Aggregation of Aβ into higher-order structures such as extracellular amyloid plaques is a necessary event in AD pathogenesis. A fundamental feature of AD is the preferential deposition of amyloid plaques in distinct brain regions. However, the mechanism by which specific brain regions are rendered vulnerable to amyloid deposition in AD has heretofore remained unknown. As Aβ aggregation is closely related to ISF Aβ concentration in vivo, elucidating the mechanisms that regulate ISF Aβ levels has important implications for understanding the etiology of region-specific Aβ deposition in AD. In the present work, we utilize in vivo microdialysis to show that ISF Aβ levels measured in several brain regions of young Tg2576 mice months prior to plaque deposition are directly proportional to the degree of subsequent plaque deposition. Moreover, ISF Aβ levels in young mice are closely associated with the level of baseline ISF lactate, a marker of neuronal activity, in a region-specific manner. Furthermore, we show that physiological changes in neuronal activity are sufficient to dynamically modulate ISF Aβ levels and amyloid plaque growth in vivo. Collectively, our results suggest that regional differences in basal neuronal activity level govern region-specific amyloid deposition through long-term regulation of steady-state ISF Aβ concentration.
Since sequential proteolysis of APP by β- and γ-secretase is necessary for Aβ generation, several previous reports have examined the regional distribution of this triad as potential mediators of region-specific Aβ deposition. However, the cerebral distribution of APP, BACE1 and PS1 transcripts is not associated with the topology of Aβ deposition in human AD31–34
or APP transgenic mice35–37
. Moreover, the regional distribution of γ-secretase activity in APP transgenic mouse brain does not correlate with the pattern of Aβ deposition38
. Consistent with these previous reports, the present results suggest that though differences in APP expression and processing are present across a subset of brain areas in Tg2576 mice, the pattern of APP expression and processing is not sufficient to account for the pattern of Aβ deposition. Together, these studies suggest that additional factors are necessary to drive region-specific Aβ deposition. In addition to region-specific differences in neuronal activity potentially contributing to region-specific differences in Aβ deposition, other possible contributors to the presence of Aβ in specific neural networks include the spreading of Aβ aggregates within and into the brain. For instance, exogenously applied Aβ aggregates can seed cerebral β-amyloidosis via intracerebral3, 39
application. Further, reports suggest that selective overexpression of APP in the cell bodies of neurons in the entorhinal cortex of transgenic mice can produce elevated soluble Aβ levels and amyloid plaque deposition in the synaptic terminal zone of the dentate gyrus41, 42
. Thus, transsynaptic propagation mechanisms as well as spread of Aβ pathology through interconnected neural networks might also contribute to the pattern of amyloid deposition observed in AD.
In healthy human brain, a specific subset of brain regions – the default network – is preferentially engaged during resting state conditions. Intriguingly, amyloid deposition in AD brain is most prominent in regions that comprise the default network. These findings raise the possibility that patterns of normal cerebral metabolism may influence the distribution of amyloid deposition in AD. In accord with this hypothesis, recent data suggests that the spatial distribution of resting state aerobic glycolysis (i.e. glucose utilization in excess of that used for oxidative phosphorylation despite sufficient oxygen to oxidize glucose to carbon dioxide and water) in cognitively normal adults overlaps closely with the topology of amyloid deposition in AD. Our present data demonstrating that pharmacological manipulation of neuronal activity directly regulates ISF lactate levels are consistent with previous animal and human studies and suggest that neuronal activity stimulates aerobic glycolysis in vivo. Given that neuronal activity also increases Aβ production and secretion into the ISF, regional differences in neuronal activity may underlie the spatial relationship between resting state aerobic glycolysis and amyloid deposition in AD.
We have previously demonstrated that ISF Aβ levels exhibit diurnal fluctuation - ISF Aβ concentration is significantly greater during waking conditions compared to sleep26
. Given that neuronal activity directly regulates ISF Aβ levels, we reasoned that diurnal fluctuation of neuronal activity may contribute to diurnal fluctuation of ISF Aβ. Indeed, previous data suggest that periods of wakefulness are associated with net synaptic potentiation and periods of sleep are associated with synaptic downscaling and depression28
. Here we show that ISF lactate concentration exhibits diurnal fluctuation and is closely associated with the diurnal fluctuation of ISF Aβ. These results are consistent with the notion that fluctuation of neuronal activity levels during the sleep/wake cycle contributes to the diurnal fluctuation of ISF Aβ levels in vivo. This observation further supports the hypothesis that physiological neuronal activity dynamically regulates ISF Aβ levels in vivo.
More generally, our results suggest that factors that elevate endogenous neuronal activity over prolonged periods may accelerate progression of Aβ deposition. In support of this hypothesis, recent fMRI data suggests that cognitively normal APOE
ε4 carriers exhibit elevated resting state activity in the default network and increased hippocampal activation during a memory encoding task compared to non-carriers43
. Thus, in addition to apoE isoform-dependent differences in apoE-Aβ interactions44
, elevated lifetime neuronal activity may be a complementary mechanism by which ε4 expression increases AD risk. Conversely, factors that reduce activity in vulnerable neural networks may slow the progression of β-amyloidosis and reduce AD risk. For example, epidemiological evidence suggests that educational attainment is negatively correlated with AD risk45
. As default network activity is suppressed during cognitively demanding tasks, one possibility is that education reduces AD risk by reducing neuronal activity and Aβ generation within the default network.
It is noteworthy that the regional distribution of lesions in mouse models of glutamate-induced excitotoxicity is strikingly similar to the distribution of Aβ deposition described herein46
. As neuronal activity regulates both Aβ and glutamate release at the synapse, the presently described regional differences in basal neuronal activity may also underlie the distribution of glutamate-induced excitotoxic lesions. A more detailed understanding of the regional distribution of pre- and post-synaptic elements may provide insight regarding the structural basis of neuronal activity patterns in normal brain and how patterns of neuronal/synaptic activity might contribute to a variety of neurological lesions.