Bioinorganic chemistry of copper and zinc ions coordinated to amyloid-β peptide

Abstract

In the present perspective, we give a critical review on the coordination chemistry of the metal ions copper and zinc to the amyloid-β (Aβ) peptide; such complexes have been linked to Alzheimer's disease. We focus on two main issues: the identification of the coordination sphere of the Cu(II) and Zn(II) ions and the affinity of these metal ions towards the peptide. With the aim to come up with as few as possible valuable structural models and binding affinity values, we critically review the divergent propositions reported in the literature and take into account the experimental differences and the limits of the methods used in the published studies. We propose that: (i) the conditional dissociation constant of the Cu(Aβ) complexes lies in the range of 10 pM to 100 nM, with a preference for the region between 100 pM to 1 nM. (ii) Two most likely coordination modes for the predominant form of the Cu(Aβ) complexes at physiological pH can be retained, both being 3N1O distorted square planar. In the first model, the Cu(II) ion is coordinated by the Nτ atoms of the three His residues and the carboxylate of the Asp1. In the second model, both the N-terminus and the carboxylate functions of Asp1 are ligated together with the Nτ of His6 and of His13 (or His14). An equilibrium between these two forms at room temperature, and a preferentially freezing out of the second one would explain most of the divergences in the published results (in particular, between those obtained by EPR and NMR). (iii) The apparent dissociation constants of Zn(Aβ) in various buffers are in the range of 1 to 20 μM (a 10 times lower conditional dissociation constant can be estimated. (iv) For the Zn(II) coordination, the implication of the three His and the Asp1 residues is consensual. The Asp1 can be coordinated by the carboxylate and/or the N-terminus functions. Additional ligands are possible, such as Glu11 or H₂O.

---

Peter Faller

Peter Faller studied biochemistry at the University of Zürich (CH), where he did his PhD with M. Vasak working on metallothionein-3. He worked as a postdoc on photosystem II, with A. W. Rutherford (CEA Saclay, F) and with A. Liszkay (Freiburg, D) on tyrosine/carotenoid radicals, and Cd-toxicity in photoactivation, respectively. In 2003 he was appointed as a Professor at the University Paul Sabatier in Toulouse (F) and he is a group leader at the Laboratoire de Chimie de Coordination du CNRS.

Christelle Hureau

Christelle Hureau was born in Charleville-Mezières (F) in 1976. She studied physical chemistry at the University of Paris XI, where she also did her PhD on Mn-based inorganic models of photosystem II under the supervision of Prof. Girerd and Drs Anxolabéhère-Mallart and Blondin. She focused on electrochemistry and advanced EPR techniques during her postdoctoral stays at the University of Paris VII (Dr Limoges) and at the CEA-Saclay (Dr Un), respectively. She also studied Cu-peptide models of the prion protein at the University Joseph Fourier (Prof. Charlet). She is now a permanent researcher in the group of Prof. Faller.

---

Alzheimer's disease

Societal impact and biological aspects

Alzheimer's disease (AD) is a progressive and devastating neurodegenerative pathology that is the most common cause of dementia. It is currently estimated that AD is striking 24 million victims worldwide: 1 person in 20 over the age of 65 and 1 in 5 over the age of 80 (ref: http://www.alz.co.uk/alzheimers/faq). No curative pharmacological treatment is available at present. Hence, because the median age of the industrialized world's population is increasing gradually, the number of AD victims is expected to increase in the near future, which will place the health-care system under an increasing pressure.

The disease can begin many years before it is eventually diagnosed. In its early stages, short-term memory loss is the most common symptom. Later symptoms include confusion, anger, mood swings, language breakdown, and long-term memory loss. Gradually the sufferer loses minor, and then major bodily functions, until death occurs. These events are accompanied by a diminution in the size of the brain regions involved in memory and learning due to death of neurons and synaptic degeneration. The morphological hallmarks of AD are the extracellular amyloid plaques (or senile plaques)¹ and the intracellular neurofibrillary tangles. Neurofibrillary tangles are fibrillary aggregates of microtubule associated tau protein that shows hyperphosphorylation and oxidative modifications.^2,3 The plaques are mainly constituted of fibrils and amorphous aggregates of a peptide called amyloid-β (Aβ).^4,5

Aβ peptide and the amyloid cascade hypothesis

The Aβ peptide is cleaved off from a membrane protein called amyloid precursor protein (APP) by the β- and γ-secretase enzymes (see Fig. 1).^4,6 APP is a transmembrane protein existing in different isoforms of 695–770 amino acids. The location of the Aβ is partly in the transmembrane region (amino acids 597–639 for the isoform APP 695). Aβ is typically a 39–43 residue polypeptide consisting of a largely hydrophilic N-terminal domain (1–28) and a C-terminal hydrophobic domain (29–39/43) (see Fig. 1 and 2).

Fig. 1 Simplified view of the amyloid cascade hypothesis.

Fig. 2 Sequences of the native Aβ peptides (Aβ40/42) and of the truncated forms Aβ16/28. Amino acid residues involved in ROS production (↓), absent in the rat that doesn't develop AD (↑), possibly involved in the Cu(II) (blue square) or Zn(II) (black square) binding.

Aβ is present in healthy brains in a soluble form.⁷ Since the amyloid plaques occur particularly in AD patients, the aggregation process from Aβ to the plaques is the key event. According to the amyloid cascade hypothesis,^8,9 an increased Aβ production and accumulation lead first to the formation of Aβ-oligomers, then to protofibrils and ultimately to fibrils which are the main constituent of the amyloid plaques¹⁰ (see Fig. 1). These oligomers are supposed to provoke neuronal dysfunction and later on dementia, likely via the production of reactive oxygen species (ROS). In vivo, the most prevalent forms of Aβ consist of 40 (Aβ40) and 42 amino acids (Aβ42) (see Fig. 2). The ratio of Aβ42/Aβ40 is increased in amyloid plaques compared to the soluble fraction. Both forms play an important role in AD, the two hydrophobic terminal residues of Aβ42 render it more prone to aggregation than the Aβ40. Aβ42 is also more toxic to neurons than Aβ40, in agreement with the amyloid cascade hypothesis.⁸

It is currently conjectured that soluble, oligomeric forms of Aβ are the most toxic species, rather than more aggregated fibrils or protofibrils.^8,10 This is supported by studies indicating that biochemically-measured levels of soluble Aβ, including soluble oligomers, correlate much better with the presence and degree of cognitive deficits than do simple amyloid plaques counts. However, there is still some debate if Aβ aggregation is the cause or only a consequence of AD and whether the oligomers are the toxic moieties responsible for synaptic dysfunction and neuronal cell loss in AD.¹¹

The structure of Aβ40 and Aβ42 in buffer at physiological pH is largely random coil.^12,13 In membrane-mimicking media, it is mainly α-helical.^14,15 The aggregation process is accompanied by the transconformation towards β-sheet structure,^16,17 an event that is assumed to trigger the abnormal fibrillar amyloid deposition associated with AD. A model of the 3D structure of Aβ fibrils has been deduced from solid-state NMR investigations.^18,19 Such misfoldings of protein or peptide leading to their aggregation are a common feature of numerous degenerative disorders, that include Parkinson's disease, prion diseases, and amyotrophic lateral sclerosis in addition of AD.^20,21

Fe, Cu and Zn ions

Another similarity of these neurodegenerative diseases is the role of transition metal ions. A large body of evidence indicates that they are directly involved in the neurodegenerative process.^22–24 More particularly, several studies have shown dyshomeostasis in the brain levels of iron, copper and zinc ions and abnormalities in their metabolism in AD disease.^25–27

Zn, Cu, and Fe ions are found in the amyloid plaques at high concentrations (∼mM)²⁸ and recent studies on AD tissues showed that Zn and Cu are co-localized with the amyloid-β deposits.^27,29In vitro studies revealed that these same metals generally promote the aggregation of Aβ,^30–33 and that metal chelators can dissolve the proteinaceous deposits from post-mortem AD brain tissues.^34,35 Recent studies^36–38 even provided evidence that contaminating trace amounts of metals (sub μM concentrations) were necessary for fibril formation of 10–20 μM Aβ and another report show that Aβ at physiological concentrations (<nM) do not aggregate by itself.³⁹ The implication of metal ions in the amyloid cascade linked to AD is schematised in Fig. 1.

Fe ions are found localized in a high concentration in human amyloid plaques (∼1 mM).^28,40 However, Fe ions are not likely to interact directly with Aβin vivo. Although in vitro studies indicate that Fe is able to interact with Aβ, Fe does not co-purify with Aβ extracted from plaques⁴¹ unlike Zn and Cu ions, and is predominantly located in neuritic processes within the plaque themselves associated with ferritin. A number of studies suggest that Fe homeostasis is altered in AD, however this is likely to be a secondary effect.²⁶ Hence, we have decided not to discuss the role of the Fe ions in the neurodegenerative processes associated with AD in the present review.

Cu ions are normally bound to Cu enzymes or proteins under normal physiological conditions. Cu can be released into the synapse upon presynaptic excitation reaching up 15 μM in the synaptic cleft, although the nature of the released Cu complex is not known. Cu was found in high concentrations in amyloid plaques (∼400 μM)²⁸ compared to the normal brain extracellular concentration of 0.2–1.7 μM (reviewed in ref. 26). There is also a clear relationship between the redox active Cu ion and reactive oxygen species (ROS), e.g. H₂O₂, and HO˙ one of the most reactive ROS in nature, that is generated by the Fenton-type reaction between the reduced transition metal ion and H₂O₂.^42–45 ROS generated under oxidative stress conditions play a key role in AD.^42,46 The Cu-induced Aβ aggregation is still controversial. Reports of accelerating and inhibiting effects have been published. It seems to depend on the conditions (e.g. pH was identified) and type of aggregated state.^31,47,48

Zn ion levels can be as high as ∼1 mM in amyloid plaques.²⁸ Histological staining of normal brain indicates that the distribution of Zn resembles the areas of the brain most prone to amyloid deposition and neuropathy in AD which includes the hippocampus, amygdala and parietal lobe.⁴⁹ The primary source of labile Zn in the brain is from Zn released into the synapse during transmission and can reach concentrations up to ∼200–300 μM.⁵⁰Zn ions can promote Aβ aggregation and plaques formation, and this activity may not be as a neurotoxic modulator but rather as a neuroprotective agent since Zn can attenuate Aβ toxicity in cortical cultures.⁵¹ The precise mechanism of cytoprotection is not clear, though possible mechanisms include competing with Cu for Aβ binding and thereby inhibiting Aβ initiated redox chemistry (reviewed in ref. 26).

Experiments with neuronal cell cultures indicated the important role of the Cu to Aβ interaction: the Cu(II) ion was not toxic and the Aβ peptide alone was less toxic to primary neuronal cell culture and HEK cells than Cu(Aβ) complexes.^41,42,52 Recently, we also showed that the use of metallothionein 3 (MT3) can reduce the Cu(Aβ) induced toxicity towards cell cultures by a metal swap.⁵³ This mechanism consists in the reduction and sequestration of the Cu ion in MT3, and concomitant transfer of a Zn ion from MT3 to the Aβ peptide. Transgenic mice, which express human APP serve as a model for AD because they are able to develop the amyloid plaques pathology. In such mice, the lack of the Zn-transporter ZnT3 that transports Zn ions into synaptic vesicles reduced the plaque load. Hence, it was concluded that endogenous Zn contributed to the amyloid deposition in transgenic mice.⁵⁴ Also in the case of Cu, studies with such AD model mice showed the mutual influence of Cu and Aβ, i.e. dietary supplement of Cu or the increasing of the Cu content in the body (by a mutant of a Cu-transporter) affected the Aβ metabolism.^55,56 Similar results were described with an AD rabbit model in which trace amounts of Cu in water induce amyloid plaque.⁵⁷ However, it has also been reported that both dietary exposure and endogenous Cu elevation reduces amyloid burden in vivo,^55,56 thus pointing out an intricate role for the metal ions in AD.

A clear consequence of what was detailed in the last few paragraphs is that in vitro, in cello and in vivo studies indicate an important role of metal ions in AD. This led to the hypothesis that metal ions could interact with Aβ, thus being involved in the aggregation process. In their aggregated states, the metallated Aβ peptides could be toxic to cellsvia the production of ROS (see Fig. 1). In this context and with the ultimate aim to rationalize the design of therapeutics, the deep knowledge of the coordination chemistry of the Cu(Aβ) and Zn(Aβ) complexes are of great importance.

In the present review, we focus on the bioinorganic chemistry of the Zn and Cu ions with Aβ, in particular on the identification of their coordination sphere and on the estimation of their affinity to Aβ (see ref. 58–60 for reviews with different focuses). We tried to critically review the literature and to distil a few of the most reasonable conclusions regarding the Cu(Aβ) and Zn(Aβ) complexes. As the aim was to give some clear ideas about the subject, we had to sort out contradicting results by weighting the different reports.

Structural characterizations of the Cu and Zn coordination to the Aβ peptide

Full length and truncated Aβ peptides

It has been proposed that the metal binding site lies in the first 16 amino acids of Aβ (Aβ16).^61–63 This truncated peptide shows no tendency to aggregate or to form fibrils under moderate concentrations, thus being a decent model for the soluble metallated Aβ peptides. Another truncated Aβ peptide consisting of amino acids 1–28 (Aβ28) has been widely studied. Indeed, it contains the metal ions binding part and it can undergo aggregation and fibril formation under moderate conditions, although slower than those obtained with the native Aβ40 and Aβ42 peptides. Sequences of the Aβ16 peptide, model of the metal ion binding site, and of the Aβ28, also model for the aggregation behaviour, are compared in Fig. 2 with those of the native Aβ peptides.

Cu(II) binding site(s) in soluble form

Stoichiometry of the Cu(Aβ) complexes. The coordination chemistry of Cu(II) ion has been extensively studied in the past few years both with truncated Aβ16 and Aβ28 and with native Aβ40 and Aβ42 peptides. However, no real consensus has been reached until now, especially on the identity of the coordinating ligands, a difficulty that can be attributed to the dynamics of the Cu(Aβ) complexes and the different conditions (buffer, Cu(II) concentrations, truncated or native peptides, etc.) used in the various studies. In the present review, we discuss the different hypotheses reported in the literature with the objective of proposing one or a few reasonable model(s).

Several studies using EPR, CD and ITC showed that the Aβ peptide can bind two equivalents of Cu(II) in a sequential and ratiometric way,^36,64–66 where the first Cu(II) equivalent shows about a 100 times stronger affinity for the Aβ peptide than the second one⁶⁵ (see also below). Other studies reported only one binding site,^67,68 but this is likely because the lower affinity of the second Cu(II) equivalent precludes its detection. The second binding site is more difficult to detect in a buffer with a strong affinity to Cu(II) like Tris, compared to a buffer with a weaker affinity like Hepes (see below). In mass spectrometry weaker metal-peptide complexes can be disrupted, which would explain why Jiang et al.⁶⁷ did not detect the binding of a second Cu(II) whereas Maet al. did.⁶⁹ So the question is not if Aβ can bind a second Cu(II) but if this second weaker binding site will be ever occupied under physiological conditions, and hence if it is physiologically relevant. The relatively low affinity, and the fact that the Aβ content in the plaques exceeds the one of Cu (see below) indicate that the second binding site is not occupied.

The vast majority of reports on Cu(II) titration were in line with the subsequent formation of a monomeric Cu₁(Aβ)₁ complex prior to aggregation (that we note Cu(Aβ) unless required for clarity purpose). In general no indication of a stable dimeric or multimeric complex was found.^64–66,70 Only Barnham et al.^71,72 reported a cooperative formation of a binuclear species Cu₂(Aβ)₂ where the two metal centers were linked together via a bridging histidine residue. The proposition of such a Cu₂(Aβ)₂ species relied on the broadening of the EPR signatures from 0.3 eq. to 1.0 eq of Cu(II) per peptide. The broadening of the EPR signal was only observed in phosphate buffered saline (PBS). If the same authors used ethylmorpholine as buffer, only the monomeric species was observed and no broadening of the EPR signal due to a dimeric Cu₂(Aβ)₂ species was detected.⁷³ So the monomeric species is formed in the absence of buffer,⁶⁴ in Hepes,⁶⁵ in ethylmorpholine,⁷³ in Tris,⁷⁰ and in PBS with 50% glycerol.⁶² Indication of a dimeric species was only observed in PBS and in the absence of glycerol. This could be due to some peculiarities of the PBS buffer. For instance, the dimeric form could be induced by a pH change, because phosphate buffer in presence of sodium shows an important decrease in pH (up to 3 units) upon freezing. In contrast, the Hepes buffer shows only a moderate change (−0.4 units) and the Tris buffer can increase up to 2 pH units.⁷⁴ The other possibility is that there exists some interference between the phosphate buffer and the Cu(Aβ) complex, e.g. a binding of the phosphate to the Cu(II) ion. Also other parameters may be important, such as the way of freezing (could explain the difference in the presence and absence of glycerol). As a consequence, the presence of a histidine-bridged Cu₂(Aβ)₂ system similar to the one found in the structure encountered in the Cu–Zn superoxide dismutase is still hypothetical.

We conclude that (i) the coordination of the first Cu(II) equivalent to the Aβ peptides leads to a mononuclear complex Cu₁(Aβ)₁; (ii) the second Cu(II) equivalent can be bound to form a Cu₂(Aβ)₁ complex (without much changing the first site, see below), but it is unlikely that this happens under physiological conditions; (iii) the existence of the histidine-bridged Cu₂(Aβ)₂ complex at room temperature needs still to be demonstrated.

Cu(Aβ) complexes at physiological pH. Potentiometric and EPR measurements indicate that around physiological pH (i.e. between pH 6 and pH 8), two Cu(Aβ) species prevail, that we shall note species I (low pH) and II (high pH). Generally, the transition between species I and II has been reported at pH 8 ± 1,^{61,64,65,75–77} where the variation may be due to different experimental conditions (in particular the use of different buffers leading to changes in pH upon freezing, see above).

Most studies of Cu(II) in the high-affinity binding site (Cu(Aβ)) have focused on the characterization of species I, which is the one that predominates around physiological pH. There is now a consensus on a type II Cu with a 3N1O coordination mode, a proposition mainly based on EPR (see Table 1) and EXAFS measurements.^{63–65,70,81,82} However, one has to keep in mind that neither the use of the Peisach Blumberg correlation⁷⁹ nor the EXAFS studies allow a clear cut distinction between oxygen or nitrogen ligands. Moreover, EPR signature is also sensitive to the geometry and slight tetrahedral distortions from the square based geometry can induce significant change in the parameters. Hence a 2N2O or a 4N coordination mode cannot be excluded.

Table 1 EPR parameters for the different species observed around physiological pH

Site 1 (high affinity site)
	Species I (“low pH”)					Species II (“high pH”)
	A//^a	g//	g⊥	xNyO ^b	Ref.	A//^a	g//		xNyO ^b	Ref.
Soluble form	170–177	2.26		3N1O	61,64,70	156–170	2.23		3N1O/4N	61,64,77
Insoluble form	160	2.27^c		3N1O	78
Both forms	168	2.27	2.06	3N1O	66

---

Site 2 (low affinity site)
	Species I (“low pH”)					Species II (“high pH”)
	A//^a	g//	g⊥	xNyO ^b	Ref.	A//^a	g//	g⊥	xNyO ^b	Ref.
a A in Gauss. b From Peisach and Blumberg correlation.^79 c We used the parameters provided for a Cu(Aβ) soichiometry of 0.1,^78 because as reported in their Erratum^80 the Cu(II) concentration in the original article was 10 off, i. e. the ratio of 0.1 is in reality a ratio of 1.
Soluble form	154	2.28	2.07	2N2O/3N1O	65	173	2.23	2.06	3N1O	65
Insoluble form	160	2.27		2N2O/3N1O	78
Both forms	157	2.30	2.07	2N2O/3N1O	66

---

Before discussing the identity of the possible N and O ligands, some general considerations may be outlined. Firstly, the principle of hard/soft acids and bases proposes a preference of Cu(II) towards N over O ligands. Secondly, Cu binding is in competition with protonation, which means that Cu(II) binds preferentially to N or O ligands with a low pK_a. For N, such ligands include His (pK_a∼ 6.5), N-terminus (pK_a∼ 8) and to a much less extent Lys (pK_a∼ 10.5), Arg (pK_a∼ 12.5), or backbone amide (pK_a > 15). So His and then N-terminus are preferred. Ligands with high pK_a are only possible if the Cu(II) ion has been brought in their proximity. Indeed, presence of the Cu(II) in a close vicinity of a ligand will induce a decrease in its pK_a value. This could be accomplished by a preorganisation of the binding site in a stable 3D structure as found in proteins or by the anchorage of the Cu(II) ion via a former ligand. In the latter case, formation of a stable five-, six- or seven-membered metallacycle between the N-terminus or an histidine residue as the anchoring sites and an adjacent deprotonated amide function from the peptide backbone is often encountered. As Aβ is not much structured in its soluble monomeric form, N ligands other than His and N-terminus are very likely to be limited to amide functions next to His. Similar considerations can be made for the oxygen ligands, indicating that Ser, Thr and Tyr are unlikely due to their high pK_a values and the impossibility to form a stable metallacycle with a His residue (or another strong anchor). Possible are the carboxylate functions from Asp, Glu or C-terminus. Also possible but less strong are amide carbonyls. However, stable rings are possible with carbonyls of the backbone amides upon anchoring with adjacent His. Increasing the pH makes ligands with high pK_a more easily available. In the case of the Cu(II) ion, the number of deprotonated amide functions from the backbone, which have often been found as ligands, increases as a function of pH (see for instance ref. 83).

N ligands of Cu(Aβ) species I “low pH”. Concerning the origins of the three nitrogen ligands in the most likely 3N1O binding site, two main hypotheses come up by surveying the literature. The first one consists in a coordination made by the three His while the second one involves the N-terminus and two His residues. The latter hypothesis was proposed by Kowalik et al.⁶¹ and is supported by the following studies. Deletion of the N-terminus in Aβ16 (leading to Aβ2–16) lead to modifications in the EPR signature of the Cu(II) corresponding species.⁶² However this can be attributed either to the implication of the N-terminus or to residues from lateral chain of the Asp1 residues (see below). A more direct evidence of the implication of the N-terminus ligand is based on the observation that its acetylation lead to profound changes in the speciation curve and spectroscopic data of N-blocked Cu(Aβ16) complex⁶¹ and in the CD signatures of the N-blocked Cu(Aβ28) complex⁶⁴ compared to those of the corresponding unprotected Cu(Aβ) complexes. An indirect confirmation is given by the fact that the affinity constant for Cu(II) is about 3 to 4 time weaker in case of N-blocked Cu(Aβ28) complex.⁶⁴ Implication of three histidines residues in the binding of species I relies on EXAFS, NMR and CD studies. The chemical shifts (in ¹H NMR) of the three histidines residues (for both Aβ16 and Aβ28 peptides) are affected by addition of substoichiometric quantity of Cu(II).^64,65,71,84 In the articles of the EXAFS experiments, the authors reported only the fit with 3 His and a Tyr residues. No model involving the N-terminus as a ligand has been provided, and hence one can not tell which model fits better.^63,82 Indirect implication of the three histidine was also proposed based on alanine mutants (His6Ala, His13Ala, His14Ala) that all show modification in their CD signature compared to the non-mutated species.⁶⁴ Lastly, one very recent study using pulsed EPR techniques (ESEEM and HYSCORE) on ¹⁵N-His labelled Aβ16 peptides showed the involvement of the three histidine residues in the Cu(II) coordination.⁸⁵ Although this study clearly shows that the three His coordinate to Cu(II), it does not prove that the 3His is the predominant binding mode. Indeed, quantification of the nitrogen atoms involved in the coordination (including also the N-terminus) would be required. Moreover the study was done at pH 7.4 and did not provide the corresponding EPR signature, so one cannot tell on which species (I or II) or mixture of species the experiments were performed.

Consequently, we can conclude that the two most likely N ligands of the Cu(II) ion are the three His residues or the two His and the N-terminus. Generally, strong evidences for the former ligands set come from NMR while for the latter ligands set, they come from EPR. In fact, these two possibilities are not irreconcilable since NMR experiments showed that the broadening does not reflect quantitatively the binding since Cu(II) exchanges rapidly between the peptides.⁶⁴ A transiently bound ligand can thus be highly affected as well. In other words, this means that the Cu(II) binding is highly dynamic and that several coordination spheres with different ligands are in rapid exchange. Although this reflects the reality at ambient temperature, it makes it very difficult to deduce the predominant binding mode, including the determination of the ligands, by NMR. In contrast, the most stable binding mode(s) will likely be frozen out (but thus will also depend on the way of freezing) and detected by EPR. Thus an equilibrium between the two coordination spheres (three His and two His/N-terminus) at room temperature and freezing out predominantly the sphere with two His/N-terminus would explain most of the spectroscopic studies (depicted in Scheme 1a and 1b, see also Fig. 3).

Fig. 3 Models of the two most likely coordination spheres of Cu(II) to Aβ:{N-term, His6, His13 or His14, Asp1-COO⁻} (left panel) and {3 His, Asp1-COO⁻} (right panel). Images of the interaction between Cu(II) and Aβ were produced using VMD 1.8.6.^89,90

Scheme 1 (a) and (b) Proposed equilibriums between the possible equatorial binding sites of Cu(II) at “low” pH, and (c) proposed involvement of the third histidine in axial postion; (d) proposed binding sites of Cu(II) at “high” pH. (e, e′) The two possible coordinating modes of the imidazole ring from the histidine residue, with the two usual conventions.

Another explanation, although less convincing, for the divergence between EPR and NMR studies is that two His residues are involved in the equatorial plane, the third one being in axial position, a binding position that can not be directly sensed by conventional EPR but might perhaps induce modification in CD and NMR spectra (see Scheme 1c).

In this context, in the case of only two His coordinated to Cu one can further question which of the three His residue are indeed found in the equatorial plane and what is the binding mode of the His residue (N_πvs.N_τ). Studies of Aβ with methylated His either on the N_π or on N_τ imidazole nitrogen atoms (see Scheme 1e) showed that the binding properties are modified when the methylation occurs on the N_τ nitrogen atoms whereas no change is detected upon methylation on the N_π nitrogen atoms, thus strongly favouring a coordination of the Cu(II) ion via the Nτ nitrogen atoms of the His residues.⁷² Insights on the identity of the His residues involved in the coordination is given by the study of the substituted His6Ala, His13Ala and His14Ala that shows a decrease by about a factor of 2 in the affinity for Cu(II) of the His6Ala-Aβ28 while only a decrease by about a factor of 1.5 is reported for both the His13AlaAβ28 and His14Ala-Aβ28.⁶⁴ This suggests that His6 is present in the native site of the Aβ28 peptide while His13 and His14 are less necessary. A similar trend has been deduced from the high similarity between soluble and fibrils binding sites, indicating that both His13 and His14 can not be simultaneously involved in the Cu(II) coordination, because this would induce disruption of the β-sheet structure in the fibrils.⁶²

His13 is absent in the Aβ peptide of the rat that does not develop AD. Recently, a structure of the Cu(ratAβ28) complex in SDS, where the rat Aβ that show three substitutions (i.e. R5G, Y10F and H13R, see Fig. 2), was proposed by NMR and the Nterm function of Asp1 was identified as a ligand of the Cu(II) ion in addition to the two N_π (and not the N_τ as for the human Aβ) nitrogen atoms of His6 and His14.⁸⁶ The fourth ligand was conjectured to be a water molecule. Consequently, the absence of His13 induces significant changes in the coordination chemistry between human and rat Aβ peptides and may be responsible for the lower aggregation propensity of Cu(ratAβ) complex. This would suggest that His13 is involved in the Cu(II) coordination by the human Aβ peptide. However, it has to be underlined that the two others mutations (R5G and Y10F), despite not being directly involved in the Cu(II) coordination, can also be responsible for the different aggregating behaviour.

We conclude that: (i) different Cu(Aβ) complexes can be in equilibrium at room temperature but unequally populated. As a consequence “one” binding site does not exist. It is rather the most populated structure what is sought; (ii) the two proposed models for the most populated state are {N-term, 2 His, O} and {3 His, O} and an equilibrium between them would explain most experimental results.

O ligand of Cu(Aβ) Species I “low pH”. Concerning the oxygen donor ligand, there are several candidates, among which ligands from the peptide, either carbonyl groups from the amide backbone or lateral chain residues such as the carboxylate group from Asp1, Glu3, Glu11, phenolate group from the Tyr10 or the carbonyl group from the glutamine amide side chains. Exogenous ligands, such as water or phosphate (buffer) can also be considered. Principally, two hypotheses are found in the literature: implication of either the carboxylate function from Asp1 or of the phenolate group from the Tyr10. A survey clearly favours Asp1 over Tyr10. The strongest argument for Asp1 as the O ligand comes from the mutation of Asp1 to Asn1, i.e. the replacement of the side chain carboxylate by an amide function. This mutant showed a heavily affected EPR spectrum.⁷⁷ The straight forward interpretation of the modified EPR spectrum is that the carboxylate group of Asp1 is linked to the Cu(II) ion. Karr and Szalay proposed another interpretation, i.e. that COO⁻ of Asp1 is indirectly involved in Cu(II) coordination. Although this interpretation is novel and intriguing, it is based on the assumption that the resolution of X-band EPR is high enough to discriminate between COO⁻ from Asp1 and COO⁻ from another residue (e.g. Glu3).

Involvement of phenolate arm from Tyr10 was proposed based on EXAFS measurements,^63,82EPR studies of the Cu(Tyr10AlaAβ42) complex,⁸⁷ and Raman investigations.⁷⁵ However, these measurements have been strongly contested by several other studies with various techniques (NMR,^64,65EPR,^62,64 and CD⁶¹). Probably the strongest argument against Tyr as a ligand comes from UV-visible spectroscopy, because a tyrosinate group linked to a Cu(II) ion would exhibit a relatively specific and intense electronic phenolate to Cu(II) charge transfer band around 400–500 nm with an extinction coefficient above 1000 cm⁻¹ M⁻¹ (see e.g. ref. 88). Such a band has never been observed.^61,65,69 Moreover, the pK_a of the Tyr residue was measured by UV-visible.⁶⁵ The identical value found for the free and the metallated peptides indicated that the Tyr residue is not coordinated nor in a close vinicity of the Cu(II) ion. The strongest result in favour of the involvement of the Tyr residue was the report that replacement of Tyr by Ala led to a change in the EPR spectrum of Cu(Aβ)vs.Cu(AβTyr10Ala) complexes. However, the spectra were recorded in PBS (see above) and a later study detected no change upon replacement of Tyr by a Phe.⁶² The EXAFS studies reported^63,82 only a fit with the Tyr residue, so it is not known if other O-ligand would give a better fit or not. Indeed the authors mentioned that when they fitted the data replacing the Tyr residue by the N-terminus group, i.e. a O by a N ligand the fit was only marginally worse. This clearly shows that it is difficult to unambiguously implicate the Tyr residue as a ligand by EXAFS. Also the Raman study⁷⁵ did not included measurements on isotopically labelled Tyr residue, which would have allowed an unambiguous assignment of the Tyr bands.

Implication of the carboxylate from Glu3 and of water ligand was rejected based on EPR study of the Glu3GlnAβ16⁷⁷ and of the¹⁷O-labelled H₂O⁶² coordination, respectively. Also azide was not able to bind to Cu(Aβ), indicating that no labile ligand (e.g.water) is bound to the Cu(II).⁶⁵ As a consequence, we favour the implication of the side-chain of the Asp1 as the oxygen donor ligand for species I (see Scheme 1). Together with the proposed role of the N-terminus group, the carboxylate of the Asp1 would generate a stable six-membered metallacycle around the Cu(II) center.

Illustration of our two favourite models for the Cu(II) coordination at low pH (between pH close to 5 and to 7.5) is given in Fig. 3. Actually, it is quite difficult from these two possible Cu(II) binding sites to anticipated how they are related to the precipitation of the Cu(Aβ) complex.

Ligands of Cu(Aβ) species II “high pH”. At higher pH value (pH > 8, see above) another species (species II) formed from the deprotonation of either a lateral side chain or an amide backbone is detected. In fact, there is a consensus for the deprotonation of an amide group from the backbone based on Raman and CD studies,^64,75 which also show a concomitant modification in the binding mode of the His residues from N_τ to N_π.⁷⁵ In that case, the formation of six-membered metallacycle with the N_π atoms of the histidine and the histidine main chain deprotonated amide is likely the driving force of such induced conformational change. Another indication is given by the comparison between Aβ16 and the shorter Aβ10 peptide that do not contain the histidine diad and that however show similar spectroscopical signatures than those observed with Aβ16,^61,91 allowing to consider the absence of His13 and His14 from the coordination sphere of the Cu(II) ion upon increasing the pH. Involvement of the N-terminus group has been evidenced as for species I by the study of the N-blocked Aβ28 peptide that show dramatic modification in their CD signature compared to the Cu(Aβ28) complex.⁶⁴ All this is best in line with the ligands N-terminus, backbone amidyl and His6. A last question remains on the nature of the fourth ligand. The published EPR data^62,64,65 are more in favour of a 3N1O than a 4N coordination mode.⁷⁹ In that case, the carboxylate group of Asp1 as observed in species I is possible, but also a carbonyl group adjacent to the deprotonated amide was proposed,⁶¹ a coordination which would maintain the charge of the Cu centre during the transition between species I and II. Coordination spheres of species II (obtained between pH close to 7.5 and 9.5) are shown in Scheme 1d).

Assuming the most likely models for species I and II, a change in the pH value around physiological pH would lead to a strong reorganization of Cu(II) binding site precluding the coordination of the His13-His14 diad. The absence of coordination of His13 and/or His14 lead to a confined Cu(II) binding site (at amino acid 1–6) and to the absence of a loop between His 6 and His13 that might be responsible for the lower propensity to aggregate.¹²

Ligands of the second Cu(II) ion in Cu₂(Aβ)₁. Only a few EPR studies deal with the coordination of the second Cu(II) equivalent to the Aβ peptide.^64–66,70EPR spectra did not show any broadening or silencing due to coupling between the two Cu ions upon addition of the second equivalent of Cu(II) ion, indicating that the two Cu(II) ions are rather distant (>7 Å). The EPR parameters, as determined in the well-resolved spectra reported recently⁶⁶ indicate either a 2N2O or 3N1O coordination mode in agreement with previous studies that however show more heterogeneous EPR traces (see also Table 2).^64,65 Generally addition of the second equivalent does not modify the EPR signature of the first Cu(II) ion.^65,66 The two binding sites can thus be considered to be independent. Possible ligands are the His13 or His14 if we considered the {N-term, His6, His13 or His14, COO⁻} donor set for the first binding site or the N-term if we considered the {3 His, COO⁻} donor set. However, this latter possibility appears less likely due to structural constrains. Deprotonated amido functions, carboxylate functions from aspartate and/or glutamate lateral chains or C-term are other candidates. It is also possible that the coordination sphere is not entirely provided by the Aβ peptide, as suggested by coordination of azide to the Cu(Aβ28) complex,⁶⁵ in line with a labile exogenous ligand such as H₂O or the buffer. This is supported by the study of Jun and Saxena,⁷⁸ when using the corrected ratio from their Erratum,⁸⁰ showing that water is bound at Cu/Aβ ratios above one.

Table 2 Dissociation constants reported for the stoichiometric coordination of Cu(II) to Aβ peptides

Aβ fragment	^aKd/μM	Calculated ^cKd/nM	pH	Buffer/competing ligand	Experiments	Ref.	^cKd (nM) calculated according to:
							Ref. 109	Ref. 110	Ref. 111
a The number without brackets corresponds to the stoichiometric, but lower affinity binding site. The higher affinity value given in brackets was reported to correspond to substoichiometric metal content, and at least in case of Aβ42 to be an artefact of the measurement.
1–16/28		0.21/0.024	7.4		Potentiometry	61
1–40/42	1.6–2.0		7.4	10 mM Tris	Tyr fl.	68	0.63	73–91
1–16/28		100	7.8	Gly (His)	Tyr fl.	64	0.4
1–16/28/40	11–47		7.4	100 mM Tris	Tyr fl.	62	∼0.002	12–32
1–16		10–100	7.8	Gly (His)	Tyr fl.	69	∼0.4
1–16/28		∼100	7.4	Gly (His)	Tyr fl.	65	∼0.4
1–40	8		7.2	50 mM PO4^−	Tyr fl.	48
1–40	2.5		7.2	10 mM Hepes	Tyr fl.	104			300
	0.4		7.2	10 mM PO4^−	Tyr fl.	104
1–40	1.2/3.8/30		7.4	20/50/100 mM Tris	Tyr fl.	110	0.4/0.003/0.002	35/24/54
	0.6/0.9/2.5		7.4	20/50/100 mM Hepes	Tyr fl.	110			24–36
1–16/28	0.1		7.4	50 mM Hepes	ITC	65			2.5
1–16	8		7.4	50 mM Tris	ITC	Fig S1	0.006	50
1–40	16		7.3	5 mM PO4^− (at 4°)	NMR	84
1–16/40		0.4	7.4	Gly	ITC	109
1–40		13 (0.05)^a	7.4	Different chelator of known ^cKd	Chelator/Aβ separation	30
1–42		5 (6 × 10^−9)^a	7.4	Different chelator of known ^cKd	Chelator/Aβ separation	30
1–40/42	4/0.3		7.4	20mM CH3COO^−	Abs 214nm	31

---

Cu(II) binding site(s) in insoluble forms

Knowledge about the coordination chemistry of the aggregates is a very important question, because this could explain why aggregates are toxic but not the soluble monomeric Aβ. However, there are only a few studies in the literature that deal with this issue.^{62,66,70,75,78,92} Comparison of EPR signatures of fibers^62,66,70 as well as of oligomeric forms⁶⁶ with those of the corresponding soluble forms indicate a similar coordination mode (see also Table 1), whatever the way of the incorporation of the Cu(II) ion into the fibers or oligomeric forms, i.e. the Cu(II) ion was added in the soluble forms and the oligomeric states were then studied or the Cu(II) ion was added directly to the oligomeric forms. A stoichiometry of 1 : 1 Cu(II) : Aβ peptide was found after washing though a second Cu(II) ion was proven to be coordinated by the Aβ in the oligomeric forms in the same manner as in the corresponding soluble forms.⁶⁶ These studies also show that the mononuclear character of the Cu(II) binding site is maintained upon aggregation and thus that no cross-linking is induced by the Cu(II) coordination. This is partially confirmed by Raman studies that show no coupling viahistidinate bridge.⁷⁵ The fact that the aggregated forms of Cu(Aβ) showed the same spectroscopic features as the soluble implicates the same coordination sphere as discussed above for the monomeric Cu(Aβ) species.

Cu(I) binding site

To our best knowledge no article has been published on the coordination chemistry of the reduced state Cu(I) with Aβ. Recently two articles appeared dealing with the Cu(I) interaction with the peptide His-His,^93,94 a motif present in Aβ. However, there is a lot of evidence that Cu(I) could play an important role in the interaction with Aβ, because: (i) the brain is a rather reducing environment both in the intra-cellular (potential of −300 mV vs.NHE)⁹⁵ and the extra-cellular media (ascorbate concentration of several hundreds of μM⁹⁶), and (ii) the production of ROS by Cu(Aβ) involves the redox cycling of the Cu between Cu(I) and Cu(II).

Zn(II) binding site in soluble forms

It has been reported that the metal-binding site for the Zn(II) ion, as it is for the Cu(II) ion, lies in the N-terminal hydrophilic region Aβ16 of the Aβ peptide.⁹⁷ Insights into its coordination mode have been mainly gained by NMR^71,98–100 and EXAFS spectroscopy,⁶³ and by indirect data relying on the comparison between the Aβ16 or Aβ28 peptide and their modified forms (either mutation or acetylation of the N-terminus).^97,98 The more commonly accepted Zn(II) : Aβ stoichiometry is 1 : 1 with a mononuclear binding site, i.e.Zn(Aβ).^97–105 In contrast to what is described for the Cu(II) ion, the involvement of the three His residues in the coordination of the Zn(II) ion is consensual. However, we should note that Zn(II) is not EPR active, the technique providing strongest evidence for two instead of three His in the Cu(II) coordination. In addition to direct evidence of the involvement of the three His residues as given by NMR,^{71,98–100,104} indirect insights were also reported. The replacement of His13 by an Arg diminished the Zn affinity and the Zn-induced aggregation in Aβ28,¹⁰⁶ and the replacement of either His13 or His14 by Ala eliminated the Zn-induced conformation change to β-sheet and aggregation of Zn(Aβ28).¹⁰⁷

The identity of the further ligand(s) is still controversial. Zn-coordination involves classically four to six ligands, i.e. one to three additional ligands are possible. The candidates proposed in the literature are the Asp1,^99,104,108Arg5,¹⁰³Glu11^100,108 and the Tyr10^63,71,82 residues and an exogenous water molecule.¹⁰² Tyr10 has been ruled out by UV-visible titration of the deprotonation of the phenol group that occur at the same pH value for the apo- and the holo-form of the Aβ16 peptide.¹⁰²Arg5 as suggested by ESI-MS is very unlikely considering its very high pK_a, the flexibility of the peptide and the fact that Arg has not been described as a ligand to Zn(II) in biological systems. The best candidates are the lateral chains of Glu11 and Asp1 and the N-terminus group of Asp1. The NMR structure of the Aβ with an acetylated N-terminus clearly showed Glu11 as a ligand.¹⁰⁰ However, native Aβ is not acetylated and NMR studies showed that acetylation affects Zn-binding.⁹⁸NMR measurements of native species (truncated and full length) proposed Asp1 as ligand.^98,99,104 Several studies used full length ¹⁵N labelled Aβ to investigate the Zn(II) binding site, but in the ¹H-¹⁵N-HSQC Asp1 and the three His were not detected, so the effect of Zn-binding could not be assessed.^84,104,105 In contrast, ¹H-¹³C HSQC study of ¹³C labelled Aβ40 agreed with Asp1 as a ligand and favoured slightly the N-terminus over the carboxylate side chain.¹⁰⁴ A heterogeneity in the Zn(II) coordination with an equilibrium between two species as a function of pH, as was observed for the Cu(II) ion, can also be assumed, making it conceivable that Asp1 is bound towards the carboxylate group at lower pH and towards the N-terminus at higher. Actually, as the Zn(II) ion can easily become pentacoordinated, the two groups can also be involved simultaneously as we proposed above for the Cu(II) ion (species I, see Scheme 1a) and 1b)), thus generating a stable six-membered metallacycle. In a very recent publication,¹⁰⁸ the NMR data of Zn(Aβ28) was measured in SDS and evidences for the simultaneous binding of Glu11 and Asp1 were found. Lastly, we should note that a putative second site lying between residues 23–28 was proposed by NMR study.¹⁰⁴

We conclude: (i) the coordination site of Zn(Aβ) is most likely made of 4 to 6 ligands including the three His; (ii) Asp1 is likely to be another ligand, either with N-terminus or/and the carboxylate group; (iii) the carboxylate side chain of Glu11 is clearly a ligand when the N-terminus is blocked and may be a ligand in the native Aβ; (iv) Tyr10 and Arg5 are very unlikely; (v) water may also be involved.

Zn(II) binding site in insoluble forms

The main insights into the Zn(II) coordination to Aβ in insoluble forms have been obtained by Raman spectroscopy. Studies were reported on Zn-induced aggregates of Aβ40 and Aβ16 generated by addition of 2 to 4 fold excess of Zn to the peptide.⁷⁵ The analysis suggested that all three histidines residues provide the primary metal binding sites and that Zn(II) is bound to the N_τ of the histidine residues. The data indicated also that Tyr10 was ligand at pH 7.4 (at least partially), but no sample with isotopic labelling of Tyr was reported (see also above). It has also been suggested that the peptide aggregates through intermolecular His(Nτ)-Zn(II)-His(Nτ) bridges. Lastly, it has to be noted that similar Zn(II) binding to the histidine Nτ residues was also reported on senile plaques.⁹²

Biological significance of the Cu/Zn to Aβ interaction

Affinity of Cu(II) and Zn(II) ions for Aβ peptides

The metal binding affinity of proteins and peptides is an important parameter in biology. It gives information about the physiological significance of the interaction of a protein/peptide with a given metal ion. A binding affinity that is too low may indicate that the peptide is not able to bind the metal ion in vivo due to the presence of other ligands with stronger affinity.

The affinity of Cu and Zn(II) ions for the Aβ peptides is also crucial in order to design chelators able to retrieve the metal ions from the Aβ peptides. Such chelators, also called metal-protein attenuating compounds (MPAC), have been considered as potential drugs, since the withdrawal of the metal ion from the Aβ peptide should slow down or reverse its aggregation and hence its toxicity or can sequester the Cu ion in a redox inactive state. These chelators should be strong enough to compete with the metal-binding affinity of Aβ, but not so strong that they withdraw metal ions from essential metalloproteins.

In this respect, one has to distinguish between the different dissociation constants of the metal ion (Cu or Zn)–ligand (L) species. From the stoichiometric dissociation constant (Kd, see Equation 1 in Scheme 2), an apparent dissociation constant can be calculated at a given pH value when the pK_a of the coordinating groups are known. For purpose of clarity, we decided to note this particular apparent dissociation constant, which only depends on the pH value, as the conditional dissociation constant (^cKd, see Equation 2 in Scheme 2). More precisely, in the present report, the ^cKd is the apparent constant calculated in a 0.1 M NaCl aqueous medium at physiological pH (see Table 2 for the precise pH values). For large biomolecules the exact binding site and/or the pK_as of the ligands are not known. For that reason, the conditional dissociation value is determined at a given pH. More often the dissociation constant (^aKd, see Equations 3 and 4 in Scheme 2) is measured in a given buffer (B) or in the presence of competing ligand. In that case, the dissociation constant is called apparent dissociation constant. The apparent ^aKd can be transferred to what we noted the conditional ^cKd by taking into account competition with the extra ligands or buffer. So the apparent ^aKd allows only comparison between values measured under the very same conditions (in particular, buffer is important). To compare apparent ^aKd determined in different buffers, the buffers contribution has to be corrected and the conditional ^cKd could then be used.

Scheme 2 Definition of: (1) stoichiometric (Kd), (2) conditional (^cKd), and (3) and (4) apparent (^aKd) dissociation constants. In (3) B is independent of pH and in (4) B depends on pH. L = Ligand, B = buffer or competing ligand. The exponent ^B refers to physical parameters related to species B, others being related to L. We consider the simpler cases where M : L and M : B stoichiometries were 1 : 1 and where there is only one protonation equilibrium for both L and B species.

In the case of Aβ, there is only one article reporting stoichiometric dissociation constants for Cu(II) coordination to the truncated peptides Aβ16 and Aβ28.⁶¹ No stoichiometric dissociation constants relative to the Zn(II) ion are available. In contrast, several reports on the apparent dissociation constants for both Cu(II) and Zn(II) ions are published (see Table 2 and 3).

Binding constants of Cu(II) to Aβ peptides

The apparent ^aKd for Cu(Aβ) shows a large variation between the different reports (see Table 2). This is very likely due to the variety of techniques used to determine this value and even more significantly to the conditions (buffer) of the experiments. It seems that the length of the Aβ peptide has only a small⁶² (the longer the peptide the higher the affinity) or even no effect.¹⁰⁹ Two recent publications tried to reconcile the apparent ^aKd values from the literature by taking into account interactions with buffers. Although both were able to bring together a variety of studies on a narrow range of values, the two values differ by almost two orders of magnitude, i.e. conditional ^cKd values of 35 nM¹¹⁰ and 0.625 nM.¹⁰⁹.‡ In the following section, we will discuss the different experiments and try to analyse why the two efforts of reconciliation provided such different values.

For the Cu(II) ion, apparent ^aKd value were deduced (i) by Cu(II) titration, where the Cu(II) binding was assessed by quenching of Tyr fluorescence, NMR or isothermal titration calorimetry (ITC); and (ii) by a competition with His, Gly and Phen Green ligands followed by fluorescence or ITC. In the first case, the buffer has to be considered, because buffers can interact with Cu(II) whereas with a competing ligand stronger than the buffer, the interaction can be neglected. The buffers used were phosphate, Tris and Hepes. We discuss shortly the measurements in the three buffers, and we propose that the measurements in the Tris buffer are the most reliable:

Phosphate buffer . Cu₃(PO₄)₂ complex is very insoluble (solubility constant of 1.4 × 10⁻³⁷) and hence precipitation could occur, which may invalidate the results.

Hepes buffer. Recently Bal and co-workers measured the coordination of Cu by Hepes.¹¹¹ They determined a stoichiometry of 1 : 1 as well as the binding and pK_a constants. Thus Cu to ligand interaction can be measured in Hepes and easily corrected (see Equation 4 in Scheme 2). However, in the case of Cu binding to Aβ, Hepes buffer is not optimal because in order to get an accurate ^aKd, the measurements have to be done with a ligand (or copper) concentration in the range of the ^aKd, which is estimated to be in the nM range in Hepes. Indeed, in the fluorescence quenching experiments, done at 4–10 μM, the titration of Cu(II) led to an almost linear decrease in fluorescence as a function of the Cu(II) concentration,^65,110 making it difficult to measure accurately the ^aKd. Moreover, it is also difficult under this condition to separate the quenching from the Cu(II) ion in the high affinity site (site 1) from the one in the low affinity binding site (site 2), since Cu(II) in site 1 does not quenche the Tyr fluorescence completely and Cu(II) in site 2 quenches further. Indeed, the apparent ^aKd of site 2 in Hepes is about 10 μM, and hence under the conditions used the Cu(II) ion starts to occupy site 2 during the titration. Note, that in our publication,⁶⁵ we did not deduce the ^aKd for the high affinity site due to this problem and that when we tried to further reduce the Aβ concentration closer to the ^aKd of site 1, the fluorescence signal intensity dropped to a level where it was difficult to measure the Cu(II) quenching. Also our measurement of ^aKd by ITC in Hepes suffers from similar problems: (i) in order to get a measurable signal the Aβ concentration was 70 μM, which is very high for site 1 but optimal for site 2 (ITC should be best measured under conditions where the value of Ka multiplied by concentration of molecule is between 10–100); (ii) moreover, the titration isotherm was a superposition of the first and second binding sites of Cu(II) where site 1 had a smaller signal and was hence partially buried. The fact that the ^aKd values obtained by several groups with various techniques are quite different in Hepes buffer (at least when compared to Tris, see Table 2 and below), demonstrates the difficulty of such measurements.

Tris buffer . As Tris buffer is a stronger chelator of Cu(II) than Hepes, the apparent ^aKd of Cu(Aβ) in Tris is higher and hence more easy to measure accurately (see above). Indeed as listed in Table 2, the values obtained by different groups and techniques do match quite well (see also Fig. S1†).

So taken together, the measurements in Tris buffer are the most reliable. In order to obtain the conditional ^cKd in a 0.1 M ionic strength (which correspond to the stoichiometric Kd without buffer at pH 7.4 and 0.1 M NaCl) the measured apparent ^aKd has to be corrected for the coordination of the Cu(II) ion to the Tris buffer. However, the two recent general “reconciling” publications proposed two different correction factors, which explains why they obtained divergent conditional ^cKds (^cKd of 35 nM¹¹⁰ and 0.625 nM¹⁰⁹). In other words, this discrepancy is due to calculations and not to measurements. Indeed the stoichiometry and nature of the complex obtained between Cu(II) and Tris do not match between the different publications (see e.g. ref. 112 and references therein). Tougu et al. took only the Cu₁(Tris)₁ complex into account while Hatcher et al. accounted for a Cu₁(Tris)_1–4 complexes where the Cu(II) ion was bound to the amine functions of Tris. However, there is also evidence that Tris may bind through the alcoholate function and form only the Cu₁(Tris)₂ complex.¹¹² This last consideration might explain the divergence between the conditional ^cKd values calculated by Hatcher et al.¹⁰⁹ for the different Tris concentrations (see Table 2).

Competition experiments with glycine ligands. Experiments with Gly as competing ligand was initially reported by Symeet al.⁶⁴ The two groups that used quenching of the Tyr fluorescence by Cu(II) found similar results, i.e. half recovery of the fluorescence was obtained with 20 or 40 equivalents of Gly ligands and deduced a ^cKd of 100 nM by taken a stoichiometry of 1 : 1 into account.^64,69 Note that in our article (ref. 65) Gly was used to probe the second low affinity Cu(II) binding site and the titration does not provide results complete enough for measuring the first strong binding site. Very recently, Hatcher et al.¹⁰⁹ studied the competition between Gly and Aβ by measuring ITC of the addition of complex Cu(Gly)₂ to the Aβ peptide. They deduced a ^cKd of 0.6 nM. Moreover, they explained the discrepancy between their value and the 100 nM determined by quenching of the Tyr fluorescence by the fact that in the fluorescence study the stoichiometry of 1 : 2 for the complex between Cu and Gly was neglected. They showed that by taking into account the 1 : 2 Cu(Gly)₂ complex the data of ref. 64 and 69 give a ^cKd of about 0.4 nM.

Conclusions. (i) The measurements of different groups and/or by various techniques agree when measured in Tris buffer or by competition with Gly. They agree much less in Hepes, likely due to measurements at concentrations far from the ^aKd. Measurements in phosphate buffer was neglected here, due to the potential risk of Cu₃(PO₄)₂precipitation. (ii) Taking into account the two possible complexes of Cu(II) with Gly (i.e.Cu(Gly) and Cu(Gly)₂), and calculations of ref. 109 and 113, the measurements of ^cKd using competition with Gly are in the range of 0.1–1 nM. Competition with Gly might be the most accurate measurement so far. (iii) In the case of Tris, Hatcher et al. considered the four possible complexes Cu(Tris)_1–4 leading to a conditional ^cKd value of 0.1 to 630 pM, whereas Tougu et al. only considered the Cu(Tris) complex and estimated conditional ^cKd value between 12–91 nM. It is the ambiguities concerning the nature of the complex(es) formed between Cu(II) and Tris, that hampered a unifying conditional ^cKd.

Binding constants of Zn(II) to Aβ peptides

Table 3 gathers the measured ^aKd for Zn(II) to full length and truncated Aβ peptide. The reported values are between 1 and 300 μM. Very high values origin from Tyr fluorescence measurements, without them the values are in the 1–20 μM range. Indeed, the effect of Zn(II) on Tyr fluorescence are difficult to reproduce. Ricchelli et al.¹¹⁴ and Garzon-Rodriguez et al.⁶⁸ reported an increase in the fluorescence upon Zn(II) addition whereas Tougu et al.¹¹⁰ reported a decrease. We did experiments with the relatively soluble Aβ16 peptide and were not able to find a reproducible, robust effect of Zn(II) binding on Tyr fluorescence (unpublished). Indeed as pointed out by Tougu et al. the fluorescence change may be due to aggregation rather than Zn(II) binding.

Table 3 Dissociation constants reported for the stoichiometric coordination of Zn(II) to Aβ peptides

Aβ fragment	^aKd/μM	^cKd/nM	pH	Buffer/competing ligand	Experiments	Ref.
a As for the Cu(II) ion, the number without brackets corresponds to the stoichiometric, but lower affinity binding site. The higher affinity value given in brackets corresponds to a substoichiometric binding of Zn(II) to Aβ40 with an apparent ^aKd of ∼100 nM, which was determined by displacement assay with radioactive and cold Zn(II) binding to blotted peptide.^116 However, in a subsequent study also using blotted peptide, but on a different membrane, no evidence was found for a sub-μM binding but confirmation of a ^aKd of ∼5 μM for Aβ40 was given.^115
1–40/42	300/57		7.4	10 mM Tris	Tyr fl.	68
1–16/28/40	22/10/7		7.4	Hepes and Tris	ITC	102
Soluble 1–16/28/40/42		14/12/7/7	7.4	Hepes	Competition with Zincon	102
Aggregated 1–40/42		3/3	7.4	Hepes	Competition with Zincon	102
1–28	6.6		7.2	10 mM Hepes	Tyr fl. of Zn/Cu competition NMR	104
	1.2		7.2	10 mM PO4^−	Tyr fl. of Zn/Cu competition NMR	104
1–40	1.1		7.2	10 mM PO4^−	Tyr fl. of Zn/Cu competition NMR	104
1–40	60/184		7.4	10/100 mM Tris	Tyr fl.	110
	65		7.4	20 mM Hepes	Tyr fl.	110
1–40/42		2	7.4	Zincon	Competition with zincon after 30′ incubation	110
		>11	7.4	Zincon	Before incubation	110
1–40	3.2		7.4	10 mM Hepes or 10 mM Tris	Displacement assay with cold and radioactive Zn	115
	5 (0.1)^a		7.4	20 mM Tris	Displacement assay with cold and radioactive Zn	116
1–16/40/42	1–10		7.4	100 mM Tris	Tyr fl.	114

---

Other reported values of the apparent ^aKd are in the 1.0 to 20 μM range. Although different methods have been applied by different groups (Table 3), this range is relatively narrow, at least when compared to the range determined for the Cu(II) ion. The reasons for that is a lower ^aKd (μM range) that place the ideal Aβ concentration for direct titration experiments also in the μM range (see above) and a weaker binding constants of Zn(II) to Tris and Hepes, making the correction factor due to the Zn(II) binding to the buffer lower and hence the apparent ^aKd closer to the conditional ^cKd. This is supported by ITC measurements that show similar apparent ^aKds in Tris and Hepes buffers.¹⁰² They are also similar to the conditional ^cKd obtained by competition with zincon (Table 3). Nevertheless, there is perhaps scope for narrowing further down the conditional ^cKd, by using other Zn(II) chelators than zincon (that is slowly degrading with time) and considering interaction with buffer.

Conclusion. It is most likely that the apparent ^aKd of Zn-binding to Aβ is in the range of 1–20 μM.

Do the Kd values of Cu(Aβ) and Zn(Aβ) complexes change with the aggregation state ?

Data available on binding constants of Cu(II)⁶⁶ or Zn(II)^102,110 show that the affinity is roughly the same for both the soluble Aβ peptides and preformed Aβ fibrils. This is in line with the idea that the metal ion binding sites are located in a region of the peptide that is not directly involved in the fibrils assembly. Besides, a two to three times higher affinity has been reported for Zn(II) induced-aggregates compared to preformed fibrils and native soluble Aβ peptides.¹⁰² This result has been recently corroborated by the study of the time-dependent binding of Zn(II) to Aβ40 that showed a increase in the affinity when the Aβ40 sample is pre-incubated with Zn(II) prior to measurement, thus allowing aggregation to occur.¹¹⁰ In that case, an apparent dissociation constant of ≈2 μM was reported.

Biological relevance of the metal ions binding sites. We have previously mentioned that the affinity constants are prone to significant modifications due to the conditions of experiments (buffer, pH, etc.). A direct consequence is that these values are only indicative of what would be determined in vivo (making the following paragraph subject to discussion).

Concentrations of Cu and Zn(II) ions in amyloid plaques were determined to 400 μm and 1 mM, respectively,²⁸ while the one of the Aβ peptides is likely much higher (at least mM), but depends on the model of packing in the fibrils and between the fibrils.¹⁹ Another study proposed a metal ion : Aβ peptides ratio ranging from 1 : 200 to 1 : 1 for Zn(II) and 1 : 10 to 1 : 2000 for Cu.¹¹⁷ Hence, Cu and Zn ions are found in amyloid plaques with a substoichiometric coordination to Aβ peptides but high local concentrations. However, the Aβ peptides are not the only constituent of the plaques. One of the major issues is thus to know if the metal ions are able to bind to soluble Aβ peptides, i.e. if the free metal ion concentration is in the range of the dissociation constants that we reported in the previous paragraph.

The intracellular concentrations of free Cu(II) and Zn(II) ions are tightly controlled to be maintained at very low values,¹¹⁸ and it can be conjectured that this also holds for neurons and other brain cells. For instance, metallothioneins that are widely distributed in the brain at higher μM concentrations show very high affinity for Zn(II) and Cu(I) ions (10¹¹ M⁻¹ and 10¹⁸ M⁻¹, respectively).^119,120

In contrast, the extracellular conditions are different: the Zn(II) and Cu ion concentrations are very low in the majority of the brain regions¹²¹ but can be present at μM concentrations in certain brain tissues (hippocampus, amygdala, cortex). Indeed, in the case of the Zn(II) ion, difference in spatial repartition of the metal centre is due to a labile weakly bound Zn-pool stored in vesicles that can be released in the synaptic cleft upon excitation. The estimated maximal local Zn(II) concentration is 300 μM.^122,123 Thus, it seems reasonable that the Zn(II) ion binds to the Aβ peptides in the extracellular space, which is also the place where amyloid plaques deposit. Because the Zn-enriched brain regions are those most affected in AD, it has been proposed that synaptic Zn have a neurotoxic role.^124,125 There is evidence that, similar to Zn(II) ion, Cu ions can be released from vesicles of neurons during synaptic transmission,^126–128 and concentrations of 15 μM of extracellular Cu have been reported. The total Cu concentration in different regions of the hippocampus were recently measured.¹²⁹ Remarkable differences in concentrations ranging from 20 μM or below (the detection limit) up to 200 μM were found. Thus, it seems well possible that, as in the case of Zn(II) ion, a relatively labile Cu-pool exists in certain brain regions and would be accessible for coordination to the Aβ peptides.

Concluding remarks and future directions

Cu(II) vs. Zn(II) binding

In terms of the Kd, Cu(II) binding to Aβ is several order of magnitudes stronger than Zn(II) binding. In terms of the binding sites, it is clear that the two metal ions are at least partially bound to the same ligands. In the case of ligand set {Asp1(COO⁻), 3 His} (see Scheme 1b), Zn(II) and Cu(II) bind to the very same ligands, with perhaps an additional ligand for the Zn(II). In the case of ligand set {Asp1(NH₂, COO⁻), 2 His} (see Scheme 1a), the overlap is only partial. This means that Cu(II) is principally able to displace Zn(II) from the Zn(Aβ) complex. If this happens in the brain is not clear, but as in the plaques the concentration of Aβ is higher than those of Cu and Zn together, it is likely that the two metal ions are bound in their site.¹¹⁷ In this respect, it is important to note that the aggregation of Zn(Aβ) and Cu(Aβ) are very different. Zn(II) is able to strongly induce Aβ aggregation, but forms perhaps more amorphous than fibrilar aggregates.^130,131 In contrast, Cu(II) promotes Aβ aggregation less strongly but more fibrilar forms are observed.⁶⁶ It would be interesting to decipher what are the differences in metal binding responsible for the different aggregation behaviour, in particular because the structure of the aggregates may be directly linked to the neurotoxicity. An explanation for that would be that Zn(II) and Cu(II) do not bind to the same ligands, which would be the case when Cu(II) is bound to the {Asp1(NH₂, COO⁻), 2 His} set of ligands (see Scheme 1a). Another possibility is that Zn(II) is coordinated by an additional ligand, perhaps from another peptide molecule leading to Zn-bridged Aβ.

Chelator compounds as Alzheimer drug: what is the desired Kd ?

Aberrant interactions of Cu(II) and Zn(II) ions with Aβ potentate AD by participating in the aggregation process of Aβ and in the case of Cu in the generation of reactive oxygen species (ROS). The reduction of aberrant metal-protein interactions by MPAC (metal-protein attenuating compounds) has been proposed as potential therapeutic interventions in AD. Such chelators should be strong enough to compete with the metal-binding affinity of Aβ, but not that strong that they withdraw metal ions from essential metalloproteins. In the case of Cu(II) the conditional ^cKd of Aβ lies in the widest range of 10 pM to 100 nM. Thus a conditional ^cKd of about 1–10 pM should be sufficient to retrieve Cu(II) completely from Aβ. This ^cKd is likely still high enough to not compete with Cu(II) sites in enzymes. This is supported by the recently proposed conditional ^cKd of 1 pM for Cu(II) in the human serum albumin (HSA),¹³² which is only loaded with 1% by Cu(II). This means that HSA has a ^cKd which is not strong enough to retrieve Cu(II) from Cu(II)-proteins. Although this value is from the blood, the fact that HSA is quite abundant in the brain (μM range) and is able to cross the blood-brain barrier, suggests that the situation in the brain may be similar (see ref. 132 and references therein).

In the case of Zn(II), a chelator with a apparent ^aKd of about 100 nM seems to be reasonable (which would give about a estimated conditional ^cKd of 10 nM). This Kd value is not low enough to compete with Zn(II) bound to proteins (typically a Kd around 10 pM), and is similar to the low affinity binding site of metallothionein, which is not occupied under normal conditions (see ref.^133,134 and references therein).

There is also evidence in the literature that Cu(II) bound to Aβ produces ROS and Zn(II) binding suppresses ROS production and is hence beneficial. In that case, a chelator with a conditional ^cKd of 1 pM for Cu(II), but a ^cKd for Zn(II) above 1–10 μM would be required.

Cu(I) binding to Aβ

Very little is known concerning the Cu(I) binding to Aβ, although it is very likely to be biologically relevant, as redox cycling of Cu bound to Aβ may contribute to the oxidative stress in AD (see also above). The nature of the coordination sphere of Cu(I) and value of the Kd are important parameters to determine in the future. In particular, the reduction of aberrant metal-protein interactions by MPAC so far neglects the Cu(I) state, although a Cu(I) chelator may be neuroprotective as shown by the study of MT3.⁵³

Note added in proof

The calculated conditional Kd at pH = 7.4 of Aβ16 and Aβ28 based on the potentiometric data of ref. 61 are 0.2072 and 0.02442 nM, respectively, (personal communication by Bela Gyurcsik, University of Szeged) (see also Table 2). These values are in the range of the values determined by glycine competition, reinforcing that the latter was the most accurate measurement.

Acknowledgements

We would like to thank Pierre Dorlet (CEA Saclay, France), Wojciech Bal (Polish Academy of Sciences, Warsaw, Poland), Lian Hong and John Simon (Duke University, USA) for helpful discussions and the present and past group members without whose work this review would not have been possible. Laurent Sabater is acknowledged for producing Fig. 3. We thank Bela Gyurcsik, University of Szeged, for the calculation of the conditional Kd at pH 7.4 of Aβ16 and Aβ28.

References

1. D. J. Selkoe, Science, 2002, 298, 789–791.
2. Y.-H. Suh and F. Checler, Pharmacol. Rev., 2002, 54, 469–525.
3. M. P. Mattson, Nature, 2004, 430, 631–639.
4. G. G. Glenner and C. W. Wong, Biochem. Biophys. Res. Commun., 1984, 120, 885–890.
5. C. L. Masters, G. Simms, N. A. Weinman, G. Multhaup, B. L. McDonald and K. Beyreuther, Proc. Natl. Acad. Sci. U. S. A., 1985, 82, 4245–4249.
6. J. Kang, H. G. Lemaire, A. Unterbeck, J. M. Salbaum, C. L. Masters, K. H. Grzeschik, G. Multhaup, K. Beyreuther and B. Muller-Hill, Nature, 1987, 325, 733–736.
7. C. Haass, M. G. Schlossmacher, A. Y. Hung, C. Vigo-Pelfrey, A. Mellon, B. L. Ostaszewski, I. Lieberburg, E. H. Koo, D. Schenk, D. B. Teplow and D. J. Selkoe, Nature, 1992, 359, 322–325.
8. J. Hardy and D. J. Selkoe, Science, 2002, 297, 353–356.
9. D. J. Selkoe, Physiol. Rev., 2001, 81, 741–766.
10. W. L. Klein, W. B. Stine, Jr. and D. B. Teplow, Neurobiol. Aging, 2004, 25, 569–580.
11. C. Haass and D. J. Selkoe, Nat. Rev. Mol. Cell Biol., 2007, 8, 101–112.
12. L. Hou, H. Shao, Y. Zhang, H. Li, N. K. Menon, E. B. Neuhaus, J. M. Brewer, I. J. Byeon, D. G. Ray, M. P. Vitek, T. Iwashita, R. A. Makula, A. B. Przybyla and M. G. Zagorski, J. Am. Chem. Soc., 2004, 126, 1992–2005.
13. R. Riek, P. Guntert, H. Dobeli, B. Wipf and K. Wuthrich, Eur. J. Biochem., 2001, 268, 5930–5936.
14. C. Morgan, M. Colombres, M. T. Nunez and N. C. Inestrosa, Prog. Neurobiol., 2004, 74, 323–349.
15. H. Shao, S. Jao, K. Ma and M. G. Zagorski, J. Mol. Biol., 1999, 285, 755–773.
16. S. B. Malinchik, H. Inouye, K. E. Szumowski and D. A. Kirschner, Biophys. J., 1998, 74, 537–545.
17. L. P. Choo, D. L. Wetzel, W. C. Halliday, M. Jackson, S. M. LeVine and H. H. Mantsch, Biophys. J., 1996, 71, 1672–1679.
18. R. Tycko, Biochemistry, 2003, 42, 3151–3159.
19. R. Tycko, Q. Rev. Biophys., 2006, 39, 1–55.
20. C. M. Dobson, Nature, 2003, 426, 884–890.
21. F. Chiti and C. M. Dobson, Annu. Rev. Biochem., 2006, 75, 333–366.
22. L. M. Sayre, G. Perry and M. A. Smith, Curr. Opin. Chem. Biol., 1999, 3, 220–225.
23. G. Multhaup and C. L. Masters, Met. Ions Biol. Syst., 1999, 36, 365–387.
24. L. M. Sayre, P. I. Moreira, M. A. Smith and G. Perry, Ann Ist Super Sanita, 2005, 41, 143–164.
25. A. I. Bush, Trends Neurosci., 2003, 26, 207–214.
26. D. G. Smith, R. Cappai and K. J. Barnham, Biochim. Biophys. Acta, 2007, 1768, 1976–1990.
27. F. Li, N. Y. Calingasan, F. Yu, W. M. Mauck, M. Toidze, C. G. Almeida, R. H. Takahashi, G. A. Carlson, Flint M. Beal, M. T. Lin and G. K. Gouras, J. Neurochem., 2004, 89, 1308–12.
28. M. A. Lovell, J. D. Robertson, W. J. Teesdale, J. L. Campbell and W. R. Markesbery, J. Neurol. Sci., 1998, 158, 47–52.
29. L. M. Miller, Q. Wang, T. P. Telivala, R. J. Smith, A. Lanzirotti and J. Miklossy, J. Struct. Biol., 2006, 155, 30–37.
30. A. I. Bush, W. H. Pettingell, G. Multhaup, M. Paradis, J. P. Vonsattel, J. F. Gusella, K. Beyreuther, C. L. Masters and R. E. Tanzi, Science, 1994, 265, 1464–1467.
31. C. S. Atwood, R. D. Moir, X. Huang, R. C. Scarpa, N. M. Bacarra, D. M. Romano, M. A. Hartshorn, R. E. Tanzi and A. I. Bush, J. Biol. Chem., 1998, 273, 12817–12826.
32. P. W. Mantyh, J. R. Ghilardi, S. Rogers, E. DeMaster, C. J. Allen, E. R. Stimson and J. E. Maggio, J. Neurochem., 1993, 61, 1171–1174.
33. W. P. Esler, E. R. Stimson, J. M. Jennings, J. R. Ghilardi, P. W. Mantyh and J. E. Maggio, J. Neurochem., 1996, 66, 723–732.
34. R. A. Cherny, C. S. Atwood, M. E. Xilinas, D. N. Gray, W. D. Jones, C. A. McLean, K. J. Barnham, I. Volitakis, F. W. Fraser, Y. Kim, X. Huang, L. E. Goldstein, R. D. Moir, J. T. Lim, K. Beyreuther, H. Zheng, R. E. Tanzi, C. L. Masters and A. I. Bush, Neuron, 2001, 30, 665–676.
35. R. A. Cherny, J. T. Legg, C. A. McLean, D. P. Fairlie, X. Huang, C. S. Atwood, K. Beyreuther, R. E. Tanzi, C. L. Masters and A. I. Bush, J. Biol. Chem., 1999, 274, 23223–23228.
36. C. S. Atwood, R. C. Scarpa, X. Huang, R. D. Moir, W. D. Jones, D. P. Fairlie, R. E. Tanzi and A. I. Bush, J. Neurochem., 2000, 75, 1219–1233.
37. X. Huang, C. S. Atwood, R. D. Moir, M. A. Hartshorn, R. E. Tanzi and A. I. Bush, JBIC, J. Biol. Inorg. Chem., 2004, 9, 954–960.
38. E. House, J. Collingwood, A. Khan, O. Korchazkina, G. Berthon and C. Exley, J. Alzheimers Dis., 2004, 6, 291–301.
39. P. Sengupta, K. Garai, B. Sahoo, Y. Shi, D. J. E. Callaway and S. Maiti, Biochemistry, 2003, 42, 10506–10513.
40. M. A. Smith, P. L. Harris, L. M. Sayre and G. Perry, Proc. Natl. Acad. Sci. U. S. A., 1997, 94, 9866–9868.
41. C. Opazo, X. Huang, R. A. Cherny, R. D. Moir, A. E. Roher, A. R. White, R. Cappai, C. L. Masters, R. E. Tanzi, N. C. Inestrosa and A. I. Bush, J. Biol. Chem., 2002, 277, 40302–40308.
42. X. Huang, C. S. Atwood, M. A. Hartshorn, G. Multhaup, L. E. Goldstein, R. C. Scarpa, M. P. Cuajungco, D. N. Gray, J. Lim, R. D. Moir, R. E. Tanzi and A. I. Bush, Biochemistry, 1999, 38, 7609–7616.
43. L. Guilloreau, S. Combalbert, A. Sournia-Saquet, H. Marzaguil and P. Faller, ChemBioChem, 2007, 8, 1317–1325.
44. T. Kowalik-Jankowska, M. Ruta, K. Wisniewska, L. Lankiewicz and M. Dyba, J. Inorg. Biochem., 2004, 98, 940–950.
45. S. I. Dikalov, M. P. Vitek and R. P. Mason, Free Radical Biol. Med., 2004, 36, 340–347.
46. X. Huang, R. D. Moir, R. Tanzi, A. I. Bush and J. T. Rogers, Ann. N. Y. Acad. Sci., 2004, 1012, 153–163.
47. J. Zou, K. Kajita and N. Sugimoto, Angew. Chem., Int. Ed., 2001, 40, 2274–2277.
48. B. Raman, T. Ban, K. Yamaguchi, M. Sakai, T. Kawai, H. Naiki and Y. Goto, J. Biol. Chem., 2005, 280, 16157–16162.
49. A. I. Bush, Neurobiol. Aging, 2002, 23, 1031–1038.
50. C. J. Frederickson, Int. Rev. Neurobiol., 1989, 31, 145–238.
51. M. P. Cuajungco and K. Y. Faget, Brain Research Review, 2003, 41, 44–56.
52. Y. Yoshiike, K. Tanemura, O. Murayama, T. Akagi, M. Murayama, S. Sato, X. Sun, N. Tanaka and A. Takashima, J. Biol. Chem., 2001, 276, 32293–32299.
53. G. Meloni, V. Sonois, T. Delaine, L. Guilloreau, A. Gillet, J. Teissie, P. Faller and M. Vasak, Nat. Chem. Biol., 2008, 4, 366–372.
54. J.-Y. Lee, T. B. Cole, R. D. Palmiter, S. W. Suh and J.-Y. Koh, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 7705–7710.
55. A. L. Phinney, B. Drisaldi, S. D. Schmidt, S. Lugowski, V. Coronado, Y. Liang, P. Horne, J. Yang, J. Sekoulidis, J. Coomaraswamy, M. A. Chishti, D. W. Cox, P. M. Mathews, R. A. Nixon, G. A. Carlson, P. S. George-Hyslop and D. Westaway, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 14193–14198.
56. T. A. Bayer, S. Schafer, A. Simons, A. Kemmling, T. Kamer, R. Tepest, A. Eckert, K. Schussel, O. Eikenberg, C. Sturchler-Pierrat, D. Abramowski, M. Staufenbiel and G. Multhaup, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 14187–14192.
57. D. L. Sparks and B. G. Schreurs, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 11065–11069.
58. D. R. Brown and H. Kozlowski, Dalton Trans., 2004, 1907–1917.
59. P. S. Donnelly, Z. Xiao and A. G. Wedd, Curr. Opin. Chem. Biol., 2007, 11, 128–133.
60. E. Gaggelli, H. Kozlowski, D. Valensin and G. Valensin, Chem. Rev., 2006, 106, 1995–2044.
61. T. Kowalik-Jankowska, M. Ruta, K. Wisniewska and L. Lankiewicz, J. Inorg. Biochem., 2003, 95, 270–282.
62. J. W. Karr, H. Akintoye, L. J. Kaupp and V. A. Szalai, Biochemistry, 2005, 44, 5478–5487.
63. V. Minicozzi, F. Stellato, M. Comai, Dalla M. Serra, C. Potrich, W. Meyer-Klaucke and S. Morante, J. Biol. Chem., 2008, 283, 10784–10792.
64. C. D. Syme, R. C. Nadal, S. E. Rigby and J. H. Viles, J. Biol. Chem., 2004, 279, 18169–18177.
65. L. Guilloreau, L. Damian, Y. Coppel, H. Mazarguil, M. Winterhalter and P. Faller, JBIC, J. Biol. Inorg. Chem., 2006, 11, 1024–1038.
66. J. W. Karr and V. A. Szalai, Biochemistry, 2008, 47, 5006–5016.
67. D. Jiang, L. Man, J. Wang, Y. Zhang, S. Chickenyen, Y. Wang and F. Zhou, Biochemistry, 2007, 46, 9270–9282.
68. W. Garzon-Rodriguez, A. K. Yatsimirsky and C. G. Glabe, Bioorg. Med. Chem. Lett., 1999, 9, 2243–2248.
69. Q.-F. Ma, J. Hu, W.-H. Wu, H.-D. Liu, J.-T. Du, Y. Fu, Y.-W. Wu, P. Lei, Y.-F. Zhao and Y.-M. Li, Biopolymers, 2006, 83, 20–31.
70. J. W. Karr, L. J. Kaupp and V. A. Szalai, J. Am. Chem. Soc., 2004, 126, 13534–13538.
71. C. C. Curtain, F. Ali, I. Volitakis, R. A. Cherny, R. S. Norton, K. Beyreuther, C. J. Barrow, C. L. Masters, A. I. Bush and K. J. Barnham, J. Biol. Chem., 2001, 276, 20466–20473.
72. A. K. Tickler, D. Smith, G. D. Ciccotosto, D. J. Tew, C. C. Curtain, D. Carrington, C. L. Masters, A. I. Bush, R. A. Cherny, R. Cappai, J. D. Wade and K. J. Barnham, J. Biol. Chem., 2005, 280, 13355–13363.
73. D. P. Smith, D. G. Smith, C. C. Curtain, J. F. Boas, J. R. Pilbrow, G. D. Ciccotosto, T. L. Lau, D. J. Tew, K. Perez, J. D. Wade, A. I. Bush, S. C. Drew, F. Separovic, C. L. Masters, R. Cappai and K. J. Barnham, J. Biol. Chem., 2006, 281, 15145–15154.
74. D. L. Williams-Smith, R. C. Bray, M. J. Barber, A. D. Tsopanakis and S. P. Vincent, Biochem. J., 1977, 167, 593–600.
75. T. Miura, K. Suzuki, N. Kohata and H. Takeuchi, Biochemistry, 2000, 39, 7024–7031.
76. C. C. Curtain, F. E. Ali, D. G. Smith, A. I. Bush, C. L. Masters and K. J. Barnham, J. Biol. Chem., 2003, 278, 2977–2982.
77. J. W. Karr and V. A. Szalai, J. Am. Chem. Soc., 2007, 129, 3796–3797.
78. S. Jun and S. Saxena, Angew. Chem., Int. Ed., 2007, 46, 3959–3961.
79. J. Peisach and W. E. Blumberg, Arch. Biochem. Biophys., 1974, 165, 691–708.
80. S. Jun and S. Saxena, Angew. Chem., Int. Ed., 2007, 46, 5263.
81. X. Huang, M. P. Cuajungco, C. S. Atwood, M. A. Hartshorn, J. D. A. Tyndall, G. R. Hanson, K. C. Stokes, M. Leopold, G. Multhaup, L. E. Goldstein, R. C. Scarpa, A. J. Saunders, J. Lim, R. D. Moir, C. Glabe, E. F. Bowden, C. L. Masters, D. P. Fairlie, R. E. Tanzi and A. I. Bush, J. Biol. Chem., 1999, 274, 37111–37116.
82. F. Stellato, G. Menestrina, M. D. Serra, C. Potrich, R. Tomazzolli, W. Meyer-Klaucke and S. Morante, Eur. Biophys. J., 2006, 35, 340–351.
83. C. Hureau, L. Charlet, P. Dorlet, F. Gonnet, L. Spadini, E. Anxolabéhère-Mallart and J.-J. Girerd, JBIC, J. Biol. Inorg. Chem., 2006, 11, 735–744.
84. L. Hou and M. G. Zagorski, J. Am. Chem. Soc., 2006, 128, 9260–9261.
85. B.-k. Shin and S. Saxena, Biochemistry, 2008, 47, 9117–9123.
86. E. Gaggelli, Z. Grzonka, H. Kozlowski, C. Migliorini, E. Molteni, D. Valensin and G. Valensin, Chem. Commun., 2008, 341–343.
87. K. J. Barnham, F. Haeffner, G. D. Ciccotosto, C. C. Curtain, D. Tew, C. Mavros, K. Beyreuther, D. Carrington, C. L. Masters, R. A. Cherny, R. Cappai and A. I. Bush, FASEB J., 2004, 18, 1427–1429.
88. E. W. Ainscough, A. M. Brodie, J. E. Plowman, S. J. Bloor, J. S. Loehr and T. M. Loehr, Biochemistry, 1980, 19, 4072–4079.
89. W. Humphrey, A. Dalke and K. Schulten, J. Mol. Graphics, 1996, 14, 33–38.
90. http://www.ks.uiuc.edu/Research/vmd/ .
91. T. Kowalik-Jankowska, M. Ruta-Dolejsz, K. Wisniewska and L. Lankiewicz, J. Inorg. Biochem., 2001, 86, 535–545.
92. J. Dong, C. S. Atwood, V. E. Anderson, S. L. Siedlak, M. A. Smith, G. Perry and P. R. Carey, Biochemistry, 2003, 42, 2768–2773.
93. R. A. Himes, G. Y. Park, A. N. Barry, N. J. Blackburn and K. D. Karlin, J. Am. Chem. Soc., 2007, 129, 5352–5353.
94. D. F. Raffa, G. A. Rickard and A. Rauk, JBIC, J. Biol. Inorg. Chem., 2007, 12, 147–164.
95. A. Voronova, W. Meyer-Klaucke, T. Meyer, A. Rompel, B. Krebs, J. Kazantseva, R. Sillard and P. Palumaa, Biochem. J., 2007, 408, 139–148.
96. M. E. Rice, Trends Neurosci., 2000, 23, 209–216.
97. S. A. Kozin, S. Zirah, S. Rebuffat, G. H. Hoa and P. Debey, Biochem. Biophys. Res. Commun., 2001, 285, 959–964.
98. C. D. Syme and J. H. Viles, Biochim. Biophys. Acta, 2006, 1764, 246–256.
99. Y. Mekmouche, Y. Coppel, K. Hochgrafe, L. Guilloreau, C. Talmard, H. Mazarguil and P. Faller, ChemBioChem, 2005, 6, 1663–1671.
100. S. Zirah, S. A. Kozin, A. K. Mazur, A. Blond, M. Cheminant, I. Ségalas-Milazzo, P. Debey and S. Rebuffat, J. Biol. Chem., 2006, 281, 2151–2161.
101. C. Talmard, L. Guilloreau, Y. Coppel, H. Mazarguil and P. Faller, ChemBioChem, 2007, 8, 163–165.
102. C. Talmard, A. Bouzan and P. Faller, Biochemistry, 2007, 46, 13658–13666.
103. S. Zirah, S. Rebuffat, A. Kozin, P. Debey, F. Fournier, D. Lesage and J.-C. Tabet, Int. J. Mass Spectrom., 2003, 228, 999–1016.
104. J. Danielsson, R. Pierattelli, L. Banci and A. Graslund, FEBS J., 2007, 274, 46–59.
105. K. H. Lim, Y. K. Kim and Y.-T. Chang, Biochemistry, 2007, 46, 13523–13532.
106. S. T. Liu, G. Howlett and C. J. Barrow, Biochemistry, 1999, 38, 9373–9378.
107. D. S. Yang, J. McLaurin, K. Qin, D. Westaway and P. E. Fraser, Eur. J. Biochem., 2000, 267, 6692–6698.
108. E. Gaggelli, A. Janicka-Klos, E. Jankowska, H. Kozlowski, C. Migliorini, E. Molteni, D. Valensin, G. Valensin and E. Wieczerzak, J. Phys. Chem. B, 2008, 112, 100–109.
109. L. Q. Hatcher, L. Hong, W. D. Bush, T. Carducci and J. D. Simon, J. Phys. Chem. B, 2008, 112, 8160–8164.
110. V. Tõugu, A. Karafin and P. Palumaa, J. Neurochem., 2008, 104, 1249–1259.
111. M. Sokołowska and W. Bal, J. Inorg. Biochem., 2005, 99, 1653–1660.
112. S. Canepari, V. Carunchio and R. Schina, Polyhedron, 1999, 18, 3263–3267.
113. L. Hong, W. D. Bush, L. Q. Hatcher and J. Simon, J. Phys. Chem. B, 2008, 112, 604–611.
114. F. Ricchelli, D. Drago, B. Filippi, G. Tognon and P. Zatta, Cell. Mol. Life Sci., 2005, 62, 1724–1733.
115. A. Clements, D. Allsop, D. M. Walsh and C. H. Williams, J. Neurochem., 1996, 66, 740–747.
116. A. I. Bush, W. H. d. Pettingell, M. Paradis and R. E. Tanzi, J. Biol. Chem., 1994, 269, 12152–12158.
117. D. Beauchemin and R. Kisilevsky, Anal. Chem., 1998, 70, 1026–1029.
118. L. A. Finney and T. V. O'Halloran, Science, 2003, 300, 931–936.
119. D. W. Hasler, L. T. Jensen, O. Zerbe, D. R. Winge and M. Vasak, Biochemistry, 2000, 39, 14567–14575.
120. Z. Xiao, F. Loughlin, G. N. George, G. J. Howlett and A. G. Wedd, J. Am. Chem. Soc., 2004, 126, 3081–3090.
121. C. J. Frederickson, L. J. Giblin, A. Krezel, D. J. McAdoo, R. N. Muelle, Y. Zeng, R. V. Balaji, R. Masalha, R. B. Thompson, C. A. Fierke, J. M. de Sarvey, M. Valdenebro, D. S. Prough and M. H. Zornow, Exp. Neurol., 2006, 198, 285–293.
122. S. Y. Assaf and S. H. Chung, Nature, 1984, 308, 734–736.
123. G. A. Howell, M. G. Welch and C. J. Frederickson, Nature, 1984, 308, 736–738.
124. M. A. Deibel, W. D. Ehmann and W. R. Markesbery, J. Neurol. Sci., 1996, 143, 137–142.
125. P. A. Adlard and A. I. Bush, J. Alzheimers Dis., 2006, 10, 145–163.
126. D. E. Hartter and A. Barnea, Synapse (N. Y.), 1988, 2, 412–415.
127. J. Kardos, I. Kovács, F. Hajós, M. Kálmán and M. Simonyi, Neurosci. Lett., 1989, 103, 139–144.
128. M. L. Schlief, A. M. Craig and J. D. Gitlin, J. Neurosci., 2005, 25, 239–246.
129. J. S. Becker, M. V. Zoriy, C. Pickhardt, N. Palomero-Gallagher and K. Zilles, Anal. Chem., 2005, 77, 3208–3216.
130. D. Noy, I. Solomonov, O. Sinkevich, T. Arad, K. Kjaer and I. Sagi, J. Am. Chem. Soc., 2008, 130, 1376–1383.
131. K. Garai, B. Sahoo, S. K. Kaushalya, R. Desai and S. Maiti, Biochemistry, 2007, 46, 10655–10663.
132. M. Rozga, M. Sokolowska, A. M. Protas and W. Bal, JBIC, J. Biol. Inorg. Chem., 2007, 12, 913–918.
133. A. Krezel and W. Maret, J. Am. Chem. Soc., 2007, 129, 10911–10921.
134. A. Krezel and W. Maret, J. Biol. Chem., 2008, 13, 401–409.

---

Footnotes

† Electronic supplementary information (ESI) available: Isothermal Titration Calorimetry of Cu(Aβ16). See DOI: 10.1039/b813398k

‡ There is only one report that is in very large disagreement with all other values.³⁶ An apparent ^aKd of 6.3 aM was determined for Aβ42. However this Kd accounted for a substoichiometric binding site and in the same work a second and not substoichiometric binding site with ^aKd of 5 nM for Aβ42 was reported, which agreed with the other reports. In the present work, we consider the stoichiometric binding with a ^aKd in the low nM range as the significant one.

---