A challenge in tissue engineering is the in vitro generation of human cartilage. To meet standards for in vitro-engineered cartilage, such as prevention of immune response and structural as well as functional integration to surrounding tissue, we established a three-dimensional cell culture system without adding exogenous growth factors or scaffolds. Human chondrocytes were cultured as spheroids. Tissue morphology and protein expression was analyzed using histological and immunohistochemical investigations on spheroid cryosections. A cartilage-like tissue similar to naturally occurring cartilage was generated when spheroids were cultured in medium supplemented only with human serum. This in vitro tissue was characterized by the synthesis of the hyaline-specific proteins collagen type II and S-100, as well as the synthesis of hyaline-specific mucopolysaccharides that increased with prolonged culture time. After 3 months, cell number in the interior of in vitro tissues was diminished and was only twice as much as in native cartilage. Additionally, spheroids quickly adhered to and migrated on glass slides and on human condyle cartilage. The addition of antibiotics to autologous spheroid cultures inhibited the synthesis of matrix proteins. Remarkably, replacing human serum by fetal calf serum resulted in the destruction of the inner part of the spheroids and only a viable rim of cells remained on the surface. These results show that the spheroid culture allows for the first time the autogenous in vitro engineering of human cartilage-like tissue where medium supplements were restricted to human serum.
SELF-REPAIR OF human hyaline cartilage does not occur. Therefore, cartilage injuries initiate a progressive degradation that eventually results in osteoarthritis.(1–3) An accepted approach for the regeneration of hyaline cartilage after traumatic cartilage damage is the autologous chondrocyte transplantation.(4–6) However, the in vitro engineering of three-dimensional hyaline cartilage tissue with the respective structure and function is still a challenge for cartilage repair in contrast to using cell suspensions as transplants. For three-dimensional in vitro engineering, cell-seeded scaffolds have been tested. The in vitro culturing of chondrocytes in the presence of growth factors on various three-dimensional scaffolds resulted in the maintenance of the cartilage-specific phenotype.(7–9) In animal models, these cell-seeded scaffolds allowed a formation of repair tissue similar to hyaline cartilage.(8,10,11) Unfortunately, the repair is often accompanied by considerable fibrocartilage formation.(12,13) Further demands on in vitro-engineered tissues are the integration into native tissue and the tissue formation-adapted resorption of scaffolds. However, neither the integration of these cell-seeded scaffolds into surrounding host cartilage, nor predictable resorption of the scaffold polymers, has been optimized for application in patients.(14)
Few approaches have been advanced to overcome these problems by renouncing any scaffolds. The generation of cartilage-like structures was observed when human chondrocytes isolated from osteoarthritic hips were cultured in a gyratory shaker.(15) However, to maintain the differentiated phenotype, the addition of growth factors was necessary.
The aim of this study was to engineer three-dimensional hyaline human cartilage that has the capacity to integrate with native cartilage and prevent immune responses. For that purpose, we established an autologous spheroid system to culture human chondrocytes without adding xenogenous serum, growth factors, or scaffolds, considering that several growth factors and scaffolds are not permitted for use in humans. Only human serum was added to the cell culture medium. In this autologous spheroid system, cells form three-dimensional aggregates and generate their own extracellular matrix that is similar to the natural matrix of hyaline cartilage. These in vitro-engineered cartilage tissues adhere to and integrate into native tissue. We further show that the addition of fetal calf serum (FCS) or antibiotics to culture medium delays or even inhibits the engineering of cartilage-like tissue in vitro. Our methodology describes, for the first time, the in vitro engineering of three-dimensional autogenous cartilage that seems suitable for the treatment of cartilage lesions, degenerative changes in cartilage, and pharmaceutical test systems.
MATERIALS AND METHODS
Articular cartilage was obtained from human articular condyles in volunteer patients undergoing knee surgery. Cartilage (60-100 mg) was minced and digested in a 50-ml Falcon tube using 20-25 U/mg collagenase type II (Biochrome, Berlin, Germany) at 37°C for 8 h in a gyratory shaker (110 rpm). Isolated cells were washed and resuspended in culture medium with the addition of FCS (n = 2), autologous serum (n = 3), or pooled human serum (n = 6) from separate human volunteers. No growth factors, cytokines, or other supplements were added. Chondrocytes were seeded in Falcon culture flasks (75 cm2, 15 ml medium) and maintained at 37°C in a humidified atmosphere and 5% CO2. Medium was changed twice weekly. After reaching confluence, the cells were trypsinized using trypsin-EDTA (PAA-Laboratories GmbH, Cölbe, Germany) and cultured in larger Falcon flasks (225 cm2, 35 ml medium). Experiments were performed with chondrocytes between the second and seventh monolayer passage. For generation of spheroids, chondrocytes were seeded in hydrogel-coated 96-well plates. For hydrogel coating, agarose was melted in cell culture medium (2% wt/vol) and pipetted into the wells. After a jelling time of 2 h at room temperature, cell suspensions (1 × 105 and/or 2 × 105 cells/well in 250 μl medium) were added. Starting with 1 × 105 and 2 × 105 cells/well should hint at a possible change in size and quality of cell aggregates. Cell aggregates of the appropriate experimental set-ups were analyzed after 5 days, 2 weeks, and 1, 2, and 3 months (n = 2). After aggregation of chondrocytes, 2-10 single spheroids were transferred into one well, allowing coalesce of spheroids. For co-culture, single spheroids were placed on cartilage tissue of isolated human femoral condyles (medial and lateral, 3.5 × 2 × 0.8 cm) from volunteer patients (n = 3 patients) undergoing knee replacement surgery because of osteoarthritis. Condyles with adherend spheroids were surrounded and covered with medium (50 ml) and cultured in petri dishes under standard conditions. After different time points (45 minutes, 3 weeks), condyles were frozen and histologically analyzed. To assess cell growth and cell differentiation in the presence of antibiotics, chondrocytes and spheroids were cultured in the addition of penicillin (100 U/ml) and streptomycin (100 mg/ml).
Cell proliferation was assessed by BrdU incorporation into monolayer (incorporation time: 6 h, 24 h) and aggregated cells (incorporation time: 6 h, 24 h, 2 days) using a cell proliferation kit (Amersham, Freiburg, Germany). Three samples per time point were analyzed.
To analyze chondrocytes in the monolayer culture, cell suspensions were added to glass slides in appropriate dishes. The cells adhere and proliferate directly on the glass surface. Slides were washed twice in phosphate buffered saline (PBS) and fixed in methanol/acetone (1:2) at −20°C for 10 minutes. Spheroid specimens were embedded in Tissue-Tek (Miles, Naperville, IL, USA), snap-frozen in liquid nitrogen, and cut into 5- to 7-μm sections using a cryomicrotome (Microm, Walldorf, Germany). Sections were mounted on pretreated slides (Superfrost Plus; Menzel Gläser, Braunschweig, Germany), air dried, and fixed in concentrated acetic acid:ethanol (1:20) at room temperature (RT) for 20 minutes. Hematoxylin/eosin (HE), safranin O, and Goldner-trichrome staining were performed on serial sections of spheroids and monolayer cells directly grown on slides using standard histochemical techniques.
Collagen and S-100 antigen(16) expression was assessed on serial sections of snap-frozen spheroids and monolayer slides using the avidin biotin complex (ABC) method (DAKO, Hamburg, Germany). As primary antibodies, polyclonal rabbit antisera recognizing human types I and II collagen (1:30 and 1:15; Novo Castra, Newcastle upon Tyne, UK) and S-100 protein (1:100, DAKO) were used. To avoid nonspecific binding of the antibodies, slides were first incubated with tris(hydroxymethyl)aminomethane (TRIS) buffer containing 5% normal porcine serum and 0.1% bovine serum albumin (BSA) at RT for 30 minutes. Incubation with primary antisera followed at 4°C in a humidified chamber for 12 h. After three washes in TRIS, bound primary antibody was detected using the DAKO ESAB + System AP, with fuchsin as the substrate for alkaline phosphatase (DAKO). Cell nuclei were counterstained with hematoxylin.
Control procedures paralleled each step. TRIS was applied to the sections instead of the primary antibodies as a control for the secondary antibody. Articular cartilage-bone sections were used as respective controls for the specificity of the primary antibodies against collagen type I and II. For the S-100 antiserum, a neuroblastoma cell line (SK-N-SH; American Type Culture Collection, Rockville, MD, USA) and fibroblasts were used as positive and negative controls, respectively.
For in vitro engineering of three-dimensional hyaline cartilage tissue, we investigated the differentiation of human chondrocytes in aggregate culture. After seeding chondrocytes in hydrogel-coated wells, cells aggregated to form a disc measuring 900-1200 μm in diameter after 1 day. During the next 2 weeks, discs rounded and became more compact, with diameters ranging from 350 to 500 μm (Figs. 1A and 1B). Coalescence of several spheroids was initiated by an active migration of surface cells of adjacent aggregates (Fig. 1B). Through 8 days, remodeling of the aggregates resulted in a more homogenous spherical structure, and gaps between aggregates were filled (Fig. 1C). Separate from the ability to merge, aggregates also migrated and contacted tissue culture flasks or glass slides (Fig. 1D).
Spheroid morphology and differentiation
During the initial 2 weeks, spheroid chondrocytes produced high amounts of acidic mucopolysaccharides, indicated by the green intercellular matrix on Goldner-stained sections (Fig. 2B). Aggregates cultured in medium with autologous and pooled serum revealed a homogenous distribution of intact spherical cells in the interior and flattened cells on the surface (Figs. 2A, 2B, and 3A–3C). Viable cells and round nuclei in the interior of spheroids suggest that sufficient nutrient supply is available for all cells. As individual aggregates merged, an adhesion zone consisting of several flattened cell layers between two aggregates was still observable after 24 h in close contact (Fig. 2A, arrowhead). However, by 5 days, these cells had achieved a histotypical morphology similar to that of internal cells (Fig. 2A, arrow).
Chondrocytes in monolayer culture loose the expression of collagen type II and the cartilage specific intracellular protein S-100(16) completely during their third or fourth passage, corresponding with a high expression of collagen type I (data not shown). However, when chondrocytes of the second to the seventh monolayer passage were cultured as spheroids, a reexpression of collagen type II and S-100 occurred (Figs. 3A and 3B). This chondrocyte differentiation was accompanied by the loss of cell proliferation, analyzed by BrdU incorporation into cell nuclei (data not shown). In the presence of autologous serum, a weak expression of collagen type II in the outer cell layer of spheroids was detectable after 1 week. After 6 weeks, a weak expression of collagen type II was also observed in the interior (Fig. 3A). In medium supplemented with allogenous pooled human serum, reexpression of collagen type II in the outer cell layer was delayed and not as high as in autologous culture. Collagen type II was detectable only after the third week in culture (data not shown). The S-100 reexpression was generally similar to that of collagen type II under autologous and human pooled serum conditions (Fig. 3B).
In contrast to the high expression of collagen type I in all cells in monolayer culture, a reduced expression or complete loss of collagen type I was found in the interior cells of the aggregate (Fig. 3C). Throughout the time course of three-dimensional culture, and independent of serum type, the outer cell layers expressed the highest amount of collagen type I. The expression of collagen type I was higher than that of collagen type II.
Under prolonged culture conditions, the in vitro generated tissues more closely resembled the typical features of in vivo cartilage: solid and elastic aggregates with a bright white surface (Fig. 4A). Cryosections revealed flattened cells on the surface (Figs. 4B and 4E) and an increased amount of intercellular matrix in the core of spheroids (Figs. 4B–4D). This core area stained positive with safranin O, affirming the synthesis and extracellular deposition of hyaline cartilage specific proteoglycans (Fig. 4B). Cell number per given matrix area varied with the depth in the aggregate. In the core area, cells were widely separated by matrix, and the cell number per given matrix area was approximately twice that of native cartilage (Figs. 4B and 4F). Nearer the surface, the number of cells per matrix area increased and was twice that of the core. In contrast to the randomly distributed cells in the core area of aggregates, cells on both sides of a merging zone were oriented vertical to the contact area (Fig. 4B). This orientation is known for native articular cartilage where chondrocytes are oriented perpendicular to the tidemark.
Influence of antibiotics and FCS on cell condition
Usually, antibiotics are added to cell culture medium. In the presence of penicillin and streptomycin in chondrocyte monolayer cultures with autologous serum, we observed a prolongation of culture time to reach confluence compared with the absence of antibiotics. Furthermore, the aggregation of chondrocytes was delayed for up to 5 days (data not shown). During the first 4 weeks in aggregation culture, the expression of cartilage-specific proteins and the cell density were similar to aggregates cultured in the absence of antibiotics. However, there is no augmentation in the expression of collagen type II, S-100, and acidic mucopolysaccharides in aggregated chondrocytes up to 3 months (data not shown). This parallels the still high and unchanged cell density in spheroids cultured in the presence of antibiotics observed after 3 months (data not shown).
While growing human chondrocytes in monolayer and aggregate culture in addition of FCS, several differences were apparent. In the monolayer, the proliferation rate of chondrocytes decreased (data not shown). Furthermore, when cells were transferred in aggregate culture, the aggregation of cells was delayed significantly. By 7 days, a central hole in the aggregate was present, and surface cells were more densely packed (Fig. 5A). During the first 6 weeks in aggregate culture with FCS, no quantitative differences in the expression of collagen type I and polysaccharides compared with autologous cultured aggregates were found (data not shown). However, the expression of collagen type II and S-100 protein was delayed and reduced with only a weak expression after 4 weeks (Fig. 5B, shown for collagen type II). After 3 months in FCS culture, the inner part of the aggregates was disintegrated and crumbled. HE staining revealed the absence of viable cells. Only outer cells retained the typical morphology and expressed type I collagen and S-100 (Figs. 5C and 5D). However, cells and matrix were loosely organized, and a shedding of the outer cells was detected (Figs. 5C and 5D).
To assess the integrative capacity of in vitro-engineered tissues, a cartilage-explant spheroid co-culture system was used. After only 45 minutes, multiple focal adhesion points were formed connecting the in vitro-generated tissue with native cartilage (Fig. 6A). Surface cells of the 3-week-old aggregates in the contact area had already changed their morphology from flattened to spherical cells. However, the cartilage on the opposing side of the aggregate retained its flattened surface cells (Fig. 6A). Over the course of 3 weeks in culture, the spheroids became more flattened. By active migration, aggregate cells were widely distributed on the surface of the degenerated cartilage. The cells not only migrated on the native cartilage surface but also synthesized new matrix (Fig. 6B). The migratory capacity enabled chondrocytes to integrate in surface fissures in addition to covering the surface (Fig. 6B). No active invading growth of spheroid derived cells was observed. The newly formed matrix on the cartilage explant surface is characterized by stronger HE staining (Fig. 6B). This layer shows a higher cell density than native tissue, indicating the cartilage forming capacity of spheroids.
This study shows the potential to use a three-dimensional, autologous in vitro culture system to engineer human articular cartilage. The formation of three-dimensional cartilage-like tissue was achieved without using any scaffolds. The only supplement to culture medium was patient-specific serum. No growth factors or other additives were used to induce chondrogenic differentiation and maintain long-term stability of the tissue constructs. The engineered tissue constructs attach, migrate, and integrate with native tissues, thereby meeting important requirements for tissue reconstruction and/or regeneration.
For generating cartilage in vitro that typifies not only the morphology but also sustains the physiological behavior of chondrocytes, it is necessary to culture the cells in a three-dimensional arrangement.(17,18) It is known that a three-dimensional arrangement leads to a specific cell shape and environmental conditions determining gene expression and behavior of cells.(17) The present work shows that a close three-dimensional contact of human spherical chondrocytes enables them to arrange themselves in three-dimensional cell aggregates, to synthesize cartilage-specific proteins and matrix components, and to deposit the components in the intercellular space (Fig. 4). This behavior parallels the natural process of chondrogenesis: aggregation of chondroprogenitor cells followed by the synthesis of a cartilaginous extracellular matrix.(19) Therefore, it is obvious that the cellular environment and cell shape of human chondrocytes in aggregate culture is responsible for this histotypical organization.
A multitude of studies have been done using different ways to culture chondrocytes in three-dimensional systems. First of all, various scaffold materials were used to create a three-dimensional system (e.g., agarose, alginate, collagen, fibrin glue, polyglycolic acid [PGA], and polylactide acid [PLA]).(20–26) All these cell-seeded scaffolds have the growing of cells in a three-dimensional environment accompanied by a regaining or maintenance of cartilage-specific features in common. However, with respect to clinical application, naturally occurring xenogenous and allogenous materials (e.g., type I collagen, hyaluronic acid, or fibrin glue) may be immunogenic and bear safety risks, and their application is partially forbidden by the Drug Act. One attempt to resolve these problems is the use of atelocollagen, where the antigenic determinants on the peptide chain of type 1 collagen (telopeptide) are removed.(9,11) The phenotype of freshly isolated chondrocytes could be maintained in the atelocollagen gel, and a cartilage-like tissue developed after implantation in rabbit and in human. However, L-ascorbic acid is necessary to culture the cell-seeded scaffolds, and patients have to be tested concerning their allergic reaction to atelocollagen. Furthermore, chondrocytes were directly seeded into the gel after isolation from the biopsy, wherefore a rather large cartilage specimen has to be harvested from healthy cartilage tissue from the patient. Other scaffolds are not biodegradable or are resorbed with a greater time constant than cartilage regeneration (e.g., hyaluronic acid).(27) During this resorption process, polymer scaffolds may produce harmful degradation products.(28) Furthermore, the integration of the cell-seeded scaffold to the adjacent normal cartilage is sometimes not shown or often incomplete.(11,14,29)
With the aim to avoid scaffold materials for three-dimensional culture, chondrocytes were grown in micromass or high-density systems. However, cartilage-like morphology and reexpression of cartilage-specific proteins could only be maintained for up to 4 weeks and only in the presence of transforming growth factor β1 (TGF-β1), bone morphogenetic protein (BMP)-2, or ascorbic acid.(30–36) Additionally, most of the growth factors are not permitted for the processing of human cell-based drugs. Using our described autologous spheroid culture system, neither the addition of scaffolds, growth factors, or cytokines nor physical manipulations were necessary to induce the formation of stable cell aggregates and the specific chondrogenic phenotype of cells.
The in vitro-generated cartilage-like tissue is characterized by a time-dependent increased expression of collagen type II, S-100, and cartilage-specific proteoglycans, paralleled by a reduction of the cell-matrix-ratio. This indicates a progressive phenotypical differentiation of chondrocytes and a potential for matrix maturation. The extracellular matrix deposition starts in the core of the spheroids (Fig. 4B), resulting in a gradient of matrix deposition. This initial pattern of matrix deposition is also observed in studies using chondrocyte-seeded scaffolds, where the gradient is equalized in long-time culture.(18) Independent of culture time, the locally different matrix production in spheroids is paralleled by a heterogenous cell morphology comparable with native cartilage tissue: round cells in deeper regions and flattened cells in the outer zone.(37) This stable hyaline-like in vitro tissue morphology indicates optimized tissue culture conditions.
Keeping in mind that tissue engineering-based therapies necessitate high amounts of cells, but only small biopsy specimens with a low yield of cells are available, an augmentation of cell number before a tissue engineering process is unavoidable. Additionally, transferring a small amount of freshly isolated chondrocytes directly into a three-dimensional system leads to a proliferation stop (see Results), resulting in an insufficient cell number and density for tissue regeneration processes.(38) Expanding chondrocytes in monolayer culture results in the loss of their cartilage-specific phenotype and matrix protein expression, and a modified cell behavior (e.g., responsiveness to growth factors).(39–42) Our experiments also show that human chondrocytes cultured as a monolayer shift their collagen expression from type II to type I, paralleled by a loss of the intracellular protein S-100. However, even after seven passages in monolayer culture, chondrocytes restored their cartilage-specific phenotype after transferring into autologous three-dimensional culture. The reexpression of collagen type II shows that the shift in collagen expression is only a transient phenotype. Therefore, the monolayer culture seems to be a useful tool for cell expansion.(43)
To increase the size of in vitro cartilage-like tissue, the initial cell number could be changed or the ability of aggregates to coalesce could be used. From work done in tumor spheroid cultures, it is known that nutrient diffusion is not sufficient to maintain viability of core cells in spheroids larger than 800-1000 μm in diameter.(44–46) For that reason, this study focused on aggregates smaller than 800 μm in height. When smaller aggregates were fused, the fusion process was mediated by the flattened surface cells (Fig. 2A). Morphological changes of chondrocytes accompanying the coalescence of spheroids indicate the potential of the cells to adapt to changed environmental conditions. Furthermore, the migration of outer spheroid chondrocytes on artificial surfaces, as well as on native tissue, showed a potentially capacity of spheroids to adhere and integrate with appropriate structures such as cartilage. This integrative property of spheroidal in vitro cartilage fulfills one of the major challenges confronting in vitro-engineered tissues in regenerative medicine.
Standard cell culture procedures use FCS as medium supplement because of the limited availability of human autologous serum. After replacing autologous serum by FCS in the spheroid culture system, aggregates failed to produce stabile compact spheroids with the chondrocyte-specific phenotype. Neither long-time stability of aggregates, viability of core chondrocytes, nor cartilage-specific protein expression could be maintained (Figs. 5C and 5D). It is assumed that the deviating composition and/or amount of serum components(47) (e.g., in regard to growth factors, hormones, or further xenogenous proteins) are responsible for the altered aggregation, differentiation, and finally the death of human chondrocytes (Fig. 5).
Further standard supplements in cell culture systems are antibiotics to suppress bacterial contamination. The addition of antibiotics to autologous/allogenous culture medium allowed cells in spheroids to survive, but a higher cell/matrix ratio was displayed compared with those cultures maintained in the absence of antibiotics. This inhibited differentiation of chondrocytes may be caused by the inhibition of protein synthesis by streptomycin or an effect of penicillin on extracellular matrix deposition.(48) Taking together, these results show that in vitro three-dimensional explant or tissue engineering studies using human cells or tissues may be optimized by the presence of donor-specific serum and the absence of antibiotics. However, the availability of donor-specific serum is limited. Our results also showed that the supplementation of cell culture medium with pooled human serum is a suitable alternative for in vitro tissue engineering studies.
Prior uses of the three-dimensional spheroid culture system are well established in tumor biology, where cells are cultured as multicellular tumor spheroids (MTS). MTS were developed to maintain tumor cell physiology in vitro that is altered in monolayer culture.(49,50) Important differences between MTS and nontumorigenic spheroids are the proliferation and invasion into neighboring tissue. Tumor cells in the rim of the MTS continue to divide, resulting in a continuous growth of spheroids. In contrast, proliferation of human chondrocytes is inhibited when cultured as spheroids. In contrast to MTS cells, cells from chondrocyte spheroids do not invade adjacent tissue.(51) Results show that chondrocytes from spheroids seem to have the capacity to migrate on the surface of tissue explants and recover osteoarthritic fissures.
This aggregate culture system is a very effective method to generate in vitro cartilage-like tissue without using any scaffold, growth factors, or further additives. Using the aggregate culture technique supplemented only with autologous serum, chondrocytes formed a hyaline-like three-dimensional cell-matrix arrangement. The in vitro engineered tissues are characterized by a long-time stability and show the capacity for integration with native tissue. Reaching a size of approximately 1 mm, the in vitro tissues are suitable for clinical use, pharmaceutical test systems, and scientific studies.
At this stage of our investigations, underlying mechanisms of morphology and physiology of cells and matrix maturation in spheroids are not known. Further studies should clarify if the different environmental conditions of surface/core cells and a gradient of nutrients, oxygen, and metabolics within the spheroids are responsible for this cartilage engineering process.
We are very grateful to Dr. Tim Ganey (Medical Center, Atlanta, GA, USA) for discussion and helpful comments and to Prof. A. Herrmann (Humboldt University of Berlin, Institute of Biology/Biophysics, Berlin, Germany) for critical discussion and referring the manuscript.