Stanford's xenocortication opens new window into human brain development

In September, neuroscientist Sergiu Pașca's team at Stanford published results of an experiment that pushes the boundary between what we can observe about human brain development and what we can ethically attempt. They grew human brain tissue from stem cells, transplanted it into mice whose cerebral cortex had been genetically prevented from forming, and watched it take root. The tissue matured, produced several types of human brain cells, became electrically active, wired itself into the mouse nervous system, and correlated with measurable changes in how the animals moved and behaved.

But the researchers were careful to name what they had not done. They had not created a human brain inside a mouse. The transplanted tissue remained immature, lacked the ordered layering of a normal human cortex, and the study could not prove that human neurons directly caused any particular behaviour. What they had achieved was narrower and, in its way, more useful: a living window into how human brain tissue develops when given space and support inside a functioning nervous system.

The problem they were solving is old and stubborn. Understanding how the human cortex develops matters for evolutionary biology—the cortex expanded dramatically during human evolution, acquiring the specialised cells and connections that underpin language, planning, and complex social behaviour. It also matters medically, because many neurological and psychiatric disorders begin during development, long before symptoms appear. Yet scientists face a fundamental constraint: they cannot experimentally manipulate a developing human brain the way they can in laboratory animals. Imaging can track structural changes. Tissue samples from foetuses and autopsies provide snapshots. Neither allows a researcher to alter a gene, observe the consequences over months, and measure the result. Human developmental neuroscience has long operated with a major blind spot.

Stem cell technology opened a second window. Scientists can reprogram ordinary human cells into induced pluripotent stem cells—a flexible, earlier state—and coax them to become nerve cells. Grown together in three dimensions, these cells self-organise into structures called organoids. A brain organoid is a small piece of laboratory-grown neural tissue that mimics some features of early brain development. Organoids let researchers watch how human neurons develop, introduce disease-related genetic changes, and test drug responses. But lab-grown tissue lacks key features of a living brain: normal blood circulation, sensory input, and long-distance connections to other parts of the nervous system. When researchers transplanted human brain organoids into rodents to improve survival and maturation, they ran into a new constraint: the rodent already had its own cortex. The transplanted human tissue had to compete for limited space and neural connections, and never amounted to more than a fragment of the host's brain.

The Stanford team opened a third possibility by re-engineering the host instead of just the graft. Using genetic techniques, they produced mice in which much of the neocortex and hippocampus never formed. They also bred the mice to lack a functioning immune system, so human tissue would not be rejected. Human cortical organoids were transplanted into the vacant space soon after birth. The researchers called the approach xenocortication—'xeno' for something from another species, cortex for the brain's outer region involved in higher functions.

Freed from competition with mouse cortical tissue, the human grafts expanded substantially. By three months, they made up about 92 percent of the animal's cortical tissue. They produced several types of developing cortical neurons, along with progenitor cells and astrocytes at a stage of maturity roughly equivalent to the human cortex late in the second trimester of pregnancy. Notably, the grafts generated deep-layer neurons carrying the molecular signature of von Economo neurons—large, distinctively shaped cells found in humans and a few other large-brained mammals, but absent in rodents, and long associated with social cognition and psychiatric illness. Such neurons are almost impossible to grow in a dish. The grafts became electrically active, producing slow waves that swept across the tissue every few minutes, a pattern characteristic of developing rather than mature cortical networks. They formed connections with surviving parts of the mouse nervous system, and human nerve fibres reached as far as the cervical spinal cord—something not seen in mice with an intact cortex.

The next question was whether this integration had any functional effect. Mice lacking most of their neocortex and hippocampus—roughly half the brain by volume—remained active and mobile, ran at normal speeds, and responded normally to heat and touch. To detect subtler differences, the researchers used a machine-learning system called Motion Sequencing, or MoSeq. A depth camera recorded freely moving mice, and software divided their movements into 60 recurring sub-second patterns—rearing, pausing, turning—creating a detailed behavioural fingerprint. By this measure, mice lacking much of their cortex behaved differently from normal animals. Xenocortical mice showed an intermediate pattern, suggesting that transplantation altered behaviour without fully restoring it. In a Y-shaped maze used to test working memory, normal and xenocortical mice performed above chance, whereas mice lacking much of their cortex did not. This suggests possible partial functional compensation, though it does not prove that human neurons restored memory.

To demonstrate the platform's potential, the researchers exposed the mice to five hours of severe oxygen deprivation. The response was confined to the human tissue: a molecular marker of oxygen stress appeared in the graft but not in the adjacent mouse brain, and not at all in animals without a graft. Two days later, the xenocortical mice walked differently, shifting to a more stable stance without any change in speed. Imaging ten days on showed changes in the graft's blood vessels. Under the microscope, the graft's supporting cells had multiplied and become more complex. This creates a potentially powerful experimental pathway: introduce a disease-related mutation into human cells, allow the resulting neural tissue to develop within a living nervous system, observe changes in cells and circuits, assess whether behaviour is affected, and test possible treatments. Such models could help study developmental brain disorders, oxygen deprivation around birth that can lead to cerebral palsy, human-specific neuronal diseases, and evolutionary differences between human and animal brains.

Yet the model remains far from reproducing a developing human brain. The grafts did not reproduce the full balance of excitatory and inhibitory neurons. Human neurons mature far more slowly than mouse neurons, creating a mismatch between the developmental timelines of graft and host. Although anatomical connections were observed, this does not prove they directly controlled behaviour. Any behavioural changes arise from a hybrid system in which human tissue interacts with the remaining mouse brain. There is no evidence that the animals developed human consciousness or a human-like mind. But future models may contain larger, more mature, and better-organised human neural tissue, with greater integration into the host brain. At what point might this alter an animal's cognition, perception, or capacity for experience in ethically significant ways? The researchers consulted bioethicists at Stanford before beginning and argued in their paper that going further—toward grafts with mature cortical layering, or transplantation into embryos or non-human primates—will require greater scrutiny before such experiments begin. Ultimately, xenocortication matters not because it creates a human brain in another species, but because it may provide a third window to study how human brain development unfolds.