The role of phenolic OH groups of flavonoid compounds with H-bond formation ability to suppress amyloid mature fibrils by destabilizing β-sheet conformation of monomeric Aβ17-42

Alzheimer’s disease (AD) is a kind of brain disease that arises due to the aggregation and fibrillation of amyloid β-peptides (Aβ). The peptide Aβ17–42 forms U-shape protofilaments of amyloid mature fibrils by cross-β strands, detected in brain cells of individuals with AD. Targeting the structure of Aβ17–42 and destabilizing its β-strands by natural compounds could be effective in the treatment of AD patients. Therefore, the interaction features of monomeric U-shape Aβ17–42 with natural flavonoids including myricetin, morin and flavone at different mole ratios were comprehensively studied to recognize the mechanism of Aβ monomer instability using molecular dynamics (MD) simulations. We found that all flavonoids have tendency to interact and destabilize Aβ peptide structure with mole ratio-dependent effects. The interaction free energies of myricetin (with 6 OHs) and morin (with 5 OHs) were more negative compared to flavone, although the total binding energies of all flavonoids are favorable and negative. Myricetin, morin and flavone penetrated into the core of the Aβ17–42 and formed self-clusters of Aβ17-42-flavonoid complexes. Analysis of Aβ17-42-flavonoids interactions identified that the hydrophobic interactions related to SASA-dependent energy are weak in all complexes. However, the intermolecular H-bonds are a main binding factor for shifting U-shape rod-like state of Aβ17–42 to globular-like disordered state. Myricetin and morin polyphenols form H-bonds with both peptide’s carbonyl and amine groups whereas flavone makes H-bonds only with amine substitution. As a result, polyphenols are more efficient in destabilizing β-sheet structures of peptide. Accordingly, the natural polyphenolic flavonoids are useful in forming stable Aβ17-42-flavonoid clusters to inhibit Aβ17–42 aggregation and these compounds could be an effective candidate for therapeutically targeting U-shape protofilaments’ monomer in amyloid mature fibrils.


Introduction
Protein misfolding and aggregation is one of the most concerning problems in applied biophysics [1,2] and molecular medicine [3]. Several human brain diseases occur due to the protein misfolding and aggregation considered as hallmark pathognomonic features of various neurodegenerative diseases, including Alzheimer, Parkinson, Huntington and Prion diseases [4,5]. Alzheimer's disease (AD) is a progressive memory loss causes destruction and death of brain cells. Unfortunately, despite various attempts to combat the disease, an effective therapeutic intercession is lacking for this common cause of mortality in the world [6]. The disease is estimated to afflict upward 5.2 million Americans and more than 25 million individuals worldwide [7]. Although the underlying cause of AD is not completely clear, it is apparent that the major pathological features are because of the presence of macroscopic structures aggregation and deposition of β-amyloid peptides (Aβ). The aggregated peptides are derived from amyloid precursor protein by sequential cleaving of β-and γ -secretases [8]. γ-secretases produce the multiple alloforms of peptides such as Aβ40 and Aβ42, with 40 and 42 amino acid residues respectively. Aβ40 peptide is the most abundant form, but in the disease state the level of Aβ42 is raised up and more readily formed aggregates in solution [9]. The aggregation of Aβ42 peptides is necessary in the progression of AD associated with amyloid fibrils [2]. In addition, the major component of brains plaques in AD, are amino-terminal truncated Aβ17-42 peptide corresponded to the amyloid mature fibrils [10,11].
Efforts to design new compounds based on γ-secretase's inhibitors have met with modest success, since the activities of secretases are vital for normal neuronal function and inhibiting could strengthen the symptoms of AD [12]. Recent efforts are focused on the numerous possibilities that have been directly explored in targeting Aβ peptides using natural compounds [13,14]. The inhibition of Aβ peptides aggregation [15,16] and destabilizing amyloid mature fibrils by disturbing cross-β strands [17,18] are promising therapeutic strategies for progression in early and late stages of AD, respectively. The characterized structural motif in amyloid fibrils is known as cross-β [19]. The cross-β strands are defined the strands of a β-sheet run vertically to the axis of fibril forming mature fibrils [20]. As schematically shown in Fig 1, there are two main strategies to destabilize and/or dissipate the amyloid mature fibrils including; a) changing cross-β strands [17,[21][22][23] mainly based on alteration of Aβ peptides' quaternary structure (Fig 1, pathways 1 and 1'), and b) focusing on the dissociated β strands from fibrils and converting β-sheet conformations to other secondary structures (changing tertiary and predominantly secondary structures of peptides, pathway 2). The disruption of mature fibrils is an effective procedure to decrease the toxicity of amyloid's plaques at therapeutic mediation in the late phase of AD. Natural compounds, such as flavonoids may be useful in inhibiting Aβ aggregation and destabilizing preformed fibrils because of their essential bioavailability and low toxicity at therapeutically suitable levels [14]. Various investigations have been conducted on the disruption of mature amyloid fibrils's cross-β strands and reduction of the toxicity of amyloid's plaques in the living cells by natural polyphenolic compounds [24,25]. In vitro evidences have proposed that polyphenolic antioxidants are useful as anti-aggregation substances in targeting Aβ peptides [14,18]. Furthermore in vivo studies showed that oral administration of grape derived polyphenolic extracts decrease amyloid plaque and can improve memory and cognitive ability [26,27]. Molecular dynamics (MD) simulations have been used here for studying atomic details of Aβ peptide interaction with natural excipients [15]. Based on the MD simulation data, Lemkul et al. [17] revealed that the flavonoid morin can bind to the ends of the fibrils and preclude the attachment of an incoming peptide, followed by the reduction of the polymerization rate. The effects of various amyloid aggregation inhibitors on different segments of Aβ peptides have been widely examined by MD simulations [15,17,28].
Both experimental and theoretical literatures currently suggest that flavonoids-related compounds inhibit early stage of Aβ aggregation which also disrupt the late stages of the of Alzheimer's cross-β fibrils, providing the motivation for the present study. The structural feature of oligomeric protofilament state of Aβ17-42 peptides and destabilizing cross oligomeric-β by the interaction with flavonoids (pathways 1 and 1') have been extensively studied using MD simulation methods [17,21,23,29]. Besides, despite of wide studies on Aβ peptides' fibrillogenesis, the conformational stability and affinity of the fibril Aβ17-42 peptide in the case of  1'). The excipients interact with assembled cross-β fibrils of Aβ17-42 peptides and reversibly destroys the fibril structure of the peptide that can "retard" late stage of amyloid plaques. Pathway 2 is based on the almost irreversible fibrils suppression by converting β-sheet conformations. Reversible destroyed assembled cross-β fibrils are accompanying irreversible β-sheet conformational changing of the Aβ17-42 peptide that can result in crush and suppress late stage of amyloid plaques. interaction with excipients remains unclear. In this study, hypothesizing that an excipient interaction with Aβ17-42 peptide (monomeric strands of cross-β protofilaments) destroys the fibril's formation and suppresses the late stage of amyloid plaques (suppression of late stage plaques, pathways 2). We aimed to investigate the effectiveness of three different flavonoids including myricetin (with 6 OHs), morin (with 5 OHs) and flavone (with no OH) in the interaction and destabilization effects on the conformation of monomeric cross-β element of Aβ17-42, using MD simulations method.

Aβ peptides setup
Aβ17-42 peptide works as a suitable model of fibrils and is similar to model systems that previously applied for MD simulation studies [17]. The highly hydrophobic monomer of Aβ17-42 peptide would rapidly evolve into fibrils related to amyloid's plaques with no soluble intermediate forms [30]. We believe the monomeric structural model of Aβ17-42 peptide is the best representative of the fibrils' cross-β strands conformations of late stage plaques, pathways 2, in order to study the flavonoids interaction. Therefore, the monomeric subunit of Aβ17-42 fibril, a protofilament structure determined by solid-state NMR (PDB code 2BEG) [31] was applied for MD simulations. To have an uncharged N-termini, amino terminus Leu17 residue of Aβ17-42 peptide was acetylated before any MD simulations.

Ligands geometries optimizing
The initial geometries of different flavonoids including myricetin, morin and flavone were generated by Hyperchem 7 software. The structures were pre-optimized using molecular mechanic MM + force field. Then the geometries optimization was obtained with the semiempirical AM1 followed by ab initio DFT method with Becke's three-parameter hybrid functional (B3LYP) level using 6-311+G (d,p) basis set. Optimized conformations of myricetin, morin and flavone were provided in Fig 2. The optimized structures were presented into PRODRG 2 server [32] without any coordinate changing to achieve ligands' topology. Some of the main parameters in topology files such as partial charges were handled based on DFT optimized data. Therefore, the refined topology of ligands from PRODRG 2 server was applied for using MD simulations.

Simulations setup
A number of 13 diverse systems were prepared and 100 ns simulations of each system have been conducted using GROMACS 5.0.4 package. The details for each system were explained in Table 1. All simulations were conducted using GROMOS96 53A6 force field [33] by applying SPC water model [34]. The systems were separately solvated in explicit water molecules that extend up to 10 Å from any edge of the cubic box to the solute atoms. Previously, different ligands with 10:1 mole ratio were examined to test their anti-aggregation properties on Aβ42 using MD simulation method, such as EGCG peptide [35], naproxen [36], ibuprofen [37] and morin [15]. For theoretical examination of any concentration dependent effects of different flavonoids with Aβ17-42 interactions, wide ranges of mole ratios were applied for flavonoids/ Aβ17-42 complexes including 1:1, 2:1, 6:1 and 10:1. The model containing 2:1 mole ratio of flavonoids/Aβ17-42 is almost similar to in vitro conditions. For example, 2:1 mol ratio of morin/Aβ was used as the highest concentration of morin to explore anti-amyloidogenic and fibril destabilizing effects of polyphenols in vitro [18]. Also, theoretical morin-treated systems containing 2:1 mol ratio of morin/Aβ42 showed that morin inhibits β-peptide aggregation by altering tertiary interactions [17]. Therefore, we have applied different mole ratios of flavonoids/Aβ17-42 from unity to 10 fold higher that could help to examine any possible concentration dependent effects of flavonoid-Aβ interactions. The simulation systems consist of Aβ and Aβ complexes with myricetin, morin and flavone molecules. All systems were neutralized by NaCl counterions. The concentration of NaCl in all simulations systems was 100 mM, by adding appropriate number of Na + and Clions. In all cases, short-range nonbonded interactions were truncated at 1.2 nm, applying long-range dispersion correction to the energy and pressure terms to account for truncation of the van der Waals interactions. The Particle Mesh Ewald (PME) method was utilized for the calculations of long range electrostatic interactions. The LINCS algorithm [38,39], was used for all bond constraints, allowing an integration time step of 2 fs. Periodic boundary conditions were applied in all directions. The temperature of the systems was preserved at 300 K by using Berendsen weak coupling method and pressure was maintained at 1 bar by utilizing Parrinello-Rahman barostat in constant pressure ensemble. All systems were energy-minimized using the steepest descent method. The minimized systems were equilibrated under NVT (constant volume) and NPT (constant pressure)  [40] and UCSF Chimera [41] were used to visualize the simulation results. All of the molecular graphical presentations were generated using the UCSF Chimera package.

MM-PBSA binding free energy
Gibbs free energies of flavonoids/Aβ17-42 complexes were calculated using MM-PBSA (molecular mechanics Poisson Boltzmann surface area) tool of Gromacs method [42]. The binding affinities were investigated during equilibrium phase by taking 200 snapshots at an interval of 100 ps from 80 to 100 ns MD simulations. Therefore, for each simulated system, total 200 snapshots were taken from the last 20 ns of the trajectory. This method is widely used for estimating binding free energies of ligand-protein complex in solvent from the snapshots of MD trajectory based on the estimation by; Where G complex is the total free energy of the Aβ-flavonoid complex, G Aβ17-42 and G flavonoid are total energy of separated Aβ17-42 and flavonoid in solvent, respectively.

Assessment of conformational stability of Aβ17-42
The conformational features of free and complex Aβ17-42 were evaluated by the root-mean squared deviation (RMSD) and root mean square fluctuation (RMSF) assays. The backbone RMSD of Aβ17-42 in different systems has been shown in  (Fig 4). RMSF deviations of Aβ-myricetin, Aβ-morin and Aβ-flavone systems are the smallest at 10:1 mole ratio. Therefore, it is expected that the destabilizing Aβ structure and toxicity reduction effectiveness by different natural flavonoids will be associated with similar functionality at higher molar ratios. In other words, different flavonoid compounds have similar conformational feature change effects at high molar ratios.

H-bond formation possibility between Aβ17-42 and flavonoids
Experimental and theoretical studies have suggested that the strong affinity between protein/ polyphenols is driven by hydrogen-bonding interactions and possibly hydrophobic contacts [43]. These groups of flavone molecule will be attracted to partial positive charge of H atom in peptide amine (NH 2 ) since the nitrogen of peptide amine is highly negative. Computations of the electronic structure of amides and the peptide bond showed that the partial negative charge values of the nitrogen atom in peptide bonds are varying between the range of -0.20 to -1.23 depending on the applied force field and/or basis set [44]. The partial negative charge on nitrogen atom is because of the relatively electron negative nitrogen atom attracting electrons toward itself from adjacent atoms. Therefore, carbonyl and ether groups of flavone act as H-bond acceptors, accepting H atom from NH 2 group of Aβ17-42 peptide backbone. But, the hydroxyl groups of myricetin and morin have the original possibility to interact with the peptide's carbonyl (C = O) and amine (NH 2 ) groups serving as the hydrogen bond donors. Consequently, data clearly denote that the flavonoids with hydroxyl group (myricetin and morin) makes H-bonds with both peptide's carbonyl and amine groups but flavone with no hydroxyl substitution group forms H-bonds only with peptide's amine portion. Formation of intermolecular H-bonds means the disruption of peptide structure by losing peptide's intramolecular backbone H-bonds between carbonyl and amine groups as important parts for the helical/sheet structure content and stability of proteins. Therefore, using polyphenols such as myricetin or morin would be more effective in destabilizing β-sheet structures of peptide compared to flavone. Analysis of the histogram plots of H-bonds in Fig 6 indicates that the normal distribution of H-bonds is corresponding to mole ratios of flavonoids. Increasing the number of flavonoids resulted in better distribution and their normality seems good. At low mole ratios of ligands/Aβ17-42 (1:1 and almost 2:1), the H-bonds histograms of all myricetin, morin and flavone have non-normal distributions. Polyphenolic myricetin and morin show better normality of H-bonds distribution compared to flavone. At the mole ratios of 6 and 10, both myricetin and morin show normal H-bonds distributions. Nevertheless, flavone shows normal distribution of H-bond histogram only at the mole ratio of 10. It should be noted that, the normal distribution is very important in probability theory [45]. Therefore, based on the probability theory, flavonoids with more OHs are better candidate to interact and disrupt Aβ structure because of providing Gaussian distribution at the lower concentrations.  Table 2. The favorable interaction energies including van der Waals, electrostatic and non-polar solvation energy (solvent accessible surface area, SASA energy) negatively contribute to the total binding free energies. These energies altogether contribute to the flavonoids/Aβ17-42 complex stability. The results indicate that polyphenolic myricetin and morin possessed more negative binding free energies than flavone for all of the applied mole ratios. In all cases, the contribution of van der Waals interactions in the total interaction energy is much larger than the electrostatic energies. Also, non-polar SASA energies contribute less than electrostatic energies in the total binding energies of all systems. The transfer energies of compounds from a nonpolar solvent to water are described by SASA-dependent energy function [42]. Then, the hydrophobic nonpolar hydration free energy, ΔG np , is decomposed as; where ΔG SASA is the cavity hydration free energy, and ΔG vdw is the free energy for establishing the solute-solvent van der Waals dispersion interactions. However, the cavity formation in water has been frequently used as a model for studying the hydrophobic effect [46]. Hence, non-polar solvation energy is the SASA-dependent hydrophobic solvation energy. While the solute-solvent van der Waals interactions are dominant in Table 2, the pure hydrophobic interactions related to SASA-dependent energy are trivial in all complexes. Moreover, logP values (lipophilicity parameter) of different species refer to larger hydrophobic property of flavone (between 2 and 2.5-fold) compared to polyphenolic flavonoids of myricetin or morin (logP flavone 3.56; logP myricetin 1.42; logP morin 1.54; https://www.ncbi.nlm.nih.gov/ pccompound). Although flavone is a more hydrophobic compound, its binding energy ( Table 2) is certainly less than myricetin or morin compounds. It means that the hydrophobic interactions of flavonoids with Aβ17-42 should be negligible. Moreover, binding energy plots of all systems (Fig 7) clearly indicate the hyperbolic behavior of flavonoids binding. Since noncooperative interaction is hyperbolic and the cooperative hydrophobic response is sigmoidal shape [47,48], then the nonsigmoid-type of interactions in all applied flavonoids denote the insignificant role of hydrophobic interactions for myricetin, morin and flavone with Aβ17-42. Consequently, we believe the role of H-bonds is dominant as clearly discussed in previous section. Table 2. Summary of calculated molecular interactions of the flavonoids with Aβ17-42 using MM/PBSA tools. Subscripts denote the molar ratios for flavonoids/ Aβ17-42. Aβ-Myc: Aβ-Myricetin, Aβ-Mor: Aβ-Morin, Aβ-Flv: Aβ-Flavone.

Secondary structure analysis
Secondary structure content including alpha helical and β-sheet structures are essential for amyloidosis and neurotoxicity [49,50]. Any agent with capability to prevent and inhibit conversion of alpha helical to β sheet structure could be a useful candidate for AD treatment [51].
To evaluate any affinity of the applied compounds to disrupt the content of β sheet structure in Aβ17-42 peptide, secondary structure analysis was accomplished by DSSP [52]. Fig 8 clarifies the temporal expansion of the secondary structure content for different systems along 100 ns trajectory. The results of secondary structure contents of systems entangled with different mole ratios of morin are almost similar to myricetin systems. The morin and myricetin treated systems exhibit that the β-sheet conformations have been disappeared at higher mole ratios. In flavone treated systems, the contents of β-sheet structures between β1 and β2 regions are reduced by increasing flavone's mole ratios. Data certainly indicate that losing the β-sheet contents of Aβ corresponds to the ligands mole ratios mainly the normality of intermolecular Hbonds distribution. Systems with normal distribution of intermolecular H-bonds clearly correspond to the suppression of any appearance of β-sheet structures. As discussed and examined at Fig 6, two mole ratios of myricetin and morin (6 and 10) and one mole ratio of flavone (10) denote normality manner in intermolecular H-bonding, namely, normal systems. The normal systems are active against β-sheet structures, resulted in losing beta sheet conformations. All of treated systems with different flavonoids exhibit other types of secondary structures mainly coil and bend conformations. S1 Table clearly shows that the initial β-sheet, β-bridge, bend and turn contents in untreated system are exchanged for random coil elements in complex systems. In this process morin content system is mainly accompanying conversion of β-sheet The role of phenolic OH groups of flavonoid compounds to suppress amyloid mature fibrils structure into bend but myricetin resulted in β-sheet conversion into coil conformations. The complete loss of beta structures mainly in morin and myricetin treated systems are more noticeable, indicating that, polyphenolic flavonoids such as morin and myricetin could have the greatest impact on secondary structure content and suppress β-sheet formation toward βamyloids.

Tertiary structure and density profile of the systems
Density function, such as mass or number densities, an analysis tool that can be derived from MD trajectories to compute one-dimensional density profiles of molecular systems. In order to evaluate any possible changes in tertiary structure of Aβ peptide by entangling various mole ratios of flavonoids, 1-D projections of peptide mass density profile along the z axis was analyzed and monitored at Fig 9. The flavonoid free system has normal distribution profile meaning that all parts of Aβ peptide structures are benefited from similar structure density at the definite range of Δz. Applying "types of flavonoids" and their "mole ratios" resulted in changing the peptide mass density profile. Myricetin and morin at the ratios of 6 and 10 caused Aβ peptide completely to unfold with increasing the peptides portions of thickness from dense core (increasing |Δz| and appearing new peaks). The process is concomitant to loss of the normal distribution of peptide mass density profile. It means that myricetin and morin caused 3D structure of Aβ peptide completely to change at the molar ratios of 6 and 10. Almost similar results were achieved for flavone at the molar ratio of 10. While, all of the treated systems with one and two flavonoids of myricetin, morin and flavone showed higher density profile. Increasing peptide mass density at low mole ratios of ligands did not demonstrate any noticeable new peak. The rigidity of Aβ peptide was obviously increased in complexed systems containing one and two ligands at all segments of the structure. Different snapshots of free Aβ and in complex with different mole ratios of myricetin, morin and flavone were prepared and shown at Fig 10 for the MD simulations. Although all systems were initially prepared by placing ligands on U-shaped conformation surface of Aβ peptide (surrounding~4 Å), the ligands were shifted inside of the peptide throughout simulations. This is mainly because of ligands tendency to interact with Aβ peptide and mainly to participate intermolecular backbone Hbonds formation with negative binding energies. The non-normal distribution of H-bonds at low mole ratios of flavonoids (Fig 6) is accompanying to firm trivial segments of Aβ conformation. Therefore, the systems containing one and two flavonoids resulting mass densities increase at small range of ΔZ. But, the systems with normal H-bonds distribution at high mole ratios of flavonoids correspond to involve almost all segments of peptides by H-bond formation capability interactions at a wider range of ΔZ (Fig 6 and Fig 9). In addition, the geometric form of Aβ peptide in different snapshots clearly support the density profile and H-bond histogram data. It shows that at higher mole ratios of flavonoids the conformation of Aβ peptide completely changed toward embracing flavonoids. Flavonoids penetrate the core of the Aβ17-42, forming large self-aggregated clusters. Because of the formation of Aβ17-42-ligands clusters, the peptide mass density profile was changed and caused to lose the sharpness and maximum values of mass density concerning wide range of ΔZ. Finally, the important findings of this manuscript are concluding that; a) all of applied flavonoids including myricetin, morin polyphenolic and flavone are mainly capable of binding to all regions of Aβ peptide including β1, β2, and turn segments. These interactions are favorable and negatively contribute to the total binding free energies. Binding affinity of flavonoids to different segment of peptide for disrupting Aβ peptide structure is essentially because of the possessing H-bond donor and acceptor groups in flavonoids structure including ether, carbonyl and hydroxyl. b) polyphenolic flavonoids with more functional OHs (myricetin and morin) are more active against β-sheet structures destabilization capability. c) The main effects of flavonoids against beta amyloid or Aβ aggregation correspond to the good distribution of The role of phenolic OH groups of flavonoid compounds to suppress amyloid mature fibrils H-bonds in the complex systems. The good distribution of H-bonds corresponds to both of the flavonoids' mole ratios and the number of their functional OH groups. Hence, flavonoids with more functional OH groups and Gaussian distribution at low mole ratio are better applicable to interact and disrupt Aβ structure. Since polyphenolic flavonoids are beneficial in inhibiting β-sheet and preventing the elongation of Aβ oligomers, the compounds myricetin and morin can be introduced as a useful candidate for AD treatment.
Supporting information S1 Table. Percentage of the secondary structural contents of Aβ-monomer in each system during 100ns simulation. (DOCX)