Kappa opioid agonists reduce oxycodone self-administration in male rhesus monkeys

Abstract

Rationale:

Combinations of mu and kappa opioid receptor (KOR) agonists have been proposed as potential analgesic formulations with reduced abuse liability. The current studies extend previous work by investigating the typical KOR agonist, salvinorin A, and the atypical KOR agonist, nalfurafine, as deterrents of oxycodone self-administration using a progressive-ratio (PR) schedule of reinforcement.

Methods:

In separate experiments, adult male rhesus monkeys (N=4/experiment) were trained under a PR schedule of reinforcement to self-administer cocaine (0.1 mg/kg/injection) and saline on alternating days. Oxycodone (0.01–0.1 mg/kg/injection) alone and combined with salvinorin A (Experiment 1; 0.006, 0.012 mg/kg/injection) or nalfurafine (Experiment 2; 0.0001–0.00032 mg/kg/injection) were tested within the alternating cocaine and saline baseline. The mechanism of nalfurafine’s effects on oxycodone self-administration was investigated via pretreatment with the KOR antagonist, nor-Binaltorphimine (nor-BNI; 10 mg/kg; i.m.).

Results:

All subjects self-administered oxycodone alone above saline levels at sufficiently large doses, and combining salvinorin A or nalfurafine with oxycodone reduced the mean number of injections per session to saline levels (Experiment 1) or to levels that were significantly lower than oxycodone alone (Experiment 2). The ability of nalfurafine to reduce oxycodone self-administration was reversed by pretreatment with nor-BNI.

Conclusions:

These results demonstrate that KOR agonists, including the clinically-used KOR agonist, nalfurafine, can punish self-administration of a prescription opioid analgesic, oxycodone, in rhesus monkeys, and that nalfurafine’s punishing effect is KOR dependent. Combinations of KOR agonists with prescription opioids may have reduced abuse liability.

Introduction

Rates of prescription-opioid deaths have risen sharply since the turn of the century (Chen et al. 2019; Rudd et al. 2016). Efforts to discourage prescription-opioid abuse include the development of abuse-deterrent formulations (ADFs), the designs of which include physical barriers to deter tampering for intravenous or intranasal use, antagonists to offset the drug’s reinforcing effects, and extended-release formulations that constrain peak drug levels through mechanisms that slow the drug’s release (Friedmann et al. 2018; Raffa et al. 2012). However, while these strategies may decrease tampering for non-oral use, they may not deter oral misuse of pain medications (Mastropietro and Omidian 2015).

The Food and Drug Administration (FDA) has released a comprehensive set of guidelines to define specific strategies for the development of ADFs for prescription opioids (FDA 2015). One strategy describes an “aversion” approach in which the medication can be formulated with an agent that causes an unpleasant effect if the medication is taken orally at higher doses than prescribed. In behavioral science terms, decreasing consumption of a reinforcer by coupling its presentation with the delivery of an aversive stimulus may be considered a form of punishment, defined as a decrease in the probability of a behavior as a function of the consequence of that behavior (Azrin and Holz 1966). A small but growing body of preclinical work has examined the effects of drugs as punishers of drug self-administration, mostly using intravenous histamine as a drug punisher (Freeman et al. 2014a; Holtz and Carroll 2015; Minervini et al. 2019; Negus 2005; Woolverton et al. 2012). Using self-administration approaches to study drugs as punishers is a rational approach for the preclinical development of ADFs for prescription medications because the deterring agent is coupled and administered contingently with the drug reinforcer, thus modeling the situation in which humans would encounter the effects of a deterring agent when overusing prescription drugs formulated with the agent.

Kappa opioid receptor (KOR) agonists are a class of drugs that have received attention as potential therapeutics for drug abuse (Chavkin 2011; Mello and Negus 2000; Prisinzano et al. 2005; Shippenberg et al. 2007; Wee et al. 2010). Unlike mu opioid receptor (MOR) agonists, KOR agonists produce aversive subjective effects including dysphoria and psychotomimesis (MacLean et al. 2013; Ranganathan et al. 2012). Numerous studies have reported that KOR agonists decrease extracellular dopamine in brain-reward centers and mitigate increases in brain dopamine produced by drugs of abuse (Al-Hasani et al. 2015; Chefer et al. 2013; Gross et al. 2019). In terms of behavior, pretreatment with KOR agonists reduces drug self-administration in monkeys and rats (Cosgrove and Carroll 2002, 2004; Glick et al. 1995; Mello and Negus 2000; Schenk et al. 2001) and decreases the rewarding effects of intracranial self-stimulation in rats (Todtenkopf et al. 2004; Tomasiewicz et al. 2008). KOR agonists also decrease cocaine-induced reinstatement of operant responding previously maintained by cocaine in monkeys and rats (Morani et al. 2009; Ruedi-Bettschen et al. 2010; Schenk et al. 2000) as well as behavioral sensitization and locomotor activity resulting from cocaine administration in rats (Collins et al. 2001; Heidbreder and Shippenberg 1994; Morani et al. 2012; Vanderschuren et al. 2000).

Fewer reports have investigated KOR agonists as punishers of drug self-administration, wherein KOR-agonist administration is contingent upon the organism’s behavior. The first study to demonstrate a punishing effect with a KOR agonist reported that U69,593, when offered as a mixture with the MOR agonist, fentanyl, reduced fentanyl self-administration in rhesus monkeys (Negus et al. 2008). Subsequently, we reported that the KOR agonist, salvinorin A, when self-administered as a mixture with the MOR agonist, remifentanil, reduced remifentanil choice in rhesus monkeys in a drug vs. drug choice procedure (Freeman et al. 2014b), an approach that is useful for distinguishing punishing effects of drugs from nonspecific, rate-decreasing effects (Banks and Negus 2017). Typical KOR agonists like U69,593 and salvinorin A produce untoward psychotropic effects such as dysphoria and psychotomimesis, and while they may be effective at punishing drug self-administration, the robustness of the side effects of typical KOR agonists precludes their feasibility for clinical use.

In recent years, KOR agonists have been developed that exhibit behavioral profiles that are atypical of the class. One example is nalfurafine, which is the only KOR agonist that has been approved for clinical use in humans (Remitch®; for the treatment of pruritis; Nagase and Fujii 2011). Nalfurafine’s atypical behavioral profile may be related to its reported potency “bias” for G-protein signaling relative to ß-arrestin₂ recruitment at the KOR (Schattauer et al. 2017), as has been suggested for other recently-developed KOR agonists (see Mores et al. 2019 for a review). However, nalfurafine produces KOR-typical thermal antinociception in rodents and monkeys (Hasebe et al. 2004; Ko and Husbands 2009; Townsend et al. 2017) and reduces the rewarding effects of morphine and cocaine in rodents in conditioned place-preference (CPP) tests (Hasebe et al. 2004), effects that are consistent with those produced by “typical” KOR agonists (see Kivell et al. 2014). We recently reported that nalfurafine, when combined in a mixture with oxycodone, reduced oxycodone self-administration to saline levels and increased oxycodone’s thermal antinociceptive potency in rats (Townsend et al. 2017), suggesting that combinations of oxycodone and nalfurafine should be further investigated as potential analgesics with reduced abuse liability. However, nalfurafine’s signaling profile at the KOR has been shown to vary greatly between rat and human KORs, with G-protein bias at the human KOR exceeding bias at the rat KOR by an order of magnitude (Schattauer et al. 2017). As such, it is possible that nalfurafine’s apparent abuse-deterring effects in rats may not translate to the same degree in human and nonhuman primates (NHPs).

Rhesus monkeys are closer in phylogeny to humans than rodents, and along with other NHPs, they have provided key translational information as animal models of substance abuse and dependence (Lynch et al. 2010; Platt et al. 2005). As stated above, we have used rhesus monkeys to study drugs as punishers in drug-choice designs (e.g., Freeman et al. 2014b; Woolverton et al. 2012). However, we have also demonstrated that punishing doses of drugs like histamine that reduce drug choice do not necessarily decrease the maximum reinforcing effectiveness of drug reinforcers in a progressive-ratio (PR) design (Freeman et al. 2014a). Reinforcing effectiveness as determined in PR designs is an important measure because it has been shown to rank order drugs in a manner that is predictive of their abuse liability in humans (Wee et al. 2005; Wilcox et al. 2000). Therefore, the objective of the current study was to investigate the punishing effects of two KOR agonists, one typical (salvinorin A) and one atypical (nalfurafine), in male rhesus monkeys self-administering oxycodone under a PR schedule of reinforcement.

Methods and Materials

All animal-use procedures were approved by the University of Mississippi Medical Center’s (UMMC) Animal Care and Use Committee and were conducted in accordance with the National Research Council’s Guide for Care and Use of Laboratory Animals (8^th edition 2011). Enrichment protocols were approved by UMMC’s veterinarian staff.

Subjects and Apparatus

Seven adult male rhesus monkeys (Macaca mulatta) weighing between 10–12 kg served as subjects. Three of the subjects (MR4351M, 4299, and 314001) had prior experimental histories with cocaine self-administration (Freeman et al. 2014a; Huskinson et al. 2016; and unpublished data, respectively). Four of the subjects (1356AS, 1010AL, 9512L, and 314206B) had no history of drug self-administration prior to the current study. All subjects were provided with sufficient food to maintain stable body weights (200–300 g/day, Teklad 25% Monkey Diet; Harlan/Teklad, Madison, WI, USA) and had unlimited access to water. Fresh fruit or vegetables and foraging material were provided daily, and a chewable vitamin supplement was given three times per week. Lighting was cycled to maintain 16 h of light and 8 h of dark, with lights on at 0600 hours. Each subject was fit with a mesh jacket (Lomir Biomedical, Malone, NY, USA) that was attached by a tether to the rear wall of the experimental cubicle (1.0 m³, Plaslabs, Lansing, MI, USA). The front door of the cubicle was made of transparent plastic, and the remaining walls were opaque. Two response levers (PRL-001, BRS/LVE, Beltsville, MD, USA) were mounted inside the door. Four stimulus lights, two red and two white, were mounted above each lever. Drug injections were delivered by a peristaltic infusion pump (7540X; Cole-Parmer, Chicago, IL, USA). A Macintosh computer (Macintosh Quadra 950, Apple Inc., Cupertino, CA, USA) with custom interface and software controlled experimental events and data collection.

Surgery

Each subject had a single-lumen silicone catheter (Cole-Parmer Co., Chicago, IL, USA) implanted according to a previous protocol (Freeman and Woolverton 2009). Briefly, subjects were injected with atropine sulfate (0.04 mg/kg, i.m.) and ketamine hydrochloride (10–20 mg/kg, i.m.) followed by inhaled isoflurane and preoperative antibiotics (cefazolin; 20–25 mg/kg, i.m.) and analgesics (carprofen, 2–4 mg/kg, s.c. and/or buprenorphine SR, 0.05 mg/kg, s.c.). Under aseptic conditions, the catheter was surgically implanted into a major vein (brachial, jugular, or femoral) with the tip terminating near the right atrium. The distal end of the catheter was passed subcutaneously to the mid-scapular region, where it exited the subject’s back. After surgery, subjects were returned to their home cage where the catheters were passed through a tether and connected to a swivel (Lomir Biomedical, Malone, NY, USA) mounted to the rear wall of the cubicle. From the outside of the swivel, a catheter line was connected to an infusion pump for drug delivery. When recommended by veterinary staff, an antibiotic (usually Kefzol; Eli Lilly & Company, Indianapolis, IN, USA) was administered twice daily (22.2 mg/kg, p.o.) post-surgery for five to seven days to prevent infection. If a catheter became nonfunctional during the experiment, it was removed, and the monkey was removed from the experiment for a 1–2 week period. After health was verified, a new catheter was implanted. Between sessions, catheters were filled 40 units/ml heparin to deter clotting.

General Procedure

Sessions began at 1100 h and were conducted daily. At least 30 min prior to each session, each subject’s catheter was filled with saline or drug, corresponding to the test or baseline condition, in an amount sufficient to make the solution available for injection after completion of the first response requirement. The white lights above the right lever indicated the start of the session. Pressing the right lever resulted in saline or drug delivery over a 10-s infusion, and responses on the left lever were recorded but had no programmed consequences. During injections, the white lights were darkened, and the red lights were illuminated. Each injection was followed by a 30-min inter-trial interval during which all lights were darkened and responses on either lever were recorded but had no programmed consequences.

Lever pressing during baseline and test sessions were maintained under a PR schedule of reinforcement similar to previous protocols (Freeman et al. 2014a; Rowlett et al. 1996). Each session consisted of five components, with four trials per component, resulting in a maximum of 20 injections per session. The response requirement within each component was fixed and doubled across components. For example, if the response requirement in the first component was 50, the requirements in subsequent components were 100, 200, 400, and 800. Response requirements were tailored for each subject to render a comparable number of cocaine injections during baseline sessions. Specifically, the response requirements for the first component was 50 for subjects MR4351M, 1356AS, 1010AL, 9512L; 100 for subject 314001; 125 for subject 314206B; and 300 for subject 4299. A 30-min limited-hold was in effect for each trial; if 30 min passed without successful completion of the response requirement, the white lever lights darkened, and the trial ended with a 30-min inter-trial interval as described above. Sessions ended with the completion of 20 injections or after two consecutive trials in which the response requirement was not completed.

All subjects were trained on an alternating sequence of cocaine (0.1 mg/kg/injection) and saline availability in which cocaine was offered for a session followed by a session of saline availability the following day. Training was considered complete when subjects consistently earned ≥ 12 injections on cocaine sessions and ≤ 4 injections on saline sessions. Once this stable baseline was achieved, substitution tests were inserted into an alternating daily sequence as follows: saline, test, saline, cocaine, test, random (saline or cocaine), cocaine. Cocaine (0.1 mg/kg/inj) was used as a baseline drug condition rather than an opioid such as oxycodone to avoid complications that could occur with serial opioid self-administration (e.g., tolerance, dependence). We and others have found this approach to be highly stable within a study and replicable across studies (Freeman et al. 2014a; Rowlett et al. 1996).

Experiment 1: Effects of Salvinorin A on Oxycodone Self-Administration.

During test sessions, a dose of oxycodone alone (0.012–0.1 mg/kg/injection) or the same doses of oxycodone mixed with salvinorin A (0.006 or 0.012 mg/kg/injection) were made available for self-administration. Cocaine (0.1 mg/kg/injection) and saline also were substituted within the test sequence. Saline and each drug or drug-combination was tested twice: once after a cocaine baseline session and once after a saline baseline session to ensure that previous-day conditions did not affect the test outcomes. An average of the two tests constituted a data point for a drug condition for a particular subject. Drug conditions were tested in irregular order within and across subjects. When the number of injections within two test sessions for a particular dose were variable (the number of injections exceeded ± 3 injections of the mean), the dose was tested two more times (four times total), once after a saline baseline session and once after a cocaine baseline session, and the average of all 4 test sessions were used regardless of outcome. Subjects MR4351M, 314001, 4299, and 1356AS were used for this experiment, and 1356AS first completed Experiment 2 before completing Experiment 1.

Experiment 2: Effects of Nalfurafine on Oxycodone Self-Administration.

The test conditions for Experiment 2 were the same as Experiment 1 with three exceptions: 1) the KOR agonist combined with oxycodone was nalfurafine (0.0001–0.00032 mg/kg/injection), 2) nalfurafine also was tested alone, and 3) the doses of oxycodone and nalfurafine were arranged on a quarter-log scale (rather than log base 2 as in Experiment 1). Tests were conducted in an irregular order within and across subjects, and subjects 1356AS, 1010AL, 9512L, and 314206B were used in this experiment.

To determine if nalfurafine’s effects were mediated by the KOR, the peak reinforcing dose of oxycodone alone (0.056 mg/kg/injection for 1356AS and 9512L; 0.032 mg/kg/injection for 1010AL) or combined with nalfurafine (0.0001–0.00032 mg/kg/injection) was re-determined in three subjects to verify that these points were consistent with those determined in the dose-response test above. Once the re-determinations were complete, the KOR antagonist, nor-Binaltorphimine (nor-BNI; 10 mg/kg i.m.; dose and route based on Butelmann et al. 1998), was administered to all subjects at least 24 h prior to retesting the same conditions (i.e., the peak dose of oxycodone alone or combined with nalfurafine). The i.m. injections of nor-BNI were delivered under methohexital sedation (3.3 mg/kg, i.v.). Drug conditions before and after nor-BNI treatment were tested in an irregular order within and across subjects.

Drugs

Oxycodone hydrochloride and cocaine hydrochloride were generously provided by the National Institute on Drug Abuse (NIDA). Methohexital was purchased from UMMC’s Pharmacy. Nalfurafine hydrochloride, nor-BNI, and salvinorin A were synthesized and provided by Dr. Thomas E. Prisinzano. Cocaine, oxycodone, and nalfurafine were prepared in 0.9% sterile saline solution. Mixtures of oxycodone and salvinorin A were dissolved in a vehicle of 1:1:18 ethanol:Tween 80:sterile saline. Nor-BNI was dissolved in a vehicle of 1:99 lactic acid:sterile water for i.m. injections. All solutions were passed through a 0.22 μm Millipore filter prior to administration to an animal.

Data Analysis

Experiment 1: Effects of Salvinorin A on Oxycodone Self-Administration.

In Experiment 1, each dose of oxycodone alone (0.012–0.1 mg/kg/injection) or in combination with salvinorin A (0.006 and 0.012 mg/kg/injection) was functionally determined for individual subjects, and all oxycodone doses were not experienced by all subjects. Because this resulted in missing data in some instances, three separate repeated-measures analyses of variance (ANOVAs) were conducted. The first ANOVA was conducted for oxycodone alone (saline and 0.012–0.1 mg/kg/injection of oxycodone). The second was conducted with saline and two oxycodone doses (0.025 and 0.05 mg/kg/injection) combined with the smallest salvinorin A dose (0.006 mg/kg/injection) because these dose combinations were experienced by all subjects, and a single paired t-test was used to compare the largest dose of oxycodone (0.1 mg/kg/injection) combined with this dose of salvinorin A (0.006 mg/kg/injection) to saline because only three subjects completed this combination. The third ANOVA was conducted with saline and three oxycodone doses (0.025, 0.05, and 0.1 mg/kg/injection) when combined with the largest salvinorin A dose (0.012 mg/kg/injection). For all ANOVAs, Dunnett’s multiple comparisons were used to compare each oxycodone dose or each oxycodone and salvinorin A combination to saline. Geisser-Greenhouse method corrected for violations of sphericity, and outcomes were considered significant when p<0.05.

Experiment 2: Effects of Nalfurafine on Oxycodone Self-Administration.

In Experiment 2, each dose of oxycodone alone (0.01–0.1 mg/kg/injection) and each dose of oxycodone combined with each dose of nalfurafine (0.0001–0.00032 mg/kg/injection) were experienced by all subjects. As a result, a single two-way repeated-measures ANOVA was conducted with oxycodone dose (saline and 0.01–0.1 mg/kg/injection) and nalfurafine dose (0.0001–0.00032 mg/kg/injection) as within-subject factors. Dunnett’s multiple comparisons were used to compare each oxycodone dose, alone or combined with each nalfurafine dose, relative to saline. Because some combinations of oxycodone and nalfurafine were statistically different from saline (i.e., functioned as reinforcers), another two-way repeated-measures ANOVA with oxycodone dose (0.018–0.1 mg/kg/injection) and nalfurafine dose (0.0001–0.00032 mg/kg/injection) as within-subject factors was conducted. For this analysis, only oxycodone doses that were significantly different from saline were used to determine which oxycodone and nalfurafine combinations were different from oxycodone alone. Dunnett’s multiple comparisons were used to compare oxycodone alone to the corresponding dose of oxycodone combined with each nalfurafine dose.

For nalfurafine alone, one subject did not experience the smallest nalfurafine dose (0.0001 mg/kg/injection). Therefore, a one-way repeated-measures ANOVA was conducted with nalfurafine dose (saline and 0.00018–0.00032 mg/kg/injection) as a within-subjects factor, and Dunnett’s multiple comparisons were used to compare each nalfurafine dose to saline. A single paired t-test was used to compare nalfurafine (0.0001 mg/kg/injection) to saline for the three subjects that completed this dose.

To evaluate effects of nor-BNI on oxycodone and nalfurafine self-administration, a repeated-measures ANOVA was conducted with each test point (saline, peak dose of oxycodone alone and combined with nalfurafine before and after nor-BNI) as a within-subjects factor. Dunnett’s multiple comparisons were used to compare each test point relative to saline. Because some doses of oxycodone combined with nalfurafine (before and after nor-BNI administration) were statistically different from saline, another repeated-measures ANOVA was conducted with only those test points (i.e., only with functional reinforcers) and Dunnett’s multiple comparison’s to determine which oxycodone and nalfurafine combinations (before and after nor-BNI administration) were statistically different from oxycodone alone.

Results

Figure 1 illustrates the mean number of injections and standard error (SEM) per session for saline tests (lower shaded portion of the figure), cocaine tests (0.1 mg/kg/injection; upper shaded portion of the figure), each dose of oxycodone alone (circles), and when combined with 0.006 mg/kg/injection (upright triangles) and 0.012 mg/kg/injection (upside down triangles) of salvinorin A. Oxycodone alone was self-administered above saline levels, indicated by a significant main effect of oxycodone dose [F(1.9,5.8)=25.6, p<0.05]. Specifically, mean injections per session were greater compared with saline at 0.025, 0.05, and 0.1 mg/kg/injection of oxycodone (p’s<0.05). However, when mixed with salvinorin A, oxycodone was self-administered at levels that were not significantly different from saline (p’s>0.05).

Fig. 1.

The mean number of injections/session is shown on the y-axis as a function of the dose of oxycodone (0.012–0.1 mg/kg/injection) on the x-axis. The lower shaded portion represents the mean and standard error (SEM) for saline tests and the upper shaded portion represents the mean and SEM for cocaine (0.1 mg/kg/injection) tests. Data are shown for oxycodone alone (circles) and when combined with 0.006 (upright triangles) and 0.012 (upside down triangles) mg/kg/injection of salvinorin A. Filled symbols represent data points that are significantly different from saline. Error bars represent one SEM and numbers in parenthesis indicate the number of subjects that contributed to a data point when all subjects did not complete a particular dose or dose combination.

Figure 2 illustrates the mean number of injections and SEM per session for saline tests (lower shaded portion, both panels), cocaine tests (0.1 mg/kg/injection; upper shaded portion, both panels), nalfurafine alone (left panel, diamonds), oxycodone alone (right panel, circles), and oxycodone combined with 0.0001 (right panel; upright triangles), 0.00018 (right panel; upside down triangles), and 0.00032 (right panels; squares) mg/kg/injection of nalfurafine. The mean number of injections per session when nalfurafine was available without oxycodone (left panel) were not significantly different from saline at any nalfurafine dose evaluated (p>0.05). However, oxycodone alone was self-administered above saline levels, indicated by a significant main effect of oxycodone dose [F(5,15)=13.7, p<0.05]. Mean injections per session for all doses of oxycodone alone except the 0.01 mg/kg/injection dose were significantly greater compared with saline (p’s<0.05). When oxycodone was mixed with nalfurafine, nalfurafine decreased the mean number of injections per session in a dose-dependent manner, and this was indicated by a significant main effect of nalfurafine dose [F(3,9)=24.7, p<0.05] and a significant oxycodone dose x nalfurafine dose interaction [F(15,45)=4.6, p<0.05]. While nalfurafine significantly reduced oxycodone self-administration in most cases, the mean number of injections per session for many oxycodone and nalfurafine combinations were still significantly greater compared with saline. The two exceptions were 0.032 and 0.056 mg/kg/injection of oxycodone combined with 0.00032 mg/kg/injection of nalfurafine (p’s>0.05).

Fig. 2.

The mean number of injections/session is shown on the y-axis as a function of dose for nalfurafine (0.0001–0.00032 mg/kg/injection; left panel) and oxycodone (0.01–0.1 mg/kg/injection; right panel) on the x-axis. The lower shaded portion in each panel represents the mean and SEM for saline tests and the upper shaded portion in each panel represents the mean and SEM for cocaine (0.1 mg/kg/injection) tests. Data are shown for nalfurafine alone (circles, left panel), and oxycodone alone (circles, right panel) and when combined with 0.0001 (upright triangles), 0.00018 (upside down triangles), and 0.00032 (squares) mg/kg/injection of nalfurafine. Filled symbols represent data points that are significantly different from saline. Minus symbols (–) represent no significant difference from oxycodone, asterisk symbols (*) represents a significant difference from oxycodone, and minus and asterisk symbols are organized from the smallest nalfurafine dose at the top, intermediate dose in the middle, and largest nalfurafine dose at the bottom. Error bars represent one SEM.

Because some combinations of oxycodone and nalfurafine were self-administered at levels significantly greater compared to saline, a second overall analysis was conducted to determine which oxycodone and nalfurafine combinations were different from oxycodone alone (Figure 2). At the 0.018 mg/kg/injection dose of oxycodone, none of the oxycodone and nalfurafine combinations were self-administered at levels significantly different from oxycodone alone (p’s>0.05). At the 0.032 mg/kg/injection dose of oxycodone, oxycodone when combined with 0.00018 and 0.00032 mg/kg/injection of nalfurafine were self-administered at levels significantly lower than oxycodone alone (p’s<0.05). At both the 0.056 and 0.1 mg/kg/injection doses of oxycodone, all combinations of oxycodone and nalfurafine were self-administered at levels significantly lower than oxycodone alone (p’s<0.05).

Figure 3 shows the mean number of injections per session for saline, the peak dose of oxycodone alone or combined with nalfurafine before and after nor-BNI administration. Prior to nor-BNI administration, oxycodone, whether administered alone or combined with 0.0001 mg/kg/injection of nalfurafine, was self-administered at levels significantly greater than saline (p’s<0.05). Oxycodone combined with 0.00018 and 0.00032 mg/kg/injection of nalfurafine was self-administered at levels not significantly different from saline (p’s>0.05). However, following nor-BNI administration, oxycodone combined with 0.00018 and 0.00032 mg/kg/injection of nalfurafine was self-administered at levels significantly greater than saline (p’s<0.05), indicating a reversal of nalfurafine’s punishing effect. When test points that were significantly greater than saline were compared to oxycodone alone, neither oxycodone combined with 0.0001 mg/kg/injection of nalfurafine nor oxycodone combined with 0.00018 and 0.00032 mg/kg/injection of nalfurafine after nor-BNI administration were significantly different (p’s>0.05).

Fig. 3.

The mean number of injections/sessions is shown on the y-axis and separate test conditions are shown on the x-axis (saline, oxycodone alone and combined with each dose of nalfurafine before and after nor-BNI administration). Plus and minus symbols are used to indicate the presence (+) or absence (−) of oxycodone (0.032 or 0.056 mg/kg/injection) and nor-BNI (10 mg/kg, i.m.). The presence of nalfurafine is indicated by the dose and the absence of nalfurafine is indicated by a minus symbol. For example, three minus symbols indicates saline while two plus symbols and a number in the nalfurafine row indicates the presence of oxycodone, nalfurafine, and nor-BNI. Filled bars represent conditions that are significantly different from saline. Error bars represent one SEM.

Discussion

The results of the current study demonstrate that the KOR agonists, salvinorin A and nalfurafine, can punish oxycodone self-administration in rhesus monkeys when they are taken contingently with oxycodone under a PR schedule of reinforcement. Salvinorin A and nalfurafine reduced the number of oxycodone injections earned in a dose-dependent manner, indicating that combinations of oxycodone and the KOR agonists are less effective reinforcers than oxycodone alone in this species, similar to what we have previously reported in rats (Townsend et al. 2017). Nalfurafine’s punishing effect was reversed by pretreatment with the KOR antagonist, nor-BNI, thus demonstrating that nalfurafine reduces oxycodone’s reinforcing effects through a KOR-dependent mechanism.

Previous reports have demonstrated that the abuse potential of drugs can be predicted by their relative reinforcing effectiveness in a PR design (Wee et al. 2005; Wilcox et al. 2000). Using the same PR design as the current approach, we previously reported that combining the drug punisher, histamine, with cocaine not only reduced cocaine’s potency as a reinforcer but also reduced cocaine’s effectiveness as a reinforcer (i.e., decreased breakpoints in an insurmountable manner; Freeman et al. 2014a). Given that reductions in reinforcing effectiveness may reflect abuse potential, an important implication from the current experiments is that the abuse potential of oxycodone may be diminished when it is coupled with punishing drugs like KOR agonists, a concept that translates well to the development of ADFs for MOR agonist medications.

A critical consideration for combining KOR agonists with MOR agonists is the KOR agonist to be used in the formulation. Salvinorin A is useful for providing proof-of-concept information because it is highly selective for the KOR and is short acting (Schmidt et al. 2005), the latter of which mitigates concerns about drug-accumulation across injections. However, most KOR agonists, including salvinorin A, produce untoward side effects of sufficient severity to limit their clinical utility (Jones et al. 2016; Prisinzano et al. 2005). Nalfurafine is unusual among KOR agonists in that it does not produce the perceptual distortions or dysphoric effects that are typical of the drug class, despite being a fully efficacious agonist at the KOR (Kaski et al. 2019; Schattauer et al. 2017) and an effective punisher of oxycodone self-administration. As such, psychotomimesis and/or dysphoria do not appear to be necessary components of KOR-mediated punishment of drug self-administration, which increases nalfurafine’s feasibility for clinical use in an ADF. A notable difference between our current results with rhesus monkeys and our previous report studying self-administration of oxycodone/nalfurafine mixtures in rats is that the ratio of nalfurafine in the mixture that was required to reduce oxycodone self-administration was much lower in rhesus monkeys (1:175 nalfurafine:oxycodone in monkeys; 1:18 in rats; Townsend et al., 2017), indicating that primates may be more sensitive to the deterring effects of this drug. This 1:175 ratio also falls within a ratio range of nalfurafine/morphine mixtures that were reported to produce additive thermal antinociception in rhesus monkeys (Ko and Husbands 2009), suggesting that ratios of oxycodone and nalfurafine that do not exhibit abuse-related effects may also produce greater antinociception than oxycodone alone.

Self-administration studies using a PR schedule of reinforcement are useful for scaling drugs in terms of reinforcing effectiveness and for measuring the mitigating effects of treatments on a drug’s abuse potential (Richardson and Roberts 1996). However, interpreting results from single-lever operant designs like the one used here must take into consideration the possibility that the treatments reducing self-administration may be doing so by mechanisms other than a selective diminishment of reinforcing effect. This “nonspecific” disruption of responding can be caused by a number of drug-induced effects such as stereotypies, nausea, or sedation (see Lynch and Carrol, 2001). One approach used to control for nonspecific effects is to test the effects of the treatment on self-administration of a nondrug reinforcer such as food. However, drugs that produce punishing effects would be expected to reduce self-administration of all reinforcers with which they are paired, including food. As such, a food comparison is not an appropriate control for the current study. Drug-choice procedures also are useful for ruling out nonspecific effects of treatments because measures of drug choice are proportion based and not dependent on response rate (Banks & Negus, 2017). As stated above, we have reported that salvinorin A reduces remifentanil and cocaine self-administration in a choice procedure without affecting response rate (Freeman et al. 2014a), which demonstrates that a KOR agonist can produce selective punishing effects. Work is currently underway in our laboratory to test the punishing effects of nalfurafine on oxycodone choice, but preliminary data indicate that nalfurafine functions as a punisher of cocaine and oxycodone self-administration in a drug vs. food choice design in rhesus monkeys (unpublished data). As such, the accumulating evidence suggests that the reductions in oxycodone self-administration produced by contingent KOR agonist administration are due, at least in part, to a punishing effect. However, it is important to note here that punishment as a dependent measure is derived from the observation of a stimulus-response relation (i.e., a behavior) and does not measure or infer the qualitative nature of accompanying subjective effects (Azrin and Holz 1966). Thus, it is unknown to what degree aversion or other effects (e.g., perceptual masking) underlie the ability of one or both KOR agonists to reduce the effectiveness of oxycodone as a reinforcer. Establishing the stimulus nature of the deterring effects of these drugs, and whether they differ within the drug class, will be an interesting next step.

There are a number of feasibility issues that must be considered when proposing the use of KOR agonists (or any punishing agent) as formulations with prescription opioids for abuse deterrence. First, punishing agents that reduce nonmedical use of prescription opioids could also negatively affect patient compliance if the stimulus effect of the punishing agent is too “heavy handed” or is not selectively applied to circumstances of misuse. Ideally, a punishing stimulus should only emerge if the patient exceeds the prescribed dose or decreases the inter-dosing interval. However, occasional occurrences of inadvertent misuse by patients could result in the experience of a punishing effect, which will require further patient education at the time of prescribing. Another consideration is the complementarity of the pharmacokinetic profiles of the drugs to be paired in the formulation. We know of no published work that has directly compared durations of behavioral effects for nalfurafine and oxycodone in nonhuman primates. However, morphine, which is similar to oxycodone in efficacy and effect duration (Villesen et al. 2007), was reported to produce durations of behavioral effects in cynomolgus monkeys, including thermal antinociception, that were comparable to nalfurafine (up to 6 h; Endoh et al. 2001). Thus, the kinetics of the two drugs would be expected to be reasonably matched. However, we have previously reported that a drug punisher (histamine) that has a much shorter duration of effect than the drug reinforcer with which it was paired (cocaine) was just as effective at reducing drug self-administration using the current PR design in rhesus monkeys as the KOR agonists in the current report (Freeman et al. 2014a), suggesting that comparable durations of effect between a drug punisher and the reinforcer with which it is paired may not be a critical parameter for effective punishment. A final consideration is the fact that nalfurafine is already prescribed to humans as an oral medication (Nakao and Mochizuki 2009). Thus, the feasibility of using nalfurafine and oxycodone as an oral formulation is high. An important next step will be to test the subjective and/or reinforcing effects of orally administered combinations of oxycodone and nalfurafine in a human laboratory setting.

The primary goal of the current study was to determine if KOR agonists, one typical and one atypical in terms of behavioral profiles, could decrease the reinforcing effectiveness of oxycodone in rhesus monkeys. The results suggest that an atypical KOR agonist that does not produce dysphoria or psychotomimesis is effective at decreasing the abuse-related effects of oxycodone. As newer KOR agonists with “milder” profiles than nalfurafine are developed (e.g., triazole 1.1; Brust et al. 2016), it will be interesting to determine if there is a point at which the behavioral profile of a KOR agonist becomes sufficiently “mild” to lose effectiveness at producing punishment of drug self-administration. Alternatively, it may be the case that the effectiveness of KOR agonists at decreasing the reinforcing effects of other drugs does not require the typical KOR-mediated adverse effects for which the drug class is known. The next step will be to systematically compare KOR agonists of varying behavioral and pharmacological profiles (e.g., along a continuum of G-protein signaling bias) as punishers of MOR agonist self-administration.