Ambient RNA from CSF to capture brain cell‐type signatures in AD and other neurodegenerative diseases

Abstract

AbstractBackgroundWe leveraged digital deconvolution methods to determine cellular changes from brain transcriptomics of Alzheimer’s Disease(AD), and identified a change in cellular proportion between neuropathology, and severity of the disease as determined by Clinical Dementia Rating[5]. We expanded these approaches to Cerebral Spinal Fluid(CSF), to evaluate cellular changes associated with disease progresses.MethodWe used RNAseq data from published sources to test and refine a gene panel to deconvolve ambient RNA from CSF. Candidate marker genes were selected via literature search and verified using the brainseq [1] along with the human brain single‐cell gene expression [2]. We optimized gene panels using the leave‐one‐out cross‐validation method to predict proportions of neurons, astrocytes, microglia, oligodendrocytes, macrophages and endothelial cells, using bulk and ambient RNA sequenced in single‐cell transcriptomics.ResultWe leveraged our deconvolution to ascertain whether ambient RNA captured in single‐cell gene expression from the CSF captures the cellular population structure of the brain. Preliminary results show that cellular signatures for non‐immune cells are present in CSF in AD (Gate et al. (n = 9, Control = 3, MCI = 3, AD = 3)[3] and Multiple Sclerosis(MS)(Ramesh et al. (n = 15, Control = 3, MS = 12)[4]), and the proportions of those non‐immune cells change depending on the disease state. In MS, there is an increase in the proportion of Astrocytes (p = 3.4×10e‐2) in MS samples. Similar results were found for Neurons (p = 4.4×10e‐2) and Oligodendrocyte Precursor Cells (OPC) (p = 8.2×10e‐3). In AD, we identified significant differences between AD, MCI and controls (p = 0.01) in OPC.ConclusionThese results show that ambient RNA in the CSF can capture subtle differences in cellular population structure in AD and MS. These analyses were done on a small sample size, but show the potential to analyze larger cohorts and cell‐type specific transcriptional states. We also plan to expand the process to other neurologic diseases.