Voluntary Exercise Ameliorates Neuropathic Pain by Suppressing Calcitonin Gene-Related Peptide and Ionized Calcium-Binding Adapter Molecule 1 Overexpression in the Lumbar Dorsal Horns in Response to Injury to the Cervical Spinal Cord

Neuropathic pain (NP) is a frequent nding in patients diagnosed with spinal cord injuries (SCIs). We aimed to investigate the effects of voluntary exercise on NP after SCI and to elucidate its potential mechanisms. A rat model of post-SCI NP induced by compression of the posterior or lateral cervical spinal cord was used to evaluate the effects of voluntary exercise by measuring the bilateral withdrawal of the hind paws using the Von Frey lament and Hargreaves tests. The place escape/avoid paradigm was used to evaluate supraspinal pain processing and somatosensory evoked potentials (SEPs) were used to examine disturbances in proprioception. Locomotor function was evaluated using Basso, Beattie, and Bresnahan (BBB) scoring. Pathologic ndings in hematoxylin and eosin-stained tissue and magnetic resonance imaging were used to evaluate the morphological changes after SCI. The lesion size within the cervical spinal cord was evaluated by staining with Eriochrome cyanine R. Quantitative polymerase chain reaction and immunohistochemistry were used to assess the expression of calcitonin gene-related peptide (CGRP) and ionized calcium-binding adapter molecule 1 (Iba-1) in the lumbar dorsal horns. dorsal horns of rats exhibiting symptoms of NP suggests that microglial activation might play a crucial role in its development. Collectively, voluntary exercise may be a promising therapeutic modality to treat NP that develops clinically in response to SCI.


Introduction
Spinal cord injury (SCI) is a devastating neurological event that is characterized by motor, sensory, and autonomic impairment of varying severity. Neuropathic pain (NP) is one of the most common accompanied by reductions in CGRP and ionized calcium-binding adapter molecule 1 (Iba-1) expression in the lumbar dorsal horn. Methods

Animal subjects
All experimental procedures were approved by the Research Ethics Committee of Sun Yat-sen University, Guangzhou, China, and conformed to all relevant regulatory standards. The ethic number is [2020]015.
Eighty-six female Sprague-Dawley rats (initial weights, 250-300 g, average weight 287 g from the Laboratory Animal Center of Sun Yat-sen University, Guangzhou, China) were used to carry out the experiments. Rats were provided with access to food and water ad libitum and housed in groups of four per cage on a 12/12-hour light-dark cycle. Two rats were sacri ced before completion of the experimental protocol because of an anesthesia accident. The remaining 84 rats were divided into six groups: (1) no compression (sham, n = 12), (2) posterior spinal cord compression injury (P-SCI, n = 15), (3) lateral spinal cord compression injury (L-SCI, n = 15) groups, (4) sham with voluntary exercise (sham + V, n = 12), (5) P-SCI with voluntary exercise (P-SCI + V, n = 15), and (6) L-SCI with voluntary exercise (L-SCI + V, n = 15). Individual rats were assigned to voluntary exercise or no voluntary exercise groups based on paw withdrawal responses to the 2g von Frey hair lament at 1 week post-injury (1wpi) in order to create matched groups of exercised and not exercised animals. All rats were sacri ced 4 weeks after surgery.

Spinal cord injury
Rats in the sham, P-SCI, and L-SCI groups were anesthetized with a cocktail (3 mg/kg) containing ketamine (31.25 mg/kg), xylazine (1.58 mg/kg), and acepromazine (0.31 mg/kg) (Guangzhou FISCLAB Environ. Sci-Tech. Co., Ltd., Guangzhou, China). Brie y, following exposure of the C4-C6 spinal process and laminae via a posterior incision, a C5 laminectomy was performed to gain access to the epidural space. A sheet (1 × 3 × 5 mm 3 ) of water-absorbable polyurethane polymer was implanted into the C6 epidural space in the dorsal region of the spinal cord of the P-SCI group. For the L-SCI group, the sheet was implanted into the C6 epidural space in the lateral region of the spinal cord. For the sham group, only C5 laminectomy was performed. After surgery, all rats received a subcutaneous (s.c.) injection of Ringer's solution (200 µl) to prevent dehydration during the surgical procedure. Post-surgical care included close observation, manual bladder voiding twice per day, and treatment with antibiotics to prevent bladder infection (100 µl ampicillin via s.c. injection twice per day) for about 7 days after surgery until re ex bladder function was restored. All surgical procedures were performed by the same experienced investigator.

Voluntary exercise
Running wheels were placed in individual cages (Huaibei Jiubai Electronic Technology Co., Ltd. China) of the sham + V, P-SCI + V, and L-SCI + V groups at 1 wpi. Identical running wheels that were locked in place (i.e., immobile), were placed into the cages of the control groups (sham, P-SCI, and L-SCI). A counter was connected to the wheel that recorded the frequency of running sessions during a 24 h period. The distance covered during each session was approximately 1 m. In all voluntary exercise groups, rats had free access to the running wheel freely over a 3-week training period. In the P-SCI + V and L-SCI + V groups, the average amount of exercise for 3 weeks was approximately 20,155 m between all rats and approximately 22,349 m in the sham + V group. Rats were acclimated to the running wheel for 2 days before the beginning of exercise protocol to avoid stress responses.

Magnetic Resonance Imaging (MRI)
A 3.0-T MR imager (Siemens Trio) was used to evaluate the location of the polyurethane polymer sheet used to generate the compression injury. Rats undergoing MRI evaluation were anesthetized with an intraperitoneal injection of 10% chloral hydrate (300 mg/kg) and placed in a prone position with a surface coil over the cervical spine regions to facilitate the acquisition of anatomical T2-weighted images (T2WIs). T2WIs were acquired with the following MRI parameters [11]: echo time/repetition time (TR) = 35/2500 ms (T2W) and 115/2500 ms (proton density-weighted [PDW]), slice thickness = 1 mm, interslice distance = 1.1 mm, and number of excitations (NEX) = 4. Fifteen axial scans were acquired from each rat that covered C3-C7 of the cervical spinal cord. Fifteen sagittal images of the cervical spinal cord (C3 to C7) were also acquired in an identical fashion.

Behavioral testing
Rats were acclimated to the individual testing set-up for two days for at least 1 h per day before behavioral testing and 30-60 min in each set-up on the days in which the tests were performed. All behavioral testing was performed on awake, unrestrained rats by the same investigator who was blinded to group identity.

Mechanical sensitivity determined by von Frey Filament testing
To evaluate changes in mechanical sensitivity after SCI with or without voluntary exercise, von Frey hair laments with force calibrated to 2, 6, and 15g (Bioseb, USA) were applied to the hind paws of each rat. The lament sizes were chosen based on ndings reported in previously published research on SCI in rats [12] [13]. Each lament was applied ve times to both the contralateral and ipsilateral hind paws. The frequency of positive responses to stimulation in the form of a percentage (i.e., the number of paw withdrawals divided by the number of applied stimuli) was used to document mechanical sensitivity. Testing with these laments was performed on two consecutive days before surgery to obtain baseline levels of mechanical sensitivity of each rat. After surgery, mechanical allodynia, and hyposensitivity were examined at 1-week post-injury (7 dpi) as well as at 2, 3, and 4 weeks (14 dpi, 21 dpi, and 28 dpi, respectively) after the opportunity for voluntary exercise was (or was not) provided ( Fig. 1).
Mechanical sensitivity using Place Escape/Avoidance Paradigm Paw withdrawal in response to stimulation with a von Frey hair lament is a measure of stimulus-evoked behavior. Therefore, paw withdrawal in response to the different laments might also be the result of evoked hyperre exia. Given this possibility, supraspinal processing of nociceptive stimuli was evaluated by testing based on a place escape/avoidance paradigm (PEAP). Brie y, rats were placed in a chamber that included a dark and a light side. A removable partition that was laminated with black (facing the dark side) and white (facing the light side) adhesive foil was positioned in the middle of the chamber. The partition separated the dark from the light side and created a passage between them. The rats were free to move from the dark to the light side and vice versa. PEAP testing is based on the assumption that the rats have the active choice between the naturally preferred dark side versus pain relief in the more aversive light side. Each rat underwent 25 min of PEAP testing. To begin the test, rats were placed in the middle of the chamber and the time they spent on the dark side was measured. The rst 10 min phase that involved no external stimuli was de ned as the exploration phase (i.e., baseline). This was followed by a 15 min testing phase (3 x 5 min blocks) during which the hind paws were stimulated with a 2g von Frey lament every 15 s when a rat was present on the dark side of the chamber. But no nociceptive stimuli were administered to the rats while they remained in the light side of the chamber. The hind paws that were contralateral and ipsilateral to the spinal cord lesion were stimulated alternatively during this period. The PEAP was used to evaluate changes in mechanical sensitivity changes after injury at 4 wpi.

Thermal hypersensitivity using the Hargreaves test
The Hargreaves test (Hargreaves Apparatus; Ugo Basile) was used to examine thermal sensitivity after SCI with or without voluntary exercise. In brief, we used an infrared heat beam to stimulate the plantar surface of the hind paws of each rat ve times on each side. The withdrawal latency to this heat stimulation was measured. Before testing, the baseline latencies to the heat stimulus were adjusted to approximately 13-14 s. A delay of at least 2 min between the two trials that test the same paw on the same rat, as well as a cut-off time of 15 s was set to avoid tissue damage. Similar to the von Frey testing, the baseline withdrawal latency of each rat was examined on two consecutive testing days before surgery. After surgery, thermal sensitivity was assessed at one week (7 dpi) post-injury, as well as at 2, 3, and 4 weeks (14 dpi, 21 dpi, and 28 dpi, respectively) after the opportunity for voluntary exercise was or was not provided.

Motor recovery using BBB scoring
The motor function of rats after SCI with or without voluntary exercise was assessed using Basso, Beattie, and Bresnahan (BBB) scoring. A single rat was placed randomly in an open eld for at least 5 min and observed by two experienced investigators who were blinded to group identity. A maximum score of 21 points indicates normal function or complete recovery, whereas a score of 0 re ects complete paralysis of the hind limb. Lower scores (0-7) were used to document isolated joint movements with little to no hindlimb movement. Intermediate scores (8-13) re ected intervals of uncoordinated gait, while higher scores (14)(15)(16)(17)(18)(19)(20)(21) were used to document appropriate coordination between the forelimb and the hindlimb.
6. Somatosensory evoked potentials (SEPs) The somatosensory conductive function of the spinal cord was evaluated using an electrophysiological monitoring system (Nicolet Endeavor CR) at 4 weeks post-injury (wpi). The animals were evaluated under general anesthesia that was induced by intraperitoneal injection with 10% chloral hydrate at a dose of 300 mg/kg. The scalp and posterior neck skin of each rat was shaved and sterilized with topical iodine. The two SEP recording electrodes were inserted subcutaneously at the midline of the skull, including one at 2 mm anterior to the bregma (Fz) and another at the midpoint of the ears (Cz). Two SEP stimulation electrodes were placed on each side of the sciatic nerve. Constant current with a magnitude of 6 mA, duration of 0.02 ms, and frequency of 3.43 Hz was used to elicit SEPs. Cortical SEPs were recorded from the skull at Cz-Fz. SEP signals were ltered using a bandpass lter of 10 Hz to 250 Hz. We averaged the results from 256 SEP trials to improve the signal-to-noise ratio. A sensitivity of 20 lV/div and a time base of 5 ms/div were used to display the SEP responses.
Onset latency and peak-to-peak amplitude were measured at the endpoint of the study. The onset latency was measured as the time interval between the delivery of the stimulus to the rst N wave from the baseline. Peak-to-peak amplitude is de ned as the vertical distance between the P and N points. Each SEP test was repeated 3 times, and average values were recorded. We de ne an immeasurable SEP (i.e., below detectable limits) as a waveform that could not be identi ed by averaging over 500 sweeps. SEP responses recorded from each limb were classi ed individually. For the nal analysis, the lowest SEP amplitudes detected for each of the four limbs were identi ed and used as the de nitive data point.

Quantitative polymerase chain reaction (qPCR)
Brie y, rats were deeply anesthetized with pentobarbital and perfused transcardially with phosphatebuffered saline (PBS). The L4-L6 segment of the spinal cord was removed immediately thereafter. Total RNAs were isolated with RNA Isolator Total RNA Extraction Reagent (Vazyme, R401-01) according to the manufacturer's protocol and PrimeScipt RT Master Mix (Takara, RR036A) was used to synthesize cDNA. QPCR reactions (10 µL each) were prepared from 5 µl Premix Ex Taq II (Takara, RR820A), 0.5 µl primer ( nal concentration 10 nM), 2 µl DEPC water, and 2 µl cDNA. Reactions were run in a LightCycler 480 qPCR instrument (Roche) using the standard conditions, including 95°C for 5 min, followed by 40 cycles (95°C for 15 s, 56°C for 30 s, and 72°C for 30 s) and a melting curve. We documented relative levels of gene expression using the 2 −ΔΔCt method with data normalized to GAPDH [14]. Sequences for all RT-qPCR primers are shown in Table S1.

Eriochrome cyanine R (EC) staining of myelin
Eriochrome cyanine R (EC) staining was used to identify myelinated regions in the cervical spinal cord and to quantify spared tissue at the epicenter of each lesion. Spinal cord sections (25 µm thickness) were subjected to EC staining as previously described [15]. The digital images of EC-stained sections were obtained at 20× magni cation using an Olympus BX53 microscope equipped with an XC camera. The percentage of the spared matter was determined by dividing the area representing the white matter area by the total cross-sectional area. Imaging and results were analyzed by the same investigator who was blinded to the group identity of each sample using ImageJ (U. S. National Institutes of Health, Bethesda, MD, USA).

Hematoxylin and eosin staining and Immunohistochemistry
At the endpoint of the study, rats were euthanized with an overdose of intravenous sodium pentobarbital and perfused transcardially with 0.9% saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PFA). Spinal cord segments C5-C7 and L4-L6 were collected, xed overnight with 4% formaldehyde in PBS at 4°C, and embedded in para n. A series of 25-µm thick spinal cord sections were prepared for Hematoxylin and eosin (H&E) and immunohistochemical (IHC) staining. Microwavemediated epitope retrieval was performed before staining. Sections were incubated overnight at 4•C with rabbit anti-CGRP (1:150; Abcam, Cambridge, UK) and rabbit anti-Iba-1 (1:200) antibodies. The sections were then incubated with reagents provided in a ready-to-use DAKO ChemMate EnVisionTM kit (K500711; Dako; Agilent Technologies, Inc., Santa Clara, CA, USA) for 30 min at room temperature. Images of each section of the perilesional spinal cord were acquired at 100× or 200× magni cation using an XC30 camera mounted on an Olympus microscope (Olympus Corporation, Tokyo, Japan).
The labeling densities of both CGRP and Iba-1 were quanti ed in the lumbar dorsal horn using ImageJ software (U. S. National Institutes of Health). Images were converted to grayscale and subjected to threshold evaluation. Labeling densities of CGRP and Iba-1 were quanti ed as the fraction of the area that exceeded the threshold level within the region of interest.

Statistical analysis
Two-way analysis of variance (ANOVA) with Fisher's least signi cant difference (LSD) post hoc test was used to analyze behavioral test data and to reveal overall group differences and signi cant changes over time. One-way ANOVA with Fisher's LSD post hoc test was used to analyze lesion size and expression levels of CGRP and Iba-1. All data are presented as mean ± standard error of the mean (SEM). Prism 7 software (GraphPad Software Inc., La Jolla, CA, USA) was used to perform statistical analysis. The threshold for signi cance was set at p < 0.05.

Assessment of SCI-associated damage by MRI and H&E staining
MRI and H&E staining were performed to determine whether the compression protocol used in this study resulted in damage to the posterior and lateral spinal cord. MRI analysis identi ed the compression materials at sites that were either directly posterior or lateral to the spinal cord where they elicited P-SCI or L-SCI, respectively ( Fig. 2A). H&E-stained sections of spinal cord tissue from rats in the sham group revealed normal histology. By contrast, H&E-stained spinal cord tissue from rats subjected to compression protocols exhibited vesicular degeneration and cavitation within the compressed regions (Fig. 2B).

Voluntary exercise has no impact on the recovery of motor function after SCI
Motor recovery was evaluated weekly beginning at 1 wpi using BBB scoring. Rats with SCIs exhibited signi cant de cits in hindlimb motor function compared with rats that underwent sham procedures at all time-points evaluated in this study (Fig. 2C, two-way ANOVA, p < 0.0001). Voluntary exercise had no impact on motor recovery (Fig. 2C). These results are consistent with the patterns of tissue sparing observed in the spinal cord lesions of rats that did or did not participate in voluntary exercise (see below). Although the sizes of the spared epicenters of these lesions were signi cantly different in rats with P-SCI versus L-SCI, no differences in the recovery of locomotor function were observed.

Voluntary exercise suppresses mechanical hypersensitivity that develops in response to SCI
To determine whether rats develop NP after compressive SCI and whether voluntary exercise results in reduced pain behaviors, rats were tested for mechanical and thermal sensitivity both before the injury and weekly thereafter.
Our ndings revealed that rats with compressive SCIs developed both mechanical hypersensitivity and hyposensitivity in the contralateral hind paw at 1 wpi ( Fig. 3A and 3B). Rats with P-SCI developed mechanical hypersensitivity and hyposensitivity in the ipsilateral hind paw at 1 wpi, while those with L-SCI developed these responses at 2 wpi.
Stimulation of the contralateral hind paw using 2g and 4g von Frey laments revealed that rats with P-SCI or L-SCI exhibited a signi cantly higher overall response rate compared to their sham-treated counterparts (two-way ANOVA, p < 0.0001). No differences were detected between the responses of rats with P-SCI versus L-SCI. However, the response rate of rats with P-SCI or L-SCI that participated in voluntary exercise was lower than that observed in rats with SCI that did not exercise from 2 through 4 wpi ( Fig. 3A and 3B, oooo p < 0.0001, vvv p < 0.001, vvvv p < 0.0001). ipsilateral hind paw revealed that rats with L-SCI required an additional week to reach the same response rate to 2g and 4g von Frey laments when compared to the response rate of rats with P-SCI rats ( Fig. 3D and 3E). Rats that participated in voluntary exercise after P-SCI exhibited lower levels of mechanical hypersensitivity of the ipsilateral hind paw. Voluntary exercise also suppressed the mechanical hypersensitivity that developed in rats with L-SCI.
Stimulation of the contralateral hind paw of rats with P-SCI or L-SCI using a 15g von Frey lament revealed an overall signi cantly lower response rate than that observed in sham-treated animals (two-way ANOVA, **** p < 0.0001, #### p < 0.0001). Voluntary exercise had no impact on the response rate to this larger diameter lament in either of these groups (Fig. 3C). Testing of the ipsilateral hind paw revealed that rats with L-SCI had a lower response rate to 15g von Frey lament compared with sham-treated animals at 2 wpi, while rats with P-SCI rats exhibited a reduced response rate after 1 wpi (Fig. 3D). 4. Mechanical hypersensitivity that develops in response to SCI is associated with aversive supraspinal responses PEAP testing was performed at the nal time point (4 wpi) to evaluate the supraspinal processing of nociception. Brie y, rats that are allowed to move freely between dark and light chambers will naturally spend more time in the dark. However, when the dark chamber is paired with painful stimuli (von Frey 2g lament), the rats may ultimately choose to spend more time on the light side. Analysis of the time spent on the dark versus light side revealed that rats with P-SCI and L-SCI spent less time in the dark chamber at 4 wpi (Fig. 3G). However, after three weeks of voluntary exercise, all rats with SCIs spent less time on the dark side compared with those that did not participate in voluntary exercise (Fig. 3G).

Voluntary exercise suppresses thermal hypersensitivity that develops in response to SCI
The Plantar Test (Hargreaves method) was used to examine the thermal sensitivity of the rat hind paws.
Signi cant differences were detected between various groups (two-way ANOVA: **** p < 0.0001, #### p < 0.0001). Rats with P-SCI and L-SCI developed signi cant thermal hypersensitivity at 1 wpi when compared with their baseline values and to the responses of sham-treated rats. Thermal sensitivity persisted through the end of the experiment (4 wpi). All rats with SCI exhibited reductions in the withdrawal latency at the contralateral hind paw of ~8-9 s from baseline values at 1 wpi. By contrast, the withdrawal latency decreased only by ~3-4 s in the ipsilateral hind paw of rats with L-SCI. Interestingly, voluntary exercise had a positive impact on the thermal hypersensitivity that developed in response to SCI (Fig. 4). The withdrawal latency increased by ~3-4 s in the contralateral hind paws of rats with SCI that participated in voluntary exercise compared to the sham-treated group and rats with SCI that did not participate in voluntary exercise. However, none of the response rates of rats with SCI ultimately returned to baseline levels (Fig. 4A). Voluntary exercise increased the withdrawal latency in the ipsilateral hind paw by 3-4 s in rats with P-SCI rats and suppressed the more severe thermal hypersensitivity exhibited by L-SCI rats (Fig. 4B).
6. Voluntary exercise resulted in improved somatosensory evoked potentials in rats with SCI Somatosensory evoked potentials (SEPs) were examined in order to evaluate the functionality of ascending sensory and proprioception pathways after SCI (Fig. 5). As shown in Fig. 5A, categories of SEP responses that have been modeled in rats include N and P points, which refer to the upward (N; negative) and downward (P; positive) peaks of the far-eld cortical potential. Onset latency was measured as the time delay before N and the amplitude was measured as the voltage differences between N and P. Rats with L-SCI exhibited signi cant decreases in amplitude and prolonged latency at the side that was ipsilateral to the compression compared to the sham-treated controls (two-way ANOVA, ** p < 0.01, **** p < 0.0001). Greater deterioration of SEPs was observed on the ipsilateral versus the contralateral side in response to the compression lesion in rats with L-SCI. By contrast, similar deterioration was observed on both sides of the lesion in rats with P-SCI.
Rats with L-SCI showed no signi cant differences in onset latency on the side that was contralateral to the lesion when compared to the responses of sham rats. Participation in voluntary exercise resulted in signi cant improvements in SEP amplitudes on both sides of the lesion in rats with L-SCI (two-way ANOVA, # p < 0.05, #### p < 0.0001). After voluntary exercise, the differences between the ipsilateral and contralateral SEP amplitudes were not signi cant, although the SEP latency detected on the side that was ipsilateral to the lesion remained signi cantly longer than responses detected on the contralateral side.
Rats with P-SCI exhibited similar amplitude decreases and prolongation of onset latency on both sides of the compression lesion. Participation in voluntary exercise resulted in improvements in the amplitude but had minimal impact on SEP latency of rats with P-SCI.

Voluntary exercise does not result in modi cation of the structural changes that result from a compressive SCI
To exclude the possibility that differences in lesion size might in uence the changes in pain sensitivity secondary to SCI and examine the impact of voluntary exercise on processes that promote neuroprotection, white matter sparing was evaluated in Eriochrome cyanine R (EC)-stained coronal sections of the rat spinal cord (Fig. 6A) The extent of white matter sparing was quanti ed at the lesion epicenter (Fig. 6B). We found no signi cant differences in white matter sparing at the lesion epicenters when comparing spinal cord tissue from rats with SCI (either P-SCI or L-SCI) that have or have not participated in voluntary exercise (Fig. 6B). Interestingly, spinal cord tissues from injured rats exhibit a comparatively large percentage of white matter and complete loss of gray matter at the lesion epicenters compared to sham-treated controls (one-way ANOVA, *** p < 0.001). Tissues from rats with L-SCI exhibit a smaller fraction of white matter loss at the lesion epicenter compared with tissues from rats with P-SCI (one-way ANOVA, *** p < 0.001).

SCI-induced increases in calcitonin gene-related peptide (CGRP) expression are suppressed by voluntary exercise
Calcitonin gene-related peptide (CGRP) is a neuropeptide that may be expressed in response to peptidergic C and Aδ innervation and that has been detected at high levels in the skin, blood, cerebrospinal uid, and spinal cord tissue in association with a variety of chronic pain conditions. Therefore, we quanti ed CGRP expression bilaterally in lumbar laminae I and II by using qPCR and IHC (Fig.7).
We detected a signi cant increase in CGRP expression was signi cantly increased in laminae I and II of the lumbar spinal cord in rats with P-SCI or L-SCI. Both groups of rats with SCI exhibited signi cant 1.7fold increases in CGRP expression compared to sham-treated controls (one-way ANOVA, ***p < 0.001). No signi cant differences were detected between rats with P-SCI vs. L-SCI. Interestingly, voluntary exercise resulted in signi cant reductions in CGRP expression in lumbar laminae I and II in rats with SCI. By contrast, no differences in CGRP expression were detected in sham vs. sham+V controls.

SCI-induced increases in microglial activation are suppressed by voluntary exercise
CGRP is a mediator of microglial activation and may promote nociceptive signaling in the dorsal horn [16]. Based on our detection of CGRP overexpression in the lumbar spinal cords of rats with P-SCI and L-SCI, we proceeded to evaluate microglial activation in the lumbar dorsal horn in experiments designed to detect and quantify expression of ionized calcium-binding adapter molecule 1 (Iba-1).
We found signi cantly increased expression of Iba-1 bilaterally in the lumbar spinal cord of rats with P-SCI compared to sham-treated controls (Fig. 8). Participation in voluntary exercise suppressed Iba-1 overexpression in these rats (one-way ANOVA, *** p < 0.001). By contrast, Iba-1 expression was signi cantly increased in the ipsilateral lumbar dorsal horn, but not in the contralateral lumbar dorsal horn of rats with L-SCI. Voluntary exercise also suppressed Iba-1 overexpression in the ipsilateral lumbar dorsal horn in rats with L-SCI. Similar to our ndings for CGRP expression, we detected no statistically signi cant differences in Iba-1 expression in the lumbar spinal cord when comparing results from the sham vs. the sham+V groups..

Discussion
The ndings presented in our study demonstrate clearly that voluntary exercise initiated during the early phase (1 wpi) of recovery from a compressive SCI results in reductions in mechanical allodynia and thermal hyperalgesia. We also found that compressive SCI-induced NP is associated with increased expression of both CGRP and Iba-1 in the lumbar dorsal horn, which are reduced in rats that participated in voluntary exercise. Thus, we conclude that voluntary exercise selectively in uences disturbed mechanical and thermal sensation that develops after posterior and lateral compressive SCIs and reduces the extent of injury-induced maladaptive changes in the lumbar dorsal horn.
An understanding of how neuropathic pain arises after SCI and what might be done to it will require ongoing mechanistic studies performed in animal models [17]. hyperalgesia that develop in response to peripheral and central nerve injury [19][20] [21]. Furthermore, we found that rats subjected to a lateral cervical compression injury rats developed mechanical allodynia, hypoalgesia, and thermal hyperalgesia in the ipsilateral hind paw after a delay of two weeks. By contrast, the contralateral hind paws of rats with both P-SCI and L-SCI developed to NP within 1 wpi. Hatashita et al. [22] reported that contralateral mechanical allodynia could be detected as soon as two days after a hemilateral spinal nerve injury. Similarly, Kang et al. [23] demonstrated that spinal cord hemisection resulted in prominent mechanical allodynia and thermal hyperalgesia at both contralateral and ipsilateral hind paws after only one day post-injury and that the administration of diluted bee venom could suppress pain-related behavior at the ipsilateral hind paw. The reasons why rats with P-SCI vs. L-SCI rats presented with contralateral and ipsilateral NP at different timepoints required further exploration.
Interestingly, participation in voluntary exercise ameliorated the mechanical allodynia and thermal hypersensitivity but not loss of sensory function in both contralateral and ipsilateral hind paws of rats with compressive cervical SCIs. These results suggest that loss of sensory function (i.e., mechanical hyposensitivity) that develops in response to a compressive cannot be recovered via these actions. It is not clear why voluntary exercise has a selective in uence on mechanical allodynia and thermal hyperalgesia but not mechanical hypoalgesia. This observation might be explained by the fact that sensory neurons are unable to renew themselves. In a previous study that featured a clip compression SCI rat model, early initiation of intensive treadmill training attenuated the development of thermal hypersensitivity after 2 weeks and reduced mechanical allodynia after 7 weeks [24]. Clip compression models typically result in more severe neurological dysfunction than the model featured in our current study. Thus, it may take longer to detect exercise-associated reversals of NP. Similarly, ndings reported by Li et al. [25] revealed that treadmill training could signi cantly mitigate both mechanical allodynia and thermal hyperalgesia at 1 wpi in a rat spinal cord contusion model. Furthermore, our results are consistent with those of Dugan et al. [24] who found that intensive treadmill training had no impact on overall functional locomotor recovery. A more precise study of locomotor responses, including an analysis of gait with Catwalk, may reveal differences between the groups that could not be detected using BBB scoring. However, this is beyond the scope of our current research.
In addition to reductions in both mechanical and thermal hypersensitivity, we also found that participation in voluntary exercise signi cantly improved SEP at 4 wpi. These results suggested that voluntary exercise results can ameliorate SCI-induced disturbances in proprioception. This is the rst study to evaluate the effects of exercise on disturbances in proprioception that developed in response to SCI in rats. Of note, Qaiser et al.
[26] demonstrated that it was possible to alter proprioception in patients diagnosed with incomplete SCI using a passive training protocol that was combined with feedback. More studies will be needed to con rm and extend these observations.
To the best of our knowledge, this is the rst research study that focused on the impact of voluntary exercise on NP that develops in response to posterior and lateral cervical SCIs in rats. Our ndings can serve as a basis for future studies designed to explore the molecular and cellular mechanisms underlying the development and persistence of NP in patients with OPLL, CSM, or spinal tumors and the role of voluntary exercise in promoting effective rehabilitation.
Because the lumbar dorsal horn plays an important role in the development and persistence of NP that develops in response to SCI, we also evaluated maladaptive changes of the dorsal horn at this site, which is below the level of the injury. There is increasing evidence that CGRP plays a role in the development of NP [27]. Results from a mouse model of spinal cord contusion injury suggested that early treadmill training can suppress NP by reducing CGRP expression, detected in this study by its labeling density in laminae III-IV of lumbar dorsal horn [28]. In this study, we also detected increased CGRP expression in laminae I-II, but not in laminae III-IV after a cervical spinal cord compression injury. Furthermore, we found that participation in voluntary exercise suppressed injury-induced CGRP overexpression in both ipsilateral and contralateral lumbar dorsal horns. Future studies will be needed to explore whether SCI induces aberrant plasticity of the neural circuitry involving peptidergic nociceptive bers in laminae I-II that may contribute to the development of NP [29].
CGRP has been implicated in the development of neurogenic in ammation and is upregulated in association with in ammation and neuropathic pain [30]. Activation of spinal microglia also contributes to the development of NP following peripheral and central nerve injury [31] [32]. Therefore, we explored the impact of spinal cord compression and voluntary exercise on microglial activation in the lumbar dorsal horn. The primary phase of microglial activation reportedly peaks at 7 dpi. Microglia undergo reactivation after 14 dpi followed by a peak at 60 dpi that persists through 180 dpi [33]. Similar to the results presented in a previous study, Iba-1 expression was increased in the lumbar spinal cord at 4 wpi; voluntary exercise resulted in a signi cant decrease in Iba-1 overexpression in the lumbar dorsal horn. Qi et al. [34]demonstrated that an increase in Iba-1 activation in the dorsal horn may account for the occurrence of stress-induced hyperalgesia in a model of chronic in ammatory pain. Many studies have demonstrated that Aβ-bers can produce and release CGRP in in ammatory states and that this response might contribute to NP that can develop in response to nerve injury and in in ammatory pain models [35] [36]. CGRP is localized within the primary afferent terminals and neurons in laminae I and II that mainly receive signals from peptidergic nociceptive bers. Therefore, CGRP overexpression might be the result of a phenotypic switch of Aβ-bers to a nociceptor phenotype. The relationship between CGRP and Iba-1 expression in response to SCI requires con rmation and further study.
We were greatly intrigued by the results of the immunohistochemical studies performed with anti-Iba-1 that revealed microglial activation in both ipsilateral and contralateral lumbar dorsal horns of rats with P-SCI, but only in the contralateral dorsal horn of rats with L-SCI. These ndings may explain why the pain on the ipsilateral hind paw emerged one week later compared with the contralateral hind paw. Additional studies will be needed to con rm this nding. The mechanisms underlying differential Iba-1 expression remain unclear and may be related to changes in the microenvironment that develop at levels below initial spinal cord injury. These responses and associated mechanisms are studied only rarely.
Experiments performed in transgenic reporter mice will be needed to pursue this issue in our future work. Moreover, Iba-1 overexpression and thus microglial activation in the lumbar dorsal horn that develops after SCI was suppressed in response to voluntary exercise. These results are consistent with those reported in many previous studies that examined the impact of exercise and its role in suppressing microglia activation in a variety of neurological diseases. For example, Xiong et al. [37] reported that long-

Availability of data and materials
The data that support the ndings of this study are available from corresponding author on reasonable request.

Competing interests
The authors declare they have no competing interests.  scoring. Posterior and lateral spinal cord injuries (P-SCI and L-SCI, respectively) in the rats in the voluntary exercise and no voluntary exercise groups both showed signi cant motor de cits compared to the two sham groups (two-way ANOVA, ****p < 0.0001) although there were no signi cant differences between the two groups.

Figure 3
Effects of a compressive spinal cord injury (SCI) and voluntary exercise on mechanical sensitivity as assessed by stimulation with von Frey laments linked to a place escape/avoidance paradigm. (A-B) L-SCI-induced mechanical hypersensitivity detected in the ipsilateral hind paws in response to light laments (2g and 6g) developed one week later compared with the analogous responses observed in rats with P-SCI. SCI-induced mechanical hypersensitivity was suppressed in rats that participated in voluntary exercise (two-way ANOVA ####p < 0.001 for L-SCI vs. sham; ****p < 0.0001 for P-SCI vs. sham; oooop < 0.0001 for L-SCI+V vs. L-SCI; vvvvp < 0.0001 for P-SCI+V vs. P-SCI). (C) Stronger laments (15g) revealed signi cant hyporesponsiveness of the ipsilateral hind paws in experiments with rats with SCI compared to those that were sham-treated regardless of voluntary exercise (two-way ANOVA, ****p < 0.0001 and ####p < 0.0001 for P-SCI and L-SCI vs. sham, respectively). (D-E) P-SCI-and L-SCI-induced mechanical hypersensitivity detected in the contralateral hind paws in response to light laments (2g and 6g) appears at 1 wpi. SCI-induced mechanical hypersensitivity was suppressed in rats that participated in voluntary exercise (two-way ANOVA, ####p < 0.0001 for L-SCI vs. sham; ****p < 0.0001 for P-SCI vs. sham; oooop < 0.0001 for L-SCI+V vs. L-SCI; vvvp < 0.001 and vvvvp < 0.0001 for P-SCI+V vs. P-SCI). (F) Voluntary exercise had no impact on SCI-induced mechanical hyposensitivity detected in the contralateral hind paws. (G) In the PEAP study, rats with P-SCI and L-SCI that did not participate in voluntary exercise spent signi cantly less time in the dark chamber with light mechanical stimuli compared with rats that were sham-treated and those with SCI that participated in voluntary exercise (P-SCI+V and L-SCI+V; two-way ANOVA, **p < 0.01 and ****p < 0.0001 for SCI vs. sham; #p < 0.05, ##p < 0.01, ###p < 0.001, and ####p < 0.0001 for all SCI+V vs. SCI; sham, n = 12; P-SCI, n = 15; L-SCI, n = 15; sham+V, n = 12; P-SCI+V, n = 15; L-SCI+V, n = 15).