We've updated our Privacy Policy to make it clearer how we use your personal data. We use cookies to provide you with a better experience. You can read our Cookie Policy here.


Full Atlas of the Mammalian Brain Reveals Unique Structure of Ancient Regions

Brain cells highlighted in red.
Millions of cells were grouped by single-nuclei analysis into nearly 5,000 cell types (left), which were then mapped onto the Slide-seq spatial dataset (right). Here, one cell type is denoted with saturated colored clusters on the left, and then mapped onto an exemplary slice from the mouse brain on the right, indicated by colored dots on the tissue slice images. Credit: Chen and Macosko labs
Listen with
Register for free to listen to this article
Thank you. Listen to this article using the player above.

Want to listen to this article for FREE?

Complete the form below to unlock access to ALL audio articles.

Read time: 4 minutes

A world-first cell atlas has opened a window into the complexity of the mammalian brain. The map details more than 5,000 cell-type clusters in the mouse brain and marks out their location. The work is the most comprehensive classification of the mammalian brain ever completed.

The study showed that the mouse brain’s complexity varied between evolutionarily ancient deep brain structures and more recent cortical brain structures.1 The work is part of a collection of nine papers published in Nature this week.

The research is part of the National Institutes of Health (NIH)’s BRAIN initiative, a multibillion-dollar project to advance our understanding of the brain’s structure and function.

One of the main priorities of the BRAIN initiative was to understand the cellular diversity of the brain, said Dr. Hongkui Zeng, director of the Allen Institute for Brain Science and a co-author of the study that details the new map. As part of this effort, the Brain Initiative Cell Census Network (BICCN) was started in 2017. “The main goal of the BICCN is to do a complete job in a mouse brain – within five years, finish a cell atlas at an anatomically precise, spatially-resolved level,” said Zeng in an interview with Technology Networks.

The new publications represent the culmination of that goal.

The cell atlas doesn’t map every cell in the mouse brain. But it does provide a representative guide to how those cells are grouped into genetically distinct types, using data from roughly 4.3 million cells. With this information, researchers now know what cell types are present in the mouse brain and at what location = they can be found.

The “what” and “where” of brain cells

The map’s creation marries together two revolutionary technologies that together provide deep data of the “what” and “where” of brain cells. The first method is single-cell RNA sequencing (scRNA-seq). This technique sequences a cell’s transcriptome, which is coded using RNA and details all of the genetic information that will be translated into proteins. The information in the transcriptomes defines the form and function of a cell. By sequencing huge amounts of scRNA-seq data, researchers can work out what defines different types of cells. But this data’s use in understanding how the brain works is massively boosted by location information said Zeng.

“It's like the genome; you can know about all 20-30,000 genes, but if you don't know how they are organized on the genome in a sequence, you don't know how they are regulated. That’s because the location of the genes in different chromosomes contains a lot of regulatory information about what genes to turn on and off and the interaction of different genes,” said Zeng.

That’s where spatial transcriptomics comes in. This links transcriptomic information to location within a tissue sample. The technology won Nature’s Method of the Year in 2020 and has gone on a meteoric rise since then. Spatial platforms like MERFISH allow researchers to simultaneously record gene information on hundreds to thousands of RNA molecules alongside their location data. “The scalability of spatial transcriptomics approaches was really the breakthrough over the last two years,” said Zeng.

Zeng explains that spatial transcriptomics cannot yet study the transcriptomic information in the same level of detail as scRNA-seq, necessitating a combination approach. But together, the two techniques enabled deep insights into the brain’s complexity.

BICCN in Numbers

The study analyzed:

  • 4.3 million cells; divided into
  • 34 classes
  • 338 subclasses
  • 1,201 supertypes
  • 5,322 clusters

Inside the “playground of evolution”

One key finding from the work was that there were differences in complexity between the dorsal and ventral brain. The dorsal brain includes structures like the cortex, thought to be the seat of higher thought. The ventral brain, including the hypothalamus and midbrain, controls functions like breathing and metabolism. These are ancient regions of the brain that have remained stable through millions of years of evolution. These areas, said Zeng, “cannot change much, because whenever you make a change, the mutation disrupts the circuits and jeopardizes the survival of the animal.”

These older areas contained a larger number of cell types that were more closely related to each other. The dorsal areas, which Zeng calls the “playground of evolution”, had fewer different cell types, but those cells had changed to take on the challenge of learning and adaptive behavior. This gave them a wide evolutionary diversity in comparison to the ventral areas.

This information will be made open access, allowing researchers to interrogate the brain’s diversity at an unprecedented level. Without this kind of insight, curing brain diseases like dementia is infinitely harder, because researchers cannot be sure what cell types are being affected by disease, or cured by treatments.

To this end, the project’s next step has the human brain in its sights.

Mapping the human brain

Under the slightly switched moniker of the Brain Initiative Cell Atlas Network (BICAN), the next stage of the work will look to create a brain cell atlas for the primate and human brain. A battery of papers published earlier this year mapped 3000 cell types in these brains – a huge step forward for the field. Nevertheless, this analysis lacked spatial context, said Zeng. By once again combining scRNA-seq and spatial transcriptomics, the BICAN team hopes to improve this human brain map.

The human brain contains billions of cells, a magnitude greater than the mouse brain. But Zeng’s prediction, based on the limited data published so far, is that this higher number won’t translate to many more cell types. “We probably will have 5000 to 10,000 cell types, but no more than that,” she stated.

What is certain is that the pace of work in this area has sped up from a crawl to a sprint. The BICAN team has five years from 2022 to finish their human brain map. That work is only just getting started.


Zhang M, Pan X, Jung W, et al. Molecularly defined and spatially resolved cell atlas of the whole mouse brain. Nature. 2023;624(7991):343-354. doi:10.1038/s41586-023-06808-9