Mice Grow Human Brain Tissue Where Their Own Cortex Should Be

A MOUSE pup, ten days old, is anesthetized and set into a tiny surgical frame with almost no cortex or hippocampus of its own, its skull built instead around a fluid filled cavity where those structures should sit. A researcher threads four pea sized clusters of human brain tissue, grown from stem cells in a dish, into that empty space. Stanford scientists have shown that human brain tissue transplanted into a mouse engineered to lack most of its own cortex will grow to fill the cavity, wire itself into the animal’s nervous system and reach all the way to the spinal cord, giving researchers a living, behaving animal in which to study human brain disorders that no dish of cells can replicate. The mice that receive this transplant, which the team calls xenocortical mice, grow up mostly indistinguishable from ordinary mice, right up until the point someone looks inside.

Human stem cell models have promised, for over a decade, a way around the central problem of brain science: living human brain tissue is almost never available to study. Lab grown organoids solved part of that, and transplanting them into newborn rodents solved another part, but competition for space and speed with the host brain always left the human tissue outnumbered and outpaced.

The team’s fix was to remove that competition before it starts. Using a genetic trick, they deleted a cell survival gene called Esco2, but only in the mouse cells that would otherwise build the neocortex and hippocampus, killing those cells off early in development and leaving what the researchers call an apallial mouse, after the pallium, the tissue layer those structures grow from. Careful breeding and extra feeding kept the pups alive despite losing 50% of total brain volume, and by adulthood their brains held only a sliver of the neurons found in an ordinary mouse cortex. Transplant four human organoids, two per hemisphere, into that cavity in a newborn apallial pup, and 86.2% of the surgeries succeed, with the graft growing 4.7-fold in volume between two and three months afterward.

By three months, the human tissue makes up 91.9% of all the cortical volume in an engrafted mouse’s brain, at a density of roughly 32,000 neurons per cubic millimeter. That is not a patch of foreign cells sitting inertly in a cavity.

The graft wires itself in. Fibers grow out through the mouse’s existing subcortical pathways, and imaging of water movement along the tissue shows the same organized directionality seen in an intact mouse cortex. In some animals, the researchers traced human axons all the way down the spinal cord, evidence that the graft had built long range connections a rodent brain would ordinarily supply itself.

A Neuron Never Grown in a Dish Before

Sergiu Pasca, the senior author, has spent years arguing that some things about the human brain simply will not show up in cultured tissue. “We couldn’t study complex human behavior in a dish,” he said. Inside the xenocortical grafts, his team found a striking confirmation of that argument: a population of large, cigar shaped brain cells called von Economo neurons.

Von Economo neurons had previously turned up only in autopsied human brains, never in cultured organoids, and the new analysis found essentially none in a large public catalog of organoid single cell data drawn from many different labs and protocols. Inside the mouse grafts, though, they appeared reliably, in all three animals the team checked closely, with roughly triple the fraction of a related projection neuron type compared with earlier transplant approaches. These cells are unusually large and sparse, projecting widely from regions tied to social behavior, and in humans they are thought to be an early casualty of frontotemporal dementia. “This cell type appears to be particularly vulnerable in frontotemporal dementia, a neurodegenerative disorder that can begin in midlife,” Pasca said, adding that early symptoms often show up as changes in personality or language well before memory loss does.

Getting a cell type this rare to appear at all took more than transplantation; it took removing the crowding that earlier approaches never solved. Whether the mouse’s own subcortical circuitry, rather than anything intrinsic to the graft, is what coaxes these neurons into forming is a question the paper leaves open.

What Low Oxygen Does to a Human Cortex Inside a Mouse

The clearest demonstration of what the model can do came from an injury experiment. The researchers dropped the oxygen level around xenocortical mice to 5% for five hours, mimicking the kind of oxygen deprivation that, in a human infant, is a recognized risk factor for cerebral palsy. In all three animals tested this way, a stress protein called HIF1a lit up specifically inside the human graft and barely at all in the mouse’s own remaining brain tissue, and follow up scans ten days later showed new patches of altered blood vessels confined to the same region.

The behavioral consequence tracked the biology closely. After the hypoxic exposure, xenocortical mice, but not mice with an intact mouse cortex, changed how they walked, planting three or four paws on the ground at once rather than their usual two, a gait shift the team measured using automated paw tracking.

None of this proves the model captures every feature of human cerebral palsy, and the authors are careful to say so. The graft, even at its most developed, still lacked a fully layered cortical architecture, a complete population of inhibitory neurons and the areal specialization that distinguishes, say, visual cortex from frontal cortex in an intact human brain.

For disease researchers, the appeal is less about literal fidelity than about access. A drug candidate, a patient derived cell line, or a suspected genetic risk factor can now be tested inside an animal that behaves, ages and can be scanned repeatedly, rather than only in a dish. Pasca described the payoff in practical terms: cells carrying a patient’s own genetic makeup can be transplanted and their downstream effects tracked in a living circuit, something no culture dish allows. That combination, a working nervous system built substantially from human cells, is what makes the platform useful well beyond this one injury experiment. Alison Singer, president of the Autism Science Foundation, who was not involved in the study, called individualized organoid modeling “a critical step toward precision medicine.”

The team is already thinking about how the model should be governed as much as how it should be used. Stanford convened outside ethicists, primate and human brain specialists, and patient advocates well before publication, and Pasca has said the group’s animal review process treated the ethical questions as seriously as the technical ones. Future versions of the platform, engrafted earlier in development or into animals with less crowded brains, could let the human tissue mature further, which is exactly the scenario the ethics conversation was convened to anticipate.

FAQ