In October 2023, a set of 21 papers published simultaneously in the journals Science, Science Advances, and Science Translational Medicine reported the largest, most-detailed maps to date of the cells making up portions of human and nonhuman primate brains [
1,
2]. Accomplished by multiple large teams of neuroscientists in the United States and Europe, the collaborative effort also included the classification of 3300 different cell types in the human brain. Two months later, in December 2023, in a set of ten papers published simultaneously in the journal Nature, a related research collaboration identified more than 5300 cell types and pinpointed their locations within a fully completed map of the more than 32 million cells in the adult mouse brain [
3,
4]. The work holds great promise for advancing the field of neuroscience, providing new tools and insights that could be key to unlocking the mysteries of the brain and finding cures for neurological diseases.
"Understanding the human brain at such unprecedented resolution will help reveal which cell types are most affected by specific mutations that underlie neurological diseases," said Ed Lein, senior author on several of the papers and an affiliate professor in the Departments of Neurological Surgery and Laboratory Medicine and Pathology at the University of Washington and senior investigator at the Allen Institute for Brain Science in Seattle, WA, USA. "It should ultimately point to new treatments and offer a new understanding of who we are as a species."
The groundbreaking science has been supported primarily by the US National Institutes of Health under its Brain Research Through Advancing Innovative Neurotechnologies (BRAIN) Initiative, a 3 billion USD program launched in 2014 to develop tools and technology to understand and map the human brain [
5]. Researchers in the BRAIN Initiative Cell Census Network (BICCN), established in 2017 with the initial goal of generating a comprehensive three-dimensional (3D) reference brain cell atlas for mice [
6], have performed the painstaking classification and mapping that underpin the new brain atlases.
For more than a century after they first began looking at brain tissue under a microscope in the late 1800s, scientists had primarily classified brain cells according to their shape (e.g., star-shaped astrocytes or pyramidal cortical cells) or activity (e.g., fast-spiking interneurons). More recently, neuroscientists have derived other cell classification schemes based on gene and protein expression.
These methods, though, had been limited to investigating cells one at a time. However, the human brain contains 86 billion neurons, Lein said. Trying to classify large swaths of these neurons has been "a very challenging task in need of a technological breakthrough," he said.
That breakthrough has mostly come over the last decade from the field of genomics, where it became possible in the 2010s to sequence the deoxyribonucleic acid (DNA) and ribonucleic acid (RNA) of individual cells [
7]. This advance allowed scientists to see not only all the genes in a cell, but the complete set of gene readouts, or the DNA code transcribed to RNA molecules used to assemble proteins, information collectively called a transcriptome.
At the same time, sequencing capabilities have exponentially advanced to improve throughput to the point where it is possible to obtain transcriptomes for individual cells in a single brain relatively quickly and inexpensively-around one million per day at a cost under 1000 USD [
8⇓-
10]. Using techniques collectively referred to as spatial transcriptomics, neuroscientists can now combine that single-cell transcriptomic information into an atlas that is hierarchically organized, detailing cell classes, subclasses, and thousands of individual cell clusters within the brain. These techniques, the most common of which is multiplexed error-robust fluorescence in situ hybridization (MERFISH), generally rely on fluorescent imaging of slices of brain tissue in which multiple probes that fluoresce in a rainbow of colors cling to highly specific sections of RNA. The color combinations detected reflect unique cell types and can be used to directly visualize and count as many as a billion RNA transcripts from hundreds to thousands of genes across
brain tissue slices with single-cell resolution [
11].
Using such single-cell spatial transcriptomics, BICCN researchers in 2021 reported on their work building a cell census of the mouse, marmoset, and human motor cortices (
Fig. 1), and a cell-level map of the mouse motor cortex [
12]. Among the findings was that while many cell types are conserved across the three species, there is a large set of species-specific cell-type gene expression profiles that reflect evolutionary specialization.
Building on other technological advances in engineering and computer science, most of this work is now automated and relies on sophisticated software for data collection and analysis, greatly accelerating the process of discovery. "In just a couple of years, we went from one mouse brain region to a complete map of the mouse brain and a very good draft of the human brain," Lein said. "That really highlights the transformative scalability of these methods. Things are taking off in our ability to look across the entire organ."
Spatial transcriptomics is just the beginning of what is possible in neuron classification. Researchers are now working on ways to further classify cells across a variety of biochemical characteristics. For example, they can include the cell epigenome [
13,
14]-the collective chemical modifications to a cell’s DNA and chromosomes that determine how the cell’s genetic information gets expressed-and are close to acquiring the cell proteome, a catalog of the cell’s expressed proteins. Researchers are also working on an accounting of the neurotransmitters and neuropeptides used by different cells. These additions to the brain atlas could be used to explain the initiation and transmission of chemical signals within the brain, which are the basis for brain circuit operation and overall brain function [
5].
One of the major revelations coming out of the human and mouse brain atlases is the deep connection between a cell’s genetic identity and its spatial position, said Hongkui Zeng, director of the Allen Institute for Brain Science and senior author of the 2023 Nature paper reporting the complete mouse brain atlas [
3]. This relationship offers clues into the evolutionary history of different brain regions and how they work together, Zeng said. To assist other researchers studying the mouse brain, the new atlas has been made publicly available through a searchable online platform that can be queried using the generative artificial intelligence tool ChatGPT [
15⇓-
17].
In one of the human brain atlas papers published in October 2023, a comparison of the cellular and molecular properties of the human brain and several nonhuman primate brains-chimpanzee, gorilla, macaque, and marmoset-revealed apparent similarities in the types, proportions, and spatial organization of cells in the cerebral cortex of humans and nonhuman primates [
18]. The genetic expression of cortical cells across species suggests that relatively small changes in gene expression in humans gave rise to changes in neuronal wiring and synaptic function that likely paved the way for the human brain’s superior capacity for learning and adapting to changing environmental conditions.
Another paper examined cellular diversity across human brain structures. While neuroscientists have focused much of their research on the brain’s outer cortical shell, which is responsible for memory, learning, language, and other functions, the reported research suggests that most cellular diversity lies in evolutionarily older structures deep inside the brain [
19]. Lein said the finding is not surprising. "The cortex is a very complicated part of the brain, but it is effectively a repeating computational unit; it takes in input and has output," he said. "The lower parts of the brain, on the other hand, control everything that we do behaviorally, including very basic things like respiration, appetite, sleep, and many other behaviors.”
Though the main goal of the BRAIN Initiative has been the development of tools to study the brain, many researchers are already exploring ways to combine classic genomics with what is being learned about cell censuses and whole brain atlases in efforts to identify root causes of-and treatments for-human disease. Genome-wide association studies, for example, have highlighted dozens of genes that appear to be related to neurological and psychiatric diseases in people. Scientists will soon be able to map those genes onto high-resolution maps of the developing brain to better see where they are expressed and at what stages of development, said Arnold Kriegstein, senior author of several BICCN papers and professor of neurology and director of the Developmental and Stem Cell Biology Program at the University of California, San Francisco (San Francisco, CA, USA). "We can pinpoint which regions to look at during specific stages of development for the onset of diseases like autism or schizophrenia," Kriegstein said. "We can also see what cell types are vulnerable and how they are affected by genetic changes that predispose people for the disease."
Other researchers are developing precision medicine tools to genetically target specific cell types involved in neurological diseases. "For something like gene therapy, where you are trying to replace a defective gene of some sort, it has typically been done in a very broad way," said Margarita Behrens, another senior author of several BICCN papers and research professor in the Computational Neurobiology Laboratory at the Salk Institute for Biological Studies in La Jolla, CA, USA. Behrens said that the gene therapy technology that would be guided by the brain atlas is getting closer to being able to target viruses to specific cell types, which could help avoid the potentially toxic side effects of a broader viral infection.
According to Zeng, the single-cell spatial transcriptomics work now reported by the collaboration’s researchers sets the stage for capturing other characteristics of individual cells ranging from protein expression and cell physiology to cell function and how cells are wired together. Understanding the complexity of the brain along all these dimensions, she said, will provide scientists and clinicians with important insights about neurological and psychiatric disorders and how to develop personalized medicine approaches for treating them.
The next major step for Behrens, Kriegstein, Lein, Zeng, and their BRAIN Initiative colleagues is the BRAIN Initiative Cell Atlas Network (BICAN), a five-year, 500 million USD project launched in 2022 [
20]. The aim of this collaboration is to generate a complete reference atlas of cell types in the human brain across the lifespan and to map the cell interactions that underlie a wide range of brain disorders. Lein said he anticipates near-complete maps of human and nonhuman primate brains will be generated in the next three to four years. "The job is not yet done for the human brain," Lein said. "We have a draft that is one thousand times bigger than anything that came before, and now we have the challenge of scaling everything up. We are in an exciting new phase."
PDF
(430KB)
