Selective targeting of peripheral cannabinoid receptors prevents behavioral symptoms and sensitization of trigeminal neurons in mouse models of migraine and medication overuse headache.

Abstract

Migraine affects ~15% of the world’s population greatly diminishing their quality of life. Current preventative treatments are effective in only a subset of migraine patients, and while cannabinoids appear beneficial in alleviating migraine symptoms, central nervous system (CNS) side effects limit their widespread use. We developed peripherally-restricted cannabinoids (PRCBs) that relieve chronic pain symptoms of cancer and neuropathies, without appreciable CNS side effects or tolerance development. Here we determined PRCB effectiveness in alleviating hypersensitivity symptoms in mouse models of migraine and medication overuse headache (MOH). Chronic glyceryl trinitrate (GTN, 10 mg/kg) administration led to increased sensitivity to mechanical stimuli, and increased expression of phosphorylated protein kinase A (p-PKA), neuronal nitric oxide synthase (nNOS), and transient receptor potential ankyrin 1 (TRPA1) proteins in trigeminal ganglia. PRCB pretreatment, but not posttreatment, prevented behavioral and biochemical correlates of GTN-induced sensitization. Low pH- and allyl isothiocyanate-activated currents in acutely isolated trigeminal neurons were reversibly attenuated by PRCB application. Chronic GTN treatment significantly enhanced these currents. Chronic sumatriptan treatment also led to development of allodynia to mechanical and cold stimuli which was slowly reversible after sumatriptan discontinuation. Subsequent challenge with a previously ineffective low-dose GTN (0.1–0.3 mg/kg) revealed latent behavioral sensitization and increased expression of p-PKA, nNOS, and TRPA1 proteins in trigeminal ganglia. PRCB pretreatment prevented all behavioral and biochemical correlates of allodynia and latent sensitization. Importantly, chronic PRCB treatment alone did not produce any behavioral or biochemical signs of sensitization. These data validate peripheral cannabinoid receptors as potential therapeutic targets in migraine and MOH.

1. Introduction

Migraine is a debilitating disorder characterized as an episodic, throbbing cephalic pain, commonly accompanied by nausea and hypersensitivity to light, sound, and touch ¹. Migraine is ~3-times more prevalent in females than males; women experience more frequent, longer lasting and more painful attacks, have more disability and greater risk of transition from episodic to chronic migraine than men ⁴⁵. One of the leading hypothesized mechanisms of migraine pain is the sensitization of nociceptive dural afferents by calcitonin gene related peptide (CGRP) and a nitric oxide (NO) donor ^47,56,65. Nociceptor sensitization leads to sensitization of 2^nd- and 3^rd-order neurons in the trigeminal nucleus caudalis and thalamus, respectively, resulting in headache as well as cephalic and extracephalic cutaneous allodynia ^{22–24,64,100}.

Triptans inhibit the release of pronociceptive transmitters, including CGRP, through actions on 5-hydroxytryptamine (5HT)1B/1D receptors on trigeminal afferents, hence their effectiveness in terminating migraine attacks ^{21,25,40,50,102}. For example, sumatriptan alleviates migraine headache induced in migraineurs by cilostazol, an inhibitor of cyclic adenosine monophosphate (cAMP) degradation ^43,49. However, frequent use of triptans can increase the risk of medication overuse headache (MOH) ^{17,31,37,87,105}. Evidence suggests that MOH, and perhaps transformation of episodic to chronic migraine by triptans, may result from increased expression of CGRP and neuronal nitric oxide synthase (nNOS) within trigeminal afferents, leading to chronically lowered thresholds to stimuli that trigger migraine headache ^32,33,112. In chronic inflammation, increased cAMP production by adenylyl cyclase and subsequent increases in phosphorylated protein kinase A (p-PKA) mediate increased expression of CGRP and nNOS ¹⁰⁸. Hyperalgesic priming, a model of chronic pain development, involves a transition from cAMP/p-PKA to cAMP and phosphorylated protein kinase Cε (p-PKCε)-dependent signaling ^18,89,90. P-PKCε-dependent hyperalgesic priming is also induced by sumatriptan administration in male, but not female rats, and prevented by pretreatment with pertussis toxin or isolectin B4–positive nociceptor toxin, IB4–saporin ⁸.

Cannabinoids acting on G_i/o-coupled cannabinoid 1 and 2 receptors (CB1Rs and CB2Rs) alleviate migraine symptoms in humans ^10,29 and in animal models ^5,6,57,84,86. However, side effects mediated by CB1Rs in the central nervous system (CNS) limit their widespread use ⁹⁹. Previous studies with conditional knockouts of peripheral CB1Rs showed that cannabinoid analgesia in inflammatory and neuropathic pain states is mediated largely by peripheral CB1Rs in nociceptors ⁴. We developed peripherally-restricted cannabinoids (PRCBs), demonstrating their effectiveness in relief of cancer and neuropathic pain symptoms in animal models, without CNS side effects ^81,97,113. These compounds allowed us for the first time, to test the utility of selective peripheral CBR activation in alleviating hypersensitivity behaviors in mouse models of migraine and MOH. We hypothesized that increased expression of trigeminal sensory neuron sensitization biomarkers such as p-PKA and subsequent increases in phosphorylation and activation of acid-sensing ion channels (ASICs) and transient receptor potential ankyrin 1 (TRPA1) channels contribute to the increased excitability of trigeminal afferents and hypersensitivity behaviors. We further hypothesized that peripheral CBR activation by the lead PRCB, 4-{2-[-(1E)-1[(4-propylnaphthalen-1-yl)methylidene]-1H-inden-3-yl]ethyl}morpholine (PrNMI), may protect against sensitization of ASIC and TRPA1 by inhibiting adenylyl cyclase/cAMP signaling, thereby preventing migraine- and MOH-like symptoms in mice.

2. Materials and methods

2.1. Animals

Female and male C57BL/6J mice (Jackson, Bar Harbor, ME) 6–8 weeks old, weighing 18–25 g at the beginning of the experiments were used. All mice were group housed (3–4 per cage, same sex) to avoid social isolation stress, in the vivarium under a 12-hr light/dark cycle (lights on at 6 AM) and had ad libitum access to food and water during the entire experiment. All experimental procedures were carried out in accordance with the National Institute of Health guidelines for the handling and use of laboratory animals and with approval from the Animal Research Committee of the University of California, Los Angeles. In addition, experiments were performed in accordance with the guidelines of the International Association for the Study of Pain to ensure minimal animal use and discomfort.

2.2. Behavioral testing

Prior to experiments, mice were randomly assigned to groups. All behavioral experiments were performed between 9 a.m. and 6 p.m. Investigators involved in behavioral testing, electrophysiological recordings, and data analysis were blinded to the nature and dose of the administered experimental drugs.

2.2-1. Assessment of mechanical and cold allodynia

Mice were initially habituated to the testing environment for 2 hrs daily on 3 consecutive days. For periorbital testing, mice were placed in 89-ml paper cups and placed in 5 cm length × 7.6 cm width × 15 cm height Plexiglas containers with a 0.8 mm thick wire mesh (0.6 cm wide vertical slits) side, with the cup opening facing the mesh. For hindpaw testing, mice were placed in 9.5 cm length x 9.5 cm width x 12.5 cm height Plexiglass containers with lids (IITC Life Sciences, Woodland Hills, CA) and a 0.8 mm wire mesh flooring (0.6 cm² spacing). All mice were tested at the same time of the day on 2–3 separate occasions to establish stable baseline responses to stimuli before any treatment. Withdrawal thresholds to von Frey filaments (Touch Test Sensory Evaluators, North Coast, Morgan Hill, CA, USA; bending force ranging from 0.008 to 2 g) applied to either the plantar region of the hindpaw or the periorbital region were determined after acclimating the mice in test cages for 15 minutes. Tests were carried out by an investigator in a soundproofed room, at 300 lux soft white fluorescent illumination, with minimal acoustic or smell disturbances capable of affecting the results. The von Frey filaments were applied perpendicularly to the plantar surface of hindpaws or the periorbital region until it buckled slightly and held for several seconds or until a withdrawal occurred, termed a positive response. If a response was not noted, a denser filament was tested; in the presence of a response, a lighter filament was tested. Withdrawal thresholds were determined by the “up-down” method ^27,39. The first filament tested for periorbital sensitivity was 0.008 g and the first filament tested for hindpaw sensitivity was 0.4 g. For hindpaw testing, mice received 2 trials on each hindpaw and the average was taken of the 4 trials. For periorbital testing, mice received 3 trials and the average was taken of the 3 trials. Each trial was performed at least 5 minutes apart to prevent sensitization to the testing filaments. The method to evaluate withdrawal responses to cold stimulation was adapted from ⁶¹. To test cold sensitivity, 10 µL of acetone at room temperature was applied to the periorbital region, with a 1 mL syringe connected to PE-10 tubing. Care was taken to avoid mechanical stimulation by the syringe tip. The withdrawal response (head shake(s) followed by facial grooming behavior involving bilateral forepaws and hindpaws) was observed for up to 30 seconds. Mice received 3 trials performed at least 5 min apart and the average was taken of the 3 trials. The responses to mechanical and cold stimuli were observed by two different investigators, respectively. All criteria for behavioral testing were determined a priori and followed throughout the study.

2.2-2. Administration of drugs

Glyceryl trinitrate (GTN, Sigma-Aldrich, St. Louis, MO) supplied as 200 mg per ampule was diluted in propylene glycol to a final concentration of 16.6 mg/ml and stored at 4°C as aliquots in amber-colored air-tight glass vials. Immediately before injecting mice, each aliquot was further diluted in an equal volume of 95% ethanol, and sterile saline. The final amount of propylene glycol and ethanol was adjusted to ≤ 6% with sterile bacteriostatic saline for all GTN or vehicle injections administered at 10 ml/kg. PrNMI was synthesized by Dr. Seltzman at Research Triangle Institute (RTI, Research Triangle Park, NC) ^46,97. SR144528 was also synthesized by Dr. Seltzman at RTI as part of the National Institute of Drug Abuse (NIDA) drug supply program. The peripherally-restricted rimonabant analogue, 18A, was synthesized by Dr. Fulp at RTI ⁴⁶. All cannabinoids were dissolved in DMSO: Tween 80: saline (1:1:80). Sumatriptan (Cayman Chemical, Ann Arbor, MI) was dissolved in bacteriostatic saline.

2.2-3. Acute and chronic migraine model

The acute and chronic migraine model was adapted from ⁹¹. For studies of PrNMI pre-treatment effectiveness (n=10–12/group), animals were tested for baseline responses immediately prior to intraperitoneal (i.p.) injection with drugs: vehicle, PrNMI (0.1–0.6 mg/kg), 18A (6 mg/kg), SR141716 (6 mg/kg), and SR144528 (6 mg/kg), or a combination thereof. One hour later, mice were administered GTN (10 mg/kg) and tested for mechanical sensitivity 2 hrs later. For studies of PrNMI post-treatment effectiveness, animals were tested for baseline responses immediately prior to injection with GTN (10 mg/kg, i.p.) followed 30 min later by PrNMI (0.6 mg/kg, i.p.) and tested again 2 hrs after PrNMI administration. Allodynia suppression was calculated using a modified formula (% Allodynia suppression = [Post-drug response – Pre-drug response]/[Baseline – Pre-drug response]*100) and normalized to GTN-induced responses in the vehicle-treated group. Prior to any additional drug testing, baseline sensitivity (measured at 2-day intervals) was allowed to return back to control values to ensure full recovery from the effects of the previously administered drug. For chronic migraine model studies, drug injections were repeated 7 times at 2-day intervals.

2.2-4. Medication Overuse Headache (MOH) model

We adapted a previously developed rat model of triptan overuse-induced allodynia ^32,33 to mice by administering sumatriptan (SUMA; 0.6 mg/kg, i.p.) every 2 days for a total of 7 doses. Groups (vehicle/SUMA, PrNMI/SUMA, vehicle/vehicle, PrNMI/vehicle, n=10/group) were tested for baseline mechanical and cold sensitivity, followed by either vehicle or PrNMI (0.6 mg/kg, i.p.) administration, followed 90 min later by vehicle or SUMA administration. Behavioral tests were repeated 30 min after SUMA injections. After the last injection of SUMA, baseline sensitivity testing continued at 2-day intervals.

For latent sensitization testing, after recovery of baseline sensitivity to pre-SUMA levels, mice were challenged with subthreshold doses of GTN (0.1 mg/kg and 0.3 mg/kg for hindpaw sensitivity, 0.3 mg/kg for periorbital sensitivity) which previously did not trigger mechanical or cold allodynia. Behavioral tests were conducted before and 2 hrs after GTN injection. Baseline sensitivity continued to be tested at 2-day intervals.

2.3. Quantitative Protein Analysis

At the conclusion of behavioral testing, all mice were euthanized and the trigeminal ganglia were harvested for biochemical analysis. Bilateral trigeminal ganglia were lysed in 100 μL of RIPA lysis buffer (Invitrogen, Long Island, NY, USA) supplemented with PMSF (0.5 μM) and Halt™ protease and phosphatase inhibitors cocktail (1:100) (Invitrogen). For quantification of target proteins, we utilized the Wes system (ProteinSimple, San Jose, CA, USA), an automated size-based separation and nano-immunoassay platform for detection and characterization of protein molecular weights in denatured protein lysates. The Wes system automatically performs all the manual steps associated with traditional Western analysis and provides true quantification of results. Briefly, samples (0.5 μg/μl) were diluted in a Wes sample buffer, reduced, and denatured according to the manufacturer’s protocol, and run on a 12–230 KDa separation module. Primary antibodies for target proteins were diluted in Wes antibody diluent 2 as described : β-actin (1:100), GAPDH (1:2500), and PKA (1:100) (Cell Signaling Technology, Danvers, MA), ASIC1 (1:50) and nNOS (1:10) (Millipore Sigma, Burlington, MA), PAR2 (1:10) and p-PKA (1:10) (Santa Cruz Biotechnology, Dallas, TX), ASIC3 (1:50) (Alomone labs, Jerusalem, Israel), CB1R (1:50) (Ken Mackie), and TRPA1 (1:150) (Novus Biologicals, Centennial, CO). Wes anti-rabbit/anti-mouse detection modules (Part # DM-001/DM-002, ProteinSimple, San Jose, CA) were used for chemiluminescent detection. Spectral peaks were quantified on Compass for Simple Western software. All target proteins were normalized to total protein for quantification.

2.4. Electrophysiology

Acutely dissociated TG neurons from female C57BL/6J mice were prepared for recordings as described previously ¹¹⁰, with minor modifications. Briefly, the excised TGs were placed into cold (4ºC) modified Tyrode’s solution containing (in mM): NaCl 130, NaHCO₃ 20, KCl 3, CaCl₂ 4, MgCl₂ 1, HEPES 10, glucose 12, and antibiotic/antimycotic solution (0.5%; Fisher Scientific, Hanover Park, IL). For ASIC current recording, tissues were incubated in collagenase I (1 mg/ml, Fisher Scientific) for 1 hour and then in collagenase with trypsin/EDTA (0.2%, Fisher Scientific) for another 1 hour at 37°C. For TRPA1 current recording, tissues were incubated in collagenase II (Invitrogen) and dispase II (1 mg/ml each, Sigma-Aldrich) for 1 hour at 37°C as the latter were shown to favor expression and function of TRPA1 channels after cell isolation ^7,96. TG cells were washed twice with the modified Tyrode’s solution and triturated gently using fire-polished Pasteur glass pipettes. The cell suspension was mixed with bovine serum albumin (15%; Fisher Scientific), centrifuged at 900 rpm for 10 min, the pellet resuspended with Neurobasal A medium containing B27 (2%), L-glutamine (0.2%), and antibiotic/antimycotic solution (0.1%), and cells plated onto glass coverslips pre-coated with poly-D-lysine/laminin (all from Fisher Scientific), incubated at 37°C in a humidified 5% CO₂ chamber and used within 8 hours after plating. We did not add exogenous growth factors, such as nerve growth factor (NGF), because they have been shown to increase the functional expression of TRP channels, including TRPA1 ^2,16.

At 1 hour after plating, coverslips containing the TG neurons were placed in a recording chamber (Warner Instruments) on the stage of an upright microscope (Leica) equipped with phase, interference contrast, and fluorescence optics. Whole-cell patch clamp recordings were obtained with patch micropipettes filled with internal solution containing (in mM): KGluconate 100, KCl 40, EGTA 1, HEPES 10, CaCl₂ 0.3, MgATP 2, and Na₂GTP 0.2, pH 7.2 (adjusted with KOH, 290 mOsm). For ASIC current recordings, the external solution contained in (mM): NaCl 140, KCl 4, HEPES 5, MES 5, CaCl₂ 2, MgCl₂ 2, glucose 10, pH 7.4 (adjusted with NaOH, 315mOsm). For TRPA1 current recordings, the external solution containing in (mM): NaCl 140, KCl 4, HEPES 10, CaCl₂ 0.5, MgCl₂ 3.5, glucose 10, pH 7.4 (adjusted with NaOH, 315mOsm). All internal and external solution chemicals were from Sigma-Aldrich. Rounded, phase-bright neurons were selected for patch-clamp experiments. The cell body diameter was determined with a calibrated eyepiece reticule. ASIC currents were evoked by fast switching to gravity-fed external solutions with pH adjusted to 6.0–6.5 with 1 M HCl with a computer-controlled fast-switching 3-barreled pipette (0.7 mm inside barrel diameter) positioned 50 μm away from the target neuron (SF-77B, Warner Instruments, Hamden, CT). To activate TRPA1 currents, AITC (10–300 μM) was applied (after appropriate dilution in the external solution) by fast switching perfusion. For TRPA1 current recordings, fluorescein isothiocyanate (FITC)-conjugated isolectin B4 (IB4, 10 μg/ml, Vector, CA, USA) was incubated for 10 min and then rinsed for 2 min before starting recordings. IB4 +ve neurons were visually identified using an FITC filter (excitation 465–495 nm; emission barrier filter 515–555 nm) ^73,101. A neuron was considered IB4 +ve if it had a continuous green ring around the perimeter at 10× magnification. The recording chamber was continuously perfused with external solution at a constant rate of 2 ml/min using a peristaltic perfusion system (Ismatec Reglo ISM-1B, Cole Palmer, Vernon Hills, IL, USA). PrNMI (1 μM) with or without 18A (10 μM) were bath applied after appropriate dilution to determine their effect on ASIC and TRPA1 currents in TG neurons. All recordings were conducted at room temperature. Voltage-clamp data were acquired using an amplifier (MultiClamp, Model 700B), A/D converter (Digidata 1440A) controlled with a PC running pCLAMP 10 software (Molecular Devices, San Jose, CA, USA). After whole-cell access was established, an equilibration period of 2–3 min was allowed prior to recording. Current data were sampled at 10 kHz and filtered at 2 kHz. Patch electrodes were pulled from 1.5 mm OD borosilicate glass (Warner Instruments) on a horizontal puller (P-87, Sutter Inst., Novato, CA, USA), and had 2–5 MΩ resistances when filled with internal solution. The junction potential associated with all test solutions was less than 5 mV and therefore not corrected. Whole-cell capacitance was compensated with amplifier circuitry. Neurons were held at −60 mV. Current density was determined by dividing peak evoked current by membrane capacitance. Concentration–response data were fitted with a Hill equation to estimate 50% effective concentration (EC₅₀). Allyl isothiocyanate (AITC), 18A, and PrNMI were dissolved in dimethyl sulfoxide (DMSO) to make a stock solution and were kept at −20 °C until the day of experiment. The final concentration of DMSO was less than 0.1% (v/v). The drugs were diluted to their final concentration with the extracellular solution just before application.

2.5. Statistical analysis

Data are expressed as mean ± SEM. The Shapiro–Wilk test was used to check data normality. For two group comparisons, paired or unpaired Student’s t-test were used. For multiple comparisons, one-way and two-way ANOVA or repeated measures (RM) ANOVA followed by the appropriate post-hoc test was performed using Prism 8 software (GraphPad, San Diego, CA, USA). Significance was set at P < 0.05 for all data analyses. Power analysis was conducted to estimate sample size with 0.8 power to reach a significance level of 0.05. The statistical analyses for individual experiments are described in figure legends.

3. Results

3.1. PrNMI pretreatment suppresses GTN-induced acute mechanical allodynia in male and female C57BL/6J mice

Significant mechanical allodynia was consistently observed in the periorbital region of male (Fig. 1A) and female (Fig. 1B) mice, respectively at 2 hrs after GTN (10 mg/kg, i.p.) administration. Although the comparative effect size was small, females were significantly more sensitive to the acute effect of GTN treatment (* P < 0.05, Males vs Females, two-way ANOVA, Šidák’s multiple comparisons test). Hindpaw withdrawal thresholds tested in another cohort of female mice were also reduced after GTN (Fig. 1C). Pretreatment with the peripherally restricted cannabinoid, PrNMI (0.6 mg/kg, i.p.), 1 hr prior to GTN injection, resulted in almost complete suppression of periorbital (Fig. 1A, B) and hindpaw (Fig. 1C) mechanical allodynia. This PrNMI dose was initially selected based on its effectiveness at suppression of allodynia symptoms in animal models of traumatic peripheral nerve injury ⁹⁷, cancer ¹¹³ and chemotherapy-induced neuropathy ⁸¹ with minimal CNS-mediated side effects. The pretreatment time point was based on the time course of changes in plasma PrNMI concentration, with a peak at 3 hrs after PrNMI (0.6 mg/kg, i.p.) administration in male C57BL/6J mice (Supplemental Fig. S1).

Figure 1. PrNMI suppresses GTN-induced acute periorbital and hindpaw mechanical allodynia.

A: changes in head withdrawal responses to mechanical stimuli of female C57BL/6J mice (n=10) 2 hrs after injection of GTN (10 mg/kg, i.p.) and 3 hrs after pretreatment with either vehicle or PrNMI (0.6 mg/kg, i.p.). B: changes in hindpaw withdrawal responses to mechanical stimuli in a separate group of female C57BL/6J mice (n=10) 2 hrs after injection of GTN (10 mg/kg, i.p.) and 3 hrs after pretreatment with either vehicle or PrNMI (0.6 mg/kg, i.p.). * P < 0.05, *** P < 0.001 (Vehicle+GTN vs PrNMI+GTN, two-way ANOVA, Šidák’s multiple comparisons test).

3.2. PrNMI suppresses GTN-induced acute mechanical allodynia via CB1R and CB2R activation

GTN-induced allodynia suppression by PrNMI was steeply dose-dependent, with 0.1 mg/kg being ineffective (Fig. 2A). Dose-response curve fitting yielded an ED₅₀ of 0.36 mg/kg for mechanical allodynia suppression (Fig 2A). To determine the contribution of each CBR subtype to the alleviation of GTN-induced mechanical allodynia, we co-administered PrNMI, a full agonist at CB1R and a partial agonist at CB2R ⁹⁷, with either a CB1R or a CB2R selective antagonist. The anti-allodynic effects of PrNMI (0.6 mg/kg) were effectively blocked by the peripherally restricted selective CB1R antagonist, 18A (6 mg/kg), or by the selective CB2R antagonist, SR144528 (6 mg/kg), to a similar extent in both females (Fig. 2B) and males (not shown). Given the comparatively lower sensitivity of male C57BL/6J mice to GTN-induced allodynia symptoms, all subsequent experiments were performed in female mice.

Figure 2. PrNMI suppresses GTN-induced acute mechanical allodynia via both CB1R and CB2R.

A: changes in withdrawal thresholds of female mice (n=10) obtained at 2 hrs after injection of Vehicle or GTN (10 mg/kg, i.p.) and 3 hrs after pretreatment with either vehicle, or PrNMI (0.1, 0.3, or 0.6 mg/kg, i.p.). B: dose-response of PrNMI in preventing GTN-induced acute mechanical allodynia. ED₅₀ = 0.36 mg/kg. C, D: changes in withdrawal thresholds of female (C) and male (D) mice (n=10/group) obtained at 2 hrs after injection of GTN (10 mg/kg, i.p.) and 3 hrs after pretreatment with either vehicle, PrNMI (0.6 mg/kg, i.p), peripherally-restricted selective CB1R antagonist 18A (6 mg/kg, i.p.) + PrNMI, or selective CB2R antagonist SR144528 (6 mg/kg, i.p.) + PrNMI. Data in (C) were obtained from the same group of mice as in (A). ** P < 0.01, *** P < 0.001 (Pre vs Post, two-way ANOVA, Tukey’s multiple comparisons test). Note the similar block of PrNMI’s actions by the CB1R antagonist, 18A, and the CB2R antagonist, SR144528.

3.3. PrNMI suppresses ASIC and TRPA1 currents via CB1R activation in TG neurons.

Considerable evidence points to the involvement of ASICs and TRPA1 channels in migraine pathophysiology ^{12–14,41,58,111}. Both ASICs and TRPA1 channels are present in mouse TG dural afferents ¹¹², and their activation was shown to be positively modulated downstream of adenylyl cyclase/cAMP-mediated phosphorylation of PKC and PKA ^67,74,107. To address potential mechanisms by which PrNMI pretreatment prevents GTN-induced behavioral hypersensitivity, we first examined the effects of PrNMI on ASIC currents using acutely isolated TG neurons from naïve female C57BL/6J mice as surrogates of their activation at peripheral terminals. In some neurons, a 2-second application of a pH 6.0 solution evoked an inward current (I_pH6.0), with medium (25–45 μm) and large size (> 45 μm) cells responding more frequently than cells of small size (< 25 μm) (Fig. 3A). Bath application of PrNMI (1 μM) had no effect on the holding current of neurons voltage-clamped at −60 mV. However, during application of PrNMI, I_pH6.0 was suppressed by ~50% and was reversible upon wash (Fig. 3B). To investigate the contribution of CB1R activation to I_pH6.0 suppression, we co-applied PrNMI (1 μM) with the selective CB1R antagonist, 18A (10 μM); this also had no effect on the holding current, but completely prevented PrNMI suppression of I_pH6.0 (Fig. 3C). For statistical comparisons, I_pH6.0 amplitudes (evoked at 5-min intervals) were normalized to cell capacitance, before, during, and after switching to vehicle (0.1% DMSO), PrNMI, or PrNMI+18A (Fig. 3D–F). There was no appreciable rundown of I_pH6.0 when recorded at 5-min intervals (Fig. 3D).

Figure 3. PrNMI suppresses ASIC currents via CB1R activation in TG neurons.

A: graph of TG neurons separated by size and responsiveness to a fast switch from a pH 7.4 to a pH 6.0 perfusate. B: representative traces of currents evoked at 5-min intervals by a 2-sec switch to a pH 6.0 solution (bar) before, during, and after a 5-min perfusion of PrNMI (1 μM). C: representative pH 6.0-evoked currents before, during, and after a 5-min perfusion of PrNMI (1 μM) with the selective CB1R antagonist, 18A (10 μM). D, E, F: graphs of pH 6.0-evoked currents before, during, and after switching to vehicle (0.1% DMSO), PrNMI, or PrNMI+18A, respectively. Recordings in (D) were from 1 large, 2 medium, and 3 small sized neurons; in (E) from 1 large, 6 medium, and 2 small sized neurons; in (F) from 1 large, 2 medium, and 5 small sized neurons. * Adjusted P < 0.05 (paired t-test followed by Holm-Šidák correction). Note the significant reversible decrease in currents during perfusion with PrNMI, but not with PrNMI+18A.

Next, we examined the effects of PrNMI on TRPA1 channel activation. Previous studies revealed that TRPA1 channels are localized primarily to small diameter TG neurons in mice ^9,51,83. It was also demonstrated that whole cell TRPA1 currents induced by brief application of low concentrations of allyl isothiocyanate (AITC) can be reversed readily by washout, whereas saturating concentrations of AITC induced tachyphylaxis of the TRPA1 current, which was more pronounced in the presence of extracellular Ca^{2+ 93}. Therefore, we concentrated on recording responses to AITC (30–300 μM) from small diameter (< 25 μm) TG neurons. Concentration-response curve fitting yielded an estimated EC₅₀ of 52 μM for the AITC-induced current amplitude normalized to the cell capacitance (Fig. 4A). A 5-second application of AITC (60 μM) evoked a slow inward current (I_AITC) which was suppressed during application of PrNMI (1 μM), and was partly reversible during wash (Fig. 4B). To investigate the contribution of CB1R activation to I_AITC suppression, we perfused PrNMI (1 μM) with the selective CB1R antagonist, 18A (10 μM); this prevented PrNMI suppression of I_AITC (Fig. 4C). For statistical comparisons, I_AITC amplitudes (evoked at 5-min intervals) were normalized to cell capacitance, before, during, and after switching to vehicle (0.1% DMSO), PrNMI, or PrNMI+18A (Fig. 4D–F). There was a significant rundown of I_AITC when recorded at 5-min intervals. This was compensated by estimating the average rundown experienced by vehicle-treated cells over time, adding that value to the I_AITC peak amplitude measured during PrNMI or 18A co-application (Supplemental Fig. S2). After such compensation, the PrNMI (1 μM)-induced suppression of the I_AITC was determined to be 60.1 ± 5.4 % (Supplemental Fig. S2).

Figure 4. PrNMI suppresses TRPA1 currents via CB1R activation in TG neurons.

A: AITC concentration-response graph in TG neurons. Data points are fit by least squares non-linear regression to estimate EC₅₀ value (n = 3–11 neurons/data point). B: representative traces of currents evoked at 5-min intervals by a fast 5-sec switch to AITC (60 μM) solution (bar) before, during, and after a 5-min perfusion of PrNMI (1 μM). C: representative AITC-evoked currents before, during, and after a 5-min perfusion of PrNMI (1 μM) with the selective CB1R antagonist 18A (10 μM). D, E, F: graphs of AITC-evoked currents before, during, and after switching to vehicle (0.1% DMSO), PrNMI, or PrNMI+18A, respectively. * P < 0.05, ** P < 0.01 (paired t-test followed by Holm-Šidák correction to obtain adjusted P-values).

3.4. PrNMI pretreatment suppresses development of GTN-induced chronic mechanical allodynia

Repetitive administration of GTN (10 μM) at 2-day intervals resulted in gradual development of chronic mechanical allodynia in the periorbital region (Fig. 5A) and hindpaws (Fig. 5B). Previous studies demonstrated the predictive validity of this model has with respect to sumatriptan effectiveness in aborting acute migraine attacks and topiramate effectiveness in preventing development of chronic migraine, respectively ^91,104. Repetitive pretreatment with PrNMI significantly reduced repetitive GTN-induced acute (not shown) and chronic mechanical allodynia in the periorbital region (Fig. 5A) and hindpaws (Fig. 5B), respectively. At the peak of periorbital hypersensitivity, animals were euthanized (↑ in Fig. 5A), trigeminal ganglia collected and stored at −80 °C for later biochemical experiments. Animals tested for hindpaw withdrawal were followed until return to pre-injection baseline thresholds (Fig. 5B). These results demonstrated that selective activation of peripheral cannabinoid receptors prevents GTN-induced acute and chronic mechanical allodynia.

Figure 5. PrNMI suppresses GTN-induced chronic mechanical allodynia in head and hindpaw.

A: baseline head withdrawal responses to mechanical stimuli (mean ± SEM) of 3 groups of C57/BL6J female mice (n=10/group) obtained at 2-day intervals just before injections (arrows). Interaction F (10, 135) = 5.447, P < 0.0001; time F (3.127, 84.42) = 14.58, P < 0.0001; treatment F (2, 27) = 101.1, P = 0.0001. * P < 0.05, ** P < 0.01, ***, P < 0.001, Vehicle+GTN vs Vehicle+Vehicle; †† P < 0.01, ††† P < 0.001, Vehicle+GTN vs PrNMI+GTN; ‡ P < 0.001, PrNMI+GTN vs Vehicle+Vehicle (two-way RM ANOVA, Tukey’s multiple comparisons test). B: baseline hindpaw withdrawal responses to mechanical stimuli (mean ± SEM) of 2 groups of C57/BL6J female mice (10/group) obtained at 2-day intervals just before injections (arrows) and continuing until recovery to pre-injection responses. Interaction F(19, 342) = 5.624, P < 0.0001; time F(4.234, 76.22) = 18.58, P < 0.0001; treatment F(1, 18) = 8.233, P < 0.05. *** P < 0.001 (Vehicle + GTN vs PrNMI + GTN, two-way RM ANOVA, Šidák’s multiple comparisons test).

3.5. PrNMI posttreatment does not suppress GTN-induced acute or chronic mechanical allodynia

We next determined if posttreatment with PrNMI will be effective in suppressing GTN-induced behavioral hypersensitivity. Significant mechanical allodynia was observed in the periorbital region of female mice at 2.5 hrs after GTN (10 mg/kg, i.p.) administration (Fig. 6A). Posttreatment with PrNMI (0.6 mg/kg, i.p), 30 min after GTN injection, had no significant effect on development of acute mechanical allodynia (Fig. 6A). Subsequent posttreatment with PrNMI after each consecutive administration of GTN at 2-day intervals (7 doses total) had no effect on the gradual development of persistent allodynia symptoms (Fig. 6B). In our previous studies, PrNMI was highly effective at reversible suppression of fully-developed chronic pain symptoms induced by peripheral nerve entrapment, bone cancer, or chemotherapeutic drug treatment ^81,97,113. Therefore, we also tested the effect of PrNMI administration on the periorbital thresholds at 2 days after the last dose of GTN. PrNMI (0.6 mg/kg) administration increased withdrawal thresholds in the vehicle and PrNMI posttreatment groups, but the magnitude of allodynia symptom relief was quite small (Fig. 6C) compared to its effectiveness as a pretreatment (cf. Figs. 1, 2, and 5). Behavioral testing continued until return to pre-injection thresholds. Noteworthy, animals tested for periorbital thresholds took considerably longer to recover to baseline than animals tested for hindpaw thresholds (compare Fig. 5B with Fig. 6B).

Figure 6. Posttreatment with PrNMI does not prevent development of acute or chronic GTN-induced mechanical allodynia.

A: changes in head withdrawal responses (mean ± SEM) to periorbital mechanical stimuli of female C57BL/6J mice (n=10/group) 2.5 hrs after injection of GTN (10 mg/kg, i.p.) and 2 hrs after posttreatment with either vehicle or PrNMI (0.6 mg/kg, i.p). *** P < 0.001 (Pre vs Post, two-way ANOVA, Šidák’s multiple comparisons test). B: baseline head withdrawal thresholds (mean ± SEM) of the same 2 groups of C57/BL6J female mice (n=10/group) as in A obtained at 2-day intervals before injections (arrows). Treatment F(1, 18) = 0.918, P = 0.35; time F(3.63, 65.37) = 34.53, P < 0.0001; interaction F(18, 324) = 0.234, P = 0.9996. ♣ P < 0.05 (measurement before 1^st GTN injection vs pre-injection, two-way RM ANOVA, Šidák’s multiple comparisons test). C: changes in head withdrawal thresholds (mean ± SEM) obtained 48 hrs after the 7^th dose of GTN and 2 hrs after injection of PrNMI (0.6 mg/kg, i.p., ▲ in B). * P < 0.05 (Pre vs Post, two-way ANOVA, Šidák’s multiple comparisons test).

3.6. Chronic GTN administration potentiates ASIC and TRPA1 currents in TG neurons

Non-invasive hyperalgesic priming of dural afferents was recently demonstrated to increase behavioral responsiveness of mice to mild acidic stimulation of dural afferents ²⁰. To determine how chronic administration of GTN affects the function of ASICs, we first recorded low pH (6.0 and 6.5)-evoked currents in TG neurons acutely dissociated from chronic vehicle- and GTN (10 mg/kg, every 2 days, 7 doses)-treated mice. Acute dissociation procedures commenced at 24 hrs after the last vehicle or GTN dose. Applications of pH 6.0 or 6.5 mainly activate ASIC1-like and ASIC3-like currents, as ASIC2-like and TRPV1 currents have been described to be activated by more drastic acidifications ⁷⁰. While ASIC1a and ASIC2 are widely distributed both in the periphery and CNS, the expression of ASIC1b and ASIC3 is restricted to peripheral sensory neurons ^11,28,106. In our recordings, we separated low pH-evoked responses into ASIC1- and ASIC3-like, based on the absence or presence of a low-amplitude sustained current that persists beyond the period of low pH application, respectively ³. Comparison revealed that both ASIC1- and ASIC3-like currents were significantly larger in neurons from GTN- versus vehicle-treated mice (Fig. 7A, B). We also compared responses to low pH in neurons of different size, demonstrating significantly larger ASIC currents in small and medium, but not large, TG neurons from GTN- versus vehicle-treated mice (Fig. 7C). To determine if the sensitivity of ASICs to low pH was also affected by chronic GTN treatment we determined the proportion of TG neurons responding to pH 6.0 and pH 6.5 application from vehicle- versus GTN-treated mice. Analysis revealed that 68/115 neurons from vehicle-treated mice (59%) were responsive to pH 6.0 application compared to 59/92 neurons (64%) from GTN-treated mice (P >0.05, Fisher’s exact test), while only 9/112 neurons from vehicle-treated mice (8%) were responsive to pH 6.5 application compared to 21/92 neurons (23%) from GTN-treated mice (P < 0.01, Fisher’s exact test), respectively, indicating a significant increase in pH sensitivity of TG afferent neurons after chronic GTN treatment.

Figure 7. Increased activation of pH 6.0-evoked ASIC currents in TG neurons from mice after chronic GTN administration.

A: representative traces of ASIC1-like currents evoked by pH 6.0 in TG neurons from vehicle- and GTN-treated mice. B: ASIC1-like peak current amplitudes (normalized to cell capacitance) from vehicle- (n=42 neurons/20 mice) and GTN- (n=39 neurons/17 mice) treated mice. ** P < 0.01, unpaired t-test. C: representative traces of ASIC3-like currents in TG neurons from vehicle- and GTN-treated mice. D: ASIC3-like peak current amplitudes (normalized to cell capacitance) from vehicle- (n=25 neurons/14 mice) and GTN- (n=20 neurons/14 mice) treated mice. *** P < 0.001, unpaired t-test. Note the significantly larger amplitude and faster desensitizing ASIC1-like currents compared to ASIC3-like currents. E: ASIC current amplitudes (normalized to cell capacitance) recorded in small-, medium-, and large-size TG neurons from vehicle- and GTN-treated mice. * P < 0.05, *** P < 0.001, unpaired t-test.

In separate experiments, we compared activation of TRPA1 currents in TG neurons acutely dissociated from chronic vehicle- and GTN (10 mg/kg, every 2 days, 7 doses)-treated mice. TRPA1 channels are activated by a variety of endogenous and exogenous stimuli including AITC, reactive oxygen and nitrogen species, reactive prostaglandins, and many environmental irritants such as chlorine, formaldehyde, cigarette smoke, and acrolein ¹³. Recent studies revealed a central role of TRPA1 channels in the development of periorbital mechanical allodynia after systemic GTN injection ⁷². In C57BL/6J mouse sensory neurons, TRPA1 is expressed predominantly in small-sized IB4 +ve nociceptors ^9,51,83. Therefore, we focused on recording AITC-evoked currents from small-diameter (< 25 mm) TG neurons followed by assessment of their IB4 labeling (Fig. 8A). Twenty three of twenty nine (79.3%) small-diameter neurons that responded to AITC were IB4 +ve (Fig. 8B), which was consistent with the previously reported findings that 79% of all small-diameter neurons from the dorsal root ganglia (DRG) of male adult C57BL/6J mice that responded to AITC were IB4-positive ⁹. TRPA1 currents were also significantly potentiated in TG neurons from GTN-treated mice compared to vehicle-treated mice (Fig. 8C, D).

Figure 8. Chronic GTN administration increases activation of TRPA1 currents in small diameter TG neurons.

A: brightfield and fluorescent images of IB4 +ve (arrows) and IB4 –ve (arrowheads) TG neurons. B: responsiveness of small diameter (< 25 μm) TG neurons (n = 29) to AITC (60 μM) application. C: representative traces of current responses to 5-sec AITC (60 μM) application in TG neurons from vehicle- and chronic GTN-treated mice. D: amplitudes of AITC (60 μM)-evoked currents are significantly larger in TG neurons from chronic GTN-treated mice (n = 14) compared to vehicle-treated mice (n = 15). ** P < 0.01, unpaired t-test.

3.7. PrNMI pretreatment prevents chronic GTN-induced increases in expression of sensitization biomarkers within the TG

To explore the second messenger mechanisms involved in the development of chronic GTN-induced hypersensitivity, we studied the changes in expression of various sensitization biomarkers within trigeminal ganglia. For this, the Vehicle+Vehicle, Vehicle+GTN, and PrNMI+GTN groups of mice were euthanized at 48 hrs after the 7th injection and bilateral TG were collected for biochemical analysis. We focused on ASICs and TRPA1 channels, as well as those enzymes and receptors whose activation is known to modulate or be modulated by the adenylyl cyclase/cAMP system. All of the antibodies we used have been extensively characterized in numerous publications ^{32,34,54,63,75,108,112}. When available, we also incubated samples with a blocking peptide during antibody optimization to ensure specificity of target protein bands and their peaks in the spectral graphs (see Supplemental Fig. S3). Chronic GTN administration resulted in significant increases in the expression of nNOS, p-PKA, and TRPA1 proteins in the TG from the Vehicle+GTN group as compared to the Vehicle+Vehicle group (Fig. 9A–D). These GTN-induced increases in nNOS, p-PKA, and TRPA1 protein levels in TG were prevented in the PrNMI+GTN group. Additionally, in the absence of changes in ASIC1 expression, there was a trend to increased expression of ASIC3, PAR2, and CB1R proteins which did not reach statistical significance; these trends were also prevented by PrNMI pretreatment (Fig. 9E–H).

Figure 9. PrNMI prevents GTN-induced alterations in expression of sensitization biomarkers in trigeminal ganglia.

(A–H): spectral graphs on left show group averages (n=9–10 mice/group) of target peaks: GAPDH, PKA (A); GAPDH, phospho-PKA (B); β-actin, nNOS (C); β-actin, TRPA1 (D); β-actin, ASIC1 (E); β-actin, ASIC3 (F); β-actin, PAR2 (G); GAPDH, CB1 (H). Bar graphs on the right are group averages of target proteins normalized to total protein and to the mean of the Vehicle+Vehicle group. * P < 0.05, ** P < 0.01, one-way ANOVA, Šidák’s multiple comparisons test.

3.8. PrNMI pretreatment prevents chronic triptan-induced allodynia and latent sensitization

Female mice received repeated administration of sumatriptan (SUMA; 0.6 mg/kg, i.p.) or vehicle and pretreatment with PrNMI (0.6 mg/kg, i.p.) or vehicle 90 minutes prior to SUMA or vehicle injections. SUMA was administered every 2 days for a total of 7 doses. Basal sensitivity to mechanical and cold stimuli applied to the periorbital region was determined prior to SUMA administration and head withdrawal thresholds were measured every 2 days immediately before injections of SUMA. Repeated injections of SUMA led to increases in baseline sensitivity to periorbital mechanical and cold stimuli in the vehicle-pretreated group, whereas the PrNMI-pretreated group and the Vehicle+Vehicle group maintained stable baseline thresholds (Fig. 10A, B). Upon discontinuation of SUMA injections, thresholds to mechanical and cold stimuli for the Vehicle+SUMA group slowly recovered to pre-triptan levels. Prior to the repetitive injections of SUMA, we determined that a GTN dose of 0.5 mg/kg did not elicit significant acute allodynia responses to mechanical (Fig. 8C) or cold (Fig. 8D) stimuli. After SUMA administration and once thresholds returned to pre-SUMA levels, mice were challenged with a lower dose of GTN (0.3 mg/kg, i.p.). Upon challenge with 0.3 mg/kg GTN, the Vehicle+SUMA group now exhibited significant acute allodynia, indicative of latent sensitization, whereas the PrNMI+SUMA, PrNMI+Vehicle, and Vehicle+Vehicle groups showed no significant responses to 0.3 mg/kg of GTN (Fig. 10C, D). Importantly, repetitive injections of PrNMI alone did not produce changes in baseline thresholds or latent sensitization. In separate groups of mice, PrNMI pretreatment also prevented SUMA-induced hindpaw mechanical allodynia and latent sensitization (Supplemental Fig. S4).

Figure 10. PrNMI prevents triptan overuse-induced allodynia and latent sensitization.

(A, B): Changes in baseline responses to mechanical (A) and cold (B) stimuli in Vehicle+sumatriptan (SUMA), PrNMI+SUMA, Vehicle+Vehicle, and PrNMI+Vehicle (arrows) treatment groups (n=10/group). Mechanical: interaction F(42, 504) = 6.48; time F(14, 504) = 9.08; treatment F(3, 36) = 19.16. Cold: interaction F(42, 504) = 4.58; time F(14, 504) = 5.07; treatment F(3, 36) = 23.65. * P < 0.05, two-way RM ANOVA). Vehicle SUMA group vs. other groups. (C, D): Acute responses to mechanical (C) and cold (D) stimuli before (Pre) and 2 hrs after (Post) GTN (0.5 mg/kg) administration. Tests were performed before and after repetitive SUMA administration in Vehicle and PrNMI pretreatment groups. * P < 0.05 (Pre vs Post of after SUMA (GTN 0.3 mg/kg), paired t-test). Note the lack of significant responses to GTN (0.3 mg/kg) in the PrNMI+SUMA and the PrNMI+Vehicle groups.

3.9. PrNMI prevents chronic triptan-induced increases in expression of sensitization biomarkers in TG

At the conclusion of behavioral testing for triptan-induced latent sensitization, mice were euthanized and bilateral TG were collected from all groups for biochemical analysis. After chronic sumatriptan administration there were significant increases in the expression of nNOS, and TRPA1 proteins in TG from the Vehicle+SUMA group as compared to the Vehicle+Vehicle group (Fig. 11A, B). These increases in nNOS and TRPA1 protein levels due to SUMA exposure were prevented in TG from the PrNMI-treated groups. Additionally, there was a trend to increased expression of p-PKA and p-PKA/PKA ratio which did not reach statistical significance; these trends were also prevented by PrNMI pretreatment (Fig. 11C, D). Chronic administration of PrNMI alone did not result in any statistically significant increase in any sensitization biomarker protein levels, as compared to the Vehicle+Vehicle group; however, expression of p-PKA, p-PKA/PKA ratio, and TRPA1 was significantly reduced compared to the Vehicle+SUMA group (Fig. 11B, C, E).

Figure 11. Triptan-induced alterations in expression of TG neuron sensitization biomarkers are prevented by PrNMI.

A–H: bar graphs are group (n=9–10/group) averages of target proteins normalized to total protein and to the mean of the Vehicle + Vehicle group. In the absence of changes in PKA (A), there was a trend to increased expression of p-PKA (B) and consequently the ratio of p-PKA/PKA (C) in the Vehicle + SUMA group. This trend was not seen in the PrNMI +SUMA group, while both p-PKA and p-PKA/PKA were significantly reduced in the PrNMI + Vehicle compared to the Vehicle + SUMA group. The nNOS (D) and TRPA1 (E) proteins were significantly increased in the Vehicle + SUMA group compared to the Vehicle + Vehicle group. These increases were not significant in the PrNMI+SUMA group, while TRPA1 expression was significantly reduced in the PrNMI + Vehicle compared to the Vehicle + SUMA group. Expression of ASIC3 (F), PAR2 (G), and CB1R (H) was not significantly altered by SUMA and PrNMI treatments. * P < 0.05, ** P < 0.01, one-way ANOVA, Šidák’s multiple comparisons test.

4. Discussion

4.1. Pretreatment with PrNMI prevents GTN-induced mechanical allodynia

We demonstrate that pretreatment with the synthetic peripherally-restricted mixed CB1R/CB2R agonist, PrNMI ⁹⁷, suppresses development of acute periorbital and hindpaw allodynia in the GTN-induced mouse migraine model (Fig. 1). The preventative anti-allodynic effects of PrNMI are mediated by both CB1R and CB2R (Fig. 2), with no apparent sex differences in its effectiveness. This contrasts with the primarily CB1R-mediated posttreatment effectiveness of PrNMI in other chronic pain models ^81,97,113, indicative of differences in migraine pathophysiology from that of other chronic pain states. We also show that PrNMI suppresses ASIC and TRPA1 currents in TG neurons entirely by CB1R activation (Figs. 3 and 4). Given the virtual absence of CB2R mRNA in TG neurons ⁹², this also suggests that the CB2R-mediated analgesic effect of PrNMI is likely due to activation of peripheral CB2Rs on immunocompetent cells. Dural nociceptors release various pro-inflammatory neuropeptides, including CGRP, substance P, and proteases. GTN-induced release of CGRP at various sites of the trigeminovascular complex, including the dura mater ^59,95 and trigeminal ganglia ^26,60, in turn, results in mast cell degranulation and secretion of vasoactive, pro-inflammatory, and neurosensitizing mediators, thereby contributing to migraine pathogenesis ^68,94,103. Indeed, a recent study demonstrated contribution of both CB1Rs and CB2Rs to the anti-allodynic effects of a brain-permeant synthetic cannabinoid, methanandamide, in the GTN-induced migraine model ⁵⁹. Importantly, they showed that methanandamide prevented CGRP release via CB1R activation, but mast cell degranulation was prevented by activation of CB2Rs ⁵⁹.

4.2. PrNMI suppresses ASIC and TRPA1 currents via CB1R activation in TG neurons.

Both ASICs and TRPA1 channels have been implicated in trigeminovascular sensitization in migraine headaches ^13,41,111. Their increased activation in sensitized dural afferents could allow for the generation of action potentials with previously ineffective stimuli, resulting in increased transmission of pain signals to the CNS ^20,42,66. We show that PrNMI, via CB1R activation, reversibly suppresses these currents in TG neurons isolated from naïve mice (Figs. 3 and 4). Our data are consistent with a previous demonstration of CB1R-mediated suppression of ASIC currents by the brain-permeant cannabinoid, WIN 55,212–2 ⁷¹. However, we are the first to demonstrate CB1R-mediated suppression of TRPA1 currents by PrNMI, since all of the previous studies have shown TRPA1 to be activated by various endogenous and exogenous cannabinoids ^55,80.

We are also first to show that chronic GTN treatment potentiates ASIC1- and ASIC3-like currents in small and medium TG neurons (Fig. 7), and TRPA1 currents in small diameter (mostly IB4+ve) neurons (Fig. 8). Noteworthy, the sensitivity of ASICs to pH 6.5 was also increased in neurons from sensitized mice. Since ASICs ^67,108 and TRPA1 channels ⁷⁴ are downstream targets of cAMP signaling produced by adenylyl cyclase, our results suggest that peripheral activation of G_i-protein-coupled CB1Rs prevents acute and chronic sensitization, at least in part, by decreasing cAMP-dependent phosphorylation of ASICs and TRPA1 channels in trigeminal neurons.

4.3. PrNMI prevents GTN-induced increases in expression of sensitization biomarkers in TG.

Chronic GTN treatment increased p-PKA, nNOS, and TRPA1 proteins in trigeminal ganglia (Fig. 9). This confirms previously demonstrated increases in expression of nNOS ¹⁰² and TRPA1 ^35,72,85 in chronic migraine models, and demonstrates for the first time, chronic GTN-induced increases in p-PKA, a downstream target of cAMP produced by adenylyl cyclase ⁶⁹. Phosphorylated PKA, in turn, increases the activity of numerous downstream targets such as voltage-gated Na^{+ 48} and Ca^{2+ 38,77,82} channels, enzymes such as nNOS ^52,108, and ligand-gated ASIC ⁶⁷, TRPA1 ⁷⁴, and TRPV1 ⁵³ channels. All these components could amplify the inflammatory cascade and promote activation and sensitization of trigeminal afferents, leading to the observed behavioral hypersensitivity. For example, increased nNOS activity producing excessive amount of nitric oxide (NO) ¹¹⁵ is consistent with NO’s causative role in initiating and maintaining migraine headache ⁸⁸. The TRPA1 channel is a major sensor of reactive oxygen ¹⁵ and nitrogen species, which includes NO ⁷⁶ produced by nNOS. NO was shown to release CGRP, a known mediator of dural afferent sensitization ^56,65, in vitro ¹⁰⁹ and in vivo ⁴⁴ through activation of TRPA1.

The GTN-induced increases in TRPA1 expression could account for the increased magnitude of AITC-evoked currents. However, functional TRPA1 channels are also expressed in satellite glial cells (SCGs) of sensory ganglia and their activation in SGCs is enhanced after both peripheral inflammation and nerve injury ⁹⁸. Since we measured protein content in homogenized TG samples, it is possible that the observed increases in TRPA1 protein are due entirely to the increased TRPA1 expression in SGCs. If so, the GTN-induced increases in AITC-evoked currents may be due to increases in TRPA1 phosphorylation rather than increases in TRPA1 expression. The lack of significant GTN-induced increases in expression of neuronal ASIC1 and ASIC3, suggests that the increases in their sensitivity and activation by low pH may be due entirely to increased channel phosphorylation ³⁶. Additional studies are needed to resolve between these possibilities.

Importantly, the increases in biomarkers of neuronal sensitization were prevented by pretreatment with PrNMI. Our data provide further support for the hypothesis that peripheral CB1R activation prevents acute and chronic sensitization, at least in part, by decreasing cAMP-dependent phosphorylation of nNOS, ASICs, and TRPA1 channels in trigeminal neurons.

4.4. PrNMI posttreatment does not suppress GTN-induced acute or chronic mechanical allodynia

The ineffectiveness of PrNMI as a posttreatment is in marked contrast to our previous studies where PrNMI effectively suppressed fully-developed chronic pain symptoms induced by peripheral nerve entrapment, bone cancer, or chemotherapeutic drug treatment ^81,97,113. However, it agrees with a recent study which demonstrated suppression of AITC-evoked migraine-like symptoms by Δ⁹-tetrahydrocannabinol (Δ⁹-THC) only when concurrently administered with AITC, but not as a posttreatment ⁵⁷. These findings are further indication of the differences in migraine pathophysiology from other chronic pain states and suggest that cannabinoids may be more effective as preventative medications rather than abortive treatments of acute migraine episodes. The animal data are supported by a retrospective human study where self-medicating with different cannabis strains after the onset of acute migraine episodes reduced their severity but did not abort the attacks ³⁰.

4.5. PrNMI prevents chronic triptan-induced allodynia and latent sensitization

Our results are consistent with previous findings of gradual decreases in head and hindpaw withdrawal thresholds and increased susceptibility to previously ineffective migraine triggers after repeated sumatriptan administration, confirming that triptan overuse produces both cephalic and extra-cephalic hypersensitivity symptoms ^19,32,79,114. Our results are also consistent with findings of increased nNOS levels in TG after sumatriptan exposure ³². We also observed increased TRPA1 expression and a trend to increased p-PKA expression, which has not been previously reported. These endured >25 days after the last sumatriptan dose, suggesting that such neuroplastic changes likely contribute to the development of latent sensitization. Pretreatment with PrNMI completely suppressed the development of triptan-induced hypersensitivity and latent sensitization, and prevented the increases in nNOS, TRPA1, and p-PKA. Repetitive PrNMI administration alone did not induce any hypersensitivity symptoms or latent sensitization, in contrast to the previously reported development of latent sensitization after repetitive administration of brain-permeant cannabinoids WIN 55,212–2 and Δ⁹-THC ⁶². This underscores the differences between brain-permeant and peripherally-restricted cannabinoids, and suggests that central, but not peripheral activation of CB1Rs may have contributed to the development of latent sensitization in that study ⁶². Noteworthy, a recent study revealed that while some tolerance to effects of cannabis on headache and migraine was detected after chronic treatment, increases in MOH were not detected ³⁰. Given the side effects and MOH-predisposing actions of brain-permeant opioids, triptans, and cannabinoids, further development of peripherally-targeted treatments, exemplified by the efficacy of anti-CGRP monoclonal antibodies in migraine prevention, is warranted.

4.6. Study limitations

The GTN migraine model ^78,91 involves high-dose GTN injection which may cause various adverse systemic changes (e.g., cardiovascular effects) not related to hyperalgesia, as well as widespread nociceptive effects not related to migraine. Future studies will need to consider using migraine models which produce selective sensitization of dural afferents (e.g., ref. ²⁰). We demonstrated increased function and expression of only a few ligand-gated ion channels and other biomarkers of trigeminal neuron sensitization. Future studies will need to address the involvement of other ligand- and voltage-gated ion channels (e.g., voltage-gated sodium channel 1.9, recently shown to contribute to GTN- and sumatriptan-induced behavioral sensitization ¹⁹) and other second messengers (e.g., PKCε, shown to be critical in hyperalgesic priming, a model of acute to chronic pain transition ^18,89,90). Also, we only examined allodynia symptoms, but not the autonomic symptoms of migraine such as nausea, photophobia, and phonophobia. Further study will need to determine the efficacy of PRCBs such as PrNMI in alleviating the autonomic symptoms of migraine.

5. Conclusions

In conclusion, our data provide further insights into the cellular mechanisms that underlie chronic migraine development. We also highlight the efficacy of selective peripheral CBR activation in preventing the behavioral symptoms and sensitization of trigeminal neurons in mouse models of migraine and MOH. These data validate peripheral CBRs as potential therapeutic targets in migraine and MOH. Further studies will be necessary to pursue clinical development of PRCBs as a preventative treatment of this neurological disorder.

Supplementary Material

Supplementary Figure S1

Supplementary Figure S1. Changes in plasma [PrNMI] following its administration in C57BL/6J male mice (n=2–3 mice/data point).

Supplementary Figure S2. Supplementary Figure S2. CB1 receptor mediated suppression of AITC-induced TRPA1 current by PrNMI (1 μM).

(A and B): Compensated currents were obtained by estimating the average rundown experienced by vehicle-treated cells over time, adding that value to the I_AITC peak amplitude measured during PrNMI (A) or PrNMI + 18A (B) co-application. Note the lack of suppressive effect of PrNMI during co-application with the selective CB1R antagonist, 18A. * P < 0.05, *** P < 0.001, paired t-tests followed by Holm-Šidák correction to obtain adjusted p-values.

Supplementary Figure S3. Supplementary Figure S3. Wes-based analysis of TG samples.

(A and C): examples of target protein bands from capillaries loaded with p-PKA (1:10) or ASIC3 (1:25) primary antibodies (Ab) with or without pre-adsorption with a control antigen (CA) at 2×Ab concentration. (B and D): spectral graphs representing the protein bands in (A and C). Note the loss of signal in the relevant bands in the presence of the control antigen.

Supplementary Figure S4. Supplementary Figure S4. PrNMI prevents triptan overuse-induced hindpaw allodynia and latent sensitization.

A: changes in baseline withdrawal thresholds in response to GTN and sumatriptan (SUMA in vehicle and PrNMI pretreatment groups, n = 10/group) injections (arrows). Treatment F(1, 18) = 0.7503, P = 0.2262; time F(9.404, 169.3) = 1.322, P = 0.2262; interaction F(29, 522) = 0.8846, P = 0.6423; * P < 0.05 (last measurement before 1^st SUMA injection vs subsequent measurements, two-way RM ANOVA, Šidák’s multiple comparisons test). B: acute response to administration of GTN (0.1 or 0.3 mg/kg) before and after repetitive SUMA administration in vehicle and PrNMI pretreatment groups. * P < 0.05 (Pre vs Post of after SUMA (GTN 0.1 or 0.3 mg/kg), paired t-test). Note the lack of significant responses in the PrNMI group after repetitive SUMA administration.