Chronic Exposure to High Levels of Zinc or Copper has Little Effect on Brain Metal Homeostasis or Aβ Accumulation in Transgenic APP-C100 Mice

Abstract

Aberrant metal homeostasis may enhance the formation of reactive oxygen species and Aβ oligomerization and may therefore be a contributing factor in Alzheimer’s disease. This study investigated the effect of chronic high intake of dietary Zn or Cu on brain metal levels and the accumulation and solubility of Aβ in vivo, using a transgenic mouse model that over expresses the C-terminal containing Aβ fragment of human amyloid precursor protein but does not develop amyloid deposits. Exposure to chronic high Zn or Cu in the drinking water resulted in only slight elevations of the respective metals in the brain. Total Aβ levels were unchanged although soluble Aβ levels were slightly decreased, without visible plaque formation, enhanced gliosis, antioxidant upregulation or neuronal loss. This study indicates that brain metal levels are only marginally altered by long term oral exposure to extremely high Cu or Zn levels, and that this does not induce Aβ-amyloid formation in human Aβ expressing, amyloid-free mice, although this is sufficient to modulate Aβ solubility in vivo.

Introduction

Metals have been postulated to pay a role in the pathogenesis of Alzheimer’s disease (AD) and related neurodegenerative diseases (Atwood et al. 1999; Bush 2000; Maynard et al. 2005; Barnham and Bush 2008). Abnormal metal levels are found in AD plasma and brain tissue, with high concentrations of Cu and Zn in and around the amyloid plaques in AD brain. Zinc (Zn) and copper (Cu) both facilitate the oligomerisation of Aβ, and imbalanced metal levels may contribute to the degenerative nature of AD via the uncontrolled generation of reactive oxygen species (ROS) by redox active metals like Cu (reviewed by Barnham et al. 2004). Furthermore, metal homeostasis can be modulated by the amyloid precursor protein (APP) and its cleavage product Aβ, and hence their abnormal metabolism in AD may further contribute to abnormal brain metal homeostasis.

Aβ possesses Cu and Zn binding sites capable of reducing Cu²⁺ (Huang et al. 1999). APP also has a Cu binding site in the N-terminal cysteine-rich domain, which reduces Cu²⁺ to Cu¹⁺ (Multhaup et al. 1996), as well as a Zn binding site which is believed to have a structural role (Bush et al. 1993). APP and amyloid precursor-like protein 2 (APLP2) knockout mice have elevated brain Cu levels (White et al. 1999) whilst APP transgenic mice or TgC100 mice (expressing the C-terminal fragment of APP containing the Aβ peptide) have decreased Cu levels (Maynard et al. 2002) demonstrating that both APP and Aβ have a role in Cu homeostasis.

Interaction with Cu²⁺ and Zn²⁺ mediates the aggregation of Aβ in vitro (Atwood et al. 1998, 2000a, b; Bush et al. 1994). Zinc is a more powerful inducer of Aβ aggregation than Cu, or indeed any other metal and this process is pH dependent (Atwood et al. 1998; Clements et al. 1996; Miura et al. 2000). The balance between Zn and Cu levels, and the maintenance of physiological pH, may be important to prevent Aβ aggregation and amyloid formation. The dynamics of these metal ion interactions with Aβ in vivo however, remain largely enigmatic. Synaptic Zn has been shown to play an important role in amyloid formation in Tg2576 mice, as its depletion, via genetic ablation of the synaptic Zn transporter ZnT3, reduces amyloid plaque formation (Lee et al. 2002). Furthermore, trace amounts of Cu, but not Zn, or other metals in drinking water have been shown to promote Aβ amyloid formation in hypercholesterolemic rabbits and amyloid forming APP/PS1 mice (Sparks et al. 2006). Chelation of metal ions on the other hand, reverses the aggregation of synthetic Aβ peptide and dissolves amyloid in post-mortem human brain specimens (Atwood et al. 1998; Cherny et al. 2001; Huang et al. 1997). Treatment of the Tg2576 transgenic mice with clioquinol, an orally bio-available metal chelator, also induced a marked inhibition of cortical amyloid accumulation (Cherny et al. 2001).

The catalytic capacity of Aβ to reduce Cu²⁺ may further contribute to the accumulation of Aβ via the formation of covalently linked protein aggregates that are resistant to clearance (discussed in Perry et al. 2002), generating larger protein aggregates that are difficult to clear.

AD brain exhibits marked oxidative damage of proteins, lipids and nucleic acids (Hensley et al. 1998; Nunomura et al. 1999; Pappolla et al. 1992; Pratico et al. 2001; Sayre et al. 1997; Smith et al. 1994, 1998a, b, 2000), and Tg2576 mice display oxidative damage similar to that found in AD brain (Smith et al. 1998a, b), with an elevation in oxidative stress markers preceding amyloid formation, and increasing with the age-dependent development of amyloid pathology (Pratico et al. 2001). The oxidative damage is highly concentrated in and around amyloid plaques, but also extends to the neuropil which is devoid of Aβ deposits.

Dietary metal overload has been postulated as a contributing factor for AD. Although the brain has mechanisms to tightly control the levels of metals entering the brain, these mechanisms may become corrupted with ageing. The formation of Aβ deposits in rabbits as a result of a high cholesterol diet could be prevented by feeding demineralised water, instead of tap water, which contained traces of Cu (0.12 ppm), and a cocktail of other metals (Sparks et al. 2006; Sparks and Schreurs 2003). This suggests that dietary metals may influence Aβ deposition in the brain under certain conditions, and at least in certain species, and where Aβ clearance is already challenged by other factors. This present study investigated whether chronic high doses of dietary Zn or Cu could alter brain metal levels, and also the aggregation or solubility of Aβ in vivo, using a transgenic mouse model that overexpresses the C-terminal 100 amino acids of human amyloid precursor protein (APP-C100; Li et al. 1999). We chose to use the TgC100 mouse line in this study as this model does not express any familial AD mutations, and does not normally form amyloid deposits in the brain up to 17 months of age, and we wished to see whether the metal feeding could modulate the Aβ solubility or induce amyloid formation in this normally amyloid-free human Aβ expressing mouse.

Methods

Feeding Mice with Metals

Mouse strains TgC100.wt^+/+ (also referred to as TgC100) and C57BL6/DBA, were used for metal feeding experiments. The TgC100.wt mice expressing the wild type sequence of C-terminal 100 residues of human APP was generated as described previously (Li et al. 1999), and bred to homozygosity in a C57BL6/DBA background. Non-transgenic mice (NTg) of the same background as the transgenic mice (C57BL6/DBA) were used as controls.

Mice were fed standard GR2 laboratory rat and mouse chow (Barastoc, Pakenham, Vic, Australia) and tap water ad libitum, until the beginning of the metal feeding, at which point metals were dissolved in tap water or demineralized water (dH₂O). Both water sources contained similar levels of metals as measured by ICP-MS (tap water: 0.022 ppm Zn, 0.024 ppm Cu and 0.006 ppm Fe; distilled water, 0.032 ppm Zn, 0.017 ppm Cu and 0.002 ppm Fe). GR2 laboratory rat and mouse chow is reported to contain 7.3 ppm Cu, 52.0 ppm Zn and 29.0 ppm Fe. The ratio of food:water intake for a mouse is ~5:6 (an average 20–30 g mouse eats 5 g food and drinks 6 ml water per day).

Before initiating large-scale metal feeding experiments, a pilot feeding experiment was set up. The transgenic mice were given drinking water containing varying amounts of ZnSO₄ (500, 1000 and 5,000 ppm Zn) or CuSO₄ (100 and 300 ppm Cu) for periods of up to 8 weeks, and their body-weights were monitored and general health was observed on a daily basis. Maximum metal levels chosen for the pilot experiment were based on published experiments in which mice or rats had been fed with Cu or Zn salts (Cunnane et al. 1986; Massie and Aiello 1979; Olafson 1983; Pocino et al. 1991, 1990; Sato et al. 1997; Zhang et al. 1994). After 2 weeks, mice fed with 5,000 ppm Zn showed a considerably reduced water intake, and appeared sickly and dehydrated, at which point this feeding dose was ceased. After 6 weeks, mice fed 300 ppm Cu showed a decline in water consumption and elevated heart rates. Based on these data, three experiments were set up. (1) High Zn diet: metal feeding consisted of tap water to which was added either 300 ppm Zn, or 1,000 ppm Zn, in the form of dissolved ZnSO₄. Analysis of the “1,000 ppm Zn in tap water” was found to contain 1,006 ppm Zn, 7 ppm Cu and 108 ppm Fe. The “300 ppm Zn in tap water” was found to contain 271 ppm Zn, 4 ppm Cu and 9 ppm Fe. Metal feeding was commenced at 7 weeks and continued until 17 months when the mice were euthanased for analysis. A subset of mice were euthanased and analysed at 7 months to see if there was any pathology developing, while others continued to age. (2) High Cu or Zn diet: mice at 7 to 11 weeks were fed dH₂O containing either 500 ppm Zn, or 100 ppm Cu in the form of dissolved ZnSO₄ or CuSO₄ until euthanasia at 17 months. Measurement of “100 ppm Cu in dH₂O” was found to contain 116 ppm Cu, 13 ppm Zn and 6 ppm Fe. Measurement of “500 ppm Zn in dH₂O” was found to contain 425 ppm Zn, 0.5 ppm Cu and 7.9 ppm Fe. A subset of mice were euthanased and analysed at 5 months to see if there was any pathology developing. (3) 500 ppm Zn or 150 ppm Cu diet: TgC100 mice at 2 months were fed dH₂O containing either 500 ppm Zn, or 150 ppm Cu, in the form of dissolved ZnSO₄ or CuSO₄. Metal feeding was continued until euthanasia at 11 months. Balanced numbers of males and females were included in the different feeding groups for all experiments. All the experimental procedures involving animals were performed in accordance with guidelines established by the animal ethics committee at the University of Melbourne, according to the National Health and Medical Research Council of Australia.

Preparation of Mouse Brain Tissue

Mice were killed by anesthetisation with halothane, followed by transcardial perfusion with PBS (pH 7.4) at 100–120 mmHg until the perfusate ran clear. Whole brains were isolated after removal of olfactory bulb and cerebellum. The brains were sagitally halved, and the right cerebral hemisphere of all mice used for Western blot and ELISA analysis. Left hemispheres were used for either immunohistochemical or metal analysis. In order to minimise variation due to sex differences, immunohistochemistry was performed only on the female tissues, whereas all male left hemispheres were used for metal analysis. Balanced quotas of surplus female tissues not required for immunohistochemical analysis were also used for metal analysis. Tissues isolated for Western blotting were snap frozen on dry ice and stored at −80°C. Tissues for metal analysis were weighed into pre-weighed vials to obtain wet tissue weight, and stored at –80°C until use. In order to minimise metal contamination of samples for metal analysis, all tubes and equipment were pre-soaked in 1% nitric acid and rinsed in dH₂O water prior to use. Brain halves for immunohistochemical analysis were fixed by soaking in 10% formalin at 4°C for 24 h, and processed through 70, 90% and three changes of 100% ethanol, followed by two grades of 100% xylene, two grades of 100% chloroform and four grades of parafin wax, all at 60°C on a Tissue-Tek ® VIP (Sakura). Processed tissue was then embedded in parafin.

Metal Analysis

Metals were analysed according to the procedure which was described previously (Maynard et al. 2002). Freeze dried brain samples were dissolved overnight in 0.6 ml of concentrated HNO₃ (Aristar, BDH) in 5 ml poly-propylene, acid washed tubes, followed by heating to 80°C for 20 min, and allowing them to cool to room temperature. In order to dissolve lipid components, an equal volume of H₂O₂ (Aristar, BDH) was added, and once effervescing had ceased (~30 min) samples were heated to 70°C for 15 min., and allowed to cool. Each sample was diluted in triplicate in 1% HNO₃, and metal levels were measured by inductively coupled mass spectrometry (ICP-MS) with an Ultramass 700 (Varian, Vic, Australia) in peak-hopping mode with spacing at 0.100 AMU, one point per peak, 50 scans per replicate, and three consecutive replicates per sample. Plasma flow was 15 L/min with an auxiliary flow 1.5 L/min. RF power was 1.2 kW. Each sample was introduced using a glass nebulizer at a flow of 0.88 L/min. The instrument was calibrated using a 1% HNO₃ mixed calibration standard (Merck Pty. Ltd.) containing 10, 50 and 100 ppb of all metals measured in 1% HNO₃, and compared to an internal control Standard Reference Material (SRM; NIST Bovine Liver SRM 1557B; National Institute of Standards and Technology, USA). Each tissue sample was diluted and measured in triplicate, and the average value was used for analysis. The metal values are expressed as µg/g or pMol/g wet weight of the original pre-frozen tissue sample. Error bars represent standard error of the mean (SEM) of each group of mouse brains analysed.

Tissue Homogenisation and Fractionation

Frozen brain tissue was thawed on ice and homogenised in ice-cold homogenisation buffer containing PBS (pH 7.4), and a protease inhibitor cocktail (Sigma) and centrifuged for 5 min at 1000×g. Sedimenting material was discarded and the remaining homogenates were used for analysis.

For preparation of the PBS-soluble fraction, an aliquot of homogenate was centrifuged at 100,000 g for 1 h at 4°C, and the supernatant termed the “PBS-soluble” fraction. Protein concentration was determined by Bichinolic Acid (BCA) Protein Assay (Pierce) using BSA standards.

ELISA Analysis of Human Aβ Levels

Aβ levels were determined using the established DELFIA assay format as described previously (George et al. 2004; Li et al. 2006), using mAbG210 (for Aβ40) as the capture antibody, and mAbWO2 (Aβ1-16) as the detection antibody. The Aβ42 species was beneath detection limits and hence could not be quantitated as reported previously (Li et al. 2006). To measure total Aβ40, an aliquot of homogenate was mixed with 8 M guanidine hydrochloride to a final concentration of 5 M to solubilize proteins. The mixtures were diluted 1:10 in the ELISA buffer for Aβ measurement. Soluble Aβ40 was measured in the PBS-soluble fractions without the addition of guanidine hydrochloride. Aβ40 levels were calibrated against Aβ1-40 peptide standards, after subtraction of background fluorescence in the absence of sample. Samples were analysed in triplicate.

Western Blot Analysis

To analyse the expression of APP and other protein markers, equal amounts of protein from homogenates were electrophoresed on 10% tris-tricine poly-acrylamide gels, as described previously (George et al. 2004; Li et al. 1999). Proteins were transferred to 0.2μm nitrocellulose membranes, and assayed by immunoblotting with enhanced chemi-luminescence detection. Primary antibodies used: neurofilament 200 (1:1,000, Sigma), GFAP (1:2000, Dako), SOD1-100 (Stressgen), β-tubulin (1:1000, Sigma), polyclonal 369 (1:2000, a generous gift from Prof Sam Gandy) and WO2 (1:100 ((George et al. 2006). Bands were quantitated by densitometry using NIH-Image 1.61 (public domain software). For analysis of bands of similar molecular weight (such as NF200 and APP), membranes were stripped of antibody with 3–4 washes in 0.1 M NaOH between applications of different antibodies.

Immunohistochemistry

Paraffin embedded brains were sectioned coronally at 7 μm using a “Mictom” microtome, and collected on AAS coated glass slides. The immunostaining procedure has been described previously (Li et al. 1999; Li et al. 2006). Sections were de-waxed and brought to water with three changes of Shellex (Access Fuels and Lubricants, Mulgrave, Vic, Australia), followed by decreasing gradations of ethanol, 100, 90, 70 and 0% (dH₂O). Epitopes were revealed by treating with 80% formic acid at RT (except for GFAP staining, where trypsin treatment was used: 1 mg/ml CaCl₂, 1 mg/ml trypsin in TBS pH 7.4, at 37°C 15min). Endogenous peroxidase activity was blocked by treating with 3% H₂O₂ for 5 min at room temperature. Slides were then washed in dH₂O and equilibrated in TBS pH 7.4. Non-specific sites were blocked with blocking serum (20% pre-immune serum from species in which secondary antibody was raised, in TBS pH 7.4) which was then applied to sections for 30 min at room temperature.

Primary antibody was then applied (mAbWO2 for APP, 1:1000 (George et al. 2006), GFAP, 1:50, Dako, Denmark) and diluted in 20% blocking serum for 1 h at 37°C. Slides were then washed two times in TBS, pH 7.4. Biotinylated secondary antibody was applied (rabbit anti-mouse or swine anti-rabbit) at 1:500 in 20% serum for 30 min at RT, followed by two washes in TBS, pH 7.4, and streptavidin 1:1,000 in 20% serum for 30 min at RT. After two washes in TBS, pH 7.4, staining was developed by applying freshly prepared DAB for up to 5 min. Slides were then promptly washed in TBS, pH 7.4, and counterstained in Mayers haematoxylin for 1 min to stain nuclei, washed in water, followed by 30 s in Scott’s tap water, followed by tap-water. Sections were dehydrated in 100% ethanol, followed by Shellex, and mounted under coverslips in D.P.X.

Statistical Analysis

For ELISA and Western-blot analyses, Student’s two-tailed t-test was used. For metal level analysis, where multiple variables existed (genotype and diet), two-way analysis of variance (ANOVA) was performed, with metal feeding group and genotype classed as independent variables. Post-hoc Scheffé tests were then used to test the metal-feeding effect in each mouse line individually. ANOVA tests were performed using Statistica™ for the Macintosh (StatSoft™).

Results

Effects of High Dietary Zinc and Copper Intake on Metal Levels in Mouse Brain

A pilot feeding experiment established the maximal tolerable level of Zn in the drinking water for these mice to be 1,000 ppm, and Cu to be 150 ppm. To best model the conditions of prolonged or life-long exposure to high dietary metals, metal feeding was commenced in young adult mice at 2 months and continued until advanced age. Mice were fed with high Zn (1,000 ppm, 500 ppm or 300 ppm) or Cu (150 ppm and 100 ppm) and euthanasised at 5, 7, 11 or 17 months of age (Table 1). We observed a trend but no significant differences in brain metal levels with the small numbers of mice that were sacrificed in each group during the feeding period to monitor whether there was any pathology developing. Therefore, to obtain an overall indication of whether the high metal diets had affected brain metal levels, a combined analysis of data from all metal feeding groups was performed after normalisation (Fig. 1). Concentrations of Cu, Zn and the Cu/Zn ratio in the brains of mice were normalised to the average values of the corresponding control group, matched for age, sex, genotype and base drinking-water type (tap or demineralised). The percentage difference from control values for each mouse was then used for an overall combined analysis of the effects of Zn and Cu feeding on brain Zn and Cu levels using two-way ANOVA. Mouse strain was included as an independent variable to determine whether APP.C100 transgene expression had any effect on brain metal level changes. Differences within each mouse strain were tested for significance with a post-hoc Scheffé test. Mice fed high Zn diets comprised: TgC100 n=31 and non-transgenic (NTg) n=19, with demineralised or tap water controls comprising: TgC100 n=25 and NTg n=19. Mice fed high Cu diets comprised: TgC100 n=11 and NTg n=14, with demineralised water controls comprising TgC100 n=11 and NTg n=9.

Table 1.

Summary of mouse numbers in feeding groups used for metal analysis

High Zn diet	Zn 300 ppm		Zn 500 ppm			Zn 1000 ppm
Age at end of feeding	7 months	17 months	5 months	11 months	17 months	11 months	17 months
TgC100	n=2 Zn	n=7 Zn	n=2 Zn	n=4 Zn	n=6 Zn	n=2 Zn	n=8 Zn
	n=2 water	n=5 water	n=2 water	n=4 water	n=5 water	n=2 water	n=5 water
BL6/DBA	n= 2 Zn	n=3 Zn	n =2 Zn		n=8 Zn	n= 2 Zn	n= 2 Zn
	n=2 water	n=3 water	n=2 water		n=7 water	n=2 water	n= 3 water

High Cu diet	Cu 100 ppm		Cu 150 ppm
Age at end of feeding	5 months	17 months	11 months
TgC100	n=2 Cu	n=5 Cu	n=4 Cu
	n=2 water	n=5 water	n=4 water
BL6/DBA	n=2 Cu	n=12 Cu
	n=2 water	n=7 water

Metal feeding was commenced at ~7 weeks and continued to 11 or 17 months when the mice were euthanased for analysis. A subset of mice was euthanased and analysed at 5–7 months to check for any pathology developing

Fig. 1.

High dietary copper intake increases brain copper levels. Brain Zn and Cu levels were measured in TgC100 and NTg (BL6/DBA) mice fed various high Zn diets [300–1,000 ppm] (a) and high Cu diets [100–150 ppm] (b) and compared with age-matched control drinking water mice. Metal levels are expressed as ‘percentage difference from control feeding group’, in box and whisker plots. Significant differences between mice fed high metal diets and controls ■ were determined by ANOVA with genotype as an independent variable, and values are indicated above each box. Post-hoc Scheffé tests were used to test for significant differences in TgC100 and NTg mouse lines individually (*P < 0.05). Boxes represent average ±SEM, and whiskers represent 1.96*SEM

A significant overall increase in brain Zn levels (5±1%, P = 0.001) was found in high Zn fed mice, with no significant effect of mouse strain. Both TgC100 and NTg mice showed elevated brain Zn levels in high Zn drinking water groups (6±1 and 3±1%, respectively), however, this increase only reached statistical significance in the TgC100 line (P < 0.05; Fig. 1a). Cu levels in the brain were not significantly altered by high Zn feeding, although a non-significant decrease in Cu was observed in TgC100 mice fed high Zn diets (−6±3%, P = 0.07), whereas levels were unchanged in NTg mice (0±2%). There was no statistically significant mouse strain effect on Cu level alterations due to high Zn feeding (Fig. 1a). The Cu/Zn ratio in the brains of mice fed a high Zn diet was significantly decreased compared with water controls (−8±2%, P = 0.004). Although both TgC100 and NTg mouse lines showed a decreased average ratio of Cu/Zn levels due to a high Zn diet, this decrease was greater and reached statistical significance only in the TgC100 line (−11±3%, P < 0.05). However, no statistically significant effect of mouse strain was found.

High dietary Cu feeding resulted in non-significantly elevated brain Cu levels in both TgC100 and NTg mice (5±6 and 5±2%, respectively). These non-significant Cu elevations were however, accompanied by a similar elevation in brain Zn levels (4±1 and 5±1% in TgC100 and NTg lines, respectively, P < 0.05 by two way-ANOVA). The similar elevations in Cu and Zn levels resulted in an unchanged Cu/Zn ratio. No significant effects of mouse strain were observed for either metal or the Cu/Zn ratio with high Cu feeding (Fig. 1b).

Human Aβ Levels in Mice Fed High Dietary Zinc or Copper

To determine whether metal feeding had altered brain Aβ levels or solubility in TgC100 mice, a sandwich ELISA for the detection of Aβ was employed. Whereas Aβ40 levels were clearly detectable, the Aβ42 species was beneath detection limits and hence could not be quantitated as reported previously (Li et al. 2006). The sandwich ELISA used mAbG210 for capture of Aβ40 C-termini, and mAbWO2 (recognises hAβ5-8) was employed for detection. Total brain homogenates, and PBS-soluble fractions from 17 month TgC100 mice that had been fed either high Zn (500 ppm, n = 11), high Cu (100 ppm, n = 9) or demineralised water (n = 7) were analysed.

Aβ40 levels in total homogenates showed no significant alterations due to high Zn or Cu feeding (n.s. increases of 6 and 5%, respectively) whereas, soluble Aβ40 levels tended to be decreased in Zn and Cu fed groups compared with demineralised water controls. High Cu feeding resulted in a significant 18% decrease in soluble Aβ levels (P = 0.02), and high Zn feeding resulted in a non-significant 13% decrease in soluble Aβ40 levels (P = 0.070; Fig. 2a). We have determined that soluble Aβ comprises ~30% of the total Aβ pool in TgC100 mice, consistent with previous data (Li et al. 1999). This explains why the impact of a reduction in soluble Aβ upon the total Aβ pool would be of smaller magnitude.

Fig. 2.

ELISA detection of Aβ40 in mice fed high zinc or copper diets. a Aβ40 levels were measured in total brain homogenate and PBS-soluble fractions of TgC100 mice fed either 500 ppm Zn, 100 ppm Cu or pure dH₂O up to 17 months of age. Aβ40 concentration is reported as ng Aβ40 per g total protein. Asterisk represents significant difference from water control by two-tailed t-test (*P < 0.05). Error bars represent SEM

Protein Markers of Cellular Stress, or Neuronal Loss in TgC100 Mice Fed High Copper or Zinc

A number of marker proteins were measured in brain homogenates by Western blotting, as indicators of various cellular stresses that may result from altered brain metal homeostasis. Glial fibrillary acidic protein (GFAP) was measured as an indicator of glial-cell proliferation, a general marker of cellular stress or injury, the 200 kDa neurofilament protein (NF200) was measured to detect neuron loss, and SOD1 (Cu/Zn superoxide dismutase) was measured to detect an elevation in oxidative stress levels. Endogenous mouse APP expression and the transgene product C100 was also measured since APP and C100 are known to be involved in metal regulation (Maynard et al. 2002; White et al. 1999). Densitometric analysis of band intensities revealed no significant differences in NF200, GFAP, SOD1, APP or C100 in the brain homogenates of TgC100 mice fed high Zn (500 ppm) or Cu (100 ppm) diets compared with water controls (Fig. 3). Furthermore, these protein markers were also not altered in the NTg mice fed with 500-1,000 ppm Zn or 100 ppm Cu (not shown).

Fig. 3.

Protein markers in the brains of 17-month TgC100 mice fed high zinc or copper. Brain homogenates from 17-month TgC100 mice fed 500 ppm Zn, 100 ppm Cu or pure dH₂O were immunoblotted for detection of NF200, GFAP, SOD1, APP and C100. β-tubulin was measured as a loading control. a Representative lanes of Western blot from mice fed water control, Zn 500 ppm (Zn) and Cu 100 ppm (Cu) diets. Densitometric analysis was performed on band intensities from mice of both sexes fed 500 ppm Zn (n=11) (b), and 100 ppm Cu (n=9) (c) compared with water controls (n=7). No significant differences were found in any of the proteins by t-test. Error bars represent SEM

Immunohistochemical Analysis of Aβ and GFAP in TgC100 Mice Fed High Zinc or Copper in Drinking Water

Immunostaining of brain sections from 17 month TgC100 mice fed high Zn (500 ppm), high Cu (100 ppm) or pure dH₂O (n = 5, 4 and 2, respectively), showed no marked differences in the intensity or distribution of Aβ staining due to metal feeding (Fig. 4). TgC100 mice exhibited dense granular staining in the perinuclear region of pyramidal neurons of the CA2 region of the hippocampus, as observed previously in younger mice of the same strain (Li et al. 1999). Numerous glial cells in the striatum and the occasional glial cells in cortex also showed dense perinuclear staining (Fig. 4b), although there were no apparent differences in mice fed high Zn or Cu diets. Importantly, we detected no Aβ -amyloid deposits.

Fig. 4.

Aβ accumulation and GFAP distribution are not affected by zinc or copper feeding in mouse brain. TgC100 mice were given pure dH₂O containing no additives (left), 500 ppm Zn (centre) or 100 ppm Cu (right) as their sole source of drinking water, from the age of 2 months until euthanisia at 17 months. Brain sections show representative female mice from each feeding group, immunostained with WO2 (a and b) or anti-GFAP (c and d). a 20× magnification of hippocampus (CA2 region). b 20× magnification of striatum. Bars, 0.1 mm. NTg brain shows no WO2 immunoreactivity (not shown). Brain sections of NTg (c) and TgC100 (d) mice immunostained with GFAP antibody. Brain sections in figures show representative sections from mice examined in each group (BL6/DBA n=3, 4 and 4, TgC100 n=2, 5 and 4 in water control, Zn, and Cu fed groups, respectively). Bar, 2 mm

Immunostaining for GFAP was performed to stain glial cells, the proliferation of which is an indication of stress, damage or toxicity. GFAP immunoreactivity was compared between TgC100 and NTg mice fed high Zn (500 ppm), high Cu (100 ppm) and pure dH₂O controls. No marked differences in the intensity or distribution of GFAP staining were observed due to high Zn or Cu feeding in agreement with Western blot data. Furthermore, we observed no differences in GFAP staining between TgC100 and NTg lines. This latter observation indicates that the level of expression of the human APP.C100 fragment and its Aβ product in these mouse models, are sub-pathological in mice as old as 17 months. Neither chronic exposure to high levels of Zn nor Cu in drinking water up to 17 months, resulted in any marked glial or astrocyte proliferation in NTg (BL6/DBA) mouse brain, or in TgC100 mouse brain overexpressing human Aβ. TgC100 mice fed with 1,000 ppm Zn also displayed no distinct differences in Aβ immunoreactivity compared with controls.

Discussion

Chronic exposure to high levels of Zn or Cu in drinking water resulted in small elevations in brain levels of the respective metals, and a 13–18% decrease in soluble Aβ levels.

Exposure to chronic high Zn in the drinking water resulted in a small 5±1% (P > 0.05) elevation in brain Zn levels and a concomitant decrease in Cu resulting in an 11% reduction in the Cu/Zn ratio in mice overexpressing APP.C100 (P > 0.05). Non-transgenic mice displayed a smaller and non-significant Zn elevation accompanied by no alteration in Cu levels. Chronic exposure to high Cu levels in the drinking water resulted in 5% elevations in average brain Cu levels in both APP.C100 and NTg mice, and this was matched by 5% elevations in brain Zn levels in both mouse lines, and thus no alteration in the Cu/Zn ratio.

Previous metal feeding studies using rats have demonstrated increased brain Cu levels (11%) as a result of a high Cu diet, and increased brain Zn levels (13%) as a result of a high Zn diet (Davis 1997). The high Zn diet of Davis (1997) also resulted in a trend towards decreased brain Cu levels, as was found in this study. The antagonistic affect of Zn on Cu levels is well known, and appears to be mediated primarily by competition for uptake at the level of the intestinal mucosa (Fischer et al. 1984). In contrast to the abovementioned rat studies, but in consolidation with the present mouse study, previous studies using mice have reported no change in brain Cu levels due to life-long administration of >300 ppm Cu in drinking water (Massie and Aiello 1979), or systemic administration of Cu approaching lethal doses (Matsuda et al. 1989), suggesting tighter regulation of brain Cu levels in mice than rats. The analysis of large numbers of mice in this study that had been chronically exposed to high levels of Zn or Cu, indeed revealed only minor alterations in brain Zn and Cu level that would not be detected using smaller mouse numbers.

Considering that APP.C100 expressing mice already exhibit decreased brain Cu levels (Maynard et al. 2002), one might consider that increasing Cu availability through the diet may be able to supplement this proposed deficit. However, the elevation of Cu in TgC100 mice was not significantly greater than in the NTg controls. A recent study found that supplementation of Cu in the drinking water of transgenic APP23 mice, which also exhibit decreased brain Cu levels, rescued them from glucose-induced mortality (Bayer et al. 2003) suggesting correction of the Cu deficit. However our data do not support any such normalisation of TgC100 brain Cu levels via high dietary Cu feeding.

Of importance from this study is the finding that brain Zn and Cu homeostasis was modified by chronic feeding of high levels of Zn and Cu in the drinking water of TgC100 mice, albeit by a small amount. Immunohistochemical analysis however, revealed no induction of amyloid pathology and no alterations in Aβ staining in TgC100 brain due to high Zn or Cu feeding.

In our study, human Aβ was overexpressed in the absence of any amyloid-promoting FAD mutations, as a model of sporadic, rather than familial AD. None the less, a significant decrease in Aβ40 levels in the PBS-soluble fraction was observed due to high Zn or Cu feeding. A similar decrease in soluble Aβ levels was also observed by Western blot analysis (data not shown). This decrease could represent either decreased generation of Aβ, increased clearance of soluble Aβ, an increased recruitment of Aβ into the PBS-insoluble fraction, or a decreased detectability of Aβ. Since PBS-soluble Aβ represents only around 30% of total Aβ in TgC100 brain, the effect on the total homogenate Aβ levels and immunohistochemically detectable Aβ may be undetectable. A decrease in the PBS-solubility of Aβ in Cu fed mouse brain may reflect increased oligomerisation of Aβ, or incorporation into insoluble aggregates, which could be elicited by altered metal homeostasis. The decrease in detectable Aβ by both ELISA and Western blot methods could also be caused by oxidative modifications to Aβ (Atwood et al. 2000a, b; Schoneich and Williams 2002) blocking the WO2 epitope. The formation of Aβ dimers, or association with small aggregates in the PBS-soluble fraction, could also contribute to the decreased detection of Aβ under the native ELISA conditions. Alternatively, the reduction in soluble Aβ levels detected may be a result of accelerated degradation by metalloproteinases (White et al. 2006). Indeed Cu and Zn levels have been recently shown to regulate Aβ degradation (Strozyk et al. 2007).

No evidence for any toxic effects of the small alterations in brain Zn/Cu homeostasis in TgC100 mice were found by measurement of an array of marker proteins for oxidative stress, proliferation or glia, or neuron loss. This indicates that either the Cu and Zn alterations were not great enough to cause damage or that sufficient cellular compensation had occurred. The level at which elevated Cu would become toxic in the brain however remains unclear, as it likely relates to the capacity for binding or sequestering the excess Cu, and for quenching any excess reactive oxygen species. Remarkably, chronic Cu overload in fibroblasts of rat has shown no evidence of oxidative stress (Aburto et al. 2001; Armendariz et al. 2004), further indicating tight intracellular regulation of this redox-active metal.

If the reduction in the soluble pool of Aβ we observed in the brains of high Cu fed mice reflects a shift to the insoluble fraction, the level of Aβ is evidently beneath the threshold required to seed amyloid deposition in TgC100 mice, which express low human Aβ levels compared with other transgenic mouse lines that do develop amyloid-like pathology with age (Li et al. 1999), and importantly, these mice express no pro-amyloidogenic FAD mutations. Were similar feeding experiments performed using a mouse model that expresses higher Aβ levels and develops amyloid plaques, such as the Tg2576 mouse model, it is possible that such a shift in Aβ solubility may be sufficient to result in a measurable effect on amyloidogenesis.

Dietary metals like Cu and Zn may play important, but conflicting roles in the development of AD pathology. Confounding the arguably undesirable action of Cu and Zn in the formation of amyloid deposits and free-radical generation, is the apparent Cu deficit in the AD and transgenic mouse brain that may contribute to the malfunctioning of normal cellular processes in AD. The development of therapeutic strategies targeting metals must therefore take into account these opposing factors. A promising finding was that treatment of aged Tg2576 mice with the Cu–Zn chelator clioquinol reduced amyloid plaque load, and increased total brain Cu levels (Cherny et al. 2001). Furthermore, the second-generation 8-hydroxy quinoline analogue PBT2 also targets metal-induced aggregation of Aβ and restores cognitive function in the AD mice (Adlard et al. 2008).

The findings of this study demonstrate that although the brain’s Cu and Zn levels are only altered to a small extent due to chronic exposure to high metal levels in the drinking water, this was sufficient to modulate the solubility of wild-type human Aβ in mice. Over a longer time period, and in a more pro-amyloidogenic environment, it is conceivable that alteration of Aβ solubility might hasten amyloidogenesis.

Abbreviations

AD

Alzheimer’s disease

Aβ

Amyloid β-peptide

APP

Amyloid precursor protein

APP-C100

The C-terminal fragment of amyloid precursor protein containing the Aβ peptide, also referred to as βCTF

GFAP

Glial fibrillary acidic protein

NTg

Non-transgenic

SOD1

Cu/Zn superoxide dismutase

SDS

Sodium dodecyl sulfate

PBS

Phosphate buffer saline