Introduction
Alzheimer’s disease (AD) is the most common neurodegenerative disease and dementia disorder. Beta-amyloid (Aβ) plaques, neurofibrillary tangles, neuroinflammation and neuronal loss in the brain are pathological hallmarks of AD that are thought to play a key role in the progressive decline in episodic memory and cognition [
1,
2]. Recent advancements in molecular imaging using positron emission tomography (PET) have allowed the visualization of fibrillar Aβ plaques and the monitoring of disease progression in vivo in AD patients [
3]. Amyloid tracers developed for human use include
11C-Pittsburgh compound B (
11C-PIB) [
4],
11C-BF-227 [
5],
11C-AZD2184 [
6],
18F-FDDNP [
7],
18F-florbetapir [
8],
18F-flutemetamol [
9],
18F-florbetaben [
10] and
18F-AZD4694 [
11]. Recently,
18F-florbetapir,
18F-flutemetamol and
18F-florbetaben have been approved by the European Medicines Agency (EMA) and by the US Food and Drug Administration (FDA) to visualize amyloid plaques. We recently reported that florbetaben, PIB and florbetapir all bind to a high-affinity binding site in postmortem AD brain tissues [
12], demonstrating their reliability for detecting fibrillar Aβ deposits. Longitudinal amyloid PET imaging in large at-risk and patient populations has shown that it takes ≈ 20 years for build-up of pathological amyloid plaque load in the brain [
13].
Neuroinflammation is increasingly recognized to play an early role in AD [
14‐
16]. Studies in postmortem AD brains have demonstrated abundant reactive astrocytes and microglia around amyloid plaques [
17], but little is known about their in vivo distribution or function. PET imaging of astrocytosis in subjects with mild cognitive impairment (MCI) and in AD patients using
11C-deuterium-
L-deprenyl (
11C-DED) has suggested that astrocytosis occurs early in AD [
18].
A wide range of transgenic mice harbouring familial AD mutations that model different aspects of AD pathogenesis are currently available. Although none of these fully replicates the disease, they have provided important insights into the pathophysiology of Aβ toxicity. The development of pathology varies among the mouse strains. APPswe mice carrying the APP Swedish mutation develop amyloid pathology more slowly than APP/PS1 and APP23 mice [
19].
In this study, we investigated the time course of astrocytosis and amyloid plaque deposition by multitracer in vivo microPET imaging in APPswe mice aged 6–24 months. The postmortem brains were then analysed using correlative immunohistochemistry and autoradiography.
Discussion
The use of multitracer PET imaging in an AD transgenic mouse model to investigate the evolution of and relationships between astrocytosis and amyloid plaque deposition in early AD has, to our knowledge, not been described before, as indicated also in a recent review [
27]. Previous amyloid PET studies in AD transgenic mice have reported various findings [
27]. For example, retention of
11C-PIB was no greater in the brains of 12-month-old APP/PS1 mice than in wt mice in one study [
28], while subsequent studies reported increased
11C-PIB retention from 17 to 18 months in APP23 mice [
29,
30], but not in 22-month-old APPswe mice [
30,
31]. Other similar studies in APP/PS1 mice showed increased retention at 9 months [
32] and no difference at 15–22 months compared to age-matched wt mice [
30]. Increased PET retention of
18F-florbetaben in 16-month-old APPswe mice [
33] and
18F-florbetapir in 5-month-old APP/PS1 mice [
34] has been reported.
In this study, we used PET imaging to detect increased
11C-AZD2184 retention in older APPswe mice. These in vivo studies showed a significant increase in retention in the cortices of 18- to 24-month-old APPswe mice versus age-matched wt mice, and a similar increase in both the cortex and the hippocampus was observed postmortem with
3H-AZD2184 and
3H-PIB in vitro autoradiography. Age-related increases in
3H-AZD2184 binding densities, as well as with
3H-PIB, and Aβ
42 immunopositive deposits were also observed in the cortices and hippocampi of APPswe mice, in agreement with previously published findings of Aβ deposition in postmortem immunohistochemical studies in this model [
35].
The gene expression and cellular morphology of reactive astrocytes are highly heterogeneous across different brain regions in both healthy and neurodegenerative- diseased brains [
36,
37]. Studies in AD transgenic mice have reported widespread astroglial atrophy in the brain in the earlier stages of AD prior to Aβ plaque deposition [
38‐
40], but have also indicated the presence of hypertrophic astrocytes surrounding amyloid plaques [
41]. The Aβ aggregates in the brain increase the number of reactive astrocytes and/or phenotypic changes and upregulate GFAP expression [
42,
43].
Our 11C-DED PET imaging findings demonstrated different time courses for astrocytosis in APPswe and wt mice; 11C-DED binding significantly decreased with age in APPswe mice in cortex and hippocampus, while it was not correlated with age in wt mice. 11C-DED binding was also significantly higher in the cortices and hippocampi of 6-month-old APPswe mice compared to other age groups of APPswe and compared to 8- to 15-month-old wt mice. These findings and the observed absence of Aβ plaques in 6-month-old APPswe mice indicate that astrocytosis occurs prior to Aβ plaque deposition.
Previous studies have measured reactive astrocytes using
3H-
L-deprenyl in vitro in AD postmortem brain tissue [
26,
44,
45] and
11C-DED in vivo [
18,
46]. Interestingly, early
11C-DED PET binding was observed in prodromal AD patients [
18], consistent with our results in APPswe mice. Astrocytosis as measured by
11C-DED PET was also observed to be negatively correlated with grey matter density in the parahippocampus at early prodromal AD stages [
47]. Early
11C-DED PET binding has also been reported in presymptomatic carriers of autosomal dominant AD mutations several decades before onset of symptoms and earlier than Aβ deposition [
48]. Therefore, our findings of early
11C-DED PET binding in APPswe mice before Aβ deposition are consistent with reported in vivo PET findings in humans, highlighting the translational aspects of this study.
In contrast to the in vivo
11C-DED PET binding results, there were no significant differences between APPswe and wt mice in the in vitro
3H-
L-deprenyl autoradiography results. Possible explanations for this include either a reduction in monoamine oxidase B (MAOB) enzyme activity when measured in vitro compared to in vivo [
46] or technical limitations due to a reduced in vitro enzyme activity in transgenic APPswe mice compared to humans.
The level of GFAP upregulation in astrocytes has been reported to be dependent on both the brain region and the context, which could in part underlie the heterogeneity of functions mediated by astrocytes in the central nervous system, evidenced by their diverse neuroprotective or neurotoxic functions across different stages in AD [
49]. Our immunostaining results in APPswe mice revealed more GFAP reactive astrocytes in the cortices and hippocampi of 18- to 24-month-old than of 6-month-old APPswe mice, although the age-dependent increase was significant only in the hippocampus. There is increasing evidence that GFAP is not expressed uniformly by all astrocytes [
50] and that different GFAP isoforms develop in response to plaque-related gliosis, as shown in APP/PS1 and 3xTg AD transgenic mice [
51]. It has been reported [
52] that about 80 % of the astrocytes in the hippocampus express GFAP, compared to only 15–20 % of those in the cortex, which could at least partially explain our immunohistochemical results showing a significant increase in GFAP with age in the hippocampus. Other studies in AD transgenic mice have reported GFAP expression prior to Aβ plaque deposition [
53] and also as an age-dependent process that is correlated with oligomeric Aβ but not with plaque burden [
54]. GFAP expression in astrocytes has also been reported as a late event related to plaque formation and maturation, and as a neuroprotective event that limits Aβ plaque growth [
55].
In our present study, in contrast to the early increase in
11C-DED binding in the APPswe mouse brain cortex, GFAP immunoreactivity was predominantly a late event. There was a significant age-dependent increase in the number of GFAP
+ reactive astrocytes in the hippocampus, possibly indicating the presence of distinct subpopulations of astrocytes and/or different stages of reactive astrocytosis. The different time course of increases in
11C-DED binding and GFAP upregulation observed in APPswe mice is consistent with findings in human AD. Quantitative autoradiography studies in postmortem AD brains have shown a strong regional correlation between the number of GFAP
+ reactive astrocytes and the extent of in vivo
11C-PIB and in vitro
3H-PIB binding, but there appears to be no regional correlation between postmortem
3H-
L-deprenyl and in vivo
11C-PIB binding [
45]. The regional and laminar distribution patterns of
3H-
L-deprenyl reactive astrocytes also differed from those of fibrillar Aβ in AD autopsy brains [
26]. Thus, our observation of early
11C-DED binding in the cortices of APPswe mice might reflect the presence of a subset of activated astrocytes that is functionally different from those measured by GFAP at later stages.
The small, soluble Aβ forms that have been observed in APPswe mice from birth [
56,
20] could influence astrocyte function. For example, Aβ
25–35 peptides caused overexpression of MAOB in cultured rat astrocytes [
57]. Further, MAOB overexpression in astrocytes led to production of proinflammatory molecules contributing to exacerbated neuroinflammation and Aβ plaque formation at later stages [
58]. Our observation of early elevated
11C-DED binding might thus reflect a reaction of astrocytes to these small Aβ forms, with potential beneficial and/or neurotoxic consequences. Activated astrocytes appear to play a role in the clearance of Aβ [
59,
60]. Whether the reactive astrocytes that were observed early in the development of AD, as measured by elevated
11C-DED binding, have a phagocytic function requires further investigation.
One limitation of this study was the limited spatial resolution of small animal PET imaging relative to the mouse brain regions selected for quantification, which contained some thin shapes being susceptible to partial volume effects. This might account for the lower sensitivity of our in vivo versus in vitro images. Inevitably, transgenic AD mice are an inherently limited model for human disease and APPswe mice might phenotypically reflect only some aspects of AD [
19]. Astrocytosis was elevated at 6 months and it subsequently declined with age in APPswe mice, consistent with findings of increased astrocytosis in the prodromal phase of AD followed by decreases in later stages in humans [
18]. The cross-sectional design of this study allowed a parallel in vivo/in vitro comparison of the regional and temporal distributions of Aβ deposition and astrocytosis in the same APPswe and wt mice at a given age interval. Further multitracer longitudinal PET imaging studies in AD transgenic animal models could provide additional insights into the temporal evolution of neuropathological changes during AD progression and could be useful for testing new therapies targeting astrocytes, especially at the earliest stages.
In conclusion, we provide in vivo evidence that astrocytosis occurs early in AD and precedes Aβ plaque deposition. The increasing recognition of heterogeneous and context-dependent astrocytosis in the progression of AD indicates that more research is needed to elucidate the functions of the different astrocyte populations in the brain at different stages of the disease.