Our limited understanding of neuronal co-transmission complicates predicting the effects of pharmacological interventions from neuronal wiring diagrams [1, 2]. For example, dopamine neurons (DAN), whose degeneration causes the motor and cognitive deficits of Parkinson's Disease (PD) [3], communicate with all their targets by dopamine (DA) and in addition use the neurotransmitters GABA, Glutamate and secreted peptides including the morphogen Sonic Hedgehog (Shh) to signal to select subsets of neurons [4][5][6][7][8]. It is unknown whether Levodopa (L-Dopa) induced dyskinesia (LID), a debilitating side effect of DA supplementation therapy in PD [9], might appear because DAN targets become exposed to high levels of DA but diminished levels of other DAN produced signaling factors. Here we show that restoring the balance of DA and Shh signaling by agonists of the Shh signaling effector G-protein coupled receptor Smoothened (Smo [10, 11]) attenuates LID in mouse-and macaque-models of PD. Using conditional genetic loss of function approaches of pre-and post-synaptic Shh signaling we found that reducing the activity of Smo selectively in cholinergic neurons facilitates LID. Conversely, the expression of a constitutive active form of Smo (SmoM2 [12]) in cholinergic neurons is sufficient to render the sensitized aphakia model of PD resistant to LID. Furthermore, the acute diminishment of Shh emanating from DAN by optogenetic means in the otherwise intact brain and in the absence of L-Dopa results in LID-like abnormal involuntary movements. We reveal that Shh co-transmission counteracts DA action in a DAN target-and striatal domain-specific manner. Our results suggest an un-anticipated mechanism -reduced Shh signaling -by which DAN degeneration sensitizes PD patients to LID. We anticipate that our models provide a starting point for assessing the contributions of Shh to the "teaching" signal that emanates from DAN and testing the hypothesis that LID are caused by aberrant learning [13][14][15].
Main:Previous work has implicated aberrant cholinergic activity in LID in PD models but mechanistic details remain ill-defined since both boosting as well as inhibiting cholinergic signaling were found to attenuate LID [16][17][18][19][20][21][22]. These divergent results are likely a reflection of the post synaptic complexity of cholinergic signaling in the striatum [23]. Hence, normalizing cholinergic activity at the level of CIN themselves might prevent the formation of LID and preserve the contribution of regulated CIN activity to striatal function. In the healthy brain, DAN engage CIN as a relay hub to control striatal circuits by patterning cholinergic activity [24,25]: DA inhibits CIN through binding to D2 receptor (D2R) resulting in the pause of the "burst-pause-burst" firing pattern of CIN in response to reward-related cues or outcomes [26,27] [28,29]. Unfortunately, how DAN degeneration results in aberrant CIN activity is also not well understood [30]. For example, as DA levels fall in the PD brain, cholinergic signaling s...