NMR Solution Structure of Rat Aβ(1–16): Toward Understanding the Mechanism of Rats' Resistance to Alzheimer's Disease

Abstract

In an attempt to reveal the mechanism of rats' resistance to Alzheimer's disease, we determined the structure of the metal-binding domain 1–16 of rat β-amyloid (rat Aβ(1–16)) in solution in the absence and presence of zinc ions. A zinc-induced dimerization of the domain was detected. The zinc coordination site was found to involve residues His-6 and His-14 of both peptide chains. We used experimental restraints obtained from analyses of NMR and isothermal titration calorimetry data to perform structure calculations. The calculations employed an explicit water environment and a simulated annealing molecular-dynamics protocol followed by quantum-mechanical/molecular-mechanical optimization. We found that the C-tails of the two polypeptide chains of the rat Aβ(1–16) dimer are oriented in opposite directions to each other, which hinders the assembly of rat Aβ dimers into oligomeric aggregates. Thus, the differences in the structure of zinc-binding sites of human and rat Aβ(1–16), their ability to form regular cross-monomer bonds, and the orientation of their hydrophobic C-tails could be responsible for the resistance of rats to Alzheimer's disease.

Introduction

According to the widely accepted amyloid hypothesis (1), the key pathogenic process of Alzheimer's disease (AD) is the transition of soluble β-amyloid (Aβ) monomer into neurotoxic oligomers and then into insoluble fibrillar polymeric aggregates, which form the amyloid plaques. The details of the molecular mechanism of this aggregation remain unknown; however, it was reported that zinc and copper ions may play a critical role in this pathological transition (2, 3). Human Aβ interacts with zinc ions through its metal-binding domain 1–16 (4, 5, 6, 7, 8), and it has been shown that the stoichiometry of Zn²⁺ binding to the soluble forms of human Aβ is 1:1. Three histidine residues (His-6, His-13, and His-14) participate in the coordination of the zinc ion (4, 5, 6, 7, 9, 10, 11, 12, 13, 14, 15). The fourth residue that fills up the Zn²⁺ coordination sphere was reported to be either Asp-1 (9, 10, 11) or Glu-11 (5, 10) depending on the pH and solvent. The minimal zinc-binding site of Aβ in water was found to consist of residues 6–14 (7). It was also shown that region 11–14 of the human Aβ mediates zinc-induced conglutination of human Aβ (16), which, as expected, leads to further formation of stable Aβ aggregates with ordered monomer chain packing (17, 18, 19).

As opposed to other mammals, rats and mice are invulnerable to AD (20, 21). The key factors in such resistance could be the three amino acid substitutions (R5G, Y10F, and H13R) in the metal-binding domain (4, 22), which are the only differences between human and rodent Aβ sequences (Fig. 1). It was found that the coordination of Zn²⁺ by His-13 is critical for zinc-induced aggregation of human Aβ (23). NMR solution studies of rat Aβ(1–28) in dimethyl sulfoxide (DMSO)-d₆ (24) and in sodium dodecyl sulfate (SDS) micelles (10) showed that the affinity of Zn²⁺ for rat Aβ(1–28) is lower than that for the analogous human peptide. His-6 and His-14 were identified as the primary zinc chelators in both media. Additionally, Arg-13 was suggested to bind zinc ion in DMSO-d₆ (24). In contrast, the terminal amino group of Asp-1 and the carboxyl group of Glu-11 of rat Aβ(1–28) were found to bind the zinc ion when the peptide was solubilized in water/SDS micelles (10). In this case the authors found that rats and humans Aβ(1–28) exhibited very similar effects upon interaction with Zn²⁺. However, in contrast to the results obtained with human Aβ(1–16) (5), the spatial structures of the rat Aβ species containing the metal-binding domain were determined in nonphysiological conditions (organic solvent or SDS micelles), which may compromise accurate comparison of these two species.

Figure 1.

Sequences of human and rat Aβ. The metal-binding domain 1–16 is shown by a rectangle. Differences in the amino acid sequences are marked.

To investigate possible molecular mechanisms in rodents' resistance to pathogenic Aβ aggregation, we obtained the structure of the complex of rat Aβ metal-binding domain in the presence of Zn²⁺ ions in aqueous solution and at a physiological pH. To determine the amino acids that participate in zinc ion coordination, we used two independent methods: NMR and isothermal titration calorimetry (ITC).

Materials and Methods

All chemicals and solvents were of high-performance liquid chromatography grade or better and were obtained from Sigma-Aldrich (St. Louis, MO). All synthetic peptides (purity > 98% checked by reverse-phase high-performance liquid chromatography) were purchased from Biopeptide, LLC (San Diego, CA). We confirmed the amino acid sequence of each peptide on an ultrahigh-resolution Fourier transform ion cyclotron resonance mass spectrometer (7T Apex Qe; Bruker Daltonics, Billerica, MA) using a de novo sequencing approach based on the collision-induced dissociation fragmentation technique.

Isothermal titration calorimetry

ITC measurements were carried out on an iTC₂₀₀ instrument (MicroCal, Northampton, MA) at 25°C in 50 mM Tris buffer, pH 7.3. Aliquots of ZnCl₂ solution (2 μl) were injected into the 0.2 mL cell containing the peptide solution to achieve a complete binding isotherm (Fig. 2 and Fig. S1 of the Supporting Material). The peptide concentration in the cell ranged from 0.1 to 1 mM, and the ZnCl₂ concentration in the syringe ranged from 5 to 15 mM. The heat of dilution was measured by injecting the ligand (ZnCl₂) into the buffer solution or by additional injections of the ligand after saturation; the values obtained were subtracted from the heat of the reaction to obtain the effective heat of binding. We fitted the resulting titration curves using MicroCal Origin software (Fig. S1), and determined the affinity constants (K_a), binding stoichiometry (N), and enthalpy (ΔH) using a nonlinear regression fitting procedure. The ITC measurements for each peptide were repeated at least three times at different peptide concentrations and yielded similar thermodynamic parameters.

Figure 2.

ITC binding isotherms of zinc interactions with rat Aβ(1–16) (a) and its mutants D1A (b), D7A (c), E3A (d), E11A (e), H14A (f), and H6A (g) at 25°C in 50 mM Tris buffer, pH 7.3.

NMR experiments

A 5 mM solution of rat Aβ(1–16) in a 20 mM Tris-d₁₁ buffer solution (pH 7.1) was used for NMR measurements. We added 0.1% NaN₃ to the samples to prevent peptide biodegradation. We prepared samples of rat Aβ(1–16) in the presence of zinc ions by adding, in several steps, a concentrated stock solution of the highest analytical grade ZnCl₂ (Aldrich) to the peptide to achieve a Zn²⁺/peptide ratio of ∼1.0–1.5. Changes in the chemical shifts of the peptide signals after stepwise addition of ZnCl₂ were monitored. The absence of pH variation upon zinc addition was confirmed after the final addition. NMR spectra were measured at 278 K either in D₂O or in 90% H₂O/10% D₂O on a Bruker AVANCE spectrometer (Bruker, Karlsruhe, Germany) operating at 600 MHz ¹H frequency, equipped with a triple-resonance (¹H, ¹³C, and ¹⁵N), pulsed-field, z-gradient probe. Two-dimensional (2D) NMR spectra were processed by NMRPipe (25) and analyzed with the use of SPARKY (26).

NMR signal assignment

We obtained the ¹H, ¹⁵N, and ¹³C signal assignments of the rat Aβ(1–16) and its complex with zinc using the following 2D spectra: DQF-COSY, TOCSY (mixing time of 70 ms), NOESY (mixing time of 200 and 500 ms), ¹³C-¹H HSQC, and ¹⁵N-¹H HSQC. Heteronuclear experiments were measured at the natural abundance of the ¹³C and ¹⁵N isotopes (Fig. S2). For the signal assignments we used the classical approach of Wuthrich (27) (see Fig. S3 and Fig. S4) supplemented by analysis of the HSQC spectra. Signal assignments of the free rat Aβ(1–16) and its complex with zinc are presented in Table S1 and Table S2, respectively.

Restrained molecular dynamics

We determined the structures of the rat Aβ(1–16) and its complex with zinc ions using a restrained molecular-dynamics (MD) approach with sets of NOE restraints. Several dihedral angle restraints for the peptide backbone were determined for the complex of the rat Aβ(1–16) with Zn²⁺ with the use of AngleSearch software (28). For the zinc-peptide complex, distance and torsion angle restraints representing the coordination site of the zinc ion were also used in MD calculations. These parameters were obtained by quantum-mechanical (QM) geometry optimization (see below). We identified the His nitrogen atoms that participate in the chelation of the zinc ion (Nδ1 for His-6 and Nε2 for His-14) by analyzing possible orientations of the His imidazole ring for each type of chelating center, and comparing the interatomic distances with the observed intensities of the NOEs. Thus, the stronger intraresidue NOEs of Hδ2–HN and Hδ2–Hα for His-14, and the equally strong NOEs for Hδ2–Hβs in both His-6 and His-14 residues (Fig. 3) unambiguously confirm the selection of the chelating atoms.

Figure 3.

Fragments of the 2D NOESY spectrum (200 ms mixing time) of the complex of rat Aβ(1–16) with zinc ions, showing several intraresidue NOEs involving the Hδ2 signals of His-6 and His-14.

Determining the structure of small, charged, and mobile peptides is a relatively difficult task due to the lack of experimentally determined restraints and their conformational averaging. To obtain reliable structural data, we carried out MD simulations in explicit water with appropriate calculations of electrostatic interactions between charged groups of the peptide. The extended chain conformation of the peptide was used as the initial structure in MD calculations. A peptide molecule was placed into a cubic cell of 0.5 × 0.5 × 0.5 nm, and 13,710 water molecules of the SPC/E model were added to the system (29). Because the total charge of the system is zero, neutralization of the molecular system was not required. We minimized the potential energy in two stages, first using the steepest-descent algorithm and subsequently the conjugated gradient algorithm. We used GROMACS 3.3.1 software (30) and AMBER 03 force-field (31) for all calculations. The system obtained after 20 ps of MD trajectory with restricted positions of all polypeptide atoms was used as the initial structure in subsequent structure calculations. A Berendsen thermostat and a barostat with a rescale velocity function were used in the MD calculations. The lengths of covalent bonds during MD were controlled by the LINCS algorithm (32). The energy of electrostatic interactions was calculated using the Ewald particles grid algorithm. The maximum distances of the Coulomb and van der Waals forces were 1.0 and 1.4 nm, respectively. The integration step for MD calculations was 2 × 10⁻¹⁵ s. Calculations of the rat Aβ(1–16) structures were performed according to a previously described protocol (33). The total length of the MD trajectory was 142 ps, and it took ∼16 min to calculate a family of 20 structures using 320 processors. We determined the convergence of the calculations using the root mean-square deviation (RMSD) of the coordinates of the C, Cα, and N atoms of protein chain structures, calculated for the whole family. We defined the quality of the calculated structures on the basis of the percentage of hits of the main dihedral angles φ and ψ in the most favorable and prohibited areas of a Ramachandran map using Procheck_NMR (34). Structures were visualized (see Fig. 5 and Fig. S5) and analyzed with the use of InsightII software (Accelrys).

Figure 5.

Structure of the rat Aβ(1–16) dimer complexed with Zn ions in solution (the 20 NMR conformers have been deposited in the Protein Data Bank with accession code 2LI9). Only the backbone atoms (Cα, C, and N) and the side chains of the His residues are shown. Chains A and B of the dimer are shown in red and blue, respectively. The N- and C-termini of both chains are labeled. (A) The family of 20 calculated NMR structures. (B) Coordination of the zinc ion by the histidine residues in the representative structure after additional QM/MM geometry optimization. The average distance between the Zn²⁺ ion and the nitrogen atoms of the His residues (Nδ1 and Nε2) is 2.07 ± 0.05 Å.

Quantum-mechanical/molecular-mechanical calculations

The binding of metal ions to a protein can cause strong polarization and charge transfer effects and coordination geometries that are not easily described within standard force fields. In many cases, an adequate simulation can only be performed with an explicit QM electronic-structure calculation. In addition, the problem of adequate definition of the metal-binding sites in proteins requires the use of an approach that takes into account the entire protein environment and incorporates finite temperature effects that are known to be crucial for biological function (35). To fulfill all of these conditions, we applied the QM/molecular-mechanical (MM) Car-Parrinello simulation approach (36, 37), which was previously shown to be successful in simulations of a number of metal-containing systems (38, 39, 40). Furthermore, recent results from Zn²⁺-Aβ complexes obtained by the QM/MM method showed good agreement with NMR data (16, 41, 42).

A representative structure of the rat Aβ(1–16) peptide complexed to a zinc ion was used as the starting conformation for QM/MM calculations. We applied the MM approach using the parameters from the parm99 force field with corrections by Duan et al. (43). To describe the QM system, we employed the density functional theory in the generalized gradient approximation using the PW91 functional. The side length of a cubic cell for allocation of the quantum subsystem was set to 45 Å. The cutoff radius for the MM calculations was set to 15 Å, and van der Waals interactions, which are poorly described by default density functional theory, were corrected with the Grimme analytical potential (44, 45). Each simulation system was filled with water molecules presented by the TIP4P model. We equilibrated water molecules around the peptide-ion complex by carrying out a 100 ps MD simulation with restrained positions of the peptide and the zinc ion. The prepared systems were subjected to QM/MM simulations with the use of the GROMACS/CPMD package (36). The integration time step was 0.12 fs (∼5 a.u.). The radius of calculation of close-range, charge-to-charge interactions was set to 10 Å. The list of neighboring atoms was renewed at each step. Long-range charge interactions were calculated with the PME algorithm (46). Temperature coupling with the Nose-Hover (47, 48) scheme allowed the determination of the systems at body temperature (300 K). The interactions between valence electrons and ionic cores were described by an ultrasoft VDB pseudopotential (49, 50). Because we applied ultrasoft pseudopotentials, the basis set for the valence electrons consisted of plane waves expanded up to a cutoff of 30 Å.

Results and Discussions

The NMR spectra of free rat Aβ(1–16) contain sharp signals typical of small monomeric peptides (Fig. 4 A), but they become broader upon addition of Zn²⁺ (Fig. 4 B). The observed changes in the spectra indicate the formation of a complex of rat Aβ(1–16) with Zn²⁺ ions. The nature of the zinc-induced line-broadening for various soluble forms of Aβ was previously examined in detail (5, 10, 12, 51). It was shown that this phenomenon originates mainly from the exchange between different conformations of the Aβ/Zn²⁺ complex. Increasing the temperature makes the exchange rates faster, and the observed NMR lines in the rat Aβ(1–16)/Zn²⁺ complex become sharper and similar to those described for the rat Aβ(1-28)/Zn²⁺ complex (10). However, the temperature increase leads to a decrease in the lifetime of each conformational state, an increase in the exchange rates of amide protons with water, and an increase in the amplitudes of the molecular motions. As previously noted (10), none of these temperature-dependent processes are favorable for obtaining structural information. Therefore, we chose the optimal temperature range by analogy with an earlier study on the human Aβ(1–16)/Zn²⁺ complex (5) in which all NMR measurements were carried out at 278 K.

Figure 4.

Low-field region of the 600 MHz ¹H NMR spectra of the rat Aβ(1–16) in its free state (A) and after addition of a twofold molar excess of ZnCl₂ (B). Spectra were collected at 10°C in Tris-d₁₁, 90 H₂O/10% D₂O, pH 7.1.

Sets of 2D homonuclear (NOESY, ROESY, TOCSY, and DQF-COSY) and heteronuclear (¹³C-¹H HSQC, HSQC-TOCSY, HMBC, and ¹⁵N-¹H HSQC) NMR spectra were collected for the free Aβ(1–16) and its complex with Zn²⁺ at the natural abundance of the ¹³C and ¹⁵N isotopes. Resonance assignments for nearly all ¹H and ¹³C nuclei were determined for rat Aβ(1–16) in both the free state (Table S1) and in its complex with zinc (Table S2). For free Aβ(1–16), we were able to obtain ¹⁵N resonance assignments for the backbone and side-chain amide groups, but we could not obtain such information for the complex of rat Aβ(1–16) with Zn²⁺ at the natural abundance of ¹⁵N due to substantial zinc-induced signal broadening.

An analysis of the changes in chemical shifts of rat Aβ(1–16) observed upon zinc binding (Fig. S7) highlights two His residues (His-6 and His-14) that participate in the metal ion coordination. Significant broadening of the resonances of Hε1 of these His residues upon addition of Zn²⁺ to the peptide (Fig. S8) also confirms this conclusion. However, due to the high mobility of the rat Aβ(1–16) in solution, zinc-induced changes of the chemical shifts are relatively small and could not be used for unambiguous identification of other residues that form the metal-binding site. ITC studies on a set of rat Aβ(1–16) alanine mutants were used to determine the amino acids that take part in zinc coordination (Fig. 2 and Fig. S1). ITC also provided information about the stoichiometry and energy of zinc binding (Table 1). Zinc binding to Aβ(1–16) and its alanine mutants is both enthalpy-driven and entropically favorable. The stoichiometry of Zn²⁺ binding to all mutants except E3A is ∼1:2 (Table 1), indicating that a peptide dimer containing one Zn²⁺ ion is formed. At the same time, the stoichiometry of Zn²⁺ binding to E3A is close to one, suggesting that substitution of Glu-3 by Ala provides an opportunity to gain favorable hydrophobic contacts within a monomer upon Zn²⁺ binding. This is also in good agreement with the highest favorable entropy of zinc binding for this mutant in comparison with the other mutants. It can be seen that substitution of Asp-1, Glu-3, and Asp-7 to Ala does not substantially affect Zn²⁺ affinity for Aβ(1–16), indicating that these residues do not participate in metal coordination. In contrast, during titration of the H6A and H14A mutants with Zn²⁺ ions, no heat exchange was detected, signifying that these His residues chelate Zn²⁺. Thus, the ITC studies unambiguously show that residues His-6 and His-14 coordinate the zinc ion, and that rat Aβ(1–16) dimerizes upon metal chelation. This information was used for the calculations of the structure of the rat Aβ(1–16) complex with zinc.

Table 1.

Thermodynamic parameters of zinc ions binding to rat Aβ(1–16) and its mutants obtained by ITC at 25°C in 50 mM Tris buffer, pH 7.3

Peptide	N	Ka × 104 (M−1)	ΔH (kcal M−1)	TΔS (kcal M−1)
Wt	0.60	1.53	−4.8	0.9
D1A	0.70	0.98	−3.5	1.9
E3A	1.20	0.71	−1.8	3.6
H6A	No binding
D7A	0.70	1.01	−2.9	2.6
E11A	1.00	1.44	−1.4	4.3
H14A	No binding

Standard deviations of binding stoichiometry (N) and enthalpy (ΔH) measurements do not exceed 10%, and those of the affinity constant (K_a) measurements do not exceed 20%.

Sets of distance and torsion angle restraints (Table 2), obtained from an analysis of NMR spectra recorded at 278 K, were used for the structure calculations of rat Aβ(1–16) and its complex with zinc. It should be noted that no long-range NOEs exist in the NOESY spectra of rat Aβ(1–16) and its complex with Zn²⁺, which is typical for small, flexible peptides.

Table 2.

Numbers of NMR restraints used for the structure calculations and the statistics of the calculated families of rat Aβ(1–16) structures

	Rat Aβ(1–16)	Dimer of rat Aβ(1–16)/Zn2+
Number of NOE restraints	111	120
Intraresidue	55	72
Sequential	45	33
Medium-range	11	15
Long-range (|i-j|>4)	0	0
Number of dihedral angle restraints	0	3
Number of Zn-chelating restraints	0	8
Ramachandran map statistic
Residues in most favored regions (%)	79.8	68.1
Residues in disallowed regions (%)	0.0	0.0
Number of conformers in a family of structures	40	20
Number of NOE violations (>0.5 Å) per structure	2	0
RMSD of coordinates of atoms C′, Cα, and N for superposition of the family of structures (Å)	2.31 ± 0.55	1.46 ± 0.48

We calculated an NMR structure of the free peptide using restrained MD and a protocol optimized to determine the structure of flexible peptides (33). An explicit water environment was used for more-accurate calculations of the electrostatic interactions between the charged peptide groups. The family of NMR structures of the free rat Aβ(1–16) is relatively loose (the RMSD of coordinates of heavy backbone atoms (C′, Cα, N, and O) is 2.31 ± 0.55 Å; Fig. S5 A). When the two terminal parts of the peptide are superimposed, the RMSD becomes smaller (1.98 ± 0.36 Å for the N-terminal (1, 2, 3, 4, 5, 6) and 1.66 ± 0.36 Å for the C-terminal (7, 8, 9, 10, 11, 12, 13, 14, 15, 16)), which is compatible with the existence of two small subdomains (Fig. S6) and a more flexible linker between them.

We calculated the NMR structures of the rat Aβ(1–16) dimer complexed with Zn²⁺ (Fig. 5) using the NMR restraints for the free peptide and a set of distance constraints between the zinc ion and His residues (Table 2). QM/MM calculations were carried out to optimize the geometry of the Zn²⁺ coordination site. The constraints obtained for the Zn-peptide interaction were used in the restrained MD calculations. Amino acids that belong to the different peptide chains in the Aβ(1–16) dimer could not be distinguished by their chemical shifts. Therefore, two possible orientations of the peptide chains (parallel and antiparallel) were tested in the structure calculations. We found that only the parallel orientation (Fig. 5) fits the set of NMR distance restraints. Structures calculated on the assumption of an antiparallel chain orientation (Fig. S9) have a much larger number of NOE violations. Moreover, for an antiparallel dimer, one would also expect to see several well-resolved interchain NOEs, which were absent in the NOESY spectra. Fig. 5 A represents a final family of 20 calculated NMR structures. (The bundle of 20 conformers representing the NMR structure has been deposited in the Protein Data Bank with accession code 2LI9, and the assignments are in the Biological Magnetic Resonance Bank (ID 17884)). The RMSD of the coordinates of the heavy backbone atoms (Cα, C′, and N) of the whole dimer is 1.46 ± 0.48 Å, and for separate superpositions of each chain the RMSDs are 1.13 ± 0.59 Å (chain A) and 1.06 ± 0.27 Å (chain B).

A representative structure of this family was subjected to an additional QM/MM optimization step (Fig. 5 B). The final structure, fixed by the His-metal interactions, is irregular and consists of bend, coil, and turn regions with no other regular secondary structural elements such as helices and β-strands. The side chains of the hydrophobic residues Phe-4, Phe-10, and Val-12 are oriented outward from the dimerization interface and do not interact with each other, in contrast to the human Aβ(1–16), where Phe-4 and Tyr-10 form a hydrophobic contact (5). As noted above, the two peptide chains in the dimer structure of rat Aβ(1–16) are packed in a parallel manner, with their N-terminal fragments being close to each other. At the same time, the C-tails of the two polypeptide chains in the rat Aβ(1–16) dimer are oriented in opposite directions to each other (Fig. 5 A). It should also be emphasized that only the His residues participate in zinc ion coordination by the rat Aβ(1–16), despite the presence of several other potential chelating residues (Asp and Glu). This distinguishes rat Aβ(1–16) from its human analog, where His-6, Glu-11, His-13, and His-14 form the coordination sphere of the metal ion under similar experimental conditions (5). The structural distinction between the zinc complexes of the human and rat Aβ(1–16) is clearly confirmed by the differences in the distribution and number of NOEs in the corresponding NMR spectra recorded under identical conditions (Fig. S10). The ITC data also show reproducible differences in the stoichiometry of zinc binding to rat (Table 1) and human (7) Aβ(1–16).

An important feature of the behavior of the Aβ(1–16) peptides both in the free state and complexed with zinc is the relatively fast (on the NMR timescale) exchange between several conformational states. This feature was previously discussed in terms of the zinc-induced line-broadening observed in human and rat Aβ(1–28) (10). In earlier studies, investigators established the existence of the free human peptides Aβ(1-40) (52) and Aβ(1-42) (53) in water solution as an ensemble of distinct conformational species undergoing fast exchange between each other, using replica-exchange MD simulations based on NMR-derived experimental data. They found that several structured regions exist in these peptides; otherwise, the peptide chains are flexible. NMR and MD studies of the Cu²⁺-binding sites in human Aβ(1-40) fibrils indicated the coexistence of several peptide conformations and metal coordination modes even in the solid state (8). It was proposed that conformational selection and the resulting population shift could provide a primary molecular mechanism for the transformation of monomeric Aβ into organized fibril geometries (19).

The energetically favorable coordination of the zinc ion by two rat Aβ(1–16) molecules results in the formation of a stable, well-defined dimer structure (Fig. 5) that corresponds to the most populated conformational state of the molecular complex. However, the observed zinc-induced line-broadening in the complex of rat Aβ(1–16) with Zn²⁺ indicates the presence of some minor species with different structures relative to the most populated conformers. For an ensemble of conformational states, all of the NMR and ITC parameters are weighted by the population of each state. It should be noted that the measured stoichiometry of zinc binding (0.60 ± 0.06) is a little greater than the value expected for a dimer (N = 0.5). This may reflect the existence of a small population of monomeric states (N = 1) in the conformational ensemble.

According to current knowledge, fibril nucleation pathways include interactions of the metal-binding domains of neighboring Aβ molecules followed by the aggregation of hydrophobic chains Aβ(17-42) into a polymeric amyloid structure (18, 19, 54). One can assume that the opposite orientation of the two C-tails in the rat Aβ(1–16) dimer, constrained by the zinc ion, will hinder the assembly of Aβ dimers into oligomeric aggregates (Fig. 6).

Figure 6.

Structural differences between human and rat Aβ that may control the molecular mechanism of rodents' resistance to AD. Full-length Aβ with the metal-binding domains, presented as curves, are shown. The N- and C-termini of the Aβ chains are labeled.

In contrast, the conformation of the human Aβ(1–16) in its complex with zinc (5) allows its C-tail to achieve an orientation that is suitable for its oligomerization into the ordered amyloid aggregates (18). The EVHH fragment of the human Aβ (residues 11–14) forms the primary metal-binding site (7). This fragment contains three potential chelating amino acids, in contrast to its analogous fragment, EVRH, in rat Aβ, which has only two such groups. This additional group in human Aβ (His-13) could be responsible for the monomer cross-linking, as suggested by earlier studies of Aβ aggregation (13, 55). In the absence of this residue in rat Aβ(1–16), the most energetically favorable coordination of the zinc ion is attained only by dimer formation at the very first stage. This presumably prevents further nucleation of the metal-binding domains and subsequent oligomerization of the whole rat Aβ.

Conclusions

NMR and ITC studies of the interactions of the rat Aβ metal-binding domain with a zinc ion, described here and compared with earlier published results on the human Aβ domain, show clear differences between the rat and human domains in the set of residues that coordinate the zinc ion, and in the structural organization of the zinc-peptide complex under close-to-physiological conditions. Thus, the observed differences between human and rat Aβ(1–16) in the structure of zinc-binding sites, the ability to form regular cross-monomer interactions, and the orientation of the hydrophobic C-tails could be responsible for rats' resistance to AD.

Editor: Heinrich Roder.