Content uploaded by Soumya Ranjan Mahapatra
Author content
All content in this area was uploaded by Soumya Ranjan Mahapatra on May 08, 2022
Content may be subject to copyright.
The potential of plant-derived secondary metabolites as novel drug
candidate against Klebsiella pneumoniae: Molecular docking and
simulation investigation
Soumya Ranjan Mahapatra
a
, Jyotirmayee Dey
a
, T.Kiran Raj
b
, Vijay Kumar
c
, Mrinmoy Ghosh
d
,
Krishn Kumar Verma
d
, Taranjeet Kaur
e
, Mahipal Singh Kesawat
f
, Namrata Misra
a,d,
*,
Mrutyunjay Suar
a,d,
*
a
KIIT School of Biotechnology, Kalinga Institute of Industrial Technology (KIIT), Deemed to be University, Campus-11, Patia, Bhubaneswar, Odisha 751024, India
b
Department of Biotechnology & Bioinformatics, School of Life Sciences, JSS Academy of Higher Education & Research, Mysuru, India
c
Department of Biotechnology, Lovely Faculty of Technology and Sciences, Lovely Professional University, Phagwara, Punjab 144402, India
d
KIIT-Technology Business Incubator (KIIT-TBI), Kalinga Institute of Industrial Technology (KIIT), Deemed to be University, Bhubaneswar, Odisha 751024, India
e
Biotechnology Industry Research Assistance Council (BIRAC), India
f
Department of Genetics and Plant Breeding, Sri University, Cuttack, Odisha 754006, India
ARTICLE INFO
Article History:
Received 1 December 2021
Revised 21 April 2022
Accepted 22 April 2022
Available online xxx
Edited by: Dr V. Kumar
ABSTRACT
Klebsiella pneumoniae is an important multidrug-resistant pathogen affecting humans and a major source for
hospital infections associated with high morbidity and mortality due to limited treatment options. The path-
ogenesis of this microbe is caused by several key proteins; one of them is type 3 fimbrial proteins, which has
been known for its crucial role in the host cell invasion. Therefore, targeting fimbriae protein can be a solu-
tion for treating this pathogenic disease. Plant Secondary metabolites with diverse chemical scaffolds have
been recognized as an invaluable source of compounds in drug discovery and development. However, sys-
tematic identification of drug targets for plant secondary metabolites at the human proteome level via vari-
ous experimental assays is highly expensive and time-consuming. In this study, we have employed plant
derived secondary metabolites to predict new drug targets of type 3 fimbrial protein of K. pneumoniae. Subse-
quently, structure-based virtual screening was performed to identify compounds showing the best binding
confirmation with the target enzyme and forming a stable complex. Molecular dynamics simulations of the
protein-ligand complex indicated that the intermolecular hydrogen bonds formed between the protein and
ligand complex remain stable during the simulation time. The interaction shown by the compounds provides
an important insight into the mechanism involved in the ligand-receptor interaction. The model will aid in
the development of a drug with improved efficacy and reduced side effects. Thus, vernolide could serve as a
drug for treating pneumonia infections in humans. The drug likeliness prediction on this derivative also sup-
ports its suitability as a drug candidate. However, an in vitro and in vivo analysis of the selected compound is
necessary for further validation before administration of the drug to human beings.
© 2022 Published by Elsevier B.V. on behalf of SAAB.
Keywords:
Klebsiella pneumoniae
Fimbrial protein
Plant secondary metabolites
Drug design
Virtual screening
1. Introduction
Klebsiella pneumoniae is one of these clinically important patho-
gens that have attracted public concern (Martin and Bachman, 2018;
Peterson and Kaur, 2018). K. pneumoniae is a prominent Enterobac-
teriaceae pathogen that causes a wide variety of ailments
(Mahapatra et al., 2021;Effah et al., 2020). Due to the
complex pathogenicity of the diseases, present vaccines and treat-
ments available for Klebsiella infections are currently unsuccessful
(Opoku-Temeng et al., 2019). Furthermore, these therapies were
observed to have adverse effects on the host immune response
both during its activation and manifestation. K. pneumoniae, on the
other hand, has been found to develop drug resistance to a variety of
antibiotics (i.e., beta-lactam antibiotics, aminoglycosides, and fluoro-
quinolones) through a variety of genetic mechanisms (Ferreira et al.,
2019). It is evident that the currently available antibiotics are insuffi-
cient to control these antibiotic-resistant bacteria. Therefore, an alter-
native safe and effective antibacterial medication is urgently needed
(Cassir et al., 2014).
* Corresponding authors at: KIIT School of Biotechnology, Kalinga Institute of Indus-
trial Technology (KIIT), Deemed to be University, Campus-11, Patia, Bhubaneswar,
Odisha 751024, India.
E-mail addresses: namrata@kiitincubator.in (N. Misra), mrutyunjay@kiitincubator.
in,msuar@kiitbiotech.ac.in (M. Suar).
https://doi.org/10.1016/j.sajb.2022.04.043
0254-6299/© 2022 Published by Elsevier B.V. on behalf of SAAB.
ARTICLE IN PRESS
JID: SAJB [m5G;May 3, 2022;22:43]
Please cite this article as: S.R. Mahapatra, J. Dey, T.K. Raj et al., The potential of plant-derived secondary metabolites as novel drug candidate
against Klebsiella pneumoniae: Molecular docking and simulation investigation, South African Journal of Botany (2022), https://doi.org/
10.1016/j.sajb.2022.04.043
South African Journal of Botany 000 (2022) 19
Contents lists available at ScienceDirect
South African Journal of Botany
journal homepage: www.elsevier.com/locate/sajb
Due to fimbrial adhesins and a thick capsule that acts as a putative
antiphagocytic factor, K. pneumoniae is intrinsically virulent and
capable of invasive infection (Pendleton et al., 2013). Type 1 and type
3fimbriae are the most commonly encoded fimbriae types in K. pneu-
moniae (Stahlhut et al., 2012b). The majority of immunological treat-
ments against K. pneumoniae fimbriae have targeted type 3 fimbriae,
also known as MrkA, which is a promising target for therapies
(Choi et al., 2019). The mrkA encodes the fimbrial subunit and a
major protein of the type III fimbriae complex on the surface of K.
pneumoniae, which is involved in host cell adhesion and biofilm
development (Murphy et al., 2012). In vitro adhesion to epithelial
cells, as well as kidney and lung tissues, is mediated by Type 3 fim-
briae, most likely in a mannose-resistant way (Stahlhut et al., 2012a).
K. pneumoniae biofilm-associated urinary infections are also caused
by type 3 fimbriae protein (Li et al., 2014). K. pneumoniae fimbriae
proteins have been shown to be excellent protein transporters and
immunogens, as well as being extensively conserved among Entero-
bacteriaceae family (Nezafat et al., 2016).
Plant-based phytocompounds play an essential role in treating
infectious diseases in order to overcome any shortcomings
(Benarba and Pandiella, 2020). Plants are the primary source of natu-
ral products that can be explored because of abundant metabolite
content and common pathways which can be easily manipulated
(Atanasov et al., 2021). Because of their cosmopolitan nature, these
plants can be easily cultivated throughout the world. Although a
good percentage of all pharmaceutical products present in the world
are derived from plants, very few are used as antimicrobials
(Gonelimali et al., 2018). Plant-derived natural compounds are
attracting more attention in the present age because of their potential
efficacy and lack of adverse effects. Plants are a rich source of second-
ary metabolites such as quinones, tannins, terpenoids, alkaloids, fla-
vonoids, and polyphenols. These compounds are synthesized in
response to pathogen attacks and are required to protect the plant
from pathogens, which might also play a role in regulatory function,
which in turn indirectly increases the level of resistance of the plant
(Bhuiyan et al., 2020). Plant extracts have the ability to increase the
efficacy of antibiotics. The synergistic interactions between plant
extracts and antibiotics have been reported to be caused by drug
efflux inhibition and alternate modes of action (Khameneh et al.,
2019). Plants that are reported to be of use in ethnomedicine and
also a large number of phytocompounds whose properties are so far
unexplored can be targeted in drug discovery using the bioinformat-
ics tools and softwares (Biswas et al., 2021).
While traditional methods of drug discovery could take years,
computational techniques aid in accelerating the drug discovery
process. A computational method is a reliable approach to identify active
phytocompounds from various databases. This strategy is currently
employed in drug discovery research (Thomford et al., 2018). Recently,
Paiva et al., have reported on the promising phytocompounds identified
from different plant sources through the computational approaches for
the treatment of bacterial diseases (Paiva et al., 2010). Nowadays, with
the continuous advancement of computer science, successful examples
of finding drugs from natural products using computer-aided drug
designmethodshavebecomemorefrequent,suchasDozamide
(Approved by FDA in 1995), Imatinib (Approved by FDA in 2001), Dasati-
nib (Approved by FDA in 2006), and Ponatinib (Approved by FDA in
2012) (Yi et al., 2018). With the continuous maturation of computer
technology, the in silico approach of utilizing a computer platform to cal-
culate the combinations of simulated compounds and targets have
become increasingly accurate (Katsila et al., 2016).
In the drug discovery process with the help of a computational
approach, the selection of structural proteins is one of the major tasks
(Lin et al., 2020). Among all the virulence factors of K. pneumoniae,fim-
brial adhesin proteins can be utilized to develop the drug as the other
virulence factors are mostly carbohydrates (Paczosa and Mecsas, 2016).
The adhesion of K. pneumoniae to mammalian tissue is conciliated by
two varieties of bacterial pili, type 1 and type 3 (Khater et al., 2015).
Type 1 fimbriae play an important role in bacterial adhesion to the D-
mannose moiety on mammalian cell surfaces due to the specificaffinity
of fimbrial protein at the tips of fimbriae (Struve et al., 2009). Type 3 fim-
briae are characterized by their affinities for a range of mammalian cells,
which include bladder epithelial cells, uroepithelial cells, and endothelial
cells (Tarkkanen et al., 1997).
In this study, we identified novel phytocompounds by employing
structure-based drug designing and recommended their role in Klebsi-
ella pathogenesis. We computationally screened six plant secondary
metabolites extracted from the PubChem database against the mrkA
protein using the PyRx, based on a recent research paper. A structure-
based pharmacophore was developed and validated based on the pro-
tein-ligand interaction of phytocompounds and the target complex. For
the post-docking investigations, we used Discovery Studio Visualizer.
The best two compounds from the virtual screening were chosen for
ADMET prediction. In the ADMET various parameters like BBB, HIA,
Caco-2, Pgpsubstrate/inhibitor, Cytochrome P450-substrate/inhibitor,
toxicity,carcinogenicity,andlethaldose,etc.werepredicted.TheADMET
analysis revealed that one compound performed well with these phar-
macokinetic parameters. Following that, Molecular Dynamic (MD) simu-
lation was used to confirm the screened putative inhibitor affinity for
the docked site and to better understand the dynamic behavior of the
complex. The novel inhibitor showed a stable root mean square devia-
tion (RMSD), stable binding orientation, and hydrogen bond interactions
with the catalytic residues of the target protein. Therefore, this study rec-
ommends vernolide as the novel inhibitor of type 3 fimbrial protein of K.
pneumoniae. Our in silico results support the prediction that the pre-
dicted compound can efficiently suppress the pneumonia infection and
block bacterial virulence. However, before administering the drug to
humans, in vitro and in vivo investigations of the identified compound is
required for additional validation. The overall workflow has been dem-
onstrated in Fig. 1.
2. Materials and methods
2.1. Protein preparation
The Uniprot database was searched for the required retrieval of the
amino acid sequence of the type 3 fimbriae protein of Klebsiella pneumo-
niae. The tertiary structure of the protein was determined using the
Robetta server (Kim et al., 2004). The Pymol visualization tool was uti-
lized to visualize the 3D structure of the target protein. The tertiary struc-
ture generated by Robetta was submitted to ‘GalaxyRefine2’(http://
galaxy.seoklab.org/cgi-bin/submit.cgi?type=REFINE2) available at Galax-
yWEB server (http://galaxy.seoklab.org/index.html)fortherefinement
of protein structure obtained from the top model generated by Robetta.
GalaxyRefine2 is a more advanced version of GalaxyRefine that employs
local and global operators, as well as local error estimates and homology
structure information, to increase the accuracy of the input protein's
structure (Lee et al., 2019). It generates 10 refined models and provides
information related to RMSD, MolProbity, Clash score, Poor rotamers,
Rama favored, and ‘GALAXY energy’for all the ten models in comparison
with the user-submitted model. Ramachandran plot embedded in PRO-
CHECK tool (Laskowski et al., 1993) was used to evaluate the phi/psi dis-
tribution of each amino acid in the refined model and to quantify amino
acids into favorable and non-favorable zones. Verify-3D (Eisenberg et al.,
1997) were employed to estimate the compatibility of an atomic model
(3D) with its own amino acid sequence (1D). The ProSA web tool, which
uses a Z score (Wiederstein and Sippl, 2007) was used to evaluate the
overall model quality of the modeled structures.
2.2. Active site identification
Active site determination was done for the modeled protein to
further work on its docking studies. Active site determination gives
ARTICLE IN PRESS
JID: SAJB [m5G;May 3, 2022;22:43]
S.R. Mahapatra, J. Dey, T.K. Raj et al. South African Journal of Botany 00 (2022) 19
2
us an idea to make a grid before docking. Theoretical active site pre-
diction of the target protein receptor was predicted using the CASTp
server. CASTp (Computed Atlas of Surface Topography) gives an
online asset to finding, portraying, and estimating sunken surface
areas on three-dimensional structures of proteins (Tian et al., 2018).
These incorporate pockets situated on protein surfaces and voids cov-
ered in the inside of proteins. The estimation incorporates the terri-
tory and volume of pocket or void by the dissolvable available
surface model (Richards’surface) and by the atomic surface model
(Connolly’s surface), all determined systematically. CASTp can be uti-
lized to examine surface highlights and utilitarian districts of pro-
teins. CASTp incorporates a graphical UI, adaptable intelligent
perception, just as on the flyfiguring for client transferred structures.
Here we input the modeled protein for predicting the ligand-binding
sites and the CASTp server predicts the amino acids crucial for bind-
ing interactions.
2.3. Ligand preparation
The chemical structures of the selected compounds were obtained
from PubChem an online repository of chemical compounds (https://
pubchem.ncbi.nlm.nih.gov/). We managed to retrieve a total of 6phy-
tocompounds from the library in 3D SDF format. Energy minimiza-
tion of the ligands was carried out by enforcing the uff force-field and
Conjugate Gradients algorithm of PyRx.
2.4. Virtual screening and molecular docking
In structural biology, molecular docking is renowned as a reliable
technique, especially in CADD processes. The technique ensures the
best prediction of binding mode between a small molecule and a spe-
cific macromolecule. The small molecules of six secondary plant
metabolites (in Single Data File (SDF) format) were obtained from the
PubChem database with the particular subsets file and virtually
screened using PyRx 0.8 tool. PyRx is an open-access virtual screening
tool dedicated to the screening of a vast range of libraries of com-
pounds against a particular drug target. AutoDock4 and AutoDock
Vina are the two most effective tools under the banner of PyRx which
offer a user-friendly interface to carry out molecular docking for
CADD. To figure out the best binding interaction of our desired
protein and the hit compounds, we used the AutoDock wizard of the
PyRx tool for molecular docking. Initially, the library of natural mole-
cules in SDF format along with three known inhibitors was prepared
using PyRx 0.8 tool. This includes (a) addition of Gasteiger charges,
(ii) adding polar hydrogens, (iii) merging non-polar hydrogens, and
(iv) finally saving into PDBQT format (XYZ coordinates ÞPartial
charges ÞAtom type).
In addition to this, the optimized model of type 3 fimbriae from K.
pneumoniae was also prepared by using this tool which includes (a)
addition of Kollmann charges, (2) merging non-polar hydrogens, (iii)
adding polar hydrogens, and finally saved into PDBQT format. The
AutoDock tool is located in the PyRx 0.8 interface option, which is
used to perform molecular docking calculation to investigate the
interaction between protein and ligand molecule based on the esti-
mated free energy of binding (DG), estimated inhibition constant
(Ki), number, and type of intermolecular interactions. In this virtual
screening, the site-specific or direct docking calculation was per-
formed using AutoDock based on the predicted binding site of type 3
fimbriae for all the ligands. All the default configurations and param-
eters of PyRx were retained during the docking process based on the
highest binding energy (kcal/mol) for evaluation. In the docking cal-
culation, the grid box parameter was set to X= 24.21, Y= 17.02,
Z=25.04. We used the BIOVIA Discovery Studio visualizer tool to
observe the binding pose of the protein-ligand complex.
2.5. Prediction of drug-likeness and ADME properties
Early-stage preclinical evaluation can save both time and money
with an increased positive outcome, needed for successful drug dis-
coveries. Accordingly, to characterize our designed derivatives for
drug-likeness parameters, we performed in silico prediction calcula-
tions. Drug kinetics and exposure of tissues to drug influences the
pharmacological activity and the performance of a drug candidate,
which is eventually determined by its ADME properties. The ultimate
objective of the in silico ADME studies is to accurately predict in vivo
pharmacokinetics of a possible and potential drug molecule. The
ADME (Absorption, distribution, metabolism, and excretion) profiling
of compounds was predicted using an online server, SwissADME
(http://www.swissadme.ch). The freely accessible web server helps
to predict the physicochemical, and pharmacokinetics, of the selected
Fig. 1. The schematic representation of the in silico drug design of the type 3 fimbrial protein of K. pneumoniae.
ARTICLE IN PRESS
JID: SAJB [m5G;May 3, 2022;22:43]
S.R. Mahapatra, J. Dey, T.K. Raj et al. South African Journal of Botany 00 (2022) 19
3
drug candidates. The compound set for testing contained different
compounds along with our designed derivatives.
2.6. Toxicity analysis
Toxicological determination is the most prime considerations in
the case of the development of new drugs. The toxicity profile of
drug candidates gives an idea about the health and environmental
risks and safety/toxicity of a chemical’s substances. Nowadays, com-
puter-aided in-silico toxicity testing is playing an important role in
the assessment of compounds toxicity more accurately without using
experimental animal models. Therefore, to evaluate the early-stage
toxicity of the selected drug candidates ProTox-II (http://tox.charite.
de/protox_II) web server has been used, which helps to determine
acute toxicity, hepatotoxicity, cytotoxicity, carcinogenicity, mutage-
nicity, and immunotoxicity of the selected compounds.
2.7. Molecular dynamics simulation analysis
Gromacs Version 2019.4 with the Gromos force field and default
cubic box parameters was used to run a 100 ns MD simulation of the
confined form of target protein and secondary metabolite complex.
The following processes are involved in MD simulation of the docked
complex: (i) generation of GRO formated files from input PDB format,
topology (or force field parameter), and positional restraint files using
PDB2GMX, (ii) Building a cubic box around the complex with EDIT-
CONF, (iii) Solving the docked complex protein with SOLVATE, (iv)
Energy minimization with the steepest descent approach, (v) Equili-
bration in two phases (NVT (Constant Number Particles, Volume, and
Temperature) and NPT (Constant Number Particles, Pressure, and
Temperature)), and (vi) finally producing MD for 100 ns. The systems
are energy reduced and equilibrated for 500 PS in NVT and 500 PS in
NPT ensembles. In the instance of ligand simulations, the PRODRG
software created the topology and GRO files for the specified ligand.
For protein-ligand simulations, the input parameters of explicit SPC
water molecules and counterions (to neutralize the system) were left
at its default values. After the MD simulations for the bounded form
of the target protein are completed, several Gromacs built-in func-
tions are utilised to analyze various structural parameters such as
RMSD and RMSF.
2.8. Binding free energies of complex
For analysing the systems after simulations, the system trajecto-
ries were concatenated and recentred within the periodic box by
using GROMACS tools. Molecular Mechanics Poisson-Boltzmann Sur-
face Area (MMPBSA) was used to compute the binding free energy
for top docked complex using g_mmpbsa tool (Kumari et al., 2014).
In MM-PBSA approach, calculation of the binding free energy
(DG
bind
) between a protein and a ligand are often performed as:
DGbind ¼DHTDS¼DEMM þDGsol TDSð1Þ
EMM þEbonded þEnonbonded ¼Ebonded þEvdw þEeOrðÞDEMM
¼DEinternal þDEelectrostatic DEvdW ð2Þ
DGsolv ¼DGel
solv þDGvdW
solv ð3Þ
Where, the total gas phase energy (sum of the
DE
internal
+DE
electrostatic
+DE
vdW
) on the binding of MM energy is
shown as DEMM, the free energy of solvation as DGsolv and the
entropy contribution as TDS. Electrostatic solvation energy term is
computed in a continuum solvent using Poisson-Boltzmann model. If
the binding energy value is negative, it favours complex formation in
water, positive value denotes unfavorable binding.
3. Result and discussion
3.1. Overall structure assessment of mrkA protein
The most basic stage in rational drug discovery is to get a 3D
structure of the target protein. The 3D structure has been utilized to
decipher the structural features and molecular function of the target
enzyme, as well as to find the strong inhibitors (Kopec et al., 2005).
Since the 3D structure of mrkA is not yet available in the Research
Collaboratory Structural Bioinformatics - Protein Data Bank
(RCSB-PDB), the amino acid sequence of K. pneumoniae type 3 fim-
brial protein was obtained from the UniProt database (https://www.
uniprot.org) and modeled using the Robetta web server. During
energy minimization, we have utilized Swiss PDB Viewer, a well-
known and efficient minimization server. For further investigation,
the energy-minimized output structure was chosen. Finally, the
estimated tertiary structure was refined using the GalaxyWEB
structure refinement server.
Ramachandran plot analysis was used to evaluate the predictabil-
ity of the structure. Following structure refinement, the majority of
the residues (92.1%) of the structure were placed in the most favored
region, with just 0.5% of the residues lying in the disallowed region,
indicating high structure quality (Fig. 2). The stereochemical charac-
teristics of the model were also confirmed via the Verify3D web
server. Verify3D examined the model's local environment and inter-
residue interactions. The model's Verify 3D score was estimated to be
90.84%, with a Z-Score of 6.04 from the ProSA server (Fig. 2).
3.2. Binding site identification and receptor grid generation
A binding site is a particular amino acid (AA) residue in a protein
to which ligands may bind and is crucial for directing drug design.
Identifying the location of protein binding sites is critical during
molecular docking simulation, as it aids in the generation of adequate
contact points with the protein and considerably improves docking
effectiveness (Meng et al., 2011). Binding sites develop to be adapted
to bind a certain substrate; hence, the protein's binding site has been
identified in this study. The CASTp server predicted V12, S15, S16,
A19, H20, S81, Q82, K84, A85, R88, T180, N181, V182, G234, and P235
as the active site amino acids.
3.3. Virtual screening
Protein-ligand docking is critical for accurately anticipating the
orientation of a ligand with its target protein (Morris and Lim-
Wilby, 2008). Understanding the possible interactions between a pro-
tein-ligand complex is an essential element of the drug development
process, as it aids in the identification of hits to leads as prospective
drug candidates. The modeled structure of the fimbrial type 3 protein
was employed for the docking investigation. The six secondary plant
metabolite compounds were extracted from the PubChem database
for docking with the target protein (Table 1). PyRx was used to do vir-
tual screening on the modeled structure of the mrkA protein and
ligands. Following the virtual screening docking study, we only
examined compounds with the highest binding energy for the best
results. As a consequence, we chose the best two compounds from
the results. Our results reveal that Glucosinolates (PubChem CID:
6,537,198) and Vernolide (PubChem CID: 5,281,508) had the highest
docking scores of 6.9 and 6.6 Kcal/mol, respectively.
3.4. ADME properties of selected compounds
The pharmacological activity and performance of a drug candidate
are affected by drug kinetics and tissue exposure, which are ulti-
mately determined by its ADME characteristics. The ultimate goal of
in silico ADME research is to properly anticipate the in vivo
ARTICLE IN PRESS
JID: SAJB [m5G;May 3, 2022;22:43]
S.R. Mahapatra, J. Dey, T.K. Raj et al. South African Journal of Botany 00 (2022) 19
4
pharmacokinetics of a prospective pharmacological molecule
(Paul Gleeson et al., 2011;Daina et al., 2017). Early-stage ADME
screening has been shown to minimize attrition rates throughout the
clinical drug development phase. The majority of medicines fail in
clinical trials due to poor pharmacokinetic characteristics. As a result,
the SwissADME web tool was used in the study's early-stage evalua-
tion of ADME properties for two compounds. Based on lipophilicity,
solubility, pharmacokinetics, medicinal chemistry, and drug-likeness
features, the server evaluated the ADME qualities of two compounds
(PubChem CID: 6,537,198, 5,281,508) (Table 2).
The molecule vernolide has maintained an optimal pharmacoki-
netics property among the two selected compounds based on binding
affinity, while the compound Glucosinolates has a negative Log Po/w
value and violated the maximum Lipinski's rule of five (RO5) men-
tioned in Table 2. As a result of the ADMET study, the compound Glu-
cosinolates was not identified to be acceptable drug candidate
(Table 2). SwissADME provides a drug-likeness graph in the form of a
radar chart (Fig. 3), with each peak representing a parameter defining
pharmacological characteristics. Within the chart, the pink area
reflects the ideal range for each characteristic. Any compound in the
pink zone might be evaluated as a possible contender. The final mole-
cule chosen to float inside the pink zone indicates its potency as a
drug (Fig. 3).
3.5. Toxicity predictions
Toxicology analysis is a vital and one of the most significant pro-
cesses in drug development since it aids in identifying the negative
effects of chemical compounds on humans, animals, plants, or the
environment. Traditional toxicity testing of chemicals necessitates
the use of an in vivo animal model, which is time-consuming, costly,
and prone to ethical concerns (Banerjee et al., 2018). As a result, com-
puter-aided in silico toxicity evaluation of chemical compounds can
be valuable in the drug design process. The ProTox-II webserver was
used in the study to determine the toxicity of the drug computation-
ally, since it is quick, cheap, and does not involve any ethical
concerns. The final compound vernolide (PubChem CID: 5,281,508)
was uploaded to the ProTox-II online site, which calculates the acute
toxicity, hepatotoxicity, cytotoxicity, carcinogenicity, and mutagenic-
ity of the compounds mentioned in Table 3. There was no evidence of
oral toxicity or organ toxicity with the compound chosen
(Pratheeshkumar and Kuttan, 2012). According to the estimated
acute oral toxicity levels, the chemical belongs to ''class IV.'' The com-
pounds in this class have LD50 values less than 5000 mg/kg and are
typically thought to be druggable. As a result, the final compound
chosen is powerful and has the potential to be employed as a promis-
ing drug with high oral bioavailability and safety.
3.6. Molecular docking
The approach of molecular docking predicts the optimal orienta-
tion of a ligand when gets bound in an active site to produce a stable
complex. All the ligand molecules were docked with in the generated
grid box around the active site of the protein. Binding affinities and
interaction pattern of the molecules are summarized in Table 1.Itis
clear from the table that the binding affinity of the compounds varied
from 5.4 to 6.9 kcal mol. The interaction between the final chosen
compound and the desired mrkA protein was studied using the BIO-
VIA Discovery Studio Visualizer tools after the ADME and toxicity
parameters were determined. The hydrogen bonding interactions
were represented by green dash lines. The amino acids residue His20,
Ala19, Ser15, and Ser81 engaged with protein ligand interactions are
represented in the Fig. 4. The surface view of the structure and com-
pound are shown in Fig. 5. Analysis of the complex structure revealed
several bonding interactions, including hydrogen bonds (conven-
tional H-B) and hydrophobic interactions (Alkyl, Pi-Alkyl) between
the protein and ligand, as shown in Fig. 4. Vernolide complex struc-
ture analysis revealed four bonding interactions, including two
hydrogen bonds (two conventional H-B) and two hydrophobic (one
Pi-Alkyl, one Alkyl) bonds to the protein's different binding site resi-
dues, with bond distances ranging from 2.03 A
to 5.22 A
, as shown in
Fig. 4. It is worth mentioning that this compound demonstrated
Fig. 2. Validation of the target protein 3D model results confirmed the model to be reliable and accurate. PROSA 3D structure validation showing Z-score (6.04). Ramachandran
analysis showing 92.1% of residues in most favored region.
ARTICLE IN PRESS
JID: SAJB [m5G;May 3, 2022;22:43]
5
S.R. Mahapatra, J. Dey, T.K. Raj et al. South African Journal of Botany 00 (2022) 19
Table 1
Chemical structures, plant sources, docking scores, and two-dimensional (2D) structure of six secondary plant metabolites.
PubChem CID Compound Name Formula Chemical Structure Plant Source Binding Affinity (kcal/mol)
14,282,633 Helihumulone C26H32O5 Helichrysum cymosum 5.9
72,388 Anolignan B C18H18O2 Terminalia sericea 5.4
139,587,894 Sesquiterpenoid C15H26O3 Warburgia salutaris 6.5
5,281,508 Vernolide C19H22O7 Vernonia colorata 6.6
442,318 vernodalol C20H24O8 Vernonia colorata 6.1
6,537,198 Glucosinolates C16H20N2O9S2 Aurinia sinuata 6.9
ARTICLE IN PRESS
JID: SAJB [m5G;May 3, 2022;22:43]
6
S.R. Mahapatra, J. Dey, T.K. Raj et al. South African Journal of Botany 00 (2022) 19
excellent Log Po/w value in ADME analysis. Further to investigate the
conformational flexibilities of drug-receptor complex MD simulation
analysis of 100 ns was performed. Conformational changes were
observed in the drug-receptor complex based on the RMSD, RMSF
and binding free energy analysis.
3.7. MD simulation
MD simulations were carried out to assess the stability of all sys-
tems, analyzing the Root-Mean Square Deviation (RMSD), and Root-
Mean-Square Fluctuations (RMSF). In our study, the protein model
has shown good accuracy of the structure as revealed by the virtual
screening. It has been earlier described that the virtual screened
model should have a stable molecular dynamic behavior. We evalu-
ated the docked complex of vernolide and target protein using the
molecular dynamic stability criteria by performing the molecular
dynamic simulation study using GROMACS 2019.4 software package.
In our case, molecular dynamics simulation study revealed a consis-
tent value of the energy of the molecule throughout the time period
of simulation, which is an indication of the strong basis of the fact
that the ligand-receptor complex has a stable structure as required
for the drug designing processes. Root mean square deviation
(RMSD) was evaluated during the simulation, and it was observed
that it was increasing at a high rate in the beginning, but after 30 ns it
showed a constant growth for the rest of the duration of the simula-
tion, which suggests that the ligand-receptor complex has lesser
RMSD for the complex backbone and has less flexibility, indicating
the stable dynamic behavior structure of the secondary metabolite
vernolide (Fig. 6). It was observed that the fluctuations in the root
mean square (RMS) were very low and most of the atoms were free
from the RMS fluctuations (Fig. 6). Furthermore, there were only a
few atoms, which showed RMS fluctuation at C and N terminals due
to the loop region, which strongly suggests that the secondary
metabolite vernolide has an accurate and stable structure.
3.8. MM-PBSA binding free energy analysis
The free energy of binding (DG
bind
) is considered to be a key ther-
modynamic quantity for assessing the favourable protein-ligand
interaction as well as their affinity for accurate biological system
modeling. The g_mmpbsa tool was used to compute the free binding
energy of the MD simulation in this study. The MM-PBSA compo-
nents such as Coulomb energy, van der Waals, Polar solvation, and
SASA energies were evaluated, found that these energies affect the
overall binding energies. The MM/PBSA computed free energy of
binding for the system was estimated as 66.663 §29.026 kJ/mol
(Table 4). The results demonstrated that the docking was energeti-
cally possible, as evidenced by negative values of Gibbs free energy
(DG). The observed negative free binding energy value shows that
the drug complex firmly binds to the receptor protein, making it a
promising candidate as a drug therapeutic against K. pneumoniae.
4. Conclusion
To our knowledge, this is the first study to use compressive in sil-
ico techniques to identify promising natural antibacterial drug candi-
dates against K. pneumoniae. This study evaluates the antibacterial
properties of plants against K. pneumoniae by deciphering the inter-
action of existing secondary metabolites activity using an in silico
technique for target identification and molecular docking analysis.
Vernolide was identified as a possible therapeutic candidate using an
integrated structure-based pharmacophore modeling, virtual screen-
ing, molecular docking, ADMET, toxicity, and MD simulation strategy.
Fig. 3. The final selected compound (PubChem CID: 5,281,508) is within colored zone that indicates suitable physio-chemical space for oral bioavailability and show the POLAR
(Polarity), INSOLU (Insolubility), INSATU (Instauration), and FLEX (Rotable bond flexibility), LIPO (Lipophilicity), SIZE (Molecular Weight), parameters.
Table 2
List of pharmacokinetics (ADME) properties includes physicochemical properties,
lipophilicity, water-solubility, drug-likeness, and medicinal chemistry of selected
two compounds.
Properties PubChem
CID5281508
PubChem
CID6537198
Physicochemical
Properties
MW (g/mol) 362.37 448.47
Heavy atoms 26 29
Aro. atoms 0 9
Rotatable bonds 3 7
H-bond acceptors 7 10
H-bond donors 1 6
TPSA (A
2) 94.59 215.58
Lipophilicity Log Po/w (Cons) 1.52 0.45
Water Solubility Log S (ESOL) Soluble Soluble
Pharmacokinetics GI absorption High Low
BBB permeant No No
P-GP substrate No No
Drug likeness Lipinski violations 0 2
Medi. Chemistry Synth. accessibility Medium Medium
Table 3
List of compounds toxicity endpoints includes acute toxicity, hepato-
toxicity, cytotoxicity, carcinogenicity, and mutagenicity of selected
one compound.
Classification Target PubChem CID5281508
Oral toxicity LD50 (mg/kg) 1190
Toxicity Class 4
Organ toxicity Hepatotoxicity Inactive
Toxicity endpoints Carcinogenicity Inactive
Mutagenicity Inactive
Cytotoxicity Inactive
ARTICLE IN PRESS
JID: SAJB [m5G;May 3, 2022;22:43]
7
S.R. Mahapatra, J. Dey, T.K. Raj et al. South African Journal of Botany 00 (2022) 19
Fig. 4. Protein-ligand interaction profile of the dock complex. Where (A) represents the interaction between type 3 fimbrial protein and compound (PubChem CID: 5,281,508)
whereas (B) shows non bonded interaction between them.
Fig. 5. Surface representation of the type 3 fimbrial protein and the compound vernolide. The target protein is shown in surface, and vernolide is shown in stick.
Fig. 6. Depicted the RMSD values extracted from the Caatoms of the selected compound in complex with the type 3 fimbrial protein. Showing the RMSF values extracted from the
Caatoms of the selected complex structure.
ARTICLE IN PRESS
JID: SAJB [m5G;May 3, 2022;22:43]
8
S.R. Mahapatra, J. Dey, T.K. Raj et al. South African Journal of Botany 00 (2022) 19
Because of the limitations of the experimental tests, additional in
vitro and/or in vivo analysis of the powerful natural metabolites
under investigation is strongly recommended as a promising target
for the development of natural medicines targeting type 3 fimbrial
protein.
Funding statement
The present study is an inhouse exploratory research work;
authors received no funding support from an external source.
Availability of data and materials
The datasets generated during and/or analyzed during the current
study are all present in the supplementary files as Supplementary
Tables.
Declaration of Competing Interest
The authors declare that there are no conflicts of interest.
Acknowledgments
We acknowledge infrastructure support available through DBT-
BUILDER program (BT/INF/22/SP42155/2021) at KIIT Deemed to Be
University, Bhubaneswar.
References
Atanasov, A.G., Zotchev, S.B., Dirsch, V.M., Supuran, C.T., 2021. Natural products in drug
discovery: advances and opportunities. Nat. Rev. Drug Discov. 20 (3), 200–216.
Banerjee, P., Eckert, A.O., Schrey, A.K., Preissner, R., 2018. ProTox-II: a webserver for the
prediction of toxicity of chemicals. Nucleic Acids Res. 46 (W1), W257–W263.
Benarba, B., Pandiella, A., 2020. Medicinal plants as sources of active molecules against
COVID-19. Front. Pharmacol. 11, 1189.
Bhuiyan, F.R., Howlader, S., Raihan, T., Hasan, M., 2020. Plants metabolites: possibility
of natural therapeutics against the COVID-19 pandemic. Front. Med. (Lausanne) 7,
444.
Biswas, S.S., Browne, R.B., Borah, V.V., Roy, J.D., 2021. In silico approach for phytocom-
pound-based drug designing to fight efflux pump-mediated multidrug-resistant
mycobacterium tuberculosis. Appl. Biochem. Biotechnol. 193 (6), 1757–1779.
Cassir, N., Rolain, J.M., Brouqui, P., 2014. A new strategy to fight antimicrobial resis-
tance: the revival of old antibiotics. Front. Microbiol. 5, 551.
Choi, M., Tennant, S.M., Simon, R., Cross, A.S., 2019. Progress towards the development
of Klebsiella vaccines. Expert Rev. Vaccines 18 (7), 681–691.
Daina, A., Michielin, O., Zoete, V., 2017. SwissADME: a free web tool to evaluate phar-
macokinetics, drug-likeness and medicinal chemistry friendliness of small mole-
cules. Sci. Rep. 7 (1), 1–13.
Effah, C.Y., Sun, T., Liu, S., Wu, Y., 2020. Klebsiella pneumoniae: an increasing threat to
public health. Ann. Clin. Microbiol. Antimicrob. 19 (1), 1–9.
Eisenberg, D., L€
uthy, R., Bowie, J.U., 1997. [20]VERIFY3D: assessment of protein models
with three-dimensional profiles. Methods Enzymol. 277, 396–404.
Ferreira, R.L., da Silva, B., Rezende, G.S., Nakamura-Silva, R., Pitondo-Silva, A.,
Campanini, E.B., Pranchevicius, M.C.D.S., 2019. High prevalence of multidrug-resis-
tant Klebsiella pneumoniae harboring several virulence and b-lactamase encoding
genes in a Brazilian intensive care unit. Front. Microbiol. 9, 3198.
Gonelimali, F.D., Lin, J., Miao, W., Xuan, J., Charles, F., Chen, M., Hatab, S.R., 2018. Anti-
microbial properties and mechanism of action of some plant extracts against food
pathogens and spoilage microorganisms. Front. Microbiol. 9, 1639.
Katsila, T., Spyroulias, G.A., Patrinos, G.P., Matsoukas, M.T., 2016. Computational
approaches in target identification and drug discovery. Comput. Struct. Biotechnol.
J. 14, 177–184.
Khameneh, B., Iranshahy, M., Soheili, V., Bazzaz, B.S.F., 2019. Review on plant antimi-
crobials: a mechanistic viewpoint. Antimicrob. Resist. Infect. Control 8 (1), 1–28.
Khater, F., Balestrino, D., Charbonnel, N., Dufayard, J.F., Brisse, S., Forestier, C., 2015. In
silico analysis of usher encoding genes in Klebsiella pneumoniae and characteriza-
tion of their role in adhesion and colonization. PLoS One 10 (3), e0116215.
Kim, D.E., Chivian, D., Baker, D., 2004. Protein structure prediction and analysis using
the Robetta server. Nucleic Acids Res. 32, W526–W531 suppl_2.
Kopec, K.K., Bozyczko-Coyne, D., Williams, M., 2005. Target identification and validation in
drug discovery: the role of proteomics. Biochem. Pharmacol. 69 (8), 1133–1139.
Kumari, R., Kumar, R., Open Source Drug Discovery Consortium, Lynn, A., Open Source
Drug Discovery Consortium, 2014. g_mmpbsa–a GROMACS tool for high-through-
put MM-PBSA calculations. J. Chem. Inf. Model. 54, 10–1021 1951.
Laskowski, R.A., MacArthur, M.W., Moss, D.S., Thornton, J.M., 1993. PROCHECK: a pro-
gram to check the stereochemical quality of protein structures. J. Appl. Crystallogr.
26 (2), 283–291.
Lee, G.R., Won, J., Heo, L., Seok, C., 2019. GalaxyRefine2: simultaneous refinement of
inaccurate local regions and overall protein structure. Nucleic Acids Res. 47 (W1),
W451–W455.
Li, B., Zhao, Y., Liu, C., Chen, Z., Zhou, D., 2014. Molecular pathogenesis of Klebsiella
pneumoniae. Future Microbiol. 9 (9), 1071–1081.
Lin, X., Li, X., Lin, X., 2020. A review on applications of computational methods in drug
screening and design. Molecules 25 (6), 1375.
Mahapatra, S.R., Dey, J., Kaur, T., Sarangi, R., Bajoria, A.A., Kushwaha, G.S., Suar, M.,
2021. Immunoinformatics and molecular docking studies reveal a novel Multi-Epi-
tope peptide vaccine against pneumonia infection. Vaccine 39 (42), 6221–6237.
Martin, R.M., Bachman, M.A., 2018. Colonization, infection, and the accessory genome
of Klebsiella pneumoniae. Front. Cell. Infect. Microbiol. 8, 4.
Meng, X.Y., Zhang, H.X., Mezei, M., Cui, M., 2011. Molecular docking: a powerful
approach for structure-based drug discovery. Curr. Comput. Aided Drug Des. 7 (2),
146–157.
Morris, G.M., Lim-Wilby, M., 2008. Molecular docking. Molecular Modeling of Proteins.
Humana Press, pp. 365–382.
Murphy, C.N., Clegg, S., 2012. Klebsiella pneumoniae and type 3 fimbriae: nosocomial
infection, regulation and biofilm formation. Future Microbiol. 7 (8), 991–1002.
Nezafat, N., Karimi, Z., Eslami, M., Mohkam, M., Zandian, S., Ghasemi, Y., 2016. Design-
ing an efficient multi-epitope peptide vaccine against Vibrio cholerae via com-
bined immunoinformatics and protein interaction based approaches. Comput.
Biol. Chem. 62, 82–95.
Opoku-Temeng, C., Kobayashi, S.D., DeLeo, F.R., 2019. Klebsiella pneumoniae capsule
polysaccharide as a target for therapeutics and vaccines. Comput. Struct. Biotech-
nol. J. 17, 1360–1366.
Paczosa, M.K., Mecsas, J., 2016. Klebsiella pneumoniae: going on the offense with a
strong defense. Microbiol. Mol. Biol. Rev. 80 (3), 629–661.
Paiva, P.M., Gomes, F.S., Napole~
ao, T.H., S
a, R.A., Correia, M.T.S., Coelho, L.C.B.B., 2010.
Antimicrobial activity of secondary metabolites and lectins from plants. Curr. Res.
Technol. Educ. Top. Appl. Microbiol. Microb. Biotechnol. 1 (2), 396–406.
Paul Gleeson, M., Hersey, A., Hannongbua, S., 2011. In-silico ADME models: a general
assessment of their utility in drug discovery applications. Curr. Top. Med. Chem. 11
(4), 358–381.
Pendleton, J.N., Gorman, S.P., Gilmore, B.F., 2013. Clinical relevance of the ESKAPE
pathogens. Expert Rev. Anti Infect. Ther. 11 (3), 297–308.
Peterson, E., Kaur, P., 2018. Antibiotic resistance mechanisms in bacteria: relationships
between resistance determinants of antibiotic producers, environmental bacteria,
and clinical pathogens. Front. Microbiol. 9, 2928.
Pratheeshkumar, P., Kuttan, G., 2012. Antimetastatic potential of vernolide-A, a sesqui-
terpenoid from Vernonia cinerea L. Hum. Exp. Toxicol. 31 (1), 66–80.
Stahlhut, S.G., Struve, C., Krogfelt, K.A., 2012. Klebsiella pneumoniae type 3 fimbriae aggluti-
nate yeast in a mannose-resistant manner. J. Med. Microbiol. 61 (3), 317–322.
Stahlhut, S.G., Struve, C., Krogfelt, K.A., Reisner, A., 2012. Biofilm formation of Klebsiella
pneumoniae on urethral catheters requires either type 1 or type 3 fimbriae. FEMS
Immunol. Med. Microbiol. 65 (2), 350–359.
Struve, C., Bojer, M., Krogfelt, K.A., 2009. Identification of a conserved chromosomal
region encoding Klebsiella pneumoniae type 1 and type 3 fimbriae and assessment
of the role of fimbriae in pathogenicity. Infect. Immun. 77 (11), 5016–5024.
Tarkkanen, A.M., Virkola, R., Clegg, S., Korhonen, T.K., 1997. Binding of the type 3 fim-
briae of Klebsiella pneumoniae to human endothelial and urinary bladder cells.
Infect. Immun. 65 (4), 1546–1549.
Thomford, N.E., Senthebane, D.A., Rowe, A., Munro, D., Seele, P., Maroyi, A., Dzobo, K.,
2018. Natural products for drug discovery in the 21st century: innovations for
novel drug discovery. Int. J. Mol. Sci. 19 (6), 1578.
Tian, W., Chen, C., Lei, X., Zhao, J., Liang, J., 2018. CASTp 3.0: computed atlas of surface
topography of proteins. Nucleic Acids Res. 46 (W1), W363–W367.
Wiederstein, M., Sippl, M.J., 2007. ProSA-web: interactive web service for the recogni-
tion of errors in three-dimensional structures of proteins. Nucleic Acids Res. 35
(suppl_2), W407–W410.
Yi, F., Li, L., Xu, L.J., Meng, H., Dong, Y.M., Liu, H.B., Xiao, P.G., 2018. In silico approach in
reveal traditional medicine plants pharmacological material basis. Chin. Med. 13
(1), 1–20.
Table 4
MM/PBSA free energy calculations of docked complexes.
Energies Energies (KJ/mol) Std. Dev (KJ/mol)
van der Waal energy 111.983 +/44.682
Electrostatic energy 23.323 +/15.993
Polar solvation energy 80.160 +/45.186
SASA energy 11.518 +/4.628
Total Binding energy 66.663 +/29.026
ARTICLE IN PRESS
JID: SAJB [m5G;May 3, 2022;22:43]
9
S.R. Mahapatra, J. Dey, T.K. Raj et al. South African Journal of Botany 00 (2022) 19
