Abstinence and Extinction Drive Opposing Changes in Striatal Activity and Dopamine Signaling During Alcohol Relapse

48 Relapse remains a major challenge in the treatment of alcohol use disorder (AUD) .1,2 49 Although sustained abstinence is the optimal outcome for individuals with AUD,3 forced cessation 50 of alcohol use often triggers progressively greater craving over time, increasing vulnerability to 51 relapse.4-6 Behavioral interventions such as extinction training have shown promise in reducing 52 relapse vulnerability.7,8 In clinical populations, extinction-based approaches can attenuate striatal 53 responses to alcohol-related cues, while in animal models, extinction training reliably suppresses 54 relapse-like behavior.9-13 Mechanistically, extinction of cocaine seeking has been shown to reduce 55 glutamatergic synaptic transmission onto ventral striatal dMSNs, in contrast to abstinence, which 56 enhances excitatory input to this population. 14 Extinction of heroin seeking also dampens 57 thalamostriatal plasticity in ventral striatal dMSNs.15 However, despite these insights, the precise 58 neural mechanisms by which extinction alters striatal circuit function to suppress alcohol relapse 59 remain poorly understood, limiting its broader translational application. 60 The striatum, particularly its dorsomedial region (dorsomedial striatum, DMS), plays a 61 critical role in encoding goal -directed behaviors and driving relapse to alcohol seeking. 16-19 The 62 principal neurons within the striatum are medium spiny neurons (MSNs), which are classically 63 divided into two major subtypes based on their dopamine receptor expression and projection 64 targets.20,21 Direct-pathway MSNs (dMSNs) express dopamine D1 receptors and project directly 65 to the substantia nigra pars reticulata; their activation facilitates positive reinforcement and 66 promotes alcohol consumption. 21-23 In contrast, indirect-pathway MSNs (iMSNs) express 67 dopamine D2 receptors and project to the external segment of the globus pallidus; their activation 68 is associated with aversive outcomes and suppression of alcohol drinking.22-25 69 All addictive substances, including alcohol, trigger striatal dopamine release, 26 which 70 preferentially activates dMSNs and strengthens their corticostriatal synaptic connections. 18,27-29 71 Notably, optogenetic inhibition of dorsostriatal dMSNs has been shown to suppress operant 72 alcohol seeking and reduce relapse.30 Moreover, alcohol-induced synaptic plasticity between the 73 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 6 medial prefrontal cortex (mPFC) and dMSNs plays a bidirectional role in modulating operant 74 alcohol-seeking behaviors.31-33 Despite these advances, the precise neural dynamics of dMSN 75 activity and dopamine release during operant self-administration (OSA) of alcohol remain unclear. 76 Furthermore, how abstinence or extinction experiences influence these dynamics has yet to be 77 fully elucidated. 78 In this study, we employed genetically encoded sensors to monitor real -time activity of 79 DMS dMSNs and dopamine release dynamics in freely moving rats during operant self-80 administration (OSA) of alcohol. We further extended these measurements to relapse conditions 81 following either forced abstinence or extinction training to determine how these experiences 82 shape striatal signaling. Lastly, we utilized dual-channel optogenetics to bidirectionally manipulate 83 corticostriatal plasticity after abstinence or extinction, and tested its causal role in regulating 84 relapse vulnerability. Together, these experiments aim to uncover circuit- and synapse-level 85 mechanisms underlying alcohol relapse and provide insights into targeted interventions that may 86 prevent it. 87 88 89

90 DMS dMSN activity and dopamine release peak after lever press but before alcohol 91 consumption during operant self-administration 92 Striatal dMSNs are preferentially activated during substance use, likely via enhanced 93 dopamine release.28,34,35 However, the precise temporal dynamics of dMSN activity and dopamine 94 release during alcohol OSA remain unclear. To address this, we first examined when and how 95 DMS dMSNs are activated, and when dopamine is released to the DMS during alcohol OSA. This 96 paradigm closely models human patterns of alcohol use and relapse.36,37 97 In our paradigm, rats were trained to press an active lever to receive alcohol, which was 98 delivered into a central magazine (Fig. 1A). A discrete cue light within the delivery port 99 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 7 accompanied each delivery and was later used to induce relapse. To monitor neural dynamics, 100 we unilaterally expressed AAV -FLEX-jGCaMP7f in DMS dMSNs and AAV -GRABDA2m in the 101 contralateral DMS for dual-site fiber photometry recordings in D1-Cre rats (Fig. 1B-1D). 102 Training began under a fixed-ratio 1 (FR1) schedule and progressed to FR3 once stable 103 responding was established. Each alcohol delivery was followed by a 20 -sec time-out period to 104 allow consumption. Fiberphotometry was performed under the FR3 schedule. Behaviorally, 105 animals took approximately 1.5 seconds to enter the magazine after the final active lever press 106 (Fig. 1E). Notably, the dMSN GCaMP signal exhibited a distinct peak immediately after the last 107 active press but before magazine entry (Fig. 1F, G; between 0 and 1.5 sec ), consistent with 108 previous reports.38,39 This peak quickly transitioned to a dip once the animals began to enter the 109 magazine to collect alcohol (Fig. 1 F, G; after 1.5 sec ). In contrast, dopamine signals peaked in 110 synchrony with alcohol delivery (Fig. 1H, I). Cosine similarity analysis revealed strong correlations 111 between dMSN calcium signals (area-under -curve [AUC] of the GCaMP peak) and task -related 112 behaviors (active pressing and magazine entry) (Fig. 1J). Therefore, we focused on the AUC of 113 the GCaMP signal for subsequent comparisons. GCaMP and DA2m signals were comparable 114 between hemispheres (Fig. 1K). 115 Together, these results show that DMS dMSNs are most active immediately following 116 lever pressing—prior to alcohol collection—while DMS dopamine release is tightly time-locked to 117 the delivery of alcohol. 118 119 dMSN activity is elevated during cue-alcohol memory retrieval after abstinence but 120 normalized by extinction training 121 Having established the activation pattern of dMSN and release pattern of dopamine during 122 alcohol OSA, we next investigated whether abstinence or extinction training differently affects 123 dMSN activity and dopamine release during cue-induced reinstatement. After ~6 weeks of alcohol 124 OSA training, rats were separated into a forced abstinence group, which stayed in their home-125 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 8 cage for 9 d, or an extinction group, which w as subjected to 9 d of extinction training (Fig. 2A). 126 This was followed by a 2-d cue-induced reinstatement test. Fiber photometry recordings were 127 conducted during the final FR3 training sessions and reinstatement testing (Fig. 2A). 128 In the abstinence group (Fig. 2B), dMSN activity—measured as the GCaMP peak AUC —129 was significantly elevated during reinstatement compared to the FR3 baseline (Fig. 2C). 130 Dopamine release also increased markedly during reinstatement (Fig. 2D). This suggests that 131 dMSN activity is enhanced during the retrieval of a cue-alcohol memory after abstinence, 132 potentially attributable to amplified dopamine release. Conversely, in the extinction group (Fig. 133 2E), there was no notable change in the AUC of the GCaMP peak between the FR3 baseline and 134 reinstatement (Fig. 2F). Intriguingly, dopamine signal s exhibited a marked decrease during 135 reinstatement (Fig. 2G). These results indicated that extinction training dampened dopamine 136 release, which could consequently alter corticostriatal plasticity and normaliz e dMSN activity 137 during memory retrieval. 138 Further comparison between the abstinence and extinction groups revealed that the 139 extinction group had significantly fewer active presses during reinstatement (Fig. 2H). The 140 increase in dMSN activity from baseline to reinstatement was significantly lower in the extinction 141 group (Fig. 2I), as was the change in dopamine release (Fig. 2J). Notably, GCaMP signals 142 remained correlated with lever pressing in the abstinence group (Fig. 2K), but this association 143 was weaker following extinction training (Fig. 2L). 144 Overall, these findings suggest that abstinence enhances dMSN activity and dopamine 145 release during cue-induced relapse, while extinction training reduces both, potentially mitigating 146 relapse vulnerability. 147 148 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 9 Optogenetic induction of mPFC-to-dMSN synaptic plasticity after abstinence or 149 extinction oppositely controls cue-alcohol memory retrieval 150 Having demonstrated that abstinence and extinction training differentially modulate dMSN 151 activity and dopamine release during cue-induced reinstatement, we next sought to elucidate the 152 underlying mechanism connecting these physiological changes to relapse behaviors. Because 153 dopamine release critically modulates corticostriatal synaptic plasticity onto dMSNs, 21,28,32,40 the 154 observed alterations in dopamine signaling and dMSN activity may reflect underlying changes in 155 corticostriatal synaptic strength. Therefore, we hypothesized that directly manipulating 156 corticostriatal plasticity —specifically, inducing LTD after abstinence to mimic extinction-like 157 plasticity, or inducing LTP after extinction to mimic abstinence-like potentiation—would 158 bidirectionally regulate the propensity for cue-induced relapse. 159 To test this hypothesis, we aimed to experimentally alter corticostriatal plasticity following 160 periods of abstinence or extinction training and assess the subsequent impact on cue-induced 161 reinstatement. D1-Cre rats received bilateral infusions of AAV -Chronos-GFP in the mPFC and 162 AAV-FLEX-Chrimson-tdTomato in the DMS, with optical fibers implanted bilaterally above the 163 DMS (Fig. 3A, 3B). Rats were trained to self-administer alcohol in operant chambers until stable 164 performance was achieved under an FR3 schedule. Subsequently, the rats were categorized into 165 abstinence or extinction groups and subjected to a cue-induced reinstatement test (Fig. 3C). 166 In the abstinence group, animals remained in their home-cages for 9 d. On the final day 167 of abstinence, corticostriatal LTD was induced in a neutral chamber via pre-injection of the NMDA 168 receptor antagonist MK801 (i.p., 0.1 mg/kg) 15 min before the pairing of oHFS and oPSD,32 aiming 169 to depotentiate corticostriatal plasticity and mimic the effect of extinction (Fig. 3D). Control animals 170 in the abstinence group did not receive LTD induction. Previous studies confirmed that MK801 171 administration itself did not alter corticostriatal plasticity nor alcohol -seeking behaviors in rats.32 172 Although both the control and optogenetically-induced LTD (oLTD) groups displayed similar active 173 presses during FR3 (Fig. 3E), rats in the oLTD group exhibited significantly fewer active presses 174 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 10 during the reinstatement test (Fig. 3F). This difference persisted even when the active presses 175 were normalized to those during the baseline FR3 sessions (Fig. 3G). These findings demonstrate 176 that depotentiation of corticostriatal plasticity following abstinence effectively mimics the 177 behavioral outcomes of extinction training, reducing relapse vulnerability. 178 In the extinction group, animals underwent extinction training in the operant chamber for 179 9 d. After the last extinction training, corticostriatal LTP was induced in a neutral chamber by 180 pairing oHFS with oPSD (Fig. 4A), with the aim of potentiating corticostriatal plasticity and mimic 181 abstinence. Rats in both the control and optogenetically -induced LTP (oLTP) groups displayed 182 similar performance during baseline FR3 and extinction training (Fig. 4B). However, rats 183 subjected to oLTP showed a pronounced increase in active presses during the reinstatement test 184 (Fig. 4C). This effect persisted even after normalizing the presses to those seen during the 185 baseline FR3 sessions (Fig. 4D). These results indicate that experimentally potentiating 186 corticostriatal synapses after extinction training partially reverses the extinction effect and 187 facilitates relapse. 188 In summary, our fiber photometry results suggest that abstinence enhances dopamine 189 release and dMSN activity during cue-induced relapse, whereas extinction training reduces 190 dopamine signaling and normalizes dMSN activity. Although we did not directly measure synaptic 191 changes, these physiological patterns are consistent with a model in which corticostriatal plasticity 192 is differentially shaped by abstinence and extinction history. Specifically, the increase in dopamine 193 during abstinence may potentiate mPFC-to-dMSN synapses, thereby facilitating relapse behavior 194 (Fig. 4E). In contrast, reduced dopamine signaling following extinction may weaken these 195 synapses, contributing to relapse suppression (Fig. 4E). Our optogenetic manipulations support 196 this interpretation by demonstrating that mimicking synaptic potentiation or depression can 197 bidirectionally influence relapse vulnerability. 198 199 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 11

200 In this study, we identified distinct neural signatures and synaptic mechanisms underlying 201 alcohol relapse after abstinence or extinction, and demonstrated that bidirectional modulation of 202 corticostriatal synaptic strength causally controls cue-induced reinstatement of alcohol -seeking 203 behavior. Using in vivo calcium and dopamine recordings, we found that abstinence enhances 204 DMS dMSN activity and dopamine release during relapse, whereas extinction training normalizes 205 dMSN activity and reduces dopamine release. Importantly, manipulating mPFC-to-dMSN synaptic 206 plasticity was sufficient to either attenuate or reinstate relapse-like behavior, revealing a causal 207 role for corticostriatal dynamics in the expression of alcohol-associated memories. 208 Our fiber photometry recordings, conducted in freely -moving rats during an operant task, 209 provide compelling in vivo evidence of neural activity linked to behavior. The task used reflects a 210 classic operant conditioning paradigm,37,41 in which active lever pressing (reward-seeking) leads 211 to alcohol delivery, followed by magazine entry (reward-taking). dMSNs are known to be involved 212 in action initiation,42 but in our paradigm, we observed that dMSN activity did not precede the start 213 of an action sequence (i.e., active lever press). Instead, activity dipped during the lever presses 214 and peaked at the transition from pressing to magazine entry, suggesting a role in goal evaluation 215 or outcome expectation rather than movement initiation.43-45 This aligns with prior studies showing 216 DMS activation during reward delivery rather than action onset. 39 Notably, similar activation 217 patterns have been observed in food-reward paradigms, where peak striatal activity corresponds 218 with the movement toward reward collection, and not with movements of similar motor demands 219 like lever pressing.38 Importantly, this population-level activation of dMSNs may also suggest the 220 formation of neuronal ensembles that encode the learned association between lever -press 221 actions and alcohol reward. 46 Dopamine release exhibited a concurrent or slightly earlier peak, 222 ramping up during lever pressing and peaking after reward delivery. This signal likely reflects 223 reward expectation and valuation, 47-49 potentially integrating multisensory cues 48 (e.g., sound of 224 dipper, cue light, smell of alcohol) and reinforcing synaptic connections between active 225 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 12 glutamatergic inputs (e.g., mPFC) and dMSNs. Given that dopamine is a key modulator of 226 corticostriatal plasticity, 21,28,29,48,50 and that memory is thought to be encoded by synaptic 227 changes,51 this convergence of dopamine release and dMSN activity may facilitate the formation 228 and stabilization of alcohol -associated corticostriatal plasticity that encodes operant associative 229 memories, ultimately driving later relapse. 230 Our result that increased activity in dMSNs during cue-induced reinstatement of alcohol 231 seeking following a 9-day abstinence reinforces previous findings of heightened striatal neuron 232 activity during drug relapse. 52-54 Cue-induced dopamine release in the dorsal striatum is a well -233 documented phenomenon across multiple drugs,66,67 including alcohol. Recent work even shows 234 that re-exposure to cocaine cues after abstinence amplifies dopamine signaling in the pallidum.55 235 In our study, similar cue-induced dopamine elevations were observed in the DMS following 236 abstinence, potentially driving increased dMSN activity. Additionally, alcohol -induced glutamate 237 spillover—due to impaired perisynaptic astrocytic transporter function—could prolong cue-evoked 238 glutamatergic signaling, 56 further enhancing dMSN activation and potentially recruiting new 239 ensemble members. These mechanisms may underlie the incubation of craving, a phenomenon 240 wherein cue-induced drug seeking intensifies over time following abstinence.6 241 Conversely, extinction training appears to dampen the neural processes supporting 242 relapse. Extinction is known to share features with synaptic depression and to impair plasticity 243 induction.15,57,58 Previous studies have shown that extinction can occlude LTD in the amygdala 244 and prevent LTD induction in ventrostriatal dMSNs. 15,57,59 These findings support the idea that 245 extinction weakens memory retrieval by disrupting corticostriatal synaptic efficacy. We found that 246 diminished dopamine release during cue-induced relapse after extinction may reflect reduced 247 reward prediction and contribute to synaptic weakening, ultimately decreasing dMSN recruitment 248 and relapse behavior. 249 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 13 Critically, we provide causal support for this framework through optogenetic manipulation 250 of corticostriatal plasticity. Inducing LTD after abstinence reduced relapse, likely by depotentiating 251 alcohol-strengthened synapses. 32,33 This effect may be mediated through presynaptic 252 mechanisms that lower glutamate release, 32 thereby decreasing dMSN activation. Conversely, 253 inducing LTP after extinction reversed the protective effect of extinction, promoting relapse 254 through potentiation of mPFC -to-dMSN synaptic transmission. These results demonstrate that 255 corticostriatal plasticity is not only shaped by behavioral history but can also be targeted to 256 manipulate relapse vulnerability. 257 Together, our findings reveal a mechanistic framework in which abstinence and extinction 258 differentially shape striatal circuit dynamics and dopamine signaling, and where mPFC -to-dMSN 259 synaptic plasticity acts as a bidirectional switch controlling alcohol relapse. This work advances 260 our understanding of relapse neurobiology and identifies specific synaptic processes that may be 261 targeted to enhance the durability of extinction and reduce the risk of relapse in alcohol addiction. 262 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 14 Figure legends 263 Figure 1. DMS dMSN calcium activity and dopamine release peak after active presses but 264 before magazine entries during operant self-administration of alcohol 265 A, Schematic of the operant self-administration (OSA) paradigm for alcohol (EtOH). Rats were 266 trained to press an active lever to receive 0.1 mL of 20% (v/v) alcohol, delivered alongside a cue 267 light above the port. Rats then entered the magazine (magazine entry; MgE) to collect the alcohol. 268 Training continued until stable performance was achieved under a fixed ratio 3 (FR3) schedule, 269 in which three active lever presses triggered an alcohol delivery. 270 B, Schematic of virus infusion for in vivo fiber photometry recordings. In D1-Cre rats, one 271 hemisphere of the DMS was infused with AAV -FLEX-jGCaMP7f (GCaMP), and the other with 272 AAV-GRABDA2m (DA2m). Virus assignments were counterbalanced across animals. 273 C, Sample confocal images show GCaMP expression in the cytosol of dMSNs (left) and DA2m 274 expression on the membrane of DMS neurons (right). 275 D, Representative raw traces of calcium and dopamine signals recorded during an OSA session. 276 E, Behavioral distribution from all rats under the FR3 schedule. On average, rats entered the 277 magazine ~1.5 sec after the onset of alcohol delivery. 278 F, Representative heatmaps showing calcium activity during an FR3 session from one rat. The 279 dashed lines at 0 sec and 1.5 sec indicate the timing of alcohol delivery and the average time of 280 first magazine entry, respectively. 281 G, Averaged across all animals , dMSN calcium peaked during the transition between active 282 pressing and magazine entry, followed by a dip at magazine entry. 283 H, Representative heatmaps showing dopamine release during an FR3 session in one rat. 284 I, Averaged across all animals, dopamine signal peaked immediately after active pressing and 285 delivery. 286 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 15 J, The cosine similarity heatmap reveals a high correlation between the behaviors and the area-287 under-curve (AUC) of the dMSN GCaMP peak, as enclosed by the dashed rectangle. X-axis from 288 left to right: T 1st active press , time of the first active press ; MgE half width, half width of averaged 289 duration of magazine entry. Y-axis: Pmax, z-score of GCaMP peak; T(Pmax), time of GCaMP peak; 290 Pmin, GCaMP dip; T(Pmin), time of GCaMP dip; AUC(P), AUC of GCaMP peak. 291 K, The AUC of the GCaMP peak (left; t(11) = -0.01, p > 0.05) or the z-score of the DA peak (right; 292 t(13) = 1.76, p > 0.05) between the left and right hemisphere had no difference. 293 Cosine similarity test (J) and unpaired t test (K). n = 11 rats (G), 5 rats (K; GCaMP-left), 8 rats (K; 294 GCaMP-right), 7 rats (K; DA-left), 8 rats (K; DA-right). 295 296 297 Figure 2. Abstinence enhances dMSN activity and dopamine release during cue-induced 298 reinstatement, an effect that is normalized by extinction training 299 A, Experimental timeline. D1-Cre rats underwent ~6 weeks of alcohol OSA training under an FR3 300 schedule. Baseline recordings of calcium and dopamine signals were performed over the last four 301 sessions. Rats were then assigned to either the abstinence group (Abs; 9 days of home-cage 302 abstinence) or the extinction group (Ext; 9 days of operant lever-extinction training). Cue-induced 303 reinstatement (Rein) tests were conducted across two days, with GCaMP and DA2m recorded on 304 separate days. 305 B, Active lever presses during the last four OSA sessions and during reinstatement (average of 306 two sessions) in the abstinence group. 307 C, In abstinent rats, dMSN GCaMP exhibited an increased AUC of the peak ( t(5) = -2.86, *p < 308 0.05) during reinstatement compared to the OSA. 309 D, In abstinent rats, the z -score of the DMS dopamine peak increased during reinstatement 310 compared to the OSA. t(8) = -2.74, *p < 0.05. 311 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 16 E, Active lever presses during the last four OSA sessions and reinstatement (average) in the 312 extinction group. 313 F, No difference in GCaMP peak AUC was observed during reinstatement compared to OSA 314 following extinction training. t(6) = 0.519, p > 0.05. 315 G, After extinction, the z -score of the DMS dopamine peak reduced during reinstatement 316 compared to the OSA. t(5) = 3.79, *p < 0.05. 317 H, Extinction training significantly reduced active lever presses during reinstatement compared to 318 abstinence. t(18) = 3.98, ***p < 0.001. 319 I, Percentage change in dMSN GCaMP peak AUC from OSA to reinstatement was smaller in the 320 extinction group than in the abstinence group. t(11) = 2.77, *p < 0.05. 321 J, Percentage change in dopamine peak from OSA to reinstatement was smaller in the extinction 322 group than in the abstinence group. t(13) = 2.71, *p < 0.05. 323 K,L, The correlation between the GCaMP signal and the first active press was still present in the 324 abstinence group (K) but absent in the extinction group (L), as marked by dashed rectangles. 325 Sample traces (left of C, D, F, G) were averaged data from all animals in the corresponding group 326 during OSA or reinstatement. Paired t test (C, F, D, G), unpaired t test (H-J), and cosine similarity 327 test (K). n = 10 rats (B), 6 rats (C), 9 rats (D), 10 rats (E), 7 rats (F), 6 rats (G), 10 rats per group 328 (H), 6 rats (I; Abs), 7 rats (I; Ext), 9 rats (J; Abs), 6 rats (J; Ext), 6 rats (K), and 7 rats (L). 329 330 331 Figure 3. mPFC-to-dMSN LTD after abstinence reduces cue-induced reinstatement 332 A, Schematic illustrating virus infusion and fiber implantation in D1-Cre rats. 333 B, Representative confocal images of AAV-Chronos-GFP expression in the mPFC (left) and AAV-334 FLEX-Chrimson-tdTomato expression in the DMS (right). The insets show single neurons from 335 enclosed areas. M, medial; V, ventral. 336 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 17 C, Experiment timeline. After ~6 weeks of alcohol OSA training, rats were divided into abstinence 337 or extinction groups. On the last day of abstinence or extinction, we delivered optogenetic long-338 term depression (oLTD) or long-term potentiation (oLTP) induction protocols, followed by a cue-339 induced reinstatement test the subsequent day. 340 D, The schematic illustrating the schedule and protocol for oLTD induction in the abstinence group. 341 E-G, The oLTD and control (Ctrl) groups exhibited similar levels of active presses during the OSA 342 (E; F(1,15) = 0.03, p > 0.05). However, oLTD induction decreased the number of active presses (F; 343 t(15) = 6.46, ***p < 0.001) and normalized active presses (G, normalized to the OSA ; t(15) = 2.64, 344 *p < 0.05) during the reinstatement. 345 Two-way RM ANOVA followed by Tukey post hoc test (E), unpaired t test (F, G). n = 8 rats (Ctrl) 346 and 9 rats (oLTD). 347 348 349 Figure 4. mPFC-to-dMSN LTP after extinction potentiates cue-induced reinstatement 350 A, The schematic outlining the schedule and protocol for oLTP induction following 9 d of extinction 351 training. 352 B-D, The oLTP and Ctrl groups exhibited similar levels of active presses during the OSA and 353 extinction training (B; F(1,12) = 0.05, p > 0.05). However, oLTP induction increased the number of 354 active presses (C; t (12) = -2.74, *p < 0.05) and normalized active presses (D, normalized to the 355 OSA; t(15) = -2.46, *p < 0.05) during reinstatement. 356 E, Schematic of the working hypothesis: glutamatergic transmission onto DMS dMSNs is 357 strengthened during abstinence, promoting relapse; conversely, extinction weakens this 358 transmission, reducing relapse. 359 Two-way RM ANOVA followed by Tukey post hoc test (B), unpaired t test (C, D). n = 8 rats (Ctrl) 360 and 6 rats (oLTP). 361 362 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 18

363 Reagents 364 AAV9-Syn-FLEX-jGCaMP7f (#104491) and AAV9-hSyn-GRAB_DA2m (#140553) were 365 obtained from Addgene. AAV8-Syn-Chronos-GFP and AAV8-Syn-FLEX-Chrimson-tdTomato were 366 purchased from the University of North Carolina Vector Core. MK801 were purchased from Sigma. 367 Animals 368 D1-Cre rats were obtained from Rat Resource and Research Center .60 Genotypes were 369 determined by PCR analysis of Cre gene in tail DNA.18,61,62 All animals that were used are in mixed 370 gender and aged from 3-8 months. Animals were housed in a temperature- and humidity -371 controlled vivarium with a reversed 12-h light/dark cycle (lights on at 10 :00 P .M. and off at 10:00 372 A.M.). All behavior experiments were conducted in their dark cycle, starting approximately 1 h 373 after the light went off. Food and water were available ad libitum. All animal care and experimental 374 procedures were approved by the Texas A&M University Institutional Animal Care and Use 375 Committee. 376 Experiment procedures 377 Intermittent-access to 20% alcohol two-bottle-choice drinking procedure 378 To acclimate rats with alcohol and establish high levels of alcohol consumption, we utilized 379 the intermittent-access to 20% alcohol two-bottle-choice drinking procedure. 18,23,31,33,63 D1-Cre 380 rats were given free access to two bottles containing water or 20% alcohol for three 24-h sessions 381 (Mondays, Wednesdays, and Fridays), with 24-h or 48 -h withdrawal periods (Tuesdays, 382 Thursdays, Saturdays, and Sundays) each week. During the withdrawal periods, the rats had 383 unlimited access to water bottles. This procedure was followed for ~3 weeks. 384 Operant self-administration of alcohol 385 Animals were trained to self-administer 20% alcohol in operant chambers using a fixed 386 ratio (FR) schedule, as described previously.30,32,64 The rat operant system (Coulbourn Instrument) 387 include two levers: active and inactive in each chamber, on the opposite sides of the same wall . 388 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 19 During training sessions, a house light centered above the operant illuminated. Each operant 389 experiment commenced with continuous reinforcement (FR1) for approximately two to three 390 weeks, wherein pressing the active side resulted in a 0.1 mL delivery of 20% alcohol. Actions on 391 the inactive side were documented but did not initiate a programmed event. The alcohol solution 392 was dispensed into a stainless steel reservoir situated within the magazine port between active 393 and inactive levers. Each alcohol delivery persisted for 20 s ec, during which the magazine port 394 was lit by a discrete yellow cue light for the same duration; both levers were withdrawn during the 395 delivery period to facilitate alcohol consumption. When animals were able to achieve at least 10 396 deliveries under the FR1 schedule, the criterion to receive the alcohol escalated to FR2 (around 397 one to two weeks), eventually progressing to FR3 (around two weeks). Each operant session for 398 rats lasted 30 min. 399 Extinction training. The extinction training was conducted under the FR3 schedule. However, 400 pressing on the active side did not result in actual alcohol delivery, nor cue light presentation. 401 Each extinction session lasted 30 min, and in total for 9 d. 402 Cue-induced reinstatement of alcohol seeking. The reinstatement test was conducted as 403 described.65 It was carried out at the FR3 schedule, where three actions on the active side 404 triggered the illumination of the cue light within the magazine port and the dipper action sound, 405 with the exception that the first cue presentation and dipper action was immediately after the first 406 active press. Considering that many rats stopped active responses following the 9-day extinction 407 training, a 20 µL volume of alcohol was pipetted into the magazine port prior to the session's 408 commencement to instigate the animals' responses. However, once the session had started, 409 alcohol was not available. 410 Stereotaxic virus infusion and fiber implantation 411 The stereotaxic virus infusion procedure was conducted as described previously. 32,33,62,66 412 Where required for the experimental design, AAV-Chronos-GFP (0.8 µL/site) was infused into the 413 mPFC (AP: +2.9 mm, ML: ±0.65 mm, DV: -3.5 mm from the Bregma), AAV -FLEX-Chrimson-tdT 414 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 20 (1.2 µL/site) was infused into the DMS (AP: +0.36 mm, ML: ±2.3 mm, DV: -4.7 mm from the 415 Bregma). For fiberphotometry rats, one side of DMS was infused with AAV-FLEX-jGCaMP7f (1.2 416 µL) and the other side with AAV -GRAB_DA2m (1.2 µL)(AP: +0.36 mm, ML: ±2.3 mm, DV: -4.7 417 mm from the Bregma; the assignment of jGCaMP7f and GRAB_DA2m was counterbalanced). 418 Optical fiber implants (300-μm diameter optical fiber secured to a 2.5- mm stainless steel ferrule) 419 were placed bilaterally in DMS through the injection tract (DV: -4.6 mm). Animals were 420 anesthetized with 3-4% isoflurane at 1.0 L/min and mounted in a stereotaxic surgery frame. The 421 head was leveled and craniotomy was performed using stereotaxic coordinates adapted from the 422 brain atlas.67 The virus was infused at a rate of 0.08 µL/min. At the end of the infusion, the injectors 423 remained at the injection site for an additional 10-15 min before removal to allow for virus diffusion. 424 For fiber implants, four metal screws were fixed into the skull to support the implants, which were 425 further secured with dental cement (Henry Schein). The scalp incision was then sutured, and the 426 animals were returned to their home cage for recovery. 427 In vivo fiber-photometry recording 428 Fiberphotometry was conducted as previously described. 38,68 In each D1-Cre rats, one 429 side of the DMS contained FLEX-jGCaMP7f and the other side contained GRAB_DA2m. The left 430 or right assignment of these two sensors were counterbalanced. Rats were connected with fiber 431 for the entire training periods to acclimate to the fiber patch cable. Baseline recordings occurred 432 during the last 4 sessions of FR3 training (each sensor in each animal was recorded twice and 433 then the signals were averaged). Following 9-d extinction training or abstinence, cue-induced 434 reinstatement was tested for 2 sessions (1 session for GCaMP measurement and 1 session for 435 DA2m measurement; the measurement was counterbalanced). The spectrum data was recorded 436 continuously at 10 Hz. At the same time, behavior data were collected. Fiber -photometry data 437 was collected using OceanView 1.6.7. 488 nm laser was delivered to excite jGCaMP7f expressing 438 in dMSNs or GRAB_DA2m. The Z-score (Z) was calculated by 439 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 21 𝑍𝑍= Δ𝐹𝐹/𝐹𝐹 − 𝜇𝜇 σ 440 using MATLAB. To anal yze the fiberphotometry data in the context of rat behavior, MATLAB 441 scripts were developed. Cosine similarity between the behavior data and the photometry data 442 was analyzed, and the Pearson correlation coefficients ( \rho) between the behavior data (x) and 443 the fiber photometry data (y) is calculated by 444 ρ𝑥𝑥,𝑦𝑦= (𝑥𝑥 − 𝑥𝑥) ⋅ (𝑦𝑦 − 𝑦𝑦) |𝑥𝑥 − 𝑥𝑥| ∗ |𝑦𝑦 − 𝑦𝑦| 445 using MATLAB. B oth the extinction and abstinence groups had 10 rats. However, due to recording 446 errors or fiber misplacement, 4 recordings of GRAB_DA2m and 3 recordings of jGCaMP7f were 447 missed in the extinction group; 1 recording of GRAB_DA2m and 4 recordings of jGCaMP7f were 448 missed in the abstinence group. 449 In vivo LTP and LTD induction 450 During the last day of abstinence or after the last session of extinction training, D1-Cre 451 rats received an LTD/LTP-inducing protocol in a neutral Plexiglass chamber, with no visual cues. 452 LTP induction consists of paired oHFS + oPSD using the following protocol: 100 pulses at 50 Hz 453 of 473-nm light (2 ms) with constant 590-nm light for 2 sec. Each stimulation was repeated 7 times 454 with a 20-sec interval, forming a bout of stimulation. A total of 6 bouts were delivered with a 3-min 455 interval. LTD induction employed the following protocol: animals were injected with MK801 (0.1 456 mg/kg) 15 min before delivery of oHFS and oPSD (same as above) . Control animals for both 457 inductions were exposed to the Plexiglass chamber and connected to the patch cable, but no light 458 was delivered. The complete LTP/LTD -inducing procedure was performed once, and animals 459 were tested for cue-induced reinstatement 24 h later. 460 Histology 461 Animals were perfused intracardially with 4% formaldehyde in phosphate-buffered saline. 462 The brains were removed and post-fixed overnight in 4% formaldehyde in phosphate-buffered 463 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 22 saline at 4°C prior to dehydration in 30% sucrose solution. The brains were cut into 50-µm coronal 464 sections using a cryostat. All images were acquired using a confocal microscope (FluoView 3000, 465 Olympus, Tokyo, Japan) and analyzed using IMARIS 8.3.1 (Bitplane, Zürich, Switzerland), as 466 previously reported.61,62 467 Statistical analysis 468 All data are expressed as the mean ± the standard error of the mean. Data were analyzed 469 by two-tailed t test (unpaired or paired) and two-way ANOVA with repeated measurement, 470 followed by the Tukey post hoc test. Significance was determined if p < 0.05. Statistical analysis 471 was conducted by the SigmaPlot program. Graphs were constructed using the OriginPro program. 472 473 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 23

474 This research was supported by NIAAA Grant R01AA021505, R01AA027768, U01AA025932 and 475 X-grant from Texas A&M University to J.W., McGovern Fellowship from Texas Research Society 476 on Alcoholism (TRSA) and Doctoral Student Small Grant from Research Society on Alcohol (RSA) 477 to X.X. 478 479 Declaration of interests 480 The authors declare no competing interests. 481 482 Inclusion and diversity 483 We support inclusive, diverse, and equitable conduct of research. 484 485 Author contributions 486 Conceptualization: X.X., J.W.; Methodology: X.X., R.C., J.W.; Investigation and Formal analysis: 487 X.X., R.C.; Software: R.C.; Writing - Original Draft: X.X.; Writing – Review & Editing: J.W., X.X; 488 Resources: X.W.; Funding acquisition: J.W., and X.X.; Supervision: J.W. 489 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 24

490 1. Torregrossa, M.M., Corlett, P.R., and Taylor, J.R. (2011). Aberrant learning and 491 memory in addiction. Neurobiology of learning and memory 96, 609- 623. 492 10.1016/j.nlm.2011.02.014. 493 2. Nicholas W. Gilpin, P.D., and George F. Koob, Ph.D. (2008). Neurobiology of 494 Alcohol Dependence. Alcohol research & health : the journal of the National 495 Institute on Alcohol Abuse and Alcoholism. 496 3. Connor, J.P., Haber, P.S., and Hall, W.D. (2016). Alcohol use disorders. Lancet 497 387, 988-998. 10.1016/S0140-6736(15)00122-1. 498 4. Heilig, M., Egli, M., Crabbe, J.C., and Becker, H.C. (2010). Acute withdrawal, 499 protracted abstinence and negative affect in alcoholism: are they linked? Addiction 500 biology 15, 169-184. 10.1111/j.1369-1600.2009.00194.x. 501 5. Bienkowski, P., Rogowski, A., Korkosz, A., Mierzejewski, P., Radwanska, K., 502 Kaczmarek, L., Bogucka- Bonikowska, A., and Kostowski, W. (2004). Time-503 dependent changes in alcohol -seeking behaviour during abstinence. Eur 504 Neuropsychopharmacol 14, 355-360. 10.1016/j.euroneuro.2003.10.005. 505 6. Pickens, C.L., Airavaara, M., Theberge, F., Fanous, S., Hope, B.T., and Shaham, 506 Y. (2011). Neurobiology of the incubation of drug craving. Trends Neurosci. 34, 507 411-420. 10.1016/j.tins.2011.06.001. 508 7. Drummond, D.C., Cooper, T., and Glautier, S.P. (1990). Conditioned learning in 509 alcohol dependence: implications for cue exposure treatment. Br J Addict 85, 725-510 743. 10.1111/j.1360-0443.1990.tb01685.x. 511 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 25 8. Drummond, D.C., and Glautier, S. (1994). A controlled trial of cue exposure 512 treatment in alcohol dependence. Journal of consulting and clinical psychology 62, 513 809-817. 10.1037//0022-006x.62.4.809. 514 9. Torregrossa, M.M., and Taylor, J.R. (2013). Learning to forget: manipulating 515 extinction and reconsolidation processes to treat addiction. Psychopharmacology 516 (Berl) 226, 659-672. 10.1007/s00213-012-2750-9. 517 10. Vollstadt-Klein, S., Loeber, S., Kirsch, M., Bach, P., Richter, A., Buhler, M., von 518 der Goltz, C., Hermann, D., Mann, K., and Kiefer, F. (2011). Effects of cue -519 exposure treatment on neural cue reactivity in alcohol dependence: a randomized 520 trial. Biological psychiatry 69, 1060-1066. 10.1016/j.biopsych.2010.12.016. 521 11. Milton, A.L., and Everitt, B.J. (2012). The persistence of maladaptive memory: 522 addiction, drug memories and anti-relapse treatments. Neurosci Biobehav Rev 36, 523 1119-1139. 10.1016/j.neubiorev.2012.01.002. 524 12. Cofresi, R.U., Lewis, S.M., Chaudhri, N., Lee, H.J., Monfils, M.H., and Gonzales, 525 R.A. (2017). Postretrieval Extinction Attenuates Alcohol Cue Reactivity in Rats. 526 Alcohol Clin Exp Res 41, 608-617. 10.1111/acer.13323. 527 13. Ray, L.A., Grodin, E.N., Leggio, L., Bechtholt, A.J., Becker, H., Feldstein Ewing, 528 S.W., Jentsch, J.D., King, A.C., Mason, B.J., O'Malley, S., et al. (2021). The future 529 of translational research on alcohol use disorder. Addiction biology 26, e12903. 530 10.1111/adb.12903. 531 14. Roberts-Wolfe, D., Bobadilla, A.C., Heinsbroek, J.A., Neuhofer, D., and Kalivas, 532 P.W. (2018). Drug Refraining and Seeking Potentiate Synapses on Distinct 533 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 26 Populations of Accumbens Medium Spiny Neurons. J. Neurosci. 38, 7100- 7107. 534 10.1523/Jneurosci.0791-18.2018. 535 15. Giannotti, G., Gong, S., Fayette, N., Heinsbroek, J.A., Orfila, J.E., Herson, P.S., 536 Ford, C.P., and Peters, J. (2021). Extinction blunts paraventricular thalamic 537 contributions to heroin relapse. Cell reports 36, 109605. 538 10.1016/j.celrep.2021.109605. 539 16. Yin, H.H., Ostlund, S.B., Knowlton, B.J., and Balleine, B.W. (2005). The role of the 540 dorsomedial striatum in instrumental conditioning. Eur J Neurosci 22, 513- 523. 541 10.1111/j.1460-9568.2005.04218.x. 542 17. Balleine, B.W., Delgado, M.R., and Hikosaka, O. (2007). The role of the dorsal 543 striatum in reward and decision- making. J Neurosci 27, 8161- 8165. 544 10.1523/JNEUROSCI.1554-07.2007. 545 18. Wang, J., Cheng, Y., Wang, X., Roltsch Hellard, E., Ma, T., Gil, H., Ben Hamida, 546 S., and Ron, D. (2015). Alcohol Elicits Functional and Structural Plasticity 547 Selectively in Dopamine D1 Receptor -Expressing Neurons of the Dorsomedial 548 Striatum. J Neurosci 35, 11634-11643. 10.1523/JNEUROSCI.0003-15.2015. 549 19. Wang, J., Ben Hamida, S., Darcq, E., Zhu, W.H., Gibb, S.L., Lanfranco, M.F., 550 Carnicella, S., and Ron, D. (2012). Ethanol -Mediated Facilitation of AMPA 551 Receptor Function in the Dorsomedial Striatum: Implications for Alcohol Drinking 552 Behavior. J. Neurosci. 32, 15124-15132. 10.1523/Jneurosci.2783-12.2012. 553 20. Lu, J., Cheng, Y., Xie, X., Woodson, K., Bonifacio, J., Disney, E., Barbee, B., 554 Wang, X., Zaidi, M., and Wang, J. (2021). Whole- Brain Mapping of Direct Inputs 555 to Dopamine D1 and D2 Receptor -Expressing Medium Spiny Neurons in the 556 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 27 Posterior Dorsomedial Striatum. eNeuro 8 , ENEURO.0348 -0320.2020. 557 10.1523/ENEURO.0348-20.2020. 558 21. Gerfen, C.R., and Surmeier, D.J. (2011). Modulation of striatal projection systems 559 by dopamine. Annu Rev Neurosci 34, 441 -466. 10.1146/annurev-neuro-061010-560 113641. 561 22. Kravitz, A.V., Tye, L.D., and Kreitzer, A.C. (2012). Distinct roles for direct and 562 indirect pathway striatal neurons in reinforcement. Nat Neurosci 15, 816- 818. 563 10.1038/nn.3100. 564 23. Cheng, Y., Huang, C.C.Y., Ma, T., Wei, X., Wang, X., Lu, J., and Wang, J. (2017). 565 Distinct Synaptic Strengthening of the Striatal Direct and Indirect Pathways Drives 566 Alcohol Consumption. Biological psychiatry 81, 918- 929. 567 10.1016/j.biopsych.2016.05.016. 568 24. Hong, S.I., Kang, S., Chen, J.F., and Choi, D.S. (2019). Indirect Medium Spiny 569 Neurons in the Dorsomedial Striatum Regulate Ethanol -Containing Conditioned 570 Reward Seeking. J Neurosci 39, 7206-7217. 10.1523/JNEUROSCI.0876-19.2019. 571 25. Isett, B.R., Nguyen, K.P., Schwenk, J.C., Yurek, J.R., Snyder, C.N., Vounatsos, 572 M.V., Adegbesan, K.A., Ziausyte, U., and Gittis, A.H. (2023). The indirect pathway 573 of the basal ganglia promotes transient punishment but not motor suppression. 574 Neuron. 10.1016/j.neuron.2023.04.017. 575 26. Luscher, C., and Ungless, M.A. (2006). The mechanistic classification of addictive 576 drugs. PLoS Med 3, e437. 10.1371/journal.pmed.0030437. 577 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 28 27. Maltese, M., March, J.R., Bashaw, A.G., and Tritsch, N.X. (2021). Dopamine 578 differentially modulates the size of projection neuron ensembles in the intact and 579 dopamine-depleted striatum. Elife 10, e68041. 10.7554/eLife.68041. 580 28. Bromberg-Martin, E.S., Matsumoto, M., and Hikosaka, O. (2010). Dopamine in 581 motivational control: rewarding, aversive, and alerting. Neuron 68, 815- 834. 582 10.1016/j.neuron.2010.11.022. 583 29. Surmeier, D.J., Shen, W., Day, M., Gertler, T., Chan, S., Tian, X., and Plotkin, J.L. 584 (2010). The role of dopamine in modulating the structure and function of striatal 585 circuits. Progress in brain research 183, 149-167. 10.1016/S0079-6123(10)83008-586 0. 587 30. Roltsch Hellard, E., Binette, A., Zhuang, X., Lu, J., Ma, T., Jones, B., Williams, E., 588 Jayavelu, S., and Wang, J. (2019). Optogenetic control of alcohol -seeking 589 behavior via the dorsomedial striatal circuit. Neuropharmacology 155, 89- 97. 590 10.1016/j.neuropharm.2019.05.022. 591 31. Ma, T., Barbee, B., Wang, X., and Wang, J. (2017). Alcohol induces input-specific 592 aberrant synaptic plasticity in the rat dorsomedial striatum. Neuropharmacology 593 123, 46-54. 10.1016/j.neuropharm.2017.05.014. 594 32. Ma, T., Cheng, Y., Roltsch Hellard, E., Wang, X., Lu, J., Gao, X., Huang, C.C.Y., 595 Wei, X.Y., Ji, J.Y., and Wang, J. (2018). Bidirectional and long- lasting control of 596 alcohol-seeking behavior by corticostriatal LTP and LTD. Nat Neurosci 21, 373-597 383. 10.1038/s41593-018-0081-9. 598 33. Xie, X., Lu, J., Ma, T., Cheng, Y., Woodson, K., Bonifacio, J., Bego, K., Wang, X., 599 and Wang, J. (2023). Linking input- and cell-type-specific synaptic plasticity to the 600 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 29 reinforcement of alcohol -seeking behavior. Neuropharmacology 237, 109619. 601 10.1016/j.neuropharm.2023.109619. 602 34. Bobadilla, A.C., Dereschewitz, E., Vaccaro, L., Heinsbroek, J.A., Scofield, M.D., 603 and Kalivas, P.W. (2020). Cocaine and sucrose rewards recruit different seeking 604 ensembles in the nucleus accumbens core. Molecular psychiatry 25, 3150- 3163. 605 10.1038/s41380-020-00888-z. 606 35. Wang, W., Xie, X., Zhuang, X., Huang, Y., Tan, T., Gangal, H., Huang, Z., 607 Purvines, W., Wang, X., Stefanov, A., et al. (2023). Striatal mu -opioid receptor 608 activation triggers direct -pathway GABAergic plasticity and induces negative 609 affect. Cell reports, 112089. 10.1016/j.celrep.2023.112089. 610 36. Panlilio, L.V., and Goldberg, S.R. (2007). Self -administration of drugs in animals 611 and humans as a model and an investigative tool. Addiction (Abingdon, England) 612 102, 1863-1870. 10.1111/j.1360-0443.2007.02011.x. 613 37. Gardner, E.L. (2000). What we have learned about addiction from animal models 614 of drug self -administration. Am J Addiction 9 , 285- 313. Doi 615 10.1080/105504900750047355. 616 38. Cui, G., Jun, S.B., Jin, X., Pham, M.D., Vogel, S.S., Lovinger, D.M., and Costa, 617 R.M. (2013). Concurrent activation of striatal direct and indirect pathways during 618 action initiation. Nature 494, 238-242. 10.1038/nature11846. 619 39. Fanelli, R.R., Klein, J.T., Reese, R.M., and Robinson, D.L. (2013). Dorsomedial 620 and dorsolateral striatum exhibit distinct phasic neuronal activity during alcohol 621 self-administration in rats. Eur J Neurosci 38, 2637-2648. 10.1111/ejn.12271. 622 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 30 40. Allichon, M.C., Ortiz, V., Pousinha, P., Andrianarivelo, A., Petitbon, A., Heck, N., 623 Trifilieff, P., Barik, J., and Vanhoutte, P. (2021). Cell -Type-Specific Adaptions in 624 Striatal Medium-Sized Spiny Neurons and Their Roles in Behavioral Responses 625 to Drugs of Abuse. Frontiers in synaptic neuroscience 13, 799274. 626 10.3389/fnsyn.2021.799274. 627 41. Sanchis-Segura, C., and Spanagel, R. (2006). Behavioural assessment of drug 628 reinforcement and addictive features in rodents: an overview. Addiction biology 11, 629 2-38. 10.1111/j.1369-1600.2006.00012.x. 630 42. Tecuapetla, F., Jin, X., Lima, S.Q., and Costa, R.M. (2016). Complementary 631 Contributions of Striatal Projection Pathways to Action Initiation and Execution. 632 Cell 166, 703-715. 10.1016/j.cell.2016.06.032. 633 43. Lau, B., and Glimcher, P.W. (2008). Value representations in the primate striatum 634 during matching behavior. Neuron 58, 451-463. 10.1016/j.neuron.2008.02.021. 635 44. Shin, J.H., Kim, D., and Jung, M.W. (2018). Differential coding of reward and 636 movement information in the dorsomedial striatal direct and indirect pathways. 637 Nature communications 9, 404. 10.1038/s41467-017-02817-1. 638 45. Kwak, S., and Jung, M.W. (2019). Distinct roles of striatal direct and indirect 639 pathways in value-based decision making. Elife 8. 10.7554/eLife.46050. 640 46. Cruz, F.C., Rubio, F.J., and Hope, B.T. (2015). Using c -fos to study neuronal 641 ensembles in corticostriatal circuitry of addiction. Brain research 1628, 157-173. 642 10.1016/j.brainres.2014.11.005. 643 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 31 47. Watabe-Uchida, M., Eshel, N., and Uchida, N. (2017). Neural Circuitry of Reward 644 Prediction Error. Annu Rev Neurosci 40, 373- 394. 10.1146/annurev -neuro-645 072116-031109. 646 48. Cox, J., and Witten, I.B. (2019). Striatal circuits for reward learning and decision -647 making. Nat Rev Neurosci 20, 482-494. 10.1038/s41583-019-0189-2. 648 49. van Elzelingen W, G.J., Warnaar P, Denys D, Arbab T, Willuhn I (2022). A 649 unidirectional but not uniform striatal landscape of dopamine signaling for 650 motivational stimuli. Proc Natl Acad Sci U S A 119 , e2117270119. 651 10.1073/pnas.2117270119 652 653 50. Bamford, N.S., Wightman, R.M., and Sulzer, D. (2018). Dopamine's Effects on 654 Corticostriatal Synapses during Reward- Based Behaviors. Neuron 97, 494- 510. 655 10.1016/j.neuron.2018.01.006. 656 51. Han, D.H., Park, P., Choi, D.I., Bliss, T.V.P., and Kaang, B.K. (2022). The essence 657 of the engram: Cellular or synaptic? Seminars in cell & developmental biology 125, 658 122-135. 10.1016/j.semcdb.2021.05.033. 659 52. Gipson, C.D., Kupchik, Y.M., Shen, H., Reissner, K.J., Thomas, C.A., and Kalivas, 660 P.W. (2013). Relapse induced by cues predicting cocaine depends on rapid, 661 transient synaptic potentiation. Neuron 77, 867 -872. 662 10.1016/j.neuron.2013.01.005. 663 53. Cruz, F.C., Babin, K.R., Leao, R.M., Goldart, E.M., Bossert, J.M., Shaham, Y., and 664 Hope, B.T. (2014). Role of Nucleus Accumbens Shell Neuronal Ensembles in 665 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 32 Context-Induced Reinstatement of Cocaine-Seeking. J. Neurosci. 34, 7437-7446. 666 10.1523/Jneurosci.0238-14.2014. 667 54. Caprioli, D., Venniro, M., Zhang, M., Bossert, J.M., Warren, B.L., Hope, B.T., and 668 Shaham, Y. (2017). Role of Dorsomedial Striatum Neuronal Ensembles in 669 Incubation of Methamphetamine Craving after Voluntary Abstinence. J Neurosci 670 37, 1014-1027. 10.1523/JNEUROSCI.3091-16.2016. 671 55. Pribiag, H., Shin, S., Wang, E.H., Sun, F., Datta, P., Okamoto, A., Guss, H., Jain, 672 A., Wang, X.Y., De Freitas, B., et al. (2021). Ventral pallidum DRD3 potentiates a 673 pallido-habenular circuit driving accumbal dopamine release and cocaine seeking. 674 Neuron 109, 2165-2182 e2110. 10.1016/j.neuron.2021.05.002. 675 56. Bobadilla, A.C., Heinsbroek, J.A., Gipson, C.D., Griffin, W.C., Fowler, C.D., Kenny, 676 P.J., and Kalivas, P.W. (2017). Corticostriatal plasticity, neuronal ensembles, and 677 regulation of drug- seeking behavior. Brain Research in Addiction 235, 93- 112. 678 10.1016/bs.pbr.2017.07.013. 679 57. Hong, I., Song, B., Lee, S., Kim, J., Kim, J., and Choi, S. (2009). Extinction of cued 680 fear memory involves a distinct form of depotentiation at cortical input synapses 681 onto the lateral amygdala. Eur J Neurosci 30, 2089- 2099. 10.1111/j.1460-682 9568.2009.07004.x. 683 58. Lee, J.H., Kim, W.B., Park, E.H., and Cho, J.H. (2022). Neocortical synaptic 684 engrams for remote contextual memories. Nat Neurosci. 10.1038/s41593- 022-685 01223-1. 686 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 33 59. Kim, J., Lee, S., Park, K., Hong, I., Song, B., Son, G., Park, H., Kim, W.R., Park, 687 E., Choe, H.K., et al. (2007). Amygdala depotentiation and fear extinction. Proc 688 Natl Acad Sci U S A 104, 20955-20960. 10.1073/pnas.0710548105. 689 60. Pettibone, J.R., Yu, J.Y., Derman, R.C., Faust, T.W., Hughes, E.D., Filipiak, W.E., 690 Saunders, T.L., Ferrario, C.R., and Berke, J.D. (2019). Knock -In Rat Lines with 691 Cre Recombinase at the Dopamine D1 and Adenosine 2a Receptor Loci. Eneuro 692 6. Artn 0163-19.2019 10.1523/Eneuro.0163-19.2019. 693 61. Wei, X., Ma, T., Cheng, Y., Huang, C.C.Y., Wang, X., Lu, J., and Wang, J. (2018). 694 Dopamine D1 or D2 receptor -expressing neurons in the central nervous system. 695 Addiction biology 23, 569-584. 10.1111/adb.12512. 696 62. Lu, J., Cheng, Y., Wang, X., Woodson, K., Kemper, C., Disney, E., and Wang, J. 697 (2019). Alcohol intake enhances glutamatergic transmission from D2 receptor -698 expressing afferents onto D1 receptor -expressing medium spiny neurons in the 699 dorsomedial striatum. Neuropsychopharmacology 44, 1123- 1131. 700 10.1038/s41386-019-0332-9. 701 63. Cheng, Y., Xie, X., Lu, J., Gangal, H., Wang, W., Melo, S., Wang, X., Jerger, J., 702 Woodson, K., Garr, E., et al. (2021). Optogenetic induction of orbitostriatal long-703 term potentiation in the dorsomedial striatum elicits a persistent reduction of 704 alcohol-seeking behavior in rats. Neuropharmacology 191, 108560. 705 10.1016/j.neuropharm.2021.108560. 706 64. Huang, C.C.Y., Ma, T., Roltsch Hellard, E.A., Wang, X., Selvamani, A., Lu, J., 707 Sohrabji, F., and Wang, J. (2017). Stroke triggers nigrostriatal plasticity and 708 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Xie et al. Page 34 increases alcohol consumption in rats. Sci Rep 7 , 2501. 10.1038/s41598- 017-709 02714-z. 710 65. Visser, E., Matos, M.R., van der Loo, R.J., Marchant, N.J., de Vries, T.J., Smit, 711 A.B., and van den Oever, M.C. (2020). A persistent alcohol cue memory trace 712 drives relapse to alcohol seeking after prolonged abstinence. Sci Adv 6, eaax7060. 713 10.1126/sciadv.aax7060. 714 66. Xie, X., Chen, R., Wang, X., Smith, L., and Wang, J. (2023). Activity -dependent 715 labeling and manipulation of fentanyl-recruited striatal ensembles using ArcTRAP 716 approach. STAR Protoc 4, 102369. 10.1016/j.xpro.2023.102369. 717 67. Franklin, K.B.J., and Paxinos, G. (2008). The mouse brain in stereotaxic 718 coordinates, 3rd Edition (Boston : Elsevier/Academic Press). 719 68. Gangal, H., Xie, X., Huang, Z., Cheng, Y., Wang, X., Lu, J., Zhuang, X., Essoh, A., 720 Huang, Y., Chen, R., et al. (2023). Drug reinforcement impairs cognitive flexibility 721 by inhibiting striatal cholinergic neurons. Nature communications 14, 3886. 722 10.1038/s41467-023-39623-x. 723 724 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Figure 1 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Figure 2 -4 0 4 -2 0 2 -4 0 4 -2 0 2 0 200 0 14AUC (peak) * 0 3z-score (peak) * Time (sec) GCaMP 0.38 z C F I OSA Rein Abstinence group Time (sec) Extinction group DA2m 0.5 z D G J Time (sec) Time (sec) 0 2 z-score (peak) * 0 14AUC (peak) n.s. Active presses (30 min) OSA(FR3) Rein OSA (~6 w) Abstinence / Extinction (9 d) .... Rein test AAV Fiber Rec. Rec. A dichroic 535-nm 470-nm LED Photoreceiver B E H Abs Ext0 40 80 Active presses (Rein, 30 min) *** Comparison Abs Ext-80 0 400 AUC (% of OSA) * Abs Ext-80 0 250 z-score (% of OSA) * 0 200Active presses (30 min) OSA(FR3) Rein Pmax T(Pmax) Pmin T(Pmin) AUC(P) T1st active press K L Correlation between GCaMP signal and behaivor -1 1 -1 1 T 1st active press Pmax T(Pmax) Pmin T(Pmin) AUC(P) was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Figure 3 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint Figure 4 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted June 6, 2025. ; https://doi.org/10.1101/2025.06.02.657507doi: bioRxiv preprint