Design, computational studies, synthesis and biological evaluation of thiazole-based molecules as anticancer agents

Article abstract of DOI:10.1016/j.ejps.2019.04.005

Background: Abolition of cancer warrants effective treatment modalities directed towards specific pathways dysregulated in tumor proliferation and survival. The antiapoptotic Bcl-2 proteins are significantly altered in several tumor types which position them as striking targets for therapeutic intervention. Here we designed, computationally evaluated, synthesized, and biologically tested structurally optimized thiazole-based small molecules as anticancer agents. Methods: The virtually designed 200 molecules were subjected to rigorous docking and in silico ADME-Toxicity studies. Out of this, 23 skeletally diverse thiazole-based molecules which passed pan assay interference compounds (PAINS) filter and were synthetically feasible were synthesized in 3 steps using cheap and readily available reagents. The molecules were in vitro evaluated against Bcl-2-Jurkat, A-431 cancerous cell lines and ARPE-19 cell lines. Molecular Dynamics (MD) simulation studies were performed to analyse conformational changes induced by ligand 32 in Bcl-2. Flow cytometry analysis of compound 32 treated Bcl-2 cells was done to check apoptosis. Results: The molecules exhibited appreciable interactions with Bcl-2 and were having acceptable drug like properties as tested in silico. The multi step synthesis yielded 23 skeletally diverse thiazole-based molecules in up to 80% yield. The molecules simultaneously inhibited Bcl-2 Jurkat cells in vitro without causing detectable toxicity to normal cells (ARPE-19 cells). Among them molecules 32, 50, 53, 57 and 59 showed considerable activities against Bcl-2 Jurkat and A-431cell lines at concentrations ranging from 32–46 μM and 34–52 μM, respectively. The standard doxorubicin exhibited IC₅₀ in Bcl-2 Jurkat and A-431cell lines at 45.87 μM and 42.37 μM, respectively. The molecule 32, almost equipotent in both the cell lines was subjected to molecular dynamics (MD) simulation with Bcl-2 protein (4IEH). It was shown that 32 interacted with protein majorly via hydrophobic interactions and few H-bonding interactions. Fluorescence-activated cell sorting (FACS) analysis established that molecule is dragging cancerous cells towards apoptosis. Discussion and conclusion: The chemical intuition was checked by computation coupled with biological results confirmed that thiazole-based hits have the potential to be developed downstream into potent and safer leads against antiapoptotic Bcl-2 cells.

Full text of DOI:10.1016/j.ejps.2019.04.005

Accepted Manuscript
Design, computational studies, synthesis and biological
evaluation of thiazole-based molecules as anticancer agents
AnuRadha, Sagar Kumar Patel, Raj Kumar Patle, Preethi
Parameswaran, Alok Jain, Amit Shard
PII:
DOI:
Reference:
S0928-0987(19)30143-5
https://doi.org/10.1016/j.ejps.2019.04.005
PHASCI 4900
To appear in:
European Journal of Pharmaceutical Sciences
Received date:
Revised date:
Accepted date:
22 November 2018
27 March 2019
2 April 2019
Please cite this article as: AnuRadha, S.K. Patel, R.K. Patle, et al., Design, computational
studies, synthesis and biological evaluation of thiazole-based molecules as anticancer
agents, European Journal of Pharmaceutical Sciences, https://doi.org/10.1016/
j.ejps.2019.04.005
This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.

ACCEPTED MANUSCRIPT
Design, Computational Studies, Synthesis
and Biological Evaluation of Thiazole-
based Molecules as Anticancer Agents
AnuRadhaaΩ, Sagar Kumar PatelaΩ, Raj Kumar patlea, Preethi Parameswarana, Alok Jainband Amit Sharda
a Department of Medicinal Chemistry, National Institute of Pharmaceutical Education and Research
Ahmedabad, Opposite Air Force Station, Gandhinagar, Gujarat 382355, India
b Department of Biotechnology, National Institute of Pharmaceutical Education and Research Ahmedabad,
Opposite Air Force Station, Gandhinagar, Gujarat 382355, India
Ω-Both authors have contributed equally
ABSTRACT
Background: Abolition of cancer warrants effective treatment modalities directed towards
specific pathways dysregulated in tumor proliferation and survival. The antiapoptotic Bcl-2
proteins are significantly altered in several tumor types which position them as striking targets
for therapeutic intervention. Here we designed, computationally evaluated, synthesized, and
biologically tested structurally optimized thiazole-based small molecules as anticancer agents.
Methods: The virtually designed 200 molecules were subjected to rigorous docking and in
silico ADME-Toxicity studies. Out of this, 23 skeletally diverse thiazole-based molecules
which passed pan assay interference compounds (PAINS) filter and were synthetically
feasible were synthesized in 3 steps using cheap and readily available reagents. The molecules
were in vitro evaluated against Bcl-2-Jurkat, A-431 cancerous cell lines and ARPE-19 cell
1

ACCEPTED MANUSCRIPT
lines. Molecular Dynamics (MD) simulation studies were performed to analyse
conformational changes induced by ligand 32 in Bcl-2. Flow cytometry analysis of
compound 32 treated Bcl-2 cells was done to check apoptosis.
Results: The molecules exhibited appreciable interactions with Bcl-2 and were having
acceptable drug like properties as tested in silico. The multi step synthesis yielded 23
skeletally diverse thiazole-based molecules in up to 80% yield. The molecules
simultaneously inhibited Bcl-2 Jurkat cells in vitro without causing detectable toxicity to
normal cells (ARPE-19 cells). Among them molecules 32, 50, 53, 57 and 59 showed
considerable activities against Bcl-2 Jurkat and A-431cell lines at concentrations ranging from
32-46 µM and 34-52 µM, respectively. The standard Doxorubicin exhibited IC50 in Bcl-2
Jurkat and A-431cell lines at 45.87 µM and 42.37 µM, respectively. The molecule 32, almost
equipotent in both the cell lines was subjected to molecular dynamics (MD) simulation with
Bcl-2 protein (4IEH). It was shown that 32 interacted with protein majorly via hydrophobic
interactions and few H-bonding interactions. Fluorescence-activated cell sorting (FACS)
analysis established that molecule is dragging cancerous cells towards apoptosis.
Discussion and conclusion:
The chemical intuition was checked by computation coupled with biological results confirmed
that thiazole-based hits have the potential to be developed downstream into potent and safer
leads against antiapoptotic Bcl-2 cells.
Keywords: Anticancer, Bcl-2 Jurkat cells, Docking, Molecular Dynamics Simulations and
Thiazole.
Abbreviations:Bcl-2: B cell lymphoma-2, ADMET: Absorption, Distribution, Metabolism,
Excretion and Toxicity, MD: Molecular Dynamics.
2

ACCEPTED MANUSCRIPT
1. Introduction
Cancer is a multifactorial, multigenic and multiple pathway directed disease (Singh et al.,
2016; Xu et al., 2016). Taking into consideration the notoriously convoluted biology and
epidemiology of cancer it is an immensely alarming situation that only a meagre
concentration of new drug leads were selected for clinical trials (Lu et al., 2017). Although
several proteins and hallmarks of cancer are reported, the protein B cell lymphoma (Bcl-2) is
overexpressed in several cancers and its overexpression helps recalcitrant cancer cells
becoming immortal (antiapoptotic) and resistant (Frenzel et al., 2009). This also means that
Bcl-2 functions as a potent anti-death molecule (Akl et al., 2014). Henceforth Bcl-2 has
surfaced as potential molecular target for the design of an entirely novel class of skeletally
diverse anticancer drugs aimed to drag the cancerous cells towards apoptosis and overcoming
the chemoresistance (Azmi et al., 2011; Ferrao et al., 2015); (Perini et al., 2018). The only
FDA approved drug approved as Bcl-2 inhibitor is venetoclax (ABT-199) (Bodur and
Basaga, 2012; Sleebs et al., 2013; Souers et al., 2013) (Figure S1, Supplementary
Information) which is chemically obese and involves multistep tedious synthesis that
escalates the cost and time of synthesis (Ali et al., 2015; Huber et al., 2014; Liu et al., 2014).
ABT-737 and navitoclax (ABT-263) are novel inhibitors of Bcl-2, Bcl-xL and Bcl-w in
clinical trials but yet to hit the market. These are flanked with side effects, so the discovery of
novel, safer and small molecule-based Bcl-2 inhibitor candidates should be endorsed to
replenish drying drug discovery pipeline.
Several heteroaromatic rings have been employed in anticancer drugs out of which thiazole
ring has benchmarking presence in several anticancer drugs such as epothilones, ixabepilone,
bleomycin, tiazofurin, dasatinib, and Kud 773 (Figure 1) (T Chhabria et al., 2016).
3

ACCEPTED MANUSCRIPT
Literature also reports that thiazole-based molecules inhibit Bcl-2 and may drag cancerous
cells towards apoptosis (Romagnoli et al., 2012). The derivatives may stall progression,
proliferation and vasculature of cancer cells through plethora of mechanisms and
simultaneous action on multiple therapeutic targets. Therefore functionally diverse set of
molecules inspired by thiazole-based scaffold may be synthesized (Leoni et al., 2014). The
two hetero atoms sulphur and nitrogen having lone pair of electrons which may interact via
hydrogen bonding, with amino acid residues of receptor proteins and the embedded electron
cloud may be stacked/intercalated with aromatic residues of target proteins (Ayati et al.,
2015; Nogrady and Weaver, 2005). All these features of interaction may guide the cancerous
cells towards apoptosis.
Considering the significance of apoptosis, small molecules with novel mode of action will
facilitate design of additional apoptotic regulators (Palchaudhuri et al., 2015; Wang et al.,
2017). It is therefore clear that thiazole-based heterocycles are ubiquitous in bioactive
molecules, show impressive functional versatility and more akin to biology (Figure 1).
Figure 1. Anticancer molecules containing thiazole core.
4

ACCEPTED MANUSCRIPT
However it’s not only thiazole scaffold that lends anticancer activity but the interplay of
appropriately positioned atoms in 3-D space which leads to drug like molecules (Duran-
Frigola et al., 2013). In this paradigm carbonitrile substituted phenylamino-thiazoles were
synthesized. It is reported that incorporation of a NH linkage between aromatic and
heteroaromatic rings lead to synthesis of potent water soluble compounds (Lu et al., 2011).
The nitrile group is electrophilic towards free nucleophiles and glutathione, generally found
elevated in cancerous cells (Fleming et al., 2010). Several crystal structures of nitrile
containing drugs with receptors have showed that nitrile group keeps itself projecting into
narrow clefts to establish polar interactions or form hydrogen bonds in sterically congested
environments (Lu et al., 2011).
To foresee how a molecule interacts with a receptor of interest computational tools are
gaining lot of momentum. They are becoming increasingly important to provide useful
predictions that match experimental results (Bartocci and Lió, 2016) and provide new
insights. Molecular modeling has become a valuable and integral part of investigating, and
predicting the properties of small molecules as potential drug candidates (Ferreira et al.,
2015). In this study, potential binding interactions between synthesized compounds and anti-
apoptotic Bcl-2 were explored by in-silico techniques such as molecular docking and
molecular dynamics (MD) simulation studies. Herein, we describe our efforts for the
synthesis of thiazole-based novel molecules through a multistep synthetic approach and
discovery of promising anticancer hits. Firstly, the thiazole- based molecules designed by
intuition were made to pass through rigorous in-silico filters like docking and ADMET
studies. The molecules showing crucial interactions and docking scores were synthesized.
The synthesized compounds were evaluated for their potential anti-cancer activities centred
on Bcl-2 Jurkat and A-549 cell lines. It was found that five compounds were potent inhibitors
5

ACCEPTED MANUSCRIPT
of Bcl-2 cells. To check what conformational changes the molecules are inducing a detailed
MD simulation study was performed. It was found that ligand 32 was inducing functionally
important conformational changes involving helices α3, α4 and α5, in the antiapoptotic Bcl-2
typically involved in onset of apoptosis in cells. It is pertinent to mention that 32 was chosen
for simulations and fluorescence-activated cell sorting (FACS) studies as it was almost
equipotent in both the cell lines.
It is also important to state that the biological activity of molecules must be the inherent
property of molecule and it should not be a me too drug (Garattini, 1997). To ascertain this it
has been suggested that Pan-assay interference compounds (PAINS) cause false positive
assay signals due to reactivity under assay conditions (Baell and Holloway, 2010).To check
whether the compounds fall under this category, they were rapidly screened using
computational software which ascertained that molecules have potential to be modified into
drug like leads (Baell and Holloway, 2010). It’s also pertinent to mention that recently one of
our promising molecule of the same series was subjected for bio-analytical method
development and it was found that it has high plasma protein binding affinity which has been
determined by analyzing the plasma samples (Biswas et al., 2018). The molecule showed
pKa, LogP and LogD values which are favorable towards its further exploitation as a lead
candidate for generation of effective of structure activity relationship (SAR). We established
logP, LogD7.4, pKa and plasma protein of this molecule as 3.01, 2.96, 11 and ˃90%.
Therefore, this novel small molecule library restores potential to be explored further as
anticancer agents (Biswas et al., 2018).
2. Materials and method
2.1. Molecular modeling
6

ACCEPTED MANUSCRIPT
The virtually designed molecules were in-silico evaluated by molecular docking and were
docked with the protein (PDB ID: 4IEH) B cell leukemia/lymphoma (Bcl-2). Molecular
docking of virtually synthesized molecules was performed using Schrödinger software
(Maestro 11.1.011) (Tourꢀ et al., 2013). MD simulation of protein-ligand-32 complex was
carried out from molecular docking structure using GROMACS simulation suit (Abraham et
al., 2015). It is important to state that initially large library of two hundred molecules was
considered for virtual design and finally we narrowed down to synthesis of 23 molecules with
considerable dock scores and their synthetic feasibility.
2.1.1. Molecular docking
2.1.1.1.Protein preparation: Bcl-2 protein (PDB ID: 4IEH) structure was retrieved from
protein data bank (Berman et al., 2000; Goodsell et al., 2015) (https://www.rcsb.org) having
resolution of 2.1 Å (Toure et al., 2013). We selected the structure (PDB ID: 4IEH) based on
best available resolution, R (Free) value and number of residues resolved. Missing loop
(residues 32-48) was modelled by SWISS-MODEL (Schwede et al., 2003). Missing N
(residues 1-8) and C-terminus (residues 164-166) residues were not modelled. Similar
procedure was adopted in previously published study for a protein from same family (Lama
and Sankararamakrishnan, 2008). Initial coordination of Bcl-2 with ligands was analysed
using Schrödinger’s Protein preparation wizard panel (Schrödinger, LLC, New York, NY,
USA). Protein exists as monomeric form so without further modification of protein
crystallized water molecules were removed and kept ligand as such 1E9 (201). OPLS3
(Harder et al., 2015) force field was used for protein energy minimization using protein
preparation wizard module of the Schrödinger. After protein preparation 10 Å grid was
generated around the ligand 1E9 (201) using receptor grid generation of Glide module and
applied standard parameter of Glide tool (Schrödinger, LLC, New York, NY, USA).
7

ACCEPTED MANUSCRIPT
2.1.1.2. Ligand preparation: Two hundred designed ligands were drawn in Maestro 2D
Workspace (Schrödinger, LLC, New York, NY, USA). All designed ligands as well as the
standard doxorubicin were prepared using LigPrep panel. Ionization states were generated at
pH 7.0 +/-2.0. Firstly we attempted to generate tautomers. Further for each tautomer we tried
to generate at maximum thirty two stereoisomers using OPLS3 force field.
2.1.1.3. Ligand docking: Receptor grid and prepared ligands were subjected to Glide XP
docking using standard protocol of Glide (Schrödinger, LLC, New York, NY, USA).
Standard parameters were applied including van der Waals radii scaling factor 0.80 for
nonpolar atoms with a partial charge cut off of 0.15. The number of poses to be reported is
ten poses per ligand. Per residue interaction scores were calculated for residues within 5 Å of
grid center.
2.2. ADMET property predictions: QikProp module of the Schrödinger software (Maestro
11.1.011) was used for predicting the pharmacokinetic and metabolic profiles of 23 screened
molecules based on dock scores and their synthetic feasibility. All these ligands were
subjected to QikProp module. A total of 16 molecular descriptors were used for calculating
the drug metabolism and pharmacokinetic (DMPK) properties. A total of 16 molecular
descriptors were used for calculating the drug metabolism and pharmacokinetic (DMPK)
properties (Detailed in Supplementary Information). Finally, these molecules were taken
further for synthesis and synthetic strategy was postulated as follows.
2.3. Synthetic procedure for the preparation of compounds
2.3.1. Preparation of
aromatic isothiocyantes (Ar-NCS): Synthesis of
isothiocyanatobenzene (8):Solution of aniline (1) (1 equiv.) and triethylamine (3.2 equiv.) in
tetrahydrofuran was cooled to 0-5 °C with continuous stirring. Carbon disulphide (1.1 equiv.)
was added to this cold reaction mixture and allowed to stir for 6 h. The completion of the
8

ACCEPTED MANUSCRIPT
reaction (formation of salt) was checked by TLC in petroleum ether:ethyl acetate (9:1)
mobile phase system. Dithiocarbamate salt was filtered, washed with tetrahydrofuran and
dried. Then it was dissolved in chloroform and triethylamine (1 equiv.) was added, followed
°
by addition of ethylchloroformate (1.1 Equiv.) at 0-5 C. After stirring for 1.5 h, 3M
hydrochloric acid solution was added to the reaction mixture and it was allowed to stir for
further 10 min. Then organic layer was separated, dried over anhydrous sodium sulphate and
concentrated under vacuum to yield oily liquid. At last, it was purified by column
chromatography (Yield: 90%) (HODGKINS and ETTLINGER, 1956).
2.3.2. Preparation of tetramethylguanidine (TMG) adduct: Synthesis of
isothiocyanatobenzene-tetramethylguanidine intermediate (18): To the stirred solution of
isothiocyanatobenzene (8) (1 Equiv.) dissolved in minimum amount of ethyl acetate, then
1,1,3,3 tetramethylguanidine (17) (1 Equiv.) was added very slowly by drop wise at room
temperature. Stirring was continued for 2 h. After completion of reaction the precipitate was
filtered, the final product was dried and weighed (Yield: 80%)(Rajappa et al., 1982).
2.3.3. Preparation of final products: 4-(dimethylamino)-2-(phenylamino)thiazole-5-
carbonitrile) (27): To the stirred solution of isothiocyanatobenzene-tetramethylguanidine
intermediate (8) (1 Equiv.) dissolved in minimum amount of dimethylformamide, then
chloroacetonitrile (26) (1 Equiv.) was added drop wise. The reaction was continued for 15 h
at room temperature. Completion of reaction was checked by TLC. After completion of
reaction, the reaction mixture was poured in ice and precipitates were filtered off and dried,
which was further purified by column chromatography (Yield: up to 90%). Rest all the
derivatives were synthesised by aforementioned procedure(Rajappa et al., 1982).
2.4. In vitro studies
9

ACCEPTED MANUSCRIPT
2.4.1. In-vitro cytotoxicity evaluation:Cell culture of Bcl-2 Jurkat cells: Bcl-2 Jurkat cells
were cultured at 37 ºC in humidified CO2 incubator with 5% CO2 and 95% air in the RPMI-
1640 containing 10% Fetal Bovine Serum (FBS), and 200 µg/mL geniticin (G418). The cells
were maintained by addition of fresh medium or replacement of medium.1×104 cells per well
were plated in 96 well plate followed by addition of freshly prepared different doses (1, 10,
25 and 50 μM, respectively) of synthesized compounds in triplicates. The cells were also
treated with positive control (doxorubicin hydrochloride) and incubated for 24 h. Although
several mechanisms have been reported for the doxorubicin induced apoptosis (Huigsloot et
al. 2002) it is reported that it induces mitochondrial-dependent apoptosis by down-regulation
of Bcl-xL and up- regulation of Bax (Leung and Wang, 1999; Sharifi et al., 2015). At the end
of the incubation time, cells were incubated further for 4 h with 100 μl (1 mg/mL) of 3-(4,5-
Dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT). The plates were
centrifuged and media was removed. Then DMSO (100 μl) was added and plates were shaken
for 10 minutes so as to ensure complete dissolution of the formazan crystals. Absorbance
values were then recorded at 570 nm using the microplate spectrophotometer system. Cell
culture of ARPE-19 cells and A431cells: ARPE-19 cells were cultured at 37 ºC in humidified
CO2 incubator with 5% CO2 and 95% air in the DMEM: F12 containing 10% Fetal Bovine
Serum (FBS), 100 IU/mL penicillin and 100 IU/mL streptomycin. The cells were maintained
by addition of fresh medium or replacement of medium. 1×104 cells per well were plated in
96 well plate and incubated for 24 h. Twenty hours later, cells were treated with different
doses (1, 10, 25 and 50 μM) of synthesized compounds in triplicate, respectively. The cells
were also treated with positive control (doxorubicin hydrochloride,) and incubated for 24 h.
At the end of the incubation time, cells were incubated further for 4 h after addition of 100 μL
(1 mg/mL) of MTT. Media was removed and DMSO (100 μL) was added. Plates were
allowed to gentle shaking for 10 minutes so as to ensure complete dissolution of the formazan
10

ACCEPTED MANUSCRIPT
crystals. Absorbance values were then recorded at 570 nm using the microplate
spectrophotometer system. % cell viability was calculated by following equation % Cell
viability = At/Ac) x 100, where At: absorbance values of test compound, Ac: absorbance
values of control (Ayalew et al., 2017).
2.5. Molecular dynamics (MD) simulation
Further molecular dynamics simulation of anti-apoptotic Bcl-2 protein (PDB ID: 4IEH) with
designed ligand was carried out. Classical MD simulation was performed with the
GROMACS molecular modeling simulation package 2018.3 using the grooms 54a7 force
field (Abraham et al., 2015). Initial protein-ligand complex structure was directly taken from
the final docked structures as discussed above. Force-field parameters for compound 32 was
generated
by
the
Automated
Topology
Builder
(ATB)
(http://compbio.chemistry.uq.edu.au/atb/)(Malde et al., 2011; Stroet et al., 2018). The starting
structure was solvated using the SPC water model (Schmid et al., 2011). Protein was placed
in the centre of the cubical box where the distance between any atom of the solute and the
edge of the solvent box was at least 1.5 nm. Counter ions were added to neutralize the
system. Solvated, system was energy minimized using steepest descent algorithm. Next the
system was subjected to two short equilibration runs. In the first equilibration run, all the
heavy atoms of the protein complex were position restrained and simulation was performed
for 100 ps in the constant-temperature, constant-volume ensemble (NVT). In the subsequent
equilibration set up system was subjected to short 100 ps run in the constant-temperature,
constant-pressure ensemble (NPT) setting. Finally a 100 ns production run was performed in
the NPT ensemble at 300 K coupled with the velocity-rescale thermostat (Berendsen et al.,
1981)with a coupling constant 0.1 ps. Parrinello-Rahman coupling algorithm(Bussi et al.,
2007)was used for keeping the pressure constant at 1 bar with a coupling constant of 2 ps.
11

ACCEPTED MANUSCRIPT
Long range non-bonded interactions in the systems were evaluated using the Particle Mesh
Ewald method (Parrinello and Rahman, 1981) with a real-space cut-off of 1.4 nm van der
Waals interactions were evaluated up to 1.4 nm. Periodic boundary conditions were applied
in all three (x,y,z) directions. The LINCS algorithm (Essmann et al., 1995) was used to
constrain the bonds involving hydrogen atoms. Equations of motion for the water molecules
were solved analytically with the SETTLE algorithm (Hess et al., 1997).
Binding free energy calculations: Binding free energy for the complex was calculated from
the Molecular Mechanics Poisson−Boltzmann Surface Area (MM-PBSA) method(Homeyer
and Gohlke, 2012). Binding free energy of the protein with ligand in solvent are expressed as
follows:
ΔGbinding= Gcomplex− (Gprotein+ Gligand) …………………….(Equation I)
where, ΔGbinding
=
, Gcomplex = energy of complex, Gprotein = energy of
free energy of binding
protein, Gligand= energy of ligand
………………………………(Equation II)
Gx= EMM + Gsolvation-TSMM
Where, x = receptor or ligand or receptor-ligand complex, EMM = Molecular mechanics
potential energy in vacuum, Gsolvation = free energy of solvation, TS Entropic contribution to
the free energy in vacuum, T and S represent the temperature and entropy
EMM = Ebonded+ Enon-bonded= Ebonded+ (Evwd+ Eelec)…………(Equation III)
Where EMM is bonded and non-bonded interactions energy, Ebonded = bonded energy, Enon-
= non-bonded energy, Eelec = electrostatic energy, Evdw = van der Waals interactions
bonded
energy
………………………………(Equation IV)
Gsolvation = Gpolar + Gnonpolar
12

ACCEPTED MANUSCRIPT
Where, Gpolar= free energy of polar interactions, Gnonpolar = free energy of non-polar
interactions
MM-PBSA approach has been extensively employed to study biological relevant systems
(Rastelli et al., 2010; Weis et al., 2006). We have used g_mmpbsa tools to compute the
binding
free
energy
on
the
ensemble
of
structures
generated by gromacs (Kumari et al., 2014). The current implementation of the MMPBSA
within g_mmpbsa does not include the contribution of the entropic term. It is shown
previously that the net entropy contribution is often small and exhibits large variation
(standard error) compared with other energetic term mentioned above. The free energy of
binding was calculated for full 100-ns simulation time using the structures saved at every 100
ps interval.
2.6. Flow Cytometric analysis
Further to ascertain whether 32 is dragging cancerous cells towards apoptosis, 1×105 Jurkat
cells per well incubated in 6 well plate having RPMI 1640 medium. Cells are treated with
ligand 32 and doxorubicin respectively at their respective IC50 of 34.77 µM and 45.87 µM
followed by incubation in 5% CO2 at 37 oC for 24 h. Preparation of sample and stock solution
was followed by standard protocol of Invitrogen Fluorescein isothiocyanate (FITC) annexin
V/ dead cell apoptosis kit with FITC annexin V and propidium iodide (PI) for flow cytometry
(Darzynkiewicz et al., 1997).
3. Results and discussion
3.1 Molecular docking
13

ACCEPTED MANUSCRIPT
Figure 2. Representative structure is showing the final docked structure of 32 with Bcl-2 (PDB ID:
4IEH). Protein is displayed in ribbon representation. Ligand 32 is shown in ball and stick
representation. Modeled loop is shown in purple colour. All the helices and termini are marked based
on the X-ray structure. Carbon, nitrogen, sulphur and hydrogen atoms are shown in green, blue,
yellow and white colour, respectively.
We have first validated the docking protocol by redocking the originally complexed ligand
(PDB ID: 4IEH) to its binding site. It docked perfectly well with rmsd value.of 0.53 Ao
compared with the X-ray structure. All the virtually designed 200 compounds were docked
using the protocol discussed in the method section. More than 150 compounds were
discarded based on the docking score (< -2.9 kcal/mol). Some of the compounds were not
considered because of lack of synthetic feasibility. Finally, we selected 23 compounds based
14

ACCEPTED MANUSCRIPT
on the docking score and feasibility of synthesis. It is important to mention that we have
thoroughly searched for other potential binding sites using the sitemap module of
Schrödinger and found two more possible binding pockets. However, our docking results
suggest that BH3 binding groove that is majorly formed by BH1, BH2 and BH3 regions
(Petros et al., 2004) gives the best score compared with the other putative binding sites for
our ligand. This also confirms that our designed ligand binds to the same groove where both
pro and anti apoptotic protein/peptide [See Supplementary Information Figure S6]
Molecular docking data suggests that the designed molecules form pi-pi interactions with the
Bcl-2 residues F63 and W67 and H-bonds with residues R66 and G104 (Table 1). Docking
pose of 32 is a representative example of the interactions of the thiazole-based ligands with
the Bcl-2 (Figure 2).
Table 1. Representative synthesized molecules with their IC50 values against Bcl-2 Jurkat cell
and A-431 cell lines and key interacting residues of Bcl-2 according to the docking study.
IC50 (µM)
Jurkat
Bcl-2
IC50
(µM)
A-431
Key interactions of Bcl-2
residues at the ligand binding
site
Compound
no
Xp G-score
(kcal/mol)
59.38
30.69
31.25
32.70
200.80
181.15
61.19
-4.403
-5.447
-5.423
-5.404
π-π: PHE63
π-π: PHE63, H bond: ARG105
π-π: TYR161
27
28
29
30
71.63
π-π: TYR161
15

ACCEPTED MANUSCRIPT
39.43
34.77
32.59
64.43
46.72
33.99
N.T
N.T
34.31
91.41
N.T
-3.165
-4.151
-5.386
-2.902
-4.995
-4.195
-4.413
-5.386
-5.507
-5.455
-3.882
-5.289
π-π: PHE63, H bond: ASP70
H bond: GLY104
31
32
33
34
50
51
52
53
54
55
56
57
π-π: PHE63
H bond: ALA59
44.64
109.17
67.65
40.98
180.50
69.44
70.42
39.77
π-π: PHE63, TYR161
π-π: PHE63
H bond: ARG105
39.15
50.86
42.80
N.T
π-π: PHE63, H bond: ARG105
π-π: TYR161
π-π: PHE63, H bond: ARG105
π-π: TYR161
45.99
π-π: PHE63, H bond: ARG105
16

ACCEPTED MANUSCRIPT
N.T
32.44
N.T
35.63
52.52
168.9
89.09
28.45
158.73
24.75
-5.347
-3.213
-5.341
-3.932
-3.005
-5.136
-4.974
π-π: PHE63, H bond: ARG105
π-π: PHE63, H bond: ARG105
H bond: ARG105,
58
59
60
61
62
67
68
41.56
62.03
46.25
57.42
No interaction
H bond: ALA59
π-π: PHE63, H bond: ARG105
π-π: PHE63, H bond: ARG105
Salt bridge:ASP99, H bond:
Glu-95, H bond: Asp70
45.87
42.37
-7.631
Doxorubicin
N.T = Not tested due to precipitation in the media
The computational studies suggest that the compounds would interact in a meaningful
manner with the clinically significant Bcl-2 protein. Thereafter, we predicted ADMET profile
of our molecules in silico.
3.2. Results of in silico ADMET studies: Majority of drugs candidates fail in clinical trials
because of their poor ADMET profiles. The pharmacokinetic profiles supplemented with
biological activity are must for a molecule to be considered as a drug candidate. There are
various in silico tools available for predicting the pharmacokinetics of a molecule. Here,
17

ACCEPTED MANUSCRIPT
"drug-likeness" parameter was evaluated taking into consideration Lipinski's "Rule of Five"
(Ro5; Lipinski et al., 1997; Ntie-Kang, 2013). All parameters for tested molecules were well
within prescribed ranges (Detailed in Supplementary Information). These were considered as
drug like candidates and their synthesis was carried out as given in section 3.3.
3.3 Chemistry:
After the computational studies the synthetic strategy was prepared. Initially the
isothiocyanate of aniline were prepared followed by preparation of tetramethylguanidine
adduct. It was followed by reaction with chloroacetonitrile to yield products 27-34 (Scheme
1
13
1) in moderate to good yields. The compounds formed were characterized by H and C
NMR and purity was assessed by HPLC.
Scheme1. General scheme for the synthesis of N, N-dimetthylamine substituted thiazole derivatives.
To check the variability and increase the skeletal diversity in the library, the dimethylamino
moiety in thiazole ring was replaced with methyl and phenyl ring, respectively. For this
reaction acetamidine and benzamidine were synthesized and reacted with variously
substituted isothiocyanates (9-16, Scheme 2) to yield 37-49. The products upon reaction with
chloroacetonitrile yielded 50-62 in up to 90% yield. Thus to synthesize the derivatives (27-
34, Scheme 1a) diverse library of skeletally diverse anilines were chosen. To synthesize the
18

ACCEPTED MANUSCRIPT
derivatives 50-62 (Scheme 2) benzimidines and acetamidines were chosen to get distinct
products.
Scheme 2. General scheme for the synthesis of phenyl and methyl substituted thiazole derivatives.
In order to prepare amide substituted derivatives, benzoyl chloride (63)was taken which upon
reaction with ammoniumthiocyanate (NH4SCN), tetra-n-butylammonium bromide (TBAB) in
chloroform led to 64. The derivatives finally ended up with dimethylamino and phenyl
substitution at the thiazole ring to yield 67-68 (Scheme 3).
Scheme3. General synthetic scheme for the synthesis of Benzoyl substituted thiazole derivatives.
3.4. Structure activity relationship (SAR):
19

ACCEPTED MANUSCRIPT
Cytotoxicity of synthesized compounds was determined against Bcl-2 Jurkat, A-431, ARPE-
19 cell lines representing acute T-leukemia, epidermoid carcinoma and normal cell lines
(ARPE) (Supplementary Information). Cytotoxicity of synthesized compounds compared
with doxorubicin used as standard. All the molecules were evaluated using variable
concentrations ( 1, 10, 25, 50 μM) and it was found that the molecules 32, 50, 53, 57 and 59
showed potent activity against the two Bcl-2 Jurkat and A-431cancerous cell lines at lower
concentration in comparison to doxorubicin as standard.
Initially 27 was synthesized having metabolically robust CN group on thiazole ring adjacent
to N, N dimethyl amino group and IC50 of 59.38 µM and 200.00 µM against Bcl-2 Jurkat Cell
line and A-431 cell lines, respectively. To see the impact of electron donating groups on the
phenyl ring with methyl group at para position exhibited IC50 of 32.70 µM and 71.63 µM in
case of 30. Further enriching the ring A with electron donating ortho located dimethyl groups
in case of 32 exhibited IC50 of 34.77 µM and 34.31 µM. In case of 33 meta located dimethyl
groups retained activity with exhibited IC50 of 32.59 µM and 91.41 µM. Further morpholino
group was introduced to enhance the conformational flexibility in 34 that furnished IC50 of
64.43 µM, replacing the dimethylamino group by phenyl ring provided 50 with IC50 of 46.72
µM and 44.64 µM against Bcl-2 Jurkat Cell line and A-431 cell line, respectively. Replacing
N, N dimethyl amino group of 28 by phenyl provided 52 with IC50 of 67.65 µM activity
against A-431 whereas replacing N, N dimethyl by methyl in case of 53 furnished IC50 of
39.15 µM and 40.98 µM against Bcl-2 Jurkat Cell line and A-431 cell line, respectively. The
replacement of the phenyl group on thiazole of compound 54 with a methyl led to 55 with a
decreased activity. The acetamidines derived products 59 and 61 had IC50 of 32.44, 52.52 µM
and 41.56 and 89.09 µM, respectively against two cell lines. The derivatives prepared from
acetamidines 59 and 61 having IC50 32.44, 52.52 µM and 41.56 and 89.09 µM, respectively.
20

ACCEPTED MANUSCRIPT
It was found clear from the SAR that 32 is almost equipotent against Bcl-2 and A549 cells
with IC50 of 34.77 µM and 34.31 µM, respectively. The molecules spared normal cells
(ARPE-19) (detailed in Supplementary Information Figure S3), while the activity against the
cancerous cells was comparable to the standard doxorubicin. The library of molecules
synthesized and their percentage yields are presented in Figure 3.
Figure 3.Thiazole-based synthesized molecules and their percentage yields
3.5. Molecular dynamics simulation
21

ACCEPTED MANUSCRIPT
The docking complex of the thiazole based ligand 32 and Bcl-2 protein was further subjected
to nanosecond-scale molecular dynamics simulation. The study explores the possible key
interactions of our ligand 32 with Bcl-2 protein.
Figure 4. Superposed starting and final conformations of Bcl-2-ligand 32 complex displaying the
conformational change occurring during the MD simulation. Starting structure and the structure after
100 ns simulation time are displayed in tan and blue colour, respectively. Ligand 32 is displayed in
ball and stick representation showing the positional transition of ligand during the simulation. All the
helices are numbered based on the Bcl-2 X-ray structure (PDB ID: 4IEH). Nitrogen, sulphur and
hydrogen atoms are shown in blue, yellow and white colour, respectively. Carbon atoms are displayed
22

ACCEPTED MANUSCRIPT
in tan and blue colour for starting and final simulated structure, respectively. This figure and
subsequent molecular plots were created using the UCSF Chimera package (Pettersen et al., 2004)
Different structural and energy parameters such as root mean square deviation (RMSD)
profile, number of hydrogen bonds between ligand 32 and Bcl-2 protein, and binding energy
were evaluated. Backbone RMSD profile of compound 32 does not exhibit any significant
change during the course of simulation [Supplementary Information, Figure S2a]. RMSD
profile of all the backbone atoms of the protein exhibited a sharp peak around 38 ns which
suggests some conformational changes in the structure [Supplementary Information, Figure
S2b]. It is important to mention that after these conformational changes ligand 32 had gone
through positional transition and slowly adopted the same binding mode like pro apoptotic
peptides (Bim, Bad, Bax) as depicted in Figure S6. Further, to identify the ligand 32 induced
structural changes, starting and final structures were compared. A closer look of the starting
and final structures identified the significant rearrangements in the helices α3, α4 and α5. It
also revealed some minor changes in the position of helix α6 and α7 as shown in Figure 4.
We have also quantified the conformational changes by calculating the root mean square
fluctuation (RMSF) profile (Supplementary Information, Figure S7). Major contributor of the
conformational changes comes from residues 82-105. This suggested that a new H-bonding
was the driving force for the observed conformational changes. Modelled loop and termini
regions always contributed to the RMSF profile but not specific to 38 ns time interval. The
aforesaid conformational changes are shown to be functionally important as qualitatively
similar structural transition was observed after binding of anti-apoptotic peptide (Bax, Bim,
Bad) with Bcl-2 (Pang et al., 2012). In the initial complex there were one to two H-bonds
(Figure 5) between protein and ligand, largely contributed by Gly102, and partly by
Arg97/Asn102. Percentage stability of these H-bonds was quantified and is shown in Figure
6a and 6b.
23

ACCEPTED MANUSCRIPT
Figure 5. Number of hydrogen bonds between ligand 32 and Bcl-2 during the 100-ns simulation time.
Furthermore, these conformational changes not only increase the total number of hydrogen
bonding (Figure 6a and 6b) but also enhance the overall hydrophobic contacts between
protein Bcl-2 and ligand-32 (Figure 6c and 6d).
After the ligand induced conformational transition, total number of H-bonds increased to
three to four as depicted in Figure 5. During 50-100 ns simulation time these H-bonds were
quite stable and that in turn enhanced the overall complex stability.
24

ACCEPTED MANUSCRIPT
Figure 6.Overall interaction pattern of ligand 32 with the Bcl-2 protein. (a, b) H-bond interactions and
(c, d) Hydrophobic contacts are displayed. Percentage stability (in bracket) of interactions during (a,c)
0-50 ns and (b,d) 50-100 ns simulation time. Only the prominent interactions were shown for clarity.
Ligand 32 is displayed in ball and stick and interacting protein residues are shown in stick
representation. Carbon, nitrogen, oxygen, sulphur, hydrogen and colour coded as green, blue, red,
yellow and white, respectively.
MD simulation studies also found the transition of ligand from outer cavity of receptor to
interior cavity during the course of simulation (Figure 7a-7b). It is interesting to observe that
repositioning of ligand 32 with respect to Bcl-2 not only driven by new H-bonding
interactions but also the hydrophobic contacts in the complex. Percentage stability of these
contacts during the first half and second half of the simulation were compared and are
displayed in Figure 6c and 6d. Initially, ligand 32 formed the hydrophobic interactions
primarily with the residues Y161, Y67, F63 and partially with F112 and L96 (Figure 6c).
After these transitions ligand 32 break some of these contacts and formed additional new
hydrophobic interactions primarily with residues F63, Y67, F71, L96 and partially with Y161
as displayed in Figure 6d. All these residues formed a hydrophobic pocket (Figure 7) which
provides the strong binding with the hydrophobic moiety of the ligand 32.
25

ACCEPTED MANUSCRIPT
Figure 7. MD simulation studies showing ligand-32 moves from outer cavity a) of receptor at 0 ns to
the b) interior cavity of receptor at 100 ns time point. The yellow area represents hydrophobic region
formed by residues: leucine, valine, isoleucine, phenylalanine, tyrosine, tryptophan, methionine, and
alanine of the protein. The blue area reflects the positively charged residues (arginine, lysine, and
histidine), the red surface represents the negatively charged residues (aspartic acid and glutamic
acid). Remaining residues are coloured grey. Ligand is shown in ball & stick representation. Carbon,
nitrogen, oxygen, sulphur, hydrogen of the ligand 32 and colour coded as green, blue, red, yellow and
white, respectively. Positions of the residues which formed hydrophobic contacts or H-bonding are
shown.
In summary, transition of ligand 32 with respect to the protein increased the overall
electrostatic and hydrophobic interactions. Further, to quantify this effect and estimate the
MM-PBSA binding energy of protein-ligand complex g_mmpbsa tool was used as discussed
in the method section. Electrostatic, van der Waals (vdW), solvation (polar and nonpolar) and
total binding energy profile of the complex is illustrated in Figure 8.
26

ACCEPTED MANUSCRIPT
Figure 8. Binding energy profile for the Bcl-2-ligand 32 complex and the contributions from various
energy components.
As discussed above, contribution of both electrostatic and vdW energy terms was enhanced
after the repositioning of ligand 32 and conformation transition within the protein. Energy
contribution was increased from -30 to -60 kJ/mol and -114 to -124 kJ/mol for electrostatic
and vdW, respectively. However, it did not affect the overall binding energy as it was
compensated by unfavourable contribution from the polar solvent energy as shown in Figure
8. Nevertheless, overall ligand binding with protein was strong enough (-77.87 kJ/mol) to
induce the functionally important conformational changes in the protein.
The cytotoxicity of 32 is described in section 2.4.1.To check apoptosis compound 32 was
tested on Bcl-2 Jurkat cells using fluorescence-activated cell sorting (FACS) and it revealed
that it is indeed inducing apoptosis in cells as discussed below.
3.6. Flow cytometric analysis:
Phosphatidyl serine (PS), is a protein located on the inner leaflet of the plasma membrane in
healthy cells (Oancea et al., 2006). During the early phases of apoptosis, it translocates to the
outer layer of the plasma membrane where it is exposed at the external surface of the cells.
27

ACCEPTED MANUSCRIPT
Annexin V has a high affinity for PS, and fluorochrome-tagged Annexin V staining is used as
an indicator of apoptosis. The staining is done in combination with propidium iodide (PI), a
nucleic acid-specific stain that is excluded from live and early apoptotic or necrotic cells but
stains nucleic acids once the plasma membrane is disrupted. Therefore, it is possible to
distinguish live, healthy cells (negative for both Annexin V and PI) from early apoptotic cells
(Annexin V positive/PI negative) and late apoptotic or necrotic cells (positive for both
Annexin V and PI) by flow cytometry(Chen et al., 2000).
Flow cytometric analysis showed that standard drug doxorubicin is shows 2.60% cells as
propidium iodide (PI) positive after 24 hrs at 45.87µM. This means this population represents
dead cells. 97.38% of cells after 24 hrs treatment with standard drug doxorubicin at 45.87µM
are annexin positive and PI positive (Figure 9) which means late apoptosis.
Figure 9a. Flow cytometry analysis of Doxorubicin against Bcl-2 Jurkat cells after 24 h treatment.
Figure 9b. Flow cytometry analysis of Ligand 32 against Bcl-2 Jurkat cells after 24 h treatment.
In contrast compound 32 was 87.66% annexin positive, corresponding to early apoptosis
inherently associated with Bcl-2 proteins. Ligand 32 is not showing PI positive results
indicate that cells do not undergo necrotic pathway. 7.60% population of cells are showing
annexin positive and PI positive underwent cell death. Thus it was proven that ligand 32
28

ACCEPTED MANUSCRIPT
predicted to have crucial interactions with the death defying Bcl-2 protein in silico, is
engaged in setting on apoptosis in cancerous cells.The results corroborated well with in silico
studies. The in vivo examination of ligand 32 is currently under investigation and will be
published in due course of time.
4. Conclusion:
In summary, out of a large virtual library of molecules, 23 thiazole-based molecules have
been successfully synthesised in moderate to good yields. The molecules which showed
critical interactions with Bcl-2 in docking as well as qualified ADMET parameters were
biologically evaluated for their cytotoxicity against Bcl-2 Jurkat cell line (acute T-cell
leukemia) and A-431 (epidermal skin cancer) cancerous cell lines. Among them 32, 50, 53,
57 and 59 showed potent activities against Bcl-2 Jurkat and A-431cell lines at 34.77, 46.72,
39.15, 45.99, 32.44 and 34.31, 44.64, 40.98, 39.77, 52.52 µM respectively and emerged as
hits. The standard doxorubicin showed IC50 at 45.87 and 42.37 µM, respectively. In-silico
analysis suggested that synthesized molecules are showing promising interactions with Bcl-2
protein. The ligand 32-Bcl-2 complex stability was monitored during the 100 ns simulation
run. Our data suggested a strong binding (-77.87 kJ/mol) between ligand 32 and protein. The
in-silico approach also elucidated the atomistic level mechanism of ligand binding with
protein and induced conformational changes that might be functionally relevant to induce
apoptosis in cancerous cells. These observations, coupled with the reported experimental
data, suggest that ligand 32 induces formation of a tight but transient binding to the
hydrophobic groove of Bcl-2 protein which is crucial to drag them towards apoptosis. Flow
cytometry analysis also suggests compound 32 was inducing apoptosis in Bcl-2 cells. The
molecules can be optimized down the line to specifically target Bcl-2 and these may be
developed as future anticancer leads. Although under physiological conditions there may be
29