Adult (8–11 weeks) C57BL/6N mice(Vital River, Beijing)were fed with food and water ad libitum and housed five per cage, at 21°C, 50% humidity, on a 12 h light:12 h dark schedule in the standard animal facility in accordance with the guidelines of Tongji University. All animal work was conducted under ethical permission from the Tongji University ethical review panel.

Before measuring the mechanical threshold, the mice were placed on a metal mesh and covered with transparent plexiglass, allowing them to acclimate for 30 min. Mechanical threshold was measured according to the previously described procedure (Sun et al., 2021). Briefly, the paw-withdrawal threshold of the ipsilateral hind paws of sham and SCI mice was measured with a set of calibrated monofilaments (von Frey hairs) in order of increasing forces from 0.008 g to 2 g; the force which caused paw withdrawal for three times during continuous stimulation was recorded as the paw-withdrawal threshold. Each monofilament was applied for a maximum of five times. The response ratio of paw withdrawal was the ratio of the number of times the animal withdrew the paw to the total number of measurements.

Spinal cord injury (SCI) surgery was performed under isoflurane-induced anesthesia according to the protocol previously described (Sun et al., 2020) with a modification. Briefly, lumbar spinal segment L5 was exposed by laminectomy at vertebral level T13 (Figure 1A). Under a dissecting microscope, a small cut was made to the dura and arachnoid membrane followed by a defined lesion with a 26G needle (0.3 mm depth) unilaterally to the left dorsal horn, avoiding the dorsal root entry zone. Finally, skin was closed with 5–0 Nylon sutures. The surgical sham procedure was the same as SCI procedures except for injury of the dura, arachnoid membrane, and spinal cord. Mechanical pain threshold was measured at day 0 pre-surgery, and day 3, day 5, and day 8 after surgery. The mice with paw-withdrawal thresholds ≤0.4 g on day 8 after surgery were designated into the SCI group with NP and used for later experiment.

From day 8 to day 11 after surgery, Nav1.7 channel blockers including PF-05089771 (Tocris, #5931, 2 mg/kg, and 4 mg/kg) and GNE-0439 (ProbeChem, # PC-62325, 10 μg/kg, 20 ug/kg, and 30 ug/kg) were administrated individually via IP injection into adult C57BL/6 SCI mice and mechanical pain threshold was tested at 30- and 60-min post drug administration. The voltage-gated calcium channel blocker Gabapentin (Sigma, #G154, 50 mg/kg) was administrated via IP injection into SCI mice and mechanical threshold was tested at 30- and 60-min post drug administration. Each drug test had a vehicle (saline or 20% DMSO) group and pre-treatment threshold was firstly measured each day.

On day 12 after surgery, mice were anaesthetized and perfused using saline via left ventricle, and then L4-6 DRGs, L4-6 SDH, or L5 SDH from SCI mice, sham mice, and control mice were dissected and were immediately lysed with RIPA lysis buffer (Beyotime, China). The lysates were centrifuged at 4°C and 12,000 g for 10 min to pellet remaining cells and the cellular debris. About 20 μg DRG protein or 40 μg SDH protein from each animal were separated by SDS-PAGE in 10 and 6% (3:4) mixed cast gels and transferred onto PVDF membranes (Merck, USA). After 1h-blocking with 5% BSA in PBS, the membranes were incubated with primary antibody at 4 degree for overnight. Primary antibodies included: rabbit antibodies against Nav1.2 (Alomone labs, ASC-022, 1:100), NGF (Abcam, ab52918, 1:1,000), Phospho-c-JUN (Cell signaling technology, #9261, 1:1,000), and GAPDH (Proteintech, 60004-1-Ig, 1:3,000), and mouse antibodies against Nav1.7 (Abcam, ab85015, 1:800) and Nav1.8 (NeuroMab, 75–166, USA, 1:1,000). After washing three times in TBST, the membranes were further incubated with horseradish peroxidase (HRP)-conjugated horse anti-mouse secondary antibodies (Cell signaling technology, 7076S, 1:3,000), or HRP-conjugated goat anti-rabbit secondary antibodies (Cell signaling technology, 7074S, 1:3,000) for 1 h at room temperature. The proteins were visualized by chemiluminescent method using the ECL and detected using Western blot imaging machine (CLiNX, China). The intensity of each protein band was normalized to the intensity of GAPDH band to get the relative expression level of the interested protein.

On day 12, mice were anaesthetized and perfused via left ventricle using saline, followed by 4% PFA, and then L4-6 DRGs, L4-6 SDH, or L5 SDH from SCI mice, sham mice, and control mice were dissected and were immersed in 4% PFA overnight. They were then soaked in 15% sucrose for 24 h and 30% sucrose for 24 h; afterwards, tissues were embedded in OCT. DRG and spinal cord tissues were sectioned in pieces of 14 μm thickness using a cryostat, and immunostaining was performed as previously described (Peng et al., 2017). The spinal cord sections and dorsal root ganglia sections were incubated overnight (2 days for Nav1.7 antibodies) at 4°C with primary antibodies, including rabbit antibodies against Nav1.7 (Proteintech Group, 20257-1-AP, 1:200), NGF (Abcam, ab52918, 1:300), and Phospho-c-JUN (Cell signaling technology, #9261, 1:300). The sections were further incubated with IgG Alexa Fluor 488 (Invitrogen, 1:1,000) and IgG Alexa Fluor 555 (Invitrogen, 1:1,000) secondary antibodies for 1.5 h at room temperature. According to the manufacturer’s instructions, we first used tyramide signal amplification (TSA) kits (PerkinElmer, NEL701A001KT, 1:50) to probe Nav1.7 (Proteintech Group, 20257-1-AP, 1:3,000) on the spinal cord sections, then incubated the rabbit antibodies against FOS (Abcam, ab190289, 1:1,000). The sections were further incubated with IgG Alexa Fluor 555 (Invitrogen, 1:1,000). Finally, the sections were counterstained with DAPI (Sigma, MBD0015, 1:10,000). Fluorescent images were captured by confocal laser scanning microscope (LSM780 or LSM880, Carl Zeiss, Oberkochen, Germany).

All quantitative data were presented as the mean ± standard deviation (SD). The data were analyzed using GraphPad Prism 9.0.0 software (GraphPad Software Inc., CA, USA), The data collected from von Frey test, protein level quantification, and immunofluorescence staining cell number were analyzed by unpaired t-test or one-way ANOVA, and the response ratios of paw withdrawal of SCI mice were analyzed by two-way ANOVA. A value of * p < 0.05 was considered as statistically significant.

To investigate whether the expression levels of sodium channels changed in SDH and DRG after SCI, we generated SCI mouse models by making a restricted narrow lesion on the unilateral dorsal horn of lumbar segment 5 of the spinal cord while keeping the dorsal column and ventral horn intact (Figures 1A,B; Sun et al., 2020). Mechanical pain developed in the ipsilateral paws of SCI mice three days post operation (DPO), and the threshold of mechanical pain further lowered from three to eight DPO, but the paw-withdrawal threshold of sham mice did not significantly change after the sham operation (Figures 1C,D). These results indicated that the SCI model was successfully generated and NP was developed in these SCI mice.

We then examined the expression level of Nav1.2 (SCN2A), Nav1.8 (SCN10A), and Nav1.7 (SCN9A) in SDH, and the result of the Western blot showed that none of the expression levels of Nav1.2, Nav1.8, or Nav1.7 in L4-6 segment SDH of SCI mice was significantly changed when compared to that in naive mice (Figures 2A–C, uncropped image of membrane showed in Supplementary Figure S1). However, these results showed an upregulation trend of Nav1.7 (1.1 ± 0.18) and Nav1.2 (1.1 ± 0.39) in SDH of the SCI mice (Figures 2A–C). It is possible that spinal cord injuries might affect the expression of Nav1.7 and Nav1.2 in DRG, so we detected the expression levels of Nav1.7 and Nav1.2 in DRG using Western blot. The results showed that Nav1.7 was significantly upregulated 1.5-fold in DRG of SCI mice when compared to that in naive mice (Figure 2D, uncropped image of membrane showed in Supplementary Figure S1), but the expression of Nav1.2 was undetectable in the DRG of both SCI and naive mice (data not shown).

We previously found that sham operations (cutting open the skin of the leg and destroying the peripheral nerve of DRG neurons in the skin) altered the expression Scn9a mRNA in SDH. To decipher the contributions of spinal injuries and sham operations, which include peripheral injuries in the back skin and drilling a hole in vertebrate to NP, we employed sham group mice and found that there was an increase of Nav1.7 expression in L4-6 DRG of sham mice (1.1-fold), but no statistical significance when compared to that in naive mice; the expression of Nav1.7 in L4-6 DRG was significantly upregulated 1.3-fold more in SCI mice than that in naive mice (Figures 3A,B, uncropped image of membrane showed in Supplementary Figure S2). These data indicated that spinal cord injuries indeed contributed to the upregulation of Nav1.7 in DRG although the sham operation of SCI surgery also mildly enhanced the expression of Nav1.7 in L4-6 DRG.

As the damaged area of the spinal cord was small due to only one minor lesion being made in our model, in order to measure the change of gene expression in injured tissue we then dissected injury-site-contained L5 segment SDHs only and excluded the L4 and L6 segments of SDH which were further from the injury site. The Western-blot results showed that the expression level of Nav1.7 was increased 1.5-fold and 2.2-fold in SDH of sham and SCI mice, respectively, although it was not statistically significant (Figures 3A,B). In order to understand if SDH neurons in SCI mice gained expression of Nav1.7, we performed immunostaining for Nav1.7 and found that the number of Nav1.7-expressing neurons significantly increased 2.9-fold and 12.9-fold in SDH of sham (15 ± 3.1 cells per section) and SCI (64 ± 4.9 cells per section) mice, respectively, when compared to that in naive mice (5 ± 1.7 cells per section) (Figures 3C,D). The increased Nav1.7-expressing neurons were distributed from laminae I-VI in SCI mice (Figure 3D). Immunostaining on DRG sections also showed upregulation of Nav1.7 in DRG neurons of SCI mice and sham mice when compared to that in naive mice (Figure 3E).

In order to confirm if the upregulation of Nav1.7 in DRG contributed to NP induced by SCI, we employed the blood–brain-barrier (BBB) non-permeable blocker of Nav1.7, PF-05089771(McDonnell et al., 2018). Intraperitoneal injection of 2 mg/kg PF-05089771 significantly alleviated mechanical pain when compared to vehicle (Figure 4A), and 4 mg/kg of PF-05089771 did not increase the efficacy of pain relief in SCI mice (Figure 4A). These data indicated that the upregulation of Nav1.7 in DRG indeed participated in mechanical allodynia. To investigate if the ectopic expression of Nav1.7 in SDH neurons contributed to NP in SCI mice, we employed another Nav1.7 blocker GNE-0439 (Chernov-Rogan et al., 2018) which is able to penetrate BBB (data not shown). GNE-0439 also attenuated mechanical pain in SCI mice significantly at a dose range from 10 to 30 μg/kg, with maximal efficacy at a dose of 20 μg/kg, but the effects of GNE-0439 declined 30 to 60 min post drug administration (Figure 4B). Intraperitoneal injection of 50 mg/kg Gabapentin also significantly relieved mechanical pain in SCI mice (Figure 4C). However, Gabapentin at this dose caused sedation and movement disability in one out of seven tested mice (data not shown). Although there was not a statistically significant difference, the maximal efficacy of GNE-0439 and Gabapentin in the SCI mice was slightly better than that of PF-05089771 (PF-05089771 0.4 ± 0.20 g, GNE-0439 0.5 ± 0.28 g, Gabapentin 0.5 ± 0.36 g) (Figure 4D). GNE-0439 had equivalent efficacy of pain relief as Gabapentin, but no sedation side effect. We further compared the response ratio of SCI mice to mechanical pressing stimulus from 0.07 g Von Frey to 1.0 g Von Frey after administration of Nav1.7 blockers, and found that GNE-0439 showed better effects to alleviate mechanical pain than PF-05089771 did (Figure 4E). These data suggested that the upregulation of Nav1.7 in DRG and SDH may contribute to NP in SCI mice.

Nav1.7 plays an important role in determination of action potential threshold (Bennett et al., 2019), and we previously found that mild mechanical pressing caused mechanical allodynia and activated FOS expression in SDH neurons with ectopic expression of Nav1.7 in Gad2^CreERT2/+; miR-96^flox/flox mice which had ablation of miR-96 in GAD2 neurons in SDH, but not in DRG (Sun et al., 2021). To investigate if SDH neurons with ectopic expression of Nav1.7 in SCI mice participated in pain conductions, we performed double immunostaining for Nav1.7 and FOS, which is a marker indicating the activation of neurons. The results showed the number of activated neurons with FOS expression increased significantly more in SCI mice than that in sham and naive mice (Figures 5A–M). Moreover, the number of Nav1.7/FOS double-positive neurons was also significantly elevated in laminae III–V of SCI mice (Figures 5D–F,N). Sham mice and naive mice had only a few and no neurons expressing both Nav1.7 and FOS in SDH, respectively. Further analysis showed that more than one quarter (26 ± 8.2%) of Nav1.7-expressing SDH neurons were activated in SCI mice which had mechanical stimuli mainly from walking in their home cages (Figures 5O,P). These data together suggested that SCI mice had mechanical pain during walking in their home cages and that ectopic expression of Nav1.7 made neurons in deep laminae layers hyperactive and participatory in the conduction of mechanical pain.

It was reported that the expression of Nav1.7 in DRG could be induced by nerve growth factor (NGF) (Toledo-Aral et al., 1997; Liu et al., 2021), so we examined whether the expression of NGF was upregulated after SCI. The results of the Western blot showed that the expression level of NGF increased 1.7-fold and 1.2-fold in DRG and SDH, respectively, in SCI mice when compared to that in naive mice (Figures 6A,B, uncropped image of membrane showed in Supplementary Figure S3), although it was not statistically significant. We then detected the expression of NGF on a cellular level by immunostaining and found that the number of NGF-expressing SDH neurons significantly increased in laminae IV–VI of SCI mice when compared to those in both sham and naive mice (Figures 6C,D). The immunostaining also demonstrated that the number of neurons expressing high levels of NGF (NGF^H+) was upregulated in L4-6 DRG of SCI and sham mice when compared to that in naive mice (Figures 6E,F), which was in line with previous reports that the expression of NGF in DRG was induced by injuries of the skin and muscle (Liu et al., 2021). We then asked how NGF activated transcription of Scn9a. It is known that NGF can activate MAPK/ERK signaling pathways in DRG neurons (Obata et al., 2004) and that MAPK/ERK activates JUN (Raitano et al., 1995). We found that there are predicted binding sites of JUN and FOS on the promoter of Scn9a ¹ , so it is possible that NGF induces the upregulation of Nav1.7 through MAPK/ERK/JUN/FOS. Therefore, we examined the expression of phosphorylated JUN by Western blot and immunostaining. The result of Western blot showed that the expression levels of phosphorylated JUN was not significantly increased in DRG and SDH of SCI mice when compared to that in naive mice (Figures 7A,B, and uncropped image of membrane showed in Supplementary Figure S4). However, the number of phosphorylated JUN-expressing SDH neurons significantly increased in laminae I–VI of SCI mice when compared to naive mice, and SCI mice had more phosphorylated JUN-expressing SDH neurons on laminae IV than sham mice (Figures 7C–O). The immunostaining also demonstrated that the number of neurons expressing high levels of phosphorylated JUN (Phos-JUN ^H+) was upregulated in L4-6 DRG of SCI mice when compared to that in naive and sham mice (Figures 7P–V). Moreover, phosphorylated JUN was found in the nucleus of DRG and SDH neurons of SCI mice (Figures 7N,U). These data suggest that NGF induced Nav1.7 expression via transcription factor JUN in both SDH and DRG after spinal cord injury.