Serum and brain natural copper stable isotopes in a mouse model of Alzheimer’s disease

Alzheimer’s disease is associated with the production of Cu rich aβ fibrils. Because monitoring the changes in Cu level of organs has been proposed to follow the evolution of the disease, we analyzed the copper isotopic composition of serum and brain of APPswe/PSEN1dE9 transgenic mice, a model of Alzheimer’s disease, and wild-type (WT) controls. Serum composition of 3, 6, 9 and 12-month-old mice, as well as the composition of 9 brains of 12-month-old mice are reported. In WT mice, brains were ~1‰ isotopically heavier than serum, and the Cu isotopic composition of the serum was isotopically different between males and females. We propose that this effect of sex on the Cu isotopic budget of the serum may be related to a difference of Cu speciation and relative abundance of Cu carriers. Brains of APPswe/PSEN1dE9 mice were slightly lighter than brains of WT mice, while not statistically significant. This trend may reflect an increase of Cu(I) associated with the formation of Aβ fibrils. The Cu isotopic composition of the brains and serum were correlated, implying copper transport between these two reservoirs, in particular a transfer of Cu(I) from the brain to the serum. Altogether, these data suggest that Cu stable isotopic composition of body fluid may have the potential to be used as detection tools for the formation of Aβ fibrils in the brain, but further work has to be done.

Alzheimer's disease (AD) is one of the main causes of death in high-income countries and the fifth leading cause of death of US citizen over 65 1 . It is characterized by a loss of brain function affecting reasoning, memory and comportment. The major physiological features of AD are the formation of neurofibrillary tangles by neuronal accumulations of abnormal hyperphosphorylated tau filaments and the formation of senile plaques by extracellular deposits of amyloid β (Aβ) fibrils, mostly the 1 to 42 peptide (Aβ . Early diagnosis of AD is a major challenge and current work focuses on the detection of excess of total tau and Aβ in cerebrospinal fluid, and on using imaging of the brain by positron emission tomography and magnetic resonance imaging (MRI) for Aβ plaques (e.g. refs [2][3][4]. The homeostasis of metals, such as Fe, Zn, Al or Cu, is modified in patients with AD (e.g. ref. 5 ). These metals are concentrated in the Aβ plaques and in consequences in the neocortex of AD patients (e.g. refs 6,7 ). The changes in metal homeostatis in AD have been tested as a diagnosis tool as the concentrations in the Aβ plaques should impact the concentration in other organs and fluids (e.g. red blood cells and serum). However, studies focussing on elemental abundance measurements in the serum of AD patients are inconsistent, showing overall Zn abundance decreasing 5 , increasing 8 or unchanged 9 . Elemental abundance variations in the serum may be affected by factors unrelated to AD, such as intestinal absorption of Zn 10 , which greatly complicates the interpretation of this data for AD diagnosis.
We and others recently showed that the natural stable isotopic composition of metals such as Zn, Cu and Fe naturally varies between body organs [11][12][13] . In particular, blood (red blood cells and/or serum) are isotopically distinct in Zn, Cu and Fe from the brain [11][12][13] ). Isotopic fractionation is due to the difference of bonding environments of the elements between different organs 12 . For example, in normal conditions, Zn is preferentially bound to cysteine-rich protein (metallothionein) in the brain, which preferentially concentrates light isotopes, while in red blood cells (RBCs) Zn is bound to histidine-rich proteins (carbonic anhydrase) that concentrate heavy isotopes 12,13 . On the other hand, in Aβ plaques Zn is bonded to isotopically heavy amino-acids (histidine and glutamate) 14  studied the Zn isotopic composition of brain and blood samples during the development of transgenic APPswe/ PSEN1dE9 and wild type controls mice and found that, as predicted, Zn was isotopically heavier in the brain of AD mice than in age-matched wild types. However, the isotopic variations among serum samples were not large enough to allow for the detection of an isotopic effect.
While Zn is only present under one redox state (Zn 2+ ), Cu is present in both 1+ and 2+ forms. The presence of redox variations of Cu potentially enhances the magnitude of isotopic variations compared to Zn (e.g. see ab initio calculations for various Cu(I) and Cu(II) species in ref. 16 ). In addition, Balter et al. 13 showed that Cu is isotopically heavier in the brain of mice than in the whole blood (by about 1 permil for the 65 Cu/ 63 Cu ratio) suggesting that, as done with Zn isotopes 15 , Cu isotopes could be used as tracers of metal dyshomeostasis in the brain. This potential has been confirmed by the discovery of Cu isotopic fractionation in the cerebrospinal fluid of patients with amyotrophic lateral sclerosis compared to control patients 17 .
In the case of AD, Cu participates in the structure of the Aβ plaques; Cu is enriched in the plaques in mice (e.g. ref. 18 ) and humans 19 compared to normal adjacent brain zones. Copper bound to the Aβ plaques could be formed under both +I and +II form 20,21 . In the brain copper is principally stored in proteins under its reduced form Cu(I) binded to metallothionein and glutathione both cystetine-rich proteins 22,23 and a fraction of Cu(II) is present in synapses 24 . The accumulation of Cu(II) in Aβ plaques changes the speciation of Cu in the brain in the brain which may affect Cu in body fluids such as blood, as suggested from the global Cu level increase observed in the serum of AD patients 25 . This change may be traceable with Cu stable isotopic composition. Furthermore, formation of Aβ plaques affect the global redox state of Cu in the brain, and can lead to the increase of reactive oxygen species, oxidative stress and loss of cognitive functions (e.g. ref. 26 ).
Here we tested whether Cu stable isotopic composition of serum and brain are modified by the formation of Aβ plaques in mice with late stage AD. We report the Cu isotopic composition of the serum from 23 transgenic APPswe/PSEN1dE9 27,28 and wild type (WT) mice at various stages (3 month, 6 month, 9 month and 12 month). In addition, we report the Cu isotopic composition of the brains of 9 of the 12-month mice. Mice. The animals used for this study correspond to some of the mice reported in Moynier et al. 15 , for which we still had samples after the analysis of Zn isotopic composition: APPswe/PSEN1dE9 (AD) and wild-type (WT) littermate controls on a C57BL/6 and C3H mixed background. At 3, 6, 9 and 12 months, blood samples were collected. After 12 months, the mice were killed and their brains were collected. All mice were housed under specific pathogen-free conditions in the Washington University animal facilities in accordance with institutional guidelines, and were given the same diet from birth. The diet is the one reported in Moynier et al. 15 . AD mice were originally obtained from The Jackson Laboratory, then bred and housed in the Washington University animal facilities by crossing to B3C3F1/J (from The Jackson Laboratory).

Method
Sample collection. Samples preparation was described in Moynier et al. 15 . Mice were anesthetized with ketamine and xylazine. Anesthesia was assessed by the toe pinch method. The thoracic cage was opened and blood was collected in polypropylene microtubes by a cardiac puncture. The vena cava was sectioned under the kidneys, and PBS was injected in the left ventricle for organ perfusion. Death was assessed by cervical dislocation. Serum was separated from red blood cells by centrifugation. After blood collection, mice were perfused by injecting phosphate buffered saline (PBS) through the heart with a dissected hepatic vein to remove blood from organs. Brains were harvested using instruments in stainless steel. All samples were stored in polypropylene microtubes or cryogenic vials (Corning). Harvested organs were snap-frozen and kept frozen until dissolution.
Chemical purification and mass-spectrometry. The Cu purification and isotopic measurements were performed at the Institut de Physique du Globe de Paris (IPGP). The cleaning and dissolution of the samples is similar as the one used previously for Zn isotopic measurements 15 . Brain were cleaned twice using 18.2 MΩ cm water and the brain and serum were dissolved using a mixture of HNO 3 (4 times sub-boiled) and Seastar© optima grade H 2 O 2 in closed Teflon beakers for one to two days. This method provides a complete dissolution of the organs with minimum contamination (as opposed to the combustion of the organs in an oven).
Copper is purified by ion exchange chromatography following a procedure adapted from Maréchal et al. 29 and similar to what we used recently 30,31 . The samples are loaded in 7 N HCl on a 1.6 mL AG-MP1 resin. The resin is further washed by passing 8 mL of 7 N HCl and the Cu is collected in 22 mL of 7 N HCl. This procedure is reproduced twice to ensure a clean Cu fraction. Copper isotopic data were measured using the MC-ICP-MS (Thermo-Finnigan Neptune Plus) located at IPGP following the method used in Savage et al. 31 . Briefly, copper is injected using a glass spray chamber and a 100 μL/min ESI Teflon micro nebulizer. The MC-ICP-MS was operated in low resolution mode with 65 Cu and 63 Cu collected in the central and L2 faraday cups, respectively.
The sample dilutions were adjusted (to within ~ ±5%) to match the concentration of the standard. The total yield of Cu is >99%. The blank of the full procedure is ~5 ng which is negligible in comparison to the amount of Cu (>200 ng) present in each samples. Typical external reproducibility estimated from numerous replicates was estimated to be better than 50 ppm (2 standard deviation) 31 .
The isotopic composition of Cu is reported as parts per 1,000 deviations relative to the NIST SRM 976 standard 16 .: We used the standard bracketing method that consists of measuring a standard before and after each sample and use the average of the two standards for the normalizing ratio 16 . During analysis session the standard utilized was IPGP-Cu and relative to NIST SRM 976, the Cu_IPGP standard has δ 65 Cu = +0.271 ± 0.006‰ (2sd; n = 55). All the data presented here were converted to δ 65 Cu relative to NIST SRM 976 by adding 0.271. Statistical analysis. Statistical analysis was performed using GraphPad Prism. Data are represented as box-and-whiskers graphs, with the box always extending from the 25th to 75th percentiles, the line representing the median, and the whiskers showing the minimal and maximal values. Mann Whitney non-parametric tests were used to compare two groups of unpaired data sets. Two-way ANOVAs were performed to test the effect of various factors (age, sex and genotype), followed by Sidak's multiple comparison test, or Wilcoxon matched-pairs signed rank test for paired data sets.

Results
Distinct copper isotopic composition of serum and brain in wild-type mice. We measured the Cu isotopic data for 75 serum and 9 brain samples, reported in Table S1. The serum samples represent nineteen 3-month-old, twenty-three 6-month-old, twenty 9-month-old and thirteen 12-month-old mice. The nine brains have been taken from 12-month-old mice. The average of the different groups of samples sorted by ages and sex are reported in Table 1.
Old AD mice have a slightly lighter (but not statistically different) copper isotopic composition in their serum and brain.
We investigated if AD would lead to a different copper isotopic composition in serum and/or brain by analyzing the isotopic composition of organs from the APPswe/PSEN1dE9 transgenic mouse model. The serum of 12-month-old AD females (δ 65 Cu = −0.59 ± 0.28, sd) was isotopically lighter than the serum of age-matched WT female littermates (δ 65 Cu of −0.31 ± 0.19, sd), even if the difference is not significant (p = 0.20, Sidak's multiple comparison test) (Fig. 3B). There was no other significant difference at any age between WT and AD mice in serum. Similarly, the brains of the 12-month-old AD females (δ 65 Cu = 0.51 ± 0.15, sd) also appear modestly lighter (but not statistically significant) than the brain of the female WT littermate (δ 65 Cu = 0.63 ± 0.26, sd; see Fig. 3C).

Discussion
Isotopic difference between brain and serum. Our observation that brains (9 samples) are typically 1‰ heavier than serum (74 samples) in mice is consistent with previous data obtained on a smaller sample pool of samples. Balter et al. 13 reported 4 brain samples and 9 whole blood samples from 5-month-old mice with an isotopic difference of ~0.8‰. On the other hand, human serum are heavier than what is reported here for mice by ~0.5‰ (e.g. refs [32][33][34][35] ).
The differences of Zn isotopic composition between organs were previously attributed to the difference of bonding environment 12,13 . In mice, Zn is isotopically lighter in brain than in serum, which is explained by the fact that Zn is principally bound to cysteine-rich protein in the brain (metallothionein) while it is bound to histidine-rich protein in the serum (albumin) 36 . Copper shows the reverse isotopic effect (brain heavier than serum). The bonding properties and the associated isotopic fractionation is more complex in the case of Cu than for Zn because of the multiple redox state of Cu (I and II). Copper is principally found as Cu(I) in metallothionein and glutathione in the brain 23 and as Cu(II) in synapses 24 . In human's serum, Cu is mainly bound to  www.nature.com/scientificreports www.nature.com/scientificreports/ ceruloplasmin (~90%; ref. 37 ) with the rest being bound to β-macroglobulin, albumin, and in some small molecules [37][38][39] . In ceruloplasmin, Cu is principally found under its +II form and bounds to cysteine and histidine 40 . It is also found under its +II in β-macroglobulin 41 , while in albumin it could be under the +I form 42 . Since the  Correlation of the Cu isotopic composition between brain and serum. Data show δ 65 Cu values from serum of WT and AD 12-month-old female mice. Linear regression showed a significant positive correlation between serum and brain δ 65 Cu values for WT animals (r 2 = 0.97; slope 1.19 ± 0.14, se, p = 0.01), but only a trend for AD animals (r 2 = 0.62; slope 0.49 ± 0.44, se, p = 0.38). www.nature.com/scientificreports www.nature.com/scientificreports/ majority of Cu in the brain should be stored under its isotopically light reduced form in metallothionein or glutathione 22,23 while it should be mostly found as its isotopically heavy oxidised form in ceruloplasmin in the serum, the clearly isotopically heavier brain is surprising. In the absence of quantitative study on the speciation of Cu in the brain it is not possible to conclude on this topic here, but one possible explanation to this conundrum is that Cu(II) involved in synaptic transmission 24 represent a large fraction of the total Cu stored in the brain.
The observation that serum in human (e.g. ref. 35 ) is isotopically heavier than in mouse by ~0.5‰ suggests that the speciation of Cu differs in mice compare to humans. This is supported by previous HPLC-ICPMS measurements on plasma doped with 65 Cu 37 who found that ceruloplasmin is a more important ligand in human's plasma compared to mice's plasma. We can therefore conclude that Cu speciation in mouse serum is different than in human. Therefore, it will be important to test other AD animal models (e.g. minipigs 43 ) as well as human patients.
The correlation that is observed between the δ 65 Cu of the brain and serum of 12-month old mice (Fig. 2) suggests that the Cu from which we analysed the isotopic composition in the serum has been interacting with the brain and therefore must be mobile. Copper bound to ceruloplasmin is not mobile, as it was demonstrated that Cu homesotatis is unchanged in ceruloplasmin-deficient mice compared to WT 44 . This implies that ceruloplasmin does not play a role in the transport and transfer of Cu between serum and organs. Therefore, low molecular weight molecules must carry the mobile Cu fraction. This is concordant with the fact that the main Cu transporter, CRT1, bounds both Cu(I) and Cu(II) and can bound large fraction of Cu in the serum 45 .
Effect of sex. The systematic difference of ~0.4‰ of the δ 65 Cu between WT male and female mice suggests an effect of sex on the Cu isotopic budget of the serum. Such a sex difference has been previously observed for Cu concentration in mouse plasma 46 . Our data represent the first example of a significant isotopic difference for mice of different sex, even if the limited data from previous work on Cu isotopes in mice hinted that there was a heavy isotope enrichment in the blood of females compared to males (~0.15‰; ref. 13 , n = 5 for the females and 4 for the males, compared to n = 19 and n = 13, respectively, in the present study). Such an effect of sex was not observed for Zn isotopes in mice 12,13,15,47 .
In human, a difference between sexes was previously reported for Fe and Cu isotopes [47][48][49] . A difference of ~0.20‰ in Cu isotopic composition was also previously seen between men and premenopausal women in the whole blood (−0.52 vs −0.74) 50 . The origin of the Cu isotopic fractionation between men and premenopausal women is usually attributed to the larger Cu turnover in women due to menstruations (e.g. refs 51,52 ). Such origin is not applicable to male and female mice since the reproductive cycle of mice does not include bleeding 53 . The sex isotopic difference in mice may be related to a difference of Cu speciation and relative abundance of Cu carrier in serum. However, to our knowledge, there is no study on the effect of sex on the speciation of Cu in mouse blood.
Effect of AD on the Cu isotopic composition of the brain and serum. Twelve-month old AAPswe/ PSEN1de9 female mice have slightly isotopically lighter brains in average than WT, even if this difference is not statistically significant. Isotope fractionation in 12-month old AAPswe/PSEN1de9 mice was previsouly observed for Zn 15 . Aβ plaques starts to develop in AAPswe/PSEN1de9 mice after 6 months with profuse plaques in the hippocampus and cortex at nine months 27 and a proliferation until 12 months 28 . The small enrichment in light Cu isotopes of 12-month AAPswe/PSEN1de9 mice brains compared to WT suggests that the change of speciation associated with the proliferation of Aβ plaques modifies the global Cu isotopic composition of the brain. It is known that the formation of Aβ plaques increases the production of free Cu+ in the brain and therefore induces oxidative damages 26 . Such increase in Cu+ in the brain should globally decrease the δ 65 Cu value of the whole brain. On the other hand, copper is primarily bound to metallothionein in normal brain while in Aβ plaques it is bound to histidine 54,55 . Metallothionein is a cysteine rich protein and at 298 K it has a logarithm of the reduced partition function (lnβ) of ~3.3 (refs 16,56 ) while in the same conditions lnβ of histidine is ~4.4. As a result, Aβ plaques should be isotopically heavier than normal brain. However, it was also shown that Aβ has a bounding site that favors Cu(I) 20 , which would conversely enrich the Aβ in the light isotope of Cu. If confirmed, the enrichment in the light isotopes of the AD mouse brains would imply that the production and increase of free Cu+ ion in the brain globally, together with the bounding of Cu(I) to Aβ, decrease the 65 Cu/ 63 Cu ratio.
Similarly to the brain, the Cu isotopic composition of the serum of the 12-month old AAPswe/PSEN1de9 mice tends to be isotopically different from the WT mice (but again statistically not different), while it is unchanged from 3 to 9 month in both the WT and AD mice. It is enriched in the lighter isotopes compared to WT mouse serum, such that the δ 65 Cu value in the serum and in the brain correlate (Fig. 2). This correlation between the δ 65 Cu value of the brain and the serum of the 12-month-old mice suggests that the change of composition of the serum is linked to the change happening in the brain. It was actually shown that the Cu level of the serum is higher in AD patients than in controls 25,57 . In particular, it was shown that AD serum has an increase of non ceruloplasmin Cu. An increase of Cu(I) species would decrease the total δ 65 Cu value of the serum. The origin of this non ceruloplasmin Cu was actually suggested to be a transfer of Cu(I) from the brain to the serum 20,58 which would explain our observations. conclusion Copper isotopic composition of serum is a potential tool to study the change of homeostasis associated with AD. Here we show that the δ 65 Cu value of WT males is ~0.4 different from WT females. This suggests an effect of sex on the Cu isotopic budget of the serum, which may be related to a difference of Cu speciation and relative abundance of Cu carrier in serum. We also find a trend that suggests that Aβ plaques induced a change in the Cu isotopic composition of the whole brain and of the serum. This tendency could be explained by the increase of isotopically light Cu(I) species in the brain. The Cu isotopic composition of the brain and of the serum correlate, which can be explained by a transfer of Cu(I) species from the brain to the serum and suggest that monitoring