Efficient Identification and Dissection of Key Neural Circuits Controlling Behavior Across Different Time Scales

Abstract

A fundamental question in neuroscience is how neural networks generate behavior. The lack of genetic tools and unique promoters to functionally manipulate specific neuronal subtypes makes it challenging to determine the roles of individual subtypes in behavior. I describe a compressed sensing-based framework in combination with non-specific genetic tools to infer key neurons controlling behaviors with fewer measurements than previously thought possible. I first tested this framework by inferring interneuron subtypes regulating the speed of locomotion of the nematode Caenorhabditis elegans. Then, I demonstrated that this method could identify key neurons in an artificial recurrent neural network, suggesting its validity in nonlinear networks (Chapter 1). We developed a real-time stabilization microscope for accurate long-term, high-magnification imaging and targeted perturbation of neural activity in freely moving animals to validate our inferences. We show that a circuit of three interconnected interneuron subtypes, RMG, AVB, and SIA, controls different aspects of locomotion speed at different timescales as the animal navigates its environment (Chapter 2). Our work suggests that compressed sensing approaches can be used to identify key nodes in complex biological networks.

Besides fast motor actions, animals exhibit long-lived behavioral states. To understand neural circuit mechanisms underlying state transition between these persistent behavioral states, I investigated local to global foraging states in C. elegans (Chapter 3). Using long-term calcium imaging and optogenetic perturbations of neural dynamics, I discovered slow dynamics in the sets of interneurons AIY and RIM controls the transition from local to global search states. In contrast, fast modulating neurons encode the sub-behavioral component at short time scales. I demonstrated that this core circuit could be tuned by upstream glutamate and insulin-like neuropeptide receptor activity to change the dynamics of the transition from one state to another.