Nobody notices the handshake until it slips. A neuron firing in the hippocampus does not make all of its own fuel; an astrocyte wrapped around it hands over lactate, an oligodendrocyte down the axon burns ATP to keep the myelin insulation intact, and a microglial cell patrols the neighbourhood on a metabolism it can reconfigure within hours. That constant trade is so ordinary that most models of dementia have looked past it, preferring to ask which single cell type broke first. A narrative review published online on June 1, 2026 in Mechanisms of Ageing and Development, by Kai Wang, Yongchao Liu, Haojia Zhang, Xiaomin Li, Rui Zhou, Wei Shao and Zijin Sun of Beijing University of Chinese Medicine and the Wangjing Hospital of the China Academy of Chinese Medical Sciences, makes the opposite case: ageing of the brain is better read as a systemic collapse of metabolic crosstalk between neurons and glia, and the thing governing that collapse is the shape and turnover of mitochondria.
Mitochondrial dynamics is the unglamorous housekeeping that decides whether a cell's power plants fuse into efficient networks, split into fragments, or get hauled off and recycled by mitophagy. The review's argument is that this housekeeping is the control layer for the whole multicellular network, and that when it drifts, each cell type fails in its own characteristic way. Senescent astrocytes, the authors write, disrupt the astrocyte-neuron lactate shuttle, the supply line that keeps hard-working neurons solvent. Oligodendrocytes develop ATP deficits, and myelin begins to break down. Microglia undergo what the authors call maladaptive immunometabolism and metabolic reprogramming, driven by excessive Drp1-mediated mitochondrial fission, and the result is chronic neuroinflammation that both starts and amplifies neurodegenerative damage. This is the grammar of the thing: not one broken word but a sentence losing its agreement.
The most striking element is the rescue mechanism. The review highlights intercellular mitochondrial transfer, in which glial cells donate working mitochondria to stressed neurons. That is not speculation invented for the review; it rests on experimental work such as the 2016 Nature paper by Kazuhide Hayakawa and colleagues at Massachusetts General Hospital and Harvard Medical School, which reported astrocytes releasing functional mitochondria that entered neurons after stroke in mice, with supporting co-culture experiments in rodent cells. It was contested at the time: a commentary in Cell Metabolism in September 2016 argued the transfer interpretation should be read with caution. A decade on, the idea has hardened enough that a 2026 review can build a framework around it, which is how cell biology usually moves.
What does that mean for a family sitting with a diagnosis? Not a treatment, and nothing to change tomorrow. It means the research question is shifting from "why do neurons die" to "why did the neighbourhood stop feeding them", and that shift has practical consequences for what gets measured and what gets targeted. If the lactate shuttle and the myelin energy budget are the early casualties, then the slow, diffuse changes many families describe long before a formal diagnosis, the loss of speed, the loss of stamina for complicated thought, may be a supply problem as much as a structural one. The review closes by synthesising emerging strategies aimed at what the authors term the glia-neuron mitochondrial social network, framed as a holistic approach to restoring brain bioenergetic balance. Emerging is the operative word: these are directions of travel described in a review article, not therapies with results in people.
It is worth being precise about what kind of document this is. A narrative review is a synthesis and an argument, written by researchers who select and interpret the literature; it enrolls no patients, runs for no months, and reports no new measurements. There is no cohort here, no trial arm, no number of participants, because there was never meant to be one. Its value is as a map, and its weakness is the same as any map: it can be drawn confidently over terrain that has only been surveyed in mice, in flies and in dishes. The neuron-glia uncoupling story is now coherent enough to be worth testing in people. It has not yet been tested in people.