Male Sprague–Dawley rats (Charles River) arrived in our animal facility at either postnatal day (P) 21 (total n = 62) or 63 (total n = 49). Adolescent rats were at P 35 and adult rats were at P 77 at the beginning of the self-administration training. In Experiment 1, we group-housed rats 4 per cage before surgery and then housed them individually after surgery (single-housed condition). In Experiment 2, rats were group-housed four per cage before and after surgery (group-housed condition). In the case that we lost cage mates due to health-related issues, we ensured that there were no less than three rats per cage in the group-housed condition. All rats were maintained under a reverse 12:12-h light/dark cycle with food and water freely available. We performed the experiments under the protocols approved by the University of Maryland College Park Animal Care and Use committee. We excluded rats due to catheter patency (adult = 6, adolescent = 5), poor training (adult = 9, adolescent = 13), and health-related issues (adult = 1, adolescent = 4). The number of rats reported herein refers to rats included in the statistical analysis.

We anesthetized the rats with isoflurane gas (5% induction; 2–3% maintenance) and inserted silastic catheters into the rat’s jugular vein, as previously described (Caprioli et al., 2017; Li et al., 2018). We injected the rats with ketoprofen (2.5 mg/kg, s.c.) after surgery to relieve pain and inflammation; we allowed the rats to recover 5–7 days before oxycodone self-administration training. During the recovery and training phases, we flushed the catheters every 24–48 h with gentamicin (Butler Schein; 5 mg/ml) dissolved in sterile saline.

We trained the rats in self-administration chambers located inside sound-attenuating cabinets and controlled by a Med Associates (Georgia, VT, United States) system. Each chamber has two levers located 8–9 cm above the floor. During self-administration training, presses on the retractable (active) lever activated the infusion pump (which delivered an oxycodone infusion); presses on the stationary (inactive) lever were not reinforced with the drug. For oxycodone intravenous infusions, we connected each rat’s catheter to a liquid swivel (Instech) via polyethylene-50 tubing, protected by a metal spring. We then attached the liquid swivel to a 20-ml syringe via polyethylene-50 tubing and to a 22-gauge modified needle (Plastics One, VA). To promote lever pressing for adolescents, we set the active lever 3 cm lower during the first 4 days of training.

We used a training procedure previously described by Altshuler et al. (2021). We trained the rats to self-administer oxycodone for 6 h per day, under a fixed-ratio-1 (FR1) with 20-s timeout reinforcement schedule. Each training session included six 1-h sessions with 10 min off in between sessions. We dissolved oxycodone (kindly provided by National Institute on Drug Abuse Drug Supply Program) in saline (10 mg/ml), and the rats self-administered oxycodone at a dose of 0.1 mg/kg/infusion over 3.5 s (0.10 ml/infusion). We trained the rats for 10 sessions over an 11-day period (off day between the 5th and 6th day). We used Brevital (3 to 4 mg/kg) to check catheter patency for low responders during the training. We only included rats with an average number of infusions above 20 across 10 training days and across the last 3 training days for the following behavioral procedures.

The daily training sessions started at the onset of the dark cycle and began with the extension of the active lever and the illumination of the red house light. The house light remained on for the duration of the 6-h session. During training, active lever presses led to the delivery of an oxycodone infusion and a compound 5-s tone-light cue (the tone and light modules were located above the active lever). During the 20-s timeout, we recorded the non-reinforced lever presses. We set 90 infusions as the maximum for each 6-h session to prevent overdose. The red house light was turned off and the active lever retracted after the rats received the maximum infusions or at the end of the 6-h session.

During the abstinence phase, we housed the rats either individually or in groups in the animal facility and handled them two to three times per week.

We conducted all relapse tests immediately after the onset of the dark cycle. The sessions began with the extension of the active lever and the illumination of the red house light, which remained on for the duration of the session. Active lever presses during testing (the operational measure of drug seeking in incubation of craving studies) resulted in contingent presentations of the tone-light cue, previously paired with oxycodone infusions, but did not result in drug infusions.

All rats first received intravenous surgeries. Next, we trained them to self-administer oxycodone for 10 days. On abstinence day 1, we tested the rats for relapse in a 30-min session. On abstinence day 15, we tested the rats for relapse in a 60-min session. Note that we used a 30-min session on abstinence day 1 to minimize the effect of extinction learning that could decrease drug seeking on abstinence day 15 (Li et al., 2018; Fredriksson et al., 2020). We performed Experiment 1 (single-housed condition) in four independent runs (total adult = 21, total adolescent = 26). We performed Experiment 2 (group-housed condition) in three independent runs (total adult = 12, total adolescent = 14). The detailed experimental timeline is shown in Figure 1.

We analyzed the data with SPSS (version 24) mixed ANOVAs or t-test as appropriate. We followed significant interaction or main effects with Fisher PLSD tests. For the repeated measures analyses of the training data, we replaced 24 outlier values of inactive lever presses with the group mean for a given training day. We defined outliers as three median absolute deviations (MADs) above the group median (Leys et al., 2013), and we only replaced one outlier (the highest value above the threshold) for each training day. For rats disconnected from swivels on training day 8 or 9, we extended their training for an additional 1 to 2 days. We replaced their day 10 training data with the mean of the last two or three training days. For lever presses during relapse tests and incubation slopes, we also excluded one outlier above and/or below the threshold (defined by above or below 3 MADs from the group median). In addition, we used the data from the day 1 relapse test and the first 30 min of the day 15 relapse test to calculate incubation slope. The equation is “Incubation slope = (active lever presses on abstinence day 15 − active lever presses on abstinence day 1)/(15 − 1),” Finally, we indicate the between- and within-subject factors of the different analyses in the Results section. All statistical comparisons are listed in Supplementary Table 1.

The goal of Experiment 1 was to examine whether adolescent-onset oxycodone self-administration leads to changes in oxycodone seeking after abstinence compared with adult-onset oxycodone self-administration under the single-housing condition. To achieve this goal, we used a within-subject design and tested rats for oxycodone seeking (relapse tests) on abstinence day 1 and 15 (Figure 1).

Based on a recent study suggesting that housing condition contributes to the age differences in incubated cocaine craving (Li and Frantz, 2017), we performed Experiment 2 using the same design as Experiment 1, but under the group-housing condition.

In both experiments described above, we observed lower non-incubated oxycodone seeking in adolescents than adults, but no differences in incubated oxycodone seeking. Based on this observation, we are interested in comparing the development of incubation, indexed by incubation slope, following adult-onset and adolescent-onset oxycodone self-administration. We analyzed the incubation slope using the between-subject factors of Age and Housing. We found no interaction between these two factors, but a significant main effect of Age (F_1,65 = 4.084, p = 0.047, Figure 4). Post hoc analysis showed that under the single-housing condition, adolescent rats exhibited significantly elevated incubation slope compared with adults (t₄₄ = −2.097, p = 0.042), while no statistical significance was detected under group-housing condition (p > 0.05). In addition, there was a trend toward significant main effect of Housing (F_1,65 = 3.090, p = 0.083). These results suggest that single-housed adolescent rats may have an accelerated incubation compared with single-housed adult rats, which may be interfered under the group-housing condition.

Based on substantial evidence demonstrating that environmental enrichment such as social housing attenuates drug seeking across different drug classes (Thiel et al., 2009, 2010; Hofford et al., 2014; Imperio et al., 2018; Frankowska et al., 2019; Westenbroek et al., 2019; Cho et al., 2020; Grimm and Sauter, 2020; Mastrogiovanni et al., 2021), we were interested in examining whether this is also the case in our study. Note that different from most studies in which housing condition is manipulated only during abstinence periods, here the manipulation of housing condition started after intravenous surgeries. First, we observed no effect of housing conditions on oxycodone self-administration across training days (Figure 5A, p > 0.05). We analyzed the active lever data from the 30-min relapse test on abstinence day 1, or from the 60-min relapse test on abstinence day 15, using the between-subject factors of Housing condition (single-housing and group-housing) and Age (adult and adolescent). On abstinence day 1, we observed main effects of Age (F_1,65 = 11.439, p = 0.001), but no effects of Housing condition or interaction between these two factors (p > 0.05, Figure 5B). On abstinence day 15, we observed main effects of Housing condition (F_1,67 = 8.243, p = 0.005), but no effects of Age or interaction between these two factors (p > 0.05, Figure 5B). Post hoc analysis revealed that in adolescent rats, the group-housing condition significantly attenuated oxycodone seeking compared with the single-housing condition (t₃₇ = 2.381, p = 0.023), while there was a trend toward a decrease in oxycodone seeking in group-housed compared with single-housed adult rats (t₄₅ = 1.745, p = 0.090).

Taken together, these analyses demonstrated that group-housing condition attenuated oxycodone seeking on abstinence day 15, but not on day 1. This attenuating effect of group-housing on oxycodone seeking is more robust in adolescents than adults.

In both studies, we found no age differences in oxycodone self-administration in rats at a dose of 0.1 mg/kg under extended access condition (6 h/day). We chose this experimental condition based on previous studies from us and others detecting reliable oxycodone self-administration and incubation of oxycodone craving in adult rats (Fredriksson et al., 2020, 2021; Altshuler et al., 2021). In contrast, an early dose-response study showed that adolescent mice self-administered less oxycodone than adult mice across various doses (Zhang et al., 2009). However, direct comparison should be made with caution because of differences in species and experimental procedures between two studies. Regarding the latter, a morphine study previously showed that while adolescent rats self-administer less morphine under the short-access condition (1 h/day), there are no age differences in morphine self-administration under the extended-access condition (Doherty et al., 2009), which is consistent with our finding here. In addition, conflicting results have also been observed under the same heroin self-administration procedures in adolescent rats (Doherty and Frantz, 2012; Doherty et al., 2013), suggesting that the variabilities in reinforcing effects of opioids exist across cohorts in adolescent rats.

Under single-housing condition, we found no age differences in incubated oxycodone seeking on abstinence day 15, which is in line with the previous finding showing no age differences in incubation of cocaine craving following adult-onset and adolescent-onset cocaine self-administration in rats (Madsen et al., 2017). In contrast, previous work from the Frantz group demonstrated that adolescent-onset, compared with adult-onset, drug self-administration leads to attenuated incubation of cocaine and heroin craving in rats (Li and Frantz, 2009; Doherty and Frantz, 2012; Doherty et al., 2013). Three factors could potentially contribute to these discrepancies. The first is the self-administration procedure. While the Frantz group used the limited access drug self-administration procedure (2–3 h/day) (Li and Frantz, 2009; Doherty and Frantz, 2012; Doherty et al., 2013), we and Madsen et al. (2017) used the extended-access procedure (6 h/day), which produces more robust incubation than limited access self-administration (Lu et al., 2004). The second is the housing condition throughout the experiment. Recent work from the Frantz group showed that single-housing condition, used here and also by Madsen et al. (2017), could offset the age differences in incubated cocaine seeking previously observed under group-housing condition (Li and Frantz, 2017). To explore this possibility, we examined incubation of oxycodone craving under group-housing condition, similar to the experimental condition used by Li and Frantz (2009). Again, we found no age differences in incubated oxycodone seeking on abstinence day 15, which further validates our finding under single-housing condition.

The third potential factor is the presence of cues during relapse tests. Doherty and Frantz (2012) found that while adolescent-onset heroin self-administration leads to attenuation of incubated heroin seeking without the presence of cues, there are no age differences in extinction responding in the presence of cues, which is in line with our findings. Stimuli-specific age differences are also observed previously in cocaine studies. For example, adolescent-onset cocaine self-administration leads to decreased cue-induced reinstatement (Li and Frantz, 2009, 2017; Anker and Carroll, 2010), increased stress-induced reinstatement (Anker and Carroll, 2010; Wong and Marinelli, 2016), and no change in context-induced reinstatement of cocaine seeking (Cho et al., 2020) in rats. Taken together, these results suggest that instead of enhancing vulnerability for all types of future relapse, adolescent-onset drug exposure might bidirectionally alter the individuals’ sensitivity to certain relapse-eliciting stimuli in a drug-specific manner.

An unexpected finding here is that while no age differences were observed regarding total oxycodone intake, adolescent rats exhibited attenuated oxycodone seeking on abstinence day 1, compared with adult rats. This result suggests that adolescents might be resistant to oxycodone seeking during early abstinence after oxycodone self-administration, which contrasts with no age differences in oxycodone seeking observed on abstinence day 15. A key difference between these two time points is that on abstinence day 1, the adolescent group was still in adolescent period, but they reached young adulthood on abstinence day 15 (Spear, 2000; McCutcheon and Marinelli, 2009). Therefore, we speculate that this resistance to oxycodone seeking may be due to behavioral characteristics in the adolescent period, such as enhanced exploratory behavior, novelty-seeking, impulsivity, and cognitive and behavioral flexibility (Douglas et al., 2003; Doremus-Fitzwater et al., 2012; Simon et al., 2013; Westbrook et al., 2018). Consistent with our finding, several studies have shown that adolescent rats, compared with adults, are resistant to forming reward-seeking habits (Anderson and Spear, 2011; Doremus-Fitzwater and Spear, 2011; Naneix et al., 2012; Serlin and Torregrossa, 2015; Rode et al., 2019). This suppressed drug-seeking behavior during early abstinence in adolescent rats might also be specific to oxycodone, because previous cocaine and heroin studies all found no differences in drug seeking on abstinence day 1 (Li and Frantz, 2009; Doherty and Frantz, 2012; Doherty et al., 2013; Madsen et al., 2017).

Finally, taking data from abstinence days 1 and 15 together, we identified an elevated incubation slope in adolescent-onset rats, compared with adult-onset rats, suggesting that adolescent-onset rats accelerated the development of incubation during abstinence. Therefore, one possibility is that adolescent-onset rats might exhibit enhanced oxycodone seeking than adult-onset rats in relapse tests beyond abstinence day 15. Although we cannot exclude this possibility, we believe this is unlikely because incubated oxycodone seeking in adult rats plateaus on abstinence day 15, without future increase on abstinence day 30 (Fredriksson et al., 2020). Alternatively, we speculate that this acceleration only occurs between abstinence days 1 and 15, during which adolescent rats transition into young adulthood and start exhibiting adult-like behaviors. It is of note that post hoc analysis identified the significant differences of the incubation slope between ages only under the single-housing condition, but not under the group-housing condition. Together with the trending significant main effect of Housing (p = 0.083), it is possible that the suppressing effect of the housing condition on oxycodone seeking on abstinence day 15 (see below) interferes with the development of incubation.

First, we observed no main effect of housing conditions on oxycodone self-administration, which is in line with a recent study showing no effect of environmental enrichment on heroin self-administration in adult rats (Imperio et al., 2018). In contrast, an early study showed that group-housing decreases acquisition of heroin self-administration in adult rats (Bozarth et al., 1989). This discrepancy may be due to different drug doses and number of rats in the group-housing condition across these studies. It is noted that substantial literatures have demonstrated that rats reared in enriched environment show decreased self-administration to cocaine (Green et al., 2010), methamphetamine (Westenbroek et al., 2019), amphetamine (Bardo et al., 2001), methylphenidate (Alvers et al., 2012), and remifentanil (Hofford et al., 2017). However, in these studies, self-administration behavior was assessed after several weeks of environmental enrichment when rats reached young adulthood, instead of being assessed concurrently during development. Therefore, our data suggest that reinforcing effect of oxycodone may overwrite the protective effect of social housing against drug intake during development.

On abstinence day 15, we found that group-housing attenuated oxycodone seeking. Several studies previously demonstrated that environmental enrichment, using a combination of group-housing and enrichment apparatuses in home cages, decreases cue-induced seeking to cocaine (Thiel et al., 2009, 2010, 2012), methamphetamine (Hofford et al., 2014), and sucrose (Grimm and Sauter, 2020), and heroin-priming induced heroin seeking (Imperio et al., 2018), compared with single-housing conditions in rats. Recent studies also showed group-housing alone in rats decreases cue-induced cocaine (Frankowska et al., 2019), methamphetamine (Westenbroek et al., 2019), nicotine, and sucrose (Mastrogiovanni et al., 2021) seeking, and context-induced cocaine seeking (Cho et al., 2020). Our result here is in line with these previous findings. It is of note that post hoc analysis identified the significant differences of oxycodone seeking on abstinence day 15 between housing conditions only in adolescent rats, with a trend toward statistical significance in adult rats (p = 0.090). The modest effect in adult rats may be due to the timing of housing manipulation, as discussed in detail below.

On abstinence day 1, we found no main effect of housing condition. The lack of statistical significance in adolescent rats could be due to a floor effect attributed by low lever responding on abstinence day 1. In adult rats, our finding is inconsistent with previous studies showing that environmental enrichment with a combination of social housing and enrichment apparatuses in home cages decreases cue-induced cocaine seeking (Chauvet et al., 2012) and sucrose seeking (Grimm and Sauter, 2020) on abstinence day 1, respectively. Beside the possible contribution from the enrichment apparatuses, another main factor that contributes to this discrepancy could be the timing of housing manipulations. In our study, housing conditions were maintained throughout the study, to examine whether social housing differentially affects incubation of oxycodone craving in adolescents and adults. However, Chauvet et al. (2012) and Grimm and Sauter (2020) employed an intervention approach in which environmental enrichment was implemented during abstinence after self-administration under single-housing conditions. Consistently, an early study showed that the protective effect of environmental enrichment against sucrose seeking after abstinence in adult rats is robust with the latter approach, but absent when housing conditions were maintained throughout the study (Grimm et al., 2008). Taken together, our data suggest that group-housing throughout oxycodone self-administration might protect against oxycodone seeking in adolescents during early abstinence, but not adults. This is possibly because compared with adults, social interaction and play is more rewarding and essential to development in adolescent rats (Panksepp, 1981; Douglas et al., 2004; Vanderschuren et al., 2016; Burke et al., 2017).