Information encoding by spatiotemporal patterns of dopamine release

Project Summary The activity of midbrain dopamine (DA) neurons, and bulk DA release in the ventromedial striatum (VMS), encodes reward prediction errors (RPE). However, there is also evidence that DA activity reflects value- orthogonal aspects of learning, including value-less sensory prediction errors (SPE). Reconciling these two bodies of data would greatly advance our understanding of DA transmission in normal and pathological conditions. One proposition is that DA release might signal a multi-factorial prediction-error. This would explain recent findings demonstrating that DA neuron ensembles encode the sensory properties of outcomes that violate learned predictions. This requires that this information, observable at the level of neuronal firing, be transmitted to downstream regions via DA release. This proposal will test the hypothesis that DA neurons encode a multi- dimensional error signal by examining specific spatiotemporal patterns of DA release in downstream regions. The current proposal takes advantage of a new methodological approach that I developed during my postdoc to image DA release in freely moving rats across a wide area at cellular resolution, as well as new technologies for gene editing and neuronal manipulation. Rats will perform an odor-guided choice task that induces both RPEs and SPEs, while DA release in the VMS will be recorded using dLight1.2 combined with GRIN lens implants and the UCLA miniscope. These recordings will be analyzed with machine learning algorithms to determine whether the specific spatiotemporal pattern of DA release encodes both value-based and value-orthogonal information (Aim 1). It will also be examined if these signals may be degraded in a pathological condition, by performing recordings in rats with chronic cocaine use experience (Aim 2). For the independent phase of the proposal, I will first examine the role of upstream cortical circuits in the encoding of this information by using chemogenetics to silence the orbitofrontal cortex (OFC) (Aim 3). I will then focus on how a specific biophysical property of DA neurons may shape this information encoding, by using gene editing techniques to delete a key K+ modulator in DA neurons while imaging DA signals in the proposed task (Aim 4). During the mentored phase of the current proposal, I will receive training critical for my short- and long-term success, including analysis of high-dimensional neural data, genetic editing using CRISPR-Cas9, and rodent self-administration models of drug use. The proposed training program combines hands-on training with expert experimenters, independent study, formal coursework, establishment of an independent collaboration, and professional scientific meetings. This program will equip me to lead a laboratory focused on the interaction of behavioral, computational, and molecular approaches to study the function of neural circuits.