Spatially resolved cell atlas of the mouse primary motor cortex by MERFISH

Brian Zingg,Hongkui Zeng,Zizhen Yao,Xiaowei Zhuang,Hongwei Dong,Stephen W Eichhorn,Meng Zhang,Kaelan Cotter

doi:10.1038/s41586-021-03705-x

Abstract

A mammalian brain is composed of numerous cell types organized in an intricate manner to form functional neural circuits. Single-cell RNA sequencing allows systematic identification of cell types based on their gene expression profiles and has revealed many distinct cell populations in the brain1,2. Single-cell epigenomic profiling3,4 further provides information on gene-regulatory signatures of different cell types. Understanding how different cell types contribute to brain function, however, requires knowledge of their spatial organization and connectivity, which is not preserved in sequencing-based methods that involve cell dissociation. Here we used a single-cell transcriptome-imaging method, multiplexed error-robust fluorescence in situ hybridization (MERFISH)5, to generate a molecularly defined and spatially resolved cell atlas of the mouse primary motor cortex. We profiled approximately 300,000 cells in the mouse primary motor cortex and its adjacent areas, identified 95 neuronal and non-neuronal cell clusters, and revealed a complex spatial map in which not only excitatory but also most inhibitory neuronal clusters adopted laminar organizations. Intratelencephalic neurons formed a largely continuous gradient along the cortical depth axis, in which the gene expression of individual cells correlated with their cortical depths. Furthermore, we integrated MERFISH with retrograde labelling to probe projection targets of neurons of the mouse primary motor cortex and found that their cortical projections formed a complex network in which individual neuronal clusters project to multiple target regions and individual target regions receive inputs from multiple neuronal clusters.

Highlights

The mammalian cerebral cortex is a highly organized structure that supports sensory, motor and cognitive functions
By integrating multiplexed error-robust fluorescence in situ hybridization (MERFISH) with retrograde labelling, we further revealed the complexity of projection patterns of these molecularly defined cell types
We selected a panel of 258 genes for MERFISH imaging, including canonical marker genes for major neuronal and non-neuronal cell types in the cortex selected based on previous knowledge[12,13,21], as well as marker genes selected based on differential gene expression and mutual information[22] analyses of neuronal clusters identified by a companion single-cell and single-nucleus RNA sequencing

Summary

L6 CT 1 L6 CT 8 L6 CT 5

Our MERFISH data classified the IT cells into 20 clusters, 19 of which belonged to the L2/3 IT, L4/5 IT, L5 IT and L6 IT subclasses and the remaining one formed a distinct cell type, L6 IT Car[3] (Fig. 3a, Extended Data Fig. 7a–c). Using pseudotime analysis[39] to order the IT cells on the basis of their expression profiles, we observed that the pseudotime of cells was highly correlated with their cortical depths, and individual cells formed a largely continuous cloud along the pseudotime and cortical depth axes, with a more appreciable separation in pseudotime between L2/3 and L4/5 IT clusters (Fig. 3f, Extended Data Fig. 9c, d) Together, these results suggest that the IT neurons adopt a gradient distribution across the cortical depth, with correlated gene expression profiles and cortical depths of individual cells. We sought to integrate MERFISH with retrograde tracing to simultaneously determine the expression profiles and spatial organization of cell types in the MOp and their projection targets To this end, we injected retrograde tracers, cholera toxin subunit b (CTb) labelled with spectrally distinct dyes, into three cortical regions, the ipsilateral MOs, the SSp and the temporal association area (TEa), all of which receive direct inputs from the MOp40,41. Among the three L6 IT clusters, the MOs mostly received input from L6 IT 1, whereas TEa–ECT–PERI mostly received input from the L6 IT 3, despite the similar gene expression profiles and the substantially overlapping spatial distributions of these L6 IT clusters (Fig. 4e)

Discussion

Methods