How do neurons coordinate alternative energy sources to meet
the demands of neural computation?
The brain is energetically expensive, a metabolic cost that is intrinsic to neural activity and hence a
defining feature of how the brain computes. As a result of this energy intensive operation, the main methods
for measuring changes in neural activity in humans, such as functional magnetic resonance imaging (fMRI),
actually infer neural activity by measuring changes in blood flow, a proxy for local energy consumption.
Moreover, many diseases that alter the efficiency and balance of energy production are characterized by
profound deficits in brain function. However, how neural activity shapes energy production at the level of
individual cells, circuits and across the brain are only incompletely understood, particularly in the context of
active sensation and behavior.
Longstanding work in the field, based in vitro models of single cells and human neuroimaging, have
revealed how different pathways for energy production react to changes in neural activity, responding when
increases in neural activity cause depletion of ATP, a core cellular energy currency. Our recent work using the
intact brain of the behaving fruit fly build on these results, and revealed a new element to the coupling between
metabolism and energy production, namely that cells use current levels of neural activity to predict future
energy needs. Thus, this project seeks to answer how the reactive and predictive elements of neural-metabolic
energy coupling interact.
The proposed work focuses on three key questions. First, do different neuron types, with distinct
patterns of activity in the intact brain, display differences in how they react to, and predict, metabolic load?
Second, how do neurons balance energy production via two alternative energy sources, namely glycolysis and
oxidative phosphorylation, to both react to metabolic cost and predict future expenditures? Finally, how are
these metabolic loads coordinated across circuits in behaving animals detecting sensory stimuli? We
hypothesize that because neuronal activity levels differ substantially across cell types, and because glycolysis
and oxidative phosphorylation can produce ATP with different latencies and efficiencies, subcellular
compartments, neurons and circuits dynamically switch between alternative energy sources to both react to
computational demand and predict future metabolic need. To test this hypothesis, we propose to use two
photon imaging of fluorescent sensors of neural activity and metabolic flux, combined with genetic and
optogenetic perturbations of specific cell types, using the adult fruit fly brain as a model.
As many of the genes involved in energy metabolism are evolutionarily conserved between humans
and flies, deepening our understanding of how neural activity couples to energy metabolism in vivo will
increasing our understanding of the neural impacts of metabolic diseases, possibly opening new therapeutic
avenues for future investigation.