Adult (PND 60+) male and female rats (n=266) from 15 HRDP strains were used for these experiments. Two strains (F344/NCrl and WKY/NCrl) were obtained from Charles River Laboratories, and the remaining 13 strains were obtained from the Medical College of Wisconsin (provided by Dr. Melinda Dwinell, R24OD024617). Strains and sample sizes are described in Table 1 , and the attrition rate is depicted in Supplementary Figure 1 . Upon arrival, rats were allowed to habituate for at least one week. Before intravenous surgery and self-administration procedures, all animals were tested for somatosensory and analgesia-related measures, as described in the Supplementary Methods and previously (31). Rats were single-housed in a temperature (22°C) and humidity (40%) controlled animal vivarium on a 12-h light/12-h dark cycle. Rats were provided with standard rat chow (Teklad 2918, Envigo, Indianapolis, IN) and water ad libitum. Animals were tested in 22 cohorts (n=15-24/cohort). The ACI/EurMcwi strain was used as a control strain in all batches obtained from the Medical College of Wisconsin to allow for the detection and minimization of batch effects. This strain was chosen due to preliminary studies identifying that the ACI/EurMcwi strain displays robust behavioral responses in all outcome measures. All procedures were conducted in accordance with institutional guidelines outlined by the Institutional Animal Care and Use Committee at the University of Colorado Boulder, an Association for Assessment and Accreditation of Laboratory Animal Care International accredited institution.

Chronic indwelling intra-jugular catheters were implanted using a modified procedure described previously (34). Animals were anesthetized (2-4% isoflurane), and intravenous catheters were implanted in the right jugular vein under aseptic conditions. Once the vein was isolated, it was punctured with a 22-gauge needle and the tubing was inserted and secured with suture thread. Sterile catheters (Access Technologies, Skokie, IL) consisted of 14 cm of polyurethane tubing (0.51 mm inner diameter, 1.12 mm outer diameter) secured to a 22-gauge back mount pedestal (Protech International). Immediately prior to surgery, animals were administered the analgesic carprofen (5 mg/kg, s.c.) and antibiotic enrofloxacin (5 mg/kg, s.c.). Animals received daily subcutaneous injections of carprofen (5 mg/kg) for two days following surgery. Catheters were flushed daily with 0.2 mL heparinized (20 IU/ml) bacteriostatic saline containing gentamicin sulfate (0.33 mg/ml; Hospira). Catheter patency was evaluated using 0.2 mL propofol (10 mg/mL propofol, Sagent) solution before the acquisition phase of oxycodone self-administration, before the escalation phase, and after the final session of the escalation phase.

Self-administration procedures were conducted in operant chambers (Med Associates, St. Albans, VT) containing two retractable levers (a drug-paired left lever and an inactive right lever), stimulus cue lights above each lever, and a sound-attenuating fan. The acquisition and escalation of oxycodone intake was measured using a two-phase self-administration paradigm ( Figure 1A ). All self-administration sessions began at the start of the dark phase of the light/dark cycle. During the acquisition phase, rats were trained to self-administer oxycodone (Oxycodone HCl; B&B pharmaceutical, Englewood, CO) on a fixed-ratio 1 (FR1) reinforcement schedule in ten 2-hour “short-access” (ShA) sessions. Responses on the drug-paired lever resulted in an oxycodone infusion (0.15 mg/kg/infusion) delivered over 5-s and a cue light (7.5-W white light, 20-s). Each oxycodone infusion was followed by a 20-s timeout period (TO20). Responses on the inactive lever were recorded but had no consequences. After the acquisition phase, rats were tested under a progressive ratio (PR) schedule to evaluate motivation to obtain oxycodone. Due to a procedural error during the PR sessions, the data are unusable and not shown. Rats were then transitioned to the escalation phase that included ten 12-hour “long-access” (LgA) sessions under the same FR1:TO20 schedule of reinforcement (35).

The coefficient of determination (R²) from a 2-way ANOVA with strain and sex as the predictors was used to estimate the broad-sense heritability (H²) for the following behavioral phenotypes during the acquisition and escalation phases: total oxycodone intake, early and late oxycodone intake, change in oxycodone intake, early and late lever discrimination, early and late normalized timeout responses.

All data are presented as mean ± SEM. Longitudinal data for hourly intake or intake per session was analyzed using a repeated measures ANOVA with session as the within-subject factor and strain and sex as the between-subject factors. Oxycodone intake, lever discrimination, and normalized timeout responding were analyzed with repeated measures ANOVAs with timepoint (early or late) as the within-subject factor and strain and sex as the between-subject factors. Results from repeated measures ANOVAs are reported in Supplementary Table 1 . Two-way ANOVAs were run to test the effects of strain, sex, and their interaction for total oxycodone intake and the change in oxycodone intake. Results from two-way ANOVAs are presented in Supplementary Table 2 . To examine sex differences in total oxycodone intake during acquisition and escalation, a simple main effects analysis was performed in which Student’s t-tests were used to test the significance of sex within each strain. Pearson correlations were used to evaluate the relationship between oxycodone analgesia and self-administration phenotypes, using strain means calculated within each sex for each behavioral phenotype. Pairwise correlations between oxycodone behavioral phenotypes are presented in Supplementary Table 3 .

Outliers were identified and winsorized prior to statistical analysis, as described in the Supplementary Methods . Missing data (< 2% of self-administration session data) was imputed as described in the Supplementary Methods . Statistical significance was set as p < 0.05. For the sex differences analysis, a Bonferroni correction was applied based on the number of tests for 15 strains during acquisition (p < 0.003) and 14 strains during escalation (p < 0.004). All data analysis was conducted in R v4.3.2.

Male and female rats from 15 HRDP strains underwent a behavioral protocol designed to measure the acquisition and escalation of i.v. oxycodone self-administration ( Figure 1A ). The average hourly oxycodone intake was calculated to describe and compare the general pattern of oxycodone consumption in males and females during acquisition and escalation sessions collapsed across strains. We observed a general pattern of increasing oxycodone intake across the entire procedure ( Figure 1B , F_6.37,1368.52 = 45.55, p < 0.001). Both males and females increased their oxycodone intake throughout the procedure, with males generally showing higher intake across sessions (F_1,215 = 9.82, p = 0.002). Next, we analyzed how the interaction between genetic background and sex influenced various aspects of self-administration.

To assess strain and sex differences in the acquisition of oxycodone self-administration rats were trained to self-administer oxycodone in 10 daily short-access sessions. The general pattern of oxycodone intake per session during the acquisition phase ( Figure 2 ) was shaped by a strain-sex interaction (F_14,236 = 3.24, p < 0.001) and a strain-session interaction (F_{76.12,1283.21} = 3.11, p < 0.001). Males and females from a few strains (e.g., HXB2/IpcvMcwi) displayed divergent patterns of oxycodone consumption, with males having greater intake than females during acquisition. Strains also differed in the general pattern of self-administration over time. Some strains (e.g., HXB23/Ipcv and HXB31/IpcvMcwi) displayed relatively stable responding while others (e.g., HXB1/Ipcv) rapidly increased oxycodone intake across the acquisition phase. To simplify the strain and sex comparisons, we calculated the total oxycodone dose consumed during the acquisition phase. Figure 3 illustrates that total oxycodone intake was shaped by a strain-sex interaction (F_14,236 = 3.24, p < 0.001). Males from the HXB2/IpcvMcwi strain consumed more oxycodone than females (t_8.29 = -6.59, p < 0.001). Males from the ACI/EurMcwi (t_29.71 = -2.89, p = 0.007) and WKY/NCrl (t_15.73 = -2.56, p = 0.02) strains also displayed increased oxycodone intake compared to their female counterparts, but these sex differences did not survive multiple testing correction.

To further characterize how patterns of oxycodone consumption change across the acquisition period the mean intake during the first three acquisition sessions (“early”) and last three acquisition sessions (“late”) was calculated ( Supplementary Figure 2 ). Within these windows, oxycodone intake was shaped by a strain-sex interaction (F_14,236 = 3.66, p < 0.001) and a strain-timepoint interaction (F_14,236 = 6.99, p < 0.001). To further explore these strain differences, the change in intake was calculated as the difference in intake between early and late acquisition ( Figure 4 ). Genetic background influenced the change in oxycodone intake across acquisition, as indicated by a main effect of strain (F_14,236 = 6.97, p < 0.001). Although most strains increased oxycodone intake, some strains (e.g., BXH6/Cub, HXB31/IpcvMcwi) showed a minimal change in intake during the acquisition phase.

Following acquisition, rats underwent 10 daily long-access sessions to measure the escalation of oxycodone self-administration. The BN-Lx/Cub strain is included in Figure 5 to qualitatively describe their pattern of self-administration but was excluded from statistical analysis due to the low sample size in the males (n=1). Throughout escalation, oxycodone intake ( Figure 5 ) was influenced by a strain-sex (F_13,184 = 4.44, p < 0.001) and a strain-session (F_70.49,997.76 = 2.33, p < 0.001) interaction, suggesting that there is a complex interplay between how genetic background and sex influence the escalation of oxycodone intake. To simplify this interaction, the total oxycodone dose during the escalation phase was calculated ( Figure 6 ). Here, we again see that sex and strain contribute to overall oxycodone intake during the escalation sessions as indicated by a strain-sex interaction (F_13,184 = 4.44, p < 0.001). A striking difference was observed in HXB2/IpcvMcwi males that self-administered significantly more oxycodone than females (t_11.33 = -5.74, p = 0.001). Nominal sex differences were observed in the WKY/NCrl strain (t_7.32 = -2.73, p = 0.03) but did not survive multiple testing correction.

Similar to the analysis of acquisition of oxycodone intake, the mean intake during the first three escalation sessions (“early”) and last three escalation sessions (“late”) was calculated ( Supplementary Figure 3 ). Oxycodone intake was influenced by a strain-sex (F_13,184 = 3.83, p < 0.001) and a strain-timepoint (F_13,184 = 3.46, p < 0.001) interaction suggesting that sex differences were not consistent across strains and the magnitude of escalation was not consistent across strains. To further describe strain differences the change in oxycodone intake during the escalation phase was calculated ( Figure 7 ). A main effect of strain (F_13,184 = 3.48, p < 0.001) highlights the significant variation between the strains in their propensity to increase oxycodone intake during the escalation phase.

As another measure of behavioral responding during the oxycodone self-administration procedure, we calculated each rat’s ability to discriminate between the lever paired with oxycodone delivery and the inactive lever. Lever discrimination was evaluated by calculating the ratio of correct responses to total responses for each rat within individual sessions. Therefore, a value close to one indicates that the rat is almost exclusively pressing the active lever. To evaluate how lever discrimination changes across sessions, we calculated the mean lever discrimination during the first three sessions (“early”) and last three sessions (“late”) of the acquisition and escalation phases. During acquisition ( Figure 8A ), lever discrimination was shaped by strain-sex (F_14,236 = 2.31, p = 0.005) and strain-timepoint (F_14,236 = 3.44, p < 0.001) interactions. During escalation ( Figure 8B ), lever discrimination was influenced by a strain-timepoint interaction (F_13,184 = 2.08, p = 0.017). These findings suggest that learning to discriminate the outcomes between the two levers generally improves or stays the same throughout the procedure but is also dependent on the strain and sex of the animal. It is interesting that rats from the genetically related SHR/OlaIpcv and WKY/NCrl strains (36) displayed divergent patterns of lever discrimination across both phases of self-administration ( Figure 8C ). Lever discrimination in WKY/NCrl rats improved throughout both phases of self-administration while SHR/OlaIpcv rats show persistent deficits in lever discrimination hovering around 50% lever discrimination.

We also evaluated the persistence of responding during periods of drug unavailability by evaluating drug-paired lever responding during the 20-s timeout period following each oxycodone infusion. Because animals that self-administer more oxycodone have more opportunities for timeout responding, the number of timeout responses was normalized to the number of oxycodone infusions. Normalized timeout responding during acquisition ( Figure 9A ) was influenced by a strain-sex-timepoint interaction (F_14,236 = 1.78, p = 0.043). Although most animals displayed low levels of timeout responding throughout acquisition, it was striking that some strains (e.g., HXB1/Ipcv) were more likely to engage in persistent responding during the timeout period, especially during the late timepoint. Normalized timeout responding during escalation ( Figure 9B ) was influenced by a strain-sex (F_13,184 = 3.79, p < 0.001) and a strain-timepoint (F_13,184 = 3.11, p < 0.001) interaction. Here, males from a few strains (e.g., HXB2/IpcvMcwi) displayed a robust increase in timeout responding.

To evaluate the relationships between oxycodone-related behavioral phenotypes, Pearson correlations were calculated using strain means within each sex ( Figure 10 ). Phenotypes related to oxycodone-induced analgesia, as described previously in (31), were included to describe the relationship between oxycodone analgesia and self-administration phenotypes. The maximal possible efficacy (% MPE) was used as a measure of acute analgesia 15- and 30-minutes after oxycodone administration, where higher values represent a greater degree of oxycodone-induced analgesia ( Supplementary Methods ). All pairwise comparisons between phenotypes are included in Supplementary Table 3 . Select correlations between phenotypes are displayed in Figure 11 . The magnitude of oxycodone-induced analgesia at 30-minutes positively correlated with total oxycodone intake during escalation (r(26) = 0.39, p = 0.04) ( Figure 11A ). This suggests that drug-naïve animals that experience a greater degree of acute oxycodone-induced analgesia may go on to consume higher levels of oxycodone. The normalized timeout responding during early acquisition sessions negatively correlated with lever discrimination during late acquisition sessions (r(26) = -0.42, p = 0.02) ( Figure 11B ). This suggests that high levels of responding during a period of drug unavailability correlate with poor lever discrimination during the acquisition phase.

The broad-sense heritability (H²) of several behavioral phenotypes was calculated by a 2-way ANOVA using strain and sex as the independent variables ( Table 2 ). Measures of oxycodone intake were highly heritable, ranging between H² = 0.26-0.54. Other aspects of oxycodone self-administration, such as lever discrimination and normalized timeout responding, were more moderately heritable, ranging between H² = 0.25-0.42.