Author contributions: B.A.S.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; X.Z. and J.A.S.: conception and design, collection and/or assembly of data, and data analysis and interpretation; A.C.B., P.A.G., S.Z., and S.C.L.: collection and/or assembly of data and data analysis and interpretation; J.M.F.-P.: collection and/or assembly of data; R.W.B. and L.M.: data analysis and interpretation; A.V.K.: conception and design and data analysis and interpretation; B.A.B.: conception and design, financial support, administrative support, data analysis and interpretation, and final approval of manuscript.
Globoid cell leukodystrophy (GLD) is a common neurodegenerative lysosomal storage disorder caused by a deficiency in galactocerebrosidase (GALC), an enzyme that cleaves galactocerebroside during myelination. Bone marrow transplantation has shown promise when administered to late-onset GLD patients. However, the side effects (e.g., graft vs. host disease), harsh conditioning regimens (e.g., myelosuppression), and variable therapeutic effects make this an unsuitable option for infantile GLD patients. We previously reported modest improvements in the twitcher mouse model of GLD after intracerebroventricular (ICV) injections of a low-dose of multipotent stromal cells (MSCs). Goals of this study were to improve bone marrow-derived MSC (BMSC) therapy for GLD by increasing the cell dosage and comparing cell type (e.g., transduced vs. native), treatment timing (e.g., single vs. weekly), and administration route (e.g., ICV vs. intraperitoneal [IP]). Neonatal twitcher mice received (a) 2 × 105 BMSCs by ICV injection, (b) 1 × 106 BMSCs by IP injection, (c) weekly IP injections of 1 × 106 BMSCs, or (d) 1 × 106 lentiviral-transduced BMSCs overexpressing GALC (GALC-BMSC) by IP injection. All treated mice lived longer than untreated mice. However, the mice receiving peripheral MSC therapy had improved motor function (e.g., hind limb strength and rearing ability), twitching symptoms, and weight compared to both the untreated and ICV-treated mice. Inflammatory cell, globoid cell, and apoptotic cell levels in the sciatic nerves were significantly decreased as a result of the GALC-BMSC or weekly IP injections. The results of this study indicate a promising future for peripheral MSC therapy as a noninvasive, adjunct therapy for patients affected with GLD. STEM Cells2013;31:1523–1534
Lysosomal storage diseases (LSDs) affect 1 out of every 5,000–7,000 live births and, collectively, represent some of the most common inherited diseases in pediatric populations [1–3]. Globoid cell leukodystrophy (GLD; Krabbe's disease) is an autosomal recessive neurodegenerative LSD caused by a deficiency of galactocerebrosidase (GALC), an enzyme responsible for the catabolism of myelin sphingolipids, such as galactocerebroside (GalCer). Although asymptomatic at birth, infantile GLD patients present with symptoms before 6 months of age, deteriorating rapidly due to toxic substrate accumulation (e.g., psychosine) and extensive demyelination of the central and peripheral nervous systems [4–6]. The symptoms resulting from such pathology include febrile episodes, inconsolable fits, arrest of mental and motor development, hypertonicity, seizures, and paralysis followed by premature death within 2 years [5, 7–9].
The twitcher mouse, a model of GLD, has striking pathological similarities to the human patient, including axonopathy, severe gliosis, globoid cell formation, widespread apoptosis and inflammation, and a diffuse demyelinating peripheral neuropathy [10–12]. Between postnatal days (PND) 15–20, the GALC deficient mouse begins to show symptoms of motor function deterioration, weight reduction, decreased locomotion, and behavioral abnormalities [13–15]. If left untreated, twitcher mice experience rapid motor and neurological deterioration and death by PND30–40. The severity of the twitcher phenotype makes these mice a useful genetic model for high-throughput development and screening of therapeutic approaches for GLD and other LSDs.
Hematopoietic stem cell transplantation (HSCT) has recently shown promising results (e.g., delayed onset of symptoms and increased lifespan) when administered to asymptomatic late-onset GLD patients [16, 17]. This therapy has also been supportive in several infantile GLD patients and twitcher mice [16, 18–20]; these patients and mice, however, have all succumbed to a progressive disease [9, 21, 22]. Furthermore, there are serious risks associated with HSCT due to complications of the procedure (e.g., graft vs. host disease, GVHD) or the aggressive conditioning regimens [23, 24]. Studies using the twitcher mouse have shown that HSCT, specifically bone marrow transplantation (BMT), exacerbates the tremor, fails to provide substantial peripheral nerve improvement, and has a 20% mortality rate due to severe hemorrhage [19, 25]. The lack of effective treatment options, in addition to the severity of GLD, highlights the need for development of innovative therapeutic approaches.
Recent studies have demonstrated that enzyme replacement therapy (ERT) [26, 27], substrate reduction therapy , vector therapy [29–31], and anti-inflammatory agents  improve the twitcher mouse pathology and modestly prolong lifespan. However, these therapeutic approaches are only partially successful even when combined with BMT [22, 29]. An alternative approach—transplantation of multipotent mesenchymal stromal cells (MSCs)—has been studied for several LSDs due to the potential of these cells to migrate to sites of tissue injury, deliver therapeutic gene products and trophic factors, replace affected cells through engraftment and differentiation, and decrease inflammation in neurodegenerative disease states [33–37]. MSCs, which can undergo both self-renewal and multilineage differentiation, are easily harvested and expanded ex vivo and are capable of providing neuroprotection and immunomodulation while evading host immune surveillance [35, 38, 39]. Interestingly, MSCs are currently being used in conjunction with BMT to decrease the risk of GVHD and improve engraftment in the clinical setting [40–42]. Thus, alleviation of symptoms in the twitcher mouse using MSC therapy would likely lead to development of a potentially safer and more effective HSCT treatment for GLD patients.
Our group recently reported that administration of a single, low-dose of MSCs through intraperitoneal (IP) or intracerebroventricular (ICV) injections improved the phenotype of the twitcher mouse by reducing the levels of inflammation [13, 43]. This study aims to enhance MSC therapy for GLD by increasing the functional GALC levels and anti-inflammatory effects in the twitcher mouse. To accomplish these goals, twitcher mice received peripheral or central-directed MSC therapy in higher cell numbers or increased injection frequency (i.e., weekly IP injections) compared to our previous studies. All MSC treatments were evaluated for effectiveness by assessing improvements in apoptosis and inflammation in peripheral nerves, twitching symptoms, body weight, and hind limb strength. The goal of the novel stem cell therapy in this study was to improve pathology or phenotype in the twitcher mouse in order to develop a preclinical model for a noninvasive adjunct therapy for patients affected with LSDs.
Materials and Methods
Adult (3-month-old) heterozygote (Het; Galctwi/+) C57Bl/6J (B6.CE-Galctwi/J) mice were originally obtained from The Jackson Laboratory (Bangor, ME Jackson Laboratories, Bangor, ME, http://jaxmice.jax.org/) and used as breeder pairs to generate homozygous (Twi; Galctwi/twi) twitcher mice and wild-type (WT) C57Bl/6J mice. All animal procedures were approved by the Institutional Animal Care and Use Committee at Tulane University and conformed to the requirements of the Animal Welfare Act. In total, there were six mouse groups studied: 15 ICV bone marrow-derived MSC (BMSC)-treated WT (Galc+/+), 26 untreated twitcher (Twi, Galc−/−), 18 single IP BMSC-treated twitcher (IP, Galc−/−), 8 weekly IP BMSC-treated twitcher (Weekly, Galc−/−), 9 IP GALC-transduced BMSC-treated twitcher (GALC, Galc−/−), and 7 ICV BMSC-treated twitcher (ICV, Galc−/−) male and female mice were used in this study (Fig. 1). The specific Galc mutation was confirmed as previously described .
Harvesting, Culture, and Characterization of Murine eGFPTgBMSCs
BMSCs were obtained from male eGFP transgenic mice (C57Bl/6-Tg(UBC-GFP)30Scha/J strain; Jackson Laboratory) between 4 and 6 months of age. BMSCs were isolated, characterized, and cultured from the femurs and tibiae of each mouse as previously described . Briefly, the ends of each tibia and femur were removed to expose the marrow. The marrow was pushed out of the bone using a syringe with complete expansion media (CEM), resuspended in CEM, and filtered through a 70 μm nylon mesh filter. The mixture was then centrifuged at 400g for 10 minutes at 4°C, and the pellet was resuspended in 3 ml CEM. CEM consists of Iscove's Modified Dulbecco's Medium (IMDM, Invitrogen, Carlsbad, CA, www.invitrogen.com) supplemented with 9% fetal bovine serum (FBS; Atlanta Biologicals, Lawrenceville, CA, www. atlantabio.com), 9% horse serum (HS; Hyclone Laboratories, Logan, UT, www.thermoscientific.com/hyclone), 100 U/ml penicillin (Invitrogen), 100 μg/ml streptomycin (Invitrogen), 0.25 μg/ml amphotericin B (Invitrogen), and 12 μM l-glutamine (Invitrogen). The cells were then plated, washed with media, and stored in liquid nitrogen or expanded further exactly as described in Ripoll and Bunnell . The cells were characterized as MSCs using osteogenesis, adipogenesis, and chondrogenic differentiation assays, fluorescence-activated cell sorting analysis for detection of cell surface markers, growth rate examination, and a colony-forming unit assay as previously described .
IP Injections of Murine eGFPTgBMSCs
Passage-eight (P8) murine eGFPTgBMSCs were recovered from cryopreservation in CEM, cultured to 70% confluence, and harvested for infusion as previously described . A 50 μl Hamilton syringe with a 30-gauge stainless steel needle was used to administer 50 μl of the cell suspension (1×106 total eGFPTgBMSCs) or Hanks' balanced saline solution (HBSS) into the left side of the peritoneal cavity on PND5–6. The pups were then returned to their mothers.
Lentiviral Transduction of eGFPTgBMSCs
The GALC viral particles were produced by calcium phosphate transient transfection of 293T cells  with the GALC transgene-containing plasmid (pNL-EF1α-GALC-TMH/WPREΔ03, a gift from Dr. Jakob Reiser, FDA/CBER) and two helper plasmids (pMD.G, pHR- Δ8.9). Briefly, the 293T cells were seeded in 15-cm cell culture plates with a concentration of 8 × 106 cells per plate. The transfection by calcium phosphate with the three plasmids in the presence of 25 μM chloroquine (Sigma-Aldrich, St. Louis, MO, www.sigmaaldrich.com) was carried out 24 hours later. The transfectate was replaced by fresh ultraculture serum-free medium (Lonza, Basel, Switzerland, www.lonza.com) after 16–18 hours, and the unconcentrated GALC viral supernatant was collected 48 hours later. The titer of the viral particles was determined by the lentiviral titer kit from MellGen Laboratories, Inc. (Surrey, BC, Canada MellGen Laboratories, Surrey, BC, Canada, www.mellgenlabs.com).
For cell transduction, the green fluorescent protein (GFP+) BMSCs were seeded on six-well cell culture plates with 8 × 104 cells per well. After a 24-hour incubation period, the cells in each well were treated with 1 ml GALC lentivirus in the presence of 0.5 ml IMDM with 10% heat-inactivated FBS and 8 μg/ml polybrene. After 24 hour, the medium containing the GALC lentivirus was replaced with new medium and another 1 ml of GALC lentivirus. The same procedure was repeated again 24 hours later, and then new medium without GALC lentivirus was added and the cells were incubated for another 72 hours. The GALC-transduced GFP+ BMSCs were isolated by single-cell colony selection. The cells were seeded on 10-cm cell culture plate with 100 cells per plate, and cultured for 14 days. The Nikon Object Marker of the microscope was used to mark the desired colonies, and a cloning ring (Fisher Scientific, Pittsburgh, PA, www.fishersci.com) with vacuum grease was put around each desired colony. Isolated colonies were lifted by trypsinization and seeded onto 48-well cell culture plates for expansion. After expansion, each selected colony was stained by a modified β-galactosidase assay  to determine the GALC transduction efficiency, and eGFPTgBMSCs with more than 90% GALC transduction efficiency were used for IP injections as described.
ICV Injection of Murine eGFPTgBMSCs
GFP+ BMSCs were harvested, cultured, and lifted as previously described for the IP injections. The cell suspension was taken up via a sterile 30G needle that was attached by plastic tubing to a 10 μl Hamilton syringe. A stereotaxic instrument was used to immobilize the heads of the cryoanesthesized twitcher and unaffected littermate pups, while 2 μl of the BMSC suspension (5 × 104 BMSCs/μl in HBSS) or HBSS was injected into each lateral ventricle on PND3–4 using previously defined coordinates . The needle was kept immobilized for 2 minutes before withdrawing. The pups were then resuscitated on a heating pad before being returned to their mothers.
Motor Function Testing
Beginning on PND14, motor function tests were performed three times per week to monitor the twitching frequency and severity, wire hang, and hind stride length. Mice were scored based on the scoring systems explained in our previous studies . Video and behavior recording of each mouse was conducted weekly using a LifeCam Cinema webcam with True 720p HD video and Debut Video Capture Software (Microsoft Corp., Redmond, WA, www.microsoft.com) as described in our previous study [13, 43]. Briefly, all video files were uploaded for analysis using the EthoVision XT7 software for quantification of various spatial endpoints . Twitcher mice were euthanized once they lost 20% of their maximum body weight or became moribund.
Tissue Processing and Histological Staining
Animals were euthanized by CO2 asphyxiation and perfused with sterile phosphate buffered saline (PBS) through the left ventricle after rupture of the right atrium. After intracardial perfusion, the brain and sciatic nerves were removed and stored for up to 3 months at 4°C in Allprotect tissue reagent solution (Qiagen, Valencia, CA, www.qiagen.com) or stored at room temperature (RT) in 10% neutral buffered formalin. Paraffin-embedded sections were cut at 5 μm, mounted, and subsequently stained with hematoxylin and eosin (H&E). Slides were scanned using an Aperio ScanScope CS instrument (Aperio Technologies, Inc., Vista, CA, www.aperio.com), and the images were analyzed with Fiji/Image J software (National Institutes of Health, Bethesda, MD, http://imagej.nih.gov/ij/) . For each treatment group, H&E stains of three mice (10 sections/mouse) were analyzed.
Transferase dUTP Nick End Labeling Assay to Detect DNA Fragmentation
Unstained slides of three mice per treatment group (three sections/mouse) for terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) and immunohistochemistry (IHC) were prepared by sectioning paraffin-embedded samples at a thickness of 5 μm. All slides were warmed on a heating platform at 57°C for 30 minutes prior to deparaffinization. For deparaffinization, all slides were submerged in HistoChoice (Amresco, Solon, OH, www.amresco-inc.com) twice for 5 minutes, 100% ethanol twice for 2 minutes, 95% ethanol twice for 2 minutes, 70% ethanol for 2 minutes, 50% ethanol for 2 minutes, distilled (DI) water twice for 2 minutes, and 1× tris-buffered saline (TBS, pH 10.0) twice for 5 minutes. The slides were rinsed with DI water, air dried, outlined using an ImmEdge Pen (Vector Laboratories, Burlingame, CA, www.vectorlabs.com), washed with 20 mM Tris-HCl (pH 8.0), and then incubated with a 1:1,000 proteinase K solution in Tris-HCl pH 10.0 (20 μg/ml) for 15 minutes at 37°C in a humidified chamber.
For any given slide, one section was designated as the positive control and incubated at RT with a DNase I (Invitrogen) solution in PBS (15 U/ml) for 10 minutes. Using the in situ Cell Death/Fluorescein Detection Kit (Roche Diagnostics, Indianapolis, IN, www.rocheUSA.com), all slides were incubated with 50 μl of TUNEL solution for 1 hour at 37°C in a humidified chamber. The slides were washed three times in 1× PBS for 5 minutes before incubation with a 0.4 mM DAPI/TBS solution. ProLong Gold Antifade Reagent (Invitrogen) was then used to mount coverslips. Fluorescent images were acquired at 5× and 10× using a Leica DMRXA2 deconvolution microscope (Leica Microsystems, Grove, IL, www.leica-microsystems.com).
The deparaffinized slides were submerged in 700 ml of citrate buffer pH 6.0 (10 mM) and heated for 20 minutes in a microwave using a low heat setting. After cooling, the slides were washed for 5 minutes in 1× PBS and subsequently washed with PBS-FSG-Tx-100 (10% [vol/vol] 10× PBS, 0.2% [vol/vol] fish skin gelatin, and 0.1% [vol/vol] Triton X-100) for 5 minutes before incubation for 1 hour in a humidified chamber at RT with blocking solution, which consisted of 10% normal goat serum (NGS) in PBS-FSG (10% [vol/vol] 10× PBS and 0.2% [vol/vol] fish skin gelatin). The primary antibody to EGFP (anti-GFP; 1:100, Invitrogen: A-11121 or 11122), mature macrophages (F4/80; 1:10, Santa Cruz: SC-59171 Rat IgG2b), neuronal nuclei (NeuN; 1:50, Chemicon: MAB377 Ms IgG1), neural crest cells (S-100; 1:1,000, Sigma: S-2644 Rb), or astrocytes (glial fibrillary acidic protein [GFAP]; 1:200, Sigma: C9205 Ms IgG1) was diluted in 10% NGS solution and applied to appropriate experimental sections for 1 hour incubation in a humidified chamber at RT. Control slides were treated with secondary antibody-only (2° only). Following incubation, the slides were washed in PBS-FSG-Tx-100 and PBS-FSG, each for 10 minutes. The sections were then incubated in a humidified chamber at RT for 1 hour with the secondary antibody (e.g., anti-rat, anti-rabbit, or anti-mouse IgG antibody conjugated with Alexa 488 or 594) in 10% NGS solution. Slides were then washed twice in PBS-FSG-Tx-100 and once in PBS-FSG. The slides were incubated with a DAPI (50 μM) or TO-Pro-3 (1:1,000)/10% NGS solution, mounted with ProLong medium, and imaged using either a deconvolution or a confocal microscope.
For protein isolation, Allprotect-preserved brain, kidney, and sciatic nerve samples were homogenized in 0.2% Nonidet P40 in PBS using a plastic motorized pestle. Approximately 10 ml of fluid was used per gram of sample. All samples were exposed to 10 rounds of sonication, where each round consisted of 5 seconds of sonication followed by 20 seconds on ice without sonication. The solutions were cleared twice by centrifugation at 12,000g; the supernatants were removed and saved in aliquots at −20°C. Protein concentration was determined using the bicinchoninic acid assay kit (Pierce, Rockford, IL, www.piercenet.com), and all protein lysates were centrifuged using centrifugal filter units with microporous membranes (Millipore, Billerica, MA, www.millipore.com).
Protein lysate (30–100 μg), NuPage Reducing Sample Agent (10×, Invitrogen), NuPage LDS Sample buffer (4×, Invitrogen), and DI water were mixed, centrifuged at 1,000g for 1 minute, then boiled for 5 minutes in a PCR PTC-200 thermal cycler (MJ Research, Waltham, MA, www.gmi-inc.com). The samples (32 μl total) and 2–3 μl of a MagicMark XP ladder (Invitrogen) were run through a NuPage 4%–12% Bis-Tris 1.5 mm gel in 1× 3-(N-morpholino)propanesulfonic acid (MOPS) (Invitrogen) running buffer for 1 hour at 200 V; NuPage antioxidant (0.5 ml, Invitrogen) was added to the MOPS in the inner chamber. An iBlot system (Invitrogen) was used with the iBlot transfer stack (Invitrogen) using program 3 for 7 minutes to transfer the protein to a nitrocellulose membrane. The membrane was then incubated with 10 ml of premade blocking solution (Bløk noise canceling reagent for chemiluminescent detection, Millipore) at RT, rinsed, and then incubated in PBST (1× PBS with 0.1% Tween-20) for 1 hour at RT followed by overnight at 4°C with the primary antibody to GALC (anti-GALC; 1:2,000, ProteinTech: 11991-1-AP), GFP (anti-GFP; 1:1,000, Invitrogen: A-11122), or Actin (anti-actin; 1:3,000, Sigma-Aldrich: A2066). The membranes were washed three times in 10 ml of PBST with shaking before being gently rocked at RT for 1 hour with an HRP (horseradish peroxidase)-conjugated anti-rabbit IgG secondary antibody (1:1,000; Cell Signaling) in blocking solution. After three washes with PBST, the membrane was incubated for 2 minutes with the Novex ECL HRP chemiluminescent substrate reagents (Invitrogen) and then developed using an ImageQuant LAS 4000 imager (GE Healthcare Biosciences, Pittsburgh, PA, www.gelifesciences.com).
For detection of GFP-specific antibodies in the sera of cell-transplanted mice, purified GFP (BioVision, San Francisco, CA, www.biovision.com) was run in a single-well NuPage 10% Bis-Tris 1.0 mm gel. After transfer of the protein to a nitrocellulose membrane, the membrane was cut into strips, which were individually blocked and incubated for 1 hour at RT followed by overnight at 4°C with mouse serum (1:20 in PBST) of a single treatment group. After washing, the membranes were incubated with HRP-conjugated anti-mouse IgG secondary antibody (1:1,000, Cell Signaling: 7076) in blocking solution and then imaged with an ImageQuant LAS 4000 imager (GE Healthcare Biosciences) as described above.
GALC Activity Assay
For preparation of the BMSCs or GALC-BMSCs, the cells were trypsinized and pelleted as performed for injection purposes and then lysed by sonication in 200 μl of 20 mM acetate buffer (pH 4.5) while on ice. A total of 150 μl lysed cell suspension or 500 μl brain protein lysates in 25 μl of 200 mM acetate buffer (pH 4.5) was incubated with 50 μl of a 3H-GalCer (Moravek Biochemicals, Brea, CA, www.moravek.com) substrate mixture (approximately 31,000 cpm) that was prepared as previously described . All enzymatic reactions were stopped, vortexed, centrifuged, extracted, and measured for radioactivity using a scintillation counter (TRI-CARB 1600CA, PACARD) as previously described without modifications . GALC activity is expressed as nanomoles of GalCer hydrolyzed per hour per milligram of protein (nmol/hour per mg protein) or per 1 × 106 cells (nmol/hour per 1 × 106cells).
Fluorescent Substrate Cleavage Assay
In order to calculate the activity of three endogenous enzymes, β-galactosidase (β-gal), α-galactosidase (α-gal), and β-hexosaminidase (β-hex), a total of 5 μl of the lysed BMSC or GALC-BMSC cell suspension was incubated with 10 μl (20 nmol) of a 4-methylumbelliferyl-β-d-galactopyranoside (4MU-β-gal), 4MU-α-d-galactopyranoside, or 4MU-β-N-acetyl-d-glucosaminide, 40 μl of 50 mM acetate buffer at pH 4.0 or 7.0, and 50 μl DI water for 0.5–3 hours. At the preset time, the enzymatic reactions were quenched by adding 0.7 ml of 20 mM borate buffer pH 9.8. A Turner Quantech fluorometer (Barnstead/Thermolyne Corporation, Dubuque, IA, www.thermoscientific.com) was used to measure the fluorescence of the liberated 4MU in each sample . The nanomoles of substrate cleaved per hour per 10,000 cells (nmol/hour per 1 × 104 cells) were calculated based on a standard curve.
For comparisons of body weight, wire hang, twitching, and hind stride length, mixed models regression methods were implemented via PROC MIXED in Statistical Analysis Software (version 9.2). For tests with a significant (p < .05) time × group interaction, pairwise comparisons of least square means were made to further investigate the interaction. When applicable, the WT group was compared to all twitcher groups. The untreated twitcher group was compared to the treated twitcher groups. Lastly, the treated groups were compared to each other. To correct for multiple pairwise comparisons, the p-values were adjusted using a Bonferroni correction, and significance was determined if p < .01 for the weight, hind stride, wire hang, and twitching studies. For the wire hang and twitching data, exact chi-square tests were also performed at each time point to test for differences between the treatment groups.
For Kaplan-Meier survival curves, analyses were performed using the log-rank test, and specific lifespan differences between the Twi group and the multiple treatment groups were determined using the Sidak-adjusted log-rank test. For the behavioral, neurophenotyping, and histological studies, statistical analysis of three or more groups was performed using one-way analysis of variance (ANOVA) followed by pairwise comparisons of the mouse groups using post hoc testing with Bonferroni correction. For any study where the mouse groups were monitored for 3 consecutive weeks, two-way ANOVA (factors: time and mouse group) with repeated measures and subsequent Bonferroni post hoc testing was performed. Significance for the overall group effect and individual pairwise comparisons was defined as p < .05. One- and two-way ANOVA tests were performed using GraphPad Prism 4.0b for Macintosh (GraphPad Software, San Diego, CA, www.graphpad.com), and all values were reported as mean ± SEM.
Weight and Longevity
The Twi mice had a median lifespan of 36 days, whereas the ICV, IP, Weekly, and GALC groups had median survival times of 43, 42, 41.5, and 44 days, respectively (Fig. 1A), compared to a 44-day median lifespan of mice receiving BMT with low radiation conditioning in a previous study . Using Sidak-adjusted differences, we found that the Twi group had a significantly lower median survival time when compared with any of the four treated groups (p < .0001). There were no significant differences between the survival curves of the treatment groups. At PND42, only 3.6% of Twi mice was alive compared to 37.5%, 38.9%, 66.7%, and 77.8% for the Weekly, IP, ICV, and GALC mice, respectively. Weight was an objective measurement used to compare all mouse groups between PND16 and PND40, and there was a significant (p < .0001) group × time interaction (Fig. 1B). By PND28, the Twi group was significantly lower in weight than the Weekly and GALC groups, and the Twi mice had lower weights than every treatment group between PND30 and 40. The Weekly and GALC groups had increased weight after PND30 and PND36, respectively, when compared with the IP group. The treatment also had significant effects on the maximum body weight achieved by each mouse group (F (5, 72) = 126.90, p < .0001); there was a significant increase in maximum weight for all treatment groups compared to Twi mice (p < .05–.001; Fig. 1B).
Assessment of Motor Function Using Observational and Automated Phenotyping Methods
Twitching is an early presenting symptom in twitcher mice, and comparison of twitching severity across mouse groups showed a significant group × time interaction (p = .0003; Fig. 1C). Pairwise comparisons of twitching severity indicated that the Twi mice had more severe twitching compared to the IP, Weekly, and GALC groups starting at PND18 and the ICV group after PND32. Although the Weekly and GALC groups achieved lower (i.e., less severe) scores for every time point compared to the IP group, only after PND38 did these groups differ significantly. Furthermore, the treatment type significantly affected the day of twitching onset (F (4, 59) = 61.80; p < .0001); the Twi group presented with twitching symptoms at PND18.70 ± 0.50, which was significantly lower than the onset of all treatment groups (p < .001). The IP, Weekly, and GALC groups had twitching onsets at PND29.88 ± 0.92, 34.25 ± 0.45, and 30.20 ± 0.96; the ICV group exhibited this symptom at PND24.57 ± 1.56, which was significantly earlier than all IP treated groups (p < .01–.001). Additionally, the twitching onset of the Weekly group was delayed compared to the mice receiving only a single IP injection (p < .05; Fig. 1C).
The wire hang (Fig. 1D), hind stride length (Fig. 1E), and rearing abilities (Fig. 1F) were tested for each mouse group. For the wire hang study, there was a significant group × time interaction (p = .0004), and no differences were found between any mouse group until PND24, at which time the Twi mice had significantly less ability to hang on the wire compared to all treated mice (p < .01). The Weekly and IP groups had improved wire hang ability compared to the Twi mice at all time points between PND26 and PND38. Compared to the Twi mice, the GALC group also had improved hanging ability for all time points, except PND28 and PND38. In contrast, the ICV mice showed no wire hang improvement after PND26. Additionally, the ICV group had decreased hanging ability compared to the IP, Weekly, and GALC groups at PND34–38 (Fig. 1D). The group × time interaction was also significant for the hind stride study (p < .0001). The hind stride of the WT was significantly greater than the Twi and IP-treated twitcher mice starting at PND28, and it was not until PND32 that the ICV group differed from the WT mice. Interestingly, the Weekly and GALC groups had similar hind stride lengths compared to the WT group until PND38. The Twi mice had shorter hind stride lengths compared to the GALC group at PND24–38 and the Weekly group at PND34. Furthermore, the GALC and Weekly groups were shown to have longer hind stride lengths compared to the IP group at PND24 and PND34, respectively.
Twi animals demonstrated deficient locomotor activity during the test period, and this deficit was rescued by IP, Weekly, and GALC treatments. Specifically, the Twi mice had significantly fewer rearing (protected and unprotected) versus WT (p < .001) at PND23–29. The Weekly and GALC-treated mice showed improved vertical rearing compared to Twi mice (p < .05; Fig. 1F). As shown by the representative track visualizations (Fig. 2A), the Twi mice showed marked thigmotaxis for all weeks and minimal movement in the arena after PND30. In contrast, all IP-treated mice showed striking differences from both the ICV and Twi mouse groups. The improvements seen in the track densities were further supported by improvements in the total distance traveled (F (5, 60) = 2.76, p < .026), average velocity (F (5, 60) = 2.76, p < .02), and time spent in the inner zone (F (5, 60) = 6.21, p < .0001) at PND23–29 in the Weekly and GALC mouse groups (Fig. 2B, 2C). Comparing angular velocity over 3 weeks for all mouse groups, significant differences in group × time interaction (F (10, 144) = 2.41, p = .01) were found with all groups compared to WT mice at PND30–35 (Fig. 2D; p < .01–.001). The Twi group was the only twitcher cohort that had increased angular velocity compared to the WT mice at PND23–29 (p < .001), and both the GALC and IP treatments significantly corrected this behavior at PND30–35 (p < .01), although not to the baseline level.
Histological Analysis of the Sciatic Nerve
The H&E sciatic nerve images were analyzed with Fiji/Image J software to quantify the number of total cells (i.e., cells of area ≥5 μm2) and the number of large cells (i.e., cells of area ≥40 μm2) per field at 400× magnification . Treatment had a significant effect on both the total number of cells (F (5, 152) = 63.57, p < .0001) and large cells (F (5, 152) = 44.26, p < .0001) in the sciatic nerve. These large cells (Fig. 3A, white arrow) were positive for F4/80, a macrophage surface marker (data not shown). Compared to the Twi group, the IP, Weekly, and GALC treatments led to a significant (p < .001) decrease in total cell number (Fig. 3B). Additionally, all IP-treated groups had a significant (p < .001) decrease in peripheral globoid cell formation (Fig. 3C). The sciatic nerves from the ICV mice had globoid cell formation and inflammatory infiltration levels that were not different from the Twi group. Interestingly, only the GALC mice had infiltrating and globoid cell numbers that were corrected to WT levels. Taken together, these data showed significant improvements in sciatic nerve pathophysiology for all three IP-treated groups compared to the ICV treated and control Twi mice (Fig. 3A--3C).
TUNEL Staining of Sciatic Nerve for Detection of Apoptosis
The presence of darkly stained nuclei in the Twi and ICV H&E stained sciatic nerves indicated that apoptotic or necrotic cells were likely present in these samples. To investigate this further, unstained sciatic nerves were tested for apoptosis or necrosis activity using the TUNEL assay. All sciatic nerves, regardless of mouse group, displayed similar positive staining when exposed to DNase treatment as a positive control (Fig. 4A). The percentage of apoptotic cells based on the total number of DAPI+ cells was quantified using Image J for all experimental samples. There was no treatment effect on the percentage of TUNEL+ cells (F (5, 29) = 2.19, p = .07). However, by visual inspection, the experimental WT and IP-treated samples showed minimal TUNEL activity in any section (Fig. 4A, 4B). Conversely, the Twi showed DNA fragmentation in every section tested, as evidenced by the presence of 17.4% ± 8.4% apoptotic or necrotic cells detected by the TUNEL assay per field (Fig. 4B).
GALC Protein Levels and Enzymatic Activity in the Brain
The presence of GALC protein was determined for all treatment groups using Western blotting (Fig. 5A). Each lane in Figure 5A represents the brain lysates of three mice pooled together, and the results showed GALC-specific bands at 50 and 30 kDa for the WT, GALC, and Weekly mouse groups. Comparisons of the actin-normalized 30 kDa band intensities for GALC and Weekly groups showed 1.8- and 1.3-fold increases in GALC protein levels, compared to WT levels (Fig. 5B). The normalized bands for the ICV and IP groups were 21% and 10% the intensity of the 30 kDa WT band.
A radioactive GalCer assay was performed to assess the activity of the GALC enzyme (Fig. 5C). This assay involved incubating a 3H-GalCer substrate in the presence of brain lysate, and any detected radioactivity represented the presence of cleaved substrate and, thus, active GALC enzyme. For each treatment group, brain lysates of three to five mice were tested, and it was found that the GALC and Weekly groups had GALC activity that was 4.4% and 2.4% of the WT activity. The ICV, IP, and Twi groups had negligible enzymatic activities, with 0.5%, 0.0%, and 0.0% of WT activity, respectively.
To confirm that the transduced and untransduced eGFPTgBMSCs produce functional GALC enzyme, three experiments were performed. First, the cells were expanded in culture and stained with a modified β-galactosidase assay . The GALC-BMSCs had functional β-galactosidase enzyme capable of cleaving the X-gal substrate in more than 90% of the cells (Fig. 5D); as expected, the staining of the untransduced BMSCs was of a lesser intensity and in fewer cells. Second, the cells were lysed and immediately incubated with 4Mu-glycosides. After normalization of the β-gal levels to both β-hex and α-gal, it was found that the GALC-BMSCs had β-gal activity that was 2.4 times greater than that of the untransduced cells (Fig. 5E). Furthermore, the enzymatic activities were calculated at either pH 4.0 or pH 7.0 (Fig. 5F), and the ratio of activities of the GALC-BMSCs at 4.0 versus 7.0 was 1.8-fold greater than the BMSCs. Lastly, the radioactive 3H-GalCer assay was performed on eGFPTgBMSCs and GALC-transduced eGFPTgBMSCs to determine the activity of the GALC enzyme present in these cell types (Fig. 5G). The GALC-BMSCs had a GALC activity of 2.9 nmol of cleaved substrate per hour per 1,000,000 cells, whereas the untransduced BMSCs cleaved the substrate at 1.05 nmol/hour per 1,000,000 cells.
Detection of Green Fluorescence Protein in Stem Cell Injected Mice by Western Blotting and Immunofluorescence
To determine whether eGFP was present in brain lysates at the time of euthanasia (PND33–48), Western blotting was performed using a GFP-specific antibody for all treatment groups (Fig. 6A). A specific GFP band at 27 kDa was detected for the ICV and WT mice that received eGFPTgBMSCs administered directly to the brain. None of the IP treatment groups showed evidence of GFP signal in their respective lanes, indicating an absence of the cells at this terminal time point. Liver and kidney lysates were also tested using Western blotting with the same GFP-specific antibody. No treatment group had detectable GFP protein in the kidney lysates, and the Weekly group was the only treatment group to have a GFP-specific band in the liver samples (data not shown).
GFP+ BMSCs were identified in frozen sections of the ICV mouse brains by IHC using a GFP antibody. These cells were found in the hind brain, periventricular region, and forebrain; however, there was no evidence of colocalization when stained for NeuN, GFAP, or S-100 (Fig. 6B--6D).
To determine whether an antibody response formed in the mice against the GFP+ BMSCs after multiple injections, the sera of the Twi, IP, and Weekly mice were incubated as the primary antibody solution for Western blotting of purified GFP (Fig. 6E). When compared with a control lane, which had been incubated with a known antibody against GFP, the Weekly treatment group had a similar band (white arrows), although of a lesser intensity.
To develop a safer, more effective therapy for GLD patients, this study investigates the therapeutic effects of central and peripheral MSC injections in the twitcher mouse. Affected mice were grouped according to the injection conditions (e.g., administration route, injection frequency, and cell transduction status) used to administer the BMSCs, and all treated cohorts were compared to Twi mice for assessment of therapeutic effectiveness. The effects of (a) increased injection frequency or (b) transduced BMSCs overexpressing the GALC enzyme have not, to our knowledge, been assessed as potential treatments. For this, stem cell persistence, brain and sciatic nerve pathology, levels of GALC, and GLD-related symptoms were assessed for all mice to compare various transplantation conditions. This study demonstrated that twitcher mice receiving weekly IP injections of BMSCs or a single IP injection of GALC-transduced BMSCs have significant improvements in lifespan, twitching onset and severity, body weight, motor function (e.g., wire hang ability and hind stride length), general motor activity (e.g., velocity and distance traveled), and sciatic nerve histopathology (e.g., decreased globoid cell formation and apoptotic cell death) versus Twi mice and/or ICV mice. Such results corroborate recent studies demonstrating the therapeutic potential of peripherally injected MSCs in LSDs [41, 51–53] and indicate a promising future for MSC therapy in the GLD patient.
Previous studies in our laboratory by Ripoll et al. , Wicks et al. , and Scruggs et al.  have shown modest improvements in the twitcher phenotype and/or lifespan after suboptimal doses of BMSCs were injected ICV, intrastriatally, or IP, respectively. In an effort to further improve the outcomes of MSC therapy, the number of cells administered in this study was increased fivefold, yet all treatment groups received doses (e.g., ∼1 × 107 cells per kilogram) that are comparable to those currently used in MSC-related clinical trials. Increasing the ICV administered cell number from 40,000  to 200,000 total BMSCs achieved greater persistence and distribution of the stem cells in the twitcher brain, as was evident by the presence of GFP+ cells in all ICV injected mice even at the time of euthanasia. Furthermore, the ICV mice had a significant increase in median lifespan compared to mice receiving fewer cells . However, the increased cell persistence in the brain did not result in increased GALC activity or transdifferentiation of the BMSCs to neurons, astrocytes, or glial cells; such results are consistent with recent findings that BMSCs provide neuroprotection through anti-inflammatory and immunomodulatory actions, not cell or enzyme replacement [38, 43].
The majority of GLD symptoms in twitcher mice result from severe peripheral nervous system (PNS) inflammation and demyelination [12, 55, 56]. Twitcher mice display symptoms related to widespread peripheral neuropathy, including severe muscle wasting, hind limb weakness, gait abnormalities, spatial abnormalities, and decreased rearing [13, 14]. Similarly, GLD patients often show decreased reflexes and limb strength, muscle atrophy, early coordination disturbance, poor head control, spasticity, and slowed motor nerve conduction velocity [7, 8, 57, 58]. Such symptoms result from the extensive myelin sheath degeneration, axonal death, and inflammatory infiltration (e.g., globoid cells, eosinophils, and mast cells) present in peripheral nerves. Striking differences were detected between the sciatic nerves of the ICV and IP-treated mice. These results suggest that MSC do not exert an anti-inflammatory effect on peripheral nerves when administered directly into the brain. Globoid cell and infiltrating cells in the PNS may be a main contributor to the poor motor function of the ICV mice. Conversely, peripheral MSC injections have been shown to be both anti-inflammatory and antiapoptotic in the PNS, which is supported by the motor function results. In support of this conclusion, the GALC sciatic nerves showed no histological evidence of inflammation. The most significant motor function improvements were seen after IP administration of BMSCs or GALC-BMSCs, yet the results showed that the IP-injected cells did not home to the CNS. Recent studies have tracked MSCs after systemic administration in neurodegenerative and autoimmune mouse models, and all have demonstrated similar beneficial effects after MSC injection due to the cells locating to secondary lymphatic organs and mediating immune responses [59–63]. Therefore, it is likely that the BMSCs and GALC-BMSCs of this study were capable of exerting potent systemic effects after IP administration by homing to sites, such as lymph nodes, and acting as immune effectors.
Many LSDs with peripheral manifestations have been successfully treated with ERT [33, 64], and recent studies using the twitcher mouse show a beneficial, albeit transient, effect on the presentation and neuropathology of the affected mice . In an effort to treat twitcher mice by providing a more stable supply of GALC enzyme, BMSCs that were engineered to overexpress GALC were delivered or the frequency of delivery of naïve BMSC was increased. Despite increased GALC protein in the brain and modest improvement in lifespan, enzyme activity assays showed no detectable GALC activity in the brains of the treated mice. These findings suggest that the GALC enzyme, over time, is unable to cleave its substrates that have accumulated in the brain at the time of euthanasia; this result was consistent with the observation that all treated mice died before PND50. The combination of increased GALC protein, active GALC enzyme in the BMSCs, and absent GALC activity in the transplanted brain may indicate that the detected enzyme, although initially active, is not transported by cross-correction mechanisms to inside the affected brain cells . Such faulty localization has been documented previously , and this may cause the enzyme to become partially degraded, incompletely processed, incorrectly folded, and/or irreversibly bound to an inhibitor. Our laboratory has ongoing experiments aimed to understand this phenomenon for improving cross-correction. Based on the improved motor function and pathology in the GALC and Weekly groups, it is expected that weekly administration of GALC-transduced BMSCs would increase the amount of GALC protein in the PNS and CNS more than either group alone. This novel treatment group will be tested in our laboratory upon production of adequate numbers of GALC-transduced BMSCs, and based on these results we expect that this treatment will improve the cross-correction of GALC, lifespan, and quality of life for twitcher mice.
BMSCs are exceptional candidates for restoration of damaged tissue [35, 66] and have the potential, if injected under optimized conditions, to significantly alleviate peripheral inflammation and apoptotic cell death in the twitcher mouse. Specifically, BMSCs secrete a variety of cytokines and factors that have paracrine activities , promote differentiation of the host's stem cells, and modulate immune reactions [35, 37, 66–69]. Although these beneficial effects are clearly demonstrated in this study using peripheral MSC transplantation, all treated mice eventually died in a moribund state after a rapid decline in health. Based on the GALC activity results, it is likely that the psychosine levels in the brain were not decreased, leading to increased oxidative stress and apoptotic death of myelin producing cells and other brain cells [6, 65]. Taken together, these results suggest a promising future for peripherally administered MSCs if combined with a therapy that increases GALC levels in the brain.
Transplantation of BMSCs requires extensive optimization in animal models before being implemented in clinical trials as a novel therapy for neurodegenerative diseases. In order to understand the mechanism of action of these cells and to improve their therapeutic efficacy in the twitcher mouse model of GLD, this study aimed to increase MSC distribution, persistence, and anti-inflammatory effects through optimization of the cell type, injection frequency, and administration route. Significant improvements in lifespan, motor function, and GLD-related symptoms after IP transplantation of MSCs were associated with a significant reduction in inflammation in the PNS. The results of this study indicate that repeated injections of BMSCs or GALC-BMSCs in the periphery could be an effective adjunct therapy for HSCT to decrease risks of GVHD and improve PNS outcomes of patients affected with GLD.
We thank Siddharth Gaikwad, Jeremy Green, Evan Kyzar, and the Neurophenotyping Core at Tulane University School of Medicine for their instruction, advice, and suggestions regarding the video recording studies. We also thank the veterinary and vivarial staff of the TUHSC Animal Facility for the daily care of the mice. We also thank Michelle E. Scarritt for help with the immunofluorescence experiments, Kathleen MP Imhof for assistance with the motor function tests, Krystal Brown for help with the deconvolution microscope, Dr. Xavier Alvarez of the Tulane National Primate Research Center (TNPRC) for assistance in analyzing the histological samples using the confocal microscope, Alan Tucker for performing flow cytometry for characterization of the MSC populations, and Dina Gaupp and Claire Llamas in the Tulane Histology Core for performance of the histological staining for this study. The multiplex plates were run by the staff at the Pathogen Detection and Quantification Core (PDQC) at the TNPRC under Grant number RR00164. The GALC activity assay was performed with the assistance of Dr. Yu-Teh Li of the Tulane University Biochemistry Department. This research was supported by Grant R21-NS059665 from the National Institutes of Neurological Disorders and Stroke (NIH) and by Tulane University.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.