The Effects of A21G Mutation on Transmembrane Amyloid Beta (11

Article pubs.acs.org/JPCB

The Effects of A21G Mutation on Transmembrane Amyloid Beta (11− 40) Trimer: An In Silico Study Son Tung Ngo,*,†,‡ Minh Tung Nguyen,§ Nguyen Thanh Nguyen,∥ and Van V. Vu*,⊥ †

Computational Chemistry Research Group, Ton Duc Thang University, Ho Chi Minh City, Vietnam Faculty of Applied Sciences, Ton Duc Thang University, Ho Chi Minh City, Vietnam § Binh Duong University, Thu Dau Mot City, Binh Duong Province, Vietnam ∥ Department of Theoretical Physics, University of Science, Ho Chi Minh City, Vietnam ⊥ NTT Hi-Tech Institute, Nguyen Tat Thanh University, Ho Chi Minh City, Vietnam ‡

S Supporting Information *

ABSTRACT: Familial Alzheimer’s disease (FAD) is passed down in family, which account for 2−3% of about 40 million AD cases worldwide. The Flemish (A21G) mutant of amyloid β (Aβ) exhibits unique properties among all hereditary mutants of FAD, including the lowest aggregation rate. Recent studies showed that Aβ oligomers play a key role in Alzheimer’s disease (AD) pathogenesis. They could insert themselves in brain cell membrane, disrupting the membrane integrity and ion homeostasis. However, experimental studies of transmembrane Aβ oligomers have been limited due to their intrinsic heterogeneity. In this work, we extensively studied the A21G mutant of the transmembrane 3Aβ11−40 (A21G 3Aβ11−40) using temperature replica exchange molecular dynamics (REMD) simulations. Results provide detailed information on the conformational distribution and dynamics of transmembrane A21G 3Aβ11−40. Minimal local change from A to G leads to significant conformational changes and wider free energy holes on the free energy surface as well as altered surface charges that lead to weaker affinity to the dipalmitoylphosphatidylcholine (DPPC) lipid bilayers. These results are consistent with experimental data that showed that A21G mutants of Aβ peptides have lower aggregation rates and membrane binding rates.

■

INTRODUCTION Alzheimer’s disease (AD) is a highly prevalent chronic neurodegenerative disease that causes memory loss in more than 40 million people worldwide.1,2 AD usually starts early and slowly progresses over 20−30 years before clear symptoms are observed in patients at the age of ca. 65. Amyloid hypothesis, or Amyloid cascade hypothesis, is the most widely accepted model for AD pathogenesis that is supported by a substantial amount of preclinical and clinical data.1,2 In this cascade, oligomers of beta-amyloid (Aβ) peptides activate microglia and astrocytes, which leads to progressive synaptic and neuritic injury. Aβ oligomers may also directly injure synapses and neurites of brain neurons. Over the last two decades, Aβ oligomers have been extensively studied.1−6 Several methods have been developed to detect the cellular localization of Aβ oligomers.3 Besides interacting with synaptic receptors, Aβ oligomers have also been found in neuron cell membrane. Membrane bound Aβ oligomers disrupt the membrane, as well as form pore-like structures that alter calcium ion homeostasis.6−8 While the pore structures comprising large numbers of amyloid peptides have been detected in the membrane using atomic force microscopy,9 how these pores are formed from Aβ monomers © 2017 American Chemical Society

and smaller oligomers has not been elucidated. In addition, Aβ oligomers are also found in the cytosol, within organelles, as well as at organellar membranes, causing damage to the mitochondria, altered proteolysis and calcium homeostasis, elevated ER stress, and apoptosis.3 Aβ trimer is regarded as one of the most neurotoxic forms of low-molecular-weight Aβ oligomers.3,10 Inhibition of Aβ oligomerization is one of the most common approaches in developing AD therapy.3,11 However, this approach is hindered by the lack of structural understanding of Aβ oligomer.3,10,12,13 Because of the inherent heterogeneity and metastability of Aβ oligomers, structural determination of both soluble and transmembrane Aβ oligomers has yielded little success. Thus, computational methods, especially molecular dynamics simulations, have been instrumental in characterizing Aβ oligomers.14−16 Familial AD (FAD) is the form of AD passed down through family and accounting for 2−3% of all cases of AD. Most of the mutations in FAD occur in the Aβ precursor protein (APP) external of the Aβ sequence. However, mutations in the Received: June 15, 2017 Revised: August 17, 2017 Published: August 17, 2017 8467

DOI: 10.1021/acs.jpcb.7b05906 J. Phys. Chem. B 2017, 121, 8467−8474

Article

The Journal of Physical Chemistry B hydrophobic region of Aβ peptides, including A21G (Flemish),17 E22Q (Dutch),18 E22K (Italian),19 E22G (Arctic),20 and D23N (Iowa),21 have drawn significant interest due to altered biochemistry. Among these mutants, Flemish mutant has unique properties, including the lowest aggregation rate22 and increased Aβ production.23 Experimental and computational studies have shown that A21G mutation causes conformational changes in Aβ42 monomer.24,25 Computational studies have also been carried out for A21G mutants of Aβ40 and Aβ42.26 Nevertheless, the effects of A21G mutation on the neurotoxic transmembrane Aβ oligomers, including trimer, have not been studied. In an attempt to understand the conformations of the neurotoxic Aβ trimer in brain cell membrane, we recently carried out extensive temperature replica exchange molecular dynamic (REMD) simulations of Aβ11−40 trimer (3Aβ11−40) in both solution and dipalmitoylphosphatidylcholine (DPPC) lipid bilayer, a membrane mimic. Results revealed significant differences in structural and kinetic properties of 3Aβ11−40 in the DPPC lipid bilayer27 compared to that in solution.28 In this work, we aimed to understand the effects of the A21G mutation on the conformations of transmembrane 3Aβ11−40. The mutant A21G Aβ11−40 trimer (A21G 3Aβ11−40) with starting structure obtained from Aβ fibril was inserted into a DPPC lipid bilayer. The solvated mutant trimer was then simulated using REMD simulations with 32 different temperatures ranging from 321 to 423 K. The conformational change in the mutant trimer was monitored over 400 ns of REMD simulations. The metastable structures of transmembrane A21G 3Aβ11−40 were deduced using the combination of free energy surface and clustering methods. Our results provide detailed information on the conformations of transmembrane A21G 3Aβ11−40 and how they differ from those of transmembrane 3Aβ11−40 obtained in our previous studies,27 which further our understanding of AD pathogenesis.

Figure 1. First snapshot of the T-REMD simulations of the solvated system, in which an A21G 3Aβ11−40 penetrates the DPPC membrane bilayers. The mutation points are noted.

Collision Cross Section (CCS). The IMPACT39 was employed to compute the mutant trimer CCS. Computational Analysis Tools. The clustering method38,40 was carried out with a Cα RMSD of 0.12 nm. The nonbonded contact between heavy atoms of different residues were counted if the distance is smaller than 0.45 nm. The hydrogen bonds between different residues were recorded when the acceptor (A)−hydrogen atom (H)−donor (D) angle is larger than 135° and the A−D distance is smaller than 0.35 nm. The lipid order parameters were computed using the 1 formula SCD = 2 3 cos2 θ − 1, where C is carbon, D is deuterium, and θ is the angle between the molecular axis given by the Ci−1 − Ci+1 vector and the bilayer normal, which is investigated over the computational period. The results were averaged over the membrane during simulation.

■

METHODS Temperature-REMD Simulations. The conformation of transmembrane 3Aβ11−40 inserted in DPPC membrane bilayers29 was taken from a previous study,27 from which A21G 3Aβ11−40 was generated using PYMOL tools.30 The mutant trimer was then represented using the united atom GROMOS 53a6 force field.31 The system was solvated using the simple point charge (SPC) water model.32 The solvated system was neutralized with three Na+ ions. The starting conformation of transmembrane A21G 3Aβ11−40 is shown in Figure 1, in which Na+ ions are hidden and the mutant points are noted. The entire solvated transmembrane A21G 3Aβ11−40 system consists of 16,795 atoms, including 3293 water molecules and 3 Na+ atoms. GROMACS version 5.1.333 was employed to simulate the system. Free Energy Perturbation (FEP) Method. The binding free energy between the mutant trimer and DPPC bilayer was predicted using the FEP method34 as described in a previous study.27 Secondary Structure Analysis. The secondary structure parameters of A21G 3Aβ11−40, including coil, beta, turn, and helix contents, were predicted using DSSP.35,36 Free Energy Surface (FES). The free energy surface of the mutant trimer was constructed using the “gmx sham” tools37,38 of GROMACS with root-mean-square deviation (RMSD) and radius of gyration (Rg) serving as reaction coordinates.

■

RESULTS AND DISCUSSION REMD Simulations of Transmembrane A21G 3Aβ11−40. Currently, REMD is one of the most powerful sampling methods to investigate the protein folding/misfolding problems, including the oligomerization of Aβ peptides.41−44 REMD has been utilized to produce appropriate results for Aβ oligomers both in solution and in membrane.27,28 Because simulating the Aβ aggregation starting from monomers would take a tremendous amount of CPU time, we used the Aβ trimer conformation in previously deduced fibril-like structure27,28 as the initial structure for A21G 3Aβ11−40. In this context, the transmembrane A21G 3Aβ11−40 was simulated using the REMD method with 32 different replicas involving 32 different temperatures ranging from 321 to 423 K. The individual temperatures were generated using a Web server.45 A total of 12,800 ns of MD simulations were achieved, in which 400,000 exchange times were completed. The results were obtained in the last 150 ns of REMD simulations. The first 250 ns of REMD simulations was removed to dodge any starting bias. 8468

DOI: 10.1021/acs.jpcb.7b05906 J. Phys. Chem. B 2017, 121, 8467−8474

Article

The Journal of Physical Chemistry B The mean exchange rates between bordering clones were found to be appropriate values, ranging from 19 to 25% (Figure S1). Examples of the dispersion of clones over the temperature space are shown in Figure S2. The results indicated that every clone disperses in the entire temperature space. Furthermore, the secondary structure terms, including coil, beta, turn, and helix contents, range from 44 to 76%, 24 to 51%, 0 to 5%, and 0−0%, respectively, at 250 ns in the temperature space (Figure 2). These dispersions indicate that our simulations were

Figure 3. Convergence of REMD simulations at 324 K. The black and red lines represent the populations of measured metrics in the time windows of 250−320 and 330−400 ns, respectively. Green lines indicate the measure metrics in the time window of 200−350 ns in the wild-type 3Aβ11−40 (reproduced from ref 27 with permission from the Royal Society of Chemistry).

higher than 1.45 nm, while that in 3Aβ11−40 has a Rg smaller than 1.45 nm (Figure 3). Second, the average RMSD of A21G 3Aβ11−40 (0.69 ± 0.11 nm) is significantly higher than that of 3Aβ11−40 (0.47 ± 0.07 nm). The majority of RMSD in 3Aβ11−40 is smaller than 0.6 nm, while in A21G 3Aβ11−40 the main population (>60%) has an RMSD of 0.69 nm and a sizable population has an RMSD of ∼0.92 nm (Figure 3). Third, the average surface area of A21G 3Aβ11−40 (68.38 ± 2.58 Å2) is larger than that in 3Aβ11−40 (64.73 ± 3.07 Å2). Lastly, A21G 3Aβ11−40 does not form a well-defined population with D23− N27 polar contacts, as found in 3Aβ11−40 (see below). D23− N27 polar contacts have been shown to play an important role in stabilizing the structures of Aβ peptides and amyloid precursor protein.27,53 These differences indicate that A21G 3Aβ11−40 has more bulky and widespread structures that are less stable than those in 3Aβ11−40. These results are consistent with previous experimental studies that showed that A21G Aβ folds more slowly than wild-type Aβ.22 “Per-Residue” Distribution of Secondary Structures. Figure S5 shows the per-residue distribution of the secondary structure of transmembrane A21G 3Aβ11−40 obtained in the last 150 ns of REMD simulations at 324 K. This distribution is very similar to that previously found for the wild-type transmembrane 3Aβ11−40.27 All three chains can be divided into five main sequences, in which sequences 11−13, 20−30, and 38−40 exhibit mostly random coil structures, while sequences 14−19 and 31−37 exhibit rigid β-structures (Figure S5). The turn structures are distributed in the three coil-dominated sequences. The negligible amount of helical structure forms in the middle of chain A (residues 24−28). Further details on the overall structures of A21G Aβ11−40 are discussed below. Interactions of Aβ11−40 Chains with the Other Chains and with the Lipid Bilayer. The maps of side chain contacts and hydrogen bonds between neighboring chains transmembrane A21G 3Aβ11−40 were constructed over the equilibrated snapshots (Figure S4). These maps are similar to

Figure 2. Dispersion of secondary structure terms including coil, beta, turn, and helix content at 250 ns of temperature REMD simulations. The results were obtained from DSSP tools.

unbiased. During the simulation, the lipid bilayer was stable (Figure S3), as observed in the case of transmembrane 3Aβ11−40.27 Here we will present and discuss the results on transmembrane A21G 3Aβ11−40 in comparison with wild-type transmembrane 3Aβ11−40. The investigated snapshots were collected between 250 and 400 ns of REMD simulations at 324 K, since 315 K is the phase transition temperature of the membrane DPPC lipid bilayer. Our simulations reached the equilibrium states at 324 K after 250 ns of REMD simulations, as all measured metrics remained unchanged in two different time windows of 250−320 and 330−400 ns (Figure 3). These metrics include coil content (57 ± 4%), beta content (41 ± 4%), turn content (2 ± 2%), helix content (∼0%), radius of gyration (Rg) (1.46 ± 0.03 nm), RMSD (0.69 ± 0.11 nm), and surface area (68.38 ± 2.58 Å2), as well as D23−N27 polar contacts. The helix content is negligible, which is consistent with previous studies showing that α-helices are intermediates of the Aβ aggregation process.13,46−49 Notably, the beta contents deduced from our simulations for both 3Aβ11−4027 and its A21G mutant are similar to those experimentally found for transmembrane Aβ oligomers50,51 and soluble 3Aβ11−40.52 The coil and turn contents found for transmembrane A21G 3Aβ11−40 are also similar to those in transmembrane 3Aβ11−40. Upon close comparison of other metrics of transmembrane A21G 3Aβ11−40 and 3Aβ11−40,27 we found significant differences. First of all, the average Rg of A21G 3Aβ11−40 (1.46 ± 0.03 nm) is larger than that in 3Aβ11−40 (1.42 ± 0.02 nm).27 The majority (68%) of the A21G 3Aβ11−40 population has a Rg 8469

DOI: 10.1021/acs.jpcb.7b05906 J. Phys. Chem. B 2017, 121, 8467−8474

Article

The Journal of Physical Chemistry B those previously found for wild-type transmembrane 3Aβ11−40.27 Corresponding to the rigid beta-content regions of the trimer, the constituting chains frequently adopt both side chain and hydrogen bond contacts. These regions form the central hydrophobic core of the trimer, which are highly likely to play the key role in stabilizing the trimer as previously found for other oligomers.54 The D23−N27 polar contacts are known to contribute significantly to the stabilization of the Aβ monomers and fibril structures in solution.55−57 The amount of polar contacts in Aβ oligomers is quantified in these studies as the populations of the chains having the distances between the polar groups of these residues equal to or less than 0.46 nm (Figure 4). In wild-type

Figure 5. Probability of intermolecular contacts between phosphorus atoms of bilayer DPPC lipids and heavy atoms of truncated A21G 3Aβ11−40 (A, this work) and wild-type 3Aβ11−40 (B, reproduced from ref 27 with permision from the Royal Society of Chemistry). The results are investigated at 324 K throughout the last 150 ns REMD simulations.

phosphate groups of DPPC lipid bilayers increases upon A21G mutation. Furthermore, there are also significant changes in the contacts of some key residues that form the most contacts with the membrane. Upon A21G mutation, residues 11, 16, and 28 in all chains form contacts in almost all of the conformations. K16 and K28 were previously shown to form the most regular contacts with lipid phosphate headgroups.58,59 Notably, K28 in chain C forms contacts with the membrane in 30 and ∼99% populations of 3Aβ11−40 and A21G 3Aβ11−40, respectively, resulting in the decrease of the D23−K28 polar contact from 71 to ∼1%. Likewise, the increase in N27− membrane contacts can also be related to the decrease in D23− N27 polar contact in chain C. Altogether, A21G mutation results in significant changes in peptide−membrane contacts and decreases in essential salt bridges, which likely leads to the decrease in aggregation rate.22 Free Energy Surface and the Representative Structures of A21G 3Aβ11−40. The metastable conformations of transmembrane A21G 3Aβ11−40 were derived from the combination of free energy surface and clustering methods that were previously applied on both solvated and transmembrane Aβ11−40 trimer.27,28 The FES of A21G 3Aβ11−40 is shown in Figure 6, which contains five minima, designated as A1−A5, at (RMSD, Rg) coordinates of (0.66; 1.43), (0.54; 1.48), (0.69; 1.50), (0.70; 1.47), and (0.91; 1.45), respectively. As described in the previous section, except for the A2 region, most of the A21G 3Aβ11−40 conformations fall in the range with RMSD higher than 0.6 nm, while all conformations of 3Aβ11−40 have a RMSD smaller than 0.6 nm. A2 corresponds to the population with a RMSD peak around 0.55 nm. A1, A3, and A4 correspond to the RMSD peak around 0.69 nm. A5 corresponds to the RMSD peak near 0.91 nm. The structures of A1−A5 are shown in Figure 7. The snapshots of REMD simulations revealed that each chain of A21G 3Aβ11−40 contains two β strands that are well separated by the random coil region in the middle of the chain. These β strands form strong interchain interactions, resulting in the overall β core of the trimer with the longitudinal axis parallel to the lipid bilayer normal. The random coil regions are placed at

Figure 4. Distributions of the distances between the charge groups of D23 and N27 (A) and D23 and K28 (B) in chain A (black), chain B (red), and chain C (green) of transmembrane mutant A21G 3Aβ11− 40.

transmembrane 3Aβ11−40, the amounts of D23−N27 polar contacts in chains A, B, and C are approximately 77, 60, and 3%, respectively.27 The amounts of D23−K28 polar contacts in these chains are 1, 18, and 71%, respectively. In contrast, A21G 3Aβ11−40 does not form well-defined populations of these polar contacts. In this mutant trimer, the amount of D23−N27 polar contacts in chains A, B, and C is 30, 4, and 9%, respectively (Figure 4A). The D23−K28 polar contacts are not observed in this mutant (0, 0, and ∼1% in chains A, B, and C, respectively) (Figure 4B). These results suggest that A21G mutation would destabilize transmembrane Aβ11−40 trimer, resulting in significant conformational changes. The decrease in the amount of D23−N27 polar contacts and the lack of the D23−K28 polar contacts in transmembrane A21G 3Aβ11−40 is also consistent with experimental studies that showed that A21G Aβ peptide folds more slowly than corresponding Aβ peptides.22 Figure 5 shows the map of per-residue interactions of A21G 3Aβ11−40 with the phosphate groups of the DPPC lipid bilayer (A) in comparison to that in 3Aβ11−40. There are some notable changes in the membrane contacts of several regions of 3Aβ11−40. In A21G 3Aβ11−40, the residues 20−23, which form random coils in the middle of the chain (Figure S5), form significant contacts with the lipid bilayer. In contrast, these residues in 3Aβ11−40 do not form sizable contacts with the membrane. In addition, there are significantly less contacts of the residues 17−18 and 37−40 with the membrane upon A21G mutation. Overall, the total amount of contacts to the 8470

DOI: 10.1021/acs.jpcb.7b05906 J. Phys. Chem. B 2017, 121, 8467−8474

Article

The Journal of Physical Chemistry B

Figure 6. FES constructed with two reaction coordinates including Rg and RMSD for transmembrane A21G 3Aβ11−40 (A) and for transmembrane 3Aβ11−40 (B, reproduced from ref 27 with permission from the Royal Society of Chemistry). There are five minima that are noted from A1 to A5 that structures are shown in Figure 7. The results were obtained from the analysis of all equilibrated snapshots.

either end of the β core, forming contacts with the lipid phosphate head groups. This result is consistent with previous studies that showed the formation of β structures of Aβ peptides inside membranes51 and random coils on the membrane surface.50,60 The secondary structure metrics of A1−A5 are shown in Table 1. Overall, the average contents of coils, beta structure, turns, and helices of transmembrane A21G 3Aβ11−40 are almost the same as those in transmembrane 3Aβ11−40.27 However, average CCS (1400 Å2) and surface area (SA) (69.27 nm2) values of A21G 3Aβ11−40 are larger than those in 3Aβ11−40 (1340 Å2 and 64.18 nm2, respectively).27 In A1, A2, and A4, the beta strands form two beta-sheets that are facing each other in antiparallel fashions (Figure 7). In A3, these two sheets are almost perpendicular to each other, resulting in a larger Rg value (Figure 6), as well as significantly larger CCS, and SA (Table 1). Notably, in A5, these two sheets are almost placed in the same plane, resulting in a larger sheet with a significantly higher RMSD value (Figure 6). The free energy barriers deduced from free energy surfaces for both A21G 3Aβ11−40 and wild-type transmembrane 3Aβ11−4027 are 15.3 kJ/mol (Figure 6). While all minima in 3Aβ11−40, except for M5, are located close to each other, the five minima in A21G 3Aβ11−40 are well separated. The “free energy holes” on the FES of A21G 3Aβ11−40 are significantly larger than those in wild-type 3Aβ11−40 (Figure S6). The number of conformations having a free energy higher than 13.3 kJ/mol in A21G 3Aβ11−40 almost doubles that in wild-type 3Aβ11−40. As

larger free energy holes are associated with slower folding kinetics, as previously shown for small proteins,61 the folding kinetics of A21G 3Aβ11−40 can be expected to be slower than that of wild-type 3Aβ11−40. This result is consistent with the slower aggregation rate of A21G amyloid peptides.22 Free Energy Difference of Binding between DPPC Lipid Bilayer and A21G 3Aβ11−40. Although various schemes have been developed to assess the free energy difference of binding,62−67 the binding free energy between settle monomer to other of A21G 3Aβ11−40 was calculated using the free energy perturbation (FEP) method that was previously used for 3Aβ11−40.27 In this method, the optimized structures of 3Aβ11−30 (M1 in ref 27) and A21G Aβ11−40 (A1, Figure 6) were subjected to FEP. Free binding energy (ΔGbind) was calculated from the difference in the annihilation energy of the transmembrane peptide and that of the corresponding soluble peptide (Figure 8). ΔGbind is expressed in two terms, electrostatic interaction energy ΔGcou and van der Waals interaction energy ΔGvdW (eq 1). ΔG bind = ΔGcou + ΔGvdW

(1)

For A21G 3Aβ11−40, the calculated values for ΔGcou and ΔGvdW are 159.87 ± 19.55 and −189.32 ± 4.47 kcal/mol, respectively, resulting in a ΔGbind value of −29.44 ± 15.61 kcal/ mol. The ΔGbind of A21G 3Aβ11−40 is significantly smaller than that determined for 3Aβ11−40 (−70 ± 18 kcal/mol) using the same method. The ΔGcou and ΔGvdW values for 3Aβ11−40 are 8471

DOI: 10.1021/acs.jpcb.7b05906 J. Phys. Chem. B 2017, 121, 8467−8474

Article

The Journal of Physical Chemistry B

Figure 8. Annihilation free energy of transmembrane (black) and soluble (red) A21G 3Aβ11−40. The energy was calculated using the BAR method.

Figure 7. Structures of five minima shown in Figure 6. The populated structures of transmembrane mutant A21G trimer were evaluated using a combination of FES and clustering methods. In total, 52% of all equilibrated snapshots are located.

Figure 9. Surface charges of A21G 3Aβ11−40 (A1, Figure 6) and WT 3Aβ11−40 (M1, ref 27) transmembrane, which were estimated using APBS calculation.69 The conformations were the representative structures of two systems, which were employed as initial structures to estimate the binding free energy of the trimer to the membrane. The membrane and solvation were hidden in this figure.

Table 1. Secondary Structure Terms and Collision Cross Section of Five Optimized Structure That Were Obtained Using DSSP and IMPACT Protocols minima A1 A2 A3 A4 A5 average average of wild-type

coil content (%)

beta content (%)

turn content (%)

helix content (%)

CCS (Å2)

SA (nm2)

51 57 61 52 60 56 54

46 39 39 46 38 42 44

3 4 0 2 2 2 2

0 0 0 0 0 0 0

1322 1382 1485 1409 1402 1400 1340

65.23 68.11 72.77 70.30 69.95 69.27 64.18

experimental results that showed A21G Aβ peptides bind to the membrane more slowly than the wild-type peptides.68

■

CONCLUSIONS We found that the A21G mutation caused remarkable changes to transmembrane A21G 3Aβ11−40. Although the beta, helix, coil, and turn contents did not change to sizable extents, we found that Rg, RMSD, and SA all increased significantly upon A21G mutation. In addition, the essential polar contacts of 3Aβ11−40 were also almost diminished in A21G 3Aβ11−40. There were changes in the sequences and residues of the trimer interacting with the DPPC lipid membrane, and the total amount of contacts to the charged phosphate groups increased. Analysis of representative structures revealed great changes in the beta cores of the trimer. All 3Aβ11−40 minimum structures contain two antiparallel sheets of parallel beta strands that are facing each other. In the A3 minimum of A21G 3Aβ11−40, the two beta sheets are almost perpendicular to each other, while in the A5 minimum these two sheets are placed next to each other in a plane. The minima of A21G 3Aβ11−40 are well separated, which forms significantly wider free energy holes on the free energy surface compared to those of wild-type 3Aβ11−40. The binding free energy calculated for the global minima is significantly reduced by ∼40 kcal/mol, mostly due to the

114 ± 18 and −183 ± 3 kcal/mol. The significant difference in ΔGbind of A21G 3Aβ11−40 and 3Aβ11−40 largely arises from the difference in the electrostatic interactions between these two trimers with the DPPC lipid bilayer. Interestingly, although replacing the nonpolar residue A21 by a nonpolar residue G does not alter the total charge of the trimer, the surface charges of the trimers increase significantly (Figure 9). The negative charges at both coil regions of the trimer increase, leading to stronger repulsion with the negative charges of the phosphate groups. This is consistent with the increase in the contacts with the phosphate group found in the 20−23 sequence discussed above (Figure 5). As a result, the A21G 3Aβ11−40 would be repelled more strongly from the inside of the membrane compared to 3Aβ11−40. This result is in good agreement with 8472

DOI: 10.1021/acs.jpcb.7b05906 J. Phys. Chem. B 2017, 121, 8467−8474

Article

The Journal of Physical Chemistry B

(6) Williams, T. L.; Serpell, L. C. Membrane and Surface Interactions of Alzheimer’s Aβ Peptide − Insights into the Mechanism of Cytotoxicity. FEBS J. 2011, 278, 3905−3917. (7) Quist, A.; et al. Amyloid Ion Channels: A Common Structural Link for Protein-Misfolding Disease. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 10427−10432. (8) Jang, H.; Zheng, J.; Nussinov, R. Models of B-Amyloid Ion Channels in the Membrane Suggest That Channel Formation in the Bilayer Is a Dynamic Process. Biophys. J. 2007, 93, 1938−1949. (9) Connelly, L.; et al. Atomic Force Microscopy and Md Simulations Reveal Pore-Like Structures of All-D-Enantiomer of Alzheimer’s B-Amyloid Peptide: Relevance to the Ion Channel Mechanism of Ad Pathology. J. Phys. Chem. B 2012, 116, 1728−1735. (10) Jana, M. K.; Cappai, R.; Pham, C. L. L.; Ciccotosto, G. D. Membrane-Bound Tetramer and Trimer Aβ Oligomeric Species Correlate with Toxicity Towards Cultured Neurons. J. Neurochem. 2016, 136, 594−608. (11) Citron, M. Strategies for Disease Modification in Alzheimer’s Disease. Nat. Rev. Neurosci. 2004, 5, 677−685. (12) Bitan, G.; et al. Amyloid B-Protein (Aβ) Assembly: Aβ40 and Aβ42 Oligomerize through Distinct Pathways. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 330−335. (13) Kirkitadze, M. D.; Condron, M. M.; Teplow, D. B. Identification and Characterization of Key Kinetic Intermediates in Amyloid BProtein Fibrillogenesis. J. Mol. Biol. 2001, 312, 1103−1119. (14) Ngo, S. T.; Li, M. S. Curcumin Binds to Abeta1−40 Peptides and Fibrils Stronger Than Ibuprofen and Naproxen. J. Phys. Chem. B 2012, 116, 10165−75. (15) Ngo, S. T.; Li, M. S. Top-Leads from Natural Products for Treatment of Alzheimer’s Disease: Docking and Molecular Dynamics Study. Mol. Simul. 2013, 39, 279−291. (16) Ngo, S. T.; et al. Anti-Arrhythmic Medication Propafenone a Potential Drug for Alzheimer’s Disease Inhibiting Aggregation of Aβ: In Silico and in Vitro Studies. J. Chem. Inf. Model. 2016, 56, 1344− 1356. (17) Hendriks, L.; et al. Presenile Dementia and Cerebral Haemorrhage Linked to a Mutation at Codon 692 of the B-Amyloid Precursor Protein Gene. Nat. Genet. 1992, 1, 218−221. (18) Levy, E.; et al. Mutation of the Alzheimer’s Disease Amyloid Gene in Hereditary Cerebral Hemorrhage, Dutch Type. Science 1990, 248, 1124−1126. (19) Bugiani, O.; et al. An Italian Type of Hchwa. Neurobiol. Aging 1998, 19, S238. (20) Nilsberth, C.; et al. The ’Arctic’ App Mutation (E693g) Causes Alzheimer’s Disease by Enhanced Aβ Protofibril Formation. Nat. Neurosci. 2001, 4, 887−893. (21) Grabowski, T. J.; et al. Novel Amyloid Precursor Protein Mutation in an Iowa Family with Dementia and Severe Cerebral Amyloid Angiopathy. Ann. Neurol. 2001, 49, 697−705. (22) Yagi-Utsumi, M.; Dobson, C. M. Conformational Effects of the A21g Flemish Mutation on the Aggregation of Amyloid B Peptide. Biol. Pharm. Bull. 2015, 38, 1668−1672. (23) Tang, T.-C.; et al. Conformational Changes Induced by the A21g Flemish Mutation in the Amyloid Precursor Protein Lead to Increased Aβ Production. Structure 2014, 22, 387−396. (24) Chong, S.-H.; Yim, J.; Ham, S. Structural Heterogeneity in Familial Alzheimer’s Disease Mutants of Amyloid-Beta Peptides. Mol. BioSyst. 2013, 9, 997−1003. (25) Xu, L.; Shan, S.; Wang, X. Single Point Mutation Alters the Microstate Dynamics of Amyloid B-Protein Aβ42 as Revealed by Dihedral Dynamics Analyses. J. Phys. Chem. B 2013, 117, 6206−6216. (26) Huet, A.; Derreumaux, P. Impact of the Mutation A21g (Flemish Variant) on Alzheimer’s B-Amyloid Dimers by Molecular Dynamics Simulations. Biophys. J. 2006, 91, 3829−3840. (27) Ngo, S. T.; Hung, H. M.; Tran, K. N.; Nguyen, M. T. Replica Exchange Molecular Dynamics Study of the Amyloid Beta (11−40) Trimer Penetrating a Membrane. RSC Adv. 2017, 7, 7346−7357. (28) Ngo, S. T.; Hung, H. M.; Truong, D. T.; Nguyen, M. T. Replica Exchange Molecular Dynamics Study of the Truncated Amyloid Beta

change in the electrostatic interactions. Consistently, the surface charges were found to increase upon A21G mutation, especially the negative charges near the coil regions, resulting in the greater repelling force on the trimer from the inside to the outside of the membrane. Altogether, the computational data obtained in these studies indicate that A21G mutations cause 3Aβ11−40 to be much more dynamic with larger overall structure, greater surface charges, and lower affinity to the DPPC lipid bilayer. These results are consistent with experimental studies that showed that A21G amyloid peptides fold more slowly and bind more weakly to the membrane compared to the wild-type peptides.17,68

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcb.7b05906. List of temperatures of simulations and additional figures showing the exchange rate between neighboring replicas, dispersion representative replicas in the entire temperature space, stability of the lipid bilayer, nonbonded and hydrogen bonded contact maps, secondary structure contents per residue of A21G Aβ11−40 trimer, and comparison of free energy holes containing conformations having a free energy greater than 13.3 kJ/mol in transmembrane A21G 3Aβ11−40 with that in transmembrane 3Aβ11−40 (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Son Tung Ngo: 0000-0003-1034-1768 Van V. Vu: 0000-0003-0009-6703 Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research is supported by Nguyen Tat Thanh University institutional grant #2016.03.01 to V.V.V. and by Vietnam National Foundation for Science & Technology Development (NAFOSTED) grant #103.01-2016.48 to S.T.N.

■

REFERENCES

(1) Hardy, J.; Selkoe, D. J. The Amyloid Hypothesis of Alzheimer’s Disease: Progress and Problems on the Road to Therapeutics. Science 2002, 297, 353−356. (2) Selkoe, D. J.; Hardy, J. The Amyloid Hypothesis of Alzheimer’s Disease at 25 years. EMBO Mol. Med. 2016, 8, 595−608. (3) Lee, S. J. C.; et al. Towards an Understanding of Amyloid-[Small Beta] Oligomers: Characterization, Toxicity Mechanisms, and Inhibitors. Chem. Soc. Rev. 2017, 46, 310−323. (4) Sakono, M.; Zako, T. Amyloid Oligomers: Formation and Toxicity of Aβ Oligomers. FEBS J. 2010, 277, 1348−1358. (5) Walsh, D. M.; Selkoe, D. J. Aβ Oligomers − a Decade of Discovery. J. Neurochem. 2007, 101, 1172−1184. 8473

DOI: 10.1021/acs.jpcb.7b05906 J. Phys. Chem. B 2017, 121, 8467−8474

Article

The Journal of Physical Chemistry B (11−40) Trimer in Solution. Phys. Chem. Chem. Phys. 2017, 19, 1909− 1919. (29) Nagle, J. F. Area/Lipid of Bilayers from Nmr. Biophys. J. 1993, 64, 1476−1481. (30) The PyMOL Molecular Graphics System, Version 1.5.0.4; Schrödinger, LLC. (31) Oostenbrink, C.; Villa, A.; Mark, A. E.; Van Gunsteren, W. F.; Biomolecular Force, A. Field Based on the Free Enthalpy of Hydration and Solvation: The Gromos Force-Field Parameter Sets 53a5 and 53a6. J. Comput. Chem. 2004, 25, 1656−1676. (32) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Hermans, a. J. Intermolecular Forces. Reidel, Dordrecht: Jerusalem, Israel, 1981. (33) Abraham, M. J.; et al. Gromacs: High Performance Molecular Simulations through Multi-Level Parallelism from Laptops to Supercomputers. SoftwareX 2015, 1−2, 19−25. (34) Zwanzig, R. W. High-Temperature Equation of State by a Perturbation Method. I. Nonpolar Gases. J. Chem. Phys. 1954, 22, 1420−1426. (35) Touw, W. G.; et al. A Series of Pdb-Related Databanks for Everyday Needs. Nucleic Acids Res. 2015, 43, D364−D368. (36) Joosten, R. P.; et al. A Series of Pdb Related Databases for Everyday Needs. Nucleic Acids Res. 2011, 39, D411−D419. (37) Mu, Y.; Nguyen, P. H.; Stock, G. Energy Landscape of a Small Peptide Revealed by Dihedral Angle Principal Component Analysis. Proteins: Struct., Funct., Genet. 2005, 58, 45−52. (38) Papaleo, E.; et al. Free-Energy Landscape, Principal Component Analysis, and Structural Clustering to Identify Representative Conformations from Molecular Dynamics Simulations: The Myoglobin Case. J. Mol. Graphics Modell. 2009, 27, 889−899. (39) Marklund, E. G.; et al. Collision Cross Sections for Structural Proteomics. Structure 2015, 23, 791−799. (40) Daura, X.; et al. Peptide Folding: When Simulation Meets Experiment. Angew. Chem., Int. Ed. 1999, 38, 236−240. (41) Morriss-Andrews, A.; Shea, J. E. Computational Studies of Protein Aggregation: Methods and Applications. Annu. Rev. Phys. Chem. 2015, 66, 643−666. (42) Sugita, Y.; Okamoto, Y. Replica-Exchange Molecular Dynamics Method for Protein Folding. Chem. Phys. Lett. 1999, 314, 141−151. (43) Strodel, B.; Lee, J. W. L.; Whittleston, C. S.; Wales, D. J. Transmembrane Structures for Alzheimer’s Aβ1−42 Oligomers. J. Am. Chem. Soc. 2010, 132, 13300−13312. (44) Davis, C. H.; Berkowitz, M. L. Structure of the Amyloid-B (1− 42) Monomer Absorbed to Model Phospholipid Bilayers: A Molecular Dynamics Study. J. Phys. Chem. B 2009, 113, 14480−14486. (45) Patriksson, A.; van der Spoel, D.; Temperature, A. Predictor for Parallel Tempering Simulations. Phys. Chem. Chem. Phys. 2008, 10, 2073−2077. (46) Fezoui, Y.; Teplow, D. B. Kinetic Studies of Amyloid B-Protein Fibril Assembly: Differential Effects of A-Helix Stabilization. J. Biol. Chem. 2002, 277, 36948−36954. (47) Jang, S.; Shin, S. Amyloid B-Peptide Oligomerization in Silico: Dimer and Trimer. J. Phys. Chem. B 2006, 110, 1955−1958. (48) Jang, S.; Shin, S. Computational Study on the Structural Diversity of Amyloid Beta Peptide (Aβ10−35) Oligomers. J. Phys. Chem. B 2008, 112, 3479−3484. (49) Klimov, D. K.; Thirumalai, D. Dissecting the Assembly of a Beta(16−22) Amyloid Peptides into Antiparallel Beta Sheets. Structure 2003, 11, 295−307. (50) Terzi, E.; Hölzemann, G.; Seelig, J. Self-Association of BAmyloid Peptide (1−40) in Solution and Binding to Lipid Membranes. J. Mol. Biol. 1995, 252, 633−642. (51) Terzi, E.; Hoelzemann, G.; Seelig, J. Alzheimer B-Amyloid Peptide 25−35: Electrostatic Interactions with Phospholipid Membranes. Biochemistry 1994, 33, 7434−7441. (52) Ono, K.; Condron, M. M.; Teplow, D. B. Structure− Neurotoxicity Relationships of Amyloid B-Protein Oligomers. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 14745−14750.

(53) Miyashita, N.; Straub, J. E.; Thirumalai, D.; Sugita, Y. Transmembrane Structures of Amyloid Precursor Protein Dimer Predicted by Replica-Exchange Molecular Dynamics Simulations. J. Am. Chem. Soc. 2009, 131, 3438−3439. (54) Han, W.; Schulten, K. Fibril Elongation by Aβ17−42: Kinetic Network Analysis of Hybrid-Resolution Molecular Dynamics Simulations. J. Am. Chem. Soc. 2014, 136, 12450−12460. (55) Petkova, A. T.; et al. A Structural Model for Alzheimer’s BetaAmyloid Fibrils Based on Experimental Constraints from Solid State Nmr. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 16742−16747. (56) Luhrs, T.; et al. 3d Structure of Alzheimer’s Amyloid-Beta(1− 42) Fibrils. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 17342−17347. (57) Petkova, A. T.; Yau, W. M.; Tycko, R. Experimental Constraints on Quaternary Structure in Alzheimer’s B-Amyloid Fibrils. Biochemistry 2006, 45, 498−512. (58) Lemkul, J. A.; Bevan, D. R. A Comparative Molecular Dynamics Analysis of the Amyloid B-Peptide in a Lipid Bilayer. Arch. Biochem. Biophys. 2008, 470, 54−63. (59) Lemkul, J. A.; Bevan, D. R. Lipid Composition Influences the Release of Alzheimer’s Amyloid B-Peptide from Membranes. Protein Sci. 2011, 20, 1530−1545. (60) McLaurin, J.; Chakrabartty, A. Characterization of the Interactions of Alzheimer B-Amyloid Peptides with Phospholipid Membranes. Eur. J. Biochem. 1997, 245, 355−363. (61) Yang, W. Y.; Gruebele, M. Folding at the Speed Limit. Nature 2003, 423, 193−197. (62) Morris, G. M.; et al. Autodock4 and Autodocktools4: Automated Docking with Selective Receptor Flexibility. J. Comput. Chem. 2009, 30, 2785−2791. (63) Holliday, J. D.; Ranade, S. S.; Willett, P. A Fast Algorithm for Selecting Sets of Dissimilar Molecules from Large Chemical Databases. Quant. Struct.-Act. Relat. 1995, 14, 501−506. (64) Srinivasan, J.; et al. Continuum Solvent Studies of the Stability of DNA, Rna, and Phosphoramidate−DNA Helices. J. Am. Chem. Soc. 1998, 120, 9401−9409. (65) Ngo, S. T.; Hung, H. M.; Nguyen, M. T. Fast and Accurate Determination of the Relative Binding Affinities of Small Compounds to Hiv-1 Protease Using Non-Equilibrium Work. J. Comput. Chem. 2016, 37, 2734−2742. (66) Truong, D. T.; Nguyen, M. T.; Vu, V. V.; Ngo, S. T. Fast Pulling of Ligand Approach for the Design of B-Secretase 1 Inhibitors. Chem. Phys. Lett. 2017, 671, 142−146. (67) Ngo, S. T.; Nguyen, M. T.; Nguyen, M. T. Determination of the Absolute Binding Free Energies of Hiv-1 Protease Inhibitors Using Non-Equilibrium Molecular Dynamics Simulations. Chem. Phys. Lett. 2017, 676, 12−17. (68) Sureshbabu, N.; et al. Lipid-Induced Conformational Transition of Amyloid B Peptide Fragments. J. Mol. Neurosci. 2010, 41, 368−382. (69) Baker, N. A.; et al. Electrostatics of Nanosystems: Application to Microtubules and the Ribosome. Proc. Natl. Acad. Sci. U. S. A. 2001, 98, 10037−10041.

8474

DOI: 10.1021/acs.jpcb.7b05906 J. Phys. Chem. B 2017, 121, 8467−8474