Copper Redox Cycling in the Prion Protein Depends Critically on Binding Mode

Abstract

The prion protein (PrP) takes up four to six equivalents of copper in its extended N-terminal domain, composed of the octarepeat (OR) segment (human sequence residues 60–91) and two mononuclear binding sites (at His96 and His111; also referred to as the non-OR region). The OR segment responds to specific copper concentrations by transitioning from a multi-His mode at low copper levels to a single-His, amide nitrogen mode at high levels (Chattopadhyay et al. J. Am. Chem. Soc., 127, 12647–12656, 2005). The specific function of PrP in healthy tissue is unclear, but numerous reports link copper uptake to a neuroprotective role that regulates cellular stress (Stevens et al. PLoS Pathogens, 5(4): e1000390, 2009). A current working hypothesis is that the high occupancy binding mode quenches copper’s inherent redox cycling, thus protecting against the production of reactive oxygen species from unregulated Fenton type reactions. Here, we directly test this hypothesis by performing detailed pH-dependent electrochemical measurements on both low and high occupancy copper binding modes. In contrast to the current belief, we find that the low occupancy mode completely quenches redox cycling, but high occupancy leads to the gentle production of hydrogen peroxide through a catalytic reduction of oxygen facilitated by the complex. These electrochemical findings are supported by independent kinetic measurements that probe for ascorbate usage and also peroxide production. Hydrogen peroxide production is also observed from a segment corresponding to the non-OR region. Collectively, these results overturn the current working hypothesis and suggest, instead, that the redox cycling of copper bound to PrP in the high occupancy mode is not quenched, but is regulated. The observed production of hydrogen peroxide suggests a mechanism that could explain PrP’s putative role in cellular signaling.

1. INTRODUCTION

The misfolding and aggregation of the prion protein (PrP), a membrane-bound glycoprotein present in the central nervous system (CNS) of mammalian and avian species, leads to the development of the prion diseases.^1,2 The human forms of prion diseases,³ including Creutzfeldt-Jakob disease and kuru, share similar neuropathologies to other more prevalent neurodegenerative disorders such as Alzheimer’s disease (AD) and Parkinson’s disease (PD).^4,5 PrP has been shown to interact with a number of divalent metal ions,^6,7 and with moderate to high affinity to Cu²⁺ and Zn^2+.7,8 The linkage between PrP and bioavailable Cu²⁺ is particularly well established by in vitro and in vivo⁹ studies and described in a number of recent reviews.^10–14 Although copper is an essential metal and exists in a number of proteins, free Cu²⁺ exhibits high cytotoxicity, stemming from its ability to initiate redox cycling, in turn generating reactive oxygen species (ROS) via the Harber-Weiss reaction:^15–18

$${\text{Cu}}^{2+}+{\text{H}}_2{\text{O}}_2\to {\text{Cu}}^{+}•{\text{O}}_2\text{H}+{\text{H}}^{+}$$ (1)

$${\text{Cu}}^{+}•{\text{O}}_2\text{H}+{\text{H}}_2{\text{O}}_2\to {\text{Cu}}^{+}+{\text{O}}_2+\text{OH}•+{\text{H}}_2\text{O}$$ (2)

$${\text{H}}^{+}+{\text{Cu}}^{+}+{\text{H}}_2{\text{O}}_2\to {\text{Cu}}^{2+}+\text{OH}•+{\text{H}}_2\text{O}$$ (3)

The highly toxic and reactive hydroxyl radical (OH•) can readily damage DNA and proteins and cause lipid peroxidation. Reaction (3) is also referred to as a Fenton-like reaction because of its similarity to the Fenton reaction with Fe²⁺ and H₂O₂ as the reactants.¹⁶

In biological milieu, free or bound Cu²⁺ could initiate another redox cycle, which is dependent on the thermodynamic properties of the reactants and the structure of the Cu²⁺ complexes. The cycle starts with the reduction of free or Cu²⁺ bound to a ligand (L) by a biological reductant. The resultant Cu⁺ can be subsequently oxidized by O₂, generating H₂O₂ as a product. Using ascorbic acid (AA, which exists as ascorbate at neutral pH and oxidizes to dehydroascorbate or DA) as a representative reductant, this cycle can be described as:^19–22

$$2\text{L}-{\text{Cu}}^{2+}(\text{free}{\text{Cu}}^{2+})+\text{AA}+{\text{H}}_2\text{O}=2\text{L}-{\text{Cu}}^{+}(\text{free}{\text{Cu}}^{+})+\text{DA}+2{\text{H}}^{+}$$ (4)

$$2\text{L}-{\text{Cu}}^{2+}(\text{free}{\text{Cu}}^{2+})+{\text{O}}_2+2{\text{H}}^{+}=2\text{L}-{\text{Cu}}^{2+}(\text{free}{\text{Cu}}^{2+})+{\text{H}}_2{\text{O}}_2$$ (5)

Notice H₂O₂ produced in reaction (5) can in turn react with any remaining free or bound Cu²⁺ through reactions (1)–(3) to produce OH•.

Given the abundance of oxygen^23,24 and biological reductants (e.g. AA and glutathione or GSH) in the CNS (e.g., cerebrospinal fluid (CSF) in particular), if a redox active metal were not properly regulated, the large quantity of ROS produced by the Cu²⁺-initiated redox cycling would likely lead to significant oxidative damage. On the basis that PrP is a Cu²⁺-binding protein,⁹ a major biological function of PrP has been hypothesized to complex Cu²⁺ to attenuate or buffer its redox activity so that ROS is prevented from forming, or converted into less harmful species.^11,12 A unique feature of PrP is that distinct Cu²⁺ binding modes are populated in response to different Cu²⁺ concentrations. PrP contains a highly conserved repeat of an octapeptide (OP) sequence within its N-terminal domain that constitutes its primary sites for Cu²⁺ uptake.^11,12 Human PrP has four tandem copies of the specific sequence PHGGGWGQ (referred to as the octarepeat¹¹ or OR within residues 60–91). In vitro studies have shown that when the concentration ratio between PrP and Cu²⁺ is larger than 1:1, four histidines (His) in the OR domain coordinate a single Cu²⁺ center forming the low Cu²⁺ occupancy binding mode OR–Cu²⁺ (Figure 1).²⁵ When the Cu²⁺ concentration rises to four or more equivalents per protein, each of the four OPs binds one Cu²⁺, giving rise to the so-called high occupancy or OR–Cu²⁺₄ (Figure 1).²⁵ In OR–Cu²⁺₄, each Cu²⁺ is ligated by one His, the amide nitrogens on the two adjacent and deprotonated glycines, and a glycine carbonyl.²⁵ The OR–Cu²⁺ and OR–Cu²⁺₄ binding modes take up Cu²⁺ with highly disparate dissociation constants (0.1 nM for the former and ~10 μM for the latter) and consequent negative cooperativity.^11,26 Immediately outside of the OR domain, the segment encompassing two His residues (in human PrP they are His96 and His111 and in mouse PrP His96 and His110) can also bind Cu²⁺. This segment is referred to as the non-OR Cu²⁺-binding site.^27–29 It is recently shown that when solution pH becomes sufficiently low (~6.0 or lower), the wild-type (WT) PrP binds a single Cu²⁺ in the OR–Cu²⁺ mode.³⁰ This is conceivable since His imidazoles have a lower pK_a than glycine amide nitrogens (even when they are bound to metal ions), and consequently the Gly amide nitrogens, once protonated, are the first to lose their metal-binding ability.¹¹

Figure 1.

Structures of OR–Cu²⁺ (low Cu²⁺ occupancy) and OR–Cu²⁺₄ (high Cu²⁺ occupancy).

The multiple binding modes between Cu²⁺ and PrP and their disparate affinities have long been linked to the various functions of PrP^11,12 in modulating or inhibiting Cu²⁺ redox cycling and oxidative stress under different conditions. For example, a recent analysis of insertional mutations in the OR domain linked early onset prion disease to hindered copper uptake and decreased population of the high occupancy mode.³¹ Delineation of the respective roles and/or functions of the distinct binding copper binding modes requires a clear understanding of their redox behaviors, which has been confounded by several contradictory studies. For example, Miura et al.³² and Ruiz et al.³³ reported that the tryptophan (Trp) residues in the OR domain appears to readily reduce Cu²⁺ to Cu⁺ at both physiological and acidic pH. It was proposed that the OR–Cu²⁺ mode may allow one Trp residue to be in close vicinity to the Cu²⁺ center for a facile Cu²⁺ reduction.³² However, many EPR studies have found that the Cu²⁺ centers bound by OR in both the low and high occupancies (even the transitory modes or stoichiometries such as OR–Cu²⁺₂ and OR–Cu²⁺₃) are stable.^25,26,34,35 Bonomo et al. determined the redox potential of the OP–Cu²⁺ complex, which contains only one OP but has an equivalent Cu²⁺ coordination to that in OR–Cu²⁺₄.³⁶ The surprisingly negative potential (~ −0.3 V vs. NHE) would silence the redox cycling of bound Cu²⁺ completely as it is lower (more negative) than those of many biological relevant reductants (e.g., AA and GSH). In other words, reaction (4) would be fully quenched. This contradicts the finding that OR–Cu²⁺₄ catalyzes the production of OH• from H₂O₂.³⁷ In addition, Nishikimi and coworkers³⁷ and Requena and coworkers³⁸ demonstrated that the Cu²⁺ centers in the OR domain are responsible for the oxidation of AA to DA. Owing to the above conflicting results, it has been difficult to rationalize cell-based and in vivo findings regarding the functions of PrP and its copper complexes.

A current working hypothesis, based on some of these published reports,^32,33,39 suggests that the high occupancy binding mode (OR–Cu²⁺₄) quenches the redox cycling of free Cu²⁺, thus protecting against the production of ROS.^11,39 In addition, this hypothesis does not consider the detailed cellular milieu in which the catalytic cycle represented by reactions (4) and (5) might be more relevant. Such a cycle can proceed only if (I) the redox potential of L–Cu²⁺ is higher than that of the biological reductant and (II) the redox potential of L–Cu⁺ is lower than that of the O₂/H₂O₂ couple. In this work, we determined redox potentials of OR–Cu²⁺, OR–Cu²⁺₄ and non-OR–Cu²⁺ under an anaerobic environment and compared their voltammetric characteristics to those observed in the presence of oxygen. We demonstrate that the redox cycling observed for these complexes can be reliably predicted from redox potentials and verified by measurements of the oxidation of AA (as a common reductant) and H₂O₂ generated from the catalytic reduction of O₂. Our finding clearly shows that redox cycling of Cu²⁺ is highly dependent on the binding mode of the PrP–Cu²⁺ complexes. Moreover, contrary to the current hypothesis, we find that OR–Cu²⁺ is the binding mode that completely quenches any copper redox cycling. OR-Cu²⁺₄ leads to the gentle and controlled production of H₂O₂, which may serve as a signaling molecule to initiate various cellular events. We also examined the redox reactions of these complexes at different pH to assess how redox activity may be modulated by PrP trafficking through early and late endosomes.

2. EXPERIMENTAL SECTION

2.1 Materials

The octapeptide PHGGGWGQ (OP), an OR-containing peptide PrP(23–28, 57–91), and the PrP(91–110) peptide representing the non-OR site in mouse PrP, were all prepared using solid-phase fluorenylmethoxycarbonyl (Fmoc) methods.^{34, 40} These peptides were acetylated at their N-termini and amidated at C-terminus. The crude products were purified by reverse-phase high-performance liquid chromatography (HPLC) and characterized by mass spectrometry. The α-synuclein (α-Syn) peptide, α-Syn(1–19), was also synthesized and purified similarly. Amyloid beta (Aβ) peptide Aβ(1–16) was purchased from American Peptide Co. Inc. (Sunnyvale, CA). Table 1 lists the five peptides used in this work.

Table 1.

Sequences of peptides used in this work

KKRPKPWGQ(PHGGGWGQ)4	PrP(23–28, 57–91) or OR
PHGGGWGQ	OP
GGGTHNQWNKLSKPKTNLKH	PrP(91–110) or non-OR
DAEFRHDSGYEVHHQK	Aβ(1–16)
MDVFMKGLSKAKEGVVAAA	α-Syn(1–19)

2.2 Electrochemical Measurements

Voltammetric measurements of OR–Cu²⁺, OR–Cu²⁺₄, and non-OR–Cu²⁺ were performed on a CHI 411 electrochemical workstation (CHI Instruments, Austin, TX) using a homemade plastic electrochemical cell. A glassy carbon disk (3 mm in diameter), a platinum wire, and a Ag/AgCl electrode served as the working, auxiliary, and reference electrodes, respectively. The electrolyte solution was a 10 mM phosphate buffer (pH 7.4) comprising 50 mM Na₂SO₄. Prior to each experiment, the glassy carbon electrode was successively polished with diamond pastes of 1 and 0.3 μm in diameter (Buehler, Lake Bluff, IL) and sonicated in deionized water. Voltammetric experiments under the oxygen-free condition were carried out in a glove box (Plas Lab, Lasing, MI) that had been thoroughly purged with and kept under N₂. The oxygen concentration in solution housed in this glove box was measured to be less than 0.05 ppm by a portable conductivity meter (Orion 3-Star Plus; Thermo Electron Corp., MA). Voltammetry was also performed in solutions purged with O₂ for a few minutes and the concentration of O₂ was maintained at ca. 8.5 ppm. In order to minimize free Cu²⁺ in solution and avoid its interference on the voltammograms of complexes formed between Cu²⁺ and the PrP peptides, [Cu²⁺]/[OR] was maintained at 0.9/1 for OR–Cu²⁺ and 3.6/1 for OR–Cu²⁺₄

2.3 Kinetic measurement

The AA oxidation rate was determined by monitoring the change in AA absorbance at 265 nm using a UV-vis spectrophotometer (Cary 100 Bio, Varian Inc., Palo Alto, CA). The absorbance values plotted have excluded those contributed by the individual peptides at 265 nm (measured in separate peptide solutions of same concentrations). The kinetic experiments were conducted at 25 °C in phosphate buffer, as described by our previous studies.²² To form exclusively OR–Cu²⁺, it is critical to maintain the [OR]/[Cu²⁺] ratio much higher than 1:1 so that the amount of free Cu²⁺ remaining in solution is negligible.

2.4 Detection of Hydrogen Peroxide

The H₂O₂-detection kit was purchased from Bioanalytical System Inc. (West Lafayette, IN)⁴¹ and the detection was conducted in a thin-layer electrochemical flow cell (Bioanalytical System Inc.) using a hydrogel-peoxidase-modified glassy carbon electrode by following our previously published protocol.⁴⁰ Briefly, 100 μM of a given peptide was mixed with Cu²⁺ (5 μM) for 10 min. The resultant mixture was then incubated with AA (200 μM) for 2 h. The sample mixture was diluted five-fold with phosphate buffer, prior to being injected into the electrochemical flow cell.

3. RESULTS

3.1 Redox behavior of Cu²⁺ complexes with PrP peptides in aerobic and anaerobic solutions

The cyclic voltammogram (CV) of OR in an oxygen-free solution reveals that it is redox inactive between −0.3 and 0.6 V vs. Ag/AgCl (dotted line curve in Figure 2A). In contrast, voltammograms of OR–Cu²⁺ exhibit a redox wave with the midpoint potential (E_1/2)⁴² at 0.126 V (black curve in Figure 2A). The potential value is about 0.05 V higher than that of a complex formed between a cyclic peptide (Gly-His)₄ and copper⁴³ for mimicking the putative role of superoxide dismutase-like activity of PrP.^{44, 45} It is also 0.06 V lower than the theoretically calculated redox potential of OR–Cu²⁺.⁴⁶ Detailed CV studies (Figure S1 and Table S1 in the Supporting Information) indicate that the cycling between the Cu²⁺ and Cu⁺ centers is quite reversible. This CV wave remains essentially unchanged when the solution is saturated with O₂ (red curve). Notice that the higher background of the CV at potential beyond −0.2 V is attributable to electrochemical reduction of dissolved O₂. Interestingly, the voltammetric behavior of OR–Cu²⁺₄ (Figure 2B) is different from that of OR–Cu²⁺ in the following aspects. First, in the absence of O₂, E_1/2 (black curve) of OR–Cu²⁺₄ was determined to be −0.025 V, a shift by 0.151 V in the negative direction with respect to that of OR–Cu²⁺. A small shoulder wave is discernable at ca. 0.075 V, which is 0.051 V more negative than the wave of OR–Cu²⁺. The two waves are more distinguishable in their differential pulse voltammorams (Figure S2). We assign the wave at −0.025 V to OR–Cu²⁺₄ and that at 0.051 V to the transitory binding modes in which the stoichiometry between OR and Cu²⁺ ranges between 1:2 and 1:3.²⁵ No Cu⁰ stripping peak (vide infra) was observed, suggesting that free Cu²⁺ in the mixture is minimal. Previously an EPR study conducted by one of us²⁵ has shown that OR–Cu²⁺₄ coexists with these less abundant transitory species. The second aspect is that, when O₂ is present in solution, the OR–Cu²⁺₄ CV becomes sigmoidal (i.e., the reduction peak is enhanced at the expense of the oxidation wave). Such a behavior was also observed for non-OR–Cu²⁺ (red curve in Figure 2C), albeit the latter exhibits a plateau at an even more negative potential. These sigmoidal CV waves have the same characteristics as that of free Cu²⁺ (red curve in Figure 2D), which is indicative of a catalytic process⁴² in which the electrogenerated Cu⁺ is rapidly reoxidized by O₂ in solution. Therefore, the Cu⁺ centers in both OR–Cu²⁺₄ and non-OR–Cu²⁺ are oxidizable by O₂, a direct consequence of the negative (cathodic) shifts of their redox potentials with respect to that of OR–Cu²⁺. In deaerated solution (Figure 2D), there is no oxygen to regenerate Cu⁺ reduced from free Cu²⁺ and consequently Cu⁺ is further reduced to Cu⁰ (since Cu⁺ is easier to reduce than Cu²⁺).⁴² Cu⁰ deposited onto the glassy carbon electrode can then be oxidized back to Cu²⁺ during the scan reversal. The sharp anodic peak is resulted from the stripping of the electrodeposited Cu⁰.⁴²

Figure 2.

Cyclic voltammograms of (A) OR–Cu²⁺, (B) OR–Cu²⁺₄, (C) non-OR–Cu²⁺ and (D) free Cu²⁺ in N₂-saturated (black curve) and O₂-purged solutions (red curve), respectively. The Cu²⁺ concentration in all cases was 90 μM, while the OR and non-OR concentrations were 100, 25, and 100 μM in panels (A), (B) and (C), respectively. The dotted line curve in (A) corresponds to the CV of OR. The scan rate was 5 mV/s.

Another point worth noting is that the shift of the plateau of the red curve in Figure 2C relative to that in Figure 2B is originated from the more negative redox potential of non-OR–Cu²⁺. In the absence of O₂ (black curve in Figure 2C), the non-OR–Cu²⁺ CV is less reversible, as evidenced by a smaller anodic peak compared to its cathodic counterpart and a wider separation in potential between the two peaks than that in the CV of OR–Cu²⁺₄ (cf. more details in Table S2).

3.2 pH-dependent redox behavior of PrP–Cu²⁺ complexes

An interesting property of PrP and its peptidic variants is that their Cu²⁺-binding is sensitive to solution pH.^11,12,30,32 At values lower than physiological pH (e.g., ~6.5 or lower),³⁰ the amide nitrogens become protonated, disfavoring the OR–Cu²⁺₄ binding mode. However, for OR–Cu²⁺, an even lower pH is required to release the only Cu²⁺ ion, owing to the fact that Cu²⁺ is strongly coordinated by the four His residues in the OR domain (cf. Figure 1). We therefore examined the voltammetric responses of OR–Cu²⁺, OR–Cu²⁺₄, and non-OR–Cu²⁺ at different pH values and also studied the effects of O₂. Panels (A) and (B) in Figure 3 are overlaid voltammograms of OR–Cu²⁺ and OR–Cu²⁺₄ in deaerated solutions at representative pH values, respectively. As can be seen, CVs of OR–Cu²⁺ are essentially congruent between pH 7.4 and 6.5 and only till pH 6.0 does the CV begin to show a drastic change. At pH 6.0 and lower, the CV is analogous to that of free Cu²⁺ under anaerobic condition (cf. the black curve in Figure 2D) and peaks associated with OR–Cu²⁺ are no longer observable. For OR–Cu²⁺₄, the Cu²⁺ release occurs at a higher pH (6.5), consistent with the abovementioned fact that protonation of the Gly amide nitrogens destabilizes the OR–Cu²⁺₄ complex. A close examination of the CVs collected at pH 7.4 (black curve) and 7.0 (red curve) reveals that even with such a small pH variation, some Cu²⁺ release appears to have taken place. This is reflected by the small increase of the wave at around 0.051 V (red curve), which is given rise by the transitory species such as OR–Cu²⁺₂ and/or OR–Cu²⁺₃. As mentioned above, these species are better resolved in their differential pulse voltammograms (Figure S2). Thus, as soon as some Cu²⁺ ions are released from OR–Cu²⁺₄, OR might have rearranged to bind the remaining Cu²⁺ ions with the available His residue(s). As for non-OR–Cu²⁺, a 0.4-unit pH decrease causes the redox wave to shift by 0.01 V in the anodic direction (cf. the red and black curves in Figure 3C).

Figure 3.

Cyclic voltammograms of (A) OR–Cu²⁺, (B) OR–Cu²⁺₄, and (C) non-OR–Cu²⁺ at different pH values. In panel (A) the curves correspond to pH 7.4 (black), 6.5 (red) and 6.0 (blue) and in panels (B) and (C) the curves correspond to 7.4 (black), 7.0 (red) and 6.0 (blue). The concentrations of Cu²⁺ and peptides used are the same as those in Figure 2. Panel (D) contrasts the voltammograms of OR–Cu²⁺ at pH 5.5 in the presence (red curve) and absence (black curve) of O₂.

To verify that in aerated solution Cu²⁺ released from the complexes behaves the same as free Cu²⁺ in ligand-free solutions, we also collected CVs of the three complexes at lower pH values. Figure 3D compares the CVs of OR–Cu²⁺ in deaerated (black) and aerated (red) solutions at pH 5.5. The trend is essentially the same as that of free Cu²⁺ (cf. Figure 2D). OR–Cu²⁺₄ or non-OR–Cu²⁺ also displayed very similar behaviors at higher pH (data not show). Thus Cu²⁺ ions, once released from these complexes, are capable of initiating redox cycles. The relevance of our data to the hypothesis that PrP acts as a Cu²⁺ transporter via endocytosis is described in the Discussion section.

3.3 Relationship between the Cu²⁺ binding mode and the H₂O₂ production

Through voltammetric studies of Cu²⁺ complexes with amyloid β (Aβ) peptides,⁴⁷ which are linked to the neuropathology of AD, and α-synuclein (α-syn),⁴⁰ which is believed to lead to the development of PD, we have demonstrated that H₂O₂ is indeed a product generated through reactions (4) and (5).^40,47 Therefore we are interested in probing whether H₂O₂ can be generated by one or more of the PrP–Cu²⁺ complexes. In Table 2, we contrast the redox potentials of OR–Cu²⁺, OR–Cu²⁺₄, and non-OR–Cu²⁺ to those of the O₂/H₂O₂ and AA/DA couples. It is evident that, from the thermodynamic viewpoint O₂ is incapable of oxidizing OR–Cu⁺. In other words, the single Cu⁺ center is so “stabilized” by OR that O₂ cannot turn it over to Cu²⁺ (i.e., reaction (5) would not occur). In contrast, both the OR–Cu²⁺₄ (high occupancy) and non-OR–Cu²⁺ modes have lower redox potentials than the O₂/H₂O₂ couple and therefore are expected to behave similarly to the Cu²⁺ complexes of Aβ and α-syn to facilitate the catalytic O₂ reduction.

Table 2.

Redox potentials of Cu²⁺ complexes with select amyloidogenic molecules and biological redox couples.

Species	E1/2 or E0 vs. Ag/AgCl	vs. NHE	References
OR–Cu2+/OR–Cu+	0.126 V	0.323 V	This work
O2/H2O2	0.099	0.296	48
Aβ–Cu2+/Aβ–Cu+	0.080	0.277	47
PrP–Cu52+/PrP–Cu5+	0.070	0.267	30
α-Syn–Cu2+/α-Syn–Cu+	0.018	0.215	40
OR–Cu2+ 4/OR–Cu+4	−0.025	0.172	This work
Free Cu2+/Cu+	−0.040	0.157	42
non-OR–Cu2+/non-OR–Cu+	−0.087*	0.110*	This work
AA/DA	−0.145	0.052	49
GSH/GSSG	−0.425	−0.228	50

^*Accuracy is affected by the less reversible redox wave of non-OR–Cu²⁺.⁴²

The above thermodynamic prediction was verified by two experiments. First, we determined the kinetics of AA oxidation by O₂ in the presence of OR–Cu²⁺, OR–Cu²⁺₄, or non-OR–Cu²⁺ by monitoring the change of AA absorbance at 265 nm. As shown by the green curve in Figure 4A, the rate of AA autooxidation was found to be exceedingly slow (0.051 ± 0.004 nM s⁻¹). The rate did not change with the addition of 5 μM OR–Cu²⁺ (0.048 ± 0.004 nM s⁻¹), but AA becomes rapidly consumed when the same amount of OR–Cu²⁺₄ (42.2 ± 4.7 nM s⁻¹), non-OR–Cu²⁺ (34.8 ± 3.7 nM s⁻¹) or free Cu²⁺ (191 ± 17.2 nM s⁻¹) is present. However, AA can be instantaneously and significantly oxidized by OR–Cu²⁺ at a concentration stoichiometrically more comparable to that of AA. Notice that the absorbance of a 200 μM AA solution decreased by ~28% immediately upon being mixed with 100 μM OR–Cu²⁺ (magenta curve). The oxidation of AA is quantitative, since such a decrease is close to the theoretical value (~25%). This observation is also confirmed by a separate differential pulse voltammetric experiment (Figure S3). Thus, the inability of O₂ to oxidize OR–Cu⁺ results in a halt of the Cu²⁺ redox cycling, preventing additional AA from being continuously oxidized. In the second experiment, we demonstrated that (1) H₂O₂ is indeed a product of the redox cycle and (2) the amount of H₂O₂ produced can be correlated to the redox potential of the Cu²⁺ complexes with select peptides. Specifically, using a commercial electrochemical H₂O₂ detection kit,⁴⁰ we found that after 2 h of reaction the amount of H₂O₂ generated in the presence of OR–Cu²⁺ is essentially negligible, whereas H₂O₂ was detected in the presence of other species. The amount of H₂O₂ increases in the order of Aβ(1–16)–Cu²⁺ < α-syn(1–19)–Cu²⁺ < non-OR–Cu²⁺ < OR–Cu²⁺₄ (Figure 4B), which is largely consistent with the order of the redox potentials of the respective complexes (Table 2). It is worth noting that we chose the Cu²⁺-binding peptidic segments of the full-length Aβ peptide and the α-syn protein, because the resultant complexes are more comparable to OR–Cu²⁺ or OR–Cu²⁺₄ in terms of their sizes and O₂ accessibility to the copper center.⁴⁷

Figure 4.

(A) Change of AA (200 μM) absorbance as a function of reaction time in the absence (green curve) and presence of different Cu²⁺-containing species: OR–Cu²⁺ (red curve), non-OR–Cu²⁺ (cyan), OR–Cu²⁺₄ (blue), and free Cu²⁺ (black). The magenta curve corresponds to the AA absorbance variation recorded after the addition of 100 μM OR–Cu²⁺ to 200 μM AA solution (B) Amounts of H₂O₂ generated by OR–Cu²⁺, Aβ(1–16)–Cu²⁺, α-syn(1–19)–Cu²⁺, OR–Cu²⁺₄, and non-OR–Cu²⁺. [Cu²⁺] was 5 μM, and concentrations of the peptide molecules were all 100 μM except for the solution of OR–Cu²⁺₄ wherein [OR] = 1.25 μM.

3.4 Time and pH dependences of the H₂O₂ production

At different solution pH, we also monitored the production of H₂O₂ as a function of time. Again, OR–Cu²⁺ generates little H₂O₂ (even over a period of 2 h as shown by Figure 5A). In contrast, H₂O₂ generated by OR–Cu²⁺₄ or non-OR–Cu²⁺ increases gradually with time. However, the amounts of H₂O₂ produced by these complexes are substantially lower than that by free Cu²⁺, especially at the beginning of the redox cycle (magenta curve in Figure 5A). We also measured the amount of H₂O₂ generated by the Cu²⁺-glutamine complex, on the basis that glutamine is the most abundant amino acid in CSF (more than 10 times higher than other amino acids and also albumin, a major Cu²⁺-transport protein²⁴). Using 10^12.5 for the formation constant of the Cu(Gln)₂ complex⁵¹ and adjusted to pH 7.0, the Cu(Gln)₂ complex has an affinity constant of 0.32 μM, which is between those of OR–Cu²⁺ (0.1 nM) and OR–Cu²⁺₄ (7.0–12.0 μM)²⁶. That much more H₂O₂ was generated by the Cu(Gln)₂ complex (see the green curve in Figure 5A) suggests that, in addition to the binding affinity, other structural factors (e.g., accessibility to the Cu²⁺ center(s) by cellular reductants and O₂) are also important. Thus, OR–Cu²⁺ completely quenches the Cu²⁺ redox cycling, while OR–Cu²⁺₄ and non-OR–Cu²⁺ are capable of reducing the amount of H₂O₂ production.

Figure 5.

(A) Time-dependence of H₂O₂ generation in different solutions: OR–Cu²⁺ (black), OR–Cu²⁺₄ (red), non-OR–Cu²⁺ (blue), Cu²⁺/glutamine (green) and free Cu²⁺ (magenta). All solutions contained 5 μM Cu²⁺ and 200 μM AA and the ligand concentration was 100 μM except for the solution of OR–Cu²⁺₄ wherein [OR] = 1.25 μM. (B) Variations of [H₂O₂] generated by OR–Cu²⁺ in solutions of different pH values: 7.4 (black), 6.5 (red), 6.0 (green) and 5.5 (blue). Each data point is the average of three replicates and the error bars are the standard deviations.

The initial sharp rises of the blue and green curves in Figure 5A can be attributed to the rapid production of H₂O₂ via reactions (4) and (5) by free Cu²⁺ or the glutamine-Cu²⁺ complex. The more gradual decay is a convoluted result of the continuous H₂O₂ production via this redox cycle and the decomposition of H₂O₂ by free Cu²⁺ or the glutamine-Cu²⁺ (Glu–Cu²⁺) complex via reactions (1)–(3). Thus, as H₂O₂ builds up in the solution, reactions (1)–(3) will accelerate to decrease the amount of H₂O₂. However, in this case the more reactive and pernicious hydroxyl radicals are formed. The biological implication of the H₂O₂ generation as well as the complete inhibition of the H₂O₂ and hydroxyl radical productions by these Cu²⁺ complexes will be presented in the Discussion section.

Finally, we determined the time dependence of H₂O₂ production by the Cu²⁺ complexes at different solution pH. Figure 5B displays representative results measured and plotted for OR–Cu²⁺ at pH 7.4, 6.5, 6.0 and 5.5. When Cu²⁺ is bound, no (for OR–Cu²⁺) or a small amount of H₂O₂ (for OR–Cu²⁺ and non-OR–Cu²⁺; data not shown) was produced. However, as the solution pH is lowered, all complexes exhibit the same trend as depicted by the green and blue curves in Figure 5B, suggesting that Cu²⁺ released from these complexes becomes predominant in producing H₂O₂ and possibly hydroxyl radicals. Thus, at physiological pH, all the binding domains possess the ability of preventing a large amount of H₂O₂ from being produced and completely annihilating the possibility of hydroxyl radical generation. This aspect will be reiterated below in connection with the discussion of the possible functions of PrP.

4. DISCUSSION

Using carefully controlled anaerobic and aerated solutions, we conducted a systematic investigation on the electrochemical behaviors of OR–Cu²⁺, OR–Cu²⁺₄, and non-OR–Cu²⁺. Comparison of the respective redox potentials of these complexes to those of the O₂/H₂O₂ and AA/DA couples (cf. Table 2) clearly indicates that, thermodynamically, the Cu²⁺ centers in these complexes can all be reduced by common cellular reductants (e.g., AA and GSH). However, the occurrence of the Cu²⁺ redox cycle described by reactions (4)–(5) to generate H₂O₂ also requires the Cu⁺ centers be oxidizable by O₂. Based on the shifts in the reduction potentials of the complexes with respect to that of free Cu²⁺, we determined that in the OR–copper complex (low occupancy), binding of Cu⁺ by OR is about three orders of magnitude stronger than that of Cu²⁺ (Supporting Information). Thus, by significantly shifting the redox potential of the bound Cu²⁺ in the anodic direction, the low occupancy binding mode completely inhibits the H₂O₂ generation. In contrast, binding affinities of Cu²⁺ and Cu⁺ in the high occupancy mode are comparable. We found that non-OR binds Cu⁺ about 4.7 times less strongly than Cu²⁺ (Supporting Information). As a consequence, non-OR–Cu²⁺ can also facilitate the H₂O₂ generation. The reported redox potentials of complexes of shorter non-OR peptides with copper are 0.24 V for the GGGTH–Cu²⁺ complex²⁹ and 0.32 V for the GGGTHSQW–Cu²⁺ complex⁵², both of which are much higher than 0.110 V we measured for non-OR–Cu²⁺. From the thermodynamic viewpoint, GGGTHSQW–Cu²⁺ would not be capable of facilitating the redox cycling of Cu²⁺ if its redox potential were 0.32 V. The results from both the UV-vis spectrophotometric measurements of AA oxidation and electrochemical quantification of H₂O₂ (Figure 4) are fully consistent with our thermodynamic reasoning. This firmly establishes a correlation between a specific Cu²⁺ binding mode in PrP and the protein’s ability to ameliorate oxidative stress. It is worth mentioning that the properties of the complexes not only depend on what and how many amino acid residues are involved in the binding, but also are governed by the environment and steric hindrance under which the residues are bound to the Cu²⁺ center. We noticed that the complex formed between (Gly-His)₄ and Cu²⁺ has a potential that is 0.05 V more negative than OR–Cu²⁺.⁴³ Such a difference is expected to make this complex behave similarly to the Aβ–Cu²⁺ complex (cf. Table 2), leading to the Cu²⁺ redox cycling to produce H₂O₂. It is also interesting to note that the redox potential of OR–Cu²⁺ (0.323 V vs. NHE; cf. Table 2) is very close to that of superoxide dismutase or SOD (0.32 V).⁴³ While this may be used to argue for the SOD-like behavior of the PrP–Cu²⁺ complex,⁴⁴ we should point out that the reported “SOD-like activity” of PrP–Cu²⁺ is at least two orders of magnitude smaller than that of native SOD.⁴⁵ In addition, there are other large differences in structures and properties between the PrP–Cu²⁺ complex and native SOD that make the correlation of the PrP–Cu²⁺ complex to SOD questionable.^11,12

The significance of our findings is three fold. First, with a better understanding of the redox behavior of the PrP–Cu²⁺ complexes, several major discrepancies in the literature may now be clarified. As mentioned in the introduction, the current paradigm suggesting that OR–Cu²⁺₄, instead of OR–Cu²⁺, is the more effective binding mode for quenching Cu²⁺ redox cycling now becomes questionable.^11,12 Our new paradigm provides theoretical supports for the work of Shiraishi et al.³⁷ and Requena and coworkers³⁸ in that AA can be directly oxidized by both OR–Cu²⁺₄ and OR–Cu²⁺, but only the latter of which completely quenches the copper redox cycling. Regarding the reduction of Cu²⁺ by Trp inherent in the OR domain of PrP, we believe that the use of bathocuproine (BC) to assay the redox activity of PrP-copper complexes can be problematic. Despite the cautions advocated by Sayre⁵³ and Ivanov et al.⁵⁴, BC and its derivatives are still widely used for studying reduction of Cu²⁺ bound by biomolecules to Cu⁺. With a strong binding affinity, BC can seize Cu²⁺ from certain biomolecules and bind it in the form of (BC)₂–Cu²⁺, which has a much higher oxidation potential than free Cu²⁺ and many other Cu²⁺ complexes.⁵³ Consequently, when oxidizable ligands are present in the system (e.g., His, Tyr, Trp, and Cys residues on protein molecules), Cu⁺ is produced and bound by BC, with the simultaneous oxidation of these redox-active moieties. Another important issue is that the Cu²⁺ redox cycling (or lack of) shown by reactions (4) and (5) is not only dependent on the conversion of Cu²⁺ to Cu⁺ by an endogenous reductant, but also governed by the requirement that Cu⁺ remaining bound to PrP must then be oxidized back by O₂ to Cu²⁺. Thus, to simply ascribe the absence or presence of a redox cycle to the “stabilization” of Cu²⁺ or Cu⁺ might not be the most accurate interpretation on the role of PrP in inhibiting Cu²⁺-induced ROS production. Finally, OR–Cu²⁺ is also more effective in inhibiting ROS production than the Aβ–Cu²⁺ and α-syn–Cu²⁺ complexes. High levels of Cu²⁺ have been found to be complexed by Aβ peptides in senile plaques⁵⁵ and by α-syn in CSFs of patients⁵⁶. Such observations have been used to suggest the pro- or anti-oxidation functions of these amyloidogenic species.^17,55–59

The second outcome of our study is that the voltammetric responses of the three PrP–Cu²⁺ complexes each exhibit a different dependence on solution pH. The pH sensitivity of PrP–Cu²⁺ and the demonstration that high copper levels (e.g., 100 μM⁶⁰) stimulate PrP endocytosis have been used to advance the hypothesis that PrP may function as a copper transporter. Once becoming internalized and incorporated by endosomes, Cu²⁺ could be released from the complex due to the acidic environment of the endosome.^11,32 Although this copper-transport model is still under debate^11,12 and the fate of internalized PrP or the bound and released Cu²⁺ is not known, our results suggest that PrP with Cu²⁺ bound in different modes may undergo different endocytotic processes. It is well established that pH is only slightly acidic (~6.3–6.8) in the tubular extension of the early endosome, whereas it decreases from 6.2 to ~5.5 in the lumen of multi-vesicular bodies.⁶¹ The tubular extension is mainly responsible for the recycling of proteins, while the multi-vesicular bodies, a feature characteristic of the late endosome, are involved in sorting its cargo to the degradation pathway.⁶² PrP fully loaded with Cu²⁺, once encountering the slightly acidic environment of the early endosome, can release most of its Cu²⁺ complement. In contrast, Cu²⁺ is strongly withheld by OR in the OR–Cu²⁺ mode, even at the much lower pH inside the multi-vesicular bodies. Indeed, Brown and colleagues recently showed that at pH 5.5 or even lower, a single Cu²⁺ remains bound to the WT PrP in the OR domain. At a higher pH (~6.0), non-OR becomes capable of binding Cu²⁺ and the incorporation of additional Cu²⁺ ions into the OR domain occurs only after the solution pH has approached to the physiological value.³⁰ Based on our pH-dependence study (Figure 3), we hypothesize that Cu²⁺ released from the fully loaded PrP, through either subsequent complexation with other copper-binding proteins or generation of H₂O₂, may trigger the accumulation of PrP in the tubular extension of the early endosome, which ultimately leads to recycling of PrP back to plasma membrane. As for PrP containing only one Cu²⁺, it could remain bound to PrP even in the late endosome, which is eventually fused with the lysosome for the final protein and metal degradation.⁶²

Finally, we note that the distinct electrochemical features associated with each binding mode may provide insight into PrP function. PrP has been suggested to serve as a neuroprotectant by scavenging weakly complexed Cu²⁺ in vivo and thereby alleviating oxidative stress that might be induced by free Cu²⁺.^31,63 The gradual generation of H₂O₂ by either OR–Cu²⁺₄ or non-OR–Cu²⁺ does not contradict the proposed neuroprotecting function of PrP. As can be seen from Figure 5, the amount of H₂O₂ generated by OR–Cu²⁺₄ or non-OR–Cu²⁺ is much lower than those generated by free Cu²⁺ and the Cu²⁺-glutamine complex (glutamine is the dominant amino acid in CSF), especially at the beginning of the redox cycle. Although H₂O₂ as a ROS could also inflict oxidative stress/damage, we note that in vivo H₂O₂ is readily decomposed by enzymes such as catalase and GSH peroxidase or reacted away with antioxidants such as AA and GSH. Therefore a harmful buildup of H₂O₂ given rise by the PrP–Cu²⁺ complexes is unlikely. It was estimated that up to 250 μM H₂O₂ could be produced within brain neuropil every minute.⁶⁴ The H₂O₂ concentration depicted in Figure 5A (sub-μM levels produced by AA at a physiologically relevant concentration at the beginning) is only a small fraction of the known endogenous H₂O₂ concentration in the CNS. Therefore, the capacity of brain homeostasis of endogenous H₂O₂ appears to be more than sufficient to reduce any adverse effects that might be caused by H₂O₂ generated by the PrP–Cu²⁺ complexes. More importantly, our observation of the in vitro production of H₂O₂ is supported by the cell culture work conducted by Nishimura et al.⁶⁵ In their work, the PrP-deficient neurons were found to be more susceptible to Cu²⁺ toxicity than normal neurons. In addition, in the presence of Cu²⁺, more intracellular H₂O₂ was detected in the media of normal neurons and the viability of these neurons is not greatly impacted. This is consistent with the common belief that H₂O₂ is much less potent than the oxygen-containing radicals (e.g., OH•) and does not readily cause oxidative damage.^66–68 In fact, Rice and coworkers did not detect lipid peroxidation in cell membranes when dopaminergic neurons were exposed to high dosage of H₂O₂.⁶⁴ Taken together, as summarized in Figure 6, we propose that H₂O₂ generated at the PrP-residing membrane could act as an important signaling molecule. There is a growing body of evidence demonstrating that H₂O₂ is a signaling messenger capable of modulating synaptic activities and triggering a variety of cellular events.^23,67–70 Indeed, Herms et al. have reported that 0.001% exogenous H₂O₂ enhances inhibitory synaptic activity in wild-type mouse Purkinje cells.⁷¹ More relevantly, a prominent mechanism for H₂O₂-triggered cell signaling involves diffusion of H₂O₂ from extracellular space to cytosol to enhance protein tyrosine phosphorylation via simultaneous inactivation of tyrosine phosphotases and activation of tyrosine kinases.⁶⁹ In support of this mechanism, we note that Mouillet-Richard and coworkers suggested a role for PrP in signal transduction that might be facilitated by coupling of the membrane-anchored PrP to the intracellular tyrosine kinase Fyn through an unidentified signaling molecule.⁷²

Figure 6.

Schematic representation of the possible roles of PrP–Cu²⁺ complexes in quenching the Cu²⁺ redox cycling or gradual production of H₂O₂ for signal transduction. PrP is tethered to cell membrane via the glycosylphosphatidylinositol (GPI) anchor (green) with its α-helices in the C terminus shown in orange, N-linked carbohydrates in purple, and the OR and non-OR domains near the N-terminus depicted in white. When [Cu²⁺] is at a low level (nM or lower) and at pH ranging between 5.5 and 7.4, Cu²⁺ (blue sphere) remains bound in the OR–Cu²⁺ mode (left), quenching the Cu²⁺ redox cycling. At higher [Cu²⁺] (μM) and a pH closer to the physiological value, the binding mode transitions to OR–Cu²⁺₄ (right), leading to a gradual and controlled production of H₂O₂. Coordinates for the PrP C-terminal domain, along with carbohydrates, GPI anchor and membrane were kindly provided by Professor Valerie Daggett (U. Washington).

In conclusion, we demonstrate that the occurrence/inhibition of the redox cycling of Cu²⁺ by PrP is dependent on binding mode. The voltammetric data, kinetic studies, and H₂O₂ detection and quantification unequivocally indicate that OR–Cu²⁺ is the most effective mode in inhibiting the Cu²⁺ redox cycling and in sequestering uncomplexed Cu²⁺, thus minimizing possible oxidative damage. Specifically, at nanomolar or lower Cu²⁺ concentrations, which favor the formation of OR–Cu²⁺, the complete abolishment of redox cycling of Cu²⁺ by PrP is realized by shifting the redox potential of the resultant complex out of the range wherein the complexed Cu⁺ is oxidizable by O₂ (Figure 6). As free or weakly complexed Cu²⁺ concentration increases to micromolar levels, the binding mode transitions to high Cu²⁺ occupancy, which facilitates the redox cycling of Cu²⁺ and conversion of O₂ to H₂O₂. Such results challenge the currently held hypothesis that the high occupancy mode quenches Cu²⁺ redox cycling. We show that both OR–Cu²⁺₄ and non-OR–Cu²⁺ produce H₂O₂ in a controlled and gradual manner. Instead of producing a large amount of H₂O₂ that might inflict cellular damage, H₂O₂ generated at a lower level probably serves as a cellular signal to initiate important cellular processes. We therefore suggest that OR–Cu²⁺, OR–Cu²⁺₄ and non-OR–Cu²⁺ all contribute to the maintenance of neuronal integrity by inhibiting the formation of the radical forms of ROS and regulating the endocytotic processes, perhaps through an H₂O₂ dependent signaling mechanism.