ReportsAlzheimer’s Disease

Changes in Amyloid-β and Tau in the Cerebrospinal Fluid of Transgenic Mice Overexpressing Amyloid Precursor Protein

See allHide authors and affiliations

Science Translational Medicine  17 Jul 2013:
Vol. 5, Issue 194, pp. 194re2
DOI: 10.1126/scitranslmed.3006446

Abstract

Altered concentrations of amyloid-β (Aβ) peptide and Tau protein in the cerebrospinal fluid (CSF) are thought to be predictive markers for Alzheimer’s disease (AD). Transgenic mice overexpressing human amyloid precursor protein (APP) have been used to model Aβ pathology, but concomitant changes in Aβ and Tau in CSF have been less well studied. We measured Aβ and Tau in the brains and CSF of two well-characterized transgenic mouse models of AD: one expressing human APP carrying the Swedish mutation (APP23) and the other expressing mutant human APP and mutant human presenilin-1 (APPPS1). Both mouse models exhibit Aβ deposition in the brain, but with different onset and progression trajectories. We found an age-related 50 to 80% decrease in Aβ42 peptide in mouse CSF and a smaller decrease in Aβ40, both inversely correlated with the brain Aβ load. Surprisingly, the same mice showed a threefold increase in total endogenous murine Tau in CSF at the stages when Aβ pathology became prominent. The results mirror the temporal sequence and magnitude of Aβ and Tau changes in the CSF of patients with sporadic and dominantly inherited AD. This observation indicates that APP transgenic mice may be useful as a translational tool for predicting changes in Aβ and Tau markers in the CSF of AD patients. These findings also suggest that APP transgenic mouse models may be useful in the search for new disease markers for AD.

INTRODUCTION

The pathology of Alzheimer’s disease (AD) is thought to start 10 to 20 years before the onset of the first clinical symptoms in both sporadic and familial AD patients (13). Thus, disease-modifying compounds are probably most effective when given at a preclinical disease stage, before neurodegeneration has become severe enough to induce a clinical phenotype. Amyloid-β (Aβ) peptide and Tau protein, the constituents of the pathological hallmarks of AD, amyloid plaques and neurofibrillary tangles, respectively, have shown promise as cerebrospinal fluid (CSF) markers of early AD (4, 5). Lower concentrations of CSF Aβ42 (the Aβ species that stops at amino acid 42) and higher concentrations of total Tau protein (t-Tau) have been used to distinguish AD patients from cognitively normal age-matched controls and to predict the conversion of mild cognitive impairment (MCI) to AD (2, 6). It has been presumed that the decrease in CSF Aβ42 reflects its aggregation and deposition in the brain parenchyma (sequestration hypothesis), whereas the increase in CSF Tau reflects its extracellular release after neuronal degeneration and neurofibrillary tangle formation. Thus, these markers are considered to be directly linked to the molecular pathogenesis of AD (4). However, the correlation between CSF markers, brain lesions, and neurodegeneration are indirect and based on CSF measurements during life versus neuropathological assessment at autopsy (4, 5).

Transgenic mice that overexpress human amyloid precursor protein (APP) are widely used models of AD that could be useful for developing CSF markers for AD. However, because these models do not develop neurofibrillary tangles and exhibit only region-specific and overall modest neuron loss, they have mainly been used as models of β-amyloidosis (7) and rarely to assess CSF markers (8, 9). To address this, we measured human Aβ and murine Tau concentrations in the CSF of two widely used APP transgenic mouse models: one expressing mutant human APP (APP23) and one expressing mutant human APP and presenilin-1 (APPPS1). Both models exhibit the age-related deposition of Aβ in the brain but with different onset and progression trajectories (10, 11). In both models, we show a steady decrease in CSF Aβ42 concentration that is inversely related to brain Aβ load. In the same mice, endogenous murine t-Tau showed a marked increase in CSF at the stage when Aβ pathology becomes prominent.

RESULTS

CSF Aβ concentrations decrease as Aβ pathology progresses in APPPS1 and APP23 mice

Measurement of human Aβ42 concentrations in the CSF of APPPS1 mice revealed a marked age-related decline (Fig. 1A). At 6 months of age, CSF Aβ42 had dropped 50%, and a more than 80% decline was noted at 18 months of age. Aβ40 also declined, but less prominently, decreasing by 45% at 18 months of age (Fig. 1B). This unequal decrease in different Aβ isoforms resulted in a decrease in the ratio of Aβ42 to Aβ40 with aging (Fig. 1C).

Fig. 1 APPPS1 mice exhibit decreased CSF Aβ and increased CSF t-Tau.

Male and female APPPS1 mice and nontransgenic littermates (1.5 to 18 months of age) were used (10 to 14 mice per group; no gender effect was found in any parameter, and thus, data were combined). (A and B) Measurement of human Aβ42 and Aβ40 in CSF revealed a significant trend with aging: F1,56 = 86.4 (P < 0.001) and 4.47 (P < 0.05). Post hoc group comparisons were always done between the youngest group of mice and all other groups. Nontransgenic control mice were not analyzed because they do not express human Aβ. (C) Human Aβ42/Aβ40 ratio in CSF; age trend F1,56 = 284.1, P < 0.001. (D and E) Human Aβ42 and Aβ40 in mouse brain homogenates; age trend F1,56 = 511.6 (P < 0.001) and 475.1 (P < 0.001). (F) Ratio of human Aβ42/Aβ40 in mouse brain; age trend F1,56 = 16.0, P < 0.001. (G) Endogenous murine t-Tau in CSF. APPPS1 mice were significantly different from nontransgenic (Non-tg) control mice [analysis of covariance (ANCOVA) Age trend × Genotype F1,116 = 40.6, P < 0.001]. (H) ROC curves comparing CSF t-Tau in APPPS1 and control mice at 3, 6, 12, and 18 months. AUC, area under the curve. (I) t-Tau/Aβ42 ratio in CSF; age trend F1,56 = 188.8, P < 0.001. All data are represented as group means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001. For Aβ measurements, the age trend test used was derived from the analysis of variance (ANOVA), whereas for t-Tau, an Age trend × Genotype interaction (ANCOVA) was calculated. Differences between the youngest APP transgenic mouse group and all other age groups were analyzed using Bonferroni’s post hoc test for multiple comparisons.

Concomitant with the decrease in CSF Aβ, we found an age-related increase in human Aβ42 (Fig. 1D) and human Aβ40 (Fig. 1E) in the brains of the same mice. This was in line with the robust deposition of Aβ in the brain parenchyma when assessed by both immunocytochemistry and Congo red staining (fig. S1). The Aβ42/Aβ40 ratio in brain increased initially at the time of Aβ deposition, but then remained constant with a minor decline (Fig. 1F).

In the second mouse model (APP23), human Aβ42 in CSF (Fig. 2A) remained unchanged until the age of 16 months and then declined. At 30 months of age, there was a 60% decrease in Aβ42 in CSF. Similar to the APPPS1 model, the decrease in Aβ40 was less prominent (Fig. 2B), leading to a decline in the Aβ42/Aβ40 ratio in CSF after 16 months of age (Fig. 2C). Analysis of Aβ load in the brains of the APP23 mice (Fig. 2, D and E) revealed that Aβ deposition in the brain precedes the drop in CSF Aβ. This is in line with the start of amyloid deposition at 6 to 8 months of age in these mice (10, 12). No consistent change in the Aβ42/Aβ40 ratio was found in the brains of the APP23 mice (Fig. 2F).

Fig. 2 APP23 mice exhibit decreased CSF Aβ and increased CSF t-Tau.

For CSF and brain Aβ measurements, male APP23 mice (3 to 30 months of age; n = 8 per group) were used. For CSF t-Tau, a new cohort of male APP23 mice and nontransgenic control mice were used (6 to 26 months of age; n = 9 to 12 per group for APP23 and 6 to 7 per group for nontransgenic controls). (A and B) Measurement of human Aβ42 and Aβ40 in CSF revealed a significant age trend F1,35 = 139.3 (P < 0.001) and 19.9 (P < 0.001). Post hoc group comparisons were always done between the youngest group and all other age groups. (C) Ratio of human Aβ42/Aβ40 in CSF; age trend F1,35 = 244.5, P < 0.001. (D and E) Human Aβ42 and Aβ40 in mouse brain homogenates; age trend F1,31 = 1411 (P < 0.001) and 3445 (P < 0.001). Note that the values of four mice are missing because of experimental processing errors. (F) Ratio of human Aβ42/Aβ40 in brain; age trend F1,30 = 2.43, P > 0.05. (G) Endogenous murine t-Tau in CSF of APP23 mice. APP23 mice were significantly different from nontransgenic control mice (ANCOVA Age trend × Genotype F1,47 = 16.9, P < 0.001). (H) ROC curve comparing CSF t-Tau in APP23 and control mice in the 24- to 26-month-old group. (I) The CSF t-Tau/Aβ42 ratio was analyzed in a subset of APP23 mice; n = 4 to 5 per group; age trend F1,11 = 97.4, P < 0.001. All data are represented as group means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

CSF Tau concentrations increase as Aβ pathology progresses in APP transgenic mice

The same CSF samples used to determine Aβ concentrations were analyzed for endogenous murine t-Tau. Results for the APPPS1 mice revealed a marked increase in CSF t-Tau starting at 6 months of age, with a delay after the onset of Aβ deposition. A more than fivefold increase was reached at 18 months of age, corresponding to a more than threefold increase compared to age-matched nontransgenic control mice (Fig. 1G). Receiver operating characteristic (ROC) curve analysis revealed that CSF t-Tau was able to discriminate between transgenic and control mice at 6 months of age and older (Fig. 1H). When the ratio of CSF t-Tau to Aβ42 was plotted—a potential predictor of AD pathology and conversion to dementia in patients with MCI (5)—a 45-fold increase was found in APPPS1 mice between 1.5 and 18 months (Fig. 1I). Correlation analyses between CSF t-Tau and both Aβ load [electrochemiluminescence-linked immunoassay (ECL-immunoassay)] and immunohistochemistry] and congophilic amyloid showed positive correlations among and within age groups (fig. S2).

In brain homogenates, endogenous murine t-Tau did not significantly change with advancing age in either APPPS1 mice or nontransgenic littermates (fig. S3). No fibrillar Tau inclusions were found at any age in APPPS1 mice, although many phosphorylated Tau–positive neuritic structures were observed in the vicinity of Aβ deposits (fig. S3), an observation consistent with previous reports (10, 11).

For the analysis of endogenous murine t-Tau in CSF of APP23 mice, a new cohort of mice was used (Fig. 2, G to I) because the CSF remaining did not allow additional measurements. Similar to the APPPS1 mice, an age-related increase in murine t-Tau was found. At 24 to 26 months of age, CSF t-Tau in APP23 mice was almost threefold greater than that in age-matched, nontransgenic control mice (Fig. 2, G and H). At this age, the ratio of CSF t-Tau to Aβ42 showed a 10-fold increase compared to young APP23 mice without Aβ deposits (Fig. 2I).

CSF α-synuclein remains unaltered

To exclude the possibility of an overall increase in neuronal proteins in the CSF of APP transgenic mice with Aβ deposits, we measured endogenous murine α-synuclein. Similar to Tau, α-synuclein is a highly abundant neuronal protein in brain that is secreted from cells under normal and pathological conditions and has been found in CSF (13, 14). Results revealed that CSF α-synuclein concentrations remained stable at all ages and did not differ significantly from those in nontransgenic mice (fig. S4).

DISCUSSION

Here, we have adopted methods to collect CSF from mice (15) following the standards set by the Alzheimer’s Association external quality control program for human CSF biomarkers (16). We then analyzed two commonly used APP transgenic mouse models, APPPS1 and APP23, that show different onset and progression times for Aβ pathology in the brain (10, 11).

Our results revealed a decrease in Aβ42 concentrations and, to a lesser extent, Aβ40 concentrations, which were both inversely related to the Aβ load in the brain, in agreement with previous CSF Aβ measurements in mice (8). CSF Aβ42 concentrations started to decline shortly after the onset of Aβ deposition in the mouse brain and decreased by as much as 50 to 80%. The decrease exceeded the 30 to 60% decrease in CSF Aβ42 reported in symptomatic AD patients compared to nondemented controls (1, 17). CSF Aβ40 also decreased with increasing Aβ deposition in mouse brain, but the decline was less marked than that in Aβ42.

In parallel with or shortly after the Aβ42 decrease in mouse CSF, endogenous murine t-Tau in the CSF of both mouse models increased up to threefold compared to age-matched control mice. These changes mirror the increases in CSF Tau that have been observed in patients with sporadic and familial AD compared to nondemented controls (1, 17). In mice, however, the increase in CSF t-Tau appears not to be related to neurofibrillary tangle formation and neuron loss because APPPS1 and APP23 mice do not develop fibrillar Tau lesions and exhibit only region-specific and modest overall neuron loss (18, 19). APP transgenic mice exhibit severe hyperphosphorylated Tau–positive neuritic and synaptic dystrophy in the vicinity of congophilic amyloid plaques (10, 11), but it is unclear how this pathology may contribute to the release of intracellular Tau. Nevertheless, the finding that the correlation between congophilic amyloid and CSF t-Tau was somewhat stronger than that between total Aβ deposits and CSF t-Tau (fig. S2) suggests that neuritic dystrophy may have a role in the increase in CSF t-Tau. Our observation that the CSF concentration of α-synuclein, which is also found in dystrophic boutons in the vicinity of amyloid plaques (20), did not change, indicates that the observed increases in CSF Tau were specific. However, further studies in these mice with γ-secretase or β-secretase inhibitors are required to better understand the mechanistic relationship between Aβ aggregation and the Tau increases in CSF.

Tau is physiologically secreted by cultured cells and in vivo (21, 22). Although the release mechanism is not understood, it is tempting to speculate that it is promoted by Aβ aggregates. Such a potential mechanism is not contradicted by the lack of t-Tau changes in brain homogenates because tissue t-Tau concentrations are about 105 times higher than t-Tau concentrations in the CSF (22). Consistent with such a hypothetical Aβ-mediated cellular Tau release, human tauopathies (in the absence of extracellular amyloid) do not show elevated CSF Tau concentrations to the extent observed in AD patients (14).

Although more time points need to be examined to determine precisely the temporal sequence of the changes in t-Tau and Aβ in the CSF of APP transgenic mice, our findings suggest that they may follow the same dynamics as predicted in humans (Fig. 3). In humans, CSF Aβ42 declines as Aβ deposits in the brain increase, and this decline is followed by a steady increase in t-Tau in CSF (1, 23). The absence of neurofibrillary tangles and global neuronal loss in the mouse models challenges the long-standing assumption that CSF t-Tau is a marker of neuronal loss and tangle pathology in AD. This is further reinforced by the recent studies in human patients, showing that in both familial and sporadic AD cases, CSF t-Tau increases 10 to 15 years before any clinical phenotype is noticeable, during which period significant neuronal loss would not be expected (1, 24). Our study, although requiring confirmation in other mouse models, does suggest that APP transgenic mice may be useful for predicting changes in protein markers in clinical treatment trials, particularly for preclinical AD (25). Our findings also suggest that APP transgenic mouse models may be useful in the search for new disease markers for AD.

Fig. 3 Protein profiles in mouse CSF with aging.

(A and B) Relative changes in CSF Aβ42 (red), CSF t-Tau (blue), and Aβ42 + Aβ40 concentrations in the brains of APPPS1 (A) and APP23 (B) mice with age as measured by ECL-immunoassay (green). (A) Aβ plaque load quantified by stereological analysis in histological sections (orange) in APPPS1 mice. The curves are based on the values shown in Figs. 1 and 2, respectively.

MATERIALS AND METHODS

APPPS1 mice

Male and female APPPS1 mice of ages 1.5 to 18 months (11) and gender- and age-matched nontransgenic control mice were bred at the Hertie Institute for Clinical Brain Research. APPPS1 mice have been initially generated and are maintained on a C57BL/6 background and coexpress K670M/N671L-mutated APP and L166P-mutated presenilin 1 (PS1) under the control of a neuron-specific Thy1 promoter element. The mice developed first Aβ plaques after 6 weeks of age, and no effect of gender was found (11). All mice were kept under specific pathogen–free conditions. The experimental procedures were undertaken in accordance with the veterinary office regulations of Baden-Württemberg (Germany) and approved by the local Animal Care and Use Committees.

APP23 mice

Male APP23 mice of ages 3 to 30 months (10) were bred at the Novartis Mouse facility (Basel, Switzerland) and used for Aβ assessment in brain and CSF. Additional 6- to 26-month-old male APP23 and age-matched nontransgenic control mice were bred at the Hertie Institute for Clinical Brain Research and were used for combined Aβ and t-Tau CSF measurements. APP23 mice express the K670M/N671L mutated APP under the control of the neuron-specific Thy1 promoter element. The mice were generated on a B6D2 background, but have since been bred with C57BL/6J mice for more than 20 generations. APP23 mice have been reported to develop first plaques at 6 to 8 months of age (10, 12), and plaque development is faster in females than in males. The experimental procedures were in accordance with the veterinary office regulations of Basel (Switzerland) and Baden-Württemberg (Germany) and were approved by the local Animal Care and Use Committees.

APP-, Tau-, and α-synuclein–deficient mice

The specificity of the Aβ, Tau, and α-synuclein immunoassay and immunohistochemistry was validated with CSF and brain tissue from APP-deficient mice (26), Tau-deficient mice (27) (gift of H. Dawson and E. Mandelkow, Chicago and Bonn, respectively), and α-synuclein–deficient mice (28) (provided by O. Riess, Tübingen).

CSF and tissue harvesting

We have standardized the methods for in vivo CSF collection in mice (15) following the standards set by the Alzheimer’s Association external quality control program for human CSF biomarkers (16). CSF collection was always performed between 10 a.m. and 2 p.m. Mice were injected with a mixture of ketamine (100 mg/kg) and xylazine (10 mg/kg) and placed in a 37°C warm chamber (Schülke & Mayr) until they were deeply anesthetized. CSF was then immediately collected from the cisterna magna with the help of magnification glasses. To this end, the overlying skin was incised to expose the skull and the posterior neck muscles. The latter were cut off layer by layer until the cisterna magna was visible through the translucent dura mater. Microbleedings were stopped with the help of a small vessel cauterizer (Fine Science Tools). After cleaning any blood residue from the surface with a cotton swab [adapted from DeMattos et al. (15)], the dura was perforated with a 25-gauge needle (B. Braun) and CSF was collected with a 20-μl gel loader tip (Eppendorf; shortened by about 2 cm at the tip) in polypropylene tubes and immediately placed on ice. For each mouse, the CSF was collected immediately after puncturing the meninges and again after a 2-min interval to allow the cisterna magna to refill with CSF. For uniformity, no CSF collection was performed after a longer time interval. CSF samples were then centrifuged at 13,000g for 30 s, assessed macroscopically for blood contamination, aliquoted (5 μl), and stored at −80°C until use. Typically, a total of 15 to 20 μl of CSF was collected. Blood-contaminated samples were discarded. Thereafter, mice were perfused with ice-cold sterile phosphate-buffered saline (PBS). The brain was removed, and one hemibrain (left) was snap-frozen in dry ice and stored at −80°C until use. The other hemibrain (right) was fixed in 4% paraformaldehyde with 0.1 M PBS (pH 7.6) for 48 hours at 4°C, immersed in 30% sucrose for an additional 24 hours at 4°C, snap-frozen in 2-methylbutane, and stored at −80°C.

Biochemical analysis of brain tissue

Hemibrains from APPPS1 and wild-type mice were homogenized at 10% (w/v) in homogenization buffer [50 mM tris (pH 8.0), 150 mM NaCl, 5 mM EDTA, and Complete protease inhibitor cocktail from Roche Molecular Biochemicals] at 4°C two times for 10 s at 5500 rpm in 7-ml lysing tubes with 2.8-mm ceramic beads using the Precellys 24-Dual homogenizer and Cryolys cooling device (Bertin). The homogenized brain tissue was aliquoted and stored at −80°C until use. For Aβ measurements, the homogenates were sequentially extracted as follows: Aliquots were thawed on ice and spun at 25,000g at 4°C for 1 hour. The supernatant was collected as the “tris-buffered saline (TBS)–soluble fraction,” and the pellet was resuspended in 70% formic acid (FA) (Sigma) to the original volume, sonicated for 35 s at 4°C, and spun at 25,000g at 4°C for 1 hour. The supernatant was collected as the “FA-soluble fraction” and equilibrated (1:20) in neutralization buffer (1 M tris base, 0.5 M Na2HPO4, 0.05% NaN3).

The brain tissue of the APP23 mice was similarly prepared with the following deviations: First, forebrains (hemibrains without the cerebellum) were used, and second, homogenization was done at 10% (w/v) in TBS [30 mM tris-HCl (pH 7.6), 137 mM NaCl, Complete protease inhibitor cocktail, Roche] by vigorous shaking with metal beads in a Retsch mill followed by brief sonication.

Electrochemiluminescence-linked immunoassay for Aβ

Aβ concentrations in CSF and brain extracts of APPPS1 mice were determined with an ECL-immunoassay using the MSD 96-Well MULTI-SPOT Human (6E10) Aβ Triplex Assay (Meso Scale Discovery). Aβ detection was conducted according to the manufacturer’s instructions. In brief, 96-well plates prespotted with capture antibodies against Aβx–38, Aβx–40, and Aβx–42 were blocked for 1 hour with 1% blocking solution [1% bovine serum albumin (BSA) in tris buffer] and washed three times with 1× tris buffer. In a second step, CSF samples (5 μl) were diluted 1:14 and FA brain extracts were diluted up to 1:1000 (depending on Aβ load, to stay within the linear range of the assay) in blocking solution and coincubated with the SULFO-TAG 6E10 detection antibody solution on the plate for 2 hours. After washing, MSD Read Buffer T was added and the plate was read immediately on a Sector Imager 6000. Data analysis used MSD DISCOVERY WORKBENCH software 2.0. Every sample was tested in duplicate, and those with a coefficient of variance (CV) more than 20% were excluded from the analysis. Internal reference samples were used as a control in every plate, and the results were adjusted for interplate variability. APP23 CSF samples and brain extracts were processed in a similar way, except that Aβ40 and Aβ42 were determined in separate plates with MSD 96-Well Human (6E10) Aβ40 and Aβ42 Ultra-Sensitive Kits (Meso Scale Discovery). Aβ concentrations were read from the standard curves using a point-to-point fit with the software SoftMax Pro 4.0 (Molecular Devices Corp.).

Electrochemiluminescence-linked immunoassay for t-Tau

Murine t-Tau in CSF and brain extracts of APPPS1 and wild-type mice was determined with an ECL-immunoassay using the MSD 96-Well Mouse Total Tau assay (Meso Scale Discovery) according to the manufacturer’s instructions. This assay uses one reference and two capture antibodies, which recognize all isoforms of human, murine, and bovine Tau independently of their phosphorylation status by targeting the protein between exon 3 and 10. The assay was validated with recombinant murine Tau (gift of M. Mandelkow, Bonn, Germany). In brief, CSF samples (5 μl) were diluted 1:14 in 10% blocking solution (10% BSA in tris buffer) and incubated for 1 hour on the plate. Soluble fractions of the brain extracts were diluted 1:1000 in 3% blocking solution (3% BSA in tris buffer) and also incubated for 1 hour on the plate. After washing the plate, it was incubated for another hour with the SULFO-TAG t-Tau detection antibody solution and analyzed as described above.

ELISA assay for α-synuclein

Murine α-synuclein was measured in the CSF with the AnaSpec ultra-sensitive ELISA assay according to the manufacturer’s protocol. Briefly, murine CSF (10 μl) was diluted 1:25 in dilution buffer containing protease inhibitors (Roche). The samples were incubated together with the detection antibody (horseradish peroxidase–conjugated rabbit polyclonal anti–α-synuclein) overnight at 4°C in a 96-well plate precoated with the capture antibody (monoclonal anti–α-synuclein). Absorbance was measured at 450 nm on an ELISA plate reader (Mithras, Berthold Technologies). Standard curves were prepared with recombinant murine α-synuclein (Promega) in the same dilution buffer as the CSF samples. Murine α-synuclein concentrations were obtained using a calculated linear equation for the concentration-absorbance relation. Every sample was assessed in duplicate, and samples with CVs more than 20% were disregarded in the analysis. An internal control was used in every plate, and results were adjusted for interplate variability.

Histology and immunohistochemistry

After freezing, fixed brains were cut into serial 25-μm-thick sagittal sections with a freezing-sliding microtome. The sections were collected in 0.1 M TBS (pH 7.4) and stained immunohistochemically according to previously published protocols with anti-Aβ polyclonal antibody CN3 (29). Sections were also stained with Congo red according to standard protocols and viewed under cross-polarized light. The monoclonal antibody AT8 (Thermo Scientific) directed against phospho-Tau position 191/194 (murine) or 202/205 (human) was used according to previous published protocols (10, 11).

Quantification of total Aβ load and congophilic amyloid load

Aβ load was quantified on an Aβ-immunostained set of every 12th systematically sampled, serial, sagittal section throughout the entire neocortex. Congophilic amyloid load was performed on a second Congo red–stained set of every 12th systematically sampled sections. Researchers who were blinded to the age groups performed the analysis. Stereological analysis was performed with a microscope equipped with a motorized x-y-z stage coupled to a video microscopy system and the Stereo Investigator software (MicroBrightField Inc.) as previously described (30). The Aβ load and congophilic amyloid load were determined by calculating the areal fraction occupied by CN3-positive immunostaining or Congo red staining with a 20× objective (0.45 numerical aperture) and a Zeiss Axioskop 2 microscope. Note that the areal fraction was not determined at a single focal plane and was calculated from the 25-μm-thick sections and thus overestimates the true Aβ or amyloid load.

Statistical analysis

The distribution of quantitative data was assessed with the Shapiro-Wilk test. Nonnormally distributed variables were logarithmic-transformed. To examine whether CSF and brain Aβ levels change with aging in APP transgenic mice, a trend test derived from an ANOVA analysis was calculated. For the congophilic amyloid load, normal distribution could not be confirmed, because all measurements were equal to zero at 1.5 months. Thus, nonparametric testing (Jonckheere-Terpstra trend test, Mann-Whitney test for pairwise comparisons with Bonferroni correction) was applied. To examine whether CSF t-Tau and α-synuclein measurements were different between transgenic and nontransgenic control mice, an Age trend × Genotype interaction (ANCOVA) was calculated. Differences between the youngest APP transgenic mouse group and all other age groups were analyzed with Bonferroni’s post hoc test for multiple comparisons. If the measured individual values were below the assay detection limit, a fixed value (detection limit of the plate where the sample was measured/√2) was imputed. ROC curves were drawn by plotting the true-positive fraction (sensitivity) against the false-positive fraction (100% − specificity) for varying cutoff values. The area under the curve was calculated. In all cases, statistical significance was set at P < 0.05. GraphPad Prism version 5 was used to generate the graphics, and SPSS version 19 was used for statistical analysis.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/5/194/194re2/DC1

Fig. S1. Aβ deposition in APPPS1 mice.

Fig. S2. Correlation analyses between brain Aβ and CSF t-Tau.

Fig. S3. Tau concentrations and Tau pathology in APPPS1 mouse brain.

Fig. S4. CSF murine α-synuclein is unaltered in APPPS1 and APP23 mice.

REFERENCES AND NOTES

  1. Acknowledgments: We thank U. Obermüller, A. Bosch, C. Krüger, C. Schäfer, J. Odenthal, W. Maetzler, H. Wolburg (Tübingen), and T. Golde (Gainesville, FL) for experimental help. The comments on this manuscript from Y. Eisele (Tübingen), L. Walker (Atlanta), and M. Goedert (Cambridge, UK) are greatly appreciated. We thank L. Binder (Chicago) for various Tau antibodies; R. Umek (Meso Scale Discovery, Rockville), E. Mandelkow (Bonn), and H. Dawson (Durham) for recombinant murine Tau and Tau-deficient control mice; and O. Riess (Tübingen) for α-synuclein–deficient control mice. Funding: This work was supported by grants from the Competence Network on Degenerative Dementias (BMBF-01GI0705) and Fundação para a Ciência e Tecnologia (SFRH/BD/66216/2009). Author contributions: L.F.M., S.A.K., M.S., and M.J. designed the study; L.F.M. and S.A.K. developed and standardized the CSF collection protocol; L.F.M. and J.R. collected the CSF and prepared the brain tissue; S.A.K., J.R., and M.S. established the protocol for the assessment of Aβ species and Tau in mouse tissue and CSF by enhanced chemiluminescence (ECL) assays; M.H. and J.R. performed the ECL measurements; L.F.M. validated and performed the α-synuclein ELISA; L.F.M. and P.M. did the statistical analysis; L.F.M., S.A.K., P.M., M.S., and M.J. interpreted the data; M.J., M.S., L.F.M., and S.A.K., with the help of all the authors, wrote the manuscript. U.O., C.S., A.B., and C.K. (listed in the Acknowledgments) were instrumental in the histology and stereological analysis. Competing interests: J.R. and M.S. are employees of Novartis Pharma and hold Novartis stock. The other authors declare that they have no competing interests.
View Abstract

Navigate This Article