Although it has been shown that aggrecanases are involved in aggrecan degradation, the role of MMP (matrix metalloproteinase) aggrecanolysis is less well studied. To investigate MMP proteolysis of human aggrecan, in the present study we used neoepitope antibodies against MMP cleavage sites and Western blot analysis to identify MMP-generated fragments in normal and OA (osteoarthritis/osteoarthritic) cartilage, and in normal, knee injury and OA and SF (synovial fluid) samples. MMP-3 in vitro digestion showed that aggrecan contains six MMP cleavage sites, in the IGD (interglobular domain), the KS (keratan sulfate) region, the border between the KS region and CS (chondroitin sulfate) region 1, the CS1 region, and the border between the CS2 and the G3 domain, and kinetic studies showed a specific order of digestion where the cleavage between CS2 and the G3 domain was the most preferred. In vivo studies showed that OA cartilage contained (per dry weight) 3.4-fold more MMP-generated FFGV fragments compared with normal cartilage, and although aggrecanase-generated SF-ARGS concentrations were increased 14-fold in OA and knee-injured patients compared with levels in knee-healthy reference subjects, the SF-FFGV concentrations did not notably change. The results of the present study suggest that MMPs are mainly involved in normal aggrecan turnover and might have a less-active role in aggrecan degradation during knee injury and OA.
Articular cartilage covers the ends of bones in synovial joints, and its major function is to provide a smooth lubricated surface between the bones and to withstand compressive loads during joint movements. The structural integrity of articular cartilage ECM (extracellular matrix) is important for maintaining its function.
Aggrecan is the major ECM proteoglycan in articular cartilage and, together with type II collagen, aggrecan provides the cartilage with its mechanical properties of reversible compressibility [1,2]. The aggrecan core protein has a molecular mass of approximately 250 kDa, and consists (starting from the N-terminus) of two globular domains (G1 and G2) with an IGD (interglobular domain) sequence between the G1 and G2 domains, followed by an extended region between G2 and a third globular domain (G3) which is heavily substituted with negatively charged sGAG [sulfated GAG (glycosaminoglycan)] . The sGAG region between G2 and G3 is further organized into three distinct structures: the KS (keratan sulfate) region and CS (chondroitin sulfate) regions 1 and 2.
Degradation of cartilage ECM is a hallmark in arthritic diseases and in joint injuries, and one of the first proteins to be degraded is aggrecan [4–10]. Aggrecan degradation occurs mainly through proteolysis, and there are several proteases that cleave aggrecan at multiple proteolytic cleavage sites: the G1 domain is cut by cathepsin K ; the IGD is cut by aggrecanases [12–17], calpain [18,19], cathepsins B and K [11,20], MMPs (matrix metalloproteinases) [21–25] and the serine protease HtrA1 ; the KS region is cut by calpain [18,19] and MMPs ; the CS1 region is cut by calpain [18,19] and MMPs ; and the CS2 region is cut by aggrecanases [15–17,29], calpain , cathepsin D  and MMPs . Also, the different MMPs have different affinities to cut aggrecan: the IGD IPEN↓FFGV site is cut by MMPs 1–3, 7–9, 12–16, 19 and 20, as observed in different species [25,31,32]; the CS1 region in bovine aggrecan is cut at multiple GVED↓I/(L)SGL sites by MMP-3 ; the end of the CS2 region in bovine aggrecan is cut at RPAE↓ARLE by MMPs 2, 3, 7 and 12 . Aggrecan proteolysis in the IGD is detrimental for the cartilage function, since it results in a loss of aggrecan fragments containing the negatively charged sGAG chains, decreasing the water-holding capacity of the tissue. On the other hand, aggrecan proteolysis in the CS1 and CS2 regions, common in natural turnover in mature cartilage, is believed not to generally affect cartilage function, as the molecule retains some sGAG chains.
In mice, ADAMTS (a disintegrin and metalloproteinase with thrombospondin motifs)-5 (also known as aggrecanase 2) is the major aggrecanase responsible for aggrecan degradation [33,34]. In humans, both ADAMTS-4 (also known as aggrecanase 1) and -5, but also MMPs, seems to be important for aggrecan degradation [25,35]. However, the specific roles of different proteases in cleaving human aggrecan during normal turnover, after joint injuries and in arthritis are still to be determined.
In the present study, we investigate the possible role for MMP digestion of human aggrecan, identifying different MMP cleavage sites and looking at the preferred order of cleavage in vitro. We also analyse the in vivo content of MMP-generated aggrecan fragments in knee articular cartilage and SF (synovial fluid) and compare these with fragments generated by aggrecanase digestion. By analysing the proteolysis of one of the major key extracellular matrix proteins, aggrecan, we hope to learn more about cartilage degradation.
MATERIALS AND METHODS
Antibodies and enzymes
Chondroitinase ABC (EC 22.214.171.124), keratanase (EC 126.96.36.199) and keratanase II (Bacillus sp. Ks36) was from Seikagaku. Human recombinant ADAMTS-4 and an aggrecan neo-specific anti-ARGS monoclonal antibody, characterized previously [36–38], were gifts from M. Pratta and Dr S. Kumar (GlaxoSmithKline). Trypsin-activated recombinant human MMP-3 was a gift from Dr M. Lark (Merck ). Neo-specific monoclonal anti-FFGV antibodies BC14 (from Abcam, catalogue number ab3776, or from MD Bioproducts, catalogue number 1042004) and AF 28 (from Millipore, catalogue number MAB19310) were used and are characterized (Supplementary Figures S1 and S2 at http://www.BiochemJ.org/bj/446/bj4460213add.htm). Neo-specific polyclonal anti-EDLS sera and anti-IPEN sera have been characterized previously [19,39]. Polyclonal anti-GVEE and anti-THLE sera were made (Innovagen) by injecting the immunogen peptides CAPGVEE and THLEIESSC (where the N- and C-terminal cysteine residues were conjugated to keyhole-limpet haemocyanin) into rabbits. Sera was collected at day 42 (1st bleed sera, used herein), day 63 (2nd bleed) and at day 84 (3rd bleed). With regard to the specificity of the anti-GVEE antibody, Western blot blocking experiments suggested that the antibody recognized both the aggrecan neoepitope (APGVEE) and the internal sequence (TTAPGVEEEISGLP) (results not shown). With regard to specificity of the anti-THLE antibody, Western blot blocking experiments showed that the antibody recognized the aggrecan neoepitope THLEIE (results not shown) and to some minor extent also full-length aggrecan (results not shown). Polyclonal antibodies against aggrecan G1 domain (catalogue number PA1-1747) and against aggrecan G3 domain (catalogue number PA1-1745) antibodies were from Affinity BioReagents. The specificity of these antibodies was confirmed by Western blot blocking experiments using immunogen peptides CATEGQVRV-NSIYQKVSL (G1)  and CDGHPMQFENWRPNQPDN (G3) (results not shown). Peroxidase-conjugated secondary antibodies were from Cell Signaling Technology (horse anti-mouse IgG, catalogue number 7076) and from KPL (goat anti-rabbit IgG, catalogue number 074-1516).
Cartilage and SF aggrecan
A human OA (osteoarthritis/osteoarthritic) knee cartilage pool (n=10 subjects) and SF OA pool (n=47 subjects; mean age=48 years, age span=16–89 years) was made and has been described previously . Knee patella (normal) cartilage was obtained from a newborn subject and from a youth (14 year old) subject, both with no record of knee disease or injury. Preparation of cartilage Gu-PG [guanidinium-extracted PG (proteoglycan)] and cartilage A1D1–D3 fractions has been described previously .
Mini SF-D1 preparations (collecting the lower half of the fraction) from the SF OA pool and of individual SF samples from knee healthy reference subjects (n=10; mean age=32 years, age span=20–61 years), acute knee injury patients (n=40; mean age=37 years, age span=19–58 years), chronic knee injury patients (n=3; mean age=30 years, age span=21–38 years) and OA patients (n=5; mean age=49 years, age span=41–54 years) were prepared as described previously . Then, reference and acute knee injury SF-D1 pools were prepared from these individual samples by mixing the corresponding SF-D1 samples so that equal SF volumes of subjects was represented in the pools. The SF samples of these patients have been characterized and used in previous studies [9,38].
sGAG determination (using Alcian Blue precipitation) of cartilage (Gu-PG and A1D1–D3) and SF (neat and D1) samples was carried out as described previously [36,40]. Molar amounts of aggrecan were calculated from sGAG values assuming that the total molecular mass of aggrecan (i.e. the molecular mass for the protein core plus carbohydrate substitution) was 1.5×106 g/mol and that 75% of this weight was represented by sGAG.
All patient-related procedures were approved by the ethics review committee of the Medical Faculty of Lund University, Lund, Sweden.
MMP in vitro proteolysis
Human OA cartilage aggrecan (A1D1 fraction) was digested by MMP-3 using three different enzyme to substrate ratios: 1:40, 1:100 and 1:250 (μg of enzyme/μg of aggrecan A1D1 dry weight) corresponding to the following mol of enzyme (molecular mass=42.8 kDa) per mol of substrate (molecular mass=1500 kDa) ratios: 1:1.1, 1:2.9 (standard digestion) and 1:7.1. Digestion started when MMP was added to the A1D1 sample in the presence of Ca2+ [MMP digestion buffer (final concentrations): 50 mM Tris/HCl, 100 mM NaCl and 10 mM CaCl2 (pH 7.5)] and the sample was incubated for up to 24 h at 37°C. Aliquots were taken from the sample at different time points and the reactions were stopped by adding an excess of EDTA. The time samples were diluted with sodium acetate and Tris acetate solutions [giving final concentrations of 50 mM sodium acetate, 50 mM Tris-acetate and 10 mM EDTA (pH 7.5)] and were deglycosylated with chondroitinase ABC, keratanase and keratanase II as described previously .
In some experiments, deglycosylation was conducted before MMP in vitro digestion as described below. Human OA cartilage aggrecan (A1D1 fraction) was incubated with (+sample) or without (−sample) deglycosylation enzymes (chondroitinase ABC, keratanase and keratanase II) at 37°C as described previously . The samples were then dialysed (10000 Da cut-off) against MMP digestion buffer, in-vitro-digested for 16 h (standard digestion), and the reactions were stopped by the addition of EDTA. The −sample was then incubated with deglycosylation enzymes (as above), whereas the +sample was incubated without the enzymes.
MMP in situ proteolysis
Knee cartilage from one patient was obtained from replacement surgery due to OA. Directly after surgery, 50 mg of full-depth punches from the non-fibrillated lateral femur condyle was harvested. The cartilage punches were washed three times with 200 μl of PBS and three times with 200 μl of DMEM (Dulbecco's modified Eagle's medium, Gibco, catalogue number 41965). The punches were divided into halves and each was incubated with 500 μl of DMEM at 37°C for 2 h. The DMEM was collected (named 0 h baseline), and the cartilage pieces were then either incubated in 200 μl of digestion buffer containing MMP-3 (45 ng/mg of cartilage) or in 200 μl of MMP-DMEM solution (DMEM containing 45 ng of MMP-3/mg of cartilage) in a 96-well plate (Costar, catalogue number 3599) at 37°C. Medium (i.e. both DMEM and the digestion buffer) was collected after 3 h, and a fresh 200 μl aliquot of digestion buffer containing MMP-3 or MMP-DMEM solution was added to each cartilage piece which was incubated further. After 9 h from baseline, the second medium was collected, and a fresh 200 μl aliquot of digestion buffer containing MMP-3 or MMP-DMEM solution was added to each cartilage piece which was incubated further; after 16 h from baseline, the third medium was collected and the cartilage was harvested. To stop the protease reactions, EDTA was added to the medium (a final concentration of 20 mM), incubated for 20 min in room temperature (22°C) and then frozen.
Western blot analysis
Deglycosylated samples and molecular mass markers (10–250 kDa; Precision Plus Protein Standards, Bio-Rad Laboratories) were denaturated and separated on NuPAGE® 3–8% Tris-acetate SDS mini or midi gels, and Bis-Tris 4–12% SDS mini-gels (Invitrogen) as described previously . Proteins were electrophoretically transferred (Criterion Blotter, Bio-Rad Laboratories) on to PVDF membranes (Bio-Rad Laboratories, catalogue number 1620230) as described previously . Immunoreactions were performed as described previously  using anti-G1 (1.5 μg/ml), anti-ARGS (7 μg/ml), anti-FFGV [AF28, Millipore (0.5 μg/ml); BC14, MD Bioproducts (0.5 μg/ml); or BC14, Abcam (1:500)], anti-GVEE (sera, 1:500), anti-IPEN (sera, 1:10000), anti-EDLS (sera, 1:1000), anti-THLE (sera, 1:1000) and anti-G3 (2 μg/ml) antibodies, together with peroxidase-conjugated secondary antibodies of horse anti-mouse IgG (6.8–17 ng/ml) or goat anti-rabbit IgG (2.5–13 ng/ml). The immunobands were visualized using ECL (enhanced chemiluminescence) Plus (GE Healthcare, catalogue number RPN2132) or ECL SuperSignal West Femto (Thermo Scientific, catalogue number 34095) together with exposure to film (Amersham Hyperfilm ECL, catalogue number 28906837) or a luminescence image analyser (Fuji Film LAS-1000).
All of the amino acid numberings in the present paper are based on the full-length human aggrecan amino acid sequence starting with the N-terminal M1TTL and finishing with the C-terminal STAH2415  (NCBI accession number P16112). This aggrecan reference sequence has 29 VNTRs (variable number of tandem repeats) in the CS1 region and a G3 splice variant which includes all of the sub-domains [EGF (epidermal growth factor)-like, C-type lectin and Sushi domains]. All of the aggrecan fragments discussed in the present paper were compared with this reference sequence.
Quantification of aggrecan fragments by Western blot analysis
ARGS and FFGV standards were made by maximum conversion of sGAG-containing aggrecan G1 fragments (human OA cartilage A1D1 fraction) into ARGS fragments (with corresponding G1-TEGE) using ADAMTS4 digestion , or FFGV fragments (with corresponding G1-IPEN), using MMP-3 digestion (24 h at 37°C, 1.1 mol of substrate per mol of enzyme). According to our previous studies , the human OA cartilage A1D1 fraction contains only 39% of full-length G1–G3 monomers , therefore we used an aggrecan molecular mass of 1.5×106 g/mol which corresponds to 0.667 nmol of ARGS or FFGV fragments per mg of aggrecan (dry weight). The standards were deglycosylated and different amounts were loaded on to gels (Figure 5B and Supplementary Figure S3 at http://www.BiochemJ.org/bj/446/bj4460213add.htm). The Western blot analysis was conducted as described above and the standards and samples were visualized using an image analyser. The positioning and analysis of regions of interest of individual bands was performed using Fujifilm software Image Gauge version 3.2, and molar quantities were calculated from the linear range of the standard curve.
The Western blot linearity of the FFGV standard was 0.012–0.204 μg of dry weight aggrecan (mean R2=0.93, n=7 Western blots), and the FFGV inter-Western blot CV (coefficient of variation) was 27% (mean value from the SF-D1 pools and three Western blots). The linearity of ARGS standards and CV for the ARGS Western assay was as described previously .
MMP in vitro and in situ digestion of human aggrecan
The aggrecan A1D1 fraction prepared from human OA pool cartilage was used as a substrate for MMP-3 in vitro digestion. In these assays high molecular mass (>200 kDa) G1 fragments (G1–G3, G1–SELE/KEEE and G1–CS1; characterized previously ) were proteolysed to low molecular mass 52 kDa G1-IPEN and a 41 kDa G3 fragment (Figure 1A). Using a calculation model , the N-terminal of the 41 kDa G3 fragment was estimated to S2146IPA located at the end of the CS2 region. Using the knowledge of a similar G3 fragment, detected after MMP in vitro digestion of bovine aggrecan , the N-terminal of the human 41 kDa G3 fragment was further determined to T2117HLEI by a neoepitope anti-THLE antibody (Figure 1A), indicating an MMP cleavage site at RPAE2116↓2117THLE, generating a T2117HLE-G3 fragment.
MMP in vitro and in situ digestion of aggrecan
The AF28 anti-FFGV antibody detected FFGV-G3 (390 kDa) and FFGV-SELE/KEEE (321 kDa) fragments in the cartilage A1D1 aggrecan (Figure 1B), where the N-termini had been produced in vivo by MMP cleavage at the IPEN360↓361FFGV site. The C-terminal of these fragments was confirmed by Western blot analysis using anti-G3, anti-SELE and anti-KEEE antibodies (results not shown).
Western blot analysis showed that the AF28 anti-FFGV antibody was approximately 200-fold more sensitive against MMP-3 in-vitro-generated aggrecan fragments as compared with the BC14 anti-FFGV antibody (Supplementary Figure S2). Therefore in-vivo-generated FFGV aggrecan fragments in cartilage A1D1 and SF-D1 samples were not detectable in Western blot analysis with the anti-BC14 antibody (results not shown).
Both AF28 (Figure 1B) and BC14 (results not shown) antibodies detected five FFGV fragments in the MMP-3 in-vitro-digested A1D1 aggrecan sample: FFGV-CS1 (303 kDa), FFGV-CS1 (173 kDa), FFGV-CS1 (153 kDa), FFGV-KS/CS1 (128 kDa) and FFGV-GVEE (79 kDa). The 303, 173 and 153 kDa FFGV fragments had their C-termini estimated by the calculation model , suggesting MMP cleavage at GAED1470↓1471LSGL, GVED1028↓1029ISGL and GVGD952↓953LSGL respectively (Figure 2), MMP sites previously described in bovine aggrecan . The MMP cut which produced the C-terminal end of the 128 kDa FFGV fragment was estimated by the calculation model to be located in the border between the KS and CS1 regions (Figures 1B and 2), an MMP site previously indicated in bovine in vitro experiments .
MMP, aggrecanase and calpain cleavage sites in human aggrecan
The anti-GVEE antibody did not detect any fragments in the cartilage A1D1 aggrecan sample; although, after MMP in-vitro-digestion, two GVEE fragments with molecular masses of 79 and 134 kDa were produced (Figure 1B). The 79 kDa GVEE fragment had its N-terminal confirmed by the anti-FFGV antibody, whereas the N-terminal of the 134 kDa GVEE fragment was estimated to be located in the border between the KS and CS1 regions (Figures 1B and 2).
In situ digestion by the addition of MMP-3 to freshly prepared cartilage plugs (from arthroplasty) showed in the culture medium and in the cartilage tissue the same MMP-generated FFGV fragments as observed in the MMP-3 in vitro assays (Figures 1B and 1C). Similarly, the G1-DIPEN and the THLE-G3 fragments were generated by MMP-3 in situ digestion, and were detected in the medium (i.e. G1-IPEN and THLE-G3) and in the cartilage (i.e. G1-IPEN) (results not shown). Also, in these MMP-3 in situ digestions, the distribution of total amounts of FFGV fragments between the cartilage plugs and the medium was between 1:200 and 1:2000, suggesting that only a minor part of the FFGV fragments stay in the tissue.
An anti-EDLS antibody, recognizing both the calpain-generated EDLS neoepitope (SGVEDLS) and the internal EDLS sequence (SASGVEDLSRLPSG) located in the CS1 region of human aggrecan , detected two G1 fragments in the aggrecan A1D1 sample (Figure 1D). After MMP-3 in vitro digestion of the aggrecan sample, the anti-EDLS antibody failed to detect these epitopes (Figure 1D). Since no low molecular mass EDLS fragments were detected after MMP digestion (Figure 1D), this suggests that MMPs have a specific cleavage site within the human aggrecan sequence SASGVEDLSRLPSG1416, generating a cut at SASGVED1409↓1410LSRLPSG, which destroys the EDLS epitope. In support of this, almost identical sites [GVED↓I/(L)SGL] have been shown as MMP sites in bovine aggrecan .
These results suggest that human aggrecan has several MMP cleavage sites in different regions of the aggrecan molecule, identified both by in vitro digestion with purified aggrecan and by in situ digestion of aggrecan localized within the cartilage tissue.
There are six MMP cleavage regions in human aggrecan
Comparison of the MMP in-vitro-generated human aggrecan fragments detected by Western blot (Figure 1) with analysis of the human aggrecan reference amino acid sequence, suggested six separate MMP sites or regions (called a–f; Fig. 2): site-a at IPEN360↓361FFGV in the IGD; site-b at GVEE698↓699WIVT in the KS region; site-c at the border between the KS and CS1 regions; region-d at GVGD952↓953LSGL and at GVED1028↓1029ISGL in CS1; region-e at GVEE1294↓1295ISGL, GVEE1332↓1333ISGL, GVED1409↓1410LSRL and GAED1470↓1471LSGL in CS1; and site-f at RPAE2116↓2117THLE at the border between the CS2 region and the G3 domain. Interesting, five out of the six MMP cleavage sites/regions in human aggrecan (i.e. a, b and d–f) are adjacent to calpain or aggrecanase cleavage sites, suggesting alternative proteolysis options on the aggrecan molecule (Figure 2).
Using GVEDISGL as a consensus sequence for MMP digestion resulted in 26 potential cleavage sites (including minor sequence variations) for MMP in the CS1 region (Supplementary Figure S4 at http://www.BiochemJ.org/bj/446/bj4460213add.htm). However, only six of these sites, two in the beginning and four in the end of CS1, was cleaved by MMP-3 in vitro (Figure 2).
Time-dependent MMP digestion of aggrecan
Time-dependent in vitro proteolysis of aggrecan by MMP-3 showed fast degradation of G1–G3 monomers (reduced by 50% after approximately 10 min), and at the same time G1-IPEN, THLE-G3 and different FFGV fragments were produced (Figures 3A and 3B). The FFGV-GV(G/E)D and FFGV-GAED fragments were the most dominant MMP-generated FFGV products, and at maximum production levels (after 12–24 h digestion) these fragments were between 2- and 12-fold higher than the levels found for the FFGV-GVEE and FFGV-KS/CS1 fragments (Figure 3B). Higher or lower enzyme-to-substrate ratios compared with the standard in vitro digestion or longer digestion times did not generate any new aggrecan fragments (results not shown).
Time-dependent MMP in vitro digestion of aggrecan
MMP cuts aggrecan in a preferred order
To determine in which specific order MMP cuts aggrecan, we analysed (using OA cartilage A1D1 as the substrate) in vitro the initial kinetic speed of proteolysis and the time-dependent 20% increase (compared with the start level) of MMP-3-generated fragments. It took only 20 min for the THLE-G3 fragment to increase by 20%, whereas for the FFGV-GVEE fragment it took 6.4 h (Figure 4). From these analyses, the following preferred MMP cleavage order for human aggrecan was obtained: 1st, site-f at the border between the CS2 region and the G3 domain; 2nd, site-a in the IGD and multiple site-e in CS1; 3rd, double site-d in CS1; 4th, site-c at the border between KS and CS1 regions; and 5th, site-b in the KS region.
Determination of the preferred order of MMP cleavage in aggrecan
In A1D1 aggrecan, prepared from normal cartilage of a young subject, MMP-3 generated in vitro the same aggrecan fragments as obtained from OA cartilage, and the cleavage order resembled that of OA aggrecan (results not shown). The difference in MMP cleavage of aggrecan between young normal and adult OA cartilage was that in aggrecan prepared from normal cartilage the IGD cut (site-a) was the most preferred, whereas the CS2/G3 cut (site-f) was the second-most preferred MMP cleavage site.
Deglycosylation of aggrecan with chondroitinase, keratanase and keratanase II (A1D1 from OA cartilage) before MMP in vitro digestion, did not change the Western blot immuno-pattern, or the immuno-intensity, of the MMP-generated fragments (results not shown).
These results suggest that MMP has a preferred order of cleavage of aggrecan, and that the MMP activities against aggrecan are not affected by sGAG substitutions.
MMP-generated fragments in vivo
The THLE-G3 and KS/CS1-GVEE fragments, generated in vitro by MMP digestion (Figure 1), were not detected in vivo in any of the SF fractions (A1–A3, D1–D3) or in the cartilage fractions (A1D1–D3, A2 and A3) prepared from OA pools (results not shown). On the other hand, Gu-PG and A1D1–D3 fractions from OA and normal (youth and newborn) cartilage contained several FFGV fragments (Figure 5A). Most of these in-vivo-generated FFGV fragments had approximately the same molecular mass as was found for MMP in-vitro-digested aggrecan: FFGV-G3, 390 kDa; FFGV-CS2, 321 kDa; FFGV-CS1, 303 kDa; FFGV-CS1, 153–173 kDa; and FFGV-KS, 79 kDa (Figure 1B). Interesting, some of these FFGV fragments were also found in SF-D1 samples prepared from knee-injured, OA and knee-healthy subjects, and in the medium of human cartilage explant cultures (Figure 5B). However, the SF-FFGV fragments from the individual subjects showed molecular mass variations between individuals and compared with the in-vitro-generated FFGV fragments prepared from the OA cartilage pool (Figure 5B).
FFGV fragments detected by Western blot analysis in cartilage and SF
Compared with the FFGV fragments generated by MMP in vitro digestion of purified aggrecan (Figure 2), some additional fragments were detected in vivo. FFGV fragments with molecular masses between 130 and 190 kDa constituted a major part of the FFGV fragments found in SF, explant medium and cartilage (Figures 5A and 5B). The C-termini of these FFGV fragments were calculated, using the model , to be located within the CS1 region, suggesting that MMPs (site-d and -e) or calpains (close to MMP sites) have generated these C-termini. A 76 kDa FFGV fragment with a calculated C-terminal within the KS region, present in SF-D1 samples and in the cartilage A1D3 fraction (Figures 5A and 5B), did not react with the C-terminal neoepitope anti-GVEE (MMP site) and anti-PGVA (calpain site) antibodies (results not shown), indicating that the C-terminus was generated from a cut in an unknown cleavage site within the KS region. Calculation of the C-termini of the 49 and 57 kDa FFGV fragments (Figure 5A), detected in the cartilage A1D3 fraction, suggests a novel protease cleavage site within the G2 domain. In support of this, FFGV fragments with molecular masses of 30–50 kDa and 150–250 kDa have been detected in cytokine-stimulated porcine cartilage explant cultures .
These results suggest that the MMP cleavage sites in aggrecan found by MMP-3 in vitro and in situ digestions most likely also occur in vivo, as seen by the FFGV fragments found in cartilage and SF samples. These results also suggests that MMP cleavages at the IPEN↓FFGV site and at sites in the KS and CS regions of aggrecan occur both in adult, youth and newborn cartilage, and that FFGV fragments from these cuts migrate out into the SF.
Quantitative amounts of aggrecan, ARGS and FFGV fragments
Aggrecan A1D1–D3 cartilage fractions, prepared from the OA pool and from a normal (knee-healthy) youth subject, were quantified for the total amount of aggrecan estimated from sGAG values, and FFGV fragments using Western blot analysis and FFGV standards. The FFGV fragments in OA cartilage was approximately 3.4-fold the amounts found in normal youth cartilage, although the FFGV content in cartilage was below 1% out of the total aggrecan in both normal and OA cartilage (Table 1).
|Cartilage||sGAG (μg/μg)||Aggrecan (pmol/mg)||FFGV (pmol/mg)||Norm||FFGV (%)|
|SF pool||sGAG (μg/ml)||Aggrecan (pmol/ml)||FFGV (pmol/ml)||Norm||FFGV (%)||ARGS (pmol/ml)||Norm||ARGS (%)||ARGS/FFGV|
|Cartilage||sGAG (μg/μg)||Aggrecan (pmol/mg)||FFGV (pmol/mg)||Norm||FFGV (%)|
|SF pool||sGAG (μg/ml)||Aggrecan (pmol/ml)||FFGV (pmol/ml)||Norm||FFGV (%)||ARGS (pmol/ml)||Norm||ARGS (%)||ARGS/FFGV|
Next we compared MMP and aggrecanase cleavage in the IGD of aggrecan in vivo. Since aggrecanase-generated ARGS fragments (from cleavage at the TEGE392↓393ARGS site) are not detected by Western blot analysis in cartilage samples , the best way to compare the contribution of these proteases in aggrecanolysis is to measure the ARGS and FFGV fragments in SF (i.e. in the SF-D1 fraction). The amount of ARGS fragments found in the SF-D1 knee-healthy reference sample was approximately 2-fold the level of FFGV fragments, whereas in SF-D1 from knee-injured or OA patients, the ARGS levels were 47- and 82-fold higher respectively, compared with the levels of FFGV (Table 1). Also, the level of SF-ARGS increased approximately 14-fold in the injury and OA samples compared with the level found in the reference sample, whereas the levels of SF-FFGV slightly decreased in these samples (Table 1). In these SF samples, the proportion of FFGV fragments represented 1–8% out of the total aggrecan, whereas the proportion of ARGS fragments was much higher (Table 1). No ARGS and FFGV fragments were found (using Western blot analysis) in the SF-D2 and SF-D3 fractions (results not shown), indicating that the only ARGS and FFGV fragments present in SF are those that were analysed (Table 1, Figure 5B and Supplementary Figure S3). SF samples from different patient groups contain ARGS fragments that in Western blot analysis migrate into two distinct regions: 120–160 and 280–320 kDa (Supplementary Figure S3) [9,38]. Interesting, the major SF-FFGV fragments, detected by Western blot analysis (Figure 5B), also migrated into two similar molecular mass regions: 130–190 and 320–330 kDa. Importantly, no cross-reactivity between the anti-FFGV (AF28) and anti-ARGS antibodies was observed (results not shown).
These results suggest that MMP digestion in the IGD of aggrecan, seen as FFGV fragments, is present both in normal and OA cartilage. These data also suggests that the contribution of MMP cleavage in the IGD of aggrecan, generating FFGV fragments, in knee cartilage from injured and OA subjects is low compared with the contribution of aggrecanase cleavage in the IGD, generating ARGS fragments. Even though it has been shown in porcine experiments that MMP-generated FFGV fragments are not a substrate for aggrecanase cleavage in the TEGE↓ARGS site , we cannot rule out that the quantification of SF-FFGV and/or SF-ARGS fragments are underestimated due to further processing by exo- or endo-peptidases, and therefore escape detection by our antibodies. Also, even though the major SF-FFGV and SF-ARGS fragments detected by Western blot analysis have similar molecular masses, we cannot rule out that there could be a restriction in clearance from the joint cavity for one of the fragment types, due to other factors than fragment size, which will in this case affect the quantitative comparison of ARGS and FFGV fragments.
By Western blot analysis we identified six MMP cleavage sites (a–f) in human aggrecan, and these cuts were independent of GAG substitutions. Kinetic analysis showed a specific order of cleavage where site-f (at the border between CS2 and G3) was the most, and site-a and site-e (in the IGD and in CS1) the second-most preferred proteolytic site in aggrecan. MMP cleavage in the IGD and CS sites generated FFGV fragments which were detected in vivo in OA and normal cartilage, and in SF from normal, OA and knee-injured patients.
The aggrecan CS1 region contains a conserved VNTR of 19 amino acids, which exist in multiple allelic forms ranging from 13 to 33 repeats . VNTRs of 25–29 are the most common repeats corresponding to 96% of all allele forms . Owing to this CS1 polymorphism, the ends of some of the aggrecan fragments that had their N- (134 kDa GVEE) or C- (128, 153, 173 and 303 kDa FFGV) termini estimated by the calculation model  (using a VNTR=29) are shifted 76 amino acids (i.e. for VNTR=25) from what is proposed in Figure 2. Importantly, this shift from VNTR 29 to 25 will most probably not erase any of the MMP cleavage sites in the CS1 (i.e. site-d and -e), only move the position to similar cleavage sites. Individual differences in the molecular mass of aggrecan fragments has been observed in the ARGS-CS1 (molecular mass=120–160 kDa) [9,38] and in the FFGV-CS1 fragments (molecular mass=130–190 kDa) (Figure 5B), and can be explained by diversities in the CS1 sequence. Another explanation for molecular mass discrepancies in these aggrecan fragments could be due to differences in the CS substitutions within the CS1 repeats and/or in the CS2 region. Owing to these possible CS1 sequence and CS substitution variations between individuals, we only estimated these fragments as domain structures (e.g. FFGV-CS1; Figure 5B).
The AF28 anti-FFGV antibody detected several FFGV aggrecan fragments in neat and MMP-3 in-vitro-digested A1D1 cartilage and in SF from different subject groups (Figure 5). This is in line with our previous observations where anti-FFGV sera detected a similar pattern of FFGV aggrecan fragments in OA cartilage and SF samples . MMP-generated aggrecan FFGV fragments have also been detected in SF from different patient groups using ELISA and Western blot analysis together with the AF28 antibody [44,45].
The retention in the cartilage of FFGV fragments that lack the G3 domain is difficult to explain. ARGS fragments are detected in SF but not in the cartilage , therefore neither the G2 domain nor the KS- and CS-enriched regions appear to mediate tissue retention of FFGV fragments. The FFGV neoepitope and/or the N- and O-linked substitutions located between the MMP and aggrecanase site within the IGD could be the mediators of this retention [22,23]. In support of this, cartilage low molecular mass FFGV fragments (49, 57 and 76 kDa) migrated in the associative caesium chloride gradient to the bottom A1 fractions and were then collected in the A1D3 fractions (Figure 5A), which indicates an association between these FFGV fragments and hyaluronan, or other molecules binding to hyaluronan.
The opposing IPEN fragment, generated by MMPs after the IPEN360↓361FFGV cut, has been detected in normal and arthritic human cartilage by immunohistochemistry  and by Western blot analysis [19,25,36,46]. It has been shown that resident IPEN fragments accumulate in immature cartilage, reaching steady-state levels after an age of 25 . In view of this, considerable amounts of both MMP-generated IPEN and aggrecanase-generated TEGE fragments have been found in normal and arthritic knee cartilage seen by immunohistochemistry , and it has been shown that the relationship between G1-IPEN and G1-TEGE fragments is approximately 1:3 in both OA (i.e. pool) and normal cartilage (i.e. a knee-healthy youth) .
Deglycosylation with chondroitinase ABC and keratanase (I and II) did not affect the MMP activity against the different cleavage sites (a–f) of aggrecan. In line with this, the MMP activity against the bovine aggrecan IPEN↓FFGV site was also unaffected by the removal of sGAG prior to digestion . On the other hand, aggrecanase digestion against the bovine aggrecan TEGE↓ARGS site was abolished by keratanase treatment prior to digestion , and likewise the calpain digestion was abolished by chondroitinase ABC and keratanase deglycosylation prior to digestion of human aggrecan . These results suggest that the substrate regulation of MMP aggrecanolysis is different from that of aggrecanases and calpains.
In fact, there are several lines of evidence (discussed below) which point to a role for MMPs in general cartilage turnover or in the late stage of joint diseases, whereas aggrecanases have a role in the early part of pathological aggrecan degradation. (i) The MMP activity towards aggrecan is insensitive to sGAG substitutions, whereas aggrecanase activity against the IGD is abolished by deglycosylation. (ii) The MMP cut in the IGD site does not increase in the SF of OA patients and in the acute injury phase as compared with SF from knee-healthy references, which is opposite to what is found for cuts by aggrecanases [8,38]. Even though our MMP in situ digestion results suggest that the major part of FFGV is released into the medium, and in comparison only low amounts of the fragments are retained in the cartilage, still both OA and normal cartilage contain low, but measureable, amounts of FFGV fragments. (iii) The majority of mature aggrecan is normally truncated at the C-terminal end missing the G3 domain [48,49], and these truncations are due to MMP digestion, but also to digestion by calpains, cathepsins and aggrecanases [16,19,30]. (iv) Using bovine and porcine explant cultures, several reports suggest that MMPs play a major role in the late stages of cartilage degradation, whereas aggrecanases have a pivotal role in the beginning of the diseases [10,31,37,50]. Also, mouse studies suggest that MMP degradation of aggrecan correlates with severe cartilage damage occurring in the final stages of cartilage degradation .
It is interesting to note that the MMP cleavage sites in aggrecan detected in the present study are adjacent to calpain and aggrecanase sites in the IGD, KS and CS1 regions, and in the border between CS2 and the G3 domain (Figure 2). Also, it has been reported that cathepsin D cuts aggrecan at four positions in the CS regions which are close to the aggrecanase cleavage sites in CS2 . These cathepsin D sites are within sGAG-free regions of the aggrecan core. On the other hand, except for the MMP CS2/G3 site, the MMP cleavage sites in the CS1 region of aggrecan are all located at or close to a serine-glycine consensus sequence with a putative attached sGAG on the serine residue (Supplementary Figure S4). In the CS1 sequence presented (Supplementary Figure S4) there are 26 putative MMP cleavage sites, but only six of these sites turned out to be cleaved by MMP. Maybe the degree of CS substitution adjacent to these sites regulates the preference of MMP cleavage?
In addition to different interpretations of the results in the present study, the study has some general limitations: (i) statistical comparisons between subject groups were not possible since pooled samples were used; (ii) the Western blot quantification method has relatively high CVs and is therefore not as sensitive in measuring differences as methods such as ELISA, however, the advantage with Western blot is that it discriminates between different proteolytic fragments carrying the same epitope, allowing a more detailed quantitative analysis. (iii) normal cartilage was from the patella and obtained from only two subjects, and the cartilage explant was only from one subject; (iv) Durigova et al.  showed that MMP-12 is more efficient in cutting at the IGD site compared with MMP-3 , in our MMP in vitro analysis we only used the MMP-3 enzyme since it is believed to be a general MMP involved in cartilage degradation; and (v) the kinetic proteolysis studies were done with purified aggrecan and MMP-3, and it is possible that the specificity and kinetics of MMP against aggrecan in vivo is different compared with what was found in our in vitro experiments.
The strength of the present study is that: (i) all of the aggrecan fragments discussed were either verified using well-characterized aggrecan neoepitope or structural antibodies (using peptide blocking experiments), or by estimation using a calculation model ; and (ii) the individual and pooled SF samples, and cartilage samples from different subject groups are homogenous and well-characterized [9,17,38].
In the present study we have shown that human aggrecan contains several MMP cleavage sites, besides the well studied IPEN↓FFGV site, in different regions along the core protein, and that these cuts are regulated in a preferred order of cleavage. The MMP-generated FFGV fragment can be detected in SF and cartilage using monoclonal antibody AF28. Comparing the amounts of FFGV fragments between OA and normal cartilage, and between SF obtained from acute knee-injured, OA and knee-healthy subjects suggests that the main role of MMPs in mature articular cartilage is in normal cartilage turnover. However, from these data we cannot rule out that MMPs play an import role in other joint disease and/or joint injury stages not studied herein.
a disintegrin and metalloproteinase with thrombospondin motifs
coefficient of variation
Dulbecco's modified Eagle's medium
variable number of tandem repeat
Maria Hansson performed the experiments and analysed the results. André Struglics designed the study, analysed the results and wrote the paper.
We thank Dr Sanjay Kumar (GlaxoSmithKline, Collegeville, PA, U.S.A.) and Michael Pratta (Centocor, Malvern, PA, U.S.A.) for the gift of ADAMTS-4, antibodies and for the explant culturing samples. We also thank Dr Michael Lark (Trevena, King of Prussia, PA, U.S.A.) for the gift of MMP-3 and Dr Stefan Lohmander (Department of Orthopaedics, Lund University, Lund, Sweden) for critical reading of the paper before submission.
This work was supported by the Crafoord Foundation, the Faculty of Medicine Lund University, the King Gustaf V 80-year Birthday Fund, the Kock Foundation, the Swedish Rheumatism Association and the Alfred Österlunds Foundation (all to A.S.).