Influence of neuromodulators on brain state transitions in larval zebrafish

With the rapid growth of optical imaging technologies, the simultaneous observation of brain-wide activity and behavior in small organisms offers a unique opportunity to study how distributed neuronal networks interact to process sensory information and generate behavior. While neuromodulators have been shown to grant flexible state-switching mechanisms to circuits, their influence on global network dynamics is poorly characterized. Using resonant-scanning two-photon microscopy, we measured brain-wide calcium dynamics in head-restrained larval zebrafish while monitoring tail movements to investigate how dopamine, norepinephrine and serotonin drive neuronal and behavioral states. We first developed an unsupervised clustering approach to extract recurrent regional activation patterns, or brain states, across all fish and recording sessions. The approach identifies short-lived states, as well as states lasting over tens of seconds, the majority of which having no behavioral manifestation. The spatial distribution of state activity is modular and constrained by structural communities of strongly interconnected regions. Transition probabilities between states are conserved across the population, while state dwell times vary. Interestingly, the identity of individual fish is recovered across imaging sessions using a subset of fingerprint states. To investigate how neuromodulators might be driving these states and individual differences, we use post hoc immunolabeling and image registration to identify cell types of interest in calcium imaging datasets, from which we extract neuromodulatory signaling. Robust correlations are observed between tail movements and the firing of dopaminergic/noradrenergic cells, recapitulating their well-studied role in driving locomotor networks. Our current framework will be expanded to different sensorimotor contexts using visual stimulation, while controlling neuromodulators with pharmacology. The receptor distributions will be characterized with fluorescence in situ hybridization to constrain recurrent neural network models with neuromodulatory interactions. We aim to give a brain-wide description of how neuromodulators interact and lead to the emergence of recurrent macroscopic states, hallmarks of multiple imaging modalities.