Chondrocyte Isolation from Loose Bodies—An Option for Reducing Donor Site Morbidity for Autologous Chondrocyte Implantation

Loose bodies (LBs) from patients with osteochondritis dissecans (OCD) are usually removed and discarded during surgical treatment of the defect. In this study, we address the question of whether these LBs contain sufficient viable and functional chondrocytes that could serve as a source for autologous chondrocyte implantation (ACI) and how the required prolonged in vitro expansion affects their phenotype. Chondrocytes were isolated from LBs of 18 patients and compared with control chondrocyte from non-weight-bearing joint regions (n = 7) and bone marrow mesenchymal stromal cells (BMSCs, n = 6) obtained during primary arthroplasty. No significant differences in the initial cell yield per isolation and the expression of the chondrocyte progenitor cell markers CD44 + /CD146+ were found between chondrocyte populations from LBs (LB-CH) and control patients (Ctrl-CH). During long-term expansion, LB-CH exhibited comparable viability and proliferation rates to control cells and no ultimate cell cycle arrest was observed within 12 passages respectively 15.3 ± 1.1 mean cumulative populations doublings (CPD). The chondrogenic differentiation potential was comparable between LB-CH and Ctrl-CH, but both groups showed a significantly higher ability to form a hyaline cartilage matrix in vitro than BMSC. Our data suggest that LBs are a promising cell source for obtaining qualitatively and quantitatively suitable chondrocytes for therapeutic applications, thereby circumventing donor site morbidity as a consequence of the biopsies required for the current ACI procedure.


Introduction
The most common disorder in which loose bodies occur is osteochondritis dissecans (OCD). It might result most commonly in pain, predominantly in late stages, due to physical activities. Other symptoms include swelling, joint locking or a decrease in the range of motion. Overall, there are no general typical clinical signs [1]. This disease predominantly

Histological Analysis of a Loose Body
A qualitative histological analysis of the native loose body specimens shows a typical tri-zone of organization of cartilage tissue (Figures 1 and A1). This starts with the outer superficial zone, e.g., clearly visible in the elastic staining and the outstretched chondrocytes parallel to the border. It is followed by the transitional zone with rounded chondrocytes and a higher collagen (Movat's Pentachrome) content. The highest proteoglycan (Alcian blue) and cartilage (Safranin O/Movat's Pentachrome) prevalence are seen in the inner radial zone [25]. Nuclei are seen well distributed in all zones in bright red (Alcian blue), blue black to purple (Hematoxylin & Eosin) or brown to black (Elastica) depending on the staining. Strikingly, in the Movat's Pentachrome staining some outer regions of the LB still show osteoid in red. Cytoplasmic parts stained in pink to red (Hematoxylin & Eosin), brick red (Masson-Goldner Trichrom) and yellow (Elastica). Connective tissue is seen in red (Elastica), yellow to red (Movat's Pentachrome) and green (Masson-Goldner Trichrom). Tissue in general is stained in green via Safranin O/Fast Green. All histological evidence exhibits a healthy and highly organized cartilage structure.
LB still show osteoid in red. Cytoplasmic parts stained in pink to red (Hematoxy Eosin), brick red (Masson-Goldner Trichrom) and yellow (Elastica). Connective tis seen in red (Elastica), yellow to red (Movat´s Pentachrome) and green (Masson-Go Trichrom). Tissue in general is stained in green via Safranin O/Fast Green. All histol evidence exhibits a healthy and highly organized cartilage structure.

Isolation and Long-Term Proliferation
We deliberately chose not to completely digest the cartilage samples (e.g., overn but instead performed a partial digestion of the fragments to facilitate outgrowth an ability. Outgrowth at passage 0 (P0) took 13 ± 6 days for the LB_CH group and 21 ± 4 for the CTRL-CH samples (p = 0.09). All isolated cells showed a small, spindle-sh and fibroblast-like morphology corresponding to that of dedifferentiated chondro ( Figure A2) [26]. Cell yield after P0 was comparable between LB-CH and CTR groups, with a mean cell number/isolation of 2.5 ± 1.0 × 10 5 and 2.7 ± 1.2 x 10 5 , respect Isolated chondrocyte populations from LB and CTRL donors showed comparable v ity (85.8% ± 4.6 and 87.4% ± 6.2, respectively) in P1 and achieved similar total cell nu (5.3 ± 2.1 × 10 6 and 5.6 ± 2.x10 6 cells/isolation, respectively). Regardless of the origi isolated cells maintained the fibroblast-like morphology ( Figure A2). It is worth men ing that the yield could be further increased, either by multiple primary seeding reseeding the fragments after the initial digestion. In this study, fragments were sepa from the single cell suspension using a cell strainer and discarded after the first pas For in vitro expansion, chondrocytes were seeded at 33,333 cells/cm 2 (2.5 × 1 T75 flask) in each passage and subcultured for a constant duration of seven days b being trypsinized and transferred to the next passage. LB-CH and CTRL-CH exh similar proliferation rate and viability throughout the whole expansion process (F 1a). All chondrocyte cultures were maintained for up to 12 passages without ultim reaching the state of cell cycle arrest. Additionally, during the whole long term cu process cell have always been in a subconfluent state. No significant differences number of cumulative PD were observed between LB-CH (15.3 ± 1.1) and CTRL-CH ± 3.6) (PD and CPD, Figure 2a).
To determine the amount of chondrocyte-derived progenitor cells (CDPCs) in cell isolation, the expression of cell surface marker CD44 and CD146 were quantifi flow cytometry [27][28][29]. For this purpose, cells were seeded at fourfold higher den (2000 cells/cm 2 ) than low-density seeding conditions to avoid active selection and p

Isolation and Long-Term Proliferation
We deliberately chose not to completely digest the cartilage samples (e.g., overnight), but instead performed a partial digestion of the fragments to facilitate outgrowth and viability. Outgrowth at passage 0 (P0) took 13 ± 6 days for the LB_CH group and 21 ± 4 days for the CTRL-CH samples (p = 0.09). All isolated cells showed a small, spindle-shaped, and fibroblast-like morphology corresponding to that of dedifferentiated chondrocytes ( Figure A2) [26]. Cell yield after P0 was comparable between LB-CH and CTRL-CH groups, with a mean cell number/isolation of 2.5 ± 1.0 × 10 5 and 2.7 ± 1.2 × 10 5 , respectively. Isolated chondrocyte populations from LB and CTRL donors showed comparable viability (85.8% ± 4.6 and 87.4% ± 6.2, respectively) in P1 and achieved similar total cell number (5.3 ± 2.1 × 10 6 and 5.6 ± 2.5 × 10 6 cells/isolation, respectively). Regardless of the origin, the isolated cells maintained the fibroblast-like morphology ( Figure A2). It is worth mentioning that the yield could be further increased, either by multiple primary seeding or by reseeding the fragments after the initial digestion. In this study, fragments were separated from the single cell suspension using a cell strainer and discarded after the first passage.
For in vitro expansion, chondrocytes were seeded at 33,333 cells/cm 2 (2.5 × 10 5 per T75 flask) in each passage and subcultured for a constant duration of seven days before being trypsinized and transferred to the next passage. LB-CH and CTRL-CH exhibited similar proliferation rate and viability throughout the whole expansion process (Figure 1a). All chondrocyte cultures were maintained for up to 12 passages without ultimately reaching the state of cell cycle arrest. Additionally, during the whole long term culture process cell have always been in a subconfluent state. No significant differences in the number of cumulative PD were observed between LB-CH (15.3 ± 1.1) and CTRL-CH (17.3 ± 3.6) (PD and CPD, Figure 2a).
To determine the amount of chondrocyte-derived progenitor cells (CDPCs) in each cell isolation, the expression of cell surface marker CD44 and CD146 were quantified by flow cytometry [27][28][29]. For this purpose, cells were seeded at fourfold higher densities (2000 cells/cm 2 ) than low-density seeding conditions to avoid active selection and proliferation of CDPCs [30]. The focus was on the initial purity, the percentage of CDPCs and their change due to long-term cultivation effects under standard conditions (Figure 2b). LB-CH and CTRL-CH show a high expression of the CD44 and CD146 marker in early in vitro cultures (P2: 44.6% ± 13.5 and 54.5% ± 20.2, respectively) and no significant difference were observed between both groups (p = 0.165). The number of CD44 and CD146 double positive cells decreased with increasing cultivation time, so that the amount of CDPCs in both group at P12 were only 30.9% ± 16.6 (LB-CH) and 33.7% ± 10.1 (CTRL-CH).
their change due to long-term cultivation effects under standard conditions (Figur LB-CH and CTRL-CH show a high expression of the CD44 and CD146 marker in ea vitro cultures (P2: 44.6% ± 13.5 and 54.5% ± 20.2, respectively) and no significant diffe were observed between both groups (p = 0.165). The number of CD44 and CD146 d positive cells decreased with increasing cultivation time, so that the amount of CDP both group at P12 were only 30.9% ± 16.6 (LB-CH) and 33.7% ± 10.1 (CTRL-CH).

Apoptosis Rate, Short-Term Proliferation, and Metabolic Activity at Early and Late Pa
To assess the number of apoptotic cells in early (P2) and late (P12) LB-CH, CTRL and BMSC cultures, caspase 3/7 activity assays were performed. This analysis rev significantly higher caspase 3/7 activity levels in early BMSC cultures compared t corresponding early LB-CH (p = 0.0004) and CTRL-CH (p < 0.0001) populations (F 3a). This difference diminished during in vitro culture, so that the different gr showed a comparable apoptosis rate in P12.
To further characterize the impact of in vitro expansion on LB-CH and CTRL-C comparison to BMSCs, the proliferation capacities of early and late cultures were d mined (Figure 3b,c). Early cultures in P2 achieved comparable population doubling within nine days (LB-CH: 1.7 ± 0.3, CRTL-CH: 1.5 ± 0.1, and BMSC: 1.6 ± 0.4). The po tion time [h] showed also no significant difference between the early chondrocyte BMSCs cultures (LB-CH: 56.6 h ± 1.4; CRTL-CH: 57.3 h ± 0.5, and BMSCs: 56.5 h ± 2 line with the low apoptosis rate, late chondrocyte and BMSC cultures in P12 showed parable PD and PD time (LB-CH: 56.9 h ± 1.7 with 1.5 ± 0.3 PD, CTRL-CH: 56.2 h ± 1.7 1.7 ± 0.4 PDs, and BMSCs: 58 h ± 1.5 with 1.3 ± 0.3). Subsequently the metabolic activ the cultures was determined using a PrestoBlue™ Cell Viability assay (Figure 3d) analysis revealed that the metabolic activity is significantly lower in late passage L (p = 0.0118) and BMSC cultures compared to their early passage counterparts, sugge increased cellular senescence. However, no significant differences were observed bet the three different groups at the same passage.

Apoptosis Rate, Short-Term Proliferation, and Metabolic Activity at Early and Late Passage
To assess the number of apoptotic cells in early (P2) and late (P12) LB-CH, CTRL-CH, and BMSC cultures, caspase 3/7 activity assays were performed. This analysis revealed significantly higher caspase 3/7 activity levels in early BMSC cultures compared to the corresponding early LB-CH (p = 0.0004) and CTRL-CH (p < 0.0001) populations ( Figure 3a). This difference diminished during in vitro culture, so that the different groups showed a comparable apoptosis rate in P12.
To further characterize the impact of in vitro expansion on LB-CH and CTRL-CH in comparison to BMSCs, the proliferation capacities of early and late cultures were determined (Figure 3b,c). Early cultures in P2 achieved comparable population doubling rates within nine days (LB-CH: 1.7 ± 0.3, CRTL-CH: 1.5 ± 0.1, and BMSC: 1.6 ± 0.4). The population time [h] showed also no significant difference between the early chondrocyte and BMSCs cultures (LB-CH: 56.6 h ± 1.4; CRTL-CH: 57.3 h ± 0.5, and BMSCs: 56.5 h ± 2.1). In line with the low apoptosis rate, late chondrocyte and BMSC cultures in P12 showed comparable PD and PD time (LB-CH: 56.9 h ± 1.7 with 1.5 ± 0.3 PD, CTRL-CH: 56.2 h ± 1.7 with 1.7 ± 0.4 PDs, and BMSCs: 58 h ± 1.5 with 1.3 ± 0.3). Subsequently the metabolic activity of the cultures was determined using a PrestoBlue™ Cell Viability assay ( Figure 3d). This analysis revealed that the metabolic activity is significantly lower in late passage LB-CH (p = 0.0118) and BMSC cultures compared to their early passage counterparts, suggesting increased cellular senescence. However, no significant differences were observed between the three different groups at the same passage.

Chondrogenic Re-Differentiation
The chondrogenic phenotype and (re-)differentiation potential of chondrocytes and BMSCs was tested using a pellet culture assay and differentiation was induced with specific chondrogenic induction medium (ChM). Cell pellet cultures in ChM without addition of TGF-ß served as negative control.

Chondrogenic Re-Differentiation
The chondrogenic phenotype and (re-)differentiation potential of chondrocytes an BMSCs was tested using a pellet culture assay and differentiation was induced with spe cific chondrogenic induction medium (ChM). Cell pellet cultures in ChM without add tion of TGF-ß served as negative control.
Alcian blue staining for qualitative detection of sulfated glycosaminoglycans (suc as hyaluronic acid or chondroitin sulfate) showed pronounced chondrogenic differentia tion of all TGF-ß-treated cultures compared to the corresponding negative control (with out TGF-ß). Both LB-CH and CRTL-CH exhibited a similar Alcian blue staining an formed larger pellet cultures then BMSC cultures (Figure 4a,b).
Immunohistochemical staining of collagen type II confirmed successful chondro Alcian blue staining for qualitative detection of sulfated glycosaminoglycans (such as hyaluronic acid or chondroitin sulfate) showed pronounced chondrogenic differentiation of all TGF-ß-treated cultures compared to the corresponding negative control (without TGF-ß). Both LB-CH and CRTL-CH exhibited a similar Alcian blue staining and formed larger pellet cultures then BMSC cultures (Figure 4a,b).
Immunohistochemical staining of collagen type II confirmed successful chondrogenic (re-)differentiation of TGF-ß-stimulated LB-CH pellets cultures with a three-zone separation compared to the negative controls. The outermost layer reflecting the superficial zone and is clearly distinguishable from the underlying transition/radial zone. The inner part of the pellets appeared to be less dense and structured, as did the other two zones (Figure 4c). produced significantly more chondrogenic matrix (340.1 ± 194.1 µ g PG per mg total pro tein) as corresponding BMSC cultures (113.5 ± 53.6 µ g PG per mg total protein, p = 0.004 but no difference to CTRL (169 ± 122.7 µ g PG per mg total protein, p = 0.1058) (Figure 4d Late LB-CH and CTRL-CH cultures in P12 were also successfully (re-)differentiated an produced a similar amount of chondrogenic matrix (LB-CH: 263.3 ± 142.3µ g and CTRL CH 300.5 ± 101.8 µ g PG/mg protein) (Figure 4e). Quantitative PCR was used to assess mRNA expression of essential chondrogen ECM components in early passage cultures after (re-)differentiation. Gene expression wa normalized to beta-2-microglobulin as a housekeeper gene, and actin (ACTB) and hypo xanthine-guanine phosphoribosyltransferase (HPRT) served as additional controls (Fig  ure 5). BMSC cultures without TGF-ß served as reference to define baseline expressio To quantitatively assess the chondrogenic (re-)differentiation, the total proteoglycan content was determined (Figure 4d,e). This analysis revealed that LB-CH cultures in P2 produced significantly more chondrogenic matrix (340.1 ± 194.1 µg PG per mg total protein) as corresponding BMSC cultures (113.5 ± 53.6 µg PG per mg total protein, p = 0.004), but no difference to CTRL (169 ± 122.7 µg PG per mg total protein, p = 0.1058) (Figure 4d). Late LB-CH and CTRL-CH cultures in P12 were also successfully (re-)differentiated and produced a similar amount of chondrogenic matrix (LB-CH: 263.3 ± 142.3 µg and CTRL-CH 300.5 ± 101.8 µg PG/mg protein) (Figure 4e).
Quantitative PCR was used to assess mRNA expression of essential chondrogenic ECM components in early passage cultures after (re-)differentiation. Gene expression was normalized to beta-2-microglobulin as a housekeeper gene, and actin (ACTB) and hypoxanthine-guanine phosphoribosyltransferase (HPRT) served as additional controls ( Figure 5). BMSC cultures without TGF-ß served as reference to define baseline expression values of the undifferentiated cell state. In line the total proteoglycan content, LB-CH exhibited significantly higher aggrecan and collagen II mRNA expression compared to BMSC (ACAN p = 0.0263; COL II p = 0.0104). While no significant difference in mRNA expression were found between LB-CH and CTRL-CH, both groups showed significant higher SOX9 expression compared to BMSCs. In contrast, the mRNA expression of the osteochondral marker collagen X was significantly higher in BMSCs compared to both LB-CH and CTRL-CH (p = 0.0108 and p = 0.048). Similarly, the expression of MMP13 was elevated in BMSCs compared to both chondrocyte cultures, but did not reach statistical significance.
values of the undifferentiated cell state. In line the total proteoglycan content, LB-CH e hibited significantly higher aggrecan and collagen II mRNA expression compared BMSC (ACAN p = 0.0263; COL II p = 0.0104). While no significant difference in mRN expression were found between LB-CH and CTRL-CH, both groups showed significa higher SOX9 expression compared to BMSCs. In contrast, the mRNA expression of t osteochondral marker collagen X was significantly higher in BMSCs compared to both L CH and CTRL-CH (p = 0.0108 and p = 0.048). Similarly, the expression of MMP13 w elevated in BMSCs compared to both chondrocyte cultures, but did not reach statistic significance.
In summary, these analyses showed that LB-CH exhibited a similar chondrogen phenotype as the corresponding CTRL-CH, and maintained their differentiation capaci even after prolonged in vitro culture.

Discussion
Although ACI for e.g., OCD treatment is hitherto the optimal solution, it is rath ironic to apply additional damage to a patient´s cartilage to harvest cells to heal the p mary cartilage defect. Especially given the fact that a massive cell source (the loose bod is discarded in that process. Even after longer time periods of detachment the in vivo su vival and supply of nutrition for LBs is archived via the synovial fluid [31] and active by the subchondral bone as shown in animal studies [32,33]. Some other publications r port that it is not possible to establish chondrocyte cultures from all LBs, especially patients whose disease onset dates back more than one year [34]. We were able to isola chondrocytes from all LBs obtained and only in two exceptional cases had to discard is lated cells because of early bacterial contamination. Compared with chondrocytes fro healthy cartilage regions, the isolated cells showed an excellent proliferation rate, cartila marker expression, and chondrogenic matrix production and were even superior to h man BMSCs. The reason for the difference between our results and those of others may that a longer overnight digestion was chosen in these studies, which in our experien In summary, these analyses showed that LB-CH exhibited a similar chondrogenic phenotype as the corresponding CTRL-CH, and maintained their differentiation capacity even after prolonged in vitro culture.

Discussion
Although ACI for e.g., OCD treatment is hitherto the optimal solution, it is rather ironic to apply additional damage to a patient's cartilage to harvest cells to heal the primary cartilage defect. Especially given the fact that a massive cell source (the loose body) is discarded in that process. Even after longer time periods of detachment the in vivo survival and supply of nutrition for LBs is archived via the synovial fluid [31] and actively by the subchondral bone as shown in animal studies [32,33]. Some other publications report that it is not possible to establish chondrocyte cultures from all LBs, especially in patients whose disease onset dates back more than one year [34]. We were able to isolate chondrocytes from all LBs obtained and only in two exceptional cases had to discard isolated cells because of early bacterial contamination. Compared with chondrocytes from healthy cartilage regions, the isolated cells showed an excellent proliferation rate, cartilage marker expression, and chondrogenic matrix production and were even superior to human BMSCs. The reason for the difference between our results and those of others may be that a longer overnight digestion was chosen in these studies, which in our experience may lead to decreased viability [34]. In addition, the isolated cell solutions are often subsequently filtered, which may also lead to damage or destruction of the cells by shear forces, especially after the prolonged digestion process.
The limitation of our study is, of course, the origin and size of the control (CTRL) group. Even though we isolated the control cells from non-weight-bearing healthy cartilage areas, a perfect control cohort would have included biopsies derived from each LB patient individually, which is not feasible considering ethical criteria and would entail additional damage to the patients (Table 1). Although appropriate biopsies for standard ACI were also taken from the patients in the LB group and cells were isolated under GMP conditions, we could not include them in our research as they were completely required for the treatment according to the routine protocol. The aim of our work was to circumvent the collection of cartilage biopsies from intact (presumably healthy) non-loaded areas of the knee joint by using cells from the LB. Other treatment options are of course available for these patients (e.g., microfracture, etc.), but several clinical trials (including phase III) show that ACI is superior to these alternative procedures [35]. A first insight into the feasibility of using LB as a cell source for ACI was delivered by our qualitative histological analysis of native loose bodies and their corresponding synovial pocket (Figures 1 and A1). These revealed an intact and healthy cartilage structure of the loose bodies. After successful isolation, in high quantities, the long-term cultivation and analysis were conducted to simulate long-term survival and the impact of in vitro aging as an approximation for an in vivo survival. This represents the foundation for any successful long-term therapy. Surprisingly, we did not detect growth arrest in long-term settings, or any differences between the LB and CTRL group (Figure 2a).
Analysis of CD44, hyaluronan receptor, marker expression reflects the purity and potential of the isolated cells [36,37]. The high rate of CD44/CD146 positive cells shows that the isolation protocol yields high potential chondrocytes. As described by Jiang and colleagues [30] we checked early and late passages of LB and CTRL chondrocytes for chondrocyte-derived progenitor cells via the expression of the CD146 marker (Figure 2b.). Positive cells are postulated to have a greater regenerative potential as compared to CD146 negative cells. Surprisingly we saw that in the CTRL cohort there is a significant drop in positive cells (20.8% p = 0.037) between passage 2 and 10 but not in LB (13.7% p = 0.058). In respect of this, the detrimental effect of in vitro aging correlates with our data of redifferentiation and its markers. For BMSC the high impact of in vitro aging was shown by our group before [38], surprisingly in low passage (≤4) chondrocytes show only an in vitro difference based upon their donor age [39].
We then went on to confirm cell survival with a specific focus on long time-culture periods (P12) and in short-term culture (P2). This was carried out via checking for the apoptosis hallmark cascade of caspase 3/7 (Figure 3a). Here BMSC demonstrate a significant higher base level in P2 compared to LB and CTRL and to a lesser, non-significant, extend in the high passage (P12). No group changed significantly from P2 to P12, which supports the long-term data as well as all microscopical observations. A crucial timeframe for therapies are always the first passages (P1-P2) for massive ex vivo expansion followed by the re-implantation procedure. We therefore checked extensively for short-term effects such as population doublings and time and metabolic activity, as a marker for viability. These were conducted over nine days in the early passages as well as in the late passages for long-term representation. No significant differences were observed in early or late passages. Surprisingly, BMSC had the longest population doubling time of all three groups in the late passage (Figure 3b,c). Metabolic activity as an indicator for viability was highest in early passage BMSC and the only significances were found between early and late passage, as expected, likely due to in vitro aging effects (Figure 3d).
Proteoglycan measurement after the re-differentiation process was carried out again in early and late passages. The only significance in proteoglycan levels could be detected between the LB and BMSC group in the early passage (P2) (Figure 4d,e). The high patient's variance in the LB cohort is most likely due to the age and loose body variety (size, duration disease state, pathophysiology etc.) in this group ( Table 1).
As quality has to be chosen over quantity, in our study ex vivo expansion was never a limiting factor, and the re-differentiation process yielded very good results for LB in the crucial early passage. Additionally, we histologically confirmed the quality after the re-differentiation via Alcian blue (proteoglycans) and collagen II staining (Figure 4a-c). Overall, both cell sources were viable and capable of proliferation even until P12 (Figures 2 and 3).
The significantly higher aggrecan and SOX9 mRNA levels of LB correlate with the higher proteoglycan levels of early chondrogenic markers, which indicate an advanced progression in the differentiation process for LB and CTRL cells in comparison to BMSC ( Figure 5) [40]. MMP13 levels in BMSC indicate an inferior quality of cartilage in terms of hypertrophy. Evidence suggests that BMSCs tend to form hypertrophic cartilage [41]. Our study supports this idea as we see a high COL X/COL II ratio and higher MMP13 levels in BMSC, which reflects transient cartilage. This seems logical if one considers the "true" nature of BMSCs as a pre-state for bone (regeneration). So the value of a BMSC-based therapy in cartilage regeneration might be by default inadequate in terms of endochondral ossification [22]. An allogenic chondrocyte therapy seems much more promising in producing the proper kind of cartilage. Furthermore, the in vitro performances of BMSCs were inferior within most parameters of our study. Our results support the conclusion that BMSCs are less suitable for cartilage reconstruction therapies and that cells of chondrogenic origin should be used for cartilage defect therapy.
Our data support the usage of loose bodies as a cell source under specific isolation conditions (partial digestion and sub-standard seeding densities), which leads to an excellent ex vivo amplification. This underlines the importance of choosing the right cell source for the right therapeutical approach to achieve the most beneficial outcome for the patient with the lowest drawbacks.

Ethics Statement
Patient recruitment and sample harvest was approved by the local ethical committee (EA2/089/20), and all donors gave written informed consent.

Chondrocyte Isolation and Culture
Loose bodies of 18 patients (mean age 31.2 ± 11.2 years), which were not eligible for re-fixation of the fragment, were included in this study. Arthroscopy of the knee was performed, and the detached osteochondral fragment was removed. Chondrocytes from healthy non-weight-bearing cartilage regions of patients undergoing a knee TEP (control, CTRL) served as references (n = 5; mean age 64.2 ± 7.8 years). All specimens were harvested under sterile conditions and stored in sterile tubes filled with phosphate-buffered saline (PBS) until further use. Processing started immediately after cartilage harvest. Therefore, isolated cartilage was cut into small fragments of approximately 2-4 mm diameter and digested sequentially, first using 350 U/mL pronase E (Sigma Aldrich, St. Louis, MO, USA) and then 100 U/mL collagenase type II (Biochrom, Berlin, Germany), each 1 h at 37 • C, under continues agitation, respectively. Cartilage fragments were centrifuged (300× g, 15 min), re-suspended in cell culture medium and seeded on culture flasks. The cell culture medium was Dulbecco's modified Eagle's medium (DMEM) low glucose (Sigma Aldrich, D5546, USA) with 10% fetal calf serum 'superior' (Biochrom, S0615, Germany), 2 mM GlutaMAX™ (Gibco, Thermo Fisher Scientific, Waltham, MA, USA), penicillin (10 U/mL)/streptomycin (10 µg/mL) (Biochrom, Germany). Sub-cultured cells were seeded at 2400 cells/cm 2 in cell culture flasks. Cell number and viability of chondrocytes and BMSCs were determined using the cell counter CASY TT (OLS OMNI Life Science, Bremen, Germany). All assays were carried out with primary cells either in passages 2-4 or 10-12, respectively.

BMSC Isolation and Culture
Primary age-and gender-matched, to the CTRL group, human bone marrow-derived BMSCs (n = 6; mean age 57.7 ± 22.1 years), isolated as described [42] were used as an additional reference. The supplemented DMEM cell culture medium was the same as for chondrocytes above. Cells were counted by using CasyTT for standard cell culture and cryopreserved until further use.

Viability and Proliferation
Cell viability was measured by using PrestoBlue ® cell viability reagent (Life Technologies, Carlsbad, CA, USA). Proliferation rates and cell population doublings (PD) were acquired by using a CyQUANT ® cell proliferation assay kit (Life Technologies, USA). Assays were conducted according to the manufacturer's instructions. In brief, 2500 cells/cm 2 were seeded per 48-well for each individual time point, starting 24 h after seeding marked the time points zero (d0), day three (d3), day six (d6) and day nine (d9). Media change was performed every three days. All determinations were executed in triplicates or quadruplicates using a multimode microplate reader (m200 pro, Tecan, Maennedorf, Switzerland). Relative fluorescence units (RFU) were used to calculate population doublings and time. Long-term cumulative population doublings were determined by plating defined cell numbers per cm 2 (2.5 × 10 5 per T75 cell culture flask). After seven days, cells were trypsinized and counted with a cell counter CASY TT (OLS OMNI Life Science, Bremen, Germany).

RNA Extraction and qPCR
After 21 days of chondrogenic differentiation, two pellets were pooled with 250 µL TRIzol ® (Thermo Fisher Scientific, USA) within a 2.0 mL screw cap micro tube (Sarstedt, Nuembrecht, Germany). Three 2.8 mm and twenty-five 1.4 mm Precellys ceramic beads (Peqlab, Erlangen, Germany) were included. A Minilys homogenizer (Peqlab, Erlangen, Germany) was used for mechanical decomposition of the pellets (3 × 1 min, max speed). Afterwards, the solution was transferred to a new 1.5 mL Eppendorf tube and each screw cap micro tube was washed three times with 250 µL TRIzol ® . RNA was than extracted from each pool (1 mL) according to the TRIzol ® manufacturer's instructions. Glycogen (10 µg) was used to increase the yield and visibility of the RNA pellet, which was fi-nally reconstituted in 20 µL RNase-free water. RNA concentration was determined using a NanoDrop 1000 (Thermo Fisher Scientific, USA). RNA (500 ng) was then transcribed with iScript™ cDNA synthesis kit (Bio-Rad, Hercules, CA, USA) according to the manual. Next, qPCR was performed using the LightCycler ® 480 SYBR Green I Master Mix (Roche, Basel, Switzerland) with 5 ng cDNA and 200 nM primers (Table A1). All qPCR runs were executed on a LightCycler ® 480 II machine (Roche; Switzerland). Further analysis was carried out with a LightCycler ® 480 SW 1.5.1 software and the fit point analysis method with a fixed threshold of 1.0 for all samples. Normalization was conducted by relating the genes to the housekeeping gene beta-2 microglobulin (b2 MG) yielding ∆Ct. Afterwards each gene was normalized to the corresponding gene of the TGF-ß negative BMSC culture resulting in the ∆∆Ct.

Proteoglycan Determination
As described before by Davis et al. [43], we extracted proteoglycan (PG) from two pooled pellets of each condition, as in the RNA preparation, except that 150 µL proteoglycan extraction buffer (PEB) was used instead of TRIzol ® . Immediately after adding the DMMB assay reagent, absorption at 516 nm was acquired on a Tecan microplate reader. PG data was normalized to the total protein content of each sample, determined with a Coomassie (Bradford) protein assay kit (Thermo Fisher Scientific, USA) according to the manual.

Histology and Immunohistology
Four µm sections of formalin-fixed, paraffin-embedded tissue were used for analysis. Automated hematoxylin and eosin (H&E) staining was carried out in a linear slide stainer (Leica ST4040) using Mayer's Haemalaun (Merck; 1.09249.2500, Rahway, NJ, USA) and Eosin Y (Merck; 1.15935.0100) [46]. Movat's Pentachrome (Verhoeff) staining was performed according to the manufacturer's protocol (Morphisto 12061, Offenbach, Germany). Safranin O/Fast Green staining was performed according to a standard protocol. Briefly, after deparaffinising and hydrating the paraffin sections, they were incubated in Weigert's Iron Hematoxylin (Hematoxylin, 517-28-2; Ferric Chloride 7705-08-0, Merck) for five minutes and afterwards washed in distilled water three times. The sections were then differentiated in 1% acid-alcohol for 10 s and rinsed in distilled water three times. Subsequently the sections were incubated for one minute in 0.2% Fast Green (Morphisto 10267), 15 s in 1.0% acetic acid, followed by 30 min in 1.0% Safranin O (477-23-6, Merck). Afterwards the slides were briefly rinsed in 96% ethanol and dehydrated with two changes of 96% ethanol and 100% ethanol. Afterwards the sections were washed in acetic acid n-butyl ester (P036.1 Roth) and mounted. Cartilage (proteoglycan specific) stains bright red and the backgrounds stains in green [47]. For Masson-Goldner trichrome and elastica staining, kits were used according to the manufacturer's recommendations (Morphisto 12043, Morphisto 14604) [48]. Alcian blue/nuclear fast red staining was performed according to a standard protocol, see below. Slides were scanned automatically in 20× or 40× magnification using the VS-120-L Olympus slide scanner 100-W system. Pictures were processed using the Olympus VS-ASW-L100 program and OlyVIA software.
One pellet of each differentiation was paraffin-embedded. Slices (3 µm) of the midsection were further processed. One set was stained with Alcian blue for glycosaminoglycans, as described by Yang et al. [49]. Counterstaining was performed with nuclear fast red. The other set of slices was processed for collagen type II immunohistochemistry. Briefly, samples on slides were deparaffinized, treated with hyaluronidase (0.02% in PBS) for 2 h at 37 • C, digested with pepsin (0.1% in 0.01 M HCl) for 30 min at 37 • C, washed, and then blocked with 2% BSA/2% normal goat serum in PBS. Primary collagen II antibody (Quartett, Potsdam, Germany, Clone 2B1.5) was diluted 1: 50 in 2% BSA/2% normal goat serum in PBS and incubated overnight at 4 • C. After washing, goat anti-mouse Alexa Fluor ® 488 (Thermo Fisher Scientific, USA) secondary antibody was added (1: 200, in PBS with 2% BSA/2% normal goat serum) for 1 h at RT. Nuclei were stained afterwards with 4 ,6-diamidin-2-phenylindol (DAPI 1: 1500 in distilled water) for 10 min. Microscope slides were covered with Fluoromount G (Southern Biotech, Birmingham, AL, USA) and glass slides. All pictures were taken with a Zeiss Axio Scope (Oberkochen, Germany) microscope.

Statistical Analysis
All statistical analysis was performed with the software Prism 9.41 (Dotmatics). Oneway ANOVA with multiple comparisons and Tukey compensation was utilized for all analysis, except for the CPD analysis, which was performed using a two-way ANOVA with multiple comparison and Sidak compensation for more robustness.

Conclusions
Loose bodies show no impaired behavior compared to primary cartilage. This might make them a promising tissue source and efficient alternative for harvesting otherwise intact cartilage for ACI based upon our results, which would also circumvent the need of further biopsies and donor side morbidity. The latter is of high importance considering the immense prevalence of e.g., osteochondritis dissecans of young and sport-active adolescents.