Full thickness cartilage defects have poor intrinsic healing capacity and if left untreated may progress to osteoarthritis (OA) finally requiring total joint replacement. Different treatment options exist such as microfracturing,1 transplantation of osteochondral cylinders (OATS) and autologous chondrocyte implantation (ACI).2 In the classical 1st generation ACI procedure, chondrocytes are isolated from a minor weight bearing area of the joint, expanded in monolayer culture and re-injected under an autologous periosteal flap sutured onto the defect.3 However, there are major drawbacks to this technique. Besides donor site morbidity and chondrocyte de-differentiation, hypertrophic changes of the periosteum are described in up to 36% of ACI using periosteal flaps causing operative re-interventions.4, 5 Periosteum is composed of two different layers: a thick fibrous outer layer and, adjacent to the bone a thin cambium layer containing osteochondral progenitor cells which seem to contribute to tissue regeneration,6 and have the potential to undergo chondrogenesis in vitro and in vivo after appropriate induction by growth factors.7–9 Besides its biomechanical function retaining the transplanted cells within the defect, periosteum is known to secrete bioactive factors such as TGF-β, retinoid acid, pro-inflammatory cytokines as IL-6 and IL-8 which affect growth and the metabolic status of epiphyseal chondrocytes in a paracrine manner.10, 11
Little is known about matrix remodeling, maturation and turnover processes occurring at the chondrocyte/cartilage implant interface or within the repair tissue itself. Roberts et al.12 demonstrated the presence of matrix metalloproteinases (MMPs) activity in the graft tissue after ACI suggesting active remodeling at the implant site. MMP-induced cleavage of collagen II fibrils allows swelling of the repair tissue and permit molecules from the neighboring intact cartilage to be incorporated into the newly formed matrix. Growth factors, such as TGF-β and BMPs, which are sequestered by the fibrillar network may be redistributed and activated, and thereby induce production of new collagens and proteoglycans necessary to generate cartilage-like repair tissue.
In pathological conditions such as rheumatoid arthritis (RA) and OA, the expression of several MMPs is elevated in cartilage.13 Periosteal and cartilage tissues express and produce MMPs, including MMP-2 and -13.14, 15 Together with other collagenases MMP-13 plays a significant role in collagen turnover by cleaving collagen II.14 The gelatinases MMP-2 and -9 are key enzymes in both, inflammatory and degenerative joint diseases.16 MMP-2 is known to be increased in osteoarthritic cartilage17 while MMP-9 is rate limiting in endochondral ossification, where it regulates apoptosis of hypertrophic chondrocyte functioning as a key regulator of growth plate angiogenesis.18 Both MMPs can activate TGF-β from its latent form, thereby playing a role in physiological tissue remodeling as well as in tumor invasion and angiogenesis.19
Pro-inflammatory cytokines such as IL-6 are generally known to reduce the expression of cartilage-specific collagens,20 proteoglycans,21 and tissue inhibitors of MMPs, while concurrently causing an increase in the expression and secretion of MMPs, cyclooxygenases, and nitric oxide (NO).22, 23 Healthy and osteoarthritic chondrocytes continuously secrete IL-6 whose synthesis increases in response to inflammatory cytokines such as IL-1β, TNF-α, or INF-γ produced by activated synoviocytes, mononuclear cells, or articular cartilage itself. As a consequence, elevated levels of IL-6 were detected in synovial fluids from degenerative and inflammatory arthropathies.24
We hypothesize that periosteal cells and chondrocytes mutually affect each others metabolic activities. The aim of the current study was to analyze to which extent periosteal cells affect proteolytic extracellular matrix (ECM) remodeling activity of chondrocytes and vice versa. For this purpose, we have employed two coculture models including human periosteal explants and expanded human articular chondrocytes kept in 3D micromass pellets to determine their effects on expression and production of MMPs and their potent inductor IL-6. These coculture models allowed direct physical contact between periosteum and micromass pellets or paracrine interactions via soluble factors only by separating periosteum and micromass pellets using a 1 µm membrane.25
MATERIALS AND METHODS
Cartilage was obtained from patients giving informed consent following the standards of the local Ethics Commission of the University of Regensburg which approved to this study. Full thickness cartilage slices were aseptically dissected from femoral condyles of OA patients aged 50–76 years who underwent total knee arthroplasty. Chondrocytes were prepared from cartilage slices as described earlier,26 plated in T175 flasks at a density of 105 cells/cm2 (passage 0) and kept in DMEM/F12, 10% FCS, 1× vitamins, 1× antibiotic/antimycotic solution (Invivogen, Toulouse, France) until reaching confluence where cells were split 1:3. After reaching confluence (passage 1), micromass pellets were generated from 2 × 105 chondrocytes in 1.5 ml Eppendorf tubes and transferred into 12-well culture plates where they were kept in chondrogenic medium27 (DMEM high-glucose (4.5 g/L) + 1% pen/strep supplemented with 0.1 µM dexamethasone, 50 µg/ml ascorbate-2-phosphate, 40 µg/ml proline, 110 µg/ml pyruvate, and ITS + 1 Premix (100 µg/ml insulin, 55 µg/ml transferrin, 50 ng/ml selenium, 0.5 mg/ml bovine serum albumin, 4.7 µg/ml linoleic acid, Sigma-Aldrich, St. Louis, MO)). The medium was changed three times per week for all culture set ups.
Only cartilage from areas classified as “intact” was dissected and used as a cell source. Cartilage was classified as “intact” using the criteria described previously.28
Periosteal explants harvested from the proximal medial tibia were stored immediately in chondrogenic medium at 37°C/5% CO2 for a maximum of 2 days prior to coculture with micromass pellets. Explants were then cut into squares of about 5 mm × 5 mm and transferred either into contact or paracrine cell culture. Periosteum and chondrocytes used in these experimental set ups were always from different patients due to the time required for chondrocytes to reach confluency in passage 1 (about 4–5 weeks). No pooling of periosteum or chondrocytes from different donors was performed, neither was cartilage harvested from osteophytes or chondrogenic cysts.
Paracrine coculture was performed in 12-transwell plates containing cell culture inserts (1 µm porous membrane) as described earlier.25 Periosteal explants were placed at the bottom of the wells, and the cell culture inserts placed above were loaded with 10 micromass pellets per insert. These culture set ups were maintained in chondrogenic medium for up to 28 days and samples were taken at days 7, 14, 21, and 28. As a control, micromass pellets and periosteal explants were kept in monocultures under the same conditions.
Physical Contact Coculture
For physical contact coculture, micromass pellets were directly transferred onto periosteal explants placed into 12-transwell cell culture inserts and allowed to adhere for 10 min at 37°C, 5% CO2 in a medium droplet. After adding chondrogenic medium cell culture was performed as for paracrine cell culture. The cambium layer of the periosteal explants faced the micromass pellets.
RNA Extraction, Generation of Plasmid Standards, and Quantitative PCR
Micromass pellets were digested for max. 4 h in a 3:1 mixture of chondrogenic medium and CTH (collagenase II, trypsin, hyaluronidase).29 Afterwards RNA was extracted using a Qiagen RNeasy Mini Kit (Qiagen, Hilden, Germany). Periosteum was blotted onto sterile paper towels, weighted, cut into 1 mm pieces and homogenized in 1 ml Trizol (Invitrogen, Darmstadt, Germany)/50 mg tissue using a polytron (Kinematica, Switzerland) and RNA was isolated as described earlier.25 For cDNA synthesis 5 µg RNA were subjected to DNase digest and cDNA synthesis was performed as described previously.27 Generation of plasmid standards for quantitative real-time PCR was performed as described earlier.25 Quantitative RT-PCR was performed using qPCR SuperMix-UDG (Invitrogen). After addition of 50 ng cDNA in 5 µl H2O, quantitative PCR was conducted in duplicates (Applied Biosystems, Foster City, CA, ABI 7000). For absolute quantification, a plasmid standard curve (10 to 106 copies of the target gene) was included in each PCR plate. PCR efficiency E [%] was calculated as follows: E = 100 × (10−1/(Slope) − 1). Primer sequences for MMP-2, -7, -9, 13 and IL-6 are described elsewhere.30
Fifty microliter aliquots of culture supernatants were subjected to gelatin zymography as described elsewhere.26 Casein zymography was performed similar with the following modifications: gels contained 2 mg/ml β-casein instead of gelatin and the gels were pre-run for 4 h allowing excess casein to migrate out of the gels.31
Following enzymes were used for identification of MMPs according to molecular weight: Recombinant pro-MMP-7 and active MMP-7 were obtained from ProSpec-Tany TechnoGene (Rehovot, Israel) recombinant active MMP-2 and pro-MMP-13 and purified native human pro-MMP-9 were obtained from Calbiochem, Nottingham, UK.
Cell culture supernatants used for zymography and ELISA were pooled from medium changes between days 2–7 (day 7), days 8–14 (day 14), days 15–21 (day 21), days 22–28 (day 28). Culture supernatants from biopsies of different patients were kept separately at any time.
IL-6 protein concentration was analyzed in supernatants of micromass pellets and periosteum on days 7, 14, 21, and 28 of culture by sandwich-ELISA using human IL-6 duo kits according to manufacturer's instructions (R&D Systems, Minneapolis, MN). IL-6 concentration was adjusted to cell number of chondrocytes in micromass pellets (2 × 106 cells) and weight of periosteal explants (50 mg). Color development was initiated by adding 100 µl of POD substrate (0.5 mg/ml o-phenylenediamine) in 1× stable peroxide buffer (Pierce, Rockford, IL) to each well. After incubation for 30 min at RT in the dark, the reaction was stopped by adding 50 µl 2.5 M H2SO4 and optical density of the product was measured at 492 nm with a Genios microplate reader (Tecan, Männedorf, Switzerland).
Sample Preparation for Histological Staining
Micromass pellets from monocultures, from paracrine cell culture, and pellets cultured on periosteal explants (physical contact cultures) were harvested after 7, 14, 21, and 28 days, washed with PBS, and fixed in 4% paraformaldehyde in PBS (pH 7.4) over night at 4°C. The samples were dehydrated through a graded series of ethanol solutions and embedded in paraffin using standard procedures. Five micrometer sections were cut and stained with hematoxylin–eosine (HE).
Real-time PCR expression data and IL-6 ELISA results were adjusted by setting the highest value within one experiment to 100% to compensate for inter-patient variation.
Data (mean ± SEM) were analyzed by two-way analysis of variance (ANOVA). Bonferroni post hoc tests were conducted to determine the differences in each group. A p value less than 0.05 was considered significant.
IL-6 Gene Expression and Protein Secretion
Representative for the presence of pro-inflammatory cytokines we have analyzed expression and secretion of IL-6 in our culture set ups. In periosteum (Fig. 1A), the highest IL-6 gene expression was observed at day 7, with a decrease in all set ups afterwards. A significant higher gene expression was observed in paracrine and physical contact cocultures versus monocultures at day 7.
In chondrocytes cultured in micromass pellets, IL-6 gene expression remained stable in physical contact cocultures while decreased in paracrine and monoculture set ups during time line of culture (Fig. 1B). Compared to monocultures and paracrine cocultures gene expression was induced in physical contact cocultures at days 14 and 28.
IL-6 protein secretion (Fig. 1C) in all culture conditions declined after day 7 and remained at that level until the end of the culture. A significantly higher IL-6 concentration was detected in paracrine coculture supernatants compared to all other culture set ups at days 7 and 14 while on day 7 both coculture conditions induced a higher IL-6 secretion compared to monocultures. Secretion in paracrine cocultures remained higher, however not significant, compared to other culture regimen during the whole culture timeline.
Gene Expression of MMPs in Periosteum
Representative for collagen degrading MMPs we have analyzed gene expression of MMP-2, MMP-7, MMP-9, and MMP-13 in periosteal cells. MMP-2 gene expression in periosteum (Fig. 2A) increased significantly in monoculture conditions from day 7 on while gene expression level in physical contact cultures was suppressed. Paracrine cocultures revealed elevated gene expression level compared to physical contact coculture, however, significant only on day 28.
MMP-7 gene expression (Fig. 2B) was increased during culture time in paracrine coculture and physical contact coculture conditions while monocultures remained statistically unchanged. Paracrine cocultures displayed elevated gene expression level compared to periosteal monocultures at days 21 and 28 while physical contact cocultures displayed higher MMP-7 gene expression which, however, remained not significant.
MMP-9 gene expression (Fig. 2C) showed a significant decrease over culture time under all culture conditions. Gene expression level was elevated in both coculture set ups at day 7 compared to monocultures and remained statistically significant elevated in physical contact cultures until day 14.
MMP-13 gene expression (Fig. 2D) increased from day 7 until day 21 in paracrine coculture compared to monocultures and physical contact cultures. At days 14 and 21 gene expression was profoundly higher in paracrine cultures compared to the other two culture conditions but reached significance only at day 14.
Gene Expression of MMPs in Micromass Pellets
Representative for collagen degrading MMPs we have analyzed gene expression of the above mentioned MMPs in chondrocytes kept in micromass pellets. In micromass pellets, MMP-2 gene expression remained mostly unaltered in all culture conditions. No statistical significant differences were detectable between the three culture regimens (Fig. 3A).
MMP-7 gene expression (Fig. 3B) was suppressed in physical contact cocultures compared to monoculture regimen at day 21 while being upregulated at day 28.
MMP-9 gene expression level (Fig. 3C) declined over culture time in all culture set ups. No significant differences were detectable between the various culture conditions except for day 7 where gene expression was upregulated in physical contact cocultures compared to paracrine cocultures.
MMP-13 gene expression (Fig. 3D) remained mostly unchanged during the experimental time course with no statistical differences between the different culture set ups except for day 7 with physical contact cocultures induced.
MMP Secretion and Activity Profile
Secretion of MMPs into cell culture supernatant was assayed by gelatin (Fig. 4) and casein zymography (Fig. 5). Gelatin zymography revealed pro- and active MMP-2 in all four culture regimen at any time-point investigated; however, monocultured micromass pellets convert only small, almost undetectable amounts of the inactive enzyme to its active form (Fig. 4C). Pro-MMP-2 secretion and its conversion to the active enzyme appeared to increase in all set ups over the culture time. Pro-MMP-9 was detected in supernatant of periosteum monocultures (Fig. 4D) at all time-points, however, in much lower intensity than MMP-2. Monocultured micromass pellets did not secrete MMP-9 (Fig. 4C).
Casein degrading MMPs such as MMP-7 and MMP-13 were detected with casein zymography in culture supernatants (Fig. 5). Pro-MMP-13 was detected in all culture set ups, however, with low band intensity compared to the other casein degrading MMPs (Fig. 5A,B). Pro-MMP-7 together with activated MMP-7 was detected at increasing concentrations during culture time line in all set ups while activated MMP-7 (according to the molecular weight of recombinant MMP-7) could not be detected in periosteum monocultures (Fig. 5A,B). Additional bands at higher molecular weight (>50 kDa) presumably correspond to MMP-1 and MMP-3 according to molecular weight and tissue origin.32
Histology of Micromass Pellets in Cocultures Compared to Monoculture
Expanded articular chondrocytes from patients retained their capacity to develop a cartilage-like matrix if they were cultured in subsequent micromass pellet cultures. While some cells underwent apoptosis in the initial pellet phase other chondrocytes produced a matrix containing glycosaminoglycans and collagens fibrils. HE staining of the micromass pellets demonstrated similar cell morphology at all time-points. The majority of cells occupied a lacuna within the matrix and displayed a chondrocyte-like morphology with round shape. From day 7 to day 28 the lacunae increased in size indicating a maturation process of the cells and an increased matrix deposition. This matrix deposition occurred in all dimensions leading to an increased distance between the cells. Therefore pellets from day 21 onwards seemed to have fewer cells than pellets of earlier time-points. Notably, the interface between periosteum and micromass pellets was mainly acellular and connected both tissues tightly (Fig. 6).
The aim of this study was to shed light on interactions between human articular chondrocytes and periosteum in an in vitro model system which resembles the ACI situation in its conventional periosteal based set up originated by Brittberg et al.33 Our coculture regimens aimed to detect effects on matrix remodeling and turnover presumably mediated by soluble and matrix-associated factors originating from both the periosteal tissue and the chondrocytes.
For long-term repair and regeneration of focal cartilage defects chondrocytes have been implanted at the site of injury and a periosteal flap has been used as cover, however, not much attention has been paid to microenvironmental effects generated by this arrangement which resembles a bioactive chamber. Despite of cartilage metabolic autonomy, an extensive paracrine communication between different regions of cartilage or between cartilage and tissues of the immediate environment exist mutually affecting cartilage metabolism. Embryonic epiphyseal longitudinal bone growth is critically dependent on specific signals from the cartilage proper itself and the tibial periosteum. The same is observed in adults during fracture repair.34–36 The prospect of implanting dedifferentiated chondrocytes into cartilage defects without time consuming and cost effective ex vivo re-differentiation steps appears mostly intriguing for future cartilage repair methods. Here, periosteum could serve as a natural scaffold and as a source of bioactive factors for these cells. Apart from paracrine factors direct physical contact and cell–ECM interactions are believed to mutually affect the differentiation and proliferation status of the cocultured cells.
Both, micromass pellets and periosteal explants express and produce IL-6. During culture time line, IL-6 secretion in paracrine coculture is higher than in physical contact coculture or in the corresponding monocultures. The elevated concentration of IL-6 in paracrine coculture supernatants does not simply represent an additive effect of combining two monocultures but a real induction in periosteum because under physical contact coculture conditions no such increase is detectable. One might speculate that direct physical contact exerts an inhibitory effect on IL-6 secretion which is abolished once tissue and cells are physically separated and which is absent in monocultures. This inhibitory effect might compensate for the inducing ability of soluble factors once direct cell/matrix contact is missing. This observation is intriguing when considering that TGF-β secretion is induced by physical contact in this coculture system, which might be regulated through similar mechanisms.25
Our data demonstrate that MMP gene expression profiles are partly dependent on culture time and, partly depend on culture conditions. Induction of gene expression of both MMP-2 and MMP-13 in periosteal cells in paracrine coculture set up occurs concomitant to secretion of IL-6. Mainly periosteal explants in paracrine cocultures are affected whereas the gene expression profile in chondrocytes remains often unaltered irrespective of culture set up. We suggest that pro-inflammatory cytokines among other, yet unidentified soluble factors are responsible for these changes in mRNA expression and either affect periosteum directly or indirectly. It is well documented that paracrine factors influence the phenotype of cells and tissues, that is, chondrocytes of normal articular cartilage do not undergo terminal differentiation whereas growth plate chondrocytes do. By coculturing articular and growth plate chondrocytes in a set up which restricted interaction to paracrine communication, Jikko et al.37 have demonstrated that soluble paracrine factors alone are responsible for alteration of the chondrogenic phenotype. We demonstrated earlier that COL1A1 gene expression was induced in periosteal explants by paracrine coculture. Protein deposition of collagen I was induced in micromass pellets in paracrine cocultures while remained repressed in physical contact cocultures.25
MMP-9 gene expression declined in periosteum and chondrocyte micromass pellets over culture time under all culture conditions. This data points to a different regulation mechanism compared to the other MMPs investigated. Nothing is known about MMP-9 gene and protein expression and its role in matrix turnover in adult periosteal tissue. Here, we demonstrate for the first time that periosteal cells express and secrete pro-MMP-9 and also can activate it in culture supernatants. mRNA level of pro-MMP-9 in micromass pellets presumably remained too low as to result in protein synthesis and secretion while periosteum contains an elevated mRNA level resulting in biosynthesis and subsequent secretion of MMP-9. This is in agreement with previous findings where pro-MMP-9 is activated but not released by articular chondrocytes.38, 26
Of note, a series of post-transcriptional regulatory processes have also been described as relevant modulators of MMP expression including mRNA stability, protein translational efficiency and miRNA which might be causative for the partly opposing effects seen in mRNA and protein levels of the mediators analyzed. MMP transcripts harbor putative binding sites within the 5′ and 3′-untranslated regions (UTR) for diverse UTR-binding proteins affecting mRNA stability.39 Additionally, MMP expression is also controlled by regulation of translational efficiency. It has been shown previously that nucleolin recruits inactive MMP-9 mRNA into the rER and enhances translation efficiency.40
Another MMP which could be detected in our system is MMP-7 which is expressed during chondrogenesis of mesenchymal stem cells and is overexpressed in OA-cartilage.41, 42 MMP-7 gene expression increased over culture time in periosteal cells and in micromass pellets in all experimental set ups and is produced by both, periosteal explants and by micromass pellets. This is intriguing since this is the first study which reports about MMP-7 gene expression and its biosynthesis in periosteum. MMP-7 is unique in that it has high specific activity for cleavage in the collagen IIA specific cystein rich region.43 Collagen IIA is highly expressed during early stages of endochondral ossification in embryonic tissues and is not produced by differentiated chondrocytes which instead express the adult form collagen IIB.44 Notably, collagen IIA is re-expressed in OA cartilage and deposited around the chondrocytes.45 Among other MMPs, MMP-7 might possibly be needed for the specific degradation of the IIA variant of collagen II.
Keeping in mind that OA is often associated with inflammatory events such as synovitis or metabolic syndrome46 it is possible that pro-MMP-9 and the high levels of MMP-7 secreted by periosteum were at least in part due to the diseased state of the tissue.
In RA and OA, MMP-7 and MMP-9 are known to be upregulated in joint tissues, however, other than periosteum. In osteoarthritic cartilage, Ohta et al.42 detected increased immunostaining and gene expression for MMP-7 compared to healthy cartilage. MMP-9 mRNA and protein was detected in human osteoarthritic cartilage, but not in normal adult articular cartilage.47
MMP-7 activation seems to be dependent on chondrocyte factors as the pro-enzyme is produced in the periosteum monocultures but is not activated there in. In all other culture regimens active MMP-7 was detectable with difference between physical contact coculture and paracrine coculture conditions. These data suggest that soluble factors from chondrocytes are involved in pro-MMP-7 activation. One possible candidate protease is MMP-3,48 which presumably was detected in the casein zymographs according to molecular weight. MMP-7, on the other hand, is the activator of pro-MMP-1 and pro-MMP-9, which both are essential to completely degrade aggrecan and collagen II. Both, chondrocytes and periosteal cells might complement each other with respect to their proteolytic activities but are in addition involved in mutual activation of these enzymes. Soluble and matrix-associated factors derived from both tissues can exert mutual influence on gene expression of MMPs and may modulate IL-6 protein production. Our data indicate that both, periosteal cells and chondrocytes in some instances alter their proteolytic activities in a coculture set up. Over all, direct physical contact can do both increase and repress gene expression according to class of MMP. Paracrine coculture affects MMP gene expression generally by inducing MMP mRNA level presumably mediated by IL-6 among other cytokines. It remains speculative why both culture conditions exert opposite effects in some instances. Possibly, the distance between the two tissues plays a role, because it determines the time soluble factors need to reach their target cells and, it defines the nature of concentration gradients of effector molecules. This mechanism possibly plays also a role in defining collagen I gene and protein expression in periosteum and micromass cocultures, however, does not seem to affect aggrecan expression and deposition.25 In physical contact situations cell–matrix interactions might be additionally involved since the micromass pellet and the periosteum form a tight tissue interface. Moreover, these effects are cell type-dependent because coculture effects are more pronounced in periosteal cells than in chondrocytes.
We conclude that metabolic interactions between periosteum and chondrocytes may be not only beneficial for the regenerating tissue in vivo and may lead to some of the clinically observed complications of ACI. While an early excessive IL-6 and MMP secretion might slow down or hinder the development of a functional neo-cartilage, a certain amount of MMP activity is likely required at a later time to achieve sufficient integration of the regenerate tissue within the surrounding healthy cartilage. We speculate that a temporal sequence of implant/periosteal interactions will be of importance for the integration of repair tissue. Interactions between periosteum and chondrocytes in early phases might promote excessive proteolytic activity which is needed for remodeling and subsequent maturation of the repair tissue, but delays integration of the graft. At later stages, communication between the implanted chondrocytes and the periosteal graft attenuates proteolytic activities and thus promotes integration of repair tissue into the focal cartilage defect.
We thank Anja Pasoldt for her superior technical assistance. This work was supported by the ReForM C joint proposal “Analysis and modulation of interactions at articular interfaces” of the University Hospital Regensburg. The grant A2 of this joint proposal was assigned to S.G. and S.A.