Molecular insights into the inhibitory mechanism of

18 downloads 0 Views 3MB Size Report
Apr 25, 2017 - The 2D structure of C1–C4 were drawn using ChemDraw Ultra. (v10.0) package (Fig. 2) [33]. The GROMOS96 force field parameters.
International Journal of Biological Macromolecules 102 (2017) 1025–1034

Contents lists available at ScienceDirect

International Journal of Biological Macromolecules journal homepage: www.elsevier.com/locate/ijbiomac

Molecular insights into the inhibitory mechanism of rifamycin SV against ␤2 –microglobulin aggregation: A molecular dynamics simulation study Simranjeet Singh Narang, Suniba Shuaib, Bhupesh Goyal ∗ Department of Chemistry, School of Basic and Applied Sciences, Sri Guru Granth Sahib World University, Fatehgarh Sahib 140406, Punjab, India

a r t i c l e

i n f o

Article history: Received 28 February 2017 Received in revised form 24 April 2017 Accepted 24 April 2017 Available online 25 April 2017 Keywords: ␤2 -Microglobulin Rifamycin SV Molecular dynamics

a b s t r a c t Dialysis-related amyloidosis (DRA) is a severe condition characterized by the accumulation of amyloidogenic ␤2 –microglobulin (␤2 m) protein around skeletal joints and bones. The small molecules that modulate ␤2 m aggregation have been identified in vitro, however, the underlying inhibitory mechanism remain elusive. In the present study, molecular docking and molecular dynamics (MD) simulations were performed to elucidate the inhibitory mechanism of an antibiotic, rifamycin SV (C1 ) reported for its in vitro anti–aggregation activity against ␤2 m. The molecular docking analysis highlight that C1 display hydrophobic contacts with residues in the aggregation prone region of ␤2 m. MD simulations reveal enhanced structural stability of ␤2 m in the presence of C1 . C1 inhibit the conformational transition of the C–terminal region of ␤2 m from a ␤–sheet to random coil conformation, which is reported for the initiation of fibrillogenesis of ␤2 m. The results of the present study provide insight into the key interactions and underlying inhibitory mechanism of a small molecule against ␤2 m aggregation that will help in the design and development of more potent, novel inhibitors of ␤2 m aggregation. © 2017 Elsevier B.V. All rights reserved.

1. Introduction The deposition of soluble proteins into insoluble fibrillar plaques are characteristic of over 20 different protein misfolding diseases [1–5] that include neurodegenerative and systematic amyloidosis like Alzheimer’s, Parkinson’s, Huntington’s disease as well as type II diabetes and dialysis-related amyloidosis (DRA) [6–8]. The proteins that aggregate in these diseases include amyloid ␤ (A␤), ␤2 –microglobulin (␤2 m), platelet-rich plasma (PrP); possess random coil, an all ␤–topology and ␣–helical structures, respectively. These proteins misfold and aggregate into amyloid fibrils which contain a characteristic cross ␤–sheet structure stabilized by hydrogen bond interactions [9,10]. The protein folding studies provide key insights into the factors that dictate conformation transition in the misfolded proteins involved in the amyloid formation [11–14]. The structural analysis of amyloid assembly mechanism is challenging due to the heterogeneity, dynamic properties and transient nature of intermediates involved in the amyloid formation [15]. The molecular dynamics (MD) simulations provide a clear view of conformational dynamics at an atomic level of details, which is difficult to probe using experimental methods. The stud-

∗ Corresponding author. E-mail address: [email protected] (B. Goyal). http://dx.doi.org/10.1016/j.ijbiomac.2017.04.086 0141-8130/© 2017 Elsevier B.V. All rights reserved.

ies of protein aggregation using MD simulations have provided key insights into the mechanism of amyloid formation [16]. ␤2 m is a 99-residue small globular protein that constitutes the light chain component of class I major histocompatibility complex (MHC-I). ␤2 m folds natively into a ␤–sandwich fold consisting of seven ␤–strands which are conventionally labelled as A–G starting with A from the N–terminus. The ␤–strands A, B, D, and E form the first ␤–sheet while the ␤–strands C, F, and G form the second ␤–sheet (Fig. 1) [17]. A disulphide bond between two cysteine residues (Cys25 and Cys80) covalently links two ␤–sheets [17]. The highly aromatic-rich region of ␤2 m comprising residues ∼60–70 has been reported as a key region for the nucleation of fibril assembly at acidic pH conditions [18,19]. Jones, et al., reported that increase in the local dynamics of the N– and C–terminal strands at acidic pH and the consequent decrease in cooperativity of the native structure are important features in initiating fibrillogenesis of ␤2 m [20]. The C–terminal fragments 72–99, 83–89, and 91–96 are reported to form amyloid fibrils in vitro and play a significant role in fibrillization of the full-length ␤2 m protein under acidic pH conditions [21–24]. DRA is a conformational disease that affects individuals undergoing long-term hemodialysis [25]. In healthy individuals, ␤2 m is shed into the serum and degraded by the kidneys, maintaining the serum ␤2 m concentration at 0.09–0.17 ␮M [6]. However, in patients with renal failure, the hemodialysis membrane is unable

1026

S.S. Narang et al. / International Journal of Biological Macromolecules 102 (2017) 1025–1034

partition coefficient value (log P) evaluated using thermodynamic integration (TI) method [35–37] was used to validate the parameters generated by PRODRG. The detailed information about the calculation of log P is provided in the supplementary material. 2.2. Molecular docking Molecular docking was performed using AutoDock 4.2 [38]. The crystal structure of ␤2 m was retrieved from the protein data bank (PDB ID: 1A1M, chain B) [39]. The water molecules were removed and structure was viewed using PyMOL [40]. The grid spacing was kept default (0.375 Å) and grid dimensions were set to 126 Å × 114 Å × 94 Å with grid centre defined at 14.987, 43.887, 44.316 in x, y, and z dimensions, respectively. The population of 150 individuals was used to generate 80 conformations for 27000 generations with 2500000 energy evaluations. The mutation rate of 0.02, a crossover rate of 0.80 and reference root-mean-square deviation (RMSD) were kept as default. Among stochastic search algorithms available in AutoDock, Lamarckian Genetic Algorithm (LGA) which utilize global search (Genetic Algorithm) and local search (Solis and Wets algorithm) were chosen [41,42]. The docked poses were clustered using a tolerance of 0.20 nm for RMSD and ranked on the basis of binding free energy. The molecular docking results were analysed using AutoDock Tools (ADT) [38], PyMOL [40], and LigPlot+ software [43]. 2.3. Molecular dynamics simulation and analysis

Fig. 1. Cartoon representation of the native structure of the ␤2 m protein (PDB ID: 1A1M, chain B). ␤2 m folds into a sandwich like structure formed by seven ␤-strands labelled from A to G starting from the N-terminal. The residues in the aggregation prone region (Trp60–Phe70) of ␤2 m are shown in the stick representation. The figure is rendered using PyMol.

to remove ␤2 m resulting in the increase of ␤2 m concentration by 25- to 60-fold [6,26,27]. This rise in concentration alone is not sufficient for fibril formation [24]. The studies reported that a decline in pH of the serum contributes in initiating the ␤2 m aggregation [18,22,23,28]. Thus, progressive accumulation of ␤2 m in the osteoarticular system, followed by its assembly into amyloid fibrils, eventually lead to tissue erosion and destruction [29]. Woods, et al., reported rifamycin SV (C1 ) as an inhibitor of ␤2 m aggregation at acidic pH conditions by screening small molecule libraries [30]. The inhibitory activity of small molecules for protein aggregation is investigated in a number of recent studies [31,32]. C1 bind to acid-unfolded ␤2 m conformers and induces the formation of non-toxic spherical oligomers. However, the inhibitory mechanism of C1 against ␤2 m aggregation remains elusive. To elucidate key interactions and the underlying inhibitory mechanism of C1 , molecular docking and MD simulations were performed. The results of the present study will provide key insights into the design of more potent inhibitors of ␤2 m aggregation. 2. Computational details 2.1. Parameterization of compounds The 2D structure of C1 –C4 were drawn using ChemDraw Ultra (v10.0) package (Fig. 2) [33]. The GROMOS96 force field parameters for C1 were derived using PRODRG server [34]. The octanol/water

The initial coordinates of C1 were taken from the docked complex. The protonation state of ␤2 m was assigned to mimic the experimental pH conditions (pH 2.5). The H++ server was used to determine the protonation states of charged amino acids [44,45]. The residues Lys, Arg, His, Asp, and Glu were protonated (Lys+ , Arg+ , His+ , Asp0 and Glu0 ), while N–termini (NH3 + ) was protonated and C–termini (COO•) was deprotonated for the simulation. MD simulations were performed using GROMACS 5.0 package with GROMOS96 54a7 force field [46–48]. GROMOS96 force field has been widely used for the conformational analysis of proteins [13,14,49,50]. The ␤2 m and a docked complex of ␤2 m–C1 were chosen for MD simulations and the two systems are named as ␤2 m and ␤2 m–C1 complex, respectively. The systems were placed in a cubic box of dimension 7.4 nm × 7.4 nm × 7.4 nm and the minimum distance from the complex to the edge of the box was kept 1.2 nm. The single point charge (SPC) water model was used and two systems were solvated with 13056 and 13047 water molecules, respectively [51]. The long-range electrostatic interactions were calculated using particle mesh Ewald (PME) method and shortrange van der Waals interactions cutoff was kept 1.0 nm [52,53]. The bond lengths were constrained using LINCS algorithm with an integration time step of 2 fs [54]. The system was equilibrated under NVT conditions for 100 ps at 300 K followed by 100 ps at 1 bar under NPT conditions. Both pressure and temperature were controlled using Berendson coupling protocols [55]. Each system was simulated for 50 ns and trajectories were sampled at 10 ps interval. To test the reproducibility and reliability of the MD simulations, three simulations with different initial velocities were performed for ␤2 m–C1 complex. The analysis of trajectories was carried out using GROMACS tools as well as visual molecular dynamics (VMD) [56], PyMOL [40] and dictionary of secondary structure of proteins (DSSP) program [57]. The MD ensembles were clustered using Daura and co-workers algorithm with an RMSD cutoff of 0.12 nm over backbone atoms [58]. The structural analysis was performed using radius-of-gyration (Rg ), RMSD, C␣ root-mean-square fluctuation (RMSF). The hydrogen bonds were computed using gmx hbond and

S.S. Narang et al. / International Journal of Biological Macromolecules 102 (2017) 1025–1034

1027

Fig. 2. The 2D chemical structures of C1 (Rifamycin SV), C2 (Rifaximin), C3 (5-hydroxynapthoquinone) and C4 (5,8-dihydroxynapthoquinone) are shown in panel a, b, c, and d, respectively.

a hydrogen bond was considered formed when the donor–acceptor distance was less than 0.35 nm. 2.4. Binding free energy calculations and free-energy landscape (FEL) analysis The binding free energy and contribution of each residue of ␤2 m to the binding free energy were computed using MM–PBSA tool implemented in the g mmpbsa program [59]. The detailed information about binding free energy calculations is provided in the supplementary material. The free-energy landscape (FEL) is a useful method to understand the conformational changes associated with different energy states. In the FEL, stable conformations are represented with free energy minima while the metastable states are represented by energy barriers connecting the minima. The FEL was constructed by projecting energy landscapes onto the three-dimensional space defined by Rg and RMSD with free energy (kcal/mol) as third co-ordinate for ␤2 m and ␤2 m–C1 complex at 300 K. The GROMACS utility gmx sham was used to construct the FEL. 3. Results and discussion 3.1. Prediction of binding regions and key interactions with molecular docking Woods, et al., reported that rifamycin SV (C1 ) bind to acid-unfolded ␤2 m conformers and induces the formation of non-toxic spherical oligomers [30]. Rifaximin (C2 ), a close analog of rifamcyin SV, did not abolish the formation of fibrils, however, increases the lag time of ␤2 m aggregation by 2.3-fold. 5-hydroxynaphthoquinone (juglone, C3 ) and 5,8- dihydroxynaphthoquinone (C4 ), having the naphthohydroquinone functionality of rifamycin SV, display anti-amyloid properties [30], however, had no effect on ␤2 m aggregation. The compounds C1 –C4 (Fig. 2) have similar parent moiety in their structures, however, the inhibitory effect of C1 –C4 on ␤2 m aggregation was reported to be different in the in vitro studies [30]. The docking analysis was performed to

understand the diverse inhibitory activities of C1 –C4 on the basis of their interactions and binding modes with ␤2 m. As listed in Table 1, the binding energy of C2 (–7.86 kcal/mol) is higher as compare to C1 , C3 and C4 (–6.78 kcal/mol, −5.51 kcal/mol and −5.34 kcal/mol, respectively). C1 is an antibiotic that contains a naphthohydroquinone moiety attached to an aliphatic ‘ansa’ chain. No hydrogen bonds were observed in the best docked pose of C1 with ␤2 m (Fig. 3a). The hydrophobic contacts of C1 were observed with Lys41, Ile46, Glu47, Lys48, Thr68, Glu69, Phe70 and Tyr78 of ␤2 m (Fig. 4a). The region comprising Trp60–Phe70 in ␤2 m is prone to self–assembly and residues spanning this region play a critical role in ␤2 m aggregation [18,60]. Thus, interactions in aggregation prone region explain the observed in vitro inhibitory activity of C1 . C2 (rifaximin) contain a dihydroxynaphthoquinone moiety attached to the aliphatic ‘ansa’ chain. The analysis of intermolecular interactions in the docked complex of C2 with ␤2 m confirms the presence of two hydrogen bonds (Fig. 3b). The hydroxyl group and carbonyl moiety of C2 participate in hydrogen bond formation with Lys94. C2 display hydrophobic contacts with Ile7, Gln8, Lys91, Val93, Asp96 and Met99 of ␤2 m (Fig. 4b). C2 binds to a region that does not directly participate in the aggregation process and this is consistent with the experimental studies as C2 does not inhibit ␤2 m aggregation in vitro. C3 (5-hydroxynaphthoquinone, juglone) and C4 (5,8–dihydroxynaphthoquinone) contain naphthohydroquinone functionality of rifamycin SV, however, there is no aliphatic ‘ansa’ chain as in C1 and C2 . C3 form four hydrogen bonds, three using oxygen of carbonyl moieties with Ser11, Ala15, and Trp95. In addition, C3 form hydrogen bond with the main chain oxygen atom of Met99 (Fig. 3c). C3 display hydrophobic contacts with Pro14 and Arg97 (Fig. 4c). As depicted in Fig. 3d, C4 form three hydrogen bonds, two using hydrogen atoms of the hydroxyl groups with Thr73 and Trp95 and one between its oxygen atom of a hydroxyl group and Tyr78. C4 display hydrophobic contacts with Glu77, Asp96, and Arg97 (Fig. 4d). Although C3 and C4 contain the same parental aromatic ring, however, they do not display interaction in the aggregation prone region of ␤2m. The interactions of C1 with key residues in the aggregation prone region (Trp60–Phe70) explain the potent inhibitory action of C1

1028

S.S. Narang et al. / International Journal of Biological Macromolecules 102 (2017) 1025–1034

Table 1 Molecular docking analysis of C1 –C4 with ␤2 m protein. The binding energies were evaluated using AutoDock. Compound

AutoDock binding energy (kcal/mol)

␤2 m residues participating in intermolecular hydrogen bonds

Residue

Atomsa

Distance (nm)

C1

Rifamycin SV

−6.78







C2

Rifaximin

−7.86

Lys94

O: H72

0.18

C3

5–hydroxy naphthoquinone

−5.51

Ser11

NH: O32 NH: O11

0.19 0.21

−5.34

Ala15 Trp95 Met99 Thr73

NH: O12 NH␧1: O11 O: H19 O: H19

0.16 0.20 0.21 0.17

Tyr78 Trp95

NH: O14 O: H20

0.19 0.22

C4

a

5,8–dihydroxy naphthoquinone

␤2 m residues involved in hydrophobic contacts

Inhibition constant (␮M)

Ligand efficiency

Lys41, Ile46, Glu47, Lys48, Thr68, Glu69, Phe70, Tyr78 Ile7, Gln8, Lys91, Val93, Asp96, Met99

10.7

–0.14

1.74

–0.14

Pro14, Arg97

92.0

0.42

Glu77, Asp96, Arg97

121

–0.38

The atoms on left side correspond to ␤2 m residues and atoms on right side correspond to compound.

Fig. 3. The docked complex of C1 , C2 , C3 , and C4 with ␤2 m protein is shown in panel a, b, c, and d, respectively. ␤2 m is shown in the cartoon representation and compounds are shown in the stick representation. The ␤2 m residues participating in hydrogen bond interactions are shown in orange sticks and the hydrogen bonds are shown as purple dashed line with distance in nm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

against ␤2 m aggregation in the in vitro studies. The Asp59–Thr71 region of ␤2 m is important in the self–association of partially folded ␤2 m into amyloid fibrils [19]. The observed inhibitory activity of C1

can be attributed to the presence of hydrophobic contacts between ‘ansa’ chain of C1 with Thr68, Glu69, and Phe70 in the aggregation prone region of ␤2 m (Fig. 4a). C1 was selected for further investi-

S.S. Narang et al. / International Journal of Biological Macromolecules 102 (2017) 1025–1034

1029

Fig. 4. The 2D interaction maps displaying hydrophobic contacts of C1 , C2 , C3 , and C4 with ␤2 m protein are shown in panel a, b, c, and d, respectively. The ␤2 m residues involved in hydrophobic contacts are shown in red semicircles and residues involved in hydrogen bond interactions are shown in green. The hydrogen bonds between ␤2 m residues and compounds are shown as green dashed lines. The maps are generated using LigPlot+ software. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

gation using MD simulation accredited to its interaction with key residues in aggregation prone region of ␤2 m. 3.2. Molecular dynamics simulations of ˇ2 m and ˇ2 m–C1 complex To get an insight into the detailed inhibitory mechanism of C1 against ␤2 m aggregation, MD simulations were performed. MD simulations have been employed to elucidate the inhibitory mechanism of small molecules against protein aggregation [61–63]. Two explicit solvent MD simulations of ␤2 m and ␤2 m–C1 complex in the acidic pH conditions (pH 2.5) were performed in the present study. To determine whether conformational ensembles produced by MD simulations matches with the experimental structures, NMR chemical shifts values for the C␣ and C␤ atoms of ␤2 m were calculated using SHIFTX program and compared with the experimental chemical shift values [64]. The calculated chemical shift values, ␦sim , of the C␣ and C␤ atoms of ␤2 m are highly correlated (R = 0.87, R = 0.98 for C␣, C␤ atoms, respectively) with experimental chemical shifts, ␦exp, as depicted in Supplementary Fig. S1 (panel a, b). The correlation between ␦sim and ␦exp values indicate that the MD ensembles for the ␤2 m are in close agreement with the experimental data. The thermodynamic integration (TI) method was used to calculate the value of partition coefficient of C1 in a water–octanol system. The details of log P calculations are reported in the supplementary material. In the absence of any relevant experimental data for C1 , the calculated log P was compared with average log P calculated by different computational programs, e.g., XLOGP [65], X-SCORE [66], miLogP (www.molinspiration.com), KowWin and ALOGPS (http:// www.vcclab.org/lab/alogps) [67,68]. The calculated value of log P from TI method is 3.04 which matches well with the average log P value (3.70) calculated by different computational programs.

3.2.1. Structural stability of ˇ2 m and ˇ2 m–C1 complex The ensembles were prepared with MD using GROMOS96 54a7 force field as described in computational details. The ensembles were compared as macrostates and over the microstates. The conformational ensembles were investigated using radius-ofgyration (Rg ), RMSD, C␣ RMSF, cluster analysis, secondary structure assessment, and free energy landscape analysis of the trajectories generated during simulation. The protein conformers were clustered in Cartesian space with an RMSD cutoff 0.12 nm over backbone atoms, giving microstates in decreasing order of population, viz., decreasing thermodynamic stability. The evolution of microstates for three simulations with different initial velocities for ␤2 m–C1 complex converge in a similar manner (Supplementary Fig. S2). To characterize the structural stability of ␤2 m and ␤2 m–C1 complex, Rg , RMSD and C␣ RMSF were evaluated. The Rg for ␤2 m–C1 complex fluctuates at a lower value as compare to ␤2 m (Fig. 5a). The average value of Rg was noted to be 1.45 nm and 1.42 nm for ␤2 m and ␤2 m–C1 complex, respectively, which indicate more compact ␤2 m structure in presence of C1 (Fig. 5a). The Rg values for ␤2 m in the presence of C1 are nearly identical for the three simulations with different initial velocities (Supplementary Fig. S3, panel a). The RMSD for ␤2 m–C1 complex fluctuates at a lower value as compare to ␤2 m (Fig. 5b). The average RMSD was noted to be 0.22 nm and 0.16 nm for ␤2 m and ␤2 m–C1 complex, respectively, which highlight overall structural stabilization of ␤2 m in presence of C1 . As depicted in Supplementary Fig. S3 (panel b), RMSD converge to nearly similar values for three simulations with different initial velocities for ␤2 m–C1 complex. The dynamic behaviour of ␤2 m and ␤2 m–C1 complex was assessed by evaluating C␣ RMSF for each residue (Fig. 5c, d). The average C␣ RMSF analysis highlight that about 50 residues of ␤2 m–C1 complex display lower fluctuations

1030

S.S. Narang et al. / International Journal of Biological Macromolecules 102 (2017) 1025–1034

Fig. 5. The radius-of-gyration (Rg ), root-mean-square deviation (RMSD) of ␤2 m (black) and ␤2 m–C1 complex (red) during simulation is shown in panel a, and b, respectively. The C␣ root-mean-square fluctuation (RMSF) of each residue for ␤2 m and ␤2 m–C1 complex is shown in panel c, and d, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

in their C␣ atoms as compare to ␤2 m. The C␣ RMSF of the residues in the ␤–strand C, D, E (Phe62–Tyr66), F, G and CD loop (Asp38–Lys48), DE loop (Ser55–Lys58), FG loop (Cys80–Val82, Val85–Ile92), and C–terminal residues (Asp96–Met99) decrease in ␤2 m–C1 complex which highlight lower flexibility of these regions (Fig. 5c, d). C1 decrease the local dynamics of the C–terminal strand and preserve the edge strand protective feature of the terminal ␤–strand G. These are consistent with the experimental results which emphasize the perturbation of C–terminal ␤–strand G as an important feature in the generation of assembly-competent states that initiate fibrillogenesis of ␤2 m [20]. 3.2.2. Conformational stability of ˇ2 m in the presence of C1 The conformations obtained from the MD simulation of ␤2 m and ␤2 m–C1 complex were clustered and representative structures of the most-populated microstates are shown in Supplementary Fig. S4. The conformational ensemble of ␤2 m consists of 66 microstates with 15.7% conformers in the most-populated microstate, while the number of microstates decreased to 38 with 22.8% conformers in the most-populated microstate for ␤2 m–C1 complex. The percentage population of the most-populated microstate increases significantly from 15.7% in ␤2 m to 22.8% in ␤2 m–C1 complex that indicates higher conformational stability of ␤2 m in presence of C1 . The most-populated conformation of ␤2 m highlight conformational transition of the ␤–strand E from a ␤–strand to random coil conformation (Supplementary Fig. S4). In comparison, the most-populated conformation of ␤2 m–C1 possess higher ␤–sheet content in the ␤–strand E (Supplementary Fig. S4) as compare to ␤2 m. Jones, et al., highlighted that the sequence corresponding to ␤–strand E of ␤2 m form amyloid-like fibrils in vitro and represent a

recognition surface for ␤2 m aggregation [19]. Thus, conformational stabilization of ␤–strand E in presence of C1 highlight enhanced overall stability of ␤2 m. The secondary structure analysis was performed using DSSP for ␤2 m and ␤2 m–C1 complex. The secondary structure profile of ␤2 m and ␤2 m–C1 complex is significantly different at the C–terminal region (Fig. 6). ␤2 m–C1 complex display conserved ␤–sheet content in the ␤–strand G region comprising Lys92–Lys94, while the residues in this region are populated in the random coil conformation in ␤2 m (Fig. 6). The results of secondary structure analysis correlate well with the conformational clustering results that highlight a loss in the ␤–content of ␤–strand G in the five most-populated conformations of ␤2 m (Supplementary Fig. S4). By employing ThT fluorescence studies, Jones, et al., reported that an increase in the local dynamics of the C–terminal strand and the consequent decrease in cooperativity of the native structure are important features in initiating fibrillogenesis of ␤2 m [20]. Thus, an edge strand protective feature of the terminal ␤–strand G remain preserved in presence of C1 that inhibit generation of specific aggregation-competent intermediates which initiate fibrillogenesis of ␤2 m. The time evolution of secondary structure depicts changes in secondary structure elements, i.e., coil, ␤−sheet, ␤−bridge, bend and turn in the presence of C1 (Fig. 6). During simulation, ␤2 m–C1 complex had a tendency to adopt ␤–strand, ␤–bridge and bend conformations at the C–terminal, while coil conformation is dominant at the C–terminal of ␤2 m. Statistically, 32% coil, 47% ␤–sheet, 2% ␤-bridge, 10% bend, and 9% turn components were observed for ␤2 m during simulation, while 28% coil, 49% ␤–sheet, 2% ␤-bridge, 12% bend, and 9% turn components were observed for ␤2 m–C1 complex (Table 2). The lower coil and higher

S.S. Narang et al. / International Journal of Biological Macromolecules 102 (2017) 1025–1034

1031

Fig. 6. The evolution of secondary structure for ␤2 m and ␤2 m–C1 complex during simulation are shown in panel a, and b, respectively. The Y-axis represents ␤2 m residues and the X-axis represents simulation time in ns. The secondary structure of ␤2 m is color-coded as shown underneath.

Table 2 The secondary structural component statistics for ␤2 m and ␤2 m–C1 complex during simulation. Secondary structure

␤2 m

␤2 m–C1 complex

Coil ␤-sheet ␤-bridge Bend Turn

32 47 2 10 9

28 49 2 12 9

␤–sheet content in ␤2 m–C1 complex highlight lower aggregation tendency of ␤2 m in presence of C1 . The secondary structural components are very similar as observed from the MD simulation of ␤2 m–C1 complex with different initial velocities (Supplementary Table S1). The time evolution of a number of residues involved in the secondary structure profile of ␤2 m and ␤2 m–C1 complex is shown in Supplementary Fig. S5. The residues involved in the coil conformation decreased significantly and the residues employed in the ␤-sheet conformation increased in presence of C1 (Supplementary Fig. S5). The noteworthy decrease in the number of residues in the coil conformation and concomitant increase in the number of residues in the ␤-sheet conformation highlights higher conformational stability of ␤2 m in presence of C1 . The formation of a stable three-dimensional structure of a protein was contributed by various intramolecular interactions. Among these, hydrogen bond interactions play a key role in the local folding of ␣-helices and ␤-sheets in a protein. Chandrasekaran, et al., highlighted the key role of hydrogen bonds in maintaining the structural stability of ␤2 m [69]. Thus, hydrogen bond analysis

was performed for ␤2 m and ␤2 m–C1 complex (Supplementary Fig. S6). The average total number of hydrogen bonds was noted to be 57.72 and 60.19 for ␤2 m and ␤2 m–C1 complex, respectively (Supplementary Fig. S6 (panel a), Table 3). The flexibility of a protein is directly proportional to intramolecular hydrogen bonds [70,71]. Valerio, et al., reported that conformational flexibility is the main mechanistic determinant for the aggregation propensity of a protein [72]. Thus, a higher average number of hydrogen bonds in ␤2 m–C1 complex highlight lower flexibility in the structure and thus lower aggregation propensity as compare to ␤2 m. In addition, hydrogen bonds were classified into two types based on the atomic interactions, i.e., main chain–main chain (MC–MC) and side chain–side chain (SC–SC) hydrogen bonds. The average number of MC–MC hydrogen bonds increased significantly from 41.03 to 44.12 in ␤2 m and ␤2 m–C1 complex, respectively, while the average number of SC–SC hydrogen bonds is decreased marginally from 6.74 to 6.34 in ␤2 m and ␤2 m–C1 complex, respectively. The ␣–helix, ␤–sheets and turns in proteins are stabilized by the MC–MC hydrogen bonds [69,73]. Thus, increase in average number of MC–MC hydrogen bonds in ␤2 m–C1 complex results in enhanced stability as compare to ␤2 m. To elucidate the stabilizing interactions between ␤2 m and C1 , the hydrogen bond interactions between ␤2 m and C1 were analysed during simulation. The hydrogen bonds between ␤2 m and C1 range between 1 and 5 (Supplementary Fig. S6, panel d) and the average number of hydrogen bonds is 1.26. Glu47 and Lys48 display hydrogen bond interactions with C1 during simulation (Supplementary Fig. S7). In addition, Glu44, Arg45, and Thr71 display hydrogen bonds with C1 during simulation for a short time interval (Supplementary Fig. S8).

Table 3 The total and an average number of hydrogen bonds during simulation for ␤2 m and ␤2 m–C1 complex. Hydrogen bondinteractions

Total MC–MC SC–SC a b

Total number of hydrogen bondsa

Average number of hydrogen bondsb

␤2 m

␤2 m − C1 complex

␤2 m

␤2 m − C1 complex

288583 205166 33709

300934 220600 31708

57.72 41.03 6.74

60.19 44.12 6.34

Depict hydrogen bonds summed over each frame of the trajectory during simulation. Depict an average number of hydrogen bonds in each frame of the trajectory during simulation.

1032

S.S. Narang et al. / International Journal of Biological Macromolecules 102 (2017) 1025–1034

Fig. 7. The binding free energy (kcal/mol) contribution of each residue in ␤2 m–C1 complex.

Fig. 8. Free energy landscape (kcal/mol) generated by projecting RMSD and Rg for ␤2 m and ␤2 m–C1 complex is shown in panel a, and b, respectively. The representative conformations for the minimum-energy basins are shown in the cartoon representation. The blue region display the minimum–energy basin with lowest energy conformations, while the red region display the high energy basin with least populated conformations. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 4 The different components of binding free energy (kcal/mol) between ␤2 m–C1 complex evaluated using the MM-PBSA method. Energy components

Binding free energy (kcal/mol)

EvdW Eelec EMM a Gps Gnps Gsolv b Gbind c

−46.17 ± 0.118 −61.98 ± 0.168 −108.15 ± 0.286 70.63 ± 0.351 −31.03 ± 0.278 39.60 ± 0.629 −68.55 ± 0.396

a b c

EMM = EvdW + Eelec . Gsolv = Eps + Enps . Gbind = EMM + Gsolv .

3.2.3. Insights into the molecular interactions between ˇ2 m and C1 using binding free energy analysis and ˇ2 m residue contribution to binding free energy To get an insight into the molecular interactions between ␤2 m and C1 , the binding free energy for ␤2 m–C1 complex was evaluated using MM–PBSA method. The non-bonded van der Waals (EvdW ), non-bonded electrostatic (Eelec ) interactions and non-

polar solvation (Enps ) favour the formation of ␤2 m–C1 complex (Table 4). The van der Waals, EvdW (–46.17 kcal/mol) and electrostatic, Eelec (–61.98 kcal/mol) interactions are the dominant forces that are involved in the stabilization of ␤2 m–C1 complex. The residues with interaction energy equal to or more than −1 kcal/mol with the substrate are considered to be important in substrate binding [74,75]. As illustrated in Fig. 7, Arg12, Lys19, Lys41, Arg45, Ile46, Glu47, Lys48, Thr68, Glu69, Phe70, Thr71, Lys75, Arg81, Lys94, and Arg97 are hot residues that are involved in the binding of C1 and in the stabilization ␤2 m–C1 complex.

3.2.4. Interactions of C1 with ˇ2 m remarkably alter the free-energy landscape of ˇ2 m To investigate the conformational changes in ␤2 m in presence of C1 , free-energy landscape (FEL) was plotted as a function of RMSD and Rg (Fig. 8). The global minimum-energy basin of the conformational states is shown in blue and the green region displays the meta-stable states. The minimum-energy basin for ␤2 m was located at an RMSD, Rg (0.23 nm, 1.44 nm) (Fig. 8a) whereas for ␤2 m–C1 complex the minima was located at lower values of RMSD, Rg (0.14 nm, 1.42 nm) (Fig. 8b). Thus, minimum-energy con-

S.S. Narang et al. / International Journal of Biological Macromolecules 102 (2017) 1025–1034

formations of ␤2 m–C1 complex was observed at lower RMSD and Rg values than ␤2 m (Fig. 8). FEL result highlight that C1 shifts the ␤2 m conformation from a disordered state to a more ordered compact state. 4. Conclusion In the present study, the effect of C1 (rifamycin SV) was investigated on the structural stability of ␤2 m using molecular docking and MD simulations. The molecular docking analysis reveals that C1 display interactions with key residues (Thr68, Glu69, and Phe70) in aggregation prone region of ␤2 m. MD simulations of ␤2 m and ␤2 m–C1 complex reveal several key results, (i) hydrogen bonds and hydrophobic contacts between ␤2 m and C1 stabilizes the ␤2 m–C1 complex; (ii) C1 inhibit the conformational transition of the C–terminal region of ␤2 m from ␤–sheet to random coil conformation, which is reported for the initiation of ␤2 m aggregation; (iii) FEL analysis depict that C1 shift the minimum-energy basin for ␤2 m from a disordered state to a more ordered compact state. The binding free energy analysis using MM–PBSA reveal that Arg12, Lys19, Lys41, Arg45, Ile46, Glu47, Lys48, Thr68, Glu69, Phe70, Thr71, Lys75, Arg81, Lys94, and Arg97 are involved in the binding of C1 and in the stabilization ␤2 m–C1 complex. The results of the present study will provide key insights into the rational design of more potent small molecules against ␤2 m aggregation in the DRA disease. Acknowledgements Bhupesh Goyal gratefully acknowledges Science and Engineering Research Board (SERB), Department of Science & Technology, Government of India for the award of SERB Start–Up Research Grant (Young Scientists) (Sanction No: SB/FT/CS-013/2014). Simranjeet Singh Narang acknowledges University Grants Commission (UGC) and Ministry of Minority Affairs, Government of India for the award of Maulana Azad National Fellowship (MANF) (Code No: MANF–2014–15–SIK–HIM–32950). The authors acknowledge C-DAC, Pune for providing the C-DAC’s supercomputing resources (PARAM Yuva-II) for the computational facilities. The authors acknowledge Department of Chemistry, Sri Guru Granth Sahib World University, Fatehgarh Sahib, Punjab, India for providing the research facilities. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ijbiomac.2017. 04.086. References [1] P. Salahuddin, M.T. Fatima, A.S. Abdelhameed, S. Nusrat, R.H. Khan, Structure of amyloid oligomers and their mechanisms of toxicities: targeting amyloid oligomers using novel therapeutic approaches, Eur. J. Med. Chem. 114 (2016) 41–58. [2] I. Pallarès, S. Ventura, Understanding and predicting protein misfolding and aggregation: insights from proteomics, Proteomics 16 (2016) 2570–2581. [3] T.P.J. Knowles, M. Vendruscolo, C.M. Dobson, The amyloid state and its association with protein misfolding diseases, Nat. Rev. Mol. Cell Biol. 15 (2014) 384–396. [4] D. Narang, P.K. Sharma, S. Mukhopadhyay, Dynamics and dimension of an amyloidogenic disordered state of human ␤2 -microglobulin, Eur. Biophys. J. 42 (2013) 767–776. [5] S.J. Hamodrakas, C. Liappa, V.A. Iconomidou, Consensus prediction of amyloidogenic determinants in amyloid fibril-forming proteins, Int. J. Biol. Macromol. 41 (2007) 295–300. [6] G.W. Platt, S.E. Radford, Glimpses of the molecular mechanisms of ␤2 -microglobulin fibril formation in vitro: aggregation on a complex energy landscape, FEBS Lett. 583 (2009) 2623–2629.

1033

[7] I. Moreno-Gonzalez, C. Soto, Misfolded protein aggregates: mechanisms, structures and potential for disease transmission, Semin. Cell Dev. Biol. 22 (2011) 482–487. [8] D. Eisenberg, M. Jucker, The amyloid state of proteins in human diseases, Cell 148 (2012) 1188–1203. [9] R. Nelson, M.R. Sawaya, M. Balbirnie, A.Ø. Madsen, C. Riekel, R. Grothe, D. Eisenberg, Structure of the cross-␤ spine of amyloid-like fibrils, Nature 435 (2005) 773–778. [10] J.D. Sipe, A.S. Cohen, Review History of the amyloid fibril, J. Struct. Biol. 130 (2000) 88–98. [11] P.G. Wolynes, Evolution, energy landscapes and the paradoxes of protein folding, Biochimie 119 (2015) 218–230. [12] Y. Cote, G.G. Maisuradze, P. Delarue, H.A. Scheraga, P. Senet, New insights into protein (un)folding dynamics, J. Phys. Chem. Lett. 6 (2015) 1082–1086. [13] B. Goyal, A. Kumar, K.R. Srivastava, S. Durani, Computational scrutiny of the effect of N-terminal proline and residue stereochemistry in the nucleation of ␣-helix fold, RSC Adv. 6 (2016) 74162–74176. [14] B. Goyal, K.R. Srivastava, S. Durani, Examination of the effect of N-terminal diproline and charged side chains on the stabilization of helical conformation in alanine-based short peptides: a molecular dynamics study, ChemistrySelect 1 (2016) 6321–6327. [15] T.K. Karamanos, A.P. Kalverda, G.S. Thompson, S.E. Radford, Mechanisms of amyloid formation revealed by solution NMR, Prog. Nucl. Magn. Reson. Spectrosc. 88–89 (2015) 86–104. [16] A. Morriss-Andrews, J.E. Shea, Computational studies of protein aggregation: methods and applications, Annu. Rev. Phys. Chem. 66 (2015) 643–666. [17] T. Eichner, S.E. Radford, Understanding the complex mechanisms of ␤2 -microglobulin amyloid assembly, FEBS J. 278 (2011) 3868–3883. [18] G.W. Platt, K.E. Routledge, S.W. Homans, S.E. Radford, Fibril growth kinetics reveal a region of ␤2 -microglobulin important for nucleation and elongation of aggregation, J. Mol. Biol. 378 (2008) 251–263. [19] S. Jones, J. Manning, N.M. Kad, S.E. Radford, Amyloid-forming peptides from ␤2 -microglobulin—insights into the mechanism of fibril formation in vitro, J. Mol. Biol. 325 (2003) 249–257. [20] S. Jones, D.P. Smith, S.E. Radford, Role of the N and C-terminal strands of beta 2-microglobulin in amyloid formation at neutral pH, J. Mol. Biol. 330 (2003) 935–941. [21] C. Liang, P. Derreumaux, G. Wei, Structure and aggregation mechanism of ␤2 -microglobulin (83–99) peptides studied by molecular dynamics simulations, Biophys. J. 93 (2007) 3353–3362. [22] L. Skora, S. Becker, M. Zweckstetter, Molten globule precursor states are conformationally correlated to amyloid fibrils of human ␤-2-microglobulin, J. Am. Chem. Soc. 132 (2010) 9223–9225. [23] S.E. Radford, W.S. Gosal, G.W. Platt, Towards an understanding of the structural molecular mechanism of ␤2 -microglobulin amyloid formation in vitro, Biochim. Biophys. Acta 2005 (1753) 51–63. [24] C.M. Eakin, A.D. Miranker, From chance to frequent encounters: origins of ␤2-microglobulin fibrillogenesis, Biochim. Biophys. Acta 2005 (1753) 92–99. [25] K.M. Koch, Dialysis-related amyloidosis, Kidney Int. 41 (1992) 1416–1429. [26] S.-H. Chong, J. Hong, S. Lim, S. Cho, J. Lee, S. Ham, Structural and thermodynamic characteristics of amyloidogenic intermediates of ␤-2-microglobulin, Sci. Rep. 5 (2015) 13631. [27] F. Gejyo, T. Yamada, S. Odani, Y. Nakagawa, M. Arakawa, T. Kunitomo, H. Kataoka, M. Suzuki, Y. Hirasawa, T. Shirahama, A.S. Cohen, K. Schmid, A new form of amyloid protein associated with chronic hemodialysis was identified as ␤2 -microglobulin, Biochem. Biophys. Res. Commun. 129 (1985) 701–706. [28] H. Katou, T. Kanno, M. Hoshino, Y. Hagihara, H. Tanaka, T. Kawai, K. Hasegawa, H. Naiki, Y. Goto, The role of disulfide bond in the amyloidogenic state of ␤2 -microglobulin studied by heteronuclear NMR, Protein Sci. 11 (2002) 2218–2229. [29] K. Ohashi, Pathogenesis of ␤2 -microglobulin amyloidosis, Pathol. Int. 51 (2001) 1–10. [30] L.A. Woods, G.W. Platt, A.L. Hellewell, E.W. Hewitt, S.W. Homans, A.E. Ashcroft, S.E. Radford, Ligand binding to distinct states diverts aggregation of an amyloid-forming protein, Nat. Chem. Biol. 7 (2011) 730–739. [31] P. Alam, S.K. Chaturvedi, M.K. Siddiqi, R.K. Rajpoot, M.R. Ajmal, M. Zaman, R.H. Khan, Vitamin k3 inhibits protein aggregation: implication in the treatment of amyloid diseases, Sci. Rep. 6 (2016) 26759. [32] A.J. Doig, P. Derreumaux, Inhibition of protein aggregation and amyloid formation by small molecules, Curr. Opin. Struct. Biol. 30 (2015) 50–56. [33] N. Mills, ChemDraw ultra 10.0, J. Am. Chem. Soc. 128 (2006) 13649–13650. [34] A.W. Schüttelkopf, D.M.F. van Aalten, PRODRG: a tool for high-throughput crystallography of protein-ligand complexes, Acta Crystallogr. D60 (2004) 1355–1363. [35] J.A. Lemkul, D.R. Bevan, Destabilizing Alzheimer’s A␤42 protofibrils with morin: mechanistic insights from molecular dynamics simulations, Biochemistry 49 (2010) 3935–3946. [36] S.E. DeBolt, P.A. Kollman, Investigation of structure, dynamics, and solvation in 1-octanol and its water saturated solution: molecular dynamics and free energy perturbation studies, J. Am. Chem. Soc. 117 (1995) 5316–5370. [37] P. Sassi, M. Paolantoni, R.S. Cataliotti, F. Palombo, A. Morresi, Water-alcohol mixtures: a spectroscopic study of the water saturated 1-octanol solution, J. Phys. Chem. B 108 (2004) 19557–19565. [38] G.M. Morris, D.S. Goodsell, R.S. Halliday, R. Huey, W.E. Hart, R.K. Belew, A.J. Olson, Automated docking using a Lamarckian genetic algorithm and an

1034

[39]

[40] [41]

[42] [43]

[44]

[45]

[46]

[47]

[48] [49]

[50]

[51]

[52] [53] [54] [55]

[56] [57]

[58]

S.S. Narang et al. / International Journal of Biological Macromolecules 102 (2017) 1025–1034

empirical binding free energy function, J. Comput. Chem. 19 (1998) 1639–1662. K.J. Smith, S.W. Reid, K. Harlos, A.J. McMichael, D.I. Stuart, J.I. Bell, E.Y. Jones, Bound water structure and polymorphic amino acids act together to allow the binding of different peptides to MHC class I HLA-B53, Immunity 4 (1996) 215–228. The PyMOL Molecular Graphics System, Version 1.3, LLC, Schrödinger, 2017. R. Huey, G.M. Morris, A.J. Olson, D.S. Goodsell, A semiempirical free energy force field with charge-based desolvation, J. Comput. Chem. 28 (2007) 1145–1152. F.J. Solis, R.J.-B. Wets, Minimization by random search techniques, Math. Oper. Res. 6 (1981) 19–30. A.C. Wallace, R.A. Laskowski, J.M. Thornton, LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions, Protein Eng. Des. Sel. 8 (1995) 127–134. R. Anandakrishnan, B. Aguilar, A.V. Onufriev, H++ 3.0: automating pK prediction and the preparation of biomolecular structures for atomistic molecular modeling and simulation, Nucleic Acids Res. 40 (2012) W537–W541. J. Myers, G. Grothaus, S. Narayanan, A. Onufriev, A simple clustering algorithm can be accurate enough for use in calculations of pKs in macromolecules, Proteins: Struct. Funct. Bioinf. 63 (2006) 928–938. M.J. Abraham, T. Murtola, R. Schulz, S. Páll, J.C. Smith, B. Hess, E. Lindahl, GROMACS: high performance molecular simulations through multi-level parallelism from laptops to supercomputers, SoftwareX 1–2 (2015) 19–25. N. Schmid, A.P. Eichenberger, A. Choutko, S. Riniker, M. Winger, A.E. Mark, W.F. van Gunsteren, Definition and testing of the GROMOS force-field versions 54A7 and 54B7, Eur. Biophys. J. 40 (2011) 843–856. Z. Lin, W.F. van Gunsteren, Refinement of the application of the GROMOS 54A7 force field to ␤-peptides, J. Comput. Chem. 34 (2013) 2796–2805. B. Pagano, P.D. Vecchio, C.A. Mattia, G. Graziano, Molecular dynamics study of the conformational stability of esterase 2 from Alicyclobacillus acidocaldarius, Int. J. Biol. Macromol. 49 (2011) 1072–1077. I. Autiero, E. Langella, M. Saviano, Insights into the mechanism of interaction between trehalose-conjugated beta-sheet breaker peptides and A␤(1–42) fibrils by molecular dynamics simulations, Mol. Biosyst. 9 (2013) 2835–2841. H.J.C. Berendsen, J.P.M. Postma, W.F. van Gunsteren, J. Hermans, Interaction models for water in relation to protein hydration, in: B. Pullman (Ed.), Intermolecular Forces, Reidel Publishing Company, Dordrecht, The Netherland, 1981, pp. 331–342. T. Darden, D. York, L. Pedersen, Particle mesh Ewald: an N.log(N) method for Ewald sums in large systems, J. Chem. Phys. 98 (1993) 10089–10092. U. Essmann, L. Perera, M.L. Berkowitz, T. Darden, H. Lee, L.G.A. Pedersen, A smooth particle mesh Ewald method, J. Chem. Phy. 103 (1995) 8577–8593. B. Hess, H. Bekker, H.J.C. Berendsen, J.G.E.M. Fraaije, LINCS: a linear constraint solver for molecular simulations, J. Comput. Chem. 18 (1997) 1463–1472. H.J.C. Berendsen, J.P.M. Postma, W.F. van Gunsteren, A. DiNola, J.R. Haak, Molecular dynamics with coupling to an external bath, J. Chem. Phys. 81 (1984) 3684–3690. W. Humphrey, A. Dalke, K. Schulten, VMD: visual molecular dynamics, J. Mol. Graph. 14 (1996) 33–38. W. Kabsch, C. Sander, Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features, Biopolymers 22 (1986) 2577–2637. X. Daura, K. Gademann, B. Jaun, D. Seebach, W.F. van Gunsteren, A.E. Mark, Peptide folding: when simulation meets experiment, Angew. Chem. Int. Ed. 38 (1999) 236–240.

[59] R. Kumari, R. Kumar, Open source drug discovery consortium, A. Lynn, g mmpbsa–a GROMACS tool for high-throughput MM-PBSA calculations, J. Chem. Inf. Model. 54 (2014) 1951–1962. [60] K.E. Routledge, G.G. Tartaglia, G.W. Platt, M. Vendruscolo, S.E. Radford, Competition between intramolecular and intermolecular interactions in an amyloid-forming protein, J. Mol. Biol. 389 (2009) 776–786. [61] A. Ganesan, M.L. Coote, K. Barakat, Molecular dynamics-driven drug discovery: leaping forward with confidence, Drug Discov. Today 22 (2017) 249–269. [62] S. Shuaib, B. Goyal, Scrutiny of the mechanism of small molecule inhibitor preventing conformational transition of amyloid–␤42 monomer: insights from molecular dynamics simulations, J. Biomol. Struct. Dyn. (2017), http:// dx.doi.org/10.1080/07391102.2017.1291363. [63] S. Shuaib, R.K. Saini, D. Goyal, B. Goyal, Insights into the inhibitory mechanism of dicyanovinyl-substituted J147 derivative against A␤42 aggregation and protofibril destabilization: a molecular dynamics simulation study, ChemistrySelect 2 (2017) 1645–1657. [64] S. Neal, A.M. Nip, H. Zhang, D.S. Wishart, Rapid and accurate calculation of protein 1 H, 13 C and 15 N chemical shifts, J. Biomol. NMR 26 (2003) 215–240. [65] R. Wang, Y. Fu, L. Lai, A new atom-additive method for calculating partition coefficients, J. Chem. Inf. Comput. Sci. 37 (1997) 615–621. [66] R. Wang, L. Lai, S. Wang, Further development and validation of empirical scoring functions for structure-based binding affinity prediction, J. Comput. Aided. Mol. Des. 16 (2002) 11–26. [67] I.V. Tetko, V.Y. Tanchuk, Application of associative neural networks for prediction of lipophilicity in ALOGPS 2.1 program, J. Chem. Inf. Comput. Sci. 42 (2002) 1136–1145. [68] G. Klopman, H. Zhu, Recent methodologies for the estimation of n-octanol/water partition coefficients and their use in the prediction of membrane transport properties of drugs, Mini. Rev. Med. Chem. 5 (2005) 127–133. [69] P. Chandrasekaran, R. Rajasekaran, A systematic molecular dynamics approach to the structural characterization of amyloid aggregation propensity of ␤2-microglobulin mutant D76N, Mol. Biosyst. 12 (2016) 850–859. [70] K. Balu, V. Rajendran, R. Sethumadhavan, R. Purohit, Investigation of binding phenomenon of NSP3 and p130Cas mutants and their effect on cell signalling, Cell Biochem. Biophys. 67 (2013) 623–633. [71] R. Purohit, V. Rajendran, R. Sethumadhavan, Studies on adaptability of binding residues and flap region of TMC-114 resistance HIV-1 protease mutants, J. Biomol. Struct. Dyn. 29 (2011) 137–152. [72] M. Valerio, A. Colosimo, F. Conti, A. Giuliani, A. Grottesi, C. Manetti, J.P. Zbilut, Early events in protein aggregation: molecular flexibility and hydrophobicity/charge interaction in amyloid peptides as studied by molecular dynamics simulations, Proteins: Struct. Funct. Bioinf. 58 (2005) 110–118. [73] N. Eswar, C. Ramakrishnan, Deterministic features of side-chain main-chain hydrogen bonds in globular protein structures, Protein Eng. Des. Sel. 13 (2000) 227–238. [74] H. Chen, Y. Zhang, L. Li, J.G. Han, Probing ligand-binding modes and binding mechanisms of benzoxazole-based amide inhibitors with soluble epoxide hydrolase by molecular docking and molecular dynamics simulation, J. Phys. Chem. B 116 (2012) 10219–10233. [75] W. Li, Y. Tang, H. Liu, J. Cheng, W. Zhu, H. Jiang, Probing ligand binding modes of human cytochrome P450 2J2 by homology modeling, molecular dynamics simulation, and flexible molecular docking, Proteins: Struct. Funct. Bioinf. 71 (2008) 938–949.