Pharmacokinetic and Pharmacodynamic Characterization of Tetrahydrocannabinol-Induced Cannabinoid Dependence After Chronic Passive Cannabis Smoke Exposure in Rats

Abstract

Introduction: Cannabis is the most widely used illicit drug in the US, and cannabis use among young adults continues to rise. Previous studies have shown that chronic administration of delta 9-tetrahydrocannabinol (THC), the main psychoactive component of cannabis, induces dependence in animal models. Because smoking is the most frequent route of THC self-administration, it is critical to investigate the effects of cannabis smoke inhalation. The goal of the current study was to develop a rat model to characterize the pharmacokinetics (PKs) of THC after cannabis smoke inhalation, and to determine if chronic cannabis smoke inhalation leads to the development of cannabis dependence.

Materials and Methods: For the PK study, male Wistar rats were administered THC intravenously (1 mg/kg) or exposed to smoke from 5 or 10 sequentially smoked cannabis cigarettes (5.3% THC) in an automated smoking machine. Plasma samples were collected from 10 min to 10 hours post smoke exposure (or intravenous administration) and analyzed using liquid chromatography–mass spectrometry to characterize the PK of THC. A three-compartment PK model was used to characterize the PKs. In a separate study, three groups of male Wistar rats were trained in an intracranial self-stimulation (ICSS) procedure, and exposed to smoke from burning 5 or 10 cannabis cigarettes (or clean air control conditions), 5 days/week for 4 weeks.

Discussion and Conclusions: Across exposure days, the change from baseline in ICSS thresholds for cannabis smoke-exposed groups was significantly lower and response latencies were significantly faster in the cannabis smoke-exposed groups compared to controls, suggesting that chronic cannabis smoke exposure has rewarding properties. Acute administration of the CB1 receptor antagonist rimonabant (0.3, 1.0, 3.0 mg/kg) induced a dose-dependent increase in ICSS thresholds in the smoke-exposed rats, suggestive of dependence and withdrawal. Finally, an effect compartment PK-pharmacodynamic model was used to describe the relationship between THC concentrations and changes in ICSS thresholds after cannabis smoke exposure.

Introduction

The use of cannabis and synthetic cannabinoids such as “spice” is on the rise. It is estimated that ∼24 million people aged ≥12 used cannabis on a regular basis in the United States in 2017, and 4.1 million people aged ≥12 in the United States were treated for cannabis use disorder in 2017.¹ Orally administered cannabis and cannabis products that are prescribed for medicinal use have also been shown to produce dependence.² Hence, the endocannabinoid system can be considered a double-edged sword when viewed as a potential target in the process of drug discovery. Cannabis toxicity can lead to hallucinations, psychosis, and suicidal symptoms, and the American College of Clinical Pharmacology recently called for United States Food & Drug Administration oversight of orally consumed cannabis products such as cookies and candies laced with cannabis.³ Such data highlight the unmet need to understand the underlying mechanisms of cannabinoid dependence and abuse potential.

The cannabis plant contains >60 cannabinoids, of which delta-9 tetrahydrocannabinol (THC) is primarily responsible for its psychoactive effects. Smoking cannabis is the preferred means of THC consumption, and the amount of plant material burnt during smoking can be directly linked to the amount of THC consumed. THC is a lipophilic compound that gets readily distributed into excessively perfused organs such as brain and spinal cord, and eventually gets stored in adipose tissue. THC is mainly metabolized in the liver by cytochrome P450 CYP2C9 and CYP3A4 enzymes; however, the rate limiting step in THC metabolism and elimination is back-diffusion of THC from fatty tissues into the blood, which contributes to its long half-life. The major THC metabolite is 11-hydroxy-tetrahydrocannabinol, which is an active metabolite. It is further metabolized into a water-soluble glucuronide conjugate before it gets eliminated. Most THC is metabolized, and <5% is eliminated unchanged. THC has a pharmacokinetic (PK) half-life of 1.3–7.3 h in rats.^4–6

As many effects of cannabinoids in humans are subjective and influenced by factors such as experiences and expectancies, various animal models have been used to characterize the behavioral effects of cannabinoids. Other advantages of laboratory animals include experimental access to brain and other tissues that cannot be readily achieved in human subjects. Self-administration studies in rodents and nonhuman primates have suggested that THC, like other drugs of abuse, has rewarding/reinforcing effects.^7–11 These findings led to research to test the effects of cannabinoid administration under different animal paradigms such as intracranial self-administration, intracranial self-stimulation (ICSS), and conditioned place preference to evaluate and understand different behavior exhibited by the animals under various conditions.¹² In particular, ICSS studies show that THC administration lowers the current threshold required to maintain ICSS behavior, suggestive of an increase in the sensitivity of the neural substrates of reward and, by extension, indicative of positive hedonic effects of THC.¹³ In contrast, withdrawal from THC leads to an increase in ICSS thresholds, which is interpreted as a dysphoric state.¹⁴ The ICSS is an established and validated method to investigate the development of dependence in rats exposed to tobacco smoke, psychostimulants, opioids, and other drugs of abuse.^15–19 Importantly, the effects of THC on ICSS threshold are blocked by antagonists at cannabinoid 1 receptors (CB1Rs), indicating a critical role for these receptors in THC-mediated reward.^20–24

Dose–effect relationships have been characterized in animal models after parenteral THC administration, but only a few studies have investigated effects of THC using a smoked route of administration.^21,25,26 Inhalation studies suggest that repeated cannabis smoke exposure can induce dependence, as indicated by the presence of somatic withdrawal signs after CB1R blockade (i.e., precipitated withdrawal).^21,26,27 Specifically, in a previous study in rats,²⁷ we showed that acute administration of the CB1 antagonist rimonabant (SR141716A) induced somatic withdrawal signs, including forepaw fluttering, grooming, ptosis, and shakes in cannabis smoke-exposed rats after 2 weeks of daily smoke exposure. In addition to these studies, an assessment of the effects of second-hand cannabis smoke exposure in nonsmokers concluded that its subjective effects are directly related to the amount of smoke exposure, and that passive exposure to high levels of cannabis smoke can mimic active smoking.²⁸

The purpose of this study was to use a rat model to predict the PKs of THC, and to use this model to assess THC dependence and characterize exposure–response relationships after chronic exposure to cannabis smoke containing THC.

Materials and Methods

Subjects

Male Wistar rats prepared with jugular vein catheters (n=10; Envigo) and weighing 300–350 g upon arrival were used to evaluate THC PKs. Male Wistar rats (n=36), weighing 175–275 g upon arrival, were obtained from Charles River (Raleigh, NC) for the dependence study. All rats were housed (2 per cage) under a reversed light–dark cycle (lights off 8AM–8PM) in the AALAS-approved vivarium in the McKnight Brain Institute at the University of Florida (UF). Animal procedures were approved by the UF Institutional Animal Care and Use Committee and followed NIH guidelines. Rats were left undisturbed for at least 1 week after arrival, before initiation of experimental procedures, and they had free access to food and water at all times.

Drugs

Cannabis cigarettes (5.3% THC, <0.001% CBD) and THC were kindly provided by the NIDA Drug Supply Program. The cannabis cigarettes were stored at −20°C. Approximately 24 h before being used, the cigarettes were removed from the freezer and placed in an airtight humidity chamber with a small dish containing saline and kept at room temperature (22–24°C). The cigarettes were used within 1 h of being removed from the humidity chamber.

The THC was dissolved in a mixture of ethanol (30% v/v) and propylene-glycol-400 (40% v/v), and the volume was made up with saline to obtain the desired concentration of 1 mg/mL. Rimonabant hydrochloride (SR 141716A) was purchased from Tocris Bioscience (Bristol, United Kingdom). Rimonabant was dissolved in a vehicle of Tween 80 (5% volume/volume, v/v), DMSO (20% v/v), and sterile saline (75% v/v). Freshly prepared vehicle and rimonabant were administered at 3 different doses (0.3, 1.0, and 3.0 mg/kg). These doses were based on a previous study in which we used rimonabant to evaluate precipitated withdrawal in cannabis smoke-exposed rats.²⁷

Drug analysis

Plasma THC and rimonabant levels were determined using a method that was developed and validated in accordance with the USFDA guidelines.²⁹ In brief, calibration curve samples from 0.1 to 100 ng/mL and quality control samples (lower limit of quantification, lower quality control (LLOQ, LQC), middle quality controls (M1QC, M2QC), and high-quality control (HQC)) were prepared on the day of the analysis. Fifty microliters of plasma was then spiked with 100 μL of acetonitrile containing Internal Standard (IS). To this mixture, 1000 μL of hexane was added and 900 μL was separated out after centrifugation for 10 min. The supernatant was dried and reconstituted in 100 μL of methanol. Twenty microliters of the sample was injected in the Shimadzu UFLC-Nexera X2 system (Kyoto, Japan) and analyzed using ABSCIEX API 5500 QTRAP (ABSCiex, Framingham, MA). Ten millimolars ammonium formate buffer containing 0.1% formic acid and methanol as the aqueous and organic mobile phases at a flow rate of 1 mL/min (90:10) was used to achieve chromatographic separation using isocratic elution.³⁰

ICSS apparatus

Testing for ICSS was conducted in 12 computer-controlled operant chambers (30.5×30×17 cm; Med Associates, Georgia, VT), each housed in a sound-attenuating cabinet (Med Associates, Georgia, VT). Each chamber had a metal grid floor and a metal wheel (5 cm wide) centered on a side wall. A detector was attached to the response wheel and recorded every 90 degrees of rotation. Brain stimulation was delivered by constant current stimulators (Model 1200C; Stimtek, Acton, MA). Rats were connected to the stimulation circuit through bipolar leads (Plastics One, Roanoke, VA) attached to commutators (Model SL2C Plastics One).^17,31

Surgery for electrode implantations

Rats were anesthetized with an isoflurane and oxygen vapor mixture (2% isoflurane) and placed in a stereotaxic frame (David Kopf Instruments, Tujunga, CA) with the incisor bar set 3.3 mm below the interaural line (flat skull). The electrodes (11 mm in length; Plastics One) were implanted in the medial forebrain bundle with the incisor bar 5 mm (angled skull) above the interaural line (anterior–posterior −0.5 mm, medial–lateral ±1.7 mm, dorsal–ventral −8.3 mm from dura as in our previous work³²). After the intracranial surgeries, the rats were allowed to recover for ∼1 week before starting behavioral testing.

ICSS procedure

Rats were trained on a modified discrete-trial ICSS procedure as described previously.^14,31 Rats were trained initially to turn the wheel on a fixed ratio 1 (FR1) schedule of reinforcement. Each quarter turn resulted in delivery of a 0.5-sec train of 0.1 msec cathodal square-wave pulses at a frequency of 100 Hz. Upon obtaining successful acquisition of responding, the rats were trained on a discrete-trial current threshold procedure. Each trial started with the delivery of a noncontingent stimulus, followed by a 7.5-sec response window during which a response from the rat (a quarter turn of the wheel) resulted in a second contingent stimulus of the same intensity. A response during this 7.5-sec response window was considered a “positive response.” The intertrial interval, which followed either a positive response or termination of the response window, had an average duration of 10 sec (7.5–12.5 sec). A test session consisted of four alternating series of descending and ascending current intensities, which typically lasted 45 min and provided two variables: brain reward threshold, defined as the minimum current (μA) necessary to elicit a positive response, and response latency, defined as the time between the noncontingent stimulus and a response for contingent stimulation.

Smoke exposure

Rats were exposed to smoke from burning either 5 or 10 sequentially smoked cannabis cigarettes, each weighing ∼0.9 g. Smoke exposure was conducted in a Teague Enterprises TE-10 Smoking Machine (Davis, CA) as described previously.^27,30,33,34 The total duration of smoke exposure was 50 or 100 min, as each cigarette took ∼10 min to burn completely. Ten puffs (2 sec per puff, 1 min interpuff interval) were obtained from each cigarette. Mainstream smoke from each puff was directed into the exposure chamber, in which rats were placed in standard rat home cages (four cages in total). For the PK studies, rats were housed individually inside the cages, and for the ICSS studies two to four rats were placed inside each cage. The air inside the exposure chamber was monitored for carbon monoxide (CO) and total suspended particulate (TSP) levels. CO levels were monitored using a Monoxor III (Bacharach, New Kensington, PA), which can measure between 0 and 2000 PPM. TSP was measured by passing the smoke in the exposure chambers through a preweighed filter (Pallflex Emfab Filter; Pall Corporation, Port Washington, NY) for 1–2 min. The TSP per cubic meter was determined by dividing the weight increase of the filter by the volume of the airflow through the filter.

Experimental design

PK studies

Rats were randomly assigned to the three treatments (1 mg/kg THC [i.v.], 5 cigarettes, or 10 cigarettes, n=3–4 per group). For the cigarette groups, rats were exposed to smoke from burning 5 or 10 sequentially smoked cannabis cigarettes. Immediately after the last puff, rats were removed from the exposure chamber for blood sample collection. Each sample withdrawn was replaced by an equal volume of heparinized 0.9% saline (20 IU heparin/mL). Serial blood samples were collected in heparinized microtubes at 10, 20, 40, 60, 120, 240, 480, and 600 min after drug administration. For i.v. administration of THC, an additional 24 h time point was collected. Plasma (200 μL) was separated from blood by centrifugation at 10,000 rpm for 5 min and frozen at −80°C until analysis.

ICSS studies

The goal of this experiment was to determine if chronic exposure to cannabis smoke has rewarding effects and leads to development of dependence, as assessed by changes from baseline in ICSS thresholds and response latencies. Rats (n=36) were divided into three groups: air exposed (n=9), 5 cigarettes (n=15), and 10 cigarettes (n=12). The rats underwent surgery to implant ICSS electrodes, and were trained on the ICSS procedure until brain reward thresholds were stable before the start of smoke exposure (see Fig. 1 for experiment design). During weeks 1 and 2, rats were exposed to cannabis smoke daily (5 days/week) followed by ICSS testing (exposure and ICSS testing were conducted between 900 and 1200 h). Rats were removed from the exposure chamber immediately after the final cigarette, and were placed inside the ICSS chamber 45–60 min later. Blood samples were collected from the dorsal pedal vein 60 min postexposure on exposure days 4 and 9 and stored for subsequent analysis. The effects of the CB1 antagonist rimonabant on ICSS thresholds and response latencies were evaluated during weeks 3 and 4. These sessions were conducted in a manner identical to those in weeks 1 and 2 (5 days/week smoke exposure followed by ICSS testing), except that rimonabant or vehicle was administered 45 min after the last puff, and rats were placed in the ICSS chambers 10 min later. Rimonabant (0, 0.3, 1.0, and 3.0 mg/kg) was administered i.p. according to a randomized, Latin square design, with at least a 3-day washout period between successive injections. This schedule allowed sufficient time for brain reward thresholds to return to baseline levels as the plasma half-life of rimonabant is 3.7 h as suggested by our previous work.³⁰ Rats continued to undergo smoke exposure followed by ICSS testing on washout days.

FIG. 1.

The experimental design for the ICSS and cannabis smoke exposure experiment. Upon obtaining a stable baseline, rats were exposed to cannabis smoke or clean air control conditions for four consecutive weeks (5 days/week, Monday–Friday). During weeks 3 and 4, rats received i.p. injections of different doses of rimonabant (0, 0.3, 1.0, and 3.0 mg/kg) using a randomized, Latin square design. Injections took place on Tuesdays and Fridays. ICSS, intracranial self-stimulation.

Dose metrics

Dose calculations for inhalation studies are usually not straightforward. Some of the sources of variability include particle size, location of the deposited fraction, and mode of exposure. A theoretical dose estimate can be obtained using various approaches. Only a fraction of the total amount of THC present is available for inhalation, as some is lost due to pyrolysis and more is lost as sidestream smoke. Previously published data suggest that ∼50% of the THC is lost due to pyrolysis, and that only 30–45% of the remaining drug is available for inhalation after accounting for loss due to sidestream smoke.³⁵ Based on these assumptions, we calculated the concentration of THC that would be present in the exposure chamber (264 L) for the animals to inhale (four cages inside the exposure chamber). After obtaining the concentration of THC inside the exposure chamber, we calculated the lung deposited dose using the following Equations (1) and (2)³⁶:

where C is the concentration of the drug (mg/L), T is the total time of exposure (min), RMV is the respiratory mean volume (L/min), DF is the fraction deposited, and BW is the body weight (kg).

The delivered dose of THC to each animal was calculated to be 0.05 and 0.2 mg/kg for 50 and 100 min of continuous smoke exposure, where 10% of the total drug was assumed to be the deposition factor.³⁶ Similar calculations have been used to determine the dose of THC delivered through cannabis smoke inhalation in mice.²⁶

Pharmacokinetic-pharmacodynamic modeling

Noncompartment analysis

Noncompartment analysis (NCA) was implemented in Phoenix WinNonlin. When analyzing PK data, the first step is to perform a NCA. The purpose of using NCA is to determine the degree of exposure after administration of a drug (such as total area under exposure curve [AUC]), and the drug's associated PK parameters, such as clearance, elimination half-life, T_max, and C_max. This analysis has the advantage that it requires fewer assumptions compared with standard nonlinear regression modeling or compartment modeling.

Pharmacokinetic–pharmacodynamic modeling

The objective of PK–pharmacodynamic (PD) modeling was to capture the change from baseline values (weeks 1 and 2) in the ICSS thresholds, which would aid in understanding the effect of dose (5 cigarettes vs. 10 cigarettes) and exposure duration on the extent of ICSS-induced hedonia experienced by the animals. Since we were more interested in the basic model structure, mean data could suggest the true profile pattern for changes in ICSS thresholds over time after exposure to cannabis smoke. The mean change (3-day averages) from baseline (day 1) in ICSS thresholds over time for the control group (air exposed) was subtracted from the mean change in ICSS thresholds for the cannabis smoke-exposed groups. This procedure helped facilitate understanding of the true effect of THC after exposure to cannabis smoke. To relate the changes in THC concentrations to changes in ICSS thresholds, various structurally different PK–PD models were evaluated. R-Programming software (3.4.2) and NONMEM software version 7.0 or greater (ICON, Dublin 18, Ireland) were used for model fitting and goodness-of-fit plots. Objective function values were evaluated during model development. A visual predictive check was performed for internal validation. Usually, there is no observable delay between inhalation of cannabis smoke and plasma concentrations, hence inhalation was modeled as bolus administration in the observation compartment, which is consistent with previously published work.³⁷ Therefore, a three-compartment PK model was used to characterize the PKs of THC after intravenous administration and cannabis smoke inhalation. A proportional error model was used to explain residual error in the PKs of THC. An effect compartment model was used to describe the relationship between changes in ICSS threshold and plasma drug concentration. Drug distribution to the effect site was characterized on the basis of a hypothetical effect compartment model. An additive error model was used to describe the residual variability in the final PK–PD model.

The change in ICSS thresholds that is induced by THC after cannabis smoke exposure is characterized using a sigmoid E_max model using Equations (3) and (4):

In these equations, E₀ represents the baseline response in the absence of the drug and is set to 100, E_max is the maximum activity of the drug, and EC₅₀ is the concentration required to achieve half maximal effect. C_e is the concentration in the hypothetical effect compartment that is responsible for the pharmacological action of the drug. When the concentrations of the drug are relatively low compared with the EC₅₀ value of the drug, a linear model can be used to describe the pharmacological effect in that given dose range [Equation (6)]. γ is the parameter used to describe the steepness of the pharmacological response curve and is called the Hill factor.

Statistical analysis

Absolute brain reward thresholds and response latencies before the onset of the smoke exposure sessions were compared with one-way ANOVA. The effects of cannabis smoke and number of days of exposure on change from baseline (day 1) for ICSS thresholds and response latencies were analyzed using two-factor ANOVA, with exposure condition (air exposed, 5 cigarettes, 10 cigarettes) as a between-subjects factor and day of exposure as a within-subjects factor. The effects of cannabis smoke and rimonabant on mean change from baseline for ICSS thresholds and response latencies were analyzed using two-factor repeated-measures ANOVA, with exposure condition (air exposed, 5 cigarettes, 10 cigarettes) as a between-subjects factor and rimonabant dose (0, 0.3, 1.0, and 3.0 mg/kg) as a within-subjects factor. ICSS thresholds in the session immediately before rimonabant administration were used as baseline values. For all statistical analyses, significant main effects and interactions in the ANOVA were followed by Tukey's HSD post hoc tests. p Values ≤0.05 were considered significant. Data were analyzed with R-programming 3.4.2 in R-studio.

Results

PKs of THC after cannabis smoke exposure and intravenous injection of 1.0 mg/kg THC

Upon obtaining the PK profile, a noncompartmental analysis was performed to estimate the PK parameters for both the routes of administration. Results from the NCA are presented in Table 1. The mean maximum concentrations (C_max) of THC after smoke exposure were 18.2 and 29.6 ng/mL for 5 and 10 cigarettes, respectively, with a terminal half-life of 3.7 h. Total AUC_0−t (h×μg/L) of THC after 5 and 10 cannabis cigarettes and 1 mg/kg THC was 12.2, 38.2, and 276, respectively. The calculated r² for dose versus AUC was 0.99, suggesting linear PKs in these dose ranges. The clearance was calculated to be 1.2 L/h and the volume of distribution was 3.5 L.

Table 1.

Summary of Exposure Parameters of THC After Intravenous Administration and Cannabis Smoke Inhalation

PK parameter	THC IV	THC inhalation	THC inhalation
Dose (mg/kg)	1	0.05	0.2
Half-life (h)	1.90±0.3	3.00±0.8	3.30±0.7
CL (L/h)	1.13±0.0	1.10±0.1	1.40±0.1
VolumeSS (L)	3.09±0.5	3.30±1.0	4.10±0.1
AUC0−t (h×μg/L)	259.10±4.5	12.40±1.7	38.20±3.8
AUCinf (h×μg/L)	263.05±3.9	13.10±1.6	41.40±4.7

PK, pharmacokinetic; THC, delta 9-tetrahydrocannabinol.

Effects of chronic exposure to cannabis smoke on ICSS

There were no differences among the three exposure conditions (air exposed, 5 cigarettes/session, and 10 cigarettes/session) in mean ICSS threshold and response latency on the first day of cannabis smoke exposure (one-factor ANOVA, F_(2,31)=2.33, p=0.114 and F_(2,31)=0.464, p=0.63, respectively) as shown in Table 2. A two-factor repeated-measures ANOVA comparing the change from baseline in ICSS thresholds across days revealed main effects of both day (F_(25,794)=2.754, p<0.001) and exposure conditions (F_(2,794)=3.372, p=0.0348), indicating that exposure condition significantly affected ICSS thresholds. Post hoc analyses showed that on days 4, 6, and 8 after starting the smoke exposure, the changes from baseline for the cannabis smoke exposure groups were significantly greater than the change from baseline for the control (Fig. 2). A similar analysis conducted on the change in ICSS response latencies from baseline revealed main effects of both day (F_(25,796)=2.339, p=0.0264) and exposure conditions (F_(2,796)=18.629, p<0.001), with post hoc analyses revealing significant effects on days 11, 18, 20, and 22 (Fig. 3). These changes in ICSS threshold and response latencies suggest that chronic inhalation of cannabis smoke enhances brain reward function, indicating that chronic cannabis smoke exposure has rewarding effects. The plasma THC levels after 5 and 10 cigarettes continuous smoke exposure on days 4 and 9 are presented in Table 3.

Table 2.

Reward Thresholds and Response Latencies for Treatments Prior to Starting Smoke Exposure

Day	Treatment	Absolute ICSS threshold (μA)	Latency to first response (msec)
1 (Baseline)	Air exposed	129.70	3.29
	5 Cigarettes	103.38	3.19
	10 Cigarettes	125.16	3.24

ICSS, intracranial self-stimulation.

FIG. 2.

The post hoc comparisons for changes (% change from baseline) in ICSS threshold (y-axis) for each exposure group (x-axis) on days 4, 6 and 8 of exposure compared to baseline ICSS thresholds. Baseline was ICSS thresholds on day 1 prior to starting smoke exposure. The horizontal line within the box indicates the median, boundaries of the box indicate the 25th and 75th-percentile, and the whiskers indicate the highest and lowest values of the results.

FIG. 3.

The post hoc comparisons for changes (% change from baseline) in ICSS latencies (y-axis) between exposure groups (x-axis) for days 11, 18, 20 and 22 of exposure compared to baseline ICSS latencies. Baseline was ICSS latencies on day 1 prior to starting smoke exposure. The horizontal line within the box indicates the median, boundaries of the box indicate the 25th and 75th-percentile, and the whiskers indicate the highest and lowest values of the results.

Table 3.

Plasma THC Concentration After 5 and 10 Cigarettes Smoke Exposure on Days 4 and 9

Dose (cigarettes)	Duration (min)	Concentration of THC in plasma (ng/mL)
		Day 4	Day 9
5	50	—	6.51±3.13
10	100	4.43±1.65	6.82±2.90

The effects of acute rimonabant administration on ICSS thresholds are shown in Figure 4. A two-factor repeated-measures ANOVA, with rimonabant dose as a within-subjects variable and exposure condition as a between-subjects variable, was used to assess the effects of rimonabant. This analysis revealed a main effect of exposure condition, such that the change from baseline in ICSS thresholds after rimonabant administration was greater in smoke-exposed animals (F_(2,99)=3.609, p=0.0307), as well as a main effect of rimonabant, such that change from baseline in ICSS thresholds increased as a function of dose of rimonabant (F_(3,99)=31.523, p<0.001). Most importantly, there was a significant interaction between these two variables (F_(6,99)=5.847, p<0.001), indicating that the increase from baseline in ICSS thresholds was greater in the cannabis smoke-exposed groups. Post hoc comparisons revealed that across exposure conditions, all doses of rimonabant were significantly different from vehicle and the changes in ICSS thresholds for 5 cigarettes/session group upon administration of 3 mg/kg rimonabant were significantly different from the air-exposed group when given the same dose of rimonabant. No significant changes were observed in response latencies after rimonabant administration.

FIG. 4.

The post hoc comparison of changes (% change from baseline) in ICSS thresholds from baseline for each exposure group after rimonabant administration. The horizontal dashed line indicates 100% (no change). Baseline values were the ICSS threshold values in the session before that in which rimonabant (or vehicle) was administered. The horizontal line within the box indicates the median, boundaries of the box indicate the 25th and 75th percentiles, and the whiskers indicate the highest and lowest values of the results. The changes in the ICSS thresholds upon administration of rimonabant were significantly different from the vehicle condition. **p<0.01 and ****p<0.00001 compared with vehicle.

Sequential PK–PD modeling of THC-induced changes in ICSS threshold

The observed plasma concentration–time course for THC data obtained from the PK study was described by a three-compartment PK model (Figs. 5–7; Appendix Fig. A1). After obtaining a robust PK model to describe the PKs of THC after intravenous administration and smoke exposure, the final population model for PK was used to link the changes in THC concentration over time to the changes in ICSS thresholds over the course of cannabis smoke exposure. The exact study design was used to characterize the exposure–response relationship. The effect compartment model best described the changes in ICSS threshold data. Appendix Table A1 shows the summary of the PK/PD parameters. Results suggest that the change from baseline in ICSS thresholds after smoke exposure was delayed due to the distribution of THC into the effect compartment as suggested by an equilibration half-life of ∼20 h (Appendix Figs. A2 and A3).

FIG. 5.

The PK–PD model used to describe the changes in ICSS threshold after exposure to cannabis smoke. A three-compartment model was used to describe the PKs. The mass transfer between the compartments was described by intercompartmental clearances (Q2 and Q3) and elimination from central compartment (plasma) is given by CL. K_e0 describe the rate of transfer of drug from the central to effect compartment. The concentration in the effect compartment is the driver for lowering the ICSS threshold after smoke exposure. PD, pharmacodynamic; PK, pharmacokinetic.

FIG. 6.

The observed concentration of THC in red circles and the model population prediction (blue lines) for each subject based on their respective cohorts. THC, delta 9-tetrahydrocannabinol.

FIG. 7.

The visual predictive check is shown for the three-compartment pharmacokinetic model, which was used to describe the pharmacokinetics of THC after intravenous administration and inhalation of THC after exposure to cannabis smoke. The blue circles represent the observed data. The dashed blue line and the two black lines present the median, 0.05 and 0.95 percentiles for the observed data. The shaded region in gray shows the median, 0.05 and 0.95 percentile for the predictions.

Discussion

Results from the PK study suggest that this animal model of smoke exposure can be utilized to study THC exposure in rats, as the PKs after intravenous administration and cannabis smoke exposure were both well captured by the model. ICSS has been used to study the rewarding and aversive properties of a number of drugs of abuse, including opioids and nicotine.^38,39 This study investigated the effects of dose and duration of cannabis smoke exposure on development of cannabinoid dependence. The results suggest that chronic exposure to even low doses of THC inhaled through cannabis smoke potentiates brain reward function as indicated by lowering of ICSS thresholds, and can induce dependence as indicated by rimonabant-precipitated increases in ICSS thresholds.

THC PKs after i.v. administration and cannabis smoke inhalation

In this study, the PKs of THC in male Wistar rats after exposure to smoke from 5 or 10 cannabis cigarettes was investigated, and a population three-compartment linear model was proposed to describe the data. Validation suggests that the model robustly predicted the population PKs of this compound in rats. Terminal elimination half-life for THC was consistent with values reported in the literature, which range from 1 to 8 h in rats.^4–6 The parameter estimates for clearance were also similar to those reported recently (0.9 L/h).⁵ A two-compartment model was used previously to model the PKs of THC after intravenous administration in rats.⁶ THC is highly lipophilic and is rapidly distributed in highly perfused organs such as lung and brain, followed by distribution into adipose tissue where it is stored for a prolonged period of time. Hence, a three-compartment model would better explain the distribution of THC compared with a two-compartment model, as it accounts for the redistribution from fatty tissues back to plasma. The population PK in humans has also been described as a three-compartment model.⁴⁰ Therefore, based on these results we can say that the final three-compartment model assuming linear PK in this given dose range is suited to predict the plasma concentration of THC in rats.

Cannabis smoke inhalation and ICSS

It was initially suggested that THC administered systemically at a dose of 1.5 mg/kg has rewarding effects in the ICSS paradigm,¹³ after which several cannabinoids were tested in this paradigm to provide insight into their abuse potential.^41,42 Other research found no rewarding effects of THC in this paradigm,^43,44 and it was not until recently that a biphasic response to THC was suggested, wherein THC produces reward-facilitating effects at low doses (0.1 mg/kg) and aversive effects at higher doses (1 mg/kg).²³ These discrepancies can be attributed at least in part to the drug dose and rodent strain employed.^20,45 In addition, however, there is uncertainty regarding potentially different effects of acute versus chronic THC exposure. This study showed that chronic exposure to cannabis smoke, with estimated doses of 0.05 and 0.2 mg/kg for 5 and 10 cigarettes, respectively, can potentiate brain reward function as early as 4 days after the onset of smoke exposure, with maximal effects observed at 8 days of smoke exposure. Dependence is indicated by rimonabant-precipitated elevations in ICSS thresholds, such that changes from baseline for smoke-exposed rats were greater than those in air-exposed control animals, further confirming that CB1 receptors play a role in the development of cannabinoid dependence.

The present finding that cannabis smoke exposure lowers ICSS thresholds is not in agreement with some previous studies in which THC and synthetic CB1 receptor agonists did not produce rewarding effects in the ICSS paradigm in rodents.^41,44 Notably, however, Vlachou et al. used doses of THC ranging from 0.5 to 2.0 mg/kg, whereas this study used a dose range 10-fold lower (0.05–0.2 mg/kg),⁴⁴ and the study by Grim et al. used a CB1 agonist rather than THC.⁴¹ In contrast, Katsidoni et al. suggested a biphasic relationship between THC dose and its affective consequences, such that 0.1 mg/kg decreases and 10 mg/kg increases ICSS threshold.²³ Although the higher dose used falls out of the range of doses used in this study, the results from the lower dose are in good agreement with those shown here.

The prolonged half-life of THC can mask withdrawal symptoms in cannabis users. To determine if chronic exposure to cannabis smoke leads to dependence, after 2 weeks of smoke exposure, rats received acute injections of the CB1 receptor antagonist rimonabant before ICSS testing. Rimonabant induced a dose-dependent increase in ICSS thresholds in rats exposed to cannabis smoke. This finding is consistent with previous work from our laboratories demonstrating rimonabant-precipitated somatic withdrawal signs in rats after several weeks of daily exposure to cannabis smoke²⁷ as well as with other studies showing rimonabant-precipitated withdrawal after chronic cannabinoid administration in animal models.^21,27,46,47 Overall, these findings indicate that exposure to cannabis smoke leads to the development of dependence.

PK–PD modeling of THC-induced changes in ICSS thresholds

The plasma or serum concentration of THC increases in a dose-dependent manner, as do its pharmacological effects.⁴⁸ Recently, Awasthi et al. related the psychoactive effects upon smoking cannabis to the plasma concentrations of THC using an effect compartment approach.⁴⁹ A similar approach was used here to link plasma THC to changes in ICSS thresholds with repeated exposure to cannabis smoke. Various outcomes have different equilibration half-lives, indicating that different systems are involved with the production of the drug's response. An equilibration half-life of ∼20 h was calculated based on the estimated parameters for the PK–PD model. This can be attributed to the lipophilicity of the drug and the location of CB1 receptors. THC is a lipophilic compound that can cross the blood–brain barrier and accumulate in the brain for long durations.⁵⁰ This model-based approach with an equilibration half-life of 20 h suggests that THC might have rewarding effects even when it cannot be detected in plasma. Studies in our laboratory showed that somatic withdrawal signs in rats exposed to cannabis smoke containing THC can be observed as long as 4 h after completion of cannabis smoke inhalation, further providing evidence that the hedonic effects of cannabis develop slowly and last for long durations.²⁷ Another recent study in humans reported that the euphoric effects of cannabis are dose dependent, and can be observed as long as 5–6 h after smoking.⁴⁸ Because this study assessed hedonic effects of cannabis smoke at only one time point (1 h after exposure), future research should assess these effects at longer time points to determine how dose and chronicity of exposure interact with time since exposure.

Finally, PK–PD models may enable prediction of the time course of the analyzed effects, after coadministration of THC and antagonist leads in pharmacological proof-of-mechanism counteraction studies. With all these features, this integrated modeling approach offers insights into underlying mechanism of the endocannabinoid system. In this way, PK–PD modeling can contribute to the development of novel compounds targeting the cannabinoid system, and eventually lead to the development of therapies to treat cannabis use disorder.

Conclusion

In conclusion, the results from this study indicate that like other drugs of abuse, cannabis smoke potentiates brain reward, suggesting that it has reinforcing properties. The increase in brain reward threshold in smoke-exposed rats after administration of rimonabant is suggestive of a negative mood state or anhedonia. In combination with the results from our previous work showing that rimonabant precipitates withdrawal symptoms in rats undergoing the same chronic smoke exposure regimen used here,²² these data indicate that chronic exposure to cannabis smoke induces dependence. This study also highlights that the pharmacological effects of THC are present even when no THC is detected in plasma as suggested by an equilibration half-life of 20 h. With no currently approved medications for cannabis use disorder and high relapse rates after treatment, there is an urgent need to understand and alleviate cannabis withdrawal symptoms, which include lack of motivation, appetite loss, restlessness, and depressed mood, and can contribute significantly to relapse. Taken together, a whole-body cannabis smoke exposure model can be used to explain exposure–response relationships and help understand some of the underlying mechanisms of cannabinoid dependence and withdrawal. These types of novel animal models of cannabis smoke exposure may also aid in developing candidates to treat cannabinoid withdrawal.

Abbreviations Used

CB1R

cannabinoid 1 receptor

CBD

cannabidiol

CO

carbon monoxide

ICSS

intracranial self-stimulation

NCA

noncompartment analysis

PD

pharmacodynamic

PK

pharmacokinetic

THC

delta 9-tetrahydrocannabinol

TSP

total suspended particulate

UF

University of Florida

Appendix

APPENDIX FIG. A1. Diagnostics plot PK. The goodness-of-fit plots for the final PK model. PK, pharmacokinetic.

APPENDIX FIG. A2. PK–PD model predictions. The model prediction of the change in baseline upon repeated cannabis smoke administration. The red dots are the observations and the blue line represents the prediction (PRED) for each cohort. PD, pharmacodynamic.

APPENDIX FIG. A3. Diagnostics plot PK–PD model. The goodness-of-fit plots for the final PK–PD model use to describe affective dependence after chronic cannabis smoke exposure.

Appendix Table A1.

Pharmacokinetic–Pharmacodynamic Parameter Estimates Obtained After Sequential Modeling

Parameter	Estimate	RSE %
CL (L/h)	0.85	5
V1 (L)	0.56	15
V2 (L)	0.49	16
Q2 (L/h)	0.46	24
V3 (L)	1.26	10
Q3 (L/h)	0.14	12
Ke0 (1/h)	0.03	34
E0	100 Fixed	—
Slope	19.1	9.3
Gamma	0.23	2

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported by the McKnight Brain Institute and DA039349 (A.W.B., B.S., M.F.).

Cite this article as: Ravula A, Chandasana H, Jagnarine D, Wall SC, Setlow B, Febo M, Bruijnzeel AW, Derendorf H (2019) Pharmacokinetic and pharmacodynamic characterization of tetrahydrocannabinol-induced cannabinoid dependence after chronic passive cannabis smoke exposure in rats, Cannabis and Cannabinoid Research 4:4, 240–254, DOI: 10.1089/can.2019.0049.