C. Voermans, PhD, Department of Experimental Immunohaematology, Sanquin, Plesmanlaan 125, 1066CX Amsterdam, The Netherlands. E-mail: C.Voermans@sanquin.nl
Mesenchymal stromal cells (MSC) are potential cells for cellular therapies, in which the recruitment and migration of MSC towards injured tissue is crucial. Our data show that culture-expanded MSC from fetal lung and bone marrow, adult bone marrow and adipose tissue contained a small percentage of migrating cells in vitro, but the optimal stimulus was different. Overall, fetal lung-MSC had the highest migratory capacity. As fetal bone marrow-MSC had lower migratory potential than fetal lung-MSC, the tissue of origin may determine the migratory capacity of MSC. No additive effect in migration towards combined stimuli was observed, which suggests only one migratory MSC fraction. Interestingly, actin rearrangement and increased paxillin phosphorylation were observed in most MSC upon stromal cell-derived factor-1α or platelet-derived growth factor-BB stimulation, indicating that this mechanism involved in responding to migratory cues is not restricted to migratory MSC. Migratory MSC maintained differentiation and migration potential, and contained significantly less cells in S- and G2/M-phase than their non-migrating counterpart. In conclusion, our results suggest that MSC from various sources have different migratory capacities, depending on the tissue of origin. Similar to haematopoietic stem cells, cell cycle contributes to MSC migration, which offers perspectives for modulation of MSC to enhance efficacy of future cellular therapies.
Due to their multilineage differentiation and immunosuppressive capacities (Kopen et al, 1999; Glennie et al, 2005), MSC are increasingly being considered as a potential cell source for regenerative medicine and immune therapies (Prockop, 1997; Pittenger et al, 1999). When injected into irradiated mice, MSC seem to preferentially home to damaged tissue (Francois et al, 2006), although the observed frequency of engraftment was very low and many MSC were found trapped in the lungs after intravenous injection. Nevertheless, in pre-clinical and clinical settings, transplanted MSC have been shown to improve haematopoietic stem cell (HSC) engraftment (Koc et al, 2000; Noort et al, 2002), reduce clinical symptoms of osteogenesis imperfecta (Horwitz et al, 1999) and graft-versus-host disease (Le Blanc et al, 2004) and further stimulate the repair of pancreatic islets (Lee et al, 2006) and the infarcted myocardium (Chen et al, 2004; Kawada et al, 2004).The migration of MSC towards the site where they are required is crucial for most applications.
Many studies on leucocyte (Butcher, 1991; Springer, 1994) and HSC (Voermans et al, 1999, 2001) migration have provided insight into common mechanisms of migration. Chemokines, cytokines and growth factors released upon injury provide migratory cues for cells. They induce upregulation of selectins and integrins on the cell surface, enabling cells to interact with the endothelium. Cells subsequently adhere and transmigrate across the endothelial layer into tissues. The homing efficiency of expanded MSC in vivo is very low (Francois et al, 2006), and therefore high cell numbers are needed. Better insight in the migratory mechanisms in MSC is important in order to provide strategies for future, more efficient therapies.
The migratory capacity of bone marrow-derived MSC has been previously studied (Ozaki et al, 2007; Ponte et al, 2007). Stromal cell-derived factor (SDF)-1α (Bhakta et al, 2006; Son et al, 2006), platelet-derived growth factor (PDGF)-BB (Fiedler et al, 2002, 2004), hepatocyte growth factor (HGF) (Forte et al, 2006) and basic fibroblast growth factor (bFGF) (Schmidt et al, 2006) have been reported as migratory stimuli for these cells. For MSC derived from adipose tissue, SDF-1α has also been described as a potent chemokine (Sengenes et al, 2007). However, comparing the various studies on MSC migration is difficult because the MSC are often cultured under different conditions and the set up of migration assays varies. Moreover, the properties of the migratory cell fraction(s) and the underlying mechanism involved in MSC migration remain unclear.
This study evaluated the migratory potential of MSC derived from adult bone marrow, adult adipose tissue, fetal lung and fetal bone marrow, whilst performing the cell culture and migration experiments under identical conditions. In addition, the migratory MSC population and the molecular mechanism of migration were investigated. As ongoing clinical trials have indicated that large cell numbers need to be transplanted, the studies mentioned above and the experiments described in the current paper were performed using culture expanded MSC. Our results indicate that MSC derived from various tissues contain a cell fraction that is able to migrate, although the optimal stimulus and the percentage of migrating cells varied. The migratory MSC were observed to maintain differentiation and migration potential and they contained fewer cells in S- and G2/M-phase of the cell cycle as compared to non-migrating counterparts.
Materials and methods
Isolation and culture of MSC
Fetal lung and fetal bone were obtained after informed consent from legally terminated second trimester pregnancies. The protocol for collecting fetal tissues for research purposes was approved by the medical ethical review board of the Academic Medical Centre (AMC) (MEC: 03/038). Fetal lung MSC (FLMSC) were derived from magnetic bead selected CD34+ fetal lung cells, which were subsequently cultured in M199 containing 10% fetal calf serum (FCS), Penicillin streptomycin, endothelial cell growth factor (ECGF) and heparin as described by Noort et al (2002). To obtain fetal bone marrow MSC (FBMSC), fetal bones were flushed with Iscove’s modified Dulbecco’s medium (IMDM; Lonza, Verviers, Belgium) containing 10% FCS (Bodinco, Alkmaar, The Netherlands) and 1% penicillin-streptomycin (Gibco, Paisley, UK). The remaining erythrocytes in the cell suspension were lysed using NH4Cl for 10 min on ice. Subsequently, cells were rinsed in PBS. 1·6 × 106 cells were seeded per well in 6-well dishes in M199 (Gibco) supplemented with 10% FCS, 1% penicillin-streptomycin, 20 μg/ml ECGF (Roche diagnostics, Indianapolis, IN, USA) and 8 IU/ml heparin (Leo Pharma, Breda, The Netherlands). The obtained cells were considered to be FBMSC.
Adipose tissue derived MSC (ASC) were a kind gift from Dr FJ. van Milligen and were derived as previously described (Varma et al, 2007). Briefly, adipose tissue obtained from healthy donors was enzymatically digested. The obtained cell suspension was neutralized by Dulbecco’s modified Eagle medium (DMEM) containing glucose and FBS and then centrifuged. The cell pellet was resuspended in PBS and subjected to Ficoll density centrifugation. The cell-containing interface was harvested. 1 × 105 cells/cm2 were seeded in DMEM containing 10% FBS, penicillin streptomycin and l-glutamine (Varma et al, 2007).
Adult bone marrow MSC (BMSC) were isolated from bone marrow aspirates obtained after informed consent from the sternum of patients undergoing cardiac surgery, according to the protocol for collecting bone marrow for research purposes approved by the medical ethical review board of the AMC (MEC:04/042#04·17·370). Briefly, MSC were isolated by density gradient centrifugation (Ficoll-paque, 1·077 g/ml; GE Health care Bio-Sciences AB, Uppsala, Sweden). 5 × 106 cells were seeded per well in a 6-well dish in M199 containing 10% FCS, 1% penicillin-streptomycin, 20 μg/ml ECGF and 8 IU/ml heparin.
After 48 h, the non-adherent cells were removed. The remaining cells were cultured for an additional 12 d or until reaching 80–90% confluence.
Upon reaching 80–90% confluence after initial plating, MSC derived from all sources were replated and further cultured under identical conditions in T80 tissue culture flasks at an initial density of 2500 cells/cm2. For all experiments, 80–90% confluent passage 4 to passage 8 MSC were used.
All sources of MSC were characterized for surface expression of various receptors by flow cytometry. Cells were rinsed trypsinized, washed and resuspended in PBS containing 0·2% bovine serum albumin (BSA) prior to incubation (20 min at room temperature) with the following monoclonal antibodies. Antibodies purchased from BD, San Jose, CA, USA: CD73 (clone AD2), CD90 (clone 5E10), CD45 (clone HI30), CD14 (clone M5E2), CD68 (clone Y1/82A), CD44 (clone G44-26), CD49e (clone VC5), CD54 (clone HA58), CD106 (clone 51-10C9), CD146 (clone 8G12), CXCR4 (clone 12G5). Purchased from Sanquin, Amsterdam, The Netherlands: CD3 (clone CLB-T3/2), CD19 (clone CLB-B4/1), CD31 (clone CLB-HEC/75), CD38 (clone CLB-1D5), CD11a (clone CLB-LFA-1/2), CD11b (clone CLB-mon-gran/1), CD18 (clone CLB-LFA-1/1), CD49b (clone 10G11), CD49f (clone GoH3). Antibodies from other companies: CD271 (clone ME20.4-1.H4; Miltenyi Biotec, Gladbach, Germany), CD105 (clone SN6; Ancell, Bayport, MN, USA), CD34 (clone 581; IQ-products, Groningen, The Netherlands), CD29 (clone P4C10; Chemicon, Millipore, Billerica, MA, USA), CD49d (clone 44H6; Imgen, ITK diagnostics, Uithoorn, The Netherlands), CD166 (clone 3A6; RDI, Concord, MA,USA), CXCR7 (Clone 358426, R&D systems, Minneapolis, MN, USA), Santa Cruz biotechnology, Santa Cruz, CA, USA; PDGFRα (Clone 16A1) and PDGRβ (Clone P-20). Secondary antibodies; Goat anti mouse IgG (Dako, Glostrup, Denmark), goat anti rabbit IgG (Invitrogen molecular probes, Paisley, UK). As a negative control, cells were labelled with isotype control IgG1, IgG2, IgG2a monoclonal antibodies (Sanquin, BD). A minimum of 10 000 events was recorded, using a FACS LSR II flow cytometer (BD).
To study the multilineage differentiation capacity, MSC were cultured under conditions promoting differentiation towards osteoblasts or adipocytes as previously described (Noort et al, 2002). For differentiation experiments, MSC were plated in a 24-well dish at a plating density of 2·5 × 104 cells/cm2 in α-minimal essential medium (α-MEM; Gibco). For osteogenic induction, the α-MEM was supplemented with 10% FCS and penicillin-streptomycin to which ascorbic acid (50 μg/ml; Sigma, St Louis, MI, USA) and dexamethasone (10−7 mol/l; Sigma) were added. From day 7 onwards, β-glycero-phosphate (5 mmol/l; Sigma) was added. Cultures were incubated in a humidified atmosphere of 5% CO2 at 37°C. Medium was replaced every 4th and 7th day of the week. For induction of adipogenesis, indomethacine (50 μmol/l; MP Biomedicals, Solon, OH, USA), 3-Isobutyl-1-methylxanthine (IBMX, 0·5 mmol/l; Sigma) and insulin (1·6 μmol/l; Sigma) were used.
At day 21, the cells induced towards osteogenic differentiation were stained for alkaline phosphatase and calcium deposition. Cells were incubated with a substrate solution (0·2 mg/ml α-naphthyl-1-phosphate (Sigma), 3 mg/ml sodium borate, 0·3 mg/ml magnesium sulphate and 0·8 mg/ml fast blue RR acid (Sigma)) for 15 min, resulting in the formation of an insoluble purple reaction product. To detect calcium deposition, cells were fixed with 3·7% formaldehyde (Merck, Darmstadt, Germany) for 10 min, and stained with 2% Alizarin Red S (ICN Biomedicals, Aurora, OH, USA) and 0·1 NH4OH (pH 5·4) for 1 min. Mineralization was indicated by the presence of red depositions. To demonstrate the presence of adipocytes, expanded cells were fixed as described above. Cytoplasmic inclusions of neutral lipids were stained with Oil-Red-O (3 mg Oil-Red-O/ml 60% isopropanol; Sigma) for 10 min.
In vitro migration experiments
Migration experiments were performed using 12 μm pore size Transwell plates (Corning Costar, Cambridge, MA, USA). The upper side of the insert was coated overnight at 37°C with fibronectin (20 μg/ml; Sigma) or collagen I (50 μg/ml; BD Biosciences) in PBS. 100 000 cells were seeded into the upper compartment in 500 μl IMDM supplemented with 0·25% BSA, and the stimuli were added to the lower compartment in 1·5 ml IMDM with 0·25% BSA. Optimal concentrations for migration were determined. Stimuli evaluated were; SDF-1α (600 ng/ml; Strathmann, Hamburg, Germany/R&D systems/Peprotech, Rocky Hill, NJ, USA), bFGF (100 ng/ml; R&D systems,), PDGF-BB (5 ng/ml; R&D systems) and HGF (40 ng/ml; R&D systems/Peprotech). Checkerboard migration assays were performed by adding stimuli or combinations of stimuli in both upper and lower compartment. Treatment of cells and concentrations of stimuli were as described above.
After 4 h incubation at 37°C, the cells in the upper compartment were removed with a cotton swab. Subsequently, the inserts were carefully rinsed twice in PBS prior to fixation in 3·7% formaldehyde and further stained with Hoechst 33258 (1:500 dilution, Invitrogen). The Transwell filter membranes were cut out and mounted onto glass slides using Vectashield (Vector Laboratories, Burlingame, CA, USA). The total number of migrating cells per view field was counted using fluorescence microscopy by counting nuclei. Data were expressed as the percentage of migrating cells related to the total number of cells loaded into the upper compartment.
Ten thousand MSC were seeded on fibronectin-coated coverslips and cultured for 2 d. Subsequently, cells were put in serum-free IMDM supplemented with 0·25% BSA. MSC were treated with SDF-1α or growth factors as indicated or left untreated. The cells were fixed for 10 minutes on ice, then permeabilized using 0·2% Triton-X-100 (Sigma) and stained for paxillin (Clone 165/Paxillin; BD) and phospho-paxillin (pY31, Rabbit polyclonal; Sigma), followed by incubation with Alexa 488 phalloidin (Invitrogen Molecular Probes; 1 IU/ml) for F-actin staining, goat-anti-mouse Alexa 633 (20 μg/ml; Invitrogen Molecular probes) and goat-anti-rabbit Alexa 568 (20 μg/ml; Invitrogen Molecular probes). Subsequently, the cells were stained with Hoechst and coverslips were then mounted using Mowiol (Sigma). Immunofluorescent staining was detected using a LSM 510 META confocal microscope (Zeiss, Jena, Germany) using a 40× oil-objective. Images were captured by ZEN 2007 confocal software (Zeiss).
Cell lysis and Western blot
MSC were seeded in 6-well culture dishes at a density of 2500 cells/cm2 and grown up to 8090% confluence. Subsequently, cells were serum starved for 30 min and treated with SDF-1α, PDGF-BB or FCS as indicated or left untreated. Next, MSC were lysed in 250 μl nonidet P40 (NP40) buffer (50 mmol/l TRIS, 100 mmol/l NaCl, 10 mmol/l MgCl2, 1% NP40, 10% Glycerol, pH 7·4, containing protease and phosphatase inhibitors) for 10 min on ice. Lysates were clarified by centrifugation at 14 000 rpm at 4°C for 10 min. The supernatant was aspirated and further analysed by Western blot.
For Western blotting, protein samples were separated by electrophoresis using a 10% sodium dodecyl sulphate-polyacrylamide gel and transferred onto polyvinylidene membranes (Bio-Rad Laboratories, Hercules, CA, USA). Membranes were incubated with antibodies for paxillin (Mouse monoclonal, Clone 165/Paxillin) and phospho-paxillin (pY31, Rabbit polyclonal; Sigma, pY118, mouse monoclonal; BD Transduction laboratories, Franklin Lakes, NJ, USA) for 1 h in TBST (Tris-buffered saline, Tween 20) containing either 5% BSA (Sigma) for antibodies to detect phosphorylated proteins or 5% nonfat dry milk for others, followed by 45 min incubation with horseradish peroxidase-conjugated goat-anti-mouse (1:7000; Pierce, Rockford, IL, USA) or goat-anti-rabbit (1:5000; Dako) secondary antibodies. Immunoreactive bands were revealed using an enhanced chemiluminescence kit (ECL; Pierce).
Cell cycle analysis
MSC were seeded in T80 tissue culture flasks at an initial density of 2500 cells/cm2. Upon reaching 80–90% confluence, MSC were trypsinized and 100 000 cells were used for cell cycle analysis. The remaining MSC were allowed to migrate as described above. After migration, migrating and non-migrating MSC were harvested for cell cycle analysis. MSC were fixed in 70% ethanol on ice and subsequently incubated with Ki67 fluorescein isothiocyanate (Clone MIB-1; Dako) for 30 min at 4°C in PBS containing 0·1% Triton-X-100. Thereafter, the MSC were incubated for 15 min at 37°C with labelling reagent containing propidium iodide (1 μg/ml) or Hoechst 33 258 (1 μg/ml), RNase A (Sigma) and 0·1% Triton-X-100. A minimum of 10 000 cells was analysed by flow cytometry as described above and further analysed using Modfit LT 3.0 software (Verity Software House, Topsham, ME, USA). Ratios were calculated for each cell cycle phase from the percentage of cells in a certain phase in migrating MSC divided by the percentage of cells in the same phase in non-migrating MSC.
Statistical significance was determined by Mann–Whitney U-test, using the Statistical Package for the Social Sciences (spss) v.15.0 (SPSS Inc, Chicago, IL, USA), except for cell cycle data, which were analysed by One-Sample Kolmogorov–Smirnov test. Results were considered to be significant when P ≤ 0·05.
Characterisation of MSC
MSC obtained from fetal lung (FLMSC), fetal bone marrow (FBMSC), adult bone marrow (BMSC) and adipose tissue (ASC) all had a spindle shaped morphology, expressed the marker combination CD73, CD90 and CD105, and lacked expression of hematopoietic markers CD34 and CD45 (Fig S1A). All MSC sources also lacked expression of other hematopoietic markers (data not shown) in agreement with the definition of the ISCT (Dominici et al, 2006). Osteoblast and adipocyte differentiation was successfully induced in all MSC described. Based on morphology, FLMSC and FBMSC seemed to be less efficient in adipogenic differentiation (Fig S1B).
MSC derived from various tissues require extracellular matrix proteins for migration
In order to determine optimal conditions for in vitro migration experiments, FLMSC, BMSC- and ASC were allowed to migrate over 12 μm Transwell membranes coated with fibronectin, collagen I, or left untreated. Hardly any migration was observed for all MSC sources across bare filters, but coating with fibronectin or collagen significantly increased the percentage of spontaneous migration and SDF-1α-induced migration (Fig 1). These data indicate that MSC originating from three different tissues all require extracellular matrix proteins for in vitro migration, which is in agreement with previous observations (Ozaki et al, 2007; Thibault et al, 2007). Coating with fibronectin was used to study the migration dynamics of MSC in more detail.
The expression of integrins, required for binding to extracellular matrix proteins, was evaluated. None of the MSC sources expressed CD11a, CD11b or CD18 (data not shown), while homogenous expression of CD29 and CD49e (VLA-5) was detected for all MSC (Table I). Expression of CD49b (VLA-2), CD49d (VLA-4) and CD49f (VLA-6) was also detected in all sources. Overall, a higher expression of these integrins was observed in fetal-derived MSC (Table I). FBMSC had a significantly higher expression of CD49b and CD49d than BMSC. CD49b was significantly higher expressed on FLMSC as compared to BMSC and ASC. FLMSC also expressed CD49f to a significantly higher level than FBMSC, BMSC and ASC (Table I).
Table I. Integrin expression on MSC.
ASC (n = 4)
BMSC (n = 6)
FBMSC (n = 3)
FLMSC (n = 4)
The data represent the percentage (mean ± SD) of positive cells for the indicated integrin, determined by flow cytometry.
ASC, adipose tissue-derived MSC; BMSC, adult bone marrow-derived MSC; FBMSC, fetal bone marrow-derived MSC; FLMSC, fetal lung-derived MSC.
*Significant differences were observed between FBMSC and BMSC, P ≤ 0·05.
†Significant difference was observed between FLMSC, and ASC, BMSC, P ≤ 0·05.
‡Significant differences were observed between FLMSC and ASC, BMSC and FBMSC, P ≤ 0·05.
22·1 ± 18·8
15·8 ± 6·3
54·9 ± 17·8*
94·8 ± 2·6†
22·9 ± 24·4
4·3 ± 2·0
29·8 ± 2·9*
21·6 ± 15·8
99·8 ± 0·1
99·9 ± 0·1
100 ± 0
99·7 ± 0·2
8·4 ± 11·5
14·6 ± 14·9
15·8 ± 5·2
40·8 ± 5·2‡
Migratory capacity of MSC derived from various tissues towards different stimuli
To examine whether MSC originating from different tissues display similar migratory behaviour as compared to the extensively studied BMSC, the different MSC were allowed to migrate towards SDF-1α, PDGF-BB, HGF, bFGF or FCS. As demonstrated in Fig 2, FLMSC showed a significant enhanced migratory capacity towards SDF-1α when compared with BMSC and ASC, whereas ASC showed an increased migratory capacity towards FCS compared with FLMSC and BMSC. These data show that MSC originating from all tissues are able to migrate, although the percentage of migrating cells and the optimal stimulus differs among MSC sources. Overall, FLMSC showed the highest response to the given stimuli (Fig 2). To determine whether this enhanced migratory capacity was due to its fetal origin, the migratory capacity of adult and fetal bone marrow-derived MSC was compared. No significant differences in migratory potential were observed between adult and fetal BMSC towards any of the stimuli, whereas FLMSC had a significantly increased migratory potential when compared with FBMSC in four out of five stimuli evaluated (Fig 2): SDF-1α, PDGF-BB, HGF and bFGF. These observations suggest that the migratory potential of various MSC is determined by the tissue of origin rather than the maturity of the tissue they have been derived from. The differences in migratory potential of MSC from various tissues could not be explained by chemokine- or growth factor receptor-expression. Although slight differences were observed in surface expression of CXCR4, CXCR7, PDGFRα and PDGFRβ, high intracellular expression of these receptors was detected in all MSC (Table SI).
No additive effect in migration of FLMSC towards SDF-1α, PDGF-BB and FCS
As it was observed that only a small percentage of the cultured MSC were able to migrate towards the various stimuli, a checkerboard migration experiment was set up to elucidate whether an additive effect could be observed in migration of MSC towards a combination of stimuli. The experiments were performed using FLMSC because they possessed the highest migratory capacity. Optimal concentrations of stimuli were used. As depicted in Fig 3, no additive effect was observed between SDF-1α, PDGF-BB and FCS in migration of FLMSC, when both stimuli were put together in the lower compartment of the Transwell. Interestingly, the presence of SDF-1α in the upper compartment partly inhibited migration towards PDGF-BB in the lower compartment and PDGF-BB in the upper compartment partly blocked migration towards SDF-1α in the lower compartment (not significant). Together, these data suggest that one subset of MSC is able to migrate towards various stimuli.
Actin rearrangement and enhanced paxillin phosphorylation in stimulated MSC
As only a small percentage of expanded MSC was able to migrate, it was evaluated whether only these cells were able to rearrange the actin cytoskeleton and focal adhesions in response to migratory cues, which is required to enable cell migration (Mitchison & Cramer, 1996; Zigmond, 1996). Strikingly, morphological changes and altered F-actin distribution were observed in the majority (80%) of the BMSC after stimulation with PDGF-BB and SDF-1α (Fig 4A). MSC formed top ruffles after 5 min (Fig 4A). After 30 min, lateral membrane ruffles were observed (Fig 4A). Paxillin is a focal adhesion-associated adapter protein, and its phosphorylation at tyrosine residues Y31 and Y118 is associated with focal adhesion turnover and migration (Schaller, 2001; Iwasaki et al, 2002). After stimulation with FCS and SDF-1α for 60 min, increased levels of phosphorylated paxillin were observed at the periphery of the cells (Fig 4B). In addition, the effects of FCS and SDF-1α on the phosphorylation levels of paxillin on the total cell population were studied using Western blot. These data show that FCS and, to a lesser extent, SDF-1α increased paxillin phosphorylation (Fig 4C); PDGF-BB also increased paxillin phosphorylation. Similar results were obtained for FLMSC and ASC and were confirmed by pY118 staining (data not shown). These data indicate that, as in other cells, paxillin phosphorylation is enhanced upon stimulation with various migratory cues in MSC.
Migratory MSC maintain differentiation and migratory capacity
In order to study the characteristics of the migratory MSC in more detail, MSC were trypsinized from the Transwell membranes after migration and put in culture as described. Migratory and non-migratory MSC were observed to maintain their proliferation capacity and when seeded in osteogenic or adipogenic supporting medium 2 weeks after migration, migrating and non-migrating MSC originating from all sources were still able to differentiate (Fig 5A). As a control, MSC from the same passage that were not used for migration experiments were differentiated simultaneously. No variations in differentiation potential could be observed between migrating, non-migrating and the cultured MSC subset in any of the sources.
To evaluate whether the migratory MSC also maintained their migratory capacity, the migrating- and non-migrating MSC obtained from the first migration run were cultured for 2 weeks before being allowed to migrate again towards a gradient of SDF-1α for 4 h. FLMSC that had migrated towards SDF-1α in the first experiment had a significantly increased migratory potential (P ≤ 0·009) in the second experiment compared to FLMSC that did not migrate in the first run (Fig 5B). This indicates that the heterogeneous population of expanded MSC contains a subset of cells with a higher intrinsic migratory capacity.
To define these MSC, which would enable enriching for migratory MSC, surface expression of the following markers was evaluated: CD44, CD49b, CD49d, CD49f, CD54, CD73, CD106, CD146, CD166, CD271, CXCR4, CXCR7 and PDGFRα and PDGFRβ. None of these markers was exclusively, neither differentially expressed on migratory MSC (data not shown).
The migratory MSC fraction contains less cells in S- and G2/M-phase
Cell cycle has been identified to influence homing and repopulation of HSC (Glimm et al, 2000) and S- and G2/M-phase were found to negatively influence these processes. Thus it was evaluated whether the distribution of the phases of the cell cycle in migratory FBMSC was different from the distribution in non-migrating FBMSC. Immediately after migration, cell cycle analysis was performed for both fractions (Fig 6A). Ratios of migrating versus non-migrating MSC were calculated for each cell cycle phase (Fig 6B). It was observed that migrating MSC contained significantly less cells in S- (0·81 ± 0·13, P < 0·028) and G2/M-phase (0·75 ± 0·13, P < 0·031; Fig 6B). Expression of the Ki67 antigen, which enables discrimination between G0- and G1-phase, revealed a trend of more cells in G1-phase in migratory MSC (1·34 ± 0·50) as compared to non-migratory MSC, however this did not reach significance (data not shown). These data show that cell cycle is also involved MSC migration and that as in HSC, S- and G2/M-phase negatively influences migration.
Similar to BMSC, in all other MSC sources evaluated, only a small percentage of the MSC population was able to migrate towards the stimuli provided, although the percentage of migrating cells and the optimal stimulus differed between MSC sources. FLMSC had the highest migratory potential towards all specific stimuli as compared to the other sources tested. This is most probably not due to the fact that they are derived from fetal tissue, because FBMSC had a similar migratory capacity when compared with BMSC. These data suggest that the migratory capacity of culture-expanded MSC varies with the tissue of origin rather than the maturity of this tissue. The observation that MSC of various origin display differential responsiveness to growth factors and chemokines in vitro, also suggests that these MSC are possibly attracted by different chemokines and/or growth factor combinations released upon injury in vivo and this may be due to conservation of niche-induced characteristics. Similar suggestions have been made by others on differences between MSC sources in terms of colony frequency, differentiation and gene expression (Wagner et al, 2005; Kern et al, 2006; da Silva et al, 2006; Covas et al, 2008). As all MSC expressed the chemokine or growth factor receptors involved at similar levels, both at the cell membrane and intracellular, this could not explain differences in migratory potential. The large intracellular receptor pools indicate that MSC derived from all sources should be able to rapidly alter surface expression in response to stimuli, however the process of receptor cycling/trafficking may differ between MSC.
In vitro and in vivo experiments have demonstrated that only subpopulation(s) of culture-expanded MSC are capable of specific homing (Francois et al, 2006; Ponte et al, 2007). It was not clear whether one or multiple migratory subpopulations exist. Our experiments suggest that there is one migratory subpopulation because no additional effect was observed when PDGF-BB and SDF-1α were both present in the lower compartment of the Transwell compartment. It is possible that suboptimal concentrations of the stimuli would in act in concert, while maximal levels would not do so.
As only few MSC were able to migrate, it was evaluated whether actin polymerization and focal adhesion formation over in response to migratory cues, required to enable cell migration (Mitchison & Cramer, 1996; Zigmond, 1996) was restricted these cells. In contrast, the majority of the MSC show actin membrane ruffles in response to SDF-1α or PDGF-BB. Similar observations were made whilst studying the focal adhesion protein paxillin, which has a role in the turnover of focal adhesions formed during migration (Geiger et al, 2001; Deakin & Turner, 2008). Paxillin has multiple phosphorylation sites that regulate its function (Brown & Turner, 2004; Deakin & Turner, 2008). Enhanced phosphorylation at residues Y31 and Y118 is related to increased focal adhesion turnover and migration (Petit et al, 2000; Iwasaki et al, 2002). To our knowledge, this study showed, for the first time, that stimulation of MSC with migratory cues, such as SDF-1α and PDGF-BB, increases paxillin Y31 phosphorylation and induces its redistribution to the cell periphery, indicating that also in MSC, phosphorylation at this residue is related to focal adhesion turnover and migration. FCS increased pY31 paxillin to a higher extent than SDF-1α, possibly due to the multiple growth factors and chemokines, including SDF-1α, present in FCS. Therefore, phosphorylation could be induced by several pathways, while SDF-1α only signals through its receptor CXCR4 (Bleul et al, 1996; Oberlin et al, 1996). Paxillin has been related to MSC differentiation (Kundu & Putnam, 2006; Luo et al, 2008) and redistribution of total paxillin was observed upon stimulation with the chemoattractant sphingosine-1-phosphate (Meriane et al, 2006), but a role for phospho-paxillin was not described previously. Phosphorylation at pY31 and actin polymerization were detected in the majority of the MSC, while migration only occurred in a small percentage of MSC, indicating that functionality of the machinery involved in the initial response to migratory cues is not restricted to the migratory MSC subset. Future studies will reveal whether differential phosphorylation of other paxillin residues in migratory and non-migratory MSC causes different paxillin-mediated signalling in these cells (Brown & Turner, 2004; Deakin & Turner, 2008).
Studies on the characteristics of the migratory MSC in vitro revealed that they maintained proliferation and differentiation capacity after migration. Moreover, in secondary migration experiments, the migratory subpopulation maintained its migratory capacity and this was significantly higher compared to MSC that did not migrate in the first run. The migration percentages in the second migration run were lower than in the first migration run, which has been described for HSC as well (Voermans et al, 2001). It remains to be established whether the migratory subset also has an enhanced migratory capacity in vivo and whether these cells contribute to a better recovery after injury. To be able to enrich for the migratory MSC from the heterogeneous MSC population, markers that distinguish migratory MSC from the non-migratory MSC are required, but none of the markers evaluated in the current study were exclusively or differentially expressed on migratory MSC. These results, together with observations that many MSC are able to respond to migratory cues, lead to the hypothesis that migratory MSC are not a specific subpopulation that can be identified by surface marker expression, but are in a different intracellular state, which enables these cells to translate the initial response to migratory cues into migratory behaviour.
A better understanding of the (molecular) mechanisms involved in MSC migration will enable modulation of MSC to enhance their homing efficiency (Sackstein et al, 2008). Cell cycle has been linked to migration and homing in HSC (Glimm et al, 2000; Cashman et al, 2002). HSC in S- and G2/M phase loose homing and engraftment capacity (Glimm et al, 2000), and SDF-1/CXCR4 is involved in regulating HSC quiescence and cycling (Chabanon et al, 2008). Interestingly, mobilized HSC were predominantly in G0/G1-phase of the cell cycle (Uchida et al, 1997). Here we showed that the cell cycle also contributes to MSC migratory potential. There is a trend of more cells in G1-phase and significantly less cells S- and G2/M-phase in migrating MSC as compared to their non-migrating counterparts. Experiments on modulating the cell cycle in MSC should therefore be considered to enhance in vitro and in vivo migration of MSC. Indeed, when the proportion of FBMSC in G1-phase (34·8 ± 9·4%) was increased by harvesting cells at 50–70% confluence (n = 3), as compared to 4·9 ± 1·9% in G1-phase in cells harvested at 80–90% confluence (n = 5), both spontaneous and SDF-1α induced migration was significantly higher (spontaneous 13·6 ± 1·1% vs. 7·2 ± 0·5%, P < 0·025, SDF-induced 20·3 ± 1·2% vs. 12·0 ± 0·7%, P < 0·025).
In conclusion, our results suggest that ex-vivo expanded MSC derived from various adult and fetal tissues have different migratory capacity towards growth factor and chemokine stimuli, which is likely to be related to the tissue of origin rather than the developmental stage of this tissue. Migratory MSC not only maintain differentiation and migration capacity, but also contain fewer cells in S- and G2/M phase, which shows a relationship between cell cycle and MSC migration. Identification and modification of processes that favour MSC migration will make a significant contribution to increasing efficacy of future cellular therapies.
M.W.M. is financially supported by DPTE grant nr 06728. J.D.vB. is financially supported by the Dutch Heart Association (Grant nr. 2005T039) and NOW-ZonMW Veni grant 916·76·053. Furthermore, the authors wish to thank Dr F.J. van Milligen for providing adipose tissue-derived MSC, the staff of the Bloemenhove Kliniek (Heemstede, The Netherlands) for providing fetal tissues, Erik Mul and Floris van Alphen for flow cytometry and confocal microscopy assistance.
M.W.M.: Designed and performed research, analysed data and wrote the manuscript.
W.A.N.: Designed research, analysed data and wrote the manuscript.
M.K.: Performed research.
C.J.A.K.: Performed research.
K.W.: Provided study material.
J.D.vB.: Designed research, analysed data and wrote the manuscript.
EvdS.: Designed research, analysed data and wrote the manuscript.
C.V.: Designed research, analysed data and wrote the manuscript.
The authors indicate no potential conflicts of interest.