Anatomically shaped osteochondral constructs for articular cartilage repair

Few successful treatment modalities exist for surface-wide, full-thickness lesions of articular cartilage. Functional tissue engineering offers a great potential for the clinical management of such lesions. Our long-term hypothesis is that anatomically shaped tissue constructs of entire articular layers can be engineered in vitro on a bony substrate, for subsequent implantation. To determine the feasibility, this study investigated the development of bilayered scaffolds of chondrocyte-seeded agarose on natural trabecular bone. In a series of three experiments, bovine chondrocytes were seeded in (1) cylindrical bilayered constructs of agarose and bovine trabecular bone, 0.53 cm² in surface area and 3.2 mm thick, and were cultured for up to 6 weeks; (2) chondrocyte-seeded anatomically shaped agarose constructs reproducing the human patellar articular layer (area=11.7 cm², mean THICKNESS=3.4 mm), cultured for up to 6 weeks; and (3) chondrocyte-seeded anatomically shaped agarose constructs of the patella (same as above) integrated into a corresponding anatomically shaped trabecular bone substrate, cultured for up to 2 weeks. Articular layer geometry, previously acquired from human cadaver joints, was used in conjunction with computer-aided design and manufacturing technology to create these anatomically accurate molds. In all experiments, chondrocytes remained viable over the entire culture period, with the agarose maintaining its shape while remaining firmly attached to the underlying bony substrate (when present). With culture time, the constructs exhibited positive type II collagen staining as well as increased matrix elaboration (Safranin O staining for glycosaminoglycans) and material properties (Young's modulus and aggregate modulus). Despite the use of relatively large agarose constructs partially integrated with trabecular bone, no adverse diffusion limitation effects were observed. Anatomically shaped constructs on a bony substrate may represent a new paradigm in the design of a functional articular cartilage tissue replacement.     

Age-related changes in the composition and mechanical properties of human nasal cartilage

Mutations in the tuberous sclerosis 2 (TSC2) gene product have been genetically linked to the pathology of both tuberous sclerosis (TSC) and the gender-specific lung disease, lymphangioleiomyomatosis (LAM). Both diseases are classified as disorders of cellular migration, proliferation, and differentiation. Earlier studies from our laboratory (1) linked TSC2 with steroid/nuclear receptor signaling. Studies presented here provide evidence for calmodulin (CaM) signaling in the propagation of this TSC2 activity. Far Western screening of a lambda phage human brain cDNA library to identify interacting proteins for the TSC2 gene product (tuberin) yielded multiple clones encoding human CaM. Direct binding with 32P-labeled tuberin demonstrated Ca2+-dependent binding to CaM-Sepharose which was lost upon deletion of the C-terminal 72 residues. The sequence (1740)WIARLRHIKRLRQRIC(1755) was identified as one capable of forming a basic amphipathic helix indicative of CaM binding domains in known calmodulin binding proteins. Studies with a synthetic peptide of this sequence demonstrated very tight Ca2+-dependent binding to CaM as judged by tryptophan fluorescence perturbation studies and phosphodiesterase activation by CaM. Deletion mutagenesis studies further suggested that this CaM binding domain is required for tuberin modulation of steroid receptor function and that mutations in this region may be involved in the pathology of TSC and LAM.     

Applied osmotic loading for promoting development of engineered cartilage

This study investigated the potential use of static osmotic loading as a cartilage tissue engineering strategy for growing clinically relevant grafts from either synovium-derived stem cells (SDSCs) or chondrocytes. Bovine SDSCs and chondrocytes were individually encapsulated in 2% w/v agarose and divided into chondrogenic media of osmolarities 300 (hypotonic), 330 (isotonic), and 400 (hypertonic, physiologic) mOsM for up to 7 weeks. The application of hypertonic media to constructs comprised of SDSCs or chondrocytes led to increased mechanical properties as compared to hypotonic (300 mOsM) or isotonic (330 mOsM) media (p<0.05). Constant exposure of SDSC-seeded constructs to 400 mOsM media from day 0 to day 49 yielded a Young's modulus of 513±89 kPa and GAG content of 7.39±0.52%ww on day 49, well within the range of values of native, immature bovine cartilage. Primary chondrocyte-seeded constructs achieved almost as high a Young's modulus, reaching 487±187 kPa and 6.77±0.54%ww (GAG) for the 400 mOsM condition (day 42). These findings suggest hypertonic loading as a straightforward strategy for 3D cultivation with significant benefits for cartilage tissue engineering strategies. In an effort to understand potential mechanisms responsible for the observed response, cell volume measurements in response to varying osmotic conditions were evaluated in relation to the Boyle–van't Hoff (BVH) law. Results confirmed that chondrocytes behave as perfect osmometers; however SDSCs deviated from the BVH relation.     

Modeling the dynamic composition of engineered cartilage

Mathematical models to describe extracellular matrix (ECM) deposition and scaffold degradation in cell-polymer constructs for the design of engineered cartilage were developed and validated. The ECM deposition model characterized a product-inhibition mechanism in the concentration of cartilage molecules, collagen and glycosaminoglycans (GAG). The scaffold degradation model used first-order kinetics to describe hydrolysis (not limited by diffusion) of biodegradable polyesters, polyglycolic acid and polylactic acid. Each model was fit to published accumulation and degradation data. As experimental validation, cell-polymer constructs (n=24) and unseeded scaffolds (n=24) were cultured in vitro. Biochemical assays for ECM content and measurements of scaffold mass were performed at 1, 2, 4, 6, 8, or 10 weeks (n=8 per time point). The models demonstrated a strong fit with published data and experimental results (R(2)=0.75 to 0.99) and predicted the temporal total construct mass with reasonable accuracy (30% RMS error). This approach can elucidate mechanisms governing accumulation/degradation and may be coupled with structure-function relationships to describe time-dependent changes in construct elastic properties.     

Synthesis rates and binding kinetics of matrix products in engineered cartilage constructs using chondrocyte-seeded agarose gels

Large-sized cartilage constructs suffer from inhomogeneous extracellular matrix deposition due to insufficient nutrient availability. Computational models of nutrient consumption and tissue growth can be utilized as an efficient alternative to experimental trials to optimize the culture of large constructs; models require system-specific growth and consumption parameters. To inform models of the [bovine chondrocyte]−[agarose gel] system, total synthesis rate (matrix accumulation rate+matrix release rate) and matrix retention fractions of glycosaminoglycans (GAG), collagen, and cartilage oligomeric matrix protein (COMP) were measured either in the presence (continuous or transient) or absence of TGF-β3 supplementation. TGF-β3's influences on pyridinoline content and mechanical properties were also measured. Reversible binding kinetic parameters were characterized using computational models. Based on our recent nutrient supplementation work, we measured glucose consumption and critical glucose concentration for tissue growth to computationally simulate the culture of a human patella-sized tissue construct, reproducing the experiment of Hung et al. (2003). Transient TGF-β3 produced the highest GAG synthesis rate, highest GAG retention ratio, and the highest binding affinity; collagen synthesis was elevated in TGF-β3 supplementation groups over control, with the highest binding affinity observed in the transient supplementation group; both COMP synthesis and retention were lower than those for GAG and collagen. These results informed the modeling of GAG deposition within a large patella construct; this computational example was similar to the previous experimental results without further adjustments to modeling parameters. These results suggest that these nutrient consumption and matrix synthesis models are an attractive alternative for optimizing the culture of large-sized constructs.     

In vitro culture increases mechanical stability of human tissue engineered cartilage constructs by prevention of microscale scaffold buckling

Many studies have measured the global compressive properties of tissue engineered (TE) cartilage grown on porous scaffolds. Such scaffolds are known to exhibit strain softening due to local buckling under loading. As matrix is deposited onto these scaffolds, the global compressive properties increase. However the relationship between the amount and distribution of matrix in the scaffold and local buckling is unknown. To address this knowledge gap, we studied how local strain and construct buckling in human TE constructs changes over culture times and GAG content. Confocal elastography techniques and digital image correlation (DIC) were used to measure and record buckling modes and local strains. Receiver operating characteristic (ROC) curves were used to quantify construct buckling. The results from the ROC analysis were placed into Kaplan-Meier survival function curves to establish the probability that any point in a construct buckled. These analysis techniques revealed the presence of buckling at early time points, but bending at later time points. An inverse correlation was observed between the probability of buckling and the total GAG content of each construct. This data suggests that increased GAG content prevents the onset of construct buckling and improves the microscale compressive tissue properties. This increase in GAG deposition leads to enhanced global compressive properties by prevention of microscale buckling.     

Dynamic compression of cartilage constructs engineered from expanded human articular chondrocytes

Recent works have shown that mechanical loading can alter the metabolic activity of chondrocytes cultured in 3D scaffolds. In this study we determined whether the stage of development of engineered cartilaginous constructs (expanded adult human articular chondrocytes/Polyactive foams) regulates the effect of dynamic compression on glycosaminoglycan (GAG) metabolism. Construct maturation depended on the culture time (3-14 days) and the donor (4 individuals). When dynamic compression was subsequently applied for 3 days, changes in GAG synthesized, accumulated, and released were significantly positively correlated to the GAG content of the constructs prior to loading, and resulted in stimulation of GAG formation only in the most developed tissues. Conversely, none of these changes were correlated with the expression of collagen type II mRNA, indicating that the response of chondrocytes to dynamic compression does not depend directly upon the stage of cell differentiation, but rather on the extracellular matrix surrounding the cells.     

Human polymer-based cartilage grafts for the regeneration of articular cartilage defects

The use of autologous chondrocyte implantation (ACI) and its further development combining autologous chondrocytes with bioresorbable matrices may represent a promising new technology for cartilage regeneration in orthopaedic research. Aim of our study was to evaluate the applicability of a resorbable three-dimensional polymer of pure polyglycolic acid (PGA) for the use in human cartilage tissue engineering under autologous conditions. Adult human chondrocytes were expanded in vitro using human serum and were rearranged three-dimensionally in human fibrin and PGA. The capacity of dedifferentiated chondrocytes to re-differentiate was evaluated after two weeks of tissue culture in vitro and after subcutaneous transplantation into nude mice by propidium iodide/fluorescein diacetate (PI/FDA) staining, scanning electron microscopy (SEM), gene expression analysis of typical chondrocyte marker genes and histological staining of proteoglycans and type II collagen. PI/FDA staining and SEM documented that vital human chondrocytes are evenly distributed within the polymer-based cartilage tissue engineering graft. The induction of the typical chondrocyte marker genes including cartilage oligomeric matrix protein (COMP) and cartilage link protein after two weeks of tissue culture indicates the initiation of chondrocyte re-differentiation by three-dimensional assembly in fibrin and PGA. Histological analysis of human cartilage tissue engineering grafts after 6 weeks of subcutaneous transplantation demonstrates the development of the graft towards hyaline cartilage with formation of a cartilaginous matrix comprising type II collagen and proteoglycan. These results suggest that human polymer-based cartilage tissue engineering grafts made of human chondrocytes, human fibrin and PGA are clinically suited for the regeneration of articular cartilage defects.     

Cartilage tissue engineering and bioreactor systems for the cultivation and stimulation of chondrocytes

Damage to and degeneration of articular cartilage is a major health issue in industrialized nations. Articular cartilage has a particularly limited capacity for auto regeneration. At present, there is no established therapy for a sufficiently reliable and durable replacement of damaged articular cartilage. In this, as well as in other areas of regenerative medicine, tissue engineering methods are considered to be a promising therapeutic component. Nevertheless, there remain obstacles to the establishment of tissue-engineered cartilage as a part of the routine therapy for cartilage defects. One necessary aspect of potential tissue engineering-based therapies for cartilage damage that requires both elucidation and progress toward practical solutions is the reliable, cost effective cultivation of suitable tissue. Bioreactors and associated methods and equipment are the tools with which it is hoped that such a supply of tissue-engineered cartilage can be provided. The fact that in vivo adaptive physical stimulation influences chondrocyte function by affecting mechanotransduction leads to the development of specifically designed bioreactor devices that transmit forces like shear, hydrostatic pressure, compression, and combinations thereof to articular and artificial cartilage in vitro. This review summarizes the basic knowledge of chondrocyte biology and cartilage dynamics together with the exploration of the various biophysical principles of cause and effect that have been integrated into bioreactor systems for the cultivation and stimulation of chondrocytes. Dedicated to Prof. K. Arnold on the occasion of his 65th birthday.     

Mechanical Stimulation of Chondrocyte-agarose Hydrogels

Articular cartilage suffers from a limited repair capacity when damaged by mechanical insult or degraded by disease, such as osteoarthritis. To remedy this deficiency, several medical interventions have been developed. One such method is to resurface the damaged area with tissue-engineered cartilage; however, the engineered tissue typically lacks the biochemical properties and durability of native cartilage, questioning its long-term survivability. This limits the application of cartilage tissue engineering to the repair of small focal defects, relying on the surrounding tissue to protect the implanted material. To improve the properties of the developed tissue, mechanical stimulation is a popular method utilized to enhance the synthesis of cartilaginous extracellular matrix as well as the resultant mechanical properties of the engineered tissue. Mechanical stimulation applies forces to the tissue constructs analogous to those experienced in vivo. This is based on the premise that the mechanical environment, in part, regulates the development and maintenance of native tissue^1,2. The most commonly applied form of mechanical stimulation in cartilage tissue engineering is dynamic compression at physiologic strains of approximately 5-20% at a frequency of 1 Hz^1,3. Several studies have investigated the effects of dynamic compression and have shown it to have a positive effect on chondrocyte metabolism and biosynthesis, ultimately affecting the functional properties of the developed tissue^4-8. In this paper, we illustrate the method to mechanically stimulate chondrocyte-agarose hydrogel constructs under dynamic compression and analyze changes in biosynthesis through biochemical and radioisotope assays. This method can also be readily modified to assess any potentially induced changes in cellular response as a result of mechanical stimuli.     

Theoretical analysis of engineered cartilage oxygenation: influence of construct thickness and media flow rate

A novel parallel-plate bioreactor has been shown to modulate the mechanical and biochemical properties of engineered cartilage by the application of fluid-induced shear stress. Flow or perfusion bioreactors may improve tissue development via enhanced transport of nutrients or gases as well as the application of mechanical stimuli, or a combination of these factors. The goal of this study was to complement observed experimental responses to flow by simulating oxygen transport within cartilage constructs of different thicknesses (250 μm or 1 mm). Using numerical computation of convection–diffusion equations, the evaluation of the tissue oxygenation is performed. Four culture conditions are defined based on tissue thickness and flow rates ranging from 0 to ∼25 mL min⁻¹. Under these experimental conditions results show a mean oxygen concentration within the tissue varying from 0.01 to 0.19 mol m⁻³ as a function of the tissue thickness and the magnitude of the applied shear stress. More generally, the influence of shear stress varying (via flow rate modification) from 10⁻³ to 10 dynes cm⁻² on the tissue oxygenation is studied. The influence on the results of important physical parameters such as the maximal oxygen consumption rate of cells is discussed. Lastly, the importance of oxygen concentration in the lower chamber and its relevance to tissue oxygenation are highlighted by the model results.     

Effect of reduced oxygen tension and long-term mechanical stimulation on chondrocyte-polymer constructs

We have investigated the influence of long-term confined dynamic compression and surface motion under low oxygen tension on tissue-engineered cell-scaffold constructs. Porous polyurethane scaffolds (8 mm × 4 mm) were seeded with bovine articular chondrocytes and cultured under normoxic (21% O₂) or hypoxic (5% O₂) conditions for up to 4 weeks. By means of our joint-simulating bioreactor, cyclic axial compression (10–20%; 0.5 Hz) was applied for 1 h daily with a ceramic ball, which simultaneously oscillated over the construct surface (±25°; 0.5 Hz). Culture under reduced oxygen tension resulted in an increase in mRNA levels of type II collagen and aggrecan, whereas the expression of type I collagen was down-regulated at early time points. A higher glycosaminoglycan content was found in hypoxic than in normoxic constructs. Immunohistochemical analysis showed more intense type II and weaker type I collagen staining in hypoxic than in normoxic cultures. Type II collagen gene expression was slightly elevated after short-term loading, whereas aggrecan mRNA levels were not influenced by the applied mechanical stimuli. Of importance, the combination of loading and low oxygen tension resulted in a further down-regulation of collagen type I mRNA expression, contributing to the stabilization of the chondrocytic phenotype. Histological results confirmed the beneficial effect of mechanical loading on chondrocyte matrix synthesis. Thus, mechanical stimulation combined with low oxygen tension is an effective tool for modulating the chondrocytic phenotype and should be considered when chondrocytes or mesenchymal stem cells are cultured and differentiated with the aim of generating cartilage-like tissue in vitro. This work was supported by the Swiss National Science Foundation (grant no. 3200B0-104083).     

Redifferentiation of chondrocytes and cartilage formation under intermittent hydrostatic pressure

Since articular cartilage is subjected to varying loads in vivo and undergoes cyclic hydrostatic pressure during periods of loading, it is hypothesized that mimicking these in vivo conditions can enhance synthesis of important matrix components during cultivation in vitro. Thus, the influence of intermittent loading during redifferentiation of chondrocytes in alginate beads, and during cartilage formation was investigated. A statistically significant increased synthesis of glycosaminoglycan and collagen type II during redifferentiation of chondrocytes embedded in alginate beads, as well as an increase in glycosaminoglycan content of tissue-engineered cartilage, was found compared to control without load. Immunohistological staining indicated qualitatively a high expression of collagen type II for both cases.     

Temporal assessment of ribose treatment on self-assembled articular cartilage constructs

Articular cartilage cannot repair itself in response to degradation from injury or osteoarthritis. As such, there is a substantial clinical need for replacements of damaged cartilage. Tissue engineering aims to fulfill this need by developing replacement tissues in vitro. A major goal of cartilage tissue engineering is to produce tissues with robust biochemical and biomechanical properties. One technique that has been proposed to improve these properties in engineered tissue is the use of non-enzymatic glycation to induce collagen crosslinking, an attractive solution that may avoid the risks of cytotoxicity posed by conventional crosslinking agents such as glutaraldehyde. The objectives of this study were (1) to determine whether continuous application of ribose would enhance biochemical and biomechanical properties of self-assembled articular cartilage constructs, and (2) to identify an optimal time window for continuous ribose treatment. Self-assembled constructs were grown for 4 weeks using a previously established method and were subjected to continuous 7-day treatment with 30 mM ribose during culture weeks 1, 2, 3, or 4, or for the entire 4-week culture. Control constructs were grown in parallel, and all groups were evaluated for gross morphology, histology, cellularity, collagen and sulfated glycosaminoglycan (GAG) content, and compressive and tensile mechanical properties. Compared to control constructs, it was found that treatment with ribose during week 2 and for the entire duration of culture resulted in significant 62% and 40% increases in compressive stiffness, respectively; significant 66% and 44% increases in tensile stiffness; and significant 50% and 126% increases in tensile strength. Similar statistically significant trends were observed for collagen and GAG. In contrast, constructs treated with ribose during week 1 had poorer biochemical and biomechanical properties, although they were significantly larger and more cellular than all other groups. We conclude that non-enzymatic glycation with ribose is an effective method for improving tissue engineered cartilage and that specific temporal intervention windows exist to achieve optimal functional properties.     

Enhancement of cell proliferation and differentiation by combination of ascorbate and dexamethasone in thermo-reversible hydrogel constructs embedded with rabbit chondrocytes

The effects of inducible materials (dexamethasone and ascorbate) on chondrogenic differentiation of rabbit chondrocytes have been examined. A hydrogel construct containing dexamethasone and ascorbate up-regulated gene expression of the cartilage matrix component of collagen to give three times the collagen content per construct at day 56 as compared to controls. Alcian Blue and Safranin-O staining revealed that these constructs also had formed more hyaline cartilage than other hydrogel constructs.     

Articular cartilage regeneration with microfracture and hyaluronic acid

Microfracture used to treat articular cartilage injuries can facilitate access to stem cells in the bone marrow and stimulate cartilage regeneration. However, the regenerated cartilage is fibrocartilage as opposed to hyaline articular cartilage and is thinner than native cartilage. Following microfracture in rabbit knee cartilage defects, application of hyaluronic acid gel resulted in regeneration of a thicker, more hyaline-like cartilage. The addition of transforming growth factor-β3, an inducer of chondrogenic differentiation in mesenchymal stem cells, to the treatment with microfracture and hyaluronic acid did not significantly benefit cartilage regeneration.     

Utility of NucleoCounter for the chondrocyte count in the collagenase digest of human native cartilage

In cartilage tissue engineering, viable cell numbers should be correctly counted in the collagenase digest of the biopsied cartilage. However, this is a difficult task due to the presence of matrix debris, cell ghosts and their aggregates. To search for the correct cell counting method in this situation, we evaluated the utility of an automatic cell counting device, the NucleoCounter, and compared it with conventional staining using the LIVE/DEAD® kit. We first measured the cell numbers of a standard chondrocyte sample by the NucleoCounter, which showed a high accuracy (R² = 0.9999) and reproducibility (%CV: 2.00–8.66). We then calculated the cell numbers and viability in some collagenase digests of native cartilage using either the NucleoCounter or LIVE/DEAD® kit, revealing that the total cell numbers, viable ones and viability were highly correlated between them (R² = 0.9601, 0.9638 and 0.917, respectively). However, both the intrapersonal and interpersonal variabilities in the NucleoCounter was significantly decreased to about 1/20–1/5, compared to that of the LIVE/DEAD® kit. The NucleoCounter was regarded as a useful tool for simple, rapid, and highly reproducible cell counts, which may not only provide constant experimental data in a certain laboratory, but also contribute to the high reproducibility of the clinical results of cartilage tissue engineering among multiple institutions.     

Mesenchymal progenitor cells derived from chorionic villi of human placenta for cartilage tissue engineering

Human mesenchymal stem cells are currently being studied extensively because of their capability for self-renewal and differentiation to various connective tissues, which makes them attractive as cell sources for regenerative medicine. Herein we report the isolation of human placenta-derived mesenchymal cells (hPDMCs) that have the potential to differentiate into various lineages to explore the possibility of using these cells for regeneration of cartilage. We first evaluated the chondrogenesis of hPDMCs in vitro and then embedded the hPDMCs into an atelocollagen gel to make a cartilage-like tissue with chondrogenic induction media. For in vivo assay, preinduced hPDMCs embedded in collagen sponges were subcutaneously implanted into nude mice and also into nude rats with osteochondral defect. The results of these in vivo and in vitro studies suggested that hPDMCs can be one of the possible allogeneic cell sources for tissue engineering of cartilage.     

Development of large engineered cartilage constructs from a small population of cells

Confronted with articular cartilage's limited capacity for self‐repair, joint resurfacing techniques offer an attractive treatment for damaged or diseased tissue. Although tissue engineered cartilage constructs can be created, a substantial number of cells are required to generate sufficient quantities of tissue for the repair of large defects. As routine cell expansion methods tend to elicit negative effects on chondrocyte function, we have developed an approach to generate phenotypically stable, large‐sized engineered constructs (≥3 cm²) directly from a small amount of donor tissue or cells (as little as 20,000 cells to generate a 3 cm² tissue construct). Using rabbit donor tissue, the bioreactor‐cultivated constructs were hyaline‐like in appearance and possessed a biochemical composition similar to native articular cartilage. Longer bioreactor cultivation times resulted in increased matrix deposition and improved mechanical properties determined over a 4 week period. Additionally, as the anatomy of the joint will need to be taken in account to effectively resurface large affected areas, we have also explored the possibility of generating constructs matched to the shape and surface geometry of a defect site through the use of rapid‐prototyped defect tissue culture molds. Similar hyaline‐like tissue constructs were developed that also possessed a high degree of shape correlation to the original defect mold. Future studies will be aimed at determining the effectiveness of this approach to the repair of cartilage defects in an animal model and the creation of large‐sized osteochondral constructs. © 2012 American Institute of Chemical Engineers Biotechnol. Prog., 2013     

Tissue engineering strategies for cartilage generation--micromass and three dimensional cultures using human chondrocytes and a continuous cell line

Utilizing ATDC5 murine chondrogenic cells and human articular chondrocytes, this study sought to develop facile, reproducible three-dimensional models of cartilage generation with the application of tissue engineering strategies, involving biodegradable poly(glycolic acid) scaffolds and rotating wall bioreactors, and micromass pellet cultures. Chondrogenic differentiation, assessed by histology, immunohistochemistry, and gene expression analysis, in ATDC5 and articular chondrocyte pellets was evident by the presence of distinct chondrocytes, expressing Sox-9, aggrecan, and type II collagen, in lacunae embedded in a cartilaginous matrix of type II collagen and proteoglycans. Tissue engineered explants of ATDC5 cells were reminiscent of cartilaginous structures composed of numerous chondrocytes, staining for typical chondrocytic proteins, in lacunae embedded in a matrix of type II collagen and proteoglycans. In comparison, articular chondrocyte explants exhibited areas of Sox-9, aggrecan, and type II collagen-expressing cells growing on fleece, and discrete islands of chondrocytic cells embedded in a cartilaginous matrix.