Bardoxolone methyl Ameliorates Chemotherapy-induced Neuropathic Pain by Activation of Phosphorylated Nuclear Factor Erythroid 2-Related Factor 2 in the Dorsal Root Ganglia

Abstract

BACKGROUND:

Many chemotherapeutic drugs, including paclitaxel, produce neuropathic pain in cancer patients, which is a dose-dependent adverse effect. Such chemotherapy-induced neuropathic pain (CINP) is difficult to treat with existing drugs. Nuclear factor erythroid 2-related factor 2 (Nrf2) is a major regulator of anti-oxidative responses and activates to phosphorylated Nrf2 (pNrf2). We determined analgesic effects of bardoxolone methyl (BM), a Nrf2 activator, and the role of pNrf2 on CINP.

METHODS:

CINP was induced in rats by intraperitoneally injecting paclitaxel on 4 alternate days in rats. BM was injected systemically as single or repeated injections after pain fully developed. RNA transcriptome, mechanical hyperalgesia, levels of inflammatory mediators and pNrf2, and location of pNrf2 in the dorsal root ganglia (DRG) were measured by RNA sequencing, von Frey filaments, Western blotting, and immunohistochemistry in rats and human DRG samples. In addition, the mitochondrial functions in 50B11 DRG neuronal cells were measured by fluorescence assay.

RESULTS:

Our RNA transcriptome of CINP rats showed a downregulated Nrf2 pathway in the pain condition. Importantly, single and repeated systemic injections of BM ameliorated CINP. Paclitaxel increased inflammatory mediators, but BM decreased them and increased pNrf2 in the DRG. In addition, paclitaxel decreased mitochondrial membrane potential and increased mitochondrial volume in 50B11 cells, but BM restored them. Furthermore, pNrf2 was expressed in neurons and satellite cells in rat and human DRG.

CONCLUSIONS:

Our results demonstrate the analgesic effects of BM by Nrf2 activation and fundamental role of pNrf2 on CINP, suggesting a target for CINP and therapeutic strategy for patients.

INTRODUCTION

Many chemotherapeutic drugs, including taxanes (e.g., paclitaxel), vinca alkaloids (e.g., vincristine), and platinum agents (e.g., cisplatin), produce peripheral neuropathy—especially pain—in the distal extremities of cancer patients and survivors; such pain is called chemotherapy-induced neuropathic pain (CINP).¹ For example, paclitaxel, a chemotherapy for breast, cervical, ovarian, and lung cancers, can induce numbness, tingling, and burning pain in hands and feet.² CINP is a dose-limiting adverse effect and occurs during chemotherapy and even after its termination.³ Present analgesic drugs such as opioids, nonsteroidal anti-inflammatory agents, sodium channel blockers, anticonvulsants, and antidepressants show little or no analgesic effect against CINP.³

We reported that systemic injection of reactive oxygen species (ROS) scavengers such as phenyl N-t-butylnitrone, tempol, and vitamin E reduced pain in animal models of paclitaxel-induced neuropathic pain (PINP) and spinal nerve ligation pain, which means ROS is a major regulator of neuropathic pain.^4–7 The expression of antioxidant proteins is regulated by the transcription factor nuclear factor erythroid 2-related factor 2 (Nrf2).⁸ In the Nrf2 pathway, phosphorylated Nrf2 (pNrf2) at serine 40 and heme oxygenase-1 (HO-1) are major regulators, ⁹ as is the Nrf2 activator bardoxolone methyl (BM), which has been used to treat chronic obstructive pulmonary disease and various malignancies such as lung, liver, colon, breast, and ovarian cancers.⁸ In previous studies, Nrf2 agonists have demonstrated analgesic effects in several neuropathic pain models, including models of chronic constriction and diabetic neuropathic pain.^10–12 However, the role of pNrf2 and analgesic effect of BM in CINP and related mechanisms have not been studied. Thus, the purposes of our study were to 1) determine the analgesic effects of BM on PINP in rats, 2) identify the role of BM and pNrf2 in the dorsal root ganglia (DRG) and 50B11 DRG neuronal cell line, and 3) find the location of pNrf2 in 50B11, rat, and human DRG.

METHODS

Animals

Experiments were approved by the Institutional Animal Care and Use Committee of The University of Texas MD Anderson Cancer Center. Male adult Sprague-Dawley rats (200-350 g, Harlan Sprague-Dawley Company, Houston, TX) were housed with food and water.

PINP model

Rats were injected intraperitoneally with paclitaxel (GenDepot; 2 mg/kg) or vehicle (4% DMSO and 4% Tween 80 in sterile saline) on days 0, 2, 4, and 6.¹³

RNA sequencing

The L1-6 DRG were collected from vehicle- and paclitaxel-injected rats on day 20 after the first vehicle or paclitaxel injection. Briefly, The RNA samples were extracted from the rats using an RNeasy Plus Universal Mini Kit (Qiagen, Germany).¹⁴ RNA sequencing was performed by Novogene (US) using the Illumina HiSeq platform.

The raw reads of the RNA sequencing were aligned to the Rattus norvegicus Genome Sequencing Consortium Rnor_6.0 (GCA_000001895.4) reference genome using Tophat2 (version 2.1.1). After alignment, the gene expressions were quantified into counts using HTseq-Count and transformed to log₂ counts per million for visualization purpose using edgeR package in R (version 3.6.0). Differentially expressed genes were identified by fitting gene-wise negative binomial generalized log-linear models, and p values were adjusted using the false discovery rate (FDR). Preranked gene set enrichment analysis (GSEA) was performed using GSEA software (version 4.0.3) with Molecular Signatures Database (MSigDB version 7.4).

Mechanical hyperalgesia

The mechanical threshold of the left hind paw was measured with von Frey filaments (0.45-14.45 g) using the up-down technique under blind conditions, and then a 50% mechanical threshold was calculated.^15–17

Sedation

Sedation of rats was measured using 5-point scales of posture (0 = normal, 4 = flaccid atonia) and righting reflexes (0 = struggles, 4 = no movement) and assessed after the mechanical hyperalgesia.¹⁵

Single or repeated intraperitoneal injections of BM

BM (Selleck Chemicals, Houston, TX) was dissolved in vehicle (5% DMSO and 10% Tween 80 in phosphate-buffered saline without calcium and magnesium) and intraperitoneally injected into rats after PINP had fully developed. To determine the single and extended analgesic effects of BM, we administered as a single intraperitoneal injection (1, 3, 10 mg/kg) on day 21 and repeated injections (10 mg/kg) twice daily starting on day 21 for 4 days. The mechanical threshold was measured before and after injection.

Western blot analysis

On day 22 after the first injection of vehicle or paclitaxel, we collected lumbar DRG from vehicle-injected rats or paclitaxel-injected rats.¹⁵ For the BM-treated paclitaxel-injected rats, BM (10 mg/kg) was injected intraperitoneally twice on day 21 and in the morning on day 22. One hour after the last injection of BM on day 22, the lumbar DRG were collected. The DRG were homogenized and centrifuged, and the supernatants were loaded and transferred to membranes. The blots were incubated with primary antibodies against Nrf2 (#ab31163, Abcam), pNrf2 (#ab76026, Abcam), HO-1 (#sc-136960, Santa Cruz Biotechnology), phosphorylated nuclear factor kappa B (pNFκB; #3033, Cell Signaling Technology, Danvers, MA), monocyte chemoattractant protein 1 (MCP-1; #sc-28879, Santa Cruz Biotechnology), and glyceraldehydes-3-phosphate dehydrogenase (GAPDH; #sc25778, Santa Cruz Biotechnology) overnight at 4°C. The blots then were incubated with anti-rabbit IgG-HRP (#W3902, GenDepot) or anti-mouse IgG-HRP (#W3903, GenDepot). The blots were analyzed with a chemiluminescence detection and normalized to GAPDH.

50B11 DRG neuronal cell line culture

The 50B11 DRG cells were kindly provided by Dr. Ahmet Höke (Johns Hopkins University, Baltimore, MD) and were cultured in Neurobasal medium (NBM) with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 1 X B-27 supplement, and 0.2% glucose in a CO₂ incubator (5% CO₂, 37°C).¹⁸

To measure the mitochondrial membrane potential (MMP) and mitochondrial volume, the cells were cultured in NBM with 5% FBS at 1 × 10⁴ cells per well in a 96-well plate for 1 day, treated with BM (0.3, 1.0 μM) or vehicle (0.1% dimethyl sulfoxide [DMSO] in NBM with 5% FBS) for 4 hours, and then changed to media with paclitaxel (1, 10, 20 μM) in NBM with 5% FBS or vehicle (0.2% DMSO in NBM with 5% FBS) for 24 hours. To measure MMP, cells were treated with 5 μM JC-1 for 1 hour, after which we measured the fluorescence at excitation at 475 nm and emission at 590 nm and 530 nm, which we calculated as a ratio of 590 nm to 530 nm.¹⁹ To measure mitochondrial volume, cells were treated with 200 nM MitoTracker Green FM probe for 30 minutes and measured the fluorescence at excitation at 490 nm and emission at 516 nm.²⁰

To find the location of pNrf2 in 50B11 cells by immunocytochemistry, 50B11 cells were plated in a chambered cover glass (8 well, Thermo Fisher Scientific, Waltham, MA) in NBM with 5% FBS at 5 × 10⁴ cells per well for 1 day, treated with BM (1.0, 3.0 μM) or vehicle for 4 hours, and then changed to media with paclitaxel (10 μM) or vehicle for 24 hours.^4,19 The cells were fixed and then incubated with the primary antibodies, including anti-pNrf2 antibody (#ab76026, Abcam, San Francisco, CA) and anti-calcitonin gene-related peptide (CGRP; #sc-57053, Santa Cruz Biotechnology, Dallas, TX), and then secondary antibodies conjugated with Alexa Fluor 568 (#SA101-015, GenDepot, Katy, TX) and Alexa Fluor 488 (#SA802, GenDepot). In addition, the sections were applied with ProLong Diamond antifade mountant (Thermo Fisher Scientific) with 4′,6-diamidino-2-phenylindole (DAPI). The cells were viewed using a CELENAR S digital cell imaging system (Logos Biosystems, Annandale, VA). The intensity of pNrf2 was analyzed with image J software (1.48 version, National Institutes of Health, Bethesda, MD, USA).

Immunohistochemical analysis of rat DRG

Immunohistochemical analysis was performed to find the localization of pNrf2 in the L5 DRG from vehicle- and paclitaxel-injected rats.²¹ Briefly, the L5 DRG were collected, frozen, cryosectioned, and mounted on microscope slides. The sections were incubated with the primary antibodies, including anti-pNrf2, anti-NeuN (#ab104224, Abcam), anti-CGRP, and anti-GFAP (#sc-51908, Santa Cruz Biotechnology), and then secondary antibodies conjugated with Alexa Fluor 568 and Alexa Fluor 488.

Immunohistochemical analyses of human DRG

We collected DRG from one patient donor at MD Anderson Cancer Center who provided written consent. This protocol was approved by MD Anderson Cancer Center. This patient had a history of cancer-related neuropathic pain including chemotherapy-induced neuropathic pain, neural compression, and physiological effects of the peritumoral environment with radiation therapy involving the spine. During standard-of-care resection of metastatic disease involving ligation of spinal nerve roots, this patient donated painful DRG. Immediately after each ganglion in pain area was excised in the operating room, ganglia were obtained, fixed, cryoprotected, and mounted. The three sections were staining with the primary antibodies, including anti-pNrf2, anti-NeuN, anti-CGRP, and anti-GFAP antibodies, and then secondary antibodies.¹³ Sections were viewed with a fluorescent microscope. To measure the cell size, each neuron was highlighted with color. The total numbers of pNrf2 positive neurons were counted in the three sections of all cells.

Statistical analysis

Data were expressed as 1) median with interquartile range for the MMP and mitochondrial volume because of their skewed distribution, 2) means with standard errors of the means for the behavioral tests because of their large variability of continuous value. Data analyses were performed using the GraphPad Prism 8.0 (GraphPad Software, La Jolla, CA). We performed 1) the Kruskal-Wallis test for the MMP and mitochondrial volume, 2) a 2-way repeated measures analysis of variance (ANOVA) with repeated measures on rats at various time points followed by the Tukey post hoc test for the behavioral test, 3) the Kruskal-Wallis test for the Western blotting and immunocytochemistry analyses, and 4) an ordinary 1-way ANOVA followed by the Tukey post hoc test for the intensity of pNrf2 in the 50B11 cell. p values less than 0.05 were considered statistically significant. This study was focus on pain behavior in CINP rats. For the RNAseq analysis, given the small sample size per group, the results are more exploratory. We may do a follow-up study later for a bigger population to verify our findings. In addition, the animal number of each group was minimized for the pain study and then sample size in the present study was N=6 per group in the level of 0.05 and a power of 0.8.

RESULTS

RNA sequencing analysis of paclitaxel-injected rat DRG

In the RNA sequencing experiments, paclitaxel upregulated 1,244 genes and downregulated 1,251 genes among 14,522 genes (Figure 1A–C; false discovery rate ≤ 0.05). GSEA showed upregulation of genes of inflammatory response pathway and downregulation of genes of the Nrf2 pathway (Figure 1D, e; p ≤ 0.05). The data suggest that paclitaxel downregulated genes of the Nrf2 pathway.

Figure 1.

Transcriptomic profiling of rat dorsal root ganglia (DRG). Paclitaxel (PAC, 2 mg/kg) or vehicle (VEH, 4% dimethyl sulfoxide and 4% Tween 80 in saline) was injected intraperitoneally on days 0, 2, 4, and 6 in 2 groups of rats. (A) Heat map view of significant DEGs (FDR≤0.05 and fold change (FC) >2) at the intersection of genes in the DRG of VEH and PAC. (B) Of the 14,522 genes identified, PAC upregulated 1,244 genes and downregulated 1,251 genes, with statistical significance (FDR ≤ 0.05). (C) Volcano plot for all genes. (D, E) Enrichment plot of the inflammatory response pathway and Nrf2 pathway.

Analgesic effects of BM

All rats treated with BM had scores of 0 on posture and righting reflex, indicating that BM yielded no sedative effects. A single intraperitoneal injection of BM (3 or 10 mg/kg) increased the mechanical threshold in a dose-dependent manner (Figure 2A). The 10 mg/kg dose increased the mechanical threshold from 0.76 g at before injection to 11.49 g and 1.68 g at 1.5 and 4 hours (Figure 2A). Detail, the 10 mg/kg dose significantly increased the mechanical threshold to 5.79 g (difference of mean 0.85; 95% CI, 0.56~1.15; p < 0.0005), 7.69 g (difference of mean 1.02; 95% CI, 0.72~1.31; p < 0.0001), 11.49 g (difference of mean 1.14; 95% CI, 0.84~1.43; p < 0.0008), 5.64 g (difference of mean 0.82; 95% CI, 0.52~1.11; p < 0.0002), 2.58 g (difference of mean 0.49; 95% CI, 0.19~0.78; p ~ 0.0001) at 0.5, 1, 1.5, 2, and 3 hours compared to the vehicle injected group by using a 2-way repeated measures analysis of variance (ANOVA) with 1 repeated factor (time) followed by the Tukey post hoc test. Repeated injections of 10 mg/kg BM significantly increased the mechanical threshold from 1.34 g on day 21 (before the first injection) to 8.47 g on days 23 (Figure 2B). Compared to the vehicle group, repeated injections of 10 mg/kg twice a day for 4 days significantly increased the mechanical threshold to 5.15 g (difference of mean −0.58; 95% CI, −0.82 ~ −0.36; p < 0.00006), 8.47 g (difference of mean −0.89; 95% CI, −1.13 ~ −0.67; p < 0.00001), 6.87 g (difference of mean −0.71; 95% CI, −0.94 ~ −0.47; p < 0.00002), 7.94 g (difference of mean −0.88; 95% CI, −1.11 ~ −0.64; p < 0.00001) on days 22, 23, 24, and 25. These findings suggest that treatment with BM may result in analgesia but not sedation.

Figure 2.

Analgesic effects of single and repeated intraperitoneal injections of bardoxolone methyl (BM) on paclitaxel (PAC)-induced neuropathic pain in rats. (A) On day 21 after PAC injection, rats received a single intraperitoneal injection of vehicle or 1, 3, or 10 mg/kg BM (arrow). The mechanical thresholds of the rats that received 10 mg/kg BM remained significantly higher than those of the vehicle-injected rats for 4 hours. (B) Starting on day 21 after the first PAC injection, rats received twice-daily intraperitoneal injections of vehicle or 10 mg/kg BM (arrow) for 4 days (hatched box). On days 22 to 25, the mechanical thresholds of the rats that received repeated intraperitoneal injections of bardoxolone methyl were significantly higher than those of the vehicle-injected rats. Data are means with standard errors; asterisks indicate significant differences (p < 0.05) compared with the vehicle group as determined by a 2-way repeated measures analysis of variance with 1 repeated factor (time) followed by the Tukey post hoc test. (C, D) PAC increases and BM subsequently decreases levels of phosphorylated nuclear factor kappa B (pNFκB) and monocyte chemoattractant protein 1 (MCP-1) in rat dorsal root ganglia (DRG). (E) BM does not change total nuclear factor erythroid 2-related factor 2 (Nrf2) in rat DRG. (F, G) BM subsequently increases levels of phosphorylated Nrf2 (pNrf2) and heme oxygenase-1 (HO-1) in rat DRG. Data are the means with standard deviations for 3 rats; asterisks indicate significant differences (p < 0.05) compared with the vehicle (VEH)-treated group as determined by the Mann-Whitney U test.

BM decreased paclitaxel-induced inflammatory mediators by increasing pNrf2 and HO-1 in the DRG

The protein levels of pNFκB and MCP-1 in paclitaxel-treated rats were 1.8 times (p = 0.022, Kruskal-Wallis test) and 1.7 times (p = 0.021, Kruskal-Wallis test) higher, respectively, than those in vehicle-treated rats on Western blot (Figure 2C, D). However, paclitaxel did not significantly change levels of Nrf2 (adjusted p > 0.99, Kruskal-Wallis test), pNrf2 (adjusted p = 0.54, Kruskal-Wallis test), and HO-1 (adjusted p = 0.41, Kruskal-Wallis test) in DRG than those in vehicle-treated rats (Figure 2E–G). Subsequent treatment with BM significantly decreased levels of pNFκB and MCP-1 in DRG compared with those of paclitaxel-treated rats that did not receive BM (Figure 2C, D). Notably, subsequent treatment with BM significantly increased levels of pNrf2 (1.5 times, p = 0.034, Kruskal-Wallis test) and HO-1 (2.4 times, p = 0.018, Kruskal-Wallis test) in DRG compared with those of paclitaxel-treated rats that did not receive BM (Figure 2F, G). These data indicate that BM decreased paclitaxel-induced levels of pNFκB and MCP-1 by increasing pNrf2 and HO-1.

BM decreased paclitaxel-induced mitochondrial damage in 50B11 cells

Paclitaxel (1, 10, 20 μM) significantly decreased MMP, but BM (0.3, 1.0 μM) did not decrease MMP in 50B11 cells (Figure 3A–C). The 20 μM of paclitaxel significantly decreased MMP to 94% compared to vehicle (adjusted p = 0.006, Kruskal-Wallis test). In addition, paclitaxel increased mitochondrial volume, but BM (1.0, 3.0 μM) did not reduce mitochondrial volume (Figure 3D–F). Paclitaxel at 20 μM significantly increased mitochondrial volume to 110% compared to vehicle (adjusted p = 0.013, Kruskal-Wallis test). These data indicate that paclitaxel induces mitochondrial damage in the 50B11 cells and BM subsequently reverses the mitochondrial damage.

Figure 3.

The change of mitochondrial membrane potential (MMP) and mitochondrial volume after treatment with bardoxolone methyl (BM) followed by paclitaxel in 50B11 cells. (A) Paclitaxel decreases MMP in cells treated with vehicle of BM only. (B, C) BM reverses MMP. (D) Paclitaxel increases the mitochondrial volume in cells treated with vehicle only. (E, F) BM restores mitochondrial volume. BM was applied for 4 hours prior to paclitaxel application for 24 hours. Data were normalized to MMP (A-C) or mitochondrial volume (D-F) of cells treated with vehicle and paclitaxel. Data are the median with interquartile range for 3 independent experiments; asterisks indicate significant differences (p < 0.05) compared with the vehicle group as determined by the Kruskal Wallis test with post-hoc Dunn multiple comparisons test.

BM increased paclitaxel-decreased pNrf2 intensity in the 50B11 cells

Paclitaxel (10 μM) significantly decreased pNrf2 intensity (difference of mean −87.97; 95% CI, −105.7 ~ −70.20; p < 0.0001) in 50B11 cells (Figure 4A–I). However, BM (0.3, 1.0 μM) significantly increased pNrf2 intensity in 50B11 cells treated with vehicle (Figure 4C, D, I) or paclitaxel (Figure 4G, H, I). In detail, the 1.0 μM of BM significantly increased pNrf2 intensity in paclitaxel-treated 50B11 cells (difference of mean −18.4; 95% CI, −36.2 ~ −0.66; p = 0.035). These data indicate that paclitaxel decreased the pNrf2 intensity in 50B11 cells, and BM increased it.

Figure 4.

Expression of pNrf2 (green, Alexa Fluor 488) in 50B11 cells treated with vehicle of paclitaxel (PAC) or PAC 10 μM. (A) Cells without treatment. (B) Cells treated with vehicle of bardoxolone methyl (BM) followed by vehicle of PAC. (C) Cells treated with BM 0.3 μM followed by vehicle of PAC. (D) Cells treated with BM 1.0 μM followed by vehicle of PAC. (E) Cells treated with PAC 10 μM without BM. (F) Cells treated with vehicle of BM followed by PAC 10 μM. (G) Cells treated with BM 0.3 μM followed by PAC 10 μM. (H) Cells treated with BM 1.0 μM followed by PAC 10 μM. (I) The bar graph shows the intensity of pNrf2 in panels A-H. Scale bars, 50 μm. N, normal; V, vehicle. Data are the means with standard deviations; asterisks indicate significant differences (p < 0.05) compared within the PAC 10 μM group or PAC vehicle group as determined by 1-way ANOVA followed by the Tukey post hoc test.

Co-localization of pNrf2 in the 50B11 cells

Immunocytochemical analysis revealed that pNrf2 was expressed in 50B11 cells treated with vehicle (Figure 5A–D) or paclitaxel (Figure 5E–H). Furthermore, the expression of pNrf2 was localized with the CGRP- or DAPI-expressing cells (Figure 5A–P). These data indicate that pNrf2 was co-localized with CGRP-expressing peptidergic sensory cells.

Figure 5.

Co-localization of phosphorylated nuclear factor erythroid 2-related factor 2 (pNrf2, green), CGRP (red), and DAPI (blue) in 50B11 cells treated with paclitaxel (PAC) 10 μM for 24 hours. (A) pNrf2 without bardoxolone methyl (BM)/PAC. (B) CGRP without BM/PAC. (C) DAPI without BM/PAC. (D) pNrf2, CGRP, and DAPI without BM/PAC. (E) pNrf2 with vehicle of BM and PAC 10 μM. (F) CGRP with vehicle of BM and PAC. (G) DAPI with vehicle of BM and PAC. (H) pNrf2, CGRP, and DAPI with vehicle of BM and PAC. (I) pNrf2 with BM 0.3 μM and PAC 10 μM. (J) CGRP with BM 0.3 μM and PAC. (K) DAPI with BM 0.3 μM and PAC. (L) pNrf2, CGRP, and DAPI with BM 0.3 μM and PAC. (M) pNrf2 with BM 1.0 μM and PAC 10 μM. (N) CGRP with BM 1.0 μM and PAC. (O) DAPI with BM 1.0 μM and PAC. (P) pNrf2, CGRP, and DAPI with BM 1.0 μM and PAC. Arrows indicate CGRP-expressing neurons with pNrf2 and DAPI. Scale bars, 50 μm.

Co-localization of pNrf2 in rat DRG

Immunohistochemical analysis revealed that pNrf2 was expressed in the L5 DRG from both vehicle- (Figure 6A–D) and paclitaxel-injected rats (Figure 6E–H). The expression of pNrf2 was co-localized with NeuN-positive neurons (Figure 6E–H), CGRP-expressing neurons (Figure 6I–L), and GFAP-expressing satellite cells (Figure 6M–P) of DRG from the paclitaxel group. pNrf2 was expressed in the nuclei and cytosol of neurons, CGRP-expressing neurons, and satellite cells (Figure 6H, L, P). The findings show that pNrf2 was co-localized in rat DRG neurons, CGRP-expressing cells, and satellite cells.

Figure 6.

Co-localization of phosphorylated nuclear factor erythroid 2-related factor 2 (pNrf2), NeuN, CGRP, GFAP, and DAPI in rat dorsal root ganglia (DRG). (A) pNrf2 (green, Alexa Fluor 488) in the L5 DRG of a vehicle (VEH)-injected rat. (B) NeuN (red, Alexa Fluor 594) in the L5 DRG of a VEH-injected rat. (C) DAPI (blue) in the L5 DRG of a VEH-injected rat. (D) pNrf2 (green, Alexa Fluor 488), NeuN (red, Alexa Fluor 594), and DAPI (blue) in the L5 DRG of a VEH-injected rat. Arrows indicate NeuN-expressing neurons with pNrf2 (insert, scale bar = 10 μm). Scale bar = 100 μm. (E) pNrf2 (green, Alexa Fluor 488) in the L5 DRG of a paclitaxel (PAC)-injected rat. (F) NeuN (red, Alexa Fluor 594) in the L5 DRG of a PAC-injected rat. (G) DAPI (blue) in the L5 DRG of a PAC-injected rat. (H) pNrf2 (green, Alexa Fluor 488), NeuN (red, Alexa Fluor 594), and DAPI (blue) in the L5 DRG of a PAC-injected rat. Arrows indicate neurons with pNrf2. (I) pNrf2 (green, Alexa Fluor 488) in the L5 DRG of a PAC-injected rat. (J) CGRP (red, Alexa Fluor 594) in the L5 DRG of a PAC-injected rat. (K) DAPI (blue) in the L5 DRG of a PAC-injected rat. (L) pNrf2 (green, Alexa Fluor 488), CGRP (red, Alexa Fluor 594), and DAPI (blue) in the L5 DRG of a PAC-injected rat. Arrows indicate CGRP-expressing neurons with pNrf2. (M) pNrf2 (green, Alexa Fluor 488) in the L5 DRG of a PAC-injected rat. (N) GFAP (red, Alexa Fluor 594) in the L5 DRG of a PAC-injected rat. (O) DAPI (blue) in the L5 DRG of a PAC-injected rat. (P) pNrf2 (green, Alexa Fluor 488), GFAP (red, Alexa Fluor 594), and DAPI (blue) in the L5 DRG of a PAC-injected rat. Arrows indicate GFAP-expressing satellite cells with pNrf2.

Co-localization of pNrf2 in human DRG

Immunohistochemistry revealed that the pNrf2 in the pain area was expressed in the neurons (Supplemental Figure 1A–D), CGRP-expressing neurons (Supplemental Figure 1E–H), and GFAP-expressing satellite cells (Supplemental Figure 1I–L). pNrf2 was expressed in the nuclei and cytosol of cells (Supplemental Figure 1D, H, L). pNrf2-positive neurons were found across all sizes of neurons, including approximately 40% in small-diameter (< 15 μm) DRG neurons (Supplemental Figure 1M). Neurons positive for both pNrf2 and CGRP were also found in small-diameter DRG neurons (Supplemental Figure 1N). These data indicate that pNrf2 was co-localized in human DRG neurons, CGRP-expressing neurons, and satellite cells.

DISCUSSION

Our results indicate that systemic injections of BM produce an analgesic effect on PINP without sedation by inhibiting pNFκB and MCP-1 and increasing pNrf2 and HO-1 in DRG. In addition, BM increased pNrf2 expression and restored damaged mitochondria in 50B11 DRG neuronal cells. In detail, the expression of pNRf2 was located in the nuclei and cytosol in neurons, CGRP-expressing neurons, and satellite cells in rat and human painful DRG. Taken together, these results suggest that activation of pNrf2 by BM produces analgesic effects in CINP by inhibiting mitochondrial damage, pNFκB, and MCP-1 in the DRG.

Nrf2, the product of the NFE2L2 gene, contains 6 highly conserved domains and is degraded by keap1.²² Previous studies reported that neuropathic pain in models of chronic constriction injury, CINP, and spared nerve injury and in spinal cord injury and diabetic neuropathic pain was ameliorated by various compounds, including oltipraz, dimethyl fumarate, and sulforaphane.^10–12 These compounds activate Nrf2 through electrophilic modification of Keap1-Cys-151.¹⁰ In the present study, we demonstrate the important role of pNrf2 in PINP. We recently found that paclitaxel increases ROS and protein kinase A in the DRG.^4,23 Protein kinase C can phosphorylate Nrf2 at serine 40 in cytosol, promote its translocation into the nucleus, and then bind to antioxidant response element (ARE) in the promoter regions of target genes, including HO-1, which leads to production of the HO-1 protein.^24,25 We previously reported that neuropathic pain-producing chemotherapeutic drugs, including paclitaxel, increased pro-inflammatory mediators such as pNFκB and MCP-1, which are involved in CINP.^4,15,21,23 These mediators may be inhibited by HO-1. In addition, BM can bind to Cys-179 of IκB kinase and inhibit the activation of NFκB, thereby inhibiting the downstream pro-inflammatory pathways.²⁶ In the present study, chemotherapy including paclitaxel produces neuropathic pain by increasing inflammatory mediators including pNFκB and MCP-1 in the DRG. The pNrf2 was produced from Nrf2 in the cytosol and mitochondria of neurons, including CGRP-expressing neurons, and satellite cells in the DRG and then translocated into the nucleus. BM produces an analgesic effect in chemotherapy-induced neuropathic pain by inhibiting inflammatory mediators by inducing pNrf2 and HO-1 and restoring mitochondrial functions in the DRG (Supplementary Figure 1).

Nrf2 has various target genes involved in cellular processes such as redox regulation, mitochondrial function, and DNA repair.⁸ In this study, BM decreased the level of pNFκB but increased the level of pNrf2. Nrf2 can crosstalk with other signaling pathways such as the NFκB, cellular tumor antigen p53, AMP-activated protein kinase, and mTOR pathways.⁸ NFκB can suppress transcription of ARE as NFκB components Nrf2 and p65/Rela require CREB-binding protein to transcribe their target genes.²⁷ Thus, binding of p65 to CREB-binding protein prevents Nrf2-driven transcription of ARE. In addition, several Nrf2 target genes were modified by NFκB.²⁸ Therefore, the Nrf2 pathway is closely related to the NFκB pathway.

In PINP model, oltipraz, a Nrf2 activator, attenuated established mechanical allodynia by upregulated Nrf2 and HO-1 in the spinal cord.¹⁰ Also dimethyl fumarate, a Nrf2 activator, significantly attenuated paclitaxel-induced thermal hyperalgesia and cold/mechanical allodynia by both decreasing pro-inflammatory cytokines in the sciatic nerves and p38 mitogen-activated protein kinase and brain-derived neurotrophic factor in the spinal cord.²⁹ Rosiglitazone, a selective agonist of PPARγ, attenuated PINP through induction of Nrf2/heme oxygenase-1 (HO-1) signaling pathway in the spinal cord.³⁰ However, paclitaxel produced peripheral neuropathy because it did not cross the blood-brain barrier. Therefore, paclitaxel can damage to peripheral nerve including DRG and sciatic nerve and then induced the changes in the dorsal horn of spinal cord. BM is contributed to the whole body including peripheral and central nerve systems because it can cross the blood-brain barrier.²⁶ Therefore, we think PINP may be involved in both first DRG and then spinal cord. In this article, we measure the levels of Nrf2 in the DRG.

We agree that the motor performance is important for the interpretation of von Frey data and posture/righting reflex is not sufficient. However, BM showed no toxicity after the oral administrations of BM once daily for 21 days at doses of 150, 200, 250, and 300 mg/day in pancreatic adenocarcinoma patients in the Phase I clinical trials.^26,31 Therefore, we think that BM did not produce the alteration of motor function and ameliorated CINP.

In present study, for the BM-treated paclitaxel-injected rats, BM (10 mg/kg) was injected intraperitoneally twice on day 21 and in the morning on day 22. One hour after the last injection of BM on day 22, the lumbar DRG were collected. After the injection of BM, the mechanical threshold was 5.15 g, 8.47 g, 6.87 g, and 7.94 g on days 22, 23, 24, and 25 (Figure 2). Therefore, we think that DRG were collected on day 22 was available for WB analysis.

In our study, pNrf2 was expressed in CGRP-expressing neurons in the DRG, which means that pNrf2 was expressed in most unmyelinated C fiber but some myelinated Aβ fiber.^32,33 Liu et al. reported that CGRP reduces apoptosis of DRG cells induced by oxidative stress injury through HO-1 and Nrf-2 expression.³⁴ CGRP co-localizes with various peptides, including substance P and galanin, in the DRG.^35,36 In addition, pNrf2 was expressed in GFAP-expressing satellite cells in human DRG. Various pro-inflammatory cytokines, including tumor necrosis factor-α and interleukin-1β, were produced by satellite cells in CINP.^4,23 Together, these results indicate that pNrf2 is associated with pain perception in the DRG and thus presents an important target for analgesic effect.

We used the 50B11 DRG neuronal cell line that is differentiated by forskolin, nerve growth factor, and capsaicin and labelled with pan-neuronal markers, TRPV1, TRPA1, and peripherin, which means that 50B11 cells are characterized as a nociceptor.^18,37 In the present study, paclitaxel produced mitochondrial damage in the 50B11 cells, including decreased MMP and increased mitochondrial volume. We previously reported paclitaxel increased mitochondrial superoxide level in cells.⁴ During damage to mitochondria, MMP is associated with loss of the electrochemical gradient. Thus, MMP can be used as an indicator of mitochondria.³⁸ We measured mitochondrial volume using MitoTracker Green FM, which can measure mitochondrial mass.³⁹ In the present study, we found that BM may reverse paclitaxel-induced mitochondrial damage in 50B11 cells.

In the present study, we measured the damage of mitochondria in 50B11 cells because paclitaxel can damage the sensory nerve including CGRP-expressing neurons, which is 50B11 DRG cells. Therefore, we chose to assess paclitaxel-induced mitochondrial damage in 50B11 cells.

In the present study, single and repeated injections of BM significantly increased mechanical thresholds without sedation. This is the first report of BM’s effect on neuropathic pain. BM has been used for the several cancers including leukemia, which means that BM may available for both the treatment of cancer and CINP.²⁶ Therefore, BM may be an excellent candidate for CINP treatment.

This study has clear limitations. First, the present study was performed with a small number of rats; a higher number of rats is needed for greater statistical power. In this study, transcriptomic profiling of DRG had 2 groups (e.g., VEH group and PAC group), which is low group numbers. Therefore, we used this transcriptomic data as the preliminary data and then added pain behavioral, western blotting, mitochondrial membrane potential, and immunohistochemical data for the clear of results. Second, the present study was performed with only male rats. Therefore, we will perform our next study with female rats. Third, the mechanisms of CIPN vary greatly depending on chemotherapeutic drugs; this study’s results are limited to paclitaxel. Although the present results had those limits, our results suggest that activation of pNrf2 may be a good candidate for patients with CINP.

In conclusion, systemic administration of activation of pNrf2 by BM ameliorated PINP, inhibited pNFκB and MCP-1 in the DRG, and increased HO-1 in the DRG. In addition, pNrf2 was expressed in neuronal cells, including CGRP-expressing neuronal cells and satellite cells, in rat and human DRG. Taken together, our results suggest that BM produces analgesic effects in CINP by inducing pNrf2 and inhibiting pNFκB and MCP-1 in the DRG and that the activation of pNrf2 plays an important role in the treatment of CINP.

Supplementary Material

Supplementary Figure. 1

Supplemental Figure 1. Co-localization of phosphorylated nuclear factor erythroid 2-related factor 2 (pNrf2), NeuN, CGRP, GFAP, and DAPI in painful human dorsal root ganglia (DRG). (A) pNrf2 (green, Alexa Fluor 488) in the human DRG. (B) NeuN (red, Alexa Fluor 594) in the human DRG. (C) DAPI (blue) in the human DRG. (D) pNrf2 (green), NeuN (red), and DAPI (blue) in the human DRG. Arrows indicate neurons with pNrf2. (E) pNrf2 (green) in the human DRG. (F) CGRP (red, Alexa Fluor 594) in the human DRG. (G) DAPI (blue) in the human DRG. (H) pNrf2 (green), CGRP (red), and DAPI (blue) in the human DRG. Arrows indicate CGRP-expressing neurons with pNrf2. (I) pNrf2 (green) in the human DRG. (J) GFAP (red, Alexa Fluor 594) in the human DRG. (K) DAPI (blue) in the human DRG. (L) pNrf2 (green), GFAP (red), and DAPI (blue) in the human DRG. Arrows indicate GFAP-expressing satellite cells with pNrf2. (M) The bar graph shows the numbers of pNrf2-positive neurons in various sized neurons in the human DRG. (N) The bar graph shows the numbers of both pNrf2- and CGRP-positive neurons in the human DRG. Scale bars = 100 μm (in insert, scale bars = 20 μm).

KEY POINTS SUMMARY.

• Question: Can bardoxolone methyl (BM) ameliorate chemotherapy-induced neuropathic pain?

• Findings: BM ameliorated mechanical hyperalgesia by 1) decreasing inflammatory mediators, 2) decreasing mitochondrial damage, and 3) increasing pNrf2 in the dorsal root ganglia.

• Meaning: pNrf2 activator, like BM, may be useful drugs for the treatment of chemotherapy-induced neuropathic pain.

GLOSSARY OF TERMS

BM

bardoxolone methyl

CGRP

calcitonin gene-related peptide

CINP

chemotherapy-induced neuropathic pain

DAPI

4′,6-diamidino-2-phenylindole

DMSO

dimethyl sulfoxide

DRG

dorsal root ganglia

FBS

fetal bovine serum

GAPDH

glyceraldehydes-3-phosphate dehydrogenase

GFAP

glial fibrillary acidic protein

HO-1

heme oxygenase-1

MCP-1

monocyte chemoattractant protein 1

MMP

mitochondrial membrane potential

NBM

Neurobasal medium

Nrf2

Nuclear factor erythroid 2-related factor 2

PINP

paclitaxel-induced neuropathic pain

pNFκB

phosphorylated nuclear factor kappa B

pNrf2

phosphorylated Nuclear factor erythroid 2-related factor 2