The cannabinoid dehydroxylcannabidiol suppresses neuropathic pain by upregulating a spinal glycine receptor-mediated compensation mechanism

Abstract

Neuropathic pain (NP) is well-known to be complicated and challenging to manage. Organisms have been shown to initiate various intrinsic compensatory strategies to alleviate NP, which could be utilized to aid the development of therapeutic treatments. However, these mechanisms that alleviate NP have not been fully elucidated. Here we show that spinal glycine receptor α1 (GlyRα1) is a key element of an intrinsic compensatory mechanism to alleviate NP. This spinal GlyRα1-mediated compensatory mechanism is manifested in many ways, including enhanced inhibitory glycinergic transmission and increased GlyRα1 protein expression levels. We show such GlyRα1 upregulation is achieved via an intracellular calmodulin kinase II/IV (CaMKII/IV)-cAMP response element binding protein signaling pathway. We demonstrate that dehydroxylcannabidiol, a synthetic nonpsychoactive cannabinoid that specifically targets GlyRs, significantly potentiates spinal GlyR function, inhibits spared nerve injury–induced spinal neuronal hyperactivity, and ultimately abrogates the development of hyperalgesia. These results suggest that the compensatory upregulated spinal GlyRα1 is efficacious in alleviating NP. Thus, our study reveals a potential strategy for the clinical treatment of NP by utilizing this central compensatory mechanism.

Neuropathic pain (NP) is a complicated pathological maladaptive response arising from peripheral or central nervous system injuries (1, 2), commonly characterized by a range of pain symptoms and complications such as allodynia and hyperalgesia (3, 4). Nerve injury-induced NP affects almost 7% to 10% of the general population, leading to a serious debilitation of life quality (1). Peripheral nerve injuries trigger a series of mutually related events in the spinal cord and dorsal root ganglion, including hypersensitivities, facilitation, and altered expression profiling of inflammatory factors, such as cytokines and prostaglandins, and neurotransmitters such as glutamate and γ-aminobutyric acid (2, 4, 5). Although great efforts have been devoted to revealing the pathological mechanism of nerve injury-induced NP, the effective treatment options are still limited (6). To date, a considerable number of patients fail to achieve sufficient pain relief after receiving various pharmacological management (7). This increases the urgency to develop new and satisfactory therapeutic strategies for NP.

Although nerve injury-induced NP is intractable and sometimes refractory to many traditional analgesics (7, 8), several endogenous compensatory mechanisms to suppress NP have already been elucidated in considerable detail. The endogenous compensatory mechanisms are achieved mainly via modulating the abundance and function of cytokines, neurotrophins, neurotransmitters, and synaptic receptors (9, 10, 11, 12, 13). Although these endogenous compensatory processes are often insufficient to completely abolish NP, the pivotal components involved in these processes are critical analgesic targets such as opioid receptors and ion channels (13, 14, 15).

Glycine receptors (GlyRs) have been considered a major inhibitory ligand-gated ion channel in the spinal cord (16, 17). Generally, GlyRs are functional as pentameric assemblies containing homomeric α1-3 subunits or heteromeric α/β subunits which are widely distributed in the spinal cord (18, 19, 20, 21). GlyR α1 and α3 subunits have been largely reported to be involved in pain processing (22, 23, 24, 25, 26). Evidence has shown that functional impairment of spinal GlyR α3 subunit seems to be the cause of chronic pain (23, 25). However, the accurate characteristic of GlyR α1 subunit in chronic pain, especially NP, remains unclear. In this study, we sought to answer two key questions: Whether the function and level of GlyR α1 change in NP? Are these changes a cause or a compensation and can they be utilized to treat NP?

Results

SNI causes NP and spinal neuronal hyperactivity

We first established a previously described SNI model of NP by ligating and transecting the common peroneal and sural nerves of C57BL/6J mice (27). A series of von Frey hair stimulations (0–2.0 g) were applied to the lateral plantar surface of the injured hind paw to test the paw withdrawal thresholds. The von Frey test was performed 1 day prior to the operation and repeated for 14 days following surgery, and the spinal cord slice patch-clamp recording test was performed on Day 3 after SNI surgery (Fig. 1A). The results showed that there was no statistical difference in the paw withdrawal threshold (PWT) between the sham mice and the SNI mice 1 day before surgery. The withdrawal threshold of SNI mice showed a gradual decrease and reached its bottom on the third day after surgery, and such SNI-induced mechanical allodynia lasted for at least 14 days (Fig. 1B).

Figure 1.

Construction and validation of the SNI model in mice.A, schematic diagram showing the processing of SNI, von Frey test, and slice recording. B, time course of withdrawal threshold in the hind paw of sham and SNI mice based on von Frey test. (sham, n = 6 mice; SNI, n = 6 mice; Interaction: F_14,60 = 4.537, p < 0.0001; Time: F_14,60 = 6.673, p < 0.0001; Treatment: F_1,60 = 657.8, p < 0.0001). C–F, sample traces, firing rates, and average values of threshold and RMP of AP recorded in L3-L5 SCDH neurons of sham and SNI mice (sham, n = 16 cells from five mice; SNI, n = 16 cells from five mice; D, interaction: F_20,151 = 3.916, p < 0.0001; Injected current: F_20,151 = 28.91, p < 0.0001; Treatment: F_1,151 = 409.5, p < 0.0001; E, threshold: t₃₀ = 7.652, p < 0.0001; F, RMP: t₃₀ = 8.005, p < 0.0001). G–I, the relationship between the withdrawal threshold on the third day and firing rate, threshold, and RMP of AP recorded in Lumbar 3-Lumbar 5 (L3-L5) SCDH neurons of sham and SNI mice (n = 10 mice). The data for linear correlation analysis were obtained from all cells of the sham and SNI mice. For x-axis, each dot represents the average value of the firing rate (at 120 pA) (G), threshold (H), and RMP (I) of multiple cells from individual mice. For y-axis, each dot represents the average value of the withdrawal threshold (G–I) from individual mice. Significance was assessed by two-way ANOVA followed by Dunnett’s post hoc test between groups in B and D and a two-tailed unpaired Student’s t test in E and F. The data are expressed as the mean ± SD. ∗∗∗∗p < 0.0001. RMP, resting membrane potential; SNI, spared nerve injury.

The spinal cord dorsal horn (SCDH) has been recognized as a pivotal region for processing multiple types of pain information. Large amounts of evidence have shown that the neuronal excitability in SCDH is significantly increased in NP (28, 29). We next examined the neuronal activity of SCDH by performing slice recording on the third day after the SNI operation. Consistently, the results showed that SNI indeed significantly increased the neuronal excitability in SCDH, reflected by the increased firing rate, decreased threshold, and reduced resting membrane potential (RMP) of electrical stimulation evoked action potentials (APs) (Fig. 1, C–F). Furthermore, there was a strong correlation between the PWT and the neuronal activity of SCDH (Fig. 1, G–I). Together, these results suggest the effectiveness of the SNI model.

Spinal GlyRα1-mediated inhibitory neurotransmission is upregulated in NP

Glycine has been reported to be the major inhibitory neurotransmitter in the spinal cord (30, 31, 32). We therefore examined the characteristics of glycinergic transmission in SCDH of SNI and sham mice. Both the frequency and the amplitude of the glycinergic miniature inhibitory postsynaptic currents (mIPSCs) in SNI mice were significantly higher than those in sham mice (Fig. 2A). Such glycinergic mIPSC was completely blocked by strychnine, the specific GlyR antagonist, indicating a GlyR-mediated mechanism (Fig. 2A). The SNI-induced increase in frequency and amplitude of glycinergic mIPSCs was also evidenced by the leftward-shifted cumulative fraction of inter-event intervals and the rightward-shifted cumulative fraction of amplitudes (Fig. 2, B and C), suggesting an enhanced glycinergic transmission in NP.

Figure 2.

Alteration of spinal glycine neurotransmission and GlyRα1 expression in NP.A, trace records, average frequency and amplitude of glycinergic mIPSCs recorded in L3-L5 SCDH neurons of sham and SNI mice (n = 12 cells each group; frequency: t₃₀ = 7.652, p < 0.0001; amplitude: t₂₄ = 6.487, p < 0.0001). B and C, cumulative probability plot for the inter-event interval and amplitudes for glycinergic mIPSCs recorded in L3-L5 SCDH neurons of sham and SNI mice (n = 12 cells each group). D, trace records of I_Gly recorded in L3-L5 SCDH neurons of the sham and SNI mice. The currents of glycine at 10 μM, 100 μM and 1000 μM were selected as representatives. E and F, dose-response curves of I_Gly in L3-L5 SCDH neurons of sham and SNI mice. The data were normalized to maximum currents (Imax) of the sham group or its own group (sham, n = 8 from three mice; SNI, n = 8 from three mice). G, EC₅₀ values of I_Gly in L3-L5 SCDH neurons of sham and SNI mice (sham, n = 8 from three mice; SNI, n = 8 from three mice; t₁₄ = 1.031, p = 0.3198). H and I, representative Western blot images and quantitative analysis of GlyRα1 protein in sham and SNI mice. GlyRα1 intensity was normalized to the GAPDH and then all the data was normalized to the value of 0 days in the sham group. The L3-L5 SCDH tissue was extracted on Day 0, Day 1, Day 3, Day 7, and Day14, respectively (I, n = 6 mice each group; sham, F_4,6 = 1.164, p = 0.4127; Day 1, p = 0.8925; Day 3, p = 0.7163; Day 7, p = 0.8356; Day 14, p = 0.8588 all relative to Day 0; SNI, F_4,25 = 10.2, p < 0.0001; Day 1, p = 0.0006; Day 3, p < 0.0001; Day 7, p = 0.0022; Day 14, p = 0.0014 all relative to Day 0). J and K, effects of strychnine (i.p., 0.2 mg kg⁻¹) on the withdrawal threshold of SNI and sham mice based on von Frey test (veh, n = 6 mice; stry, n = 6 mice; J, interaction: F_6,70 = 1.027, p = 0.4150; Time: F_6,70 = 1.644, p = 0.1479; Treatment: F_1,70 = 6.199, p = 0.0152; K, interaction: F_6,70 = 0.563, p = 0.7583; Time: F_6,70 = 1.275, p = 0.2802; Treatment: F_1,70 = 1.006, p = 0.3193). Significance was assessed by a two-tailed unpaired Student’s t test in A and G and one-way ANOVA with Tukey’s post hoc test in I and two-way ANOVA followed by the Dunnett’s post hoc test between groups in J and K. The data are expressed as the mean ± SD. ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. ns, no significance. Stry, strychnine; Veh, vehicle.

We next examined the electrophysiological function of GlyRs in SCDH neurons. The glycine-activated currents (I_Gly) recorded in the SCDH neurons of the SNI mice were significantly higher than those in the sham mice (Fig. 2, D and E). However, there was no difference in the dose–response curve (Fig. 2F) and half-maximal effective concentration (EC₅₀) values of I_Gly (Fig. 2G) between the sham mice and the SNI mice, indicating that the glycine binding affinity was unaffected. These results suggest that GlyR functioning is enhanced in NP, and such enhancement has no relationship with the binding affinity of glycine.

The enhancement of glycine neurotransmission motivated us to examine whether there is a compensatory increase in GlyRα1 protein in NP. The Western blot analysis showed that the protein abundance of GlyRα1 exhibited a gradual increase and reached its peak on the third day after surgery in SNI mice when compared with the sham mice (Fig. 2, H and I). Pharmacological blockade of GlyRα1 via intraperitoneal (i.p.) injection of strychnine significantly aggravated the SNI-induced mechanical allodynia, reflected by the further decreased PWT in the SNI mice (Fig. 2J) but did not affect the PWT in the sham mice (Fig. 2K). These results suggest that the upregulation of spinal GlyRα1 protein is a compensatory mechanism associated with NP.

The CaMKII/IV-CREB signaling pathway contributes to spinal GlyRα1 upregulation in NP

Plenty of studies have corroborated that the activated calcium/calmodulin-dependent protein kinase II/IV (CaMKII/IV)-cAMP response element-binding (CREB) (CaMKII/IV-CREB) cascade is a common response in various types of NP, and such activation can mediate the transcriptional regulation of multiple genes that encode peptides and proteins (33, 34, 35, 36, 37, 38). Considering this, we wonder whether the CaMKII/IV-CREB pathway also contributes to the upregulation of GlyRα1 protein in NP. The phosphorylation state of CREB (p-CREB) is a hallmark of CREB protein activation (33). Indeed, the Western results showed that the abundance of p-CREB protein was significantly higher in SNI mice than that in sham mice, while the total expression level of CREB protein was unaltered (Fig. 3, A–D). Intrathecal (i.t.) infusion of 666-15, a specific inhibitor for CREB phosphorylation, significantly blocked the SNI-induced increase in the protein abundance of p-CREB and GlyRα1 (Fig. 3, B–E).

Figure 3.

Involvement of CaMKII-CREB pathway in SNI-induced GlyRα1 up-regulation.A, scheme diagram showing the processing of SNI and Western blot. B–E, representative Western blot images and quantitative analysis of CREB, p-CREB, and GlyRα1 protein with or without an intrathecal infusion of 666-15 in sham and SNI mice. The intensity of CREB, p-CREB, and GlyRα1 was first normalized to the intensity of respective GAPDH and then the data was normalized to their respective value in the sham + veh group. The L3-L5 SCDH tissue was extracted on Day 3 after SNI or sham surgery (n = 6 mice each group; C, F_3,12 = 0.8952, p = 0.4718, sham + 666-15, p = 0.1124, SNI + veh, p = 0.9801, SNI + 666-15, p = 0.9968 all relative to sham + veh group, SNI + 666-15, p = 0.9337 relative to SNI + veh group; D, F_3,4 = 14.67, p = 0.0127, sham + 666-15, p = 0.0026, SNI + veh, p < 0.0001, SNI + 666-15, p = 0.0866 all relative to sham + veh group, SNI + 666-15, p < 0.0001 relative to SNI + veh group; E, F_3,12 = 18.56, p < 0.0001, sham + 666-15, p = 0.0326, SNI + veh, p = 0.0227, SNI + 666-15, p < 0.0001 all relative to sham + veh group, SNI + 666-15, p < 0.0001 relative to SNI + veh group). F–K, representative Western blot images and quantitative analysis of CaMKII, p-CaMKII, CREB, p-CREB and GlyRα1 protein with or without intrathecal infusion of KN93 in the sham and the SNI mice. The intensity of CaMKII, p-CaMKII, p-CREB, and GlyRα1 was first normalized to the intensity of respective GAPDH and then the data were normalized to their respective value in the sham + veh group (n = 6 mice each group; G, F_3,4 = 1.617, p = 0.3192, sham + KN93, p = 0.9301, SNI + veh, p = 0.7630, SNI + KN93, p = 0.8420 all relative to sham + veh group, SNI + KN93, p = 0.3009 relative to SNI + veh group; H, F_3,4 = 13.26, p = 0.0152, sham + KN93, p = 0.0009, SNI + veh, p < 0.0001, SNI + KN93, p < 0.0001 all relative to sham + veh group, SNI + KN93, p = 0.0015 relative to SNI + veh group; I, F_3,4 = 0.8278, p = 0.5346, sham + KN93, p = 0.9867, SNI + veh, p = 0.9152, SNI + KN93, p = 0.4022 all relative to sham + veh group, SNI + KN93, p = 0.7784 relative to SNI + veh group; J, F_3,4 = 47.96, p = 0.0014, sham + KN93, p = 0.0002, SNI + veh, p < 0.0001, SNI + KN93, p = 0.8166 all relative to sham + veh group, SNI + KN93, p < 0.0001 relative to SNI + veh group; K, F_3,5 = 1.437, p = 0.3366, sham + KN93, p = 0.0327, SNI + veh, p = 0.0003, SNI + KN93, p = 0.8949 all relative to sham + veh group, SNI + KN93, p < 0.0001 relative to SNI + veh group). Significance was assessed by a one-way ANOVA with Tukey’s post hoc test in C, D, E, G, H, I, J, and K. The data are expressed as the mean ± SD. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. ns, no significance.

The T286 phosphorylation state of CaMKII (p-CaMKII) or T196/200 phosphorylation state of CaMKIV (p-CaMKIV) is hallmark for CaMKII and CaMKIV activation, respectively (37, 38, 39). Western blotting results showed that the levels of both p-CaMKII and p-CaMKIV protein were also significantly increased in the SNI mice when compared with the sham mice (Figs. 3, F–H and S1, Q–S). I.t. infusion of KN93, an inhibitor for CaMKII/IV and their phosphorylation, significantly blocked the phosphorylation of CaMKII/IV and CREB and the upregulation of GlyRα1 protein in the spinal cord of SNI mice, without affecting the total expression level of CaMKII/IV and CREB protein (Figs. 3, F–K and S1, Q–S).

Cannabinoid promotes glycine neurotransmission

Considering its compensatory upregulation, GlyR could be a critical drug target for treating NP. Dehydroxylcannabidiol (DH-CBD), a synthetic nonpsychoactive cannabinoid, has been widely reported to specifically enhance the function of GlyRα1 and relieve pain (22, 23, 40, 41). We therefore evaluated the effects of DH-CBD on the GlyR function and the glycine neurotransmission in NP. DH-CBD at 10 μM significantly increased I_Gly in SCDH neurons of the sham mice (Fig. 4, A and B). Notably, such an increase was significantly higher in the SNI mice than that in sham mice (Fig. 4, A and B). DH-CBD exerts its potentiating effect mainly through interacting with the site of Ser296 (S296) on GlyRα1 (22, 23, 24, 40, 42, 43). Indeed, mutating the S296 to alanine (S296A) significantly blocked the DH-CBD-induced potentiation of I_Gly in SCDH neurons of both the sham and the SNI mice (Fig. 4, C and D).

Figure 4.

Effects of DH-CBD on the glycinergic transmission in L3-L5 SCDH of SNI mice.A and B, trace records and average values of I_Gly and percentage of potentiation of DH-CBD on I_Gly recorded in L3-L5 SCDH neurons of the sham- and SNI-GlyRα1^WT mice with or without DH-CBD pre-incubation. Glycine at 30 μM and DH-CBD at 10 μM were used (sham, n = 12–13 cells from three mice; SNI, n = 12–13 cells from three mice; B, left, F_3,44 = 69.75, p < 0.0001, sham + DH-CBD, p = 0.0262, SNI, p < 0.0001, SNI + DH-CBD, p < 0.0001 all relative to sham group, SNI + DH-CBD, p < 0.0001 relative to SNI group; B, right, t₂₂ = 26.87, p < 0.0001). C and D, trace records and average values of I_Gly recorded in SCDH neurons of the sham- and SNI-GlyRα1^S296A mice with or without DH-CBD pre-incubation (sham, n = 12 cells from three mice; SNI, n = 12 cells from three mice; D, F_3,44 = 17.48, p < 0.0001, sham + DH-CBD, p = 0.9963, SNI, p < 0.0001, SNI + DH-CBD, p < 0.0001 both relative to sham group, SNI + DH-CBD, p = 0.8285 relative to SNI group). E, trace records and average values of frequency and amplitude of glycinergic mIPSCs recorded in SCDH neurons of the sham- and SNI-GlyRα1^WT mice with or without DH-CBD pre-incubation. (n = 12 cells each group; frequency: F_2,33 = 197.6, p < 0.0001, SNI, p < 0.0001, SNI + DH-CBD, p < 0.0001 both relative to sham group, SNI + DH-CBD, p < 0.0001 relative to SNI group; amplitude: F_2,31 = 93.93, p < 0.0001, SNI, p < 0.0001, SNI + DH-CBD, p < 0.0001 both relative to sham group, SNI + DH-CBD, p = 0.0011 relative to SNI group). F and G, cumulative probability plot for the inter-event interval and amplitudes of glycinergic mIPSCs recorded in SCDH neurons of the sham- and SNI-GlyRα1^WT mice with or without DH-CBD pre-incubation. H, trace records and average values of frequency and amplitude of glycinergic mIPSCs recorded in SCDH neurons of the sham- and SNI-GlyRα1^S296A mice with or without DH-CBD pre-incubation. (n = 12 cells each group; frequency: F_2,33 = 101.4, p < 0.0001, SNI, p < 0.0001, SNI + DH-CBD, p < 0.0001 both relative to sham group, SNI + DH-CBD, p = 0.1961 relative to SNI group; amplitude: F_2,33 = 72.72, p < 0.0001, SNI, p < 0.0001, SNI + DH-CBD, p < 0.0001 both relative to sham group, SNI + DH-CBD, p = 0.1004 relative to SNI group). I and J, cumulative probability plot for the inter-event interval and amplitudes of glycinergic mIPSCs recorded in SCDH neurons of sham- and SNI-GlyRα1^S296A mice with or without DH-CBD pre-incubation. Significance was assessed by a two-tailed unpaired Student’s t test in B right and a one-way ANOVA with Tukey’s post hoc test in B left, E, and H. The data are expressed as the mean ± SD. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗∗p < 0.0001. ns, no significance.

We also examined the effects of DH-CBD on glycinergic transmission by recording glycinergic mIPSCs in SCDH slices of the sham and the SNI mice. As demonstrated earlier, both the frequency and the amplitude of mIPSC in the SNI mice were significantly increased when compared with the sham mice (Fig. 4, E–G). Such an increase was significantly enhanced by the bath application of DH-CBD (Fig. 4, E–G). The effects of DH-CBD on glycinergic mIPSCs were significantly diminished in the GlyRα1^S296A mice (Fig. 4, H–J). These results suggest that DH-CBD could be used to enhance the GlyRα1-mediated compensation to treat NP.

DH-CBD suppresses the neuronal hyperactivity of SCDH and the SNI-induced hyperalgesia in NP

Glycinergic regulation of neuronal excitability in SCDH exists in various models of chronic pain (30, 31). Given that neuronal hyperactivity is a hallmark of NP, we next examined the effects of DH-CBD on the neuronal activity of SCDH in SNI mice. As indicated above, SNI significantly increased the SCDH neuronal activity in both the GlyRα1^WT mice and the GlyRα1^S296A mice. Furthermore, DH-CBD remarkably inhibited the SNI-induced neuronal hyperactivity in SCDH, reflected by the decreased firing rate and increased threshold and RMP of APs (Fig. 5, A–D). Such effects of DH-CBD were significantly suppressed in the GlyRα1^S296A mice (Fig. 5, E–H). These results indicate that DH-CBD can suppress neuronal hyperactivity in NP by acting on GlyRα1.

Figure 5.

Effects of DH-CBD on SNI-induced hyperexcitability of L3-L5 SCDH neurons and hyperalgesia.A–D, sample traces, firing rates, and average values of threshold and RMP of AP recorded in L3-L5 SCDH neurons of sham- and SNI-GlyRα1^WT mice with or without DH-CBD pre-incubation. (sham, n = 12 cells from six mice; SNI, n = 12 cells from six mice; SNI + DH-CBD, n = 12 cells from six mice; B, Interaction: F_40,315 = 5.944, p < 0.0001; Injected current: F_20,315 = 56.84, p < 0.0001; Treatment: F_2,315 = 341.9, p < 0.0001; SNI: p < 0.0001, SNI + DH-CBD: p < 0.0001 both relative to sham group; SNI + DH-CBD: p < 0.0001 relative to SNI group; C, F_2,33 = 107.6, p < 0.0001; SNI, p < 0.0001, SNI + DH-CBD, p < 0.0001 both relative to sham group, SNI + DH-CBD, p < 0.0001 relative to SNI group; D, F_2,33 = 83.6, p < 0.0001; SNI, p < 0.0001, SNI + DH-CBD, p = 0.4125 both relative to sham group, SNI + DH-CBD, p < 0.0001 relative to SNI group). DH-CBD at 10 μM were used. E–H, sample traces, firing rates, average values of threshold, and RMP of AP recorded in L3-L5 SCDH neurons of sham- and SNI-GlyRα1^S296A mice with or without DH-CBD pre-incubation. (sham, n = 12 cells from three mice; SNI, n = 12 cells from three mice; SNI+DH-CBD, n = 12 cells from three mice, F, interaction: F_40,357 = 8.653, p < 0.0001; Injected current: F_20,357 = 99.22, p < 0.0001; Treatment: F_2,357 = 611.3, p < 0.0001; SNI: p < 0.0001, SNI + DH-CBD: p < 0.0001 both relative to sham group; SNI + DH-CBD: p = 0.1692 relative to SNI group; G, F_2,33 = 132.5, p < 0.0001; SNI, p < 0.0001, SNI + DH-CBD, p < 0.0001 both relative to sham group, SNI + DH-CBD, p = 0.1796 relative to SNI group; H, F_2,33 = 234, p < 0.0001; SNI, p < 0.0001, SNI + DH-CBD, p < 0.0001 both relative to sham group, SNI + DH-CBD, p = 0.3531 relative to SNI group). DH-CBD at 10 μM were used. I and J, Time course of withdrawal threshold in the injured hind paw of sham and SNI mice with or without various DH-CBD administration based on von Frey test. DH-CBD at 3 mg kg⁻¹, 10 mg kg⁻¹, and 30 mg kg⁻¹ were i.p. injected (n = 6 mice each group, I, interaction: F_18,120 = 2.059, p = 0.0113; Time: F_6,120 = 9.806, p < 0.0001; Treatment: F_3,20 = 1.316, p = 0.2970; 3 mg kg⁻¹: p = 0.9918; 10 mg kg⁻¹: p = 0.6978; 30 mg kg⁻¹: p = 0.2018 all relative to veh group; J, interaction: F_18,120 = 8.813, p < 0.0001; Time: F_6,120 = 28.34, p < 0.0001; Treatment: F_3,20 = 15.67, p < 0.0001; 3 mg kg⁻¹: p = 0.9341; 10 mg kg⁻¹: p = 0.0053; 30 mg kg⁻¹: p < 0.0001 all relative to veh group). K and L, time course of withdrawal threshold in the injured hind paw of sham and SNI GlyRα1^WT and GlyRα1^S296A mice with or without various DH-CBD administration based on von Frey test (n = 6 mice each group, K, interaction: F_6,60 = 0.6846, p = 0.6627; Time: F_6,60 = 2.518, p = 0.0306; Treatment: F_1,10 = 0.1427, p = 0.7135; L, interaction: F_6,60 = 10.46, p < 0.0001; Time: F_6,60 = 12.03, p < 0.0001; Treatment: F_1,10 = 15.13, p = 0.003). M and N, time course of withdrawal threshold in the ipsilateral hind paw of sham and SNI mice with or without various DH-CBD administration based on von Frey test. DH-CBD at 0.1 μg, 1 μg, 5 μg and 10 μg were i.t. injected (n = 6 mice each group; M, interaction: F_24,150 = 1.699, p = 0.0299; Time: F_6,150 = 8.607, p < 0.0001; Treatment: F_4,25 = 0.2806, p = 0.8877; 0.1 μg: p = 0.9838; 1 μg: p = 0.9988; 5 μg: p = 0.9999; 10 μg: p = 0.9340 all relative to veh group; N, interaction: F_24,150 = 10.09, p < 0.0001; Time: F_6,150 = 58.06, p < 0.0001; Treatment: F_4,25 = 20.57, p < 0.0001; 0.1 μg: p = 0.9872; 1 μg: p = 0.0093; 5 μg: p < 0.0001; 10 μg: p < 0.0001 all relative to veh group). O and P, time course of withdrawal threshold in the ipsilateral hind paw of sham and SNI GlyRα1^WT and GlyRα1^S296A mice with or without various DH-CBD administration based on von Frey test. DH-CBD at 10 μg were i.t. injected (n = 6 mice each group; O, interaction: F_6,60 = 0.08591, p = 0.9975; Time: F_6,60 = 1.048, p = 0.4038; Treatment: F_1,10 = 2.647, p = 0.1348; P, interaction: F_6,60 = 20.21, p < 0.0001; Time: F_6,60 = 19.98, p < 0.0001; Treatment: F_1,10 = 32.13, p = 0.0002). Significance was assessed by a one-way ANOVA with Tukey’s post hoc test in C, D, G, and H and two-way ANOVA followed by the Dunnett’s post hoc test between groups in B, F, and I–P. The data are expressed as the mean ± SD. ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. ns, no significance. PWT, paw withdrawal threshold.

To test the effects of DH-CBD on the NP, both the sham and the SNI mice received i.p. administration of various doses of DH-CBD on Day 14 after surgery. Although DH-CBD had no effects on the PWT of sham mice (Fig. 5I), it significantly increased the PWT of SNI mice in a dose-dependent manner (Fig. 5J). The increase in the PWT developed and peaked at 45 min, and persisted for almost 90 min following DH-CBD administration. Such increase was substantially intercepted in the GlyRα1^S296A mice (Fig. 5, K and L).

To further examine whether the analgesic effect of DH-CBD is achieved through targeting spinal GlyRα1, DH-CBD was directly micro-infused into the L3-L5 section of the spinal cord. I.t. infusion of various doses of DH-CBD failed to affect the PWT of the sham mice (Fig. 5M). However, i.t. infusion of DH-CBD significantly increased the PWT of SNI mice in a dose-dependent manner (Fig. 5N). The increase in the PWT developed and peaked at 30 min and persisted for at least 90 min following DH-CBD administration. Such an effect was significantly blocked in the GlyRα1^S296A mice (Fig. 5, O and P). Taken together, these results suggest that DH-CBD can rescue the SNI-induced NP via enhancing spinal GlyRα1-mediated compensation.

Discussion

GlyRs, as the major inhibitory ionotropic receptors in the spinal cord, regulate varieties of physiological functions such as motor control, muscle contraction, and pain processing (17, 23, 25, 26). These effects are mainly accomplished through the post-synaptic located GlyRs which increase the intracellular chloride ion concentration and reduce cell excitability, and the presynaptic membrane-located GlyRs which facilitate the release of inhibitory transmitters. In the present study, we find that the spinal GlyRα1 is a pivotal element of the compensatory mechanism for NP, reflected by the increased frequency and amplitude of glycine mIPSCs and the elevated protein expression level of spinal GlyRα1. These results indicate that GlyRα1-mediated compensation probably occurs at both the pre-synaptic and post-synaptic levels. The upregulation of pre-synaptic GlyRα1 leads to increased glycine release, while the upregulation of post-synaptic GlyRα1 causes more chloride ions to enter and inhibits the activity of neurons in SCDH.

Three major GlyR α subunits (α1-3) have been identified in humans (19, 44). Except for α2 subunit, which is abundantly and widely expressed in embryonic period (44, 45), both α1 and α3 are mainly distributed in the adult spinal cord and involved in inflammatory pain (22, 23, 25, 40, 46). A previous study indicated that GlyRα3 is functionally inhibited during zymosan A- or complete Freund’s adjuvant–induced peripheral inflammation (25). In contrast with GlyRα3, GlyRα1 is significantly upregulated in inflammatory pain and NP revealed by previous studies and this study (40). Thus, different subunits of GlyRs seem to be assigned to distinct roles in the development of chronic pain. The GlyRα3 dysfunction is the cause of hyperalgesia in inflammation, while the GlyRα1 upregulation is used to compensatively diminish the neuronal hyperexcitability and produce analgesic effects in NP. Such differences may be related to the different ways of intracellular signal transduction regulation. GlyRα3 can be directly phosphorylated and inhibited by spinal prostaglandin E2 (PGE2), a key mediator of inflammation in chronic pain. In the present study, we find that GlyRα1 upregulation is mainly achieved through phosphorylating and activating the upstream CaMKII/IV-CREB signaling pathway.

In the present study, the superficial laminar II/III neurons in L3-L5 SCDH were recorded. Previous studies have reported that superficial laminar II/III neurons in L3-L5 SCDH are complex and consist of various populations with distinct firing properties, which are mainly described as a tonic, delayed, and gap neurons (47, 48, 49). Thus, it is hard for us to perform linear correlation analysis between such a complex network of neurons with distinct firing properties. The delayed neurons such as PKCγ neurons within the excitatory cohort in the lumbar spinal cord have been reported to be involved in neuropathic injury-induced mechanical allodynia (50, 51). Thus, only the delayed neurons are selected for linear correlation analysis in the present study.

Here, we show that activation of the CaMKII/IV-CREB pathway increases the protein level of GlyRα1 in SCDH. However, whether such an effect can directly affect pain sensitivity and glycinergic transmission in the SNI model remains uninvestigated in this study. It is noteworthy that the evidence of other studies may give some hints. For example, several previous reports have reported that activation of the CaMKII/IV-CREB pathway in the SCDH contributes to NP through increasing the level of brain-derived neurotrophic factor, neuropeptides, and NMDA receptor or decreasing GABAergic transmission in SCDH (33, 34, 52, 53, 54). In addition, i.t. injection of the KN93 significantly increases the sensitivity to a mechanical stimulus in diabetic neuropathy (55). Although evidence has shown that the CaMKII/IV-CREB pathway contributes to NP, further studies are still needed to evaluate its effects on glycinergic transmission and the analgesic effect of DH-CBD.

Our results showed that GlyRα1 might mediate a partial compensatory effect for NP. For instance, strychnine caused more severe hyperalgesia in the SNI mice when compared with the sham mice, hinting that the function of GlyRα1 is compensatorily enhanced in NP. However, such intrinsic compensatory action of GlyRα1 is far from curing NP, probably due to the following reasons. First, the upregulation of spinal GlyRα1 protein is limited and is not enough to completely counteract the enhanced neuronal activity and block the NP. Second, the endogenous agonist glycine is also too limited to maximize the function of the upregulated GlyRs. Third, the agonist binding affinity of GlyRs looks unaltered, weakening the effect of the GlyR upregulation. Therefore, enhancing GlyRα1 function with specific drugs could be a good way to effectively utilize the compensatory GlyRα1 upregulation.

Emerging evidence has suggested cannabinoids as a high-efficiency potentiator of GlyRs (23, 24, 26, 56, 57). It is worth noting that the potentiating effect of DH-CBD on spinal GlyR function is further increased in NP, which is likely due to the up-regulation of GlyRα1 protein. Massive evidence has suggested that DH-CBD is also a high-efficiency potentiator of GlyRα3 (22, 23, 25, 46). Based on this, it is likely that DH-CBD achieves its analgesic effect in SNI-induced NP partially through potentiating the function of GlyRα3. Besides, DH-CBD may interact with 5HT_1A and GABA_A receptors which are also important targets for NP (58, 59, 60, 61). However, whether the function and level of GlyRα3, 5HT_1A, and GABA_A receptors is changed in NP remains unknown. Thus, further studies are needed to clarify the characteristics of these receptors in the analgesic effect of DH-CBD. DH-CBD is a synthetic nonpsychoactive cannabinoid modified from tetrahydrocannabinol but does not bind to cannabinoid receptors. Thus, DH-CBD can produce significant analgesic effects without causing any psychoactive adverse effects (22, 23, 24, 41). Such GlyR-hypersensitive nonpsychoactive cannabinoids may represent an efficacious clinical therapeutic strategy for treating NP.

Experimental procedures

Animals

All procedures were approved by the Institutional Animal Use and Care Committee of the University of Science & Technology of China. C57BL/6J male mice (7–8 weeks old) used for mechanical allodynia test, Western blot, and electrophysiological recordings, were obtained from Vital River Laboratory Animal Technology Co, Ltd. GlyRα1^S296A mice and their wild-type littermates (7–8 weeks old) were also used. GlyRα1^S296A mice were constructed as previously described (40). Genotyping of the GlyRα1^S296A mice was done using the following primers: forward: 5′-GAATCTTCCAGGCAACATTTCAG-3′; reverse: 5′-AGTATCCCACCAAGCC AGTCTTT-3′. All mice were housed under a normal 12-h dark/light cycle with ad libitum access to water and food.

SNI neuropathic pain model

The animal model of chronic neuropathic pain was induced by unilateral SNI as previously described (62). Mice were anesthetized via intraperitoneal injection of sodium pentobarbital (50 mg kg⁻¹). Fully anesthetized mice were then placed on a heating pad to maintain body temperature at 37 °C during surgery. The surgical site is shaven by using an electrical razor to minimize contaminations. The surgery started by making an incision (1 cm) for the mouse through the skin and biceps femoris in the left hind limb by using a surgical scalpel. After the exposure of the sciatic nerve through the biceps femoris muscle, the three terminal branches of the sciatic nerve (tibial, common peroneal, and sural nerves) were carefully separated while minimizing any contact with or stretching of the sural nerve. The tibial and common peroneal nerves were then individually ligated with 6.0 silk and cut distally 2 to 3 mm of each nerve distal to the ligation. The muscle incision was closed with silk sutures and the skin with surgical staples. The muscular layer and the incision in the shaved layer of the skin are stitched by using sutures. Sham controls involve the sciatic nerve that was merely exposed but not ligated and dissected. The lesion results in evident hypersensitivity in the lateral area of the paw, which is innervated by the spared sural nerve.

Mechanical allodynia test

von Frey test was used to examine the mechanical allodynia of mice after SNI surgery. von Frey filaments (Ugo basile) were applied to the injured hind paw of the mice. Various von Frey filaments, including 0.008 g, 0.02 g, 0.04 g, 0.07 g, 0.16 g, 0.4 g, 0.6 g, 1.0 g, 1.4 g, and 2.0 g of bending force, were used to apply increasing amounts of bending force to the lateral plantar surface (the lateral plantar surface was innervated by the sural nerve). The mice were placed in a cage with an open wire mesh base and allowed to habituate for 30 min and then the filaments were pressed vertically against the plantar surface to cause a slight bending against the paw and were held for a maximum of 10 s with a 10-min interval between two stimulations. Brisk withdrawal or paw flinching was considered positive response. Each mouse was tested five times per stimulus strength. The filament causing three or more positive responses was regarded as the paw withdrawal mechanical threshold of the mice. For constructing the SNI model, ipsilateral plantar tests were performed for 14 consecutive days. For evaluating the effects of strychnine and DH-CBD, ipsilateral plantar tests were performed at various time points, including 0 min, 15 min, 30 min, 45 min, 60 min, 75 min, and 90 min, after drug administrations on Day 14.

Spinal cord slices preparation and recording

L3-L5 SCDH slices from C57BL/6J, GlyRα1^S296A mice and their wild type (WT) littermates (7–8 weeks old) were used in this experiment. Briefly, the spinal cord was quickly removed after laminectomy and immersed in pre-oxygenated cutting solution. The L3-L5 SCDH was then transversely sliced (300 μm) by Leica Vibratome in ice-cold cutting solution containing (in mM) 30 NaCl, 26 NaHCO₃, 10 Glucose, 194 sucrose, 4.5 KCl, 1.2 NaH₂PO₄, 1 MgCl₂ and continuously bubbled with carbogen (95% O₂-5% CO₂). Slices were then transferred to a perfusion chamber containing artificial cerebrospinal fluid (ACSF) (in mM): 124 NaCl, 4.5 KCl, 1 MgCl₂, 2 CaCl₂, 1.2 NaH₂PO₄, and 26 NaHCO₃, continuously bubbled in carbogen. The L3-L5 SCDH slices were transferred to a recording chamber continuously perfused with ACSF (2–3 ml min⁻¹) after 30-min incubation at room temperature. The glass pipettes filled with an internal solution containing 120 mM CsCl, 4 mM MgCl₂, 10 mM EGTA, 10 mM HEPES, 0.5 mM Na-GTP, and 2 mM Mg-ATP (pH 7.2 with CsOH, ∼280 mOsm) were used for glycinergic mIPSCs and glycine-activated currents recording. Meanwhile, tetrodotoxin at 1 μM, kynurenic acid at 4 mM, and bicuculline at 50 μM were added in continuously perfused ACSF.

For the AP recording of L3-L5 SCDH neurons, the superficial laminar II/III neurons in L3-L5 SCDH were randomly selected. The laminar II/III neurons in L3-L5 SCDH and their neuronal networks are diverse in terms of their electrophysiological properties and firing pattern (47, 48, 63, 64). Based on their firing pattern in response to the depolarizing current, these neurons can be mainly described as tonic, delayed, phasic, and gap neurons (47, 49). Here, among all the recorded neurons, only the delayed neurons such as PKCγ neurons were selected for linear correlation analysis since they are important in conveying neuropathic injury-induced mechanical allodynia within the excitatory cohort in the lumbar spinal cord (50, 51). Depolarizing currents (20 sweeps) were injected at the soma from 0 pA and increased with a steps of 10 pA. The glass pipettes were filled with an internal solution containing 145 mM K-gluconate, 5 mM HEPES, 5 mM Mg-adenosine triphosphate (ATP), 0.2 mM Na-guanosine 5′-triphosphate (GTP), and 10 mM EGTA (pH 7.2 with KOH, ∼280 mOsm) were used for AP recording. The pClamp 10.4 software (Molecular Devices, Sunnyvale, CA) was used for data analysis. The data for linear correlation analysis were obtained from all cells of the sham and SNI mice. The parameters of AP induced at 120 pA were averaged when a mouse has multiple valid data points. GlyRα1^S296A mice and their wild-type littermates were also used to evaluate the effects of DH-CBD at 10 μmol on glycinergic mIPSCs, glycine-activated currents, and APs.

Western blotting

Spinal cords were extruded with ice-cold PBS (pH 7.4) on Day 0, Day 1, Day 3, Day 7, and Day 14 after SNI unless otherwise indicated. The spinal cord L3–L5 segments were removed rapidly and collected in 1.5 ml tubes, followed by incubation for 60 min on ice. The tubes were then centrifuged at 12,000 rpm for 15 min at 4 °C. Samples were adjusted 1:1 with Laemmle loading buffer and denatured at 99 °C for 10 min. The samples were separated on 12% SDS/PAGE gels for 90 min at 120 V and then were electrotransferred to a 0.45 μm PVDF membrane at 240 mA for 1.5 h. The membranes were then blocked in 5% skim milk for 2 h at room temperature, followed by overnight incubation at 4 °C with the following primary antibodies: rabbit anti-GlyR (1:1000, H00002741-M02, NOVUS), rabbit anti-p-CaMKII (1:1000, ab124880, Abcam), rabbit anti-CaMKII (1:1000, ab22609, Abcam), mouse anti- CaMKIV(1:1000, sc55501, Santa Cruz), rabbit anti-p-CaMKIV(1:1000, ab59424, Abcam), rabbit anti-CREB (1:1000, 9197S, Cell Signaling), rabbit anti-p-CREB (1:5000, ab32096, Abcam) and rabbit anti-GAPDH (1:10,000, ab181602, Abcam) diluted in TBST. On the following day, the membranes were washed six times with TBST buffer and incubated with secondary goat anti-rabbit antibody conjugated with horseradish peroxidase (1:5000, SA00001-2, Proteintech) for 2 h at room temperature. Immunoblots were detected using the ECL system (Millipore Corporation). GAPDH was used as a loading control for total protein. Images of the Western blotting protein bands were collected and analyzed using Image J software (National Institutes of Health). To evaluate the effects of 666-15 and KN93 on the expression level of GlyRα1 protein, the L3-L5 SCDH tissue was extracted on Day 3 after SNI or sham surgery.

Intrathecal injections (DH-CBD, 666-15, and KN93)

C57BL/6J mice were intrathecally injected around lumbar vertebral segments L3-L5 at a 30° to 45° angle using a Hamilton syringe (30 gauge) eliciting a Straub tail response. All intrathecal injections were delivered in a total volume of 10 μl. DH-CBD at 0.1 μg, 1 μg, 5 μg, and 10 μg were i.t. injected on Day 14 after surgery. 666-15 at 10 μg was i.t. injected 1 day before surgery for four consecutive days. KN93 at 10 μg was i.t. injected 1 day before surgery for four consecutive days.

Drugs

All chemicals, including kynurenic acid, glycine, bicuculline, strychnine, 666-15, KN93, and tetrodotoxin were obtained from Sigma-Aldrich. The external solution was prepared the day before the recording experiment. Before electrophysiological recordings, the agonists, antagonists, and modulators were further diluted with an external solution. DH-CBD was synthesized according to the procedure described previously (40, 65). DH-CBD was dissolved in ethanol and diluted by an external solution before recording.

Quantification and statistical analysis

For behavioral tests, animals with different genotypes were picked randomly. For patch-clamp experiments, L3-L5 SCDH neurons were randomly picked for recordings. Data were statistically compared by unpaired t test, One-Way ANOVA, and Two-Way ANOVA using GraphPad Prism 8.0 (GraphPad Software), as indicated in the specific figure legends. Average values are expressed as the mean ± SD.

Data availability

All the data are contained within the article and supporting information. Data can be made available upon request of the lead contact.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Roger Colbran