Modeling a Cellular Cascade of Alzheimer's Disease
Alzheimer's disease is complex and puzzling, and massively funded, high-profile efforts to find treatments for the condition have been failing for decades. The research community has focused on clearance of amyloid-β, as this protein accumulates and misfolds in Alzheimer's patients. Yet some old individuals exhibit high levels of amyloid-β and do not suffer Alzheimer's, while clearance of extracellular amyloid-β fails to meaningfully improve the condition of patients. It may be that intracellular amyloid-β is the real target, or that amyloid-β accumulation is a side-effect of the real pathological mechanisms.
Of late, more attention is being given to overly active or senescent glial cells in the brain and their contribution to rising levels of inflammation. Chronic inflammation may well turn out to be the most important mechanism in Alzheimer's disease, and thus senolytic therapies to clear senescent cells and their inflammatory secretions may turn out to be quite effective as a treatment. We'll find out whether this is the case in the years ahead.
Today's open access paper delves into post-mortem human brain tissue in order to model the cascade of changing glial cell population characteristics. The data is supportive of a focus on glial cells and their contribution to inflammation. That activation of the immune system may be the cause of increased amounts of amyloid-β in its role as an antimicrobial peptide, a part of the innate immune response. At the end of the day, the only really compelling data in the context of Alzheimer's disease is a narrowly focused treatment that produces a reversal of pathology: that would settle the debate over which of the many possibilities is the most important pathological mechanism.
Alzheimer's Disease (AD) is a progressive neurodegenerative disease seen with advancing age. Recent studies have revealed diverse AD-associated cell states, yet when and how they impact the causal chain leading to AD remains unknown. To reconstruct the dynamics of the brain's cellular environment along the disease cascade and to distinguish between AD and aging effects, we built a comprehensive cell atlas of the aged prefrontal cortex from 1.64 million single-nucleus RNA-seq profiles. We associated glial, vascular, and neuronal subpopulations with AD-related traits for 424 aging individuals, and aligned them along the disease cascade using causal modeling. We found two predicted trajectories in the cellular landscape, termed (a) progression of AD (prAD) and (b) Alternative Brain Aging (ABA).
At the subpopulation level, microglial nuclei profiles were partitioned into 16 subpopulations, including proliferative (Mic.1), surveilling (Mic.2-5; expressing CX3CR1), reacting (Mic.6-8; TMEM163), enhanced-redox (Mic.9-10; FLT1), stress response (Mic.11; NLRP1, TGFBR1, upregulating genes of heat response, cellular senescence and NLRP1 inflammasome), interferon response (Mic.14, IFI6), inflammatory (Mic.15; CCL3/CCL4, NFKB1, NLRP3), SERPINE1 expressing (Mic.16) and lipid-associated (Mic.12-13; APOE) subpopulations. The lipid-associated Mic.12 and Mic.13 both expressed the AD risk genes APOE and GPNMB, with Mic.13 also expressing high levels of SPP1 and TREM2 compared to other subpopulations.
Astrocytes were partitioned into 10 subpopulations - homeostatic-like (Ast.1-2), enhanced-mitophagy (Ast.3; PINK1), reactive-like Ast.4 (GFAP, ID3) and Ast.5 (GFAP, SERPINA3, OSMR), interferon-responding (Ast.7; IFI6), and stress response (Ast.8-10): Ast.8, expressing heat stress and DNA damage, calcium, and sterol metabolism genes; Ast.9, expressing heat and oxidative stress response, tau binding and necroptosis genes; and Ast.10 (SLC38A2), expressing oxidative stress and ROS, metallothioneins and zinc ion homeostasis genes.
Oligodendrocyte lineage cells were partitioned into 13 subpopulations of mature oligodendrocytes (Oli.1-13), such as the stress response Oli.13 (SLC38A2), 3 subpopulations of oligodendrocyte precursor cells (OPC.1-3), one committed oligodendrocyte precursor (COP) subpopulation and one newly formed oligodendrocytes (NFOL) subpopulation. The newly discovered diversity of OPCs are of particular interest, and included an enhanced-mitophagy subpopulation (OPC.1; PINK1), which had higher expression of AD risk genes (e.g. APOE, CLU), and an axon projection/regeneration associated subpopulation (OPC.3; SERPINA3, OSMR).
Specifically, we suggest the following sequence of events underlying the prAD trajectory: At the early stages, selective homeostatic glial subpopulations decrease in proportion alongside an increase of, first, the lipid-associated microglia subpopulation APOE+ Mic.12 subpopulation that is itself influenced by advancing age and contributes to the accumulation of amyloid-β proteinopathy, and up-regulates immune activation pathways. Then, a distinct but related Mic.13 subpopulation of APOE+TREM2+ microglia that are influenced by APOEε4 (the strongest genetic risk factor for AD) contributes to the subsequent accumulation of tau proteinopathy.
At the next stage of the AD cascade, with the accumulation of tau proteinopathy, we observed a transient increase in the proportions of Ast.3 and OPC.1, which both upregulate genes associated with high energy demand and enhanced-mitophagy, as well as oxidative phosphorylation and glutamate secretion. OPC.1 further upregulates genes associated with response to oxidative stress aligning with reports suggesting the increased vulnerability of OPCs to oxidative stressors that are rising during this phase.
At the last stage of the AD cascade, we observed further increase in Mic.12 and Mic.13 proportions, together with a coordinated increase of Ast.10 and Oli.13, with Ast.10 playing an important role mediating the effect of tau proteinopathy on the increased rate of cognitive decline. Both Ast.10 and Oli.13 express stress response genes, with Ast.10 mainly demonstrating response to oxidative stress while Oli.13 showing response to heat stress and unfolded protein. Cognitive decline appears to be directly affected by Ast.10 that is driven by both tau and Mic.13, suggesting that the proportion of this astrocyte subpopulation may be a point of convergence for different processes leading to cognitive dysfunction.
While Mic.12 offers a good target with which to perturb the accumulation of Aβ proteinopathy to enhance therapeutic options centered on anti-amyloid-β antibodies, preventing polarization of microglia and astrocytes into Mic.13 and Ast.10 respectively, may have more immediate impact in helping to prevent cognitive impairment. The latter strategy would also be better suited for individuals who are already Aβ+ and are at risk of tauopathy.
On the other hand, these subpopulations do not appear to be relevant to the alternative brain aging (ABA) trajectory, where we found a selective decrease in homeostatic glial subpopulations and an increase in reactive microglial subpopulations. At the next stage, we observed increased proportions of reactive-like Ast.5 and OPC.3, both expressing the markers SERPINA3 and OSMR. Among participants in ABA, we found constant levels of neocortical amyloid, very limited neocortical tau and varying dynamics of cognitive decline. Thus, more work is needed to better understand this trajectory and its interesting, defining glial subpopulations.