Relative messenger RNA expression profiling of collagenases and aggrecanases in human articular chondrocytes in vivo and in vitro

Authors


Abstract

Objective

Osteoarthritic (OA) cartilage destruction depends on collagen- and aggrecan-degrading proteases such as collagenases (MMP-1 and MMP-13), stromelysin (MMP-3), MMP-14, as well as the so-called aggrecanases (ADAM-TS4 and ADAM-TS5). In this study, we tried to clarify whether these proteases are expressed in vivo in human normal and OA cartilage (and whether they are up-regulated or down-regulated during the disease process) and in interleukin-1β (IL-1β)–stimulated chondrocytes in vitro.

Methods

Quantitative polymerase chain reaction assays were developed and performed on RNA isolated directly from normal and degenerative cartilage tissue as well as from primary human articular chondrocytes cultured with and without IL-1β.

Results

In vivo, MMP-1 was detectable only at very low levels in any condition. MMP-13 expression was low in normal and early degenerative cartilage but was strongly up-regulated in late-stage OA specimens. MMP-1 and MMP-13 were expressed much higher in vitro than in vivo and were up-regulated by IL-1β. Among all proteases, MMP-3 was by far the most strongly expressed, although it was strongly down-regulated in late-stage OA specimens. Expression of MMP-3 was higher in vitro than in vivo and was up-regulated by IL-1β. ADAM-TS5 and MMP-14 were expressed in all sample groups. Expression of ADAM-TS4 was very low in vivo and was induced in vitro after stimulation by IL-1β.

Conclusion

Our expression data clearly support MMP-13 as the major collagenase in OA cartilage. The most strongly expressed aggrecanase was ADAM-TS5. ADAM-TS4 was expressed only at a very low level in normal cartilage and was only slightly up-regulated in OA cartilage, casting doubt on this enzyme being the relevant aggrecanase of articular cartilage. Results of our study show that expression of many enzymes is significantly different in vitro and in vivo and suggest that IL-1β stimulation of articular chondrocytes might not be a good model for the matrix catabolism in OA cartilage.

Osteoarthritic (OA) cartilage degeneration as well as cartilage destruction in rheumatoid arthritis depend, at least to a significant degree, on enzymatic degradation of matrix components. Two types of molecules, collagen fibrils and the proteoglycan aggrecan, represent the major targets in terms of enzymatic activity and functional loss of the cartilage matrix as a result of damage to these molecules. Whereas degradation of collagen fibrils leads to matrix instability with tissue swelling, degradation of proteoglycans leads to cartilage softening and loss of fixed charges (1), both of which are classic features of cartilage destruction.

Catabolism of both types of molecules is supposed to involve different types of enzymes, largely from 2 protease families: matrix metalloproteinases (MMPs) and the “A disintegrin and metalloproteinases with thrombospondin type 1 motif” (ADAM-TS). The initial degradation of collagen fibrils (within the triple-helical region) depends on cleavage at the collagenase site, for which there exist 2 major candidate enzymes: collagenase 1 (MMP-1) and collagenase 3 (MMP-13). Expression of a third candidate, collagenase 2 (MMP-8 or neutrophil collagenase), was not observed in previous experiments (2) and was, therefore, not investigated in this study. Expression of both MMP-1 and MMP-13 in articular cartilage has been reported (3–5).

Aggrecan degradation involves cleavage at many sites within the core protein and also the carbohydrate side chains (6, 7). However, 2 defined sites in the interglobular domain are thought to play an important role in normal and OA cartilage (8, 9): the so-called MMP site (Asn341–Phe342) (10), at which cleavage by many MMPs and in particular stromelysin (MMP-3) is described (10–12), and the so-called aggrecanase site (Glu373–Ala374) (13–16). For the latter site, enzymes such as aggrecanase 1 (ADAM-TS4) (17, 18) and aggrecanase 2 (ADAM-TS5) (19), as well as MMP-14 (20) and more recently ADAM-TS1 (21), are thought to be responsible. At least, all these enzymes have been shown to cleave aggrecan correspondingly and to be expressed in articular chondrocytes in vivo or in vitro (22, 23). However, no quantitative data on expression levels of these enzymes in normal and OA cartilage are thus far reported allowing estimations of the relative importance of these proteases in normal and diseased tissue.

Data on protease expression are valuable not only for understanding immediate matrix degradation, but also because MMPs are important in the processing of other proteases and growth factors (e.g., MMP-3 activates tumor necrosis factor α), and for growth factor–modulating proteins (e.g., degradation of insulin-like growth factor–binding proteins by MMP-2 and MMP-3) (for review see ref. 24).

Knowledge of the level of expression of potentially involved enzymes, as well as changes in the level of expression, is of primary relevance for understanding the degenerative processes as well as identifying potential targets for therapeutic intervention. Previous in vitro studies have addressed these issues, but, depending on the culture conditions, levels of expression vary considerably (23). Also, use of animal cells further complicates the picture of detected levels of expression, because gene expression levels often are not fully comparable (23).

Thus, to determine whether a molecule is significantly expressed in human normal and OA cartilage and whether it is up-regulated or down-regulated during the disease process, in situ expression must be observed directly, using a quantitative method in a representative number of samples. Because results of such analyses have not yet been reported, the currently available information is mostly speculative and, to a large extent, contradictory.

In this study, we used online quantitative polymerase chain reaction (PCR) technology, as provided by the TaqMan system (Perkin Elmer, Weiterstadt, Germany), on a series of normal, early degenerative, and late-stage OA cartilage samples in order to provide solid data on these 2 important questions. Additionally, we compared in vivo levels and in vitro expression in 2 classic in vitro culture systems (alginate beads and short-term high-density monolayer culture), with and without stimulation by interleukin-1β (IL-1β).

MATERIALS AND METHODS

Cartilage samples.

Cartilage from human femoral condyles of the knee joints was used. Samples of normal articular cartilage (n = 14) from donors ages 40–83 years (mean 60.4 years) and early degenerative cartilage (n = 12) from donors ages 49–91 years (mean 68.3 years) were obtained at autopsy, within 48 hours of death. Samples of OA cartilage (n = 15) were obtained from patients with late-stage OA joint disease (ages 60–79 years, mean 69.9 years) who were undergoing total knee replacement surgery. Immediately after removal, the cartilage samples were frozen in liquid nitrogen and stored at −80°C until required for RNA isolation.

Cartilage was considered normal if it showed no significant softening or surface fibrillation. Early degenerative cartilage was defined as cartilage that showed moderate fibrillation and softening but no advanced erosion; cartilage from peripheral areas showing no obvious signs of degeneration was not used. Late-stage OA cartilage was always derived from patients who were undergoing knee arthroplasty because of complete destruction of the articular cartilage in major portions of the joints. Patients with rheumatoid arthritis were excluded from the study. Only primary degenerative and not regenerative cartilage (osteophytic tissue) was used for this study.

RNA isolation from articular cartilage.

Total RNA from cartilage tissue was isolated as described previously (25). Isolated RNA was controlled for quality by electrophoresis and by spectrophotometry.

Cell isolation.

Samples of normal human knee articular cartilage were obtained from 9 donors, at autopsy (within 48 hours of death) or at the time of amputation (alginate bead culture: 61-year-old man, 59-year-old man, 71-year-old man; monolayer culture without serum: 61-year-old man, 39-year-old woman, 50-year-old man; monolayer culture with serum: 43-year-old man, 52-year-old man, 60-year-old man). Cartilage pieces were finely chopped, and chondrocytes were enzymatically isolated from associated matrix. Sliced cartilage pieces were first digested with 1 mg/ml pronase (Roche, Basel, Switzerland) in Dulbecco's modified Eagle's medium (DMEM)/Ham's F-12 (Gibco BRL, Eggenstein, Germany) with 10% fetal calf serum (FCS; Biochrom, Berlin, Germany) for 30 minutes and subsequently with 1 mg/ml collagenase P (Boehringer Mannheim, Ingelheim, Germany) in Ham's F-12 (Gibco BRL) with 10% FCS. Finally, cells were washed several times in Ham's F-12 and counted and checked for viability using trypan blue staining.

High-density monolayer cultures in serum-containing media, and stimulation with IL-1β.

After isolation, chondrocytes were seeded at 2 × 106 cells/well in 6-well tissue culture plates and maintained for 48 hours in DMEM/Ham's F-12 medium (Gibco BRL) supplemented with 10% FCS and 50 μg/ml penicillin–streptomycin solution (Gibco BRL) and 50 μg/ml ascorbate (Sigma, Deisenhofen, Germany). Thereafter, chondrocytes were stimulated with 1 ng/ml recombinant human IL-1β (rHuIL-1β; Biomol, Hamburg, Germany) in DMEM/Ham's F-12 medium containing 10% FCS (Biochrom) or cultivated in medium containing 10% FCS alone for 3 days. The medium was changed every day. At the end of the stimulation period, cells were washed in sterile phosphate buffered saline (PBS), lysed in 350 μl of lysis buffer RLT/106 cells (Qiagen, Hilden, Germany), and stored at −80°C.

High-density monolayer cultures in serum-free media, and stimulation with Il-Lβ.

The cells were processed as described above, but in serum-free medium.

Alginate bead culture, and stimulation with IL-1β.

After isolation, chondrocytes were encapsulated in alginate beads at a density of 4 × 106 cells/ml in a manner similar to the protocol described by Häuselmann et al (26). Briefly, cells were suspended in sterile 0.15M NaCl containing 1.2% medium-viscosity alginate (Sigma), then slowly pushed through a 22-gauge needle in a dropwise manner into a 0.1M CaCl2 solution. The beads were allowed to polymerize in the CaCl2 solution for 10 minutes, then were washed 3 times with 0.15M NaCl and once with DMEM/Ham's F-12. Afterward, the beads were placed and maintained in culture medium (DMEM/Ham's F-12 plus 10% FCS, 50 μg/ml penicillin–streptomycin, and 50 μg/ml ascorbic acid) for 2 days.

Thereafter, chondrocytes were stimulated with 1 ng/ml rHuIL-1β (Biomol) in DMEM/Ham's F-12 medium with 10% FCS (Biochrom) or were cultivated in 10% FCS (Biochrom) alone for 3 days. The medium was changed every day. At the end of the stimulation period, the alginate beads were dissolved with sodium citrate buffer (55 mM sodium citrate, 0.15M NaCl, 20 mM EDTA, pH 7.4), and the cells were recovered by centrifugation, washed in sterile PBS, lysed in 350 μl of lysis buffer RLT/106 cells (Qiagen), and stored at −80°C.

RNA isolation.

RNA was isolated from cultured cells using the RNeasy mini kit (Qiagen) (using an on-column DNase digestion step, according to the manufacturer's instructions). Briefly, cells were passed through a QIAshredder (Qiagen), and the eluted lysate was mixed 1:1 with 70% ethanol. The lysate was applied to a mini column, and, after washing and DNase digestion, the RNA was eluted in 30–50 μl of RNase-free water. The quantity and quality of RNA were assessed by ethidium bromide staining of RNA separated on 1.2% agarose gels.

Synthesis of complementary DNA (cDNA).

First-strand cDNA was synthesized using 1 μg of total RNA, 400 units of Moloney murine leukemia virus reverse transcriptase, RNase H Minus (both from Promega, Mannheim, Germany), 2 mM deoxyribonucleoside triphosphate (Roth, Karlsruhy, Germany), and 200 ng random primers (Promega) in a total volume of 40 μl.

TaqMan PCR.

TaqMan PCR was used to detect human MMP-1, MMP-3, MMP-13, MMP-14, as well as ADAM-TS4 and ADAM-TS5 in human articular cartilage RNA samples. The primers (MWG Biotech, Ebersberg, Germany) and TaqMan probes (Eurogentec, Luik, Belgium) were designed using Primer Express software (Perkin Elmer). In order to be able to obtain quantifiable results for all genes, specific standard curves using sequence-specific control probes were performed in parallel to the analyses. Thus, for each gene, a gene-specific cDNA fragment was amplified by the gene-specific primers (Table 1) and cloned into pGEM-T Easy (Promega) or pCRII-TOPO (Invitrogen, Karlsruhe, Germany).

Table 1. Sequences of primers and probes for quantitative online PCR experiments*
 Accession no.PrimernmProbeMg, mM
  • *

    PCR = polymerase chain reaction; fw = forward; rv = reverse; MMP = matrix metalloproteinase.

GAPDHNM_002046fw: GAAGGTGAAGGTCGGAGTC50CAAGCTTCCCGTTCTCAGCC5.5
  rv: GAAGATGGTGATGGGATTTC900  
MMP-1NM_002421fw: CTGTTCAGGGACAGAATGTGCT300ACGGATACCCCAAGGACATCTACAGCTCC6.5
  rv: TCGATATGCTTCACAGTTCTAGGG900  
MMP-3NM_002422fw: TTTTGGCCATCTCTTCCTTCA900AACTTCATATGCGGCATCCACGCC4
  rv: TGTGGATGCCTCTTGGGTATC50  
MMP-13NM_002427fw: TCCTCTTCTTGAGCTGGACTCATT900TCCTCAGACAAATCATCTTCATCACCACCAC7
  rv: CGCTCTGCAAACTGGAGGTC50  
MMP-14NM_004995fw: TGCCTGCGTCCATCAACACT900AAGACGAATTTGCCATCCTTCCTCTCGT7
  rv: CATCAAACACCCAATGCTTGTC50  
ADAM-TS4AF148213fw: TGCCCGCTTCATCACTGA900ACAGTGCCCATAGCCATTGTCCAGGA6
  rv: CAATGGAGCCTCTGGTTTGTC50  
ADAM-TS5AF142099fw: CGCTGCCACCACACTCAA300AAGTGGCAGCACCAACACAACCAGC4.5
  rv: CGTAGTGCTCCTCATGGTCATCT900  

The cloned amplification product was sequenced for confirmation of correct cloning. Cloned standard probes were amplified using the QIAfilter Midi plasmid kit (Qiagen) and linearized by restriction digest. Linearized standard probes were gel-purified using the QIAquick gel extraction kit (Qiagen). Purified probes (fragments) were quantified using a fluorimetric assay (Picogreen; Molecular Probes, Eugene, OR). Concentrations were confirmed (checked) by measuring the absorbance at 260 nm in a spectrophotometer and comparison with DNA bands of known concentration (MassRuler DNA Ladder; MBI Fermentas, St. Leon-Rot, Germany) in an ethidium bromide–stained agarose gel. For the standard curves, concentrations of 10, 100, 1,000, 10,000, 100,000, as well as 1,000,000 molecules per assay were used (all in triplicate).

For the analyses of the different genes, a separate master mix was prepared for each of the primer pairs and contained a final concentration of 200 μM nucleoside triphosphate, 600 nM ROX buffer (Applied Biosystems, Weiterstadt, Germany), and 100 nM TaqMan probe. For all genes, the final reaction mix contained (besides cDNA and 1 unit polymerase [Eurogentec]), forward and reverse primers, the corresponding probes, and MgCl2 at the concentrations shown in Table 1. All experiments were performed in triplicate. The assay for GAPDH was described previously (27).

Statistical analysis.

For the in vivo investigations, statistical evaluation of significant differences in levels of expression was performed using the nonparametric Wilcoxon-Mann-Whitney test. This nonparametric test is more likely to be appropriate than, for example, the t-test, because it is not based on assumptions regarding the distribution of expression values (e.g., normal distribution). Because of the limited number of cases to be compared in the in vitro investigations, the t-test for pairwise comparison was used. P values less than 0.05 were considered significant.

RESULTS

TaqMan assay development.

For exact quantification of levels of messenger RNA (mRNA) expression for the MMP and ADAM-TS of interest, TaqMan assays for the various gene expression products were developed. Primers and probes were designed using Primer Express software (Perkin Elmer). Primers, probes, and various assay conditions were tested according to the manufacturer's recommendations. The sequences of primers and probes as well as optimal assay conditions are shown in Table 1. For all genes, gene-specific control plasmid standard probes were generated (using the primers indicated in Table 1). These amplified, cloned, and sequenced cDNA fragments were used to perform standard curves in parallel to every assay, in order to obtain quantitative data on all genes investigated (including GAPDH). For standardization of levels of gene expression as determined by TaqMan analysis, mRNA ratios relative to GAPDH were calculated (by dividing the gene copy number by the copy number obtained for GAPDH).

Expression analysis of MMP-1 and MMP-13.

The MMPs able to degrade native triple-helical collagen by cleaving at the collagenase site (MMP-1 [collagenase 1] and MMP-13 [collagenase 3]) were investigated first. MMP-1 was detectable only at a very low level in normal articular cartilage and early degenerative cartilage and was not significantly up-regulated in late-stage OA samples (Figure 1a).

Figure 1.

Quantitative TaqMan analysis for expression of messenger RNA in normal, early degenerative (early deg), and late-stage osteoarthritic (late OA) cartilage. ∗ = P < 0.05 versus normal articular cartilage; ∗∗∗ = P < 0.001 versus normal articular cartilage. MMP = matrix metalloproteinase. Bars show the mean and SD.

MMP-1 was expressed much higher in vitro than in vivo (190–1,000-fold, depending on culture conditions; P < 0.001) (Figure 2 and Table 2) and was u p-regulated even more (5–6-fold, depending on culture conditions) by IL-1β (Figures 3A–C) (Pmonolayer < 0.001) (Table 3).

Figure 2.

Levels of protease messenger RNA (mRNA) expression in normal articular chondrocytes in vivo compared with normal articular chondrocytes cultured in vitro in alginate beads or high-density monolayers with (monolayer +) and without (monolayer−) serum. Values are the mRNA ratios relative to GAPDH.

Table 2. Levels of mRNA expression in normal articular chondrocytes in vivo versus normal articular chondrocytes cultured in vitro in alginate beads or high-density monolayers with and without serum*
ProteaseIn vivoAlginate beads with serumHigh-density monolayer cultures
With serumWithout serum
Ratio to GAPDHPRatio to GAPDHPRatio to GAPDHP
  • *

    Values are the ratio of mRNA molecules to GAPDH molecules. P values were determined by t-test. MMP = matrix metalloproteinase; NS = not significant.

MMP-10.0021.2<0.0010.37<0.0011.1<0.001
MMP-311.955.0<0.00137.2<0.00569.9<0.005
MMP-130.0020.74<0.0010.12<0.0010.25<0.001
MMP-140.210.22NS0.08<0.010.01<0.05
ADAM-TS40.0040.001NS0.001NS<0.001NS
ADAM-TS50.050.22<0.0010.07NS0.56<0.005
Figure 3.

Quantitative TaqMan analysis for levels of messenger RNA expression of proteases in isolated normal articular chondrocytes after stimulation with interleukin-1β (IL-1β). Values are the ratios of IL-1β–stimulated molecules to non–IL-1β–stimulated molecules. Three completely independent experiments were performed for each condition. A, Experiments B1, B2, and B3, alginate beads with fetal calf serum (FCS). B, Experiments M1, M2, and M3, high-density monolayer culture with FCS. C, Experiments M4, M5, and M6, high-density monolayer culture without FCS. MMP = matrix metalloproteinase; c+ = with serum; c− = without serum.

Table 3. Significance levels for changes in expression of mRNA after interleukin-1β stimulation*
 Alignate bead cultureShort-term high-density monolayer cultures
With serumWithout serumTotal
  • *

    Significance levels are P values as determined by t-test. Monolayer total = values obtained by combining all monolayer cultures independent of the addition of serum to the culture medium. MMP = matrix metalloproteinase.

MMP-10.1630.0090.012<0.001
MMP-30.0310.0360.0620.001
MMP-130.0930.0830.0830.007
MMP-140.0940.1600.0040.005
ADAM-TS40.1340.2050.2750.085
ADAM-TS50.4810.1660.7500.157

MMP-13 was also hardly detectable in normal and early degenerative cartilage, but was very much up-regulated (47-fold) in late-stage OA specimens (P < 0.001) (Figure 1c). In vitro expression of MMP-13 was also considerably higher compared with in vivo expression (60–700-fold depending on culture conditions used; P < 0.005) (Figure 2 and Table 2) and was up-regulated by IL-1β (3–11-fold, depending on culture conditions) (Figures 3A–C) (Pmonolayer < 0.01) (Table 3).

Expression analysis of aggrecan-degrading enzymes stromelysin 1 (MMP-3), aggrecanase 1 (ADAM-TS4), and aggrecanase 2 (ADAM-TS5), as well as membrane type 1 MMP (MT1-MMP; MMP-14).

After the collagenases, aggrecan-degrading enzymes were investigated. In this respect, MMP-3 represents the major enzyme cleaving aggrecan at the so-called MMP cleavage site, with ADAM-TS4 and ADAM-TS5 as well as MMP-14 cleaving at the so-called aggrecanase cleavage site of aggrecan.

MMP-3 was by far the most strongly expressed gene in all samples investigated (P < 0.001). Of note, the highest levels of expression were found in normal (mean 11.9 molecules/molecules GAPDH) and early degenerative (mean 10.7 molecules/molecules GAPDH) specimens, whereas a significant down-regulation (7-fold) was observed in late-stage OA samples (P < 0.001; mean 1.6 molecules/molecules GAPDH) (Figure 1b). Expression of MMP-3 was even higher in vitro than in vivo (37–59-fold, depending on culture conditions used; P < 0.001) (Figure 2 and Table 2) and was up-regulated by IL-1β (3–5-fold, depending on culture conditions) (Figure 3) (Pmonolayer < 0.002) (Table 3).

Expression of ADAM-TS4 was very low in all 3 in vivo sample groups investigated (Figure 1a). Expression was slightly but significantly increased in late-stage OA specimens (P < 0.05) compared with normal specimens but remained very low compared with ADAM-TS5 (Figure 1C). In vitro, ADAM-TS4 was also hardly detectable in any culture model used. However, ADAM-TS4 was induced by IL-1β (Figure 3 and Table 3).

ADAM-TS5 was constitutively expressed in all in vivo sample groups investigated (Figure 1c). A slight increase was found in late-stage OA cartilage compared with normal specimens (3-fold; P < 0.05). Overall expression of ADAM-TS5 was much stronger than that of ADAM-TS4 (P < 0.001). Expression of ADAM-TS5 either was not different in vitro and in vivo (high-density monolayer with serum) or was increased (alginate beads, high-density monolayer without serum). Expression of ADAM-TS5 either was not influenced by IL-1β (alginate beads, high-density monolayer with serum) or was down-regulated (high-density monolayer without serum).

In all specimens investigated, membrane-bound MT1-MMP (MMP-14) was expressed in vivo at roughly the same level (0.21–0.28 molecules/molecules GAPDH) (Figure 1c). In vitro expression of MMP-14 was either similar to that in vivo (alginate beads) or was slightly stronger (4.6-fold; high-density monolayer culture) and was weakly up-regulated by IL-1β (2–3-fold) (Figure 3) (Pmonolayer < 0.01) (Table 3).

DISCUSSION

To understand the molecular basis of OA cartilage degeneration and potential approaches to therapeutic intervention, increased knowledge of involved matrix-degrading proteases is of obvious importance (for review, see ref. 28). In this respect, enzymes able to cleave at the crucial sites of the 2 main components of adult articular cartilage, collagen type II and aggrecan, are of central interest. Despite many reports in the literature on the levels of expression of various collagenases and aggrecanases, no systematic comparative analysis of levels of expression of these enzymes is currently available, particularly in a representative series of in vivo cartilage samples. Most data relate to in situ hybridization analysis (3, 29) or to in vitro studies based on a small number of cases (23). However, in situ hybridization is not quantitative and is prone to nonspecific reactions, and isolated chondrocytes show a largely altered pattern of expression, as was also documented by this study. Thus, the previously reported high levels of expression of MMP-1 in isolated OA (30) as well as normal articular chondrocytes (0.4–2 molecules/molecules GAPDH) are not reflected by high levels of expression in normal and OA chondrocytes in vivo (0.003 molecules/molecules GAPDH), although this was previously suggested by in situ hybridization experiments (29). Also, MMP-13 and MMP-3 are expressed much more highly in vitro than in vivo.

Obviously, the data presented here relate primarily to mRNA, although mRNA and protein appear to be well-correlated for many proteases, such as MMP-1 and MMP-13 (4). Nonetheless, mRNA expression data must be interpreted with caution, because translational and posttranslational modifications, proteolytic activation as well as inhibitors (e.g., tissue inhibitors of metalloproteinases [TIMPs]), are highly relevant for the activation of proteases in normal and diseased tissue. Also, proteases that are relevant for articular cartilage degradation might primarily diffuse in from the synovial fluid, originating in synovial cells (31, 32) rather than articular chondrocytes. It is well known from biochemical studies that OA cartilage shows increased collagenolytic activity (9, 33, 34) and severe collagen denaturation and degradation (35).

Our expression data clearly support the notion that the responsible protease for collagen breakdown is MMP-13 (4), which is in line with the fact that collagenolytic activity can be largely blocked by an MMP-13–specific inhibitor (9). In contrast, MMP-1 was expressed only at very low levels in both normal (3) and OA articular chondrocytes in vivo. MMP-1 may be involved in the basic pericellular collagenous matrix turnover (36) that is found in both normal and OA cartilage (37). However, our data support previous indirect biochemical evidence (9) suggesting that MMP-1 is not primarily involved in interterritorial collagen breakdown in OA cartilage, although a role in disease stages that were not represented in our specimens cannot be excluded.

The aggrecan-degrading protease that was most strongly expressed in all specimens was clearly MMP-3. Strong expression of MMP-3 was suggested in some previous studies (2, 3), whereas other studies could not confirm this (29). Of note, MMP-3 was significantly down-regulated in late-stage OA cartilage, which is consistent with the observation that chondrocytes isolated from OA cartilage express less MMP-3 than those isolated from normal articular cartilage (3), although other authors reported contradictory results (29, 30). Despite the clear down-regulation of MMP-3 in OA cartilage, the overall abundance of MMP-3 in OA cartilage retains this molecule as a highly interesting target for further investigation.

Among the proteases able to cleave aggrecan at the aggrecanase site, ADAM-TS5 was expressed most strongly. ADAM-TS4, the originally reported aggrecanase 1 (17), was expressed only at a very low level in normal articular cartilage and was up-regulated in OA cartilage only to a very low degree. Both of these findings cast severe doubt on the notion that this enzyme represents the major aggrecanase of normal or OA articular cartilage. This doubt is further supported by the fact that, in contrast to very low in vivo expression of other proteases (e.g., MMP-1 and MMP-13), in vitro expression of ADAM-TS4 is not significant in isolated articular chondrocytes either. MMP-14, the MT1-MMP that can also cleave aggrecan at the aggrecanase site (20), was similar to ADAM-TS5 in that it was constitutively expressed in all cartilage specimens at a moderate level, without any significant alteration in OA cartilage specimens, confirming previous in situ hybridization data (22). Other reports of an increase in mRNA levels of MMP-14 in OA cartilage (38) are not supported by our data. Besides its aggrecanase activity, this enzyme is also known to have limited collagenase activity (36, 39) and more importantly to cleave and activate proMMP-2 (40) and proMMP-13 (41, 42) as well as proforms of other matrix metalloproteinases.

In addition to our in vivo studies, we were interested in the expression pattern of chondrocytes in vitro and how inducible the proteases of interest were by IL-1β (the prototypic catabolic stimulus of articular chondrocytes). The first important aspect was that the expression level of many enzymes was significantly higher in vitro than in vivo, even though we also used the alginate bead system, which is supposed to best stabilize the in vivo phenotype of articular chondrocytes (43, 44). One striking example that has led to considerable confusion in the literature is the rather strong expression of MMP-1 in normal articular chondrocytes in vitro (30), which was confirmed in this study and is not found in vivo. Although these data contradict previous reports of nonexpression of MMP-1 in cultured normal articular chondrocytes, they are consistent with results of our own recent cDNA-array experiments (Aigner et al: unpublished observations) as well as those of other studies (5). Another example is MMP-13, which also is barely expressed in normal articular chondrocytes in situ but is significantly expressed in vitro (5, 23). In a second class of enzymes, expression appears to be equal in vitro and in vivo (e.g., MMP-14, ADAM-TS5) (45) or barely detectable (e.g., ADAM-TS4). Both in vivo and in vitro, MMP-3 was, by far, the protease with the highest level of expression, but expression was even higher in vitro than in vivo (30).

With respect to modulation by IL-1β in vitro, 2 classes of proteases were detectable. The first showed strong up-regulation by IL-1β and included MMP-1 (4, 46–48), MMP-3 (3), and MMP-13 (4, 5, 48), genes known to contain promoter elements that are responsive to IL-1 stimulation. ADAM-TS4 was also induced, which was observed in some studies (49) but was missed in others (45). The second class of proteases, represented by ADAM-TS5, was not induced or even down-regulated by IL-1β (45).

Our study, although based on only a limited number of genes, suggests that IL-1β stimulation of articular chondrocytes is not a good model for the destructive processes occurring in OA cartilage, although this model is used in many studies. In contrast to OA cartilage, MMP-3 is up-regulated by IL-1β and is expressed more strongly in vitro than in vivo. MMP-1 is also strongly expressed in vitro and even increased after IL-1β stimulation, whereas it is hardly expressed in vivo and is not up-regulated in OA cartilage. MMP-14 is also induced by IL-1β in vitro, whereas no significant change was found in OA cartilage in vivo.

Our study shows interesting correlations with (and contrasts to) previous observations in the animal model of anterior cruciate ligament transection that was recently characterized on the molecular level (50). In that model system, collagenases 1 and 3 appeared to be expressed at very low levels in normal articular cartilage and, in particular, MMP-13 was strongly up-regulated in degenerating cartilage. In addition, the aggrecanases ADAM-TS4 and ADAM-TS5 were not up-regulated in the diseased state. However, the comparison with our data also shows the limitations of using animal models for simulating human disease. Whereas MMP-3 is already abundant in normal human articular cartilage, it was hardly expressed in the model before transection and was up-regulated in the diseased state, which strongly contrasts with the situation in human OA cartilage. Strong up-regulation of MMP-1 is also not found in the human condition.

Overall, our data extend and specify a recently suggested picture of cartilage matrix degradation in the normal and OA joint based on cDNA-array data (2). Both datasets show many similarities, but the TaqMan assays are clearly more sensitive and show less technical variability, which is documented by, for instance, lower P values for expression differences (e.g., MMP-3 and MMP-13) (2). In both investigations, MMP-3 (stromelysin 1) was, by far, the most abundantly expressed protease, and MMP-3 was significantly down-regulated in OA cartilage. These data suggest that MMP-3 does not represent the driving force in final matrix degradation in OA cartilage, which is further documented by the fact that MMP-3 knockout mice and their wild-type littermates are equally susceptible to cartilage destruction (51). Rather, MMP-3 appears to be a major contributor to physiologic matrix turnover in normal articular cartilage (52, 53). Whereas the cDNA-array data showed no expression of MMP-1 and MMP-13 in normal cartilage, very low levels were detectable by TaqMan analysis. Up-regulation of MMP-13 was found in late-stage OA cartilage in both systems, suggesting that MMP-13, together with MMP-2 (2), is involved in the breakdown of collagen fibers in the setting of OA cartilage degeneration (4, 54).

Results of our study clearly support the possibility that MMP-3 and ADAM-TS5 are the enzymes responsible for the physiologic turnover of aggrecan in normal joints, whereas scarcely any collagenases are expressed and active. In OA cartilage, MMP-13, together with MMP-2, appears to be the enzyme that is critical for collagen fibril breakdown. The increased aggrecan-degrading activity in OA cartilage remains enigmatic, because no strong, highly significant increase in any aggrecan-degrading protease that has thus far been assumed to be important could be identified. Clearly, this could be related to the fact that enzymatic activity is only partly linked to levels of mRNA expression, because enzyme processing and the balance with inhibitory agents (e.g., TIMPs) might be more important than levels of expression. Alternatively, our data indicate that “the” aggrecanase might not yet be identified. In fact, it might be ADAM-TS1, because this enzyme was shown to have aggrecanase activity (21) and is expressed in articular cartilage (Aigner et al: unpublished observations).

Acknowledgements

We thank Drs. G. Zeiler and W. Eger (Orthopedic Hospital Rummelsberg) for providing OA cartilage specimens and Ms F. Boggasch for excellent technical assistance. We are particularly grateful for the expert advice of Dr. A. Zien (Fraunhofer Institute for Algorithms and Scientific Computing, Bonn, Germany) on biostatistical evaluation of the data.

Ancillary