Read our COVID-19 research and news.

Lab-grown organoids capture many aspects of brain development, but their cells fail to mature like real brain cells.

Madeline Andrews/Arnold Kriegstein Lab/University of California, San Francisco

Lab-grown ‘minibrains’ differ from the real thing in cell subtypes, gene expression

Glance at an intricately structured blob of human brain cells in a lab dish, and it’s tempting to dub it a “minibrain.” That’s the popular term for cerebral organoids, complex 3D tissues made from stem cells that are revolutionizing how researchers study neural development and conditions from autism to Zika. But the most comprehensive genetic comparison yet of cells from real brains and cerebral organoids, published today, reveals important differences between them.

Arnold Kriegstein, a neuroscientist at the University of California, San Francisco, and his colleagues first cataloged the genes turned on in individual cells from different parts of human fetal brains at 6 to 22 weeks of gestation. They then compared these patterns of gene expression to those of cells from cerebral organoids created using several previously published methods.

When it came to broad categories of brain cells—like neurons and nonneuronal cells called glia—the gene expression generally matched up. But when the researchers examined more precise subtypes of cells—a subset cells known as outer radial glia, for example—the comparisons started to break down. The organoid cells weren’t reliably maturing and expressing the specific combinations of genes that distinguish one subtype of cell from another.

That’s a potential limitation for studies that use organoids to model disease. Many such studies “reprogram” cells from a person with a disease into stem cells that can form an organoid. But nervous system diseases are highly cell-type specific, says Kriegstein, making it hard to draw conclusions about the disease if your lab-grown model doesn’t contain the exact cell type affected in the brain.

Though the organoid cells didn’t generally mature into precise subtypes, they did become specialized in another way: They took on genetic signatures that correspond to different regions of the brain. But those identities came on unpredictably. Signatures that you’d expect to see on opposite sides of the brain were sometimes right next to each other in an organoid, the team found. That could complicate studies of certain diseases, says Kriegstein, such as a form of dementia that affects only particular brain regions.

So why don’t organoids perfectly reproduce the cell types of the brain? Kriegstein’s team suggests a key factor is the stress of growing in a dish instead of a body. The organoid cells expressed genetic markers of metabolic stress, but this expression was reduced when the cells were grafted into a mouse brain. What’s more, this relocation seemed to help resolve these cells’ identity crisis, the team reports in Nature; it prompted them to take on the genetic signatures of the more developed cell types in the human brain.

That means there might be ways to make organoids more brainlike by tweaking the broth of nutrients used to nourish them in a dish, says Madeline Lancaster, a developmental geneticist at the Medical Research Council’s Laboratory of Molecular Biology. For example, researchers often culture organoids in high concentrations of glucose, “equivalent to a very, very diabetic person,” she says. “You could imagine this may not be very good for a developing brain.”

Future studies might also reduce the unnaturally high levels of oxygen present in lab conditions, says Flora Vaccarino, a developmental biologist at Yale University. That’s a relatively easy variable to tweak, she says, using incubators that control how much oxygen flows to the organoid cells.

It’s no surprise that organoids don’t perfectly replicate the human brain, she says. But the new study lays out how you can “convince yourself that you have a valid model of human development,” she says. “It gives a very optimistic message.”