Molecular Atlas of the Adult Mouse Brain

Mapping the molecular diversity of brain regions and neuron subtypes is at the core of understanding the organization of brain circuits and thereby also their function. The recent development of methods that can define the transcriptome of single neurons with a high degree of accuracy has allowed the mapping of neuron subtypes. The spatial mapping of gene expression has so far relied on performing large-scale in situ hybridization (ISH) approaches.

Spatial transcriptomics (ST) is a method that combines high-resolution tissue imaging with high-throughput in-situ transcriptomics data. The ST method enables the study of the transcriptome at specific locations in the tissue and with the possibility to visualize the data in-situ. We have applied ST on 75 consecutive coronal sections from adult mice covering the entire brain. The sections have been aligned to a reference brain atlas, which has enabled us to create a 3D fully interactive molecular atlas of the mouse brain.

We performed unsupervised clustering of our atlas based entirely on gene expression patterns. This clustering enabled us to accurately map our gene expression data onto anatomical brain regions and also reveal the 3D organization of regions like the thalamus, hippocampus and amygdala. Our atlas allows the spatial mapping of brain subregions and identification of the spatially defining gene expression, and we find that spatial markers for specific regions can be confirmed by ISH annotations found in the Allen Brain Atlas. We have further applied machine learning to recover with a high degree of accuracy the spatial identity of single neurons using scRNA-seq datasets mapped onto our molecular atlas. In summary, this molecular atlas establishes an unbiased classification of the adult mousse brain and reveals the molecular profile underlying the brain subregions.