Male Sprague Dawley rats, weighing between 350 and 400 g were procured from Charles River, Kingston, NY, United States. The rats were housed in a controlled setting with a reversed 12-h light/dark cycle with free access to food and water. All self-administration sessions began (∼9:00 a.m. everyday) at the start of the dark phase of the light/dark cycle. All experimental procedures adhered to the guidelines outlined in the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals and were approved by the NIDA (National Institute of Drug Abuse) Animal Care and Use Committee at the Intramural Research Program (IRP).

Rats were first anesthetized using a combination of ketamine (50 mg/kg) and xylazine (5 mg/kg). A polyurethane catheter was surgically inserted into the right jugular vein while the external end was mounted to the back of the rat (Blackwood et al., 2019a). The rats were then given a recovery period of 7 days before beginning self-administration training.

Following an established protocol, drug-naive rats were trained to self-administer oxycodone (0.1 mg/kg/infusion) or saline within a sound-attenuated chamber using a FR1 schedule (Blackwood et al., 2019b). A total of 36 rats were divided into three groups: Saline (Sal) (n = 8), Short-access (ShA) (n = 10), or Long-access (LgA) (n = 18). Short-access rats were allowed to self-administer oxycodone for only one 3-h session for the entirety of the study (days 1–20). Long-access and saline rats were scheduled to self-administer for three sessions using, one 3-h session during days 1–5, followed by two 3-h sessions during days 6–10 and then three 3-h sessions for the rest of the study (days 11–20). The 20-day timeframe was selected based on previous paradigms in the literature that reliably produce escalation and allow for the emergence of compulsive-like drug intake behaviors (Blackwood and Cadet, 2021; Blackwood et al., 2019a; Blackwood et al., 2019b).

We gradually increased access to oxycodone over 4 weeks to prevent any adverse effects of oxycodone intake. There was a 20 s timeout between each infusion. Each 3-h session for LgA and Sal from day 6 to day 20 was separated by a 30-min timeout. This 30-min break was implemented to prevent overdoses as there was no limit to the number of infusions a rat could take during a session. We also included a 48h weekend abstinence period between every 5 days of SA to prevent significant weight loss that might have led to the elimination of some rats from the study. Catheter patency was tested thought the experiment. Rats were euthanized 2 hours after the first session of the last day. Saline animals underwent similar surgical procedures as oxycodone rats, were placed in the identical operant chambers, and experienced similar cue presentations during SA sessions.

Two hours after the final self-administration session, rats were euthanized via rapid decapitation using a guillotine. PFC, NAc and HIP tissues were dissected using precise neuroanatomical coordinates using the Atlas (Paxinos and Watson, 2006) and then immediately snap-frozen on dry ice before being stored at −80 °C (Blackwood et al., 2019b). Total RNA was extracted from tissue using Qiagen RNeasy Mini kit (Qiagen, Valencia, CA, United States). A half microgram (0.5 μg) of total RNA was reverse-transcribed to cDNA with oligo dT primers using Advantage RT-for-PCR kit (Clontech, Mountain View, CA, United States).

qRT-PCR was carried out using a Roche LightCycler 480 II (Roche Diagnostics, Indianapolis, IN) with Luna Universal qPCR SYBR GREEN (NEB Inc, Ipswich, MA) following the manufacturer’s protocol. We purchased gene-specific primers from Integrated DNA Technologies (IDT) (Coralville, IA, United States). These primers were designed using Thermo Fisher Scientific (OligoPrefect Primer Designer software). We normalized mRNA using beta-2 microglobulin (B2M), Clathrin, and ornithine decarboxylase antizyme (OAZ1) as reference genes and the mRNA expression of target genes were reported as fold changes. The primer sequences used for PCR are listed in Supplementary Table S1.

Behavioral data were analyzed with the statistical program GraphPad Prism 10 using factorial ANOVA with repeated measures. Independent variables were the rat reward types (Sal, ShA, LgA-L, LgA-H), within-subject factor SA day (training days 1–20). Oxycodone intake served as the dependent variable. A second-degree polynomial regression model was used to identify potential non-linear patterns in oxycodone intake over 20 days of SA for individual animals, to segregate the rats into LgA-H and LgA-L subgroups. The rats which escalated their intake were termed as LgA-H where those who did not were termed as LgA-L. Biochemical data were analyzed using one-way ANOVA followed by the Tukey’s multiple comparisons test if the main effect was significant. The slopes of all the regression lines were calculated using one-way ANOVA. Statistical significance for all hypothesis tests was set at p < 0.05.

As shown previously (Blackwood et al., 2019a) we analyzed the behavioral data using repeated measures two-way ANOVA with groups (ShA vs. LgA) and training weeks as factors. We observed significant effects for group [F_{(1, 49)} = 37.6, p < 0.0001], training week [F_{(1.994, 97.71)} = 26.56, p < 0.0001], and group × training week [F_{(3, 147)} = 23.51, p < 0.0001]. Post-hoc test showed LgA rats had greater total oxycodone intake than ShA and Sal rats (Figure 1A).

Interestingly, not all LgA rats self-administered oxycodone to the same degree. We performed regression analyses to compare oxycodone acquisition and rate of change of oxycodone intake over time (Blackwood et al., 2019b). We found that some LgA rats significantly escalated their oxycodone intake over the period of 20 days, whereas others did not escalate their intake. Animals that escalated their oxycodone intake over 20 days of SA were called Long-access High (LgA-H), whereas those that did not escalate were named Long-access Low (LgA-L) (Figure 1B). We reanalyzed the behavioral data with four phenotypes LgA-H, LgA-L, ShA, and Sal. Two-way ANOVA showed significant effects for groups [F_{(2, 48)} = 86.30, p < 0.0001], oxycodone intake [F_{(2.594, 124.5)} = 82.67, p < 0.0001], and group × oxycodone intake interaction [F_{(6, 144)} = 34.17, p < 0.0001] (Figure 1B).

Previous studies reasoned that the differences at the behavioral level may be due to different drug-induced molecular neuroadaptations in potassium channel expression between the different phenotypes (McCoy et al., 2021). Building on these findings, we sought to investigate the molecular mechanisms underlying potassium channel differences to better understand their role in OUD vulnerability.

The ANOVA for voltage gated potassium channels (K_v) in the PFC showed significant effects of treatment group on Kcna1 [F_{(3, 25)} = 3.612, p = 0.0271], Kcna3 [F_{(3, 26)} = 3.097, p = 0.0442], Kcna4 [F_{(3, 26)} = 4.735, p = 0.0091], Kcnb1 [F_{(3, 27)} = 3.344, p = 0.0338], Kcnb2 [F_{(3, 23)} = 4.037, p = 0.0192], Kcnd2 [F_{(3, 27)} = 4.028, p = 0.0172], Kcnd3 [F_{(3, 29)} = 22.43, p < 0.0001], Kcnq1 [F_{(3, 28)} = 14.87, p < 0.0001], Kcnq2 [F_{(3, 26)} = 3.615, p = 0.0264], Kcnq3 [F_{(3, 28)} = 6.843, p = 0.0013], Kcnq4 [F_{(3, 27)} = 5.265, p = 0.0054], and Kcnq5 [F_{(3, 26)} = 5.099, p = 0.0066] (Figures 2A–L). For Kcna1 and Kcnb1, LgA-H rats revealed decreased mRNA expression compared to Sal (Figures 2A,D). LgA-L rats showed elevated mRNA levels for Kcna3 and Kcna4 when compared to LgA-H (Figures 2B,C). Likewise LgA-L rats displayed increased expression of Kcnq4, and Kcnq5 when compared to Sal and LgA-H (Figures 2K,L). LgA-L rats and ShA rats exhibited higher expression of Kcnb2 compared to the LgA-H phenotype (Figure 2E). Only ShA animals unveiled increased Kcnd2 expression when compared to Sal and LgA-H rats (Figure 2F), while for Kcnq3 this increase was only if compared to the Sal phenotype (Figure 2J). Kcnd3’s mRNA level was found to be decreased among LgA-H and ShA rats when compared to Sal, whereas LgA-L rats showed increased expression compared to Sal, LgA-H and ShA (Figure 2G). The mRNA level of Kcnq1 was found to be decreased among LgA-H, LgA-L and ShA, in addition there was a significant difference between the magnitude of LgA-L’s downregulation and that of LgA-H and ShA (Figure 2H). We also observed increased expression of Kcnq2 in the PFC of ShA rats than Sal (Figure 2I). No significance was found for Kcna2, Kcna5, Kcna6, Kcnd1, Kcng1, Kcng2, Kcng3, and Kcng4 (Supplementary Figure S1A–H).

We also analyzed the results for calcium-activated potassium channels (K_Ca), ANOVA revealed significant effects for Kcnma1 [F_{(3, 28)} = 34.98, p < 0.0001], Kcnn1 [F_{(3, 28)} = 3.923, p = 0.0186], Kcnn2 [F_{(3, 26)} = 20.96, p < 0.0001], Kcnt1 [F_{(3, 28)} = 3.65, p = 0.0244], and Kcnt2 [F_{(3, 28)} = 4.493, p = 0.0107] (Figures 3A–E). LgA-L revealed increased mRNA expression for Kcnma1 and Kcnn2 in their PFC when compared to Sal, LgA-H and ShA rats (Figures 3A,C). LgA-L also displayed higher expression then LgA-H for Kcnt2 (Figure 3E). Interestingly ShA showed decreased mRNA level for Kcnn1 when compared with LgA-L (Figure 3B) and increased expression for Kcnt1 when compared to LgA-H rats (Figure 3D).

We observed a significant one-way ANOVA results in the rats NAc for Kcnb1 [F_{(3, 27)} = 3.268, p = 0.0365], Kcnb2 [F_{(3, 27)} = 3.785, p = 0.0218], Kcnd1 [F_{(3, 25)} = 5.122, p = 0.0067], Kcnd2 [F_{(3, 28)} = 4.043, p = 0.0166], Kcnd3 [F_{(3, 26)} = 7.580, p = 0.0008], Kcng2 [F_{(3, 29)} = 4.037, p = 0.0163], Kcnq2 [F_{(3, 27)} = 4.868, p = 0.0078], and Kcnq3 [F_{(3, 25)} = 3.841, p = 0.0217] (Figures 4A–H). Moreover, LgA-H rats revealed a significant increase for Kcnb1, Kcnd1, Kcnq2, and Kcnq3 when compared to LgA-L (Figures 4A,C,G,H) and exhibited higher Kcnb2 expression when compared with ShA (Figure 4B). Higher mRNA level for Kcnd2 and Kcng2 was also seen in LgA-H rats than LgA-L and ShA rats (Figures 4D,F). Likewise, only LgA-H rats had higher expression of Kcnd3 than Sal, LgA-L and ShA rats (Figure 4E). No significance changes were found for Kcna1, Kcna2, Kcna3, Kcna4, Kcna5, Kcna6, Kcna10, Kcng1, Kcng3, Kcng4, Kcnq1, Kcnq4, and Kcnq5 (Supplementary Figure S2A–M).

Like that of K_v, K_Ca revealed significant effects of treatment group on Kcnn2 [F_{(3, 29)} = 3.398, p = 0.0309], Kcnt1 [F_{(3, 27)} = 6.787, p = 0.0015] and Kcnt2 [F_{(2, 25)} = 3.499, p = 0.0302] (Figures 5A–C). mRNA expression levels of Kcnn2 and Kcnt2 was higher in LgA-H animals when compared to Sal (Figures 5A,C). LgA-H rats also revealed a higher level for Kcnt1 when compared with LgA-L and ShA rats (Figure 5C). No changes were seen for Kcnn1 (Supplementary Figure S3A).

When looking at K_v in the HIP we found a significant ANOVA effect of treatment group on Kcna2 [F_{(3, 29)} = 10.08, p = 0.0001], Kcna5 [F_{(3, 28)} = 5.066, p = 0.0063], Kcna10 [F_{(3, 27)} = 3.778, p = 0.0220], Kcng1 [F_{(3, 29)} = 3.342, p = 0.0328], Kcng2 [F_{(3, 30)} = 3.132, p = 0.0401], and Kcnq3 [F_{(3, 30)} = 3.834, p = 0.0195] (Figures 6A–F). Further LgA-H rats displayed significant increase in the expression of Kcna2 compared to Sal, LgA-L and ShA (Figure 6A) and for Kcnq3 then ShA (Figure 5F). Whereas ShA animals revealed higher mRNA level for Kcna5 and Kcna10 than LgA-H and LgA-L (Figures 5B,C) and for Kcng1 only when compared to LgA-L (Figure 5D). While LgA-L phenotype revealed elevated mRNA level for Kcng1 when compared with LgA-H rats (Figure 5E). No significance difference was found for Kcna1, Kcna3, Kcna6, Kcnb1, Kcnb2, Kcnd1, Kcngd2, Kcnd3, Kcng3, Kcng4, Kcnq1, Kcnq2, Kcnq4, and Kcnq5 (Supplementary Figure S4A–N).

Moreover, we also saw a significant effect for Kcnma1 [F_{(3, 30)} = 5.024, p = 0.0061], Kcnn1 [F_{(3, 29)} = 4.851, p = 0.0074], and Kcnn2 [F_{(3, 29)} = 5.134, p = 0.0057] (Figures 7A–C). ShA rats displayed significant higher mRNA levels for Kcnn1 and Kcnn2 than LgA-H rats (Figures 7B,C). Interestingly, expression of Kcnma1 mRNA found to be significantly increased in the LgA-H group compared to Sal, LgA-L and ShA (Figure 7A). No significance was found for Kcnt1, Kcnt2 and Kcnn3 (Supplementary Figure S5A–C).

In the present study, we observed significant differences among the experimental groups, including: (1) LgA-H compared to LgA-L and/or ShA; (2) LgA-L compared to LgA-H and/or ShA; and (3) ShA compared to LgA-H and/or LgA-L. These differences may reflect inherent phenotypic variability or adaptive processes not directly attributable to the self-administration (SA) procedure. Rather than representing direct drug-induced molecular changes, such variability could result from pre-existing individual differences or secondary effects of prolonged drug exposure and behavioral stratification. Therefore, we propose that future studies should not limit comparisons to control groups alone but also include contrasts among drug-exposed groups to better capture the contribution of intrinsic variability and adaptive mechanisms to substance use disorder.

Rats that took the highest levels of oxycodone, LgA-H, were suspected to be the most vulnerable to OUD. They showed a very interesting expression pattern for potassium channels in their NAc. The Nucleus Accumbens is a key hub for reward circuitry involved in drug-taking behaviors (Kuhn et al., 2014), and can enhance the reinforcing effects of opioids (Cornish et al., 1999). It also plays a significant role in processing pleasurable experiences like eating, drug use, and social interactions (Bassareo and Di Chiara, 1999; Carelli, 2002; Olsen, 2011). Potassium channels in the NAc regulate reward behavior by influencing synaptic plasticity (Fernández-Fernández and Lamas, 2021; Kim and Hoffman, 2008). Potassium channels help control inhibitory signaling within the brain (Qiu et al., 2025), which is crucial for suppressing impulsive behaviors. Dysfunction in potassium channel activity can lead to hyperexcitability of neurons, heightening drug-cue reactivity and impairing self-regulation (Humphries and Dart, 2015; Wu et al., 2024). The symptoms you can expect to see because of this dysregulation are irregular heartbeats (Fanoe et al., 2009; Meents et al., 2018) respiratory depression (Montandon et al., 2016; Wei and Ramirez, 2019) and antinociceptive affects (Nakamura et al., 2014). Dysregulation of potassium channels can also lead to seizures (Gross et al., 2016; Zhang et al., 2021), autism (Liu et al., 2022), and ataxia (Pollini et al., 2020).

In the NAc, LgA-H rats revealed consistent upregulation of potassium channels, when compared to LgA-L rats, supporting the link between potassium channel expression and increased vulnerability to OUD. There was higher level of Kcnd2, Kcnd3, Kcng2 and Kcnt1 in NAc of LgA-H rats compared to ShA and LgA-L and expression of Kcnn2 and Kcnt2 was different then Sal. Moreover, administration of a potassium channel inhibitor in the NAc of rats was found to attenuate cocaine seeking behaviors in them (Xia et al., 2024), further supporting their role in addiction-related neuroadaptations. Additionally, increased potassium channels in the NAc would cause a hyperexcitable reward circuit making it easier to succumb to oxycodone abuse (Yuferov et al., 2018). These findings suggest that reducing the frequency of potassium channels in the NAc, could serve as a treatment for OUD.

We thought studying molecular neuroadaptations in LgA-L rats might be interesting because they failed to increase their oxycodone intake despite having the same extended access as the LgA-H group. LgA-L phenotypes displayed higher mRNA levels for Kcna3, Kcna4, Kcnd3, Kcnq4, Kcnq5, Kcnma1 and Kcnn2 in their PFC. The PFC is a brain region of significant interest for its role in decision making, cognitive function (Bausch et al., 2015; Yu et al., 2016; Yu et al., 2019), and impulse control (Arnsten, 2009; Ramnani and Owen, 2004). The PFC, along with subcortical circuits, also plays a crucial role in self-control, and social behavior, influencing both drug-cue reactivity and the regulation of craving and drug-seeking in substance use disorders (Daiwile et al., 2022; Jasinska et al., 2015). Potassium channels control the flow of potassium ions across cell membranes, which is essential for maintaining the resting membrane potential. This means they can play a significant role in regulating self-control and decision-making, by influencing the excitability of neurons (Greene and Hoshi, 2017; Humphries and Dart, 2015). By stabilizing neuronal excitability, potassium channels play a vital role in maintaining balanced neural activity, necessary for recovery in individuals with substance use disorders.

The results we found in the PFC contrasts the pattern we saw in the NAc, where upregulated potassium channel expression was associated with LgA-H rats. The connection between high potassium channel expression in the PFC and low intake behavior suggest the presence of a compensatory mechanism that can reduce a subject’s susceptibility to OUD. Previous studies have shown that enhanced potassium channel activity in the PFC is linked to reduced compulsive drug use (Buchta and Riegel, 2015). Studies on the alcohol consumption in drosophila (Cavaliere et al., 2012), rodents (Padula et al., 2015), and humans (Clarke et al., 2011) have found K_Ca and K_v levels to be decreased in compulsive animals (Cannady et al., 2018), meaning increasing their levels may decrease OUD susceptibility. Other studies show that acute or chronic drug exposure can decrease potassium channel expression in the PFC (Dong et al., 2005). Taken together, these results show that increased expression of potassium channels in the PFC might lead to a decreased oxycodone intake in LgA-L rats. More importantly, these discoveries highlight the PFC as another fundamental brain region where potassium channel modulation plays a role in the neurobiology of OUD (Padula et al., 2015) and suggest that targeting potassium channels may be an advantageous therapeutic approach.

Lastly, we investigated neuro-molecular changes in the brains of ShA rats, as this group can give us insights into molecular adaptations that are due to the mere exposure of oxycodone. The results show multiple K_v (Kcna5, Kcna10, Kcng1) and K_Ca (Kcnn1, Kcnn2) to be upregulated in the Hip of ShA rats. The hippocampus is responsible for the rats’ abilities in learning and memory (O'Dell et al., 2015), functions that we know are impacted by oxycodone intake (Cherrier et al., 2009). Specifically, the VTA to hippocampus neuronal projection plays a vital role in the primary reward circuit and is even more essential for memory formation (Atweh and Kuhar, 1977; Bird and Burgess, 2008). Potassium channels contribute to these functions by maintaining the balance between excitation and inhibition in this circuit (Tsuboi et al., 2024). By maintaining proper neuronal excitability, they ensure that signals associated with reward experiences are accurately processed. Dysfunction in this system will contribute to impaired reward learning (Faulkner et al., 2024), potentially influencing behaviors seen in substance use disorder. This result is important to note as these same genes were not seen to be impacted in other brain regions by any behavioral group, indicating that their activation is linked to a protective neural state. This implies a connection between these genes and an early neurobiological response that reduces vulnerability to compulsive drug-seeking behavior. Moreover, the absence of these changes in LgA-H rats implies that the loss of this protective mechanism contributes to addiction progression.