Genetically encoded nanosensors such as the FRET-based adenosine triphosphate (ATP) sensor ATeam enable the measurement of changes in ATP levels inside cells, promoting our understanding of metabolic interactions between astrocytes and neurons. The sensors are usually well characterized in vitro but display altered properties when expressed inside cells, precluding a meaningful conversion of changes in FRET ratios into changes in intracellular ATP concentrations ([ATP]) on the basis of their in vitro properties.
Here, we present an experimental strategy for the intracellular calibration of two different variants of ATeam in organotypic tissue slice culture of the mouse brain. After cell-type-specific expression of the sensors in astrocytes or neurons, slices were first perfused with a saline containing the saponin β-escin to permeabilize plasma membranes for ATP.
Next, cells were depleted of ATP by perfusion with ATP-free saline containing metabolic inhibitors. Finally, ATP was re-added at defined concentrations and resulting changes in the FRET ratio recorded. When employing this protocol, ATeam1.03 expressed in astrocytes reliably responds to changes in [ATP], exhibiting an apparent KD of 9.4 mM. The high-affinity sensor ATeam1.03YEMK displayed a significantly lower intracellular KD of 2.7 mM.
On the basis of these calibrations, we found that induction of a recurrent neuronal network activity resulted in an initial transient increase in astrocytic [ATP] by ~0.12 mM as detected by ATeam1.03YEMK, a result confirmed using ATeam1.03. In neurons, in contrast, [ATP] immediately started to decline upon initiation of a network activity, amounting to a decrease by an average of 0.29 mM after 2 min. Taken together, our results demonstrate that ATeam1.03YEMK and ATeam1.03 display a significant increase in their apparent KD when expressed inside cells as compared with in vitro.
Moreover, they show that both ATeam variants enable the quantitative detection of changes of astrocytic and neuronal [ATP] in the physiological range. ATeam1.03YEMK, however, seems preferable because its KD is close to baseline ATP levels. Finally, our data support the idea that synchronized neuronal activity initially stimulates the generation of ATP in astrocytes, presumably through increased glycolysis, whereas ATP levels in neurons decline.