Negative allosteric modulation of CB₁ cannabinoid receptor signaling suppresses opioid-mediated reward

Abstract

Blockade of cannabinoid type 1 (CB₁)-receptor signaling decreases the rewarding properties of many drugs of abuse and has been proposed as an anti-addiction strategy. However, psychiatric side-effects limit the clinical potential of orthosteric CB₁ antagonists. Negative allosteric modulators (NAMs) represent a novel and indirect approach to attenuate CB₁ signaling by decreasing affinity and/or efficacy of CB₁ ligands. We hypothesized that a CB₁-NAM would block opioid reward while avoiding the unwanted effects of orthosteric CB₁ antagonists. GAT358, a CB₁-NAM, failed to elicit cardinal signs of direct CB₁ activation or inactivation when administered by itself. GAT358 decreased catalepsy and hypothermia but not antinociception produced by the orthosteric CB₁ agonist CP55,940, suggesting that a CB₁-NAM blocked cardinal signs of CB₁ activation. Next, GAT358 was evaluated using in vivo assays of opioid-induced dopamine release and reward in male rodents. In the nucleus accumbens shell, a key component of the mesocorticolimbic reward pathway, morphine increased electrically-evoked dopamine efflux and this effect was blocked by a dose of GAT358 that lacked intrinsic effects on evoked dopamine efflux. Moreover, GAT358 blocked morphine-induced reward in a conditioned place preference (CPP) assay without producing reward or aversion alone. GAT358-induced blockade of morphine CPP was also occluded by GAT229, a CB₁ positive allosteric modulator (CB₁-PAM) and absent in CB₁-knockout mice. Finally, GAT358 also reduced oral oxycodone (but not water) consumption in a two-bottle choice paradigm. Our results support the therapeutic potential of CB₁-NAMs as novel drug candidates aimed at preventing opioid reward and treating opioid abuse while avoiding unwanted side-effects.

Graphical Abstract

1. Introduction

The opioid epidemic has profoundly increased drug overdose deaths globally[1–3]. Opioid overdose death rates have been further exacerbated by the COVID-19 pandemic[4]. While opioid monotherapy is effective for acute pain management[5], its chronic clinical use is limited by adverse side-effects, including addiction, tolerance, physical dependence and, in extreme cases, a fatal overdose[6–9]. Despite recent changes in prescribing guidelines, opioids continue to be used in patients at elevated risk for abuse[10], highlighting the urgent need to identify novel therapeutic strategies to circumvent opioid abuse liability while sparing therapeutic efficacy.

Opioid reward and reinforcement require crosstalk between the endocannabinoid and endogenous opioid systems of the brain[11–15]. Cannabinoid-type 1 receptors (CB₁) and μ-opioid receptors are co-localized in several brain regions implicated in opioid reward[16–18]. Lower levels of opioid seeking-, reinforcement-, dependence-, and relapse-like behaviors accompany the attenuation of CB₁ signaling[19–23]. Unfortunately, clinical use of direct (orthosteric) CB₁ antagonists (inverse agonists) is limited by adverse side-effects, including depression, anxiety, and suicidality[24–27].

The characterization of alternative and distinct allosteric binding sites on CB₁[28, 29] has led to the development of novel CB₁-positive and negative allosteric modulators (PAMs and NAMs, respectively). The binding of a pure NAM to an allosteric site decreases the affinity and/or efficacy of signaling of the orthosteric agonist or endogenous ligand at the orthosteric binding site[28, 30–32]. Crucially, a CB₁-NAM, by reducing endogenous signaling in a spatially and temporally-controlled fashion, may produce a more optimal spectrum of physiological effects compared to direct CB₁ antagonism, resulting in a more desirable pharmacological profile[33–35].

In the present study, we characterized the in vivo pharmacological profile of GAT358, a novel CB₁-NAM. GAT358, is a 3,4-diaminocyclobut-3-ene-1,2-dione derivative which shows minimal CB₁ inverse agonist-related side-effects and displays functional selectivity in the β-arrestin assay[36, 37]. GAT358 was developed by a focused structure-activity relationship study on the classical CB₁-NAM, PSNCBAM-1[36]. In preliminary studies, GAT358 reduced alcohol preference in mice but was not aversive in the taste reactivity or the light dark emergence tests[37]. However, whether a CB₁-NAM influences opioid pharmacology remains unknown.

We characterized the effects of GAT358 on cardinal signs of CB₁ activation induced by the orthosteric CB₁ agonist CP55,940. Next, we assessed the impact of GAT358 on evoked dopamine release in the nucleus accumbens (NAc) shell, a key component of the mesocorticolimbic reward pathway[38, 39], using fast-scan cyclic voltammetry (FSCV). We also asked whether GAT358 would block opioid reward in a conditioned place preference (CPP) assay consistent with a CB₁ allosteric mechanism. Finally, we tested the impact of GAT358 on volitional oral oxycodone consumption in a two-bottle choice drinking in the dark (TBC-DID) paradigm. Our results show that a CB₁-NAM blocks opioid-induced reward without producing unwanted cannabimimetic side-effects or attenuating cannabinoid antinociception.

2. Materials and Methods

2.1. Animals

Adult male Sprague Dawley rats (weighing 200–250g and ~9–12 weeks old) used herein were purchased from Envigo (Indianapolis, IN). Adult male (weighing 25–30g and ~12–18 weeks old) C57BL/6J wildtype (WT) mice purchased from The Jackson Laboratory (Bar Harbor, ME). CD1 wildtype and CB₁-knockout (CB₁KO) mice used herein were bred at Indiana University from heterozygous breeding pairs and congenic for at least 40 crosses[21]. All animals were maintained on a regular 12h light/dark cycle (except for CPP and TBC-DID experiments on a reverse light/dark cycle) and given ad libitum access to food and water (except as noted in the TBC-DID experiments). All procedures were approved by the Indiana University Bloomington Animal Care and Use Committee, followed the International Association for the Study of Pain Guidelines for the Use of Animals in Research[40] and complied with the ARRIVE guidelines 2.0[41].

2.2. Drugs

GAT358 and GAT229 (Fig. 1) were synthesized at Northeastern University and dissolved in a vehicle containing 20% DMSO; it was comprised of dimethyl sulfoxide (DMSO), ethanol, emulphor, and saline in a 5:2:2:16 ratio, respectively. CP55,940 (Sigma Aldrich (St. Louis, MO)) was dissolved in a vehicle containing 3% DMSO for use in cumulative dosing; it was comprised of DMSO, ethanol, emulphor, saline in a 3:5:5:87 ratio, respectively[42]. The vehicle was used to dissolve cannabinoid compounds due to their poor solubility and to assist with drug absorption based on published protocols for in-vivo assays using cannabinoid compounds[43]. Previous work from our laboratory has shown that the 20% DMSO vehicle did not impact the neurochemical and behavioral parameters assessed in the present study[44–47]. Morphine sulfate (NIDA Drug Supply Program (Bethesda, MD) or Sigma Aldrich) was dissolved in saline. Oxycodone hydrochloride (Sigma Aldrich) was dissolved in standard drinking water that was treated using reverse osmosis, deionized, and irradiated with a bactericidal UV lamp. All drug and combination treatments were administered via intraperitoneal (i.p.) injection so that the overall injection volumes did not exceed 5ml/kg in rats and 10ml/kg in mice. All test drug pretreatments were performed 15–20 min prior to the morphine injection and control groups received the same vehicle used to dissolve the drug.

Fig. 1. Chemical structure of CB₁ receptor allosteric modulators.

The figure shows the structures of (A) the CB₁-NAM, GAT358 and (B) the CB₁-PAM, GAT229. Figures based on[36, 106].

2.3. CB₁ tetrad assay

The ability of GAT358 to produce cardinal signs of CB₁ activation (i.e. hypolocomotion, catalepsy, hypothermia, and antinociception) were measured as described previously by our group and others[42, 48–50]. Pre-drug baseline measures were obtained on all tests except the activity meter test to avoid habituation effects. Subsequently, separate groups of mice received a single injection of GAT358 (20mg/kg i.p.), GAT229 (20mg/kg i.p.), rimonabant (10mg/kg i.p.) or vehicle in a time-staggered manner. Starting 15 min post injection, the following tests were run at 5 min intervals in the following order: activity meter, ring test, rectal temperature, and tail-flick antinociception.

Locomotor Activity:

Hypolocomotion was assessed by placing mice in an Omnitech Superflex Node activity meters (Dimensions: 42 × 42 × 30 cm) and the total distance travelled (cm) during a 5 min testing interval was assessed using the Fusion 6.5 software (Omnitech Electronics, Columbus, OH).

Ring test:

Catalepsy was assessed by the amount of time the mouse spent immobile (minus respiratory movements) on an elevated wire ring (6.35 cm diameter wire ring suspended 16 cm above a flat platform) over a 5 min interval.

Hypothermia:

Hypothermia was assessed by the body temperature (°C) measured using a thermometer (Physitemp Instruments, Inc., Clifton, NJ) attached to a rectal probe (Braintree Laboratories, Inc., Braintree, MA).

Antinociception:

The hot water tail-immersion test was performed by submerging the distal 2 cm of the tail of a mouse in a hot water bath (52°C) and the latency to elicit a „flick‟ response was measured as described previously[51, 52]. Prior to injection, three baseline values were obtained with a 10 min inter-trial interval and a maximum cut-off latency of 15 s to avoid tissue damage.

2.4. Cumulative CP55,940 dosing triad assay

The impact of GAT358 on the cannabimimetic effects of the CB₁ agonist CP55,940 was measured using a modified version of the cumulative dosing CB₁-triad paradigm[42] and procedures published previously by our laboratory[48–50]. Following pre-drug baseline measures, separate groups of mice received an acute i.p. injection of GAT358 (1, 10 or 20mg/kg) or vehicle in a time-staggered manner. In a separate study, the impact of GAT358 (20mg/kg), GAT229 (20mg/kg), rimonabant (10mg/kg), or vehicle was evaluated on the cannabimimetic effects of CP55,940. Successive doses of CP55,940 were administered every 40 min to achieve cumulative doses of 0.1, 0.3, 1mg/kg; consequently, actual doses of CP55,940 administered were 0.1, 0.2, 0.7mg/kg at concentrations of 0.01, 0.02, 0.07 mg/ml respectively. The CB₁-triad assessments consisted of performing three tests on each mouse in the following order: ring test for catalepsy, rectal temperature monitoring for hypothermia and a hot-water tail immersion test for antinociception (described above in the CB₁ tetrad assay).

2.5. FSCV

The effect of GAT358 on morphine-induced increase in electrically-evoked dopamine efflux was measured using FSCV at a carbon-fiber microelectrode in the NAc shell under conditions in which dopamine efflux was evoked by a biphasic stimulus application in the medial forebrain bundle (MFB)[47, 53–55]. Rats were anesthetized with urethane (1.6g/kg i.p.) and placed in a stereotaxic frame (David Kopf Instruments, Tujunga, CA) with a heating pad underneath to maintain optimal body temperature. An incision was made to expose the skull surface and multiple craniotomies (~2 mm diameter) were made to allow for the placement of a reference (Ag/AgCl) electrode, a twisted bipolar stimulating electrode, and a carbon-fiber microelectrode. The carbon-fiber microelectrode was fabricated by inserting a single carbon fiber (T650, 6.0 μm diameter, Cytec Engineering Materials, West Paterson, NJ) into a borosilicate capillary tube (1.2 mm o.d.; Sutter Instruments, Novato, CA) pulled to a taper using a vertical micropipette puller (PF-2 Narishige, Tokyo, Japan). The stimulating electrode was positioned in the MFB (−4.6 Anterior-posterior (AP), ±1.3 Medial-lateral (ML), −7.5 Dorsal-ventral (DV)), the carbon-fiber microelectrode was positioned in the NAc shell (+1.2 AP, ±1.4 ML, −6.6 DV), and reference electrode was placed in the contralateral cortex with all co-ordinates referenced to bregma. Final co-ordinates for the carbon-fiber microelectrode and the stimulating electrode were based on the optimization of the electrically-evoked dopamine signal and were not changed for the duration of the experiment.

The carbon-fiber microelectrode was held at a potential of −0.4V, and every 100ms the potential was ramped up to +1.3V and back at a rate of 400V/s. The currents were amplified and recorded using a potentiostat (EI-400 Cypress systems, Lawrence, KS), computer-controlled interface boards (National Instruments, Austin, TX), and the TarHeel CV software (ESA Bioscience Chelmsford, MA). The electrical stimulation was computer generated and passed through a custom-made constant-current generator. Software-controlled biphasic stimulation pulses were applied to the twisted bipolar electrode (Plastics One, Roanoke, VA), which had its tips spread approximately 1 mm apart. Trains of biphasic pulses with a duration of 4 ms (2 ms for each phase) and stimulus intensity of 300µA, were applied at a frequency of 60Hz for 0.4 s (i.e. a total of 24 pulses). Recordings were sampled with an approximately 5 min interval between stimulating trains. Because dopamine oxidizes at approximately +0.6V and reduces at −0.3V, these oxidation and reduction peaks create a unique chemical signature and voltammogram[56]. Voltammograms were obtained every 100ms at a scan rate of 400V/s to create a color plot to confirm the presence of DA. To determine changes in the amplitude of electrically-evoked dopamine, signals were recorded under baseline conditions and were analyzed to determine the average peak amplitude (i.e. the maximal concentration of dopamine evoked by electrical stimulation;[DA]_max).

Following a pre-drug baseline (15 min prior to i.p. drug injection), rats were pretreated with GAT358 (20mg/kg i.p.) or vehicle followed 20 min later by morphine (6mg/kg i.p.). This approach allowed for the assessment of the impact of GAT358 or vehicle alone and its impact on morphine-induced dopamine dynamics. Carbon-fiber microelectrodes were calibrated at the end of the experimental data collection and currents recorded at peak oxidative potential for dopamine (~ +0.6V) were converted to dopamine concentration based on post-calibration peak currents using dopamine [1 μM] buffer solution containing 12mM Tris-HCl, 1.2mM CaCl₂, 1.25mM NaH₂PO₄, 1.2mM MgCl₂ and, 2mM Na₂SO₄. Data was collected for the 90 min observation interval following injection of morphine. Recordings from baseline and following drug injections were analyzed, and values were normalized to each rats‟ baseline[DA]_max to allow for the detection of maximal drug effects as described previously[47, 57–61].

2.6. CPP

The impact of GAT358 on conditioned morphine reward and mediation by CB₁ was assessed using a randomized, unbiased 12-day CPP paradigm as described previously[44, 46]. The CPP apparatus (Med-Associates, St. Albans, VT) consisted of two side chambers with distinct visual cues (vertical or horizontal black and white stripes) separated by a central neutral chamber (gray). The chambers were equipped with computer-controlled (Med-PC software, Med-associates) guillotine doors and photobeams to detect the location of the mouse in all chambers.

Briefly, mice were maintained on a reverse light cycle and were habituated for 20 mins in the experimental room prior to placement in the CPP apparatus. On days 1–3, mice were placed into the central chamber to habituate for 5 min. Next, the flanking guillotine doors opened, and mice were allowed to freely explore all chambers for 30 min (1800 s). The time that the mice spent in each chamber was recorded. On day 3, an initial baseline assessment (i.e. pre-conditioning) was conducted to confirm that mice did not present with a specific chamber bias prior to drug pairings. Mice spending more than 1440 s (i.e. 80% of time) or less than 360 s (i.e. 20% of time) in either conditioning chamber were excluded from the experiment and were never subjected to pharmacological manipulations. Next, mice were restricted to a single, randomly selected side chamber after receiving each of four repeated pairings of the assigned drug condition on alternate days (i.e. on day 4, 6, 8 and 10) with vehicle administered in the opposite chamber on other days (i.e. on day 5, 7, 9 and 11). On day 12, all mice were evaluated in the drug free state (i.e. post-conditioning test) for the time spent in either chamber to assess the development of CPP or conditioned place aversion (CPA). Comparisons were made between the drug-paired chamber time and the vehicle-paired chamber time on the post-conditioning test day versus the pre-conditioning baseline day.

The above protocol was used to evaluate the dose response for morphine to produce CPP, as well as the impact of the CB₁-NAM GAT358 on morphine. Allosteric modulators were also evaluated alone to determine whether the CB₁-NAM GAT358 or the CB₁-PAM GAT229 produced aversion or reward, respectively. In a separate group of mice, we assessed whether effects of GAT358 on morphine CPP were occluded by GAT229. The following drug conditions were evaluated in separate groups of WT mice: morphine alone (4mg/kg i.p.), morphine alone (8mg/kg i.p.), GAT358 (20mg/kg i.p.) + morphine (4mg/kg i.p.), GAT358 (20mg/kg i.p.) + morphine (8mg/kg i.p.), GAT358 alone (20mg/kg i.p.), GAT229 alone (20mg/kg i.p.), (GAT358 (20mg/kg i.p.) + GAT229 (20mg/kg i.p.)) + morphine (8mg/kg i.p.), (GAT358 (20mg/kg i.p.) + GAT229 (20mg/kg i.p.)) alone, GAT229 (20mg/kg i.p.) + morphine (4mg/kg i.p.), GAT229 (20mg/kg i.p.) + morphine (8mg/kg i.p.), rimonabant (3mg/kg i.p.) + morphine (8mg/kg i.p.) or rimonabant (3mg/kg i.p.) alone. A separate group of WT mice assessed concurrently received vehicle in both chambers (data not shown).

We also asked whether the observed effects of GAT358 on morphine CPP were dependent upon CB₁ activation using CB₁KO mice. The following drug conditions were evaluated in CB₁KO mice: morphine alone (4mg/kg i.p.), morphine alone (8mg/kg i.p.), GAT358 (20mg/kg i.p.) + morphine (8mg/kg i.p.) or GAT358 alone (20mg/kg i.p.).

CPP studies were conducted concurrently with interleaved positive and negative controls by the same blinded experimenter (VI). The total injection volume and vehicle used remained consistent across all groups and treatment assignments were randomized to ensure equal baseline chamber times and unbiased pairings to the left and right chambers (counterbalanced). A chamber preference score was calculated and used to compare effects of different pharmacological treatments as follows: Preference score = Time in drug chamber paired chamber post-conditioning (Test) - Time in drug chamber paired chamber pre-conditioning (Baseline).

2.7. Oxycodone TBC-DID

The impact of GAT358 on levels of volitional oral oxycodone and water consumption was evaluated using a 9-day limited access TBC-DID paradigm developed in our laboratory[51]. Single-housed mice on a reverse light cycle were provided with two bottles (10ml serological pipettes with lixit sipper tubes); one “treated” and one “untreated”. The initial left/right positions of the bottles were randomly distributed and swapped daily to avoid positional bias[51]. The bottles were placed into the cages 3h into the dark cycle and the fluid consumption was noted 4h later. Mice continued to receive ad libitum food for the duration of the testing. On days 1–3, both bottles were filled with water, to allow for habituation and a baseline assessment. Next, from day 4–9, mice were randomized into groups and received once daily injections of either vehicle or GAT358 (20mg/kg i.p.) prior to the placement of an “untreated” bottle containing drinking water and a “treated” bottle containing 0.5 mg/ml oxycodone dissolved in drinking water. Mice were weighed on alternate days to calculate the daily oral oxycodone dose (mg/kg) and the volume of water (ml/kg) consumed. The oxycodone preference (%) was calculated as follows: % preference = (treated bottle volume / total volume *100).

2.8. Statistical Analyses

Data were analyzed by one-way, two-way, or repeated-measures ANOVA (followed by Bonferroni‟s post hoc or Holm-Sidak‟s multiple comparisons tests) or by paired or unpaired t-tests, as appropriate. All statistical analyses were performed using GraphPad Prism version 7.05 (GraphPad Software Inc., La Jolla, CA, www.graphpad.com). p<0.05 was considered significant.

3. Results

3.1. GAT358 does not produce characteristic CB₁-mediated tetrad behaviors

As expected, GAT358 did not itself produce cardinal signs of CB1 activation in the tetrad (See Supplemental Results and Supplemental Fig. 1 and 2).

3.2. High dose GAT358 attenuates the cataleptic and hypothermic but not the antinociceptive effects of the CB₁ agonist CP55,940

CP55,940 increased ring test immobility time (GAT358 dose: F_{3, 26}= 35.9, p < 0.0001; CP55,940 dose: F_{3, 78}= 689.6, p < 0.0001; Interaction: F_{9, 78}= 20.05, p < 0.0001; Fig. 2B), lowered body temperature (GAT358 dose: F_{3, 26}= 3.592, p = 0.0270, CP55,940 dose: F_{3, 78}= 105.2, p < 0.0001; Interaction: F_{9, 78}= 4.791, p < 0.0001; Fig. 2C) and increased tail-flick latencies (GAT358 dose: F_{3, 26}= 4.937, p = 0.0076; CP55,940 dose: F_{3, 78}= 129.1, p < 0.0001; Interaction: F_{9, 78}= 1.405, p = 0.2006; Fig. 2D), consistent with cannabimimetic effects, and the interactions were significant. High dose GAT358 (20mg/kg) attenuated catalepsy and hypothermia but enhanced antinociception induced by 0.3mg/kg and 1mg/kg CP55,940 doses (p < 0.05 for all comparisons). The high dose of GAT358 (20 mg/kg i.p.) also produced a greater suppression of cannabimimetic effects compared to lower doses of GAT358 (1 and 10mg/kg) and vehicle-treatment (p < 0.05 for all comparisons). We, consequently, compared effects of the high dose of GAT358 (20 mg/kg) with GAT229 (20 mg/kg) and rimonabant (10 mg/kg) in the same triad of tests of cardinal signs of CB₁ activation. In a separate cohort of mice receiving CP55,940, pharmacological manipulations increased ring test immobility time (Drug treatment: F_{3, 22}= 220.6, p < 0.0001; CP55,940 dose: F_{3, 66}= 820.6, p < 0.0001; Interaction: F_{9, 66}= 120.7, p < 0.0001; Fig. 2E), reduced body temperature (Drug treatment: F_{3, 22}= 16.09, p < 0.0001; CP55,940 dose: F_{3, 66}= 95.97, p < 0.0001; Interaction: F_{9, 66}= 14.14, p < 0.0001; Fig. 2F) and elevated tail-flick latencies (Drug treatment: F_{3, 22}= 11.32, p < 0.0001; CP55,940 dose: F_{3, 66}= 59.44, p < 0.0001; Interaction: F_{9, 66}= 8.616, p < 0.0001; Fig. 2G); these cannabimimetic effects differed as a function of pharmacological treatment and CP55,940 dose and the interaction was significant. GAT358 (20mg/kg) attenuated catalepsy and hypothermia and enhanced antinociception induced by 0.3mg/kg and 1mg/kg CP55,940 dose compared to baseline levels and vehicle treatment (p < 0.05 for all comparisons). In contrast, rimonabant (10mg/kg) attenuated the cataleptic, hypothermic and antinociceptive effects of CP55,940 doses compared to baseline and vehicle treatment. GAT229 (20mg/kg) failed to alter the cataleptic and hypothermic but potentiated the antinociceptive effects of CP55,940 doses compared to baseline and vehicle treatment (p < 0.05 for all comparisons).

Fig. 2. GAT358 attenuates cataleptic and hypothermic but not antinociceptive effects of the orthosteric CB₁ agonist CP55,940.

(A) The schematic shows the experimental timeline for examining impact of drug treatments on the cumulative CP55,940 dosing triad assay. (B) GAT358 attenuated CP55,940-induced catalepsy in a dose-dependent manner. The highest GAT358 dose (20mg/kg) was most effective in reducing CP55,940-induced catalepsy compared to lower GAT358 doses and vehicle treatment. (C) GAT358 attenuated CP55,940-induced hypothermia in a dose-dependent manner with low and high dose GAT358 attenuating hypothermia induced by the high dose CP55,940 (1mg/kg). (D) GAT358 enhanced CP55,940-induced antinociception in a dose-dependent manner with the high dose GAT358 most effectively enhancing CP55,940-induced antinociception compared to lower GAT358 doses and vehicle treatment. (E) GAT358 and rimonabant attenuated while GAT229 failed to alter CP55,940-induced catalepsy. (F) GAT358 and rimonabant attenuated while GAT229 failed to alter CP55,940-induced hypothermia. (G) Rimonabant attenuated while GAT358 trended to and GAT229-treatment enhanced CP55,940-induced antinociception. Data are expressed as mean ± S.E.M. (n = 6–8 per group). “*” indicates vs. pre-drug baseline where ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05, “+” indicates vs. vehicle-treated group with the same symbol indications, two-way repeated measures ANOVA followed by Bonferroni’s post hoc test.

3.3. GAT358 attenuates morphine-induced dopamine efflux in the nucleus accumbens shell evoked by electrical stimulation of the medial forebrain bundle (MFB)

In the anesthetized rat, application of biphasic MFB stimulation (arrow) elicited a rapid and transient release of extracellular dopamine (Fig. 3B). The corresponding voltammogram confirmed that the analyte was dopamine, with a peak oxidation potential of approximately +0.6V and a reduction peak of approximately - 0.2V (Fig. 3B inset). Vehicle pretreatment alone did not significantly alter the evoked [DA]max signal relative to the pre-injection baseline at any timepoint (p > 0.05) (Fig. 3C, 3D left). In these same vehicle-treated rats, a subsequent injection of morphine (6mg/kg) produced a robust increase in the evoked [DA]max signal relative to the pre-injection baseline (Fig. 3C, 3D right) (Drug treatment: F_3,16= 3.112, p = 0.0558, Time (F_7,112= 2.043, p = 0.0557, Interaction: F_21,12= 2.543, p = 0.0009). Morphine significantly increased the extracellular [DA]max signal relative to baseline at the 15, 30 and 45 mins post-injection consistent with morphine-induced increases in dopamine efflux (p < 0.05 for all timepoints).

Fig. 3. GAT358 blocks morphine-induced enhancement of electrically-evoked dopamine efflux without altering dopamine efflux when administered alone.

(A) A carbon-fiber microelectrode (CFM) was lowered into the nucleus accumbens (NAc) shell. A triangular waveform was applied to the CFM from −0.4 V to +1.3 V and back at a scan rate of 400 V/s. A bipolar stimulating electrode (SE) was subsequently lowered over the ipsilateral medial forebrain bundle (MFB) to evoke endogenous dopamine release by application of a biphasic current (60Hz, 0.4s, 300μA). (B) A representative scan shows the electrically-evoked current changes converted to dopamine concentration. The inset shows the corresponding cyclic voltammograms confirming peak of dopamine oxidation potential (~ +0.6V) after electrical stimulation (taken from the peak signal of current change after the stimulus). (C) Neither GAT358 (20mg/kg) alone (empty red triangles) nor vehicle (empty black circles) changed the dopamine signal during the 90 mins following administration that was evoked by MFB stimulation. Morphine (6mg/kg) elicited a robust time-dependent increase (20 min post-injection) in the maximal concentration of electrically-evoked dopamine release ([DA]_max) following vehicle pretreatment (filled black squares). Pretreatment with GAT358 (20mg/kg) prevented morphine enhancement of the dopamine signal (filled red triangles). Pseudo-color plots (x-axis: time; y-axis: applied potential; z-axis: recorded current) display representative serial cyclic voltammograms (from left to right; top and bottom panels) for (D) vehicle and 20 min following morphine (6mg/kg) administration and for (E) GAT358 (20mg/kg) and 20 min following morphine (6mg/kg) administration in the separate rats. The dotted white horizontal line indicates the potential, and the dotted white vertical line indicates the time at which[DA]_max (i.e., the peak DA oxidation potential), occurred. Vehicle or GAT358 was injected at 0 min and the black arrow indicates the subsequent morphine injection at 100 min. Data are expressed as mean ± S.E.M. (n = 4–6 per group). “+” indicates vs. pre-drug baseline (−15 min) where ++++p<0.0001, +++p<0.001, ++p<0.01, and +p<0.05. “*” indicates GAT358+morphine group vs. morphine alone group with the same symbol indications, two-way repeated measures ANOVA followed by Bonferroni’s post hoc or planned comparisons test.

In a separate group of rats, GAT358 alone (20mg/kg) did not alter the evoked [DA]max signal relative to the pre-injection baseline and vehicle-treated rats at any timepoint (p > 0.05) (Fig. 3C, 3E left). In these GAT358-treated rats, a subsequent injection of morphine (6mg/kg) failed to significantly change the evoked [DA]max signal relative to the pre-injection baseline (Fig. 3C, 3E right). The combination treatment of GAT358 (20mg/kg) + morphine (6mg/kg) did not reliably alter the extracellular [DA]max signal relative to baseline at any timepoint over the entire recording interval (p > 0.05 at all timepoints from 0–90 mins post injection). The GAT358+morphine group also showed a significantly lower evoked [DA]_max signal compared to the vehicle + morphine group at the 15, 30 and 45 mins post-injection (p < 0.05 for all timepoints).

3.4. GAT358 blocks morphine-induced CPP without producing preference or aversion via a CB₁-allosteric dependent mechanism

We asked whether GAT358 would block morphine-induced reward using a CPP approach (See Table 1 for detailed statistical results). Both low dose (4mg/kg; Fig. 4B) and high dose (8mg/kg; Fig. 4C) morphine produced robust CPP for the drug-paired chamber relative to the vehicle-paired chamber and this effect was dependent upon the conditioning phase. Mice spent more time in the morphine-paired chamber compared to the vehicle-paired chamber on the CPP test day (4mg/kg: p = 0.0007; 8mg/kg: p < 0.0001) but not the baseline day, consistent with the development of morphine-induced CPP. GAT358 (20mg/kg) blocked CPP to both low dose (4mg/kg; Fig. 4D) and high dose morphine (8mg/kg; Fig. 4E). Neither GAT358 (20mg/kg; Fig. 4F) nor GAT229 (20mg/kg; Fig. 4G) treatment alone altered time spent in the drug-paired chamber relative to the vehicle-paired chamber on the CPP test day. Moreover, CPP to high dose morphine was preserved in mice receiving pretreatment with GAT358 (20mg/kg) + GAT229 (20mg/kg) prior to administration of the high dose of morphine (8mg/kg; Fig. 4H). Mice spent more time in the drug-paired chamber compared to the vehicle-paired chamber post-conditioning (p = 0.0126) but not the baseline day, indicating that the combined pretreatment with GAT358+GAT229 did not prevent development of morphine-induced CPP. Combination treatment with GAT358 (20mg/kg) + GAT229 (20mg/kg) alone did not alter the time spent in the drug-paired relative to the vehicle-paired chamber on the CPP test day (Fig. 4I). In a separate cohort of mice, run concurrently, the time spent in either chamber post-conditioning did not differ when mice received a vehicle injection in both chambers (data not shown).

Table 1.

Impact of GAT358 and GAT229 alone and in combination on morphine CPP

Figure	Group	Strain	n	Main effect – Drug	Main effect - Conditioning Phase	Interaction – Drug x Conditioning Phase	Post hoc (drug- vs. vehicle-paired chamber
4B	Morphine (4)	C57BL/6J WT	10	F 1,16 = 7.362, p = 0.0153	F1,16= 0.9739, p = 0.3384	F 1,16 = 8.675, p = 0.0095	p = 0.0007
4C	Morphine (8)	“	10	F 1,18 = 10.85, p = 0.0040	F1,18= 2.737, p = 0.1154	F 1,18 = 64.85, p < 0.0001	p < 0.0001
4D	GAT358 (20) + Morphine (4)	“	10	F1,16= 0.00067, p = 0.9795	F1,16= 0.1636, p = 0.8998	F1,16= 0.192, p = 0.6671	-
4E	GAT358 (20) + Morphine (8)	“	10	F1,18= 0.05504, p = 0.8172	F1,18= 0.385, p = 0.5427	F1,18= 0.02569, p = 0.8745	-
4F	GAT358 (20)	“	10	F1,18= 0.2953, p = 0.5935	F1,18= 0.8643, p = 0.3651	F1,18= 0.173, p = 0.6824	-
4G	GAT229 (20)	“	10	F1,18= 0.03136, p = 0.8614	F1,18= 2.229, p = 0.1528	F1,18= 0.7581, p = 0.3954	-
4H	GAT358 (20) + GAT229 (20) + Morphine (8)	“	10	F1,18= 2.514, p = 0.1302	F1,18= 2.5, p = 0.1312	F 1, 18 = 11.56, p = 0.0032	p = 0.0126
4I	GAT358 (20) + GAT229 (20)	“	10	F1,18= 0.0413, p = 0.8412	F1,18= 0.499, p = 0.5113	F1,18= 0.1038, p = 0.7511	-
4J	Morphine (4)	CD1 CB1KO	10	F1,18=1.146, p = 0.2985	F1,18= 0.9021, p = 0.3548	F1,18= 0.9415, p = 0.3448	-
4K	Morphine (8)	“	10	F 1,18 = 5.44, p = 0.0315	F1,18= 0.5552, p = 0.4658	F 1,18 = 9.094, p = 0.0074	p = 0.0016
4L	GAT358 (20) + Morphine (4)	“	10	F 1,18 = 4.419, p = 0.0499	F1,18= 0.06871, p = 0.7962	F 1,18 = 9.474, p = 0.0065	p = 0.0025
4M	GAT358 (20)	“	8	F1,14= 0.1363, p = 0.7175	F1,14= 1.816, p = 0.3356	F1,14= 0.2083, p = 0.1992	-
Suppl. 3A	GAT229 (20) + Morphine (4)	C57BL/6J WT	9	F 1,16 = 14.04, p = 0.0018	F 1,16 = 6.182, p = 0.0243	F 1,16 = 15.39, p = 0.0012	p < 0.0001
Suppl. 3B	GAT229 (20) + Morphine (8)	“	10	F 1,18 = 6.584, p = 0.0194	F1,18= 0.6737, p = 0.4225	F 1, 18 = 15.17, p = 0.0011	p = 0.0003
Suppl. 3C	Rimonabant (3) + Morphine (8)	“	7	F1,12= 0.5992, p = 0.4539	F1,12= 0.7368, p = 0.4075	F1,12= 1.139, p = 0.3069	-
Suppl. 3D	Rimonabant (3)	“	6	F1,10= 0.3294, p = 0.5787	F1,10= 0.0029, p = 0.9577	F1,10= 0.9787, p = 0.3458	-
-	Vehicle	“	10	F1,18= 0.2187, p = 0.6457	F1,18= 2.403, p = 0.1385	F1,18= 0.4332, p = 0.5188	-

Fig. 4. GAT358 blocks morphine-induced CPP via a putative CB₁ allosteric mechanism.

(A) The schematic shows the timeline of the CPP paradigm. The black and grey vertical arrows indicate the drug- and vehicle-pairing days, respectively. In WT mice, both (B) low dose (4mg/kg) and (C) high dose morphine (8mg/kg) increased the time spent in the drug-paired chamber relative to the vehicle-paired chamber on the test day indicative of the development of opioid-induced reward. In WT mice, (D) GAT358 (20mg/kg) pretreatment prior to low dose morphine (4mg/kg) and (E) high dose morphine (8mg/kg) reduced the time spent in the morphine-paired chamber, consistent with the blockade of opioid-induced reward. In WT mice, (F) GAT358 (20mg/kg) and (G) GAT229 (20mg/kg) treatment in the absence of morphine did not produce CPP or CPA. (H) In WT mice, the combination of GAT358 (20mg/kg) and GAT229 (20mg/kg) did not alter morphine CPP (8mg/kg), indicating GAT229 occludes GAT358 blockage of opioid-induced reward. (I) In WT mice, in the absence of morphine, the combination of GAT358 (20mg/kg) and GAT229 (20mg/kg) did not produce CPP or CPA. In CB₁KO mice, high dose (8mg/kg) (K), but not low dose (4mg/kg) (J) morphine increased the time spent in the drug-paired chamber relative to the vehicle-paired chamber on the CPP test day, indicating the development of opioid-induced reward for high dose morphine alone. (L) In CB₁KO mice, GAT358 (20mg/kg) did not alter CPP to morphine (8mg/kg). (M) In CB₁KO mice, GAT358 (20mg/kg) in the absence of morphine did not produce CPP or CPA for the drug-paired chamber. All pretreatment drugs were administered 20 min prior to the morphine injection. No difference in chamber preference times were observed on the pre-conditioning baseline day in any cohort of mice. Data are expressed as mean ± S.E.M. (n = 8–10 per group) “*” indicates vs. vehicle-paired chamber time where ****p<0.0001, ***p<0.001, **p<0.01, and *p<0.05, two-way repeated measures ANOVA followed by Bonferroni’s post hoc test. See Table 1 for detailed statistical results.

In CB₁KO mice, low dose morphine (4mg/kg; Fig. 4J) did not produce CPP. However, high dose morphine (8mg/kg; Fig. 4K) produced robust CPP for the drug-paired chamber relative to the vehicle-paired chamber and this effect was dependent upon the conditioning phase. Mice spent more time in the drug-paired compared the vehicle-paired chamber on the CPP test day (p = 0.0016) but not the baseline day, consistent with the development of morphine-induced CPP. GAT358 did not attenuate CPP to high dose morphine (8mg/kg; Fig. 4L), an effect which was also dependent upon conditioning in CB₁KO mice. Mice spent more time in the drug-paired chamber compared to the vehicle-paired chamber on the CPP test day (p = 0.0025) but not the baseline day. GAT358 (20mg/kg) alone did not alter time spent in the drug-paired chamber relative to the vehicle-paired chamber on the CPP test day in CB₁KO mice (Fig. 4M).

3.5. The combination of GAT358+morphine reduced drug chamber preference scores compared to morphine alone

In WT mice, pharmacological treatments altered drug chamber preference scores (F_4,42= 2.858, p = 0.0350; Fig. 5A). The high dose morphine (8mg/kg) group showed increased preference scores compared to vehicle (p = 0.0190), GAT358 (20mg/kg) + morphine (8mg/kg) (p = 0.0439), and GAT358 (20mg/kg) alone (p = 0.0439) treatment groups. However, chamber preference scores were similar in groups receiving morphine (8 mg/kg alone) and rimonabant (3mg/kg) + morphine (8mg/kg) (p > 0.05) treatment groups. In contrast, chamber preference scores were similar in vehicle-treated groups compared to GAT358 (20mg/kg) + morphine (8mg/kg), GAT358 (20mg/kg) alone, and rimonabant (3mg/kg) + morphine (8mg/kg) treatment groups (p > 0.05 for all comparisons). In a separate cohort of WT mice, pharmacological treatments altered drug chamber preference scores (F_2,25= 5.637, p = 0.0095; Fig. 5B). The low dose morphine (4mg/kg) group showed increased preference scores compared to vehicle (p = 0.0074) and GAT358 (20mg/kg) + morphine (4mg/kg) (p = 0.0187) treatment groups. Chamber preference scores were similar in the GAT358 (20mg/kg) + morphine (4mg/kg)-treated group compared to vehicle-treated group (p > 0.05). In CB₁KO mice, preference scores did not differ between high dose morphine (8mg/kg) alone and GAT358 (20mg/kg) + morphine (8mg/kg) groups (t₁₈= 0.1538, p = 0.8795; Fig. 5C). Thus, GAT358 blocked CPP to morphine.

Fig. 5. GAT358 but not rimonabant reduced preference for the morphine-paired chamber.

(A) In WT mice, the drug chamber preference score for the GAT358 (20mg/kg) + high dose morphine (8mg/kg) group was lower than the high dose morphine alone (8mg/kg) but did not differ reliably from the vehicle alone group. (B) In WT mice, the drug chamber preference score for the GAT358 (20mg/kg) + low dose morphine (4mg/kg) group was lower than the low dose morphine alone (4mg/kg) but did not differ reliably from the vehicle alone group. (C) In CB₁KO mice, the drug chamber preference score for the GAT358 (20mg/kg) + high dose morphine (8mg/kg) group did not differ reliably from the high dose morphine alone (8mg/kg) group. Data are expressed as mean ± S.E.M. (n = 7–10 per group) “*” indicates vs. morphine alone group where **p<0.01 and *p<0.05, one-way ANOVA followed by Bonferroni’s post hoc or multiple comparisons test, n.s. indicates non significance.

3.6. GAT358 reduces oxycodone but not water consumption in an oral oxycodone TBC-DID paradigm

GAT358 reduced the daily orally consumed oxycodone dose (mg/kg) consumed in our TBC-DID paradigm, and oxycodone consumption trended to differ across sessions, but the interaction was not significant (GAT358 treatment: F_1,18= 26.26, p < 0.0001; Session: F_5,90= 2.301, p = 0.0513, Interaction: F_5,90= 0.8083, p = 0.5467; Fig. 6B). Body weight did not differ between groups (data not shown). In the same mice, water consumption (ml/kg) was not reliably altered by GAT358 treatment, trended to differ across sessions and the interaction was not significant either during the baseline (data not shown) or the TBC-DID (GAT358 treatment: F_1,18= 0.3039, p = 0.5882; Session: F_5,90= 2.278, p = 0.0534, Interaction: F_5,90= 0.2042, p = 0.9599) sessions (Fig. 6C). As expected, percent preference for the oxycodone bottle (i.e. the oxycodone bottle % preference) did not differ as a function of the drug treatment, was not altered across sessions and the interaction was not significant (Drug: F_1,18= 2.858, p = 0.1081; Session: F_5,90= 0.2819, p = 0.9219, Interaction: F_5,90= 0.3239, p = 0.8974; Fig. 6D). Thus, GAT358 selectively reduced oral oxycodone but not water consumption throughout the observation interval.

Fig. 6. GAT358 reduces oral oxycodone consumption but not the water intake in a two-bottle choice drinking-in-dark (TBC-DID) paradigm.

(A) The schematic shows the timeline for the TBC-DID paradigm with a 3-day water TBC-DID baseline and a 6-day oxycodone TBC-DID. The vertical black arrows indicate GAT358 (20mg/kg) or vehicle (i.p.) treatment. (B) The oxycodone dose consumed (mg/kg) was lower in the GAT358-treated group compared to the vehicle-treated group overall indicating that daily GAT358 treatment (20mg/kg/day i.p. x 6 days treatment) reduced orally consumed oxycodone throughout the observation interval. (C) GAT358 treatment did not reliably alter the water consumption volume (ml/kg). (D) Mice subjected to a six-day TBC-DID paradigm did not show preference for the oxycodone-treated bottle and this % preference did not differ as a function of GAT358 treatment. Data are expressed as mean ± S.E.M. (n = 10 per group). “*” indicates GAT358 vs. vehicle-treatment where ****p<0.0001, two-way repeated measures ANOVA followed by Bonferroni‟s post hoc test. BL- baseline.

4. Discussion

CB₁-NAMs have been designated by NIDA‟s Division of Therapeutics and Medical Consequences as one of the top ten pharmacological targets for rapid development of therapeutics in response to the opioid crisis[62]. However, to our knowledge, no prior study has evaluated in vivo efficacy of a CB₁-NAM on opioid-mediated reward. Our studies support negative allosteric modulation of CB₁ signaling as a potential therapeutic strategy for fine-tuning unwanted pharmacological effects of opioid signaling to reduce opioid-induced reward.

The CB₁-NAM GAT358 was developed by a focused structure-activity relationship study of PSNCBAM-1, a well-characterized CB₁-NAM, and has comparable activities in the cAMP and β-arrestin assays to PSNCBAM-1[36]. However, there are no published reports to date that enumerate the in vivo pharmacological effects of GAT358 either by itself or in the presence of an orthosteric CB₁ agonist such as CP55,940. Several previous studies indicate that putative CB₁-NAMs have differential CB₁ signaling effects in vivo compared to their in vitro profiles. PSNCBAM-1 did not alter Δ⁹-THC-induced effects on activity, hypothermia, or catalepsy but decreased Δ⁹-THC’s anti-nociceptive effect[63]. Conversely, Org27569 did not alter CP55,940 induced catalepsy or anti-nociception but altered the development of hypothermia[64]. Such differential effects could be due to the fact that the activity of most allosteric modulators depends on the orthosteric probe being used (i.e. probe-dependence) and underscores the difficulty in elucidating the in vivo effects of CB₁ allosteric modulators[65, 66]. GAT358 did not produce cardinal signs of CB₁ activation or inactivation (i.e., hypomotility, catalepsy, hypothermia, anxiety, aversion, etc.) by itself. Unlike other CB₁-NAMs, GAT358 also suppressed the cataleptic and hypothermic effects of the classical CB₁ agonist CP55,940 but did not alter the ability of CP55,940 to produce antinociception in the spinally-mediated tail-flick test in our studies. These results are consistent with in vitro data showing that GAT358 acts as a CB₁-NAM in an arrestin-biased manner[37] and does not engage the orthosteric CB₁ binding site or elicit cannabimimetic effects.

Abuse liability remains one of the main limitations of the therapeutic utility of opioid analgesics. GAT358 eliminated morphine-induced increases in electrically-evoked dopamine efflux in the mesocorticolimbic pathway while not affecting basal electrically-evoked dopamine efflux. The NAc and MFB are two principal components of the mesocorticolimbic reward pathway and dopamine release in this circuit is a critical mediator of drug reward-related processes[15, 67–70]. The endocannabinoid system regulates afferent inputs to dopaminergic neurons to influence their activity[71] and previous FSCV findings indicate that endocannabinoids modulate dopaminergic neurotransmission[72, 73]. Furthermore, dopamine release in the NAc produced by nicotine, ethanol, and cocaine is attenuated by rimonabant[74]. Taken together, these observations implicate a role for CB₁ in the development of drug-context association and suggest that CB₁ inhibition may attenuate drug reward[75]. GAT358 could inhibit CB₁ located on VTA GABAergic[76] or glutamatergic[77] presynaptic terminals and thereby prevent the retrograde signaling of endocannabinoids produced from dopaminergic VTA neurons. Alternatively, activation of cortical afferents to the VTA can produce long term adaptations in dopaminergic cells that result in behavioral sensitization and consequently drug addiction[78, 79]. The circumscribed allosteric effects of GAT358 may result in endocannabinoid-mediated self-regulation of dopaminergic neurons that reduces drug reward[80]. Such mechanisms could restore the balance between inhibitory and excitatory synaptic signaling and promote homeostasis following aberrant opioid-induced dysregulation in the mesocorticolimbic reward pathway[81]. Because GAT358 was administered systemically in our FSCV experiments, several of these potential mechanisms could mediate the suppression of opioid-induced dopamine efflux and reward by disruption of CB₁ signaling[31].

Studies in pre-clinical animal models highlight several potential mechanisms through which cellular interactions between the CB₁ and opioid receptors occur to modulate reward and addiction-related behaviors. While chronic opioid treatment results in altered endocannabinoid production in several brain regions associated with reward, the extent and direction of these changes vary considerably[82–85]. Several factors such as the species used (mice or rats), type of opioid, dosage and duration of treatment, pattern of drug administration (experiment administered or free-choice self-administration) may play a role in regulating this expression[85–87]. CB₁ and μ-opioid receptors also co-localize on the same neurons[17] and can potentially compete for the same pool of Gi/o-proteins[88]. Heterodimers of CB₁ and μ-opioid receptors have also been reported in the NAc, and stimulation of these receptor complexes can cause synergistic suppression of GABA release[89]. CB₁ is also localized to terminals at both inhibitory and excitatory synapses, where endocannabinoids act as synaptic circuit breakers[90]. Additional work is necessary to determine the specific neuronal populations and potential mechanisms and sites of action involved in the effects described in our study.

Our neurochemical findings using FSCV are complemented by our behavioral results. GAT358 blocked CPP induced by both low and high dose morphine without producing preference or aversion when administered by itself. Attenuation of CB₁ signaling suppresses reward and reinforcement of multiple drugs of abuse including morphine, cocaine, and nicotine[91–94]. CB₁ antagonists/inverse agonists such as rimonabant and AM251 decrease acquisition, expression, duration of extinction and reinstatement of morphine-induced CPP[93, 95–97], but direct CB₁ antagonism may alter endogenous reward pathways[13, 97–101]. By contrast, GAT358 lacked intrinsic effects on reward in our studies and blocked CPP to both low and high dose morphine. Systemic administration of rimonabant attenuated morphine-induced CPP in rats (4mg/kg) and mice (1mg/kg) and also attenuated heroin (0.06mg/kg per infusion) and morphine (0.2µg/kg per infusion) self-administration in rats and mice, respectively, at doses that were not active by themselves[94, 95, 102]. Moreover, intra-accumbal and intra-prefrontal cortex infusions of rimonabant reduced heroin (0.02mg/infusion) self-administration in rats[103, 104]. Systemic administration of the CB₁ neutral antagonist AM4113 also dose-dependently inhibited heroin (0.5 mg/kg per infusion) self-administration in rats[105]. The dosages of morphine used in the present report were significantly higher than those used in these previous studies examining morphine-induced CPP in mice, raising the possibility that GAT358 may have a much bigger effect size in suppressing opioid reward.

GAT358-induced blockade of opioid reward was absent in CB₁KO mice and occluded by a CB₁-PAM, raising the possibility that a CB₁ allosteric mechanism of action mediates these effects. GAT229 is the S-(−)-enantiomer of the racemic ago-PAM compound, GAT211[106]. In vitro and in vivo evidence indicates that GAT229 shows no independent activity at CB₁ but potentiates the effect of endogenous CB₁ orthosteric agonists, consistent with a PAM profile[106, 107]. Our findings must, nonetheless, be parsimoniously interpreted as the in vitro profile of other CB₁ allosteric modulators has not necessarily translated in vivo[30, 63, 108]. In vivo effects of a CB₁-PAM on a CB₁-NAM have not been previously evaluated. Occlusion of the effects of GAT358 by a CB₁-PAM could involve alternate mechanisms (e.g. physiological antagonism) that are not necessarily allosterically-mediated. Finally, CB₁KO mice showed CPP to high but not low dose morphine, suggesting that CB₁ receptor knockout itself attenuates but does not eliminate opioid reward[109, 110]. Nonetheless, CB₁KO mice did show robust CPP to high dose morphine, and this effect was not blocked by GAT358. These observations provide another line of evidence to support our hypothesis that the observed in vivo pharmacological effects of GAT358 are mediated by CB₁.

Our limited access oral oxycodone TBC-DID paradigm induces physical dependence and mesocorticolimbic region-dependent increases in ΔFosB expression without producing dose escalation or oxycodone-induced preference[51]. Here, GAT358 reduced the daily oral oxycodone dose consumed, without altering water consumption, further suggesting that suppression of CB₁ signaling attenuates the reinforcing properties of opioids.

Blockade of opioid reward by GAT358 in our behavioral paradigms is unlikely to be result of memory impairment for several reasons. Downregulation of endogenous CB₁ signaling by antagonists has been suggested to facilitate memory and dose-dependently reverse memory deficits induced by of cannabimimetics[111, 112]. CB₁KO mice exhibit enhanced long term potentiation in the Schaffer collateral–CA1 synapses indicating an enhanced capacity to strengthen synaptic connections in a brain region crucial for memory formation[113]. CB₁-NAMs such pregnenolone and cannabidiol also ameliorate Δ⁹-THC-induced memory impairments in mice without side-effects[114, 115]. The vehicle employed here did not impair memory in our previous studies evaluated at multiple retention intervals (unpublished data). Finally, results of our CPP and oral oxycodone consumption studies are consistent with results of our in vivo voltammetry studies that directly measure morphine-evoked DA efflux in the NAc. Taken together, these results suggest that it is unlikely that GAT358 produces memory impairment that confounds the results of our behavioral tests.

Our behavioral and neurochemical studies thus collectively suggest that CB₁-NAMs such as GAT358 represent a class of novel drug candidates with potential to reduce the rewarding properties of opioids. Our findings provide a better understanding of how the CB₁ cannabinoid and opioid systems can be targeted therapeutically and, in the context of the current opioid epidemic, have significant translational relevance. Currently approved opioid-based medications for treating opioid use disorder have limitations such as the lack of a ceiling effect[116], lethal effects in combination with central nervous depressants[117–121], poor adherence[122–124] and re-narcotization[125, 126]. There is, therefore, an urgent need for novel non-opioid therapies to complement current approaches [127].

While, theoretically, a CB₁-NAM provides numerous benefits over CB₁ orthosteric antagonists and inverse agonists[128], it‟s in vitro effects must translate in vivo as well. From the standpoint of translation and relevance to human opioid addiction several key factors need to be addressed. First, future studies will seek to extend these findings to female rodents, as the impact of sex on cannabinoids[129] and opioid addiction[130–132] must be considered. Second, we specifically employed naïve mice in our study so that the impact of GAT358 on the positive reinforcing effects (i.e. reward) of opioids could be assessed without the possible confound of negative reinforcing effects (i.e. removal of an aversive pain state). However, several studies suggest a complex interplay between the presence of a pain state in the development of opioid reward and unwanted side-effects[133, 134] and the effects of a pathological pain state on the ability of GAT358 to modulate opioid reward need to be evaluated. Finally, our evaluation of GAT358 does not directly address the adverse side-effects normally associated with direct CB₁ downregulation such as depression, anxiety, and suicidality[24, 26, 27] and these factors may increase the translational relevance of our findings. More work is necessary to elucidate the impact of allosteric modulation on CB₁ signaling and its response to natural reward or other drugs of abuse.

We postulate that a CB₁-NAM could potentially suppress aberrant signaling cascades involved in opioid-seeking behavior, while preserving mechanisms required for normal pharmacotherapeutic functions. Our results highlight the clinical potential of novel multimodal cannabinoid-opioid combination therapies and may help accelerate solutions to the current opioid epidemic.

Highlights:

• CB₁ negative allosteric modulator GAT358 does not produce cannabimimetic effects

• GAT358 decreases CB₁ agonist-induced catalepsy, hypothermia but not antinociception

• GAT358 eliminates morphine-induced phasic dopamine release in the NAc shell

• GAT358 blocks conditioned place preference to morphine via a CB₁ mechanism

• GAT358 reduces oral oxycodone but not water intake in a two-bottle choice assay

Abbreviation List:

CB₁

Cannabinoid-type 1 receptor

PAM

Positive allosteric modulator

NAM

Negative allosteric modulator

NAc

Nucleus accumbens

FSCV

fast-scan cyclic voltammetry

CPP

conditioned place preference

TBC-DID

two-bottle choice drinking in the dark

WT

wildtype

CB₁KO

CB1-knockout

MFB

medial forebrain bundle