Amygdalin as multi-target anticancer drug against targets of cell division cycle: double docking and molecular dynamics simulation

Abstract Cell-division protein kinases (CDKs) are gorgeous examples of targets for the helpful treatment of cancer by using multi-target inhibitors. Specifically, targeting cell-division protein kinase1/cyclin B (CDK1/Cyclin B), cell-division protein kinase 2/cyclin A (CDK2/Cyclin A) and cell-division protein kinase 4/cyclin D1 (CDK4/Cyclin D1) are considered a safe strategy to over the toxicity complications which are emerging from low specificity. In this work, we conducted the double docking and molecular dynamics to explicate the effect of amygdalin upon conformational modifications of selected targets. Moreover, the principal component analysis (PCA) was employed to inspect the effect of amygdalin on the fundamental motions of the each protein as target. Docking results illustrated that the binding free energies of amygdalin (AMY) to CDK1/Cyclin B, CDK 2/Cyclin A and CDK 4/Cyclin D1 were to be −9.41, −9.02 and −10.6 kcal/mol, respectively. The PCA results disclosed that binding of the AMY minimized the fundamental dynamics of CDK1/Cyclin B and CDK2/Cyclin A. The obtained results can give an insight into inhibitory activity of amygdalin that could help in designing of potential inhibitors. In the other word, it can be used AMY to inhibit other mechanisms and/or hallmarks of cancer. Communicated by Ramaswamy H. Sarma


Introduction
For decades, one of the most complicated issues is to find an impressive treatment for a complex disease, cancer. Cancer disease is characterized as a distinct lack by a functional redundancy of many signaling pathways affected by abnormal mutations (Raghavendra et al., 2018). However, protein kinases procure from conveying of the phosphate group of ATP to into particular residues of the binding site; As a result, from the phosphorylation of many cellular, cancer can develop rapidly (Qian et al., 2015). The protein kinases which encoded by a human genome are approximately five hundred, serving as the central pillars in biological signal transduction and regulation (Cohen & Alessi, 2013). They are classified hierarchically into two groups: (i) 134 families and (ii) 201 subfamilies (Manning et al., 2002). Cell-division protein kinases (CDKs) are one of the protein kinases families, and they dominate the progress steps of the cell division cycle by regulating of heterodimeric cyclin partners. Specific CDK-cyclin complexes helm the systematic progress through the G0-G1-S-G2-M cell cycle phases (Satyanarayana & Kaldis, 2009;Senderowicz, 2003).
Cell-division protein kinases (CDKs) are serine/threonine proline-directed kinases. One phase is only inoperative, means the CDKs are almost inactive monomer unless accompanied by specific cyclin (form functional heterodimeric complexes) (Morgan, 1997). Thus far, early investigations have identified twenty of both different CDKs and different cyclins in mammalian cells (Malumbres & Barbacid, 2005). While CDK1, CDK2, CDK4 and CDK6 are determined as regulators of cell-cycle progression (Obaya & Sedivy, 2002), other CDKs are embroiling in on-cell-cycle, unrelated biological pathways (Lim & Kaldis, 2013). Data from several sources have identified that decreasing in some CDKs associated with human cancers (Malumbres & Barbacid, 2009). As the S-phase begins, CDK1/Cyclin B accumulates at centrosomes (Pines & Hunter, 1989). First, the active CDK1/Cyclin B gives a signal to start the onset of mitosis system and it is in demand for successful completion (Brown et al., 2015). During G2 phase, CDK1/Cyclin B commences the phosphorylation of vast number of substrates in the G2 phase. It induces early phase of mitosis by duplicating the centrosome, bringing the spindle, stimulating the chromosome condensation and so on (Chashoo & Saxena, 2014;Holt et al., 2009). APC/C E3 ubiquitin ligase and the proteasome degrade the cyclin A and cyclin B at the beginning of anaphase. In the end, CDK1 activity is depleted (Barford, 2011;Clute & Pines, 1999;Peters, 2006). Due to CDK1 losses of its functions (Zhang et al., 2011) or be overexpressed (Shi & Zhang, 2009) is exhibited in different types of cancers. Phosphorylation of different endogenous substrates through early S-phase by active CDK2/Cyclin A, drive to DNA replication and inactivation of G1 transcription factor E2F. Inactivation of E2F is vital for S-phase completion. When E2F exists in the lack of CDK2/Cyclin A, can be essential to procure cellular apoptosis. Based on these cases, cytotoxicity can be reached by inhibiting the CDK2 instead of cell arrest (Davies et al., 2002;Knockaert et al., 2002;Lapenna & Giordano, 2009). Moreover, many cancers resulted from either by inactivating the Cip-Kip endogenous inhibitors or overexpression of CDK2, including lung cancer, melanoma, osteosarcoma, ovarian cancer, sarcomas, and pancreatic cancer (Lapenna & Giordano, 2009). The first cyclin/CDK active protein complex is CDK4/Cyclin D1 and is discharged after from a dormant state cyclin D1 and CDK4 based on the cell type. CDK4-pRb pathway is an important key regulator in the cell cycle progress, and its deregulation plays a key role in many types of cancers (Vasanthakumari et al., 2019). Cyclin D1/CDK4 or 6/INK4/pRb/E2F pathway is organizing the transition from G1 phase to S phase during the cell cycle. Consequently, aiming CDK4 for the therapeutic designing of new molecules to halt and/or improve cell cycle control is an emerging field (Dickson, 2014). CDK4/Cyclin D1 is of particular interest because of the mutation in its genome level (Hall & Peters, 1996) or overexpression of either the proteins (CDK4/Cyclin D1). The holoenzyme is very significant in the growing of different types of cancers. Retinoblastoma protein (pRb) (Fry et al., 2004;Konstantinidis et al., 1998), a tumor suppressor gene phosphorylates the active holoenzyme and releases several transcriptional factors (E2Fs). Involvement of other activated CDKs/Cyclins are recognized as therapeutic targets for finding effective treatment of malignancy during the later stages of cell cycle (Bruy ere & Meijer, 2013;Lapenna & Giordano, 2009). Despite CDKs are expressed continuously during the cell cycle, their corresponding cyclins are periodically expressed and degraded after finishing their functions in healthy cells (Morgan, 1997).
Amygdalin (AMY) is the major ingredient of the Chinese-traditional medicine in bitter almond, that approximately two hundred years it has been under research (Lee & Moon, 2016). The compound Amygdalin, D-mandelonitrile-b-D-glucoside-6-b-glucoside (Figure 1), is from aromatic cyano glycoside family derived from sources plant seed. Growing evidence has boosted the anticancer activities of amygdalin Makarevi c, Rutz, Juengel, Kaulfuss, Reiter et al., 2014;Park et al., 2005;Qian et al., 2015). The previous investigations showed that amygdalin has inhibitory actions against to CDK1/Cyclin B, CDK2/Cyclin A and CDK4/Cyclin D1 Makarevi c, Rutz, Juengel, Kaulfuss, Reiter et al., 2014). Despite all the results of experimental considerations, the anticancer effects of AMY are considered as one of the mysterious puzzles of how amygdalin works against cancer (Al-Khafaji & Taskin Tok, 2020). The popular belief in medical community that the anticancer effects of amygdalin result from hydro cyanide due to its hydrolysis (Bhatnagar et al., 2017;Song & Xu, 2014). However, the deep understanding of molecular mechanisms about how amygdalin affects their function by inhibiting targets up to now it is still undiscovered. Accordingly, double docking and molecular dynamics simulations represent the backbones of computational techniques that can represent the details about the dynamics feature of inhibition for the ligand and each protein in this study.

Ligand and proteins preparations and optimizations
Amygdalin structure recouped from the PubChem database made out of the National Center for Biotechnology Information with ID: 656495 (https://pubchem.ncbi.nlm.nih. gov). The three dimensional (3D) coordinate of the AMY was optimized by hiring the 'prepare ligand' protocol of Discovery Studio 2019 (BIOVIA D. S., 2016). The protocol is employed to systematize the charges for common groups, numeration of charges in a specialized pH, generation of tautomer and isomers and calculation of 3D coordinates for all AMY's atoms. Then the PDBQT format of the ligand had generated using Autodock Tools (Morris et al., 2009).
3D structures of selected targets, ( Table 1) were downloaded from PDB (http://www.rcsb.org/pdb). As needed for  the Lamarckian Genetic Algorithm docking compilation, all water atoms were evacuated from the local structure of over-specified proteins with the ensuing expansion of polar hydrogen atoms followed by the calculation of Gasteiger charges. Ancillary, the missing residuals were corrected exploitation the repair missing atoms possibility in AutoDock Tools (ADT) and converted to (PDBQT) format (Morris et al., 2009).

Ligand-protein docking
The first step of double docking was started from execution pf AutodockFR (ADFR) software (Ravindranath et al., 2015) in conjunction with the AutoGridFR (AGFR version 1.0) (Ravindranath et al., 2015) is utilized for acquisition rotatable ligand bonds. ADFR calculations were employed to control flexible side-chains of proteins' binding pockets, by setting the corresponding residues in the configuration file. AGFR 1.0 hired for making flexible receptor's side chains more flexible, to enable the amygdalin reaches buried grooves (Ravindranath et al., 2015). AutoGrid maps were notably estimated using the position of reference ligand inside the pocket of the holo-structure for selected targets' docking boxes. What's more, the best pose of AMY and reference ligand within each target (for validation of our docking method and comparison amygdalin with reference ligands) were obtained by running of AutoDockFR with presumptive parameters for all complexes (Ravindranath et al., 2015). Secondly, all obtained complexes doubly docked (Al-Khafaji & Taskin Tok, 2020) utilizing Autodock 4.2 for more-precise estimation of ligand pose within the pockets of targets, also to enhance the determination of binding energies that available in the generated DLG files. Not only, Autodock 4.2 has scoring function which uses the AMBER force field to estimate the binding energy of ligand-receptor but also AutoDock 4.2 is an automated and durable docking algorithm based on the Lamarckian Genetic Algorithm (LGA) (Pandey et al., 2017). The active sites were input and a grid parameter file for each protein was generated by fixing the number of grid points on the x, y and z axes to 40 Â 40 Â 40 with 0.375 Å grid spacing. Docking parameters: the number of energy evaluations was set to 250,000, and the number of generations was 50. Another key thing to remember, other docking parameters were set to the software's default values (Sathishkumar et al., 2012). The AutoGrid 4.2 and AutoDock 4.2 programs adopted to produce grid maps and to get docking results in generated DLG files. We selected the high negative values of the estimated binding energy with the lowest RMSD value for amygdalin from 50 diversified conformers from gross docking GA runs. Discovery Studio 2019 (BIOVIA D. S., 2016) was used for visualization of the docked conformations.

Molecular dynamics simulation
Molecular dynamics, (MD) possess a significant role in the elucidation of the effects of small molecules upon the dynamical stability of complexes when they bind to targets (Kumar et al., 2019). It provides an understandable image for the changes in conformations during the time of simulation, where we can find that the target-ligand complex is stable or not (Shukla et al., 2019). In this study, MD simulations were executed for selected targets in both of ligand-free and ligand-bound proteins. The obtained protein-ligand complex structures from double docking were introduced to MD simulations, according to their ligand interaction profile with high docking scores obtained from lowest binding energies of AMY with CDK1/Cyclin B, CDK2/Cyclin A and CDK4/Cyclin D1. MD were performed by using GROMACS 2018.1 package (Abraham et al., 2015), with Charmm 27 force field for all atoms (Bjelkmar et al., 2010). SwissParam (Zoete et al., 2011) was employed to generate the AMY's topologies. Proteinamygdalin systems were solvated with three-point transferable intermolecular potential (TIP3P) and placed in the center of a cubic box of size 24 Â 24 Â 24 A 3 . Minimum 1.0 Å distance was maintained between the protein and the edge of the simulation box so that protein can fully immerse with water and rotate freely. Their charges were equalized by adding sodium or chloride counter ions, where four chloride ions added for CDK2/Cyclin A and CDK2/Cyclin A-AMY, and one chloride ion for CDK4/Cyclin D1 and CDK4/Cyclin D1-AMY systems. Then, Particle Mesh Ewald, (PME) method (Wang et al., 2010) was utilized for assessment the electrostatic energy calculation. It permits the use of the Ewald summation at a computational cost comparable to that of a simple truncation method of 10 Å. In the next step, the energy minimization, (EM) was proceeded by the steepest descent algorithm at the tolerance value of 1000 kJ/mol.nm. The steepest descent approach (1000 ps) was used for each protein-AMY complex for energy minimization. Further NVT and NPT were performed for 100 ps to equilibrate the system of protein and AMY for constant volume, pressure (1 atm) and temperature (300 K). After all the optimizations, the simulation was performed without any constraint on the protein molecules or ligand for 10 ns to determine the stability. Finally, Molecular dynamics simulation was performed at a timescale of 10 ns with1000 steps (time step of 0.01 ns). After completion of simulation, the MD trajectories were examined by drawing up the root mean square deviation (RMSD) in Angstrom (Å) for each frame against time in nanosecond (ns), as well as the root mean square fluctuation (RMSF) in Angstrom (Å) of the protein backbone by utilizing the GROMACS utilities.

Principal component analysis
The principal component analysis, (PCA) approach based on the protocol (Amadei et al., 1993) within the GROMACS 2018.1 package (Abraham et al., 2015) was used to carry out the computations of the eigenvectors and eigenvalues and their projection along with the first two principal components. It is applied to decrease the complexity of the data. Besides, it extracts the dynamics motion in simulations that are presumably meaningfully for biological function (Amadei et al., 1993). In the principal component analysis, a variance/ covariance matrix was obtained from the molecular dynamics trajectories after removing undesired motions (rotational and translational). A series of eigenvectors and eigenvalues were specified by diagonalizing the matrix. The eigenvalues are amplitude of the eigenvector forward with the multi-dimensional space and the displacement of atoms along each eigenvector. It shows the planned motions of protein along each direction. The dynamics motions of proteins in the fundamental subspace were specified by projecting the Cartesian trajectory coordinates onward, the most important eigenvectors from the analysis. Trajectories of the backbone of the targets were gained by using gmx covar and gmx anaeig of Gromacs scripts. The eigenvalues are amplitude of the eigenvector along with the multidimensional space, and the displacement of atoms along each eigenvector shows the concerted motions of protein along each direction.

Cdk1/cyclin B with amygdalin
The double docking results of amygdalin with CDKs/Cyclins are given in Table 2. From this table, it is noticeable that the predicted binding energies are surprisingly close and very high in negative magnitude between CDKs/Cyclins and amygdalin. The double docking generated a conformation of amygdalin inside the binding pocket of CDK1/Cyclin B with À9.41 kcal/mol ( Figure 2 and Table 2). The conformation of AMY obtained from DLG file depending upon two criteria: First, choosing the smallest value of RMSD and second, the highest score in negative magnitude (kcal/mol). The pose selection from all the docking solutions was furthermore analyzed. The interaction between AMY and CDK1/Cyclin B complex shows that all of amygdalin entered all ATP-binding grooves (Supplementary material, Figure S1). Furthermore, the Figure S1 (Supplementary material) shows the superimposing of the experimental crystal conformation, including the inhibitor Cks1 (Brown et al., 2015) and the optimized docked conformation of amygdalin. The compound, AMY interacted with CDK1/Cyclin B and made three p-alkyl interactions between the benzene ring with Ala145, Leu135, and Val18 of CDK1's chain (Supplementary material, Table S1 and Figure S2). While the two glucose rings of amygdalin form four hydrogen bonds with Glu51, Asp146, Lys33, and Leu83 of CDK1. Furthermore, the stability of amygdalin during the 10 ns MD simulation was performed to better understand both of the effects of AMY upon CDK1/Cyclin B complex.

Cdk2/cyclin A with amygdalin
The double docking was used to investigate the conformation of amygdalin inside the ATP-binding pocket of CDK2/ Cyclin A complex (Supplementary material, Figure S3). Docking results revealed that amygdalin interacted with residues in active site of the CDK2/Cyclin A with calculated binding energy value, (À9.02 kcal/mol) as given Table 2. The compound, amygdalin formed three conventional H-bonds with Leu83, Glu81 and Gln131 residues in CDK2 chain ( Figure 3) (Detail information is available Supplementary material, Table S1 and Figure S4). Moreover, Figure S4 (Supplementary material) exhibits the superimposition of experimental crystal conformation of 4-((6-([1,1 0biphenyl] À 3-yl)-9H-purin-2-yl)amino) benzene sulfonamide inhibitor (Coxon et al., 2017). The obtained conformation of amygdalin from the double docking, used as starting for MD simulation, to investigate the stability of CDK2/Cyclin A-Amygdalin complex.

Cdk4/cyclin D1 with amygdalin
The docking results revealed that AMY binds to the groove of CDK4 (Dong et al., 2017) with the lowest binding energy (most negative) with value-10.6 kcal/mol ( Table 2). The major interactions between CDK4/Cyclin D1 and amygdalin are Hbonds with Val14, Tyr17, Gly18, Glu144, Asn145 and Asp158 residues in CDK4 (Supplementary material, Table S1 and Figure 4). The groups involved in H-bonding were hydroxyl (hydrogen donor) and carbonyl (hydrogen acceptor) groups of Amygdalin. This docking conformation of CDK4/Cyclin D1-AMY complexes, generated by double docking, was taken as initial conformation for Molecular Dynamics simulation.

Cdk1/cyclin B with amygdalin
The CDK1/Cyclin B-AMY complex was subjected for carrying out MD simulation to assess the stability of the CDK1/Cyclin B's backbone stability and also the stability of amygdalin inside the pocket. We have analyzed the time-dependent behavior of MD trajectories for CDK1/Cyclin B-AMY complex including root mean square deviation (RMSD) for all backbone atoms and root mean square fluctuations of the residues (RMSF). The RMSD of backbone is an indicator to appreciate the stability of the target during the simulation. Figure 5A provides the RMSD for CDK1/Cyclin B's backbones for amygdalin-unbound and amygdalin-bound profiles. From the data in Figure 5A, it is apparent that RMSD for both systems have the similar fashion of deviation during 10 ns. The RMSD values were constantly fewer than 2.5 Å for the entire simulation which proposed that the CDK1/Cyclin B-amygdalin complex is stable. The single most striking observation to emerge from the Figure 5B the RMSD of AMY was less than 1 Å. A possible explanation for these situations might be that AMY forms stable interaction inside the pocket of CDK1/Cyclin B. These results are likely to be related to the stability of CDK1/Cyclin B's backbone stability.
Another significant aspect of Molecular Dynamics simulation is the assessment of the protein flexibility with help of the RMSF script. The RMSF for amygdalin -bound CDK1/Cyclin B and amygdalin -unbound CDK1/Cyclin B were  Figure 6). These results show that the CDK1/Cyclin A-AMY complex is stable as in amygdalin -unbound CDK1/ Cyclin A. Taken together, RMSD and RMSF results suggest that amygdalin has a stable interaction with CDK1/Cyclin B.

Cdk2/cyclin A with amygdalin
As applied the same process for the related target, Figure 7A presents the RMSD for CDK2/Cyclin A-amygdalin complex and CDK2/Cyclin A. The RMSD profiles were constantly fewer than 2.9 Å for the entire simulation which suggested the stability of CDK2/Cyclin A-AMY complex as shown in Figure 7A. In contrast to the steep of RMSD value for the AMY-unbound CDK2/Cyclin A, the RMSD steep was much more shaper. This state shows CDK2/Cyclin A-AMY complex needs 1.5 ns from MD simulation's beginning to reach an equilibrium state. After 2.5ns th AMY-bound and AMY-unbound CDK2/Cyclin A showed approximately the same average RMSD values. Turning now to the RMSD profile of AMY showed overall less than 1.3 Å ( Figure 7B). It is indicated that amygdalin is stable inside the ATP-binding pocket during MD simulation. Furthermore, it is meaningful to diagnose the flexible parts of the protein by measuring RMSF of backbone residues. The  complex (AMY-bound CDK2/Cyclin A) showed conspicuously the same fluctuations as compared to the unbound form ( Figure 8). This result is shown that the CDK2/Cyclin B-AMY complex is stable as in AMY-unbound CDK2/Cyclin B.

Cdk4/cyclin D1 with amygdalin
The RMSD of backbone for AMY-bound CDK4/Cyclin D1 and AMY-unbound CDK4/Cyclin D1were computed to elucidate the influence of amygdalin on conformational stability through an overall time of MD simulation of the CDK4/Cyclin D1. Figure 9A shows that the RMSD profile for CDK4/Cyclin D1-AMY was increasing till 6.6 ns to reach $3. From 6.6-8.7 ns the RMSD jumped to 4 and fell to 3 for two intervals. After 8.7 ns RMSD had the same fashion of free ligand CDK4/Cyclin D1. Before explaining this behavior, it is necessary to examine the RMSD of amygdalin which shows the variation of AMY's position inside the binding site. In Figure  9B there is a clear unstable trend in the RMSD of amygdalin where it fluctuated between 1 Å and 3 Å. it occurs because of the interaction of almost polar molecule (AMY) inside the binding pocket which comprises hydrophobic grove Figure  S5 (Supplementary material). Also, the fluctuation of RMSD of AMY between 1 and 3 is another evidence for interaction of two different natures of amygdalin and CDK4/Cyclin D1 complex needs the time to be stable. Further, RMSF of backbone residues from its time-averaged position was analyzed to investigate the flexible regions of the protein. The CDK4/Cyclin D1-AMY showed conspicuously the same fluctuations as compared to AMY-unbound CDK4/Cyclin D1 ( Figure 10). This situation indicates that the CDK4/Cyclin D1-AMY complex seems to be stable as in AMY-unbound CDK1/Cyclin A.

Principal component analysis
In particular, every protein molecule has ingrained movements; they arise from the complex correlations among its   atomic motions. These internal motions are needed by the protein for its proper functioning, e.g. substrate binding, conformational modifications to various biological environments, etc. Protein molecules have many of the internal motions difficult to interpret. PCA is a method of shortening the huge dimensions of the data set to the major principal components, showing the major variations that would display the global motion of the protein with the essential information. The PCs collected during MD simulation are originally the eigenvector values collected from the covariance matrix, each of which corresponds to the change in protein trajectory. The eigenvector values were collected from the covariance matrix of the simulation were first determined. And later the protein motions are dominant which were identified after filtering the different trajectories (Mitra et al., 2019). A two dimensional (2 D) PCA conducted to compare the dynamics of protein-AMY systems to understand the role of amygdalin binding upon the proteins' motions. It can be seen from the graph in Figure 11 that the motions of CDK1/ Cyclin B were restricted to a lesser phase space in case of amygdalin -bound than AMY-unbound CDK1/Cyclin B and reduced the arbitrary in protein's motion. Overall, this result shows that binding of amygdalin in the active site lead to diminish the dynamics of CDK1/Cyclin A and reduce its biological activity. This behavior was observed for the effect of amygdalin on dynamic motion of CDK2/Cyclin A and reduced it to a lower degree (Figure 12). But with CDK4/ Cyclin D1, the arbitrary in the dynamics motion is increased due to the presence of amygdalin when it compared with AMY-free form of CDK4/Cyclin D1 as presented in Figure 13.

Putative mechanism of amygdalin as drug against targets of cell division cycle
Several reports have shown that the route of amygdalin administration can affect the nature of effect. For example, Jaswal et al. (2018) showed that the amount of released hydrogen cyanide (HCN) is different depending species of intestinal microbes which hydrolyze the amygdalin. Another study, exemplifies that intravenous infusion of AMY in human body releases neither cyanidemia nor signs of toxicity (Moertel et al., 1981). The amygdalin led to down regulate number of cell cycle controlling proteins like, cyclin A, cyclin B and cyclin D3 as well as CDK1, CDK2 and CDK4 in all cancer cells after14 days of treatment. As a consequence, amygdalin regulates cell cycle, 'CDK-Cyclin-axis': anticancer activity of amygdalin (Makarevi c et al., 2016). The compound amygdalin is also boosting the immunity where amygdalin has the capability to prompt human immune system in the prostate cancerous cells and WBCs. It has been proved that a high capability of patient's WBCs to attach to prostate cancerous cells. It heralds that amygdalin provoked the immune system to fight against cancer cells (Halenar et al., 2013). This means that amygdalin was safe for patients with advanced cancer (Moertel et al., 1981). Therefore, computational techniques are demanded to probe the real mechanism of amygdalin at atomic level. The present research disclosed that, through double docking, the amygdalin forms stable interactions with of CDK1/Cyclin B, CDK2/Cyclin A, and CDK4/Cyclin D1 complexes. After then the MD simulations were run to dissect the mechanism of the amygdalin in deep level. From RMSD, and RMSF and PCA parts, it's obvious that amygdalin has high stability in the binding sites CDK1/Cyclin B, CDK2/Cyclin A.   results showed the proficiency of amygdalin to inhibit CDK1/ Cyclin B and CDK2/Cyclin A. Also clinical examinations disclosed that amygdalin does not inhibit the CDK4/Cyclin D1 in different cell line of cancers Makarevi c, Rutz, Juengel, Kaulfuss, Reiter et al., 2014;Makarevi c et al., 2016). Surprisingly, the obtained results from MD simulations (RMSD, RMSF and PCA), showed that amygdalin doesn't exhibit stable interactions with CDK4/ Cyclin D1. In addition, the dynamics motion of CDK4/Cyclin D1 is more dynamics after binding with amygdalin. This can consider evidence that amygdalin can inhibit targets without any transformation of amygdalin structure.

Conclusion
The computational methods, double docking and molecular dynamics simulation studies were performed to explain the effect of amygdalin on to reveal the binding mechanism of three proteins behaviors of CDKs/Cyclin A, CDK2/Cyclin B and CDK4/Cyclin D1. Docking results revealed that amygdalin     strongly binds to ATP-binding site of CDK2/Cyclin A with binding free energy À9.41 kcal/mol, CDK1/Cyclin B with binding free energy À9.02 kcal/mol and CDK4/Cyclin D1 with binding free energy À10.6 kcal/mol. The obtained results from MD simulations of complexes demonstrated that amygdalin is stable with CDK1/Cyclin B and CDK2/Cyclin A and less stable with CDK4/Cyclin D1 during 10 ns MD simulation. The investigation of the dynamic behavior provided reliable information on the effect of amygdalin on the conformational variations of CDKs/Cyclin targets. The PCA results revealed that amygdalin reduced the conformational motions for both CDK1/Cyclin B and CDK2/Cyclin A. Before this study, evidence of AMY's anticancer activity comes from its hydrolysis. The present study has confirmed the findings of (Moertel et al., 1981), which found that the amygdalin in human body releases neither cyanidemia nor signs of toxicity. Finally, confirming that amygdalin has a unique structure can be a reference to design new molecules against cancer has the same structure and also to apply this molecule in extensive in vitro, in vivo studies.