Rapid eye movement (REM) sleep has been hypothesized to promote emotional resilience, but any neuronal circuits mediating this have not been identified. We find that in mice, somatostatin (Som) neurons in the entopeduncular nucleus (EPSom)/internal globus pallidus are predominantly active during REM sleep. This unique REM activity is both necessary and sufficient for maintaining normal REM sleep. Inhibiting or exciting EPSom neurons reduced or increased REM sleep duration, respectively. Activation of the sole downstream target of EPSom neurons, Vglut2 cells in the lateral habenula (LHb), increased sleep via the ventral tegmental area (VTA). A simple chemogenetic scheme to periodically inhibit the LHb over 4 days selectively removed a significant amount of cumulative REM sleep. Chronic, but not acute, REM reduction correlated with mice becoming anxious and more sensitive to aversive stimuli. Therefore, we suggest that cumulative REM sleep, in part generated by the EP → LHb → VTA circuit identified here, could contribute to stabilizing reactions to habitual aversive stimuli.

The ventral pallidum (VP) is a central node in the reward system that is strongly implicated in reward and addiction. Although the majority of VP neurons are GABAergic and encode reward, recent studies revealed a novel glutamatergic neuronal population in the VP [VP neurons expressing the vesicular glutamate transporter 2 (VPVGluT2)], whose activation generates aversion. Withdrawal from drugs has been shown to induce drastic synaptic changes in neuronal populations associated with reward, such as the ventral tegmental area (VTA) or nucleus accumbens neurons, but less is known about cocaine-induced synaptic changes in neurons classically linked with aversion. Here, we demonstrate that VPVGluT2 neurons contact different targets with different intensities, and that cocaine conditioned place preference (CPP) training followed by abstinence selectively potentiates their synapses on targets that encode aversion. Using whole-cell patch-clamp recordings combined with optogenetics in male and female transgenic mice, we show that VPVGluT2 neurons preferentially contact aversion-related neurons, such as lateral habenula neurons and VTA GABAergic neurons, with minor input to reward-related neurons, such as VTA dopamine and VP GABA neurons. Moreover, after cocaine CPP and abstinence, the VPVGluT2 input to the aversion-related structures is potentiated, whereas the input to the reward-related structures is depressed. Thus, cocaine CPP followed by abstinence may allow VPVGluT2 neurons to recruit aversion-related targets more readily and therefore be part of the mechanism underlying the aversive symptoms seen after withdrawal.SIGNIFICANCE STATEMENT The biggest problem in drug addiction is the high propensity to relapse. One central driver for relapse events is the negative aversive symptoms experienced by addicts during withdrawal. In this work, we propose a possible mechanism for the intensification of aversive feelings after withdrawal that involves the glutamatergic neurons of the ventral pallidum. We show not only that these neurons are most strongly connected to aversive targets, such as the lateral habenula, but also that, after abstinence, their synapses on aversive targets are strengthened, whereas the synapses on other rewarding targets are weakened. These data illustrate how after abstinence from cocaine, aversive pathways change in a manner that may contribute to relapse.

Cocaine-driven changes in the modulation of neurotransmission by neuromodulators are poorly understood. The ventral pallidum (VP) is a key structure in the reward system, in which GABA neurotransmission is regulated by opioid neuropeptides, including dynorphin. However, it is not known whether dynorphin acts differently on different cell types in the VP and whether its effects are altered by withdrawal from cocaine. Here, we trained wild-type, D1-Cre, A2A-Cre, or vGluT2-Cre:Ai9 male and female mice in a cocaine conditioned place preference protocol followed by 2 weeks of abstinence, and then recorded GABAergic synaptic input evoked either electrically or optogenetically onto identified VP neurons before and after applying dynorphin. We found that after cocaine CPP and abstinence dynorphin attenuated inhibitory input to VPGABA neurons through a postsynaptic mechanism. This effect was absent in saline mice. Furthermore, this effect was seen specifically on the inputs from nucleus accumbens medium spiny neurons expressing either the D1 or the D2 dopamine receptor. Unlike its effect on VPGABA neurons, dynorphin surprisingly potentiated the inhibitory input on VPvGluT2 neurons, but this effect was abolished after cocaine CPP and abstinence. Thus, dynorphin has contrasting influences on GABA input to VPGABA and VPvGluT2 neurons and these influences are affected differentially by cocaine CPP and abstinence. Collectively, our data suggest a role for dynorphin in withdrawal through its actions in the VP. As VPGABA and VPvGluT2 neurons have contrasting effects on drug-seeking behavior, our data may indicate a complex role for dynorphin in withdrawal from cocaine.SIGNIFICANCE STATEMENT The ventral pallidum consists mainly of GABAergic reward-promoting neurons, but it also encloses a subgroup of aversion-promoting glutamatergic neurons. Dynorphin, an opioid neuropeptide abundant in the ventral pallidum, shows differential modulation of GABA input to GABAergic and glutamatergic pallidal neurons and may therefore affect both the rewarding and aversive aspects of withdrawal. Indeed, abstinence after repeated exposure to cocaine alters dynorphin actions in a cell-type-specific manner; after abstinence dynorphin suppresses the inhibitory drive on the \"rewarding\" GABAergic neurons but ceases to modulate the inhibitory drive on the \"aversive\" glutamatergic neurons. This reflects a complex role for dynorphin in cocaine reward and abstinence.