Astrocytes, the Brain's Support Cells, Build Their Own Wiring That Neurons Never Touch

Astrocytes, the brain's support cells, build their own wiring that neurons never touch

The brain has a second communication network. Neurons get all the attention, but a study published this week in Nature shows that astrocytes, the star-shaped support cells that ferry nutrients and carry away waste, build their own far-reaching wiring that can connect brain regions neurons never touch. The finding is a genuine paradigm challenge to a century of neuron-centric neuroscience, and the tool the NYU Langone team built to see it is cheap enough that other labs will follow.

"For more than a century, neuroscientists have thought of neurons as the main actors in the brain," said lead author Melissa Cooper, a postdoc in neuroscience at NYU Grossman School of Medicine. "Yet our findings suggest that astrocytes, which are usually viewed as merely support cells, are also running their own widespread signaling pathway, adding another layer to how brain regions stay connected."

The paper, doi:10.1038/s41586-026-10426-6, comes from Cooper's lab under co-senior authors Shane A. Liddelow, associate professor of neuroscience and ophthalmology, and Moses V. Chao, professor of cell biology, neuroscience, and psychiatry. It is the first study to map active, brain-wide communication networks built by astrocytes and demonstrate that these pathways are highly specific rather than diffuse and nonspecific.

To trace astrocyte connections, the team engineered a fusion protein combining Connexin 43 (Cx43), the main gap junction protein used by astrocytes, with TurboID, a promiscuous biotinylating enzyme. They delivered it via AAV5 under a shortened Gfap promoter. When the fusion protein incorporates into a gap junction, molecules passing through are tagged with biotin, letting the researchers identify which astrocytes share a signaling pathway. They then made brains transparent and used light-sheet microscopy to image the full 3D network. The approach was designed to be reproducible, with a cost low enough for other labs to adopt it.

The results across hundreds of mice showed two types of astrocyte networks: local ones confined to single brain regions, and long-range networks that robustly interconnect multiple regions across hemispheres, often following patterns distinct from neuronal networks. Crucially, some of these pathways linked brain regions with no direct neuronal connection between them. In mice genetically engineered to lack astrocyte gap junctions, the communication networks largely disappeared, confirming the pathways depend on physical gap junction bridges.

The finding matters because decades of research using gap junction knockout mice had already established that these structures are necessary for memory formation, synaptic plasticity, coordination of neuronal signaling, and closing the visual and motor critical periods. But nobody could see the full architecture of the network they form. This study shows that architecture is specific and experience-dependent, not a generic syncytium.

When the team trimmed whiskers on one side of a mouse's face, a pathway from the region processing whisker touch got smaller and reconnected to different astrocyte partners. The network remodeled. "The fact that astrocyte networks shrink and reroute after a loss of sensory signals suggests they may be shaped by experience," said Chao. "It also raises the possibility that each of us has a somewhat unique pattern of connections molded by what our brains have learned and lived through."

That is a striking claim from a co-senior author, and it will need strong evidence before it can be taken beyond mice. Liddelow was direct about the caveat: while gap junctions and astrocytes exist in humans, it remains unknown whether the networks link the same regions in the same way. The field has a history of astrocyte hype. The tripartite synapse model, proposed in the 1990s and widely cited as a reconceptualization of how neurons and astrocytes communicate, was oversold relative to what the data could support. Liddelow himself has been a careful voice in this space, and his explicit statement of uncertainty is worth taking seriously.

What is less speculative is the therapeutic angle. Astrocyte gap junctions are already a credible target for neurodegeneration. Liddelow maintains financial interests in AstronauTx and Synapticure, companies working on Alzheimer's and dementia care. The paper notes these interests are unrelated to the current study, managed under NYU Langone's conflict policies. But readers should know the landscape: any finding positioning astrocyte gap junctions as central to brain-wide signaling will attract commercial attention, and the distance from mouse network architecture to human therapeutic is long.

The tool the team built may matter as much as the finding. Current methods for studying astrocyte connectivity, including slice electrophysiology and dye diffusion, disrupt native connectivity and are limited to local environments. The AAV5-Cx43-TurboID approach with brain clearing and light-sheet imaging can be reproduced at relatively low cost. If other labs adopt it, the field will quickly know whether these networks look the same across disease models, developmental stages, and species.

Liddelow and Chao plan to investigate which molecules move through the networks and apply the tool to models of brain disorders. They want to examine how the webs change during development and aging. That program of work is sensible and will be worth watching. The immediate claim, that astrocytes form their own organized brain-wide signaling infrastructure distinct from neuronal networks, is robustly supported by the data and hard to walk back.

The neuron-centric model of brain connectivity is not wrong. But it is incomplete in a way that matters for anyone trying to understand or manipulate neural circuits. A second network has been hiding in plain sight, and now there is a map.