Knowing how the human brain develops is critical to understanding how things can go awry in neurodevelopmental disorders, from intellectual disability and epilepsy to schizophrenia and autism. But between the fact that researchers cannot poke around inside growing human brains and the inadequacies of animal models, scientists currently do not fully understand the process. “We know a bit about the early stages because [the situation is] very similar to what happens in rodents,” says psychiatrist Sergiu Paca of Stanford University. “But everything beyond the second trimester [of pregnancy] and soon after birth is poorly understood.”

Enter the invention of brain “organoids”: cells grown in 3-D clusters in the lab and designed to mimic the composition of the organ’s tissue. The technology recently reached the point where specific brain regions can be modelled for sufficiently long periods to allow researchers to study their development. Paca and his colleagues have now used organoid models of parts of the human forebrain—the seat of higher cognitive abilities such as complex thought, perception and voluntary movement—to peer into how gene activity drives brain development. “The work brings new understanding of how, as the brain is formed, distinct regulatory regions of the genome are used to execute specific tasks—for example, the generation of specific types of neurons,” says neuroscientist Paola Arlotta of Harvard University, who was not involved in the new study. The researchers used their findings to map genes associated with certain disorders to specific cell types at specific stages, giving insight into the origins of conditions such as autism and schizophrenia.

In the study, published Thursday in Science, Paca’s team coaxed human induced pluripotent stem cells into 3-D cultures mimicking the earliest stage of two parts of the forebrain. These stem cell structures included two types: cortical spheroids, which mimicked the rear, or dorsal, forebrain, and subpallial spheroids, which represented the front, or ventral, forebrain. The researchers induced the cells to form nascent brain tissue—neural progenitor cells—using a cocktail of drugs and proteins. The ventral area of real brains contains inhibitory, or brakelike, neurons that are not initially present in the dorsal region but that migrate there later. The team created a model of the ventral region by adding an extra protein to cortical spheroids at just the right time. From there, the organoids develop automatically, first generating different types of neurons and then forming other brain cells called glia.

The researchers used a genetic sequencing technique called ATAC-seq to assess which genes were accessible for forming proteins, which of them were active, and which stages and cell types they were active in. (DNA is wrapped in a structure called chromatin, which changes shape to control access to genes.) The team has made these data available to the research community online. “This is one of the first analyses of the chromatin landscape: the parts of the DNA that are open, which then impact gene expression and [which] proteins are made and, therefore, the functions of cells,” says neuroscientist Madeline Lancaster of the University of Cambridge, who was not involved in the study. “This is very powerful. And to have it at various times during brain organoid development is a great resource.”

The researchers grew the organoids for 20 months, covering prenatal and postnatal stages of development. “Human brain development is an incredibly long process,” Paca says. It takes 27 weeks just to make all of the neurons in the cerebral cortex. Then, starting in the third trimester, glial cells are produced, which continues long after birth, he says. The researchers compared their data with existing data sets from human tissue samples to confirm that the changes they saw mirrored those occurring in real brains. Such organoids still have many limitations, though. “There's no vasculature, sensory input or microglia—the brain’s immune-surveillance cells, which are derived outside the brain,” Paca says. “Despite this, we find there’s an intrinsic program: cells know what specific cell type to become and when.”

Much of this development occurs during a “wave” of activity between 80 and 260 days, during which genetic changes generate the different cell types. “It’s very exciting to be able to watch these dynamic regulatory events unfold as organoids develop,” Arlotta says. “There’s so much new information here to guide years of work aimed at decoding the mechanisms that form the human brain.” Further analyses identified some proteins as potential key players driving this program—a critical first step.

The team then used its data to find when genes linked to schizophrenia and autism were active and which cell types they were active in. “Some of these processes are going awry early in development, but the timing is different for autism and schizophrenia,” Paca says. Many autism risk genes appeared in progenitor cells, whereas schizophrenia genes were more active later—in glia and inhibitory neurons, for instance. “A stronger, earlier genetic insult [such as a mutation] to the developing brain probably is associated with autism,” Paca says. “A later one that’s a bit subtler but can have long-term consequences is associated with schizophrenia.”

The next step is to look at how genetic variants found in patients with a range of neurological disorders affect brain organoid development. “This study provides a nice resource to look at when a particular disease gene might be playing a role,” Lancaster says. “That tells us what stage to look at if we mutate that gene.” Paca has the same goal in mind: “We first have to create this map of the developing brain but ultimately use it to understand disease, because that’s the promise,” he says.