNRS, 67000 Strasbourg, France
Corresponding Authors:
E-mail: cong.ma@polyu.edu.hk (C.M.), xiaoyang@cuhk.edu.hk (X.Y.),
joanzuo@cuhk.edu.hk (Z.Z.)
ABSTRACT: Discovery of antimicrobial agents with a novel model of action is in urgent
need for the clinical management of multidrug-resistant bacterial infections. Recently, we
reported the identification of a first-in-class bacterial ribosomal RNA synthesis inhibitor,
which interrupted the interaction between the bacterial transcription factor NusB and NusE.
In this study, a series of diaryl derivatives were rationally designed and synthesized based on
the previously established pharmacophore model. Inhibitory activity against the NusB-NusE
binding, circular dichroism of compound treated NusB, antimicrobial activity, cytotoxicity,
hemolytic property and cell permeability using Caco-2 cells were measured. Structure-
activity relationship and quantitative structure–activity relationship were also concluded and
discussed. Some of the derivatives demonstrated improved antimicrobial activity than the hit
compound against a panel of clinically important pathogens, lowering the minimum
inhibition concentration to 1-2 µg/mL against Staphylococcus aureus, including clinical
strains of methicillin-resistant Staphylococcus aureus at a level comparable to some of the
2

ACCEPTED MANUSCRIPT
marketed antibiotics. Given the improved antimicrobial activity, specific inhibition of target
protein-protein interaction and promising pharmacokinetic properties without significant
cytotoxicity, this series of diaryl compounds have high potentials and deserve for further
studies towards a new class of antimicrobial agents in the future.
Keywords: antimicrobial activity, diarylimine, diarylamine, protein-protein interaction,
inhibitor, methicillin-resistant Staphylococcus aureus
1
. Introduction
Antibiotic-resistant Staphylococcus aureus such as methicillin-resistant S. aureus (MRSA)
strains re-emerged to become problematic causes of severe infections. Bacterial transcription
is a valid but underutilized target for antibiotic discovery [1]. We previously discovered
inhibitors interrupting the protein-protein interaction (PPI) between the core enzyme in
bacterial transcription, RNA polymerase and the initiation factor sigma [2,3]. Recently, we
reported a first-in-class inhibitor as a hit compound identified by in silico screening of a
pharmacophore model designed based on the crystal structures of two bacterial proteins,
NusB and NusE [4]. These two proteins are essential and highly conserved transcription
factors specifically involved in the bacterial ribosomal RNA (rRNA) transcription. By
examining the PPI interface of the crystal structure of Escherichia coli NusB-NusE complex
[
5], the interactions between NusB Glu81 and NusE His15, NusB Tyr18 and NusE Asp19,
NusB Glu75 and NusE Arg16, and NusB Leu22 and NusE Val26 (E. coli amino acid residue
numbering) were chosen to design the pharmacophore model (Figure 1), followed by in silico
screening [4]. One of the short-listed compounds, (E)-2-{[(3-ethynylphenyl)imino]methyl}-
4
-nitrophenol (1) was able to specifically inhibit the NusB-NusE interaction, decrease the
3

ACCEPTED MANUSCRIPT
level of bacterial rRNA and demonstrate antimicrobial activities with a minimum inhibition
concentration (MIC) at 8 µg/mL against an antibiotic susceptibility control strain S. aureus
ATCC 25923, as well as a number of clinical MRSA strains (MIC 16-64 µg/mL) without
significant cytotoxicity against human A549 and HaCaT cell lines. It was the first time the
NusB-NusE interaction has been used for inhibitor discovery to identify a small molecule
with promising antimicrobial activities and cytotoxicities as a hit compound. Herein we
extend our study to the structure-activity relationship (SAR) of this compound for
comprehensive understanding of the inhibitory mechanism, and the possibility of further
development as antimicrobial drug candidates.
Figure 1. Structure of the E. coli NusB-NusE co-crystal complex (left; PDB: 3D3B),
conserved interactions across bacterial species at the interface (middle) and pharmacophore
model construction (right).
2
. Chemistry
2
.1 Derivative Design
4

ACCEPTED MANUSCRIPT
The bioactive hit compound, 1 has a relatively low molecular weight of 263 and a specific
conjugated diphenyl imine system. As predicted by the pharmacophore model, 1 falls into a
basin-like surface of NusB, which is a hydrophobic semi-pocket (Figure 2, left). As indicated
in Figure 2, the Glu75 residue of NusB forms a hydrophobic interaction with the ethynyl
group of 1, Glu81 bonds to the phenolic hydroxyl group of 1 by hydrogen bond, and the nitro
group of 1 accepts a hydrogen bond donation from Tyr18 of NusB.
Figure 2. The docking model showing the interaction of the hit compound 1 with E. coli
NusB (PDB:3D3B) with the hydrophilicity of NusB surface is colored in blue (left) and the
modification plan of the hit compound 1 (right).
To examine the pharmacophore model and study the SAR, we planned to modify first the left
and right benzene rings of the hit compound 1 (Figure 2, right). Various substituents such as
alkyl, hydroxyl, chloride, trifluoromethyl have been used to replace ethynyl on the left
benzene ring of the hit compound 1 to test the suitability for hydrophobic interaction with
Glu75 of NusB.
5

ACCEPTED MANUSCRIPT
New analogues with or without the nitro group, hydroxyl of phenol, masked hydroxyl by
methoxy, replacement of nitro by other hydrogen bond accepters, electron-withdrawing or –
donating groups to alter the pKa of phenol have been tested to confirm the important
hydrogen bonds between right benzene ring of 1 with Glu81 and Tyr18 of NusB.
The imine linker of the hit compounds has also been reduced to amine to study how the
relative position of two benzene rings of hit compounds affects the NusB-NusE interaction
and thus the inhibitory activity against bacteria (Figure 2). As the flexibility of imine and
amine linkers are distinct, while bond lengths are also slightly different, we intended to
examine whether the entropy penalty played an important role in protein target affinity and
antimicrobial activity because of the flexible diaryl system with an amine linker.
2
.2 Synthesis
Synthesis of the left ring modified derivatives 1-35 is shown in Scheme 1. Reaction of amines
or substituted aniline 1-19a with 5-nitrosalicylaldehyde in ethanol at room temperature
afforded Schiff base type derivatives 1-19 in 60-93% yields. Reductive amination of the
above derivatives using sodium triacetoxyborohydride in dichloromethane gave the
corresponding amines 20-35 in 86-94% yields. The intermediate 2a was not commercially
available, so we prepared it by a Sonogashira coupling reaction from 2-iodoaniline in 59%
yield [6].
6

ACCEPTED MANUSCRIPT
Scheme 1. Synthesis of a) the target compounds 1-19, 20-35 and b) intermediate 2a
Synthesis of the right ring modified derivatives 36-60 was performed following the same
synthetic method of imine formation and reductive amination as outlined in Scheme 2.
Regarding the commercially unavailable substituted benzaldehydes as reaction partners, a
modified Casnati-Skattebøl ortho-formylation furnished aldehyde intermediates 36a, 39a and
4
0a in 52-78% yields [7]. Although 39a was obtained as a 7:3 mixture of 2-(3-formyl-4-
hydroxyphenyl)acetonitrile and the starting material 4-hydroxybenzyl cyanide, as determined
1
by H NMR analysis, this mixture could be used directly in the subsequent reaction step, as 4-
hydroxybenzyl cyanide would not participate in the imine formation reaction and could be
then easily removed through filtration. Nucleophilic aromatic substitution of 2-
7

ACCEPTED MANUSCRIPT
fluorobenzaldehydes using sodium methoxide in methanol provided benzaldehydes 38a and
4
3a in 85-88% yields. Demethylation of compound 38a with boron tribromide in
dichloromethane provided cyanobenzaldehyde 38b in 71% yield [8]. In summary, Schiff base
type derivatives 36-48 and amines 49-60 were synthesized in 65-87% and 86-94% yields,
respectively.
Scheme 2. Synthesis of a) the target compound 36-48, 49-60; b) intermediates 36a, 39a, 40a;
c) intermediates 38b, 43a.
8

ACCEPTED MANUSCRIPT
3
. Results and Discussion
3
.1 Inhibitory activity against the NusB-NusE interaction.
A biochemical inhibitory assay for testing the capacity of disrupting the NusB-NusE PPI has
been developed using the luciferase complementation technique [9]. The NusB and NusE
proteins were each fused to one of the luciferase complementation fragment and their
interaction resulted in reformation of the natural luciferase. Binding of NusB to the inhibitor
prevented its interaction with NusE, therefore caused decreased luminescence released from
complemented luciferase catalyzed reaction. As the percentage inhibition at a single
concentration correlated perfectly with the IC value [9], we maintained the previous testing
5
0
conditions of compounds at 100 µM and displayed the percentage of inhibition values in
Tables 1 and 2.
Table 1 includes the testing results of compounds 1-19 with modifications of the left ethynyl
benzene ring from the hit compound, and the corresponding analogues 20-35 containing an
amine bond reduced from the imine group. As shown in Table 1, the modifications of the left
ethynyl benzene ring were able to affect the inhibitory activity against the NusB-NusE PPI
with the percentage inhibition values varied from 16.8% ± 0.7% (compound 5) to 96.3% ±
0
.6% (compound 9) compared to the hit compound 1 at 85.3% ± 1.5%. Compound 5 contains
a t-butyl group at 2-position related to imine, which seems too bulky for binding to NusB,
and compound 9 has a 3-hydroxyl group as highly preferred. Other derivatives with relatively
high inhibitory values includes compound 2 with 2-ethynyl (71.4% ± 1.8%), compound 3
with 4-ethynyl (76.3% ± 3.0%), compound 10 with 4-hydroxyl (89.1% ± 1.1%), and
9

ACCEPTED MANUSCRIPT
compound 17 with 3-carboxylic acid (78.5% ± 2.6%). In summary, the substitution of the left
benzene ring is necessary as most of the substituted compounds demonstrated improved
inhibitory activities compared to compound 4 with an unsubstituted benzene ring (39.2% ±
3
.0%). Additionally, our pharmacophore docking model indicated the binding of the
substituents to Glu75 of NusB played an important role in the inhibition, and the binding type
belongs to the hydrophobic interaction with the residue chain of Glu75. Effectively, the
percentage inhibition values suggested that the inhibitory activity could be enhanced
regardless hydrophobic or hydrophilic substituents, which correlates the hydrophobic
interaction dominate the binding of Glu75 to our compounds using the aliphatic chain. By
summarizing the relative positions of substitutions on the left benzene ring to the inhibitory
activities, it is suggested that the 3-position was preferred for t-butyl, hydroxyl, chloride,
ethynyl, and 4-position for CF3.
The inhibitory activity of amines 20-35 obtained from reduction of imines 1-19 were then
subjected to the testing. As the molecular conformation of 1-19 was released from a quasi-
plane to a freely rotatable structure in 20-35, we can draw such a prediction that some
compounds may express improved inhibitory activities because of flexible structures
promoting better binding affinity with NusB, such as compound 22. This compound
containing a t-butyl group at 2-position on the left benzene ring has a significant
improvement on percentage inhibition at 72.2% ± 2.7% from its imine analogue 5 at 16.8% ±
0
.7% as the imine with the lowest activity in Table 1. By turning the rigid imine to flexible
amine, the bulky t-butyl group at the 2-position proved to adjust the conformation to
accommodate this steric hindrance and displayed appropriate hydrophobic interaction with
the amino acid residue of NusB. Another case is compound 25 with an increased percentage
10

ACCEPTED MANUSCRIPT
inhibition at 81.5% ± 4.0%, compared to the imine analogue 8 at 57.7% ± 2.5%. In the
contrast, other compounds display similar or slightly weakened inhibitory activities due to
entropy penalty. A special case is amine compound 21 with an unsubstituted benzene ring,
which lost completely the protein affinity to NusB with a percentage inhibition at 4.6% ±
3
.4% compared to the imine analogue 4.
Table 1. Inhibitory and antimicrobial activities of compounds 1-35
a
Compound
R
MIC
8
% Inhibition
85.3 ± 1.5
47.7 ± 7.9
71.4 ± 1.8
ClogP
3.32
1
2
0
16
4
3.02
2
3.32
3
4
4
76.3 ± 3.0
3.32
4
16
4
39.2 ± 3.0
4.6 ± 3.4
16.8 ± 0.7
3.05
2.75
4.88
2
1
5
2
2
1
4
72.2 ± 2.7
51.3 ± 1.5
4.58
4.88
6
2
3
2
39.5 ± 2.7
4.58
7
4
2
38.8 ± 2.5
10.8 ± 5.4
4.88
4.58
2
4
8
4
16
16
57.7 ± 2.5
81.5 ± 4.0
96.3 ± 0.6
2.38
2.08
2.38
2
5
9
2
6
64
66.7 ± 3.5
2.08
11

ACCEPTED MANUSCRIPT
1
0
4
89.1 ± 1.1
2.38
1
2
1
1
7
2
2
2
2
45.7 ± 2.2
31.0 ± 9.2
66.5 ± 0.7
3.93
3.77
3.93
2
8
4
34.9 ± 5.8
3.77
1
2
1
3
9
4
8
8
8
50.0 ± 3.8
53.6 ± 2.0
53.1 ± 2.9
3.93
3.77
4.23
3
1
0
5
2
61.5 ± 4.0
56.2 ± 2.7
4.16
4.23
16
3
1
4
61.7 ± 3.7
4.16
1
3
6
2
8
8
4
68.9 ± 2.1
54.1 ± 3.6
4.23
4.16
1
7
78.5 ± 2.6
2.99
>256
3
3
50.8 ± 3.0
2.84
1
3
1
8
4
9
4
>256
4
68.3 ± 2.2
58.2 ± 5.9
43.4 ± 3.0
3.87
3.20
4.22
3
5
4
47.4 ± 2.8
3.93
a
%
Inhibition displayed the percentage of inhibition on the NusB-NusE
interaction using the previously described conditions [9]: the compound was
used at 100 µM in the NusB-NusE (1:1) solution at 1 µM in PBS buffer.
The imine compounds 36-48 with modified right benzene rings and the corresponding amine
analogues 49-60 have been tested and results were shown in Table 2. In this study, we
focused on the modification of hydroxyl and nitro groups. Regarding the hit compound 1
with a percentage inhibition value of 85.3% ± 1.5%, both hydroxyl and nitro groups of 1
seemed necessary as compound 41 without nitro and 42 without hydroxyl showed significant
loss of inhibitory activity with respective percentage inhibition values of 25.7% ± 7.9% and
12

ACCEPTED MANUSCRIPT
2
4.7% ± 1.9%, as predicted by our rational design using the pharmacophore model. The
transformation from phenol to methyl ether 43 also led to lowered activity (40.4% ± 1.9%).
m-nitrophnol 44, or p-nitrophenol 46 with relatively changed position to the imine group,
exhibited decreased inhibitory activity (46.7% ± 2.5%, 52.2% ± 3.1%). Nitro is a critical
group, as it processes the dual effect of serving as a hydrogen bond acceptor and electron-
withdrawing group affecting the acidity of p-nitrophenol, which is supposed to bind to Glu81
of NusB by the design of hit compound. The replacement of nitro by methoxy in compound
4
0 kept the hydrogen bonding capability, however, lost the inhibitory activity (29.3% ±
5
.6%) as methoxy is an electron-donating group and capable of increasing the pKa of phenol.
When nitro was replaced by other hydrogen bond acceptors and electron-withdrawing groups
such as carboxylate, fluoride, nitrile, acetonitrile and chloride, the respective compounds 36
(41.4% ± 3.7%), 37 (28.1% ± 2.4%), 38 (61.3% ± 2.0%), 39 (49.9% ± 1.2%) and 48 (54.2%
±
2.8%) displayed noticeably decreased inhibitory activity compared to the hit compound 1.
It is suggested that the pKa of phenol is more important for the inhibitory activity, as o-
nitrophenol 45 that lost the hydrogen bond at p-position of hydroxyl but maintained a similar
pKa to p-nitrophenol still displayed excellent inhibitory activity (87.7% ± 0.6%).
Additionally, an extra o-bromo group to p-nitrophenol 47 trying to lower further the pKa did
not affect the activity.
Amine 49-60 demonstrated more favorable changes related to inhibitory activity compared to
amine compounds with modifications on the left benzene ring. Amines 49, 50, 53, 55, 56, 59
possessed significant improvement over the imine analogues, especially 59 with a measured
9
9.6% ± 0.1% as the best percentage inhibition so far, indicating that the low pKa of phenol
by additional o-bromo group is favored, however, affected by the structural rigidity of 47.
13

ACCEPTED MANUSCRIPT
When the molecule can freely rotate, the improved binding affinity overpowered the entropy
penalty to provide a better inhibitory activity. While amines 51, 52, 54, 57, 60 with various
substituents showed similar activity to the imine analogues, suggesting the binding force are
mainly from hydrogen bonds, which were sufficiently robust to maintain the binding to
NusB, in contrast to the left benzene ring modifications that were more sensitive to entropy
penalty, indicating the Van der Waals interactions. The only dramatic decrease in activity
occurred on compound 58 (64.5% ± 1.7%). Despite of maintaining pKa of o-nitrophenol, the
loss of hydrogen bond between the nitro group and Tyr18 of NusB could not retain the
binding of 58 with NusB very firmly as compound 45 by the conformation control as a Schiff
base.
We have measured the IC value of 27, which was 69.2 ± 2.4 µM. This value was at the
50
same range to the hit compound 1 [4], suggesting a similar inhibitory activity targeting the
NusB-NusE interaction.
Table 2. Inhibitory and antimicrobial activities of compounds 36-60
a
Compound
Ar
MIC
%Inhibition
ClogP
3
4
6
9
128
32
41.4 ± 3.7
57.1 ± 0.8
3.51
3.15
3
5
7
0
256
32
28.1 ± 2.4
61.8 ± 2.0
3.60
3.08
3
5
3
8
1
9
32
32
256
61.3 ± 2.0
64.8 ± 5.4
49.9 ± 1.2
3.03
2.76
2.78
14

ACCEPTED MANUSCRIPT
5
4
5
4
5
4
5
4
5
4
5
4
2
0
3
1
4
2
5
3
6
4
7
5
64
>256
64
256
>256
256
>256
>256
>256
64
44.7 ± 1.4
2.06
3.46
2.74
3.35
2.64
2.86
3.10
3.31
3.32
3.32
3.02
3.32
29.3 ± 5.6
58.3 ± 0.9
25.7 ± 7.9
20.3 ± 1.9
24.7 ± 1.9
33.1 ± 2.6
40.4 ± 1.9
56.1 ± 1.4
46.7 ± 2.5
48.2 ± 1.9
87.7 ± 0.6
8
256
5
8
128
256
64.5 ± 1.7
3.02
4
6
52.2 ± 3.1
3.32
4
5
4
6
7
9
8
0
32
2
74.1 ± 1.6
99.6 ± 0.1
54.2 ± 2.8
50.6 ± 4.9
3.94
3.70
4.59
4.14
32
2
a
%
Inhibition displayed the percentage of inhibition on the NusB-NusE interaction using
the previously described conditions [9]: the compound was used at 100 µM in the NusB-
NusE (1:1) solution at 1 µM in PBS buffer.
3
.2 Circular dichroism.
We considered using circular dichroism spectroscopy to detect the influence of representative
derivatives on the NusB protein folding, as different secondary structure has a specific
circular dichroism signature [10]. Compounds 5 and 22 were chosen as they represent two
different linkers: imine and amine that may affect the special position of the two benzene
rings and thus the binding affinity to NusB. We first examined the NusB protein at 2.0 µg/mL
in neutral phosphate buffer, which displayed a spectrum showing a typical α-helical character
(Figure 3), consistent to the multiple α-helices observed by in the X-ray crystallography
structure (PDB: 3D3B) [11].
15

ACCEPTED MANUSCRIPT
When compounds 5 and 22 were respectively mixed with NusB at 1:1 ratio, the changes on
spectra could be observed as a result of the protein-ligand binding. The chiral character of
NusB decreased by binding to one of the derivatives. When compounds 5 and 22 were
respectively added to NusB, similar changes of the CD spectra were observed, which
indicated that the derivatives induced similar structural changes of NusB. In summary, the
CD spectra showed that the two representative compounds were able to cause similar
conformational changes of NusB, suggesting they may bind to the same binding site on the
protein.
Figure 3. CD spectrum of NusB before and after addition of compounds 5 or 22
3
.3 Antimicrobial activity.
16

ACCEPTED MANUSCRIPT
The antimicrobial activity of the compounds was determined by the microdilution method
according to the Clinical & Laboratory Standards Institute (CLSI) guidelines against S.
aureus ATCC 25923 [12].
As shown in Table 1, amongst the compounds 1-19 with imine as the linker, most of the
modifications of left benzene ring led to equal or superior antimicrobial activity (MIC 2-8
µg/mL) compared to the hit compound 1, in which compounds 11 and 12 with chloride
displayed the best activity with MIC values of 2 µg/mL. The only compounds demonstrating
inferior activity were 9 and 15 with hydroxyl and trifluoromethyl at 3-position. The MIC
values of 9 and 15 against S. aureus ATCC 25923 were 16 µg/mL, similar with the hit
compound regarding the data interpretation in antibiotic susceptibility testing. These results
proved again that the binding of left benzene ring to NusB belongs to the hydrophobic
interaction. It is worth noting that the in vitro inhibitory activity of compounds does not fully
correspond to the antimicrobial activity, with an extreme example for comparison between
compounds 5 and 9 demonstrating the lowest and highest inhibitory activity, but they
displayed reversed antimicrobial activity with MIC values of 4 and 16 µg/mL, respectively.
As the cell system is much more complicated than the in vitro biochemical assay system, we
need to consider the impact of cell permeability, efflux system and metabolic enzymes
besides the interaction of two partner proteins. Effectively, the clogP values of compounds 5
and 9 are diverse at 4.88 and 2.38, respectively, which may affect their permeability.
When compounds 20-35 with amine as the linker were subjected to the antimicrobial testing,
most of the compounds exhibited equal or inferior antimicrobial activity to their imine
analogues, except compounds 22-24 with t-butyl at 2-, 3- and 4-position of left benzene ring,
which presented superior antimicrobial activity with MIC values of 1-2 µg/mL. Compound
17

ACCEPTED MANUSCRIPT
2
2 displayed the best antimicrobial activity, with an MIC of 1 µg/mL. In summary, the
antimicrobial activity of these series of compounds follows the same trend of their inhibitory
activity, which strengthened our hypothesis that the NusB-NusE interaction is the target of
these compounds.
As shown in Table 2, compounds 36-48 with modifications of the right benzene ring changed
nitro and hydroxyl as the hydrogen binding groups, led to dramatic decrease of antimicrobial
activity with MIC values of 32-256 µg/mL, in which compound 40 with methoxy replacing
nitro, and 43 with methyl ether masking hydroxyl lost their antimicrobial activity at all (>256
µg/mL). As these compounds have imine as the linker, once hydrogen bonding groups were
modified, they were not capable of adjusting the rigid structures to accommodate the required
binding in a more dynamic cellular environment. The possible interactions of NusB with
other proteins that may also competitively change the conformation of NusB. However, when
imine was replaced by amine, compounds 49-60 with flexible structures were able to improve
significantly the antimicrobial activity with MIC values of 2-64 µg/mL. The best
antimicrobial activity (2 µg/mL) of these series of compounds were obtained from compound
5
9 with an additionally bromide at o-position to phenol, and compound 60 with dichloride
replacing nitro and bromide in compound 59. This series of compounds again proved the
importance of hydrogen bonding groups, as compounds 41-43, 54-56 with truncated or
masked pharmacophores lost completely the antimicrobial activity regardless imine or amine
as linkers.
With 60 derivatives of the hit compound in hand, we can conclude a preliminary SAR of the
derivatives on the basis of their antimicrobial activity against S. aureus ATCC 25923 (Figure
4
).
18

ACCEPTED MANUSCRIPT
Figure 4. SAR of synthesized derivatives of the hit compounds against S. aureus ATCC
2
5923
On the other hand, we also tried to align the activity to the structure and cell permeability. As
shown in Figure 5, we summarized the in vitro inhibitory activity of NusB-NusE interaction
related to the antimicrobial activity against S. aureus 29523 in a bubble chart with the bubble
size representing clogP of compounds. It is clear that the MIC values are inversely
proportional to the percentage inhibition of compounds, which indicated that the
antimicrobial activity and in vitro inhibitory activity are in direct proportion. When we
plotted the clogP values against the antimicrobial activity with bubble sizes demonstrating
percentage inhibition, we could find out that the clogP between 2.5 and 5.0 is also inversely
proportional to the MIC values. The statistics of these data suggested that the antimicrobial
activity of our inhibitors has a visible relation to the inhibitory activity and cell membrane
permeability of compounds.
19

ACCEPTED MANUSCRIPT
Figure 5. The trend of the inhibitory activity (left y axis) and ClogP (right y axis) are
inversely proportional to the antimicrobial activity shown by MIC values (x axis). The size of
spheres represents the propotional values of ClogP (left) or inhibitory activity (right).
2
Both of the values of the coefficient of determination (R ) of inhibitory activity – MIC and
ClogP – MIC were > 0.1, which suggested a certain degree of correlations. While these
values were still modest, we need to consider that there are a number of other factors
contributing to a compound’s antimicrobial activities, such as solubility, membrane
permeability, potential bacterial enzyme modification/degradation, and bacterial efflux.
Investigations of other possible bacterial targets for our compounds would also provide
information on the factors affecting the antimicrobial activity.
We then intended to examine the antibacterial spectrum of this series of molecules targeting
the NusB-NusE interaction. The first bacteria set used is top six pathogens amongst the most
20

ACCEPTED MANUSCRIPT
recent “WHO priority pathogens list for guiding R&D of new antibiotics” containing three
Gram-positive and three Gram-negative bacteria: Enterococcus faecium, S. aureus,
Streptococcus pneumonia, Acinetobacter baumannii, Pseudomonas aeruginosa, and
Enterobacter spp.. We used the standard strains for antibiotic susceptibility testing and
conditions to perform the experiment, and the result was shown in Table 3: the compounds
demonstrated broad-spectrum antimicrobial activity against both Gram-positive and -negative
bacteria, as the NusB-NusE interaction is essential and conserved in bacteria. Additionally,
these compounds displayed specific activities against S. aureus, S. pneumonia, E. faecium
and A. baumannii over other pathogens. S. aureus, S. pneumonia, and E. faecium are all
Gram-positive bacteria, suggesting Gram-positive bacteria may be more susceptible to these
compounds. Encouragingly, A. baumannii is top one bacterium of the WHO pathogen list that
requires new antibiotic development, our compounds may be developed as a specific
antimicrobial agent to address this urgent need. In the contrast, other Gram-negative bacteria
were less responsive to the compound treatment, probably due to reduced cell permeability
and vigorous efflux system, which need further testing in the future.
Table 3. Antimicrobial activity evaluation of compounds 1-60 against seven representative
bacterial strains including two S. aureus strains in the WHO priority pathogens list for
guiding R&D of new antibiotics.
a
b
Compound
EFAE
>256
>256
128
>256
>256
>256
>256
>256
>256
>256
SAUR
SAUR
16
16
16
16
16
16
32
16
SPNE
16
32
64
32
ABAU
64
>256
>256
256
>256
128
>256
128
PAER
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
ECLO
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
1
2
3
4
5
6
7
8
9
8
4
4
4
4
4
4
4
16
4
64
256
128
64
256
64
64
8
>256
>256
1
0
21

ACCEPTED MANUSCRIPT
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
>256
>256
256
256
>256
256
>256
>256
>256
16
128
8
32
>256
256
>256
32
2
2
8
8
16
8
4
4
4
16
16
1
2
2
16
64
2
4
8
16
16
32
16
64
16
16
8
16
16
32
4
4
4
32
128
2
4
64
64
64
32
32
16
16
16
32
8
64
8
8
8
64
128
8
8
16
8
16
16
>256
>256
16
128
256
128
>256
>256
>256
>256
>256
128
256
>256
128
8
256
256
>256
>256
256
256
64
256
128
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
256
>256
>256
>256
256
>256
>256
>256
>256
>256
>256
256
>256
>256
>256
>256
256
>256
>256
>256
>256
>256
>256
>256
>256
>256
256
128
256
>256
>256
>256
128
>256
128
128
>256
128
64
32
>256
>256
256
64
256
>256
64
128
>256
16
>256
>256
>256
256
>256
256
256
128
256
32
32
8
16
4
2
4
8
2
4
8
256
256
32
>256
>256
16
>256
>256
4
>256
>256
4
>256
>256
64
>256
>256
>256
128
>256
256
256
256
32
64
16
32
32
32
32
>256
>256
>256
8
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
0.5
>256
>256
256
>256
>256
>256
>256
>256
>256
256
>256
>256
256
32
32
64
32
64
>256
>256
>256
32
>256
64
128
256
32
256
>256
256
256
>256
64
256
256
32
32
32
32
32
64
64
>256
>256
>256
8
256
>256
>256
>256
>256
>256
>256
>256
256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
2
256
128
>256
128
8
16
16
8
>256
>256
>256
>256
>256
>256
64
>256
256
>256
>256
>256
32
8
>256
>256
>256
16
256
32
256
0.5
2
128
2
2
16
2
4
1
0.25
0.5
0.0125
32
1
16
16
VAN
OXA
GEN
RIF
2
0.5
0.5
0.0125
32
0.0625
2
4
>256
>256
22

ACCEPTED MANUSCRIPT
CIP
4
0.5
0.5
0.5
2
0.5
0.0125
a
b
EFAE: Enterococcus faecium ATCC 19433, SAUR : S. aureus ATCC 25923, SAUR : S.
aureus ATCC 29213, SPNE: Streptococcus pneumonia ATCC 49619, ABAU: Acinetobacter
baumannii ATCC 19606, PAER: Pseudomonas aeruginosa ATCC 27853, ECLO:
Enterobacter cloacae ATCC 13047, VAN: vancomycin, OXA: oxacillin, GEN: gentamicin,
RIF: rifampicin, CIP: ciprofloxacin.
A series of compounds were then chosen to test against representative clinically significant
Gram-positive pathogens such as Enterococcus casseliflavus causing bacteremia,
Staphylococcus epidermidis able of growing biofilms on plastic devices, most commonly on
intravenous catheters, Staphylococcus saprophyticus causing urinary tract infections,
Streptococcus
pneumoniae
as
the
major
cause
of community
acquired
pneumonia and meningitis, Streptococcus pyogenes causing pharyngitis (strep throat),
localized skin infection (impetigo), necrotizing fasciitis, and neonatal infections,
Streptococcus agalactiae causing neonatal infections. As shown in Table 4, most of the
compounds displayed antibacterial activity against these Gram-positive pathogens with
clinical complexity for treatment, which demonstrated the potentials of our compounds.
Table 4. Antimicrobial activity evaluation of compounds 22, 23, 24, 27, 28, 30 and 59
against representative pathogenic Gram-positive bacteria.
Compound
ECAS
16
32
32
32
32
64
64
4
SEPI
SSAP
2
1
8
8
16
4
16
0.5
SPNE
2
8
4
8
8
8
32
0.5
SPYO
8
SAGA
8
2
2
2
2
2
3
5
2
3
4
7
8
0
9
1
16
4
8
8
2
4
16
16
16
16
16
32
0.5
16
16
32
32
16
32
4
CIP
0.0625
23

ACCEPTED MANUSCRIPT
GEN
OXA
VAN
4
32
4
0.125
0.25
1
0.0625
32
2
0.5
8
32
0.5
0.5
1
1
0.0625
0.5
ECAS: Enterococcus casseliflavus ATCC 25788, SEPI: Staphylococcus epidermidis ATCC
1
2228, SSAP: Staphylococcus saprophyticus ATCC 15305, SPNE: Streptococcus
pneumoniae ATCC 49619, SPYO: Streptococcus pyogenes ATCC 19615, SAGA:
Streptococcus agalactiae ATCC 12386.
MRSA represents one of the most difficult bacterial strains for treatment due to antibiotic
resistance. We also tested the above compounds against a series of globally spread MRSA
strains in hospitals and communities: S. aureus ST239 [13], S. aureus ATCC BAA-43 [14], S.
aureus ATCC BAA-44 [14], S. aureus HA W-231 ST45 [15], CA-MRSA W-47 ST30 [15],
CA-MRSA W-45 ST59 [15], CA-MRSA W-46 ST59 [15], S. aureus USA300/ST8 [16], S.
aureus 11B086169 ST22 [17], CA-MRSA W-4 ST338 [17], CA-MRSA W-48 ST217 [18], S.
aureus HA W-235 ST5 [19]. As shown in Table 5, all of compounds displayed consistent
antibacterial activity against these MRSA strains, in which some of them developed
resistance to oxacillin, as the first line antibiotic drug used in the USA for MRSA, as well as
other antibiotics such as ciprofloxacin and gentamicin. Our compounds were able to maintain
stable MIC values against diverse MRSA strains, suggesting the compounds may be
developed further to treat MRSA caused infectious diseases.
Table 5. Antimicrobial activity evaluation of compounds 22, 23, 24, 27, 28, 30 and 59
against representative globally spread HA- and CA-MRSA strains.
Compound
2
2
2
2
3
4
1
2
2
4
4
4
2
8
4
2
2
4
2
8
4
2
8
4
2
4
4
2
8
4
2
8
4
2
8
4
2
8
4
2
4
4
2
4
8
2
4
4
24

ACCEPTED MANUSCRIPT
2
2
3
5
7
8
0
9
2
4
2
2
1
1
1
2
2
4
2
2
1
1
0.25
1
8
16
4
4
4
>256
>256
1
2
4
4
2
2
8
8
4
4
8
4
2
2
16
4
0.5
4
8
4
8
8
4
4
8
2
4
8
2
4
4
0.5
32
1
4
4
4
4
4
0.25
64
0.5
4
8
2
4
8
2
2
4
0.5
64
0.5
4
8
2
2
4
16
4
>256
>256
1
2
2
4
2
CIP
GEN
OXA
VAN
0.5
0.25
4
0.25
0.5
8
0.25
0.5
8
0.5
0.25
0.5
0.5
>256
>256
1
>256
>256
0.5
0.5
0.5
1
3
.4 Time kill assessment
The time kill kinetics were used to demonstrate the in vitro reduction of a microbial
population by certain antimicrobial(s), and compounds 1 and 27 were tested against S. aureus
ATCC 25923 (Figure 6). The hit compound 1 started to inhibit bacteria growth at ¼ MIC,
and showed bacteriostatic effect at MIC and 4 MIC. 27 demonstrated similar inhibitory effect
at the ¼ and 1 MIC, and started to demonstrate slight bactericidal activities at 4 MIC.
Figure 6. Time-kill kinetics of compounds 1 and 27 at different concentrations against S.
aureus ATCC 25923
3
.5 Cytotoxicity
25

ACCEPTED MANUSCRIPT
The compounds within the best range of antimicrobial activity against S. aureus ATCC
2
5923 (MIC 1-2 µg/mL) were subjected to cytotoxicity testing against A549 human lung
carcinoma cell line and HaCaT human keratinocytes, as the major S. aureus infection sites
often occur in lung and skin [4]. As shown in Table 6, compounds 11 and 12 with imine
linker did not show significant cytotoxicity (≥200 µM), whereas compounds 22, 23, 24, 27,
3
0 with amine linker demonstrated mild cytotoxicity (13-40 µM). The result is
understandable as aniline is a known cytotoxic structure. The additional substitution on the
right benzene ring of compounds 59 and 60 can respectively reduce the cytotoxicity to about
8
0 and 110 µM even in the presence of aniline structure. The preliminary therapeutic index
were also calculated with 69-119 for imines and 6-12 for amines, which demonstrated very
promising safety range. When additional substitutions were installed on the right benzene
ring, the therapeutic index of amine compounds could be significantly improved to 25-33.
Note that the cell-based cytotoxicity of compounds is not always consistent with in vivo
toxicity, as the latter is associated with more complexed factors such as whole body
pharmacokinetics, organ or tissue distribution and the corresponding specific toxicity.
Therefore, the in vivo safety needs to be determined using animal models in the future.
Table 6. Cytotoxicity of selective compounds against A549 and HaCaT
b
a
CC (µM)
Therapeutic Index
MIC
50
Cpd.
(
µg/mL)
A549
HaCaT
A549
HaCaT
11
12
22
23
24
27
30
59
2
2
1
2
2
2
2
2
431.69 ± 30.14
274.16 ± 4.69
13.42 ± 1.53
18.32 ± 1.75
37.23 ± 4.44
22.52 ± 2.35
19.98 ± 2.11
79.14 ± 15.97
346.59 ± 28.93
249.52 ± 22.73
15.07 ± 1.81
20.24 ± 2.42
38.89 ± 4.18
20.09 ± 2.05
18.31 ± 1.97
72.15 ± 13.31
119
76
8
6
11
6
96
69
9
6
12
6
6
27
6
25
26

ACCEPTED MANUSCRIPT
6
0
2
-
112.18 ± 21.56
7.70 ± 0.58
108.10 ± 18.46
7.83 ± 0.66
33
32
-
c
DDP
-
a
b
c
S. aureus ATCC 25923, Calculated by CC (µg/mL) / ½ MIC (µg/mL), DDP: cisplatin
5
0
3
.6 Hemolytic property
With the antibacterial activity on hand, we intended to explore the potentials of using these
compounds in vivo. To initiate the study, we need to confirm that these compounds would not
cause hemolysis when injected into blood veins or transported by plasma proteins.
Compounds 27 and 28 were tested for their hemolytic property using human blood cells [20].
As shown in Figure 7, compounds 27 and 28 demonstrated very low haemolytic potential at
all tested concentrations (0.1, 1 and 10 µg/mL). The hemolytic values of all tested
compounds at all tested concentrations were lower than the suggested nonhemolytic cut-off
value.
27

ACCEPTED MANUSCRIPT
Figure 7. in vitro hemolysis activity of compounds 27 and 28
3
.7 Caco-2 cell permeability
If drugs are potentially orally available, the intestinal absorption is critical. The human
intestinal epithelial cell line Caco-2 was used to determine the cell permeability of
compounds 27 and 28 [20]. As shown in Table 7, compounds 27 and 28 possessed a high
-
5
apparent permeability, which is larger than 10 cm/s. Furthermore, the deduced apparent
permeability of compounds had a similar magnitude to that of the predicted values.
Table 7. The deduced and predicted apparent permeability of compounds 27 and 28
Compound
Papp Average (n=4) (cm/s)
Papp Predicted (cm/s)
-
5
-6
2
7
8
1.61 ± 0.53 x 10
1.49 ± 0.79 x 10
2.57 x 10
-5
-6
2
7.93 x 10
4
. Conclusions
In this study, we described the rational design and synthesis of a new class of diarylimine and
amine compounds based on the pharmacophore model derived from structure-based virtual
-
screening against the bacterial NusB-NusE protein. The entire library of hit compound
derivatives were assessed for their inhibition of NusB-NusE interaction, as well as the
antimicrobial activities, among which a few compounds demonstrated promising growth
inhibitory effect against a panel of bacterial pathogens, including clinical MRSA strains. The
cytotoxicity study demonstrated that these compounds are not very toxic to human cell lines,
while cell-based pharmacokinetic studies suggested the high druggability of the compounds.
28