During the process of endochondral bone formation, proliferating chondrocytes give rise to hypertrophic cells, which then deposit a mineralized matrix to form calcified cartilage prior to replacement by bone. Previously, we reported that a clonal cell line, ATDC5, undergoes efficient chondrogenic differentiation through a cellular condensation stage. Here we report that the differentiated ATDC5 cells became hypertrophic at the center of cartilage nodules, when the cells ceased to grow. Formation of hypertrophic chondrocytes took place in association with type X collagen gene expression and a dramatic elevation of alkaline phosphate (ALPase) activity. After 5 weeks of culture, mineralization of the culture could be discerned as Alizarin red-positive spots, which spread throughout the nodules even in the absence of β-glycerophosphate. Electron microscopy and electron probe microanalysis revealed that calcification was first initiated at matrix vesicles in the territorial matrix and that it advanced progressively along the collagen fibers in a manner similar to that which occurs in vivo. The infrared spectrum of the mineralized nodules indicated two absorption doublets around 1030 cm−1 and 600 cm−1, which are characteristic of apatitic mineral. Calcifying cultures of ATDC5 cells retained responsiveness to parathyroid hormone (PTH): PTH markedly inhibited elevation of ALPase activity and calcification in the culture in a dose-dependent manner. Thus, we demonstrated that ATDC5 cells keep track of the multistep differentiation process encompassing the stages from mesenchymal condensation to calcification in vitro. ATDC5 cells provide an excellent model to study the molecular mechanism underlying regulation of cartilage differentiation during endochondral bone formation.
In the formation of cartilage, mesenchymal cells pass through at least three distinct differentiation stages: (1) prechondrogenic cells,1 (2) proliferating chondrocytes,2,3 and (3) hypertrophic and calcifying chondrocytes.4,5 During endochondral bone development, mesenchymal cells undergo an orderly series of events which involves the transitions of prechondrogenic cells to proliferating chondrocytes through cellular condensation (early phase differentiation). Chondrocytes at the proliferating stage then proceed to the hypertrophic stage (late phase differentiation). In permanent cartilage, e.g., normal articular cartilage on the joint surface, chondrocytes do not undergo the late phase conversion of phenotype. The late phase differentiation is characterized by a several-fold increase in cell volume and a marked increase in alkaline phosphatase (ALPase) activity.6 Hypertrophic chondrocytes eventually mineralize the surrounding cartilage matrix to allow invasion of blood vessels, leading to the replacement of cartilage by bone.
This biphasic transition of the cellular phenotype is accompanied by a change in collagen gene activation: prechondrogenic mesenchymal cells express type I collagen mRNA, and the early phase differentiation is characterized by inductive expression of type II and IX collagen genes as well as the aggrecan gene.2,7 The late phase differentiation is characterized by the onset of expression of the short-chain collagen type X gene, which thus far has been found only in hypertrophic and calcifying chondrocytes.7,8 Hypertrophy of chondrocytes is also accompanied by reduction of aggrecan and type II collagen expression.7,8 To date, there is no report describing a clonal cell line that mimics the late phase differentiation giving rise to calcifying chondrocytes in vitro, although a few cell lines are known to be chondrogenic.9,10
The clonal cell line, ATDC5, was isolated from the feeder-independent teratocarcinoma stem cell line AT805 on the basis of chondrogenic potentials in the presence of insulin.11 In the preceding study, we reported that ATDC5 cells reproducibly undergo the early phase differentiation of chondrocytes to form numerous cartilage nodules in the presence of insulin.12 Cartilage nodules increased in size due to the proliferation of chondrocytes and then spontaneously ceased to grow.12 After the cessation of growth, we found that hypertrophic cells appeared in the center of cartilage nodules in association with type X collagen gene expression and a dramatic elevation of ALPase activity in culture. Matrix vesicles, ALPase, and type X collagen are all implicated in the mineralization of cartilage.13–15 Under the appropriate culture conditions, hypertrophic ATDC5 cells initiated mineral deposition via the matrix vesicle (MV)-mediated mechanism in the preformed cartilage matrix. Mineralization in the culture markedly propagated with time in culture.
β-glycerophosphate has been a culture supplement used for the stimulation of mineralization in vitro, although it sometimes gives artifactual mineralization. β-glycerophosphate supplementation was not required for mineralization of the differentiated ATDC5 cells. Physicochemical analyses were carried out to characterize the mineralized cartilage matrix formed in the culture of ATDC5 cells. Here we demonstrate that ATDC5 cells reproduce the orderly transition of the differentiation stages of chondrocytes observed in vivo during endochondral bone formation. Cellular responses upon activation of parathyroid hormone (PTH)/PTH-related peptide (PTHrP) receptor during mineralization are also discussed. To our knowledge, ATDC5 cells are the first example of a prechondrogenic stem cell line which provides an excellent mineralizing culture in vitro without β-glycerophosphate supplementation and offers an opportunity to analyze the molecular mechanism underlying cartilage mineralization during endochondral bone formation.
MATERIALS AND METHODS
Cell lines and culture conditions
For chondrogenic induction, ATDC5 cells were cultured in a 1:1 mixture of Dulbecco's modified Eagle's (DME) and Ham's F12 (DME/F12) medium (Flow Laboratories, Irvine, U.K.) containing 5% fetal bovine serum (FBS: GIBCO, New York, NY, U.S.A.), 10 μg/ml bovine insulin (I; Wako Pure Chemical, Osaka, Japan), 10 μg/ml human transferrin (T; Boehringer Mannheim, Mannheim, Germany), and 3 × 10−8 M sodium selenite (S; Sigma Chemical Co., St. Louis, MO, U.S.A.) at 37°C in a humidified atmosphere of 5% CO2 in air for the initial 3 weeks, as previously described.11,12 Inoculum density of the cells was 2 × 104 cells/well in a 24-multiwell plate, 4 × 104 cells/well in a 12-multiwell plate, or 6 × 104 cells/well in a 6-multiwell plate (Corning, New York, NY, U.S.A.). On day 21, the culture medium was switched to alpha modified essential medium (α-MEM) containing 5% FBS plus ITS, and the CO2 concentration was shifted to 3% to facilitate mineralization in culture. The medium was replaced every other day. For comparison, growth plate chondrocytes were isolated from the ribs of young male New Zealand rabbits, as previously described.16 The cells were plated at an initial density of 8 × 104 cells/well in a 12-multiwell plate. The cells were grown in α-MEM and 10% FBS at 37°C in 5% CO2 in air for the initial 2 weeks. The CO2 concentration was then shifted to 3% to facilitate mineralization in culture.17 The medium was replaced every other day.
ATDC5 cells were plated at an initial density of 4 × 104 cells/well in a 12-multiwell plate and cultured in the presence of 5% FBS plus ITS. At each time point, cells were rinsed with PBS and fixed with 95% methanol for 20 minutes. They were then stained with 1% Alizarin red S (Wako Pure Chemical) for 5 minutes and destained with distilled water five times, according to the method of Dahl.18
For electron microscopic (EM) analysis,19 ATDC5 cells were inoculated on Thermanox cover slips (22 × 60 mm; Nunc, Naperville, IL, U.S.A.) at an initial density of 8 × 104 cells/well. On days 24, 28, and 35, cells were rinsed with PBS and fixed in cold 2.5% glutaraldehyde (pH 7.2) for 2–4 h and postfixed in 1% OsO4 solution containing 1.5% potassium ferrocyanide. After being dehydrated through a graded ethanol series, these specimens were embedded in Epon 812. Ultrathin sections were cut with a diamond knife on an LKB-ultramicrotome (LKB-8800, Bromma, Sweden) and stained with uranyl acetate and lead citrate prior to observation with a JEM-100B electron microscope (JEOL, Tokyo, Japan). Mineral deposits in the cultures were analyzed by scanning electron microscopy using an electron probe microanalyzer (JXA-8900L; JEOL). Unfixed specimens were prepared after 35 days of culture for microanalysis; the cells were rinsed first with PBS and subsequently with distilled water. After the specimens were air dried, they were carbonated and scanned. Component images and the electron probe spectrum were obtained.
Fourier transform infrared spectroscopy and measurement of calcium content
ATDC5 cells were cultured in the presence of 5% FBS plus ITS in 12-multiwell plates. On day 21 and day 52 of culture, cells were rinsed with PBS, fixed with 95% methanol for 20 minutes, and rinsed with distilled water. Then the cell layers were air dried. Cartilage nodules were scraped off and were incorporated into KBr discs, which were subjected to Fourier transform infrared (FT-IR) spectroscopy (JIR-7000; JEOL). Thirty spectral scans were performed over the wavenumber range of 4000–400 cm−1 with 4 cm−1 resolution. For comparison, bone-derived hydroxyapatite (a generous gift from Dr. K. Kawauchi, JEOL) and the mineralized culture of primary rabbit growth plate chondrocytes were also analyzed. For measurement of calcium, ATDC5 cells were fixed with ethanol after rinsing with ice-cold PBS. The cell layers were extracted with 4 M HCl containing 1% LaCl. Calcium content in the solution was measured by atomic absorption spectrophotometry at 422.7 nm with AA-660 atomic absorption/flame emission spectrophotometer (Shimazu, Kyoto, Japan). Calcium standard solution was from Wako Pure Chemical.
Measurement of ALPase activity
ATDC5 cells were plated in 24-multiwell plates and cultured in the presence of 5% FBS plus ITS. At each time point, cells were rinsed with PBS. In some experiments, cells were plated in 12-multiwell plates and cultured. On day 28, the cells were treated with [Nle8, Nle18, Tyr34]bovine PTH(1–34) amide (bPTH[1–34]; Bachem, Torrance, CA, U.S.A.) for an additional week and then rinsed with PBS. The culture medium was replaced every other day. ALPase activity was measured by a modification of the method of Bessey et al.20 using p-nitrophenyl phosphate (pNPP) as a substrate. Cell layers were homogenized in 0.9% NaCl/0.2% Triton X-100 at 0°C and centrifuged for 15 minutes at 12,000g. The supernatant, which contained 95% of the total activity, was assayed in a mixture of 0.5 mM pNPP and 0.5 mM MgCl2. The reaction mixture was incubated for 5–15 minutes at 37°C. The reaction was terminated by the addition of 0.25 volume of 1 M NaOH, and the concentration of p-nitrophenol generated was determined by spectrophotometry at 410 nm. p-nitrophenol was used as a standard. The enzyme activity was expressed in nanomoles of pNPP cleaved per minute per microgram of DNA. DNA content was determined by the method of Johnson-Wint.21
RNA extraction and hybridization analysis
ATDC5 cells were cultured in the presence of 5% FBS plus ITS in 6-multiwell plates. Total RNA was prepared from the cultures by a single-step method according to Chomczynski and Sacchi.22 Total RNA (20 μg) was denatured with 6% formaldehyde, separated by 1% agarose electrophoresis, and transferred to Nytran membranes (Schleicher & Schuell, Dassel, Germany). Hybridization was performed overnight at 42°C with an appropriate probe (106 cpm/ml) in solutions containing 50% formamide, 6× SSPE (2× SSPE contained 0.3 M NaCl, 20 mM NaH2PO4, 2 mM EDTA at pH 7.4), 0.2% bovine serum albumin (BSA), 0.2% Ficoll 400, 0.2% polyvinylpyrrolidone, 0.1% SDS, and 100 μg/ml denatured salmon sperm DNA. Hybridization probes were prepared by the random-primer method with a BcaBEST Labeling kit (Takara, Shiga, Japan) using the appropriate cDNA fragments: 1.4 kb EcoRI fragment of pKT118023 as a probe for α1(II) collagen mRNA; 0.9 kb EcoRI fragment of pKT164323 as a probe for α1(IX) collagen mRNA; 0.65 kb HindIII fragment of pSAm10h24 as a probe for α1(X) collagen mRNA; 0.87 kb PstI fragment of p135525 as a probe for aggrecan mRNA; 2.2 kb EcoRI-XhoI fragment of pcDNA1R15B26 as a probe for PTH/PTHrP-receptor mRNA; 2.5 kb EcoRI fragment of pBS+ harboring rat bone-liver-kidney type ALPase cDNA27 as a probe for ALPase mRNA; and 1.2 kb EcoRI fragment of pBS+ harboring rat osteopontin cDNA28,29 as a probe for osteopontin mRNA. The filters were washed twice for 15 minutes each time at 55°C in 0.1× SSPE and 0.1% SDS and exposed to Kodak X-OMAT film at −80°C with a Cronex lightening plus intensifying screen (DuPont, Boston, MA, U.S.A.). For slot-blot hybridization analysis, RNA (4 μg or 1 μg) was isolated and denatured with 6% formaldehyde and applied to Nytran membranes with a Schleicher & Schuell Manifold II slot-blot apparatus. Quantification was performed by scanning densitometry of the autoradiographs with a Hoefer GS-300 Scanning Densitometer (Hoefer Scientific Instruments, San Francisco, CA, U.S.A.).
Treatment of cells with bPTH(1–34)
ATDC5 cells were plated in 12− or 24-multiwell plates. After the cells were grown in the differentiation medium for the initial 28 days, they were treated with bPTH(1–34) for another 1–2 weeks. The culture medium was replaced every other day. The effect of bPTH(1–34) on calcification was monitored by Alizarin red S staining of the cells. The ALPase activity in the bPTH(1–34)-treated culture was also determined.
Cellular hypertrophy and mineralization of differentiated ATDC5 cells
ATDC5 cells retained the properties of chondroprogenitor cells and rapidly proliferated to form a confluent monolayer in vitro12 (Fig. 1a). In the presence of insulin (10 μg/ml), the culture initiated chondrogenic differentiation to form cartilage nodules through a cellular condensation process that gave rise to proliferating chondrocytes. The cartilage nodules increased in size due to proliferation of chondrocytes, which continued for about 2 weeks and then ceased by day 21 of culture.12 By day 24, hypertrophic cells were discernible in the central regions of the nodular structures under a phase-contrast microscope (Fig. 1b). On day 35, the central regions of the cartilage nodules looked dark under a phase-contrast microscope (Fig. 1c). The hypertrophic cells occupied the cartilage nodules as the major cell population. Alizarin red S densely stained the light-impenetrable regions of the nodules, indicating mineralization of hypertrophic chondrocytes (Fig. 1d).
Figure 2 shows the time-course of Alizarin red S staining of ATDC5 cell culture. The cells became progressively hypertrophic and no Alizarin red S positive nodules were seen in the culture by day 28. On day 33, mineralized regions began to appear in the preexisting cartilage nodules as tiny spots and they propagated throughout the cartilage nodules with time. No mineral deposition was seen in the flat cell monolayer surrounding the cartilage nodules. Calcium content of the culture increased from 2.94 μg/well on day 30 to 22.97 μg/well on day 42. Standard deviations was within 5% of the average in the triplicate assay. By day 55, the culture was filled with mineralized cartilage nodules. Heavily mineralized matrix was readily recognized as white precipitates covering the culture plates without staining (Fig. 2; nonstaining). In the preliminary study, mineralization was markedly facilitated by the shift of the CO2 concentration from 5 to 3% in air in primary cultures of rabbit growth plate chondrocytes (data not shown). Similarly, mineralization of ATDC5 cell cultures was greatly propagated in the atmosphere of 3% CO2 in air. When cultured in 5% CO2 in air, the cultures were only poorly mineralized in the absence of β-glycerophosphate (Fig. 2).
The ultrastructure of ATDC5 cells was examined throughout the course of differentiation by transmission EM. Cells prior to chondrogenic differentiation retained an ultrastructure characteristic of undifferentiated immature cells: the cells were preserved with large oval euchromatin-rich nuclei containing prominent nucleoli, the profile of rough endoplasmic reticulum, and several mitochondria. The Golgi apparatus was relatively small and consisted of several groups of thin stacks. The territorial matrix was composed of amorphous fine materials (data not shown). On day 24, differentiated cells inside the cartilage nodules revealed an ultrastructure characteristic of early hypertrophic chondrocytes seen in vivo; the cells were preserved with large euchromatic nuclei, lysosomal dense bodies, vacuoles, small mitochondria, and Golgi-endoplasmic reticulum systems (Fig. 3a). The well-developed Golgi apparatus contained many condensing vacuoles and vesicles. Rough endoplasmic reticulum developed at the periphery of the cell bodies, and usually one or more vacuoles were formed in association with glycogen aggregates. Cells in lacunae were surrounded by a well-developed extracellular matrix of randomly distributed collagen type II fine fibrils (about 20 nm in diameter). As shown in Fig. 3b, dense aggregates of fine needle-like crystals were found in association with matrix vesicles that were in the territorial matrix. By day 35, the cells modulated to the small spherical cells but still retained relatively abundant organelles such as mitochondria, dense bodies, and Golgi apparatus. The nuclei were usually eccentric and showed a peripheral rim of heterochromatin. The Golgi apparatus consisted of a stack of short cisternae. Mineralized matrix vesicles in the extracellular matrix were fused with one another and formed calcospherites at the early stage of mineralization (Fig. 3c).
Cultures of differentiated ATDC5 cells reproducibly mineralized, and mineral deposits were located in extracellular matrix, principally associated with collagen fibrils. Ascorbic acid facilitated the appearance of hypertrophic cells in the cartilage nodules, as well as the elevation of ALPase activity and expression of type X collagen mRNA. In its absence, mineralization was rarely observed. The supplementation of DME/F12 medium with ascorbic acid (50 μg/ml) induced a phenotypic transition of proliferating chondrocytes to hypertrophic cells, although mineralization did not propagate well. Iscove's modified Dulbecco's medium supplemented with 50 μg/ml ascorbic acid also worked well for the expression of the hypertrophic phenotype. In the present study, we employed α-MEM, which contains 50 μg/ml ascorbic acid in the standard formula, for induction of the hypertrophic phenotype in the cells. Since ATDC5 cells similarly undergo chondrogenic differentiation in α-MEM in the presence of insulin, the cells exhibited the entire spectrum of cellular stages from undifferentiated to mineralized in α-MEM containing 5% FBS plus ITS. However, the growth curve of the cells suggested that proliferation of chondrocytes ceased by day 15 in α-MEM, 1 week earlier than that in DME/F12 medium (data not shown). This resulted in the formation of smaller cartilage nodules where mineral deposition occurred. In the presence of 5% FBS plus ITS, the final cell density at the end of the proliferating stage in α-MEM was about 4.1 × 105 cells/cm2, as compared with approximately 9.1 × 105 cells/cm2 in DME/F12 medium. Thus, in the present study, ATDC5 cells were cultured in optimum conditions that included DME/F12 medium for the initial 3 weeks and then α-MEM in the presence of 5% FBS and ITS for the remaining time in culture.
β-glycerophosphate (usually 4–20 mM) has been an important culture supplement for the stimulation of extracellular mineralization of chondrocytes as well as osteoblasts,8,30,31 although it sometimes results in artifactual mineralization in culture. It should be noted here that mineralization was initiated and propagated successfully in the absence of β-glycerophosphate or other particular supplements in the culture of ATDC5 cells (Fig. 2), as in the case of rabbit growth plate chondrocytes in primary culture.17 In contrast, mineralization was markedly facilitated in the atmosphere of the lower CO2 concentration (3% CO2 in air), as shown in Fig. 2. Since we found a slight reduction in the efficiency of chondrogenesis and the subsequent growth of cartilage nodules under 3% CO2 in air, the cells were maintained in 5% CO2 in air for the initial 3 weeks. The culture was then shifted to 3% CO2 in air for mineralization.
Characterization of mineralized matrix surrounding ATDC5 cells
ATDC5 cells were analyzed at the early stage of mineralization (day 35) by the electron probe microanalyzer (Fig. 4). As shown by the secondary electron image of the culture (Fig. 4a), mineral deposits were recognized as electron-reflecting granular materials in the territorial matrix surrounding the cells. The electron probe spectrum of the granules is shown in Fig. 4b. Calcium and phosphorus were identified as the major elemental components of the granular materials. Peak positions and relative peak heights were superimposable with those of the control spectrum obtained from the mineralizing culture of rabbit growth plate chondrocytes (data not shown). Heavily mineralized cell layers were scraped off and subjected to FT-IR spectroscopy. The absorption doublets at 900–1250 cm−1 (the v3 PO43− domain) and 500–650 cm−1 (the v4 PO43− domain) were characteristic of apatitic mineral among the four vibrational modes of phosphate ion, as exemplified by the IR spectrum of purified hydroxyapatite derived from bone (Fig. 5a). The doublet at 560–600 cm−1 is characteristic of crystalline calcium phosphate, and the splitting of the 1030–1130 cm−1 P-O stretch band is characteristic of apatites formed via an octacalcium phosphate intermediate.15 These P-O stretch bands were absent in unmineralized ATDC5 cell layers (Fig. 5b). Absorption bands in the spectrum were mostly derived from organic components in the cell layers (Fig. 5b). Many of them were assigned to vibrational modes of peptide bonds. In contrast, mineralized ATDC5 cell layers (day 52) evidently revealed the characteristic splitting in the v3 and the v4 PO43− domains (Fig. 5c). The spectrum was also superimposable with that obtained from a heavily mineralized culture of rabbit growth plate chondrocytes (Fig. 5d). These physicochemical analyses of mineral deposits suggested that the ATDC5 cell culture reproduced matrix mineralization of cartilage during endochondral bone formation in vivo.
Elevation of alkaline phosphatase activity and induction of type X collagen gene expression during cellular hypertrophy and mineralization
The marked elevation of ALPase activity and expression of type X collagen were closely associated with phenotypic transition of the cells from the growing stage to the hypertrophic and mineralizing stages. The time course of the ALPase activity was determined during the chondrogenic differentiation of ATDC5 cells, and each value was normalized for the DNA content in the culture (Fig. 6a). The enzyme activity was detectable but low (<0.1 nmol/minute/μg DNA) in undifferentiated ATDC5 cells on day 3. An increase in ALPase activity was first recognized during days 16–20 at the expense of growth in culture, even in the absence of ascorbic acid. The enzyme activity during this time period remained approximately 2 nmol/minute/μg DNA. ALPase activity progressively increased as mineralization propagated in the culture on day 30 or later. On day 47, ALPase activity was over 200-fold higher than that of cultures at the proliferating stage of chondrocytes on day 13 or earlier (Fig. 6a).
As shown in Fig. 6b, we examined the expression of the bone/liver/kidney-type ALPase gene during the chondrogenic differentiation of ATDC5 cells. A low level of ALPase mRNA was expressed in undifferentiated cells (Fig. 6b, lane 1). Upon chondrogenic differentiation, the expression of ALPase mRNA significantly up-regulated during the growth of cartilage nodules (Fig. 6b, lanes 2 and 3). It reached a maximum as the cells matured to become hypertrophic, from day 24 to day 35 (Fig. 6b, lane 4), and then gradually declined in mineralized chondrocytes (Fig. 6b, lane 5). Thus, a dramatic elevation of ALPase activity in culture did not simply correlate with the level of the ALPase gene transcripts, but it was associated with an accumulation of the ALPase-rich matrix vesicles in culture.
Type II collagen gene was expressed in association with the chondrogenic differentiation of ATDC5 cells (Fig. 7a), as previously described.12 The expression of type IX collagen mRNA followed a time course parallel to that of type II collagen mRNA (Fig. 7a). Figure 8 summarizes the temporal pattern of the cartilage marker gene expression monitored by slot-blot analysis. Expression of the aggrecan gene was initiated by chondrogenic differentiation of ATDC5 cells. As previously reported in chick limb-bud chondrogenesis,3 a dramatic increase in the accumulation of aggrecan mRNA was preceded by that of type II collagen mRNA (Fig. 8). Expression of the PTH/PTHrP receptor gene occurred in close association with early chondrogenesis.12 Transcripts for these three genes reached a maximal level by the time that cartilage nodules stopped growing, on day 21. Thereafter, mRNA levels for these genes declined as the cells proceeded to the hypertrophic stage.
Expression of the type X collagen gene, a marker of hypertrophic chondrocytes, could not be detected in undifferentiated ATDC5 cells or in proliferating chondrocytes (Figs. 7b and 8). Transcripts for the type X collagen gene were first detected on day 21 at the end of the proliferating stage, and then levels of transcripts increased as the culture became mineralized (Figs. 7b and 8). In contrast, the osteopontin gene transcripts were detected in the cells at the proliferating stage on day 7 and day 14 (Fig. 7b). A low level of osteopontin mRNA was also detected in the undifferentiated ATDC5 cells on day 2 by a longer exposure of a film of a Northern blot. However, the expression of osteopontin mRNA was significantly up-regulated at the end of the proliferating stage and then gradually increased as the cells matured to be hypertrophic (Fig. 7b).
Effect of PTH on mineralization of ATDC5 cells
We studied the effect of PTH on mineralization of ATDC5 cells. Cells were grown in DMEM/F12 and then α-MEM in the presence of 5% FBS and ITS for the initial 28 days to become hypertrophic. The culture was then incubated in the presence of various concentrations of bPTH(1–34) for 1 week. As shown in Fig. 6a, ALPase activity gradually increased in the control culture during this time period. However, treatment with bPTH(1–34) markedly inhibited the elevation of ALPase activity in culture in a dose-dependent manner (Fig. 9). The inhibitory effect became evident with the incubation of 10−10 M bPTH(1–34). With 10−8 M bPTH(1–34), the ALPase activity was only 30% that in the untreated culture. In parallel, the effect of bPTH(1–34) on mineralization was monitored by Alizarin red S staining (Fig. 10). Treatment with bPTH(1–34) dose-dependently inhibited formation of mineralized nodules in the culture. Mineralization was completely blocked in the culture treated with 10−8 M bPTH(1–34). After treatment for 2 weeks, the inhibitory effect of the agent was clearly recognized on day 42 as a reduction of white mineral deposits (Fig. 10, nonstaining). Accordingly, the mRNA levels of ALPase and type X collagen were markedly reduced within 48 h by the treatment of 10−8 M bPTH(1–34) (data not shown). There was considerable shrinkage of hypertrophic cells and diminution of cartilage nodules during the PTH treatment in culture.
The process of cellular hypertrophy and mineralization of chondrocytes has been morphologically and biochemically studied by using the primary cultures of isolated chondrocytes in vitro.8,15,30,32,33 Through these studies, late phase differentiation of chondrocytes became defined by the several molecular markers. However, multiple passages of the cells resulted in a rapid loss of the differentiated phenotype of chondrocytes. It has been difficult to generate stable cell lines that express the phenotype of chondrocytes, since the phenotype is not stably maintained. It was recently reported that the constitutive expression of the myc oncogene kept avian chondrocytes in the proliferating stage without loss of type II collagen synthesis.34 Expression of the myc oncogene also allowed for the isolation of immortalized mouse limb bud cells with chondrogenic potentials.35 Mallein-Gerin and Olsen36 reported generation of MC615 cells having the phenotype of proliferating chondrocytes by retroviral expression of the large T antigen of simian virus 40 (SV40) in mouse rib chondrocytes. The cells expressed type II, IX, and XI collagen as well as aggrecan and link protein. However, there was no sign of cellular hypertrophy or type X collagen expression.36 Immortalization of chondrocytes prevented these cells from phenotypic conversion into hypertrophic and calcifying chondrocytes. In contrast, in chondrocytes immortalized by a temperature-sensitive SV40 large T antigen, expression of type X collagen was up-regulated upon growth arrest by shifting the temperature to a nonpermissive temperature.37 Therefore, withdrawal from growing cell cycles may be an important prerequisite for the late phase differentiation of chondrocytes.
In the culture of ATDC5 cells, formation of hypertrophic cells in cartilage nodules was also preceded by the growth arrest of chondrocytes (Fig. 1b). The phenotypic conversion of ATDC5 cells into hypertrophic cells proceeded in association with the induction of type X collagen gene expression and the reduction of a level of type II collagen mRNA (Figs. 7 and 8). This observation was compatible with the previous findings during differentiation of chondrocytes in vitro and in vivo.7,38,39 Ascorbic acid has been suggested to be an important supplement of culture medium for induction of the hypertrophic phenotype in chondrocytes and type X collagen expression.8,31,40 In the present study, the supplementation of culture medium with ascorbic acid resulted in the early onset of type X collagen gene expression in differentiated ATDC5 cells (data not shown). However, even in the absence of ascorbic acid, the expression of type X collagen mRNA was evident on day 21 in cells that had been cultured in DME/F12 medium (Figs. 7b and 8). Thus, type X collagen expression per se does not require the presence of ascorbic acid in culture, although the presence of ascorbic acid facilitated a higher level of the gene expression and maintenance of hypertrophic phenotype.
Ultrastructural examination revealed that differentiated ATDC5 cells on day 24 retained cellular and extracellular morphology similar to that of hypertrophic chondrocytes seen in vivo and in vitro (Fig. 3a).8,41,42 In the extracellular matrix, occurrence of matrix vesicles was also noted in association with differentiated cells. Matrix vesicles were found throughout the various zones of the epiphyseal growth plate.41 Their association with mineral deposition was typically initiated in the zone of hypertrophy in vivo. As shown in Fig. 3b, initial mineralization first occurred in matrix vesicles that were abundant in the territorial matrix. On day 35, many of the matrix vesicles were heavily mineralized (Figs. 3c and 4a). Thereafter, mineralization of the culture advanced progressively. By day 55, virtually all cartilage nodules in the culture were covered with the thick mineral phase that was readily discernible without staining (Fig. 2). The electron probe spectrum and FT-IR spectrum were identical to those of mineral deposits formed in the control culture of rabbit growth plate chondrocytes (Figs. 4 and 5).17 These data were compatible with the formation of hydroxyapatite in ATDC5 culture.15,43,44
Matrix vesicles, ALPase, and collagen have all been implicated in the mineralization process.15 Since the expression of the bone/liver/kidney-type ALPase gene is not specific to the mineralizing tissues, ATDC5 cells expressed the gene transcripts even in the undifferentiated stage. The mRNA level for the gene, however, was gradually increased as the cells became matured to be hypertrophic and mineralized until days 35–40 (Fig. 6b). When the culture was heavily calcified, the level of ALPase mRNA appreciably declined (Fig. 6b, lane 5). ALPase activity was low but detectable in undifferentiated or early differentiated ATDC5 cells. High levels of ALPase activity are normally associated with hypertrophic chondrocytes and matrix vesicles which accumulate in the extracellular matrix. There is morphological evidence that matrix vesicles are generated by pinching off the membrane processes on the surface of epiphyseal growth plate chondrocytes.13 Therefore, ALPase activity in culture is mainly correlated with the release and accumulation of ALPase-rich matrix vesicles in the extracellular space. Probably, ALPase activity in undifferentiated or early differentiated cells was very low because of the lack of mechanism for accumulation of the enzyme such as matrix vesicles. Accordingly, ALPase activity elevated much more dramatically (from day 30 or later) due to the accumulation of matrix vesicles, compared with up-regulation of the ALPase mRNA level, during the propagation of mineralization in the culture of ATDC5 cells (Fig. 6a). Type X collagen gene expression was also markedly up-regulated during this time period (Fig. 8). The presence of ascorbic acid in the culture medium was critical for the propagation of mineralization.
There is a growing body of evidence suggesting that there is probably considerable overlap in the osteoblast and hypertrophic chondrocytes. Expression of osteopontin gene was reported to occur in cultured primary chondrocytes in association with cellular hypertrophy and mineralization.45–48 Expression of osteopontin mRNA was also readily detected in the cultures of ATDC5 cells by Northern blot analysis as the cells became hypertrophic and mineralized (Fig. 7b). The level of mRNA increased as late phase differentiation proceeded. However, the very low level of osteopontin mRNA could also be detected in the culture after a longer exposure of the membrane by Northern blot analysis. Roach and others suggested a possibility that the mineralizing chondrocytes underwent trans-differentiation to the osteoblastic cells.45,48,49 However, it is not clear at present whether the differentiated ATDC5 cells undergo trans-differentiation into the cells morphologically resembling osteoblastic cells.
Ballock and Reddi studied the induction of the hypertrophic phenotype of chondrocytes in a chemically defined medium.50 They suggested that insulin and thyroxine were important regulators for the survival and hypertrophy of chondrocytes, although there was no indication of mineralization. In the present study, once initial mineralization took place in the culture of ATDC5 cells, mineralization propagated similarly in α-MEM in the absence of serum as long as insulin was present (data not shown). In the absence of insulin, hypertrophic morphology could not be maintained.
An amplified anabolic action of bone morphogenetic protein (BMP) has been documented during the phenotypic conversion of chondrocytes to hypertrophic.35,50,51 BMP may play a role in the process as an internal mediator for cell–cell interactions. ATDC5 cells express BMP-4 and its receptor transcripts (data not shown). Exogenous BMP was not required for chondrogenic differentiation and mineralization in the ATDC5 cell culture. However, when added as a culture supplement, recombinant BMP-2 facilitated the induction of chondrogenic differentiation and mineralization (Ohta et al., manuscript in preparation).
With regard to β-glycerophosphate supplementation, this agent has been shown by numerous researchers to facilitate mineralization of extracellular matrix.44 Although the regulation of phosphate metabolism is important in mineralization, supplementation with this agent was not required for the progressive mineralization in the culture of ATDC5 cells. In contrast, the CO2 content in the culture environment profoundly affected mineralization. Reducing the CO2 content in air from 5 to 3% greatly facilitated the propagation of mineralization of ATDC5 cells (Fig. 2) as well as rabbit growth plate chondrocytes. The role of nutrition and oxidative metabolism as it relates to the oxygen tension of cartilage may be of considerable importance in vivo, since the mineralized cartilage undergoes progressive vascularization during endochondral bone development.52–55 It was shown in the earlier studies that parietal and long bones were required to be maintained in an aerobic environment by placing them on the wire mesh or on the rocking incubator to keep the bone near the surface of the culture medium in vitro.56,57 These results seemed to indicate that the higher oxygen tension would be favorable for bone cells. In contrast, chondrogenic differentiation was favored in an anaerobic environment during fracture healing or repair of the defects of articular cartilage. For example, generation of cartilaginous tissue always occurred in the deeper zone filled by undifferentiated mesenchymal cells of the full thickness defects of articular cartilage in rabbit.58,59 Thus, we speculate that the shift of the environmental CO2 content conforms with the change of the metabolic states of hypertrophic chondrocytes during mineralization. The present results suggest that the CO2 concentration is one of the important factors influencing the onset of mineralization in culture.
There is accumulating evidence suggesting that PTHrP is an important regulator of endochondral bone development.60–62 As we reported in the preceding paper,12 acquisition of PTH/PTHrP responsiveness was closely associated with the early phase differentiation of chondrocytes. Binding capacity of ATDC5 cells for [125I]bPTH(1–34) markedly increased over the course of differentiation, as did the PTH/PTHrP receptor mRNA level. Differentiated ATDC5 cells have abundant high-affinity receptors for PTH (Kd = 3.9 nM; 3.2 × 105 sites/cell) on their cell surface.12 Lee and coworkers reported the highest expression of PTH/PTHrP receptor mRNA at the transitional zone from the proliferating zone to the hypertrophic zone of cartilage in the developing bone.63 As shown in Fig. 8, the level of PTH/PTHrP receptor mRNA in ATDC5 cells peaked during the transition of the cells from the proliferating to the hypertrophic stage of the culture. Thus, the expression pattern of the gene transcripts in ATDC5 cells appeared to be compatible with the localization of the PTH/PTHrP receptor transcripts in the developing bone in vivo. In the previous report, we demonstrated that undifferentiated ATDC5 cells expressed PTHrP protein into the culture medium and that the level of PTHrP gradually declined as the cells underwent differentiation.12 The in vivo study by in situ hybridization demonstrated expression of PTHrP mRNA in perichondrium and the surrounding mesenchyme in the developing bone.63 This in vivo situation of PTHrP and PTH/PTHrP receptor expression may be reflected in the reciprocal expression pattern of PTHrP and PTH/PTHrP receptor in the ATDC5 cell culture. However, there is the intensive expression site of PTHrP gene in vivo in the perichondrium and the neighboring mesenchymal cells covering epiphyseal cartilage in the developing long bone,64,65 which is absent in the ATDC5 culture. The presence of this intensive source of PTHrP production may modify the expression pattern of PTH/PTHrP receptor mRNA in the resting and growth plate cartilage in vivo. Although expression of the PTH/PTHrP receptor gene transcripts declined as mineralization proceeded (Fig. 8), the binding capacity of the cells remained unchanged until at least day 47 of culture (C. Shukunami, unpublished data). Thus, the hypertrophic cells retained full capacity for PTH binding.
Incubation with bPTH(1–34) profoundly affected cell morphology and the accumulation of ALPase activity at this stage (Fig. 9). These changes may lead to the inhibition of mineralization (Fig. 10). Taken together with the previous findings concerning PTHrP action on early chondrogenesis, the present data suggest that PTHrP signaling negatively affects both early and late phase transitions of the chondrocyte phenotype. Homozygous mutant mice carrying the PTHrP null mutation exhibited a diminution of cartilage growth, associated with premature transitions of the chondrocyte phenotype from proliferating to hypertrophic chondrocytes and accelerated replacement of cartilage by bone.60,61 The early onset of mineralization in cartilage may trigger a premature replacement of cartilage by bone. As demonstrated here, the genetic program for cartilage differentiation per se was carried out in the absence of PTH/PTHrP receptor activation. The expression of PTHrP has been documented at varying levels in cartilage and its surrounding tissues in fetal and postnatal animals.66–68 ATDC5 cells also express PTHrP immunoreactivity in culture medium, the level of which declines during mineralization in the culture.12 PTHrP signaling may modulate endochondral bone development by preventing premature phenotypic transitions of chondrocytes in vivo and may play a role in the formation of the columnar structure of epiphyseal growth plate chondrocytes by ensuring properly coordinated transitions of the cartilage phenotype.
Cartilage mineralization has been extensively studied by using primary cultures of chondrocytes isolated from chicken growth plate cartilage or sterna.33,69 Mineralization appears to be a prerequisite for vascular invasion into cartilage during endochondral bone formation. However, little is known about the regulatory mechanism underlying cartilage mineralization. The molecular mechanism of chondrocyte differentiation has been hampered by a lack of the clonal cell line that traces the multiple stages of differentiation in growth plate cartilage in vitro. Poliard and his coworkers recently reported the mesodermal tripotent progenitor cell line derived from teratocarcinoma.70 Under certain conditions, the cells differentiated into chondrocytes expressing type X collagen after long-term culture. The cells mineralized with expression of the osteoblastic phenotype, but there was no sign of cartilage mineralization. In the preceding12 and present studies, we demonstrated that the confluent monolayer culture of ATDC5 cells was induced to differentiate into proliferating chondrocytes through a cellular condensation process and undergo cellular hypertrophy and mineralization in the absence of β-glycerophosphate. Thus, the mouse clonal cell line ATDC5 exhibits the entire spectrum of chondrocytes, i.e., early and late phase differentiation, and is a useful in vitro model for deciphering the molecular mechanism of endochondral bone development.
We thank Dr. T. Kimura (Osaka University) for rat type II and type IX collagen cDNAs, Dr. B.R. Olsen (Harvard Medical School) for mouse type X collagen cDNA, Dr. G.V. Segre (Massachusetts General Hospital) for PTH/PTHrP receptor cDNA, Dr. G.A. Rodan (Merck Sharp & Dohme Research Laboratories) for ALPase and osteopontin cDNAs, and Dr. Y. Yamada (National Institute of Dental Research) for human aggrecan cDNA. Special thanks are due to Dr. T. Takao (Institute for Protein Research, Osaka University) and Mr. K. Kawauchi (JEOL) for FT-IR spectroscopy and for providing us with authentic bone-derived hydroxyapatite. Thanks are also due to Ms. Ann and Dr. Kevin McCluskey for their critical reading of our manuscript and Miss S. Yamamoto for her valuable secretarial help. This work was partly supported by the Senri Life Science Foundation. Chisa Shukunami is a recipient of a Research Fellowship of the Japan Society for the Promotion of Science for Young Scientists.