Reduction of cholesterol synthesis in the mouse brain does not affect amyloid formation in Alzheimer's disease, but does extend lifespan

Abstract

Alterations in cellular cholesterol synthesis or content in cultured neurons affect the cleavage of amyloid precursor protein to amyloidogenic Aβ₄₀ and Aβ₄₂ peptides characteristic of Alzheimer's disease. To determine whether a decrease in cholesterol synthesis affects amyloid precursor protein processing in vivo, we crossed cholesterol 24-hydroxylase knockout mice, which exhibit a 50% reduction in brain sterol synthesis, with transgenic mice [B6.Cg-Tg(APPswe, PSEN1E9)85Dbo/J] that develop Alzheimer's disease-like pathology. Amyloid precursor protein expression and amyloid plaque deposition in the cortex and hippocampus of male and female Alzheimer's disease mice between the ages of 3 to 15 months were similar in the presence and absence of cholesterol 24-hydroxylase. A modest but statistically significant decline in insoluble Aβ₄₂ peptide levels was detected in the hippocampus of 12-month-old knockout/Alzheimer's disease males. The levels of insoluble Aβ₄₀ and Aβ₄₂ peptides in 15-month-old knockout/Alzheimer's disease females were also reduced slightly. Although amyloid plaque accumulation did not affect brain sterol or fatty acid synthesis rates in 24-hydroxylase WT or knockout mice, loss of one or both cholesterol 24-hydroxylase alleles increased longevity in Alzheimer's disease mice. These studies suggest that reducing de novo cholesterol synthesis in the brain will not substantially alter the course of Alzheimer's disease, but may confer a survival advantage.

Alzheimer's disease (AD) is a progressive neurodegenerative disease of the central nervous system that affects a growing percentage of the elderly population. AD is characterized by gradual loss of cognitive function leading to death, and at the biochemical level, by the deposition of aberrantly folded proteins in the cortex and hippocampus of the brain (1). These protein deposits, termed amyloid plaques and neurofibrillary tangles, contain proteolytic fragments of the amyloid precursor protein (APP) and hyperphosphorylated forms of the tau microfilament protein, respectively (2, 3).

Mutations in genes encoding proteins that produce amyloid plaques such as APP, presenilin 1, and presenilin 2 cause early onset forms of AD (1), as do altered versions of several proteins with less well-defined roles in the pathogenesis of the disorder, such as apolipoprotein E (ApoE). There are 3 major isoforms of ApoE, referred to as ApoE2, E3, and E4, and individuals who inherit one of more copies of the gene specifying ApoE4 are predisposed to earlier ages of AD onset (4). ApoE plays an important role in cholesterol metabolism in peripheral tissues (5), and this function has led to the notion that alterations in brain cholesterol metabolism may affect the inception or progression of AD. In support of this hypothesis, some but not all epidemiological studies show that subjects who take statins, a class of drugs that reduce plasma cholesterol levels by inhibiting de novo cholesterol synthesis, have altered susceptibilities to AD (6).

Cholesterol metabolism in the brain differs from that in peripheral tissues. All cholesterol in the brain is obtained thorough de novo synthesis and not from the metabolism of cholesterol-rich lipoprotein particles (7). In contrast to peripheral tissues that can turn over cholesterol via exchange with circulating lipoprotein particles, the brain does not have access to these particles and instead relies on an alternate catabolism pathway in which the neuronal enzyme cholesterol 24-hydroxylase converts cholesterol to 24S-hydroxycholesterol. This oxysterol spontaneously crosses cellular membranes, gains access to the blood, and is cleared by the liver (8). The 24-hydroxylase pathway is active in the hippocampal and cortical neurons that are affected by AD (9, 10). Mice lacking 24-hydroxylase exhibit a 50% reduction in cholesterol turnover and compensate for this loss by decreasing de novo cholesterol synthesis by an equivalent amount (9, 11). This response maintains steady-state levels of cholesterol in the brain, but has the unintended consequence of reducing the synthesis of the polyisoprenoid geranylgeraniol, which is required for normal learning and hippocampal synaptic plasticity in mice (12, 13).

In the current study, we determine the effects of altered brain cholesterol metabolism brought about by loss of 24-hydroxylase on the development and progression of AD. Mice deficient in 24-hydroxylase were crossed to a readily available line of transgenic mice [B6.Cg-Tg(APPswe, PSEN1E9)85Dbo/J (line 85)] that accumulate amyloid plaques in the cortex and hippocampus at an early age and manifest cognitive defects (14, 15). We find that reductions in neuronal cholesterol synthesis and turnover do not markedly affect rates or extent of amyloid plaque formation or levels of amyloidogenic peptides, but these metabolic perturbations significantly prolong longevity in this strain of mice with AD-like pathology.

Results

Analysis of Proteins Related to Cholesterol Metabolism and AD.

A series of immunoblotting experiments using whole brain homogenates from aged animals were performed to examine protein expression in WT, 24-hydroxylase knockout (KO), and WT and KO mice expressing the AD transgenes (WT/AD and KO/AD, respectively). The amounts of 24-hydroxylase did not differ between WT and WT/AD mice (Fig. 1, lanes 1 and 3), and as expected, KO mice did not express this enzyme (lanes 2 and 4). Previous examination of neuropathology in this mouse model of AD has revealed limited neuritic dystrophy without widespread loss of neurons (16), which may explain the normal levels of 24-hydroxylase in WT/AD animals. Blotting with an antiserum that recognized the humanized APP revealed expression in AD mice only and showed that the level of this protein was not altered by the presence or absence of 24-hydroxylase (see lanes 3 and 4). No evidence of reactive astrogliosis was observed in any of the mouse lines as judged from the amounts of glial fibrillary acidic protein, nor were levels of tau protein different (see lanes 1–4). The signals from a loading control (α-tubulin) indicated that equal amounts of protein were present in each lane of the blot.

Fig. 1.

Protein expression in brains of 24-hydroxylase KO and AD mice. Brain homogenates were prepared from 11- to 13-month-old female mice of the indicated genotype and aliquots (35 μg of protein) were subjected to immunoblot analysis for the indicated proteins using antibodies as described in Materials and Methods and the SI. Film was exposed for 5 to 30 s. The positions to which standards of known molecular mass (× 10³) migrated to on the polyacrylamide gel are indicated on the left of each panel.

In Vivo Lipid Synthesis.

To determine the effect of AD gene expression on brain cholesterol metabolism, rates of de novo cholesterol synthesis were determined in WT and KO female animals aged 11 to 13 months using a quantitative tritiated water incorporation assay. Females were chosen for the experiment because levels of 24-hydroxylase are not sexually dimorphic (9), and female mice accumulate amyloid plaque to a greater degree than male AD mice of the same age (see below). The data of Fig. 2A show that expression of the AD genes did not significantly alter levels of cholesterol synthesis in 24-hydroxylase WT or KO mice. As expected from previous studies (9, 11), cholesterol synthetic rates were reduced in animals lacking 24-hydroxylase, and this reduction was specific to cholesterol as fatty acid synthesis was similar among all genotypes examined (Fig. 2B). Additionally, rates of cholesterol synthesis were indistinguishable in the liver among these genotypes (data not shown), emphasizing that the reduction in synthesis observed in KO and KO/AD animals was limited to the brain.

Fig. 2.

In vivo synthesis of sterols (A) and fatty acids (B) in 24-hydroxylase KO and AD mice. Mice of the indicated genotypes (11- to 14-month-old females, 10 per group) were maintained on a normal chow diet containing 0.02% cholesterol. On the day of the experiment, each animal was injected peritoneally with 50 mCi of tritium-labeled water; 1 h later tissues were removed and processed for measurement of radio-labeled sterols and fatty acids as described in the SI. Student's t test was performed to assess differences between experimental groups. In (A), KO values were significantly (P = 0.022) lower than WT values; KO/AD values were not significantly (P = 0.06) lower than WT/AD values. Error bars indicate SEM values.

Amyloid Plaque Accumulation in AD Mice.

To qualitatively assess the effects of 24-hydroxylase genotype on amyloid plaque development in AD mice, brain tissues from 9-month-old male WT/AD mice were stained with thioflavine S to visualize plaques and with DAPI to identify cell nuclei. WT/AD mice exhibited abundant plaques in the cortex and hippocampus (Fig. 3 A and D). Similarly, KO/AD mice of the same age demonstrated advanced plaque accumulation (Fig. 3 B and E). In contrast, plaques were not observed in WT or KO mice lacking the AD transgenes (Fig. 3 C and F).

Fig. 3.

Amyloid staining in 9-month old male 24-hydroxylase KO and AD mice. Cortical (Ctx) and hippocampal (Hpc) sections from mice of the indicated 24-hydroxylase/AD genotypes were stained with thioflavine S and DAPI to visualize amyloid plaques (green) and cell nuclei (blue), respectively. No differences in plaque load were detected in AD mice with (A and D) or without (B and E) 24-hydroxylase, and no plaques were detected in the absence of the AD transgenes and 24-hydroxylase (C and F).

To determine if rates of plaque accumulation were altered by the presence or absence of 24-hydroxylase, brains from WT/AD and KO/AD male and female mice (≥ 6 animals per genotype and sex) were collected at 3-month intervals between the ages of 3 and 15 months. Coronal tissue sections were prepared from the cortex and hippocampus, stained with Congo Red to detect β-amyloid, and counterstained with hematoxylin to distinguish nuclei. Plaque load was quantified using stereology as described in the supporting information (SI). A pilot study was first conducted in which plaque load was quantified in 18 to 21 tissue sections from each of 6 animals of different genotypes. The results of this exploratory experiment indicated that a minimum sampling frequency of 3 evenly spaced sections per brain region was necessary to obtain an accurate measurement of plaque burden. The pilot study also revealed that plaque accumulation occurred randomly throughout the cortex, but was more abundant in the polymorphic layer of the dentate gyrus than in other regions of the hippocampus (data not shown).

Plaque load was measured subsequently in WT/AD and KO/AD male and female mice between the ages of 3 and 15 months using methods established in the pilot study. Percent-plaque area was plotted as a function of age (Fig. 4). All genotypes exhibited a linear rate of plaque accumulation over time in both the cortex and hippocampus. Female mice exhibited plaque formation earlier and accumulated plaques to a greater extent than male mice of the same age (compare Fig. 4 B vs. A). No significant differences in cortical or hippocampal plaque accumulation were detected between male and female WT/AD and KO/AD mice at any age.

Fig. 4.

Amyloid deposition in 24-hydroxylase KO and AD mice. Coronal sections from the hippocampus and cortex of female (A) and male (B) mice of the indicated genotypes and ages (≥ 6 animals per group) were prepared and stained with Congo Red as described in Materials and Methods and the SI. Plaque area was determined by stereology and plotted as a function of the age of the animal. Error bars indicate SEM values.

Quantification of Aβ Peptides.

To examine APP processing in 24-hydroxylase WT/AD and KO/AD mice, soluble and insoluble pools of Aβ₄₀ and Aβ₄₂ peptides in hippocampal and cortical homogenates from males and females of different ages were quantified by ELISA. In the cortex, insoluble Aβ peptides were first detected in both sexes at 6 months of age (Fig. 5 A and B). This appearance was earlier than in the hippocampus, where insoluble Aβ peptides were not detected until 9 months of age (Fig. 5 C and D). Levels of insoluble Aβ₄₀ rose rapidly in the cortex and reached steady-state levels in females by 9 months, and in males by 12 months (see Fig. 5A). By 15 months, the amount of Aβ₄₀ in the insoluble fraction was similar in both sexes. A linear rate of accumulation of insoluble Aβ₄₂ in the cortex was also observed in males and females with increasing age (see Fig. 5B); however, unlike the hippocampus, there were no differences in rates of cortical Aβ₄₂ peptide deposition between sexes.

Fig. 5.

Insoluble Aβ peptide levels in 24-hydroxylase KO and AD mice. Cortical (A and B) and hippocampal (C and D) Aβ peptide levels in mice (≥5 animals per group) of the indicated genotype, sex, and age were quantified by ELISA as described in Materials and Methods and the SI. Tissues were homogenized in RIPA buffer containing protease inhibitors. Insoluble Aβ peptides were isolated by centrifugation and solubilized with 70% formic acid before measurement. Error bars indicate SEM values.

By 15 months of age, there was ≈3 to 4 times more Aβ₄₂ than Aβ₄₀ in the cortex. In contrast, the amount of insoluble Aβ₄₀ in the hippocampus was roughly equal to that of Aβ₄₂ at 15 months and there was a linear rate of deposition of these 2 peptides into this pool between 9 and 15 months of age. Females exhibited a trend toward higher levels of both Aβ₄₀ and Aβ₄₂ than males (see Fig. 5 C and D), and this trend reached statistical significance at 15 months of age in 3 of the 4 groups examined (Student's t test; WT/AD: Aβ₄₀ P = 0.0073, Aβ₄₂ P = 0.0006; KO/AD: Aβ₄₀ N.S.; KO/AD Aβ₄₂ P = 0.0184). Additionally, at 12 months of age, WT/AD males had modestly higher hippocampal levels of Aβ₄₂ than their KO/AD counterparts (Student's t test; P = 0.0469). This difference in males was not significant at 15 months of age; however, at this age WT/AD females demonstrated slightly higher hippocampal levels of Aβ₄₀ and Aβ₄₂ than their KO/AD counterparts (Student's t test; Aβ₄₀ P = 0.0239, Aβ₄₂ P = 0.0059).

Aβ₄₀ peptide levels in the detergent-soluble fractions of hippocampal and cortical brain homogenates were only slightly above the limit of detection at 15 months of age in all groups tested (data not shown). Similarly, Aβ₄₂ peptide was detected solely in the insoluble fraction, with peptide in the soluble fraction failing to exceed the level of detection at all ages examined.

Survival Study.

Multiple transgenic mouse models of AD, including the line used in the current study, are known exhibit premature death of unidentified causes (17). To examine the effects of 24-hydroxylase on mortality in AD mice, a survival study was performed. Animals of different genotypes were housed in same-sex groups of 2 to 4 animals per cage and longevity monitored. A minimum of 12 animals was included for each genotype and mice were monitored over a 550-day period. The numbers of female WT/AD mice in the study declined steadily over time, with only ≈50% of the animals surviving past the 550-day censor date (Fig. 6). Male WT/AD mice demonstrated a similar, although less severe, accelerated death rate. The absence of 24-hydroxylase prolonged life in AD mice, despite having little or no effect on plaque accumulation or Aβ peptide levels. Approximately 90% of KO/AD male and female animals survived past the censor date. Additionally, mice missing only 1 copy of the 24-hydroxylase gene (HET/AD) were protected from premature death to the same degree as homozygous KO/AD mice. All non-AD animals exhibited normal rates of survival (data not shown), consistent with the observation that 24-hydroxylase KO mice display no outward differences from WT in a variety of physiological read-outs, including life span (9, 11).

Fig. 6.

AD mice live longer in absence of 24-hydroxylase. Virgin male and female AD mice of the indicated 24-hydroxylase genotypes (HET = ±, KO = –/–; ≥12 animals per group) were maintained under standard vivarium conditions (single-sex housing, 2–4 animals per shoebox; 12-h light/dark cycle) and given food (7001 diet, Harlan Teklad) and water ad libitum. Survival was assessed over a 550-day period and all genotypes were confirmed at death or at the end of the experiment.

Discussion

The data in the current study indicate that decreasing de novo cholesterol synthesis by genetic means does not markedly influence the formation of amyloid plaque in the cortex and hippocampus of a mouse AD model. Conversely, the deposition of amyloid does not have a major effect on cholesterol biosynthetic rates in the central nervous system of these animals. The loss of 1 or more 24-hydroxylase genes, and presumably the accompanying decrease in cholesterol synthesis and turnover, does prolong the life of male and female AD mice.

Several genes with documented or suspected roles in peripheral tissue cholesterol metabolism, including ApoE (18), clusterin (19), and ABCA1 (20, 21), modulate amyloid plaque formation in the mouse brain; however, knockout of any one of these genes with the exception of clusterin, which has not yet been tested, does not affect cholesterol metabolism in the brain (22). This negative correlation provides support for the hypothesis that the roles of these proteins in affecting plaque deposition may not be related to the binding, transport, and turnover of cholesterol, but rather may involve effects on APP trafficking, APP processing, or Aβ peptide metabolism (23, 24).

In contrast to these findings, 24-hydroxylase KO mice exhibit an ≈50% decrease in de novo cholesterol synthesis and a corresponding 50% decrease in cholesterol excretion from the brain (9, 11), but these alterations do not affect rates of amyloid deposition, at least as judged by Congo Red staining (see Fig. 4). This outcome is similar to that observed in ABCG1 transgenic and KO mice in which cholesterol synthesis rates are apparently decreased and increased, respectively, but neither of these changes affect APP levels or processing to amyloidogenic Aβ peptides (25). Inasmuch as data from mice can be extrapolated to humans, these outcomes suggest that even if drugs such as statins were able to cross the blood–brain barrier and reduce cholesterol synthesis, they would not alter the deposition of amyloid in AD.

The reduced cholesterol flux through the brains of 24-hydroxylase KO mice also does not markedly affect the processing of APP to Aβ₄₀ and Aβ₄₂ peptides (see Fig. 5). These in vivo results imply that levels of secretase activity are similar in WT/AD and KO/AD mice, and they suggest that intracellular cholesterol levels and distribution are normal in the KO mice. An extensive body of literature indicates that acute changes in intracellular cholesterol levels can alter the processing of APP in vitro (26), in part by modifying membrane lipid composition, which in turn increases or decreases secretase activity (27–29). The difference between the in vivo results reported here and these in vitro results is possibly explained by the harsh pharmacological treatments used to change intracellular cholesterol levels, which can temporarily override the intricate regulatory mechanisms that normally maintain membrane cholesterol content within a narrow range (30). In contrast, these regulatory mechanisms are intact in 24-hydroxylase KO mice based on the observation that brain cholesterol levels in the mutant mice are not different from those of WT mice (9, 11). Similarly, in vitro studies suggest that high concentrations of Aβ peptides can directly modulate cholesterol synthesis in cultured neurons and other cell types (31), whereas here, no effect of extensive amyloid accumulation was observed on in vivo lipid synthesis (see Fig. 2). These disparate findings may reflect differences in rates or extent of Aβ peptide accumulation.

AD is a progressive, fatal neurodegenerative disorder in humans, and some mouse models such as the one used in the current studies recapitulate the premature death associated with the human disease. The effect of deleting the 24-hydroxylase gene on the lifespan of these AD mice seems clear cut; loss of 1 or more 24-hydroxylase alleles confers an essentially normal lifespan to both male and female animals (see Fig. 6). There are several caveats to consider in interpreting these findings (32). First, the number of mice in each genotype and sex analyzed was limited to ≥12 animals and the experiment was done only once. Second, the animals were of mixed genetic background (BL/6J;129SvEv;C3HeJ), and they were maintained in 2 different animal rooms, in 1 of which sentinel mice tested positive for exposure to mouse hepatitis virus, but no other common rodent pathogens. Third, the numbers of animals in each experimental group necessitated the use of nonlittermates, and fourth, although the genotyping procedures tested for the presence of both AD transgenes, the method did not assess transgene copy number. Any one of these variables could have generated noise that confounded interpretation of the survival study data; nevertheless, the observation that increased longevity was observed in 4 experimental groups (HET/AD males and females, and KO/AD males and females) suggests that loss of this gene does confer a true survival advantage. KO mice with 24-hydroxylase exhibit a codominant phenotype, with heterozygotes manifesting intermediate levels of decreased cholesterol synthesis and excretion (R.W.H., unpublished observations). This phenotypic effect in turn indicates that the increase in lifespan observed in 24-hydroxylase heterozygous and KO/AD mice may be linked to the alteration in cholesterol metabolism. Future experiments are needed to determine the mechanism of this protective effect.

Materials and Methods

Mice.

The 24-hydroxylase KO mice with a null mutation in the gene were of mixed strain (C57BL/6J;129S6/SvEv) (11). B6.Cg-Tg(APPswe, PSEN1E9)85Dbo/J (line 85) AD mice (14) were purchased from Jackson Laboratories (stock #005864). To produce the animals used in this study, 24-hydroxylase KO mice were mated initially to AD hemizygotes and subsequently, 24-hydroxylase heterozygous/AD hemizygous animals were crossed to 24-hydroxylase heterozygous animals to generate 24-hydroxylase WT, KO, and AD mice. All AD mice studied were hemizygotes. The genotypes of offspring were determined by PCR at weaning (between 3 and 5 weeks of age) and, for animals used in a given experiment, confirmed at the time of killing. The University of Texas Southwestern Institutional Animal Care and Research Advisory committee approved all experimental procedures. Further details concerning genotyping are described in the SI.

Amyloid Plaque Detection.

Thioflavine S and Congo Red staining were performed as described in the SI.

Amyloid Plaque Quantification.

Experimental and sampling parameters were developed based on the methods of Mouton (33) and Gundersen and Osterby (34) and are described in the SI. To compare results within and among experimental animals, data were converted to a fraction equal to the area represented by plaque divided by the total surface area of the brain region. Student's t test was used to assess differences between 2 experimental groups.

Aβ Peptide Quantification.

Brain tissues were harvested and homogenates prepared in ice-cold RIPA buffer (150 mM NaCl, 1.0% (vol/vol) Nonidet P-40, 0.5% (wt/vol) sodium deoxycholate, 0.1% (wt/vol) SDS, and 50 mM Tris-HCl, pH 8.0). Ice-cold 70% (vol/vol) formic acid was used to solubilize amyloid, and Aβ₄₀ and Aβ₄₂-peptide levels were determined by quantitative ELISA using a modification of published protocols (35). Anti-amyloid β (1–16) clone 6E10 (EMD Bioscience and Covance) was used as the capture antibody, Aβ₄₀ and Aβ₄₂ specific sera were from Calbiochem (PC149 or PC150, respectively), and antigen-antibody complexes were detected using a kit (1-Step ABTS; Pierce Biotechnology). A Synergy 4 microplate reader employing Gen 5 software (BioTek) was used to quantitate signals, and data were exported to Microsoft Excel for analysis. Additional experimental details regarding Aβ peptide quantification appear in the SI.

Cholesterol Synthesis.

Rates of cholesterol synthesis in living animals were determined by measuring the incorporation of tritiated water into cholesterol over time using methods described in (36) and modified as described in the SI.

Immunoblotting.

Pooled brain homogenates were prepared from female mice aged 11 to 13 months (n = 3 per genotype) and subjected to immunoblotting as described in the SI.

Survival Study.

Mice were committed to the experiment at the time of weaning (n ≥ 12 for all genotypes and sexes). Animals were housed in groups of 2 to 4 and were not used for further investigation, nor were they mated. Identification of individual living mice was by unique ear markings. Dead animals were removed from the colony and genotypes confirmed. Data were plotted and survival curves generated using Prism software.