Neuronal LXR Regulates Neuregulin-1 Expression and Sciatic Nerve-Associated Cell signaling

Neuropathic pain caused by peripheral nerve injury significantly affects sensory perception and quality of life. Accumulating evidence strongly link cholesterol and inflammation with development and progression of Obesity and Diabetes associated-neuropathies. However, the exact mechanisms of how lipid metabolism in peripheral nervous system (PNS) contributes to the pathogenesis of neuropathy remains poorly understood. Dysregulation of LXR pathways have been identified in many transcriptomic analyses in neuropathy models. LXR α/β expressed in sensory neurons are necessary for proper peripheral nerve function. Deletion of LXR α/β from sensory neurons lead to pain-like behaviors. In this study, we identified that LXR α/β expressed in sensory neurons regulates neuronal neuregulin-1 (Nrg1). Using in vivo cell-specific approaches, we observed that loss of LXR from sensory neurons altered genes regulating lipid metabolism in non-neuronal cells potentially representing Schwann cells (SC). Our data suggest that neuronal LXR may regulates SC function via a Nrg1-dependent mechanism. The decrease in Nrg1 expression in DRG neurons of WD-fed mice may suggest an altered Nrg1-dependent neuron-SC communication in Obesity. The communication between neurons and non-neuronal cell such as SC could be a new biological pathway to study and to treat Obesity-associated neuropathy and PNS dysfunction.

7 synthesis (Srebf1/Srebp-1c), and confirmed that their expressions were increasing following 1 LXR activation (Figure3E). These data demonstrate that LXRs are transcriptionally active in 2 DRG neurons. To test the activity of LXRs in the presence of saturated fatty acid, transduced 3 neurons with the lentivector reporter were treated with palmitic acid followed by GW3965 4 stimulation. Palmitate treatment significantly lowered the increase in luminescence induced by 5 GW3965 (Figure3D). These findings confirm that LXR activity and its canonical gene 6 expressions are altered in Obesity in neurons as in many other tissues 16,40,41 . Next, we hypothesized that LXR could regulate Nrg1 gene expression. To evaluate the effect of 9 LXR activation on the Nrg1-type III expression in normal and in an obesogenic environment, we 1 0 treated mice fed either NC or WD for 8 weeks with LXR agonist, GW3965 (25mg/Kg of body 1 1 weight) twice a week for 3 weeks. Compared to NC-fed mice, DRG of WD-fed mice had lower 1 2 level of Nrg1-type III mRNA. GW3965 administration increased Nrg1-type III mRNA level in the To test whether GW3965 regulates Nrg1-type III expression in sensory neurons, NC or WD-fed 2 0 RiboTag-Nav mice or palmitate-stimulated DRG organotypic cultures from RiboTag-Nav mice 2 1 were treated with GW3965 as describe above. Compared to NC-fed RiboTag-Nav, sensory and GW3965 treatment attenuated this decrease (Figure4D,E). Overall data suggest that LXR 2 4 α /β regulate Nrg1 expression in the sensory neurons of the DRG. Our findings also suggest that 2 5 the decrease in Nrg1 mRNA level observed in vivo in the nociceptors of obese mice i) may be 2 6 8 the consequence of a decrease in LXR activity(induced by diet and/or saturated fatty acids) and 1 ii) could be rescued by GW3965. processed to assess the mRNA levels of LXR canonical pathway and Nrg1. As shown in 1 2 Figure5B, Abca1 was decreased more than two fold in both NC and WD-fed LXRabnav mice.

3
Srebf1/Srebp-1c were decreased in DRG of WD-fed mice but in contrast to abca1, no difference 1 4 was observed between LXRab and LXRabnav suggesting that, in sensory neurons, LXR does 1 5 not drive Srebf1/Srebp-1c. Interestingly, Nrg1-type III was significantly reduced when LXRs are 1 6 absent from sensory neurons demonstrating that endogenous LXRs regulate Nrg1-type III 1 7 expression in sensory neurons in vivo and that WD feeding may alter this regulation (Figure5B). As mentioned above, DRG contains cell bodies of axons that compose the sciatic nerve and are 2 0 associated with myelinating or non-myelinating SC. As Axonal Nrg1 is known to determine SC 2 1 function 26,27 , we isolated sciatic nerves from LXRab and LXRabnav mice to assess gene 2 2 expression involved in lipid metabolism that were previously shown to be downstream of 2 3 Nrg1/ErbB 25-27,30 : Srebf1/Srebp-1c, myelin protein zero (MPZ), and peripheral myelin protein 2 4 (PMP22) were evaluated. We rigorously dissected the nerve to harvest the same piece of tissue 2 5 in each experiment. In pilot data, the number of cells isolated were counted, the total number of 2 6 9 total cells were similar between all experiments (not shown). Interestingly, Srebf1/Srebp-1c 1 expression increased in WD-fed mice sciatic nerve compared to NC-fed mice (Figure5C). We 2 did not observe any differences between genotypes suggesting that Srebp1f change following 3 WD in nerve-associated cells is independent of neuronal LXR (Figure5C). We observed an 4 increase of Mpz and Pmp22 mRNA in sciatic nerve of LXRabnav mice compared to control 5 mice. Compared to LXRab, we also observed a decrease in ErbB2 and ErbB3 mRNA in the 6 sciatic nerve of LXRabnav mice (Figure5D). This decrease was exacerbated when fed WD 7 (Figure5D). These data suggest loss of LXR in sensory neurons may affect signaling in nerve-8 associated cells and may change myelin structure. We evaluated the sciatic nerve of LXRab 9 and LXRabNav mice using electron microscopy, but we did not see any significant changes in 1 0 SC number or structure in either Remak bundles (non-myelinated SC) or myelinating SC 1 1 associated with the nerve (Figure5E). It is also possible that the change in gene expression is 1 2 seen at an earlier time point than that of the change in structure or that the EM fails to capture 1 3 these changes. Cell-specific models (e.g -cre) allowing to specifically purify SC non-myelinating 1 4 and myelinating would be necessary to quantify in vivo the numbers of all SC and also to 1 5 directly purify mRNA. Unfortunately, these models do not exist, the current knowledge on SC  Human studies showed correlation between circulating cholesterol and lipids in development ligand-activated nuclear receptors that bind metabolites of cholesterol 18,19 . While LXRα/β is 2 4 described in many reports as being an important pathway involved in many neurodegenerative 2 5 diseases 36,37 , its specific role in CNS and PNS cells remains unclear and likely pleiotropic. Our 2 6 current data suggest that LXRα/β may drive a unique transcriptional program regulating neuron-1 nerve-associated cells (likely SC) interactions that may sustain peripheral nerve function. Our 2 findings using mice models (WD-fed and LXRα/β sensory neuron specific deletion) and ex vivo 3 (DRG organotypic and primary neuron cultures), show that WD alters LXR activity that changes 4 Nrg1 expression leading to neuron-SC communication impairment in the PNS. LXR activation 5 had been shown to improve WD-induced neuropathy 16 . It is possible that LXR/Nrg1/ErbB 6 pathway explain in part this improvement. functionally modulated by Nrg1. ErbB2 and ErbB3 receptors are required for signal transduction in the Schwann cells, but 2 4 ErbB3, but not ErbB2, binds extracellular ligands with high affinity, however ErbB3 is 2 5 catalytically inactive, and ErbB2 contributes to the tyrosine kinase activity essential for  provide another mechanism to explore and understand the role of LXR in the SC function. Previous studies from others and our group showed that signaling pathway in the neurons of the 1 9 DRG is disrupted inducing cellular stress 16,32,38,47 . Our previous data showed that activation of 13 total) (Envigo, Indiana, USA) for 12 weeks starting at weaning 16,17 . All studies mentioned were 1 done exclusively using male mice to avoid confounding effect of hormones with experimenter 2 blinded to both treatment and genotype.  In vivo agonist treatment. WT and RiboTag mice were treated with vehicle or LXR agonist 1 6 (GW3965; 25mg/kg BW) (Axon Medchem, Virginia, USA) by i.p. twice a week for 3 weeks 1 7 starting at 8 weeks on WD as reported before 16 . Tissues were rapidly dissected and frozen in 1 8 liquid nitrogen before analysis. Tissue from RiboTag mice were harvested and processed as 1 9 detailed below. replacing every other day in a 37ºC and 5% CO 2 incubator. After an overnight incubation in low 2 5 serum (2.5%) MEM supplemented with GlutaMAX (2mM), DRG were stimulated with either 2 6 14 vehicle or 15µM GW3965 for 24 hrs before palmitate treatment (400 µM) for another 24hrs as 1 described previously 16 . RNA was extracted using Acturus PicoPure RNA Extraction Kit (Applied 2 Biosystems, California, USA).

4
Enrichment of transcripts from sensory neurons. DRG from RiboTag+/+:Nav1.8Cre+/-mice 5 were either freshly harvested for RNA isolation or harvested to perform organotypic culture 6 followed by RNA isolation. To isolate RNA associated with HA-tagged ribosomes in sensory 7 neurons, immunoprecipitation (IP) followed by mRNA purification following the procedure 8 published by Sanz et al. was used 16,49 . Briefly, DRG were homogenized in homogenization 9 buffer and supernatant removed after centrifuging at 10,000g for 10 min at 4ºC. 10% of the 1 0 homogenate was saved (input) for mRNA isolation. Remaining volume was incubated at 4ºC 1 1 with anti-HA antibody (Biolegend, #901513) at 1:150 dilution for 4hrs on a gentle spinner. This is 1 2 followed by an overnight incubation at 4ºC on a gentle spinner with above sample transferred to axotomized DRG were then transferred to a collagenase A/trypsin mix (1.25mg/ml each) and 2 5 incubated for 30min. Partially digested DRG were then passed through fire polished glass 2 6 pipettes followed by 3min spin at 3000g. After careful removal of supernatant, cells were 1 resuspended in advanced DMEM with 10% FBS and 4mM GlutaMAX, and plated onto a poly-l-2 lysine coated plates. Neuronal cultures were maintained in a 37ºC and 5% CO 2 incubator for 3-4 3 days changing above media supplemented with Ara-C (2µM) to inhibit replicative cells every 4 other day before treating the cells to extract RNA as described above.  Total RNA was extracted from Sciatic nerve of NC-or WD-fed mice with Arcturus PicoPure RNA 1 7 isolation kit (Applied Biosystems). Two biological replicates were used for each group. Total 1 8 RNA was quantified by Qubit and assessed for quality on an Agilent Bioanalyzer using Total 1 9 RNA Pico Chip. Total RNA samples that passed QC were used as input for library construction. Analysis done with STAR and DESeq2. The quality of reads, in FASTQ format, was evaluated 1 using FastQC. Reads were trimmed to remove Illumina adapters from the 3' ends using 2 cutadapt 52 . Trimmed reads were aligned to the Mus musculus genome (mm10) using STAR 53 .

3
Read counts for each gene were calculated using htseq-count 54 in conjunction with a gene 4 annotation file for mm10 obtained from Ensembl (http://useast.ensembl.org/index.html).

5
Normalization and differential expression were calculated using DESeq2 that employs the Wald 6 test 55 . The cutoff for determining significantly differentially expressed genes was an FDR-7 adjusted p-value less than 0.05 using the Benjamini-Hochberg method. and RB were involved in data collection, assembly, analysis and interpretation of data. CKG and 2 5 VM-A drafted the manuscript. CKG, VM-A, and RB reviewed the manuscript. The authors declare no competing interests.   RiboTag-Nav mice) stimulated with or without palmitate (vehicle group treated as 100%) (n=3 4 experiments in triplicate). All data are Mean±S.E.M. *p<0.05; **p<0.005; ***p<0.0005 GW3965 with NC group treated as 100% (n=6/group). B) Nrg1-type III mRNA level in 2 2 organotypic culture of DRG stimulated with or without palmitate and treated with or without 2 3 GW3965 (vehicle group treated as 100%) (n=4 experiments in triplicate). C) Nrg1-type III mRNA 2 4 level in primary neuronal culture of DRG stimulated with or without palmitate and treated with or III mRNA level in DRG sensory neurons of RiboTag-Nav mice fed either NC or WD and treated 1 with or without GW3965 with NC-veh group treated as 100% (n=6/group). E) Nrg1-type III 2 mRNA level in organotypic culture of DRG sensory neurons (from RiboTag-Nav mice) 3 stimulated with or without palmitate and treated with or without GW3965 (vehicle group treated 4 as 100%) (n=3 experiments in triplicate). All data are Mean±S.E.M. *p<0.05; **p<0.005; 5 ***p<0.0005