Behavioral flexibility and gut microbiome as potential predictors of oral oxycodone self-administration

Abstract

Prescription opioids such as oxycodone have been widely used in the United States and have contributed to the ongoing opioid epidemic. While many individuals limit use to prescribed contexts, a subset transitions to misuse and, in some cases, to illicit opioid use. Identifying behavioral and biological factors that predict this vulnerability is critical for improving prevention and intervention strategies. Here, we investigated whether individual differences in behavioral flexibility and gut microbiome composition are associated with future oxycodone intake using a translationally relevant model of oral oxycodone self-administration in male and female Long-Evans rats. We established a model in which distinct intake phenotypes emerged, characterized by animals with high versus low oxycodone consumption. Behavioral flexibility, assessed using a contingency degradation task, was associated with oxycodone intake, identifying it as a potential behavioral biomarker of vulnerability. In parallel, oral oxycodone exposure altered gut microbiome composition, and microbiome features were associated with both behavioral flexibility and drug-taking behavior. These findings support a framework in which individual differences in opioid intake arise from the interaction of pre-existing behavioral traits and biological states, including gut microbiome composition which provides a foundation for identifying predictive biomarkers and developing individualized strategies to mitigate risk for opioid misuse.

INTRODUCTION

Prescription opioids such as oxycodone are a main driver of the current opioid epidemic in the United States. While prescription opioids are a critical factor influencing the opioid epidemic, many individuals go no further than prescription use while a subset of individuals begin illicit use of prescription opioids and/or transition to heroin or fentanyl^1–5. Prescription opioids such as oxycodone are commonly administered and consumed orally in humans. However, the majority of preclinical oxycodone self-administration studies have been utilized intravenous short (2 hours)⁶ or long (6 hours)^7–15 access oxycodone self-administration paradigms. As such, there is little known about the motivation for oral oxycodone (i.e., schedules of reinforcement). Individual differences in drug-taking behaviors may be dependent on prior differences in behavioral and biological factors¹⁶. While a history of drug use impairs goal-directed behavior in behavioral flexibility tasks^17–20, it remains unclear the relationship between intrinsic differences in behavioral flexibility and future drug taking behaviors. Thus, it is critical to characterize a preclinical model of oral oxycodone intake to investigate clinically relevant features of prescription-derived oxycodone access to develop biomarkers and effective therapeutics for substance use disorders.

An emerging mediator of substance use disorders is the gut microbiome. Dysbiosis of the gut microbiome has been linked to substance use disorders in humans^21,22 and preclinical studies using rodents for multiple drugs of abuse such as cocaine^23,24, fentanyl²⁵, morphine²⁶, and oxycodone²⁷. Notably, these studies used antibiotics to decrease the gut microbiome in intravenous or experimenter-delivered drug studies^23–27. Therefore, it is unknown how the baseline gut microbiome (i.e., without antibiotics) affects oral oxycodone self-administration, and how this may change over the course of oral oxycodone self-administration. Therefore, the studies here are critical as animals are self-administering and consuming oxycodone voluntarily and oxycodone is being delivered directly into the gastrointestinal tract to influence microbiome.

Here, we utilized an oral oxycodone self-administration paradigm to investigate the potential of a behavioral flexibility task²⁸ as a behavioral biomarker of future oral oxycodone intake. Additionally, we investigated the overall effects of oral oxycodone on gut microbiome bacteria levels and the relationship between the baseline gut microbiome behavioral flexibility, and oral oxycodone intake.

MATERIALS AND METHODS.

Subjects.

Male (weight upon arrival: 225–300 g, age in weeks at start of self-administration: 12–15, n=34) and female (weight upon arrival: 175–225 g, age in weeks at start of self-administration: 12–15, n=43) adult Long-Evans were purchased from Envigo (Indianapolis, IN, USA). Animals were maintained on a reverse 12:12 hour light/dark cycle (lights off at 9 A.M., lights on at 9:00 P.M.). During water magazine training and initial oral oxycodone self-administration, animals were maintained at 85% of their body weight by regulating water access (females: 17.5–22.25 mL of water; males: 20.5–25.5 mL of water; [ad libitum food]). We have previously reported contingency degradation data from a subset of rats (n=50; 21 males, 29 females)²⁸. We have not previously reported the oral oxycodone self-administration or gut microbiome data. All procedures were approved by the Rowan University Institutional Animal Care and Use Committee (IACUC) and conducted in accordance with the US Public Health Service Guide for the Care and Use of Laboratory Animals.

Behavioral training and testing.

Animals were trained and tested in standard Med Associates operant chambers as described previously²⁸. All chambers were equipped with two spatially distinct levers with unilluminated cue lights above and an internal food receptacle between the spatially distinct levers on one side of the chamber that was used for all instrumental conditioning, contingency degradation training, and contingency degradation test day. All chambers were also equipped with two spatially distinct nose poke ports with cue lights and an internal trough located between the nose poke ports on the opposite side of the chamber, which was used for all operant oral oxycodone self-administration sessions. During the contingency degradation portion of the experiment, responses in the nose poke ports were without consequence. During the operant oral oxycodone self-administration portion of the experiment, movement into the external food trough was without consequence. Animals underwent instrumental conditioning, contingency degradation training and testing in an operant chamber. Animals were then moved over one box to a new chamber for oral oxycodone self-administration and underwent all self-administration sessions in the new box.

Contingency Degradation.

Animals were first assessed for their behavioral flexibility using a contingency degradation task^28–32 to determine if performance on this task is associated with future oral oxycodone intake. In this task, an action (i.e., left lever press) was degraded, and a different action (i.e., right lever press) remained intact after stable lever pressing is established. An animal’s ability to update this information in response to changes in their environment provides us with a measure of an animal’s inherent behavioral flexibility. Estrous cycle monitoring was conducted throughout this phase of the experiment²⁸.

Instrumental conditioning.

As described previously²⁸, animals underwent magazine training for 2 days to receive dustless precision sucrose pellets (Bioserve, Flemington NJ; males: 45 mg; females: 20 mg) into the internal food receptacle over 30 trials with a variable intertrial interval. During instrumental conditioning, animals underwent one training session per day (68 minutes), with each session consisting of four 15-minute blocks with a two-minute inter-block interval. During each block, one lever was extended. The lever extended was alternated between the two levers in each block (e.g., Block 1-left, Block 2-right, Block 3-left, Block 4- right, left and right as the starting lever, counterbalanced). Rats were first trained on a fixed ratio (FR) 1 schedule until they earned 50 pellets per lever over two days, followed by FR5 training (50 pellets per lever). Lastly, animals were trained on a variable ratio (VR) 9 schedule (range, 5–13 presses) for 4–5 days.

Contingency degradation training.

Contingency degradation training consisted of a daily session (68 minutes) for 4 days as described previously²⁸. One of the two levers was randomly selected as the “contingent” or “non-degraded” lever and outcome delivery (sucrose pellet) was dependent upon the lever press under a VR9 schedule of reinforcement, while the other lever was selected as the “non-contingent” or “degraded” lever and outcome delivery did not depend on the animal’s actions (i.e., lever press; sucrose pellets were delivered every 30–40 seconds).

Contingency degradation test day.

Next, animals were tested under extinction on their behavioral flexibility (i.e., ability to learn and update the new information of lever contingencies). Both levers were extended, and lever presses were recorded but did not result in the delivery of a pellet for 15 minutes.

Oral Oxycodone Self-Administration.

Animals underwent magazine training and oral oxycodone self-administration in standard Med Associates operant chambers as above; however, animals were not in different chambers from their contingency degradation test. Chambers were equipped with two spatially distinct nose pokes and an internal water trough between the two nose pokes. One nose poke was designated as the active hole. Nose poking into the active hole resulted in the presentation of an illuminated cue light (20 s) within the nose poke port and delivery of water or oxycodone into the water trough dependent on the fixed-ratio (FR) schedule and stage of training. The second nose poke port was designated as the inactive hole to control for selectivity for the drug-associated nose poke port and general locomotor behavior. Nose poking into the inactive hole resulted in no consequence. An automated syringe pump (PHM-100VS-2, Med Associates) was located outside of each chamber and tubing connected the syringe to the internal trough such that nose pokes into the active hole, dependent on the FR schedule, were delivered into the internal water trough. The delivery of one reinforcer resulted in 0.1 mL (males) or 0.07 mL (females) of water or oxycodone into the internal trough. If an animal left liquid (i.e., water or oxycodone) in the internal trough at the end of the session, the amount was collected and measured to determine the total volume (water, mL) or amount (oxycodone, mg/kg) consumed. All animals were water regulated prior to the start of magazine training (1–2 days), throughout magazine training (3 days), and for the first four days of self-administration for a total of 8 to 9 days of total water regulation.

Magazine training.

Rats were water regulated one to two days prior to the start of magazine training to increase exploratory behavior and engagement with the chamber. All animals underwent magazine training for three days on an FR1 schedule where a nose poke into the active hole resulted in no cue illumination but the presentation of water into the water trough. Nose pokes into the inactive hole resulted in no consequence.

Water or oral oxycodone self-administration.

Following magazine training, rats underwent water (n=31) or oral oxycodone (n=46) self-administration for 2 hours, 5 days/week (2 days off) for a total of 21 self-administration days (Fig. 1). From self-administration days 1–10, animals were allowed to self-administer for 2 hours or until they reached 50–60 reinforcers to reduce the chance of an animal overdosing during initial acquisition of oral oxycodone self-administration. From self-administration days 11–21, there was no maximum number of reinforcers and animals were allowed to self-administer for a total of 2 hours regardless of the amount of reinforcers earned. Animals were on an FR1 schedule for self-administration days 1–13 where one nose poke into the active hole resulted in cue illumination and delivery of either water or oxycodone into the internal trough. All animals were then switched to an FR3 schedule for days 14–21 of self-administration where 3 nose pokes resulted in delivery of water or oxycodone into the water trough. After a reinforcer was earned, there was a 20 second time-out period. All animals were water regulated for the first 4 days of self-administration and at the end of the 4th day of self-administration were given free access to water. For the first 6 days of self-administration, the oxycodone dose was maintained at 0.3 mg/ml for all animals³³. From days 7–10, a subset of animals (n=12) self-administered for 1 mg/ml of oxycodone and were water regulated for these four days to ensure animals would poke for the reinforcers. This dose was chosen to determine if an increased dose during early self-administration would influence self-administration for a lower dose during later self-administration. All other oxycodone animals (n=34) were maintained at an oxycodone dose of 0.3 mg/ml from self-administration days 7–10. From self-administration days 11–21, all animals self-administered at a dose of 0.1 mg/ml³³. Stability for responding into the active hole during last four days of self-administration (SAD18–21) based on the criteria that there was less than about 35% variability in active hole responses and less than about 20% variability in reinforcers earned during these days.

Figure 1. Self-administration timeline.

All groups were water regulated prior to self-administration training (dark blue) and underwent magazine training (light blue) to obtain a water reinforcer prior to the start of oral oxycodone or water self-administration. All oral oxycodone animals self-administered oxycodone at a dose of 0.3 mg/kg from self-administration day (SAD) 1 to 7. A subset of animals (n=12) self-administered oral oxycodone at a dose of 1 mg/kg from SAD7 to 10, while the majority of the animals (n= 34) self-administered oral oxycodone at a dose of 0.3 mg/kg from SAD7 to SAD10. The dose was decreased to 0.1 mg/kg for both groups at SAD10. All groups self-administered oral oxycodone at a dose of 0.1 mg/kg from SAD11 to SAD21. All groups were moved to a fixed-ratio (FR) 3 schedule of reinforcement on SAD14 to SAD21. Fecal samples were collected prior to self-administration (Pre) and during the last days of self-administration (Post).

Low or high-taking phenotypes during oral oxycodone self-administration.

Oxycodone-exposed animals were separated into a low or high intake group as previous studies have shown distinct phenotypes emerge during intravenous oxycodone self-administration^34–36. To determine the low or high phenotype, the average amount of oxycodone consumed during the last four days of self-administration, when baseline responding was reached, was calculated for each animal. The amount consumed for each animal was calculated based on body weight, number of reinforcers earned, dose (0.1 mg/kg), and amount of oxycodone (mL) left in the internal trough. There was no significant difference in the median amount consumed between males and females (unpaired t-test; t(44) = 0.46, p = 0.65), so males and females were combined and the median for all animals was taken. Animals were then split where any animal at (n=1) or below (n=22) the median was classified as an oxycodone low animal (n=23) and any animal above the median (n=23) was classified as an oxycodone high animal.

Drug.

Oxycodone HCl was purchased from Sigma Aldrich (#1485205). Oxycodone was dissolved in tap water and filtered (#565–0010, Thermo Fisher Scientific) to create a stock solution of 100 mg/ml. Dilutions of the stock solution were made using tap water based on the stage of self-administration for a final dose upon delivery of: 0.3 mg/ml, 1 mg/ml, or 0.1 mg/ml.

Estrous cycle stage classification.

Estrous cycle was determined in freely cycling females during self-administration. Vaginal swabs were taken from all female rats immediately before each behavioral session. As described previously by us and other laboratories^6,28,37–40, vaginal samples were collected by gently swabbing the vaginal canal using a saline-dipped, cotton tipped applicator. Samples were then smeared on glass microscope slides. Males were handled on an identical schedule. Estrous cycle stage was determined by the presence and morphology of cells^41,42. Each smear sample was classified into one of four stages: proestrus, estrus, metestrus (diestrus I), or diestrus (diestrus II). The proestrus stage was identified by the presence of ≥75% of nucleated epithelial cells, the estrus stage (vaginal estrus) was classified by the presence of ≥75% of non-nucleated cornified cells, the metestrus stage was identified by the presence of equal amounts of nucleated epithelial cells, non-nucleated cornified epithelial cells, and leukocytes, and the diestrus stage was classified by the observation of fewer cells and the presence of leukocytes and epithelial cells. All rats were observed to have a typical 4–5 day estrous cycle during self-administration.

Estrous cycle analysis.

We investigated the effects of the estrous cycle on the last 4 days of oral oxycodone self-administration, when stable active hole responding was reached. Additionally, since self-administration occurred 5 days/week with 2 days off we did not capture proestrus in some of the females during the last 4 days of self-administration as they were in proestrus during the days off from self-administration and they were excluded from this analysis (n=21). We examined the number of active hole responses and intake in a within-subjects design (n=23) such that the active hole responses and intake were analyzed when an individual female was in proestrus compared to when that same individual female was in non-proestrus stages. We analyzed the estrous cycle data as females in the proestrus stage compared to non-proestrus stages based on previous studies indicating that there is decreased opioid self-administration behaviors and intake during proestrus compared to all other stages^6,43,44.

Fecal collection.

Fecal samples were collected from a subset of animals (water: n=12, oxycodone: n=20) at two timepoints: prior to water or oxycodone self-administration (Pre) and during the final self-administration days (SAD19–21; Post). Fecal samples were collected either from the animal’s home cage on cage change day to ensure the fecal samples were collected on the same day as excretion, or from the operant chamber following that day’s self-administration. Fecal samples were collected into 1.5 mL tubes, immediately put on dry ice, and then stored at −80°C.

16S rRNA Sequencing, processing and analysis.

Samples were shipped on dry ice to Omega Bioservices (Norcross, GA) for processing and analysis. All samples underwent and passed quality control analysis. 16S rRNA sequencing was performed using V3, V4 primers to detect bacteria and were analyzed using NextGen Sequencing software. All data were compiled on Illumina BaseSpace for further analysis.

Data Analysis.

For analyses of the effect of a higher dose of oxycodone during early self-administration on oxycodone intake, a repeated measures two-way ANOVA was used to assess self-administration day as the within-subjects factor and oxycodone group (subset receiving 1.0 mg/kg from SAD7–10 vs animals receiving 0.3 mg/kg from SAD7–10) as the between-subjects factor.

Following the previous analyses where we found no difference between these groups, both groups of animals were combined, and self-administration data was assessed as follows. Self-administration was comprised of three portions, self-administration day 1–4 (SAD1–4), self-administration days 5–13 (SAD5–13) and self-administration days 14–21 (SAD14–21). To assess the effect of sex on self-administration behaviors for each portion, a repeated measures fourway ANOVA with nose poke hole (active, inactive) and self-administration day as within-subjects factors and sex (male vs female), drug exposure (water vs oxycodone) as between-subjects factors. A repeated measures three-way ANOVA was used to assess the effect of sex on number of reinforcers earned with self-administration day as the within-subjects factor and sex (male vs female) and group (water vs oxycodone) as between-subjects factors. Since we did not find a difference in these measures between males and females, males and females were combined for further analyses of the self-administration portions.

Active hole and inactive hole responding for each self-administration portion was assessed using a repeated measures three-way ANOVA as nose poke hole (active hole, inactive hole) and self-administration days as within-subjects factors and drug exposure group (water vs oxycodone or water vs oxycodone low vs oxycodone high) as between-subjects factor. The number of reinforcers earned for each portion was assessed using a two-way ANOVA with self-administration day as within-subjects factor and drug exposure group (water vs oxycodone, or water vs oxycodone low vs oxycodone high) as the between-subjects factor. The average responding for self-administration was reached during the last four days of self-administration (SAD18–21) and a two-way ANOVA was used to analyze these data with nose poke (active, inactive) as a within-subjects factor and drug exposure group (water vs oxycodone or water vs oxycodone low vs oxycodone high) as the between-subjects factor. Further analyses of these data when the data was binned into 30-minute intervals were done using a three-way ANOVA on nose poke responding (active, inactive and time bin as within-subjects factors and drug exposure group [water vs oxycodone or water vs oxycodone low vs oxycodone high] as the between-subjects factor). The number of reinforcers earned during these time bin analyses were analyzed using a two-way ANOVA with time bin as the within-subjects factor and drug exposure group as the between-subjects factor. For active pokes and reinforcers, we created a difference score between the last 4 days of self-administration and the 4 days of self-administration (average across days 18–21 – average across days 1–4) for both water and oxycodone treated animals. Difference scores were calculated to determine if either group (water or oxycodone) had change in responding for the active hole (drug-associated nose poke) or in reinforcers earned throughout self-administration. We ran unpaired t-test or one-way ANOVAs were used on these data. When a significant effect was revealed, Bonferroni post-hoc analyses were used.

A degradation index was calculated to normalize lever presses to correlate performance on the contingency degradation task with oral oxycodone intake using the following formula: (Contingent Lever Presses - Non-Contingent Lever Presses) / (Contingent Lever Presses + Non-Contingent Lever Presses)^19,28. A degradation index of 1 indicates that all presses were on the contingent lever, an index of 0 indicates an equal number of presses were on the contingent and non-contingent lever, and an index of 1 indicates that all presses were on the non-contingent lever. The ability to respond preferentially on the contingent lever (a positive degradation index) is interpreted as an animal having flexible behavior. Meanwhile, degradation indices equal to or less than 0 indicate inflexible behavior. The relationship between degradation indices and oxycodone intake, degradation indices and oxycodone intake in males and females, and degradation indices and oxycodone intake in oxycodone high and low animals were assessed using simple linear regressions.

To analyze the effects of water and oral oxycodone exposure on the number of species in the gut, a two-way ANOVA was used with time point of fecal sample (pre, post) as the within-subjects factor and drug exposure group (water vs oxycodone) as the between-subjects factor. Planned comparisons were done (Bonferroni) as we had hypothesized prior to analysis that there would be a difference in the number of species between time points in oral oxycodone exposed animals. Simple linear regressions were used to assess the relationship between degradation indices and the number of species (pre), the number of species prior to and post self-administration and active hole nose pokes during the last four days of self-administration (SAD18–21) in water and oral oxycodone animals, the number of species prior to and post self-administration and the amount of oral oxycodone consumed, and the number of species post self-administration and the amount of oral oxycodone consumed in oxycodone high and low animals.

Raw 16S rRNA gene sequencing reads were processed using a standard bioinformatics pipeline to generate high-quality taxonomic profiles. Briefly, sequences were quality filtered, denoised, and merged, and amplicon sequence variants (ASVs) were inferred. Taxonomy was assigned using a curated reference database, and ASVs were aggregated to the species level to generate a count-based feature table for downstream analysis. Differential abundance analysis was performed in R using DESeq2 (oxycodone relative water control in the post-treatment fecal samples). Species-level count data were imported into DESeq2 and modeled using a negative binomial distribution with treatment condition (oxycodone and water control post-treatment relative to their pre-treatment taxa) as the main factor. Library size normalization was performed internally by DESeq2 using size factor estimation. For each species, log2 fold change (log2FC) values were calculated to quantify differences in abundance between oxycodone-exposed and water-treated groups. Statistical significance was assessed using Wald tests and resulting p-values were adjusted for multiple comparisons using the Benjamini–Hochberg procedure to control the false discovery rate.

To identify taxa jointly associated with oxycodone exposure and behavioral flexibility, genus-level relative abundance data were analyzed across two dimensions. First, genus-level sensitivity to oxycodone was assessed using DESeq2 as described above, generating log₂ fold change (log₂FC) values and adjusted p-values for oxycodone versus water comparisons. Second, genus-level associations with behavioral flexibility were calculated using Spearman correlation coefficients (ρ) between baseline relative abundance and behavioral flexibility. Statistical significance for each dimension was determined independently (false discovery rate-corrected p-values for oxycodone effects; nominal p-values for behavioral correlations). Mean relative abundance was calculated for each genus to provide context regarding its prevalence and contribution to the overall microbial community.

RESULTS.

Rats self-administer oral oxycodone, and oxycodone-exposed rats respond more on a fixed-ratio 3 schedule than water control animals.

We found no differences in self-administration between on water and oral oxycodone self-administration during the early self-administration when rats were on an FR1 schedule (Fig. 1; see Supplement). We found that once animals were moved to a higher effort schedule (FR3) oxycodone-exposed animals showed an increase in nose pokes and reinforcers earned compared to water controls (SAD14–21, Fig. 2). Specifically, A three-way ANOVA conducted on active and inactive hole responses during these days (Fig. 2A) revealed a significant effect of drug exposure (oxycodone vs water, F_{(1, 75)} = 6.62, p = 0.012), nose poke hole (active hole vs inactive hole, F_{(1, 75)} = 189.02, p < 0.001) and self-administration day (F_{(7, 525)} = 6.88, p < 0.0001). We found no significant interaction between self-administration day and drug exposure (oxycodone vs water; F_{(7, 525)} = 1.06, p = 0.39), and self-administration day, nose poke hole, and drug (active hole vs inactive hole, oxycodone vs water; F_{(7, 525)} = 0.93, p = 0.48). There was a significant interaction between self-administration day and nose poke hole (active hole vs inactive hole; F_{(7, 525)} = 8.57, p < 0.001), and nose poke hole and drug exposure (active hole vs inactive hole, oxycodone vs water; F_{(1, 75)} = 4.73, p = 0.033). We observed that oxycodone-exposed animals poked more than water controls in both the active hole (p = 0.019) and inactive hole (p = 0.028). Importantly, we found a main effect of drug exposure on the number of reinforcers earned when animals are on an FR3 schedule of reinforcement (Fig. 2B; two-way ANOVA, oxycodone vs water; F(1, 75) = 5.20, p = 0.025). There was a significant main effect of self-administration day (F(4.45, 333.7) = 6.45, p < 0.0001). There was no significant interaction between self-administration day and drug exposure (oxycodone vs water; F(4.45, 333.7) = 1.04, p = 39).

Figure 2. Oral oxycodone animals have increased responding and earn more reinforcers on a fixed-ratio 3 schedule compared to water controls.

A. Nose pokes into the active and inactive holes of animals self-administering water (blue) or oral oxycodone (orange) during the 21 days of self-administration. Oral oxycodone animals respond differently than water control animals. Oral oxycodone animals have increased active and inactive hole nose pokes compared to water controls from self-administration day (SAD) 14–21 (three-way ANOVA; F_{(1, 75)} = 6.62, **p = 0.01). active hole vs inactive hole, F_{(1, 75)} = 189.02, p < 0.001B. Reinforcers earned during the 21 days of self-administration. Oral oxycodone animals earn more reinforcers from SAD14–21 compared to water controls (two-way ANOVA; F_{(1, 75)} = 5.20, *p = 0.025). C. Average number of active and inactive hole pokes and reinforcers earned during the last four days of self-administration (SAD18–21), when baseline self-administration is reached. Oral oxycodone animals poked more into the active hole (two-way ANOVA; F_{(1, 75)} = 6.21, *p = 0.015) and earned more reinforcers (unpaired t-test;; t₍₇₅₎ = 2.47, *p = 0.016) than water controls. D. Difference score of active hole nose pokes (left) and reinforcers earned (right) during the last four days of self-administration and first four days of self-administration. Oxycodone exposed animals have increased active hole (unpaired t-test, t₍₇₅₎ = 4.03, ***p = 0.0001) and reinforcers earned (unpaired t-test, t₍₇₅₎ = 4.74, ***p < 0.0001) difference scores compared to water controls. Data are expressed as mean ± SEM. Water: n = 31; oxycodone: n = 46.

Oxycodone-exposed animals reached stability in their responding during the last four days of self-administration (days 18–21; Fig. 2A–C). A two-way ANOVA revealed a significant effect of nose poke hole (active vs inactive hole; F_{(1, 75)} = 227.5, p < 0.0001) and drug exposure (oxycodone vs water; F_{(1, 75)} = 6.21, p = 0.015) and a significant interaction between nose poke hole and drug (active hole vs inactive hole, oxycodone vs water; F_{(1, 75)} = 5.07, p = 0.027) during the last self-administration days. Notably, oxycodone-exposed animals had increased active hole responses compared to water controls (p = 0.002) during the last four days of self-administration (days 18–21). Additionally, oxycodone animals earned more reinforcers than water control animals (unpaired t-test; t₍₇₅₎ = 2.47, p = 0.016; Fig. 2C). As baseline responding and intake was reached during the last four days of self-administration (days 18–21), we used these days for further analyses of the relationship between oral oxycodone behaviors and behavioral and biological biomarkers.

Next, we calculated a difference score (Fig. 2D) between the last days of self-administration when responding was stabilized compared to the first days of self-administration. Oxycodone-exposed animals showed higher active hole difference score (Fig. 2D, left; unpaired t-test; t₍₇₅₎ = 4.03, p = 0.0001) and reinforcer difference score (Fig. 2D, right; unpaired t-test; t₍₇₅₎ = 4.74, p < 0.0001) compared to water controls, indicating that oxycodone-exposed animals were responding more and receiving more reinforcers than water controls at the end of 21 days of self-administration compared to the initial phase of self-administration. We found no sex differences in responding in oxycodone-exposed or water controls (active hole, inactive hole; Fig. S1) or in the amount of oral oxycodone consumed (Fig. S1; see Supplement). Therefore, male and female animals were combined for each drug exposure group (oxycodone, water) for the remainder of analyses.

Two distinct phenotypes emerge during oral oxycodone self-administration.

We examined the 21 days of self-administration data considering these phenotypes (oxycodone high vs oxycodone low). When we divided the oxycodone-exposed animals based on the median amount of oxycodone consumed after stable responding occurred (SAD18–21), we found two distinct phenotypes emerge when the schedule of reinforcement changes from FR1 to FR3 (SAD14–21; Fig. 3, see Supplement). Specifically, we found that a group of oxycodone exposed animals have similar active hole responding and reinforcers as water controls (oxycodone low) and a group of animals that have increased active hole responding and reinforcers compared to both groups (oxycodone low, water) from self-administration days 14–21 (oxycodone high). A three-way ANOVA on self-administration days 14–21 (Fig. 3A) revealed a significant main effect of group (F(2, 74) = 23.69, p < 0.001), nose poke hole (F(1, 74) = 325.32, p < 0.001) and self-administration day (F(7, 518) = 7.70, p < 0.001). There was no significant interaction between group, nose poke hole, and self-administration day (F(14, 518) = 1.48, p = 0.11). There were significant interactions between group and nose poke hole (F(2, 74) = 19.47, p < 0.001), group and self-administration day (F(14, 518) = 2.18, p = 0.008), and self-administration day and nose poke hole (F(7, 518) = 9.71, p < 0.001). Importantly, oxycodone high animals poked more into the active hole compared to oxycodone low (p < 0.001) and water control (p < 0.001) animals and poked more into the inactive hole compared to oxycodone low (p = 0.004) and water control (p < 0.001) animals. We also found there was a significant main effect of group (oxycodone low, oxycodone high, water) on the number of reinforcers earned during self-administration days 14–21 (Fig. 3B; two-way ANOVA; F(2, 74) = 13.64, p < 0.001). There was also a significant main effect of self-administration day (F(7, 68) = 8.06, p < 0.001) and a significant interaction between group and self-administration day (F(14, 518) = 2.12, p = 0.1).

Figure 3. Distinct oral oxycodone phenotypes emerge during 21 days of self-administration.

A. Active and inactive hole responses for water (blue), oxycodone low (light orange) and oxycodone high (dark orange). Oxycodone high animals have increased responding during the first four days of self-administration (SAD1–4; three-way ANOVA; F(2, 74) = 3.90, *p = 0.03) compared to oxycodone low animals and increased responding during SAD14–21 compared to oxycodone low and water controls (three-way ANOVA; F(2, 74) = 23.69, ***p < 0.001). B. Reinforcers earned during the 21 days of self-administration. Oxycodone animals earn more reinforcers from SAD14–21 compared to oxycodone low and water controls (two-way ANOVA; F(2, 74) = 13.64, ***p < 0.001). C. Average number of active and inactive hole responses and reinforcers earned during the last four days of self-administration. Oxycodone high animals respond more (two-way ANOVA; (F(2, 74) = 21.93, p < 0.001) into the active (***p < 0.001) hole compared to oxycodone low and water animals, and poke more into inactive hole compared to water controls (*p = 0.03). Oxycodone high animals earn more reinforcers compared to oxycodone low and water control animals (one-way ANOVA; F(2, 74) = 16.89, ****p < 0.0001). D. Difference score of active hole responses (left) and reinforcers (right) during the last four days of self-administration and the first four days of self-administration. Oxycodone high animals have an increased active hole difference score compared to oxycodone low (p = 0.0001) and water controls (p < 0.0001) groups (one-way ANOVA; F(2, 74) = 19.59, ****p < 0.0001). There are differences between groups in reinforcer difference scores (one-way ANOVA; F(2, 74) = 14.74, ****p < 0.0001) where oxycodone low (p = 0.0095) and oxycodone high (p < 0. 0001) have increased reinforcer difference scores compared to water controls. Data are expressed as mean ± SEM. Water: n = 31; oxycodone low: n = 25; oxycodone high: n = 21.

We found that oxycodone high animals have increased responding compared to oxycodone low and water control animals during the last four days of self-administration (Fig. 3C). A two-way ANOVA revealed a significant main effect of group (F(2, 74) = 21.93, p < 0.001), and nose poke hole (F(1, 74) = 386.09, p < 0.001). There was no significant effect of self-administration day (F(3, 222) = 0.32, p = 0.81) and no significant interactions between group and self-administration day (F(6, 222) = 0.66, p = 0.69), between nose poke hole and self-administration day (F(3, 222) = 0.29, p =0.83), or between group, nose poke hole and self-administration (F(6, 222) = 0.81, p = 0.56). There were significant interactions between group and nose poke hole (F(2, 74) = 21.15, p < 0.001). Importantly, oxycodone high animals had increased active hole responses compared to oxycodone low (p < 0.001) and water (p < 0.001) animals and increased inactive hole responses compared to water animals (p = 0.03) during the last four days of self-administration. We also found a significant effect of group on number of reinforcers earned (one-way ANOVA; F(2, 74) = 16.89, p < 0.0001) where oxycodone high animals earned more reinforcers than oxycodone low (p < 0.0001) and water (p < 0.0001) animals. Active hole, inactive hole, and reinforcers were binned into 30-minute intervals for further analyses of responding and reinforcers earned across the 2-hour self-administration sessions (Fig. S2; See Supplement).

Additionally, we calculated a difference score to determine if there were group differences in responding and reinforcers earned from the first four days of self-administration compared to the last four days of self-administration (Fig. 3D). We found a difference between the groups (one-way ANOVA; F(2, 74) = 19.59, p < 0.0001), where oxycodone high animals had an increased active hole difference score compared to oxycodone low (p = 0.0001) and water (p < 0.0001) animals. We also found a difference in the reinforcers difference score between groups (one-way ANOVA; F(2, 74) = 14.79, p < 0.0001), where both oxycodone low (p = 0.0095) and oxycodone high (p < 0.0001) had increased reinforcer difference scores compared to water controls and there was almost a significant difference in reinforcer difference scores between oxycodone low and oxycodone high animals (p = 0.0508).

Performance on the contingency degradation task prior to drug exposure correlates with future oral oxycodone measures.

Animals’ performance on the contingency degradation task was calculated based on their raw lever presses to determine a degradation index²⁸ to correlate performance with the amount of oral oxycodone consumed. First, we investigated if contingency degradation performance correlated with future oral oxycodone intake during baseline self-administration (days 18–21). We found that an animal’s score on the task negatively correlated with the amount of oral oxycodone consumed during the last four days of self-administration (days 18–21), where animals that had higher degradation indices (closer to 1.0, indicative of a greater behavioral flexibility) consumed less oral oxycodone during the last four days of self-administration (Fig. 4A; simple linear regression, F(1, 33) = 4.66, R² = 0.12, p < 0.038). Next, we examined if sex influenced this correlation as previous studies have shown that performance on a decision-making task is more closely associated with addictive-like behaviors in females than in males⁴⁵. A simple linear regression showed a negative correlation between degradation index and amount of oral oxycodone consumed in females (F(1, 16) = 6.15, R² = 0.28, p = 0.02) but no correlation males (Fig. 4B; F(1, 15) = 0.11, R² = 0.007, p = 0.74). Lastly, we examined if performance on the contingency degradation task correlated with future oral oxycodone phenotype (oxycodone low vs oxycodone high). We found that performance on the contingency degradation task negatively correlated with amount consumed during the last four days of self-administration in oxycodone high animals (simple linear regression; F(1, 18) = 6.59, R² = 0.27, p = 0.02) but did not correlate in oxycodone low animals (Fig. 4C F(1, 13) = 0.31, R² = 0.02, p = 0.59).

Figure 4. Behavioral flexibility correlates with future oral oxycodone consumed.

A. Degradation index (performance on the contingency degradation task) negatively correlates with the amount of oral oxycodone consumed during the last four days of self-administration (SAD18–21; simple linear regression, R² = 0.12, *p = 0.038). B. Degradation index negatively correlates with the amount of oral oxycodone consumed from SAD18–21 in females (simple linear regression; R² = 0.28, *p 0.025) but not in males (R² = 0.007, p = 0.74). C. Degradation index negatively correlates with the amount of oral oxycodone consumed from SAD18–21 in oxycodone high (simple linear regression; R² = 0.36, **p = 0.007) but not oxycodone low (R² = 0.023, p = 0.59) animals. Oxycodone: n = 39; males: n = 17, females n = 18; oxycodone high: n = 20, oxycodone low: n = 15.

Oral oxycodone self-administration disrupts the gut microbiome compared to water self-administration.

First, we examined the number of species after the contingency degradation task and prior to self-administration (pre), we found that the number of species negatively correlated with performance on the contingency degradation task (Fig. 5A; simple linear regression; F(1, 28) = 9.7, R² = 0.26, p = 0.004). Next, we examined the effect of oral oxycodone on the number of species in the gut microbiome prior to (pre) and at the end of (post) self-administration. We found that oral oxycodone exposure decreases the number of species compared to pre levels and water animals’ post levels of species (Fig. 5B). A two-way ANOVA revealed a significant main effect of drug exposure (F(1, 30) = 13.03, p = 0.001). There was no effect of time point (pre vs post, F(1, 30) = 1.43, p = 0.24). There was a trend towards a significant interaction between drug exposure and time point (F(1, 30) = 3.20, p = 0.08). Planned comparisons of drug exposure and time point showed a significant difference between the number of species prior to oral oxycodone exposure and post oral oxycodone exposure (p = 0.042), where there was a decrease in the number of species after oral oxycodone exposure (Fig. 5B). Additionally, there was a significant difference between the number of species post self-administration between oxycodone-exposed and water control animals (p = 0.0008). Interestingly, when we investigated if the number of species in the gut microbiome prior to (pre) and at the end of self-administration (post) correlated with the amount of oral oxycodone consumed in the last four days of self-administration (Fig. 5C) we found the number of species (pre) did not correlate with the amount consumed (simple linear regression; F(1,17) = 0.37, R² = 0.02, p 0.55) but the number of species at the end of self-administration (post) positively correlated with the amount consumed during the last four days of self-administration (F(1, 17) = 18.21, R² = 0.52, p = 0.0005). Lastly, oxycodone exposure resulted in broad alterations in the gut microbiome at the species level. Differential abundance analysis revealed a greater number of species with significantly decreased abundance in the oxycodone group relative to water controls than species that were increased (post treatment), suggesting a general reduction in specific microbial taxa following oxycodone exposure (Fig. 5D).

Figure 5. Oral oxycodone exposure decreases the number of species in the gut and correlates with the number of species in the gut following oral oxycodone exposure.

A. Degradation index negatively correlated with the number of species in the gut following contingency degradation (prior to self-administration; simple linear regression; R² = 0.26, p = 0.04). B. There was a decrease in the number of species in the gut following oral oxycodone exposure (Post) compared to levels before drug exposure (Pre; planned comparisons, *p = 0.02) and compared to the number of species following water self-administration (Post; planned comparisons, ***p = 0.0004) C. The number of species in the gut prior to self-administration (Pre) did not correlate with the amount of oral oxycodone consumed from SAD18–21 (simple linear regression; R² = 0.01, p = 0.67) but the number of species in the gut at the end of self-administration (Post) positively correlated with the amount of oral oxycodone consumed from SAD18–21 (R² =0.52, ***p = 0.0005). D. Volcano plot showing differential abundance of gut microbiome species following oxycodone exposure relative to water control. Red circles indicate species that are significantly different between oxycodone- and water-exposed groups (false discovery rate < threshold). The majority of significant species exhibit negative log₂(fold change) values, indicating decreased abundance following oxycodone exposure.

To determine whether specific microbial features are associated with behavioral flexibility and oxycodone exposure, we performed genus-level analyses examining relationships between relative abundance, behavioral correlation, and oxycodone sensitivity. Across genera, relative abundance was associated with both positive and negative correlations with behavioral flexibility (Fig. 6A), with 14 genera showing significant positive correlations and 52 genera showing significant negative correlations (Spearman correlation, p < 0.05). Differential abundance analysis revealed that oxycodone exposure resulted in selective changes at the genus level, with a greater number of genera exhibiting decreased abundance (n=60) relative to increased abundance (n=10, Fig. 6B), consistent with a net reduction in specific microbial taxa following oxycodone exposure. Notably, genera associated with behavior and sensitive to oxycodone were distributed across a range of abundances (Fig. 6A–B).

Figure 6. Genus-level relationships between relative abundance, oxycodone exposure, and behavioral flexibility.

A. Genus-level relative abundance (log 10 mean relative abundance) is associated with behavioral flexibility (Spearman ρ). Genera exhibiting significant correlations with behavioral flexibility are indicated by blue circles (14 positively correlated, 52 negatively correlated; p < 0.05). B. Differential abundance analysis at the genus level following oxycodone exposure relative to water control. Red circles indicate genera that are significantly different between oxycodone- and water-exposed groups (false discovery rate < threshold). C. Integrated analysis showing genus-level relationships between oxycodone-induced changes and behavioral flexibility. Each point represents a genus plotted by oxycodone effect (-log false discovery rate with directionality; left: decreased with oxycodone, right: increased) and behavioral correlation (Spearman ρ). Genera that are significantly associated with both oxycodone exposure and behavioral flexibility are depicted as black circles, revealing distinct subsets of taxa that are reduced following oxycodone exposure and are either positively or negatively associated with behavioral flexibility.

To integrate these findings, genera were mapped according to both their oxycodone-induced change and their association with behavioral flexibility (Fig. 6C). This analysis identified a subset of genera exhibiting coordinated effects across both dimensions (n=9). Within this subset (Table 1), the majority of genera were reduced following oxycodone exposure and negatively associated with behavioral flexibility (n=7), whereas a smaller group of oxycodonereduced genera showed positive associations with flexibility (n=2).

Table 1.

Genus taxa that are oxycodone sensitive and correlated with behavioral flexibility.

Genus	log2FC_Oxy	Spearman_rho	Relative_Abundance	Interpretation
Oxalophagus	−1.27E+00	5.00E-01	2.36E-06	↓ oxy / ↑ flexibility
Jonquetella	−1.10E+00	4.77E-01	5.35E-04	↓ oxy / ↑ flexibility
Geothrix	−1.02E+00	−6.27E-01	1.12E-06	↓ oxy / ↓ flexibility
Rhodopila	−9.72E-01	−5.37E-01	9.36E-06	↓ oxy / ↓ flexibility
Phenylobacterium	−8.79E-01	−6.21E-01	5.35E-06	↓ oxy / ↓ flexibility
Pontibacillus	−7.18E-01	−8.02E-01	1.72E-06	↓ oxy / ↓ flexibility
Tannerella	−7.00E-01	−5.41E-01	3.05E-06	↓ oxy / ↓ flexibility
Fontibacter	−6.32E-01	−5.81E-01	3.73E-07	↓ oxy / ↓ flexibility
Actinomycetospora	−6.26E-01	−5.77E-01	1.80E-06	↓ oxy / ↓ flexibility

DISCUSSION.

Many individuals are prescribed oxycodone and majority of these individuals will terminate use upon completion of the prescription⁴⁶. Only a subset of individuals prescribed oxycodone will abuse the drug^47,48, indicating individual differences in the motivation to obtain and continue to use oxycodone after prescriptions have ended. Extended-access intravenous studies have demonstrated the emergence of distinct phenotypes during oxycodone self-administration^34–36, however we are the first, to our knowledge, to show these phenotypes emerge during a short-access oral oxycodone self-administration paradigm. Animals that consumed more oxycodone during self-administration (oxycodone high) had increased nose poke responding and reinforcers earned compared to animals that consumed less oxycodone (oxycodone low) and water control animals (Fig. 3A, B). Identifying these phenotypes in an oral oxycodone model is critical as we may be modeling the human scenario where oxycodone is ingested orally and a subset of individuals have increased motivation to obtain oxycodone. Furthermore, opioid users have cited prescription oxycodone as the drug they first became dependent on before transitioning to other drugs of abuse such as heroin and/or fentanyl^1–5. Therefore, this model is capturing a subset of individuals that may have an increased susceptibility to oxycodone dependence and future use of other opioids such as heroin and/or fentanyl. Future studies investigating if the phenotypes that emerge during oral oxycodone self-administration are associated with future illicit opioid self-administration (heroin or fentanyl) are crucial to further delineate if these preclinical phenotypes map on to reported findings shown in humans. Taken together, these findings highlight the importance of investigating the behavioral differences between these phenotypes as well as the neurobiological underpinnings that underlie these phenotypes to develop comprehensive and effective interventions and therapeutics at a more individualized level.

To better manage the possibility of dependence and abuse of oxycodone in human patients, identifying behavioral biomarkers of potential dependence prior to prescriptions is a critical avenue to investigate. Decision-making has been previously found to be predictive of oral oxycodone intake, particularly in female rats⁴⁵. Here, we have extended those findings to include the behavioral flexibility task, contingency degradation, as another behavioral biomarker that is associated with the amount of oral oxycodone consumed. Animals with intrinsic inflexible behavior (lower degradation indices) consumed more oral oxycodone during the last four days of self-administration, while animals with inherent flexible behavior consumed less oral oxycodone during the last four days of self-administration. These findings highlight that individual differences in inherent flexible behavior are a behavioral biomarker for future oral oxycodone intake. Notably, contingency degradation tasks have been utilized in human populations⁴⁹. Therefore, this behavioral task is a translationally relevant option that could be utilized prior to prescription use to inform dosing options in a human population. Interestingly, performance on the contingency degradation task and amount of oral oxycodone consumed is strongly correlated in females but not in males, similar to previous findings⁴⁵. One caveat is that we have previously shown an effect of the estrous cycle on contingency degradation²⁸, which may be influencing the correlation in females. Since we have demonstrated the emergence of distinct phenotypes during oral oxycodone self-administration (Fig. 3A, B), we also examined if there was a relationship between an animal’s phenotype during self-administration and intrinsic behavioral flexibility. Interestingly, we found that an animal’s intrinsic flexible behavior negatively correlated with the amount of oral oxycodone consumed during the last four days of self-administration. These findings highlight the relationship between inherent flexible behavior and the subset of individuals who will abuse oxycodone following a prescription’s termination. Importantly, the neurobiological underpinnings between performance on the contingency degradation task and oral oxycodone have not yet been investigated. The underlying circuitry involved in inherent flexible behavior (i.e., the strength of these circuits at baseline in each phenotype) and the influence of oral oxycodone on these circuits (i.e., oral oxycodone strengthening systems involved in inflexible behavior) are critical mechanisms to understand to improve treatment options for patients. As such, examining the relationship between these two mechanisms may provide an enhanced therapeutic avenue for oxycodone prescriptions.

Previous studies utilizing operant oral oxycodone self-administration paradigms have shown increased self-administration and addictive-like behaviors in females compared to males^33,45. Interestingly, we found no difference between males and females in nose poke responding or the amount of oral oxycodone consumed, which could be due to different length of programs, doses utilized and/or controls (saccharin fade vs water)⁴⁵. While our findings differ from previous operant oral oxycodone self-administration findings^33,45, our findings fall in line with short access intravenous oxycodone studies that have observed no effect of sex on oxycodone self-administration^6,50–52. We also observed no sex differences in water control animals in nose poke responding or number of reinforcers earned, similar to previous findings⁴⁵. Hormonal fluctuations across the rat reproductive cycle (estrous cycle) have been shown to drive self-administration behaviors in opioid self-administration paradigms^6,43,44,53. Specifically, female rats in the proestrus stage of the estrous cycle compared to non-proestrus stages have decreased self-administration behaviors^6,43,44. While we observed no differences in nose poke responding and reinforcers earned when females were in the proestrus stage of their cycle compared to non-proestrus stages during the last four days of self-administration, we did observe a significant effect of estrous cycle on the amount of oral oxycodone consumed. These findings indicate that the estrous cycle influences oral oxycodone consumption and therefore is an avenue to investigate to understand the effects of ovarian hormones on oxycodone consumption in a translationally relevant model.

Future studies investigating variables in testing conditions that are contributing to these effects would be useful to understand under what conditions the emergence of sex differences occur and/or how hormonal fluctuations influence oxycodone intake.

Critically, substance use disorders (SUDs) impact the whole body⁵⁴, with the gut being a primary region impacted in oral drug use due to ingestion of the drug directly into the gastrointestinal tract. We found that the number of species in our “pre” samples correlated with performance on the behavioral flexibility. Specifically, we found that a higher number of species predicted worse performance on the contingency degradation task (i.e., less flexible). These findings are intriguing, as the interplay between the gut-brain axis and baseline behavioral flexibility has not been extensively studied. Prior studies showed the antibiotic-induced dysbiosis of the microbiome impaired cognitive flexibility in mice⁵⁵ or diet (i.e., high-fat diet, Western diet) influences gut microbiome and how diet exposure then influences gut microbiome and cognitive flexibility in mice^56,57. Thus, our study focuses on individual differences within flexible behavior and gut microbiome environment. Further investigation into the species and/or genera that underlie performance on behavioral flexibility may prove to be useful in better understanding key factors in flexible behavior.

Similar to previous studies investigating the effects of opioids on the gut microbiome, we found that oral oxycodone self-administration impacted the gut microbiome^58–60, where there was a decrease in the number of species in oxycodone-exposed animals. Interestingly, we found that higher oxycodone taking (i.e., amount consumed) was correlated with a higher number of species in the gut microbiome. It is possible that this relationship may be due to animals that consumed less were more sensitive to the effects of oral oxycodone such that oxycodone dysregulated their gut microbiome more. Lastly, we found taxa that are associated with behavioral flexibility and altered by oxycodone where there are more decreases than increases distributed across the relative abundance (i.e., not just rare or common taxa but a mix of both common and rare taxa). Finally, there is a subset of nine taxa that are linked to both behavioral flexibility and oxycodone (Table 1).

Together, these findings demonstrate that oral oxycodone self-administration reveals individual differences in drug intake that are tightly linked to pre-existing behavioral and biological variability. Specifically, intrinsic differences in behavioral flexibility, i.e., contingency degradation performance, emerge as a predictor of future oxycodone consumption, with a particularly strong relationship observed in females. These results support the idea that variability in decision-making processes reflects underlying vulnerability to opioid misuse rather than being a consequence of drug exposure alone. These behavioral differences are paralleled by alterations in the gut microbiome, highlighting a bidirectional relationship between gut composition, behavioral flexibility, and drug-taking behavior. While overall taxa richness alone does not predict oxycodone intake prior to exposure, associations with both behavioral flexibility and post-exposure drug consumption suggest that microbiome dynamics may reflect or contribute to individual susceptibility. Our work supports a framework in which individual differences in opioid intake arise from the interaction of pre-existing behavioral traits and biological states, including gut microbiome composition. By identifying behavioral flexibility as a translationally relevant biomarker and linking it to microbiome variation, these findings provide a foundation for developing individualized strategies to predict risk and guide therapeutic interventions for opioid use disorders. Future studies targeting the neural and microbial mechanisms underlying these relationships will be essential for determining causality and for leveraging these insights into clinically actionable approaches.

Funding Statement

This work was funded in part by the National Institute on Drug Abuse (R00DA042934 to EAW) and the Brain and Behavior Research Foundation (EAW) and Whitehall Foundation (EAW) as well as Rowan University School of Osteopathic Medicine internal funds.

DATA AVAILABILITY STATEMENT

The datasets generated and analyzed during the current study are available in the GEO repository, GEO accession GSE326997.