Exercise rapidly alters proteomes in mice following spinal cord demyelination

Exercise affords broad benefits for people with multiple sclerosis (PwMS) including less fatigue, depression, and improved cognition. In animal models of multiple sclerosis (MS), exercise has been shown to improve remyelination, decrease blood–brain barrier permeability and reduce leukocyte infiltration. Despite these benefits many PwMS refrain from engaging in physical activity. This barrier to participation in exercise may be overcome by uncovering and describing the mechanisms by which exercise promotes beneficial changes in the central nervous system (CNS). Here, we show that acute bouts of exercise in mice profoundly alters the proteome in demyelinating lesions. Following lysolecithin induced demyelination of the ventral spinal cord, mice were given immediate access to a running wheel for 4 days. Lesioned spinal cords and peripheral blood serum were then subjected to tandem mass tag labeling shotgun proteomics workflow to identify alteration in protein levels. We identified 86 significantly upregulated and 85 downregulated proteins in the lesioned spinal cord as well as 14 significantly upregulated and 11 downregulated proteins in the serum following acute exercise. Altered pathways following exercise in demyelinated mice include oxidative stress response, metabolism and transmission across chemical synapses. Similar acute bout of exercise in naïve mice also changed several proteins in the serum and spinal cord, including those for metabolism and anti-oxidant responses. Improving our understanding of the mechanisms and duration of activity required to influence the injured CNS should motivate PwMS and other conditions to embrace exercise as part of their therapy to manage CNS disability.


Lysolecithin (lysophosphatidylcholine)-induced demyelination.
Demyelination was accomplished as previously described 26 . Mice were anaesthetized using ketamine (100 mg/kg) and xylazine (10 mg/kg) injected intraperitoneally. Skin overlying surgical site was shaved and disinfected with 70% ethanol and iodine. Ophthalmic gel was applied to both eyes to prevent drying, and buprenorphine (0.05 mg/kg) was injected subcutaneously immediately prior to surgery and 12 h post-surgery as an analgesic. Animals were positioned on a stereotaxic frame and a midline incision approximately 5 cm long was made between the shoulder blades using a #15 scalpel blade. A retractor was used to separate the muscle and adipose tissue to expose the spinal column. The prominent T2 vertebra was used as a landmark to find the T3-T4 intervertebral space. Tissue in the T3-T4 gap was then blunt dissected apart using forceps and spring scissors, and the dura was removed using a 30-gauge metal needle. Using a 32-gauge needle attached to a 10 µL Hamilton syringe, 0.5 µL of 1% lysolecithin/lysophosphatidylcholine (LPC) (Sigma-Aldrich, L1381) was injected into the ventral column of the spinal cord at a rate of 0.25 µL/min for 2 min. The needle was left in place for 2 min following the injection to avoid back flow, followed by suturing of the muscle and skin. Mice were then placed in a thermally controlled environment for recovery. Exercise paradigm. Immediately following recovery from surgery (~ 1-2 h post-surgery), exercising animals were singly housed in modified rat cages 19 with 5 inch running wheels mounted to the wire lid for 4 days. Wheels were connected to a computer system running a software to monitor wheel revolutions (developed in LabVIEW). Rotations per minute (rpm) were recorded in 10-min bins generating an average rpm over that 10-min period. Sedentary control animals were singly housed in modified rat cages with the running wheel placed on the ground as environmental enrichment control. Naïve animals were given access to running wheel cages at the same time as LPC animals.
Serum and tissue collection for proteomics. For the proteomics experiment, serum and lesioned spinal cords were collected from eight different female C57BL6 mice per group and two mice were pooled per sample: naïve without exercise (n = 4), naïve with exercise (n = 4), LPC lesioned without exercise (n = 4) and LPC lesioned with exercise (n = 4). Tissue was harvested 4 h into the dark cycle when animals had reached peak running for 2 h. Mice were anaesthetized using ketamine (100 mg/kg) and xylazine (10 mg/kg) injected intraperitoneally, and 500 µL of whole blood was drawn from the heart using a 1 mL syringe with a 25-gauge needle and collected in a 1.5 mL microcentrifuge tube. Mice were then transcardially perfused with 15 mL of room temperature phosphate-buffered saline (PBS) solution. Spinal tissue from the lower cervical to lower thoracic region was dissected. Tissue was placed on a dissecting microscope set up with dry ice and 2-methylbutane (Sigma-Aldrich) placed underneath the stage allowing for the tissue to be frozen to avoid protein degradation. Spinal cords were trimmed to 5 mm around the lesion site then cut into quarter-sections to include the lesion area. Lesioned quarter-sections from 2 mice were pooled into one sample in a 1.5 mL microcentrifuge tube to allow for sufficient protein levels and left on dry ice until harvesting was finished. Tissue was stored at − 80 °C until the proteomics experiments were performed. Tissue collection for immunohistochemistry followed a different methodology described below.
High performance liquid chromatography (HPLC) and mass spectrometry. All liquid chromatography and mass spectrometry experiments were carried out by the Southern Alberta Mass Spectrometry (SAMS) core facility at the University of Calgary, Canada. To assure consistency in data collection, HPLC and mass spectrometry data acquisition was replicated from our previous publication 27 . In detail, analysis was performed on an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific) operated with Xcalibur (version 4.0.21.10) and coupled to a Thermo Scientific Easy-nLC (nanoflow Liquid Chromatography) 1200 system. Tryptic peptides (2 μg) were loaded onto a C18 trap (75 μm × 2 cm; Acclaim PepMap 100, P/N 164946; Thermo Fisher Scientific) at a flow rate of 2 μL/min of solvent A (0.1% formic acid and 3% acetonitrile in LCmass spectrometry grade water). Peptides were eluted using a 120 min gradient from 5 to 40% (5% to 28% in 105 min followed by an increase to 40% B in 15 min) of solvent B (0.1% formic acid in 80% LC-mass spectrometry grade acetonitrile) at a flow rate of 0.3 μL/min and separated on a C18 analytical column (75 um × 50 cm; PepMap RSLC C18; P/N ES803; Thermo Fisher Scientific). Peptides were then electrosprayed using 2.3 kV voltage into the ion transfer tube (300 °C) of the Orbitrap Lumos operating in positive mode. The Orbitrap first performed a full mass spectrometry scan at a resolution of 120,000 FWHM to detect the precursor ion having a mass-to-charge ratio (m/z) between 375 and 1575 and a + 2 to + 4 charge. The Orbitrap AGC (Auto Gain Control) and the maximum injection time were set at 4e5 and 50 ms, respectively. The Orbitrap was operated using the top speed mode with a 3 s cycle time for precursor selection. The most intense precursor ions presenting a peptidic isotopic profile and having an intensity threshold of at least 2e4 were isolated using the quadrupole (isolation window of m/z 0.7) and fragmented with HCD (38% collision energy) in the ion routing Multipole. The fragment ions (MS2) were analyzed in the Orbitrap at a resolution of 15,000. The AGC, the maximum injection time and the first mass were set at 1e5, 105 ms and 100, respectively. Dynamic exclusion was enabled for 45 s to avoid of the acquisition of same precursor ion having a similar m/z (plus or minus 10 ppm).
Proteomic data and bioinformatics analysis. Spectral data were matched to peptide sequences in the murine UniProt protein database using the Andromeda algorithm 28 as implemented in the MaxQuant 29 software package v.1.6.0.1, at a peptide-spectrum match false discovery rate (FDR) of < 0.01. Search parameters included a mass tolerance of 20 p.p.m. for the parent ion, 0.5 Da for the fragment ion, carbamidomethylation of cysteine residues (+ 57.021464 Da), variable N-terminal modification by acetylation (+ 42.010565 Da), and variable methionine oxidation (+ 15.994915 Da). TMT 6-plex labels 126-131 were defined as labels for relative quantification. The cleavage site specificity was set to Trypsin/P (search for free N-terminus and for only lysines), with up to two missed cleavages allowed. Next, quantified proteins were filtered using the Perseus software 30 . Proteins that have more than two replicates with zero values in more than one group were removed. Filtered proteins were normalized using the CycLoess approach via NormalizerDE package 31 using the R language (v3.6.0) 32 . An average of the normalized results for each group was calculated and followed by the ratio of each group comparison. The ratios were log(2) transformed and the significant outlier cut-off values were determined after log(2) transformation by boxplot-and-whiskers analysis using the BoxPlotR tool 33  www.nature.com/scientificreports/ analyzed using the Mus musculus as our model organism at a false discovery rate of 5%. Proteins belonging to specific pathways were selected and their ratios were plotted as heatmaps.
Immunohistochemistry. For histology, lesioned spinal cords were collected from five different female C57BL6 mice per group: LPC lesioned without exercise (n = 5) and LPC lesioned with exercise (n = 5). The experiment was performed in duplicates, for a total of twenty mice. Tissue was harvested 4 h into the dark cycle when animals had reached peak running for 2 h. Mice were transcardially perfused with 15 mL of room temperature PBS followed by 15 mL of ice cold 4% paraformaldehyde (PFA). The spinal cord was dissected from the lower cervical to lower thoracic regions and post fixed in 4% PFA overnight at 4 °C, then cryoprotected in 30% sucrose for 3 days. Cords were cut into 5 mm sections with the lesion site in the centre and frozen in FSC 22 Frozen Section Media (Leica). Spinal cords were cut coronally into 20 μm sections on a cryostat (Thermo Fischer Scientific) onto Superfrost Plus microscope slides (VWR). Sections were stored at − 20 °C until staining and analysis. Slides with coronal sections were warmed to room temperature for 5 min. For myelin basic protein (MBP) staining, slides were delipidated with concurrent washes of 50%, 70%, 90%, 95%, 100%, 95%, 90%, 70%, and 50% ethanol. Slides were rehydrated for 10 min in PBS, then permeabilized with 0.2% Triton-X100 in PBS for 10 min. Slides were then blocked for 1 h at room temperature using horse blocking buffer (PBS, 10% horse serum, 1% BSA, 0.1% cold fish stain gelation, 0.1% Triton X-100, 0.05% Tween-20) and IgG Fab fragments (20 µg/mL). Staining for NFH, Olig2, PDGFRα, CC, and CX32 did not require delipidating steps. After blocking, primary antibodies were added in antibody dilution buffer (PBS, 1% BSA, 0.1% cold fish stain gelation, 0.1% Triton X-100) overnight at 4 °C. Sections were washed three times with 0.05% Tween-20 in PBS for 5 min each, followed by 1 h of incubation with secondary antibodies in antibody dilution buffer (1/400 from manufacturer stock) and Nuclear Yellow (1/800) at room temperature. Sections were then washed three more times with 0. Confocal microscopy. Confocal images were taken on the Leica TCS Sp8 laser confocal microscope using the 10 × air objective and 25 × 0.5 NA water objective. The 405 nm, 488 nm, 552 nm, and 640 nm lasers were used to excite fluorophores with detection by 2 low dark current Hamamatsu PMT detectors and two high sensitivity hybrid detectors. Images were acquired in 8-bits, in a z-stack bidirectionally, with 2-times line average, and at 0.57 μm optical sections. Laser levels were maintained across all sections of common stains, and secondary alone controls were used to account for non-specific secondary fluorescence. Images were blinded and analyzed using ImageJ (RRID: SCR_003070).
Statistical analysis. For sample size calculation, it was used previously published studies as a reference 19 , using as a baseline the least sensitive method. The Shapiro-Wilk test was used to assess data normality and the F-test was used to evaluate if the variances of two groups are equal. Normal data were evaluated using the Student t-test, while non-normal data was assessed using the Mann-Whitney test. For the proteomics analysis, it was applied an interquartile boxplot analysis to determine the proteins differentially expressed 33 . The FDR was generated by the bioinformatic tools used for pathway analysis. The R software 34 , GraphPad Prism version 9, and Microsoft Excel were used for the statistical analysis.

Ethical standards.
Our study was carried out in compliance with the ARRIVE guidelines.

Results
Exercise in naïve mice alters the spinal cord proteome. To evaluate the impact of exercise on the CNS and serum, we subjected mice to short-term bouts of exercise and compared to sedentary control. Running animals had free access to a running wheel over 4 days, while sedentary animals were given a locked wheel as environmental enrichment control. Mice ran during their night cycle and rested during the day as determined from the number of runs per 10 min bin (Fig. 1A,B). Four hours into day 4 and during the animal's active night cycle, serum and spinal cords at T3-T4 were collected and subjected to a quantitative shotgun proteomics workflow labelled with tandem mass tags (TMT) 6-plex kits (Fig. 1A). Mice subjected to LPC-surgery had reduced exercise activity in the first and second day after surgery but attained similar levels by the end of the third day (Fig. 1B); over the 4 days, LPC-surgery mice ran significantly shorter distances than the naïve animals ( Fig. 1C). We first examined protein changes in non-lesioned naïve mice that exercised on the running wheel to determine if short term exercise could alter the CNS proteome. Compared to their sedentary naïve controls, there were 115 significantly upregulated proteins and 67 significantly downregulated proteins in the spinal cord following exercise as determined by an interquartile boxplot analysis ( Fig. 2A,B; Supplementary Tables 1-2). Pathway enrichment analyses were performed using STRING 35  In the serum of naïve exercising animals there were 9 upregulated proteins including those related to detoxification of oxygen species (Gpx3: glutathione peroxidase 3, Prdx2: peroxiredoxin-1), and platelet degranulation (Fgb: fibrinogen beta chain, Ecm1: extracellular matrix protein 1) (Supplementary Figure 2A Interquartile box plot analysis was used for statistics. Data analysis was accomplished using the R software 32 . The plot was generated using the heatmap.2 function from the gplots package 34 . www.nature.com/scientificreports/ LPC into the spinal cords of mice 26 and compared to naïve uninjured mice. LPC-injected mice were placed in individual cages with a running wheel immediately after recovery from surgery in the LPC group. Following LPC injection, there was demyelination as verified by eriochrome cyanine staining, and a loss of mature OLIG2 + CC1 + oligodendrocytes within the lesion; the number of OLIG2 + PDGFRα + oligodendrocyte progenitor cells (OPC) in the lesion appeared increased at 3 days post lesion (dpi) compared to the contralateral uninjured side (Fig. 3A,B). There was increased detection of MBP positive myelin debris, visible loss of axons, and a lack of GFAP positive astrocytes within the lesion (Fig. 3C). These results affirm the creation of a demyelinated lesion against which exercise in previous experiments 19 promoted oligodendrogenesis at later time points, and they provide the substrate to evaluate the proteomics of exercise following demyelination.
In the serum of exercising animals, proteins associated with the innate immune system were downregulated (Orm1: Alpha-1-acid glycoprotein-1, Orm2), and anti-oxidant proteins (Sod1: superoxide dismutase 1, Prdx2: peroxiredoxin-2) were upregulated (Fig. 5C, Supplementary Figure 4). This shows that acute amounts of exercise are sufficient to profoundly alter the proteome of the spinal cord and serum in both naïve and LPC demyelinated animals.

Validation by immunofluorescence microscopy of decrease of connexin-32 and myelin basic protein in 4-day lesion of exercising animals.
We sought to validate the findings of proteomics by immunofluorescent staining of coronal spinal cord sections. An antibody each to a marker of metabolism (Pgam2) and anti-oxidant enzyme (Mgst1) was first attempted but signals of immunofluorescence were unconvincing in any sections. We next chose antibodies to structural proteins such as the gap junction protein, connexin-32, and MBP, which have been pointed out above to be reduced in exercising demyelinated mice. Using immunofluorescence, we found that while lesion size at 4 days post-LPC did not differ between exercising and sedentary groups (Fig. 7A,B), there was a significant decrease in the amount of both connexin-32 and MBP immunoreactivity within the lesion environment of exercising animals (Fig. 7C,D). As MBP immunoreactivity within the LPC lesion at early time points post-injury (and which is devoid of eriochrome cyanine as in Fig. 3) is indicative of myelin debris 37 , the decrease in MBP in lesion of exercising mice may represent the more rapid clearance of myelin debris that is inhibitory to repair processes.

Discussion
There is unanimity that exercise benefits general health. For brain health, several reports demonstrate that exercise modulates the brain's connectome and even its structure [38][39][40] . There is now an extensive literature that exercise improves cognition, fatigue, depression and mood in PwMS 3,7 . In animal studies, outcomes of exercise in healthy mice or those modeling neurological conditions include reduced neuroinflammation or blood-brain barrier breakdown, neurogenesis, oligodendrogenesis, neuroprotection and remyelination 1   www.nature.com/scientificreports/ by which exercise promotes CNS wellbeing appear to be multiple, such as the generation of brain-derived neurotrophic factor, reducing pro-inflammatory responses, modulating microglia activity, and ameliorating oxidative stress (reviewed in 12,14,41,42 ). Endurance (treadmill running) training in rodents decreased oxidative stress in the spinal cord, while increasing levels of the Nrf2 transcription factor and downstream anti-oxidant enzymes 43 . Exercise also promotes neurotransmission and multiple signaling pathways in brain cells 17,42 . Molecular mediators of the benefits of exercise include muscle-derived myokine FNDC5/irisin that prevents neurodegeneration and rescues memory impairment in a model of Alzheimer's disease, and liver-derived glycosylphosphatidylinositolspecific phospholipase D1 that improves cognition in aged mice 44,45 .
Despite the increasing knowledge of the mechanisms of exercise, it is not known how quickly they are affected. Here, using quantitative shotgun proteomics, we report that a short bout of exercise changes numerous proteins in blood and spinal cord of naïve and demyelinated mice (Fig. 8). Running activity of demyelinated mice was low to modest in the first 2 days and was similar to naïve controls only by the fourth day. Thus, the proteomic changes from tissues harvested 4 h into running at day 4 ( Fig. 1) are reasonably the result of less than 2 days of running. Unanswered questions are whether the proteomic changes would be affected after a single running episode, or whether some would be long-lasting after exercise is stopped. Regardless, with a short bout of running, the results are remarkable, as over 150 proteins are up-or down-regulated in the spinal cord of both naïve and demyelinated mice. Fewer altered proteins are detected in the serum, likely the result of our technical detection of a smaller number of proteins (248) in the serum compared to the spinal cord (1920 proteins) of LPC demyelinated mice (Figs. 4A, 5A).
The quick alteration of numerous proteins in the CNS may help explain the brain changes in humans soon after exercise. For example, a single 50-min vigorous intensity cycling on a stationary bicycle in healthy subjects improved their prefrontal cortex-dependent cognition that lasted up to 2 h after exercise 46  www.nature.com/scientificreports/ of harness-supported treadmill walking in highly progressed PwMS increases brain excitability, which indicates higher potential for neuroplasticity, and it is more pronounced in individuals with higher aerobic fitness. Although this feature was observed only on the hemisphere less affected by the disease, it shows the importance of exercise in PwMS and the potential for changes in the CNS even following acute exercise 47 .
In evaluating the changes to the proteomes in the spinal cord of demyelinated mice after exercise, we found that markers of anti-oxidant responses, metabolism, neurotransmission and proteolytic remodeling of the extracellular matrix are commonly elevated by exercise. The last is notable since subsequent repair after injury would require extracellular matrix remodeling 48 . Elevation of anti-oxidant responses are also notable. Markers of oxidative stress are increased within MS lesions where they have been implicated in axonal damage and demyelination 1,49,50 . In the EAE model of MS, the primary anti-oxidant glutathione is upregulated following voluntary wheel running 51 . Consistent with this report, we found elevation of Mgst1 and Ggt7 in the exercising LPC-demyelinated spinal cord. Mgst1 has both glutathione transferase and peroxidase functions which makes it particularly potent at attenuating oxidative stress injuries 52,53 . Thus, exercise-induced upregulation of these antioxidants may help mitigate injury-related oxidative damage. We attempted to corroborate these findings by looking at markers of oxidative injury such as malondialdehyde (   www.nature.com/scientificreports/ and 4-hydroxynonenal (4-HNE). Due to limitations with high antibody background staining and thus specificity we were unable to corroborate differences in overall oxidative stress. However, we hypothesize that any difference would likely be seen later in lifespan of the lesion after these oxidative stress response proteins have had time to function. We also found increased expression of Sod1 and Prdx2 in the serum of exercising mice. Decreased total antioxidant capacity has been noted in the serum of PwMS 54,55 . Taken together, the combined spinal cord and serum upregulation of antioxidants suggests that exercise produces the necessary components to handle the pro-oxidants commonly seen in MS. Changes in metabolism influence cell function 56,57 . Metabolic reprogramming such as the reliance of aerobic glycolysis for rapid energy supply occurs in cells when responding to injury and infection [57][58][59] . For example, macrophages rely on aerobic glycolysis to transmigrate into the parenchyma from perivascular cuffs in the EAE model 60 . Additionally, microglia are thought to use oxidative phosphorylation for homeostatic or tissue repair functions 61,62 but switch to aerobic glycolysis when pro-inflammatory 63 . Our proteomic analysis identified significant changes in metabolism-related proteins. For example, upregulation of Pgam2 suggests a general rise in glycolysis while increased expression of proteins associated with mitochondria (Ndufb9, Ndufv3, mt-Nd3) suggests elevated oxidative phosphorylation. These metabolic changes could be downstream of the transcriptional co-activator PGC1α (peroxisome proliferator-activated receptor gamma coactivator 1-alpha), which mediates exercise-enhanced remyelination in mice 19 .
Impressively, exercise in naïve mice also elevated numerous proteins (115) in the naïve spinal cord of mice, with smaller changes (9 elevated) noted in the serum. This finding may help account for the benefit of exercise not only for general health, but also for CNS wellbeing in the absence of any pathology. Common pathways elevated by exercise in the naïve or demyelinated CNS include those related to anti-oxidants and metabolism. Namely, the upregulation of metabolism upon exercise was identified by 16 and 18 proteins in the LPC (FDR = 3.16 × 10 -2 ) and naïve (FDR = 1.29 × 10 -2 ) samples, respectively. Among the 34 proteins, they shared only Ggt7, Mgst1 and Rbp1, indicating that exercise has a wide effect on protein expression in the spinal cord tissue, modulating the levels of multiple proteins within a specific biological function. For the proteins commonly enriched in sedentary mice, Fxyd6 had the highest log fold change, -1.56 naïve and -0.93 LPC. Experimental evidence indicates that the FXYD family of transmembrane proteins can associate and modulate the transport properties of Na,K-ATPase 64,65 , an enzyme responsible for regulating the Na + and K + gradients across the cell membrane 66 . Dysfunction of Na,K-ATPase may play a central role in MS, as loss of axonal Na,K-ATPase is linked to neurological decline in chronic stages of the disease 66 . Although the exact impact of Fxyd6 on Na,K-ATPase is not completely known, www.nature.com/scientificreports/ its regulation by exercise, independently of LPC, suggests a beneficial and protective role for PwMS. These findings may also help explain why MS is less common in those that exercise versus those that do not, supporting the suggestion that exercise may prevent MS 7,67 . We note that while several pathways have been identified by Metascape and STRING analyses for exerciseinduced changes, and several proteins have been highlighted, these proteins may serve several functions in different pathways. For instance, Mgst1 is highlighted in both anti-oxidant and metabolism pathways in our analyses. Also, not all existing proteins have been identified in the serum and spinal cord in this study, a limitation of mass spectrometry, so other exercise-induced changes remain to be discovered.
When comparing the proteomics findings to the transcriptomics results from a previous study published by our group 19 , it is not possible to see a clear correlation between both techniques, even though it was analyzed in the same type of LPC-induced demyelinating lesions. However, this lack of correlation may be explained by the number of days post lesion (dpl) analyzed in each experiment, as the proteomics approach was performed on 3 dpl and the transcriptomics analyzed samples at 5 and 10 dpl. Nonetheless, proteomics identified 4 members of the mitogen-activated protein kinases (Mapk) family (Mapk1, Mapk3, Mapk15, and Mapk10) with minimal alterations in their fold change, while the ERK/MAPK signaling pathway was enriched at 5 dpl on the transcriptomics data and presented the lowest p-value (2 × 10 -5 ). This suggests that the MAPK pathway is part of a delayed response induced by exercise and most likely to be detected at the protein level on day 5 and onwards.
LPC demyelination provides a useful model for subsequent remyelination 26 . However, it models the repopulation of oligodendrocyte lineage cells after demyelination in MS and does not replicate every facet of the disease. We acknowledge that this is not a classical immune mediated lesion as is seen in MS or EAE, but rather a toxin induced injury. LPC has been useful because of the well characterized injury and a reliable course of resolution. Following cell death and lesion formation there is a rapid increase in the innate immune response by day 3. Some limitations of this model include the dearth of an adaptive immune response. While lymphocytes have been shown to be present early in lesion formation, the short lifespan and focal nature of the lesion restricts the immune response to an innate one. By day 7 following lesion formation, the repair process has begun and OPCs accumulate in the lesion. By two weeks, remyelination begins to occur and is robust by three weeks. For the purposes of this study, to investigate what changes exercise induces in naïve and demyelinated tissue, LPC is a useful model. LPC produces rapid and complete demyelination within the first day of application 68 , and the eriochrome cyanine staining shows a complete loss of intact myelin at day 4 (Fig. 3A). Nonetheless, myelin debris is still manifest within the lesion at this point, as noted by MBP immunoreactivity (Fig. 7) within the lesion that is intense due presumably to exposure of antibody epitopes in the degraded myelin. In previous work, we found that exercise promotes myelin debris removal by 3-4 days of injury 19 , likely resulting in the lower MBP immunoreactivity in the lesion of exercising mice herein (Fig. 7). This is important in demyelinating injuries in which myelin debris is inhibitory to OPCs 37,69 and would need to be cleared. However, we did www.nature.com/scientificreports/ not find phagocytosis-associated proteins to be elevated in the exercise LPC group, but endocytosis-associated proteins were increased (Chmp5: Charged multivesicular body protein 5, Dnm1: Dynamin-1, Ehd1: EH domaincontaining protein 1, Epn1: Epsin-1, Git1: ARF GTPase-activating protein GIT1, and Stam: Signal transducing adapter molecule 1) (Supplementary Table 1).
In the lysolecithin-injury, OPCs begin repopulating within 3 days but maturation to oligodendrocytes occurs later (approximately day 7). This may reconcile the result that Myrf, a transcriptional factor expressed in mature myelinating oligodendrocytes that promotes the expression of myelin genes 70 , is downregulated in the LPC spinal cord by exercise, as the day four post-LPC time point is not yet timely for myelin formation. Conversely, Myrf is elevated by exercise in the naïve spinal cord (Supplementary Figure 1), suggestive of activity-dependent myelin formation in the homeostatic CNS 17 .
In summary, we provide a protein level landscape of how exercise alters the CNS. The rapid elevation of numerous proteins involved in several pathways, particularly anti-oxidative and metabolic responses, not only following demyelination but also in the healthy state, is remarkable (Fig. 8). These rapid changes help reconcile the observations that short bouts of exercise can influence the brain. A finer dissection of the crucial proteins may provide direction for future studies of exercise to promote repair responses in MS and other neurological conditions. Finally, the profound changes to the proteome induced by exercise may provide a lesion milieu particularly conducive for a pro-remyelinating medication to act upon. This integration of Medication and eXercise or "MedXercise" may lay the foundation for future strategies to maximize regeneration of the injured CNS.

Data availability
The proteomics data are publicly available and are deposited in PRIDE Archives, accession number: PXD020782. www.nature.com/scientificreports/