Alzheimer’s disease (AD) is a multifactorial neurodegenerative condition constituting the majority of dementias. Primary feature of AD is neuronal cell loss in the hippocampus and cerebral cortex, areas involved in memory and cognition (Bernardo et al., 2007; Kim et al., 2008). Histopathological characteristics are depositions of amyloid plaques, comprising extracellular accumulations of fibrillar amyloid β (Aβ)-peptide, and the formation of intracellular neurofibrillary tangles composed of hyperphosphorylated tau (P-tau; Hooijmans et al., 2009; Kim and Tsai, 2009). Aβ is produced by cleavage of the amyloid precursor protein (APP) by β-secretase (BACE-1) and γ-secretase, which is comprised of four proteins: presenilin (PS) -1 or -2, PEN, Aph-1 and Nicastrin. Cleavage of APP by β-secretase results in a N-terminal soluble fragment and a C-terminal fragment that is further cleaved by γ-secretase resulting in Aβ peptides. Missense mutations in either APP or PS-1 can cause accumulation of Aβ in hereditary AD. The mechanism leading to Aβ accumulation in the majority of sporadic AD patients is unclear (Mare et al., 2007). Extracellular aggregation of the Aβ-peptide is considered a central and causative phenomenon of AD (Yoshiike et al., 2001; Hooijmans et al., 2009; Zatta et al., 2009; Finder et al., 2010). However, in AD patients, Aβ is also present in elevated amounts within the degenerating neurons, and this may contribute to cell death (Hasegawa et al., 2005).

In vivo Aβ has two predominant forms: Aβ1-40 and Aβ1-42 with two additional hydrophobic residues at the carboxyterminus. Aβ1-40 is the main soluble species, whereas Aβ1-42 is the predominant species found in amyloid plaques. The latter is more toxic to neurons and is considered the most amyloidogenic species, most likely responsible for the neuropathology in AD (Hasegawa et al., 2005; Mare et al., 2007; Finder et al., 2010). Amyloid β aggregation is believed to happen in phases: first, Aβ monomers associate into soluble oligomers that then form insoluble oligomers (initial slow nucleation or “seeding”), generating protofibrils, and fibrils (Finder and Glockshuber, 2007; Tõugu et al., 2009).

Whether Aβ forms fibrils in vitro in the presence of copper and the nature of these fibrils is currently a subject of debate. Main question is the accelerating or preventing role of copper in amyloid fibril formation, and whether this role is dependent on Cu²⁺ or Aβ concentration and stoichiometry. It has been presented that sub-stoichiometric concentrations of Cu²⁺ accelerate amyloid fibril formation, and supra-stoichiometric concentrations of Cu²⁺ prevent fibrillization (Viles, 2012). There are several studies where Cu²⁺ was reported to inhibit fibril formation and rather form amorphous aggregates (Yoshiike et al., 2001; Raman et al., 2005; Tõugu et al., 2009; Innocenti et al., 2010). On the contrary, the opposing arguments rely mainly on the study of Sarell et al. (2010) where the substoichiometric levels of Cu²⁺ were shown to accelerate fibril formation of Aβ. A recent study of Mold et al. (2013) addresses this dilemma by fluorimetry and transmission electron microscopy (TEM). In this study they show that Cu²⁺, independent of stoichiometry, prevented the formation of ThT-positive amyloid fibrils of Aβ42.

Amyloid plaques are composed of fibrillar Aβ, small amounts of other proteins and transition metals like copper and zinc (Tõugu et al., 2009). Several studies have shown that homeostasis of the transition metals copper and zinc can greatly influence Aβ misfolding and plaque formation. Furthermore, restoring metal ion homeostasis dissolved Aβ plaques in mice and delayed cognitive deficits in AD patients (Zatta et al., 2009). Thus, an interaction between Aβ and copper may be involved in AD pathology (Klevay, 2007a,b).

We have previously shown that homocysteine (Hcy) binds copper, and that this may be an important mechanism of the neurotoxicity of Hcy, as the presence of Hcy can lead to deficiency of copper-dependent enzymes like cytochrome-C-oxidase (White et al., 2001; Apostolova et al., 2003; Linnebank et al., 2006). Hcy is a non-proteinogenic sulfhydryl-containing amino acid formed as an intermediate in the metabolism of methionine (Hasegawa et al., 2005; Bernardo et al., 2007; Kim et al., 2008). Deficiencies of vitamin B12 or folate, common conditions in the elderly, can lead to hyperhomocysteinemia, which is a risk factor for cardio- and cerebrovascular diseases as well as neurodegenerative disorders such as AD (White et al., 2001; Irizarry et al., 2005; Linnebank et al., 2006; Bernardo et al., 2007; Kim et al., 2008). In hyperhomocysteinemic patients, blood copper levels are elevated, possibly due to binding to increased amounts of Hcy (Apostolova et al., 2003; Linnebank et al., 2006). In cell culture, Hcy sensitizes neurons to Aβ toxicity by induction of intraneuronal Aβ accumulation due to speculative mechanisms (Hasegawa et al., 2005). In addition, hyperhomocysteinemia increases Aβ production in rats, probably through enhanced expression of γ-secretase and APP phosphorylation, placing hyperhomocysteinemia upstream of increased Aβ production (Zhang et al., 2009). In this study, we aimed at modelling the interaction between copper, Hcy and Aβ fibril formation.

Hyperhomocysteinemia is a risk factor for AD, in which Aβ fibrillization plays an important role. Our study suggests that copper is a link between Hcy and Aβ. First, via a ThT assay and TEM, we confirmed that in vitro CuCl₂ prevents and reverts Aβ fibril formation (House et al., 2004; Bolognin et al., 2011; Chen et al., 2011; Figures 1A–C, 5A–C). Addition of CuCl₂ results in decreased Aβ fibril length and complexity i.e., lack of higher order aggregates (Figure 2C). The underlying mechanism is most likely connected to the ability of CuCl₂ to prevent the formation of Aβ42 beta-sheets in vitro (Yoshiike et al., 2001; House et al., 2004), but due to the nature of ThT assay we cannot completely rule out the possibility of the formation of oligomers or amorphous aggregates. However, Hcy did not affect the fibrillar structure of Aβ as seen in ThT assay and in TEM images (Figures 1B, 2B).

Our ThT aggregation assay results of CuCl₂ and Aβ are in-line with other published results with similar experimental setup. It has been presented that in experimental conditions where fibril formation is fast, metal ions lower concentration of free peptide and thus inhibit fibrillization (Tõugu et al., 2009). In conditions with slow fibril formation metal ions enhance fibril formation by metal-induced aggregates that can turn into fibrils (Sarell et al., 2010). Differences in experimental setup are also greatly influencing the outcome of fibrillization studies. This is summarized in a recent publication by Viles (2012) and shows multiple differences associated with changes in stoichiometry, used peptide preparation, concentration and study technique.

Cytotoxicity experiments performed in the absence of Aβ42 showed that CuCl₂ or Hcy alone had no effect on cell viability (Figure 3A). In accordance to previous studies, cytotoxicity was elevated when cells were co-incubated with CuCl₂ and Hcy, confirming the toxicity of Hcy-copper-complexes (Figure 3A; White et al., 2001). In accordance with our ThT assay data and TEM images, toxicity of mature Aβ42 fibrils to primary neurons was high, and Aβ fibrils formed in the presence of Hcy and CuCl₂ showed similar toxicity (Figures 1C, 2A–C, 3B,C). Aβ42 co-incubated in 1:1 molar ratio with CuCl₂ showed the same level of toxicity as monomeric Aβ42 underlining that CuCl₂ effectively inhibited the formation of bigger, more mature fibrils during ThT aggregation assays (Figures 2C, 3B; House et al., 2004; Chen et al., 2011). Is has been reported that the fibrillar status of Aβ42 affects its cytotoxicity in cultured neurons: small soluble Aβ42 dimers and oligomers cause higher toxicity than bigger fibrillar forms (Klyubin et al., 2005; Lesné et al., 2006; Agnati et al., 2007; Ferreira et al., 2007; Haass and Selkoe, 2007; Ono et al., 2009).

As a certain time-span is required for Aβ preparations in contact with primary neurons to exert the cytotoxic properties, we cannot completely rule out that during the 72 h incubation a modification of the Aβ species can occur. However, as a result of our Aβ ThT aggregation assays we observed that the addition of CuCl₂ almost completely prevented the formation of higher order aggregates (sensitive to ThT fluorescence). As CuCl₂ itself did not have a measurable influence on cell viability, we infer that the observed cytotoxicity is due to Aβ monomers and potentially aggregates of lower molecular weight, such as Aβ di- and oligomers, Aβ-derived diffusible ligands or protofibrils (all probably being ThT fluorescence negative) that may have formed during the incubation period.

Copper has four coordinating ligands in human Aβ: 3N and 1O, which involve His6, His13, His14 and possibly Tyr10, carboxylate group of Asp1, the amide of Ala2, and the N-terminal amine. To date, no consensus of exact coordinating ligands exists. When single histidines were mutated to alanine, Hong et al. (2010) observed that His6 has three times higher copper binding constant than His13 or His14. His6 also requires less conformational changes upon copper binding, making it more entropy-favored. His6 was concluded to be ubiquitously involved in copper binding, accounting for 50% of the Aβ bound Cu(II). In our experiments, Hcy had only minor effect on the fibril formation of H6A mutant of Aβ42 with reduced affinity to copper (Figure 4A). Also the effect of CuCl₂ was weaker in H6A fibrillization, and the concentration of Hcy needed to restore H6A fibril formation was reduced compared to wild type Aβ42 (Figures 4B, 1C). This allows the speculation that the amount of mature fibrils i.e., higher order aggregates in this sample was higher compared to the wild type Aβ42. In accordance to this, H6A fibrils also showed less pronounced cytotoxicity in the presence of CuCl₂ (Figure 4D). The control, ScAβ, expectedly showed no fibrils in ThT-assay (Figure 4C). Accordingly, ScAβ was not cytotoxic and this was not influenced by incubation with CuCl₂ or CuCl₂ plus Hcy (Figure 4D). In ScAβ the copper binding sites are lost, confirming that binding of copper to Aβ was the decisive mechanism for the observed differences in the ThT assay and the cytotoxicity experiments of the different combinations.

The addition of CuCl₂ to already fibrillized Aβ led to untangling of fibrils (Figures 5A–C). This does not necessarily mean that copper is relevant for Aβ fibrillization in vivo, however, mice with defective copper transport have decreased brain copper levels together with increased amounts of amyloid plaques. When these mice are crossed with Wilson’s disease mouse model, the offspring have increased brain copper levels, less amyloid plaques and a longer life span (Phinney et al., 2003). When copper sulphate was added to the drinking water of mice susceptible to amyloid accumulation, less accumulation was observed (Bayer et al., 2003). In patients with mild to moderate AD, plasma copper negatively correlates with cognitive abilities (Pajonk et al., 2005; Kessler et al., 2006). In addition, AD patients show elevated serum levels of free copper (serum copper not bound to ceruloplasmin) (Squitti et al., 2004, 2005, 2006, 2009, 2011; Capo et al., 2008) while autopsy samples of hippocampus and amygdala from AD patients showed generally reduced copper contents (Deibel et al., 1996; Klevay, 2008).

In contrast, in the amyloid plaques, copper concentration can be as high as 400 μM, although normal brain extracellular concentration is 0.2–1.7 μM (Gutteridge, 1984; Kardos et al., 1989; Linder and Hazegh-Azam, 1996; Lovell et al., 1998; Schümann et al., 2002; White et al., 2004; Squitti et al., 2006). This data seems to be in contrast to our observation that CuCl₂ prevents amyloid fibril formation. However, in our experiments, CuCl₂ alone prevented and reverted aggregation, whereas Hcy plus CuCl₂ did not. Thus, it would be interesting to analyze whether copper in amyloid plaques of AD patients is bound to Hcy or other molecules. Moreover, the inhibitory effect of CuCl₂ on fibrillization was concentration-dependent (Figures 5B,C). Thus, one may speculate that in the copper-rich plaques of AD patients, copper levels may not have reached the necessary concentration.

Serum Hcy concentrations over 14 μM are an independent risk factor for the development of AD (Seshadri et al., 2002). In our study, the addition of Hcy alone did not change Aβ42 fibril formation (Figure 1B), although the addition of Hcy slightly reduced Aβ42 toxicity in vitro (Figure 3B). In co-incubation experiments, Hcy concentration-dependently reduced the inhibitory effects of CuCl₂ on Aβ42 fibrillization suggesting that Hcy and Aβ compete for copper binding, i.e., homocysteine-bound copper has reduced or no effects on Aβ42 fibril formation (Figure 1C). Two different complexes are possible between Hcy and copper, showing molar ratios of 1:1 or ≤1:3 (Apostolova et al., 2003). Similarly, the majority of the copper-Aβ complexes form with a 1:1 stoichiometry at physiological pH (Karr et al., 2005; Syme and Viles, 2006; Tõugu et al., 2008; Faller and Hureau, 2009). For H6A, Hcy was more effective in neutralizing the effect of CuCl₂ on fibril formation confirming that CuCl₂ is less effective in preventing fibrillization of this mutated peptide due to its reduced affinity to copper (Figure 4B).

In summary, this study shows that both Hcy and Aβ42 bind and compete for copper. Copper prevents and reverts fibril formation by binding to Aβ42 and thereby increases Aβ toxicity. Homocysteine builds toxic complexes with copper and concentration-dependently prevents the effects of copper on Aβ42 fibrillization. In the presence of Aβ42, neurotoxicity of copper is reduced giving rise to the speculation that one physiological Aβ function might be the prevention of copper neurotoxicity. Due to complex building, hyperhomocysteinemia reduces the availability of free copper, which in the light of our results, likely increases amyloid plaque formation. Acute presence of high copper concentrations untangle aggregates leading to high concentrations of mono- or oligomeric Aβ42-copper complexes causing marked neurotoxicity. The analysis of the interaction between Aβ42, copper and Hcy in patients may lead to novel therapeutic strategies in the prevention and treatment of AD.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.