Vinorelbine causes a neuropathic pain-like state in mice via STING and MNK1 signaling associated with type I interferon induction

Summary Type I interferons (IFNs) increase the excitability of dorsal root ganglia (DRGs) neurons via MNK-eIF4E signaling to promote pain sensitization in mice. Activation of stimulator of interferon response cGAMP interactor 1 (STING) signaling is pivotal for type I IFN induction. We hypothesized that vinorelbine, a chemotherapeutic and activator of STING, would cause a neuropathic pain-like state in mice via STING signaling in DRG neurons associated with IFN production. Vinorelbine caused tactile allodynia and grimacing in wild-type (WT) mice and increased p-IRF3, type I IFNs, and p-eIF4E in peripheral nerves. Supporting our hypothesis, vinorelbine failed to induce IRF3-IFNs-MNK-eIF4E in StingGt/Gt mice and, subsequently, failed to cause pain. The vinorelbine-elicited increase of p-eIF4E was not observed in Mknk1−/− (MNK1 knockout) mice in peripheral nerves consistent with the attenuated pro-nociceptive effect of vinorelbine in these mice. Our findings show that activation of STING signaling in the periphery causes a neuropathic pain-like state through type I IFN signaling to DRG nociceptors.


INTRODUCTION
Vinorelbine is a chemotherapeutic agent that belongs to the semisynthetic vinca alkaloid family. 1 This drug is Food and Drug Administration (FDA) approved for non-small cell lung cancer (NSCLC) treatment. 2Vinorelbine is also used to treat metastatic breast carcinoma and has shown effectiveness in advanced forms of melanoma and renal cell carcinoma. 1NSCLC and breast cancer are estimated to be the top two cancer incidences in 2023 in the US. 3 Despite serving as a first-line treatment in advanced NSCLC, 4 vinorelbine has been reported to cause peripheral neuropathy [5][6][7][8][9][10] and induce pain in cancer patients, mostly at the tumor site or in areas innervated by nerves compressed by the tumor. 1 Vinorelbine mainly acts as a destabilizing agent of microtubules. 4,11Microtubule destabilizers have been shown to activate stimulator of interferon (IFN) genes (STING)-IFN regulatory factor (IRF) 3-dependent production of IFN-b and downstream IFN-stimulated genes (ISGs) in both immune and triple negative breast cancer cells. 12Type I IFNs are powerful monomeric cytokines that induce an antiviral response upon infection through the upregulation of molecules that antagonize viral replication. 13Regulation of type I IFNs by STING depends on the IRF family members IRF3/7. 14The expression of IRF7 in most cells is low, while IRF3 is ubiquitously expressed triggering the phosphorylation of IRF3 upon viral infection in nearly all cell types. 15Loss-of-function mutation in the IRF3 gene impairs IFN expression in patients, suggesting a critical role of this transcription factor in the antiviral IFN responses. 16,17reviously we showed that IFN-a and IFN-b activate IFN receptors (IFNRs) localized on dorsal root ganglion (DRG) nociceptors driving hyperexcitability and nociception in mice. 18,19Type I IFN-induced nociception in mice depends on mitogen-activated protein (MAP) kinase interacting kinase (MNK1) signaling to elongation initiation factor 4E (eIF4e).The phosphorylation of eIF4e catalyzed by MNK1 induces the translation regulation of mRNAs associated to antiviral response, inflammation, and cancer. 20Additional evidence from preclinical models 21 and clinical studies 22 supports the pro-nociceptive actions of type I IFNs.Recently published evidence shows that increased eIF4E-dependent mRNA translation resulting from the genetic ablation of 4E-BP1 increases the E3 ubiquitin ligase tripartite motif protein 32, TRIM32, which promotes type I IFN signaling and mechanical hypersensitivity in mice. 23In that paper, the authors acknowledge the possibility that TRIM32 is regulating STING ubiquitination in DRG, but this hypothesis needs to be tested.However, there is evidence that type I IFN induction mediated by TRIM32 is stimulated by targeting STING ubiquitination in HEK293 human cells 24 and with viral infections such as herpes simplex virus of cultured human epithelial and of mouse cornea. 25On the other hand, the role of STING in nociception is far from clear.Some findings have shown that STING-mediated induction of type I IFN signaling attenuates nociception via an action in the spinal dorsal horn. 26,27Other studies show a clear peripheral sensitization effect of STING signaling activation or the attenuation of nerve damage-induced Vinorelbine induces p-IRF3 in DRG neurons in mice p-IRF3 is the main downstream effector of STING inducing the transcription of Ifna1 and Ifnb1 genes. 34,35We found that vinorelbine induced an increase in the phosphorylated form of IRF3 in DRG neurons compared to vehicle (Figure 2A).A clear trend toward increased p-IRF3 was observed in all neuronal sizes, but the increase was significant only in small-and large-sized neurons (Figures 2A and 2B).Peripherin-, which is expressed only in unmyelinated cells, positive neurons had an enhancement of 3.5 times the p-IRF3 mean intensity value in DRGs from vinorelbine-treated versus vehicle mice (Figure 2C).These results are consistent with our western blot experiments that showed an increase in p-IRF3 in DRG due to vinorelbine treatment and suggest that this change is mostly neuronal.Our results are in line with experiments showing that STING is localized mainly in small diameter and larger neurons, including Ab neurons in the mouse DRG. 26

Vinorelbine-induced nociception and type I IFN induction require STING signaling
We sought to assess whether vinorelbine-induced nociceptive effects depend on STING activation.To this end, we took advantage of the mutant mouse strain, Goldenticket (Sting Gt/Gt ) also known as Tmem173 Gt , that fails to produce type I IFNs upon activation with c-di-guanosine monophosphate (GMP) or infection with pathogens. 36,37Sting Gt/Gt animals harbor a chemically induced (N-ethyl-N-nitrosourea) mutation in exon 6 of Sting resulting in an isoleucine to asparagine change in amino acid 199 located in the carboxyl-terminal end (CTT) of the protein, which is an important site to recruit TBK1, the kinase that phosphorylates IRF3; thus signaling via IRF3 is lost in Sting Gt/Gt mice.As expected, western blotting experiments revealed a profound reduction (78.9% in DRG and 76.9% in sciatic nerve) in the levels of STING in Sting Gt/Gt mice, which render STING barely detectable (Figure S3).
We assessed the effects of vinorelbine in Sting Gt/Gt mice and their WT controls.Because no sex differences were observed in previous experiments, male and female mice were used as part of the same cohort in these experiments.As expected, vinorelbine induced an increase in grimace score in WT mice from day 5 after first administration (day 0) and lasting until day 10.However, vinorelbine had no effect in Sting Gt/ Gt mice (Figure 3A).This genotype difference was also observed in the effect size (Figure 3B).Additionally, a reduction in the paw withdrawal threshold was observed in WT mice that was not observed with vinorelbine in Sting Gt/Gt mice (Figures 3C and 3D).These results indicate that STING activation is needed for the pro-nociceptive effects of vinorelbine in mice.
We performed western blots to further elucidate the changes of the canonical STING-IFN pathway underlying the blunted hypersensitivity of vinorelbine in Sting Gt/Gt mice.In the absence of a functional STING pathway, the administration of vinorelbine was not able to result in the increased production of IFN or p-IRF3 in DRG (Figures 3F-3H) and sciatic nerve (Figures 3K-3M), which aligns with the lack of hypersensitivity induced by the chemotherapeutic in Sting Gt/Gt mice.These results support the conclusion that vinorelbine induces type I IFN production via STING activation, consistent with the loss of nociceptive sensitization in Sting Gt/Gt mice.

Loss of type I IFN-MNK-eIF4E signaling reduces the vinorelbine-elicited pain in MNK1 knockout (KO) mice
9][40][41] Previously, we showed that type I IFNs stimulate extracellular signal-regulated kinase (ERK)/MAP kinase-MNK-eIF4E signaling in mouse DRG neurons. 18Since vinorelbine promotes the production of type I IFNs in DRG and sciatic nerve, we tested the hypothesis that vinorelbine-evoked nociceptive behaviors require MNK1 signaling.Vinorelbine was administered to MNK1 knockout (KO) mice and their WT controls (C57BL/6J), and its effects were compared with vehicle in behavioral assays.As expected, we found an increase in the grimace score in WT mice from day 1 after administration to day 10 (Figure 4A).A reduction in the vinorelbine-induced pro-nociceptive effect was observed in MNK1 KO mice with a significantly lower grimace score the day after the second administration compared to WT animals (Figure 4A).The statistical difference in the genotype effect can also be observed in the effect sizes in vinorelbine-treated animals (Figure 4B).Similarly, we found a reduction of paw withdrawal threshold in WT animals that reached almost 0.2 g at its peak.While a significant decrease in von Frey threshold was observed in the MNK1 KO mice 1 day after treatment, this effect was short lived and not seen on subsequent days (Figure 4C).In line with this, the paw withdrawal threshold effect sizes of vinorelbine were significantly greater in WT animals compared to MNK1 KO mice (Figure 4D).
Considering the reduction of the pro-nociceptive effects of vinorelbine in MNK1 KO mice, in an independent set of experiments we measured the effects of vinorelbine on STING-TBK1-IRF3-IFN signaling.Vinorelbine treatment was associated with a trend toward decreased levels of p-IRF3 in MNK1 KO mice (Figure 4G) in DRG, and it did not modify the levels of p-STING (Figure 4E) or type I IFN production (Figures 4H and 4I).Vinorelbine provoked similar changes in p-IRF3 in sciatic nerve (Figure 4M), and in addition the drug also downregulated p-TBK1 and IFN-a levels in MNK1 KO mice (Figures 4L and 4N).Together, these results show an interplay between the STING-IRF3-IFN pathway and MNK1 signaling, highlighting that vinorelbine fails to induce increases in IFN production in peripheral nerves in the absence of MNK1.
Because we observed a reduction in the pro-nociceptive effect of vinorelbine in MNK1 KO animals as well as abrogated associated changes in the STING pathway, we investigated the effect of vinorelbine on p-eIF4E in WT, Sting Gt/Gt , and MNK1 KO animals.Moreover, translation regulation through p-eIF4E has been shown to be associated with antiviral response and translation of mRNAs encoding proteins such as type I IFNs. 20An increase in p-eIF4E in DRG neurons was observed across all neuronal sizes (Figures 5A and 5B).Specifically, peripherin-positive neurons from vinorelbine-administered animals had an increase in p-eIF4E of 0.4 times the mean intensity value compared to control (Figure 5C).To give more insight about p-eIF4E changes induced by vinorelbine, western blot experiments in animals of the different genotypes were performed in DRG and sciatic nerve.In line with our immunohistochemistry experiments we found a clear trend toward an increase of p-eIF4E in DRG from WT animals (Figure 6A).On the other hand, no changes were found in DRGs of Sting Gt/Gt and MNK1 KO mice (Figures 6B and 6C).An increase in p-eIF4E was observed in sciatic nerve in WT mice (Figure 6D).Administration of vinorelbine resulted in no change in p-eIF4E levels in Sting Gt/Gt in sciatic nerve (Figure 6E).Further, a trend toward reduction in p-eIF4E with vinorelbine administration was observed in MNK1 KO sciatic nerve (Figure 6F).Both MNK1 and MNK2 are expressed in mouse DRG neurons, 42 so the remaining p-eIF4E in peripheral nerves in MNK1 KO mice is likely supported by the MNK2 isoform, which is known to have constitutive activity. 43ltogether, our results reinforce the idea that IFNs cause pain by acting on nociceptors and explain, mechanistically, how chemotherapeutics that are STING activators and microtubule destabilizers such as vinorelbine have the potential to cause pain and peripheral neuropathy.Furthermore, this study gives insight into the complexity of IFN induction within peripheral nerves and provides additional evidence that translational regulatory mechanisms play a critical role in chemotherapy-induced pain. 38

DISCUSSION
We reach several primary conclusions from the experiments described here.First, we have discovered a plausible mechanism through which vinorelbine induces pain hypersensitivity in male and female mice by activating the STING signaling pathway, including IRF3 phosphorylation  (S396), and subsequent type I IFN induction in the peripheral nervous system.This conclusion is supported by the finding that vinorelbineinduced pain is abrogated in Sting Gt/Gt mice and peripheral nerve induction of type I IFN by vinorelbine was absent in these mice.Second, our work shows that a downstream effector involved in vinorelbine-induced pain is MNK1-eIF4E signaling, a finding consistent with our previous work on mechanisms through which type I IFNs cause pain in mice. 18Therefore, our work also provides a clear mechanistic understanding of  vinorelbine-induced peripheral neuropathy and leads us to the third conclusion that MNK signaling is a therapeutic avenue for reducing the neuropathy caused by this chemotherapeutic.Future work will be aimed at understanding how MNK inhibition can be used as a therapeutic approach to reduce chemotherapy neuropathies and improve cancer chemotherapeutic efficacy.
Our work was primarily motivated by the consistent clinical finding that vinorelbine causes painful peripheral neuropathy in humans treated with this drug.Our work fills an important gap in knowledge by providing a mechanistic explanation for vinorelbine-induced pain.Neuropathy has been observed in oncological patients receiving vinorelbine in phase II/III trials for metastatic breast cancer, with neurotoxicity ranging from 33% to 43.6%. 44,45Notably, a heightened severity of neurotoxicity (87%) after 4 cycles of vinorelbine has also been reported. 6Patients who undergo treatment with vinorelbine report pain, primarily localized at the tumor site or in regions affected by nerve compression caused by the tumor. 1,44,46Moreover, some patients also present with paresthesia and myalgia consisting of acute pain in lower back muscles and extremities. 44There is also evidence that vinorelbine causes a peripheral neuropathy because the drug decreases the amplitude of evoked action potentials in sensory nerve bundles in patients, 6 one of the main neurophysiological abnormalities observed in neuropathy. 47Vincristine and vinblastine, which are vinca alkaloid chemotherapeutic agents with a shared multi-ringed chemical structure, induce neuropathy.Among them vincristine causes the most severe neurotoxicity, 48 a deleterious effect also observed in preclinical studies on vincristine-induced peripheral neuropathy and mechanical allodynia, 49 but also in the clinical studies, resulting in peripheral neuropathy in 18%-70% of adult patients. 50Compared to vincristine, vinorelbine is slightly less potent and less neurotoxic 48 and has a better therapeutic index than vincristine and vinblastine probably due to its higher affinity for mitotic microtubules. 51Finally, vinorelbine treatment is associated with the worsening of peripheral neuropathy in patients with pre-existing diabetes or chemotherapy-induced peripheral neuropathy (CIPN). 52Despite the induction of neuropathy, vinorelbine is approved for advanced NSCLC treatment 4 and is used in clinical trials for a wide variety of cancer types. 1,5,44,45ased on the rationale that vinorelbine causes neuropathic pain and is used alone or in combination with other chemotherapeutic drugs, 53 we studied the signaling pathways associated with vinorelbine-related pain to find new strategies to mitigate or prevent the concomitant pain associated with its use.Our results provide a detailed mechanistic rationale for blocking MNK signaling as a therapeutic strategy to mitigate the pain-promoting effects of vinorelbine.
A second motivation for our work was to better understand the role of STING and type I IFN signaling in painful peripheral neuropathy.Recently, it was demonstrated that microtubule-targeted agents (MTAs), including vinorelbine and eribulin, activate the STING pathway leading to type I IFN production in human breast cancer and leukemia monocytic cell lines. 12,54The STING pathway is an intracellular sensor system of external or self-origin DNA that is critical for recognizing pathogen infection and/or cellular damage to establish an effective host defense. 55,56Our results clearly demonstrate that vinorelbine induces STING activation in mouse DRG neurons and peripheral nerve as reflected in increased IRF3 phosphorylation and type I IFN induction. 57Recently published work shows that STING induces peripheral nociceptive sensitization via nuclear factor kB (NF-kB) activation in DRG neurons in a model of bone cancer pain in rats. 29Furthermore, the STING inhibitor C-176 has been shown to reduce mechanical hypersensitivity induced by nerve injury in rats suggesting a pro-nociceptive role for STING. 28ounting evidence has shown that STING is anomalously activated in models of peripheral neuropathic pain, 30,58 central neuropathic pain, 59,60 and inflammatory pain 29,61 in rodents.9][60][61] Collectively, our work is consistent with the emerging view that activation of STING signaling in the peripheral nervous system is pro-nociceptive and may play an important role in multiple types of painful peripheral neuropathies.
Evidence of an antinociceptive role of STING has also been reported. 26,27,62Activators of STING, including ADU-S100, DMXAA, and 3 0 3 0 -cyclic GMP-AMP, have been shown to reverse CIPN-, nerve injury-, 26 and bone cancer-induced pain in mice 27 via suppression of nociceptor excitability.Under physiological conditions, the role of STING has been shown to regulate steady-state nociception, since intrathecal administration of the STING agonists ADU-S100 or DMXAA produced antinociception in naive mice.Moreover, naive Sting Gt/Gt mice showed increased sensitivity to mechanical and cold stimuli, although we did not find this in our experiments for von Frey threshold, and DRG nociceptors from these mice displayed increased action potential firing and reduced rheobase suggesting nociceptor hyperexcitability. 26Thus, the aforementioned studies suggest an antinociceptive role for STING. 27A recent review concluded that STING signaling causes bidirectional effects on nociception but did not reach firm conclusions on why STING activation may be associated with increased pain in some contexts and antinociception in others. 56One hypothesis to explain this controversy is that whether STING is activated in the peripheral or central nervous system (CNS) is critical for the direction of the effects of the pathway on nociceptive signaling.From this perspective, it is important to note that the chemical properties of vinorelbine, like other vinca alkaloids such as vincristine, render the drug unable to cross the blood-brain barrier. 63,64P-glycoprotein limits vinorelbine distribution to the brain causing this drug to have poor antitumor activity on CNS metastasis. 65ccordingly, we only observed pro-nociceptive effects of vinorelbine in mice.
There is controversy over the effects of type I IFNs on nociception, which may also be explained by peripheral nervous system versus CNS sites of action. 66We have previously shown that type I IFNs activate their receptors (IFNRs) localized on nociceptors driving neuron hyperexcitability and pro-nociceptive actions in mice. 18Likewise, systemic IFN-a increases formalin-evoked nociceptive behavior in mice. 21In the same line, patients with hepatitis undergoing IFN-a therapy reported experiencing pain with IFN infusion. 22On the other hand, other studies have shown that STING-type I IFN signaling attenuates DRG neuron excitability via suppression of Na + and Ca 2+ channels activity, 26,27 suggesting an antinociceptive role for IFNs.Notably, IFN-a administered in the periphery (i.pl.) led to mechanical hypersensitivity, which was alleviated by intrathecal administration of IFN-a, 26 again consistent with different actions of this signaling pathway depending on the site of administration.8][69] Our work adds to the growing body of evidence that peripheral type I IFN signaling is pro-nociceptive.We did not test CNS-mediated effects of vinorelbine because this is not relevant for the clinical effects of this peripherally restricted drug.
Findings in Sting Gt/Gt mice unequivocally implicate STING signaling in the pro-nociceptive effects of vinorelbine.Accordingly, we did not observe changes in STING downstream effectors in either the peripheral nerve or DRG with vinorelbine administration.In contrast, in WT mice, we consistently observed increased IRF3 phosphorylation using western blotting and immunofluorescence.We also observed increases in type I IFNs in DRG and sciatic nerve in WT mice, which likely occurred due to the presence of IRF3-driven transcription of Ifna1 and Ifnb1 genes. 34,35On the other hand, we did not find changes in the phosphorylation of STING (Ser366), which is essential to trigger STING pathway signaling. 70A possible explanation is that, after trafficking to the Golgi apparatus, STING is phosphorylated by serine/threonine UNC-51-like kinase (ULK1/ATG1) to facilitate its degradation and prevent the deleterious effect of a long-lasting inflammatory state induced by persistent STING activation. 71Thus, our failure to observe changes in phosphorylation of STING could be due to degradation of the activated protein at the time point at which we harvested tissues for analysis.This could also explain the lack of changes in p-TBK1 in WT mice.
The exact mechanism by which the MTAs such as vinorelbine or eribulin enhance STING signaling has not been fully uncovered.Some studies have suggested that the STING pathway is activated by mitochondrial or nuclear DNA release induced by cellular damage resulting from the treatment with MTAs, but it has also been proposed that STING activation occurs downstream of the microtubule disruption triggered by these chemotherapeutics. 12,54,72Whatever the precise mechanism turns out to be, the discovery that vinorelbine-induced pain depends on STING-type I IFN signaling makes it possible to implement strategies to alleviate the neuropathic pain state caused by the chemotherapeutic.We demonstrate that this can be achieved by targeting MNK1-eIF4E signaling.Vinorelbine caused an increase in eIF4E phosphorylation at a site that is specifically regulated by MNK in both the peripheral nerve and DRG cell bodies.Moreover, grimacing and mechanical hypersensitivity were profoundly reduced in MNK1 KO mice.Translation regulation via MNK-eIF4E signaling has been shown to participate in the development and maintenance of pain hypersensitivity in many different preclinical models. 18,38,39,41Notably, we have previously shown that type I IFNs induced the phosphorylation of ERK1/2, which phosphorylates the kinase MNK inducing eIF4E phosphorylation in mouse DRG.Thus, MAPK/ERK serves as the link between IFNR activation and MNK-eIF4E hypersensitivity induction. 18MNK1 has been shown to be expressed at the mRNA level in nearly all human nociceptors 42 indicating that this mechanism is likely to translate to humans.Our findings support the idea that MNK represents a promising mechanistic target for pharmacological relief or prevention of chemotherapyinduced pain.It will be important to understand whether MNK inhibition will interfere with the anticancer activity of the chemotherapeutic, but this is unlikely since increased MNK activity and increased eIF4E phosphorylation are associated with many cancers 73,74 and MNK inhibitors are currently in clinical trials for cancer treatment. 75It is crucial to emphasize that the combination of vinca alkaloids with potent pain management medications has the potential to substantially enhance the tolerability of cancer treatments, offering a prophylactic solution for chemotherapy-induced pain.Instead of advocating for reduced clinical use of vinca alkaloids, which may happen as cancer therapeutic development continues to progress, our work seeks to illustrate that, due to the multifaceted effects of these chemotherapeutic agents on signaling pathways, we can harness these mechanisms as therapeutic targets for pain, serving as proactive alternatives to mitigate chemotherapy-induced pain.Blocking IFN signaling would not be a favorable approach in the context of vinorelbine chemotherapy, as IFN production plays a role in the anticancer response of these drugs.IFN-a, with its immunomodulatory, antiproliferative, and antiangiogenic effects, is FDA approved for treating some types of leukemia, lymphoma, and melanoma. 76Moreover, antitumor effects of STING require type I IFN induction to activate dendritic cells and subsequently T cells, 77 improving outcomes for cancer patients.Clinical trials have even combined vinorelbine with IFN-a2, enhancing the chances of partial remission in renal cancer. 78Thus, we believe that blocking MNK signaling is the best avenue for intervention in our view.

Limitations of the study
There are several limitations of this study.We did not directly interfere with type I IFNs, using either IFN blocking antibodies or IFN KO mice, which would have allowed us to dissect STING activation versus type I IFN downstream effects.While this can be addressed in future studies, our main goal was to understand whether STING activation was linked to vinorelbine-induced pain, and our use of Sting Gt/Gt mice provides a clear link between vinorelbine and STING signaling.In some cases, we found trends for signaling changes in sciatic nerve or DRG (p values between 0.06 and 0.09).These trends were consistent with findings using transgenic and KO mice for signaling pathways and were observed in a complex tissue where there are a mix of cell types that contribute to the overall changes in protein levels.We do not think that these observed trends weaken the overall conclusions of the study.Unanswered questions remain to be uncovered regarding other STING activated pathways, which we did not examine.For instance, it is known that STING signaling can activate inhibitor of kB kinase (IKKε)/NF-kB, which synergizes with IRF3 to induce higher levels of type I IFNs. 79Future studies can address nuances of these signaling effects now that our work has unequivocally demonstrated that vinorelbine-induced pain requires STING signaling.Finally, while we provide genetic evidence for MNK1-eIF4E signaling as a causative factor in vinorelbine-evoked pain, it will be interesting to assess the extent to which MNK inhibitors like eFT508 can reproduce this effect.Given that pharmacological inhibition of MNK has thus far matched genetic manipulation in behavioral endpoints in neuropathic and other pain models, 18,38,39,41,80,81 it is likely that these MNK inhibitors will be effective in reducing pain-promoting effects of vinorelbine.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

EXPERIMENTAL MODEL AND ANIMAL DETAILS
All experiments were carried out in male and female mice.The following mouse strains were used: ICR outbred mice purchased from Envigo Laboratories (4-6 weeks, weighing from 20 to 25 g, RRID: IMSR_CRL:022); Mknk1 À/À (MNK1 KO), a gift from the Sonenberg laboratory at McGill University and their C57BL/6J WT (WT) littermates 43,82 bred at the University of Texas at Dallas (UTD); Sting Gt/Gt also known as Goldenticket mice (Tmem173Gt) and the suggested controls C57BL/6J WT (Lot: 000664) purchased from Jackson laboratory.
Purchased mice were used in experiments starting one week after arrival at the animal facilities at the University of Texas at Dallas.Animals had free access to food and water before the experiments.Mice were housed in groups of 4 per cage in non-environmentally enriched cages with food and water ad libitum on a 12 h non-inverted light/dark cycle.All animal procedures were approved by The University of Texas at Dallas IACUC.Experiments were in compliance with the National Institutes of Health Guide for Care and Use of Laboratory Animals (Publication No. 85-23, revised 1985).Animals were monitored for health according to IACUC guidelines at The University of Texas at Dallas prior and during experimentation.This study was not pre-registered.

Drug administration scheme
Chemotherapeutic drugs are administered to patients using restricted doses.In this work, we administered the maximum tolerated dose (MTD) of vinorelbine (10 mg/kg, i.v; 32 ) to male and female mice.MTD refers to the highest dose of a drug that causes a therapeutic effect but not unacceptable side effects.The dose 10 mg/kg of vinorelbine is well tolerated after a single-dose or 3 repeated cycles at a singledose. 32][85] Our drug administration scheme was based on pilot experimental trials, which showed that vinorelbine (10 mg/kg, i.v.) but not vehicle (DMSO 3% in PBS) causes tactile allodynia and facial grimace responses.These effects were aggravated by the second administration on day 7. Based on these pilot results and the previous clinical and preclinical literature, we decided to do two i.v.administrations of vinorelbine separated by a week for the rest of the experiments.

Immunohistochemistry (IHC)
Mouse DRG sections were cut from frozen blocks with a cryostat (Leica CM1950) at 20 mm thickness and mounted on to charged slides.The tissues were fixed with 10% formalin for 10 min followed by serial permeabilization with 50% ethanol, 75% ethanol and 100% ethanol for 5 min each.The slides were allowed to dry at room temperature until moisture from the slides evaporated.Hydrophobic boundaries were drawn around the sections with a Pap-pen (ImmEdge Hydrophobic Barrier Pap-Pen, Cat.#H-4000, Vector Laboratories) and the slides were allowed to further dry for 30 min at room temperature.Sections were blocked with 10% normal goat serum with 0.3% Triton X-100 in 0.1M PB pH 7.4 for 1 h at room temperature.The sections were then incubated overnight at 4 C in blocking solution containing primary antibodies.

FLIR imaging
Changes in temperature either on the site of administration (tail base) or abdomen wall were assessed using an FLIR T-Series Thermal Imaging Camera (FLIR Systems, Inc).Animals were allowed to acclimate in the testing room for 1 h (ambient temperature of 21 C G 2 C) in Plexiglas chambers (11.4 3 7.6 3 7.6 cm) before testing.We captured colorized infrared thermogram images containing the abdomen wall and tail of mice before experimental treatment and at 1, 3, 5 and 8 days after the first administration of vinorelbine.The temperature was obtained drawing a straight line at the base of the tail or a square on the abdomen of mice.The mean temperature was recorded from the average of each pixel along the drawn line or in the square.Three thermograms were averaged to obtain the mean temperature of the tail and the abdomen per animal.

Mouse behavior
We performed behavioral experiments in wild type (WT) or transgenic male and female mice.Animals from each cage were randomized into groups (control or treatment).All behavioral measurements were performed in age-matched animals.Animals were habituated to the experimental setup (Plexiglas chambers (11.4 3 7.6 3 7.6 cm)) for the von Frey test 1 h before each experiment.Behavioral testers were blinded to treatment and genotype in all experiments.Mechanical paw withdrawal thresholds were measured using the up-down method with calibrated von Frey filaments (Stoelting Company, Wood Dale, IL).We calculated this threshold by using the formula: 50% g threshold = 10(Xf+kd)/10000, where Xf = the value (in log units) of the final von Frey filament used, k = the value from look up table for the pattern of positive and negative responses published previously, 87 and v = the mean difference (in log units) between stimuli.Increasing or decreasing forces with different von Frey filaments were applied during 3 s to the mouse paw in order to calculate the 50% withdrawal threshold.
Mouse grimace scoring was performed as a behavioral test for examination of spontaneous pain response. 88Mice were placed individually in a Plexiglas chamber and allowed to acclimate for 1 h, and then scored by blinded scorers at 1, 3, 5, 8, and 10 days after the first vinorelbine administration.The scores of each animal were averaged at each time point.

QUANTIFICATION AND STATISTICAL ANALYSIS
Data were analyzed with Graphpad Prism V9 (Graphpad, San Diego, CA).All data are shown as mean G standard error of the mean (SEM).Repeated measures two-way ANOVA was used to analyze behavioral data plotted as group by time.Student's t test was used to assess effect sizes and Western blot data.Other statistical tests are described in figure legends.We used at least 5 mice per group of each sex in behavioral experiments based on power analysis predicated on our previous work with type I IFNs. 18Because no sex differences were observed in initial behavioral experiments with vinorelbine, we used male and female mice throughout our experiments with all data pooled from both sexes.We did not test for outliers or exclude behavioral data from analysis.The ROUT method (Q = 1%) was used to identify outliers in Western blot analysis.

Figure 1 .
Figure 1.Vinorelbine induced mechanical hypersensitivity and spontaneous pain in mice associated with activation of STING-IRF3-IFN pathway in DRG and sciatic nerve (A) Time course of grimacing score in WT ICR mice at 1, 3, 5, 8, and 10 days after the first administration of vinorelbine (10 mg/kg, i.v.) or vehicle (3% DMSO i.v.).Arrow heads show the time of administration (days 0 and 7).(B) Effect size (AUC) of vinorelbine-induced grimacing score in WT ICR mice.(C) Time course of mechanical sensitivity in WT ICR mice at 1, 3, 5, 8, and 10 days after the first administration of vinorelbine (10 mg/kg i.v.) or vehicle (3% DMSO i.v.) in paw withdrawal threshold.(D) Effect size (AUC) of vinorelbine-mediated mechanical sensitivity in WT ICR mice.(E-I) Western blot analysis showing STING-IRF3-IFN pathway in WT DRGs from vinorelbine administered animals compared to vehicle.(J) Representative western blot images showing STING-IRF3-IFN pathway in vinorelbine, or vehicle administered WT DRGs.(K-O) Western blot analysis showing STING-IRF3-IFN pathway in vinorelbine-treated WT sciatic nerve compared to vehicle.(P) Representative western blot images showing STING-IRF3-IFN pathway in vinorelbine-treated WT sciatic nerve compared to vehicle.Data are presented as the mean G SEM. **p < 0.01, ***p < 0.001, ****p < 0.0001 (n = 13 per group, with 7 male and 6 female mice) as determined by two-way RM ANOVA followed by Bonferroni's test in A, C. *p < 0.05, ***p < 0.001 as determined by t test in B, D, E-I, and K-O (n = 13 per group in B, D; n = 8-12 per group in E-O).Vino: vinorelbine, veh: vehicle.See also Figures S1 and S2.

Figure 2 .
Figure 2. Vinorelbine increases p-IRF3 immunoreactivity in DRG neurons (A) Representative confocal micrographs showing p-IRF3 (red) and peripherin (green) immunofluorescence in WT ICR mouse DRG neurons day 3 post-second administration of vinorelbine (10 mg/kg i.v.) or vehicle (3% DMSO i.v.).DAPI (cyan) stains nuclei in the tissue.Respective bottom rows show zoomed-in images of a few neurons inside the white dashed-lined box.Scale bar -10 mm.(B) Analysis of immunoreactivity of p-IRF3 across all neuronal sizes in vinorelbine-and vehicle-administered WT ICR mice.The mean intensity values are plotted as a function of different neuronal sizes.(C) Mean intensity value of p-IRF3 in peripherin-positive neurons in vinorelbine-and vehicle-administered WT ICR mice.Data are presented as mean G SEM. Section thickness -20 mm.Scale bar -50 mm.*p < 0.05, **p < 0.01, ***p < 0.001 (n = 3 in vehicle group, n = 4 in vinorelbine group) as determined by two-way ANOVA followed by Bonferroni's multiple comparisons test in B. ****p < 0.0001 as determined by t test in C.

Figure 3 .
Figure 3. Sting Gt/Gt mice show decreased vinorelbine-induced nociception and lose p-IRF3 and type I IFN induction in peripheral nerves (A) Time course of grimacing score in Sting Gt/Gt and C57BL/6J WT mice at 1, 3, 5, 8, and 10 days after the first administration of vinorelbine (10 mg/kg, i.v.) or vehicle (3% DMSO i.v.).Arrow heads show the time of administration (days 0 and 7).(B) Effect size (AUC) of vinorelbine-induced grimacing score in Sting Gt/Gt and C57BL/6J WT mice.(C) Time course of mechanical threshold in Sting Gt/Gt and WT C57BL/6J mice at 1, 3, 5, 8, and 10 days after the first administration of vinorelbine (10 mg/kg i.v.) or vehicle (3% DMSO i.v.) in paw withdrawal threshold.(D) Effect size (AUC) of vinorelbine-mediated mechanical sensitization on paw withdrawal threshold in Sting Gt/Gt and C57BL/6J mice.(E-H) Western blot analysis showing STING downstream pathway in DRGs from vinorelbine-administered Sting Gt/Gt mice compared to vehicle.(I) Representative western blot images showing STING downstream pathway in DRGs from vinorelbine-administered Sting Gt/Gt mice compared to vehicle.(J-M) Western blot analysis showing STING downstream pathway in sciatic nerve from vinorelbine-administered Sting Gt/Gt mice compared to vehicle.(N) Representative western blot images showing STING downstream pathway in sciatic nerve from vinorelbine-administered Sting Gt/Gt mice compared to vehicle.Data are presented as the mean G SEM. *p < 0.05, **p < 0.01, ***p < 0.001 vs. veh WT, and #p < 0.05, ##p < 0.01 vs. Vino C57BL/6J (n = 10-12 per group, 6 males and 4-6 females per group) as determined by two-way RM ANOVA followed by the Bonferroni's test in A,C.***p < 0.001, ****p < 0.0001 as determined by twoway ANOVA followed by Bonferroni's test in B,D (n = 10-12 per group).t test in E-H and J-M (n = 3-6 mice per group).Vino: vinorelbine, veh: vehicle.See also Figure S3.

Figure 5 .
Figure 5. Vinorelbine increases p-eIF4E in WT DRG neurons and sciatic nerve but not in Sting Gt/Gt and MNK1 KO mice (A) Representative confocal micrographs showing p-eIF4E (magenta) and peripherin (green) immunofluorescence in WT mouse DRG neurons at day 3 postsecond administration of vinorelbine (10 mg/kg, i.v.) or vehicle (3% DMSO i.v.).DAPI (cyan) stains nuclei in the tissue.Respective bottom rows show zoomedin images of a few neurons inside the white dashed-lined box.Scale bar -10 mm.

Figure 5 .
Figure 5. Continued (B) Analysis of immunoreactivity of p-eIF4E across all neuronal sizes in vinorelbine-and vehicle-administered WT mice.The mean intensity values are plotted as a function of different neuronal sizes.(C) Mean intensity value of p-eIF4E in peripherin-positive neurons in vinorelbine-and vehicle-administered WT mice.Data are presented as mean G SEM. Section thickness -20 mm.Scale bar -50 mm.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 (n = 3 per group) as determined by two-way ANOVA followed by Bonferroni's multiple comparisons test in B. Data are presented as mean G SEM ****p < 0.0001 as determined by t test in C.

Figure
Figure Vinorelbine enhances p-eIF4E in WT DRG and sciatic nerve but not in Sting Gt/Gt and MNK1 KO mice (A-C) Western blot analysis of p-eIF4E in DRGs of WT (n = 8-10 per group), Sting Gt/Gt (n = 6 per group), and MNK1 KO (n = 10-12 per group) vinorelbine-or vehicle-administered mice.(D-F) Western blot analysis of p-eIF4E in sciatic nerve of WT (n = 10 per group), Sting Gt/Gt (n = 3-4 per group), and MNK1 KO (n = 8-10 per group) vinorelbine-or vehicle-administered mice.(G) Representative western blot images showing p-eIF4E bands in DRGs of WT, Sting Gt/Gt , and MNK1 KO vinorelbine-or vehicle-administered mice.(H) Representative western blot images showing p-eIF4E mean intensity levels in sciatic nerve of WT, Sting Gt/Gt , and MNK1 KO vinorelbine-or vehicleadministered mice.Data are presented as mean G SEM *p < 0.05 as determined by t test in D.

TABLE d
B Drugs B Drug administration scheme B Western blot B Immunohistochemistry (IHC) B FLIR imaging B Mouse behavior d QUANTIFICATION AND STATISTICAL ANALYSIS