Recent Patents on Anti-Cancer Drug Discovery, 2009, 4, 000-000 1
1574-8928/09 $100.00+.00 © 2009 Bentham Science Publishers Ltd.
Basis for Resistance to Imatinib in 16 BCR-ABL Mutants as Determined
Using Molecular Dynamics
Tai-Sung Lee
1*
, Steven J. Potts
2
and Maher Albitar
3
1
Consortium for Bioinformatics and Computational Biology, and Department of Chemistry, University of Minnesota,
Minneapolis, Minnesota; MN 55455, USA,
2
Aperio Technologies, Vista, California; CA 92081, USA,
3
Quest Diagnostics
Nichols Institute, San Juan Capistrano, California, CA 92690-6130 USA
Received: September 3, 2008; Accepted: December 3, 2008; Revised: December 5, 2008
Abstract: Large-scale (~36,000 atoms) long-time (30 ns each) molecular dynamics (MD) simulations on the complex of
imatinib and 16 common mutants of the ABL tyrosine kinase domain have been performed to study the imatinib resistance
mechanisms at the atomic level. MD simulations show that long time computational simulations could offer insight
information that static models, simple homology modeling methods, or short-time simulations cannot provide for the
BCR-ABL imatinib resistance problem. Three possible types of mutational effects from those mutants are found: the direct
effect on the contact interaction with imatinib (e.g. some P-loop mutations), the effect on the conformation of a remote
region contacting with imatinib (e.g. T315I), and the effect on interaction between two regions within the BCR-ABL
domain (e.g. H396P). Insights of possible imatinib resistance mechanisms, not consistent with current consensus, are
revealed from various analyses and our findings suggest that drugs with different binding modes may be necessary to
overcome the drug resistance due to T315I and other mutations. The relevant patents are discussed.
Keywords: BCR-ABL, imatinib resistance, molecular dynamics simulation, MD simulation
INTRODUCTION
Chronic myelogenous leukemia (CML) is a form of
chronic leukemia caused by a characteristic chromosomal
translocation called the Philadelphia chromosome [1-5].
CML is triggered by the BCR-ABL oncoprotein containing a
constitutively activated BCR-ABL tyrosine kinase domain [6,
7]. Hence inhibitors of BCR-ABL domain have been used
successfully to treat most chronic phase CML [8].
Imatinib is an inhibitor of the ABL tyrosine kinase
domain and is currently used to treat CML [9-11]. It is
currently marketed as Gleevec or Glivec (Europe/Australia).
Although imatinib is an effective inhibitor in most cases,
mutations of BCR-ABL often show different degrees of
resistance to imatinib, especially the T315I mutation and
mutations in the ATP phosphate-binding loop (P-loop) [12-
15]. Patients who have progressed to advanced-stage CML
frequently fail to respond or lose their response to therapy
due to imatinib resistance [12, 16, 17]. The understanding of
the mechanism of imatinib resistance clearly is a critical step
toward imatinib modification or new inhibitors design [18-
22].
There have been significant amount of analysis done
based on the crystal structure of the complex of imatinib and
BCR-ABL domain [18, 19, 23-30]. The crystal structure is
very valuable information for understanding the molecular
basis of the imatinib binding. However, the crystal structure
*Address correspondence to this author at the Consortium for
Bioinformatics and Computational Biology, and Department of Chemistry,
University of Minnesota, Minneapolis, Minnesota, MN 55455, USA; Tel:
+1-612-624-1772; Fax: +1-612-626-7541; E-mail: taisung@umn.edu
is an average structure and does not necessary represent the
true structure, especially in the case that the structure
undergoes a rapid equilibrium within few conformations.
Furthermore, mutation effect analysis based on the static
structure normally ignores possible short or long range
conformational changes neither any dynamics effect due to
thermal motions is included. Finally, the crystal structure,
although potentially close to the true structure in vivo or in
vitro, might be significantly different from the true structure
in certain regions since the experimental conditions of a
crystal structure are very different from the real-life
conditions.
Computational simulations can provide structure details,
energy landscape, dynamics behaviors, and other properties
which experiments cannot or have difficulties to offer.
Relative few computational simulations have been done on
the imatinib resistance problem probably due to the size of
the system. Monte Carlo simulations have been reported to
study tyrosine kinases binding specificity and drug resistance
[31]. Molecular dynamics (MD) simulations [32] and MM-
PBSA [33] have been proven useful to study protein-ligand
interactions. A 0.4ns MD simulation with cutoff models has
been reported [34] and some insights of imatinib resistance
due to T315I have been revealed. Such small-scale and/or
short-time simulations could provide relatively fast ways to
estimate binding free energies or establish simple models to
evaluate mutational effects or to help design new drugs.
However, small-scale and/or short-time simulations are
more valuable only in local regions, and can not account for
larger area structural changes common in flexible molecules
like kinases. We recently have performed 20ns MD simu-