Dorsal Raphe 5-HT Neurons Utilize, But Do Not Generate, Negative Aversive Prediction Errors

Abstract

The dorsal raphe nucleus (DRN) contains the largest population of serotonin (5-HT) neurons in the central nervous system. 5-HT, synthesized via tryptophan hydroxylase 2 (Tph2), is a widely functioning neuromodula-tor implicated in fear learning. Here, we sought to investigate whether DRN 5-HT is necessary to reduce fear via negative prediction error (–PE). Using male and female TPH2-cre rats, DRNtph2+ cells were selectively deleted via cre-caspase (rAAV5-Flex-taCasp3-TEVp) in experiment 1. Rats then underwent fear discrimination during which three cues were associated with unique foot shock probabilities: safety p = 0.00, uncertainty p = 0.375, and danger p = 1.00. Rats then received selective extinction to the uncertainty cue, a behavioral ma-nipulation designed to probe –PE. Deleting DRNtph2+ cells had no impact on initial discrimination but slowed selective extinction. In experiment 2, we used a within-subjects optogenetic inhibition design to causally impli-cate DRNtph2+ cells in prediction error signaling. Male and female TPH2-cre rats received intra-DRN infusions of cre-dependent halorhodopsin (rAAV5-Ef1a-DIO-eNpHR3.0-eYFP) or cre-YFP. DRNtph2+ cells were inhibited specifically during the time of prediction error or a control period. Illumination during either positive prediction error (+PE) or control periods had no impact on fear to the uncertainty cue. Inhibition of DRNtph2+ cells at the time of –PE did not impact immediate fear, but facilitated selective extinction in postillumination sessions. Together, these results demonstrate a role for DRNtph2+ cells in using, but not generating, –PE to weaken cue-shock associations.