Single-cell and nucleus RNA-seq in a mouse model of AD reveal activation of distinct glial subpopulations in the presence of plaques and tangles

Abstract

Multiple mouse models have been generated that strive to recapitulate human Alzheimer′s disease (AD) pathological features to investigate disease mechanisms and potential treatments. The 3xTg-AD mouse presents the two major hallmarks of AD, which are plaques and tangles that increase during aging. While behavioral changes and the accumulation of plaques and tangles have been well described in the 3xTg-AD mice, the subpopulations of neurons and glial cells present throughout disease progression have not been characterized. Here, we used single-cell RNA-seq to investigate changes in subpopulations of microglia, and single-nucleus RNA-seq to explore subpopulations of neurons, astrocytes, and oligodendrocytes in the hippocampus and cortex of aging 3xTg-AD as well as 5xFAD mice for comparison. We recovered a common path of age-associated astrocyte activation between the 3xTg-AD and the 5xFAD models and found that 3xTg-AD-derived astrocytes seem to be less activated. We identified multiple subtypes of microglia, including a subpopulation with a distinct transcription factor expression profile that showed an early increase in Csf1 expression before the switch to disease associated microglia (DAM). We used bulk RNA-seq in the hippocampus of 3xTg-AD mice across their lifespan to identify distinct modules of genes whose expression increases with aging and worsening pathology. Finally, scATAC-seq revealed multiple subpopulations of cells with accessible chromatin in regions around genes associated with glial activation. Overall, differences between the main glial groups point to a slower activation process in the 3xTg-AD model when compared to the 5xFAD. Our study contributes to the identification of progressive transcriptional changes of glial cells in a mouse model that has plaques and tangles, thus providing information to aid in targeted AD therapeutics that could translate into positive clinical outcomes.

Systematic Phenotyping and Characterization of the 3xTg-AD Mouse Model of Alzheimer’s Disease

“…Bulk tissue gene expression was measured via RNA-seq, from micro-dissected hippocampi, and can be explored in an interactive and searchable fashion at https://modeladexplorer.org/ . Comparisons between bulk tissue and single cell/single nucleus RNA-seq from 3xTg-AD brains are explored extensively in a companion publication ( Balderrama-Gutierrez et al, 2021 ).…”

“…We explored Thy1 promotor expression, which drives the transgenes in both 3xTg-AD and 5xFAD mice [data derived from Balderrama-Gutierrez et al (2021) ] and ( Forner et al, 2021 , respectively. Thy1 expression was elevated in 3xTg-AD cortex relative to wild-type Thy1 (which does not drive a transgene but sets the background level) only at 18 months of age, consistent with the lack of pathology that we observe there ( Figure 9F ).…”

MAC2 is a long‐lasting marker of peripheral cell infiltrates into the mouse CNS after bone marrow transplantation and coronavirus infection

“…However, a recent study has shown that enzymatic dissociation methods can induce aberrant gene expression signatures in microglia (Marsh et al, 2020), indicating that some scRNAseq findings may be confounded and an artifact of isolation methods, analysis, or sequencing platform. A recent investigation utilizing scRNAseq, snRNAseq and scATACseq identified distinct microglia subpopulations in 3xTgAD and 5xFAD mice, including a DAM cluster containing Cst7 , but lacking Lgals3 ; instead, Lgals3 was found in a small cluster containing Cd14 and Itgal (Balderrama‐Gutierrez et al, 2021). Galectin‐3 was also shown to be significantly elevated in human AD patients, exhibit preferential expression in microglia in contact with plaques, and serve as a ligand for TREM2 (Boza‐Serrano et al, 2019).…”

A comparison of marker gene selection methods for single-cell RNA sequencing data

“…We also do not consider the ability of selected marker genes to be used in designing spatial transcriptomics analyses [7,25] . Finally, we only directly consider the quality of the marker genes in their use for annotation, rather than for other purposes, such as to gauge the quality of a clustering [8] or to provide markers that can be used in wet-lab applications [6] . We believe, however, that methods that select high quality marker genes for annotation will likely also perform well at these other tasks.…”

“…Annotation of the cell-type of clusters is critical both to guide the clustering and to interpret the results of downstream analyses performed with respect to the clustering, such as differential expression testing [4] or single-cell eQTL mapping [5] . In addition to cluster annotation, some authors treat the marker genes as new biological discoveries in their own right [6] , as candidates for further perturbation or differential analyses [7] , or as evidence for the quality of a particular clustering [8] . As well as their direct use, marker genes are a common component of computational approaches that aim to annotate clusters automatically [9,10] .…”