Molecular Mechanism Behind the Resistance of ... - ACS Publications

Subscriber access provided by - Access paid by the | UCSB Libraries

B: Biophysical Chemistry and Biomolecules

The Molecular Mechanism Behind Resistance of the G1202RMutated Anaplastic Lymphoma Kinase to the Approved Drug Ceritinib Chaohong Chen, Zhifeng He, Deyao Xie, Liangcheng Zheng, Tianhao Zhao, Xinbo Zhang, and Dezhi Cheng J. Phys. Chem. B, Just Accepted Manuscript • DOI: 10.1021/acs.jpcb.8b02040 • Publication Date (Web): 12 Apr 2018 Downloaded from http://pubs.acs.org on April 14, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

The Molecular Mechanism Behind Resistance of the G1202R-Mutated Anaplastic Lymphoma Kinase to the Approved Drug Ceritinib

Chaohong Chen,† Zhifeng He,† Deyao Xie,† Liangcheng Zheng,† Tianhao Zhao,† Xinbo Zhang,† Dezhi Cheng*,†

†

Department of Thoracic Cardiovascular, The First Affiliated Hospital of Wenzhou

Medical University, Wenzhou, 325000, China.

* Correspondence Dezhi Cheng Department of Thoracic Cardiovascular, The First Affiliated Hospital of Wenzhou Medical University, Southern white elephant town, Ouhai district, Wenzhou city 325000, Zhejiang province, China E-mail: [email protected] Tel: +86-0755-55579242; Fax: +86-0755-55579242

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT: Anaplastic lymphoma kinase (ALK) has been regarded as an essential target for the treatment of non-small cell lung cancer (NSCLC). However, the emergence of G1202R solvent front mutation that confer resistance to the drugs were reported for the first as well as the second generation ALK inhibitors. It was supposed that the G1202R solvent front mutation might hinder the drug binding. In this study, a different fact could be clarified by multiple molecular modeling methodologies through a structural analogue of ceritinib (compound 10, Cpd-10) that is reported to be a potent inhibitor against the G1202R mutation. Herein, molecular docking, accelerated molecular dynamics (aMD) simulations in conjunction with principal component analysis (PCA) and free energy map calculations were used to produce reasonable and representative initial conformations for the conventional MD simulations. Compared with Cpd-10, the binding specificity of ceritinib between ALK wild type (ALKWT) and ALK G1202R (ALKG1202R) are primarily controlled by conformational change of the Ploop and A-loop induced energetic re-distributions, and the variation is non-polar interactions as indicated by conventional MD simulations, PCA, dynamic crosscorrelation map (DCCM) analysis and free energy calculations. Furthermore, the umbrella sampling (US) simulations were carried out to make clear the principle of the dissociation processes of ceritinib and Cpd-10 towards ALKWT and ALKG1202R. The calculation results suggest that Cpd-10 has similar dissociation processes from both ALKWT and ALKG1202R, but ceritinib is more easily dissociate from ALKG1202R than from ALKWT, thus less residence time is responsible for the ceritinib resistance. Our results suggest both of the binding specificity and the drug residence time should be emphasized in rational drug design to overcome the G1202R solvent front mutation of ALK resistance.


Page 2 of 36

Page 3 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. INTRODUCTION With the program of ‘Precision Medicine initiative’ promoted day by day, several smallmolecule tyrosine kinase inhibitors, such as gefitinib and crizotinib, have been considered as the standard first-line therapies for epidermal growth factor receptor (EGFR)-mutated non-small cell lung cancer (NSCLC) and anaplastic lymphoma kinase (ALK)-rearranged NSCLC.1-3 Crizotinib, a small-molecule inhibitor that targeting ALK, hepatocyte growth factor receptor (Met), and c‑ros oncogene 1 (ROS1) tyrosine kinases, is the first FDA-approved drug for NSCLC patients harboring chromosomal rearrangements of ALK or ROS1.4-6 However, acquired resistance is a major limitation to the efficacy of crizotinib in clinic. Similar with other TKIs, secondary ALK kinase domain mutations have been observed in patients whom turn into non-effective with crizotinib.7-8 Then, the second-generation ALK inhibitors ceritinib and alectinib have been approved for use in crizotinib-resistant ALK-fusion-positive NSCLC patients.9-11 Although the second-generation ALK inhibitors show efficacy in crizotinib-resistant ALK-fusion-positive NSCLC patients, unfortunately, resistance to these inhibitors have emerged once again. In these cases, relapsed tumors often express the ALK kinase domain G1202R solvent front mutation.12-13 ALK kinase domains divided into a smaller N-terminal domain and a larger Cterminal domain (Figure 1A). The two domains are connected by a flexible linker (kinase hinge) and the G1202R solvent front mutation is in the kinase hinge ((Figure 1A, green stick model). The N-terminal lobe adopts five stranded antiparallel β-sheets, an important regulatory αC-helix, and an ATP-binding glycine-rich loop (P-loop). The adenosine triphosphate (ATP) binding site is located in the cleft between the two domains (Figure 1A, magenta surface). The C-terminal lobe contains activation loop, eight α-helices (αD-αI) and two β stranded sheets.14-15 A full understanding of resistance mechanism is of great significance to overcome the troublesome resistance problem of the ALK inhibitors used in clinic presently, and will guide drug development for the treatment of NSCLC in the future. Some resistance mechanisms between ALK kinase domain mutations and small-molecule drugs have been investigated, however, the



resistance mechanism of ALK G1202R solvent front mutation (ALKG1202R) to ceritinib remains unknown.16-19 Based on the crystal structure of the ALKWT/ceritinib complex, it is supposed that the G1202R mutation is located at hinge region and can obstruct the key protein-ligand interactions. However, the above explanation is relatively ambiguous. Several previous studies have carried out molecular dynamics simulations to investigate the drug resistance mechanism induced by single point mutation and provide an insightful basis for rational drug design.20-23 Therefore, in this study, a comprehensive molecular modeling study was applied to reveal the resistance mechanisms.

Please insert Figure 1

In this study, ceritinib (Zykadia, LDK378) and its structural analogue of Cpd-10 (5-chloro-N2-(4-(2-(dimethylamino)ethyl)-2-methoxy-5-methylphenyl)-N4-(2 (isopropylsulfonyl)phenyl)pyrimidine-2,4-diamine) were used to study the drug resistance mechanism of ALKG1202R (Figure 1B). Ceritinib is a highly selective oral tyrosine kinase small-molecule inhibitor that has demonstrated efficacy against several crizotinib resistance mutations (in-vitro and in-vivo), and shown potent efficacy in patients harboring crizotinib-resistant ALK mutations, such as L1196M, G1269A and S1206Y.10, 24 Whereas, as same as most reported ALK inhibitors, ceritinib display very low activity on G1202R solvent front mutation.10 Recently, it was reported that the structural analogue of ceritinib (Cpd-10) that exhibited equivalent activities in vitro as ceritinib and shows encouraging activities against wild-type ALK as well as G1202R mutation.25 In the present study, molecular docking, accelerated molecular dynamics (aMD) simulations, conventional molecular dynamics (MD) simulations, principal component analysis (PCA), dynamical cross-correlation map (DCCM) analysis, molecular mechanics/generalized Born solvent area (MM/GBSA) binding free energy calculations and decompositions, umbrella sampling (US) simulations were carried out to make clear the resistance mechanism of G1202R mutation toward ceritinib. This study may provide valuable information for the design of more specific and novel


Page 4 of 36

Page 5 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ALKG1202R inhibitors for the treatment of NSCLC.

2. MATERIALS AND METHODS 2.1. Construct the Initial Structures of ALKG1202R/Ceritinib, ALKWT/Cpd-10 and ALKG1202R/Cpd-10 The atomic coordinates of the ALKWT/ceritinib complex were obtained from Protein Data Bank (PDB) database (PDB entry: 4MKC).10 The structure of ALK G1202R mutation were mutated by using Swiss-Pdb Viewer software by substituting specific residues with the wild-type ALK.26 Then, the modeled structures were refined by Chimera software, including remove all non-bonded hetero-atoms, water molecules, adding missing hydrogen atoms.27 Finally, the prepared WT and mutant structure of ALK were used to construct the ALKG1202R/ceritinib, ALKWT/Cpd-10 and ALKG1202R/Cpd-10 models by the latest version of AutoDock 4.2.6 software.28 The preparation protocol are as follows: AutoDock4 atomic radii and Gasteiger partial charges were assigned to the prepared protein structures and ligand structures, respectively. The affinity maps of ALKWT and ALKG1202R were calculated by using AutoGrid4 with a grid of 60, 60 and 60 in x, y and z directions that surrounding the binding site of ALK. The modified docking protocol are as follows: trials of 200 dockings which were clustered according to the RMSD tolerance of 2.0 Å, population size of 300, maximum number of evaluation 25000000 and other parameters were set as default. The docking results showed that the crystal structure of ALKWT/ceritinib complex and docked ALKWT/Cpd-10 had high similarity (Figure S1). Therefore, the most reliable binding mode were selected as the initial structures for further MD simulations.

2.2. Accelerated MD Simulations The crystal structure of ALKWT/ceritinib, modeled structures of ALKG1202R/ceritinib, ALKWT/Cpd-10 and ALKG1202R/Cpd-10 were used as the initial structures for the aMD simulations. The partial atomic charges for ceritinib and Cpd-10 were calculated using



Page 6 of 36

the restrained electrostatic potential (RESP) protocol with a level of quantum mechanical based on HF/6- 13G* basis set. The proteins and ligands were described by the Amber ff14SB force field and General AMBER Force Field (GAFF) by LEaP modules in Amber 14 package, respectively.29-30 Subsequently, an appropriate number of counter ions were added to maintain the electro-neutrality for all the studied systems, and then, each system was immersed in a rectangular box by TIP3P water mode with at least a 15 Å distance around the complex. Prior to the productive aMD simulations, the systems underwent energy minimization, heating and equilibration. Initially, energy minimization was carried out for each system via three steps. Firstly, a harmonic constraint potential with 5 kcal mol1

Å-2 were applied to all the water molecules with 3000 steps of steepest descent and

3000 steps of conjugate gradient. Then, the backbone atoms of the protein were constrained by a harmonic constraint potential with 4 kcal mol-1 Å-2, which enables the amino acid side chains to relieve any structural clash in the solvated systems, including 3000 steps of steepest descent and 3000 steps of conjugate gradient. Finally, all atoms were allowed to move freely without any restraint consist of 3000 steps of steepest descent and 3000 steps of conjugate gradient. After minimization, all systems were gradually heated up from 0 to 300 K in the NVT ensemble for 100 ps with a force constant of 3.0 kcal mol-1 Å-2. After heating, we performed a 2 ns equilibration simulation to adjust the solvent density under 1 atm pressure in the NPT ensemble simulation by restraining all heavy atoms with a harmonic restraint weight of 1.0 kcal mol-1 Å-2. Subsequently, the system was equilibrated for 2 ns within the NPT collective, followed by 10 ns within the NVT collective. Lastly, each system was submitted to 300 ns aMD simulation. The aMD simulation was performed using the dual-boost in which a bias is applied to both the dihedral energetic terms and the total potential energy of the system. aMD simulation modifies the energy landscape by adding a boost potential ΔV(r) to the original potential energy surface, when V(r) is below a predefined energy level E, as defined in equation. 𝛥𝛥𝛥𝛥(𝑟𝑟) = 0

𝛥𝛥𝛥𝛥(𝑟𝑟) ≥ 𝐸𝐸 ACS Paragon Plus Environment

(1)

Page 7 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


[𝐸𝐸 − 𝛥𝛥𝛥𝛥(𝑟𝑟)]2 𝛥𝛥𝛥𝛥(𝑟𝑟) = 𝛼𝛼 + [𝐸𝐸 − 𝛥𝛥𝛥𝛥(𝑟𝑟)]

𝛥𝛥𝛥𝛥(𝑟𝑟) < 𝐸𝐸

(2)

where α is the acceleration factor, E is the threshold energy. The energy threshold

is determined by equations (3) and (4), and the acceleration factor is determined by equations (5) and (6). 𝐸𝐸𝑑𝑑𝑑𝑑ℎ𝑒𝑒𝑒𝑒 = 𝑉𝑉𝑑𝑑𝑑𝑑ℎ𝑒𝑒𝑒𝑒_𝑎𝑎𝑎𝑎𝑎𝑎 + (𝜆𝜆 × 𝑉𝑉𝑑𝑑𝑑𝑑ℎ𝑒𝑒𝑒𝑒_𝑎𝑎𝑎𝑎𝑎𝑎 ) 𝐸𝐸𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 = 𝑉𝑉𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡_𝑎𝑎𝑎𝑎𝑎𝑎 + (0.2

𝑉𝑉𝑑𝑑𝑑𝑑ℎ𝑒𝑒𝑒𝑒_𝑎𝑎𝑎𝑎𝑎𝑎 5 𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘 = 0.2 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 −1 × 𝑁𝑁𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑚𝑚𝑚𝑚𝑚𝑚

𝛼𝛼𝑑𝑑𝑑𝑑ℎ𝑒𝑒𝑒𝑒 = 𝜆𝜆 × 𝛼𝛼𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡

𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 −1 × 𝑁𝑁𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 ) 𝑚𝑚𝑚𝑚𝑚𝑚

(3) (4) (5) (6)

The selections of the boost parameters E and α for the dihedral boost (𝐸𝐸𝑑𝑑𝑑𝑑ℎ𝑒𝑒𝑒𝑒 and

𝛼𝛼𝑑𝑑𝑑𝑑ℎ𝑒𝑒𝑒𝑒 ) and the total boost (𝐸𝐸𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 and 𝛼𝛼𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 ) were based on the corresponding

average dihedral energy and total potential energy obtained from combined

conventional MD simulation production runs. During the productive aMD simulations, Particle Mesh Ewald (PME) algorithm was employed to calculate the long-range electrostatic interactions of a periodic box with cutoff of 10 Å and the hydrogen atoms involved in covalent bonds were constrained by the SHAKE algorithm.31-32 The temperature was maintained using the Langevin temperature scalings.33 The coordinates were saved every 1 ps for further analysis. After aMD simulation calculations, the bias were removed by a ten order Maclaurin series expansion reweighting of each configuration to recover the canonical ensemble.34-35 The PyReweighting toolkit was used to reweight the aMD simulations for calculating PMF profiles.36 Free energy maps were constructed using the projection of the structures extracted from the aMD on the main components PC1 and PC2 computed by the principal component analysis.

2.3. Conventional MD Simulations The snapshots with lowest energy minimum reweighted by the aMD simulations were used as the initial structures for the conventional MD simulations. During the productive conventional MD simulations, PME algorithm was employed to calculate



Page 8 of 36

the long-range electrostatic interactions and the hydrogen atoms involved in covalent bonds were constrained by the SHAKE algorithm.31-32 The pressure was controlled by using a Berendsen barostat and the temperature was controlled by the Langevin temperature equilibration scheme.33, 37-38 The coordinates were saved every 10 ps for further analysis.

2.4. Principal Component Analysis (PCA) PCA method was usually used to reduce the dimensionality of the data and recognize the conformational space that a molecule occupies during MD simulation. The translational and rotational motions of all Cα atoms were eliminated by alignment of the structures from MD simulation trajectories, and the 3N × 3N covariance matrix was created with the Cartesian coordinates. The sets of eigenvectors and eigenvalues that represent the motion were generated by the diagonalization of the covariance matrix. In this study, PCA was performed for the 300 ns aMD simulation trajectories and last 20 ns conventional MD simulation trajectories by using Bio3D package of R.39

2.5. Dynamic Cross-Correlation Map (DCCM) Analysis DCCM was performed to evaluate the dynamic correlation differences for the four systems. DCCM give the pairwise correlations in the motions of the amino acid residues in the protein during the MD simulations and are calculated based on the simulation trajectories by using the Bio3D package of R.39 In this study, the cross-correlation matrix (Cij) between residues i and j were calculated based on last 20 ns conventional MD simulation trajectories with a total of 2000 snapshots and only Cα was used for analysis by averaging motions of Cα atoms deviating from the mean structure. The cross-correlation matrix (Cij) is determined by the following equation: 〈∆𝑟𝑟𝑖𝑖· ∆𝑟𝑟𝑗𝑗 〉 𝐶𝐶𝑖𝑖𝑖𝑖 = �〈∆𝑟𝑟𝑖𝑖 2 ∆𝑟𝑟𝑗𝑗 2 〉

（7）

where Cij is the fluctuant covariance of Cα atom of two residues i and j, Δr is the

distance between the current position and the average position of an atom. ⟨⟩ denotes

the time average. The value of Cij fluctuated from -1 to 1. Positive Cij value (Cij>0)


Page 9 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


represents a positive correlated motion between the residue i and the residue j, while negative Cij value (Cij

Molecular Mechanism Behind the Resistance of ... - ACS Publications

Recommend Documents