A Cell Atlas for the Mouse Brain

Abstract

Despite vast numbers of studies of stained cells in the mouse brain, no current brain atlas provides region-by-region neuron counts. In fact, neuron numbers are only available for about 4% of brain of regions and estimates often vary by as much as 3-fold. Here we provide a first 3D cell atlas for the whole mouse brain, showing cell positions constructed algorithmically from whole brain Nissl and gene expression stains, and compared against values from the literature. The atlas provides the densities and positions of all excitatory and inhibitory neurons, astrocytes, oligodendrocytes, and microglia in each of the 737 brain regions defined in the AMBA. The atlas is dynamic, allowing comparison with previously reported numbers, addition of cell types, and improvement of estimates as new data is integrated. The atlas also provides insights into cellular organization only possible at this whole brain scale, and is publicly available.

Matrix Inversion and Subset Selection (MISS): A novel pipeline for mapping of diverse cell types across the murine brain

“…Next we examine the quantitative concordance of MISS against literature-derived cell counts, using studies listed in Table 1, with the addition of total neurons 22 and total cells 3 -see scatter plot in Figure 2e. MISS achieved excellent overall concordance between the predicted cell counts per region and those measured by single-cell counting methods (Rc=0.69), and compares favorably to a previous computational approach to derive cell counts 5 . Additionally, the similarity between R and Rc shows that there is no global bias in our methodology.…”

“…MISS stands out from existing approaches for quantifying cell types because it bypasses the cell type specificity/spatial coverage tradeoff that has emerged in the field ( Table 1). Previous attempts to quantify specific cell types across the whole brain at single-cell resolution have relied upon broad cell classes with at most two molecular markers 1,5,[11][12][13][14] . Other groups have addressed this limitation directly and mapped hundreds of individual genes to achieve single-cell maps with an impressive depth of cell type coverage 4,7,8 , but only for small fractions of the total mouse brain volume.…”

“…Characterizing whole-brain distributions of different neural cell types, especially different subtypes of GABAergic (inhibitory), glutamatergic (excitatory) cells, is a topic of keen interest in modern neuroanatomy with many applications to both basic and clinical neuroscience research. Advances in molecular methods for quantifying gene expression and data analytic cell clustering techniques based on morphologic or genetic profiles [1][2][3][4][5][6][7][8] are beginning to enable the mapping of meso-and-microscale neuronal and non-neuronal cell type architecture at a whole-brain scale. Pioneering work mapping single-cell RNA sequencing (scRNAseq) data from aquatic flatworms and zebrafish onto in situ hybridization (ISH) expression provides a plausible route to mammalian cell-type mapping in the nervous system 9,10 .…”

“…Whole-brain mapping of broad classes of neuronal and glial subpopulations at single-cell resolution was demonstrated using purely computational methods 5 , with the limitation that the mapped cell types required unique identification via a single molecular marker. Other groups have successfully mapped more specific cell types in small fractions of the total mouse brain volume, including the visual cortex 4 , somatosensorty cortex 7 , and hypothalamus 8 at single-cell resolution.…”

“…Three key methodological innovations make this possible. First, we employ a matrix of known single-cell gene expression obtained from scRNAseq data, thereby moving away from single cell markers 1,5,[11][12][13][14] and instead incorporating all available and relevant information across genes and cell types. We perform non-negative constrained inversion of this matrix, similar to Grange et al 17 , who first applied this approach using microarray data.…”