GLT1 gene delivery based on bone marrow-derived cells ameliorates motor function and survival in a mouse model of ALS

Amyotrophic lateral sclerosis (ALS) is an intractable neurodegenerative disease. CD68-positive bone marrow (BM)-derived cells (BMDCs) accumulate in the pathological lesion in the SOD1(G93A) ALS mouse model after BM transplantation (BMT). Therefore, we investigated whether BMDCs can be applied as gene carriers for cell-based gene therapy by employing the accumulation of BMDCs. In ALS mice, YFP reporter signals were observed in 12–14% of white blood cells (WBCs) and in the spinal cord via transplantation of BM after lentiviral vector (LV) infection. After confirmation of gene transduction by LV with the CD68 promoter in 4–7% of WBCs and in the spinal cord of ALS mice, BM cells were infected with LVs expressing glutamate transporter (GLT) 1 that protects neurons from glutamate toxicity, driven by the CD68 promoter, which were transplanted into ALS mice. The treated mice showed improvement of motor behaviors and prolonged survival. Additionally, interleukin (IL)-1β was significantly suppressed, and IL-4, arginase 1, and FIZZ were significantly increased in the mice. These results suggested that GLT1 expression by BMDCs improved the spinal cord environment. Therefore, our gene therapy strategy may be applied to treat neurodegenerative diseases such as ALS in which BMDCs accumulate in the pathological lesion by BMT.


Gene transduction efficiency in BM cells after lentiviral infection and BMT of wild type mice.
To evaluate cell-based gene transduction efficiency in BMDCs after infection with a lentiviral vector (LV) and BMT, we first constructed a yellow fluorescence protein (YFP) expression vector driven by elongation factor (EF)-1 (BOS) promotor (Fig. 1a, LV-BOS-YFP). After applying the LV-BOS-YFP vector to total BM cells, the cells were used for BMT into wild type adult mice. Three and 6 weeks later, flow cytometric analysis of peripheral blood from the mice was performed. We observed 18.2% and 12.4% YFP positive-cells in a representative mouse white blood cells (WBCs) at 3 and 6 weeks after BMT, respectively (Fig. 1b,c), while 86-91% of WBCs were a GFP-positive population from a representative positive control mouse after BMT from GFP mouse (Fig. 1d,e). The population of YFP-positive cells was significantly decreased from 3 (18.2 ± 1.1%) to 6 weeks (13.2 ± 0.9%) after BMT (Fig. 1e).

Gene transduction efficiency in BM cells after lentiviral infection and BMT of SOD1(G93A)
mice. Next, the cells were used for BMT into 8-week-old SOD1(G93A) mice after applying the LV-BOS-YFP vector to total BM cells (Fig. 2). We observed 13.0% and 11.2% YFP-positive cells among mouse peripheral WBCs at 3 and 6 weeks after BMT (Fig. 2a,b). The population of YFP-positive cells was significantly decreased from 3 (14.4 ± 1.5%) to 6 weeks (12.1 ± 1.1%) after BMT in ALS mice (Fig. 2c). And, we observed spinal cord tissues to confirm whether these cells had migrated into nervous system tissues. Spinal cords were isolated from SOD1(G93A) mice at 6 weeks after BMT. YFP fluorescent signals were observed beside neurons marked with Nissl stain in sections of the spinal cords in the LV-BOS-YFP group, but we did not detect YFP-positive signals in the SOD1(G93A) mice after BMT transduced with empty vector (LV-BOS) as control group (Fig. 2d). These results indicated that YFP-positive BMDCs had migrated and accumulated in the pathological lesion of the spinal cord under the ALS condition. This suggests the potential of the therapeutic strategy for ALS by cell-based gene delivery to the spinal cord using BMDCs.
Furthermore, to develop more specific gene delivery to the spinal cord in ALS mice, the CD68 promoter was used to construct an LV (Fig. 3). Whole BM cells were prepared from wild type mice and infected with the LV expressing YFP protein driven by the CD68 promoter (Fig. 3a, LV-CD68-YFP). These cells were then transplanted into SOD1(G93A) mice. Populations of 7.1% and 4.5% showed YFP gene transduction among peripheral WBCs of SOD1(G93A) mice at 3 and 6 weeks after BMT (Fig. 3b,c). The population of YFP-positive cells was significantly decreased from 3 (7.3 ± 1.1%) to 6 weeks (4.8 ± 0.7%) after BMT in SOD1(G93A) mice (Fig. 3d). In the histological analysis of the spinal cord at 6 weeks after BMT, YFP-positive cells were observed beside neurons marked with Nissl stain in the LV-CD68-YFP group similarly to the LV-BOS-YFP group, whereas we did not detect YFP-positive signals in the empty vector control group (LV-CD68) (Figs. 2d, 3e). YFP positive population in LV-CD68-YFP group was less than in LV-BOS-YFP group in peripheral blood because CD68 population was a part of WBCs. However, YFP positive cells in LV-BOS-YFP group were similarly observed in LV-CD68-YFP group in spinal cord of SOD1(G93A) mice (see Supplementary Fig. 1). These points seem to show that CD68 promoter is more efficient than BOS promoter for gene transduction based on BMDCs in ALS. Additionally, CD68 immunostaining was performed on spinal cord sections of ALS mice at 10 weeks after BMT to confirm that YFP was expressed in CD68-positive cells (Fig. 3f). In the LV-CD68-YFP group, YFP signals were merged with 38.1 ± 2.9% of CD68-positive cells, whereas no YFP signals were observed in the LV-CD68 group (Fig. 3f,g). These results suggested that the LV expressing a therapeutic gene driven by the CD68 promoter may be a powerful tool as a therapeutic strategy for ALS by cell-based gene delivery to the spinal cord using BMDCs.
Expression level of the transduced gene gradually increases during disease progression in SOD1(G93A) mice. To investigate how the expression level of a delivered gene changed during disease progression of ALS, histological analysis was performed on spinal cords of SOD1(G93A) mice at 0 week as pre-symptomatic state, 6 weeks as mid stage of disease and 10 weeks as end stage of disease, after BMT (Fig. 4). Immunostaining of ionized calcium binding adaptor molecule 1 (Iba1), as a microglial marker, was performed and its correlation with YFP signals was observed at 0, 6 and 10 weeks after BMT (Fig. 4a). Iba1 positive cells were diffusely observed in the spinal cords of ALS mice (Fig. 4a). Most YFP signals were merged with Iba1 staining. Many more Iba1-positive cells in the 10-week group were found compared with the 0-week and 6-week group. Then, to evaluate the origin of Iba1-positive cells of spinal cord in ALS mice, we counted the cell number of Iba1 and YFP double positive cells as BMDCs (Iba1 + YFP +), and that of Iba1 positive and YFP negative cells  (Fig. 4b,c). The cell number of both Iba1 + YFP + and Iba1 + YFP − cells gradually increased from 0-week to 6-week group, and from 6-week to 10-week group (Fig. 4b). Additionally, the percentage of Iba + YFP + cells in Iba1 positive population gradually increased as the disease progressed (Fig. 4c). These data suggested that overall YFP expression had increased with the elevation of accumulated Iba1-positive cells in the spinal cord as the disease progressed. Therefore, incorporation of a therapeutic gene into the LV-CD68-YFP vector was considered to deliver more therapeutic gene expression to the spinal cord of ALS mice as the disease progresses.
GLT1 gene delivery based on BMDCs improved motor function and survival in SOD1(G93A) mice. For treatment of ALS model mice, GLT1 was inserted into the LV-CD68-YFP vector (LV-CD68-GLT1-YFP, Fig. 5a). To investigate whether GLT1 suppressed disease development, BM cells from wild type mice were infected with LV-CD68-GLT1-YFP and BMT using these cells was performed in 8-week-old SOD1(G93A) mice. After BMT, motor functions and survival were evaluated by the rotarod test and the physiological condi-  www.nature.com/scientificreports/ tion was monitored each week until physiological death and compared with SOD1(G93A) mice transplanted with non-infected BM from SOD1(G93A) mice or LV-CD68-YFP-infected BM from wild type mice (Fig. 5b,c). SOD1(G93A) mice in the LV-CD68-GLT1-YFP group showed significant improvement of motor behaviors compared with those in the SOD1(G93A) and in the LV-CD68-YFP control groups at 16-22 week (Fig. 5b). And LV-CD68-YFP control group showed significant improvement of motor behaviors compared with those in the SOD1(G93A) group at 16-18 week (Fig. 5b). In addition, Kaplan-Meier curves showed that the survival rate was significantly prolonged in the LV-CD68-GLT1-YFP group compared with the LV-CD68-YFP group, and in the LV-CD68-YFP group compared with SOD1(G93A) group, similarly to the trend of motor functions (Fig. 5c).
These results indicate that BMDCs expressing GLT1 is most effective, however BMT alone from wild type mice is also effective.
GLT1 protein and gene expression and glutamate contents in SOD1(G93A) mice after gene therapy. To analyze the mechanism underlying the beneficial effects of LV-CD68-GLT1-YFP in SOD1(G93A) mice, immunohistochemistry of GLT1 was performed on the spinal cords of LV-CD68-YFP and LV-CD68-GLT1-YFP groups, and the immunostaining and YFP signals were observed together (Fig. 6a). A large number of YFP-positive areas were diffusely observed throughout the spinal cords of both LV-CD68-YFP and LV-CD68-GLT1-YFP groups. However, GLT1-positive cells in the LV-CD68-GLT1-YFP group were clearly and abundantly present, which overlapped with YFP signals (Fig. 6a, arrowheads), whereas GLT1-positive staining was only slightly detected in the LV-CD68-YFP group. Additionally, because few GLT1-positive areas in the LV-CD68-YFP group did not overlap with YFP (Fig. 6a, arrows), these findings suggested that most of the GLT1-positive areas in the LV-CD68-GLT1-YFP group had a BM origin, whereas those areas in the LV-CD68-YFP group had an endogenous origin rather than BM. Quantitative analysis revealed that the intensity of the GLT1-positive area www.nature.com/scientificreports/ was approximately three times larger in the LV-CD68-GLT1-YFP group than in SOD1(G93A) disease control and LV-CD68-YFP groups (Fig. 6b). To clarify the relation of Iba1 positive cells and bone marrow-derived YFP positive cells with GLT1 expression, double staining of Iba1 and GLT1 was performed in LV-CD68-GLT1-YFP treatment group (Fig. 6c). Many Iba1 positive cells were diffusely observed in over the ventral horn of the spinal cord and some population of them were partly merged with YFP and GLT1 (Fig. 6c, arrowheads). And many YFP negative and Iba1 positive cells were also observed. And small number of YFP negative Iba1 negative cells expressed GLT1 (Fig. 6c, arrows). Additionally, quantitative analysis of GLT1 mRNA was performed on the spinal cords of SOD1(G93A), LV-CD68-YFP, and LV-CD68-GLT1-YFP groups. Similarly, GLT1 gene expression was particularly increased in the LV-CD68-GLT1-YFP group compared with the other groups (Fig. 6d). Furthermore, to confirm the protective action of GLT1, we measured glutamate contents in cerebrospinal fluid (CSF) of SOD1(G93A), LV-CD68-YFP, and LV-CD68-GLT1-YFP groups (Fig. 6e). The concentration of glutamate in CSF was significantly decreased in the LV-CD68-GLT1-YFP group compared with the other two groups (Fig. 6e). These data suggested that more exogenous GLT1 was expressed in the spinal cords of the LV-CD68-GLT1-YFP group, which indicated that cells expressing the therapeutic gene was efficiently delivered to the pathological lesion in the spinal cord of SOD1(G93A) mice.
Histological analysis of motor neuron survival, astrogliosis, muscle atrophy and nerve degeneration in SOD1(G93A) mice after gene therapy. To confirm the protective effects of GLT1 expression by BMDCs, motor neuron survival, astrogliosis, muscle atrophy and nerve degeneration were analyzed in the LV-CD68-GLT1-YFP group and compared with the SOD1(G93A) and the LV-CD68-YFP groups (Fig. 7). In spinal cord sections, Nissl stain for survival neurons was performed at ventral horn in the SOD1(G93A), in the LV-CD68-YFP and in the LV-CD68-GLT1-YFP groups ( Fig. 7, Neuron survival). Motor neurons were preserved in LV-CD68-GLT1-YFP group more than other two groups, and intensity of Nissl stain was approximately twice higher than the other groups ( Fig. 7, most upper two rows and their right-side bar graph). For the evaluation of astrogliosis, GFAP immunostainings and their quantification were performed in same three groups ( Fig. 7, Astrogliosis). GFAP staining in the LV-CD68-GLT1-YFP group was the weakest among the three groups, and the staining was suppressed to approximately half that of the other two groups (Fig. 7, third and fourth rows and their right-side bar graph). Next, we evaluated skeletal muscle and peripheral nerve as the peripheral compartment ( Fig. 7, Skeletal muscle and Peripheral nerve). In both muscle and nerve, YFP positive signals were observed patchy in the LV-CD68-YFP and in the LV-CD68-GLT1-YFP groups, and they were similar frequent in the two groups ( Fig. 7, fifth and seventh rows, and Supplementary Fig. 2). However, muscle fiber area and S100 protein, as a myelin marker, were significantly preserved in the LV-CD68-GLT1-YFP group compared to the other two control groups (Fig. 7, sixth and eighth rows and their right-side bar graphs). In addition, the distribution of muscle fiber area was shifted to right side in the LV-CD68-GLT1-YFP group compared to the other two groups (See Supplementary Fig. 3). These results suggested that GLT1 expression provided by BMDCs protected pathological changes both in the spinal cord and in peripheral nervous system.

Gene expression of cytokines and neuroprotective microglia markers in SOD1(G93A) mice after gene therapy.
To clarify the mechanism of the therapeutic effects by GLT1 gene delivery and expression, we performed quantitative PCR analyses of inflammatory and anti-inflammatory cytokines, and neuroprotective microglia markers in the spinal cord tissues of treated SOD1(G93A) mice (Fig. 8). Expression of IL-1β as an inflammatory cytokine was significantly suppressed in the LV-CD68-GLT1-YFP group, and that of IL-4 as an anti-inflammatory cytokine was significantly increased in the LV-CD68-GLT1-YFP group compared with the other two groups (Fig. 8). Additionally, expression of arginase (Arg) 1 and FIZZ as neuroprotective microglia markers was significantly increased in the LV-CD68-GLT1-YFP group (Fig. 8). However, expression of TNF-α as an inflammatory cytokine and that of IL-10 as an anti-inflammatory cytokine were not different in the LV-CD68-GLT1-YFP group compared with the other two groups (Fig. 8). These results suggested that expression of GLT1 in the pathological lesion induced a beneficial environment for neurons in the spinal cord.
Numerous studies have reported that novel treatments including stem cell therapy, administration of growth factors, and gene therapy might prolong survival and delay the progression of symptoms [3][4][5][6][7][8]19 . Stem cells have emerged as an attractive option to treat ALS because they tend to migrate to damaged nerves, which offers a means to deliver therapeutic genes where they are needed 8,31,32 . Gene therapy may facilitate curing ALS if vectors can carry therapeutic genes to salvage dying nerve cells. Our current study combined gene therapy with stem cells, which showed that BMDCs expressing GLT1 by the LV in transgenic SOD1(G93A) mice resulted in expression of GLT1 at segments of the spinal cord and led to an extension of lifespan and improved motor functions. The pathogenesis involved in motor neuron death of ALS is complex and neuroinflammation has been accepted as a major contributor to motor neuron degeneration and disease progression 33 . Recently, it was noted that non-neuronal cells, such as immune cells, endothelial cells, and glial cells surrounding neurons, are deeply involved in the pathogenesis of neurodegenerative diseases in addition to the autonomous neuronal cell death caused by accumulated abnormal proteins 34,35 . This theory of "non-cellular autonomic nerve cell death" is also attracting attention in terms of ALS, a neurodegenerative disease 34,35 . Microglia are distributed throughout the brain and spinal cord parenchyma, which account for 10%-20% of the total glial cell population 36 .
We have reported that a large number of microglia accumulate in the spinal cord as the stage progresses in an ALS model animal 19 . Similarly, our results showed that YFP expression increased with the elevation of accumulated Iba1-positive cells in the spinal cord as the disease progressed in this study. Therefore, the migrating www.nature.com/scientificreports/ YFP-positive cells acted as gene carriers and the number of migrating cells was increased gradually in the spinal cord as the stage progressed, which increased the expression of therapeutic genes. Microglia are extremely sensitive to physiological changes in their environment and become "activated" following exposure to specific cytokines and growth factors, which indicates infection, trauma, neuronal insult, or inflammation 37 . The levels of several proinflammatory cytokines are altered during ALS, which suggest the presence of inflammation. During chronic neuroinflammation, CNS-infiltrating macrophages express microglial markers and convert to significantly different phenotypes [38][39][40] . Microglia primarily have two different phenotypes, proinflammatory and neuroprotective, in response to various microenvironmental signals 41 . On the basis of different activation stimuli, microglia polarize to either the proinflammatory or neuroprotective type. During the early stage of motor neuron injury in ALS models, the surveying microglia exhibit the neuroprotective phenotype 42 . However, as the disease progresses, microglia shift to the proinflammatory phenotype and injure motor neurons 43 . In this study, expression of IL-1β by proinflammatory microglia was significantly suppressed in the LV-CD68-GLT1-YFP group. This is in agreement with reports showing that chronic administration of IL-1β results in neurodegeneration 44 , whereas IL-1β depletion attenuates inflammation and prolongs the lifespan of ALS mouse models 45 . Additionally, expression of IL-4, Arg1, and Fizz as the neuroprotective microglial markers 46,47 was significantly increased in the LV-CD68-GLT1-YFP group, which may have been expressed by the infiltrating BMDCs transformed to Iba1-positive microglia-like cells, or the induced resident neuroprotective microglia in spinal cord with gene therapy. CNS delivery of IL-4 in SOD1(G93A) mice via a lentiviral-mediated gene therapy strategy has resulted in general amelioration of clinical outcomes during the early slowly progressive phase of the disease 48 . Therefore, the migration of BMDCs and expression of GLT1 in the spinal cord are thought to have improved the environment around neurons from inflammatory to non-inflammatory. In addition, as a direct effect of GLT1 expression in BMDCs, glutamate content was significantly reduced in CSF from the treated mice. This reduction could be caused by taking glutamate into BMDCs through GLT1. It has been reported that the loss of GLT1 induce increased extracellular levels of glutamate and cause motor neuron toxicity and muscle paralysis in animal models 49 , and the pharmacological stimulation of GLT1 rescue motor neuron degeneration in SOD1(G93A) mice 50 . Excess glutamate, which has been observed in CSF in ALS patients, has been reported to be toxic to neuron 51 . Therefore, these results support that the treatments have protected neurons from neurotoxicity of glutamate. Furthermore, as astrogliosis, muscle atrophy and peripheral nerve degeneration were significantly suppressed, the infiltrating BMDCs and their expression of GLT1 may have affected astrocyte in spinal cord, muscle cells and Schwann cells in peripheral tissues in ALS mice.
As a represent study of BM cell-based gene therapy, treatment of adrenoleukodystrophy (ALD) has been performed. ALD is a genetic disease linked to X chromosome and caused by the dysfunction of peroxisomal ATP-binding cassette (ABC) half-transporter ALD protein due to the mutation of ABC, subfamily D, member 1 (ABCD1) gene 52 . Cerebral ALD shows demyelination and neurodegeneration 52 . Progression of the disease results in loss of neurological functions and death, which can be stopped only by the transplantation of allogeneic hematopoietic stem cells 52 . Recently, a cell transplantation therapy of the autologous CD34 + cells transduced by an elivaldogene tavalentivec (Lenti-D) LV has been reported 53 . This study indicates that Lenti-D gene therapy safely and effectively halts the progression of ALD and may be offered instead of allogeneic stem cell transplantation in patients with cerebral ALD 53 . This study supports the feasibility of our strategy because it is a clinical application of BM cell-based gene therapy. Recently, a literature reported that bone marrow derived-cells were not accumulated into spinal cord of ALS mice without irradiation with BMT 54 . However, our finding that many BMDCs accumulate in the spinal cord as the disease progresses is very attractive, even with the effects of irradiation. In fact, BMT has been previously performed on ALS patients after total body irradiation 55 . Therefore, it is believed that our strategy seems to be acceptable as a clinical application while further safety should be confirmed.

Isolation of bone marrow cells and infection of lentiviral vectors. Whole BM cells were isolated
from wild type C57BL/6 mice as described previously 19 . Before BMT into wild type or SOD1(G93A) mice, 1 × 10 6 BM cells per mouse were infected with LVs at a concentration of 10 MOI (100 vp/cell) in 1 ml Span Serum-free expansion medium (STEMCELL Technologies, Vancouver, Canada) with stem cell factor (100 ng/ ml; R&D Systems, Minneapolis, MN), Flt3 (100 ng/ml; R&D Systems), IL-3 (100 ng/ml; R&D Systems), and IL-6 (200 ng/ml; R&D Systems). After 12 h of incubation at 37 °C, the lentivirus-infected BM cells were used for BMT as gene therapy for SOD1(G93A) mice. Before BMT, transduction efficiency of all LVs to BM cells were confirmed.
Bone marrow transplantation and gene therapy of SOD1(G93A) mice. For BMT, 8-week-old female SOD1(G93A) and wild type female C57BL/6 mice were irradiated (9 Gy) and then injected from tail vein with 1 × 10 6 whole BM cells after lentiviral infection. After BMT, these mice were used to evaluate gene transduction efficiency and gene therapy.

Flow cytometric analysis.
Transduction efficiency of the transferred gene was evaluated by flow cytometric analysis in WBCs from wild type and SOD1(G93A) mice at 3 and 6 weeks after BMT because it takes approximately three to four weeks for the stable engraftment of bone marrow cells after transplantation. WBCs in 200 µl blood were collected from each mouse by cutting the tip of tail, and centrifuged at 100×g for 5 min at 4 °C. After the pellet of blood cells was resuspended with 1 ml lysis buffer (154 mM NH 4 Cl, 2 mM NaHCO 3 and 0.1 mM EDTA 2 Na), that was centrifuged at 100×g for 5 min at 4 °C. Then, 1 ml PBS (−) was added to the cell pellet, followed by centrifugation at 100×g for 5 min at 4 °C again. After 500 µl PBS (−) was added to the cell pellet, for the cells were analyzed by flow cytometry. Emission at 530 was induced by excitation of a 488 nm laser. The gate of the fluorescence threshold was determined using WBCs from C57BL/6 mice after BMT from GFP or wild type mice as positive and negative controls, respectively. A FACS Canto II with FACS DIVA software (BD Biosciences) was used for data collection and analysis.
Quantitative RT-PCR analysis. The mice were transcardially perfused with PBS and the spinal cord tissues at the lumbar level were collected under deep anesthesia by intraperitoneal injection of medetomidine  Lower row shows Hematoxylin-eosin stain of anterior tibial muscle in the same three group. The first and second row from the bottom: S100 immunohistochemistry (red) with YFP (green) signals of the spinal cords in the same three group. Upper row shows the color images and lower row shows black and white images of red color (S100 staining) isolated from the corresponding upper row. Scale bar = 100 µm. The bar graph shows relative intensity of Nissl, GFAP and S100 staining as seen in the same three group (n = 5 in each group). The intensity of Nissl, GFAP and S100 staining was measured in the black and white image using the Image J software and the ratio was calculated against the intensity of that in SOD1(G93A) group. The bar graph shows relative area of muscle fiber in the same three groups. The average areas of muscle fibers were compared among the three groups (n = 5 in each group). Error bars represent the mean + SD. **p < 0.01 between the treatment group and others. Behavior test. Rotarod tests (Ugo Basile, Comerio-Varese, Italy) were conducted once a week from 1 week before the beginning of treatment to physiological death (when the rotarod test result was 0 s, the mice were considered as physiologically dead). Rotarod tests were performed at a rotation speed of 5 rpm/min to a maximum of 50 rpm/min for 5 min (acceleration was 9 rpm/min 2 ) as previously described 19 . The averages of three medians in five trials for each mouse with an interval of > 3 min were calculated and used to evaluate motor functions 19 . SOD1(G93A) mice were grouped in a random manner. Rota-rod test was performed in a non-blinded manner. Mice were adjusted age and sex, and the tests were performed by the same enforcer. The number of living mice was counted in accordance with the definition of physiological death until all mice were recognized as such for the analysis of Kaplan-Meier survival curve 19 . And there was no censoring of mice due to treatment-related death.
Histological analysis. For histological analysis, SOD1(G93A) mice after treatment were transcardialy perfused with PBS and fixed with 4% paraformaldehyde. Then, the spinal cord at the lumbar level, anterior tibialis muscle and sciatic nerves were isolated and their sections were prepared by embedding with OCT compound (Sakura Finetek Japan, Tokyo, Japan) and cutting. For immunostaining, the sections were incubated with a primary antibody (anti- www.nature.com/scientificreports/ signals and GFAP-GLT1-or S100-positive immunostaining were converted to the black white image after splitting to single color, and the intensity was measured in over ten scenes of each mouse by ImageJ software version 1.51 (National Institutes of Health, Bethesda, MD) 57 . To analyze the muscle degeneration, the sections were stained with hematoxylin-eosin or mounted with the Vectashield mounting medium with DAPI (Vector Laboratories) 57 . The muscle fibers' area was measured in over ten scenes of each mouse by ImageJ software version 1.51 (National Institutes of Health).
Glutamate assay. For glutamate assay, the cerebrospinal fluid was collected from the cisterna magna of 18-week-old SOD1(G93A) mice after transplantation. The concentration of the glutamate levels in the cerebrospinal fluid were measured by Glutamate Assay Kit according to the manufacturer's protocol (ab83389, abcam).
Statistical analysis. Statistical analysis was performed using SPSS 25.0 software (IBM Corp., Armonk, NY). All data are shown as the mean ± standard deviation (S.D.). One-way ANOVA and Tukey's test was used for analysis of statistical significance for multiple datasets. The log-rank test was performed for statistical analysis of the Kaplan-Meier curve for a univariate survival. The analysis was performed with GLT1 treatment as an independent variable and the survival as a dependent variable. Data were considered significantly different at p < 0.05.

Data availability
All the data for this study will be available upon reasonable request to the corresponding author.