Abstract region increases the tyrosine kinase activity and the


Inhibition of
BCR-ABL kinase domain can be utilized to treat chronic myeloid leukemia. Some
inhibitors like imatinib, desatinib and nilotinib are efficient drug but are
resistant to some mutations. New generation of inhibitors revealed better
inhibition than the previously known inhibitors. In the present study, we have
studied the molecular dynamics simulations and binding affinities of the
wild-type and four mutants, E373K, F359C, F359L and L364I, complexed with imatinib.
According to results, simulations suggested the fluctuations in residues from
P-loop, A-loop and C-terminal residues comes from remote residue mutations
indicating global effects of mutations on the structure and drug resistance of
the BCR-ABL kinases. Analysis of binding energy revealed that the binding of
all mutants to imatinib is weaker compared to wild-type. Furthermore, in all
mutations, number of hydrogen bonds is decreased indicating the inhibition of
the formation of hydrogen bonds with imatinib and therefore, impaired or
totally abolishing drug binding. These results demonstrated large-scale simulation
is a powerful method to clarify and predict the resistance to potential drugs
or clinical response.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now



or excessive division of granulocyte cells brings about a stem cell disorder
which is called CML including three phases: Chronic, accelerated and blast 1, 2. Based on the previous studies, transfer of ABL gene of 9th
to BCR of 22nd chromosome give rise to hybrid Philadelphia
chromosome with chimeric BCR-ABL genes 3. The result of the mentioned gene is a cytoplasmic fusion BCR-ABL
active tyrosine kinase oncoprotein which advances the proliferation of CML
cells. To treat most chronic-phase of CML, inhibiting of ABL kinase domain
could be targeted 4. Besides, the protein has a pivotal role in processes such as
signal transduction and cellular growth 5-7.
N-terminal region of ABL protein is consisted of SH2 and SH3 domains with the
function of regulating tyrosine kinase activity of the protein 8. On the other hand, C-terminal region of the ABL protein includes
DNA and actin binding sites 9. Abnormality in SH2 domain brings about delaying of
phosphorylation and in SH3 domain promotes the transformation process 10. In case of BCR, N-terminal region increases the tyrosine kinase
activity and the capacity of binding ABL protein to actin 4. BCR-ABL fundamental proliferations and their abnormal expressions
arrive to different abnormalities in the peripheral blood and bone marrow
followed by changing the number of granular leukocytes. Based on the previous
studies, treatments like chemotherapy and bone marrow transplantation is having
high side effects 11. Some of the reasons affecting drug resistance are including the
amplification of the oncogenic protein kinase gene 12 or other mechanisms, however, by choosing of cancer cells
containing secondary mutations in the aimed kinase, resistance can be tracked
down. Furthermore, in order to attenuate or prevent interactions with
inhibitor, these resistance mutations often emerge in the kinase catalytic
domain 13. To figure out and address resistance in other targets,
development of BCR-ABL kinase inhibitors as a significant model is under
investigation 14. Therefore, new studies have been shifted to BCR-ABL tyrosine
kinase domain inhibitors 15. Most inhibitors influence the ATP binding with SH2 domain by
changing the tyrosine kinase domain of ABL protein. Discovery of imatinib, a type-II
tyrosine kinase inhibitors with special impact on cancerous cells and less
influences the normal cells, made a productive insight for CML treatment 16-18.
These kinds of inhibitors inhibit the target in the inactive form and binds to
the hydrophobic pocket which is in the vicinity of ATP binding site 17, 19. With taking up of tyrosine kinase domain of ABL by imatinib, ATP
is not able to contribute its phosphate group and ABL is unable to activate
downstream signaling process which intensifies CML 20. Imatinib is mainly effective in initial phases of CML, namely
chronic and accelerated, and is having less impact in blast phase 8, 17. Based on the reports, 85 % of patients treated with imatinib have
survived 21-23.
However, numerous cases regarding resistance of developed cancerous cells to imatinib
have been reported 12. As causing agent for drug resistance, about 15 point mutations
including T315I, Y253F/H, E255K/V, M351T, G250E, F359C/V, H396R/P, M244V,
E355G, F317L, M237I, Q252H/R, D276G, L248V and F486S are recognized, so far 21. On the other hand, second generation inhibitors, for inhibition
of BCR_ABL tyrosine kinase, such as nilotinib and befatinib have also been
identified 24. Since these inhibitors are effective in initial phases of chronic
and accelerated and have a low influence in blast phase of CML, therefore;
comprehensive study regarding target based therapy of CML should be considered 25. Although the crystal structures are close to in vivo or in
vitro structures, but as a result of different conditions of crystal
structure from the true structures, they probably significantly different from
the true structure26. On the other hand, by computational simulations more detailed description
of atomic level of structural details, dynamic behaviors, and other features
which are difficult to be achieved from the experimental studies, could be

Materials and

The crystal
structure of the BCR-ABL complexed with imatinib (PDB code 1IEP) was obtained
from the RCSB Protein Data Bank as initial structure 27.

All MD
simulations and molecular mechanics (MM) minimization were performed using the Gromacs 5.1.2 27 package under ff99SB force field 28 under periodic boundary conditions. The GAFF topologies for
imatinib were created by Antechamber software through restrained electrostatic
potential method and converted to GROMACS format using acpype tool. On the
other hand, all mutants were produced by SwissPdb-Viewer 4.129. The cut off of van der waals forces were considered 10 Å and PME
method was utilized with 10 Å cut off. At first, the systems were energy minimized
by steepest descent followed by conjugate gradients method. In the
equilibration step, heavy atoms were restrained by a force constant of 1000
kJ/mol nm and the solvent and ions were developed which is carried out through
minimization and MD in NVT and NPT ensembles for 100 ps. Then the temperature
was increased and the velocities, according to Maxwell-Boltzmann distribution
were reassigned at 298 K and equilibrated for 100 ps. During the equilibration
step, the Berendsen algorithm was utilized for thermostat and barostat. In the
final step, a run of 50 ns of MD simulation under NPT ensemble was carried out.
To retain a stable temperature and pressure in the production phase, Nosé-Hoover
thermostat and Parrinello-Rahman barostat was applied. The temperature set
to 298 K with the time step of 2 fs.

Relative Binding Free Energy Analysis: To analyze
differences in the relative binding free energy, ??Gbinding, of
mutants compared to wild-type the MMPBSA approach 30, 31 was used. The following
equation was applied to calculate the binding free energy. ??Gbinding=
?Eelectrostatic + ?EvdW + ??GPB + ??Gnonpolar
where ?Eelectrostatic is the electrostatic interactions, ?EvdW
is the van der waals contributions, ??GPB is the polar solvation
contributions and ??Gnonpolar is the nonpolar contributions.

Also, the number of hydrogen bonds and accessible surface
area calculations were retrieved from VADAR 1.832.


In order to
assess the structural behavior and flexibility of the native and mutants of
BCR-ABL, MD simulations of all structures was performed by Gromacs 5.1 for 50
ns. Based on our data, since RMSD analysis disclosed at least 10 ns is needed
before the system reaches a stable situation, therefore, long-time simulations
are suggested for studying the mutants of BCR-ABL system.

The wild-type
and mutant BCR-ABL kinase-imatinib complexes surrounded by explicit water molecules
and sodium ions for charge neutralization were subjected to 50 ns MD simulation
to achieve optimal interactions and molecular basis for binding.

Global Structural Stability: The
conformational stability of the wild-type and BCR-ABL mutants was carried out
by analyzing the RMSD of C? of the backbone atoms over the course of
simulation at 323 K. Fig 1 compares the RMSD values of the wild-type and
mutants over the time of simulation to their initial structure. For native
structure, results of RMSD indicate the structure is stable during 50 ns of simulation
and no significant structural alteration was observed. In E373K, there is a
structural change during initial 10 ns of simulation and the structure become
stable from 10 to 30 ns, after 30 ns another structural change was observed and
after 35 ns this mutant become stable during the rest of simulation. L364I
reveals a conformational change in 9 ns and then the structure become stable
until 30 ns. Another structural alteration was observed at 30 ns and after that
this mutant becomes already stable to 50 ns. In case of F359C, a jump of RMSD
observed at 20 ns and then F359C already became stable and for the last mutant,
namely F359L, a shift of RMSD during initial and final 10 ns of the simulation
is observed. Overall, because of the relatively small fluctuations of the
models, the trajectories are stabilized over the time and finally converged to
less than 0.4 nm and closeness of fluctuations implying the minor impact of the
mutations on the overall flexibility.

RMSF Analysis: In order to evaluate
and compare the thermal motions of a residue, the average RMSF values (per
residue) of the wild-type and mutants were calculated as a global flexibility. According
to Fig 2, different mutants show different impacts on the local motions of
proteins indicating, due to mutation, RMSF of the protein will be easily
altered. While mutations are selected between residues 359 to 373, the mobility
of the whole structure has changed. In E373K, F359C and F359L, compared to
native state, an increase of fluctuation in N-terminal residues and a decrease
in residues of 445 to 455 is observed. Furthermore, in L364I, besides of these
regions, the mobility in the region of 400 and 460 is also intensified. All of
these mobility change in different regions of the protein implying the global
impact of the mutations on the structure of the protein.

Related Binding Free Energy
Calculations: MMPBSA is applied to analyze the relative binding free energy,
??Gbinding of mutants to wild-type and usually is in excellent
agreement with experimental data. The methodology deals with thermodynamic
feature of the association of a ligand with its target protein. According to
Table 1, E373K, L364I, F359L and F359C show the values of -221.96 +/- 21.26,
-223.79 +/- 20.88, -211.32 +/- 21.23 and -94.06 +/- 88.60 kJ/mol, respectively
which are more positive compared to wild-type value -224.53 +/- 14.92 kJ/mol.
Besides, to study the binding change more deeply, electrostatic and van der
waals energy of each mutations was investigated and compared to wild-type
state. The electrostatic part describes the long-range columbic interaction
which is containing charge-charge and other multiple interactions. On the other
hand, the vdW part is the contact energy. In all cases, a positive value
representing a reduction in binding energy and more negative values implying
the binding is stronger. Table 1 shows a detailed analysis of electrostatic and
vdW energy for all mutations and their comparison to wild-type. Results
indicating in all mutants the values of vdW and electrostatic energy are more
positive than wild-type implying the stronger binding of imatinib and the protein
in wild-type situation. Furthermore, analysis of hydrogen bonds unveiling the
number of hydrogen bonds is decreased after mutations which the highest
decrease is observed in L364I. On the other hand, comparison of SASA of the
wild-type and mutants revealing this parameter is increased in all mutations
compared to native state.

Secondary Structure: Although
simulated structures did not come together to match the experimental structures
but since these structures obtain a rough form before folding, they may be
precisely predicted by MD simulations 33. To determine secondary structure, simulation analysis program
DSSP (Dictionary of Protein Secondary Structure) was used to create and visualize
secondary structure plot and the evolution of secondary structure over 50 ns
was measured. Based on the results representing in Fig 3, no difference in
secondary structures between the wild-type and variants was observed.


One of the major challenges for
targeted cancer therapies like tyrosine kinase inhibitors (TKIs) is drug
resistance. It can bring about by different factors such as point mutations in
the kinase receptor 13, 34, 35. The
impact of point mutations can be evaluated in different ways, both
experimentally and theoretically. Computational techniques can be utilized to
both assessment of experimental data and predict probable results in clinical
practice 31, 36, 37. By
theoretical studies, instances like the dynamics of proteins 38-40, the
shape and location of binding pockets 41-43 and
the anticipated interactions between the protein and ligand 44-46 can
be inspected. In drug resistance studying, mutated structures would be compared
to native states to identify remarkable differences 47, 48 which could consist of loss of interactions or change of
conformation which could be as a result of ligand binding 31, 49. In the present study, computational methods such as molecular
dynamics simulation 50, 51 to study the impact of point mutations on the BCR-ABL kinase and
how these mutations confer resistance have been applied.

Based on previous studies, there are
possible mechanisms of mutational impacts on ligand resistance including
eliminating important hydrogen bonds or preventing BCR-ABL from achieving a
specific conformation required for high-affinity ligand binding or mutations in
regulatory motifs like activation loop which may stabilize an active
conformation that is inaccessible to ligand 52.

In order to investigate structural
behavior and flexibility of the wild-type and BCR-ABL mutants, all lead
complexes were analyzed by Gromacs ….. and MD simulation was performed for 50 ns
for each complexes. Root mean square deviations (RMSD) of the mentioned
complexes were calculated against their initial structure and to compare the
flexibility of the backbone graphs were created. During the simulation period,
no significant fluctuations in the backbone of all lead compounds were observed
indicating not only binding of ligand to the active site is stable and strong
but also the protein backbone stability does not interrupted.

The RMSF data of the wild-type and
mutants reveals the mutational impact on the mobility of different regions
ranging from N- to C-terminal which is implying on global impact of mutations
on the structure of the protein. The more common mutations occur in hot spots
within the BCR-ABL kinase domain including the ATP binding P-loop (amino acids
248-256), the catalytic domain (amino acids 350-363) and the activation
(A)-loop (amino acids 381-402). From these regions, the A-loop is a critical
regulator of BCR-ABL kinase activity through adopting a closed (inactive) or
open (active) conformation and mutation in this region often destabilize the
inactive conformation, the situation which is necessary for ligand binding 53. In our study, L364I brings about an increasing flexibility in the
A-loop region including amino acids 394-398. The higher flexibility of this
region likely weakens the interaction between imatinib and BCR-ABL. Furthermore,
based on MMPBSA data, the more negative binding energy, the higher affinity of
ligand and protein and would result in more stability of the structure. As
mentioned in Table 1, all mutants unveil more positive value compared to
wild-type implying the binding of imatinib to the protein is weaker. Also,
analysis of hydrogen bonds interactions disclosing reduction of hydrogen bonds
between mutated states of the protein and imatinib that likely may bring about
disruption of the interactions between imatinib and BCR-ABL and as a consequence,
drug binding is impaired or totally abolished 54. In addition, SASA data shows in all mutations this parameter has
increased compared to native state indicating mutation may result in a
conformational change such as partial unfolding which increases the solvent
accessible surface area. Therefore, catalytic domain mutations studied in our
experiment can cause a global impact on the overall structure of the protein
and may bring about drug resistance.


In the present study, analysis based
on large-scale long time molecular dynamics simulations on wild-type and 4
mutations of BCR-ABL complexed with ligand are presented. All mutations located
in catalytic domain and the effect of these mutations on the whole of structure
was investigated. Our findings propose the following possible solutions to dominate
the current drug-resistance problems. All mutations bring about alterations of the
flexibility in activation loop which may destabilize inactive conformation and
hence attenuating ligand binding. Also, reduction of hydrogen bonds and more
positive values of ??Gbinding and electrostatic and vdW energy in
all mutations are implying on weaker or inappropriate binding of imatinib to
BCR-ABL. Further simulations for discovering more specific inhibitors and
designing of next generation of inhibitors for BCR-ABL tyrosine kinases would
be helpful.