Ligand- and structure-based in silico studies to identify kinesin spindle protein (KSP) inhibitors as potential anticancer agents

Article abstract of DOI:10.1080/07391102.2017.1396255

Kinesin spindle protein (KSP) belongs to the kinesin superfamily of microtubule-based motor proteins. KSP is responsible for the establishment of the bipolar mitotic spindle which mediates cell division. Inhibition of KSP expedites the blockade of the normal cell cycle during mitosis through the generation of monoastral MT arrays that finally cause apoptotic cell death. As KSP is highly expressed in proliferating/cancer cells, it has gained considerable attention as a potential drug target for cancer chemotherapy. Therefore, this study envisaged to design novel KSP inhibitors by employing computational techniques/tools such as pharmacophore modelling, virtual database screening, molecular docking and molecular dynamics. Initially, the pharmacophore models were generated from the data-set of highly potent KSP inhibitors and the pharmacophore models were validated against in house test set ligands. The validated pharmacophore model was then taken for database screening (Maybridge and ChemBridge) to yield hits, which were further filtered for their drug-likeliness. The potential hits retrieved from virtual database screening were docked using CDOCKER to identify the ligand binding landscape. The top-ranked hits obtained from molecular docking were progressed to molecular dynamics (AMBER) simulations to deduce the ligand binding affinity. This study identified MB-41570 and CB-10358 as potential hits and evaluated these experimentally using in vitro KSP ATPase inhibition assays.

Full text of DOI:10.1080/07391102.2017.1396255

Journal of Biomolecular Structure and Dynamics
ISSN: 0739-1102 (Print) 1538-0254 (Online) Journal homepage: http://www.tandfonline.com/loi/tbsd20
Ligand and structure based in silico studies to
identify kinesin spindle protein (KSP) inhibitors as
potential anticancer agents
Chandrasekaran Balakumar, Muthusamy Ramesh, Chuin Lean Tham,
Samukelisiwe Pretty Khathi, Frank Kozielski, Cherukupalli Srinivasulu,
Girish A. Hampannavar, Nisar Sayyad, Mahmoud E. Soliman & Rajshekhar
Karpoormath
To cite this article: Chandrasekaran Balakumar, Muthusamy Ramesh, Chuin Lean Tham,
Samukelisiwe Pretty Khathi, Frank Kozielski, Cherukupalli Srinivasulu, Girish A. Hampannavar,
Nisar Sayyad, Mahmoud E. Soliman & Rajshekhar Karpoormath (2017): Ligand and structure
based in silico studies to identify kinesin spindle protein (KSP) inhibitors as potential anticancer
agents, Journal of Biomolecular Structure and Dynamics, DOI: 10.1080/07391102.2017.1396255
To link to this article: http://dx.doi.org/10.1080/07391102.2017.1396255
Accepted author version posted online: 24
Oct 2017.
Submit your article to this journal
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=tbsd20
Download by: [The Library at Queen's University]
Date: 24 October 2017, At: 07:28

Publisher: Taylor & Francis
Journal: Journal of Biomolecular Structure and Dynamics
DOI: http://doi.org/10.1080/07391102.2017.1396255
Ligand and structure based in silico studies to identify kinesin spindle
protein (KSP) inhibitors as potential anticancer agents
a
a
b
ChandrasekaranBalakumar, MuthusamyRamesh, Chuin Lean Tham, Samukelisiwe Pretty
a
b
a
Khathi, Frank Kozielski, CherukupalliSrinivasulu, Girish A.
a
a
a
a
Hampannavar, NisarSayyad, Mahmoud E. Soliman, and Rajshekhar Karpoormath *
aDiscipline of Pharmaceutical Sciences, College of Health Sciences, University of KwaZulu-
Natal (UKZN), Westville, Durban 4001, South Africa.
bDepartment of Pharmaceutical and Biological Chemistry, The School of Pharmacy,
University College London, 29-39 Brunswick Square, London WC1N 1AX, U.K.
*
Corresponding author: karpoormath@ukzn.ac.za (R. Karpoormath)
Abstract
Kinesin spindle protein (KSP) belongs to the kinesin superfamily of microtubule-
based motor proteins. KSPis responsible for the establishment of the bipolar mitotic spindle
which mediates cell division. Inhibition of KSP expedites the blockade of the normal cell
cycle during mitosis throughthe generation of monoastralMTarrays that finally cause
apoptotic cell death. As KSP is highly expressed in proliferating/cancer cells, it has gained
considerable attention as a potential drug target for cancer chemotherapy. Therefore, this
study envisaged to design novel KSP inhibitors by employing computational techniques/tools
such as pharmacophore modeling, virtual database screening, molecular docking, and
1

molecular dynamics. Initially, the pharmacophore models were generated from the dataset of
highly potent KSP inhibitors and the pharmacophore models were validated against in
housetest set ligands.The validated pharmacophore model was then taken for database
screening (Maybridgeand ChemBridge) to yield hits, which were further filtered for their
drug-likeliness. The potential hits retrieved from virtual database screening were docked
using CDOCKER to identify the ligandbinding landscape. The top-ranked hits obtained from
molecular docking were progressed to molecular dynamics (AMBER) simulations to deduce
the ligand binding affinity. This study identified MB-41570 and CB-10358 as potentialhits
and evaluated experimentally using in vitroKSP ATPase inhibition assays.
Key words: KSP inhibitors; pharmacophore modeling;CDOCKER; molecular dynamics;
KSP ATPase enzyme inhibition.
List of abbreviations
ADP
AMBER
ATP
CB
adenosine diphosphate
assisted model building with energy refinement
adenosine triphosphate
chembridge
CG
conjugate gradient
DS
discovery studio
GI
gastro intestinal
GAFF
KSP
MB
generalized amber force field
kinesin spindle protein
maybridge
MT
microtubule
MD
molecular dynamics
MM/GBSA molecular mechanics/generalized Born and surface area continuum solvation
PDB
RoG
protein data bank
radius of gyration
RMSD
RMSF
SD
root mean square deviation
root mean square fluctuations
steepest descent
2

3
D
three dimensional
van der Waals
vdW
Introduction
Kinesins are microtubule (MT) based motor proteins that belong to the kinesin
superfamily. In mitotic cells, they are involved in various key cellular events such as mitotic
spindle assembly, re-modeling of MTs, the positioning of subcellular organelles, and
chromosome segregation(Hirokawa, 1998).Kinesin spindle protein (also known as KSP, Eg5
or KIF11) is a plus-end-directed N-terminal MT motor and is responsible for the
establishment of the bipolar organization of the mitotic spindle(Goodson, Kang, & Endow,
1
994).The mitotic spindle segregates chromosome to process through mitosis. If a functional
spindle fails to form, normal segregation of chromosomes does not take
place.Consequently,checkpoint proteins inhibit cell division resulting in mitotic arrest
(DeBonis et al., 2003) followed by cell death.KSP is highly expressed in proliferating human
tissues and tumors of the colon, breast, uterus, ovary,and lungs(Compton, 2000).Therefore,
KSP is a potential drug target in cancer chemotherapy. Inhibitors of KSP are mitosis-specific
that preferably act on cells undergoing cell division(Sarli & Giannis, 2008).Hence,KSP
inhibitorsmay likely to have fewer side effects (with the exception of neutropeniaa main side
effect) while comparing with other agents affecting MT-based processes (Sakowicz et al.,
2
004; Song, Zhou, Wang, & Li, 2013).
The KSP inhibitorsare structurally diverse and include (i) six-membered fused
heterocycles (CK-0106023,monastrol, and dimethylenastron), (ii) five-membered nitrogen-
containing heterocycles(dihydro-pyrazolothiophenes, dihydro-thiadiazoles,and thiazoles),
(iii)biaryl compounds (GSK-1 and GSK-2) and (iv)natural products (adociasulfate-2)(El-
Nassan, 2013). Few of the KSP inhibitors were studied under different stages of clinical
trials.
SB-743921, a chromen-4-one derivative was evaluated in clinical trials(phase-II) to
treat non-Hodgkin lymphoma and Hodgkin lymphoma(https://clinicaltrials.gov).Further, it
has demonstrated 5-fold more potency than ispinesib in cell-based assays (Holen et al., 2011;
Yin et al., 2015).ARQ-621 is an allosteric inhibitor evaluated against solid metastatic
tumours and hematologic malignancies in a phase-Istudy (https://clinicaltrials.gov).
Similarly, AZD-4877 was evaluated under phase-II clinical trial for the treatment of
3

advanced bladder cancer (https://clinicaltrials.gov). A pyrrole analogue, MK-0711 was
studied in patients with advanced solid tumours under phase-I clinical trial
(
https://clinicaltrials.gov). Filanesib (also known as ARRY-520) is now under phase-III
clinical studies in the management of multiple myeloma in combination with carfilzomib
https://clinicaltrials.gov)(Figure 1). Ispinesib (SB-715992) was investigated in phase-II
studies in treating patients with metastatic or unresectable kidney cancer
https://clinicaltrials.gov) (Figure 2). Similarly, litronesib(known as LY2523355) is an ATP
non-competitive, and allosteric inhibitor evaluated in patients with breast cancer (phase-II)
https://clinicaltrials.gov) and reported as a promising anticancer agent (Wakui et al.,
014)(Figure 2).
(
(
(
2
KSP encompasses 1057 amino acid residues and is divided into three major domains
namely the motor domain (N-terminal), stalk and tail (C-terminal) domains (Figure 3). A
short stretch of initial 20 amino acids precedes the motor domain (21-356 amino acids)
followed by a neck linker (357-363),the internal stalk responsible for oligomerisationand the
C-terminal tail (364-1057),which includes a phosphorylation-site (Thr927 in human
KSP)(Blangy, Arnaud, & Nigg, 1997; Turner et al., 2001).The motor domain is the highly
significant segment as it is responsible for the hydrolysis of adenosine triphosphate (ATP) to
adenosine diphosphate (ADP) thus generating the required energy for movement of the motor
along MTs.KSP utilizes the energy to propel spindle MTs in an antiparallel manner through
its motor activity and sliding them apart. Thus, it contributes to the separation of centrosomes
and the generation of bipolar spindle assembly during cell division(Civelekoglu-Scholey &
Scholey, 2010; Saunders, Powers, Strome, & Saxton, 2007; Wordeman, 2010).As the motor
domain of KSP plays a central role in cell division, many research groups determined the
structure of the motor domain and studied the mode of action of small organic moleculesas
KSP inhibitors(Barsanti et al., 2010; Fraley et al., 2006; Kaan, Ulaganathan, Rath, et al.,
2
010; Kaan, Ulaganathan, Hackney, & Kozielski, 2010; Wang et al., 2012). Recently, the X-
ray crystal structure of the KSP motor domain in complex with SB-743921 a second
generation ispinesib analogue and ADP was reported in the protein data bank (PDB ID:
4
BXN)(Figure 4). The inhibitor was found to bind at the allosteric binding-siteformed by
helix α2/loop L5/α3(Talapatra, Anthony, Mackay, & Kozielski, 2013).
Computational methods such as ligand-based (pharmacophore model and 3D QSAR),
structure-based (molecular docking or molecular dynamics simulations) and integrated
approaches are generally employed in designing therapeutically active new chemical entities.
4

These approaches have been employed successfully to discover drugs such as dorzolamide
carbonic anhydrase inhibitor), captopril (angiotensin-converting enzyme inhibitor),
(
saquinavir, ritonavir, indinavir and raltegravir (anti-HIV drugs) and tirofiban (fibrinogen
antagonist)(Gu, Zhang, & Yuan, 2014; Hartman et al., 1992; Talele, Khedkar, & Rigby,
2
010; Vijayakrishnan, 2009)as well as drug metabolism profiles(Ramesh & Bharatam, 2014).
First,a quantitative 3D pharmacophore model with good correlation coefficient (r = 0.965) for
KSP inhibitors was developed by Liu and co-workers. The studyreportedthe best hypothesis
bearing four different chemical features (ring aromatic, hydrogen bond acceptor, hydrogen
bond donor, and hydrophobic) using HypoGenmodule implemented in Catalyst software(Liu,
You, & Chen, 2007). Subsequently, Jiangand his colleagues reported docking (structure-
based) studies of 15 KSP inhibitors and proposed the significance of a ‘minor pocket’ that is
situated away from the ‘main pocket’ in enhancing the binding affinity of ligands(Jiang,
Chen, Wang, & You, 2007).
Till date, there are49 solved X-ray crystal structures ofKSP available in the
PDB(http://www.rcsb.org/pdb/).Thus, these structures paved a way towards the determination
of novel KSP inhibitors using structure-based drug design method. In 2012, Nagarajan and
collaborators reported combined ligand- and structure-based virtual screening approaches
toidentify novel KSP inhibitors. In their study, the interaction-based pharmacophore models
(IBPs) were developed and validated using six different complexes of KSP crystal structures.
The generated models were appliedto filter the ChemDiv database (0.7 million) to
identifyinitial hits. The IBP-based hits were subjected to docking (GOLD and
GLIDE),biological screening (inhibition of tumor cell proliferation and basal Eg5 ATPase
activity),and phenotypic analysis. These integrative approaches identified substituted
imidazolidinecompound as a novel KSP inhibitor(Nagarajan, Skoufias, Kozielski, & Pae,
2
012)(9,Figure5). Similarly, Carbajales and co-workers analyzedKSP allosteric binding-
sitefrom the 24 crystal structures of allostericinhibitor complexesand developed planar
heterocycles based on the crystal structures of KSP (PDB ID: 2GM1). This structure-guided
design led to the discovery of benzimidazole-based KSP inhibitors throughmulticomponent
synthesis(10and 11, Figure 5)(Carbajales et al., 2014).Recently (2015), Yokoyama and co-
workers reported the crystal structure (PDB ID: 4WPN) and biochemical characterisation of
the Eg5 motor domain in complexwith a new type of allosteric inhibitor (PVZB-1194). This
biphenyl-type inhibitor was proposed to bind within the novel α4/α6 allosteric pocket,
whereas residues in the allosteric pocket formed by α2/loop L5/α3 are responsible for the
binding of conventional KSP inhibitors (Yokoyama et al., 2015).In recent literature reports
5

(2016), chemical feature-based pharmacophore modelswere generated and validated.
Subsequently, the best validated model was considered for virtual screening including the
drug-like parameters to yield hits. Based on molecular interactions (docking) and computed
electronic parameters (density functional theory calculations), authors reported six lead
compounds as novel KSP inhibitors(Karunagaran, Subhashchandrabose, Lee, & Meganathan,
2
016).
Currently, there are no specific KSP inhibitor drugs in the market. However, few KSP
inhibitors (Figure 1 and 2) are at different stages of clinical trials. This prompts medicinal
chemists to design and develop new KSP inhibitors through rational drug discovery
approaches. Therefore, in the present study, we envisaged to develop novel KSP inhibitors by
employing ligand- and structure-based drug design techniques. In addition, in vitro enzyme
assay evaluations were carried out for the identified hits against the basal and MT-stimulated
ATPase activity of KSP.
2
. Materials and methods
2
.1. Collection of training and test set
The selection of a suitable training set is crucialfor the generation of ligand-based
pharmacophore models. Additionally, it is necessary to consider an appropriate test set of
ligands that are not employed for the generation of models but may assistin the validation of
the obtained pharmacophore models. In this study, a data set of highly active KSP inhibitors
incorporating clinical/preclinical trial molecules (Cox et al., 2008; Gerecitano et al., 2009;
Roecker et al., 2007; Rosen et al., 2010; Theoclitou et al., 2011; Woessner et al., 2009)as a
training set (Figure1). Further, it was taken into an account that the selected ligands were
pharmacologically screened under similar experimental conditions (in vitro), and structural
diversity. Many of the reported active and inactive KSP inhibitors exhibited a stereocenter in
their chemicalstructure. Hence,appropriate stereochemistry of ligands was carefully
considered in the training as well as the test sets.
2
.2. Generation of pharmacophore models
Pharmacophore modeling relates pharmacological activity with the 3D spatial
arrangement of various chemical features in a set of active molecules.HipHop is a well-
known application for common feature-based pharmacophore model generation. It is
incorporated in Discovery Studio (DS) software package as ‘Common Feature
6

PharmacophoreGeneration’ protocol(DassaultSystèmes BIOVIA, Discovery Studio Modeling
Environment, Release 4.5, San Diego: DassaultSystèmes, 2015).The present study includes
the six most active ligands in the training set to build/construct suitablepharmacophore
models.To generate the pharmacophore models, aconformational analysis was carried out for
all molecules of thetraining set using the same software package. These conformations were
generated by a fast conformational method with a maximum limit of 255,having an energy 20
kcal/molhigher than the global energy minimum(Neves, Dinis, Colombo, & Sá e Melo,
2
009). Based on the presence of certain chemical features in the training set, hydrogen bond
acceptor (A), hydrogen bond donor (D), hydrophobic (H), hydrophobic aliphatic (HY-Ali),
positively ionizable (P) and ring aromatic (R) chemical functions were considered for feature
analyses. The minimum inter-feature distance was set to the default value (2.97 Å).
2
.3. Validation of pharmacophore model
The pharmacophore model usually consists of certain fixed parameters/filters that
every new chemical entity has to satisfy inorder to be considered as an active inhibitor. The
validation of pharmacophore model is performed by screening an in house test set consisting
of both known active and inactive molecules. This is an indispensable step for the validation
of the pharmacophore model. The test set was thoroughly verified to exclude high molecular
weight compounds, prodrugs, and peptides. The in house test set constituted of 54 active KSP
inhibitors (IC = 0.1-100.0 nM) and 755 inactive molecules (IC >100.0 nM). From the 54
5
0
50
active KSP inhibitors, 36 ligands were identified as ‘highly active’ (IC values < 20 nM)
5
0
while the remaining 18 ligands were arbitrarily classified as ‘moderately active’ (IC = 20 to
5
0
1
00 nM). Although, from the pharmacological point of view the ligands with IC values 100
50
to 500 nM are also considered as active inhibitors, however in our study the cut off IC value
5
0
was set at 100 nM as one of the crucial parameter to identify highly potent inhibitors
Supplementary Table S1).
(
2
.4. Virtual screening
Virtual 3D database screening can offer a valuable route to identify potential lead
molecules for further drug development.Eventually, database searchprovides an advantage
that theidentified or retrieved molecules can be procured handily or synthesized conveniently
for pharmacological screeningprotocols when compared to other de novodrug design
methodologies(Michaux et al., 2006).The approach uses 3D features of the pharmacophore
7

model asa query in the database screening to discover novelhit molecules. Twodatabase
search methods (i) Fast/Flexible (diverse low-energy conformation) and Best/Flexible(best
coverage of conformational space)are incorporated in DS. In general, Best/Flexible search
method is employed for virtual database screening to obtain better results (Karunagaran,
Subhashchandrabose, Lee, & Meganathan, 2016)and the same method was implemented in
our study. Two different databases, namely Maybridge-HitDiscover (MB)and ChemBridge
(
CB) were searched to identify potential KSP targeting hits. Further, the molecules were
filtered fortheir drug-like properties using Lipinski’s rule of five[molecular weight less than
00, less than 5 hydrogen bond donor groups, less than 10 hydrogen bond acceptor groups
5
and an octanol/water partition co-efficient (LogP) value of less than 5], Veber rule [10 or less
2
than 10 rotatable bonds and polar surface area equal to or less than 140 Å ](Veber et al.,
2
002)andADMET (Absorption, Distribution, Metabolism, Excretion, andToxicity) filters
using DS. Hit ligands which qualified for all of thesescreening protocols were
takenforwardfor molecular docking studies.
2
.5. Molecular docking
Molecular docking is a structure-based in silico approach to identify the possible
ligand binding landscape inside the binding-site of a macromolecular protein/receptor.
Molecular docking was conducted using the CDOCKER module under the receptor-
ligandinteractions protocol implemented in DS. It is a grid-based docking method
thatemploys CHARMm force field. In this protocol, a macromolecule is kept under constraint
while ligands are enabled for their conformational flexibility during the refinement step. A
high-temperature molecular dynamics (MD) pursued by random rotations resulted in different
conformations for initial ligand structure. The final refinement of random conformations was
implemented through a grid-based simulated annealingfollowed by either grid-based or full
forcefield minimization(Wu, Robertson, Brooks, & Vieth, 2003).To perform molecular
docking,a crystal structure of humanKSP with bound inhibitor (SB-743921) was used (PDB
ID: 4BXN)(Talapatra et al., 2013).Prior to the molecular docking, KSPwas prepared using
‘prepare protein’ implemented under ‘Macromolecules’ module of DS. All the water
molecules, chloride ions, and cadmium ions present in co-crystallized structure ofthe protein
were removed and the missing hydrogen atoms were added to the KSP structure (Figure
6
A).The co-factor (ADP)present in the crystal structure was retained for moleculardocking
protocol in which the pH for protonation was set at 7.4. To optimize the molecular docking
8

protocol, several docking runs were performed. During the optimization, various parameters
were set for theactive-site radius (15Å), scoring function (CDOCKER energy), algorithm
(CDOCKER), physiological state of the ligands (ionized state), root mean square deviation (<
1
.5) to enhance the docking accuracy. By employing receptor-ligand interactions tool, the
total protein was defined as a receptor and a binding-site sphere was generated with the
radius of 15Å around the inhibitory ligand present in the protein structure (Figure 6B).
Further, the ligands were prepared for ionization change, Lipinski’s filter, and 3D
conformation generation. For the purpose of the validation of the molecular docking protocol,
the bound ligand (SB-743921) was extracted from the co-crystallized complex and was
docked into the definedbinding-site of the protein. Similarly, molecules present in the training
set, test set and the hits retrieved from virtualscreeningwere also docked into the pre-defined
binding-site of KSP (target protein) with default settings of DS. After completion of docking,
the protein-ligand complexes were analyzed to investigate the type of interactions and
CDOCKER energy values(Yang, Yang, Zou, Li, & Zhu, 2016).The final hits of molecular
docking were selected based on their higher (most negative, thus favourable to
binding)CDOCKER energy and selected complexes were subjected to molecular dynamics
simulations.
2
.6. Molecular dynamics
Molecular dynamics (MD) is a widely used structured-based in silico approach to
determine the binding efficiency of ligands inside the binding-site of macromolecules(Zhao
Caflisch, 2015). All MD simulations were run in AMBER 14.0 software package in
&
SANDER(Case et al.,2014).The protein-ligand complexes of 4BXN-SB743921 (native ligand
in the crystal structure)(Talapatra et al., 2013) and the top-ranked hit molecules of molecular
dockingfrom each database weresubjected to MD simulations. The ligand binding pose
obtained from top-ranked molecular docking was considered as starting point to runMD
simulations. Initially, the geometry of the protein and bound ligands from the docked
complexes were separated using Chimera Software package(Pettersen et al., 2004).The
ligands were defined with Generalized Amber Force Field (GAFF) and the partial atomic
charges were derived from AM1-BCC method of Antechamber. The protein was defined with
9

AMBER99SB force field and was prepared in Leap module to add missing hydrogen and to
generate the parameters for a solvated protein-ligand complex(Kholmurodov, Smith,
Yasuoka, Darden, & Ebisuzaki, 2000).The complex was enclosed with TIP3P water
molecules in a box which was extended up to 10 Å from the center of the
molecule(Jorgensen, Chandrasekhar, Madura, Impey, & Klein, 1983). The solvated complex
+
+
was electro-neutralized with counter ion Na . Four Na ions were added to neutralize the
system.To remove the bad contacts, partial minimization was carried out with a restraint
-1
-2
potential of 500 kcal mol Å . Partial minimization was run for 1000 steps of steepest
descent (SD) followed by 500 steps of conjugate gradient (CG). Further, the complex was
energetically minimized using 50 steps of SD followed by 150 steps of CG. After
minimization, heating was conducted graduallyfrom 0 to 300 K for 5 psusing
LangevinDynamics. Then the system was equilibrated at 300 K in a constant pressure and
temperature ensemble for 500 ps. The pressure of the system was maintained at 1 bar using
Berendsenbarostat. All the bonds involving hydrogens were constrained using
SHAKEalgorithm(Ryckaert, Ciccotti, & Berendsen, 1977).MD was run without restrain at a
temperature of 300 K and a pressure of 1 bar. A5 ns MD was run with a time step of 2fs and a
distance cutoff of 12.0 Å for the non-bonded interactions(Berendsen, Postma, van Gunsteren,
DiNola, & Haak, 1984).The trajectories were saved and analyzed for every 1 ps.
Visualization of trajectories was carried out using Chimera software package. After MD
simulations, binding free energies were computed.
2
.7. MM/GBSA and other post dynamics analyses
Binding free energies were defined as bound and unbound states of protein-ligand
complexes.After 5ns of MD simulation, the binding free energies were computed by
MM/GBSA method[Eqn 1-5](Hou, Wang, Li, & Wang, 2011; Lyne, Lamb, & Saeh,
2
006).Binding affinities were estimated by averaging snapshots taken from the MD
trajectories. The binding free energies were averaged over 1000 snapshots of MD trajectories.
ΔGbind = Gcomplex – Greceptor - Gligand…………………………………. (1)
ΔGbind = Egas + G – TS………………………………..………………..(2)
sol
Egas = E + E + Eele…………………………………………………….(3)
int
vdw
G = G + GSA………………………………………………………………(4)
sol
GB
G = SASA…………………………………………………..………………(5)
SA
ϒ
E : gas-phase energy that consists of the internal energy(E ), coulomb energy (E ),
gas
int
ele
1
0

and thevan der Waals energies (EvdW).Egas was directly estimatedfrom the FF99SB force field
terms.The solvationfreeenergy (G ) was estimated from the energy contributionfrom the
sol
polar states (G ) and nonpolar states (G ). Thenonpolar solvation energy(G ) was
GB
SA
SA
determined from theSASA using a water probe radius of 1.4 Å, whereasthe polar solvation
G ) contribution was estimated bysolving the GB equation.The total entropy of the
(
GB
soluteand temperature was denoted by S and T, respectively.Post-MD analyses (root mean
square deviation, root mean square fluctuation, and radius of gyration), the snapshot of the
structures during simulation and average structure were carried out using CPPTRAJ and
PTRAJ module of AMBER 14(Roe & Cheatham, 2013; Duan, Liu, & Zhang, 2016).
2
2
.8. Chemical Synthesis and Characterization
.8.1. General
Melting points of all the synthesized compounds were recorded in open capillaries
using Electro-thermal (IA9300) digital melting point apparatus and are uncorrected. A Bruker
FT-IR spectrophotometer with universal ATR sampling accessory was used to record all IR
1
13
spectra. H and C NMR spectra were recorded on NMR spectrometer (Bruker Avance IV)
at 400 and 101 MHz, using CDCl with TMS as an internal standard and reported in parts per
3
million (ppm). The progress of the reactions and the purity of the synthesized compounds
were monitored by Thin Layer Chromatography (TLC) on precoated silica gel 60 F254 (mesh)
(E. Merck) and spots were visualized under UV light (long and short wavelength). The
analytical grade (AR) chemicals and reagents were obtained from Merck and Sigma-Aldrich
and were used as such without any further purification.The ChemBridge ligand (CB-10358)
was procured directly from ChemBridge Corporation, San Diego, USA.
2
.8.2. Synthesis of diethyl 2-amino-4,7-dihydrothieno[2,3-c]pyridine-3,6(5H)-dicarboxylate
(3)
Ethyl 4-oxo-piperidine-1-carboxylate 1 (0.04 mol; 6.0 ml), ethyl cyano acetate 2 (0.04
mol; 4.52 ml), powdered sulphur (0.04 mol; 1.28 gm), and absolute ethanol (40 ml) were
taken in a conical flask and warmed at 50-60 °C. Then diethylamine (4 ml) was added
dropwise with constant stirring until all sulphur went into solution. The stirring was
continued for an additional one hour until brownish-yellow coloured solid separated, then it
was cooled to room temperature and filtered. The crude product was dried and recrystallized
from ethanol to yield pure diethyl 2-amino-4,7-dihydrothieno[2,3-c]pyridine-3,6(5H)-
1
1

dicarboxylate (3).
Brownish-yellow solid, yield: 9.1 gm (87%), mp: 160-162 C; IR (KBr): 3422, 3318
◦
-1 1
(
NH ), 1685 (ester-C=O), 1665 (piperidine-C=O) cm ; H NMR (400 MHz, CDCl ): δ 1.24-
2
3
1
.29 (t, J = 7.02 Hz, 3H, CH CH ), 1.30-1.35 (t, J = 7.10 Hz, 3H, CH CH ), 2.80 (s, 2H, CH ,
2 3 2 3 2
piperidine), 3.64 (s, 2H, CH , piperidine), 4.12-4.18 (q, J = 7.06 Hz, 2H, CH CH ), 4.21-4.27
2
2
3
1
3
(
q, J = 7.13 Hz, 2H, CH CH ), 4.38 (s, 2H, CH , piperidine), 5.5 (br s, 2H, NH ) ppm; C
2 3 2 2
NMR (100 MHz, CDCl , CPD): δ 14.58, 14.86, 27.22, 41.44, 42.78, 59.78, 61.71, 155.71,
3
1
62.51, 165.94 ppm.
2
.8.3. Synthesis of diethyl 2-(2-chloroacetamido)-4,7-dihydrothieno[2,3-c]pyridine-3,6(5H)-
dicarboxylate (4)
The compound 3 (0.02 mol; 5.967 gm) was dissolved in dry N,N-dimethyl formamide
40 ml) and added triethylamine (0.02 mol; 2.8 ml) followed by the addition of a solution of
(
chloroacetyl chloride (0.024 mol; 1.9 ml) in DMF (10 ml) slowly dropwise. The reaction
mixture was stirred at room temperature for additional two hours until all starting material
converted to product (monitored by TLC). Then the mixture was kept at room temperature to
precipitate dark coloured solids. The product was isolated by filtration under vacuum, dried
and recrystallized with absolute alcohol to obtain pure intermediate 4.
◦
Cream coloured solid, yield: 6.4 gm (85%), mp: 126-128 C; IR (KBr): 3201 (NH),
1
1
695 (ester-C=O), 1676 (piperidine-C=O), 1655 (amide-C=O); H NMR (400 MHz, CDCl ):
3
δ 1.26-1.30 (t, J = 7.08 Hz, 3H, CH CH ), 1.37-1.40 (t, J = 7.18 Hz, 3H, CH CH ), 2.89-2.92
2
3
2
3
(
m, 2H, CH , piperidine), 3.68-3.71 (m, 2H, CH , piperidine), 4.15-4.20 (q, J = 7.12 Hz, 2H,
2
2
CH CH ), 4.26 (s, 2H, CH -acetamido), 4.34-4.39 (q, J = 7.11 Hz, 2H, CH CH ), 4.56 (s, 2H,
2
3
2
2
3
1
3
CH , piperidine), 12.10 (br s, 1H, NH) ppm; C NMR (100 MHz, CDCl , CPD): δ 14.51,
2
3
1
4.88, 41.26, 42.35, 42.78, 61.22, 61.86, 112.79, 146.99, 155.67, 163.83, 165.86 ppm.
2
.8.4. Synthesis of diethyl 2-(2-(4-methylpiperazin-1-yl)acetamido)-4,7-dihydrothieno[2,3-
c] pyridine-3,6(5H)-dicarboxylate (5)
To a suspension of the intermediate 4 (0.001 mol; 0.375 gm) in 3 ml of dry
tetrahydrofuran (THF) added 1-methylpiperazine (0.0015 mol) dropwise and stirred at 60 °C
for two hours. After completion of the reaction (as monitored by TLC), the excess THF was
evaporated under reduced pressure, the resulting residue was triturated with water and filtered
to obtain crude product. Recrystallization of crude compound with absolute alcohol resulted
1
2

in pure final compound 5 (Maybridgehit MB-41570).
◦
Golden-yellow solid, yield: 350 mg (80%), mp: 95-97 C; IR (KBr): 3229 (NH), 1659
1
(
ester-C=O), 1562 (piperidine-C=O), 1516 (amide-C=O); H NMR (400 MHz, CDCl ): δ
3
1
.26-1.29 (t, J = 7.01 Hz, 3H, CH CH ), 1.36-1.40 (t, J = 7.20 Hz, 3H, CH CH ), 2.45 (s, 3H,
2 3 2 3
N-CH ), 2.75 (s, 8H, 4CH -piperazine), 2.89 (s, 2H, CH , piperidine), 3.28 (s, 2H, CH -
3
2
2
2
acetamido), 3.68-3.69 (m, 2H, CH , piperidine), 4.14-4.19 (q, J = 7.06 Hz, 2H, CH CH ),
2
2
3
4
.33-4.38 (q, J = 7.06 Hz, 2H, CH CH ), 4.55 (s, 2H, CH , piperidine), 12.24 (br s, 1H, NH)
2 3 2
1
3
ppm; C NMR (100 MHz, CDCl , CPD): δ 14.34, 14.67, 41.64, 45.45, 52.68, 54.61, 60.50,
3
6
0.74, 61.61, 147.34, 155.50, 165.34, 168.14 ppm.
2
.8.5. Procurement and characterization of ChemBridgehit (CB-10358).
The ChemBridgehit “CB-10358” was procured directly from the drug database
provider (ChemBridge Corporation, San Diego, USA) and confirmed its structure based on
1
13
H and C NMR.
1
H NMR (400 MHz, CDCl ): δ 1.23-1.29 (t, J = 7.01 Hz, 3H, CH CH ), 2.17-2.45 (m,
3
2
3
6
H, piperidine), 2.79 (m, 1H, piperidine), 2.9-3.0 (m, 1H, piperidine), 3.18-3.22 (m, 2H,
CH ), 3.38-3.46 (m, 2H, CH 1H, piperidine), 3.64-3.74 (br s, 6H, OCH ), 4.15-4.20 (q, J =
2
2,
3
7
.06 Hz, 2H, CH CH ), 6.20 (s, 1H, Ar-H), 6.93 (s, 2H, Ar-H), 11.46 (br s, 1H, NH) ppm;
2 3
1
3
C NMR (100 MHz, CDCl , CPD): δ 14.07 (CH -CH ), 25.29 (CH -piperidine), 31.98 (CH -
3
3
2
2
2
piperidine), 32.09 (CH -CH ), 34.98 (CH-piperidine), 50.45 (CH -piperidine), 52.74 (CH -
2
2
2
2
piperidine), 53.04 (O-CH ), 53.23 (O-CH ), 55.34 (CH -CH ), 61.31 (CH CH ), 96.86, 98.15,
3
3
2
2
2
3
9
8.22 (Ar-C), 139.81 (Ar-C-N), 167.50 (C-OCH ), 167.73 (C-OCH ), 171.87 (C=O), 172.92
3 3
(C=O) ppm.
2
.9. Pharmacological evaluation
.9.1. Evaluation of hits against the steady-state basal and MT-stimulated KSP ATPase
2
activities
The experiments were performedat 25 °C using a 96-well Sunrise photometer (Tecan,
Mannesdorf,Switzerland) at a final volume of 100 μL per well. Steady-state basal and MT-
stimulated ATPase rates were measured using the pyruvate kinase/lactate
dehydrogenaselinkedassay in buffer A25A [25 mmol/L potassium ACES (pH 6.9),2 mmol/L
magnesium acetate, 2 mmol/L potassium EGTA, 0.1 mmol/L potassium EDTA, 1 mmol/L β-
mercaptoethanol]. The amounts of KSP were optimized at 80-100 nM for the basal and 5 nM
1
3

for the MT-stimulated activity assays. The IC values for the inhibition of the basal and MT-
5
0
stimulated ATPase activities of Eg5 were measured for K-858 (positive control) between 1
and 200 μM. The ATP concentration was fixed at 1 mM. For the inhibition of the MT-
stimulated ATPase activity, the MT concentration was 1μM. The data were analyzed
usingKaleidagraph 3.0 (Synergy SoftwareReading, PA) and Microsoft Excel to obtain IC50
values (Talapatra et al., 2013; Nagarajan et al., 2012).
3
.Results and discussion
3
.1. Common feature pharmacophore generation
Ten common feature-based pharmacophore models were generated using a training
set of ligands. The scores of the obtained pharmacophore models ranged from 74.029to
8.245and each of the pharmacophore models had a minimum of four features(Table 1).
6
These models vary in their orientation,vector directions, and composition. Essentially, the
major structural requisite determined by DS consists of ring aromatic (R), hydrogen bond
acceptor (A), positively ionizable (P) or hydrophobic (H) feature for favourable interactions.
The Direct Hitbit mask is the ligand that directly match the pharmacophore and the value ‘1’
indicates that the molecule maps to every feature in the pharmacophore. The Partial Hit bit
mask is the ligand that partially match the pharmacophore and the value ‘0’ indicates the
complete matching with the pharmacophore. The positively ionizable (P) pharmacophoric
feature and hydrogen bond acceptor (A) were observed as common features among all the ten
pharmacophore models. This is in agreement with the presence of basic amino groups and
hydrogen bond acceptor atoms in structural features of all the training set molecules. The top-
ranked pharmacophore model (Hypo1) was chosen as the best model based on scores, fit
values, the presence of the number of pharmacophoric features and feature diversity (Tables 1
and 2). The proposed modes of pharmacophore mapping of the training set ligands on this
model were found to be good and in accordance with fit values. The best model (Hypo1)
contains one ring aromatic (R), one positively ionizable (P), one hydrophobic (H), and two
hydrogen bond acceptors (A and A ) features. Hypo1 was found to becompetentadequately
1
2
to map the training set molecules and canbe applied effectively to recognize the potential
KSP inhibitory ligands from drug databases.The distance between two hydrogen bond
acceptors (A and A ) was determined as 3.510 Å. The hydrogen bond acceptor 1 (A ) was
1
2
1
located at a distance of 4.654 Å from the hydrophobic feature (H) and the feature P (6.080
Å), respectively. The hydrogen bond acceptor 2 (A ) was distanced from the ring aromatic
2
(R) feature at 3.653 Å and hydrophobic (H) feature at 5.048 Å. Three features (P, R, and H)
1
4

are positioned in atriangular fashionwherein the distance between ‘P’ and ‘R’ was found to be
.947 Å, while the distance between ‘P’ and ‘H’ features was determined to be 8.364 Å
Figure 7A).The highly active molecule of the training set (2) mapped all the
pharmacophoricfeatures effectively on Hypo1 (Figure 7B).
7
(
3
.2. Validation
The pharmacophore models were validated using a test set comprising of active and
inactive KSP inhibitors (Table 3 and Supplementary Table S1). The test set constitutes a total
of 809 ligands (active: 54andinactive: 755). The top-ranked pharmacophore model (Hypo1)
was validatedcorrespondingly using a test set by defining Maximum Omitted Feature value as
‘0’. Hypo1 screened 48 compounds (highly active=34 and moderately active=14) as active
out of 54 known active ligands (88.88%) and 13 compounds as inactive out of 755 known
inactive ligands (1.72%) from the test set (Table 3). Hypo1was observed toidentify the active
molecules and to distinguish the active from inactive molecules effectively. Finally, Hypo1
was selected as the best pharmacophore model of KSP inhibitors.
3
.3.Database screening/searching
The validated pharmacophore model (Hypo1) was subjected to searching in two
different databases[Maybridge(MB) and ChemBridge(CB)]constituting 52610 and 74917
ligands, respectively. During the database searching,MaxOmit Feature ‘0’ and fit value > 3
were applied so that molecules that match all the features would be screened. This screening
process identified 389 and 632ligands from MB and CB,respectively (Figure 8). Further,
molecules were filtered through (i) Lipinski’srule offive and (ii)Veberrule. The
implementation of drug-likeliness filter reduced the hits of drug unlikeliness profiles and
retrieved 340 and 603 hits from MBand CB databases as drug-like molecules. Further, these
hit molecules were filtered to retain the hits with efficient ADMET profiles. The approach
filtered 43 and 217 (Total: 260) ligands respectively as hits of necessary pharmacophoric,
drug-likeliness, and ADMET profiles. These hits were subjected to molecular docking
studies(Figure 8).
3
.4. Molecular docking
Molecular docking was employed as a structure-based filter to assess the potential
ligands interacting with binding-site residues as well as to predict the probable binding mode
of ligands withinthe protein structure. All the hits retrieved from the two different
1
5

databaseswere subjected to molecular docking to analyzethe capability of binding ofligands
inside the binding pocket of KSP.For each of the input ligands, CDOCKER generated
numerousreasonable binding conformations and rankedthese on the basis of
CDOCKERenergy values. Thus, the ligand-bound conformationbearing the most favourable
energy was considered as the best orientation.
Before performing the molecular docking for all the hit molecules, SB-743921 (bound
ligand of KSP) and training set (includes SB-743921) ligands were also docked into the
crystal structure of KSP (PDB ID: 4BXN) as a part of the validation of the molecular docking
protocol.Molecular docking reproduced almost the same binding mode of SB-743921 inside
the defined binding pocket with a docking energy of -26.05 kcal/mol(Table 2). The docking
pose of SB-743921 (pink coloured ball and stick model) showed favourable interactions with
crucial residues within the binding-site of KSP as observed in the crystal structure(yellow
coloured ball and stick model) pose (Figure 9). The chromen-4-one moiety of SB-743921
was surrounded by the crucial residues of Trp127, Try211, Glu215, and Ala218, whereas the
chloro substituent demonstrated hydrophobic interactions with Leu214 and Lys216,
respectively. Similarly, the methyl group on the aromatic ring exhibited hydrophobic
interactions with the residues of Leu132 and Ala133 (Figure 9). In 2007, Roeckerand co-
workers crystallized KSP with an inhibitor MKR (PDB: 2Q2Y) and revealed a weak
hydrogen bond between the carbonyl oxygen of MKR with the hydroxyl group of an aromatic
amino acid Tyr211 (3.26 Å)(Roecker et al., 2007).In 2010, Kaan and co-workers reported the
phenolic hydroxyl group of (S)-enastron interacting with Glu118 of KSP through a hydrogen
bond (2.69 Å). Further, a phenyl substituent of the same ligand interacted with aromatic
residues (Trp127 and Tyr211) through a π-π stacking interaction and an acidic residue
Glu116 exhibited vdW interactions thereby contributing to the higher ligand binding affinity
(PDB ID:2X7C)(Kaan, Ulaganathan, Rath, et al., 2010).The same research group crystallized
another KSP-inhibitor complex (PDB ID: 2WOG) and identified the existence of a
hydrophobic interaction of STLC with the side chain of Glu116(Kaan, Ulaganathan,
Hackney, et al., 2010).Recently (2013), the authors deposited the structure of the KSP-
ispinesib complex (PDB ID: 4A5Y) in which a strong hydrogen bond prevailed between the
free amino group of ispinesib with Glu116 (2.88 Å)(Yi et al., 2013).This has been further
evidenced by another KSP structure crystallized with a β-carboline inhibitor (PDB ID:3K3B)
which demonstrated the crucial involvement of two different hydrogen bonds for β-carboline
NH- and phenolic OH group (substituent) with Glu116 (3.00 Å) and Glu118 (2.66 Å),
respectively(Barsanti et al., 2010). The above literature clearly indicated the involvement of
1
6

the crucial residues Glu116, Glu118, Trp127 and Tyr211 in ligand binding and similar
molecular interactions were observed for the docked SB-743921 ligand.
The CDOCKER energies ranged from -26.05 to -33.78 kcal/mol for the training set
molecules (which are highly potent) suggesting favourable binding at the KSP binding-
site(Table 1). Further, molecular docking was performed for all the hit molecules resulting
from database screening. Molecular docking experiments indicated the most favourable
binding for the top-rankedhit molecules of Maybridge(MB-41570,CDOCKER energy: -42.70
kcal/mol)and ChemBridge (CB-10358, CDOCKER energy: -39.13 kcal/mol). It was observed
that the two hits displayed higher CDOCKER energies indicating the most favourable
binding. The KSP docked complexes of MB-41570 (Maybridge)(Figure 10)and CB-10358
(ChemBridge)(Figure 11) were found to have top-ranked CDOCKER energy values. These
complexes were then subjected to MD simulations to further substantiate the CDOCKER
energies of the hits. The CDOCKER energy values and molecular interactions of bound
ligand, training set ligands, and top-ranked hits (Figure 12) from each database are presented
in Table S2 of the supplementary material section. The top-ranked complex from each
database and the corresponding fit values are also depicted (Supplementary Table S3).
3
.5. Binding free energies
The MD studies were conducted for the top-ranked hits retrieved from molecular
docking approach (Figure 9) as well as the co-crystallized ligand (SB-743921). After 5ns MD
simulation, the binding free energies were computed to identify the inhibitory potential of the
two hits and SB-743921. SB-743921 has shown very high in vitropotency against KSP
(
IC =0.1 nM) and the computed binding free energy of SB-743921 was used as a reference
50
for the assessment of binding affinity of other hit molecules(Talapatra et al., 2013). The
calculated binding free energy for SB-743921 at the binding-site of KSP was -31.54
kcal/mol.The binding free energies of hit molecules were found to be -42.89 and -47.53
kcal/molfor MB-41570 (Maybridge) and CB-10358 (ChemBridge)respectively. These two hit
molecules therefore show stronger binding free energy values than the reference molecule
SB-743921. Further, MB-41570 (Maybridge) and CB-10358 (ChemBridge) were predicted as
the most promising candidatesfor KSP inhibition (Table 5). The van der Waals (vdW) and
electrostatic energies of MB-41570 and CB-10358 were found to be -52.15±4.54kcal/mol and
-51.73±3.53kcal/mol, respectively. The vdWinteractions energy has highly contributed to the
ligand binding affinity of the identified hits.In the future, this study may help us to designand
1
7

develop variousanalogs of MB-41570 and CB-10358asprobable potent KSP inhibitors
forcancer chemotherapy.
3
.6. Post molecular dynamics analysis
The stability of the protein-ligand complexes (i) 4BXN-SB743921, (ii) 4BXN-
MB41570,and (iii) 4BXN-CB10358 was analyzed over 5ns MD simulation by computing
RMSD, RMSF, and RoG.The ligand-protein complex of hit molecules showeda marginal
variation as compared to reference during the entire simulation suggesting the ligand bound
complex of hit molecules as stable as the reference. The RMSD values for the
complexes4BXN-SB743921, 4BXN-MB41570,and 4BXN-CB10358 were determined as
2
.45, 2.79, and 2.37,respectively(Table 6 and Figure 13).
The flexibility of amino acids for the complexes 4BXN-SB743921, 4BXN-
MB41570,and 4BXN-CB10358 was analyzed by assessing the root mean square fluctuations
RMSF) from MD trajectories (Figure 14). The overall RMSF was found to be 0.92, 0.98 and
.93,respectively for 4BXN-SB743921, 4BXN-MB41570and 4BXN-CB10358 complexes.
(
0
The complex of 4BXN-MB41570and 4BXN-CB10358 indicatedslightly higher amino acid
flexibility than the reference. However, the identical trend in the fluctuation of amino acid
residues was observed in all three complexes. The highest magnitude of flexibility was
observed around residues10-25, 35-45, and 230-245.
The radius of gyration (RoG) is defined as the moment of inertia of the group of
atoms from their centroid of mass. Computation of RoG during the simulation sheds light on
the stability of the system.Figure 15presents theRoG over 5ns simulations in the4BXN-
SB743921, 4BXN-MB41570and 4BXN-CB10358 complexes. The complex of 4BXN-
MB41570has shown highestRoG after the 0.5ns simulation.The simulation beyond
2
nsresulted in a decreased RoG. The average RoG for the complexes of 4BXN-
SB743921,4BXN-MB41570 and 4BXN-CB10358 was 20.97, 21.03and 20.87, respectively.
The results suggestedthat the relatively close stability of the complexes during the MD
simulations.
3
.7. Synthesis and characterization of Maybridgehit
The Maybridgehit (MB-41570) was synthesized through a convenient synthetic route
(Ismail et al., 2012) as shown in scheme-1. Structures of all the synthesized compounds were
1
13
characterized based on their physicochemical and spectral (IR, H NMR and C NMR)
analysis. The spectra of the newly synthesized compounds along with their anticipated
1
8

structures are presented in the supplementaryinformation. Ethyl 4-oxo-piperidine-1-
carboxylate 1 reacted with ethyl cyanoacetate2 and sulphur in absolute ethanol in the
presence of diethylamine afforded the key starting material, diethyl 2-amino-4,7-
dihydrothieno[2,3-c]- pyridine-3,6(5H)-dicarboxylate (3) quantitatively. The FT-IR spectrum
of 3, we observed a reasonably sturdy and characteristic bands around 3422, 3318 cm-1
accounting for N-H stretching of primary amine. Further, a couple of prominent carbonyl
-1
stretching frequencies were observed at 1685 and 1665 cm which was attributed to the ester-
1
C=O and piperidine-C=O, respectively. The structure of 3 was elucidated by H NMR
spectrum (400 MHz), wherein the appearance of an informative broad singlet signal at δ 5.5
ppm attributing to NH protons. Further, appearance of additional triplet (δ = 1.30-1.35 ppm)
2
and quartet (δ = 4.21-4.27 ppm) signals corresponding to CH -CH and CH -
2
3
2
CH ,respectivelyindicated the formation of the desired starting material 3 via the well-known
3
Gewald reaction. This has been further validated by the appearance of deshielded carbon
signals at δ 165.94, 162.51, and 155.71 ppm attributable to C-NH , N-C=O, and C-C=O
2
1
3
respectively in C NMR. Moreover, carbon signals in respect to that of ethyl groups
appeared at δ 14.58, 14.86, 59.78, and 61.71 ppm in the shielded region confirmed the
formation of 3. Reacting the amino group in 3 with chloroacetyl chloride in the presence of
an organic base (triethylamine) resulted an intermediate, diethyl 2-(2-chloroacetamido)-4,7-
dihydrothieno[2,3-c]pyridine-3,6(5H)-dicarboxylate 4. Thisamide formation was further
-1
confirmed by the appearance of NH stretching band at 3201 cm anddisappearance of
primary amino stretching band in its FT-IR spectrum. Moreover appearance of additional
-1
1
C=O bond stretching at 1655 cm has been assigned as carbonyl group of amide. The H
NMR of 4 showed a broad singlet in the shielded region at δ 12.1 ppm for amide -NH apart
from the appearance of acetamido methylene proton at δ 4.26 ppm as a singlet. This has been
1
3
further evidenced in the C NMR where the signal corresponding to acetamido carbonyl was
appeared at δ 163.83 ppm apart from the appearance of acetamido methylene signal in the
shielded region (δ = 41.26 ppm). Nucleophilic displacement of chlorine in 4 with secondary
heterocyclic amine such as N-methyl piperazine in tetrahydrofuran yielded the final
1
compound 5. The H NMR of 5 indicated the appearance of additional alicyclic proton in the
shielded region as singlet at δ 2.75 ppm corresponding to eight protons of piperazinemoiety,
apart from the appearance of one singlet signal at δ 2.45 ppm for N-CH protons. Similarly,
3
13C NMR data corroborated well with the appearance of characteristic N-methyl carbon at δ
4
5.45 ppm apart from the carbon signals of piperazine moiety in the upfield region. All the
above data confirmed the successful formation of the final compound 5 (Maybridgehit MB-
1
9

4
1570).
3
3
.8. Pharmacological evaluation of hit compounds
.8.1. Inhibition of hits against the steady-statebasal KSP and MT-stimulated KSP ATPase
activities
In silico identified hits (Maybridge: MB-41570and ChemBridge: CB-10358) were
evaluated against the basal and MT-stimulated KSP ATPase activities using standard
reported protocols(Talapatra et al., 2013; Nagarajan et al., 2012). From the concentration-
response curves obtained(seesupplementary information), it was observed that the two
identified hitswere unable to displayinhibition against the basal or the MT-stimulated KSP
ATPase.
4
. Conclusion
In this study, pharmacophore models were developed using ligand-based
approachesand validated against a test set of KSP inhibitors. The validated pharmacophore
model (Hypo1) exhibited one ring aromatic (R), one positively ionizable (P), one
hydrophobic (H), and two hydrogen bond acceptors (A and A ) asessential features for KSP
1
2
inhibition. This Hypo1 was used for virtual screening to identify potentKSP inhibitor
scaffolds. Subsequently, thesescaffoldswere subjected to moleculardocking and molecular
dynamics followed by the calculations of binding free energies. Finally, two ligands (MB-
4
1570 and CB-10358) wereidentified as potentialand easily obtainable or
synthesizable(Nankervis et al., 2011; Plater, Murdoch, Morphy, Rankovic, & Rees,
000)KSP inhibitors. Since the two identified hits (MB41570 and CB10358) demonstrated
2
similar interactions as compared to SB-743921 and exhibited higher binding energies, we
synthesized these and evaluated them against the basal and the MT-stimulated KSP ATPase
activities. Even though the identified hits had pharmacophoric feature similarities to SB-
7
43921, they displayed no in vitro KSP enzyme inhibition.However, the pharmacophore
model, interaction analysis, and the systematic protocol of the present study contributes to
our understanding of KSP inhibition and may help future medicinal chemists in developing
new KSP inhibitors by taking our research findings into account.
Acknowledgments
2
0

Authors are thankful to Discipline of Pharmaceutical Sciences, College of Health
Sciences, University of KwaZulu-Natal (UKZN), Durban, South Africa for providing the
necessary facilities. One of the authors (CB) gratefully acknowledgesNational Research
Foundation (DST-NRF), South Africafor research funding in the form of Innovation Post
DoctoralResearch Fellowship (ID: 99546).Authors also sincerely thank Centre for High
Performance Computing (CHPC), Cape Town, South Africa for computational resources.
Disclosure statement
Authors hereby declare that there are no financial/commercial conflicts of interest.
References
Barsanti, P. A., Wang, W., Ni, Z. J., Duhl, D., Brammeier, N., Martin, E., … Walter, A. O.
(
2010). The discovery of tetrahydro-β-carbolines as inhibitors of the kinesin Eg5.
Bioorganic and Medicinal Chemistry Letters, 20, 157–
60.https://doi.org/10.1016/j.bmcl.2009.11.012
1
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. The Journal of Chemical
Physics, 81, 3684–3690. https://doi.org/10.1063/1.448118
Blangy, A., Arnaud, L., & Nigg, E. A. (1997). Phosphorylation by p34(cdc2) protein kinase
regulates binding of the kinesin-related motor HsEg5 to the dynactin
subunitp150(Glued). Journal of Biological Chemistry, 272, 19418–19424.
https://doi.org/10.1074/jbc.272.31.19418
Carbajales, C., Prado, M. Á., Gutiérrez-de-Terán, H., Cores, Á., Azuaje, J., Novio, S., …
Coelho, A. (2014). Structure-based design of new KSP-Eg5 inhibitors assisted by a
targeted multicomponent reaction. ChemBioChem, 15, 1471–1480.
https://doi.org/10.1002/cbic.201402089
Case, D. A., Babin, V., Berryman, J. T., Betz, R. M., Cai, Q., Cerutti, D. S., … Kollman, P.
A. (2014). AMBER 14, University of California, San Francisco. Retrieved from http://
ambermd.org/
Civelekoglu-Scholey, G., & Scholey, J. M. (2010). Mitotic force generators and chromosome
segregation. Cellular and Molecular Life Sciences, 67, 2231–2250.
https://doi.org/10.1007/s00018-010-0326-6
Compton, D. A. (2000). Spindle assembly in animal cells. Annual Review of Biochemistry,
6
9, 95–114. https://doi.org/10.1146/annurev.biochem.69.1.95
2
1

Cox, C. D., Coleman, P. J., Breslin, M. J., Whitman, D. B., Garbaccio, R. M., Fraley, M. E.,
Hartman, G. D. (2008). Kinesin spindle protein (KSP) inhibitors. 9. Discovery of
2S)-4-(2,5-Difluorophenyl)-N-[(3R,4S)-3-fluoro-1-methylpiperidin-4-yl]-2-
…
(
hydroxymethyl)-N-methyl-2-phenyl-2,5-dihydro-1H-pyrrole-1-carboxamide (MK-0731)
for the treatment of taxane-refractory cancer. Journal of Medicinal Chemistry, 51, 4239–
4
252. https://doi.org/10.1021/jm800386y
DassaultSystèmes BIOVIA, Discovery Studio Modeling Environment, Release 4.5, San
Diego: DassaultSystèmes, 2015.
DeBonis, S., Simorre, J.-P., Crevel, I., Lebeau, L., Skoufias, D. A., Blangy, A., …
Kozielski,F. (2003). Interaction of the mitotic inhibitor monastrol with human kinesin
Eg5. Biochemistry, 42, 338–349. https://doi.org/10.1021/bi026716j
Duan, L., Liu, X., & Zhang, J. Z. H. (2016). Interaction entropy: A new paradigm for highly
efficient and reliable computation of protein-ligand binding free energy. Journal of the
American Chemical Society, 138, 5722–5728. https://doi.org/10.1021/jacs.6b02682
El-Nassan, H. B. (2013). Advances in the discovery of kinesin spindle protein (Eg5)
inhibitors as antitumor agents. European Journal of Medicinal Chemistry, 62, 614–631.
https://doi.org/10.1016/j.ejmech.2013.01.031
Fraley, M. E., Garbaccio, R. M., Arrington, K. L., Hoffman, W. F., Tasber, E. S., Coleman,
P. J., … Hartman, G. D. (2006). Kinesin spindle protein (KSP) inhibitors. Part 2: The
design, synthesis, and characterization of 2,4-diaryl-2,5-dihydropyrrole inhibitors of the
mitotic kinesin KSP. Bioorganic & Medicinal Chemistry Letters16, 1775-1779.
https://doi.org/10.1016/j.bmcl.2006.01.030
Gerecitano, J. F., O'Connor, O., Van Deventer, H., Hainsworth, J., Leonard, J., Afanasayev,
B., … Conlan, M. (2009). A phase I/II trial of the kinesin spindle protein (KSP)
inhibitor SB-743921 dosed q14d without and with prophylactic G-CSF in non-Hodgkin
lymphoma (NHL) or Hodgkin lymphoma (HL). Journal of Clinical Oncology, 27, (No.
1
5_suppl, abstr) 8578. https://doi.org/10.1200/jco.2009.27.15_suppl.8578.
Goodson, H. V, Kang, S. J., & Endow, S. a. (1994). Molecular phylogeny of the kinesin
family of microtubule motor proteins. Journal of Cell Science, 107, 1875–1884.
Retrieved from http://jcs.biologists.org/content/107/7/1875.abstract
Gu, W.-G., Zhang, X., & Yuan, J.-F. (2014). Anti-HIV drug development through
computational methods. The AAPS Journal, 16, 674–80. https://doi.org/10.1208/s12248-
0
14-9604-9
Hartman, G. D., Egbertson, M. S., Halczenko, W., Laswell, W. L., Duggan, M. E., Smith, R.
2
2

L., … Lynch, R. J. (1992). Non-peptide fibrinogen receptor antagonists. 1. Discovery
and design of exosite inhibitors. Journal of Medicinal Chemistry, 35, 4640–4642.
https://doi.org/10.1021/jm00102a020
Hirokawa, N. (1998). Superfamily and Dynein Kinesin of and the Mechanism Proteins
Organelle Transport. Science, 279, 519–526.
https://doi.org/10.1126/science.279.5350.519
Holen, K. D., Belani, C. P., Wilding, G., Ramalingam, S., Volkman, J. L., Ramanathan, R.
K., … Ho, P. T. C. (2011). A first in human study of SB-743921, a kinesin spindle
protein inhibitor, to determine pharmacokinetics, biologic effects and establish a
recommended phase II dose. Cancer Chemotherapy and Pharmacology, 67, 447–454.
https://doi.org/10.1007/s00280-010-1346-5
Hou, T., Wang, J., Li, Y., & Wang, W. (2011). Assessing the performance of the molecular
mechanics/Poisson Boltzmann surface area and molecular mechanics/generalized Born
surface area methods. II. The accuracy of ranking poses generated from docking.
Journal of Computational Chemistry, 32, 866–877. https://doi.org/10.1002/jcc.21666
http://www.rcsb.org/pdb/results/results.do?qrid=CA488289&tabtoshow=Current.(n.d.).Retrie
ved
fromhttp://www.rcsb.org/pdb/results/results.do?qrid=CA488289&tabtoshow=Current
https://clinicaltrials.gov (accessed on 09/09/2016).
Ismail, M. M., Kamel, M. M., Mohamed, L. W., Faggal, S. I. & Galal, M. A. (2012).
Synthesis and biological evaluation of thiophene derivatives as acetylcholinesterase
inhibitors. Molecules17, 7217-7231. https://doi:10.3390/molecules17067217
Jiang, C., Chen, Y., Wang, X., & You, Q. (2007). Docking studies on kinesin spindle protein
inhibitors: an important cooperative `minor binding pocket’ which increases the binding
affinity significantly. Journal of Molecular Modeling, 13, 987–992.
https://doi.org/10.1007/s00894-007-0219-2
Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., & Klein, M. L. (1983).
Comparison of simple potential functions for simulating liquid water. The Journal of
Chemical Physics, 79, 926. https://doi.org/10.1063/1.445869
Kaan, H. Y. K., Ulaganathan, V., Hackney, D. D., & Kozielski, F. (2010). An allosteric
transition trapped in an intermediate state of a new kinesin-inhibitor complex. The
Biochemical Journal, 425, 55–60. https://doi.org/10.1042/BJ20091207
Kaan, H. Y. K., Ulaganathan, V., Rath, O., Prokopcovà, H., Dallinger, D., Kappe, C. O., &
Kozielski, F. (2010). Structural basis for inhibition of Eg5 by dihydropyrimidines:
2
3

Stereoselectivity of antimitotic inhibitors enastron, dimethylenastron and fluorastrol.
Journal of Medicinal Chemistry, 53, 5676–5683. https://doi.org/10.1021/jm100421n
Karunagaran, S., Subhashchandrabose, S., Lee, K. W., & Meganathan, C. (2016).
Investigation on the isoform selectivity of novel kinesin-like protein 1 (KIF11) inhibitor
using chemical feature based pharmacophore, molecular docking, and quantum
mechanical studies. Computational Biology and Chemistry, 61, 47–61.
https://doi.org/10.1016/j.compbiolchem.2016.01.002
Kholmurodov, K., Smith, W., Yasuoka, K., Darden, T., & Ebisuzaki, T. (2000). A smooth-
particle mesh Ewald method for DL_POLY molecular dynamics simulation package on
the Fujitsu VPP700. Journal of Computational Chemistry, 21, 1187–1191.
https://doi.org/10.1002/1096-987X(200010)21:13<1187::AID-JCC7>3.0.CO;2-7
Liu, F., You, Q.-D., & Chen, Y.-D. (2007). Pharmacophore identification of KSP inhibitors.
Bioorganic & Medicinal Chemistry Letters, 17, 722–726.
https://doi.org/http://dx.doi.org/10.1016/j.bmcl.2006.10.083
Lyne, P. D., Lamb, M. L., & Saeh, J. C. (2006). Accurate prediction of the relative potencies
of members of a series of kinase inhibitors using molecular docking and MM-GBSA
scoring. Journal of Medicinal Chemistry, 49, 4805–4808.
https://doi.org/10.1021/jm060522a
Michaux, C., de Leval, X., Julémont, F., Dogné, J. M., Pirotte, B., & Durant, F. (2006).
Structure-based pharmacophore of COX-2 selective inhibitors and identification of
original lead compounds from 3D database searching method. European Journal of
Medicinal Chemistry, 41, 1446–1455. https://doi.org/10.1016/j.ejmech.2006.07.017
Nagarajan, S., Skoufias, D. A., Kozielski, F.,& Pae, A. N. (2012). Receptor–ligand
interaction-based virtual screening for novel Eg5/kinesin spindle protein inhibitors.
Journal of Medicinal Chemistry, 55, 2561–2573. https://doi.org/10.1021/jm201290v
Nankervis, J. L., Feil, S. C., Hancock, N. C., Zheng, Z., Ng, H. L., Morton, C. J., … Parker,
M. W. (2011). Thiophene inhibitors of PDE4: Crystal structures show a second binding
mode at the catalytic domain of PDE4D2. Bioorganic and Medicinal Chemistry Letters,
2
1, 7089–7093. https://doi.org/10.1016/j.bmcl.2011.09.109
Neves, M. A. C., Dinis, T. C. P., Colombo, G., & Sá e Melo, M. L. (2009). An efficient
steroid pharmacophore-based strategy to identify new aromatase inhibitors. European
Journal of Medicinal Chemistry, 44, 4121–4127.
https://doi.org/10.1016/j.ejmech.2009.05.003
Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C.,
2
4

&
Ferrin, T. E. (2004). UCSF Chimera-A visualization system for exploratory research
and analysis. Journal of Computational Chemistry, 25, 1605–1612.
https://doi.org/10.1002/jcc.20084
Plater, M. J., Murdoch, A. M., Morphy, J. R., Rankovic, Z., & Rees, D. C. (2000). Solid
phase synthesis of tertiary amines on amide REM resins: Grignard and metal hydride
compatible resins. Journal of Combinatorial Chemistry, 2, 508–512.
https://doi.org/10.1021/cc000023
Ramesh, M., & Bharatam, P. V. (2014). Importance of hydrophobic parameters in identifying
appropriate pose of CYP substrates in cytochromes. European Journal of Medicinal
Chemistry, 71,15–23. https://doi.org/10.1016/j.ejmech.2013.10.023
Roe, D. R., & Cheatham, T. E. (2013). PTRAJ and CPPTRAJ: Software for processing and
analysis of molecular dynamics trajectory data. Journal of Chemical Theory and
Computation, 9, 3084–3095. https://doi.org/10.1021/ct400341p
Roecker, A. J., Coleman, P. J., Mercer, S. P., Schreier, J. D., Buser, C. A., Walsh, E. S., …
Hartman, G. D. (2007). Kinesin spindle protein (KSP) inhibitors. Part 8: Design and
synthesis of 1,4-diaryl-4,5-dihydropyrazoles as potent inhibitors of the mitotic kinesin
KSP. Bioorganic & Medicinal Chemistry Letters, 17, 5677–5682.
https://doi.org/10.1016/j.bmcl.2007.07.074
Rosen, L., Chen, L.-C., Iyengar, T., Goldman, J., Lahr, S., Chen, C.-R., … Von Hoff, D. D.
(
2010). ARQ 621, a novel potent and selective inhibitor of Eg5: Preclinical data and
early results from a clinical phase 1 study. Cancer Research, 70, 2750. http://dx.doi.org/
0.1158/1538-7445.AM10-2750
1
Ryckaert, J. P., Ciccotti, G., & Berendsen, H. J. C. (1977). Numerical integration of the
cartesian equations of motion of a system with constraints: Molecular dynamics of n-
alkanes. Journal of Computational Physics, 23, 327–341. https://doi.org/10.1016/0021-
9
991(77)90098-5
Sakowicz, R., Finer, J. T., Beraud, C., Crompton, A., Lewis, E., Fritsch, A., … Wood, K. W.
2004). Antitumor activity of a kinesin inhibitor. Cancer Research, 64, 3276–3280.
(
Sarli, V., & Giannis, A. (2008). Targeting the kinesin spindle protein: Basic principles and
clinical implications. Clinical Cancer Research, 14, 7583–7587.
https://doi.org/10.1158/1078-0432.CCR-08-0120
Saunders, A. M., Powers, J., Strome, S., & Saxton, W. M. (2007). Kinesin-5 acts as a brake
in anaphase spindle elongation. Current Biology, 17, R453–R454.
https://doi.org/10.1016/j.cub.2007.05.001
2
5

Song, H., Zhou, S., Wang, R., & Li, S. (2013). Kinesin spindle protein (KSP) inhibitors in
combination with chemotherapeutic agents for cancer therapy. ChemMedChem, 8,
1
736–1749. https://doi.org/10.1002/cmdc.201300228
Talapatra, S. K., Anthony, N. G., Mackay, S. P., & Kozielski, F. (2013). Mitotic kinesin Eg5
overcomes inhibition to the phase I/II clinical candidate SB743921 by an allosteric
resistance mechanism. Journal of Medicinal Chemistry, 56, 6317–6329.
https://doi.org/10.1021/jm4006274
Talele, T. T., Khedkar, S. A., & Rigby, A. C. (2010). Successful applications of computer
aided drug discovery: Moving drugs from concept to the clinic. Current Topics in
Medicinal Chemistry, 10, 127–141. https://doi.org/10.2174/156802610790232251
Theoclitou, M.-E., Aquila, B., Block, M. H., Brassil, P. J., Castriotta, L., Code, E., … Zheng,
X. (2011). Discovery of (+)-N-(3-Aminopropyl)-N-[1-(5-benzyl-3-methyl-4-oxo-
[1,2]thiazolo[5,4-d]pyrimidin-6-yl)-2-methylpropyl]-4-methylbenzamide (AZD4877), a
kinesin spindle protein inhibitor and potential anticancer agent. Journal of Medicinal
Chemistry, 54, 6734–6750. https://doi.org/10.1021/jm200629m
Turner, J., Anderson, R., Guo, J., Beraud, C., Fletterick, R., & Sakowicz, R. (2001). Crystal
structure of the mitotic spindle kinesin Eg5 reveals a novel conformation of the neck-
linker. Journal of Biological Chemistry, 276, 25496–25502.
https://doi.org/10.1074/jbc.M100395200
Veber, D. F., Johnson, S. R., Cheng, H., Smith, B. R., Ward, K. W., & Kopple, K. D. (2002).
Molecular properties that influence the oral bioavailability of drug candidates. Journal
of Medicinal Chemistry, 45, 2615–2623. https://doi.org/10.1021/jm020017n
Vijayakrishnan, R. (2009). Structure-based drug design and modern medicine. Journal of
Postgraduate Medicine, 55, 301–304. https://doi.org/10.4103/0022-3859.58943
Wakui, H., Yamamoto, N., Kitazono, S., Mizugaki, H., Nakamichi, S., Fujiwara, Y., …
Tamura, T. (2014). A phase 1 and dose-finding study of LY2523355 (litronesib), an Eg5
inhibitor, in Japanese patients with advanced solid tumors. Cancer Chemotherapy and
Pharmacology, 74, 15–23. https://doi.org/10.1007/s00280-014-2467-z
Wang, F., Good, J. A. D., Rath, O., Kaan, H. Y. K., Sutcliffe, O. B., MacKay, S. P., &
Kozielski, F. (2012). Triphenylbutanamines: Kinesin spindle protein inhibitors with in
vivo antitumor activity. Journal of Medicinal Chemistry, 55, 1511–1525.
https://doi.org/10.1021/jm201195m
Woessner, R., Tunquist, B., Lemieux, C., Chlipala, E., Jackinsky, S., Dewolf, W., … Walker,
D. (2009). ARRY-520, a novel KSP inhibitor with potent activity in hematological and
2
6

taxane-resistant tumor models. Anticancer Research, 29, 4373–4380.
Wordeman, L. (2010). How kinesin motor proteins drive mitotic spindle function: Lessons
from molecular assays. Seminars in Cell and Developmental Biology, 21, 260–268.
https://doi.org/10.1016/j.semcdb.2010.01.018
Wu, G., Robertson, D. H., Brooks, C. L., & Vieth, M. (2003). Detailed analysis of grid-based
molecular docking: A case study of CDOCKER - A CHARMm-based MD docking
algorithm. Journal of Computational Chemistry, 24, 1549–1562.
https://doi.org/10.1002/jcc.10306
Yang, Y. S., Yang, B., Zou, Y., Li, G., & Zhu, H. L. (2016). Design, biological evaluation
and 3D QSAR studies of novel dioxin-containing triaryl pyrazoline derivatives as
potential B-Raf inhibitors. Bioorganic and Medicinal Chemistry, 24, 3052–3061.
https://doi.org/10.1016/j.bmc.2016.05.012
Yi, H., Kaan, K., Major, J., Tkocz, K., Kozielski, F., & Rosenfeld, S. S. (2013). “ Snapshots ”
of ispinesib-induced conformational changes in the mitotic kinesin Eg5. Journal of
Biological Chemistry, 288, 18588–18598. https://doi.org/10.1074/jbc.M113.462648
Yin, Y., Sun, H., Xu, J., Xiao, F., Wang, H., Yang, Y., … Wang, L. (2015). Kinesin spindle
protein inhibitor SB743921 induces mitotic arrest and apoptosis and overcomes imatinib
resistance of chronic myeloid leukemia cells. Leukemia & Lymphoma, 56, 1813–1820.
https://doi.org/10.3109/10428194.2014.956319
Yokoyama, H., Sawada, J., Katoh, S., Matsuno, K., Ogo, N., Ishikawa, Y., … Asai, A.
(2015). Structural basis of new allosteric inhibition in kinesin spindle protein Eg5. ACS
Chemical Biology, 10, 1128–1136. https://doi.org/10.1021/cb500939x
Zhao, H., & Caflisch, A. (2015). Molecular dynamics in drug design. European Journal of
Medicinal Chemistry, 91, 4–14. https://doi.org/10.1016/j.ejmech.2014.08.004
Figure Legends
Figure 1.The training set employed for the generation of pharmacophore models.
Figure 2. KSP inhibitors under different stages of clinical development.
Figure3. Pictorial map of KSP representing the three major domains (motor, stalk, and tail).
Figure4. Three-dimensional (3D) arrangement of KSP (ribbon, coloured in grey), inhibitor
(SB-743921; magenta sticks) and ADP (shaded in blue) in the crystal structure (PDB ID:
2
7

4
BXN).
Figure 5. Novel KSP inhibitors designed through molecular modeling approaches.
Figure 6.KSP for molecular docking (A) and location of the binding-site sphere (B).
Figure 7. (A) Best pharmacophore model (Hypo1) and (B) Mapping of 2 upon Hypo1.
Figure 8.Flowchart of database screening, drug-like filtrations, molecular docking and
molecular dynamics in identifying novel KSP hits.
Figure 9. The orientation of the docked pose (pink coloured ball and stick model) with the
cognate ligand (yellow coloured ball and stick model) in the binding-site of KSP.
Figure 10. Interactions of MB-41570 with KSP (Maybridge database).
Figure 11. Interactions of CB-10358 with KSP (ChemBridge database).
Figure 12.Top-ranked hits from each database after molecular docking/MD-MMGBSA.
Figure 13. Root mean square deviation (RMSD) for the complexes of 4BXN-SB743921
(black), 4BXN-MB41570 (red) and 4BXN-CB10358 (green).
Figure 14. Root mean square fluctuations (RMSF) of C-α atoms for 4BXN-SB743921
black), 4BXN-MB41570 (red) and 4BXN-CB10358 (green) complexes.
(
Figure 15. Radiation of gyration (RoG) for the complex of 4BXN-SB743921 (black), 4BXN-
MB41570 (red) and 4BXN-CB10358 (green).
Scheme 1: Synthetic route for the synthesis of Maybridge hit: Reagents and conditions (a) S,
diethylamine, EtOH, 60-70 °C, 2h; (b)ClCH COCl, N,N-DMF, rt, 2h; (c) N-methyl-
2
piperazine, THF, 50-60 °C, 2 h.
Table Legends
Table 1. Description of common feature pharmacophore models.
Table 2. A training set of molecules for the generation of ligand-based pharmacophore
model.
Table 3. Validation of pharmacophore model (Hypo1) using test set.
Table 4.Statistical analysis of top-ranked pharmacophore model (hypo1).
Table 5. MM/GBSA binding free energies of a reference molecule and the best-docked hits in
complexation with kinesin spindle protein.
2
8

Table 6. Post-dynamics analysis of a reference molecule and the hits in complexation with
kinesin spindle protein.
2
9