Zonisamide ameliorates progression of cervical spondylotic myelopathy in a rat model

Cervical spondylotic myelopathy (CSM) is caused by chronic compression of the spinal cord and is the most common cause of myelopathy in adults. No drug is currently available to mitigate CSM. Herein, we made a rat model of CSM by epidurally implanting an expanding water-absorbent polymer underneath the laminae compress the spinal cord. The CSM rats exhibited progressive motor impairments recapitulating human CSM. CSM rats had loss of spinal motor neurons, and increased lipid peroxidation in the spinal cord. Zonisamide (ZNS) is clinically used for epilepsy and Parkinson's disease. We previously reported that ZNS protected primary spinal motor neurons against oxidative stress. We thus examined the effects of ZNS on our rat CSM model. CSM rats with daily intragastric administration of 0.5% methylcellulose (n = 11) and ZNS (30 mg/kg/day) in 0.5% methylcellulose (n = 11). Oral administration of ZNS ameliorated the progression of motor impairments, spared the number of spinal motor neurons, and preserved myelination of the pyramidal tracts. In addition, ZNS increased gene expressions of cystine/glutamate exchange transporter (xCT) and metallothionein 2A in the spinal cord in CSM rats, and also in the primary astrocytes. ZNS increased the glutathione (GSH) level in the spinal motor neurons of CSM rats. ZNS potentially ameliorates loss of the spinal motor neurons and demyelination of the pyramidal tracts in patients with CSM.

The histological features of CSM include gliosis, loss of anterior horn (AH) cells, and cystic necrosis resulting from the chronic and mechanical compression [15][16][17] . However, the exact cellular mechanisms for insidious and progressive deterioration remain not to be fully elucidated 9 . Recent studies show that chronic compression of the spinal cord sequentially causes disruption of the blood-spinal cord barrier, ischemia, and neuroinflammation 5,16 . CSM is characterized by loss of spinal motor neurons, degeneration of axons constituting the corticospinal tracts, demyelination of oligodendrocytes, astrocytic scarring, and cavitation 4,5,18 . Neuronal and oligodendrocytic apoptosis is observed in human CSM and in the Tiptoe Walking Yoshimura (twy/twy) hyperostotic mice [15][16][17] . Multicellular apoptosis in CSM is accounted for by glutamatergic toxicity, free radicals (oxygen radicals and lipid peroxidation), and cationic cell injury 16 .
Zonisamide (ZNS) is an anti-epileptic agent widely used as an adjunctive therapy for partial seizures in adults and an anti-Parkinsonian agent particularly for resting tremor 19 . We previously reported that ZNS enhanced neurite elongation of primary spinal motor neurons in mice, and facilitated nerve regeneration of an autograft of the sciatic nerve in mice 20,21 . We also reported that ZNS protected primary spinal motor neurons against oxidative stress 20,21 . In a mouse model of Parkinson's disease, ZNS counteracts neurodegeneration partly by increasing expressions of cystine/glutamate exchange transporter (xCT) in astrocytes and glutathione (GSH) in the dopaminergic neurons in the striatum 22 . xCT in astrocytes takes up cystine, which is converted into two cysteines 23,24 . Neurons then take up astrocyte-derived cysteine, and synthesize GSH. Thus, the synthesis of GSH in neurons is dependent on the expression of xCT in astrocytes 22 . In addition to xCT, ZNS also induces the expression of metallothionein 2A, which captures radical residues as an anti-oxidant, in response to oxidative stress, in astrocytes 22,25 . ZNS is thus likely to exert its neuroprotective effect as an anti-oxidant.
In this study, we made a rat model of CSM at cervical (C) levels 5 and 6. The CSM rats showed progressive motor impairments, loss of spinal motor neurons, and increased lipid peroxidation represented by 4-hydroxynonenal (4-HNE). The effects of ZNS on nerve regeneration 20,21 and on preventing degeneration of dopaminergic neurons 22,25 prompted us to examine the effects of ZNS on a rat CSM model. We found that oral administration of ZNS delayed progression of motor impairments, rescued loss of spinal motor neurons, and spared myelination of the pyramidal tracts. ZNS also upregulated xCT and metallothionein 2A in the spinal cord, which is associated with increased GSH production in the motor neurons.

Materials and methods
Animal maintenance and group design. All rat studies were approved by the Animal Care and Use Committee of Nagoya University, and were performed in accordance with relevant guidelines. All rats were housed in environmentally adequate cages with less than 3 animals. The temperature was maintained at 22 °C and the room was kept at a 12/12-h light/dark cycle.
Adult female Wistar rats aged 12 weeks were purchased from Japan SLC, Inc. Two independent experiments were performed. The first experiment included two groups of sham-operated (n = 13) and CSM (n = 7) rats. The second experiment included CSM rats with daily intragastric administration of 0.5% methylcellulose (n = 11) and ZNS (30 mg/kg/day) in 0.5% methylcellulose (n = 11).
Surgical implantation of the expandable material. We used an expandable material made of a waterabsorbing polyurethane elastomer (Aquaprene Dx, Sanyo Chemical Industries) 26 . The polymer sheet reached a volume of 150% over 24 h when embedded under the skin of a rat (Supplemental Fig. S1A). The polymer was dissected in a rectangular shape with 0.80 mm (ventrodorsal) × 3 mm (mediolateral) × 5 mm (rostrocaudal) using a Mcilwain Tissue chopper (Campden Instruments LTD, England). The sizes of the implant were essentially taken from previous reports 27,28 except that we increased the thickness from 0.70 mm to 0.80 mm. To avoid injury to the spinal cord, the edge of the polymer was slanted by removing 1 mm from the corner using a surgical A representative T2-weighted mid-sagittal MRI image of a CSM rat at 10 weeks after surgery shows dorsal-ventral cervical spinal cord compression at C5 and C6 levels while keeping laminae of C4, 5, 6, and 7 intact. The area in the red dotted square is schematically shown in the inset. (B) Upper panels schematically show the spinal cord around the C5-6 level. Hematoxylin and eosin (HE) staining of axial sections at the red broken lines at the C5-6 disc level are shown in the lower panels. Blue boxed areas are enlarged in (E). Bar = 600 μm. (C,D) Stride length (C) and duty cycle (D) of fore-and hind-limbs analyzed by the CatWalk system of sham-operated and CSM rats from 1 to 10 weeks postoperatively. (C) A representative stride length of the right forelimb is indicated by a doubleheaded arrow in the right panel. (D) Duty cycle is a ratio in time when a paw stays on the glass plate in a single step cycle. CSM rats started to exhibit gait disturbance from 5 weeks after surgery. Mean and SE are indicated (n = 6 and 4 for sham-operated and CSM rats, respectively). *p < 0.05, **p < 0.01, and ***p < 0.001 by two-way repeated measures ANOVA followed by Bonferroni correction. (E,F) HE staining at 5 weeks postoperatively (E) and 4-hydroxynonenal (4-HNE) staining at 10 weeks postoperatively (F) of an axial section of the spinal cord at the C5-6 disc level in sham-operated and CSM rats. (E) Images are enlargement of the blue boxed areas in (B). A representative neuron (white arrow) in the anterior horn (AH) region is enlarged in the inset. Purple-stained Nissl bodies are indicated by yellow arrowheads. The numbers of cells with Nissl bodies in the AH region of sham-operated and CSM rats are shown. Mean and SD are indicated (n = 3 rats in each group). *p < 0.05 by Mann-Whitney U-test. (F) Representative images of the spinal cord stained with 4-hydroxynonenal (4-HNE). Arrows point to 4-HNE-positive large cells (diameter > 40 μm), which were recognized as motor neurons, in the AH region. A border between the gray and the white matters is indicated by a red dotted line. Bar = 50 μm. Mean and SD are indicated (n = 3 rats in each group). *p < 0.05 by Mann-Whitney U-test. www.nature.com/scientificreports/ knife (Supplemental Fig. S1B). We attached a surgical thread (3-0 Ethicon, Johnson & Johnson K. K.) and used it to guide the passage of the blade underneath the C5 and C6 laminae. Anesthesia was induced by inhaling 2.5% isoflurane gas (Fujifilm Wako Pure Chemical) through a facial mask. We implanted the blade essentially following a protocol by Kim et al. 28 . Briefly, a 4.0-cm midline skin incision was made from C2 to the thoracic (T) level 2. The C4-7 laminae were identified by counting spinous processes from T2, and were carefully exposed. The ligamenta flava at C4-5, C5-6, and C6-7 were removed to expose the underlying dura. The blade was epidurally implanted underneath the C5 and C6 laminae using microsurgical techniques (Supplemental Fig. S1C-E), and multilayer tissue closure was performed. The sham-operated group had mock surgery without implantation.
Wire mesh walking test. The numbers of forepaw and hindpaw slips were counted using a 100 cm × 10 cm walkway with a 3.0 × 3.0-cm grid made of 1.6-mm-thick wires. Videos were recorded for each animal while completing the walk three times 9,29 . A slip was counted when one of paws fell below the plane of the grid, and the average number of slips during a 100-cm walk was calculated. The percentage of paw slips was calculated by dividing the total number of slips by the total number of steps 30 .

Automated gait analysis (CatWalk).
Gait analysis was performed using the CatWalk system XT Ver. 9 (Noldus Information Technology, Wageningen, Netherlands) at 1 to 10 weeks after the surgical implantation, as described previously [31][32][33][34] . Briefly, the system was consisted of a horizontal glass plate with computer-operated video-recording equipment placed underneath the glass plate. The following criteria were used for walkway crossing: (1) the rat needed to cross the walkway without any interruption, and (2) a minimum of three complete crossings per animal were required. We evaluated the stride length and the duty cycle with the CatWalk system. The duty cycle indicates a ratio (%) of the time when each paw is in contact with the glass plate in a single step cycle. We also measured the print position, which is the distance between the placements of the ipsilateral fore-and hind-paws. Data of gait factors were obtained from the both side limbs and the values were calculated independently.
Histopathology. CSM rats were histologically analyzed at 5-10 weeks after surgical implantation. Rats were transcardially perfused with 4% paraformaldehyde (PFA) in cold phosphate-buffered saline (PBS) at pH 7.4. The spine at C2-T2 was removed and fixed in 4% PFA and 10% sucrose overnight at 4 °C. Laminae and vertebral bodies were removed, and a 5-mm-thick segment of the spinal cord at C5-6 was isolated. The C5-6 segment was embedded in paraffin and sectioned into 9 μm-thick slices. The specimens were stained with hematoxylin/eosin (HE) and Luxol Fast Blue/Nissl (LFB/Nissl). Neurons in the anterior horn region were identified by the presence of large nuclei and the Nissl bodies in the cytoplasm. Number of the neurons was counted in the both sides of anterior horn region and evaluated independently. For frozen sections, a 5-mm-thick segment at C5-6 was similarly prepared as stated above, and was secondarily fixed with 4% PFA and 20% sucrose in PBS overnight at 4 °C. Subsequently, the C5-6 segment was sectioned into 20 μm-thick slices.
immunohistochemistry. Lipid peroxidation was evaluated by immunostaining for 4-HNE of paraffin sections of the spinal cord. The sections were blocked with 10% goat serum (Fujifilm Wako Pure Chemical, WDR2410) in PBS, and were incubated with an anti-4-HNE monoclonal antibody (1:40, JaICA, MHN-100P) at 4 °C overnight. The sections were then incubated with a biotinylated anti-mouse IgG antibody (1:100, Vector Laboratories, BA-2000) followed by incubation with an avidin-biotin-horseradish peroxidase complex (1:500, Vector Laboratories, the Vectastain ABC kit), and were visualized with the ImmPACT DAB Substrate (Vector Laboratories). The number of 4-HNE-positive cells were counted in the both side of anterior horn region and evaluated independently. The frozen sections were used for staining GSH, choline acetyltransferase (ChAT), and 4,6-diamidino-2-phenylindole (DAPI). After blocking with 5% horse serum (Invitrogen) and 0.3% Triton X-100 in 0.1 M PBS, the sections were air-dried at room temperature for 10 min, placed in cold acetone for 10 min, and dried again. The sections were then incubated overnight at 4 °C with the following primary antibody: mouse anti-GSH (1:1,000, Abcam, ab19534) or goat anti-ChAT (1:100, Millipore, AB144P). The sections were washed in PBS three times and incubated with biotinylated anti-goat IgG (1:100, Vector laboratories, BA-9500) for 2 h, washed in PBS three times, and incubated with Alexa 546-conjugated streptavidin (1:100, Invitrogen, S11225) or FITC-conjugated anti-mouse IgG (1:100, Vector Laboratories, FI-2000) for 30 min. The sections were mounted on cover slips with Mowiol mounting medium containing DAPI to counterstain the nuclei. Immunostaining was quantified by two blinded observers using the BZ-9000 microscope (Keyence, Woodcliff Lake, NJ). In 4-HNEstained sections, a diameter of > 40 μm with clear cytoplasm was used as a hallmark of spinal motor neurons. The intensity of GSH signals co-localized with ChAT signals in the gray matter in CSM rats was quantified by www.nature.com/scientificreports/ two blinded observers using a confocal laser scanning microscope system (TiE-A1R, Nikon) and the Image-J software (NIH). The intensity of GSH signals co-localized with ChAT signals were counted in the both side of anterior horn region and evaluated independently.
Total RNA extraction and quantitative RT-PCR. Sham-operated rats and CSM rats with or without ZNS at 10 weeks after surgery were transcardially perfused with PBS. The spinal cord at C5-6 was isolated as stated in the section for histopathology. Total RNA of the spinal cord was isolated using QIAzol (QIAGEN). First strand cDNA was synthesized with ReverTra Ace (Toyobo  Figures 3C-E, 4D, 5A, and S2A-C were analyzed by Kruskal-Wallis test followed by Dunn-Bonferroni HSD test, as the sample sizes were less than 10 in each group. P values of 0.05 or less were considered to be statistically significant.

CSM rats exhibit gait dysfunctions from 5 weeks after surgery.
In patients with CSM, motor and sensory symptoms develop slowly and progress steadily for several years [35][36][37] . To recapitulate human CSM in an animal model, we epidurally implanted a blade-shaped water-absorbing polyurethane elastomer (Aquaprene Dx) 26 underneath the C5 and C6 laminae in a rat, while keeping the laminae uncut (Fig. 1A,B and Supplemental Fig. S1). The sham-operated rats had mock surgery without implantation. We evaluated motor dysfunctions and histopathology of the spinal cords in our CSM rats (Supplemental Fig. S1C-E). First, we performed gait analysis with the CatWalk system to evaluate the stride length and the duty cycle, which represents a ratio of standing duration of each paw in one cycle of walking. In 1-3 weeks after surgery, the stride lengths of the forelimbs and hindlimbs in CSM rats were not different from those in sham-operated rats. In 5-10 weeks after surgery, the stride lengths gradually increased in sham-operated rats, whereas they remained essentially unchanged in CSM rats (Fig. 1C). Similarly, in 5-10 weeks after surgery, the duty cycles gradually decreased in sham-operated rats, whereas they remained stable in CSM rats (Fig. 1D). We also measured the print position, which is the distance between the forepaw step and the hindpaw step. The print positions were substantially unchanged in sham-operated rats, whereas they gradually increased in 5-10 weeks after surgery in CSM rats (Supplementary Fig. 1E). To summarize, CSM rats required a higher number of steps to finish the walkway and kept their paws on the floor for a longer time in walking compared to sham-operated rats.
At 5 weeks after surgery, the number of Nissl body-positive cells in the AH region at C5-6 in CSM rats was lower than that in sham-operated rats at 5 weeks after surgery (Fig. 1B,E). Twy/twy mice develop calcified ligamentum flavum at C2-3, which chronically compresses the cervical cord. The compressed cervical cord shows apoptotic neurons and oligodendrocytes 15 . Oxidative stress is one of the inducers of apoptosis in CSM 16,34 . We therefore evaluated cellular oxidation by staining for 4-HNE, a marker of lipid peroxidation. The number of 4-HNE-positive neurons was higher at 10 weeks after the surgery in CSM rats compared to that in sham-operated rats (Fig. 1F). As has been reported in previous CSM animal models 28,38 , our CSM model also demonstrated the presence of oxidative stress at the compressed site.
ZNS ameliorates gait dysfunctions in CSM rat. Zonisamide (ZNS) has a protective effect on the dopaminergic and peripheral neurons against cell death 20,22,25 , and improves motor symptoms in patients with Parkinson's disease 39 . We previously reported in mice that ZNS enhanced neurite elongation of primary spinal motor neurons, and facilitated regeneration of an autograft of the sciatic nerve 20,21 . We thus examined the effects of ZNS on our CSM rats. We intragastrically administered 30 mg/kg/day of ZNS starting at 5 weeks after surgery, when CSM symptoms became overt ( Fig. 2A). Daily administration of ZNS for 5 weeks from 5 to 10 weeks after  www.nature.com/scientificreports/ surgery increased the stride lengths in both forelimbs and hindlimbs (Fig. 2B,C). ZNS also decreased the duty cycle of hindlimbs (Fig. 2E). Additionally, the wire mesh walking test showed that daily ZNS administration for 5 weeks suppressed the rate of limb slippage ( Fig. 2F and Supplementary Fig. S2). These results demonstrate that daily ZNS administration for 5 weeks improved motor dysfunctions caused by CSM operation.

ZNS has a neuroprotective effect at the compressed cervical spinal cord.
To measure the number of motor neurons, we performed HE staining (Fig. 3A), LFB/Nissl staining (Fig. 3B), and immunohistochemistry with an anti-ChAT antibody (Fig. 4A). The number of cells with Nissl bodies in the AH region in CSM rats without ZNS [ZNS(−)] was lower than that that in sham-operated rats at 10 weeks after surgery (Fig. 3C).
In contrast, ZNS administration [ZNS( +)] spared the number of cells with Nissl bodies compared to that in ZNS(−) rats. These results suggest that ZNS reduced the loss of motor neurons in the AH region of CSM rats. In CSM, myelopathy is caused by demyelination of the pyramidal tracts 40, as well as the loss of motor neurons. We thus examined the size of the pyramidal tracts in the CSM rats, which are located in the dorsal spinal cord in rodents 41 . We blindly measured the pyramidal tract area by HE staining, and the myelinated area by LFB staining, and found that CSM operation decreased both the pyramidal tract area (Fig. 3D) and the ratio of LFB-positive myelinated area in the pyramidal tract area (Fig. 3E). ZNS administration had no effect on the pyramidal tract area, which was likely because ZNS could not mitigate the compression (Fig. 3D). In contrast, ZNS administration spared the ratio of LFB-positive area in the pyramidal tract area (Fig. 3E). Next, we observed that ZNS administration increased the number of ChAT-positive motor neurons in the spinal cord (Fig. 4A,B). These results suggest that ZNS improved the survival of spinal motor neurons and spared the myelination of motor axons in the spinal cord in CSM rats.

ZNS increases expressions of anti-oxidants in the spinal cord of CSM rat.
We previously reported that ZNS rescued cell death of primary motor neurons exposed to oxidative stress by H 2 O 2 treatment 20,22,25 . Proliferating cell nuclear antigen (PCNA) is a hallmark of proliferation of reactive astrocytes 42 . A previous report shows that ZNS increases the PCNA expression in 6-hydroxydopamine-treated substantia nigra modeling for Parkinson's disease 25 . Here we showed that H 2 O 2 reduced PCNA expression in primary astrocyte derived from normal spinal cord, and ZNS rescued the reduction (Supplementary Fig. S3A). ZNS is thus protective against the oxidative stress in astrocytes, as we previously observed in spinal motor neurons 20,22,25 .
Astrocytes take up cystine via xCT to generate two cysteines, which are then taken up by motor neurons to generate GSH 23,24 . Metallothionein 2A is an anti-oxidant expressed in astrocytes, as well as in motor neurons to a lesser extent 25 . We observed that ZNS increased gene expressions of xCT and metallothionein 2A in primary astrocytes and minimally in the NSC34 motor neuron cell line (Supplemental Figs. S3B-C). Similarly, in the spinal cord of CSM rats at 10 weeks after surgery, ZNS increased gene expressions of xCT and metallothionein 2A (Fig. 5A). We also observed by double immunofluorescence staining for ChAT and GSH that ZNS increased GSH signal intensity in ChAT-positive motor neurons of CSM rats at 10 weeks after surgery ( Fig. 5 and Supplemental Fig. 4). Taken together, ZNS induced expressions of anti-oxidants and was likely to have exerted a neuroprotective effect on spinal motor neurons against oxidative stress in CSM rats.

Discussion
Five different animal models of CSM have been previously reported. First, twy/twy mouse develops ectopic ossification of ligamentum flavum, which compresses the spinal cord. However, the mouse also develops joint ankylosis, which makes it difficult to evaluate motor dysfunctions associated with CSM 15 . Second, axial growth of the cervical spine was restricted by tying a surgical thread around the C4 level. However, this method is technically challenging, and is difficult to induce CSM invariably 43 . Third, a titanium-screw-based device was used to compress the cervical spinal cord. However, elimination of the cervical laminae and step-wise compression of the spinal cord did not recapitulate human CSM 44 . Fourth, synthetic aromatic polyether adsorbs phosphate anions to induce calcium phosphate sedimentation leading to osteoid formation. The degrees of ossification were predicted to be variable from rat to rat, and we had difficulty in obtaining the material 34,38 . Fifth, waterabsorbing polyurethane elastomers (Aquaprene C, Aquaprene G, and Aquaprene DX) have been used. We also used Aquaprene DX in our rat model. We hope to note three unique features of our model. First, a 0.7-mm-thick implant has been previously used 9,26,28 . We increased the thickness of the implant to 0.8 mm to enable invariable generation of a rat CSM model. Second, cutting the implant in a blade-shape markedly reduced a chance of developing spinal cord injury. Third, evaluation of motor deficits by the CatWalk system enabled early detection of CSM phenotypes, as has been reported by another group 34,38 . In contrast, evaluation of motor deficits with conventional methods require 8-17 weeks after surgery 26,28 . We confirmed reproducibility of CSM phenotypes in our rat model, and used our model to examine the effect of ZNS.
In our rat CSM model, ZNS rescued motor deficits caused by chronic compression of the spinal cord at C5-6 levels ( Fig. 2 and Supplemental Fig. S1). Histologically, ZNS spared the number of spinal motor neurons and ameliorated demyelination of the pyramidal tracts (Figs. 3 and 4). We previously reported that ZNS mitigated H 2 O 2 -induced cell death of primary spinal motor neurons in naive mice 20 . We observed a similar effect of ZNS in primary astrocytes in naive rats (Supplemental Fig. S2A). Amelioration of the development of CSM by ZNS was likely to be accounted for at least partly by induction of expressions of xCT and metallothionein 2A in astrocytes, as well as in neurons to a lesser extent (Supplemental Figs. S2B,C), and also by induction of GSH expression in spinal motor neurons (Figs. 5B,C). ZNS is a pre-approved drug for epilepsy and Parkinson's disease with Scientific RepoRtS | (2020) 10:13138 | https://doi.org/10.1038/s41598-020-70068-0 www.nature.com/scientificreports/