Background: Multipotent mesenchymal stromal cells (MSC) are of particular interest for their potential clinical use in cartilage engineering, but a consistent model is missing in large animals.
Objective: In the absence of any detailed study reporting a complete characterisation of the mesenchymal cells isolated from sheep bone marrow, we fully characterised adherent stromal cells and developed a pre-clinical model of cartilage engineering by implantation of autologous MSC in the Merinos sheep.
Methods: Ovine MSC (oMSC) were isolated from bone marrow, expanded and further characterised according to the recently proposed definition of the MSC. The experimental model consists of partial-thickness lesions created in the inner part of the patellae of the posterior legs. Lesions were filled with oMSC with or without chitosan, with or without transforming growth factor (TGF)β-3, in a fibrin clot.
Results: oMSC were shown to display the three main characteristics of MSC: adherence to plastic, phenotypic profile (positive for CD44, CD105, vimentin and negative for CD34 and CD45), and trilineage differentiation potential. We also report two other important functional characteristics of MSC: support of long-term haematopoiesis and immunosuppressive capacity. In vivo, 2 months after implantation the histological analysis revealed chondrocyte-like cells surrounded by a hyaline-like cartilaginous matrix that was integrated to the host cartilage when oMSC were combined with chitosan and TGFβ-3.
Conclusions: This study provides for the first time a strong characterisation of oMSC and establishes the basis for a model of cartilage engineering in a large animal.
Statistics from Altmetric.com
Articular cartilage damage frequently results from injury or diseases such as rheumatoid arthritis (RA) or osteoarthritis (OA). Due to its poor intrinsic capacity to heal, the articular cartilage cannot fully regenerate and its continuous degradation will lead to large defects reaching the subchondral bone. At present, there is no known treatment that has proven to be superior in overcoming the limitations of the methods currently used in clinical practice. Tissue engineering is a promising alternative to the conventional approaches for cartilage repair, but has to be validated in large animal pre-clinical models providing mechanical constraints similar to those of human joints.
Tissue engineering involves the use of various cell types that can produce a cartilaginous matrix characterised by the presence of specific collagens and proteoglycans.1 The cells generally used are allogeneic or autologous chondrocytes. Although autologous chondrocyte implantation has exhibited efficiency for cartilage regeneration,2 the limited source of these cells, their poor capacity to proliferate and their tendency to dedifferentiate in vitro prompted the search for an alternative source of cells. Due to their availability and the ease of isolation, skeletal muscle and adipose-derived stem cells could be used, as they were reported to undergo chondrogenesis in vitro and in vivo.3 4 The well characterised multipotent mesenchymal stromal cell (MSC) may be a better alternative however, and is currently under investigation for its potential to generate cartilage.
MSC are pluripotent precursors localised in the bone marrow (BM) but also in fat tissue or synovium. They are characterised by their capacity to support haematopoiesis and differentiate into multiple mesenchymal lineages: cartilage, bone and adipose tissues (for a review, see Jorgensen et al).5 Whereas no specific markers are available to characterise MSC, they are recognised on the basis of the expression of a number of cell surface markers such as CD44, CD73, CD90, CD105 and the absence of haematopoietic stem cell markers (such as CD34 and CD45).6 MSC have been isolated from many mammals7 but ovine MSC (oMSC) are poorly characterised. BM-derived oMSC were shown to stain positive for CD105, α-smooth muscle actin, desmin and negative for CD31.8 Cord blood-derived oMSC were reported to differentiate into osteoblasts, adipocytes and chondrocytes9 and their capacity to form cartilage in vivo was demonstrated after seeding onto polyglycolic acid scaffolds.10 However, the data are scattered and a complete characterisation of the cells is lacking.
In the present study, we have established six primary adult oMSC cultures from BM aspirates and evaluated their phenotypic and functional properties. Using these fully characterised oMSC, we have developed a pre-clinical model of cartilage engineering using autologous implantation onto a chitosan-based scaffold.
MATERIALS AND METHODS
BM aspirates were obtained from the humeral head of six 1-year-old female Merinos sheep after approval by the Regional Ethical Committee. Total BM cells (0.5–1.5×107 cells/ml) were plated at the density of 1–2×107 cells/cm2 in complete medium consisting of α-minimum essential medium (α-MEM), supplemented with 10% foetal bovine serum (FBS; Perbio Science France SAS, Brebières, France), 100 U/ml penicillin and 100 μg/ml streptomycin. Cells were subsequently plated at the density of 103 cells/cm2.
The ovine peripheral blood mononuclear cells (oPBMC) were isolated by density gradient centrifugation using the lymphocyte separation medium (density 1.077; Eurobio, Courtaboeuf, France) and centrifuged at 670 g for 30 min at ambient temperature. Cells at the surface of the separation medium were suspended in 50 ml phosphate buffered saline (PBS), centrifuged at 1000 g for 10 min and subsequently suspended at 5×107 cells/ml FBS:dimethyl sulfoxide solution (90:10) for freezing.
The frequency of colony forming unit-fibroblast (CFU-F) in fresh BM samples was determined by plating the mononuclear cells at the concentration of 0.5×106, 1×106, 2×106 cells in a 100 mm culture dish. After 14 days, cells were fixed in methanol, stained with the Giemsa solution (Sigma, Saint Quentin Fallavier, France) and colonies (containing at least 50 cells) were scored macroscopically.
For evaluating the number of clonogenic cells at various passages, cells were diluted by serial limited dilution and plated at the density of 1 cell every 3 wells in a 96 well plate for 14 days. The number of clonogenic cells was determined as the percentage of single cells able to form one colony per well.
FACS analysis of cultured MSC
Cultured oMSC (passages 1–5) were harvested by treatment with 0.05% trypsin and 0.53 mM EDTA, suspended in PBS containing 0.1% bovine serum albumin (BSA) and 0.01% sodium azide and processed as previously described.17 Briefly, cell aliquots (105/100 μl) were incubated for 30 min on ice with monoclonal antibodies (mAb) or isotypic controls followed by two washes in PBS containing 0.1% BSA and when necessary, another incubation with conjugated secondary antibodies for 30 min on ice. FITC-conjugated mAbs against hCD34, hCD44 (Immunotech, Marseille, France), oCD45 (Serotec, Cergy Saint Christophe, France), unconjugated mAbs specific for mammal vimentin (BD, Le Pont de Claix, France) and hCD105 (Santa Cruz, Teby, Le Perray en Yveline, France) and the secondary FITC-conjugated goat F(ab’)2 Fragment anti-mouse IgG antibody (BD) were used. Cross-reactivity of hCD44 mAbs was confirmed with oPBMC. Flow cytometry was performed on a fluorescence activated cell sorter (FACS Scan), and data analysed with the Cellquest software (BD Pharmingen).
Long-term haematopoiesis support
oMSC at passage 1 and the MS5 murine stromal cell line were plated at the concentration of 2000 cells/cm2 until confluency. Human umbilical cord blood CD34+ cells were purified as described11 and 10 000 cells co-cultured with the stromal cells in the presence of 10 ng/ml stem cell factor, 15 ng/ml FLT3 Ligand, 20 ng/ml thrombopoietin (human recombinant proteins; Peprotech, Neuilly-sur-seine, France). Co-cultures were demi-depopulated twice a week and half of the cells were counted. Proliferation of CD34+ cells is expressed as the cumulative cell number when cultured in the presence of three individual oMSC populations or the MS5 cell line. After 63 days, both non-adherent and adherent cells were recovered and plated in SYN.S2 methylcellulose based medium with foetal bovine serum (FBS) containing a number of recombinant human cytokines (AbCys, Paris, France). This medium is optimised for evaluation of myeloid and erythroid clonogenic progenitor cells CFU-G, CFU-GM, CFU-M, CFU-Mix and BFU-E and was used to determine the total clonogenic cell content of each culture condition. Colonies were scored 14 days later and colony-forming cell (CFC) enumeration was performed as described previously.12 13 Results are expressed as the number of CFC per plated 105 cells calculated at day 63 and the fold increase of CFC corresponding to total number of CFC d63/CFC d0. The number of CFC calculated at d0 was calculated to be 2500 per 104 CD34+ plated cells. The data are representative of four experiments.
In vitro differentiation
Chondrogenic differentiation was induced by culture in micropellet for 21 days in chondrogenic medium containing 10 ng/ml human transforming growth factor β3 (hTGFβ3) (R&D Systems, Lille, France). Osteogenic differentiation was induced by culture for 21 days in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS, 0.1 μM dexamethasone (Sigma, l’Isle d’Abeau, France) and 50 μg/ml ascorbic acid (Sigma). For mineralisation, 3 mM NaH2PO4 was added to the osteogenic medium. Adipogenic differentiation was induced for 21 days in the presence of DMEM containing 10% FBS, 1 mM dexamethasone, 17 mM acid ascorbic-2-phosphate, β-glycerophosphate and hBMP-2 in a conditioned supernatant from a producing cell line.
Pellets and tissues were fixed in 4% formaldehyde and embedded in paraffin. Presence of proteoglycans was visualised by incubation with a 0.1% Safranin O solution for 5 min. For evaluation of mineralised matrix, cells were fixed with ethanol 70% for 1 h, rinsed with PBS and stained with Alizarin red S (Sigma) for 5 min, followed by a rapid wash with acetone/methanol solution (1:1). To evaluate the presence of neutral lipids, cells were fixed with 3% glutaraldehyde for 1 h, stained with oil red O solution for 2 h and washed with 60% isopropanol. Immunohistochemical analysis was performed for type II collagen and aggrecan, using the primary antibodies from Interchim (Montluçon, France) and Chemicon (Hampshire, UK), respectively. Immunostaining was then performed using the “Ultravision Mouse Tissue Detection System” kit from Lab Vision Corporation (Fremont, California, USA).
Total RNA isolation and reverse transcriptase (RT)-PCR
Total RNA was extracted using the RNeasy mini kit (Qiagen S.A., Courtaboeuf, France). One μg of total RNA was reverse transcribed to cDNA and 1/6 of the reverse transcription was used for primer amplification as a two-step procedure with the GeneAmp® RNA PCR Core Kit (Applied Biosystems, Courtaboeuf, France). The reactions were performed using an Eppendorf Mastercycler using the primers listed in table 1. The cycling profiles were as follows: 94°C for 5 min; 94°C for 30 s, annealing temperature (depending on the primers; see table 1) for 30 s, 72°C for 30 s (for 35 cycles, except for β2-microglobulin: 30 cycles); 72°C for 5 min. PCR products were analysed in 1% agarose gel, stained with ethidium bromide and visualised by ultraviolet transillumination.
Mixed lymphocyte reaction
The proliferation of oPBMC (passage 3) in the presence of autologous or allogeneic oMSC was determined as previously described.17 Mixed lymphocyte reactions (MLR) were performed as described above, except that responder oPBMC (105 cells) were stimulated with allogeneic oPBMC (105 cells) in the presence of oMSC (105 cells). Allogeneic stimulator oPBMC (107 cells/ml) were treated with 50 μg/ml mitomycin C (Sigma) at 37°C for 45 min, followed by five extensive washes with FBS-containing RPMI 1640 medium. Results are given as the percentage of proliferation relative to maximal proliferative response after allogeneic stimulation (MLR) and expressed as the mean of triplicates. One representative out of three independent experiments performed with six different oMSC populations is shown.
In vivo tumourigenesis model
CB17-SCID/bg mice were purchased from Harlan (Gannat, France) and cared for according to the laboratory animal care guidelines. For cell injection, mice were anaesthetised with pentobarbitol (50 mg/kg) and MSC were administered subcutaneously. At least five mice were included per group of treatment using 106 cells in 100 μl PBS and maintained for 4 months.
In vivo model of cartilage repair
Three female sheep were used in this study, in accordance with the Regional Ethic Committee on Animal Research and Care. Blood was withdrawn from a jugular vein and 200 μl was immediately delivered onto 25 μl of a suspension containing chitosan (45 mg) and/or hTGFβ3 (50 ng) and/or autologous oMSC (107 cells at passage 1) for 30 min at room temperature. In the meantime, a lateral parapatellar skin incision was made and the knee joint was exposed via lateral dislocation of the patella. The internal groove of the patella was exposed and a 4 mm-diameter biopsy punch was used to create a partial-thickness defect in the cartilaginous tissue (two defects per patella). The defects were then filled with the clots and the patella put back in place. Animals were killed 9 weeks after and the repaired tissues were cut in 5 mm large cubes for histology.
Growth characteristics and phenotype of oMSC
A total of (2.8 (1))×108 (8.7×106 cells/ml (3.2)×106 cells/ml) nucleated cells were obtained from a 25 to 35 ml sample of BM. After isolation, fibroblast-like cells could be observed (fig 1A). The cells were expanded by more than 1010-fold for 8 months with a linear growth curve until passage 10 (fig 1B). For the first 10 passages the mean doubling time was about 12 h, then it rapidly increased indicating that cells underwent senescence. The mean (SD) clonogenic efficiency of the nucleated cells from fresh BM was 1.50 (0.36) CFU-F in 105 mononuclear cells. The clonogenic potential of oMSC during culture was stable for at least four passages (42% (0.03)) and decreased thereafter to reach 22% (0.083) at passage 9. Moreover, after subcutaneous implantation of oMSC (passage 2 and 7) in SCID mice, no tumour was observed on the various organs suggesting that cultured MSC were not tumourigenic, at least until passage 7.
The phenotype of oMSC was analysed on few markers because a restricted number of antibodies was available for sheep. Cells (passage 2 and 5) were shown to be negative for CD45 and CD34, indicating that the cells were not of haematopoietic origin (fig 1C). The cells were found to be highly positive for CD44 and vimentin and, low positive for CD105.
Differentiation potential of oMSC
The chondrogenic potential was evaluated in micropellet culture after 3 weeks. The accumulation of sulphated proteoglycans was visualised by Safranin O staining (fig 2A, left panel). The expression of mRNA for type II collagen and aggrecan was detected by RT-PCR at d21 whereas the mRNA for biglycan was only very slightly induced (fig 2A, right panel).
When cultured in osteogenic medium, the spindle shape of oMSC flattened and broadened with time. At day 21 in the presence of proliferation medium, no mineralisation was observed using Alizarin red S staining (fig 2B, left panel). By contrast, the formation of mineralised nodules was detected as shown by a highly positive Alizarin red S staining (fig 2B, middle panel) but no increase of the osteopontin mRNA was observed by RT-PCR (fig 2B, right panel). Transcripts for type II collagen and aggrecan mRNA were not detected by RT-PCR at d21 (data not shown). Due to the lack of sequence identification of other osteoblast specific genes, these results show that oMSC are able to secrete a mineralised matrix, which is the main function that characterises osteoblasts.
Morphologic changes and formation of neutral lipid vacuoles were observed 2 weeks after adipogenesis induction and visualised by Oil Red O staining (fig 2C, middle panel). The adipogenic differentiation was confirmed by the induction of PPARγ mRNA on day 21 (fig 2C, right panel). In these conditions, we could not detect the expression of transcripts for type II collagen and aggrecan mRNA at d21 (data not shown). Importantly, proliferation medium or standard adipogenic media containing 3-isobutyl-1-methylxanfine (IBMX), dexamethasone and indomethacin or insulin and dexamethasone, did not induce the formation of lipid droplets (fig 2C, left panel). Altogether, these data confirm the trilineage potential of oMSC.
Ability of oMSC to support long-term haematopoiesis
To determine whether oMSC were capable of supporting the proliferation and differentiation of haematopoietic stem-progenitor cells, the oMSC were cultured with human CD34+ haematopoietic stem cells. As controls, CD34+ stem cells were maintained with the MS5 murine stromal cell line or in the absence of a stromal layer. CD34+ stem cells cultured in the presence of either the MS5 stromal cell line or the three different populations of oMSC grew exponentially during the 63-day culture period (fig 3A). Importantly, oMSC were able to support the expansion of human LTC-IC cultures (fig 3B). The haematopoietic cell expansion rate with the oMSC was approximately twofold lower than with the MS5 stromal cells but 18-fold higher than in the absence of a stromal layer, demonstrating that oMSC are able to support long-term haematopoiesis.
Immunosuppressive properties of oMSC
The immunosuppressive property of six oMSC populations was tested in a proliferative assay in the presence of oPBMC. All the five allogeneic oMSC failed to induced the proliferation of T cells since the response was in the same range that the basal proliferation (data not shown). In a second step, oPBMC from two individuals were cultured in a mixed lymphocyte reaction (MLR). In comparison to the maximal proliferative response normalised to 100% (MLR), the proliferation of oPBMC was inhibited in the presence of oMSC from autologous or allogeneic origin (fig 3C). Altogether, the results demonstrate that the oMSC are able to suppress the proliferation of allogeneic T lymphocytes induced by cells from a third party.
In vivo capacity of differentiation of oMSC
To evaluate the potential of tissue regeneration of oMSC, we implanted the cells within a fibrin matrix into cartilaginous lesions of sheep joints. Macroscopical examination of the cartilage defects was performed 9 weeks after surgery. Control defects filled with fibrin appeared rosy and the margins of the lesions were clearly distinguishable (fig 4A). In the lesions receiving chitosan, the lesions appeared partly as a glossy white or a fibrous-like tissue (fig 4B). By contrast, the defects treated with oMSC either alone or with chitosan with or without TGFβ3 were filled with glossy white repaired tissue, although moderately well integrated with the surrounding cartilage (fig 4C–E). When transforming growth factor (TGF)β3 alone was applied, the defects were partly filled with a white appearing tissue at the margins and a scarred tissue in the central region (fig 4F).
Histological observations confirmed that, in the defects filled with fibrin alone (fig 4G) or with fibrin and oMSC (fig 4I), the repaired tissue appeared poorly integrated within the host cartilage. Most of the tissue was fibrotic (even necrotic; see blue staining material) and the matrix was not stained by safranin O (fig 4G,I), did not express aggrecan or collagen type II in the lesions (fig 5A,B). By contrast, when the chitosan powder was included within the fibrin clot, in the presence or in absence of oMSC, a tissue resembling cartilage with cells included in lacunae was observed (fig 4H,J). However, the integration of the repaired tissue was very poor and in same areas, a fibrous tissue overlapped the repaired tissue. The safranin O staining was slightly positive whereas collagen type II and aggrecan were expressed in the matrix (data not shown). Addition of TGFβ3 increased the cellularity in the repaired tissue and, in absence of exogenously added oMSC, a large area of tissue highly positive for Safranin O seemed to go deeper into the subchondral bone, suggesting the recruitment of endogenous cells from the bone marrow (fig 4L). Low but positive staining for aggrecan as well as a high staining for type II collagen confirmed that normal hyaline cartilage was observed at the margins of the neotissue (fig 5E,F). The best repaired tissue in terms of integration to the host cartilage, presence of chondrocyte-like cells surrounded by a matrix which stained positive for Safranin O (fig 4K), collagen type II and aggrecan (fig 5C,D) resulted from the injection of the combination of oMSC, scaffold and growth factor although a fibrous tissue overlapped the neocartilage in some areas.
Previous studies reported the isolation of oMSC from BM or cord blood but no or poor investigations were performed to characterise the adherent stromal cells obtained after subcultures.10 14 Consequently, the aim of this study was first to accurately investigate the properties of the oMSC according to the criteria recently defined.15 Second, we developed a new model of cartilage repair aiming at evaluating the capacity of oMSC to form a neotissue with the characteristics of functional hyaline cartilage.
Here, we demonstrate that oMSC display similar phenotypic and functional properties to MSC isolated from other species. First, we evaluated the number of CFU-F and the clonogenic capacities of the cells along the passages. The frequency of CFU-F is similar to that reported from human bone marrow (1 in 104 to 105 mononuclear cells).16 Second, we performed a phenotypic analysis of these cells using the few commercially available antibodies specific for ovine or cross-reacting human epitopes. The cells are negative for the haematopoietic markers and positive for CD44, CD105 and vimentin which are characteristic of MSC.17 Third, the growth properties of oMSC are higher than hMSC (120 cell doublings vs 40–50 respectively).16 18 Importantly, bFGF, which greatly enhances the proliferation and maintains the multipotentiality of hMSC, was omitted in our conditions.19 Although yet to be formally demonstrated, it may be hypothesised that the high proliferation potential of oMSC is due to longer telomeres or the expression of embryonic genes as suggested for human MSC.20 Nevertheless, oMSC cultured in vitro for at least seven passages are not tumourigenic when implanted in immunodeficient mice. Fourth, we demonstrate for the first time that three different MSC populations possess the three main properties specifying the MSC: (1) capacity to support long-term haematopoiesis, (2) immunosuppressive function, and (3) multilineage differentiation potential. Although the properties of immune tolerance and haematopoietic support may not be relevant to the model of cartilage repair, they confirm that these cells are MSC. The adipogenic or osteogenic potential of oMSC has already been shown,8 9 as well as their chondrocytic differentiation using various scaffolds in vitro10 or in vivo.14 However, this is the first report reporting the trilineage potential of single MSC populations, their characterisation using newly designed primers, and chondrogenesis induction in the micropellet. Under our conditions, 3 weeks were sufficient to induce chondrogenic differentiation whereas 12 weeks was required to differentiate the cord blood-derived oMSC into chondrocytes.10 BM-derived oMSC are thus similar to hMSC in their capacity of differentiation when cultured in 3D chondrogenic conditions. By contrast, adipogenic differentiation did not occur when oMSC were cultured in media typically used to induce adipogenesis of MSC from human or murine origin whereas the addition of BMP-2 was sufficient to trigger the differentiation process. BMP-2 has been shown to induce adipogenesis of murine or rat MSC when cultured in the presence of dexamethasone or rosiglitazone.21 22 In this study, it was suggested that BMP-2 could act as a potent adipogenic agent if presented together with activators of PPARγ. After osteogenesis induction the expression of osteopontin was not enhanced, but the high basal levels in absence of stimulation may have hampered the up-regulation of the transcript. However, the high capacity of oMSC to mineralise the matrix, which is the main function of osteoblasts, was sufficient to demonstrate their osteogenic differentiation potential. In summary, oMSC possess morphologic, immunophenotypic and functional characteristics similar to those previously described for their human or murine counterparts. Their higher in vitro proliferation potential makes them interesting sources of cells for evaluating cartilage-engineering approaches in a large pre-clinical animal model.
Sheep may be a good model to investigate the feasibility of cartilage repair using MSC, as models of meniscectomy OA and of large osteochondral defects in the femoral groove are available.23–25 However, OA changes involved biochemical, biomechanical and histopathological alterations that complicate the evaluation of cartilage repair in a preliminary setting. The large osteochondral defect model involves the destruction of the subchondral bone and the immigration of endogenous MSC from the bone marrow into the defect contributing to an accelerated healing process.23 In this study, we aimed at developing a model of partial-thickness lesions in the cartilaginous tissue to better evaluate the impact of the exogenously added oMSC on the repair process. We demonstrate that the fibrin glue already used as a suitable matrix to settle the cells inside the lesions,26 the oMSC, or the scaffold alone have poor properties to repair partial-thickness defects. By contrast, when the oMSC were seeded in the presence of the chitosan-based scaffold and TGFβ3, a repaired tissue staining for aggrecan and type II collagen was obtained. Addition of TGFβ3 enhanced the cell proliferation and favoured the integration of the neotissue within the host cartilage. The improvement of cartilage repair by local delivery of a growth factor using engineered stem cells has already been demonstrated in similar models.3 27 28 Here, the chitosan powder has proven to provide a biomechanical support for the differentiation of oMSC to hyaline-like cartilage. The advantages of this scaffold include the fact that it is resorbable and can be melt-processed into 3D scaffolds with different morphologies, sizes, shapes and properties adaptable to various tissue-specific environments.29 However, the scaffold has still to be optimised, and the development of chitosan-derived discs fitting to the size of the defects that may be pre-coated in vitro with the oMSC previous to implantation is under investigation. Development of chitosan discs is thought to limit the formation of fibrous tissue at the cartilage surface, which may hamper the formation of a fully functional tissue in the long term.
In conclusion, although the repair tissue needs to be analysed at later timepoints with an optimised combination of growth factor and scaffold and the evaluation of the biomechanical properties of the neocartilage, this model sets up the conditions for evaluating the repair of a partial-thickness defect in a pre-clinical setting. Further improvement of the technology appears to be feasible and has to be achieved prior to therapeutic applications.
This work was supported by the European Community (key action LSH 1.2.4-3, integrated project: “Adult mesenchymal stem cells engineering for connective tissue disorders. From the bench to the bed side”, contract no. 503161).
CJ and DN contributed equally to this work.
Funding: This work was supported by the European Community.
Competing interests: None declared.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.