Generalized cue reactivity in rat dopamine neurons after opioids

Abstract

Cue reactivity is the maladaptive neurobiological and behavioral response upon exposure to drug cues and is a major driver of relapse. A widely accepted assumption is that drugs of abuse result in disparate dopamine responses to cues that predict drug vs. natural rewards. The leading hypothesis is that drug-induced dopamine release represents a persistently positive reward prediction error that causes runaway enhancement of dopamine responses to drug cues, leading to their pathological overvaluation. However, this hypothesis has not been directly tested. Here, we develop Pavlovian and operant procedures in male rats to measure firing responses within the same dopamine neurons to drug versus natural reward cues, which we find to be similarly enhanced compared to cues predicting natural rewards in drug-naive controls. This enhancement is associated with increased behavioral reactivity to the drug cue, suggesting that dopamine neuronal activity may still be relevant to cue reactivity, albeit not as previously hypothesized. These results challenge the prevailing hypothesis of cue reactivity, warranting revised models of dopaminergic function in opioid addiction, and provide insights into the neurobiology of cue reactivity with potential implications for relapse prevention.

Fig. 1. Pavlovian experimental procedure.

a Schematic of operant chamber for training/recording in Experiment 1. b Example waveform showing stability over a two-hour recording session. Color corresponds to action potential amplitude. c Pavlovian conditioning of three 5-s tones predicting sucrose (blue), remifentanil (green), or nothing (neutral, red). d Example responses of putative dopamine responses to cued sucrose reward (top), reward omission (middle), and uncued reward (bottom). e Rasters and post-stimulus time histogram (PSTH) showing example dopamine neuron firing after remifentanil infusion (2 infusions of 4 μg/kg remifentanil) with onset of remifentanil effect at ~10 s post-infusion. f PSTH for neuronal data of onset and duration of effect of remifentanil at doses of 0.5 to 16 μg/kg/infusion prior to training (activation terminates by ~2.5 min for 4 μg/kg). g Total time (left) and probability (right) of rat in sucrose reward well during sucrose and neutral cue presentations in opioid-naive rats (N = 4 rats, n = 22 units, two-tailed Wilcoxon test, z = -2.105, p < 0.0001). Lines and shades represent means ± SEM. h Same as h but in opioid-exposed animals after sucrose, remifentanil, and neutral cues (N = 7 rats, n = 50 units, Friedman test, F = 75.36, p < 0.0001 with Dunn’s multiple comparison test: sucrose vs. remifentanil p > 0.9999, sucrose vs. neutral p < 0.0001, remifentanil vs. neutral, p < 0.0001). Source data are provided as a Source Data file.

Fig. 2. Similar within-neuron dopamine firing responses to sucrose and remifentanil predictive cues.

a Dopamine neuron identification (N = 11 rats). (i) Heatmap showing functional activation of each unit (rows) aligned to sucrose cue onset (0 s). Activity normalized using area under receiver-operator curve (auROC) method, compared against baseline. Scales from 0 to 1. (ii) First three components extracted via principal component analysis. (iii) Dendrogram showing results from hierarchical clustering. (iv) Mean auROC for units classified into each cluster. (v) PSTH and raster plots from example units for each cluster. b Example responses from a single unit to sucrose (blue, top), remifentanil (green, middle), and neutral (red, bottom) cues. Red and blue shaded regions indicate distinct phases of cue response reflecting detection (30–180 ms) and valuation (180–500 ms). Inset waveform represents mean of sorted unit. Scale bars: 50 μV and 500 ms. c PSTH traces showing average dopamine firing during sucrose (blue, 40 μL), remifentanil (green, 4 μg/kg), and neutral trials (red). All traces show mean ± SEM. d Mean baseline-subtracted firing rates during cue response to sucrose, RMF, and neutral cues (Friedman test, F = 31.71, n = 21, p < 0.0001, with Dunn’s multiple comparison test: sucrose vs. remifentanil p > 0.9999, sucrose vs. neutral p < 0.0001, remifentanil vs. neutral, p < 0.0001). All bars show mean + SEM. e Close-up of cue period from Fig. 2c. f Scatter plot of individual units’ responses to sucrose (abscissa) and RMF (remifentanil; ordinate) cues during the detection phase. Inset: cue responses during the detection phase (Friedman test, F = 16.29, n = 21, p = 0.0003, with Dunn’s multiple comparison test: sucrose vs. remifentanil p > 0.9999, sucrose vs. neutral p = 0.0006, remifentanil vs. neutral, p = 0.0036). g Individual units’ responses during the evaluation phase presented as in Fig. 2f. Note that identical Chi-squared statistics to 2 f are due to equal numbers of units exhibiting firing to remifentanil > sucrose in both intervals. Inset: cue responses during the evaluation phase Friedman test, F = 28.67, n = 21, p < 0.0001, with Dunn’s multiple comparison test: sucrose vs. remifentanil p > 0.9999, sucrose vs. neutral p < 0.0001, remifentanil vs. neutral, p < 0.0001). Source data are provided as a Source Data file.

Fig. 3. Sensitized dopamine neuron responses in opioid-exposed rats in the Pavlovian procedure.

a PSTH and raster plot examples from a single opioid-exposed unit. Vertical lines indicate cue onset (0 s) and cue offset/sucrose delivery (5 s). Red = neutral trials, green = remifentanil (4 µg/kg), and blue = sucrose (40 µL). All traces show mean ± SEM. b PSTH and raster plot examples from a single opioid-naive unit. Display similar to a, with purple = sucrose and pink = neutral trials. c Mean firing rates prior to the start of session for each putative opioid-exposed (blue, N = 5 rats) vs. opioid-naive (purple, N = 4 rats) units (two-tailed unpaired t-test, t(97) = 2.060, n = 99, p = 0.0421). All bars show mean + SEM. d Comparison of previous training quantified as number of sucrose trials prior to the recording session (two-tailed Mann-Whitney test, n = 75, U = 499, p = 0.2586). e (i) Average dopamine firing during sucrose and neutral (red) trials in opioid-exposed units (same data as Fig. 2). (ii) Close-up of cue response in opioid-exposed units. (iii) Close-up of reward delivery response in opioid-exposed units. f (i) Average dopamine firing during sucrose and neutral (red) trials in opioid-naive units. (ii) Close-up of cue response in opioid-naive units. (iii) Close-up of reward delivery response in opioid-naive units. g Comparison of mean cue response to sucrose cues in opioid-exposed (blue) vs. opioid-naive (purple) units two-tailed Mann-Whitney test, n = 43, U = 131, p = 0.0145). h Comparison of mean cue response to neutral cues in opioid-exposed (red) vs. opioid-naive (pink) units (two-tailed Mann-Whitney test, n = 43, U = 120, p = 0.0064). i Comparison of mean response to sucrose delivery (0 – 400 ms after cue termination) (two-tailed Mann-Whitney test, n = 43, U = 118, p = 0.0054). j Trace of dopamine responses at time of reward omission in opioid-exposed (blue) and opioid-naive units (purple). Red and blue shaded regions reflect positive (0-200 ms) and negative (200-1000 ms) response periods 1 and 2, respectively. k Comparison of omission firing responses during period 1 and 2 between opioid-exposed and opioid-naive units (2-way RM ANOVA, Period factor: F(1,33) = 25.05, p < 0.0001; Exposure factor: F(1,33) = 6.087, p = 0.0190; Period x Exposure interaction F(1,33) = 9.022, p = 0.0051, Fisher’s LSD test exposed vs. naive P1 p = 0.0002, exposed vs. naive P2 p = 0.6376.). Source data are provided as a Source Data file.

Fig. 4. Operant experimental procedure demonstrates behavioral drug-cue reactivity.

a Schematic of experimental setup during training (top) and recording (bottom) sessions in Experiment 2. Boxes equipped with infusion pump (a), swivel (b), speaker (c), trial light (d), feeder lights I, feeders (f), retractable lever (g), dippers (h), commutator (i), and recording computer (j). b Experimental design. Gray shading indicates omitted reward in block two of recording trials. c Behavior training results for opioid-exposed animals (n = 5) = 5). Circles indicate % correct responses, Xs indicate % incorrect, and diamonds indicate % not responded to. Error bars = SEM. d Similar to 4c for opioid-naive rats (n = 6). e Total number of trials completed per recording session for opioid-exposed and opioid-naive rats (All bars = mean + SEMn = 106 sessions from N = 11 rats,two-tailed Mann-Whitney test, U = 1135, n = 106 sessions from N = 113 sessions from N = 11 rats, = 11 rats, p = 0.0975). f Number of correct responses per session to Cue A and Cue B (2-way RM ANOVA, Cue factor: F(1,114) = 4.786, n = 113 sessions from N = 11 rats, p = 0.0307. g Trial accuracy response index for opioid-exposed and opioid-naive animals (two-tailed one-sample t tests, H_o: μ = 1. Opioid-exposed: t(= 64, 47) = 2.168, n = 48, p = 0.0353; Opioid-naive: t(63) = 0.8954, n = 64, p = 0.3740). h Probability that first feeder entry after cue onset is congruent with cue identity (2-way RM ANOVA, opioid exposure factor: n = 100 sessions from Nn = 47, = 11 rats, F(1,101) = 5.032, p = 0.027; interaction F(1,101) = 8.621, p = 0.0041). i Ratio of total congruent feeder entries for Cue B/Cue A in both groups (two-tailed one-sample t tests. Opioid-exposed: t(46) = 3.343, n = 47, p = 0.0017; Opioid-naive: t(5 = 110 sessions from N = 11 rats, 4) = 0.6371, n = 55, p = 0.5268). j Cumulative congruent feeder entries following auditory cue onset (F-test for difference of slopes, F(3,156) = 5.482, n = 10 p = 0.0023; rat 8: t2 sessions from = 1.730, N = 11 rats, p = 0.0013). k Left axis: Session-mean of lever press hazard rate. 2-way Mixed effects model, Opioid-exposure factor: F(1,108) = 24.04, n = 110 sessions from N = 11 rats, n = 49 sessions from N = p < 0.0001). Right axis: cumulative % of sessions with lever press at time t is shifted to the left in opioid-exposed rats. All traces show mean ± SEM. l–n As in g–i respectively, but for rat 8 (red) and all other opioid-exposed animals: l (Opioid-exposed: t(36) = 3.284, n = 37, p = 0.0023; rat 8: t(11) = 1.730, n = 12, p = 0.1115); m (interaction F(1,47) = 24.4, n = 49 sessions from N = 5 rats, p = <0.0001); n (Opioid-exposed: t(34) = 5.352, n = 35, p < 0.0001; rat 8: t(10) = 5.000, n = 11, p = 0.0005). o Mean hazard rate for lever press during 0-2 s after lever extension in rat 8 vs. other opioid-exposed rats (two-tailed Mann-Whitney test, n = 49, U = 91, p = 0.0038). p Left: First two principal components (PCs) extracted from composite behavior measures for opioid-naive rats (purple), rat 8 (red filled), and all other opioid-exposed rats (teal). Crosses represent corresponding centroids. Right: Mean difference of rat 8 centroid under true labels (red dashed line) compared to bootstrapped distribution under shuffled labeling shows greater similarity of rat 8 to the opioid-naive group. Black dashed lines indicate two-tailed 95% threshold of null distribution. Source data are provided as a Source Data file.

Fig. 5. Sensitized dopamine responses to drug and non-drug cues in opioid-exposed rats with behavioral drug-cue reactivity.

a PSTH and raster plot examples from a single opioid-exposed unit. Vertical red lines indicate timeline discontinuity due to variable intervals within trials. Vertical black lines indicated discrete events: lever extension, lever press, and rewarded feeder entry. Purple trace includes all trials, which subsequently split based on trial identity to corresponding colors. Purple and orange dots indicate left and right feeder entries respectively. Green dots indicate lever presses. All traces show mean ± SEM. b Similar to a but for an opioid-naive unit. c Mean PSTH from opioid-exposed units (n = 31, N = 5 rats) displayed like in (a). d Mean PSTH from opioid-naive units (n = 38, N = 6 rats). e Top: PSTH of baseline-subtracted firing rate around lever extension for opioid-exposed and opioid-naive units. Bottom: mean firing rate for both groups (two-tailed Mann-Whitney test, n = 62, U = 124, p < 0.0001). All bars show mean + SEM. f Baseline firing rate for opioid-exposed (teal) and opioid-naive (purple) units (two-tailed unpaired t-test,t(67) = 2.117, n = 69, p = 0.0379). g Responses of opioid-exposed and opioid-naive units to the auditory cues. Right: mean firing rates (2-way RM mixed-effects analysis, n = 67, Cue factor: F(1,64) = 10.79, p = 0.0017; Exposure factor: F(1,65) = 8.157, p = 0.0058, Fisher’s LSD exposed A vs exposed B p = 0.0226, exposed A vs naive A p = 0.0170, exposed B vs naive B p = 0.0096, naive A vs naive B p = 0.0237). h Responses of opioid-exposed and opioid-naive units to light cue. Right: mean firing rates (2-way RM mixed-effects analysis, n = 69, trial-type factor: F(1,60) = 2.110, p = 0.1515; Exposure factor: F(1,62) = 11.96, p = 0.0010, Fisher’s LSD exposed A vs exposed B p = 0.0571, exposed A vs naive A p = 0.0070, exposed B vs naive B p = 0.0002, naive A vs naive B p = 0.9430). i Responses of opioid-exposed and opioid-naive neurons to water reward delivery. Right: mean firing rates (2-way RM mixed-effects analysis, n = 69, trial-type factor: F(1,61) = 0.1357, p = 0.7139; Exposure factor: F(1,62) = 38.41, p < 0.0001, Fisher’s LSD exposed A vs exposed B p = 0.8570, exposed A vs naive A p < 0.0001, exposed B vs naive B p < 0.0001, naive A vs naive B p = 0.7218). j Mean responses of neurons from rat 8 (red) vs. all other opioid-exposed units in response to lever extension (t(27) = 3.695, n = 29, p = 0.0010), auditory cue (A and B; t(27) = 2.915, n = 29, p = 0.0071), light cue (A and B; t(28) = 2.209, n = 30, p = 0.0355) and reward delivery (A and B; n = 30, U = 8, p = 0.0008). k Left: First two principal components (PCs) extracted from composite firing measures for opioid-naive rats (purple), rat 8 (red filled), and all other opioid-exposed rats (teal). Crosses represent corresponding centroids. Right: Mean difference of rat 8 centroid under true labels compared to bootstrapped distribution under shuffled labeling plotted like in Fig. 4p shows greater similarity of rat 8 to the opioid-naive group. l Scatter plot of principal component 1 from neuron firing activity analysis in K vs. principal component 1 from behavior analysis from Fig. 4p shows moderate positive correlation (two-tailed t-test for Ho: ρ = 0) between behavioral cue reactivity and dopamine firing response. Source data are provided as a Source Data file.