The dose of CNO was chosen based on a previous study reporting high in vivo efficacy at doses between 1 and 5 mg/kg on locomotor activity and stereotypy in a transgenic mouse expressing the hM3Dq receptor within the forebrain (Alexander et al., 2009). CNO reduced the firing rate of a substantial portion

of the MD units (Figures 2B–2D). To quantify this effect, we first INK1197 cost calculated the ratio of firing rates after saline and CNO for each unit; the distribution of ratios in the sample was significantly different than that expected by chance (p < 0.01, Wilcoxon signed-rank test) (Figure 2C). We next compared firing rates after saline and CNO injections for each neuron independently. CNO decreased firing rate significantly (p < 0.05 by paired t test) in 30 neurons (48%). In an additional 11 units (17%), CNO increased the firing rate. Using a more stringent Bonferroni correction (p < 0.0008) 16 (25%) and 6 (9.5%) units were inhibited or activated

by CNO, respectively. For both analyses neurons with decreased firing rate were overrepresented as a consequence of CNO treatment (Binomial test: p < 0.01, p < 0.05 for Bonferroni-corrected values). Importantly, CNO treatment did not completely silence MD neurons; the neurons with decreased firing rate showed an average decrease of 38.7% ± 5.3%. This decrease in firing rate was not related to changes in locomotor activity or to differences in the isolation of single units (Figures S2A–S2C). The CNO-mediated decrease in firing rate was not observed HIF-1 activation in wild-type mice that do not express hM4D, demonstrating its dependence on hM4D (Figure 2C, inset). While hyperpolarization of thalamic cells can induce a shift in the firing pattern from a tonic to a bursting mode due to the activation of T-type Ca2+ channels (Jahnsen and Llinás, 1984), we did not observe a significant change in the fraction of burst firing in vivo after CNO injection (Figure S2D). Due to the strong projections of the MD to the orbitofrontal cortex (OFC) (Figure 1D), we first tested whether decreasing MD activity affects reversal learning, a cognitive process of executive the function that is sensitive to orbitofrontal lesions (Schoenbaum et al., 2002). To address reversal

learning, we developed an operant-based reversal learning task for the mouse in which lever presses are rewarded in the presence of one visual cue (S+), but not in the presence of another visual cue (S−) (discrimination phase). After mice reached criterion the contingencies were reversed (reversal phase) (Figure 3A). The acquisition of the discrimination phase was not affected by decreasing MD activity (repeated ANOVA p = 0.17) though CNO-treated MDhM4D mice showed a tendency for a delay in learning during the first three days of acquisition (Figure 3B). In contrast, reversal learning was clearly impaired in CNO-treated MDhM4D mice when compared to the three control groups (repeated ANOVA followed by Bonferroni correction for group comparisons ∗p < 0.05, ∗∗p < 0.