The Role of Human HtrA1 in Arthritic Disease*

Human HtrA1 belongs to a widely conserved family of serine proteases involved in various aspects of protein quality control and cell fate. Although HtrA1 has been implicated in the pathology of several diseases, its precise biological functions remain to be established. Through identification of potential HtrA1 targets, studies presented herein propose that within the context of arthritis pathology HtrA1 contributes to cartilage degradation. Elevated synovial HtrA1 levels were detected in fluids obtained from rheumatoid and osteoarthritis patients, with synovial fibroblasts identified as a major source of secreted HtrA1. Mass spectrometry analysis of potential HtrA1 substrates within synovial fluids identified fibronectin as a candidate target, and treatment of fibronectin with recombinant HtrA1 led to the generation of fibronectin-degradation products that may be involved in cartilage catabolism. Consistently, treatment of synovial fibroblasts with HtrA1 or HtrA1-generated fibronectin fragments resulted in the specific induction of matrix metalloprotease 1 and matrix metalloprotease 3 expression, suggesting that HtrA1 contributes to the destruction of extracellular matrix through both direct and indirect mechanisms.

Human HtrA1 (L56) is a member of the HtrA 3 (High temperature requirement) family of serine proteases, a well defined group of proteases sharing many of the characteristics associated with bacterial HtrAs (1). Such features include a highly conserved trypsin-like serine protease domain and at least one PDZ domain at the C terminus. In addition, HtrA1 contains an insulin-like growth factor-binding protein domain and a Kazal-type serine protease inhibitor motif at its N terminus (2). Originally identified as a gene down-regulated in SV40-transformed fibroblasts (2), HtrA1 has since been implicated in the modulation of various disease pathologies. Recent reports suggest that HtrA1 plays a protective role in various malignancies because of its tumorsuppressive properties (3)(4)(5)(6). Studies have shown that HtrA1 is downregulated in cancerous tissue as compared with normal tissue and that overexpression results in the inhibition of tumor cell growth and proliferation both in vitro and in vivo (5). In contrast to tumor tissue, HtrA1 expression is up-regulated in skeletal muscle of Duchenne muscular dystrophy (7) and in cartilage of osteoarthritic joints (8). Therefore, up-regulation of HtrA1 in osteoarthritic joints may contribute to the development of this debilitating disease.
Progressive degradation of components of the extracellular matrix plays an important role in the pathogenesis of arthritic diseases (9,10). The destruction of the major cartilage components is driven by members of all classes of proteases, including serine proteases, although the matrix metalloproteases (MMPs) are considered to be the primary instigators (11)(12)(13). Elevated levels of various MMPs have been identified in the diseased joints of both osteoarthritis (OA) (14 -16) and rheumatoid arthritis (RA) (17) patients, originating primarily from synovial fibroblasts and chondrocytes (9,18,19). Within the cartilage matrix, interstitial collagens are the main targets of degradative collagenases such as MMP-1 (collagenase-1) (11,12). The primary function of these MMPs is in the degradation of native fibrillar collagen, resulting in the generation of collagen fragments that are then further cleaved by gelatinases, MMP-2 and MMP-9, and stromelysin (MMP-3) (12). However, for collagenases to gain access to these substrates, small proteoglycans and interfibrillar cross-links must first be removed (12). Recently, it was suggested that several proteoglycans and glycoproteins in the extracellular matrix may serve as potential substrates for HtrA1 (20 -22) and that this protease may therefore be pivotal in the onset of destructive joint pathology seen in arthritic disease. In the present study, we have demonstrated a potential direct and indirect involvement of HtrA1 in cartilage destruction in arthritic diseases.
Isolation of Human Synovial Fibroblasts (HSF)-HSF were isolated, harvested, and cultured using a method previously described (24). Briefly, synovial tissue was obtained after synovectomy from patients with osteoarthritis or rheumatoid arthritis under approval of the local Ethics Committees. Samples were washed with Dulbecco's calciumand magnesium-free phosphate-buffered saline prior to digestion with collagenase (750 units/ml in phosphate-buffered saline) for 1 h at 37°C. After digestion, the synovial fibroblasts were expanded in culture flasks containing Dulbecco's modified Eagle's medium and nutrient mix F12 (1:1) supplemented with 10% fetal calf serum, penicillin (50 international units/ml), streptomycin (50 g/ml), L-glutamine (0.3 mg/ml), hydrocortisone (4 g/ml), insulin (250 g/ml), and transferrin (250 g/ml). Cells were grown in a humidified incubator at 37°C containing 5% CO 2 in air. At least four separate cell lines were cultured and used between passages 3 and 5.
Zymography-HtrA1 (2 g) was loaded on a precast zymogram containing casein as substrate (Bio-Rad) and developed following the manufacturer's instructions.
Identification of Potential HtrA1 Substrates in Synovial Fluid-Synovial fluids (10 l) identified as containing low levels of HtrA1 were incubated with or without recombinant HtrA1 (5 g) for 3 h at 37°C and then analyzed by SDS-PAGE. Candidate protein substrates were then extracted from the SDS gel and identified using mass spectrometry.
Degradation of fibronectin was determined by incubation of various amounts of recombinant HtrA1 (0.5, 1, 3, and 5 g) with 10 g of fibronectin in 50 mM Tris-HCl, pH 8.5, 150 mM NaCl for 3 or 18 h at 37°C. Fibronectin (10 g) and HtrA1 (5 g) were incubated separately under the same conditions and served as controls. HtrA1 inhibitor was preincubated with HtrA1 for 20 min at room temperature at a final concentration of 1, 2, or 5 M prior to the addition of fibronectin substrate. Samples were analyzed on a Coomassie-stained Schaegger gel.
Quantification of Secreted MMP-3-MMP-3 protein levels in culture supernatants were determined using an MMP-3-specific ELISA kit according to the manufacturer's instructions (R&D Systems Inc.).
Effect of Fibronectin Fragments on MMP and TIMP Expression-To prepare fibronectin fragments, 10 g of fibronectin was incubated with 5 g of HtrA1 for 16 h at 37°C in buffer in 50 mM Tris-HCl, pH 8.5, 150 mM NaCl. Subsequently, these samples were applied with and without 5 M HtrA1 inhibitor to synovial fibroblasts for 24 h before determining MMP and TIMP mRNA levels by RT-PCR.
Statistical Analysis-Two-tailed Student's t-test was used to determine statistical significance between values. A p value of Ͻ0.05 was considered statistically significant. Values are expressed as the mean Ϯ S.E. thritic patients as compared with normal individuals. To investigate the potential importance of HtrA1 in the progression of arthritis, initial studies were carried out to determine the level of HtrA1 in synovial fluids. In the present study, the role of HtrA1 in arthritic diseases was further investigated and its possible mechanisms of action elucidated.
Identification of HtrA1 in Synovial Fluid-HtrA1 levels within synovial fluids from either OA or RA patients were determined by ELISA using purified recombinant HtrA1 as a standard reference (Fig. 1, inset). HtrA1 levels in arthritic patients were compared with those detected in synovial fluids taken from non-arthritic trauma patients. HtrA1 levels were elevated in OA (38 Ϯ 6 ng/ml) and RA (19 Ϯ 4 ng/ml) synovial fluids as compared with non-arthritic individuals (5 Ϯ 1 ng/ml) (Fig. 1).
HtrA1 Secretion by Human Synovial Fibroblasts-To identify potential sources of HtrA1 in synovial fluids, the secretion of HtrA1 by HSF isolated from either OA or RA patients was determined using the HtrA1-specific ELISA. Levels of HtrA1 secreted by human foreskin fibroblasts were also analyzed and served as a non-arthritic control. HtrA1 levels were significantly elevated in supernatants from OA and RA HSF as compared with human foreskin fibroblasts at all time points tested (Fig. 2).
Production, Purification, and Inhibition of Recombinant HtrA1-To further analyze the effects of HtrA1 in the context of arthritis, we generated a recombinant His-tagged HtrA1 lacking the N-terminal insulinlike growth factor-binding protein and serine protease inhibitor domain in E. coli (Fig. 3A). Affinity-purified HtrA1 was Ͼ98% pure as determined by SDS-PAGE (Fig. 3B, lane 1) and was recognized by the monoclonal HtrA1 antibody (lane 2). In addition, this HtrA1 construct was confirmed to be proteolytically active as shown by zymography (lane 3). This truncated version of HtrA1 is thought to be of physiological relevance as HtrA1 possesses autoproteolytic activity generating N-terminal truncations in in vitro translation (8) as well as cell culture systems (data not shown). An additional tool for these studies was a potent inhibitor of HtrA1 that was obtained from a high throughput screen (Fig. 3C). In the presence of this HtrA1 inhibitor, proteolytic activity was inhibited in a dose-dependent manner with an IC 50 of 0.21 M as determined by HtrA1-dependent digestion of resorufin-labeled casein (Fig. 3D).
Identification of Potential Substrates of HtrA1-The identification of substrates naturally occurring within the joint would be beneficial for investigating the role of HtrA1 in this destructive disease. The potential of cartilage matrix degradation has been previously demonstrated by the ability of HtrA1 to digest small proteoglycans such as decorin and biglycan (20,22). To identify possible substrates of HtrA1 within arthritic joints, OA and RA synovial fluids low in HtrA1 (OA, 6.2 ng/ml; RA 2.2 ng/ml) were digested with purified recombinant HtrA1, and degraded proteins were identified by SDS-PAGE followed by mass spectrometry. Among the candidate substrates identified, fibronectin was considered to be of particular interest because of its involvement in maintenance of cartilage matrix integrity through its interaction with collagen (28). In addition, elevated levels of fibronectin fragments produced following proteolytic degradation have been detected in both OA and RA synovial fluids (29 -34). These fragments may play an important role in arthritic diseases because of their ability to stimulate chondrocytes and synovial fibroblasts to produce MMPs (35)(36)(37). Protease assays with purified components were performed to confirm fibronectin degradation. Various amounts of recombinant HtrA1 were incubated with 10 g of fibronectin for 3 and 18 h at 37°C. A fragment of fibronectin migrating at ϳ30 kDa could be detected after 3 h of incubation with 3 and 5 g of HtrA1 (Fig. 4A). Further incubation of fibronectin with HtrA1 for 18 h led to generation of additional fibronectin fragments ranging from 50 to 175 kDa that increased in intensity with increasing amounts of HtrA1 (Fig. 4A). The most prominent fibronectin fragment generated after 18 h was the 30-kDa fragment. Fibronectin degradation was completely abolished by addition of 5 M HtrA1 inhibitor. In addition, inhibition of HtrA1-dependent fibronectin digestion was investigated by preincubating HtrA1 with 1, 3, or 5 M HtrA1 inhibitor prior to adding fibronectin. Degradation of fibronectin was completely abolished by addition of 3 and 5 M HtrA1 inhibitor (Fig.  4B). The appearance of HtrA1 as one or two bands after prolonged incubation is due to its autoproteolytic activity. Inhibition of HtrA1 also inhibits autoproteolytic activity resulting in only one HtrA1 band.
Effect of Recombinant HtrA1 on MMP/TIMP Production by HSF-Fibronectin fragments are present in micromolar levels in synovial fluid of arthritic joints and have been shown to up-regulate MMP production in human synovial fibroblasts and chondrocytes (34,35,37,38). To investigate the potential regulatory effects of HtrA1 on cellular functions  resulting from the production of fibronectin fragments, we examined MMP expression in HSF following incubation with recombinant HtrA1. In addition, we investigated the effects of HtrA1 on expression of the naturally occurring inhibitors of MMPs (tissue inhibitor of matrix metalloproteinases, TIMPs). Both OA and RA HSF were incubated for 24 h in serum-free conditions with and without increasing amounts of recombinant HtrA1. The expression of MMP-1, -2, -3 and TIMP-1 and -3 was determined by semi-quantitative PCR. MMP-1 and -3 mRNA levels were markedly increased in both OA and RA HSF treated with HtrA1. In contrast, expression of MMP-2, TIMP-1, and TIMP-3  remained unaffected by HtrA1 (Fig. 5, A and B). These stimulatory effects of HtrA1 were almost completely abolished following addition of 5 M HtrA1 inhibitor (Fig. 5, A and B), suggesting that the proteolytic activity of HtrA1 is crucial for up-regulation of MMPs. In contrast, the HtrA1 inhibitor had no effect on the regulation of MMP synthesis in response to the pro-inflammatory cytokine interleukin 1 (Fig. 5C), confirming the specificity of the HtrA1 inhibitor. To determine whether the stimulatory effects of HtrA1 on MMP regulation could also been observed at the protein level, MMP-3 release was monitored by a specific ELISA. MMP-3 was constitutively secreted by both OA (130 Ϯ 20 pg/ml) and RA (70 Ϯ 20 pg/ml) HSF (Fig. 4, D and E). These levels were increased Ͼ100-fold in OA (15,720 ng/ml Ϯ 4,370 pg/ml) and 7-fold in RA (500 pg/ml Ϯ 40 pg/ml) HSF following incubation with 2.5 g/ml of HtrA1 for 24 h (Fig. 5, D and E). Consistent with our previous results, these stimulatory effects of HtrA1 were abolished by the addition of 5 M HtrA1 inhibitor.
Effect of Fibronectin Fragments on MMP Production by HSF-To confirm that the up-regulation of MMP-1 and MMP-3 production by HtrA1 was mediated through the generation of fibronectin fragments, HSF were incubated with HtrA1-digested human fibronectin. Stimulation of HSF with fibronectin fragments for 24 h resulted in a marked increase in the expression of MMP-1 and MMP-3 (Fig. 6). These effects were specifically due to fibronectin fragments, as undigested fibronectin alone had little or no effect. Furthermore, incubation with HtrA1 inhibitor confirmed that these effects were predominantly because of the fibronectin fragments and not contaminating HtrA1. Fibronectin fragments had no effect on TIMP-1 or -3 expression levels. We point out that in the absence of HtrA1 or fibronectin fragments the basal levels of MMP mRNA detected in cells derived from OA patients can be heterogeneous (Figs. 5 and 6). The reason for the observed differences is unknown, although they might be best explained by the heterogeneity of the clinical condition itself, for example by alterations in disease phenotype between individuals (39).

DISCUSSION
HtrA1 levels in cartilage explants from OA patients have previously been shown to be increased 7-fold as compared with cartilage from non-arthritic individuals (8). In addition, in vitro studies have demonstrated that cartilage damage can induce a significant increase in HtrA1 production by resident chondrocytes (20). By using an HtrA1-specific ELISA, we have demonstrated that HtrA1 levels are also increased 3-7fold in the synovial fluids from both OA and RA patients, with a significantly higher level of HtrA1 being detected in OA synovial fluid. Thus, HtrA1 levels could serve as an additional marker for diagnosis of disease. Further analyses revealed that HSF extracted from OA and RA joint tissue constitutively secrete HtrA1. This up-regulation of HtrA1 production appeared to be disease specific, as human foreskin fibroblasts secreted 2-3-fold less HtrA1 than OA and RA HSF. The secretion of HtrA1 by OA and RA HSF suggests that the synovial membrane may also be an important source of HtrA1 within the arthritic joint, in addition to the articular cartilage. Although both OA and RA HSF secrete  HtrA1, it is perhaps the highly active state of OA chondrocytes that accounts for the differences seen in the levels of HtrA1 in OA and RA synovial fluid (20).
Destruction of articular cartilage is a common feature of OA and RA (9, 10). We identified the extracellular matrix glycoprotein fibronectin as a natural substrate of HtrA1, suggesting a direct role of HtrA1 in matrix degradation. HtrA1 effectively degraded purified human fibronectin, generating fragments of various sizes including several prominent fragments ranging from 83 to 170 and 29 -30 kDa. Elevated levels of fibronectin fragments ranging from 24 to 200 kDa have been identified both in OA and RA synovial fluid in micromolar concentrations (34,38) and are involved in the regulation of numerous cellular activities (40,41). The central involvement of fibronectin fragments in cartilage catabolism is highlighted by their ability to decrease proteoglycan synthesis (42) and enhance the release of several MMPs (34,37). In the present report, we have demonstrated that HtrA1 has the potential to up-regulate MMP expression and secretion in arthritic joints through activation of HSFs. Regulation of MMP-1 and -3 expression in HSF by HtrA1 was shown to be dependent on the production of fibronectin fragments. Therefore, we suggest that HtrA1 degrades fibronectin present within the cell culture system and the resulting fibronectin fragments instigate the expression and secretion of MMPs. Additional evidence was provided from the findings that neither HtrA1 nor fibronectin fragments had any effect on TIMP-1 and -3 expression by HSF as has previously been reported (34).
The present study provides evidence for a detrimental role of HtrA1 in both OA and RA, leading to a working hypothesis for its biological functions in this context. Not only does HtrA1 have the potential to directly degrade cartilage through proteolytic cleavage of extracellular matrix components such as fibronectin, cartilage oligomeric matrix protein, biglycan, decorin, fibromodulin, aggrecan, and reduced collagen (20 -22), but it seems also to act indirectly through its ability to stimulate the overproduction of MMPs by synovial fibroblasts. Therefore, specific inhibition of HtrA1 production or activity in arthritic joints may serve as a novel therapeutic strategy for treatment of arthritic diseases.