Hepatocyte growth factor mediates MSCs stimulated functional recovery in animal models of MS

MSC -CM modulates neural development and disease in EAE

Long-term functional recovery in demyelinating disease depends on myelin repair and suppression of pathogenesis both of which appear to be influenced by MSCs. To determine whether the neural responses and functional recovery seen after treatment of animals with infused MSCs reflected release of soluble factors, the effects of MSC conditioned medium (MSC-CM) on neural cell development and functional recovery were assayed. Exposure of neurosphere cultures to MSC-CM resulted in a reduction in the proportion of astrocytes and an increase in the proportion of A2B5+ oligodendrocyte precursor cells (OPCs), oligodendrocytes and neurons that developed (Fig. 1a). In control cultures after 3 days in vitro, approximately 30% of cells were GFAP+ astrocytes and A2B5+ OPCs while 25% were O4+ cells and less than 5% neurons. By contrast after 72 hrs in MSC-CM, the proportion of astrocytes was reduced to 20%, the proportion of O4+ cells (50%) and neurons (25%) increased significantly (Fig. 1b)(p<0.05).

Injection of MSC-CM significantly reduced disease load in EAE animals (Fig. 1c). Injection of MSC-CM 17 days after MOG_35–55 immunization, at the peak of disease resulted in a rapid reduction in functional deficits in 11/ 11 animals. While control animals maintained a clinical score between 2.5 and 3, all animals that received MSC-CM showed some functional improvement. Four of the 11 animals improved to a clinical score of 1 and 2 other animals appeared completely normal within 5 days. This functional recovery was correlated with a reduction in the extent of regions of demyelination seen in spinal cord sections stained with Luxol Fast Blue (Fig. 1d).

MSC-CM stimulated recovery depends on a 50–100Kd activity

Size fractionation studies demonstrated the disease modifying activity of MSC-CM required a fraction with a molecular mass between 50–100kD. Fractions with masses 1–50kD did not significantly modify disease progression in MOG_35–55-EAE in vivo (data not shown). By contrast, 1–100kD (MSC-CM_100kD) fractions reduced disease burden in the EAE model (n=11) (Fig. 2a) and increased the proportion of oligodendrocytes and neurons from neurosphere cultures (Fig. 2b). This functional improvement was correlated with improvements in the histology of the spinal cord that were comparable to that seen with complete MSC-CM. Injection of control medium or heat inactivated CM (n=8) had no obvious effect on either functional outcome or histology.

Isolation of mononuclear cells from the spinal cords of control and MSC-CM_100kD treated EAE animals demonstrated alterations in the relative levels of inflammatory cytokines (Fig. 2c). The level of the pro-inflammatory cytokines interferon-γ (IFN-γ), (IL)-17, tumor necrosis factor- α (TNF-α), IL-2, and IL-12p70 were decreased in animals that received MSC-CM_100kD while levels of anti-inflammatory cytokines such as IL-10 and IL-4 were increased at day 35 after EAE onset.

Recombinant HGF modulates disease progression in EAE

Based on the size fractionation studies, known MSC expression data^27,28 and previously described biological properties^29,30 hepatocyte growth factor (HGF) was identified as a potential candidate molecule within the MSC-CM_100kD fraction that could modulate disease in EAE. Western blot analyses of active MSC-CM_100kD identified HGF (Mwt 62kD alpha chain + 34kD beta chain) at a concentration of between 600–800ng/ml in all active samples (n= 6) but not in controls (Fig. 3a). To determine whether recombinant HGF treatment could promote functional recovery in EAE, animals at the peak of disease (clinical score=3) received injections of either PBS (n=10) or HGF at two different concentrations (50 and 100ng/animal) (n=10) every other day over a five-day period (days1, 3 and 5) and their clinical scores monitored for 30 days post immunization. The clinical score of control animals remained relatively constant between 3–3.5 while those that received HGF improved (Fig. 3b). The higher dose of HGF appeared to provide slightly better functional recovery although both doses reduced the mean functional score below 1.5 (Fig. 3b). Compared to IV injections the functional improvement of animals following IP injection of HGF (100ng) was reduced. For example, at 30 days post immunization the mean score of IP treated animals was approximately 2.4 compared to 1.5 for IV treated animals (Fig. 3b). All subsequent studies utilized IV delivery.

The HGF-induced functional improvement was closely correlated with histological improvement. In control EAE animals, Luxol Fast Blue staining revealed extensive regions of white matter pathology along the length of the spinal cord 30 days after immunization (Fig. 3c). In HGF treated animals at the same stage, the degree of demyelination was dramatically reduced and the majority of white matter appeared relatively normal (Fig. 3c). The level of inflammatory cell infiltration in EAE animals was also reduced in HGF treated animals. In control EAE animals, H & E staining revealed white matter regions with a high density of infiltrated cells (Fig. 3c) and this was confirmed by labeling with anti-CD3 antibodies (Fig. 3c). Both the extent of the infiltrate and the density of cells were reduced in animals treated with HGF. Myelin integrity was confirmed in 1um Epon sections labeled with Toluidine Blue and by ultrastuctural analyses (Fig. 3c) that demonstrated a decrease in the number of unmyelinated axons. In lesion areas of EAE animals a majority of axons were unmyelinated (80±20%) while in lesion areas of HGF treated EAE animals a minority (10±4% p=0.05) were unmyelinated. Analyses of myelin thickness in lesions areas showed that HGF treatment resulted in substantially thicker myelin sheaths around small diameter axons, as well as an increase in the relative number of small diameter (>1μm) axons (Fig. 3d). Together these studies suggest that treatment of EAE animals with HGF results in significant functional improvement and an enhancement in the extent of spinal cord remyelination. To assess the timing of HGF-induced histological improvement, animals were sacrificed at the time of treatment initiation and 3, 11 and 17 days after treatment initiation and their histology assessed on 1um sections. Spinal cords had extensive demyelinated lesions and cellular infiltrates at initiation of treatment that were still apparent after 3 days of treatment (Sup. Fig. 1). After a further 8 days, the lesions were largely resolved although there was a residual elevated cellularity. By 17 days after initiation of treatment, light microscopy (Fig. 3c) and ultrastructural analyses (Fig. 3c) reveal extensive remyelination in HGF treated animals with essentially all fibers including small diameter fibers extensively myelinated. These data suggest HGF both reduces pathological pressure and enhances or facilitates myelin repair.

HGF and MSC-CM stimulated recovery is mediated by cMet

In other systems the biological properties of HGF are mediated through the tyrosine receptor kinase cMet, the best characterized receptor for HGF³¹. To determine if the functional benefits of HGF in EAE were also mediated by cMet, EAE animals were treated with a function blocking cMet antibody ^32,33 (2× 250ng/animal IV 24hr prior to HGF treatment). In animals that received control IgG followed by 3 injections of HGF at the peak of disease (n=8), functional improvements were evident after 24 hrs that plateaued after 5–7 days (Fig. 4a). By contrast, animals treated with cMet antibodies (n=10) failed to show any significant improvement over the subsequent 14 days. Histologically the lesions in animals treated with cMet antibodies appeared similar to controls with only limited remyelination (data not shown).

The capacity of HGF to modulate the immune response in EAE animals was also negated by treatment with cMet antibodies. For example, compared to untreated EAE animals relatively low levels of the proinflammatory cytokines IFN-γ, IL-17, TNFα, and IL-2 were detected in HGF treated animals (Fig. 4c) and these were significantly elevated in animals that received cMet antibodies. Likewise the level of anti-inflammatory cytokine IL-10 and IL-4 elevated in HFG treated animals were reduced following treatment with cMet antibodies (Fig. 4c). ELISPOT assays demonstrated that HGF treatment resulted in a reduction in the frequency of spinal cord derived MOG35– -specific proinflammatory cytokine producing cells (IFNγ IL17) and an increase in IL-10 producing cells that was reversed by cMet antibody treatment (Fig. 4e). These results confirm that in the setting of EAE, the biological activity of HGF is mediated through cMet.

Treatment with cMet antibodies blocked MSC-CM mediated functional recovery. In control IgG treated animals (2X injections, 24hr prior to treatment), MSC-CM stimulated a rapid and sustained functional recovery (Fig. 4b) (n=9). By contrast, animals treated with cMet antibodies showed little or no functional improvement over the subsequent 15 days (Fig. 4b) (n=11). Treatment of MSC-CM with function blocking anti-HGF for 2 hrs prior to injection also inhibited its capacity to reduce functional deficits in EAE (Fig. 4b, black line). Cytokine analyses revealed that the decrease in pro-inflammatory cytokines and the increase in anti-inflammatory cytokines associated with MSC-CM were largely reversed by prior treatment with cMet antibodies (Fig. 4d) as were the alterations in the frequencies of antigen specific IFNγ, IL17 and IL-10 secreting cells shown by ELISPOT analysis (Fig. 4f). These data strongly suggest that HGF signaling in MSC-CM is critical for mediating functional recovery and promoting histological improvement in MOG induced EAE.

cMet mediates HGF/MSC-CM stimulation of neural development

To assess the role of HGF in MSC-C_M100kD induced neural cell development the relative generation of the different classes of neural cells from postnatal neurospheres and the dependence on cMet signaling was compared between HGF and MSC-CM_100kD. In control neurosphere-derived cultures grown for 7 days, the proportion of A2B5+ OPCs, O4+ oligodendrocyte lineage cells and beta-tubulin+ neurons was below 10% while the proportion of GFAP+ astrocytes was greater than 60% (Fig. 5a–b). By contrast, in the presence of MSC-CM_100kD the proportion of A2B5+ OPCs was increased to approximately 25%, the proportion of O4+ cells was greater than 30% and the proportion of beta-tubulin+ neurons was greater than 40% while the proportion of astrocytes was dramatically reduced (approx. 10%). In both cases, cells tended to form clusters although this was more prevalent in controls and total cell numbers were not significantly different in either condition. Similar changes in cell proportions were seen in cultures treated with HGF (45ng/3×10⁵ cells) although the reduction in GFAP+ astrocytes was less pronounced with HGF than with MSC-CM_100kD. Blocking cMet negated the effects of both MSC-CM100kD and HGF and the cellular composition of the cultures reverted to that of controls (Fig. 5a–b). Similarly, incubation of CM_1–100kD with function blocking anti-HGF antibodies negated its effects on neural cell development consistent with a major role for MSC-CM-derived HGF in the regulation of neural cell development (Fig. 5a–b).

Repopulation of demyelinated lesions depends on OPCs migration that can be stimulated by MSCs³⁴. To assess if EAE pathology based responses were dependent on cMet, neurospheres were established from the subventricular zone of animals with ongoing EAE and cell migration compared in the presence or absence of MSC-CM_100kD or HGF±cMet supplemented medium. Cell migration was enhanced in the presence of HGF and this was inhibited by cMet antibodies (Fig. 5c). Similarly, cell migration from EAE-derived neurospheres was enhanced by MSC-CM_100kD and this effect was blocked by anti-cMet or anti-HGF antibodies (Fig. 5d). The majority of the migratory cells gave rise to either neurons (Fig. 5c–d) or oligodendrocytes (Fig. 5d) and very few generated astrocytes (data not shown). Consistent with HGF mobilizing OPCs and enhancing remyelination, in vivo EAE lesions of animals treated with HGF contained greater than twice as many EGFP-PLP+ oligodendrocytes (3.1±0.3/200μm²)than lesions of control animals (1.8±0.2/200μm²) (Fig. 5e). The average number of lesions was also significantly reduced in animals treated with HGF compared to controls. For example, in control EAE animals the average section of spinal cord had 6±0.4 areas of lesion that was reduced to 3.4±0.6 in animals treated with HGF(p<0.05).

HGF promotes repair in non-immune mediated demyelination

The restitution of histological integrity seen in EAE following treatment with HGF is likely a combination of suppression of the immune response and the promotion of myelin repair. To determine whether HGF promoted myelin repair in non-immune mediated demyelination in vivo, spinal cord LPC lesion and slice culture models were employed. In animals injected iv with HGF (0.8μ on days 4, 6 and 11 after an LPC lesion and examined on 14 or 28 days post lesion (dpl), a 30% reduction in overall lesion volume was seen by Luxol Fast Blue staining compared to saline controls (Fig. 6a–b). Labeling of frozen sections with antibodies to myelin basic protein (MBP) demonstrated only limited expression in control lesions while in the HGF treated lesions significant MBP staining was seen throughout the lesion at 14 dpl (Fig. 6c–d). Similarly the density of NG2+ cells was increased approximately 8 fold in HGF treated lesions (4±2 to 33±5 /unit area)(p<0.005) (Fig. 6e,f and p) and greater than 25% of these cells were proliferating as shown by Ki67 double labeling (data not shown). By contrast, while control lesions showed an extensive astrocyte response with intense GFAP labeling around the lesion perimeter this was significantly attenuated in HGF treated animals (Fig. 6g–h).

The increase in NG2+ and MBP+ labeling in the HGF treated lesions was accompanied by a dramatic increase in the extent of remyelination (Fig. 6i–m). In control lesions, (n=9) few myelinated axons were present in lesions, even at the lesion edge (Fig. 6i) while in HGF treated animals (n=8) myelinated axons were distributed throughout the lesion interspersed with cells full of myelin debris (Fig. 6j). Ultrastructural analyses showed that control lesions contained largely unmyelinated axons some of which were ensheathed in glial processes (Fig. 6k) while in HGF treated animals lesions contained axons of different sizes at different stages of remyelination with some relatively thick myelin sheaths (Fig. 6l) but the majority with thin myelin (Fig. 6m). Calculation of G ratios confirmed the abundance of thinly myelinated axons in HGF treated lesions (Fig. 6n) and showed a greater than 5 fold increase in the density of remyelinated axons (18± 2 to 130±6/unit area) (p<0.0002). Treatment of animals with 0.4μl HGF resulted in increased but less profound enhancement of remyelination and by day 28dpl some limited remyelination was apparent in controls although the beneficial effects of HGF were still evident (data not shown).

Consistent with the HGF induced functional recovery in LPC lesions, treatment of LPC demyelinated slice cultures with HGF enhanced recovery. Slices of P7 cerebellum myelinate extensively in vitro over a 7day period (Sup. Fig. 2A,a–b) and treatment with LPC induces a rapid and sustained demyelination (Sup. Fig. 2A, c–d). Exposure of LPC treated slices to 50ng/ml HGF promoted re-ensheathment by oligodendrocytes and reestablished the cytoarchitecture of the slice (Sup. Fig. 2A,e–f). Lower doses of HGF (10ng/ml) had little effect while higher doses (90ng/ml) resulted in marginal additional improvement. A similar HGF-induced recovery of MBP profiles was seen in slice cultures from cerebral cortex treated with HGF (Sup. Fig. 2A,g–i.). Quantification of the extent of myelination/oligodendrocyte demonstrated a 2.5 fold decrease after LPC treatment that was largely reversed by HGF treatment (Sup. Fig. 2b). This HGF induced functional recovery was negated by exposure to anti-cMet antibodies (Sup. Fig. 2b). This recovery likely reflects stimulation of oligodendrocyte development since growth of P0 spinal cord dissociated cell cultures in 50ng/ml HGF significantly increased the proportion of O4+ and O1+ cells (Sup. Fig. 2c–d) from approximately 3% (O4+) and 1% (O1+) in control cultures to greater than 15% (O4+) and 23% (O1+) after 3 days in HGF. To confirm the expression of cMet on OPCs, cells were double labeled with A2B5 and anti-cMet in vitro after 3 days in culture. Virtually all OPCs in culture expressed cMet (Sup Fig. 3a–c). By contrast in vivo in naïve animals very few cells of the oligodendrocyte lineage expressed detectable cMet while in the setting of demyelinating lesions virtually all cells of the oligodendrocyte lineage expressed high levels of cMet (sup. Fig. 3d–f).

Taken together these studies suggest that the ability of MSCs to reduce disease burden in models of EAE is dependent on their promotion of HGF signaling that regulates both immunomodulation and enhances remyelination in the setting of demyelinating disease.

Immunization

All studies were approved by Case Western Reserve University School of Medicine's IACUC and animals were maintained in a NIH/AAALAC-approved facility. Chronic EAE was induced in 8–12 week old female C57BL/6 mice by subcutaneous immunization with 200μL of 200μg MOG_35–55 with complete Freund's adjuvant. Pertussis toxin (500ng) (List Biological Laboratories, Campbell, CA) was injected IP at 0 and 2 days. Animals were sacrificed at Day 35 by transcardial perfusion with 4% paraformaldehyde. Animals were graded according to a standard clinical index: 1=limp tail; 2=hind limb weakness; 3=one limb plegic; 4=plegia of 2 limbs; 5=moribund or dead..

Induction of local demyelination and HGF treatment

LPC-induced demyelination of the spinal cord was performed as described previously⁴⁹. Animals were anesthetized with 0.001 mL per kg of body weight of 3 ml ketamine (Fort Dodge), 3 ml xylazine (Lloyd Laboratories), 1 mL acepromazine (Webster Vet). A T10 laminectomy was performed and 1.5 μl of 1% freshly prepared LPC (Sigma) solution was infused at 15 ml h⁻¹. Rats received either 0.4, 0.8μg/Kg HGF or vehicle control at on d5, d6 and d11 by tail vein. Animals were sacrificed at d14 and 28.

Preparation of Human mesenchymal stem cells conditioned medium and treatment protocols

Human mesenchymal stem cells (hMSCs) were isolated from bone marrow aspirates and grown in DMEM-low glucose supplemented with 10% FBS. This growth medium was collected between days 11–14 (MSCs-CM). The MSCs used to prepare conditioned medium were derived from 5 separate donors to insure that the results were not donor specific. Similar medium lacking conditioning by hMSC was collected as control medium (Ctrl-CM)⁵⁰. Conditioned growth medium was concentrated 10-fold through YM-3 centrifugal filter devices with a YM-100 (100,000MW) devices (Millipore, Billerica, MA). This 100kD fraction was termed MSC-CM_100kD. Protein concentrations were estimated using the BCA Protein Assay Kit (Pierce, Rockford, IL) and 0.5mg/protein was used for cell culture and in vivo studies. Recombinant HGF (HGF), cMet and anti-HGF antibodies were purchased from R&D Systems (Minneapolis, MN). Two doses of rHGF (50ng and 100ng) were injected intravenously on day 14 at peak disease every other day for a total of three injections. For cMet antibody treatment, animals were injected 24 hours prior to treatment. Conditioned media was incubated with function blocking anti-HGF for 1hr prior to use. In culture, HGF was used at 45ng/2×10⁵ cells and anti cMet at 50ng/2×10⁵ cells added 1hr prior to HGF.

Western blot

Conditioned medium proteins were separated on 10% SDS-Polyacrylamide gels transferred to a polyvinylidene difluoride membrane, and incubated with anti-HGF monoclonal primary antibodies (Cell Signaling Technology, Beverly, MA) and HRP-conjugated secondary antibodies. Visualization was performed with an enhanced chemiluminescence system (ECL, GE Healthcare Biosciences, Piscataway, NJ) according to the manufacturer's instructions.

Neuropathology

Sections (4% PFA fixed) were stained with Hematoxylin and Eosin (H&E) for detection of inflammatory infiltrates and Luxol Fast Blue (LFB) for myelin detection. For electron microscopy, animals were perfused with 2.5% glutaraldehyde. CNS tissues were postfixed in 1% osmium tetroxide for 1 hour, dehydrated, and embedded in epoxy resin. One-micron epoxy sections were stained with Toluidine Blue and examined by light microscopy. For EM, thin sections were stained with uranyl acetate and lead citrate and examined in a Joel 100CX microscope at 80Kv. For LFB, H&E and Toludine blue analysis 3 different animals were assayed with at least 16 sections/animal. For quantification of the proportion of myelinated and unmyelinated axons was taken from multiple sections at least 2 different spinal cord regions from 2 animals. G ratios were determined by analysis of at least 200 axons in each condition. For analysis of lesion load the data was obtained from 10 sections taken from 3 different animals of each control and HGF treated EAE. Data was collected using image j and the units represent cm² ± standard deviation. Sample sizes were determined based on power estimates. Lesions were identified as shadow areas in LFB stained sections. All images were taken by an investigator blinded to the treatment of the individual animals.

Neural Cell Cultures

Neurosphere cultures were prepared from newborn mouse cortex and adult SVZ and grown in 10ng/mL EGF. Second passage neurospheres were dissociated and plated at 2×10⁵cells/coverslip in the absence or presence of whole (MSC-CM) or fractioned (MSCs-CM 100kD), rHGF ± anti-cMet. (1:50; Santa Cruz Biotechnology, Santa Cruz, CA) for 3 days prior to analyses. Cells were labeled with antibodies to Nestin (rat-401, Developmental Studies Hybridoma Bank, Iowa City, IA), A2B5, O4, β-tubulin III (Sigma Aldrich, St. Louis, MO) and GFAP (R&D Systems, Minneapolis, MN). The proportion of positive cells was determined in 3 randomly selected areas/coverslip in at least three separate experiments derived from 3 independent neurosphere preparations. Data assessed by independent t-test. To determine the effects of MSC CM1-100 and HGF on the migration of neurosphere derived cells, neurospheres were grown for 14 days and replated on poly-L-lysine coated coverslips in the presence of CM_100kD, CM_100kD + anti-cMet, CM_100kD + anti-HGF, and HGF + anti-cMet. Cells were allowed to migrate for 24hrs and the preparations fixed with cold methanol and processed for immunohistochemistry as above. Cerebellar slice cultures were established from 300μm saggital slices of P9 cerebellum cut on a Leica Vibratome. Slices were grown for 3 days in DMEM/BME with N2 and PDGF -AA (20ng/ml) before being switched to media containing 15% horse serum. Slices were treated with Lysolecithin (LPC, Sigma) at a final concentration of 0.5mg/ml for 16 hours rinsed and then incubated in either control media or control media containing 50ng/ml HGF for 3 to 7 days. Oligodendrocyte ensheathment was assayed by MBP expression (Covance, SMI99, Invitrogen Alexa-594). Cortical slices were treated similar but obtained from coronal sections of P4 brain and grown for 4 days prior to LPC treatment.

ELISPOT assay for MOG_35–55-specific CD4+ T cells in the CNS

Cells were isolated from the spinal cords of animals with EAE following gross manual disruption, filtered, washed and counted. ELISPOT plates (Millipore, Billerica, MA) were coated with IFN-γ antibody (eBioscience, San Diego, CA), anti-IL-10 (eBioscience) or anti-IL-17 (eBioscience) overnight. Cells were stimulated with medium alone or with 10μg/ml autoantigen. Subsequently, biotinylated anti-IFN-γ (Mabtech, Cincinnati, OH), anti-IL-10 or anti-IL-17 (eBioscience) was added, followed by Strepavidin-HRP (Mabtech) and developed with TMB ELISPOT substrate (Carl Zeiss Vision, Hallbergmoos, Germany). Spots were counted by AID ELISPOT plate reader version 4.0. Individuals were designated as responders if the numbers of spots in the presence or absence of HGF or HGF plus cMet-Ab were significantly altered (t-test) (P < 0.05) from controls and the frequency of myelin reactive CD4+ cells calculated.

Cytokines analysis

IL-17, IFN-γ, TNF-α, IL-2, IL-10 and IL-4 were analyzed with the Qiagen LiquiChip System (BioChipNet, Reutlingen, Germany) using anti-mouse monoclonal antibodies with supernatant samples and standards in duplicate. Data represent mean±standard deviation assayed by t-test (n=3). Biotinylated secondary antibodies were added followed by streptavidin-horseradish peroxidase (Strep- HRP) and peroxide-chromagen substrate following the manufacturer's instructions.