VTA dopamine neurons drive spatiotemporally heterogeneous striatal dopamine signals during learning

Abstract

Environmental cues, through Pavlovian learning, become conditioned stimuli that guide animals toward the acquisition of rewards (for example, food) by invigorating and directing seeking behavior. Dopamine neurons are crucial for this process, by signaling information related to rewards, as well as associated actions, states, and cues. Dopamine neurons are embedded in a reciprocally connected network with their targets in the striatum, the functional relevance of which remains poorly understood. Here we investigated how cue-evoked behavior generated by dopamine neurons is signaled by dopamine release downstream in striatal subregions and how they evolve across stages of learning. We made use of a genetically encoded dopamine biosensor (dLight) to monitor dopamine signaling in the nucleus accumbens (NAc) core and shell, dorsomedial striatum (DMS), and dorsolateral striatum (DLS) with fiber photometry, while tracking detailed movement features during optogenetic Pavlovian cue conditioning of VTA dopamine neurons. We demonstrate spatially heterogeneous, learning-dependent changes across striatal regions. Cues paired with optogenetic activation of VTA dopamine neurons evoked dopamine release preferentially in the NAc core, but not shell, early in training, when behavioral responses were slower and directed toward the cue. Critically, these NAc core signals remained stable, and as training progressed, VTA cue-evoked dopamine signals also emerged in the DMS. Further, prediction error signals, reflecting a violation in the expectation of dopamine neuron activation, only emerged in the NAc core and DMS. Despite the development of rapid, vigorous movement late in training, conditioned dopamine signals did not emerge in the DLS. Together our studies show broad dissociation in the fundamental prediction and reward-related information signaled by dopamine across the striatum. Further, they offer new insight into how larger-scale plasticity across the striatal dopamine network emerges during Pavlovian learning to coordinate behavior.