Protection against oxaliplatin-induced mechanical and thermal hypersensitivity in Sarm1â/â mice

Chemotherapy-induced peripheral neuropathy (CIPN) is a common dose-limiting side effect of cancer treatment, often associated with degeneration of sensory axons or their terminal regions. Presence of the slow Wallerian degeneration protein (WLDS), or genetic deletion of sterile alpha and TIR motif containing protein 1 (SARM1), which strongly protect axons from degeneration after injury or axonal transport block, alleviate pain in several CIPN models. However, oxaliplatin can cause an acute pain response, suggesting a different mechanism of pain generation. Here, we tested whether the presence of WLDS or absence of SARM1 protects against acute oxaliplatin-induced pain in mice after a single oxaliplatin injection. In BL/6 and WldS mice, oxaliplatin induced significant mechanical and cold hypersensitivities which were absent in Sarm1 /− mice. Despite the presence of hypersensitivity there was no significant loss of intraepidermal nerve fibers (IENFs) in the footpads of any mice after oxaliplatin treatment, suggesting that early stages of pain hypersensitivity could be independent of axon degeneration. To identify other changes that could underlie the pain response, RNA sequencing was carried out in DRGs from treated and control mice of each genotype. Sarm1 /− mice had fewer gene expression changes than either BL/6 or WldS mice. This is consistent with the pain measurements in demonstrating that Sarm1 /− DRGs remain relatively unchanged after oxaliplatin treatment, unlike those in BL/6 and WldS mice. Changes in levels of four transcripts – Alas2, Hba-a1, Hba-a2, and Tfrc – correlated with oxaliplatin-induced pain, or absence thereof, across the three genotypes. Our findings suggest that targeting SARM1 could be a viable therapeutic approach to prevent oxaliplatin-induced acute neuropathic pain.

Proposed mechanisms for chemotherapy-induced axon degeneration by different chemotherapeutics include; altered axon transport and axon degeneration by vincristine and paclitaxel (Boyette-Davis et al., 2013;LaPointe et al., 2013;Siau et al., 2006), altered mitochondrial function and Ca 2+ homeostasis (Fukuda et al., 2017), and bortezomib-induced transcriptional changes leading to SARM1 activation (Geisler et al., 2019b) in addition to its microtubule-targeting properties (Pero et al., 2019). The mechanism of oxaliplatin-induced pain is unclear. Whilst OIPN could be due to axon degeneration, symptoms in patients receiving oxaliplatin tend to subside between treatment cycles (Extra et al., 1998) and can completely resolve after cessation of treatment (Kanat et al., 2017) suggesting involvement of non-degenerative pain-inducing mechanisms.
After rapid non-enzymatic conversion of oxaliplatin to oxalate and a platinum-containing derivative (Starobova and Vetter, 2017), the oxalate moiety induces cold-hypersensitivity, whereas the platinumcontaining derivative induces mechanical allodynia in rats (Sakurai et al., 2009). Platinum compounds accumulate in dorsal root ganglia (DRGs), where they form DNA adducts (Faivre et al., 2003). Therefore, damage to nuclear or mitochondrial DNA in DRGs or other cells, or other (unknown) actions of platinum-based compounds could underlie the acute pain response.
Here we investigated whether the presence of WLD S or absence of SARM1 protects against acute oxaliplatin-induced pain in mice. Cold and mechanical sensitivity, as well as assessing intraepidermal nerve fiber (IENF) density as a readout of distal axon degeneration were employed, and transcriptional changes were observed via RNA Sequencing. We demonstrate that acute oxaliplatin-treatment induced significant mechanical and cold hypersensitivity in wild-type mice. This was attenuated in Sarm1 − /− mice but not in Wld S mice. Hypersensitivities were not associated with detectable loss of IENFs in footpads of any genotype after oxaliplatin treatment which suggests that the early stages of pain hypersensitivity may be independent of axon degeneration. However, we noted transcriptional changes that correlate with induction of pain phenotypes.

Animals
All animal work was undertaken at the Babraham Institute in accordance with the Animals (Scientific Procedures) Act 1986 under project licence number 70/7620. Wild-type C57BL/6Babr (BL/6), Wld S and Sarm1 − /− male mice (both on a C57BL/Babr background) aged 12-16 weeks were used for the study. Mice were housed under controlled conditions (12 h light-dark cycle) in individually ventilated cages (IVCs) in a pathogen-free environment with standard rodent chow and water available ad libitum (except when behavioural assessments were being made). Most mice were group housed (2-3 mice per cage), though very occasionally a mouse was housed individually.

Behavioural tests
Mechanical allodynia was assessed using a dynamic plantar aesthesiometer and thermal allodynia using a dynamic cold plate. For the plantar aesthesiometer test, mice were placed into individual sections of a plastic box with a wire mesh floor and left to habituate for 30-60 min. Then, a dynamic plantar aesthesiometer (Ugo Basile) was used to apply pressure up to a maximum of 5 g at a rate of 0.5 g/s to the plantar region of the right hind paw. If the mouse did not withdraw its paw after 30 s, the experiment was stopped manually and the probe retracted. This was repeated five times with at least 10 min between each measurement. The average of the five values for each mouse was taken forward for analysis. This was repeated 48 h after oxaliplatin administration. For the cold plate test, an individual mouse was placed onto a dynamic hot/cold plate (Ugo Basile) at 18 • C. The plate was cooled at a rate of 0.3 • C/min to 2 • C and the number of jumps (defined as both rear paws off the plate simultaneously) was recorded and grouped into 2 • C bins (Young et al., 2014). The baseline measurement is the number of jumps prior to oxaliplatin injection. This process was repeated 4 days after mice received the injection to obtain the post-oxaliplatin measurements, based on a previous study which noted that cold sensitivity peaked 4 days after oxaliplatin administration (Young et al., 2014). We refer to this as allodynia (rather than hyperalgesia) since in our mice, no increase in jumping behaviour was observed at lower temperatures before oxaliplatin administration, suggesting these temperatures are not painful in naïve mice. Cold plate data is presented as the area under the curve alongside individual time courses.

Sample sizes and exclusions
In all behavioural studies, group sizes were 12 mice. Control vehicletreated groups were also included. No wild-type or Sarm1 − /− mice were excluded from any of the studies. One Wld S mouse was excluded from the study as it performed abnormally in the cold plate test at baseline compared to the other mice. The mouse was excluded from all successive assessments (including other behavioural tests and histological assessment of intraepidermal nerve fiber density).

Epidermal nerve fiber analysis
Evaluation of intraepidermal innervation (IENF) was performed using immunohistochemical staining for PGP 9.5, a pan-axonal marker, as described previously (Melli et al., 2006). Briefly, 4 days postinjection, 2 mm punch biopsies were taken from the medial footpads of the right hind limb and fixed in PLP fixative. After cryosectioning, the slides were stained with anti-PGP 9.5 antibody (Sigma-Aldrich AB5925). IENFs were counted in 4-6 sections for each mouse and averaged. The data were presented as the mean number of fibers per linear millimetre of epidermis.

Animals
Mice of three experimental genotypes (C57BL/6Babr, Wld s and Sarm1 − /− − ; n = 8 animals per genotype) were processed for transcriptomic studies. Two-month-old mice were administered either a single intraperitoneal injection of 6 mg/kg oxaliplatin or saline (n = 4 mice per treatment).

RNA isolation
Dorsal root ganglia (L3-L5) were sub dissected in RNase-free conditions (RNaseZap, Sigma Aldrich) from freshly culled mice and flash frozen until further use. For RNA extraction, tissue was thawed directly in TRIsure reagent (Bioline) and RNA isolated following manufacturer's instructions. RNA was purified (RNeasy kit, Qiagen) with on-column DNase treatment and analysed on an Agilent 2100 Bioanalyzer.

RNA-seq library preparation
Only RNA samples with RIN > 8 were used for sequencing. Libraries were prepared using the TruSeq (unstranded) mRNA kit (Illumina) following the manufacturer's low sample protocol. Following library amplification and bead purification, the final fragment size was analysed, and libraries quantified using the Universal KAPA Library Quantification kit (Kapa Biosystems) and a Bio-Rad C100 thermal cycler. An equal amount of cDNA was used to pool up to four samples, which were sequenced in one lane. Sequencing was carried out to a depth of 20 million 25-bp single-end reads per library.

Bioinformatics pipeline and statistics
FastQ files were trimmed with Trim Galore v0.4.3 using default settings then aligned against the mouse GRCm38 genome assembly with hisat2 v2.0.5 using options -no-mixed and -no-discordant. Mapped positions with MAPQ values of < 20 were discarded.
Gene expression was quantified using the RNA-Seq quantification (A) Both BL/6 and Wld S , but not Sarm1 − /− mice exhibit increased mechanical hypersensitivity when pressure was applied to the paw using the plantar aesthesiometer. Statistically significant difference between groups is indicated (***p < 0.001; **p < 0.01, two-way ANOVA followed by Sidak's post-hoc pairwise comparison). (B) Mean difference in mechanical hypersensitivity between baseline and 2 days post-oxaliplatin treatment for the three genotypes. P values are shown on the graph, with statistically significant difference for BL/6 and Wld S (paired t-tests followed by Bonferroni correction for multiple comparisons). Data are presented as mean ± s.e.m. n = 12 per group.
pipeline in SeqMonk v1.37.0 in unstranded library mode using gene models from Ensembl v67. For count based statistics, raw read counts over exons in each gene were used. For visualization and other statistics, log2 RPM (reads per million reads of library) expression values were used. Differentially expressed genes were selected using pairwise comparisons with DESeq2 with a cut-off of P < 0.05 following multiple testing correction. A secondary intensity filter akin to a dynamic foldchange filter was applied to DESeq2 hits. DESeq2 comparisons were between saline and oxaliplatin treated mice of each experimental genotype.

Statistical analysis
Microsoft Excel was used to store raw data and calculate mean values. Statistical analysis and graphical visualisations were made using GraphPad Prism 6 (USA). For cold plate analysis, a Mann Whitney test followed by Bonferroni analysis was performed on area under the curve (AUC) data. For the plantar aesthesiometer, two-way ANOVA followed by Sidak's multiple comparison test was used (for Fig. 1.A) in order to identify and quantify differences between genotypes and/or treatment overall. For Fig. 1.B, mean difference in mechanical hypersensitivity was analysed using paired t-tests followed by Bonferroni correction for Fig. 2. Oxaliplatin-induced cold hypersensitivity is prevented by removal of SARM1, but not the presence of WLD s . (A) Number of jumps made by BL/6, Wld S and Sarm1 − /− mice in response to a cold ramp (cooling from 18 • C to 2 • C) before (baseline) and 4 days after a single dose of oxaliplatin. (B) BL/6 and Wld S mice exhibit a significant increase in the number of jumps after oxaliplatin treatment while there is no change in Sarm1 − /− mice. Statistically significant difference between groups is indicated (*p < 0.05 and *** p < 0.001, AUC-area under the curve showing difference in the total number of jumps in response to declining temperature ramp from 18 • C to 2 • C). Data are presented as mean ± s.e.m. n = 12 per group. multiple comparisons. For assessment of intraepidermal nerve fiber density, two-way ANOVA followed by Sidak's multiple comparison test was used.

Results
The present study has assessed thermal and mechanical allodynia (tested on different days to minimise interference between testing sessions; Supplementary Fig. 1) and intraepidermal nerve fiber (IENF) density after a single administration of oxaliplatin in vivo. Changes to RNA expression were also assessed after oxaliplatin administration.

Sarm1 deletion blocks the development of mechanical hypersensitivity following acute oxaliplatin treatment
In order to assess whether absence of SARM1 or presence of WLD S protects against mechanical allodynia, following acute oxaliplatin treatment, we injected all three mouse genotypes with a single dose of oxaliplatin (6 mg/kg). Forty-eight hours after oxaliplatin administration, wild-type and Wld S mice displayed increased sensitivity to pressure; they withdrew their paws at a significantly lower pressure than they did at baseline assessment ( Fig. 1.A, B). Remarkably, the Sarm1 − /− mice displayed no significant difference in sensitivity to pressure as their threshold for removing their paws was the same as baseline ( Fig. 1.A, B).
None of the three genotypes tested differed significantly in the pressure they withdrew their paws at baseline assessment ( Fig. 1.A).

Sarm1 deletion prevents the development of cold hypersensitivity following acute oxaliplatin treatment
At baseline, there was no difference in cold-induced jumping behaviour of wild-type, Wld S or Sarm1 − /− mice at any of the temperatures measured. However, 4 days after a single 6 mg/kg injection of oxaliplatin, wild-type mice showed significantly increased jumping compared to baseline measures ( Fig. 2.A, B). Similarly, Wld S mice exhibited significantly increased jumping behaviour after oxaliplatin treatment compared to baseline and this increase was greater than that seen in the wild-type mice (Fig. 2.A, B). Interestingly, the jumping behaviour of Sarm1 − /− did not change from baseline after oxaliplatin treatment ( Fig. 2.A, B).

Acute oxaliplatin treatment does not induce intraepidermal nerve fiber (IENF) loss
There was no difference in IENF density between the three genotypes treated with saline (Fig. 3). No change in IENF density was detected for any genotypes post-oxaliplatin treatment. Although we cannot exclude highly localised effects outside the regions studied, this suggests that general axon degeneration does not underpin the pain phenotype after acute oxaliplatin administration.

RNA expression changes after acute oxaliplatin administration
To explore the mechanistic link between acute oxaliplatin treatment and differences in behaviour we collected RNA from dorsal root ganglia (DRGs) of mice treated with either saline or oxaliplatin (4 days after treatment). Mice used for the RNA sequencing study exhibited similar levels of mechanical allodynia as those used in the behavioural and histological studies (Supplementary Fig. 2). DRGs were chosen for RNA Sequencing as they have previously been implicated in the pathophysiology of the oxaliplatin pain mechanism. We hypothesised that acute oxaliplatin treatment would induce changes in DRG gene expression that are either a consequence of pathological events or could even be part of the pathogenic mechanism in both mechanical and cold hypersensitivity observed in these animals. Transcriptomic differences between C57BL/ 6, Wld S and Sarm1 − /− mice in response to oxaliplatin treatment could underlie the differences seen in pain response and the resistance of Sarm1 − /− mice to cold and mechanical hypersensitivity.
Following acute oxaliplatin treatment we observed 46 differentially expressed genes in wild-type mice relative to saline treatment (Fig. 4.A). This included 26 upregulated and 20 downregulated genes. The number of differentially expressed genes in Wld S mice in response to oxaliplatin treatment was lower with a total of 4 upregulated and 9 downregulated genes (Fig. 4.A). Consistent with the absence of significant behavioural changes, only three differentially expressed genes were noted in Sarm1 − /− mice following treatment with oxaliplatin ( Fig. 4.A) and these genes were not shared with either C57BL/6 or Wld S mice (Fig. 4.B). This lack of transcriptomic perturbation in Sarm1 − /− mice is in keeping with their resistance to oxaliplatin-induced mechanical and cold hypersensitivity. Decreased expression of four transcripts correlated with oxaliplatin-induced pain across the three genotypes: Alas2, Hba-a1, Hba-a2, and Tfrc. (Fig. 4.C).

Discussion
Here, using a previously described protocol (Young et al., 2014), we report that acute oxaliplatin administration induces mechanical and cold hypersensitivities in wild-type (BL/6) and Wld S mice. This reflects acute peripheral neuropathies comprising cold-induced paraesthesia, dysaesthesia and pain that appear abruptly after administration in the clinic (Cersosimo, 2005). We associate acute oxaliplatin-induced pain with a plethora of transcriptional changes without loss of intraepidermal nerve fibers (IENFs) in BL/6 and Wld S mice. Genetic deletion of Sarm1 prevents development of these acute painful neuropathies and transcriptional changes are largely absent in DRGs from Sarm1 − /− -oxaliplatin-treated mice. The lack of detectable IENF loss in any genotype, and the absence of protection against pain in Wld S mice suggest a mechanism of pain generation independent from programmed axon death. In support of this, no known genes involved in programmed neurone or axon death were identified through our RNA sequencing. The common transcriptional changes occurring in mice that developed pain were downregulation of four iron metabolism and haem synthesis genes: Alas2, Hba-a1, Hba-a2, and Tfrc. A full list of gene expression changes observed in each genotype can be viewed in Supplementary Table 1. While Alas2, Hba-a1, Hba-a2, and Tfrc have known functions in erythrocytes, the absence of altered erythroid-specific gene expression, GATA binding protein 1 (Gata1) and erythroid-associated factor (Eraf), or endothelial specific genes, CD34 and Vascular Endothelial Growth Factor Receptor 2 (KDR), indicate that differential gene expression was not due to contamination by erythrocytes (Richter et al., 2009;Vanni et al., 2018). Interestingly, literature describes roles of ALAS2 and haemoglobin in central and peripheral neurones, including DRGs (Biagioli et al., 2009;Haraguchi et al., 2011;Richter et al., 2009;Stephens et al., 2019;Walser et al., 2017). Genetic mutations in Alas2, haemoglobin, and Tfrc genes have been observed in patients with painful or neurodegenerative conditions (Bishop et al., 2013;Kallianpur et al., 2014;Vanni et al., 2018): Haemoglobin in human brain (Biagioli et al., 2009;Richter et al., 2009) is differentially expressed in frontal cortex of humans with prion-related neurodegenerative diseases (Vanni et al., 2018). Mouse models of depression and chronic stress exhibit increased expression of Alas2 (Stankiewicz et al., 2015;Stankiewicz et al., 2014;Yamamoto et al., 2015). Animal models of pain show that chemotherapeutic paclitaxel causes downregulation of Hba and Hbb in rat DRGs (Nishida et al., 2008), and there is an association between chronic constriction injury-induced neuropathic pain with downregulation of Alas2, Hba-a1, and Hba-a2 (amongst other genes) in DRGs of female mice (Stephens et al., 2019). This evidence along with changes we noted in our acute OIPN model supports a role of Alas2, Hba-a1, and Hba-a2 downregulation in painful neuropathies. However, whether these are causal or reflect other pathogenic events is a crucial question for future studies to address.
Oxaliplatin accumulates in rat DRG mitochondria (Nishida et al., 2018) causing downregulation of mitochondrial electron transport chain proteins and increased reactive oxygen species (ROS) which is associated with oxaliplatin-induced mechanical allodynia (Cheng et al., 2019). Since Alas2, Hba-a1, and Hba-a2 are important for mitochondrial function, their downregulation could indicate mitochondrial stress. Indeed, mitochondrial toxin rotenone causes transcriptional downregulation of neuronal haemoglobin genes Hba-a2 and Hbb in rats (Richter et al., 2009). Removal of SARM1 protects against axon degeneration triggered by mitochondrial toxins rotenone and CCCP (Loreto et al., 2020;Summers et al., 2014) and sub-degenerative concentrations of CCCP induce SARM1 activation without causing degeneration (Sasaki et al., 2020). Taken together, we speculate that removal of SARM1 could protect against oxaliplatin-induced mitochondrial dysfunction associated with SARM1 activation in a context that leads to pain in the absence of axon degeneration.
Oxaliplatin forms protein adducts with many proteins (Graham et al., . Comparison: DESeq2 (differential gene expression analysis based on negative binomial distribution) saline vs. oxaliplatin treatment (n = 4 mice per treatment per genotype). Blue dots indicate significant changes (DESeq2) and red dots indicate significant changes that are also intensity hits. (B) Venn diagram highlighting overlapping DEGs between BL/6, Wld S and Sarm1 -− /− mice after treatment with oxaliplatin. (C) Common DEGs in BL/6, and Wld S mice following oxaliplatin treatment are highlighted. All differentially expressed genes are listed in Supplementary Table 1. 2000; Pendyala and Creaven, 1993) and oxaliplatin-haemoglobin adducts have been shown to affect cytoplasmic pH and sensitise TRP channels in DRGs (Peng et al., 2005;Potenzieri et al., 2020). Activation of TRP channels is associated with pain (Jang et al., 2018), including oxaliplatin-induced mechanical allodynia (Sakurai et al., 2009), and mice lacking TRPM2 exhibit decreased mechanical allodynia after paclitaxel (So et al., 2015). SARM1 products ADPR and cADPR (Essuman et al., 2017;Horsefield et al., 2019) can activate the TRPM2 channel (Yu et al., 2019). Therefore, another speculative pathway through which SARM1 may act in pre-degeneration pain pathophysiology could be TRP channel activation or sensitisation via SARM1 products. Finally, SARM1 is also reported as a positive regulator of cytokine and chemokine production in the innate immune system (Hou et al., 2013;Szretter et al., 2009;Wang et al., 2018). Acute oxaliplatin treatment increases C-C chemokine ligand 2 (CCL2) in small rat DRGs whereas its receptor (CCR2) is increased in medium and large DRGs, associated with mechanical hypersensitivity (Illias et al., 2018). However, a recent study questions the extent to which SARM1 is truly involved in mediating cytokine responses (Uccellini et al., 2020) and no altered chemokine RNA transcripts were associated with acute oxaliplatin administration in the present study. However, the heterogeneity of our DRG population could confound our ability to detect altered transcription that may occur in specific DRG sub-types, of which there are 11 possessing distinct mRNA profiles (Li et al., 2016;Usoskin et al., 2015).

Conclusions
The present study contributes data towards an emerging role of Alas2, Hba-a1, and Hba-a2 gene downregulation in pain pathophysiology and a role of SARM1 in acute chemotherapy-induced peripheral neuropathy. This work raises several interesting questions which future studies should address, including: Do these gene expression changes drive pain pathophysiology or are they a consequence of mitochondrial dysfunction or other pathological changes? What role does SARM1 activation have in the absence of programmed axon death? Are other axonal proteins such as tubulin or pro-survival protein NMNAT2 modified in the presence of oxaliplatin and do they play a causative role in CIPN?

Funding
This work was supported by the United Kingdom Medical Research Council project grant (MR/L003813/1); Biotechnology and Biological Sciences Research Council Institute Strategic Programme Grant; and the John and Lucille van Geest Foundation.

Declaration of Competing Interest
None.