Proteolysis Involving Matrix Metalloproteinase 13 (Collagenase-3) Is Required for Chondrocyte Differentiation That Is Associated with Matrix Mineralization


  • C. William Wu,

    1. Joint Diseases Laboratory, Shriners Hospitals for Children, Canadian Hospital and Department of Surgery, Division of Surgical Research, McGill University, Montreal, Quebec, Canada
    2. Present address: Arthritis Unit, Department of Medicine, Massachusetts General Hospital, Charlestown, Massachusetts, USA
    Search for more papers by this author
  • Elena V. Tchetina,

    1. Joint Diseases Laboratory, Shriners Hospitals for Children, Canadian Hospital and Department of Surgery, Division of Surgical Research, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Fackson Mwale,

    1. Joint Diseases Laboratory, Shriners Hospitals for Children, Canadian Hospital and Department of Surgery, Division of Surgical Research, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Karen Hasty,

    1. Department of Anatomy and Neurobiology, University of Tennessee, Memphis, Tennessee, USA
    Search for more papers by this author
  • Isabelle Pidoux,

    1. Joint Diseases Laboratory, Shriners Hospitals for Children, Canadian Hospital and Department of Surgery, Division of Surgical Research, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Agnes Reiner,

    1. Joint Diseases Laboratory, Shriners Hospitals for Children, Canadian Hospital and Department of Surgery, Division of Surgical Research, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Jeffrey Chen,

    1. Inflammatory Diseases Unit, Roche Bioscience, Palo Alto, California, USA
    Search for more papers by this author
  • Harold E. Van Wart,

    1. Inflammatory Diseases Unit, Roche Bioscience, Palo Alto, California, USA
    Search for more papers by this author
  • A. Robin Poole Ph.D., D.Sc.

    Corresponding author
    1. Joint Diseases Laboratory, Shriners Hospitals for Children, Canadian Hospital and Department of Surgery, Division of Surgical Research, McGill University, Montreal, Quebec, Canada
    • Joint Diseases Laboratory, Shriners Hospitals for Children, Canadian Hospital, 1529 Cedar Avenue, Montreal, Quebec H3G 1A6, Canada
    Search for more papers by this author

  • The authors have no conflict of interest.


Collagenases are involved in cartilage matrix resorption. Using bovine fetal chondrocytes isolated from physeal cartilages and separated into a distinct prehypertrophic subpopulation, we show that in serum-free culture they elaborate an extracellular matrix and differentiate into hypertrophic chondrocytes. This is characterized by expression of type X collagen and the transcription factor Cbfa1 and increased incorporation of45Ca2+ in the extracellular matrix, which is associated with matrix calcification. Collagenase activity, attributable only to matrix metalloproteinase (MMP) 13 (collagenase-3), is up-regulated on differentiation. A nontoxic carboxylate inhibitor of MMP-13 prevents this differentiation; it suppresses expression of type X collagen, Cbfa1, and MMP-13 and inhibits increased calcium incorporation in addition to inhibiting degradation of type II collagen in the extracellular matrix. General synthesis of matrix proteins is unaffected. These results suggest that proteolysis involving MMP-13 is required for chondrocyte differentiation that occurs as part of growth plate development and which is associated with matrix mineralization.


DURING DEVELOPMENT there is extensive growth and remodeling of tissues. This involves proteolysis of the extracellular matrix.(1) One such event is the process of endochondral ossification such as that observed in the primary and secondary growth plates and in fracture repair. Growth plate chondrocytes establish an extracellular matrix containing type II collagen fibrils and noncollagenous proteins, which they then resorb and remodel as calcification ensues. This requires chondrocyte differentiation from prehypertrophic to hypertrophic cells.(2,3) During hypertrophy and the associated matrix resorption, chondrocytes synthesize type X collagen. There is associated denaturation and loss of type II collagen(4,5) resulting from increased collagenase activity.(5–8)

The role of collagenases in matrix remodeling in endochondral ossification is unclear. The transcription factor Cbfa1 has been shown to regulate expression of matrix metalloproteinase (MMP) 13 (collagenase-3).(9,10) In mice lacking Cbfa1, chondrocyte hypertrophy and matrix mineralization often is impaired and MMP-13 expression is suppressed.(11,12) Of the collagenases, MMP-13 is the most effective at cleaving type II collagen.(13–15) Recent work on mice has revealed that MMP-8(16,17) and two genes that are closely related to MMP-1(18) are expressed also. Presently, there is no information to indicate that they are expressed in the growth plate.

In this study we show that MMP-13 is up-regulated on hypertrophy of bovine chondrocytes. MMP-1 is not expressed. We find that a specific nontoxic inhibitor of MMP-13 can suppress hypertrophy. This results in inhibition of the increase in45Ca2+ incorporation, which is associated with matrix mineral deposition, in addition to suppressing degradation of type II collagen. The implication of these observations is that MMP activity, involving that of MMP-13, is required for chondrocyte differentiation (hypertrophy) that is involved in growth plate development.


Source of cells and tissues

Bovine fetuses were obtained from a local abattoir (Colbex, St. Cyrille, QC, Canada) and transported to the laboratory. Within 1.5 h of death, femora, tibias, and humeri were removed. Fetal age was determined by measurement of tibial length as described previously.(19) Ages ranged from 190 to 221 days. Bovine fetal skin fibroblasts were used that had been isolated in this laboratory.

Cell isolation and culture

Bovine fetal growth plate chondrocytes were isolated and fractionated into subpopulations, with minor modifications of a procedure as previously described.(20,21) Before Percoll gradient separation, isolated cells were washed twice in Dulbecco's modified Eagle's medium (DMEM; Gibco BRL, Burlington, Ontario, Canada) containing antibiotics at room temperature, as described previously,(20) and once in citrate buffer (125 mM of NaCl, 18 mM of citric acid, and 10 mM of K2HPO4, pH 6.0) to dissolve residual mineral. Chondrocyte subpopulation C, which represents a population of prehypertrophic chondrocytes,(20–22) was seeded (2 ml/well) on gelatin-coated 12-well flat-bottomed plates (Falcon, Becton Dickinson, Franklin Lakes, NJ, USA) at a density of 1 × 106 cells/ml. Cells were cultured in 2 ml of DMEM supplemented with 50 μg/ml of ascorbic acid, 5 mM of sodium β-glycerophosphate (both were freshly prepared at each medium change), 5 μg/ml of insulin, 5 μg/ml of transferrin, 5 ng/ml of sodium selenite (ITS), 1 mg/ml of hydrolyzed bovine serum albumin (BSA), and 10 nM of triiodothyronine (T3).(22) Medium was changed every 2 days. For MMP-13 inhibition studies, subpopulation C was cultured as mentioned previously in the presence of 1 nM or 10 nM of RS 102,481, a carboxylate inhibitor (Ki = 0.08 nM) of collagenase-3(23,24) in a final concentration of 0.1% dimethylsulfoxide. This inhibitor has no effect on collagenase-1 at the concentrations used in this study(23,24) Control cultures contained only the solvent.

For maximal MMP induction, chondrocytes and fibroblasts were cultured where indicated in DMEM supplemented with 1 U/ml or 10 U/ml of human recombinant interleukin (IL)-1α (R & D Systems, Minneapolis, MN, USA) for 24 h. IL-1α was used because it is our experience it is more potent for induction of collagenases and proteolysis in chondrocyte populations.

Total RNA extraction and isolation

Total RNA was isolated from fibroblasts or chondrocytes using a modification of a published method.(25) Briefly, fibroblast or chondrocyte cultures (1 × 106 cells) or cartilage (100 mg) were solubilized in solution D (4 M of guanidine isothiocyanate, 20 mM of sodium acetate, pH 5.2, 0.1 M of 2-mercaptoethanol, and 0.5% N-lauroylsarcosine).(25) One volume of isopropanol was added to this mixture and all proteins and nucleic acids were precipitated at 20°C overnight. After centrifugation, the pellet containing the proteins and nucleic acid was digested with 1 mg/ml of proteinase K (molecular biology grade; Gibco BRL; in 10 mM Tris, 5 mM of EDTA, and 1% sodium dodecyl sulfate [SDS], pH 8.0) for 2 h at 65°C. After digestion, the mixture was extracted with 1 vol of phenol and 0.1 vol of chloroform/alcohol (49:1). The aqueous phase was recovered after centrifugation and precipitated with 1 vol of isopropanol. The RNA and remaining contaminating glycosaminoglycans (GAGs) were recovered by centrifugation. This pellet was washed in 4 M of LiCl,(25) which solubilizes the GAGs but not the RNA. The RNA was recovered by centrifugation, resuspended in solution D, and extracted with phenol/chloroform. Pure total RNA is recovered by precipitating the aqueous phase and washing with 70% ethanol to remove any excess salt. Then, the total RNA pellet was resuspended in diethylpyrocarbonate-treated (DEPC) water and the optical density at 260 nm (OD260) was used to quantitate (OD260 1.0 = 40 μg of RNA) and assess the purity (ratio 260/280 must be ≥1.8) of the preparation. Total RNA was isolated from fibroblasts using the guanidine/acid/phenol/chloroform method.(25)

Reverse transcription

First-strand complementary DNA (cDNA) was prepared by incubating 5 μg of total RNA (from fetal fibroblasts or chondrocytes) with a reaction mixture of 20 μl containing 50 mM of Tris/HCl, pH 8.3; 75 mM of KCl; 3 mM of MgCl2; 10 mM of dithiothreitol (DDT); 50 μM each of deoxyadenosine triphosphate (dATP), deoxyguanosine triphosphate (dGTP), deoxycytosine triphosphate (dCTP), and deoxythymidine triphosphate (dTTP); 0.5 of μg Oligo (dT)12-18; 5 μg of total RNA; and 200 U of SuperScript TMII H Reverse Transcriptase (Gibco BRL) at 42°C for 1 h in a thermal cycler.

Polymerase chain reaction

One microliter of reverse-transcribed total RNA was incubated with 2.5 U of Taq polymerase (Perkin Elmer, Woodbridge, Ontario, Canada) in 50 mM of KCl, 1.5 mM of MgCl2, 10 mM of Tris/HCl, pH 8.3, 400 μM of deoxynucleoside triphosphates (dNTPs), and 0.8 μM each of upstream and downstream oligonucleotide primers in 25-μl reaction tubes. polymerase chain reaction (PCR) amplifications were carried out in a thermal cycler. The PCR protocol involved heating for 1 minute at 95°C and then for 1 minute at 50°C and 5 minutes at 72°C. This sequence was repeated for 30 cycles followed by a 10-minute incubation at 72°C. The oligonucleotides for each cDNA are described in Table 1. Those used for individual PCR studies of MMP-13 are indicated in the Results section. PCR product sizes and amounts were determined by electrophoresis of samples containing 3 μl of a 1-mg/ml solution of ethidium bromide in water in 1.6% agarose gels in 40 mM of Tris, 20 mM of acetic acid, and 1 mM of EDTA. The digital images of the gels were analyzed using National Institutes of Health (NIH) 1.60 software to determine the pixel intensity of each band of the synthesized PCR products. The autobackground subtraction was used to control for the background signal. 18sRNA was used as reference for gel loading. By varying the cycle numbers, the band intensities were determined to be below saturation. PCR products were sequenced (McGill University Central Sequencing Facility, Montreal, Quebec, Canada) to confirm identities.

Table Table 1.. Summary of Oligonucleotide Primers for PCR
original image

Northern blot analysis

Total RNA was fractionated on a 1% agarose/2.2 M formaldehyde gel.(26) Even loading of the RNA was verified by ethidium bromide staining. After electrophoresis, the gel was washed in deionized water to remove excess formaldehyde. RNA was transferred onto Hybond-N (Amersham, Oakville, Ontario, Canada) or N+ (Amersham) membrane by capillary blotting overnight with 20× SSC. The membrane was air-dried and then the RNA was cross-linked onto the membrane by UV radiation (Stratagene cross-linker; Stratagene, La Jolla, CA, USA). cDNA probes were labeled with [32P]dCTP using a random priming kit (Stratagene) and unincorporated radionucleotides were removed by using a gel filtration column (Nick columns; Amersham, Baie D'Urfe, Quebec, Canada). The specific activity of the probe was determined by counting in a Beckman scintillation spectrometer (Packard, Meridan, CT, USA). The membranes were prehybridized in a commercially available hybridization buffer (Quickhyb; Stratagene) for 30 minutes and then hybridized for 2 h at 65°C in hybridization buffer containing 200 μl/ml of denatured salmon sperm DNA and heat-denatured [32P]-labeled probe (specific activity >1 × 108 counts per minute [cpm]/μg) in a hybridization oven. The membranes were washed sequentially twice in 2× SSC at room temperature for 5 minutes, twice in 2× SSC and 1% SDS at room temperature for 15 minutes, and twice in 0.1× SSC and 0.1% SDS at 65°C for 15 minutes. The membranes were dried and exposed to Eastman Kodak Co. X-AR film (Eastman Kodak Co., Rochester, NY, USA) at −70°C with an intensifying screen for varying time periods.

Cloning of bovine MMP-1 and MMP-13

cDNAs of bovine MMP-1 and MMP-13 were obtained by PCR of reversed transcribed messenger RNA (mRNA) from bovine fibroblasts or chondrocytes using oligonucleotide primers (Table 1). MMP-1 primer sequences were selected following sequence alignment of the published human(27) porcine(28) and bovine(29) cDNAs. MMP-13 primer sequences were selected by using conserved regions of the human,(30) rat,(31) and mouse(32) cDNAs.

DNA sequencing

cDNA products were purified by a phenol/chloroform extraction to remove the Taq polymerase, followed by a molecular cut-off spin column (Centricon 3000; Amicon, Inc., Beverly, MA, USA) to remove unused oligonucleotide primers. Purified cDNAs were subcloned into pCR-Script SK(+) vector (Stratagene) according to manufacturers instructions. Purified plasmids containing the cDNA insert of interest were purified from minipreps(22) from Escherichia coli. Plasmid DNA was sequenced by the dideoxy chain termination method(33) using [35S]dATP (Amersham) and a commercial kit (Sequenase II sequencing kit; U.S. Biochemicals Corp., Cleveland, OH, USA). The primers used in the sequencing reactions were either specific oligonucleotides found in the cDNA of interest or the M13 universal sequencing primer (5′-GTAAAACGACGGCCAGT-3′). Samples were denatured by boiling for 5 minutes and electrophoresed on a 6% or 8% polyacrylamide/7 M urea gel. Gels were transferred to Whatman paper, dried for autoradiography, and exposed to Eastman Kodak Co. X-OMAT film (Eastman Kodak Co.).

Preparation of rabbit polyclonal antibodies to collagenases MMP-1 and MMP-13

For generation of polyclonal antisera against MMP-1 and MMP-13, the amino acid sequences of all known collagenases (MMP-1, MMP-8, and MMP-13) from different species were aligned and compared using a computer program, MACAW (National Center for Biotechnology Information, Washington D.C.). Specific sequences were identified for MMP-1 (AIEKAFQLWSNV), corresponding to amino acids 133-144, and for MMP-13 (PNPKHPKTPEK), corresponding to amino acids 275-285. These are unique to each MMP and highly conserved among different species. The specificity of the epitope was compared further with all of the Swiss protein database using the BLAST search program that is available from the National Center for Biotechnology Information. A cysteine residue was added to the amino terminal to both peptides to allow for conjugation to ovalbumin.

For anti-MMP-1 and anti-MMP-13 antisera production, female New Zealand white rabbits weighing 2.5-3.0 kg (Ferme des Chenes Bleus, Inc., Montreal, QC) were immunized initially intramuscularly with 0.5 mg of each peptide sequence conjugated to ovalbumin (OVA; Sigma, St. Louis, MO, USA) in 0.25 ml of phosphate-buffered saline (PBS). The conjugation was performed with an amino terminal spacer arm as described.(15) Peptide conjugates were emulsified in 0.25 ml of Freund's complete adjuvant (Difco, Detroit, MI, USA). Booster injections were given intramuscularly in Freund's incomplete adjuvant (Difco) every 2 weeks. After the second booster, test bleeds were performed and antibody titers were determined by ELISA as described in the following section. After two boosters, the antisera produced good titers and the rabbits were exsanguinated by cardiac puncture and ∼80 ml of serum was obtained from each animal.

Electrophoresis and immunoblotting

For Western blots, 500 μl of conditioned medium from chondrocytes or fibroblasts was precipitated with 2 vol of 100% ethanol. The proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE) as previously described(34) under denaturing conditions using 10% 1-mm-thick 7 cm × 8 cm mini-Protean gels (Bio-Rad Laboratories, Mississauga, ON, Canada). Gels were either stained with Coomassie blue R-250 (Bio-Rad Laboratories) or electrophoretically blotted onto nitrocellulose (Bio-Rad Laboratories) in methanol-free buffer containing 25 mM of Tris and 0.2 M of glycine, pH 8.3, overnight at 4°C. Nitrocellulose membranes were blocked with PBS-3% BSA and incubated either overnight or for 1 h with the appropriate dilution of affinity-purified immunoglobulin G (IgG) at 4°C. Affinity-purified IgG was obtained by using a peptide affinity column. After three 10-minute washes in PBS-1% BSA-1% Tween, the membranes were incubated for 1 h at room temperature with alkaline phosphatase-conjugated goat anti-rabbit IgG (Sigma) diluted 1:30,000 with PBS-3% BSA-1% Tween. The membranes were washed again three times for 10 minutes each in PBS-1% BSA-1% Tween and then rinsed with distilled water. Alkaline phosphatase substrate solution (Bio-Rad Laboratories) was added to visualize the immune complex. After optimal color development (10-20 minutes), the reaction was stopped by washing off the substrate solution with distilled water.


DNA was determined fluorometrically with a modification of a published method using the dye bisbenzimide (Hoechst, Frankfurt, Germany).(35) Briefly, chondrocyte cell layers were digested with proteinase K (0.5 mg/ml) for 24 h at 56°C in 0.1 M of sodium phosphate buffer, pH 6.5, containing 0.01% EDTA. DNA standard curves were prepared from calf thymus DNA (0.1-1.0 mg) in phosphate/EDTA buffer.

Collagen purification

Human fetal cartilage was collected from therapeutic abortions. Human and bovine type II collagen were prepared by pepsin digestion and differential salt precipitation as described.(36)

Collagenase assay

Acid-soluble telopeptide-free human type II collagen was acetylated using14C-acetic anhydride and collagenase activity was assayed essentially as described(37) with some modifications. Briefly,14C-telopeptide-free collagen substrate (specific activity 3.7 × 106 cpm/mg collagen at 3.1 mg/ml, in 50 mM of Tris/HCl, pH 7.6, and 0.3 M of NaCl) was used in 1.5-ml Microtubes (Diamed Lab Supplies, Mississauga, Ontario, Canada). Aminophenylmercuric acetate (APMA; 0.5 mM) was added to activate procollagenase. One millimolar of 1,10-phenanthroline was added to inhibit collagenase and it served as a blank control. After 18 h of incubation at 30°C, the reaction was terminated by the addition of 0.5 mM of EDTA. Collagenase cleavage products were digested further by 35 μg of trypsin (Sigma) and 35 μg of α-chymotrypsin (Sigma) in the presence of 200 μg of bovine type II collagen as a carrier protein for 2 h at 30°C. The undigested type II collagen was precipitated in 10% trichloracetic acid (TCA) at 4°C. After centrifugation, 100 μl of supernatant was added to 1 ml of Aquasol scintillation fluid (New England Nuclear, Boston, MA, USA) and counted in a Beckman scintillation spectrometer (Packard). The percent digestion was calculated as the total counts per minute in the supernatant minus counts per minute in the 1,10-phenanthroline blank divided by original counts in the14C-type II collagen × 100. One unit of collagenase is defined as the amount of collagenase that digests 1 μg of14C- type II collagen in 1 minute at 30°C.

Before detection of enzyme activity in all samples, tissue inhibitor of MMPs (TIMPs) in conditioned media were inactivated by reduction in 2 mM of DTT at 37°C for 30 minutes. To prevent spontaneous reactivation of TIMP, the reduced TIMP was alkylated with 5 mM of iodoacetamide at 37°C for 30 minutes as described.(38)

Determination of matrix mineralization using calcium (45Ca2+) incorporation

Prior studies with this culture system have revealed that the formation of hydroxyapatite is accompanied by an increase in uptake of45Ca2+.(20) Chondrocyte cell layers were labeled(20,22) at every medium change (every 2 days) with 1.25 μCi/ml of45Ca2+ (Amersham). On completion of the culture period, the cell layers were collected, washed twice in DMEM alone, and dried under vacuum for 1 h. Then, the cell layers were solubilized at 70°C by the addition of 1 ml/well of 90% formic acid for 45 minutes before liquid scintillation counting.

Collagen biosynthesis and general protein turnover

To assess collagen biosynthesis in chondrocyte cell layers, [3H]proline (25 μCi/ml; Amersham) was added to cultures at every medium change (every 2 days) in the presence of 70 μg/ml of β-aminopropionitrile. Collagens from the media were precipitated by the addition of (NH4)2SO4 to a final concentration of 33% (of saturation) for 24 h at 4°C. After centrifugation, pellets were washed twice with 70% ethanol. General protein synthesis and release from cultures was examined in cultures in methoin-free DMEM (Gibco BRL) supplemented with(35)S-methionine (50 μCi/ml) (Amersham) for the times indicated using the same methods. The dry pellets were analyzed directly by SDS-PAGE using 10% gels followed by fluorography.(20,22)

Determination of type II collagen denaturation

Type II collagen degradation in bovine fetal growth plate chondrocyte cell layers was measured using the COL2-3/4m ELISA.(39) This assay measures an intrachain epitope that arises from the unwinding of the triple helical collagen.

Histological analysis of chondrocyte cell layers

Chondrocyte cell layers were fixed in 2% glutaraldehyde in 0.1 M of sodium cacodylate, pH 7.2, overnight at 4°C. Tissue was dehydrated with graded ethanols and embedded in Spurr resin (Polysciences, Inc., Washington, PA, USA). Semithin sections (1-2 μm) were stained with Von Kossa's reagent (for mineral) or with toluidine blue.

Statistical analyses

Student's t-test was used. A value of p ≤ 0.05 was considered significant. Experiments were repeated at least once with similar results.


Cloning of bovine MMP-13

To develop reagents to study both MMP-1 and MMP-13 mRNA expression and protein synthesis, bovine MMP-1 and MMP-13 were cloned. Full-length bovine MMP-13 and MMP-1 cDNAs were obtained by PCR using IL-1-treated fibroblasts (MMP-1) or bovine hypertrophic chondrocytes (MMP-13). The complete coding sequence was obtained by DNA sequencing. The published sequence for bovine MMP-1(29) was confirmed. The complete cDNA sequence and deduced amino acid sequence for MMP-13 is shown in Fig. 1 (GenBank accession number AF072685). Sequence alignment of MMP-13 cDNAs showed bovine MMP-13 to have 92% sequence identity to human MMP-13 cDNA and 82% sequence identity to mouse and rat MMP-13. Sequence alignment of MMP-13 protein sequences showed that the bovine protein had 90% homology to human protein and 85% to mouse and rat protein. Comparison of bovine MMP-1 and MMP-13 sequences showed only a 28% homology. Comparison of bovine MMP-1 and MMP-13 protein sequences showed 48% homology.

Figure FIG. 1..

Complete coding sequence and deduced amino acid sequence for bovine MMP-13.

Expression of MMP-1 and MMP-13 in fibroblast and growth plate chondrocyte cultures analyzed by reverse transcription (RT)-PCR

Expression of MMP-1 (primer pair BTMMP1S5 and BTMMP1S3, Table 1) was clearly detectable in fibroblasts cultured with IL-1 (Fig. 2, lane 4). In contrast, it could not be detected in growth plate chondrocyte cultures, either prehypertrophic or hypertrophic (Fig. 2, lane 5), or in those chondrocytes cultured with IL-1α (data not shown). In contrast, MMP-13 (primer pair BTMMP13S5 and BTMMP13S3, Table 1) was clearly expressed by hypertrophic chondrocytes (Fig. 2, lane 7) but not by fibroblasts ± IL-1 (Fig. 2, lane 6). Hypertrophic chondrocytes (Fig. 2, lane 3) but not fibroblasts (Fig. 2, lane 2) expressed type X collagen (primer pair BTTYPEX5 and BTTYPEX3, Table 1).

Figure FIG. 2..

Expression of MMP-13 is restricted to hypertrophic chondrocytes and is not expressed in fibroblasts stimulated by IL-1 (fetal age, 206 days). Bovine fetal fibroblasts stimulated with IL-1α are present in lanes 2, 4, and 6. Hypertrophic chondrocytes (subpopulation C, days 16-18 in culture) are present in lanes 3, 5, and 7. Lane 1 contains a 100-basepair (bp) DNA ladder. PCR products were analyzed on a Tris/acetic acid/EDTA 1% agarose gel. MMP-1 mRNA expression is only detectable in IL-1α-stimulated fetal bovine skin fibroblasts (1 ng/ml for 24 h). Type X collagen and MMP-13 mRNA are expressed only in hypertrophic chondrocytes. PCR product sizes were as predicted by the primers shown in Table 1.

Expression of MMP-13 is up-regulated in hypertrophic chondrocytes

Northern analyses

Total RNA from cultured chondrocytes was probed using MMP-13 cDNA during the culture of subpopulation C cells. Expression was not detected until days 12-14, which coincided with the initiation of synthesis of type X collagen (Fig. 3A). Expression of MMP-13 increased at days 16-18 and was maintained at days 18-20. Two transcripts (sizes 2.8 kilobase [kb] and 2.0 kb) were detected, consistent with previous reports for human MMP-13 mRNA from human cartilage.(14)

Figure FIG. 3..

(A) Northern blot of MMP-13 expression during maturation of subpopulation C (fetal age, 206 days). (B) The ethidium bromide-stained gel is shown also to establish RNA loading, together with time of initiation of type X collagen synthesis. Five micrograms of RNA was loaded per lane. The membrane was probed with [32P]-labeled MMP-13 (291 basepairs [bp]) cDNA and exposed for 24 h.

Western blotting

Analyses of conditioned media from subpopulation C cells revealed an MMP-13 protein as a band of 65 kDa, which was first clearly detected on days 16-18 (Fig. 4) after initiation of type X collagen synthesis (days 12-14). This was increased in content on days 18-20. There also was an immunoreactive lower molecular weight species (∼ 30 kDa) seen at days 18-20 which probably represents an autolytic fragment of MMP-13. Blotting for MMP-1 produced negative results although it was detectable in IL-1-treated fibroblast cultures (data not shown).

Figure FIG. 4..

Western blot of MMP-13 protein during maturation of subpopulation C (fetal age, 206 days). Conditioned medium was collected every 2 days from growth plate chondrocyte cultures. Conditioned media (500 μl) were precipitated with 2 vol of 100% ethanol and were analyzed by SDS-PAGE. Proteins were blotted onto nitrocellulose and probed with anti-MMP-13 affinity-purified IgG diluted 1/250. MMP-13 was detected only on days 16-18 and days 18-20. The initiation of type X collagen synthesis is indicated (12-14 days). Molecular weight marks (kDa) are shown.

Collagenase activity

This was first detected on days 12-14 in which the majority was latent and activatable by APMA. Total and active enzyme increased up to days 18-20 with the preponderance shifting to active enzyme (Fig. 5). It is clear from a comparison of Figs. 4 and 5 that collagenase activity can be detected with much more sensitivity than the detection of MMP-13 as protein by Western blotting.

Figure FIG. 5..

Collagenase activity and activation increase during maturation of subpopulation C (fetal age, 206 days). Conditioned medium was collected every 2 days. Conditioned media (100 μl) were added to14C-labeled human type II collagen and assayed for collagenase activity. (1 U = 1 μg14C collagen digested/minute), which was detected at the same time as type X collagen synthesis and MMP-13 protein detection (Fig. 4). Results for total activity (+; APMA activated) and active enzyme (−; nonactivated) media are shown. Data points represent the mean ± SD of triplicate wells.

Effects of MMP inhibitor

Arrest of hypertrophy

Collagen synthesis was examined by analysis of the incorporation of3H-proline into type II and type X collagens. Type II collagen synthesis was observed at all times. However, synthesis of type X collagen was observed first in control cultures on days 14-16 (Fig. 6). Previously, we have established the identity of the type X collagen band in these cultures.(20) In the presence of either 1 nM or 10 nM of RS102481, there was no obvious effect on type II collagen synthesis. However, type X synthesis was never observed even up to days 20-22 when the experiment was terminated (Fig. 6). Expression of MMP-13, (primer pair BMMP13D and BMMP13R, Table 1), Cbfa1 (Table 1), and type X collagen (primer pair BCOLXD and BCOLXR, Table 1) mRNA determined using RT-PCR revealed that the MMP-13 mRNA was elevated on day 10 over days 2, 4, 12, and 16 and was expressed maximally in association with expression of the hypertrophic phenotype, and after the transient increase in Cbfa1 and type X collagen mRNA (Fig. 7). The presence of the inhibitor at 10 nM inhibited these increases in MMP-13, Cbfa1, and type X collagen mRNA expression associated with hypertrophy (Fig. 7). PCR product sizes were as predicted by the primer sequences. The differences observed between the times of maximal expression of MMP-13seen here compared with Fig. 3 are caused by differences in fetal age, which influence the maturation of chondrocytes observed in any one experiment.

Figure FIG. 6..

The inhibition of type X collagen synthesis by RS 102,481 in subpopulation C during maturation (fetal age, 221 days). [3H]proline-labeled collagens were precipitated from conditioned medium (500 μl) every 2 days. Collagens were analyzed by SDS-PAGE and visualized by autoradiography. At 1 nM and 10 nM RS 102,481, inhibited type X collagen synthesis in C subpopulation cells as compared with control. Type II collagen synthesis was unaffected.

Figure FIG. 7..

The inhibitory effect of RS 102,481 (10 nM) on mRNA expression during maturation of subpopulation C (fetal age, 211 days). MMP-13, Cbfa1, and type X collagen expression were analyzed by RT-PCR. Ethidium bromide-stained 1.6% agarose gel. The bands show the amount of PCR product synthesized at a single representative dilution of the sample cDNA after amplification for 30 cycles in all cases. Molecular weight standards (base pairs) are shown. PCR product sizes were as predicted by the primers described in the text and in Table 1.

Total protein synthesis

Although there was no evidence of toxicity as indicated by the lack of effect on type II collagen synthesis, we determined whether overall protein synthesis, reflected by [35S]methionine incorporation, was affected. Culture media from chondrocytes cultured with and without the inhibitor were examined by SDS-PAGE and autoradiography. There was no detectable effect on protein synthesis as reflected by release of proteins that had incorporated [35S]methionine at 1 nM and 10 nM (data not shown), indicating a lack of evidence for a general toxic effect of the inhibitor at either concentration, or for a general effect on the release of newly synthesized proteins in these cultures.

Matrix assembly, degradation, and45Ca2+ incorporation

As chondrocytes matured and expressed type X collagen on days 14-16 there was a progressive increase in type II collagen degradation (as measured by denaturation) initiated on days 8-11 (Fig. 8A), a progressive increase in type II collagen content (Fig. 8B), and onset of incorporation of45Ca2+ on days 14-16 (Fig. 8D). At both doses, the inhibitor suppressed the increase in collagen denaturation (reflective of its degradation; Fig. 8A), suppressed the increases in total type II collagen content (Fig. 8B), and also markedly suppressed the increase in45Ca2+ incorporation (Fig. 9C).

Figure FIG. 8..

The inhibitory effects of RS 102,481 on matrix assembly (fetal age, 221 days). Type II collagen (A) denaturation, (B) content, and (C)45Ca2+ incorporation were measured in the presence and absence of the inhibitor at 1 nM and 10 nM. There was a reduction in collagen denaturation and content in the presence of the inhibitor.45Ca2+ incorporation was inhibited also (fetal age, 221 days).

Figure FIG. 9..

Mineralization in chondrocyte C subpopulation cell layers cultured with or without RS 102,481 (fetal age, 221 days). (A, C, and E) Chondrocyte morphology (toluidine blue) and (B, D, and F) mineral (Von Kossa's stain) in the extracellular matrix of cell layers grown in the presence of (A and B) 1 nM, (C and D) 10 nM, (E and F) or absence (control) of RS 102,481 inhibitor. C subpopulation cells have enlarged after 20 days in culture. There was no significant difference in cell size in cells grown in the (A-D) presence or (E and F) absence of inhibitor. However, there is no matrix mineralization detectable in cultures containing (A-D) either 1 nM or 10 nM of RS 102,481 as compared with (E and F) control. (E and F) Arrowheads indicate areas of mineral in control cultures (bar = 50 μm).

There was never any evidence for the presence of the ¾ and ¼ fragments generated by initial collagenase cleavage of type II collagen. This probably is because of rapid secondary proteolysis of these cleaved denatured α-chains by MMP-13 (an excellent gelatinase) and other MMPs.

Evidence for mineral deposition in chondrocyte cell layers was examined using von Kossa stain. On day 22 (Fig. 9), there were no pronounced differences in cell size between the control cells and cells cultured in the presence of the inhibitor at 1 nM or 10 nM. However, extracellular staining for mineral in the control cells was observed with Von Kossa's reagent. Toluidine blue staining for cell morphology also revealed similar extracellular staining. This staining was absent in cultures with the inhibitor. In previous studies, we showed that the formation of hydroxyapatite (characterized by X-ray diffraction) corresponded to the increased incorporation of45Ca2+ in association with expression of type X collagen. (20)


MMP-13 and not MMP-1 was detected in these chondrocyte cultures, confirming recent human observations using in situ hybridization.(40,41) In agreement with these human studies and those of the mouse,(11,42,43) the expression and synthesis of MMP-13 mRNA was up-regulated on chondrocyte differentiation into hypertrophic cells. When type X collagen was synthesized, Western blotting revealed the detection of pro-MMP-13 (65 kDa). Although the active form of MMP-13 at 55 kDa was not observed, a proteolytic fragment of MMP-13 at 30 kDa was detectable. This is the same as the kind described for MMP-1(44) and suggests rapid autodegradation of MMP-13. The fragment also reflects the activation of the proenzyme that was shown by the measurements of the latent and active enzyme, which were clearly detectable in conditioned medium at hypertrophy. Total activity and activation of pro-MMP-13 increased with time after type X collagen synthesis, resulting eventually in almost complete activation of the proenzyme.

The synthesis and activity of MMP-13 by hypertrophic chondrocytes was associated with increased degradation of type II collagen, as observed in the hypertrophic zone in situ(4) and on hypertrophy in culture.(5) Inhibition of matrix degradation by RS 102,481 not only arrested collagen II degradation (measured as denaturation), but also had a profound inhibitory effect on chondrocyte differentiation as revealed by the suppression of type X collagen and Cbfa1 gene expression, type X collagen synthesis, and suppression of45Ca2+ incorporation, which normally is increased markedly at this time. At the concentrations we used, RS 102,481 also can inhibit membrane type 1 (MT1)-MMP (MMP-14), which has been reported to have collagenase activity(45) and to be capable of activating MMP-13.(46) However, MMP-14 has not been found in human growth plate chondrocytes.(41) Moreover, MMP-14 null mice(47) do not exhibit impaired cleavage of type II collagen by collagenase within growth plates (A.R. Poole, I. Pidoux, H. Birkedahl-Hansen, and K. Holmbeck, unpublished data, 2001). Thus, the arrest of chondrocyte differentiation and incorporation of45Ca2+ by this inhibitor indicates that the activity of MMPs, including the collagenase MMP-13, is required for chondrocyte differentiation, the latter being associated in situ in the growth plate with matrix mineralization. It also suggests that MMP activity serves to autoregulate MMP-13 expression and that without efficient matrix degradation, chondrocyte differentiation and matrix calcification is suppressed. Cbfa1 is an MMP-13 transcription factor.(9,10) In growth plates of mice in which Cbfa1 is not expressed, MMP-13 also is not expressed. This is associated with impaired chondrocyte maturation and matrix mineralization in endochondral development.(11,12) Also, overexpression of Cbfa1 causes acceleration of endochondral ossification because of precocious chondrocyte differentiation maturation in vivo.(48) In vitro this overexpression stimulates chondrocyte maturation.(49) Together, these and our observations indicate that the Cbfa/MMP-13 pathway plays an important role in chondrocyte differentiation as well as in matrix turnover.

Recently, we have shown that degradation products of type II collagen can induce expression and activation of MMP-13 and MMP-1 in articular chondrocytes (M. Kobayashi, T. Yasuda, and A.R. Poole, unpublished data, 2001). Thus, it appears that inhibition of MMP-13, Cbfa1, and type X collagen gene expression by the collagenase inhibitor may be because of arrest of generation of collagen II degradation product(s) of type II collagen as a result of inhibition of MMP activity, preventing this positive feedback loop from operating.

This inhibition of an MMP(s) leading to the arrest of its gene expression and chondrocyte differentiation is reminiscent of another study in which doxycycline was shown to inhibit chondrocyte differentiation (onset of type X collagen synthesis) of chick hypertrophic chondrocytes.(50,51) Doxycycline can inhibit collagenase and gelatinase activity in articular cartilage(52) leading to down-regulation of mRNA expression of MMP-13 as well as MMP-1 and MMP-8.(53)

Together, these observations suggest that the expression and extracellular activity of MMPs such as MMP-13 are integrated carefully as a result of cell-extracellular matrix interactions and that MMP-mediated matrix resorption involving MMP-13 is required for chondrocyte differentiation, which results in matrix mineralization in growth and development.


This study was funded by the Shriners Hospitals for Children, the Canadian institutes of Health Research, Canadian Arthritis Network, and the National Institute for Aging, National Institutes of Health (to A.R.P.).