Editorial NexoScience
A syringe, four tiny balls of human brain tissue and a newborn mouse with a hollow where most of its cortex should be. That’s the whole recipe. A few months later, the human tissue had taken over almost all of the empty space, hooked itself up to the mouse’s nervous system and started sending waves of electrical activity across its surface.
The tissue came from human brain organoids, small clusters of brain cells grown in the lab from stem cells. Scientists at Stanford University call the procedure “xenocortication,” and they described it Wednesday (Sept. 16) in the journal Nature. The goal isn’t to build a smarter mouse. It’s to give researchers a living model of the developing human cortex, big enough to study how it wires up, how it gets injured and how it might be treated.
Organoids grow better inside an animal. In a dish, brain organoids stay small and fairly simple. Inside a living host they get signals that no one knows how to copy in the lab. “There are some cues that are present in vivo that are really important, and we simply don’t know what to add in, in vitro,” Sergiu Pașca, a professor of psychiatry and behavioral sciences at Stanford and a co-author of the study, told Live Science. His team showed this in 2022, when organoids transplanted into newborn rats grew larger and more active than their lab-dish twins.
But there was a catch. Human neurons mature far more slowly than rodent ones, about 20 times slower, according to Pașca. So the host’s own cortex races ahead, grabs the connections first and leaves the human cells squeezed for room. In the rat work, the graft never filled more than about a third of one side of the cortex.
The fix was to clear the room first. The team bred mice that barely build a cortex at all. They switched off a gene called Esco2, which dividing cells need to keep their freshly copied chromosomes together, but only in the cells that would normally become the neocortex and the hippocampus (the brain’s main memory hub). Without it, those cells die early in development. Brain scans showed these “apallial” mice had about half as much brain tissue as normal lab mice. The mice also lacked a working immune system, so they wouldn’t reject human cells.
Here’s the thing: the mice were fine. Or close to it. They walked, ran and explored at the same speed as normal mice, and reacted to heat and touch the same way. That matters, because a model animal that can’t move would be useless for studying behavior.
Four organoids went in, and most of them stayed. When the pups were one to two weeks old, the researchers injected two organoids into each side of the empty skull. The grafts took hold in about 86% of the 29 mice tested, the same success rate the team had seen in rats with an intact cortex. Then the tissue grew fast. Between the second and third month, graft volume rose nearly fivefold. By three months, human tissue made up about 92% of all the cortical tissue in the mouse’s head.
Think of it as moving into a house where the upstairs was never built. In the rat experiments, the human cells had to squeeze in among tenants who arrived first. Here, the whole floor was empty and the plumbing (the connections running down to the rest of the brain) was waiting to be hooked up.
The human cells wired themselves into the mouse. Using fluorescent tags and a modified rabies virus that tracks connections one step back, the team saw human nerve fibers running deep into the mouse brain, and mouse neurons from the thalamus and other regions sending signals into the graft. Some human fibers went even farther. They reached the mouse’s spinal cord in the neck, something the researchers did not see when they placed an organoid into a mouse with a normal cortex.
That long-distance wiring points to a special type of cell. The grafts were rich in “layer 5 extratelencephalic” neurons, the class that includes the cells sending commands from the cortex down to the spinal cord. These neurons almost never show up in organoids grown in a dish. In a database of more than 200,000 lab-grown organoid neurons, only three carried their genetic signature. The mouse grafts held more than three times the share of these cells seen in earlier transplant methods. A handful even looked like von Economo neurons, large spindle-shaped cells found in big-brained mammals such as humans but not in rodents, and linked to social behavior and psychiatric disease.
The graft fired in waves. To watch it work, the team made the human neurons glow when they were active and then cleared the mice’s skulls optically so a camera could see through. The grafts produced big, coordinated bursts of activity lasting tens of seconds, followed by minutes of quiet. Each wave started in one spot and swept across the whole graft in about a tenth of a second. Electrodes placed inside the tissue picked up the same rhythm. The bursts also lined up with movements of the mouse’s face.
This kind of synchronized firing is typical of an immature brain. Genetic comparisons with human fetal tissue placed the grafted neurons at roughly the late second trimester of pregnancy. In other words, the mice carried a large sheet of fetal-stage human cortex.
The mice with human tissue remembered better. In a Y-shaped maze, healthy mice tend to explore a new arm instead of going back into the one they just visited. That takes a short-term memory of where they’ve been. Normal mice and those with human grafts beat chance on this test. Mice missing their cortex and receiving no graft did not. The result held in two separate groups of mice tested months apart.
The picture wasn’t all tidy, though. Both groups without a normal cortex had trouble learning to fear a tone paired with a mild foot shock. Both also lost the neat timing of diagonal paws that normal mice show when they walk. And when the team tracked every small movement the mice made in an open arena using a machine learning tool, the grafted mice landed somewhere in between: different from normal mice, different from the mice with no cortex, and less consistent from one session to the next.
A lack of oxygen hit the human tissue hardest. The most practical test came last. Oxygen starvation around birth can damage the developing brain and is linked to cerebral palsy. The team kept grafted mice and normal mice in air with just 5% oxygen, about a quarter of normal, for five hours. A protein that cells make under low oxygen, called HIF1α, lit up in the human graft but not in the neighboring mouse tissue, and not in normal mice given the same treatment. In the weeks after, the human tissue filled with support cells that typically swarm to a brain injury. Two days after the exposure, the grafted mice also changed how they walked, keeping three or four paws on the ground more often. Normal mice didn’t.
That’s the kind of readout the researchers were after: a human brain cell response and a change in behavior, in the same animal.
This is not a human cortex in a mouse. The authors list the gaps plainly. The graft didn’t form the six neat layers of a real cortex, didn’t divide into specialized regions, had few of the inhibitory neurons that balance brain activity and stayed at a fetal stage. The organoids were at most 177 days old. The researchers wrote that the behavior they saw probably comes from the human graft working together with the mouse’s remaining circuits, and that it’s still unclear whether the human neurons are needed for any specific behavior. The experiments also didn’t include grafts of non-cortical tissue, which would help show whether the effects come from cortex cells in particular or from any graft at all. Many of the tests used small groups of mice, some with only three animals.
The ethics question is built into the design. Putting human brain tissue into animals raises an obvious worry: could it change what the animal can do? The study was reviewed by Stanford bioethicists and an outside ethics committee, and the mice went through a battery of movement and memory tests partly to check for new abilities. “We have not seen emergence of any new properties,” Pașca told Live Science. The authors wrote that building mice with a more mature, more human-like cortex will need ethical guidance agreed in advance. Pașca said that trying the method in monkeys is “an experiment that I don’t see is justified at this point.”
For now, the next steps are more down to earth. The team sees the mice as a way to test drugs on a large chunk of living human cortex, to study injuries like low oxygen or toxins before birth, and to swap in organoids carrying genetic changes linked to conditions such as autism. The mouse still grows up in weeks. The human tissue inside it is still on a timeline measured in months, and closing that gap is the problem left to solve.
Source: Kaganovsky, K., Kelley, K. W., Gschwind, T., Harary, P. M., Kochalka, J., White, A. D., et al. (2026). “Developmental xenocortication using human-derived organoids in mice.” Nature. DOI: 10.1038/s41586-026-11032-2
This article summarizes published scientific research for general information. It is not medical advice and does not replace consultation with a qualified health professional.
