Prolonged Diet Induced Obesity Has Minimal Effects Towards Cerebral Amyloid Angiopathy: Implications for Studying Obesity-Brain Interactions In Mice

Abstract

Cerebral amyloid angiopathy (CAA) occurs in nearly every individual with Alzheimer’s disease (AD) and Down’s syndrome, and is the second largest cause of intracerebral hemorrhage. Mouse models of CAA have demonstrated evidence for increased gliosis contributing to CAA pathology. Nearly two thirds of Americans are overweight or obese, with little known about the effects of obesity on the brain, although increasingly the vasculature appears to be a principle target of obesity effects on the brain. In the current study we describe for the first time whether diet induced obesity (DIO) modulates glial reactivity, amyloid levels, and inflammatory signaling in a mouse model of CAA. In these studies we determine that DIO does not significantly alter gross Aβ levels, astrocyte (GFAP) or microglial (IBA-1) gliosis in the CAA mice. However, within the hippocampal gyri a localized increase in reactive microglia were increased in the CA1 and stratum oriens relative to CAA mice on a control diet. DIO was observed to selectively increase IL-6 in CAA mice, with IL-1β and TNF-α not increased in CAA mice in response to DIO. Taken together, these data show that prolonged DIO does not significantly alter gross levels of Aβ or gliosis in CAA mice, but appears to elevate some localized microglial reactivity within the hippocampal gyri and selective markers of inflammatory signaling. These data are consistent with many of the studies in the existing literature using other models of Aβ pathology, which surprisingly show a mixed profile of DIO effects towards pathological processes in mouse models of neurodegenerative disease. The importance for considering the potential impact of ceiling effects in Aβ pathogenesis within the various mouse models, the lack of consistency for DIO based experimentation, and the current experimental limitations for DIO in mice to fully replicate metabolic dysfunction present in human obesity, are discussed.

1. Introduction

Despite the fact nearly two thirds of Americans are overweight or obese, and the fact that obesity is known to contribute to multiple morbidities and dysfunction of multiple tissues, relatively little is known in terms of the effect of obesity on the brain [1,2]. Epidemiological and cross sectional studies, while not uniform, suggest that presence of obesity increases the incidence of numerous neurodegenerative conditions including Alzheimer’s disease (AD), vascular dementia, and cerebral stroke [2–7]. The presence of prolonged obesity, starting in midlife, is particularly linked to the development of neurodegenerative disorders such as AD [8–13]. Despite such advances, very little is known in terms of the potential mechanisms by which obesity potentially promotes the development of neurodegenerative diseases, or which specific neuropathological features may potentially be exacerbated by the presence of obesity.

A large and robust number of publications involving findings from rodent studies have been insightful in understanding the molecular and pathophysiological basis for obesity, and the metabolic complications of obesity [14–23]. While the impact of obesity on multiple tissues of the body are well established in human and rodent studies (muscular, hepatic, cardiac, nephritic systems), relatively little is known in terms of how obesity impacts the central nervous system. In particular, there is a paucity of data on the effects of diet induced obesity (DIO) as modulators of specific neuropathological processes in established mouse models of neurodegenerative disease. Such data is essential in order to develop a mechanistic understanding as to if, and how, obesity modulates neurodegenerative processes involved in diseases such as AD.

Cerebral amyloid angiopathy (CAA) occurs in nearly all cases of Alzheimer’s disease (AD), Down’s syndrome, and is the second leading cause of intracerebral hemorrhage [24–28]. CAA is caused by the accumulation of amyloid within the cerebrovasculature, and is associated with numerous alterations relevant to AD and other neurodegenerative conditions including neuroinflammation, oxidative stress, and gliosis. No studies to date have examined the impact of DIO towards CAA. In the current study, we examined the effects of prolonged DIO towards multiple neuropathological processes in an established CAA mouse model [29–31].

Taken together, our data demonstrate that prolonged DIO does not exacerbate any of the gross measures of Aβ levels or glial reactivity. Localized changes in microglial reactivity, and a selective elevation in IL-6 (but not IL-1 or TNF-α) was observed in CAA mice in response to DIO. The lack of gross DIO effects towards Aβ pathogenesis is surprisingly consistent with a number of studies in the literature [32–36], and points to a need to carefully consider the current experimental limitations in studying the effects of DIO on the brain. The potential for ceiling effects in Aβ pathogenesis within the various mouse models, the lack of consistency for DIO based experimentation, and the current experimental limitations for DIO in mice to fully replicate metabolic dysfunction present in human obesity, are discussed.

2. Materials and Methods

2.1. Animals and Dietary Treatments

All animal experiments were approved by the Institutional Animal Care and Use Committee of Pennington Biomedical Research Center. Male and female C57BL/6 mice (“Control”; Charles River Laboratories) and male and female Tg-SwDI (“CAA”) mice were studied. The TgSwDI mouse was generated as described previously [29]. The CAA and corresponding control mice were maintained and genotyped as described previously [29–31,37]. Briefly, the transgenic mouse model expressed the human AβPP770 isoform including the Swedish, Dutch and Iowa mutations under the Thy-1 promoter. Both genotypes were randomly divided and assigned to either a high fat diet (HFD; D12492, Research Diets) or control diet (CD; D12450B) at 2–3 months of age. The HFD provided 60% kcals from fat while the CD provided 10% kcals from fat. The CD was developed as an appropriate, balanced control to the HFD: both contained the necessary mineral and vitamin mixes, 20% kcals from protein and the fat source was soybean oil and lard. These diets were administered for 9–11 months. Mice were housed in standard caging with a 12:12 light/dark cycle and ad libitum access to water and their respective diet unless otherwise noted. Total body fat content was measured via nuclear magnetic resonance (NMR) spectroscopy (Minispec, Brucker Optics, Billerica, MA).

2.2. Immunohistochemistry

The formalin-perfused mouse brain tissues were kept in formalin for 24–48 hours and then processed for paraffin embedding. Samples were sectioned at 5 µm. For immunohistochemistry staining, paraffin was removed by washing slides three times in xylenes, tissue was re-hydrated in a series of alcohol washes (100%, 95%, 70%, and 50%), and then tissue was treated for heat-mediated antigen retrieval. Slides were washed, blocked in 10% serum and incubated overnight with primary antibody anti-collagen IV (Cat. #: ab6586, Abcam, Cambridge, MA, USA) in 3% serum. The next day slides were washed, incubated with biotinylated secondary antibody (Vector, Burlingame, CA, USA), and developed using DAB Peroxidase Substrate Kit (Vector). For double labeling, the slides were re-blocked again in 10% serum and incubated overnight with primary antibody anti-IBA-1(Cat. #: 019-19741, Wako, Osaka, Japan) or anti-GFAP (Cat. #: ab48050, Abcam) in 3% serum. The next day slides were washed, incubated with biotinylated secondary antibody, and developed using VIP Peroxidase Substrate Kit (Vector). Slides were scanned using a Hamamatsu NanoZoomer Digital Slide Scanning System (Hamamatsu City, Japan) at 20× magnification.

2.3. RT-PCR

RT-PCR experiments were performed as done previously by our laboratory [38]. Total RNA from 30 mg mouse striatum was isolated using RNeasy Mini Kit (Qiagen, Valencia, CA) following the manufacturer’s instructions with minor modifications. The corresponding cDNA was made from 2 µg of extracted total RNA by M-MuLV transcriptase (New England Biolabs, Ipswich, MA) using 20 µL of the reverse transcription system according to the manufacturer’s instructions. For quantitative real-time PCR (qPCR) analysis, aliquots of cDNA were subjected to qPCR in 20 µl of 1×Brilliant II QPCR & QRT-PCR Reagents (Agilent Technologies, Santa Clara, CA), 1× primers and TaqMan probe (6-FAM/ZEN/IBFQ mode) and 10 ng of cDNA. Primers and probes for TNF-α, IL-1β, IL-6 and Gapdh were ordered through the PrimeTime Pre-designed qPCR Assays system of Integrated DNA Technologies (IDT, Coralville, IA). Each sample was loaded in triplicate, negative and positive controls were included. Amplification of Gadph was used as an internal reference gene. PCR amplifications were performed as follows: 50°C for 2 minutes, 95°C for 15 minutes, and 40 cycles each with 95°C for 15 seconds and 60°C for 45 seconds using an ABI PRISM 7000 sequence detector according to the manufacturer’s instructions (Applied Biosystems, Foster City, CA). For data analysis, the _ΔΔC_t method was used. Relative mRNA expression of each gene was expressed as the mean and STD of 6 independent total RNA extractions and real-time PCR analyses.

2.4. Western Blotting

Protein lysates from mouse striatum were made in RIPA buffer using a tissue homogenizer and the amount of protein in lysates was estimated using BCA reagent (Thermo Fisher Scientific, Rockford, IL, USA). Protein samples were analyzed by SDS-PAGE and then transferred onto a nitrocellulose membrane. The membrane was then probed with antibodies anti- IBA1, GFAP as described previously and β-actin (Cat. #: sc-47778, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). All electrophoresis and immunoblotting reagents were purchased from Bio-Rad Laboratories (Hercules, CA, USA). The HRP-conjugated secondary antibodies were purchased from the Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA, USA).

2.5. Aβ analysis

The amount of PBS and SDS soluble Aβ was quantified as outlined previously in numerous studies from our laboratories [39–42].

2.6. Statistical Analysis

Statistical significance between two groups was determined using a paired t-test. All data are shown as mean ± standard error of measurement. For all other analyses, statistical significance was determined by ANOVA followed by Fisher's least significant difference post hoc test. P < 0.05 was considered statistically significant (labeled as *).

3. Results

Placement of CAA and control mice on a DIO regiment resulted in the predicted increase in both body weight and adiposity (Fig. 1), as compared to maintaining animals on a control diet (CD). It is important to point out that the current study differed from our recent publication examining the effects of DIO on adiposity in CAA mice [43], in the fact the diet duration in the current study was nearly 2 times longer duration, and the fact the previous study used only male mice (current study an equal mix of male/female mice). We next analyzed the ability of DIO to alter the levels of Aβ₄₀ and Aβ₄₂. As anticipated CAA mice exhibited significant increases in both Aβ₄₀ and Aβ₄₂ (Table 1), as compared to control mice. Surprisingly, prolonged DIO did not significantly alter the levels of in the Aβ₄₀/Aβ₄₂ in either the soluble (PBS) or insoluble (SDS) fractions (Table 1).

Fig. 1. Body weights and body fat following diet induced obesity.

(A) Control HFD mice (C-HFD) were significantly heavier than all other groups (p<0.05) while CAA control diet (CAA-CD) mice weighed significantly less than all other groups (p<0.05). (B) Body fat as measured by NMR revealed increased total body fat (p<0.05) for both groups fed the HFD. The CAA-CD mice had less fat compared to all other groups (p<0.05).

Table 1.

Dietary induced obesity effects on Aβ levels in the context of a mouse model of cerebral amyloid angiopathy

	PBS		SDS
	Aβ40	Aβ42	Aβ40	Aβ42
CAA Control Diet	202 (37)	1,006 (168)	92,040 (8,216)	12,399 (832)
CAA High Fat Diet	98 (41)	746 (205)	84,564 (7,358)	10,251 (1,050)
WT Control Diet	70 (13)	497 (125)	1,091 (62)	10,892 (518)
WT High Fat Diet	12 (12)	542 (70)	862 (46)	11,847 (539)

Data presented as the mean (nM) and SEM of soluble (PBS) and insoluble (SDS) levels of Aβ.

Next we analyzed the effects of DIO towards gross inflammatory signaling in CAA mice. Interestingly, HFD exposure selectively increased IL-6 in CAA mice (Fig. 2), and did not significantly alter either TNF-α or IL-1β in CAA mice following administration of DIO (Fig. 2).

Fig. 2. Effects of the diet induced obesity on the gross inflammatory signaling pathway in mouse brains.

The mRNA level of TNF-α (A), IL-1 (B) and IL-6 (C) in different mouse brain was examined. HFD exposure selectively increased IL-6 in CAA mice (p<0.05).

We next analyzed the impact of DIO towards the levels of microglial and astrocyte gliosis in CAA mice. As expected, CAA mice exhibited a significant increase in both IBA-1 and GFAP levels as compared to control mice (Fig. 3); consistent with elevated levels of microglial and astrocyte reactivity, respectively. DIO did not significantly alter the gross levels of either IBA-1 or GFAP in CAA mice (Fig. 3). While gross levels of GFAP and IBA-1 were not altered by DIO, we next sought to elucidate if any change in the level of these markers could be detected using immunohistochemistry. In these studies we observed that, consistent with our Western blot analysis no significant change in the levels or distribution of GFAP immuoreactivity was observed in CAA mice response to DIO (Fig. 4). This observation was consistent whether analyzing the entire hippocampal gyri, as well as a subset of analysis in CA1 and stratum oriens within the hippocampal gyri. We next analyzed IBA-1 immunoreactivity and observed an apparent increase in the number of IBA-1 labeled, were observed in stratum oriens of the hippocampal gyri (Fig. 5). The increase in large phagocytic microglia near cerebral arterioles was selective for stratum oriens and not seen in the CA1 or other areas of the hippocampal gyri (Fig. 5).

Fig. 3. Effects of the diet induced obesity on astrocyte and microglial reactivity.

Diet induced obesity did not significantly alter the gross protein levels of either astrocyte (GFAP) or microglial (IBA-1) reactivity.

Fig. 4. Diet induced obesity did not significantly alter immunoreactivity and distribution of GFAP in the hippocampus of CAA mice.

The levels of astrocyte reactivity were measured in the brain using GFAP immunoreactivity (purple) and co-localized with cerebral blood vessels using collagen IV immunoreactivity (brown).

Fig. 5. Effects of diet induced obesity on the immuoreactivity and distribution of IBA1 in the hippocampus of CAA mice.

The levels of microglial reactivity were measured in the brain using IBA-1 immunoreactivity (purple) and co-localized with cerebral blood vessels using collagen IV immunoreactivity (brown).

4. Discussion

Our results indicate that DIO does not significantly alter gross Aβ levels in either soluble or insoluble fractions of CAA mice. These data are consistent with previous studies which identified that DIO in mouse models of Aβ neurodegeneration do not alter the gross levels of Aβ [33–35]. Interestingly, studies which combine obesity with hypercholesterolemia are known to more consistently increase Aβ pathology in mice [44–46]. Taken together, these data suggest that more than just obesity may be required to consistently alter gross levels of Aβ levels in mice. These observations are particularly important when considering the preclinical use of mice for the study of obesity-Aβ interactions. For example, it is well known that mouse models of DIO do not recapitulate the pleotropic array of metabolic syndrome observed in obese humans. Obesity in humans is accompanied by the development of metabolic syndrome [47–49] including the presence of hypercholesterolemia, diabetes, hyperlipidemia and often accompanied by morbidities such as atherosclerosis and hypertension [48]. In mice, most of these modifications do not occur in response to DIO, and those that do are only modestly elevated (insulin resistance, hyperlipidemia). For these reasons, the preclinical studies of DIO effects on brain neuropathology will likely require an approach in which obesity is accompanied by hypercholesterolemia or other cardiovascular disease risk factor. Despite the differences between mice and human DIO studies, it appears that the vasculature is a key target of obesity induced effects on the various tissues of the body [50–52]. Most studies to date have focused on the effects of obesity on the vasculature of the heart and peripheral organs, although it is highly likely cerebrovasculature will be a principle target of DIO effects on the brain.

Studies are needed in mice to identify which aspects of Aβ regulation (synthesis, processing, degradation, oligomerization, deposition) are consistently altered by obesity in humans and clinically relevant mouse models. Currently, little to nothing is known in terms of the effects of obesity on key regulatory Aβ enzymes such as BACE and γ-secretase, nor have studies identified if obesity promotes meaningful changes the rates/levels of Aβ assembly and/or deposition. In the current study while we did not see a significant change in Aβ levels in either the soluble or insoluble pools, it is important to point out that these studies were conducted following prolonged DIO, and the CAA mice at this age are known to have near maximal levels of Aβ. These studies raise the likelihood of ceiling effects (in terms of potentially maximal Aβ levels/changes) reducing the chances for identifying DIO modulation of Aβ. Careful consideration of ceiling effects for pathology are clearly an important consideration in the designing of future studies looking at the ability of DIO to modulate Aβ pathogenesis. Studies in the future need to carefully design studies (time points, diet duration, Aβ model) in a manner in which there is significant room in the modulation of Aβ levels to identify the ability of DIO to increase, as well as decrease Aβ levels in the brain.

Additional concerns for studies of DIO effects on the brain include the fact that there is little consistency in the types of diets used, the duration of diets, the control diets used for the majority of studies in the literature (Table 2). The importance of control diets utilized is critical as the use of chow diet as controls in DIO studies does not allow for the control of the types of macronutrients utilized (source, quality, etc). This can make a huge difference in terms of experimental outcomes and the interpretation of whether DIO is capable of modulating specific aspects of AD related pathogenesis. Pair feeding studies, in which the amount of calories is held constant between a control diet and high fat diet, can be useful in identifying whether it is an increase in obesity or dietary fat which is responsible for mediating a given aspect of pathology. To date, pair feeding studies have not been used in studies of brain pathogenesis, although it is important to point out that pair feeding studies do have a potential confound in the fact that access to food is restricted. This restriction of food access is necessary to control for calories, but can induce stress in rodents, with stress potentially modulating brain homeostasis.

Table 2.

Diet induced obesity and alterations in mouse brain homeostasis

Diet	Duration	Mouse Strain	End Point & Results	Ref.
60%-HFD, 10%-CD	16 weeks	C57BL/6	increased reactive oxygen species (ROS); decreased cognition	[43]
60%-HFD, 5%-CD	16 weeks	C57BL/6	increased IR, ROS; no AD histopathology, no increase in Aβ or phospho-tau, no impairments in IGF signaling or acetylcholine homeostasis	[33]
10% sucrose in water	25 weeks	B6C3-Tg(APPswe, PSEN1dE9)85Dbo/J	Metabolic dysfunction, memory impairment, increased apoE and aggregation of Aβ in the brain	[34]
40%-WD, 10%-CD	4 weeks	APPswe/PS1M146V; PS1M146V; C57BL/6	increased cerebral ROS, alterations in brain Aβ	[36]
60%-HFD	16 weeks	C57BL/6	impairment of fear-conditioning test	[53]
60%-HFD, 10%-CD	20 weeks	APPSwe/Ind	metabolic dysfunction; decreased cognition	[54]
21.2%-HFD, 5.5%-CD	22 weeks	C57BL/6	increased pro-inflammatory factors and APP expression in brains and fat tissues	[55]
55% -HFD	8 weeks	WT, PSEN1, APP/PSEN1	AD increases body weight gain, glucose intolerance and insulin resistance induced by HFD in mice	[56]
60%-HFD	20 weeks	C57BL/6J, MC4R−/−, Tie2-GFP	enhanced GFAP-immunoreactivity in the medial preoptic, paraventricular and dorsomedial nuclei of the hypothalamus.	[57]
60%-HFD	22 weeks	C57BL/6	impaired hippocampus-dependent spatial memory, alteration in the expression of corresponding genes	[58]
58%-HFD, 10.5%-CD	12 weeks	C57BL/6	promotion of negative emotional states and depressive-like symptomology	[59]
45%-HFD, 10%-CD	14 weeks	C57BL/6	HFD selectively protects against the effects of chronic social stress in the mouse	[60]

It is well established that the different mouse strains exhibit considerable variability in regards to their susceptibility to DIO [33–36]. Additionally, even within the same strain there is known to be considerable heterogeneity in body weight gain following DIO that appears to be due to genetic influences [61–63]. While traditionally, these variances have been the focus of understanding the causes of obesity, they have considerable influence on how mouse studies are used to study the consequences of DIO on multiple tissues, including the brain. For example, recent studies have demonstrated that increases in APP occur in response to obesity in both brain and adipose tissue [55–64], with decreases in obesity contributing to lower APP levels [65]. These studies did not share significant overlap in their design, nor did they account for the influences of genetic susceptibility to DIO, making it difficult to understand what is truly occurring in terms of DIO effects on APP levels.

Similar to Aβ, we believe that the lack of an effect of DIO on gross inflammatory signaling and glial reactivity in CAA mice is also due in large part to a ceiling effect in the level of Aβ pathology in CAA mice. Previous studies from our laboratory and others has firmly established the ability of DIO to modulate both glial reactivity and inflammatory signaling in the rodent brain [55,66–70]. It is highly likely that the modulation of these two aspects of brain homeostasis contributes to the mechanism by which DIO modulates brain pathology and brain function, with studies currently underway in our laboratory to establish this experimentally in mouse models of Aβ pathogenesis. Both microglia and astrocytes are key to the neurovascular unit, raising the potential for DIO modulating neurovascular unit homeostasis [71–73] and potentially serve as the basis for DIO increasing susceptibility to Alzheimer’s disease [74,75].

It is important to point out that in humans and rodent studies, increases in visceral adiposity are preferentially linked to the morbidities and mortality associated with obesity [76–78]. The ability of visceral adiposity to preferentially mediate the negative effects of DIO appears to be due to both the fact it is biochemically distinct from visceral adiposity, as well as the fact it is near vital internal organs of the body. Establishing the specific role of different adipose depots in mediating the negative effects of DIO on the brain is a critical area of future research. Similarly, it is well established that gastric surgery in humans and rodents (gastric sleeve, gastric bypass, roux en y) has significant beneficial effects on mobidity and mortality [79–82]. Interestingly, these benefits often occur before significant weight loss has occurred, and in individuals who remain morbidly obese. It is presumed these rapid beneficial effects are mediated by changes in gut hormones such as GLP-1, although it remains to be determined what beneficial effects of gastric surgery are on Aβ pathogenesis and other aspects of neurodegenerative disease.

Adipose tissue is the largest endocrine tissue in most Americans, and the adipokines secreted by adipose tissue almost certainly contribute to brain homeostasis. Understanding the influence of adipokines like leptin on brain homeostasis fairly limited to date. Interestingly, studies are showing complex roles for these adipokines in regulating Aβ processing and Aβ levels [2,83–85]. Understanding the effects of physiological levels of adipokines and their targets in the brain, as related to neurodegenerative disease, is another crucial area of research. Db/Db and Ob/Ob mice have been useful in understanding adipose biology and adipose modulation of other tissues of the body, in particular the role of leptin signaling in different aspects of biology and pathological processes [14,20]. Studies involving crossing Ob/Ob diabetic mice with mouse models of Aβ pathology have demonstrated evidence for accelerated neurodegeneration and cognitive impairment [86]. It is important to point out that these studies involve acute experimentation (animals less than 6 months old) because of the severity of obesity and extreme insulin resistance. The magnitude of metabolic disturbances and obesity appear to limit the use of this model to study obesity mediated effects on the brain, and make it difficult to discern whether the effects on the brain are a byproduct of a mouse with a myriad of morbidities and short life span.