Oxidative stress induced by NOX2 contributes to neuropathic pain via plasma membrane translocation of PKCε in rat dorsal root ganglion neurons

Background Nicotinamide adenine dinucleotide phosphate oxidase 2 (NOX2)-induced oxidative stress, including the production of reactive oxygen species (ROS) and hydrogen peroxide, plays a pivotal role in neuropathic pain. Although the activation and plasma membrane translocation of protein kinase C (PKC) isoforms in dorsal root ganglion (DRG) neurons have been implicated in multiple pain models, the interactions between NOX2-induced oxidative stress and PKC remain unknown. Methods A spared nerve injury (SNI) model was established in adult male rats. Pharmacologic intervention and AAV-shRNA were applied locally to DRGs. Pain behavior was evaluated by Von Frey tests. Western blotting and immunohistochemistry were performed to examine the underlying mechanisms. The excitability of DRG neurons was recorded by whole-cell patch clamping. Results SNI induced persistent NOX2 upregulation in DRGs for up to 2 weeks and increased the excitability of DRG neurons, and these effects were suppressed by local application of gp91-tat (a NOX2-blocking peptide) or NOX2-shRNA to DRGs. Of note, the SNI-induced upregulated expression of PKCε but not PKC was decreased by gp91-tat in DRGs. Mechanical allodynia and DRG excitability were increased by ψεRACK (a PKCε activator) and reduced by εV1-2 (a PKCε-specific inhibitor). Importantly, εV1-2 failed to inhibit SNI-induced NOX2 upregulation. Moreover, the SNI-induced increase in PKCε protein expression in both the plasma membrane and cytosol in DRGs was attenuated by gp91-tat pretreatment, and the enhanced translocation of PKCε was recapitulated by H2O2 administration. SNI-induced upregulation of PKCε was blunted by phenyl-N-tert-butylnitrone (PBN, an ROS scavenger) and the hydrogen peroxide catalyst catalase. Furthermore, εV1-2 attenuated the mechanical allodynia induced by H2O2 Conclusions NOX2-induced oxidative stress promotes the sensitization of DRGs and persistent pain by increasing the plasma membrane translocation of PKCε. Supplementary Information The online version contains supplementary material available at 10.1186/s12974-021-02155-6.


Background
Neuropathic pain is one of the most intractable neurological diseases due to the lack of effective clinical treatments to date [1]. Oxidative stress generated by superoxide anion (·O 2 − ), hydroxyl radical (·OH), and hydrogen peroxide (H 2 O 2 ) plays a crucial role in multiple pathological pain states, such as neuropathic pain [2,3], migraine-related pain [4], and inflammatory pain [5]. A proposal that O 2 − was intimately involved in the inflammatory response was raised as early as the 1970s [6]. It implicates that reactive oxygen species (ROS) is one of mediators of neuroinflammation [7] as·O 2 − , H 2 O 2 , and·OH occurs at the site of inflammation [8]. Accordingly, scavenging reactive oxygen species (ROS) with phenyl-N-tert-butylnitrone (PBN) or catalyzing H 2 O 2 with catalase in the spinal dorsal horn (SDH) has been shown to alleviate nerve injury-induced mechanical allodynia or antinociceptive tolerance to morphine [2,9]. Nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (NOX) is one of the major sources of ROS in the nervous system and includes seven members: NOX1-5 and Duox1-2 [10]. Our recent study demonstrated that nicotinamide adenine dinucleotide phosphate oxidase 2 (NOX2)-induced ROS production participates in the development of persistent pain through increasing the excitability of SDH neurons of rats under a high-frequency stimulation (HFS) at the sciatic nerve [11]. Moreover, Kallenborn-Gerhardt et al. showed that nerve injury activated NOX2-mediated oxidative stress in dorsal root ganglia (DRGs), leading to neuropathic pain behavior, which were reduced in Nox2-deficient mice [12]. This is suggested that NOX2 plays a critical role in neuropathic pain. However, whether NOX2-ROS could promote neuropathic pain though regulating the peripheral sensitization in DRGs is still unclear. = In addition to oxidative stress, the activation of protein kinase C (PKC) in DRGs plays a crucial role in pathological pain [13][14][15]. The specific distribution of PKC via plasma membrane translocation is a marker of PKC activation. In small, medium, and large DRG neurons, plasma membrane translocation of PKC has been shown to produce chemotherapy-induced peripheral neuropathy [16] or bone cancer-induced hyperalgesia [17]. Moreover, ROS production by NOX2 can modulate PKC activity, resulting in pain hypersensitivity in the SDH [18]. However, whether NOX2-induced oxidative stress in rodent DRGs after nerve injury can promote plasma membrane translocation of PKC remains unknown.
In this study, we investigated whether oxidative stress induced by NOX2 in DRGs leads to mechanical allodynia in spared nerve injury (SNI) models of neuropathic pain and the underlying mechanisms. We found that NOX2-mediated ROS and H 2 O 2 activation induced hyperexcitability of DRG neurons in SNI rats. In addition, this activation promoted PKCε activation and plasma membrane translocation, which led to mechanical allodynia. In summary, activation of NOX2-ROS-PKCε signaling in DRGs was necessary and sufficient to generate neuropathic pain.

Animals and surgery
Male Sprague-Dawley (SD) rats weighting 200-250 g were used unless otherwise mentioned. They were obtained from the Animal Center of Nanchang University. The rats were housed in a temperature-controlled room, maintaining at 24 ± 1°C and 50-60% humidity, with a 12:12-h light/dark cycle. Food and water were available ad libitum. Experimental procedures were approved by Institutional Animal Care and Use Committee of Nanchang University and according to ARRIVE guidelines [19]. Spared nerve injury (SNI) was performed as previously described [20]. Briefly, under isoflurane (1.5-2.5%) anesthesia, the sciatic nerve of the left hindlimb was exposed to common peroneal, tibial, and sural nerves, and then the former two nerves were ligated and sectioned, whereas the sural nerve was kept intact. In the sham group, the sciatic nerve was only exposed without ligation or cut. All of the animals were assigned to different treatment groups randomly.
To verify if the treatment of drug was limited to the specific DRG, CFSE (21888, Sigma; 100 μM in 2 μl each DRG), a fluorescent dye, was applied onto L5 DRG by gelatin sponge. One hour after application, the animal was sacrificed, and L5 DRG and proximal spinal cord segment were excised. The DRG sections (16 μm) and spinal cord sections (25 μm) were cut in a cryostat (CM1950, Leica, Nussloch, Germany). Fluorescent images were obtained with a laser confocal microscope (LSM700, Zeiss, Germany).

Assessment of mechanical sensitivity
Mechanical sensitivity of rats were assessed with the up-down method following the previous study [23]. After acclimation for 30 min on a plastic box, a set of Von Frey filaments (North coast, USA) with logarithmically incremental stiffness from 0.4 to 15.0 g were applied alternately to the plantar surface of the hind paw. Each stimulus consisted of a maximum of 6 s application of the filament, quick withdrawal or licking on the paw in response to the stimulus was considered a positive response. The behavioral test was performed on doubleblinded design.

Whole-cell patch-clamp recording
Whole-cell patch-clamp recording was randomly selected from small size (< 25 μm) and medium size (25-35 μm) DRG neurons, and was obtained by an EPC-10 amplifier and the PatchMaster program (HEKA Electronics, Lambrecht, Germany) as previously described [25]. The membrane capacitance was read from the amplifier by PatchMaster. Patch pipettes with 3-5 MΩ resistance were pulled from borosilicate glass (World Precision Instruments, Sarasota, FL, USA) using a Sutter P-97 puller (Sutter Instruments, Novato, CA). For current-clamp recording, the extracellular solution contained (in mM) 117 NaCl, 3. . The action potentials (APs) of DRG neurons were elicited by a series of depolarizing currents every 10 s with step intervals of 10 pA from 0 to 400 pA over a period of 500 ms to measure the current threshold (rheobase). The current that induced the first AP was defined as 1×rheobase. The peak potential and half-width of APs were analyzed by Clampfit software (pClamp10, Molecular Devices). The neuronal input resistance was determined by measuring the voltage response to a depolarizing current (10 pA, 500 ms) from RMP in current-clamp mode.

Immunocytochemistry and immunohistochemistry
DRG neurons on coverslips were fixed by 4% paraformaldehyde (PFA) for 10 min at room temperature. After blocking with 10% donkey serum and 0.3% Triton X-100 for 1 h, the neurons were incubated with rabbit anti-PKCε (1:200, ab63638, Abcam) overnight at 4°C. The cultured neurons were then incubated with FITC secondary antibody (1:200, Jackson ImmunoResearch, West Grove, PA) for 1 h at room temperature.
For frozen section, rats were transcardially perfused with 4% PFA. The L4-L6 DRGs were dissected and postfixed in 4% PFA for 1 h and dehydrated in 30% sucrose for over 3 nights. The DRG sections (16 μm) were cut in a cryostat (CM1950, Leica, Nussloch, Germany) and then incubated with primary antibodies against NOX2 (1:100, DF6520, Affinity Biosciences, China), PKCε Fluorescent images of DRG neurons on coverslips or DRGs sections were obtained with the laser confocal microscope (LSM700, Zeiss, Germany). Determination of the PKCε membrane translocation was based on the fluorescence intensity of the cells as previous reports [5,16]. Briefly, the fluorescence intensity along a straight line across the neuronal soma was measured by Zen software (Zeiss, Germany). The ratio of averaged fluorescence intensity of the membrane region against the total fluorescence intensity of neuronal soma was demonstrated. The size of cultured DRG neurons was classified as small: < 600 μm 2 , medium: 600-1200 μm 2 , and large: > 1200 μm 2 [16]. For quantification of immunofluorescence staining, 8-OHG-immunoreactivity (IR) in L4-L6 DRGs per section was measured by ImageJ (Version: k 1.45).

Measurement of hydroxyl radical (·OH)
Level of ·OH was analyzed by Cell Hydroxyl Radical Colorimetrical Assay Kit (GMS10124.4, GenMed Scientifics Inc., MA, USA) according to the manufacturer's instructions. Briefly, DRGs tissue in 400 ml medium was collected in a 5 ml glass tube, and reacted with reaction mixtures provided by the kit at room temperature for 30 min. Absorbance at a wavelength of 420 nm was measured by a spectrophotometer (U-2900, HITACHI, Japan).

Statistics analysis
Data were shown as means ± SEM, and were analyzed by GraphPad Prism 8.4 software. The sample sizes were based on our previous research. The Shapiro-Wilk test was used to check the normality of data, which were considered normally distributed when p > 0.05; otherwise, non-parametric tests were used. Homogeneity of variance was be verified by Brown-Forsythe tests, when p < 0.05, data was analyzed by one-way ANOVA followed by Dunnett's T3 multiple comparisons test. Two-tailed unpaired Student's t test was used for analyses with two groups; one-way ANOVA followed by Tukey's post hoc test was used to compare multiple groups. Nonparametric analysis was used when datasets did not pass the normality test. Kruskal-Wallis or Friedman test was used for multiple group statistical evaluation followed by Dunn's multiple comparisons test. The Mann-Whitney test was used for analyses with two groups. p < 0.05 was considered statistically significant.
Local application of NOX2-blocking peptides or shRNA attenuates SNI-induced mechanical allodynia and hyperexcitability of DRG neurons It has been reported that gp91-tat, a 20-amino acid NOX2-blocking peptide that prevents the binding of p47 phox to gp91 phox and blocks enzyme assembly and activation [26]. To examine whether increased NOX2 expression in DRGs was responsible for SNI-induced pain hypersensitivity, we locally applied gp91-tat to ipsilateral L4-L6 DRGs 2 hours before SNI surgery. Site-specific drug delivery was verified by CFSE staining, as the CFSE signal was confined to the L5 DRG ( Fig. 2a shows that gp91-tat pretreatment significantly decreased the 50% PWT at day 3, and the effect persisted until day 14 after SNI. In contrast, the 50% PWT did not markedly differ between vehicle-and gp91-tat-treated SNI rats when applied 5 days after surgery (Fig. 2c). Consistently, Western blot analysis revealed that gp91tat pretreatment reversed the SNI-induced upregulation of NOX2 in DRGs at day 7 after surgery ( Fig. 2b).
Because the hyperexcitability of DRG neurons is critical for neuropathic pain development following peripheral nerve injury [27], we next tested whether the attenuation of SNI-induced mechanical allodynia by the NOX2-blocking peptide gp91-tat occurred by decreasing the hyperexcitability of DRG neurons. Whole-cell patch clamp recordings were obtained from small (< 25 μm) and medium (25-35 μm) L4-L6 DRG neurons to examine the characteristics of the APs of DRG neurons. To do this, DRG neurons were prepared from sham rats, rats at 7 days after SNI, and gp91-tat-pretreated rats at 7 days after SNI. As shown in Fig. 2e and g(a), the rheobase was significantly decreased in SNI rats at 7 days after surgery (88.89 ± 17.75 pA) compared with that of sham rats (196.00 ± 23.49 pA, p < 0.01, one-way ANOVA followed by Tukey's test). However, compared with SNI rats, pretreatment with gp91-tat induced a profound increase in the rheobase (160 ± 17.6 pA, p < 0.05, one-way ANOVA followed by Tukey's test). Next, we examined the number of APs elicited by 1× rheobase, which was increased by SNI to 6.22 ± 1.77 ( Fig. 2f and  g(b), p < 0.05, Kruskal-Wallis test followed by Dunn's multiple comparison test). Pretreatment with gp91-tat decreased the number of APs (1.82 ± 0.40, p = 0.3411, Kruskal-Wallis test followed by Dunn's multiple comparison test) but did not reach statistical significance, indicating a trend toward inhibiting DRG overexcitation following SNI. The membrane capacitance, peak potential, half width of APs, and input resistance were not Fig. 1. SNI induces NOX2 upregulation in DRG neurons. a Paw withdrawal threshold (PWT) after SNI (n = 5-6/group). Mann-Whitney U test. *p < 0.05, **p < 0.01 versus the sham group. b NOX2 expression in L4-L6 DRGs from rats in (a) at 1, 7, and 14 days after SNI and sham surgery was determined by Western blotting (n = 4/group). n.s. indicates not significant. One-way ANOVA followed by Tukey's test. *p < 0.05 versus the sham group. c Representative double-immunofluorescence staining showing the colocalization of NOX2 with IB4 (c(a)), CGRP (c(b)), NF-200 (c(c)), and CD11b (c(d)) but not GFAP (c(e)) at day 7 after SNI. c(f) The histogram shows the percentage of NOX2-positive cells among the various cell markers (n = 6/group) different among the three groups ( Fig. 2g(c, d)). Taken together, these results confirm that NOX2 activation in DRG neurons plays an essential role in SNI-induced neuropathic pain by increasing DRG excitability.
To further confirm the effect of NOX2 on neuropathic pain, we used a cybb-shRNA to knockdown the expression of NOX2. AAV-cybb-shRNA-mCherry or AAVscramble-shRNA-mCherry was injected into the DRGs Fig. 2 Pretreatment with NOX2-blocking peptides or shRNA attenuates SNI-induced mechanical allodynia and hyperexcitability in DRG neurons. a Images of the L5 DRG (a(a, b)) and proximal spinal cord (a(c, d)) after the local application of CFSE (100 μM in 2 μl) by gelatin sponge. b, c Local application of gp91-tat (26 μg or 52 μg) to L4-L6 DRGs before (b) but not after (c) SNI surgery significantly attenuated mechanical allodynia at 3-14 days after SNI. The dotted arrow indicates the time of gp91-tat or vehicle application. Kruskal-Wallis test followed by Dunn's multiple comparison test. *p < 0.05, **p < 0.01, ***p < 0.001 versus the sham group; #p < 0.05, ##p < 0.01 versus the vehicle-treated SNI group. d Western blotting showed that pretreatment with gp91-tat decreased NOX2 expression in L4-L6 DRGs at 7 days after SNI (n = 7). One-way ANOVA followed by Tukey's test. *p < 0.05 versus the sham group, #p < 0.05 versus SNI group. e Representative traces illustrating the action potentials and their rheobase currents recorded in L4-L6 DRG neurons. f Representative traces showing the action potentials elicited by 1×rheobase in L4-L6 DRG neurons. g Quantification analysis of rheobase currents (a), one-way ANOVA followed by Tukey's test. **p < 0.01 versus the sham group, #p < 0.05 versus the SNI group. Numbers of action potentials (b), Kruskal-Wallis test followed by Dunn's multiple comparison test. **p < 0.01 versus the sham group. Membrane capacitances (c). Peak amplitudes (d) and half-widths (e) of action potentials, as well as input resistance (f) in each group (n = 9-11/group). h Experimental diagram showing the timeline of AAV-shRNA injection and behavioral tests. i(a-b) Representative immunofluorescence staining showing the expression of cybb-shRNA-mCherry or scramble-shRNA-mCherry in transduced cells. j Western blotting analysis showed that cybb-shRNA injected into the DRGs diminished the protein expression of NOX2 (n = 4). Two-tailed unpaired Student's t test. ***p < 0.001 versus scrambled shRNA. k NOX2 cybb-shRNA but not scrambled shRNA significantly abolished the initiation of mechanical allodynia after SNI (n = 5-6/group). Kruskal-Wallis test followed by Dunn's multiple comparison test. *p < 0.05, **p < 0.01 versus the sham group; #p < 0.05 versus the cybb-shRNA group of the operation side 21 days before SNI (Fig. 2h). The immunohistochemical results demonstrated that cybb-shRNA-mCherry and scramble-shRNA-mCherry were expressed in DRGs at day 35 after SNI (Fig. 2i). Western blot further showed that the expression of NOX2 was downregulated after cybb-shRNA compared to scramble shRNA application (Fig. 2j). Behavioral tests also demonstrated that cybb-shRNA induced a marked reduction in the 50% PWT at 5, 7, and 14 days after SNI (Fig. 2k).

Oxidative stress induces mechanical allodynia and hyperexcitability of DRG neurons
It has been well documented that oxidative stress caused by ROS (e.g., superoxide and ·OH) and H 2 O 2 , which may be generated by NOX, is crucial in the development and maintenance of neuropathic pain in the central and peripheral nervous systems [3,28,29]. To investigate the role of SNI-induced oxidative stress in DRGs, we used 8-OHG-IR to examine oxidized nucleic acids produced by cellular ROS damage. We found that the levels of 8-OHG-IR in the DRGs were increased at days 4 and 7 after SNI, whereas pretreatment with gp91-tat reversed the increase in 8-OHG-IR at day 7 after SNI ( Fig. 3a and  b). Figure 3c shows that nonnuclear 8-OHG immunostaining significantly increased at day 4 after SNI, whereas pretreatment with gp91-tat reversed this increase. Nonnuclear 8-OHG immunostaining indicates oxidative stress in mitochondria [30], which produce superoxide that is converted to H 2 O 2 by superoxide dismutase (SOD) [7]. Double immunostaining demonstrated that 8-OHG colocalized with IB4-, CGRP-, and NF-200-positive DRG neurons but not GFAP-positive glial cells at 7 days after SNI (Fig. S2A). The level of ·OH, which is the protonated form of ·O 2 − , was increased in the DRG tissue of the SNI group. This increase at 7 days after SNI was blocked by pretreatment with gp91-tat (Fig. 3d).
Next, we investigated the effect of the direct application of H 2 O 2 to DRGs on pain behavior in naïve rats. Notably, the 50% PWT dose-dependently showed a rapid (1 h) and persistent (14 to 35 days) decrease after the local application of H 2 O 2 (Fig. 3e). Accordingly, the level of ·OH was increased at 4 days after the administration of 60 μg of H 2 O 2 (Fig. S2B). However, Western blotting showed that the same dose of H 2 O 2 had no effect on the protein expression of NOX2 (Fig. S2C).
To explore the cellular mechanism underlying the rapid effect of H 2 O 2 on mechanical allodynia, we next examined whether H 2 O 2 could modulate the excitability of DRG neurons. As shown in Fig. 3f and g, compared to the sham control (193.33 ± 26.79 pA), the 10-min bath application of 500 μM H 2 O 2 decreased the rheobase (105.00 ± 21.87 pA) (p < 0.01, one-way ANOVA followed by Tukey's test) and increased the number of APs in response to the 1× rheobase (5.00 ± 0.73, p = 0.0628, Friedman test followed by Dunn's multiple comparison test). Similar results were observed after the 20min bath application of H 2 O 2 . No significant difference was observed in the membrane capacitance, peak potential, half width of the AP, or input resistance in response to H 2 O 2 perfusion (Fig. S2C). Collectively, these data strongly indicate that ROS generated by NOX2 can rapidly induce mechanical allodynia and increase the excitability of DRG neurons.

Pretreatment with the NOX2-blocking peptide gp91-tat inhibits SNI-induced upregulation of PKCε
The interaction between oxidative stress and PKC is critically important in chronic pain and can amplify the development of central sensitization [28]. We hypothesized that NOX2 and PKC in DRGs interact with each other to impact neuropathic pain. To confirm this hypothesis, we investigated the effect of SNI-induced neuropathic pain on PKC expression in L4-L6 DRGs. However, Western blotting did not show any profound alterations in the protein expression of PKC after SNI (Fig. 4a). Of note, we observed that PKCε was significantly upregulated at 7 days after SNI (Fig. 4b), and this effect was inhibited by pretreatment with the NOX2-blocking peptide gp91-tat 2 h before SNI (from 170.21 ± 20.20% to 99.35 ± 14.92%, p 0.05, one-way ANOVA followed by Tukey's test).
Next, to elucidate the up-and downstream relationship between these factors, we applied εV1-2, a PKCεspecific inhibitor, to L4-L6 DRGs 2 h before SNI. Similar to what was observed in gp91-tat-pretreated SNI rats, blocking PKCε significantly attenuated SNI-induced mechanical allodynia, which persisted to day 14 after SNI (Fig. 4c, at day 14 after SNI operation, 16.34 ± 5.08 g, p < 0.01 compared with the vehicle-treated SNI group, Kruskal-Wallis test followed by Dunn's multiple comparison test). In parallel, we tested the effect of εV1-2 on NOX2 expression. We found that the SNI-induced increase in NOX2 protein expression in DRGs was not affected by pretreatment with εV1-2 (Fig. 4d, from 157.40 ± 12.51% in the SNI group to 184.18 ± 20.84% in the εV1-2-pretreated group, p > 0.05, one-way ANOVA followed by Dunnett's T3 multiple comparisons test), suggesting that PKCε activation is preceded by NOX2 activation. Consistently, the local application of εV1-2 at day 7 after SNI partially decreased the SNI-induced 50% PWT, which persisted for 7 days (Fig. 4e, 9.79 ± 5.61 g at day 8 after SNI, p < 0.01, 8.81 ± 6.71 g at day 14 after SNI, p < 0.01, Mann-Whitney U test). However, local application of the specific PKCε activator ψεRACK to L4-L6 DRGs rapidly induced mechanical allodynia in naïve rats in a dose-dependent manner. As shown in Fig.  4f, 16 μg and 64 μg of ψεRACK reduced the 50% PWT at 2 h, and the effect persisted for 21 days, whereas 4 μg of ψεRACK had no effect on the PWT. To further verify that the activation of PKCε was followed by NOX2 activation, we next examined whether pharmacological blockade of NOX2 could inhibit PKCε-induced mechanical allodynia. However, pretreatment with gp91-tat 2 h before ψεRACK injection did not have a significant effect on the 50% PWT decrease (Fig. 4g).

Pretreatment with gp91-tat prevents the SNI-induced plasma membrane translocation of PKCε in DRG neurons
Double immunofluorescence staining showed that PKCε colocalized with IB4-, CGRP-, and NF-200-labeled neurons ( Fig. 5a(a-c), (d-f), and (g-i), respectively) but not with GFAP-labeled cells (Fig. 5a(j-l)) in the L4-L6 DRGs of sham rats, rats at 7 days after SNI, and gp91-tatpretreated rats at 7 days after SNI. Interestingly, we found that SNI caused the translocation of PKCε from the cytosol to the plasma membrane in L4-L6 DRG neurons, which was significantly attenuated by gp91-tat pretreatment (Fig. 5b(a-f)). To further confirm the membrane translocation of PKCε, we next performed WB on the subcellular fraction and verified that PKCε protein expression was upregulated in both the plasma membrane and cytosol at day 7 after SNI (119.38 ± 2.93% and 139.28 ± 11.89% of sham control, p < 0.01 and p < 0.05, respectively, one-way ANOVA followed by Tukey's test, Fig. 5c, d), and this effect was significantly decreased by gp91-tat pretreatment (101.55 ± 2.50% and 106.19 ± 5.50%, p < 0.01 and p < 0.05, respectively, oneway ANOVA followed by Tukey's test).

H 2 O 2 causes the translocation of PKCε to the plasma membrane
Next, we examined whether the expression and subcellular distribution of PKCε could be regulated by oxidative stress through the administration of H 2 O 2 (Fig. 6a). As shown in Fig. 6b and c, at day 4 after H 2 O 2 treatment, the protein expression of PKCε was significantly upregulated to 139.28 ± 9.34% and 136.65 ± 6.64% in the plasma membrane and cytosol of cultured DRGs, respectively, compared to those of vehicle-treated rats (100.00 ± 4.43% in the membrane; 100.00 ± 7.87% in the cytosol; p < 0.01, two-tailed unpaired Student's t test).
To further confirm the effect of H 2 O 2 on the subcellular distribution of PKCε, we performed immunostaining of PKCε in frozen sections of DRGs obtained from control and H 2 O 2 -treated rats. Interestingly, H 2 O 2 treatment induced membrane translocation of PKCε in DRGs, which began at 1 h and persisted for at least 4 days after H 2 O 2 treatment (Fig. 6d). Furthermore, we investigated the translocation of PKCε in cultured DRG neurons. As shown in Fig. 6e and f, at day 1 after treatment with H 2 O 2 , the average distribution ratio of the membrane to soma was higher than that of the control rats and was maintained at a higher level at day 4 in H 2 O 2 -treated DRGs (from 0.99 ± 0.02 in the control group to 1.62 ± 0.05 in the 4-day H 2 O 2 -treated group, p < 0.001, one-way ANOVA followed by Dunnett's T3 multiple comparisons test). Moreover, the translocation of PKCε was observed in small (< 25 μm), medium (25-35 μm), and large (> 35 μm) DRG neurons (Fig. 6g). Taken together, these results indicated that the direct application of H 2 O 2 may induce the plasma membrane translocation of PKCε in DRG neurons.

Oxidative stress involvement in pain behavior is mediated by the plasma membrane translocation of PKCε
Treatment with the ROS scavenger PBN (100 μg in 2 μl per DRG) and the specific H 2 O 2 catalyst catalase [31] (500 units in 2 μl per DRG) decreased SNI-induced PWT after 1 day (from 0.71 ± 0.07 g in the vehicletreated SNI group to 10.66 ± 3.72 g in the PBN-treated SNI group, p = 0.1513, and 12.09 ± 5.99 in the catalasetreated SNI group, p = 0.0906, Kruskal-Wallis test followed by Dunn's multiple comparison test, Fig. 7a). The Western blot analysis showed that PBN and catalase treatment 1 day after SNI abrogated the SNI-induced increase in PKCε protein in both the plasma membrane and cytosol (Fig. 7b and c). Moreover, treatment with εV1-2 after H 2 O 2 reversed the decrease in PWT from 1-(1.03 ± 0.09 in the H 2 O 2 treatment with vehicle group compared with 12.79 ± 5.99 in the H 2 O 2 treatment with εV1-2 group, p < 0.01, Mann-Whitney U test, Fig. 7d) until the end of the pain behavior test. Taken together, these data suggest that the activation of superoxide and H 2 O 2 is required for the upregulation of PKCε protein in both the plasma membrane and cytosol, which ultimately leads to SNI-induced neuropathic pain.

Discussion
In this study, we proposed a novel mechanism underlying the generation of neuropathic pain. SNI-induced upregulation of NOX2 generates ·O 2 − , H 2 O 2 , and ·OH, which in turn promote the plasma membrane translocation of PKCε in neurons. Inhibition of NOX2 and PKCε or treatment with ROS-scavenging agents can attenuate the hyperexcitability of DRG neurons and mechanical allodynia. The suppression of NOX2 by local pretreatment of DRGs with gp91-tat or NOX2-shRNA effectively reduced SNI-induced mechanical allodynia, which was also attenuated by PBN, catalase, and εV1-2 (Fig. 8). Moreover, the hyperexcitability of DRG neurons and increased levels of 8-OHG, ·OH, and PKCε after SNI were reversed by gp91-tat administration. We further showed that PKCε translocated from the cytosol to the plasma membrane in DRG neurons after SNI, and this effect was blocked by gp91-tat and mimicked by H 2 O 2 administration in naïve animals. Finally, H 2 O 2 -induced mechanical allodynia was suppressed by εV1-2. Therefore, we concluded that PKCε membrane trafficking mediated by NOX2-induced oxidative stress after peripheral nerve injury promoted peripheral nociceptive sensitization.

Oxidative stress is a key link between upstream NOX2 and downstream PKCε in DRGs
Accumulating evidence indicates that NOX2 plays an important role in neuropathic pain by altering the balance of pro-and anti-inflammatory cytokines, such as TNFα, IL-1β, and IL-10 [32,33]. Our previous results showed that NOX2 was required for synaptic plasticity in the SDH and contributes to HFS-induced mirrorimage pain [11]. In the present study, we showed that NOX2 was upregulated in DRGs after SNI. This upregulation was suppressed by the gp91-tat peptide or shRNA treatment before SNI, which attenuated mechanical allodynia, suggesting a critical role for NOX2 in initiating neuropathic pain. It reported that gp91-tat prevented phosphorylation of p47PHOX last to 7 day, treated with gp91-tat had a significant behavior improvement at 14-  (a-c)), CGRP (a(d-f)), and NF-200 (a(g-i)) but not GFAP (a(j-l)) in the sham, SNI, and SNI pretreated with gp91-tat groups (n = 3/group). b Representative staining showing the plasma membrane translocation of PKCε in the SNI group (b(c, d)) but not in the sham (b(a, b)) and gp91-tat-pretreated groups (b(e, f)) presented as the fluorescence intensity of PKCε in the cell. c, d PKCε expression in both the plasma membrane (c) and cytosolic fractions (d) of DRGs from the sham, SNI, and gp91-tatpretreated SNI groups was determined by Western blotting (n = 5/group). One-way ANOVA followed by Tukey's test. *p < 0.05, **p < 0.01 versus the sham group; #p < 0.05, ##p < 0.01 versus the SNI group and 28-day time points after spinal cord injury (SCI), suggesting a long-term effect of gp91-tat in SCI mice [34]. In contrast, gp91-tat failed to alleviate pain if it was administered after SNI; the behavior test also found that application of PBN or catalase had a limited effect on mechanical allodynia induced by SNI. It implies that the activation of NOX2-ROS in DRGs by peripheral nerve injury triggered multiple downstream mechanisms. is illustrated. f Ratios of the PKCε fluorescence intensity in the membrane and total soma of control cells and at 1 day and 4 days after drug application (n = 32-48/group). One-way ANOVA followed by Dunnett's T3 multiple comparisons test. ***p < 0.001 versus the control group. g Quantitative measurements showing the fluorescence intensity ratios of membrane and total soma in small, medium, and large DRGs (n = 12-25/group). One-way ANOVA followed by Dunnett's T3 multiple comparisons test. ***p < 0.001 versus the control group Fig. 7 Oxidative stress induces mechanical allodynia via the plasma membrane translocation of PKCε. a Effects of PBN and catalase on the 50% PWT after drug application. The dotted arrow indicates the time of PBN, catalase, or vehicle application (n = 5-6/group). Kruskal-Wallis test followed by Dunn's multiple comparison test. *p < 0.05, **p < 0.01, ***p < 0.001 versus the sham group. b, c Treatment with PBN or catalase 1 day after SNI inhibited the upregulation of PKCε protein expression in both the plasma membrane (b) and cytosol (c) of DRGs (n = 6/group). One-way ANOVA followed by Tukey's test. **p < 0.01 versus the vehicle-treated sham group; #p < 0.05, ##p < 0.01 versus the vehicle-treated SNI group. d εV1-2 reversed H 2 O 2 -induced mechanical allodynia (n = 6/group). Mann-Whitney U test. **p < 0.01 versus the vehicle-treated H 2 O 2 group Fig. 8 The mechanism of NOX2-ROS-PKCε signaling pathway contributes to peripheral sensitization and mechanical allodynia These results indicate that NOX2 is a good target for neuropathic pain prevention. It has been reported that NOX2 in DRG macrophages contributes to neuropathic pain by releasing proinflammatory cytokines to stimulate sensory neurons [12]. Additionally, it has been shown that NOX2 is expressed in the soma of trigeminal ganglion neurons [4]. In this study, we found that in rat DRGs, NOX2 was mainly expressed in neurons, with less expression in macrophages and no expression in satellite glial cells. These results indicated that NOX2 induced oxidative stress regulated neuro-excitability in DRG through direct neuronal interactions and the indirectly interactions of macrophage and neuron. As expected, the blockade of NOX2 deceased the hyperexcitability of DRG neurons. Therefore, we concluded that NOX2 promoted neuropathic pain by modulating neural excitability. However, we could not exclude the role of the macrophage-neuronal NOX2 signaling in SNI.
NOX2 is an important source of both extracellular and intracellular ROS [35]. Increase ROS induce pain by reducing GABA inhibitory transmission in substantia gelatinosa neurons of the SDH [2]. In DRG neurons, the sensitization of transient receptor potential ankyrin 1 to H 2 O 2 leads to cold sensitivity [36]. Our data showed that SNI increased 8-OHG and ·OH levels in DRGs, and this effect was inhibited by the NOX2-blocking peptide. Nonnuclear 8-OHG immunostaining, which indicates the exposure of mitochondrial DNA to excessive oxidative stress [30], was also inhibited by pretreatment with gp91-tat before SNI. Moreover, we found that the application of H 2 O 2 to DRGs dose-dependently induced mechanical allodynia and hyperexcitability in DRG neurons. These results suggest that ROS and H 2 O 2 activation is an important mediator of NOX2 signaling in DRGs in the context of neuropathic pain.
It was reported that NOX2-derived ROS in the spinal cord could modulate PKC activity through sigma-1 receptors [18]. In addition, PKC is a regulator of NOX activity, as it induces the phosphorylation of NOX subunits such as gp91phox [37]. In this study, we found that gp91-tat inhibited the upregulation of PKCε but not PKC in DRGs after SNI. As PKCε is a key mediator of pain in DRGs [38], we next focused on the interaction of NOX2 and PKCε in this study. Our data showed that the PKCε inhibitor εV1-2 failed to affect the protein level of NOX2 and that mechanical allodynia induced by the PKCε activator ψεRACK was not attenuated by gp91-tat. Furthermore, εV1-2 treatment was sufficient to alleviate SNI-induced mechanical allodynia. These findings indicate that the upregulation of NOX2 after SNI is an important upstream activator of PKCε that ultimately contributes to the development of neuropathic pain in DRGs. ROS may be a key link between upstream NOX2 and downstream PKCε.

Oxidative stress induced by NOX2 enhances the translocation of PKCam PKCεportant upstream activator of bitor gp91
Rather than the upregulation of PKCε, the translocation of PKC from the cytosolic (soluble) fraction to the plasma membrane is a signal of its activation [39]. The translocation of PKC isoforms has been reported in many neural diseases, including postmortem brain samples of patients with bipolar disorder [40] and the DRGs of rats with paclitaxel-induced peripheral neuropathy [16]. Consistently, we found plasma membrane translocation of PKCε in DRG neurons and upregulated protein expression in the plasma membrane after SNI. However, the functions of PKCε in the plasma membrane of DRG neurons are still unknown. Wu et al. showed that ψεRACK enhanced the Nav1.8 current and induced mechanical hyperalgesia [13]. In contrast, PKCε-I, a selective PKCε-anchoring inhibitor, prevented the reduction in the peak Na + current in the mouse hippocampus [41]. In addition, TRPV1 currents in DRG neurons could also be reduced by a PKCε translocation inhibitor (PKCε TIP) in the recording pipette or pretreatment with the PKC inhibitor BIM [17]. Furthermore, it showed that motor protein kinesins, such as Kinesin superfamily proteins (KIFs), played an important role in neuronal cargo trafficking [42]. The KIF3 could motor transports atypical protein kinase C (aPKC) to the tip of nascent axons [43]. The effect of motor protein KIFs might be a trafficking mechanism that underling the plasma membrane translocation of PKCε induced by ROS. These reports raise the possibility that the plasma membrane translocation of PKCε allows it to be anchored to specific subcellular sites, which enables inhibitors and activators to be more effective. This possibility was further validated in our study by the intracellular application of a PKCε inhibitor or activator. As expected, εV1-2 decreased the excitability of DRG neurons in SNI rats, whereas ψεRACK rapidly increased the excitability of DRG neurons in naïve rats within minutes. However, whether the mechanism underlying the rapid regulation of DRG excitability by PKCε translocation occurs via interactions with other channels, such as sodium channels or TRPV1, in the plasma membrane needs to be further studied.
PKCε translocation in DRG neurons is affected by a series of factors. It was reported that ROS derived from NOX1 accelerated PKCε translocation by modulating the redox state of cysteine residues in DRG neurons and enhanced inflammatory pain [5]. Moreover, TGF-β which could be activated by redox imbalance [44] also induced the translocation of PKCε in DRG neurons [17]. In addition, bradykinin (BK), a peptide that enhances the membrane ionic current in nociceptive neurons, induces PKCε translocation to the cell membrane when it is activated by heat stimulation [45]. Furthermore, BK, activin A, and phorbol-12-myristate-13-acetate (a potent PKC activator) were reported to be involved in the translocation of PKCε, followed by TRPV1 channel sensitization and thermal hyperalgesia generation [46]. Notably, BK may induce NOX to generate ROS [47], and activin A is also related to ROS production [48]. These reports suggest that ROS may be an important source of PKCε cargo to the plasma membrane. In this study, we demonstrated that the plasma membrane translocation of PKCε and the upregulation of PKCε protein levels in both the plasma membrane and cytosol after SNI were blocked by gp91-tat pretreatment in DRGs. Similar results were observed after the local application of H 2 O 2 to DRGs. To determine the unique roles of ROS and H 2 O 2 , PBN and catalase were used to scavenge ROS and catalyze H 2 O 2 , respectively. We found that both treatments decreased the upregulation of PKCε protein expression in the plasma membrane and cytosol. Most importantly, εV1-2 attenuated H 2 O 2induced mechanical allodynia. Collectively, the generation of ROS and H 2 O 2 by NOX2 after SNI is a key step in the activation of PKCε, which ultimately leads to neuropathic pain.

Conclusion
Our study demonstrated that NOX2-induced oxidative stress in DRG neurons is an important upstream factor of PKCε in the context of SNI conditions. In particular, oxidative stress induces DRG neuron hyperexcitability and the plasma membrane translocation of PKCε, ultimately contributing to peripheral sensitization and pain sensitivity in rats.