Single Neurons Can Induce Phase Transitions of Cortical Recurrent Networks with Multiple Internal States

Abstract

Fluctuations of membrane potential of cortical neurons, referred to here as internal states, are essential for brain function, but little is known about how these internal states emerge and are maintained, or what determines transitions between these states. We performed intracellular recordings from hippocampal CA3 pyramidal cells ex vivo and found that neurons display multiple and hierarchical internal states, which are linked to cholinergic activity and are characterized by several power law structures in membrane potential dynamics. Multiple recordings from adjacent neurons revealed that the internal states were coherent between neurons, indicating that the internal state of any given cell in a local network could represent the network activity state. Repeated stimulation of single neurons led over time to transitions to different internal states in both the stimulated neuron and neighboring neurons. Thus, single-cell activation is sufficient to shift the state of the entire local network. As the states shift to more active levels, theta- and gamma-frequency components developed in the form of subthreshold oscillations. State transitions were associated with changes in membrane conductance but were not accompanied by a change in reversal potential. These data suggest that the recurrent network organizes the internal states of individual neurons into synchronization through network activity with balanced excitation and inhibition, and that this organization is discrete, heterogeneous and dynamic in nature. Thus, neuronal states reflect the 'phase' of an active network, a novel demonstration of the dynamics and flexibility of cortical microcircuitry.