Role of Copper in the Onset of Alzheimer’s Disease Compared to Other Metals

Alzheimer’s disease (AD) is a neurodegenerative disorder that is characterized by amyloid plaques in patients’ brain tissue. The plaques are mainly made of β-amyloid peptides and trace elements including Zn 2+ , Cu 2+ , and Fe 2+ . Some studies have shown that AD can be considered a type of metal dyshomeostasis. Among metal ions involved in plaques, numerous studies have focused on copper ions, which seem to be one of the main cationic elements in plaque formation. The involvement of copper in AD is controversial, as some studies show a copper deficiency in AD, and consequently a need to enhance copper levels, while other data point to copper overload and therefore a need to reduce copper levels. In this paper, the role of copper ions in AD and some contradictory reports are reviewed and discussed.

The key event in AD is the formation of fibrils and plaques in AD patients’ brain. Plaques are mainly made of β-amyloid peptide, the natural peptide that is produced in the brain and exists at nanomolar concentration levels in cerebrospinal fluid (CSF) and serum ( Masters et al., 1985 ; Vigo-Pelfrey et al., 1993 ). On the other hand, a high concentration of trace metals, including copper, is observed in amyloid plaques ( Miller et al., 2006 ). Interestingly, some data show that copper distribution in the brain does not correspond to β-amyloid plaques distribution in TASTPM mice model ( Torres et al., 2016 ) with plaque pathology but not appreciable neuronal loss ( Howlett et al., 2004 ). Copper is a necessary trace metal in nervous system development since disruption of its homeostasis leads to neurodegenerative disorders like Menkes and Wilson’s diseases ( Waggoner et al., 1999 ). Cu 2+ ions bind to β-amyloid peptides with high affinity ( Atwood et al., 2000 ; Sarell et al., 2009 ; Barritt and Viles, 2015 ; Mital et al., 2015 ; Drew, 2017 ) and increase the proportions of β-sheet and α-helix structures in amyloid peptides, which can be responsible for β-amyloid aggregation ( Dai et al., 2006 ). Various concentrations of Cu 2+ ions enhance fibril formation while binding of copper ions to β-amyloid noticeably increases its toxicity for cells ( Dai et al., 2006 ; Sarell et al., 2010 ). In addition, substoichiometric concentrations of Cu 2+ are more toxic to cells ( Sarell et al., 2010 ). Fibril formation is highly pH-dependent and Cu 2+ ions cause it to occur at physiological pH. However, the formation of amorphous aggregates dominates in acidic conditions ( Jun et al., 2009 ; Sarell, 2010 ; Lv et al., 2013 ). In a proton-rich environment, β-amyloid (Aβ 40 ) possesses two copper binding sites, and its second bound Cu 2+ ion causes the formation of amorphous aggregates by preventing the conformational transition of β-amyloid into amyloid fibrils ( Jun et al., 2009 ). The production of Reactive Oxygen Species (ROS) is a key factor in β-amyloid toxicity toward neurons, which is dependent on metal ion redox properties. Copper ions in complex with β-amyloid fibrils produce hydrogen peroxide, in the presence of biological reducing agents ( Parthasarathy et al., 2014 ). When the ratio of copper to peptide increases, hydrogen peroxide levels and the production of hydroxyl radicals increase, and the morphology of aggregates changes from fibrillar to amorphous ( Mayes et al., 2014 ). Although previous studies have presented ROS as fatal molecules provoking neurodegeneration, the accumulated evidence shows that some ROS act as essential molecules in processes underlying cognition and memory formation ( Klann, 1998 ; Yermolaieva et al., 2000 ; Knapp and Klann, 2002a , b ; Kamsler and Segal, 2003 , 2004 ; Hu et al., 2006 ; Kishida and Klann, 2007 ). On the other hand, some results imply that the copper-amyloid complex produces fewer ROS than free copper ions ( Nakamura et al., 2007 ). According to in vitro data, oligomeric and fibrillar forms of β-amyloid inhibit H 2 O 2 generation at higher concentrations of Cu 2+ . In addition, the fibrillar form generates less H 2 O 2 than the oligomeric form ( Fang et al., 2010 ). Copper toxicity in AD brains is attributed to the oxidized form of copper ions, i.e., Cu 2+ , ( Brewer, 2015 ; Greenough et al., 2016 ). In contrast, other data show that copper ions are only transported in their reduced form, i.e., Cu 1+ , ( Macreadie, 2008 ). Some studies suggest that Cu 2+ bypasses the liver ( Brewer, 2015 ). Otherwise, some data show that the removal of Cu 1+ from β-amyloid, hinders the formation of oligomers and prevents ROS production ( Atrián-Blasco et al., 2015 ). A study of the affinity of the soluble copper-binding domain of the β-amyloid peptide for Cu 1+ shows that it binds to β-amyloid stronger than Cu 2+ suggesting Cu 1+ is the relevant in vivo oxidation state ( Feaga et al., 2011 ). Both Cu 1+ and Cu 2+ inhibit β-amyloid degradation by insulin-degrading enzyme, but Cu 1+ cations act as irreversible inhibitors ( Grasso et al., 2011 ). Copper ion by reduction from Cu 2+ to Cu 1+ protects proteins against free radicals ( Deloncle and Guillard, 2014 ). A meta-analysis ( Schrag et al., 2011 ) and subsequent studies ( James et al., 2012 ; Szabo et al., 2016 ) demonstrated that the concentration of total copper is decreased in the brain of AD patients, while the concentration of labile copper is increased in the most affected regions of the AD brain ( James et al., 2012 ). In addition, AD cortical tissues ( James et al., 2012 ) and the cortex of mice with Traumatic Brain Injury show an elevated binding capacity for Cu 2+ ( Peng et al., 2015 ). Another study shows that in APP sw/0 mouse model of AD, which shows parenchymal plaques but no neuronal loss ( Elder et al., 2010 ; Sasaguri et al., 2017 ), unlike in the control mouse in which the metal accumulates in the capillaries, copper i

A significant elevation in Zn 2+ is found in the AD hippocampus and amygdale area ( Deibel et al., 1996 ), and in the AD neuropil compared to controls ( Lovell et al., 1998 ), although some data indicate an elevation in Zn 2+ content in the AD brain cortex ( Religa et al., 2006 ). Accordingly, a meta-analysis revealed no significant changes in zinc content in the AD neocortex ( Schrag et al., 2011 ). Nevertheless, three meta-analyses ( Ventriglia et al., 2015 ; Wang et al., 2015 ; Li et al., 2017 ) were recently carried out on plasma and serum zinc levels in AD patients and healthy controls. One study ( Ventriglia et al., 2015 ) (the pooled sample size included 777 AD vs. 1728 healthy controls) analyzed a total of 16 studies and concludes that AD patients show a decrease in serum zinc levels compared to healthy controls. Ventriglia et al.’ (2015) study (the pooled sample size included 287 AD vs. 166 healthy controls) analyzed a total of five studies on plasma zinc: two indicate a decrease and three no significant differences in plasma zinc in AD patients compared to controls. The other study ( Wang et al., 2015 ) (the pool of subjects included 862 AD patients vs. 1705 controls) analyzed a total of 17 studies: 10 indicate a decrease, and seven no significant differences in serum zinc in AD patients compared to controls. The most recent meta-analysis by Li et al. (2017) analyzed 22 studies with a total pool of 1027 patients with AD and 1949 healthy controls: three studies indicate an increase, 18 studies a decrease, and one study no differences in serum zinc between AD and healthy controls. Interestingly, an older study found that a rise in serum Zn 2+ occurs in a special type of AD ( González et al., 1999 ). As the absence of fibrinogen in serum is the main difference between serum and plasma compositions and fibrinogen-β-amyloid interactions are involved in AD progression ( Cortes-Canteli et al., 2012 ; Derakhshankhah et al., 2016 ), it is likely that the reduction in serum Zn 2+ in AD patients is due to its interactions with fibrinogen. The rise in zinc levels is accompanied by a rise in tissue amyloid levels ( Religa et al., 2006 ). AD brain tissue contains hot spots of metal ions especially enriched by copper and zinc ions. In 2006, it was reported for the first time that β-amyloid plaques and hot spots of accumulated metal ions co-localize ( Miller et al., 2006 ). It has been suggested that zinc ions induce β-amyloid aggregation in vitro whereas Cu 2+ ions inhibit it through competing with zinc for histidine residues. The strongest inhibitory effect occurs at a copper: β-amyloid molar ratio of about four. Above this value, copper ions themselves induce aggregation ( Suzuki et al., 2001 ). Interestingly, previous studies have shown that β-amyloid binds to three or four Cu 2+ ions at pH 7.0 ( Atwood et al., 1998 ). Studies on synthetic Aβ(1–40) and Aβ(1–42) peptides show that at physiological pH, β-amyloid binds the same ratio of copper and zinc ions, whereas in acidic conditions copper ions replace zinc ions ( Figure 2 ; Atwood et al., 2000 ; Roberts et al., 2012 ). Cu 2+ and Zn 2+ ions inhibit β-amyloid fibrillization, promoting instead the formation of non-fibrillar aggregates in vitro . In addition, zinc ions have a threefold stronger inhibitory effect than copper ions ( Tõugu et al., 2009 ). Both Cu 2+ and Zn 2 ions prevent the formation of soluble fibrils if incubated with Aβ(1–42) in vitro ( Bolognin et al., 2011 ). Some authors have also reported that small, soluble oligomers play the main role in β-amyloid neurotoxicity ( Tabner et al., 2011 ). Cu 2+ ions form stable and soluble 1:1 complexes with β-amyloid (Aβ 40 ) while Zn 2+ ions cause their partial aggregation ( Tõugu et al., 2008 ). Cu 2+ -induced aggregates are toxic to neurons only in the presence of ascorbate, while monomers and zinc-induced aggregates are not toxic ( Tõugu et al., 2009 ). FIGURE 2 Copper and zinc ions in AD. (A) Zinc ions promote β-amyloid aggregation. (B) Copper ions have an inhibitory effect on aggregation induced by zinc ions. Under physiological conditions, β-amyloid binds the same ratio of copper and zinc ions, but in acidic pH copper ions replace zinc ions overall. The role of zinc supplementation in AD remains controversial according to a systematic review in 2012 reporting that Zn 2+ in the diet has no effect on cognitive decline ( Loef et al., 2012 ) even though the Zenith and the Zincage studies suggested some effects ( Simpson et al., 2005 ; Marcellini et al., 2006 ; Maylor et al., 2006 ). However, 3xTg-AD mice model with background neuropathology similar to AD patients ( Sasaguri et al., 2017 ) displayed a delay in memory impairment ( Corona et al., 2010 ). Furthermore, Tg2576, TgCRND8 and CRND8/E4 models exhibited a potentiationin memory impairment ( Linkous et al., 2009 ; Railey et al., 2011 ; Flinn et al., 2014 ). Moreover, Tg2576 mice and Sprague Dawley rats showed a lower brain copper and amyloid burden

There is considerable evidence that the amyloidogenic pathway takes place in lipid rafts, which are specific membrane domains enriched in cholesterol. The enzymes that cleave APP to β-amyloid peptides are found in lipid rafts ( Riddell et al., 2001 ; Hung et al., 2009 ). Cellular copper deficiency results in an accumulation of copper ions in cholesterol-rich lipid rafts. In fact, the copper level in lipid rafts is inversely related to cellular copper, which results in enhanced copper-amyloid complex formation under copper deficiency conditions in AD ( Cater et al., 2008 ; Hung et al., 2009 ). Clearly, diet is a critical factor in the progression of AD. Several studies indicate the key role of fat in AD ( Grant, 1997 ; Sparks and Schreurs, 2003 ; Morris et al., 2006 ; Brewer, 2012 ). The effective role of Cu 2+ and cholesterol overload in neurodegeneration has been reported ( Arnal et al., 2013b ; Wong et al., 2014 ). In rats treated with copper and cholesterol, a significant change in visuo-spatial memory is detected ( Arnal et al., 2013b ) while the administration of dietary cholesterol plus copper-supplemented drinking water (in the form of copper sulfate) induces an accumulation of β-amyloid in rabbit brain ( Larry, 2004 ). The binding of Cu 2+ ions to β-amyloid causes oxidation of cholesterol, and the generation of H 2 O 2 ( Hung et al., 2009 ) and other lipid peroxidation products accumulating in AD patients’ brain ( Murray et al., 2007 ) and in Tg2576 transgenic mouse model ( Puglielli et al., 2005 ). In fact, previous unsuccessful therapeutic attempts and recent findings regarding β-amyloid accumulation at lipid rafts have led to a new hypothesis that neurotoxicity in AD is the result of the association of small soluble amyloid oligomers with the plasma membrane ( Drolle et al., 2014 ; Kotler et al., 2014 ; Arbor et al., 2016 ). B-amyloid membrane binding is mediated by its interactions with phosphatidylserine ( Ciccotosto et al., 2011 ). A highly toxic isoform of β-amyloid that accumulates in AD patients’ brain has a great capacity to induce lipid peroxidation. It also alters the calcium influx by binding to cell membranes. This isoform does not lead to an increase in ROS but it causes instead a reduction in plasma membrane integrity and an increase in dityrosine-β-amyloid oligomers ( Gunn et al., 2016 ). Other studies have suggested that the toxicity of the cross-linked dimer is due to enhanced membrane binding ( Ciccotosto et al., 2004 ). Simulation experiments show that increased cell membrane cholesterol results in some changes in the membrane; namely, an enhancement of surface hydrophobicity and a reduction in bilayer mobility. These membrane changes induce amyloid binding to the cell membrane and cause β-amyloid to refold into a helical or unstructured form. B-amyloid is stabilized on the membrane surface or inserted into the bilayer with the help of calcium ions ( Yu and Zheng, 2012 ). In fact, β-amyloid peptides compose the oligomeric pores in the membrane via the cholesterol-binding domain ( Lashuel et al., 2002 ; Di Scala et al., 2014 ). Channel formation is cholesterol-dependent and occurs in the presence of at least 30% cholesterol in lipid bilayer membranes. These pores are ion channels that disrupt calcium homeostasis in neural cells and have led to the return of the “calcium hypothesis” of AD ( Di Scala et al., 2014 , 2016 ; Figure 3 ). In the absence of copper, β-amyloids are cleared to the blood even if there are increased cholesterol levels, while in the presence of copper β-amyloid accumulate in the brain ( Sparks, 2007 ). Otherwise, copper ions in the absence of β-amyloid are not toxic to cells ( Sarell et al., 2010 ). FIGURE 3 Schematic diagram of the relationship between copper and lipid rafts in AD. (A) The enzymes that cleave APP to β-amyloid peptides are found in lipid rafts. (B) Cellular copper deficiency results in copper ions accumulating in cholesterol-rich lipid rafts, and β-amyloid production increases because of greater enzyme activity due to a rising copper concentration. (C) β-amyloid peptides compose the oligomeric pores in membranes via the cholesterol-binding domain. These pores are calcium channels that disrupt calcium homeostasis in neural cells.

The review was written by SB with assistance and feedback from AS, RS, TH, and MS. All authors approved the final version of the manuscript.