Activation of cartilage matrix metalloproteinases by activated protein C

Authors

  • Miriam T. Jackson,

    1. University of Sydney at Royal North Shore Hospital, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Margaret M. Smith,

    1. University of Sydney at Royal North Shore Hospital, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Susan M. Smith,

    1. University of Sydney at Royal North Shore Hospital, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Christopher J. Jackson,

    1. University of Sydney at Royal North Shore Hospital, St. Leonards, New South Wales, Australia
    Search for more papers by this author
    • Dr. C. J. Jackson is an inventor of two products involving activated protein C, for which the University of Sydney holds the patents.

  • Meilang Xue,

    1. University of Sydney at Royal North Shore Hospital, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Christopher B. Little

    Corresponding author
    1. University of Sydney at Royal North Shore Hospital, St. Leonards, New South Wales, Australia
    • Raymond Purves Research Laboratories, Level 10 Kolling Building, B6, University of Sydney at Royal North Shore Hospital, St. Leonards, New South Wales 2065, Australia
    Search for more papers by this author

Abstract

Objective

To investigate the role of activated protein C (APC) in cartilage degradation.

Methods

Chondrocyte expression of protein C, endothelial protein C receptor (EPCR), and thrombomodulin (TM) were evaluated by reverse transcription–polymerase chain reaction (RT-PCR). APC was immunolocalized in developing joints and in osteoarthritic (OA) cartilage from humans. The effect of APC on aggrecan and collagen degradation was examined in explant cultures of ovine cartilage in control cultures and in cultures stimulated with interleukin-1α (IL-1α), tumor necrosis factor α (TNFα), or retinoic acid (RetA), using colorimetric assays and Western blotting. Chondrocyte expression of matrix metalloproteinases (MMPs), ADAMTS, and tissue inhibitor of metalloproteinases (TIMPs) was measured by RT-PCR. MMP-2 and MMP-9 activity was evaluated by gelatin zymography and MMP-13 by fluorogenic assay.

Results

Positive cellular immunostaining for APC was found at sites of MMP activity in developing joints and in OA, but not normal, cartilage. Chondrocytes expressed messenger RNA for protein C, EPCR, and TM, with the latter 2 levels increased by IL-1α and TNFα stimulation. APC augmented aggrecan release and initiated collagen breakdown in IL-1α–treated and TNFα-treated cartilage, but not in normal or in RetA-treated cartilage. APC-stimulated aggrecan and collagen breakdown were due to MMP activity but were not associated with modulation of MMP, ADAMTS, or TIMP expression. APC resulted in MMP-13 activation in cartilage cultures. APC could not directly activate proMMP-13, but it was associated with increased MMP-2 and MMP-9 activity.

Conclusion

APC may be a relevant activator of MMPs in cartilage and may play a role in progressive cartilage degradation in arthritis.

Elevated levels of activated protein C (APC) have been found in the synovium and synovial fluid from the joints of patients with rheumatoid arthritis (RA), but the consequences of this increase have not been defined (1). APC is a plasma serine protease that binds to the cofactor protein S, leading to inactivation of clotting factors VIIIa and Va (for review, see ref. 2). It also acts as a feedback inhibitor of the coagulation cascade by inhibiting thrombin generation (3). APC is derived from its precursor, protein C, on the endothelial cell surface through binding to the endothelial protein C receptor (EPCR) (2). The activation peptide of the EPCR-bound protein C is released by thrombin that is bound to thrombomodulin (TM), and the serine protease domain is thus converted to its active conformation (2).

In addition to its anticoagulant role, APC has significant antiinflammatory properties, such as decreasing proinflammatory cytokine synthesis and reducing leukocyte recruitment (4). In vivo, APC protects against the lethal effects of endotoxemia in animal models and suppresses lipopolysaccharide-induced tumor necrosis factor α (TNFα) production (5). APC accelerates cutaneous wound healing due to increased cell proliferation and migration, and it attenuates keratinocyte apoptosis (4). These antiinflammatory responses of APC are EPCR-dependent and are largely mediated by cleavage of the protease-activated receptors (PARs) (6). While PAR-1 activation is antiinflammatory in some tissues (7), it is associated with disease progression in others (8). In the context of arthritis, PAR-1 activation may play a role in promoting joint inflammation and associated cartilage damage (9). Significantly, APC, unlike thrombin, can also cleave PAR-2, which is increased in osteoarthritic (OA) cartilage and, when activated, leads to the up-regulation of matrix metalloproteinases (MMPs) implicated in cartilage degradation (10, 11).

Increased APC levels correlate with MMP-2 activity in RA joints, and APC up-regulates MMP-2 expression and activation in a number of cells types (1, 12). MMP-2–knockout mice show increased inflammation and associated proteoglycan depletion from cartilage in an inflammatory model of arthritis, suggesting an antiinflammatory role of MMP-2 in RA (13). Within cartilage however, MMP-2 acts as part of a cascade for the activation of collagenases, and MMP-2 expression and protein levels are elevated in association with increased collagen breakdown in human OA cartilage (14, 15). In light of these conflicting observations, it remains unclear what role elevated levels of APC may play in the initiation and progression of arthritis. In particular, the effect of APC on cartilage pathology and as a potential activator of MMPs in this tissue has not been investigated. It is important to differentiate between the potential beneficial effects of APC in reducing joint inflammation from possible deleterious actions in promoting cartilage degradation, and this is the focus of the current investigation.

MATERIALS AND METHODS

In vitro models of cartilage degeneration.

Full-depth articular cartilage explants of ∼5 mm2 (average wet weight 40 mg) were removed from the trochlear groove of 6-month-old ovine knee joints and placed in Dulbecco's modified Eagle's medium (DMEM; Sigma, Castle Hill, New South Wales, Australia) buffered with 3.7 gm/liter of sodium bicarbonate (Fronine, Riverstone, New South Wales, Australia) and containing 10% (volume/volume) fetal calf serum (FCS; Trace Biosciences, Castle Hill, New South Wales, Australia), 2 mML-glutamine (ICN Biochemicals, Aurora, OH), and 50 μg/ml of gentamicin (Pharmacia, Bentley, Western Australia, Australia). The explants were allowed to equilibrate for 48 hours at 37°C in an atmosphere of 5% (v/v) CO2, after which the FCS was removed by 3 washes (5 minutes each at 37°C) in serum-free DMEM.

For the majority of experiments, cartilage explants were subsequently cultured individually for 4 days in 48-well culture plates containing 500 μl of serum-free DMEM in the presence or absence of 10 ng/ml of interleukin-1α (IL-1α; PeproTech, Rocky Hill, NJ), 100 ng/ml of TNFα (PeproTech), or 10–6M retinoic acid (RetA; Sigma) and in the presence or absence of 0.2–20 μg/ml of APC (drotrecogin alfa [activated]; Eli Lilly, Indianapolis, IN). In some cultures, a broad-spectrum MMP inhibitor (PGE3162689; synthesized at Procter & Gamble, Cincinnati, OH) was added to the medium at a concentration of 300 nM, the concentration at which PGE3162689 has been shown to inhibit MMPs 1, 2, 3, 7, 8, 9, 13, and 14, but not ADAMTS, in cartilage cultures (16).

Time course of APC effects.

To determine whether IL-1α and APC needed to be present simultaneously and how long after IL-1α treatment chondrocytes continued to respond to APC, cartilage explants were stimulated with IL-1α (10 ng/ml) for 24 hours. After this time, the IL-1α was washed from the cultures (3 washes in serum-free DMEM for 10 minutes each), and the cartilage was maintained in DMEM for a further 24 hours, 96 hours, 7 days, or 14 days. In the extended cultures, the medium was changed every 3–4 days.

After each of these time periods, the cartilage was then incubated for a further 4 days with DMEM containing 20 μg/ml of APC. The conditioned medium from this final 4 days of culture with APC was collected and stored at −20°C until further analyzed. The cartilage explants were blotted dry, weighed, and either snap-frozen in liquid nitrogen for RNA extraction or papain-digested for biochemical analysis.

Quantification of proteoglycan and collagen release.

The proteoglycan content of the cultured medium and papain digests was measured as sulfated glycosaminoglycan (sGAG) by colorimetric assay using dimethylmethylene blue (Sigma) and using chondroitin sulfate C from shark cartilage (Sigma) as a standard (17, 18). Collagen in the medium and papain digests was measured using a simplified microtiter plate adaptation of the hydroxyproline assay described by Stegemann and Stalder (19). Release of sGAG and hydroxyproline into the medium was expressed as a percentage of the total (media plus papain digest of the explant). In some cases, when the explant was used for other analyses (e.g., RNA extraction), release of sGAG or hydroxyproline into the medium was expressed as micrograms per milligram of wet weight cartilage.

Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting.

Volumes of conditioned media that gave equal concentrations (30 μg) of sGAG were deglycosylated overnight at 37°C with 0.01 unit of chondroitinase ABC, 0.01 units of keratanase I, and 0.0001 units of keratanase II (Seikagaku Kogyo, Tokyo, Japan) in 0.1M Tris acetate buffer (pH 6.5). Samples were then dialyzed against ultrapure (Milli-Q) water at 4°C overnight and freeze-dried. Samples were separated electrophoretically under reducing conditions on 10% Bis-Tris Gels (Invitrogen, Melbourne, Victoria, Australia), transferred to nitrocellulose membranes (Invitrogen), blocked with 5% (weight/volume) bovine serum albumin (Sigma), and immunoblotted overnight with a 1:100 dilution of monoclonal antibodies BC-3 (anti-ARG...) or BC-14 (anti-FFG...) (gifts from Prof. B. Caterson, Cardiff University, Cardiff, UK) as described previously (20).

RNA extraction and reverse transcription–polymerase chain reaction (RT-PCR).

Total RNA was extracted from ovine articular cartilage explants or human OA cartilage and isolated using RNeasy Mini Columns and reagents (Qiagen, Valencia, CA) with an on-column DNase digestion as previously described (21). Total RNA was quantified using SYBR Green II (Cambrex Bioscience, Rockland, ME), with ribosomal RNA as a standard, and 0.5–1 μg of total RNA from samples within groups for comparison underwent simultaneous RT using the Omniscript kit (Qiagen) according to the manufacturer's instructions, with the inclusion of 1 units/μl of RNase Guard (Amersham Biosciences, Buckinghamshire, UK). Semiquantitative PCR was performed as described previously (21), using a PTC-100 Peltier thermal cycler (GeneWorks, Adelaide, South Australia, Australia) and previously described primers (22–24). A single product was identified for all primers, and the specificity of the PCR was confirmed by sequencing.

MMP activity.

MMP-2 and MMP-9 protein secretion and activation in pooled culture medium from triplicate cultures was measured using gelatin zymography under nonreducing conditions as described elsewhere (25). MMP-13 activity in the conditioned medium was detected using a SensoLyte Plus 520 MMP-13 assay kit (AnaSpec, San Jose, CA) according to the manufacturer's instructions. ProMMP-13 used as a positive control was generously provided by Prof. Gillian Murphy (Cambridge University, Cambridge, UK).

Immunohistochemistry.

Sections (4 μm) from archival formalin-fixed, decalcified, paraffin-embedded specimens (medial tibial plateaus from 4 patients undergoing total knee replacement for OA and joints from 12–14-week-old human fetuses; used with approval of the Northern Sydney Health Human Research Ethics Committee, Sydney, New South Wales, Australia) were cut using a rotary microtome and attached to Superfrost Plus microscope slides. Sections were deparaffinized in xylene and rehydrated through graded alcohols. Immunohistochemical evaluation was performed essentially as previously described (26). Sections were pretreated with protease-free chondroitinase ABC (0.1 unit/ml; Sigma) and keratanase I (0.1 unit/ml: Seikagaku Kogyo) prior to overnight incubation with monoclonal anti-human protein C (Sigma) at 5 μg/ml and with equivalent concentrations of mouse IgG as a negative control.

Statistical analysis.

Data were analyzed using the Mann-Whitney U test. All analyses were performed using Stata SE 10.1 software (StataCorp, College Station, TX). The alpha level was set at 0.05, and the Benjamini-Hochberg post hoc correction for multiple comparisons was performed to correct for false-positive results (27).

RESULTS

Chondrocyte expression of protein C at sites of cartilage resorption.

Protein C/APC was immunolocalized to chondrocytes and their immediate pericellular matrix in human fetal tissues at sites of joint cavitation and vascular invasion of the physis and in hypertrophic chondrocytes in the growth plate (Figure 1A). Chondrocytes in human articular cartilage in areas of fibrillation but not in nonfibrillated regions in OA joints showed positive protein C/APC staining (Figure 1A).

Figure 1.

Presence of protein C, endothelial protein C receptor (EPCR), and thrombomodulin (TM) in chondrocytes. A, Immunolocalization of protein C/activated protein C (APC) in developing human fetal joints and in human osteoarthritic (OA) cartilage. Protein C/APC was immunolocalized to sites of joint cavitation and vascular invasion of the epiphysis and in hypertrophic chondrocytes in the growth plate of developing human fetal joints. Toluidine blue–stained macroscopic section of a tibial plateau from a patient with OA (left) demonstrates similar cartilage proteoglycan loss in regions with an intact (a) and fibrillated (b) surface. Immunolocalization of protein C/APC in these 2 regions (right) demonstrates positive staining of chondrocytes in the fibrillated cartilage. (Magnification × 100.) B, Semiquantitative reverse transcription–polymerase chain reaction (RT-PCR) showing expression of protein C, EPCR, and TM in ovine and human articular cartilage chondrocytes. C, Semiquantitative RT-PCR of TM and EPCR gene expression in ovine articular cartilage cultured for 4 days with interleukin-1α (IL-1α; n = 9), tumor necrosis factor α (TNFα; n = 3), or retinoic acid (RetA; n = 3 without APC and n = 8 with APC) in the presence or absence of APC. Values are the mean and SD. = P < 0.05 versus corresponding control (i.e., with or without APC; n = 12).

The positive immunostaining well-removed from blood vessels suggested that chondrocytes themselves might be synthesizing protein C. Semiquantitative PCR analysis of messenger RNA (mRNA) isolated from ovine cartilage explants and human OA cartilage (Figure 1B) confirmed that these cells express mRNA for protein C, EPCR, and TM.

TM expression in ovine chondrocytes was significantly increased by stimulation with 10 ng/ml of IL-1α, 100 ng/ml of TNFα, and 10–6M RetA as compared with control cultures. APC did not modulate TM expression when added alone or in combination with IL-1α or TNFα (Figure 1C). Stimulation with IL-1α or TNFα increased EPCR expression, with the difference reaching statistical significance for IL-1α plus APC and for TNFα with or without APC as compared with their respective controls (Figure 1C). Expression of protein C mRNA by ovine chondrocytes was not significantly regulated by APC, IL-1α, TNFα, or RetA (data not shown).

APC augmentation of cytokine-stimulated proteoglycan and collagen degradation.

Since all components of the protein C pathway that can lead to the activation of protein C to APC are present in cartilage, we examined whether this serine protease may contribute to cartilage degradation. APC induced a dose-dependent (0.2–20 μg/ml) augmentation of sGAG and hydroxyproline release from cartilage over 4 days of culture in the presence of 10 ng/ml of IL-1α (Figures 2A and B, respectively). Preliminary studies comparing sGAG release at 8 hours and at 1, 2, 3, and 4 days indicated the effect of APC was not evident until 2 days and was maximal at 4 days (data not shown). While significant induction of sGAG release above that induced by IL-1α alone was not observed with APC concentrations of <1 μg/ml (Figure 2A), significant collagen release was stimulated with as little as 0.2 μg/ml of APC (Figure 2B). Collagen release over 4 days was significantly greater with 20 μg/ml of APC than with lower doses (Figure 2B). However, continual exposure of cartilage to 0.2 μg/ml of APC in the presence of IL-1α for 21 days, resulted in cumulative collagen release (mean ± SD 3.3 ± 4.4 μg/ml) equivalent to that observed after 4 days with 20 μg/ml of APC (Figure 2B).

Figure 2.

Effect of activated protein C (APC) and interleukin-1α (IL-1α) on the release of sulfated glycosaminoglycan (sGAG) and hydroxyproline (HyPro) from cultured ovine articular cartilage. Dose-response effect of A and B, APC (0.2–20 μg/ml; n = 6) and C and D, IL-1α (0.01–10 ng/ml; n = 6) on the release of sGAG and hydroxyproline after 4 days of culture. Results are expressed as μg/mg of cartilage wet weight, since the explants were used for RNA extraction and gene expression analysis. Release of E, sGAG (n = 3) and F, hydroxyproline (n = 3) from ovine articular cartilage explants preincubated with or without 10 ng/ml of IL-1α for 24 hours, followed by a further 24 or 96 hours of culture of all explants in control medium, and then APC treatment for a further 4 days. Results are expressed as the percentage of total sGAG or total hydroxyproline, respectively. Values are the mean and SD. = P < 0.05 versus control and, as shown by the connecting horizontal bars, between the indicated treatments; # = P < 0.05 versus the same cytokine treatment without APC.

In the absence of IL-1α, incubation with APC even at the highest concentration (20 μg/ml) did not directly stimulate proteoglycan or collagen release from cartilage (Figures 2C and D). The effect of treatment with IL-1α alone on sGAG release was dose-dependent, being maximal at 10 ng/ml (Figure 2C). APC at 20 μg/ml significantly increased the IL-1α–induced sGAG release at all doses of IL-1α tested, with maximal effects at 10 ng/ml of IL-1α, but no significant differences between 10, 1, and 0.1 ng/ml of IL-1α (Figure 2C). APC was able to stimulate collagen release from cartilage coincubated with 0.01–10 ng/ml of IL-1α, with no significant differences observed between the 3 highest doses of IL-1α (Figure 2D).

To investigate the mechanisms whereby APC has its effects within a reasonable time period and to avoid issues of chondrocyte dedifferentiation and long-term IL-1α exposure, the doses of IL-1α and APC (10 ng/ml and 20 μg/ml, respectively) that induced maximal release of sGAG and collagen over 4 days were used for further experimentation.

IL-1α and APC do not need to be present simultaneously.

The effect of APC on proteoglycan and collagen proteolysis required living chondrocytes, since no effect was seen with freeze–thawed cartilage (data not shown). The effect of APC did not require coincubation with IL-1α, as shown by experiments in which cartilage was incubated for 24 hours with IL-1α, washed, cultured in serum-free medium for 1–14 days, and then cultured for a final 4 days in fresh cytokine-free medium with or without APC, after which sGAG and collagen release were measured. In the absence of APC, sGAG release in IL-1α–pretreated explants was elevated above that in explants that had never been incubated with cytokine (Figure 2E). Addition of APC to explants that had been precultured with IL-1α significantly augmented the release of sGAG when it was added 24 or 96 hours after IL-1α pretreatment (Figure 2E).

Hydroxyproline release was increased by the addition of APC to control cultures after both 24 and 96 hours. However, at both time points when APC was added following pretreatment with IL-1α, hydroxyproline release was significantly increased compared with both control and IL-1α–preincubated cultures (Figure 2F). The stimulation of sGAG release by the addition of APC for 4 days was still significant following further incubation for up to 14 days in the absence of IL-1α (mean ± SD 63.6 ± 26.4% versus 7.70 ± 1.03% in control cultures). Hydroxyproline release was significantly stimulated by the addition of APC on day 7 after removal of the cytokine (mean ± SD 2.05 ± 1.50% versus 0.07 ± 0.01% in control cultures) but not on day 14 (0.29 ± 0.41% versus 0.08 ± 0.02% in control cultures).

To determine whether the effects of APC were specific to IL-1α, we compared the results in replicate cultures stimulated with 10 ng/ml of IL-1α, 100 ng/ml of TNFα, or 10–6M RetA in the presence or absence of 20 μg/ml APC (Figure 3). In these cultures, APC alone did not significantly increase the release of sGAG or collagen. IL-1α, TNFα, and RetA alone significantly increased sGAG, but not collagen, release. In cultures treated with IL-1α or TNFα, but not RetA, APC significantly increased both sGAG and collagen release above the release with the catabolic agent alone.

Figure 3.

Release of A, sulfated glycosaminoglycan (sGAG; n = 9) and B, hydroxyproline (collagen; n = 9) from ovine articular cartilage explants cultured for 4 days with 10 ng/ml of interleukin-1α (IL-1α), 100 ng/ml of tumor necrosis factor α (TNFα), or 10–6M retinoic acid (RetA) in the presence or absence of 20 μg/ml of activated protein C (APC). Results are expressed as the percentage of total sGAG or total hydroxyproline, respectively. Values are the mean and SD. = P < 0.05 versus corresponding control (i.e., with or without APC); # = P < 0.05 versus the same catabolic agent without APC.

MMP activity and APC-augmented aggrecan and collagen release.

To determine whether the APC-stimulated augmentation of aggrecan and collagen release from the cartilage explants was due to the action of MMPs, a broad-spectrum MMP inhibitor (PGE3162689) was included in cultures with IL-1α in the presence or absence of APC. MMP inhibition did not significantly modulate sGAG release from control or IL-1α–stimulated cultures, but significantly reduced the release of sGAG from cartilage cocultured with IL-1α plus APC down to the levels observed with IL-1α alone (Figure 4A). PGE3162689 almost completely inhibited hydroxyproline release from cartilage cultured with IL-1α plus APC (Figure 4B).

Figure 4.

Matrix metalloproteinase (MMP) activity and the augmentation of cytokine-stimulated aggrecan and collagen release by activated protein C (APC). Ovine articular cartilage explants were treated for 4 days with interleukin-1α (IL-1α) in the presence or absence of APC and in the presence or absence of a broad-spectrum MMP inhibitor (MMPI), and release of A, sulfated glycosaminoglycan (sGAG; n = 5) and B, hydroxyproline (HyPro; n = 5) into the medium was determined. Results are expressed as the percentage of total sGAG or total hydroxyproline, respectively. Values are the mean and SD. = P < 0.05 versus control; # = P < 0.05 versus IL-1α without APC. C, Western blot analysis was performed on aggrecan catabolites released into the medium from cartilage explants that had been treated for 4 days with IL-1α, tumor necrosis factor α (TNFα), or retinoic acid (RetA) in the presence or absence of APC. Neoepitope antibodies BC-3 (anti-ARG...) and BC-14 (anti-FFG...) were used to distinguish between ADAMTS generated and MMP-generated fragments, respectively. Migration positions of prestained molecular mass markers are shown on the left.

Consistent with these data, proteoglycan release from ovine articular cartilage stimulated with either IL-1α, TNFα, or RetA was associated with increased ADAMTS cleavage of aggrecan and the release of high molecular mass fragments with an ARG... N-terminus (BC-3 immunoblot in Figure 4C), as previously described for other species (20). In association with APC-augmented sGAG release in cultures treated with IL-1α or TNFα, but not RetA, there was a decrease in the size of the BC-3–positive metabolites and detection of fragments initiating with the FFG... neoepitope generated by MMPs (BC-14 immunoblot in Figure 4C).

We investigated whether the increase in aggrecan and collagen proteolysis in cultures stimulated with cytokine plus APC, but not RetA plus APC, was associated with altered and/or differential expression of MMPs 1, 2, 3, 9, 13, and 14, ADAMTS 1, 4, and 5, and TIMPs 1 and 3. APC alone did not significantly modulate mRNA expression of any of the genes investigated as compared with control cultures (Table 1). Each of the 3 catabolic agents increased the expression of MMP-1 (20–30-fold) and MMP-2 (∼4-fold) to a similar extent. The increases in MMP-9 and MMP-13 (2–5-fold) expression were not significant with any catabolic agent. MMP-14 was modestly (∼2-fold), but not significantly, increased by IL-1α and RetA, while MMP-3 was significantly up-regulated by IL-1α (17-fold) but not RetA. There was a differential regulation of ADAMTS enzymes, with IL-1α and TNFα significantly increasing ADAMTS-4 (20–26-fold) and ADAMTS-5 (5–20-fold). None of the agents significantly modulated TIMP-1, whereas TIMP-3 was decreased by IL-1α (by 75%), with this reduction being partially abrogated by APC, although it remained significantly below control values.

Table 1. Results of semiquantitative reverse transcription-polymerase chain reaction analysis of ovine cartilage*
GeneAPCIL-1αIL-1α plus APCTNFαTNFα plus APCRetARetA plus APC
  • *

    Values are the mean ± SD fold change as compared with control cultures. Numbers in parentheses are the number of samples tested. APC = activated protein C; IL-1α = interleukin-1α; TNFα = tumor necrosis factor α; RetA = retinoic acid; NT = not tested.

  • P < 0.05 versus control cultures.

  • P < 0.05 versus corresponding cytokine without APC.

MMP-10.22 ± 0.12 (8)35.1 ± 9.04 (5)23.3 ± 9.58 (6)30.0 ± 2.58 (3)12.63 ± 2.67 (3)22.14 ± 13.4 (6)13.1 ± 4.68 (3)
MMP-21.39 ± 1.39 (12)4.47 ± 3.60 (8)3.77 ± 3.03 (9)3.88 ± 1.52 (3)4.87 ± 1.58 (3)3.61 ± 2.58 (8)2.54 ± 0.86 (3)
MMP-30.85 ± 0.76 (9)17.3 ± 11.7 (9)14.9 ± 9.91 (9)NTNT0.78 ± 0.47 (8)1.04 ± 0.39 (3)
MMP-90.62 ± 0.39 (12)1.94 ± 1.17 (7)0.87 ± 0.40 (7)5.08 ± 5.86 (3)3.24 ± 4.31 (3)4.27 ± 4.99 (7)9.13 ± 4.39 (3)
MMP-130.86 ± 0.69 (8)2.14 ± 2.35 (8)0.78 ± 0.64 (8)2.57 ± 1.98 (3)0.81 ± 0.60 (3)5.09 ± 5.20 (8)11.34 ± 5.66 (3)
MMP-141.12 ± 0.39 (6)1.61 ± 0.53 (6)1.21 ± 0.35 (6)NTNT2.26 ± 0.50 (6)2.74 ± 0.47 (3)
ADAMTS-10.89 ± 0.79 (8)0.75 ± 0.52 (6)1.42 ± 2.06 (5)0.35 ± 0.05 (3)0.49 ± 0.10 (3)24.2 ± 20.4 (5)NT
ADAMTS-40.91 ± 0.51 (9)21.1 ± 18.5 (6)14.6 ± 16.7 (5)26.0 ± 7.33 (3)20.5 ± 8.59 (3)3.33 ± 4.16 (5)NT
ADAMTS-50.45 ± 0.41 (6)5.00 ± 2.15 (3)5.00 ± 0.39 (3)20.6 ± 6.99 (3)19.0 ± 6.50 (3)15.9 ± 7.87 (3)NT
TIMP-11.28 ± 0.52 (5)0.43 ± 0.29 (3)0.49 ± 0.46 (3)1.04 ± 0.17 (3)1.15 ± 0.25 (3)2.36 ± 0.65 (3)NT
TIMP-31.04 ± 0.3 (9)0.27 ± 0.15 (9)0.48 ± 0.15 (9)NTNT1.14 ± 0.47 (7)NT

In general, there was little effect of APC on gene expression when added to catabolically stimulated cultures. APC down-regulated MMP-1 expression under all experimental conditions, but this did not reach statistical significance. APC reduced MMP-2 expression in RetA- and IL-1α–treated cultures, and reduced MMP-9 and MMP-13 mRNA in IL-1α– and TNFα-treated cultures. Given the observed variance in gene expression, the current analyses were underpowered to detect significant differences in the expression of ADAMTS-1, MMP-2, MMP-13, and MMP-14 (>80% power to detect all other genes). Nevertheless, the effect of APC in augmenting aggrecan and collagen proteolysis and release from cartilage did not appear to be associated with up-regulated expression of key enzymes or with down-regulated expression of inhibitors.

MMP-2, MMP-9, and MMP-13 activation in APC/cytokine costimulated cultures.

In control cultures, proMMP-2 and proMMP-9 were present, with some active MMP-2, but not MMP-9, also being detected (Figure 5A). IL-1α, TNFα, or RetA alone failed to increase the levels of active MMP-2 or MMP-9. The addition of APC with or without IL-1α, TNFα, or RetA increased active MMP-2 levels in conditioned medium (Figure 5A). The relative increase in active MMP-2 was not as high in the RetA-treated cultures as in those treated with IL-1α or TNFα. The addition of APC to control or cytokine-stimulated cultures resulted in 100% activation of proMMP-9, with only partial activation evident in RetA-treated cultures (Figure 5A).

Figure 5.

Matrix metalloproteinase (MMP) activity in conditioned medium from ovine articular cartilage explants cultured for 4 days with interleukin-1α (IL-1α), tumor necrosis factor α (TNFα), or retinoic acid (RetA) in the presence or absence of activated protein C (APC). A, MMP-2 and MMP-9 activity in conditioned medium, as determined by gelatin zymography. Migration positions of the pro forms and active forms of MMP-2 and MMP-9 are shown on the left. B, MMP-13 activity in conditioned medium from cultures incubated with or without APC, following further incubation for 1 hour in the presence or absence of the MMP activator APMA (n = 6). Values are the mean and SD fluorescence ratio. = P < 0.05 versus corresponding control (i.e., with or without APC); # = P < 0.05 versus the same catabolic agent without APC; ∧ = P < 0.05 versus the same catabolic agent without APMA. C, Activity of proMMP-13 incubated for 240 minutes either alone or in the presence of either APMA or 20 μg/ml of APC. Values are the mean ± SD of 2 experiments.

Using a fluorescence activity assay, no MMP-13 activity was detected in media under control conditions or in the presence of APC, IL-1α, TNFα, or RetA (Figure 5B). When the media were incubated with the generic MMP activator APMA, there was a very small increase in MMP-13 activity in control and Ret A–stimulated cultures, whereas there was a marked increase in IL-1α–stimulated and TNFα-stimulated cultures, indicating that these 2 cytokines up-regulated proMMP-13 synthesis. MMP-13 activity was seen when APC was included in cultures treated with IL-1α and with TNFα, but not RetA, with no further increase in activity with the addition of APMA.

APC was unable to directly activate proMMP-13 in vitro when incubated for >4 hours at 37°C, as compared with the rapid activation by APMA used as a positive control (Figure 5C). Even when cultures were incubated for 24 hours, there was no difference between proMMP-13 alone or in the presence of 20 μg/ml APC (data not shown). These results demonstrate that although APC cannot directly activate proMMP-13, its addition to IL-1α–stimulated or TNFα-stimulated cultures does result in increased MMP-13 activity, which correlated with the augmented sGAG and collagen proteolysis (r = 0.73 and r = 0.66, respectively, P < 0.01 for each comparison).

DISCUSSION

Activated protein C was initially identified as an anticoagulant; however, recent studies have revealed that it has important antiinflammatory properties (4). We have now shown that APC may be detrimental to articular cartilage through its promotion of the degradation of aggrecan and collagen by activating MMPs.

Expression profile studies have shown that articular chondrocytes express EPCR (in the mouse [28]) and TM (in the rat [29]), with TM also being increased in surgically induced OA in the rat. The results of the present study extend these previous observations to show the expression of protein C, EPCR, and TM in both human and ovine articular chondrocytes. Together with previous studies demonstrating that chondrocytes express thrombin (30), it is apparent that these cells express all of the components necessary for local generation of APC in tissues distant from the circulatory system. Furthermore, our immunostaining data showed increased chondrocyte protein C/APC in regions of active cartilage turnover during growth and during OA. Taken together, these findings strongly implicate APC as a potential activator of MMPs in cartilage.

The collagenolysis and generation of FFG-bearing aggrecan fragments along with C-terminal truncation of ADAMTS-generated ARG... fragments observed in explants stimulated for 4 days with APC addition to IL-1α or TNFα (Figure 4C) mimics the features previously observed in prolonged (3–4-week) IL-1α–stimulated bovine nasal and equine articular cartilage cultures (16, 31). It is interesting that the augmented sGAG release in APC plus IL-1α cultures could be abrogated by PGE3162689 treatment (Figure 4A), suggesting that aggrecan cleavage by MMPs was occurring in the cartilage, rather than after its release into the medium. Furthermore, APC was able to activate proMMPs sequestered in the cartilage matrix for up to 2 weeks after IL-1α stimulation in the absence of accelerated ADAMTS-driven aggrecanolysis.

The addition of APC to cultures stimulated with IL-1α or TNFα is a useful in vitro model and offers the potential to investigate the role of proMMP activation in cartilage aggrecan breakdown. Culture of mouse femoral head explants has proved useful for the evaluation of mechanisms of aggrecan proteolysis in wild-type and genetically modified mice (32, 33). As in the current ovine cartilage cultures, we have found that APC activates significant collagenolysis in mouse femoral head explant cultures (Jackson CM, et al: unpublished observations), and this may therefore prove a useful system in which to study the role of specific proteinases in cartilage collagen breakdown using genetically modified mice.

We used APC at a concentration of 20 μg/ml for most experiments in order to maximize MMP activation; however, increased hydroxyproline release was seen over 4 days with as little as 200 ng/ml, which is near the concentration observed in synovial fluid from OA joints (mean ± SD 136 ± 42 ng/ml) and well below that in synovial fluid from RA joints (462 ± 112 ng/ml) (1). Moreover, prolonged exposure to concentrations of APC found in arthritic human joints (200 ng/ml) ultimately induced total collagenolysis, similar to that after 4 days of 20 μg/ml of APC, supporting the potential relevance of this MMP-activation mechanism in cartilage in vivo.

The mechanism whereby APC activates MMPs and, in particular, collagenases in these cartilage cultures remains to be established. The MMP-generated collagen release in cultures containing APC plus IL-1α or APC plus TNFα correlated with MMP-13 activity in the medium. This is consistent with the data suggesting that MMP-13 is the principal collagenase in cartilage (34) and with its superior efficiency in cleaving type II collagen (35). We cannot rule out a role of MMP-1 in collagenolysis in the present studies, since all catabolic agents increased MMP-1 mRNA levels after 4 days of culture (Table 1). However, the reduced MMP-1 expression with addition of APC is not consistent with this enzyme being responsible for collagenolysis in these cultures. MMP-13 expression was not significantly increased by any of the catabolic agents, but as noted above, the PCR analysis may be underpowered to detect a difference. However, it may also be a temporal effect associated with expression levels only being acquired at the conclusion of the 4-day culture period, as a previous study showed a peak increase in MMP-13 mRNA with IL-1α and oncostatin M stimulation at 12 hours before decreasing to baseline levels (36). Further analysis of the temporal change in gene expression with a larger number of samples is required to determine if a similar pattern is operant in our ovine explant cultures.

Importantly, APC was unable to directly activate proMMP-13 (Figure 5C), suggesting that additional activation steps were required. APC has been shown to up-regulate and activate proMMP-2 in skin fibroblasts and human umbilical vein endothelial cells (4, 37), and MMP-2 is an efficient activator of MMP-13 (38). We also demonstrated significant activation of proMMP-2 in our cartilage cultures in the presence of APC (Figure 5A). The absence of a partially active band of MMP-2 indicated that MMP-14 was not involved in proMMP-2 activation (37). In support of the idea that MMP-2 activation by APC is critical for subsequent collagenase activity, the failure of APC to augment sGAG release or to stimulate collagen release in RetA-treated cultures was coordinated with a decrease in MMP-2 mRNA expression and less activation of proMMP-2 with RetA. Buisson-Legendre et al (1) found a strong correlation between levels of APC and MMP-2 activity in synovial fluid. Colocalization of APC with MMP-2 in the synovium suggested that APC might be responsible for the activation of MMP-2 in the RA joint (1). However, collagenolytic activity in our cultures was also associated with the degree of MMP-9 activation (Figure 5A). MMP-9 activation mechanisms are largely unknown (for review, see ref. 39), and in other cell types, APC does not induce activation of MMP-9 (12, 40), suggesting that it is likely not a direct effect in our cartilage cultures.

An additional pathway through which APC may potentiate cartilage breakdown is via cleavage of PARs. Of the 4 PARs, the one most strongly implicated in the pathogenesis and progression of inflammatory arthritis is PAR-2 (41, 42). Importantly, PAR-2 is also increased in OA human cartilage, can be up-regulated by IL-1α and TNFα, and its activation leads to increased MMP-1 and MMP-13 synthesis (10, 11). Unlike the other family members, PAR-2 is not cleaved/activated by thrombin, but it can be activated through canonical cleavage by APC (43). Activation of PARs is closely regulated by EPCR and TM, increased concentrations of which favor endogenous protein C activation by TM-bound thrombin and subsequent proteolysis of PARs and other substrates by APC (44). We found that chondrocyte expression of EPCR and TM was induced by IL-1α and TNFα, both of which are implicated in cartilage degradation in OA (45). A critical role of EPCR and PAR activation in the effects of APC has been demonstrated in keratinocytes (46), and it will be important to determine if similar pathways are active in chondrocytes.

In conclusion, we identified APC as a potential activator of MMPs in cartilage. All of the proteins necessary for the autocrine production and activation of APC are expressed by chondrocytes and may be regulated by cytokines implicated in cartilage degradation in vivo. The APC pathway may be a novel target for the control of MMP activation and cartilage degradation in OA.

AUTHOR CONTRIBUTIONS

Dr. Little 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. M. T. Jackson, M. M. Smith, C. J. Jackson, Little.

Acquisition of data. M. T. Jackson, S. M. Smith, Xue.

Analysis and interpretation of data. M. T. Jackson, M. M. Smith, C. J. Jackson, Little.

Manuscript preparation. M. T. Jackson, M. M. Smith, C. J. Jackson, Xue, Little.

Statistical analysis. M. T. Jackson, M. M. Smith.

Ancillary