Characterizing Disrupted Cellular Crosstalk Signaling Networks in Alzheimer’s Disease Using Single‐Nuclei Transcriptomics

Abstract

AbstractBackgroundGenetics and molecular studies have implicated multiple genes, pathways, and cell types in Alzheimer’s disease (AD) risk and progression. How these genes and pathways perturb and are modulated by brain cellular crosstalk networks have not been deeply characterized.MethodWe generated single‐nuclei transcriptomic profiles (snRNA‐seq) of parietal lobes from 67 donors from the Knight ADRC and DIAN brain banks, representing neuropathological‐free controls and AD cases from pre‐symptomatic and AD[1]. We estimated cellular crosstalk patterns among the brain cell types based on the expression of known ligand‐receptor pairs[2] and reconstructed the co‐expression network upstream and downstream of these ligand‐receptor pairs to determine how these signals are propagated across cells[3].ResultWe analyzed ∼294K high‐quality nuclei and identified six major cell populations[1]. We observed changes in cellular crosstalk patterns between controls and AD, with the largest involving microglial interactions (increased in AD, OR = 1.31, p = 8.94e‐11). Cellular interactions directly involving AD‐related genes[4] as either the receptor or the ligand were enriched for neuron‐microglia pairs (OR = 2.74, p = 4.41e‐15), and the majority (64.9%) codified for microglial cell membrane receptors, supporting the role for these cells in AD. We observed an increase in the frequency of a subset of interactions involving AD‐related genes in microglia when comparing pre‐symptomatic and AD individuals, suggesting correlation with pathological burden. These included TREM2‐semaphorin (neuron‐microglia, 4.38‐fold increase). We performed a comprehensive study of the microglial co‐expression networks up/downstream of these interactions and provided additional evidence of their association with Braak stage. The TREM2‐semaphorin neuron‐microglia interaction is predicted to directly propagate signals into a sub‐network associated with microglia activation, involving four genes from the HLA family previously implicated in AD. Furthermore, the neuronal co‐expression network had significant enrichment for AD‐related genes including CELF2 (OR = 2.10, p = 5.32e‐3), consistent with AD‐related genes in neurons propagating signals into microglia. We replicated these results in public snRNA‐seq studies and are currently validating these findings experimentally.ConclusionThis study reveals the role of cellular crosstalk in AD biology and identifies disruptions in neuron‐microglia interactions as an important component of AD pathology. In addition, we show that cellular crosstalk networks can directly modulate genes previously associated with AD (Figure 1).