Differential recruitment of ventral pallidal e-types by behaviorally salient stimuli during Pavlovian conditioning

Abstract

The ventral pallidum (VP) is interfacing striatopallidal and limbic circuits, conveying information about salience and valence crucial to adjusting behavior. However, how VP neuron populations with distinct electrophysiological properties (e-types) represent these variables is not fully understood. Therefore, we trained mice on probabilistic Pavlovian conditioning while recording the activity of VP neurons. Many VP neurons responded to punishment (54%), reward (48%), and outcome-predicting auditory stimuli (32%), increasingly differentiating distinct outcome probabilities through learning. We identified e-types based on the presence of bursts or fast rhythmic discharges and found that non-bursting, non-rhythmic neurons were the most sensitive to reward and punishment. Some neurons exhibited distinct responses of their bursts and single spikes, suggesting a multiplexed coding scheme in the VP. Finally, we demonstrate synchronously firing neuron assemblies, particularly responsive to reinforcing stimuli. These results suggest that electrophysiologically defined e-types of the VP differentially participate in transmitting reinforcement signals during learning.

Keywords: Behavioral Neuroscience; Cellular Neuroscience; Cellular Physiology.

Graphical abstract

Figure 1

Targeting VP in mice performing auditory Pavlovian conditioning (A) Coronal section from a ChAT-Cre mouse showing the tetrode tracks (DiI, yellow; ChAT+, green) through the VP. The tetrodes were advanced 0–100 μm between recording days. Although the images show the full extent of the electrode tracks, only those sessions that were conducted strictly within VP boundaries based on post hoc histological reconstruction (see transparent methods) were included. Scale bar, 1 mm. (B) Left, magnified view of the target area. Scale bar, 500 μm. Right, confocal image of a cholinergic neuron located near the electrode track (20× magnification, z stack of 8 planes, maximal intensity projection). Scale bar, 10 μm. (C) Reconstructed location of the electrode tracks. Only neurons recorded inside the VP were included (see transparent methods, histology). (D) Top, schematic of the auditory Pavlovian task setup. Bottom, trial structure with possible outcomes. After the mouse stopped collecting the previous reward (“no lick”), a variable inter-trial interval started, signaled by turning an light-emitting diode off, in which no licking was allowed. Then two cue tones of well-separated pitch predicted likely reward or likely punishment. (E) Raster plots of licking activity in an example session of one mouse. The cue predicting likely reward (top, green) elicited stronger anticipatory licking than the cue that signaled likely punishment (bottom, red). (F) Peri-event time histograms (PETH) of licking activity from the same session. Purple arrow indicates average reinforcement delivery time (note that reinforcement time was randomized between 400 and 600 ms after cue offset according to a uniform distribution to prevent full temporal predictability of reinforcement delivery). (G) Average licking activity (PETH) of N = 5 mice shows stronger anticipatory licking after the cue that predicted likely reward (green). Data are represented as mean ± SEM. (H) The reward-predicting cue elicited significantly more licks in 4/5 mice. Data are represented as median ±SE of median. ∗∗∗p < 0.001, Wilcoxon signed rank test. CPu, caudate putamen; HDB, horizontal nucleus of the diagonal band of Broca; MS, medial septum; NAc, nucleus accumbens; VDB, vertical nucleus of the diagonal band of Broca.

Figure 2

Ventral pallidal neurons are modulated by reward, punishment, and outcome-predictive cues during Pavlovian conditioning (A–F) Example single VP neurons activated by cue stimuli (A), reward (B) or punishment (C), or inhibited by cue (D), reward (E), or punishment (F). Top, spike raster; bottom, PETH aligned to the respective behavioral events. (G–I) Pie charts showing the number of all VP neurons activated and inhibited by cue (G), reward (H), and punishment (I), pooled across mice. (J–L) Average Z-scored PETH of all recorded VP neurons modulated by cue (J), reward (K), or punishment (L), tested separately for the three events. Data are represented as mean ± SEM. (M) Excitatory response latencies to predictive cues, reward, and punishment. Box-whisker plots show median, interquartile range, non-outlier range, and outliers. ∗∗p < 0.01; ∗∗∗p < 0.001, Mann-Whitney U test. (N) Inhibitory response latencies to predictive cues, reward, and punishment. Box-whisker plots show median, interquartile range, non-outlier range, and outliers. ∗p < 0.05; ∗∗p < 0.01, ∗∗∗p < 0.001, Mann-Whitney U test. (O) Number of neurons showing different combinations of all possible responses to cues and reinforcers. +, activation; -, suppression; 0, no significant firing rate change. For instance, the color in line 1, column 1 indicates the number of neurons that were activated by all three salient events tested (cue, reward, punishment); color in line 2, column 2 indicates the number of neurons that were activated after reward but inhibited both after sound cues and punishment. See also Figures S1 and S2.

Figure 3

Ventral pallidal neurons are modulated by expectation (A) Left, average PETH of cue-activated VP neurons after cues predicting likely reward (pink) or likely punishment (purple). Right, box-whisker plot of average spike count difference between the two outcome probability conditions in VP neurons activated by cue presentations. Data are represented as mean ± SEM. ∗∗∗p < 0.001, Wilcoxon signed rank test. (B) Average PETH and spike count difference for VP neurons inhibited after cue presentations. Data are represented as mean ± SEM. ∗∗p < 0.01, Wilcoxon signed rank test. (C and D) Same as in (A and B), but showing responses to reward presentations. Left, reward-activated VP neurons; right, reward-suppressed VP neurons. Pink, expected reward (reward delivery after the likely reward cue); purple, surprising reward (reward delivery after the likely punishment cue). Data are represented as mean ± SEM. ∗p < 0.05, Wilcoxon signed rank test. (E and F) Same as in (A and B), but showing responses to punishment presentations. Left, punishment-activated VP neurons; right, punishment-suppressed VP neurons. Pink, surprising punishment (punishment delivery after the likely reward cue); purple, expected punishment (punishment delivery after the likely punishment cue). Data are represented as mean ± SEM. Recordings with less than 5 trials in either of the tested conditions were excluded. Box-whisker pots show median, interquartile range, and non-outlier range in all panels. See also Figures S1 and S2. See also Figure S1 and S2.

Figure 4

Non-bursting ventral pallidal neurons respond to reinforcers more frequently (A) Top, Example autocorrelograms of a single bursting and a non-bursting VP neuron. Bottom, average across all bursting and non-bursting neurons recorded from the VP. (B) Pie chart showing the proportion of bursting and non-bursting neurons, pooled across mice. (C) Pie charts showing the number of all bursting and non-bursting VP neurons activated and inhibited by cue, reward, or punishment. (D–I) Average, Z-scored PETHs of bursting (D–F) and non-bursting (G–I) VP neurons aligned to cue (D and G), reward (E and H), and punishment (F and I). Data are represented as mean ± SEM. See also Figures S4–S6.

Figure 5

Salient events mostly recruit non-rhythmic neurons (A) Example autocorrelogram of a rhythmically firing neuron. (B) Pie chart showing the proportion of rhythmic and non-rhythmic neurons, pooled across mice. (C) Pie charts showing the number of rhythmic and non-rhythmic VP neurons activated or inhibited by cue, reward, or punishment, pooled across mice. (D–I) Average, Z-scored PETHs of rhythmic (D–F) and non-non-rhythmic (G–I) VP neurons aligned to cue (D and G), reward (E and H), and punishment (F and I). Data are represented as mean ± SEM. See also Figure S7.

Figure 6

Dissociation of single spike and burst firing in a subset of VP neurons (A) Example PETH of a VP neuron activated upon cue onset. (B and C) The neuron showed increased burst (B) and single spike firing (C) after the cue. (D–F) (D) Example of another cue-activated neuron, which showed decreased burst firing (E) and increased single spike firing (F) after cue presentation. (G) Pie chart showing the number of neurons responding with significant rate change of both bursts and single spikes after cue (left), reward (middle), or punishment (right), pooled across mice. Insert, number of neurons where a significant change was detected for both bursts and single spikes (green). (H–J) Average Z-scored PETH of burst and single spike responses of neurons where bursts showed an increase, whereas single spikes showed a decrease after cue (H), reward (I), or punishment (J). Data are represented as mean ± SEM. See also Figure S8.

Figure 7

VP neurons form co-firing assemblies (A) An example cell assembly formed by VP neurons during Pavlovian conditioning. Brown, narrow (1–2 ms) zero-lag synchrony; pink, broad (≥3 ms) zero-lag synchrony; blue, putative monosynaptic excitation. (B) Example cross-correlograms of neuronal pairs of this assembly. Gray, 95% confidence intervals.

Figure 8

Synchronous neurons respond more to reinforcement (A) Top, example cross-correlogram of synchronously activated neurons. Bottom, average cross-correlation of all synchronously active pairs. (B) Pie chart showing the proportion of synchronous and asynchronous VP neurons, pooled across mice. (C) Pie charts showing the number of synchronous and asynchronous VP neurons activated or inhibited by cue, reward or punishment, pooled across mice. (D–I) Average, Z-scored PETHs of synchronous (D–F) and asynchronous (G–I) VP neurons aligned to cue (D and G), reward (E and H), and punishment (F and I). Data are represented as mean ± SEM. See also Figure S9.

Figure 9

Comparison of ventromedial (VPvm) and lateral (VPl) subregions (A) Fluorescent images of anti-ChAT (red), anti-NT (green), and anti-SP (magenta) triple immunohistochemical staining. Left, coronal section of a hemisphere. Scale bar, 1,000 μm. Middle, VP area. Scale bar, 500 μm. Right, high-magnification images of VP cholinergic neurons. Scale bars, 20 and 50 μm. (B) Pie charts showing the proportion of VPvm versus VPl neurons within all recorded VP neurons and broken down to the different e-types examined. (C) Pie charts showing the proportion of bursting versus non-bursting, rhythmic versus non-rhythmic, and synchronous versus asynchronous neurons across the VPvm and VPl subpopulations. (D) Pie charts showing the proportion of neurons activated or suppressed after presentation of the cue, the reward, or the punishment. Bursting versus non-bursting neurons are shown for VPvm and VPl separately. (E–G) Correlation of the burst index (E), beta-rhythmicity index (F), and gamma-rhythmicity index (G) with dorsoventral position of recorded VP neurons. Burst index showed significant negative (p < 0.001) and beta-rhythmicity index showed significant positive correlation (p = 0.01) with dorsoventral position.