Mice missing 98% of their cerebral cortex recovered memories after receiving human brain cell grafts
Researchers transplanted four clumps of brain tissue grown from human stem cells into mice whose cerebral cortex was almost entirely absent. Months later, these mice not only regained lost maze memories, their stiff, awkward gait also improved. This work comes from a team led by Stanford neuroscientist Sergiu Pașca, published in Nature.
Many readers react with alarm: Did human brain cells take over these mice? Have scientists created “human-brained mice”? Before jumping to sci-fi conclusions, the actual findings are far more grounded.
To understand the experiment, start with a basic question: How could these mice survive with nearly no cerebral cortex at all?
Early in embryonic development, Pașca’s team used genetic engineering to disable key genes governing cerebral cortex and hippocampus formation. The resulting animals are known as apallial mice: roughly 98% of their cortex and hippocampus never develop, leaving large cerebrospinal-fluid-filled cavities inside the skull.
If an adult animal suddenly lost this much brain tissue, the outcome would be catastrophic. But these mice were born without the cortex. Breathing and heartbeat are controlled by the brainstem; basic locomotion relies on evolutionarily older structures: brainstem, striatum and cerebellum. That is how they stayed alive.
Survival through neural compensation comes with clear costs: First, they move cautiously, with a stiff, uncoordinated gait. Second, they lose spatial memory entirely in mazes, wandering aimlessly with no sense of direction.
These two deficits — impaired gait and lost maze memory — became the core behavioural readouts for testing the treatment.
Why implant human brain tissue into these empty cranial cavities?
Pașca’s group set out to solve a longstanding neuroscience bottleneck: many neuropsychiatric drugs work well in standard animal models yet fail repeatedly in human clinical trials. One major reason is that ordinary mouse brains lack human-specific advanced neurons. Meanwhile, brain organoids grown in lab dishes cannot connect into intact, living neural circuits.
Scientists badly needed a platform to observe living human neural circuits inside a living animal. Studying human brain organoids in culture hits two hard physical limits:
The first limit is oxygen starvation. In nutrient media, organoids stop growing once they reach two to three millimetres across. Without embedded microvessels, oxygen and nutrients cannot penetrate deep into the tissue, and inner cells rapidly die from hypoxia.
The second limit is incomplete maturation. Without real sensory input — light, movement, physical feedback — neurons in a dish stall in an early developmental state.
If culture dishes cannot support large, mature organoids, why not transplant them into a living host?
The researchers differentiated human pluripotent stem cells in vitro for roughly 40 days to generate four small human cortical organoids. These tissue constructs were then surgically implanted into the vacant cranial cavities of apallial mice. The resulting animals are called XCX xenocortical mice.
This specific mouse strain solves the two culture-dish bottlenecks perfectly.
Take oxygen supply first. After implantation, the mouse’s own blood vessels quickly grow into the organoid, delivering steady oxygen and nutrients. The empty cavity carries another critical advantage: since nearly all native cortex and hippocampus is absent, grafted human cells face little competition for space from native mouse neurons.
The result? Within three months, the four human tissue clumps expanded rapidly, filling over 90% of the original void. The size ceiling imposed by dish culture was eliminated by live blood perfusion.
The second bottleneck also falls away: once embedded in a living animal, the neurons receive authentic physiological sensory input, which proves crucial later.
Next comes neural wiring.
These human neurons do not grow in isolation. They form functional synapses with the mouse thalamus and send long axonal projections down through the brainstem, all the way into the spinal cord.
Recall the two original deficits.
For gait recovery: those long human nerve fibres extending to the spinal cord reconnected broken motor pathways, producing clear improvements in movement.
For memory: control apallial mice without grafts kept wandering blindly in mazes. XCX mice carrying human organoids recovered nearly normal spatial memory, performing close to healthy wild-type mice when searching for exits.
The team also made a notable discovery: they detected mature von Economo neurons inside these in-vivo grafts. These large, spindle-shaped neurons exist naturally in humans, great apes, whales and elephants. They are not exclusive to humans, yet researchers had rarely succeeded in maturing them reliably in culture. Only in this living vascularised environment, with constant sensory input, could these cells fully develop.
So the second developmental barrier was overcome.
This brings back the unsettling opening question: if the grafts connect to the spinal cord, restore memory and grow specialised advanced neurons, do these mice possess human-like cognition?
Jürgen Knoblich, a neuroscientist at the Institute of Molecular Biotechnology in Austria, gives a clear answer: no.
Scans show the four grafted tissue masses lack organised internal layering. A true human cerebral cortex has a precise six-layered architecture plus sharply defined sensory, motor, language and visual subregions. The implanted human tissue in these mice has none of that ordered structure or functional partitioning. While many cell types are present, it is simply a disorganized cluster of neurons — not a miniature human brain.
This raises another puzzle: if it is just unstructured human neural tissue, how does it rescue maze memory?
The mouse’s native brainstem, thalamus and other subcortical structures remain intact. What the animals lack is an intermediate relay layer for signal processing. As long as the implanted human neurons can fire action potentials and form synaptic connections, they can plug into the mouse’s existing circuit and fill that missing relay node. This remains a working hypothesis.
The core goal of Pașca’s multi-year platform is not to create chimeric sci-fi creatures. It provides a living assay system for studying Alzheimer’s, Huntington’s disease, schizophrenia and autism, allowing direct observation of intact human neural circuits. Whether this platform will accelerate successful drug development is still unproven.
What, then, is that maze-running mouse?
It is no monster endowed with human thought. It still senses the world through mouse whiskers and mouse eyes. Human cells live and grow inside its skull and blood supply, yet it remains, fundamentally, a mouse.
If you enjoy what I do, consider supporting me! Every little bit means the world! ❤️
☕ Buy me a coffee