The Development and Maintenance of Paclitaxel-induced Neuropathic Pain Require Activation of the Sphingosine 1-Phosphate Receptor Subtype 1*

Background: Chemotherapy-induced peripheral neuropathy (CIPN) is a critical dose-limiting side effect of many chemotherapeutic agents, including paclitaxel. Results: Spinal activation of the S1P-to-S1PR1 axis contributes to the development and maintenance of paclitaxel-induced neuropathic pain through enhanced neuroinflammatory processes. Conclusion: Inhibition of S1PR1 blocks and reverses paclitaxel-induced neuropathic pain without interfering with anticancer effects. Significance: Targeting the S1PR1 signaling pathway could be an effective approach for the treatment of CIPN. The ceramide-sphingosine 1-phosphate (S1P) rheostat is important in regulating cell fate. Several chemotherapeutic agents, including paclitaxel (Taxol), involve pro-apoptotic ceramide in their anticancer effects. The ceramide-to-S1P pathway is also implicated in the development of pain, raising the intriguing possibility that these sphingolipids may contribute to chemotherapy-induced painful peripheral neuropathy, which can be a critical dose-limiting side effect of many widely used chemotherapeutic agents. We demonstrate that the development of paclitaxel-induced neuropathic pain was associated with ceramide and S1P formation in the spinal dorsal horn that corresponded with the engagement of S1P receptor subtype 1 (S1PR1)-dependent neuroinflammatory processes as follows: activation of redox-sensitive transcription factors (NFκB) and MAPKs (ERK and p38) as well as enhanced formation of pro-inflammatory and neuroexcitatory cytokines (TNF-α and IL-1β). Intrathecal delivery of the S1PR1 antagonist W146 reduced these neuroinflammatory processes but increased IL-10 and IL-4, potent anti-inflammatory/neuroprotective cytokines. Additionally, spinal W146 reversed established neuropathic pain. Noteworthy, systemic administration of the S1PR1 modulator FTY720 (Food and Drug Administration-approved for multiple sclerosis) attenuated the activation of these neuroinflammatory processes and abrogated neuropathic pain without altering anticancer properties of paclitaxel and with beneficial effects extended to oxaliplatin. Similar effects were observed with other structurally and chemically unrelated S1PR1 modulators (ponesimod and CYM-5442) and S1PR1 antagonists (NIBR-14/15) but not S1PR1 agonists (SEW2871). Our findings identify for the first time the S1P/S1PR1 axis as a promising molecular and therapeutic target in chemotherapy-induced painful peripheral neuropathy, establish a mechanistic insight into the biomolecular signaling pathways, and provide the rationale for the clinical evaluation of FTY720 in chronic pain patients.

The ceramide-sphingosine 1-phosphate (S1P) rheostat is important in regulating cell fate. Several chemotherapeutic agents, including paclitaxel (Taxol), involve pro-apoptotic ceramide in their anticancer effects. The ceramide-to-S1P pathway is also implicated in the development of pain, raising the intriguing possibility that these sphingolipids may contribute to chemotherapy-induced painful peripheral neuropathy, which can be a critical dose-limiting side effect of many widely used chemotherapeutic agents. We demonstrate that the development of paclitaxel-induced neuropathic pain was associated with ceramide and S1P formation in the spinal dorsal horn that corresponded with the engagement of S1P receptor subtype 1 (S1PR 1 )dependent neuroinflammatory processes as follows: activation of redox-sensitive transcription factors (NFB) and MAPKs (ERK and p38) as well as enhanced formation of pro-inflammatory and neuroexcitatory cytokines (TNF-␣ and IL-1␤). Intrathecal delivery of the S1PR 1 antagonist W146 reduced these neuroinflammatory processes but increased IL-10 and IL-4, potent anti-inflammatory/ neuroprotective cytokines. Additionally, spinal W146 reversed established neuropathic pain. Noteworthy, systemic administration of the S1PR 1 modulator FTY720 (Food and Drug Administration-approved for multiple sclerosis) attenuated the activation of these neuroinflammatory processes and abrogated neuropathic pain without altering anticancer properties of paclitaxel and with beneficial effects extended to oxaliplatin. Similar effects were observed with other structurally and chemically unrelated S1PR 1 modulators (ponesimod and CYM-5442) and S1PR 1 antagonists (NIBR-14/15) but not S1PR 1 agonists (SEW2871). Our findings identify for the first time the S1P/S1PR 1 axis as a promising molecular and therapeutic target in chemotherapy-induced painful peripheral neuropathy, establish a mechanistic insight into the biomolecular signaling pathways, and provide the rationale for the clinical evaluation of FTY720 in chronic pain patients.
Paclitaxel (Taxol) is a widely used chemotherapeutic agent indicated for treating breast, ovarian, non-small cell lung carcinomas, and Kaposi sarcoma. Unfortunately, the dose-limiting side effect and leading cause of discontinuation of this highly efficacious anticancer drug is peripheral neuropathy accompanied by chronic neuropathic pain (CIPN) 2 that may resolve within weeks or could persist for years after drug termination (1). The clinical management of these patients becomes difficult as current pain drugs are marginally effective and display unacceptable side effects (1). Identification of target-directed therapeutic approaches based on mechanistic insight is of paramount significance as the estimated incidence of CIPN is 30 -90% of patients treated with taxanes and combinational chemotherapies (1). * This work was supported by the Leukemia and Lymphoma Society Transla-As a desirable anticancer mechanism, paclitaxel can activate the sphingolipid pathway to produce the potent pro-apoptotic ceramide, which can induce apoptosis in a variety of neoplasms (2). Ceramide is then hydrolyzed by ceramidases to sphingosine and then phosphorylated by sphingosine kinases (SphK1 and -2) to produce sphingosine 1-phosphate (S1P) (3). S1P levels are further regulated by its dephosphorylation by S1P phosphatases and lipid phosphate phosphatases or cleaved by S1P lyase (3). Once released, S1P initiates signaling through a family of five cognate G protein-coupled receptors (S1PR [1][2][3][4][5] leading to various cellular responses (3,4). In contrast to ceramide, S1P is a potent anti-apoptotic sphingolipid; it is hypothesized that the ceramide/S1P rheostat plays a critical role in regulating cancer cell fate with elevated levels of ceramide inducing cell death, whereas elevated levels of S1P lead to survival and proliferation (2). To this end, therapeutic strategies to treat various types of cancer by increasing the levels of ceramide (e.g. with ceramidase inhibitors), reducing S1P bioavailability (e.g. with SphK inhibitors), or attenuating S1P/S1PR 1 signaling with anti-S1P antibodies or S1P 1 modulators, such as FTY720, are an active area of investigation and are moving forward as novel anticancer agents (2,5). FTY720 (fingolimod/Gilenya) is the first orally available agent approved by the Food and Drug Administration for the treatment of relapsing-remitting multiple sclerosis (MS) (6), an autoimmune disorder characterized by neuroinflammation in the central nervous system (CNS), demyelination, and neurodegeneration. In addition to their well established roles in inflammation and cancer, ceramide and S1P are emerging as important modulators in the development of peripheral and central sensitization associated with enhanced pain processing (7,8). For example, peripheral ceramide and S1P (acting via S1PR 1 ) increase the excitability of small diameter sensory neurons and contribute to nerve growth factor-induced sensitization of sensory neurons (9 -13). Intraplantar injection of ceramide (14 -16), S1P, or S1PR 1 agonists (15,17) in rats or mice evoke profound mechano-hypersensitivity via activation of the S1P 1 receptors and subsequent formation of a peripheral inflammatory response (14,15,18). In the CNS, these sphingolipids also appear to be important mediators in the development of spinal sensitization associated with increased nociceptive input. For example, ceramide/S1P levels are elevated in the spinal dorsal horn of neuropathic animals (19) and in morphine-tolerant rats where they contribute to the development of central sensitization by hyperactivating glial cells and increasing the production of pro-inflammatory/neuroexcitatory cytokines and nitro-oxidative species (20,21). Furthermore, Yan and Weng (22) recently reported that IL-1␤ generated in the spinal cord of neuropathic rats contributes to central sensitization; the activity of presynaptic NMDA receptors is enhanced by activation of the sphingomyelinase/ceramide signaling pathway that results in increased glutamate release from the primary afferent terminals. Whereas the underlying causative mechanisms of CIPN following paclitaxel are multifactorial and include neuropathological changes in the periphery (23), prominent neuropathological changes in the CNS have been documented to contribute through the development of neuroinflammation and dysregulation of neuroglia communication in the spinal cord (24). We hypothesize that if paclitaxel-in-duced neuropathic pain is dependent on the activation of the S1P/S1PR 1 axis, then anti-S1PR 1 -targeted approaches should provide an effective means to mitigate CIPN without interfering with anticancer effects. Indeed, our results identify for the first time S1PR 1 as a promising molecular target in CIPN, establish a mechanistic link into potential biomolecular signaling pathways, and provide the foundation to consider "fast-track" clinical use of FTY720 as a therapeutic agent in CIPN patients. Animal Models of Chemotherapy-induced Neuropathic Pain-Paclitaxel (Parenta Pharmaceuticals, Yardley, PA) or its vehicle (Cremophor EL and 95% EtOH in 1:1 ratio, Sigma) was injected intraperitoneally in rats on 4 alternate days (D0, 2, 4, and 6; cumulative dose of 8 mg/kg) (25). Oxaliplatin (Oncology Supply, Dothan, AL) or its vehicle (5% dextrose) was injected intraperitoneally in rats on 5 consecutive days (D0 -4) for a final cumulative dose of 10 mg/kg. This low dose paradigm does not cause kidney injury (26). S1PR 1 -induced Mechano-hypersensitivity-Chronic intrathecal cannulae were placed in rats using the L5/L6 lumbar approach, as described previously (27). To induce mechanoallodynia and mechano-hyperalgesia, SEW2871 (10 l) or its vehicle (2% DMSO in saline) was administered via these cannulae and flushed with sterile physiological saline (10 l). Behavior was assessed before any treatment (baseline) and then periodically after drug administration for 2-5 h.
Osmotic Pump Implantation-On D16, rats were lightly anesthetized with isoflurane (3% in 100% O 2 ), and their backs were shaved and scrubbed with Nolvasan. An incision was made in the interscapular region for subcutaneous implantation of primed osmotic minipumps (Alzet 2001; Alza) that infused 1 l/h over a 7-day period. Minipumps were filled according to the manufacturer's specifications with FTY720, NIBR-14, CYM5442, or their vehicle, 100% DMSO. Immediately following surgery, FTY720-treated rats were injected with a loading intraperitoneal dose of 0.03 mg/kg FTY720 or its vehicle (2% DMSO), and the minipump was allowed to deliver FTY720 (0.03 and 0.1 mg/kg/day) or vehicle for 6 days. Likewise, NIBR-14-treated rats were given a loading intraperitoneal dose of 0.3 mg/kg/day NIBR-14 or vehicle (2% DMSO), and the minipump was allowed to deliver NIBR-14 (1 or 3 mg/kg/day) or vehicle for 6 days. For CYM-5442-treated rats, a loading dose of 3 mg/kg CYM-5442 or vehicle (5% DMSO) was given intraperitoneally, and the minipump was allowed to deliver CYM-5442 (3 mg/kg/day) or vehicle for 6 days.
Behavior Testing-Assessment of behavior was always started in the morning. Rats were randomized into treatment groups before any behavioral assessment was performed. Behavioral testing was done prior to any drug administration (D0 or baseline) and then subsequently at selected time points. For mechano-allodynia, rats were placed in elevated Plexiglas chambers (28X40X35-cm) upon a wire mesh floor and allowed to acclimate for 15 min prior to measuring the mechanical paw withdrawal thresholds, grams (PWT (g)), using calibrated von Frey filaments (Stoelting, Wood Dale, IL, ranging from 3.61 (0.407 g) to 5.46 (26 g) bending force) according to the "up-anddown" method (29) or with an electronic version of the von Frey test (dynamic plantar aesthesiometer, model 37450, Ugo Basile, Geminio, Italy) with a cutoff set at 50 g. The development of mechano-allodynia is evidenced by a significant (p Ͻ 0.05) reduction in mechanical mean absolute PWT (g) at forces that failed to elicit withdrawal responses before treatment (baseline or D0). For mechanical hyperalgesia, PWTs (g) were measured by the Randall and Sellitto paw pressure test (30) using a Ugo-Basile analgesiometer (model 37215) that applies a linearly increasing mechanical force to the dorsum of the rat's hind paw. The nociceptive threshold was defined as the force (gram) at which the rat withdrew its paw (cutoff set at 250 g). Animals receiving chemotherapeutic agents in the presence or absence of the experimental test substances did not display signs of toxicity, i.e. they exhibited normal posture, grooming, locomotor behavior, hair coat was normal, no signs of piloerection or ocular porphyrin discharge, and they gained body weight normally and comparable with vehicle-treated rats. If testing coincided with a day when rats received test substance, behavioral measurements were always taken before the injection of the test substance.
Tail Flick-The tail flick test, which measures the latency (in seconds) of tail withdrawal from a noxious radiant heat source (Ugo Basile; model number 37360), was used to measure acute thermal nociceptive sensitivity in rats with baseline latencies of 2-5 s and a cutoff time of 15 s to prevent tissue injury (31). Tail flick latencies were taken before and at 15, 30, 60, and 90 min after injection of test compounds.
Tissue for Biochemical Analysis-For all biochemical analyses, rats were sacrificed at a time of peak mechano-hypersensitivity (D16). The lower lumbar enlargement of the spinal cord (L4 to L6) was harvested, flash-frozen in liquid nitrogen, and kept at Ϫ80°C until the appropriate assay could be performed.
Sphingomyelinase Activity-Spinal cord tissues were homogenized in specific buffers, as described previously (32). Following homogenization, samples were preincubated with sphingomyelin for 30 min at 37°C before the sphingomyelinase activity was measured for 1.5 h using an Amplex red sphingomyelinase assay kit (Molecular Probes, Eugene, OR) in a fluorescence microplate reader. The sphingomyelinase activity was expressed as milliunits/s and normalized by protein concentration (g/ml). Hydrogen peroxide and purified sphingomyelinase were used as positive controls.
Serine Palmitoyltransferase (SPT) Activity-SPT activity was determined by measuring the incorporation of [ 3 H]serine into 3-ketosphinganine following a previously described method (33). SPT activity was measured by number of counts/min and normalized by protein concentration (g/ml).
Sphingosine Kinase Activity-Spinal cord tissue was homogenized in SK1 buffer (containing 20 mM Tris-HCl, pH 7.4, 1 mM EDTA, 0.5 mM deoxypyridoxine, 15 mM NaF, 1 mM ␤-mercaptoethanol, 1 mM sodium orthovanadate, 40 mM ␤-glycerophosphate, 0.4 mM phenylmethylsulfonyl fluoride, 10% glycerol, 0.5% Triton X-100, and a complete protease inhibitor mixture) using a rotor homogenizer (61). After brief sonication and quantitation of protein concentration by the BCA method, 30 g of protein was incubated in 90 l of reaction mixture containing sphingosine (50 M, delivered in 4 mg/ml fatty acid-free bovine serum albumin), [␥-32 P]ATP (5 Ci, 1 mM dissolved in 10 mM MgCl 2 ), and SK1 buffer for 30 min at 37°C. The reaction was terminated by the addition of 10 l of 1 N HCl and 400 l of chloroform/methanol/HCl (100:200:1, v/v/v). Subsequently, 120 l of chloroform and 120 l of 2 M KCl were added, and samples were centrifuged at 3000 ϫ g for 5 min. The organic phase (200 l) was transferred to new glass tubes and dried. Samples were resuspended in chloroform/methanol/HCl (100: 100:1, v/v/v). Lipids were then resolved on silica thin layer chromatography plates using 1-butanol/methanol/acetic acid/water (8:2:1:2, v/v/v/v) as the solvent system and visualized by autoradiography. The radioactive spots corresponding to S1P were scraped from the plates and counted for radioactivity. Background values were determined in negative controls in which sphingosine was not added to the reaction mixture.
Immunofluorescence-After behavioral measurements, rats were anesthetized with ketamine/xylazine and transcardially perfused with phosphate-buffered saline (PBS, pH 7.4) followed by 10% buffered neutral formalin. Spinal cord was harvested and post-fixed in the same fixative for 16 h at 4°C. After rinsing in PBS, the tissue was infiltrated with 30% (w/v) sucrose in PBS at 4°C for 48 h, rinsed again in PBS, transferred to OCT, and frozen on dry ice. Transverse sections (20 m) were cut in a cryostat, collected on gelatin-coated glass microscope slides,

S1PR 1 Involvement in Paclitaxel-induced Neuropathic Pain
air-dried overnight, and stored at Ϫ20°C. Spinal cord sections were blocked (10% normal goat serum, 2% bovine serum albumin, 0.2% Triton X-100 in PBS) and then immunolabeled as described previously (20,34,35) using a well characterized primary antibody, rabbit IgG polyclonal anti-ceramide (36) (1:50, incubated 18 h at 4°C), followed by several PBS washes and incubation (2 h room temperature in the dark) with a goat antirabbit IgG antibody conjugated to Alexa Fluor 568 (1:250, Invitrogen). The coverslips were mounted with Fluorogel II containing DAPI (Electron Microscopy Sciences, Hatfield, PA) and photographed with an Olympus FV1000 MPE confocal microscope (multiline argon lasers with excitation at 405 and 543 nm) using a 10ϫ objective (UPLSAPO; 0.4 NA) for regional fluorescence intensity image analysis and with 60ϫ oil immersion objective (PLAPON; 1.42 NA) and 2.4ϫ optical zoom (0.1 m pixel dimensions in the X-Y plane and the pinhole set at 1 Airy unit) for higher magnification images. Images were acquired within the dynamic range of the microscope (i.e. no pixel intensity values of 0 or 255 in an 8-bit image). Sections treated with rabbit IgG at equivalent concentrations to primary antibodies were used as controls yielding only nonspecific background fluorescence. The fluorescence intensity of immunolabeled ceramide was calculated as reported previously (34,35,37) in selected regions of the lumbar spinal cord. The mean fluorescence intensity (MFI) was calculated by Equation 1, where i is the mean gray value; pp is the positive pixel area, and p is total pixel area. Image analysis was performed using the freeware program ImageJ (version 1.43, National Institutes of Health) (38). Images received background threshold corrections prior to analysis using the automated ImageJ intermode histogram-based threshold function. The superficial dorsal horns (laminae I and II), dorsal horns, and ventral horns at the L4, L5, and L6 levels were outlined on images bilaterally using the ImageJ region of interest tool. The borders of these regions were determined and confirmed using cresyl violet-stained sections of regions adjacent to immunolabeled sections and an atlas (39). There were no significant differences bilaterally, so MFI was calculated as a combined value and reported as fold change compared with the vehicle group. Data are expressed as arbitrary units.
ELISA or Multiplex Assay for Cytokines-The levels of cytokines in spinal cord lysates were either assessed using commercially available ELISA kits (R&D Systems) or by using a commercially available magnetic multiplex cytokine kit (Bio-Rad). Samples were processed according to the manufacturer's protocol.
survivability ϭ ͑ A 560-570 nm of chemotherapeutic vehicle ϩ FTY720͒/͑mean A 560-570 nm of the naive control wells͒ ϫ 100 The LD 50 of each chemotherapeutic agent ϩ FTY720 or its vehicle was calculated using three-parameter nonlinear analysis using Equation 3. The top and bottom plateaus were constrained using GraphPad Prism Version 5.03 (GraphPad Software, Inc.). S1P ELISA-Spinal cord tissue was homogenized. The level of S1P was measured in spinal cord lysates using commercially available ELISA kits (Echelon Biosciences, Salt Lake City, UT) according to the manufacturer's protocol.
Leukocyte Counts-Six days following osmotic pump implantation, whole blood was collected via intracardiac puncture into K 3 EDTA Vacutainer tubes and sent to Advanced Veterinary Labs (St. Louis) for a complete blood count panel with white blood cell differential.
Statistical Analysis-Data are expressed as mean Ϯ S.D. for n animals. Differences in behavioral data from the full time course studies were analyzed by two-way repeated measures ANOVA with Bonferroni comparisons. All other biochemical data were analyzed by one-way ANOVA with Dunnett's comparisons. Differences in fluorescence intensity and single comparison behavioral tests as noted were analyzed using the unpaired Student's t test. Significant differences were defined at p Ͻ 0.05.

S1P/S1PR 1 Signaling Pathway Is Activated in Spinal Cord
Dorsal Horn in Response to Paclitaxel-Consistent with previous studies (45,46), paclitaxel produced a time-dependent development of mechano-allodynia and mechano-hyperalgesia (i.e. mechano-hypersensitivity) (Fig. 1, B and C) as evidenced by a decrease in PWTs (g) by D12 (onset), which peaked by D16, and plateaued throughout our observation period (D25; times relative to first injection). The delay between the last exposure to paclitaxel and onset of mechano-hypersensitivity mimics the clinical "coasting" phenomenon described in patients (1). Ceramide is generated by enzymatic hydrolysis of sphingomyelin by sphingomyelinases (sphingomyelin pathway) and from de novo synthesis by SPT and ceramide synthase (de novo pathway) (47). Ceramide, in turn, is hydrolyzed to sphingosine, which is phosphorylated by SphK1 and -2 to form S1P (47). SphK1 activity is the major determinant of S1P levels in inflammatory diseases and is known to be activated by several cytokines, in particular TNF-␣ (48). As shown in Fig. 2, the peak of paclitaxel-induced mechano-hypersensitivity was associated with increased SPT and neutral/acidic sphingomyelinase activities in lumbar spinal cord (L4 -6) (Fig. 2, A and B) as well as enhanced bilateral ceramide immunolabeling within laminae I and II of the L4 -6 superficial dorsal horn (discrete pain-processing region that undergoes neuroinflammatory changes during chronic pain Fig. 2, C-E) (49, 50)). The intensity of spinal ceramide immunofluorescence (MFI) was compared between vehicle and paclitaxel groups using automated image analysis (38) of confocal microscopy images by calculating the product of the mean gray value and percentage of positive pixel area for each region of interest (e.g. superficial dorsal horn). Compared with vehicle, paclitaxel was associated with increased MFI in the superficial dorsal horn (Fig. 2C) but not the ventral horn (data not shown). In addition to the activation of the enzymes involved in the biosynthesis of ceramide, paclitaxel also led to increased SphK1 activity and S1P formation in spinal cord (Fig. 3, A and B). Daily (D0 -D15) intrathecal injections of a commonly used and well characterized SphK1/2 inhibitor, SK-I (0.3 M/d) (51), given at a dose previously shown to block increased spinal production of S1P during central sensitization evoked by prolonged use of opioids (20) blocked the following: 1) the increased activation of SphK1 (Fig. 3A); 2) the increased production of S1P (Fig. 3B) in spinal cord; and 3) the development of mechano-allodynia and mechano-hyperalgesia (Fig. 3, C and D). These results establish the role of the sphingosine kinase to S1P signaling pathway in spinal cord during CIPN (Fig. 3E).

S1PR 1 Involvement in Paclitaxel-induced Neuropathic Pain
attenuated paclitaxel-induced mechano-allodynia and mechanohyperalgesia (Fig. 1, B and C) in a dose-dependent manner; doses were chosen from previous studies (15). Although chemotherapy is completed within a few days, we continued dosing until peak mechano-allodynia and mechano-hyperalgesia were achieved because the delay to symptom onset introduces uncertainty regarding the exact initiation of relevant pathological processes making continued treatment prudent. When W146 treatment was discontinued on D15, mechano-hypersensitivity did not emerge through D25 (Fig. 1, B and C). The ED 50 (effective dose providing 50% effect) of W146 for mechano-allodynia and mechano-hyperalgesia at D25 was 0.4 nmol/day (95% CI: 0.2-0.7) and 0.8 nmol/day (95% CI: 0.5-1. 3).
Because all drugs tested had comparable effects on mechanoallodynia and mechano-hyperalgesia, we will only show the latter for simplicity in the subsequent figures, and we refer to it as mechano-hypersensitivity in the text.
Intrathecal Injection of Selective S1PR 1 Agonist SEW2871 Causes Mechano-hypersensitivity-Taken together, these results suggest that the formation of S1P in spinal cord in response to paclitaxel engages the S1P 1 receptor to evoke mechano-hypersensitivity. Therefore, does spinal activation of S1PR 1 with exogenous application of an S1PR 1 agonist mimic these behavioral outcomes? In normal rats with previously implanted i.th. catheters, injection of the selective S1PR 1 agonist, SEW2871 (0.8 nmol) (53), led to a time-dependent development of mechano-hypersensitivity that peaked by 2 h (Fig. 4). These effects were blocked with W146, but not W140 (2.2 nmol, i.th.; given 30 min before SEW2871 treatment; Fig. 4A), confirming an S1PR 1 mechanism of action. These results suggest that activation of S1PR 1 in spinal cord by an exogenous S1PR 1 agonist can recapitulate behavioral features associated with paclitaxel administration Intrathecal Injection of S1PR 1 Modulators Block SEW2871induced Mechano-hypersensitivity-Binding of SEW2871 (or the endogenous ligand S1P) to the S1P 1 receptor (53) allows S1PR 1 recycling back to the plasma membrane after internalization. In contrast, other S1PR 1 agonists potently induce irreversible down-regulation of S1PR 1 resulting in the ubiquitinylation and proteosomal degradation of the receptor, yielding a net decrease in S1PR 1 expression at the plasma membrane (54). S1PR 1 agonists in this class referred to as S1PR 1 modulators include the following: FTY720 (a sphingosine analog that is phosphorylated in vivo by SphK2 to produce its bioactive Sisomeric monophosphate ester FTY720-P, which is an S1P

S1PR 1 Involvement in Paclitaxel-induced Neuropathic Pain
mimetic capable of binding to all S1PRs except S1PR 2 (6)); CYM-5442 (55); and ponesimod (28). These S1PR 1 modulators therefore act as functional antagonists (unlike SEW2871 or S1P) at S1PR 1 to block S1P/S1PR 1 signaling and exert effects similar to those observed with other S1PR 1 antagonists (56). This raises the following question. Do these S1PR 1 modulators block the sensitizing effects of SEW2871? Intrathecal injection of FTY720 or CYM-5442 (0.8 nmol) 30 min before SEW2871 blocked its ability to elicit mechano-hyperalgesia and mechano-allodynia (Fig. 4, B and C). Similar results were obtained when FTY720 or CYM-5442 was given systemically consistent with their high CNS permeability (57,58) and were corroborated with NIBR-14 (3 mg/kg; Fig. 4D). NIBR-14 is a methyl ester pro-drug that is rapidly hydrolyzed in vivo to its corresponding carboxylic acid (NIBR-15) that acts as a potent and selective S1PR 1 antagonist (40). The suppressive effects of FTY720 and CYM-5442 therefore mimic those observed with other S1PR 1 antagonists and support their role as functional antagonists for their mechanism of action at S1PR 1 .
Systemic Administration of S1PR 1 Modulators and S1PR 1 Antagonists Block the Development of Paclitaxel-induced Neuropathic Pain-Results gathered so far define the important contribution of the S1P/S1PR 1 axis in the spinal cord to the development of CIPN; blocking this signaling pathway mitigates CIPN. This raises the question as to whether these anti-S1PR 1 agents would be effective when given systemically, thereby paralleling potential clinical conditions. Systemic FTY720 (0.003, 0.01, and 0.03 mg/kg/day; i.p), CYM-5442, ponesimod, or NIBR-14 (all at 0.3, 1, and 3 mg/kg/day; p.o.) blocked mechano-hypersensitivity in a dose-dependent fashion (Fig. 5, A-D). ED 50 values for mechano-allodynia and mechano-hyperalgesia on D25 for each agent are reported in Table 1. We also examined whether restricting the dosing regimen of the anti-S1PR 1 agents to coincide with only the paclitaxel treatment would afford protection as dosing patients only when they receive the chemotherapeutic agent would be a preferred regimen. FTY720 (0.03 mg/kg/day), CYM-5442, ponesimod, or NIBR-14/-15 (all at 3 mg/kg/day), when given orally at

Contribution of Spinal NFB and MAPKs to Paclitaxel-induced
Neuropathic Pain-We previously reported that the development of paclitaxel-induced neuropathic pain is associated with increased formation of TNF-␣ and IL-1␤ in the spinal cord and that inhibiting these pro-inflammatory cytokines blocks CIPN (45). We now extend these findings and demonstrate that at peak mechano-hypersensitivity on D16, additional canonical pro-inflammatory signaling pathways are activated. These include the activation of the NFB signaling pathway as indicated by p65 phosphorylation at serine 536 (Fig. 6A), nuclear translocation of the p65 subunit of NFB (Fig. 6B), and the activation of the MAPK signaling pathway as evidenced by the increased phosphorylation of ERK1/2 (Fig. 6C) and p38 (Fig.  6D). Because the function of each of these signaling pathways in the context of CIPN is not known, selective inhibitors for each of these pathways were used to establish their possible contribution. The selective inhibitor of NFB, SN50, is a small peptide that binds to and prevents the translocation of NFB to the nucleus (59), whereas the MAPK inhibitors U0126 and SB203580 prevent MAPK/ERK1/2 (MEK1/2) phosphorylation of ERK (60) and phospho-p38 catalytic activity (61), respectively. As can be seen in Fig. 6, i.th. delivery of SN50 (2 ng/kg/ day) (62), U0126 (1 g/kg/day) (63), and SB203580 (10 g/kg/ day) (62, 64) attenuated their respective pathways (Fig. 6, A-D), as well as the development of mechano-hypersensitivity (Fig. 6,  E-G).
Spinal Neuroinflammation Is Blocked by Anti-S1PR 1 Approaches-Inhibition of CIPN by W146 (i.th.; 2.2 nmol/day) at peak mechano-hypersensitivity was associated with inhibition of the activation of NFB (p65 phosphorylation at serine 536 and nuclear translocation of the p65 subunit of NFB; Fig.  7, A and B) and MAPKs (reflected as phosphorylation of

S1PR 1 Involvement in Paclitaxel-induced Neuropathic Pain
ERK1/2 and p38 (Fig. 7, D and E) as well as the overproduction of TNF-␣ and IL-1␤ (Fig. 7F). Conversely, W146 increased formation of anti-inflammatory cytokines (IL-10 and IL-4) (Fig.  7F). Supporting a common anti-S1PR 1 mechanism of action in the modulation of spinal neuroinflammation in the genesis of CIPN, systemic FTY720, CYM-5442, ponesimod, or NIBR-14 given at the highest dose attenuated TNF-␣ and IL-1␤ but increased IL-10 and IL-4 (Fig. 8, A-D). S1PR 1 Modulators and S1PR 1 Antagonists, but Not S1PR 1 Agonists, Reverse Established Paclitaxel-induced Neuropathic Pain in a Naloxone-independent Manner-We next sought to define whether activation of spinal S1PR 1 contributes to the maintenance of CIPN. Intrathecal administration of W146 on D25, but not W140, caused a rapid reversal of mechano-hypersensitivity, which was maximal within 2 h (Fig. 9A). In contrast, SEW2871 (i.th.; 0.8 nmol; Fig. 9A) had no effect, reinforcing the notion that S1PR 1 antagonism and not agonism provides anti-nociception. In added support, the administration of a single oral dose of FTY720 (0.1 mg/kg) or NIBR-14 (3 mg/kg) on D25 reversed mechano-hypersensitivity and this was not blocked by a high dose (intraperitoneal) of the nonselective opioid receptor antagonist naloxone (2 mg/kg), thus excluding the contribution of an endogenous opioid pathway (Fig. 9, B and C). It is noteworthy that a subcutaneous mini-pump infusion of FYT720 (0.1 mg/kg/day for 6 days), CYM-5442, or NIBR-14 (both 3 mg/kg/day for 6 days) following an intraperitoneal loading dose completely reversed mechano-hypersensitivity (Fig. 10A) without any evident analgesic tolerance or alteration to the peripheral blood leukocyte differential (Table 2).
FTY720 Blocks Oxaliplatin-induced Neuropathic Pain-To examine whether the beneficial effects of an anti-S1PR 1 strategy can be extended to neuropathic pain resulting from other chemotherapeutic agents with different mechanisms of anticancer action, we used oxaliplatin, which is widely used for FIGURE 7. Inhibition of spinal neuroinflammation by the S1PR 1 antagonist W146. On D16 and when compared with vehicle (V-V, open bars), administration of paclitaxel (P-V, black bars) increases cytosolic phosphorylation of NFB p65 (A), nuclear translocation of NFB p65 (B), and phosphorylation of ERK1/2 (C) and p38 (D). These events were blocked by daily i.th. delivery (D0 -15) of W146 (2.2 nmol/day, gray bars). E, when compared with vehicle on D16, paclitaxel treatment also increased TNF-␣ and IL-1␤ production, which was blocked by W146; W146 increased the levels of IL-4 and IL-10. W146 alone had no effect in the spinal cord of vehicle-treated rats (light gray bars). Results are expressed as mean Ϯ S.D. for n ϭ 5-6 rats and analyzed by one-way ANOVA with Dunnett's comparisons. *, p Ͻ 0.05 versus V-V; †, p Ͻ 0.05 versus P-V. JULY 25, 2014 • VOLUME 289 • NUMBER 30 metastatic colon cancer and other gastrointestinal cancers. We decided to examine this question with FTY720 only because this drug is already used clinically and therefore provides an advantage over the other S1PR 1 antagonists/ modulators that are in preclinical discovery or clinical trials. Confirming our previous studies (65), oxaliplatin-induced neuropathic pain was significant by D11 (onset), peaked by D17, and plateaued throughout our observation period (D25). Mechano-hypersensitivity was attenuated in a dosedependent manner by FTY720 with ED 50 values reported in Table 1. Furthermore, and mimicking effects seen with paclitaxel, at the highest dose tested, FTY720 (0.01 mg/kg/day) blocked the increased production of TNF-␣ and IL-1␤, whereas the levels of anti-inflammatory cytokines IL-10 and IL-4 increased significantly (data not shown).

S1PR 1 Involvement in Paclitaxel-induced Neuropathic Pain
Lack of Effects on Acute Nociception-In contrast to morphine, which was used as a positive control at 5 mg/kg, the highest doses of FTY720 (3 mg/kg, intraperitoneally), CYM-5442, ponesimod, or NIBR-14 (3 mg/kg, oral gavage) tested had no effects on acute nociception as measured by tail flick latency (Fig. 10B).
FTY720 Does Not Alter Anticancer Activity in Vitro-The anticancer effects of FTY720 and other anti-S1P/S1PR 1 agents are well documented (2,5) and are therefore not expected to interfere with the chemotherapeutic effectiveness of paclitaxel or oxaliplatin. Based upon the reported pharmacokinetic studies in rats (6), the anticipated plasma levels of FTY720 at doses providing maximal blockade of CIPN are about 0.5-0.7 nM. When tested at doses at least 10-fold higher, FTY720 (10 nM) did not diminish the in vitro anticancer effects of paclitaxel (1-100 nM) or oxaliplatin (0.3-30 M) in human breast adenocarcinoma (SKBr3) or colon carcinoma (SW480) cells, respectively. FTY720 (10 nM) alone caused a 3-20% decrease in anticancer cell survival, which is not surprising ( Table 3).

DISCUSSION
Despite extensive research efforts, chronic neuropathic pain remains a largely unmet medical need, and novel therapies are needed. Several important findings have emerged from our study.
First, our results establish that the S1P-S1PR 1 axis is a critical determinant in the development and maintenance of paclitaxel-induced neuropathic pain and identify the following two viable therapeutics with potential for fast translational impact to patient care: FTY720, which is Food and Drug Administration-approved (6), and ponesimod, which is in phase 2 clinical trials (66). In addition to the clinically feasible prevention treatment paradigm, the fact that anti-S1PR 1 -targeted approaches can be used in an intervention paradigm allows therapy to be extended to those patients who have developed the neuropathy. It is reasonable to hypothesize that anti-S1PR 1 approaches may provide added therapeutic opportunities because these are proving useful in anticancer therapies as stand-alone drugs or as adjuncts to chemotherapeutic agents (2,5). Thus, drugs, such as FTY720, may enhance the anticancer effects of chemotherapeutic agents while minimizing the major dose-limiting toxicity neuropathic pain as shown in this study. Noteworthy, these beneficial effects were not restricted to paclitaxel only but also FIGURE 8. Systemic delivery of S1PR 1 modulators and S1PR 1 antagonists blocks paclitaxel-induced pro-inflammatory cytokine production in spinal cord.  extended to other chemotherapeutic agents, raising the exciting possibility of a more widespread impact to mitigate CIPN. Effects were naloxone-independent and therefore are not mediated by activation of an endogenous opioid system reducing the potential for the development of analgesic tolerance. Because the S1P/S1PR 1 axis is critically involved in the egress of T and B cells from lymphoid tissues (67), it was initially thought that the efficacy of FTY720 and other S1PR 1 functional antagonists was mediated by their ability to block the infiltration of autoreactive T cells into the CNS where they play an important role in the progression of MS (10). However, phase II clinical trials revealed that the peak effectiveness of FTY720 in the treatment of MS occurs at doses that are suboptimal for the induction of lymphopenia (6). These paradoxical human clinical data, coupled with the findings obtained in the CNS that S1P is produced by both neurons and glia expressing S1PR 1 , suggest that the beneficial effects of these agents may also be exerted by nonimmunocompetent cells in the CNS (68). It is known that FTY720 efficiently crosses the blood-brain barrier where it is phosphorylated by SphK2 (predominantly expressed in neurons and glial cells) (69) to yield FTY720-P (70). The CNS is therefore an important site for mechanism of action of not only FTY720 but also of other S1PR 1 modulators, including CYM-5442 (6,57,71). In our studies, the maximal efficacy regarding the reduction of mechano-hypersensitivity was detected with doses of FTY720 at least 10-fold lower than those shown to cause lymphopenia (6) or bradycardia (72) and for NIBR-14/15 to cause pulmonary vascular leakage (40). Furthermore, we did not observe any changes in differential white blood cell counts following a 6-day infusion of FTY720 that provided near-tomaximal inhibition of CIPN. Similar findings were reported by Coste et al. (73), where they demonstrated that FTY720 at doses devoid of immunosuppressive effects was capable of reversing the neuropathic pain produced by a chronic constriction injury of the sciatic nerve.

S1PR 1 Involvement in Paclitaxel-induced Neuropathic Pain
tivity resulting from the injection of formalin into the rat's hind paw (74). This lack of effect is consistent with the fact that SEW2871 is not a functional antagonist and agrees with our findings that SEW2871 does not reverse CIPN. In addition, our results support previous findings in various animal models for transplantation and autoimmune diseases, including MS and arthritis, where functional antagonism of S1PR 1 rather than persistent agonism/signaling at S1PR 1 was identified as the mechanism of action (56). Third, our results identify the spinal cord as one site of action for anti-S1PR 1 agents and inhibition of spinal neuroinflammation as a molecular target in their pharmacological activity. This parallels the documented CNS mechanisms of action of anti-S1PR 1 therapies in MS (71). Initial studies using FTY720 in animal models of MS revealed the importance of the S1P/S1PR 1 axis to neuroinflammatory processes in the CNS, as well as the protective actions of various S1PR 1 functional antagonists or selective S1PR 1 antagonists (56). In a rodent model of MS, FTY720 (75) or CYM-5442 (57) inhibits the S1P/S1PR 1 activation and signaling in astrocytes and the downstream formation of pro-inflammatory cytokines, including IL-1␤ and IL-6. Pharmacological and genetic manipulations reveal that these beneficial effects are exerted through S1PR 1 functional antagonism in astrocytes and neurons (56). Likewise, accumulating evidence implicates neuroinflammatory processes in the alteration of glia-neuronal communication in the dorsal horn of the spinal cord that is associated with paclitaxel-induced neuropathic pain. Specifically, hyperactivation of glial cells (45,76,77), enhanced TNF-␣ and IL-1␤ (45,78,79), and nitroxidative species production (45) as well as dysregulation of glutamate homeostasis (45,80,81) have been documented. A recent study revealed that increased activation of GSK-3␤ in spinal cord contributes to the development and maintenance of paclitaxelinduced neuropathic pain by activating astrocytes and causing the overproduction of IL-1␤ (79). It is, however, known that in addition to GSK-3␤, redox-activated transcription factors such as NFB and MAPKs regulate the levels of TNF-␣ and IL-1␤ (and vice versa) and contribute to the development of several different neuropathic pain states (82). However, to date their functional contribution to CIPN was not documented. Our results now reveal that these signaling pathways are activated in response to paclitaxel and contribute to CIPN because their inhibition attenuates mechano-hypersensitivity. Moreover, our findings that S1PR 1 antagonism blocks the activation of NFB and MAPKs and shifts the cytokine environment in the spinal cord from pro-inflammatory to anti-inflammatory suggest that inhibition of spinal neuroinflammation is an important component in their beneficial armamentarium. Noteworthy, similar enhancement in the expression of anti-inflammatory IL-10 following treatment with FTY720 has been described with activated dendritic cells (83) and in colitis (84). In addition to inhibition of neuroinflammatory processes in spinal cord, it is important to recognize that blocking neuronal S1PR 1 signaling may also contribute to the beneficial effects of anti-S1PR 1 approaches (8). For example, blocking S1PR 1 may also attenuate the increased neuronal excitability in spinal dorsal horn observed during paclitaxel-induced neuropathic pain (85).
It has been previously reported in other animal models that cytokines and nitroxidative species for example can activate SphK1 (48) and sphingomyelinases (47) as well as regulate the enzymatic activities of SPT, sphingomyelinases, and ceramidases (86 -88). Reciprocally, ceramide and S1P can increase the formation of cytokines, and nitroxidative species from glial cells (89) and can also regulate cellular redox homeostasis (87). Therefore, the ceramide/S1P pathway may not only be a target for the action of neuroinflammatory species, it may also serve as a trigger for their formation. Together, these nitroxidative-and cytokine-rich environments may synergize with S1P signaling to prolong and exacerbate CIPN by instituting feedback loops that sustain local neuroinflammatory processes in spinal cord.
Although our studies focused on changes in the spinal cord, we cannot exclude the likely contributions of S1P/S1PR 1 in the periphery. For example, S1PR 1 activation increases the excitability of rat sensory neurons, and blocking S1P/S1PR 1 signaling reduces their excitability to inflammatory stimuli (90,91). Therefore, further understanding of the S1P/S1PR 1 axis could provide the framework for clinical development of S1PR 1 antagonists/modulators in CIPN and perhaps other chronic neuropathic pain states such as diabetic neuropathy, whose underlying pathology shares commonalities with CIPN (92).