Summary

Researchers developed a mouse model in which human stem-cell-derived cortical organoids occupy most of the cerebral cortical space and connect with the host nervous system. The grafts showed organized electrical activity, influenced behaviour and produced measurable responses to experimentally induced hypoxic injury.

Researchers have developed a mouse model in which human stem-cell-derived cortical organoids occupy most of the space normally taken by the mouse neocortex and hippocampus. The grafts formed connections with the host nervous system, generated coordinated electrical activity and produced behavioural readouts, creating an in vivo platform for studying aspects of human cortical development and injury.

The study, published in Nature on 16 September 2026, calls the approach xenocortication. It uses genetically modified, immunodeficient mice whose developing dorsal and medial pallium—the embryonic region that gives rise to much of the neocortex and hippocampus—is severely depleted.

Contents

A larger space for human cortical grafts

The researchers deleted the Esco2 gene in cells expressing Emx1, a marker of the dorsal and medial pallium, on an immunocompromised SCID mouse background. This caused targeted cell death during development and removed most mouse cortical tissue. MRI showed an approximately 50% reduction in total brain tissue compared with control mice, while whole-brain single-nucleus RNA sequencing confirmed a sevenfold depletion of dorsal-pallium-derived glutamatergic neurons.

The resulting cavity was engrafted shortly after birth with four human cortical organoids, three-dimensional tissues produced from human induced pluripotent stem cells. Grafts survived in 86.2% of 29 transplanted mice across three stem-cell lines. Between two and three months after transplantation, graft volume increased approximately 4.7-fold. At three months, human-derived tissue made up 91.9% of the combined cortical tissue volume in the sampled mice.

The graft contained several classes of developing human cortical cells, including deep- and superficial-layer glutamatergic neurons, progenitor cells, oligodendrocyte progenitor cells and astrocyte-lineage cells. Transcriptomic comparison placed the glutamatergic neurons at a maturation stage corresponding approximately to the late second trimester of human cortical development when assessed 24 weeks after differentiation.

The researchers also identified layer 5 extratelencephalic, or L5-ET, neurons. These include projection-neuron types that are difficult to generate reliably in organoids grown only in vitro. A small population of cells had the large soma and bipolar or corkscrew-like morphology associated with von Economo neuron-like cells. These cells were found in all three sampled xenocortical mice and represented 0.16% of fluorescently labelled somas.

Human neurons connected to the mouse nervous system

Tracing experiments showed that axons from the human graft formed organised pathways through subcortical structures and reached the superior colliculus. Retrograde tracing identified host inputs to the graft from the mouse palaeocortex, thalamus and pallidum. Human-derived projections also extended into the mouse cervical spinal cord in all three mice examined, whereas comparable projections were not detected from a focal organoid graft placed in an intact mouse cortex.

The grafts were electrically active in awake animals. Widefield calcium imaging revealed bursts that began in local regions and propagated across the dorsal graft in roughly 100 milliseconds. Bursts lasted tens of seconds, recurred every few minutes and contained individual events separated by approximately one second. The activity was accompanied by local-field-potential bursts recorded with a 32-channel probe placed within the graft, supporting the presence of coordinated network activity rather than isolated cellular signals.

The study did not find a continuous, canonical cortical-layer structure throughout the graft. Instead, spatial transcriptomics identified local regions enriched for upper- or deep-layer neuronal subclasses, with related cell types positioned closer together than expected by chance. This pattern was consistent with local self-organisation, although the researchers did not observe a clear anterior–posterior gradient of cortical arealisation at the measured time point.

Behaviour and experimental injury

The mice retained broadly preserved locomotion. Control, cortex-depleted and xenocortical mice moved at similar rates in activity-chamber and CatWalk tests, but the experimental groups showed altered paw coordination and changes in the organisation of spontaneous behaviour. Unsupervised analysis of three-dimensional movement divided behaviour into 60 subsecond “syllables”; xenocortical mice occupied a behavioural state distinct from controls and generally closer to the cortex-depleted group.

In a spontaneous-alternation Y-maze test, control and xenocortical mice performed above chance, while cortex-depleted mice did not. Both cortex-depleted and xenocortical groups showed reduced freezing in tests of aversive trace conditioning. These findings indicate that the graft influenced selected behavioural functions, while not restoring every feature of the missing mouse cortex.

The researchers then exposed mice to 5% oxygen for five hours, following a one-hour gradual reduction from atmospheric oxygen. In xenocortical mice, the human graft showed HIF1α immunoreactivity, MRI changes associated with deoxygenated blood and increased densities and morphological changes in astroglia and microglia. Two days after the exposure, xenocortical mice also showed altered gait-related measures, including greater three- or four-paw support, without a change in average running speed.

Xenocortication therefore combines a large human cortical graft with circuit-level recording and behavioural testing in a living animal. The model uses immunodeficient mice, severely alters the host brain and produces tissue at a mid-gestation-like developmental stage, so its results are best interpreted as a tool for studying human neural development and disease mechanisms rather than as a direct model of the adult human cortex.

Sources