We previously reported the identification in a nude mouse assay of molecular markers predictive of the capacity of articular cartilage–derived cells (ACDCs) to form ectopic stable cartilage that is resistant to vascular invasion and endochondral ossification. In the present study, we investigated whether in vitro–differentiated mesenchymal stem cells (MSCs) from the synovial membrane (SM) express the stable-chondrocyte markers and form ectopic stable cartilage in vivo.
Chondrogenesis was induced in micromass culture with the addition of transforming growth factor β1 (TGFβ1). After acquisition of the cartilage phenotype, micromasses were implanted subcutaneously into nude mice. Alternatively, cells were released enzymatically and either replated in monolayer or injected intramuscularly into nude mice. Marker analysis was performed by quantitative reverse transcription–polymerase chain reaction. Cell death was detected with TUNEL assay.
Cartilage-like micromasses and released cells expressed the stable-chondrocyte markers at levels comparable with those expressed by stable ACDCs. The released cells lost chondrocyte marker expression by 24 hours in monolayer and failed to form cartilage when injected intramuscularly into nude mice. Instead, myogenic differentiation was detected. When intact TGFβ1-treated micromasses were implanted subcutaneously, they partially lost their cartilage phenotype and underwent cell death and neoangiogenesis within 1 week. At later time points (15–40 days), we retrieved neither cartilage nor bone, and human cells were not detectable.
The chondrocyte-like phenotype of human SM MSCs, induced in vitro under specific conditions, appears to be unstable and is not sufficient to obtain ectopic formation of stable cartilage in vivo. Studies in animal models of joint surface defect repair are necessary to evaluate the stability of the SM MSC chondrocyte-like phenotype within the joint environment.
Autologous chondrocyte implantation represents a paradigm in the biologic repair of joint surface defects (1). Despite the promising clinical results (2, 3), the limitations of this procedure may be related to the source and the in vitro expansion of the cells. Chondrocytes are obtained by articular cartilage biopsy of an “uninvolved” and minor load-bearing area of the same joint. This results in additional injury to the joint surface, creating a potential locus minoris resistentiae and, possibly, an increased risk of developing osteoarthritis, especially in genetically predisposed individuals. Moreover, since the clinical indications may extend to joints with early osteoarthritis, the availability of healthy articular cartilage may be a limiting factor.
Chondrocyte expansion is essential for obtaining a sufficient number of cells for implantation. In vitro expansion of mature articular chondrocytes poses its challenges. The expandability of chondrocytes, as with most somatic cells, is limited by the occurrence of cell senescence (4). More importantly, chondrocyte expansion in monolayer cultures results in dedifferentiation (5) and loss of the capacity to form stable hyaline cartilage in vivo that is resistant to vascular invasion and endochondral ossification (6). This may account for some reported variability of autologous chondrocyte implantation and, more particularly, may affect the quality of the repair tissue and the long-term outcome of this treatment. Indeed, in some patients, the repair tissue consists of poorly differentiated and disorganized fibrocartilage (2).
Mesenchymal stem cells (MSCs) can be obtained without irreversible damage, are easily expandable with limited senescence, and are “phenotypically stable” throughout expansion and after storage in liquid nitrogen, as assessed by a number of molecular markers (7–9). The use of MSCs is restricted by insufficient knowledge of the long-term stability of the repair tissue and by their tendency to differentiate into other cell lineages, with the associated risk of heterotopic tissue formation (10). Autologous chondrocyte implantation can indeed be performed using cells that are stably committed to the appropriate phenotype and naturally resistant to vascular invasion, mineralization, and ossification (6).
We previously described an in vivo assay for measuring the capacity of adult human articular cartilage–derived cells (ACDCs) to form ectopic stable cartilage in vivo when injected intramuscularly into nude mice (6). In the same study, we identified molecular markers that reproducibly predict the outcome of this in vivo assay, independently of donor age. Stable ACDCs, as defined by the expression of these markers, were capable of forming ectopic stable cartilage in a per se nonchondrogenic environment, the skeletal muscle (6), and thereby partially overriding environmental signals (11). This marker set included type II collagen, specifically, α1(II) collagen (COL2A1), fibroblast growth factor receptor 3 (FGFR-3), and bone morphogenetic protein 2 (BMP-2) as positive markers, and activin receptor-like kinase 1 (ALK-1) as a negative marker (6). We have also characterized a population of adult human MSCs from the synovial membrane (SM) that are capable of differentiation into cartilage, bone, skeletal muscle, and adipose tissue in vitro (9). Subsequently, we demonstrated that SM MSCs can differentiate into skeletal muscle in vivo when injected intramuscularly into nude mice (12).
In the present study, we investigated whether human SM MSCs can acquire in vitro the expression of the reported markers of the stable-chondrocyte phenotype of ACDCs (6). We also examined whether this is associated with the capacity to form ectopic stable cartilage in vivo.
MATERIALS AND METHODS
Cartilage harvest and ACDC isolation.
Normal articular cartilage was obtained from the femoral condyles of 3 adult human donors (ages 18, 24, and 36 years) within 12 hours postmortem. Release of cells from articular cartilage and in vitro culture of ACDCs were performed as described previously (6).
Isolation and culture of SM MSCs.
Random biopsy specimens of SM (wet weight 10–50 mg) were obtained by aseptic technique from the knee joints of 3 age-matched human donors within 12 hours postmortem. MSCs were isolated and expanded in monolayer on plastic in growth medium (high-glucose Dulbecco's modified Eagle medium [DMEM; Life Technologies, Merelbeke, Belgium] containing 10% fetal bovine serum [BioWhittaker, Verviers, Belgium] and antibiotics [Life Technologies]), as described previously (9).
Assay for in vitro chondrogenesis.
The in vitro chondrogenesis assay was performed as described elsewhere (9). After expansion in monolayer, SM MSCs at passage 5 were released with trypsin, counted, tested for viability by trypan blue exclusion, and resuspended in growth medium at a density of 2.0 × 107 viable cells/ml. Micromass cultures were obtained by pipetting 100-μl droplets of cell suspension into individual wells of 12-well plates. After cells were allowed to attach without medium for 3 hours, a chemically defined serum-free medium (13) was added. The day of plating in micromass culture was designated as day 0.
Starting on day 1, when the culture medium was changed, recombinant human transforming growth factor β1 (TGFβ1; R&D Systems, Abingdon, UK), dissolved in 4 mM HCl containing 1 mg/ml bovine serum albumin (BSA; Serva, Heidelberg, Germany), was added to the culture medium every other day for 3 weeks at a final concentration of 10 ng/ml. Identical amounts of 4 mM HCl containing 1 mg/ml BSA were added to parallel cultures as controls for the treatment.
Release of cells from micromasses.
After 3 weeks of TGFβ1 treatment, micromasses were digested with trypsin for 5 minutes and then with 0.2% crude type IV collagenase (Life Technologies) in DMEM for 2 hours. Following release, cells were washed with growth medium. Cell viability was >97% by trypan blue exclusion test. Cells were either replated at high density (30,000 cells/cm2) or injected intramuscularly into the thigh of nude mice. A small aliquot was used for gene expression analysis.
Animal care and maintenance.
All the procedures on animals were approved by the local ethics committee. Eight-week-old, female NMRI nu−/− nude mice were maintained in isolator cages in pathogen-free conditions until the end of the experiments.
In vivo assays.
After 3 weeks of TGFβ1 treatment, micromasses were washed with phosphate buffered saline (PBS) and implanted subcutaneously into the back of anesthetized nude mice. Alternatively, following enzymatic release, 5 × 106 viable cells were resuspended in 50 μl of PBS and injected intramuscularly into the posterior compartment of the thigh of nude mice, as described elsewhere (6). At different time points (between 3 and 40 days), the mice were killed, and the tissues at the site of implantation were excised and either fixed for histologic, histochemical, or in situ hybridization experiments or used for gene expression analysis by reverse transcription–polymerase chain reaction (RT-PCR).
Histochemistry and detection of cell death.
Tissue samples were fixed overnight at 4°C in 10% neutral buffered formalin, embedded in paraffin, and sectioned at 5 μm. Staining with toluidine blue or Masson's trichrome was performed according to standard protocols. To detect cell death, TUNEL staining was performed using ApopTag in situ apoptosis detection kit (Oncor, Gaithersburg, MD). Slides were then mounted with Mowiol containing 4′,6-diamidino-2-phenylindole (ICN, Asse-Relegem, Belgium).
In situ hybridization for human-specific Alu genomic repeats.
In situ hybridization for human Alu genomic repeats was performed as described elsewhere (11).
Total RNA extraction and RT-PCR analysis.
Total RNA was isolated using TRIzol (Life Technologies). After DNase treatment, complementary DNA (cDNA) were obtained by reverse transcription of 2 μg of total RNA (Thermoscript; Life Technologies) using oligo(dT)20 as primer. Semiquantitative PCR was performed as described previously (9). Gene expression of human cells within mouse tissues was evaluated using primers specific for human cDNA. When mouse/human chimeric samples were equalized for the expression of human β-actin, control mouse samples without human cells were normalized to the mouse/human chimeric sample of the series with the highest mouse/human β-actin (12). Real-time quantitative PCR was carried out using the ABI Prism 7700 detection system (TaqMan; Applied Biosystems, Lennik, Belgium). The sequences of the primers as well as the TaqMan probes are listed in Table 1. For BMP-2, quantitative RT-PCR was performed using SYBR Green.
Table 1. Primers and probes used for RT-PCR analysis and expected sizes of PCR products*
Primer and probe sequence
RT-PCR = reverse transcription–polymerase chain reaction; COL2A = collagen type IIA; COL2B = collagen type IIB; MyHC-IIx/d = myosin heavy chain type IIx/d; BMP-2 = bone morphogenetic protein 2; FGFR-3 = fibroblast growth factor receptor 3; ALK-1 = activin receptor-like kinase 1. The prefix “mh-” means that the primer set does not allow the distinction of mouse cDNA from human cDNA. The prefix “h-” means that the primer set allows specific amplification of the human cDNA.
Expression of the stable-chondrocyte markers by human SM MSCs at levels comparable with those of stable ACDCs.
Human SM MSCs can form in vitro a cartilage-like tissue that displays metachromatic staining with Safranin O and toluidine blue, with expression of chondrocyte markers such as type II and type IX collagen (9). We investigated whether SM MSCs that have been induced to form cartilage in vitro acquired a set of molecular markers that we previously identified as being predictive of the capacity of expanded ACDCs to form stable cartilage ectopically in a nude mouse assay (6). We compared the messenger RNA (mRNA) expression levels of COL2A1, FGFR-3, BMP-2, and ALK-1 in the TGFβ1-treated SM MSC micromasses, in the cells released from them, and in stable ACDCs by quantitative RT-PCR. This experiment was performed in quadruplicate, with ACDCs and SM MSCs obtained from 3 age-matched donors.
TGFβ1-treated SM MSC micromasses displayed levels of COL2A1, FGFR-3, BMP-2, and ALK-1 mRNA that were comparable with those of stable ACDCs. Immediately after enzymatic digestion, the cells released in suspension displayed similar expression levels (Figure 1). These results indicate that under specific in vitro culture conditions, SM MSCs acquired the expression of the markers previously reported to be associated with the stable-chondrocyte phenotype of human ACDCs (6).
Rapid loss of the chondrocyte-like phenotype of SM MSCs in monolayer.
To investigate whether, after removal of the chondrogenic conditions, the induced chondrocyte-like phenotype of human SM MSCs is stable in vitro, cells were released from TGFβ1-treated micromasses and replated at high density in monolayer. After 24 hours, the number of cells was unchanged; neither proliferation, cell death, nor senescence was detected, as evaluated by 3H-thymidine incorporation, TUNEL assay, and staining for senescence-associated β-galactosidase (8, 9, 14), respectively (results not shown). These cells were analyzed for the expression of the molecular markers of the stable-chondrocyte phenotype using quantitative RT-PCR.
In all 3 donors tested, the expression of COL2A1, FGFR-3, and BMP-2 decreased to the basal levels of the original SM MSC populations in monolayer, while the expression of ALK-1 remained unchanged (Figure 2). Other chondrocyte markers tested, such as type IX and type XI collagen, became undetectable (data not shown). These results indicate that the chondrocyte-like phenotype induced in human SM MSCs by combining micromass culture and TGFβ1 treatment is rapidly lost in vitro under the culture conditions we used.
The chondrocyte-like phenotype of SM MSCs induced in vitro by combining micromass culture and TGFβ1 treatment is not sufficient for obtaining ectopic cartilage formation in vivo.
The rapid loss of the chondrocyte-like phenotype in monolayer by SM MSC–derived chondrocyte-like cells could be analogous to the dedifferentiation of ACDCs during culture expansion, although it happens faster (6). To test the hypothesis that the loss of the chondrocyte-like phenotype was due to the monolayer culture conditions, we injected 5 × 106 cells (obtained immediately after enzymatic release from TGFβ1-treated micromasses) intramuscularly into the thigh of nude mice. At the time of injection, these cells expressed the markers of the stable-chondrocyte phenotype at levels comparable with those of stable ACDCs, as assessed by quantitative RT-PCR (Figure 1).
Upon macroscopic dissection, no implant was retrieved at 3 weeks. Histologically, we observed neither cartilage nor bone formation at any time point examined up to day 40. Semiquantitative RT-PCR analysis at 3 weeks revealed myogenic differentiation, since the muscle injected with the SM MSC–derived chondrocyte-like cell population expressed human myosin heavy chain type IIx/d (MyHC-IIx/d), whereas human COL2A1 transcripts were not detected (Figure 3A). Human nuclei were localized within the mouse muscle by in situ hybridization for human Alu genomic repeats (Figure 3B). Under the same conditions, the injection of human ACDCs expressing the stable-chondrocyte markers (Figure 1) resulted in the formation of a human COL2A1–positive cartilage implant, with no detectable expression of human MyHC-IIx/d (Figure 3A). Together, these data indicate that the expression in human SM MSCs of the stable-chondrocyte markers induced by combining micromass culture and TGFβ1 treatment is not sufficient for cartilage formation in this mouse assay. Instead, a certain number of human SM MSCs were recruited to the muscle differentiation program, which further confirms their competence to respond to environmental cues from the surrounding muscle (12).
Preservation of tissue integrity is not sufficient for maintaining the stable-cartilage phenotype in vivo.
Intramuscular injection of the released cells in suspension resulted in dispersion of the SM MSC–derived chondrocyte-like cells within the skeletal muscle. To investigate whether the disruption of the cartilage-like tissue integrity due to the enzymatic digestion was responsible for the loss of the cartilage-forming capacity, we implanted intact TGFβ1-treated micromasses subcutaneously into nude mice.
The implants retrieved at 1 week were mostly contributed by human cells, as evaluated by in situ hybridization for human Alu genomic repeats (Figure 4A). Histochemical analysis revealed them to be weakly metachromatic with toluidine blue staining (Figure 4B) as compared with TGFβ1-treated micromasses before implantation (Figure 4D), which suggests a loss of cartilage-specific proteoglycans (15). Masson's trichrome staining revealed the presence of capillaries within the implants (Figure 4C), demonstrating that, as opposed to stable cartilage tissue, the SM MSC–derived cartilage-like tissue is not resistant to vascular invasion. Together, these data indicate that preserving tissue integrity is not sufficient to maintain the stable-cartilage phenotype in vivo after subcutaneous implantation.
Detection of cell death in TGFβ1-treated SM MSC cartilage-like tissue in vivo.
The loss of the cartilage phenotype in the TGFβ1-treated micromasses after subcutaneous implantation could be due to either dedifferentiation or death of the cells. Cell death was detected using the TUNEL assay. TUNEL-positive cells were observed mostly within the area occupied by human cells (Figures 4A and 5A). The presence of pycnotic nuclei at high magnification (results not shown) suggests that cell death was at least partly due to apoptosis. Parallel micromasses that were not implanted subcutaneously and were maintained in culture for an additional week did not display detectable cell death (Figure 5B). The cell death observed in vivo did not seem to be due to the manipulations needed for subcutaneous implantation, since cell death was not detected at 3 days after implantation (results not shown). At time points later than 15 days and up to 40 days, no implants were identified macroscopically at the site of subcutaneous implantation, and neither cartilage nor bone tissues were detected at histologic examination. At the same time points, we did not detect human β-actin expression by RT-PCR, nor did we observe human nuclei by in situ hybridization for Alu repeats (results not shown), indicating complete loss of the implanted human cells.
As compared with the skeletal muscle, the subcutaneous environment does not permit cartilage formation by implanted human stable ACDCs (Dell'Accio F, et al: unpublished observations). Therefore, TGFβ1-treated micromasses were implanted intramuscularly in the posterior compartment of the thigh of nude mice. At 1 week, we did not detect human cells, as evaluated by RT-PCR for human β-actin and in situ hybridization for human Alu repeats (results not shown). Histologically, we observed neither cartilage nor bone formation at any time point examined between 7 and 40 days.
Autologous chondrocyte implantation is becoming a generally accepted technique for the repair of focal joint surface defects and has proved superior to mosaicplasty in a prospective controlled clinical trial (3). However, several constraints, including those related to the source and the manufacture of the cellular product, represent important limitations for daily clinical practice.
SM MSCs are a promising alternative to ACDCs for the repair of joint surface defects. A small SM biopsy specimen represents an easily accessible source of autologous MSCs in the context of an explorative, diagnostic, or therapeutic arthroscopy. Expanded human SM MSCs maintain a stable molecular profile, as assessed by a number of markers, and retain the capacity to form a cartilage-like tissue in vitro (under specific conditions) over at least 10 passages as well as after cryopreservation (9). These properties of SM MSCs represent an opportunity to increase the consistency, reproducibility, and flexibility of the autologous cell implantation procedure. Yet, the use of multipotent MSCs implies the risk of unforeseen events, such as heterotopic tissue formation (10). Predifferentiation has been suggested as a useful step for inducing proper lineage commitment and reducing the risk of undesired tissue formation by restricting alternate differentiation pathways. However, it is not clear to what extent the phenotype acquired in vitro is maintained once the cells are implanted in vivo.
In this study, we investigated the stability of the chondrocyte-like phenotype obtained in SM MSCs by combining micromass culture and TGFβ1 treatment, using the nude mouse assay for stable-cartilage formation in vivo in an ectopic site validated with ACDCs and the associated molecular markers (6). We demonstrated that SM MSC–derived chondrocyte-like cell populations obtained after 3 weeks of TGFβ1 treatment in micromass culture acquired the expression of the markers associated with the stable-chondrocyte phenotype at levels comparable with those of stable ACDC populations. However, the chondrocyte-like phenotype of SM MSCs was transient in vitro upon removing the chondrogenic conditions and replating the released cells in monolayer. In addition, despite the expression of the stable-chondrocyte markers, cells released from TGFβ1-treated micromasses failed to form stable cartilage when injected intramuscularly into nude mice, suggesting that the loss of the chondrocyte-like phenotype is not solely related to the monolayer culture conditions.
The cell fate of the SM MSCs differed in the two nude mouse assays adopted in this study. In TGFβ1-treated micromasses implanted subcutaneously, we observed cell death and neoangiogenesis within the cartilage-like tissue. Although the experiments were performed in immunodeficient nude mice, it cannot be excluded that some form of inflammatory/immunologic reactions might have contributed to the death of the human cells after micromass implantation. However, no obvious inflammatory infiltrates or foreign-body reactions were documented at histologic examination at all time points examined. Alternatively, these data could be interpreted as showing the formation of a transient cartilage tissue in the context of endochondral ossification. Indeed, during endochondral ossification, the cartilage mold is invaded by blood vessels, undergoes apoptosis, and is replaced by bone (16, 17). However, we were unable to detect bone formation up to day 40, in contrast to bone seen in cartilage tissue obtained by implantation of epiphyseal chondrocytes (ref. 18 and Dell'Accio F, et al: unpublished observations). It is possible that bone did not form because of either an absence of competent osteoprogenitors within the implanted human micromasses or a failure in their recruitment from the mouse host.
Although we have proved that human SM MSCs are multipotent at the single-cell level, and are therefore also osteogenic, we cannot exclude that micromass culture and TGFβ1 treatment may have limited their differentiation potential. Also, when implanted within the skeletal muscle of the thigh of nude mice, TGFβ1-treated micromasses did not give rise to any retrievable cartilage or bone tissues, suggesting that the lack of stability of the implanted cartilage-like tissue and the absence of bone are not related to the site of implantation. When the cells were released and injected intramuscularly, we were unable to retrieve any cartilage-like tissue, although human cells were detected for at least 40 days. In contrast, after subcutaneous implantation of the TGFβ1-treated micromasses, human cells were not retrieved at time points later than 15 days. The released cells, but not the cells aggregated in micromasses, displayed the capacity to survive and to respond to environmental signals. RT-PCR detection of human MyHC-IIx/d at 3 weeks suggests muscle differentiation of the intramuscularly injected released cells. SM MSCs are known to be competent for myogenesis after transplantation into skeletal muscle (12). The acquisition of a muscle phenotype seems to indicate that myogenic environmental signals were sufficient to override the induced chondrocyte-like phenotype of the injected cell population.
To respond to inductive signals, competent cells must be surrounded by similar cells (19). This phenomenon, called the “community effect,” has been described in the mammalian embryo as a prerequisite for mesodermal cells to undergo differentiation and maturation (20) and has been shown to be important for the maintenance of a stable-cartilage phenotype by ACDCs (11). The capacity to form cartilage in vivo by stable ACDC populations is dependent on the number of cells, since injection of 5 × 105 cells did not result in the formation of any retrievable cartilage, and although cells persisted in the muscle, as determined by in situ hybridization for human Alu repeats, no cartilage tissue was detected in serially sectioned injected muscles (ref. 11 and Dell'Accio F, et al: unpublished observations). This suggests that specific threshold levels and, possibly, autocrine/paracrine mechanisms are required for in vivo cartilage tissue formation by stable ACDCs. A threshold might also be required for SM MSC–derived chondrocyte-like cell populations and could be higher than the threshold for stable ACDCs, thereby explaining, at least in part, the failure to form stable cartilage in this assay.
Two lines of evidence, however, plead against this possibility. First, 10 × 106 viable cells from TGFβ1-treated SM MSC micromasses proved to be insufficient to form cartilage in vivo (data not shown). Second, undigested intact micromasses, in which the SM MSC–derived chondrocyte-like cells remained tightly clustered together, also failed to retain their cartilage-like phenotype when implanted in vivo. It remains to be investigated whether the same cell populations would make durable cartilage tissue when implanted into a cartilage-inducing/maintaining environment, such as an articular cartilage defect. Moreover, the proportion of cells with a stable-chondrocyte phenotype and the expression of each marker at the single-cell level might differ between the stable ACDC population and the SM MSC–derived chondrocyte-like cell population. The aim of this study was to compare the SM MSC–derived chondrocyte-like cells with the stable ACDCs as whole cell populations in terms of the expression of the molecular markers reported to be associated with the stable-chondrocyte phenotype of ACDCs and in terms of the related capacity to form ectopic stable cartilage in vivo (6). A phenotype analysis at the single-cell level is an ongoing effort in our laboratory.
We previously found that when injected into regenerating anterior tibialis muscles of nude mice, stable ACDC populations did not form a retrievable cartilage pellet, but maintained their expression of type II collagen in vivo for at least 4 weeks (11). In the present study, SM MSC–derived chondrocyte-like cells did not retain the expression of type II collagen (undetectable at 3 weeks). This might be due to either selective cell death or nuclear reprogramming with, possibly, a shift from the chondrocyte-like phenotype to the myocyte-like phenotype. The data obtained in vitro demonstrate that the chondrocyte-like phenotype of SM MSCs is transient and is rapidly lost upon removal of the chondrogenic conditions. Whether the chondrocyte-like cells dedifferentiate to the original phenotype of expanded SM MSCs, undergo cell death, or are overgrown by the undifferentiated cells remains to be addressed. The absence of detectable senescence, cell death, or proliferation would lend support to a rapid dedifferentiation process, making the cells competent to respond to myogenic signals. This is not surprising, and it is especially not specific to SM MSCs, since human ACDCs also undergo a myogenic fate in this assay after they have lost their phenotypic stability due to serial passaging (11). We recently reported that myogenesis of SM MSCs injected into regenerating skeletal muscle takes place in the absence of nuclear fusion in a multistep process where myogenic commitment happens very early and before integration into host myofibers (12). However, we cannot rule out the possibility that under different experimental conditions, the expression of human MyHC-IIx/d may be the consequence of reprogramming of human nuclei secondary to fusion into host muscle fibers (21).
The molecular markers identified so far as being predictive of the in vivo stable-cartilage–forming capacity are specific to ACDC populations (6) and, under our experimental conditions, appear not to be directly applicable to other chondrogenic cell populations, such as SM MSCs. This emphasizes the basic question related to the identification of the signaling pathways that are necessary and sufficient to obtain and maintain the stable-chondrocyte phenotype (22).
The clinical relevance of our findings pertains to the use of MSCs for articular cartilage repair. Although SM MSCs can be converted into a cartilage phenotype (9), our data indicate that predifferentiation in vitro is not sufficient to guarantee stable lineage commitment and restriction of differentiation. Indeed, the cartilage-like phenotype induced in vitro was not stable in nonchondrogenic environments and did not prevent muscle differentiation when the cells were implanted into skeletal muscle. The stabilization of cell lineage commitment is a challenge in modern tissue engineering and can be achieved by means of signaling molecules, genetic manipulations, biomechanics, and the implantation of cells within 3-dimensional (bioactive) scaffolds (23). Finally, we are aware that the mouse assay for stable-cartilage formation is not representative of the joint environment. Clearly, studies in animal models of joint surface defect repair (24) are necessary to evaluate the potential use of SM MSCs and their phenotypic behavior within the articular cartilage microenvironment.