During brain anoxia or ischaemia a fall of ATP level leads to a sudden run-down of transmembrane ion gradients (the anoxic depolarization or AD). This releases glutamate by reversed uptake (Rossi et al. 2000), which triggers the death of neurons. By whole-cell clamping pyramidal cells (using a KCl-based, BAPTA and ATP containing internal solution) to sense the rise of [glutamate]_o accompanying the AD, we have investigated the energetic factors controlling the latency of the AD in area CA1 of hippocampal slices (from P12 rats killed humanely by cervical dislocation).

When oxygen and glucose (10 mM) were removed, glycolytic ATP production was blocked with iodoacetate (2 mM), and mitochondrial ATP production was blocked using rotenone (100 µM) and antimycin (100 µM), the AD occurred in ~7.5 mins (440 ± 17 (S.E.M.) s n = 29) at 33 °C. Allowing glycolysis fuelled by glycogen, when glucose and oxygen were removed and mitochondria were blocked, delayed the AD by a further ~5.5 mins (to 771 ± 37 s n = 8), and if external glucose was present the AD was prevented altogether (latency > 3922 ± 811 s n = 4). Allowing mitochondria to function when glycolysis was blocked increased the latency to the AD by over 6 mins (786 ± 78 s n = 13, compared with 409 ± 37 s n = 7, when mitochondria were blocked, both experiments done with oxygen present, glucose removed and iodoacetate added), suggesting that metabolites feeding mitochondria downstream of glycolysis (pyruvate and citric acid cycle intermediates) provide a significant energy store. Superfusing lactate (5 mM) in this situation did not prevent the AD occurring, and provided no extra increase in the latency to the AD (n = 7). Simply removing oxygen and glucose from the superfusion solution produced an AD after 17 mins (1021 ± 84 s n = 5). By comparison in vivo ischaemia produces an AD after ~2 mins, implying a ~3 mins latency at the 33 °C used here.

These data show that glycolysis alone can prevent the anoxic depolarization, and do not support the idea that most brain ATP production is powered by lactate supplied to neurons by glia. The long latency to the AD seen on simply removing oxygen and glucose may result from: (i) the presence of residual oxygen in superfusion solutions passing through the recording chamber, which has to be open to the air to allow electrode access; (ii) a lower metabolic rate for sliced tissue compared to tissue in vivo; (iii) the use of a non-physiologically high glucose concentration (10 mM) which gives the tissue an abnormally high reserve of glycogen.

NA and RK are in the 4 year PhD Programme in Neuroscience at UCL. This work was supported by the Wellcome Trust, the EU and a Wolfson-Royal Society award.