Finding. Goldfish horizontal cells turn down their glutamate receptors in hypoxia, and they need a mitochondrial potassium channel to do it. Ramadan, Jonz and colleagues patch dissociated horizontal cells, the inhibitory interneurons that eat photoreceptor glutamate and give GABA back. Twenty minutes of hypoxia cuts the peak glutamate inward current about 25% (P = 0.0059). Block mitochondrial ATP-sensitive potassium channels with 100 micromolar glibenclamide and the cut disappears. Block mitochondrial calcium uptake with ruthenium red, or ER calcium release through ryanodine receptors, or calmodulin, or PP1/PP2A, and the cut disappears the same way. GABA release, measured by mass spectrometry, does not change. The authors' reading: mKATP starts a calcium signal that dephosphorylates or otherwise quiets iGluRs, demand falls, the cell keeps working, and the GABA tap stays open.
Why this paper matters
Hypoxia-tolerant retinas are a demand-management problem. Cutting excitatory current while keeping inhibitory output is a specific solution. Putting mKATP, the mitochondrial calcium uniporter, and PP1/PP2A on one path is a circuit you can steal for other ischemia-tolerant neurons. It is also a reminder that glibenclamide, a diabetes drug, can erase a mitochondrial hypoxic adaptation in a dish.
What they actually measured
Perforated patch, a short hypoxia, a clean inhibitor panel, GABA mass spec. No ATP assay, no intact-loop ERG.
How to read the score
Low 80s. Named mitochondrial channel, a number, a calcium path, a spared transmitter. Confidence is high for the dissociated-cell pharmacology.
Caveats
Goldfish. Dissociated. Glibenclamide promiscuity. Demand-lowering is an inference.
What to do with it
If you study retinal hypoxia, add mKATP and MCU to the iGluR protocol and measure GABA separately. If you use glibenclamide in ischemia models, watch for erased adaptations. Do not export 25% as a human number.