The totipotent embryonic stem cell generates various mesodermal cells when stimulated with BMP4. Among the resulting cells, those expressing flk-1 and/or PDGFRα displayed chondrogenic activity in the presence of TGFβ3 and expressed cartilage-specific genes in 7 to 16 day pellet cultures. Depositions of cartilage matrix and type II collagen were detected by day 14. TGFβ-stimulated chondrogenesis was synergistically enhanced by PDGF-BB,resulting in a larger cartilage particle filled with a cartilaginous area containing type II collagen, with a surface cell layer expressing type I collagen. In contrast, noggin inhibited both the TGFβ- and TGFβ+PDGF-stimulated cartilage formation, suggesting that a BMP-dependent pathway is involved. In fact, replacement of TGFβ3 with BMP4 on days 10 to 12 markedly elevated the cartilage matrix deposition during the following 7 to 8 days. Moreover, culture with TGFβ3 and PDGF-BB, followed by the incubation with BMP4 alone, resulted in a cartilage particle lacking type I collagen in the matrix and the surface layer, which suggests hyaline cartilage formation. Furthermore, such hyaline cartilage particles were mineralized. These studies indicate that the PDGFRα+ and/or flk-1+ cells derived from embryonic stem cells possess the full developmental potential toward chondrocytes, in common with embryonic mesenchymal cells.
Introduction
Chondrogenesis is initiated during embryogenesis by the commitment of two types of mesoderm, paraxial mesoderm and lateral plate mesoderm, to differentiate into the sclerotome of the somite and limb mesenchyme,respectively. The former leads to the axial skeleton and the latter to the appendicular skeleton. A number of gene products playing important roles in skeletogenesis have been identified and characterized, and chondrogenesis culture methods using embryonic mesenchymes have also been established(Cancedda et al., 1995). However, the specific cellular events occurring during the transformation from a totipotent epiblast to a chondrocyte remain largely undefined. Taking advantage of the capacity of totipotent embryonic stem (ES) cells to differentiate in vitro into cell types derived from all three germ layers, we have begun to address the cellular events involved by selectively differentiating ES cells into mesodermal lineages and by identifying,isolating and characterizing various progenitor cells for their differentiation potential.
Upon differentiation of ES cells in a conventional serum-containing medium,some cells starts to express the vascular endothelial growth factor (VEGF)receptor-2 (VEGFR-2 or flk-1) protein, whose function is essential for embryonic hematopoiesis and vasculogenesis in mice(Shalaby et al., 1997; Shalaby et al., 1995), and some begin to express the platelet-derived growth factor receptor α(PDGFRα) protein (Kabrun et al.,1997; Nishikawa et al.,1998a). The ES-cell-derived flk-1+ cells contain the hemangioblast, the common progenitor cell for both the hematopoietic and endothelial cell lineages (Choi et al.,1998; Faloon et al.,2000; Nishikawa et al.,1998a; Ogawa et al.,1999). Embryo-derived cells with similar cell-surface characteristics also show hemogenic and angiogenic activities(Nishikawa et al., 1998b; Ogawa et al., 1999). Consistently, both the extraembryonic mesoderm, the origin of the primitive endothelial cells and hematopoietic cells, and the intraembryonic mesoderm,which includes the presumptive endocardium and the proximal lateral mesoderm,express flk-1 (Dumont et al.,1995; Kataoka et al.,1997; Takakura et al.,1997; Yamaguchi et al.,1993). Therefore, flk-1+ cells are considered to represent the hemoangiogenic lateral plate mesoderm, that is, the splanchnopleuric mesoderm. Another part of the lateral plate mesoderm, the somatopleuric mesoderm, participates in the formation of limb bone. However,the potential of flk-1+ cells to generate cartilage and bone has not been reported yet.
During early mouse embryogenesis, the PDGFRα gene is expressed in most of mesodermal cells(Orr-Urtreger et al., 1992; Schatteman et al., 1992), and PDGFRα signaling is essential for normal development of many mesodermal tissues, including the axial skeleton(Soriano, 1997). Interestingly, protein expression appeared to be limited to the paraxial mesoderm, and later to the somite(Takakura et al., 1997). The strong expression of the PDGFRα gene in the somitic block is later restricted in the dermatome and sclerotome, but not in the myotome. PDGFRα is also expressed in the lateral mesoderm, the proximal limb mesenchyme, and the perichondrium. Thus, the ES-cell-derived PDGFRα+ cells may form the paraxial mesoderm, somite cells or the chondrogenic mesenchymal cells, including the sclerotome. However, the chondrogenic potential of the ES-cell-derived PDGFRα+ cells has not yet been demonstrated.
Chondrogenesis from undifferentiated mesenchymal cells can be divided into several stages (Cancedda et al.,1995; Hall and Miyake,2000). The first stage is precartilage condensation, in which the mesenchymal cells are closely packed and start to differentiate into chondroblasts and chondrocytes. Chondroblasts/chondrocytes proliferate and secrete increasing amounts of cartilage matrix macromolecules, until each single cell is completely surrounded by a matrix. Cells at the periphery of condensation are relatively undifferentiated and later form a sheath of spindle-shaped cells, which becomes the perichondrium, around the cartilage rudiment. Each stage requires a different and/or an overlapping set of extracellular factors. For example, transforming growth factor (TGF)β is a potent factor for driving the differentiation of mesenchymal cells toward chondrocytes in vitro (Kulyk et al.,1989). It seems to be necessary for the initial mesenchymal condensation stage (Leonard et al.,1991). In addition, TGFβ also plays an inhibitory role in chondrocyte maturation (Ballock et al.,1993; Kato et al.,1988; Serra et al.,1997; Yang et al.,2001), for which the importance of the perichondrium has been demonstrated (Alvarez et al.,2001). On the other hand, bone morphogenetic protein (BMP)function is also implicated in several steps in cartilage formation. Like TGFβ, BMP has early positive roles, that is, proliferation, maintenance and maturation of chondrocytes, and a negative role at a later stage, that is,delay of terminal hypertrophic differentiation in the growth plate(Capdevila and Johnson, 1998; Duprez et al., 1996; Enomoto-Iwamoto et al., 1998; Minina et al., 2001; Pathi et al., 1999; Pizette and Niswander, 2000; Zou et al., 1997). Different BMPs are detected around the mature chondrocytes in the mid to deep zone of the articular cartilage and the hypertrophic/calcifying zone of the growth plate (Anderson et al.,2000).
BMP4, required for mesoderm formation in mice(Winnier et al., 1995) and for hematopoietic cell genesis during the early embryogenesis of vertebrates, is a strong hematopoietic-cell-inducer for ES cells in a serum-free environment(Johansson and Wiles, 1995; Nakayama et al., 2000). We have previously demonstrated that treatments with BMP4 followed by VEGF, a flk-1 ligand, synergistically stimulate the generation of lymphohematopoietic progenitor cells from ES cells in a serum-free medium, which supports the idea that BMP4-induced mesoderm produces hematopoietic cell progenitors in a flk-1-dependent manner like it does in vivo(Nakayama et al., 2000). Here,we have demonstrated that two types of mesodermal cells, flk-1+cells and PDGFRα+ cells, are induced by BMP4 in serum-free medium, from which macroscopic cartilage particles are generated in vitro. Hematopoietic progenitor cells were generated predominantly from the flk-1+PDGFRα– cells. However, under the optimal conditions established with TGFβ3, BMP4 and PDGF-BB, the flk-1–PDGFRα+ cells, the flk-1+PDGFRα+ cells and the flk-1+PDGFRα– cells produced hyaline cartilage particles, which were further mineralized. Therefore, the ES-cell-derived mesodermal cells seem to possess the full developmental potential to form mature cartilage.
Materials and Methods
Cells, factors and antibodies
The E14 ES cells and the OP9 stromal cell line were obtained and cultured as described previously, except that ES cells were cultured under 7.5%CO2 (Nakayama et al.,1998). Fluorescein isothiocyanate (FITC)- or R-phycoerythrin(PE)-conjugated monoclonal antibodies and their isotype controls for the fluorescence-activated cell sorting (FACS), fetal calf serum (FCS), media,buffers, media supplements, cytokines and all of the tissue culture flasks and plates were obtained as described previously(Nakayama et al., 1998; Nakayama et al., 2000). Added to this study were human insulin-like growth factor-1 (IGF1), human PDGF-AA,human PDGF-BB, mouse thrombopoietin/megakaryocyte growth and differentiation factor (TPO) and human TGFβ3, purchased from R&D Systems(Minneapolis, MN), the rat monoclonal antibodies for mouse flk-1 (clone Ava12)conjugated with PE (Flk-1-PE), purchased from Pharmingen (San Diego, CA), and the monoclonal antibody for mouse PDGFRα (clone APA5), kindly provided by S. Nishikawa (Kyoto, Japan) and conjugated with biotin at Amgen(PDGFRα-bio). The PDGFRα+ cells were detected with allophycocyanin-conjugated streptavidin (SA-APC), purchased from Pharmingen. For the collagen immunostaining analysis, the anti-type II collagen(anti-COL2) monoclonal antibodies (clones 2B1.5 and 6B3) were obtained from NeoMarker, the anti-type X collagen (anti-COL10) monoclonal antibody (clone X53) was from Quartett GmbH, the control mouse IgG was from Pharmingen, the rabbit anti-mouse type I collagen (anti-COL1) polyclonal antibody (AB765P) was from Chemicon (Temecula, CA) and the control rabbit IgG was from Sigma. Mouse noggin protein fused with the human IgG1-Fc portion (noggin-Fc) was purchased from R&D, and human IgG1-Fc fragment (IgG-Fc) was from Jackson ImmunoResearch Lab (West Grove, PA).
Induction of differentiation in a serum-free medium by the embryoid body formation method
Differentiation was performed as described previously(Nakayama et al., 2000). Briefly, E14 ES cells were cultured on a fibronectin-coated 60 mm plate(Becton Dickinson) for 2 days in the KnockOut-SR (Gibco)-based serum-free medium containing 10 ng/ml human leukemia inhibitory factor. The cells were then differentiated at 4500 cells/ml in the serum-free medium containing 0.9%methylcellulose (Stem Cell Technology) in the presence of 1.9 ng/ml BMP4 under 5% CO2, 5% O2. On days 3.6 and 4.6, embryoid bodies(EBs) were collected, combined and treated with 0.5 mM EDTA, and single cell suspensions were obtained by passing the EBs through a 22-gauge needle. The cells were then stained with 2 μg/ml PDGFRα-bio and 4 μg/ml Flk-1-PE or with 2 μg/ml SA-APC. The stained samples were analyzed and sorted using a Vantage SE cell sorter (Becton Dickinson).
Serum-free erythro-myeloid colony-forming cell analysis
Each sorted EB cell fraction was mixed with a serum-free colony-forming cell (CFC) medium, containing 1% methylcellulose and supplemented with a mixture of hematopoietic cytokines, and was distributed into two to four 35 mm bacterial-grade dishes at 2.5×104 cells/plate. The serum-free CFC medium was as described previously(Heyworth and Spooncer, 1993),except that linoleic acid and soybean lipids were replaced with 2% chemically defined lipids (Gibco). The cytokines included were the same as described before for the FACS-purified EB cells, except that 10 ng/ml TPO was added(Nakayama et al., 2000). On days 10 to 12, burst-forming-unit erythrocyte, colony-forming unit (CFU)macrophage/monocyte, CFU neutrophil, CFU mast cell and CFU mix were counted.
Co-culture method for developing erythro-myeloid CFCs,lymphokine-activated killer cells and pre-B cells from sorted EB cells
The OP9 cell co-culture method for generating the erythro-myeloid CFC activity and the pre-B and/or lymphokine-activated killer (LAK) lymphocytes was as described previously (Nakayama et al., 1998). For the former, sorted EB cells were plated at 5×104 cells/well onto a confluent layer of OP9 cells within a six-well plate and were cultured for 3 days without interleukin (IL)-2 and IL-7. Both non-adherent cells and loosely attached cells were then mechanically harvested and subjected to CFC analysis. Lymphoid potential was detected with the same culture with IL-2 and IL-7, except that the sorted cells were plated at 1×104 cells/well. On days 10 to 12, both non-adherent cells and loosely attached cells were mechanically harvested,followed by FACS phenotyping (FACScan, Becton Dickinson) using the Sca-1-FITC and B220-PE antibodies to confirm the generation of B220+lymphocytes: pre-B cells (Sca-1–) and LAK cells(Sca-1+), as described previously(Nakayama et al., 1998).
Serum-free micromass cultures for chondrogenesis
The FACS-purified EB cells were re-suspended at 2×107cells/ml in a serum-free chondrogenesis medium. The medium was a modification of that reported by Mackay et al. (Mackay et al., 1998), in that Dulbecco's modified Eagle's medium (DMEM)was replaced with DMEM (high glucose): Ham's F12=1:1 (Gibco), 0.3% glucose(Sigma), 50 μM monothioglycerol (MTG, Sigma) and 50 ng/ml IGF1 were added. Then, 7.5 μl (1.5×105 cells)/well of the cell suspensions were spotted onto a fibronectin-coated 24-well plate (Becton Dickinson),incubated for 1 hour, and cultured in 1 ml/well of the same medium under 5%CO2, 5% O2 for 7 to 8 days. The culture was washed twice with PBS and the total protein was extracted with LDS gel-loading buffer(Invitrogen) including the protease-inhibitor mix (Boehringer Mannheim) for western blot analysis. For the COL2 analysis, 1 μg/ml 6B3 was used. To detect the sulfated glycosaminoglycans, the culture was fixed with 10%formalin, stained with 1% Alcian blue at pH 1 (American Master Tech Scientific) and destained with acetic acid, according to Tallquist et al.(Tallquist et al., 2000).
The pellet culture was performed as described previously(Mackay et al., 1998) except that the serum-free chondrogenesis medium included 50 μM MTG. The 3-4×105 FACS-purified EB cells were first re-suspended in 0.5 ml of the chondrogenesis medium supplemented with or without 10 ng/ml TGFβ3 and/or other factors indicated. Cells were then centrifuged and cultured as a pellet. On days 14 to 20, each cartilage particle was fixed with 10% buffered zinc formalin (Anatech) or Gendre's fluid(Bedossa et al., 1987) for less than 24 hours at room temperature, paraffin-embedded, sectioned through the center part of each particle and stained with 0.1% Toluidine blue (Sigma)(Sheehan and Hrapchak, 1987). Two additional sections were also made through different parts of each particle, to confirm reproducibility. Some sections were immunostained with 4μg/ml 2B1.5 for COL2 or with 2 μg/ml AB765P for COL1, according to the manufacturer's recommendation, and were counterstained with Gill 2 Hematoxylin(Shandon) (Sheehan and Hrapchak,1987).
Mineralization was induced in the modified hypertrophic differentiation medium described by Mackay et al. (Mackay et al., 1998). We added 50 μM MTG and replaced 50 ng/ml thyroxine with 10 nM T3 (Sigma), as described previously(Alini et al., 1996). After 5 days of culture, the particle was fixed with 10% buffered zinc formalin,paraffin-embedded, sectioned, stained with von Kossa and counterstained with Nuclear FastRed (Sheehan and Hrapchak,1987). Some sections were immunostained with 1:20 diluted X53 for COL10, according to the manufacturer's recommendations.
Cartilage-specific gene expression by the reverse transcriptase-polymerase chain reaction method
Two to five aggregates were harvested at each designated time point and were disrupted immediately in the guanidine isothiocyanate solution provided with the RNeasy Kit (Qiagen). Total RNAs were purified using the protocol recommended by the manufacturer, including the DNase I treatment step. Essentially the same materials and protocols were used for reverse transcription (RT) and nested polymerase chain reaction (PCR) as previously described (Nakayama et al.,1998). Modifications were (1) approximately 0.1 μg cDNA per reaction was used, (2) the annealing temperature was set at 68°C and (3)the cycle number was 21 with an outside primer set, followed by another 21 cycles with the corresponding inside primer set. The primers for cartilage–specific genes, including parathyroid hormone-related protein(PTHrP), aggrecan, cartilage oligomeric matrix protein (COMP), COL2, COL10 and Chordin-like 1 (CHL1), are shown in Table 1.
Gene . | . | Sequence . | Product (bp) . | |||
---|---|---|---|---|---|---|
Outside primers | ||||||
PTHrP | sense | 5′-CGGTTTTCGCTCTTCTTTCAG-3′ | 798 | |||
antisense | 5′-GTTTCCTGGGGAGACAGTTTGAT-3′ | |||||
Aggrecan | sense | 5′-CTTGGTGGCGCTGTAACAAAC-3′ | 1047 | |||
antisense | 5′-TTTCTTCTGCCCGAGGGTTCTA-3′ | |||||
COMP | sense | 5′-ACAGATGCTGCGAGAACTTCA-3′ | 1683 | |||
antisense | 5′-TCATCAGTGGCGGTGTTTACAT-3′ | |||||
COL2 | sense | 5′-AGCAGGGCGAGAGGGGACT-3′ | 1224 | |||
antisense | 5′-GGGAGGACGGTTGGGTATCATC-3′ | |||||
COL10 | sense | 5′-GGCGGAAAGGGGAAACAGGATA-3′ | 1310 | |||
antisense | 5′-CATCGTAGGCGTGCCGTTCTTAT-3′ | |||||
CHL1 | sense | 5′-CATATTGCGTGTTTCAAGACAAGA-3′ | 817 | |||
antisense | 5′-GATTGGGGCAGTGGATTTTC-3′ | |||||
β-actin | sense | 5′-GTGACGAGGCCCAGAGCAAGAG-3′ | 541 | |||
antisense | 5′-GAGGAAGAGGATGCGGCAGTG-3′ | |||||
Inside primers | ||||||
PTHrP | sense | 5′-GCTGCGGAGGCTGGTTCA-3′ | 373 | |||
antisense | 5′-CTTCTTCCCGGGTGTCTTGAGTG-3′ | |||||
Aggrecan | sense | 5′-CCGGAAGTGAGTGGAGAATCTA-3′ | 403 | |||
antisense | 5′-CAGGAAAGTGGCGGTAACAGT-3′ | |||||
COMP | sense | 5′-GTTGCGACACGAGGTCAAGGAG-3′ | 410 | |||
antisense | 5′-GCTGCCCGGTCTCACACTCATT-3′ | |||||
COL2 | sense | 5′-TGCCAAGACCTGAAACTCTGC-3′ | 425 | |||
antisense | 5′-TCGGCCCTCATCTCTACATCAT-3′ | |||||
COL10 | sense | 5′-ACCCAGGGGCTCCAGGAATAG-3′ | 791 | |||
antisense | 5′-TACCCCGTGGTTAGCACTGACAA-3′ | |||||
CHL1 | sense | 5′-TGCGAATACAATGGAACCACTTA-3′ | 506 | |||
antisense | 5′-ACAATGCCAAATGCTCGTAGAT-3′ | |||||
β-actin | sense | 5′-ACTGGGACGACATGGAGAAG-3′ | 336 | |||
antisense | 5′-TGAGGTAGTCCGTCAGGTCC-3′ |
Gene . | . | Sequence . | Product (bp) . | |||
---|---|---|---|---|---|---|
Outside primers | ||||||
PTHrP | sense | 5′-CGGTTTTCGCTCTTCTTTCAG-3′ | 798 | |||
antisense | 5′-GTTTCCTGGGGAGACAGTTTGAT-3′ | |||||
Aggrecan | sense | 5′-CTTGGTGGCGCTGTAACAAAC-3′ | 1047 | |||
antisense | 5′-TTTCTTCTGCCCGAGGGTTCTA-3′ | |||||
COMP | sense | 5′-ACAGATGCTGCGAGAACTTCA-3′ | 1683 | |||
antisense | 5′-TCATCAGTGGCGGTGTTTACAT-3′ | |||||
COL2 | sense | 5′-AGCAGGGCGAGAGGGGACT-3′ | 1224 | |||
antisense | 5′-GGGAGGACGGTTGGGTATCATC-3′ | |||||
COL10 | sense | 5′-GGCGGAAAGGGGAAACAGGATA-3′ | 1310 | |||
antisense | 5′-CATCGTAGGCGTGCCGTTCTTAT-3′ | |||||
CHL1 | sense | 5′-CATATTGCGTGTTTCAAGACAAGA-3′ | 817 | |||
antisense | 5′-GATTGGGGCAGTGGATTTTC-3′ | |||||
β-actin | sense | 5′-GTGACGAGGCCCAGAGCAAGAG-3′ | 541 | |||
antisense | 5′-GAGGAAGAGGATGCGGCAGTG-3′ | |||||
Inside primers | ||||||
PTHrP | sense | 5′-GCTGCGGAGGCTGGTTCA-3′ | 373 | |||
antisense | 5′-CTTCTTCCCGGGTGTCTTGAGTG-3′ | |||||
Aggrecan | sense | 5′-CCGGAAGTGAGTGGAGAATCTA-3′ | 403 | |||
antisense | 5′-CAGGAAAGTGGCGGTAACAGT-3′ | |||||
COMP | sense | 5′-GTTGCGACACGAGGTCAAGGAG-3′ | 410 | |||
antisense | 5′-GCTGCCCGGTCTCACACTCATT-3′ | |||||
COL2 | sense | 5′-TGCCAAGACCTGAAACTCTGC-3′ | 425 | |||
antisense | 5′-TCGGCCCTCATCTCTACATCAT-3′ | |||||
COL10 | sense | 5′-ACCCAGGGGCTCCAGGAATAG-3′ | 791 | |||
antisense | 5′-TACCCCGTGGTTAGCACTGACAA-3′ | |||||
CHL1 | sense | 5′-TGCGAATACAATGGAACCACTTA-3′ | 506 | |||
antisense | 5′-ACAATGCCAAATGCTCGTAGAT-3′ | |||||
β-actin | sense | 5′-ACTGGGACGACATGGAGAAG-3′ | 336 | |||
antisense | 5′-TGAGGTAGTCCGTCAGGTCC-3′ |
Results
BMP4 induced flk-1+ and PDGFRα+ progeny from ES cells in a serum-free medium
First, using the EB formation method in a serum-free medium, we determined whether BMP4 would allow ES cells to generate the flk-1+ cells and the PDGFRα+ cells, as demonstrated using the two-dimensional differentiation method in a serum-containing medium(Nishikawa et al., 1998a). As shown in Fig. 1A, 1.9 ng/ml of BMP4 induced flk-1+ and PDGFRα+ cells from ES cells by day 4.6 of differentiation.
After FACS isolation, single positive cells(flk-1–PDGFRα+ and flk-1+PDGFRα–), double positive (DP) cells(flk-1+PDGFRα+) and double negative (DN) cells(flk-1–PDGFRα–) were individually examined for their erythro-myeloid CFC activities and lymphoid potentials. Prior to the OP9 culture, CFCs were hardly detected in any of these cell fractions. However, the CFC activities were strongly induced from the flk-1+PDGFRα– cells and, to a lesser degree,from the DP cells in 3 days on OP9 (Fig. 1B). In contrast, the flk-1–PDGFRα+ and DN cell fractions were virtually devoid of such a CFC-generating activity. Similarly, the flk-1+PDGFRα– cells and, to a lesser degree,the DP cells, eventually generated B220+ lymphocytes during the 10 to 12 day culture on OP9 in the presence of IL-2 and IL-7(Fig. 1C). The flk-1–PDGFRα+ and DN cell fractions never generated B220+ pre-B or LAK cells. Endothelium-like cell clusters were also observed during the culture of flk-1+PDGFRα– cells and DP cells on OP9,whereas flk-1–PDGFRα+ cells never generated such cells (data not shown).
These results were consistent with the previous observations that the FACS-purified flk-1+ cells have CFC-generating (hemogenic) activity and angiogenic activity. Thus, the mesodermal progenitor cells expressing flk-1 and/or PDGFRα generated with BMP4 in the serum-free medium are likely to be equivalent to those made in the serum-containing medium described before.
In vitro cartilage formation with the EB cells in the presence of TGFβ
Both the paraxial mesoderm and the lateral plate mesoderm form cartilage in vivo. Therefore, we addressed whether the ES-cell-derived mesodermal cells were also capable of forming cartilage. The sorted EB cell fractions were subjected to the serum-free pellet culture with 10 ng/ml TGFβ3, as described for mesenchymal stem cells(Mackay et al., 1998). 24 hours later, the cells pelleted in a tube at the beginning of the culture were detached from the bottom to form a round cell aggregate. By day 14, areas that stained light- to dark-blue with Alcian blue or light- to dark-red/purple with Toluidine blue (metachromatic staining) became noticeable within the cell particle, which is indicative of cartilage-like extracellular matrix accumulation.
At frequencies of 96%, 87% and 65%(Table 2), the flk-1–PDGFRα+ cells,flk-1+PDGFRα– cells and DP cells,respectively, formed a particle containing one to three cartilage nodules that consisted of well separated, round cells and large intercellular spaces that stained metachromatically with Toluidine blue(Fig. 2, Fig. 3A). The other areas contained loosely attached, spindle-shaped cells and showed weak to no metachromatic staining with Toluidine blue. For further confirmation, the contiguous sections were stained with an anti-COL1 antibody as well as an anti-COL2 antibody (Fig. 3A). Most of the areas that stained weakly with Toluidine blue expressed both COL1 and COL2, whereas some cartilage nodules that strongly stained with Toluidine blue consisted predominantly of COL2 (Fig. 3A), suggesting that the particles were heterogeneous, in that they consisted of hyaline cartilage nodules, fibro cartilage areas and non-cartilaginous areas. On the other hand, the DN cells formed a large particle, consisting mostly of a non-cartilaginous area that was not metachromatically stained (Fig. 2G). The DN cell population was expected to be heterogeneous,because APA5 is known to recognize part of the PDGFRα mRNA-expressing mesoderm (Takakura et al.,1997).
Factor* added . | flk-1+PDGFRα- . | flk-1+PDGFRα+ (DP) . | flk-1-PDGFRα+ . | flk-1-PDGFRα- (DN) . |
---|---|---|---|---|
No factor | 0/2 | 0/6 | 7/10 (0)† | ND‡ |
TGFβ3 | 26/30 (0) | 11/17 (0) | 44/46 (0) | 2/2 (0) |
BMP4 | ND | ND | 0/3 | ND |
PDGF | 0/1§ | 0/2 | 3/3 (0) | ND |
TGFβ3+BMP4 | 3/3 (3) | 1/1 (1)§ | 5/5 (5) | ND |
TGFβ3+PDGF | 4/4 (3) | 5/5 (1) | 6/6 (5) | ND |
TGFβ3+noggin | 0/1§ | ND | 0/4 | ND |
TGFβ3+PDGF+noggin | 0/1§ | 0/1§ | 0/2 | ND |
Factor* added . | flk-1+PDGFRα- . | flk-1+PDGFRα+ (DP) . | flk-1-PDGFRα+ . | flk-1-PDGFRα- (DN) . |
---|---|---|---|---|
No factor | 0/2 | 0/6 | 7/10 (0)† | ND‡ |
TGFβ3 | 26/30 (0) | 11/17 (0) | 44/46 (0) | 2/2 (0) |
BMP4 | ND | ND | 0/3 | ND |
PDGF | 0/1§ | 0/2 | 3/3 (0) | ND |
TGFβ3+BMP4 | 3/3 (3) | 1/1 (1)§ | 5/5 (5) | ND |
TGFβ3+PDGF | 4/4 (3) | 5/5 (1) | 6/6 (5) | ND |
TGFβ3+noggin | 0/1§ | ND | 0/4 | ND |
TGFβ3+PDGF+noggin | 0/1§ | 0/1§ | 0/2 | ND |
ES-cell-derived mesoderm cells cultured for 14 to 20 days were stained with Toluidine blue. The resulting particles containing cartilage nodules displaying weak to strong metachromatic staining were scored and are indicated per total number of particles.
TGFβ3 10 ng/ml, BMP4 50 ng/ml, PDGF (i.e. PDGF-BB) 50 ng/ml, and noggin (i.e. noggin-Fc) 1 μg/ml.
Not determined.
Single experiment.
In conclusion, the pellet culture in the presence of TGFβ3 allowed fractions of ES-cell-derived mesodermal cells, such as flk-1–PDGFRα+ cells,flk-1+PDGFRα– cells and DP cells, to form a cartilage-nodule-containing particles.
Cartilage-specific gene expression during the pellet culture of the ES-cell-derived mesodermal cells
For further validation, we investigated whether the in-vitro-formed cartilage expressed cartilage-specific genes by nested RT-PCR(Fig. 3B). Prior to the induction of chondrogenesis (day 0), transcripts for aggrecan COMP, COL10 and PTHrP were hardly detected in the flk-1–PDGFRα+ and flk-1+PDGFRα– cell fractions, whereas the COL2 transcript was readily detectable. Upon differentiation, all of them were upregulated and became detectable by days 7 to 16. Interestingly, the CHL1 transcript, which is preferentially expressed in condensing mesenchymes as well as mature chondrocytes of the developing skeleton(Nakayama et al., 2001), and the PTHrP transcript were induced earlier (from day 3). Consistent with the previous histological observations, these results supported the idea that the ES-cell-derived mesodermal cells were capable of generating cartilage-containing particles in vitro.
TGFβ-dependent and TGFβ-independent chondrogenic activities
Without TGFβ3, the efficiency of the initial aggregate formation was significantly reduced (data not shown), and the size of the particles formed became smaller (Fig. 2A,C,E). Furthermore, in the case of flk-1+PDGFRα–cells and DP cells, no area was stained metachromatically with Toluidine blue within the particles (Fig. 2C,E). In contrast, in the case of the flk-1–PDGFRα+ cells, seven out of 10 small particles formed contained one to two small cartilaginous nodules that stained strongly with Toluidine blue (Fig. 2A and Table 2).
To quantify the TGFβ-independent cartilage-forming activity, three chondrogenic EB cells, flk-1–PDGFRα+ cells,flk-1+PDGFRα– cells and DP cells, were individually subjected to a serum-free micromass culture. In the presence of 50 ng/ml PDGF-BB, the cells survived or slowly proliferated during the first 8 days, as observed with the somite cells(Tallquist et al., 2000). Consistent with the results shown in Fig. 2, the addition of TGFβ3 stimulated the accumulation of sulfated glycosaminoglycan in all three cultures(Fig. 4A). Furthermore, the addition of TGFβ3 induced COL2 accumulation during the culture of the flk-1+PDGFRα– cells and the DP cells(Fig. 4Ca). The flk-1–PDGFRα+ cell culture that accumulated low levels of COL2 without TGFβ3 also displayed an approximately three-fold increase in the COL2 level with TGFβ3.
Without TGFβ, however, approximately 79 small cartilage nodules that stained positively with Alcian blue were detected per 2×105flk-1–PDGFRα+ cells initially seeded(Fig. 4B). Smaller cartilage nodules were detected in the DP cell fraction at a lower frequency(approximately 14 per 2×105 cells). No cartilage nodule was found with the flk-1+PDGFRα– cells(Fig. 4B). Therefore, in contrast to the lymphohematopoietic potential, the TGFβ-independent chondrogenic activity was highly enriched in the flk-1–PDGFRα+ EB cell fraction and was absent from the flk-1+PDGFRα– cell fraction.
Thus, there seem to be two modes of chondrogenesis. In one type, exogenous TGFβ is absolutely essential (for flk-1+PDGFRα– as well as DP mesodermal cells) and in the other, chondrogenesis occurs as small, distinct nodules without the exogenous TGFβ, and the addition of TGFβ expanded the chondrogenic area (for flk-1–PDGFRα+cells).
PDGF-BB augments TGFβ-induced chondrogenesis
TGFβ alone was not sufficient to form a particle filled with the cartilage matrix macromolecules (Figs 2, 3). Since PDGF-BB sustained the viability of the ES-cell-derived mesodermal cells during the serum-free micromass culture, we further investigated the effect of PDGF. The addition of 20 to 50 ng/ml PDGF-BB significantly elevated the COL2 accumulation during the micromass culture of all three cell types tested in the presence of TGFβ3(Fig. 4Cb). Twenty to 100 ng/ml of PDGF-AA did not significantly elevate the COL2 level.
On the other hand, the addition of 50 ng/ml PDGF-BB to the pellet culture of flk-1–PDGFRα+ cells(Fig. 5A),flk-1+PDGFRα– cells(Fig. 5B) or DP cells (data not shown) in the presence of TGFβ3 markedly enlarged the particle volume as well as the areas staining with the anti-COL2 antibody and metachromatically with Toluidine blue, so that, except for a thin surface cell layer, the cartilage matrix was distributed throughout the particle. We refer to such particles as `full' cartilage in this report. The COL2+ interior of the large `full' cartilage particle formed in the presence of TGFβ3 and PDGF-BB (Fig. 5Ad-f, 5Bj-l) was weakly stained with the anti-COL1 antibody. However, a layer of cells on the surface of the particle showed relatively high levels of COL1, and COL2 was also detected at a level similar to that in the interior COL2+area. Thus, even in the presence of PDGF-BB, the `full' cartilage was still fibro cartilage.
In contrast, PDGF-BB treatment alone resulted in small particles with either flk-1+PDGFRα– cells(Fig. 5Bg-i) or DP cells (data not shown) that were devoid of cartilage matrix. The TGFβ-independent chondrogenesis of the flk-1–PDGFRα+ cells was still observed in the presence of PDGF-BB alone(Fig. 5Ba-c). However, a large`full' cartilage particle was formed only with TGFβ3(Table 2).
Thus, PDGF-BB seems to be a synergistic factor for the TGFβ-induced chondrogenesis of the ES-cell-derived mesodermal cells, leading to `full'cartilage particle formation. Further analyses were performed, mostly with two interesting cell types that also efficiently formed `full' cartilage(Table 2): the hemo-chondrogenic flk-1+PDGFRα– cells (75%)and the chondrogenic flk-1–PDGFRα+ cells(83%).
Inhibitory effect of noggin
The BMP family is also implicated in chondrogenesis(Cancedda et al., 1995). To determine whether TGFβ-induced or TGFβ+PDGF-induced chondrogenesis involved BMP activity, we added a pan-BMP-binding inhibitor, noggin, to the pellet culture. Noggin interacts with BMP2, BMP4 and BMP7, but does not bind to Activin, TGFβ1 (Zimmerman et al.,1996) and TGFβ3 (data not shown).
When added at the beginning of the culture, noggin-Fc inhibited the TGFβ-induced cartilage formation in a dose-dependent manner(Fig. 6; Table 2). One μg/ml of noggin-Fc was sufficient for complete inhibition of the formation of areas stained metachromatically with Toluidine blue, and 0.1 μg/ml produced weaker inhibitory effects (Fig. 6A). A similar inhibitory effect on the TGFβ+PDGF-induced`full' cartilage formation was observed(Fig. 6B), in that the area immunostained with the anti-COL2 antibody was also abolished. Moreover, the noggin-Fc at 1 μg/ml significantly retarded the growth of particles formed either with TGFβ3 alone or with TGFβ3 and PDGF-BB, resulting in a small particle. At 0.1 μg/ml, no effect on the particular volume was observed.
When TGFβ3 was removed on days 10 to 12, which otherwise would have resulted in a cartilage-nodule-containing particle, the addition of noggin-Fc at 1 μg/ml at this point prevented the formation of cartilaginous areas(0-20% of particles contained areas showing metachromatic staining) during the following 7 to 8 days (Fig. 7C; Table 3). An inhibitory effect on the particle growth was also noted.
Factor* added . | flk-1+PDGFRα- . | flk-1+PDGFRα+ (DP) . | flk-1-PDGFRα+ . |
---|---|---|---|
TGFβ3† | 26/30(0)‡ | 11/17 (0) | 44/46 (0) |
TGFβ3, then none | 12/17 (4) | 5/6 (0) | 19/19 (5) |
TGFβ3, then BMP4 | 4/4 (3) | 2/2 (2) | 6/6 (6) |
TGFβ3, then TGFβ3+BMP4 | ND§ | ND | 3/3 (3) |
TGFβ3, then noggin | 0/3 | 0/2 | 1/5 (0) |
Factor* added . | flk-1+PDGFRα- . | flk-1+PDGFRα+ (DP) . | flk-1-PDGFRα+ . |
---|---|---|---|
TGFβ3† | 26/30(0)‡ | 11/17 (0) | 44/46 (0) |
TGFβ3, then none | 12/17 (4) | 5/6 (0) | 19/19 (5) |
TGFβ3, then BMP4 | 4/4 (3) | 2/2 (2) | 6/6 (6) |
TGFβ3, then TGFβ3+BMP4 | ND§ | ND | 3/3 (3) |
TGFβ3, then noggin | 0/3 | 0/2 | 1/5 (0) |
Particles including areas morphologically resembling cartilage were scored,and are indicated per total number of particles examined, as in Table 2. This series of experiments involved a medium switch on days 10 to 12 of culture, in that the serum-free chondrogenesis medium with TGFβ3 was changed to medium alone,with BMP4, with TGFβ3+BMP4, or with noggin, and the culture was continued for another 7 to 8 days.
TGFβ3 10 ng/ml, BMP4 50 ng/ml, and noggin (i.e. noggin-Fc) 1μg/ml.
This is a control for no medium change. The same data are shown in Table 2.
Definitions of numbers in parentheses are the same as those in Table 2.
Not determined.
Thus, the BMP family of proteins seems to be involved in the chondrogenesis of the ES-cell-derived mesodermal cells, probably downstream of TGFβ or TGFβ+PDGF.
Synergistic enhancement of the TGFβ-induced cartilage formation by BMP4
The roles of BMP were further investigated. First, 50 ng/ml BMP4, added in place of TGFβ3 at the beginning of the culture, resulted in a small particle without cartilage nodules, although it supported the initial aggregate formation. Rather, BMP4 prevented the TGFβ-independent formation of the small cartilage nodules in the flk-1–PDGFRα+-cell-derived particle(Fig. 7Ab; Table 2).
By contrast, the addition of 20 to 50 ng/ml BMP4 to the TGFβ-containing culture markedly enhanced the depositions of cartilage matrix (Fig. 7B) and COL2 (data not shown), leading to `full' cartilage formation. The enhancement was dose dependent, and 5 ng/ml of BMP4 were not effective(Fig. 7Bb). The frequency of`full' cartilage formation reached 100% in the presence of TGFβ3 and 50 ng/ml BMP4 (Table 2). A similar synergistic effect was observed even when TGFβ3 was replaced with 50 ng/ml BMP4 after the cells were treated with TGFβ3 for 10 to 12 days(Fig. 7C). This resulted in a large `full' cartilage particle filled with a cartilage matrix and COL2 at a frequency of 75-100% (Table 3)in the following 7 to 8 days of culture(Fig. 7B).
Thus, the exogenously added BMP4 synergistically enhanced the TGFβ-induced cartilage formation from the ES-cell-derived mesodermal cells. However, TGFβ treatment seemed to be pre-requisite for BMP4 to stimulate chondrogenesis.
Potential inhibitory effects of TGFβ on chondrocyte maturation
The cartilage nodules formed in the presence of TGFβ3 stained fainter with Toluidine blue than those formed in the absence of it(Fig. 2A,B). This raised the possibility that TGFβ suppresses chondrocyte maturation at a later stage of cartilage formation, as is the case for the embryonic limb mesenchymes. Therefore, TGFβ3 was removed on days 10 to 12 of culture, and the resulting particles were examined on days 17 to 20(Fig. 7C). Under these conditions, the flk-1–PDGFRα+ cells gave rise to particles containing a larger cartilaginous area that consisted of larger chondrocytes and displayed stronger metachromatic staining with Toluidine blue. The `full' cartilage was also obtained at a frequency of 26%(Table 3). The TGFβ+PDGF-induced chondrogenic culture gave rise to a `full' cartilage particle in 17 to 20 days (Fig. 5). When both TGFβ3 and PDGF-BB were removed on days 10 to 12, no increase in either the COL2 level or the strength of Toluidine blue staining was noted on days 17 to 20. However, an increase in the size of the chondrocytes and a reduction in the COL1 level in the cartilaginous area were apparent (Fig. 8A).
These observations suggest that, in a later phase of the chondrogenic culture of the ES-cell-derived mesodermal cells, the continuous presence of TGFβ3 prevented or retarded the chondrocyte maturation and, even in the presence of PDGF-BB, maintained the COL1 expression within the cartilage matrix, thereby preventing the hyaline cartilage formation.
Hyaline cartilage formation and induction of mineralization
When TGFβ3 and PDGF-BB were removed, a distinct, thin layer of cells that strongly stained with the anti-COL1 antibody, but did not stain with either Toluidine blue or the anti-COL2 antibody, became apparent on the surface of a particle (Fig. 7Cc, Fig. 8Ad-f). It was composed of spindle-shape cells resembling undifferentiated mesenchymal cells. Upon the addition of 50 ng/ml BMP4, the thin cell layer became a thicker layer containing round chondrocytic cells and a matrix that was rich in COL2 and stained lightly with Toluidine blue(Fig. 7Cd, Fig. 8Ag,i). Moreover, BMP treatment resulted in a reduced level of COL1 in the surface cell layer as well as in the interior cartilaginous area(Fig. 8Ah), resulting in a hyaline cartilage particle. A similar thick layer that stained lightly with Toluidine blue was evident when the cells were cultured from the beginning with TGFβ3 and 20-50 ng/ml BMP4 (Fig. 7Bc,d,f).
The hyaline cartilage particles formed by culturing the flk-1–PDGFRα+ cells or the flk-1+PDGFRα– cells with TGFβ3 and PDGF-BB for 10 days, and with BMP4 alone for the following 6 days, were further tested for mineralization. In the presence of T3 and b-glycerophosphate an area in which the intercellular space was stained with von Kossa became apparent within the particle after 5 days of culture,indicating that mineral deposits in the cartilage matrix were induced(Fig. 8B). The anti-COL10 antibody detected COL10 deposition as a larger area, in which the mineralizing matrix was included and a lower level of COL2 was accumulated.
These observations indicate that the ES-cell-derived mesodermal cells are able to form a hyaline cartilage particle and that the hyaline cartilage generated with the ES-cell-derived mesodermal cells mature and mineralize the matrix. They also suggest that the COL1+ mesenchymal cells in the surface layer could be BMP-sensitive chondroprogenitors that serve as the source of chondrocytes during the growth of the cartilage particle, which is equivalent to the perichondrium.
Discussion
We have demonstrated that the ES-cell-derived mesodermal cells are chondrogenic and that TGFβ and PDGF synergistically function in in vitro development of cartilage through BMP activity. As far as we know, this is the first report demonstrating the full chondrogenic activity of ES-cell-derived mesodermal cells in vitro.
In vitro chondrogenesis of the ES-cell-derived mesodermal cells as a faithful culture model for chondrogenesis
First, to address whether chondrogenesis was achieved through a developmental pathway similar to that in vivo, we examined the roles of three factors that are known to be involved in embryonic chondrogenesis: TGFβ,PDGF and BMP.
TGFβ and BMP are implicated in several steps during limb cartilage development, and their effects are both positive and negative, depending on the developmental stage. For example, TGFβ has an early positive function and a late negative function. Our pellet culture of either flk-1+PDGFRα– or flk-1–PDGFRα+ EB cells produced similar TGFβ effects: (1) supporting aggregate growth and formation of cartilaginous areas within the particle generated (Figs 2, 5) and (2) inhibiting of the generation of hypertrophic chondrocytes and the formation of hyaline cartilage, both of which were induced by removing TGFβ later(Fig. 7C, Fig. 8A). However, the detection of the COL10 transcript by PCR suggested that some degree of hypertrophic differentiation might occur in the presence of TGFβ(Fig. 3). BMP is also involved in multiple steps of chondrogenesis. We provided evidence that (1) BMP is involved in TGFβ-induced as well as TGFβ+PDGF-induced chondrogenesis(Fig. 6, Fig. 7C), (2) exogenous BMP enhanced the cartilage matrix deposition in the TGFβ-treated particles,leading to a dramatic increase in the volume of the cartilage particles(Fig. 7), and (3) BMP suppressed the COL1 level and elevated COL2 expression, resulting in a hyaline cartilage particle (Fig. 8A),thus supporting the positive roles of BMP. It is worth mentioning that no significant changes in mineral deposition within such hyaline cartilage particles were observed after 5 days of hypertrophic differentiation induced with or without 100 ng/ml BMP6, although COL10 accumulation was enhanced by BMP6 (data not shown).
Despite the positive effects of BMP4 at a late stage of chondrogenesis culture, BMP4 added alone from the beginning was rather detrimental(Fig. 7A), which implies that chondrogenesis of the ES-cell-derived mesodermal cells occurs through two steps requiring different proteins: TGFβ followed by BMP. This two-step model agrees with the previous notion that, during the micromass culture of chicken limb mesenchymal cells, a stimulatory effect of BMP4 is observed 24-48 hours after culture, whereas that of TGFβ3 was readily detectable from the beginning of the culture (Roark and Greer, 1994).
PDGF-BB also has striking synergy with TGFβ and forms a large `full'cartilage particle (Fig. 5)through BMP action (Fig. 6B). Therefore, the role of PDGF may also be placed upstream of that of BMP. The direct effect of PDGF on chondrogenesis has been demonstrated with primary mouse somitic cells (Tallquist et al.,2000). PDGF-BB, and to a lesser degree PDGF-AA, increased the accumulation of Alcian-blue-positive cartilage matrix in the absence of TGFβ without stimulating cell proliferation. Enhancement of COL2 accumulation by PDGF-BB but not by PDGF-AA(Fig. 4C) is consistent with this observation, although our experiment was done in the presence of TGFβ3.
Thus, as judged by the effects of TGFβ, BMP and PDGF, the serum-free chondrogenesis culture using isolated ES-cell-derived mesodermal cells seems to reproduce some, if not all, aspects of the in vivo process.
The ES-cell-derived PDGFRα+ mesodermal cells are chondrogenic but not hemogenic
The TGFβ-independent chondrogenic activity was highly enriched in the flk-1–PDGFRα+ EB cell fraction(Fig. 4), although no lymphohematopoietic cell potential was detected(Fig. 1), suggesting that the flk-1–PDGFRα+ cells are committed to non-hematopoietic cell lineages, including the chondrocyte lineage. This is consistent with the notion that the PDGFRα signaling is essential for normal axial skeletogenesis (Soriano,1997) and that PDGFRα is highly expressed in the somites and later in the sclerotomal mesenchymes, the condensing limb mesenchymes and the perichondrium, which are considered to be non-hematopoietic, chondrogenic cells (Orr-Urtreger et al.,1992; Schatteman et al.,1992).
The biological significance of the two types of chondrogenic activities,the TGFβ-independent activity and the TGF-dependent activity, is not clear. The flk-1–PDGFRα+ EB cell fraction may contain TGFβ-producing cells or mature chondroprogenitors that no longer require TGFβ to form cartilage nodules. Candidates for the latter are the PDGFRα+ sclerotomal mesenchymes and the limb mesenchymes. Nevertheless, we have concluded that the PDGFRα+EB cell fraction contains chondrogenic mesoderm.
The ES-cell-derived flk-1+ mesodermal cells containing the hemangioblasts are also chondrogenic
The ES-cell-derived flk-1+ hemangioblast is characterized by its ability to form a blast cell colony in the presence of VEGF and to generate hematopoietic CFCs (Faloon et al.,2000; Kabrun et al.,1997; Nishikawa et al.,1998a). In this respect,flk-1+PDGFRα– is likely to be the hemangioblast fraction, because it was enriched for CFC-generating activity(Fig. 1B), whereas no spontaneous chondrogenic activity was detected in it (Figs 2, 4, 5). The TGFβ-dependent chondrogenic activity of the flk-1+PDGFRα–cells thus implies that the hemangioblast may be multi-potential at this stage of development. The observations that the ES-cell-derived flk-1+cells produce vascular smooth muscle-like cells in the presence of PDGF-BB(Yamashita et al., 2000), and that a purer preparation of hemangioblasts(flk-1+VE-cadherin+PDGFRα– cells)was also chondrogenic in the presence of TGFβ3 and BMP4 (data not shown),supported this possibility.
Another possibility is that the flk-1+PDGFRα– cell fraction is a heterogeneous mixture of cell types that are differentially committed to either the hematopoietic/endothelial cell lineages or the chondrogenic cell lineage. However, the result that flk-1–PDGFRα+ cells, DP cells and flk-1+PDGFRα– cells accumulated similar levels of COL2 during 8 days of micromass culture in the presence of TGFβ3 and PDGF-BB (Fig. 4C) would argue against the possibility that cross contamination by neighboring cells, such as DP cells, accounts for the TGFβ-dependent cartilage forming activity of the flk-1+PDGFRα– cells.
An appropriate expansion culture method may have to be established to answer these questions definitively and is currently under way.
In vitro chondrogenesis in the whole EB and with the ES-cell-derived mesodermal cells
Kramer et al. demonstrated that chondrogenesis can be performed directly within an EB (Kramer et al.,2000). The culture medium contained 15% FCS, unlike ours, and a high concentration of BMP (100 ng/ml) stimulated chondrogenesis when added within the first 5 days of culture. In contrast, TGFβ displayed either no effect or an inhibitory effect. These results are inconsistent with our observation that chondrogenesis with the FACS-purified mesodermal cells from day 3.6-4.6 EBs either required TGFβ or was augmented by TGFβ(Fig. 2) and that the effect of BMP4 (50 ng/ml) was observed even after 10 days of the pellet culture(Fig. 7C; Table 3). The discrepancy can be explained by the serum effect as well as by the difference in the cell populations used. For example, the requirement of TGFβ in our culture condition may be caused by the removal of TGFβ-producing cells and/or a TGFβ-inducing factor, FCS (Mummery and van den Eijnden-van Raaij, 1999). Nevertheless, as judged by the temporal order of the factor requirements, the serum-free chondrogenesis culture using isolated ES-cell-derived mesodermal cells seems to represent the in vivo process better. Furthermore, considering the heterogeneity in the DN cell population (Fig. 2G),macroscopic cartilage containing a homogeneous population of chondrocytes would be formed easier with isolated mesodermal cells.
By taking advantage of the ease of obtaining large numbers of purified mesodermal cells, the ES cell differentiation culture described here would provide an important research platform for molecular and cellular dissections of the chondrogenesis pathway in the context of macroscopic cartilage. This method may lead to a better strategy for cartilage repair and cartilage engineering. Applications to the human ES cells may provide an important tissue source for cartilage replacement/repair therapy.
Acknowledgements
We would like to acknowledge D. Chang for the biotinylation of APA5. We would also like to give warm thanks to W. Boyle and S. Simonet for support.