Mesenchymal Stromal Cells Rescue Cortical Neurons from Apoptotic Cell Death in an In Vitro Model of Cerebral Ischemia

Abstract

Cell therapy with mesenchymal stromal cells (MSCs) was found to protect neurons from damage after experimental stroke and is currently under investigation in clinical stroke trials. In order to elucidate the mechanisms of MSC-induced neuroprotection, we used the in vitro oxygen–glucose deprivation (OGD) model of cerebral ischemia. Co-culture of primary cortical neurons with MSCs in a transwell co-culture system for 48 h prior to OGD-reduced neuronal cell death by 30–35%. Similar protection from apoptosis was observed with MSC-conditioned media when added 48 h or 30 min prior to OGD, or even after OGD. Western blot analysis revealed increased phosphorylation of STAT3 and Akt in neuronal cultures after treatment with MSC-conditioned media. Inhibition of the PI3K/Akt pathway completely abolished the neuroprotective potential of MSC-conditioned media, suggesting that MSCs can improve neuronal survival by an Akt-dependent anti-apoptotic signaling cascade. Using mass spectrometry, we identified plasminogen activator inhibitor-1 as an active compound in MSC-conditioned media. Thus, paracrine factors secreted by MSCs protect neurons from apoptotic cell death in the OGD model of cerebral ischemia.

Introduction

Transient focal cerebral ischemia (stroke) results in severe irreversible loss of neuronal cells with persistent neurological deficits in affected individuals. Since current concepts of stroke treatment are confined to acute systemic or local thrombolysis therapy, cell-based therapeutic approaches have received considerable attention as a delayed treatment strategy.

Mesenchymal stromal cells (MSCs) are particularly attractive candidates for cytotherapy due to their neuroprotective properties and the low immunogenicity (Gnecchi et al. 2005; Dezawa et al. 2004). Several pre-clinical animal studies in rats have demonstrated that administration of MSCs after experimental stroke results in smaller infarct volumes and improved functional recovery (Chen et al. 2001, 2003a; Kurozumi et al. 2005; Nomura et al. 2005; Zhao et al. 2006). However, the regenerative potential of MSCs is usually temporally restricted because MSCs tend to disappear quickly after transplantation—usually within a few days (Lee et al. 2009). The paradox of low cell engraftment after intravenous MSC delivery and improved tissue repair redirected attention to paracrine mechanisms, by which MSCs may facilitate tissue protection (Lee et al. 2009). Today, it is widely accepted that MSCs enhance tissue regeneration by the secretion of growth factors and cytokines (Wagner et al. 2007), including glial-derived neurotrophic factor (GDNF), nerve growth factor (NGF), and vascular endothelial growth factor (VEGF) (Chen et al. 2002). Moreover, MSCs secrete cytokines like interferon-gamma, tumor necrosis factor-alpha, transforming growth factor-beta (TGF-β), and interleukin (IL)-6 (Karnoub et al. 2007). The trophic factors secreted by MSCs prevent apoptosis and enhance cell survival in damaged tissues (Caplan and Dennis, 2006; Parekkadan et al. 2007). In addition, MSCs can activate tissue self-repair by stimulating mitosis and differentiation of endogenous stem cell populations (Munoz et al. 2005). Importantly, intracerebral MSC administration after an ischemic event was found to improve neovascularization of damaged tissue (Chen et al. 2003b), which is tightly linked to adult neurogenesis. MSCs also have immunomodulatory and immunosuppressive capacities (Maitra et al. 2004; Gerdoni et al. 2007), which may explain some of their therapeutic effects in immune-mediated disorders like multiple sclerosis (Zappia et al. 2005).

The paracrine functions of MSCs in cerebral ischemia remain incompletely understood, although MSCs are already administered to stroke patients in clinical trials. The aim of this study was to identify the molecular mechanisms mediating neuroprotection by bone marrow-derived MSCs.

Experimental Procedures

Cell Culture

Mesenchymal Stromal Cells

Murine MSCs (mMSCs) were obtained from the bone marrow of tibias and femurs of C57BL/6 mice aged 8–12 weeks (BfR, Berlin, Germany). mMSC isolation and culture techniques, characterization by cell surface epitope expression and differentiation assays into fat, bone, and cartilage are described in Scheibe et al. (2011).

Human MSCs (hMSCs) were kindly provided by the laboratory of Dr. D. Prockop, Tulane University, Center for Gene Therapy, New Orleans, LA, USA. hMSCs were expanded and characterized as previously described (Colter et al. 2000; Sekiya et al. 2002).

Neural Stem Cells (NSCs)

Subventricular region embodying the lateral ventricles was obtained from 2-mm thick brains slices derived from male C57BL/6 mice aged 8–10 weeks. A thin layer surrounding the ventricles was prepared, cut into small pieces, and incubated for 30–60 min in a papain-DNase-solution (47.2 mg papain (Cellsystems, Troisdorf, Germany), 9 mg cysteine (Sigma, Schnelldorf, Germany), 9 mg EDTA (Sigma) in 50 ml EBSS (Invitrogen, Darmstadt, Germany) at 37°C. Cells were pelleted by centrifugation at 110×g for 10 min. Tissue was dissociated in an ovomucoid solution (0.7 mg/ml ovomucoid (Sigma) in NBM-A, 2% B27 w/o retinoic acid, 1% l-glutamine; Invitrogen). Single cells were centrifuged again at 110×g for 10 min and resuspended in growth medium [NBM-A, 2% B27 w/o retinoic acid, 1% l-glutamine, 10 ng/ml EGF, 20 ng/ml β-FGF (both from Biochrom AG, Berlin, Germany)]. Cells were seeded in 25 cm² flasks at a density of 4,000 cells per cm² to obtain neurospheres. After the first splitting of neurospheres, cells were cultivated in low-attachment flasks (Corning, VWR International, Darmstadt, Germany). Experiments were performed with cells from passages 4–5.

Primary Neuronal Cells

Primary rat cortical neurons were prepared from the cerebral cortex of Wistar rats at embryonic day 17. Animals were purchased from FEM (Charité, Berlin, Germany), and all media and supplements were from Biochrom AG if not otherwise noted. Neuronal cultures were prepared according to modified protocols from Brewer (1995) and Lautenschlager et al. (2000). After removal of the meninges, cerebral cortices were dissected and incubated with 0.5%/0.2% (w/v) trypsin/EDTA in a water bath for 15 min at 37°C, rinsed twice with PBS, and once with dissociation medium (consisting of modified Eagle’s medium (MEM) with 10% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM l-glutamine, 10 mM HEPES, 44 mM glucose, 100 IE/l insulin). Thereafter, they were dissociated with a Pasteur pipette in dissociation medium, pelleted by centrifugation at 210×g for 2 min at room temperature (RT), redissociated in starter medium (NBM supplemented with B27 (both from Invitrogen), 100 U/ml penicillin, 100 μg/ml streptomycin, 0.5 mM l-glutamine, 25 μM glutamate) and were seeded into 24-well plates at a density of 200,000 cells/cm². Wells were coated with poly-l-lysine (0.5% w/v in PBS) for 1 h at RT, rinsed with PBS, followed by incubation with coating medium (MEM, 5% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, 10 mM HEPES, and 1% w/v collagen G) for 1 h at 37°C in the incubator. Wells were rinsed with PBS and filled with 500 μl/well starter medium. Cells were added in a volume of 100 μl/well and cultures were kept at 37°C and 5% CO₂. Feeding began after 4 days in vitro (div) with neuronal cell culture medium (starter medium without glutamate) by replacing half of the medium twice a week.

Oxygen–Glucose Deprivation (OGD)

Induction of Neuroprotection by MSC Co-culture or MSC-Conditioned Medium (CM)

Neuronal cultures were kept under the conditions described above until a medium change was performed after 8 div. 300 μl medium were removed from each well and 250 μl of fresh NBM + B27 supplemented with 4% FCS (end concentration of 2% FCS/well) were added to each well. Finally, each well contained approximately 500 μl as total volume. In all experiments, four wells of a 24-well plate were used for each approach and each plate contained a control group to exclude plate to plate variability.

For investigation of MSC-mediated neuroprotective effects, rat cortical neurons were treated either with CM from mMSCs or hMSCs that was added in varying concentrations (0.1, 0.25, 0.5, 1, and 5%) and at different time points (−48 h, −30 min, immediately post-OGD) to the neurons. Unconditioned medium consisting of NBM + B27, 100 U/ml penicillin, 100 μg/ml streptomycin, 0.5 mM l-glutamine, and 2% FCS served as negative control. To generate MSC supernatants, mMSCs and hMSCs were allowed to grow to 70–80% confluency in 145 cm² petri dishes in their regular growth medium. After rinsing with PBS, 25 ml medium consisting of NBM + B27, 100 U/ml penicillin, 100 μg/ml streptomycin, 0.5 mM l-glutamine, and 2% FCS were added to each dish. After 24 h, supernatants from MSC cultures were collected, debris was removed by rinsing through a 0.2-μm filter system, and aliquots of MSC supernatants were frozen at −20°C.

In another approach, cortical neurons were directly pre-conditioned by co-culture with MSCs for 48 h. In a transwell co-culture system, rat cortical neurons were located in the bottom well and co-culture was initiated by seeding of either 10,000 mMSCs or 1,000 hMSCs into inserts. Varying MSC cell numbers resulted from different proliferation characteristics of hMSCs and mMSCs in the inserts, resulting in distinct factor secretion with different dose–effect curves. In general, the cells were seeded in 150 μl medium into 0.4-μm pore size inserts (Falcon; BD Biosciences) with a polyethylene terephthalate (PET) membrane. The membrane was permeable to soluble factors, but prevented cell-to-cell contact between the compartments. 100 μl of medium were added to the bottom well in the transwell experiments (total volume of 250 μl was added to each well).

Blocking Experiments

To investigate molecular pathways involved in MSC-mediated neuroprotection, the same pre-conditioning regimen was applied to neuronal cultures as described above. To inhibit neuroprotective effects of 1% hMSC- or 1% mMSC-CM, the respective CM with or without 1 μM Akt inhibitor X (AktX; Merck, Darmstadt, Germany) were added to neuronal cultures 48 h prior to OGD.

To block soluble factors secreted by hMSCs, 2 μg/ml plasminogen activator inhibitor (PAI)-1 antibody (R&D systems, Wiesbaden-Nordenstadt, Germany) were added to neuronal cultures either with or without 1% hMSC-CM at 48 h prior to OGD. Application of vehicle or 1% hMSC-CM served as negative or positive control in these experiments.

OGD Experiments

Primary rat cortical neurons were used for OGD experiments after a cultivation period of 10 div. Before OGD, inserts with MSCs and medium from all wells were removed, wells were rinsed once with 300 μl PBS, and OGD was started by addition of 500 μl/well of a deoxygenated aglycemic solution (DAS: 143.8 mM Na⁺, 5.5 mM K⁺, 1.8 mM Ca²⁺, 1.8 mM Mg²⁺, 125.3 mM Cl⁻, 26.2 mM HCO³⁻, 1.0 mM PO₄ ³⁻, 0.8 mM SO₄ ²⁻; pH 7.4) in an anoxic atmosphere (0% O₂). Anoxia was generated in a humidified and gas-tight incubator (Concept 400, Ruskinn Technologies, Bridgend, UK) at 37°C that was flushed with a gas mixture consisting of 5% CO₂, 85% N₂, and 10% H₂. OGD lasted for the duration of 90–120 min. In control experiments, cells were washed with 300 μl PBS and were then exposed to 500 μl basic salt solution (BSS: 143.8 mM Na⁺, 5.5 mM K⁺, 1.8 mM Ca²⁺, 1.8 mM Mg²⁺, 125.3 mM Cl⁻, 26.2 mM HCO³⁻, 1.0 mM PO₄ ³⁻, 0.8 mM SO₄ ²⁻, 4.5 mg/l glucose, 2% FCS; pH 7.4) in a normoxic atmosphere with 5% CO₂ at 37°C. Immediately after OGD and in control experiments, medium was removed and replaced with 400 μl neuronal cell culture medium (50% fresh NBM + B27 and 50% conditioned NBM + B27 that was collected from neuronal cell cultures after 8 div) with or without MSC-CM.

Lactate Dehydrogenase (LDH) Assay

LDH release was measured as described previously (Bruer et al. 1997). 24 h after OGD, the LDH activity was detected in the supernatant of OGD-treated and control cultures to assess neuronal cell damage. Then, 20 μl of a 10% Triton X-100 solution were administered to each well to initiate full kill of cells for quantification of the maximum LDH release. Cells were incubated for 20 min at 37°C in an incubator and LDH values were measured again. LDH values of OGD-treated and BSS-stimulated cultures were divided by full kill LDH values to calculate cytotoxicity. Afterward, cytotoxicity values of control cultures were subtracted as background from OGD-treated cultures. Normoxic BSS stimulation was set to 0% cell damage and OGD stimulation to 100% cell death, respectively.

Ethidium Bromide (EB) and Acridine Orange (AO) Staining

Twenty-four hours after OGD, 2 μg/ml AO and 2 μg/ml EB (both from Sigma) were added to neuronal cultures to quantify apoptotic and necrotic cell death of primary neurons. After 5 min incubation, dead cells were counted in three wells and five visual fields (400× magnification) using a standard fluorescence microscope (Leica DM-RA).

Western Blotting

Neuronal cells with or without exposure to MSC-CM were harvested after 5, 15, or 30 min in cell lysis buffer (New England Biolabs, Frankfurt am Main, Germany) supplemented with complete mini protease inhibitor cocktail (New England Biolabs) and 1 mM phenylmethylsulfonylfluoride (PMSF; Sigma). After incubation on ice for 10 min and cell scraping, lysates were centrifuged at 14,000×g at 4°C for 5 min. Protein-containing supernatants were diluted in SDS sample buffer (end concentrations: 62.5 mM Tris–HCl, pH 6.8, 0.5% SDS, 10% glycerol, 5% β-mercaptoethanol, 0.004% bromphenol blue; all from Sigma) and boiled for 5 min before samples were stored at −80°C. Protein concentrations were determined by BCA assay (Pierce, Thermo Fisher Scientific, Bonn, Germany). Approximately, 20 μg of protein/lane were separated on SDS polyacrylamide gels by electrophoresis. Proteins were blotted onto nitrocellulose membranes (Roth, Karlsruhe, Germany). Blocking of membranes was carried out with TBST (10 mM Tris, 0.05% Tween 20, 150 mM NaCl, pH 7.6; all from Sigma) and 5% BSA for 1 h at RT. After rinsing with TBST, membranes were incubated with primary antibodies in TBST and 5% BSA overnight at 4°C. The following primary antibodies were applied: p42/44 MAPK antibody (1:1,000), pAkt (Thr308) antibody (1:2,000), pan-Akt antibody (1:1,000), pSTAT3 (Tyr 705) antibody (1:1,000; all from Cell Signaling, Frankfurt am Main, Germany), and β-actin antibody (1:1,000; Santa Cruz, Heidelberg, Germany). The next day, nitrocellulose membranes were washed with TBST and a secondary HRP-linked anti-rabbit antibody (1:1,000; Amersham Biosciences, Freiburg, Germany) or HRP-linked anti-goat antibody (1:10,000; Santa Cruz) was added at RT for 2 h. Signals on membranes were visualized on X-ray films (Kodak, Amersham Biosciences) after incubation with a chemiluminescence kit (Santa Cruz). All experiments were repeated at least three times.

Protein Identification and Mass Spectrometry (MS)

For preparation of hMSC supernatants, hMSCs were cultured in their regular proliferation medium until 80% confluency. Thereafter, cells were rinsed with PBS thrice and serum-free proliferation medium was added for 72 h. Collected hMSC-CM was filtered through a 0.2-μm filter and frozen at −20°C. Thawed samples were concentrated with 4 ml 10 kDa MWCO Amico filters (Millipore, Billerica, MA). For protein identification by MS, 15 μl concentrated extract was separated by SDS-PAGE and stained using a MS-compatible silver staining protocol (Zabel and Klose 2009). Protein bands were excised from gels and subjected to in-gel tryptic digestion. Tryptic fragments were analyzed by nanoflow high-performance liquid chromatography (nanoHPLC; Proxeon Easy-nLC, Denmark, Odense)/electrospray ionization (ESI)-MS and ESI-MS/MS on a LCQ Deca XP ion trap instrument (Thermo Finnigan, Waltham, MA, USA). Nano-HPLC was directly coupled to ESI-MS analysis. Protein bands eluates of 18 μl were loaded onto a SSPE traps C18 pre-column (5 μm, 120 Å, 100 μm I.D. × 20 mm); NanoSeparations, Nieuwkoop, Netherlands) using 0.1% v/v trifluoroacetic acid at a flow rate of 20 μl/min. Peptides were separated onto an analytical C18 columns (5 μm, 120 Å, 75 μm I.D. × 10 cm). The elution gradient was created by mixing 0.1% v/v formic acid in water (solvent A) and 0.1% v/v formic acid in acetonitrile (solvent B) and run at a flow rate of 200 nl/min. The gradient was started at 5% v/v solvent B and increased linearly up to 50% v/v solvent B after 40 min. ESI-MS data acquisition was performed throughout the LC run. Three scan events: (i) full scan, (ii) zoom scan of most intense ion in full scan, and (iii) MS/MS scan of the most intense ion in full scan were applied sequentially. No MS/MS scan on single charged ions was performed. Raw data were extracted by the TurboSEQUEST algorithm, and trypsin autolytic fragments and known keratin peptides were subsequently filtered. All DTA files generated by BioWorks version 3.2 (Thermo Scientific, Waltham, MA, USA) were merged and converted to MASCOT generic format files (MGF). Mass spectra were analyzed using our in-house MASCOT software package license version 2.2 automatically searching the SwissProt database for Homo sapiens (human) (SwissProt 51.8, 513,877 sequences). MS/MS ion search was performed with this set of parameters: (i) taxonomy: H. sapiens (human), (ii) proteolytic enzyme: trypsin, (iii) maximum of accepted missed cleavages: 1, (iv) mass value: monoisotopic, (v) peptide mass tolerance 0.8 Da, (vi) fragment mass tolerance: 0.8 Da, and (vii) variable modifications: oxidation of methionine and acrylamide adducts (propionamide) on cysteine. No fixed modifications were considered. Only proteins with scores corresponding to P < 0.05, with at least two independent peptides identified were considered. The cut-off score for individual peptides using ESI identification was equivalent to P < 0.05 for each peptide and usually in a MOWSE score >31. This number was calculated by the MASCOT software.

Immunocytochemistry

Neuronal cultures grown on coverslips were fixed after 10 div with 4% paraformaldehyde (PFA) for 15 min at RT. Blocking and permeabilization were carried out in 5% normal donkey serum (NDS), 1% BSA, and 0.3% Triton X-100 (Sigma) in PBS. Primary antibodies against MAP2 (1:1,000; Sigma) GFAP (1:1,000; Dako, Hamburg, Germany), Iba1 (1:2,000; Wako, Neuss, Germany), and CNPase (1:500; Covance, Munich, Germany) were incubated in a PBS solution with 3% NDS at 4°C overnight. The following secondary antibodies (all from Invitrogen) were incubated at a 1:800 dilution in PBS with 3% NDS for 1 h at RT: Alexa 594 donkey anti-mouse, Alexa 488 donkey anti-rabbit, and Alexa 594 donkey anti-rat. Cells were counterstained with DAPI before mounting. Omission of primary antibodies served as negative controls. Neuronal cultures were examined by fluorescence microscopy using a Leica microscope (DM-RA). For each group and experiment, cells of three coverslips in 10 visual fields were counted at 100× magnification to quantify neuronal cell culture composition of the respective treatment group.

Statistical Analysis

For statistical analysis of OGD data, Kruskal–Wallis one-way ANOVA on ranks test with Dunn’s method as post-hoc test or one-way ANOVA with Bonferroni’s post-hoc test was used. Immunocytochemistry data were tested for statistical significance by ANOVA with Bonferroni’s post-hoc test. P values <0.05 were considered as statistically significant. Results are expressed as means + standard deviation (SD).

Results

MSCs Rescue Cortical Neurons from OGD-Induced Cell Death

For neuroprotection studies of MSCs and their conditioned media, 90–120 min OGD experiments were performed to mimic the pathophysiological conditions of cerebral ischemia in vitro. Figure 1a illustrates the experimental setup.

Fig. 1.

Transwell co-culture with MSCs increases neuronal cell survival after OGD. a Illustration of the experimental setup. b Co-culture of neuronal cells with hMSCs or mMSCs in a two-compartment system for 48 h resulted in neuroprotection as shown by the significant reduction of LDH release (n = 5). c Co-culture of neuronal cells with murine NSCs failed to confer neuroprotection (n = 3). **P < 0.001

Pre-conditioning Experiments

In order to prevent cell-to-cell contacts while allowing the diffusion of soluble factors, hMSCs, or mMSCs were co-cultured with cortical neurons in a transwell system for 48 h prior to OGD. This resulted in a significant reduction of OGD-induced neuronal cell death by 30–35% as determined by LDH assay (Fig. 1b). In order to investigate whether the neuroprotective effects are specific for MSCs or can also be conferred by other stem cell populations, murine NSCs were co-cultured with neurons prior to OGD. However, NSCs did not provide any protection from OGD at up to twice the concentration of MSCs (Fig. 1c).

Given the paracrine actions of MSCs, media conditioned by hMSCs and mMSCs were added to neuronal cultures at different time points prior to OGD (−48 h, −30 min) at varying concentrations (0.1–5%). When CM was added to neurons 48 h prior to OGD, dose-dependent neuroprotective effects were observed. Low concentrations of hMSC- and mMSC-CM (0.1%) failed to prevent neuronal cell death after OGD, whereas higher concentrations of hMSC-CM (0.25–5%) and mMSC-CM (1–5%) conferred significant reductions of LDH release by 29–36% (Fig. 2a). The degree of neuroprotection achieved with MSC-CM was comparable to the results of direct co-culture experiments (Fig. 1b). Addition of hMSC-CM (0.1–5%) and mMSC-CM (1–5%) to cortical neurons only 30 min prior to OGD resulted in reductions of LDH release by 23–25% (Fig. 2b).

Fig. 2.

MSC-conditioned media exhibit neuroprotective effects. a Addition of varying concentrations (0.1–5%) of hMSC-CM (left panel) or mMSC-CM (right panel) to neuronal cultures at 48 h prior to OGD protected neurons from cell death (n = 5). b Significant reduction of LDH release was also observed after addition of hMSC- and mMSC-CM at 30 min prior to OGD (n = 5). c Post-OGD treatment with hMSC- or mMSC-CM contributed to increased neuronal survival after OGD (n = 5). *P < 0.05, **P < 0.001

Post-OGD Treatment with MSC-Conditioned Media

In order to determine whether MSC-CM also protect neurons from cell death when given after OGD, varying concentrations of hMSC- and mMSC-CM were added to neuronal cultures immediately after OGD. hMSC-CM and mMSC-CM (0.25–1%) reduced neuronal cell death by 20–25% post-OGD (Fig. 2c).

MSCs and MSC-CM do not Alter the Cellular Composition of Neuronal Cultures

The primary cultures used in this study consisted of approximately 89% MAP2-immunoreactive neurons, 10% GFAP-immunoreactive astrocytes, <0.3% CNPase-immunoreactive oligodendrocytes, and <0.05% Iba1-immunoreactive microglia (data not shown). The composition of primary neuronal cell cultures was neither altered by the co-culture with hMSCs and mMSCs nor by the addition of hMSC-CM and mMSC-CM (Fig. 3a). Thus, the reduction of LDH release in MSC- or MSC-CM-treated groups did not result from altered neuronal cell culture composition with a modified LDH release pattern.

Fig. 3.

Neuroprotection by MSC-conditioned media is mediated by an anti-apoptotic mechanism. a Immunocytochemical characterization of neuronal cell cultures revealed that transwell co-culture with hMSCs or mMSCs for 48 h did not affect the relative percentages of neurons and glial cells in the cultures, nor did the addition of MSC-CM for 48 h. b AO and EB staining was used to distinguish between apoptotic and necrotic cell death of neurons after OGD. Addition of hMSC- and mMSC-CM at 48 h prior to OGD or post-OGD resulted in a remarkable reduction of apoptotic neurons, whereas necrotic neuronal cell death was not affected (n = 6)

MSC-Conditioned Media Protect Neurons from OGD-Induced Apoptotic Cell Death

AO and EB staining revealed that apoptosis is the predominant cell death mechanism during and after OGD, whereas necrosis only accounts for less than 5% of OGD-induced cell death (Fig. 3b). Treatment of neurons with hMSC-CM and mMSC-CM at 48 h prior to OGD or immediately post-OGD resulted in significant reductions of the rate of apoptosis, whereas the rate of necrosis was unaffected (Fig. 3b), suggesting that MSCs protect neurons from OGD-induced apoptotic cell death.

Molecular Mechanisms of MSC-Induced Neuroprotection

Western blot studies were performed to investigate the downstream molecular mechanisms involved in MSC-mediated neuroprotection. Treatment of neuronal cultures with hMSC-CM and mMSC-CM for 5–30 min resulted in increased levels of pSTAT3 (Tyr705) and pAkt (Thr308) in neurons (Fig. 4a). In contrast, p42/44 MAPK levels remained unchanged (Fig. 4a). The data suggest that soluble factors in MSC-CM activate intracellular signaling cascades in neurons, which may be involved in anti-apoptotic effects. To further elucidate the role of Akt in MSC-mediated neuroprotection, neuronal cultures were pre-treated with hMSC-CM and mMSC-CM (1%) in the presence or absence of 1 μM AktX for 48 h prior to OGD. AktX completely abrogated the anti-apoptotic effects of MSC-CM (Fig. 4b), suggesting that PI3K/Akt is a major pathway for MSC-mediated neuroprotection.

Fig. 4.

Neuroprotection by MSC-conditioned media involves the Akt pathway and PAI-1. a Western blot studies indicated an upregulation of pSTAT3 (Tyr 705) and pAkt (Thr 308) in neuronal cell lysates after exposure of primary neurons to mMSC- or hMSC-CM for 5–30 min. In contrast, the p42/44 MAPK pathway was not activated by MSC-CM. β-actin and Akt (pan) served as internal controls. b Blocking of the Akt pathway with AktX (1 μM) abolished the neuroprotective effects of 1% hMSC-CM and 1% mMSC-CM (pre-conditioning time: 48 h, n = 5–6). c Blocking of PAI-1 by addition of 2 μg/ml anti-PAI-1 antibody partly reverted the neuroprotective effect of 1% hMSC-CM at 48 h prior to OGD (n = 5). *P < 0.05, **P < 0.001

Next, we tried to analyze the secretome of hMSCs by MS to identify putative factors, which mediate neuroprotection. Only factors that reached statistical significance in the hMSC proteomic study are summarized in Table 1. Twelve proteins were identified by MS, of which PAI-1 was considered an interesting candidate since it is known to prevent apoptosis though the Akt pathway. We also detected cytokines and growth factors like VEGF, TGF-β, macrophage colony-stimulating factor (M-CSF), and insulin-like growth factor-binding proteins (IGFBPs) in hMSC-CM by antibody arrays (data not shown). As proof-of-concept, 2 μg/ml anti-PAI-1 antibody was added to neuronal cultures at 48 h prior to OGD, which partly reverted the neuroprotective effect of hMSC-CM (Fig. 4c).

Table 1.

Proteomic characterization of human MSC-conditioned media

Protein	ID	MW (Da)	Function
Collagen alpha-1 (I) chain	P02452	138,827	ECM component
Collagen alpha-2 (I) chain	P08123	129,209	ECM component
Chitinase-3-like protein 1	P36222	42,598	ECM component
Hyaluronan and proteoglycan link protein 1	P10915	40,140	ECM component
Basement membrane-specific heparan sulfate proteoglycan core protein	P98160	468,501	ECM component
72 kDa type IV collagenase (MMP2)	P08253	73,835	Protease, collagen metabolism
Metalloproteinase inhibitor 2 (TIMP2)	P16035	24,383	Protease, metalloproteinase inhibiting activity
Plasminogen activator inhibitor 1 (PAI-1)	P05121	45,031	Protease, fibrinolysis inhibitor
Vimentin	P08670	53,619	Cytoskeletal component, cell growth
Profilin-1	P07737	15,045	Cytoskeletal component, cell growth
Golgin subfamily A member	Q13439	260,980	Intracellular transport protein
Serum albumin	P02768	69,321	Oncotic pressure in blood, transport protein
Secreted protein acidic cysteine rich glycoprotein (SPARC)	P09486	34,610	Interaction with ECM components, inhibition of cell cycle, cell adhesion

Discussion

Here, we show that human and murine MSCs rescue cortical neurons from apoptotic cell death in an in vitro model of cerebral ischemia. The neuroprotective effects are specific for MSCs and are mediated by paracrine factors, like PAI-1.

Our data are in line with previous in vitro reports, suggesting that astrocytes and oligodendrocytes are rescued from cell death after OGD by transwell co-culture with bone marrow stromal cells (Gao et al. 2008; Zhang et al. 2008). Moreover, intravenously administered MSCs reduced apoptosis and improved functional recovery after stroke in vivo (Chen et al. 2003a). MSCs were found to protect irradiated fibroblasts from apoptosis (Block et al. 2009). However, MSC-CM did not prevent apoptosis in this model, and MSC-mediated activation of fibroblasts by direct cell-to-cell contact was required to induce production of protective factors (Block et al. 2009). In our OGD model, robust neuroprotection was achieved by addition of diluted MSC-CM at a concentration range of 0.25–5%. Incubation with MSC-CM at concentrations higher than 10% yielded toxic effects on neuronal cultures, as did higher MSC numbers (>5,000 for hMSCs and >1,000 for mMSCs, respectively) in transwell co-culture experiments (data not shown). We performed all experiments in the non-toxic range as demonstrated by phase-contrast microscopy and LDH measurements of the cultures. Our results are in line with previous observations that pre-treatment of organotypic hippocampal slice-cultures with undiluted or 1:2-diluted MSC supernatants induced neurotoxicity after OGD (Horn et al. 2009). We further observed a dose-dependent neuroprotection by 0.25–5% MSC-CM when added to cortical neurons 48 h prior to OGD. Neuroprotection was achieved with even lower concentrations of MSC-CM (0.1%) when added 30 min prior to OGD, but the degree of neuroprotection was generally lower at 30 min even with high concentrations of MSC-CM. This may suggest a combination of direct neuroprotective effects of MSC-CM subject to protein degradation, and delayed neuroprotection by glial cells or secondary induction of gene expression and protein translation in neurons.

Several previous studies aimed to identify the factors and mechanisms by which MSCs mediate not only their cyto- and neuroprotective effects but also their immunomodulatory and angiogenic functions. One secretome study using human adipose tissue-derived MSCs revealed interesting molecular candidates, e.g., anti-inflammatory molecules (follistatin-like 1, pentraxin-related gene), but also antioxidants (gluthatione-S-transferase P, peroxiredoxin 6, thioredoxine) (Chiellini et al. 2008). Other proteomic studies found that MSCs secrete a variety of cytokines, chemokines, growth factors, and their receptors (Wagner et al. 2007). One study reported that different isolation and culture techniques of MSCs, as well as different species and organ origins may explain the variability observed in the secretory profiles of MSCs (Wagner et al. 2007). We identified only 12 factors that met the criteria of statistical significance for MS. The low number of detected molecules may result from technical limitations of MS and the serum-free culture conditions of MSCs, which were required for precise protein identification.

The molecular complexity of MSC-CM suggests that protection of cortical neurons after OGD may be mediated by different mechanisms: (1) by direct effects of MSC-CM on neurons, (2) indirectly by soluble MSC-derived factors that stimulate contaminating astrocytes (~10%) to secrete neuroprotective factors, and (3) by delayed conditioning mechanisms in neurons. Besides proteins like cytokines and growth factors, smaller molecules such as lipids, peptides, or antioxidants are also putative candidates for MSC-CM-mediated neuroprotection.

In our proteomic MS approach, we identified PAI-1 (~45 kDa) as a putative candidate for MSC-mediated neuroprotection. Previous studies observed that PAI-1 produced by astrocytes in mixed neuronal/astrocytic cultures protected neurons against N-methyl-d-aspartate (NMDA) receptor-mediated excitotoxicity by modulating the NMDA-evoked calcium influx (Docagne et al. 2002; Gabriel et al. 2003). In contrast, a recent study observed that astrocytes increase tissue plasminogen activator (tPA) activity and downregulate PAI-1 in response to MSCs or MSC-CM in the OGD model (Xin et al. 2010). We found that application of blocking antibodies against PAI-1 significantly reduced MSC-mediated neuroprotection after OGD. The inhibition was not complete, which supports our assumption that neuroprotection by MSCs is likely mediated by a combination of different paracrine factors secreted by MSCs. Along these lines, we also detected SPARC/osteonectin in hMSC-CM, which has been shown to provide protection from apoptosis via an Akt-dependent pathway (Shi et al. 2004; Chang et al. 2010).

On the level of intracellular signal transduction, we observed that MSC-CM increased the phosphorylation of STAT3 and Akt in neurons. This is in line with findings that human aortic endothelial cells are protected from hypoxia-induced apoptotic cell death by MSC supernatants after activation of PI3K/Akt and STAT3 signaling cascades (Hung et al. 2007). In this study, the authors found relevant levels of VEGF, MCP-1, and IL-6 in MSC-CM. However, experiments with neutralizing antibodies against VEGF, MCP-1, and IL-6 failed to confirm that these factors were involved in the prevention of apoptotic cell death by MSC-CM (Hung et al. 2007). Moreover, treatment of cultures with MSC-CM also increased the levels of pERK, but an inhibitor of the ERK1/2 pathway had no effect on hypoxia-induced apoptosis (Hung et al. 2007). In another OGD model, oligodendrocytes were protected from apoptosis by the activation of Akt and p75 (Zhang et al. 2008). These data are in line with our results, since we found that the Akt pathway is one of the major mechanisms that underlie the anti-apoptotic effects of MSC-CM. It remains to be resolved whether activation of the STAT3 and PI3K/Akt transduction pathways interrelate to provide neuroprotection. Thus, a recent study demonstrated that cytokines such as IL-6 have anti-apoptotic effects in models of NMDA-induced excitotoxicity by combined activation of the STAT3 and PI3K/Akt signaling pathways (Liu et al. 2011). IL-10 was also found to reduce the apoptosis of cortical neurons after OGD by up-regulation of phosphorylated STAT3 and Akt (Sharma et al. 2011).

In conclusion, our data suggest that MSCs provide neuroprotection by paracrine mechanisms. MSC-derived soluble factors either directly or indirectly activate STAT3- and Akt-dependent anti-apoptotic pathways in neurons that enhance survival after OGD.