Transplantation of progenitor cells like mesenchymal stem cells (MSCs) from the bone marrow, with an innate potential for chondrogenic differentiation, is a promising strategy to treat articular cartilage defects in patients . Yet, the use of MSCs in such settings is still restrained by the low percentage of cells that enter appropriate chondrocyte lineage-differentiation pathways to produce a reparative tissue of proper quality. It is well known that MSC-derived cells tend to undergo premature terminal differentiation, hypertrophy, and ossification [4, 5]. Such limitations might be overcome by directing the cells toward an adequate phenotype by application and stable expression of candidate genes capable of controlling chondrocyte differentiation. Among the potentially beneficial agents, the transcription factor SOX9 is a strong candidate to adjust chondrogenesis, as a key regulator of chondrocyte differentiation and cartilage formation  that can delay hypertrophic maturation at certain stages of differentiation [23–27]. Instead of using classic nonviral [26, 30], adeno-, retro-, and lentiviral vectors [24, 31–33], we focused on rAAV systems that advantageously genetically modify hMSCs at very high efficiencies and for extended periods without affecting their potential for differentiation [8, 16, 34]. This finding was confirmed here when applying the rAAV-FLAG-hSOX9 vector to undifferentiated monolayer and chondrogenically differentiated hMSC cultures (70% to 85% transduction efficiencies for up to 21 days, with about a ninefold difference in SOX9 expression levels compared with control treatments that showed a similar evolution for all parameters in the evaluation). Equally important, we further demonstrate that the efficient and sustained SOX9 expression levels achieved here with rAAV were capable of promoting and enhancing chondrogenic differentiation of hMSCs in suitable aggregate cultures, with an increased production of major extracellular matrix components (proteoglycans, type II collagen) compared with control conditions, as already seen in human osteoarthritic chondrocytes  and in agreement with the properties of this factor [22, 24, 26, 27, 31, 32]. Interestingly, administration of rAAV SOX9 did not further modify the levels of proliferation and viability of hMSCs in all the systems tested compared with control treatments, as also reported with chondrocytes , and instead, these parameters decreased over the course of the evaluation. It remains to be seen whether too elevated levels of SOX9 expression will not cause toxicity on cells transduced through rAAV . This is, however, consistent with previous observations when expanding similar controls of hMSC transduction  and, more important, with findings showing the lack or opposing effects of SOX9 on the proliferation and cell-cycle progression of hMSCs in the adult [23, 31, 32].
Strikingly, the present results also indicate that prolonged, elevated rAAV-mediated expression of SOX9 significantly reduced the expression and activities of several markers of hypertrophy and terminal or osteogenic differentiation (type I and type X collagen, ALP, MMP13, OP, matrix mineralization), concordant with previous reports showing contrasting effects of SOX9 on osteogenesis, bone formation, terminal differentiation, and calcification and on the expression of these markers [23–27, 31, 52–58]. Remarkably, these effects of SOX9 treatment via rAAV were associated with significant decreases in the levels of RUNX2, a transcription factor essential for bone formation, terminal maturation, and mineralization that stimulates the expression of osteoblast-related genes (COL1A1, COL10A1, ALP, MMP13, OP) [53, 55, 56, 58–62], again in good agreement with the effects of SOX9 on RUNX2 expression [31, 63, 64]. Interestingly, Ikeda et al.  reported that SOX9 gene transfer in hMSCs failed to suppress the expression of such hypertrophic and osteogenic markers. However, it is important to mention that, in this previous study, less efficient adenoviral vectors were used and mediated gene expression for about 5 days, whereas sustained and very high levels of transgene SOX9 expression were detected here for at least 21 days. Of further note, we also observed that application of rAAV-FLAG-hSOX9 led to a decrease in β-catenin expression, a mediator of the Wnt signaling pathway known to stimulate osteoblast lineage differentiation . In addition, we noted that the vector increased the levels of PTHrP, an inhibitor of hypertrophic maturation and calcification that has a significant impact on the regulation of gene expression for COL1A1, COL10A1, ALP, and RUNX2, delays bone formation [15, 53, 55, 56, 58, 66–72], and activates SOX9 transcriptional activities . The effects of SOX9 evidenced here on these signaling pathways are consistent with reports showing enhanced β-catenin degradation and PTHrP activation mediated by the cartilage-specific transcription factor [23, 52, 73].
Altogether, the data demonstrate that concurrent activation and inhibition of different signaling pathways by rAAV SOX9 gene transfer might permit a significant reduction of osteogenic processes in hMSCs. Still, in the present study, evaluations were not performed beyond day 21, and it remains to be seen whether the SOX9-transduced cells will not undergo hypertrophy and terminal or osteogenic differentiation over time if they lose SOX9 expression [28, 29], an issue that might have consequences for an adequate treatment of cartilage lesions. Also noteworthy, the candidate treatment here also significantly decreased the expression of adipogenic markers (accumulation of lipid droplets, LPL, and PPARG2 levels), allowing containing the adipogenic differentiation of hMSCs, again in good agreement with previous findings [27, 57]. To our best knowledge, this is the first evidence showing that overexpression of SOX9 via rAAV stimulates hMSC chondrogenic differentiation with an important delay in terminal differentiation and hypertrophy, while affecting osteogenic and adipogenic differentiation over a continuous period. Apart from SOX9, the use of other members of the SOX family (SOX5, SOX6) [52, 56, 74] might be of further benefit to favor chondrogenic versus osteo-/adipogenic differentiation of hMSCs, as proposed by various groups who delivered the SOX trio by more classic, less efficient nonviral or adenoviral vectors [24, 30]. Delivery of additional factors displaying proliferative activities might be also valuable to generate high numbers of hMSCs for transplantation in articular cartilage defects. Among various agents with such effects, IGF-I , FGF-2 [16, 17], hTERT [19, 20], or Bcl-xL  may be potentially provided along with SOX sequences. Again, rAAV are powerful vectors, as they conveniently permit separate expression of multiple genes at the same time within their targets .
In addition, it will be important to test further whether transplantation of rAAV SOX9-modified MSCs in articular cartilage defects allows for an effective healing of the lesions in vivo, in association with competent chondrogenic differentiation that avoids premature terminal differentiation as noted in vitro. Interesting findings have been reported by Cao et al. , who showed that implantation of MSCs modified to overexpress SOX9 in a polyglycolic acid (PGA) scaffold led to better repair of osteochondral defects in rabbits, although the gene-transfer system used was a relatively low efficiency adenoviral vector compared with rAAV that might prove even more effective because of higher levels and duration of transgene expression. Regarding the value of the present approach for cartilage repair, this strategy with rAAV will have to be translated in rabbit MSCs before transplantation of genetically modified cell platforms within cartilage defects in vivo. Parameters to consider will include the amounts of cells to provide, the potential use of control elements to contain transgene expression (lineage-specific or regulatable promoters), and the selection of the best-suited supportive matrix for cell containment in the lesions. Also, long-term evaluations will be necessary to test the mechanical quality of the repair tissue within the defects, as other cells (periosteum-, perichondrium-, adipose-, muscle-derived stem cells, bone marrow aspirates, tissue grafts, or even chondrocytes) might be applied as engineered platforms [75–78] compared with direct gene-transfer strategies [35, 79, 80].