1. Top of page
  2. Abstract
  6. Acknowledgements


To study the changes in patterns of gene expression exhibited by human chondrocytes as they dedifferentiate into fibroblastic cells in culture in order to better understand the mechanisms that control this process and its relationship to the phenotypic changes that occur in chondrocytes during the development of osteoarthritis (OA).


Human fetal epiphyseal chondrocytes (HFCs) were cultured either on poly-(2-hydroxyethyl methacrylate)–coated plates (differentiated HFC cultures) or in plastic tissue culture flasks as monolayers (dedifferentiated HFC cultures). Following 11 days of culture under either condition, poly(A+) RNA was isolated from the two cell populations and subjected to a gene expression analysis using a microarray containing ∼5,000 known human genes and ∼3,000 expressed sequence tags (ESTs).


A ≥2-fold difference in the expression of 62 known genes and 6 ESTs was observed between the two cell types. The differences in expression of several of the genes detected by the microarray hybridization were confirmed by Northern analyses. Two transcription factor genes, TWIST and HIF-1α, and a cellular adhesion protein gene, cadherin 11, were markedly regulated in response to differentiation and dedifferentiation. Expression of these genes was also detected in adult normal and OA cartilage and chondrocytes. Analysis of the gene expression profile of HFCs revealed a complex pattern of gene expression, including many genes not yet reported to be expressed by chondrocytes.


Chondrocytes in monolayer become dedifferentiated, acquiring a fibroblast-like appearance and changing their pattern of gene expression from one of expression of chondrocyte-specific genes to one that resembles a fibroblastic or chondroprogenitor-like pattern. Changes in gene expression associated with the process of dedifferentiation of HFCs in vitro were observed in a wide variety of genes, including genes encoding extracellular matrix proteins, transcription factors, and growth factors. At least 3 of the genes that were regulated in response to dedifferentiation were also found to be expressed in adult normal and OA articular cartilage and chondrocytes.

Articular cartilage (AC) chondrocytes are the highly specialized cells responsible for the production and maintenance of the integrity of the cartilage extracellular matrix (ECM). The ECM of AC imparts the tissue with its unique properties of almost frictionless motion under high load conditions (1). To produce and maintain a properly functional cartilage matrix, the chondrocyte displays a specific pattern of gene expression both during development and in the adult (2, 3). Injured or diseased chondrocytes alter their pattern of gene expression in response to changes in their surrounding matrix, in the mechanical properties of the tissue, and in response to various growth factors, cytokines, and inflammatory mediators (4, 5). The chondrocyte response may lead, in certain situations, to long-term changes in the phenotype of the cell, and therefore to an inability to properly repair or maintain the cartilage ECM (4). This is exemplified by the phenotypic changes in the patterns of production or in the temporal or spatial distribution of the synthesis of interstitial collagens, fibroblast-type proteoglycans, and production of ECM proteins associated with fetal development that occur during the development of osteoarthritis (OA), dedifferentiation in culture, or in response to certain types of cartilage injury (6–13).

Normal chondrocytes exhibit a remarkable range of phenotypic plasticity during development, in adult tissues, and in culture. For example, chondrocytes within the different zones of adult AC display pronounced morphologic and phenotypic differences (1). Chondrocytes in the superficial layers of AC have a flattened, fibroblast-type appearance, while those in the deeper zones are more spherical. Chondrocytes from the different zones of AC isolated and cultured separately retain these differences, in that they show differences in proliferation rates and proteoglycan and collagen synthesis and in their response to cytokines such as interleukin-1 (IL-1).

Generally, chondrocytes from the deeper zones show higher rates of proliferation and collagen and proteoglycan synthesis compared with chondrocytes isolated from the superficial and middle zones (14, 15). Furthermore, the proteoglycans synthesized by the chondrocytes from the deeper zones have a higher keratan sulfate content, while those from the superficial zone are more sensitive to the catabolic effects of IL-1 (16). During development, mammalian growth plate chondrocytes exhibit several distinct phenotypes, such as those of resting, proliferating, and hypertrophic chondrocytes. Progression through each of these phases is accompanied by profound changes in gene expression patterns (3). Thus, the maintenance of the chondrocyte-specific phenotype is of crucial importance to the preservation of the normal structure and biomechanical properties of AC and during repair of injured and diseased tissue. In addition, the ability of the chondrocyte to exhibit a high degree of phenotypic plasticity is also crucial for the proper development and functioning of cartilaginous tissues.

Numerous studies have shown that chondrocytes propagated under culture conditions that allow them to attach and spread on a 2-dimensional surface, such as on tissue culture plastic, undergo a phenotypic change both in morphology and in gene expression pattern. Studies that have examined the plasticity of the chondrocyte phenotype from many different species have consistently shown that culture of these cells in monolayers on plastic substrata for prolonged periods or upon repeated passages leads to the loss of their spherical shape and to the acquisition of an elongated fibroblast-like morphology (7, 17–28). These morphologic alterations are accompanied by profound biochemical changes, including loss of the cartilage-specific phenotype, as evidenced by an arrest of the synthesis of the cartilage-specific collagens (types II, IX, and XI) and proteoglycans (aggrecan), initiation of synthesis of the interstitial collagens (types I, III, and V), and increase in the synthesis of fibroblast-type proteoglycans (versican) at the expense of aggrecan (7, 17, 19–29).

The chondrocyte phenotype can be reexpressed in these cells by culturing them in suspension, in agarose, with alginate beads, or on a hydrogel substrate (17, 19, 25, 28, 30–36). These changes in the biosynthetic profile of dedifferentiated chondrocytes resemble some of the phenotypic changes displayed by OA chondrocytes, and the matrix they produce is similar to that synthesized by chondroprogenitor cells (7–13, 17, 37–40). However, the mechanisms responsible for the phenotypic instability of chondrocytes and the regulation of these different chondrocyte-specific phenotypes remain poorly understood.

To better understand the mechanisms that underlie the phenotypic instability of chondrocytes, we performed a microarray gene expression analysis of human fetal chondrocytes (HFCs) cultured either under conditions that allow them to preserve their differentiated phenotype or under conditions that lead to their dedifferentiation. These studies showed remarkable differences in the patterns of gene expression between cells cultured under the two conditions. The results we obtained have important implications for interpretation of studies conducted with chondrocytes cultured as monolayers on plastic as well as for the use of such chondrocytes in cell therapies for correction of cartilage defects or for the treatment of OA.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Isolation of human and bovine fetal chondrocytes. HFCs were obtained from epiphyseal cartilage removed under sterile conditions from femoral heads, knee condyles, and tibial plateaus of fetuses ages 20–24 weeks (Anatomical Gift Foundation, Laurel, MD). The fetal tissues were obtained by the Anatomical Gift Foundation with previous informed consent following protocols reviewed and approved by the National Institutes of Health and in accordance with the National Organ Transplant Act. Bovine fetal chondrocytes were isolated from epiphyseal cartilage (knee joint) obtained from late–second- trimester fetuses (Animal Technologies, Tyler, TX). The chondrocytes were isolated according to the method of Reginato et al (36). Briefly, to remove adherent fibrous tissues, the cartilage was incubated in Hanks' medium containing trypsin and bacterial collagenase (2 mg/ml each) for 1 hour at 37°C. The medium was discarded and the tissue fragments minced and digested overnight at 37°C in Dulbecco's minimum essential medium (DMEM) with 4.5 gm/liter glucose containing 10% fetal bovine serum (FBS) and 0.5 mg/ml bacterial collagenase. The released cells were filtered through a 70μ nylon membrane into a vessel containing fresh DMEM and 10% FBS. The cells were collected by centrifugation at 250g for 5 minutes and washed 4 times with collagenase-free medium. The average yield of chondrocytes obtained with this procedure is 3.0 ± 0.4 × 108/gm wet weight cartilage.

Culture of chondrocytes. HFCs from 3 separate specimens were isolated as described above. For each experiment, one-half of the cells were plated in 60-mm petri dishes that had been precoated with 0.9 ml of a 10% solution of poly-(2-hydroxyethyl methacrylate) (poly-HEMA; Polysciences, Malvern, PA) as previously described (36). Chondrocytes cultured under these conditions retain their differentiated phenotype for at least 180 days in culture (36). In order to induce dedifferentiation, the remaining one-half of the chondrocytes from each sample were plated into 162-cm2 tissue culture flasks at a density of 30,000/cm2. This resulted in 3 sets each of differentiated (poly-HEMA) and dedifferentiated (plastic) HFC cultures. All cultures were grown in DMEM containing 10% FBS, 2 mM glutamine, 1% vitamin supplements, 40 μg/ml ascorbic acid, 100 units/ml penicillin, 100 μg/ml streptomycin, and 2.5 μg/ml amphotericin B.

Microarray hybridization. The three sets of HFCs were cultured under the two conditions described above for 11 days, at which time they were separately processed for poly(A+) RNA isolation. Total RNA was first isolated by the TRIzol method (Gibco BRL, Rockville, MD), and then poly(A+) RNA was obtained by passing the total RNA twice through oligo(dT) columns (Oligotex columns; Qiagen, Valencia, CA). The average yields of poly(A+) RNA were 0.56 μg/100 × 106 differentiated chondrocytes and 3 μg/100 × 106 dedifferentiated chondrocytes. Seven hundred fifty nanograms of poly(A+) RNA obtained from one set of differentiated HFCs and 750 ng of poly(A+) RNA obtained from the corresponding set of dedifferentiated HFCs was employed for complementary DNA (cDNA) synthesis and hybridization to the human UniGEM V microarray, which contains ∼5,000 known human genes and ∼3,000 expressed sequence tags (ESTs) (Genome Systems, St. Louis, MO). A ≥2-fold difference in expression levels was deemed significant and reproducible by the manufacturer following extensive testing of their procedure prior to its availability.

Northern hybridization analysis. Human cDNA for insulin-like growth factor 2 (IGF-2), IGF binding protein 4 (IGFBP-4), bone morphogenetic protein 6 (BMP-6), hypoxia-inducible factor 1α (HIF-1α), and the retinoic acid receptor RXR-β for use as probes in Northern analyses were obtained by reverse transcriptase–polymerase chain reaction (RT-PCR) using the TITAN RT-PCR kit (Roche Molecular Systems, Alameda, CA), human chondrocyte poly(A+) RNA, and the primers shown in Table 1. A human cDNA clone for the basic helix-loop-helix (bHLH) transcription factor Twist was obtained from Genome Systems. All cDNA clones and RT-PCR products were sequenced with an ABI automatic sequencer (Perkin-Elmer, Norwalk, CT) for verification of their authenticity.

Table 1. Primers used to obtain cDNA probes by reverse transcriptase–polymerase chain reaction (PCR) for Northern analyses*
GenePrimers (5′[RIGHTWARDS ARROW]3′)Size (position), bp
  • *

    IGF-2 = insulin-like growth factor 2; IGFBP-4 = IGF binding protein 4; BMP-6 = bone morphogenetic protein 6; HIF-1α = hypoxia-inducible factor 1α.

  • Size of the PCR product and its position in the corresponding mRNA.


For Northern blot analyses, 100–200 ng poly(A+) RNA from the differentiated and dedifferentiated HFCs was electrophoresed on 1% agarose–formaldehyde gels and transferred to Hybond-N+ or XL membranes (Amersham Pharmacia Biotech, Piscataway, NJ) as described previously (36). The blots were hybridized in 5× saline–sodium citrate (SSC), 2× Denhardt's solution, 50% formamide, 0.1–0.5% sodium dodecyl sulfate (SDS), and 0.1 mg/ml denatured salmon sperm DNA at 42°C for 18 hours with 32P-labeled cDNA probes. The blots were washed to a stringency of 0.2× SSC/0.1% SDS at 65°C for 30 minutes and then exposed to x-ray film or to a phosphorimaging cassette. The hybridization signals were normalized to those of GAPDH, and the fold (magnitude) difference in expression levels was determined by using a Storm phosphorimager and the ImageQuant analysis software (Molecular Dynamics, Sunnyvale, CA).

RT-PCR analysis. The RT-PCR analysis of RNA from human OA cartilage was carried out as described in Hu et al (41), utilizing the RNA described therein. First-strand cDNA was synthesized with an oligo(dT) primer and Moloney murine leukemia virus RT. AmpliTaq Gold Polymerase (Perkin-Elmer) was used for PCR, and the PCR conditions were 32 cycles of 94°C for 30 seconds, 63°C for 30 seconds, and 72°C for 2 minutes. PCR products were electrophoresed on 10% polyacrylamide gels and stained with SYBR Green I (Molecular Probes, Eugene, OR).

For the RT-PCR analysis of TWIST, total RNA was either isolated from chondrocytes that had been liberated from cartilage slices obtained from patients undergoing knee replacement surgery as a result of OA, or from the knee of a patient undergoing above-the-knee amputation for peripheral vascular disease (normal). To isolate the RNA, the remaining cartilage was excised from the femoral condyles and tibial plateaus with a scalpel, and the chondrocytes were liberated from the surrounding matrix by an overnight digestion with collagenase as described above. RNA was isolated from the freed chondrocytes by using the TRIzol reagent (Gibco BRL) following the manufacturer's specifications. First-strand cDNA was prepared from 1–5 μg of total RNA with Superscript RT (Gibco BRL) and oligo(dT) primer in a 20-μl reaction. PCR was carried out in a volume of 50 μl with Taq polymerase and 2–5 μl of the first-strand cDNA reaction mixture with the primers shown in Table 2. The conditions for the PCR reaction were 35 cycles of 94°C for 1 minute, 58°C for 45 seconds, and 72°C for 45 seconds.

Table 2. Primers for reverse transcriptase–PCR analysis of RNA isolated from adult human osteoarthritic and normal articular cartilage and chondrocytes*
GenePrimers (5′[RIGHTWARDS ARROW]3′)Size (position), bp
  • *

    See Table 1 for definitions.

  • Size of the PCR product and its position in the corresponding mRNA.

  • Size is 933 bp with intron.

Cadherin 11Forward: CAACAGTTCTGAGCTGTAATTTCGCC246(3427–3672)

Preparation of nuclear extracts, electrophoretic mobility shift assays (EMSAs), and Western analysis. Nuclear extracts were prepared from freshly isolated bovine fetal chondrocytes according to the method of Dignam et al (42). Briefly, cells were lysed by Dounce homogenization in hypotonic buffer (10 mM HEPES [pH 7.9], 1.5 mM MgCl2, 10 mM KCl, and 0.5 mM dithiothreitol [DTT]). Nuclei were recovered by centrifugation at 3,300g for 15 minutes at 4°C and extracted in buffer C (20 mM HEPES [pH 7.9], 0.42M NaCl, 25% glycerol, 1.5 mM MgCl2, 0.2 mM EDTA, and 0.5 mM DTT) for 30 minutes at 4°C by gentle shaking. The extract was then centrifuged for 30 minutes at 25,000g, and the supernatant was then frozen at –70°C. All buffers also contained a protease inhibitor cocktail (2 mM 4-(2-aminoethyl)benzenesulfonylfluoride, 1.4 μM trans-epoxysuccinyl-l-leucylamido[4-guanidino]butane, 130 μM bestatin, 1 μM leupeptin, and 0.3 μM aprotinin; Sigma, St. Louis, MO).

EMSAs were carried out as follows. Binding reactions consisted of 12.5 mM HEPES (pH 7.9), 50–100 mM NaCl, 5% glycerol, 2 mg/ml bovine serum albumin, 1–2 μg poly(dI-dC), 0.1 mM EDTA, 0.1 mM DTT, 1 ng 32P–end-labeled double-stranded oligonucleotide probe, and 10 μg nuclear protein. The probes consisted of wild-type and mutated HIF-1 binding sites from the promoter region of the human VEGF gene (wild type ACAGTGCATACGTGGGCTCCAAC; mutated ACAGTGCACGATGTGGCTCCAAC) (differences in sequences are in boldface type) (43). Binding reaction mixtures were incubated for 50 minutes at 4°C and then loaded onto 4% acrylamide/0.25× Tris–borate–EDTA gels and electrophoresed at 200V for 1.5 hours.

For Western analysis, 50 μg of bovine fetal chondrocyte nuclear extract was separated by SDS–polyacrylamide gel electrophoresis (44) and transferred to a Hybond-P membrane (Amersham Pharmacia Biotech) in 20 mM Tris, 150 mM glycine, and 20% methanol at 40V for 18 hours at 4°C. The membrane was blocked in phosphate buffered saline (PBS) containing 10% nonfat milk, 0.1% normal goat serum, and 0.2% Tween 20. The primary antibody was a mouse monoclonal anti–HIF-1α antibody (1/250; Novus Biologicals, Littleton, CO) and the secondary antibody was a sheep anti-mouse IgG–horseradish peroxidase conjugate (1/1,000; Novus Biologicals). The membrane was washed 4 times for 7 minutes each between antibody incubations in PBS plus 0.2% Tween 20. The blot was developed using the ECL detection kit (Amersham Pharmacia Biotech).


  1. Top of page
  2. Abstract
  6. Acknowledgements

Assessment of the differentiated and dedifferentiated phenotype by cellular morphology and COL2A1 and COL1A1 messenger RNA (mRNA) expression. To examine the gene expression profile of differentiated versus dedifferentiated chondrocytes in vitro, we utilized the poly-HEMA culture system to preserve the chondrocyte-specific phenotype and culture as monolayers on tissue culture plastic to induce the loss of this phenotype. In other investigations, our group has documented that the differentiated chondrocyte phenotype can be maintained on poly-HEMA for at least 180 days (36). Because we were interested in the detection of early changes in gene expression that occur during the dedifferentiation process and in the expression of gene regulatory factors that may control the chondrocyte-specific phenotype, we chose to isolate mRNA from the cell populations after only 11 days of culture either on poly-HEMA or as primary monolayer cultures on plastic.

Figure 1 shows photomicrographs representative of the morphology of chondrocytes cultured on poly-HEMA or on tissue culture plastic for 11 days. To evaluate the extent of dedifferentiation that occurred as a result of the two culture conditions, we performed Northern analyses with a COL2A1 cDNA fragment corresponding to the COOH-terminal and 3′-untranslated regions and with a COL1A1 cDNA probe also corresponding to the COOH-terminal region. Figure 2 shows the results with the mRNA isolated from the differentiated and dedifferentiated cultures that were utilized for the microarray hybridization assay (see below). As expected, there was a profound down-regulation of mRNA transcripts for COL2A1 and a corresponding increase in transcripts for COL1A1 in the dedifferentiated cells, in accord with previously reported results (36).

thumbnail image

Figure 1. Morphology of human fetal chondrocytes (HFCs). HFCs were isolated by enzymatic digestion as described in Materials and Methods and were cultured for 11 days, either as differentiated chondrocytes on poly-(2-hydroxyethyl methacrylate) (poly-HEMA)–coated dishes or as dedifferentiated chondrocytes on tissue culture plastic. A,HFCs cultured on poly-HEMA–coated dishes. B,HFCs cultured on tissue culture plastic. (Original magnification × 100.)

Download figure to PowerPoint

thumbnail image

Figure 2. Northern analysis of poly(A+) mRNA isolated from differentiated and dedifferentiated HFCs. Chondrocytes were cultured on poly-HEMA or tissue culture plastic for 11 days, and poly(A+) RNA was isolated as described in Materials and Methods. Northern blotting was performed with 100 ng of poly(A+) RNA, and the membrane was hybridized with cDNA probes for human COL2A1 (A) and human COL1A1 (B). Hybridization with GAPDH was performed to control for differences in sample loading. pH = poly-HEMA (differentiated chondrocytes); Pl = tissue culture plastic (dedifferentiated chondrocytes) (see Figure 1 for other definitions).

Download figure to PowerPoint

Microarray hybridization findings. Using the mRNA samples shown in Figure 2, a microarray hybridization analysis was performed employing the UniGEM Human V Microarray (Genome Systems). This array contains ∼5,000 known genes and ∼3,000 ESTs. Table 3 lists the 281 genes most frequently expressed by HFCs in poly-HEMA culture, grouped into 18 different categories. The list encompasses a wide variety of genes, including 22 different ECM genes and 26 different transcription factors. Table 4 lists relevant genes which showed a ≥2-fold difference in expression between differentiated and dedifferentiated HFCs, grouped into the following 4 categories: 1) ECM proteins, 2) transcription/gene regulatory factors, 3) growth factors/cytokines/extracellular signaling molecules, and 4) cell adhesion proteins.

Table 3. List of 283 genes most frequently expressed by human fetal epiphyseal chondrocytes in poly-HEMA culture*
  • *

    Poly-HEMA = poly-(2-hydroxyethyl methacrylate); ID = Unigene reference number; C = category (1 = ribosomal proteins, 2 = nuclear proteins, 3 = transcription factors, 4 = membrane proteins/receptors, 5 = signal transduction/G-proteins, 6 = extracellular matrix proteins, 7 = proteases, 8 = other, 9 = cytoskeletal proteins, 10 = channel proteins, 11 = metabolic enzymes, 12 = growth factors/cytokines, 13 = kinases/phosphatases, 14 = translation associated, 15 = lysosomal/ubiquitin, 16 = cell adhesion, 17 = apoptosis related, 18 = iron metabolism); DE = differential expression (positive is more highly expressed in poly-HEMA; negative is more highly expressed in monolayer); REL = relative expression level (fluorescence level of Cy3-labeled cDNA corresponding to mRNA from the poly-HEMA culture).

Ribosomal protein L41Hs.1081241−1.411352
Ribosomal protein, large, P0Hs.737421−111197
Ribosomal protein L10Hs.297971−110113
Ribosomal protein L4Hs.28611.110045
Ribosomal protein S4Hs.7534411.37609
Ribosomal protein L23aHs.1847761−1.37605
Ribosomal protein L13aHs.1191221−1.16206
Ribosomal protein S8Hs.151604115980
Ribosomal protein L29Hs.1836981−1.35382
Ribosomal protein S24Hs.1804501−1.14768
Ribosomal protein L19Hs.758791−1.34412
Ribosomal protein L14Hs.1586751−1.23637
Ribosomal protein S10Hs.762301−1.13633
Ribosomal protein S23Hs.346311.13383
Ribosomal protein S27aHs.329711.13119
Ribosomal protein L6Hs.1741311−1.23091
Ribosomal protein L18Hs.754581−1.13075
Ribosomal protein, large, P1Hs.1775921−1.52785
Ribosomal protein S20Hs.81021−1.12727
Histidine triad nucleotide-binding protein (HINT)Hs.2566972−1.212466
Primase, polypeptide 1 (49 kd)Hs.827412−1.112822
DNA fragmentation factor, 45 kd, alpha polypeptideHs.1553442−111835
Marenostrin proteinHs.1737302−1.310729
Human zinc-finger protein (ZNF139)Hs.13239021.310621
Nucleolar autoantigen (55 kd)Hs.20725121.38127
Heterogeneous nuclear ribonucleoprotein A1Hs.2494952−17720
PRP4/STK/WD splicing factorHs.8551217712
RNA binding motif protein 3 (RBM3)Hs.1822252−1.26753
Histone deacetylase 3 (HDAC3)Hs.2797892−1.16623
MDM2-like p53-binding protein (MDMX)Hs.10187421.16341
Zinc-finger protein (ZnF20)Hs.17034121.35852
Translin-associated factor XHs.9624721.25315
Zinc-finger protein 33aHs.7061721.25256
Heterogeneous nuclear ribonucleoprotein A2/B1Hs.755982−1.64995
Zinc-finger protein 184 (Kruppel-like)Hs.15817421.14634
71-kd heat-shock cognate proteinHs.1804142−1.64633
Poly(rC)-binding protein 1Hs.28532−1.44607
Zinc-finger protein 175Hs.11901421.24599
Heat-shock 90-kd protein 1, betaHs.743352−1.24330
Heterogeneous nuclear ribonucleoprotein KHs.1295482−1.23858
Heat-shock 70-kd protein 5Hs.754102−1.33634
RNA pol II transcriptional mediator (Med6)Hs.16773821.13459
Zinc-finger protein 267Hs.14549821.53374
Milk fat globule-EGF factor 8 proteinHs.37452−1.33352
Zinc-finger protein 135Hs.15958221.33179
Zinc-finger protein 173Hs.12782−1.33174
Heterogeneous nuclear ribonucleoprotein A1Hs.2494952−1.43041
E1B-55-kd–associated protein 5Hs.1552182−1.23029
High-mobility group protein 1Hs.1443212−1.52982
Damage-specific DNA binding protein 1 (127 kd)Hs.1083272−1.12913
DNA repair endonuclease subunit (XPF)Hs.892962−1.12908
HBV pX–associated protein 8Hs.2050921.72879
Zinc-finger protein 140Hs.15420521.12868
Heterogeneous nuclear ribonucleoprotein FHs.8082−1.52751
Ataxia telangiectasia (ATM)Hs.19438221.22669
Rod1 polypyrimidine tract binding proteinHs.14507821.42649
Zinc-finger protein 136Hs.1828282−12582
NF kappa B-2 (p49/p100)Hs.7309031.113284
Transcription factor IIH (44-kd subunit)Hs.19135631.17508
(TBP)-associated factor IIA (250 kd)Hs.117931.25292
T cell nuclear receptor NOTHs.8212031.24703
PBX/knotted 1 homeobox 1Hs.15822531.14451
Hypoxia-inducible factor 1, alpha subunitHs.19754033.43669
p300/CBP-associated factorHs.1990613−1.23454
Transcription factor Dp-1Hs.793533−1.43150
Transforming protein sno-NHs.3878332.72935
ETS-related protein 71Hs.194061312903
Forkhead protein (FKHRL1)Hs.14845312858
Helix-loop-helix protein (ld-2)Hs.18091931.12843
Insulin promoter factor 1Hs.329383−1.22832
Far upstream element (FUSE) binding protein 3Hs.15363631.12819
Rel protooncogeneHs.44313312764
Wilms tumor 1Hs.114531.22745
Myogenic repressor I-mf (MDFI) mRNAHs.15320331.92642
SRY (sex-determining region Y)-box 9Hs.231633.52616
Far upstream element (FUSE) binding protein 1Hs.1189623−1.32605
Forkhead box F1Hs.1555913−1.42600
Syntaxin 16Hs.13320741.210370
Choroideremia (Rab escort protein 1)Hs.201041.210310
Stromal cell proteinHs.1799994−1.29790
Cubilin (intrinsic factor-cobalamin receptor)Hs.16620641.29265
Platelet-activating factor receptorHs.4641.26993
Interleukin-17 receptorHs.129751416563
CD68 antigenHs.2463814−1.36350
GM-CSF receptorHs.18237841.26144
Major histocompatibility complex, class I, BHs.779614−15762
Leukocyte immunoglobulin-like receptor 1 (LIR1)Hs.20404041.25162
Prostate apoptosis response protein par-4Hs.1760904−1.14750
CD22 antigenHs.17176341.14727
Laminin receptor 1 (67 kd), ribosomal protein SA)Hs.1813574−1.24687
CD63 antigenHs.762944−1.44030
CD43 (sialophorin)Hs.807384−13967
Coated vesicle membrane proteinHs.28882941.43560
Interferon-induced transmembrane protein 3Hs.1822414−1.93494
CD44 antigenHs.1696104−2.13471
Discoidin domain receptor family, member 1Hs.7566241.43412
Interleukin-3 receptor, alphaHs.17268941.13187
Human thyrotropin receptor (TSH)Hs.123078413043
Beta-3-adrenergic receptorHs.254941.32947
Low-density lipoprotein–related protein 1Hs.891374−22926
RAB interacting factor (RABIF)Hs.908754−1.12911
Clathrin, heavy polypeptide (Hc)Hs.1787104−1.22845
Tumor necrosis factor receptor p55Hs.159 /41.32838
Epithelial membrane protein 3Hs.99994−1.42816
GM2 ganglioside activator proteinHs.28908241.32797
Fibroblast activation protein, alphaHs.41841.32779
Lymphocyte antigen 75 (LY75)Hs.1535634−1.12749
Urokinase plasminogen activator receptorHs.1796574−1.42749
Glutamate receptor, metabotropic 2Hs.1215104−1.32717
Glycoprotein (transmembrane) nmbHs.822264−1.32703
Epithelial membrane protein 1Hs.793684−1.12629
Ciliary neurotrophic factor receptorHs.1947744−1.52619
Tuberous sclerosis 2Hs.903035−1.110327
SMAD 5Hs.3750151.210234
G protein–coupled receptor 64 HE6Hs.1849425−1.17058
GNB2L1 (G protein)Hs.56625−1.36842
Phospholipase C-alphaHs.2891015−1.56107
V-crk oncogeneHs.3707851.15527
Human 14-3-3 protein tauHs.7440551.15494
Human 14-3-3 protein zeta/deltaHs.751035−1.75368
TNF receptor–associated factor 1Hs.213451.24846
Mannose 6 phosphate receptor binding proteinHs.1404525−1.34843
Phospholipase C, epsilonHs.15332251.24595
GNB1 (G protein)Hs.2155955−1.53921
Human G protein–coupled receptor (GPR4)Hs.1717051.23094
Protein kinase C substrate 80K-HHs.14325−1.53016
Synaptic Ras GTPase-activating protein 1Hs.208305−1.42984
GTPase-activating protein–likeHs.1843675−1.12974
G protein–coupled receptor CKR-L3Hs.464685−1.22955
Rab6 GTPase-activating proteinHs.5509951.12867
rho guanine nucleotide exchange factor (GEF) 1Hs.2522805−1.12723
Collagen, type XI, alpha 1Hs.8277264.911067
Aggrecan 1Hs.215964.49942
Collagen, type XI, alpha 2Hs.121509611.39935
Fibronectin 1Hs.1181626−1.38190
Collagen, type V, alpha 1Hs.14642861.56583
Cartilage-linking protein 1Hs.2799612.26570
Collagen, type IX, alpha 2Hs.3716564.95961
Human cartilage glycoprotein 39Hs.751846−4.45715
Collagen, type VI, alpha 3Hs.809886−1.95434
Matrilin 3Hs.278461627.64395
Cartilage oligomeric matrix protein (COMP)Hs.158462.14121
Collagen, type XIV, alpha 1Hs.361316−1.33644
Collagen, type I, alpha 1Hs.1729286−2.13371
Fibulin 1Hs.797326−2.53106
Matrilin 1Hs.15036662.12979
Fibulin-like extracellular matrix protein 1Hs.762246−1.42751
Hexabrachion (tenascin-C)Hs.2891146−4.72615
Matrix metalloproteinase–like 1Hs.1982657−19594
Carboxypeptidase MHs.169765719523
Matrix metalloproteinase 2Hs.1113017−1.17632
Alpha-1-antichymotrypsin precursorHs.23472679.87029
Caspase 2Hs.10813171.26944
Caspase 10Hs.535371.15089
Cathepsin BHs.2499827−3.35061
Protease, serine, 11 (IGF binding)Hs.751117−1.83137
Caspase 6Hs.328071.12789
Calpain 4, small subunitHs.744517−1.82576
Liprin beta 1Hs.10217881.311649
Pulmonary-associated surfactant protein BHs.7630581.27335
Galectin 6–binding proteinHs.79339814045
Galectin 3Hs.62181.13369
Reticulocalbin 1Hs.1677918−1.13244
FK506 binding protein 8 (38 kd)Hs.1734648−1.12757
Cellular retinoic acid binding proteinHs.767881.22610
Coronin actin binding protein 2A (CORO2A)Hs.443969−111470
Kinesin family member 3BHs.168212918363
Tubulin, betaHs.1796619−1.77950
Tubulin, beta 5Hs.1080149−1.86055
Tropomyosin 4Hs.2506419−1.55487
Transgelin 2Hs.757259−1.95077
Tropomyosin 2 (beta)Hs.1802669−2.34066
Calponin 2Hs.1697189−1.53788
Actin, alpha 2, smooth muscleHs.1955819−2.23627
Actin bundling protein (HSN)Hs.1184009−1.32771
Potassium inwardly-rectifying channelHs.3716910−121344
Solute carrier family 24 (Na+/K+/Ca2+ exchanger)Hs.173092101.116302
Solute carrier family 31 (copper transporters)Hs.73614101.110541
Solute carrier family 15 (oligopeptide transporter)Hs.2217101.18957
ATPase, Na+/K+ transporting, alpha 2Hs.34114101.17122
Solute carrier family 17 (anion/sugar transporter)Hs.117865101.26127
Solute carrier family 25Hs.16428010−1.14425
Solute carrier family 35Hs.2189910−1.33251
Solute carrier family 2Hs.7594101.13247
Pancreatic lipase–related protein 1 (PLRP1)Hs.7392311−1.111672
Lactate dehydrogenase A (LDHA)Hs.279511−1.211360
Dihydrofolate reductaseHs.8376511110154
Carbonic anhydrase XIIHs.533811−18518
Procollagen proline 4-hydroxylase (beta)Hs.7565511−1.28160
Fatty acid desaturase MLDHs.18597311−1.38097
Enolase 1 (alpha)Hs.25410511−18016
Triosephosphate isomerase 1Hs.83848111.87677
Sialyltransferase 8Hs.17018011−1.17655
Aldolase A, fructose-bisphosphateHs.273415111.87350
Prostaglandin 12 (prostacyclin) synthaseHs.6133311−1.16485
Cysteine sulfinic acid decarboxylase–related proteinHs.279815111.15826
Glucose phosphate isomeraseHs.18053211−1.15294
NADH dehydrogenase (ubiquinone)Hs.183435111.15085
Fatty acid desaturase 1Hs.1328981114957
Cytochrome P450, subfamily IIBHs.1360111.34670
Diaphorase (NADH/NADPH cytochrome b5 reductase)Hs.80706111.44660
Procollagen-lysine lysine hydroxylase 2Hs.41270113.13860
Phosphoglycerate kinase 1Hs.787711133854
Tyr 3-monooxygenase/trp 5-monooxygenase APHs.7554411−1.53428
Peptidylglycine alpha-amidating monooxygenaseHs.83920111.13422
Nifedipine oxidase (cytochrome P450)Hs.1787381113390
Sterol-C4-methyl oxidase–likeHs.2399261113341
ATP synthase-H+ transporting (beta)Hs.2511−1.13266
5′-nucleotidase (purine), cytosolic type BHs.138593111.13232
Serine hydroxymethyltransferase 2Hs.75069111.33146
N-acetylgalactosaminidase, alphaHs.7537211−13135
Nitric oxide synthase 1Hs.46752111.13101
Enoyl-CoA hydratase/3-OHacyl-CoA dehydrogenaseHs.7586011−1.12843
Procollagen (type III) N-endopeptidaseHs.18313811−1.12745
ADP-ribosylation factor 3Hs.119177111.22742
Peptidylprolyl isomerase A (cyclophilin A)Hs.18293711−1.62722
Steroid 11-beta-hydroxylaseHs.261011−1.12698
mRNA for ATP synthase alpha subunitHs.15510111−1.12626
Aldehyde dehydrogenase 10Hs.159608111.22586
Glutathione peroxidase 1Hs.7668611−1.82577
Cytochrome c heme-lyaseHs.211571111.12566
Acyl-coenzyme A oxidase 2Hs.979511−1.52549
Transforming growth factor, beta-induced, 68 kdHs.118787121.911941
Interferon-induced proteinHs.1818741219003
Insulin-like growth factor 2Hs.251664126.38249
Insulin-like growth factor binding protein 3Hs.7732612−2.66150
Teratocarcinoma-derived growth factor 1Hs.75561121.34937
Neuregulin 2Hs.113264121.14605
Connective tissue growth factorHs.75511121.74484
Cyr61 protein (CYR61)Hs.886712−13600
Insulin-like growth factor binding protein 4Hs.151612−3.63536
Pleiotrophin (heparin-binding growth factor 8)Hs.44121.52639
Protein kinase, PKX1Hs.14799613−16231
Mitogen-activated protein kinase 12Hs.5503913−1.13685
SFRS protein kinase 2Hs.7835313−1.43667
Mevalonate kinaseHs.130607131.13483
Calcium/calmodulin-dependent serine protein kinaseHs.15146913−1.23398
Homeodomain-interacting protein kinase 3Hs.30148131.12911
MAPKKK1 (M3K1)Hs.29872713−1.12747
Dual-specificity phosphatase 4Hs.235913−1.22736
cdc-like kinase 2Hs.7398613−1.12647
EphB3 (receptor tyrosine kinase, HEK 2)Hs.291313−1.12556
Eukaryotic translation initiation factor 3Hs.192023141.29982
Eukaryotic translation initiation factor 4 gammaHs.18368414−1.36641
Signal sequence receptor, betaHs.7456414−1.42745
Ubiquitin BHs.18384215−1.18168
Ubiquitin-conjugating enzyme E2G 2Hs.192853151.26865
Ubiquitin CHs.18370415−1.16491
Proteasome subunit HsC7-IHs.139015−16111
Lysosomal-associated membrane protein 2Hs.826215−1.15904
Nedd-4-like ubiquitin-protein ligaseHs.166119153.74550
Polycystic kidney disease protein 1Hs.7581316−1.16939
Integrin binding protein Del-1Hs.129764163.56779
Thrombospondin 2Hs.10862316−1.45852
Catenin (cadherin-associated protein) alpha 1Hs.17845216−1.33672
Thrombospondin 1Hs.87409161.83367
Alpha integrin binding protein 63Hs.292051612617
Caspase-like apoptosis regulatory protein (CLARP)Hs.195175171.35087
Death-associated protein 6Hs.18022417−1.62728
Ferritin, light polypeptideHs.11133418−1.68964
Ferredoxin 1Hs.744181.15319
Iron-responsive element binding protein 2Hs.295789181.13815
Ferritin, heavy polypeptide 1Hs.6295418−1.23231
Table 4. Relevant genes that display a ≥2-fold difference in expression between differentiated and dedifferentiated human fetal chondrocytes (HFCs)*
GeneFold difference
Extracellular matrix proteins
 Higher expression in differentiated HFCs
  Matrilin 327.6
  Dermatan sulfate proteoglycan 36.5
  Chondroitin sulfate proteoglycan 32.5
  Cartilage linking protein 12.2
  Matrilin 12.1
 Higher expression in dedifferentiated HFCs
  Chitinase 3–like protein 14.4
  Fibrillin 12.8
  Fibulin 12.5
Transcription factors
 Higher expression in differentiated HFCs
  Hypoxia-inducible factor 1α3.4
  RING zinc-finger protein RZF2.2
  Zinc-finger protein 352.0
  MADS/MEF2 family transcription factor2.0
 Higher expression in dedifferentiated HFCs
Growth factors/cytokines/extracellular mediators
 Higher expression in differentiated HFCs
  Frizzled-related protein10.2
  Melanoma growth regulatory protein (CD-RAP)3.9
 Higher expression in dedifferentiated HFCs
  Insulin-like growth factor binding protein 43.6
  Insulin-like growth factor binding protein 32.6
Adhesion proteins
 Higher expression in differentiated HFCs
  Del-1 integrin binding protein3.5
  Epithelial V–like antigen (EVA)3.4
 Higher expression in dedifferentiated HFCs
  Cadherin 112.8

We observed that a large number of chondrocyte-specific ECM protein genes were down-regulated (some as much as 27-fold) in response to dedifferentiation. These results are in accord with a large body of previously reported results concerning the dedifferentiation of chondrocytes from multiple species. We also observed the up-regulation of numerous genes encoding ECM proteins associated with the fibroblastic or prechondrogenic phenotype displayed by dedifferentiated cells, including COL1A1 and tenascin (37–40).

Among the gene regulatory factors whose mRNA levels were differentially regulated by the two culture conditions were the chondrogenic transcription factor SOX9, the O2-regulated HIF-1α, the bHLH transcription factor Twist, the winged-helix transcription factor Freac-4, and the retinoic acid receptor RXR-β. In the category of growth factors/cytokines/extracellular signaling molecules, we observed that differentiated chondrocytes expressed higher levels of transcripts for Frizzled-related protein, IGF-2, melanoma growth regulatory protein (also known as CD-RAP [cartilage-derived retinoic acid–induced protein]), and BMP-6. Interestingly, we observed higher levels of the transcripts for two IGFBPs, IGFBP-3 and IGFBP-4, in the dedifferentiated chondrocytes. Finally, in the category of cell adhesion proteins, there was an increase in the levels of transcripts for the bone-associated adhesion protein, cadherin 11, in the dedifferentiated chondrocyte cultures and in the endothelial cell–associated adhesion protein, Del-1, in the differentiated chondrocyte cultures (45).

Validation of microarray results with Northern hybridizations. In order to validate the microarray observations, we performed Northern blot analyses to examine the levels of expression of transcripts from selected genes that showed substantial differential expression in the microarray studies. Figure 3 shows the results of Northern analyses for expression of IGF-2, BMP-6, IGFBP-4, and RXR-β transcripts in differentiated compared with dedifferentiated chondrocytes. A 24-fold difference in the level of IGF-2 transcripts between the two culture conditions was observed (Figure 3A), following normalization for the levels of GAPDH transcripts, compared with a 6.3-fold difference that was obtained with the microarray. In the case of the IGFBP-4 gene, the difference observed in the Northern analysis was 3.5-fold (Figure 3C), versus the 3.6-fold difference observed with the microarray. In the case of the BMP-6 gene (Figure 3B), we also detected a higher level of expression in the differentiated chondrocytes by Northern analysis, as we observed with the microarray. The expression differences of RXR-β (Figure 3D) and SOX9 (data not shown) by Northern analyses were approximately equal to those obtained with the microarray.

thumbnail image

Figure 3. Northern analysis of poly(A+) mRNA isolated from differentiated and dedifferentiated HFCs. Chondrocytes were cultured on poly-HEMA or tissue culture plastic for 11 days, and poly(A+) RNA was isolated as described in Materials and Methods. Northern blotting was performed with 100 ng of poly(A+) RNA, and the membranes were hybridized with cDNA probes for human insulin-like growth factor 2 (IGF-2) (A), bone morphogenetic protein 6 (BMP-6) (B), IGF binding protein 4 (IGFBP-4) (C), and the retinoic acid receptor RXR-β (D). Hybridization with GAPDH was performed to control for differences in sample loading. pH = poly-HEMA (differentiated chondrocytes); Pl = tissue culture plastic (dedifferentiated chondrocytes) (see Figure 1 for other definitions).

Download figure to PowerPoint

Although we observed some differences in the levels of expression of these transcripts compared with the microarray hybridization results, the trends remained the same. Microarray hybridization analysis is more sensitive than Northern analysis, with a detection limit of ∼2 pg versus a limit of ∼20 pg for Northern analysis. However, microarray hybridization depends on a cDNA synthesis step to generate the probe, and this could influence the relative levels of transcripts detected by the microarray analysis. Also, normalization to GAPDH transcripts, which may vary between the two culture states, could add to the observed variability.

We observed a 3.4-fold lower level of expression of HIF-1α in dedifferentiated chondrocytes with the microarray, and by Northern analysis we observed a 3.5-fold difference when normalized to GAPDH, similar to that obtained with the microarray (Figure 4A). Since we were primarily interested in gene regulatory factors that were differentially regulated by the two culture conditions, we proceeded to document HIF-1α protein levels and binding activity in freshly isolated chondrocytes. We hypothesized that as chondrocytes form clusters in poly-HEMA culture, they become hypoxic and should therefore express HIF-1α (46). In order to address this hypothesis, we performed an EMSA to detect binding of HIF-1, which is a dimer of HIF-1α and HIF-1β (also known as aryl hydrocarbon receptor nuclear translocator [ARNT]). For this experiment, we used a 20-bp probe containing either a wild-type or mutated version of a HIF-1 binding site from the human VEGF gene (43). As can be seen in Figure 4B, we detected the formation of a slowly migrating complex (see arrow) only with the wild-type probe, consistent with previous results concerning binding by HIF-1 (43).

thumbnail image

Figure 4. Analysis of hypoxia-inducible factor 1α (HIF-1α) mRNA and HIF-1 DNA-binding activity in HFCs and bovine fetal chondrocytes. Poly(A+) RNA was isolated from differentiated and dedifferentiated HFCs, and nuclear extracts were prepared from differentiated bovine fetal chondrocytes as described in Materials and Methods. A, Northern analysis of poly(A+) mRNA (100 ng) isolated from differentiated and dedifferentiated HFCs with cDNA probes for HIF-1α and GAPDH. pH = poly-HEMA (differentiated chondrocytes); Pl = tissue culture plastic (dedifferentiated chondrocytes). B, Electrophoretic mobility shift assay of bovine fetal chondrocyte nuclear extracts with a DNA probe containing an HIF-1 binding site from the human VEGF gene. WT = wild-type binding site; MUT = mutated binding site; − or + indicates absence or presence of 10 μg nuclear protein. Arrowindicates complex formed with the WT probe but not with the MUT probe. See Figure 1 for other definitions.

Download figure to PowerPoint

In contrast to the results with HIF-1α, Northern analysis for another transcription factor, Twist, showed higher expression in dedifferentiated chondrocytes (Figure 5A), as was observed in the microarray analysis. Concordant with these results, we also observed Twist transcripts in HFCs that were transformed with a retroviral vector expressing the E6 and E7 proteins from the human papilloma virus and grown for multiple passages in the dedifferentiated monolayer state and in HFCs passaged 4 times at 11-day intervals in monolayer (Figure 5B). In contrast, we did not detect Twist expression in HTB-94 human chondrosarcoma cells, which appear to maintain a more differentiated state (47).

thumbnail image

Figure 5. Northern analysis of total RNA and mRNA isolated from differentiated and dedifferentiated HFCs and from human chondrosarcoma cells and transformed HFCs. Total cellular and poly(A+) RNA was isolated from differentiated and dedifferentiated HFCs, and poly(A+) RNA was isolated from HTB-94 human chondrosarcoma cells, E6–E7–transformed HFCs, and passage 4 HFCs as described in Materials and Methods. A, Northern analysis of total RNA (20 μg) isolated from differentiated and dedifferentiated HFCs with cDNA probes for human type II collagen and the basic helix-loop-helix transcription factor Twist. Below is shown the ethidium bromide–stained 18S RNA for comparison of sample loading. B, Northern analysis of poly(A+) mRNA (1.5 μg) isolated from HTB-94 human chondrosarcoma cells (HTB), transformed HFCs (TFC), or HFCs grown as a monolayer through 4 passages of 11 days each (P4) probed with human cDNA for Twist and GAPDH. Pl = tissue culture plastic (dedifferentiated chondrocytes); pH = poly-HEMA (differentiated chondrocytes) (see Figure 1 for other definitions).

Download figure to PowerPoint

RT-PCR analyses of the expression of selected genes in adult cartilage and chondrocytes. In order to determine whether certain genes that exhibited differential expression in HFCs in response to the two culture conditions were also expressed in normal and OA adult AC, we performed RT-PCR analysis of RNA isolated from adult normal and OA cartilage from multiple individuals. As shown in Figures 6A and B, HIF-1α, Twist, IGF-2, IGFBP-3, IGFBP-4, Del-1, and cadherin 11 transcripts were expressed in OA cartilage samples. RT-PCR analysis of β-actin from the same samples is shown as a control. In a separate experiment, RT-PCR analysis for Twist expression was performed on RNA from freshly isolated human normal adult and OA chondrocytes. Twist RNA expression was detected in 1 normal adult sample and in 3 OA samples (Figure 6C).

thumbnail image

Figure 6. Reverse transcriptase–polymerase chain reaction (RT-PCR) analysis of the expression of selected genes in adult human normal and osteoarthritic (OA) cartilage and chondrocytes. Total cellular RNA was isolated directly from cartilage samples obtained from OA patients or from chondrocytes that were freshly isolated from OA cartilage as described in Materials and Methods. A,RT-PCR analysis for the expression of HIF-1α, Del-1, β-actin, IGF-2, IGFBP-3, IGFBP-4, and the basic helix-loop-helix transcription factor Twist in RNA directly isolated from the cartilage of 4 different patients with OA. B, RT-PCR analysis for the expression of cadherin 11, performed as described in A. C,RT-PCR analysis for expression of Twist in RNA from freshly isolated chondrocytes from adult human normal (N) and OA cartilage. M = 100-bp marker. The primer sets employed for amplification are shown in Table 2. For the analysis of Twist in A, the Twist-a primer set was used. For the analysis of Twist in C, the Twist-b primer set was used. The Twist-b primer set bridges an intron. See Figures 3 and 4 for other definitions.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  6. Acknowledgements

We performed a microarray analysis of differential gene expression comparing differentiated with dedifferentiated HFCs. HFCs were cultured for 11 days on poly-HEMA–coated dishes in order to preserve their differentiated phenotype, or in plastic tissue culture flasks as primary monolayer cultures in order to induce their dedifferentiation. Poly(A+) RNA obtained from these two sets of chondrocytes was subjected to a microarray hybridization analysis on the UniGEM Human V Microarray (Genome Systems). This array contains ∼5,000 known human genes and ∼3,000 ESTs.

The results of the screen showed a dramatic change in phenotype as evidenced by the down-regulation of numerous genes associated with the ECM of cartilage (types IX and XI collagen, aggrecan, COMP) and the up-regulation of ECM genes associated with an undifferentiated mesenchymal cell phenotype (COL1A1, tenascin, cadherin 11, TWIST). We also demonstrated by RT-PCR analysis the expression of 7 genes, HIF-1α, Del-1, IGFBP-3, IGFBP-4, TWIST, IGF-2, and cadherin 11, in human OA and normal AC or in chondrocytes from these tissues. To our knowledge, this is the first report of the expression of 3 of these genes in human articular chondrocytes or cartilage (HIF-1α, cadherin 11, TWIST). However, it should be noted that since only one time point (11 days) was taken, many important genes that are expressed or repressed earlier, later, or transiently during the dedifferentiation process may have been missed in this screen.

Transcription factors that showed differential expression between the two cell states were of interest, since it is likely that they might play a role in controlling the phenotypic difference. Indeed, we found important changes in the regulation of expression of the genes for SOX9, TWIST, and HIF-1α. SOX9 is a member of the sex-determining region Y–type high mobility group box protein family and has been shown to be required for chondrocyte-specific gene expression and chondrogenesis (48–50). We also observed the involvement of the SOX factors in the dedifferentiation process (51). Interestingly, recent investigations have established a connection between SOX9 and the protein CD-RAP. CD-RAP was identified as a protein that is down-regulated in response to retinoic acid treatment of chondrocytes (52). Retinoic acid causes dedifferentiation of chondrocytes, inducing many of the characteristics observed when chondrocytes become dedifferentiated by monolayer culture (53, 54). SOX9 has been shown to up-regulate the expression of CD-RAP in chondrocytes (55), and, in accordance with this observation, we found that transcripts for SOX9 and CD-RAP were down-regulated in dedifferentiated chondrocytes, while transcripts for the retinoic acid receptor RXR-β were up-regulated.

The bHLH transcription factor TWIST plays a role in determining mesodermal cell fate; however, it is also expressed in adult tissues arising from the mesoderm (56). TWIST was originally identified in Drosophila for its role in gastrulation, mesoderm formation, and myogenesis (57). The human TWIST gene encodes a 21-kd protein that shares extensive homology with the mouse and Drosophila Twist proteins (58). Basic HLH transcription factors such as Twist form either hetero- or homodimers in order to bind a DNA consensus sequence, called an E-box, that is found in the promoters of numerous genes. These factors can either inhibit or activate the transcription of other genes through binding to E-box sites, or by other mechanisms such as transcription factor sequestration (57, 59).

It has been hypothesized that TWIST might regulate chondrogenesis by promoting or maintaining the prechondrogenic cell phenotype, and that differentiation into chondrocytes proceeds through down-regulation of TWIST expression (57, 60). Indeed, TWIST expression in the developing mouse coincides with areas adjacent to chondrogenic differentiation, such as those surrounding cells expressing the embryonic type IIA procollagen variant (60). In the results presented here, we observed an increase in the expression of TWIST in dedifferentiated chondrocytes, in contrast to chondrocytes that were maintained in a differentiated state. We also observed the expression of TWIST transcripts in adult normal and OA cartilage and in chondrocytes derived from these tissues. This latter result supports the hypotheses that dedifferentiated chondrocytes may express a prechondrogenic-like phenotype and that TWIST may regulate genes that are important for the establishment of the two different phenotypes (differentiated versus dedifferentiated).

Another transcription factor that showed a differential expression pattern between differentiated and dedifferentiated chondrocytes was HIF-1α. HIF-1α is part of the heterodimeric HIF-1 transcription factor that regulates a host of genes in response to hypoxia, insulin, IGFs, IL-1β, and tumor necrosis factor α, all of which are known to be important for chondrocyte physiology and pathophysiology (for review, see ref. 61). HIF-1α is an HLH–PAS (Per, ARNT, Sim) domain–containing transcription factor that, along with HIF-1β (also known as ARNT, and also an HLH–PAS factor), forms the heterodimeric factor HIF-1 in response to hypoxia (47, 62). When cells are subjected to hypoxic conditions or agents such as Ni, Co, Mn, CO, and/or antioxidants, the HIF-1α protein becomes induced and can heterodimerize with HIF-1β, leading to the activation by HIF-1 of downstream genes involved in the response to hypoxia, such as those mediating anaerobic metabolism, oxygen transport, vasodilation, and blood vessel formation (61).

We observed a down-regulation of HIF-1α mRNA levels in dedifferentiated chondrocytes. Cartilage is considered to be a relatively hypoxic tissue, and we postulate that, concomitant with the formation of the cartilage-like nodules in the poly-HEMA culture system, the chondrocytes become hypoxic. Similarly, we observed that during the early stages of growth in monolayer, chondrocytes tend to form clusters and/or nodules which could also become partially hypoxic (63). At later times of monolayer culture, the dedifferentiated chondrocytes become fibroblastic in morphology and form only a thin layer of cells that presumably would be nonhypoxic. These differences could explain the remarkable differences in HIF-1α mRNA levels between the two culture conditions. However, the observation that HIF-1α mRNA levels are higher in differentiated chondrocytes that are growing in an environment that allows them to form a cartilage-like tissue indicates that HIF-1α may be present in cartilage in vivo. Indeed, we have detected HIF-1α transcripts in both human adult normal and OA cartilage.

We also noted that along with the genes mentioned above, the IGF-2 gene, two IGFBP genes (IGFBP-3 and IGFBP-4), and the BMP-6 gene were differentially regulated. IGFs 1 and 2 are known to stimulate cartilage matrix synthesis, glucose uptake, and growth of chondrocytes and prechondrocytes (64, 65). Evidence now indicates that the proteins encoded by these genes function as autocrine factors promoting the survival of chondrocytes (66). Our investigations indicate that dedifferentiated chondrocytes may no longer require IGF-2 or are less dependent on it for their survival. Interestingly, the mRNA levels of certain IGFBPs are also increased in OA cartilage as they are in dedifferentiated cells. The resulting increase in the corresponding proteins would further reduce the action of the IGFs, thus contributing to the loss of cartilage matrix in OA (64). Finally, a few studies have implicated BMP-6 as having a role in inducing the differentiation of chondrocytes in the growth plate (67, 68).

Current hypotheses to explain the changes in OA chondrocytes include the notion that phenotypic alterations of the cells occur in response to changing signals or matrix composition (10–13). It is well known that chondrocytes can adopt different phenotypes, including those associated with the growth plate (resting, proliferating, hypertrophic), those associated with the different zones of AC, and the dedifferentiated phenotype observed in monolayer culture (3, 7, 13, 17). Numerous studies have investigated the phenotypic alterations observed in OA cartilage at the molecular level by RT-PCR, Northern and in situ hybridizations, and immunohistologic analyses. These studies have shown that a major change in the chondrocyte phenotype in OA involves a switch in the types of collagen molecules that are synthesized. Clusters of chondrocytes in OA cartilage have been shown to express types I and III collagens, which are not normally found or found at very low levels in normal AC (10, 12, 69–71).

Recently, it has been demonstrated that OA chondrocytes initiate expression of the fetal type IIA procollagen variant that arises from alternative splicing of exon two of the COL2A1 gene (13). It has also been shown that chondrocytes within the deep zones of OA cartilage can express type X collagen, indicating that they may be undergoing hypertrophy (8, 9, 72, 73). Others have observed the synthesis by OA chondrocytes of aggrecan molecules that resemble the structure of aggrecan molecules in fetal chondrocytes (11). At least one transcription factor also shows differential expression between normal and OA cartilage. Wang et al (74) reported that the early growth response protein, Egr-1, is down-regulated in OA cartilage compared with normal cartilage. Egr-1, a zinc-finger transcription factor, has been shown to be important for cell proliferation, differentiation, and apoptosis, and is expressed in epiphyseal cartilage and AC during development (75, 76). Interestingly, Egr-1 showed a wide expression pattern in normal cartilage, but was restricted to the clusters of clonal chondrocytes observed in OA cartilage (74). These data collectively indicate that OA chondrocytes may express a more dedifferentiated phenotype than normal chondrocytes. Therefore, the study of gene expression of normal human chondrocytes undergoing dedifferentiation may have implications for the understanding of the pathogenesis of OA.

The molecular mechanisms responsible for controlling the phenotypic plasticity of chondrocytes still remain poorly understood. Based on the results presented here, and on previous investigations by us and others, we suggest that altered expression of the transcription factors SOX9 and Twist could play a role in the different phenotypes displayed by chondrocytes in culture. In support of this hypothesis, it has been shown that in chimeric Sox9−/− mice, mesenchymal cells deficient for SOX9 could not differentiate into chondrocytes (50). Mesenchymal cells in Twist knockout mice also display abnormalities in differentiation and proliferation, and overexpression of Twist in osteoblast-like cells inhibits their differentiation (77, 78). It is likely that the combined down-regulation of SOX9 and up-regulation of Twist in dedifferentiated cells is prohibitive for redifferentiation. This hypothesis is currently being examined.

In conclusion, the results described here demonstrate a profound change in the pattern of gene expression that can occur when chondrocytes are cultured under conditions that do not allow the cells to establish 3-dimensional contacts with each other or with a surrounding matrix; these changes encompass a broad range of genes, including those for matrix proteins, growth factors, and transcription factors. Since these remarkable changes were detected when chondrocytes were cultured as primary cultures in monolayers for just 11 days, these results have important implications for interpretation of studies carried out on chondrocytes that are cultured as monolayers, as well as for the use of such chondrocytes in cell therapy applications.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The expert assistance of Kate Salmon and Maria Mokan in the preparation of the manuscript and the technical assistance of Daniel Luk (Novartis) in the gene expression analysis in human cartilage samples are gratefully acknowledged.


  1. Top of page
  2. Abstract
  6. Acknowledgements