Abstract
Cannabinoid-targeted pain therapies are increasing with the expansion of cannabis legalization, however, their efficacy may be limited by pain-induced adaptations in the cannabinoid system. Cannabinoid receptor subtype 1 (CB1R) inhibition of spontaneous, GABAergic miniature IPSCs (mIPSCs) and evoked IPSCs (eIPSCs) in the ventrolateral periaqueductal gray (vlPAG) were compared in slices from naive and inflamed male and female Sprague Dawley rats. Complete Freund's Adjuvant (CFA) injections into the hindpaw induced persistent inflammation. In naive rats, exogenous cannabinoid agonists robustly reduce both eIPSCs and mIPSCs. After 5–7 d of inflammation, the effects of exogenous cannabinoids are significantly reduced because of CB1R desensitization via GRK2/3, as function is recovered in the presence of the GRK2/3 inhibitor, Compound 101 (Cmp101). Inhibition of GABA release by presynaptic μ-opioid receptors in the vlPAG does not desensitize with persistent inflammation. Unexpectedly, while CB1R desensitization significantly reduces the inhibition produced by exogenous agonists, depolarization-induced suppression of inhibition protocols that promote 2-arachidonoylglycerol (2-AG) synthesis exhibit prolonged CB1R activation after inflammation. 2-AG tone is detected in slices from CFA-treated rats when GRK2/3 is blocked, suggesting an increase in 2-AG synthesis after persistent inflammation. Inhibiting 2-AG degradation with the monoacylglycerol lipase (MAGL) inhibitor JZL184 during inflammation results in the desensitization of CB1Rs by endocannabinoids that is reversed with Cmp101. Collectively, these data indicate that persistent inflammation primes CB1Rs for desensitization, and MAGL degradation of 2-AG protects CB1Rs from desensitization in inflamed rats. These adaptations with inflammation have important implications for the development of cannabinoid-based pain therapeutics targeting MAGL and CB1Rs.
SIGNIFICANCE STATEMENT Presynaptic G-protein-coupled receptors are resistant to desensitization. Here we find that persistent inflammation increases endocannabinoid levels, priming presynaptic cannabinoid 1 receptors for desensitization on subsequent addition of exogenous agonists. Despite the reduced efficacy of exogenous agonists, endocannabinoids have prolonged efficacy after persistent inflammation. Endocannabinoids readily induce cannabinoid 1 receptor desensitization if their degradation is blocked, indicating that endocannabinoid concentrations are maintained at subdesensitizing levels and that degradation is critical for maintaining endocannabinoid regulation of presynaptic GABA release in the ventrolateral periaqueductal gray during inflammatory states. These adaptations with inflammation have important implications for the development of cannabinoid-based pain therapies.
Introduction
Cannabinoids are neuromodulators that regulate neurotransmitter release from presynaptic terminals. The synthesis of endogenous cannabinoid ligands and the expression of cannabinoid receptors are highly plastic. The cannabinoid 1 receptor (CB1R) is one of the most highly expressed G-protein-coupled receptors (GPCRs) in the brain (Herkenham et al., 1990) and is primarily localized to presynaptic terminals, where its activation inhibits neurotransmitter release (Katona et al., 1999; Vaughan et al., 2000; Vaughan and Christie, 2005; Mikasova et al., 2008). CB1Rs are activated by exogenous and endogenous ligands. Endogenous cannabinoid ligands, endocannabinoids (eCBs), are synthesized on-demand in postsynaptic neurons and travel retrogradely to activate receptors expressed on presynaptic terminals. The two most well studied eCBs are 2-arachidonoylglycerol (2-AG) and anandamide (AEA). In the brain, levels of 2-AG are >100 times higher than AEA (Stella et al., 1997). 2-AG levels are regulated by the synthesis enzyme diacylglycerol (DAGL; Bisogno et al., 2003) and the catabolism enzyme, monoacylglycerol lipase (MAGL; Dinh et al., 2002, 2004). These synthesis and degradation enzymes tightly control 2-AG levels (for review, see Ahn et al., 2008) and the subsequent CB1R suppression of neurotransmitter release from the presynaptic terminal.
CB1Rs are expressed in high levels throughout the brain and throughout the rostral–caudal axis of the ventrolateral periaqueductal gray (vlPAG; Tsou et al., 1998; Wilson-Poe et al., 2012), a key region within the descending pain modulatory circuit. Activation of CB1Rs in the vlPAG inhibit GABA release from presynaptic terminals (Vaughan et al., 2000) by reducing the probability of release (Vaughan et al., 2000; Drew et al., 2008, 2009) and shifting the mode of vesicle release from multivesicular to univesicular release (Aubrey et al., 2017). Repeated administration of exogenous cannabinoids, including tetrahydrocannabinol (THC) and WIN 55,212–2, result in decreased CB1R signaling measured with GTPγS assays (Sim et al., 1996; Breivogel et al., 1999; Rubino et al., 2000; Kouznetsova et al., 2002; Lazenka et al., 2014), suggesting CB1R desensitization; however, the synaptic mechanisms underlying this desensitization are not known. Presynaptic GPCRs are resistant to acute desensitization (Wetherington and Lambert, 2002; Fyfe et al., 2010; Pennock and Hentges, 2011; Pennock et al., 2012; Pennock and Hentges, 2016). In fact, presynaptic μ-opioid receptors (MORs) become sensitized after multiple days of morphine treatment (Ingram et al., 1998). Given that painful stimuli and stress increase the release of eCBs in the PAG (Walker et al., 1999; Hohmann et al., 2005; Suplita et al., 2005; Petrosino et al., 2007), we examined the synaptic mechanisms underlying adaptations in CB1R function after persistent inflammation.
The present results show that persistent inflammation increases eCB release in the vlPAG, leading to CB1R desensitization through a GPCR kinase 2/3 (GRK2/3)-dependent mechanism. While this desensitization is clearly observed with exogenous CB1R agonists, eCB efficacy, measured with depolarization-induced suppression of inhibition (DSI), is prolonged after inflammation. DSI-induced release of 2-AG results in CB1R desensitization if the degradation of 2-AG by MAGL is impaired. Together, these results illuminate a mechanism by which inflammation primes CB1Rs for desensitization and rapid eCB degradation by MAGL protects CB1Rs from desensitization by eCBs.
Materials and Methods
Animals
Adult male and female Sprague Dawley rats (Harlan Laboratories and bred in-house; age range, postnatal day 29–73; median age, 49 d) were used for all experiments. All procedures were performed in strict accordance with the Guide for the Care and Use of Laboratory Animals as adopted by the Institutional Animal Care and Use Committee of Oregon Health & Science University and the University of Colorado Anschutz Medical Campus. Care was taken to minimize discomfort.
Inflammation
Complete Freund's Adjuvant (CFA; heat-killed Mycobacterium tuberculosis in mineral oil, 1 mg/ml, 0.1 ml volume injected; Sigma-Aldrich) was injected subcutaneously into the plantar surface of the right hindpaw. This CFA injection produces an intense tissue inflammation of the hindpaw characterized by erythema, edema, and hyperalgesia throughout the time course of 5–7 d (Iadarola et al., 1988; Wang et al., 2006; Tsutsui et al., 2011; Eidson and Murphy, 2013; Djouhri et al., 2015). Electrophysiological recordings and tissue dissections were performed 5–7 d following CFA injection.
vlPAG slice preparation
Slices containing the vlPAG were prepared as previously described (Tonsfeldt et al., 2016; Bouchet et al., 2021; McPherson et al., 2021, 2023). Rats were deeply anesthetized with isoflurane, and the brain was rapidly removed and placed in ice-cold sucrose-based cutting buffer containing the following (in mm): 75 NaCl, 2.5 KCl, 0.1 CaCl2, 6 MgSO4, 1.2 NaH2PO4, 25 NaHCO3, 2.5 dextrose, and 80 sucrose. Slices containing the vlPAG were cut to a thickness of 220 µm on a vibratome (Leica Microsystems) in sucrose-based cutting buffer and were transferred to a holding chamber with aCSF containing the following (in mm): 126 NaCl, 21.4 NaHCO3, 22 dextrose, 2.5 KCl, 2.4 CaCl2, 1.2 MgCl2, and 1.2 NaH2PO4, and the osmolarity was adjusted to 300–310 mOsm. Slices were oxygenated with 95% O2 and 5% CO2 until transfer to a recording chamber on an upright microscope (model BX51WI, Olympus) and superfused with aCSF maintained at 32°C.
Whole-cell patch-clamp recordings
Voltage-clamp recordings (holding potential, −70 mV) were made in whole-cell configuration using an amplifier (model Axopatch 200B, Molecular Devices). Patch-clamp electrodes were pulled from borosilicate glass (diameter, 1.5 mm; WPI) on a two-stage puller (catalog #PP83, Narishige). Pipettes had a resistance of 2.5–5 MΩ. IPSCs were recorded using an intracellular pipette solution containing the following (in mm): 140 CsCl, 10 HEPES, 4 MgATP, 3 NaGTP, 1 EGTA, 1 MgCl2, and 0.3 CaCl2, at pH adjusted to 7.3 with CsOH, and osmolarity at 290–300 mOsm. QX314 (100 μm) was added to the internal solution for evoked IPSC (eIPSC) experiments to reduce action potentials in the recording cell. Access resistance was continuously monitored. Recordings in which access resistance changed by >20% during the experiment were excluded from data analysis. A junction potential of −5 mV was corrected during recording. GABAergic events were isolated in the presence of glutamate receptor antagonist 2,3-dihydroxy-6-nitro-7-sulfonyl-benzo[f]quinoxaline (NBQX; 5 μm). Spontaneous miniature IPSCs (mIPSCs) were recorded in the presence of NBQX (5 μm) and tetrodotoxin (TTX; 500 nm). Events were low-pass filtered at 2 kHz and sampled at 10–20 kHz for offline analysis (Axograph 1.7.6), and individual events were visually confirmed. In experiments using exogenous cannabinoid agonists, one neuron was recorded per slice because of the lipophilic nature of cannabinoid receptor drugs. Agonists were added to the bath after a stable baseline was achieved in NBQX for at least 5 min. Cannabinoid agonists were superfused for 15 min, after which a cannabinoid antagonist was added to the bath for 15 min. eIPSCs were averaged when maximal drug effect was achieved between 10 and 15 min. After each experiment with exogenous cannabinoid agonists or antagonists, lines were washed with 50% EtOH.
Depolarization-induced suppression of inhibition
After obtaining stable eIPSCs, a protocol for DSI collected two eIPSCs for baseline measurement, followed by a brief depolarizing step (5 s at +20 mV; Wamsteeker et al., 2010) before returning to the holding potential (−70 mV). eIPSCs were evoked at 0.2 Hz for 90 s following the depolarizing step, and amplitudes were normalized to the average baseline eIPSC amplitude. Not all cells were sensitive to the DSI protocol, which is consistent with observations from other laboratories (Ohno-Shosaku et al., 2001). Cells were grouped into “DSI” or “No DSI” by their eIPSC amplitude within 15 s of the depolarizing current (DSI, >5% inhibition of eIPSC). Because of the high level of variability, a higher n was required for the DSI experiments relative to experiments using exogenous cannabinoid drugs, consistent with previous observations (Ohno-Shosaku et al., 2001; Wilson and Nicoll, 2001; Wamsteeker Cusulin et al., 2014). The DSI protocol was also run after equilibration of the slice with the CB1R antagonist NESS-0327 (NESS; 0.5 μm). Experiments using DO34 and JZL184 incubated slices in the drugs for at least 1 h before recordings, and drugs were included in the bath for these experiments. For these datasets, all experiments were used in the analysis (i.e., no determination of whether cells responded to DSI protocol was made).
Drugs
WIN 55,212–2 (Cayman Chemical), rimonabant [RIM (SR141716A); Cayman Chemical], and NESS-0327 (Tocris/Biotechne) were dissolved in DMSO, aliquoted, and stored at −20°C. CP55,940 and AM251 (Cayman Chemical) were dissolved in methanol and stored at −20°C. DMSO and methanol at appropriate concentrations were used as vehicle controls. NBQX (Sheardown et al., 1990), [d-Ala(2),N-Me-Phe(4),Gly(5)-ol]-enkephalin (DAMGO), naloxone and, TTX were purchased from Abcam, dissolved in distilled water, and stored at 4°C. Compound 101 (Cmp101; 3-[(4-methyl-5- pyridin-4-yl-1,2,4-triazol-3-yl)methylamino]-N-[[2-(trifluoromethyl) phenyl]methyl]benzamide hydrochloride) was purchased from Hello Bio and was prepared as described previously (Leff et al., 2020). Briefly, Cmp101 (made fresh daily) was first dissolved in a small amount of DMSO (10% of final volume), sonicated, then brought to its final volume with 20% 2-hydroxypropyl)-β-cyclodextrin (HPCD) and sonicated again to create a 10 mm solution. For experiments using a higher concentration of Cmp101, Cmp101 was applied to the slice as follows: 30 μm incubation for >1 h, 1 μm maintenance while patching, and 10 μm in drug tubes (Lowe et al., 2015; Leff et al., 2020; Adhikary et al., 2022). For experiments using a lower concentration of Cmp101 (1 μm) was used for incubation (>1 h), maintenance while patching, and in drug tubes. DMSO and 20% HPCD were used as the vehicle control.
Radioligand binding assay
Tissue dissection.
Rats were deeply anesthetized with isoflurane, brains were removed and submerged in ice-cold Tris-HCl buffer, pH 7.4 at 4°C. Over ice, the brain was sectioned into 1 mm slices from which the vlPAG, dorsolateral striatum (DLS), and hypothalamus were dissected and immediately flash frozen on dry ice. Tissue samples were stored at −80°C.
Total particulate tissue preparation.
Tissue preparation was adapted from the study by Eastwood et al. (2018). Since the brain regions sampled are so small, tissue from each brain region from multiple animals (vlPAG, eight animals; DLS, two animals; hypothalamus, two animals) was pooled to ensure ample protein levels for saturation binding. Tissue was removed from −80°C and transferred to a 2 ml tube containing 0.5 ml Tris-HCl, pH 7.4 at 4°C, with protease inhibitor (protease inhibitor cocktail set #539134, EMD Millipore). Tissue was homogenized with a Polytron PT1200E Handheld Homogenizer 4× for 6 s, placing the sample on ice for 20 s between homogenizations. The Polytron was washed with water between each sample. The volume was increased to 1.5 ml, then the sample was transferred to a minicentrifuge and spun at 13,000 × g for 20 min at 4°C. The supernatant was discarded, and the pellet was resuspended in 0.5 ml of Tris-HCl with protease inhibitor. Tissue was homogenized for 7 s and spun as described above once more. After the final spin, the supernatant was discarded, the pellet was resuspended in TME Binding Buffer (200 mm Tris Base, 50 mm MgCl2, and 10 mm EDTA; pH 8.0) with protease inhibitor and homogenized for 10 s. TME Binding Buffer with protease inhibitor was added for a final volume of 1.5 ml. Samples were kept on ice throughout the preparation. Protein levels were determined with a BCA Protein Assay Kit (Thermo Fisher Scientific).
Saturation curve.
Binding assays were conducted in the absence of Na+. [3H]CP-55940 was used to measure cannabinoid receptor binding (Romero et al., 1995; Hill et al., 2008; Chanda et al., 2010; McLaughlin et al., 2013; Catani and Gasperi, 2016; Freels et al., 2020). Binding assays were conducted using five concentrations of [3H]CP-55 940 (0.1–7 nm) in a final volume of 1 ml. Assays were performed in duplicate in a 96-well plate with 50 mm TME Binding Buffer with bovine serum albumin (1 mg/ml), pH 7.4 at 30°C. Nonspecific binding was subtracted from total binding to yield specific binding. Nonspecific binding was determined with 1 μm WIN 55,212–2 and was 59%, 19%, or 55% in naive rat tissue samples, and was 53%, 17%, or 55% in CFA in vlPAG, DLS, and hypothalamus tissue samples, respectively. Prepared membranes were incubated with [3H]CP-55940 at 30°C for 60 min. The incubation was terminated using a cell harvester (Tomtec) by rapid filtration through PerkinElmer Filtermat A filters presoaked in 0.2% polyethylenimine. The filters were dried, spotted with scintillation cocktail, and radioactivity was determined using a microBetaplate 1405 scintillation counter (PerkinElmer).
Experimental design and statistical analysis
In all electrophysiological experiments, each dataset included recordings from at least three rats. For DSI experiments, “Max DSI” averaged the first four eIPSCs after depolarization and “Late DSI” averaged eIPSCs 30–45 s after depolarization. In radioligand binding experiments, three replicates per group were run. All analyses were conducted in Graphpad Prism (version 9.5; Graphpad Software). Values are presented as the mean ± SE, and all data points are shown in bar graphs to illustrate variability. Statistical comparisons were made using two-tailed paired or unpaired t test, or ANOVA when appropriate. In all summary bar graphs, each dot represents an individual cell while the numbers in the bars represent the animal number. When post hoc analysis was appropriate, the Šidák multiple-comparisons test was used. Significance was defined as p < 0.05.
Results
Persistent inflammation reduces CB1R-mediated IPSC inhibition induced by exogenous agonists
Plasticity in presynaptic CB1R function induced by persistent inflammation was examined following the injection of CFA into the hindpaw of male and female Sprague Dawley rats. All experiments were conducted in naive animals or 5-7 d post-CFA injection. Whole-cell patch-clamp recordings of eIPSCs were used to measure evoked and spontaneous GABAergic IPSCs, and the inhibition of GABA release by the nonselective cannabinoid receptor agonist WIN (3 μm; Fig. 1A,B). In tissue from naive animals, WIN reduced eIPSC amplitudes by 57 ± 5% compared with baseline. CFA-induced inflammation significantly attenuated WIN inhibition (18 ± 4%). WIN inhibition was reversed by the CB1R-selective antagonist RIM (3 μm) in slices from both naive and CFA-treated rats. When all experiments were pooled, there was no main effect of sex (two-way ANOVA; F(1,37) = 0.002, p = 0.96; naive male, n = 10; naive female, n = 9; CFA male, n = 13; CFA female, n = 9), so data from male and female rats were combined for all analyses. There were no differences in baseline eIPSC paired-pulse ratios in recordings from naive or CFA-treated rats (unpaired t test, t(16) = 1.23; p = 0.2).
Persistent inflammation reduces WIN-mediated inhibition of GABA release. A, Representative traces of eIPSCs isolated in NBQX recorded from vlPAG neurons in baseline (5 μm; teal), the cannabinoid receptor agonist WIN (3 μm; black) and the CB1R-selective antagonist RIM (3 μm; blue) from naive (left) and CFA-treated (right) animals. B, Percentage of inhibition of eIPSC amplitude by WIN compared with the average of baseline and RIM recovery in slices from naive and CFA-treated rats (unpaired t test; t(14) = 5.34, p < 0.0001). C, D, Representative traces of mIPSCs recorded from vlPAG neurons in baseline containing TTX (500 nm) and NBQX (5 μm), WIN (3 μm), and RIM (3 μm) from slices of naive (C) or CFA-treated rats (D). E, mIPSC frequency at baseline, WIN, and RIM from slices of naive and CFA-treated rats. (F) WIN percentage inhibition of mIPSC frequency from naive and CFA-treated rats (unpaired t test; t(10) = 4.65, p = 0.0005). Error bars represent the SEM, dots indicate individual recordings, and numbers represent the number of rats represented per bar.
To determine whether inflammation also affects the inhibition of spontaneous release by CB1Rs, we measured mIPSCs in the presence of TTX (500 nm). WIN suppressed mIPSC frequency by 56 ± 5% in tissue from naive animals, and this effect was significantly reduced (14 ± 6%) in slices from inflamed animals and was reversed with RIM (Fig. 1C–F). Activating CB1Rs had no effect on mIPSC amplitude (one-way repeated-measures ANOVA: F(1.6,5.5) = 0.43, p = 0.56), which is consistent with a presynaptic effect of CB1R activation.
The reduction in CB1R suppression of GABA release could be the result of a general change in presynaptic GPCR signaling. To test this, we investigated the effects of persistent inflammation on CB2R and MOR inhibition of GABA release. The CB2R agonist AM1241 (3 μm) minimally reduced mIPSC frequencies in vlPAG slices from naive animals (14 ± 4% inhibition), and this was not changed after persistent inflammation (17 ± 10% inhibition; unpaired t test: t(15) = 0.71, p = 0.5). We next tested whether the function of MORs, well known for their ability to modulate presynaptic GABA release in the vlPAG, changes after persistent inflammation. The MOR-selective agonist DAMGO (1 μm) inhibited eIPSCs to the same extent in slices from naive and CFA-treated rats (Fig. 2A,B). An EC50 concentration of DAMGO (100 nm) also did not show evidence of desensitization in slices from CFA-treated rats (data not shown). In addition, DAMGO-induced suppression of mIPSC frequency was unaffected by persistent inflammation (Fig. 2C,D). Together, these data suggest that the effects of persistent inflammation within the vlPAG are selective to CB1Rs.
Persistent inflammation does not affect MOR suppression of GABA release. A, Representative eIPSC traces at baseline: NBQX, 5 μm (teal); DAMGO, 1 μm (black), and naloxone, 1 μm (blue). B, Percentage inhibition of eIPSCs by DAMGO in naive (black bar) and CFA-treated (red bar) conditions (unpaired t test; t(10) = 0.32, p = 0.8). C, Spontaneous mIPSC frequency in slices from naive (black) or CFA-treated (red) animals during baseline, DAMGO administration (1 μm), and naloxone (1 μm). D, DAMGO inhibition of mIPSC frequency inhibition (unpaired t test: t(9) = 1.11, p = 0.3). Error bars represent the SEM, dots indicate individual recordings, and numbers represent the number of rats represented per bar.
Cannabinoid receptor expression is unchanged following persistent inflammation
Previously published work from our laboratory established that persistent inflammation downregulates CB1R protein via Western blot in the rostral ventromedial medulla (RVM; Li et al., 2017), the downstream target of the vlPAG in the descending pain modulatory pathway. Therefore, we hypothesized that our observed reduction in CB1R suppression of GABA release after persistent inflammation was because of receptor downregulation. Expression levels of cannabinoid receptors were measured using radioligand binding with [3H]CP-55 940. Since this is a different ligand than was used in the experiments in Figure 1, we first replicated our original observation. Similar to WIN, CP-55940 suppression of GABA release is significantly reduced after persistent inflammation and is reversed with the CB1R-selective antagonist AM251 (Fig. 3A,B). Next, radioligand binding was conducted in vlPAG dissected from naive and CFA-treated rats using [3H]CP-55940. Contrary to our hypothesis, there was no difference in total cannabinoid receptor binding between tissue from naive and CFA-treated animals (Fig. 3C). Cannabinoid receptor binding was also assessed in the DLS, hypothalamus, and RVM. Unfortunately, the radioligand binding data from the RVM were uninterpretable because of high nonspecific binding, likely because of the high levels of myelination in the adult RVM. Similar to the vlPAG, persistent inflammation did not impact cannabinoid receptor binding in either the DLS or hypothalamus (DLS: naive Bmax, 1550 ± 257 fmol/mg; CFA Bmax, 1626 ± 230 fmol/mg; unpaired t test: t(9) = 0.22, p = 0.8; naive Kd, 0.9 ± 0.2 nm; CFA Kd, 1.0 ± 0.2 nm; unpaired t test: t(9) = 0.7, p = 0.5; and hypothalamus: naive Bmax, 617 ± 42 fmol/mg; CFA Bmax, 616 ± 67 fmol/mg; unpaired t test: t(4) = 0.013, p = 0.99; naive Kd, 2.2 ± 0.3; CFA Kd, 1.8 ± 0.4 nm; unpaired t test, t(4) = 0.8, p = 0.5).
Cannabinoid receptor binding is unaffected by persistent inflammation. A, Representative traces of eIPSCs recorded from vlPAG neurons in baseline (NBQX 5 μm; teal), cannabinoid agonist CP55,940 (CP55; 3 μm; black), and CB1-selective antagonist AM251 (3 μm; blue) from naive and CFA-treated rats. B, Percentage inhibition of eIPSC amplitude by CP55,940 in vlPAG slices from naive or CFA-treated rats (unpaired t test; t(9) = 7.8, p < 0.0001). Error bars represent the SEM, dots indicate individual recordings, and numbers represent the number of rats per bar. C, Representative radioligand binding saturation curve with [3H]CP55,940 and vlPAG tissue from naive (black) and CFA-treated (red) rats (vlPAG from 8 rats pooled per curve; statistics run on an average of 3 curves). Naive Bmax, 708 ± 126 fmol/mg; CFA Bmax, 785 ± 61 fmol/mg; unpaired t test: t(4) = 0.55, p = 0.6. Naive Kd = 1.7 ± 0.5 nmol; CFA Kd = 1.8 ± 0.4 nmol; unpaired t test: t(4) = 0.27, p = 0.8.
Presynaptic CB1Rs do not display acute desensitization to exogenous agonist
The observation that total cannabinoid receptor binding was unchanged in slices from CFA-treated rats led us to hypothesize that CB1Rs are desensitized after persistent inflammation. To first assess acute desensitization, mIPSCs were measured while WIN (3 μm) washed over slices from naive animals for 30 min. Similar to other presynaptic GPCRs, CB1Rs do not desensitize during 30 min of activation by WIN superfusion (Fig. 4A). To test whether CB1Rs in the vlPAG desensitize after longer WIN-induced activation, slices containing vlPAG were incubated in WIN (3 μm) for 90 min up to 5.5 h, after which RIM was used to determine the extent of inhibition by WIN. RIM increased eIPSC amplitudes similarly after 15 min of WIN (275 ± 48%) or >1 h of WIN (274 ± 92%; Fig. 4B,C). These results indicate that CB1Rs are resistant to desensitization, even after several hours of WIN exposure.
CB1R function is sustained throughout 5 h WIN-induced activation. A, Inhibition of mIPSC frequency in recordings from vlPAG neurons during 30 min of WIN exposure (3 μm; n = 6 cells from 6 rats). Data are normalized to mIPSC frequency during baseline in TTX (500 nm) and NBQX (5 μm). WIN (3 μm) reduced mIPSC frequency over the first 10 min of drug application. Frequency remained reduced for the entirety of the 30 min drug application and was reversed by RIM (3 μm). B, eIPSC amplitude with bath application of CB1R-selective antagonist RIM (3 μm) after 15 min in WIN (3 μm; paired t test; t(7) = 2.42, *p = 0.046; data from 6 rats) or >1 h of WIN incubation (3 μm; paired t test; t(5) = 3.53, *p = 0.02; 5 cells from 3 rats). Average is shown in a thick black line. C, Bar graph depicting RIM percentage increase from WIN after 15 min in WIN (white bar) or >1 h in WIN (gray bar; unpaired t test; t(11) =0.2, p = 0.8).
Persistent inflammation induces GRK2/3-dependent CB1R desensitization
Although CB1Rs are resistant to desensitization with application of an exogenous agonist over multiple hours (Fig. 4), it is possible that CB1R desensitization is induced by endogenous agonists over the course of 5–7 d of CFA-induced inflammation. A key step in canonical postsynaptic GPCR desensitization is GRK phosphorylation of the GPCR C-terminal tail (Lefkowitz, 1993; Kovoor et al., 1998; Zhang et al., 1998; Wang, 2000). To determine whether GRK-mediated desensitization is occurring at the CB1R with inflammation, we blocked this step by incubating slices in Cmp101 (1 μm, ≥1 h), a potent, membrane-permeable inhibitor of GRK 2/3 (Ikeda et al., 2007; Thal et al., 2011). Incubating slices from CFA-treated rats in Cmp101 recovered CB1R suppression of GABA release after persistent inflammation (Fig. 5A). This result indicates that persistent inflammation induces GRK2/3-dependent desensitization of the presynaptic CB1R. We also tested a higher concentration of Cmp101 (Lowe et al., 2015; Leff et al., 2020; Adhikary et al., 2022) and recovered more CB1R-mediated inhibition. GRK2/3 desensitization after persistent inflammation appears to be selective to the CB1R as incubation in Cmp101 (30 μm) did not alter presynaptic MOR suppression of GABA release between slices from naive and CFA-treated rats (Fig. 5B).
Cmp101 incubation recovers CB1R inhibition of GABA release after persistent inflammation. A, WIN (3 μm) inhibition of eIPSC amplitudes from naive or CFA-treated rats. vlPAG slices were incubated in vehicle (no fill) or Cmp101 (filled bar) for >1 h. Cmp101 incubation fully recovered CB1R signaling in slices from CFA-treated rats (two-way ANOVA; main effect of Cmp101: F(1,14) = 5.9, p = 0.029; main effect of CFA: F(1,14) = 26.3, p = 0.0002; CFA × Cmp101 interaction: F(1,14) = 14.63, p = 0.002; p values on graphs are Šídák post hoc test). B, DAMGO (1 μm) inhibition of eIPSC amplitude after Cmp101 (30 μm) incubation from naive or CFA-treated rats (unpaired t test; t(8) = 0.92, p = 0.4). C, Cmp101 incubation reveals eCB tone in recordings from CFA-treated rats (two-way ANOVA; main effect of Cmp101: F(1,46) = 6.1, p = 0.02; values on graphs are from Šídák post hoc test). Experiments using either RIM or NESS are combined. Error bars represent the SEM, dots indicate individual neurons, and numbers represent the number of animals per bar.
GRK2/3-dependent desensitization after persistent inflammation suggests increased CB1R activation by agonists. The next experiments used the CB1R antagonist RIM (3 μm) to determine whether eCBs tonically activate CB1Rs in the vlPAG. In line with previous findings in the vlPAG (Aubrey et al., 2017), RIM did not consistently increase eIPSC amplitude in recordings from naive rats (paired t test; baseline vs RIM: t(14) = 1.46, p = 0.17), nor was there consistent eCB tone in CFA-treated rats (paired t test; baseline vs RIM: t11 = 1.13, p = 0.28). Since inflammation induces CB1R desensitization, we realized that it was possible that CB1R desensitization obscured our ability to detect eCB tone in slices from CFA-treated rats. This was tested by incubating slices in Cmp101 to block GRK2/3-dependent desensitization. Cmp101 incubation did not change overall eCB tone in slices from naive animals but revealed eCB tone in slices from CFA-treated animals (Fig. 5C).
RIM is an inverse agonist, so effects of RIM may be because of increased constitutive activity of CB1Rs in CFA-treated rats (Ruiu et al., 2003). To determine whether the increased effect of RIM was a result of increased CB1R constitutive activity or eCB tone, we again tested for eCB tone but with the CB1R neutral antagonist NESS-0327 (0.5 μm). After Cmp101 incubation, NESS increased eIPSC amplitudes (40 ± 8%; n = 9) to a similar extent of the increase produced by RIM (31 ± 5%; n = 5; unpaired t test NESS vs RIM: t(12) = 0.82, p = 0.43). Thus, constitutive CB1R activity does not explain the increase in eIPSC amplitude in the presence of RIM, but instead eCBs tonically activate and desensitize CB1Rs in the presence of inflammation.
Persistent inflammation prolongs 2-AG signaling in the vlPAG
To directly examine eCB actions at CB1Rs, we used DSI, a depolarization protocol that induces the synthesis and release of eCBs in other brain regions (Kreitzer and Regehr, 2001; Ohno-Shosaku et al., 2001; Wilson and Nicoll, 2001). The DSI protocol (+20 mV for 5 s; Wamsteeker et al., 2010) induced a rapid and transient suppression of presynaptic GABA release in a subset of PAG neurons that was blocked by the CB1R antagonist NESS (0.5 μm; Fig. 6A). The DSI protocol elicited inhibition of eIPSCs in fewer than half of the recorded neurons from slices from naive rats (Fig. 6B). After 5–7 d of hindpaw inflammation, the same DSI protocol induced a prolonged suppression of GABA release that was also blocked in the presence of NESS (Fig. 6C). Further, persistent inflammation significantly increased the proportion of cells that exhibit DSI (Fig. 6D). The time course of suppression was analyzed by measuring the maximal inhibition immediately following depolarization (max DSI) and inhibition 30 s after depolarization (late DSI; Patel et al., 2009). Max DSI in recordings from naive rats (male: 29 ± 4%, n = 10; female: 34 ± 7%, n = 9) is similar to inhibition observed in recordings from CFA-treated rats (male: 37 ± 4%, n = 13; female: 35 ± 4%, n = 9). After 30 s, eIPSC amplitudes returns close to baseline in naive rats but stay inhibited in recordings from CFA-treated rats (Fig. 6A,C,E). These data were analyzed for a sex difference with a three-way ANOVA (treatment × sex × time) with no main effect of sex (F(1,37) = 0.24, p = 0.62), so data from recordings from males and females are combined for all further DSI analyses.
Persistent inflammation increases 2-AG activation of CB1Rs. A, Summary of DSI (+20 mV for 5 s) in tissue from naive rats (black circles; n = 21 recordings from 15 rats). DSI is blocked by NESS (0.5 μm; gray open boxes; n = 4 recordings from 3 rats). B, Proportion of patched neurons in slices from naive rats that responded to DSI (21 neurons exhibited DSI, and 29 did not). C, Summary of DSI in tissue from CFA-treated rats (red dots; n = 22 recordings from 17 rats). DSI is blocked by NESS (0.5 μm; gray open boxes; n = 9 recordings from 5 rats). D, After persistent inflammation, 22 cells exhibit DSI, while 11 do not. The proportion of neurons that produced DSI was significantly higher in slices from CFA-treated slices than naive slices (χ2, p = 0.03). E, Quantification of inhibition of eIPSC amplitudes at max DSI and late DSI in vlPAG tissue from naive and CFA-treated rats (two-way repeated-measures ANOVA; main effect of DSI: F(1,41) = 4.42, p = 0.04; main effect of treatment: F(1,41) = 10.75, p = 0.002; DSI × treatment interaction: F(1,41) = 4.22, p = 0.046; values on graphs from Šídák post hoc test). F, DO34 (1 μm, >1 h), an inhibitor of the 2-AG synthesis enzyme DAGLα, significantly reduces maximum DSI (two-way ANOVA; main effect of DO34, F(1,77) = 92.19, p < 0.0001; Šídák post hoc test). G, Maximum and late DSI in slices from naive and CFA-treated animals after incubation in JZL184 (1 μm, >1 h) and JZL184 + Cmp101 (1 μm, >1 h; two-way repeated-measures ANOVA; main effect of treatment, F(2,53) from 6.04, p = 0.004; p values on graphs from Šídák post hoc test). Dots represent individual recordings, numbers below the bar represent number of animals; error bars represent the SEM.
The eCB 2-AG is responsible for DSI in multiple brain regions, and DSI can be prolonged by inhibiting 2-AG degradation (Straiker and Mackie, 2005; Hashimotodani et al., 2007). To determine whether 2-AG is the eCB responsible for DSI in the vlPAG and the prolonged DSI time course after persistent inflammation, we inhibited 2-AG synthesis with the DAGLα inhibitor DO34 (1 μm, >1 h). Blocking 2-AG synthesis completely blocked DSI in slices from both naive and CFA-treated rats (Fig. 6F), supporting the interpretation that 2-AG is the primary signaling molecule responsible for DSI after inflammation. To determine the role of 2-AG degradation on DSI in the vlPAG, slices from naive or inflamed rats were incubated in the MAGL inhibitor JZL184 (1 μm) for at least 1 h (Long et al., 2009; Lau et al., 2014). Blocking 2-AG degradation prolonged DSI in slices from naive rats (Fig. 6G). Interestingly, inhibiting MAGL in slices from CFA-treated rats abolished DSI, but eCB signaling was recovered in the presence of Cmp101 (Fig. 6G). Together, these data indicate that CFA-induced inflammation increases 2-AG synthesis in the vlPAG and blocking MAGL-mediated 2-AG degradation induces CB1R desensitization. Therefore, we propose that MAGL maintains a subdesensitizing concentration of 2-AG that is required to maintain prolonged CB1R signaling observed during persistent inflammation.
Discussion
Here, we describe a mechanism of persistent inflammation-induced adaptations in the endogenous cannabinoid system. Importantly, adaptations induced by inflammation differentially impact exogenous and endogenous cannabinoids actions at CB1Rs because of MAGL degradation of the eCB 2-AG. Inflammation promotes desensitization of presynaptic CB1Rs that suppress GABA release in the vlPAG. This desensitization is dependent on GRK2/3, as CB1R suppression of GABA release recovers in the presence of the GRK2/3 inhibitor Cmp101. Inflammation increases 2-AG release and prolongs DSI-induced inhibition of GABA release from vlPAG terminals. Inhibition of MAGL results in rapid CB1R desensitization, suggesting that MAGL maintains 2-AG levels at subdesensitizing concentrations.
Our data show direct evidence of GRK2/3-dependent desensitization of presynaptic CB1Rs. While postsynaptic GPCRs readily desensitize and internalize in response to agonist exposure (Williams et al., 2013), it is well established that presynaptic GPCRs are resistant to desensitization (Wetherington and Lambert, 2002; Fyfe et al., 2010; Pennock and Hentges, 2011; Pennock et al., 2012; Pennock and Hentges, 2016). Sustained signaling from presynaptic receptors during prolonged agonist exposure may be because of multiple mechanisms. One such mechanism involves protein–protein interactions with presynaptic scaffold proteins immobilizing receptors close to the plasma membrane, as observed for presynaptic GABAB receptors (Boudin et al., 2000; Vigot et al., 2006; Vargas et al., 2008; Laviv et al., 2011). An alternative mechanism, observed with presynaptic MORs, involves lateral movement of perisynaptic receptors into the synapse following agonist-induced internalization, allowing for continued signaling through rapid receptor replacement (Jullié et al., 2020, 2022). Both mechanisms result in sustained GPCR signaling from presynaptic terminals. Like MORs, CB1Rs also exhibit rapid mobility through the synapse under basal conditions; however, in contrast to MOR regulation, prolonged agonist exposure significantly reduces CB1R mobility and localization at the plasma membrane (Mikasova et al., 2008). In our studies, we find that CB1Rs and MORs are regulated differently in response to inflammation; CB1Rs desensitize, while MORs do not. Interestingly, Jullié et al. (2020) find that MOR endocytosis is mediated by GRK2/3 phosphorylation and blocked with Cmp101, but functional desensitization is not observed because of rapid receptor replacement. Consistent with findings from the study by Jullié et al. (2020), our functional studies show that Cmp101 has no impact on MOR-induced suppression of GABA release in slices from naive or CFA-treated rats. The results presented in our experiments clearly illustrate different cellular mechanisms regulating these two presynaptic Gαi/o-coupled GPCRs within the vlPAG.
Reduced CB1R activity in response to prolonged administration of exogenous agonists, such as THC or WIN, has previously been reported (Sim et al., 1996; Breivogel et al., 1999; Rubino et al., 2000; Kouznetsova et al., 2002; Lazenka et al., 2014). Long-term increases in eCBs also lead to reduced effects of CB1R activation (Long et al., 2009; Schlosburg et al., 2010; Kinsey et al., 2013; Imperatore et al., 2015; Navia-Paldanius et al., 2015). Levels of eCBs, including AEA and 2-AG, in the PAG are rapidly increased after acute inflammation induced by formalin injection into the hindpaw (Walker et al., 1999), as well as after 7 d of chronic constriction injury, a model of neuropathic pain (Petrosino et al., 2007). The observed CB1R desensitization in our study is likely a result of increased CB1R-induced G-protein signaling within the vlPAG early in inflammation (Wilson-Poe et al., 2021).
The prolonged DSI time course observed after inflammation may be the result of increased eCB synthesis, decreased activity/expression of the degradation enzyme MAGL, or both mechanisms. As mentioned above, acute inflammation and neuropathic pain increase the release of both AEA and 2-AG, albeit 2-AG levels are higher in the brain (Stella et al., 1997). We show that prolonged inhibition is blocked by inhibiting DAGLα, the enzyme responsible for 2-AG synthesis, implicating 2-AG signaling as a major component in the adaptations induced by inflammation in the vlPAG. Our data indicate that more cells are responsive to a DSI protocol, suggesting that there may be more sites of eCB synthesis and release after inflammation. It is possible AEA plays a role in priming the CB1Rs for desensitization. It is clear that increasing 2-AG levels through MAGL inhibition is not sufficient to induce CB1R desensitization or JZL184 incubation of naive slices would show appreciable desensitization and that is not observed. Further studies will examine the cellular mechanisms by which persistent inflammation regulates CB1R signaling and desensitization.
Importantly, we find that MAGL activity is required to protect CB1Rs from desensitization in slices from inflamed rats. Under normal conditions in the vlPAG, MAGL catabolizes 2-AG quickly enough that washing 2-AG over the slice does not suppress GABA release unless MAGL is blocked (Lau et al., 2014); however, prolonged DSI can be achieved in slices from naive rats by pharmacologically or genetically inhibiting MAGL (Straiker and Mackie, 2005; Pan et al., 2009; Schlosburg et al., 2010; Chen et al., 2016). Further highlighting the importance of MAGL for CB1R function, experiments using MAGL knock-out mice or pharmacological inhibition of MAGL show increases in 2-AG signaling that lead to reduced CB1R function in other brain areas (Long et al., 2009; Schlosburg et al., 2010; Kinsey et al., 2013; Imperatore et al., 2015; Navia-Paldanius et al., 2015). These results highlight the importance of eCB degradation for balance between CB1R signaling and desensitization.
THC and synthetic cannabinoids are effective at reducing edema and pain associated with inflammatory agents (Sofia et al., 1973; Smith et al., 1998; Li et al., 1999), including via microinjections directly into the vlPAG (Martin et al., 1995; Lichtman et al., 1996; Wilson et al., 2008). Systemic administration of cannabinoids effectively reduce CFA-induced hyperalgesia at acute time points (Mitchell et al., 2005; Ahn et al., 2009, 2011; Anderson et al., 2014), although fewer studies have tested cannabinoids at persistent inflammation time points (Craft et al., 2013; Britch et al., 2020). Inhibitors of fatty acid amide hydrolase and MAGL that increase levels of eCBs are also effective antihyperalgesic agents for inflammation-induced pain at early (<5 d) time points (Mitchell et al., 2005; Ahn et al., 2009, 2011; Clapper et al., 2010; Naidu et al., 2010; Anderson et al., 2014). However, prolonged MAGL inhibition leads to tolerance and dependence (Schlosburg et al., 2010), and the data presented here suggest that MAGL inhibition may not be a viable strategy to treat inflammatory conditions if MAGL activity is required to protect CB1Rs from desensitization. Interestingly, several studies find that eCB antihyperalgesia is not fully reversed with CB1 and CB2 receptor antagonists (Smith et al., 1998; Ahn et al., 2009, 2011), suggesting that CB1Rs are not engaged after prolonged inflammation, possibly because of desensitization. The more interesting question is whether prolonged increases in eCBs are compensatory adaptations that reverse pain, especially in light of CB1R desensitization. One intriguing possibility is that inflammation-induced increases in 2-AG in the vlPAG contribute to hyperalgesia and that CB1R desensitization is a compensatory response that protects synaptic function. Indeed, there is precedent for cannabinoids to contribute to hyperalgesia (Dunford et al., 2021; Khasabova et al., 2022). In this case, MAGL inhibition would be useful therapeutically. Understanding the behavioral consequences of inflammation-induced changes in eCB signaling within the vlPAG, the generalizability to other brain areas and the reversibility of this process have important implications for future drug development targeting MAGL and CB1Rs.
Footnotes
This work was supported by funding from National Institutes of Health (NIH) Grants R01-DA-042565 and R01-NS-120486 (to S.L.I.). C.A.B. was supported by NIH/National Institute on Drug Abuse (NIDA) Grants F31-DA052114 and T32-DA-007262). A.J. was supported by NIH/NIDA Grant ADA20003-001-00002, US Drug Enforcement Administration Grant D-20-OD-00), US Food and Drug Administration Grant CDER-20-I-0546, and Department of Veterans Affairs Research Career Scientist Program Grant 1IK6BX005754. We thank Dr. Amy Eshleman for expertise and guidance with radioligand binding assays. We also thank Dr. John Williams and Dr. Sweta Adhikary for help with the Compound 101 experiments. In addition, we thank members of the Ingram laboratory, as well as Dr. Mary Heinricher and laboratory, for valuable discussion and suggestions.
The authors declare no competing financial interests.
- Correspondence should be addressed to Susan L. Ingram at susan.ingram{at}cuanschutz.edu
