Animal feeding and model establishing methods used in this study complied with the International Association for the Study of Pain Guidelines and have been approved by the Laboratory Animal Ethical and Welfare Committee, the Center for New Drug Safety Evaluation and Research, Hebei Medical University (NO. IACUC-Hebyd AP-2020033). The behavioral measurements on male Sprague Dawley (SD) rats (6–8 weeks) were all done in the awake state. The SD rats weighing 180–220 g were used in this study and were housed individually in automatically controlled environmental conditions, namely 12 h light-dark cycle (lights on from 08:00 to 20:00) and free access to food and water. Prior to the experiment, the rats were placed in the above-mentioned environment for 7 days to fully adapt, and the researchers should observe the rats’ health status every day. SD rats were randomly divided into 3 groups based on mechanical pain threshold (control group, model group, and GAS group) (Qiu et al., 2014; Xiao et al., 2016; Chen et al., 2017; Ye et al., 2018), the specific operations were as follows:

Control group: rats were intraperitoneally injected with the same volume of normal saline (vehicle) daily. Vincristine group: the rats were intraperitoneally injected with vincristine (Shenzhen Wanle Pharmaceutical Co., LTD.) at 9 a.m. every day (DAY 1–7) (0.125 mg/kg, diluted in saline before injection); then normal saline was injected into the abdominal cavity of rats at the same time period (DAY 8–10); finally, vincristine was injected intraperitoneally at the same time (9 a.m.) period on day 11–14. GAS group: DAY 1–7, rats received intraperitoneal injection of vincristine (Shenzhen Wanle Pharmaceutical Co., LTD.) at 9 am daily (0.125 mg/kg, diluted in saline before injection); then rats were injected with normal saline at 9 a.m. and GAS (4-hydroxybenzyl alcohol-4-O--D-glucopyranoside, Purity > 98%, purchased from Nanjing Daosifu Biotechnology Co., Ltd., product batch number: 20170811s) at 16:00 p.m. (60 mg/kg, diluted in saline before injection) from day 8 to 10 daily; finally, rats were intraperitoneally injected with vincristine at 9 a.m. and GAS at 4 p.m. according to the above-mentioned drug dosage (DAY 11–14) (Qiu et al., 2014; Xiao et al., 2016; Chen et al., 2017; Ye et al., 2018).

The von Frey instrument was used to assess the threshold sensitivity of mechanical stimuli in rats. The calibrated nylon filaments (von Frey hair, Stoelting) with various bending forces were applied to stimulate the middle plantar surface of the right hind paw of rats. The rats were stimulated from the minimum gram of nylon yarn and the hardness gradually increased. When the hindlimb of the rats was quickly retracted, the rats were considered to have a positive reaction. Hargreaves strategy was applied to measure thermal hyperalgesia using thermal radiation meter (Mengtai Technology). The paw withdrawal latency of the right hind-paw in response to heat was measured using 30% radiant intensity, and the elapsed time was recorded.

In the DRG neuron excitability recording experiment, the neurons were selected from 32 adult male SD rats (provided by the Experimental Animal Center of Hebei Province, People’s Republic of China), and according to literature reports (Nie et al., 2017; Zhang et al., 2018), DRG neurons were selected from the L4-L6 segment of rats. DRG ganglia were digested with collagenase (3 mg/mL) and dispase (7.5 mg/mL) for 30 min at 37°C, and then mechanically triturated and washed twice with DMEM supplemented with 10% fetal calf serum, which were further plated on poly-D-lysine-coated glass coverslips. Next, the action potential, Na_V1.8 current density, and TTX-sensitive sodium current were recorded.

According to the instructions, 800 ng RNA was reversely transcribed using PrimeScriptTM RT reagent Kit with gDNA Eraser (perfect real time) kit (Takara, Japan) (Wang et al., 2017; Wang et al., 2021); then, gene-specific mRNA analyses were conducted with SYBR premix ex TaqTMⅡ (TliRnaseH plus) kit (Takara, Japan) as standard protocol. In this study, Gapdh was applied as a reference to normalize the mRNA expression of SCN9A and SCN10A. After amplification, each qPCR product was electrophoresed to ensure specificity. After the components required for the PCR reaction were configured, the cycling systems were placed on the PCR machine and preheated at 95°C for 3 min to fully denature the template DNA, and then enter the amplification cycle stage.

In each cycle, the template was denatured at 95°C for 30 s, and then the temperature was lowered to the renaturation environment of 60°C for 30 s to fully anneal the primer and template. Finally, the prepared systems were kept in 72°C for 1 min (amplifying 1 kb fragment) to make the primer extend on the template and synthesize DNA. The above cycle was repeated 39 times to accumulate a large amount of amplified DNA fragments, and then kept at 72°C for 5 min to complete the extension of the products. Finally, the products were preserved at 4°C.

The genbank accession number of SCN9A and SCN10A were NM_133289.1 and NM_017247.1, respectively. The position of primer sequences of SCN9A located between 1174 and 1293 base pairs, and the primer sequences of SCN10A were located between 581 and 737 base pairs. In addition, SCN9A was the gene expressing the Na_V1.7 channel, and the primer was detailed in the following sequence: Na_V1.7 (SCN9A): forward TAC​CTG​ATA​AAC​TTG​ATC​CTG​GC; reverse TTT​GAG​TCG​GTC​TAA​CAT​CTG​CT; SCN10A was the gene expressing the Na_V1.8 channel, and the primer was detailed in the following sequence: Na_V1.8 (SCN10A): forward GTC​TGT​CCA​TTC​CTG​GTT​CTC​C; reverse ACA​AAA​CCC​TCT​TGC​CAG​TAT​CT.

The DRG neuron lysates were prepared with RIPA lysis buffer, and the protein loading amount was 40 µg. Equal amounts of protein were separated by SDS-PAGE and electro-transferred to the polyvinylidene fluoride (PVDF) membrane. PVDF membrane was blocked with 5% nonfat dairy milk, and incubated with primary antibody of Na_V1.7, Na_V1.8, and β-actin overnight at 4°C, which was further incubated with IRDye800-conjugated secondary antibody (1: 50,000; EARTHOX) for 2 h at room temperature and subsequently scanned with the Odyssey Infrared Imaging System (LI-COR). The primary antibody of Na_V1.7 (item number ARG56140) was monoclonal antibody provided by Shanghai Bio-Platform Technology Company (Shanghai, China), and the host was mice. The production number of Na_V1.8 was ARG56141, and the rest of the information was consistent with Na_V1.7. Monoclonal antibody for mouse β-actin (product number 66009-1-Ig) was provided by proteintech (Rosemont, IL, United States). The integrated intensity of polyvinylidene fluoride membrane was detected by Odyssey Imager software (version 3.0).

The model group neurons were incubated with vincristine for 24 h, and the GAS group was incubated with the mixture of GAS and vincristine for 24 h. DRG neurons were collected after washed by PBS for 3 times, and then washed with 4% paraformaldehyde for 30 min, which were transferred to 3% BSA and 0.3% Triton solution for 60 min. DRG neurons were blocked with 10% normal goat serum containing 0.3% Triton-X-100 (co-incubation for 1 h, 37°C). The sections were incubated with primary antibody for 12 h at 4°C (anti-Nav1.7 channel antibody: ARG56140; anti-Nav1.8 channel antibody: ARG56141), which were washed 5 times with PBS solution for minutes each time. The sections were incubated with secondary antibody for 2 h at room temperature, and then the secondary antibody was removed using PBS solution. Finally, the sections containing DRG were sealed with gelatin coating and placed in the dark for SIM imaging.

The ascending aorta of SD rats was perfused with saline solution, and then perfused with 4% paraformaldehyde (PFA, pH 7.4, 4°C). Subsequently, the DRG L5 ganglion was removed and fixed in 4% PFA for 24 h (4°C), and then soaked in 30% sucrose for 48 h (4°C) to dehydrate. After that, the dehydrated ganglion was sectioned into 14 μm thick slices in the cryostat using gelatinized slides, and then subjected to immunofluorescence treatment. Then, the sections were blocked with 0.2% Triton X-100 containing 2% BSA for 1 h at room temperature, and then incubated with primary antibody (anti-Na_V1.7 channel antibody: ARG56140; anti-Na_V1.8 channel antibody: ARG56141) overnight at 4°C. Finally, the sections were washed 3 times with PBS at room temperature (5 min for each time), and then co-incubated with secondary antibodies at room temperature for 1 h.

We constructed the three-dimensional structures of Na_V1.7&1.8 using homology modeling method.

In this study, Discovery Studio 2020 Client (DS 2020) was used for the prediction of binding pockets, molecular docking, and the estimation of the binding interactions between the receptor and GAS. Molecular docking was performed by CDOCKER module with the CHARMm force field. First, the three-dimensional (3D) structures of Na_V1.7 and Na_V1.8 were constructed using homology modeling method (Waterhouse et al., 2018), and processed by “prepare proteins module,” including Loop domain optimization, protonation, removing water, energy minimization using conjugate gradient algorithm with CHARMm force field. Second, GAS was sketched by ChemDraw 19.0, and prepared by the “Prepare Ligands module,” consisting of pH-based ionization, tautomers generation, generating 3D coordinates, rearranging hydrogens, and minimization. Then, the binding sites of the receptor were defined by the “From Receptor Cavities” method within the “Define and Edit Binding Site module,” and referring to the active sites reported in the literature (Ito et al., 1989; Swain et al., 2017; Moyer et al., 2018; Wu et al., 2018; Xu et al., 2019). Finally, the CDOCKER method was applied to evaluate the binding interactions of GAS at different sites with the docking score as an important criterion (Zhang et al., 2019a; Zhang et al., 2020a).

Molecular dynamic (MD) simulation was further used to evaluate the binding interactions between the GAS and the predicted pockets. MD simulations of the constructed systems were performed using GPU-accelerated PMEMD in Amber18 on 28 cores of an array of two 2.9 GHz Intel Gold 6226R processors and 6 pieces of Nvidia Tesla v100s 32 GB graphic card. Prior to MD simulation, all components (protein-ligand complex, protein, and ligand) were processed with program tleap embedded in AmberTools to generate the corresponding coordinate files and topology files. Amber ff14SB and general Amber force fields were applied for the receptor protein and the docked ligand, and the Li/Merz ion parameters for SPC/E water model were directly used based on the previous publications (Wang et al., 2018; Wang et al., 2019; Zhang et al., 2020b; Li et al., 2021; Liang et al., 2021). Antechamber was chosen to assign the charges of the docked ligands via the restrained electrostatic potential partial charges, and Gaussian 09 was used to optimize the ligand’s geometry and calculate the electrostatic potential calculations at HF/6–31G* level. Then, the processed systems were subjected to 200 ns MD simulation, mainly consisting of 2 steps energy minimization optimization process, progressive heating process, 5 ns equilibrium process, and the final 200 ns MD process. Program cpptraj was used to analyze the RMSD and representative conformations of MD trajectories, and mm_pbsa.pl program was selected to explore the binding free energies between the ligand and receptor.

The behavioral experiment results were processed and analyzed by OriginPro (version 9.1.0). Statistical difference among multiple groups was compared by one/two-way ANOVA, and Bonferroni Test was used for two group comparisons, which was presented as mean ± SD. HEKA FitMaster (version 2x90.3) and OriginPro (version 9.1.0) were used to analyze the electrophysiological data, and the above-mentioned method was applied to analyze and compare the statistical difference, presented by mean ± SEM. Threshold represents the voltage value at the inflection point during the rising phase of the action potential, recorded with the current clamp (Zhang et al., 2019b). Additionally, G = I / ( V m − E N a ) was applied to convert the peak inward currents obtained from activation protocols (G represented the conductance; I indicated the peak inward current; V m represented the instantaneous membrane potential; E N a indicated the equilibrium potential for sodium channel.) Maximum conductance value was used to normalize the conductance data, which was further fit with Boltzmann equation, namely G = G m a x + ( G m i n − G m a x ) / ( 1 + exp [ ( V m − V 1 / 2 ) / k ] ) (G also represented the conductance; V 1 / 2 was the midpoint of activation; k indicated the slope factor.). The IC₅₀ values were calculated using logistic equation fitting. Licor Odyssey software was applied to quantify the gray values of proteins in WB, and fluorescence intensity in immunofluorescence was measured by NIS-Elements Viewer (version 4.20, Nikon, Japan) software.

Vincristine has been reported to induce various pains, such as postherpetic neuralgia, diabetic neuralgia (Feld et al., 1980), and type I complex regional pain syndrome (Benzon et al., 2003), which has been associated with damage to the peripheral nerve terminal (Rosenthal and Kaufman, 1974; Dupuis et al., 1985; Old et al., 2014). Additionally, more than half of the rats treated with vincristine have been found to respond abnormally to thermal and mechanical pain induced by C fiber stimulus (Tanner et al., 1998; Xiao and Bennett, 2008; Xu et al., 2017). Therefore, vincristine was used to construct the CINP rat model, and thresholds of mechanical and thermal pains were measured (Figure 1A).

There was no difference in the mean paw withdrawal threshold of the three groups as determined by von Frey and radiant heat stimulus. The selected experimental rats were randomly divided into three groups (control group, vincristine group, and GAS group), and the basic values of mechanical and thermal thresholds were preliminarily evaluated before drug administration. The specific values were as follows: mechanical thresholds: 7.11 ± 1.03 g (control group), 7.02 ± 0.62 g (vincristine group), 6.83 ± 1.25 g (GAS group); thermal thresholds: 19.65 ± 1.78 s (control group), 20.19 ± 2.65 s (vincristine group), 20.21 ± 1.73 s (GAS group). After 7 days of vincristine administration (ip, 0.125 mg/kg), the values of mechanical and thermal thresholds were calculated, and the mean paw withdrawal threshold was significantly reduced compared with the control group: mechanical thresholds: 7.05 ± 1.00 g (control group), 3.69 ± 1.11 g (vincristine group), 4.02 ± 1.21 g (GAS group); thermal thresholds: 19.76 ± 2.15 s (control group), 17.11 ± 1.68 s (vincristine group), 16.63 ± 1.16 s (GAS group).

Continuous administration of GAS (ig, 60 mg/kg) for 7 days significantly attenuated the development of mechanical (n = 10) and thermal hyperalgesia (n = 23), as shown in Figure 1B. Quantitative analysis showed that, compared with the vincristine model, GAS increased the mechanical pain threshold from 3.11 ± 1.34 g to 8.92 ± 1.14 g, and the thermal withdrawal latency from 14.77 ± 2.32 s to 20.97 ± 3.01 s on the 14th day of modeling (mean ± SD, ***p < 0.001 and ***p < 0.001) (Figure 1C). Apparently, the applied modeling method successfully induced mechanical pain and thermal hyperalgesia in rats. After treatment with GAS, the thresholds of mechanical and thermal hypersensitivity were greatly improved, and the mechanical tactile and thermal allergic of the model rats could be restored to normal levels. In addition, in order to better illustrate the mechanisms underlying GAS alleviating vincristine induced peripheral neuropathic pain, the influences of GAS on the mechanical and thermal thresholds in normal SD rats were also observed. The results showed that the mechanical and thermal thresholds didn’t increase significantly after 3 days of GAS administration, indicating that the mechanism of GAS reversing vincristine-induced hyperalgesia was different from that of anesthetic effects (Supplementary Figure S1). The behavioral results suggested that GAS had a good curative effect on CINP model rats, which was expected to be a small active molecule for the treatment of peripheral neuropathic pain.

As primary sensory afferent neurons, the ectopic afferent discharge of DRG neurons is widely considered to be the main cause of chronic pain after peripheral nerve injury (Seijffers et al., 2007; Norcini et al., 2016). Thus, DRG neurons were isolated to observe the changes of their firing patterns. The alleviation of vincristine-induced rat hyperalgesia by GAS indicated that GAS might directly decrease the excitability of neurons. In order to explore the mechanism of GAS, the firing spikes of action potentials were recorded on small-diameter (17–25 μm) DRG neurons, which were closely related to the afferent noxious signal. GAS (100 μM) significantly attenuated hyperexcitability of vincristine-induced DRG neurons, and reduced the number of action potentials triggered by 500 pA (Figures 2A,C). 200 μM GAS and 500 nM PF-05089771 (Cat. No.: HY-12883, MCE) notably increased the rheobase current (depolarization current threshold eliciting the 1st action potential), namely, from 86.8 ± 3.1 pA (n = 22) to 203.3 ± 10.7 pA (n = 15, **p < 0.01) and 640.0 ± 66.2 pA (n = 10, **p < 0.01) (Figure 2D); the action potential threshold voltage also remarkably increased from −36.6 ± 0.7 mV (n = 17) to −21.3 ± 1.8 mV (n = 10, ***p < 0.001) and −12.3 ± 1.8 mV (n = 10, ***p < 0.001) (Figure 2F); 200 μM GAS and 500 nM PF-05089771 significantly reduced the amplitude of action potentials in DRG neurons (**p < 0.01) (Figure 2E). Figure 2B showed that GAS decreased the rate of action potential slope rise (dV/dt). Therefore, GAS could significantly reduce the number of action potential bursts of DRG neurons in model rats, decrease the amplitude of action potentials, and increase the rheobase and threshold of action potentials (Table 1), thereby inhibiting the excitability of DRG neurons in model rats, which was similar to that of the inhibitory effects of PF-05089771 on excitability of model DRG neurons.

The repetitive firing of DRG neurons depends on the inherent resonance characteristics of the cell membrane. Most small or medium-diameter DRG neurons cannot produce repeated firing during continuous depolarization, and step depolarization stimulus only causes a single spike or transient burst (Amir and Devor, 1996; Amir et al., 2005). According to Figures 2A,C, it could be found that vincristine significantly increased the number of repetitive discharges of DRG neurons, and the increase of ectopic discharges was the internal cause of neuropathy-induced pain sensation. GAS effectively suppressed the number and amplitude of action potential, thereby reducing the occurrence of hyperalgesia. In addition, GAS also significantly increased the thresholds and rheobase of action potentials and reduced the probability of action potential bursts. Furthermore, GAS reduced the upstroke slope of action potential’s rising phase, and slowed down the depolarization of the action potential, thus retarding the occurrence process of the action potential. Furthermore, based on the effects of GAS on the excitability of DRG neurons in normal SD rats, it could be found that GAS could increase the threshold and rheobase of action potentials to a certain extent, but had fewer effects on the number and amplitude of action potential, indicating that DRG neurons from normal SD rats were less sensitive to GAS compared to that of model DRG neurons (Supplementary Figure S2). In short, based on the biological functions of GAS discussed above, GAS reduced vincristine-induced hyperalgesia of primary sensory neurons associated with pain.

In order to determine whether GAS affects protein expression, we first observed the effects of GAS on the fluorescence intensity of Na_V1.7 & 1.8 channels on DRG small-diameter neurons in the L5 segment of model rats from the slice level. And the average fluorescence intensity of each group was as follows: 1) Na_V1.7 system: control group, 1.5 ± 0.5 (n = 82, N = 4); vincristine group, 2.6 ± 0.9 (n = 80, N = 5); GAS, 1.4 ± 0.5 (n = 115, N = 5) (Figures 6A,B). 2) Na_V1.8 system: control group, 1.5 ± 0.4 (n = 79, N = 4); vincristine group: 2.5 ± 0.8 (n = 120, N = 5); GAS group, 1.2 ± 0.5 (n = 74, N = 4) (Figures 6C,D), where “n” representing the number of cells and “N” representing the number of ganglions. The above results indicated that GAS could significantly reduce the fluorescence intensity of Na_V1.7 & 1.8 protein in DRG neurons of model rats.

DRG cells were incubated with vincristine and GAS mixture for 24 h and imaged using structured illumination microscopy (SIM) and confocal fluorescence microscopy. Immunohistofluorescence utilized moiré fringes generated by high-frequency stripe illumination to separate the high-frequency and low-frequency signals of the sample, making the ultrafine structure of the cell be seen. Fluorescence staining results showed that vincristine could up-regulate the distribution of Na_V1.7 and Na_V1.8 channel proteins in the DRG neurons, as shown in Figures 6E,F. GAS mixture incubation significantly reduced the enhanced fluorescence intensity of Na_V1.7 and Na_V1.8 membrane proteins induced by vincristine (**p < 0.01, Figures 6E,F). Therefore, GAS significantly reduced the fluorescence intensity of Na_V1.7 and Na_V1.8 membrane proteins enhanced by vincristine. The results further verified that GAS inhibits the expression of Na_V1.7 and Na_V1.8 channels. Nowadays, studies also pointed out that GAS could alleviate the peripheral neuropathy to some extend by inhibiting some signal pathways or receptors, such as the inhibitory activity of GAS on the P38/MAPK signal pathway (Sun et al., 2016; Qin et al., 2021), indicating the various action mechanisms of GAS, which was worthy of further study.