Involvement of the notch pathway in the regulation of matrix metalloproteinase 13 and the dedifferentiation of articular chondrocytes in murine cartilage




To demonstrate the activation of the Notch signaling pathway during changes in the phenotype of chondrocytes in vitro, and to assess the influence of Notch on the production of chondrocyte markers.


Serial monolayer primary cultures of murine articular chondrocytes (MACs), as a model of chondrocyte dedifferentiation, were prepared. MACs were cultured with or without a Notch inhibitor and transfected with different Notch-expressing vectors. The Notch pathway and chondrocyte marker profiles were assessed by quantitative reverse transcription–polymerase chain reaction, immunoblotting, and immunocytochemistry.


Successive passages of MACs resulted in a loss of type II collagen and aggrecan (chondrocyte differentiation markers), an increase in type I collagen (dedifferentiation marker), an increase in Notch ligands, and augmented target gene activity. The Notch inhibitor decreased the type II collagen protein content but had no effect on Col2a1 messenger RNA, while transfection with the constitutive active forms of the Notch1 receptor led to a decrease in type II collagen in transfected cells. In assays to investigate the mechanism of type II collagen breakdown, matrix metalloproteinase 13 (MMP-13) synthesis was regulated in a Notch-dependent manner, whereas MMP-2 synthesis was unchanged.


The Notch signaling pathway is associated with decreased type II collagen production during the dedifferentiation of MACs in vitro. This may be correlated with the increase in MMP-13 production linked to activation of Notch.

Bone growth is a complex process that results from the time-dependent activation or inactivation of genes, including those encoding transcription factors and signaling proteins involved in chondrocyte differentiation. Osteoarthritis (OA), a degenerative disease leading to cartilage breakdown and changes in subchondral bone, can be attributed, at least in part, to the dysregulation of these factors (1). Although the breakdown of cartilage that occurs in OA has long been considered to be the result of an irreversible process, it is now clear that chondrocytes respond to the beginning of the process by attempting to repair the cartilage matrix. Since chondrocytes are the only cells that synthesize type II collagen and aggrecans, the 2 main constituents of cartilage matrix, a challenging approach has been to treat cartilage defects with chondrocytes. Although this approach has been quite successful in the treatment of cartilage defects resulting from trauma (2), there are still problems associated with the treatment of cartilage defects in OA. A major issue is that the phenotype of articular chondrocytes is altered in OA or in other processes in which chondrocytes are transplanted into a pathologic environment. A better understanding of the mechanisms involved in chondrocyte dedifferentiation should help to define the best conditions for a cell-based treatment of OA.

Unfortunately, chondrocytes respond to adverse environmental stimuli by promoting cartilage breakdown and preventing cartilage repair. Sandell and Aigner (3) described 5 patterns of chondrocyte reactions to OA development: 1) modulation of the articular chondrocyte phenotype, 2) osteophyte formation, 3) proliferation and apoptosis of chondrocytes, 4) matrix synthesis, and 5) matrix degradation. Whereas many studies have focused on chondrocyte differentiation during embryonic development (1, 4), the changes to the chondrocyte phenotype that occur in OA have received much less attention.

Differentiated chondrocytes secrete a variety of proteins involved in maintaining the capacity of the cartilage matrix to counteract joint stress due to such things as overloading or overuse. The unique biomechanical properties of cartilage rely mainly on the functions of type II collagen and aggrecan (high molecular weight proteoglycans). The characteristic changes in molecular markers during OA (3, 5) have been described as a reversion of the articular chondrocyte phenotype (6), with a decreased ratio of type II collagen to type I collagen being characteristic of chondrocyte differentiation (7). The loss of type II collagen and/or aggrecan in OA chondrocytes is also due to the synthesis of matrix-degrading enzymes, of which a major one is matrix metalloproteinase 13 (MMP-13; collagenase 3). MMP-13 is involved in type II collagen degradation (8, 9) and is abundant in both experimental OA (10) and human OA chondrocytes (11). Markers of the chondrocyte phenotype decrease when cells are cultured at low density and/or on plastic, or after passaging. This altered phenotype results in the loss of type II collagen and aggrecan (12, 13).

Notch receptors are involved in determining an alternative cell fate in both developing and adult invertebrates and vertebrates. The Notch pathway has been identified in chondrocytes, and it has been found to be involved in regulating articular cartilage development (14) and inhibiting chondrocyte differentiation (15–17). A recent report described a subset of articular chondrocytes containing Notch1 that were expressed at the surface of articular cartilage and were thought to be progenitor cells (18). The Notch receptor is a type I transmembrane molecule that is located at the cell surface as a heterodimer. Classically, Notch signaling occurs after the binding of a transmembrane ligand through cell–cell contact. These ligands belong to the Delta-like or Jagged families of Notch ligands in mammals. Ligand binding elicits the proteolytic release of the intracellular domain (ICD) of the Notch receptor via 2 sequential proteolysis events. In mammals, the Notch receptor is cleaved, first in its ectodomain and then in its transmembrane domain, by γ-secretase. The ICD is then translocated to the nucleus, where it participates in the activation of target genes (19). The present report describes the role of the Notch pathway in modulating the phenotype of murine articular chondrocytes (MACs) in vitro and evaluates the influence of the Notch pathway in the regulation of proteins involved in matrix degradation.



All reagents were purchased from Sigma-Aldrich (St. Quentin Fallavier, France), unless stated otherwise. Cell culture medium and fetal calf serum (FCS) were obtained from Gibco (Cergy Pontoise, France). Collagenase D and the Complete protease inhibitor mixture were obtained from Roche Diagnostics (Meylan, France). The γ-secretase inhibitor N-[N-(3,5-difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester (DAPT) was obtained from Calbiochem (La Jolla, CA).


Polyclonal rabbit anti–Notch1 ICD (V1744) was obtained from Upstate Cell Signaling Solutions (Milton Keynes, UK). Anti–type II collagen polyclonal antibodies, anti–MMP-2 antibodies, and anti–MMP-13 antibodies were from Tébu for Santa Cruz (Le Perray en Yvelines, France). The enhanced chemiluminescence (ECL) Western blot analysis system was purchased from Pharmacia Biotech for Amersham (Orsay, France). The Immuno-Blot polyvinylidene difluoride (PVDF) Immobilon-P membranes for Western blotting were obtained from Millipore (Molnstein, France), and Precision Plus protein standards were obtained from Bio-Rad (Ivry sur Seine, France).

Sample collection and cell culture.

All experiments were performed according to protocols approved by the French/European institutional ethics committees. Immature MACs were collected as previously described (12). Briefly, femoral heads and tibial plateaus from 12-pup litters (Swiss mice, ages 5–6 days) were dissected in sterile conditions and incubated twice, for 45 minutes each, with collagenase D (3 mg/ml in Dulbecco's modified Eagle's medium [DMEM]) at 37°C in 5% CO2. All soft tissues were removed, and the cartilage pieces were incubated overnight with collagenase D (0.5 mg/ml) at 37°C. The resulting chondrocytes were suspended in culture medium (DMEM–Glutamax [1 gm/liter glucose]; Gibco) plus 10% FCS and antibiotics. The chondrocytes were then seeded in 10-cm dishes at 5 × 105 cells/dish (considered baseline; designated passage 0 [P0]). Thereafter, the MACs were removed with trypsin just before they reached confluence, and were again seeded in 10-cm dishes at 5 × 105 cells/dish (designated P1, P2, and P3 according to the time to reach confluence, with P0 and P1 reaching confluence in 5 days, and P2 and P3 reaching confluence after ∼7 days, for a total time in culture of ∼24 days).

Real-time quantitative reverse transcription–polymerase chain reaction (RT-PCR) assays.

Total RNA was extracted from the MACs using the RNeasy kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Concentrations were determined spectrophotometrically. RT was performed on 1 μg total RNA, using the OmniScript RT kit (Qiagen). Two RTs were performed for each RNA sample to reduce intraexperiment variations. Quantitative PCR was performed using the LightCycler LC480 (Roche Diagnostics). The PCR mix included 5 μl of each reverse transcriptase (diluted 1:25) and 300 nM of each primer in 1× LightCycler DNA SYBR Green 1 Master Mix. Specific primers for complementary DNA (cDNA) were chosen with the LightCycler Probe Design2 program according to European Molecular Biology Laboratory accession numbers: for the housekeeping gene HPRT, 5′-AGGACCTCTCGAAG TGT-3′ and 5′-ATTCAAATCCCTGAAGTACTCAT-3′; for the aggrecan gene, 5′-CAGAGTTAGTGGAGGGTGTGA-3′ and 5′-AGACCCTGGGAAGTTTGT-3′; for the type I collagen gene Col1a2, 5′-TCTGTGCCTCAGAAGAACT-3′ and 5′-GAGCCCTCGCTTCCGTA-3′; for the type II collagen gene Col2a1, 5′-GGCAACAGCAGGTTCACATA-3′ and 5′-ATGGGTGCGATGTCAATAAT-3′; for the basic helix-loop-helix homologous of enhancer of split 1 gene Hes1, 5′-CTCCTGACGGCCAATTT-3′ and 5′-AAGGTGACACTGCGTTAG-3′; for Notch1, 5′-TCCTTCCCAGCAC-AGTTA-3′ and 5′-CCTCGGACCAATCAGAGA-3′; for Notch3, 5′-CTCTCAGACTGGTCTGACTCAA-3′ and 5′-GGAGGGAGGGAACAGATATG-3′; for Delta-like1, 5′-AGCAGCTTTAAGGTCCG-3′ and 5′-TGTGACTGGCACTTGGT-3′; for Jagged1, 5′-CTGGTAGACAGAGAGGAGAAG-3′ and 5′-TACGATGTATTCCATCCGGTT-3′; for MMP2, 5′-GATGCTGCCTTTAACTGGAGTA-3′ and 5′-GGAGTCTGCGATGAGCTT-3′; and for MMP13, 5′-TGATGGCACTGCTGACATCAT-3′ and 5′-TGTAGCCTT-TGGAACTGCTT-3′.

The PCRs were performed using the following thermal settings: denaturation and enzyme activation at 95°C for 8 minutes, with cycling at 95°C for 10 seconds, 64°C for 10 seconds, and 72°C for 8 seconds. Amplification was followed up online, and the PCRs were stopped after the logarithmic phase. Melting curve analyses were also performed after PCR to check the reaction specificity. Controls and water blanks were included in each run; these were negative in all cases.

The amount of each target messenger RNA (mRNA) relative to the amount of mRNA for a housekeeping reference gene, HPRT, was estimated in the logarithmic phase of the PCR, and serial dilutions were used to determine the fit coefficients of the relative standard curve. The PCR efficiencies for each of the targets were similar, so that individual cultures could be compared.

Plasmid preparation and transient transfection.

The vectors expressing Notch1 (N1ΔE and N1ICD) were generously donated by Dr. R. Kopan (School of Medicine, Washington University, St. Louis, MO). The expression vector for the dominant-negative form of the Notch1 transcriptional coactivator CSL (CSL-DN) was generously donated by Dr. A. Israël (Pasteur Institute, Paris, France). To prepare the plasmids for transient transfection, the Notch1 cDNA sequences corresponding to residues 1702–2194 (N1ΔE) or 1810–2194 (N1ICD) were cloned into the pCS2 expression vector. The CSL-DN cDNA sequence, carrying a dominant-negative mutation, was cloned into the pSG5 expression vector. Mock plasmids corresponded to pCS2 or pSG5 empty vectors. Plasmids were prepared using the QIAfilter Maxiprep kit according to the manufacturer's instructions (Qiagen).

When the cells had reached 70–80% confluence, they were transfected using Fugen HD reagent (Roche Diagnostics) according to the manufacturer's instructions. The cells were transfected with various expression constructs by adding 10 μg DNA and 30 μl Fugen HD (ratio 1:3) per 10-cm dish. Cells were harvested 24–48 hours posttransfection and used for real-time PCR or Western blot analyses. The transfection efficiency was evaluated by microscopy using a cytomegalovirus–green fluorescent protein (GFP) vector. Approximately 25% of the cells were transfected; 10% of the cells had high GFP production, and 15% had low GFP production.

Protein extraction and Western blot analysis.

Cells were grown to confluence in 10-cm dishes without or with DAPT (2.3 μM, changed every 2 days). The cells were then washed with ice-cold phosphate buffered saline (PBS) and lysed in cold lysis buffer (20 mM Tris HCl, pH 7.6, 150 mM NaCl, 2 mM EDTA, 1% Triton, and 10% glycerol) plus Complete Protease Inhibitor mixture (Roche Diagnostics) and 100 mM NaF and 10 mM Na4P2O7. The lysate was centrifuged at 13,000g for 10 minutes at 4°C. The protein concentrations in the supernatants were determined using the Bio-Rad Bradford Protein Assay (Perbio Science-Pierce, Bezons, France). Aliquots of cell lysates (∼20 μg protein) were analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis on 8% resolving gels or 4–12% gradient gels (Bio-Rad criterion system), followed by transfer to PVDF membranes. The free binding sites on membranes were blocked with Tris buffered saline containing 0.1% Tween 20 (TBS-T) (20 mM Tris HCl, pH 7.5, 100 mM NaCl, 0.1% Tween 20) and 5% fat-free milk, for 1 hour at room temperature.

The membranes were then incubated with the primary antibodies (in TBS-T with 1% milk or, for cultures with V1744 antibody, in TBS-T with 5% bovine serum albumin) overnight at 4°C. The membranes were then washed in TBS-T and incubated with horseradish peroxidase–conjugated secondary antibodies for 1 hour at room temperature. The membranes were then washed repeatedly with TBS-T. Signals were detected with the ECL detection system and exposed to Biomax MR1 films (Eastman Kodak, Rochester, NY) or Fujifilm LAS-300 (Fujifilm Medical Systems, Stamford, CT). Equal protein loading and transfer efficiency were determined by Ponceau Red staining after electrotransfer and β-actin detection. The densitometry patterns were analyzed with ImageGauge software (Science Lab 2004; Fujifilm) and normalized to the density of β-actin.


MACs were seeded directly onto glass coverslips (3.5 × 104 cells/coverslip) that were placed in 24-well culture plates. When the cells had reached 70–80% confluence, they were rinsed twice with PBS, fixed in 4% paraformaldehyde (for 10 minutes at room temperature), and incubated with the primary antibody overnight at 4°C. The MACs were then washed twice with PBS and incubated for 2–4 hours at room temperature with a fluorescein isothiocyanate–conjugated secondary antibody, diluted 1:100 in PBS. The cells were again washed in PBS (3 times), and the coverslips were mounted with Dako fluorescent mounting medium (Dako, Carpinteria, CA). Finally, the cells were examined for immunofluorescence using a Nikon Diaphot 300 microscope equipped with a mercury lamp (Nikon, Tokyo, Japan).

Statistical analysis.

All real-time PCR data are reported as the mean ± SEM. All experiments were performed in triplicate (n ≥ 3 samples per plate). The mean values were compared between groups with the Welch's unpaired, corrected t-test (for each passage of cultures with DAPT versus each passage without DAPT). One-way analysis of variance with a post hoc Bonferroni's test was used to compare the mean values among several groups (P1, P2, or P3 versus P0).


Changes in the chondrocyte phenotype of passaged MACs.

Confluent chondrocytes in P0 cultures had the typical morphologic characteristics of articular chondrocytes, including a rounded or polygonal shape with granular cytoplasm. Serial subculturing of cells from P0 to P3 resulted in flattened, fibroblast-like cells (results not shown); these cultures were used to assay chondrogenic markers.

The relative amounts of mRNA encoding the genes for aggrecan, type I collagen (Col1a2), and type II collagen (Col2a1) were assessed by quantitative RT-PCR (Figures 1A–C). The amount of mRNA encoding the aggrecan and type II collagen genes decreased 20-fold and 5-fold, respectively, from P0 to P3 (P < 0.001) (Figures 1A and C), whereas the amount of type I collagen mRNA remained unchanged (Figure 1B). The ratio of the extracellular matrix components that are commonly used for monitoring chondrocyte differentiation (Col2a1:Col1a2) also decreased (Figure 1D). This decrease in synthesis of chondrogenic markers from P0 to P3 indicates that these MAC cultures can be used as an in vitro model to study the molecular changes occurring during modifications in the chondrocyte phenotype.

Figure 1.

Chondrogenic markers in primary cultures of murine articular chondrocytes (MACs). The effects of passaging of MACs from baseline (P0) up to passage 3 (P3) on the production of mRNA encoding the genes for aggrecan (A), type II collagen (B), and type I collagen (C) were determined by quantitative reverse transcription–polymerase chain reaction on a Roche Diagnostics LightCycler LC480. In addition, the ratio of Col2a1 mRNA to Col1a2 mRNA was determined (D). As a reference, values were normalized to the levels of HPRT mRNA, which remained constant during passaging (results not shown). Bars show the mean and SEM fold induction of mRNA during passaging in cultures (n = 4 per plate) from 4 independent experiments, with each experiment assessing a pool of 2–4 plates in triplicate. ∗∗∗ = P < 0.001 versus control (P0). NS = not significant.

Notch signaling pathway in successive MAC monolayer cultures.

We assessed changes in the activity of the Notch signaling pathway during chondrocyte dedifferentiation by assaying the mRNA encoding the repressor-type gene Hes1, a major target of the Notch signaling pathway (20) that is important for maintaining the differentiated state of cells (21). The amount of Hes1 mRNA increased from P0 to P1 and remained high thereafter (maximum increase of 3.5-fold from P0 to P3; P < 0.01) (Figure 2A). MACs cultured in the presence of the γ-secretase inhibitor DAPT, which binds to the active site of presenilin in the γ-secretase complex (22) and thus inhibits Notch activation (23, 24), had significantly less Hes1 mRNA at P1 (decrease of 60%) as compared with MACs at P1 cultured without DAPT (P < 0.05) (Figure 2A).

Figure 2.

Components of the Notch signaling pathway in primary cultures of murine articular chondrocytes (MACs). The effects of passaging of MACs from baseline (P0) up to passage 3 (P3) on mRNA production of the Notch target genes Hes1 (A) and Notch3 (B), without or with incubation with N-[N-(3,5-difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester (DAPT), were determined by quantitative reverse transcription–polymerase chain reaction (RT-PCR) on a Roche Diagnostics LightCycler LC480. Notch1 receptor mRNA and Notch ligand (Delta-like1 and Jagged1) mRNA were also monitored by quantitative RT-PCR (C). As a reference, results were normalized to the levels of HPRT mRNA. Bars show the mean and SEM fold induction of mRNA during passaging in cultures (n = 3 per plate) from 3 independent experiments, with each experiment assessing a pool of 2–4 plates in triplicate. For cultures without DAPT versus those with DAPT at each passage, # = P < 0.05, ## = P < 0.01, and ### = P < 0.001. For cultures without DAPT at different passages versus the equivalent culture at P0, ∗ = P < 0.05 and ∗∗ = P < 0.01. NS = not significant.

The activity of the Notch-3 gene, a target of Notch-1 (25), increased during passaging, with a maximum increase in Notch3 mRNA of 3-fold at P2 (P < 0.01 versus P0) (Figure 2B). This increase in Notch3 mRNA was significantly inhibited by DAPT (decrease of 42% at P3; P < 0.05 versus cultures without DAPT at P3) (Figure 2B).

Last, the amount of Notch1 mRNA did not vary significantly during passaging, while the amounts of Notch ligand mRNA greatly increased from P0 to P3 (4.9-fold for Delta-like1 mRNA at P3 and 4.2-fold for Jagged1 mRNA at P3; P < 0.01 versus P0 for each) (Figure 2C). The activation of Notch target genes and the increased levels of Notch ligand mRNA suggest that the Notch pathway can be implicated in changes in the chondrocyte phenotype in vitro.

Effect of inhibition of γ-secretase on type II collagen synthesis.

We studied the effect of inactivating the Notch pathway on chondrocyte markers by investigating the influence of the γ-secretase inhibitor DAPT on type II collagen synthesis (Figure 3). The results of immunofluorescence analyses (Figures 3A and B) showed that the number of cells containing type II collagen decreased during passaging. Furthermore, immunoblotting showed that the amount of type II collagen protein decreased during passaging (Figure 3C). Adding the γ-secretase inhibitor DAPT increased the overall quantity of type II collagen protein detected by immunoblotting (Figure 3C) and increased the number of cells staining positive for type II collagen by immunofluoresence (Figures 3A and B).

Figure 3.

Influence of the γ-secretase inhibitor DAPT on type II collagen (Col2) synthesis. A, Cells from passages P0 to P3 were cultured without or with DAPT (2.3 μM), and then fixed and analyzed by immunofluorescence for type II collagen expression; the nuclei were detected by Hoechst staining (with 4′,6-diamidino-2-phenylindole [blue]). B, Type II collagen–positive cells were counted in each passage. Bars show the mean and SD percent positive cells/total cells in cultures without or with DAPT in 5 random microscope fields from 2 independent experiments. ∗∗∗ = P < 0.001. C, Immunoblots for type II collagen, with β-actin as a reference, were obtained using total cell extracts (20 μg per line) from P0 to P3 in cultures without or with DAPT, with the cultures resolved on a 4–12% Bis-Tris gradient gel. Western blot results are representative of 1 of 3 blots from 3 independent cultures. D, Type II collagen mRNA was assessed by quantitative RT-PCR in cultures without or with DAPT. As a reference, results were normalized to the levels of HPRT mRNA. Bars show the mean and SEM fold induction during passaging in cultures (n = 3 per plate) from 3 independent experiments, with each experiment assessing a pool of 2–4 plates in triplicate. See Figure 2 for other definitions.

Densitometric analysis of the immunoblots indicated that the Col2a1:β-actin ratio was greater in cells cultured with DAPT than in controls, with increases of 20% at P0, 50% at P1, 120% at P2, and 190% at P3. The amounts of other chondrocyte markers (aggrecan, Col1a2) were not affected by DAPT (results not shown). In contrast, DAPT did not alter the amounts of Col2a1 mRNA at any stage of culture (Figure 3D). This suggests that Notch target genes influence collagen synthesis at a posttranslational step.

Regulation of MMP-13 synthesis by γ-secretase.

Because DAPT did not influence the amount of Col2a1 mRNA in cells, we postulated that other proteins involved in type II collagen breakdown are dependent on γ-secretase. We therefore examined the synthesis of MMP-13, the main MMP involved in type II collagen breakdown (8). The amount of MMP13 mRNA in cells increased markedly (up to 7-fold; P < 0.001) from P0 to P3 (Figure 4A), but the amount of MMP13 mRNA decreased by 76% in cells incubated with DAPT (P < 0.01 versus cultures without DAPT at P3). Similarly, the amount of MMP-13 protein in cells increased greatly from P0 to P3, but once again, DAPT greatly reduced the amount of protein at each passage (Figure 4C). Densitometric analysis of the immunoblots indicated that the MMP13:β-actin ratio in cells incubated with DAPT was reduced by 73% at P0, 78% at P1, 96% at P2, and 50% at P3, as compared with that in control cultures without DAPT.

Figure 4.

Influence of the γ-secretase inhibitor DAPT on levels of matrix metalloproteinase 13 (MMP-13) (A and C) and MMP-2 (B and D). Cells from passages P0 to P3 were cultured without or with DAPT (2.3 μM). The cells were then removed by scraping, resolved on a 4–12% Bis-Tris acrylamide gradient gel, and assessed for MMP13 and MMP2 mRNA by quantitative RT-PCR (A and B) and for MMP-13 and MMP-2 protein by Western blotting (C and D). As a reference, results were normalized to the levels of HPRT mRNA (A and B) or β-actin (C and D). Bars show the mean and SEM fold induction of mRNA during passaging in cultures (n = 3 per plate) from 3 independent experiments, with each experiment assessing a pool of 2–4 plates in triplicate. For cultures without DAPT versus with DAPT at each passage, # = P < 0.05 and ## = P < 0.01. For cultures without DAPT at different passages versus the equivalent culture at P0, ∗∗∗ = P < 0.001. See Figure 2 for other definitions.

We also monitored the synthesis of an unrelated metalloproteinase, MMP-2, to demonstrate that the changes in MMP-13 synthesis from P0 to P3 and its inhibition by DAPT were not linked to a global effect on the synthesis of MMPs. MMP2 mRNA and MMP-2 protein were abundant in our monolayer cultures, but neither showed a change in expression levels from P0 to P3, and neither was influenced by DAPT (Figures 4B and D).

Of note, however, both Notch activation and MMP-13 synthesis were reversible, since chondrocytes from P1 cultured in alginate beads underwent redifferentiation. This led to increased type II collagen synthesis and decreased synthesis of type I collagen, Hes1, and MMP-13 (results available upon request from the corresponding author).

Effect of Notch activation or inhibition on MMP-13 synthesis.

In addition to Notch, γ-secretase activates other signaling pathways, such as APP and CD44, and therefore these pathways are also inhibited by DAPT (23, 26). We demonstrated the direct influence of the Notch signaling pathway on MMP-13 synthesis using primary cultures of chondrocytes overexpressing constitutive active forms of Notch1 or a dominant-negative form of the Notch1 transcriptional coactivator CSL (Figure 5). We used a Notch1 construct lacking its extracellular domain, N1ΔE, that is constitutively cleaved by γ-secretase, and a Notch1 construct including the ICD, N1ICD, that bypasses the effect of the γ-secretase cleavage block (27). Primary cultures of chondrocytes were transiently transfected at P0, and the expression levels of the constructs were monitored using a cleaved Notch1 antibody, V1744.

Figure 5.

Effect of constitutive activation or inhibition of the Notch pathway on synthesis of matrix metalloproteinase 13 (MMP-13). Cells at P0 (A) and P2 (B) were cultured without or with the γ-secretase inhibitor DAPT (2.3 μM) to 70% confluency, and then transfected with the indicated expression vectors. Cells were harvested 48 hours after transfection, and extracts (20 μg) were used for immunoblotting and/or quantitative RT-PCR. A, Primary cultures of chondrocytes at P0 were transfected with empty vector (mock), a Notch1-expressing vector that lacked its extracellular domain (N1ΔE), or a Notch1-expressing vector with its intracellular domain (N1ICD). Extracts were analyzed by immunoblotting with antibodies against MMP-2, MMP-13 (arrowhead), type II collagen (Col2), activated Notch (V1744), or β-actin as a reference. B, Primary cultures of chondrocytes at P2 were transfected with empty vector (mock) or with a vector expressing a dominant-negative form of CSL (CSL-DN). Protein extracts were analyzed by immunoblotting for MMP-13 protein with β-actin used as a reference (top), and by quantitative RT-PCR for MMP13 mRNA with HPRT used as a reference (bottom). Bars show the mean and SEM results from 3 independent experiments. See Figure 2 for other definitions.

The V1744 antibody recognized N1ICD but not N1ΔE in cells cultured with DAPT (Figure 5A), thus demonstrating the bona fide expression of the constructs and their regulation by the γ-secretase inhibitor. Transfection with N1ΔE led to increased MMP-13 protein synthesis as compared with the MMP-13 levels in cultures using the empty vector (Figure 5A). This increase was not completely blocked by DAPT (31% inhibition in the levels of MMP-13), probably due to the overexpression of N1ΔE. In contrast, the increased MMP-13 protein synthesis was not altered by N1ICD when cells were treated with DAPT (11% inhibition in the levels of MMP-13), since N1ICD bypasses the γ-secretase cleavage block. Expression of these constructs had no effect on the synthesis of MMP-2 protein (Figure 5A).

Transfection with N1ΔE reduced type II collagen synthesis, and DAPT blocked this reduction because N1ΔE is regulated by DAPT. Cells transfected with N1ICD also contained subnormal amounts of type II collagen protein, and the levels were not influenced by DAPT because N1ICD is downstream of the γ-secretase in the Notch cascade and is insensitive to DAPT. However, cells transfected with the mock empty vector also had reduced type II collagen synthesis as compared with that in untransfected cells, perhaps because transfection itself or features of the vectors can modify cells and basal expression of type II collagen. We also demonstrated the effect of Notch activation on type II collagen synthesis by immunofluorescence on positive-transfected cells, using constitutive active forms of Notch1 receptor (results available upon request from the corresponding author).

CSL (or RBP-J) is a DNA binding protein associated with repressors of transcription in the absence of Notch activation. When Notch is activated, CSL is associated with the ICD of Notch and other transcriptional coactivators, changing CSL from a transcriptionally inactive complex to an active one (28). The use of a dominant-negative form of CSL (29) can inhibit the activation of Notch target genes. Inhibiting the CSL coactivator of Notch transcription late in chondrocyte monolayer culture (P2 or P3) should influence MMP-13 synthesis, because Notch has been activated during successive passages. Cells transiently transfected with a dominant-negative form of CSL (CSL-DN) showed reduced synthesis of MMP13 mRNA and MMP-13 protein (Figure 5B). Thus, our results demonstrate that the Notch signaling pathway is involved in the regulation of MMP-13 synthesis, and that this process occurs as the phenotype of the chondrocytes in primary culture changes.


The dynamic, constant balance between the synthesis and breakdown of the cartilage extracellular matrix is disrupted in degenerative diseases such as OA. Degrading enzymes such as aggrecanases and collagenases are produced, and matrix components are destroyed (3). Phenotypic changes also occur, as indicated by the abundance of type II collagen in normal cartilage but lack of synthesis by OA chondrocytes, and the fact that chondrocytes produce other matrix components. These changes are enhanced by proinflammatory stimuli. The changes in the chondrocyte phenotype during OA remain controversial. Some investigators have suggested that dedifferentiation takes place, since type I collagen synthesis is increased (30), while others have suggested that hypertrophic differentiation occurs, because type X collagen is produced (31, 32).

Studies have shown that the phenotype of mouse chondrocytes, cultured in the same manner as described herein, changes. Mouse chondrocytes become fibroblast-like, synthesize less chondrocyte marker (type II collagen, aggrecan), and synthesize more type I collagen and MMP-13 (12, 33, 34). Although the cells dedifferentiated to fibroblast-like cells, our in vitro data recapitulate, to some extent, the changes in phenotype that occur in OA cartilage, in which cells lose chondrocyte markers and produce type I collagen and metalloproteinases.

We show that the Notch signaling pathway is activated (Figure 2) during passaging of mouse chondrocytes in primary culture. The results of several studies have suggested that the specific features of chondrocytes and their maturation are promoted or suppressed by the Notch cascade, in conjunction with other signaling pathways. Although it was shown that Notch signaling, acting through Delta1, negatively regulates the transition from prehypertrophic to hypertrophic chondrocytes (15, 17), a recent study has shown that Notch has the opposite function in distinct phases of mouse chondrogenesis (35). Those authors demonstrated that Notch activation promotes the development of specific chondrocyte features but blocks the maturation of mouse mesencephalic neural crest cells to hypertrophic chondrocytes. Since our results indicate that Notch signaling is involved in chondrocyte dedifferentiation, this apparent discrepancy may be due to differences in the cell populations studied. Mouse mesencephalic neural crest cells need Notch activation to express Sox9; therefore, to be committed to the chondrogenic lineage, the early Notch-sensitive period is critical. The population of cells obtained from young mouse articular cartilage is initially heterogeneous, although most of the cells synthesize type II collagen and are in a proliferative or prehypertrophic state.

Our results show that Notch is involved in dedifferentiation. It has been proposed that Notch1 signaling in immature articular cartilage maintains clonality and proliferation, allowing the maintenance of a progenitor-like subpopulation (18). The balance between the progenitor and differentiated cells, with active or inactive Notch, must be maintained for articular cartilage to function and resist disease.

Type II collagen mRNA synthesis decreases following the first passage (P1), but the drop in type II protein is more gradual, continuing until P3. The decrease in type II collagen protein could be due to decreased mRNA synthesis and/or increased levels of MMP-13 (collagenase 3), the main substrate of which is type II collagen (8). However, there may be other explanations, such as posttranscriptional or posttranslational changes that could influence the synthesis and stability of type II collagen during culture.

The cells used in this study were immature, which may make it difficult to extrapolate our results to studies of mature chondrocytes in the pathogenesis of OA. Mature chondrocytes synthesize MMP-13 (36), and its synthesis is supernormal in OA chondrocytes (37). Furthermore, MMP-13 seems to be a key enzyme in OA pathophysiology, since overexpression of human MMP-13 in transgenic mice leads to cartilage breakdown similar to that observed in human OA (38). A recent study demonstrated that the Notch signaling pathway is activated in OA cartilage, and the Notch receptor and its ligands and target genes are much more abundant than in healthy cartilage (39). Thus, both MMP-13 and Notch are active in mature chondrocytes, and their activities are supernormal in modified chondrocytes such as those found in OA. Nevertheless, a correlation between Notch activation and MMP-13 synthesis has never been previously described.

Inhibition of Notch activation by DAPT, which strongly inhibits MMP-13 synthesis, protects type II collagen from breakdown but has no effect on type II collagen mRNA synthesis. Our data therefore support the hypothesis that the γ-secretase inhibitor protects type II collagen from breakdown by blocking MMP-13 synthesis. However, we have shown that MMP-13 synthesis depends on Notch activation, and therefore the MMP-13 promoter is probably a target of the CSL/Notch ICD transactivator complex.

The mouse MMP-13 promoter sequence has a putative CSL (Notch coactivator) binding site (CCTGGGAA; underline indicates the amino acid substitution compared with the consensus sequence) at −7/+1, and this is also present in the rat MMP-13 promoter. It matches the 7/8-bp CSL consensus binding site (CGTGGGAA) as defined by Tun et al (40). There is also a putative CSL binding site (GGTGGGAA) in the MMP-2 promoter (at −61/−54), but MMP-2 is not regulated by Notch in our model. There is no evidence that the MMP-13 promoter is directly regulated by Notch1/CSL binding. Thus, Notch may well regulate MMP-13 synthesis via other transcription factors.

Many studies have described the regulation of MMP-13 synthesis. Several stimuli, such as hyaluronan oligosaccharide (41) and proinflammatory cytokines such as interleukin-1β (IL-1β) (42), activate the MMP-13 promoter via NF-κB and p38 MAPK. There may be crosstalk between these signaling pathways, which contributes to the complex manner in which the fate of a chondrocyte and its responses of cells to extracellular signals are determined. Crosstalk between the Notch and NF-κB signaling pathways (43, 44) and between the Notch and MAPK pathways (45) has been described. Furthermore, IL-1β and other proinflammatory stimulators also stimulate MMP-13 synthesis via the NF-κB and p38 MAPK pathways. Stimulation by these cytokines could enhance Notch activation by increasing γ-secretase activity (46).

Our findings therefore suggest that Notch activation leads to cartilage breakdown by MMP-13. However, further studies are needed to determine the molecular mechanisms involved in MMP-13 regulation by Notch, to investigate the possible crosstalk with other signaling pathways involved in MMP-13 synthesis, and to better understand the relevance of these phenomena to OA.

Articular chondrocytes account for only 5–10% of the total volume of articular cartilage, without any cell–cell interaction, whereas the cells used in our study grow to form confluent monolayers. Because Notch is activated via cell–cell interactions, the in vivo ligand of Notch in relation to Notch activation must be assessed.

The main hypothesis of how Notch is activated without cell–cell interaction involves the role of soluble ligands. In mammals, the Delta and Jagged ligands mature as a result of cleavages by tumor necrosis factor α–converting enzyme–like protease and γ-secretase, and these could release extracellular soluble forms (47). Furthermore, Notch ligands can also be released extracellularly via secreted vesicles (exosomes) (48). However, the Delta and Jagged families are not the only Notch ligands. A recent study demonstrated that microfibril-associated glycoproteins (MAGPs), components of extracellular microfibrils, cause dissociation of the Notch1 extracellular domain and activate the receptor (49). Immunoprecipitation studies showed that MAGP-1 interacts with the chondroitin sulfate proteoglycan decorin in the cultured medium of fetal bovine chondrocytes (50). The next major challenge may be to identify the Notch ligand in cartilage, since this will considerably advance our understanding of how the Notch signaling pathway is involved in the pathogenesis of OA. Manipulation of the Notch pathway could be critical for improving therapy using chondrocyte implants.


Dr. Berenbaum had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Blaise, Mahjoub, Berenbaum, Bausero.

Acquisition of data. Blaise, Mahjoub, Salvat, Barbe, Brou, Bausero.

Analysis and interpretation of data. Blaise, Mahjoub, Brou, Corvol, Savouret, Rannou, Berenbaum, Bausero.

Manuscript preparation. Blaise, Mahjoub, Brou, Corvol, Savouret, Berenbaum, Bausero.

Statistical analysis. Blaise, Mahjoub, Bausero.


We thank Dr. Owen Parkes for editing the English text.