Background
Alzheimer's disease (AD) is an age-related neurodegenerative disorder associated with progressive functional decline, dementia and neuronal loss. Demographics make evident that the prevalence of AD will increase substantially over the coming decades. Patients initially exhibit an inability to assimilate new information and as the disease progresses, both declarative and nondeclarative memory become ever more profoundly impaired [
1]. The pervasive societal and economic burden created by this debilitating disease should provide sufficient incentive for the development of new natural history-modifying therapeutic approaches. However, because the mechanistic underpinnings of AD are incompletely understood, the clinical disease spectrum broad, and the neuropathological features of its initiation and progression limited, the development of such potential disease modifying therapies has been relatively limited.
The pathological hallmarks of the AD brain include extracellular proteinaceous deposits (plaques), composed largely of amyloid beta (Aβ) peptides, and intraneuronal neurofibrillary tangles (NFTs), which are characterized by excessive phosphorylation of tau protein. Other AD-related histopathologic features include, but are not limited to, astrogliosis, microglial activation, and reduction of synaptic integrity. These features appear to arise in a region- and time-dependent manner (reviewed in [
2]). Amyloid pathology evolves in stages: early involvement is anatomically circumscribed to the basal neocortex, most often within poorly myelinated temporal areas; progression involves adjacent neocortical areas, the hippocampal formation, perforant path inclusive of its coursing through the subiculum and termination within the molecular layers of the dentate gyrus, and; finally the process involves all cortical areas [
3]. Neurofibrillary tangle pathology is also progressive: Initially involving projection neurons with somata in the transentorhinal region, tangles then extend to the entorhinal region proper typically in the absence of amyloid deposition. Subsequent progression to the hippocampus and temporal proneocortex, and then association neocortex, followed by superiolateral spread and ultimately extending to primary neocortical areas [
4‐
6]. Moreover, individuals diagnosed with mild cognitive impairment, a forme fruste of AD, display decreased entorhinal and hippocampal volume, primarily associated with diminished neuron number as compared to non-cognitively impaired controls [
7‐
10]. These data suggest that the entorhinal cortex and hippocampus are selectively vulnerable early during the disease process.
Gaining an enhanced understanding of why these brain regions are specifically susceptible to neurodegeneration in the context of AD and elucidating the mechanisms underlying these disease processes has been the subject of intensive investigation over the past several decades. Attention has been focused upon synaptic dysfunction, due to the previously observed diminution of cholinergic synapse density and overall synapse numbers during early stages of AD [
11,
12]. Additionally, mouse models overexpressing human amyloid precursor protein (APP), the protein from which pathogenic Aβ peptides are proteolytically derived, exhibit decreased synaptic function antecedent to plaque deposition [
13], thereby further implicating disrupted synaptic function in early stages of AD pathogenesis.
Inflammatory processes, marked by activated microglia and astrocytes in the post-mortem AD brain some of which co-localize to plaques and tangles, have long been hypothesized to contribute to AD pathogenesis [
14]. The role that this response plays in the disease process, especially during pre-symptomatic stages, is not well defined. There exist multiple means by which inflammatory processes can affect neurons and potentially synaptic function in AD. Cytokines have been shown to be expressed in response to Aβ generation and a subset of these molecules have demonstrated neurotoxic activities [
15‐
17]. Such observations imply these inflammatory molecules may serve to mechanistically link the elaboration of pathological hallmarks and synaptic dysfunction. We hypothesized that inflammation plays a role early during the disease process, at a time when synaptic dysfunction and early cognitive deficits first become evident. Disease-related inflammatory contributors to synaptic dysfunction found in early AD have long been debated, but such studies have been hampered by the lack of age-matched, early-stage human post-mortem tissue samples as well as AD-relevant animal models. In the present study, we sought to determine the temporal and region-specific expression of inflammatory molecules, previously implicated in late-stage AD, in the context of a mouse model that develops amyloid and tau pathology. A triple-transgenic model of AD (3xTg-AD) has recently been created that harbors three disease-relevant genetic alterations: a human Presenilin M146V knock-in mutation (PS1M146V), human amyloid precursor protein Swedish mutation (APPswe), and the human tauP301L mutation. These mice develop plaques and tangles in a spatial and time-dependent manner similar to pathological hallmarks observed in the brains of AD-afflicted individuals [
18,
19]. Most notably, this is the first animal model developed to date which facilitates the study of inflammation in the context of both amyloid and tau pathology. We performed region-specific quantitative transcript analyses and unbiased stereological counting to correlate regional and temporal changes in inflammatory molecule expression profiles to alterations in inflammatory cell numbers and AD-related pathologies. Our findings further implicate inflammatory processes as playing a role early during the disease process, and that regional differences exist that may elucidate why particular brain regions are more susceptible to AD-related disease mechanisms.
Materials and methods
Strains of mice
Triple transgenic (3xTg-AD) mice were created as previously described [
18,
19]. Age-matched 2, 3, and 6 month-old male mice were used in all studies (n = 6 per experimental group for biochemical assays, n = 4 per experimental group for quantitative stereological studies). Age-matched male C57BL/6 mice were used as non-transgenic controls in all experiments. All animal housing and procedures were performed in compliance with guidelines established by the University Committee of Animal Resources at the University of Rochester.
Quantitative real-time PCR analysis of pro-inflammatory molecules from brain-derived RNA
RNA was isolated from microdissected hippocampus- or entorhinal cortex-enriched tissue from 2, 3, and 6 month-old 3xTg-AD and non-transgenic mice with TRIzol solution (Invitrogen, Carlsbad, CA). RNA was treated with RQ DNAse I (Promega, Madison, WI) to selectively degrade any contaminating genomic DNA, followed by phenol:chloroform extraction and ethanol precipitation. One microgram of total RNA was reverse transcribed using Applied Biosystems High-Capacity cDNA Archive Kit. An aliquot of cDNA (100 ng) was used to assess presence of 23 inflammatory targets per mouse, and was analyzed in a standard PE7900HT quantitative PCR reaction using a Taqman Assay on Demand primer probe sets in Microfluidic cards (Applied Biosystems, Foster City, CA) and 100 μL MasterMix containing HotStart DNA polymerase (Eurogentec, Belgium). 18s RNA served as the control to which all samples were normalized (Applied Biosystems, Foster City, CA). We further analyzed the data using the ΔΔCT method, normalizing the 3 and 6 month-old 3xTg-AD and control mouse samples to the 2 month-old 3xTg-AD and non-transgenic samples, respectively.
Quantitative histochemical analysis of macrophages and microglia in brains of 3xTg-AD and non-transgenic mice
Age-matched 3xTg-AD and non-transgenic mice were sacrificed and processed with 4% paraformaldehyde (PFA)/PB trans-cardiac perfusions; brains were removed and post-fixed overnight with 4% PFA/PB. Sequentially, brains were transferred to 20% sucrose in PBS overnight and then 30% sucrose where they remained until sectioning. Brains were sectioned coronally (30 μm) on a sliding microtome, and stored in cryoprotectant until used for immunohistochemistry.
Sections were washed four times for 3 min. each in PB to remove cyroprotectant. To quench endogenous peroxidase activity, sections were incubated for 25 min. with 3% H2O2 (Sigma). Sections were mounted onto slides and allowed to dry. Slides were incubated in 0.15 M PB + 0.4% Triton-X100 for 5 min. at room temperature (RT; 22°C) to permeabilize the tissue. Then slides were incubated with blocking solution containing 3% normal goat serum, 3% bovine serum albumin, and 0.4% Triton-X 100 in 0.15 M PB for 1 hr. Slides were incubated with rat monoclonal anti-F4/80 antibody (Serotec, 1:100) overnight in blocking solution. Next, slides were washed eight times for 3 min. each with 0.15 M PB prior to incubation with Vectastain biotinylated goat anti-immunoglobulin (Vector Laboratories, Burlingame, CA) for 2 hrs. at RT. Excessive secondary antibody was washed in 0.15 M PB and incubated with A and B reagents (Vector Laboratories, Burlingame, CA) to conjugate HRP. Slides were developed using a DAB peroxidase kit, according to manufacturer's instructions for nickel enhancement (Vector Laboratories, Burlingame, CA).
Positively stained F4/80-expressing cells were visualized using an Olympus AX-70 microscope equipped with a motorized stage (Olympus, Melville, NY) and the MCID 6.0 Elite Imaging Software (Imaging Research, Inc.). Sections were tiled under 4× magnification. Five equal sections of entorhinal cortex and seven equal sections of hippocampus from each mouse (4 mice total) per timepoint were analyzed. Fifty percent of the defined region of interest in the entorhinal cortex or hippocampus was assessed, under 60× magnification. The coordinates from which sections were chosen for the entorhinal cortex were 2.92 mm to 4.04 mm posterior from Bregma. The sections counted in the hippocampus were from 1.70 mm to 3.40 mm posterior from Bregma.
Qualitative immunohistochemical analysis of amyloid deposition in 3xTg-AD and non-transgenic mice
Sections were washed three times for 5 min. each, then twice for 30 min. in PBS to remove cryoprotectant. To quench endogenous peroxidase activity, sections were incubated with 3% H2O2 and 3% methanol for 25 min. Sections were then washed twice for 5 min. each with PBS, followed by epitope retrieval treatment with 90% formic acid for 5 min. at RT. Next, sections were washed twice for 5 min. each with PBS. Tissue was permeabilized with PBS + 0.1% PBS/Triton-X 100. Sections were then incubated for 1 hr at RT with PBS + 0.1% PBS/Triton-X 100 + 10% normal goat serum. Sections were incubated overnight at 4°C with primary 6E10 antibody (Signet, 1:1000) in PBS 0.1% PBS/Triton-X 100 + 1% normal goat serum. Samples were washed twice for 10 min. each with PBS + 0.1% Triton-X 100 + 1% normal goat serum prior to addition of secondary antibody. The mouse HRP ABC kit was used according to manufacturer's protocol (Vector Laboratories, Burlingame, CA). Excessive secondary antibody was washed in PBS and developed using a DAB peroxidase kit, according to manufacturer's instructions for nickel enhancement (Vector Laboratories, Burlingame, CA) and mounted on slides.
Discussion
Dissecting the role that inflammation plays early in AD is challenging, as AD is a complex chronic disorder with varying pathologic sequelae from which the underlying causative mechanisms are unknown. Activation of microglia and astrocytes, and the presence of many inflammatory mediators, including cytokines, chemokines and complement proteins have been only identified in the post-mortem AD brain in the vicinity of senile plaques and NFTs [
20,
25]. This observation leads one to question if inflammation is involved early during the course of AD and, if so, how does it contribute to pathogenesis? Understanding the earliest events is of utmost importance, as inflammation may represent a viable therapeutic target of AD. Interestingly, retrospective studies assessing the effects of non-steroidal anti-inflammatory drugs (NSAIDs) on nondemented individuals have shown decreased risk of developing AD when these individuals utilized NSAIDs for prolonged periods of time [
26,
27]. Our studies aimed to identify the earliest period during which inflammatory processes initiate in the 3xTg-AD mouse model [
19]. Our results illustrate 3 main points: 1) Inflammatory processes precede significant extracellular amyloid plaque deposition in the 3xTg-AD brain, substantiated by increased TNF-α and MCP-1 transcript levels, coincident temporally with the production of intracellular Aβ accumulation. 2) The expression of these molecules is spatially localized to the entorhinal cortex but not hippocampus at the early time-points assessed. 3) There is a marked increase in the number of microglia and macrophages in the entorhinal cortex that correlates with when TNF-α and MCP-1 transcript levels are significantly up-regulated.
In the late-stage AD brain, it has been shown that inflammatory molecules are produced primarily by microglia and astrocytes as they respond to plaques and neuronal damage [
17]. Our finding of increased TNF-α and MCP-1 expression prior to significant plaque deposition in 3xTg-AD mice, which occurs extensively at 12 months [
19], may represent a contributory role between inflammatory processes and early AD pathogenesis. Precisely how these molecules impart effects in 3xTg-AD mice at this early time-point is not certain; however, recent evidence has suggested that TNF-α and MCP-1, as well as other pro-inflammatory molecules may play a role in inhibiting microglial phagocytosis of fibrillar Aβ
in vitro [
28]. Likewise, increased inflammatory responses and subsequent secretion of cytokines, in particular, IL-1β, may play an important role in tau phosphorylation in the 3xTg-AD model [
29]. In this study, activation of microglia by LPS only affected the tau pathology via cdk5/p25 activation, but not the amyloid pathology, further highlighting the potential pathophysiological changes that can be induced by inflammation in AD. Certainly, inflammatory mediators have been implicated as being both protective and exacerbating, depending on the model system and the levels of cytokine present [
30].
TNF-α can be expressed by astrocytes, microglia and neurons in response to various stimuli in the CNS [
17]. Initially, TNF-α is an innate mediator, promoting chemokine and cytokine expression and extravasation of other immune cells. One possible mechanism that may implicate TNF-α in contributing to AD pathogenesis is evidence that it can increase Aβ peptide production [
31]. Additionally, inflammatory molecule signaling may cause increased cleavage of APP by the γ-secretase complex, whereby TNF-α, IL-1β, and IFN-γ have been shown to enhance production of Aβ peptides via a γ-secretase-dependent mechanism
in vitro. Moreover, antagonizing TNFR1 signaling can lead to diminished γ-secretase activity [
32]. Further evidence supporting pathogenic effects of TNF-α-mediated signaling is TNFR1 and TRADD, a TNF receptor adaptor protein that allows for NF-κB and JNK activation, are both increased in AD tissue and APPswe mice. This increase is correlative with TUNEL-positive neurons in primary hippocampal cultures [
33]. Collectively, these observations suggest TNF-α contributes to aberrant APP processing and initiation of pro-apoptotic pathways.
MCP-1 is a chemokine that is expressed by microglia and astrocytes that facilitates extravasation of immune cells expressing its cognate receptor, CCR2, to cross the blood brain barrier and guides them to the site of damage. As with TNF-α, the role of MCP-1 in AD pathophysiology is uncertain. A recent study of APPswe/CCL2 (MCP-1) bigenic mice showed increased diffuse Aβ deposition, as compared to APPswe mice at 14 months of age. Since changes were not observed in APP processing, the authors concluded that MCP-1 overexpression in APPswe mice correlated with diminished clearance of Aβ [
34]. Overall, it is interesting that of the 23 immunomodulatory markers assessed in our study, TNF-α and MCP-1 were the only two that changed significantly over time, possibly signifying their importance during nascent stages of AD pathogenesis. Perhaps, the other inflammatory targets are triggered at later stages of the disease in response to further neurodegenerative events.
Exogenously applied Aβ can trigger the expression of cytokines
in vitro and when injected directly into the mouse brain [
24,
35]. However, the fascinating result in our study is that expression of TNF-α and MCP-1 was detected specifically within the entorhinal cortex and not the hippocampus, despite the fact that immunocytochemically detectable intraneuronal Aβ increased over time in both brain regions. Additionally, although not statistically significant, TNF-α and MCP-1 transcript levels were elevated at 3 months of age in 3xTg-AD mouse entorhinal cortex, and were increased to statistical significance by 6 months of age suggesting a state of chronic up-regulation and positive-feedback for the expression of both of these inflammatory molecules. Therefore, we are unable to conclude, as detected by the methodology employed in this study, that intracellular Aβ accumulation is the sole contributing factor promoting TNF-α and MCP-1 transcript expression specifically in the entorhinal cortex. It is possible that these molecules are neuronally expressed within the entorhinal cortex because they are inherently more sensitive to accumulating Aβ, as other neurons were shown previously to express both of these molecules during times of stress [
17]. Another possibility is that the entorhinal cortex elaborates inflammatory molecule expression owing to the structure of Aβ elaborated. For example, oligomeric Aβ is posited to be the more toxic structural intermediate arising during Aβ fibrillogenesis, and this form has been shown to readily induce cytokine expression
in vitro [
36]. Regional differences in intracellular and extracellular oligomeric Aβ profiles
in vivo may, therefore, account for the regional specificity of cytokine/chemokine expression and microglial activation that we observed in the 3xTg-AD mice. Conversely, the regional elaboration of TNF-α and MCP-1 may occur via an Aβ-independent mechanism and caused by an environmental stimulus, such as region-selective oxidative stress. Practico and colleagues demonstrated that lipid peroxidation in the APPswe brain can occur prior to Aβ deposition in APPswe mice [
37], suggesting that disruption of cellular membrane homeostasis could also contribute to inflammatory molecule induction, and perhaps, regional differences in lipid peroxidation profiles are responsible for our region-specific observations in 3xTg-AD mice.
The significant increase in the number of microglia and macrophages in the entorhinal cortex from 2 to 6 months of age in 3xTg-AD mice is coincident with the increase of TNF-α and MCP-1. We believe this could be due to microglial proliferation, activation of the resting resident population of brain microglia and macrophages and/or recruitment of peripheral macrophage-like cells (F4/80-positive) from outside the brain. Macrophages express CCR2 and thus, are capable of responding to a compromised entorhinal cortex via chemotaxis. Whether the observed increase in F4/80
+ cell number indicates a homeostatic or pathologic response is not clear. APPswe/CCL2 mice demonstrate enhanced microglial numbers that are concurrent with increased extracellular Aβ deposition, that the authors postulate is due to an inability to effectively clear Aβ [
34]. This may relate partially to the increased ApoE levels observed in APPswe/CCL2 mice produced by microglia and macrophages. If a similar mechanism is at play in the 3xTg-AD mouse model, this finding suggests a pathogenic role for these cells in initiating degeneration within the entorhinal cortex.
In summary, our results indicate a potential early role for inflammatory processes in the temporal and spatial evolution of AD pathogenesis. Because TNF-α and MCP-1 are produced specifically within the entorhinal cortex where human AD has been shown to arise, these molecules are likely playing an instrumental role in disease perpetuation. This work provides insight into the involvement of TNF-α and MCP-1 mediated inflammation in the temporal and spatial progression of early AD pathogenic events and may potentially herald new therapeutic targets. Our use of the 3xTg-AD model to assess these early events is unique, as all previous studies have examined inflammatory processes in the context of either amyloid or tau pathology, but not both. This transgenic mouse allows us to directly examine the dynamic interplay of inflammation, Aβ pathology, and tau dysfunction. Modulating TNF-α and MCP-1 function in future studies will elucidate how these inflammatory mediators influence the severity and progression of AD-related pathology and synaptic dysfunction.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MCJ carried out the quantitative real-time PCR, 6E10 immunohistochemistry, F4/80 immunohistochemistry, stereology, experimental analysis and data interpretation, and prepared the manuscript. MAM performed tissue microdissection, brain sectioning, and 6E10 immunohistochemistry. SO and FML conceived the design of and generated the 3xTg-AD mouse model. HJF and WJB conceived the design of the study, aided in the preparation of the manuscript, and provided critical analysis of the manuscript.