Research Article
Proteins: Structure, Function and Bioinformatics DOI 10.1002/prot.24669
Title: Dual effects of familial Alzheimer's disease mutations (D7H, D7N, and H6R) on amyloid β peptide: Correlation dynamics and zinc binding Short Title Effects of C-terminal mutations of amyloid β peptide Keywords Alzheimer’s disease; amyloid peptide; molecular dynamics simulations; free energy; mutation; Author Information: Liang Xu,*,1 Yonggang Chen,2 Xiaojuan Wang3 1
School of Chemistry, Dalian University of Technology, Dalian, China,
2
Network and Information Center, Dalian University of Technology, Dalian, China
3
School of Chemical Machinery, Dalian University of Technology, Dalian, China
Corresponding Author: [email protected]
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an ‘Accepted Article’, doi: 10.1002/prot.24669 © 2014 Wiley Periodicals, Inc. Received: May 19, 2014; Revised: Jul 17, 2014; Accepted: Aug 11, 2014
ABSTRACT Although the N-terminal region of Amyloid β (Aβ) peptides plays dual roles as metal-coordinating sites and conformational modulator, few studies have been performed to explore the effects of mutations at this region on the overall conformational ensemble of Aβ and the binding propensity of metal ions. In this work, we focus on how three familial Alzheimer disease mutations (D7H, D7N, and H6R) alter the structural characteristics and thermodynamic stabilities of Aβ42 using molecular dynamics simulations. We observe that each mutation displays increased β-sheet structures in both N and C termini. In particular, both the N terminus and central hydrophobic region of D7H can form stable β-hairpin structures with its C terminus. The conserved turn structure at Val24–Lys28 in all peptides and Zn2+-bound Aβ42 is confirmed as the common structural motif to nucleate folding of Aβ. Each mutant can significantly increase the solvation free energy and thus enhance the aggregation of Aβ monomers. The correlation dynamics between Aβ(1–16) and Aβ(17–42) fragments are elucidated by linking the domain motions with the corresponding structured conformations. We characterize the different populations of correlated domain motions for each mutant from a more macroscopic perspective, and unexpectedly find that Zn2+-bound Aβ42 ensemble shares the same populations as Aβ42, indicating the binding of Zn2+ to Aβ follows the conformational selection mechanism, and thus is independent of domain motions, even though the structures of Aβ have been modified at a residue level.
2
INTRODUCTION Amyloid β (Aβ) peptide, a typical kind of intrinsically disordered protein, has been associated with the pathogenesis of Alzheimer's disease (AD).1, 2 Among various length of Aβ, two species with 40 and 42 amino acids (i.e., Aβ40 and Aβ42) are most commonly generated by proteolytic cleavage of amyloid precursor protein.3, 4 One of fascinating characteristics of these full-length Aβ peptides lies in the observation that Aβ monomers are primarily statistical random coil in physiological conditions, whereas they form two β-sheets connected by a turn motif in C-terminal region according to Aβ fibril structures developed from experiments.3, 5 For example, in the Lührs Aβ(17–42) model (PDB ID: 2BEG),6 two β-sheet motifs involving residues 18–26 and 31–42 are linked by a U-turn formed by residues 27–30. Note that the N-terminal region Aβ(1–16) was not determined due to its disordered property. In another Aβ fibril structure Aβ(9–40) developed by Tycko et al. in terms of solid state NMR experiments,7 the first eight residues are missing. Extensive experimental and simulation studies primarily focus on the investigation of the β propensity of the C-terminal region of Aβ because the conformational transition in this domain might be directly associated with the aggregation tendency of Aβ peptides.3, 8-22 However, few studies investigated the role of the N-terminal region played in the β-sheet formation.13, 23-28 The overall content of β-sheet structure in monomeric Aβ is largely dependent on the measurement conditions, with a probability varying from 10 to 25%.4,
29
Moreover,
molecular dynamics (MD) simulations suggest that Aβ monomers can transiently sample 3
oligomer and fibril-like conformations.11 It is thus reasonably to assume that such less populated β-sheet species may become predominate structure during the oligomerization pathway by population shift.30, 31 Interestingly, both experimental and MD simulations using efficient sampling techniques suggest that the N-terminal is not entirely disordered, but can sample turn and β-sheet structures with relatively high probabilities.9, 26, 32 It is thus important to understand the correlation dynamics between the N and C terminus of Aβ. The finding that metal ions like Cu2+, Zn2+ and Fe3+ can induce Aβ aggregation further complicates the relationship between the N- and C- terminals.33-35 It is generally believed that the N-terminal Aβ(1–16) is the primary metal binding sites.36-38 In the NMR structure of Zn2+-bound Aβ(1-16) (PDB ID: 1ZE9),39 His6, Glu11, His13, and His14 are coordinated with Zn2+. Results from NMR experiments indicate that the binding of Zn2+ to Aβ40/42 leads to form nonfibrillar aggregates.40 Residues 24–28 were induced to form a rigid turn-like conformation, and consequently the intervening region and the C terminus of Aβ40/42 become more flexible.40 Previous experiments and MD simulations also suggested a significant decrease of β-sheet in Zn2+-bound Aβ40/42.41-44 Thus, a negative relation between the two terminals of Aβ can be derived: when the disordered N-terminal becomes less flexible upon metal binding, the C-terminal becomes more unstructured. As a result, the propensity of β-sheet formation in the C-terminal domain markedly decreases. Although large-scale MD simulations have been performed to investigate the effects of metal binding on the thermodynamics and kinetics of Aβ,41, 42, 45, 46 the modulation effect of N-terminal due to metal binding on the C-terminal of Aβ have not been examined so far. 4
In this study, the wide-type Aβ42, as well as three familial mutations located at Aβ N terminus, including Taiwanese (D7H),27 Tottori (D7N),47 and English (H6R),48 were used to study the correlation dynamics between Aβ42 N and C terminus by replica-exchange molecular dynamics (REMD) simulations.49,
50
Experimental studies have shown that
these mutations promote Aβ oligomerization and enhance fibril formation without increasing formation of intermediate protofibril.26,
51
Obviously, alterations in the N
terminus affect the overall structural and dynamic property of Aβ peptide. On the other hand, these mutations may have different effects on the binding of metal ions because D7H has an additional histidine, H6R lacks one metal-binding histidine, and D7N may alter the susceptibility of its neighboring histidine to metal ions. Relative to wide-type Aβ42, we observed that the conformational spaces sampled by these mutations shrink to different degrees. These mutations increase the β-sheet propensity of both termini, especially the C-terminal region, whereas the overall tertiary structures of these mutations do not change significantly. Using both the MM/GBSA52 and MM/3D-RISM methods,53 we found that Aβ42 monomer becomes less stable upon mutation, which can be attributed to the increased solvation free energy. As a consequence, these mutations promote Aβ to aggregate into low molecular weight oligomers that are more neurotoxic. We further observed that both N- and C- terminals of Aβ42 predominantly populate low to mild structured conformations simultaneously, and mutations at its N terminus alter the relative populations differently. Conformations with highly ordered N terminus but rather mobile C terminus, and extremely disordered N terminus but rather rigid C terminus, are less 5
populated. Of importance, the same conformational populations were also found for Zn2+-bound Aβ42, suggesting that the binding of metal ions is independent of the correlation dynamics of both Aβ terminal regions. MATERIALS AND METHODS REMD simulations The same simulation protocol as used in our previous studies of amyloid peptides was applied in this work.13 Briefly, the NMR structure of full-length Aβ42 (PDB ID: 1Z0Q)54 was used as the starting conformation for REMD simulations. Based on the above Aβ42 structure, the initial structures of D7H, D7N, and H6R were prepared using the mutation tools in PyMOL program.55 All structures were modeled using Amber ff99SB force field56 and the solvent effect was represented by Onufriev-Bashford-Case GB implicit solvation model57 in order to enhance conformational sampling. In each REMD simulations, eight replicas with temperatures ranging from 280 K to 400 K were applied, leading to an acceptance ratio of about 20%. The integration time step was 2 fs and exchange between replicas was attempted every 5 ps. Previous REMD simulations including our own suggest that the fluctuation of cumulative helix content throughout simulations is an excellent indicator to check the convergence of REMD simulations.10, 24, 45, 58-60 As the cumulative helix percentage reaches to a plateau, we found that the system becomes equilibrium. Structural properties such as the population of radius of gyration and secondary structures at different time intervals are very similar (see Table I). To compare the conformational ensemble obtained from our REMD simulations with 6
available experimental data. The chemical shifts were calculated using SHIFTX61 (Fig. S1) based on the trajectory collected at 280 K (closest to the experimental temperature of 278 K).62 The Pearson correlation coefficients are 0.98, 0.84, and 0.90 for Cα, Hα, and N, respectively, in good agreement with experimental values. In addition, the J-coupling constants were also calculated using the Karplus equation,63, 64 and compared with the experimental data.65 Fig. S2 shows that a qualitative agreement between experimental and simulated J-coupling constants was obtained. It is still quite challenging to calculate the J-coupling constants of Aβ from simulations that are quantitatively in line with experiment.22 The total simulation time varies for each system, as it was found that 60 ns might be sufficient to reach convergence for Aβ42,13, 32 whereas much longer simulation time is necessary for its mutations. The total simulation times for D7H, D7N, and H6R are 7.52 µs (940 ns/replica), 4.21 µs (526 ns/replica), and 4.21 µs (526 ns/replica), respectively (Fig. S3). The last 100-ns MD simulations were used for data analysis. For comparison, we extended our previous REMD simulations of Aβ42 from 100 ns/replica to 200 ns/replica, and used the last 100-ns trajectory for all data analysis. All REMD simulations were performed using AMBER12 software package.66 Each trajectory containing 100,000 conformations collected at physiological temperature 310 K was used for data analysis. Free energy calculations The classical MM/GBSA method52 was applied to estimate the relative stability of Aβ and its mutations by calculation of their free energies (GMM/GBSA) in terms of the following 7
formula: G MM/GBSA = H MM/GBSA − TS NM ,
where T is the temperature, and the enthalpy HMM/GBSA is calculated in terms of MM/GBSA H MM/GBSA = EMM + Gsolvation
where EMM is the internal energy (a sum of electrostatic and van der Waals interaction MM/GBSA includes the electrostatic contribution and energies). The solvation free energy Gsolvation
the non-polar part that is proportional to the solvent accessible surface area of Aβ peptides. The entropy SNM was calculated by applying normal mode (NM) analysis on trajectories obtained from REMD simulations. However, previous studies have shown that MM/GBSA method may not be suitable for the evaluation of free energies of disordered proteins.24, 45 In particular, the NM method cannot discriminate the entropy difference between Aβ and its various mutations. Therefore, an alternative method, the three-dimensional reference interaction site model (3D-RISM) was applied to calculate the free energies of all peptides. This approach is based on the integral-equation theory of liquids and is able to account for both polar and non-polar features of the solvation structure.67, 68 However, the absolute value of solvation free energy varies with the closure relation used and only relative values are reasonably accurate.69-71 The robust Kovalenko-Hirata closure and the Gaussian fluctuation approximation as suggested by Chandler et al. was used to calculate the solvation free energy.53 Recently, a new method of computing the conformational entropy of Aβ peptides has been derived from the 3D-RISM theory.69, 70 8
TS 3D-RISM =
β 2
< ( H 3D-RISM − < H 3D-RISM >) 2 >
where β=1/(kBT) and kB is the Boltzmann constant, and denotes the ensemble average. In this method, the enthalpy (H3D-RISM) and entropy (TS3D-RISM) contributions to the free energy G3D-RISM are therefore different from those in the MM/GBSA method. The enthalpy H3D-RISM was obtained by combining the internal energy (EMM, the same as in the MM/GBSA method) and the solvation free energy calculated using the 3D-RISM method (Table 2). RESULTS AND DISCUSSION D7H, D7N, and H6R mutations do not change the intrinsic dynamics of Aβ42 The conformational dynamics of Aβ are crucial for its relevant functions. It is intriguing if mutations at its N terminus alter the overall intrinsic disorder and consequently enhance their aggregation tendency. Previously, we used the trajectory from REMD simulations to characterize the internal dynamics of Zn2+-bound Aβ40 and Aβ42.72 We calculated the NMR 15N-1H nuclear Overhauser enhancement (NOE) by rescaling the time scale of the trajectory at the highest temperature. The simulated NOE values were nearly in quantitative agreement with available experimental data. Here, we applied the same strategy to investigate the effect of point mutations on the overall tumbling motions of Aβ42 (Fig. S4). It is found that these mutations have little effect on the global tumbling dynamics of Aβ42, suggesting that these mutations are still intrinsically flexible. However, due to the difficulty of extracting reliable motional correlation times from MD simulations
9
and the limitations in interpreting experimental NMR relaxation data by use of a simple model, the subtle difference of local dynamics may be obscure. An alternative simple but insightful parameter, the relative solvent-accessible surface area (RSASA),73, 74 was used here to measure the relative flexibility of Aβ and its mutations. The RSASA is defined as the observed SASA calculated directly from conformational ensemble, divided by the correspondingly expected/idealized value of SASA. Figure 1 shows the RSASA for each residue of Aβ and its mutated species. The RSASA for all residues is not equal to one, clearly demonstrating that all Aβ peptides are intrinsically flexible. As expected, there are great fluctuations of conformations for each residue. Except residues Asp1 and Ala42, the remaining part with RSASA
Proteins: Structure, Function and Bioinformatics DOI 10.1002/prot.24669
Title: Dual effects of familial Alzheimer's disease mutations (D7H, D7N, and H6R) on amyloid β peptide: Correlation dynamics and zinc binding Short Title Effects of C-terminal mutations of amyloid β peptide Keywords Alzheimer’s disease; amyloid peptide; molecular dynamics simulations; free energy; mutation; Author Information: Liang Xu,*,1 Yonggang Chen,2 Xiaojuan Wang3 1
School of Chemistry, Dalian University of Technology, Dalian, China,
2
Network and Information Center, Dalian University of Technology, Dalian, China
3
School of Chemical Machinery, Dalian University of Technology, Dalian, China
Corresponding Author: [email protected]
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an ‘Accepted Article’, doi: 10.1002/prot.24669 © 2014 Wiley Periodicals, Inc. Received: May 19, 2014; Revised: Jul 17, 2014; Accepted: Aug 11, 2014
ABSTRACT Although the N-terminal region of Amyloid β (Aβ) peptides plays dual roles as metal-coordinating sites and conformational modulator, few studies have been performed to explore the effects of mutations at this region on the overall conformational ensemble of Aβ and the binding propensity of metal ions. In this work, we focus on how three familial Alzheimer disease mutations (D7H, D7N, and H6R) alter the structural characteristics and thermodynamic stabilities of Aβ42 using molecular dynamics simulations. We observe that each mutation displays increased β-sheet structures in both N and C termini. In particular, both the N terminus and central hydrophobic region of D7H can form stable β-hairpin structures with its C terminus. The conserved turn structure at Val24–Lys28 in all peptides and Zn2+-bound Aβ42 is confirmed as the common structural motif to nucleate folding of Aβ. Each mutant can significantly increase the solvation free energy and thus enhance the aggregation of Aβ monomers. The correlation dynamics between Aβ(1–16) and Aβ(17–42) fragments are elucidated by linking the domain motions with the corresponding structured conformations. We characterize the different populations of correlated domain motions for each mutant from a more macroscopic perspective, and unexpectedly find that Zn2+-bound Aβ42 ensemble shares the same populations as Aβ42, indicating the binding of Zn2+ to Aβ follows the conformational selection mechanism, and thus is independent of domain motions, even though the structures of Aβ have been modified at a residue level.
2
INTRODUCTION Amyloid β (Aβ) peptide, a typical kind of intrinsically disordered protein, has been associated with the pathogenesis of Alzheimer's disease (AD).1, 2 Among various length of Aβ, two species with 40 and 42 amino acids (i.e., Aβ40 and Aβ42) are most commonly generated by proteolytic cleavage of amyloid precursor protein.3, 4 One of fascinating characteristics of these full-length Aβ peptides lies in the observation that Aβ monomers are primarily statistical random coil in physiological conditions, whereas they form two β-sheets connected by a turn motif in C-terminal region according to Aβ fibril structures developed from experiments.3, 5 For example, in the Lührs Aβ(17–42) model (PDB ID: 2BEG),6 two β-sheet motifs involving residues 18–26 and 31–42 are linked by a U-turn formed by residues 27–30. Note that the N-terminal region Aβ(1–16) was not determined due to its disordered property. In another Aβ fibril structure Aβ(9–40) developed by Tycko et al. in terms of solid state NMR experiments,7 the first eight residues are missing. Extensive experimental and simulation studies primarily focus on the investigation of the β propensity of the C-terminal region of Aβ because the conformational transition in this domain might be directly associated with the aggregation tendency of Aβ peptides.3, 8-22 However, few studies investigated the role of the N-terminal region played in the β-sheet formation.13, 23-28 The overall content of β-sheet structure in monomeric Aβ is largely dependent on the measurement conditions, with a probability varying from 10 to 25%.4,
29
Moreover,
molecular dynamics (MD) simulations suggest that Aβ monomers can transiently sample 3
oligomer and fibril-like conformations.11 It is thus reasonably to assume that such less populated β-sheet species may become predominate structure during the oligomerization pathway by population shift.30, 31 Interestingly, both experimental and MD simulations using efficient sampling techniques suggest that the N-terminal is not entirely disordered, but can sample turn and β-sheet structures with relatively high probabilities.9, 26, 32 It is thus important to understand the correlation dynamics between the N and C terminus of Aβ. The finding that metal ions like Cu2+, Zn2+ and Fe3+ can induce Aβ aggregation further complicates the relationship between the N- and C- terminals.33-35 It is generally believed that the N-terminal Aβ(1–16) is the primary metal binding sites.36-38 In the NMR structure of Zn2+-bound Aβ(1-16) (PDB ID: 1ZE9),39 His6, Glu11, His13, and His14 are coordinated with Zn2+. Results from NMR experiments indicate that the binding of Zn2+ to Aβ40/42 leads to form nonfibrillar aggregates.40 Residues 24–28 were induced to form a rigid turn-like conformation, and consequently the intervening region and the C terminus of Aβ40/42 become more flexible.40 Previous experiments and MD simulations also suggested a significant decrease of β-sheet in Zn2+-bound Aβ40/42.41-44 Thus, a negative relation between the two terminals of Aβ can be derived: when the disordered N-terminal becomes less flexible upon metal binding, the C-terminal becomes more unstructured. As a result, the propensity of β-sheet formation in the C-terminal domain markedly decreases. Although large-scale MD simulations have been performed to investigate the effects of metal binding on the thermodynamics and kinetics of Aβ,41, 42, 45, 46 the modulation effect of N-terminal due to metal binding on the C-terminal of Aβ have not been examined so far. 4
In this study, the wide-type Aβ42, as well as three familial mutations located at Aβ N terminus, including Taiwanese (D7H),27 Tottori (D7N),47 and English (H6R),48 were used to study the correlation dynamics between Aβ42 N and C terminus by replica-exchange molecular dynamics (REMD) simulations.49,
50
Experimental studies have shown that
these mutations promote Aβ oligomerization and enhance fibril formation without increasing formation of intermediate protofibril.26,
51
Obviously, alterations in the N
terminus affect the overall structural and dynamic property of Aβ peptide. On the other hand, these mutations may have different effects on the binding of metal ions because D7H has an additional histidine, H6R lacks one metal-binding histidine, and D7N may alter the susceptibility of its neighboring histidine to metal ions. Relative to wide-type Aβ42, we observed that the conformational spaces sampled by these mutations shrink to different degrees. These mutations increase the β-sheet propensity of both termini, especially the C-terminal region, whereas the overall tertiary structures of these mutations do not change significantly. Using both the MM/GBSA52 and MM/3D-RISM methods,53 we found that Aβ42 monomer becomes less stable upon mutation, which can be attributed to the increased solvation free energy. As a consequence, these mutations promote Aβ to aggregate into low molecular weight oligomers that are more neurotoxic. We further observed that both N- and C- terminals of Aβ42 predominantly populate low to mild structured conformations simultaneously, and mutations at its N terminus alter the relative populations differently. Conformations with highly ordered N terminus but rather mobile C terminus, and extremely disordered N terminus but rather rigid C terminus, are less 5
populated. Of importance, the same conformational populations were also found for Zn2+-bound Aβ42, suggesting that the binding of metal ions is independent of the correlation dynamics of both Aβ terminal regions. MATERIALS AND METHODS REMD simulations The same simulation protocol as used in our previous studies of amyloid peptides was applied in this work.13 Briefly, the NMR structure of full-length Aβ42 (PDB ID: 1Z0Q)54 was used as the starting conformation for REMD simulations. Based on the above Aβ42 structure, the initial structures of D7H, D7N, and H6R were prepared using the mutation tools in PyMOL program.55 All structures were modeled using Amber ff99SB force field56 and the solvent effect was represented by Onufriev-Bashford-Case GB implicit solvation model57 in order to enhance conformational sampling. In each REMD simulations, eight replicas with temperatures ranging from 280 K to 400 K were applied, leading to an acceptance ratio of about 20%. The integration time step was 2 fs and exchange between replicas was attempted every 5 ps. Previous REMD simulations including our own suggest that the fluctuation of cumulative helix content throughout simulations is an excellent indicator to check the convergence of REMD simulations.10, 24, 45, 58-60 As the cumulative helix percentage reaches to a plateau, we found that the system becomes equilibrium. Structural properties such as the population of radius of gyration and secondary structures at different time intervals are very similar (see Table I). To compare the conformational ensemble obtained from our REMD simulations with 6
available experimental data. The chemical shifts were calculated using SHIFTX61 (Fig. S1) based on the trajectory collected at 280 K (closest to the experimental temperature of 278 K).62 The Pearson correlation coefficients are 0.98, 0.84, and 0.90 for Cα, Hα, and N, respectively, in good agreement with experimental values. In addition, the J-coupling constants were also calculated using the Karplus equation,63, 64 and compared with the experimental data.65 Fig. S2 shows that a qualitative agreement between experimental and simulated J-coupling constants was obtained. It is still quite challenging to calculate the J-coupling constants of Aβ from simulations that are quantitatively in line with experiment.22 The total simulation time varies for each system, as it was found that 60 ns might be sufficient to reach convergence for Aβ42,13, 32 whereas much longer simulation time is necessary for its mutations. The total simulation times for D7H, D7N, and H6R are 7.52 µs (940 ns/replica), 4.21 µs (526 ns/replica), and 4.21 µs (526 ns/replica), respectively (Fig. S3). The last 100-ns MD simulations were used for data analysis. For comparison, we extended our previous REMD simulations of Aβ42 from 100 ns/replica to 200 ns/replica, and used the last 100-ns trajectory for all data analysis. All REMD simulations were performed using AMBER12 software package.66 Each trajectory containing 100,000 conformations collected at physiological temperature 310 K was used for data analysis. Free energy calculations The classical MM/GBSA method52 was applied to estimate the relative stability of Aβ and its mutations by calculation of their free energies (GMM/GBSA) in terms of the following 7
formula: G MM/GBSA = H MM/GBSA − TS NM ,
where T is the temperature, and the enthalpy HMM/GBSA is calculated in terms of MM/GBSA H MM/GBSA = EMM + Gsolvation
where EMM is the internal energy (a sum of electrostatic and van der Waals interaction MM/GBSA includes the electrostatic contribution and energies). The solvation free energy Gsolvation
the non-polar part that is proportional to the solvent accessible surface area of Aβ peptides. The entropy SNM was calculated by applying normal mode (NM) analysis on trajectories obtained from REMD simulations. However, previous studies have shown that MM/GBSA method may not be suitable for the evaluation of free energies of disordered proteins.24, 45 In particular, the NM method cannot discriminate the entropy difference between Aβ and its various mutations. Therefore, an alternative method, the three-dimensional reference interaction site model (3D-RISM) was applied to calculate the free energies of all peptides. This approach is based on the integral-equation theory of liquids and is able to account for both polar and non-polar features of the solvation structure.67, 68 However, the absolute value of solvation free energy varies with the closure relation used and only relative values are reasonably accurate.69-71 The robust Kovalenko-Hirata closure and the Gaussian fluctuation approximation as suggested by Chandler et al. was used to calculate the solvation free energy.53 Recently, a new method of computing the conformational entropy of Aβ peptides has been derived from the 3D-RISM theory.69, 70 8
TS 3D-RISM =
β 2
< ( H 3D-RISM − < H 3D-RISM >) 2 >
where β=1/(kBT) and kB is the Boltzmann constant, and denotes the ensemble average. In this method, the enthalpy (H3D-RISM) and entropy (TS3D-RISM) contributions to the free energy G3D-RISM are therefore different from those in the MM/GBSA method. The enthalpy H3D-RISM was obtained by combining the internal energy (EMM, the same as in the MM/GBSA method) and the solvation free energy calculated using the 3D-RISM method (Table 2). RESULTS AND DISCUSSION D7H, D7N, and H6R mutations do not change the intrinsic dynamics of Aβ42 The conformational dynamics of Aβ are crucial for its relevant functions. It is intriguing if mutations at its N terminus alter the overall intrinsic disorder and consequently enhance their aggregation tendency. Previously, we used the trajectory from REMD simulations to characterize the internal dynamics of Zn2+-bound Aβ40 and Aβ42.72 We calculated the NMR 15N-1H nuclear Overhauser enhancement (NOE) by rescaling the time scale of the trajectory at the highest temperature. The simulated NOE values were nearly in quantitative agreement with available experimental data. Here, we applied the same strategy to investigate the effect of point mutations on the overall tumbling motions of Aβ42 (Fig. S4). It is found that these mutations have little effect on the global tumbling dynamics of Aβ42, suggesting that these mutations are still intrinsically flexible. However, due to the difficulty of extracting reliable motional correlation times from MD simulations
9
and the limitations in interpreting experimental NMR relaxation data by use of a simple model, the subtle difference of local dynamics may be obscure. An alternative simple but insightful parameter, the relative solvent-accessible surface area (RSASA),73, 74 was used here to measure the relative flexibility of Aβ and its mutations. The RSASA is defined as the observed SASA calculated directly from conformational ensemble, divided by the correspondingly expected/idealized value of SASA. Figure 1 shows the RSASA for each residue of Aβ and its mutated species. The RSASA for all residues is not equal to one, clearly demonstrating that all Aβ peptides are intrinsically flexible. As expected, there are great fluctuations of conformations for each residue. Except residues Asp1 and Ala42, the remaining part with RSASA
