Dopamine dynamics are dispensable for movement but promote reward responses

Abstract

Dopamine signalling modes differ in kinetics and spatial patterns of receptor activation^1,2. How these modes contribute to motor function, motivation and learning has long been debated^3-21. Here we show that action-potential-induced dopamine release is dispensable for movement initiation but supports reward-oriented behaviour. We generated mice with dopamine-neuron-specific knockout of the release site organizer protein RIM to disrupt action-potential-induced dopamine release. In these mice, rapid in vivo dopamine dynamics were strongly impaired, but baseline dopamine persisted and fully supported spontaneous movement. Conversely, reserpine-mediated dopamine depletion or blockade of dopamine receptors disrupted movement initiation. The dopamine precursor L-DOPA reversed reserpine-induced bradykinesia without restoring fast dopamine dynamics, a result that substantiated the conclusion that these dynamics are dispensable for movement initiation. In contrast to spontaneous movement, reward-oriented behaviour was impaired in dopamine-neuron-specific RIM knockout mice. In conditioned place preference and two-odour discrimination tasks, the mice effectively learned to distinguish the cues, which indicates that reward-based learning persists after RIM ablation. However, the performance vigour was reduced. During probabilistic cue-reward association, dopamine dynamics and conditioned responses assessed through anticipatory licking were disrupted. These results demonstrate that action-potential-induced dopamine release is dispensable for motor function and subsecond precision of movement initiation but promotes motivation and performance during reward-guided behaviours.

Extended data figure 1.. Slice amperometry in dorsal striatum and additional movement analyses of RIM cKO^DA mice.

a. Schematic of slice amperometry in subareas of the dorsal striatum. b, c. Example traces (b) and quantification (c) of peak amplitudes of dopamine release evoked by electrical stimulation before and after 1 μM tetrodotoxin (TTX), 16 slices from 4 mice. d-e. Example traces (d) and quantification (e) of peak amplitudes as in b+c, RIM control 9 slices from 4 mice, RIM cKO^DA 10/4. f-i. As d+e, but in the other subareas shown in a, RIM control 10/4, RIM cKO^DA 10/4. j-t. Quantification of parameters of gait, RIM control 1122 cycles from 10 mice, RIM cKO^DA 1206/10. u-z. Schematics (u, w, y) and analyses (v, x, z) of horizontal bar, vertical bar and Rotarod (after four days of training) tests, RIM control 7 mice, RIM cKO^DA 7. aa-ab. Representative trajectories (aa) and quantification of distance traveled in 30 min (ab) before (day 1) and after i.p. injection of PBS (day 2) and DMSO (day 4), RIM control 4, RIM cKO^DA 3. ac, ad. Schematic (ac) of unilateral 6-OHDA lesions followed by analyses of rotations before and after i.p. injection of the D1 and D2 receptor agonist apomorphine in 6-OHDA pre-treated mice (1 mg/kg) and quantification of net contralateral rotations (ad), RIM control 6, RIM cKO^DA 6. ae, af. Schematic of videography (ae) and analyses of distance traveled in 30 min (af) for the mice in Fig. 1p–t, RIM control 13, RIM cKO^DA 14. Some of the data are from the baseline condition shown in Fig. 1m and these data points are replotted here. Data are mean ± SEM; *** p < 0.001, assessed by: two-sided Mann-Whitney rank-sum tests for c, e, g, i, j-r, v, x, af; two-way ANOVA for z, ad; and Kruskal-Wallis analysis of variance with post-hoc Dunn’s tests for ab. For videos of gait, see Supplementary video 1.

Extended data figure 2.. Pharmacological inhibition of locomotion by drug infusion and assessment of excitation power output.

a. Schematic of assessment of movement with bilateral drug infusion in dorsal striatum. b, c. Representative trajectories (b) and quantification of total distance traveled in 15 min (c) before and after local infusion of ACSF or D1 (SCH23390, 20 μM, 1 μl for each site) and D2 (haloperidol, 40 μM, 1 μl for each site) receptor antagonists, 4 mice. d, e. Example traces (d) and quantification (e) of fluorescence variation quantified as standard deviation (SD) of ΔF/F₀ of GRAB_DA fluorescence at variable output power in freely moving mice, 7 mice. f, g. As in d+e, but for tdTomato, 7 mice; d-g reveal that ΔF/F₀ variation is similar across excitation output powers. h. Schematic of the measurement of dopamine dynamics in freely moving mice. Fiber photometry and drug delivery were in the right dorsal striatum with an optofluid cannula. i-l. Example traces (i+j) and quantification of the variation of ΔF/F₀ of GRAB_DA (k) and of tdTomato (l) fluorescence before and after local infusion of the sodium channel blocker TTX (500 nM, 1 μl), 6 mice. Data are mean ± SEM; * p < 0.05, ** p < 0.01, assessed by two-sided Mann-Whitney rank-sum tests for c, k; Kruskal-Wallis analysis of variance with post-hoc Dunn’s tests were used for e, g.

Extended data figure 3.. Haloperidol injection and additional analyses of GRAB_DA dynamics in RIM cKO^DA mice.

a-c. Example traces (a) and quantification of the variation of ΔF/F₀ of GRAB_DA (b) and of tdTomato (c) fluorescence before and after i.p. injection of the D2 receptor antagonist haloperidol (2 mg/kg), RIM control 5 mice, RIM cKO^DA 5. d. Average GRAB_DA and tdTomato signals registered to the artificially shifted instantaneous velocity plotted in polar coordinates for the experiment shown in Fig. 2g+h. The shifting of the velocity time course to earlier or later time points relative to the photometry illustrates that the GRAB_DA fluorescence signal peaks after the velocity and suggests that dopamine signaling tracks the velocity time course, RIM control 5, RIM cKO^DA 5. e. Analyses of distance traveled for the experiment shown in Fig. 2, n as in a–c. f. Quantification of tdTomato fluorescence during contralateral movement initiations shown in Fig. 2i+j, event heatmaps are sorted by the order of the corresponding velocity signals in Fig. 2i, RIM control 354 events from 5 mice, RIM cKO^DA 455/5. g-i. Quantification of time courses of velocity amplitudes (g), and of GRAB_DA (h) and tdTomato (i) fluorescence changes during ipsilateral movement initiations (right turns, velocity angles between 180° and 360°). Event heatmaps were sorted by the peak velocity amplitude in g, RIM control 405/5, RIM cKO^DA 469/5. Data are mean ± SEM; ** p < 0.01, assessed by two-sided Mann-Whitney rank-sum tests for b, e.

Extended data figure 4.. In vivo GCaMP6s fluctuations in RIM cKO^DA mice.

a. Strategy for dual-color fiber photometry of dopamine axonal GCaMP6s and tdTomato. b, c. Example traces (b) and quantification (c) of fluorescence variation as SD of ΔF/F₀ of GCaMP6s and of tdTomato in freely moving mice, RIM control 6 mice, RIM cKO^DA 6. d-f. Example traces (d) and quantification of fluorescence variation as SD of ΔF/F₀ of GCaMP6s (e) and of tdTomato (f) before and after i.p. injection of the D1 and D2 receptor agonist apomorphine (1 mg/kg). The suppression of GCaMP6s fluorescence changes by apomorphine indicates that GCaMP6s fluorescence changes reflect activity-dependent depolarizations of dopamine axons that are inhibited by D2 auto-receptors, n as in b+c. g-i. Time course of fluorescence in individual trials (event heatmaps, top) and average data (bottom) for GCaMP6s (g) and tdTomato (h) fluorescence aligned to the sensory stimulation (dashed line), and peak GCaMP6s per mouse (i). Event heatmaps in g+h were sorted by the peak amplitude in g, RIM control 93 events from 6 mice, RIM cKO^DA 93/6. The finding that axonal Ca²⁺ dynamics are not detectably changed in RIM cKO^DA mice is unexpected given RIMs role in targeting Ca_V2 channels to presynaptic active zones. It could be due to distinct RIM functions in dopamine neuron axons^,, due to a distinct set of Ca²⁺ channels in dopamine axons^,, due to compensatory effects in RIM cKO^DA mice because of loss of the D2 auto-receptor feedback, due to technical differences in experiments, or due to Ca²⁺ entry away from active zones and not under the control of RIM, which might be enhanced if Ca_V2 channels are mislocalized. Data are mean ± SEM; ** p < 0.01, assessed by two-sided Mann-Whitney rank-sum tests for c, e, i.

Extended data figure 5.. Assessment of dopamine axon GCaMP6s and striatal RdLight1 dynamics in RIM cKO^DA mice.

a. Schematic for assessment of fluorescence of GCaMP6s expressed in striatal dopamine axons and RdLight1 expressed in striatal neurons. b, c. Example traces (b) and cross-correlation (c) of GCaMP6s and RdLight1 signals, RIM control 4 mice, RIM cKO^DA 4. d, e. Individual (event heatmaps, top) and average (bottom) time courses of GCaMP6s (d) and RdLight1 (e) fluorescence aligned to the light flash (dashed line). Event heatmaps in d+e were sorted by the peak amplitude in d, RIM control 204 events from 4 mice, RIM cKO^DA 202/4. f. Correlation analyses of transients in d+e (area under the curve, 0 to 600 ms after stimulation), RIM control 203/4, RIM cKO^DA 198/4. g-i. Individual (event heatmaps, top) and average (bottom) time courses of velocity amplitudes (g), and of GCaMP6s (h) and RdLight1 (i) fluorescence during contralateral movement initiations. Event heatmaps in g were sorted by the peak velocity amplitude, and h+i by the peak amplitude of GCaMP6s in h, RIM control 563/4, RIM cKO^DA 599/4. j-n. Analyses of peaks from d, e, and g-i for each mouse, n as in b+c. There was a strong positive correlation in RIM control mice between GCaMP6s and RdLight1 fluctuations that was disrupted in RIM cKO^DA mice (c). Field illuminations induced dopamine axonal GCaMP6s transients in both genotypes, but failed to trigger dopamine release in RIM cKO^DA mice (d-f). Axonal Ca²⁺ dynamics and striatal dopamine fluctuations correlated during contralateral turns in RIM control mice. In RIM cKO^DA mice, only Ca²⁺ transients, not movement-associated dopamine transients, were detected (g-i). Together with Extended data fig. 4, the data suggest that dopamine neuron firing and the underlying regulatory network are not strongly disrupted in RIM cKO^DA mice. Data are mean ± SEM; * p < 0.05, *** p < 0.001, assessed by: two-sided Mann-Whitney rank-sum tests for areas under the curve in c, j-n.

Extended data figure 6.. Ablating the Ca²⁺ sensor Syt-1 in dopamine neurons does not disrupt in vivo dopamine dynamics or locomotor behaviors.

a-c. Analyses of motor behaviors as in Extended data fig. 1u–z, but for Syt-1 control and Syt-1 cKO^DA mice. For dopamine release analyses in Syt-1 cKO^DA, see. The time spent to cross a horizontal bar (a), to climb down a vertical bar (b), or the latency to fall from a Rotarod after four days of training (c) are quantified, Syt-1 control 10 mice, Syt-1 cKO^DA 10. d-i. In vivo fiber photometry performed as in Fig. 2 and Extended data fig. 3, but for Syt-1 cKO^DA mice with example traces (d) and quantification of variation of ΔF/F₀ of GRAB_DA (e) and of tdTomato (f) fluorescence before and after i.p. injection of the D2 receptor antagonist haloperidol (2 mg/kg), and with individual (event heatmaps, top) and average (bottom) time courses of GRAB_DA (g) and tdTomato (h) fluorescence aligned to the sensory stimulation (dashed line) and peak GRAB_DA per mouse (i). Event heatmaps are sorted by the peak GRAB_DA amplitude in g; Syt-1 control 200 events from 4 mice, Syt-1 cKO^DA 200/4. Altogether, knockout of Syt-1 from dopamine neurons did not disrupt motor function. Despite the strong impairment in dopamine release in brain slices^,, in vivo dopamine fluctuations were maintained, likely due to the remaining release after Syt-1 knockout, presumably asynchronous release, that is detected with in vivo microdialysis or in brain slices after dopamine transporter (DAT) blockade (striatum), or in response to stimulus trains (somatodendritic release). Hence, removing the fast Ca²⁺ sensor from dopamine neurons does not suffice to abolish in vivo dopamine dynamics and Syt-1 cKO^DA mice cannot be used to test behavioral roles of these dynamics. Data are mean ± SEM; * p < 0.05, assessed by: two-sided Mann-Whitney rank-sum test for a, b, e, i; two-way ANOVA for c.

Extended data figure 7.. Analyses of GRAB_DA F₀ and locomotion after L-DOPA treatment in reserpine-depleted mice.

a. Schematic of the experiment. b, c. Assessment of GRAB_DA F₀ (b) and quantification of distance traveled (c, analyzed in 900 s bins for the first 900 s and from 950–9950 s) before and after i.p. injection of L-DOPA (250 mg/kg L-dopa methyl ester with 25 mg/kg carbidopa) in mice treated with reserpine (3 mg/kg reserpine) ~18 h before L-DOPA injection, GRAB_DA F₀ is normalized to the first 5 min, 3 mice. Data are mean ± SEM.

Extended data figure 8.. Additional analyses L-DOPA treated mice.

a. Schematic of the experiment. b-d. Example traces (b) and quantification of the variation of ΔF/F₀ of GRAB_DA (c) and of tdTomato (d) fluorescence before and after i.p. injection of L-DOPA (250 mg/kg L-dopa methyl ester with 25 mg/kg carbidopa), 5 mice. e-g. Individual (event heatmaps, top) and average (bottom) time courses of GRAB_DA (e) and tdTomato (f) fluorescence aligned to the sensory stimulation (dashed line), and peak GRAB_DA (g) for each mouse, before and after i.p. injection of L-DOPA. Data in a-g establish that GRAB_DA fluorescence increases can be detected when L-DOPA is present without reserpine depletion, indicating that GRAB_DA fluorescence is not saturated after L-DOPA injection. Event heatmaps in e+f were sorted by the peak GRAB_DA amplitude in e, baseline 136 events from 5 mice, L-DOPA 118/5. h-j. Individual (event heatmaps, top) and average (bottom) time courses of GRAB_DA (h) and tdTomato (i) fluorescence aligned to the sensory stimulation (dashed line), and peak GRAB_DA (j) for each mouse, before and after i.p. injection of reserpine (3 mg/kg) and L-DOPA (250 mg/kg L-dopa methyl ester with 25 mg/kg carbidopa). Data were recorded during the experiment that is shown in Fig. 3b–j. Event heatmaps in h+i were sorted by the GRAB_DA peak amplitude in h, baseline 393/4, reserpine + L-DOPA 223/4. Data are mean ± SEM; * p < 0.05, assessed by two-sided Mann-Whitney rank-sum tests in c, g, j.

Extended data figure 9.. Additional analyses of GRAB_DA fluorescence during movement initiation in reserpine and L-DOPA treated mice.

a. Average GRAB_DA and tdTomato signals registered to the artificially shifted instantaneous velocity plotted in polar coordinates before and after i.p. injection of reserpine (3 mg/kg) and L-DOPA (250 mg/kg L-dopa methyl ester with 25 mg/kg carbidopa) for the experiment shown in Fig. 3b–j, 4 mice. b. Quantification of tdTomato fluorescence during contralateral movement initiations shown in Fig. 3g+h, event heatmaps are sorted by the order of the corresponding velocity signals in Fig. 3g, baseline 316 events from 4 mice, reserpine + L-DOPA 378/4. c-e. Individual (event heatmaps, top) and average (bottom) time courses of velocity amplitudes (c) and GRAB_DA (d) and tdTomato (e) fluorescence changes during ipsilateral movement initiations (right turns, velocity angles between 180° and 360°) for the experiment shown in Fig. 3b–j. Event heatmaps in c-e were sorted by the peak velocity amplitude in c, baseline 317/4, reserpine + L-DOPA 557/4. f-j. Individual (event heatmaps, top) and average (bottom) time courses of velocity amplitudes (f), and of GRAB_DA (g) and tdTomato (h) fluorescence during contralateral movement initiations (left turns, velocity angles between 0° and 180°), and peak velocity (i) and GRAB_DA (j) per mouse, for the experiment shown in Fig. 3k–n, RIM control data is replotted from Fig. 2j+l. Event heatmaps in f-h were sorted by the peak velocity amplitude in f, baseline 385/4, reserpine + L-DOPA 362/4. The observations that L-DOPA restored movement in RIM cKO^DA mice to pre-reserpine levels (f, Fig. 3k–n), and that dopamine denervation followed by apomorphine-induction of rotations was unaffected (Extended data figs. 1ac+ad), indicate that there is no strong sensitization of dopamine receptors or dopamine-modulated circuits after RIM ablation. Data are mean ± SEM; * p < 0.05, assessed by two-sided Mann-Whitney rank-sum tests in i, j.

Extended data figure 10.. GRAB_DA analyses in ventral striatum and additional analyses during the probabilistic cue-reward association task in RIM cKO^DA mice.

a. Schematic of slice imaging. b-d. Representative images (b) and quantification (c+d) of dopamine release monitored by GRAB_DA fluorescence in slices containing the ventral striatum (dashed lines outline the striatum), evoked by a single stimulus (b+c) or 10 stimuli (d, 10 Hz), RIM control 10 slices from 4 mice, RIM cKO^DA 10/4. e, f. Example traces (e) and quantification (f) of ventral striatum fluorescence variation quantified as standard deviation (SD) of GRAB_DA and of tdTomato raw fluorescence on day 1 of the task in Fig. 5, RIM control 6 mice, RIM cKO^DA 7. g, h. As in e and f, but for dorsal striatum on day 2, n as in e+f. The in vivo deficits are overall similar in dorsal and ventral striatum. There might be an enhanced GRAB_DA signal in ventral striatal brain slices compared to dorsal striatum (Fig. 1a–d) in RIM cKO^DA mice. This was not observed with amperometry, and could be because of differences in the roles of RIM, or technical differences in experiments, or because of detection of other transmitters by GRAB_DA, for example norepinephrine for which innervation is prominent in ventral but not dorsal striatum^,,. i. The number of habituation days for the experiment shown in Fig. 5. RIM control 6, RIM cKO^DA 7. j. Number of trials that the mice completed during each training phase. For analyses, only completed blocks were used, n as in i. k, l. Anticipatory licks (k) and peak licks to expected reward during the 1 s time window from water onset (l) for each odor during training days 1 to 7, n as in i. m. Average ventral striatum GRAB_DA odor responses (within 3 s from odor onset), n as in i. n-q. Average ventral striatum GRAB_DA reward responses (n, p, within 201 to 1200 ms from water onset) and reward omission responses (o, q, within 1501 to 2500 ms from water onset) for odors 1 (n+o) and 2 (p+q), n as in i. r, s. Total licks (r, within 1 to 3000 ms after water onset) and average GRAB_DA fluorescence (s, within 201 to 1200 ms after water onset) for free water, n as in i. Data are mean ± SEM; ** p < 0.01, *** p < 0.001, assessed by two-sided Mann-Whitney rank-sum tests in c, d, f, h, i, j.

Figure 1.. Normal motor function after disrupting evoked dopamine release.

a. Schematic of imaging in parasagittal slices of RIM cKO^DA and RIM control mice. b-d. Representative images (b) and quantification (c+d) of dopamine release monitored by GRAB_DA fluorescence (dashed lines outline the striatum), evoked by paired (b+c, 1 Hz) or 10 stimuli (d, 10 Hz), RIM control 8 slices from 3 mice, RIM cKO^DA 11/3. e. Schematic of gait analyses. f-h. Average speed (f) and cadence (g) across gait cycles, and scatter plot and linear regression thereof (h), RIM control R = 0.88 in 1122 cycles from 10 mice, RIM cKO^DA 0.85 in 1206/10. i, j. As f-h, but for stride length against speed, RIM control R = 0.42 in 1122/10, RIM cKO^DA 0.57 in 1206/10. k. Schematic of movement initiation analyses in a round arena. l, m. Representative trajectories (l) and quantification of distance traveled in 30 min (m) before and after i.p. injection of reserpine (2 mg/kg), RIM control 7 mice, RIM cKO^DA 8. n, o. As l+m, but for D1 (SCH23390, 1 mg/kg) and D2 (haloperidol, 2 mg/kg) receptor antagonists, RIM control 6, RIM cKO^DA 6. p-r. Individual (p) and average time courses of movement initiations (q), and peak speed per mouse (r). Event heatmaps (p) were sorted by the peak speed amplitude, RIM control 3615 events from 13 mice, RIM cKO^DA 4352/14. s, t. Frequency of peak speeds (s) and durations (t) of detected movement events above the indicated cutoffs, n as in p-r. Data are mean ± SEM; ** p < 0.01, *** p < 0.001 as assessed by: two-sided Mann-Whitney rank-sum tests in c, d, f, g, i, m, o, q, r; two-way ANOVA for s, t. For dopamine amperometry, and additional analyses of motor function, see Extended data fig. 1. Exact p-values for this and all following figures are in the Supplementary table 1.

Figure 2.. Disrupted in vivo dopamine dynamics but unaltered movement initiation in RIM cKO^DA mice.

a. Schematic of the experiment. The fiber photometric canula was in the right dorsal striatum. b, c. Example traces (b) and quantification (c) of fluorescence variation quantified as standard deviation (SD) of ΔF/F₀ of GRAB_DA and tdTomato fluorescence, RIM control 5 mice, RIM cKO^DA 5. d-f. Time course of fluorescence in individual trials (event heatmaps, top) and average data (bottom) for GRAB_DA (d) and tdTomato (e) fluorescence aligned to the sensory stimulation (dashed line), and peak GRAB_DA per mouse (f). Event heatmaps in d+e were sorted by the peak amplitude in d, RIM control 117 events from 5 mice, RIM cKO^DA 115/5. g. Schematic of movement initiation analyses as established before. h. Average GRAB_DA and tdTomato signals registered to their concurrent velocity; each velocity vector was plotted in polar coordinates, and the corresponding fluorescence changes (ΔF/F₀) were mapped onto these velocity vectors for each genotype and fluorophore. i-l. Individual (event heatmaps, top) and average (bottom) time courses of velocity amplitudes (i) and GRAB_DA fluorescence changes (j) during contralateral movement initiations (left turns, velocity angles between 0° and 180°), and peak velocity (k) and GRAB_DA (l) per mouse. Event heatmaps in i+j were sorted by the peak velocity amplitude, RIM control 354/5 mice, RIM cKO^DA 455/5. Data are mean ± SEM; ** p < 0.01, assessed by two-sided Mann-Whitney rank-sum test in c, f, k, l. For assessment of locomotion after dopamine receptor blockade, ΔF/F₀ variation as a function of excitation power output, ΔF/F₀ variation after sodium channel blockade, ΔF/F₀ variation before and after GRAB_DA blockade, additional movement-related GRAB_DA fluorescence of RIM cKO^DA mice, dopamine axon Ca²⁺ dynamics with GCaMP6s of RIM cKO^DA mice, and dopamine release with RdLight1, see Extended data figs. 2–5.

Figure 3.. L-DOPA restores movement but not dopamine dynamics.

a. Schematic of the experiment and of the time-course of drug treatment, i.p. dosage: reserpine 3 mg/kg, L-DOPA methyl ester 250 mg/kg, carbidopa 25 mg/kg. b-e. Representative traces of trajectories (b) and GRAB_DA signals (c), and quantification of distance traveled over 30 min (d) and GRAB_DA variation (e) in mice in an open field, 4 mice. f-j. Average GRAB_DA signals registered to concurrent velocity in polar coordinates (f), individual (event heatmaps, top) and average (bottom) time course of velocity amplitudes (g) and GRAB_DA fluorescence changes (h) during contralateral movement initiations (left turns, velocity angles between 0° and 180°), and peak velocity (i) and GRAB_DA (j) per mouse. Event heatmaps in g+h were sorted by the peak velocity amplitude, baseline 316 events from 4 mice, reserpine + L-DOPA 378/4. k-n. As b-e, but for RIM cKO^DA mice; in n, the RIM control condition is replotted from Fig. 2c for comparison, RIM cKO^DA 4 mice. Data are mean ± SEM; * p < 0.05, ** p < 0.01, assessed by: Kruskal-Wallis analysis of variance with post hoc Dunn’s tests for d, e, m, n (vs. reserpine); two-sided Mann-Whitney rank-sum tests for i, j. For assessment of Syt-1 cKO^DA mice, F₀ and movement after reserpine + L-DOPA, additional GRAB_DA fluorescence in control and reserpinized mice, and additional GRAB_DA fluorescence in RIM cKO^DA mice, see Extended data figs. 6–9.

Figure 4.. RIM cKO^DA mice learn to make binary choices but the vigor of reward-oriented behavior is reduced.

a. Schematic of food-conditioned place preference (food-CPP) in which the less preferred side at baseline (day 1) is associated with food on days 2, 4, 6 and 8. On days 3, 5, 7 and 9, place preference is assessed. b-e. Representative example heatmaps of mouse location in the CPP arena (b) and quantification of percentage of time spent on test days in the food-associated chamber (c), of frequency of entries into the center of the CPP apparatus (d), and of movement speed of during these center entries (e). In the heatmaps (b), the time spent of each mouse is normalized in each session, RIM control 6 mice, RIM cKO^DA 6. f. Schematic and time course of the perceptual decision-making task in which the correct choice after exposure to odor A or B is rewarded by a water drop. The data were collected while mice learned to discriminate two odors. g-j. Quantification of the number of center port pokes (g), trial initiation times (h+i) and the number of trials it took to achieve > 90% accuracy for two consecutive sessions (j) over all trials, RIM control 9, RIM cKO^DA 9. k-m. Quantification of the reaction time (k), movement time (l) and trial completion time (m) in the first two sessions; times were calculated as the median of all trials for each mouse, n as in g-j. Data are mean ± SEM; * p < 0.05, ** p < 0.01, assessed by: two-way ANOVA for c, d; two-sided Mann-Whitney rank-sum tests for e, g-m.

Figure 5.. Disrupted anticipatory behavior of RIM cKO^DA mice in a probabilistic cue-reward association task.

a. Schematic of the task in head-fixed mice. b-e. Quantification of free water trials with the time course of GRAB_DA fluorescence (b, top) and licking (b, bottom) aligned to free water delivery (dashed line) on days 1 and 17 of the training (b), and peak GRBA_DA (c), peak licks (d) and mean delayed licks (e), RIM control 6 mice, RIM cKO^DA 7. f, g. Quantification of odor trials with anticipatory licks during the 2 s time window between odor and water delivery (f) and peak licks to reward within the first second of water delivery (g) for odors 1 (80% reward), 2 (40% reward) and 3 (no reward) from day 8 to day 20, n as in b-e. h, i. Quantification of odor trials with time course of GRAB_DA fluorescence (h, heatmaps), average GRAB_DA fluorescence (i, top) and licking (i, bottom) aligned to odor and water delivery per mouse. Heatmaps in h are sorted by the peak amplitude in odor 1 for each genotype, n as in b-e. j-l. Average GRAB_DA for each odor (j, within 3 s from odor onset), and average GRAB_DA for reward trials (k, within 201 to 1200 ms from water onset) and omission trials (l, within 1501 to 2500 ms from water onset), n as in b-e. Data are mean ± SEM; * p < 0.05, ** p < 0.01, *** p < 0.001 assessed by: two-sided Mann-Whitney rank-sum tests for c, d, e, j-l; two-way ANOVA for f, g. For additional assessment of licking and of GRAB_DA fluorescence in slices and in vivo, see Extended data fig. 10; for analyses of licking and dopamine dynamics on each training day, see Supplementary figures 1 to 4.