Transcriptomic cytoarchitecture reveals principles of human neocortex organization

Abstract

Variation in cortical cytoarchitecture is the basis for histology-based definition of cortical areas, such as Brodmann areas. Single cell transcriptomics enables higher-resolution characterization of cell types in human cortex, which we used to revisit the idea of the canonical cortical microcircuit and to understand functional areal specialization. Deeply sampled single nucleus RNA-sequencing of eight cortical areas spanning cortical structural variation showed highly consistent cellular makeup for 24 coarse cell subclasses. However, proportions of excitatory neuron subclasses varied strikingly, reflecting differences in intra- and extracortical connectivity across primary sensorimotor and association cortices. Astrocytes and oligodendrocytes also showed differences in laminar organization across areas. Primary visual cortex showed dramatically different organization, including major differences in the ratios of excitatory to inhibitory neurons, expansion of layer 4 excitatory neuron types and specialized inhibitory neurons. Finally, gene expression variation in conserved neuron subclasses predicts differences in synaptic function across areas. Together these results provide a refined cellular and molecular characterization of human cortical cytoarchitecture that reflects functional connectivity and predicts areal specialization.

A Meta-Atlas of the Developing Human Cortex Identifies Modules Driving Cell Subtype Specification

“…These efforts are facilitated by our meta-module activity metric, which enables the comparison of meta-modules from our developing meta-atlas with recently released atlas-scale datasets of the adult human brain 28 (Fig. S3A, Supplementary Table 10).…”

“… A) Meta-module 156 was identified in an analysis comparing the cell type distribution of meta-module activity across our developing neocortical meta-atlas and in a transcriptomic atlas of the adult human cortex 28 . The left UMAP displays the activity of meta-module 156 during development, highlighting its enrichment among progenitors.…”

Joint inference of discrete cell types and continuous type-specific variability in single-cell datasets with MMIDAS

“…The dataset comprises 42 transcriptional types, referred to as "supertypes." These supertypes were identified through an iterative projection of MTG data from young neurotypical reference donors 44 to itself, and by using scANVI 23 to predict the class label, as explained in. 27 In line with previous datasets, we performed a gene selection process to identify genes relevant to neuronal types.…”

Integrated multimodal cell atlas of Alzheimer’s disease

“…Over 100 cell types can be reliably identified using single nucleus RNA-seq in any cortical area, and alignment across species shows strong conservation of cellular architecture 30,31 that allows inference of human cellular properties from studies in experimentally tractable mouse and non-human primate models. Systematic BICCN and BICAN efforts are now producing the first brain-wide cell atlases of the mouse 28,32,33 and human brain [25][26][27]34,35 , providing robust and highly curated genomically-based reference cell classifications, spatial maps of cellular distributions, and characterization of cellular properties in normal brain. These reference classifications provide an extremely powerful foundational reference, akin to the human genome, to understand the cellular, molecular, and epigenomic underpinnings of Alzheimer's disease.…”