This article was downloaded by: [University of York]

On: 14 November 2014, At: 02:07

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Journal of Biomolecular Structure and Dynamics

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tbsd20

In silico approach towards identification of potential

inhibitors of Helicobacter pylori DapE
a a
Rahul Shubhra Mandal & Santasabuj Das
a
Biomedical Informatics Centre, National Institute of Cholera and Enteric Diseases, P-33,
C.I.T. Road, Scheme XM, Beleghata, Kolkata 700 010, India
Published online: 09 Sep 2014.

To cite this article: Rahul Shubhra Mandal & Santasabuj Das (2014): In silico approach towards identification of potential
inhibitors of Helicobacter pylori DapE, Journal of Biomolecular Structure and Dynamics, DOI: 10.1080/07391102.2014.954272

To link to this article: http://dx.doi.org/10.1080/07391102.2014.954272

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 Journal of Biomolecular Structure and Dynamics, 2014
http://dx.doi.org/10.1080/07391102.2014.954272

In silico approach towards identification of potential inhibitors of Helicobacter pylori DapE

Rahul Shubhra Mandal and Santasabuj Das*
Biomedical Informatics Centre, National Institute of Cholera and Enteric Diseases, P-33, C.I.T. Road, Scheme XM, Beleghata,
Kolkata 700 010, India

Communicated by Ramaswamy H. Sarma

(Received 7 February 2014; accepted 10 August 2014)

Helicobacter pylori is a gastric mucosal pathogen and is associated with diseases like peptic ulcer and gastric cancer. To
combat H. pylori infection, there is an urgent need for new class of antibiotics due to the emergence of drug-resistant
strains. Enzymes involved in bacterial lysine biosynthetic pathways may be potential targets for antibacterial drug
development, since lysine is an essential component of the bacterial peptidoglycan cell wall. No pathway exists for
Downloaded by [University of York] at 02:07 14 November 2014

lysine biosynthesis in humans; hence, the inhibitors targeting bacterial enzymes may have selective toxicity.
dapE-encoded N-succinyl-L,L-diaminopimelic acid desuccinylase (DapE) is a critical enzyme of this pathway and
deletion of DapE gene is lethal to H. pylori, since the organism has no alternative pathway for lysine biosynthesis. In
this study, we reported a 3D model structure of H. pylorie DapE, which consisted of a catalytic domain and a dimeriza-
tion domain generated by MODELLER software. We also confirmed the stability of the modeled structure through 10 ns
molecular dynamics simulation using GROMACS software. Next, to identify potential small molecule inhibitors of
DapE, drug-like small molecule-screening library was generated. This was performed by Tanimoto-based similarity
searching in the PubChem Database with DapE substrate L,L-SDAP as a query molecule, followed by fragment-based
docking approach using GLIDE XP. This approach identified two potential substrate-competitive small molecule
inhibitors of DapE. These new molecules may provide a starting point to search for novel therapeutics.
Keywords: Helicobacter pylori; DapE; homology modeling; fragment docking; inhibitor

Introduction further suggesting the importance of this pathway for drug

Helicobacter pylori is a Gram-negative enteric bacterium
that colonizes the human gastric mucosa and is implicated
in peptic ulcer diseases (Blaser et al., 1991; Covacci et al.,
1993; Crabtree et al., 1991) and malignant neoplasms of
the stomach (Blaser et al., 1995; Nomura et al., 1991;
Parsonnet et al., 1991; Talley et al., 1991). Enzymes
involved in the lysine biosynthetic pathway in bacteria
may be a potential drug target, as they are absent in
humans, while bacteria cannot supplement lysine from the
outside sources. In addition, another product of this path-
way, meso-diaminopimelate (mDAP), is an essential com-
ponent of the peptidoglycan layer of the bacterial cell wall
and critical for bacterial survival. This may be the reason
why multiple pathways (Succinylase, Dehydrogenase and
Acetylase pathway) are present in bacteria for lysine bio-
genesis. However, Helicobacter pylori has only the Suc-
cinylase pathway for lysine biosynthesis. Targeting this
pathway for inhibitor designing may be an effective
approach to combat H. pylori infection. Several enzymes
of this pathway, such as dihydrodipicolinate synthase
(dapA) and dihydrodipicolinate reductase (dapB) were tar-
geted earlier for inhibitor designing (Garg, Tewari, &
Raghava, 2010; Paiva et al., 2001; Scapin et al., 1997),
targeting. More importantly, deletion of dapE-encoded
N-succinyl-L,L-diaminopimelic acid (L,L-SDAP) desuc-
cinylase (DapE; EC 3.5.1.18) gene was lethal to H. pylori
(Karita et al., 1997). Many pathogenic Gram-positive and
Gram-negative bacteria, such as Bordetella pertussis,
Acinetobacter baumannii (MDRAB), Mycobacterium
tuberculosis, Pseudomonas aeruginosa, Escherichia coli
(O157:H7), Yersinia pestis, Vibrio cholerae, Rickettsia
prowazekii, H. pylori, Staphylococcus aureus (strain
MRSA252), Streptococcus pneumonia, Enterococcus
faecium, Salmonella enterica, and Haemophilus
influenzae possess DapE (Blaser et al., 1995; Gillner,
Armoush, Holz, & Becker, 2009; Nomura et al., 1991).
Several multidrug-resistant bacterial strains are also
included in this list, indicating that inhibitors of DapE
may be used as broad-spectrum antibacterial agents. There
are reports of substrate-competitive inhibitors developed
against DapE, and chemical molecules like penicillamine,
captopril, and some Boronic acid analogs were tested for
DapE inhibitory activities (Gillner et al., 2009; Gillner,
Becker, & Holz, 2013). Among these inhibitors,
L-Captopril exhibited lowest inhibitory concentration
(IC50 = 3.3 μM) and a measured Ki of 1.8 μM (Gillner

*Corresponding author. Email: dasss@icmr.org.in

© 2014 Taylor & Francis

 2 R.S. Mandal and S. Das
et al., 2009). However, published data suggested that
Captopril lacks specificity towards DapE (Uda & Creus,
2011). There is ample scope to review the proposed DapE
inhibitors and develop new set of molecules with higher
specificity. Here, we took a different approach to develop
DapE inhibitors through in silico modeling and docking,
and finally came up with a set of related molecules having
better binding efficacy than the known DapE substrate L,
L-SDAP. These inhibitors may be used as substrate-com-
petitive chemical agents against H. pylori.
Materials and methods
Homology modeling
The protein sequence of DapE of H. pylori was retrieved
from National Center for Biotechnology Information
Downloaded by [University of York] at 02:07 14 November 2014
(NCBI) sequence (Wheeler et al., 2008) database. The
template structure was selected from RCSB Protein Data
Bank (PDB) using PSI BLAST (Position-Specific Iter-
ated BLAST) and Hidden Markov Model-based predic-
tion method, HHPRED. The target-template alignment
between DapE sequences from H. pylori and H. influ-
enzae was performed by dynamic programming-based
Align2D script of MODELLER (Sali & Blundell, 1993).
Three-dimensional (3D) model of DapE was generated
through MODELLER program and the disordered loops
were refined using ModLoop server (http://modbase.
compbio.ucsf.edu/modloop/). To get better relaxation and
more correct arrangement of the side chain and main
chain atoms, optimization was performed with Protein
Preparation Wizard in Maestro 9.3 using OPLS 2005
force field with a .3 Å RMSD tolerance of backbone
atoms. Finally, models were energy minimized using
steepest descent followed by Conjugate Gradient method
in Discovery Studio 2.5 software. Analysis and valida-
tion of the structures were performed with the aid of
Structural Analysis and Verification Server (SAVES)
(nihserver.mbi.ucla.edu/SAVES/) by analyzing ERRAT
score (Colovos & Yeates, 1993) VERIFY_3D (Lüthy,
Bowie, & Eisenberg, 1992) score, PROCHECK
(Laskowski, MacArthur, Moss, & Thornton, 1993), and
Ramachandran Plot (Hooft, Sander, & Vriend, 1997).
Figures were prepared using Discovery Studio 2.5 and
PyMOL software.
Molecular dynamic simulation
The structural stability of the generated model was
evaluated using molecular dynamic (MD) simulation in
GROMACS 4.5 software (Van Der Spoel et al., 2005).
Before the simulation process, the modeled protein
structure was optimized using Protein Preparation
Wizard in Maestro 9.3. Protein backbone and side
chains were optimized using OPLS 2005 force field
with a tolerance of backbone RMSD deviation of .3 Å.
The simulation was performed using cubic cell geome-
try, and the default simple point charge (SPC) water
was added to the box. The model was subjected to
molecular dynamics simulation in water at 300 K tem-
perature and for 10 nanoseconds using GROMOS 43al
force field. A total of 12,832 water molecules and, to
neutralize the overall charge of the system, 4 Na+ ions
were added to the system. Periodic boundary condition
was applied and the distance between the grid box and
the protein was set to .8 nm. First, a Steepest Descent
minimization for 10,000 steps was performed to remove
bad van der Waals contacts. Then, a 20 ps position-
restrained simulation was performed by keeping the
protein coordinates fixed and making the water mole-
cules to soak into the macromolecule. Finally, a 10 ns
MD simulation was executed for the modeled DapE
proteins. During simulation, the bond length was con-
strained by the application of LINCS algorithm (Hess,
Bekker, Berendsen, & Fraaije, 1997), while SETTLE
algorithm (Miyamoto & Kollman, 1992) was used to
restrain the water molecules. A 2 femtoseconds time
step was used for the simulation. Leapfrog algorithm
was employed in the NPT ensemble to separately cou-
ple each of the components like protein, water mole-
cules, and ions. The Berendsen temperature and
pressure coupling (Berendsen, Postma, van Gunsteren,
DiNola, & Haak, 1984) constants were set to .1 and .5,
respectively, to keep the system in a stable environment
(300 K temperature and 1 bar pressure). During these
steps, the particle-mesh Ewald method for long-range
electrostatics was used, whereas a 10 Å cut-off was
applied to truncate the short-ranged components. A 14
Å cut-off for van der Waals interactions were applied,
since GROMOS-96 force field was parameterized with
a Lennard-Jones cut-off of 14 Å and rvdw is not
restricted by the force-field parameterization scheme.
The visual inspection and analysis were performed by
Discovery Studio 2.5. The secondary structure profile
of the protein during simulation period was visualized
using Define Secondary Structure of Proteins (DSSP)
program (Kabsch & Sander, 1983). Microsoft Excel
program was used for preparation of the graphs. Both
energy and root mean square deviation (RMSD) plots
were derived from the respective trajectory output file.
Small molecule library generation for virtual screening
The small molecule library for virtual screening was gen-
erated by similarity searching in PubChem Database
(http://pubchem.ncbi.nlm.nih.gov/) using DapE substrate
L,L-SDAP (PubChem CID: 25202447). In the PubChem,
we used Tanimoto-based similarity searching method,
which utilizes the 2D structure fingerprints of small mol-
ecules. Similarity was measured using the Tanimoto
 Identification of potential inhibitors of Helicobacter pylori DapE
equation and the PubChem dictionary-based binary fin-
gerprint, which come with PubChem as default. The
Tanimoto cut-off for compound selection was set at .8 or
80%. The resultant molecules were further scaled down
by applying Rule of 5 criteria to obtain the drug-like
compounds only.
Molecular docking
Molecular docking and scoring calculations were per-
formed using Schrodinger 2012 suite (Maestro 9.3).
Drug-like similar molecules, including L,L-SDAP and
known inhibitor Captopril were prepared for docking
using LigPrep (Brooks, Daniel, Sung, & Guida, 2008).
Protein Preparation Wizard in Maestro 9.3 was used to
prepare the receptor (modeled DapE), and a 20 Å grid
box was generated on the receptor by picking the active
Downloaded by [University of York] at 02:07 14 November 2014
site residues in Glide 5.8. Before running virtual screen-
ing, different ADME constraints were applied, such as
Lipinski filter and Reactive filter to filter ligands with
suitable pharmacological properties and no reactive func-
tional group. Finally, docking was performed using Stan-
dard Precision method, and the ligands having better
binding affinity than L,L-SDAP were selected. The
resulted molecules were further docked using Glide Extra
Precision (XP) method and the molecules having better
binding affinity than known inhibitor Captopril were
selected.
Fragment-based molecule generation and screening
Selected molecules from the previous step were consid-
ered for fragment generation using inbuilt script in
Schrodinger suit, fragment_molecule.py. This script
broke up a set of molecules into fragments based on
some simple RECAP rules (Lewell, Judd, Watson, &
Hann, 1998). All these fragments were again docked into
the DapE active site using Glide XP method. Finally, the
potential fragments were selected on the basis of ligand
efficiency metric and spatial diversity using frag-
ment_selector.py script, which was used in post-process
results from a Glide XP fragment-docking calculation. It
first applied a ligand efficiency score to each pose and
then selected the top fragment for each region of the
active site based on the ligand efficiency. The selected
fragments were used to generate novel molecules by
joining them using combine fragments method in
Schrodinger 2012 suite, which accepted docked poses of
fragments and combined the neighboring fragments to
build a new molecule.
Results and discussion
Homology modeling
The sequence of H. pylori DapE retrieved from NCBI
(gi number: 2258460) consisted of 388 amino acids. We
3
employed two independent search methods, namely the
PSI BLAST (Position-Specific Iterated BLAST) and
HHPRED to find out suitable template structures in
RCSB PDB. DapE of H. influenzae containing two Zn
ions Zn(II) (PDB ID : 3IC1_A) and mono-zinc-bound
DapE structures, 3ISZ_A and 1VGY_A were the best
hits identified by both the search methods. However,
3IC1_A was considered as the template for model build-
ing, since both the metal ions were important for the full
enzymatic activity of DapE (Nocek, Gillner, Fan, Holz,
& Joachimiak, 2010). However, we truncated 80 residues
form the N-terminal end, which fell outside the active
site and were not aligned properly. This improved the
sequence identity between the two DapE molecules from
37 to 42%.
Homology model of H. pylori DapE, constructed
using MODELLER, generated 20 independent structures.
The best model was selected on the basis of minimum
DOPE (Discrete Optimized Protein Energy) score (Shen
& Šali, 2006), and the disordered loops of the modeled
structure were optimized using ModLoop server. Active
site Zn(II) ions were added by superposing the modeled
structure with the template (PDB ID : 3IC1_A) and sav-
ing the coordinates of Zn(II) ions. It was found that five
Zn-binding residues, such as H67, D100, E135, E163,
and H349 in the template structure were well conserved
and identical to H84, D115, E147, E175, and H360,
respectively, of in H. pylori DapE (Figure S1). The 3D
structure of the truncated DapE of H. pylori consisted of
one catalytic and one dimerization domain (Figure 1).
The domains were connected by a small hinge region
(residues 188-190 and 304-309) allowing movement of
the dimerization domain with respect to the catalytic
domain. The Catalytic domain was composed of residues
1-191 and 303-387, and contained 5 α-helices and 9
β-sheets, while the dimerization domain had 4 antiparal-
lel β-sheets and 2 α-helices. A similar folding pattern
was also reported for other DapE structures (Nocek
et al., 2010). Search for structural homologues of our
modeled structure using the DALI Server (Holm &
Sander, 1998) identified several closely related homologs
of DapE from other bacterial species, such as Neisseria
meningitidis MC58 (PDB ID: 4O23, Z-score = 34.3,
RMSD = 1.6) and V. cholera DapE (PDB ID: 4OP4,
Z-score = 23.1, RMSD = 1.5). This further confirmed
the structural integrity of the DapE model.
Validation of the model
Different structure verification programs, such as
PROCHECK, VEFIFY3D, and ERRAT (http://nihserver.
mbi.ucla.edu/SAVES) were used to evaluate the 3D
model of H. pylori DapE. Ramachandran plot drawn
through PROCHECK program resulted in 92.2% resi-
dues falling within the most favored regions [A,B,L] and
 4 R.S. Mandal and S. Das
Downloaded by [University of York] at 02:07 14 November 2014
Figure 1. Homology model of truncated (residue number 81-
387) DapE protein from Helicobacter pylori. Individual
domains are labeled and the secondary-structure elements are
colored (yellow, β-strand; red, α-helices; green, loop). Six
β-sheets (β1, β4, β5, β6, β11, and β12) are making the core of
the catalytic domain and are sandwiched between 5 α-helices,
α2 and α6 on one side and α1, α5, and α7 on the other side.
remaining 7.8% residues being in the additional allowed
regions [a, b, l, p] (Figure 2). This specifies that the
protein backbone dihedral angles phi (Φ) and psi (Ψ)
occupied reasonably accurate positions in the 3D model.
Protein main chain and side chains atoms were also posi-
tioned accurately in the model (Figure S2a and S2b). In
addition, VERIFY3D score (80.86%) and ERRAT score
(94.898) strongly supported the compatibility and overall
quality of the 3D atomic structure of the model
(Figure S3). Finally, the backbone conformation of the
model was confirmed by superimposition with the
template structure (3IC1), which resulted in an RMSD
value of .210 Å that fell within the acceptable range
(Figure S4) (Maiorov & Crippen, 1994).
Molecular dynamics simulation study
During 10 ns molecular dynamics (MD) simulation, it
was observed that the potential energy of the modeled
protein was initially high, but came down during the
Figure 2. Ramachandran plot for the DapE model.
course of simulation. The RMSD profile was quite stable
after 6 ns simulation period and showed linear deviation
till 10 ns. The RMSD fluctuation plot showed that the
C-alpha backbone deviation during the simulation
process was within the acceptable range (.3–.4 nm)
(Figure 3). The modeled DapE structure was also ana-
lyzed by DSSP profile, backbone root-mean-square
deviation (RMSD), and root-mean-square fluctuation
(RMSF) of the backbone C-alpha atom. DSSP profile
suggested that the model was quite stable, since no
deformation was observed in the conserved α-helices and
β-sheet regions (Figure S5). RMSF plot showed that the
α-helices and β-sheets were more stable (fluctuation
~.2 nm) than the loop regions (Figure S6), suggesting
that the modeled protein structure was quite stable under
aqueous environment. Finally, molecular docking was
performed with the structure extracted from the last
frame of the 10 ns MD run.
Active site analysis
The active site residues are usually conserved between
functionally related proteins and the catalytic activity
often depends on one or more specific and highly con-
served residues (Milik, Szalma, & Olszewski, 2003). By
multiple sequence alignments and structural superimposi-
tion of DapE enzymes from different sources, we found
that the active site residues of these enzymes were sig-
nificantly conserved across bacterial species (Figure 4).
H84, D115, E147, E175, and H360 residues over the cat-
alytic domain of H. pylorie DapE were structurally con-
served in all other DapE proteins (Figure S7) and these
residues were critical for metal binding (Figure S1). All
 Identification of potential inhibitors of Helicobacter pylori DapE
Figure 3. RMSD trajectory of the modeled DapE structure during 10 ns simulation, showing a linear backbone deviation, especially
after 6 ns.
DapE proteins characterized to date are medium-sized,
Downloaded by [University of York] at 02:07 14 November 2014
dimeric enzymes (41.6 kDa/subunit), consisting of active
site zinc ions required for their activity (Nocek et al.,
2010). The active site residues that function as metal
ligands in the structurally characterized M28 family mem-
bers, such as the leucine aminopeptidase (AAP) from Aer-
omonas proteolytica and the carboxypeptidase (CPG2)
from Pseudomonas spp. strain RS-16 are conserved in all
DapE sequences (Born, Zheng, & Blanchard, 1998;
Makarova & Grishin, 1999). Both CPG2 and AAP pos-
sess a (μ-aquo) (μ-carboxylato) dizinc (II) core with one
terminal carboxyl group and one histidine residue at each
metal site (Chevrier et al., 1994; Greenblatt et al., 1997),
which is similar to DapE active site (Bienvenue, Gilner,
Davis, Bennett, and Holz 2003; Born et al., 1998;
Cosper et al., 2004; Davis et al., 2006). It was also
reported that similar to AAP, DapE enzymes that con-
tained only one tightly bound Zn ion, exhibited ∼60% of
the total activity (Born et al., 1998; Bienvenue et al.,
2003). Thus, both metal ions seem to be required for full
enzymatic activity of DapE (Nocek et al., 2010).
Docking of the substrate and known inhibitor into the
DapE active site
Since no co-crystal structure is available for DapE-bound
substrate L,L-SDAP, the actual orientation of the sub-
strate within the active site is unknown. To get the
approximate pose of the substrate-bound DapE structure,
we docked the L,L-SDAP structure (retrieved from Pub-
Chem database, CID 25202447) with DapE using Glide
XP in Schrodinger 2012 suite. The final pose of the
docked substrate within the active site was chosen based
on Glide Emodel score and there were no constrains
imposed on the chemicals to form bonds with metal ions.
The binding affinity, denoted by G Score (Glide, version
5.5, Schrödinger, LLC, New York, NY, 2009) of the
resultant interaction was −5.17 kcal/mol, and the sub-
strate bound in an extended mode (Nocek et al., 2010)
5
within the smile-shaped active site covering the whole
length of the pocket (Figure 5(a)). By analyzing the
interaction of the substrate within the pocket, we found
that the peptide bond was positioned over the active site
metal ions, while the rest of the substrate was further sta-
bilized by the interaction of carboxyl groups of the sub-
strate with the negatively charged residue E146
(equivalent to E134 of H. influenzae ZnZn_DapE)
through H-bonding. The second metal ion also coordi-
nated with the carboxyl side chain of the substrate,
which supported the hypothesis proposed by Nocek
et al. (2010) (Figure 5(b)). Importantly, E146 residue
was found to be structurally conserved in other known
DapE crystal structures (Figure S7), and was proposed to
act as the general acid/base during the hydrolysis reac-
tion catalyzed by DapE (Davis et al., 2006). These facts
proved the reliability of the docked conformation. Gillner
et al. reported several micromolar inhibitors of DapE
from H. influenzae through a screen biased toward the
compounds containing zinc-binding groups (ZBG’s),
such as thiols, carboxylic acids, boronic acids, phospho-
nates, and hydroxamates (Gillner et al., 2009). Out of
these inhibitors, L-captopril was found to be the most
potent. In vitro antimicrobial activity of L-captopril was
demonstrated against E. coli. It inhibited DapE with
IC50 of 3.3 μM and a measured Ki of 1.8 μM. To get
an approximate idea about H. pylori DapE-Captopril
interaction, we docked L-captopril into the DapE active
site. The carboxyl moiety of L-captopril interacted with
the second Zn(II) through metal coordination and occu-
pied the similar active site region as L,L-SDAP with
higher binding affinity (G Score = −7.24 kcal/mol)
(Figure S8). However, Captopril is not a specific inhibi-
tor of DapE and may have additional cellular targets,
since it inhibits bacterial growth in DapE-mutant strain
(Uda & Creus 2011). To find out novel inhibitors of
DapE with higher affinity and more specificity, we used
DapE–Captopril interaction as the reference for virtual
screening of the small molecule library.
 6 R.S. Mandal and S. Das
Downloaded by [University of York] at 02:07 14 November 2014

Figure 4. Multiple sequence alignments of H. pylori DapE with other DapE sequences having crystal structures (V. cholerae (PDB
ID: 3T68), N. meningitidis (PDB ID: 1VGY), and H. influenzae (PDB ID: 3IC1)) generated using ClustalX software, showing con-
served active site residues marked by Red arrow.

Small molecule library generation and virtual
screening
We performed virtual screening of the PubChem data-
base to find out molecules similar to DapE substrate L,
L-SDAP. The scheme of our approach is shown in
Figure 6. Using 80% similarity threshold for the screen,
a total of 3697 molecules were obtained, out of which
2160 molecules had drug-like properties according to the
Lipinski Rule of 5 (Lipinski, Lombardo, Dominy, &
Feeney 2001). We screened all the drug-like molecules
by Glide SP method using a binding affinity cut-off of
~−5.0 kcal/mol to search for the top hits, which identi-
fied 15 molecules (Supp. Table 1). Next, we re-docked
these molecules to DapE using Glide XP to get better
approximation about the interactions. As expected, we
found better binding affinity and pose within the DapE
active site. On the basis of electrostatic interactions with
the critical residues of the active site and better binding
 Identification of potential inhibitors of Helicobacter pylori DapE 7
Downloaded by [University of York] at 02:07 14 November 2014

Figure 5a. Interaction of L,L-SDAP within the smile-shaped active site of DapE (H bonds are shown by yellow dotted lines).

Figure 5b. Interaction of L,L-SDAP within the smile-shaped active site of DapE (H bonds are shown by pink arrows and the metal
coordination shown by straight lines).

affinity estimates than Captopril (<−7 kcal/mol), we
selected top 11 hits (Supp Table 2). The glide ligand effi-
ciency and rotatable bond count of these molecules ran-
ged from −.354 to −.729 and 5–12, respectively, with 9
molecules out of 11 having >6 rotatable bonds (Figure 7).
In contrast, the known inhibitor Captopril has three rotat-
able bonds and a ligand efficiency of −.517. We decided
to design molecules, which would be able to make
strong interactions within the active site pocket while
keeping minimum amount of flexibility and would have
higher specificity than captopril. Similar virtual screening
approach was employed for other enzymes like dihydro-
dipicolinate synthase (DHDPS) of M. tuberculosis lysine
biosynthesis pathway. The inhibitors identified in this
screen had GScore values between −8 and −9 kcal/mol
and the molecular weight ranged between 162–291 Da
(Garg et al., 2010).
Fragment-based molecule generation and screening
Increasing number of rotatable bonds and poor ligand
efficiency indicate the lack of specificity towards the
 8 R.S. Mandal and S. Das
Downloaded by [University of York] at 02:07 14 November 2014
Figure 6. The overall scheme for the screening of substrate-competitive DapE inhibitor.
Figure 7. Binding affinity (XP GScore) and the ligand effi-
ciency of 11 molecules form PubChem having better binding
affinity than the known DapE inhibitor L-Captopril is plotted
along the Y- and X-axis, respectively. Each molecule is colored
according to their rotatable bond count number. PubChem CID
for molecules having comparatively better ligand efficiencies is
labeled.
target. The designed ligands should have less number of
heavy atoms, higher binding affinity value, and mini-
mum rotatable bond counts to increase the efficiency and
specificity of inhibitor activity (Veber et al., 2002).
Hence, they are expected to be small sized molecules.
To design such inhibitors, we fragmented 11 selected
molecules using fragment_molecule.py script to generate
novel molecular scaffolds, which resulted in 339 frag-
ments (Figure S9). Fragment molecules are represented
here as the respective PubChem CID number followed
by fragment number within parentheses. These fragments
were again docked into the DapE active site to get the
best pose of each fragment within the pocket. From the
docking results, we successfully selected 21 potential
fragments having optimal interactions within the pocket
using fragment_selector.py script (Supp Table 3 and
Figure S10). Further analysis of these 21 fragment-like
molecules revealed eight potential fragments (out of
which six are unique structures) with lower molecular
weight (<200 Da), less number of rotatable bonds (<6),
and higher binding affinities (<−10 kcal/mol), reported
as critical parameters for drug-like inhibitors (Veber
et al., 2002), compared with the previous 11 molecules
(Figure 8). To find out if any small molecule that was
similar to the above fragments already existed, we
searched the PubChem database and found that frag-
ments 5289281 (frag-35), 23644489 (frag-32), and
51351649 (frag 27) were reported in PubChem with CID
numbers 1715066, 22558288, and 12679627, respec-
tively. Among these three fragments, 5289281 and
23644489 yielded GScore values of −10.6 kcal/mol and
−10.0 kcal/mol, respectively. In contrast, despite a G
Score value of −10.3 kcal/mol, fragment 51351649 had
more number (6) of rotatable bonds compared with the
 Identification of potential inhibitors of Helicobacter pylori DapE
Figure 8. The plot showing potential fragments (labeled) hav-
ing low molecular weight, minimum number of rotatable
bonds, and higher ligand efficiency. Fragments 23644489 (frag
63) and 51351649 (frag 28) are identical with 5289281 (frag
Downloaded by [University of York] at 02:07 14 November 2014
35) and 11948933 (frag 26). The positional overlap of these
identical molecules on the plot also reflects docking accuracy.
other two fragments, 5289281(frag 35) and 23644489
(frag 32) with five and three bonds, respectively. These
fragments had comparable number of rotatable bonds as
Captopril and were able to make metal coordination with
the Zn2 atom through carboxylic acid side chain, but
interacted more strongly with the critical active site resi-
dues of DapE due to the higher binding affinity. More-
over, they interacted well with the active site critical
residues including E146 and E147 through H-bonding.
Interestingly, we found a conserved H-bond interaction
with the T336 backbone of DapE that was similar to L,
Figure 9a. Interaction of 5289281 (frag-35) [CID 1715066] within the DapE active site of H. pylori showing H-bond interaction
with the conserved active site residues E146 and E147 and metal coordination with Zn2 through carboxyl side chain.
9
L-SDAP interaction (Figures 9(a)–9(c) and 5(b)). Thus,
fragments 5289281 (frag 35) and 23644489 (frag 32)
may be more specific inhibitors of DapE than Captopril,
which lacks specificity.
In a parallel approach to generate novel molecular
scaffolds out of the above 339 fragments, we joined
the fragments using combine_fragments.py script in
Schrodinger 2012 suite. In this method, the said script
took an input of 339 docked fragments and tried to
join neighboring fragments to build a new molecule.
The joined fragment-based molecules were again
docked using Glide XP and analyzed on the basis of
mainly four physiochemical parameters, such as molec-
ular weight, rotatable bond count, binding affinity of
interaction, and ligand efficiency. This analysis enabled
us to successfully select 80 molecules with binding
affinity <−10 kcal/mol. However, none of the joined
fragments showed better drug-like properties than the
previously identified fragments, 5289281 (frag 35) and
23644489 (frag 32), since the joined fragments had
higher number of rotatable bonds (between 10 and 20)
and heavy atom counts, resulting in less specificity
and poor ligand efficiency (ranges between .25 and
.60) (Figure S11).
Docking of fragment-like small molecules with other
DapE crystal structures
Since DapE active site residues are conserved across spe-
cies, we evaluated if the fragments we generated were
capable of binding to other DapE proteins. For this
 10 R.S. Mandal and S. Das
Downloaded by [University of York] at 02:07 14 November 2014

Figure 9b. Interaction of 23644489 (frag-32) [CID 22558288] within the DapE active site of H. pylori showing H-bond interaction
with the conserved active site residues E146 and E147 and metal coordination with Zn2 through carboxyl side chain.

Figure 9c. Interaction of 51351649 (frag 27) [CID 12679627] within the DapE active site of H. pylori showing H-bond interaction
with the conserved active site residues E146 and E147 and metal coordination with Zn2 through carboxyl side chain.

purpose, we preformed molecular docking with the DapE ions in the active site. Docking results were similar to
crystal structures of V. cholerae (PDB ID: 3T6 M) and that of H. pylori DapE and the binding affinities of
H. influenzae (PDB ID: 3IC1), which possess two Zn the complexes were also comparable. The interacting
 Identification of potential inhibitors of Helicobacter pylori DapE 11
Downloaded by [University of York] at 02:07 14 November 2014

Figure 10a. Interaction of 5289281 (frag-35) [CID 1715066] within the DapE active site of H. influenzae showing H-bond interac-
tion with the conserved active site residues E134 and E135 and metal coordination with Zn2 through carboxyl side chain.

Figure 10b. Interaction of 23644489 (frag-32) [CID 22558288] within the DapE active site of H. influenzae showing H-bond inter-
action with the conserved active site residues E134 and E135 and metal coordination with Zn2 through carboxyl side chain.

residues were also similar, suggesting that not only the
active site residues, but also the binding cavity possessed
similar architectures. The binding affinities of 5289281
(frag-35) and 23644489(frag-32) with H. influenzae
DapE were –8.5 kcal/mol and –8.6 kcal/mol, respectively
(Figures 10(a) and 10(b)). In contrast, the binding affini-
ties of these two molecules with V. cholerae DapE was
lower, −5.85 kcal/mol for 5289281(frag-35) and
−5.6 kcal/mol for 23644489(frag-32) (Figure 11(a) and
11(b)). The above results indicate that the binding affin-
ity of our chosen fragment-sized molecules varies with
DapE from different bacterial species. However, this
could be a good starting point to search for novel
broad-spectrum therapy against bacterial infections.
 12 R.S. Mandal and S. Das
Downloaded by [University of York] at 02:07 14 November 2014

Figure 11a. Interaction of 5289281 (frag-35) [CID 1715066] within the DapE active site of V. cholerae showing H-bond interaction
with the conserved active site residues E135 and E136 and metal coordination with Zn2 through carboxyl side chain.

Figure 11b. Interaction of 23644489 (frag-32) [CID 22558288] within the DapE active site of V. cholerae showing H-bond interac-
tion with the conserved active site residues E135 and E136 and metal coordination with Zn2 through carboxyl side chain.

Conclusions
In this study, using homology modeling technique we
generated the 3D structure of H. pylori DapE and vali-
dated its structural integrity using different structure vali-
dation servers and molecular dynamics simulation.
Employing small molecule docking and fragment-based
approach, we successfully identified three new drug-like
small molecules (PubChem CID: 1715066, 22558288,
and 12679627) with a potential to bind to the active site
of H. pylori DapE and enforce substrate-competitive
inhibition. The docked complexes showed molecular
interactions of the inhibitors with the critical active site
residues and two metal ions of DapE. It was also evident
from the interactions that the negatively charged resi-
dues, E146 and E147 of DapE were playing an impor-
tant role in ligand–protein interaction. Our molecules
showed higher affinity and ligand-efficiency towards
DapE active site than Captopril. We also evaluated the
binding potential of CID1715066 and CID22558288
with other known DapE crystal structures from H. influ-
enzae and V. cholerae having Zn(II) ions in their active
site. Interaction of inhibitors with DapE from different
bacterial spp resulted in differential intensity of interac-
tions, but all these interactions involved conserved active
 Identification of potential inhibitors of Helicobacter pylori DapE
site critical residues and Zn ions, which point towards a
promise as broad-spectrum inhibitors of DapE. This
study will help medicinal chemists to develop new thera-
peutics based on the small molecules we have reported.
Supplementary material
The supplementary material for this paper is available
online at http://dx.doi.10.1080/07391102.2014.954272.
Funding
This work was supported by Indian Council of Medical
Research [IRIS ID: 2013-1551G].
References
Downloaded by [University of York] at 02:07 14 November 2014
Berendsen, H. J. C., Postma, J. P. M., van Gunsteren, W. F.,
DiNola, A., & Haak, J. R. (1984). Molecular-dynamics
with coupling to an external bath. Journal of Chemical
Physics, 81, 3684–3690.
Bienvenue, D. L., Gilner, D. M., Davis, R. S., Bennett, B., &
Holz, R. C. (2003). Substrate specificity, metal binding
properties, and spectroscopic characterization of the dapE-
encoded-N-succinyl-L, L-diaminopimelic acid desucciny-
lase from Haemophilus influenzae. Biochemistry, 42,
10756–10763.
Blaser, M. J., Perez-Perez, G. I., Kleanthous, H., Cover, T. L.,
Peek, R. M., Chyou, P. H., … Nomura, A. (1995). Infec-
tion with Helicobacter pylori strains possessing cagA is
associated with an increased risk of developing adenocarci-
noma of the stomach. Cancer Research, 55, 2111–2115.
Blaser, M. J., Perez-Perez, G. I., Lindenbaum, J., Schneidman,
D., Van Deventer, G., Marin-Sorensen, M., & Weinstein,
W. M. (1991). Association of infection due to Helicobacter
pylori with specific upper gastrointestinal pathology.
Reviews of Infectious Diseases, 13, S704–S708.
Born, T. L., Zheng, R., & Blanchard, J. S. (1998). Hydrolysis
of N-succinyl-L, L-diaminopimelic acid by the Haemophi-
lus influenzae dapE-encoded desuccinylase: Metal activa-
tion, solvent isotope effects, and kinetic mechanism.
Biochemistry, 37, 10478–10487.
Brooks, W. H., Daniel, K. G., Sung, S. S., & Guida, W. C.
(2008). Computational validation of the importance of
absolute stereochemistry in virtual screening. Journal of
Chemical Information and Modeling, 48, 639–645.
Chevrier, B., Schalk, C., D’Orchymont, H., Rondeau, J.-M.,
Moras, D., & Tarnus, C. (1994). Crystal structure of
Aeromonas proteolytica aminopeptidase: A prototypical
member of the co-catalytic zinc enzyme family. Structure,
2, 283–291.
Colovos, C., & Yeates, T. O. (1993). Verification of protein
structures: Patterns of nonbonded atomic interactions.
Protein Science, 2, 1511–1519.
Cosper, N. J., Bienvenue, D. L., Shokes, J. E., Gilner, D. M.,
Tsukamoto, T., Scott, R., & Holz, R. C. (2004). The
dapE-encoded N-succinyl-L, L-diaminopimelic acid
desuccinylase from Haemophilus influenzae is a dinuclear
metallohydrolase. Journal of the American Chemical
Society, 125, 14654–14655.
Covacci, A., Censini, S., Bugnoli, M., Petracca, R., Burroni, D.,
Macchia, G., Massone, A., Papini, E., Xiang, Z., Figura, N.,
& Rappuoli, R. (1993). Molecular characterization of the
13
128-kDa immunodominant antigen of Helicobacter
pylori associated with cytotoxicity and duodenal ulcer.
Proceedings of the National Academy of Sciences, 90,
5791–5795.
Crabtree, J. E., Taylor, J. D., Wyatt, J. I., Heatley, R. V.,
Shallcross, T. M., Tompkins, D. S., & Rathbone, B. J.
(1991). Mucosal IgA recognition of Helicobacter pylori
120 kDa protein, peptic ulceration, and gastric pathology.
Lancet, 338, 332–335.
Davis, R., Bienvenue, D., Swierczek, S. I., Gilner, D. M.,
Rajagopal, L., Bennett, B., & Holz, R. C. (2006). Kinetic
and spectroscopic characterization of the E134A- and
E134D-altered dapE-encoded N-succinyl-l,l-diaminopimelic
acid desuccinylase from Haemophilus influenzae. JBIC
Journal of Biological Inorganic Chemistry, 11, 206–216.
Garg, A., Tewari, R., & Raghava, G. P. (2010). Virtual Screen-
ing of potential drug-like inhibitors against Lysine/DAP
pathway of Mycobacterium tuberculosis. BMC Bioinformat-
ics, 11, S53. doi:10.1186/1471-2105-11-S1-S53
Gillner, D., Armoush, N., Holz, R. C., & Becker, D. P. (2009).
Inhibitors of bacterial N-succinyl-L, L-diaminopimelic acid
desuccinylase (DapE) and demonstration of in vitro antimi-
crobial activity. Bioorganic & Medicinal Chemistry Letters,
19, 6350–6352. doi:10.1016/j.bmcl.2009.09.077
Gillner, D. M., Becker, D. P., & Holz, R. C. (2013). Lysine
biosynthesis in bacteria: A metallodesuccinylase as a
potential antimicrobial target. JBIC Journal of Biological
Inorganic Chemistry, 18, 155–163. doi:10.1007/s00775-
012-0965-1
Greenblatt, H. M., Almog, O., Maras, B., Spungin- Bialik, A.,
Barra, D., Blumberg, S., & Shoham, G. (1997). Streptomy-
ces griseus aminopeptidase: X-ray crystallographic structure
at 1.75 Å resolution. Journal of Molecular Biology, 265,
620–636.
Hess, B., Bekker, H., Berendsen, H. J. C., & Fraaije, J. G. E.
M. (1997). LINCS: A linear constraint solver for molecular
simulations. Journal of Computational Chemistry, 18,
1463–1472.
Holm, L., & Sander, C. (1998). Touring protein fold space with
Dali/FSSP. Nucleic Acids Research, 26, 316–319.
Hooft, R. W. W., Sander, C., & Vriend, G. (1997). Objectively
judging the quality of a protein structure from a Ramachan-
dran plot. Computer Applications in the Biosciences, 13,
425–430. doi:10.1093/bioinformatics/13.4.425
Kabsch, W., & Sander, C. (1983). Dictionary of protein second-
ary structure: Pattern recognition of hydrogen-bonded and
geometrical features. Biopolymers, 22, 2577–2637.
doi:10.1002/bip.360221211
Karita, M., Etterbeek, M. L., Forsyth, M. H., Tummuru, M. K.,
& Blaser, M. J. (1997). Characterization of Helicobacter
pylori dapE and construction of a conditionally lethal dapE
mutant. Infection and Immunity, 65, 4158–4164.
Laskowski, R. A., MacArthur, M. W., Moss, D. S., & Thorn-
ton, J. M. (1993). PROCHECK: A program to check the
stereochemical quality of protein structures. Journal of
Applied Crystallography, 26, 283–291.
Lewell, X. Q., Judd, D. B., Watson, S. P., & Hann, M. M.
(1998). RECAP–retrosynthetic combinatorial analysis pro-
cedure: A powerful new technique for identifying privi-
leged molecular fragments with useful applications in
combinatorial chemistry. Journal of Chemical Information
and Computer Sciences, 38, 511–522.
Lipinski, C. A., Lombardo, F., Dominy, B. W., & Feeney, P. J.
(2001). Experimental and computational approaches to esti-
mate solubility and permeability in drug discovery and
 14 R.S. Mandal and S. Das
development settings. Advanced Drug Delivery Reviews,
46, 3–26.
Lüthy, R., Bowie, J. U., & Eisenberg, D. (1992). Assessment
of protein models with three-dimensional profiles. Nature,
356, 83–85.
Maiorov, V. N., & Crippen, G. M. (1994). Significance of root-
mean-square deviation in comparing three-dimensional
structures of globular proteins. Journal of Molecular
Biology, 235, 625–634.
Makarova, K. S., & Grishin, N. V. (1999). The Znpeptidase
superfamily: Functional convergence after evolutionary
divergence. Journal of Molecular Biology, 292, 11–17.
Milik, M., Szalma, S., & Olszewski, K. A. (2003). Common
structural cliques: A tool for protein structure and function
analysis. Protein Engineering, 16, 543–552.
Miyamoto, S., & Kollman, P. A. (1992). Settle: An analytical
version of the SHAKE and RATTLE algorithm for rigid
water models. Journal of Computational Chemistry, 13,
952–962. doi:10.1002/jcc.540130805
Nocek, B. P., Gillner, D. M., Fan, Y., Holz, R. C., & Joachi-
Downloaded by [University of York] at 02:07 14 November 2014
miak, A. (2010). Structural basis for catalysis by the mono-
and dimetalated forms of the dapE-encoded N-succinyl-L,
L-diaminopimelic acid desuccinylase. Journal of Molecular
Biology, 397, 617–626. doi:10.1016/j.jmb.2010.01.062
Nomura, A., Stemmermann, G. N., Chyou, P. H., Kato, I.,
Perez-Perez, G. I., & Blaser, M. J. (1991). Helicobacter
pylori infection and gastric carcinoma among Japanese
Americans in Hawaii. New England Journal of Medicine,
325, 1132–1136.
Paiva, A. M., Vanderwall, D. E., Blanchard, J. S., Kozarich, J.
W., Williamson, J. M., & Kelly, T. M. (2001). Inhibitors of
dihydrodipicolinate reductase, a key enzyme of the diami-
nopimelate pathway of Mycobacterium tuberculosis.
Biochim Biophys Acta, 1545, 67–77.
Parsonnet, J., Friedman, G. D., Vandersteen, D. P., Chang, Y.,
Vogelman, J. H., Orentreich, N., & Sibley, R. K. (1991).
Helicobacter pylori infection and the risk of gastric
carcinoma. New England Journal of Medicine, 325, 1127–
1131.
Sali, A., & Blundell, T. L. (1993). Comparative protein model-
ling by satisfaction of spatial restraints. Journal of Molecu-
lar Biology, 234, 779–815.
Scapin, G., Reddy, S. G., Zheng, R., & Blanchard, J. S.
(1997). Three-dimensional structure of Escherichia coli di-
hydrodipicolinate reductase in complex with NADH and
the inhibitor 2,6-pyridinedicarboxylate. Biochemistry, 36,
15081–15088.
Shen, M., & Sali, A. (2006). Statistical potential for assessment
and prediction of protein structures. Protein Science, 15,
2507–2524.
Talley, N. J., Zinsmeister, A. R., Weaver, A., DiMagno, E. P.,
Carpenter, H. A., Perez-Perez, G. I., & Blaser, M. J.
(1991). Gastric Adenocarcinoma and Helicobacter pylori
infection. Journal of the National Cancer Institute, 83,
1734–1739.
Uda, N. R., & Creus, M. (2011). Selectivity of inhibition of n-
succinyl-l, l-diaminopimelic acid desuccinylase in bacteria:
The product of dapE-gene is not the target of l-captopril
antimicrobial activity. Bioinorganic Chemistry and Applica-
tions, 2011, 306465. doi:10.1155/2011/306465
Van Der Spoel, D., Lindahl, E., Hess, B., Groenhof, G., Mark,
A. E., & Berendsen, H. J. C. (2005). GROMACS: Fast,
flexible, and free. Journal of Computational Chemistry, 26,
1701–1718. doi:10.1002/jcc.20291
Veber, D. F., Johnson, S. R., Cheng, H. Y., Smith, B. R., Ward,
K. W., & Kopple, K. D. (2002). Molecular properties that
influence the oral bioavailability of drug candidates. Jour-
nal of Medicinal Chemistry, 45, 2615–2623.
Wheeler, D. L., Barret, T., Benson, D. A., Bryant, S. H.,
Canese, K., Chetvernin, V., … Yashenko, E. (2008). Data-
base resources of the National Center for Biotechnology
Information. Nucleic Acids Research, 36, D13–D21.