Page 106 - IMO-2-2

Innovative Medicines & Omics Modeling Aurora-B kinase inhibitors

checkpoints are the major characteristics of most tumor employed in this study, including rigorous statistical
cells. Aurora kinases are highly expressed in mitotically analysis, virtual screening, and MD simulations, provide
active cells such as the spleen, thymus, testis, bone marrow, valuable insights into the identification and optimization
intestine, and fetal liver. Homo sapiens express three of Aurora-B inhibitors, thereby accelerating the drug
2
3-5
Aurora kinase paralogues: Aurora-A, -B, and -C. All discovery process.
three kinases show 67 – 76% sequence identity. Aurora-A
6,7
primarily localizes to spindle poles and centrosomes, and 2. Materials and methods
its disruption leads to centrosome separation and mitotic 2.1. Protein preparation
errors. In contrast, Aurora-B, known as a chromosomal
8,9
passenger protein, initially resides at chromosome Ligand-based pharmacophore modeling studies were
centromeres during early mitosis and later shifts to carried out using the PHASE application implemented
microtubules during anaphase. 10,11 Aurora-B is essential for with Maestro (Phase, version 3.4, and Maestro, version 9.3,
chromosome segregation, alignment, spindle-checkpoint 2014, Schrödinger, LLC, New York, NY). 25,26 The 3D
function, and cytokinesis, interacting with partners such coordinates of the crystallographic structure of the
as inner centromere protein (INCENP), survivin, and human Aurora-B-INCENP complex (PDB ID: 4AF3) were
12
borealin. The precise function of Aurora-C in mammals downloaded from Brookhaven Protein Data Bank (www.
27
remains unclear, but it is known to complement Aurora-B rscb.com). The protein complex was pre-processed and
and aid cytokinesis. Dysregulation of Aurora-A and -B prepared using Protein Preparation Wizard in Maestro
13
28
is associated with various cancers, making Aurora-B (Schrödinger software). The minimization of the
a promising oncogene target. 14-19 Notable Aurora complex was continued using the Optimized Potential
kinase inhibitors include Hesperadin, ZM447439, and for Liquid Simulations (OPLS)–2005 force field until the
VX680, with Hesperadin and VX680 inhibiting all three root mean square deviation (RMSD) reached the value of
29
kinases. 20-22 ZM447439 selectively targets proliferating 0.3 Å. The preprocessing steps included the correction of
tumor cells during mitosis, sparing non-proliferating cells. heavy atoms, water molecules, cofactors, and metal ions,
VX680 exhibits promising results in arresting tumor cell as well as the addition of missing hydrogen atoms, side
proliferation in vivo and in animal models. chains, and protons. The Schrodinger suite’s EPIK and
IMPREF programs were employed for further refinement
In the quest to discover novel scaffold inhibitors for a and minimization. To define the active site vicinity, the
range of targets, various in silico approaches, including three- receptor grid generation module was utilized, creating
dimensional quantitative structure activity relationship a grid that encapsulated the ligand’s surroundings. The
(3D-QSAR), play a pivotal role. 23,24 For Aurora-B inhibition, resulting workspace configuration, appearing as a centroid
a ligand-based pharmacophore model was employed to in a cubic shape, precisely delineated the protein’s active
pinpoint the essential chemical features shared by effective site. The molecular docking studies of hit molecules were
inhibitors. To construct these hypotheses, a collection of conducted using Glide software (Glide, version 5.8, 2014,
known Aurora-B inhibitors was sourced from the Binding Schrödinger, LLC, New York, NY). MD simulation studies
30
Database website (www.bindingdb.org/). The selection were carried out using the DESMOND module (Desmond
of the most robust pharmacophore hypothesis was based MD System, 2014, D.E. Shaw Research, version 3.1, and
on rigorous statistical criteria, including the coefficient of Maestro–Desmond Interoperability Tools, version 3.1,
determination (R ), cross-validated correlation coefficient Schrödinger, New York, NY). 31
2
2
(Q ), standard deviation (SD), variance ratio (F), and root
mean square error (RMSE). This selected hypothesis was 2.2. Pharmacophore modeling
subsequently employed as a 3D query to initiate virtual A total of 58 known inhibitors of Aurora-B with quantitative
screening. The screened compounds then underwent
further refinement to ensure optimal pharmacokinetic bioactivity data were used to design the pharmacophore
properties, after which they were subjected to molecular model. The inhibitors were randomly divided into the
docking studies to identify the best-fit molecules. The training set and test set for structure-based pharmacophore
generation and validation. Highly active and least active
ultimate decision on lead compounds was made through
a comprehensive comparison of docking analysis and compounds were included in both training and test sets.
pharmacokinetic properties. These chosen leads were The training set ligands provided critical information for
then subjected to rigorous evaluation through molecular pharmacophore design.
dynamics (MD) simulations to elucidate their binding The pharmacophore modeling process consisted of five
mode interactions. The advanced in silico methods steps, as described in the following sections.

Volume 2 Issue 2 (2025) 100 doi: 10.36922/imo.6547