Evidence for widespread, severe brain copper deficiency in Alzheimer's dementia

21 Datasets comprising simultaneous measurements of many essential


Introduction
AD is the predominant cause of ageing-related dementia 1 .Some 35 million people world-wide were living with dementia in 2010 and the estimated global prevalence will reach ~115 million affected individuals by 2050 2 .
Consequently, AD represents the largest unmet medical need in neurology 3 .
AD is characterized by the presence of extracellular amyloid-β (Aβ) plaques and intra-neuronal neurofibrillary tangles formed from tau protein in affected regions of the brain.These aggregates have been implicated in the causation of neuronal loss and neurodegeneration, which ultimately lead to the characteristic global impairment of cognitive function but their role in pathogenesis remains unresolved and they may represent phenomena down-stream from the aetiological insult(s) that cause the disease 4 .Moreover, therapeutic interventions aimed at these protein-level phenomena have not yielded an effective diseasemodifying therapy 5 .
Perturbed brain-metal homeostasis has been associated with cognitive decline and neurodegeneration and hence may play a role in AD pathogenesis [6][7][8][9] .However, the contribution of defective metal regulation to the pathogenesis of AD is unresolved and no available therapy targeting metal metabolism has an accepted place in the pharmacotherapy of AD 10 .The brain is a highly oxidative organ, and contains abundant antioxidant mechanisms that may counteract the detrimental effects of reactive oxygen species (ROS) produced during the reactions of redox-active transition metals with molecular oxygen, superoxide and other oxygen-containing species 11 .Metals may not only affect rates of synthesis, degradation and clearance of Aβ, but are also thought to play potential roles in the aggregation of Aβ and tau 7,12 .Furthermore, as an integral component of many metalloenzymes, transition metals can modulate the function of numerous important pathways and, therefore, the metabolic processes they support 13 .Results of therapeutic interventions aimed at Aβ or tau have been uniformly disappointing to date, so new ideas concerning the pathogenesis of AD leading to identification of more tractable targets and interventions, are required to meet the continuing unmet clinical need 5,14 .
Previous studies investigating changes in metal concentrations in human AD-brain have typically reported tissue content of one or a few metals (often iron (Fe), zinc (Zn), and/or copper (Cu)) in seriously affected regions such as the hippocampus.Furthermore, results of studies comparing levels of metals in the brains of AD patients and controls reported in the literature are somewhat heterogeneous 15 , possibly due to experimental variability arising from factors such as differences in tissue processing: (e.g.fresh or formalin-fixed), handling, numbers of samples, and intrinsic variation in the distribution of metals in the brain tissue, resulting perhaps from both compartmentalization and inter-regional differences 7,16,17 .Therefore, further mapping of the regional distribution of metals in the brain is required to substantiate and extend our understanding of the content of metals in various brain regions, and to associate these with region-specific variation in AD pathology.As one example, Cu, Fe and Zn are enriched in amyloid plaques from AD brain 18,19 : by contrast, measurement of levels of these metals in whole-brain tissue from affected regions has yielded somewhat inconsistent values 20,21 .
To provide further evidence concerning the changes in concentration and regional distribution of metals in brain from both non-demented aged and cases with AD, we report a study that we performed using short-delay, fresh-frozen post-mortem human brain tissue by applying a validated inductively-coupled-plasma mass spectrometry (ICP-MS) method 22 .We made parallel measurements of eight metals and selenium (Se), and compared and contrasted results across seven functionally distinct brain regions from cases with AD and matched controls without clinical dementia.Since some brain regions are more heavily affected by AD than others 23 , we further aimed to contrast representative brain regions that are considered to be severely-affected: hippocampus (HP), entorhinal cortex (ENT), and middle-temporal gyrus (MTG) 23,24 ; moderately-affected: sensory cortex (SCX), motor cortex (MCX), and cingulate gyrus (CG); along with one control region, cerebellum (CB), which is relatively spared in the disease process 25,26 .

Acquisition and Sampling of Human Brain
Whole brains from cases and matched controls were obtained from the New Zealand Neurological Foundation Douglas Human Brain Bank, in the Centre for Brain Research, Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand 27 .All procedures in this study were approved by the University of Auckland Human Participants Ethics Committee with informed consent from all families, and its performance was consistent with international best practice for such studies.
After receipt into the Brain Bank, each brain was dissected under the supervision of neuroanatomists, to ensure accurate identification of each brain region to be studied 27,28 .Two sets of tissue samples (each of 50 ± 5 mg) were dissected from each region and stored at -80 o C until analysis.

Diagnosis and Severity
All cases had clinical dementia, whereas age-and sex-matched controls did not.Study-group characteristics are shown in Table 1, and each patient's details, including cause(s) of death as certified by post-mortem examination, are summarized in Table 2.A consultant neuropathologist diagnosed or excluded AD by applying the Consortium to Establish a Registry for AD Disease (CERAD) criteria 29 , and also determined the neuropathological severity by assigning the Braak stage 25 to each brain (Table 1, 2).

Brain Extracts
We determined metal concentrations normalized on a per-dry-weight basis [30][31][32][33] .Brain samples of 50 ± 5 mg wet-weight were first dried to constant weight in a centrifugal concentrator (Savant Speedvac; Thermo-Fisher, Waltham, MA) and dry-weight was measured using a Mettler Toledo New Classic ML balance (ML204/01): the coefficient of variation (CV) across 25 x 10-mg measurements was 0.68%.Wet-weight/dry-weight ratios are recorded in Table 1.Samples were digested in 2-mL microcentrifuge tubes (eppendorf) as described below.

Digestion
Tissue was digested using concentrated nitric acid (A509 Trace Metal Grade; Fisher, Loughborough, UK) with added 5% (v/v) Agilent Internal Standard mixture (5183-4681; Agilent Technologies, Cheadle, UK).An internal standard solution was added to the digestion solution to account for evaporation-associated volume loss during the digestion protocol.This internally standardized acid was also used at appropriate dilutions to provide rinse and calibration solutions, at 2% (v/v) final nitric acid concentration.Calibration solutions were produced by appropriate dilutions of Environmental Calibration Standard (Agilent 5183-4688).
Acid digestion was carried out using an 'open-vessel' method.Tissue aliquots were briefly centrifuged to ensure that the tissue sat at the bottom of the tube.The tube lids were punctured to prevent pressure build-up, and 0.2 mL standard-containing nitric acid added.Tubes were then inserted into a "Dri-block" heater, which was initially at room temperature.Plasma (SRM 1950) from the National Institute of Standards and Technology (NIST), Gaithersburg, U.S.A was processed in parallel with experimental samples as standard reference material (Supplementary Table 1).Tubes with standard-containing acid but no sample were also processed in each batch to provide "digestion" blanks (Supplementary Table 2).Temperature was then set to 60 o C and the block switched on.After 30 min, the set temperature was increased to 100 o C, and digestion continued for a further 210 min.After digestion, the tubes were allowed to cool overnight.
Aliquots of 100 μL were taken from each digestion and added to 15-mL Falcon tubes (Greiner) containing 5-mL LC-MS grade water, to produce solutions for analysis at a final nitric acid concentration of 2% (v/v).

ICP-MS
Metal concentrations were measured using an Agilent 7700x ICP-MS spectrometer equipped with a MicroMist nebulizer (Glass Expansion, Melbourne, Australia) and a Scott double-pass spray chamber.Nickel sample and skimmer cones were used.Sample introduction was by using an Agilent Integrated autosampler (I-AS) with helium as the collision gas.A multi-element method including all those present in the calibration solution was applied as previously reported 22 .Calibration solutions were produced by appropriate dilutions of Environmental Calibration Standard (Agilent 5183-4688).Scandium was used as the internal standard for all elements except Zn and Se, where germanium was used, and Mo, where indium was used.Two collision cell gas modes were applied, and all elements were analysed in helium mode (5.0 mL.min -1 helium), except for Se which was analysed in high-energy helium mode (10 mL.min -1 helium).Germanium and indium internal standards were analysed in both modes.Mode selection followed Agilent's recommendations to minimize interference for measured elements by e.g.isobaric cluster ions.Integration times were 0.1 s for sodium (Na), magnesium (Mg), potassium (K) and calcium (Ca), 0.3 s for manganese (Mn), Cu, Zn and molybdenum (Mo), 0.01 s for Fe, and 3 s for Se.For each analytical batch, multi-element calibration was performed using serial dilutions of the calibration standard (Supplementary Figure 1).An intermediate concentration from this calibration series was used as a periodic quality-control (QC) sample throughout each analytical batch.
Instrument and digestion blanks were also interspersed through each set of randomized samples.Detection limit, limit of quantitation and background equivalent concentration for each physiological metal measured in this study (Supplementary Table 3) were automatically calculated by the software employed (Mass Hunter, Agilent).ICP-MS generates measurements of metals in their elemental states, from samples that have been rendered into plasmas: therefore, where the result of an ICP-MS-based measurement is referred to in this manuscript, the elemental symbol has been used (e.g.K for potassium).On the other hand, most metals are present in the body as cations: therefore, when the physiological role(s) of metals and metal-related processes are discussed, the symbol for the physiological cation has been employed (e.g.K + for the potassium ion, or Cu(II) for the physiological divalent-Cu cation).

Data analysis
Datasets were exported to Microsoft Excel worksheets and individual values of each sample normalized by the corresponding sample dry-weight.Weight-adjusted datasets were then log-transformed for statistical analysis.
Means (± 95% CI) of the log-transformed data were calculated and the significance of between-group differences was examined by unpaired Welch's t-tests to allow for unequal variances and sample sizes.Means (± 95% CI) were back-transformed to reflect the actual elemental concentrations of elements.Statistical calculations were performed using GraphPad v6.04 (Prism; La Jolla, CA).P-values <0.05 have been considered significant, and those of 0.05 ≤ P < 0.10 are also tabulated.

Results
We measured concentrations of eight essential metals, Na, Mg, K, Ca, Mn, Fe, Cu, and Zn, and the metalloid Se, in seven regions of human post-mortem brains from nine AD cases and 13 controls matched for age and sex (Table 1, 2).Median (range) brain weight was 1062 g (831-1355) in the patient group and 1260 g (1094-1461; P = 0.005) in the controls (Table 1, 2): the ~16% decrease in mean brain weight in cases is consistent with the histological severity 34 .The post-mortem delay in the AD group, 7 h (4.0-12.0), was significantly shorter than in the control group 12 h (5.5-15.0;P = 0.005; Table 1, 2).The AD group comprised sporadic cases only, as determined by clinical, CERAD and Braak criteria.One control case also had neuropathological findings consistent with early-stage AD (Braak II; Table 2) and was therefore diagnosed with premanifest disease: this finding is consistent with the known frequency of asymptomatic AD in similarly-aged groups in the study population 35 .Wet-wt/dry-wt ratios did not differ significantly between cases or controls (Table 1).
In AD, measured Na levels were higher in severely affected regions (HP, ENT and MTG, where values ranged from 0.55 to 0.59 mmol/kg dry-wt) compared with corresponding regions in control brains (whose values ranged from 0.30 to 0.41 mmol/kg dry-wt; Table 3 -5).Mean Na levels were the highest in CB of normal brain (Fig. 1a, Table 3 -9).Mg levels were significantly higher in ENT and MTG, but lower in CB of AD-brain (Fig. 1b, Table 4 -5,    9).In these brain regions, Mg concentrations were all around 0.03 mmol/kg dry-wt in AD-brain.However, in normal brain, Mg levels varied among different brain regions.In the control brains, mean Mg concentrations were highest in CB (0.036 mmol/kg dry-wt; Table 9), and much lower in ENT and MTG (0.025 -0.027 mmol/kg dry-wt; Table 4, 5).Similar to Na and Mg, K showed the highest mean concentrations in CB (0.56 mmol/kg drywt) in normal brain (Fig. 1c, Table 9).The mean concentration of K was the second highest in CG in the normal brain (0.40 mmol/kg dry-wt; Fig. 1c, Table 8).Compared to normal brain tissue, levels of K were significantly lowered in CG and CB of AD (Fig. 1c, Table 8, 9).
Compared to controls, mean concentrations of Ca in AD brain trended higher in all brain regions except for CB (although values reached statistical significance only in ENT; Fig. 1d, Table 3 -9).
In normal brain, Fe levels were highest in CB (6.1 mmol/kg dry-wt) and lowest in HP (3.7 mmol/kg drywt; Fig. 1g, Table 3 -9).AD tissue showed significantly higher levels of Fe in HP, ENT, and MTG (Table 3 -5), whereas no significant differences were observed in other brain regions.Interestingly, mean Fe levels were lower in CG and CB in AD compared to controls (Table 8 -9).Similarly, Zn concentrations were higher in heavily affected regions, ENT and MTG (Table 4 -5) but lower in CB (Table 9) in AD compared to control.Mean Zn concentrations were highest in CB (1.9 mmol/kg dry-wt) and HP (1.3 mmol/kg dry-wt; Fig. 1h, Table 1 -9).
Mean Cu concentrations in controls were highest in CB (710 μmol/kg dry-wt) and lowest in ENT (310 μmol/kg dry-wt).Remarkably, Cu levels were significantly decreased in all seven regions of AD-brain compared with corresponding controls (Fig. 1i, Table 3

Discussion
Here we report measurements of physiologically-essential metal concentrations in seven regions of postmortem brain in cases with AD and controls matched for age and male-to-female ratio, whose ante-mortem brain structure and function had been assessed as normal.Although post-mortem tissue metal values may be affected by ion gradients between cells and extracellular compartments, they are neither metabolized nor exchanged with other tissues after death, due to cessation of blood circulation.There is no pronounced correlation between post-mortem delay and tissue metal content: hence, it has been suggested that total brain-tissue metal content does not change significantly after death 36 .Therefore, the relatively small differences in post-mortem delay between case and control groups in this study are unlikely to have affected the accuracy of our measurements and hence, the validity of comparisons between AD and control groups.
The maintenance of physiological intracellular Na + and K + levels is critical for controlling the homoeostasis of cell volume and prevention of apoptosis 37 .The marked increase of Na levels in the mostaffected regions of AD-brain (HP, ENT, and MTG) observed here is consistent with another report of increased Na concentrations in AD in the medial temporal lobes 38 , and frontal and parietal cortex 36 .We also observed, in moderately affected regions (MCX, SCX, and CG), that there were non-significant trends towards increased Na levels.Na levels in CB did not differ between groups, consistent with another report that cerebellar Na levels did not differ between post-mortem human AD and control brain 36 .Increased Na content has also been inversely correlated with brain volume in the HP 38 .Therefore, given that heavily-affected brain regions reportedly undergo volume losses at greater rates during disease progression 39 , our observation is consistent with the idea that Na + levels increase in regions of severe neurodegeneration/volume loss in AD, perhaps due to diminished energy supply required for aspects of cell-Na + homeostasis.Shrinkage of grey matter is commonly regarded as a consequence of neuronal cell death 40 .Cell death and cell shrinkage are distinguishable processes that are intrinsically linked, in that cell shrinkage is a universal feature of apoptosis that is conserved among species.Intracellular Na + reportedly increases transiently to initiate apoptotic signalling prior to cell-volume loss, and a significant decrease in intracellular K + and Na + is known to occur during apoptotic cell-shrinkage 37,41 .Whether the elevation in Na levels observed here is associated with neuronal apoptosis is undetermined.While a previous study has reported decreased brain K levels in AD 42 , here we found no changes of K levels in the severely affected regions of AD-brain.Instead, we identified significant lowering of K in CG and CB, at possible variance with a reported increase in K concentration in CB of AD-brain 36 .It was also suggested that K + efflux mediates cell death by apoptosis 37,43 .
Consistently, our observations may represent a result of K + efflux associated with apoptosis in these regions prior to death in AD patients, possibly due to insufficient energy supply to maintain K + homeostasis.CSF levels of Na and K did not differ between AD cases and controls in another report 36 , consistent with the altered concentrations of both metals observed in our current study, which may reflect changes in the concentrations of both in the intracellular compartment.
Neuronal Mg 2+ not only participates in energy production (e.g. as a component of purine nucleoside complexes), intracellular signalling and synaptic neurotransmission, but also modulates synaptic density and plasticity 44 .Mg 2+ displays a non-homogeneous distribution among different regions of normal brain, and significant decreases have been reported in AD 45 .Here we found that brain Mg showed a heterogeneous distribution, with CB having the highest levels in normal brain.In AD however, Mg concentrations were increased in ENT and MTG but relatively decreased in CB, resulting in a more homogenous distribution of Mg across different brain regions as compared with normal brain.Prolonged exposure to elevated Mg 2+ is said to exert an overall inhibitory effect on neuronal transmission resulting in decreased neuronal survival 44 , which could be consistent with increased Mg levels in the heavily affected regions of ENT and MTG.On the contrary, whether the markedly decreased Mg concentration in the CB is associated with improved neuronal survival remains uncertain.
Calcium homeostasis is critical for maintenance of normal brain function since Ca 2+ ions mediate neuronal signal transduction and modulate processes including synaptic plasticity and apoptosis [46][47][48] .
Characteristic lesions of AD have been associated with increased intracellular Ca 2+ , including Aβ accumulation, tau hyperphosphorylation, and neuronal death 48 .Excessive Ca 2+ is implicated as a cause of acute neuronal injury, possibly through neurotoxicity induced by excitatory amino acids 47,49 .Here, Ca levels were significantly increased only in ENT, possibly reflecting a Ca 2+ -associated neurotoxicity localized in this brain region, which is known to be an initial site of neurodegeneration in AD.Our findings also suggest alteration in total Ca concentrations in the affected regions of AD tissue following progressive neurodegeneration, although they do not exclude the possibility that disturbed Ca 2+ homeostasis occurs before neurodegeneration.
As an important antioxidant, Se plays a protective role against oxidative stress, and its concentration is well-maintained by the brain, even during dietary Se deficiency 50 .Se deficiency has been correlated with altered cognitive function 51,52 and decreased serum and plasma Se is reported in AD [53][54][55] .However, systematic examination of Se levels in AD-brain has not been reported to our knowledge.Here we found that Se concentrations were not altered in the heavily or moderately affected brain regions.Rather, significant decreases were observed in CG and CB of AD-brain.This finding may be consistent with previously observed co-localization of the Se-transport protein, selenoprotein P, with Aβ plaques and neurofibrillary tangles 56 , possibly serving as a compensatory mechanism to maintain Se levels in affected locations in AD brain.
Oxidative damage is an important aspect of AD pathology.Cu, Zn and Mn participate in oxidative defence mechanisms as necessary components of the superoxide dismutase (SOD) enzymes, including Mn-SOD (SOD2), present mainly in mitochondria 57 , and two Cu/Zn-SODs (SOD1, SOD3), present respectively in the cytosol and mitochondria, and in the extracellular space 58 .
Mitochondrial oxidative damage and impaired energy production are considered to be early pathological events that lead to neurodegeneration.Mice with brain-specific Mn-SOD deficiency exhibit spongiform neurodegeneration of the brain accompanied by increased lipid peroxidation 59 and Mn-SOD may play a protective role against apoptosis and neuronal degeneration 60 .Furthermore, Mn-deficient animals exhibit depressed Mn-SOD, with associated increase in lipid peroxidation 61 .Additionally, Mn excess confers neurotoxicity in the central nervous system, through both oxidative damage and involvement of Mn in the activation of the enzyme glutamine synthetase and therefore, in glutaminergic and GABAergic neurotransmission 62,63 .Previous studies reported decreased serum Mn levels in AD patients 55 .However, data concerning altered brain levels of Mn remain inconclusive in the current literature; independent studies have reported either increased Mn levels 64 or no significant alteration 65 in Mn levels in AD post mortem brain.This is not surprising considering the inter-regional difference in Mn levels in the brain as observed in this study.
Our results indicate that Mn homeostasis is disturbed in both HP and CB, but evidently in opposite directions, since Mn levels were increased in HP and decreased in CB of AD-brain.Also, acting through mitochondrial electron transport chain (mETC) complex II, Mn reportedly elicits mitochondrial hydrogen peroxide production 66,67 .The increase in Mn concentration observed here in the HP could be associated with severe neurodegeneration through increased oxidative stress via this mechanism.
The metal hypothesis of AD pathogenesis was originally built upon a triad of altered homeostasis of the transition elements: Fe, Zn and Cu.In AD brain, dysregulation of metal homeostasis is postulated to engender amyloidogenic and oxidative stress as a consequence, and, in reverse, AD-related proteins (for example, APP and tau) may play significant roles in brain-metal regulation 8,16 .Previous studies reported elevated levels of iron 68 and zinc 69 in extracts of AD brain and increased levels of labile Fe in the serum of AD patients 55 .Furthermore, Fe levels were reported to be markedly increased in degenerating regions of AD brain 70 .Here we also found significant increases in Fe and Zn levels that were localized in heavily affected regions.This effect could possibly be associated with overall load of plaques and tangles in these regions, considering the reported enrichment of Fe and Zn in and around amyloid plaques in AD 19 .Decreased transferrin in cortex may impair brain-Fe utilization and subsequently drive oxidative damage and neuronal degeneration in affected regions 71 .
In the brain, Zn 2+ is co-localized with glutamate and modulates both excitatory and inhibitory neurotransmission 72 .Zn 2+ not only elicits potential roles as a (putative) neurotransmitter and modulator of synaptic plasticity, but is also thought to play a potential role in AD pathogenesis related to its interactions with APP and Aβ, reportedly participating in APP processing, Aβ aggregation and Aβ clearance.Enrichment of Zn in the core of Aβ plaques might disrupt Zn homeostasis in brain regions important for memory and vulnerable to AD pathogenesis 73,74 .Our observations of increased Zn in ENT and MTG mirrors previous reports of its elevation in AD brain 69 , which may accord with the (proposed) increase of Zn 2+ -sequestration in heavily affected brain regions, potentially leading to functional unavailability of Zn 2+ required for physiological processes, regardless of any increase in total tissue-Zn.Whether or how the previously reported decrease in serum Zn levels in AD 55 is associated with altered brain-Zn regulation is unclear.A previous study has suggested that HP may contain the highest Zn content among those brain-regions examined 17 : consistently, we found Zn levels to be higher in HP than other regions, although they approximated those in CB.

Cu is a necessary component of certain metalloenzymes that play key roles in essential metabolic
processes such as energy metabolism, antioxidant-defence mechanisms, Fe metabolism, and neurotransmitter synthesis 75 .It is still controversial as to whether it is Cu excess or Cu deficiency that might contribute to the pathogenesis of AD 76 .Here we found Cu levels to be significantly depressed in all seven regions of AD-brain examined (Fig. 1).There are several reports that tissue-Cu concentration is decreased in several regions of post-mortem brain in AD 68,69 .These findings, taken together, are consistent with reported depression of cytochrome C oxidase (COX) activity in mitochondria isolated from AD-brain 77 .COX is a metalloenzyme that acts at the terminus of the mitochondrial respiratory chain (in complex IV); its catalytic core is formed by three subunits encoded by the mitochondrial genome: two of these, COI and COII, which cooperate to catalyse the reduction of molecular oxygen to form water, are linked Cu enzymes that together contain three bound Cu atoms.Marked decreases in tissue-Cu content will not only have a significant impact on cellular energy production, but are also likely to facilitate increased oxidative damage in the AD-brain by impairing Cu-mediated antioxidant defences, for example that catalysed by SOD1.Cu also plays roles in regulation of other transition metals: therefore, the possibility that altered Cu homeostasis in AD could play roles in observed alterations in levels of some of the other transition metals merits consideration.
Increased concentrations of caeruloplasmin (the major plasma ferroxidase and Cu-containing protein in plasma) in HP, ENT and frontal cortex of AD-brain, have been suggested to represent a localized acute phasetype response and/or a compensatory increase to counter oxidative stress 78 .In AD patients, caeruloplasmincopper dysregulation has been associated with elevated labile-Cu levels in the blood 55,76,79 .In light of this observation, global decreases in brain Cu as observed in this study, combined with previously-described Cu enrichment in and around amyloid plaques in AD-brain 19 , have been interpreted as consistent with marked depletion of Cu-availability for normal cellular functions (such as energy production and anti-oxidant defence) in all brain regions examined in the current study.
Another interesting pattern is evident in our data.There were markedly higher levels of Cu in the normal CB, compared to all other brain regions examined.It is noteworthy that the metal concentrations in normal CB measured in this study (on a wet-weight basis) are consistent with CB metal concentrations previously reported (for Cu, Fe, Zn, Mn, and Se only) 80 .The intrinsically higher levels of several metals measured in CB, including Ca, Mg, Zn, Fe and Cu may be attributable to higher mitochondrial density in this region.Consistently, AD-brain reportedly exhibits significant reduction in mitochondrial membrane fluidity in cerebral regions that may result from lipid peroxidation: the exception to this observation was noted to be the CB 81 .
What might be the significance of lowered brain Cu in AD? Cu plays central roles in antioxidant defence via its structural and functional actions in SOD1, SOD3, and COI and COII.We have previously shown that, in the context of diabetic cardiomyopathy (DCM, another chronic disease linked to altered Cu regulation), deficient cardiac Cu impairs antioxidant defences that are restored by drug treatment, thereby establishing the pathogenic action of lowered tissue Cu 58,82 .In DCM, lowering of myocardial Cu may be caused by the action of divalent Cu bound to N-ε-carboxymethyllysine (CML) groups in modified collagen, to suppress CTR1-mediated Cu uptake, leading to deficient Cu insertion into metalloenzymes [82][83][84] .The depression in brain-Cu levels in AD is similar to that in the diabetic heart 82 , and may well be caused by a similar mechanism; for example, extracellular CML is also elevated in AD 85 , where it could well serve as a marker for this Cu-related pathogenic process 84 .Therefore, these low brain-Cu levels are consistent with the causation of neurodegeneration and may well be pathogenic in AD.
Lowered brain-Cu levels, measured here in AD cases, approximate to those reported in Menkes' disease, an Xlinked recessive disorder caused by mutation in ATP7A, which encodes a Cu-transporting ATPase, causing severe neurodegeneration via brain-Cu deficiency 86 .Similar lowering of brain Cu with neurodegeneration also occurs in animals with spontaneous or induced mutations in Atp7a 87 .Metabolic studies reveal that trafficking of ATP7A after NMDA (N-methyl-D-aspartate) receptor activation triggers rapid release of copper ions from hippocampal neurons; ATP7A is directly required for this copper efflux, because parallel studies in hippocampal neurons derived from mice lacking functional ATP7A do not release copper 88 .Thus, ATP7A plays a critical role in the availability of an NMDA receptor-dependent, releasable pool of copper in hippocampal neurons and affords a unique mechanism linking copper homeostasis to neuronal activation in the central nervous system.In addition, NMDA receptor activity is modulated by cellular prion protein (PrPC) in a copperdependent manner 89 .Moreover, copper ions potently modulate NMDA-receptor kinetics via modulation of PrP C , providing a mechanism that potentially links defective copper regulation to neurodegeneration.Thus, disruption of the NMDA synaptic system could provide a mode of copper dysregulation in AD.Furthermore, brain-Cu deficiency also causes neurodegeneration in other contexts, for example, insufficient dietary-Cu uptake [90][91][92][93] .Taken together, these observations suggest that a treatment that can safely and effectively restore brain Cu might exert beneficial effects in the treatment of AD 10 .However, elevating oral Cu intake had no effect on cognition in patients with mild AD in a pilot phase 2 clinical trial 94 although, in that study, restoration of brain-Cu levels was not demonstrated.An effective method for in vivo measurement of brain Cu would be helpful for monitoring future intervention trials that aim to restore brain Cu but, to our knowledge, such a method is not yet available.Also required is an effective means of therapeutically raising brain-Cu levels: in this regard, the divalent-Cu-selective chelator, triethylenetetramine demonstrably elevates tissue Cu in another metabolic context, namely localized myocardial Cu deficiency in diabetic cardiomyopathy 10,82,95 .
Recent data demonstrating severe diabetes-like metabolic perturbations in the brain in AD may also be relevant to this objective 9,96 .
In conclusion, we have presented a study of post-mortem levels of nine essential elements in seven regions of the human brain in cases of AD and matched controls.The inter-regional difference in metal concentrations in both diseased and healthy brains as observed in this study: 1) demonstrates the importance of spatial resolution when studying a complex organ such as brain, and 2) provides potential explanations for some of the apparently differing findings in the literature.AD-associated alterations in metal homeostasis were evident in all brain regions studied, but changes in metals other than Cu were present only in a subset.By contrast, these data provide compelling evidence for pan-cerebral Cu deficiency in AD.Widespread brain-Cu deficiency may well contribute to the pathogenesis of neurodegeneration and dementia, probably acting via Cu-deficiency-induced defects in energy utilization and anti-oxidant defences.We propose that new therapeutic interventions that can safely and effectively restore brain Cu levels, could have a place in the experimental therapeutics of Alzheimer's dementia.Aliquots of plasma (SRM 1950) from the National Institute of Standards and Technology (NIST, Gaithersburg, U.S.A.) were processed with sample batches to provide a standard reference material.NIST SRM 1950 aliquots of 50 µl were digested and processed in the same way as samples.The standard material provides certified reference values for Na, Mg, K, Ca, Cu, Zn and Se in the units given in the table.All measurements across multiple batches were within 5% of the certified values except for zinc, which was 6.26%.The %CV across these batches was less than 5% for all metals.

Figure 1 .
Figure 1.Concentrations of nine physiologically essential elements in seven brain regions compared between AD cases and matched controls.Shown are elemental concentrations in each region of human post-mortem brain tissue from control (blue) and AD (red) subjects.Mean and between-group significance of calcium levels shown in (d), MTG, were determined from all measurements; two outlying values, one from each group, were excluded from the plot for clarity and all other data sets are complete as shown.Abbreviations: hippocampus, HP; entorhinal cortex, ENT; middle-temporal gyrus, MTG; sensory cortex, SCX; motor cortex, MCX; cingulate -9), consistent with the presence of pan-cerebral Cu deficiency.This pattern distinguished Cu from all other elements measured.Brain-metal concentrations in one control case (H241; Table 2), who had signs of premanifest AD (Braak Stage II), are presented in Table 10: these values do not enable clear differentiation from either experimental group.

Table 8 : Metal concentrations in cingulate gyrus of AD and control brains
P-values for significance of between-group differences were calculated by Welch's ttests based on measurements from control (n=9) and AD (n=9) brains.

Table 9 : Metal concentrations in cerebellum of AD and control brains
P-values for significance of between-group differences were calculated by Welch's ttests based on measurements from control (n=13) and AD (n=8) brains.