Single nuclei RNA‐seq data analysis identifies glial cell lineages associated with Alzheimer’s disease severity

Abstract

AbstractBackgroundSingle nucleus RNA sequencing (snRNA‐seq) has the potential to improve our understanding of the cellular‐specific drivers of sporadic Alzheimer’s disease. In this study we re‐analyze published snRNA‐seq data from post‐mortem dorsolateral prefrontal cortex tissue from 48 participants in the Religious Order Study and Memory and Aging Project (ROS/MAP) cohorts (Mathys et al. 2019) and snRNA‐seq data from prefrontal cortex tissue from 9 participants across the Adult Changes in Thought (ACT), University of Washington Alzheimer’s Disease Research Center (UW‐ADRC), and Seattle Longitudinal Study (SLS) cohorts.MethodAll snRNA‐seq data were normalized using scran, with normalized counts pre‐processed in Monocle 3. Pseudotime lineages were learned separately for male and female patients for each cell subtype with Monocle3. Genes with non‐zero counts in fewer than 10 cells in a subtype were discarded.ResultWe reidentified the ROS/MAP disease‐associated microglia subtype (Mic1) from Mathys et al. 2019 in the ACT, UW‐ADRC, and SLS data as determined by Fisher’s exact test overlap of subtype specific gene expression markers (p.adjusted<10−16). Furthermore, we identify a trajectory of cellular state (pseudotime) within the Mic1 population that is disease associated (p.adjusted ≤ 3.4 × 10−11 in females, and p.adjusted ≤ 4.8 × 10−4 in males) in the Mathys et al. data. In female samples in both studies we observed that APOE expression has a positive association with Mic1 pseudotime, which we do not see in males. In males we see MEF2C expression was negatively associated with Mic1 pseudotime. Lastly, we see both activation of similar Mic1 pseudotime expression patterns in a subpopulation of oligodendroglial cells, as well as a loss of oligodendroglial cells in diseased patients.ConclusionWe provide increasing evidence implicating a specific microglial subpopulation in the etiology of Alzheimer’s disease. Furthermore, we see that this population has heterogeneity in expression of key Alzheimer’s disease risk genes as a function of sex, and that we can define genes that are specific to this microglial subtype lineage for further functional investigation. Finally, we also see evidence of a similar disease‐associated expression signature in oligodendroglial cells, suggesting potentially an interaction between the disease associated microglia populations and oligodendrocytes in diseased patients that warrants further functional investigation.