Terpenes from Cannabis sativa induce antinociception in a mouse model of chronic neuropathic pain via activation of adenosine A 2A receptors

Abstract

Terpenes are small hydrocarbon compounds that impart aroma and taste to many plants, including Cannabis sativa . A number of studies have shown that terpenes can produce pain relief in various pain states in both humans and animals. However, these studies were methodologically limited and few established mechanisms of action. In our previous work, we showed that the terpenes geraniol, linalool, β-pinene, α-humulene, and β-caryophyllene produced cannabimimetic behavioral effects via multiple receptor targets. We thus expanded this work to explore the potential antinociception and mechanism of these Cannabis terpenes in a mouse model of chronic pain. We first tested for antinociception by injecting terpenes (200 mg/kg, IP) into male and female CD-1 mice with mouse models of chemotherapy-induced peripheral neuropathy (CIPN) or lipopolysaccharide-induced inflammatory pain, finding that the terpenes produced roughly equal antinociception to 10 mg/kg morphine or 3.2 mg/kg WIN55,212. We further found that none of the terpenes produced reward as measured by conditioned place preference, while low doses of terpene (100 mg/kg) combined with morphine (3.2 mg/kg) produced enhanced antinociception vs either alone. We then used the adenosine A 2A receptor (A 2A R) selective antagonist istradefylline (3.2 mg/kg, IP) and spinal cord-specific CRISPR knockdown of the A 2A R to identify this receptor as the mechanism for terpene antinociception in CIPN. In vitro cAMP and binding studies and in silico modeling studies further suggested that the terpenes act as A 2A R agonists. Together these studies identify Cannabis terpenes as potential therapeutics for chronic neuropathic pain and identify a receptor mechanism for this activity.

Figure 1:. Terpenes are efficacious in producing antinociception in a mouse model of neuropathic pain.

Male and female CD-1 mice had CIPN induced and measured as described in the Methods. Data shown is the mean ± SEM, performed in 2–6 technical replicates for each experiment, with sample sizes noted in each graph. BL = baseline. A) WIN55,212 or vehicle injected (1–10 mg/kg, IP). Each dose had a 10 mg/kg comparison performed at the same time, which are all combined here. B) The data from A was normalized to %Maximum Possible Effect (%MPE) and used to construct a dose/response curve. Linear regression revealed an A₅₀ = 1.5 mg/kg for WIN55,212 in CIPN. The dotted lines are the 95% confidence intervals for the regression. C-D) Each terpene (200 mg/kg, IP) or morphine comparison (10 mg/kg, SC) was injected. Each experiment had a morphine comparison, which are all combined here. E) The AUC for each terpene along with vehicle, morphine, and 3.2 mg/kg WIN55,212 controls are shown here. *, ** = p < 0.05, 0.01 vs. vehicle control by 1 Way ANOVA with Fisher’s Least Significant Difference post hoc test. All statistical details for the ANOVA comparisons performed in the main text are shown in Table S1.

Figure 2:. Terpenes are efficacious in producing antinociception in a mouse model of acute inflammatory pain.

Male and female CD-1 mice had inflammatory pain induced by LPS and measured as described in the Methods. Data shown is the mean ± SEM, performed in 2 technical replicates for each experiment, with sample sizes noted in each graph. BL = baseline. A) Terpene (i. geraniol, ii. linalool, iii. β-pinene, iv. α-humulene, v. β-caryophyllene)(200 mg/kg, IP) injected as noted along with vehicle control. *, **, ***. **** = p < 0.05, 0.01, 0.001, 0.0001 vs. same time point vehicle group by RM 2 Way ANOVA with Sidak’s post hoc test. B) AUC values calculated for each terpene and vehicle control, shown here. All vehicle results were combined into the group shown. *, ** = p < 0.05, 0.01 vs. vehicle group by 1 Way ANOVA with Fisher’s Least Significant Difference post hoc test.

Figure 3:. Terpenes lack reward liability as measured by conditioned place preference.

Male and female CD-1 mice had vehicle, morphine (10 mg/kg, SC), or terpene (200 mg/kg, IP) injected over a 4 day conditioning trial (see Methods), with preference measurement on day 5. Data shown represents the mean ± SEM of the % time spent in paired chamber over 3–5 technical replicates per group, with sample sizes noted in the legend. Values over 50% represent preference while values under 50% represent aversion. A) Summary data shown for each group. *, ** = p < 0.05, 0.01 vs. the baseline for each group by RM 2 Way ANOVA with Sidak’s post hoc test. Morphine shows a positive preference (reward), validating the assay. B) The baseline and post-conditioning trajectories for each mouse in each group are shown.

Figure 4:. Terpenes enhance morphine antinociception in CIPN.

Male and female CD-1 mice had CIPN induced and measured as described in the Methods. Data shown is the mean ± SEM, performed in 2–3 technical replicates for each experiment, with sample sizes noted in each graph. BL = baseline. Vehicle, morphine (3.2 mg/kg, SC), terpene (100 mg/kg, IP) or both combined were injected, with mechanical hypersensitivity measured. A) The time course data for geraniol is shown as an example; the curves for the other terpenes are all shown in Figure S2. *, **, **** = p < 0.05, 0.01, 0.0001 vs. same time point morphine or terpene group by RM 2 Way ANOVA with Dunnett’s post hoc test. B) The AUC data from A was calculated and is shown here (i. geraniol, ii. linalool, iii. β-pinene, iv. α-humulene, v. β-caryophyllene). * = p < 0.05 vs. terpene or morphine group by 1 Way ANOVA with Tukey’s post hoc test.

Figure 5:. Terpenes produce similar antinociceptive tolerance to morphine in CIPN.

Male and female CD-1 mice had CIPN induced as in the Methods, with A) morphine (10 mg/kg, SC) or B) terpene (i. geraniol, ii. linalool, iii. β-pinene, iv. α-humulene, v. β-caryophyllene)(25–200 mg/kg, IP) injection twice daily over a 4-day protocol, with daily mechanical hypersensitivity measurement after the morning injection. Data shown is the mean ± SEM, performed in 2–5 technical replicates for each experiment, with sample sizes noted in each graph. β-caryophyllene at 25 and 200 mg/kg showed high variability in the first cohorts, so additional cohort experiments were performed, explaining the wide range in sample size. The data shown here is the AUC calculated from each experimental set, with all raw data shown in Figures S5–S7. The morphine data is reproduced with permission from [29]. The 200 mg/kg dose of each terpene showed a roughly similar tolerance trajectory to 10 mg/kg morphine, with perhaps a less severe drop off on day 2. Lower doses of terpene had less initial efficacy, but also appear to have had slower tolerance development.

Figure 6:. Terpenes evoke antinociception in CIPN via the A_2AR.

Male and female CD-1 mice had CIPN induced and measured as described in the Methods. Data shown is the mean ± SEM, performed in 2–3 technical replicates for each experiment, with sample sizes noted in each graph. BL = baseline. A) Naïve mice with no CIPN had vehicle or the A_2AR antagonist istradefylline (3.2 mg/kg, IP) injected, followed by mechanical threshold measurements. Istradefylline had no impact on naïve mechanical thresholds, validating the dose. B) CIPN mice had vehicle or istradefylline (3.2 mg/kg, IP) injected, with a 10 min treatment time, followed by terpene (i. geraniol, ii. linalool, iii. β-pinene, iv. α-humulene, v. β-caryophyllene)(200 mg/kg, IP) as noted. *, **, ***, **** = p < 0.05, 0.01, 0.001, 0.0001 vs. same time point istradefylline/terpene group by RM 2 Way ANOVA with Sidak’s post hoc test. Istradefylline significantly reduced antinociception in CIPN by each terpene, implicating the A_2AR as a mechanism of action.

Figure 7:. AUC data for istradefylline/terpene treatment in CIPN.

The AUC was calculated from each pair of experiments in Figure 6 and shown here. *, ** = p < 0.05, 0.01 vs. paired vehicle/terpene group by Unpaired 1-Tailed t Test. The comparison is justified by the independent nature of each experiment along with the demonstrated reduction in antinociception with istradefylline treatment shown in Figure 6.

Figure 8:. Most terpenes evoke antinociception in CIPN via a spinal cord site of action.

Male and female CD-1 mice had CIPN induced and measured as described in the Methods. Data shown is the mean ± SEM, performed in 2–4 technical replicates for each experiment, with sample sizes noted in each graph. α-humulene showed high variability in the first experimental cohorts, so additional cohort experiments were performed, explaining the higher sample size in comparison to the other terpenes. BL = baseline. A) Vehicle or terpene (i. geraniol, ii. linalool, iii. β-pinene, iv. α-humulene, v. β-caryophyllene)(100 nmol, IT) injected into the spinal cord in CIPN mice and mechanical hypersensitivity measured. *, **, **** = p < 0.05, 0.01, 0.0001 vs. same time point vehicle group by RM 2 Way ANOVA with Sidak’s post hoc test. B) AUC values calculated from the data in A (all vehicle groups combined). **, **** = p < 0.01, 0.0001 vs. vehicle group by 1 Way ANOVA with Fisher’s Least Significant Difference post hoc test. All terpenes except β-pinene induce antinociception when injected into the spinal cord.

Figure 9:. Terpene mechanism confirmed by A_2AR CRISPR knockdown in the spinal cord.

Male and female CD-1 mice had CIPN induced along with A_2AR-targeted CRISPR or Negative Control (NC) CRISPR injections into the spinal cord so that day 8 of CIPN and day 10 of CRISPR coincided (see Methods). Data shown is the mean ± SEM, performed in 2–3 technical replicates for each experiment, with sample sizes noted in each graph. BL = baseline. Terpene injected (200 mg/kg, IP), with the exception of β-pinene due to lack of intrathecal response above, followed by mechanical hypersensitivity measurement. A) Time course data shown (i. geraniol, ii. linalool, iii. α-humulene, iv. β-caryophyllene). *, **, **** = p < 0.05, 0.01, 0.0001 vs. same time point A_2AR-CRISPR group by RM 2 Way ANOVA with Sidak’s post hoc test. B) AUC data calculated from A and compared. * = p < 0.05 vs. same terpene NC group by Unpaired 1-Tailed t Test. The comparison is justified by the independent nature of each experiment along with the demonstrated reduction in antinociception with A_2AR-CRISPR treatment shown in A.

Figure 10:. Terpenes act as A_2AR agonists in vitro.

A_2AR-HEK cells used for each experiment. Data represented as the mean ± SEM, with the sample size of independent experiments shown in each graph. Experiments performed in 3–4 technical, independent replicates. A) Vehicle, terpenes (100 μM), and NECA positive control (10 μM) used to stimulate cAMP accumulation. Data normalized to the percent stimulation caused by vehicle (0%) and NECA (100%). **** = p < 0.0001 vs. vehicle group by 1 Way ANOVA with Dunnett’s post hoc test. B) Competition radioligand binding performed with terpene (300 μM) competing against the orthosteric A_2AR ligand ³H-ZM241385. Data normalized to the percent competition caused by vehicle (0%) and saturating cold ligand (100%). Together the results suggest that terpenes act as A_2AR partial agonists without competing against the orthosteric ligand ZM241385.

Figure 11:. Molecular modeling of terpene binding with the A_2AR and the mu opioid receptor (MOR).

The 5 highest ranking poses of each terpene were docked to the A_2AR and the MOR as described in the Methods. TMH = transmembrane helix; EC = extracellular domain. Docked illustrations are shown in the top-down view of the orthosteric binding pocket. Ai) Docking poses of the terpenes in the A_2AR binding pocket, along with noted interacting residues. Aii) The 16 high confidence interacting residues for the terpenes are shown. Green = hydrophobic residues, Cyan = hydrophilic residues. Aiii) 2D ligand interactions with the receptor cavity, showing geraniol as a representative example. Geraniol showed hydrogen bonding with Ser²⁷⁷, His²⁷⁸ and aromatic amino acids; Phe¹⁶⁸, Tyr²⁷¹, Trp²⁴⁶ by π-π aromatic side chains and stacking interactions and aliphatic amino acids by van der Waals interaction. Pink and green colored circled amino acids are polar and lipophilic residues, respectively. The pale blue halo marks the area of interactions. Bi) Docking poses of the terpenes in the MOR binding pocket, along with noted interacting residues. Tyr326 and Asp147 are particularly noted, since the terpenes lack key bonds with these residues in our predictive model, which may explain their lack of activity at the MOR. Bii) The 17 high confidence interacting residues for the terpenes are shown. Green = hydrophobic residues, Cyan = hydrophilic residues. Biii) 2D ligand interactions with the receptor cavity, exemplified by linalool. Linalool showed hydrogen bonding with Asp¹⁴⁷, Tyr³²⁶ by arene-H, and Trp²⁹³, along with aromatic side chain interactions by π-π and van der Walls stacking. Pink and green colored circled amino acids are polar and lipophilic residues, respectively. The pale blue halo marks the area of ligand interactions with the receptor binding cavity.

Figure 12:. Model of terpene antinociception in CIPN.

Under the condition of paclitaxel (TAX) treatment, terpene injection results in activation of the A_2AR in the spinal cord, leading to antinociception. Separately, opioid injection can activate the MOR at an undefined site which enhances terpene antinociception.