Chemotherapy for pain: reversing inflammatory and neuropathic pain with the anticancer agent mithramycin A

2.1. Animals

C57Bl/6 male and female mice (8-12 weeks) were obtained from the Jackson Laboratory (Bar Harbor, ME) or bred and housed within the UCSF animal care facility in a climate-controlled room on a 12-hours light/dark cycle. Laboratory diet was available ad libitum, except when the mice were being tested.

2.2. Drug administration

2.3. Behavioral assessments

Efforts were made to minimize the number of mice used and their discomfort. Experimental protocols were approved by the University of California, San Francisco, Institutional Animal Care and Use Committee (IACUC) and adhered to the NIH Guidelines for the Care and Use of Laboratory Animals. Investigators who conducted behavioral tests were blinded to treatment condition, and animals were randomized to groups.

2.4. Cell culture

Primary DRG neurons were cultured from 8- to 12-week-old adult male or female mice (C57Bl/6) using methods previously described.^85,108 HCT116 cells (ATCC) were maintained in McCoy 5a media supplemented with 10% fetal bovine serum (Corning, Tewksbury, MA) and 1% penicillin-streptomycin. OVCAR3 cells (ATCC) were maintained in ATCC-formulated RPMI-1640 medium (Gibco-Thermo Fisher, Waltham, MA) supplemented with 20% fetal bovine serum (Corning), 0.01 mg/mL bovine insulin (Sigma-Aldrich), and 1% penicillin–streptomycin.

2.5. Calcium imaging and microscopy

Measurement of [Ca²⁺]_i and cell size determinations in primary DRG neurons were performed by preloading cultured neurons plated on coverslips with 5-μM cell permeant Fluo-4 AM (Invitrogen) calcium-sensitive dye (488/520 nm) with Pluronic F-127 (Invitrogen, Waltham, MA) for 45 to 50 minutes at 37°C, 5% CO₂. Dorsal root ganglion neurons were visualized through a Zeiss Axiovert microscope (10x) equipped with an Axiocam camera and quantified using Zen Pro software (Carl Zeiss, Jena, Germany). Dorsal root ganglion neurons were perfused with Hank buffered salt solution with calcium and magnesium, supplemented with 20 mM HEPES and 1% penicillin–streptomycin, at 2 mL/minute at room temperature (22°C) before cold buffer or KCl (50 mM) stimulation. Cold perfusion buffer was applied for 30 seconds with a final bath temperature of 11.3°C (range, 10.5-12.1°C) at the end of cold buffer application, using a cold perfusion unit (TC-RD Bioscience Tools, San Diego, CA). Neurons that responded within 15 seconds after cold buffer application by an increase of [Ca²⁺]_i ≥ 20% from baseline values were defined as responders. Neurons that did not respond to KCl were excluded.

2.6. qRT-PCR

Quantitative Reverse Transcription PCR RNA was isolated from mouse lumbar DRG (Trizol Reagent, Invitrogen), and first strand complementary DNA was prepared (SuperScript III First-Strand Synthesis System, Invitrogen-Thermo Fisher, Waltham, MA). Using the QuantiStudioTM 6 Flex Real-Time PCR system (Applied Biosystems, Waltham, MA), mRNA expression levels were quantified using iTaq Universal SYBR Green Supermix assays (Bio-Rad, Hercules, CA) for TRPM8 (F:GTGTCTTCTTTACCAGAGACTCCAAGGCCA;R:TGCCAATGGCCACGATGTTCTCTTCTGAGT), normalized by GAPDH (F: TGCGACTTCAACAGCAACTC; R: CTTGCTCAGTGTCCTTGCTG) expression, and represented as relative quantitation (RQ) using the comparative ∆∆C_T method.^20,88,90,108

2.7. Immunohistochemistry: intraepidermal nerve fiber

The glabrous skin of the mouse footpad (hind paws) was removed and fixated in 4% PFA at 4°C overnight. Samples were transferred to 20% (wt/vol) sucrose for at least 24 hours and embedded in optimal cutting temperature (OCT) compound, frozen, and sliced into 25-µm sections using a cryostat. Sections were washed in PBS and incubated at room temperature (RT) for 2 hours in blocking solution (5% normal donkey serum [NDS] and 0.3% Triton X-100 in PBS). Free-floating sections were incubated with primary antibody PGP9.5 (panneuronal marker) (rabbit anti-mouse polyclonal, 1:200 dilution, Thermo-Fisher) in wash buffer (1% NDS and 0.3% Triton X-100 in PBS) at RT for 2 hours. Following 3 washes in wash buffer, slices were incubated overnight at 4°C with a 1:400 dilution of donkey anti-rabbit Cy3 secondary antibody (Millipore Burlington, MA) in wash buffer solution. Sections were washed 3 times in PBS, mounted onto slides (Aqua-Polymount, Thermo-Fisher), and imaged using Keyence microscope and software (Keyence Corp. of America, Pleasant Hill, CA). Five sections from each mouse were randomly chosen for intraepidermal nerve fiber (IENF) quantification and 3 fields of view per section using a ×40 objective. The length of the epidermis within each field of view was measured using ImageJ software, and the density of fibers was determined as number of nerve fibers per millimeter. The experimenter was blinded to experimental conditions.

2.8. Apoptosis

OVCAR3 and HCT116 cells (∼ 5000 cells/per well) in a 96-well format (Falcon, Corning Tewksbury, MA) were cultured overnight. Following 24 hours of treatment, apoptosis activity was measured using Caspase Glo 3/7 Assay System (Promega, San Luis Obispo, CA) following manufacturer's protocol. Luminescence was recorded by Cytation 5 Cell Imaging Multi-Mode Reader (BioTek, Winooski, VT) in triplicate measures.

2.9. Measurement of reactive oxygen species in dorsal root ganglion neurons

Adult male mice (8-12 week old) were treated in vivo with oxaliplatin (3 mg/kg, i.p.) at day 0, followed by MTM (100 µg/kg, i.p.) 48 hours and 72 hours postinjection. On day 10, mice were euthanized under carbon dioxide, and approximately half of the total DRGs were harvested and cultured with the addition of uridine to a final concentration of 20 µM. CellROX (5 µM) Green Reagent (Invitrogen-Thermo Fisher) was added to the cells for 30 minutes following manufacturer's protocol. Then, 100-µL Live-Cell Imaging Solution (Invitrogen-Thermo Fisher) was added. Imaging and analysis were performed on Cytation5 using Gen5 software (BioTek, Agilent, Santa Clara, CA).

2.10. Statistical analysis

Nociceptive thresholds were expressed as means ± SEM. When applicable, detection of behavioral differences between multiple groups was performed by 2-way repeat-measures analysis of variance (RM-ANOVA) or 1-way ANOVA followed by Tukey post hoc test. When indicated, independent-sample t-tests and Fisher exact tests were done. A P value of <0.05 was considered statistically significant. Analysis was performed using Prism (GraphPad Software). Figures were composed using Adobe Illustrator and BioRender.

2.11. Transcriptomics

3.1. Mithramycin reverses inflammatory heat hyperalgesia

We previously demonstrated that binding of Sp4 or Sp1 to the TRPV1 promoter P2 region drives the expression of TRPV1 mRNA in cultured mouse DRG neurons.²⁰ Mithramycin, a Sp1-family transcription inhibitor, was shown to block TRPV1 expression in nociceptors.¹⁰⁸ Therefore, we examined if systemic treatment of mice with MTM would block ongoing inflammatory heat hyperalgesia, a TRPV1-dependent pain behavior.^15,16 Following induction of peripheral inflammation by intraplantar (i.pl) injections of CFA, we observed a decrease in paw withdrawal latency (heat hyperalgesia) from a baseline value of 11.6 ± 0.6 to 6.3 ± 0.3 seconds at 48 hours. After 2 sequential daily injection of MTM (100 µg/kg i.p), a dose based on its anticancer and neuroprotective activity in mice,³² a time-dependent reversal of heat hyperalgesia was observed (Fig. 1A), with heat thresholds approaching baseline values at day 6 (11.4 ± 0.6 seconds P = 0.028) and day 10 (11.6 ± 0.6 seconds P = 0.0025), as compared with the vehicle-treated CFA mice with a mean latency of 8.7 ± 0.4 seconds on day 10. Mithramycin did not change heat thresholds when given alone. Importantly, MTM-induced reversal of inflammatory heat hyperalgesia occurred without a measurable change in the development of CFA-induced paw thickness (edema) (Fig. 1B).

To determine if MTM-induced reversal of hyperalgesia in male mice was observed in female mice, we evaluated the ability of CFA to induce heat hyperalgesia in female mice compared with male mice. As shown in (Fig. 1C), male and female mice had indistinguishable baseline thermal paw withdrawal latencies. Female mice developed a greater decrease in paw withdrawal latency (heat hyperalgesia) at days 1 (P = 0.001), 6 (P< 0.0001), and 10 (P = 0.001), as compared with male mice. After 2 sequential daily injections of MTM beginning at 48 hours, a time-dependent reversal of heat hyperalgesia was observed in both male and female mice (Fig. 1D). However, female mice showed a less robust MTM reversal of heat hyperalgesia with MTM reversal latencies at days 6 (P = 0.007) and 10 (P= 0.0001), attaining values equivalent to untreated CFA-male mice (inflammatory heat hyperalgesia). By contrast, MTM treatment reversed CFA-treated male mice to baseline latency values on days 6 (P = 0.007) and 10 (P = 0.0004).

3.2. Mithramycin reverses cisplatin-induced heat and mechanical hypersensitivity

Cisplatin is associated with dose-limiting side effects primarily due to its neurotoxic effects on sensory neurons.^38,71 Behavioral models of cisplatin-induced hypersensitive states induce both heat and mechanical hypersensitivity.⁹⁹ As shown in (Fig. 1E), cisplatin treatment was associated with a reduction in heat paw withdrawal latency, from baseline values of 15.0 ± 1.2 to 9.6 ± 1.2 seconds at 48 hours. Following 2 daily treatments of MTM, heat withdrawal latencies returned to baseline values of 14.8 ± 1.2 seconds on day 4 and stayed near baseline values through day 14 (14.7 ± 0.9 seconds P = 0.0009). By comparison, vehicle-treated cisplatin mice maintained a reduced heat latency of 10.1 ± 0.5 seconds on day 14. To determine if MTM was able to reverse other modalities of cisplatin-induced hypersensitivity, changes in mechanical threshold were measured. Cisplatin, administered daily for 5 days, induced a reduction in mechanical threshold values from a baseline of 0.91 ± 0.03 g to 0.18 ± 0.03 g on day 2. Following MTM treatment (Fig. 1F), mechanical thresholds reversed towards baseline, attaining near baseline values of 0.83 ± 0.09 g on day 7 that were sustained through day 21 (P < 0.0001).

3.3. Mithramycin reverses oxaliplatin-induced cold and mechanical hypersensitivity

Chemotherapy-induced peripheral neuropathy pain is a dose-limiting side effect induced by oxaliplatin.⁵⁵ Given our previous observations that oxaliplatin-induced cold and mechanical hypersensitivity were reversed in Sp4^+/--knockdown mice,⁹⁰ we investigated whether the use of MTM would result in a similar reversal. Using doses of MTM tested in models of cancer (Ewing sarcoma)⁴⁰ and neurodegeneration (Huntington disease [HD])³² and the lowest profile of toxicity, we compared a single dose of MTM at 10, 50, or 100 µg/kg to reverse oxaliplatin (3 mg/kg)-induced cold (10°C) hypersensitivity. As shown in Fig. 2A, oxaliplatin (3 mg/kg i.p.) induced a decrease in paw withdrawal latency (cold hypersensitivity) at 48 hours (P = 0.0008) and 72 hours (P < 0.0001). Reversal of cold hypersensitivity was achieved at 72 hours following MTM at 100 µg/kg (P = 0.0024) but not at 10 or 50 µg/kg. Overall, MTM appears to reverse behavioral models of nociceptive pain at a 10-fold lower dose than required to reduce an experimental tumor model and within the working dose range to increase survival in a model of HD.^32,40 Then, we studied the persistence of oxaliplatin-induced cold hypersensitivity following MTM treatment (Fig. 2B). Oxaliplatin-induced a decrease in 10°C paw withdrawal latency in male mice from 15.9 ± 1.0 to 10.7 ± 0.7 seconds at 24 hours. Mithramycin treatment resulted in a reversal of oxaliplatin-induced hypersensitivity, achieving baseline values of 16.4 ± 0.9 seconds from day 3 to day 14 (P < 0.0001), through day 21 (P = 0.022). When testing was repeated at 4°C (Fig. 2C), MTM demonstrated a similar reversal on days 3 (P = 0.0083), 6 (P < 0.0001), 10 (P = 0.0016), 14 (P = 0.016), and 21 (P < 0.0001). Mithramycin alone (Figs. 2B and C) failed to change baseline paw withdrawal latency in control mice.

Mechanical hypersensitivity associated with oxaliplatin treatment has been modeled in rodents.^48,74,75,94 As shown in Fig. 2D, oxaliplatin induced a robust decrease in mechanical nociceptive threshold from a baseline of 0.85 ± 0.01 g to 0.20 ± 0.02 g at 48 hours after oxaliplatin treatment. Mithramycin produced marked reversal of mechanical hypersensitivity to baseline values of 0.83 ± 0.05 g by days 3 to 10 (P < 0.0001) that began to diminish on day 14 (P = 0.044). Mithramycin did not change mechanical thresholds when given alone.

To determine if MTM pretreatment could prevent oxaliplatin-induced mechanical hypersensitivity, MTM was administered on day −1 and again 45 minutes before oxaliplatin administration on day 0. As shown in (Supplementary Figure 1, available at http://links.lww.com/PAIN/B865), pretreatment with MTM reduced the magnitude of mechanical hypersensitivity on days 1 to 3 as compared with oxaliplatin-only control at 24 hours, 48 hours, and day 3 (P < 0.0001).

Given our finding that MTM reversed CFA-induced inflammatory heat hyperalgesia to a greater degree in male than female mice (Fig. 1D), we evaluated the ability of MTM to reverse oxaliplatin-induced cold and mechanical hyperalgesia in female mice. As shown in Figs. 2E, F, G, although oxaliplatin induced equivalent hypersensitivity in male and female mice, oxaliplatin-treated female mice showed trends of MTM reversal of cold and mechanical hypersensitivity, reaching significance on day 3 at 4°C (P = 0.0052). To quantitate differences in MTM rescue between male and female mice, we examined the percent efficacy of MTM reversal of CFA and oxaliplatin-induced hypersensitivity. As shown in Supplementary Figure 2, available at http://links.lww.com/PAIN/B865, MTM was approximately 50% more efficacious in male mice reversing inflammatory heat hyperalgesia on all days measured: day 3 (P = 0.021), day 4 (P = 0.014), day 6 (P = 0.0013), and day 10 (P = 0.017). Mithramycin was more efficacious in male mice at reversing oxaliplatin-induced cold (10°C) hypersensitivity on day 6 (P = 0.010) and day 10 (P = 0.034).

3.4. Mithramycin reverses oxaliplatin-induced cold responsiveness of dorsal root ganglion neurons

Platinum-based chemotherapy primarily drives neurotoxicity through their transport and accumulation in peripheral sensory neurons.^38,55,71,74 Platinum agents, such as oxaliplatin, are largely, but not completely, excluded from the CNS.⁴⁶ Therefore, painful cold and mechanical hypersensitivities arising from oxaliplatin may be driven by alterations in peripheral sensory neurons. To test our hypothesis that the target for systemically administered MTM to reverse hypersensitivity to cold resides, at least in part, in the sensory neurons, we measured changes in the percentage of DRG neurons (ex vivo) that were activated by the application of cold buffer following in vivo treatments with oxaliplatin with or without MTM. As shown in Fig. 3A, the percentage of cold responsive primary DRG neurons was higher in oxaliplatin-treated mice compared with vehicle controls (P = 0.0008). By contrast, DRG neurons derived from mice receiving oxaliplatin plus MTM showed a reduced percentage of cold activated neurons (P = 0.019). In addition, cold responsive DRG neurons had a smaller size distribution (Fig. 3B), with a mean area of 304 ± 22 µm² compared with cold nonresponders 423 ± 19 µm² (P < 0.0001). These findings support our hypothesis that MTM in vivo reverses oxaliplatin-induced hypersensitivity to a cold stimulus, in part, through its action on small-diameter sensory neurons.

3.5. Mithramycin inhibition of oxaliplatin cold hypersensitivity is associated with reversal of TRPM8 overexpression in dorsal root ganglion neurons

TRPM8, a channel activated by cold temperatures (26°C-15°C), expressed in a subset of small-diameter DRG neurons,^9,51,72 is proposed to be a transducer of painful oxaliplatin-induced cold hypersensitivity.³⁵ We investigated whether the MTM reversal of oxaliplatin-induced cold hypersensitivity was linked to changes in TRPM8 expression in DRG. We confirmed oxaliplatin's ability to induce cold hypersensitivity in vivo in mice and that DRG derived from these mice had an increase (∼4 fold) in TRPM8 mRNA (P = 0.0006) (Fig. 3C).³⁵ Importantly, DRG derived from mice tested with oxaliplatin followed by MTM rescue showed reversal of TRPM8 mRNA overexpression to baseline levels (P = 0.001).

Given our observations that MTM reversed oxaliplatin-induced cold hypersensitivity (Figs. 2A–C), we then validated our ability to pharmacologically block TRPM8 and TRPA1, a TRP channel with conflicting reports of cold activation. As shown in (Fig. 3D), TRPA1 antagonist HC-03003 completely blocked paw-licking behavior in response to AITC (P = 0.015) and TRPM8 antagonist AMG333 inhibited icilin-induced “wet dog shakes” (P = 0.0085) (Fig. 3E). Then, we tested whether blockade of TRPM8 or TRPA1 would reverse oxaliplatin-induced cold hypersensitivity. As shown in Fig. 3F, the TRPM8-specific antagonist AMG333 reversed oxaliplatin-induced cold hypersensitivity (decreased paw withdrawal latency at 10°C) two hours post treatment (P = 0.002). Whereas, the TRPA1 antagonist was without effect. Taken together, these findings support the hypothesis that MTM reversal of oxaliplatin-induced cold hypersensitivity is, in part, due to its ability to normalize TRPM8 gene expression in DRG neurons under conditions of oxaliplatin-induced neurotoxicity.

3.6. Mithramycin induces apoptosis in ovarian cancer cell line OVCAR-3 and increases oxaliplatin-induced apoptosis in human colon cancer cell line HCT116

To further validate the activity of MTM used throughout this study, we sought to determine its anticancer (proapoptotic) properties in 2 established cancer models. First, we confirmed MTM's ability to induce apoptosis in the ovarian cancer cell line OVCAR3. Mithramycin at 50 nM, 200 nM, and 400 nM, increased apoptosis of OVCAR cells at 24 hours (P < 0.0001) as shown in Supplementary Figure 3, available at http://links.lww.com/PAIN/B865.³¹ Subsequently, we examined the action of MTM in the presence of oxaliplatin. Oxaliplatin, a first-line treatment for metastatic colon cancer, mediates its cytotoxic action through complex mechanisms including disruption of DNA replication and apoptosis.¹⁰⁶ Although oxaliplatin-associated DNA adducts are maintained by covalent binding, concurrent treatment with MTM, a DNA binding transcriptional inhibitor,⁶⁵ may displace, disrupt, or compete with oxaliplatin's anticancer action. As literature to support or refute this possibility could not be identified, we tested whether MTM at 50 nM, a dose previously found to decrease TRPV1 expression in DRG neurons in vitro,¹⁰⁸ would block oxaliplatin-induced apoptosis in a model of human colon cancer.¹⁰⁶ As shown in Supplementary Figure 3, available at http://links.lww.com/PAIN/B865, MTM increased apoptosis in the human colon cancer cell line HCT116 (P < 0.0001). Moreover, MTM (50 nM) with 10 or 20 µM oxaliplatin increased apoptosis in HCT116 cells in an additive manner (P < 0.0001). These findings support the suggestion that MTM used in our studies retains its anticancer activity⁸³ and complimented rather than blocked oxaliplatin's apoptotic activity in HCT116 cells.

3.7. Persistent oxaliplatin-induced cold and mechanical hypersensitivity is reversed by mithramycin

A progressive, dose- and cycle-dependent relationship occurs with the transition from acute to chronic cold- and mechanical-evoked pain following platinum-based chemotherapy. Therefore, cumulative exposure to oxaliplatin is associated with progressive increases in severity and duration (chronicity) of CIPN pain. To determine if MTM is able to reverse a persistent neuropathy model that is associated with cold and mechanical hypersensitivity following multiple oxaliplatin cycles, we treated mice with oxaliplatin (3 mg/kg) twice weekly for 4 weeks, followed by 2 doses of MTM (100 µg/kg). As shown in Fig. 4A, repeated injections of oxaliplatin decreased withdrawal latency to cold (10°C) from a baseline of 12.8 ± 0.5 seconds, to values that persisted through day 28 (6.9 ± 0.5 seconds). Following the first of the 2 MTM treatments (day 29), the persistent decrease in paw withdrawal latency reverted to baseline values (12.4 ± 0.8 seconds) that continued for the duration of the study period (1 week) until day 35 (P < 0.0001). Similar findings were observed (Fig. 4B) with MTM reversal of persistent oxaliplatin hypersensitivity evoked by a colder stimulus (4°C) with a decrease from baseline values of 10.7 ± 0.5 to 5.9 ± 0.7 seconds by day 28. Following MTM, latency values returned to baseline (10.4 ± 0.6 seconds) on day 29 (P = 0.0004) but maintained only a partial reversal on days 30 (P = 0.0011) and 31 (P = 0.049), until day 35 (P= 0.025).

In addition, mechanical threshold testing was performed following repeated doses of oxaliplatin resulting in a decrease from 0.83 ± 0.03 to 0.16 ± 0.03 g by day 28 (Fig. 4C). Following 2 MTM treatments, oxaliplatin mechanical hypersensitivity was reversed by approximately 50% for the duration of the study period with values of 0.49 ± 0.07 g by day 35 (P < 0.0001). Overall, the MTM-directed reversal of persistent cold (10°C and 4°C) and mechanical hypersensitivity, suggesting that certain sustained transcription-dependent changes induced by oxaliplatin are shared by cold and mechanical sensory transduction and can be reversed by an inhibitor of Sp1-like factors.

3.8. Mithramycin does not reverse oxaliplatin-induced decreases in intraepidermal nerve fiber density

Several chemotherapeutic agents, such as oxaliplatin, induce peripheral neuropathy.¹⁴ A decrease in intraepidermal nerve fiber (IENF) density was reported in mice 4 weeks after paclitaxel treatment¹⁰; after 11 weeks of oxaliplatin (3 mg/kg × 10 doses over 3 weeks)¹⁰³; and at 30 days in humans (3 doses of oxaliplatin): 3 of 8 patients had a decreased IENF density that at day 180 was more pronounced.¹⁴ Given this variability, the relationship between chemotherapy-induced reduction of IENF density and painful hypersensitivity remains an open question. Therefore, we investigated whether MTM treatment (2 doses), which reversed persistent cold and mechanical hypersensitivity, would also reverse oxaliplatin-induced reduction in IENF density. As shown in Figs. 5A–E, long-term oxaliplatin treatment (8 doses over 4 weeks) reduced IENF density (30 ± 1.3 IENF/mm), compared with vehicle control (34.0 ± 1.2 IENF/mm) (P < 0.05). Although MTM reversed persistent oxaliplatin-induced cold and mechanical hypersensitivity (Fig. 4), MTM failed to reverse the oxaliplatin-induced IENF loss during the same period of our investigation (7 days following the first MTM treatment) (28.1 ± 1.0 IENF/mm) (P = 0.002). When compared with vehicle, MTM treatment alone did not change the IENF density (36 ± 0.6 IENF/mm). Moreover, oxaliplatin plus vehicle (P = 0.003) or oxaliplatin plus MTM (P < 0.0001) compared with MTM alone were both associated with persistent reduction in IENF density.

3.9. Oxaliplatin induces genome wide transcriptional changes within dorsal root ganglion

Although oxaliplatin-induced transcriptional inhibition represents a fundamental mechanism underlying its anticancer activity,³ how oxaliplatin binding to the DRG genome results in painful CIPN remains poorly understood. Therefore, we undertook an extensive analysis of the transcriptional changes induced by oxaliplatin in the DRG to identify not only potential genes of interest but also to establish a framework of gene networks that may underlie oxaliplatin-induced painful hypersensitivity. Lumbar DRG RNA was extracted from mice treated with only oxaliplatin (oxaliplatin + vehicle OV) and compared with DRG RNA from vehicle-only DRG RNA (vehicle + vehicle VV). A total of 931 DEGs were identified (FDR < 0.05). The top DEGs with a log fold change of ±0.38 (fold change for ±1.5) are displayed in Fig. 6A. A complete table of the DEGs is shown in Supplementary Table 1, available at http://links.lww.com/PAIN/B864. Assuming the transcriptome of approximately 30,000 genes, approximately 3% of genes showed significant changes. This finding suggests broad changes in gene expression occurred following the administration of oxaliplatin.

Two prior studies used RNAseq to examine the DRG transcriptomic changes because of oxaliplatin treatment.^5,98 The DEGs from our oxaliplatin-only (OV) were compared with DEGs from these studies. When compared with the data published from Starobova et al.,⁹⁸ only 7 genes appeared in our data set (ie, A330023F24Rik, Cadm2, Gm26871, Meg3, Nav2, Zbtb33, and Zfp804a). It is worth noting that Starobova et al. used intraplantar injection of oxaliplatin to induce a hypersensitive state; by contrast, our studies used systemic administration. When compared with the data published from Bangash et al.,⁵ using microarray assays for gene expression quantification, no overlapping genes were found between data sets representing oxaliplatin-treatment alone. We suggest that a combination of systemic oxaliplatin treatment and an RNA sequencing depth of approximately 50 million reads facilitated the identification of a greater number of transcriptomic changes than previously reported. As such, our data serve as an ongoing resource for the study of transcriptional changes that occur within the DRG because of oxaliplatin treatment in vivo.

As shown in Fig. 6B, using DEGs derived from oxaliplatin-treated mouse DRG, an accumulative hypergeometric statistical analysis was used to identify the ontological terms where the input genes show significant enrichment. This enrichment analysis revealed functional associations among the DEGs including the regulation of plasma membrane projection organization, behaviors, synaptic signaling, and cell morphogenesis. Analysis of which putative transcription factor(s) regulate the oxaliplatin-induced DEGs revealed 2 members (Sp1, Sp3) of the Sp1-like transcription factor family (Fig. 6C). Sp1 and Sp4 are drivers of TRPV1 promoter activity and TRPV1 expression in DRG neurons²⁰ with member Sp4 being necessary for the persistence of oxaliplatin-induced hypersensitivity.⁹⁰

3.10. Mithramycin reversal of oxaliplatin-induced hypersensitivity is associated with transcriptional changes only partially overlapping those induced by oxaliplatin

As shown in Fig. 7A, a Venn diagram illustrates the distribution of DEGs between oxaliplatin treatment only (OV), mithramycin rescue of oxaliplatin treatment (OM) and MTM only (VM). Of the 496 DEGs from the mithramycin rescue group, only 61 genes overlap with the oxaliplatin-only condition. Mithramycin treatment alone is associated with only 22 DEGs. Therefore, the majority of transcriptional changes observed with MTM-mediated reversal of painful hypersensitivity are not associated with oxaliplatin-induced DEGs. Heat maps of these DEGs are shown in Fig. 7B.

The top MTM-associated DEGs are shown in Fig. 8A, and a complete list of oxaliplatin + MTM rescue DEGS are listed in Supplementary Table 2, available at http://links.lww.com/PAIN/B864. Ontology enrichment analysis of MTM rescue DEGs (Fig. 8B) show top functional associations with Parkinson disease, transport and uptake of IGF, regulation of cellular proliferation, cellular homeostasis, and synapse organization. Notably, analysis of how putative transcription factor(s) regulate the MTM rescue–associated DEGs revealed two: Dnmt3a and Sp1. DNA methylase 3a (Dnmt3a) was found to be associated with neuropathic pain through repression of KCNa2 in primary afferent neurons.¹¹⁰ Sp1 is the well-described target of MTM—its principal inhibitor of DNA binding to GC-rich promoter sites. The 61 DEGs overlapping between oxaliplatin only and MTM rescue are shown in Supplementary Figure 4, available at http://links.lww.com/PAIN/B865. Although a survey of these DEGs (PubMed) revealed several individual pain-related genes, no overall thematic pattern of gene function was appreciated.

3.11. Mithramycin reversal is associated with differentially expressed genes whose function are at multiple levels of known nociceptor function

To better understand the potential function of the DEGs underlying MTM reversal of painful hypersensitivity at the level of molecular nociception, we compared these genes with known pain gene databases and searched in a semiautomated manner through a PubMed query. Differentially expressed genes with known association with pain were putatively assigned a category of function as depicted in Fig. 9. References that support these categorizations are found in Supplementary Table 3, available at http://links.lww.com/PAIN/B864. Based on this categorization of MTM reversal–associated DEGs, we hypothesize that these MTM-regulated genes modulate nociceptor function at multiple levels through transcription, transduction, and transmission.

3.12. Protein-protein enrichment analysis of mithramycin reversal identifies mitochondrial electron transport chain genes as a potential therapeutic target

Using the list of DEGs identified through the analysis of the MTM reversal group, a further enrichment analysis was performed at the level of their protein products. By evaluating DEGs at this level, protein–protein interaction network analysis can identify biologically relevant pathways/processes with a better signal-to-noise ratio than the ontological enrichment analysis of RNA transcript levels alone. The Metascape tool uses a network analysis algorithm (the molecular complex detection algorithm—MCODE⁴) to cluster DEGs with highly associated protein–protein interactions. As shown in Fig. 10, genes that have significant protein–protein interactions are related to the function of the mitochondrial electron transport chain (ETC)—oxidative phosphorylation, MCODE 5 (see boxed regions) (Fig. 10). Additional DEGs related to mitochondrial function under conditions of MTM reversal or oxaliplatin treatment are shown in Supplementary Figure 5, available at http://links.lww.com/PAIN/B865.¹⁰⁰ Overall, these MCODE5-identified genes include principal components of ETC's Complex 1: NADH dehydrogenase (ND1, ND3, ND4, ND4L, NDufa6, Ndufb6)¹¹¹ that is a critical production site for pathological reactive oxygen species (ROS).⁵⁷ Given that mitochondrial ROS production is known to serve a critical role in neuronal signaling, nociceptor function, and hypersensitivity, we sought to link this protein–protein interaction analysis to potential mechanisms driving both oxaliplatin-induced hypersensitivity and its reversal by MTM in vivo.

3.13. Mithramycin reverses oxaliplatin-induced increases of reactive oxygen species in dorsal root ganglion neurons

Excessive ROS generation induced by platinum agents such as oxaliplatin, that leads to oxidative stress and mitochondrial dysfunction, contribute to the development of CIPN.^45,76 DRG neurons exposed to oxaliplatin in vitro induced a dose-dependent and time-dependent mitochondrial production of ROS.^50,63 Given the observed enrichment of ETC components following MTM rescue, we investigated whether DRG neurons increased their content of ROS in response to oxaliplatin treatment in vivo and if MTM treatment could reverse this effect. As shown in Fig. 11, ROS levels were significantly increased in DRG neurons derived from oxaliplatin-treated mice (2.2 ± 0.47) 10 days after oxaliplatin treatment as compared with those from vehicle-treated mice (1.0 ± 0.10) (P = 0.0004). Although MTM (100 µg/kg, i.p.) treatment alone on day 2 and day 3 did not significantly change ROS levels, DRG neurons cultured from oxaliplatin-treated mice and subsequently with MTM on days 2 and 3 displayed ROS levels of 0.83 ± 0.09 on day 10 that did not differ from baseline to control values. We propose that as a component of its transcriptional effects in DRG neurons, MTM reversal of oxaliplatin-induced ROS overproduction underlies, in part, MTM reversal of painful hypersensitivity to oxaliplatin.

3.14. Differential alternative splice analysis reveals cytoskeletal-dependent transport genes as a major site of mithramycin regulation in oxaliplatin treated dorsal root ganglion

Analysis of gene expression as measured by DEGs alone may limit our understanding of the complexity of oxaliplatin-induced and MTM-induced changes across the DRG genome. Differential alternative RNA splicing is another critical component in the regulation of gene expression.¹⁰¹ We designed our RNAseq experiments for a depth of approximately 50 million reads per sample allowing for differential splice analysis using the established rMATs package.⁹¹ At this deeper depth of read, we aimed to identify important splice changes in gene families and pathways that may otherwise go undetected.

There were 2313 differential splice events across 1533 genes with the MTM reversal and oxaliplatin treatment groups. The principal component plot (PCA) (Fig. 12A) shows the first principal component (x-axis) in the dimension where there is the greatest variance (PC1 39.9%). As the red (MTM rescue) and blue (oxaliplatin only) conditions have the largest separation in this dimension, we interpret that MTM treatment contributes the greatest variance in splicing between the groups. Overall, the represented differential splicing is occurring within the same gene(s). However, the proposed spliced isoforms are different between the 2 treatment groups (MTM rescue vs oxaliplatin only).

Based on the identification of differences in the splice events of individual genes driven by MTM treatment, an ontological enrichment pathway analysis of the differentially alternatively spliced genes was conducted to identify pathways enriched for differential splicing events. Most of the differentially spliced genes have a function related to cytoskeletal-dependent transport, synaptic organization, and cytosolic transport that overall align with axonal transport (Fig. 12B). As axonal transport was identified as a key mechanism for CIPN, these differential splicing events may represent early changes that contribute to CIPN.⁵²

4.1. Reversal of inflammatory and neuropathic pain by mithramycin

Inflammation induced by localized injection of CFA produced heat hyperalgesia in male mice that was completely reversed by MTM (Figs. 1A and D). Similarly, MTM reversed cisplatin-induced heat hyperalgesia (Fig. 1E). Both are TRPV1-dependent processes.^15,26,78,99 These observations support our prior findings that MTM blocked TRPV1 expression through Sp1- and Sp4-dependent mechanisms.²⁰ In addition, MTM treatment of DRG neurons in vitro reduced the number of capsaicin-responsive DRG neurons rather than their response magnitude.¹⁰⁸ These findings suggest that MTM in vivo decreased the expression of TRPV1 in a subset of sensory neurons that are dependent on Sp1-like transcription factors. Sp4 is expressed predominantly in smaller-diameter DRG neurons and is critical for the maintenance of CFA-induced painful hypersensitive states.⁹⁰ We hypothesize that the action of MTM under inflammatory and cisplatin conditions is directed at this dynamic subpopulation of DRG neurons by reversing aberrant Sp4- and/or Sp1-dependent neuroplastic changes in the DRG genome. Therefore, ongoing expression of TRPV1 in sensory neurons lacking Sp1-like regulation would be insensitive to MTM's action. This model may explain why we observed no difference in baseline heat withdrawal latencies in Sp4^+/--knockout mice, whereas TRPV1^−/−-null mice had higher heat withdrawal latencies.⁹⁰

The platinum-based chemotherapy oxaliplatin, used to treat metastatic colon cancer, is associated with significant dose-limiting neuropathic side effects, including cold allodynia and hyperalgesia that are poorly understood.⁷⁵ We observed MTM reversal of oxaliplatin cold hypersensitivity without a change in control baseline withdrawal latencies (Figs. 2B and C).⁷⁷ Multiple mechanisms have been examined to both understand and develop effective therapies for oxaliplatin-induced painful CIPN,^1,27,68 including the role of TRPM8.^67,68 Our studies, support the hypothesis that MTM reverses oxaliplatin-induced cold hypersensitivity (Figs. 2A–C) by normalization of TRPM8 gene expression in nociceptors (Fig. 3).

When compared with male mice, the observations that female mice exhibited a greater CFA-induced heat hyperalgesia (Fig. 1C) but lower efficacy of MTM reversal of hypersensitivity (Supplementary Figure 2, available at http://links.lww.com/PAIN/B865) suggests sex differences in genes regulated by Sp1-like factors. Although other factors may help explain these findings,⁸⁰ a report that human female tibial nerves express higher amounts of Sp4 than those from male subjects,⁸⁴ suggesting future studies to determine what mechanism(s) underlies these observations.

4.2. Mechanisms for the reversal of pain by mithramycin

The mechanism(s) that underly MTM-dependent reversal of cisplatin (Fig. 1F) and oxaliplatin-induced mechanical hypersensitivity (Figs. 2D and 4C) remain to be determined.^49,71 Peripheral and central TRPV1 expression is required for the development and persistence of mechanical hypersensitivity in multiple models of painful hypersensitivity,^{19,23,24,29,34,36,93} including cisplatin⁹² and oxaliplatin-induced¹⁸ CIPN. Therefore, MTM's ability to block Sp4-dependent TRPV1 expression may contribute to MTM's reversal of mechanical hypersensitivity. More broadly, Sp4 is coexpressed in a subset of nociceptors that bind Isolectin B4 (IB4),⁹⁰ with mechanosensitive properties.^6,81 Notably, ablation of IB4+ DRG neurons eliminates oxaliplatin-induced mechanical hypersensitivity.⁴⁸ We hypothesize that the ability of MTM to reverse mechanical hypersensitivity induced by cisplatin and oxaliplatin includes its modulation of genes expressed in IB4+, Sp4+ primary afferent neurons. This idea is plausible given that CFA-induced inflammation increases TRPV1 expression (24% > 80%) in IB4+ DRG neurons,¹¹ and the majority of IB4+ DRG neurons express Sp4.⁹⁰

Clinically, oxaliplatin is administered in multiple cycles (doses) over time. An increasing cumulative dose leads to greater effectiveness of its anticancer activity, but with increasing risk of neurotoxicity,¹² CIPN, and peripheral neuropathies that include a decrease of intraepidermal nerve fiber (IENF) density in humans¹⁴ and in animal models.^10,103 As shown in Figs. 4A–C, persistent cold and mechanical hypersensitivity sustained over 4 weeks of oxaliplatin treatment was reversed following MTM treatment. However, MTM rescue of hypersensitivity observed after 1 month of oxaliplatin did not reverse the oxaliplatin-induced loss of IENF density 7 days after MTM treatment (Fig. 5) despite ongoing anatomic evidence of small-fiber neuropathy. This finding supports the hypothesis that loss of IENF is associated with the development of pain behaviors but is not necessary or sufficient for pain persistence.

Which genes underlie MTM's reversal of heat, mechanical, and noxious cold hypersensitivity? To address this question, we applied multiple analytic approaches to DRG RNAseq data derived from oxaliplatin-treated mice with, or without, subsequent MTM rescue. As detailed in Fig. 7, MTM treatment alone under control conditions represented the lowest number of DEGs (22) and correlated with stable baseline nociceptive thresholds (Figs. 1A, 2B–D). By contrast, MTM rescue of oxaliplatin treatment resulted in 496 DEGs (Supplementary Table 2, available at http://links.lww.com/PAIN/B864). We interpret these findings to suggest that MTM is having its greatest transcriptional impact on DRG gene expression under oxaliplatin-induced neurotoxicity, reflecting de novo access of MTM to Sp1-like regulatory sites. This idea is consistent with our findings showing that MTM-rescue is associated with enrichment of genes regulated by Sp1-like transcription factors (Figs. 6C, 8C). We assembled a model of a nociceptor with the functional classification of DEGs identified from MTM rescue that spanned a range of signaling pathways (Fig. 9). Then, we applied an unbiased analysis of protein–protein interaction enrichment to identify what genes or gene networks could underly an overarching mechanism of MTM's reversal of painful hypersensitivity. As shown in Fig. 10, the highest scoring interaction identified mitochondrial components of the electron transport chain (ETC).

4.3. Mitochondria and mithramycin

Platinum-based chemotherapy agents like cisplatin and oxaliplatin accumulate and form DNA adducts in DRG that induce DNA/mitochondrial damage, enhancing reactive oxygen species (ROS) production while activating nociceptor-associated channels.^45,63,76,102 Exactly how oxaliplatin or other platinum chemotherapy increase mitochondrial ROS is poorly understood. Nevertheless, increased mitochondrial ROS levels are associated with cisplatin's and oxaliplatin's neurotoxicity in painful CIPN.⁵⁰ As a strategy to abrogate platinum-induced elevated mitochondrial ROS, antioxidants were investigated as potential therapeutics. However, no clinically relevant protective effects were found.^54,97 We investigated the role of MTM to reverse oxaliplatin-induced hypersensitivity through mitochondrial ROS.⁷⁰ As shown in Fig. 11, oxaliplatin-induced increases in ROS in sensory neuronal cultures harvested from mice on day 10 correlated with ongoing cold and mechanical hypersensitivity. By contrast, cotreatment with MTM in vivo not only reversed behavioral hypersensitivity but decreased sensory neuronal ROS levels to baseline on day 10.

Our differential expression and enrichment analyses identified increased expression of 4 (ND1, ND3, ND4, ND4L) of 7 components of the mitochondrial DNA encoded subunits of Complex 1—a major source of ROS production of the ETC.⁸⁹ We hypothesize that oxaliplatin-mediated dysfunction of components of complex-1 increase ROS production that is reversed, in part, through MTM-mediated transcriptional regulation of the ETC. This hypothesis is plausible because Complex-1 inhibitors such as rotenone enhance oxaliplatin-induced allodynia¹⁰⁴ and genetic targeting of Complex-1 subunit NDUFA1 produces an increase in ROS.^82,89 Conversely, a point mutation of a component of Complex 1 ND6 (P25L) blocks pathological ROS production.¹⁰⁷

Certain mitochondrial genes are regulated by nuclear transcription factors,⁶¹ including Sp4, that bigenomically regulates the transcription of all mitochondrial and nuclear encoded genes of cytochrome c oxidase (COX), the terminal enzyme of the ETC.⁴⁷ Whether an increase in COX1 or a decrease in Cox4i1 observed following MTM rescue (Supplementary Figure 5, available at http://links.lww.com/PAIN/B865) contributes to the reversal of DRG ROS levels will require further investigation.⁷The ability of MTM to reverse oxaliplatin-induced increases in TRPM8 expression and its normalization of mitochondrial ROS through modulation of ETC transcription may represent an effective strategy for the relief of painful CIPN as proposed in (Fig. 13).⁶⁶

4.4. Limitations and conclusions

Several limitations warrant consideration. Although we provide evidence of MTM's action on the peripheral (nociceptive) system, MTM is known to act on the central nervous system.³² Therefore, MTM reversal of pain behaviors may involve the regulation of genes and gene networks throughout both the central and peripheral nervous systems. In addition to modulation of mitochondrial ROS production, MTM-directed regulation of alternative splicing of genes enriched for axonal transport represents another potential therapeutic mechanism (Fig. 12). Therefore, other genes and gene families may serve to amplify the reversal of painful hypersensitive states. Traumatic nerve injury and CIPN may share common mechanisms such as dependence on Sp1-like factors, which can be blocked by the action of MTM.³⁷ Although the identification of an FDA-approved anticancer drug with the ability to reverse painful hypersensitivities is clinically appealing, it needs to be approached with caution given its known toxicity.⁹⁵ Fortunately, inhibition of Sp1-like transcription does not appear to be a driver for the systemic toxicity reported with the administration of MTM.

The ongoing pain induced by platinum neurotoxicity and observed in patients with CIPN is not fixed, but is being sustained in part, by ongoing neuroplastic changes in pain transduction and mitochondria function that can be reversed through targeted transcriptional therapeutics.