Activation, but not inhibition, of the indirect pathway disrupts choice rejection in a freely moving, multiple-choice foraging task

Abstract

The dorsomedial striatum (DMS) plays a key role in action selection, but less is known about how direct and indirect pathway spiny projection neurons (dSPNs and iSPNs, respectively) contribute to choice rejection in freely moving animals. Here, we use pathway-specific chemogenetic manipulation during a serial choice foraging task to test the role of dSPNs and iSPNs in learned choice rejection. We find that chemogenetic activation, but not inhibition, of iSPNs disrupts rejection of nonrewarded choices, contrary to predictions of a simple "select/suppress" heuristic. Our findings suggest that iSPNs' role in stopping and freezing does not extend in a simple fashion to choice rejection in an ethological, freely moving context. These data may provide insights critical for the successful design of interventions for addiction or other conditions in which it is desirable to strengthen choice rejection.

Keywords: CP: Neuroscience; choice rejection; decision making; direct pathway; dorsomedial striatum; explore-exploit; indirect pathway; reinforcement learning; striatum.

Figure 1.. Select/suppress heuristic predictions for SPN manipulation in odor-guided serial choice task

(A) Task schematic. (B) Odor choices made during acquisition (top) and test (bottom) phases from representative mouse. Vertical bars indicate odor choice on single trial. (C) Putative activity patterns of dSPN and iSPN ensembles for each odor choice, illustrating basic assumptions of simple select/suppress heuristic model. (D) Select/suppress heuristic predictions for dSPN and iSPN manipulation during test phase. (C) is based on Extended Data Figure 10 in Parker et al., (2018).

Figure 2.. Establishing efficacy of DREADDS in DMS iSPNs and dSPNs

(A) Unilateral chemogenetic inhibition of dSPNs versus iSPNs had opposite effects on rotation bias: dSPN inhibition drove a significant ipsiversive bias (*p = 0.02), and iSPN inhibition drove a significant contraversive bias (***p < 0.0001). Unilateral chemogenetic activation of iSPNs drove a significant ipsiversive bias (***p < 0.0001) (n sessions/N mice = 8/4, 10/5, and 6/3). (B) D2-Cre mice were co-transduced with Cre-dependent hM4Di-mCherry and Cre-dependent ChR2-EYFP into DMS. Scale bar: 1 mm. (C) Sagittal slice containing globus pallidus externa (GPe) targeted for patch-clamp recording. (D) Cell-attached recording configuration. Asterisk indicates raster for raw trace above. (E) Peristimulus spike histogram; blue light stimulus significantly reduced spike rate (n cells/N mice = 6/3). (F) Whole-cell recording configuration: sample evoked inhibitory postsynaptic current (eIPSC) before and after CNO (10 μM) wash on. (G) Normalized eIPSC amplitude before and after CNO wash on (***p < 0.0001) (n cells/N mice = 6/3). (H) mCherry+ cells were targeted for whole-cell current clamp recording in D2-Cre mice transduced with Cre-dependent hM3Dq-mCherry or mCherry virus. CNO (10 μM) was bath applied in the presence of TTX (0.5 μM). (I) CNO depolarized hM3Dq-mCherry+, but not mCherry+, cells. (***p < 0.0001) (n cells/N mice = 5/3, 6/4). (J) Top panel: representative images show lack of co-localization of Cre-dependent mCherry (red) and choline acetyltransferase (ChAT) immunoreactivity (green) within DMS of AAV8-hSyn-DIO-mCherry-transduced D2-Cre mice (N = 4). White arrows indicate ChAT+ neurons. Bottom panel: rare co-localization of Cre-dependent mCherry (red) and ChAT (green). White arrow indicates ChAT+/mCherry+ neuron. (K) Proportion of ChAT+/mCherry+ neurons within regions of AAV8-hSyn-DIO-mCherry infection. Mean ± SEM shown in (G) and (I). See full statistics in Table S2.

Figure 3.. Chemogenetic activation of iSPNs and inhibition of dSPNs both impaired test-phase performance, while inhibition of iSPNs had no significant effects

(A) Top panel: injection site and viral spread for D2-Cre DIO-mCherry (N = 21), DIO-hM4Di (N = 12), and DIO-hM3Dq (N = 11) mice. Bottom panel: summary of behavior. (B) Acquisition (saline) choices to criterion. (C) Effect of odor identity (**p < 0.001) and virus (p = 0.50) on nonrewarded choices. (D) Test (CNO) choices to criterion (**p < 0.01). (E) Test nonrewarded choices (*p < 0.05). (F) Test choices to the training-naive preferred odor (O2) (**p < 0.01). (G) Test reward accumulation (**p < 0.01). (H) Top panel: Injection site and viral spread for D1-Cre DIO-mCherry (N = 10) and DIO-hM4Di (N = 6) mice. Bottom panel: summary of behavior. (I) Acquisition (saline) choices to criterion (p = 0.99 Mann-Whitney U test). (J) Effect of odor identity (***p < 0.001) and virus (p = 0.25) on nonrewarded choices. (K) Test choices to criterion (**p < 0.01). (L) Test nonrewarded choices (*p < 0.05). (M) Test choices to the training-naive preferred odor (O2) (*p < 0.05). (N) Test reward accumulation (***p < 0.0001). Mean ± SEM are plotted for normally distributed data. Otherwise, data plotted indicate median ± interquartile range (IQR). Data in (G) and (N) indicate linear regression line with bands plotting the 95% confidence interval. See full statistics in Table S2.

Figure 4.. Enhancing iSPN activity and reducing dSPN activity increased choice stochasticity but reduced physical serial exploration of choices before decision

(A) RL modeling of acquisition and test performance in D2-Cre (top) and D1-Cre (bottom) groups. (B) Left panel: no effect of chemogenetic manipulation on $\text{Δ}\alpha$ in D2-Cre (top) or D1-Cre (bottom) groups (p > 0.05). Right panel: chemogenetic activation of iSPNs (green) significantly increased $\text{Δ}\beta$ compared to mCherry control and chemogenetic inhibition of iSPNs (gray). Chemogenetic inhibition of iSPNs did not significantly change $\text{Δ}\beta$ compared to mCherry (top). Chemogenetic inhibition of dSPNs (orange) increased $\text{Δ}\beta$ compared to mCherry control (bottom). (C) Chemogenetic activation of iSPNs (top) (N = 21, 12, 11) or chemogenetic inhibition of dSPNs (bottom) (N = 10, 6) is associated with increased proportion of single entry choices. *p < 0.05, **p < 0.01. Hypothesis tests were conducted using Kruskal-Wallis ANOVA or Mann Whitney U test for behavior and sample-based credible interval for model parameters. See full statistics in Table S2.