Shifted Balance Between Ventral Striatal Prodynorphin and Proenkephalin Biases Development of Cocaine Place Avoidance

Cocaine use is rising in the United States and current pharmacotherapies are limited. Cocaine use disorder reflects both positive reinforcement (euphoria) and negative reinforcement (withdrawal-induced aversion). Acute intoxication can also provoke an aversive affective state (paranoia, panic) in a subset of users, potentially reducing subsequent use or motivating co-use with opioids. While rodent models robustly capture reward (e.g., CPP, self-administration), modeling acute aversion is less developed. Prior work shows that trace conditioning—presenting the context before drug administration—can yield CPA for ethanol, nicotine, and amphetamine, suggesting drugs possess bivalent affective states depending on timing. Striatal opioid systems are implicated: dynorphin/KOR signaling is linked to aversion and relapse after withdrawal, whereas enkephalin/MOR supports cocaine reward. The study aimed to: (1) identify parameters to produce cocaine CPA via trace conditioning by varying dose (15, 20, 25 mg/kg), session length (15 vs 30 min), and apparatus (two- vs three-chamber), keeping interstimulus interval minimal; and (2) test whether individual differences in CPA are related to the striatal balance of prodynorphin (Pdyn) and proenkephalin (Penk), hypothesizing higher ventral striatal Pdyn would predict CPA.

Rodent and human data indicate acute cocaine can elicit negative affect (anxiogenic responses, panic, paranoia). Runway models show approach-avoidance conflicts to cocaine-paired cues, but disentangling intoxication vs withdrawal aversion is challenging. Place conditioning with simultaneous pairing typically yields CPP for cocaine; pairing at longer latencies (e.g., 15 min pre-placement) can yield CPA, attributed to falling brain levels and withdrawal. Trace/delay conditioning—context preceding drug—has conditioned CPA for ethanol, nicotine, and amphetamine, and CPA depends on short CS–US intervals. The striatal dynorphin/KOR system increases after repeated cocaine and mediates stress-induced reinstatement; KOR activation is aversive in ventral striatal subregions. MOR signaling in the ventral striatum is essential for cocaine reward; enhancing enkephalin facilitates CPP, and MOR antagonism blocks CPP. These lines suggest ventral striatal dynorphin may promote aversion/relapse (particularly after withdrawal), while enkephalin/MOR supports acute reward.

Animals: 103 adult male and female mice (8–20 weeks; C57BL/6J background) from in-house transgenic lines: 41 Adora2a-Cre−/−, 2 MORf/f, 19 VGat:Cre, 41 VGluT2:Cre. A subset (Adora2a-Cre−/− = 24) and six saline-treated MORf/f controls were used for qPCR. For causal testing, D2-PenkKO mice were generated by crossing Adora2a-Cre+ with Penk−/− to selectively delete Penk from striatal D2-MSNs; 17 D2-PenkKO and 17 Adora2a-Cre−/− Penkf/f littermate controls were used. Mice were group-housed with ad libitum food/water, 12:12 light/dark. Drugs: Cocaine HCl in sterile saline (0.9%), i.p., 10 mL/kg volume. Doses tested: 15, 20, 25 mg/kg, chosen from doses known to induce CPP. Conditioned Place Avoidance—Apparatus: Two-chamber (Med Associates) with distinct tactile cues (grid vs hole) separated by guillotine door; three-chamber (Stoelting) with two conditioning contexts differing in tactile (smooth vs textured) and visual cues (vertical white stripes vs plain black), connected by smaller neutral chamber. Procedure—Experiment 1: Trace conditioning with immediate post-context i.p. injection. Two groups: 15-min sessions in two-chamber or 30-min sessions in three-chamber. Subjects confined to one side for the session, removed, injected with cocaine or saline, returned to home cage. Cocaine and saline days alternated over 2 weeks; total 8 sessions (4 cocaine, 4 saline). Unbiased assignment; drug-context pairing, side, and order counterbalanced. Preference tests (30 min) occurred pre-conditioning (Pretest) and after sessions 4 and 8 (Post-Test 1 and 2); after each test, mice received saline. Outcome: percent time on cocaine-paired floor. Doses assessed in separate cohorts across both apparatus groups. Experiment 2: D2-PenkKO and littermate controls conditioned to 25 mg/kg cocaine in two-chamber with 15-min sessions as above; outcomes computed identically. qPCR: After Post-Test 2, mice (25 mg/kg cocaine conditioned Adora2a-Cre−/−, n=24; saline MORf/f controls, n=6) were euthanized; dorsal and ventral striatum dissected on ice. RNA extraction (Qiagen RNeasy), cDNA synthesis (BioRad iScript). TaqMan assays (ThermoFisher) on CFX384: Penk (Mm01212875_m1), Pdyn (Mm00457573_m1), normalized to β-actin (Actb, Mm01205647_g1). Cycling: 95°C 20 s; 40 cycles of 95°C 1 s, 60°C 20 s. Triplicate runs, negative controls; relative expression via ΔΔCT. Statistics: Place conditioning analyzed as percent time on cocaine floor using multifactorial ANOVAs: repeated measure factor Test (Pretest, Post-Test 1, Post-Test 2); between-subjects factors Chamber type and Dose; zone analyses in three-chamber (conditioned vs unconditioned neutral zone) via two-way ANOVAs; sex effects via two-way ANOVA after collapsing across dose/chamber. Linear regression tested dose vs preference. Individual differences captured by stratifying subjects using ±1 SD from pretest mean to define neutral zone; groups: preferers (≥ upper threshold), avoiders (≤ lower threshold), neutral. Group proportion changes via Chi-square or Fisher’s exact test. Penk and Pdyn associations with place conditioning via Pearson’s correlation; Penk:Pdyn ratios correlated with preference. Two-way repeated-measures ANOVAs for gene (Penk vs Pdyn) × group (preferer/avoider/neutral) separately in ventral and dorsal striatum. For Experiment 2, generalized linear model with inverse Gaussian distribution (Wald Chi-square) compared pretest vs Post-Test 2; post hoc analyses with Sidak corrections; alpha=0.05.

• Manipulated parameters: Three-way RMANOVA showed mean percent time on cocaine-paired floor was stable and did not significantly vary by dose or apparatus across tests (Test × Chamber × Dose: F4,230 = 2.14, p = 0.077).
• Chamber × Test interaction (F2,230 = 7.1, p < 0.01) driven by pretest difference between two- vs three-chamber in the cohort later receiving 20 mg/kg, indicating sampling bias rather than true chamber effect (two-chamber vs three-chamber at pretest: t357 = 3.98, p < 0.001).
• Main effect of dose (F2,115 = 3.43, p < 0.05) without significant post hoc between-dose differences.
• In three-chamber apparatus at Post-Test 1, higher cocaine dose predicted greater avoidance (linear regression: F1,58 = 6.6, p < 0.05; R² = 0.10); no dose–preference relationship at Post-Test 2 for either apparatus.
• Time allocation changes reflected less time in cocaine-paired zone and more in saline-paired zone; neutral chamber time did not drive CPA (Test × Zone: F4,66 = 5.7, p < 0.001).
• Variability increased over conditioning: coefficient of variation rose from 20.8% (Pretest) to 32.5% (Post-Test 2). Stratification (Post-Test 2 mean 51.02%, SD 10.75) defined avoiders ≤ 40.27%, preferers ≥ 61.77%, neutral 40.28–61.76%.
• Group proportions shifted significantly from Pretest to Post-Test 2 (χ² [2, n = 103] = 51.08, p < 0.0001): Pretest—Preferers 12.6%, Avoiders 9.7%, Neutral 77.7%; Post-Test 2—Preferers 25.2%, Avoiders 29.1%, Neutral 45.6%. Apparatus type did not affect proportions (χ² [2, n = 103] = 1.08, p = 0.58). No sex differences (F1,95 = 1.34, p = 0.25).
• qPCR correlations: Penk mRNA did not correlate with preference in dorsal or ventral striatum (R² DS = 0.095; VS = 0.0001). Pdyn mRNA positively associated with cocaine preference in ventral striatum (R² = 0.2004, p < 0.05), not dorsal (R² = 0.055).
• Penk:Pdyn ratio in ventral striatum negatively correlated with preference (R² = 0.52, p < 0.001); no correlation in dorsal striatum.
• Ventral striatum gene expression differed by preference group (Gene × Group: F2,17 = 9.7, p < 0.01): avoiders had higher Penk relative to Pdyn (t17 = 4.32, p < 0.01); no Penk–Pdyn imbalance in preferers or neutrals. In dorsal striatum, Pdyn slightly elevated vs Penk across groups (Gene: F1,21 = 7.03, p < 0.05). β-actin control unchanged (saline vs cocaine: 20.09 ± 0.14 vs 19.95 ± 0.09; t113 = 0.78, p = 0.44).
• Experiment 2 (D2-PenkKO): GLM revealed genotype × test interaction (Wald χ² = 5.73, df = 1, p < 0.05). By Post-Test 2, Penkf/f controls exhibited significant CPA vs pretest (p < 0.05), whereas D2-PenkKO did not (p = 0.12). Genotypes differed at Post-Test 2 (p < 0.05), indicating higher relative Pdyn (via Penk deletion) mitigated CPA. Proportions shifted within genotypes across tests (Penkf/f: p < 0.0001; D2-PenkKO: p < 0.01); between-genotype proportion differences at Post-Test 2 were not significant (p = 0.12), potentially limited by strata sample sizes.

Trace conditioning revealed that acute cocaine can produce bivalent motivational states: roughly equal subsets of mice developed preference or avoidance. Dose, apparatus type, and session length had minimal and transient impact on CPA expression, supporting generalizability of the procedure across common apparatuses and parameters. The pattern of effects and dependence on short CS–US intervals support interpretation that CPA reflects Pavlovian association between a transient, aversive state immediately following cocaine administration and the preceding context, rather than neophobia or conditioned inhibition. The striatal opioid milieu appears to bias these individual differences: in ventral striatum, lower Pdyn and higher Penk:Pdyn ratios were associated with stronger CPA, while higher relative Pdyn (as modeled by D2-PenkKO) reduced CPA development. These findings suggest opposing roles for dynorphin in acute intoxication versus withdrawal contexts—high dynorphin tone is implicated in withdrawal-induced negative affect and relapse, but low ventral striatal Pdyn may predispose to acute aversion after intoxication. Enkephalin/MOR signaling supports cocaine reward; thus, the balance between enkephalin and dynorphin shapes whether acute cocaine intoxication is experienced as rewarding or aversive, with implications for risk, resilience, and potential co-use of opioids to counteract aversion.

Trace conditioning effectively captures individual differences in acute cocaine-induced preference and avoidance, identifying subjects sensitive to a transient aversive state immediately after cocaine. The balance between ventral striatal enkephalin and dynorphin is a key correlate: low Pdyn (and high Penk:Pdyn) associates with CPA, whereas higher relative Pdyn prevents avoidance. These data suggest dynorphin may have opposing roles in acute intoxication vs withdrawal. The approach can aid discovery of neurobiological markers predictive of vulnerability or protection against cocaine abuse and may inform strategies to address cocaine–opioid co-use. Future work should parse subregion-specific dynamics within the ventral striatum, measure peptide levels and release in real time, manipulate KOR/MOR signaling acutely during testing, and systematically assess dose and timing parameters to optimize CPA expression.

qPCR measured mRNA, not peptide levels or real-time peptide release; mRNA changes may not perfectly reflect peptide dynamics. Striatal dissections did not resolve nucleus accumbens subregions (core vs shell), limiting spatial specificity. It remains unclear whether Pdyn/Penk differences preceded conditioning or emerged with cocaine exposure; both interpretations may hold. Experiment 2 had smaller sample sizes and only tested the highest dose, potentially limiting generalization and statistical power to detect genotype differences in group proportions. The 30-min conditioning was only assessed in the three-chamber apparatus, confounding session length with apparatus type. Real-time assessments of dynorphin tone or MSN activity (e.g., peptide sensors, calcium imaging) were not performed.