Astrocytes, the brain's most abundant glial cells, were recently shown to form anatomically specific, plastic networks coupled by gap junctions that interconnect defined brain regions, including across hemispheres, in patterns distinct from known neuronal projections (Cooper et al., Nature, 2026). That study mapped the networks' structure with a connexin-43–TurboID viral tracer but left their function entirely open. The open problem is to determine what these networks do: identify by mass spectrometry the endogenous molecules that are actually transferred through astrocyte gap-junctional networks in vivo; establish whether the networks' physiological role is redistribution of metabolic and antioxidant support (candidate molecules discussed in the source: glutathione and phosphocreatine) from low-activity to high-demand regions, dissipation of accumulated metabolic byproducts in pathology, or something else; characterize the dynamics and non-uniform distribution of fluxed molecules, the distinct roles of local versus long-distance segments of the same interconnected chain, and their dependence on astrocyte molecular subtype; and establish the timescales and triggers of network reorganization, including whether acute energetic stress drives rapid reconfiguration consistent with the 1.5–5 hour half-life of connexin 43, within the source's broader program-level direction of determining how networks are reshaped by injury, disease, development, ageing, and experience-dependent processes such as learning and memory.

Background

Astrocytes are star-shaped glial cells that tile the entire central nervous system and support neurons metabolically and homeostatically. Neighboring astrocytes are extensively coupled through gap junctions — membrane channel complexes that directly join the cytoplasm of adjacent cells, each built from six connexin protein subunits — forming an interconnected cell network known as a syncytium. Astrocyte gap junctions are formed mainly by connexin 43 (Cx43, encoded by the gene Gja1) and connexin 30 (Cx30), and their channels pass ions and small molecules (roughly below 1–2 kDa) directly from cell to cell. Genetic studies show this coupling is functionally important: decoupling astrocytes in adult mice impairs synaptic plasticity and spatial learning (Hösli et al., Cell Reports, 2022), and astrocyte gap junctions are required for closing the visual critical period (the developmental window of heightened cortical plasticity; Ribot et al., Science, 2021). Astroglial metabolic networks sustain hippocampal synaptic transmission (Rouach et al., Science, 2008), and in a neurodegeneration model astrocyte networks redistribute metabolic resources to protect stressed neurons (Cooper et al., PNAS, 2020). Despite this, the spatial architecture of astrocyte networks in the intact brain was unknown: existing methods, such as dye diffusion in acute brain slices or reporter spreading in injury models, are confined to local environments and disrupt native connectivity, leaving it unclear whether astrocytes form one continuous brain-wide syncytium or discrete subnetworks, and whether their connectivity mirrors neuronal pathways. Cooper et al. (Nature, 2026) resolved this anatomical question with a viral tracer: an adeno-associated vector expressing, under a shortened Gfap promoter, a fusion of Cx43 with TurboID (a promiscuous biotin ligase enzyme that tags nearby molecules with small biotin tags). When the fusion protein incorporates into a gap junction channel, molecules that flux through the channel into neighboring astrocytes are biotinylated; because the mouse brain has almost no native biotin, streptavidin staining after whole-brain tissue clearing (chemical rendering of the intact brain optically transparent) with light-sheet microscopy reveals complete, intact three-dimensional astrocyte networks in awake, behaving mice. The study found that multiple discrete astrocyte networks traverse the mouse brain; that they selectively connect specific regions rather than diffusing indiscriminately, with local networks confined to single regions and long-range networks robustly interconnecting multiple regions across hemispheres, often in patterns distinct from known neuronal networks; and that networks are plastic, shrinking in the barrel cortex (the whisker representation of somatosensory cortex, with one barrel per whisker) after one month of whisker trimming. The authors emphasize that the method reports connectivity through a binary lens (connected or not), and their Discussion explicitly frames the functional questions that follow as unresolved: the physiological roles of the networks remain speculative (hypothesized redistribution of metabolic and antioxidant support such as glutathione, about 307 Da, and phosphocreatine, both small enough to pass Cx43 channels, or dissipation of locally accumulated metabolic byproducts during pathology); the identities of the specific transferred molecules await mass-spectrometric identification; molecule flux dynamics are likely non-uniform, with local and long-distance networks possibly serving distinct roles even within one interconnected chain, shaped by the molecular identity and subtype of each astrocyte; the short half-life of Cx43 (estimated 1.5–5 hours) and rapid astrocyte process remodeling make acute-timescale network reorganization plausible but untested (the one-month sensory deprivation paradigm cannot probe it); and how networks are shaped by injury, disease, development, ageing, and learning and memory remains unexplored. The study established that the tracer signal depends on gap-junctional coupling, using adult-inducible Cx43/Cx30 conditional knockout, and proposed two further transfer-specificity controls that were not executed: an N-terminal TID–Cx43 fusion expected to block the channel and unfused cytoplasmic TID.

Problem Statement

Overarching objective: determine the physiological function of the newly mapped astrocyte gap-junctional networks — that is, which endogenous molecules these networks transfer in the living mouse brain, and what those transfers accomplish. Building on Cooper et al. (Nature, 2026), the problem comprises the following linked questions, which the source states as open and which jointly characterize network function:

  1. Cargo identity: identify, by mass spectrometry, the specific endogenous molecules that are transferred through astrocyte gap-junctional networks in vivo (the biotin-tagged cargo of the Cx43–TurboID tracer). Glutathione and phosphocreatine are hypotheses about what the networks may carry, not an established cargo list; the target is which endogenous molecules the networks actually transfer.
  2. Physiological roles: test what the transferred molecules do — in particular whether, under physiological conditions, the networks redistribute metabolic and antioxidant support (candidates discussed in the source: glutathione, about 307 Da, and phosphocreatine) from regions of low neuronal activity to regions of high demand, whether, during pathology, they dissipate locally accumulated metabolic byproducts or pathogenic molecules by redistributing them for degradation or clearance, or whether the networks' physiological role is something else.
  3. Flux dynamics and network segmentation: determine the dynamics of different fluxed molecules, whose distribution across networks is expected to be non-uniform — including whether local and long-distance segments of the same interconnected chain of astrocytes serve distinct physiological roles, and how these dynamics are shaped by the molecular identity and subtype of each astrocyte in the network.
  4. Reorganization timescales and state-dependence: establish whether acute perturbations such as energetic stress trigger rapid network reorganization (consistent with the reported 1.5–5 hour half-life of Cx43 and rapid astrocyte process remodeling, and inaccessible to the one-month sensory-deprivation paradigm used). The source frames the accompanying state-dependence agenda — how network structure and function are shaped by injury, disease, development, ageing, and experience-dependent processes such as learning and memory — as a broad program-level direction for future exploration rather than as a single bounded question.

References

  1. Melissa L. Cooper, Maria Clara Selles, Michael Cammer, Chase Redd, Holly K. Gildea, Joseph Sall, Katelyn E. Chiurri, Philip Cheung, Damian G. Wheeler, Aiman S. Saab, Shane A. Liddelow, Moses V. Chao. Astrocytes connect specific brain regions through plastic networks. Nature 655(8121):183–191 (2026). DOI: 10.1038/s41586-026-10426-6. https://doi.org/10.1038/s41586-026-10426-6
  2. Melissa L. Cooper, Maria Clara Selles, Michael Cammer, Holly K. Gildea, Joseph Sall, Katelyn E. Chiurri, Aiman S. Saab, Shane A. Liddelow, Moses V. Chao. Astrocytes connect specific brain regions through plastic gap junctional networks. bioRxiv preprint (2025). DOI: 10.1101/2025.07.18.665573. https://doi.org/10.1101/2025.07.18.665573
  3. Melissa L. Cooper, Silvia Pasini, Wendi S. Lambert, Karis B. D'Alessandro, Vincent Yao, Michael L. Risner, David J. Calkins. Redistribution of metabolic resources through astrocyte networks mitigates neurodegenerative stress. Proceedings of the National Academy of Sciences 117(31):18810–18821 (2020). DOI: 10.1073/pnas.2009425117. https://doi.org/10.1073/pnas.2009425117
  4. Nathalie Rouach, Annette Koulakoff, Veronica Abudara, Klaus Willecke, Christian Giaume. Astroglial metabolic networks sustain hippocampal synaptic transmission. Science 322(5907):1551–1555 (2008). DOI: 10.1126/science.1164022. https://doi.org/10.1126/science.1164022
  5. Ladina Hösli, Noemi Binini, Kim David Ferrari, Laetitia Thieren, Zoe J. Looser, Marc Zuend, Henri S. Zanker, Stewart Berry, Martin Holub, Wiebke Möbius, Torben Ruhwedel, Klaus-Armin Nave, Christian Giaume, Bruno Weber, Aiman S. Saab. Decoupling astrocytes in adult mice impairs synaptic plasticity and spatial learning. Cell Reports 38(10):110484 (2022). DOI: 10.1016/j.celrep.2022.110484. https://doi.org/10.1016/j.celrep.2022.110484
  6. Jérôme Ribot, Rachel Breton, Charles-Félix Calvo, Julien Moulard, Pascal Ezan, Jonathan Zapata, Kevin Samama, Matthieu Moreau, Alexis-Pierre Bemelmans, Valentin Sabatet, Florent Dingli, Damarys Loew, Chantal Milleret, Pierre Billuart, Glenn Dallérac, Nathalie Rouach. Astrocytes close the mouse critical period for visual plasticity. Science 373(6550):77–81 (2021). DOI: 10.1126/science.abf5273. https://doi.org/10.1126/science.abf5273
  7. Tess C. Branon, Justin A. Bosch, Ariana D. Sanchez, Namrata D. Udeshi, Tanya Svinkina, Steven A. Carr, Jessica L. Feldman, Norbert Perrimon, Alice Y. Ting. Efficient proximity labeling in living cells and organisms with TurboID. Nature Biotechnology 36(9):880–887 (2018). DOI: 10.1038/nbt.4201. https://doi.org/10.1038/nbt.4201
  8. Jia-Qi Zhang, Wei Sun, Wei-Lin Jin. The astrocytic connectome: Gap-Junction topology beyond diffuse syncytium. Trends in Cell Biology 36(8):629–631 (2026). DOI: 10.1016/j.tcb.2026.05.005. https://doi.org/10.1016/j.tcb.2026.05.005
  9. Leanne M. Holt, Eric J. Nestler, Michelle L. Olsen. Cellular and molecular mechanisms of astrocyte plasticity in learning and memory. Trends in Neurosciences 49(5):344–357 (2026). DOI: 10.1016/j.tins.2026.03.001. https://doi.org/10.1016/j.tins.2026.03.001
  10. João Baltar, Rebecca Abati, Luísa Florido, Matthew G. Holt. Frozen in Place: Proximity Labeling Maps Glial Interactomes Across Cell States. Glia 74(10):e70216 (2026). DOI: 10.1002/glia.70216. https://doi.org/10.1002/glia.70216
  11. Katherine Bourzac. New-found brain network is a 'secret system' made of helper cells. Nature news (2026). DOI: 10.1038/d41586-026-01338-6. https://doi.org/10.1038/d41586-026-01338-6