Abstract
Aggregation of the amyloid-beta (Aβ) peptide into insoluble plaques is a major factor in Alzheimer’s disease (AD) pathology. Another major factor in AD is arguably metal ions, as metal dyshomeostasis is observed in AD patients, metal ions modulate Aβ aggregation, and AD plaques contain numerous metals including redox-active Cu and Fe ions. In vivo, Aβ is found in various cellular locations including membranes. So far, Cu(II)/Aβ interactions and ROS generation have not been investigated in a membrane environment. Here, we study Cu(II) and Zn(II) interactions with Aβ bound to SDS micelles or to engineered aggregation-inhibiting molecules (the cyclic peptide CP-2 and the ZAβ3(12–58)Y18L Affibody molecule). In all studied systems the Aβ N-terminal segment was found to be unbound, unstructured, and free to bind metal ions. In SDS micelles, Aβ was found to bind Cu(II) and Zn(II) with the same ligands and the same KD as in aqueous solution. ROS was generated in all Cu(II)/Aβ complexes. These results indicate that binding of Aβ to membranes, drugs, and other entities that do not interact with the Aβ N-terminal part, appears not to compromise the N-terminal segment’s ability to bind metal ions, nor impede the capacity of N-terminally bound Cu(II) to generate ROS.
Keywords
INTRODUCTION
Alzheimer’s disease (AD) is a progressive neurodegenerative disorder affecting hippocampal and neocortical pathways of the brain leading to memory loss, disorientation, personality changes, and eventually fatal global brain degeneration. Despite many years of research, the molecular mechanisms leading to AD remain unclear. The amyloid cascade hypothesis, stating that AD is related to self-aggregation of amyloid-beta (Aβ) peptides into extracellular plaques [1–3], is still considered valid even though the hypothesis does not in itself explain the neuronal death associated with AD [4–6]. The Aβ peptides, which are 39–43 amino acids long proteolytic cleavage products from the transmembrane amyloid-β protein precursor (AβPP), can aggregate along various pathways into insoluble amyloid fibrils [7]. The term “amyloid” structure is here used to indicate the “cross-β” structure originally identified via solid state X-ray fiber diffraction [8, 9]. In solution, certain chemical probes such as Congo Red and Thioflavin T (ThT) have been used to identify amyloid material, as these probes change their optical properties specifically when binding amyloid structures [10]. The toxicity of amyloid fibrils remains disputed [11], and neurotoxicity is sometimes attributed not to the fibrillar plaques but to intermediate oligomeric Aβ aggregates [12–14]. Also the cellular location of the toxic species, and of the aggregation event, is unclear. Although the amyloid plaques reside extracellularly, the Aβ peptides are found both inside and outside cells and also in their membranes.
The aggregation pathways of the Aβ peptides can be modulated by various factors including small molecules and metal ions [6, 15–19], where Aβ displays specific binding modes to certain metal ions including Cu(II), Fe(II), Mn(II), and Zn(II) [15, 21]. Aβ binding to Cu(II) and Zn(II) is particularly well studied, although the measurements are difficult as the binding strength not only changes with pH and ionic strength/buffer but also increases if the peptide aggregates during the course of the measurements [15]. Thus, Aβ/Cu(II) binding affinities have been reported in the micromolar to picomolar range, with a recent consensus around 100 pM that may or may not be the final verdict [6, 22–24]. Cu and Fe ions are known to interact with molecular oxygen to generate various harmful reactive oxygen species (ROS) [16, 25]. The elevated concentrations of Cu and Fe found together with Zn in Aβ amyloid plaques [26] have therefore prompted the suggestion that ROS induced by redox-active metals could induce Aβ-related cell damage [16, 25–32]. Numerous observations at population and molecular levels show correlations between neurodegenerative pathology and altered brain metal concentrations, both in AD [33–38] and other neurodegenerative disorders [39, 40], suggesting causative connections between neuronal death and metal dyshomeostasis or/and metal exposure.
Most studies on Aβ/metal-ion interactions have been performed in aqueous solutions, even though the amphiphilic Aβ peptides have membrane-binding properties and even though membrane-mediated aggregation mechanisms have been proposed to catalyze Aβ nucleation and fibrillation [41–43]. In lipid bilayers, Aβ peptides can assemble into Ca2 +-conducting nanopores with membrane-disrupting capacities [44], thereby possibly contributing to cellular toxicity. Metal ions such as Zn(II) and Al(III) appear to block the open channel activities of these Aβ nanopores in a concentration-dependent manner, indicating that certain metals can reduce the cellular toxicity of the pores [45]. In aqueous buffers, monomeric Aβ peptides are known to be unstructured and to bind metal ions at the N-terminal part, with the three histidines H6, H13, and H14 acting as specific ligands for Cu(II) and Zn(II) binding together with a fourth residue that may be A2, E11, or the N-terminal D1 [15, 47]. In membrane-mimicking environments such as sodium dodecyl sulfate (SDS) micelles, Aβ adopts a well-defined helix-loop-helix structure where residues 15–24 and 29–35 form two α-helical regions held together by an unstructured loop consisting of residues 25–28 [48–51]. Zn(II) ions have been shown to bind the N-terminal part of a truncated Aβ1–28 peptide in SDS micelles, where Aβ residues 16–24 appeared to adopt a helical structure [52].
Here, we characterize the binding of Cu(II) and Zn(II) ions to full-length Aβ1–40 peptides in SDS micelles. We also investigate Aβ40/metal-ion interactions when the Aβ peptides are bound to either of two molecules engineered to bind Aβ and prevent its aggregation, namely the cyclic peptide CP-2 [53] and the ZAβ3(12–58)Y18L Affibody molecule [54]. The focus is on binding properties, structural alterations, and Cu-induced ROS generation in thecomplexes.
MATERIALS AND METHODS
Reagents
Aβ40 peptide was purchased lyophilized from AlexoTech AB (Umeå, Sweden), either unlabeled or 15N-labeled. Fresh samples were dissolved in monomeric form before each measurement according to previously published procedures [55]. In short, the peptide was dissolved in 10 mM cold NaOH to a concentration of 1.0 mg/ml, sonicated in an ice bath, diluted to half the final sample volume with cold distilled water, sonicated again, and diluted to final concentration with the appropriate buffer. The Affibody dimer ZAβ3(12–58)Y18L and the cyclic peptide CP-2 were both synthesized in-lab according to previously published protocols [53, 56]. Copper and zinc acetate was purchased from Sigma-Aldrich Co. All experiments described below were performed under quiescent conditions, i.e., without any kind of stirring or shaking.
NMR spectroscopy
A Bruker Avance 500 MHz spectrometer equipped with a triple-resonance cryogenically cooled probehead was used to record 1D and 2D 1H,15N-HSQC NMR spectra during titrations of Cu(II) acetate and Zn(II) acetate to 115 μM 15N-labeled Aβ40 peptide in 20 mM HEPES buffer (90/10 H2O/D2O) with 50 mM SDS, 25°C. The 115 μM Aβ concentration was chosen to facilitate comparison with other similar studies on Aβ/metal-binding (e.g., [54, 56]). The Aβ concentration is typically kept low in in vitro studies to avoid aggregation during the measurements. As binding to SDS micelles and to the engineered Affibody binders appears to inhibit Aβ aggregation (see Results and Discussion below), some NMR measurements for such systems were carried out at higher Aβ concentrations. All NMR signals were referenced to TSP, and the assignment of HSQC amide crosspeaks for Aβ in SDS micelles has previously been published by our group and others [48, 57]. Chemical shift changes for the Aβ40 HSQC crosspeaks, induced by the metal ions, were calculated as the standard weighted average, i.e., Δδ= (((ΔδN/5)2+(ΔδH)2)/2)1/2 [58].
Fluorescence spectroscopy
A Jobin Yvon Horiba Fluorolog 3 (Longjumeau, France) unit was used to record fluorescence emission spectra of 10 μM Aβ40 peptide at 25°C between 290 and 350 nm (excitation wavelength 276 nm). The excitation and emission slits were set at 4 nm. A 4 mm path-length quartz cuvette was used to hold 1 ml of Aβ40 peptide sample, in 20 mM HEPES buffer at pH 7.3 with or without 50 mM SDS. Cu(II) acetate solution was titrated to the sample in small steps, thereby quenching the intrinsic fluorescence of Y10 of the Aβ peptide. The resulting fluorescence data were fitted to a binding model assuming a single binding site:
where I0 is the fluorescence intensity of the sample in the absence of Cu(II) ions, I ∞ is the fluorescence intensity of the sample when saturated with metal ions, [Aβ] is the concentration of the peptide, [Me] the concentration of the metal ion, and KD is the dissociation constant of the complex.
CD spectroscopy
Circular dichroism (CD) spectroscopy was carried out using a Chirascan CD unit (Applied Photophysics, Surrey, U.K.). A 4 mm path-length quartz cuvette with a plastic cap was used to hold 1 ml sample of 10 μM Aβ40 peptide in 20 mM sodium phosphate buffer at pH 7.3 together with 50 mM SDS at 25°C. Although most other analyses were done in HEPES buffer, which has minimal interaction with metal ions, HEPES interferes with CD measurements, hence the choice of the sodium phosphate buffer. Small amounts of Zn(II) or Cu(II) acetate was titrated to the sample, and CD spectra were recorded between 190 and 270 nm using 0.5 nm steps. Binding constants were calculated by fitting Equation 1 to the CD intensity at 208 nm as a function of added metal ions. The CD data for added Zn(II) was also fitted to the hyperbolic binding isotherm:
Hydrogen peroxide measurements
The concentration of hydrogen peroxide generated from Aβ was quantitatively determined fromreaction with tris(2-carboxyethyl)phosphine (TCEP), as described in previous studies [59, 60]. Different combinations of 10 μM Aβ40, 10 μM Zn(II) acetate, 3 μM Cu(II) acetate, 50 mM SDS, 11 μM ZAβ3(12–58)(Y18L) Affibody dimer, 100 μM CP-2 drug, 10 μM apo-SOD1 protein, and/or 100 μM EDTA were incubated at 37°C for approximately 70 min together with 50 μM TCEP and 20 mM sodium phosphate buffer, pH 7.3. Any and all H2O2 produced reacts stoichiometrically with TCEP, yielding water and oxidized TCEP. After the incubation 150 μM 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB) was added. Remaining non-oxidized TCEP reduces the disulphide bond of DTNB, thereby producing two molecules of 2-nitro-5-thiobenzoate (NTB), a yellow dye absorbing at 412 nm. UV intensities at this wavelength were measured with a Jasco V-560 UV/vis spectrophotometer, and baseline-corrected against the buffer solution. The NTB concentration was then determined using Lambert-Beers law and a molar extinction coefficient of 14,150 M–1 cm–1for the NTB anion. As each non-oxidized TCEP produces two NTB molecules, the H2O2 concentration follows as the initial TCEP concentration (50 μM) minus half the measured NTB concentration.
RESULTS
NMR spectroscopy
Figure 1 shows the amide region of the 1H,15N-HSQC 2D NMR spectrum of 115 μM 15N-labeled Aβ40 peptides positioned in SDS micelles, both in absence and presence of added Cu(II) and Zn(II) acetate. These measurements are part of a larger titration series shown in Supplementary Figures 1 and 2. The amide crosspeak intensities and chemical shifts for residues 1–20 display significant concentration-dependent changes when Cu(II) or Zn(II) ions are added, which clearly demonstrates that both metal ions selectively bind the N-terminal part of the peptide (Fig. 1 and Supplementary Figures 1 and 2). Increasing the concentration of added Cu(II) induces intensity losses not only in the N-terminal crosspeaks but also in the crosspeaks for residues G37-V40, an effect not observed when Zn(II) is added (Fig. 1 and Supplementary Figures 1 and 2). This effect arguably supports the idea that the Aβ C-terminus is located outside the SDS micelle [48], as the G37-V40 residues likely are located next to the N-terminal segment where they experience paramagnetic effects from the N-terminal-bound copper ion. The current NMR results furthermore indicate metal binding in the E22-V24 region, as the corresponding amide crosspeaks display large alterations in intensities and chemical shifts upon added Zn(II) or Cu(II) ions (Fig. 1 and Supplementary Figures 1 and 2).
The metal-induced loss of NMR signal is visible also in the 1D NMR spectra (Fig. 2). After one week of incubation under quiescent conditions at room temperature, the 1D spectrum of 280 μM Aβ with 350 μM Cu(II) acetate in 50 mM SDS remains virtually unchanged (Fig. 2). This shows that in SDS micelles, the Aβ peptides are stable over time also in presence of Cu(II) ions, which otherwise are known to efficiently promote Aβ aggregation [55].
Figure 3 shows the amide crosspeak region for 15N-labeled Aβ40 in 1 : 1 complex with an unlabeled ZAβ3(12–58)Y18L Affibody dimer, an engineered binding molecule that locks Aβ in a hairpin conformation without affecting its N-terminal part [54, 62]. The amide crosspeak assignment for Affibody-bound Aβ is known form previous work [56], and the crosspeak labeled “X” is likely one of the N-terminal histidines (Fig. 3). Addition of Zn(II) acetate induces loss of crosspeak intensity for the N-terminal residues 1–16, indicating specific Zn(II) binding to the N-terminus similar to that for free Aβ in aqueous solution.
While divalent metal ions may sometimes coordinate two or more Aβ molecules [18, 63], thereby possibly inducing Aβ aggregation/accumulation [15, 64] accompanied by loss of NMR signal, such effects are not applicable to Aβ bound to micelles or Affibody molecules, as both systems prevent aggregation by keeping Aβ monomeric (at least when the Aβ concentration is lower than the micelle concentration). Thus, the loss of NMR signal intensity upon added Zn(II) ions must be due to Zn-bound and non-Zn-bound Aβ conformations exchanging on a slow or intermediate NMR time scale [65, 66]. For Cu(II) binding, however, paramagnetic effects cannot be ruled out [54].
Fluorescence spectroscopy
To evaluate the binding strength of Cu(II) ions to the Aβ40 peptide, the intrinsic fluorescence of Aβ’s Y10 and its subsequent quenching by added Cu(II) was measured. Titrations with Cu(II) acetate were carried out both in presence and absence of SDS micelles (Fig. 4 and Supplementary Figures 3 and 4). The resulting titration curves exhibit typical patterns for a single binding site. Fitting the fluorescence data to Equation 1 yields Cu(II)/Aβ KD values of 0.59 μM and 0.58 μM in the absence and presence of micelles, respectively (Fig. 4 and Supplementary Table 1). The similar values indicate that Cu(II)/Aβ binding is not much affected by SDS micelles. The obtained KD values are not corrected for buffer composition/ionic strength, and should therefore be considered apparent dissociation constants in line with previously recorded values [15, 47].
Longer titration series starting with nanomolar additions of Cu(II) acetate initially show a sigmoidal pattern consistent with a single metal binding site (Supplementary Figure 4). At higher copper concentrations, however, i.e., at additions above 10 μM, a deviation from the sigmoidal pattern is observed that would be consistent with the presence of a secondary weaker binding site.
CD spectroscopy
The CD spectrum for monomeric Aβ40 in aqueous buffer displays, as expected, a typical random coil signal with a minimum around 198 nm. When SDS micelles are added Aβ transitions into an alpha-helical secondary structure, producing a CD signal with characteristic minima around 208 and 222 nm (Fig. 5). This is consistent with previous reports that Aβ adopts an alpha-helical conformation in membrane-like environments [7, 49]. Addition of either Cu(II) or Zn(II) ions to the Aβ/SDS system induces small structural alterations, observed as a reduction of CD signal intensity around the 208 and 222 nm minima. Although a general loss of alpha-helical structure cannot be ruled out, the intensity reductions at 208 and 222 nm are dissimilar, leading to increased [θ222]/[θ208] ratios indicative of increased helix supercoiling (i.e., two or more α-helices forming coiled coils via hydrophobic interactions [67–69]). When Zn(II) acetate is gradually added to a final concentration of 800 μM, the [θ222]/[θ208] ratio gradually increases from 0.70 to 0.78 (Supplementary Table 2). For copper, the ratio reaches a maximum of 0.87 at 6 μM Cu(II), and then slightly decreases to a final value of 0.85 at 250 μM (Supplementary Table 2). This more complex behavior is caused by the 208 nm intensity monotonically decreasing upon Cu(II) addition, while the 222 nm intensity initially remains unaffected and only decreases at higher copper concentrations. Plotting the CD intensity at 208 nm versus metal ion concentration produces standard binding curves (Fig. 5C). Fitting Equation 1 to these curves yields binding constants of 184±469 μM for Zn(II) and 0.47±0.40 μM for Cu(II) (Supplementary Figure 5). The latter value agrees very well with the KD of 0.58 μM for Cu(II) obtained from the fluorescence measurements. The roughly two orders of magnitude difference between copper and zinc ion binding is also in line with previous reports [15, 55]. For the Zn(II) data, as the concentration of the Aβ peptide is lower than the KD, it can also be fitted to a hyperbolic binding isotherm (i.e., Equation 2) [24]. This produces a KD of 216±114 μM, where the value is rather similar to that produced by Equation 1 but the error is smaller (Supplementary Figure 5).
H2O2 measurements
In order to monitor generation of reactive oxygen species (ROS), an assay involving incubation with TCEP was used [59, 70]. Over 70 min, 3 μM of free Cu(II) ions generated around 32 μM H2O2 in buffer, and around 25 μM in buffer with SDS micelles (Fig. 6). The amount of H2O2 generated was not significantly different in comparable systems containing 10 μM Aβ40 peptide: Aβ + Cu(II)+SDS produced 24 μM H2O2, Aβ + Cu(II) produced 33 μM H2O2, Aβ + ZAβ3(12–58)Y18L Affibody dimer (∼1 : 1 ratio) + Cu(II) produced 34 μM H2O2, and Aβ + CP-2 drug (∼1 : 10 ratio) + Cu(II) produced 31 μM H2O2. Control experiments without Cu(II) ions, with Cu(II) ions bound to EDTA or the apo-SOD1 protein (i.e. metal-free SOD1), and with Zn(II) ions produced no or negligible amounts of H2O2 (Fig. 6). Although the strong binding of copper to EDTA and the SOD1 protein inhibits ROS formation, the weaker Cu(II) binding to Aβ (nanomolar range) does not impede the capacity of Cu(II) to produce H2O2, even when Aβ itself is bound to micelles or other molecules such as the engineered Aβ-binders and aggregation-inhibitors CP-2 and ZAβ3(12–58)Y18L.
DISCUSSION
Being amphiphilic and generally disordered in aqueous solutions, the Aβ peptides are known to adopt different structural arrangements in different environments. In membranes and membrane-like environments Aβ forms α-helical structures [48–51] in approximately the same segments (i.e., residues 15–24 and 29–35) that make up the legs of the β-hairpin motif in aggregated Aβ fibrils [7]. In SDS micelles, which constitute a membrane-mimicking environment often used for biochemical studies in vitro, the hydrophobic Aβ residues in the central segments are directed towards the hydrophobic micelle interior, while the Aβ N-terminus appears to reside free and unstructured outside the Aβ/micelle complex [48, 49]. When SDS is titrated into a solution of Aβ40, intermediate Aβ/SDS co-aggregates are formed where the peptide displays β-structure in spectroscopic studies and where direct Aβ NMR signals are lost [71]. When the SDS level reaches the critical micellar concentration (CMC), the helical form of Aβ reappears [51, 72]. Despite Aβ being a membrane-binding peptide [48–52], and despite Aβ/metal ion interactions being implicated in AD progression [6, 18], surprisingly little research has been done on Aβ/metal ion interactions in membrane environments [52]. Here, we use excess micelle levels (i.e., Aβ/micelle ratios <1) to study binding of Cu(II) and Zn(II) ions to Aβ40 peptides positioned in SDS micelles. Just like for standard cell membranes, the anionic SDS head groups provide a negatively charged outer surface, which may affect the metal-binding properties of attached biomolecules[73].
Our current results show that addition of Cu(II) or Zn(II) ions to micelle-bound monomeric Aβ peptides reduces the NMR signal intensities for N-terminal Aβ residues. The 1H,15N-HSQC spectra show particularly strong effects on the amide crosspeaks for histidines H6, H13, and H14, indicating that these are binding ligands (Fig. 1 and Supplementary Figures 1 and 2). This is in line with previous research showing these three Aβ histidines to be involved in coordinating Cu(II) and Zn(II) ions in aqueous buffer conditions [15, 47]. The fluorescence quenching experiments produce very similar KD values for Aβ/Cu(II) binding in the presence (0.58 μM) and absence (0.59 μM) of SDS micelles, respectively (Fig. 4 and Supplementary Table 1). These KD values, calculated without compensation for buffer effects [47, 54], are in good agreement with previously reported binding constants in aqueous solution [47, 55]. Thus, the NMR and fluorescence results indicate that N-terminal metal ion binding to Aβ is not much affected by the SDS micelles— most likely the N-terminal Aβ segment resides sufficiently far outside the micelle surface to be affected. Similar NMR results were obtained for Zn(II) binding to Aβ40 in complex with the Affibody dimer ZAβ3(12–58)Y18L, an engineered binding molecule that locks Aβ in a β-hairpin conformation [7] with a free and unstructured N-terminal part [54, 62].Here, the loss of signal intensity for N-terminal crosspeaks upon addition of Zn(II) again indicates specific N-terminal binding (Fig. 3), which is in line with previous results of Cu(II) binding to this system [54]. Together with the results for Aβ in SDS, the Aβ/Affibody data suggest that as long as it remains free, the N-terminal Aβ segment will retain its metal-binding capacity, regardless of whether Aβ residues 15–40 adopt alpha-helices in micelles, or form a β-hairpin in complex with a binding protein. This notion may be of biological relevance, as the unstructured N-terminal Aβ segment seems to be free to bind metal ions under many conditions. For example, solid state NMR structure studies of aggregated Aβ peptides indicate that the Aβ hairpin conformation observed in the Aβ/Affibody complex is a prominent building block in Aβ amyloid fibers [7]. In these fiber structures the N-terminal Aβ segment typically appears unstructured [7, 74], and would therefore be available for metal-binding, which would be consistent with the elevated metal levels observed in the amyloid plaques in AD brain tissue [26].
Our CD measurements show that Aβ in SDS displays a CD spectrum typical for alpha-helical secondary structure (Fig. 5), which is in line with previous reports. Addition of Cu(II) or Zn(II) acetate reduces the CD intensities at 208 and 222 nm with different amounts (Supplementary Table 2), leading to increased [θ222]/[θ208] ratios indicative of increased helix supercoiling [67–69]. This observed structural change might explain why copper-bound Aβ peptides have been reported to more readily insert into lipid membranes [75]. Plotting the [θ208] intensity vs metal concentration yields binding curves with binding constants of 184 μM for Zn(II) and 0.47 μM for Cu(II) (Fig. 5, Supplementary Figure 5, Supplementary Table 4), where the latter value agrees very well with the KD of 0.58 μM for Cu(II) obtained from the fluorescence measurements (Supplementary Table 1). As Cu(II) displays stronger binding then Zn(II), it is perhaps not surprising that Cu(II) induces the most supercoiling, i.e., a [θ222]/[θ208] ratio of 0.85, compared to the ratio of 0.78 produced by Zn(II) (Supplementary Table 2). More surprising is the idea that metal binding to the free unstructured N-terminal would affect the helix structure in residues 15–35. Previously, secondary weaker metal binding sites have been proposed to exist both in the Aβ N-terminus [22] and around residues D23-K28 [47]. The current NMR results support metal binding in the E22-V24 region, as both the amide crosspeak intensities and chemical shifts for these residues are significantly altered upon addition of Zn(II) or Cu(II) (Fig. 1 and Supplementary Figures 1 and 2). This effect is here more pronounced than in previous studies carried out in aqueous buffer (e.g., [47]), which might be attributed to the SDS micelles: if the E22-V24 residues are located close to the anionic micelle surface, then not only is the local cation concentration higher, but the SDS sulphates may help to coordinate Cu(II) ions. The two carboxyl groups of E22 and D23 cannot provide very strong Cu(II) binding on their own, but together with two SDS sulphates a suitable 4 N coordination might be achieved, which would afford significantly stronger binding [73]. This binding arrangement is however only a speculation, and needs to be further tested. Nevertheless, the fluorescence data for Cu(II) acetate titrated to Aβ in SDS micelles show a deviation from sigmoidality at high Cu(II) levels, which suggests the existence of a secondary weaker binding mode (Supplementary Figure 4). Previous work has argued that a secondary Aβ metal ion binding site may lack physiological relevance due to its weaker binding affinity [76]. The current results, however, suggest that metal ion binding to the central Aβ region may be stronger in membrane environments, and that such binding might be responsible for the observed metal-induced supercoiling, implying that metal ions could affect in a relevant way the Aβ secondary structure in membrane environments. Yet, the metal binding affinities derived from the CD measurements are on par with the Cu(II) and Zn(II) KD’s for primary binding to the Aβ N-terminal region, and the possibility of a secondary binding mode in the N-terminal region should not be ignored [22, 76]. More research is clearly needed to elucidate the metal binding to central Aβ regions and other aspects of Aβ/metal-ion interactions in membrane environments [77, 78]. For now, we conclude that SDS micelles appear to affect the structural and metal ion-binding properties of central Aβ regions, but leave the N-terminal segment unaffected.
A further important aspect of Aβ/metal-ion interaction is generation of harmful oxygen radicals (ROS). Because the amyloid plaques in AD patients contain elevated copper and iron [26], it has been suggested that neuronal damage may by induced by redox-active Fe(III) and Cu(II) ions bound to various forms of monomeric or aggregated Aβ [25]. It is well established that H2O2 can form when Aβ-bound copper reacts with molecular oxygen [16, 79], and this H2O2 can then react to form more aggressive ROS such as HO. via Fenton Chemistry or Haber-Weiss reactions [16, 80]. By accumulating Cu(II) ions, Aβ peptides may thereby induce a large amount of ROS-related molecular damage in their immediate environment. Our H2O2 measurements show that Aβ-bound Cu(II) generates ROS also when the Aβ peptide is in complex with SDS micelles, with the aggregation-inhibitor CP-2 (a cyclic peptide), or with the Affibody binding molecule ZAβ3(12–58)Y18L (Fig. 6). Just like the SDS micelles, the two engineered molecules prevent Aβ aggregation by binding to the central and C-terminal parts of Aβ [53, 61], leaving theN-terminal histidines free to bind metals.
Thus, it appears that molecular interactions with Aβ which do not involve the N-terminal Aβ residues will not affect the metal-binding properties of the N-terminal segment, and N-terminal bound Cu(II) ions will continue to generate ROS. In addition to furthering our understanding of Aβ/metal-interactions in membranes, this notion might be important also for drug development, as most molecules designed to prevent Aβ fibrillation target residues 15–35, i.e., the region responsible for aggregation via β-sheet stacking [7].
Conclusion
Although monomeric Aβ peptides residing in SDS micelles display alterations in secondary structure and metal-ion binding for the central and C-terminal segments, compared to Aβ in aqueous solutions, binding of Aβ to SDS micelles, engineered aggregation-inhibitors such as CP-2 and ZAβ3, or other molecules that do not interact with the Aβ N-terminal part, appears not to compromise the N-terminal segment’s ability to bind metal ions, nor impede the capacity of N-terminally bound Cu(II) to generate ROS.
Footnotes
ACKNOWLEDGMENTS
The work was funded by grants from the Magnus Bergvall foundation to SW, and from the Swedish Research Council and the Brain Foundation to AG. The discussion of the results was greatly improved by the insightful and constructive comments by the anonymous reviewers.
