**4. Discussion**

Biomaterial-guided gene delivery using clinically adapted rAAV vectors [10,11,24–33] is an emerging, potent approach to treat focal cartilage lesions by non-invasive transfer and overexpression of chondroregenerative factors. In the present study, we examined the feasibility of providing independent rAAV constructs coding for the highly chondroreparative SOX9 transcription factor [43] and TGF-β [5,6] to hMSCs via carbon dots (CDs) as a means to stimulate the biological activities in these cells, an advantageous source of progenitor cells to enhance the intrinsic healing processes in sites of cartilage damage [5–7].

The present findings show, for the first time to our best knowledge, that CDs may be effective systems to successfully formulate and release rAAV gene transfer vectors. Among all the CDs tested here, CD-2, a carbonaceous nanoparticle prepared using pyrolysis at normal pressure of a mixture of CA, mPEG550, and DMEDA, allowed for the highest intracellular vector release with a good over time maintenance for at least 10 days, the longest time point examined. Equally important, CD-2 was able to promote the effective and sustained modification of hMSCs when used to deliver a reporter (rAAV-*lacZ*) gene vector for at least 10 days (up to 2.2-fold increase in *lacZ* expression relative to free vector treatment) in a safe manner (100% cell viability, presumably due to the presence of the PEG protective shield around the particles), reaching levels similar to those noted with other nano-sized systems for rAAV delivery in hMSCs [24]. In contrast, the genetic modification of hMSCs using CD-3 or CD-4 was associated with decreased levels of cell viability, while CD-1 led to a reduction of gene transfer efficiency relative to free vector administration and other control conditions.

The results next demonstrate that the optimal CD-2 nanoparticles were further capable of promoting the delivery of rAAV vectors coding for the therapeutic *sox9* and TGF-β candidate genes in hMSCs, promoting a significant overexpression of each transgene in the cells over an extended period of time (about 97.5% SOX9<sup>+</sup> cells using rAAV-FLAG-h*sox9*/CD-2 and 79.8% TGF-β<sup>+</sup> cells with rAAV-hTGF-β/CD-2 after 21 days) relative to control treatments (≤7.8% and ≤11.8% transgene-expressing cells in the -/CD-2 and rAAV-*lacZ*/CD-2 conditions, respectively), higher than upon free rAAV *sox9* application (80–85%) [44] and comparable to free TGF-β gene transfer (80%) [45]. Yet, the levels of TGF-β produced via rAAV-hTGF-β/CD-2 (155–225 pg/mL) were 4- to 56-fold higher than those achieved upon free rAAV TGF-β gene transfer (17–24 pg/mL) [45]. This result is probably due to the difference of vector doses applied (MOI = 80–133 here compared with MOI = 4–20 using free vector gene administration, i.e., a 4- to 33-fold difference), but it reflects the improvement of TGF-β production via CD-2-guided rAAV gene transfer, as application of the current vector dose in a free form would have only raised 70-160 pg/mL of growth factor in cells versus 155–225 pg/mL here via CD-2 (i.e., a 1.4- to 2.2-fold difference). Interestingly, application of rAAV-hTGF-β/CD-2 resulted in the detection of 52.8% SOX9<sup>+</sup> cells, probably due to an upregulation of SOX9 expression in response to TGF-β production via rAAV/CD-2, as previously noted when using TGF-β in its recombinant form (rTGF-β) [48] or upon free rAAV TGF-β gene transfer [45], while no effects of SOX9 overexpression were seen on the levels of TGF-β. Effective SOX9 and TGF-β overexpression via CD-2-guided gene delivery led to increased levels of cartilage matrix production in the cells (glycosaminoglycan and type-II collagen expression) over time (21 days) relative to the control conditions, concordant with the respective pro-anabolic activities of SOX9 [43] and TGF-β [5–7], with observations showing short-term effects only of nonviral SOX9 gene transfer using arginine-based CDs (14 days) [42], and with our previous findings using free rAAV *sox9* or TGF-β gene transfer [44,45]. Furthermore, application of rAAV-hTGF-β/CD-2 had a significant influence on hMSC proliferation, in good agreement with the properties of the growth factor [6] and with our previous observations using free rAAV TGF-β gene transfer [45]. In contrast, rAAV-FLAG-h*sox9*/CD-2 had no impact on such a process, consistent with the activities of SOX9 [49] and with our findings via free rAAV *sox9* gene transfer [44]. Interestingly, CD-2-guided delivery of either rAAV-FLAG-h*sox9* or rAAV-hTGF-β advantageously prevented the deposition of type-I and -X collagen in hMSCs over time versus control treatments, concordant with the effects of SOX9 [49] and with results obtained using free rAAV *sox9* gene transfer [44], but in contrast to findings using rTGF-β [5] or upon free rAAV TGF-β gene delivery [45]. This might be due to differences of culture conditions and cell environment (monolayer hMSC cultures here versus three-dimensional hMSC cultures in free rAAV TGF-β gene transfer setting) [45] or to the differences between the levels of TGF-β achieved here via rAAV-hTGF-β/CD-2 (155–225 pg/mL) and the amounts of rTGF-β applied elsewhere (10 ng/mL, i.e., a 44- to 65-fold difference) [5].

In conclusion, the present work reports the possibility of transferring therapeutic rAAV (SOX9 or TGF-β) gene vectors to reparative hMSCs using optimal carbon-based nanoparticles as a novel, off-the-shelf system for cartilage repair. It will be interesting to extend the current approach in

the future using adipose-derived hMSCs as these cells can be harvested at a 1000-fold higher yield in a less invasive manner than bone marrow-derived MSCs while displaying longer life-span and higher proliferative capacity and carrying micro-RNAs that regulate tissue inflammation and cell interplays [50,51]. Analyses are ongoing to test the value of the approach in a three-dimensional environment (high-density cultures) using single and combined CD-2-assisted rAAV SOX9/TGF-β gene transfer to potentiate the effects of the two factors on cell proliferation (TGF-β) and matrix deposition (glycosaminoglycans with TGF-β superiority and type-II collagen with SOX9 superiority) [52] and next in an orthotopic in vivo model of cartilage defect [14,16,53,54]. Such evaluations will provide insights into the potential benefits of CDs over other scaffolds (collagen, hyaluronic acid) or treatments like autologous platelet-rich plasma [50,55,56] for translational cartilage regeneration. Overall, this evaluation provides original evidence on the ability of CD-guided therapeutic rAAV gene transfer in regenerative hMSCs as platforms for therapy of cartilage defects in translational protocols.

**Author Contributions:** Conceptualization, J.K.V. and M.C. (Magali Cucchiarini); methodology, W.M., A.R.-R., M.C. (Mickaël Claudel), G.S., and S.S.-M.; software, W.M. and A.R.-R.; validation, formal analysis, investigation, data curation and visualization, W.M., A.R.-R., M.C. (Mickaël Claudel), G.S., S.S.-M., F.P., L.L., J.K.V. and M.C. (Magali Cucchiarini); writing—original draft preparation, W.M., A.R.-R., F.P., L.L., J.K.V. and M.C. (Magali Cucchiarini); writing—review and editing, W.M., A.R.-R., M.C. (Mickaël Claudel), G.S., S.S.-M., F.P., L.L., J.K.V. and M.C. (Magali Cucchiarini); resources and supervision, F.P., L.L., J.K.V. and M.C. (Magali Cucchiarini). All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We thank R. J. Samulski (The Gene Therapy Center, University of North Carolina, Chapel Hill, NC), X. Xiao (The Gene Therapy Center, University of Pittsburgh, Pittsburgh, PA), G. Scherer (Institute for Human Genetics and Anthropology, Albert-Ludwig University, Freiburg, Germany) for the human *sox9* cDNA, and E. F. Terwilliger (Division of Experimental Medicine, Harvard Institutes of Medicine and Beth Israel Deaconess Medical Center, Boston, MA, USA) for providing the genomic AAV-2 plasmid clones and the 293 cell line. We acknowledge support by the Saarland University within the funding programme Open Access Publishing.

**Conflicts of Interest:** The authors declare no conflicts of interest.
