Evidence for spontaneous cannabinoid withdrawal in mice

Abstract

Although the behavioral effects of acute and chronic exposure to cannabinoids have been extensively studied in mice, spontaneous withdrawal following exposure to cannabinoids has not been well characterized in this species. To address this issue, different groups of mice were treated for 5 days with saline, 20-36 mg/kg/day of the CB partial agonist Δ⁹-tetrahydrocannabinol (Δ⁹-THC), or 0.06-0.1 mg/kg/day of the CB high-efficacy agonist AM2389. Initial studies assessed changes in observable behavior (paw tremors) that were scored from the recordings taken at 4 or 24 h after the last injection. Subsequently, radiotelemetry was used to continuously measure body temperature and locomotor activity before (baseline), during, and after the 5-day dosing regimens. Results show that increases in paw tremors occurred following 5-day exposure to AM2389 or Δ⁹-THC. In telemetry studies, acute AM2389 or THC decreased both temperature and activity. Rapid tolerance occurred to the hypothermic effects of the cannabinoids, whereas locomotor activity continued to be suppressed following each drug injection. In contrast, increases in locomotor activity were evident 12-72 h after discontinuing daily injections of either 0.06 or 0.1 mg/kg/day AM2389. Increases in locomotor activity were also noted in mice treated daily with 30 or 36, but not 20 mg/kg/day Δ⁹-THC; these effects were smaller and appeared later than effects seen in AM2389-treated mice. These results indicate that the discontinuation of daily treatment with a CB high-efficacy agonist will yield evidence of spontaneous withdrawal that may reflect prior dependence, and that the degree of cannabinoid dependence may vary in relation to the dose or efficacy of the agonist injected daily.

The regular use of many drugs results in physical dependence, ascertained by the appearance of an associated withdrawal syndrome after drug exposure is delayed or discontinued (abstinence). Opioid and alcohol withdrawal have been widely studied and have well-delineated abstinence syndromes following the interruption of drug treatment. Each type of spontaneous withdrawal is characterized by objective physiological changes that are easily replicated in laboratory animals, and there is little controversy regarding their diagnosis. Spontaneous withdrawal following chronic use of monoaminergic stimulants, nicotine, or caffeine also has been described; however, withdrawal from these drugs is not necessarily accompanied by robust physical signs or symptoms and, thus, an abstinence syndrome has been more difficult to model in laboratory animals. Interrupted use of cannabis, the most commonly used illicit drug, likewise has been associated with withdrawal symptoms. By some estimates up to 30-70% of daily users, and 10 - 17% of all users, describe symptoms associated with spontaneous cannabis withdrawal (Agrawal et al., 2008; Bahji et al., 2020; Copersino et al., 2006). However, like nicotine or caffeine dependence, the predominant symptoms of cannabinoid abstinence such as nightmares, angry outbursts, and difficulty falling asleep are psychometric and, consequently, difficult to translate to animal models (Allsop et al., 2011).

Despite such obstacles to defining cannabinoid withdrawal in laboratory animals, more easily-studied signs of abstinence-induced cannabinoid withdrawal have been described previously, albeit infrequently. For example, decreases in food-maintained operant responding in rhesus monkeys or increase in paw tremor and head shake behaviors in mice during abstinence have been forwarded as behavioral signs of spontaneous withdrawal and, hence, cannabinoid dependence (Beardsley et al., 1986; Trexler et al., 2018). Most studies of cannabinoid dependence in laboratory animals, however, have failed to identify distinct and reproducible signs of spontaneous withdrawal (Lichtman and Martin, 2002; Maldonado, 2002). In the absence of a reliable assay of spontaneous cannabinoid withdrawal, a more common approach to identifying cannabinoid dependence in laboratory animals has been to administer the CB1 antagonist rimonabant (SR141716A) to chronically CB-treated mice (Cook et al., 1998; Huang et al., 2009), rats (Aceto et al., 1996; Tsou et al., 1995), or monkeys (McMahon and France, 2003). Remarkably, rimonabant was used as a tool to study cannabinoid dependence within a year of the first published report on its development as a selective CB1 antagonist (Rinaldi-Carmona et al., 1994) highlighting the perceived need for a laboratory assay of cannabinoid withdrawal.

The administration of rimonabant to rats or mice treated daily with Δ⁹-tetrahydrocannabinol (Δ⁹-THC) yields a constellation of behavioral effects, which has been described as precipitated cannabinoid withdrawal, and is often taken as an indication of the development of physical dependence to Δ⁹-THC (Aceto et al., 1996; Cook et al., 1998). This approach to studying cannabinoid dependence is predicated on opioid dependence studies, in which a close congruence between antagonist-precipitated withdrawal and abstinence-associated withdrawal was firmly established in opioid addicts and laboratory animals (Wikler et al., 1953; Wright et al., 1991). A similarly close congruence has not been established in cannabinoid dependence studies and, consequently, evidence that abstinence-associated withdrawal is described by the effects of rimonabant in cannabinoid-treated subjects is suggestive but not compelling.

The limited evidence of spontaneous, or abstinence-associated cannabinoid withdrawal in animals may reflect the common use of Δ⁹-THC as the drug administered daily. Although Δ⁹-THC is the major psychoactive component of cannabis, and the cannabinoid compound most associated with human use, it is a partial agonist at CB1 receptors (Sim et al., 1996; Wilson et al., 2019) and studies with a higher efficacy drug may be more likely to produce stronger evidence of abstinence-related withdrawal and cannabinoid dependence. Support for this idea comes from other pharmacological classes, e.g. opioids. For example, the potent but low efficacy opioid agonist buprenorphine is associated with milder dependence-liability than drugs with relatively higher efficacy such as morphine (Martin et al., 1976; Tompkins et al., 2014). Another complexity in translating human cannabinoid dependence to animals is that, whereas signs of cannabinoid abstinence in humans are slow to emerge, with peak effects reported 2 to 6 days after the onset of abstinence (Budney and Hughes, 2006; Haney et al., 1999), most animal studies are limited to discrete observations only up to 24 h after the last injection. Indeed, two studies that did find signs of withdrawal in monkeys or mice following the cessation of Δ⁹-THC treatment reported that peak effects occurred, on average, after 1.5 to 3 days of abstinence (Beardsley et al., 1986; Trexler et al., 2018).

In an effort to further elucidate effects of spontaneous cannabinoid withdrawal in mice, we compared the effects of daily administration of different doses of two cannabinoid receptor agonists: Δ⁹-THC, the prototypical cannabinoid agonist, and AM2389, a high efficacy cannabinoid receptor agonist (Paronis et al., 2012). In keeping with previous studies of cannabinoid dependence, we used five-day treatment regimens; however, in parallel with the expression of cannabinoid withdrawal in humans, we monitored the behavior of the animals for up to five days after the last injection, a longer period than has been commonly reported in earlier studies. Results show that spontaneous cannabinoid withdrawal is observable by measures of paw tremors and increased locomotor activity, with exposure to AM2389 producing qualitatively greater changes in behavior than observed following daily treatment with Δ⁹-THC. The present results indicate that the magnitude of cannabinoid dependence can be related to the efficacy of the cannabinoid receptor agonist.

Methods:

Subjects:

Adult male CD-1 or C57/BL6 mice, weighing 25-30g at the start of the study (Charles River Laboratories, Wilmington MA), were housed 4 per cage if used in studies of observed behavior and were housed singly if implanted with radiotelemetry emitters. Food and water were available ad libitum. Studies of observed behavior were completed during the light portion (lights on at 7AM) of the light/dark cycle; radiotelemetry data were gathered over continuous 24 h periods in mice housed under a reverse light/dark cycle (lights on at 11PM). All studies were approved by the Northeastern University or McLean Hospital Animal Care and Use Committees, in accordance with guidelines established by the National Research Council.

Observed Behavior:

CD-1 mice were injected once or twice per day (at 8AM and/or 6PM) with vehicle, 0.03-1.0 mg/kg AM2389, or 10 mg/kg Δ⁹-THC for five days. On test days, at 4 or 24 h following the last vehicle, AM2389 or Δ⁹-THC injection, mice received an injection of vehicle and were placed in clear acrylic observation chambers (8” x 4” x 5”) with mirrors located below and behind each chamber. Behavior was simultaneously recorded from four adjacent chambers with a Canon Vixia camcorder in 5 minute epochs at 15, 30, and 45 minutes after the vehicle injection. Recordings of each mouse were reviewed later by blinded observers; recordings were played back on a computer at 0.67x speed, and standardized procedures were used to score the behaviors of each subject using freeware (EthoLog v2.2). Rearing and grooming were scored according to duration of time engaged in the behavior; paw tremors, head shakes, and scratching were scored once per occurrence. The following definitions were used to categorize each behavior: grooming-both front paws involved in synchronous behavior in combination with mouth or other body part (e.g., front, back, or anogenital area); rearing-any instance of having only rear paws on the ground and no evidence of grooming; scratch-rapid movement of a rear paw against another body part (usually head or torso); head shakes-head movement, typically a rotational shake, from shoulders or neck to nose; paw tremor-quick movement of one or both forepaws, ranging from a single paw twitch to full fluttering or shaking of both paws, with no involvement of other body parts. No significant differences in grooming, rearing, scratching, or headshakes were noted and these behaviors are not discussed further.

Radiotelemetry:

Mice were implanted intraperitoneally with radiotelemetry probes (G2 or G2HR E-mitter transponders; StarrLife Sciences Corp, Oakmont, PA) under anesthesia (100mg/kg ketamine; 10mg/kg xylazine). Transponders were inserted through a 1.5cm incision in the abdominal area and were attached to the abdominal wall with absorbable 5.0 suture. Incisions were closed with 5.0 absorbable chromic gut suture. After surgery, animals were singly housed in cages placed above receivers (Model ER-4000, StarrLife Sciences Corp). Implanted transponders produced activity- and temperature-modulated signals that were registered by the receiver, processed using Vital View data software, and stored on a computer in 15 minute bins, 24 h per day. Following a 7-10 day recovery period, baseline data were gathered for five consecutive days. Next, mice received one or two daily injections of saline, 0.03 or 0.1 mg/kg AM2389, 10-30 mg/kg Δ⁹-THC, or 10 mg/kg morphine for five or five and a half days. The first and last injections of the regimen were given at 11AM, all other injections were given within 30 minutes of 11AM and 11PM daily. Data continued to be recorded for six days after the last injection.

Drugs:

Δ⁹-THC was obtained from the Drug Supply Program of the National Institute on Drug Abuse (NIDA, Rockville, MD), AM2389 was synthesized in the Center for Drug Discovery at Northeastern University, and morphine sulfate was obtained from Sigma-Aldrich Chemicals (St. Louis, MO). Δ⁹-THC and AM2389 were dissolved in a vehicle of 1:1:18 ethanol:oil:saline, and were further diluted with saline, morphine was dissolved in saline. All injections were given subcutaneously, and drug doses are expressed in terms of the base weight.

Statistical analysis:

All data were plotted, and statistical analysis completed, using GraphPad Prism v. 8.4 (GraphPad Software, San Diego California). Significance for all tests was set at p <0.05. Observed behavior. Each recording was initially scored by 2 observers; if there was ≥90% agreement, then the data were averaged. In cases where the scores differed by >10%, a third observer scored the recording and the median value was used (16% of all observations). Data obtained at 15, 30, and 45 minutes post-injection were grouped together and are expressed as the average number of observations/5 minutes. Data were analyzed using 2-way ANOVA, followed by Holm-Sidak’s or Dunnett’s multiple comparison test, except as noted. Radiotelemetry.: Temperatures were averaged over 1 h periods and are expressed as change from measures recorded at 11AM. Locomotor activity was summed over 4 and 12 h periods. Data from the five baseline days were averaged to obtain control values for each subject. Data obtained on the penultimate day of daily drug injections in groups of mice that received twice daily injections were excluded from analysis so that an equal number of drug days were compared for all groups. Data were analyzed using 2-way ANOVA, followed by Dunnett’s multiple comparison test, except where noted.

Results

Observed behavior: paw tremors

Two groups of CD-1 mice (n=8) received twice daily vehicle injections for five days and a saline injection either 4 or 24 h after the last vehicle injection and immediately prior to placement in the experimental chamber (open bars, Figure 1). The average paw tremor rates (mean ±SEM) for these two groups were 7±1 and 4±1 per five minutes and were not significantly different [t₍₁₄₎=1.75, p>0.1]. Differences in paw tremor rates were apparent in CD-1 mice that received daily injections of AM2389 or Δ⁹-THC (filled bars, Figure 1). Comparisons between vehicle-treated and drug-treated mice indicate that paw tremor rates for mice tested 4 h after the last Δ⁹-THC or AM2389 injection were lower than those in vehicle-treated mice [F_(3,28)=9.6, p<0.001]. In contrast, paw tremor rates in mice tested 24 h after the last injection of AM2389 were significantly greater than those in vehicle-treated mice [F_(4,35)=3.9, p<0.001]. Mean paw tremor rates in mice tested 24 h after the last injection of 10 mg/kg Δ⁹-THC were more than twice the rate in vehicle-treated animals but lay below values needed to reach statistical significance (Figure 1).

Figure 1.

Paw tremors in CD-1 mice injected for 5 days with vehicle, twice daily with 10 mg/kg Δ⁹-THC or 0.03 mg/kg AM2389, or once daily with 0.1 mg/kg or 1.0 mg/kg AM2389. Horizontal axis: time, 4 or 24 h, after the last daily injection. Vertical axis: paw tremors/5 minutes. Vertical bars: group means ±SEM, n=8; *p<0.05, **p<0.01, compared to time-matched vehicle control.

Radiotelemetry: Control data

Control data for temperature and locomotor activity were collected over five days and then averaged to create 24 h control values for each mouse. All mice exhibited diurnal rhythms for body temperature and locomotor counts, with values of both measures decreasing after the lights came on and increasing after lights were turned off. Averaged body temperatures were stable across days and ranged from a low of 34.4±0.5°C, at 8AM, to a high of 35.7±0.5°C, at 4PM. Averaged across days for individual subjects and compiled in 12 h bins corresponding to the light and dark cycles, locomotor counts ranged from 10,948±1,148 to 13,452±1,237 during the dark cycle and from 5,418±269 to 7,872±630 during the light cycle.

Saline treatment.

One group of CD-1 mice received one or two injections of saline every day for 5 days, and one group of C57/BL6 mice received two injections of saline for 5 days (Figure 2). Temperature and locomotor activity obtained in CD-1 mice during the daily injection or post-injection phases of the study did not differ from their control values [temperature: F_(10,55)=0.95, p>0.4; locomotor activity: F_(10,55)=0.47, p>0.9], and there was within day variation in both measures [temperature: F_(5,256)=23.5, p<0.001; locomotor activity: F_(3,177)=84.2, p<0.001]. Temperature data from the saline-treated C57/BL6 mice also did not differ from control values during the injection or post-injection phases of the study [F_(10,55)=0.28, p>0.9]. Locomotor activity was generally unaffected by saline, although slight increases were recorded during the first 4 h of the dark phase on the 3^rd and 5^th days of the daily injection period [F_(10,55)=2.1, p<0.05]. As in CD-1 mice, significant diurnal differences in both measures were evident within days [temperature: F_(5,261)=240, p<0.001; locomotor activity: F_(4,235)=233, p<0.001].

Figure 2:

Radiotelemetry data from vehicle-treated CD-1 mice (left panels) or C57/BL6 mice (right panels) before, during, and after saline treatment. Open symbols (Control) represent averaged values collected over five days prior to saline injections, grey symbols present data collected over the daily injection period, black symbols present data collected for five days starting the day after the last injection. Horizontal axis: days. Vertical axis: (Top panels) locomotor activity counts collected over 4 h, (bottom panels) change in body temperature, relative to temperature recorded at 11AM. Each point represents the mean of 6 mice, vertical bars are SEM. *p<0.05, **p<0.01, ***p<0.001 compared to time-matched control.

Radiotelemetry: temperature

CD-1 mice.

Following determination of their individual control values, different groups of mice received daily injections of 0.1 mg/kg AM2389 for five days or twice daily injections of 0.03 mg/kg AM2389, 10 mg/kg Δ⁹-THC, or 10 mg/kg morphine for five and a half days (the extra half-day allowed the first and last injection in all treatment groups to be administered at 11AM). Significant hypothermic effects were evident in all cannabinoid-treated groups following the first injection [F_(4,700)=11.0, p<0.001]. The greatest response occurred in the group that received 0.1 mg/kg AM2389 (Figure 3, top panel): in these mice, temperature decreased, on average, by 5.9±0.6°C with peak effects occurring 4 to 8 h after injection. The hypothermic effects of the first injection of 0.1 mg/kg AM2389 persisted for 24 h, however none of the subsequent injections of AM2389 reduced body temperature in this group. The effects of 0.03 mg/kg AM2389 were less than those of 0.1 mg/kg: both the first and the second injections of 0.03 mg/kg AM2389 reduced body temperature by 2.2°C, with peak effects at 4 h after injection and nearly full recovery within 12 h. As with 0.1 mg/kg AM2389, subsequent injections of 0.03 mg/kg on days 2-5 of the daily injection regimen failed to produce hypothermic effects. The first injection of 10 mg/kg Δ⁹-THC also reduced average body temperature in all mice, with the peak effect, a decrease of 2.4±1.1°C, observed 4 h after injection. Baseline temperature was nearly recovered by 12 h after injection, and subsequent injections of Δ⁹-THC had no evident hypothermic effects. Morphine injections did not significantly alter temperature at any point during the study (these data are omitted from Figure 3 for clarity). Cessation of daily injections of AM2389 or Δ⁹-THC had no effect on body temperature.

Figure 3:

Body temperature in CD-1 mice (top panel) and C57/BL6 mice (bottom panel) during the daily injection and post-injection periods. Note that the horizontal axis is scaled to enlarge the days during which cannabinoid treatments had peak hypothermic effects. Data shown in the left portion of the graph present 4 h measurement intervals obtained on days 1 and 2 of the daily injection period; data shown in the center and right portion of the graph present data obtained at 12 h intervals. The stippled bar marks the separation of the daily injection and post-injection periods. The break in the graph for C57/BL6 mice injected with 0.1 mg/kg AM2389 represents an interruption in the daily injection pattern as detailed in the text. Other details as in Figure 2.

C57/BL6 mice.

Following determination of control values, different groups of mice received daily injections of 0.1 mg/kg AM2389 or 30 mg/kg Δ⁹-THC for five days, or twice-daily injections of 0.03 mg/kg AM2389, 10 mg/kg Δ⁹-THC, or 18 mg/kg Δ⁹-THC for five and one-half days. Significant hypothermic effects were evident in all groups following the first injection [F_(5,540)=95.0, p<0.001] (Figure 3, bottom panel). The response was greatest following 0.1 mg/kg AM2389; in these mice, temperature decreased, on average, by 8.7±0.4°C and the hypothermia lasted more than 24 h. To insure the health of this group of mice, the second injection was withheld until 48 h after the first injection. Following the second injections in all groups, only 0.1 mg/kg AM2389 and 30 mg/kg Δ⁹-THC still produced hypothermic effects and, by the third injection day, no hypothermic effects were evident in any group of mice. Cessation of daily injections after the 5-day regimen had no effect on body temperature in any group of C57/BL6 mice.

Radiotelemetry: locomotor activity

CD-1 mice.

Locomotor activity was significantly suppressed in all CD-1 mice following repeated injections of AM2389 or Δ⁹-THC. These effects persisted through the five days of drug treatment, although they varied depending on whether the lights were on (light phase) or off (dark phase). During the dark phase of the day, activity in mice that received 0.1 mg/kg AM2389 was suppressed to 20 to 40% of control values throughout the daily dosing period. During the light phase of the day, locomotor activity was suppressed only following the first two daily injections of 0.1 mg/kg AM2389; on the third, fourth and fifth days of daily dosing, activity during the light phase was only increased, though not significantly, above control values (Figure 4; full results of statistical analysis are presented in Supplemental Material Table 1). After daily injections of 0.1 mg/kg AM2389 stopped, locomotor activity increased during the dark, but not light, phase of post-injection days 1 and 2; no changes from control rates were evident on subsequent days.

Figure 4:

Locomotor activity counts in CD-1 mice. Locomotor activity counts were compiled in 12 h bins in accordance with the 12 hr light/dark cycle of the day. The upper dotted horizontal lines represent the average control locomotor activity counts during the dark phase of the day; the lower dotted horizontal lines represent the average control locomotor activity counts during the light phase of the day. Data for saline-treated animals are replotted from Figure 2. Vertical axis: Summed 12 h locomotor activity counts. Other details as in Figure 2.

Mice injected twice daily with either 0.03 mg/kg AM2389 or 10 mg/kg Δ⁹-THC exhibited nearly identical changes in locomotor activity that were qualitatively similar to effects in mice treated once daily with 0.1 mg/kg AM2389 (Figure 4). Thus, in AM2389 treated mice, locomotor activity was decreased to approximately 35% of control rates during the first day and, thereafter, was decreased to 25% to 70% of control values during the dark phase of Days 2-5. In contrast, activity during the light phase was significantly decreased only on the first day of injections and returned fully to control values by the third day of injections. Activity was decreased over the first 12 h following the last injection of 0.03 mg/kg AM2389 (during the dark phase), but more than doubled during the subsequent light phase, i.e., 12-24 h later, and was similar to dark-phase control values (shown in the light bar over Drug Day 5 in Figure 4). Locomotor activity counts returned to control values on subsequent post-injection days. Treatment with 10 mg/kg Δ⁹-THC reduced locomotor activity counts to approximately 50% of control values during the dark phase of the day throughout the daily dosing period. However, locomotor activity did not differ significantly from control values when the lights were on. Daily values for locomotor activity in Δ⁹-THC-treated mice remained at control levels during both phases of the day throughout the post-injection period.

A group of mice that received 10 mg/kg morphine twice daily was included to provide comparison data with a psychoactive drug that does not act directly via CB1 mechanisms. The first injection of 10 mg/kg morphine slightly increased locomotor activity above control values, but subsequent injections had no discernible effects (Figure 4). During the post-injection period, and notwithstanding a significant linear trend [F_(1,25)=14.5, p<0.001] for changes in locomotor activity during the light phase of the day, no significant differences in locomotor activity counts were identified.

C57/BL6 mice.

Locomotor activity was suppressed in all mice following initial injections of 0.03 or 0.1 mg/kg AM2389, although statistical significance was achieved only during the dark phase (Figure 5; full results of statistical analysis are presented in Supplemental Material Table 2). Over days 2 through 5 of the daily injection period, dark-phase locomotor activity gradually recovered to control levels in mice injected twice daily with 0.03 mg/kg AM2389 but continued to be suppressed following single daily injections of 0.1 mg/kg AM2389. In both groups of AM2389-teated mice, activity during the light phase of the day was at or above control values by the second drug day. Following the last AM2389 injection, increases in locomotor activity were seen in all mice, however the magnitude and time of greatest increase varied between groups. Thus, the last injection of 0.03 mg/kg AM2389 did not have any activity-decreasing effects over the first 12 h (during the dark phase) and, as in the CD-1 mice, activity was almost doubled at 12 to 24 h after the last injection (shown in the light bar over Drug Day 5 in Figure 5). In this group of mice, locomotor activity also was increased during the dark phase of post-injection days 2 and 3. However, locomotor activity was significantly increased only during the dark phase of post-injection day 1 in mice that had received once daily injections of 0.1 mg/kg AM2389.

Figure 5:

Locomotor activity counts in C57/BL6 mice. Other details as in Figures 2 and 4.

Injections of Δ⁹-THC initially and dose-dependently suppressed locomotor activity to 30% to 70% below control values during the dark phase; these effects diminished but did not fully dissipate over following treatment days, albeit 10 mg/kg Δ⁹-THC significantly suppressed locomotor activity only following the first injection. In contrast, regardless of dose, Δ⁹-THC had negligible effects on locomotor activity during the light phase throughout the daily injection period. During the post-injection period, increases in locomotor activity appeared in C57/BL6 mice that had received one injection per day of 30 mg/kg, or two injections per day of 18 mg/kg Δ⁹-THC, however the effects were smaller, and appeared later, than following either dose of AM2389. Thus, locomotor activity in mice treated with 30 mg/kg Δ⁹-THC increased to 140% of control values during the dark phase of post-injection day 4, but otherwise did not differ significantly from control values. Small increases in activity also were observed in mice injected twice daily with 18 mg/kg Δ⁹-THC 12 to 24 h after the last injection, during the light phase of post-injection day 2, and during the dark phase of post-injection day 4. These effects, ranging from 124% to 147% of control values, were relatively smaller than those seen in the AM2389-treated mice. No significant changes in locomotor activity were observed during the post-injection period in mice injected twice daily with 10 mg/kg Δ⁹-THC,

Peak increases in locomotor activity.

To determine whether maximum increases in locomotor activity emerged at different times among subjects, the greatest increase in activity, relative to control values and regardless of whether this occurred during the light or dark phase of the day, was plotted as a percentage of phase-appropriate control activity for each subject (Figure 6, top panel). The 12 h period during the post-injection phase in which the maximum activity occurred is presented in the bottom panel of Figure 6. Overall, the magnitude of locomotor increases was greatest in AM2389-treated mice of either strain, though with greater variability than in vehicle or Δ⁹-THC-treated mice. With only two exceptions, the largest increases in locomotor activity in AM2389-treated subjects consistently occurred within 48 h after stopping the daily injections. In contrast, the timing of peak activity varied widely, across the full five day recording period, among vehicle- and Δ⁹-THC-treated mice.

Figure 6:

Top panel: peak increases in locomotor activity during the post injection period. Points present maximum increases in individual mice, expressed as a percent of control activity; horizontal bars indicate group medians. Bottom panel: 12 h time period, relative to when the last injection was administered, when the maximum increase in activity was recorded. * p<0.05, **p<0.01, ***p<0.001 compared to vehicle treated group.

Discussion

The increases in paw tremors in AM 2389-treated mice in the present study indicate that, under some conditions, the discontinuation of daily cannabinoid injections alone may lead to changes in overt behavior. This finding stands apart from most previous studies of cannabinoid withdrawal in which paw tremors were increased in cannabinoid-treated mice only after administration of rimonabant (Cook et al., 1998; Lichtman et al., 2001). The major methodological differences between the present and previous studies are that we used an agonist shown previously to have higher efficacy than Δ⁹-THC and that effects were evaluated 24 h after discontinuation. Previous studies, which were more focused on producing evidence of precipitated cannabinoid withdrawal, have commonly reported effects measured within 4 h after the last agonist injection (Castañé et al., 2004; Hutcheson et al., 1998). As most cannabinoid drugs have a long duration of action and sometimes do not reach their peak effects until several hours after injection (Davis et al., 1973; Nikas et al., 2015; Sharma et al., 2013; Tai et al., 2015), it seems unlikely that any signs of withdrawal would emerge at early time points, e.g., 4 h after the last injection. This pharmacokinetic consideration informed the decision to observe behaviors in the present study at the later time of 24 h and, as well, to use telemetry, which captures behavioral changes continuously, as a means of precluding the need to preselect times of observation.

The acute hypothermic and locomotor-suppressant effects of cannabinoids are well known and are two of the often-used tetrad of behavioral effects (in addition to catalepsy and antinociception) used to characterize drugs as cannabinoid-like (Martin et al., 1991; Wiley et al., 2012). Tolerance to these effects also has been reported, including observations that, as in the present studies, a more rapid tolerance occurs to the hypothermic than locomotor effects of cannabinoids (Fan et al., 1994; Tai et al., 2015). The present data provide further insight into the development of cannabinoid tolerance, showing that tolerance to the effects of cannabinoids on locomotor activity differs with the diurnal cycle. The animals in groups treated with 0.1 mg/kg AM2389 or 30 mg/kg Δ⁹-THC received only one injection per day, always administered at the start of the dark portion of the day. Thus, the locomotor suppressant effects of the drugs might be expected to wane 12 to 24 h after injection, and the decreased effect during the light portion of the cycle might be most readily explained by metabolism of the drugs. However, the continued suppressant effects of twice daily treatment with 0.03 mg/kg AM2389 and 10 or 18 mg/kg Δ⁹-THC during only the dark hours of the day were unexpected and clearly independent of any metabolic considerations. It is unclear how to explain this pattern of sustained effects of these cannabinoid regimens. Perhaps, diurnal fluctuation in endocannabinoid levels can provide some insight into these dramatic circadian-related results. For example, endocannabinoid concentrations in rat brain have been reported to vary over the course of a day, generally increasing in cerebrospinal fluid during the light portion of the day, and increasing in particular brain regions (pons, hippocampus, and hypothalamus) during the dark phase of the day (Murillo-Rodriguez et al., 2006). A pilot study in human subjects also has suggested a diurnal, though opposite, pattern of fluctuation in plasma anandamide concentrations; i.e., anandamide concentrations are lower during the night than during the day (Vaughn et al., 2010). While these are intriguing and possibly relevant observations, the effects of the cannabinoid treatment regimens on endocannabinoid concentrations or their diurnal variation are presently unknown. Thus, conjecture about the role of their interplay in the results reported here must be considered purely speculative.

Based upon the well-documented characterization of opioid withdrawal, it was initially hypothesized that abstinence-related signs of cannabinoid withdrawal would be opposite to the acute effects of cannabinoid injection. For example, opioids induce constipation and pupillary constriction, and two hallmark signs of opioid abstinence are increased defecation and pupillary dilation. Cannabinoids produce decreases in body temperature; thus, it was anticipated that temperature would increase upon treatment cessation. However, there was no evidence of hyperthermia after stopping the daily cannabinoid treatment. The lack of temperature changes during cannabinoid abstinence may suggest that, as with opioids, responses to drug treatment that exhibit rapid tolerance are less readily perturbed by the cessation of chronic drug treatment (Colpaert, 1996). Alternatively, withdrawal effects on body temperature simply might be less evident due to a relatively rapid and homeostatic re-adjustment of thermoregulatory mechanisms during the drawn out and variable time course of cannabinoid withdrawal.

Notwithstanding the absence of withdrawal effects on body temperature, increases in locomotor activity—a clear sign of withdrawal—were observed in mice following discontinuation of daily cannabinoid treatment, although varying in magnitude and timing depending on the cannabinoid treatment. These findings receive some support from two recent studies suggesting that locomotor activity also may increase following discontinuation of either daily CP 55,940 injections in mice or prolonged periods of daily Δ⁹-THC injections in monkeys (Navarrete et al., 2018; Wilkerson et al., 2019). However, interpretation of those data are complicated by control considerations. For example, although Navarette and colleagues reported approximately 50% increases in locomotor activity in a single 15 minute session, 12 h following the last injection of CP 55,940 and relative to activity in vehicle-treated mice, locomotor activity in the control group decreased over the course of the study, making it difficult to evaluate the results. Similarly, activity increased in rhesus monkeys over a period of 10 days after discontinuing Δ⁹-THC injections; however, no non-drug control values were provided—again complicating interpretation of these results (Wilkerson et al., 2019). In the present studies, locomotor activity increased in two strains of mice following five-day exposures to a high-efficacy cannabinoid agonist, AM2389, and these effects were robust whether compared to pre-drug baseline values or to activity in vehicle-treated animals. Moreover, this effect was not a generalized effect of drug discontinuation as daily treatment with morphine was followed by, if anything, a slight and transient decrease in locomotor activity. In conjunction, the present and previous results show that increases in locomotor activity following the discontinuation of cannabinoid treatment can serve as a clear sign of cannabinoid withdrawal and, thus, dependence.

The decision to use AM2389 in these studies was based on the reportedly higher efficacy of AM2389 than Δ⁹-THC (Paronis et al., 2012) and the proposition that, as with other drug classes (Bergman et al., 2000), administration of a higher efficacy cannabinoid would be more likely to result in observable signs of dependence. Thus, the lesser effects on locomotor activity following cessation of daily Δ⁹-THC-treatment, in comparison to those of AM2389 treatment, were not unexpected. Once the temporal parameters for the measurement of withdrawal were established with AM2389, subsequent studies established a rough equivalence between the doses of 0.03 mg/kg AM2389 and 10 mg/kg Δ⁹-THC based on the magnitude and duration of their acute effects in telemetry measures as well as their quantitatively similar effects in the paw tremor assay. However, this equivalence was limited as, notwithstanding increases in dosage, the magnitude of effect produced by Δ⁹-THC withdrawal did not increase further. Similarly, although a role for genetic differences in the effects of drugs, including cannabinoids, has been reported previously (Elmer et al., 1998; Parks et al., 2020; Stolerman et al., 1999), the magnitude of effect was not altered here by a change in mouse strain from CD-1 to C57/Bl6. Although these observations are consistent with the idea that effects, including withdrawal effects, of a partial agonist like Δ⁹-THC may plateau below those of a higher-efficacy agonist like AM2389, it is possible that extending the duration of daily treatment with Δ⁹-THC might have increased the magnitude of the withdrawal effect observed in the present studies. Cannabis withdrawal in humans was first identified in subjects that had experienced multiple periods of daily marijuana exposure over the course of years (Budney et al., 2003; Kouri and Pope, 2000), which contrasts dramatically with the limited exposure (often, a week or so) in most animal studies (e.g., (Aceto et al., 1996; Cook et al., 1998). From this perspective, five days of exposure in the present studies is not an overly long regimen. While it may be difficult to define functionally equivalent chronic regimens in human subjects and mice, it is must be considered that effects of terminating Δ⁹-THC, or AM2389, injections might have had a greater impact on locomotor activity in mice treated for longer than one week.

In conclusion, the present data indicate that mice can show changes in behavior, notably increases in paw tremor and in locomotor activity, that may be considered as signs of withdrawal following cessation of daily cannabinoid treatment. Similar to reports of cannabis withdrawal in humans, the effects seen are mild, not debilitating and also are relatively slow to emerge, 12 to 96 h after cessation of drug treatment. Another symptom associated with cannabinoid withdrawal in human subjects is sleep disruption, which we have explored in other studies [see (Missig et al., 2021) in this volume]. Consistent with other drug classes, the effects produced by a higher efficacy agonist, AM2389, were of greater magnitude than the effects produced by daily administration of Δ⁹-THC, a cannabinoid CB1-partial agonist. Thus, high efficacy cannabinoid agonists may provide better tools with which to evaluate the development and expression of cannabinoid dependence and withdrawal.