A landscape of the genetic and cellular heterogeneity in Alzheimer disease

Abstract

BackgroundAlzheimer disease (AD) has substantial genetic, molecular, and cellular heterogeneity associated with its etiology. Much of our current understanding of the main AD molecular events associated with the amyloid hypothesis (APP, PSEN1 and PSEN2) and neuroimmune modulation (TREM2 and MS4A) is based on genetic studies including GWAS. However, the functional genes, downstream transcriptional ramifications, and the cell-type-specific effects of many GWAS loci remain poorly understood. Understanding these effects can point us to the cellular processes involved in AD and uncover potential therapeutic targets.MethodsWe applied a genetic-based approach to our sample selection; our cohort included carriers of AD pathogenic mutations (APP, PSEN1), risk variants in TREM2, and the resilience variant (rs1582763) in the MS4A cluster associated with cerebrospinal fluid (CSF) soluble TREM2 levels. We performed single-nucleus RNA-sequencing (snRNA-seq) of 1,102,459 nuclei from the human parietal cortex of these carriers. Following initial unbiased clustering and cell-type annotation, we performed deep subclustering analysis per cell type to identify unique cellular transcriptional states associated with these genetic variants. We identified differentially expressed genes between cell states and genetic variant carriers/controls, and performed differential cell proportion analyses to determine key differences among these carriers. We analyzed sequencing data from human dorsolateral prefrontal cortex and mouse models to replicate the enrichment of unique cell states in genetic variant carriers. Finally, we leveraged these cell-state differential expression results to link genes in AD GWAS loci to their functional cell types.FindingsWe identified cell-specific expression states influenced by AD genetic factors for neurons and glia. Autosomal dominant AD (ADAD) brains exhibited unique transcriptional states in all cell types. TREM2 variant carrier brains were also enriched for specific microglia and oligodendrocyte subpopulations. Carriers of the resilience MS4A variant were enriched for an altered activated-microglia expression state. We mapped AD GWAS genes to their potential functional cell types, and some, including PLCG2 and SORL1, were expressed in a broader range of brain cell types than previously reported.InterpretationAD pathogenic, risk, or resilience variants are sufficient to alter the transcriptional and cellular landscape of human brains. Overall, our results suggest that the genetic architecture contributes to the cortical cellular heterogeneity associated with disease status, which is a critical factor to consider when designing drug trials and selecting the treatment program for AD patients.Our findings suggest that integrating genetic and single-cell molecular data facilitates our understanding of the heterogeneity of pathways, biological processes and cell types modulated by genetic risk factors for AD.FundingUS National Institutes of Health, Hope Center, Archer foundation, Alzheimer Association, CZI.

Higher levels of myelin are associated with higher resistance against tau pathology in Alzheimer’s disease

“…Together, these results suggest the possibility that myelin-triggered Fyn kinase may play a role in the formation of tau tangles in AD. Myelin damage has been observed previously in histochemical and neuroimaging studies in AD [38][39][40][41] , and recent single cell studies provide converging results showing that alterations in oligodendrocytes, a cell-type involved in myelination, are prominent in AD 42,43 . Recent studies in transgenic mouse models of amyloid pathology suggest that acute demyelination may occur upstream of amyloid deposition by interfering with a TREM2-related microglia activation signature, thus leading to enhanced amyloid deposition 44 .…”

Phospholipase D3 contributes to Alzheimer’s disease risk via disruption of Aβ clearance and microglia response to amyloid plaques

“…Given the association of Pld3 loss with altered microglia function in mouse models, we sought to determine whether PLD3 is altered in microglia in human brains. Nuclei were isolated and sequenced from frozen AD and age-matched control brains (Figure 6A)[17, 18]. AD brains were further classified based on the presence of TREM2 risk variants (named: TREM2).…”

“…AD brains were further classified based on the presence of TREM2 risk variants (named: TREM2). Unsupervised clustering of the brain nuclei revealed 15 cell-type specific clusters that correspond to the major cell-types found in the brain [17]. We isolated microglia from other cells and reexamined the alternative transcriptional states that we further classified into nine subclusters (Figure 6B).…”

“…Human parietal cortices were processed to isolate nuclei, and the nuclei were then sequenced using the 10X Chromium single cell Reagent Kit v3, with 10,000 cells per sample and 50,000 reads per cell for each of the 74 samples as previously described [17, 18]. The CellRanger (v3.0.2 10XGenomics) software was used to align the sequences and quantify gene expression.…”

“…To determine if there was unique functionality or potentially altered expression levels associated with disease/genetic carriers in the alternative cell states, we employed linear mixed model that predicted the expression level of each gene, modeled as zero-inflated negative binomial distributions and corrected for sex and age of death. Donors were modeled as random variables as previously described [17]. Data can be publicly accessed at http://ngi.pub/SNARE.…”