Amyloid plaques, primarily composed of aggregated Aβ, exist in the extracellular space of the brain and are a pathological hallmark of AD. However, the factors that govern the formation and growth of plaques in the living brain are unknown. In the present study, we utilized serial in vivo
multiphoton microscopy in APP/PS1 mice to directly quantify amyloid plaque formation and growth in vivo
. We found that though many plaques remained stable in size at both ages examined, plaques imaged in 6 month-old APP/PS1 mice exhibited robust growth relative to plaques found in 10 month-old mice, suggesting that plaque growth is more prominent early in disease pathogenesis. This finding is consistent with reports of biphasic development of plaque load in APP transgenic mice; compact plaque load is reported to increase with age (Sturchler-Pierrat et al., 1997
; Wengenack et al., 2000
; Jack et al., 2005
; Braakman et al., 2006
; Harigaya et al., 2006
) before stabilizing at later disease stages (Gordon et al., 2002
). Plaque load stabilization has also been hypothesized to occur in human AD patients, as amyloid burden does not correlate with disease duration (Hyman et al., 1993
; Engler et al., 2006
) and plaque burden in a subset of patients with mild cognitive impairment is indistinguishable from AD patients (Lopresti et al., 2005
; Price et al., 2005
; Mintun et al., 2006
). Our study also indicated that plaque growth was related to plaque size, as smaller plaques exhibited greater rates of growth compared to larger plaques, regardless of age.
We also found that treatment with the γ-secretase inhibitor, Compound E, markedly decreased the appearance of new plaques and growth of pre-existing plaques in APP/PS1 mice. This suppression in plaque appearance and growth was reflected in a decrease in total plaque burden in parallel cross-sectional studies. To determine the extent to which Compound E treatment reduced soluble extracellular Aβ levels in vivo
, we utilized in vivo
microdialysis to measure ISF Aβ levels in APP/PS1 treated with Compound E. Chronic dosing of Compound E over a 7-day period decreased ISF Aβx-40
levels by only 20-25% over a 24-hour period. That reduced extracellular Aβ concentration is associated with inhibition of amyloid plaque growth but not plaque regression is consistent with previous data obtained using transgenic mice that overexpress mutant APP under the regulation of a tetracycline-responsive promoter. Inhibition of mutant APP expression for 6 months after plaque formation arrested the progression of amyloid pathology but did not reduce overall plaque burden (Jankowsky et al., 2005
). A more recent report demonstrated that 3 weeks of treatment with an orally-active γ-secretase inhibitor did not reduce size of existing plaques in APP/PS1 mice (Garcia-Alloza et al., 2009
). Together, these studies suggest that a kinetic disequilibrium between Aβ plaque aggregation and dissociation may exist in vivo
The hypothesis that soluble extracellular Aβ concentration is a key determinant of Aβ aggregation in vivo
is supported by data demonstrating that areas of the brain that ultimately develop plaque pathology have higher basal ISF Aβ levels early in life relative to brain regions that do not develop pathology (Cirrito et al., 2003
). Moreover, intracerebral injection of Aβ-containing brain extract from human AD patients or APP transgenic mice induces cerebral amyloidosis in APP transgenic mice in a concentration-dependent manner (Meyer-Luehmann et al., 2006
). Though correlative, the present study is the first demonstration of the relationship between soluble ISF Aβ and amyloid plaque growth in vivo
. Moreover, it is the first report that this relationship can be modulated with pharmacological intervention. That a modest reduction of ISF Aβ levels is associated with inhibition of amyloid plaque growth and attenuation of new plaque formation is consistent with recent data demonstrating that a 30% reduction in γ-secretase activity, as seen with knocking out γ-secretase components throughout life, can attenuate plaque burden in a mouse model of AD (Li et al., 2007
). Thus, partial γ-secretase inhibition may be sufficient to arrest amyloid plaque progression in AD.
A number of potential therapeutic implications follow from the results of the present study. First, the observation that γ-secretase inhibition can prevent growth of existing plaques and attenuate new plaque formation, but not induce plaque regression suggests that anti-Aβ treatments may be most efficacious if administered early in disease pathogenesis. This hypothesis is supported by our cross-sectional study, showing that 28-day Compound E treatment reduced amyloid plaque load in 6 month-old APP/PS1 mice, but not in 10 month-old mice. Moreover, our finding that a modest reduction in soluble, extracellular Aβ is associated with a dramatic reduction in plaque growth and formation may have important implications for drug dosing and pharmacodynamic effects of anti-Aβ therapeutics to be used in clinical trials.
Our present observations of individual plaque growth over a period of weeks are in contrast to a recent report which showed that plaques reach a mature size within 24 hours after appearance (Meyer-Luehmann et al., 2008
). We hypothesized that these differences might be due to different techniques used to create transcranial windows. We used a small (<0.8 mm diameter) closed thinned-skull window in our experiments, in contrast to the larger (6 mm diameter) open-skull (craniotomy) window used by Meyer-Luehmann et al
(see supplemental Fig. 1A, B
). A direct comparison of plaque growth using these two techniques revealed robust growth under thinned-skull window preparations, but no significant growth under open-skull preparations. In addition, the open-skull preparation was associated with extensive cortical microglial and astrocytic activation, which was largely absent under the thinned-skull window. These findings are consistent with a previous comparison of these two techniques, which demonstrated that the open-skull window resulted in significant reactive gliosis and an alteration of dendritic spine dynamics compared to the closed thinned-skull window (Xu et al., 2007
). Coupled with previous reports implicating reactive gliosis in plaque size maintenance and regression (Gordon et al., 2002
; Wyss-Coray et al., 2003
; El Khoury et al., 2007
; Takata et al., 2007
; Bolmont et al., 2008
), this may explain the discrepancy between the findings using the two different techniques.
In summary, our results suggest that individual amyloid plaques grow over a period of weeks and that the rate of plaque growth is related to disease stage, plaque size, gliosis, and soluble extracellular Aβ concentration. Thus, growth of individual plaques may be a fundamental mechanism by which plaque load increases in AD. Furthermore, the present results suggest that a decrease in ISF Aβ levels by as little as 20-25% at key time points in plaque development may be sufficient to prevent the progression of amyloid pathology.