Transcriptomic architecture of nuclei in the marmoset CNS

Abstract

To understand the cellular composition and region-specific specialization of white matter — a disease-relevant, glia-rich tissue highly expanded in primates relative to rodents — we profiled transcriptomes of ~500,000 nuclei from 19 tissue types of the central nervous system of healthy common marmoset and mapped 87 subclusters spatially onto a 3D MRI atlas. We performed cross-species comparison, explored regulatory pathways, modeled regional intercellular communication, and surveyed cellular determinants of neurological disorders. Here, we analyze this resource and find strong spatial segregation of microglia, oligodendrocyte progenitor cells, and astrocytes. White matter glia are diverse, enriched with genes involved in stimulus-response and biomolecule modification, and predicted to interact with other resident cells more extensively than their gray matter counterparts. Conversely, gray matter glia preserve the expression of neural tube patterning genes into adulthood and share six transcription factors that restrict transcriptome complexity. A companion Callithrix jacchus Primate Cell Atlas (CjPCA) is available through https://cjpca.ninds.nih.gov.

A 4D transcriptomic map for the evolution of multiple sclerosis-like lesions in the marmoset brain

“…Before the day of tissue harvest, a custom-made brain holder was generated for each marmoset brain by 3D printing (Ultimaker 2 + ) to guide tissue sampling (139). Tissue dissection was carried out as described (42). Briefly, marmosets were deeply anesthetized with 5% isoflurane until no signs of breathing, then were transcardially perfused with ice-cold artificial cerebrospinal fluid (aCSF) for 5 min.…”

“…Nuclei preparation was carried out as described (42). Briefly, RNAlater preserved samples were thawed on ice, retrieved from the storage tube with a pair of clean forceps, dabbed with Kimwipes to remove residual RNAlater, and placed in a 1 mL douncer tube (Dounce Tissue Grinder, 357538, Wheaton).…”

“…Alignment. The raw snRNA-seq reads were aligned to a marmoset genome assembly, ASM275486v1 (GCA_002754865.1), with a reference package built as described in (42). CellRanger (version 3.1.0, 10x Genomics) software was used to align reads for snRNA-seq samples to generate cell barcode-to-gene feature matrix for downstream analysis, and automatic estimation of cell number was applied for most of the snRNA-seq samples unless otherwise specified ( TableS3 ).…”

“…Preprocessing and quality control . Preprocessing and quality control parameters were applied as described in (42). Briefly, Seurat v3 object was created for individual samples and DoubletFinder (142) was applied to estimate and remove putative doublets to mitigate technical confounding artifacts.…”

“…In parallel, we included leukocortical T 2 -hyperintense lesions (n = 2) along with matching healthy (n = 2) and NA (n = 1) controls, and nearby abnormal-looking lateral geniculate nucleus (LGN) tissue on MRI (n = 2) along with matching healthy controls (n = 2), to explore tissue-specific or shared responses. We implemented a hierarchical workflow comparing across cell classes (Level 1, L1) and subclusters within a class (Level 2, L2) to better realize the importance of each signaling change ( Methods , (42)). Overall, we found a remarkable expansion in the number and diversity of glial and immune cells as lesions develop ( Fig1E ).…”