Identification of novel scaffolds for potential anti- Helicobacter pylori agents based on the crystal structure of H. pylori 3-deoxy- d -manno-octulosonate 8-phosphate synthase (Hp KDO8PS)

Article abstract of DOI:10.1016/j.ejmech.2015.11.036

The crystal structure of 3-deoxy-d-manno-octulosonate-8-phosphate synthase (KDO8PS) from Helicobacter pylori (HpKDO8PS) was determined alone and within various complexes, revealing an extra helix (HE) that is absent in the structures of KDO8PS from other

Full text of DOI:10.1016/j.ejmech.2015.11.036

European Journal of Medicinal Chemistry 108 (2016) 188e202
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Research paper
Identification of novel scaffolds for potential anti-Helicobacter pylori
agents based on the crystal structure of H. pylori 3-deoxy-
D-manno-
octulosonate 8-phosphate synthase (HpKDO8PS)
a
a
a
b
c
d
Sujin Cho , Hookang Im , Ki-Young Lee , Jie Chen , Hae Ju Kang , Hye-Jin Yoon ,
c
b
b
a, *
Kyung Hoon Min , Kang Ro Lee , Hyun-Ju Park , Bong-Jin Lee
Research Institute of Pharmaceutical Sciences, College of Pharmacy, Seoul National University, Seoul 151-742, Republic of Korea
School of Pharmacy, Sungkyunkwan University, Gyeonggi-do 440-746, Republic of Korea
College of Pharmacy, Chung-Ang University, Seoul 156-756, Republic of Korea
Department of Chemistry, College of Natural Sciences, Seoul National University, Seoul 151-742, Republic of Korea
a
b
c
d
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 8 September 2015
Received in revised form
The crystal structure of 3-deoxy-D-manno-octulosonate-8-phosphate synthase (KDO8PS) from Heli-
cobacter pylori (HpKDO8PS) was determined alone and within various complexes, revealing an extra
helix (HE) that is absent in the structures of KDO8PS from other organisms. In contrast to the metal
coordination of the KDO8PS enzyme from Aquifex aeolicus, HpKDO8PS is specifically coordinated with
17 November 2015
Accepted 21 November 2015
Available online xxx
2
þ
2þ
Cd or Zn ions, and isothermal titration calorimetry (ITC) and differential scanning fluorimetry (DSF)
2
þ
2þ
revealed that Cd thermally stabilizes the protein structure more efficiently than Zn . In the substrate-
bound structure, water molecules play a key role in fixing residues in the proper configuration to achieve
a compact structure. Using the structures of HpKDO8PS and API [arabinose 5-phosphate (A5P) and
phosphoenolpyruvate (PEP) bisubstrate inhibitor], we generated 21 compounds showing potential
HpKDO8PS-binding properties via in silico virtual screening. The capacity of three, avicularin, hyperin,
and MC181, to bind to HpKDO8PS was confirmed through saturation transfer difference (STD) experi-
ments, and we identified their specific ligand binding modes by combining competition experiments and
docking simulation analysis. Hyperin was confirmed to bind to the A5P binding site, primarily via hy-
drophilic interaction, whereas MC181 bound to both the PEP and A5P binding sites through hydrophilic
and hydrophobic interactions. These results were consistent with the epitope mapping by STD. Our
results are expected to provide clues for the development of HpKDO8PS inhibitors.
Keywords:
Structure-based virtual screening
Docking simulation
Enzyme inhibitors
X-ray crystallography
NMR spectroscopy
©
2015 Published by Elsevier Masson SAS.
1
. Introduction
bacterium that colonizes the stomach. Marshall and Warren first
identified this microbe from patients suffering from chronic
gastritis and peptic ulcerations [1]. Currently, more than half of the
general population is infected with H. pylori, which is linked to
gastritis, duodenal ulcer, gastric cancer, gastric mucosa-associated
lymphoid tissue lymphoma (MALT), and sudden infant death syn-
drome (SIDS) [2e7].
Triple therapy involving a proton pump inhibitor (omeprazole)
and antibiotics (amoxicillin and clarithromycin) was initially rec-
ommended for treating H. pylori infection [8]; however, this con-
ventional multi-therapy is no longer effective due to the prevalence
of antibiotic resistance [9,10]. Moreover, these antibiotics
commonly disrupt the normal gastrointestinal flora, causing diar-
rhea as a side effect [11]. Therefore, it is necessary to develop new
anti-H. pylori agents based on the structure of novel cellular targets.
Helicobacter pylori (H. pylori) is a gram-negative, microaerophilic
Abbreviations: KDO8PS, 3-deoxy-
PEP, phosphoenolpyruvate; A5P, -arabinose-5-phosphate; HpKDO8PS, KDO8PS
from Helicobacter pylori; EcKDO8PS, KDO8PS from Escherichia coli; AaKDO8PS,
KDO8PS from Aquifex aeolicus; DAH7PS, 3-deoxy -arabino-heptulosonate-7-
phosphate synthase; apoHpKDO8PSwt, the metal- and substrate-free form of
HpKDO8PS; HpKDO8PS_H204A, mutant of HpKDO8PS (His204 to Ala204);
HpKDO8PS_C18A, a mutant of HpKDO8PS (Cys18 to Ala18); HpKDO8PS-Cd, the
cadmium-bound form of HpKDO8PS; HpKDO8PS-PEP-Zn, the PEP and zinc-bound
form of HpKDO8PS; API, an A5P and PEP bisubstrate inhibitor.
D-manno-octulosonate-8-phosphate synthase;
D
D
a
*
Corresponding author. Current address: Seoul National University, 21 dong, 417
ho, 1, Gwanak-ro, Gwanak-gu, Seoul 151-742, Republic of Korea.
E-mail address: lbj@nmr.snu.ac.kr (B.-J. Lee).
http://dx.doi.org/10.1016/j.ejmech.2015.11.036
0223-5234/© 2015 Published by Elsevier Masson SAS.

S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
189
þ
3
-Deoxy-
D
-manno-octulosonate
KDO8PS) is the enzyme that catalyzes the condensation reaction
8-phosphate
synthase
environment of the active site [29]. For example, Cd2 -bound
AaKDO8PS exhibits maximal enzymatic activity over the other
metal-bound proteins [27].
1
(
2
between arabinose 5-phosphate (A5P) and phosphoenolpyruvate
PEP) to synthesize KDO8P, the precursor of the 8-carbon sugar 3-
deoxy- -manno-octulosonate (KDO). KDO is the essential compo-
nent of gram-negative bacterial lipopolysaccharides (LPS) [12],
functioning as the linkage between lipid A and O-antigen [13]. As
the enzymes related to KDO synthesis and its incorporation within
the LPS structure play an important role in the survival and growth
of gram-negative bacteria, inhibition of KDO synthesis results in a
loss of the LPS endotoxin component and reduced pathogenicity
3
(
Based on the catalytic mechanism and AaKDO8PS structures,
Gatti and colleagues proposed novel inhibitors that mimic the in-
termediate form of the condensation reaction [26,30]. In this study,
we determined the crystal structures of HpKDO8PS alone
D
9
(apoHpKDO8PSwt), in complex with PEP and zinc (HpKDO8PS-
10
11
PEP-Zn) , and in complex with cadmium (HpKDO8PS-Cd) in an
effort to develop anti-H. pylori agents. The structure of an
12
HpKDO8PS mutant (HpKDO8PS_H204A) was also determined.
Chemical compounds binding to the HpKDO8PS active site were
evaluated by in silico virtual screening using the determined
HpKDO8PS structure. Among the 21 compounds initially selected,
three were validated for binding to HpKDO8PS via STD NMR
spectroscopy and waterLOGSY experiments. We also performed
docking simulations for HpKDO8PS in complex with the three
compounds. These compounds could serve as novel scaffolds for
the development of antibiotics that inhibit the function of
HpKDO8PS.
[
14e16]. Because the enzymes that participate in bacterial LPS
synthesis reactions are absent in mammalian systems, they are
considered to be potential targets for the development of antibi-
otics with fewer potentially undesirable cross-reactions [17].
Members of the KDO8PS family have been grouped into two classes
according to their transition metal requirements [18]. Escherichia
coli (E. coli) KDO8PS is metal independent (class I), whereas the
Aquifex aeolicus (A. aeolicus) and H. pylori enzymes (HpKDO8PS)4
require a transition metal (class II) [18,19]. KDO8PS crystal struc-
tures, both alone and in complex with various ligands, have been
previously reported (e.g., KDO8PSs from E. coli, A. aeolicus, Neisseria
meningitidis, Burkholderia pseudomallei and Pseudomonas aerugi-
2. Materials and methods
5
nosa) [20e24]. The KDO8PSs from E. coli (EcKDO8PS) and
A. aeolicus (AaKDO8PS) are homotetramers in which each mono-
2.1. Cloning, expression, and protein purification
6
8
mer forms a (b/a) barrel fold structure [20,21]. In addition, simi-
larities in the enzymatic reaction and active site structure between
The gene encoding HP0003 from H. pylori 26695 was amplified
by PCR using H. pylori genomic DNA (strain ATCC 700392/26695) as
a template. The PCR primers used to clone the HpKDO8PS expres-
sion plasmid, in which the restriction enzyme sites are underlined,
KDO8PS and 3-deoxy D-arabino-heptulosonate-7-phosphate syn-
7
thase (DAH7PS) suggest that they share a common ancestor [21].
The active site cavity in the PEP- and A5P-bound forms is closed,
whereas it is open in the substrate-free form [20]. It is possible that
its binding site allows PEP to form its distorted C-2 geometry by
adopting sp3 instead of sp hybridization, thereby facilitating the
intermediate form prior to the condensation with A5P [20]. Based
on the crystal structures of binary complexes of EcKDO8PS with PEP
0
were as follows: 5 -GGGAATTCCATATGAAAACTTCTAAAACAAAAA
0
0
CCCC-3
and 5 -CCGCTCGAGTTAAAATAAATTTTGGATTTTTAACA
0
TGTCGG-3 . The PCR product and pCold I vector (Takara, Japan)
were digested with NdeI and XhoI (NEB, UK) and ligated together.
After confirming the sequence, the recombinant plasmid was
overexpressed in chaperone-expressing E. coli cells pTf16/BL21
2
and through the use of a mechanism-based inhibitor (K
d
¼ 0.4
mM),
(
Takara, Japan) grown in LB broth. When the culture media reached
the condensation reaction between the substrates was shown to be
mediated by an intermediate form containing a transient oxo-
carbenium ion formed at the C-2 position in PEP [25]. Additionally,
the structures of AaKDO8PS in complex with R5P and PEP, along
with the bisubstrate inhibitor API [arabinose 5-phosphate (A5P)
an O.D. of 0.8, 1 mM isopropyl- -thiogalactopyranoside (IPTG)
b-D
was added to induce protein expression; cells were transferred to
ꢀ
1
5 C and grown for an additional 20 h. The cells were collected by
centrifugation at 4293ꢁ g for 10 min, and the pellet was resus-
pended in lysis buffer (20 mM Tris, 500 mM NaCl, 5 mM imidazole,
8
and phosphoenolpyruvate (PEP) bisubstrate inhibitor], suggest
ꢀ
0
.5 mM TCEP, pH 7.8) and then sonicated at 4 C using a 30% duty
that the water molecule coordinated to the metal ion on the si side
of PEP is necessary to trigger the reaction [26]. These structures also
support the hypothesis that a proton of the PEP-phosphate group is
transferred to the A5P-aldehyde group to first form the hydroxyl
group and that condensation is completed by a syn addition of
water and A5P to the si side of PEP [26].
cycle setting for 20 min (Cole Parmer Inc., USA). The lysate was
ꢀ
centrifuged at 6708ꢁ g for 60 min at 4 C, and the supernatant was
loaded into an open Ni-NTA column (Qiagen, USA) pre-equilibrated
with lysis buffer. The column was washed with a 30-fold excess
volume of buffer containing 150 mM imidazole; the protein was
eluted when the buffer reached 400 mM imidazole using an
imidazole gradient (from 150 to 500 mM). After concentrating the
protein, the buffer was changed to 20 mM MES, 200 mM NaCl, and
AaKDO8PS Cys11, His185, Glu222, and Asp233 are the residues
that interact with the aforementioned transitional metal ions.
Among them, His185 in particular directs PEP to its active site and
0
.5 mM TCEP at pH 6 by dialysis using an Amicon Ultra-15 Cen-
trifugal Filter Unit with a 10 K molecular weight cutoff (MWCO)
Millipore, USA). SDS-PAGE revealed purity up to 95%. The enzyme
2
þ
2þ
places a water molecule on the si side of PEP [27]. The Fe or Zn
ions bound to the AaKDO8PS active site are replaced by other
(
2þ
2þ
divalent metal ions, such as Cd and Cu [27e29], and these
was concentrated to 0.12 mM for crystallization.
metal ion substitutions affect enzymatic activity by altering the
The HpKDO8PS_metal-free form used for circular dichroism
(CD) and ITC experiments was prepared by dialyzing purified
HpKDO8PSwt against a buffer containing 20 mM MES, 200 mM
NaCl, 2 mM 1, 10-phenanthroline, and 2 mM EDTA at pH 6. Sub-
sequently, dialysis with a buffer containing 20 mM MES, 200 mM
1
3
-Deoxy-D-manno-octulosonate-8-phosphate synthase.
2
3
4
5
6
7
8
Arabinose 5-phosphate.
Phosphoenolpyruvate.
KDO8PS from Helicobacter pylori.
KDO8PS from Escherichia coli.
KDO8PS from Aquifex aeolicus.
9
Metal- and substrate-free form of HpKDO8PS.
PEP and zinc-bound form of HpKDO8PS.
Cadmium-bound form of HpKDO8PS.
10
11
3-Deoxy
D
-arabino-heptulosonate-7-phosphate synthase.
12
A5P and PEP bisubstrate inhibitor.
Mutant of HpKDO8PS (His204 to Ala204).

190
S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
NaCl, and 0.5 mM TCEP at pH 6 was conducted to reach a 1000-fold
dilution of the metal ion chelators. The resulting HpKDO8PS_metal-
free form was concentrated to 0.03 mM for CD spectroscopy and
(Hyogo, Japan). The dataset was processed and scaled with the
HKL-2000 program package [31]. The apoHpKDO8PSwt crystal
was identified from the C2 space group with the following unit
0
.1 mM for ITC.
cell parameters: a ¼ 137.83 Å, b ¼ 50.66 Å, c ¼ 78.67 Å, and
ꢀ
Mutagenesis primers (Table 1) targeting the four metal-binding
b
¼ 110.5 . There were two monomers in each asymmetric unit.
residues were designed, and EZchange™ Site-Directed Mutagen-
esis Kit (Enzynomics, Korea) was used to generate point mutations
according to the manufacturer's instructions. Briefly, the PCR con-
ditions consisted of 25 repeated cycles including a 30-s melting
step at 94 C, a 1-min annealing step at 55 C and a 5-min elon-
gation step at 72 C. Template removal and ligation of the PCR
M
The calculated crystal volume per protein weight (V ) was
3
ꢂ1
2.12 Å Da , and the solvent content was 42.0% [32]. The struc-
ture of apoHpKDO8PSwt was determined by molecular replace-
ment using the PHENIX Phaser-MR program [33] based on the
AaKDO8PS structure as a search model (PDB: 1FX6) [20]. The
initial structures were refined by alternately using the Refmac [34]
and Phenix.refine [33] programs. The solvent molecules were
inserted with Coot [35].
ꢀ
ꢀ
ꢀ
products were performed using EZ-MIX buffer (Enzynomics, Ko-
rea). After transformation of the mutant plasmids into DH5a-
competent cells, the sequence of the plasmid DNA sequence was
confirmed. BL21/pTF16 was used for overexpression of the mutant
proteins, and the successfully expressed mutants (HpKDO8PS_C18A
and HpKDO8PS_H204A) were purified.
The X-ray diffraction data from a single HpKDO8PS_H204A
crystal were collected at PAL (Pohang, Korea) using an ADSC
quantum 315r CCD detector on beamline 5C-SBII at 100 K and a
resolution of 2.4 Å. The raw data were processed using the HKL-
2000 program [31]. The HpKDO8PS_H204A crystal belongs to the
space group C2, with two monomers in each asymmetric unit (unit
cell parameters: a ¼ 139.86 Å, b ¼ 50.87 Å, c ¼ 78.30 Å, and
2
.2. Crystallization and structure determination
ꢀ
3
ꢂ1
b
¼ 104.73 ). Its V
M
was 2.22 Å Da , and the solvent content was
The most efficient crystallization conditions for each form of
4
4.6% [32]. The structure of the HpKDO8PS_H204A crystal was
HpKDO8PS were selected using the Crystal Screen™ and Index™
crystal screening kits (Hampton Research, USA) with the hanging
drop vapor diffusion method. The protein solution and reservoir
solution containing 18% polyethylene glycol (PEG) 3350, 0.1 M
HEPES, pH 7.5, and 0.25 M magnesium chloride hexahydrate were
mixed together in a 1:1 drop ratio. Crystals were obtained through
determined by molecular replacement using the PHENIX Phaser-
MR program [33] based on the determined model of apoHpK-
DO8PSwt. The refinement step for HpKDO8PS_H204A was con-
ducted in a manner identical to that for apoHpKDO8PSwt.
Raw X-ray diffraction data for the HpKDO8PS-Cd crystals were
collected at a resolution of 1.93 Å at 100 K using the same detector
as for the HpKDO8PS_H204A crystal at PAL (Pohang, Korea). The
data were processed and scaled using the HKL-2000 program [31].
The HpKDO8PS-Cd crystal was identified from the C2 space group,
ꢀ
drop equilibration with the reservoir solution at 20 C for 14 days.
Prior to data collection, the crystals were flash-cooled with liquid
nitrogen with protection by the addition of 25% glycerol to the cryo-
solution. An identical procedure was applied for HpKDO8PS_H204A
crystal preparation.
To obtain better diffraction-quality crystals of HpKDO8PS-Cd
and HpKDO8PS-PEP-Zn, we performed streak seeding using pul-
verized apoHpKDO8PSwt crystals as microseeds for nucleation. The
protein solution contained the protein-binding partners (Cd
Zn and PEP) at a 2-fold excess concentration over the protein
molecules (i.e., protein at 0.12 mM and protein binding partners at
0
with unit cell parameters of a ¼ 140.35 Å, b ¼ 51.02 Å, c ¼ 78.67 Å,
ꢀ
and
b
¼ 104.7 . Two monomers of HpKDO8PS-Cd were found in
3
ꢂ1
each asymmetric unit, with a V
M
¼ 2.24 Å Da and a solvent
content of 45.2% [32]. The structure was determined by molecular
replacement using apoHpKDO8PSwt as a model with the PHENIX
Phaser-MR program [33]. Refinement of the HpKDO8PS-Cd model
was performed by alternately using the Refmac [34] and Phe-
nix.refine [33] programs, as mentioned above.
2
þ
,
2
þ
.24 mM). For the crystallization drop, 1 mL of prepared protein
The HpKDO8PS-PEP-Zn diffraction dataset was collected at a
resolution of 1.68 Å at 100 K using a Saturn A200 mosaic CCD de-
tector at beamline 26B1 in SPring-8 (Hyogo, Japan). The dataset was
processed and scaled with the HKL-2000 program [31]. The
HpKDO8PS-PEP-Zn crystal was identified from the C2 space group,
solution was added to the same volume of reservoir solution. The
seed stock was prepared after washing an apoHpKDO8PSwt crystal;
the stock was continuously transferred to the original crystalliza-
tion solution for stabilization and crushed using a glass homoge-
nizer. This concentrated seed stock was serially diluted with the
stabilizing buffer by a factor of 1000. Each drop was streaked with
the diluted solution using a clean whisker-like fiber. Crystals
with unit cell parameters of a ¼ 140.3 Å, b ¼ 51.01 Å, c ¼ 78.55 Å,
ꢀ
and
b
¼ 104.37 . There were two monomers in each asymmetric
3
ꢂ1
ꢀ
unit. The V was 2.24 Å Da , and the solvent content was 45.2%
appeared in 7e14 days at 20 C after the start of the incubation and
M
[32]. The structure of HpKDO8PS-PEP-Zn was determined by mo-
were additionally soaked in a cryo-solution containing 1 mM of
2
þ
2þ
lecular replacement using the PHENIX Phaser-MR program [33]
with the apoHpKDO8PSwt structure as a search model. The initial
structures were refined by alternately using the Refmac [34] and
Phenix.refine [33] programs.
Zn , PEP or Cd for 12 h prior to the flash-freezing step.
The diffraction dataset for the apoHpKDO8PSwt crystal was
collected at 100 K at a resolution of 2.0 Å using an MAR225HE CCD
detector at beamline BL44XU in the SPring-8 radiation facility
Table 1
Primers for site-directed mutagenesis targeting the residues involved in metal binding within the active site of HpKDO8PS. In primer IDs, “F”
represents forward, and “R” represents reverse. Mutated nucleotides are underlined in the primer sequences used for site-directed mutagenesis.
Primer ID
Primer sequence
Mutated nucleotide
C18p-F
C18p-R
GTTTTAATCGCTGGGCCAGCTGTCATTGAGAGCTTAG
CTAAGCTCTCAATGACAGCTGGCCCAGCGATTAAAAC
GATTTTTGACGCTACCGCTAGCGTGCAAATGCCAG
CTGGCATTTGCACGCTAGCGGTAGCGTCAAAAATC
ATTGATGGGTTGTTTGCTGCTACGCATGTTGATCCTAAA
TTTAGGATCAACATGCGTAGCAGCAAACAACCCATCAAT
CTAAAAACGCCCTAAGCGCTGGAGCAAACATGCTAAAAC
GTTTTAGCATGTTTGCTCCAGCGCTTAGGGCGTTTTTAG
C18A
C18A
H204p-F
H204p-R
E241p-F
E241p-R
D252p-F
D252p-R
H204A
H204A
E241A
E241A
D252A
D252A

S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
191
ꢂ1
ꢂ1
2.3. Circular dichroism spectroscopy
0.001 kcal Å mol was reached.
For CD experiments, the HpKDO8PS forms (HpKDO8PS_metal-
2.7. Preparation of protein structures
bound, HpKDO8PS_metal-free, HpKDO8PS_C18A, and HpKDO8P-
S_H204A) were used at a concentration of 0.03 mM in a buffer
containing 20 mM MES, 200 mM NaCl, and 0.5 mM TCEP, pH 6.
HpKDO8PS_metal-bound represents the sample used in the crys-
tallization of apoHpKDO8PSwt. CD spectra were collected using a
Chirascan™-plus CD spectrometer (Applied Photophysics, UK) with
The crystal structure of AaKDO8PS in complex with API (PDB:
1JCX) [26] was selected to gain insight into the ligand binding site.
Superimposition of apoHpKDO8PSwt and AaKDO8PS in complex
with API revealed that the C-terminal active sites of the two en-
zymes are highly conserved (motifs C18/11, K47/41, N54/48, R55/49,
S56/50, A108/102, K130/124, H207/185, and E241/222). Therefore,
the API molecule from the AaKDO8PS-API structure was extracted
and merged into our structure. The resulting API:HpKDO8PS com-
plex was energy-minimized. Both proteins were prepared with the
Protein Preparation module of the Sybyl-X suite v. 1.3 [36] using the
default parameters.
ꢂ1
the following settings: scan rate of 120 nm min , response time of
0
0
.5 s, step size of 1 nm, and bandwidth of 1 nm. The path length was
.2 mm using a quartz cell (106-0.20-40) with detachable windows
(
(
Hellma, Germany). All scans were conducted in the far-UV range
190e260 nm). The data were processed by baseline subtraction
and smoothing using the Pro-Data Viewer program (Applied Pho-
tophysics, UK).
2.8. Docking-based virtual screening
2.4. Isothermal titration calorimetry
Receptor-based virtual screening was performed to obtain new
A 0.1 mM solution of HpKDO8PS_metal-free was prepared as
compounds harboring the desired activity profiles. The chemical
database containing natural products and synthetic compounds
was docked into the validated HpKDO8PS binding site using
Surflex-Dock [36]. First, a protomol was generated to define the
active site, with a threshold of 0.4 and bloat set of 1 around the
embedded ligand. Core interactions (Lys47, Asn54, Arg55, Ser56,
Ala108, and Lys130) were also assigned in the protomol. The
number of docking runs was set to 50, and other parameters were
set as the Surflex-Dock Geom default settings. The final hit com-
pounds were evaluated for binding by combining the consensus
scoring function Cscore3 (Cscore >3) and Surflex-Dock total score
described above. Zinc chloride and cadmium chloride were dis-
solved to a concentration of 1.5 mM in the same buffer as the
protein. The protein and metal ion solutions were degassed in a
vacuum before measurement. All ITC experiments were performed
using a MicroCal iTC200 microcalorimeter (GE healthcare, UK) at
ꢀ
2
5
C. The protein was added to the sample cell, and a metal so-
lution (zinc or cadmium chloride) was charged in the injection
syringe. During the titration, 20 aliquots of metal solution were
injected into the sample chamber. The titrations began with a 60-s
delay time and 0.4-mL injection volume, followed by a 2-mL
injection-volume with 5-s delay times. Intervals of 150 s between
injections were included at the end of each titration. The stirring
speed in the sample chamber at 25 ± 0.1 C was 1000 rpm. Heat
(ꢂlog K
d
). Visual inspection considering the important interactions
was necessary to evaluate binding.
ꢀ
generation during the titration experiments was measured using
the integrated data from the ITC calorimeter with the Origin 7.0
software package supplied by MicroCal, subtracting the heat
2.9. Synthesis and preparation of ligands
Hyperin was prepared as reported by Lee and colleagues [37],
and avicularin was purchased (Jinan Boss Chemical Industry,
China). MC181 was synthesized using the following 2-step method.
(see Scheme 1).
a
generated by the buffer. K (binding constant), DH (enthalpy
change) and N (stoichiometry) values were calculated by applying
the one-site fitting model.
2.5. Differential scanning fluorimetry
2.9.1. Step 1) ethyl 1-(3-phenylpropyl)piperidine-4-carboxylate
(
compound 1)
mixture of ethyl-4-piperidinecarboxylate (500 mg,
3.18 mmol), 3-phenylpropylbromide (576.2 L, 3.82 mmol), and
HpKDO8PS_metal-free was diluted to 5
mM in a white 96-well
A
plate using the identical buffer as mentioned above. A 5000X so-
lution of the dye SYPRO Orange (Sigma Aldrich, USA) was added to
each well to achieve a final concentration of 5X. Measurements
m
potassium carbonate (in excess) prepared in DMF (1 mL) was stir-
red at 70 C for 2 h. The reaction mixture was cooled to room
ꢀ
were performed in 100
mL in the presence and absence of 75
mM
temperature and then diluted with ethyl acetate, washed with
metal ion (zinc or cadmium chloride). The temperature was ram-
ped from 25 to 95 C at a rate of 1 C/min using an Applied Bio-
systems 7500 Fast Real-Time PCR Instrument System
water and brine, dried on anhydrous MgSO
vacuo. The residue was purified by flash chromatography on silica
gel (MeOH:CH Cl
4
, and concentrated in
ꢀ
ꢀ
2
2
¼ 1:20) to generate ethyl 1-(3-phenylpropyl)
(
ThermoFisher Scientific, USA). The raw fluorescence data were
plotted as a function of temperature, which generated a sigmoidal
curve [Fig. 5(A)]. The inflection point of the transition curve (T
midpoint melting temperature) was calculated using Prism 5
GraphPad software, USA) by applying the Boltzmann sigmoidal
fitting [Fig. 5(B)].
piperidine-4-carboxylate (863.9 mg, 98.6%), referred to as Com-
pound 1.
,
1H NMR (600 MHz, CD
3
OD): d 7.26e7.23 (m, 2H), 7.19e7.17 (m,
m
2H), 7.16e7.13 (m, 1H), 4.13e4.09 (m, 2H), 2.89 (d, J ¼ 11.46 Hz, 2H),
2.61 (t, J ¼ 15.24 Hz, 2H), 2.38e2.36 (m, 2H), 2.35e2.30 (m, 1H),
2.07 (t, J ¼ 22.02 Hz, 2H), 1.91e1.87 (m, 2H), 1.84e1.81 (m, 2H),
(
1
3
1
d
.75e1.68 (m, 2H), 1.23(t, J ¼ 14.28, 3H); C NMR (150 MHz, CDCl
3
):
2.6. Preparation of compounds for docking-based virtual screening
175.07, 142.10, 128.37, 128.27, 125.72, 60.25, 58.14, 52.94, 41.16,
3.70, 28.57, 28.20, 14.21, 14.20; LC/MS (ESI ) m/z: 276.3 [MþH] .
þ
þ
3
A total of 415 compounds from our in-house database were
processed for docking with the Sybyl-X suite v. 1.3 (Certara, USA)
36]. The compounds were prepared by adding hydrogen atoms
2.9.2. Step 2) N-(3-(furan-2-yl)phenyl)-1-(3-phenylpropyl)
piperidine-4-carboxamide (MC181)
[
and rectifying wrong valences and then energy-minimized using
the conjugated gradient method in the Tripos force field with the
Gasteiger-Hückel charge method until a convergence value of
A mixture of Compound 1 (79.6 mg, 0.289 mmol) and 3NeHCl
(2 mL) was subjected to microwave synthesis (Monowave 300) at
150 C for 20 min. The reaction mixture was concentrated in vacuo
ꢀ

192
S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
Fig. 1. ApoHpKDO8PSwt structure (PDB: 4Z1A). (A) The crystal structure of apoHpKDO8PSwt exhibited a (
with distinctive structures in different colors ( -helix, green, -strand, cyan, and loop, orange). (B) ApoHpKDO8PSwt is superimposed onto EcKDO8PS (PDB: 1D9E) and AaKDO8PS
PDB: 1FX6) for comparison. Both structures are drawn in a cartoon diagram (EcKDO8PS, pink, AaKDO8PS, brown, and HpKDO8PS, green). A hairpin in EcKDO8PS is highlighted in hot
pink. An extra helix (HE) in HpKDO8PS is highlighted in lime green. (C) Topological models of HpKDO8PS, EcKDO8PS, and AaKDO8PS. Cylinders represent -helices (EcKDO8PS, pink,
AaKDO8PS, orange, and HpKDO8PS, green), and cyan arrows represent -strands. The differences between models are highlighted in red boxes. (D) Interactions of HE (raspberry). HE
8
b/a) topology. The secondary structures are drawn in a cartoon diagram
a
b
(
a
b
interacts with neighboring residues in H5 and S6 as well as loops through both hydrophobic interaction (left) and hydrogen bonding (right). The interactions are shown in the
surface diagram by black dashed lines. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
to produce the corresponding acid, and the resulting residue was
used for the next reaction without further purification. A mixture of
the acid (20 mg, 0.070 mmol), 3-(2-furyl)aniline hydrochloride
stirred at room temperature for 1.75 h. The mixture was added to 4-
dimethylaminopyridine (3.5 mg, 0.03 mmol) and stirred at room
temperature for 75 min. The reaction mixture was diluted with
(
12.9 mg, 0.081 mmol), and triethylamine (113
CH Cl (10 mL) was added to 1-ethyl-3-(3-dimethylaminopropyl)
carbodiimide (77.7 mg, 0.405 mmol) in a dropwise manner and
m
L, 0.81 mmol) in
CH
and concentrated in vacuo. The crude product was purified by flash
chromatography on silica gel (MeOH:CH Cl
¼ 1:20) to generate
2 2 4
Cl , washed with water and brine, dried on anhydrous MgSO ,
2
2
2
2

S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
193
Fig. 2. Interactions within the active sites in four HpKDO8PS X-ray crystal structures. The residues related to the interactions are shown as sticks; water molecules (numerically
ordered) are drawn as gray balls; hydrogen bonds are drawn as black dashed lines. (A) Interactions between apoHpKDO8PSwt (green) and water molecules. (B) Interactions between
2
þ
HpKDO8PS_H204A (cyan) and water molecules (see also Fig. S1). (C) Interactions between HpKDO8PS-Cd (orange) and water molecules. The cadmium ion (Cd ) is shown as a green
2þ
ball. (D) Interactions between HpKDO8PS-PEP-Zn (light blue) and water molecules. The zinc ion (Zn ) is shown as a pink ball, and PEP is shown as a stick (carbon, green; oxygen, red;
phosphorus, olive). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
MC181 (10 mg, 37%).
irradiations were applied at chemical shifts of 0 and ꢂ30 ppm,
respectively. HpKDO8PS was saturated using a train of Gaussian-
shaped 50-ms-long pulses. The total length of the saturation train
was set to 2 s.
1
H NMR (600 MHz, CD
3
OD):
d
7.94 (t, J ¼ 4.2 Hz, 1H), 7.54 (dd,
J ¼ 1.8, 0.66 Hz, 1H), 7.43e7.41 (m, 1H), 7.31 (t, J ¼ 15.8 Hz, 1H),
7.28e7.25 (m, 2H), 7.21e7.20 (m, 2H), 7.18e7.15 (m, 1H), 6.73 (dd,
J ¼ 3.36, 0.66 Hz, 1H), 6.50 (dd, J ¼ 3.39, 1.92 Hz, 1H), 3.22 (d,
Prior to acquisition, a 15-ms spin-lock pulse with a 1-W strength
(T2 filter) was applied to remove protein signals from the STD
spectrum. The H NMR spectrum was acquired with 32 K real points
J ¼ 12.1 Hz, 2H), 2.66 (t, J ¼ 11.94 Hz, 2H), 2.67e2.62 (m, 2H),
1
3
1
2
.53e2.48 (m, 1H), 2.40 (t, J ¼ 20.7 Hz, 2H), 1.96e1.89 (m, 6H);
C
NMR (150 MHz, CD
38.90, 131.45, 128.06, 127.99, 125.7, 119.08, 118.64, 114.94, 114.93,
11.32, 111.29, 104.99, 104.97, 57.37, 52.26, 42.16, 32.90, 27.35, 27.12;
3
OD):
d
174.08, 153.51, 142.11, 142.09, 141.15,
and 6320 scans.
1
1
Single pseudo-2D data from two serial free-induction decays
were divided into two separate 1D data (on- and off-resonance). To
increase the signal-to-noise ratio, the 1D raw data were processed
with a 0.5-Hz line broadening and exponential window function
prior to Fourier transformation. For comparison, the STD effects of
individual peaks were quantified by the simple equation (Ioff ꢂ Ion)/
þ
þ
LC/MS (ESI ) m/z: 389.3 [MþH] .
2.10. NMR spectroscopy
NMR experiments [38,39] were performed at 298 K using a
Ioff, where Ion and Ioff represent the absolute intensities of the on-
Bruker Avance DRX 600 MHz spectrometer equipped with a 5-mm
and off-resonance spectra peaks, respectively.
1
13 15
The off-resonance spectrum is identical to a conventional 1H
NMR spectrum. The largest STD value was set to 100%, and the
other STD values were normalized to the largest value [40,41].
TXI ( H/ C/ N) probe. The NMR sample was prepared as a mixture
of 5 M HpKDO8PS and 0.2 mM ligand, which were dissolved in a
solution containing 98% D O and 2% DMSO. On- and off-resonance
m
2

194
S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
2
þ
Fig. 4. ITC data for Cd titration onto metal-free HpKDO8PS (HpKDO8PS_metal-free).
2
þ
Fig. 3. Metal and PEP binding sites in HpKDO8PS. Omit difference maps (Fo ꢂ Fc)
Cd injection induces an exothermal reaction and stabilizes the enzyme. The data are
2þ
contoured at 3.0
s
(gray) are shown around the metal, water and PEP. (A) Cd (green
well fitted in a single-site binding isotherm model.
ball) has a distorted octahedral coordination in HpKDO8PS, in contrast with the ge-
ometry in gas phase or AaKDO8PS, which exhibit a tetrahedral or square pyramidal
coordination, respectively. (B) Zn2þ (pink ball) shows a distorted square pyramidal
The peaks of the 1D 1H NMR spectra were assigned using a
geometry. The metal geometry of the Cd2 ion and Zn
þ
2þ
ion in HpKDO8PS is
server (www.Acdlabs.com/resources/ilab) for the assignment of
peaks and analysis of STD and WaterLOGSY data. All STD and
WaterLOGSY spectra were analyzed in a spectral range of approx-
imately 6 ppme8 ppm, where aromatic protons appear.
2þ
remarkably opposite to that in AaKDO8PS. (C) Zn (pink ball) and PEP binding sites in
HpKDO8PS. PEP (carbon, green; oxygen, red; phosphorus, olive) interacts directly with
Ser49, Lys52, and Lys130. Gln133 also participates in PEP binding via water molecules
(
gray balls) (see also Table S1). (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)
3. Results
WaterLOGSY NMR experiments [42] were conducted to confirm
the results of the STD experiments. All of the WaterLOGSY spectra
were recorded at 298 K using a Jeol ECA 600 MHz spectrometer
equipped with a 5-mm triple resonance inverse probe. The ligand
3.1. Protein expression and structure determination
Several HpKDO8PS constructs were prepared using the pCold I
vector (Takara, Japan). Five types of different chaperone-expressing
BL21 competent cells (pG-KJE8, pGro7, pKJE7, pG-Tf2, and pTf16,
Takara, Japan) were tested for the most efficient production of
soluble protein extracts. Among them, the pTf16 chaperone-
expressing BL21 competent cells provided the best results, and
soluble protein was prepared.
We determined four crystal structures, namely, apoHpK-
DO8PSwt, HpKDO8PS_H204A, HpKDO8PS-Cd, and HpKDO8PS-PEP-
Zn. The data and refinement statistics are presented in the
Supplementary material Table S1. The electron density map of
apoHpKDO8PSwt showed some poorly resolved regions. Several
residues between residues 212e218 in chain A and 211e217 in
chain B exhibited poor electron density. In the N-terminal region,
loop-forming residues 1e9 were also invisible on the electron
density map, possibly due to their level of structural disorder.
(
0.2 mM) was dissolved in the presence and absence of 5
HpKDO8PS in a solution containing 10% D O and 90% H O in a total
volume of 300 L. To selectively excite water, Gaussian-shaped 20-
mM
2
2
m
ms pulses were irradiated at approximately 4.7 ppm. The mixing
time for the magnetization transfer was set at 2 s. The NMR spectra
were acquired with 256 transients and 8 K data points. WATERGATE
pulse sequences were interleaved in the WaterLOGSY sequences to
suppress water signals. To increase the signal-to-noise ratio, free-
induction decays were processed with a 0.5-Hz line broadening
and exponential window function prior to Fourier transformation.
For STD-NMR competition experiments, PEP or A5P was added
to the NMR samples, which contained either HpKDO8PS and
hyperin or HpKDO8PS and MC181. The same acquisition and pro-
cessing parameters mentioned above were used to obtain the STD
NMR spectra.

S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
195
substrate-free forms of wild-type HpKDO8PS), as well as EcKDO8PS
and AaKDO8PS [20], are similar to each other, with r.m.s. deviations
of 1.38 Å and 1.08 Å, respectively, with 463 and 478 equivalent C
a
s
and harboring sequence identity levels of 46.9% and 51.5%,
13
respectively. An extra
between the H5 -helix and S6
10-helix between S8 and H8 (residues 246e249) was also found, as
a
-helix (HE, residues 160e166) is found
a
b
-strand in HpKDO8PS (Fig. 1); a
3
in AaKDO8PS. However, due to poor electron density, the 310-helix
was not observed in the structure of EcKDO8PS (not shown). No
hairpins were observed at the N-terminus of HpKDO8PS and
AcKDO8PS [Fig. 1(C)], in contrast with EcKDO8PS.
The crystal structure of the H204A mutant (HpKDO8PS_H204A)
was similar to that of apoHpKDO8PSwt. However, several residues
located primarily at the active site (Lys52, Asn54, Arg55, Gln133,
Asp252, and Asn255) were oriented in different directions in
HpKDO8PS_H204A compared with the corresponding residues in
apoHpKDO8PSwt [Fig. 2(A) and (B)]. These residues are charged
amino acids or have similar polar groups that can form hydrogen
bonds. Asn255 in HpKDO8PS_H204A is flipped into the active site,
forming a hydrogen bond with a water molecule (water-7)
[Fig. 2(B)]. In addition, water-7 and -8 are linked together by a
hydrogen bond. In apoHpKDO8PSwt, Glu241, Asp252, and Asn255
are moved toward the space previously filled with His204
[Fig. 2(B)]. The distance between the residues in HpKDO8PS_H204A
is shorter compared with apoHpKDO8PSwt, with the Glu241-
Asp252 distance changing from 9.0 Å to 8.2 Å, the Glu241-Asn255
distance changing from 7.2 Å to 5.3 Å, and the Asp252-Asn255
distance from 7.6 Å to 5.0 Å. In the HpKDO8PS_H204A structure, a
hydrogen bond between Arg55 and Ser251 is absent; therefore,
Arg55 moves toward the outside of the active site [Fig. 2(A) and
(B)]. Lys52, which hydrogen bonds with water-5 in apoHpK-
DO8PSwt, interacts with water-12 and Asn54 through multiple
hydrogen bonds, moving closer to Asn54 by a distance of 5.0 Å
compared with the Lys52-N-Asn54-O distance of 9.3 Å observed in
apoHpKDO8PSwt [Fig. 2(A) and (B)].
Fig. 5. Thermal shift assay of HpKDO8PS_metal-free by differential scanning fluo-
rimetry measured in the presence and absence of metal (Cd2 or Zn ). (A) Protein
þ
2þ
To analyze the metal geometry in the active site, crystals of the
Cd2 -bound form (HpKDO8PS-Cd) were obtained. A Cd ion is
þ
2þ
thermal denaturation is depicted by a fluorescence versus temperature plot in the
presence of SYPRO Orange. (B) Mid-point melting temperatures (T
the melting curve inflections shown in Fig. 5(A).
m
) calculated from
bound to the metal binding residues (Cys18, His204, Glu241, and
ꢀ
ꢀ
Scheme 1. (a) 3-Phenylpropylbromide, potassium carbonate, DMF, 70 C, 2 h; (b) i. 3NeHCl, microwave 150 C, 20 min; ii. 3-(2-furyl)aniline hydrochloride, Et
3
N, EDC, DMAP, room
temperature, 3.5 h.
In the refined model of HpKDO8PS_H204A, HpKDO8PS-Cd, and
HpKDO8PS-PEP-Zn, two monomers were observed in each asym-
metric unit, and they were well superimposed with the apoHpK-
DO8PSwt model, with root mean square (r.m.s.) deviation values of
Asp252) of each protomer, forming a distorted octahedral geometry
with a water molecule (i.e., water-19) [Figs. 2(C) and 3(A)]. In the
active site, several residues and water molecules are linked to each
other by hydrogen bonds making space for substrate binding [20].
Compared with the apoHpKDO8PSwt structure, the HpKDO8PS-Cd
Asp252 residue moves toward Asn255 when it is bound to a Cd
ion [Fig. 2(A) and (C)]; the distance between the two residues then
changes from 7.6 Å to 4.8 Å by linking a water molecule (i.e., water-
0
.50 Å, 0.36 Å, and 0.36 Å, respectively, for the 513, 514, and 514 C
atom pairs. In these structures, two loop regions involving residues
to 9 and 210 to 220, which correspond to the similar sites in
apoHpKDO8PSwt, were also invisible.
a
2
þ
1
1
7).
3.2. HpKDO8PS crystal structures
Instead of forming a hydrogen bond with Ser251, the Arg55-NH
residue is bound to water-24, and the Arg55-Nε residue is linked to
Thr56 via water-23, favoring a rigid conformation [Fig. 2(A) and
(C)]. The link between Ser49-water-5-Lys52 [Fig. 2(A)] is complex
HpKDO8PS is a 30.3-kDa enzyme with two monomers in each
asymmetric unit. The monomers adopt the (
8
b/a) barrel topology
[
Fig. 1(A)], similar to their previously reported homologs. The sec-
ondary structures were assigned using the STRIDE server [43]. Both
apoHpKDO8PSwt monomeric structures (i.e., the metal- and
13
An extra helix in the HpKDO8PS structure.

196
S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
2
þ
in HpKDO8PS-Cd due to water-18 and -20, which mediate the
interaction with Asn54 via two additional water molecules (i.e.,
water-21 and water-22) [Fig. 2(C)]. The overall structure of the
active site could become compact through polar interactions.
Consequently, the distances between Ser49, Lys52, and Asn54
become shorter. More precisely, between apoHpKDO8PSwt and
HpKDO8PS-Cd, the Ser49-Lys52 distance changes from 4.0 Å to
(Fig. 5). This result supported the hypothesis that Cd contributes
2
þ
.
more to the thermostability of HpKDO8PS than Zn
.4. Validation of the virtual docking method
Before virtually screening our in-house chemical database with
3
the Surflex-Dock program [36], the docking protocol was evaluated
for its ability to reproduce the binding modes of known KDO8PS
inhibitors [API and 2,8-bis(phosphonooxy)-octanoic acid]
3
.3 Å, the Ser49-Asn54 from 12.0 Å to 9.7 Å, and the Lys52-Asn54
from 9.3 Å to 7.0 Å.
We attempted to crystallize HpKDO8PS in complex with sub-
strates to determine which residues participate in the binding of
specific substrates, and HpKDO8PS-PEP-Zn crystals were obtained
using various substrates and metal ion combinations. In
(
Supplementary material Fig. S2). API [26], which mimics the in-
termediate form of the KDO8PS substrate condensation reaction,
was used as a reference molecule by re-docking it into the active
site of apoHpKDO8PSwt. The API phosphate and phosphonate
groups formed a hydrogen bond network with Ala108, Lys130 and
Asn54, which was highly consistent with the reported co-crystal
structure (not shown) [26], thereby validating our docking protocol.
2
þ
HpKDO8PS-PEP-Zn, a Zn ion is bound to a site identical to the one
bound by the Cd2 ion in HpKDO8PS-Cd. This binding forms a
distorted square pyramidal geometry involving neighboring resi-
dues and a water molecule (i.e., water-34) [Figs. 2(D) and 3(B)]. The
overall folding conformation in HpKDO8PS-PEP-Zn crystals appears
to be similar to that of the model apoHpKDO8PSwt, though some
residues generate differences in the active site responsible for
substrate binding. Indeed, several residues interact directly with
PEP. Notably, Ser49 and Lys52 are linked to PEP O2ʹ, whereas Lys130
captures PEP O2P, O2, and O1 [Fig. 2(C)]. More water molecules are
observed in the HpKDO8PS-PEP-Zn active site, as they form
hydrogen bonds with PEP and amino acids. Among them, water-27
and water-31 form a direct link with PEP O1P, whereas water-28
and water-30 bind to PEP O3P [Fig. 2(C)]. In particular, as in EcK-
DO8PS and AaKDO8PS, water-31 is consistently found on the PEP si
side, which is expected to trigger the condensation reaction [25,26].
As the structure of HpKDO8PS-PEP-Zn was obtained at higher res-
olution compared to apoHpKDO8PSwt, it shows a more detailed
water network together with Ser49, Lys52, Asn54, Arg55, Gln133,
þ
3.5. HpKDO8PS-targeted virtual screening
A step-wise strategy for virtual screening was employed in our
study to identify novel HpKDO8PS inhibitors, and various criteria
were applied in combination to select hit compounds, as outlined
in Fig. 6. First, 65 top-ranked compounds (out of 415 docked) were
selected based on their Surflex-Dock energy score and Cscore
(Cscore >3) [36]. Then, their interactions with the receptor active
site were then analyzed to select compounds with the required
hydrogen bond interactions. Compounds that made contact with
residues that have the desired hydrogen bonds (residues Asn54/
Arg55 and Lys130/Ala108) were considered to be real hits. The third
filter was a visual inspection of the binding pose considering the
diversity of ligand scaffolds, such as the number of hetero atoms,
hydrogen bond donor/acceptor, different types of aromatic, and
non-aromatic ring and alkyl groups. Finally, 21 compounds, all of
which have drug-like profiles obeying Lipinski's rule-of-five [45],
were selected for further bioactivity evaluation. For comparison,
the 21 compounds were docked into HpKDO8PS-Cd, EcKDO8PS,
and AaKDO8PS; the docking scores and ranks of the 21 compounds
are listed in Supplementary material Table S2. In addition, virtual
screening using those structures was conducted (Supplementary
Ser251, Asp252, and Zn2 [Fig. 2(D)].
þ
In addition, structural comparison between HpKDO8PS-Cd and
HpKDO8PS-PEP-Zn shows that the water molecules are located in
similar positions in both crystals. Water-18 in HpKDO8PS-Cd is
substituted to PEP O2ʹ in linking Ser49 and Lys52 in HpKDO8PS-
PEP-Zn. Additionally, the connections found in HpKDO8PS-Cd are
consistently observed in HpKDO8PS-PEP-Zn, starting at the trigonal
link between Ser49, Lys52, and water-20 (or water-42 in
HpKDO8PS-PEP-Zn) and extending to Asn54 via water-21 and wa-
ter-22 (or water-40 and water-43) [Fig. 2(C) and (D)].
3.3. Thermal scanning for metal and metal-free HpKDO8PS
interaction
The thermostabilizing effects of Cd2þ and Zn on HpKDO8PS
2þ
2
þ
were measured by ITC and DSF experiments. The ITC data with Cd
injection were well fitted to the single-site binding in the isotherm
2
þ
model (Fig. 4). However, Zn ion titration with the enzyme caused
an exothermic reaction up to the 6th injection and an endothermic
reaction in the following injections (from the 7th to the 20th in-
jections), generating data that could not be fitted to the single-site
binding isotherm model. Indeed, HpKDO8PS contains a single site
2þ
for a Cd ion, and the binding affinity (K
d
) is 460.5 ± 75.5 nM, with
a heat change (
D
H) of ꢂ5741 ± 61.83 cal/mol.
The DSF experiments were conducted as described in the Ma-
terials and Methods section. The fluorescent dye (SYPRO Orange)
binds to the hydrophobic residues of the protein and fluoresces.
During the thermal unfolding process, hydrophobic regions of the
protein are exposed, and fluorescence intensity increases with
enhanced dye binding [44]. Thus, thermostability can be evaluated
ꢀ
from the shift in T
m
. In the absence of metal, T
m
was up to 48 C,
suggesting destabilization. In contrast, T
m
was observed to shift to
ꢀ
2þ
ꢀ
2þ
6
3 C in the presence of Cd and was 59 C in the presence of Zn
Fig. 6. Scheme of structure-based virtual screening (see also Fig. S2 and Table S2).

S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
197
material Table S3). The ranking of Surflex-Dock scores is slightly
worse for the HpKDO8PS-Cd docking results and falls significantly
when the compounds were docked into KDO8PS from other species
(E. coli and A. aeolicus).
3.6. NMR analysis of interactions between hit compounds and
HpKDO8PS
To validate the results from our in silico virtual screening, we
applied saturation transfer difference (STD) NMR spectroscopy and
water-ligand observed via gradient spectroscopy (waterLOGSY) to
evaluate whether the individual hit compounds bind to HpKDO8PS
[38,39,42,46].
STD NMR spectroscopy is a useful approach to detect a reduction
in a ligand's NMR signal, which is caused by the nuclear Overhauser
effect (NOE) between the protein and its ligand [39]. The on-
1
resonance spectrum records a 1D H NMR signal for each com-
pound, which is reduced by the magnetization transfer from
HpKDO8PS, whereas the off-resonance spectrum serves as a refer-
ence [Figs. 7 and 8(B), and Supplementary material Fig. S3(B)] that
1
is identical to a standard 1D H NMR spectrum for ligands [47]. The
STD spectrum represents the difference between the on- and off-
resonance signals [Figs. 7 and 8(D), and Supplementary material
Fig. S3(D)], and the signals on the STD spectrum reflect the inter-
action between the protein and its ligand. Figs. 7 and 8(C) and
Supplementary material Fig. S3(C) show representative STD spectra
for each compound in the absence of HpKDO8PS.
Among the 21 potential ligands derived from our in silico virtual
screening, three ligands, avicularin (quercetin-3-O-
anoside) (Supplementary material Fig. S3), hyperin (quercetin-3-O-
-galactopyranoside) (Fig. 7), and MC181 (N-(3-(furan-2-yl)
a-L-arabinofur-
b-D
phenyl)-1-(3-phenylpropyl)piperidine-4-carboxamide) (Fig. 8),
produced STD spectra that indicated interactions with HpKDO8PS.
Because hyperin and avicularin [Fig. 7(A) and Supplementary
material Fig. S3(A)] have identical aromatic moieties involved in
HpKDO8PS binding and exhibited only a minor difference in their
auxiliary sugar moieties (b-D-galactopyranose for hyperin and a-L-
arabinofuranose for avicularin), the following analysis was per-
formed only for hyperin.
All of the five protons of hyperin exhibited STD signals, among
which proton-2 showed the largest reduction in signal ratio be-
tween the on- and off-resonance spectra [Fig. 7(D)]. For compari-
son, the levels of STD signals of the other protons were normalized
to that of proton-2, as shown in Fig. 7(A). Proton-1 exhibited the
second-largest STD signal, which represented 75 percent of that of
proton-2. The protons from the sugar moieties could not be
observed because of buffer signals, though the buffer demonstrated
no STD signals as an internal reference.
For MC181, proton-6 exhibited the largest STD effect, and the
STD effects of protons-1, -3, -5, -8, and -10 were quantified and
normalized to that of proton-6, demonstrating relatively low STD
effects [Fig. 8(A)]. To corroborate the STD data, waterLOGSY ex-
periments were performed for hyperin and MC181. This technique
uses magnetization transfer from water to the free ligand and the
ligand-bound protein via intermolecular NOE [48,49].
Fig. 7. 1H STD and WaterLOGSY NMR spectra in the aromatic region and docking
conformation of hyperin. (A) Chemical structure of hyperin and epitope mapping (each
proton is numbered in red, and the values of the normalized STD effect are presented
1
as percentages). (B) Reference H NMR spectrum of hyperin. (C) STD spectrum of
hyperin in the absence of HpKDO8PS. (D) STD spectrum of hyperin in the presence of
HpKDO8PS. (E) STD spectrum of hyperin in the presence of HpKDO8PS and PEP. (F) STD
spectrum of hyperin in the presence of HpKDO8PS and A5P. (G) WaterLOGSY spectrum
of hyperin in the absence of HpKDO8PS. (H) WaterLOGSY spectrum of hyperin in the
presence of HpKDO8PS. (I) Docking conformation of hyperin. The distances between
residues (carbon atoms, orange), shown as sticks, and the compound are less than
3.5 Å. Hyperin (carbon atoms, green) forms hydrogen bonds (black dashed lines) with
Asn54, Arg55, Asp87, His89, Lys130, Arg173, and Ser251. (J) Distances between the
protons of hyperin and residues. The protons exhibiting an STD effect are numbered as
in Fig. 7(A). Although proton-2 exhibited the largest STD effect, it is not consistent with
the distances between the proton and residues (dashed arrows and indicated in blue)
(see also Figs. S3 and S4). (For interpretation of the references to color in this figure
legend, the reader is referred to the web version of this article.)
We compared the waterLOGSY spectra of the ligand in the
absence [Figs. 7 and 8(G)] and presence [Figs. 7 and 8(H)] of
HpKDO8PS, showing the peaks in the opposite signs. The protons
corresponding to these peaks are believed to interact with
HpKDO8PS via water molecules within the active site, and they
were observed only at 6e7.5 ppm (hyperin) and 6.5e8 ppm
KDO8PS, PEP [Figs. 7 and 8(E)] and A5P [Fig. 7 and 8(F)], as com-
petitors. These substrates are well known to bind to different parts
of the same active site in KDO8PS [20,50].
(
MC181). These results were highly consistent with those from the
STD experiments.
To identify the HpKDO8PS ligand-binding site, STD competition
experiments were conducted using the natural substrates of
The hyperin competition STD data showed that the STD peak
intensities or hyperin in the presence of A5P were significantly

198
S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
decreased but that PEP was less affected in 6e7.5 ppm of the
hyperin STD spectrum compared with A5P [Fig. 7(E) and (F)]. These
results indicate that hyperin competes with A5P for HpKDO8PS
binding, suggesting that hyperin binds to the A5P binding site (A5P-
subsite) of HpKDO8PS.
For MC181, the overall intensities of the STD spectra were
significantly decreased in the presence of PEP or A5P [Fig. 8(E) and
(F)], indicating that MC181 binds to a broad range of PEP and A5P
binding sites within the HpKDO8PS active site. This result is in
agreement with the fact that MC181 is a long molecule.
3.7. Docking analyses of HpKDO8PS ligands
The binding modes of representative hit compounds were pre-
dicted by Surflex-Dock docking [36]. As shown in Fig. 7 and 8(I),
these results revealed that the hit compounds fit well in the area
covering the combined binding sites for PEP and A5P, the natural
KDO8PS substrates. The compounds were docked into the
HpKDO8PS-Cd structure, which is presumably more similar to the
2
þ
Zn -bound HpKDO8PS structure than its apo-form. The binding
modes of hyperin and MC181 were compared with that of the
known inhibitor API [25] (Supplementary material Fig. S4), which
was used as the reference compound for our docking analysis.
According to the hyperin-bound models, the hydroxyl groups of
the sugar moiety are docked into the PEP binding site (PEP-subsite)
[20,25] with hydrogen bonds to Lys130 and Arg173, which interact
with nearby His204 [Fig. 7(I)]. However, hyperin exhibits no
interaction with the metal because it is placed farther from the
2
þ
bound Cd ion compared with the API binding site. In addition, the
hydrogen bonds with Lys47, Lys52, and Ala108 in the API-bound
model disappear in the hyperin-docked conformation [Fig. 7(I)
and Supplementary material Fig. S4]. The hyperin benzopyran ring
occupies the site where A5P binds (A5P-subsite) in the AaKDO8PS
structure [20,25] [Fig. 7(I)], forming multiple hydrogen bonds with
Asn54, Arg55 and Ser251, which are not observed in the API-bound
model. The hyperin dihydroxyphenyl ring is placed near Asp87,
His89, and Pro107 via hydrogen bonds and hydrophobic in-
teractions [Fig. 7(I)].
In the MC181 binding mode analysis [Fig. 8(I)], the PEP-subsite
[20] is occupied by the MC181 furan ring, which interacts with
Gln105, Pro107, and Lys130 at a short distance. Although the furan
2
þ
ring of MC181 is farther away from the Cd ion than the PEP
moiety in the API-bound model, the phenyl group of MC181 is well
matched to API [Fig. 8(I) and Supplementary material Fig. S4].
Indeed, it binds to Leu182 via a hydrophobic interaction and also
interacts with His204, Gln207, and Asp252. MC181 covers the A5P-
subsite with its piperidine ring, and the peptide oxygen forms
hydrogen bonds with Ala53, Asn54, and Arg55 [Fig. 8(I)]. Asn54 also
forms a hydrogen bond with the peptide's nitrogen atom.
MC181 was also docked into EcKDO8PS and AaKDO8PS, and the
results showed that the key residues for MC181 binding are
different from those in HpKDO8PS. MC181 binds to Asn62, Arg63,
Ser64, Ala116, Lys138, His202, and Gln205 in EcKDO8PS, or Asn48,
Arg49, Ser50, Ala102, Lys124, His185, and Gln188 in AaKDO8PS.
Considering the protein sequence differences between HpKDO8PS,
EcKDO8PS, and AaKDO8PS, the binding key residues are 45%
(EcKDO8PS) or 52% (AaKDO8PS) identical compared with those of
HpKDO8PS.
Analysis of the three catalytic pockets (HpKDO8PS, EcKDO8PS
and AaKDO8PS) revealed that although the active sites are highly
conserved, the dimensions of the catalytic channel rim differ
greatly [Fig. 9(A)]. Fig. 9(A) shows four regions, AeD, that represent
the boundaries of the catalytic channel rim. Region A corresponds
to Asn54 in HpKDO8PS, Ser232 in AaKDO8PS and Asn26 in EcK-
DO8PS, region B to Arg55 in HpKDO8PS and Arg49, Ser50 in
Fig. 8. 1H STD and WaterLOGSY NMR spectra in the aromatic region and docking
conformation of MC181. (A) Chemical structure of MC181 and epitope mapping (each
proton is numbered in red, and the values of the normalized STD effect are presented
1
as percentages). (B) Reference H NMR spectrum of MC181. (C) STD spectrum of MC181
in the absence of HpKDO8PS. (D) STD spectrum of MC181 in the presence of
HpKDO8PS. (E) STD spectrum of MC181 in the presence of HpKDO8PS and PEP. (F) STD
spectrum of MC181 in the presence of HpKDO8PS and A5P. (G) WaterLOGSY spectrum
of MC181 in the absence of HpKDO8PS. (H) WaterLOGSY spectrum of MC181 in the
presence of HpKDO8PS. (I) Docking conformation of MC181. The distances between
residues (carbon atoms, orange), shown as sticks, and the compound are less than
3.5 Å. MC181 (carbon atoms, green) forms hydrogen bonds (black dashed lines) with
Ala53, Asn54, Arg55, and Asp87. (J) Distances between the protons of MC181 and
residues. The protons exhibiting an STD effect are numbered as in Fig. 8(A). The epitope
mapping result is consistent with the distances between the compound and enzyme,
except for proton-12 (dashed arrows and indicated in blue) (see also Figs. S3 and S4).
(
For interpretation of the references to color in this figure legend, the reader is referred
to the web version of this article.)

S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
199
Fig. 9. Topedown view of the active site of AaKDO8PS (top), HpKDO8PS (middle) and EcKDO8PS (bottom). (A) The distances between the boundaries of the catalytic channel rim are
shown with red arrows. Letters AeD denote four boundary areas surrounding the channel rim. (B) The binding poses of MC181 in the active sites of AaKDO8PS (top), HpKDO8PS
(
middle) and EcKDO8PS (bottom). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
AaKDO8PS, region C to Gln207 in HpKDO8PS, Phe103 in AaKDO8PS
and Phe117 in EcKDO8PS, and region D to Phe134 in HpKDO8PS and
Pro190 in AaKDO8PS. Variation at the A, B and C regions produces a
significantly wider channel rim in the crystal structure of
HpKDO8PS compared to the other two KDO8PSs. The structural
manifestation of this variation is that the measured BeD distance is
terminus of the loop. Moreover, Leu164, which is located in this
same loop, forms a hydrophobic interaction with Pro197 in the
H6eS7 loop [Fig. 1(D), left panel], which in turn affects the S7eH7
loop, where Gln207 and His204 form an A5P and a metal ligand-
binding site, respectively. Based on these observations, the addi-
tional HE may play a role in indirectly stabilizing the conformation
of the active site. However, further studies are needed to evaluate
the possible role of the HE.
HpKDO8PS is a metal-dependent enzyme that has four metal-
coordinating residues, namely, Cys18, His204, Glu241, and
Asp252, in the active site, similar to AaKDO8PS [20]. These residues
were mutated to determine whether they are related to protein
stability and active site coordination. Of the four mutants (C18A,
H204A, E241A, and D252A), the crystal structure of the H204A
(HpKDO8PS_H204A) was determined. As shown in our CD spec-
troscopy studies, HpKDO8PS_H204A demonstrated a better stabil-
ity in its secondary structure than the other mutants, which
presumably led to crystallization success (Supplementary material
Fig. S1). The two other mutants, E241A and D252A, were unavai-
lable due to low protein stability during purification.
6
.21 Å in AaKDO8PS compared to 11.7 Å in HpKDO8PS, whereas the
measured AeC distance is 10.06 Å in EcKDO8PS compared to 20.6 Å
in HpKDO8PS.
The binding poses of MC181 in the active site of AaKDO8PS or
EcKDO8PS revealed that the catalytic channel is too narrow to
accommodate the entire MC181 molecule, with the furan ring and
phenyl group of MC181 protruding out of the binding pocket
[
Fig. 9(B)]. Nonetheless, MC181 can bind deeply at the bottom of the
HpKDO8PS catalytic channel, which occupies both the PEP- and
A5P-subsites.
4
. Discussion
HpKDO8PS possesses an extra helix (HE) (Fig. 1) that is not
The structure of HpKDO8PS_H204A [Fig. 2(B)] showed that the
residues adjacent to His204 in the active site became closer to each
other. Moreover, the conformations of Asn54, Lys52, and Asn255
were significantly altered with respect to our other wild-type
structures. Because the residues adjacent to His204 are related to
substrate and metal binding, His204 is believed to not only serve as
a metal-binding residue but also as a contributor to active site
observed in the structures of KDO8PS from E. coli, A. aeolicus,
B. cenocepacia [23], P. aeruginosa [24], and N. meningitides [22]. This
HE interacts with its neighboring H5 a-helix, S6 b-strand, and loops
(
H6eS7 and HEꢂS6) through hydrogen bonds and hydrophobic
interactions [Fig. 1(D)]. Arg151 interacts with Gly167 at the HEꢂS6
loop [Fig. 1(D), right panel], an interaction that affects the PEP-
binding residues Lys130 and Gln133, which are located at the N-

200
S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
formation. These results are consistent with a previous report
showing that the His185 residue in the active site of AaKDO8PS is
responsible for forming the substrate binding space [27].
occupied by the benzopyran ring of hyperin, which is consistent
with the epitope mapping results from STD experiments [Fig. 7(A)],
even though the STD values could not be easily converted to the
distance between the protein and ligand [Fig. 7(J)]. This result from
the docking experiments was also confirmed by competition data,
which indicated that hyperin competes with A5P for the ligand
binding site [Fig. 7(F)]. Although the sugar moiety of hyperin binds
to the PEP-subsite, according to the docking simulation results, it
was difficult to confirm these data by NMR experiments because
the protons of the sugar moiety could not be detected on the NMR
spectra.
According to docking simulations, MC181 and API share similar
binding HpKDO8PS modes, despite their different chemical struc-
tures (Fig. 8 and Supplementary material Fig. S4). MC181, through
hydrogen bonds and hydrophobic interactions, binds to and
broadly covers the PEP- and the A5P-subsites in HpKDO8PS. These
results were highly consistent with the results obtained from STD
competition experiments against PEP or A5P [Fig. 8(E) and (F)],
which showed a decline in the STD spectral intensities. The epitope
mapping results [Fig. 8(A)] were also consistent with the observed
distances between the compounds and enzyme, as shown in
Fig. 8(J). MC181 was also docked into EcKDO8PS and AaKDO8PS to
investigate whether it specifically binds to HpKDO8PS (Fig. 9). The
active sites are highly conserved, yet the catalytic channel rims of
the enzymes differ significantly [Fig. 9(A)], with HpKDO8PS having
a wider channel rim, which is consistent with the MC181 docking
results showing that MC181 binds well to the HpKDO8PS PEP- and
A5P-subsites. Conversely, due to their narrow channel rims, MC181
juts out from the active sites of EcKDO8PS and AaKDO8PS
[Fig. 9(B)]. Based on the results, MC181 can be considered a
promising scaffold for developing new antibiotics that specifically
act against H. pylori.
2
þ
2þ
The structures of HpKDO8PS in complex with Cd or PEP/Zn
HpKDO8PS-Cd and HpKDO8PS-PEP-Zn) [Fig. 2(C) and (D)] were
(
determined, and different hydrogen bond networks with water
molecules were observed within the active site. In particular, the
HpKDO8PS-PEP-Zn active site appeared rigid and therefore stabi-
lized via its complicated hydrogen bond network [Fig. 2(D)]. By
facilitating the condensation process, the water molecules in the
structures appear to be involved in the formation of a stable com-
plex between the protein and ligands. The idea of a protein-ligand
interaction mediated by water was supported by waterLOGSY-NMR
experiments [51].
2
þ
Interestingly, in these structures, HpKDO8PS binds to Zn and
Cd2 ions in distorted square pyramidal and octahedral geometries
þ
[
Fig. 3(A) and (B)], respectively, in contrast to the AaKDO8PS
2
þ
enzyme structure [29]. Moreover, in contrast with Cd , the ITC
data (Fig. 4) for Zn2 , the naturally occurring metal ion binding to
HpKDO8PS, did not fit to the single binding site isothermal model,
possibly for the following reasons: (i) multiple binding modes
þ
2
þ
could exist during the equilibration process between Zn
and
HpKDO8PS; (ii) inevitable interactions could occur with the buffer;
or (iii) metal ion precipitation could generate heat [52] (data not
2
þ
shown). The DSF data supported the idea that Cd
enhances
2
þ
HpKDO8PS thermostability to a greater degree than Zn (Fig. 5).
According to Krosky and colleagues, the Kcat value increases by
approximately 2-fold, and the A5P K
imately 6.5-fold upon replacement of the Zn ion by Cd in Zn
bound HpKDO8PS [19]. Consistently, our metal coordination results
m
value decreases by approx-
2þ
2þ
2þ
-
2
þ
2þ
indicated that the metal geometries of Cd and Zn could affect
enzyme activity. Indeed, the properties of metal ions as Lewis acids
play a particularly important role in biology: metal ions can activate
coordinated ligands for reactivity by affecting either bond length,
The interactions observed for MC181 involved both hydrophobic
and hydrophilic binding modes, whereas hyperin primarily inter-
acted with HpKDO8PS via hydrophilic interactions. Furthermore,
the compound docking results provided clues for the modification
of these inhibitors. Hyperin and avicularin are derivatives of quer-
cetin, representing a typical subclass of flavonoids [54], which have
been reported to exert various biological effects, including anti-
microbial, antihypertensive, neuroprotective, and chemoprotective
effects [54e56]. Among those biological activities, the antimicro-
bial effect is supported by the interactions between HpKDO8PS and
these compounds (hyperin [57,58] and avicularin [59]).
2
þ
bond angles, or coordination site number [53]. In HpKDO8PS, Cd
2
þ
may favor a more ideal coordination than Zn for the enzymatic
reaction.
To date, only a few inhibitors of KDO8PS have been reported
(Supplementary material Fig. S2), all of which were designed to
mimic the intermediate form of the condensation reaction between
A5P and PEP, and the lack of known ligands has limited the use of
conventional ligand-based screening methods to identify novel
KDO8PS inhibitors. Instead, structural information for the crystal
structure of HpKDO8PS determined for the first time in our present
study was utilized to identify novel scaffolds that specifically bind
to HpKDO8PS using structure-based virtual screening.
5. Conclusions
This step-wise virtual screening approach successfully identi-
fied three novel chemotypes (avicularin, hyperin, and MC181) as
HpKDO8PS inhibitors. To the best of our knowledge, this is the first
in silico study on the identification of novel KDO8PS inhibitors. For
comparison, an API-based 3D similarity search against the same in-
house database was conducted using the SurflexSim program [36].
Notably, three active compounds that directly bind to HpKDO8PS
were highly ranked in the output of Surflex-Dock scoring (docking
ranks ranged from 16 to 21), and the ranking was significantly
lower when they were docked into EcKDO8PS and AaKDO8PS,
suggesting that the compounds may specifically bind to HpKDO8PS.
However, a simple 3D similarity search failed to select the most
active compounds (similarity ranks ranged from 77 to 146), as
shown in the Supplementary material Table S2. Taken together, the
results demonstrated that the virtual screening approach was
successful for the discovery of novel KDO8PS inhibitors.
Due to increased resistance to antibiotics and the gastrointes-
tinal side effects of conventional multiple therapy, the develop-
ment of new anti-H. pylori drugs is urgently needed. The LPS
synthesis pathway is considered to be a target for developing an-
tibiotics against H. pylori, and HpKDO8PS is an essential enzyme
that catalyzes the condensation of A5P and PEP to generate KDO8P,
the precursor of LPS biosynthesis. The current study provides in-
formation of HpKDO8PS crystal structures, and among 21 possible
ligands generated via in silico virtual screening, the capacity of 3
compounds to bind to HpKDO8PS was demonstrated through STD
and waterLOGSY experiments. It is expected that this study will
provide a basis for the design of new and selective HpKDO8PS
inhibitors.
Author contributions
The HpKDO8PS binding modes of hyperin and MC181 were
investigated using STD-NMR experiments and docking simulations.
According to the docking results, the A5P-subsite is spatially
B.L. conceived this project and designed the experiments, S.C.
contributed to the protein purification, site-directed mutagenesis,
CD spectroscopy, ITC experiments. S.C., H.I., and H.Y. performed the

S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
201
protein crystallization and structure determination. S.C. and K.L.
carried out the STD and waterLOGSY NMR experiments. H.P. and J.C.
conducted the docking simulation, virtual screening, and com-
pound preparation. K.H.M and H.J.K. synthesized MC181. S.C., H.I.,
K.L., and B.L. wrote the manuscript.
synthase is
97e306.
a zinc-metalloenzyme, Biochim. Biophys. Acta 1594 (2002)
2
[
20] H.S. Duewel, S. Radaev, J. Wang, R.W. Woodard, D.L. Gatti, Substrate and metal
complexes of 3-deoxy-D-manno-octulosonate-8-phosphate synthase from
Aquifex aeolicus at 1.9-Å resolution. Implications for the condensation mech-
anism, J. Biol. Chem. 276 (2001) 8393e8402.
21] S. Radaev, P. Dastidar, M. Patel, R.W. Woodard, D.L. Gatti, Structure and
mechanism of 3-deoxy-D-manno-octulosonate 8-phosphate synthase, J. Biol.
Chem. 275 (2000) 9476e9484.
22] F.C. Cochrane, T.V. Cookson, G.B. Jameson, E.J. Parker, Reversing evolution: re-
establishing obligate metal ion dependence in a metal-independent KDO8P
synthase, J. Mol. Biol. 390 (2009) 646e661.
[23] L. Baugh, L.A. Gallagher, R. Patrapuvich, M.C. Clifton, A.S. Gardberg,
T.E. Edwards, B. Armour, D.W. Begley, S.H. Dieterich, D.M. Dranow,
J. Abendroth, J.W. Fairman, D. Fox 3rd, B.L. Staker, I. Phan, A. Gillespie, R. Choi,
S. Nakazawa-Hewitt, M.T. Nguyen, A. Napuli, L. Barrett, G.W. Buchko, R. Stacy,
P.J. Myler, L.J. Stewart, C. Manoil, W.C. Van Voorhis, Combining functional and
structural genomics to sample the essential Burkholderia structome, PLoS One
[
[
Acknowledgments
We thank the beamline staffs at Pohang Accelerator Laboratory,
Korea (BL-5C, and BL-7A) and SPring-8, Japan [BL-26B1 (2013A1053)
and BL44XU (2012A6741)] for assistance during X-ray diffraction
experiments. This work was supported by the National Research
Foundation of Korea [NRF-2007-0056817, 2012R1A2A1A01003569,
2
008-355-E00019, 2014R1A1A3A04050250], the Korean Govern-
8
(2013) e53851.
ment, Korea Healthcare Technology R&D project, Ministry for Health,
Welfare and Family Affairs, Republic of Korea [A092006], and the
[
[
[
24] S.K. Nelson, A. Kelleher, G. Robinson, S. Reiling, O.A. Asojo, Structure of 2-keto-
3-deoxy-D-manno-octulosonate-8-phosphate synthase from Pseudomonas
aeruginosa, Acta Crystallogr. Sect. F Struct. Biol. Cryst. Commun. 69 (2013)
2
014 BK21 plus Project for Medicine, Dentistry and Pharmacy.
1084e1088.
25] O. Asojo, J. Friedman, N. Adir, V. Belakhov, Y. Shoham, T. Baasov, Crystal
structures of KDOP synthase in its binary complexes with the substrate
phosphoenolpyruvate and with a mechanism-based inhibitor, Biochemistry
Appendix A. Supplementary data
40 (2001) 6326e6334.
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.ejmech.2015.11.036.
26] J. Wang, H.S. Duewel, R.W. Woodard, D.L. Gatti, Structures of Aquifex aeolicus
KDO8P synthase in complex with R5P and PEP, and with a bisubstrate in-
hibitor: role of active site water in catalysis, Biochemistry 40 (2001)
1
5676e15683.
References
[27] J. Wang, H.S. Duewel, J.A. Stuckey, R.W. Woodard, D.L. Gatti, Function of
His185 in Aquifex aeolicus 3-deoxy-D-manno-octulosonate 8-phosphate syn-
thase, J. Mol. Biol. 324 (2002) 205e214.
[28] H.S. Duewel, R.W. Woodard, A metal bridge between two enzyme families. 3-
deoxy-D-manno-octulosonate-8-phosphate synthase from Aquifex aeolicus
requires a divalent metal for activity, J. Biol. Chem. 275 (2000) 22824e22831.
[29] F. Kona, P. Tao, P. Martin, X. Xu, D.L. Gatti, Electronic structure of the metal
[
1] B.J. Marshall, J.R. Warren, Unidentified curved bacilli in the stomach of pa-
tients with gastritis and peptic ulceration, Lancet 1 (1984) 1311e1315.
2] A.A. Ciociola, D.J. McSorley, K. Turner, D. Sykes, J.B. Palmer, Helicobacter pylori
infection rates in duodenal ulcer patients in the United States may be lower
than previously estimated, Am. J. Gastroenterol. 94 (1999) 1834e1840.
3] J.C. Atherton, The pathogenesis of Helicobacter pylori-induced gastro-duodenal
diseases, Annu. Rev. Pathol. 1 (2006) 63e96.
[
center in the Cd2 , Zn , and Cu substituted forms of KDO8P synthase:
implications for catalysis, Biochemistry 48 (2009) 3610e3630.
[30] X. Xu, J. Wang, C. Grison, S. Petek, P. Coutrot, M.R. Birck, R.W. Woodard,
D.L. Gatti, Structure-based design of novel inhibitors of 3-deoxy-D-manno-
octulosonate 8-phosphate synthase, Drug Des. Discov. 18 (2003) 91e99.
[31] Z. Otwinowski, W. Minor, Processing of X-ray diffraction data collected in
oscillation mode, Method Enzymol. 276 (1997) 307e326.
þ
2þ
2þ
[
[
[
4] Y. Yamaoka, Helicobacter pylori: Molecular Genetics and Cellular Biology,
Caister Academic, Wymondham, 2008.
5] X.C. Wu, P. Andrews, V.W. Chen, F.D. Groves, Incidence of extranodal non-
Hodgkin lymphomas among whites, blacks, and Asians/Pacific Islanders in
the United States: anatomic site and histology differences, Cancer Epidemiol.
3
3 (2009) 337e346.
6] K.M. Mohandas, Helicobacter pylori and gastric lymphoma, N. Engl. J. Med. 331
1994) 746.
7] J.R. Kerr, A. Al-Khattaf, A.J. Barson, J.P. Burnie, An association between sudden
infant death syndrome (SIDS) and Helicobacter pylori infection, Arch. Dis.
Child. 83 (2000) 429e434.
[32] B.W. Matthews, Solvent content of protein crystals, J. Mol. Biol. 33 (1968)
491e497.
[
[
(
[33] P.D. Adams, P.V. Afonine, G. Bunkoczi, V.B. Chen, I.W. Davis, N. Echols,
J.J. Headd, L.W. Hung, G.J. Kapral, R.W. Grosse-Kunstleve, A.J. McCoy,
N.W. Moriarty, R. Oeffner, R.J. Read, D.C. Richardson, J.S. Richardson,
T.C. Terwilliger, P.H. Zwart, PHENIX: a comprehensive Python-based system
for macromolecular structure solution, Acta Crystallogr. D Biol. Crystallogr. 66
(2010) 213e221.
[
8] P. Malfertheiner, F. Megraud, C.A. O'Morain, J. Atherton, A.T. Axon, F. Bazzoli,
G.F. Gensini, J.P. Gisbert, D.Y. Graham, T. Rokkas, E.M. El-Omar, E.J. Kuipers,
Management of Helicobacter pylori infectionethe Maastricht IV/Florence
consensus report, Gut 61 (2012) 646e664.
[34] G.N. Murshudov, A.A. Vagin, E.J. Dodson, Refinement of macromolecular
structures by the maximum-likelihood method, Acta Crystallogr.
Crystallogr. 53 (1997) 240e255.
[35] P. Emsley, B. Lohkamp, W.G. Scott, K. Cowtan, Features and development of
coot, Acta Crystallogr. D Biol. Crystallogr. 66 (2010) 486e501.
[36] C. L.P., Sybyl eX, 2.1 ed., SYBYL Molecular Modeling Software, 2013. St. Louis,
MO
D Biol.
[9] W. Wu, Y. Yang, G. Sun, Recent insights into antibiotic resistance in Heli-
cobacter pylori eradication, Gastroenterol. Res. Pract. 2012 (2012) 723183.
[10] E.A. Cameron, K.U. Powell, L. Baldwin, P. Jones, G.D. Bell, S.G. Williams, Heli-
cobacter pylori: antibiotic resistance and eradication rates in Suffolk, UK,
1991e2001, J. Med. Microbiol. 53 (2004) 535e538.
[
11] D. Carcanague, Y.K. Shue, M.A. Wuonola, M. Uria-Nickelsen, C. Joubran,
J.K. Abedi, J. Jones, T.C. Kuhler, Novel structures derived from 2-[[(2-pyridyl)
methyl]thio]-1H-benzimidazole as anti-Helicobacter pylori agents, part 2,
J. Med. Chem. 45 (2002) 4300e4309.
[37] S.Y. Lee, K.H. Kim, I.K. Lee, K.H. Lee, S.U. Choi, K.R. Lee, A new flavonol
glycoside from Hylomecon vernalis, Arch. Pharm. Res. 35 (2012) 415e421.
[38] M. Mayer, B. Meyer, Characterization of ligand binding by saturation transfer
difference NMR spectroscopy, Angew. Chem. Int. Ed. 38 (1999) 1784e1788.
[39] B. Meyer, T. Peters, NMR spectroscopy techniques for screening and identi-
fying ligand binding to protein receptors, Angew. Chem. Int. Ed. 42 (2003)
864e890.
[
12] D.H. Levin, E. Racker, Condensation of arabinose 5-phosphate and phos-
phorylenol pyruvate by 2-keto-3-deoxy-8-phosphooctonic acid synthetase,
J. Biol. Chem. 234 (1959) 2532e2539.
[
13] A.P. Moran, M.M. Prendergast, B.J. Appelmelk, Molecular mimicry of host
structures by bacterial lipopolysaccharides and its contribution to disease,
FEMS Immunol. Med. Microbiol. 16 (1996) 105e115.
[40] A. Viegas, J. Manso, F.L. Nobrega, E.J. Cabrita, Saturation-transfer difference
(STD) NMR: a simple and fast method for ligand screening and character-
ization of protein binding, J. Chem. Educ. 88 (2011) 990e994.
[41] A. Viegas, J. Manso, M.C. Corvo, M.M.B. Marques, E.J. Cabrita, Binding of
ibuprofen, ketorolac, and diclofenac to COX-1 and COX-2 studied by satura-
tion transfer difference NMR, J. Med. Chem. 54 (2011) 8555e8562.
[42] C. Dalvit, G. Fogliatto, A. Stewart, M. Veronesi, B. Stockman, WaterLOGSY as a
method for primary NMR screening: practical aspects and range of applica-
bility, J. Biomol. NMR 21 (2001) 349e359.
[
[
[
[
14] C.R. Raetz, C. Whitfield, Lipopolysaccharide endotoxins, Annu. Rev. Biochem.
71 (2002) 635e700.
15] T.J. Wyckoff, C.R. Raetz, J.E. Jackman, Antibacterial and anti-inflammatory
agents that target endotoxin, Trends Microbiol. 6 (1998) 154e159.
16] C.R. Raetz, C.M. Reynolds, M.S. Trent, R.E. Bishop, Lipid A modification systems
in gram-negative bacteria, Annu. Rev. Biochem. 76 (2007) 295e329.
17] M. Sarkar, L. Maganti, N. Ghoshal, C. Dutta, In silico quest for putative drug
targets in Helicobacter pylori HPAG1: molecular modeling of candidate en-
zymes from lipopolysaccharide biosynthesis pathway, J. Mol. Model. 18
[43] M. Heinig, D. Frishman, STRIDE: a web server for secondary structure
assignment from known atomic coordinates of proteins, Nucleic Acids Res. 32
(2004) W500eW502.
(
2012) 1855e1866.
[44] F.H. Niesen, H. Berglund, M. Vedadi, The use of differential scanning fluo-
rimetry to detect ligand interactions that promote protein stability, Nat.
Protoc. 2 (2007) 2212e2221.
[45] C.A. Lipinski, F. Lombardo, B.W. Dominy, P.J. Feeney, Experimental and
computational approaches to estimate solubility and permeability in drug
discovery and development settings, Adv. Drug Deliv. Rev. 46 (2001) 3e26.
[
[
18] M.R. Birck, R.W. Woodard, Aquifex aeolicus 3-deoxy-D-manno-2-octulosonic
acid 8-phosphate synthase: a new class of KDO 8-P synthase? J. Mol. Evol.
5
19] D.J. Krosky, R. Alm, M. Berg, G. Carmel, P.J. Tummino, B. Xu, W. Yang, Heli-
cobacter pylori 3-deoxy-D-manno-octulosonate-8-phosphate (KDO-8-P)
2 (2001) 205e214.

202
S. Cho et al. / European Journal of Medicinal Chemistry 108 (2016) 188e202
[
[
46] C. Dalvit, P. Pevarello, M. Tato, M. Veronesi, A. Vulpetti, M. Sundstrom, Iden-
tification of compounds with binding affinity to proteins via magnetization
transfer from bulk water, J. Biomol. NMR 18 (2000) 65e68.
47] J.L. Wagstaff, S.L. Taylor, M.J. Howard, Recent developments and applications
of saturation transfer difference nuclear magnetic resonance (STD NMR)
spectroscopy, Mol. Biosyst. 9 (2013) 571e577.
titration calorimetry in bioinorganic chemistry, J. Biol. Inorg. Chem. 15 (2010)
1183e1191.
[53] K.L. Haas, K.J. Franz, Application of metal coordination chemistry to explore
and manipulate cell biology, Chem. Rev. 109 (2009) 4921e4960.
[54] A.W. Boots, G.R. Haenen, A. Bast, Health effects of quercetin: from antioxidant
to nutraceutical, Eur. J. Pharmacol. 585 (2008) 325e337.
[
[
48] G. Otting, NMR studies of water bound to biological molecules, Prog. Nucl.
Magn. Reson. Spectrosc. 31 (1997) 259e285.
49] C. Ludwig, P.J.A. Michiels, X. Wu, K.L. Kavanagh, E. Pilka, A. Jansson,
U. Oppermann, U.L. Gunther, SALMON: solvent accessibility, ligand binding,
and mapping of ligand orientation by NMR spectroscopy, J. Med. Chem. 51
[55] S.C. Bischoff, Quercetin: potentials in the prevention and therapy of disease,
Curr. Opin. Clin. Nutr. Metab. Care 11 (2008) 733e740.
[56] K.A. Lee, S.H. Moon, K.T. Kim, A.F. Mendonca, H.D. Paik, Antimicrobial effects of
various flavonoids on Escherichia coli O157:H7 cell growth and lipopolysac-
charide production, Food Sci. Biotechnol. 19 (2010) 257e261.
[57] M. Kalegari, Antibacterial, allelopathic and antioxidant activity of extracts and
compounds from Rourea induta Planch. (Connaraceae), J. Appl. Pharm. Sci. 2
(9) (2012) 061e066.
[58] S. Lee, D.S. Shin, K.B. Oh, K.H. Shin, Antibacterial compounds from the leaves of
Acanthopanax senticosus, Arch. Pharm. Res. 26 (2003) 40e42.
[59] A.M. Metwally, A.A. Omar, F.M. Harraz, S.M. El Sohafy, Phytochemical inves-
tigation and antimicrobial activity of Psidium guajava L. leaves, Pharmacogn.
Mag. 6 (2010) 212e218.
(
2008) 1e3.
[
[
[
50] R. Vainer, V. Belakhov, E. Rabkin, T. Baasov, N. Adir, Crystal structures of
Escherichia coli KDO8P synthase complexes reveal the source of catalytic
irreversibility, J. Mol. Biol. 351 (2005) 641e652.
51] L. Unione, S. Galante, D. Diaz, F.J. Canada, J. Jimenez-Barbero, NMR and mo-
lecular recognition. The application of ligand-based NMR methods to monitor
molecular interactions, MedChemComm 5 (2014) 1280e1289.
52] N.E. Grossoehme, A.M. Spuches, D.E. Wilcox, Application of isothermal