Comparison of the reinforcing, antinociceptive, and respiratory depressant effects of prototypical and G-protein biased mu-opioid receptor agonists in male and female Sprague-Dawley rats

Abstract

Rationale:

G-protein biased mu-opioid receptor (MOR) agonists have been reported to exhibit superior therapeutic windows compared to prototypical MOR agonists when relating antinociception to respiratory depression. However, there is relatively little research on the abuse potential of G-protein biased MOR agonists in relation to other behavioral endpoints.

Objectives:

The aim of the present study was to quantitatively compare the reinforcing, antinociceptive, and respiratory-depressant effects of the prototypical MOR agonists, fentanyl and oxycodone, to the G-protein biased MOR agonists, SR14968 and SR17018, in male and female rats.

Methods:

In the self-administration study, four separate groups of Sprague-Dawley (SD) rats self-administered intravenous (i.v.) fentanyl, oxycodone, SR14968, and SR17018 under a progressive-ratio schedule of reinforcement. Using a within-subjects design, separate cohorts of SD rats were tested with i.v. fentanyl, oxycodone, SR14968, and SR17018 using a hot-plate assay, assays of neuropathic and inflammatory antinociception, and whole-body plethysmography.

Results:

All MOR agonists functioned as reinforcers, but SR14968 and SR17018 were less efficacious relative to oxycodone and fentanyl. Moreover, all MOR agonists produced dose-dependent and fully efficacious antinociception across all nociception modalities. Oxycodone and fentanyl, but not SR14968 or SR17018, produced respiratory depression in a dose-dependent manner.

Conclusion:

The present results indicate that the G-protein biased MOR agonists tested herein produce MOR-typical antinociception, exhibit reduced but apparent abuse potential, and do not produce respiratory effects at doses that are above the antinociceptive range. Atypical MOR agonists within the SR series should be further studied as foundational molecules for the development of safter analgesics.

Introduction

Mu-opioid receptor (MOR) agonists (e.g., morphine) are the most effective treatments for moderate-to-severe pain (Gross et al. 2019). However, MOR agonists produce side effects, such as respiratory depression, dependence, and abuse that add risks to their clinical use (Chan, 2017). The MOR is a member of the superfamily of G-protein coupled receptors (GPCRs) that signal through distinct second messenger systems to produce behavioral effects (Kieffer, 1999; Pasternak and Pan 2013). In recent years, two intracellular-signaling processes have received increasing attention in MOR pharmacology; the G-protein-signaling pathway and the recruitment of β-arrestin-2 with recent studies focusing on their relative roles in distinct MOR-mediated behavioral effects (Roth et al. 1998; Siuda et al. 2017). The possibility that behavioral effects of MOR agonists are differentially mediated by separate intracellular signaling processes has given rise to drug-development strategies focused on identifying molecules that can preferentially activate one pathway over another in a ligand-specific manner (Brust et. al. 2016; DeWire et. al. 2013; Varadi et. al. 2016), a concept that has been termed “biased agonism” (Kenakin 2003; Rankovic et al. 2016). According to this concept, when a G-protein biased MOR agonist binds to the receptor, it exhibits a higher potency and/or efficacy to activate the G-protein pathway over the recruitment of β-arrestin-2 compared to a standard MOR agonist (e.g., oxycodone). According to recent theories regarding biased signaling, increases the compound’s therapeutic selectivity (Pergolizzi et al. 2018).

To test the hypothesis that G-protein bias improves the therapeutic selectivity of MOR agonists, Schmid et al. (2017) evaluated in mice a series of MOR agonists that spanned the bias spectrum between G-protein signaling and β-arrestin-2 recruitment in tests of antinociception and respiratory depression. In that report, the degree of G-protein bias was positively associated with the therapeutic windows of the compounds, defined as the potency to produce respiratory depression divided by the potency to produce thermal antinociception. That is, relative to prototypical MOR agonists, G-protein biased MOR agonists produced antinociception at much lower doses than those required to produce respiratory effects.

The relation between the signaling bias of a MOR agonist and its effects in preclinical tests of abuse liability has received less experimental attention. Early reports in β-arrestin-2 knock-out (KO) mice demonstrated that morphine produced conditioned place preferences (CPP) in subjects lacking this protein, suggesting that the therapeutic and rewarding effects of MOR agonists were both mediated by G-protein signaling (Bohn et al. 1999; Bohn et al. 2000; Bohn et al. 2003). Consistent with this finding, the G-protein biased MOR agonist, TRV130, was reported to produce effects on intracranial self-stimulation (ICSS) in rats (Altarifi et. al. 2017) and subjective effects in humans (Soergel et. al. 2014) that were comparable to morphine, outcomes suggestive of abuse liability. Conversely, equi-antinociceptive doses of morphine and the G-protein-biased MOR agonists, TRV130 and PZM21, produced disparate results in mice in CPP tests, with morphine alone producing a place preference (Kudla et al. 2019; Manglik et al. 2016). However, PZM21 and TRV130 have been reported to be self-administered in nonhuman primates and rats, respectively (Ding et al., 2020; Zamarripa et al. 2018), and the biased MOR agonists, SR14968 and SR17018, were recently reported to produce CPP in mice (Kudla et al. 2021). Taken together, the above results suggest that G-protein biased agonists will have abuse potential, but may have more favorable therapeutic windows than prototypical MOR agonists when relating potencies to produce antinociception to potencies to produce abuse-related and other adverse effects.

The goal of the current study was to conduct a systematic comparison of the reinforcing, antinociceptive, and respiratory-depressant effects of fentanyl (reported to be β-arrestin-2-biased; Schmid et al. 2017), oxycodone (reported to be relatively unbiased relative to prototypical MOR agonists; Rivero et al. 2012), SR14968 (reported to be moderately G-protein biased; Schmid et al. 2017), and SR17018 (reported to be largely G-protein biased; Schmid et al. 2017) in male and female Sprague-Dawley rats. Dose-response assessments for fentanyl, oxycodone, SR14968, and SR17018 as reinforcers in self-administration were determined using a PR schedule of reinforcement. Additionally, dose-response assessments of the antinociceptive effects for fentanyl, oxycodone, SR14968, and SR17018 were determined using the hot-plate assay (thermal nociception) and the electronic von Frey approach (neuropathic and inflammatory antinociception). Finally, pulmonary effects of fentanyl, oxycodone, SR14968, and SR17018 were measured using whole-body plethysmography.

Methods and Materials

Subjects

Male and female Sprague-Dawley rats (Envigo Laboratories, New Jersey, USA) arrived at approximately 65 days of age and weighing 275-299 g and 200-224 g, respectively. Subjects were acclimated to their housing chambers for at least one week prior to beginning experiments, pair housed with ad libitum access to food and water, and maintained on a reversed 12-h light/dark cycle (lights off at 0700) with experiments occurring during the dark phase. Experiments were conducted in compliance with the National Research Council’s Guide for Care and Use of Laboratory Animals (2011) and approved by the University of Mississippi Medical Center’s Institutional Animal Care and Use Committee.

Surgery and Apparatus

All subjects received intravenous (i.v.) catheters implanted into their left femoral vein. Details regarding surgical methodology and the apparati (operant chambers, hot-plate apparatus, electronic von Frey, and whole-body plethysmography) can be found in the Supplemental Methods.

Drugs

Oxycodone hydrochloride and cocaine hydrochloride were generously provided by the National Institute on Drug Abuse (NIDA). Fentanyl citrate and naltrexone hydrochloride were purchased from Sigma. Methohexital was purchased from the UMMC Pharmacy. SR14968 and SR17018 were synthesized and provided by Dr. Bruce E. Blough. Oxycodone, fentanyl, cocaine, and naltrexone were prepared in 0.9% sterile saline solution. SR14968 and SR17018 were dissolved in a vehicle of 45% β-cyclodextrin and 55% sterile water. All vehicle conditions during procedures with within-subject designs (i.e., tests of antinociception and respiratory depression; described below) were tested using a solution made of 45% β-cyclodextrin and 55% sterile water as it was the vehicle required to dissolve the SR compounds.

Procedures

Self-administration of MOR agonists

Thirty-two subjects (16 males, 16 females) were trained to self-administer cocaine (0.56 mg/kg/injection) on a fixed-ratio (FR) then progressive-ratio (PR) schedule of reinforcement. A detailed self-administration training protocol can be found in the Supplemental Methods. Once training was completed, subjects were divided into four separate groups corresponding to one of the four MOR agonists, with each group consisting of four male and female subjects. The dose ranges for each MOR agonist were as follows: fentanyl (0.0001-0.00056 mg/kg/injection), oxycodone (0.01-0.1 mg/kg/injection), SR14968 (0.001-0.0056 mg/kg/injection), and SR17018 (0.1-0.56 mg/kg/injection). These doses were functionally determined to include doses that engendered intakes that did not differ from vehicle and up to full effect (i.e., an asymptotic point in the dose-response function). A vehicle test condition was included within the counterbalanced dose sequence for each drug group as a non-reinforcer control. Saline was used as the vehicle condition for fentanyl and oxycodone, and the 45% β-cyclodextrin vehicle was used for both SR compounds. All doses within a drug condition were tested in an irregular order. For each dose condition, stability criteria were: a) injections earned across 3 days had to be within 20% of the 3-session mean, and b) no upward or downward trends.

After each completed test of a drug condition and previous to the subsequent test, the training dose of cocaine (0.56 mg/kg/injection) was offered for one session on the PR schedule to monitor the stability of responding for the training dose over the time period required to test all drug conditions, and this “return to baseline” session was followed by a series of saline sessions sufficient to reduce responding to within 20% of each subject’s pre-test saline-injection baseline. Notably, the mean cocaine injections for each of these sessions were within 20% of the pretest baseline-injection mean for all subjects, and all subjects decreased responding to within 20% of their pre-test saline levels within 3 sessions of saline availability. As such, all test conditions were conducted within a context of comparable baseline performance.

Thermal Antinociception

Ten subjects, 5 males and 5 females, were implanted with i.v. catheters, and all test drugs were administered i.v. On Day 1 of the study, each subject was placed on the unheated apparatus for an acclimation period of 60 sec before being removed from the hot plate and placed in their home cage. Following this acclimation exposure period, the surface of the apparatus was heated to 52.5 °C. Each subject received an injection of saline, followed by a 20-min period before being placed on the hot plate for a test of nociception. This injection-test series was repeated for a total of 5 serial tests to acclimate the subjects to the cumulative-dosing procedure (described below). Notably, all subjects were removed from the hot plate immediately following a nociceptive response, which was defined as a paw lift or lick, a shake, or vocalization.

All test sessions started with a vehicle injection followed 20 min later by placement on the hot plate heated to 52.5 °C to serve as a within-session control response. Subjects were removed from the hot plate immediately after emitting a nociceptive response, and latency (in seconds) to emit the response was recorded. Following the vehicle control test, subjects received increasing doses of fentanyl (0.0018-0.1 mg/kg), oxycodone (1.0-5.6 mg/kg), SR14968 (0.1-0.56 mg/kg), or SR17018 (1.8-10.0 mg/kg) using the same injection and timing parameters described for the saline test above. Thus, all doses of a particular drug were tested within a single session using a cumulative-dosing approach that rendered a dose range in quarter-log intervals that encompassed a full range effect (≤20% to ≥80% maximum possible effect). The 20-minute inter-injection intervals for all tests of nociception (described below) were chosen based on pilot testing with the von Frey apparatus, demonstrating that this window was optimal for test administrators to administer each drug condition and complete behavioral testing without overlapping into the next administration/test period. If 60 s passed without a nociceptive response, subjects were removed from the hot plate by the experimenter. All MOR agonists were tested in all subjects on different days in a counterbalanced order, and all tests occurred at least 72 h apart. An experimenter blinded to the test conditions scored and recorded response latencies while another experimenter administered injections.

Following completion of the dose-response assessments for the four MOR agonists, each subject received an injection of naltrexone (0.032 mg/kg, i.v.) at the start of test sessions of antinociception, and the individual MOR agonists were retested in a counterbalanced order using the cumulative-dosing method described above. The dose of naltrexone was selected based on its ability to attenuate oxycodone thermal antinociception in a pilot study.

Neuropathic Antinociception

Twenty subjects, 10 males and 10 females, were implanted with i.v. catheters, and all test drugs were administered i.v. Figure 3A illustrates the experimental timeline. On Day 0 of the study, each subject was placed in the plastic enclosure and acclimated for 30 minutes. Following this acclimation period, each subject received an injection of saline, and the baseline paw withdrawal thresholds (in grams) for both hindpaws were recorded three times. Each injection was followed by a 20-min period prior to subsequent paw withdrawal measurements. This within-session test series was repeated for a total of 5 serial tests to acclimate the subjects to the cumulative-dosing procedure (described above) before the subjects were returned to their home cage. Following the acclimation testing, subjects were counterbalanced into two separate groups based on their baseline paw withdrawal thresholds: vehicle-treated or paclitaxel (PTX)-treated (and balanced for sex).

Figure 3.

Panel A illustrates the experiential time for the neuropathic antinociception study. Panel B illustrates the paw withdrawal threshold (in grams) for the vehicle-treated (downward triangle) and paclitaxel-treated (upward triangle) groups across days following paclitaxel (PTX) administration. BL = baseline paw withdrawal thresholds prior to PTX administration. Panel C illustrates the paw withdrawal threshold (in grams) for fentanyl (0.0018-0.0056 mg/kg; downward triangle), SR14968 (0.0056-0.018 mg/kg; square), oxycodone (0.1-0.32 mg/kg; circle), and SR17018 (0.18-056 mg/kg; upward triangle) for the PTX-treated groups. The lower shaded grey bar represents the paw withdrawal threshold under vehicle conditions for the paclitaxel-treated group. The upper shaded grey bar represents the paw withdrawal threshold under vehicle conditions for the vehicle-treated group. Panels D through G illustrated the paw withdrawal threshold (in grams) for fentanyl (0.0056 mg/kg), SR14968 (0.018 mg/kg), oxycodone (0.32 mg/kg), and SR17018 (0.56 mg/kg) before and after administration of the opioid antagonist naltrexone (0.032 mg/kg; i.v.). Filled bars represents a significant difference from vehicle (saline) administration. Error bars indicate ± SEM.

On Day 1, all subjects were injected intraperitoneally every other day for a total of four injection days (Days 1, 3, 5, and 7) with vehicle (saline) or PTX (2.0 mg/kg; Legakis et al. 2018). During this period, no baseline recordings occurred. Starting on Day 8, baseline paw withdrawal thresholds were recorded to measure changes in paw-withdrawal thresholds between the two groups prior to MOR agonist assessment. The criterion to initiate MOR-agonist testing was set as a significant difference between the vehicle-treated and PTX-treated groups that persisted for two consecutive days. This criterion was met on Day 13 (Figure 3B).

On Days 14 to 18, all subjects began test sessions with a saline injection followed 20 min later by six paw withdrawal threshold (three measurements per hind paw) recordings. Following the vehicle test, subjects received increasing doses of fentanyl (0.0018-0.01 mg/kg), oxycodone (0.1-0.56 mg/kg), SR14968 (0.0056-0.032 mg/kg), SR17018 (0.18-1.0 mg/kg), or vehicle. All doses of a particular drug were tested within a single session using a cumulative-dosing approach that rendered a dose range in quarter-log intervals that encompassed a full range of effect (i.e., no change from baseline up to a threshold greater than the saline-treated group baseline). Tests of nociception occurred within 20 min after each injection and were followed immediately after by the next injection in the test series. Fentanyl, oxycodone, SR14968, SR17018, and vehicle were tested in all subjects on different days in a counterbalanced order, and all tests occurred daily to ensure that all tests were completed prior to the loss of neuropathy. Neuropathy was verified daily by comparing baseline paw withdrawal thresholds to the initial baselines following PTX administration (Day 13) to confirm that each test was conducted under similar baseline conditions (data not shown). An experimenter blinded to the test conditions administered the injections and scored and recorded paw withdrawal thresholds during the session.

To verify that the effects of each MOR agonist were mediated by an opioid receptor, on Days 19 to 22 each subject completed a modified version of the procedure described above. First, only the subjects from the PTX-treated group were used during this procedure. Second, each session consisted of only three tests of nociception. Vehicle was administered and paw withdrawal thresholds were recorded 20 min later. Then, each subject received a single dose of a MOR agonist condition, and paw withdrawal thresholds were recorded 20 min later. Subjects received fentanyl (0.0056 mg/kg), oxycodone (0.32 mg/kg), SR14968 (0.018 mg/kg), and SR17018 (0.56 mg/kg) counterbalanced across four separate days. The doses for each MOR agonist were selected because they produced clear mechanical antinociception (i.e., paw withdrawal thresholds that were greater than the saline-treated group baseline) and produced comparable levels of effect. Finally, following the test of nociception for each MOR agonist, subjects received an injection of naltrexone (0.032 mg/kg; i.v.), and paw withdrawal thresholds were recorded 20 minutes later.

Inflammatory Antinociception

Twenty subjects, 10 males and 10 females, were implanted with i.v. catheters, and all test drugs were administered i.v. Figure 4A illustrates the experimental timeline for the study. On Day 0 of the study, acclimation to the apparatus and the injection-test series were identical to the neuropathic antinociception procedure described above. Following the acclimation testing, subjects were counterbalanced into two separate groups based on their baseline paw thresholds: vehicle-treated or complete Freund’s adjuvant (CFA)-treated.

Figure 4.

Panel A illustrates the experiential time for the inflammatory antinociception study. Panel B illustrates the change from baseline paw withdrawal threshold (in grams) for the untreated (circle) and CFA-treated (upward triangle) hindpaws across days for the CFA-treated group following complete Freund’s adjuvant (CFA) administration. Panel C illustrates the paw withdrawal threshold (in grams) for fentanyl (0.0018-0.01 mg/kg; downward triangle), SR14968 (0.0056-0.032 mg/kg; square), oxycodone (0.1-0.56 mg/kg; circle), and SR17018 (0.18-1.0 mg/kg; upward triangle). The lower shaded grey bar represents the paw withdrawal threshold under vehicle conditions for the CFA-treated group. The upper shaded grey bar represents the paw withdrawal threshold under vehicle conditions for the vehicle-treated group. Panels D through G illustrate the paw withdrawal threshold (in grams) for fentanyl (0.0056 mg/kg), SR14968 (0.01 mg/kg), oxycodone (0.32 mg/kg), and SR17018 (0.56 mg/kg) before and after administration of the opioid antagonist naltrexone (0.032 mg/kg; i.v.). Filled bars represents a significant difference from vehicle (saline) administration. Panel H illustrates the paw withdrawal threshold (in grams) for fentanyl (0.0056 mg/kg; downward triangle), SR14968 (0.01 mg/kg; square), oxycodone (0.032 mg/kg; circle), and SR17018 (0.56 mg/kg; upward triangle) as a function of time (in minutes). Error bars indicate ± SEM.

On Day 1, all subjects were injected with either vehicle (saline; 0.1 mL) or CFA (50% in saline, with 5 mg/mL heat-killed Mycobacterium tuberculosis, 0.1 mL; Sufka 1994) into the plantar surface of the right hindpaw. No baseline recordings occurred on Day 1. Starting on Day 2, baseline paw withdrawal thresholds following saline or CFA treatment were recorded to determine shifts in thresholds between the two groups prior to MOR agonist testing. Criterion to initiate MOR-agonist testing was set as CFA-treated group responding being less than half of the vehicle-treated responding for two consecutive days. This criterion was met on Day 3 (Figure 4B).

On Days 4 to 8, all subjects began test sessions with a vehicle injection followed 20 min later by six paw withdrawal threshold recordings (i.e., three measurements per hind paw). The cumulative dosing procedure was identical to the neuropathic antinociception study (described above). The doses for each MOR agonist were: fentanyl (0.0018-0.01 mg/kg), oxycodone (0.1-0.56 mg/kg), SR14968 (0.0032-0.018 mg/kg), SR17018 (0.18-1.0 mg/kg) or vehicle. All subjects received each drug condition in a counterbalanced order, and all tests occurred daily to ensure they were completed prior to the loss of an inflammatory response. An experimenter blinded to the test conditions administered the injections and scored and recorded paw withdrawal thresholds during the session.

To verify that the effects of each MOR agonist were mediated by an opioid receptor, on Days 9 to 12 each subject completed a modified version of the procedure described above. Only the subjects from the CFA-treated group were used during this procedure and each session consisted of only three tests of nociception under the identical procedure described in the neuropathic antinociception study. The dose of each MOR agonist used was: fentanyl (0.0056 mg/kg), oxycodone (0.32 mg/kg), SR14968 (0.018 mg/kg), and SR17018 (0.56 mg/kg). Each drug condition was counterbalanced across four separate days.

The inflammatory response was still present after the dose-response determinations and the antagonism study, enabling an assessment of the duration of action of each MOR agonist. On Days 13 to 16, all CFA-treated subjects received an injection of vehicle as previously described, and paw withdrawal thresholds were recorded 20 minutes later. Following the saline measurement, the same doses for each of the four drug conditions used within the antagonism portion of the study were administered, and paw withdrawal thresholds were recorded at specific time points. The selected time points were: 0, 10, 20, 40, 80, 160, and 320 minutes following MOR agonist administration. Subjects were tested with fentanyl (0.0056 mg/kg), oxycodone (0.32 mg/kg), SR14968 (0.018 mg/kg), and SR17018 (0.56 mg/kg), counterbalanced across four separate days, and the tests were conducted up to a time in which the response was comparable to the vehicle test point.

Respiratory Depression

Eight subjects, 4 males and 4 females, were implanted with i.v. catheters as described above. Whole-body plethysmography was used to evaluate the pulmonary function of the subjects. The principal measures of pulmonary function were ventilation frequency (breaths per minute), tidal volume (milliliters per breath), and minute volume (milliliters per minute). Test sessions occurred every 2 days and were 90 min in duration. During a test session, respiratory performance measurements occurred in 5-sec bins. Saline was administered at the start of the session and fentanyl (0.01-0.32 mg/kg), oxycodone (0.1-3.2 mg/kg), SR14968 (0.032-0.32), and SR17018 (0.32-10 mg/kg) were administered under the same cumulative-dosing procedure described above. The doses for each MOR agonist were selected based on their ability to reduce pulmonary parameters to approximately 50% of the saline baseline. In the case of both SR14968 and SR17018, the dose was increased in preliminary tests up to a point at which ictal effects occurred in some subjects that were not blocked with naltrexone treatment (i.e., an apparent off-target effect). Because of this, the doses for SR14968 and SR17018 were constrained to the ranges indicated above. All drug conditions were counterbalanced across subjects.

Data Analysis

Self-administration

The average number of injections earned per session for each drug condition was analyzed using five separate one-way repeated measures analyses of variance (ANOVA) with the within-subject factor of Dose. Specifically, all doses of fentanyl (0.0001-0.00056 mg/kg/injection), oxycodone (0.01-0.1 mg/kg/injection), SR14968 (0.001-0.0056 mg/kg/injection), and SR17018 (0.1-0.56 mg/kg/injection) were compared to the within-drug saline condition. Bonferroni’s multiple-comparison tests were used to compare mean number of injections for each dose within a drug condition to saline alone. Additionally, the highest response requirement completed for each subject in each drug condition, irrespective of dose, was determined (termed BP_max), and these values were averaged across subjects within a drug condition and compared to the BP of saline alone within a drug condition using a Wilcoxon Rank sum Test. The BP_max measure was used to provide an index of reinforcing strength that takes into account individual differences in peak BP values within a drug condition, irrespective of dose. Finally, due to non-homogeneity of variance between the four drug conditions, the BP_max for all four drug conditions were compared using a Kruskal-Wallis one-way ANOVA with the between-subject factor of Drug Condition (fentanyl, oxycodone, SR14968, SR17018). Dunn’s multiple comparisons were used to compare the BP_max of each drug condition to oxycodone.

The relative potencies of fentanyl, oxycodone, SR14968, and SR17018 in self-administration were then analyzed using non-linear regression fit to the ascending limbs of the dose-response functions for each subject to determine ED₅₀ values (i.e., the dose estimated to produce 50% effect). The top constraint for each MOR agonist was set to the mean number of injections at the dose that elicited the highest injections earned per session (i.e., peak dose on the ascending limb) and the bottom constraint was set to the mean number of saline injections (GraphPad Prism). Individual-subject ED₅₀ values were then averaged to determine the 95% confidence intervals. Aggregated ED₅₀ values were considered to be significantly different if the 95% confidence intervals did not overlap.

Thermal Antinociception

Hot-plate latencies were expressed as % maximum possible effect (%MPE) and were calculated using the following equation:

$$\%\mathrm{MPE}=[(\text{Test Latency}-\text{Saline Latency})∕(60\text{-Saline Latency})]\times 100$$

where the test latency was the latency for the rat to emit a nociceptive response after administration of a dose of a drug, and saline latency was the latency to emit a response following the initial injection (which was vehicle). %MPE for each drug condition was plotted as a function of logarithmic dose. Dose ranges were selected such that the lowest dose corresponded to ≤ 20% effect and the highest dose corresponded to ≥ 80% effect. To compare relative potencies of fentanyl, oxycodone, SR14968, and SR17018 in thermal antinociception, the dose-response data for each drug condition were aggregated across subjects and fit with non-linear regression to determine ED₅₀ values and 95% confidence intervals as described for the self-administration study above. ED₅₀ values were considered to be significantly different if the confidence intervals did not overlap. Finally, full antagonism of effect was defined as reducing antinociceptive magnitude to less than 20% or less of the max achieved for each drug condition.

Neuropathic Antinociception

The average paw withdrawal threshold (in grams) for both the vehicle-treated and PTX-treated groups prior to neuropathy induction was analyzed using a student’s t-test. The average paw withdrawal thresholds (in grams) for both the saline-treated and PTX-treated groups following treatment administration were analyzed using a two-way mixed-design ANOVA with the within-subject factor of Days (days 8 – 13) and with the between-subjects factor of Treatment Group (saline-treated, PTX-treated). Bonferroni’s multiple comparisons were used to compare the average paw withdrawal threshold for the vehicle-treated and PTX-treated groups for each day. To compare relative potencies of fentanyl, oxycodone, SR14968, and SR17018 in neuropathic antinociception, the dose-response data for each drug condition within the PTX-treated group were fit with non-linear regression to determine ED₅₀ values and 95% confidence intervals using the doses that corresponded to less than or equal to the baseline paw withdrawal threshold for the PTX-treated group and the highest dose that corresponded to greater than or equal to the baseline paw withdrawal threshold for the PTX-treated group before PTX treatment. Therefore, the doses used were: fentanyl (0.0018-0.0056 mg/kg), SR14968 (0.0056-0.018 mg/kg), oxycodone (0.1-0.32 mg/kg), and SR17018 (0.18-0.56 mg/kg). ED₅₀ values were considered to be significantly different if the confidence intervals did not overlap. Four separate one-way repeated-measures ANOVA with the within-subject factor of Drug Treatment (vehicle, MOR agonist alone, MOR agonist following 0.032 mg/kg naltrexone) were used to compare the paw withdrawal threshold following naltrexone administration. Bonferroni’s multiple comparisons were used to compare each MOR agonist alone or following naltrexone administration to saline.

Inflammatory Antinociception

The average paw withdrawal threshold (in grams) for both hindpaws for the saline-treated and CFA-treated groups prior to treatment administration were analyzed using a two-way ANOVA with the within-subject factor of Hindpaw (Left, Right), and the between-subject factor of Treatment (CFA, Saline). To verify the induction of inflammatory nociception, the average paw withdrawal threshold (in grams) for both the vehicle-treated and CFA-treated hindpaws (i.e., the right hindpaw) following treatment administration were analyzed using a two-way repeated measures ANOVA with the within-subject factor of Days (Days 2 and 3), and the between-subject factor of Treatment Group (saline-treated, CFA-treated). Bonferroni’s multiple comparisons were used to compare the average paw withdrawal threshold for the vehicle-treated and CFA-treated hindpaws for each day. To measure the stability of CFA-induced decreases in paw withdrawal thresholds, the left (untreated) and right (CFA-treated) hindpaw thresholds of the CFA-treated group only were analyzed using a two-way repeated measures ANOVA with the within-subject factors of Days (2-15) and Inflammation (Left, Right). Bonferroni’s multiple comparisons were used to compare Days 3-15 to Day 1 post-CFA treatment to determine if there were drifts in thresholds. To compare relative potencies of fentanyl, oxycodone, SR14968, and SR17018 in inflammatory antinociception, the dose-response data for each drug condition within the CFA-treated group were fit with non-linear regression to determine ED₅₀ values and 95% confidence intervals using the doses that corresponded to less than or equal to the baseline paw withdrawal threshold for the CFA-treated hindpaw and the highest dose that corresponded to greater than or equal to the baseline paw withdrawal threshold for the CFA-treated hindpaw prior to CFA treatment. Therefore, the doses used were: fentanyl (0.0018-0.0056 mg/kg), SR14968 (0.0032-0.01 mg/kg), oxycodone (0.1-0.32 mg/kg), and SR17018 (0.18-0.56 mg/kg). ED₅₀ values were considered to be significantly different if the confidence intervals did not overlap. Four separate one-way repeated-measures ANOVA with the within-subject factor of Drug Treatment (vehicle, MOR agonist alone, MOR agonist following 0.032 mg/kg naltrexone) were used to compare the paw withdrawal threshold following naltrexone administration. Bonferroni’s multiple comparisons were used to compare each MOR agonist alone or following naltrexone administration to saline.

Finally, the area under the curve (AUC) values for each individual subject were averaged across each drug condition in the time-course study. AUC values were analyzed using a one-way repeated measures ANOVA with the within subject factor of Drug Condition (fentanyl, oxycodone, SR14968, and SR17018). Bonferroni’s multiple comparisons were used to compare drug conditions to one another.

Respiratory Depression

For each drug dose, all 5-second bins from the final 10 minutes of each 20-minute injection interval were averaged for breath frequency, tidal volume, and minute volume. All variables (breath frequency, tidal volume, and minute volume) for each drug condition (fentanyl, oxycodone, SR14968, SR17018) were analyzed with separate one-way repeated measures ANOVAs with the within-subject factor of Dose. Bonferroni’s multiple comparisons were used to compare each dose to the within-session baseline (i.e., vehicle injection).

Quantification of Relative Potencies.

A major aim of the current study was to quantify potency windows for each MOR agonist between the different measures to determine if the G-protein biased compounds exhibited greater potency to produce antinociception relative to their potencies to produce the “side effects” modeled in the report (i.e., reinforcing effects and respiratory depression). ED₅₀ values for oxycodone, fentanyl, SR14968, and SR17018 from self-administration and from the antinociception studies (thermal, neuropathic, and inflammatory) were used to quantify the relative potency ratios for each MOR agonist according to the following formula:

$$=E{D}_{50}\text{Self-administration}∕E{D}_{50}\text{Antinociception}$$

Notably, ED₅₀ values could not be determined for all MOR agonists in respiratory depression because only fentanyl and oxycodone reduced respiratory performance at the doses tested. Therefore, an alternative measure was used to compare the potencies of the compounds in antinociception to their respiratory effects. Specifically, the lowest dose that produced a significant reduction in minute volume for fentanyl (0.032 mg/kg) and oxycodone (1.0 mg/kg) was used as the quantity for a measure of effect, and for the MOR agonists that produced no change in minute volume, the highest dose tested of SR14968 (0.32 mg/kg) and SR17018 (10.0 mg/kg) was used to quantify the limit of the “window” that could be determined by the current data. The calculations relating this respiration quantity to each drug’s antinociceptive potency were made according to the formula:

$$=\text{Dose Minute Volume}∕E{D}_{50}\text{Antinociception}$$

These calculations were conducted for each of the three antinociception measures described above. Higher relative potency values (i.e., greater than 1) indicate a trajectory of increased potency for a MOR agonist to produce antinociceptive effects relative to the adverse effects (i.e., abuse liability or respiratory depression), and signify a more favorable therapeutic profile.

Results

Self-administration

The current study did not detect sex differences in drug efficacy or potency within any measure (Supplementary Table 1). Therefore, all males and females were averaged and analyzed together across drug conditions. Figure 1A illustrates the average number of injections per session (± SEM) for fentanyl, oxycodone, SR14968, and SR17018. A repeated-measures ANOVA revealed a main effect of Dose for fentanyl (F [1.709, 11.96] = 15.15; p = 0.0007), oxycodone (F [1.915, 11.87] = 21.03; p = 0.0001), SR14968 (F [2.753, 19.27] = 7.140; p = 0.0025), and SR17018 (F [2.503, 17.52] = 10.59; p = 0.0005). Specifically, Bonferroni’s multiple-comparisons revealed that 0.00018-0.00056 mg/kg/injection doses for fentanyl, 0.032-0.1 mg/kg/injection doses for oxycodone, and 0.18 and 0.32 mg/kg/injection doses for SR17018 were all self-administered at levels significantly greater than saline (all p’s < 0.05). Figure 1B illustrates the BP_max for fentanyl, oxycodone, SR14968, and SR17018. Wilcoxon Rank sum Test revealed a significant increase in the median BP_max values for fentanyl (median = 52.3, interquartile range = 41.8-67.3), oxycodone (median = 54.0, interquartile range = 41.3-117.6), SR14968 (median = 32.7, interquartile range = 27.3-39.1), and SR17018 (median = 33.0, interquartile range = 30.4-36.7) when compared to the within-drug saline condition. Additionally, a Kruskal-Wallis one-way ANOVA revealed a significant main effect of Drug Condition (p = 0.0001) when the BP_max for all four MOR agonists were compared. Specifically, the BP_max for oxycodone was significantly greater than SR14968 and SR17018.

Figure 1.

Panel A illustrates the number of injections ± SEM earned per session under a progressive-ratio schedule of reinforcement for i.v. saline (lower shaded grey bar), fentanyl (0.0001-0.00056 mg/kg/inj; downward triangle), SR14968 (0.001-0.0056 mg/kg/inj; square), oxycodone (0.01-0.1 mg/kg/inj; circle), and SR17018 (0.1-0.56 mg/kg/inj; upward triangle). * indicates significant differences from saline alone. Panel B illustrates the maximum breakpoint irrespective of dose (BP_max) ± SEM under a progressive-ratio schedule of reinforcement for i.v. fentanyl, SR14968, oxycodone, and SR17018 along with the respective saline BP_max. White bars represent the saline BP_max within each drug condition. Black bars represent the BP_max for each MOR agonist. * indicate differences from saline. # indicates significant differences from the oxycodone BP_max. Error bars indicate ± SEM

Thermal Antinociception

Figure 2A illustrates the thermal antinociceptive effects (± SEM) of fentanyl, oxycodone, SR14968, and SR17018 in the hot-plate test. %MPE values are plotted as a function of dose for each drug. Fentanyl, oxycodone, SR14968, and SR17018 produced full antinociception (i.e., 100% MPE) at the highest dose tested for each drug. ED₅₀ values for each drug are listed in Table 1. The confidence intervals indicated significant differences (i.e., the confidence intervals did not overlap) in potency between all drugs to produce antinociception with a potency order of: fentanyl > SR14968 > oxycodone > SR17018. Figure 2B illustrates the thermal antinociceptive effects for fentanyl, oxycodone, SR14968, and SR17018 following administration of the opioid antagonist naltrexone. The thermal antinociceptive effects of all four MOR agonists were blocked by the administration of naltrexone (i.e., all data points were comparable to saline).

Figure 2.

Panel A illustrates the percent of maximum possible effect (%MPE) for thermal antinociception (Hot Plate) as a function of dose for fentanyl (0.0018-0.01 mg/kg), SR14968 (0.1-0.56 mg/kg), oxycodone (1.0-5.6 mg/kg), and SR17018 (1.8-10.0 mg/kg). Panel B illustrates %MPE for thermal antinociception as a function of dose for fentanyl (0.0032-0.01 mg/kg), SR14968 (0.18-0.56 mg/kg), oxycodone (1.8-.6 mg/kg), and SR17018 (3.2-10.0 mg/kg) following pretreatment with the opioid receptor antagonist naltrexone (0.032 mg/kg; i.v.). Error bars indicate ± SEM.

Table 1.

Potency of fentanyl, oxycodone, SR14968, and SR17018 in self-administration, thermal, neuropathic, and inflammatory antinociception, and respiratory depression. ED50 values for the dose-response data for all four MOR agonists are expressed in terms of mg/kg body weight

	Self- Administration	Thermal Antinociception	Neuropathic Antinociception		Inflammatory Antinociception		Respiratory Depression*
			PTX	Vehicle	CFA	Vehicle
Fentanyl (95% C.I.)	0.00035 (0.00091-0.001)	0.0073 (0.004-0.01)	0.0020 (0.001-0.003)	0.0026 (0.002-0.003)	0.0060 (0.004-0.008)	0.0018 (0.0014-0.0021)	0.81 (0.64-0.98)
Oxycodone (95% C.I.)	0.028 (0.02-0.036)	2.6 (2.1-3.1)	0.13 (0.094-0.157)	0.11 (0.048-0.17)	0.42 (0.026-0.58)	0.12 (0.059-0.18)	0.042 (0.023-0.060)
SR14968 (95% C.I.)	0.0017 (0.0013-0.0021)	0.33 (0.23-0.44)	0.0099 (0.006-0.014)	0.0065 (0.005-0.008)	0.0073 (0.005-0.009)	0.0032 (0.0021-0.0043)	n.d.
SR17018 (95% C.I.)	0.13 (0.11-0.16)	4.9 (3.8-6.0)	0.26 (0.12-0.40)	0.16 (0.093-0.23)	0.73 (0.45-1.0)	0.28 (0.20-0.37)	n.d.

C.I. = Confidence intervals. N.D.=Not determined.

^*indicates that ED₅₀ values are based only on minute volume in respiratory depression. Italics indicate a significant difference from the vehicle-control group

Neuropathic Antinociception

Figure 3B illustrates the average paw withdrawal threshold (in grams; ± SEM) for the vehicle-treated and PTX-treated subjects before and after the administration of PTX as a function of day. A student’s t-test revealed no significant difference between the treatment groups prior to PTX administration (p = 0.8131). A two-way mixed-design ANOVA revealed a main effect of Treatment Group (F [1, 18] = 8.547; p = 0.0091) and a Day x Treatment Group interaction (F [5, 90] = 6.972; p = 0.0001), but no effect of Day (F [5, 90] = 0.5538; p = 0.7351) for the groups following administration of PTX (Days 8-13). Bonferroni’s multiple comparisons revealed a significant decrease in paw withdrawal threshold in the PTX-treatment group when compared to the vehicle-treatment group on days 12 and 13.

Figure 3C illustrates the average paw withdrawal threshold (in grams) for rats when treated with varying doses of fentanyl, oxycodone, SR14968, and SR17018 (Days 14-18; Table 2). The grey shaded bars represent the paw withdrawal threshold following vehicle administration for the vehicle-treated (upper shaded) and PTX-treated (lower shaded) subjects. ED₅₀ values for each drug in the PTX-treated group are listed in Table 1. Figure 3D, E, F, and G illustrate the paw withdrawal thresholds (in grams) for fentanyl, oxycodone, SR14968, and SR17018 following naltrexone administration (Days 19-22). Four separate one-way repeated measures ANOVAs indicated a main effect of Drug Treatment for fentanyl (F [2, 27] = 118.0; p = 0.0001), oxycodone (F [1.156, 10.40] = 94.04; p = 0.0001), SR14958 (F [2, 27] = 127.0; p = 0.0001), and SR17018 (F [2, 27] = 116.7; p = 0.0001). Specifically, the paw withdrawal threshold following administration of each MOR agonist was significantly greater relative to the saline control, but not following administration of naltrexone (0.032 mg/kg).

Table 2.

Statistical outcomes from two-way repeated-measures ANOVAs in neuropathic and inflammatory antinociception.

	Fentanyl	Oxycodone	SR14968	SR17018
Neuropathic Antinociception
Dose	F [3.188, 57.39] = 100.8; p = 0.0001	F [2.358, 42.44] = 139.6; p = 0.0001	F [2.305, 41.48] = 110.0; p = 0.0001	F [1.847, 33.24] = 108.5; p = 0.0001
Treatment Group	F [1, 18] = 3.591; p = 0.0743	F [1, 18] = 15.35; p = 0.001	F [1, 18] = 21.01; p = 0.0001	F [1, 18] = 11.49; p = 0.0033
Dose X Treatment Group	F [4, 72] = 5.439; p = 0.0007	F [4, 72] = 4.206; p = 0.0041	F [4, 72] = 3.343; p = 0.0144	F [4, 72] = 3.301; p = 0.0153
Inflammatory Antinociception
Dose	F [3.188, 57.39] = 100.8; p = 0.0001	F [4, 68] = 127.6; p = 0.0001	F [2.729, 46.39] = 132.4; p = 0.0001	(F [3.269, 55.57] = 140.4; p = 0.0001
Treatment Group	F [1, 17] = 120.7; p = 0.0001	F [1, 17] = 70.47; p = 0.0001	F [1, 17] = 98.41; p = 0.0001	F [1, 17] = 56.25; p = 0.0001
Dose X Treatment Group	F [4, 72] = 5.439; p = 0.0007	F [4, 68] = 8.680; p = 0.0001	F [4, 68] = 8.695; p = 0.0001	F [4, 68] = 9.641; p = 0.0001

F values in italics met statistical significance at p < 0.05

Inflammatory Antinociception

Before administration of CFA, a two-way mixed-design ANOVA revealed no main effects for Hindpaw (F [1, 18] = 0.1051; p 0.7695), Treatment Group (F [1, 18] = 0.01561; p = 0.9019), or a Hindpaw x Treatment Group interaction (F [1, 18] = 1.577; p = 0.2253). Following administration of CFA, a two-way repeated measures ANOVA revealed a main effect of Treatment Group (F [1, 18] = 242.0; p = 0.0001), but not a main effect of Days (F [1, 18] = 0.07538; p = 7868) or a Days x Treatment Group interaction (F [1, 18] = 0.002679; p = 0.9593). Specifically, there was a significant difference between treatment groups on Days 1 and 2 after CFA treatment (data not shown). Figure 4B illustrates the change from baseline paw withdrawal threshold as a function of days for the CFA-treated group only. A two-way repeated measures ANOVA revealed a significant main effect for Inflammation (F [1.000, 9.000] = 715.1; p = 0.0001), but not a main effect of Days (F [5.037, 45.33] = 1.927; p = 0.1079), or an Inflammation X Days interaction (F [4.118, 34.42] = 1.758; p = 0.1583), indicating that nociceptive responding was stable over days (Days 2-15). Specifically, there was a significant difference in paw withdrawal threshold between the untreated and CFA-treated paws throughout the entirety of testing, indicating that all conditions occurred under a similar baseline.

Figure 4C illustrates the average paw withdrawal threshold (in grams) for fentanyl, oxycodone, SR14968, and SR17018 for each treatment group as a function of dose (Days 4-8). The grey shaded bars represent the paw withdrawal threshold baseline for the CFA-treated paw prior to CFA administration (upper shaded) and after CFA administration (lower shaded). ED₅₀ values for each drug in the CFA-treated group are listed in Table 1. Figure 4D, E, F, and G illustrate the average paw withdrawal thresholds for fentanyl, oxycodone, SR14968, and SR17018 following naltrexone administration (Days 9-12; Table 2). Four separate one-way repeated measures ANOVAs indicated a main effect of Drug Treatment for fentanyl (F [1.362, 10.90] = 240.1; p = 0.0001), oxycodone (F [1.131, 9.052] = 142.5; p = 0.0001), SR14958 (F [1.940, 15.52] = 402.8; p = 0.0001), and SR17018 (F [1.296, 10.37] = 164.7; p = 0.0001) following naltrexone administration. Specifically, for all four MOR agonists, the administration of naltrexone (0.032 mg/kg) significantly attenuated the increases in paw withdrawal threshold produced by each MOR agonist to levels no different than vehicle. Figure 4H illustrates the average paw withdrawal threshold for a single dose of fentanyl, oxycodone, SR14968, and SR17018 as a function of time (Days 13-16). A one-way repeated-measures ANOVA revealed no main effect of Drug Condition (F [2.225, 17.80] = 0.9844; p = 0.4012), indicating that all drugs were comparable in terms of duration of effect.

Respiratory Depression

Figure 5A, B, and C illustrates the average frequency, tidal volume, and minute volume for fentanyl, oxycodone, SR14968, and SR17018 as a function of dose. Three separate one-way repeated-measures ANOVAs revealed a main effect of Dose for frequency (F [1.418, 9.928] = 8.083; p = 0.0122), and minute volume (F [2.535, 17.74] = 7173; p = 0.0033), but not for tidal volume (F [2.170, 15.19] = 1.966; p = 0.1719) for fentanyl. Three separate one-way repeated-measures ANOVAs revealed a main effect of Dose for frequency (F [2.105, 14.74] = 23.47; p = 0.0001), tidal volume (F [1.914, 13.39] = 6.149; p = 0.0135), and minute volume (F [2.064, 14.45] = 11.51; p = 0.0010) for oxycodone. Three separate one-way repeated-measures ANOVA revealed no main effect of Dose for frequency (F [2.459, 17.21] = 1.140; p = 0.3527), tidal volume (F [2.568, 17.98] = 3.291; p = 0.0504), or minute volume (F [2.480, 17.36] = 1.253; p = 0.3168) for SR14968. Finally, three separate one-way repeated-measures ANOVA revealed no main effect of Dose for frequency, tidal volume, or minute volume for SR17018 (p’s > 0.05).

Figure 5.

Panels A through C represent breathing frequency (A), tidal volume (B), and minute volume (C) for fentanyl (0.01-0.032 mg/kg; downward triangle), SR14968 (0.032-0.32 mg/kg; square), oxycodone (0.1-3.2 mg/kg; circle), and SR17018 (0.32-10.0 mg/kg; upward triangle). The grey shaded bar represents the baseline under the vehicle condition. * indicates a significant different from the within-session baseline (vehicle injection). Error bars indicate ± SEM.

Quantification of Relative Potencies

Table 1 shows the ED₅₀ values from self-administration, thermal, neuropathic, and inflammatory antinociception, and respiratory depression. Figure 6 shows the relative potency ratios for oxycodone, fentanyl, SR14968, and SR17018 for self-administration and respiratory depression versus thermal, neuropathic, or inflammatory antinociception. In thermal antinociception, the potency ratio order was: SR14968 < Oxycodone < SR17018 < Fentanyl, where fentanyl displayed the largest separation in potency values. In neuropathic antinociception, the potency ratio order was: SR14968 < Fentanyl < Oxycodone < SR17018. Finally, in inflammatory antinociception, the potency ratio order was: Fentanyl < Oxycodone < SR17018 < SR14968.

Figure 6.

Panels A through C illustrate the relative potency ratio for thermal (A), neuropathic (B), and inflammatory (C) antinociception when compared to self-administration. Panels D through F illustrate the relative potency value for thermal (D), neuropathic (E), and inflammatory (F) antinociception when compared to the lowest effective respiratory-depressant dose for minute volume. Values greater than 1 indicate a trajectory of decreasing proximal potency for a MOR agonist to produce an adverse effect (i.e., abuse liability or respiratory depression) relative to the antinociceptive effects (i.e., the doses are decreasingly similar to produce the behavioral effects).

Discussion

The novel findings from the current study were that the G-protein biased MOR agonists, SR14968 and SR17018, produced reinforcing effects in self-administration, but had reduced reinforcing effectiveness compared to oxycodone. Additionally, SR14968 and SR17018 produced antinociceptive effects in a dose-dependent manner across all three pain modalities that were comparable in efficacy to oxycodone and fentanyl. The opioid antagonist, naltrexone, blocked the antinociceptive effects of each drug across all three pain modalities, indicating that all compounds produced antinociception through an opioid-dependent mechanism. Furthermore, unlike oxycodone and fentanyl, SR14968 and SR17018 did not produce respiratory depression.

One of the main goals of the study was to determine if MOR agonists reported to be G-protein biased functioned as reinforcers in a manner comparable to prototypical MOR agonists. In the present study, both SR14968 and SR17018 functioned as reinforcers under a PR schedule of reinforcement, which is in agreement with previous reports demonstrating that G-protein biased MOR agonists from other structural classes produce abuse-related effects in rodents (Zamarripa et al. 2018; Kudla et al. 2021) and nonhuman primates (Ding et al. 2020). However, when the breakpoints for fentanyl, SR14968, and SR17018 were compared to oxycodone, SR14968 and SR17018, but not fentanyl, were significantly lower. This contrasts with other findings, including those from our group, demonstrating that other novel G-protein biased MOR agonists (e.g., TRV130 and PZM21) produce reinforcing effects comparable to traditional MOR agonists (Zamarripa et al., 2018; Ding et al., 2020). However, G-protein biased agonists like TRV130 and PZM21 are considered to have low-to-moderate preference for G-protein signaling pathways (Stahl and Bohn, 2022). These results suggest that while functioning as a biased agonist does not inherently prevent a drug from acting as a reinforcer, agonists with greater G-protein signaling relative to recruitment of β-arrestin-2 (e.g., SR14968 and SR17018) may have a lower abuse liability due to their higher G-protein selectivity. Collectively, these findings suggest that molecules developed from the chemical class of SR14968 and SR17018 could have reduced (but still measurable) misuse potential as therapeutics.

While biased signaling is one possible explanation for reduced relative reinforcing efficacy, another explanation is lower intrinsic efficacy at the MOR. Recent studies indicate that the signaling profiles of G-protein biased MOR agonists more closely resemble those of partial agonists (e.g., buprenorphine) when G-protein signaling is measured in an alternate manner (Gillis et al. 2020; Singleton et al. 2021). As stated above, we and others have reported that certain G-protein biased MOR agonists produce reinforcing effects in self-administration under a PR schedule of reinforcement that are equal in efficacy to oxycodone (Ding et al. 2020; Zamarripa et al. 2018), suggesting that some G-protein biased MOR agonists produce abuse-related effects above what would be expected for partial MOR agonists. As such, factors other than signaling bias or intrinsic efficacy that are particular for the SR structural class may underlie the relatively low reinforcing efficacy of these compounds. More studies using drug self-administration and thorough dose-response assessments are needed to provide standardized comparisons between G-protein biased agonists from different structural classes to distinguish the relative roles of signaling bias, intrinsic efficacy, and potentially undiscovered actions that may be particular to a structural class (e.g., the SR series of molecules).

In the three pain modalities tested, all MOR agonists produced equi-efficacious antinociception. These findings are in agreement with previous reports demonstrating that MOR agonists produce antinociception in thermal (Cornelissen et al. 2021; Grim et al. 2020; Kudla et al. 2021; Schwienteck et al. 2019; Townsend et al. 2017), neuropathic (Pantouli et al. 2021; Pascual et al. 2014), and inflammatory (Armendariz and Nazarian 2018; Wang et al. 2006) models of nociception. All antinociceptive effects were blocked by naltrexone, indicating that the effects of each compound were mediated by an opioid receptor. SR14968 and SR17018 both produced full antinociception across the three pain modalities comparable to oxycodone and fentanyl. These findings are in agreement with past findings indicating that G-protein biased MOR agonists produce antinociception comparable to traditional MOR agonists in models of neuropathy and inflammation and that G-protein signaling bias does not reduce MOR-agonist efficacy in these measures (Pantouli et al. 2021). Further, both SR14968 and SR17018 had greater potency to produce neuropathic and inflammatory antinociception relative to thermal antinociception, which is consistent with past findings (Pantouli et al. 202 1). Notably, when comparing the single-dose determinations of all four MOR agonists in the naltrexone tests to the corresponding effects at the same doses in the cumulative series in neuropathic and inflammatory models, the single-dose injections resulted in higher paw withdrawal thresholds compared to the same doses in the cumulative series. This is likely due to drug elimination occurring during the cumulative-dosing period. Regardless, these results indicate that the G-protein biased agonists, SR14968 and SR17018, are comparable in effect to prototypical MOR agonists across different pain modalities, and that both compounds are more potent to produce these effects in clinically-translational models (i.e., neuropathy and inflammation) relative to the “side effects” (abuse potential and respiratory toxicity than oxycodone and fentanyl.

Both fentanyl and oxycodone depressed pulmonary ventilation measures in a dose-dependent manner. In agreement with previous reports (Schmid et al. 2017; Cornelissen et al. 2021), neither SR14968 nor SR17018 produced respiratory depression at any of the doses tested, which encompassed a full antinociceptive range for each compound. Taken together, these results suggest that G-protein biased MOR agonists like SR14968 and SR17018 would exhibit greater potency separation between therapeutic and respiratory-depressant effects relative to traditional MOR agonists like fentanyl and oxycodone. However, due to off-target, non-opioid mediated ictal effects observed in some subjects during preliminary testing, escalating the dose to identify respiratory effects was not possible. Therefore, the only firm conclusion that can be made is that both SR14968 and SR17018 are relatively safe in terms of respiratory effects up to the doses required to produce full thermal antinociception and over an order of magnitude higher than the doses required to produce full antinociception against neuropathic and inflammatory pain.

An overarching goal of this study was to quantitatively relate the reinforcing, antinociceptive, and respiratory depressant effects of MOR agonists that varied in terms of reported G-protein bias to determine if these compounds exhibited improved therapeutic selectivity (i.e., produced antinociception at doses that were relatively low compared to those required to produce abuse-related and respiratory effects). Within-drug ratios were calculated for potency by comparing values from self-administration and respiratory depression to each pain modality. Relative to fentanyl and oxycodone, both SR14968 and SR17018 generally exhibited improved potency ratios for antinociception versus self-administration and respiratory depression (the “side effects”), suggesting that G-protein biased MOR agonists produce more favorable therapeutic selectivity when compared to the clinically-used opioid analgesics. However, there was an exception to this finding. Specifically, fentanyl had the largest relative potency ratio when relating the “side effects” to thermal antinociception, suggesting that under these conditions, fentanyl would have the safest therapeutic index of the four tested MOR agonists. However, fentanyl has been reported to have a similar therapeutic index when compared to morphine in rats using thermal antinociception assays (Schmid et al. 2017; Gillis et al. 2020) and postoperative analgesia in humans (Ding et al. 2016), and a narrower therapeutic window when compared to reported G-protein biased MOR agonists (Schwienteck et al. 2019). However, in the current report, fentanyl did not produce larger potency windows relating the “side effects” to neuropathic and inflammatory antinociception. These differences may be related to A) differences in responsivity to the drugs between evoked (i.e., thermal) and static (i.e., neuropathic and inflammatory) nociception, and/or B) differences in responsivity to the drugs when using high dose ranges (required for thermal pain) and lower dose ranges (required for neuropathic and inflammatory nociception). Regardless, it is notable that the SR compounds produced superior therapeutic windows in the more translational models of nociception (neuropathy and inflammation), suggesting that further work with compounds from this chemical series may lead to safer analgesics with efficacy against clinically-relevant pain types.

In conclusion, the results of this study suggest that SR14968 and SR17018 confer higher therapeutic selectivity relative to fentanyl and oxycodone. This increase in therapeutic selectivity may be due to G-protein signaling bias, differences in intrinsic efficacy, or yet-to-be-discovered mechanisms. In this study, the G-protein biased agonists produced effects similar to full MOR agonists in measures of antinociception while producing effects more consistent with partial agonists in other measures (self-administration, respiratory depression). This outcome is consistent with claims that biased MOR agonists produce higher efficacy in antinociception vs. other effects such as respiratory depression and reinforcement. However, as discussed above, we and others have reported that biased agonists from other structural classes function as reinforcers at a level comparable to full MOR agonists. When considering all of the information available thus far, it is reasonable to conclude that SR14968 and SR17018 can produce robust antinociception with reduced but apparent abuse potential, but that these outcomes cannot be decisively linked to signaling bias. More work systematically comparing biased MOR agonists from a variety of structural classes as reinforcers, respiration modulators, and antinociceptive agents will be needed to distinguish between the effects of signaling characteristics at the MOR vs. effects (possibly undiscovered) peculiar to the SR class that may underlie the outcomes observed in the current report.