Sensitivity of a fentanyl-vs.-social interaction choice procedure to environmental and pharmacological manipulations

Abstract

Recent studies have shown that social interaction can serve as an alternative reinforcer to opioid self-administration under a choice context in rats. However, additional parametric studies are needed to evaluate the sensitivity of opioid-vs.-social interaction procedures relative to more established opioid-vs.-food procedures. The current study evaluated the sensitivity of a novel fentanyl-vs.-social interaction choice procedure to environmental and pharmacological manipulations previously shown to affect fentanyl-vs.-food choice.

Male and female rats (responder rats; n=6/sex) were trained to respond in a discrete-trial choice procedure for either 30-s access to a same-sex “partner” rat or an intravenous fentanyl infusion. Once trained, the effects of fentanyl unit dose (0, 0.32–10 μg/kg/inf), partner rat presence, opioid-dependence status, chronic naltrexone administration (0.032, 0.1 mg/kg/h), and response requirement for fentanyl self-administration (fixed ratio 1–320) were determined across weeks.

The fentanyl-vs.-social interaction choice procedure was sensitive to the unit dose of fentanyl, chronic naltrexone treatment, and fentanyl response requirement. However, the magnitude of these effects on fentanyl choice was smaller than those reported in published fentanyl-vs.-food choice studies. Furthermore, fentanyl-vs.-social interaction choice was not sensitive to removal of the partner rat or opioid-dependence status. Minimal sex differences were detected.

These results suggest that this fentanyl-vs.-social interaction choice procedure is less sensitive to environmental and pharmacological interventions than previously established opioid-vs.-food choice procedures. The observed discrepancy in sensitivity between the procedures suggests that social interaction may have qualitatively different reinforcing properties compared to more commonly assessed alternative reinforcers such as food (preclinical) or money (human laboratory).

1. Introduction

Insights gained through preclinical research are needed to better treat Opioid Use Disorder (OUD) (1). However, OUD is a complex psychiatric condition that cannot be fully captured with any single preclinical model. Thus, it may be beneficial for preclinical models to focus on a subset of OUD features and compare results with clinical observations to gauge predictive validity. A core feature of OUD diagnosis is the excessive allocation of behavior towards procuring and using opioids over engaging with non-drug reinforcers that were deemed enjoyable or adaptive prior to opioid use (2, 3). Behavioral allocation in the context of opioid use can be simplified and modeled with procedures that provide the subject with a choice between opioid self-administration and a non-opioid reinforcer. Therefore, it is hypothesized that opioid-vs.-non-opioid choice procedures can model aspects of OUD and may serve as a translationally relevant platform to evaluate candidate medications and other determinants of opioid use (4–6).

Under human-laboratory settings, opioid self-administration procedures often provide participants with a choice between an opioid (e.g., heroin, morphine, hydromorphone) and a monetary alternative. Under these conditions, opioid choice is dose dependent, with a fixed amount of money chosen over small opioid doses and large opioid doses chosen over the same monetary alternative (e.g., (7)). Administration of clinically effective OUD medications such as naltrexone and buprenorphine not only decrease opioid choice, but also promote choice of the monetary alternative (see (6) for review). The concordance between OUD medication effects in human-laboratory opioid-vs.-money choice studies and clinical observations provides evidence of the predictive validity of opioid-vs.-money choice procedures. Furthermore, these results with naltrexone and buprenorphine establish a benchmark to compare candidate OUD medication effects.

Opioid-vs.-non-opioid choice procedures used in preclinical animal studies most commonly use food as the alternative reinforcer. Similar to human-laboratory findings, animal subjects (i.e., non-human primates, rodents) tend to exhibit dose-dependent choice of opioid infusions over food (5), but see (8). Furthermore, treatment with naltrexone and buprenorphine promote a reallocation of behavior away from opioid infusions and towards food reinforcement (6), mirroring the human-laboratory findings and the clinical effectiveness of these OUD medications. This agreement between clinical and preclinical observations provides evidence of reverse translation and suggests that novel findings gained through preclinical opioid-vs.-food choice procedures could be translated forward towards human laboratory studies and clinical trials.

Despite encouraging evidence of the translational potential of preclinical opioid-vs.-food choice procedures, the use of food as the alternative, non-opioid reinforcer may introduce interpretive challenges. For example, drugs that suppress appetite have been proposed as candidate OUD medications (9, 10). This could be a confound because appetite suppressants would be expected to directly decrease food choice and indirectly increase opioid choice, potentially giving a “false negative” result. The use of qualitatively different (i.e., non-food) alternative reinforcers may mitigate these concerns. Recent efforts with rats have shown that social interaction can serve as an alternative reinforcer to opioid self-administration under a choice context (11). In these operant procedures, rats exhibit dose-dependent opioid choice, choosing social interaction over small opioid doses, but choosing the opioid over social interaction when high doses are available (12–14). Additionally, opioid-vs.-social interaction choice is sensitive to environmental manipulations such as punishment (15). These findings suggest that opioid-vs.-social interaction and opioid-vs.-food choice procedures may have similar sensitivity to environmental manipulations and that opioid-vs.-social procedures could be used to evaluate candidate OUD medications. However, additional parametric studies are needed to evaluate the sensitivity of opioid-vs.-social interaction procedures relative to more established opioid-vs.-food procedures. Therefore, the goal of the current study was to evaluate the sensitivity of a novel fentanyl-vs.-social interaction choice procedure to environmental and pharmacological manipulations previously shown to affect fentanyl-vs.-food choice, including fentanyl unit dose, alternative reinforcer magnitude, response requirement, opioid-dependence status, and chronic naltrexone administration.

2. Methods

2.1. Subjects

Twenty-four male and female Sprague-Dawley rats (12 male, 12 female) were bred in-house and pair housed. At approximately 12 weeks of age, rats transitioned to single housing and mean weights (±SEM) were 412 g (±6.5) for males and 267 g (±6.3) for females at the time of separation. Rats were divided into two groups (6 male, 6 female each). Group 1 consisted of the “responder” rats that were trained under the fentanyl-vs.-social interaction choice procedure. Group 2, the “partner” rats, served as the non-drug reinforcer for the responder rats. Each same-sex pair of responder and partner rats had a behavioral history of co-habitation. All 12 responder rats completed Experiments 1 through 4, while only 10 responder rats completed Experiment 5 (4 male, 6 female). Specifically, one male rat died unexpectedly, and a second male rat was retired before Experiment 5 because it severely damaged plastic components within its operant chamber, rendering the chamber inoperable. Rats were housed in clear plexiglass cages and thus had the opportunity for visual access to other rats while in the housing room. Following social access self-administration training (see below), responder rats were surgically implanted with vascular access ports (Instech, Plymouth Meeting, PA) and custom-made jugular catheters as detailed previously (16). Rats were housed in a temperature- and humidity-controlled vivarium that was maintained on a 12-hr. light-dark cycle (lights off at 6:00 PM). Food (Teklad Rat Diet, Envigo) and water were provided ad-libitum in the home cage. Animal maintenance and research protocols were approved by the Virginia Commonwealth University Institutional Animal Care and Use Committee as well as in accordance with the “Guide for the Care and Use of Laboratory Animals” 8^th Ed. (2011).

2.2. Apparatus and catheter maintenance

Twelve modular operant chambers (MED-008CT-SOCIAL, Med Associates, St. Albans, VT) based on a published design (11) were located in sound-attenuating cubicles (SAC-302216–27, Med Associates). Chambers were equipped with two retractable levers (ENV-112CM, Med Associates) on opposite sides of the chamber. Above each lever was a white stimulus light which signaled the availability of each reinforcer, referred to herein as the fentanyl- and social-associated stimulus lights. One wall of the chamber held the fentanyl-associated lever and stimulus light as well as the LED house light (ENV-215M-LED). The opposite wall of the chamber held the social-associated lever and stimulus light along with the adjoining chamber for the reinforcer rat. The responder and partner rat chambers were separated by a sliding guillotine door that allowed the responder and partner rat to interact through a metal grate when the door was raised. The grate contained cut-out sections large enough to allow physical contact between rats while preventing co-mingling (11, 12). Intravenous (IV) fentanyl infusions were delivered by activation of a syringe pump (PHM-100, Med Associates) located inside the sound-attenuating cubicle as described previously (16). Behavioral sessions were operated by custom programs written for Med-PC IV (Med Associates). After each session, catheters were flushed with 0.1 mL of gentamicin (4 mg/mL) followed by 0.1 mL of heparinized saline (30 units/mL). Catheter patency was verified for the responder rats at least every two weeks and at the conclusion of the study by the instantaneous loss of muscle tone following IV methohexital (0.5 mg) administration.

2.3. Fentanyl-vs.-social interaction choice training

Responder rats were trained on the terminal fentanyl-vs-social interaction procedure using the following stepwise methods. First, daily 30-min behavioral sessions were used to train responder rats to lever press for 30-s of social interaction under an initial fixed-ratio (FR) 1 / 20-s time out schedule of reinforcement. Each session began with a 30-s raising of the guillotine door (revealing the partner rat) followed by a 60-s time out, the illumination of the house light, extension of the social-associated lever (right-side lever), and illumination of the social-associated stimulus light. Upon completion of a response requirement, the lever retracted, the stimulus light was extinguished, and the responder and partner rats could interact for 30-s while the guillotine door was raised. This schedule of reinforcement was in effect until at least 5 reinforcers were earned for two consecutive sessions. Next, the response requirement was increased to FR2 until the same criteria were met. Finally, the FR3 schedule was maintained for at least 5 sessions and until three consecutive sessions with five or more earned reinforcers. After completion of social self-administration training, responder rats were surgically implanted with IV catheters and provided a recovery period of 5 days. Next, responder rats were trained to self-administer IV infusions (infs) of fentanyl (3.2 μg/kg/inf) during daily 2-h behavioral sessions under an initial FR1 / 20-s time out schedule of reinforcement. These sessions began with a non-contingent fentanyl infusion followed by a 60-s time out followed by the illumination of the house light, extension of the fentanyl-associated lever (left-side lever), and illumination of the fentanyl-associated stimulus light. Upon completion of a response requirement, the lever retracted, the stimulus light was extinguished, and an IV fentanyl infusion was administered. This schedule of reinforcement was in effect for two sessions. Next, the FR requirement was increased to FR3 and in effect for four sessions.

Once social- and fentanyl-maintained responding were both established in isolation, responder rats were trained on the discrete trial fentanyl-vs.-social interaction choice procedure in a manner similar to a previously published fentanyl-vs.-food choice procedure in rats (16). This occurred during daily (M-F) 106-min behavioral sessions beginning at approximately 1PM. The terminal choice session consisted of a 12 minute “sample” component followed by nine, 10-min “choice” trials. In the sample component, the house light and social-associated stimulus light were both illuminated, and the guillotine door was opened for 30-s allowing the responder rat to interact with the partner rat. This was followed by a 5-min time out during which both the house and stimulus lights were extinguished. Next, the house light and fentanyl-associated stimulus light were illuminated and an infusion of the unit dose of fentanyl available in the subsequent choice trials (3.2 μg/kg/inf during training sessions) was administered to the responder rat. This was again followed by a 5-min time out with all lights extinguished, concluding the sample component. Next, nine identical choice trials began with the extension of both levers and the illumination of the house light, fentanyl-paired stimulus light, and the social paired stimulus light. The response requirement for both reinforcers was FR3. A “carryover contingency” was in place such that the first lever press during a choice trial inactivated the response requirement for the opposite lever. The stimulus light above the inactivated lever was extinguished although the lever remained extended. Completion of the response requirement on the remaining, active lever before the end of the 10-min choice trial resulted in the presentation of the corresponding reinforcer, retraction of both levers, and extinguishing of both house and stimulus lights for the remainder of that trial. If the response requirement was not met on the active lever, the corresponding reinforcer was not presented, stimulus and house lights were extinguished, and the trial was counted as an omission. Each of these nine trials was separated by a 30-s timeout during which levers were retracted and both stimulus and house lights were extinguished. This schedule of reinforcement was in place for three sessions before beginning the fentanyl dose-response determination, described below.

2.4. Experiment 1: Fentanyl-vs.-social interaction dose-response determination

Using the fentanyl-vs.-social interaction choice procedure described above, different doses of fentanyl (0, 0.32, 1, 3.2, and 10 μg/kg/inf) were each evaluated for five consecutive days in a counterbalanced dosing order. The unit dose of fentanyl was varied by infusion duration and determined according to the individual rat’s weight as recorded weekly. Data collected during the final (5^th) day of each dose evaluation are presented.

2.5. Experiment 2: Effect of partner rat presence on fentanyl-vs.-social interaction choice

We next evaluated the effect of removing the partner rat on choice of two of the smaller fentanyl unit doses (i.e., 0.32 and 1 μg/kg/inf fentanyl; unit doses that engendered less than 90% fentanyl choice). First, choice between 0.32 μg/kg/inf fentanyl and 30-s access to an empty partner chamber was determined for five consecutive days in all rats. Choice between 0.32 μg/kg/inf fentanyl and 30-s access to the partner rat was determined for five consecutive days the following week. This experiment was then repeated for a larger fentanyl unit dose (1 μg/kg/inf) over the following two weeks. Data collected during the final (5^th) day of each condition are presented. Of note, the operant chambers were cleaned once weekly throughout the studies and this schedule was not modified for Experiment 2.

Experiment 3: Effect of opioid withdrawal on fentanyl-vs.-social interaction choice

The effect of opioid withdrawal on the choice between a small unit dose of fentanyl (0.32 μg/kg/inf) and social interaction was determined. We selected this unit dose based on our previous reports that abstinence from extended access to fentanyl robustly increased choice of 0.32 μg/kg/inf fentanyl over a food alternative in male rats (17, 18) and the results of the dose-response determination (Experiment 1) showing that this fentanyl unit dose engendered less than 50% fentanyl choice, providing an opportunity to detect increased fentanyl choice. Prior to opioid dependence, choice between 0.32 μg/kg/inf fentanyl and social interaction was evaluated from 1–2:46PM (±0.5h) for 5 consecutive days. Just prior to each daily choice test (approximately 12:55PM), responder rats were each weighed and observed for 30s for the presence of nine somatic withdrawal signs (see Supplemental Materials). Fentanyl-vs.-social interaction choice, somatic withdrawal sign scores, and bodyweights collected on the 5^th day (Friday) served as a “pre-opioid dependence baseline” for subsequent analyses. Beginning the following Sunday evening (5:00 PM) rats were provided extended access (FR3, 10-s timeout) to fentanyl (3.2 μg/kg/inf) during 12h sessions (5PM-5AM, Figure 3F). Approximately 8h after each overnight fentanyl self-administration session, rats were observed for somatic withdrawal signs, weighed, and evaluated for fentanyl-vs.-social interaction choice. Overnight self-administration tests occurred from Sunday-Thursday nights for the first week and Sunday-Wednesday nights for the following week. Daily fentanyl-vs.-social interaction choice tests with a 0.32 μg/kg/inf unit dose of fentanyl occurred Monday-Friday the first week and Monday-Thursday the following week. After the completion of this overnight regimen of extended access to fentanyl, overnight extended access to fentanyl was discontinued and choice behavior between 0.32 μg/kg/inf fentanyl and social interaction was evaluated again from 1–2:46PM (±0.5h) for 5 consecutive days (Monday-Friday).

Figure 3:

Effect of overnight (5PM-5AM) fentanyl self-administration (FR3 TO10; 3.2 μg/kg/infusion) on fentanyl-vs.-social interaction choice, somatic withdrawal signs, and bodyweight in male (n=6) and female (n=6) rats. Abscissa: Experimental Day. (A) Percent choice of 0.32 μg/kg/infusion fentanyl over social interaction. (B) Number of infusions earned during overnight sessions. (C) Number of somatic withdrawal signs present. (D) Number of choices completed per session. (E) Change in bodyweight expressed as a percentage of baseline (collected the Friday prior to initiation of the experiment). (F) Daily schedule of self-administration sessions and withdrawal sign scoring. *Denotes significant difference relative to the first session of overnight fentanyl self-administration (panel B) or baseline (all other panels), defined as p< 0.05. All points represent the mean ± SEM. See Table S1 in Supplemental Materials for statistics relevant to each panel.

2.6. Experiment 4: Effect of 7-day continuous treatment with naltrexone on fentanyl-vs.-social interaction choice

Following recovery from extended access fentanyl self-administration, the effect of 7-day continuous treatment with naltrexone (0 [no osmotic pump], 0.032 and 0.1 mg/kg/h) on the choice between a relatively large unit dose of fentanyl (3.2 μg/kg/inf) and social interaction was evaluated. In a counterbalanced order, half of the rats were aseptically implanted with a 7-day osmotic pump (0.032 mg/kg/h naltrexone) on a Friday, and the other rats were untreated. Starting the following Monday, choice between fentanyl (3.2 μg/kg/inf) and social interaction occurred for five consecutive days. On the fifth day (Friday), osmotic pumps were aseptically removed from the naltrexone-treated rats and the remaining rats were implanted with naltrexone-filled pumps. Testing continued in the same manner the following week. After completion of the counterbalance, all rats were aseptically implanted with a 7-day naltrexone osmotic pump at a larger dose (0.1 mg/kg/h) on Friday and evaluated over the following week in the same manner. Data collected during the final (5^th) day of each dose condition are presented.

Experiment 5: Effect of fentanyl response requirement on fentanyl-vs.-social interaction choice

The effect of varying the response requirement for fentanyl (3.2 μg/kg/inf) on the choice of fentanyl over social interaction was assessed as a final experiment. The social interaction FR was held constant at FR1 while the fentanyl FR was increased in the following order: 1, 3, 10, 32, 100, 320. Each successive FR increase was maintained for 5 consecutive days (Monday – Friday). Data collected during the final (5^th) day of each condition are presented.

Data analysis

The primary dependent measures for each daily session of the fentanyl-vs.-social interaction choice procedure were: 1) percent fentanyl choice, defined as “(number of choices completed on the fentanyl-associated lever/total number of choices completed on both levers) × 100”, and 2) number of choices completed per session (fentanyl choices, social interaction choices, total choices). For the extended access fentanyl self-administration sessions, the primary dependent measure was the number of infusions earned during each session. Other dependent measures included the number of somatic withdrawal signs present and changes in bodyweight relative to the pre-opioid dependence baseline. Data were compared using one- or two-way ANOVA and the Geisser-Greenhouse correction was applied as appropriate. A significant ANOVA was followed by a Dunnet’s or Bonferroni post hoc test as appropriate. Statistical significance was defined as p<0.05. See Supplemental Table S1 for all statistical tests and factors, Supplemental Table S2 for sex comparisons, and Supplemental Table S3 for Day 1 vs. Day 5 comparisons.

2.7. Drugs

Fentanyl HCl and (−)naltrexone HCl were provided by the National Institute on Drug Abuse Drug Supply Program (Bethesda, MD) and dissolved in sterile saline. All solutions were passed through a 0.22 μm sterile filter (Millex GV, Millipore Sigma, Burlington, MA) before administration. All drug doses were expressed as the salt forms listed above and delivered based on weights collected weekly.

3. Results

3.1. Experiment 1: Fentanyl-vs.-social interaction dose-response determination

Figure 1A shows that choice of fentanyl over social interaction depended on fentanyl unit dose. Here, fentanyl choice was below 50% when no fentanyl or the smallest unit dose of fentanyl (0.32 μg/kg/inf) were available during the session. As the available unit dose of fentanyl increased, choice of fentanyl progressively increased, with the largest dose of fentanyl (10 μg/kg/inf) engendering almost exclusive fentanyl choice (fentanyl unit dose: F_2.4,26.9=34.0, p<0.001). Furthermore, both 3.2 and 10 μg/kg/inf fentanyl were chosen significantly more than the 0 μg/kg/inf condition, corresponding with an increase in fentanyl choices and a decrease in social interaction choices (Figure 1B; reinforcer type × fentanyl unit dose interaction: F_33,132=36.8, p<0.0001). The number of omitted trials remained low, indicating that the increase in fentanyl choice was associated with a decrease in social interaction choice, rather than a decrease in overall response rates. No effect of sex was detected on either dependent measure under baseline conditions (Table S2). A time × fentanyl unit dose interaction was detected when choice data were plotted as “Day 1 vs. Day 5” for each condition. No other effect of “Day 1 vs. Day 5” was detected (Table S3).

Figure 1:

Baseline fentanyl-vs.-social interaction choice in male (n=6) and female (n=6) rats. (A) Percent fentanyl choice. Abscissa: Intravenous unit fentanyl dose in μg/kg/infusion. (B) Number of choices completed per session. Abscissa: Intravenous unit fentanyl dose in μg/kg/infusion. All points represent the mean ± SEM. *Denotes significant difference relative to the “0 fentanyl-vs.-social interaction” condition, defined as p<0.05. See Table S1 in Supplemental Materials for statistics relevant to each panel.

3.2. Experiment 2: Effect of partner rat presence on fentanyl-vs.-social interaction choice

Figures 2A and 2B show that fentanyl choice was not sensitive to the removal of the partner rat when the two smaller unit doses of fentanyl were available (0.32, 1 μg/kg/inf). Although not significant, removal of the partner rat resulted in a trend towards increased choice of the smaller unit dose of fentanyl (0.32 μg/kg/inf) and a decrease choice of the social-interaction associated lever (Figure 2A). Additionally, a “partner rat presence × reinforcer type” interaction was detected when the 0.32 μg/kg/inf unit dose of fentanyl was available (Figure 2C; F_2,33=5.5, p=0.009) but not when 1 μg/kg/inf fentanyl was available (Figure 2D). Fentanyl choice was greater in male rats relative to female rats at the 1 μg/kg/inf unit dose of fentanyl irrespective of partner rat presence (Table S2). No other sex differences were detected on either dependent measure (Table S2). No effect of “Day 1 vs. Day 5” was detected (Table S3, Figure S2).

Figure 2:

Effect of the presence (filled bars) or absence (empty bars) of the partner rat on fentanyl-vs.-social interaction choice in male (n=6) and female (n=6) rats. (A) Percent choice of 0.32 μg/kg/infusion fentanyl over social interaction. Abscissa: Partner rat presence. (B) Percent choice of 1 μg/kg/infusion fentanyl over social interaction. Abscissa: Partner rat presence. (C) Number of choices completed per session (at 0.32 μg/kg/inf fentanyl unit dose). Abscissa: Reinforcer type. (D) Number of choices completed per session (at 1 μg/kg/inf fentanyl unit dose). Abscissa: Reinforcer type. All points represent the mean ± SEM. See Table S1 in Supplemental Materials for statistics relevant to each panel.

3.3. Experiment 3: Effect of opioid withdrawal on fentanyl-vs.-social choice

The effects of overnight (5PM-5AM) fentanyl self-administration on daytime (1–2:46PM) fentanyl-vs.-social choice, somatic withdrawal signs, and body weight are depicted in Figure 3. Unexpectedly, the number of fentanyl infusions earned decreased during the second session of overnight fentanyl self-administration as compared to the first night but then trended towards increasing across the remainder of the overnight sessions (Figure 3B; time: F_2.9,31.4=7.2, p=0.001). During this time, somatic withdrawal signs increased (Figure 3C; time: Friedman statistic=39.4, p<0.0001) and body weights decreased (Figure 3E; time: F_41.5,16.5=53.2, p<0.0001). Despite evidence that overnight fentanyl self-administration produced opioid dependence and withdrawal, choice of 0.32 μg/kg/inf fentanyl (Figure 3A; time: F_3.3,32.4=0.34, p=0.81) was not affected. The number of choices completed for social reinforcement and the total number of choices completed per session both decreased following the first night of overnight fentanyl self-administration (Figure 3D; interaction: F_18,297=5.4, p<0.0001), but was not different from baseline at any other timepoint. No effect of sex was detected on any of the dependent measures (Table S2).

3.4. Experiment 4: Effect of 7-day continuous naltrexone treatment on fentanyl-vs.-social interaction choice

Figure 4 shows the effects of continuous naltrexone treatment (0.032 and 0.1 mg/kg/h) on choice between fentanyl (3.2 μg/kg/inf) and social interaction choice after a one-week recovery period from extended fentanyl access. Notably, bodyweights consistently increased the week of recovery from overnight fentanyl self-administration and somatic withdrawal signs were no longer present. A dose-dependent decrease in fentanyl choice was detected during naltrexone treatment when compared to the no-osmotic-pump condition (Figure 4A; naltrexone dose: F_1.5,16.6=5.1, p=0.026). These changes in choice behavior were accompanied by a significant decrease in fentanyl choices and an increase in social-interaction choices (Figure 4B; “naltrexone dose × reinforcer type” interaction: F_4,66=7.2, p<0.0001). No effect of sex was found on either dependent measure (Table S2). No effect of “Day 1 vs. Day 5” was detected (Table S3, Figure S3).

Figure 4:

Effect of 7-day continuous naltrexone treatment on fentanyl-vs.-social interaction choice in male (n=6) and female (n=6) rats. (A) Percent choice of 3.2 μg/kg/infusion fentanyl over social interaction. Abscissa: Dose of naltrexone administered via osmotic pump in mg/kg/hour. (B) Number of choices completed per session. Abscissa: Reinforcer type. All points represent the mean ± SEM. *Denotes significant difference relative to the no-osmotic-pump-baseline, defined as p< 0.05. See Table S1 in Supplemental Materials for statistics relevant to each panel.

3.5. Experiment 5: Effect of fentanyl response requirement on fentanyl-vs.-social interaction choice

Figure 5A shows that choice of fentanyl over social interaction depended on the response requirement (i.e., cost) of fentanyl. Here, choice of 3.2 μg/kg/inf fentanyl was above 90% when the response requirements between fentanyl and social interaction were balanced at FR1. As the response requirement for fentanyl increased, choice of fentanyl progressively decreased, with 10% of choices completed on the fentanyl lever when the fentanyl response requirement was FR320 (cost (FR): F_2.8,24.8=25.2, p<0.001). Figure 5B shows decreases in fentanyl choice corresponded with both decreases in the number of fentanyl choices completed and increases in the number of social interaction choices completed at the two highest response requirements tested (FR100 and FR320; reinforcer type × cost (FR) interaction: F_10,135=32.2, p<0.0001). No effect of sex was detected on either dependent measure under baseline conditions (Table S2). No effect of “Day 1 vs. Day 5” was detected (Table S3, Figure S4).

Figure 5:

Effect of fentanyl response requirement on fentanyl-vs.-social interaction choice in male (n=4) and female (n=6) rats. (A) Percent choice of 3.2 μg/kg/infusion fentanyl over social interaction. Abscissa: Fentanyl response requirement (FR). (B) Number of choices completed per session. Abscissa: Fentanyl response requirement (FR). All points represent the mean ± SEM. *Denotes significant difference relative to the “Fentanyl FR1 vs. Social Interaction FR1” condition, defined as p< 0.05. See Table S1 in Supplemental Materials for statistics relevant to each panel.

4. Discussion

The present study evaluated the sensitivity of a fentanyl-vs.-social interaction choice procedure to environmental and pharmacological manipulations previously shown to affect opioid-choice procedures that use other alternative reinforcers. Consistent with previous choice studies that used food (preclinical) or money (human laboratory) as the alternative reinforcer to opioid self-administration, the choice of fentanyl over social interaction depended on the unit dose of fentanyl, the presence of naltrexone, and the response requirement for fentanyl. However, unlike published fentanyl-vs.-food choice data, the current fentanyl-vs.-social interaction choice procedure was insensitive to partner rat presence (i.e., alternative reinforcer magnitude) or opioid-dependence status. Overall, these findings suggest that this fentanyl-vs.-social interaction choice procedure is less sensitive to interventions than more commonly used opioid-choice procedures. Nevertheless, these data illustrate that this fentanyl-vs.-social interaction choice procedure could be used to complement existing opioid-choice procedures to evaluate generality of independent-variable effects across alternative reinforcer types. Additionally, these results might also suggest that social interaction may have qualitatively different responsivity to certain manipulations relative to more commonly assessed alternative reinforcers such as food or money.

4.1. Choice of fentanyl over social interaction is dose dependent

The choice of fentanyl over social interaction depended on the dose of fentanyl, with the largest fentanyl dose (10 μg/kg/inf) maintaining ~100% choice and smallest tested fentanyl doses (0.32–1 μg/kg/inf) maintaining similar levels of choice to the “no fentanyl” condition. These results are consistent with decades of work with preclinical opioid-vs.-food and human laboratory opioid-vs.-money choice procedures (7, 13, 19–22). Furthermore, recent work with a remifentanil-vs.-social interaction choice procedure has similarly shown dose sensitivity (14). However, when the choice was between social interaction and the “no fentanyl” condition, the percentage of choices completed on the fentanyl-associated lever remained above 40%. In contrast, we have reported near-zero levels of choice of “no fentanyl” over food reinforcement in both rats and monkeys (17, 23). The dynamic range of fentanyl choice with the current fentanyl-vs.-social choice procedure (~40–100%) is smaller than the range reported in published fentanyl-vs.-food choice studies, (~0–100%; (16)), suggesting that the current fentanyl-vs.-social choice procedure has decreased sensitivity to detect changes in the relative reinforcing effects of fentanyl compared to fentanyl-vs.-food choice procedures. Furthermore, the incomplete reallocation of behavior away from the “no fentanyl” condition and towards the social-interaction alternative suggests the reinforcing effectiveness of social interaction may be lower than food. This finding aligns with a recent behavioral economic study showing that social interaction reinforcement is more sensitive to increases in response requirement than food reinforcement, even when FR1 (lowest cost) rates are normalized between reinforcer types (24). Alternatively, rats of the current study may have become “sated” on social reinforcement by fewer than the nine possible choices allotted during each choice session. Future studies could test this hypothesis by determining the effects of varying the total number of choices possible within the experimental session on the choice of fentanyl over social interaction.

4.2. Continuous naltrexone administration decreases fentanyl-vs.-social interaction choice

Our finding of dose sensitivity suggests that the current fentanyl-vs.-social interaction procedure could detect interventions that affect the potency of fentanyl. Consistent with this hypothesis, continuous administration of the mu-opioid receptor antagonist naltrexone dose-dependently decreased choice of a relatively large unit dose of fentanyl (3.2 μg/kg/inf) and increased choice of social interaction. These results extend previous studies showing naltrexone dose-dependently decreases opioid choice and increases choice of a food (preclinical) or monetary (human laboratory) alternative (see (6) for review). Given that naltrexone is FDA approved for the treatment of OUD (25), the current results demonstrate sensitivity of a fentanyl-vs.-social choice procedure to a positive-control compound. Furthermore, a strength of using choice procedures to evaluate candidate pharmacotherapies for Substance Use Disorders is that they can determine whether a drug selectively decreases reinforcing effects of a drug of abuse (desirable; e.g., naltrexone) or produces non-selective decreases in both drug and non-drug reinforcement (undesirable). In addition, the use of social interaction as the alternative reinforcer to opioid self-administration instead of food may be better suited to evaluate candidate OUD medications that may affect feeding behavior. However, we note that naltrexone effects on fentanyl-vs.-social choice were modest relative to those previously reported using a fentanyl-vs.-food choice procedure. Whereas the highest tested dose of naltrexone (0.1 mg/kg/h) decreased choice of fentanyl (3.2 μg/kg/inf) by 17% when social interaction was the alternative reinforcer, this same dose of naltrexone decreased choice of the same unit dose of fentanyl by 77% when food was the alternative reinforcer (17). Additional studies with larger doses of naltrexone and smaller unit doses of fentanyl are needed to determine if the use of social interaction as an alternative reinforcer decreases the potency and/or the efficacy of naltrexone to decrease fentanyl choice relative to food. Nevertheless, despite the apparent diminished sensitivity of the current fentanyl-vs.-social interaction procedure, these results suggest that future studies could use naltrexone as a benchmark to compare the effectiveness of candidate OUD medications to decrease choice of fentanyl over social interaction.

4.3. Choice of fentanyl over social interaction decreases with increasing response requirement

A non-pharmacological approach for decreasing drug self-administration includes increasing the response requirement (i.e., cost) for the drug. This strategy is consistent with the fundamental economic principle of an inverse relationship between price and consumption of any commodity (26). Accordingly, fentanyl self-administration decreased when the response requirement for social interaction was fixed at FR1 and the response requirement for fentanyl self-administration increased across weeks. In addition, decreases in fentanyl self-administration were accompanied by a reallocation of behavior towards the “cheaper” social-interaction alternative, consistent with both preclinical and human laboratory studies that evaluated the effect of price on choice of opioid self-administration over qualitatively different non-opioid reinforcers such as food or money (e.g., (7, 27)). However, it should be noted that this reallocation was not detected until relatively large response requirements were imposed (FR100). In contrast, a response requirement of FR30 was sufficient to promote a reallocation of behavior away from fentanyl and towards food in a previous study (27). Nevertheless, increasing the response requirement for fentanyl was the most effective strategy to promote reallocation of behavior away from fentanyl and towards social reinforcement, with fentanyl choice decreasing to 10% when the fentanyl FR was 320. Notably, the remaining fentanyl choice at FR320 was attributable to a single female rat that responded exclusively for fentanyl under these contingencies. Overall, the reallocation of behavior away from fentanyl and towards social interaction in response to increased response requirements illustrates that social reinforcement substituted for fentanyl self-administration under this context and provides evidence that social reinforcement indeed functioned as a reinforcer.

4.4. Lack of sensitivity to partner rat presence or opioid-dependence status

Fentanyl-vs.-social interaction choice was not sensitive to two manipulations previously shown to affect opioid-vs.-food choice. First, removal of the partner rat from the social interaction chamber did not significantly decrease responding on the social-interaction-associated lever. However, a non-significant trend for decreased responding on the social-interaction lever (i.e., increased fentanyl choice) was observed following removal of the partner rat. This lack of a significant effect of partner rat presence was unexpected, as previous work has shown that acquisition of social interaction self-administration accelerates with increasing degrees of social contact (28), suggesting that removal of the partner rat would be expected to robustly decrease responding on the social-interaction-associated lever and increase responding on the fentanyl-associated lever. Furthermore, analogous preclinical (opioid-vs.-food) and human laboratory (opioid-vs.-money) experiments have shown decreased alternative-reinforcer choice when the magnitude of the alternative reinforcer is decreased (7, 13, 16, 29–31). The current findings suggest that interaction with another rat was not an essential component of the reinforcing effects of “social interaction” once the rats were trained. Rather, these data suggest that the other stimuli (i.e., lever retraction, extinction of the house light, raising of the guillotine door, odor of the partner rat) may have been sufficient to maintain an equivalent level of operant behavior. This conclusion suggests that these partner-rat independent stimuli functioned as conditioned reinforcers that may take longer to extinguish than was allowed in the current study (i.e., 5 consecutive testing days), although additional studies are needed to test this hypothesis. In either case, these data illustrate that this fentanyl-vs.-social interaction choice procedure is less responsive to changes in non-opioid reinforcer magnitude than other published opioid-vs.-food and opioid-vs.-money choice procedures.

A second manipulation that failed to affect fentanyl-vs.-social interaction choice was opioid-dependence status. Rats became opioid dependent across two weeks of overnight fentanyl self-administration, as evidenced by robust weight loss and somatic signs of opioid withdrawal during afternoon observations. However, choice of fentanyl over social interaction was not affected by opioid withdrawal. In fact, during instances of reduced overall choices per session, the allocation of this diminished level of behavior between fentanyl and social interaction was not different from baseline. These findings are in contrast to previous reports of opioid withdrawal increasing choice of opioid infusions over food in male primates and rats (18–20, 32–34), and decreasing opioid choice over food in female rats (18). A limitation of the current study was that fentanyl-vs.-social choice was only evaluated at one relatively small unit dose of fentanyl (0.32 μg/kg/inf). This decision was made because choice of this unit dose most closely approximated 50% at baseline, providing an opportunity to detect both increases and decreases in fentanyl-vs.-social interaction choice. However, the possibility remains that changes in the fentanyl unit dose during the choice session or other parametric adjustments could unmask opioid-withdrawal effects on this endpoint. Alternatively, the lack of opioid-withdrawal effects on fentanyl-vs.-social interaction choice may be attributable to equivalent effects of opioid withdrawal on both fentanyl and social reinforcement, thereby negating any effects of opioid withdrawal on behavioral allocation. Nevertheless, the current results illustrate feasibility of evaluating fentanyl-vs.-social interaction choice in opioid-withdrawn rats.

4.5. Implications for the study of opioid-vs.-social interaction choice

The current results illustrate sensitivity of a novel fentanyl-vs.-social interaction choice procedure to pharmacological and environmental manipulations. These parametric results begin to define the upper and lower boundaries of this procedure and may facilitate interpretation of future experiments. However, when considered against published fentanyl-vs.-food choice results, the current procedure was less sensitive to some manipulations and insensitive to others. This diminished sensitivity may be a result of relatively weak reinforcing effects of social interaction under the current parameters, as evidenced by incomplete reallocation of choice towards social interaction in the absence of fentanyl availability. Thus, future studies may be able to increase the sensitivity of this fentanyl-vs.-social interaction procedure by altering parameters to enhance the relative reinforcing effects of social interaction (e.g., training history, social-isolation status, degree of social contact during reinforcement component). Although the current results illustrate some challenges in establishing a fentanyl-vs.-social interaction choice procedure, the choice of opioid self-administration over social interaction is an important determinant of OUD and modeling this balance with preclinical models may be of high translational relevance.

Highlights.

The fentanyl-vs.-social interaction choice procedure was sensitive to the unit dose of fentanyl, chronic naltrexone treatment, and fentanyl response requirement, albeit with a smaller magnitude than results in published fentanyl-vs.-food choice studies.

Fentanyl-vs.-social interaction choice was not sensitive to removal of the partner rat or opioid-dependence status.

These results suggest that this fentanyl-vs.-social interaction choice procedure is less sensitive to environmental and pharmacological interventions than previously established opioid-vs.-food choice procedures.