Amyloid-β and Tau
Aβ burdens were significantly variable between the 5 brain regions from the 9 individuals (Friedman test, p=0.0018). Specifically, we found that the hippocampus had the lowest amount of Aβ plaques compared to BA41/42 (p=0.0061), BA44/45 (p=0.0175), and BA46 (p=0.0365) but not BA17 (Dunn’s post-hoc test) (Fig. 1A and H). The deposition patterns between individuals varied considerably, with some showing very low Aβ burdens (SD017/16) and some showing extensive Aβ deposits throughout the grey matter (SD031/16). Strikingly, the participant with the highest Aβ burden (SD031/16) had an apolipoprotein (APOE) ε4 allele, a strong late-onset AD risk factor, and a Thal score 5, resembling AD-like pathology (Suppl. Tables 1 and 2). Moreover, cerebral amyloid angiopathy (CAA) with Aβ deposits around blood vessels was observed in some cases but was not always associated with Aβ plaques in the cortex (data not shown). Therefore, Aβ depositions are heterogeneous both between and within individuals.
As expected from the early Braak stages of the individuals, BA17 and BA46 showed almost no phosphorylated tau species, whereas BA41/42 and BA44/45 showed low NFT densities (Fig. 1B and I). In contrast to our Aβ data, the hippocampus most commonly exhibited tau pathology, and significantly higher levels of tau pathology than BA17 (p=0.0026) and BA46 (p=0.0224) (Dunn’s post-hoc test). The only individual (SD031/16) with tau spread in all 5 brain regions also had the highest NFT density across all regions. This was also the case with the highest amyloid burdens and an Apo ε4 allele. Again, these preliminary data show evidence of p-tau heterogeneity not only between brain areas but also between individuals.
Microglia and astrocytes
Microglia and astrocytes, similar to tau, were found at highest levels in the hippocampus. Iba1-positive microglia, representing total microglia numbers, showed a significant difference in burdens between brain areas (p=0.0017), with the hippocampus having a higher burden than BA17 (p=0.0006) (Dunn’s post-hoc test) (Fig. 1C and J). CD68-positive microglia, a marker of phagocytic activity, showed no significant differences between the 5 brain areas (p=0.135) (Fig. 1D and K). By exploiting the fact that Iba1 stains most microglia and CD68 only stains phagocytic microglia, we generated a ratio of phagocytic versus total microglia in all 5 regions. No statistical differences of activation status were observed between brains areas (p=0.216) (Fig. 1E).
Astrocyte burdens were significantly elevated in the hippocampus compared to both BA17 (p=0.0061) and BA41/42 (p=0.0287), but no further significant differences in burdens were found between other brain areas (Dunn’s post-hoc test) (Fig. 1F and L). Altogether, the hippocampus has statistically higher levels of glial cells than other brain areas, particularly BA17.
A significant difference in cortical thickness between cortical regions was observed (p=0.0133), with BA17 showing a significantly thinner cortex than BA41/42 (p=0.037) and BA46 (p=0.0209), but not BA44/45 (Dunn's post-hoc test) (Fig. 1G). These data show BA17 has a thinner cortex compared to more anterior areas. This observation is most likely explained by natural rostro-caudal differences in cortical thickness, rather than ageing-induced neuron loss.
In this study, we have measured Aβ, phosphorylated tau (NFT), microglia, and astrocyte levels in 5 brain regions and have demonstrated both regional and individual variability in 9 non-demented LBC1936 participants.
Our observed heterogeneity in tau and Aβ burdens heterogeneity reflects previous findings in the literature. Specifically, AD brains have higher levels of tau pathology in the hippocampus compared to cortical regions. Conversely, Aβ deposits are highest in the neocortex and moderately distributed in the hippocampus, until later stages of the disease. The greatest genetic risk factor for developing AD in non-familial cases is the possession of an APOE ε4 allele, whereas the ε2 allele appears to be protective against AD. Of note, the highest amyloid and tau pathology was observed in the only individual with an ε4 allele and the lowest amyloid pathology in the only individual with the ε2 allele (Suppl. Table 1). Nevertheless, these results have to be interpreted with caution due to the extremely low numbers involved and that the levels of Aβ and NFTs in this cohort are significantly lower than those found in AD cases. Overall, despite the small sample size, the data from this study have surprisingly, yet closely, depicted previously described features of the ageing brain.
It is currently unclear if or how Aβ and tau interact in the ageing and AD brain. It is more evident that the quantity and spread of NFTs correlates strongly with, not only, cognitive decline but also synapse loss and gliosis. Indeed, in our study, glial cells were predominantly found in the hippocampus in the LBC1936 individuals (similar to NFTs), confirming preceding findings. The increased glial coverage and p-tau in the hippocampus may indicate a causative role in the early hippocampal neurodegeneration during ageing and AD, and by extension it may explain the early memory impairments observed in the elderly and demented. Interestingly, other studies have shown that Iba1+ve microglia burdens are correlated with normal cognitive function, whereas activated (CD68) microglia are correlated with poorer cognitive function in dementia. While this study is currently too small to assess these kinds of associations, as more brains become available this cohort will provide an invaluable opportunity to discover associations between post-mortem brain changes and detailed longitudinal cognitive performance.
The hippocampus undergoes extensive spine remodelling and as a result, may require greater glial surveillance to ensure efficient temporal and spatial synaptic pruning. This could explain the higher numbers of glial cells we observe in the hippocampus compared to cortical regions (Fig. 1C). Synaptic health is critical for normal brain function, evidenced by the fact that synapse loss is the best correlate with cognitive impairment in AD. Synapse loss is also thought to correlate with poorer cognitive performance in normal ageing. During ageing, synaptic degeneration has been shown to impair electrophysiological properties of neurons by increasing the long-term depression (LTD) and reducing the long-term potentiation (LTP). Notably, Aβ and p-tau co-localize in synapses in AD brains, while microglia and astrocytes interact with synapses both in health and disease, marking synapses as critical points in normal and pathological ageing. Altogether, these age-related synaptic changes are hypothesised to occur before the onset of cognitive loss and to be key drivers of MCI and dementia. It would, therefore, be important to quantify synaptic puncta in the LBC1936 participants using high-resolution techniques such as array tomography to visualize how microglia, astrocytes, Aβ, and NFT’s interact with synapses. Furthermore, the data presented here must be compared to a younger cohort to establish if these brain area differences truly are age-associated phenomena, or region-specific throughout life.
In the future, a greater sample size will allow us to correlate longitudinal cognitive function scores to synapse integrity and tau pathology or gliosis, in order to understand how these factors may mediate age-related cognitive impairments. Furthermore, by following-up on these individuals' cognitive function, we expect to detect MCI in some individuals and through our detailed post-mortem analyses, begin to get an understanding of its neuropathological origin. By doing so, prodromal phases of AD can be detected early and described post-mortem to the single synapse level, which will greatly improve our understanding of AD progression and thus provide better avenues for treatments.