Identifying, By First-Principles Simulations, Cu[Amyloid-Î²] Species

Article pubs.acs.org/JPCB

Identifying, By First-Principles Simulations, Cu[Amyloid-β] Species Making Fenton-Type Reactions in Alzheimer’s Disease Giovanni La Penna,*,† Christelle Hureau,‡ Oliviero Andreussi,§ and Peter Faller‡ †

CNR - National Research Council of Italy, ICCOM - Institute for Chemistry of Organo−Metallic Compounds, via Madonna del Piano 10, I-50019 Sesto Fiorentino, Firenze, Italy ‡ CNRS - National Center for Scientific Research, LCC - Laboratory of Coordination Chemistry, 205 route de Narbonne, F-31077 Toulouse, France § University of Pisa, DCCI - Department of Chemistry and Industrial Chemistry, via Risorgimento 35, I-56126 Pisa, Italy S Supporting Information *

ABSTRACT: According to the amyloid cascade hypothesis, amyloid-β peptides (Aβ) play a causative role in Alzheimer’s disease (AD), of which oligomeric forms are proposed to be the most neurotoxic by provoking oxidative stress. Copper ions seem to play an important role as they are bound to Aβ in amyloid plaques, a hallmark of AD. Moreover, Cu−Aβ complexes are able to catalyze the production of hydrogen peroxide and hydroxyl radicals, and oligomeric Cu−Aβ was reported to be more reactive. The flexibility of the unstructured Aβ peptide leads to the formation of a multitude of different forms of both Cu(I) and Cu(II) complexes. This raised the question of the structure−function relationship. We address this question for the biologically relevant Fenton-type reaction. Computational models for the Cu−Aβ complex in monomeric and dimeric forms were built, and their redox behavior was analyzed together with their reactivity with peroxide. A set of 16 configurations of Cu−Aβ was studied and the configurations were classified into 3 groups: (A) configurations that evolve into a linearly bound and nonreactive Cu(I) coordination; (B) reactive configurations without large reorganization between the two Cu redox states; and (C) reactive configurations with an open structure in the Cu(I)−Aβ coordination, which have high water accessibility to Cu. All the structures that showed high reactivity with H2O2 (to form HO•) fall into class C. This means that within all the possible configurations, only some pools are able to produce efficiently the deleterious HO•, while the other pools are more inert. The characteristics of highly reactive configurations consist of a N−Cu(I)−N coordination with an angle far from 180° and high water crowding at the open side. This allows the side-on entrance of H2O2 and its cleavage to form a hydroxyl radical. Interestingly, the reactive Cu(I)−Aβ states originated mostly from the dimeric starting models, in agreement with the higher reactivity of oligomers. Our study gives a rationale for the Fenton-type reactivity of Cu−Aβ and how dimeric Cu−Aβ could lead to a higher reactivity. This opens a new therapeutic angle of attack against Cu−Aβ-based reactive oxygen species production.

■

of H2O2 and HO•.10,12 There is an active debate in the literature if Cu−Aβ is an antioxidant or a prooxidant in vivo. The antioxidant activity seems to be supported by the fact that the efficiency of Cu−Aβ as a catalyst for the production of H2O2 and HO• in the presence of O2 and ascorbate is less than that of Cu in an aqueous buffer solution.10,13 Indeed, Cu in a buffer, present as an aquo-complex or weakly bound to buffer components, is a very efficient ROS producer with O2/ ascorbate.14,15 The rate of HO• production catalyzed by “free” Cu is about 5−10 times faster than that catalyzed by Cu− Aβ.10,12 However, the free Cu concentration in biological systems is very low because of the presence of high concentrations of potential ligands, like proteins and amino acids.

INTRODUCTION Amyloid plaques are a hallmark of Alzheimer’s disease (AD) and their main component is the amyloid-β (Aβ) peptide. These plaques contain also high amounts of copper ions1 that are likely bound to Aβ as shown, for instance, by Raman spectroscopy.2 The connection between copper ions and Aβ metabolism has been confirmed by a large body of in vitro and in vivo evidence, and it was proposed that copper ions bind to Aβ under AD conditions.3 An important aspect of Cu−Aβ interactions related to AD is the catalytic role of Cu in the production of reactive oxygen species (ROS). An overproduction of ROS can induce oxidative stress, which seems to play an important role in AD progression.4−6 Although the possible in vivo mechanisms by which Cu−Aβ contributes to oxidative stress are not clear, in vitro experiments clearly showed that Cu−Aβ is able to produce ROS.7−11 In the presence of a physiological reductant like ascorbate and under aerobic conditions, Cu−Aβ catalyzes the production © 2013 American Chemical Society

Received: October 9, 2013 Revised: December 3, 2013 Published: December 6, 2013 16455

dx.doi.org/10.1021/jp410046w | J. Phys. Chem. B 2013, 117, 16455−16467

The Journal of Physical Chemistry B

Article

Cu(II)−Aβ and Cu(I)−Aβ) acting on the hypothetical relevant reaction coordinate (where reaction is the electron transfer to or from the system). The movements occur starting from configurations for the Cu(II)−Aβ complex built on the basis of experimental information. The reference scheme for the transformation is given in Figure 1. As an example, the

To compare the efficiency of ROS production by different catalysts, it makes more sense to compare Cu−Aβ with biologically relevant Cu-complexes. Indeed, such measurements indicate that the quantity of H2O2 and HO• in the presence of Cu−Aβ tends to be higher than that measured for tested Cuprotein/peptide complexes. For instance Cu−Aβ catalyzes the production of HO• more efficiently than metallothionein, serum albumin, and the peptide GHK, but is less efficient than the Cu-Gly2 complex.10,13,16 Thus, Cu−Aβ can have a prooxidant activity, and it might contribute to oxidative stress in AD. In recent years, the oligomeric forms of Aβ came into focus, as their presence correlate with the observed neurotoxicity. This led to the idea that oligomers, not monomers or amyloid fibrils, are mostly responsible for the neuronal death. Different types of toxic oligomers have been described, ranging from dimers up to relatively large structures.17−19 Recently, a new model of toxic oligomers based on several experimental methods was reported.20−22 It is interesting to note that the efficiency of oligomeric species of Cu−Aβ as catalysts for ROS production is better than that of monomers or amyloid fibrils,10,12,13,23 although the mechanism of the reaction is not known. During the past decade a lot of work has been done on the identification of the coordination of Cu(II) and Cu(I) to Aβ. A consensus was reached for the most populated conformers at neutral pH: Cu(I) is bound to imidazole N of His 13 and His 14 in a digonal fashion.5,24−26 Cu(II) is bound equatorially by the NH2 and CO (amide bond) of Asp 1, His 6 imidazole N, and either His 13 or 14 imidazole N.26,27 Apical ligands include water28 and carboxylates.29 However, the complexes in both I and II oxidation states are very flexible, and the more populated conformers are in equilibrium (fast on NMR scale) with several other less populated ones.5,26,29−31 This raises the question about the structure−function relationship, i.e., which conformation has which activity, to address central activities related to AD, like aggregation or redox reactivity of the involved chemical species. Indeed, an electrochemical study indicated that a very low populated form is responsible for the entire redox reaction. The electron transfer proceeds via an “inbetween” species, in which the coordination of both Cu redox states are very close and which are in equilibrium with the resting Cu(I) and Cu(II) species characterized by spectroscopy.32 In this work we address the question of the biological relevant Fenton-type reaction of Cu(I)−Aβ with H2O2 that produces the highly reactive hydroxyl radical HO•. The aim is to compare a multitude of different coordination spheres and to obtain an insight into the structural features needed to allow an efficient Fenton-type reaction.

Figure 1. Schematic representation of the computational experiment performed on each of the 16 configurations. In red is the manipulation of the model in the oxidized state. In blue is the process performed in the reduced state. In green is the final reoxidation process. The reaction coordinate chosen (x axis) is, in this work, the Cu coordination number (CN, see Methods for the details of its definition). The representation, in terms of atomic positions, of points 1−8 is the goal of this work.

oxidized state of one Cu(II)−Aβ complex in the sample is moved from high coordination numbers (point 1) toward low coordination numbers (point 3), these latter being favored in the reduced state (see the pathway drawn in red in Figure 1). Once the change in reaction coordinate is completed, the charge is changed according to the state favored by the reaction coordinate that has been reached (point 4, the starting point of the blue pathway in the figure). After the stabilization of the system with the new charge (point 5), the process is repeated in the opposite direction (the green pathway in Figure 1), i.e., movement in the reduced state toward higher coordination numbers (point 6) followed by reoxidation (points 7 and 8). The above pathways are performed by simulating the systems at room conditions (T = 300 K, a supercell filled with water molecules in the liquid state) and applying the extended Lagrangian, or Car−Parrinello, simulation algorithm.33 The Cu coordination is represented at the level of density-functional theory for explicit valence electrons with a setup tested on bioinorganic systems.34 As in most room-temperature calculations of this kind, the computed energy values are not accurate enough for a complete analysis of reorganization energy and reduction potential, i.e., those parameters that could be cast into the Marcus theory (see ref 35 and references therein). Thus, in this work, we limit our analysis to a classification of realistic oxidoreductive processes of Cu−Aβ, including exchange of Cu ligand atoms with water molecules in the first solvation layers. On the basis of this classification, the Cu(I)−Aβ models are investigated in terms of their reactivity with H2O2. The dissociation of peroxide models, introduced in the proximity of Cu(I), at the end of the simulation for each of the investigated models is tested, and from the 16 resulting experiments we discuss the essential characteristics of the rare Cu−Aβ complex that generates hydroxyl radicals. The result is a description of the conditions that make the Cu(II)−Aβ reduction more oriented toward the oxidation of the organic substrate and less oriented toward eventual protection from

■

METHODS The PDB notation for atoms of the imidazole His side chain is used: Nδ (sometimes identified as Nπ) is N 3 in the imidazole4-yl IUPAC notation for His side chain, while Nε (sometimes identified as Nτ) is N 1. The same notation is adopted for H atoms bonded to N atoms. The backbone amide atoms are identified as in the PDB, simply with N, H, C, and O. The Nand C-terminal capping groups, used in truncated peptide models, are acetyl and NH-methylamide groups (indicated as Ac and NHMe, respectively). The aim of the computational experiment is to move the Cu−Aβ system with a given charge (oxidation state, i.e., 16456



Article

Table 1. Summary of Coordination Topology and Geometry for Selected Points along the 16 Simulations (See Figure 1)a ; Q

point

CN

coord

ox red ox

2 5 8

4.0 4.0 3.5

sqp td sqp

ox red ox

2 5 8

4.0 3.0 3.0

− T T

ox red ox

2 5 8

4.0 2.6 3.0

sqp T T

ox red ox

2 5 8

4.0 3.0 3.0

sqp − T

ox red ox

2 5 8

4.3 3.0 4.0

− − −

ox red ox

2 5 8

4.1 3.1 4.0

− trig −

ox red ox

2 5 8

4.0 3.0 4.0

sqp − td

ox red ox

2 5 8

4.2 3.9 4.0

td − td

ox red ox

2 5 8

4.1 2.0 5.0

td dig tbp

ox red ox

2 5 8

4.1 3.0 3.7

sqp T sqp

ox red ox

2 5 8

4.0 4.0 3.2

sqp − −

ox red ox

2 5 8

4.0 3.0 3.0

td T T

ox red ox

2 5 8

4.0 3.0 4.0

sqp − sqp

ox red ox

2 5 8

4.0 3.0 3.8

sqp T −

ox red ox

2 5 8

4.0 3.0 3.3

sqp T T

ligand atoms Trajectory 1 Nδ(H6), Nε(H13), Nε(H14), Ow O(D1), Nδ(H6), Nε(H13), Nε(H14) Nδ(H6), Nε(H13), Nε(H14), Ow Trajectory 2 Nδ(H13), Nε(H14), Nε(H6), Ow Nδ(H13),Nε(H14) / Ow = Trajectory 3 N(D1), Nε(H6), Nδ(H13), Nδ(H14) Nδ(H13), Nδ(H14) / N(D1) Nε(H6), Nδ(H14) / N(D1) Trajectory 4 N(D1), Nδ(H6), Nδ(H14), Ow Nδ(H6), Nδ(H14), N(D1) Nδ(H6), Nδ(H14) / N(D1) Trajectory 5 N(D1), Nδ(H6),Nδ(H13), Ow Nδ(H6), Ow / Nδ(H13) Nδ(H6),Nδ(H13),Ow,Ow Trajectory 6 Nδ(H6), Nε(H13), Nδ(H14), Ow Nδ(H6), Nε(H13), Nδ(H14) N(D1), Nδ(H6), Nε(H13), Nδ(H14) Trajectory 7 O(D1), Nε(H6), Nδ(H14), Ow Nε(H6), Nδ(H14), Ow Nε(H6), Nδ(H14), Ow, Ow Trajectory 8 N(D1), Nε(H13), Nε(H14), Ow = = Trajectory 9 N(D1), Oδ(D1), Nδ(H6), Ow Nδ(H6), Ow N(D1), Oδ(D1), Nδ(H6), Nδ(H14), Ow Trajectory 10 N(D1), Nδ(H6), Nδ(H14), Ow N(D1), O(D1), Ow N(D1), O(D1), Nδ(H13), Ow Trajectory 11 Nδ(H6), Nε(H13), Ow, Ow Nδ(H6), Nε(H13), Nδ(H14), Ow Nδ(H6), Nδ(H14), Ow Trajectory 12 Nδ(H6), Nε(H13), Nε(H14), Ow Nδ(H6), Nε(H14) / Nε(H13) = Trajectory 13 Nδ(H13), Nδ(H14), Ow, Ow Nδ(H13), Nδ(H14), Ow Nδ(H6), Nδ(H13), Nδ(H14), Ow Trajectory 14 N(D1), Nε(H13), Nε(H14), Ow N(D1), Nε(H13) / Ow Oδ(D1), N(D1), Nε(H13), Ow Trajectory 15 Nδ(H6), Nδ(H13), Nε(H14), Ow Nδ(H6), Nδ(H13) / Nε(H14) approach of Ow (2.5 Å)

16457

, (V) −0.71 +0.11 −0.16 −0.16 +0.23 +0.25 −0.84 +0.30 −0.02 −0.97 +0.11 +0.12 −0.44 +0.31** −0.49 −0.66 +0.20 −0.37 −0.39 +0.16 −0.03 −0.55 +0.27 −0.50 −0.68 +0.53 −0.77 −0.41 +0.14 −0.32 −0.60 −0.11 +0.23 −0.59 +0.24 +0.04 −0.77* +0.22 −0.83 −0.16 +0.28 −0.60 −0.75 +0.35 +0.00



Article

Table 1. continued Q ox red ox

point 2 5 8

CN 4.0 2.3 5.0

coord sqp − tbp

ligand atoms Trajectory 16 N(D1), Nδ(H6), Nε(H13), Ow Nδ(H6) / Nε(H13) N(D1), O(D10), Nδ(H6), Nε(H13), Nε(H14)

, (V) −0.75 +0.34 −0.79

a

Symbols: oct = octahedral; sqp = square planar; tbp = trigonal bipyramidal; td = tetrahedral; trig = trigonal; T = T-like; dig = digonal; ox = oxidized (Q = 0); red = reduced (Q = −1); , = energy contribution to the reduction potential in the given sample; CN = coordination number (only Cu−N/ O). * Absolute minimum of energy in oxidized form. **Absolute minimum of energy in reduced form.

ROS damages, like superoxide dismutation. In summary, some of the trajectories represent species able to oxidize an organic substrate like ascorbate, while other trajectories represent species that are also able to reoxidize Cu itself using ROS in the vicinity. The different methods used in the model construction, simulation, and analysis are the same used previously for several reduced and oxidized Cu−Aβ models36−39 as well as for Zn− Aβ models.40,41 The details are reported in the Supporting Information (SI). Here we outline a brief summary, including the most relevant information required to understand the analysis of results. For each of the 16 models investigated in this work, the steps summarized below are performed. 1. An empirical model for the ligand peptide, in monomeric (8 configurations) and dimeric (8 configurations) forms, is used to search for nonoverlapping structures of Aβ(1− 16) satisfying the conditions imposed by experiments in terms of possible coordination topologies. Monte Carlo random walks (MC-RW) are used, followed by selection of candidate configurations, according to the presence of atoms N and O of Asp 1, and Nδ or Nε of His 6, His 13, and His 14 together within the same sphere of radius 3 Å. The ligand peptide is DAEFRHDSGYEVHHQK with the charged C-terminus and a neutral N-terminus (the group NH2). The nontruncated ligand has, therefore, a charge Q = −3. 2. The selected configurations (16, configurations 1−8 monomeric and 9−16 dimeric) of the one or two peptides (for monomeric and dimeric cases, respectively) are completed by inserting one Cu ion (in the empirical form of a dummy counterion38,42) into the center of the selection sphere. Then the configurations are merged into simulation supercells filled with water molecules, resulting in empirical models of 3871 and 16708 atoms for monomeric and dimeric samples, respectively. After thermalization of such models, the peptides are truncated to a first-principles manageable model. The 1−16 peptide is modified into two segments, H2N-DGGGGHD-NHMe (1−7 region, Q = −2) and Ac-HH-NHMe (13−14 region, Q = 0). While the truncation of most of the side chains to a single H atom is required by computational resources, the Asp 7 side chain is kept in the truncated sequence because of the interactions with the imidazole side chains observed in previous models of Cu−Aβ. In the dimeric cases (trajectories 9−16), the first segment belongs to one peptide and the second segment to the other peptide. The truncated models are again merged into a simulation cell filled by water molecules and counterions to neutralize the simulation cell. 3. An empirical model for the cell is simulated by routine molecular dynamics (MD) methods to provide a

configuration for solvent molecules and counterions approximating in vitro standard conditions (water solution of the complex). These models are all composed of 1056 atoms in the oxidized (Q = 0) state. 4. The final configuration of the previous step is then simulated within a density-functional (DFT) model and using the extended Lagrangian formalism (Car− Parrinello molecular dynamics, CP-MD).43 An external harmonic force based on the Cu-coordination number CN (see Supporting Information for details) is used to move from one oxidized model to another, crossing the reduced state.44 5. At the end of the cycle, the water molecule closest to Cu is replaced with HO2− and Cu is reduced, providing a complex with the same charge of the oxidized form. After an energy minimization of the system, the O−O distance of peroxide is stretched with an external force to probe the resistance of peroxide against dissociation and hydroxyl radical production. Steps 2−4 in the protocol reported above are repeated for 16 initial configurations obtained by step 1. Therefore, 16 CP-MD independent trajectories were performed, each collecting 2.3 ps (including thermalization steps, see tables in Supporting Information). Step 5 is performed for all of the trajectories displaying at least one water molecule in the first Cucoordination sphere (11 trajectories over 16). For the other 5 samples, peroxide replaces a second-sphere water molecule close to the Cu-bound water molecule (models 2s, 5s, 7s, 11s, and 15s derived by models 2, 5, 7, 11, and 15, respectively) to model water-mediated outer-sphere electron transfer between peroxide and Cu. Several configurations selected along the 16 trajectories are also studied in the vacuum and in mean-field models for the water solvent45 for more accurate energy calculations. The details of the procedure are reported in the Supporting Information.

■

RESULTS AND DISCUSSION The oxidoreductive process of Cu−Aβ, as schematically described in Figure 1, is analyzed for 16 different configurations represented by truncated models. The setup and behavior of the 16 different trajectories is summarized in Table 1. In all the trajectories, we started from configurations with high His crowding around Cu(II) (2 or 3 His side chains with different Cu(II)-binding topologies, i.e., with Nδ or Nε binding Cu), O and/or N of Asp 1 together with O atoms (Ow) of water molecules in the Cu(II) coordination sphere. The choice is dictated by the experimental evidence that these residues are the most likely ligands in the most populated conformers26,28 but are also involved in the redox active state.46 Indeed, it is known that in monomeric Cu(I)−Aβ the most populated conformation is Cu(I) linearly bound to His 13 and 16458



Article

His 14; other possible conformations have a lower statistical weight, where His 6 replaces one of the two His 13−14.5,26,38,47 In the oxidized state the Cu(II)−Aβ complex can involve more ligand topologies, involving, at physiological pH, Asp 1, Ala 2, the three His (6, 13, and 14), and a water molecule in apical position (see Introduction and ref 26 and references therein). On the basis of the electrochemical data,32 we assume that the transition from species involving the N-terminus (more frequent in the oxidized state) toward N-terminus-free Cu coordinations (more likely in the reduced state) corresponds to a large reorganization energy for the redox process. More precisely, the electrochemical studies suggested that a low populated conformer (called “in-between” or “entatic-like” state) is able to efficiently cycle between the two oxidation states. This “in-between” state has structural features different from those displayed by the above-described most populated conformers, which are associated with sluggish redox chemistry and that could be considered as sort of “ground” or “resting” states. The latter resting states are in equilibrium with the redox competent “in-between” state. On the basis of the assumptions above, we started with 16 different models where the main ligands were bound or at least nearby the Cu(II) center. Eight models were built from monomeric Cu(II)−Aβ(1−16) and simulated at room conditions via biased empirical models. After thermalization of the peptide environment, the models were truncated and part of the Aβ(1−7) and Aβ(13−14) fragments were kept in DFT models. These models were used to explore their redox behavior by performing a biased reduction and oxidation and by modulating the coordination sphere (see Methods and Figure 1). The other eight models were obtained as simplifications of Aβ(1−7) and Aβ′(13−14), originating from two different peptides, Aβ and Aβ′ (i.e., from a dimer Cu[Aβ]2), respectively, and one Cu ion between (see Methods in Supporting Information for details). These models underwent the same analysis. In the following, the term “point” refers to the points indicated in Figure 1. Reduction Potential of Cu−Aβ Is Modulated by the Ligand. In Figure 2, the values for the total energy (top and middle panels) and reduction energy (bottom panel, see Methods for details) for snapshots obtained within the trajectories of the 16 models are reported as a function of the coordination number CN. Even though in Figure 2 the dispersion of values for energy is high, some trends can be extrapolated by the whole set of points for the 16 trajectories. Along the oxidized curve (Figure 2, top panel), on average a decrease in coordination (CN = 3) corresponds to an increase in energy, while an increase in coordination (CN = 5) corresponds to a limited number of configurations, mostly with energy lower than that of CN = 4. This picture is consistent with the chemistry of the free Cu(II) in an aqueous solution, where the coordination number 5 is well-represented, in exchange with CN = 4. Also, in proteins where the Cu(II)-binding site is distorted by the rigid protein matrix (like in crystallized proteins), the coordination of a water molecule in the fifth coordination position is often observed.48 The behavior of energy as a function of the coordination number changes when the Cu−Aβ complex is in the reduced form (Figure 2, middle panel). The values of CN = 3 is the more represented in terms of low-energy values, and the lowest-energy values have indifferently CN = 3 and 4. In the reduced state, the energy landscape is flatter in the 2−4 CN range, whereas in the oxidized state, high coordination numbers

Figure 2. Energy for each configuration as a function of the coordination number (CN). Oxidized state (top panel) and reduced state (center panel). The zero of energy is the minimal energy in the oxidized state (trajectory 13 at CN = 4). Bottom panel: energy difference between reduced and oxidized state for each configuration. The CN is defined as a continuous variable (see Supporting Information for details). The points in red, blue, and green are points 2, 5, and 8, respectively (see Figure 1 and Table 1).

are more likely and more energetically favored. The larger coordination flexibility in the reduced state is expected because of the d10 electron configuration. However, the low chance for high coordination numbers in the reduced state cannot be accounted by the latter simplification. In both the oxidized and reduced states, the complex has its lowest energy with CN = 4. In the oxidized state the Cu coordination geometry is distorted square-planar, with His 13, His 14, and two water molecules in the coordination sphere of Cu (trajectory 13, see Figure 3, Table 1, and Results in Supporting Information for details). This is a trajectory where the 1−7 peptide fragment initially bound to Cu is expelled. In Figure 3, the minimal energy structure is displayed, representing the replacement of the 1−7 peptide with two water molecules in the coordination sphere of Cu. However, the movement of His 6 away from Cu is not large, and His 6 side-chain binding to Cu is easily recovered when the complex is reoxidized (see below). This minimal energy structure indicates that the replacement of the 1−7 segment with water molecules in the Cu binding is energetically favored in the oxidized state. Because mechanical effects in the 1−16 segment seem to increase the energy required by the folding of the Nterminus toward Cu (see below, the reconstructed empirical models), the water accessibility to the Cu coordination site is critical in increasing the statistical weight for the coordination 16459



Article

more easily reoxidized (the reduction potential is negative compared to the couple H+/H2 calculated in the same conditions) because Cu can keep the high crowding of ligand atoms around itself, and the crowding stabilizes the oxidized state more than the reduced state (as the comparison between top and center panels of Figure 2 shows). The bottom panel of Figure 2 shows that a positive reduction potential with respect to the H+/H2 couple is obtained, on average, only for CN ≤ 3 (see the interpolation curve). Even though an accurate calculation of reduction potential deserves more accurate methods (see ref 49 for applications of post Hartree−Fock methods on the same system), the approximate statistics reported here show that the reduction potential of the Cu−Aβ complex is largely modulated by the ligand mechanics. We shall see below that the nature of the ligand atoms (water molecules or part of Aβ peptides), as well as the geometry of the peptide coordination, strongly affects the chance of reorganization and the possibility for the reduced form to access a water-soluble oxidant like hydrogen peroxide. Three Different Pathways for the Oxidoreductive Process of Cu−Aβ. Surveying the trajectories of the 16 initial configurations, we were able to classify them into three groups that will be identified as A, B, and C. We define class A as including the Cu(II) → Cu(I) → Cu(II) trajectories (see Figure 4) that end up with the typical Cu(I) coordination (i.e.,

Figure 3. Minimal energy configuration obtained for the oxidized solute and Cu-bound water molecules in the water solvent mean-field model (trajectory 13, point 2, oxidized state with CN = 4). See text and Figure 1 for definition of points. Color scheme: C, gray; N, blue; O, red; Cu, orange; H white. Atomic radii are arbitrary. This and the following figures are drawn with the VMD program.52

of Asp 1 to Cu(II), as it is observed in experiments. Indeed, the configurations with the coordination topology of the minimal energy configuration (two water molecules and no atoms of Asp 1 bound to Cu) are represented in 4 cases out of a total of 48 samples for points 2, 5, and 8 (see Table 1). On the other hand, 23 configurations out of 48 display one atom of Asp 1 and one of the His side chains in the Cu-coordination sphere. However, the calculations reported in this work do not provide the thermal weight of the different binding sites because larger models, a proper sampling, and a less biased set of initial configurations must be used for a correct estimate. In the reduced state, the complex has its lowest energy with CN = 3, in a T-like coordination geometry (trajectory 5, see Table 1 and Results in Supporting Information). Copper’s affinity for water is lower in the reduced state (see below), and one of the water molecules that is present in the oxidized state is transiently replaced by the carbonyl group of one of the nearby residues. The Cu coordination of the CO group allows a rearrangement of the N−Cu−N coordination, from an angle of about 90° to a value closer to the linear angle (∼150°, see the angle evolution for trajectory 5 in Supporting Information) in the final reoxidized state. At the end of the reoxidation of trajectory 5 (point 8), the configuration becomes similar to the minimal energy configuration in the oxidized state (trajectory 13, see above): a His−Cu−His segment with two almost orthogonal Cu−water bonds. The energy difference between reduced and oxidized states (the energy contribution to the reduction potential in the given modeled samples, Figure 2, bottom panel) clearly shows that, on average, the lower the coordination number, the higher the reduction potential. Because the energy of the final reduced state is not significantly affected by the Cu coordination (Figure 2, center panel), the reduction potential of Cu−Aβ is ruled by the coordination and structure of the oxidized state: Cu(II)− Aβ can be an oxidant in certain coordination spheres, but the result of its reduction can be the stable low-coordination Cu(I)−Aβ only when the ligand mechanics allows the reorganization of ligand and water molecules in the vicinity of Cu. In the other cases the formed complex Cu(I)−Aβ can be

Figure 4. Schematic representation of the different behaviors of the 16 trajectories (see text for details).

that with Cu linearly bound by two His side chains). These trajectories show that once point 5 is reached, an end-point 8 similar to 2 (high Cu(II) coordination numbers) is not recovered and the trajectories keep the low Cu(II) coordination numbers despite the change of charge of the complex. Class B contains the trajectories close to their respective origins, i.e., the Cu complex moves easily around the starting high Cu coordination numbers (points 2, 5, and 8 are all similar). When this event is possible, even with ligand reshuffling along the pathway, a low reorganization energy is estimated. Class C includes trajectories that end up with a Cu(II) coordination that is significantly different from the starting Cu(II)−Aβ point but do not pass through a stable reduced state (the end-point of class A). These latter cases, where point 8 represents a different state from point 2 of the same trajectory, are frustrated high Cu-coordination configurations that exchange ligands in the coordination sphere of Cu. These trajectories can involve a large reorganization energy due to the change of Cu ligands in order to maintain the high CN values during the change of the complex charge. Therefore, these trajectories represent extended ligand reshuffling, keeping high ligand crowding available to Cu and providing a wider spectrum of coordination types (and therefore reactivities) available to Cu compared to the other two classes (A and B). The schematic frame that will serve to describe the following observations is displayed in Figure 4. 16460



Article

In Table 1, the geometries of the configurations sampled at three relevant points in the pathway are briefly described. A more detailed description is reported in the Supporting Information Results section. The configurations are schematically described in terms of the ligand atoms, L, that are within the first Cu-coordination sphere (d(Cu−L) < 2.5 Å). The parameters used to approximately define the geometry in terms of ideal systems are reported in the Methods section of the Supporting Information. As an example of a class A trajectory, three representative snapshots of trajectory 4 are displayed in Figure 5. The initial 3His coordination is rapidly destroyed and a square-planar coordination is formed, with the His−Cu−His bonds orthogonal to N(Asp 1)−Cu−Ow (point 2, oxidized, top panel). The linear array of His−Cu−His bonds is kept, irrespective of imposed changes in coordination number and charge (points 5, reduced, and 8, reoxidized, center and bottom panels, respectively); the third His (His 13) is too far away to be recovered by Cu when the complex is reoxidized. The initially Cu-bound water molecule that is expelled when the complex is reduced (point 5, center panel) is not recovered from the solvent bath after reoxidation. The N atom of Asp 1 assists the T-like configuration that is formed after reduction. As an example of a class B trajectory, three representative snapshots of trajectory 15 are displayed in Figure 6. The 3-His distorted Cu coordination appears to be kept irrespective of the target value for the coordination number and of the oxidation state. The effect of the change of charge (reduction) is the lower solvation of Cu (point 5, center panel). Upon reoxidation (point 8, bottom panel), the water molecule in the first coordination sphere of Cu is recovered from a direction slightly different than that of the original oxidized configuration (point 2, top panel). Although this event implies a significant reorganization energy (the estimated lower limit displayed is for this trajectory in Figure 7), the ligand reshuffling does not change the topology of the coordination. Class C does not contain a representative individual trajectory because the final points (8) can be very different among different trajectories of the same class. For instance, trajectory 5 ends with a configuration not too dissimilar to that represented by class A, i.e., a linear His−Cu−His coordination, but filled by two orthogonal Cu−water bonds (point 8, see Supporting Information); trajectory 16 displays the replacement, in the Cu-coordination sphere, of a water molecule (point 2, see Supporting Information) with the N and O ligand atoms of Asp 1 together with the recovery of one His side chain; trajectory 14 displays the exchange of one His side chain with the carboxylate group of Asp 1. In some cases, the 2-His Cu coordination that is obtained in the reduced form is neither the digonal stable one (class A) nor able to recover the third His rapidly, like in class B. In those cases where the 2-His nondigonal coordination is filled (when CN = 4 is imposed) with water molecules, the accessibility of reactant to Cu is expected to be higher than in the cases belonging to class A or B. The peptide ligand is not able to restore the Cu-coordination sphere with peptide atoms, allowing Cu to interact with reactant molecules in the solvent. More details describing the observations reported above are provided in the Supporting Information. Interestingly, trajectories of class A (trajectories 2, 3, and 4) are more likely when generated by monomeric empirical models. This indicates that monomeric forms have more chances to keep Cu more stable in the reduced form or, in

Figure 5. Three snapshots obtained along trajectory 4 (class A): point 2 (first oxidized configuration, top panel); point 5 (reduced configuration, center panel); point 8 (reoxidized configuration, bottom panel). Only the solute and Cu-bound water molecules (d(O−Cu) < 2.5 Å) are displayed. Configurations have been rotated for graphical purposes.

other words, more sluggish as a reductant. The lower limit of reorganization energy for the reduction of Cu(II) in class A trajectories is in the range of that of class B (Figure 7). Therefore, the different reactivity between the two classes is related not only to this quantity but also to the different availability of ligand atoms that can be used to increase the 16461



Article

Figure 7. Reorganization energy (λ) for each of the 16 trajectories. Circles are for the extreme points in the reduction, squares are for the extreme points in reoxidation.

activation of hydrogen peroxide by Cu(I)−Aβ is possible, a HO2− molecule was put in place of the Cu-bound water molecule in each of the final configurations obtained for Cu− Aβ in the 16 trajectories described in the previous section. When such a Cu-bound water molecule was not present, i.e., trajectories 3, 4, 6, 12, and 16, the HO2− ion was placed in the second shell of trajectories 2, 5, 7, 11, and 15, respectively. These trajectories will be identified as 2s, 5s, 7s, 11s, and 15s in the following report (the “s” suffix is for “second shell”). The insertion of peroxide in the second Cu-coordination shell in contact with a Cu-bound water molecule provides a configuration possibly mimicking an outer-sphere electron transfer mediated by water. The starting Cu−Aβ complex was in the reduced form. After the settling of the HO2− ion in contact with the Cu-coordination site or with the Cu-bound water molecule, the O−O bond was elongated at low temperature with a harmonic force associated to an equilibrium distance of 2 Å (see Methods in Supporting Information for details). This experiment addresses the reactivity of the 16 reasonable Cu(I)−Aβ complexes, obtained via the procedure reported above, as Fenton catalysts. In Figure 8, the low-temperature average O−O distances are compared to the average distances observed during the application of the external potential. The inset displays the same values for peroxide in water (with no Cu) and for 3 Cu complexes in water built with different geometries. The charge is zero for all the simulated samples (Cu is reduced and, when

Figure 6. Three snapshots obtained along trajectory 15 (class B): point 2 (first oxidized configuration, top panel); point 5 (reduced configuration, center panel); point 8 (reoxidized configuration, bottom panel). Only the solute and Cu-bound water molecules (d(O−Cu) < 2.5 Å) are displayed. Configurations have been rotated for graphical purposes. Figure 8. Average O−O distance (error-bar is the root-mean-square error) at T = 50 K (□) and at T = 150 K and in the presence of the external stretching force (○), for the 16 simulated trajectories. The inset reports the value for peroxide in water (left) and for the three Cu+ complexes in water (from left to right): peroxide bonding Cu endon in apical position, peroxide equatorial with O−O bond in the equatorial plane, and peroxide equatorial with O−O bond perpendicular to the Cu−water plane.

coordination number of Cu. Despite the large error affecting reorganization energy, it can be noticed that the scattering in the lower limit of the reorganization energy in the samples obtained by dimeric models (trajectories 9−16) is larger than that in monomeric models (trajectories 1−8, see Figure 7). Effect of Cu Coordination on the Reactivity of the Cu(I)−Aβ Complex with Peroxide. To probe whether the 16462



Article

coordination sphere of Cu with two Cu-bound His side chains forced to form N−Cu bonds at an angle smaller than 180° allows side-on access of H2O2 to Cu, easily generating hydroxyl radicals. These latter are aggressive against all the organic groups nearby. In Figure 10, two configurations are displayed within trajectory 11: the configuration before (left panel) and after (right panel) the O−O stretching. Because one of the expected products of the dissociation is a biradical form, the simple analysis of spin density is a useful tool for understanding where the unpaired electrons are displaced (see Methods in Supporting Information). Therefore, this density is also displayed in Figure 10. The HO2− ion is coordinated to Cu (d(O−Cu) = 2.1 Å) with the O−O bond approximately in the N(His)−Cu−N(His) plane. The spin density (the system is modeled with Sz =0, but with a significant component of S = 1) is concentrated around the O atom farther from Cu. The fast dissociation (right panel) is accompanied by the hydrogen extraction from a close-by water molecule, thus showing the importance of the high degree of solvation anti to the distorted His−Cu−His moiety. The larger amount of spin density is distributed among the Cu2+−OH− bond and the leaving OH group, the latter exiting from the coordination site as a OH• radical. The comparison between energies, evaluated in the vacuum, for the species involved in different spin states shows that in the initial configuration, the Sz = 0 state is more stable than the Sz = 1 state by 19.7 kJ/mol. In the final configuration, the Sz =1 state is more stable by 46 kJ/mol than in the state Sz =0. The change in potential energy between the two most stable states is −383 kJ/mol, including the extraction of a proton from the bulk water (see Methods in Supporting Information for the estimate of this quantity). The conversion from the antiferromagnetic to the ferromagnetic state occurs in the Sz = 0 state, as this is denoted by the increase in spin contamination along the CP-MD simulation of trajectory 11. Noticeably, the component with Sz = 0 is resonant with a Cu3+ species interacting with two OH− anions (a diamagnetic form); the net positive charge comes from the proton extracted from bulk water. This resonant form contributes to the formation of the biradical, decreasing the repulsion between states with the same spin component. Spin-density analysis, therefore, supports a product described as a biradical species with Cu2+ coordinated to Aβ and the hydroxyl radical exiting from the Cu coordination site. An interesting feature of Figure 10 is the arrangement of the two carboxylate groups (Asp 7 and 1 at top and bottom, respectively, of the panels). The two groups are aligned with the Cu-bound OH group, with Asp 7 close enough to form a weak hydrogen bond: d(H−Oδ(Asp 7)) = 2.0 Å at T = 50 K and 2.8 Å at T = 150 K. Asp 1 can not approach closer to the Cu-coordination site with Oδ because the N and O atoms are close to Cu. On the other hand, the entrance of carboxylate in the Cu-coordination sphere (like in trajectories 9 and 14, both of class C) hinders the Cu accessibility to peroxide, inhibiting the formation of hydroxyl radicals by a different mechanism. As expected, not all trajectories of class C can easily dissociate peroxide because this feature depends on the chance to keep the reactive His−Cu−His arrangement upon a change of oxidation state. For instance, trajectory 13, that in point 2 displays the minimal energy in both oxidized and reduced states, despite the easy detachment of the fragment 1−7, is able to recover partially the His 6 binding. In the reduced state, one water molecule is expelled from the Cu-coordination sphere and it is easily replaced by His 6 when the complex is

Cu is removed, it is replaced by H). Among the Cu−water complexes, the one with peroxide approaching on the Cu coordination end-on from the apical position (configuration number 1) is the most inert. In contrast, the most prone to dissociation is the complex where the peroxide is initially oriented side-on toward Cu. When this initial coordination is adopted, the two O atoms move more easily apart, laying approximately in a plane formed by Cu and the other two ligands. In Figure 9, the configuration corresponding to the

Figure 9. Configuration of maximally elongated O−O distance obtained for Cu(I)+HO2− in a sample of water (i.e., with no peptide ligands), and with HO2− replacing one of the four Cu-bound water molecules in equatorial position for the oxidized Cu complex. The two atoms of peroxide are on the right of Cu.

maximally elongated O−O bond distance is displayed: the hydrogen-bond network that anticipates the formation of two OH groups is shown on the right-hand side. When a peptide ligand is present around the reaction center, the dissociation can be assisted or inhibited by ligand atoms. By observing the behavior of the 16 Cu−Aβ complexes, it can be noticed that values of O−O distances consistent with enhanced dissociation ability are displayed for 3 samples in class C, i.e., trajectories 7, 8, and 11 (Figure 8). One particular complex (produced by trajectory number 11) is particularly efficient. Within species that belong to class C, the important features in order to have an efficient dissociation of peroxide are (i) two His−Cu bonds that are forced to make a small angle (∼90°) and (ii) the other ligand is a single water molecule with enough empty space around Cu to let a second O approach Cu side-on. As a reference for the effects of dissociation, the configurations where peroxide is in the second coordination sphere (beyond 2.5 Å, trajectories 2s, 5s, 7s, 11s, and 15s) are almost inert toward the forced dissociation, representing a peroxide ion weakly interacting with the Cu−Aβ complex. The complexes of class A display a low propensity to keep peroxide coordinated to Cu; in the same way, they do not keep water molecules when in the reduced form in the first coordination sphere of Cu. The results show that every class C trajectory that is able to attract at least one water molecule at close distance in the first 16463



Article

Figure 10. Configurations obtained for trajectory 11, with HO2− replacing the Cu-bound water molecule and Cu reduced. Left panel: configuration obtained after 150 fs at T = 50 K. Right panel: after an additional 150 fs at T = 150 K, together with O−O stretching of peroxide. The isosurfaces of spin density are at 0.1 within the range 0−0.33.

reoxidized. Thus, the potentially aggressive His−Cu−His distorted arrangement is rapidly silenced by His 6 that is in the vicinity of Cu also in the reduced state. This is not possible when the 1−7 fragment is replaced by solvent. Dimeric Complexes Are More Reactive than Monomeric Complexes. In the description above, the model consisted of the Cu ion and the two portions of the Aβ peptide, Aβ(1−7) and Aβ(13−14), simplified by truncation of side chains not directly related with Cu binding. However, most of the experimental results concerning the catalysis of the Fentontype reaction by Cu−Aβ were obtained with the Cu-binding domain Aβ(1−16) or with the full length Aβ(1−42). To be closer to the experimental results, we reconstructed the Aβ(1− 16) fragment over the truncated models of Aβ(1−7) and Aβ(13−14) and the reduction−oxidation cycle of Cu−Aβ(1− 16) was revisited according to the interactions that can be observed in these more relevant models. This was also motivated by our results of the structural studies of Cu(I) binding to Aβ, where it was shown, by using the full Cu-binding domain Aβ(1−16), that long-range effects were important for stabilizing the binding site as detected by experimental methods.38 First, the reconstruction (see Supporting Information for the details about the procedure) was possible for all 16 trajectories, indicating that the truncated structures studied in this work did not involve unrealistic conformations for the nontruncated peptides. Second, the effects of the explicit side chains are different in monomeric (trajectories 1−8) and dimeric (9−16) models, while the effects of changes in Cu-coordination number and oxidation state imposed along each first-principles trajectory to the truncated systems appeared less dependent on the subset of trajectories where they were applied (see previous subsection). For the monomers, the initial configurations correspond to high crowding around Cu of Asp 1, His 6, 13, and 14, where side chains come from the same peptide chain. Because the reconstruction resides in the initial complete structure (i.e.,

before truncation), the process led to a loop structure between the two parts (Aβ(1−7) and Aβ(13−14), see the last molecular drawing for each trajectory in Results of Supporting Information). This loop structure exerts mechanical forces on the peptide and, on average, tends to expose side chains to the solvent.38 In contrast, for the dimeric model, where the reconstruction needs two separate Aβ(1−16) fragments bound to the single Cu ion, no such loop structure under tension was present in trajectories 9−16 studied in this work. In the dimeric forms, the shape of the Cu−[Aβ(1−16)]2 complex is ruled by the portions of the peptide chains that are not involved in the Cu binding because these portions are not constrained by the interactions within the Cu-binding region. Nevertheless, the orientation and flexibility of these portions are limited in different extents because of the different structures of the Cu binding site that makes the peptide cross-link. Third, and most interestingly, the study of the reconstructed models confirmed the tendency observed above, that the dimeric models Cu−[Aβ(1−16)]2 fall mostly in the class C, and that they were the most reactive in the Fenton-type reaction. This might be explained by the strained loop structure in the monomer, which has the tendency to move apart the two binding units, i.e., Aβ(1−7), with Asp 1 and His 6, and Aβ(13− 14), with the two His. As such, the reoxidation is less likely as this last process needs a higher ligand crowding around Cu(II). The Cu(I) form becomes more easily trapped in the more stable digonal coordination (like His−Cu(I)−His) that is the signature of class A. In contrast, in the dimer no such constraint is present because of the absence of the loop; hence, the ligands have the tendency to stay near the Cu ion, the high ligand crowding is maintained, and an easier reoxidation of the Cu center is possible. This is very intriguing because experiments showed that the reactivity in ROS production of oligomeric species of Cu−Aβ is more than that of Cu−Aβ monomers.10,12,13,23 The present calculations give, for the first time, a molecular mechanism consistent with this observation. This is of high 16464



Article

source agreements. This material is available free of charge via the Internet at http://pubs.acs.org.

relevance in the context of Alzheimer’s disease as it explains why oligomeric species (like the dimer in our case) are more neurotoxic than the monomers.

■

■

CONCLUSIONS The aim of this work is to understand the structure−function relationship of the highly flexible Cu−Aβ complex in catalyzing the Fenton-type reaction; the latter is supposed to contribute to oxidative stress in Alzheimer’s disease. First, the redox behavior of different structures was probed by using 16 different starting models of Cu(II)−Aβ; the main ligand residues (Asp 1 and His 6, 13, and 14) were in the proximity of Cu. The trajectories could be divided into three classes: A, B, and C. On the basis of this classification, the reactivity of the Cu(I)− Aβ with H2O2 was investigated. The Fenton-type active structures all fall into class C and had in common (i) Cu that was accessible by solvent molecules (i.e., H2O2 had access to the inner sphere to bind Cu side-on); (ii) Cu(I) that was coordinated by two ligands (His, but also N(Asp 1)) in a bent geometry (angle N−Cu−N smaller than 180°); and (iii) these latter configurations displaying a high water crowding in the third and fourth available coordination positions. In contrast, Class A, with digonal, nonbent Cu(I) geometry, stabilizes Cu(I) in a form where H2O2 does not bind efficiently to Cu(I). Class B is potentially redox active, and this might be the species observed in electrochemistry.32 However, species in class B react slowly with H2O2 because there is no easy side-on access of H2O2 to the Cu(I) site. Water access to Cu is necessary as second sphere H2O2 interactions are not enough to activate the Fenton-type reaction. The access of more than one water molecule to Cu provides possible side-on access of peroxide. In general, this supports the view that the very dynamic Cu− Aβ complex has a multitude of different conformations, but only some of them are redox competent (complexes from class B) or prone to promote Fenton-type reactions (complexes from class C). This is in agreement with the experimental observations showing that HO• production for Cu−Aβ is about 5 times lower than that for Cu ion in buffer.10,11,50,51 Cu in buffer is quite reactive in the Fenton-type reaction. For Cu−Aβ, only a portion of the accessible conformations are found able to perform efficiently the H2O2 reduction. An interesting outcome of the present study is the observation that the trajectories starting from a dimeric complex (Cu−[Aβ(1−16)]2) fall with a larger extent into the Fenton-type reactive class C compared to trajectories from monomeric Cu−Aβ(1−16) that ended up more easily in the nonreactive class A. An explanation might be that in the monomeric Cu−Aβ(1−16) all ligands come from the same peptide, resulting in a strained structure; hence, there is a tendency that some of the ligands move away from Cu. In the dimeric complex (Cu−[Aβ(1−16)]2) no such strain is present. This is a new mechanism that could explain the experimentally observed higher ROS production by Cu−Aβ oligomers and hence their higher neurotoxicity compared to monomers.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work has been done within the PRACE (partnership for advanced computing in Europe) Project 063 (fourth call, 2012) “Identification of redox competent species in biological processes” and within the bilateral project CNR (I)/ CNRS (F) “Identification of active redox pathways for copper ions bonded to amyloid peptides”, 2012-2013. The computational resources offered by the HPC infrastructures CALMIP (Toulouse, France) and CINECA (Bologna, Italy) are also greatly acknowledged. G.L.P. thanks P. Giannozzi (University of Udine, Italy) and C. Cavazzoni (CINECA, Italy) for help in the efficient usage of PRACE resources. The authors thank Dr. Adam Day for the language check in the manuscript.

■

REFERENCES

(1) Lovell, M. A.; Robertson, J. D.; Teesdale, W. J.; Campbell, J. L.; Markesbery, W. R. Copper, Iron and Zinc in Alzheimer’s Disease Senile Plaques. J. Neurol. Sci. 1998, 158, 47−52. (2) Dong, J.; Atwood, C. S.; Anderson, V. E.; Siedlak, S. L.; Smith, M. A.; Perry, G.; Carey, P. R. Metal Binding and Oxidation of Amyloid-β within Isolated Senile Plaque Cores: Raman Microscopic Evidence. Biochemistry 2003, 42, 2768−2773. (3) Hung, Y.; Bush, A.; Cherny, R. Copper in the Brain and Alzheimer’s Disease. J. Biol. Inorg. Chem. 2010, 15, 61−76. (4) Smith, D. G.; Cappai, R.; Barnham, K. J. The Redox Chemistry of the Alzheimer’s Disease Amyloid β Peptide. Biochem. Biophys. Acta 2007, 1768, 1976−1990. (5) Hureau, C.; Balland, V.; Coppel, Y.; Solari, P. L.; Fonda, E.; Faller, P. Importance of Dynamical Processes in the Coordination Chemistry and Redox Conversion of Copper Amyloid-β Complexes. J. Biol. Inorg. Chem. 2009, 14, 995−1000. (6) Chassaing, S.; Collin, F.; Dorlet, P.; Gout, J.; Hureau, C.; Faller, P. Copper and Heme-mediated Aβ Toxicity: Redox Chemistry, Aβ Oxidations and Anti-ROS Compounds. Curr. Top. Med. Chem. 2012, 12, 2573−2595. (7) Huang, X.; Atwood, C. S.; Hartshorn, M. A.; Multhaup, G.; Goldstein, L. E.; Scarpa, R. C.; Cuajungco, M. P.; Gray, D. N.; Lim, J.; Moir, R. D.; et al. The Aβ Peptide of Alzheimer’s Disease Directly Produces Hydrogen Peroxide Through Metal Ion Reduction. Biochemistry 1999, 38, 7609−7616. (8) Huang, X.; Cuajungco, M. P.; Atwood, C. S.; Hartshorn, M. A.; Tyndall, J. D.; Hanson, G. R.; Stokes, K. C.; Leopold, M.; Multhaup, G.; Goldstein, L. E.; et al. Cu(II) Potentiation of Alzheimer Aβ Neurotoxicity. Correlation with Cell-Free Hydrogen Peroxide Production and Metal Reduction. J. Biol. Chem. 1999, 274, 37111− 37116. (9) da Silva, G. F. Z.; Ming, L.-J. Alzheimer’s Disease Related Copper(II)- β-Amyloid Peptide Exhibits Phenol Monooxygenase and Catechol Oxidase Activities. Angew. Chem., Int. Ed. 2005, 44, 5501− 5504. (10) Guilloreau, L.; Combalbert, S.; Sournia-Saquet, A.; Mazarguil, H.; Faller, P. Redox Chemistry of Copper-Amyloid-β: The Generation of Hydroxyl Radical in the Presence of Ascorbate Is Linked to RedoxPotentials and Aggregation State. ChemBioChem 2007, 8, 1317−1325. (11) Baruch-Suchodolsky, R.; Fischer, B. Soluble Amyloid β 1−28 Copper(I)/Copper(II)/Iron(II) Complexes Are Potent Antioxidants in Cell-Free Systems. Biochemistry 2008, 47, 7796−7806.

ASSOCIATED CONTENT

S Supporting Information *

Detailed description of the 16 trajectories and details for reproducing the data here reported, including the references to the codes used and publicly available on the basis of open16465



Article

Environment in the Amyloid-β Peptide Near Physiological pH. Angew. Chem., Int. Ed. 2009, 48, 9522−9525. (30) Faller, P. Copper and Zinc Binding to Amyloid-β: Coordination, Dynamics, Aggregation, Reactivity and Metal-Ion Transfer. ChemBioChem 2009, 10, 2837−2845. (31) Shin, B. K.; Saxena, S. Substantial Contribution of the Two Imidazole Rings of the His13−His14 Dyad to Cu(II) Binding in Amyloid-β(1−16) at Physiological pH and Its Significance. J. Phys. Chem. A 2011, 115, 9590−9602. (32) Balland, V.; Hureau, C.; Savéant, J.-M. Electrochemical and Homogeneous Electron Transfers to the Alzheimer Amyloid-β Copper Complex Follow a Preorganization Mechanism. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 17113−17118. (33) Car, R.; Parrinello, M. Unified Approach for Molecular Dynamics and Density-Functional Theory. Phys. Rev. Lett. 1985, 55, 2471−2474. (34) Giannozzi, P.; De Angelis, F.; Car, R. First-Princple Molecular Dynamics with Ultrasoft Pseudopotentials: Parallel Implementation and Application to Extended Bioinorganic Systems. J. Chem. Phys. 2004, 120, 5903−5915. (35) Sulpizi, M.; Raugei, S.; VandeVondele, J.; Carloni, P.; Sprik, M. Calculation of redox properties: Understanding short- and long-range effects in rubredoxin. J. Phys. Chem. B 2007, 111, 3969−3976. (36) Furlan, S.; La Penna, G.; Guerrieri, F.; Morante, S.; Rossi, G. Ab Initio Simulations of Cu Binding Sites On the N-Terminal Region of the Prion Protein. J. Biol. Inorg. Chem. 2007, 12, 571−583. (37) Guerrieri, F.; Minicozzi, V.; Morante, S.; Rossi, G.; Furlan, S.; La Penna, G. Modeling the Interplay of Glycine Protonation and Multiple Histidine Binding of Copper in the Prion Protein Octarepeat Subdomains. J. Biol. Inorg. Chem. 2009, 14, 361−374. (38) Furlan, S.; Hureau, C.; Faller, P.; La Penna, G. Modeling the Cu+ Binding in the 1−16 Region of the Amyloid-β Peptide Involved in Alzheimer’s Disease. J. Phys. Chem. B 2010, 114, 15119−15133. (39) Furlan, S.; Hureau, C.; Faller, P.; La Penna, G. Modeling Copper Binding to the Amyloid‑β Peptide at Different pH: Toward a Molecular Mechanism for Cu Reduction. J. Phys. Chem. B 2012, 116, 11899−11910. (40) Furlan, S.; La Penna, G. Modeling of the Zn2+ Binding in the 1− 16 Region of the Amyloid β Peptide Involved in Alzheimer’s Disease. Phys. Chem. Chem. Phys. 2009, 11, 6468−6481. (41) Giannozzi, P.; Jansen, K.; La Penna, G.; Minicozzi, V.; Morante, S.; Rossi, G. C.; Francesco, S. Zn Induced Structural Aggregation Patterns of β-Amyloid Peptides by First-Principle Simulations and XAS Measurements. Metallomics 2012, 4, 156−165. (42) Pang, Y.-P.; Xu, K.; El Yazal, J.; Prendergast, F. G. Successful Molecular Dynamics Simulation of the Zinc-Bound Farnesyltransferase Using the Cationic Dummy Atom Approach. Protein Sci. 2000, 9, 1857−1865. (43) Giannozzi, P.; Baroni, S.; Bonini, N.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Chiarotti, G. L.; Cococcioni, M.; Dabo, I.; et al. QUANTUM ESPRESSO: A Modular and Open-Source Software Project for Quantum Simulations of Materials. J. Phys.: Condens. Matter 2009, 21, 395502. (44) Bonomi, M.; Branduardi, D.; Bussi, G.; Camilloni, C.; Provasi, D.; Raiteri, P.; Donadio, D.; Marinelli, F.; Pietrucci, F.; Broglia, R. A.; et al. PLUMED: A Portable Plugin for Free-Energy Calculations with Molecular Dynamics. J. Comput. Phys. 2009, 180, 1961−1972. (45) Andreussi, O.; Dabo, I.; Marzari, N. Revised Self-Consistent Continuum Solvation in Electronic-Structure Calculations. J. Chem. Phys. 2012, 136, 064102. (46) Cassagnes, L.-E.; Hervé, V.; Nepveu, F.; Hureau, C.; Faller, P.; Collin, F. The Catalytically Active Copper-Amyloid-β State: Coordination Site Responsible for Reactive Oxygen Species Production. Angew. Chem., Int. Ed. 2013, 52, 11110−11113. (47) Alies, B.; Badei, B.; Faller, P.; Hureau, C. Reevaluation of Copper(I) Affinity for Amyloid-β Peptides by Competition with Ferrozine: An Unusual Copper(I) Indicator. Chem.Eur. J. 2012, 18, 1161−1167.

(12) Opazo, C.; Huang, X.; Cherny, R. A.; Moir, R. D.; Roher, A. E.; White, A. R.; Cappai, R.; Masters, C. L.; Tanzi, R. E.; Inestrosa, N. C.; Bush, A. I. Metalloenzyme-like Activity of Alzheimer’s Disease βAmyloid. Cu-Dependent Catalytic Conversion of Dopamine, Cholesterol, and Biological Reducing Agents to Neurotoxic H2O2. J. Biol. Chem. 2002, 277, 40302−40308. (13) Nadal, R. C.; Rigby, S. E.; Viles, J. H. Amyloid β-Cu2+ Complexes in Both Monomeric and Fibrillar Forms Do Not Generate H2O2 Catalytically but Quench Hydroxyl Radicals. Biochemistry 2008, 47, 11653−11664. (14) Rowley, D. A.; Halliwell, B. Superoxide-Dependent and Ascorbate-Dependent Formation of Hydroxyl Radicals in the Presence of Copper Salts: A Physiologically Significant Reaction? Arch. Biochem. Biophys. 1983, 225, 279−284. (15) Halliwell, B. Oxidative Stress and Neurodegeneration: Where Are We Now? J. Neurochem. 2006, 97, 1634−1658. (16) Hureau, C.; Eury, H.; Guillot, R.; Bijani, C.; Sayen, S.; Solari, P.L.; Guillon, E.; Faller, P.; Dorlet, P. X-Ray and Solution Structures of Cu(II)GHK and Cu(II)DAHK Complexes: Influence on Their Redox Properties. Chem.Eur. J. 2011, 17, 10151−10160. (17) Petkova, A. T.; Ishii, Y.; Balbach, J. J.; Antzutkin, O. N.; Leapman, R. D.; Delaglio, F.; Tycko, R. A Structural Model for Alzheimer’s β-Amyloid Fibrils Based On Experimental Constraints From Solid State NMR. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 16742− 16747. (18) Luhrs, T.; Ritter, C.; Adrian, M.; Riek-Loher, D.; Bohrmann, B.; H.Dobeli, H.; Schubert, D.; Riek, R. 3D Structure of Alzheimer’s Amyloid-β(1−42) Fibrils. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 17342. (19) Fändrich, M. Oligomeric Intermediates in Amyloid Formation: Structure Determination and Mechanisms of Toxicity. J. Mol. Biol. 2012, 421, 427−440. (20) Parthasarathy, S.; Long, F.; Miller, Y.; Xiao, Y.; McElheny, D.; Thurber, K.; Ma, B.; Nussinov, R.; Ishii, Y. Molecular-Level Examination of Cu2+ Binding Structure for Amyloid Fibrils of 40Residue Alzheimer’s β by Solid-State NMR Spectroscopy. J. Am. Chem. Soc. 2011, 133, 3390−3400. (21) Stroud, J. C.; Liu, C.; Teng, P. K.; Eisenberg, D. Toxic Fibrillar Oligomers of Amyloid-β Have Cross-β4 Structure. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 7717−7722. (22) Hane, F.; Tran, G.; Attwood, S. J.; Leonenko, Z. Cu2+ Affects Amyloid-β(1−42) Aggregation by Increasing Peptide-Peptide Binding Forces. PLOS ONE 2013, 8, e59005. (23) Fang, C. L.; Wu, W. H.; Liu, Q.; Sun, X.; Ma, Y.; Zhao, Y. F.; Li, Y. M. Dual Functions of β-Amyloid Oligomer and Fibril in Cu(II)Induced H2O2 Production. Regul. Pept. 2010, 163, 1−6. (24) Shearer, J.; Szalai, V. A. The Amyloid-β Peptide of Alzheimer’s Disease Binds CuI in a Linear Bis-His Coordination Environment: Insight Into a Possible Neuroprotective Mechanism for the Amyloid-β Peptide. J. Am. Chem. Soc. 2008, 130, 17826−17835. (25) Himes, R. A.; Park, G. Y.; Siluvai, G. S.; Blackburn, N. J.; Karlin, K. D. Structural Studies of Copper(I) Complexes of Amyloid-β Peptide Fragments: Formation of Two-Coordinate Bis(Histidine) Complexes. Angew. Chem., Int. Ed. 2008, 47, 9084−9087. (26) Hureau, C.; Dorlet, P. Coordination of Redox Active Metal Ions to the Amyloid Precursor Protein and to Amyloid-β Peptides Involved in Alzheimer Disease. Part 2: Dependence of Cu(II) Binding Sites with Aβ Sequences. Coord. Chem. Rev. 2012, 256, 2175−2187. (27) Drew, S. C.; Barnham, K. J. The Heterogeneous Nature of Cu2+ Interactions with Alzheimer’s Amyloid-β Peptide. Acc. Chem. Res. 2011, 44, 1146−1155. (28) Kim, D.; Kim, N. H.; Kim, S. H. 34 GHz Pulsed ENDOR Characterization of the Copper Coordination of an Amyloid β-Peptide Relevant to Alzheimer’s Disease. Angew. Chem., Int. Ed. 2013, 52, 1139−1142. (29) Hureau, C.; Coppel, Y.; Dorlet, P.; Solari, P. L.; Sayen, S.; Guillon, E.; Sabater, L.; Faller, P. Deprotonation of the Asp1-Ala2 Peptide Bond Induces Modification of the Dynamic Copper(II) 16466



Article

(48) Dokmanić, I.; Šikić, M.; Tomić, S. Metals in Proteins: Correlation Between the Metal-Ion Type, Coordination Number and the Amino-Acid Residues Involved in the Coordination. Acta Crystallogr., Sect. D 2008, 64, 257−263. (49) Azimi, S.; Rauk, A. On the Involvement of Copper Binding to the N-Terminus of the Amyloid-β Peptide of Alzheimer’s Disease: A Computational Study On Model Systems. Int. J. Alzheimer’s Dis. 2011, 2011, 1−15. (50) Noël, S.; Perez, F.; Pedersen, J. T.; Alies, B.; Ladeira, S.; Sayen, S.; Guillon, E.; Gras, E.; Hureau, C. A New Water-Soluble Cu(II) Chelator That Retrieves Cu From Cu(Amyloid-β) Species, Stops Associated ROS Production and Prevents Cu(II)-Induced Aβ Aggregation. J. Inorg. Biochem. 2012, 117, 322−325. (51) Jensen, M.; Canning, A.; Chiha, S.; Bouquerel, P.; Pedersen, J. T.; Ostergaard, J.; Cuvillier, O.; Sasaki, I.; Hureau, C.; Faller, P. Inhibition of Cu-Amyloid-β by Using Bifunctional Peptides with βSheet Breaker and Chelator Moieties. Chem.Eur. J. 2012, 18, 4836− 4839. (52) Humphrey, W.; Dalke, A.; Schulten, K. VMD - Visual Molecular Dynamics. J. Molec. Graphics 1996, 14, 33−38.

16467


Identifying, By First-Principles Simulations, Cu[Amyloid-Î²] Species

Recommend Documents