We propose sequences of events that are consistent with three key scenarios found in our experimental data, one in which oligomycin triggers persistent oscillations (top), one in which cells reach stable adaptation at a lower level of ATP (middle), and one in which cells reach stable adaptation at a high level of ATP (bottom). Top panel shows the condition of EGF-stimulated cells in high glucose, which display strong oscillations upon oligomycin treatment (see Figure 5A, first panel). In state (0), glucose metabolism operates at a moderately high level, and ATP levels are high. (1) When oligomycin is added, ATP production by oxidative phosphorylation is blocked, and ATP levels begin to fall; AMPK activity increases. (2) Positive feedback increases the rate of flux into upper glycolysis. (3) Increased glycolytic flux increases ATP production and AMPK activity is reduced; pools of TCA intermediates in the mitochondria become saturated. (4) Negative feedback regulation of glycolysis is triggered both by high ATP levels and by buildup of citrate in the TCA cycle, leading to reduction of glycolytic flux. (5) ATP levels fall as a result of reduced glycolysis; TCA saturation dissipates through usage of TCA intermediates in other pathways and proton gradient leakage (together indicated as ‘leak’ flux). In this state, glycolytic flux is insufficient to maintain ATP levels in the continuing presence of oligomycin, and the cell returns to state (1). The middle panel indicates the condition of EGF stimulation in the presence of intermediate levels of glucose (~2 mM; see Figure 5A, third panel). (0) External glucose levels are not limiting in the basal state, and cells are able to maintain moderately high ATP levels. (1) Oligomycin blocks ATP production by oxidative phosphorylation, and ATP levels fall. (2) Positive feedback regulation increases the rate of flux through glycolysis; however, relative to the top panel, lower available glucose levels limit the extent to which glycolytic flux can increase. (3) Increased glycolytic flux leads to a moderate rise in ATP levels, but this increase is smaller than in the top model, and negative feedback to glycolysis is not triggered. This state persists stably as glycolysis continues to operate at a rate sufficient to supply some ATP in the continuing presence of oligomycin, but not enough to return to initial levels. The bottom panel shows the condition of insulin-treated cells in the presence of high glucose and glutamine (see Figure 5D, lower right panel). In state (0), high Akt activity induced by insulin stimulates a high level of glycolytic flux and high ATP levels. (1) When oligomycin is added, ATP levels drop moderately, but not as far as in the top panels due to the higher glycolytic flux; AMPK activity increases moderately. (2) Positive feedback due to the loss of ATP stimulates further increase in glycolytic flux. (3) Increased glycolytic flux replenishes ATP and increases pools of TCA metabolites. (4) Negative feedback to upper glycolysis is triggered by an increase in ATP, but is counteracted by the insulin-stimulated high activity of Akt; glycolytic flux remains high. Meanwhile, glutamine entering the TCA cycle through anaplerotic reactions alleviates the negative feedback by allowing citrate to continue through the TCA cycle and preventing its buildup. Therefore, cells continue to be able to maintain a high level of glycolytic flux, providing high levels of ATP even in the absence of oxidative phosphorylation.