Cancer as a Neural Circuit Disease

# Neuronal activity feeds glioma growth *Evidence guide: Primary preclinical results and mechanistic interpretation. Dependencies have scope, escape routes, and possible normal-cell effects.* In 2015, the Monje Lab asked: does neuronal activity change how fast a glioma grows? Humsa Venkatesh and colleagues placed patient-derived high-grade glioma cells, including pediatric glioblastoma and DIPG, into the premotor cortex of immunodeficient mice whose cortical neurons carried channelrhodopsin. Stimulating the neurons with light in awake mice raised the fraction of dividing glioma cells from about seven to about ten percent. A week of daily stimulation increased tumor burden in the active circuit by about forty percent, while unstimulated regions showed no difference. Protein analysis and mass spectrometry pointed to a leading candidate: neuroligin-3. Neuroligins are normally synaptic adhesion proteins that help hold synapses together. Here, a released fragment was in the active fluid, and on its own was enough to drive glioma proliferation. Neuroligin-3 was not the only factor. Trapping it only partly reduced the effect, and two other secreted proteins, BDNF and GRP78, also promoted proliferation, less potently. Inside glioma cells, it activated the PI3K–mTOR growth pathway, and even boosted its own production: a feed-forward loop. Is this relevant in patients? In public data from adult glioblastoma, higher NLGN3 expression went with shorter survival. That is a correlation: it fits the mouse experiments, but cannot show cause on its own. In 2017, the lab tested necessity. In mice lacking the Nlgn3 gene, xenografts of pediatric glioblastoma, DIPG, and adult glioblastoma engrafted, then grew strikingly little for months. A breast-cancer brain-metastasis model grew regardless: not every brain tumor shares this dependency. The dependency was not absolute: by about four and a half months, a subset of tumors began growing anyway. And the growth block was stronger than known NLGN3 signaling could explain, so part of the mechanism remains unknown. NLGN3 is a powerful influence, not the sole driver. How is it released? The enzyme ADAM10 cuts it from the cell surface, and oligodendrocyte precursor cells turned out to be a major source, alongside neurons. Active neurons release ADAM10 itself. In mice, ADAM10 inhibitors reduced glioma xenograft growth: a preclinical lead for a drug target. But what does NLGN3 bind on a glioma cell? Its synaptic partners, neurexins, are not expressed in these cells. A 2026 Monje Lab study found the partner: CSPG4, also called NG2, a classic marker of oligodendrocyte precursor cells. Both shed proteins were detectable in patients’ cerebrospinal fluid. The proposed pathway is mechanical. Binding to CSPG4 increased membrane tension, which opened PIEZO1, a mechanosensitive ion channel, and the cells depolarized: the fraction showing calcium signals rose from about two to about twenty-two percent. Depolarization then triggered CSPG4 shedding and PI3K–mTOR activation. The evidence included a fluorescent membrane-tension sensor, electrical recordings of PIEZO1 currents, cells in which CSPG4 or PIEZO1 was knocked out, and DIPG xenografts lacking PIEZO1, which proliferated less in the mouse pons. The striking part: the same pathway keeps healthy OPCs in an undifferentiated, progenitor state. A normal developmental signal is co-opted by the cancer, which also means blocking it could affect healthy OPCs. Sources: [venkatesh2015](https://pmc.ncbi.nlm.nih.gov/articles/PMC4447122/), [venkatesh2017](https://pmc.ncbi.nlm.nih.gov/articles/PMC5891832/), [kim2026](https://doi.org/10.1038/s41593-026-02397-8).