Human CHK-1 is a nuclear protein and it consists of a highly conserved N-terminal kinase domain and less conserved C-terminal region (17). The X-ray crystallographic solid foundation of human CHK-1 kinase protein and its complex with ATP (an endogenous agonist) and other clinical trial inhibitors (UCN-01, CCT245737, Rabuseritib, Prexasertib, GDC-0575) briefly delineates the five essential regions namely, (a) the hinge region, (b) the ribose binding pocket, (c) the solvent exposed region, (d) the water pocket and (e) the polar (18, 19). The highly conserved hinge region containing Cys-87 and Glu-85 amino acids and the ligand binding to this region potentiates the inhibitory activity. The major amino acid residues such as Tyr-20, Lys-38, Glu-55, Asn-59, Phe-149 and Ser-147 that are present in the interior water pocket and polar region are unique for CHK-1 kinase and is occupied by three water molecules. The binding interaction of ligand functional groups in this region increases the CHK-1 kinase selectivity. There is also a ribose binding pocket that consists of hydrophilic amino acids like Glu-91, Glu-134 and Asn-135. Similarly, the exposed region is situated only a few amino acid residues away from the hinge region The presence of polar and highly electronegative groups increases the ligand stability and kinase inhibitory (18, 20). (Fig. 1a).
e-Pharmacophore Modelling
Pharmacophore modelling is a 3D special arrangement of molecular features to recognize the suitable ligands for selected biological macromolecules (21). Compared to the traditional pharmacophore modelling, e-pharmacophore hypothesis incorporates both stereo-electronic features of the ligand with the energetics of its interactions with the protein structure (22). It serves as a powerful filter tool for virtual screening and utilizes the Glide XP scoring function to characterise the protein-ligand interactions.
The present study generated an e-pharmacophore hypothesis from 3D crystal coordinates of CHK-1 protein complex with a potent and selective clinical trial inhibitor (CCT245737). Initially the protein structure was obtained from Protein Data Bank (PDB ID: 5F4N; Resolution: 1.91Å) and was prepared using protein preparation wizard. Subsequently, the prepared protein structure was validated through Ramachandran plot by visualizing the stereo-chemical spatial arrangement of the amino acid residues. From the above protein structure, the native ligand was extracted and re-docked with same binding pocket using Glide XP docking mode. The binding of the re-docked co-crystalized ligand in the ATP binding site was visualized with the help of XP-Visualizer application and simultaneously the reward and penalty regions were estimated.
The protein-ligand interactions of CHK-1 with CCT245737 ligand included (a) the two H-bond interactions in the hinge region involving the backbone NH-group of Cys-87 and backbone carbonyl group of Glu-85 amino acid residues; (b) H-bond and salt bridge interactions formed with the sidechain carbonyl group and acidic residue of Glu-91 amino acid residue, in the ribose binding pocket and (c) occupation of highly electronegative halogen group (Fluorine) in solvent exposed region which increases the protein-ligand stability (Fig. 1b & c).
Based on the insight of XP-visualizer information and protein -ligand interaction, we developed an e-pharmacophore hypothesis with seven essential molecular features AADPRRR (Two- Hydrogen bond acceptors (A), One- Hydrogen bond donor (D), One-Positively ionisable group (P) and Three- Aromatic rings (R)) to retrieve the reprovable FDA approved drug candidate from further database screening. The generated e-phamacophoric features were illustrated in Fig. 2.
e-pharmacophore hypothesis screening and molecular docking
The pharmacophore features were incorporated in PHASE module and allowed for previously prepared database screening. To retrieve a more promising FDA drug for CHK1 inhibition, we fixed 4 minimum matching criteria from the generated 7 pharmacophore features. After database screening we obtained 812 drug candidates with maximum fitness score of 1.867 and used for further molecular docking studies.
Recently, numerous studies have used molecular docking method to identify the suitable drug molecule for the target of interest (23). It is carried out to exclude those inactive drug candidates from the above e-pharmacophore based screening to obtain most favourable compounds. In our study, the RMSD value of re-docked clinical trial ligand (CCT245737) was substantially low (0.4387), and conformed the accuracy of docking program to screen and predict the binding affinity of above obtained 812 ligands with CHK-1 protein (Supplementary Fig.S1). Finally, the Extra precision (XP) molecular docking program was opted for above obtained FDA drugs, and 167 drugs were docked into the ATP binding site of CHK-1 protein. The molecular docking produced ligand docking score, Glide score along with bond interaction distance values (24). The binding modes of all 167 drug candidates were ranked based on the docking score and the top 10% ligands were visually analysed for nature of protein-ligand interactions and binding affinity (Table 1). Interaction fingerprints depends on both specific interactions at the binding site and the non-specific force outside the target binding pocket. The binding mode structures of top 10% ligands are illustrated in supplementary Fig.S2.
Table 1
Extra precision docking results of top 10% FDA compounds obtained through e-pharmacophore screening approach.
Drug name | Docking score | Glide G Score | Interaction Residue | Nature of Interactions | Fitness score |
Methotrimeprazine | -10.505 | -10.505 | Glu-91 | H-bond (1.70Å) Salt bridge (4.82 Å) | 1.750 |
Acepromazine | -10.358 | -10.385 | Glu-91 | H-bond (1.68Å) Salt bridge (4.84 Å) | 1.738 |
Promazine | -9.721 | -9.721 | Glu-91 | H-bond (1.67Å) Salt bridge (4.79 Å) | 1.757 |
Procaterol | -9.466 | -9.734 | Glu-85 Glu-91 Cys-87 | H-bond (2.47Å) H-bond (1.83Å) H-bond (1.92Å) Salt bridge (4.84 Å) | 1.568 |
Carvedilol | -9.297 | -9.406 | Glu-85 Glu-91 | H-bond (1.90Å) H-bond (1.82Å) H-bond (2.15Å) Salt bridge (4.94 Å) | 1.569 |
Alimemazine | -9.289 | -9.289 | Glu-91 | H-bond (1.73Å) Salt bridge (4.83 Å) | 1.762 |
Ropinirole | -9.026 | -9.026 | Glu-85 Cys-87 | H-bond (1.63Å) H-bond (2.05Å) | 1.621 |
Methodilazine | -9.011 | -9.011 | Glu-91 | H-bond (1.91 Å) Salt bridge (4.87 Å) | 1.797 |
Propiomazine | -8.913 | -8.913 | Glu-91 | H-bond (1.82 Å) | 1.648 |
Aceprometazine | -8.873 | -8.873 | - | - | 1.660 |
Mesoridazine | -8.738 | -8.873 | Glu-91 | H-bond (1.64 Å) Salt bridge (4.74Å) | 1.842 |
Dexrazoxane | -8.586 | -9.041 | Glu-85 Cys-87 Glu-91 | H-bond (1.88Å) Salt bridge (4.84Å) H-bond (1.99Å) | 1.522 |
Encainide | -8.489 | -8.489 | Cys-87 | H-bond (2.07Å) | 1.677 |
Promethazine | -8.408 | -8.408 | Glu-91 | H-bond (2.33Å) | 1.701 |
Cyamemazine | -8.231 | -8.231 | Glu-91 | H-bond (1.82 Å) Salt bridge (4.95Å) | 1.723 |
Isothipendyl | -8.191 | -8.191 | - | - | 1.694 |
CCT245737 | -10.190 | -11.468 | Glu-85 Cys-87 Glu-91 | H-bond (2.13Å) H-bond (2.22Å) H-bond (1.48Å) Salt bridge (4.56Å) | - |
From the top 10% scored ligands, we noticed that the Procaterol and Dexrazoxane drugs possess better binding affinity as well as important key interactions with the ATP binding pocket of CHK-1 protein. More elaborately, Procaterol showed double hydrogen bond interactions with hinge region amino acid residue Glu-85 as well as single hydrogen bond interaction with Glu-91 and Cys 87 residues present in the ribose pocket and hinge region respectively (Fig. 3a & b). Similarly, Dexrazoxane forms hydrogen bond with Glu-85 and Cys-87 amino acid residues of hinge region and Glu-91 residue of Ribose binding pocket (Fig. 3c & d). We selected Procaterol and Dexrazoxane for further computational analysis based on the knowledge of existing CHK-1 inhibitors and binding site requirements.
Further, we elucidated the selectivity and potency of Procaterol and Dexrazoxane with CHK-2 protein by molecular docking using Glide XP scoring function. The 3D crystal structure of human CHK-2 protein complex with potent inhibitor (Debromohymenialdisine) with 2.70Å resolution was obtained from “Protein Data Bank” and structural refinement was performed using Protein Preparation Wizard. Similar to that of CHK-1 protein, the RMSD value of superimposition was found to be low (0.674) and it delineated the accuracy of docking (Supplementary Fig.S3).
After validation of the docking program, both clinical trial ligand (CCT245737) and the two selected FDA drugs were docked with ATP binding pocket of CHK-2 protein using Glide Extra Precision mode. The docking score of CCT245737, Procaterol and Dexrazoxane was found to be -6.369, -3.497 and − 5.489 kcal/mol respectively (Supplementary Fig.S4). This molecular docking shows that procaterol and dexrazoxane possess less binding affinity towards CHK-2 than the clinical trial ligand.
In silico prediction of Pharmacokinetic (ADME) and toxicity parameters using Qikprop
Generally, the Qikprop module analyses and calculates physiologically suitable pharmacokinetic and toxicity parameters and provides information about the safety range for all selected drug molecules (25). In this study we performed Qikprop screening for the already existing FDA approved drugs - Procaterol and Dexrazoxaneto compare their pharmacokinetic and toxicity parameters with the clinical trial inhibitor. The results of Qikprop data are shown in Table 2. All selected drug candidates show the acceptable range of drug likeness characteristics.
Table 2
The Selected drug candidates with their physicochemical and toxicity descriptors determined by Qikprop tool.
Compound Namea | Molecular weight (g/mol)b | QPlogo/wc | QPlogSd | QPlog HERGe | QPlogBBf | QP Caco g | #metab ± h | QPlogKhsai | %human Oral Absorptionj | RO5k | RO3l |
CCT245737 | 379.34 | 0.985 | -4.046 | -5.639 | -0.775 | 57.546 | 6 | -0.385 | 64.22 | 0 | 0 |
Procaterol | 290.36 | 0.805 | -2.065 | -4.839 | -0.878 | 69.689 | 3 | -0.301 | 64.67 | 0 | 0 |
Dexrazoxane | 268.27 | -2.328 | 0.023 | -4.355 | -0.822 | 2.538 | 6 | -0.869 | 20.55 | 0 | 1 |
a Ligand name ; b Molecular weight of the compound (acceptable range: 130–725 g/ mol ); c Predicted octanol/water partition co-efficient logP (acceptable range:-2.0 to 6.5) d Predicted aqueous solubility ; S in Mol/L (acceptable range:-6.5 to 0.5); e Predicted IC50 value for Blockage of HERG K+ Channels (acceptable range: below − 5.0); f Predicted Blood Brain Barrier(BBB) permeability (acceptable range:-3.0 to 1.2); g Predicted apparent Caco-2 cell permeability in nm/sec (range: <25 poor, > 500 great) h Number of likely metabolic reactions (range: 1to 8);I Predicted Human serum albumin binding (acceptable range: -1.5 to 1.5);j Percentage of human oral absorption (< 25% is poor &>80% is high); kNumber of violations in Lipinski rule of Five; l Number of violations of Jorgensen’s rule of three.
Prime MMGBSA calculation
The Prime MMGBSA simulation tool can be utilized to predict the ligand binding energies and strain energies (26) of the selected drug candidates from previous computational steps. The low energy binding poses of CCT245737, Procaterol and Dexrazoxane with CHK-1 protein (5F4N) retrieved from Glide XP docking, were rescored based on the theoretical calculation of total binding free energies. Our results revealed the MMGBSA ΔG Bind score of CCT245737, Procaterol and Dexrazoxane to be -50.28, -48.35 and − 32.40 respectively. Interestingly, Procaterol displayed a better ΔG Bind score near the clinical trial inhibitor (CCT245737) than the Dexrazoxane molecule. Collectively MM-GBSA analysis suggested that Procaterol forms a more stable complex and possessed selective inhibition of CHK-1 protein like clinical trial inhibitor.
Molecular dynamics insight on protein-ligand stability
Molecular dynamics assign velocities and calculate forces on all atoms to provide an insight into dynamic perturbations within the protein-ligand complex and interactions of ligand with water molecules (27). It is carried out to understand the stability and conformational changes of ligand–protein complexes (28). Based on the results of Molecular docking, ADME/T prediction and MMGBSA analysis, Procaterol possesses better binding affinity, acceptable range of kinetic profile and lowest ΔG Bind score than Dexrazoxane. The protein-ligand stability of the above shortlisted Procaterol-CHK-1 complex was evaluated using Desmond MD simulation analysis and the results were compared with the reference inhibitor (CCT245737).
The MD results of clinical trial inhibitor (CCT245737)-CHK-1 complex is illustrated in Fig. 4. The dynamics trajectory events remained stable throughout the simulation process. Additionally, the stability of the complex was determined by plotting the RMSD graph during simulation. The Fig. 4a depicts the stability of CCT245737 CHK-1 complex and the PL-RMSD was found between the ranges of 0.50 to 3.2 Å. Subsequently, CHK-1 protein residues present in the hinge region, ribose binding pocket and polar region interacted with the ligand atoms and the complex was stabilized by different intermolecular interactions. Briefly, interaction patterns showed that 98% and 97% of H-bond interactions was formed with Glu-85 and Cys-87 in hinge region. Similarly, 95% of H-bond interactions was formed with Glu-91 residue present in ribose binding pocket. Also, it occupied the water pocket and forms Water Bridge (95% and 83%) and H-bond (39%) interactions with polar region residues Glu-55, Asn-59 and Asp-148 respectively.
The result obtained from the Procaterol-CHK-1 complex MD simulation and the results are showed in Fig. 5. The RMSD for the Procaterol-CHK-1 complex remained stable for 200 ns time frame with the Protein RMSD 2.25 Å and ligand RMSD within 0.25 to 4Å till the end of the simulation. Further, we examined the site at which Procaterol bound to the CHK-1 protein and found that it robustly interacted with CHK-1 residues present in the hinge region, ribose binding pocket and polar region. In short, the key interactions included, the 87% H- bond interaction as well as 71% were formed with hinge region residues Glu-85 and Cys-87 respectively. Similarly, 85% and 52% of water bridge interactions were formed in the polar and ribose binding pockets of Glu-91 amino acid residues. The results of XP-docking and Post-MD based intermolecular interactions for CHK1 protein with CCT245737 and Procaterol are summarised in Table 3. The overall MD results insisted that Procaterol is a preferable CHK-1 inhibitor since it has all the essential interactions as compared to the clinical trial inhibitor (CCT245737).
Table 3
The XP-docking and Post-MD based intermolecular interactions between protein-ligand complexes
| XP-docking based intermolecular interactions | Post MD based intermolecular interactions (100ns, 200ns) |
Drug Name | Interaction Residue | Nature of interaction | Interaction Residue | Nature of major interaction factions |
CCT245737 | Glu-85 Glu-91 Cys-87 | H-bond H-bond H-bond Salt bridge | Glu-85 Cys-87 Glu-91 Glu- 55 Asn- 59 | H-bond H-bond H-bond Water bridge Water bridge |
Procaterol | Glu-85 Glu-91 Cys-87 | H-bond H-bond H-bond Salt bridge | Glu-85 Cys-87 Glu-91 Lys-38 | H-bond Water bridge H-bond H-bond Water bridge Water bridge |