Cell-based resurfacing of large cartilage defectsLong-term evaluation of grafts from autologous transgene-activated periosteal cells in a porcine model of osteoarthritis.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 58, No. 2, February 2008, pp 475–488 DOI 10.1002/art.23124 © 2008, American College of Rheumatology Cell-Based Resurfacing of Large Cartilage Defects Long-Term Evaluation of Grafts From Autologous Transgene-Activated Periosteal Cells in a Porcine Model of Osteoarthritis Kolja Gelse,1 Christiane Mühle,2 Oliver Franke,3 Jung Park,3 Marc Jehle,1 Karsten Durst,3 Mathias Göken,3 Friedrich Hennig,4 Klaus von der Mark,3 and Holm Schneider2 healthy hyaline cartilage. Unstimulated periosteal cells and cells activated by liposomal gene transfer formed only fibrocartilaginous repair tissue with minor contact stiffness. However, within 6 months following transplantation, the AAV/Ad-stimulated cells in the superficial zone tended to dedifferentiate, as indicated by a switch from type II to type I collagen synthesis and reduced contact stiffness. In deeper zones, these cells retained their chondrocytic phenotype, coinciding with positive staining for type II collagen in the matrix. Conclusion. Large partial-thickness cartilage defects can be resurfaced efficiently with hyaline-like cartilage formed by transgene-activated periosteal cells. The long-term stability of the cartilage seems to depend on physicobiochemical factors that are active only in deeper zones of the cartilaginous tissue. Objective. To investigate the potential of transgene-activated periosteal cells for permanently resurfacing large partial-thickness cartilage defects. Methods. In miniature pigs, autologous periosteal cells stimulated ex vivo by bone morphogenetic protein 2 gene transfer, using liposomes or a combination of adeno-associated virus (AAV) and adenovirus (Ad) vectors, were applied on a bioresorbable scaffold to chondral lesions comprising the entire medial half of the patella. The resulting repair tissue was assessed, 6 and 26 weeks after transplantation, by histochemical and immunohistochemical methods. The biomechanical properties of the repair tissue were characterized by nanoindentation measurements. Implants of unstimulated cells and untreated lesions served as controls. Results. All grafts showed satisfactory integration into the preexisting cartilage. Six weeks after transplantation, AAV/Ad-stimulated periosteal cells had adopted a chondrocyte-like phenotype in all layers; the newly formed matrix was rich in proteoglycans and type II collagen, and its contact stiffness was close to that of Articular cartilage lesions often pose great clinical challenges, and current joint-preserving therapies entail many problems, including unstable repair tissue, poor anchoring of engineered cartilage grafts, or significant morbidity at the donor site. Bone marrow–stimulating techniques have been applied for many years, but have resulted, at best, in the formation of biomechanically inferior fibrocartilage (1–4). Although encouraging results have been reported for transplantation of autologous chondrocytes (5,6), its clinical use is still limited to circumscribed defects in which a surrounding intact cartilage shoulder protects the graft from mechanical shear stress and dislocation (7). None of the current cell-based methods has been proven suitable for treatment of the large lesions of osteoarthritis (OA). This study, therefore, investigates a concept for the repair of large cartilage defects as seen in OA. The repair of such defects requires either high numbers or Supported by the German Research Foundation (grant GO741/13-1) and the Interdisciplinary Center for Clinical Research Erlangen (grant C1). 1 Kolja Gelse, MD, Marc Jehle, MD: University Hospital Erlangen, and University of Erlangen-Nuremberg, Erlangen, Germany; 2Christiane Mühle, DiplBiochem, Holm Schneider, MD, PhD: University of Erlangen-Nuremberg, Erlangen, Germany, and Medical University of Innsbruck, Innsbruck, Austria; 3Oliver Franke, DiplEngr, Jung Park, PhD, Karsten Durst, PhD, Mathias Göken, PhD, Klaus von der Mark, PhD: University of Erlangen-Nuremberg, Erlangen, Germany; 4Friedrich Hennig, MD: University Hospital Erlangen, Erlangen, Germany. Address correspondence and reprint requests to Holm Schneider, MD, PhD, Medical University of Innsbruck, Department of Pediatrics, Anichstrasse 35, 6020 Innsbruck, Austria. E-mail: holm. email@example.com. Submitted for publication December 12, 2006; accepted in revised form October 19, 2007. 475 476 high proliferative capacity of chondrogenic cells. Because of their limited availability, articular chondrocytes appear unsuitable for this purpose. Furthermore, chondrocytes have a restricted proliferative capacity, particularly those isolated from the joints of elderly patients with OA. Studies have shown that chondrocytes dedifferentiate when cultured ex vivo and may undergo senescence during amplification (8–11). Therefore, mesenchymal precursor cells may represent an interesting alternative (12,13). Several types of tissue, including periosteum, bone marrow, synovial tissue, muscle, or fat, have been shown to contain progenitor cells capable of differentiating into chondrocyte-like cells (12,14–17). These potential donor tissues are often available abundantly, and cells can be isolated by minimally invasive procedures with minor donor-site morbidity. In addition, such precursor cells proliferate well and retain their chondrogenic potential, even in advanced age (18). However, recent studies have demonstrated that a chondrogenic stimulus is required, since, in the adult organism, spontaneous differentiation of mesenchymal precursors into chondrocytes is rarely observed (14,19, 20). In addition to biomechanical influences, a number of growth or differentiation factors, including transforming growth factor ␤, bone morphogenetic proteins (BMPs) 2, 4, and 7, and cartilage-derived morphogenetic proteins, have been shown to promote chondrogenesis by certain precursor cells (19,21–23). However, application of the respective recombinant proteins is very expensive, and their sustained supply in a bioactive form still represents a challenging problem. Therefore, the transfer of complementary DNA (cDNA) encoding chondrogenetic factors has been suggested as an alternative method by which bioactive proteins can be provided for a prolonged period of time directly at the site of cell implantation (23,24). In this study, we used either liposomal gene transfer or a combination of adenovirus (Ad)– and adeno-associated virus (AAV)–mediated gene transfer to deliver BMP-2 cDNA ex vivo to periosteal cells prior to seeding the cells in a polyglycolic acid (PGA) matrix to be implanted into the joint. Our previous studies have demonstrated the efficacy of such cell-mediated gene transfer techniques as well as the absence of detrimental immune responses in a small-animal model (16,19). In the present study, the feasibility of resurfacing large partial-thickness cartilage defects, a typical feature of OA, was investigated in miniature pigs. In addition to morphologic evaluation, we used nanoindentation measurements to assess both the structural and biomechanical properties of the resulting repair tissue. GELSE ET AL MATERIALS AND METHODS Adenoviral and AAV vectors. The adenoviral vector AdBMP-2, carrying human BMP-2 cDNA under the control of the cytomegalovirus (CMV) promotor, was described previously (14,16,19). Concentrations of the vector stocks, as determined by plaque assay on 911 cells, ranged from 1 ⫻ 1010 to 3 ⫻ 1010 plaque-forming units/ml. The AAV serotype 2–derived vector AAVBMP-2 was generated by inserting human BMP-2 cDNA under the control of a CMV promotor into an AAV backbone, which was produced in HEK293 cells by cotransfection of 3 plasmids, as described by Xiao et al (25). Viral particles were purified according to established protocols (26). The concentrations of the vector stocks ranged from 1.5 ⫻ 109 to 4 ⫻ 109 infectious particles/ml. Isolation and characterization of autologous periosteal cells. Twelve female adult miniature pigs (Ellegaard, Dalmose, Denmark), each age 18 months and with body weights of 35–40 kg, were used. The animals were kept in an air-conditioned animal facility in which a temperature of 20–25°C, relative humidity of 40–55%, and light/dark cycle of 12 hours were maintained. All animals were fed a standard diet ad libitum. The pigs were anesthetized by intramuscular injection of 30 mg midazolam (Dormicum; Roche, Mannheim, Germany) and 300 mg ketamine (Ketavet; Pfizer, New York, NY) followed by ventilation with isoflurane (Baxter Diagnostic, McGaw Park, IL) at 2 liters/minute. Periosteal cells were obtained from the cambium layer of the periosteum of the right proximal tibia. The harvested flap was minced with a scalpel and treated with 0.2% trypsin (Gibco Life Technologies, Grand Island, NY) for 20 minutes, followed by exposure to 0.02% clostridial collagenase (Roche) dissolved in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal calf serum (FCS) (Gibco Life Technologies) for 10 hours at 37°C. Suspended periosteal cells were purified from debris with a sterile nylon filter and resuspended in DMEM, yielding a total of ⬃1.5 ⫻ 106 cells from each animal. The washed cells were seeded on plastic dishes at a density of 106 cells per 10-cm plate, cultivated at 37°C, 5% CO2 in DMEM supplemented with 10% FCS, insulin–transferrin–selenium (Sigma-Aldrich, Bornem, Belgium), ascorbate-2-phosphate (50 g/ml), and penicillin–streptomycin (100 units/ml), and passaged once, prior to gene delivery, after reaching initial confluence. The periosteal cell population was characterized by reverse transcription–polymerase chain reaction (RT-PCR) for a number of stem cell markers. Detection of CD13, CD34, CD45, CD90, Sca-1, and c-Kit was performed as described previously (14), using the following primer pairs: 5⬘-CTCCTCAGCGTTCGACTACC-3⬘ and 5⬘-CATCCTCCAGTTGTCCTCGT-3⬘ for CD13; 5⬘-GGCCAACGGAACAGAACTTA-3⬘ and 5⬘-CGACGGTTCATCAGCAAGTA-3⬘ for CD34; 5⬘-TGACCACTTCAGCAAGCATT-3⬘ and 5⬘-TGGAGCATCTTTGCACAGTC-3⬘ for CD45; 5⬘-AGAAGGTGACCAGCCTAACGG-3⬘ and 5⬘-TCTGAGCACTGTGACGTTCTG-3⬘ for CD90; 5⬘-TATGGTTTTGTGATGTTTGTCC-3⬘ and 5⬘TAGATCCAGGGGCATTGTAG-3⬘ for Sca-1; and 5⬘-GATGCCTTCAAGGATTTGGA-3⬘ and 5⬘-TCTGAGCACTGTGACGTTCTG-3⬘ for c-Kit. BMP-2 gene delivery. Periosteal cells were transferred into three 15-cm dishes and stimulated, at subconfluence, by CELL-BASED RESURFACING OF LARGE CARTILAGE DEFECTS nonviral or viral BMP-2 gene delivery. Liposomal transfection was performed using 75 g of the plasmid pVAXCMVBMP-2 mixed with FuGene (Roche) at a concentration of 2 l/g DNA. The liposome–DNA complexes in serum-free DMEM were added dropwise to the cells. After incubation at 37°C for 4 hours, the transfection medium was replaced by complete medium. The viral vectors AAVBMP-2 and AdBMP-2 were applied at doses of 1,000 infectious particles/cell and 100 plaque-forming units/cell, respectively, in a total of 10 ml of serum-free DMEM. Two hours after infection, 10 ml of complete culture medium containing 20% FCS was added to the dish. Following incubation for another 24 hours, the cells were washed with phosphate buffered saline (PBS), mobilized using 0.1% trypsin/0.025% EDTA, resuspended in DMEM containing 10% FCS, and washed twice with DMEM to remove noninternalized viral particles. Infection assay in vitro and quantification of BMP-2 production. Periosteal cells at a density of 1 ⫻ 105 cells per well in 24-well plates were infected with AdBMP-2 (100 plaque-forming units/cell) or AAVBMP-2 (1,000 infectious particles/cell) or simultaneously with both vectors. Liposomal transfection was performed as described above. BMP-2 concentrations in the cell supernatants were determined 4, 7, 14, 21, and 28 days after infection. The medium was replaced every second day and exactly 24 hours before collecting the supernatant. BMP-2 was quantified with a commercial enzyme-linked immunosorbent assay, according to the protocol provided by the manufacturer (R&D Systems, Wiesbaden, Germany). Scaffold preparation. Following mobilization with trypsin/EDTA, 1.5 ⫻ 107 BMP-2–stimulated or untreated periosteal cells were resuspended in 30 l of fibrinogen (Beriplast; Aventis-Behring, Marburg, Germany) and seeded in a PGA matrix with an area of 2 cm2 (Soft PGA Felt; Alpha Research, Berlin, Germany). Gel formation was achieved by adding 30 l of thrombin solution to both sides of the cell-loaded scaffold, which was then incubated for another 48 hours in culture medium to allow cell attachment. Surgical procedures. Three weeks after cell isolation, the animals were anesthetized as described above. The skin around the left knee was washed, shaved, disinfected, and draped. The joint capsule was opened by a medial parapatellar incision, and the patella was displaced laterally. A large partial-thickness defect comprising the entire medial half of the articular surface of the patella was created using a custommade device in which the blade exceeded the basis by 0.6 mm. Since the cartilage thickness at this site was determined to be 0.7–0.9 mm in transverse sections, this device did not injure the subchondral bone plate, as confirmed by a complete absence of bleeding. The artificial cartilage defect was then treated by implantation of a cell-loaded PGA matrix. The animals in group 1 (n ⫽ 2) and group 4 (n ⫽ 3) received PGA scaffolds containing 1.5 ⫻ 107 unstimulated autologous periosteal cells. In group 2 (n ⫽ 2), a scaffold with 1.5 ⫻ 107 autologous periosteal cells stimulated with liposomal BMP-2 gene transfer was applied. In group 3 (n ⫽ 2) and group 5 (n ⫽ 3), animals received PGA scaffolds containing 1.5 ⫻ 107 AdBMP-2/ AAVBMP-2–infected autologous periosteal cells. In 5 animals (from groups 1 and 4), an additional partial-thickness defect, 477 which was created in the proximal fourth of the lateral half of the patella, was left untreated to serve as the control; according to clinical experience, such a small, relatively smooth chondral lesion at a peripheral site of the joint does not lead to relevant release of matrix fragments and, therefore, may not initiate a general catabolic response with progression of the lesion or alterations of the surrounding tissue. The scaffold was trimmed to the defect size and fixed with 2 PGA pins (Resor-Pin; Geistlich Biomaterials, Wolhusen, Switzerland) to the subchondral bone. To avoid damage to the opposing joint surface, the head of the pin was lowered beneath the surface level by milling the superficial part of the hole. To provide a smooth surface, a bilayer collagen membrane (Chondrogide; Geistlich Biomaterials) was fixed onto the PGA matrix by a noninterrupted suture with 5-0 Polydioxanon suture. The patella was reduced, and the stability of the graft was tested by repeated flexion of the knee. The joint capsule and skin were then closed by resorbable vicryl sutures. After this treatment, the animals were allowed to move freely in their cages. The right knee joint was left untreated. The animals were killed either 6 weeks after cell transplantation (groups 1–3) or 26 weeks thereafter (groups 4 and 5). All procedures were approved by the appropriate institutional and governmental review boards. Biomechanical characterization of the repair tissue. The knee joints were dissected carefully, and the articular surface of the patella was first assessed macroscopically. A representative osteochondral biopsy sample (0.4 ⫻ 0.4 ⫻ 0.4 cm) was obtained using a miniature rotating saw. A corresponding sample of healthy cartilage was isolated from the patella of the untreated right knee. These samples were kept in PBS and investigated within 24 hours. Since cartilage is an inhomogeneous, anisotropic, and nonlinear elastic tissue, evaluation of all biomechanical parameters would have gone beyond the scope of this study and would also overburden the diagnostic means in most clinical situations. From a clinical point of view, the contact stiffness of the articular surface is considered a suitable measure for comparing the quality of different repair tissues. To evaluate the contact stiffness in a microscopic range, as required in our miniature pig model, nanoindentation measurements were performed as described previously (27). Briefly, the subchondral bone of the osteochondral blocks was glued to a plastic vessel, with the cartilage surface perpendicular to the indentation direction. All measurements were carried out in PBS using a NanoIndenter XP system (MTS, Oak Ridge, IL) with a 3-sided pyramidal Berkovich tip. The repair tissue was analyzed on the basis of load-displacement curves recorded during loading and unloading, using at least 5 different indentation sites per sample and a complex loading cycle with 5 loading–unloading segments per indentation site. The maximum load applied in each loading segment was increased stepwise from 0.6125 mN to 10 mN, with a loading time of 15 seconds. To limit the effect of creep, a holding segment was implemented after each loading segment. The contact stiffness (S) was evaluated from the slope of the unloading curve at each unloading segment, following the approach of Oliver and Pharr (28). Because this approach is based on linear elasticity, its validity for the determination of the contact area in viscoelastic, hydrated soft tissues may be limited in the case of slow unloading. Therefore, a high 478 GELSE ET AL unloading rate of 1.3 N/second was used to minimize the influence of the viscoelastic response and to ensure that mainly elastic tissue recovery governs the measured contact stiffness after unloading. Since elastic tissue recovery occurs instantaneously, only the upper part of the unloading curve was taken into account (29). The correlation between the reduced indentation modulus (Er) and the contact stiffness (S) can be derived from Sneddon’s contact model, as follows: S⫽ dF 2 ⫽ E ␤ 冑A dh 冑 r where dF is the load applied, dh is the displacement, ␤ is the correction factor accounting for the pyramidal indenter shape, and A is the projected contact area. In an inhomogeneous tissue such as articular cartilage, there is no linear increase in contact stiffness with increasing indentation depth, because the indentation modulus (Er) depends on the cartilage zone investigated. However, the reduced indentation modulus (Er) obtained by nanoindentation can be correlated with the sample modulus (Es) for a given Poisson ratio (s), taking the elastic properties of the indenter (Ei) into consideration, as follows: 1 ⫺ i2 1 ⫺ 2s 1 ⫺ 2s 1 ⫽ ⫹ ⬇ . Er Ei Es Es Because the materials studied herein were very soft, the elastic deformation of the indenter (Ei) could be considered negligible, thus simplifying the above-described equation. Detection of vector DNA and BMP-2 transgene expression in the repair tissue. DNA or RNA was extracted from tissue specimens (1 mg) from a representative central region of AAVBMP-2/AdBMP-2–stimulated or unstimulated repair tissue. DNA was isolated with the DNeasy kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. The number of vector copies per specimen was determined by quantitative PCR as described previously (30). AAVBMP-2 was detected with the primer pair 5⬘-GCCATTGTTCAGACGTTGGT-3⬘ and 5⬘-GCCACCACCTTCTGATAGGC-3⬘, designed to amplify a 276-bp fragment of the AAV DNA containing parts of both the BMP-2 insert and the 3⬘untranslated region sequence of the AAV. The primers 5⬘-AGGACAGGCCTACCCTGCTA-3⬘ and 5⬘-ACACGGACCACGTCAAAGAC-3⬘, amplifying a 270-bp fragment of the adenoviral hexon gene, were used to detect AdBMP-2. Realtime PCR was performed in parallel to the amplification of 10-fold serial dilutions of the vector DNA (109 to 101 copies) spiked into the background of genomic pig DNA. All samples were analyzed in quadruplicate. RNA was isolated with a commercial kit and treated with DNase I for 30 minutes to remove any contaminating genomic or vector DNA. Reverse transcription was performed using an anchored oligo(dT)18 primer. Transgenic human BMP-2 messenger RNA was detected using the primers 5⬘-GCCATTGTTCAGACGTTGGT-3⬘ and 5⬘-GCTGTACTAGCGACACCCACA-3⬘ in a first round of PCR, and the primers 5⬘-GCTGTGTCCCGACAGAACTC-3⬘ and 5⬘-CAACCCTCCACAACCATGTC-3⬘ for amplification of a 100-bp fragment in the subsequent nested reaction. For each sample, 0.5 l of the cDNA was subjected to the initial 30 PCR cycles, and 0.5 l of the reaction product was then amplified in a further 30 PCR cycles using the second pair of primers. A 207-bp fragment of pig 18S ribosomal RNA, amplified with the primers 5⬘-CCTGGATACCGCAGCTAGGA-3⬘ and 5⬘CGAACCTCCGACTTTCGTTC-3⬘, served as internal control. All PCR products were analyzed by gel electrophoresis on a 2% agarose gel. Histologic and immunohistochemical assessments. After isolation of osteochondral blocks for mechanical testing, the patella was fixed in 4% paraformaldehyde for 18 hours, followed by decalcification in 0.5M EDTA for 3 months. The proximal, intermediate, and distal third of the patella were assessed separately, yielding 6–9 individual assessments per group. After standard processing, the samples were embedded in paraffin. Serial transverse 5-m sections were cut and stained with toluidine blue to estimate the proteoglycan content or with hematoxylin and eosin for further histologic investigation. For immunohistochemical detection of type I and type II collagen, deparaffinized sections were pretreated with 0.2% hyaluronidase (Roche) in PBS (pH 5.0) for 60 minutes at 37°C and with Pronase (2 mg/ml in PBS, pH 7.3; Sigma-Aldrich) for 60 minutes at 37°C. The sections were then washed with Tris buffered saline (TBS; 5 mM Tris in 0.9% NaCl, pH 7.35) and left to react at 4°C with a monoclonal mouse anti-human type I collagen antibody (MP Biomedicals, Irvine, CA) diluted in PBS at a ratio of 1:200 or with a mouse anti-human type II collagen antibody (1:500; MP Biomedicals), followed by careful washing with TBS and subsequent incubation with a biotinylated donkey anti-mouse secondary antibody (Dianova, Hamburg, Germany) for 30 minutes. After careful washing, a complex of streptavidin and biotinylated alkaline phosphatase was added. The sections were developed with fast red and counterstained with hematoxylin. Histologic sections from the 5 experimental groups were compared using a scoring system based on that of O’Driscoll et al (31) and the histologic scale suggested by the International Cartilage Repair Society (32). Statistical analysis. Data on the reduced modulus and contact stiffness were analyzed between groups using Student’s t-test. P values less than 0.01 were considered significant. RESULTS Characterization of the porcine periosteal cells. The periosteal cell populations obtained from the tibiae of miniature pigs were heterogeneous, consisting primarily of fibroblast-like cells. Expression of stem cell– related antigens, including Sca-1, CD13, and CD90, as detected by RT-PCR, indicated the presence of mesenchymal progenitor cells, whereas no expression of c-Kit, CD34, or CD45 was found. Chondrogenic differentiation of these cells was stimulated in vitro by nonviral or viral BMP-2 gene delivery. However, only simultaneous infection of the cells with AAVBMP-2 and AdBMP-2 resulted in BMP-2 production at biologically relevant levels of ⱖ10 ng/ml CELL-BASED RESURFACING OF LARGE CARTILAGE DEFECTS Figure 1. a, Experimental animal model used for periosteal cell–based grafts, and b, evaluation of vector persistence in the repair tissue at 6 weeks and 26 weeks after cell transplantation. a, In miniature pigs, partial-thickness chondral defects comprising the entire medial half of the articular surface of the patella were created using a custom-made planer. A cell-loaded polyglycolic acid (PGA) matrix was fixed with resorbable pins to the underlying bone. The upper part of the pin holes was milled out to allow lowering of the pin heads in plane with the surface. To create a smooth surface, an additional collagen membrane was sutured onto the PGA matrix. b, Quantitative polymerase chain reaction analysis revealed a time-dependent decline in the copy numbers of both adeno-associated virus (AAV) and adenovirus (Ad) vectors. However, the AAV vector was detectable much longer than the adenoviral vector. Bars show the mean and SD. BMP-2 ⫽ bone morphogenetic protein 2. (33) for more than 3 weeks (mean ⫾ SD daily secretion of BMP-2 13.8 ⫾ 1.7 ng/ml medium at 4 weeks after gene transfer). Infection with AdBMP-2 alone yielded a 479 high secretion of BMP-2 initially, but this declined within 3 weeks to ⬍3 ng/ml per 24 hours. Porcine periosteal cells infected with AAVBMP-2 alone showed sustained transgene expression at insufficient levels (mean ⫾ SD daily secretion of BMP-2 0.75 ⫾ 0.29 ng/ml at 4 weeks after gene transfer). Liposomal gene transfer allowed neither sufficient nor sustained production of transgenic BMP-2 (⬍0.2 ng/ml at 4 weeks after gene transfer). Duration of transgenic stimulation following cell transplantation in vivo. In a pilot experiment in 6 pigs, autologous periosteal cells on a bioresorbable scaffold were transplanted into chondral defects comprising the entire medial half of the patella (Figure 1a). In 4 animals, a chondrogenic stimulus was provided to the cells, prior to transplantation, using liposomal BMP-2 gene transfer or AAV/Ad-mediated BMP-2 gene transfer (groups 2 and 3, respectively). Implants of unstimulated periosteal cells served as controls (group 1). The resulting repair tissue was examined 6 weeks after transplantation. Persistence of the vector DNA in vivo was evaluated by quantitative PCR. At 6 weeks after transplantation, the adenoviral and AAV vectors were still detectable in the repair tissue formed by AdBMP-2/AAVBMP-2–stimulated cells. However, only a few copies of adenovirus were found, whereas AAV copies were present in much higher numbers (Figure 1b). In the repair tissue of additional pigs investigated 26 weeks after transplantation of AAVBMP-2/AdBMP-2–stimulated cells, only 5 AAV copies/mg tissue were detectable, but no AdBMP-2 DNA was found (Figure 1b). In accordance with the declining presence of the vectors, RT-PCR analysis of the repair tissue showed BMP-2 transgene expression only after 6 weeks of stimulation, but not after 26 weeks of stimulation, with AAVBMP-2 and AdBMP-2. Neither the vector DNA nor expression of the human transgene were observed in samples of unstimulated repair tissue (results not shown). Resurfacing large cartilage defects with a hyaline-like repair tissue. In all animals, the cell-based grafts integrated sufficiently into the preexisting cartilage and conferred resistance to joint loading, despite the lack of a protective surrounding cartilage shoulder. A whitish repair tissue covered the medial half of the articular surface and was attached firmly to the underlying cartilage. Macroscopically, no delamination of scaffold or repair tissue could be detected in any group; however, the cartilage surface was partially uneven and inhomogeneous. 480 GELSE ET AL Figure 2. Results of periosteal cell–based resurfacing of large cartilage lesions in miniature pigs at 6 weeks after transplantation. The cartilage defects were treated by implantation of a polyglycolic acid matrix loaded with autologous periosteal cells; prior to implantation, chondrogenic differentiation was stimulated by liposomal BMP-2 gene transfer or AAV/Ad-mediated BMP-2 gene transfer. Implants of unstimulated periosteal cells and untreated lesions served as controls. The resulting repair tissues were examined 6 weeks after cell transplantation by toluidine blue staining (a, d, g, j, and m). Representative parallel sections were additionally evaluated by immunohistochemistry for type I collagen (b, e, h, k, and n) and type II collagen (c, f, i, l, and o). The underlying bone showed strong type I collagen staining (b, e, h, k, and n), and thus served as the internal control. Type II collagen (c, f, i, l, and o) was always clearly detectable in intact hyaline articular cartilage and calcified cartilage. Untreated partial-thickness lesions (d–f) did not heal spontaneously. Unstimulated cells (g–i) formed fibrous repair tissue that was positive for type I collagen (h) but negative for type II collagen (i). Transplantation of periosteal cells stimulated by nonviral BMP-2 gene transfer also failed to generate hyaline cartilage (j–l). In contrast, repair tissue produced by AAVBMP-2/AdBMP-2–infected cells showed a hyaline-like phenotype and strong staining for proteoglycan (m) and type II collagen (o), whereas faint staining for type I collagen (n) was detectable only in the superficial layer of the repair tissue, which may be attributable to remnants of the type I/type III collagen membrane sewed over the implanted matrix. Bonding to the neighboring cartilage was tight (p). The former pin area was replaced by fibrous tissue (q), and at higher magnification (r), a clear border between the cartilaginous repair tissue and the ingrowing fibrous tissue can be recognized. (Original magnification ⫻ 100 in a–o; ⫻ 400 in p and q; ⫻ 800 in r.) See Figure 1 for definitions. CELL-BASED RESURFACING OF LARGE CARTILAGE DEFECTS 481 Figure 3. Long-term results of periosteal cell–based resurfacing of large cartilage lesions. The cartilage defects were treated with unstimulated autologous periosteal cells or with AAVBMP-2/AdBMP-2–infected autologous periosteal cells, and at 26 weeks after cell transplantation, the tissue was investigated by macroscopic assessment of morphologic features (a and e), toluidine blue staining for proteoglycan (b and f), and immunohistochemistry for type I collagen (c and g) and type II collagen (d and h). The medial half of the patella (right half in a and e) was covered with a firmly attached, whitish-opaque repair tissue after application of unstimulated cells (a) or had a thick, glossy-whitish layer of repair tissue after application of AAVBMP-2/AdBMP-2–infected cells (e). Histologically, the repair tissue formed by unstimulated cells was positive for type I collagen (c) but negative for the cartilage-specific type II collagen (d). Transplantation of BMP-2–stimulated cells led to repair tissue with a fibrocartilaginous superficial zone that showed moderate toluidine blue staining for proteoglycan (f) and strong staining for type I collagen (g) but not type II collagen (h), whereas the intermediate and deep zones lacked type I collagen (g) but stained strongly for proteoglycan (f) and type II collagen (h). See Figure 1 for definitions. In the animals of groups 1, 2, and 3, circumscribed signs of vascularization were observed at sites to which the PGA pins had been anchored in the subchondral bone. Apart from these areas of vascularization, the repair tissue showed a lack of blood vessels. Macroscopic examination revealed minor, if any, remnants of sutures or scaffold. Parallel sections were investigated by toluidine blue staining for proteoglycan and immunohistochemical analysis for type I collagen, a marker of fibrous tissue, and type II collagen, a major and specific component of hyaline cartilage. Transplantation of unstimulated periosteal cells (group 1) failed to generate hyaline cartilage (Figures 2g–i). The repair tissue showed only weak staining for proteoglycan and type II collagen, but strong staining for type I collagen. Formation of fibrocartilage could be observed in only the deep zone of the repair tissue, wherein some cells had a round phenotype and were embedded in a matrix with partial staining for proteoglycan and type II collagen. In group 2, stimula- tion of the cells with liposomal BMP-2 gene transfer did not yield significantly better results (Figures 2j–l). In this group, a fibrous superficial zone was again distinguishable from a fibrocartilaginous deep zone that partially contained type II collagen. In group 3, however, after stimulation of the cells with AAV/Ad-mediated BMP-2 gene transfer, the repair tissue showed strong toluidine blue staining, indicative of a proteoglycan-rich matrix (Figure 2m), together with intense staining for type II collagen in all layers (Figure 2o). Most of the cells had adopted a chondrocyte-like phenotype, although the typical columnar orientation of the cells was not apparent. In this experimental group, a slight hypercellularity and excessive tissue were observed in some regions, correlating with the partially uneven surface. Bonding of the repair tissue to the underlying cartilage was firm, without any clefts (Figure 2p). The former pin area was replaced by fibrous tissue that could be distinguished clearly from the repair cartilage (Figures 2q and r), indicating that there was no 482 GELSE ET AL Figure 4. Structural organization of the matrix in the superficial zone (a–c), intermediate zone (d–f), and deep zone (g–i) of repair tissue formed by unstimulated (a, d, and g) or viral BMP-2–stimulated (b, e, and h) periosteal cells or healthy articular cartilage (c, f, and i) after 26 weeks. At 26 weeks, remnants of the polyglycolic acid scaffold, which is degraded within 6–10 weeks, were no longer detectable. Results of phase-contrast microscopy of paraffin sections show an irregular collagen fiber orientation in the matrix produced by unstimulated cells (d and g), in comparison with the homogeneous structure of healthy hyaline cartilage (c, f, and i), but a beginning perpendicular alignment in the deep zone of BMP-2–stimulated repair tissue (h). Bar ⫽ 100 m. See Figure 1 for definitions. relevant contribution of the ingrowing cells to the hyaline repair tissue. Based on these results and on the biomechanical findings (as described below), the viral BMP-2 gene transfer approach combining an Admediated strong initial stimulation of the graft along with a more sustained AAV-mediated stimulation was further investigated in long-term followup studies. Assessment of the long-term stability of hyalinelike repair tissue. A followup experiment investigating the long-term stability of the repair tissue formed by unstimulated periosteal cells or AAV/Ad-stimulated periosteal cells was performed on 6 additional pigs (groups 4 and 5, respectively). At 26 weeks after trans- plantation, the repair tissue was found to be firmly attached to the underlying cartilage (Figure 3). In comparison with the results at 6 weeks after transplantation with unstimulated periosteal cells (group 1), the repair tissue produced by unstimulated periosteal cells had a more regular surface, but still appeared rather opaque at 26 weeks (Figure 3a). At this later time point, the sites of pin anchorage, remnants of pin material, sutures, or PGA matrix were no longer detectable. In group 5, at 26 weeks after transplantation, the repair tissue originating from BMP-2–stimulated cells covered the former defect completely in 2 animals, and covered ⬃80% of the defect in the third animal. The CELL-BASED RESURFACING OF LARGE CARTILAGE DEFECTS newly formed cartilage had a homogeneous whitish, glossy appearance, but its translucency differed slightly from that of healthy articular cartilage. In 1 animal, the repair cartilage showed a totally even surface and was tightly connected to the underlying tissue, but was more than twice as thick as the neighboring healthy cartilage (Figure 3e). The other 2 animals displayed a whitish repair tissue with some minor irregularities and a slight tendency toward excessive cartilage formation in some areas. The thickness of the repair tissue in group 5 was a mean ⫾ SD 1.75 ⫾ 0.38 mm, which differed significantly from that in the animals of group 4 (1.33 ⫾ 0.17 mm) or that of normal articular cartilage at this site (0.78 ⫾ 0.07 mm). No ossification of synovial tissue or joint membranes was observed in either group. In the samples from group 4, histologic analysis revealed weak toluidine blue staining for proteoglycan and weak staining for type II collagen (Figures 3b and d). Phase-contrast microscopy of the group 4 samples showed an irregular orientation of the fiber structures in the intermediate and deep zones of the repair tissue (Figures 4d and g), with a tendency toward horizontal alignment in the superficial zone (Figure 4a). In comparison with the results in group 1 after 6 weeks, no relevant improvement in the structural organization of the matrix was observed in group 4 after 26 weeks; in fact, initial signs of cartilage degeneration, such as fibrillation, were beginning to develop. Thus, transplantation of unstimulated periosteal cells failed to generate hyaline cartilage within 26 weeks. In contrast, transplantation with BMP-2– stimulated periosteal cell grafts, which had yielded hyaline-like repair cartilage after 6 weeks, resulted in a tissue with significantly better matrix composition and cell morphologic features in the long-term experiment. However, the specimens obtained after 26 weeks (group 5) showed a decreased intensity of metachromatic toluidine blue staining in the superficial zone, whereas in the deep zone, the hyaline-like appearance was maintained (Figure 3f). Type II collagen was no longer detectable in the superficial zone (Figure 3h), and the cells partially adopted a spindle-shaped phenotype (Figure 4b), whereas the cells of the deep zone retained their round shape and were still surrounded by a type II collagen– positive matrix (Figures 3h and 4e and h). These changes could account for the reduced scores for matrix staining and cell morphologic features in group 5 compared with group 3 (Figure 5). Bonding to the underlying calcified cartilage and to neighboring cartilage of the lateral half of the patella was tight, and mostly continuous, in all samples. No 483 Figure 5. Morphologic scores of the repair tissues in the different experimental groups, determined according to an established histologic grading scale. A score of 3 indicates an exact match with healthy hyaline cartilage. The repair tissues of the proximal, intermediate, and distal third of the patella were evaluated separately, yielding 6–9 individual assessments per experimental group. Bars show the mean and SD for unstimulated cells at 6 weeks (open bars), liposomal gene transfer–stimulated cells at 6 weeks (second solid bar in each category), AAVBMP-2/AdBMP-2–stimulated cells at 6 weeks (dark gray shaded bar), unstimulated cells at 26 weeks (light gray shaded bar), and AAVBMP-2/AdBMP-2–stimulated cells at 26 weeks (fifth solid bar in each category). The groups did not differ significantly with respect to surface architecture and integration into the adjacent preexisting cartilage. However, the repair tissue formed by AAVBMP2/AdBMP-2–stimulated periosteal cells showed improvements both in matrix staining and in cellular phenotype and also showed a tendency toward better structural organization, as compared with the tissue produced by unstimulated cells. See Figure 1 for definitions. inflammatory reactions in the synovial membrane were observed. Analysis of contact stiffness and modulus of the repair tissue by nanoindentation studies. A biomechanical evaluation of the newly formed cartilage was attempted by measuring its contact stiffness and indentation modulus. Although nanoindentation is primarily a surface-characterization technique, it represents a sensitive method for the detection of small differences in tissue stiffness, even in relatively soft cartilage specimens (27). The mechanical response of healthy articular cartilage from miniature pigs served as the referent for assessing the quality of the repair cartilage. Healthy cartilage was characterized by a relatively low contact stiffness (mean ⫾ SD 84.0 ⫾ 23.4 N/m) and a low modulus (mean ⫾ SD 0.94 ⫾ 0.45 MPa) after application of a minor load (0.6 mN), but had a much higher contact stiffness (2,836.2 ⫾ 390 N/m) and modulus (8.02 ⫾ 1.51 MPa) when exposed to a load of 10 mN, indicating a strong, nonlinear increase both in contact stiffness and in modulus with the load displacement. Figure 6a shows the contact stiffness of the 484 GELSE ET AL Figure 6. Contact stiffness and indentation modulus as functions of the load applied to healthy and repair cartilage. A scatter diagram of the contact stiffness (a) and both the contact stiffness and reduced modulus at maximum loads of 2.5 mN (b and c) and 10 mN (d and e) in healthy and repair cartilage are shown. At loads of 2.5 mN, the mean contact stiffness and reduced modulus of the repair tissue did not differ significantly from the values in healthy articular cartilage, except for the repair tissue produced by unstimulated cells in experimental group 4. At a maximum load of 10 mN, however, the repair tissue from all treatment groups proved to be significantly softer than healthy cartilage. At both 6 weeks and 26 weeks after transplantation, the repair tissue produced by unstimulated periosteal cells showed a significantly lower contact stiffness and indentation modulus than that formed by AAVBMP-2/AdBMP-2–stimulated cells. Bars show the mean and SD. ⴱ ⫽ P ⬍ 0.01. See Figure 1 for definitions. specimens from groups 1, 3, 4, and 5 as a function of the load applied, highlighting the difference between repair tissue and healthy cartilage. Interestingly, if exposed to loads of ⬍2.5 mN, the contact stiffness and the modulus of healthy cartilage were even lower than the values in the repair tissue from group 3 (Figures 6b and c). At maximum load (10 mN), healthy cartilage displayed the highest contact stiffness, followed, in descending order of stiffness, by the samples from groups 3, 1, 5, and 4 (Figure 6d). Correspondingly, CELL-BASED RESURFACING OF LARGE CARTILAGE DEFECTS the modulus was highest in healthy cartilage and lowest in group 4 cartilage (Figure 6e). Although the hydrostatic pressure state during application of 10 mN by nanoindentation differs from the physiologic situation, normal articular stress is in the same order of magnitude. Under normal joint loading, the stress approximates 1 MPa (⬃400N, ⬃4 cm2) in this model, compared with 0.3 MPa during indentation with 10 mN on healthy cartilage. At 26 weeks after transplantation, the repair tissue formed by unstimulated cells had a very low contact stiffness and modulus (Figures 6b–e), both of which appeared to be independent of the indentation depth. However, in an inhomogeneous tissue such as articular cartilage, the indentation modulus is not a constant value and is, rather, highly dependent on the indentation depth and other factors, such as hydration. Thus, the data obtained in this study are relative measures that serve to compare the different repair tissues under identical testing conditions. These values may not be compared generally with the results of other studies, because both contact stiffness and modulus are influenced by various methodologic and environmental specifications. DISCUSSION This study documents both the potential and the limitations of a cell-based cartilage repair approach for resurfacing of large partial-thickness cartilage defects as seen in OA. Although our model does not reflect the metabolic alterations of OA cartilage, it does allow the investigation of a novel cell transplantation approach for the repair of large lesions in which an intact cartilage shoulder is lacking. We herein evaluated the use of BMP-2–stimulated mesenchymal precursor cells as a means to assess the long-term stability of the chondrogenic phenotype induced by gene transfer. The rationale behind our experiments was the high incidence of OA lesions in young patients, for whom the perspective of joint replacement has to be considered very critically. In considering the possibility of resurfacing such large defects with the use of a repair tissue comparable with hyaline cartilage, attachment and tight bonding of the graft to the underlying and neighboring tissue are issues of central importance. Chondrocytes or precursor cells are certainly the most suitable for connecting the graft with preexisting tissue, because their receptormediated bonding is superior to artificial adhesives, such as fibrin glue, which are degraded gradually (34,35). A PGA matrix covered with a collagen membrane was 485 shown to be an appropriate scaffold for resurfacing large defects of the joint surface that are not shielded from mechanical stress by an intact cartilage shoulder. As observed in our recent studies on rats (14,19), viral BMP-2 gene transfer efficiently induced chondrogenic differentiation of periosteal cells within 6 weeks. AAV vectors are known to facilitate long-term transgene expression; yet, their infection efficacy is relatively low and the onset of transgene expression is typically delayed (36). Therefore, simultaneous adenoviral- and AAV-mediated BMP-2 gene transfer in conjunction with prompt short-term adenoviral transgene expression to compensate for the delayed onset of AAV-mediated effects appears to be a promising way to not only induce but also support the maintenance of chondrogenic differentiation. Furthermore, synergistic effects of adenoviral and AAV vectors have been documented in studies on other animal models (30,37). However, such dual vector administration constitutes only one possibility among several means to achieve a sustained BMP-2 supply. For example, the stimulus may also be provided by retroviral vectors or devices that allow slow protein delivery. Our previous studies showed that the stimulatory effects on chondrogenesis were due to the transgenic BMP-2 and could not be ascribed to the nonspecific effects of the vector particles (19). Adenoviral coinfection may increase the efficacy of AAV-mediated transgene expression by facilitating nuclear translocation of AAV DNA and supporting the conversion of the AAV genome into a transcriptionally active double-stranded form (35). In comparison with the combined viral gene transfer method, liposomal gene delivery was found to be far less efficient and, in this model, failed to induce complete chondrogenesis of the transplanted cells. Interestingly, the thickness of repair tissue formed by BMP-2–stimulated cells was greater than that of unstimulated tissue or normal articular cartilage. After initial maturation of the graft, no further increase in thickness between week 6 and week 26 occurred, which might be a consequence of the decline in BMP-2 transgene expression. This indicates that the final thickness of the repair cartilage may depend on dose and duration of the differentiation stimulus, which could be controlled by varying the dose and type of the vectors applied. Thus, dose-response studies will have to be performed and additional work will be required to clarify whether the final thickness of the repair tissue mainly depends on the initial differentiation signal or on other factors such as matrix or cell density. Stimulation by BMP-2 gene transfer prior to 486 transplantation clearly improved the quality of the repair tissue formed by periosteal cells. Contact stiffness and modulus, when used as parameters for biomechanical comparison of the repair tissue, showed trends similar to the trends in intensity of toluidine blue staining of the matrix and in phenotypic similarities between the transplanted cells and articular chondrocytes. The high sensitivity of the nanoindentation technique allowed basic biomechanical characterization of the upper zone of the repair tissue. However, cartilage is an inhomogeneous, anisotropic, and nonlinear elastic tissue that cannot be characterized entirely by 1 or 2 biomechanical parameters, and some important aspects, such as creep and permeability, have not been included in the evaluation. This may be a limitation of our study and is worth addressing in future studies. Furthermore, nanoindentation, as performed herein, does not allow exact determination of the properties of the deeper zones of cartilage. This issue was addressed recently by Li et al, who showed that results of nanoindentation mainly reflect properties of the superficial zone (38). Nevertheless, it has to be pointed out that the superficial layer is of highest importance for the function of repair tissue, because degeneration of the articular surface, e.g., fibrillation and fissuring, begins in the superficial zone. A proteoglycan-rich hyaline matrix on its own may not guarantee the unique mechanical properties of healthy cartilage, unless the ultrastructural orientation of the collagen fibers is adjusted to the physiologic requirements. Within 6 months, such structural remodeling of the matrix with physiologic alignment of the fibers occurs only partially and incompletely in deeper zones of the repair tissue. In this study, unstimulated cells formed, at best, fibrocartilage that was characterized by a predominance of type I collagen and a minor contact stiffness and modulus. A comparison of the 6-week samples with those obtained after 26 weeks revealed a dramatic reduction in both the contact stiffness and modulus with time, which may be explained by the assumed initial mechanical support of the transplanted collagen membrane and the PGA matrix that was followed, after 6 months, by complete degradation. No time-dependent differences with regard to cellular phenotype and matrix staining could be detected. Thus, periosteal cells failed to spontaneously undergo complete chondrogenesis in vivo, despite being exposed to mechanical stimuli under physiologic conditions. Histologic examination at 6 months after cell transplantation revealed the first signs of degeneration, indicating the limited durability of fibrous repair tissue in stressed joints. GELSE ET AL Interestingly, BMP-2–stimulated periosteal cells located in the superficial zone of the repair tissue tended to dedifferentiate from a transient chondrocyte-like phenotype to a fibroblastic phenotype in the course of 6 months. The homogeneous hyaline-like repair tissue that had been observed in all layers after 6 weeks was later transformed into a bilayered tissue characterized by a superficial fibrocartilaginous or fibrous zone and a persisting hyaline-like zone in the depth. This zonedependent deterioration was reflected by a significant decrease in contact stiffness and modulus of the upper zones of the tissue. Unfortunately, due to methodologic constraints of the nanoindentation technique, deeper areas could not be characterized biomechanically in more detail. Although transgene-activated precursor cells may temporarily adopt a chondrocyte-like phenotype with its typical gene expression pattern (14), such cells could still differ significantly from articular chondrocytes in their genetic program, since their differentiation state seems unstable, at least under conditions found in the superficial zone of a repair tissue in vivo. Similar results have been reported by Luyten and Dell’Accio (20), who found that chondrocytes, but neither periosteum- nor bone marrow–derived mesenchymal cells, produced a stable hyaline tissue following intramuscular injection in nude mice. However, despite the time-dependent decline in BMP-2 transgene expression, BMP-2–stimulated periosteal cells located in deeper zones of the repair tissue maintained their chondrocyte-like phenotype for 26 weeks, even with initial signs of physiologic remodeling of the tissue architecture, which, in this respect, indicates a significant role of physical and biochemical factors. An issue that has not been addressed in our study is the cellular changes and metabolic alterations of OA cartilage. Detrimental conditions in OA joints, which display activation of catabolic pathways, may interfere significantly with the viability, maturation, and integration of a cell-based graft. Therefore, future studies have to include an animal model that better reflects the pathophysiologic conditions of OA, and a concomitant transient antiinflammatory therapy may be necessary to protect the graft. The animal model used in the present study thus demonstrates the feasibility of a cell-based approach to the treatment of large partial-thickness cartilage defects. However, it also points out limitations in the use of mesenchymal precursor cells for that purpose. Further studies will have to clarify whether the observed dedifferentiation lies in the nature of mesenchymal progenitor cells, and which physicobiochemical factors are key CELL-BASED RESURFACING OF LARGE CARTILAGE DEFECTS players in the complex chondrogenic differentiation pathways. 487 12. ACKNOWLEDGMENTS We thank E. Koppmann for excellent technical assistance, Dr. Philippe Moullier (Genethon, Nantes, France) for providing the plasmid pRepCap, Dr. Richard Samulski (University of North Carolina, Chapel Hill) for the plasmid pXX680, and Geistlich Biomaterials (Wolhusen, Switzerland) for the Chondrogide membranes. 13. 14. 15. AUTHOR CONTRIBUTIONS Dr. Schneider had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study design. Gelse, Durst, Göken, Hennig, von der Mark, Schneider. Acquisition of data. Gelse, Mühle, Franke, Park, Jehle, Durst. Analysis and interpretation of data. Gelse, Mühle, Franke, Durst, Göken, Schneider. Manuscript preparation. Gelse, Mühle, Franke, Jehle, Durst, Göken, Schneider. Statistical analysis. Gelse, Mühle, Franke, Schneider. 16. 17. 18. REFERENCES 1. Hunziker EB. Articular cartilage repair: basic science and clinical progress. A review of the current status and prospects. Osteoarthritis Cartilage 2002;10:432–63. 2. Insall J. The Pridie debridement operation for osteoarthritis of the knee. Clin Orthop Relat Res 1974;101:61–7. 3. Steadman JR, Rodkey WG, Briggs KK, Rodrigo JJ. The microfracture technique in the management of complete cartilage defects in the knee joint. Orthopade 1999;28:26–32. In German. 4. Knutsen G, Engebretsen L, Ludvigsen TC, Drogset JO, Grontvedt T, Solheim E, et al. Autologous chondrocyte implantation compared with microfracture in the knee: a randomized trial. J Bone Joint Surg Am 2004;86-A:455–64. 5. Peterson L, Minas T, Brittberg M, Nilsson A, Sjogren-Jansson E, Lindahl A. Two- to 9-year outcome after autologous chondrocyte transplantation of the knee. Clin Orthop Relat Res 2000;374: 212–34. 6. Marlovits S, Zeller P, Singer P, Resinger C, Vecsei V. Cartilage repair: generations of autologous chondrocyte transplantation. Eur J Radiol 2006;57:24–31. 7. Behrens P, Bosch U, Bruns J, Erggelet C, Esenwein SA, Gaissmaier C, et al. Indications and implementation of recommendations of the working group “Tissue Regeneration and Tissue Substitutes” for autologous chondrocyte transplantation (ACT): recommendations for indication and application of ACT of the Joined Advisory Board of the German Societies for Traumatology and Orthopaedic Surgery. Z Orthop Ihre Grenzgeb 2004;142: 529–39. In German. 8. Von der Mark K, Gauss V, von der Mark H, Muller P. Relationship between cell shape and type of collagen synthesised as chondrocytes lose their cartilage phenotype in culture. Nature 1977;267:531–2. 9. Hayflick L. The limited in vitro lifetime of human diploid cell strains. Exp Cell Res 1965;37:614–36. 10. Parsch D, Brummendorf TH, Richter W, Fellenberg J. Replicative aging of human articular chondrocytes during ex vivo expansion. Arthritis Rheum 2002;46:2911–6. 11. Piera-Velazquez S, Jimenez SA, Stokes DG. Increased life span of 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. human osteoarthritic chondrocytes by exogenous expression of telomerase. Arthritis Rheum 2002;46:683–93. Winter A, Breit S, Parsch D, Benz K, Steck E, Hauner H, et al. Cartilage-like gene expression in differentiated human stem cell spheroids: a comparison of bone marrow–derived and adipose tissue–derived stromal cells. Arthritis Rheum 2003;48:418–29. De Bari C, Dell’Accio F, Vanlauwe J, Eyckmans J, Khan IM, Archer CW, et al. Mesenchymal multipotency of adult human periosteal cells demonstrated by single-cell lineage analysis. Arthritis Rheum 2006;54:1209–21. Park J, Gelse K, Frank S, von der Mark K, Aigner T, Schneider H. Transgene-activated mesenchymal cells for articular cartilage repair: a comparison of primary bone marrow-, perichondrium/ periosteum- and fat-derived cells. J Gene Med 2006;8:112–25. De Bari C, Dell’Accio F, Tylzanowski P, Luyten FP. Multipotent mesenchymal stem cells from adult human synovial membrane. Arthritis Rheum 2001;44:1928–42. Gelse K, Jiang QJ, Aigner T, Ritter T, Wagner K, Poschl E, et al. Fibroblast-mediated delivery of growth factor complementary DNA into mouse joints induces chondrogenesis but avoids the disadvantages of direct viral gene transfer. Arthritis Rheum 2001; 44:1943–53. Adachi N, Sato K, Usas A, Fu FH, Ochi M, Han CW, et al. Muscle derived, cell based ex vivo gene therapy for treatment of full thickness articular cartilage defects. J Rheumatol 2002;29: 1920–30. De Bari C, Dell’Accio F, Luyten FP. Human periosteum-derived cells maintain phenotypic stability and chondrogenic potential throughout expansion regardless of donor age. Arthritis Rheum 2001;44:85–95. Gelse K, von der Mark K, Aigner T, Park J, Schneider H. Articular cartilage repair by gene therapy using growth factor–producing mesenchymal cells. Arthritis Rheum 2003;48:430–41. Luyten FP, Dell’Accio F. Identification and characterization of human cell populations capable of forming stable hyaline cartilage in vivo. In: Hascall VC, editor. The many faces of osteoarthritis. Basel: Birkhaeuser; 2002. p. 67–76. Gruber R, Mayer C, Bobacz K, Krauth MT, Graninger W, Luyten FP, et al. Effects of cartilage-derived morphogenetic proteins and osteogenic protein-1 on osteochondrogenic differentiation of periosteum-derived cells. Endocrinology 2001;142:2087–94. Joyce ME, Roberts AB, Sporn MB, Bolander ME. Transforming growth factor-␤ and the initiation of chondrogenesis and osteogenesis in the rat femur. J Cell Biol 1990;110:2195–207. Gelse K, Schneider H. Ex vivo gene therapy approaches to cartilage repair. Adv Drug Deliv Rev 2006;58:259–84. Evans CH, Ghivizzani SC, Robbins PD. Gene therapy for arthritis: what next? [review]. Arthritis Rheum 2006;54:1714–29. Xiao X, Li J, Samulski RJ. Production of high-titer recombinant adeno-associated virus vectors in the absence of helper adenovirus. J Virol 1998;72:2224–32. Snyder RO, Xiao X, Samulski RJ. Production of recombinant adeno-associated viral vectors. In: Dracpoli N, Haines J, Krof B, editors. Current protocols in human genetics. New York: Wiley; 1996. p. 1–24. Franke O, Durst K, Maier V, Goken M, Birkholz T, Schneider H, et al. Mechanical properties of hyaline and repair cartilage studied by nanoindentation. Acta Biomater 2007. E-pub ahead of print. Oliver WC, Pharr GM. An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J Mater Res 1992;7:1564–83. Ebenstein DM, Kuo A, Rodrigo J, Reddi A, Ries M, Pruitt L. A nanoindentation technique for functional evaluation of cartilage repair tissue. J Mater Res 2004;19:273–81. Muhle C, Neuner A, Park J, Pacho F, Jiang Q, Waddington SN, et al. Evaluation of prenatal intra-amniotic LAMB3 gene delivery by 488 adenoviral and AAV vectors in a mouse model of Herlitz disease. Gene Ther 2006;13:1665–76. 31. O’Driscoll SW, Keeley FW, Salter RB. Durability of regenerated articular cartilage produced by free autogenous periosteal grafts in major full-thickness defects in joint surfaces under the influence of continuous passive motion: a follow-up report at one year. J Bone Joint Surg Am 1988;70:595–606. 32. Mainil-Varlet P, Aigner T, Brittberg M, Bullough P, Hollander A, Hunziker E, et al, for the International Cartilage Repair Society. Histological assessment of cartilage repair: a report by the Histology Endpoint Committee of the International Cartilage Repair Society. J Bone Joint Surg Am 2003;85-A Suppl 2:45–57. 33. Knippenberg M, Helder M, Zandieh Doulabi B, Wuisman P, Klein-Nulend J. Osteogenesis versus chondrogenesis by BMP-2 and BMP-7 in adipose stem cells. Biochem Biophys Res Commun 2006;342:902–8. GELSE ET AL 34. Peretti GM, Zaporojan V, Spangenberg KM, Randolph MA, Fellers J, Bonassar LJ. Cell-based bonding of articular cartilage: an extended study. J Biomed Mater Res A 2003;64:517–24. 35. Wang H, Kandel RA. Chondrocytes attach to hyaline or calcified cartilage and bone. Osteoarthritis Cartilage 2004;12:56–64. 36. Buning H, Braun-Falco M, Hallek M. Progress in the use of adeno-associated viral vectors for gene therapy. Cells Tissues Organs 2004;177:139–50. 37. Chen Y, Luk KD, Cheung KM, Lu WW, An XM, Ng SS, et al. Combination of adeno-associated virus and adenovirus vectors expressing bone morphogenetic protein-2 produces enhanced osteogenic activity in immunocompetent rats. Biochem Biophys Res Commun 2004;317:675–81. 38. Li C, Pruitt LA, King KB. Nanoindentation differentiates tissuescale functional properties of native articular cartilage. J Biomed Mater Res A 2006;78:729–38.