Repeated Mesenchymal Stromal Cell Treatment Sustainably Alleviates Machado-Joseph Disease

Abstract

Machado-Joseph disease (MJD) or spinocerebellar ataxia type 3, the most common dominant spinocerebellar ataxia (SCA) worldwide, is caused by over-repetition of a CAG repeat in the ATXN3/MJD1 gene, which translates into a polyglutamine tract within the ataxin-3 protein. There is no treatment for this fatal disorder. Despite evidence of the safety and efficacy of mesenchymal stromal cells (MSCs) in delaying SCA disease progression in exploratory clinical trials, unanticipated regression of patients to the status prior to treatment makes the investigation of causes and solutions urgent and imperative. In the present study, we compared the efficacy of a single intracranial injection with repeated systemic MSC administration in alleviating the MJD phenotype of two strongly severe genetic rodent models. We found that a single MSC transplantation only produces transient effects, whereas periodic administration promotes sustained motor behavior and neuropathology alleviation, suggesting that MSC therapies should be re-designed to get sustained beneficial results in clinical practice. Furthermore, MSC promoted neuroprotection, increased the levels of GABA and glutamate, and decreased the levels of Myo-inositol, which correlated with motor improvements, indicating that these metabolites may serve as valid neurospectroscopic biomarkers of disease and treatment. This study makes important contributions to the design of new clinical approaches for MJD and other SCAs/polyglutamine disorders.

Oliveira Miranda et al. show that repeated mesenchymal stromal cell (MSC) treatment, in opposition to a single treatment, sustainably ameliorates Machado-Joseph disease motor behavior and neuropathology, explaining recent clinical trial findings. Moreover, magnetic resonance spectroscopy biomarkers to measure MSC efficacy are proposed. Overall, MSCs are promising for ataxia therapy when repeated administration is employed.

Introduction

Machado-Joseph disease (MJD), also known as spinocerebellar ataxia (SCA) type 3, is a neurodegenerative disorder caused by the expansion of a CAG trinucleotide in chromosome 14q32.1 of the ATXN3/MJD1 gene, which translates into a polyglutamine (polyQ) stretch within the mutated protein ataxin-3.1, 2 Under normal conditions, ataxin-3 has 10–51 glutamines, the mutated protein carries 55 or more consecutive glutamines.³ Physiologically, ataxin-3 is a deubiquitinating enzyme that participates in the quality control of proteins within the proteasome.⁴ Although the precise mechanism behind mutant ataxin-3 toxicity is still largely unknown, it is well established that the mutant protein form gains a toxic function in neuronal cells, triggering a degenerative process and accumulating in the form of neuronal intranuclear inclusions.⁵ Neuropathologically, multiple regions of the brain can be affected, including the cerebellum and striatum.6, 7 The clinical hallmark of MJD is progressive gait and limb ataxia, resulting in severe clinical presentation and leading to premature death.

Recent studies from our and other groups have shown that molecular strategies such as gene silencing.8, 9 autophagy activation,10, 11 calpain blockage,¹² caloric restriction,¹³ or neural stem cell transplantation¹⁴ have promising therapeutic potential in MJD models and have enormously enlarged our knowledge of disease mechanisms. Despite that, no therapy that allows modification of the progression of this fatal disease is available to patients so far.

Mesenchymal stromal cells (MSCs) are promising therapeutic tools for neurodegenerative disorders15, 16 because they are easy to isolate and grow to clinical grades without major ethical issues, are considered non-tumorigenic,¹⁷ and can be used in allogeneic transplantation when autologous transplants are not possible because of their low immunogenicity.¹⁸ Moreover, they have a strong paracrine effect because they can deliver neurotrophic factors, cytokines, exosomes, or mitochondria to neighboring cells, helping their recovery.19, 20, 21

The successful use of MSC in preclinical studies regarding neurodegenerative disorders has buttressed the expansion of clinical studies.²² In fact, phase I/II clinical trials, some including MJD patients, have already started and demonstrated that MSC are safe and may delay disease progression (https://www.clinicaltrials.gov/; ClinicalTrials.gov identifiers NCT01489267, NCT01958177, NCT01360164, NCT01649687, and NCT02540655). However, recent reports suggest that many patients regress to the status prior to treatment.23, 24, 25, 26 For maximum success in clinics, it is urgent to understand what has gone wrong regarding MSC therapies.

This therapeutic approach has, however, received limited attention at pre-clinical levels, and, as a result, very little is known regarding the MSC mechanism in SCAs.27, 28 Pre-clinical studies are thus imperative to achieve better outcomes in clinical trials. In the present study, we compared the efficacy of a single intracranial injection with several systemic administrations of MSCs in alleviating MJD motor behavior defects and the neuropathology of genetic mouse models of MJD and compared the in vivo expression of cerebellar metabolites in MSC-treated and non-treated (NT) MJD mice, aiming to clarify the mechanism of action of MSC. The two mouse MJD models used in the present study presented strongly severe symptoms of the disease and were treated post-symptomatically.

Our data indicate the following: MSCs can successfully improve the MJD behavioral phenotype and neuropathology, the proof of concept of a valuable effect in the MJD context, but a single local or intracerebroventricular (i.c.v.) administration of MSC does not produce effects sustained over time; MSCs start to disappear over a short period of time after in vivo administration; multiple systemic administrations can produce effective phenotypic alleviation; and MSCs exert neuroprotective mechanisms that involve the diminishment of inflammation and preservation of neuronal populations, culminating in increased synaptic activity and/or plasticity.

Importantly, the effects observed in the present study are certainly sufficient to delay disease progression in MJD patients because the mouse models we used in the present study were strongly severe, mimicking late stages of the disease in humans. Moreover, a repeated non-invasive therapy may be easily implemented in the clinic, contrary to craniotomy procedures.

Overall, our data suggest that this promising cellular therapy must consist of repeated treatments to promote sustained effects in MJD patients or patients with other SCAs and suggest new biomarkers that can be used not only to follow up on the effects of MSC treatment but probably also for evaluating disease progression.

Results

MSCs Isolated from the Bone Marrow of Wild-Type Mice Meet the Criterion Established for MSCs after Elimination of CD45⁺ Cells by Cell Sorting

Given the heterogeneity of mouse MSC cultures and possible contamination with hematopoietic cells after initial adherence and growth until sub-confluency, we performed MSC sorting for CD45, excluding all positive cells for this hematopoietic marker. MSCs were later expanded until passage 9 and frozen for subsequent analysis (Figure 1A). MSCs were then characterized at passage 15 because all the assays of the present study were performed at passages 15–17 (Figures 1B–1E). In culture, MSCs exhibited a fibroblast-like morphology, and, after induction with the appropriate differentiation medium, could differentiate into adipocytes, osteoblasts, and chondrocytes (Figure 1B). Flow cytometry analysis revealed (Figures 1C and 1D) that MSCs were negative for the hematopoietic markers CD11b and CD45 and highly positive (between 95%–100%) for the mesenchymal markers CD73, CD105, SCA-1, CD29, and CD106 (Figure 1D), as expected. The present isolation and selection method can thus produce a population of MSCs that meets the criterion defined by the International Society for Cell Therapy for human MSCs.³³ Finally, immunocytochemical analysis revealed that MSCs expressed both neuronal and glial markers in vitro, including Nestin, β-tubulin III, MAP2, GFAP, and IBA-1 (Figure 1E), as described previously,34, 35 as well as Nanog and SSEA-4, two markers for pluripotency, at very low levels (Figure 1E).

Figure 1.

Mesenchymal Stem Cells Characterization after Ex Vivo Expansion: Validation of the Isolation Method

(A) Schematic representation of MSC isolation and sorting of CD45⁺ cells before scaling up. (B) Bright-field photomicrograph of MSCs in passage 15 (left) and fluorescence photomicrographs of MSCs after differentiation into FABP4-positive (adipocytes), osteopontin-positive (osteoblasts), and collagen II-positive (chondrocytes) after 19–21 days in contact with the respective induction medium (following). (C and D) Immunophenotype characterization by flow cytometry. Analysis was performed on gated cells from a total of 50,000 events based on forward and side light-scatter parameters (forward scatter [FSC] versus side scatter [SSC]). (C) Double staining acquisitions were performed as illustrated. The anti-CD45 and anti-CD106 antibodies were fluorescein isothiocyanate (FITC) conjugates; anti-CD11b, anti-CD73, anti-CD105, and anti-CD29 were phycoerythrin (PE) conjugates; the anti-SCA-1 antibody was a PE-cyanin 7 (PE-Cy7) conjugate. (D) Log fluorescence was collected and displayed as single-parameter histograms. Solid-line histograms represent isotype control-labeled cells, and filled histograms represent cells labeled with antibodies specific for the indicated cell surface antigens. (E) MSC expression of neuronal and pluripotent markers in vitro by immunocytochemistry. Triple staining was performed as shown. β-Tubulin III, MAP2, and Nanog are shown in green; GFAP, IBA-1, Nestin, and SSEA4 are shown in red; DAPI is shown in blue. FABP4, fatty acid-binding protein 4; GFAP, glial fibrillary acidic protein; IBA-1, ionized calcium-binding adaptor molecule 1; MAP-2, microtubule-associated protein 2; SCA-1, stem cell antigen 1; SSEA-4, stage-specific embryonic antigen-4.

MSCs Alleviate MJD Motor Behavior Deficits and Neuropathology

We first investigated whether MSCs would promote the functional and neuropathological recovery of a transgenic MJD model with severe balance and motor coordination deficits.³⁶ Lentiviral GFP-labeled MSCs were stereotaxically injected in the parenchyma of the cerebellum of Tg-ATXN3-69Q (MJD) mice 3–5 weeks of age. NT MJD littermates were used as controls. Wild-type (WT) mice treated and not treated with MSCs were also included to explore MSC safety. All mice were phenotypically evaluated as shown in the experimental design diagram (Figure 2A) and sacrificed 12 weeks post-transplantation for evaluation of neuropathology and transplant engraftment.

Figure 2.

A Single Injection of MSCs in the Parenchyma of the Cerebellum Promotes Neuropathological Mitigation and Transient Phenotypic Alleviation in Tg-ATXN3-69Q Mice

(A) Timeline of the experimental procedure after transplantation of MSCs in the cerebellar parenchyma of Tg-ATXN3-69Q or WT mice. (B–D) The motor performance of non-treated (WT) or MSC-treated wild-type mice (WT+MSC) and non-treated (NT-MJD) or MSC-treated (MJD+MSC) Tg-ATXN3-69Q mice was evaluated by rotarod (B) and swimming (C), and gait parameters were analyzed by footprint (n = 5 NT-MJD mice versus n = 5 MJD+MSC mice for all former experiments) (D); differences between NT-MJD and MJD+MSC mice are shown. (E) Schematic representation of the engraftment sites of MSCs essentially in lobule II of the cerebellum of MJD mice; a confocal representative microphotograph of MSCs (GFP, green; DAPI, blue) in lobule II is shown on the right. (F) Illustrative microphotographs of cresyl violet staining of cerebellar sections of NT-MJD and MJD+MSC mice and graphic representation of the measurement of molecular layer thickness between lobules II and III of WT+MSC, NT-MJD, or MJD+MSC mice (n = 6 WT+MSC mice, n = 4 NT-MJD mice, n = 4 MJD+MSC mice). (G–I) Spearman correlations between molecular layer thickness and motor assessments were performed 4 weeks after treatment: rotarod (G), swimming (H), and hind base (I). In (F), bars represent the mean ± SE. In (B)–(D), boxplots represent median, 25%–75% interquartile range, non-outlier range, and outliers; repeated measures ANOVAs followed by post hoc Bonferroni-corrected tests. *p < 0.05, **p < 0.01, and ***p < 0.001.

Comparisons between NT (WT) and MSC-treated WT mice (WT+MSC) demonstrated that, except for an improvement in the footprint test in MSC-treated mice 8 weeks after transplantation (F_2,8 = 4.89, p = 0.04), all other comparisons were not statistically significant (all F_2,8 < 3.11, p > 0.10; Figures 2B–2D). This allowed us to conclude that stereotaxic transplantation of MSCs is not harmful in this experimental paradigm.

Regarding comparisons between NT and MSC-treated MJD mice, 2 weeks after transplantation, MSCs improved the motor coordination and balance of treated transgenic mice; these presented significantly increased latencies to fall on a rotarod (a 4.6-fold increase relative to NT MJD) 4 weeks after transplantation (F_2,8 = 4.95, p = 0.04; Figure 2B). However, 8 weeks post-treatment, this positive effect disappeared, and the treated mice showed even worse performance than NT controls either at 8 weeks (Figure 2B) or later time points (data not shown). In the swimming test, 2 weeks post-treatment, MJD mice showed more motor difficulties in executing the test, but 4 weeks after MSC administration, mice took less time to swim and reach the platform of the swimming pool compared with NT MJD mice (F_2,8 = 10.15, p = 0.001; Figure 2C). However, in the swimming test, MSC-treated MJD mice also slightly worsened 8 weeks post-treatment (Figure 2C). In the footprint test, we measured the hind base width; i.e., the distance between the left and the right hind footprints. In MJD mice, this distance is larger relative to WT mice because of their balance defects. Four weeks after MSC administration, treated MJD mice showed a decrease in the hind base width compared with NT MJD mice (F_2,4 = 6.90, p = 0.05; Figure 2D), but not at later time points (data not shown), demonstrating again that MSCs promote a transient improvement in mice gait. In summary, an overall phenotypic alleviation was observed 4 weeks after transplantation but not at later time points.

MSC engraftment was evaluated through GFP fluorescence on sagittal brain sections of transplanted MJD mice 12 weeks after treatment. MSCs were detected near lobules II and III of the cerebellum (mostly near lobule II; Figure 2E), showing that MSCs did integrate into the cerebellar tissue and survive for this time period. However, the GFP signal was not very strong, indicating that probably only a few cells lasted that long.

Because this mouse model preferentially expresses truncated ataxin-3 with an abnormally expanded polyQ tract in cerebellar Purkinje cells,³¹ and because these neurons account for a large fraction of the molecular layer thickness, we next determined the thickness of the molecular layer between lobules II and III, where MSCs were mainly found. MSCs mediated preservation of the molecular layer of the cerebellum of MJD mice in comparison with NT mice between lobules II and III (F_2,11 = 54.33, p < 0.001; Figure 2F) and between lobules III and IV (F_2,11 = 46.28, p < 0.001; Figure S1A).

Finally, we investigated whether behavioral improvement and alleviation of cerebellar neuropathology significantly correlated with each other. The correlation between the molecular layer thickness between lobules II and III (assessed 12 weeks after transplantation) and the behavior performances 4 weeks post-transplantation (matching the best phenotypical performances) were analyzed in NT, MSC-treated MJD, and WT mice. Rotarod performance was positively correlated with the thickness of the molecular layer (Spearman correlation coefficient [r_s] = 0.60, p = 0.01; Figure 2G), whereas swimming and hind base were negatively correlated with this cerebellar measurement (r_s = −0.53, p = 0.04 and r_s = −0.56, p = 0.02; Figures 2H and 2I), as expected. Significant correlations were also observed between the molecular layer thickness determined between lobules III and IV of the cerebellum and all phenotypic parameters (rotarod: r_s = 0.68, p = 0.004; swimming: r_s = −0.55, p = 0.03; hind base: r_s = −0.57, p = 0.02; Figures S1B–S1D).

Interestingly, separated populations of mice could be individualized in the plots for each group (NT-MJD in red, n = 5; MJD+MSC in green, n = 5; WT+MSC in blue, n = 6), where MSC-treated mice were more closely positioned relative to WT than NT MJD mice (Figures 2G–2I; Figures S1B–S1D). This result may be an indication that MSCs can promote phenotypic improvements in MJD mice through preservation of Purkinje cells and the neuronal network, in accordance with a report regarding damage of Purkinje cells arborization in this MJD mouse model.³⁷ In conclusion, our data indicate that local implantation of MSCs alleviates the motor behavior and neuropathological deficits of the Tg-ATXN3-69Q mouse model. However, relief of the MJD motor phenotype is transient.

MSCs Alleviate MJD through a Paracrine Effect

To investigate whether the MJD alleviation results from a paracrine effect, we transplanted MSCs into the right lateral ventricle of the brain and evaluated behavioral performance and neuropathology (Figure 3A). It was expected that, if the MSC-mediated neuroprotection resulted from a paracrine effect, then the factors released from MSCs into the cerebrospinal fluid (CSF) would circulate throughout the entire nervous system, reaching the whole cerebellar region and mediating similar, if not better, results compared with the previously described transplantation into the brain parenchyma.

Figure 3.

A Single Injection of MSCs in the Lateral Ventricle of the Brain Promotes Equivalent Neuropathological and Transient Phenotypic Alleviation in Tg-ATXN3-69Q Mice

(A) Schematic representation of the experimental procedure of transplantation of MSCs in the right lateral ventricle of the brain in Tg-ATXN3-69Q mice. (B and C) Evaluation of the motor performance of NT-MJD or MJD+MSC Tg-ATXN3-69Q mice by rotarod (n = 13 NT-MJD mice versus n = 14 MJD+MSC mice ) (B) and pole (n = 14 NT-MJD mice versus n = 14 MJD+MSC mice) (C). (D) Microphotograph of a coronal brain section, showing the engraftment of MSCs in the walls of the lateral ventricle. (E) Illustrative microphotographs of cresyl violet staining of cerebellar sections of NT-MJD or MJD+MSC Tg-ATXN3-69Q mice and graphic representation of the measurement of the volume of lobule X (n = 5 NT-MJD mice versus n = 6 MJD+MSC mice). (F) Illustrative fluorescent microphotographs of lobule X stained with PCP4 of NT-MJD and MJD+MSC Tg-ATXN3-69Q mice and quantification of the number of Purkinje cells in lobules VI and X (n = 6 NT-MJD mice versus n = 5 MJD+MSC mice). (G) Representative membrane showing western blot of mutant ataxin-3 (both high-molecular-weight species and the soluble protein form) and β-tubulin. Quantification of mutant ataxin-3 bands corrected for individual β-tubulin (endogenous protein) signal intensity is given. PCP4, Purkinje cell protein 4. Bars represent the mean ± SE. Repeated-measures ANOVA followed by post hoc Bonferroni-corrected tests. *p < 0.05.

Again, a significant but transient phenotypic alleviation was observed in the present case (Figures 3B and 3C). In the rotarod test, MSC-treated MJD mice showed better performance 2 weeks after transplantation. However, 8 weeks after treatment, the performance of MJD mice did not differ from NT littermates (F_1,26 = 7.73, p = 0.01; Figure 3B). In the pole test, improvements were also observed only 2 weeks post-treatment (F_1,26 = 4.67, p = 0.04; Figure 3C). All phenotypic evaluations thus indicate that factors secreted by MSCs implanted in the lateral ventricle are sufficient to alleviate motor deficits. Nevertheless, these effects were temporary, disappearing 4–8 weeks after the treatment, corroborating the results found upon MSC transplantation in the parenchyma of the cerebellum.

MSC engraftment was also evaluated at the end of the experiment. We found agglomerated GFP⁺ MSCs either in the lumen of the ventricles (attached to ventricle walls) or in the parenchyma close to the ventricle walls in 70% of transplanted mice (n = 7 mice of 10) (Figure 3D). In the remaining 30% of transplanted mice (n = 3), no MSCs were detected in sagittal sections of the brain, probably indicating that, 10 weeks post-transplantation, they were no longer viable.

Neuropathology evaluation revealed that the volume of lobule X was larger in MSC-treated MJD mice compared with NT controls, indicating that MSC-mediated preservation of cerebellar neurons (Student’s t test [t₁₀] = 2.27, p = 0.04; Figure 3E). Moreover, MSC-treated MJD mice exhibited a higher number of Purkinje cell protein 4 (PCP4)-positive Purkinje cells than NT controls in lobules VI and X, highlighting the preservation of the most affected neuronal cells in this model³¹ (t₁₀ = 2.58, p = 0.03 and t₁₀ = 2.67, p = 0.03, respectively; Figure 3F). Because formation of mutant ataxin-3 neuronal inclusions is one of the hallmarks of MJD,⁵ the whole cerebellum was analyzed by western blot for determination of ataxin-3 levels. A tendency to decrease was observed for the soluble form of the protein in the cerebellum of Tg-ATXN3-69Q mice treated with MSCs, whereas no significant differences in the levels of the high-molecular-weight species of ataxin-3 protein were observed (t₁₀ = 1.99, p = 0.08; Figure 3G).

In summary, MSCs transplanted i.c.v. mediated partial recovery of cerebellar morphology, supporting the survival of Purkinje cells and improving the motor functions of ataxic mice, supporting a paracrine mechanism of action for MSCs. However, phenotype alleviation remained transitory.

MSCs Alleviate Neuropathology in an LV-Based Model

To further investigate the paracrine actions mediated by MSCs, we took advantage of structural magnetic resonance and a lentivirus (LV)-based model of MJD. LV-based models obtained by stereotaxic injection of lentiviral vectors encoding mutant ataxin-3 in the rodent brain provide robust in vivo models with neuronal degeneration in specific regions of the brain that are particularly adequate for quantitative analysis.⁶

Lentiviral vectors encoding full-length mutant ataxin-3 with 72 glutamines (LV-ATXN3-72Q) were injected into the right side of the striatum, followed by transplantation of MSCs or vehicle into the ipsilateral ventricle 2 weeks later. MRI assessments were performed 1 and 4 weeks after transplantation.

Eight weeks after MSC delivery, neuropathology was assessed in coronal brain sections (Figure 4A). In accordance with the previously described experience with the transgenic MJD model (Tg-ATXN3-69Q), there were no differences in the number of ataxin-3 inclusions in this model 8 weeks after MSC treatment (t₇ = 2.08, p = 0.08; Figure 4B). Nonetheless, preservation of striatal neurons was observed, here represented by a reduction in the volume of the region depleted of immunostaining for the dopamine-regulated neuronal protein dopamine and cyclic AMP-regulated phosphoprotein 32 (DARPP-32) (t₇ = 2.53, p = 0.04; Figure 4C). Moreover, longitudinal MRI analysis revealed that, in the first week after transplantation, the volume of the striatum decreased less in MSC-treated mice (MJD+MSC) than in NT MJD controls (NT-MJD); however, statistical differences were no longer observed in the variation of the volume of the striatum 4 weeks after transplantation, suggesting that MSCs exerted a more robust neuroprotective effect right after treatment (t₉ = 2.64, p = 0.03; Figure 4D). This paradigm thus established that MSC neuroprotection is reproduced in this LV-based model.

Figure 4.

In the LV-ATXN3-72Q Model, Intracerebroventricular Transplantation of MSCs Promotes Early Physiological Alterations that Are Responsible for Neuropathological Alleviation

(A) Schematic representation of the experimental procedure of MSC injection in the right lateral ventricle 2 weeks after the injection of LV mutant ataxin-3 in the right side of the striatum. (B) Representative coronal brain photomicrographs illustrating anti-1H9 IHC and quantification of ataxin-3 aggregates (dark dots inside the delimitated region) in non-treated (NT-MJD, n = 4 males) or MSC-treated LV-ATXN3-72Q mice (MJD+MSC, n = 4 males) 8 weeks after transplantation. (C) Representative coronal brain sections immunolabeled with DARP32 and determination of the neuron-depleted area (delimited region) in non-treated (NT-MJD, n = 4 males) or MSC-treated LV-ATXN3-72Q mice (MJD+MSC, n = 4 males) 8 weeks after transplantation. (D) Representative MRI images showing the extension of the striatum (green area) and quantification of the difference in striatal volume between the time of transplantation and 1 week after treatment (1–0 weeks) and between 1 and 4 weeks after transplantation in non-treated (NT-MJD, n = 5 males) or MSC-treated LV-ATXN3-72Q mice (MJD+MSC, n = 5 males). In (B) and (C), bars represent the mean ± SE. In (D), boxplots represent median, 25%–75% interquartile range, non-outlier range, and outliers. Repeated-measures ANOVA followed by post hoc Bonferroni-corrected tests; *p < 0.05.

MSCs Have a Short Lifetime after In Vivo Transplantation

Because the protective effect of MSC has been shown to be transient both in clinical trials after intrathecal or intravenous (i.v.) administration23, 24, 25 and in the present study in MJD mice, we investigated the lifetime of MSCs longitudinally. By MRI technology, we assessed MSCs after transplantation of cells pre-labeled with superparamagnetic iron oxide particles (SPIONs)³² (n = 5/group) (Figure 5A). In vivo image acquisition indicated that MSCs remained in the lateral ventricles upon i.c.v. injection (Figures 5B, 5D, and 5E). Nonetheless, the volume of the MSC graft was drastically reduced from 1 to 4 weeks after transplantation, showing an average reduction of 55% of the volume estimated from 1 to 4 weeks after transplantation (Figure 5C) (F_2,4 = 6.89, p = 0.05). From 4 to 7 weeks after transplantation, MSC total volume continued to decrease, although the drop rate was not significant (Figure 5C). Surprisingly, both astrocytes and microglia were found surrounding MSCs within the lateral ventricles (Figures 5D and 5E, respectively). This could be indicative of either unspecific recruitment of immune cells favored by the transplantation and by MSCs or an active immune reaction occurring against MSCs. This last hypothesis would be supportive of MSC disappearance throughout time but needs further confirmation.

Figure 5.

MSCs Remain in the Lateral Ventricles after Intracerebroventricular Implants and in the Brain Parenchyma after Intravenous Administration but Disappear in a Short Time

(A) Schematic representation of the experimental procedure of i.c.v. transplantation of MSCs in the lentivirus-induced MJD mouse model. MSCs were injected into the right lateral ventricle 2 weeks after the injection of LV mutant ataxin-3 in the right side of the striatum. (B) MRI in vivo coronal acquisitions of the brain of a transplanted mouse showing the presence of MSCs in the right lateral ventricle (dark dots) 1, 4, and 7 weeks after transplantation. For detection by the MRI equipment, MSCs were previously labeled with iron oxide nanoparticles (Rhodamine B, Molday ION). (C) Graphic representation of the total volume of MSCs, calculated from images acquired with MRI equipment, showing a significant decrease in MSC volume from 1 to 4 weeks after treatment (n = 5 MJD+MSC males). (D and E) Illustrative immunofluorescent microphotographs of MSCs surrounded by GFAP (D) and IBA-1 immuno-positive cells (E). (F) Representative images of the bioluminescent signal captured in IVIS equipment 30 min, 2hr 45 min, 2 hr, 48 hr, and 7 days after luciferase-labeled MSC administration in the tail vein of 6-week-old Tg-ATXN3-69Q mice (n = 2), confirming the MSC signal in the brain (top) and in the lung (bottom). (G) Quantification of signal intensity (photons per second) in the brains of mice. (H) MSC engrafting the parenchyma of the cerebellum (granular layer of lobule III, left) and the interlobular region (between lobules IV/V and VI/VII, right) without co-localizing with CD31, indicating MSC localization out of the blood vessels. IBA-1, ionized calcium-binding adaptor molecule 1; CD31, cluster of differentiation 31. Bars represent the mean ± SE. Repeated measures ANOVA followed by post hoc Bonferroni-corrected tests. *p < 0.05.

Furthermore, to investigate the extent of MSC penetration and/or engraftment into the brain following systemic administration, we labeled MSCs with lentiviral vectors expressing luciferase (Figures S2A and S2B). Mice (n = 2- to 6-week-old Tg-ATXN3-69Q mice) were injected i.v. with MSCs (8 × 10⁷ MSCs/kg) so that biodistribution could be assessed by bioluminescence through time on IVIS Lumina XR equipment upon injection of D-Luciferin. As expected, MSCs engrafted mainly in the lungs but could also reach the brain, even though cells only stayed in the brain (or even in the lungs) for the first hours; 24 hr after administration, MSCs were no longer detected in the brain and were barely detected in the lungs (Figures 5F and 5G and 2C). Next, to investigate whether MSCs could reach the parenchyma of the brain, we re-injected the mice (6 × 10⁷/kg) after MSCs had completely disappeared from the mouse circulation. The bioluminescent signal was captured 30 min after MSC administration, confirming the MSC signal in the brain (Figure S2D), and mice were sacrificed to perform immunohistochemistry (IHC) against luciferase and CD31 (to label the blood vessels). Co-localization confirmed that MSCs could, in fact, get out of blood vessels and reach the brain parenchyma (Figure 5H).

Repeated Systemic Administration of Syngeneic MSCs Promotes Phenotype Relief Sustained in Time and Rescue of Cerebellar Neurodegeneration

To overcome the short lifespan of MSCs and transient neuroprotection, we therefore investigated whether repeated treatment with MSCs via the minimally invasive i.v. systemic route would sustain MJD alleviation. The MSC-treated group received four tail vein administrations of 4.5–8 × 10⁷ MSCs/kg every 2–3 weeks, whereas controls were injected with the corresponding volume of Hank’s balanced salt solution (HBSS) according to their body weight. MJD mice were submitted to behavioral tests before transplantation (week 0) and between transplantations, as illustrated in the timeline (Figure 6A). At the end of the experiment, MRI acquisitions were made to assess spectroscopy, and mice were sacrificed for neuropathological assessments.

Figure 6.

Repeated Intravenous MSC Transplantation Promotes Sustained Phenotype Alleviation in Tg-ATXN3-69Q Mice

(A) Schematic representation of the experimental procedure of repeated transplantation of MSCs in the tail vein of Tg-ATXN3-69Q mice. (B–I) Motor performance evaluation of NT-MJD and MJD+MSC Tg-ATXN3-69Q mice. Coordination and balance were assessed through rotarod (n = 13 NT-MJD mice versus n = 14 MJD+MSC mice) (B), pole (n = 8 NT-MJD mice versus n = 6 MJD+MSC mice) (C), and beam walking (n = 7 NT-MJD females versus n = 9 MJD+MSC females) (D). Gait was evaluated by footprint analysis; front base (E), hind base (F), and overlap (G and H) are represented (n = 15 NT-MJD mice versus n = 13 MJD+MSC mice). The mean velocity determined in the most central area of the open field box (zone 2) was used to evaluate mouse anxiety and locomotion capacities (n = 6 NT-MJD mice versus n = 6 MJD+MSC mice) (I). Bars represent the mean ± SE. Repeated-measures ANOVA followed by post hoc Bonferroni-corrected tests. *p < 0.05 and **p < 0.01.

Despite the fact that MSCs are considered safe, possible hepatotoxicity was assessed because several MSC administrations could represent an overload for MJD mice. Serum was collected at the end of the experiment and analyzed for the activity of the transaminases alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP) 9 weeks after the first MSC administration (i.e., 2 weeks after the last MSC transplantation). After four MSC administrations, there was no increase in the activity of hepatic transaminases (Figures S3A–S3C). A slight decrease in the level of ALP was detected in MJD+MSC mice relative to NT-MJD and WT mice, which was probably due to red blood cell contamination in the serum. Spleen weights were equal among the groups (data not shown), and no abnormalities in the organs of MSC-treated mice were detected. Thus, repeated treatment with syngeneic MSCs seems to be safe.

Repeated treatment with MSCs promoted sustainable motor coordination from 2–4 weeks after treatment until the end of the experiment (Figures 6B–6D). Motor improvements were evaluated by rotarod, pole, and beam walking. In the rotarod test, improvements were noticed as soon as 1 week after the first treatment and were prolonged throughout time, showing a gradually increased latency to fall (F_2,25 = 9.22, p = 0.001; Figure 6B). Accordingly, in the pole test, treated mice spent less time to reach the floor from 1 week after the first treatment, and the improvement was maintained throughout time, although, at 8 weeks, differences between the two groups were not significant (F_2,12 = 6.94, p = 0.01; Figure 6C). In the beam walking test, MJD+MSC mice took less time to cross the beam than NT-MJD mice from 4 weeks after the first treatment on (F_2,12 = 7.01, p = 0.01; Figure 6D). All of these results show that MSC-treated Tg-ATXN3-69Q MJD mice improved their balance and locomotor activity.

Gait was evaluated by footprint measurements. Front base, hind base, and overlap (left and right), four measures that are increased in MJD mice compared with WT mice because of balance defects, were all decreased in transplanted animals 8 weeks after treatment (F_2,26 = 3.70, p = 0.04; F_2,26 = 3.81, p = 0.04; F_2,26 = 3.38, p = 0.05; F_2,26 = 4.04, p = 0.03; Figures 6E–6H, respectively), denoting additional gait improvement rather than locomotor enhancement, which could not be observed for the single treatment with MSCs. This may indicate that motor coordination improved right after the treatment, whereas gait needed more time to be established.

Finally, open field is a valued test not only to measure anxiety but also patterns of locomotion in several models of neurodegenerative disorders. Analysis of data was divided into two distinct zones: zone 1, the peripheral area of the square, where mice usually spend most of the time, and zone 2, the central area, mostly used to measure anxiety patterns. Observations in zone 2 of the open field showed that, at 8 weeks, MJD+MSC mice had a higher mean velocity in zone 2 (F_2,26 = 3.76, p = 0.04; Figure 6I), resembling the WT mice (data not shown), illustrating that anxiety levels are reduced and locomotion is improved in MJD mice treated with MSCs.

Neuropathological analysis revealed that animals treated with MSCs exhibited a significantly larger thickness of the molecular and granular layers (t₇ = 2.37, p = 0.05; Figure 7A). Similarly, there was a significantly larger volume of lobule X in MSC-treated Tg-ATXN3-69Q mice compared with the control group (t₇ = 3.55, p = 0.01; Figure 7B). The sum of the volumes of all lobules (II–X) was also significantly increased in MJD+MSC mice in comparison with NT-MJD mice (t₇ = 2.41, p = 0.05; Figure 7B).

Figure 7.

MSC Repeated Intravenous Transplantation Promotes Broad Neuropathological Mitigation in the Cerebellum of Tg-ATXN3-69Q Mice

(A) Illustrative microphotographs of cresyl violet staining of cerebellar lobule X and graphic representation of the measurement of molecular and granular layer thickness (measured together) in lobule X of NT-MJD or MJD+MSC mice (n = 4 NT-MJD mice versus n = 4 MJD+MSC mice). (B) Illustrative microphotographs of cresyl violet staining of cerebellar sections of NT-MJD or MJD+MSC mice and graphic representation of the measurement of the volume of lobule X (n = 4 NT-MJD mice versus n = 4 MJD+MSC mice) and the sum of the volume of all lobules (n = 4 NT-MJD mice versus n = 5 MJD+MSC mice). (C) Illustrative fluorescent microphotographs of lobule VIII stained with PCP4 of NT-MJD and MJD+MSC Tg-ATXN3-69Q mice and quantification of the number of Purkinje cells in the same lobule (n = 6 NT-MJD mice versus n = 5 MJD+MSC mice). (D) Representative membrane showing western blot of mutant ataxin-3 protein (either high-molecular-weight species or the soluble protein form) and β-tubulin in NT-MJD or MJD+MSC mice (n = 7 NT-MJD mice versus n = 9 MJD+MSC mice). Quantification of mutant ataxin-3 bands corrected for individual β-tubulin (endogenous protein) signal intensity is given. Bars represent the mean ± SE. Bonferroni corrected Student’s t tests; *p < 0.05 and **p < 0.01.

Quantitative analysis of the number of PCP4-positive cells revealed a significantly higher number of Purkinje cells in lobule VIII of treated mice compared with the NT group (t₁₀ = 2.36, p = 0.04; Figure 7C). Western blot analysis of the cerebellum revealed a tendency for a decrease of the levels of the aggregated form of ataxin-3 (high-molecular-weight species) (t₁₅ = 1.81, p = 0.10) concomitant with a significant decrease in the soluble form of the ataxin-3 protein (t₁₅ = 2.25, p = 0.04; Figure 7D).

These results suggest that repeated MSC transplantation via the i.v. route sustainably and effectively alleviates motor phenotype and neuropathological defects in MJD mice.

MSCs Protect GABAergic and Glutamatergic Neurons and Reduce Inflammation in MJD Animal Models

To further investigate MSC mechanisms in MJD, proton magnetic resonance spectroscopy (¹H-MRS) was performed in both MJD models. ¹H-MRS is a powerful tool to measure the metabolic profile in vivo, which allows the detection of possible alterations in the concentration of metabolites that play crucial roles in brain physiology after MSC transplantation. Because of space constraints, in the present report we focus on a small number of metabolites, including ones related to neurotransmission, the others being the subject of future studies.

In the LV-based model, ¹H-MRS average plots of the ipsilateral side of the striatum were obtained 1 and 4 weeks after i.c.v. transplantation of MSCs (n = 5 MJD+MSC mice; n = 5 NT-MJD mice) (Figure 8A). As soon as 1 week after treatment initiation, the neurotransmitter γ-aminobutyric acid (GABA) increased upon MSC treatment (MJD+MSC) and remained increased 4 weeks after transplantation (F_1,8 = 8.77, p = 0.01; Figure 8B), indicating preservation of GABAergic neurons. Glutamate was also increased in the MSC group relative to the control 1 week after treatment, although these differences disappeared 4 weeks after transplantation (F_1,8 = 5.69, p = 0.03; Figure 8B). The levels of three other metabolites, phosphocholine (PCH), Myo-inositol (INS), and taurine (TAU), are shown in Figure S4A.

Figure 8.

MSC Transplantation Promotes Modulation of Inflammation and Synaptic Activity and/or Plasticity in MJD

(A) Proton magnetic resonance spectroscopy (¹H-MRS) average plot obtained 1 week after intracerebroventricular transplantation of MSCs. (B) Comparison of the concentration of GABA and glutamate (GLU) in the ipsilateral side of the striatum of non-treated LV-ATXN3-72Q (NT-MJD) and MSC-treated LV-ATXN3-72Q males (MJD+MSC) (n = 5 NT-MJD mice versus n = 5 MJD+MSC mice) 1 and 4 weeks after transplantation. (C) ¹H-MRS plots were obtained from the cerebellum of WT (n = 7) and non-treated Tg-ATXN3-69Q mice (NT-MJD, n = 4) or MSC-treated Tg-ATXN3-69Q mice (single injection i.c.v.: MJD+MSC_icv, n = 6; four repeated i.v. transplantations: MJD+MSC_iv, n = 4) for GABA, the GLU-glutamine (GLN) complex (GLU+GLN), and INS. (D–F) Spearman’s correlations between the concentration (millimolar) of GABA (D), GLU+GLN (E), and NAA+NAAG/INS (F) and the rotarod performance of NT-MJD or MJD+MSC mice after repeated i.v. treatment. (G–I) The mRNA expression levels of the GABA receptor Gabrb2 (n = 5 NT-MJD mice, n = 6 MJD+MSC mice, and n = 7 WT mice) (G) and the GLU receptors Grm1 (H) and Gria3 (n = 5 NT-MJD mice, n = 5 MJD+MSC mice, and n = 7 WT mice) (I) were determined by real-time qPCR. Only females were used in assessments for (C)–(I). In (B) and (C), boxplots represent median, 25%–75% interquartile range, non-outlier range, and outliers. In (G)–(I), bars represent the mean ± SE. Repeated measures ANOVA (A and B) or one-way ANOVAs (C–I) followed by post hoc Bonferroni-corrected tests. *p < 0.05, **p < 0.01, and ***p < 0.001.

To investigate whether different metabolites could have variations in their levels because of MSC treatment, we established a multi-correlation diagram of central metabolites in the striatum of NT-MJD and MJD+MSC mice 1 week after transplantation. Curiously, upon treatment, several positive correlations that were either absent or weak in NT-MJD mice (Figure S4B) were established in MJD+MSC mice (Figure S4C), indicating that metabolites are co-operating with each other to promote neuronal tissue recovery. Because deep metabolic changes have been reported to occur in the cerebellum of MJD mouse models and patients bearing this severe neurodegenerative disorder,38, 39 ¹H-MRS was also performed in the Tg-ATXN3-69Q model. Animals treated either with a single administration of MSCs i.c.v. (MJD+MSC_icv) or with four i.v. injections of MSC (MJD+MSC_iv) were compared with NT Tg-ATXN3-69Q (NT-MJD) and WT mice at the end stage of the experiment to investigate whether transplanting MSCs would correct such metabolic defects (Figure 8C).

The levels of neurotransmitters playing fundamental roles in the cerebellum, such as GABA, glutamate, or the glutamate-glutamine complex, are also severely impaired in MJD.39, 40 Accordingly, in the present model, GABA and the glutamate-glutamine complex were drastically decreased in NT-MJD relatively to WT mice. Irrespective of the treatment modality (i.c.v. or i.v.), a recovery effect was observed in the levels of both GABA (F_3,17 = 8.77, p = 0.001) and the glutamate-glutamine complex (F_3,17 = 12.11, p < 0.001; Figure 8C).

On the other hand, INS, which is frequently used as a marker for glial cells in ¹H-MRS, was upregulated in MJD relative to WT mice. Importantly, INS levels were reduced in MJD+MSC_iv, probably indicating a decrease in gliosis (F_3,17 = 8.89, p = 0.001; Figure 8C). However, in MJD+MSC_icv, the levels of INS were not changed, although there were significant differences among other groups. INS levels in the LV-ATXN3-72Q model also corroborate the present data (Figure S4A), indicating that the surgical procedure of i.c.v. treatment probably elevates the levels of glial cells.

We next investigated whether metabolic changes in MJD+MSC_iv mice would correlate with phenotypic improvements. Importantly, the levels of GABA and the glutamate-glutamine complex in the cerebella of Tg-ATXN3-69Q mice were positively correlated with their rotarod performance (r_s = 0.59, p = 0.05 and r_s = 0.77, p = 0.03; Figures 8D and 8E, respectively).²⁶ Even though, in the present study, no differences were observed in the levels of N-acetylasparate (NAA) and N-acetylaspartylglutamate (NAAG) (data not shown), there was a positive correlation between the levels of (NAA+NAAG)/INS and rotarod performance (measured 8 weeks after the first treatment course), as already shown for another SCA⁴¹ (r_s = 0.87, p = 0.01; Figure 8F). Finally, we highlight that two different subpopulations can be distinguished in every plot, with the MSC-treated group exhibiting higher metabolic levels and increased rotarod performance than controls (Figures 8D–8F), suggesting that these biomarkers might be used to track the success of MSC transplantation.

Furthermore, the cerebella of MJD mice treated i.c.v. were analyzed for the expression of mRNA levels of receptors for glutamate and GABA by real-time qPCR at the end stage of the experiment (Figures 8G–8I). Further supporting the previous results showing an increase in neurotransmitters, the mRNA levels of the GABA type A receptor Beta2 subunit (Gabrb2) and glutamate metabotropic receptor 1 (Grm1), both deficient in Tg-ATXN3-69Q mice,³⁷ were increased in the cerebella of MJD+MSC mice compared with NT-MJD mice (F_2,15 = 11.34, p = 0.001 and F_2,15 = 6.42, p = 0.01; Figures 8G and 8H, respectively). These receptors are relevant for the connectivity of cerebellar neurons, including cerebellar nuclei, basket cells, stellate cells, Golgi cells, and Purkinje cells. No significant differences between groups were noticed in glutamate ionotropic receptor alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) type subunit 3 (Gria3) (F_2,14 = 3.20, p = 0.07; Figures 8I), a less abundant receptor in the mouse cerebellum.

In conclusion, the mechanistic insights gained in the present study are of crucial interest because they provide valuable cellular and molecular information regarding what occurs in vivo upon MSC treatment in Tg-ATXN3-69Q MJD mice. Furthermore, our results point to a neuroprotective action of MSCs that protect GABAergic and glutamatergic neurons, culminating in an increase of both GABA and glutamate neurotransmitters as well as in their receptors, suggesting preservation of synaptic plasticity in the MJD brain.

Discussion

In this work, we investigated whether MSCs can alleviate the phenotype of MJD preclinical models. We found that periodic but not single administration of MSCs promote sustained motor behavior and neuropathology alleviation concomitant with imaging biomarker improvement in pre-clinical MJD animal models, suggesting that clinical protocols with MSCs should include repeated administration.

Pre-clinical studies are of extreme importance to design rational protocols to be used in the treatment of human patients. These should ideally be coupled with biomarkers to enable precise unbiased monitoring of disease progression and efficacy of the employed therapy. A few pre-clinical studies have investigated whether MSCs could alleviate SCA1²⁸ and SCA2.²⁷ However, each neurological disease has specificities of mechanism, brain regions, and cell populations impaired by the disease and resulting clinical presentation, demanding individualized assessment and therapy development. Hence, investigating MSC therapeutic value in MJD upon different therapeutic regimens is of crucial importance to develop effective therapies.

Previous demonstrations that MSCs are safe and promising¹⁶ for therapy of different disorders were the grounds for several trials involving MSC transplantation in patients with neurodegenerative conditions (https://www.clinicaltrials.gov/). In SCAs, five clinical trials, taking place in Taiwan, India, and China, are either running or have finished (ClinicalTrials.gov identifiers NCT01489267, NCT01958177, NCT01360164, NCT01649687, and NCT02540655), and one (NCT02540655), promoted by Steminent Biotherapeutics, was approved by the Food and Drug Administration, emphasizing the hope and trust in this approach. However, the clinical studies report that, despite motor amelioration and delay in disease progression, some patients had regressed to the stage prior to MSC transplantation a few months after treatment.23, 24, 25, 26 This highlights the need for readjustment of the administration protocol. Given that this approach is already in clinical trials,²² it is urgent to establish the best strategy to treat MJD as well as other SCAs and possibly polyglutamine disorders.

In accordance with previous descriptions from clinical trials involving SCA patients as well as other disorders,²² here we show that a single dose of MSCs, injected either in the lateral ventricles or in the cerebellar parenchyma, mediates transient relief of the MJD phenotype. Importantly, periodic i.v. injections of MSCs sustainably alleviate the MJD phenotype, providing evidence that repeated treatment may be required to achieve better outcomes in MJD patients and, possibly, in other SCAs. This probably results from the short lifetime of MSCs after in vivo administration in the MJD context, as we demonstrate here by both MRI of SPION-labeled MSCs injected i.c.v. in the lentiviral paradigm and by bioluminescence in transgenic mice upon i.v. administration of luciferase-labeled MSCs.

Moreover, we show that treatment with MSCs promotes a paracrine effect that probably preserves neurons, astrocytes, and glia by several pathways that culminate in higher synaptogenesis and/or synaptic plasticity. Interestingly, we previously showed that cerebellar transplantation of neural stem cells improves MJD behavioral and histological deficits.¹⁴ However, the use of neural stem cells for clinical applications has some limitations related to neural stem cell (NSC) availability, expansion, and potential immune recognition of the graft.⁴² On the other hand, MSCs are easy to isolate and grow to clinical grades without major ethical issues, and allogeneic transplants can be performed. Moreover, because they exist in adult tissues such as bone marrow, adipose tissue, or dental pulp, among other tissues, it is possible to access a patient’s own cells to treat the disorders whenever MSCs are healthy.16, 43

As recognized, MSCs isolated from the bone marrow are heterogenic, and hematopoietic contaminants may persist in culture, especially in the case of MSCs isolated from mice.⁴⁴ To isolate a relatively pure population of mesenchymal cells, here we simply performed exclusion of CD45+ cells after the established method of isolation of plastic adherence, which resulted in a population of cells that fulfil the requirements defined by the International Society for Cell Therapy for mesenchymal stromal cells³³ and exhibit a phenotype similar to MSCs described in other work.⁴⁵ As expected, MSCs expressed specific neuronal and glial markers in vitro.³⁴ This neural phenotypic plasticity observed in bone marrow-derived MSCs may be explained by the presence of stem cells derived from the neural crest, identified in the adult bone marrow by Wislet-Gendebien et al.⁴⁶ In conclusion, we obtained a mixed population of both immature and more committed “neuron-like” cells, ready to assemble the neuronal environment.

After validating the MSC source, the proof of concept that MSCs are, in fact, beneficial for MJD was obtained by introducing the cells into the parenchyma of the cerebellum of Tg-ATXN3-69Q mice. The cerebellum has an important role in the maintenance of balance and posture, coordination of voluntary movements, and motor learning, and a functional defect in Purkinje cells can trigger cerebellar ataxia, including balance and gait defects.³¹ Transplantation of MSCs into the cerebella of Tg-ATXN3-69Q mice transiently alleviated motor impairments and the associated neuropathology in the region of cellular engraftment, demonstrating their value for MJD therapy. Because proximity to the site of local cell grafting might be important, we next evaluated the clout of i.c.v. transplantations of MSC. Because clinical studies performed on patients with SCAs usually use i.v. and/or intrathecal administration (here represented by i.c.v. transplantation),23, 24, 25, 26 by addressing this subject we also expected to move a step forward into the clinic. On the other hand, because MJD is a disorder affecting several regions of the nervous system, a better outcome was envisaged upon i.c.v. administration of MSCs because the secreted factors could reach the entire CNS through the CSF. The similarity of results to both intra-parenchymal and i.c.v. approaches demonstrates that cell contact is not necessary for MSC beneficial effects, suggesting a prevailing paracrine mode of action.

Behavioral improvements for longer periods have previously been reported. Matsuura et al.²⁸ showed that MSC treatment improved the morphology of the cerebellum and improved the motor behavior for 15 weeks after transplantation when injected intrathecally in a transgenic mouse model of SCA-1. In another study, MSCs from the umbilical cord were injected i.v., and implanted cells were detected in the cerebellum (Purkinje cell layer and internal granular cell layer) for at least 12 weeks with no obvious differentiation.⁴⁷ In our study, MSCs were not able to revert the MJD phenotype later than 4 weeks post-treatment. These differences could be explained by the fact that the MSC lifetime is dependent on the context they are put into because the environment where MSCs are implanted is decisive for their survival.48, 49 Likewise, different mouse models also present diverse grades of severity and disease progression. The Tg-ATXN3-69Q MJD model is extremely severe because it bears the truncated form of the human ataxin-3 protein, meaning that the achievements obtained in this model would probably represent the worst-case scenario. However, this mirrors what happens in clinics, where treatment usually starts at a late stage of the disease. However, when we performed i.c.v. transplantations in the LV-ATXN3-72Q mouse model, MSCs died in a short time frame, indicating that the MJD context per se must be hostile for MSCs.

It has previously been reported that MSCs infused intrathecally or i.c.v. circulate throughout the ventricular system via the CSF⁵⁰ or can integrate the parenchyma of nervous tissues.51, 52, 53 This was not the case in the present study because MSCs were found as a group of cells attached to the walls of the ventricles in both MJD mouse models used, as reported elsewhere.⁵⁴ In the latter study, MSCs administered i.c.v. into experimental autoimmune encephalomyelitis (EAE) mice (a mouse model for amyotrophic lateral sclerosis [ALS]) before disease onset (6 days after EAE induction) were found around the lateral ventricles 30 days after transplantation. Corroborating these results, in the present study, MSCs were detected up to 49 days, although their number was drastically reduced from 7–28 days after transplantation. Moreover, contrary to other reports,27, 47 we could not identify GFP-positive MSCs in the cerebellum 3 weeks after the last i.v. injection.

Although a single local treatment with MSCs did not produce the sustained efficacy desired in our study, four repeated administrations of MSCs, spaced by 2 weeks, could increasingly promote phenotype relief in MJD mice. This effect, despite being small, resulted not only in improved equilibrium and motor performance but also in gait improvement in this particularly severe mouse model of MJD. Therefore, such improvement is expected to result in a meaningful clinical outcome. Furthermore, this result is in accordance with a preclinical study in SCA2, demonstrating that repeated MSC transplantation could alleviate motor function deterioration in a more effective way than a single intracerebral transplantation.²⁷ Moreover, while we were in the process of submitting this paper for publication, a new report was published that re-enforces the phenotypic improvements after repeated treatment with MSCs in MJD.⁵⁵ However, in that study, the authors performed pre-symptomatic treatment, which would not be easy to implemented in the clinic. Our study proves that post-symptomatic treatment is also effective.

A repeated transplantation procedure could, however, have some safety concerns. It was recently demonstrated that repeated autologous infusion of hematopoietic stem cells is safe in human patients with liver insufficiency.⁵⁶ Corroborating the previous study, we did not find any increases in liver transaminases. However, other studies must now address the possible risks of performing repeated allogeneic transplantation. Future pre-clinical and clinical studies with MSCs in MJD and other SCAs are welcome to address this issue. Moreover, the possibility of performing autologous transplantations in MJD may also be explored in the future.²⁶ Alternatively, MSC by-products could replace MSCs themselves, which would avoid transplantation concerns.

Besides phenotypic improvements, MSC have also been reported to mitigate cerebellar neuronal disorganization, particularly in a SCA1 transgenic mouse model, which was suggestive of an alleviation of neuropathology.²⁸ In the present work, MSCs also promoted preservation of cerebellar neurons in the Tg-ATXN3-69Q MJD mouse model. This model presents mutant ataxin-3 inclusions and severe atrophy of the cerebellum with marked disarrangement of Purkinje cells.³¹ Even though alleviation of neuropathology was detected with all MSC administration protocols performed in this study, a more robust effect was observed upon repeated systemic treatment, with preservation of Purkinje cells, thickness of cerebellar layers, and cerebellar volume, indicating that several neuronal populations of the cerebellum are preserved by the action of MSCs. These observations are consistent with previous reports.47, 55 Overall, ours and previous data suggest that multiple systemic administrations of MSCs might alleviate, in a more competent fashion, the cerebellar atrophy characteristic of the Tg-ATXN3-69Q mouse model than a single local injection, whose effects were confined to a more restricted cerebellar region. Importantly, repeated systemic MSC treatment tended to downregulate both forms of ataxin-3.

How MSCs mediate neuroprotection remains unclear and was therefore investigated in this pre-clinical study. Notably, by ¹H-MRS, we show that MSCs can upregulate the expression of neurotransmitters in MJD. Modulation of the excitatory and inhibitory balance in synaptic transmission is recognized in many functions, as in implicit learning or motor control, and deregulation of this balance has been implicated in many brain dysfunctions, such as ischemic stroke, hypoglycemic brain damage, and Huntington’s disease. Interestingly, upregulation of GABA and glutamate, both severely impaired in MJD mice, was recorded after treatment with MSCs in the striatal model. Similarly, in the Tg-ATXN3-69Q model, MSCs also promoted the upregulation of both the glutamate-glutamine complex and GABA in the cerebellum. Moreover, the expression of mRNA encoding receptors for both Glu (Grm1) and GABA (Gabrb2) was increased in treated MJD mice. Clinical manifestations frequently associated with SCAs, such as ataxia, dysarthria, or dysphagia, may result from interruption of one or more connections between the Purkinje cells, the dentate nucleus, and inferior olivary nuclei, where the dentate nucleus plays a central role, controlled by inhibitory corticonuclear GABAergic connections and excitatory glutamatergic collateral connections emerging from mossy and climbing fibers.⁵⁷ Other researchers have also reported a decrease in the levels of GABA receptors or GABAergic function in both mouse models and MJD patients.40, 58 Our results show that MSCs are probably preserving these connections, playing an essential role in synaptic plasticity.

Supporting our results, previous reports have already shown that MSCs are able to enhance the GABAergic system.59, 60 Bae et al.⁵⁹ have demonstrated that MSCs were able to promote synaptogenesis within the cerebellum in a mouse model for Niemann-Pick disease. By gene microarray analysis, they also found that MSCs upregulate the genes encoding subunits of the ionotropic glutamate receptors (AMPA) GluR4 and GABA type A receptor beta2 subunit, influencing the neurotransmission of both excitatory and inhibitory pathways, analogous to our results. MSC could also increase hippocampal GABAergic synaptogenesis in in vitro culture systems by releasing soluble factors into the culture medium.⁶⁰ These results were reversed by using K252a, an inhibitor of the Trk neurotrophin receptor, as well as by TrkB receptor antibodies, and the authors of the study concluded that brain-derived neurotrophic factor (BDNF) should be the responsible neurotrophic factor for the MSC effect—the broadening of GABAergic transmission in hippocampal cultures. In the future, the role of BDNF and the preponderance of other neurotrophic factors in the MSC effect should be evaluated in MJD models.

Inflammation is another important aspect to be considered in MJD.⁶¹ In the present study, INS levels were also reduced in Tg-ATXN3-Q69 mice treated i.v., which suggests that MSC reduced gliosis in mutant mice. It is believed that glia proliferation is favored in this neurologic condition to ensure vascular and metabolic activities usually accomplished by neurons,⁶² so a reduction in this marker may also be indicative of higher neuronal preservation. Astrocytes may also play a fundamental role, acting as a couple with neurons in neurotransmitter recycling and energy supply processes. As key elements in glutamate homeostasis and metabolism, astrocytes take up glutamate (via glutamate transporters) and convert it into glutamine via glutamine synthetase; glutamine is then transported back to neurons, where it is used for glutamate and GABA synthesis. Therefore, several neurological disorders have reported a reduction in both glutamate transporters and glutamine synthase.63, 64 Furthermore, it has been reported that MSCs deliver microRNA 124 (miR-124) to neural progenitor cells and astrocytes (through MSC-derived exosomes), inducing higher expression of glutamate transporters.⁶⁵ Here, glutamate-glutamine complex levels were increased in both single i.c.v. and repeated i.v. treatments. Furthermore, higher levels of both GABA, the glutamate-glutamine complex, or the ratio of NAA+NAAG/INS were correlated with better rotarod performance, indicating that these changes in neurotransmitter levels are responsible for preservation of neuronal functioning and, consequently, motor performance in MJD. These metabolites may therefore serve as biomarkers not only to follow disease progression but also to infer treatment success in patients.

NAA is present in neurons, axons, and dendrites in the mature CNS, and, thus, with a small contribution of NAAG, is frequently considered a biomarker for neuronal integrity.⁶⁶ In MJD, loss of NAA+NAAG in the cerebellum may therefore be indicative of integrity loss and/or dysfunction of Purkinje cells. In the present report, although NAA+NAAG were not statistically increased in MSC-treated Tg-ATXN3-Q69 MJD mice (data not shown), there was a clear preservation of cerebellar neurons.⁴¹ This result is in accordance with a clinical report by Tsai et al.,²⁶ where the authors did not find any differences in the levels of the NAA/creatine (Cr) ratio measured by ¹H-MRS in MJD brains, despite motor improvements being observed after treatment with adipose-derived (AD)-MSC i.v. Further studies are necessary to assess the utility of this biomarker to evaluate the effects of MSC therapy in this paradigm.

Essentially, the results of the present investigation suggest a collection of molecular changes occurring in MJD because of MSC treatment that are spatially and chronologically inter-related and probably involve several populations of cells, such as neurons, microglia, and astrocytes, that ultimately culminate in neuroprotection, as previously suggested elsewhere for other neurodegenerative disorders.67, 68 These changes may ultimately result not only in the regulation of intracellular events that induce sparing of neurons and neurite extension but also higher expression of specific neurotransmitters, probably resulting in higher synaptic activity between neurons. Further studies of the mechanism(s) of MSCs in this disorder will be performed in the future to scrutinize these events.

Importantly, in the present study, we show that continuous therapy with MSCs in Tg-ATXN3-69Q MJD mice can sustainably alleviate motor behavior and induce the reversion of some neurometabolic defects in this fatal disorder. Future studies ascertaining the long-term safety and efficacy of repeated treatment with MSCs or MSC by-products should now be performed because we foresee that a repeated therapeutic strategy may offer a modification of the progression of MJD and other SCAs, providing hope for patients and their relatives.

Materials and Methods

MSC Isolation, Expansion, and Characterization

MSCs were isolated from the bone marrow of 6- to 8-week-old WT mice of both genders with a C57BL/6 background, as reported elsewhere,²⁹ with some modifications. Negative sorting for CD45⁺ was performed afterward (FACSAriaIII), and MSCs were expanded and frozen with standard protocols. MSCs from the same batch in passages 15–17 were used throughout the study. For multipotency assessment, the Mouse Mesenchymal Stem Cell Functional Identification Kit (R&D Systems, catalog no. SC010) was used according to manufacturer’s instructions. For MSC characterization, flow cytometry analysis and immunocytochemistry (ICC) were performed. Further information can be found in the Supplemental Materials and Methods.

MSC Infection with an LV Encoding GFP and Luciferase

The LV vector encoding GFP and luciferase was produced in 293T cells using a four-plasmid system as described previously.³⁰ MSCs were infected with 80 ng or 30 ng of LV/100,000 cells encoding GFP or Luciferase, respectively, for 24 hr. More details can be found in the Supplemental Materials and Methods.

Transgenic Mouse Model

An MJD transgenic mouse model (C57BL/6 background) expressing the N-terminal truncated human ataxin-3 with 69 glutamine repeats (Tg-ATXN3-69Q) together with an N-terminal hemagglutinin (HA) epitope, driven by the L7 promoter, was used.³¹ A colony of these mice was brought to our animal facility in the Center for Neuroscience and Cell Biology (CNC) of the University of Coimbra, and the line was maintained by backcrossing heterozygous. Animals were housed in a temperature-controlled room and maintained on a 12 hr light and dark cycle. Food and water were available ad libitum. The genotype was confirmed by PCR. Unless otherwise stated, all groups included age-matched mice of both genders.

Stereotaxic Surgery for MSC Administration in the Tg-ATXN3-69Q Model

Post-symptomatic Tg-ATXN3-69Q mice 4–7.5 weeks of age were transplanted with a single injection of 300,000 MSCs/3 μL HBSS either into the cerebellum or into the right lateral ventricle of the brain under anesthesia. Further details are provided in the Supplemental Materials and Methods.

Intravenous Injection of MSCs in the Tg-ATXN3-69Q Model

For i.v. therapy, 4- to 6.5-week-old Tg-ATXN3-69Q MJD mice were transplanted with 4.5–8 × 10⁷ MSCs/kg/70–150 μL of HBSS every 2–3 weeks four consecutive times (or twice once a week for the study of MSC biodistribution after i.v. injection).²⁷ NT mice were injected with HBSS. Failed transplants were excluded from the analysis. For more details, see the Supplemental Materials and Methods.

Behavior Assessment

Tg-ATXN3-69Q MJD mice were subjected to motor tests before the first MSC injection (0 weeks of experimental design) and in certain periods after treatment according to each timeline represented in the corresponding figures. Tests were performed after acclimatization. A battery of tests (rotarod, beam walking, pole, swimming, footprint, and activity box) were used and are described in detail in the Supplemental Materials and Methods.

LV-Based Striatal Model for MJD

The LV-induced MJD model was produced by injection of lentiviral vectors in the striatum encoding for human ataxin-3 with 72 glutamines (LV-ATXN3-72Q), as established previously in our lab.⁶ C57BL/6 mice from Charles River Laboratories were used. A detailed protocol can be found in the Supplemental Materials and Methods.

MSC Staining with SPIONs for Longitudinal Visualization by MRI

For MRI detection, MSCs were pre-incubated with SPIONs (Rhodamine B, Molday ION, catalog no. CL-50Q02-6A) according to the manufacturer’s instructions.³² A detailed description is provided in the Supplemental Materials and Methods.

Stereotaxic Surgery for MSC Administration in the LV-ATXN3-72Q Model

Mice were treated 2 weeks post-viral infection, followed by MRI longitudinally to either study MSC viability or for structural and MRS acquisition and sacrificed 8 weeks after treatment. 300,000 MSCs/animal resuspended in 3 μL of HBSS were delivered into the right lateral ventricles; saline (HBSS) was used in the case of controls.

MRI Acquisition

In vivo image acquisition was conducted with a 9.4 T magnetic resonance small animal scanner (BioSpec 94/20) with a standard Bruker cross-coil setup using a volume coil for excitation (86/112 mm of inner/outer diameter, respectively) and a quadrature mouse surface coil for signal detection (Bruker Biospin, Ettlingen, Germany) at the Institute for Nuclear Sciences Applied to Health (ICNAS), University of Coimbra. Volumetric analyses and ¹H-MRS were performed, as detailed in the Supplemental Materials and Methods.

Bioluminescence Studies to Assess the Biodistribution of MSCs after Intravenous Injection

Bioluminescence images were acquired 30 min, 2 hr 45 min, 24 hr, 48 hr, and 7 days after the injection of MSCs pre-labeled with luciferase (8 × 10⁷ MSCs/kg) with the IVIS Lumina XR imaging system using Living Image software (version 4.10, Xenogen). For each determination, mice were injected intraperitoneally (i.p.) with D-Luciferin (100 mg/kg) and anesthetized, with bioluminescence images being obtained 20 min after D-Luciferin injection. For quantification of the bioluminescent signal, a region of interest (ROI) was drawn around the chest and cranium. Values are expressed as photons per second. After a second injection of MSCs (6 × 10⁷ MSCs/kg), the bioluminescent signal was acquired, and the mice were sacrificed to assess the engraftment of cells in the brain and specifically in the cerebellum parenchyma. A detailed description of the methodology can be found in the Supplemental Materials and Methods.

Tissue Preparation

Animals were intracardially perfused with 1× PBS, followed by fixation with 4% paraformaldehyde (PFA) (Sigma), post-fixed in 4% PFA for 24 hr at 4°C, and incubated in 20% sucrose in 1× PBS for 4–5 days at 4°C. Brains were then sliced at a cryostat-microtome (Leica CM3050S, Leica Microsystems) and 35-μm sagittal (in the Tg-ATXN3-69Q model) or 25-μm coronal sections (in the LV-ATXN3-72Q model) were serially collected in PBS in 0.05 μM sodium azide at 4°C for IHC procedures to analyze MSC engraftment and neuropathological evaluations, according to the protocols described in the Supplemental Materials and Methods.

Protein Extraction and Western Blot

Protein was extracted from the cerebellum after dissection, and ATXN3 levels were analyzed by western blot, as detailed in the Supplemental Materials and Methods.

Total RNA Isolation and Real-Time qPCR

Total RNA was extracted with the NucleoSpin RNA II isolation kit (Macherey Nagel) from the cerebellum after dissection, and real-time qPCR was performed to determine the levels of Gabrb2, Gria3, Grm1, and Gapdh, as detailed in the Supplemental Materials and Methods.

Statistics

In the behavioral tests, the effect of group (e.g., treated versus NT MJD mice) and time (before and 2, 4, and 8 weeks after treatment) on all behavioral response variables (e.g., latency to fall from a rotarod) was assessed through repeated measures ANOVA, followed by post hoc Bonferroni pairwise comparisons to identify significant differences among groups and time periods. One-way ANOVA followed by post hoc Bonferroni pairwise comparisons was used to compare means among three or more groups, and Bonferroni-corrected Student’s t tests were applied to compare means between two groups. Bars representing the mean ± SE or boxplots representing median, 25%–75% interquartile range, non-outlier range, and outliers are shown for each group. Spearman correlation tests were used to study the relationship between behavioral tests results (e.g., latency to fall from a rotarod), with neuropathology results (e.g., molecular layer thickness) and diverse metabolic compounds measured by ¹H-MRS (e.g., GABA versus glutamate).

All statistical analyses were carried out in R statistical software (version 3.01). Response variables were tested for normality (quantile-quantile [Q-Q] plots) and homogeneity (Cleveland dot plots) before each statistical test and transformed when needed. All analyses were performed assuming a significance level of p < 0.05.

Study Approval

All animal experiments were carried out in accordance with the European Community Council Directive (86/609/EEC) for the Care and Use of Laboratory Animals, and were previously approved by the Responsible Organization for Animal Welfare of the Faculty of Medicine and CNC of the University of Coimbra (ORBEA and FMUC/CNC, Coimbra, Portugal; ORBEA_66_2015/22062015). All researchers were certified to perform experiments with animals (Federation of European Laboratory Animal Science Associations [Felasa]-certified course and Direcção Geral de Veterinária, Lisbon, Portugal) (General Management of Food and Veterinary [DGAV]: 0421/000/000/2015).

Author Contributions

C.O.M., research study design, experiment conduction, data acquisition, data analysis, and manuscript writing; A.M., data acquisition and data analysis; T.P.S., data acquisition and data analysis; J.B., data acquisition and data analysis; A.V.-F., data acquisition and data analysis; D.P., data acquisition and data analysis; C.N., research study design and data acquisition; J.A., data acquisition and data analysis; S.D., data acquisition; I.B., data acquisition and data analysis; J.S., data acquisition; L.I.P., data analysis; J.C., data analysis; V.H.P., data analysis; P.R.-S., data acquisition; V.A., data acquisition; I.N.-C., data acquisition and data analysis; R.J.N., data analysis and research study design; C.G., data acquisition and data analysis; M.C.-B., research study design provision of reagents; L.P.d.A., research study design, provision of reagents, and manuscript writing.

Conflicts of Interest

The authors declare no competing financial interests.