Finally, Zachary Kilpatrick expanded a classic neural circuit model of temporal-difference learning to account for glu-DA neurons, as well as the individual contributions of glutamate and dopamine from glu-DA and DA-only neurons. The model recapitulated some of our key behavioral results. 20/n

…But blocking glutamate release did not cause gross motor deficits which might explain the differences, nor impair escape responses to a simulated predator. We concluded that blocking glutamate release impairs reward- and exploration-related vigor without affecting aversion-related vigor. 19/n

Blocking glutamate release did not impact the acquisition of reward learning or fear learning/extinction. But it slowed reward retrieval latency and enhanced reward extinction, and suppressed exploratory behavior in an open field… 18/n

Knocking down TH (to block DA synthesis) didn’t impact acquisition of reward learning. But mice had deficits when learning about less-rewarding (reward CS-, reward extinction) or aversive (shock CS+) associations. 17/n

We zoomed in on glu-DA co-transmitting neurons and identified distinct functional roles for glu release and DA release in motivated behaviors. Using recombinase-dependent shRNA vectors, we blocked molecules necessary for DA synthesis (in glu neurons) or vesicular glu release (in DA neurons) 16/n

Anatomically, we also observed (as others have before) that glu-DA neurons heavily innervate NAcc medial shell, with some innervation in medial core. DA-only neurons heavily innervate NAcc core with some innervation in ventral and lateral shell. 15/n

We used ChRmine & GRABDA to measure opto-elicited NAcc DA release from VTA glu-DA or DA-only terminals. Axonal stim in both subpops increased DA release. Glu-DA axons were more sensitive to longer pulse trains in phasic magnitude. DA-only were more sensitive to longer trains in tonic duration 14/n

Differences in signaling dynamics were starker in an aversive context. Glu-DA and glu-only neurons were activated to shock and the shock-predictive cue—glu-DA neurons strongly so. When measuring the minimum GCaMP value, we saw that DA-only neurons were suppressed at shock and cue. 11/n

Interestingly, DA-only neurons responded to the neutral cue (CS-) depending on the mouse’s behavior. When the mouse correctly avoided the port, activity decreased at the cue. But when the mouse incorrectly entered the port, activity decreased at the time reward “would” have come on a CS+ trial. 9/n

When an expected reward was omitted (reward prediction error), sustained activity of glu-DA neurons prevented a suppression below baseline, while transient activity of DA-only and glu-only allowed RPE signaling. This is the first RPE signaling identified in a VTA glutamate subpopulation. 8/n

In Pavlovian reward, all subpops were activated at reward and reward-predictive cue. But glu-DA neurons exhibited sustained activation between cue and reward peaks, while DA-only and glu-only exhibited transient activation in which activity dropped toward baseline between cue and reward peaks. 7/n

The incredible @ampolter.bsky.social and Kailyn Price demonstrated that these subpops are electrophysiologically distinct. Some highlights: a subset of glutamate-only neurons exhibited Ih and D2 sensitivity; glutamate-dopamine neurons largely lacked Ih. 5/n

Additionally, we used a subtractive genetic strategy to target (for the first time) neurons that do not release dopamine and do not release GABA, which in the VTA selects for neurons that release glutamate alone. Our cellular targeting strategy is summarized here. 4/n

Honored to receive first place poster award at #wcbr2025! Thanks to all who stopped by. The poster (and my panel) explored divergent signaling dynamics in VTA dopamine subpopulations, as well as functional roles for glutamate and dopamine in co-transmitting VTA neurons. Stay tuned for more!